Spatial biology using single-cell mass spectrometry imaging and integrated microscopy

Spatial biology using single-cell mass spectrometry imaging and integrated microscopy | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Spatial biology using single-cell mass spectrometry imaging and integrated microscopy Alexander Potthoff, Jan Schwenzfeier, Marcel Niehaus, Sebastian Bessler, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6751055/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 15 Oct, 2025 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Abstract Spatial biology pursues the analysis of cells in their native microenvironment, often with regard to morphology and gene- or protein expression. The spatial analysis of lipid and metabolic profiles by matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) can add another layer of molecular information. A seamless integration of MSI data with other (multimodal) methods at a single-cell level is often hampered by insufficient co-registration and resolution. Here, we introduce a MALDI-MSI based method that integrates in-source bright-field and fluorescence microscopy, allowing for a coupled (sub)cellular investigation of the same sample in both modalities. Presented examples from cell culture and tissue analysis include the visualization of intracellular lipid distributions in macrophages during phagocytosis, and the heterogeneity of lipid profiles of tumor infiltrating neutrophils correlated to their individual microenvironments. The achieved combination of lipid profiling with morphological features and protein expression on the single-cell level constitutes a powerful method for cell biology. Biological sciences/Cell biology/Cellular imaging Biological sciences/Molecular biology Biological sciences/Biological techniques/Imaging/Molecular imaging Biological sciences/Biological techniques/Mass spectrometry Biological sciences/Biological techniques/Metabolomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Two cells in a tissue or a cell culture are never exactly identical, even if they share the same genetic code and microenvironment. This heterogeneity can be a consequence of differences in age, ontogeny, or tissue imprinting resulting in alterations of morphology, gene expression, molecular composition, and ultimately function. In living systems, individual cells are part of interdependent multicellular structures, i.e. tissues, and undergo a constant exchange within local regulatory networks as part of the almost inestimable complexity that is life. Thus, to characterize the identity and state of a cell, the description should follow a holistic approach and include information from a wide range of analytical techniques 1 . In recent years, single-cell RNAseq and spectral flow cytometry have emerged as valuable tools to profile cellular heterogeneity. Similarly, advances in single-cell proteomics provide information on the heterogeneity on the protein level, albeit with much lower throughput 2,3 . Under the umbrella of spatial biology, the emergence of spatial-omics technologies including deep visual proteomics, multiplexed immunohistochemistry and spatial RNAseq have added information on morphology and the spatial relationship of cells 4–7 . A complementary addition to these techniques is the analysis by high spatial resolution mass spectrometry imaging (MSI). Importantly, this method can augment information on active transcripts and protein expression with the resulting downstream production of lipids and further metabolites, as well as glycans and peptides 8 . The most widely used MSI approach is based on matrix-assisted laser desorption ionization (MALDI). Herein, finely focused laser pulses are applied to pre-defined pixel positions on the sample surface, resulting in spatially defined mass spectra that can be processed to reveal signal intensity maps for up to hundreds of analytes differentiated based on their molecular weight. To increase signal intensity for a wide range of analyte classes the technique can be amended with laser postionization (MALDI-2) 9 . Typically, eukaryotic cells stretch over distances between 10 and 100 µm and display a wide variety in shape and form. Prominent sub-cellular structures like the nucleus or other large organelles, but also morphological features, show sizes between 3 and 10 µm. Their spatially resolved analysis therefore requires pixel sizes of no more than 1x1 µm². The current bench-mark figure on the pixel size of MALDI-MSI, however, is at about 5x5 µm². Less established experimental techniques dedicated to high resolving power, such as transmission-mode or inverse geometry (t-)MALDI-2-MSI and others, enables pixel sizes ≤ 1x1 µm² and can thus fulfill this criterion 10–16 . Recent application of MALDI-2 and t-MALDI-2-MSI for single-cell mass spectrometry has revealed a strong molecular heterogeneity for lipids and metabolites within clonal populations of cells and seemingly homogeneous tissue regions demonstrating its high potential for cell biology research 17–20 . To contextualize these metabolic phenotypes with existing knowledge on the single-cell level, however, MALDI-MSI has to be connected to prevalent microscopy techniques conducted on the same slide. To date, the general application of this contextualization on the same slide is strongly limited by three main factors: Spatial resolving power of MALDI-MSI, a precise co-registration with optical modalities, and automation of data analysis. Especially in the context of tightly packed cellular systems, such as tissue, shortcomings of the first two factors lead to erroneous assignment of mass spectrometric information to individual cells and hamper the generation of unambiguous cell-specific molecular information, while the manual annotation of cells hampers the analysis of statistically relevant numbers of cells 17,18,21 . To address these challenges, we here present the development of hard- and software solutions on a prototype ion source coupled to an orthogonal time-of-flight (oTOF) mass analyzer. It utilizes transmission mode for ultra-high spatial resolution and is equipped with laser postionization (MALDI-2) to increase sensitivity for the analysis of lipids and metabolites 9,15,22,23 . While previous implementations of t-MALDI ion sources have only hinted at the potential to utilize the employed optical components for microscopy of the sample inside the ion source 15,24 , we here make full use of this option by including in-source bright field (BF) as well as fluorescence microscopy (FM). Results Description and benchmarking of transmission MALDI-2-MSI with integrated scanning microscopy A full technical description of the optical setup and preparation protocols is presented in the methods. In short the ion source of a Bruker timsTOF-fleX MALDI-2 was amended for t-MALDI analysis. Dedicated lighting strategies enable the use of BF as well as FM inside the ion source ( Fig. 1 A,B ) . Importantly, laser delivery and in-source microscopy share essential components of the optical beam path and for stage movement. In the xy-plane (sample surface), the site of laser ablation is therefore directly linked to a specific position on the optical microscope image. This design feature inherently co-registers both modalities by utilizing the same coordinate system for in-source microscopy and MALDI-MSI ( Fig. 1 A ) . Next to providing highly resolved orientation to set up imaging runs, FM-images acquired inside the ion source constitute an essential link for precise co-registration to other optical modalities. In z-direction the inherent coupling of the focal plane of optical microscopy and laser delivery enables a fast and easy setup of experiments based on autofocusing routines. To enable FM analysis prior to MSI analysis, we have developed dedicated staining protocols optimized to omit or strongly reduce chemical alterations and spatial delocalization of the analyte content of interest. The newly developed methods and pipelines were benchmarked with regard to microscopic image quality and conservation of chemical and morphological integrity. For this, we conducted combined FM and t-MALDI-2-MSI on coronal cryo-sections of fresh-frozen mouse cerebellum. This tissue exhibits histological features across different lateral scales each presenting distinct lipid profiles. It has been intensely used as model system across a wide range of microscopic and MSI methods 25–28 and is well characterized with regard to its lipidome based on the analysis of lipid extracts 29 . Details on sample preparation and data acquisition are described in the methods. To inspect and demonstrate the quality of our newly developed staining-based FM analysis on the same tissue section prior to t-MALDI-MSI analysis, we employed small molecule stains, targeting the nuclei and F-actin, as well as immunofluorescence (IF) to target calbindin expressed in Purkinje cell bodies ( Fig. 1 C ). Slide scanning FM recorded prior to matrix application in an external microscope (VS200, Evident) at 50X displays the expected structures and morphology for all three staining methods in high quality and corroborates the feasibility of the employed protocols ( Fig. 1 D ) . After matrix application by resublimation, the sample was transferred into the MALDI ion source followed by in-source slide scanning FM (50X, DAPI channel). Utilizing the FM image of the DAPI channel generated inside the ion source as an intermediate enables a straightforward co-registration of external FM images with MALDI-MSI ( Extended Data Fig. 1 A). In contrast, a conventional “cross-modality” co-registration of FM and MSI on a cellular or sub-cellular level is often hampered by difficulties in identifying pairs of fiducial markers across modalities that represent the exact same spatial location 30 . Subsequently, t-MALDI-2-MSI data at a pixel size of 1x1 µm² was acquired. Overview and zoomed in regions of the resulting MS images display distinct intensity distributions demonstrated on the example of ion signals of two different phosphatidylcholines (PC) ([PC(38:6) + H] + at m/z 806.57 and [PC(40:6) + H] + at m/z 834.60) and the molecular ion signal of the Hoechst stain of the nuclei at m/z 452.23 (Fig. 1 D). Direct comparison of t-MALDI-2-MSI with microscopy confirms that morphological features at the different lateral scales are largely retained during matrix application step and subsequent MSI analysis (Fig. 1 D). The overlay of both modalities reveals a strong correlation with histological features of the white matter and granular layer, discernable based on the F-actin staining and the nuclei. Similarly, signal intensity distribution of PC(40:6), associated with the Purkinje cell layer 31 shows good correlation with the IF-based specific staining of these structures (Fig. 1 D). Overall, deviations for co-registration were found to be less than 1 µm ( Extended Data Fig. 1 B-C and SI notes 1 ). To explore the depth of information for lipid analysis available from pre-stained t-MALDI-2-MSI, mass spectral information was matched against a previously published full lipidomics profile for murine cerebellum derived from bulk 29 . This analysis reveals that ~ 82% of the reported molecular lipid content of cerebellum are detected in the t-MALDI-2 measurement ( SI notes 2 ). This amounts to 66 annotated ion species that produce contrast rich intensity distributions in t-MALDI-2-MSI in positive and/or negative ion mode measurements ( Extended Data Table 1 ). Comparison of the detected signal intensities with the molar content of the different lipoforms of each of the detected lipid classes shows a good agreement with the literature 29 ( Extended Data Fig. 2 and SI notes 2 ). Notably, this includes lipid classes, such as phosphatidylethanolamine (PE) and phosphatidylserine (PS) that may be affected by the chemical fixation used during sample preparation 32 . Lipid analysis at sub-cellular resolving power To demonstrate the potential of histology guided t-MALDI-2-MSI for the analysis of sub-cellular structures, we investigated THP-1-derived macrophages. Cells were cultured and differentiated directly in chamber slides. Freshly differentiated cells were incubated with pHrodo red E. coli bioparticles for 2 h. Upon phagocytosis, these particles become fluorescent and pinpoint acidified phagolysosomes inside the cells. For analysis, cells were briefly fixed and their actin filament and nuclei stained following MALDI compatible protocols previously reported 17 . Slide scanning fluorescence microscopy on an external slide scanner prior to coating with MALDI matrix reveals heterogeneously pronounced phagocytosis with some cells showing a large number of internalized bioparticles (Fig. 2 A). After coating with MALDI matrix using resublimation, in-source FM (50X) and t-MALDI-2-MSI data at 1.5x1.5 µm² pixel size was collected for a few hundred cells (Fig. 2 B). Image co-registration between the externally generated FM image and MSI data was performed by using the in-source FM data as an intermediate ( Extended Data Fig. 1 D). The overlay of both modalities corroborates a high-fidelity co-registration on the level of the employed MALDI pixel size. Areas visualized by the pHrodo red stain in FM accurately co-align with the distribution of specific lipid signal intensities in MSI revealing little to no leakage or diffusion of lipid content or other preparation artifacts that may compromise (sub-)cellular integrity (Fig. 2 C, D) 22 . To investigate the molecular distributions specifically related to phagolysosomes, we used subcellular spatial information provided by the fluorescence signal produced by the uptake of pHrodo bioparticles. Utilizing the high-fidelty co-registration, we overlayed this histological information to t-MALDI-2-MSI data and identified those pixels that correlate with phagolysosomes to generate phagolysosome-specific mass spectra. (Fig. 2 C,D). Lipid annotation was based on previously published results for THP-1 derived macrophages and E. coli , respectively 20,33 . Comparative analysis of the resulting phagolysosome-specific molecular information revealed an increase of signal intensity for numerous lipid species (Fig. 2 E). Confirmed by the respective MS images, lipids such as PE(33:1) and PE(34:2) spatially correlate with the phagolysosomes identified by fluorescence ( Fig. 2 C,D). Both PE species are abundantly expressed in the membrane of E. coli cells 33 . Notably, PE(33:1) is not detected in non-phagocytosing THP-1 cells (Fig. 2 B-D, red trace). Its presence, confirmed by additional on-cell MS/MS experiments (described in Methods), provides a direct link to the phagocytized pathogen material and corroborates the highly accurate co-registration of both modalities. Other tentatively assigned lipid species linked to the phagolysosomes can be attributed to degradation of E. coli -derived phospholipids during phagocytosis 34,35 . These degradation products are most likely the result of enzymatic activity and reactions with reactive oxygen species inside the phagolysosomes. They include lipid metabolites such as the diacylglycerol DG(33:1), that can be directly linked to the enzymatic cleavage of PE(33:1) and lipid species tentatively identified as lysophosphatidylglycerols (LPG). The latter can be linked to the enzymatic breakdown of phosphatidylglycerols (PG) as well as cardiolipins (CL) both highly expressed in the membrane of E.coli 33,35 (Fig. 2 E). Automated single-cell analysis of lipids in their histological context in tumor tissue In a second exemplary study, we applied our methods to the single-cell molecular analysis of tumor tissue. This tissue type typically contains a large number of different cell types in a wide range of states translating to a wide variety of metabolic microenvironments (tumor microenvironment, TME) of different size and shape and a highly heterogeneous lipid profile across the tissue 36 . The employed 4T1 mouse mammary carcinoma model is highly malignant with locally aggressive growth and exhibits metastasis to regional lymph nodes and distant organs, particularly lung, liver, and bone 37,38 . It has diverse inflammatory TMEs with a large number of infiltrating immune cells such as T-cells, macrophages, neutrophils and natural killer cells present 39,40 . The site and cell-specific molecular profile of these immune cells is of great importance for their fate and function and is believed to be in constant interdependence with their microenvironment 41 . Consequently, the investigation of this crosstalk not only requires a molecular analysis at the cellular level but also an assessment of the immediate vicinity. For analysis, a heterogeneous area of interest that includes a broad range of TMEs was selected based on H&E staining and histological annotation (Fig. 3 A). A consecutive section was prepared using MALDI-compatible staining that targets cell nuclei, F-actin, and three IF stains that selectively target immune cells, (CD45), neutrophils (Ly6G), and T3 neutrophils (DcTRAIL-R1) 41,42 . All channels were recorded on the complete section using external slide scanning FM at 50X before matrix application (Fig. 3 B). Co-registration with t-MALDI-2-MSI was based on the DAPI channel recorded inside the ion source after matrix application as an intermediate ( Extended Data Fig. 1 E). After microscopy, t-MALDI-2-MSI data was recorded in seven separate measurements across the region of interest at a pixel size of 1x1 µm² resulting in a total of approx. 7 Mpixels. For data analysis, we selected 63 molecular ion species annotated to individual lipid species based on exact mass (< 3 ppm) (see Extended Data Fig. 3 for selected images and Table S2 ). Spatial distribution for a number of glycero- and glycerophospholipids such as [PC(34:1) + H] + at m/z 760.58 reveal a sizeable heterogeneity across the tumor sample (Fig. 3 C, Extended Data Fig. 3 ). Signal intensity of the ether lipid [PC(O-32:0) + H] + ,detected at m/z 720.59, correlates with the neutrophil-specific Ly6G-stain (Fig. 3 D-F). This is consistent with a significant overexpression of this lipid species in mouse neutrophils reported previously 43,44 . Other specific signal intensity distributions also directly correlate with structures visible in the microscopy data. As expected 45 , signal intensity for triacylglycerides (TG) such as [TG(52:5) + H] + correlates with the adipose tissue (Fig. 3 C). Acylcarnitine, detected at m/z 428.38 as [ACAR(18:0) + H] + is an important intermediate in lipid metabolism and a reported marker for hypoxia 46 . Its signal intensity distribution correlates with the detection of tumor-associated T3 neutrophils identified by the DcTRAIL-R1 staining that have been described to be found in hypoxic areas 41,42 (Fig. 3 E). Heme B, detected at m/z 616.18, is directly associated with erythrocytes and, hence, serves as marker for hemorrhage or blood vessels 47 (Fig. 3 F). While general assignment of lipid profiles to TMEs has been reported previously, we here demonstrate the analysis of single cells within their tissue environment in statistically relevant numbers. For this we adapted our previously reported workflow for the analysis of single cells in culture to the analysis of tissue sections 17,20 . Robust and reliable cell segmentation, necessary for automated analysis 21 , was provided based on the external fluorescence microscopy images using Mesmer by DeepCell (Fig. 3 G) 48 . This resulted in segmentation masks for each cell and enabled the calculation of a variety of morphometric parameters using CellProfiler 49 . To produce single-cell mass spectra, cell segmentation masks were projected on the t-MALDI-2-MSI data, again taking advantage of the inherent link between in-source FM and the acquired MSI data ( Extended Data Fig. 1 E). An overlay of these projected masks with the lipid neutrophil marker [PC(O-32:0) + H] + and comparison to the specific Ly6G staining demonstrates the high-fidelity co-registration and corroborates the ability of our combined method to deduct single-cell mass spectra (Fig. 3 G). Single-cell molecular data is generated for all 63 annotated lipid species by summing the respective signal intensities over all pixels allocated to a specific cell using the respective mask. Subsequently all values are normalized to the sum of all signal intensities for the respective cell. Cell segmentation and subsequent data analysis resulted in combined morphometric and molecular data for 63,408 individual cells. Notably this analysis retains the histological context of all cells and includes parameters describing location within the tissue and fluorescence intensity for all employed channels. This enables the identification of CD45 + , Ly6G + , and DcTRAIL-R1 + cells based on intensity thresholding for each specific channel. For visualization, this information can be projected onto the segmentation canvas, thereby identifying the location and local density of different sub-groups of immune cells in the tissue section (Fig. 4 A). Processing of the combined IF and MSI single-cell data enabled spatial biology analysis on the level of lipids and metabolites, by classifying cells based on molecular similarity. In this context, analysis on the single-cell basis is especially important for scarce or isolated cells like specific immune cells because their molecular profile may substantially differ from the surrounding TME. This type of analysis is demonstrated on the example of infiltrating neutrophils inside the 4T1 tumor section. To classify neutrophils based on their molecular profile independent of their microenvironment and vice versa, single-cell data was split into two groups based on Ly6G staining. The resulting subgroups contained 3,410 Ly6G + neutrophils and 59,998 Ly6G − cells, respectively. Both groups were independently analyzed using Uniform Manifold Approximation and Projection (UMAP) and appropriate clustering algorithms. To identify molecularly similar cell types within the tissue, all Ly6G − cells were subjected to k-means clustering, resulting in eight discernable clusters. By assigning a specific color to each cluster, the outlines of each cell can be filled in with the respective color and projected back onto their original location (Fig. 4 B). The resulting map of classified cells displays clearly discernable tissue micro-environments where molecularly similar cells cluster in specific areas of the tissue. Clusters IV, V, and VIII contain cells of the skin, the adipose tissue and the wider tumor periphery, respectively. Inside the viable tumor tissue, a clear separation into TMEs reveals a layer-like structure with frayed edges (clusters II, III, and VI). Cluster VII contains cells directly adjacent to the tumor edges and away from the adipose tissue but also cells inside the tumor close to its necrotic core. Cluster I contains cells allocated to hypoxic regions of the tumor. The direct connection to the morphometric data allows to characterize each cluster with regard to immune infiltration. Figure 4 G, summarizes