Systematic evaluation of PASEF acquisition strategies in complex metaproteomes

Systematic evaluation of PASEF acquisition strategies in complex metaproteomes | 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 Research Article Systematic evaluation of PASEF acquisition strategies in complex metaproteomes Feng Xian, Goran Mitulovic, Ranjith Kumar Ravi Kumar, Lukas Uhrik, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9103948/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Metaproteomics enables direct measurement of functional activity in complex microbial communities but remains technically challenging due to the extreme complexity and dynamic range of microbiome samples. Trapped ion mobility spectrometry coupled with parallel accumulation–serial fragmentation (PASEF) offers unique advantages in general proteomics and in the analysis of complex metaproteomes. In recent years, PASEF has diversified into multiple acquisition modes. However, the lack of systematic side-by-side evaluations under matched experimental conditions, together with the predominant use of low- to medium-complexity benchmark samples, hampers informed selection of acquisition strategies. Here, we benchmark five PASEF acquisition modes (DDA-, DIA-, Slice-, Synchro-, and midia-PASEF) using a complex fecal peptide background spiked with defined bacterial references. Performance was evaluated across three chromatographic gradients and input levels, comprising 540 LC–MS acquisitions processed using a single computational workflow. DIA-based strategies consistently achieved greater peptide and protein coverage than DDA-PASEF, particularly for low-abundance microbial features. Slice- and DIA-PASEF exhibited the lowest quantitative variability, minimal ratio compression, and the most consistent scaling of species abundances, while functional profiling revealed an expanded dynamic range of enzyme annotations. Application to a murine colonic injury model demonstrated that DIA- and Slice-PASEF capture highly concordant host and microbial responses. Together, this study provides a unified evaluation of PASEF acquisition strategies and illustrates how acquisition choices influence sensitivity, reproducibility, and functional resolution in proteomic analyses. Analytical Biochemistry Systems Biology General Microbiology Metaproteomics Ion mobility mass spectrometry PASEF Microbiome Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Metaproteomics is increasingly used to investigate the functional interplay between complex microbial communities (e.g., gut microbiome) and their host 1, 2 . Gut metaproteomics represents one of the most demanding application scenarios for LC–MS–based proteomics 3, 4, 5 . Fecal samples comprise dense mixtures of host and microbial proteomes spanning several orders of magnitude in abundance, providing a stringent stress test for acquisition strategies and data analysis pipelines 6 . At the same time, metaproteomic studies increasingly seek to quantify low-abundance community members and functional pathways that are central to host–microbiome interactions and diseases, placing exceptional demands on sensitivity, reproducibility, and quantitative accuracy 7, 8, 9 . Trapped ion mobility spectrometry coupled with parallel accumulation–serial fragmentation (PASEF) has substantially expanded the speed, sensitivity, and quantitative precision of liquid chromatography–mass spectrometer (LC-MS)-based proteomics 10, 11, 12, 13, 14, 15, 16 . On this platform, multiple acquisition strategies have been developed, including data-dependent (DDA-PASEF 10 ), data-independent (DIA-PASEF 11 ), and more recent variants such as Slice- 12 , Synchro- 13 , and midia-PASEF 14 , which differ in their precursor sampling logic and trade-offs between analytical depth and throughput. However, the direct side-by-side comparison under matched analytical and computational conditions is missing. Moreover, despite their benefits in relatively low- to medium-complex samples, the performance of each acquisition strategy in highly complex biological matrices, where dynamic range, co-elution, and sampling bias strongly influence both identification and quantification 17 , is unknown. PASEF technology applied to complex metaproteomes 18 improves taxonomic and functional depth, increases throughput, improves limits of detection, and discovers host–microbiome interactions in preclinical mouse models of inflammatory bowel disease 1 . As a result, DDA- and DIA-PASEF are increasingly adopted in metaproteomic workflows to improve sequencing depth and throughput in gut microbiome studies 18, 19, 20, 21, 22 . Thus, it is necessary to study if the more recent methods 12, 13, 14 (e.g., Slice-, Synchro-, midia-PASEF) could further provide improvements on identification depth, sensitivity to low-abundance species, quantitative reliability, and functional interpretation in complex microbiome samples 18, 20 . Here we established a controlled benchmarking framework to evaluate five PASEF acquisition strategies (PentaPASEF) across chromatographic regimes and sample input levels using a highly complex fecal peptide background spiked with defined reference bacteria. We systematically assessed qualitative performance, quantitative precision and accuracy, species- and function-level consistency, and false-positive control across 540 matched LC–MS acquisitions. Finally, we evaluated the biological impact of acquisition choice in a murine epithelial injury model 23 , showing that DIA- and Slice-PASEF captured highly concordant microbial and host responses while differing primarily in quantitative sensitivity-related changes. Together, this study enables direct translation of acquisition design into expected analytical performance, providing practical guidance for benchmarking PASEF strategies according to sample complexity, sensitivity requirements, and experimental throughput. Results Experimental design for benchmarking PASEF acquisition strategies in fecal microbiomes Ion mobility spectrometry adds an orthogonal dimension of separation to proteomics by resolving ions according to their collisional cross section, thereby improving selectivity, sensitivity, and quantitative precision. When combined with trapped ion mobility spectrometry (TIMS) and parallel accumulation–serial fragmentation (PASEF), this approach enables near-complete ion utilization and accelerated duty cycles, substantially enhancing proteome coverage. These advances have led to several PASEF acquisition strategies, including DDA-, DIA-, Slice-, Synchro-, and midia-PASEF, yet their relative performance in highly complex biological matrices remains insufficiently characterized. To enable a systematic comparison, we established a controlled benchmarking framework spanning different chromatographic separations and sample complexities (Fig. 1). Human fecal material was selected as the background matrix owing to its extreme proteomic complexity 7, 24 , which captures the dynamic range and compositional heterogeneity typical of real biological samples. Two reference bacteria—SILAC-labeled Ligilactobacillus murinus ( L. murinus ) and unlabeled Salinibacter ruber ( S. ruber ) 1 —were spiked into the fecal peptide background at a fixed ratio (1:3) to emulate low-abundance community members. Three input levels (50 pg : 150 pg, 5 pg : 15 pg, and 0.5 pg : 1.5 pg) were analyzed in triplicate to ensure statistical robustness, together with non-spiked controls. LC–MS/MS data were acquired using three gradient lengths (5, 22, and 45 min) to balance throughput and depth across all PentaPASEF modes, resulting in 540 injections covering all experimental combinations. DDA-PASEF data were processed with MSFragger 25 within FragPipe, whereas all other datasets were analyzed using DIA-NN 26 to maximize comparability, applying a 1% FDR at precursor and protein group levels. To ensure confident identification of spiked bacterial peptides, any heavy-labeled L. murinus or S. ruber identifications observed in non-spike controls (gradient- and method-wise) were removed from corresponding experimental runs, and background peptides falsely assigned with heavy labeling were filtered. This experimental framework provides a rigorous and biologically relevant basis to compare PASEF acquisition strategies, capturing their relative performance across different chromatographic resolutions, acquisition modes, and analytical depths in metaproteomic contexts. Acquisition strategy shapes taxonomic and functional resolution in metaproteomics We first assessed the qualitative performance of PentaPASEF acquisition strategies across LC gradients and spike-in levels. The overall number of identified peptides (Fig. 2A; Supplementary Data 1) and proteins (Fig. 2B) remained relatively stable across spike-in conditions and replicates (acquisition triplicates were averaged; n = 9 per bar), demonstrating the robustness of all PASEF methods. This stability was expected because the spiked bacterial peptides constituted only a minor fraction of the total peptide pool within the complex fecal background. Across all PentaPASEF methods, peptide and protein identifications scaled consistently with LC gradient length, with 45-min separations yielding the highest coverage (Fig. 2A-2B). DIA-based methods, particularly Slice- and DIA-PASEF, consistently exceeded DDA-PASEF in total identifications, reflecting their enhanced ion sampling efficiency and reduced precursor selection bias. Notably, even at the 5-min gradient, DIA-based modes maintained substantial peptide coverage, underscoring their suitability for high-throughput analyses. For Slice- and midia-PASEF, alternative acquisition schemes were evaluated. A one-frame Slice configuration and a 5-Da overlay in midia-PASEF yielded higher identifications (Supplementary Fig. 1A) and were therefore retained for downstream analyses. Interestingly, midia-PASEF showed fewer identifications compared to DIA- and Slice-PASEF. To exclude analysis pipeline–dependent effects, DIA-PASEF and midia-PASEF datasets were reanalyzed with Spectronaut 20, which reproduced the same pattern (Supplementary Fig. 1B). Peptide-level overlap analysis revealed a large shared core among methods at all gradients, with overlap increasing at longer separations (Supplementary Fig. 1C–E). DDA-PASEF generated the highest number of unique identifications, consistent with its semi-random targeting of abundant precursors, which resulted in lower reproducibility relative to DIA-based approaches. Focusing on the spiked L. murinus and S. ruber peptides, the PentaPASEF methods displayed distinct sensitivity profiles (Fig. 2C; Supplementary Data 2). While following the gradient-dependent trend observed for total identifications, the number of spike-in peptide identifications decreased proportionally with input amounts. Nevertheless, DIA-based strategies retained higher recovery across gradients, demonstrating superior detection of low-abundance peptides in complex matrices. Replicate variability increased at lower inputs (Fig. 2C), reflecting greater sampling fluctuation as signals approached the detection limit. Importantly, even at the sub-picogram level (0.5 pg L. murinus ), DIA-based approaches—particularly DIA- and Slice-PASEF—still yielded detectable peptide identifications (Fig. 2C; Supplementary Data 2), most prominently at the 22- and 45-min gradients. Assessment of non-spike controls confirmed excellent false-positive control for both FragPipe and DIA-NN, with mis-assigned heavy-labeled L. murinus and S. ruber peptides below 1% across datasets (Supplementary Fig. 1F). DDA-PASEF showed slightly higher false assignment frequencies, whereas DIA-based modes maintained tighter control. Given that the background matrix consisted of human fecal material, we next examined how acquisition strategy influenced taxonomic and functional representation. Genus- and species-level profiles (Fig. 2D) and functional annotations based on EC numbers and KEGG pathways (Fig. 2E) mirrored the gradient-dependent trends observed for peptide identifications. The number of confidently assigned taxa (≥3 taxa-specific peptides) and functional entries increased with gradient length but differed only marginally between 22- and 45-min runs, indicating that near-maximal profiling depth was already achieved at 22-min. The recovered community composition was consistent with previous fecal metaproteome reports (Supplementary Fig. 1G). Although DDA-PASEF detected a similar number of species as DIA-based methods (Fig. 2D), peptide-level inspection revealed that species shared across methods were supported by substantially fewer peptides in DDA datasets (Fig. 2F). This suggests lower annotation confidence and reduced taxonomic coverage for DDA-derived assignments, particularly compared with DIA- and Slice-PASEF. A comparable pattern was observed at the functional level: analysis of protein support per EC entry (Fig. 2G) showed that, even at the 5-min gradient, DIA-, Slice- and Synchro-PASEF annotated many ECs with >20 proteins, whereas DDA- and midia-PASEF annotated far fewer. Moreover, DIA-based methods displayed a larger fraction of ECs supported by >50 proteins, indicating superior capacity to define enzymatic functions through multiple converging protein evidences. This broader protein-to-function mapping effectively enhances the functional detection limit in complex metaproteomic samples. Quantitative precision and accuracy across PentaPASEF strategies Building on the qualitative benchmarks, we next compared the quantitative performance of the PASEF strategies under identical analytical conditions. Quantitative precision was assessed by calculating the coefficient of variation (CV) of peptide intensities across acquisition replicates (n = 3). Intra-group Pearson correlations were high for most datasets and gradients (Supplementary Fig. 2A). The few lower correlations observed for Slice-PASEF at 22 min (r < 0.9) originated from a single replicate (Supplementary Data 3) rather than a systematic effect. Across all experiments, the majority of quantified peptides showed CVs below 20% (Fig. 3A), indicating overall high reproducibility of both the LC–MS platform and the acquisition schemes. Interestingly, DDA-PASEF displayed a pronounced gradient dependence, with variability decreasing at longer separations, whereas DIA-based methods were less affected by gradient length. Among all strategies, Slice-PASEF consistently achieved the lowest median CVs, including at the 5-min gradient, demonstrating that high quantitative precision can be retained under high-throughput conditions. When analyzed separately for the spike-in species (Fig. 3A; Supplementary Fig. 2B), all methods showed good precision at the highest input (150 pg S. ruber , 50 pg L. murinus ). At the medium level, performance declined first for DDA-PASEF, followed by Synchro- and midia-PASEF, particularly for 5 pg L. murinus . In contrast, Slice-PASEF largely preserved its precision under these conditions. At the lowest inputs (1.5 pg S. ruber, 0.5 pg L. murinus), variability increased markedly for all methods, indicating current limits of reproducible quantification below the picogram range. Quantitative accuracy was evaluated by comparing log₂ intensity ratios between high and medium spike-in conditions (Fig. 3B; Supplementary Data 4). For the human fecal background, the log₂ ratios were largely centered around zero across all methods and gradients, indicating consistent quantification for the majority of background peptides. Spike-in peptides showed clear separation consistent with the expected 10-fold difference. Under the 5-min gradient, all methods exhibited some ratio compression, most prominently for DDA-PASEF. For L. murinus , DIA-, Slice- and Synchro-PASEF clustered closely around the theoretical log₂ ratio of 3.3 at 22- and 45-min gradients, whereas DDA-PASEF showed broader dispersion and systematic underestimation. S.ruber peptides followed a similar pattern, with most DIA-based methods maintaining distributions closely centered near the expected ratio, while DDA-PASEF exhibited broader variability and stronger ratio compression. Among all methods, Slice-PASEF showed the narrowest ratio distributions and best agreement with theoretical expectations across gradients, indicating superior quantitative accuracy. To assess quantitative consistency at the microbial compositional level, all quantified taxa were stratified into Top, Medium and Low abundance tiers, and the CV of their relative abundance ranks across replicates (n = 9) was used as a measure of stability. Overall, DIA-based methods exhibited markedly tighter rank variability than DDA-PASEF across gradients and abundance tiers (Fig. 3C; Supplementary Data 5), consistent with the beta-diversity analysis where DDA showed larger within-method dispersion (Supplementary Fig. 2C). Notably, the Top tier did not always display the lowest variability—particularly at 5 min—whereas several DIA-based modes maintained stable reproducibility even in Medium and Low tiers (Fig.3C). We next examined the ranks of the spike-in species across conditions. As expected, both S. ruber and L. murinus showed a progressive decrease in rank with lower spike-in amounts, consistent with their relative spike-in levels (Fig. 3D). Overall, the rank patterns were largely comparable among most PASEF methods at each gradient, indicating similar quantitative scaling across acquisition strategies. However, ranking deviations were observed at low inputs, particularly for DDA-PASEF, which occasionally yielded disproportionately high ranks at low inputs—for example, ranking S. ruber at the 35th in the 22-min gradient, while other methods placed it behind the 120th. At the medium input, L. murinus ranked near 