Leaping over the blood-brain barrier: DNA methylation as a link between peripheral and central immune systems

Leaping over the blood-brain barrier: DNA methylation as a link between peripheral and central immune systems | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (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];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Leaping over the blood-brain barrier: DNA methylation as a link between peripheral and central immune systems View ORCID Profile Mandy Meijer , Maggie Po Yuan Fu , View ORCID Profile Erick Isaac Navarro-Delgado , Hannah-Ruth Engelbrecht , Gustavo Turecki , Meingold Hiu-ming Chan , Michael Steffen Kobor doi: https://doi.org/10.1101/2025.08.25.672204 Mandy Meijer 1 Department of Medical Genetics, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada 2 British Columbia Children’s Hospital Research Institute, University of British Columbia , Vancouver, BC, Canada 3 Centre for Molecular Medicine and Therapeutics, University of British Columbia , Vancouver, BC, Canada 4 Edwin S. H. Leong Centre for Healthy Aging, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mandy Meijer Maggie Po Yuan Fu 1 Department of Medical Genetics, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada 2 British Columbia Children’s Hospital Research Institute, University of British Columbia , Vancouver, BC, Canada 3 Centre for Molecular Medicine and Therapeutics, University of British Columbia , Vancouver, BC, Canada 4 Edwin S. H. Leong Centre for Healthy Aging, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada BSc Find this author on Google Scholar Find this author on PubMed Search for this author on this site Erick Isaac Navarro-Delgado 1 Department of Medical Genetics, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada 2 British Columbia Children’s Hospital Research Institute, University of British Columbia , Vancouver, BC, Canada 3 Centre for Molecular Medicine and Therapeutics, University of British Columbia , Vancouver, BC, Canada 4 Edwin S. H. Leong Centre for Healthy Aging, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada BSc Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Erick Isaac Navarro-Delgado Hannah-Ruth Engelbrecht 1 Department of Medical Genetics, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada 2 British Columbia Children’s Hospital Research Institute, University of British Columbia , Vancouver, BC, Canada 3 Centre for Molecular Medicine and Therapeutics, University of British Columbia , Vancouver, BC, Canada 4 Edwin S. H. Leong Centre for Healthy Aging, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada MSc Find this author on Google Scholar Find this author on PubMed Search for this author on this site Gustavo Turecki 5 Department of Psychiatry, McGill Group for Suicide Studies, Douglas Mental Health University Institute, McGill University , Montreal, QC, Canada PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Meingold Hiu-ming Chan 1 Department of Medical Genetics, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada 2 British Columbia Children’s Hospital Research Institute, University of British Columbia , Vancouver, BC, Canada 3 Centre for Molecular Medicine and Therapeutics, University of British Columbia , Vancouver, BC, Canada 4 Edwin S. H. Leong Centre for Healthy Aging, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Michael Steffen Kobor 1 Department of Medical Genetics, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada 2 British Columbia Children’s Hospital Research Institute, University of British Columbia , Vancouver, BC, Canada 3 Centre for Molecular Medicine and Therapeutics, University of British Columbia , Vancouver, BC, Canada 4 Edwin S. H. Leong Centre for Healthy Aging, Faculty of Medicine, University of British Columbia , Vancouver, BC, Canada PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: Michael.kobor{at}ubc.ca Abstract Full Text Info/History Metrics Supplementary material Preview PDF Abstract The majority of existing DNA methylation (DNAm) studies have used peripheral surrogate tissues to research molecular mechanisms underlying brain-related traits. Epigenetic processes in the brain have yet to be fully elucidated by DNAm findings from peripheral tissues, as these processes are complex to disentangle from DNAm tissue- and cell type-specificity. Furthermore, distinct brain cell types play important roles in brain health and disease. Building on previous findings of high blood-brain correlations at some DNAm sites, we specifically aimed to 1) identify single DNAm sites associated with DNAm-estimated brain cell type proportions in both the frontal brain and peripheral blood, 2) to combine blood DNAm sites to predict brain cell type proportions through multivariate models, and 3) to examine the role of biological factors associated with blood DNAm, age and epigenetic age acceleration (EAA) on brain cell type proportions. Epigenome-wide association studies for seven distinct estimated brain cell type proportions