the total amount of cells as well as the numbers of CD45 + , Ly6G + , and DcTRAIL-R1 + cells that are found within the boundaries of the respective cluster. Next to TMEs, we investigated the molecular heterogeneity of neutrophils. For this, the Ly6G + cells were analyzed using a UMAP and classified using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) (Fig. 4 C). This results in a clear separation of eight clusters. To provide linguistic distinction from clusters described for the tissue microenvironments, results for neutrophils will be referred to as subtypes. Colored respectively and projected onto the cell map, these different subtypes can be found in distinct areas of the tumor (Fig. 4 D). Apparent from the comparison of the images depicted in Fig. 4 , some subtypes show a clear spatial correlation with a specific microenvironment and histological features ( Fig. 4 G, Extended Data Fig. 4 A ) . Subtype 6 is almost exclusively located in the skin. Neutrophils are found in skin in comparably high number and have important and specific barrier functions 50 . Our data suggests a significantly altered lipid profile for these neutrophils. For example, in contrast to all other subtypes, subtype 6 shows a low expression of PC(O-32:0) ( Extended Data Fig. 4 B,C). Neutrophils of subtypes 5 and 7 are located in specific tissue microenvironments that correlate with the adipose tissue (clusters IV and V) in the periphery of the tumor. Their lipid profiles show an increase of TG and DG lipid species (Fig. 4 H). Subtype 4 is predominantly found in non-adipose tissue close to the tumor (cluster VII). Similarly, subtype 2 also populates this peripheral region but is additionally found inside the tumor close to its necrotic core. Lipid profiles for both subtypes are low in DG and TG. While their overall cell density in this area is relatively high, they do not form agglomerates. Interestingly, neutrophils in this periphery of the tumor show strong molecular similarities to their respective microenvironment ( Extended Data Fig. 4 C). This may be explained by an uptake of extracellular vesicles or cell debris produced by the surrounding tissue and “recycling” of the lipid content by the neutrophils. Notably, the expression of lipid species that require specific enzymatic machinery for their synthesis such as PC ether lipids (PC-O) do not show similarities to the surrounding tissue but are upregulated only in the respective neutrophil populations ( Extended Data Fig. 4 C). The exclusive co-localization of these ion species to Ly6G + -cells also precludes the involvement of measurement artefacts, such as analyte diffusion into this convergence of lipid profiles between neutrophils and their respective TME (Fig. 3 ). With the exception of subtype 4 all neutrophils found in the tumor periphery and the adipose tissue show a relatively high density of immune cells in their direct vicinity (Fig. 4 A). This observation is corroborated by a neighborhood analysis that counts the average number of immune cells within a distance of 15 µm to the cell edge of each individual neutrophil (Fig. 4 G). Inside the tumor, neutrophils of the subtype 1 are found mostly in the viable tumor tissue (clusters II, III, and V) (Fig. 4 G and Extended Data Fig. 4 A). Notably, their lipid profile shows little similarity to their respective surrounding TMEs ( Extended Data Fig. 4 C). This may indicate a decreased up-take and turnover of lipids or a reduced residence time within the tumor tissue. While the overall density of subtype 1 neutrophils in the tumor increases towards the adjacent adipose tissue, they are mostly isolated with relatively few other immune cells around them (Fig. 4 G). In contrast, neutrophils of subtypes 3 and 8 form strong agglomerations (Fig. 4 E). These subtypes are located within TME I with subtype 8 being confined to the center of the region. Their neighborhood analysis results in high numbers of overall immune cells in their direct vicinity (Fig. 4 G). About 90% of subtype 8 neutrophils are identified as DcTRAIL + . The lipid profile of this subtype strongly differs from that of the surrounding hypoxic TME I ( Extended Data Fig. 4 C). It presents an increased signal for ACAR 18:0 and an increase in signal intensities for lipid metabolites such as LPCs (Fig. 4 H) and DG, while TGs are mostly absent. This may indicate an upregulation of specific lipid metabolism in this subtype. Similar to subtype 8, neutrophils of the subtype 3 are almost exclusively found in agglomerates inside TME I but are located further away from its center. In contrast to subtype 8, however, only 17% are DcTRAIL-R1 + . The lipid profile of subtype 3 also contains high signal intensities for ACAR 18:0 but reduced levels of LPC and DG. Again, in contrast to subtype 8, the lipid profile of subtype 3 neutrophils shows great similarities to its surrounding TME I. This increased molecular crosstalk may indicate a longer residence time of neutrophils within this hypoxic TME as opposed to the viable TMEs II, III, and VI. Discussion In its own right, MSI analysis can provide specific molecular information of single cells, thereby delivering an important puzzle piece towards the understanding of individual cellular behavior. For a more holistic approach, however, the resulting lipidomic or metabolomic information has to be put into the context of the cell type, differentiation status and state but also has to regard its current tissue macro- and microenvironment 1 . Here, a tight connection to optical microscopy analysis has long since been identified as a key feature in the integration of MALDI-MSI into the “biomedical toolbox” 51 . Conventionally, microscopy of the same sample prior to MALDI-MSI is restricted to the limited analytical capabilities offered by optical microscopy of unstained tissue, e.g. by autofluorescence or phase-contrast microscopy 52,53 . Staining procedures applied after application and removal of the matrix and especially laser irradiation that precedes microscopy, can damage or distort the tissue and impede binding properties for antibody based analysis 12 . High quality microscopy analysis, including IF, is therefore typically carried out on adjacent tissue sections introducing well-known challenges in accuracy of co-registration. In particular, the use of adjacent sections for microscopy strongly limits the analysis of specifically targeted single cells 30,51,54,55 . In contrast, the presented technique includes slide scanning microscopy outside and inside the ion source of the same sample recorded prior to ultra-high resolving power t-MALDI-2-MSI. Enabled by dedicated sample preparation techniques, this combination lays the groundwork for the large scale, automated, and reliable generation of information-rich highly resolved data for both modalities on the same sample. Two exploratory studies highlight the advantages of the described method for the molecular analysis with cellular and even sub-cellular resolution. Using the phagocytosis of E. coli bioparticles by macrophages as an example, we demonstrate how the technique can be used to investigate metabolic processes inside sub-cellular compartments like the phagolysosomes. Assignment of bacterial PE 33:1 and PE 34:2 confirmed the uptake of E.coli bioparticles. The additional identification of increased amounts of the respective DG species suggests their partial degradation. In addition, the tentative identification of several LPG species provides further useful leads to the metabolic breakdown of E. coli derived CL and PG by enzymatic degradation and/or oxidation of lipids inside the phagolysosome 56 . Overall, this study illustrates a new approach to analyze molecular profiles of subcellular compartments. In future applications, this largely untargeted technique may be used to investigate and monitor metabolic turnover on the organelle level of specific single cells. In our study of the 4T1 tumor model, we demonstrate spatial biology analysis on the lipid level for highly heterogeneous tissue. While previous work on single-cell MSI from tissue sections has been limited to the manual annotation of small numbers of cells 17–19,57 , automated segmentation based on slide scanning (immuno-)fluorescence microscopy resulted in the identification of more than 63K cells and allowed for the identification and classification of common immune cells (CD45 + , ~ 23.7K cells; 38% of all cells), relatively rare neutrophils (Ly6G + , ~ 3.4K cells; 5.4%), and very rare tumor associated T3-neutrophils (DcTRAIL-R1 + , 276 cells; 0.4%) within the tissue. Importantly, the segmentation, in combination with highly accurate co-registration of the microscopy with t-MALDI-2-MSI results, allowed for the generation of single-cell mass spectra for each cell. Using statistical analysis, cells were clustered based on their lipid profile, revealing eight distinct tissue microenvironments inside the tumor and around its periphery. Likewise, an independent clustering of just the Ly6G + cells revealed eight subtypes of neutrophils inside the investigated tissue region. Comparison of lipid profiles between neutrophils and their direct tissue environment revealed striking similarities in some cases. These results allow for a new perspective on the uptake and recycling of lipids by specific neutrophil subtypes. They also reveal distinctive differences between neutrophils and their direct environment such as the increased levels of ether lipid species. In addition to a general upregulation of this lipid species also described in the literature 44 we are able to identify subtype and TME specific differences in its expression. Neighborhood analysis of all investigated immune cells in the direct vicinity of the neutrophils provides a first hint at their swarming behavior. While some of the subtypes represent relatively isolated cells, others are found in agglomeration with neutrophils and other immune cells. The strength of the molecular investigation at cellular level becomes especially apparent in the examination of the hypoxic regions of the tumor. This TME shows a signal intensity increase for ACAR 18:0, as well as a high density of immune cells both described as markers for hypoxia 46 . The inner core of this TME is populated by densely packed T3-neutrophils (most of them of subtype 8), concurring with the literature 41,42 . A second group of agglomerated mostly DcTRAIL-R1 − neutrophils (subtype 3) is also located inside the same hypoxic area but further away from its center. The subtyping now enables an in-depth analysis of underlying lipid profiles and integration with the state of research. While both subtypes, for example, express increased levels of ACAR 18:0, only subtype 8 (mainly DcTRAIL-R1 + cells) shows increased levels of fully saturated LPCs. These mediator are known to be important in metastasis formation and in the recruitment of phagocytes during apoptosis 58,59 and our data points to a previously unknown tentative connection with T3-neutrophils. While a full investigation of the underlying biological mechanisms is not within the focus of this publication, the presented proof-of-concept studies showcase the great potential of the technique. The seamless integration of morphometric data with (sub-)cellular lipid and metabolite analysis significantly expands the informational space available for each individual cell inside the tissue and enables a direct contextualization with its tissue microenvironment in situ . Overall, the presented techniques constitute a versatile set of tools for an improved integration of single-cell MSI into existing workflows. Embedded within a full suite of spatial biology methods, in future applications, it will allow for the combination of IF-based information that may describe the regulation of specific genes and the expression of individual proteins with the expression of lipids or metabolites in the same cell and its direct surrounding 60 . This addition of “spatial biology of lipids and metabolites” on the single-cell level provides an important additional layer of information to complete pathway analysis from the transcript and proteome level down to the regulation of the molecular products. In can be a valuable asset to review and validate downstream effects of genetic or transcriptomic variation as well as protein inhibition but also help to investigate regulatory functions of small molecules on gene and protein expression. In this, the methods described here have great potential to help entangling the influence of endogenous and exogenous factors such as immune cell infiltration, infection, or drug delivery on cellular behavior in a specially resolved manner. They are, however, by no means limited to these applications but can be applied to decipher open biological questions in a wide range of tissue and cell types. Methods Reagents, resources and software All reagents, resources, and software used are listed in the table of resources ( Table S1 ) Cell culture preparation THP-1 ACC-16 cells (DSMZ) were cultured in T25 cell culture flasks at 37 °C and 5% CO 2 in RPMI 1640 Medium (Lonza), which was supplemented with 2 mM L-glutamine, 10% fetal bovine serum and 1 mM sodium pyruvate. About 10,000 cells per chamber were sown in Millicell EZ Slide 8-well chamber slides (Sigma-Aldrich). Differentiation to M0 macrophages was induced by exposure to 20 ng/ml phorbol 12-myristate 13-acetate (PMA, Sigma-Aldrich) for 24 h followed by a 3-day rest phase in PMA-free medium. Two hours before harvest, pHrodo™ Red E. coli BioParticles (Thermo Fisher) were added to the culture medium resulting in a concentration of 95 µg/mL. Cells were washed with PBS and fixed for 5 min using 4% formaldehyde (low-methanol, Roth). F-Actin and nuclei were stained according to the previously described protocol by Bien et al . using 1X CellMask Green Actin Tracking Stain (Thermo Fisher) diluted in PBS for 30 min and 1 µg/mL Hoechst 33342 (Sigma-Aldrich) dissolved in PBS for 5 min 1 . Cells were washed with PBS after every staining step. After staining, cells were washed using 150 mM ammonium acetate (Sigma-Aldrich) and dried at room temperature for 2 h. Mouse model All animal experiments were carried out in accordance with local animal welfare guidelines approved by the responsible authorities (North Rhine-Westphalia State Agency for Nature, Environment and Consumer Protection, LANUV, Protocol No. 81-02.04.2018.A010). Female BALB/c mice (Charles River Laboratories) were used at the age of 8 to 12 weeks. Mice were housed under a 12-hour light/dark cycle with ad libitum access to food and water. For tumor implantation, 10 6 4T1 cells were resuspended in 25 µl of DMEM cell culture medium (Thermo Fisher) and then injected orthotopically into the lower left mammary fat pad of the mice. Tumors were allowed to grow for 9 days and tumor size was monitored daily with a digital caliper. On day 9 after tumor implantation, the animals were sacrificed. Brain tissue was harvested from untreated mice used as control group for the mentioned study. Tissue sectioning Mouse brain and tumor tissue were embedded in Epredia M-1 Embedding Matrix (Thermo Fisher) and snap-frozen in liquid nitrogen. 10 µm (cerebellum) or 8 µm (tumor) thick tissue sections were produced with a cryostat (CM 3050 S, Leica Biosystems) at -20 °C. Sections were thaw-mounted onto SuperFrost glass slides (Thermo Fisher) for H&E-staining or IntelliSlides (Bruker) for t-MALDI-2 MSI measurements. Samples for t-MALDI-2 MSI were vacuum-sealed and stored at -80 °C before further use. Tissue fixation and fluorescence staining Frozen tissue sections were slowly brought to room temperature in the vacuum-sealed package. After opening the seal, sections were further dried under a gentle nitrogen stream for 10 min. With a Super PAP Pen (Science Services) a contour was drawn around each section, followed by another 10 min drying under a gentle nitrogen stream, to create a hydrophobic barrier and thus enabling pipetting aqueous solutions on and off the sample. The following washing, fixation and staining steps were all performed by pipetting ~100 µl of the accordant solution onto the sample and removing it after the specified time. Slides with tissue sections were placed on a metal plate standing in ice water. Samples were quickly fixed in 4 % formaldehyde (low-methanol, Sigma-Aldrich) in PBS for 5 minutes and subsequently washed with PBS for 30 s. After tissue fixation, the primary antibody, Ly6g (ab307167, Abcam), DcTRAIL-R1 (130-110-873, Miltenyi Biotec), CD45 (130-110-798, Miltenyi Biotec) or Anti-Calbindin (ab229915, Abcam), diluted 1:500 (v/v) in 1% BSA (Serva) in PBS was applied to the sample and incubated for 60 min. The sample was then washed 3 times for 30 s in PBS. In case of a dual antibody staining, the secondary antibody, Alexa Fluor 594 (ab307167, Abcam), diluted 1:1000 (v/v) in 1% BSA in PBS was applied and incubated for 60 minutes, followed by 3 times 30 s washing steps in PBS. 5 µg/ml Hoechst33342 (Sigma-Aldrich) and 1X CellMask Green Actin Tracking Stain (Thermo Fisher) diluted in PBS were applied and incubated for 5 minutes. After staining, the slide was lifted from the metal plate and parts of the hydrophobic barrier were removed at the bottom of the slide with a pipette tip. Samples were then washed with 2 ml of ammonium acetate (Sigma-Aldrich) each, by letting it flow carefully over the section. For final drying, samples were placed under a gentle N 2 stream for 20 minutes. Tissue H&E staining Hematoxylin and eosin (H&E) stains of tissue sections were performed on untreated consecutive sections directly after sectioning. Sections were stained for 30 s with hematoxylin (Sigma-Aldrich) with subsequent bluing in tap water for 2 min. Afterwards, the tissue was stained for 30 s with eosin (Sigma-Aldrich) and then rinsed with water. External microscopy Fluorescence images of cell cultures and tissue sections were recorded at 50X using the DAPI, FITC, TxRed, Cy3 and Cy5 channels of a slide scanning microscope (VS200, Evident) equipped with a SpectraSplit 7 filterset (Kromnigon). For each channel individually, the exposure time was set by performing the auto exposure procedure of the microscope on a region with high signal intensity for the respective fluorophore. This procedure was conducted prior to the imaging run, and the entire slide was subsequently imaged with a constant exposure time, with a maximum exposure of 500 ms. H&E-stained tissues were recorded at 20X in bright field mode. Image processing was performed with OlyVIA (Evident). Transmission MALDI optics A schematic of the optical set-up used for material ejection for the t-MALDI-2-MSI analysis and in-source brightfield and fluorescence microscopy image generation is depicted in Fig. 1A-C of the main article. The sample is mounted on a three axis piezo stage (SmarAct) and is illuminated homogeneously by 15 green LEDs (LED528EHP, Thorlabs) mounted on top of a 50X objective (M Plan Apo NUV HR 50X, Mitutoyo) with a custom-designed 3D printed LED ring. Light from the sample is collected by the objective and focused by a tube lens (TTL100-A, Thorlabs) onto the sensor of a digital camera (XCG-CG160, Sony). A simple top-view observation of the sample is in addition possible by the pre-installed camera of the prototype timsTOF fleX MALDI-2 mass spectrometer. The beam path of the MALDI laser (smartbeam, wavelength 355 nm, pulse duration 7 ns) of the timsTOF fleX MALDI-2 mass spectrometer is re-routed and condensed through a 6:1 telescope optic and co-aligned to the microscopy path with a dichroic mirror (L-07246, Laseroptik Garbsen). For recording fluorescence microscopy images, a filter cube (DFM2/M, Thorlabs) with a filter set (DFM2T1, Thorlabs) for the suitable fluorescence channel (MDF-BFP, DAPI; MDF-FITC, FITC; MDF-MCHC, mCherry; all Thorlabs) consisting of excitation and emission filters and a dichroic mirror is mounted in front of the tube lens. In fluorescence mode, the LED ring is turned off and fluorescence is excited by the respective fiber-coupled LED (M385FP1, DAPI; M455F3, FITC; MINTF4, mCherry; all Thorlabs) mounted with a collimating optic directly to the filter cube. The camera is integrated in the software by using the XC-SDK 2018 (Sony). The trigger signal for image acquisition of the camera and fluorescence excitation is provided as a “position reached” event by the controller of the piezo stage and is handled by a digital delay pulse generator (9200 Sapphire Series, Quantum Composers). With this pulse generator, the fluorescence excitation time was set from 10 ms to 2 s per image depending on the sample and fluorescence channel. MALDI-matrix coating by resublimation To achieve a reproducible and homogeneous coating of the sample slides with MALDI-matrix, a custom-designed resublimation chamber was developed. A schematic of this setup is shown in Extended Data Fig. 5A-B . Under atmospheric pressure conditions, the hot-side is preheated to 80 °C and 0.5 ml of 10 mg/ml CHCA MALDI-matrix (Sigma-Aldrich) in acetone solution is pipetted into the matrix-reservoir. With the acetone evaporating rapidly, the MALDI-matrix is left forming a uniform layer on the bottom of the matrix reservoir. The sample is then clamped onto the cold-side at room temperature and the whole device is sealed and evacuated. After reaching a final pressure of 10 -2 mbar, the cold-side is first cooled to -18 °C, then the hot side is heated to 150 °C to start the sublimation and resublimation. After 60 min, the heating of the hot-side is turned off and the cold-side is heated to room temperature before venting the system to prevent condensation of ambient humidity on the sample. The sample is then put onto a room temperature metal block in a saturated atmosphere of 0.5% ethanol in water at 70 °C for 1.5 min for recrystallization. To demonstrate differences in contrast for in-source BF microscopy induced by the choice of matrix, cerebellum samples were coated with DHB and HABA MALDI-matrix (both Sigma-Aldrich) using the same settings as described abovefor CHCA (Extended Data Fig. 5C) . Autofocus of optics The MALDI laser and microscope camera utilize the same objective for focusing and observation in the transmission-MALDI setup employed in this work. A schematic of the setup is depicted in Fig. 1A in the main article. The focal plane of the MALDI-laser is adjusted to match that of the camera by fine adjusting the second telescope lens of the laser beam path. After aligning the focal planes, all optics are kept at a constant position. The objective (M Plan Apo NUV HR 50X, Mitutoyo) has a small focal depth of 1.4 µm and samples have to be in focus during optical microscopy and MALDI-MSI. This can be achieved by methodically adjusting the z-position of the sample with regard to the objective while recording sequential optical microscope images at z-increments of 1 µm. For each of these images a Laplacian filter is applied and the variance of the result is calculated, which serves as a measure of image sharpness. The stage is then moved to