80th in DDA- and DIA-PASEF but around 130th in Slice-, Synchro- and midia-PASEF, and these differences largely diminished at the 45-min gradient, suggesting that longer separation times improve rank agreement among methods at moderate abundance levels (Fig. 3D). Across conditions, rank patterns were most consistent among DIA- and Slice-PASEF acquisitions, whereas DDA-PASEF frequently deviated at low input levels. At the functional layer, acquisition strategy strongly influenced quantitative depth. DIA-based methods, especially Slice-PASEF, exhibited a broader EC dynamic range and higher intensities at comparable abundance ranks (Fig. 3E; Supplementary Data 6), whereas DDA- and midia-PASEF showed weaker signals in low-abundance regions, suggesting reduced detection depth of enzymatic functions. This pattern was consistent across gradients, with minimal improvement beyond 22 min. When assessing functional quantification reproducibility, DIA-based methods again showed superior performance, characterized by steeper ECDF curves and a higher proportion of EC entries with CVs below 20%. Slice-PASEF consistently exhibited the most compact CV distributions across all abundance tiers (Fig. 3F), demonstrating its capacity to maintain precise functional quantification even for low-abundance ECs. DIA- and Slice-PASEF discover concordant microbial and host responses in a colon-injury mouse model DIA- and Slice-PASEF showed the best overall performance in the benchmarking experiment. We next investigated their ability to discover molecular mechanisms in an experimental design that integrates not only the complexity of a fecal microbiome but also the biological variability of an in vivo setup. Thus, we applied both acquisition strategies on a murine colon injury model created by the conditional knock-out of Hsp60 in intestinal epithelial cells combined with the presence or absence of Interleukin 10 (Il-10) expression (Fig. 4A). Conditional deletion of Hsp60 induces mitochondrial stress, leading to mucosal injury and microbial dysbiosis, while additional IL-10 deficiency impairs epithelial regeneration in the distal colon, resulting in persistent inflammatory pathology 1, 23 . Across all conditions, DIA- and Slice-PASEF identified a similar total number of peptides (Supplementary Fig. 3A; Supplementary Data 7) and number of total bacterial species detected (Supplementary Fig. 3B; Supplementary Data 8). Alpha- and beta-diversity profiles (Supplementary Fig. 3C and Fig. 4B) show how both acquisition strategies discovered analogous injury-associated compositional gradients concomitant to the tissue phenotype separation between proximal and distal colon in Hsp60 Δ/ΔIEC ;Il10 −/− mice (e.g., different clustering only at D14 in the distal colon in Fig. 4B). While species-level abundances quantified by DIA- and Slice-PASEF were highly correlated (Pearson r > 0.91; Supplementary Fig. 3D), Slice-PASEF exhibited a systematic upward shift in species abundance distributions (Fig. 4C). This shift was not driven by differences in the number of quantified species (Supplementary Fig. 3B) or species-specific peptides (Supplementary Fig. 3E). Instead, it reflected a consistently higher peptide-level intensities in Slice-PASEF (Fig. 4D), likely resulting from denser MS/MS sampling intrinsic to the Slice acquisition and its propagation into MaxLFQ-based quantification. In fact, Slice-PASEF generated substantially larger raw file sizes and required longer DIA-NN processing times (Supplementary Fig. 3F). Minimum detectable species abundances were ~0.003% for both methods, with Slice-PASEF reaching slightly lower limits in both distal (0.0031% vs. 0.0034% in DIA) and proximal samples (0.0023% vs. 0.0037% in DIA) (Supplementary Data 9). Differential abundance analysis revealed that the majority of significantly regulated species were shared between DIA- and Slice-PASEF across all comparisons and timepoints (Fig. 4E; Supplementary Data 10). Temporal patterns of differential species discovered the underpinnings correlating to the previously observed tissue pathology 23 (Fig. 4E). In the Hsp60 Δ/ΔIEC ;Il10 +/+ (injury at D4 and D8 but histopathological regeneration at D14) vs Hsp60 fl/fl ;Il10 −/− (no colitis at the timepoint in the absence of Hsp60 deletion 23 ) contrast, the number of differential species markedly decreased by D14 in both colonic regions, consistent with epithelial regeneration due to the regain of Hsp60 protein expression in the conditional Hsp60 Δ/ΔIEC ;Il10 +/+ mice at D14. In contrast, the comparison between Hsp60 Δ/ΔIEC ;Il10 −/− (injury at D4 and D8 without histopathological regeneration at D14 in the distal colon 23 ) and Hsp60 fl/fl ;Il10 −/− (no colitis at the timepoint in the absence of Hsp60 deletion 23 ) revealed a larger set of significant species at D14 in the distal colon compared to D4 and D8. These findings offer a new layer of information based on the ecosystem functional expression, which both increases the sensitivity, compared to imaging alternatives, to profile pathological changes, and offers a plausible explanation for the recovery dynamics observed in these mice. Interestingly, each method also detected a subset of unique significant species (Fig. 4E). Inspection of raw p-values revealed that many method-unique species were nominally significant in the alternative method but failed multiple-testing correction (Supplementary Fig. 3G; Supplementary Data 10), indicating that uniqueness frequently reflected borderline statistical behavior rather than discordant biology. Direct comparison of Δ|log2FC| and Δ(−log10 p-value) further showed that method-unique calls aligned with the method exhibiting both larger effect sizes and stronger statistical evidence (Supplementary Fig. 3H), suggesting that uniqueness is primarily driven by method-dependent effect size estimation rather than increased measurement variability or statistical noise. Despite partial species-level uniqueness, pathway-resolved analysis revealed strong functional concordance between DIA- and Slice-PASEF (Fig. 4F). Across contrasts and timepoints, both methods identified highly overlapping sets of significantly regulated species–KEGG associations, comprising 182 shared KEGG pathways linked to 175 shared species (Supplementary Data 11). In the Hsp60 Δ/ΔIEC ;Il10 −/− vs Hsp60 fl/fl ;Il10 −/− contrast, both methods consistently detected similar sets of inflammation-associated (e.g., NOD-like receptor signaling), stress-response metabolic (e.g., pentose phosphate and methane metabolism), and regeneration-associated pathways (e.g., biofilm formation, quorum sensing). A detailed analysis revealed how both acquisition strategies identify comparable species-level drivers underlying particular functional changes, as well as site-specific functional strategies of specific species. For example, in biofilm formation, DIA- and Slice-PASEF assigned highly similar sets of species concurrent with the lack of functional remodeling in the distal colon (Fig. 4G). Interestingly, both strategies discovered several species (e.g., Eubacterium plexicaudatum , Butyribacter intestini , and Jutongia huaianensis ) in the proximal colon as associated with decreased biofilm activity, and other species (e.g., Ligilactobacillus murinus and Anaerotruncus colihominis ) in the distal colon associated with increased biofilm formation. A similar concordance and site-specification were also observed for the two-component system pathway (Supplementary Fig. 3I). Analysis of the host proteome revealed interesting functional alteration patterns aligning to the above-described microbiome alterations. We observed highly concordant patterns of protein regulation between DIA- and Slice-PASEF (Supplementary Data 12), with both methods capturing maximal host protein dysregulation at D8, corresponding to the peak of tissue injury detected histopathologically as described previously 23 (Fig. 4H, Hsp60 Δ/ΔIEC ;Il10 −/− vs Hsp60 fl/fl ;Il10 −/− ). Moreover, a larger number of proteins remained significantly altered at D14 in the distal compared with the proximal colon. Both methods consistently identified enrichment of epithelial repair and immune defense pathways, supporting concordant biological interpretation at the host level. Proteins involved in hemostasis (e.g., Fga, Fgb, and Fgg) and epithelial membrane repair and restitution (e.g., Anxa1, Anxa2, Anxa5) were more prominently upregulated in the proximal colon (Fig. 4I, Supplementary Fig. 3J), strongly suggesting accelerated tissue regeneration relative to the distal colon. Discussion The growing diversity of PASEF acquisition strategies has created both opportunity and uncertainty for metaproteomics 18, 20, 27 , as method choice increasingly determines analytical depth, quantitative reach, and biological interpretability. By benchmarking five PASEF implementations within a unified experimental and computational framework, our study reveals that acquisition strategy is not a neutral technical decision but a primary driver of metaproteomic sensitivity, reproducibility, and functional resolution. This study provides a comprehensive and biologically grounded evaluation of five PASEF acquisition strategies across chromatographic regimes, sample input levels, and analytical depths in fecal metaproteomes. By integrating qualitative, quantitative, and biological validation analyses, our results establish a unified framework for understanding how acquisition strategy selection influences identification depth, quantitative performance, and downstream biological interpretation in complex microbiome samples. Across all acquisition modes, peptide and protein identifications scaled predictably with chromatographic separation length, confirming gradient duration as a primary determinant of analytical depth. DIA-based strategies consistently achieved broader and more stable peptide coverage than DDA-PASEF, including under high-throughput conditions (Fig. 2A-B). This advantage extended to the detection of low-abundance microbial peptides and functional annotations, where DIA-based acquisitions provided enhanced dynamic range and broader protein support per enzymatic entry (Fig. 2F-G). However, these gains come with practical trade-offs. Slice-PASEF generated substantially larger raw data files (Supplementary Fig. 3F) than other acquisition modes, increasing storage demands and computational burden during data processing. In addition, midia-PASEF currently requires proprietary licensing, which may limit accessibility and adoption despite its demonstrated performance 14 . These considerations highlight that acquisition strategy selection should balance analytical performance against data management capacity and software availability. Quantitative analyses revealed marked differences between acquisition modes. Slice-PASEF achieved the most favorable balance between sensitivity and quantitative robustness, yielding consistently low coefficients of variation, minimal ratio compression and stable species-level abundance ordering across spike-in series (Fig. 3). DIA-PASEF also maintained robust quantitative performance but exhibited slightly higher variability under low-input conditions (Fig. 3A) and narrower dynamic ranges of quantified functions (Fig. 3E). In contrast, DDA-PASEF showed increased variability especially under high throughput conditions, ratio compression and rank instability (Fig. 3B-C), underscoring inherent limitations of precursor-dependent sampling in dense fecal peptide matrices. Our accuracy assessment was based on controlled spike-in ratios of two bacterial species and relative abundance rankings, which capture internal quantitative fidelity but do not provide absolute concentration ground truth. The absence of comprehensive isotopically labeled standards across the entire microbial dynamic range limits direct calculation of absolute quantitative accuracy, and future work incorporating such standards would further refine sensitivity limits and bias estimates. Methodological differences observed in benchmarking are directly translated into biological interpretation. In a murine colonic epithelial injury model, DIA- and Slice-PASEF captured highly concordant microbial and host response patterns, which indicates that the acquisition strategy primarily modulates quantitative reach rather than biological directionality. More importantly, DIA- and Slice-PASEF datasets provided evidence, to an extent, aligning with the tissue pathology previously reported 23 for the same samples. Nevertheless, the biological validation was conducted in a single disease model without an independent molecular ground truth for taxonomic change, pathway activation, or host response. Consequently, while relative concordance between methods can be assessed, the absolute biological correctness of specific response magnitudes cannot be definitively established, especially for method-dependent findings. Additional validation using orthogonal assays or controlled perturbation models will be required to fully define biological accuracy. Although fecal samples represent one of the most compositionally complex and analytically demanding metaproteomic matrices, they do not capture the full diversity of microbiome-associated environments. Microbiomes from environmental 28, 29 , oral 30, 31 , or skin 32, 33 niches may differ in peptide complexity, dynamic range, and interference patterns, which could influence relative method performance. Importantly, for each acquisition strategy, we applied standardized and widely adopted parameter settings rather than exhaustively optimizing all method-specific acquisition parameters. Apart from exploratory testing of Slice-PASEF frame densities and midia-PASEF window overlap configurations (Supplementary Fig. 1A), acquisition windows, duty cycles, and isolation schemes were not systematically tuned to maximize performance for each individual method. Consequently, absolute performance ceilings for some strategies may be slightly higher than those observed here, and future targeted optimization could further refine sensitivity, throughput, and quantitative behavior. Our results indicate that DIA-based PASEF strategies—particularly DIA- and Slice-PASEF—define a new operational regime for metaproteomics in which high-throughput workflows can simultaneously achieve deep coverage, stable quantification, and expanded functional detectability. Together, these findings establish a quantitative reference for selecting and refining PASEF strategies and provide a foundation for future development of acquisition designs, software tools and standards that will enable reproducible, sensitive and scalable metaproteomics. Methods E thics Human fecal sampling carried out at the University of Vienna was under the ethical approval (01149). Mouse work carried out at the Technical University of Munich, as well as maintenance and breeding of mouse lines, were approved by the Committee on Animal Health Care and Use of the state of Upper Bavaria (Regierung von Oberbayern; AZ ROB-55.2-2532.Vet_02-14-217, AZ ROB-55.2-2532.Vet_02-20-58, AZ ROB-55.2-2532.Vet_02-18-37) and performed in strict compliance with the EEC recommendations for the care and use of laboratory animals (European Communities Council Directive of November 24, 1986 (86/609/EEC)). Animals and housing conditions Mice for in vivo experiments (Fig. 4) were male and housed under specific pathogen-free (SPF) conditions according to the criteria of the Federation for Laboratory Animal Science Associations (FELASA) (12-hour light/dark cycles at 24–26°C) in the mouse facility at the Technical University of Munich (School of Life Sciences Weihenstephan). All mice received a standard diet (autoclaved V1124-300, Ssniff) ad libitum , autoclaved water and were sacrificed by CO 2 or isoflurane. Details of the animal models can be found in our previous study 23 . Briefly, Hsp60 fl/fl mice and Hsp60 fl/fl x VillinCreER T2-Tg mice were generated as described previously 34 to create IEC-specific Hsp60 knockout mice via tamoxifen induction (Hsp60 Δ/ΔIEC ;Il10 +/+ ). In addition, Hsp60 fl/fl andHsp60 Δ/ΔIEC ;Il10 +/+ mice were crossed with whole body Il-10 knockout mice (Il10 -/- ) to generate Hsp60 fl/fl ;Il10 -/- andHsp60 Δ/ΔIEC ;Il10 -/- mice 23 .For conditional Hsp60 deletion,mice and appropriate control mice (6-weeks of age) were kept on phytoestrogen-reduced diet 1005 (V1154-300, Ssniff) for four weeks under SPF conditions. Afterwards, mice received 400mg tamoxifen citrate per kg chow feed (CreActive T400 (10mm, Rad), Genobios) ad libitum for 7 days. After the induction phase, tamoxifen diet was replaced with the phytoestrogen-reduced diet. During and after the induction phase, mice were monitored daily and aborted when a combined score considering weight loss, changes in stool consistency, general behaviour, and general state of health was reached. Animals were sacrificed at the indicated time points (D4, D8, and D14 after tamoxifen diet). All mice and their respective genotypes were generated and maintained on an in-house crossing of C57Bl/6N and C57Bl/6J background. Protein extraction and SP3-assisted protein digestion for metaproteomics analysis The procedures from protein extraction of gut microbiome material to final peptide preparation were performed as previously described 1, 18 . Liquid chromatography-mass spectrometry configurations Nanoflow reversed-phase liquid chromatography (Nano-RPLC) was performed on either ProElute or NanoElute2 systems (Bruker Daltonik, Bremen, Germany) coupled with timsTOF Ultra2 (Bruker Daltonik, Bremen, Germany) via CaptiveSpray ion source, respectively. Mobile solvent A consisted of 100% water containing 0.1% FA and mobile phase B of 100% acetonitrile containing 0.1% FA. Liquid chromatogram setups