in matched frontal brain and peripheral blood samples (n=104) revealed that ∼7% of brain cell type-associated DNAm sites had correlating DNAm levels in peripheral blood (p<0.05). Six DNAm sites in peripheral blood were associated with endothelial and stromal brain cell type proportions (False Discovery Rate<0.05). Brain cell type proportion predictions trained with machine learning approaches using peripheral blood DNAm showed a correlation (p<0.05) with microglia and astrocyte proportions estimated through cell deconvolution using brain DNAm. Estimated blood immune cell type proportions were significantly associated with estimated brain cell type proportions. Lastly, brain EAA was associated with different brain and blood cell types proportions (p<0.05). These results show that DNAm levels in peripheral blood can inform brain cell type proportions, even though DNAm patterns are tissue-specific. The correlations between DNAm profiles specific to immune cell types in blood and brain highlights the link between the peripheral immune system and immune functions in the central nervous system. 1. Background Brain-related disorders and diseases are common and have high lifetime prevalences. For instance, one in eight people in the world live with a psychiatric disorder ( 1 ), and the lifetime risk of receiving a diagnosis of Alzheimer’s disease (AD) or Parkinson’s disease (PD) is up to one in five ( 2 – 4 ). Furthermore, psychiatric and neurodegenerative conditions are often co-occurring ( 5 – 8 ). These brain-related conditions are multifactorial, involving both genetic as well as environmental factors in their development and persistence ( 9 – 13 ). Considerable knowledge about the molecular underpinnings of many brain-related conditions is lacking, and effective treatments for most of these conditions are not available yet. Previous studies on postmortem brain samples have identified potential roles of specific brain cell types in different brain-related conditions. Neurons are the main brain cells that receive and send information; glial cells have a supporting role in the central nervous system: microglia and astrocytes are important effector cells in the central immune system ( 14 ), whereas the main role of oligodendrocytes is to myelinate neurons ( 15 ).The involvement of brain cell differences in brain-related conditions are showcased across a range of cell types: alterations in microglia proportions and gene expression levels have been observed for autism spectrum disorder, schizophrenia, depression, and AD ( 16 – 20 ). Similar differences have been observed for astrocytes in depression ( 17 ), and for neuronal cells in autism spectrum disorder, schizophrenia, depression, and PD ( 16 , 21 – 23 ). Furthermore, based on genome- and epigenome-wide association studies, genetic and epigenetic sites associated with brain-related conditioned are enriched in different brain cell types: schizophrenia has been shown to be associated with median spiny neurons ( 24 ) and GABA- and glutamatergic neurons ( 25 ). ADHD has been associated with excitatory neurons and astrocytes ( 26 , 27 ), while PD has been genetically associated with oligodendrocytes ( 25 ). In short, different brain-related conditions have been linked to changes in a distinct palette of brain cell types proportions and functions. Cell type identity and regulation is modulated, in part, by DNA methylation (DNAm). Thus, DNAm patterns are cell type-specific ( 28 ), and more stable than transcriptomics signatures ( 29 – 31 ). However, most DNAm-association studies on brain-related outcomes are performed in surrogate tissues such as blood due to less invasive collection protocols ( 32 ), which might not fully reflect brain DNAm patterns. Fortunately, these peripheral surrogate tissue DNAm levels sometimes show statistically significant correlations with those in brain tissue: global correlations between peripheral blood and brain tissue across all measured DNAm site is reported to be between Pearson’s r=0.860 and r=0.892 ( 33 – 36 ). When individual peripheral DNAm sites were measured, 6-11% showed moderate to high blood-brain DNAm level correlations ( 33 – 36 ). Despite the cell type and tissue-specificity of DNAm, the correlation of DNAm sites across tissues suggests that some peripheral blood DNAm sites can still be informative for brain DNAm levels. On that basis, single DNAm sites with high blood-brain correlation could be informative biomarkers for brain-related phenotypes. Supporting this notion, epigenome-wide association studies have identified DNAm sites that are significantly associated with different brain-related disorders and disorders using peripheral samples ( 37 – 43 ). What remains unclear is what drives the single DNAm site correlation between blood and brain. A possible linkage is the underlying cell types in blood and brain, which might biologically interact with each other. One could argue that peripheral blood and brain DNAm levels show correlated levels due to a biological cross-talk