the z-position corresponding the maximum sharpness image and a fine-scan with 0.1 µm z-distance is performed in the range of +- 1 µm in the same way as before to find the z-position of highest sharpness at a submicron precision. Since the focal planes of the optical microscope and the MALDI laser are aligned, this optical focusing ensures an optimal focus of the MALDI-laser on the sample. This autofocus routine can be executed both BF and FM mode. The python code of this program AutoFocusClass.py is provided in Data S1 in the supplementary information. In-source microscopy workflow The scale factor to convert pixel on the camera sensor to position of the stage was calculated by using a 1951 USAF Resolution Test Target (Thorlabs). For selection of a region of interest (ROI) for an optical microscope scan in the ion source, autofocusing points are placed manually on the sample. The ROI is defined as the smallest enveloping rectangle that contains all points. The stage is then moved to each of the autofocusing points and an autofocus routine as described in “Autofocus of optics” is performed, which yields the focal z-offset of that xy-coordinate. Using a radial basis function interpolation, the z-offset is calculated for every coordinate in a 10 µm x 10 µm mesh inside the ROI and saved to a z-interpolation file. After creating the z-interpolation, the optical microscope image of the ROI is taken. For this an up to 100 µm x 100 µm wide x/y-position mesh is created in the ROI. An optical microscope image is then taken at each mesh point at the z-offset of the closest point to the xy-position from the z-interpolation file. To create a complete image of the ROI, an empty canvas of the correspondent size is created. Using the scale factor of the setup, images are cropped from the edges to a size of 100 µm x 100 µm and are placed onto the empty canvas using the center coordinates from the xy-position mesh. After creating a z-interpolation file for a ROI once, multiple modalities, such as brightfield or different fluorescence channels can be recorded without creating a new interpolation. ImageJ was used to create overlay images of different modalities. The python code of this workflow can be found in the files RecordPositions.py, ZforRecordPositions.py, Scanning.py and stitching.py in Data S1 in the supplementary information. Transmission-MALDI-2 MSI The MALDI-MSI run is setup partly using the proprietary flexImaging software (Bruker). Instead of using an external scan of the sample and teaching that to the stage coordinate system, the (multimodal) image of the ROI (see “In-source microscopy workflow”) is used. In the classic flexImaging workflow, three teaching points would be manually selected on the optical image. The corresponding stage coordinates would then be recorded by driving the sample-stage manually to the respective positions. Since center coordinates and pixel positions on the ROI image are already established from recording the ROI image, three x/y-positions from the 100 µm x 100 µm mesh (top left, bottom left, bottom right) and their corresponding stage coordinates are automatically deployed as teaching points. With these points, the setup file for the t-MALDI-2 MSI run can be automatically created, defeating any positioning errors introduced by manually selecting teaching points in the classic workflow. When the MALDI-laser is not directed at the exact middle pixel of the in-source optical microscope, fine adjustments of a few pixel offset can be applied at this stage. The automatically created setup file for the t-MALDI-2 MSI run can be opened directly in flexImaging to create measurement regions, select settings like pixel size, and laser shots per pixel. The previously acquired z-interpolation is recognized by timsControl software during t-MALDI-2 MSI measurements by providing the interpolation mesh as a geometry file. The python code for creating the flexImaging setup files can be found in Createmis.py Data S1 in the supplementary information. All t-MALDI-2 MSI experiments were performed with 25 laser shots per pixel and a pixel size of 1 x 1 µm 2 for the cerebellum and tumor sample and 1.5 x 1.5 µm 2 for cell culture samples. Signal intensity distributions are visualized using SCiLS Lab software (Bruker) using weak de-noising. Ion signal annotation For the THP-1 cells lipid signals were identified as described by Schwenzfeier et al . using shotgun lipidomics by electrospray ionization (ESI) mass spectrometry on a Q-Exactive Plus orbitrap mass spectrometer (Thermo Fisher) equipped with a dual ion funnel 2,3 . Lipid extracts of 72 h PMA stimulated THP-1 macrophages were produced by an adapted protocol of Folch et al. 4 . Approximately 10 6 cells were centrifuged at 1000 g to separate them from the medium. Cells were resuspended in 1 ml CHCl 3 /MeOH (2:1, v/v) and lyzed for 10 min in an ultrasonic bath at room temperature followed by an incubation of 50 min at 4 °C in a cryoshaker at 500 RPM. Phase separation was induced by addition of 240 μl H 2 O and centrifugation at 9880 g for 10 min. The organic phase was transferred to a glass vial and the solvent was removed using a nitrogen evaporator. Lipid extracts were vacuum sealed and stored at -80 °C. For ESI-MS analysis, extracts were re-dissolved in 500 μl MeOH and measured both in positive and negative ion modes using a flow rate of 5 µl/min. A data dependent analysis workflow (DDA) was applied using tandem MS based on collision-induced dissociation (CID) with stepped normalization collision energies (NCE) of 20, 25, 30 and an exclusion time of 300 s. The mass resolving power was set to 280,000 (at m/z 200) with an AGC target of 1E6 and a maximum injection time of 250 ms. The resulting data was merged with a t-MALDI-2-MSI dataset and imported into LipostarMSI (Molecular Horizons) for annotation. The resulting feature list was hand-curated, which involved removing of isotope peaks and addition of known species from E.coli . For mouse cerebellum and the 4T1 murine tumor section, annotations were based on accurate mass ( 3 were extracted from the mean spectrum and annotated using the LipidMaps "Bulk Structure Search” 5 . In positive ion-mode sphingolipids, glycerolipids, sterol lipids and glycerophospholipids were selected as possible lipid classes either as [M+H] + - or [M-H 2 O+H] + -ions. In negative ion-mode only [M-H] - -ions were searched for. The resulting annotation lists were manually curated to remove biologically unlikely annotation (e.g. lipids with odd side chains) or lipid classes that are usually not detected in the respective ion-mode (e.g. glycerophosphoinositols in positive ion-mode). Using this method, a total of 66 ion signals for the mouse cerebellum and 63 ion signals for the 4T1 tumor sample were annotated, respectively. The lipid annotation of mouse cerebellum was subsequently matched with data from a full lipidomics analysis reported in the literature 6 . Because t-MALDI-2-MSI without the use of additional techniques like OzID or Paternò-Büchi-reactions 7,8 is not able to differentiate isomers within a lipid class (e.g. PC(16:0/18:1) and PC(16:1/18:0)) or between lipid classes (e.g. PE(16:0/18:0) and PC(15:0/16:0)) the literature data was reduced to 255 unique lipid species according to the here employed level of specificity. Of the 66 annotated m/z -values, 63 were also detected by Fitzner et al . 6 . The remaining unmatched values primarily corresponded to sulfoglycosphingolipids and fatty acids, lipid classes that were not analyzed in the original publication. Overall, the annotation accounted for approximately 82% of the total lipid mass identified by Fitzner et al ., calculated by summing all lipids found in both datasets and using the quantitative amounts reported. Lipid nomenclature and shorthand notation is reported according to Liebisch 9 . Single-cell mass spectra from tissue The external and in-source fluorescence images were co-registered using SimpleITK 2.3.1 (NumFOCUS) 10,11 . For initialization a landmark registration was used. Based on the resulting translation matrix an affine image registration was performed using Mattes mutual information as metric and the gradient descent algorithm as optimizer 12 . The number of bins for the mutual information was set to 50, the parameters for the optimizer were set as follows: learning rate (1), number of iterations (50), convergence minimum value (1E-6) and convergence window (20). The sampling strategy was set to random and 1% of the pixels were sampled. The resulting transformation matrix was combined with the initial translation matrix to result in a global transformation matrix. Due to the inherent link between the t-MALDI-2-MSI measurement and the in-source fluorescence images, the global transformation matrix could be used to resample the external fluorescence images to fit the exact region of the t-MALDI-2-MSI measurement. The precision of the registration was determined by placing pairs of landmarks at distinct features, which are visible in the external fluorescence image as well as the MALDI image of the cerebellum dataset ( Extended Data Fig. 1B ). While the precision of the placement of each individual landmark is limited by the manual selection process and differences between modalities, overall accuracy is ensured by determining the average over a total of eleven pairs of landmarks. The target registration error (TRE) was then determined by projecting the landmarks of the MALDI image into the coordinate system of the external fluorescence image using the global transformation matrix and calculating the distance to the corresponding landmarks of the external fluorescence image ( Extended Data Fig. 1C and Note S1 ). Cell masks were created by using DeepCell Mesmer 0.12.9 (Van Valen Lab) 13 based on the DAPI channel (Hoechst 33342, nuclei) and the FITC channel (CellMask Green Actin Tracking Stain, actin). Using FISCAS 2 , mass spectra were calculated for each cell by compiling MSI pixels based on their co-localized cell masks. MSI pixels assigned to multiple cell masks were scaled according to the hit coverage of the segmentation masks. Morphometric parameters were acquired using CellProfilers 4.2.1 (Cimini Lab) 14 measurement modules. Cell classification For cell classification, immune cells were identified and divided into three groups (CD45 + , LyG6 + , DcTRAIL-R1 + ) based on thresholding the mean intensities of the corresponding antibody stains in their respective fluorescence channels (Cy5, TxRed, Cy7). Ly6G + cells showed high intensities on the TxRed and Cy5 channel, while DcTRAIL-R1 + cells showed high intensities on all three channels. The clustering of the tumor microenvironment (TME) was done using k-means clustering on the standardized MS data of all Ly6G - cells. The number of clusters was determined by visual inspection of the clustering results for values of k between 6 and 10. The Ly6G + cells were also clustered using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) with an epsilon of 0.7 and a minimum number of 7 neighbors, based on a uniform manifold approximation and projection (UMAP) of their standardized mass spectra. The clusters are visualized both directly on the cell masks as well as in a UMAP of the standardized mass spectra using default parameters. Ly6G + cells were assigned to a specific tumor region, by identifying the most frequent TME label of all neighbors within 40 µm. References Methods Bien, T., Koerfer, K., Schwenzfeier, J., Dreisewerd, K. & Soltwisch, J. Mass spectrometry imaging to explore molecular heterogeneity in cell culture. Proc. Natl. Acad. Sci. U. S. A. 119 , e2114365119 (2022). Schwenzfeier, J., Weischer, S., Bessler, S. & Soltwisch, J. Introducing FISCAS, a Tool for the Effective Generation of Single Cell MALDI-MSI Data. J. Am. Soc. Mass Spectrom. 35 , 2950–2959 (2024). Belov, M. E. et al. Design and Performance of a Novel Interface for Combined Matrix-Assisted Laser Desorption Ionization at Elevated Pressure and Electrospray Ionization with Orbitrap Mass Spectrometry. Anal. Chem. 89 , 7493–7501 (2017). Folch, J., Lees, M. & Sloane Stanley, G. H. A simple method for the isolation and purification of total lipides from animal tissues. J. Biol. Chem. 226 , 497–509 (1957). Fahy, E., Sud, M., Cotter, D. & Subramaniam, S. LIPID MAPS online tools for lipid research. Nucleic Acids Res. 35 , W606-612 (2007). Fitzner, D. et al. Cell-Type- and Brain-Region-Resolved Mouse Brain Lipidome. Cell Rep. 32 , 108132 (2020). Claes, B. S. R. et al. Mass Spectrometry Imaging of Lipids with Isomer Resolution Using High-Pressure Ozone-Induced Dissociation. Anal. Chem. 93 , 9826–9834 (2021). Bednařík, A., Bölsker, S., Soltwisch, J. & Dreisewerd, K. An On-Tissue Paternò-Büchi Reaction for Localization of Carbon-Carbon Double Bonds in Phospholipids and Glycolipids by Matrix-Assisted Laser-Desorption-Ionization Mass-Spectrometry Imaging. Angew. Chem. Int. Ed. Engl. 57 , 12092–12096 (2018). Liebisch, G. et al. Update on LIPID MAPS classification, nomenclature, and shorthand notation for MS-derived lipid structures. Journal of Lipid Research 61 , 1539–1555 (2020). Yaniv, Z., Lowekamp, B. C., Johnson, H. J. & Beare, R. SimpleITK Image-Analysis Notebooks: a Collaborative Environment for Education and Reproducible Research. J Digit Imaging 31 , 290–303 (2018). Lowekamp, B. C., Chen, D. T., Ibanez, L. & Blezek, D. The Design of SimpleITK. Front. Neuroinform. 7 , (2013). Mattes, D., Haynor, D. R., Vesselle, H., Lewellyn, T. K. & Eubank, W. Nonrigid multimodality image registration. in Medical Imaging 2001: Image Processing vol. 4322 1609–1620 (SPIE, 2001). Greenwald, N. F. et al. Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. Nat. Biotechnol. 40 , 555–565 (2022). Stirling, D. R. et al. CellProfiler 4: improvements in speed, utility and usability. BMC Bioinf. 22 , 433 (2021). Declarations Resource availability Materials availability This study did not generate new unique reagents. Data and code availability MALDI MSI data, fluorescence and brightfield microscopy images, processed data and all original code have been deposited at OMERO or in the supplementary information and are publicly available as of the date of publication. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledegements We thank Dagmar Mense for help with cell culture, Mathis Richter for helpful discussions on immunohistochemistry, Alexander Fengler and Vincent Suerdieck (all University of Münster) for help with mouse tissue sectioning, Benedikt Geier (Stanford), Robert Ernst and Hyojik Yang (both University of Maryland) for helpful discussions, Henning Peise, Arne Fütterer, Andreas Wöste, Raik Milautzki (all Bruker) for hard- and software support, and Sarah Weischer and Thomas Zobel from the Münster Imaging Network Cells in Motion Interfaculty Centre for support with image analysis. Figure 1 was created in part using biorender (agreement number GO275P34H3). Financial support from the German Research Foundation to KD and OS (Z1 project of the TRR332; project no. 449437943) and to JS (project no. 544444139) is gratefully acknowledged. Author contributions Conceptualization: KD, JH, MN, AP, JS; Data curation: SB, AP, JSch, JS; Formal analysis: SB, AP, JSch; Funding acquisition: KD, JH, OS, JS; Investigation: SB; MN, AP, JSch, JS; Methodology: SB, MN, AP, JSch, JS; Project administration: KD, JS; Resources: KD, JH, EH OS, JS; Software: AP, JSch; Supervision: KD, OS, JS; Validation: KD, JH, OS; Visualization: SB, AP, JSch, JS; Writing – original draft: AP, JS; Writing – review & editing: SB, KD, JH, MN, AP, JSch, OS, JS. Competing interests The authors declare the following competing interest(s): MN and JH are employees of Bruker Daltonik GmbH & Co KG (Bremen). KD, MN, AP, and JS have patents pending regarding parts of the hardware development (WO/2024/041681, WO/2025/067576). References Rafelski, S. M. & Theriot, J. A. Establishing a conceptual framework for holistic cell states and state transitions. Cell 187 , 2633–2651 (2024). Bennett, H. M., Stephenson, W., Rose, C. M. & Darmanis, S. Single-cell proteomics enabled by next-generation sequencing or mass spectrometry. Nat. Methods 20 , 363–374 (2023). Boekweg, H. & Payne, S. H. Challenges and Opportunities for Single-cell Computational Proteomics. Mol. Cell. Proteomics 22 , 100518 (2023). Bressan, D., Battistoni, G. & Hannon, G. J. The dawn of spatial omics. Science 381 , eabq4964 (2023). Vandereyken, K., Sifrim, A., Thienpont, B. & Voet, T. Methods and applications for single-cell and spatial multi-omics. Nat. Rev. Genet. 24 , 494–515 (2023). Bodenmiller, B. Highly multiplexed imaging in the omics era: understanding tissue structures in health and disease. Nat Methods 21 , 2209–2211 (2024). Mund, A. et al. Deep Visual Proteomics defines single-cell identity and heterogeneity. Nat Biotechnol 40 , 1231–1240 (2022). Dilmetz, B. A. et al. Novel technical developments in mass spectrometry imaging in 2020: A mini review. Analytical Science Advances 2 , 225–237 (2021). Soltwisch, J. et al. Mass spectrometry imaging with laser-induced postionization. Science 348 , 211–215 (2015). Cuypers, E. et al. ‘On the Spot’ Digital Pathology of Breast Cancer Based on Single-Cell Mass Spectrometry Imaging. Anal. Chem. 94 , 6180–6190 (2022). Taylor, M. J., Lukowski, J. K. & Anderton, C. R. Spatially Resolved Mass Spectrometry at the Single Cell: Recent Innovations in Proteomics and Metabolomics. J. Am. Soc. Mass Spectrom. 32 , 872–894 (2021). Bourceau, P. et al. Visualization of metabolites and microbes at high spatial resolution using MALDI mass spectrometry imaging and in situ fluorescence labeling. Nat. Protoc. 18 , 3050–3079 (2023). Müller, M. A., Zweig, N., Spengler, B., Weinert, M. & Heiles, S. Lipid Signatures and Inter-Cellular Heterogeneity of Naı̈ve and Lipopolysaccharide-Stimulated Human Microglia-like Cells. Anal. Chem. 95 , 11672–11679 (2023). Marques, C. et al. Global and Spatial Metabolomics of Individual Cells Using a Tapered Pneumatically Assisted nano-DESI Probe. J. Am. Soc. Mass Spectrom. 34 , 2518–2524 (2023). Niehaus, M., Soltwisch, J., Belov, M. E. & Dreisewerd, K. Transmission-mode MALDI-2 mass spectrometry imaging of cells and tissues at subcellular resolution. Nat. Methods 16 , 925–931 (2019). Kompauer, M., Heiles, S. & Spengler, B. Atmospheric pressure MALDI mass spectrometry imaging of tissues and cells at 1.4-μm lateral resolution. Nat. Methods 14 , 90 (2017). Bien, T., Koerfer, K., Schwenzfeier, J., Dreisewerd, K. & Soltwisch, J. Mass spectrometry imaging to explore molecular heterogeneity in cell culture. Proc. Natl. Acad. Sci. U. S. A. 119 , e2114365119 (2022). Rappez, L. et al. SpaceM reveals metabolic states of single cells. Nat. Methods 18 , 799–805 (2021). Capolupo, L. et al. Sphingolipids control dermal fibroblast heterogeneity. Science 376 , eabh1623 (2022). Schwenzfeier, J., Weischer, S., Bessler, S. & Soltwisch, J. Introducing FISCAS, a Tool for the Effective Generation of Single Cell MALDI-MSI Data. J. Am. Soc. Mass Spectrom. 35 , 2950–2959 (2024). Ščupáková, K., Dewez, F., Walch, A. K., Heeren, R. M. A. & Balluff, B. Morphometric Cell Classification for Single-Cell MALDI-Mass Spectrometry Imaging. Angew. Chem. Int. Ed. Engl. 132 , 17600–17603 (2020). Bien, T., Bessler, S., Dreisewerd, K. & Soltwisch, J. Transmission-Mode MALDI Mass Spectrometry Imaging of Single Cells: Optimizing Sample Preparation Protocols. Anal. Chem. 93 , 4513–4520 (2021). Soltwisch, J. et al. MALDI-2 on a Trapped Ion Mobility Quadrupole Time-of-Flight Instrument for Rapid Mass Spectrometry Imaging and Ion Mobility Separation of Complex Lipid Profiles. Anal. Chem. 92 , 8697–8703 (2020). Geng, Z. et al. Enhanced MALDI-2 Sensitivity with Reflecting Post-Ionization Laser for High-Resolution MS Imaging Combined with Real-Time Microscope Imaging. Anal. Chem. 96 , 12983–12990 (2024). Fernandes, A. M. A. P. et al. Direct Visualization of Neurotransmitters in Rat Brain Slices by Desorption Electrospray Ionization Mass Spectrometry Imaging (DESI - MS). J. Am. Soc. Mass Spectrom. 27 , 1944–1951 (2016). Shariatgorji, M., Svenningsson, P. & Andrén, P. E. Mass Spectrometry Imaging, an Emerging Technology in Neuropsychopharmacology. Neuropsychopharmacol. 39 , 34–49 (2014). Agüi-Gonzalez, P., Jähne, S. & Phan, N. T. N. SIMS imaging in neurobiology and cell biology. J. Anal. At. Spectrom. 34 , 1355–1368 (2019). Schröder, H., Moser, N. & Huggenberger, S. The Mouse Cerebellum. in Neuroanatomy of the Mouse: An Introduction (eds. Schröder, H., Moser, N. & Huggenberger, S.) 153–170 (Springer International Publishing, Cham, 2020). doi:10.1007/978-3-030-19898-5_7. Fitzner, D. et al. Cell-Type- and Brain-Region-Resolved Mouse Brain Lipidome. Cell Rep. 32 , 108132 (2020). Patterson, N. H., Tuck, M., Van de Plas, R. & Caprioli, R. M. Advanced Registration and Analysis of MALDI Imaging Mass Spectrometry Measurements through Autofluorescence Microscopy. Anal. Chem. 90 , 12395–12403 (2018). Sugiura, Y. et al. Visualization of the cell-selective distribution of PUFA-containing phosphatidylcholines in mouse brain by imaging mass spectrometry. J. Lipid Res. 50 , 1776–1788 (2009). Vos, D. R. N., Bowman, A. P., Heeren, R. M. A., Balluff, B. & Ellis, S. R. Class-specific depletion of lipid ion signals in tissues upon formalin fixation. Int. J. Mass Spectrom. 446 , 116212 (2019). Kralj, T. et al. Multi-Omic Analysis to Characterize Metabolic Adaptation of the E. coli Lipidome in Response to Environmental Stress. Metabolites 12 , 171 (2022). Elsbach, P. & Weiss, J. Phagocytosis of bacteria and phospholipid degradation. Biochimica et Biophysica Acta (BBA) - Reviews on Biomembranes 947 , 29–52 (1988). Hambrey, P. N. & Mellors, A. Cardiolipin degradation by rat liver lysosomes. Biochemical and Biophysical Research Communications 62 , 939–945 (1975). Xiao, Z., Dai, Z. & Locasale, J. W. Metabolic landscape of the tumor microenvironment at single cell resolution. Nat. Commun. 10 , 3763 (2019). DuPré, S. A., Redelman, D. & Hunter, K. W. The mouse mammary carcinoma 4T1: characterization of the cellular landscape of primary tumours and metastatic tumour foci. Int. J. Exp. Pathol. 88 , 351–360 (2007). Schrörs, B. et al. Multi-Omics Characterization of the 4T1 Murine Mammary Gland Tumor Model. Front. Oncol. 10 , 1195 (2020). Hoffmann, E. et al. Profiling specific cell populations within the inflammatory tumor microenvironment by oscillating-gradient diffusion-weighted MRI. J. Immunother. Cancer 11 , e006092 (2023). Li, P. et al. Dual roles of neutrophils in metastatic colonization are governed by the host NK cell status. Nat. Commun. 11 , 4387 (2020). Ng, M. S. F. et al. Deterministic reprogramming of neutrophils within tumors. Science 383 , eadf6493 (2024). Silvestre-Roig, C., Kalafati, L. & Chavakis, T. Neutrophils are shaped by the tumor microenvironment: novel possibilities for targeting neutrophils in cancer. Sig Transduct Target Ther 9 , 1–3 (2024). Spangenberg, P. et al. msiFlow: automated workflows for reproducible and scalable multimodal mass spectrometry imaging and microscopy data analysis. Nat Commun 16 , 1065 (2025). Morgan, P. K. et al. A lipid atlas of human and mouse immune cells provides insights into ferroptosis susceptibility. Nat. Cell Biol. 26 , 645–659 (2024). Ahmadian, M., Duncan, R. E., Jaworski, K., Sarkadi-Nagy, E. & Sook Sul, H. Triacylglycerol Metabolism In Adipose Tissue. Future Lipidol. 