Results presented in Fig. 1-Fig. 3 were generated with three LC gradients on the ProElute system. For 5-min gradient separation, peptides were loaded onto a PepSep® column (4 cm x 150 µm) packed with 1.9 µm ReproSil C18 particles (Bruker). The mobile phase B was linearly increased from 2 to 35% in 5 minutes with a flowrate of 0.3 µL/min, followed by a steep increase to 95% in 0.1 minute. The mobile phase B was maintained at 95% for the last 3 minutes 0.3 µL/min. For 22-min gradient separation, peptides were loaded onto a PepSep® column (25 cm x 75 µm) packed with 1.5 µm ReproSil C18 particles (Bruker). The mobile phase B was linearly increased from 2 to 20% in 17 minutes with a flowrate of 0.3 µL/min, followed by another linear increase to 35% within 5 minutes and a steep increase to 95% in 0.5 minute. The mobile phase B was maintained at 95% for the last 7.5 minutes 0.3 µL/min. For 45-min gradient separation, peptides were loaded onto a PepSep® column (25 cm x 75 µm) packed with 1.5 µm ReproSil C18 particles (Bruker). The mobile phase B was linearly increased from 2 to 20% in 35 minutes with a flowrate of 0.3 µL/min, followed by another linear increase to 35% within 10 minutes and a steep increase to 95% in 1 minute. The mobile phase B was maintained at 95% for the last 4 minutes 0.3 µL/min. Mouse colonic microbiome samples (Fig. 4) were analyzed on a NanoElute2 system coupled with an Aurora TM ULTIMATE column (25 cm x 75 µm) packed with 1.7 µm C18 particles (IonOpticks, Fitzroy, Australia) with a 45-min gradient. The mobile phase B was linearly increased from 2 to 20% in 35 minutes with a flowrate of 0.25 µL/min, followed by another linear increase to 35% within 10 minutes at a flowrate of 0.25 µL/min and then a steep increase to 95% in 0.5 minute with flowrate increasing to 0.4 µL/min. The mobile phase B was maintained at 95% for the last 4.5 minutes 0.4 µL/min. PentaPASEF method setups Different PASEF methods can be extacted from the raw data uploaded. A summary report of all methods with key parameters is provided in Supplementary Data 13. Microbial database construction To analyze the human fecal samples, DDA-PASEF files generated in Fig. 1 were submitted to MSfragger 35 (version 4.1) integrated in FragPipe computational platform (version 22.0), searching against the MGnify human gut protein catalogue (https://www.ebi.ac.uk/metagenomics/genome-catalogues/human-gut-v2-0-2) 36 . The decoy database was generated with reversed sequences. Trypsin was specified with a maximum of two missed cleavages allowed. The search included variable modifications of methionine oxidation and N-terminal acetylation and a fixed modification of carbamidomethyl on cysteine. The mass tolerances of 20 ppm were set for precursor and fragment. Peptide length was set to 7 to 50 amino acids with a mass range from 500 to 5000 Da. The remaining parameters were kept as default settings. During the validation, MSBooster (version 1.2.31) was used for rescoring and Percolator 37 (version 3.6.5, default parameters) was used for PSM validation. FDR level was set to 1% for PSM, peptide and protein. The identified proteins from the search formed a sample-specific protein database, containing 30854 protein sequences. To generate a sample-specific microbial database for the colonic samples (Fig. 4), we followed the recently published strategy of novoMP 1 . Briefly, a total of 60 µg pooled peptides (from both proximal and distal regions) were fractionated (Fisher Scientific, Cat. 84868) according manufactures instruction. Eight peptide factions were dried using vacuum centrifugation and then re-suspended in 30 µL of MS-grade water. The peptide concentration was measured in duplicate using NanoPhotometer N60 (Implen, Munich, Germany) at 205 nm. Peptide samples were acidified with formic acid to a final concentration of 0.1% and were stored at -20°C until LC-MS/MS analysis. 50 ng of each peptide fraction (a total of 8 fractions) were loaded on an Aurora TM ULTIMATE column (25 cm x 75 µm) packed with 1.6 µm C18 particles (IonOpticks, Fitzroy, Australia) with a total separation time of 60 minutes in DDA-PASEF mode. The TIMS analyzer was operated in a 100% duty cycle with equal accumulation and ramp times of 100 ms each. Specifically, 10 PASEF scans were set per acquisition cycle (cycle time of 1.17 s) with ion mobility range from 0.7 to 1.3 (1/k0). The target intensity and intensity threshold were set to 20000 and 500 respectively. Dynamic exclusion was applied for 0.4 minutes. Ions with m/z between 100 and 1700 were recorded in the mass spectrum. Collision energies were dependent on ion mobility values with a linear increase in collision energy from 1/K0 = 0.6 Vs/cm² at 20 eV to 1/K0 = 1.6 Vs/cm² at 63 eV. Fractionated DDA files were submitted to MSfragger 35 (version 4.3) integrated in FragPipe computational platform (version 23.1), searching against the MGnify mouse gut protein catalogue (https://www.ebi.ac.uk/metagenomics/genome-catalogues/mouse-gut-v1-0). The decoy database was generated with reversed sequences. Trypsin was specified with a maximum of two missed cleavages allowed. The search included variable modifications of methionine oxidation and N-terminal acetylation and a fixed modification of carbamidomethyl on cysteine. The mass tolerances of 20 ppm were set for precursor and fragment. Peptide length was set to 7 to 50 amino acids with a mass range from 500 to 5000 Da. The remaining parameters were kept as default settings. During the validation, MSBooster (version 1.3.17) was used for rescoring and Percolator 37 (version 3.7.1, default parameters) was used for PSM validation. FDR level was set to 1% for PSM, peptide and protein. The identified proteins from the search formed a sample-specific protein database, containing 97249 protein sequences. In addition, DDA raw files were subjected to de novo sequencing as previously described 1 . As a result, a total of 232863 protein sequences were included in the microbial database. This database was combined with the standard Mus musculus proteome (https://www.uniprot.org/proteomes/UP000000589, accessed on 2023-04-27) to process acquired datasets. Spectral library generation for DIA-NN analysis Spectral libraries for datasets acquired using DIA-, Slice-, Synchro- and midia-PASEF were generated using DIA-NN (v2.0.2) in library-free mode based on in silico–predicted spectra. Protein databases used for library prediction included a reduced microbial reference database, the standard proteomes of Ligilactobacillus murinus (UniProt UP000051612, accessed 2023-07-19), Salinibacter ruber (UP000008674, accessed 2023-07-19), Homo sapiens (UP000005640, accessed 2025-03-20), and contaminant sequences provided by DIA-NN. Peptides were generated by in silico digestion using Trypsin/P with up to two missed cleavages and N-terminal methionine excision enabled. Carbamidomethylation of cysteine was set as a fixed modification, while methionine oxidation, N-terminal acetylation, and heavy isotopic labeling of arginine (+10.0083 Da) and lysine (+8.0142 Da) were specified as variable modifications, allowing a maximum of two variable modifications per peptide. Peptide lengths were restricted to 7–30 amino acids. Precursor m/z values were limited to 250–1200 with charge states of 2–4, and fragment ions were considered in the range of 100–1700 m/z. DIA-NN’s deep learning models were used to predict fragment intensities, retention times, and ion mobility values. Proteotypicity was set to “Protein names (from FASTA),” and heuristic protein inference was disabled. mass accuracy for MS1 and MS2 was set to automatic determination. The resulting predicted library comprised 12,412,855 precursors corresponding to 109,255 protein groups. To prevent incorrect retention-time alignment, DIA-NN searches were performed separately for each acquisition method and gradient length. For the colonic samples presented in Fig. 4, pooled peptide samples (generated by combining all proximal or distal colon samples) were repeatedly analyzed using both DIA- and Slice-PASEF to monitor LC–MS performance and to support generation of a sample-specific experimental spectral library. In addition, fractionated pooled samples were acquired in DIA-PASEF mode to further increase spectral coverage. These datasets were processed using DIA-NN (v2.3.0) with a predicted library comprising 26,374,776 precursors derived from a reduced microbial database (232,863 protein sequences), the standard Mus musculus proteome (UniProt UP000000589, accessed 2023-04-27), and DIA-NN contaminant sequences. Trypsin/P digestion was specified with a maximum of one missed cleavage and N-terminal methionine excision enabled. Carbamidomethylation of cysteine was used as a fixed modification, while methionine oxidation and N-terminal acetylation were allowed as variable modifications, permitting a maximum of one variable modification per peptide. Peptide length was restricted to 7–30 amino acids, precursor m/z values were set to 250–1200 with charge states of 2–3, and fragment ions were considered between 100 and 1700 m/z. Proteotypicity was set to “Isoform IDs,” and protein inference was disabled. The resulting experimental spectral library contained 184,884 precursors corresponding to 85,044 protein groups and was used for DIA- and Slice-PASEF analysis of individual samples. For all searches conducted in DIA-NN, RT-dependent cross-run normalization and QuantUMS 38 (high precision) options were selected for quantification. All DIA-NN search outputs were further processed with the R package, DIA-NN (https://github.com/vdemichev/diann-rpackage), to calculate the MaxLFQ 39 quantitative intensities for all identified peptides and protein groups with q-value ≤ 0.01 as criteria at precursor and protein group levels. Sample preparation for PentaPASEF benchmarking The human fecal background was generated by pooling fecal peptide samples produced in-house from multiple independent studies to ensure consistency and maximize proteomic complexity. The pooled fecal peptide mixture was diluted with 0.15% (w/v) n-dodecyl-β-D-maltoside (DDM; Sigma-Aldrich) to the desired concentrations. Ligilactobacillus murinus (DSM 20452) and Salinibacter ruber (DSM 13855) were obtained from DSMZ (Braunschweig, Germany), and bacterial culture conditions were as previously reported 1 . Peptides derived from the two species were mixed at a fixed ratio of 1:3 ( L. murinus : S. ruber ) and subsequently diluted with 0.15% DDM to generate the required spike-in levels. Owing to the extended acquisition period, samples corresponding to different spike-in concentrations were further aliquoted and stored at −20 °C. For each acquisition method and LC gradient, a fresh aliquot was thawed immediately before analysis to minimize sample loss and variability associated with prolonged storage in the autosampler. Taxonomic and functional annotation and quantification Taxonomic annotation was done using Unipept web application 40 with default settings. peptides present in the peptide quantification output of each method and gradient were submitted for annotation and the resulting taxa assignments were merged back to the quantification matrix where peptide intensities from acuqisition triplicates were averaged. Taxa were considered confident when supported by at least three taxa-specific peptide entries, and taxa abundances were quantified by summing intensities of taxa-assigned peptides. The microbial protein databases used in this manuscript were annotated using EggNOG-mapper 41 (http://eggnog-mapper.embl.de/) with default settings to retrieve potential functions and pathways. The annotated function was merged with the peptide quantification matrix using Meta4P 42 . The quantification of a functional entry was done by summing up the peptide intensities where the corresponding proteins were annotated to this functional entry. For the in vivo dataset, peptides identified by DIA- and Slice-PASEF were annotated separately using Unipept and merged with the corresponding peptide quantification matrices. Confident taxa required at least three taxa-specific peptide entries, and taxa intensities were computed by summing peptide intensities after applying a content-wise detection filter, retaining peptides only if they were detected in all replicates of at least one group (colonic region × genotype × timepoint). The taxon-specific functions were analyzed using Meta4P with peptide quantification data from DIA-NN, confident taxonomic annotation, and functional annotation files from EggNOG-mapper as inputs. Quantification of taxon-specific functions was performed by summing the peptide intensities assigned to taxa and associated with specific functions. Statistical analysis Pairwise comparisons presented in Supplementary Fig. 3 were performed using a two-sided Wilcoxon rank-sum test (Mann–Whitney U test) in R (version 4.5.2). Differential abundance analysis of species, species–pathway, and host proteins (Fig. 4) were performed using the limma (version 3.66.0) framework. For each colonic content and time point, log₂-transformed abundances were modeled using linear models, genotype contrasts were tested using empirical Bayes–moderated t-statistics with robust variance estimation, and multiple testing correction was applied independently for each contrast within each colonic content and time point using the Benjamini–Hochberg procedure. Declarations Data Availability The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD073779 (reviewers can access the dataset by logging in to the PRIDE website using [email protected] , sassword: rIphcfnHZd7p) and PXD073688 (reviewers can access the dataset by logging in to the PRIDE website using [email protected] , sassword: lYIVr7eVH5US). Source data are provided with this paper. Code Availability The codes essential for the study have been deposited to Github repository Acknoledgements We thank all members of the Systems Biology of Pain laboratory, Division of Pharmacology and Toxicology, University of Vienna, for valuable discussions and suggestions. We thank the kind help from Prof. Stefan Tenzer’s group (University Medical Center of the Johannes-Gutenberg University Mainz) on our midia-PASEF data. We thank Biognosys AG for the access to Spectronaut. This research was funded in part by the University of Vienna and by the Austrian Science Fund (FWF; 10.55776/P35856 and 10.55776/P36554 to M.S.). E.U., D.A., and D.H. were funded by Deutsche Forschungsgemeinschaft (DFG; 395357507 and 469152594). For open access purposes, the author has applied a CC BY public copyright license to any author accepted manuscript version arising from this submission. The computational results presented have been achieved in part using the Austrian Scientific Cluster (ASC). Author Contributions Statement Conceptualization: D.G.V.; Experimental design: D.G.V., F.X., G.M., R.K.R.K., D.A., E.U., and D.H.; Biochemistry and mass spectrometry: F.X., R.K.R.K., and G.M.; Sample collection and preparation: F.X., R.K.R.K., E.U., and D.A.; Data processing: F.X., L.U., and R.K.R.K.; Formal analysis: F.X.; Writing: F.X., and D.G.V.; Study supervision: D.G.V.; Project administration: M.S. and D.G.V.; Funding acquisition: M.S. and D.G.V. All authors edited and approved the final manuscript. Competing Interests Statement M.S. received research awards and travel support from the German Pain Society (DGSS) both of which were sponsored by Astellas Pharma GmbH (Germany). M.S. received research awards from the Austrian Pain Society. M.S. received a one-time consulting honorarium from Grunenthal GmbH (Germany). None of these sources influenced the content of this study, and M.S. declares no conflict of interest. D.G.V. and M.S. have an ongoing scientific collaboration with Bruker (Center of Excellence for Metaproteomics University of Vienna - Bruker). F.X., R.K.R.K., L.U., E.U., D.A., D.H., M.S., and D.G.V. declare no competing interests. G.M. is an employee of Bruker Austria. Reference Xian F , et al. Ultra-sensitive metaproteomics redefines the dark metaproteome, uncovering host-microbiome interactions and drug targets in intestinal diseases. Nature Communications 16 , 6644 (2025). Li L , et al. Systematic metaproteomics mapping reveals functional and ecological landscapes of Ex vivo human gut microbiota responses to therapeutic drugs. Nature Communications 16 , 9383 (2025). Wang J, Lang T, Shen J, Dai J, Tian L, Wang X. Core Gut Bacteria Analysis of Healthy Mice. Front Microbiol 10 , 887 (2019). Armengaud J. Metaproteomics to understand how microbiota function: The crystal ball predicts a promising future. Environmental Microbiology 25 , 115-125 (2023). Clemente JC, Ursell LK, Parfrey LW, Knight R. The impact of the gut microbiota on human health: an integrative view. Cell 148 , 1258-1270 (2012). Stamboulian M, Canderan J, Ye Y. Metaproteomics as a tool for studying the protein landscape of human-gut bacterial species. PLoS Comput Biol 18 , e1009397 (2022). Duan H , et al. Assessing the Dark Field of Metaproteome. Anal Chem 94 , 15648-15654 (2022). Dumas T , et al. The astounding exhaustiveness and speed of the Astral mass analyzer for highly complex samples is a quantum leap in the functional analysis of microbiomes. Microbiome 12 , 46 (2024). Kleikamp HBC , et al. Database-independent de novo metaproteomics of complex microbial communities. Cell Syst 12 , 375-383.e375 (2021). Meier F , et al. Parallel Accumulation-Serial Fragmentation (PASEF): Multiplying Sequencing Speed and Sensitivity by Synchronized Scans in a Trapped Ion Mobility Device. J Proteome Res 14 , 5378-5387 (2015). Meier F , et al. diaPASEF: parallel accumulation-serial fragmentation combined with data-independent acquisition. Nat Methods 17 , 1229-1236 (2020). Sinn LR , et