and physical interaction of the peripheral and central nervous system ( 44 ). It has been shown recently that the body-brain axis informs the brain of an emerging inflammatory response, and that the brain can modulate the peripheral immune response. This hypothesis is supported by the fact that there is a high comorbidity and genetic correlation between brain-related traits and autoimmune diseases ( 45 – 50 ), and many brain-related conditions are thought to have an immune system component, such as schizophrenia ( 51 ), bipolar disorder ( 52 ), ADHD ( 53 ), and PD ( 54 ). Alterations in immune functioning are not only associated with brain-related conditions, but also a notable part of the aging process. As such, age is often identified as an important risk factor related to the development of psychiatric disorders (e.g., depression ( 55 )), and neurological disorders (e.g., dementias ( 2 )). Studying the molecular markers of ageing can thus provide a useful avenue to interrogate brain-related conditions, which increase with ageing ( 56 ), and their associations with immune system functioning, which decreases with ageing ( 57 ). It is important to delineate the effects of ageing on brain cell function and proportions, which can be studied through the lens of biological ageing. Accelerated biological ageing, or epigenetic age acceleration (EAA), is the relative rate of ageing for an individual, given their chronological age, estimated based on DNAm data ( 58 – 60 ). EAA has been associated with age-related brain diseases and traits ( 61 – 63 ), as well as multiple psychiatric disorders and neurodegenerative diseases (e.g., ( 61 – 71 )). It has been previously shown that differences in peripheral blood epigenetic age is driven by Natural Killer (NK) cell and T cell immunosenescence ( 72 ), and that oligodendrocytes are associated with increased epigenetic ageing in brain ( 73 ). In summary, the exact underlying molecular mechanisms of many brain-related conditions are unknown, but changes in brain cell type proportions have been identified to potentially underlie different disorders and diseases. However, access to brain samples is limited, hampering research into the molecular mechanisms of brain-related traits in humans, because most studies are relying on peripheral blood samples. The biological interpretation of the existing correlation between peripheral blood and brain DNAm could be better understood when taking cell type proportions into account. In addition to relying on single DNAm sites, multivariate analyses could also reveal new informative patterns linking blood DNAm and brain cell types that could be left undiscovered by single site approaches. Lastly, it is of importance to establish the relation of age in these correlations. Therefore, the aim of this study is to investigate how DNAm in peripheral blood can inform brain cell type-specific epigenetic signals. Specifically, we are investigating brain DNAm in the frontal cortex, as this brain region is affected in many brain-related conditions, including ADHD, schizophrenia, and bipolar disorder ( 74 ). This is crucial for studying brain-related conditions and brain health in general in-depth based on DNAm derived from surrogate tissues. We aimed to 1) identify single DNAm sites associated with estimated brain cell type proportions in both the frontal brain and peripheral blood; 2) combine blood-based DNAm sites to inform brain cell type proportions through multivariate models, and 3) to identify the role of chronological age and biological brain ageing on brain cell type proportions. 2. Methods 2.1 Samples Publicly available datasets containing paired blood and brain DNA methylation (DNAm) were used (GSE214901; GSE95049; GSE111165; GSE59685) ( Table S1 ). Detailed descriptions of the individual datasets have been published previously ( 33 – 36 ). All brain samples taken from the frontal brain were selected. All brain samples that had matched blood samples from the same individual were retained. When one dataset included more than one frontal brain sample from the same individual, one random sample was selected. In total, we included 104 matched frontal brain and peripheral blood samples from the same individuals in the current study. 2.2 DNA methylation DNA was extracted from the whole blood and brain samples as previously described ( 33 – 36 ). DNAm was assessed by either the Infinium® HumanMethylation450 BeadChip (450K array; Illumina, San Diego, CA, USA) or the Infinium® MethylationEPIC BeadChip (EPIC array; Illumina). Raw data, both for each dataset and tissue type, were processed separately in R v4.2.4. For dataset 1, 2, and 4 ( Table S1 ), raw iDAT files were available and processed, for dataset 3, raw beta values were available and processed. A complete overview of DNA methylation preprocessing and quality control (QC) can be found in the Supplementary methods . In total, 366,570 and 370,721 DNAm sites were retained in blood and brain for subsequent analysis, respectively, after extensive QC. 