2 , 229–237 (2007). Chughtai, K., Jiang, L., Greenwood, T. R., Glunde, K. & Heeren, R. M. A. Mass spectrometry images acylcarnitines, phosphatidylcholines, and sphingomyelin in MDA-MB-231 breast tumor models[S]. J. Lipid Res. 54 , 333–344 (2013). Anderson, D. M. G., Spraggins, J. M., Rose, K. L. & Schey, K. L. High Spatial Resolution Imaging Mass Spectrometry of Human Optic Nerve Lipids and Proteins. J. Am. Soc. Mass Spectrom. 26 , 940–947 (2015). Greenwald, N. F. et al. Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. Nat. Biotechnol. 40 , 555–565 (2022). Stirling, D. R. et al. CellProfiler 4: improvements in speed, utility and usability. BMC Bioinf. 22 , 433 (2021). Vicanolo, T. et al. Matrix-producing neutrophils populate and shield the skin. Nature 1–9 (2025) doi:10.1038/s41586-025-08741-5. Balluff, B., Heeren, R. M. A. & Race, A. M. An overview of image registration for aligning mass spectrometry imaging with clinically relevant imaging modalities. J. Mass Spectrom. Adv. Clin. Lab 23 , 26–38 (2022). Geier, B. et al. Connecting structure and function from organisms to molecules in small-animal symbioses through chemo-histo-tomography. Proc. Natl. Acad. Sci. U. S. A. 118 , e2023773118 (2021). Geier, B. et al. Spatial metabolomics of in situ host–microbe interactions at the micrometre scale. Nat. Microbiol. 5 , 498–510 (2020). Lin, B.-J. et al. MSIr: Automatic Registration Service for Mass Spectrometry Imaging and Histology. Anal. Chem. 95 , 3317–3324 (2023). Yang, E. et al. FluoMALDI Microscopy: Matrix Co-Crystallization Simultaneously Enhances Fluorescence and MALDI Imaging. Adv. Sci. 10 , 2304343 (2023). Yeung, T., Ozdamar, B., Paroutis, P. & Grinstein, S. Lipid metabolism and dynamics during phagocytosis. Current Opinion in Cell Biology 18 , 429–437 (2006). Nikitina, A. et al. A Co-registration Pipeline for Multimodal MALDI and Confocal Imaging Analysis of Stem Cell Colonies. J. Am. Soc. Mass Spectrom. 31 , 986–989 (2020). Schneider, G., Sellers, Z. P., Abdel-Latif, A., Morris, A. J. & Ratajczak, M. Z. Bioactive Lipids, LPC and LPA, Are Novel Prometastatic Factors and Their Tissue Levels Increase in Response to Radio/Chemotherapy. Molecular Cancer Research 12 , 1560–1573 (2014). Lauber, K. et al. Apoptotic Cells Induce Migration of Phagocytes via Caspase-3-Mediated Release of a Lipid Attraction Signal. Cell 113 , 717–730 (2003). Volpi, E. V. & Bridger, J. M. FISH Glossary: An Overview of the Fluorescence in Situ Hybridization Technique. BioTechniques 45 , 385–409 (2008). Data S1 Data S1 is not available with this version Additional Declarations Yes there is potential Competing Interest. The authors declare the following competing interest(s): MN and JH are employees of Bruker Daltonik GmbH & Co KG (Bremen). KD, MN, AP, and JS have patents pending regarding parts of the hardware development (WO/2024/041681, WO/2025/067576). Supplementary Files ExtendedDatalegends.docx pothoffetalSI.docx Supporting Information TableS1.xlsx Supporting Table 1 TableS2.xlsx Supporting Table 2 ExtendedDataFigures.docx Cite Share Download PDF Status: Published Journal Publication published 15 Oct, 2025 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6751055","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":467176286,"identity":"77b4b116-bca2-4233-8d75-e5676892f701","order_by":0,"name":"Alexander Potthoff","email":"","orcid":"","institution":"Institute of Hygiene, University of Münster, Germany","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Potthoff","suffix":""},{"id":467176287,"identity":"f8d95ea8-9f43-4335-9370-39bbe4a69e4e","order_by":1,"name":"Jan Schwenzfeier","email":"","orcid":"","institution":"Institute of Hygiene, University of Münster, Germany","correspondingAuthor":false,"prefix":"","firstName":"Jan","middleName":"","lastName":"Schwenzfeier","suffix":""},{"id":467176292,"identity":"5edad631-b910-4f1d-9362-65cf24d8d6c1","order_by":2,"name":"Marcel Niehaus","email":"","orcid":"","institution":"Bruker Daltonics GmbH \u0026 Co. KG, Bremen, Germany","correspondingAuthor":false,"prefix":"","firstName":"Marcel","middleName":"","lastName":"Niehaus","suffix":""},{"id":467176288,"identity":"8428c32f-b2a4-4b6f-8d85-1cf738a96c7a","order_by":3,"name":"Sebastian Bessler","email":"","orcid":"","institution":"Institute of Hygiene, University of Münster, Germany","correspondingAuthor":false,"prefix":"","firstName":"Sebastian","middleName":"","lastName":"Bessler","suffix":""},{"id":467176289,"identity":"f00dbe01-f1bd-4241-b5d3-11dade1f1774","order_by":4,"name":"Emily Hoffmann","email":"","orcid":"","institution":"Clinic of Radiology, University of Münster, Münster Germany","correspondingAuthor":false,"prefix":"","firstName":"Emily","middleName":"","lastName":"Hoffmann","suffix":""},{"id":467176290,"identity":"ed58f92d-e223-459c-9bc5-f1e134bb9a33","order_by":5,"name":"Oliver Soehnlein","email":"","orcid":"https://orcid.org/0000-0002-7854-0694","institution":"Institute of Experimental Pathology, University of Münster, Germany","correspondingAuthor":false,"prefix":"","firstName":"Oliver","middleName":"","lastName":"Soehnlein","suffix":""},{"id":467176293,"identity":"7c717a12-f908-4ecb-8db9-9cf2b82437d4","order_by":6,"name":"Jens Höhndorf","email":"","orcid":"","institution":"Bruker Daltonics GmbH \u0026 Co. KG, Bremen, Germany","correspondingAuthor":false,"prefix":"","firstName":"Jens","middleName":"","lastName":"Höhndorf","suffix":""},{"id":467176291,"identity":"e828e467-36dc-40f6-b7c9-dfe8c70d129b","order_by":7,"name":"Klaus Dreisewerd","email":"","orcid":"https://orcid.org/0000-0002-7619-808X","institution":"Institute of Hygiene, University of Münster, Germany","correspondingAuthor":false,"prefix":"","firstName":"Klaus","middleName":"","lastName":"Dreisewerd","suffix":""},{"id":467176285,"identity":"3a1c0355-a82f-43e5-b429-46f493bb1414","order_by":8,"name":"Jens Soltwisch","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA90lEQVRIiWNgGAWjYBACxgYwZSEHZ0IAG0EtEsYILWwEtECBRCLCCkJamNt7D3/8USGRvuF2cwPTzTabPPn5zcc+MJTZ4HZYz7k0aZ4zErkb7hxsYM5tSys2OMaWPIPhXBpuLTNyzJgZ24BabiS2/85tO5y4gY3HmIGx7TBuLfPfGH/82SaRbnAjEWTL/8T5bfyfgVr+47GFx0CCt00iAarlQGLDMR5moJYDePySYwbyi+FMkJacc8mJG46lGTMknEvGqcWw/YwxMMRs5PlupD9gzimzS5zffPgxw4cyO9xaGrAKJ+DUwMAgj0duFIyCUTAKRgEEAAALLVMQ8foPVgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-0258-1561","institution":"Institute of Hygiene, University of Münster, Germany","correspondingAuthor":true,"prefix":"","firstName":"Jens","middleName":"","lastName":"Soltwisch","suffix":""}],"badges":[],"createdAt":"2025-05-26 13:00:36","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6751055/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6751055/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41467-025-64603-8","type":"published","date":"2025-10-15T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":85368585,"identity":"19725ece-5748-48a0-9cc3-334f4f6cd2c6","added_by":"auto","created_at":"2025-06-25 07:09:07","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":4716118,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTransmission MALDI-2-MSI with integrated scanning microscopy for subcellular resolving power\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) A schematic drawing of the t-MALDI-2 setup illustrates the general layout. The sample is illuminated homogeneously by an LED ring. Light is collected by the microscope objective and focused onto the sensor of a digital camera. The MALDI laser-beam is co-aligned using a dichroic mirror and focused onto the sample by the same objective. Ions generated by the MALDI process with optional MALDI-2 are extracted to a mass analyzer.\u003c/p\u003e\n\u003cp\u003e(B) A horizontal cut through the schematic displays the fluorescence microscope components. Fluorescence is excited by a fiber-coupled LED and filtered by an excitation filter, a dichroic mirror and an emission filter.\u003c/p\u003e\n\u003cp\u003e(C) To enable fluorescence microscopy and t-MALDI-2-MSI on the same sample, different staining steps are performed following a brief fixation. Samples are imaged with an external slide scanning fluorescence microscope, and are then prepared for MALDI-MSI by homogeneous coating with MALDI matrix (\u003cstrong\u003eExtended Data Fig. 5A, B, D\u003c/strong\u003e). To enable high fidelity co-registration of the fluorescence microscopy image, a second fluorescence image is recorded inside the ion source prior to t-MALDI-2-MSI (\u003cstrong\u003eExtended Data Fig.\u0026nbsp; 1A, D-E\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e(D) An external fluorescence image of an 8 µm thick section of a mouse cerebellum recorded prior to matrix application using Hoechst 33342 (nuclei, DAPI channel, blue), CellMask Green Actin Tracking Stain (actin, FITC channel, green) and Alexa Fluor 594 Anti-calbindin antibody (calbindin, TxRed channel, red). The different layers of the brain (molecular layer, granular layer, and arbor vitae), as well as the Purkinje cells are well distinguishable. The external FM image is overlayed with the ion signal intensity distribution of [PC(40:6)+H]\u003csup\u003e+\u003c/sup\u003e at m/z 834.60 (grey) revealing a successful co-registration of both modalities. The overlay of three ion signal intensities of the molecular ion of Hoechst 33342 at \u003cem\u003em/z\u003c/em\u003e 452.23 (cyan), [PC(38:6)+H]\u003csup\u003e+\u003c/sup\u003e at \u003cem\u003em/z\u003c/em\u003e 806.67 (magenta) and [PC(40:6)+H]\u003csup\u003e+\u003c/sup\u003e at \u003cem\u003em/z\u003c/em\u003e 834.60 (yellow), affirms the ability to produce images of high contrast while preserving and reproducing the biological structure of the sample.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/d9f128b80f8cc7072ad16799.png"},{"id":85368595,"identity":"5fdba9ba-0b9a-4d97-8dc3-304639887738","added_by":"auto","created_at":"2025-06-25 07:09:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":3496759,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLipid analysis of cultured cells at sub-cellular resolving power\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Fluorescence image of THP-1 derived macrophages recorded before MALDI matrix application with a slide scanning microscope using Hoechst 33342 (nuclei, DAPI channel, blue), CellMask Green Actin Tracking Stain (actin, FITC channel, green) and pHrodo (phagolysosomes, Cy3 channel, red).\u003c/p\u003e\n\u003cp\u003e(B) An overlay of three different ion signal intensity distributions measured using t-MALDI-2-MSI with 1.5x1.5 µm² pixel size of the molecular ion of Hoechst 33342 (\u003cem\u003em/z\u003c/em\u003e 452.23, blue), [PE(33:1)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 704.52, red) and [PC(34:1)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 760.58, green) reveals a strong cell-to-cell heterogeneity.\u003c/p\u003e\n\u003cp\u003e(C) and (D) Zoom in from (A) and (B), marked in each as a blue and red square, respectively, demonstrates intracellular differences in lipid distribution. To investigate the intracellular molecular distributions related to phagocytosis, masks for the regions with pHrodo+ (red) and remaining cell (green) are created based on the FM image. The outlines of the masks identified by the pHrodo stain are overlayed with the MSI results to demonstrate the quality of co-registration\u003c/p\u003e\n\u003cp\u003e(E) Comparative analysis of the resulting region-specific mass spectra from masks as shown in (C) and (D) reveal an increase of signal intensity for a number of lipid species inside the pHrodo\u003csup\u003e+\u003c/sup\u003e regions. Identified \u003cem\u003em/z\u003c/em\u003e-values are denoted in a table with tentative assignments, where applicable.\u0026nbsp;\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/8737711f09193961cd4e3fc0.png"},{"id":85368576,"identity":"2ec65815-8d50-491e-ad27-8b866e58a04c","added_by":"auto","created_at":"2025-06-25 07:09:04","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":6364863,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e:\u003c/strong\u003e \u003cstrong\u003eAutomated single-cell analysis of lipids in their histological tissue context\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) A BF image of an H\u0026amp;E stain of a 4T1 tumor section was recorded and annotated to identify histological features and broader tissue microenvironments. The approximate ROI of the t-MALDI-2-MSI conducted on a consecutive section is marked with a blue rectangle in the H\u0026amp;E-stained section.\u003c/p\u003e\n\u003cp\u003e(B) The section was stained for FM using Hoechst 33342 (nuclei, DAPI channel, blue), CellMask Green Actin Tracking Stain (actin, FITC channel, grey), IF stains for neutrophil cells (Ly6G, TxRed channel, red), immune cells (CD45, Cy5 channel, yellow), and T3 neutrophils (DcTRAIL-R1, Cy7 channel, not pictured in (B)) prior to MALDI matrix application and imaged on a slide scanning fluorescence microscope.\u003c/p\u003e\n\u003cp\u003e(C) Ion signal intensity distribution measured with t-MALDI-2-MSI of acylcarnitine [ACAR 18:0+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 428.38, yellow), [PC(O 32:0)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 720.59, magenta), [PC(34:1)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 760.58, cyan) and [TG 52:4+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 855.75, blue) are presented as an overlay corroborating the histological annotations from (A) and (B). Acylcarnitine is a reported marker for hypoxia, PC(O-32:0) a marker for neutrophils in mice and TG correlates with the presence of adipocytes.\u003c/p\u003e\n\u003cp\u003e(D) A zoom-in on the region marked as blue squares in the FM (B) and MSI (C) images highlights a region annotated with intratumoral hemorrhage. The FM image shows Hoechst 33342 (nuclei, DAPI channel, blue), CellMask Green Actin Tracking Stain (actin, FITC channel, grey), IF stain for neutrophil cells (Ly6G, TxRed channel, red) and immune cells (CD45, Cy5 channel, yellow). For MSI data, the ion signal intensities of [cholesterol-H\u003csub\u003e2\u003c/sub\u003eO+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z \u003c/em\u003e369.35, yellow), [PC(O-32:0)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 720.59, magenta), [LPC(16:0)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 496.34, cyan) are overlayed.\u003c/p\u003e\n\u003cp\u003e(E) A zoom-in on the region marked as red squares in the FM (B) and MSI (C) images highlights a region annotated with a lymphocyte accumulation. The FM image shows an IF stain for neutrophil cells (Ly6G, TxRed channel, red), immune cells (CD45, Cy5 channel, yellow) and T3 neutrophils (DcTRAIL-R1, Cy7 channel, cyan). For MSI data, the ion signal intensities of [ACAR 18:0+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 428.38, yellow), [PC(O-32:0 )+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 720.59, magenta) and [PC(O-38:5)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 794.61, cyan) are overlayed.\u003c/p\u003e\n\u003cp\u003e(F) A zoom-in on the region marked as green squares in the FM (B) and MSI (C) images highlights a region showing a blood vessel. The FM image shows Hoechst 33342 (nuclei, DAPI channel, blue), CellMask™ Green Actin Tracking Stain (actin, FITC channel, grey), IF stain for neutrophil cells (Ly6G, TxRed channel, red) and immune cells (CD45, Cy5 channel, yellow). For MSI data, the ion signal intensities of the molecular ion of Heme B (\u003cem\u003em/z\u003c/em\u003e 616.18, yellow), [PC(O-32:0)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 720.59, magenta) and [SM(34:1;O2)+H]\u003csup\u003e+\u003c/sup\u003e (\u003cem\u003em/z\u003c/em\u003e 703.57, cyan) are overlayed.\u003c/p\u003e\n\u003cp\u003e(G) The contours of 63,408 cells (cell masks) were generated based on the FM image using DeepCell Mesmer, inside the region measured with t-MALDI-2-MSI. A zoomed in region is depicted overlayed with the cell masks. The same cell masks are also overlayed with the t-MALDI-2-MSI data based on co-registration of both modalities with the ion signal intensity of [PC(O 32:0)+H]\u003csup\u003e+\u003c/sup\u003e at m/z 720.59 selected exemplarily. Using FISCAS \u003csup\u003e20\u003c/sup\u003e, mass spectra for each cell mask are calculated by combining spectral data of co-localizing MSI pixels.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/a9e9cd7ec3d7cbcae3efd673.png"},{"id":85368575,"identity":"2298bb39-4d2d-4a85-a353-7fe7291af07a","added_by":"auto","created_at":"2025-06-25 07:09:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":6279355,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNeutrophil lipid profile in dependence of their state and localization in the tumor tissue\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Cell masks colored corresponding to their FM-based classification are projected onto their original position of the t-MALDI-2-MSI measurement region. Ly6G\u003csup\u003e+\u003c/sup\u003e cells are depicted in red, CD45\u003csup\u003e+\u003c/sup\u003e cells in yellow and DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells in cyan. All other cells are marked dark grey\u003c/p\u003e\n\u003cp\u003e(B) The 59,998 Ly6G\u003csup\u003e-\u003c/sup\u003e cells were separated to 8 different clusters (tissue microenvironments) using k-means clustering on the standardized single-cell MS data. Cell masks for Ly6G\u003csup\u003e-\u003c/sup\u003e cells are colored according to their cluster and projected onto their original position of the t-MALDI-2-MSI measurement region. All other cells are marked dark grey.\u003c/p\u003e\n\u003cp\u003e(C) 3,410 Ly6G\u003csup\u003e+\u003c/sup\u003e cells were identified based on FM and selected for analysis. They are plotted in a UMAP based on a list of 63 annotated ion signals. Each dot represents an individual cell while the distance between dots visualizes the similarity of the standardized lipid profiles of the cells. Using DBSCAN, the UMAP reveals eight clearly discernable clusters (neutrophil subtypes) which are color coded.\u003c/p\u003e\n\u003cp\u003e(D) Cell masks for Ly6G\u003csup\u003e+\u003c/sup\u003e cells are projected onto their original position of the t-MALDI-2-MSI measurement region. All other cells are marked dark grey. The colors of cell masks correlates with the clusters found in (C).\u003c/p\u003e\n\u003cp\u003e(E) Zoom-in of the region annotated with lymphocyte accumulation (red squares in A, B and D).\u003c/p\u003e\n\u003cp\u003e(F) Zoom-in of the region annotated with intratumoral hemorrhage (blue squares in A, B and D).\u003c/p\u003e\n\u003cp\u003e(G) Data matrix tabulating the number of neutrophil subtypes (D) found in each tissue microenvironment (B). For each subtype, the percentage of DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells is calculated. In a neighborhood analysis for all neutrophils, the mean number of cells categorized as CD45\u003csup\u003e+\u003c/sup\u003e cells, Ly6G\u003csup\u003e+\u003c/sup\u003e cells and DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells found in a 15 µm radius are presented. Additionally, for each microenvironment, the total number of contained cells, CD45\u003csup\u003e+\u003c/sup\u003e, Ly6G\u003csup\u003e+\u003c/sup\u003e and DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e is noted.\u003c/p\u003e\n\u003cp\u003e(H) A heat map of standardized intensities of 63 ion species for the eight neutrophil subtypes visualizes differences in lipid profiles.\u0026nbsp;\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/1fd5854785f3cf8e0fc6781e.png"},{"id":93655969,"identity":"0e279ee2-42fd-4247-8657-a7625d39b121","added_by":"auto","created_at":"2025-10-16 07:06:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":26621482,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/de9d7ba7-a486-4dd8-bb52-6460b6d0de31.pdf"},{"id":85369249,"identity":"5348ebeb-ab45-46bf-b24e-85206288b0ea","added_by":"auto","created_at":"2025-06-25 07:17:08","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":14329,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDatalegends.docx","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/4cdc897da8a943132c1e84b5.docx"},{"id":85368586,"identity":"d5ab4233-0ea7-4cd6-a615-6e51e841cf7a","added_by":"auto","created_at":"2025-06-25 07:09:07","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":40389,"visible":true,"origin":"","legend":"Supporting Information","description":"","filename":"pothoffetalSI.docx","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/665ff6bcab2f96f953e54096.docx"},{"id":85368584,"identity":"fc0bc41b-614a-4dad-bb09-beff03231ddf","added_by":"auto","created_at":"2025-06-25 07:09:05","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":49782,"visible":true,"origin":"","legend":"Supporting Table 1","description":"","filename":"TableS1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/9293228939271cc9b0d4e769.xlsx"},{"id":85368596,"identity":"dbae8bdd-1da7-4317-a602-b03ebe7d5aaa","added_by":"auto","created_at":"2025-06-25 07:09:08","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":18651,"visible":true,"origin":"","legend":"Supporting Table 2","description":"","filename":"TableS2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/a5b39d5f7a7e4cbd6f79c3bf.xlsx"},{"id":85368571,"identity":"8aaaf8cd-9587-4bcf-8a65-d89670565349","added_by":"auto","created_at":"2025-06-25 07:09:02","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":3873253,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDataFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-6751055/v1/b2b342cc611f5a670199a5ea.