al. Slice-PASEF: Maximising Ion Utilisation in LC-MS Proteomics. bioRxiv , (2025). Skowronek P , et al. Synchro-PASEF Allows Precursor-Specific Fragment Ion Extraction and Interference Removal in Data-Independent Acquisition. Molecular & Cellular Proteomics 22 , 100489 (2023). Distler U , et al. midiaPASEF maximizes information content in data-independent acquisition proteomics. bioRxiv , 2023.2001.2030.526204 (2023). Steigenberger B , et al. Benefits of Collisional Cross Section Assisted Precursor Selection (caps-PASEF) for Cross-linking Mass Spectrometry. Mol Cell Proteomics 19 , 1677-1687 (2020). Gomez-Zepeda D , et al. Thunder-DDA-PASEF enables high-coverage immunopeptidomics and is boosted by MS(2)Rescore with MS(2)PIP timsTOF fragmentation prediction model. Nat Commun 15 , 2288 (2024). Mendes ML, Borrmann KF, Dittmar G. Eleven shades of PASEF. Expert Review of Proteomics 21 , 367-376 (2024). Gomez-Varela D, Xian F, Grundtner S, Sondermann JR, Carta G, Schmidt M. Increasing taxonomic and functional characterization of host-microbiome interactions by DIA-PASEF metaproteomics. Front Microbiol 14 , 1258703 (2023). Wang A , et al. Assessing fecal metaproteomics workflow and small protein recovery using DDA and DIA PASEF mass spectrometry. Microbiome Research Reports 3 , 39 (2024). Rajczewski AT , et al. Data-Independent Acquisition Mass Spectrometry as a Tool for Metaproteomics: Interlaboratory Comparison Using a Model Microbiome. Proteomics 25 , e202400187 (2025). Samodova D , et al. Salivary proteomics and metaproteomics identifies distinct molecular and taxonomic signatures of type-2 diabetes. Microbiome 13 , 5 (2025). Xian F , et al. Ultra-sensitive metaproteomics redefines the dark metaproteome, uncovering host-microbiome interactions and drug targets in intestinal diseases. Nat Commun 16 , 6644 (2025). Urbauer E , et al. Mitochondrial perturbation in the intestine causes microbiota-dependent injury and gene signatures discriminative of inflammatory disease. Cell Host & Microbe 32 , 1347-1364.e1310 (2024). Armengaud J. Metaproteomics to understand how microbiota function: The crystal ball predicts a promising future. Environ Microbiol 25 , 115-125 (2023). Kong AT, Leprevost FV, Avtonomov DM, Mellacheruvu D, Nesvizhskii AI. MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry–based proteomics. Nature Methods 14 , 513-520 (2017). Demichev V, Messner CB, Vernardis SI, Lilley KS, Ralser M. DIA-NN: neural networks and interference correction enable deep proteome coverage in high throughput. Nat Methods 17 , 41-44 (2020). Van Den Bossche T , et al. The microbiologist's guide to metaproteomics. Imeta 4 , e70031 (2025). Arsène-Ploetze F, Bertin PN, Carapito C. Proteomic tools to decipher microbial community structure and functioning. Environmental Science and Pollution Research 22 , 13599-13612 (2015). Nebauer DJ, Pearson LA, Neilan BA. Critical steps in an environmental metaproteomics workflow. Environmental Microbiology 26 , e16637 (2024). Baker JL, Mark Welch JL, Kauffman KM, McLean JS, He X. The oral microbiome: diversity, biogeography and human health. Nat Rev Microbiol 22 , 89-104 (2024). Yuan J , et al. OSaMPle workflow for salivary metaproteomics analysis reveals dysbiosis in inflammatory bowel disease patients. npj Biofilms and Microbiomes 11 , 63 (2025). Zhou X , et al. Longitudinal profiling of the microbiome at four body sites reveals core stability and individualized dynamics during health and disease. Cell Host & Microbe 32 , 506-526.e509 (2024). Chen Y, Knight R, Gallo RL. Evolving approaches to profiling the microbiome in skin disease. Front Immunol 14 , 1151527 (2023). Berger E , et al. Mitochondrial function controls intestinal epithelial stemness and proliferation. Nature Communications 7 , 13171 (2016). Kong AT, Leprevost FV, Avtonomov DM, Mellacheruvu D, Nesvizhskii AI. MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry-based proteomics. Nat Methods 14 , 513-520 (2017). Almeida A , et al. A unified catalog of 204,938 reference genomes from the human gut microbiome. Nature Biotechnology 39 , 105-114 (2021). Kall L, Canterbury JD, Weston J, Noble WS, MacCoss MJ. Semi-supervised learning for peptide identification from shotgun proteomics datasets. Nat Methods 4 , 923-925 (2007). Kistner F, Grossmann JL, Sinn LR, Demichev V. QuantUMS: uncertainty minimisation enables confident quantification in proteomics. bioRxiv , 2023.2006.2020.545604 (2023). Cox J, Hein MY, Luber CA, Paron I, Nagaraj N, Mann M. Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol Cell Proteomics 13 , 2513-2526 (2014). Mesuere B, Willems T, Van der Jeugt F, Devreese B, Vandamme P, Dawyndt P. Unipept web services for metaproteomics analysis. Bioinformatics 32 , 1746-1748 (2016). Cantalapiedra CP, Hernandez-Plaza A, Letunic I, Bork P, Huerta-Cepas J. eggNOG-mapper v2: Functional Annotation, Orthology Assignments, and Domain Prediction at the Metagenomic Scale. Mol Biol Evol 38 , 5825-5829 (2021). Porcheddu M, Abbondio M, De Diego L, Uzzau S, Tanca A. Meta4P: A User-Friendly Tool to Parse Label-Free Quantitative Metaproteomic Data and Taxonomic/Functional Annotations. Journal of Proteome Research 22 , 2109-2113 (2023). Additional Declarations The authors declare potential competing interests as follows: M.S. received research awards and travel support from the German Pain Society (DGSS) both of which were sponsored by Astellas Pharma GmbH (Germany). M.S. received research awards from the Austrian Pain Society. M.S. received a one-time consulting honorarium from Grunenthal GmbH (Germany). None of these sources influenced the content of this study, and M.S. declares no conflict of interest. D.G.V. and M.S. have an ongoing scientific collaboration with Bruker (Center of Excellence for Metaproteomics University of Vienna - Bruker). F.X., R.K.R.K., L.U., E.U., D.A., D.H., M.S., and D.G.V. declare no competing interests. G.M. is an employee of Bruker Austria. Supplementary Files SupplementaryInformation.pdf Supplementary Figures Cite Share Download PDF Status: Posted 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-9103948","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":605036100,"identity":"273ff661-1463-4a17-aef3-3c13a9f2d4e5","order_by":0,"name":"Feng Xian","email":"","orcid":"https://orcid.org/0000-0002-8345-0108","institution":"Systems Biology of Pain, Division of Pharmacology \u0026 Toxicology, Department of Pharmaceutical Sciences, Faculty of Life Sciences, University of Vienna","correspondingAuthor":false,"prefix":"","firstName":"Feng","middleName":"","lastName":"Xian","suffix":""},{"id":605036101,"identity":"c3cf762e-412a-4c7d-b020-854e693afab4","order_by":1,"name":"Goran Mitulovic","email":"","orcid":"","institution":"Bruker Austria GmbH","correspondingAuthor":false,"prefix":"","firstName":"Goran","middleName":"","lastName":"Mitulovic","suffix":""},{"id":605036102,"identity":"bcdefcbf-40ed-40fa-af67-df94bff9abfc","order_by":2,"name":"Ranjith Kumar Ravi Kumar","email":"","orcid":"","institution":"Systems Biology of Pain, Division of Pharmacology \u0026 Toxicology, Department of Pharmaceutical Sciences, Faculty of Life Sciences, University of Vienna","correspondingAuthor":false,"prefix":"","firstName":"Ranjith","middleName":"Kumar Ravi","lastName":"Kumar","suffix":""},{"id":605036103,"identity":"008c51b3-4f60-4a0b-a72d-8665ea89351f","order_by":3,"name":"Lukas Uhrik","email":"","orcid":"","institution":"Systems Biology of Pain, Division of Pharmacology \u0026 Toxicology, Department of Pharmaceutical Sciences, Faculty of Life Sciences, University of Vienna","correspondingAuthor":false,"prefix":"","firstName":"Lukas","middleName":"","lastName":"Uhrik","suffix":""},{"id":605036104,"identity":"a624827a-9463-4d1b-8eaa-620061feaa9e","order_by":4,"name":"Elisabeth Urbauer","email":"","orcid":"","institution":"Chair of Nutrition and Immunology, Technical University of Munich","correspondingAuthor":false,"prefix":"","firstName":"Elisabeth","middleName":"","lastName":"Urbauer","suffix":""},{"id":605036105,"identity":"18e197c6-02c7-43f7-a018-268d20e98a3d","order_by":5,"name":"Doriane Aguanno","email":"","orcid":"","institution":"Chair of Nutrition and Immunology, Technical University of Munich","correspondingAuthor":false,"prefix":"","firstName":"Doriane","middleName":"","lastName":"Aguanno","suffix":""},{"id":605036106,"identity":"0ba47d14-f232-4c4a-8a9e-33ac194292e2","order_by":6,"name":"Dirk Haller","email":"","orcid":"","institution":"Chair of Nutrition and Immunology, Technical University of Munich","correspondingAuthor":false,"prefix":"","firstName":"Dirk","middleName":"","lastName":"Haller","suffix":""},{"id":605036107,"identity":"2f7c0ccf-8904-4975-8942-b39be14f8fd9","order_by":7,"name":"Manuela Schmidt","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABHElEQVRIie3QP0vDQBjH8V8JxCXoelJoXoHwhMCB4J+3cqFglii6lEKFVgrJIrjGd3EvoaVYl1TXEx2ydelQtwxFvLYoRY52dbjvcNxwH567A2y2f9v1cvEA4aIB1HoQOwn9Ego3id7tInAp2jxlJEfZZMzmdAY/mzzJsrWI5Xs/ZWUb/kHPmZYGwouri8OcmiC9UeKFLuXHMGWiQJAP3MA0hQ8SXvfIATGPqyjVREV9ilLUJGAmrzNeX1AXfr4mMa3IF84l9j6NROkpoBGg1kRoclfqT4gkPPMUNQuP7+lZvyXh+i1h8LgkYsya+ci7yY0XSwJVtTv6xwr+VrUa/r6Ky2F1e3L6kGVybiA/df9egQHOlvM2m81m29o3yQdnTp9c35IAAAAASUVORK5CYII=","orcid":"","institution":"Systems Biology of Pain, Division of Pharmacology \u0026 Toxicology, Department of Pharmaceutical Sciences, Faculty of Life Sciences, University of Vienna","correspondingAuthor":true,"prefix":"","firstName":"Manuela","middleName":"","lastName":"Schmidt","suffix":""},{"id":605036108,"identity":"28120c3b-fb80-4286-b2f8-0d014fc52f9f","order_by":8,"name":"David Gomez-Varela","email":"data:image/png;base64,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","orcid":"","institution":"Systems Biology of Pain, Division of Pharmacology \u0026 Toxicology, Department of Pharmaceutical Sciences, Faculty of Life Sciences, University of Vienna","correspondingAuthor":true,"prefix":"","firstName":"David","middleName":"","lastName":"Gomez-Varela","suffix":""}],"badges":[],"createdAt":"2026-03-12 11:02:06","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":true,"conflictsOfInterestStatement":true,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":true},"doi":"10.21203/rs.3.rs-9103948/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9103948/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104560918,"identity":"952ff694-9bac-4c82-b395-ba85b2746a04","added_by":"auto","created_at":"2026-03-13 10:13:48","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":932558,"visible":true,"origin":"","legend":"\u003cp\u003eBenchmarking design for systematic evaluation of PASEF acquisition strategies.\u003cstrong\u003e \u003c/strong\u003eHuman fecal peptides were used as a complex biological background and spiked with defined amounts of SILAC-labeled \u003cem\u003eLigilactobacillus murinus\u003c/em\u003e\u0026nbsp;(\u003cem\u003eL. murinus\u003c/em\u003e) and unlabeled \u003cem\u003eSalinibacter ruber\u003c/em\u003e\u0026nbsp;(\u003cem\u003eS. ruber\u003c/em\u003e) at three input levels (50:150 pg, 5:15 pg, 0.5:1.5 pg) in triplicate. Samples were analyzed using five PASEF acquisition strategies (DDA, DIA, Slice, Synchro and midia) across three LC gradients (5, 22 and 45 min) in triplicate, together with non-spike controls.\u003c/p\u003e","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-9103948/v1/102e1d2947df646313426869.png"},{"id":104781654,"identity":"26897612-1b2c-4efc-a866-414e8821a809","added_by":"auto","created_at":"2026-03-17 07:56:05","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2259509,"visible":true,"origin":"","legend":"\u003cp\u003eQualitative performance of PentaPASEF across gradients and spike-in levels.\u003cstrong\u003e A–B\u003c/strong\u003eTotal peptide (\u003cstrong\u003eA\u003c/strong\u003e) and protein group (\u003cstrong\u003eB\u003c/strong\u003e) identifications across acquisition methods and gradients.The bars represent the mean of spike-in conditions (n = 9), and the error bars show the standard deviations. \u003cstrong\u003eC\u003c/strong\u003eIdentification of spiked \u003cem\u003eL. murinus\u003c/em\u003e and \u003cem\u003eS. ruber\u003c/em\u003e peptides across input levels. Box plots show the distribution of identified peptides of spike-in and acquisition replicates (n = 9). Each box represents the inter-quartile range, spanning from the 25th percentile (lower bound of the box) to the 75th percentile (upper bound). The center line indicates the median (50th percentile). \u003cstrong\u003eD\u003c/strong\u003e Number of confidently annotated genera and species (≥3 taxa-specific peptides). The bars show the average number of genera and species across spike-in replicates (n = 9), and the error bars represent the standard deviations. \u003cstrong\u003eE\u003c/strong\u003e Number of enzymes (EC) and KEGG pathways detected. The bars show the average number of genera and species across spike-in replicates (n = 9), and the error bars represent the standard deviations. \u003cstrong\u003eF \u003c/strong\u003eScaled peptide counts annotated for species shared across five methods in spike-in replicates (n = 9 per method). Gray colors represents the species were not detected in the respective replicates. \u003cstrong\u003eG\u003c/strong\u003e, Distribution of protein supports per EC entry. Source data for Fig. 2 are provided in the Source Data file.\u003c/p\u003e","description":"","filename":"Fig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-9103948/v1/432f181d86a3d0da3c34b72b.png"},{"id":104560920,"identity":"84e19bad-0573-447e-b00c-653702b0f478","added_by":"auto","created_at":"2026-03-13 10:13:48","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2905692,"visible":true,"origin":"","legend":"\u003cp\u003eQuantitative performance of PentaPASEF strategies.\u003cstrong\u003e A\u003c/strong\u003e Distribution of peptide CVs across acquisition replicates (n = 3) for the background and spike-in peptides. Each box represents the inter-quartile range, spanning from the 25th percentile (lower bound of the box) to the 75th percentile (upper bound). The center line indicates the median (50th percentile). The red-dotted lines indicate the CV of 20%.\u0026nbsp;\u0026nbsp; \u003cstrong\u003eB\u003c/strong\u003e Distribution of Log₂ ratios between high and medium spike-in conditions. The red-dotted lines indicate the theoretical Log₂ ratios. \u003cstrong\u003eC\u003c/strong\u003e Variability of species abundance ranks across replicates (n = 9) stratified into Top, Medium and Low abundance tiers. The red-dotted lines indicate the CV of 20%. \u003cstrong\u003eD\u003c/strong\u003e Relative abundance ranks of spike-in species across gradients and input levels. \u003cstrong\u003eE\u003c/strong\u003e Dynamic range of EC entries across abundance ranks. The color bar represents the Log\u003csub\u003e10\u003c/sub\u003e intensity. \u003cstrong\u003eF\u003c/strong\u003e ECDF of CV distributions for quantified EC entires across abundance tiers (Top, Medium, and Low). Source data for Fig. 3C-F are provided in the Source Data file.\u003c/p\u003e","description":"","filename":"Fig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-9103948/v1/15a43c27be3081946cdde4c2.png"},{"id":104781059,"identity":"4c31d4f3-efe2-446a-bf5e-41f28ab9d18d","added_by":"auto","created_at":"2026-03-17 07:54:37","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2663016,"visible":true,"origin":"","legend":"\u003cp\u003eDIA- and Slice-PASEF reveal concordant microbial and host responses \u003cem\u003ein vivo\u003c/em\u003e.\u003cstrong\u003e A\u003c/strong\u003e Schematic of the experimental design showing the strategy for the transient deletion of mitochondrial heat shock protein 60 (Hsp60) using tamoxifen induction in mouse intestinal epithelial cells (IECs) with/without interleukin 10 (Il-10) knockout. Proximal and distal contents were sampled and analyzed at day 4 (D4), day 8 (D8), and day 14 (D14) after the tamoxifen diet. \u003cstrong\u003eB\u003c/strong\u003e Bray–Curtis beta diversity (species) of all experimental conditions. \u003cstrong\u003eC\u003c/strong\u003e Comparison of species abundance (mean) distributions between DIA- and Slice-PASEF. \u003cstrong\u003eD \u003c/strong\u003eIntensity distributions of species-specific peptides for DIA- and Slice-PASEF. \u003cstrong\u003eE\u003c/strong\u003e Number of significantly regulated species (Adjusted p value ≤ 0.05) across timepoints and contrasts. \u003cstrong\u003eF\u003c/strong\u003e Number of significantly regulated species–KEGG associations (Adjusted p value ≤ 0.05) across timepoints and contrasts. \u003cstrong\u003eG\u003c/strong\u003e Heatmap showing the Log\u003csub\u003e2\u003c/sub\u003e fold changes of species-KEGG association for biofilm formation in the contrast of Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e vs Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e. Gray color in the plot represents the species-KEGG associations were either statistically insignificant or missing.\u0026nbsp; \u003cstrong\u003eH\u003c/strong\u003e Number of significantly regulated host proteins (Adjusted p value ≤ 0.05) across timepoints in the contrast between Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e and Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e. \u003cstrong\u003eI \u003c/strong\u003eHeatmap showing the Log\u003csub\u003e2\u003c/sub\u003e fold changes of host proteins involved in the regulation of wound headling. Gray color in the plot represents that the host proteins\u0026nbsp; were either statistically insignificant or missing. Source data for Fig. 4B-I are provided in the Source Data file.