2.3 Cell type deconvolution Brain cell type proportions were estimated through epigenomic deconvolution in the HiBED (Hierachical Brain Extended Deconvolution) v0.99.6 R package ( 75 ). Proportions were estimated for astrocytes, endothelial cells, GABA- and glutamatergic neurons, microglia, oligodendrocytes, and stromal cells. The HiBED reference panel consists of two layers with categories (L1, L2A, L2B, and L2C), containing 621 probes in total. In our dataset, 580 probes were present: 75 out of 81 for L1, 174 out of 183 for L2A, 219 out of 237 for L2B, and 112 out of 120 for L2C. Blood cell type proportions were estimated for twelve different immune cell types based on the IDentifying Optimal Libraries (IDOL) extended reference dataset ( 76 ): basophils, eosinophils, neutrophils, monocytes, B naïve cells, B memory cells, CD4 + naïve cells, CD4 + memory cells, CD8 + naïve cells, CD8 + memory cells, NK cells, and T regulatory cells. Betas from dataset 1 through 4 were quantile normalized with the betas present in the reference panels, and probes used to estimate cell type proportions were selected based on t-tests, with a p-value cut-off of 0.05, and minimum delta beta of 0.05. One hundred probes were selected for each cell type estimated. Cell type proportions were estimated using constrained projection ( 76 , 77 ). Interquartile ranges (IQR) for the probes selected for blood cell type proportion estimations were calculated. 2.4 Epigenome-wide association meta-analyses To identify whether single DNAm sites in blood inform brain cell type proportions, we performed epigenome-wide association studies (EWAS) for each brain cell type, separately, in both brain and blood tissue, by using linear regression models corrected for age, sex and cell type proportions. To make results comparable across datasets 1 through 4 and across tissues (i.e., brain and blood), we also corrected for biological variability likely attributable to tissue-specific cell type proportions for the tissue involved in the EWAS through surrogate variable analysis (SVA). An overview of the number of surrogate variables for each analysis can be found in Table S2 . Results for each brain cell type in each tissue were sample size-based meta-analyzed using METAL ( 78 ). FDR-adjusted p-values were calculated, and DNAm sites with an FDR0.05 was present, as well as having a consistent direction of effects (i.e., direction of effect for a single DNAm site was constantly negative or positive in all four datasets; Figure S1 ). Inflation factors for each EWAS were calculated as the square root of the expected median of a chi-squared statistics distribution divided by the median of the chi-statistics with one degree of freedom. To determine whether percentages of correlations between blood and brain for the significant DNAm sites were more or less than expected by chance, we performed a permutation analysis. Brain DNAm sites were randomly shuffled, and the analyses were performed again. 2.5 Multivariate brain cell type proportion predictors based on blood DNAm data To assess whether a multivariate predictor of brain cell type proportions could be created out of DNAm sites in peripheral blood, Ridge regression was performed on a training dataset and validated in a testing dataset. The training superset consisted of datasets 1, 2, and 3. Prior to Ridge regression, the training superset was renormalized together with dasen and ComBat . Dataset 4 was reserved as test dataset and renormalized by itself. Additionally, Variable Methylated Regions (VMRs) were generated using the RAMEN R package ( 79 ) using the default parameters. Following, a Random Forest (RF) model performed using the ranger R v0.14.1 package with 500 trees ( 80 ). See Supplementary methods for more details on the Ridge regression, VMRs and RF. 2.6 Principal component analysis To assess whether DNAm sites in peripheral blood with high blood-brain correlation can inform brain cell types, we used the 429 probes from the brain cell type reference panel that were present in both the frontal brain and blood supersets. We used this specific set of probes as these have been identified to distinguish between different brain cell types in datasets independent of ours ( 75 ). We calculated statistically significant correlations (p<0.05) between these DNAm sites in both blood and brain. We used these, rather than the DNAm sites significantly associated with brain cell type proportions in brain tissue, as the reference probes reflect a hierarchical granularity (i.e., containing probes informative for both glial cells (Layer 1) and e.g., microglia (Layer 2), and are derived from single-cell measures ( 75 ). The DNAm sites were clustered by cell type (e.g., for microglia all sites annotated to either microglia or glial cells were clustered), and a principal component analysis (PCA) was performed on the derived DNAm sites in both brain and blood. 2.7 Epigenetic ageing in the brain Epigenetic frontal brain age and epigenetic age acceleration (EAA) was calculated with the Cortical Clock ( 60 ), with the median age of the training set at 56.53 years (see Supplementary methods ). 