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nThe authors declare the following competing interest(s): MN and JH are employees of Bruker Daltonik GmbH \u0026 Co KG (Bremen). KD, MN, AP, and JS have patents pending regarding parts of the hardware development (WO/2024/041681, WO/2025/067576).","formattedTitle":"Spatial biology using single-cell mass spectrometry imaging and integrated microscopy","fulltext":[{"header":"Introduction","content":"\u003cp\u003eTwo cells in a tissue or a cell culture are never exactly identical, even if they share the same genetic code and microenvironment. This heterogeneity can be a consequence of differences in age, ontogeny, or tissue imprinting resulting in alterations of morphology, gene expression, molecular composition, and ultimately function. In living systems, individual cells are part of interdependent multicellular structures, i.e. tissues, and undergo a constant exchange within local regulatory networks as part of the almost inestimable complexity that is life. Thus, to characterize the identity and state of a cell, the description should follow a holistic approach and include information from a wide range of analytical techniques \u003csup\u003e1\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn recent years, single-cell RNAseq and spectral flow cytometry have emerged as valuable tools to profile cellular heterogeneity. Similarly, advances in single-cell proteomics provide information on the heterogeneity on the protein level, albeit with much lower throughput \u003csup\u003e2,3\u003c/sup\u003e. Under the umbrella of spatial biology, the emergence of spatial-omics technologies including deep visual proteomics, multiplexed immunohistochemistry and spatial RNAseq have added information on morphology and the spatial relationship of cells \u003csup\u003e4\u0026ndash;7\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eA complementary addition to these techniques is the analysis by high spatial resolution mass spectrometry imaging (MSI). Importantly, this method can augment information on active transcripts and protein expression with the resulting downstream production of lipids and further metabolites, as well as glycans and peptides \u003csup\u003e8\u003c/sup\u003e. The most widely used MSI approach is based on matrix-assisted laser desorption ionization (MALDI). Herein, finely focused laser pulses are applied to pre-defined pixel positions on the sample surface, resulting in spatially defined mass spectra that can be processed to reveal signal intensity maps for up to hundreds of analytes differentiated based on their molecular weight. To increase signal intensity for a wide range of analyte classes the technique can be amended with laser postionization (MALDI-2) \u003csup\u003e9\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTypically, eukaryotic cells stretch over distances between 10 and 100 \u0026micro;m and display a wide variety in shape and form. Prominent sub-cellular structures like the nucleus or other large organelles, but also morphological features, show sizes between 3 and 10 \u0026micro;m. Their spatially resolved analysis therefore requires pixel sizes of no more than 1x1 \u0026micro;m\u0026sup2;. The current bench-mark figure on the pixel size of MALDI-MSI, however, is at about 5x5 \u0026micro;m\u0026sup2;. Less established experimental techniques dedicated to high resolving power, such as transmission-mode or inverse geometry (t-)MALDI-2-MSI and others, enables pixel sizes\u0026thinsp;\u0026le;\u0026thinsp;1x1 \u0026micro;m\u0026sup2; and can thus fulfill this criterion \u003csup\u003e10\u0026ndash;16\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eRecent application of MALDI-2 and t-MALDI-2-MSI for single-cell mass spectrometry has revealed a strong molecular heterogeneity for lipids and metabolites within clonal populations of cells and seemingly homogeneous tissue regions demonstrating its high potential for cell biology research \u003csup\u003e17\u0026ndash;20\u003c/sup\u003e. To contextualize these metabolic phenotypes with existing knowledge on the single-cell level, however, MALDI-MSI has to be connected to prevalent microscopy techniques conducted on the same slide. To date, the general application of this contextualization on the same slide is strongly limited by three main factors: Spatial resolving power of MALDI-MSI, a precise co-registration with optical modalities, and automation of data analysis. Especially in the context of tightly packed cellular systems, such as tissue, shortcomings of the first two factors lead to erroneous assignment of mass spectrometric information to individual cells and hamper the generation of unambiguous cell-specific molecular information, while the manual annotation of cells hampers the analysis of statistically relevant numbers of cells \u003csup\u003e17,18,21\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo address these challenges, we here present the development of hard- and software solutions on a prototype ion source coupled to an orthogonal time-of-flight (oTOF) mass analyzer. It utilizes transmission mode for ultra-high spatial resolution and is equipped with laser postionization (MALDI-2) to increase sensitivity for the analysis of lipids and metabolites \u003csup\u003e9,15,22,23\u003c/sup\u003e. While previous implementations of t-MALDI ion sources have only hinted at the potential to utilize the employed optical components for microscopy of the sample inside the ion source \u003csup\u003e15,24\u003c/sup\u003e, we here make full use of this option by including in-source bright field (BF) as well as fluorescence microscopy (FM).\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDescription and benchmarking of transmission MALDI-2-MSI with integrated scanning microscopy\u003c/h2\u003e \u003cp\u003eA full technical description of the optical setup and preparation protocols is presented in the methods. In short the ion source of a Bruker timsTOF-fleX MALDI-2 was amended for t-MALDI analysis. Dedicated lighting strategies enable the use of BF as well as FM inside the ion source \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eA,B\u003cb\u003e)\u003c/b\u003e. Importantly, laser delivery and in-source microscopy share essential components of the optical beam path and for stage movement. In the xy-plane (sample surface), the site of laser ablation is therefore directly linked to a specific position on the optical microscope image. This design feature inherently co-registers both modalities by utilizing the same coordinate system for in-source microscopy and MALDI-MSI \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eA\u003cb\u003e)\u003c/b\u003e. Next to providing highly resolved orientation to set up imaging runs, FM-images acquired inside the ion source constitute an essential link for precise co-registration to other optical modalities. In z-direction the inherent coupling of the focal plane of optical microscopy and laser delivery enables a fast and easy setup of experiments based on autofocusing routines.\u003c/p\u003e \u003cp\u003eTo enable FM analysis prior to MSI analysis, we have developed dedicated staining protocols optimized to omit or strongly reduce chemical alterations and spatial delocalization of the analyte content of interest.\u003c/p\u003e \u003cp\u003eThe newly developed methods and pipelines were benchmarked with regard to microscopic image quality and conservation of chemical and morphological integrity. For this, we conducted combined FM and t-MALDI-2-MSI on coronal cryo-sections of fresh-frozen mouse cerebellum. This tissue exhibits histological features across different lateral scales each presenting distinct lipid profiles. It has been intensely used as model system across a wide range of microscopic and MSI methods \u003csup\u003e25\u0026ndash;28\u003c/sup\u003e and is well characterized with regard to its lipidome based on the analysis of lipid extracts \u003csup\u003e29\u003c/sup\u003e. Details on sample preparation and data acquisition are described in the methods.\u003c/p\u003e \u003cp\u003eTo inspect and demonstrate the quality of our newly developed staining-based FM analysis on the same tissue section prior to t-MALDI-MSI analysis, we employed small molecule stains, targeting the nuclei and F-actin, as well as immunofluorescence (IF) to target calbindin expressed in Purkinje cell bodies \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eC\u003cb\u003e).\u003c/b\u003e Slide scanning FM recorded prior to matrix application in an external microscope (VS200, Evident) at 50X displays the expected structures and morphology for all three staining methods in high quality and corroborates the feasibility of the employed protocols \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eD\u003cb\u003e)\u003c/b\u003e.\u003c/p\u003e \u003cp\u003eAfter matrix application by resublimation, the sample was transferred into the MALDI ion source followed by in-source slide scanning FM (50X, DAPI channel). Utilizing the FM image of the DAPI channel generated inside the ion source as an intermediate enables a straightforward co-registration of external FM images with MALDI-MSI (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). In contrast, a conventional \u0026ldquo;cross-modality\u0026rdquo; co-registration of FM and MSI on a cellular or sub-cellular level is often hampered by difficulties in identifying pairs of fiducial markers across modalities that represent the exact same spatial location \u003csup\u003e30\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eSubsequently, t-MALDI-2-MSI data at a pixel size of 1x1 \u0026micro;m\u0026sup2; was acquired. Overview and zoomed in regions of the resulting MS images display distinct intensity distributions demonstrated on the example of ion signals of two different phosphatidylcholines (PC) ([PC(38:6)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e at \u003cem\u003em/z\u003c/em\u003e 806.57 and [PC(40:6)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e at \u003cem\u003em/z\u003c/em\u003e 834.60) and the molecular ion signal of the Hoechst stain of the nuclei at \u003cem\u003em/z\u003c/em\u003e 452.23 (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003eDirect comparison of t-MALDI-2-MSI with microscopy confirms that morphological features at the different lateral scales are largely retained during matrix application step and subsequent MSI analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). The overlay of both modalities reveals a strong correlation with histological features of the white matter and granular layer, discernable based on the F-actin staining and the nuclei. Similarly, signal intensity distribution of PC(40:6), associated with the Purkinje cell layer \u003csup\u003e31\u003c/sup\u003e shows good correlation with the IF-based specific staining of these structures (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). Overall, deviations for co-registration were found to be less than 1 \u0026micro;m (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eB-C and \u003cb\u003eSI notes 1\u003c/b\u003e).\u003c/p\u003e \u003cp\u003eTo explore the depth of information for lipid analysis available from pre-stained t-MALDI-2-MSI, mass spectral information was matched against a previously published full lipidomics profile for murine cerebellum derived from bulk \u003csup\u003e29\u003c/sup\u003e. This analysis reveals that ~\u0026thinsp;82% of the reported molecular lipid content of cerebellum are detected in the t-MALDI-2 measurement (\u003cb\u003eSI notes 2\u003c/b\u003e). This amounts to 66 annotated \u003cem\u003eion species\u003c/em\u003e that produce contrast rich intensity distributions in t-MALDI-2-MSI in positive and/or negative ion mode measurements (\u003cb\u003eExtended Data Table\u0026nbsp;1\u003c/b\u003e). Comparison of the detected signal intensities with the molar content of the different lipoforms of each of the detected lipid classes shows a good agreement with the literature \u003csup\u003e29\u003c/sup\u003e (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003e \u003cb\u003eand SI notes 2\u003c/b\u003e). Notably, this includes lipid classes, such as phosphatidylethanolamine (PE) and phosphatidylserine (PS) that may be affected by the chemical fixation used during sample preparation \u003csup\u003e32\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eLipid analysis at sub-cellular resolving power\u003c/h3\u003e\n\u003cp\u003eTo demonstrate the potential of histology guided t-MALDI-2-MSI for the analysis of sub-cellular structures, we investigated THP-1-derived macrophages. Cells were cultured and differentiated directly in chamber slides. Freshly differentiated cells were incubated with pHrodo red \u003cem\u003eE. coli\u003c/em\u003e bioparticles for 2 h. Upon phagocytosis, these particles become fluorescent and pinpoint acidified phagolysosomes inside the cells. For analysis, cells were briefly fixed and their actin filament and nuclei stained following MALDI compatible protocols previously reported \u003csup\u003e17\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eSlide scanning fluorescence microscopy on an external slide scanner prior to coating with MALDI matrix reveals heterogeneously pronounced phagocytosis with some cells showing a large number of internalized bioparticles (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003eAfter coating with MALDI matrix using resublimation, in-source FM (50X) and t-MALDI-2-MSI data at 1.5x1.5 \u0026micro;m\u0026sup2; pixel size was collected for a few hundred cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Image co-registration between the externally generated FM image and MSI data was performed by using the in-source FM data as an intermediate (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). The overlay of both modalities corroborates a high-fidelity co-registration on the level of the employed MALDI pixel size. Areas visualized by the pHrodo red stain in FM accurately co-align with the distribution of specific lipid signal intensities in MSI revealing little to no leakage or diffusion of lipid content or other preparation artifacts that may compromise (sub-)cellular integrity (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eC, D) \u003csup\u003e22\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo investigate the molecular distributions specifically related to phagolysosomes, we used subcellular spatial information provided by the fluorescence signal produced by the uptake of pHrodo bioparticles. Utilizing the high-fidelty co-registration, we overlayed this histological information to t-MALDI-2-MSI data and identified those pixels that correlate with phagolysosomes to generate phagolysosome-specific mass spectra. (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eC,D). Lipid annotation was based on previously published results for THP-1 derived macrophages and \u003cem\u003eE. coli\u003c/em\u003e, respectively \u003csup\u003e20,33\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eComparative analysis of the resulting phagolysosome-specific molecular information revealed an increase of signal intensity for numerous lipid species (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). Confirmed by the respective MS images, lipids such as PE(33:1) and PE(34:2) spatially correlate with the phagolysosomes identified by fluorescence \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eC,D). Both PE species are abundantly expressed in the membrane of \u003cem\u003eE. coli cells\u003c/em\u003e \u003csup\u003e33\u003c/sup\u003e. Notably, PE(33:1) is not detected in non-phagocytosing THP-1 cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eB-D, red trace). Its presence, confirmed by additional on-cell MS/MS experiments (described in Methods), provides a direct link to the phagocytized pathogen material and corroborates the highly accurate co-registration of both modalities.\u003c/p\u003e \u003cp\u003eOther tentatively assigned lipid species linked to the phagolysosomes can be attributed to degradation of \u003cem\u003eE. coli\u003c/em\u003e-derived phospholipids during phagocytosis \u003csup\u003e34,35\u003c/sup\u003e. These degradation products are most likely the result of enzymatic activity and reactions with reactive oxygen species inside the phagolysosomes. They include lipid metabolites such as the diacylglycerol DG(33:1), that can be directly linked to the enzymatic cleavage of PE(33:1) and lipid species tentatively identified as lysophosphatidylglycerols (LPG). The latter can be linked to the enzymatic breakdown of phosphatidylglycerols (PG) as well as cardiolipins (CL) both highly expressed in the membrane of \u003cem\u003eE.coli\u003c/em\u003e \u003csup\u003e33,35\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e2\u003c/span\u003eE).\u003c/p\u003e\n\u003ch3\u003eAutomated single-cell analysis of lipids in their histological context in tumor tissue\u003c/h3\u003e\n\u003cp\u003eIn a second exemplary study, we applied our methods to the single-cell molecular analysis of tumor tissue. This tissue type typically contains a large number of different cell types in a wide range of states translating to a wide variety of metabolic microenvironments (tumor microenvironment, TME) of different size and shape and a highly heterogeneous lipid profile across the tissue \u003csup\u003e36\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe employed 4T1 mouse mammary carcinoma model is highly malignant with locally aggressive growth and exhibits metastasis to regional lymph nodes and distant organs, particularly lung, liver, and bone \u003csup\u003e37,38\u003c/sup\u003e. It has diverse inflammatory TMEs with a large number of infiltrating immune cells such as T-cells, macrophages, neutrophils and natural killer cells present \u003csup\u003e39,40\u003c/sup\u003e. The site and cell-specific molecular profile of these immune cells is of great importance for their fate and function and is believed to be in constant interdependence with their microenvironment \u003csup\u003e41\u003c/sup\u003e. Consequently, the investigation of this crosstalk not only requires a molecular analysis at the cellular level but also an assessment of the immediate vicinity.\u003c/p\u003e \u003cp\u003eFor analysis, a heterogeneous area of interest that includes a broad range of TMEs was selected based on H\u0026amp;E staining and histological annotation (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). A consecutive section was prepared using MALDI-compatible staining that targets cell nuclei, F-actin, and three IF stains that selectively target immune cells, (CD45), neutrophils (Ly6G), and T3 neutrophils (DcTRAIL-R1) \u003csup\u003e41,42\u003c/sup\u003e. All channels were recorded on the complete section using external slide scanning FM at 50X before matrix application (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Co-registration with t-MALDI-2-MSI was based on the DAPI channel recorded inside the ion source after matrix application as an intermediate (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eE). After microscopy, t-MALDI-2-MSI data was recorded in seven separate measurements across the region of interest at a pixel size of 1x1 \u0026micro;m\u0026sup2; resulting in a total of approx. 7 Mpixels. For data analysis, we selected 63 molecular ion species annotated to individual lipid species based on exact mass (\u0026lt;\u0026thinsp;3 ppm) (see \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003e for selected images and \u003cb\u003eTable \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e).\u003c/p\u003e \u003cp\u003eSpatial distribution for a number of glycero- and glycerophospholipids such as [PC(34:1)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e at \u003cem\u003em/z\u003c/em\u003e 760.58 reveal a sizeable heterogeneity across the tumor sample (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eC, \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Signal intensity of the ether lipid [PC(O-32:0)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e,detected at \u003cem\u003em/z\u003c/em\u003e 720.59, correlates with the neutrophil-specific Ly6G-stain (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eD-F). This is consistent with a significant overexpression of this lipid species in mouse neutrophils reported previously \u003csup\u003e43,44\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOther specific signal intensity distributions also directly correlate with structures visible in the microscopy data. As expected \u003csup\u003e45\u003c/sup\u003e, signal intensity for triacylglycerides (TG) such as [TG(52:5)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e correlates with the adipose tissue (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). Acylcarnitine, detected at \u003cem\u003em/z\u003c/em\u003e 428.38 as [ACAR(18:0)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e is an important intermediate in lipid metabolism and a reported marker for hypoxia \u003csup\u003e46\u003c/sup\u003e. Its signal intensity distribution correlates with the detection of tumor-associated T3 neutrophils identified by the DcTRAIL-R1 staining that have been described to be found in hypoxic areas \u003csup\u003e41,42\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eE). Heme B, detected at \u003cem\u003em/z\u003c/em\u003e 616.18, is directly associated with erythrocytes and, hence, serves as marker for hemorrhage or blood vessels \u003csup\u003e47\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eF).\u003c/p\u003e \u003cp\u003eWhile general assignment of lipid profiles to TMEs has been reported previously, we here demonstrate the analysis of single cells within their tissue environment in statistically relevant numbers. For this we adapted our previously reported workflow for the analysis of single cells in culture to the analysis of tissue sections \u003csup\u003e17,20\u003c/sup\u003e. Robust and reliable cell segmentation, necessary for automated analysis \u003csup\u003e21\u003c/sup\u003e, was provided based on the external fluorescence microscopy images using Mesmer by DeepCell (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eG) \u003csup\u003e48\u003c/sup\u003e. This resulted in segmentation masks for each cell and enabled the calculation of a variety of morphometric parameters using CellProfiler \u003csup\u003e49\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo produce single-cell mass spectra, cell segmentation masks were projected on the t-MALDI-2-MSI data, again taking advantage of the inherent link between in-source FM and the acquired MSI data (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e1\u003c/span\u003eE). An overlay of these projected masks with the lipid neutrophil marker [PC(O-32:0)\u0026thinsp;+\u0026thinsp;H]\u003csup\u003e+\u003c/sup\u003e and comparison to the specific Ly6G staining demonstrates the high-fidelity co-registration and corroborates the ability of our combined method to deduct single-cell mass spectra (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003eG).\u003c/p\u003e \u003cp\u003eSingle-cell molecular data is generated for all 63 annotated lipid species by summing the respective signal intensities over all pixels allocated to a specific cell using the respective mask. Subsequently all values are normalized to the sum of all signal intensities for the respective cell.\u003c/p\u003e \u003cp\u003eCell segmentation and subsequent data analysis resulted in combined morphometric and molecular data for 63,408 individual cells. Notably this analysis retains the histological context of all cells and includes parameters describing location within the tissue and fluorescence intensity for all employed channels. This enables the identification of CD45\u003csup\u003e+\u003c/sup\u003e, Ly6G\u003csup\u003e+\u003c/sup\u003e, and DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells based on intensity thresholding for each specific channel. For visualization, this information can be projected onto the segmentation canvas, thereby identifying the location and local density of different sub-groups of immune cells in the tissue section (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003eProcessing of the combined IF and MSI single-cell data enabled spatial biology analysis on the level of lipids and metabolites, by classifying cells based on molecular similarity. In this context, analysis on the single-cell basis is especially important for scarce or isolated cells like specific immune cells because their molecular profile may substantially differ from the surrounding TME. This type of analysis is demonstrated on the example of infiltrating neutrophils inside the 4T1 tumor section.