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"Fig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-9103948/v1/1a4551b3723647d5ff965b9e.png"},{"id":106959593,"identity":"c57c5650-3e30-4091-9963-6114f29533a4","added_by":"auto","created_at":"2026-04-15 09:12:00","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8392622,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9103948/v1/7d60e4b2-2898-4b0a-871b-d992429fc126.pdf"},{"id":104560915,"identity":"d2810eb7-1387-46bd-9c06-5f80f1d61b4c","added_by":"auto","created_at":"2026-03-13 10:13:48","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":935784,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Figures\u003c/p\u003e","description":"","filename":"SupplementaryInformation.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9103948/v1/e02353cf434d8e0e1ac6bb3b.pdf"}],"financialInterests":"The authors declare potential competing interests as follows: M.S. received research awards and travel support from the German Pain Society (DGSS) both of which were sponsored by Astellas Pharma GmbH (Germany). M.S. received research awards from the Austrian Pain Society. M.S. received a one-time consulting honorarium from Grunenthal GmbH (Germany). None of these sources influenced the content of this study, and M.S. declares no conflict of interest. D.G.V. and M.S. have an ongoing scientific collaboration with Bruker (Center of Excellence for Metaproteomics University of Vienna - Bruker). F.X., R.K.R.K., L.U., E.U., D.A., D.H., M.S., and D.G.V. declare no competing interests. G.M. is an employee of Bruker Austria.","formattedTitle":"\u003cp\u003eSystematic evaluation of PASEF acquisition strategies in complex metaproteomes\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMetaproteomics is increasingly used to investigate the functional interplay between complex microbial communities (e.g., gut microbiome) and their host\u003csup\u003e1, 2\u003c/sup\u003e. Gut metaproteomics represents one of the most demanding application scenarios for LC–MS–based proteomics\u003csup\u003e3, 4, 5\u003c/sup\u003e. Fecal samples comprise dense mixtures of host and microbial proteomes spanning several orders of magnitude in abundance, providing a stringent stress test for acquisition strategies and data analysis pipelines\u003csup\u003e6\u003c/sup\u003e. At the same time, metaproteomic studies increasingly seek to quantify low-abundance community members and functional pathways that are central to host–microbiome interactions and diseases, placing exceptional demands on sensitivity, reproducibility, and quantitative accuracy\u003csup\u003e7, 8, 9\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eTrapped ion mobility spectrometry coupled with parallel accumulation–serial fragmentation (PASEF) has substantially expanded the speed, sensitivity, and quantitative precision of liquid chromatography–mass spectrometer (LC-MS)-based proteomics\u003csup\u003e10, 11, 12, 13, 14, 15, 16\u003c/sup\u003e. On this platform, multiple acquisition strategies have been developed, including data-dependent (DDA-PASEF\u003csup\u003e10\u003c/sup\u003e), data-independent (DIA-PASEF\u003csup\u003e11\u003c/sup\u003e), and more recent variants such as Slice-\u003csup\u003e12\u003c/sup\u003e, Synchro-\u003csup\u003e13\u003c/sup\u003e , and midia-PASEF\u003csup\u003e14\u003c/sup\u003e, which differ in their precursor sampling logic and trade-offs between analytical depth and throughput. However, the direct side-by-side comparison under matched analytical and computational conditions is missing. Moreover, despite their benefits in relatively low- to medium-complex samples, the performance of each acquisition strategy in highly complex biological matrices, where dynamic range, co-elution, and sampling bias strongly influence both identification and quantification\u003csup\u003e17\u003c/sup\u003e,\u0026nbsp;is unknown.\u003c/p\u003e\n\u003cp\u003ePASEF technology applied to complex metaproteomes\u003csup\u003e18\u003c/sup\u003e improves taxonomic and functional depth, increases throughput, improves limits of detection, and discovers host–microbiome interactions in preclinical mouse models of inflammatory bowel disease\u003csup\u003e1\u003c/sup\u003e. As a result, DDA- and DIA-PASEF are increasingly adopted in metaproteomic workflows to improve sequencing depth and throughput in gut microbiome studies\u003csup\u003e18, 19, 20, 21, 22\u003c/sup\u003e. Thus, it is necessary to study if the more recent\u0026nbsp;methods\u003csup\u003e12, 13, 14\u003c/sup\u003e (e.g.,\u0026nbsp;Slice-, Synchro-,\u0026nbsp;midia-PASEF)\u0026nbsp;could further provide improvements on identification depth, sensitivity to low-abundance species, quantitative reliability, and functional interpretation in complex microbiome samples\u003csup\u003e18, 20\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eHere we established a controlled benchmarking framework to evaluate five PASEF acquisition strategies (PentaPASEF) across chromatographic regimes and sample input levels using a highly complex fecal peptide background spiked with defined reference bacteria. We systematically assessed qualitative performance, quantitative precision and accuracy, species- and function-level consistency, and false-positive control across 540 matched LC–MS acquisitions. Finally, we evaluated the biological impact of acquisition choice in a murine epithelial injury model\u003csup\u003e23\u003c/sup\u003e, showing that DIA- and Slice-PASEF captured highly concordant microbial and host responses while differing primarily in quantitative sensitivity-related changes. Together, this study enables direct translation of acquisition design into expected analytical performance, providing practical guidance for benchmarking PASEF strategies according to sample complexity, sensitivity requirements, and experimental throughput.\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eExperimental design for benchmarking PASEF acquisition strategies in fecal microbiomes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIon mobility spectrometry adds an orthogonal dimension of separation to proteomics by resolving ions according to their collisional cross section, thereby improving selectivity, sensitivity, and quantitative precision. When combined with trapped ion mobility spectrometry (TIMS) and parallel accumulation–serial fragmentation (PASEF), this approach enables near-complete ion utilization and accelerated duty cycles, substantially enhancing proteome coverage. These advances have led to several PASEF acquisition strategies, including DDA-, DIA-, Slice-, Synchro-, and midia-PASEF, yet their relative performance in highly complex biological matrices remains insufficiently characterized.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo enable a systematic comparison, we established a controlled benchmarking framework spanning different chromatographic separations and sample complexities (Fig. 1). Human fecal material was selected as the background matrix owing to its extreme proteomic complexity\u003csup\u003e7, 24\u003c/sup\u003e, which captures the dynamic range and compositional heterogeneity typical of real biological samples. Two reference bacteria—SILAC-labeled \u003cem\u003eLigilactobacillus murinus\u003c/em\u003e (\u003cem\u003eL. murinus\u003c/em\u003e) and unlabeled \u003cem\u003eSalinibacter ruber\u003c/em\u003e (\u003cem\u003eS. ruber\u003c/em\u003e)\u003csup\u003e1\u003c/sup\u003e—were spiked into the fecal peptide background at a fixed ratio (1:3) to emulate low-abundance community members. Three input levels (50 pg : 150 pg, 5 pg : 15 pg, and 0.5 pg : 1.5 pg) were analyzed in triplicate to ensure statistical robustness, together with non-spiked controls.\u003c/p\u003e\n\u003cp\u003eLC–MS/MS data were acquired using three gradient lengths (5, 22, and 45 min) to balance throughput and depth across all PentaPASEF modes, resulting in 540 injections covering all experimental combinations. DDA-PASEF data were processed with MSFragger\u003csup\u003e25\u003c/sup\u003e within FragPipe, whereas all other datasets were analyzed using DIA-NN\u003csup\u003e26\u003c/sup\u003e to maximize comparability, applying a 1% FDR at precursor and protein group levels. To ensure confident identification of spiked bacterial peptides, any heavy-labeled \u003cem\u003eL. murinus\u003c/em\u003e or \u003cem\u003eS. ruber\u003c/em\u003e identifications observed in non-spike controls (gradient- and method-wise) were removed from corresponding experimental runs, and background peptides falsely assigned with heavy labeling were filtered.\u003c/p\u003e\n\u003cp\u003eThis experimental framework provides a rigorous and biologically relevant basis to compare PASEF acquisition strategies, capturing their relative performance across different chromatographic resolutions, acquisition modes, and analytical depths in metaproteomic contexts.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcquisition strategy shapes taxonomic and functional resolution in metaproteomics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe first assessed the qualitative performance of PentaPASEF acquisition strategies across LC gradients and spike-in levels. The overall number of identified peptides (Fig. 2A; Supplementary Data 1) and proteins (Fig. 2B) remained relatively stable across spike-in conditions and replicates (acquisition triplicates were averaged; n = 9 per bar), demonstrating the robustness of all PASEF methods. This stability was expected because the spiked bacterial peptides constituted only a minor fraction of the total peptide pool within the complex fecal background. Across all PentaPASEF methods, peptide and protein identifications scaled consistently with LC gradient length, with 45-min separations yielding the highest coverage (Fig. 2A-2B). DIA-based methods, particularly Slice- and DIA-PASEF, consistently exceeded DDA-PASEF in total identifications, reflecting their enhanced ion sampling efficiency and reduced precursor selection bias. Notably, even at the 5-min gradient, DIA-based modes maintained substantial peptide coverage, underscoring their suitability for high-throughput analyses. For Slice- and midia-PASEF, alternative acquisition schemes were evaluated. A one-frame Slice configuration and a 5-Da overlay in midia-PASEF yielded higher identifications (Supplementary Fig. 1A) and were therefore retained for downstream analyses. Interestingly, midia-PASEF showed fewer identifications compared to DIA- and Slice-PASEF. To exclude analysis pipeline–dependent effects, DIA-PASEF and midia-PASEF datasets were reanalyzed with Spectronaut 20, which reproduced the same pattern (Supplementary Fig. 1B). Peptide-level overlap analysis revealed a large shared core among methods at all gradients, with overlap increasing at longer separations (Supplementary Fig. 1C–E). DDA-PASEF generated the highest number of unique identifications, consistent with its semi-random targeting of abundant precursors, which resulted in lower reproducibility relative to DIA-based approaches.\u003c/p\u003e\n\u003cp\u003eFocusing on the spiked \u003cem\u003eL. murinus\u003c/em\u003e and \u003cem\u003eS. ruber\u003c/em\u003e peptides, the PentaPASEF methods displayed distinct sensitivity profiles (Fig. 2C; Supplementary Data 2). While following the gradient-dependent trend observed for total identifications, the number of spike-in peptide identifications decreased proportionally with input amounts. Nevertheless, DIA-based strategies retained higher recovery across gradients, demonstrating superior detection of low-abundance peptides in complex matrices. Replicate variability increased at lower inputs (Fig. 2C), reflecting greater sampling fluctuation as signals approached the detection limit. Importantly, even at the sub-picogram level (0.5 pg \u003cem\u003eL. murinus\u003c/em\u003e), DIA-based approaches—particularly DIA- and Slice-PASEF—still yielded detectable peptide identifications (Fig. 2C; Supplementary Data 2), most prominently at the 22- and 45-min gradients. Assessment of non-spike controls confirmed excellent false-positive control for both FragPipe and DIA-NN, with mis-assigned heavy-labeled L. murinus and S. ruber peptides below 1% across datasets (Supplementary Fig. 1F). DDA-PASEF showed slightly higher false assignment frequencies, whereas DIA-based modes maintained tighter control.\u003c/p\u003e\n\u003cp\u003eGiven that the background matrix consisted of human fecal material, we next examined how acquisition strategy influenced taxonomic and functional representation. Genus- and species-level profiles (Fig. 2D) and functional annotations based on EC numbers and KEGG pathways (Fig. 2E) mirrored the gradient-dependent trends observed for peptide identifications. The number of confidently assigned taxa (≥3 taxa-specific peptides) and functional entries increased with gradient length but differed only marginally between 22- and 45-min runs, indicating that near-maximal profiling depth was already achieved at 22-min. The recovered community composition was consistent with previous fecal metaproteome reports (Supplementary Fig. 1G).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough DDA-PASEF detected a similar number of species as DIA-based methods (Fig. 2D), peptide-level inspection revealed that species shared across methods were supported by substantially fewer peptides in DDA datasets (Fig. 2F). This suggests lower annotation confidence and reduced taxonomic coverage for DDA-derived assignments, particularly compared with DIA- and Slice-PASEF. A comparable pattern was observed at the functional level: analysis of protein support per EC entry (Fig. 2G) showed that, even at the 5-min gradient, DIA-, Slice- and Synchro-PASEF annotated many ECs with \u0026gt;20 proteins, whereas DDA- and midia-PASEF annotated far fewer. Moreover, DIA-based methods displayed a larger fraction of ECs supported by \u0026gt;50 proteins, indicating superior capacity to define enzymatic functions through multiple converging protein evidences. This broader protein-to-function mapping effectively enhances the functional detection limit in complex metaproteomic samples.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQuantitative precision and accuracy across PentaPASEF strategies\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBuilding on the qualitative benchmarks, we next compared the quantitative performance of the PASEF strategies under identical analytical conditions. Quantitative precision was assessed by calculating the coefficient of variation (CV) of peptide intensities across acquisition replicates (n = 3). Intra-group Pearson correlations were high for most datasets and gradients (Supplementary Fig. 2A). The few lower correlations observed for Slice-PASEF at 22 min (r \u0026lt; 0.9) originated from a single replicate (Supplementary Data 3) rather than a systematic effect. Across all experiments, the majority of quantified peptides showed CVs below 20% (Fig. 3A), indicating overall high reproducibility of both the LC–MS platform and the acquisition schemes. Interestingly, DDA-PASEF displayed a pronounced gradient dependence, with variability decreasing at longer separations, whereas DIA-based methods were less affected by gradient length. Among all strategies, Slice-PASEF consistently achieved the lowest median CVs, including at the 5-min gradient, demonstrating that high quantitative precision can be retained under high-throughput conditions.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhen analyzed separately for the spike-in species (Fig. 3A; Supplementary Fig. 2B), all methods showed good precision at the highest input (150 pg \u003cem\u003eS. ruber\u003c/em\u003e, 50 pg \u003cem\u003eL. murinus\u003c/em\u003e). At the medium level, performance declined first for DDA-PASEF, followed by Synchro- and midia-PASEF, particularly for 5 pg \u003cem\u003eL. murinus\u003c/em\u003e. In contrast, Slice-PASEF largely preserved its precision under these conditions. At the lowest inputs (1.5 pg S. ruber, 0.5 pg L. murinus), variability increased markedly for all methods, indicating current limits of reproducible quantification below the picogram range.\u003c/p\u003e\n\u003cp\u003eQuantitative accuracy was evaluated by comparing log₂ intensity ratios between high and medium spike-in conditions (Fig. 3B; Supplementary Data 4). For the human fecal background, the log₂ ratios were largely centered around zero across all methods and gradients, indicating consistent quantification for the majority of background peptides. Spike-in peptides showed clear separation consistent with the expected 10-fold difference. Under the 5-min gradient, all methods exhibited some ratio compression, most prominently for DDA-PASEF. For \u003cem\u003eL. murinus\u003c/em\u003e, DIA-, Slice- and Synchro-PASEF clustered closely around the theoretical log₂ ratio of 3.3 at 22- and 45-min gradients, whereas DDA-PASEF showed broader dispersion and systematic underestimation. S.ruber peptides followed a similar pattern, with most DIA-based methods maintaining distributions closely centered near the expected ratio, while DDA-PASEF exhibited broader variability and stronger ratio compression. Among all methods, Slice-PASEF showed the narrowest ratio distributions and best agreement with theoretical expectations across gradients, indicating superior quantitative accuracy.