3. Results 3.1 Dataset description We combined four publicly available datasets of brain and blood DNAm (GSE214901; GSE95049; GSE111165; GSE59685), and retained in total 104 frontal cortex and blood samples that were matched from the same individuals ( Table S1 ). The mean age of brain and blood samples at collection were 76.76 years and 74.04 years, respectively (Standard Deviation (SD) = 21.62 and 20.30, respectively), where the median age difference for collection was 2.00 years (range: 0-16 years). In total, 44% of the individuals were male. Using this combined dataset, we tested whether DNAm in peripheral blood can inform on brain cell type-specific epigenetic patterns, using three distinct yet overlapping broad approaches, ranging from association studies to machine learning based predictors to determining the role of biological variables ( Figure 1 ). Download figure Open in new tab Figure 1 Study overview. Aim 1: DNA methylation (DNAm) was assessed in paired frontal cortex and blood samples (upper row, left and right panels). Brain cell type proportions (astrocytes, endothelial cells, stromal cells, oligodendrocytes, GABAergic neurons, glutamatergic neurons and microglia) were estimated based on epigenomic deconvolution in frontal cortex samples (upper row, middle panel). Epigenome-wide association studies (EWAS) for each brain cell type proportion were performed in both brain and blood samples (upper row, left and right panels). The results of these EWAS were compared across tissues. Aim 2: Then, the dataset was split in a training and testing dataset and all DNAm sites in blood were used to predict brain cell type compositions through Ridge regression (middle row, left panel). We also assessed whether a subset of DNAm sites with highly correlated blood-brain DNAm levels were associated with brain cell type proportions (middle row, middle panel). Because of the consistent immune brain cell types-blood DNAm associations, blood immune cell type proportions were estimated, and correlated with estimated brain cell types (middle row, right panel). Aim 3: Then, the influence of age on these correlations was assessed (bottom row, left panel). Finally, biological ageing and epigenetic age acceleration in the cortex was estimated based on cortical epigenetic clocks, and the role of different brain and blood cell types on biological brain ageing was assessed (bottom row, right panel). As the foundation for our study, we first used bioinformatic deconvolution of DNAm data derived from the frontal brain to estimate proportions of seven different cell types in the all 104 samples. As expected, oligodendrocytes were the most abundant cell type present in the brain samples (mean = 38.15%, inter-quartile range (IQR) = 9.36%), followed by glutamatergic neurons (mean = 25.72%, IQR = 6.35) ( Figure 2 ).We noted a considerable variability for both of these estimated cell types in each of the individual datasets, whereas the other five cells types had a tighter distribution, perhaps owing to their overall lower proportions. Download figure Open in new tab Figure 2 Oligodendrocyte proportions are most abundant and variable in the frontal brain. Brain cell type proportions (astrocytes, endothelial cells, GABAergic neurons, glutamatergic neurons, microglial, oligodendrocytes, and stromal cells) in the frontal brain were estimated in the combined datasets based on the epigenomic deconvolution tool HiBED. Different colours indicate the individual sub-datasets. 3.2 DNA methylation associations with brain cell type proportions were tissue-specific Given the role of DNAm in the establishment and maintenance of cellular identity, we next tested which individual DNAm sites in bulk brain DNAm profiles were associated with brain cell type proportions in the frontal cortex specifically. To account for heterogeneity within datasets and to identify DNAm sites with consistent direction of effect over datasets, we stratified by dataset and performed a meta-EWAS ( Figure 3 , Table 1 , Figure S1-5, Table S3-4 ). In total, 370,721 DNAm sites were retained for brain tissue after extensive quality control. In the frontal brain, after removal of DNAm sites with inconsistent direction of effect for the association with brain cell types across the four different datasets, the number of brain DNAm sites significantly associated with brain cell type proportions ranged from five for stromal cells to 28,618 for oligodendrocytes, however, effect sizes of these associations per cell type were similar ( Figure S5 ). We also identified 27 GABA-DNAm associations, 155 endothelial cell-DNAm associations, 2,045 astrocyte-DNAm associations, 3,134 microglia-DNAm associations, and 16,614 GLU-DNAm associations. The number of associations detected was correlated with variability in cell type proportion. Interestingly, there was a considerable amount of overlap between GLU-DNAm and oligodendrocyte-DNAm