\u003c/p\u003e \u003cp\u003eTo classify neutrophils based on their molecular profile independent of their microenvironment and vice versa, single-cell data was split into two groups based on Ly6G staining. The resulting subgroups contained 3,410 Ly6G\u003csup\u003e+\u003c/sup\u003e neutrophils and 59,998 Ly6G\u003csup\u003e\u0026minus;\u003c/sup\u003e cells, respectively. Both groups were independently analyzed using Uniform Manifold Approximation and Projection (UMAP) and appropriate clustering algorithms.\u003c/p\u003e \u003cp\u003eTo identify molecularly similar cell types within the tissue, all Ly6G\u003csup\u003e\u0026minus;\u003c/sup\u003e cells were subjected to k-means clustering, resulting in eight discernable clusters. By assigning a specific color to each cluster, the outlines of each cell can be filled in with the respective color and projected back onto their original location (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003eThe resulting map of classified cells displays clearly discernable tissue micro-environments where molecularly similar cells cluster in specific areas of the tissue. Clusters IV, V, and VIII contain cells of the skin, the adipose tissue and the wider tumor periphery, respectively. Inside the viable tumor tissue, a clear separation into TMEs reveals a layer-like structure with frayed edges (clusters II, III, and VI). Cluster VII contains cells directly adjacent to the tumor edges and away from the adipose tissue but also cells inside the tumor close to its necrotic core. Cluster I contains cells allocated to hypoxic regions of the tumor.\u003c/p\u003e \u003cp\u003eThe direct connection to the morphometric data allows to characterize each cluster with regard to immune infiltration. Figure\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eG, summarizes the total amount of cells as well as the numbers of CD45\u003csup\u003e+\u003c/sup\u003e, Ly6G\u003csup\u003e+\u003c/sup\u003e, and DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells that are found within the boundaries of the respective cluster.\u003c/p\u003e \u003cp\u003eNext to TMEs, we investigated the molecular heterogeneity of neutrophils. For this, the Ly6G\u003csup\u003e+\u003c/sup\u003e cells were analyzed using a UMAP and classified using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). This results in a clear separation of eight clusters. To provide linguistic distinction from clusters described for the tissue microenvironments, results for neutrophils will be referred to as subtypes. Colored respectively and projected onto the cell map, these different subtypes can be found in distinct areas of the tumor (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003eApparent from the comparison of the images depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003e, some subtypes show a clear spatial correlation with a specific microenvironment and histological features \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eG, \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eA\u003cb\u003e)\u003c/b\u003e. Subtype 6 is almost exclusively located in the skin. Neutrophils are found in skin in comparably high number and have important and specific barrier functions \u003csup\u003e50\u003c/sup\u003e. Our data suggests a significantly altered lipid profile for these neutrophils. For example, in contrast to all other subtypes, subtype 6 shows a low expression of PC(O-32:0) (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eB,C).\u003c/p\u003e \u003cp\u003eNeutrophils of subtypes 5 and 7 are located in specific tissue microenvironments that correlate with the adipose tissue (clusters IV and V) in the periphery of the tumor. Their lipid profiles show an increase of TG and DG lipid species (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eH). Subtype 4 is predominantly found in non-adipose tissue close to the tumor (cluster VII). Similarly, subtype 2 also populates this peripheral region but is additionally found inside the tumor close to its necrotic core. Lipid profiles for both subtypes are low in DG and TG. While their overall cell density in this area is relatively high, they do not form agglomerates.\u003c/p\u003e \u003cp\u003eInterestingly, neutrophils in this periphery of the tumor show strong molecular similarities to their respective microenvironment (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). This may be explained by an uptake of extracellular vesicles or cell debris produced by the surrounding tissue and \u0026ldquo;recycling\u0026rdquo; of the lipid content by the neutrophils. Notably, the expression of lipid species that require specific enzymatic machinery for their synthesis such as PC ether lipids (PC-O) do not show similarities to the surrounding tissue but are upregulated only in the respective neutrophil populations (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). The exclusive co-localization of these ion species to Ly6G\u003csup\u003e+\u003c/sup\u003e-cells also precludes the involvement of measurement artefacts, such as analyte diffusion into this convergence of lipid profiles between neutrophils and their respective TME (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWith the exception of subtype 4 all neutrophils found in the tumor periphery and the adipose tissue show a relatively high density of immune cells in their direct vicinity (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). This observation is corroborated by a neighborhood analysis that counts the average number of immune cells within a distance of 15 \u0026micro;m to the cell edge of each individual neutrophil (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eG).\u003c/p\u003e \u003cp\u003eInside the tumor, neutrophils of the subtype 1 are found mostly in the viable tumor tissue (clusters II, III, and V) (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eG and \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Notably, their lipid profile shows little similarity to their respective surrounding TMEs (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). This may indicate a decreased up-take and turnover of lipids or a reduced residence time within the tumor tissue. While the overall density of subtype 1 neutrophils in the tumor increases towards the adjacent adipose tissue, they are mostly isolated with relatively few other immune cells around them (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eG).\u003c/p\u003e \u003cp\u003eIn contrast, neutrophils of subtypes 3 and 8 form strong agglomerations (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eE). These subtypes are located within TME I with subtype 8 being confined to the center of the region. Their neighborhood analysis results in high numbers of overall immune cells in their direct vicinity (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eG). About 90% of subtype 8 neutrophils are identified as DcTRAIL\u003csup\u003e+\u003c/sup\u003e. The lipid profile of this subtype strongly differs from that of the surrounding hypoxic TME I (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). It presents an increased signal for ACAR 18:0 and an increase in signal intensities for lipid metabolites such as LPCs (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e4\u003c/span\u003eH) and DG, while TGs are mostly absent. This may indicate an upregulation of specific lipid metabolism in this subtype.\u003c/p\u003e \u003cp\u003eSimilar to subtype 8, neutrophils of the subtype 3 are almost exclusively found in agglomerates inside TME I but are located further away from its center. In contrast to subtype 8, however, only 17% are DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e. The lipid profile of subtype 3 also contains high signal intensities for ACAR 18:0 but reduced levels of LPC and DG. Again, in contrast to subtype 8, the lipid profile of subtype 3 neutrophils shows great similarities to its surrounding TME I. This increased molecular crosstalk may indicate a longer residence time of neutrophils within this hypoxic TME as opposed to the viable TMEs II, III, and VI.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn its own right, MSI analysis can provide specific molecular information of single cells, thereby delivering an important puzzle piece towards the understanding of individual cellular behavior. For a more holistic approach, however, the resulting lipidomic or metabolomic information has to be put into the context of the cell type, differentiation status and state but also has to regard its current tissue macro- and microenvironment \u003csup\u003e1\u003c/sup\u003e. Here, a tight connection to optical microscopy analysis has long since been identified as a key feature in the integration of MALDI-MSI into the \u0026ldquo;biomedical toolbox\u0026rdquo; \u003csup\u003e51\u003c/sup\u003e. Conventionally, microscopy of the same sample prior to MALDI-MSI is restricted to the limited analytical capabilities offered by optical microscopy of unstained tissue, e.g. by autofluorescence or phase-contrast microscopy \u003csup\u003e52,53\u003c/sup\u003e. Staining procedures applied after application and removal of the matrix and especially laser irradiation that precedes microscopy, can damage or distort the tissue and impede binding properties for antibody based analysis \u003csup\u003e12\u003c/sup\u003e. High quality microscopy analysis, including IF, is therefore typically carried out on adjacent tissue sections introducing well-known challenges in accuracy of co-registration. In particular, the use of adjacent sections for microscopy strongly limits the analysis of specifically targeted single cells \u003csup\u003e30,51,54,55\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn contrast, the presented technique includes slide scanning microscopy outside and inside the ion source of the same sample recorded prior to ultra-high resolving power t-MALDI-2-MSI. Enabled by dedicated sample preparation techniques, this combination lays the groundwork for the large scale, automated, and reliable generation of information-rich highly resolved data for both modalities on the same sample.\u003c/p\u003e \u003cp\u003eTwo exploratory studies highlight the advantages of the described method for the molecular analysis with cellular and even sub-cellular resolution. Using the phagocytosis of \u003cem\u003eE. coli\u003c/em\u003e bioparticles by macrophages as an example, we demonstrate how the technique can be used to investigate metabolic processes inside sub-cellular compartments like the phagolysosomes. Assignment of bacterial PE 33:1 and PE 34:2 confirmed the uptake of \u003cem\u003eE.coli\u003c/em\u003e bioparticles. The additional identification of increased amounts of the respective DG species suggests their partial degradation. In addition, the tentative identification of several LPG species provides further useful leads to the metabolic breakdown of \u003cem\u003eE. coli\u003c/em\u003e derived CL and PG by enzymatic degradation and/or oxidation of lipids inside the phagolysosome \u003csup\u003e56\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOverall, this study illustrates a new approach to analyze molecular profiles of subcellular compartments. In future applications, this largely untargeted technique may be used to investigate and monitor metabolic turnover on the organelle level of specific single cells.\u003c/p\u003e \u003cp\u003eIn our study of the 4T1 tumor model, we demonstrate spatial biology analysis on the lipid level for highly heterogeneous tissue. While previous work on single-cell MSI from tissue sections has been limited to the manual annotation of small numbers of cells \u003csup\u003e17\u0026ndash;19,57\u003c/sup\u003e, automated segmentation based on slide scanning (immuno-)fluorescence microscopy resulted in the identification of more than 63K cells and allowed for the identification and classification of common immune cells (CD45\u003csup\u003e+\u003c/sup\u003e, ~ 23.7K cells; 38% of all cells), relatively rare neutrophils (Ly6G\u003csup\u003e+\u003c/sup\u003e, ~ 3.4K cells; 5.4%), and very rare tumor associated T3-neutrophils (DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e, 276 cells; 0.4%) within the tissue. Importantly, the segmentation, in combination with highly accurate co-registration of the microscopy with t-MALDI-2-MSI results, allowed for the generation of single-cell mass spectra for each cell.\u003c/p\u003e \u003cp\u003eUsing statistical analysis, cells were clustered based on their lipid profile, revealing eight distinct tissue microenvironments inside the tumor and around its periphery. Likewise, an independent clustering of just the Ly6G\u003csup\u003e+\u003c/sup\u003e cells revealed eight subtypes of neutrophils inside the investigated tissue region. Comparison of lipid profiles between neutrophils and their direct tissue environment revealed striking similarities in some cases. These results allow for a new perspective on the uptake and recycling of lipids by specific neutrophil subtypes. They also reveal distinctive differences between neutrophils and their direct environment such as the increased levels of ether lipid species. In addition to a general upregulation of this lipid species also described in the literature \u003csup\u003e44\u003c/sup\u003e we are able to identify subtype and TME specific differences in its expression. Neighborhood analysis of all investigated immune cells in the direct vicinity of the neutrophils provides a first hint at their swarming behavior. While some of the subtypes represent relatively isolated cells, others are found in agglomeration with neutrophils and other immune cells.\u003c/p\u003e \u003cp\u003eThe strength of the molecular investigation at cellular level becomes especially apparent in the examination of the hypoxic regions of the tumor. This TME shows a signal intensity increase for ACAR 18:0, as well as a high density of immune cells both described as markers for hypoxia \u003csup\u003e46\u003c/sup\u003e. The inner core of this TME is populated by densely packed T3-neutrophils (most of them of subtype 8), concurring with the literature \u003csup\u003e41,42\u003c/sup\u003e. A second group of agglomerated mostly DcTRAIL-R1\u003csup\u003e\u0026minus;\u003c/sup\u003e neutrophils (subtype 3) is also located inside the same hypoxic area but further away from its center. The subtyping now enables an in-depth analysis of underlying lipid profiles and integration with the state of research. While both subtypes, for example, express increased levels of ACAR 18:0, only subtype 8 (mainly DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells) shows increased levels of fully saturated LPCs. These mediator are known to be important in metastasis formation and in the recruitment of phagocytes during apoptosis \u003csup\u003e58,59\u003c/sup\u003e and our data points to a previously unknown tentative connection with T3-neutrophils.\u003c/p\u003e \u003cp\u003eWhile a full investigation of the underlying biological mechanisms is not within the focus of this publication, the presented proof-of-concept studies showcase the great potential of the technique. The seamless integration of morphometric data with (sub-)cellular lipid and metabolite analysis significantly expands the informational space available for each individual cell inside the tissue and enables a direct contextualization with its tissue microenvironment \u003cem\u003ein situ\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eOverall, the presented techniques constitute a versatile set of tools for an improved integration of single-cell MSI into existing workflows. Embedded within a full suite of spatial biology methods, in future applications, it will allow for the combination of IF-based information that may describe the regulation of specific genes and the expression of individual proteins with the expression of lipids or metabolites in the same cell and its direct surrounding \u003csup\u003e60\u003c/sup\u003e. This addition of \u0026ldquo;spatial biology of lipids and metabolites\u0026rdquo; on the single-cell level provides an important additional layer of information to complete pathway analysis from the transcript and proteome level down to the regulation of the molecular products. In can be a valuable asset to review and validate downstream effects of genetic or transcriptomic variation as well as protein inhibition but also help to investigate regulatory functions of small molecules on gene and protein expression.\u003c/p\u003e \u003cp\u003eIn this, the methods described here have great potential to help entangling the influence of endogenous and exogenous factors such as immune cell infiltration, infection, or drug delivery on cellular behavior in a specially resolved manner. They are, however, by no means limited to these applications but can be applied to decipher open biological questions in a wide range of tissue and cell types.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cem\u003eReagents, resources and software\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll reagents, resources, and software used are listed in the table of resources (\u003cstrong\u003eTable S1\u003c/strong\u003e)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCell culture preparation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTHP-1 ACC-16 cells (DSMZ) were cultured in T25 cell culture flasks at 37 \u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e in RPMI 1640 Medium (Lonza), which was supplemented with 2 mM L-glutamine, 10% fetal bovine serum and 1 mM sodium pyruvate. About 10,000 cells per chamber were sown in Millicell EZ Slide 8-well chamber slides (Sigma-Aldrich). Differentiation to M0 macrophages was induced by exposure to 20 ng/ml phorbol 12-myristate 13-acetate (PMA, Sigma-Aldrich) for 24 h followed by a 3-day rest phase in PMA-free medium. Two hours before harvest, pHrodo\u0026trade; Red E. coli BioParticles (Thermo Fisher) were added to the culture medium resulting in a concentration of 95 \u0026micro;g/mL. Cells were washed with PBS and fixed for 5 min using 4% formaldehyde (low-methanol, Roth). F-Actin and nuclei were stained according to the previously described protocol by Bien \u003cem\u003eet al\u003c/em\u003e. using 1X CellMask Green Actin Tracking Stain (Thermo Fisher) diluted in PBS for 30 min and 1 \u0026micro;g/mL Hoechst 33342 (Sigma-Aldrich) dissolved in PBS for 5 min \u003csup\u003e1\u003c/sup\u003e. Cells were washed with PBS after every staining step. After staining, cells were washed using 150 mM ammonium acetate (Sigma-Aldrich) and dried at room temperature for 2 h.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMouse model\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll animal experiments were carried out in accordance with local animal welfare guidelines approved by the responsible authorities (North Rhine-Westphalia State Agency for Nature, Environment and Consumer Protection, LANUV, Protocol No. 81-02.04.2018.A010). Female BALB/c mice (Charles River Laboratories) were used at the age of 8 to 12 weeks. Mice were housed under a 12-hour light/dark cycle with ad libitum access to food and water. For tumor implantation, 10\u003csup\u003e6\u003c/sup\u003e 4T1 cells were resuspended in 25 \u0026micro;l of DMEM cell culture medium (Thermo Fisher) and then injected orthotopically into the lower left mammary fat pad of the mice. Tumors were allowed to grow for 9 days and tumor size was monitored daily with a digital caliper. On day 9 after tumor implantation, the animals were sacrificed. Brain tissue was harvested from untreated mice used as control group for the mentioned study.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTissue sectioning\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eMouse brain and tumor tissue were embedded in Epredia M-1 Embedding Matrix (Thermo Fisher) and snap-frozen in liquid nitrogen. 10 \u0026micro;m (cerebellum) or 8 \u0026micro;m (tumor) thick tissue sections were produced with a cryostat (CM 3050 S, Leica Biosystems) at -20 \u0026deg;C. Sections were thaw-mounted onto SuperFrost glass slides (Thermo Fisher) for H\u0026amp;E-staining or IntelliSlides (Bruker) for t-MALDI-2 MSI measurements. Samples for t-MALDI-2 MSI were vacuum-sealed and stored at -80 \u0026deg;C before further use.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTissue fixation and fluorescence staining\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFrozen tissue sections were slowly brought to room temperature in the vacuum-sealed package. After opening the seal, sections were further dried under a gentle nitrogen stream for 10 min. With a Super PAP Pen (Science Services) a contour was drawn around each section, followed by another 10 min drying under a gentle nitrogen stream, to create a hydrophobic barrier and thus enabling pipetting aqueous solutions on and off the sample. The following washing, fixation and staining steps were all performed by pipetting ~100 \u0026micro;l of the accordant solution onto the sample and removing it after the specified time. Slides with tissue sections were placed on a metal plate standing in ice water. Samples were quickly fixed in 4 % formaldehyde (low-methanol, Sigma-Aldrich) in PBS for 5 minutes and subsequently washed with PBS for 30 s. After tissue fixation, the primary antibody, Ly6g (ab307167, Abcam), DcTRAIL-R1 (130-110-873, Miltenyi Biotec), CD45 (130-110-798, Miltenyi Biotec) or Anti-Calbindin (ab229915, Abcam), diluted 1:500 (v/v) in 1% BSA (Serva) in PBS was applied to the sample and incubated for 60 min. The sample was then washed 3 times for 30 s in PBS. In case of a dual antibody staining, the secondary antibody, Alexa Fluor 594 (ab307167, Abcam), diluted 1:1000 (v/v) in 1% BSA in PBS was applied and incubated for 60 minutes, followed by 3 times 30 s washing steps in PBS. 5 \u0026micro;g/ml Hoechst33342 (Sigma-Aldrich) and 1X CellMask Green Actin Tracking Stain (Thermo Fisher) diluted in PBS were applied and incubated for 5 minutes. After staining, the slide was lifted from the metal plate and parts of the hydrophobic barrier were removed at the bottom of the slide with a pipette tip. Samples were then washed with 2 ml of ammonium acetate (Sigma-Aldrich) each, by letting it flow carefully over the section. For final drying, samples were placed under a gentle N\u003csub\u003e2\u003c/sub\u003e stream for 20 minutes.