\u003c/p\u003e\n\u003cp\u003eTo assess quantitative consistency at the microbial compositional level, all quantified taxa were stratified into Top, Medium and Low abundance tiers, and the CV of their relative abundance ranks across replicates (n = 9) was used as a measure of stability. Overall, DIA-based methods exhibited markedly tighter rank variability than DDA-PASEF across gradients and abundance tiers (Fig. 3C; Supplementary Data 5), consistent with the beta-diversity analysis where DDA showed larger within-method dispersion (Supplementary Fig. 2C). Notably, the Top tier did not always display the lowest variability—particularly at 5 min—whereas several DIA-based modes maintained stable reproducibility even in Medium and Low tiers (Fig.3C). We next examined the ranks of the spike-in species across conditions. As expected, both \u003cem\u003eS. ruber\u003c/em\u003e and \u003cem\u003eL. murinus\u003c/em\u003e showed a progressive decrease in rank with lower spike-in amounts, consistent with their relative spike-in levels (Fig. 3D). Overall, the rank patterns were largely comparable among most PASEF methods at each gradient, indicating similar quantitative scaling across acquisition strategies. However, ranking deviations were observed at low inputs, particularly for DDA-PASEF, which occasionally yielded disproportionately high ranks at low inputs—for example, ranking \u003cem\u003eS. ruber\u003c/em\u003e at the 35th in the 22-min gradient, while other methods placed it behind the 120th. At the medium input, \u003cem\u003eL. murinus\u003c/em\u003e ranked near 80th in DDA- and DIA-PASEF but around 130th in Slice-, Synchro- and midia-PASEF, and these differences largely diminished at the 45-min gradient, suggesting that longer separation times improve rank agreement among methods at moderate abundance levels (Fig. 3D). Across conditions, rank patterns were most consistent among DIA- and Slice-PASEF acquisitions, whereas DDA-PASEF frequently deviated at low input levels.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAt the functional layer, acquisition strategy strongly influenced quantitative depth. DIA-based methods, especially Slice-PASEF, exhibited a broader EC dynamic range and higher intensities at comparable abundance ranks (Fig. 3E; Supplementary Data 6), whereas DDA- and midia-PASEF showed weaker signals in low-abundance regions, suggesting reduced detection depth of enzymatic functions. This pattern was consistent across gradients, with minimal improvement beyond 22 min. When assessing functional quantification reproducibility, DIA-based methods again showed superior performance, characterized by steeper ECDF curves and a higher proportion of EC entries with CVs below 20%. Slice-PASEF consistently exhibited the most compact CV distributions across all abundance tiers (Fig. 3F), demonstrating its capacity to maintain precise functional quantification even for low-abundance ECs.\u0026nbsp;\u003cbr\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDIA- and Slice-PASEF discover concordant microbial and host responses in\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ea\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ecolon-injury mouse model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDIA- and Slice-PASEF showed the best overall performance in the benchmarking experiment. We next investigated their ability to discover molecular mechanisms in an experimental design that integrates not only the complexity of a fecal microbiome but also the biological variability of an \u003cem\u003ein vivo\u003c/em\u003e setup. Thus, we applied both acquisition strategies on a murine colon injury model created by the conditional knock-out of Hsp60 in intestinal epithelial cells combined with the presence or absence of Interleukin 10 (Il-10) expression (Fig. 4A). Conditional deletion of Hsp60 induces mitochondrial stress, leading to mucosal injury and microbial dysbiosis, while additional IL-10 deficiency impairs epithelial regeneration in the distal colon, resulting in persistent inflammatory pathology\u003csup\u003e1, 23\u003c/sup\u003e. \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003eAcross all conditions, DIA- and Slice-PASEF identified a similar total number of peptides (Supplementary Fig. 3A; Supplementary Data 7) and number of total bacterial species detected (Supplementary Fig. 3B; Supplementary Data 8). Alpha- and beta-diversity profiles (Supplementary Fig. 3C and Fig. 4B) show how both acquisition strategies discovered analogous injury-associated compositional gradients concomitant to the tissue phenotype separation between proximal and distal colon in Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e mice (e.g., different clustering only at D14 in the distal colon in Fig. 4B). While species-level abundances quantified by DIA- and Slice-PASEF were highly correlated (Pearson r \u0026gt; 0.91; Supplementary Fig. 3D), Slice-PASEF exhibited a systematic upward shift in species abundance distributions (Fig. 4C). This shift was not driven by differences in the number of quantified species (Supplementary Fig. 3B) or species-specific peptides (Supplementary Fig. 3E). Instead, it reflected a consistently higher peptide-level intensities in Slice-PASEF (Fig. 4D), likely resulting from denser MS/MS sampling intrinsic to the Slice acquisition and its propagation into MaxLFQ-based quantification. In fact, Slice-PASEF generated substantially larger raw file sizes and required longer DIA-NN processing times (Supplementary Fig. 3F). Minimum detectable species abundances were ~0.003% for both methods, with Slice-PASEF reaching slightly lower limits in both distal (0.0031% vs. 0.0034% in DIA) and proximal samples (0.0023% vs. 0.0037% in DIA) (Supplementary Data 9).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDifferential abundance analysis revealed that the majority of significantly regulated species were shared between DIA- and Slice-PASEF across all comparisons and timepoints (Fig. 4E; Supplementary Data 10). Temporal patterns of differential species discovered the underpinnings correlating to the previously observed tissue pathology\u003csup\u003e23\u003c/sup\u003e (Fig. 4E). In the Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e+/+\u003c/sup\u003e (injury at D4 and D8 but histopathological regeneration at D14) vs Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e (no colitis at the timepoint in the absence of Hsp60 deletion\u003csup\u003e23\u003c/sup\u003e) contrast, the number of differential species markedly decreased by D14 in both colonic regions, consistent with epithelial regeneration due to the regain of Hsp60 protein expression in the conditional Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e+/+\u003c/sup\u003e mice at D14. In contrast, the comparison between Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e (injury at D4 and D8 without histopathological regeneration at D14 in the distal colon\u003csup\u003e23\u003c/sup\u003e) and Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e (no colitis at the timepoint in the absence of Hsp60 deletion\u003csup\u003e23\u003c/sup\u003e) revealed a larger set of significant species at D14 in the distal colon compared to D4 and D8. These findings offer a new layer of information based on the ecosystem functional expression, which both increases the sensitivity, compared to imaging alternatives, to profile pathological changes, and offers a plausible explanation for the recovery dynamics observed in these mice.\u003c/p\u003e\n\u003cp\u003eInterestingly, each method also detected a subset of unique significant species (Fig. 4E). Inspection of raw p-values revealed that many method-unique species were nominally significant in the alternative method but failed multiple-testing correction (Supplementary Fig. 3G; Supplementary Data 10), indicating that uniqueness frequently reflected borderline statistical behavior rather than discordant biology. Direct comparison of Δ|log2FC| and Δ(−log10 p-value) further showed that method-unique calls aligned with the method exhibiting both larger effect sizes and stronger statistical evidence (Supplementary Fig. 3H), suggesting that uniqueness is primarily driven by method-dependent effect size estimation rather than increased measurement variability or statistical noise.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDespite partial species-level uniqueness, pathway-resolved analysis revealed strong functional concordance between DIA- and Slice-PASEF (Fig. 4F). Across contrasts and timepoints, both methods identified highly overlapping sets of significantly regulated species–KEGG associations, comprising 182 shared KEGG pathways linked to 175 shared species (Supplementary Data 11). In the Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e vs Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e contrast, both methods consistently detected similar sets of inflammation-associated (e.g., NOD-like receptor signaling), stress-response metabolic (e.g., pentose phosphate and methane metabolism), and regeneration-associated pathways (e.g., biofilm formation, quorum sensing). A detailed analysis revealed how both acquisition strategies identify comparable species-level drivers underlying particular functional changes, as well as site-specific functional strategies of specific species. For example, in biofilm formation, DIA- and Slice-PASEF assigned highly similar sets of species concurrent with the lack of functional remodeling in the distal colon (Fig. 4G). Interestingly, both strategies discovered several species (e.g., \u003cem\u003eEubacterium plexicaudatum\u003c/em\u003e, \u003cem\u003eButyribacter intestini\u003c/em\u003e, and \u003cem\u003eJutongia huaianensis\u003c/em\u003e) in the proximal colon as associated with decreased biofilm activity, and other species (e.g., \u003cem\u003eLigilactobacillus murinus\u003c/em\u003e and \u003cem\u003eAnaerotruncus colihominis\u003c/em\u003e) in the distal colon associated with increased biofilm formation. A similar concordance and site-specification were also observed for the two-component system pathway (Supplementary Fig. 3I).\u003c/p\u003e\n\u003cp\u003eAnalysis of the host proteome revealed interesting functional alteration patterns aligning to the above-described microbiome alterations. We observed highly concordant patterns of protein regulation between DIA- and Slice-PASEF (Supplementary Data 12), with both methods capturing maximal host protein dysregulation at D8, corresponding to the peak of tissue injury detected histopathologically as described previously\u003csup\u003e23\u003c/sup\u003e (Fig. 4H, Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e vs Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e−/−\u003c/sup\u003e). Moreover, a larger number of proteins remained significantly altered at D14 in the distal compared with the proximal colon. Both methods consistently identified enrichment of epithelial repair and immune defense pathways, supporting concordant biological interpretation at the host level. Proteins involved in hemostasis (e.g., Fga, Fgb, and Fgg) and epithelial membrane repair and restitution (e.g., Anxa1, Anxa2, Anxa5) were more prominently upregulated in the proximal colon (Fig. 4I, Supplementary Fig. 3J), strongly suggesting accelerated tissue regeneration relative to the distal colon.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe growing diversity of PASEF acquisition strategies has created both opportunity and uncertainty for metaproteomics\u003csup\u003e18, 20, 27\u003c/sup\u003e, as method choice increasingly determines analytical depth, quantitative reach, and biological interpretability. By benchmarking five PASEF implementations within a unified experimental and computational framework, our study reveals that acquisition strategy is not a neutral technical decision but a primary driver of metaproteomic sensitivity, reproducibility, and functional resolution. This study provides a comprehensive and biologically grounded evaluation of five PASEF acquisition strategies across chromatographic regimes, sample input levels, and analytical depths in fecal metaproteomes. By integrating qualitative, quantitative, and biological validation analyses, our results establish a unified framework for understanding how acquisition strategy selection influences identification depth, quantitative performance, and downstream biological interpretation in complex microbiome samples.\u003c/p\u003e\n\u003cp\u003eAcross all acquisition modes, peptide and protein identifications scaled predictably with chromatographic separation length, confirming gradient duration as a primary determinant of analytical depth. DIA-based strategies consistently achieved broader and more stable peptide coverage than DDA-PASEF, including under high-throughput conditions (Fig. 2A-B). This advantage extended to the detection of low-abundance microbial peptides and functional annotations, where DIA-based acquisitions provided enhanced dynamic range and broader protein support per enzymatic entry (Fig. 2F-G). However, these gains come with practical trade-offs. Slice-PASEF generated substantially larger raw data files (Supplementary Fig. 3F) than other acquisition modes, increasing storage demands and computational burden during data processing. In addition, midia-PASEF currently requires proprietary licensing, which may limit accessibility and adoption despite its demonstrated performance\u003csup\u003e14\u003c/sup\u003e. These considerations highlight that acquisition strategy selection should balance analytical performance against data management capacity and software availability.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eQuantitative analyses revealed marked differences between acquisition modes. Slice-PASEF achieved the most favorable balance between sensitivity and quantitative robustness, yielding consistently low coefficients of variation, minimal ratio compression and stable species-level abundance ordering across spike-in series (Fig. 3). DIA-PASEF also maintained robust quantitative performance but exhibited slightly higher variability under low-input conditions (Fig. 3A) and narrower dynamic ranges of quantified functions (Fig. 3E). In contrast, DDA-PASEF showed increased variability especially under high throughput conditions, ratio compression and rank instability (Fig. 3B-C), underscoring inherent limitations of precursor-dependent sampling in dense fecal peptide matrices. Our accuracy assessment was based on controlled spike-in ratios of two bacterial species and relative abundance rankings, which capture internal quantitative fidelity but do not provide absolute concentration ground truth. The absence of comprehensive isotopically labeled standards across the entire microbial dynamic range limits direct calculation of absolute quantitative accuracy, and future work incorporating such standards would further refine sensitivity limits and bias estimates.\u003c/p\u003e\n\u003cp\u003eMethodological differences observed in benchmarking are directly translated into biological interpretation. In a murine colonic epithelial injury model, DIA- and Slice-PASEF captured highly concordant microbial and host response patterns, which indicates that the acquisition strategy primarily modulates quantitative reach rather than biological directionality. More importantly, DIA- and Slice-PASEF datasets provided evidence, to an extent, aligning with the tissue pathology previously reported\u003csup\u003e23\u003c/sup\u003e for the same samples. Nevertheless, the biological validation was conducted in a single disease model without an independent molecular ground truth for taxonomic change, pathway activation, or host response. Consequently, while relative concordance between methods can be assessed, the absolute biological correctness of specific response magnitudes cannot be definitively established, especially for method-dependent findings. Additional validation using orthogonal assays or controlled perturbation models will be required to fully define biological accuracy.\u003c/p\u003e\n\u003cp\u003eAlthough fecal samples represent one of the most compositionally complex and analytically demanding metaproteomic matrices, they do not capture the full diversity of microbiome-associated environments. Microbiomes from environmental\u003csup\u003e28, 29\u003c/sup\u003e, oral\u003csup\u003e30, 31\u003c/sup\u003e , or skin\u003csup\u003e32, 33\u003c/sup\u003e niches may differ in peptide complexity, dynamic range, and interference patterns, which could influence relative method performance. Importantly, for each acquisition strategy, we applied standardized and widely adopted parameter settings rather than exhaustively optimizing all method-specific acquisition parameters. Apart from exploratory testing of Slice-PASEF frame densities and midia-PASEF window overlap configurations (Supplementary Fig. 1A), acquisition windows, duty cycles, and isolation schemes were not systematically tuned to maximize performance for each individual method. Consequently, absolute performance ceilings for some strategies may be slightly higher than those observed here, and future targeted optimization could further refine sensitivity, throughput, and quantitative behavior.\u003c/p\u003e\n\u003cp\u003eOur results indicate that DIA-based PASEF strategies—particularly DIA- and Slice-PASEF—define a new operational regime for metaproteomics in which high-throughput workflows can simultaneously achieve deep coverage, stable quantification, and expanded functional detectability. Together, these findings establish a quantitative reference for selecting and refining PASEF strategies and provide a foundation for future development of acquisition designs, software tools and standards that will enable reproducible, sensitive and scalable metaproteomics.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eE\u003c/strong\u003e\u003cstrong\u003ethics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuman fecal sampling carried out at the University of Vienna was under the ethical approval (01149). Mouse work carried out at the Technical University of Munich, as well as maintenance and breeding of mouse lines, were approved by the Committee on Animal Health Care and Use of the state of Upper Bavaria (Regierung von Oberbayern; AZ ROB-55.2-2532.Vet_02-14-217, AZ ROB-55.2-2532.Vet_02-20-58, AZ ROB-55.2-2532.Vet_02-18-37) and performed in strict compliance with the EEC recommendations for the care and use of laboratory animals (European Communities Council Directive of November 24, 1986 (86/609/EEC)).