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTissue H\u0026amp;E staining\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eHematoxylin and eosin (H\u0026amp;E) stains of tissue sections were performed on untreated consecutive sections directly after sectioning. Sections were stained for 30 s with hematoxylin (Sigma-Aldrich) with subsequent bluing in tap water for 2 min. Afterwards, the tissue was stained for 30 s with eosin (Sigma-Aldrich) and then rinsed with water.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eExternal microscopy\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFluorescence images of cell cultures and tissue sections were recorded at 50X using the DAPI, FITC, TxRed, Cy3 and Cy5 channels of a slide scanning microscope (VS200, Evident) equipped with a SpectraSplit 7 filterset (Kromnigon). For each channel individually, the exposure time was set by performing the auto exposure procedure of the microscope on a region with high signal intensity for the respective fluorophore. This procedure was conducted prior to the imaging run, and the entire slide was subsequently imaged with a constant exposure time, with a maximum exposure of 500 ms. H\u0026amp;E-stained tissues were recorded at 20X in bright field mode. Image processing was performed with OlyVIA (Evident).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTransmission MALDI optics\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA schematic of the optical set-up used for material ejection for the t-MALDI-2-MSI analysis and in-source brightfield and fluorescence microscopy image generation is depicted in \u003cstrong\u003eFig. 1A-C\u003c/strong\u003e of the main article. The sample is mounted on a three axis piezo stage (SmarAct) and is illuminated homogeneously by 15 green LEDs (LED528EHP, Thorlabs) mounted on top of a 50X objective (M Plan Apo NUV HR 50X, Mitutoyo) with a custom-designed 3D printed LED ring. Light from the sample is collected by the objective and focused by a tube lens (TTL100-A, Thorlabs) onto the sensor of a digital camera (XCG-CG160, Sony). A simple top-view observation of the sample is in addition possible by the pre-installed camera of the prototype timsTOF fleX MALDI-2 mass spectrometer.\u003c/p\u003e\n\u003cp\u003eThe beam path of the MALDI laser (smartbeam, wavelength 355 nm, pulse duration 7 ns) of the timsTOF fleX MALDI-2 mass spectrometer is re-routed and condensed through a 6:1 telescope optic and co-aligned to the microscopy path with a dichroic mirror (L-07246, Laseroptik Garbsen). For recording fluorescence microscopy images, a filter cube (DFM2/M, Thorlabs) with a filter set (DFM2T1, Thorlabs) for the suitable fluorescence channel (MDF-BFP, DAPI; MDF-FITC, FITC; MDF-MCHC, mCherry; all Thorlabs) consisting of excitation and emission filters and a dichroic mirror is mounted in front of the tube lens. In fluorescence mode, the LED ring is turned off and fluorescence is excited by the respective fiber-coupled LED (M385FP1, DAPI; M455F3, FITC; MINTF4, mCherry; all Thorlabs) mounted with a collimating optic directly to the filter cube. The camera is integrated in the software by using the XC-SDK 2018 (Sony). The trigger signal for image acquisition of the camera and fluorescence excitation is provided as a \u0026ldquo;position reached\u0026rdquo; event by the controller of the piezo stage and is handled by a digital delay pulse generator (9200 Sapphire Series, Quantum Composers). With this pulse generator, the fluorescence excitation time was set from 10 ms to 2 s per image depending on the sample and fluorescence channel.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMALDI-matrix coating by resublimation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTo achieve a reproducible and homogeneous coating of the sample slides with MALDI-matrix, a custom-designed resublimation chamber was developed. A schematic of this setup is shown in \u003cstrong\u003eExtended Data Fig. 5A-B\u003c/strong\u003e. Under atmospheric pressure conditions, the hot-side is preheated to 80 \u0026deg;C and 0.5 ml of 10 mg/ml CHCA MALDI-matrix (Sigma-Aldrich) in acetone solution is pipetted into the matrix-reservoir. With the acetone evaporating rapidly, the MALDI-matrix is left forming a uniform layer on the bottom of the matrix reservoir. The sample is then clamped onto the cold-side at room temperature and the whole device is sealed and evacuated. After reaching a final pressure of 10\u003csup\u003e-2\u003c/sup\u003e mbar, the cold-side is first cooled to -18 \u0026deg;C, then the hot side is heated to 150 \u0026deg;C to start the sublimation and resublimation. After 60 min, the heating of the hot-side is turned off and the cold-side is heated to room temperature before venting the system to prevent condensation of ambient humidity on the sample. The sample is then put onto a room temperature metal block in a saturated atmosphere of 0.5% ethanol in water at 70 \u0026deg;C for 1.5 min for recrystallization. To demonstrate differences in contrast for in-source BF microscopy induced by the choice of matrix, cerebellum samples were coated with DHB and HABA MALDI-matrix (both Sigma-Aldrich) using the same settings as described abovefor CHCA\u003cstrong\u003e\u0026nbsp;(Extended Data Fig. 5C)\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAutofocus of optics\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe MALDI laser and microscope camera utilize the same objective for focusing and observation in the transmission-MALDI setup employed in this work. A schematic of the setup is depicted in \u003cstrong\u003eFig. 1A\u003c/strong\u003e in the main article. The focal plane of the MALDI-laser is adjusted to match that of the camera by fine adjusting the second telescope lens of the laser beam path. After aligning the focal planes, all optics are kept at a constant position. The objective (M Plan Apo NUV HR 50X, Mitutoyo) has a small focal depth of 1.4 \u0026micro;m and samples have to be in focus during optical microscopy and MALDI-MSI. This can be achieved by methodically adjusting the z-position of the sample with regard to the objective while recording sequential optical microscope images at z-increments of 1 \u0026micro;m. For each of these images a Laplacian filter is applied and the variance of the result is calculated, which serves as a measure of image sharpness. The stage is then moved to the z-position corresponding the maximum sharpness image and a fine-scan with 0.1 \u0026micro;m z-distance is performed in the range of +- 1 \u0026micro;m in the same way as before to find the z-position of highest sharpness at a submicron precision. Since the focal planes of the optical microscope and the MALDI laser are aligned, this optical focusing ensures an optimal focus of the MALDI-laser on the sample. This autofocus routine can be executed both BF and FM mode. The python code of this program AutoFocusClass.py is provided in Data S1 in the supplementary information.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eIn-source microscopy workflow\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe scale factor to convert pixel on the camera sensor to position of the stage was calculated by using a 1951 USAF Resolution Test Target (Thorlabs). For selection of a region of interest (ROI) for an optical microscope scan in the ion source, autofocusing points are placed manually on the sample. The ROI is defined as the smallest enveloping rectangle that contains all points. The stage is then moved to each of the autofocusing points and an autofocus routine as described in \u0026ldquo;Autofocus of optics\u0026rdquo; is performed, which yields the focal z-offset of that xy-coordinate. Using a radial basis function interpolation, the z-offset is calculated for every coordinate in a 10 \u0026micro;m x 10 \u0026micro;m mesh inside the ROI and saved to a z-interpolation file. After creating the z-interpolation, the optical microscope image of the ROI is taken. For this an up to 100 \u0026micro;m x 100 \u0026micro;m wide x/y-position mesh is created in the ROI. An optical microscope image is then taken at each mesh point at the z-offset of the closest point to the xy-position from the z-interpolation file. To create a complete image of the ROI, an empty canvas of the correspondent size is created. Using the scale factor of the setup, images are cropped from the edges to a size of 100 \u0026micro;m x 100 \u0026micro;m and are placed onto the empty canvas using the center coordinates from the xy-position mesh. After creating a z-interpolation file for a ROI once, multiple modalities, such as brightfield or different fluorescence channels can be recorded without creating a new interpolation. ImageJ was used to create overlay images of different modalities. The python code of this workflow can be found in the files RecordPositions.py, ZforRecordPositions.py, Scanning.py and stitching.py in Data S1 in the supplementary information.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTransmission-MALDI-2 MSI\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe MALDI-MSI run is setup partly using the proprietary flexImaging software (Bruker). Instead of using an external scan of the sample and teaching that to the stage coordinate system, the (multimodal) image of the ROI (see \u0026ldquo;In-source microscopy workflow\u0026rdquo;) is used. In the classic flexImaging workflow, three teaching points would be manually selected on the optical image. The corresponding stage coordinates would then be recorded by driving the sample-stage manually to the respective positions. Since center coordinates and pixel positions on the ROI image are already established from recording the ROI image, three x/y-positions from the 100 \u0026micro;m x 100 \u0026micro;m mesh (top left, bottom left, bottom right) and their corresponding stage coordinates are automatically deployed as teaching points. With these points, the setup file for the t-MALDI-2 MSI run can be automatically created, defeating any positioning errors introduced by manually selecting teaching points in the classic workflow. When the MALDI-laser is not directed at the exact middle pixel of the in-source optical microscope, fine adjustments of a few pixel offset can be applied at this stage. The automatically created setup file for the t-MALDI-2 MSI run can be opened directly in flexImaging to create measurement regions, select settings like pixel size, and laser shots per pixel. The previously acquired z-interpolation is recognized by timsControl software during t-MALDI-2 MSI measurements by providing the interpolation mesh as a geometry file. The python code for creating the flexImaging setup files can be found in Createmis.py Data S1 in the supplementary information. All t-MALDI-2 MSI experiments were performed with 25 laser shots per pixel and a pixel size of 1 x 1 \u0026micro;m\u003csup\u003e2\u003c/sup\u003e for the cerebellum and tumor sample and 1.5 x 1.5 \u0026micro;m\u003csup\u003e2\u003c/sup\u003e for cell culture samples. Signal intensity distributions are visualized using SCiLS Lab software (Bruker) using weak de-noising.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eIon signal annotation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFor the THP-1 cells lipid signals were identified as described by Schwenzfeier \u003cem\u003eet al\u003c/em\u003e. using shotgun lipidomics by electrospray ionization (ESI) mass spectrometry on a Q-Exactive Plus orbitrap mass spectrometer (Thermo Fisher) equipped with a dual ion funnel \u003csup\u003e2,3\u003c/sup\u003e. Lipid extracts of 72 h PMA stimulated THP-1 macrophages were produced by an adapted protocol of Folch et al. \u003csup\u003e4\u003c/sup\u003e. Approximately 10\u003csup\u003e6\u0026nbsp;\u003c/sup\u003e cells were centrifuged at 1000 g to separate them from the medium. Cells were resuspended in 1 ml CHCl\u003csub\u003e3\u003c/sub\u003e/MeOH (2:1, v/v) and lyzed for 10 min in an ultrasonic bath at room temperature followed by an incubation of 50 min at 4 \u0026deg;C in a cryoshaker at 500 RPM. Phase separation was induced by addition of 240 \u0026mu;l H\u003csub\u003e2\u003c/sub\u003eO and centrifugation at 9880 g for 10 min. The organic phase was transferred to a glass vial and the solvent was removed using a nitrogen evaporator. Lipid extracts were vacuum sealed and stored at -80 \u0026deg;C. For ESI-MS analysis, extracts were re-dissolved in 500 \u0026mu;l MeOH and measured both in positive and negative ion modes using a flow rate of 5 \u0026micro;l/min. A data dependent analysis workflow (DDA) was applied using tandem MS based on collision-induced dissociation (CID) with stepped normalization collision energies (NCE) of 20, 25, 30 and an exclusion time of 300 s. The mass resolving power was set to 280,000 (at \u003cem\u003em/z\u003c/em\u003e 200) with an AGC target of 1E6 and a maximum injection time of 250 ms. The resulting data was merged with a t-MALDI-2-MSI dataset and imported into LipostarMSI (Molecular Horizons) for annotation. The resulting feature list was hand-curated, which involved removing of isotope peaks and addition of known species from \u003cem\u003eE.coli\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eFor mouse cerebellum and the 4T1 murine tumor section, annotations were based on accurate mass (\u0026lt; 3 ppm). For this, signals with S/N \u0026gt; 3 were extracted from the mean spectrum and annotated using the LipidMaps \u0026quot;Bulk Structure Search\u0026rdquo; \u003csup\u003e5\u003c/sup\u003e. In positive ion-mode sphingolipids, glycerolipids, sterol lipids and glycerophospholipids were selected as possible lipid classes either as [M+H]\u003csup\u003e+\u003c/sup\u003e- or [M-H\u003csub\u003e2\u003c/sub\u003eO+H]\u003csup\u003e+\u003c/sup\u003e-ions. In negative ion-mode only [M-H]\u003csup\u003e-\u003c/sup\u003e-ions were searched for. The resulting annotation lists were manually curated to remove biologically unlikely annotation (e.g. lipids with odd side chains) or lipid classes that are usually not detected in the respective ion-mode (e.g. glycerophosphoinositols in positive ion-mode). Using this method, a total of 66 ion signals for the mouse cerebellum and 63 ion signals for the 4T1 tumor sample were annotated, respectively.\u003c/p\u003e\n\u003cp\u003eThe lipid annotation of mouse cerebellum was subsequently matched with data from a full lipidomics analysis reported in the literature \u003csup\u003e6\u003c/sup\u003e. Because t-MALDI-2-MSI without the use of additional techniques like OzID or Patern\u0026ograve;-B\u0026uuml;chi-reactions \u003csup\u003e7,8\u003c/sup\u003e is not able to differentiate isomers within a lipid class (e.g. PC(16:0/18:1) and PC(16:1/18:0)) or between lipid classes (e.g. PE(16:0/18:0) and PC(15:0/16:0)) the literature data was reduced to 255 unique lipid species according to the here employed level of specificity. Of the 66 annotated \u003cem\u003em/z\u003c/em\u003e-values, 63 were also detected by Fitzner \u003cem\u003eet al\u003c/em\u003e. \u003csup\u003e6\u003c/sup\u003e. The remaining unmatched values primarily corresponded to sulfoglycosphingolipids and fatty acids, lipid classes that were not analyzed in the original publication. Overall, the annotation accounted for approximately 82% of the total lipid mass identified by Fitzner \u003cem\u003eet al\u003c/em\u003e., calculated by summing all lipids found in both datasets and using the quantitative amounts reported.\u003c/p\u003e\n\u003cp\u003eLipid nomenclature and shorthand notation is reported according to Liebisch \u003csup\u003e9\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSingle-cell mass spectra from tissue\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe external and in-source fluorescence images were co-registered using SimpleITK 2.3.1 (NumFOCUS) \u003csup\u003e10,11\u003c/sup\u003e. For initialization a landmark registration was used. Based on the resulting translation matrix an affine image registration was performed using Mattes mutual information as metric and the gradient descent algorithm as optimizer \u003csup\u003e12\u003c/sup\u003e. The number of bins for the mutual information was set to 50, the parameters for the optimizer were set as follows: learning rate (1), number of iterations (50), convergence minimum value (1E-6) and convergence window (20). The sampling strategy was set to random and 1% of the pixels were sampled. The resulting transformation matrix was combined with the initial translation matrix to result in a global transformation matrix. Due to the inherent link between the t-MALDI-2-MSI measurement and the in-source fluorescence images, the global transformation matrix could be used to resample the external fluorescence images to fit the exact region of the t-MALDI-2-MSI measurement. The precision of the registration was determined by placing pairs of landmarks at distinct features, which are visible in the external fluorescence image as well as the MALDI image of the cerebellum dataset (\u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 1B\u003c/strong\u003e). While the precision of the placement of each individual landmark is limited by the manual selection process and differences between modalities, overall accuracy is ensured by determining the average over a total of eleven pairs of landmarks. The target registration error (TRE) was then determined by projecting the landmarks of the MALDI image into the coordinate system of the external fluorescence image using the global transformation matrix and calculating the distance to the corresponding landmarks of the external fluorescence image (\u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 1C and Note S1\u003c/strong\u003e). Cell masks were created by using DeepCell Mesmer 0.12.9 (Van Valen Lab) \u003csup\u003e13\u003c/sup\u003e based on the DAPI channel (Hoechst 33342, nuclei) and the FITC channel (CellMask Green Actin Tracking Stain, actin). Using FISCAS \u003csup\u003e2\u003c/sup\u003e, mass spectra were calculated for each cell by compiling MSI pixels based on their co-localized cell masks. MSI pixels assigned to multiple cell masks were scaled according to the hit coverage of the segmentation masks. Morphometric parameters were acquired using CellProfilers 4.2.1 (Cimini Lab) \u003csup\u003e14\u003c/sup\u003e measurement modules.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCell classification\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFor cell classification, immune cells were identified and divided into three groups (CD45\u003csup\u003e+\u003c/sup\u003e, LyG6\u003csup\u003e+\u003c/sup\u003e, DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e) based on thresholding the mean intensities of the corresponding antibody stains in their respective fluorescence channels (Cy5, TxRed, Cy7). Ly6G\u003csup\u003e+\u003c/sup\u003e cells showed high intensities on the TxRed and Cy5 channel, while DcTRAIL-R1\u003csup\u003e+\u003c/sup\u003e cells showed high intensities on all three channels. The clustering of the tumor microenvironment (TME) was done using k-means clustering on the standardized MS data of all Ly6G\u003csup\u003e-\u003c/sup\u003e cells. The number of clusters was determined by visual inspection of the clustering results for values of k between 6 and 10. The Ly6G\u003csup\u003e+\u003c/sup\u003e cells were also clustered using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) with an epsilon of 0.7 and a minimum number of 7 neighbors, based on a uniform manifold approximation and projection (UMAP) of their standardized mass spectra. The clusters are visualized both directly on the cell masks as well as in a UMAP of the standardized mass spectra using default parameters. Ly6G\u003csup\u003e+\u003c/sup\u003e cells were assigned to a specific tumor region, by identifying the most frequent TME label of all neighbors within 40 \u0026micro;m.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eReferences Methods\u003c/strong\u003e\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003eBien, T., Koerfer, K., Schwenzfeier, J., Dreisewerd, K. \u0026amp; Soltwisch, J. Mass spectrometry imaging to explore molecular heterogeneity in cell culture. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e \u003cstrong\u003e119\u003c/strong\u003e, e2114365119 (2022).\u003c/li\u003e\n\u003cli\u003eSchwenzfeier, J., Weischer, S., Bessler, S. \u0026amp; Soltwisch, J. Introducing FISCAS, a Tool for the Effective Generation of Single Cell MALDI-MSI Data. \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 2950\u0026ndash;2959 (2024).\u003c/li\u003e\n\u003cli\u003eBelov, M. E. \u003cem\u003eet al.\u003c/em\u003e Design and Performance of a Novel Interface for Combined Matrix-Assisted Laser Desorption Ionization at Elevated Pressure and Electrospray Ionization with Orbitrap Mass Spectrometry. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e89\u003c/strong\u003e, 7493\u0026ndash;7501 (2017).\u003c/li\u003e\n\u003cli\u003eFolch, J., Lees, M. \u0026amp; Sloane Stanley, G. H. A simple method for the isolation and purification of total lipides from animal tissues. \u003cem\u003eJ. Biol. Chem.\u003c/em\u003e \u003cstrong\u003e226\u003c/strong\u003e, 497\u0026ndash;509 (1957).\u003c/li\u003e\n\u003cli\u003eFahy, E., Sud, M., Cotter, D. \u0026amp; Subramaniam, S. LIPID MAPS online tools for lipid research. \u003cem\u003eNucleic Acids Res.\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, W606-612 (2007).\u003c/li\u003e\n\u003cli\u003eFitzner, D. \u003cem\u003eet al.\u003c/em\u003e Cell-Type- and Brain-Region-Resolved Mouse Brain Lipidome. \u003cem\u003eCell Rep.\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 108132 (2020).\u003c/li\u003e\n\u003cli\u003eClaes, B. S. R. \u003cem\u003eet al.\u003c/em\u003e Mass Spectrometry Imaging of Lipids with Isomer Resolution Using High-Pressure Ozone-Induced Dissociation. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e93\u003c/strong\u003e, 9826\u0026ndash;9834 (2021).\u003c/li\u003e\n\u003cli\u003eBednař\u0026iacute;k, A., B\u0026ouml;lsker, S., Soltwisch, J. \u0026amp; Dreisewerd, K. An On-Tissue Patern\u0026ograve;-B\u0026uuml;chi Reaction for Localization of Carbon-Carbon Double Bonds in Phospholipids and Glycolipids by Matrix-Assisted Laser-Desorption-Ionization Mass-Spectrometry Imaging. \u003cem\u003eAngew. Chem. Int. Ed. Engl.\u003c/em\u003e \u003cstrong\u003e57\u003c/strong\u003e, 12092\u0026ndash;12096 (2018).\u003c/li\u003e\n\u003cli\u003eLiebisch, G. \u003cem\u003eet al.\u003c/em\u003e Update on LIPID MAPS classification, nomenclature, and shorthand notation for MS-derived lipid structures. \u003cem\u003eJournal of Lipid Research\u003c/em\u003e \u003cstrong\u003e61\u003c/strong\u003e, 1539\u0026ndash;1555 (2020).\u003c/li\u003e\n\u003cli\u003eYaniv, Z., Lowekamp, B. C., Johnson, H. J. \u0026amp; Beare, R. SimpleITK Image-Analysis Notebooks: a Collaborative Environment for Education and Reproducible Research. \u003cem\u003eJ Digit Imaging\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 290\u0026ndash;303 (2018).\u003c/li\u003e\n\u003cli\u003eLowekamp, B. C., Chen, D. T., Ibanez, L. \u0026amp; Blezek, D. The Design of SimpleITK. \u003cem\u003eFront. Neuroinform.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, (2013).\u003c/li\u003e\n\u003cli\u003eMattes, D., Haynor, D. R., Vesselle, H., Lewellyn, T. K. \u0026amp; Eubank, W. Nonrigid multimodality image registration. in \u003cem\u003eMedical Imaging 2001: Image Processing\u003c/em\u003e vol. 4322 1609\u0026ndash;1620 (SPIE, 2001).\u003c/li\u003e\n\u003cli\u003eGreenwald, N. F. \u003cem\u003eet al.\u003c/em\u003e Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. \u003cem\u003eNat. Biotechnol.\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 555\u0026ndash;565 (2022).\u003c/li\u003e\n\u003cli\u003eStirling, D. R. \u003cem\u003eet al.\u003c/em\u003e CellProfiler 4: improvements in speed, utility and usability. \u003cem\u003eBMC Bioinf.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 433 (2021).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eResource availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMaterials availability\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis study did not generate new unique reagents.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eData and code availability\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eMALDI MSI data, fluorescence and brightfield microscopy images, processed data and all original code have been deposited at OMERO or in the supplementary information and are publicly available as of the date of publication. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledegements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank Dagmar Mense for help with cell culture, Mathis Richter for helpful discussions on immunohistochemistry, Alexander Fengler and Vincent Suerdieck (all University of M\u0026uuml;nster) for help with mouse tissue sectioning, Benedikt Geier (Stanford), Robert Ernst and Hyojik Yang (both University of Maryland) for helpful discussions, Henning Peise, Arne F\u0026uuml;tterer, Andreas W\u0026ouml;ste, Raik Milautzki (all Bruker) for hard- and software support, \u0026nbsp;and Sarah Weischer and Thomas Zobel from the M\u0026uuml;nster Imaging Network Cells in Motion Interfaculty Centre for support with image analysis. Figure 1 was created in part using biorender (agreement number GO275P34H3). Financial support from the German Research Foundation to KD and OS (Z1 project of the TRR332; project no. 449437943) and to JS (project no. 544444139) is gratefully acknowledged. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: KD, JH, MN, AP, JS; Data curation: SB, AP, JSch, JS; Formal analysis: SB, AP, JSch; Funding acquisition: KD, JH, OS, JS; Investigation: SB; MN, AP, JSch, JS; Methodology: SB, MN, AP, JSch, JS; Project administration: KD, JS; Resources: KD, JH, EH OS, JS; Software: AP, JSch; Supervision: KD, OS, JS; Validation: KD, JH, OS; Visualization: SB, AP, JSch, JS; Writing \u0026ndash; original draft: AP, JS; Writing \u0026ndash; review \u0026amp; editing: SB, KD, JH, MN, AP, JSch, OS, JS.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare the following competing interest(s): MN and JH are employees of Bruker Daltonik GmbH \u0026amp; Co KG (Bremen). KD, MN, AP, and JS have patents pending regarding parts of the hardware development (WO/2024/041681, WO/2025/067576).\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eRafelski, S. M. \u0026amp; Theriot, J. A. Establishing a conceptual framework for holistic cell states and state transitions. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e187\u003c/strong\u003e, 2633\u0026ndash;2651 (2024).\u003c/li\u003e\n\u003cli\u003eBennett, H. M., Stephenson, W., Rose, C. M. \u0026amp; Darmanis, S. Single-cell proteomics enabled by next-generation sequencing or mass spectrometry. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 363\u0026ndash;374 (2023).\u003c/li\u003e\n\u003cli\u003eBoekweg, H. \u0026amp; Payne, S. H. Challenges and Opportunities for Single-cell Computational Proteomics. \u003cem\u003eMol. Cell. Proteomics\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 100518 (2023).\u003c/li\u003e\n\u003cli\u003eBressan, D., Battistoni, G. \u0026amp; Hannon, G. J. The dawn of spatial omics. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e381\u003c/strong\u003e, eabq4964 (2023).\u003c/li\u003e\n\u003cli\u003eVandereyken, K., Sifrim, A., Thienpont, B. \u0026amp; Voet, T. Methods and applications for single-cell and spatial multi-omics. \u003cem\u003eNat. Rev. Genet.\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 494\u0026ndash;515 (2023).\u003c/li\u003e\n\u003cli\u003eBodenmiller, B. Highly multiplexed imaging in the omics era: understanding tissue structures in health and disease. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 2209\u0026ndash;2211 (2024).\u003c/li\u003e\n\u003cli\u003eMund, A. \u003cem\u003eet al.\u003c/em\u003e Deep Visual Proteomics defines single-cell identity and heterogeneity. \u003cem\u003eNat Biotechnol\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 1231\u0026ndash;1240 (2022).\u003c/li\u003e\n\u003cli\u003eDilmetz, B. A. \u003cem\u003eet al.\u003c/em\u003e Novel technical developments in mass spectrometry imaging in 2020: A mini review. \u003cem\u003eAnalytical Science Advances\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 225\u0026ndash;237 (2021).\u003c/li\u003e\n\u003cli\u003eSoltwisch, J. \u003cem\u003eet al.\u003c/em\u003e Mass spectrometry imaging with laser-induced postionization. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e348\u003c/strong\u003e, 211\u0026ndash;215 (2015).\u003c/li\u003e\n\u003cli\u003eCuypers, E. \u003cem\u003eet al.\u003c/em\u003e \u0026lsquo;On the Spot\u0026rsquo; Digital Pathology of Breast Cancer Based on Single-Cell Mass Spectrometry Imaging. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e94\u003c/strong\u003e, 6180\u0026ndash;6190 (2022).\u003c/li\u003e\n\u003cli\u003eTaylor, M. J., Lukowski, J. K. \u0026amp; Anderton, C. R. Spatially Resolved Mass Spectrometry at the Single Cell: Recent Innovations in Proteomics and Metabolomics. \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 872\u0026ndash;894 (2021).\u003c/li\u003e\n\u003cli\u003eBourceau, P. \u003cem\u003eet al.\u003c/em\u003e Visualization of metabolites and microbes at high spatial resolution using MALDI mass spectrometry imaging and in situ fluorescence labeling. \u003cem\u003eNat. Protoc.\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 3050\u0026ndash;3079 (2023).\u003c/li\u003e\n\u003cli\u003eM\u0026uuml;ller, M. A., Zweig, N., Spengler, B., Weinert, M. \u0026amp; Heiles, S. Lipid Signatures and Inter-Cellular Heterogeneity of Naı̈ve and Lipopolysaccharide-Stimulated Human Microglia-like Cells. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e95\u003c/strong\u003e, 11672\u0026ndash;11679 (2023).\u003c/li\u003e\n\u003cli\u003eMarques, C. \u003cem\u003eet al.\u003c/em\u003e Global and Spatial Metabolomics of Individual Cells Using a Tapered Pneumatically Assisted nano-DESI Probe. \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 2518\u0026ndash;2524 (2023).\u003c/li\u003e\n\u003cli\u003eNiehaus, M., Soltwisch, J., Belov, M. E. \u0026amp; Dreisewerd, K. Transmission-mode MALDI-2 mass spectrometry imaging of cells and tissues at subcellular resolution. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 925\u0026ndash;931 (2019).\u003c/li\u003e\n\u003cli\u003eKompauer, M., Heiles, S. \u0026amp; Spengler, B. Atmospheric pressure MALDI mass spectrometry imaging of tissues and cells at 1.4-\u0026mu;m lateral resolution. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 90 (2017).\u003c/li\u003e\n\u003cli\u003eBien, T., Koerfer, K., Schwenzfeier, J., Dreisewerd, K. \u0026amp; Soltwisch, J. Mass spectrometry imaging to explore molecular heterogeneity in cell culture. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e \u003cstrong\u003e119\u003c/strong\u003e, e2114365119 (2022).\u003c/li\u003e\n\u003cli\u003eRappez, L. \u003cem\u003eet al.\u003c/em\u003e SpaceM reveals metabolic states of single cells. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 799\u0026ndash;805 (2021).\u003c/li\u003e\n\u003cli\u003eCapolupo, L. \u003cem\u003eet al.\u003c/em\u003e Sphingolipids control dermal fibroblast heterogeneity. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e376\u003c/strong\u003e, eabh1623 (2022).\u003c/li\u003e\n\u003cli\u003eSchwenzfeier, J., Weischer, S., Bessler, S. \u0026amp; Soltwisch, J. Introducing FISCAS, a Tool for the Effective Generation of Single Cell MALDI-MSI Data. \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 2950\u0026ndash;2959 (2024).\u003c/li\u003e\n\u003cli\u003e\u0026Scaron;čup\u0026aacute;kov\u0026aacute;, K., Dewez, F., Walch, A. K., Heeren, R. M. A. \u0026amp; Balluff, B. Morphometric Cell Classification for Single-Cell MALDI-Mass Spectrometry Imaging. \u003cem\u003eAngew. Chem. Int. Ed. Engl.\u003c/em\u003e \u003cstrong\u003e132\u003c/strong\u003e, 17600\u0026ndash;17603 (2020).\u003c/li\u003e\n\u003cli\u003eBien, T., Bessler, S., Dreisewerd, K. \u0026amp; Soltwisch, J. Transmission-Mode MALDI Mass Spectrometry Imaging of Single Cells: Optimizing Sample Preparation Protocols. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e93\u003c/strong\u003e, 4513\u0026ndash;4520 (2021).\u003c/li\u003e\n\u003cli\u003eSoltwisch, J. \u003cem\u003eet al.\u003c/em\u003e MALDI-2 on a Trapped Ion Mobility Quadrupole Time-of-Flight Instrument for Rapid Mass Spectrometry Imaging and Ion Mobility Separation of Complex Lipid Profiles. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e92\u003c/strong\u003e, 8697\u0026ndash;8703 (2020).\u003c/li\u003e\n\u003cli\u003eGeng, Z. \u003cem\u003eet al.\u003c/em\u003e Enhanced MALDI-2 Sensitivity with Reflecting Post-Ionization Laser for High-Resolution MS Imaging Combined with Real-Time Microscope Imaging. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e96\u003c/strong\u003e, 12983\u0026ndash;12990 (2024).\u003c/li\u003e\n\u003cli\u003eFernandes, A. M. A. P. \u003cem\u003eet al.\u003c/em\u003e Direct Visualization of Neurotransmitters in Rat Brain Slices by Desorption Electrospray Ionization Mass Spectrometry Imaging (DESI - MS). \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 1944\u0026ndash;1951 (2016).\u003c/li\u003e\n\u003cli\u003eShariatgorji, M., Svenningsson, P. \u0026amp; Andr\u0026eacute;n, P. E. Mass Spectrometry Imaging, an Emerging Technology in Neuropsychopharmacology. \u003cem\u003eNeuropsychopharmacol.\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 34\u0026ndash;49 (2014).\u003c/li\u003e\n\u003cli\u003eAg\u0026uuml;i-Gonzalez, P., J\u0026auml;hne, S. \u0026amp; Phan, N. T. N. SIMS imaging in neurobiology and cell biology. \u003cem\u003eJ. Anal. At. Spectrom.\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 1355\u0026ndash;1368 (2019).\u003c/li\u003e\n\u003cli\u003eSchr\u0026ouml;der, H., Moser, N. \u0026amp; Huggenberger, S. The Mouse Cerebellum. in \u003cem\u003eNeuroanatomy of the Mouse: An Introduction\u003c/em\u003e (eds. Schr\u0026ouml;der, H., Moser, N. \u0026amp; Huggenberger, S.) 153\u0026ndash;170 (Springer International Publishing, Cham, 2020). doi:10.1007/978-3-030-19898-5_7.\u003c/li\u003e\n\u003cli\u003eFitzner, D. \u003cem\u003eet al.\u003c/em\u003e Cell-Type- and Brain-Region-Resolved Mouse Brain Lipidome. \u003cem\u003eCell Rep.\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 108132 (2020).\u003c/li\u003e\n\u003cli\u003ePatterson, N. H., Tuck, M., Van de Plas, R. \u0026amp; Caprioli, R. M. Advanced Registration and Analysis of MALDI Imaging Mass Spectrometry Measurements through Autofluorescence Microscopy. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e90\u003c/strong\u003e, 12395\u0026ndash;12403 (2018).\u003c/li\u003e\n\u003cli\u003eSugiura, Y. \u003cem\u003eet al.\u003c/em\u003e Visualization of the cell-selective distribution of PUFA-containing phosphatidylcholines in mouse brain by imaging mass spectrometry. \u003cem\u003eJ. Lipid Res.\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 1776\u0026ndash;1788 (2009).\u003c/li\u003e\n\u003cli\u003eVos, D. R. N., Bowman, A. P., Heeren, R. M. A., Balluff, B. \u0026amp; Ellis, S. R. Class-specific depletion of lipid ion signals in tissues upon formalin fixation. \u003cem\u003eInt. J. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e446\u003c/strong\u003e, 116212 (2019).\u003c/li\u003e\n\u003cli\u003eKralj, T. \u003cem\u003eet al.\u003c/em\u003e Multi-Omic Analysis to Characterize Metabolic Adaptation of the E. coli Lipidome in Response to Environmental Stress. \u003cem\u003eMetabolites\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 171 (2022).\u003c/li\u003e\n\u003cli\u003eElsbach, P. \u0026amp; Weiss, J. Phagocytosis of bacteria and phospholipid degradation. \u003cem\u003eBiochimica et Biophysica Acta (BBA) - Reviews on Biomembranes\u003c/em\u003e \u003cstrong\u003e947\u003c/strong\u003e, 29\u0026ndash;52 (1988).\u003c/li\u003e\n\u003cli\u003eHambrey, P. N. \u0026amp; Mellors, A. Cardiolipin degradation by rat liver lysosomes. \u003cem\u003eBiochemical and Biophysical Research Communications\u003c/em\u003e \u003cstrong\u003e62\u003c/strong\u003e, 939\u0026ndash;945 (1975).\u003c/li\u003e\n\u003cli\u003eXiao, Z., Dai, Z. \u0026amp; Locasale, J. W. Metabolic landscape of the tumor microenvironment at single cell resolution. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 3763 (2019).\u003c/li\u003e\n\u003cli\u003eDuPr\u0026eacute;, S. A., Redelman, D. \u0026amp; Hunter, K. W. The mouse mammary carcinoma 4T1: characterization of the cellular landscape of primary tumours and metastatic tumour foci. \u003cem\u003eInt. J. Exp. Pathol.\u003c/em\u003e \u003cstrong\u003e88\u003c/strong\u003e, 351\u0026ndash;360 (2007).\u003c/li\u003e\n\u003cli\u003eSchr\u0026ouml;rs, B. \u003cem\u003eet al.\u003c/em\u003e Multi-Omics Characterization of the 4T1 Murine Mammary Gland Tumor Model. \u003cem\u003eFront. Oncol.\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 1195 (2020).\u003c/li\u003e\n\u003cli\u003eHoffmann, E. \u003cem\u003eet al.\u003c/em\u003e Profiling specific cell populations within the inflammatory tumor microenvironment by oscillating-gradient diffusion-weighted MRI. \u003cem\u003eJ. Immunother. Cancer\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, e006092 (2023).\u003c/li\u003e\n\u003cli\u003eLi, P. \u003cem\u003eet al.\u003c/em\u003e Dual roles of neutrophils in metastatic colonization are governed by the host NK cell status. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 4387 (2020).\u003c/li\u003e\n\u003cli\u003eNg, M. S. F. \u003cem\u003eet al.\u003c/em\u003e Deterministic reprogramming of neutrophils within tumors. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e383\u003c/strong\u003e, eadf6493 (2024).\u003c/li\u003e\n\u003cli\u003eSilvestre-Roig, C., Kalafati, L. \u0026amp; Chavakis, T. Neutrophils are shaped by the tumor microenvironment: novel possibilities for targeting neutrophils in cancer. \u003cem\u003eSig Transduct Target Ther\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 1\u0026ndash;3 (2024).\u003c/li\u003e\n\u003cli\u003eSpangenberg, P. \u003cem\u003eet al.\u003c/em\u003e msiFlow: automated workflows for reproducible and scalable multimodal mass spectrometry imaging and microscopy data analysis. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 1065 (2025).\u003c/li\u003e\n\u003cli\u003eMorgan, P. K. \u003cem\u003eet al.\u003c/em\u003e A lipid atlas of human and mouse immune cells provides insights into ferroptosis susceptibility. \u003cem\u003eNat. Cell Biol.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 645\u0026ndash;659 (2024).\u003c/li\u003e\n\u003cli\u003eAhmadian, M., Duncan, R. E., Jaworski, K., Sarkadi-Nagy, E. \u0026amp; Sook Sul, H. Triacylglycerol Metabolism In Adipose Tissue. \u003cem\u003eFuture Lipidol.\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 229\u0026ndash;237 (2007).\u003c/li\u003e\n\u003cli\u003eChughtai, K., Jiang, L., Greenwood, T. R., Glunde, K. \u0026amp; Heeren, R. M. A. Mass spectrometry images acylcarnitines, phosphatidylcholines, and sphingomyelin in MDA-MB-231 breast tumor models[S]. \u003cem\u003eJ. Lipid Res.\u003c/em\u003e \u003cstrong\u003e54\u003c/strong\u003e, 333\u0026ndash;344 (2013).\u003c/li\u003e\n\u003cli\u003eAnderson, D. M. G., Spraggins, J. M., Rose, K. L. \u0026amp; Schey, K. L. High Spatial Resolution Imaging Mass Spectrometry of Human Optic Nerve Lipids and Proteins. \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 940\u0026ndash;947 (2015).\u003c/li\u003e\n\u003cli\u003eGreenwald, N. F. \u003cem\u003eet al.\u003c/em\u003e Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. \u003cem\u003eNat. Biotechnol.\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 555\u0026ndash;565 (2022).\u003c/li\u003e\n\u003cli\u003eStirling, D. R. \u003cem\u003eet al.\u003c/em\u003e CellProfiler 4: improvements in speed, utility and usability. \u003cem\u003eBMC Bioinf.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 433 (2021).\u003c/li\u003e\n\u003cli\u003eVicanolo, T. \u003cem\u003eet al.\u003c/em\u003e Matrix-producing neutrophils populate and shield the skin. \u003cem\u003eNature\u003c/em\u003e 1\u0026ndash;9 (2025) doi:10.1038/s41586-025-08741-5.\u003c/li\u003e\n\u003cli\u003eBalluff, B., Heeren, R. M. A. \u0026amp; Race, A. M. An overview of image registration for aligning mass spectrometry imaging with clinically relevant imaging modalities. \u003cem\u003eJ. Mass Spectrom. Adv. Clin. Lab\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 26\u0026ndash;38 (2022).\u003c/li\u003e\n\u003cli\u003eGeier, B. \u003cem\u003eet al.\u003c/em\u003e Connecting structure and function from organisms to molecules in small-animal symbioses through chemo-histo-tomography. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e \u003cstrong\u003e118\u003c/strong\u003e, e2023773118 (2021).\u003c/li\u003e\n\u003cli\u003eGeier, B. \u003cem\u003eet al.\u003c/em\u003e Spatial metabolomics of in situ host\u0026ndash;microbe interactions at the micrometre scale. \u003cem\u003eNat. Microbiol.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 498\u0026ndash;510 (2020).\u003c/li\u003e\n\u003cli\u003eLin, B.-J. \u003cem\u003eet al.\u003c/em\u003e MSIr: Automatic Registration Service for Mass Spectrometry Imaging and Histology. \u003cem\u003eAnal. Chem.\u003c/em\u003e \u003cstrong\u003e95\u003c/strong\u003e, 3317\u0026ndash;3324 (2023).\u003c/li\u003e\n\u003cli\u003eYang, E. \u003cem\u003eet al.\u003c/em\u003e FluoMALDI Microscopy: Matrix Co-Crystallization Simultaneously Enhances Fluorescence and MALDI Imaging. \u003cem\u003eAdv. Sci.\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 2304343 (2023).\u003c/li\u003e\n\u003cli\u003eYeung, T., Ozdamar, B., Paroutis, P. \u0026amp; Grinstein, S. Lipid metabolism and dynamics during phagocytosis. \u003cem\u003eCurrent Opinion in Cell Biology\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 429\u0026ndash;437 (2006).\u003c/li\u003e\n\u003cli\u003eNikitina, A. \u003cem\u003eet al.\u003c/em\u003e A Co-registration Pipeline for Multimodal MALDI and Confocal Imaging Analysis of Stem Cell Colonies. \u003cem\u003eJ. Am. Soc. Mass Spectrom.\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 986\u0026ndash;989 (2020).\u003c/li\u003e\n\u003cli\u003eSchneider, G., Sellers, Z. P., Abdel-Latif, A., Morris, A. J. \u0026amp; Ratajczak, M. Z. Bioactive Lipids, LPC and LPA, Are Novel Prometastatic Factors and Their Tissue Levels Increase in Response to Radio/Chemotherapy. \u003cem\u003eMolecular Cancer Research\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 1560\u0026ndash;1573 (2014).\u003c/li\u003e\n\u003cli\u003eLauber, K. \u003cem\u003eet al.\u003c/em\u003e Apoptotic Cells Induce Migration of Phagocytes via Caspase-3-Mediated Release of a Lipid Attraction Signal. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e113\u003c/strong\u003e, 717\u0026ndash;730 (2003).\u003c/li\u003e\n\u003cli\u003eVolpi, E. V. \u0026amp; Bridger, J. M. FISH Glossary: An Overview of the Fluorescence in Situ Hybridization Technique. \u003cem\u003eBioTechniques\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 385\u0026ndash;409 (2008).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Data S1","content":"\u003cp\u003eData S1 is not available with this version\u003c/p\u003e\n"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6751055/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6751055/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSpatial biology pursues the analysis of cells in their native microenvironment, often with regard to morphology and gene- or protein expression. The spatial analysis of lipid and metabolic profiles by matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) can add another layer of molecular information. A seamless integration of MSI data with other (multimodal) methods at a single-cell level is often hampered by insufficient co-registration and resolution. Here, we introduce a MALDI-MSI based method that integrates in-source bright-field and fluorescence microscopy, allowing for a coupled (sub)cellular investigation of the same sample in both modalities. Presented examples from cell culture and tissue analysis include the visualization of intracellular lipid distributions in macrophages during phagocytosis, and the heterogeneity of lipid profiles of tumor infiltrating neutrophils correlated to their individual microenvironments. The achieved combination of lipid profiling with morphological features and protein expression on the single-cell level constitutes a powerful method for cell biology.\u003c/p\u003e","manuscriptTitle":"Spatial biology using single-cell mass spectrometry imaging and integrated microscopy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-25 07:08:30","doi":"10.21203/rs.3.rs-6751055/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"327c16b5-67cf-4e8e-99b8-405b6128ec93","owner":[],"postedDate":"June 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":49603869,"name":"Biological sciences/Cell biology/Cellular imaging"},{"id":49603870,"name":"Biological sciences/Molecular biology"},{"id":49603871,"name":"Biological sciences/Biological techniques/Imaging/Molecular imaging"},{"id":49603872,"name":"Biological sciences/Biological techniques/Mass spectrometry"},{"id":49603873,"name":"Biological sciences/Biological techniques/Metabolomics"}],"tags":[],"updatedAt":"2025-10-16T07:05:51+00:00","versionOfRecord":{"articleIdentity":"rs-6751055","link":"https://doi.org/10.1038/s41467-025-64603-8","journal":{"identity":"nature-communications","isVorOnly":false,"title":"Nature Communications"},"publishedOn":"2025-10-15 04:00:00","publishedOnDateReadable":"October 15th, 2025"},"versionCreatedAt":"2025-06-25 07:08:30","video":"","vorDoi":"10.1038/s41467-025-64603-8","vorDoiUrl":"https://doi.org/10.1038/s41467-025-64603-8","workflowStages":[]},"version":"v1","identity":"rs-6751055","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6751055","identity":"rs-6751055","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}