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAnimals and housing conditions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMice for \u003cem\u003ein vivo\u003c/em\u003e experiments (Fig. 4) were male and housed under specific pathogen-free (SPF) conditions according to the criteria of the Federation for Laboratory Animal Science Associations (FELASA) (12-hour light/dark cycles at 24–26°C) in the mouse facility at the Technical University of Munich (School of Life Sciences Weihenstephan). All mice received a standard diet (autoclaved V1124-300, Ssniff) \u003cem\u003ead libitum\u003c/em\u003e, autoclaved water and were sacrificed by CO\u003csub\u003e2\u003c/sub\u003e or isoflurane. \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDetails of the animal models can be found in our previous study\u003csup\u003e23\u003c/sup\u003e. Briefly, Hsp60\u003csup\u003efl/fl\u003c/sup\u003e mice and Hsp60\u003csup\u003efl/fl\u003c/sup\u003e x VillinCreER\u003csup\u003eT2-Tg\u003c/sup\u003e mice were generated as described previously\u003csup\u003e34\u003c/sup\u003e to create IEC-specific Hsp60 knockout mice via tamoxifen induction (Hsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e+/+\u003c/sup\u003e). In addition, Hsp60\u003csup\u003efl/fl\u0026nbsp;\u003c/sup\u003eandHsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e+/+\u003c/sup\u003e mice were crossed with whole body Il-10 knockout mice (Il10\u003csup\u003e-/-\u003c/sup\u003e) to generate Hsp60\u003csup\u003efl/fl\u003c/sup\u003e;Il10\u003csup\u003e-/-\u0026nbsp;\u003c/sup\u003eandHsp60\u003csup\u003eΔ/ΔIEC\u003c/sup\u003e;Il10\u003csup\u003e-/-\u0026nbsp;\u003c/sup\u003emice\u003csup\u003e23\u003c/sup\u003e.For conditional Hsp60 deletion,mice and appropriate control mice\u0026nbsp;(6-weeks of age)\u0026nbsp;were kept on phytoestrogen-reduced diet 1005 (V1154-300, Ssniff) for four weeks under SPF conditions. Afterwards, mice received 400mg tamoxifen citrate per kg chow feed (CreActive T400 (10mm, Rad), Genobios) \u003cem\u003ead libitum\u003c/em\u003e for 7 days. After the induction phase, tamoxifen diet was replaced with the phytoestrogen-reduced diet. During and after the induction phase, mice were monitored daily and aborted when a combined score considering weight loss, changes in stool consistency, general behaviour, and general state of health was reached. Animals were sacrificed at the indicated time points\u0026nbsp;(D4, D8, and D14 after tamoxifen diet).\u0026nbsp;All mice and their respective genotypes were generated and maintained on an in-house crossing of C57Bl/6N and C57Bl/6J background.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProtein extraction and SP3-assisted protein digestion for metaproteomics analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe procedures from protein extraction of gut microbiome material to final peptide preparation were performed as previously described\u003csup\u003e1, 18\u003c/sup\u003e. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLiquid chromatography-mass spectrometry configurations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNanoflow reversed-phase liquid chromatography (Nano-RPLC) was performed on either ProElute or NanoElute2 systems (Bruker Daltonik, Bremen, Germany) coupled with timsTOF Ultra2 (Bruker Daltonik, Bremen, Germany) via CaptiveSpray ion source, respectively. Mobile solvent A consisted of 100% water containing 0.1% FA and mobile phase B of 100% acetonitrile containing 0.1% FA.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLiquid chromatogram setups\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eResults presented in Fig. 1-Fig. 3 were generated with three LC gradients on the ProElute system. For 5-min gradient separation, peptides were loaded onto a PepSep® column (4 cm x 150 µm) packed with 1.9 µm ReproSil C18 particles (Bruker). The mobile phase B was linearly increased from 2 to 35% in 5 minutes with a flowrate of 0.3 µL/min, followed by a steep increase to 95% in 0.1 minute. The mobile phase B was maintained at 95% for the last 3 minutes 0.3 µL/min. For 22-min gradient separation, peptides were loaded onto a PepSep® column (25 cm x 75 µm) packed with 1.5 µm ReproSil C18 particles (Bruker). The mobile phase B was linearly increased from 2 to 20% in 17 minutes with a flowrate of 0.3 µL/min, followed by another linear increase to 35% within 5 minutes and a steep increase to 95% in 0.5 minute. The mobile phase B was maintained at 95% for the last 7.5 minutes 0.3 µL/min. For 45-min gradient separation, peptides were loaded onto a PepSep® column (25 cm x 75 µm) packed with 1.5 µm ReproSil C18 particles (Bruker). The mobile phase B was linearly increased from 2 to 20% in 35 minutes with a flowrate of 0.3 µL/min, followed by another linear increase to 35% within 10 minutes and a steep increase to 95% in 1 minute. The mobile phase B was maintained at 95% for the last 4 minutes 0.3 µL/min. Mouse colonic microbiome samples (Fig. 4) were analyzed on a NanoElute2 system coupled with an Aurora\u003csup\u003eTM\u003c/sup\u003e ULTIMATE column (25 cm x 75 µm) packed with 1.7 µm C18 particles (IonOpticks, Fitzroy, Australia) with a 45-min gradient. The mobile phase B was linearly increased from 2 to 20% in 35 minutes with a flowrate of 0.25 µL/min, followed by another linear increase to 35% within 10 minutes at a flowrate of 0.25 µL/min and then a steep increase to 95% in 0.5 minute with flowrate increasing to 0.4 µL/min. The mobile phase B was maintained at 95% for the last 4.5 minutes 0.4 µL/min.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePentaPASEF method setups\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDifferent PASEF methods can be extacted from the raw data uploaded. A summary report \u0026nbsp;of all methods with key parameters is provided in Supplementary Data 13.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMicrobial database construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo analyze the human fecal samples, DDA-PASEF files generated in Fig. 1 were submitted to MSfragger\u003csup\u003e35\u003c/sup\u003e (version 4.1) integrated in FragPipe computational platform (version 22.0), searching against the MGnify human gut protein catalogue (https://www.ebi.ac.uk/metagenomics/genome-catalogues/human-gut-v2-0-2)\u003csup\u003e36\u003c/sup\u003e. The decoy database was generated with reversed sequences. Trypsin was specified with a maximum of two missed cleavages allowed. The search included variable modifications of methionine oxidation and N-terminal acetylation and a fixed modification of carbamidomethyl on cysteine. The mass tolerances of 20 ppm were set for precursor and fragment. Peptide length was set to 7 to 50 amino acids with a mass range from 500 to 5000 Da. The remaining parameters were kept as default settings. During the validation, MSBooster (version 1.2.31) was used for rescoring and Percolator\u003csup\u003e37\u003c/sup\u003e (version 3.6.5, default parameters) was used for PSM validation. FDR level was set to 1% for PSM, peptide and protein. The identified proteins from the search formed a sample-specific protein database, containing 30854 protein sequences.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo generate a sample-specific microbial database for the colonic samples (Fig. 4), we followed the recently published strategy of novoMP\u003csup\u003e1\u003c/sup\u003e. Briefly, a total of 60 µg pooled peptides (from both proximal and distal regions) were fractionated (Fisher Scientific, Cat. 84868) according manufactures instruction. Eight peptide factions were dried using vacuum centrifugation and then re-suspended in 30 µL of MS-grade water. The peptide concentration was measured in duplicate using NanoPhotometer N60 (Implen, Munich, Germany) at 205 nm. Peptide samples were acidified with formic acid to a final concentration of 0.1% and were stored at -20°C until LC-MS/MS analysis. 50 ng of each peptide fraction (a total of 8 fractions) were loaded on an Aurora\u003csup\u003eTM\u003c/sup\u003e ULTIMATE column (25 cm x 75 µm) packed with 1.6 µm C18 particles (IonOpticks, Fitzroy, Australia) with a total separation time of 60 minutes in DDA-PASEF mode. The TIMS analyzer was operated in a 100% duty cycle with equal accumulation and ramp times of 100 ms each. Specifically, 10 PASEF scans were set per acquisition cycle (cycle time of 1.17 s) with ion mobility range from 0.7 to 1.3 (1/k0). The target intensity and intensity threshold were set to 20000 and 500 respectively. Dynamic exclusion was applied for 0.4 minutes. Ions with m/z between 100 and 1700 were recorded in the mass spectrum. Collision energies were dependent on ion mobility values with a linear increase in collision energy from 1/K0 = 0.6 Vs/cm² at 20 eV to 1/K0 = 1.6 Vs/cm² at 63 eV. Fractionated DDA files were submitted to MSfragger\u003csup\u003e35\u003c/sup\u003e (version 4.3) integrated in FragPipe computational platform (version 23.1), searching against the MGnify mouse gut protein catalogue (https://www.ebi.ac.uk/metagenomics/genome-catalogues/mouse-gut-v1-0). The decoy database was generated with reversed sequences. Trypsin was specified with a maximum of two missed cleavages allowed. The search included variable modifications of methionine oxidation and N-terminal acetylation and a fixed modification of carbamidomethyl on cysteine. The mass tolerances of 20 ppm were set for precursor and fragment. Peptide length was set to 7 to 50 amino acids with a mass range from 500 to 5000 Da. The remaining parameters were kept as default settings. During the validation, MSBooster (version 1.3.17) was used for rescoring and Percolator\u003csup\u003e37\u003c/sup\u003e (version 3.7.1, default parameters) was used for PSM validation. FDR level was set to 1% for PSM, peptide and protein. The identified proteins from the search formed a sample-specific protein database, containing 97249 protein sequences. In addition, DDA raw files were subjected to de novo sequencing as previously described\u0026nbsp;\u003csup\u003e1\u003c/sup\u003e. As a result, a total of 232863 protein sequences were included in the microbial database. This database was combined with the standard \u003cem\u003eMus musculus\u003c/em\u003e proteome (https://www.uniprot.org/proteomes/UP000000589, accessed on 2023-04-27) to process acquired datasets.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSpectral library generation for DIA-NN analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpectral libraries for datasets acquired using DIA-, Slice-, Synchro- and midia-PASEF were generated using DIA-NN (v2.0.2) in library-free mode based on in silico–predicted spectra. Protein databases used for library prediction included a reduced microbial reference database, the standard proteomes of Ligilactobacillus murinus (UniProt UP000051612, accessed 2023-07-19), Salinibacter ruber (UP000008674, accessed 2023-07-19), Homo sapiens (UP000005640, accessed 2025-03-20), and contaminant sequences provided by DIA-NN. Peptides were generated by in silico digestion using Trypsin/P with up to two missed cleavages and N-terminal methionine excision enabled. Carbamidomethylation of cysteine was set as a fixed modification, while methionine oxidation, N-terminal acetylation, and heavy isotopic labeling of arginine (+10.0083 Da) and lysine (+8.0142 Da) were specified as variable modifications, allowing a maximum of two variable modifications per peptide. Peptide lengths were restricted to 7–30 amino acids. Precursor m/z values were limited to 250–1200 with charge states of 2–4, and fragment ions were considered in the range of 100–1700 m/z. DIA-NN’s deep learning models were used to predict fragment intensities, retention times, and ion mobility values. Proteotypicity was set to “Protein names (from FASTA),” and heuristic protein inference was disabled. mass accuracy for MS1 and MS2 was set to automatic determination. The resulting predicted library comprised 12,412,855 precursors corresponding to 109,255 protein groups. To prevent incorrect retention-time alignment, DIA-NN searches were performed separately for each acquisition method and gradient length.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor the colonic samples presented in Fig. 4, pooled peptide samples (generated by combining all proximal or distal colon samples) were repeatedly analyzed using both DIA- and Slice-PASEF to monitor LC–MS performance and to support generation of a sample-specific experimental spectral library. In addition, fractionated pooled samples were acquired in DIA-PASEF mode to further increase spectral coverage. These datasets were processed using DIA-NN (v2.3.0) with a predicted library comprising 26,374,776 precursors derived from a reduced microbial database (232,863 protein sequences), the standard Mus musculus proteome (UniProt UP000000589, accessed 2023-04-27), and DIA-NN contaminant sequences. Trypsin/P digestion was specified with a maximum of one missed cleavage and N-terminal methionine excision enabled. Carbamidomethylation of cysteine was used as a fixed modification, while methionine oxidation and N-terminal acetylation were allowed as variable modifications, permitting a maximum of one variable modification per peptide. Peptide length was restricted to 7–30 amino acids, precursor m/z values were set to 250–1200 with charge states of 2–3, and fragment ions were considered between 100 and 1700 m/z. Proteotypicity was set to “Isoform IDs,” and protein inference was disabled. The resulting experimental spectral library contained 184,884 precursors corresponding to 85,044 protein groups and was used for DIA- and Slice-PASEF analysis of individual samples.\u003c/p\u003e\n\u003cp\u003eFor all searches conducted in DIA-NN, RT-dependent cross-run normalization and QuantUMS\u003csup\u003e38\u003c/sup\u003e (high precision) options were selected for quantification. All DIA-NN search outputs were further processed with the R package, DIA-NN (https://github.com/vdemichev/diann-rpackage), to calculate the MaxLFQ\u003csup\u003e39\u003c/sup\u003e quantitative intensities for all identified peptides and protein groups with q-value ≤ 0.01 as criteria at precursor and protein group levels.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSample preparation for PentaPASEF benchmarking\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe human fecal background was generated by pooling fecal peptide samples produced in-house from multiple independent studies to ensure consistency and maximize proteomic complexity. The pooled fecal peptide mixture was diluted with 0.15% (w/v) n-dodecyl-β-D-maltoside (DDM; Sigma-Aldrich) to the desired concentrations.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eLigilactobacillus murinus\u003c/em\u003e (DSM 20452) and \u003cem\u003eSalinibacter ruber\u003c/em\u003e (DSM 13855) were obtained from DSMZ (Braunschweig, Germany), and bacterial culture conditions were as previously reported\u003csup\u003e1\u003c/sup\u003e. Peptides derived from the two species were mixed at a fixed ratio of 1:3 (\u003cem\u003eL. murinus\u003c/em\u003e:\u003cem\u003eS. ruber\u003c/em\u003e) and subsequently diluted with 0.15% DDM to generate the required spike-in levels. Owing to the extended acquisition period, samples corresponding to different spike-in concentrations were further aliquoted and stored at −20 °C. For each acquisition method and LC gradient, a fresh aliquot was thawed immediately before analysis to minimize sample loss and variability associated with prolonged storage in the autosampler.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTaxonomic and functional annotation and quantification\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTaxonomic annotation was done using Unipept web application\u003csup\u003e40\u003c/sup\u003e with default settings. peptides present in the peptide quantification output of each method and gradient were submitted for annotation and the resulting taxa assignments were merged back to the quantification matrix where peptide intensities from acuqisition triplicates were averaged. Taxa were considered confident when supported by at least three taxa-specific peptide entries, and taxa abundances were quantified by summing intensities of taxa-assigned peptides. The microbial protein databases used in this manuscript were annotated using EggNOG-mapper\u003csup\u003e41\u003c/sup\u003e (http://eggnog-mapper.embl.de/) with default settings to retrieve potential functions and pathways. The annotated function was merged with the peptide quantification matrix using Meta4P\u003csup\u003e42\u003c/sup\u003e. The quantification of a functional entry was done by summing up the peptide intensities where the corresponding proteins were annotated to this functional entry.\u003c/p\u003e\n\u003cp\u003eFor the \u003cem\u003ein vivo\u003c/em\u003e dataset, peptides identified by DIA- and Slice-PASEF were annotated separately using Unipept and merged with the corresponding peptide quantification matrices. Confident taxa required at least three taxa-specific peptide entries, and taxa intensities were computed by summing peptide intensities after applying a content-wise detection filter, retaining peptides only if they were detected in all replicates of at least one group (colonic region × genotype × timepoint). The taxon-specific functions were analyzed using Meta4P with peptide quantification data from DIA-NN, confident taxonomic annotation, and functional annotation files from EggNOG-mapper as inputs. Quantification of taxon-specific functions was performed by summing the peptide intensities assigned to taxa and associated with specific functions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePairwise comparisons presented in Supplementary Fig. 3 were performed using a two-sided Wilcoxon rank-sum test (Mann–Whitney U test) in R (version 4.5.2). Differential abundance analysis of species, species–pathway, and host proteins (Fig. 4) were performed using the limma (version 3.66.0) framework. For each colonic content and time point, log₂-transformed abundances were modeled using linear models, genotype contrasts were tested using empirical Bayes–moderated t-statistics with robust variance estimation, and multiple testing correction was applied independently for each contrast within each colonic content and time point using the Benjamini–Hochberg procedure.\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD073779 (reviewers can access the dataset by logging in to the PRIDE website using [email protected], sassword: rIphcfnHZd7p) and PXD073688 (reviewers can access the dataset by logging in to the PRIDE website using [email protected], sassword: lYIVr7eVH5US). Source data are provided with this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe codes essential for the study have been deposited to Github repository\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknoledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank all members of the Systems Biology of Pain laboratory, Division of Pharmacology and Toxicology, University of Vienna, for valuable discussions and suggestions. We thank the kind help from Prof. Stefan Tenzer’s group (University Medical Center of the Johannes-Gutenberg University Mainz) on our midia-PASEF data. We thank Biognosys AG for the access to Spectronaut. This research was funded in part by the University of Vienna and by the Austrian Science Fund (FWF; 10.55776/P35856 and 10.55776/P36554 to M.S.). E.U., D.A., and D.H. were funded by Deutsche Forschungsgemeinschaft (DFG; 395357507 and 469152594). For open access purposes, the author has applied a CC BY public copyright license to any author accepted manuscript version arising from this submission. The computational results presented have been achieved in part using the Austrian Scientific Cluster (ASC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: D.G.V.; Experimental design: D.G.V., F.X., G.M., R.K.R.K., D.A., E.U., and D.H.; Biochemistry and mass spectrometry: F.X., R.K.R.K., and G.M.; Sample collection and preparation: F.X., R.K.R.K., E.U., and D.A.; Data processing: F.X., L.U., and R.K.R.K.; Formal analysis: F.X.; Writing: F.X., and D.G.V.; Study supervision: D.G.V.; Project administration: M.S. and D.G.V.; Funding acquisition: M.S. and D.G.V. All authors edited and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM.S. received research awards and travel support from the German Pain Society (DGSS) both of which were sponsored by Astellas Pharma GmbH (Germany). M.S. received research awards from the Austrian Pain Society. M.S. received a one-time consulting honorarium from Grunenthal GmbH (Germany). None of these sources influenced the content of this study, and M.S. declares no conflict of interest. D.G.V. and M.S. have an ongoing scientific collaboration with Bruker (Center of Excellence for Metaproteomics University of Vienna - Bruker). F.X., R.K.R.K., L.U., E.U., D.A., D.H., M.S., and D.G.V. declare no competing interests. G.M. is an employee of Bruker Austria.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"Reference","content":"\u003col\u003e\n \u003cli\u003eXian F\u003cem\u003e, et al.\u003c/em\u003e Ultra-sensitive metaproteomics redefines the dark metaproteome, uncovering host-microbiome interactions and drug targets in intestinal diseases. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 6644 (2025).\u003c/li\u003e\n \u003cli\u003eLi L\u003cem\u003e, et al.\u003c/em\u003e Systematic metaproteomics mapping reveals functional and ecological landscapes of Ex vivo human gut microbiota responses to therapeutic drugs. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 9383 (2025).\u003c/li\u003e\n \u003cli\u003eWang J, Lang T, Shen J, Dai J, Tian L, Wang X. Core Gut Bacteria Analysis of Healthy Mice. \u003cem\u003eFront Microbiol\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 887 (2019).\u003c/li\u003e\n \u003cli\u003eArmengaud J. Metaproteomics to understand how microbiota function: The crystal ball predicts a promising future. \u003cem\u003eEnvironmental Microbiology\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 115-125 (2023).\u003c/li\u003e\n \u003cli\u003eClemente JC, Ursell LK, Parfrey LW, Knight R. The impact of the gut microbiota on human health: an integrative view. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e148\u003c/strong\u003e, 1258-1270 (2012).\u003c/li\u003e\n \u003cli\u003eStamboulian M, Canderan J, Ye Y. Metaproteomics as a tool for studying the protein landscape of human-gut bacterial species. \u003cem\u003ePLoS Comput Biol\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, e1009397 (2022).\u003c/li\u003e\n \u003cli\u003eDuan H\u003cem\u003e, et al.\u003c/em\u003e Assessing the Dark Field of Metaproteome. \u003cem\u003eAnal Chem\u003c/em\u003e \u003cstrong\u003e94\u003c/strong\u003e, 15648-15654 (2022).\u003c/li\u003e\n \u003cli\u003eDumas T\u003cem\u003e, et al.\u003c/em\u003e The astounding exhaustiveness and speed of the Astral mass analyzer for highly complex samples is a quantum leap in the functional analysis of microbiomes. \u003cem\u003eMicrobiome\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 46 (2024).\u003c/li\u003e\n \u003cli\u003eKleikamp HBC\u003cem\u003e, et al.\u003c/em\u003e Database-independent de novo metaproteomics of complex microbial communities. \u003cem\u003eCell Syst\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 375-383.e375 (2021).\u003c/li\u003e\n \u003cli\u003eMeier F\u003cem\u003e, et al.\u003c/em\u003e Parallel Accumulation-Serial Fragmentation (PASEF): Multiplying Sequencing Speed and Sensitivity by Synchronized Scans in a Trapped Ion Mobility Device. \u003cem\u003eJ Proteome Res\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 5378-5387 (2015).\u003c/li\u003e\n \u003cli\u003eMeier F\u003cem\u003e, et al.\u003c/em\u003e diaPASEF: parallel accumulation-serial fragmentation combined with data-independent acquisition. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 1229-1236 (2020).\u003c/li\u003e\n \u003cli\u003eSinn LR\u003cem\u003e, et al.\u003c/em\u003e Slice-PASEF: Maximising Ion Utilisation in LC-MS Proteomics. \u003cem\u003ebioRxiv\u003c/em\u003e, (2025).\u003c/li\u003e\n \u003cli\u003eSkowronek P\u003cem\u003e, et al.\u003c/em\u003e Synchro-PASEF Allows Precursor-Specific Fragment Ion Extraction and Interference Removal in Data-Independent Acquisition. \u003cem\u003eMolecular \u0026amp; Cellular Proteomics\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 100489 (2023).\u003c/li\u003e\n \u003cli\u003eDistler U\u003cem\u003e, et al.\u003c/em\u003e midiaPASEF maximizes information content in data-independent acquisition proteomics. \u003cem\u003ebioRxiv\u003c/em\u003e, 2023.2001.2030.526204 (2023).\u003c/li\u003e\n \u003cli\u003eSteigenberger B\u003cem\u003e, et al.\u003c/em\u003e Benefits of Collisional Cross Section Assisted Precursor Selection (caps-PASEF) for Cross-linking Mass Spectrometry. \u003cem\u003eMol Cell Proteomics\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 1677-1687 (2020).\u003c/li\u003e\n \u003cli\u003eGomez-Zepeda D\u003cem\u003e, et al.\u003c/em\u003e Thunder-DDA-PASEF enables high-coverage immunopeptidomics and is boosted by MS(2)Rescore with MS(2)PIP timsTOF fragmentation prediction model. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 2288 (2024).\u003c/li\u003e\n \u003cli\u003eMendes ML, Borrmann KF, Dittmar G. Eleven shades of PASEF. \u003cem\u003eExpert Review of Proteomics\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 367-376 (2024).\u003c/li\u003e\n \u003cli\u003eGomez-Varela D, Xian F, Grundtner S, Sondermann JR, Carta G, Schmidt M. Increasing taxonomic and functional characterization of host-microbiome interactions by DIA-PASEF metaproteomics. \u003cem\u003eFront Microbiol\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 1258703 (2023).\u003c/li\u003e\n \u003cli\u003eWang A\u003cem\u003e, et al.\u003c/em\u003e Assessing fecal metaproteomics workflow and small protein recovery using DDA and DIA PASEF mass spectrometry. \u003cem\u003eMicrobiome Research Reports\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 39 (2024).\u003c/li\u003e\n \u003cli\u003eRajczewski AT\u003cem\u003e, et al.\u003c/em\u003e Data-Independent Acquisition Mass Spectrometry as a Tool for Metaproteomics: Interlaboratory Comparison Using a Model Microbiome. \u003cem\u003eProteomics\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, e202400187 (2025).\u003c/li\u003e\n \u003cli\u003eSamodova D\u003cem\u003e, et al.\u003c/em\u003e Salivary proteomics and metaproteomics identifies distinct molecular and taxonomic signatures of type-2 diabetes. \u003cem\u003eMicrobiome\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 5 (2025).\u003c/li\u003e\n \u003cli\u003eXian F\u003cem\u003e, et al.\u003c/em\u003e Ultra-sensitive metaproteomics redefines the dark metaproteome, uncovering host-microbiome interactions and drug targets in intestinal diseases. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 6644 (2025).\u003c/li\u003e\n \u003cli\u003eUrbauer E\u003cem\u003e, et al.\u003c/em\u003e Mitochondrial perturbation in the intestine causes microbiota-dependent injury and gene signatures discriminative of inflammatory disease. \u003cem\u003eCell Host \u0026amp; Microbe\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 1347-1364.e1310 (2024).\u003c/li\u003e\n \u003cli\u003eArmengaud J. Metaproteomics to understand how microbiota function: The crystal ball predicts a promising future. \u003cem\u003eEnviron Microbiol\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 115-125 (2023).\u003c/li\u003e\n \u003cli\u003eKong AT, Leprevost FV, Avtonomov DM, Mellacheruvu D, Nesvizhskii AI. MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry\u0026ndash;based proteomics. \u003cem\u003eNature Methods\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 513-520 (2017).\u003c/li\u003e\n \u003cli\u003eDemichev V, Messner CB, Vernardis SI, Lilley KS, Ralser M. DIA-NN: neural networks and interference correction enable deep proteome coverage in high throughput. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 41-44 (2020).\u003c/li\u003e\n \u003cli\u003eVan Den Bossche T\u003cem\u003e, et al.\u003c/em\u003e The microbiologist\u0026apos;s guide to metaproteomics. \u003cem\u003eImeta\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, e70031 (2025).\u003c/li\u003e\n \u003cli\u003eArs\u0026egrave;ne-Ploetze F, Bertin PN, Carapito C. Proteomic tools to decipher microbial community structure and functioning. \u003cem\u003eEnvironmental Science and Pollution Research\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 13599-13612 (2015).\u003c/li\u003e\n \u003cli\u003eNebauer DJ, Pearson LA, Neilan BA. Critical steps in an environmental metaproteomics workflow. \u003cem\u003eEnvironmental Microbiology\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, e16637 (2024).\u003c/li\u003e\n \u003cli\u003eBaker JL, Mark Welch JL, Kauffman KM, McLean JS, He X. The oral microbiome: diversity, biogeography and human health. \u003cem\u003eNat Rev Microbiol\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 89-104 (2024).\u003c/li\u003e\n \u003cli\u003eYuan J\u003cem\u003e, et al.\u003c/em\u003e OSaMPle workflow for salivary metaproteomics analysis reveals dysbiosis in inflammatory bowel disease patients. \u003cem\u003enpj Biofilms and Microbiomes\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 63 (2025).\u003c/li\u003e\n \u003cli\u003eZhou X\u003cem\u003e, et al.\u003c/em\u003e Longitudinal profiling of the microbiome at four body sites reveals core stability and individualized dynamics during health and disease. \u003cem\u003eCell Host \u0026amp; Microbe\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 506-526.e509 (2024).\u003c/li\u003e\n \u003cli\u003eChen Y, Knight R, Gallo RL. Evolving approaches to profiling the microbiome in skin disease. \u003cem\u003eFront Immunol\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 1151527 (2023).\u003c/li\u003e\n \u003cli\u003eBerger E\u003cem\u003e, et al.\u003c/em\u003e Mitochondrial function controls intestinal epithelial stemness and proliferation. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 13171 (2016).\u003c/li\u003e\n \u003cli\u003eKong AT, Leprevost FV, Avtonomov DM, Mellacheruvu D, Nesvizhskii AI. MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry-based proteomics. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 513-520 (2017).\u003c/li\u003e\n \u003cli\u003eAlmeida A\u003cem\u003e, et al.\u003c/em\u003e A unified catalog of 204,938 reference genomes from the human gut microbiome. \u003cem\u003eNature Biotechnology\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 105-114 (2021).\u003c/li\u003e\n \u003cli\u003eKall L, Canterbury JD, Weston J, Noble WS, MacCoss MJ. Semi-supervised learning for peptide identification from shotgun proteomics datasets. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 923-925 (2007).\u003c/li\u003e\n \u003cli\u003eKistner F, Grossmann JL, Sinn LR, Demichev V. QuantUMS: uncertainty minimisation enables confident quantification in proteomics. \u003cem\u003ebioRxiv\u003c/em\u003e, 2023.2006.2020.545604 (2023).\u003c/li\u003e\n \u003cli\u003eCox J, Hein MY, Luber CA, Paron I, Nagaraj N, Mann M. Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. \u003cem\u003eMol Cell Proteomics\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 2513-2526 (2014).\u003c/li\u003e\n \u003cli\u003eMesuere B, Willems T, Van der Jeugt F, Devreese B, Vandamme P, Dawyndt P. Unipept web services for metaproteomics analysis. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 1746-1748 (2016).\u003c/li\u003e\n \u003cli\u003eCantalapiedra CP, Hernandez-Plaza A, Letunic I, Bork P, Huerta-Cepas J. eggNOG-mapper v2: Functional Annotation, Orthology Assignments, and Domain Prediction at the Metagenomic Scale. \u003cem\u003eMol Biol Evol\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 5825-5829 (2021).\u003c/li\u003e\n \u003cli\u003ePorcheddu M, Abbondio M, De Diego L, Uzzau S, Tanca A. Meta4P: A User-Friendly Tool to Parse Label-Free Quantitative Metaproteomic Data and Taxonomic/Functional Annotations. \u003cem\u003eJournal of Proteome Research\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 2109-2113 (2023).\u003cstrong\u003e\u003cbr\u003e\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"University of Vienna","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Metaproteomics, Ion mobility mass spectrometry, PASEF, Microbiome","lastPublishedDoi":"10.21203/rs.3.rs-9103948/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9103948/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eMetaproteomics enables direct measurement of functional activity in complex microbial communities but remains technically challenging due to the extreme complexity and dynamic range of microbiome samples. Trapped ion mobility spectrometry coupled with parallel accumulation–serial fragmentation (PASEF) offers unique advantages in general proteomics and in the analysis of complex metaproteomes. In recent years, PASEF has diversified into multiple acquisition modes. However, the lack of systematic side-by-side evaluations under matched experimental conditions, together with the predominant use of low- to medium-complexity benchmark samples, hampers informed selection of acquisition strategies. Here, we benchmark five PASEF acquisition modes (DDA-, DIA-, Slice-, Synchro-, and midia-PASEF) using a complex fecal peptide background spiked with defined bacterial references. Performance was evaluated across three chromatographic gradients and input levels, comprising 540 LC–MS acquisitions processed using a single computational workflow. DIA-based strategies consistently achieved greater peptide and protein coverage than DDA-PASEF, particularly for low-abundance microbial features. Slice- and DIA-PASEF exhibited the lowest quantitative variability, minimal ratio compression, and the most consistent scaling of species abundances, while functional profiling revealed an expanded dynamic range of enzyme annotations. Application to a murine colonic injury model demonstrated that DIA- and Slice-PASEF capture highly concordant host and microbial responses. Together, this study provides a unified evaluation of PASEF acquisition strategies and illustrates how acquisition choices influence sensitivity, reproducibility, and functional resolution in proteomic analyses.\u003c/p\u003e","manuscriptTitle":"Systematic evaluation of PASEF acquisition strategies in complex metaproteomes","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-13 10:11:58","doi":"10.21203/rs.3.rs-9103948/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0eb24b5d-9b68-4e32-831b-84dae0d2c503","owner":[],"postedDate":"March 13th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":64388305,"name":"Analytical Biochemistry"},{"id":64388306,"name":"Systems Biology"},{"id":64388307,"name":"General Microbiology"}],"tags":[],"updatedAt":"2026-03-13T10:11:58+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-13 10:11:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9103948","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9103948","identity":"rs-9103948","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}