associations ( Figure 3 ). Of all the significant DNAm-brain cell type proportion associations (p<0.05), 89% of associations with GABAergic neurons had a positive regression coefficient in the meta-EWAS, whereas the associations for astrocytes, glutamatergic neurons, and stromal cells primarily had negative regression coefficients (66%, 73%, and 80%, respectively). Download figure Open in new tab Figure 3 Overlapping DNA methylation associations for distinct brain cell types. A meta-EWAS from bulk frontal cortex-derived DNA methylation profiles was performed for seven different brain cell type proportions. Each cell type is represented as a row. On the left, the horizontal bar plot indicates the number of significantly associated DNAm sites per cell type as identified by the meta-EWAS. On the top, the vertical bar plot shows the intersections (i.e., significantly associated DNA methylation sites which are overlapping) per combination of cell types. The filled black dots connected by lines in the rows indicate which combinations of cell types are displayed, with single dots reflecting the unique DNAm sites significantly associated with that specific cell type. View this table: View inline View popup Table 1 Overview of significant findings for each EWAS for brain cell type proportion in frontal brain tissue We next asked whether any of the DNAm sites that were included in the publicly available cell type deconvolution algorithm actually were associated with brain cell type proportions. On average, 1.35% of the DNAm sites that showed an association with any type of brain cell type proportion were present in the list of 580 DNAm sites from the HiBED brain deconvolution reference panel. Conversely, 58% of these 580 DNAm sites (n = 336) included in the reference panel were identified as significantly associated with brain cell types in our study. These 336 DNAm sites were in the top 10% most significant EWAS results for each individual cell type. Of all the significant DNAm associations with any brain cell type proportion, 7.83% had significantly correlating DNAm levels between matched frontal brain and blood samples (p<0.05; Pearson’s r: |0.21 – 0.73|). After FDR-correction, only 0.33% of these DNAm sites showed a significant correlation (FDR<0.05). This is more than expected by chance as permutation analysis reveals that 7.39% (SD=0.15%) and 0% (SD=0.0059%) of random probes shows a correlation between blood and brain with both statistical cut-offs, respectively. Next, we assessed whether DNAm levels in peripheral blood are also associated with brain cell type proportions, and thus whether these sites could help inform on brain cell type proportions in association with a brain phenotype based on EWAS in peripheral blood. For this, we again used a similar meta-EWAS approach as for the brain DNAm-specific associations. In total, 366,570 DNAm sites were retained for blood after extensive quality control. In contrast to the brain-specific analysis, we identified only a limited number of statistically significant blood DNAm associations with distinct brain cell type proportions ( Figure S2-4 , Table S3 ): three DNAm sites in blood were significantly associated with endothelial cells and three DNAm sites were associated with stromal cells (FDR < 0.05). No DNAm sites from peripheral blood were associated with the other five predicted brain cell type proportions. To assess whether the significant associations were unique to blood or shared with the brain results, we compared the results obtained from the EWASs performed in blood and brain tissue: three out of the six blood-based DNAm sites showed blood-brain correlation (Pearson’s r: 0.257 - 0.900) ( Table S4 ), but none of these DNAm sites were significantly associated with any brain cell type proportion in the EWAS performed on DNAm retrieved from the frontal cortex ( Table S3 ). Indeed, and perhaps not surprisingly given the distinct nature of the two tissues, concordance in general between the effect sizes at the level of single DNAm sites retrieved from the meta-EWAS in blood and frontal brain tissue was low ( Figure S6 ). 3.3 Predicted brain cell type proportions in peripheral blood showed the strongest correlation with estimated immune brain cell types As an alternative to the limited information content offered by individual blood DNAm sites on brain cell proportions, we tested whether a more complex machine-learning based combination of DNAm sites might be more suitable for this purpose. To develop a predictive model of brain cell type proportions in the frontal brain based on a combination of blood DNAm sites, we used Ridge regression on a training superset consisting of combined and renormalized datasets 1, 2 and 3, whereas dataset 4 was independently normalized and used as a testing dataset ( Table S1 ), to ensure a set of samples with a diverse age rage. The root mean squared error (RMSE) of the model trained using Ridge regression for the different cell type percentages were the following: astrocytes = 1.503, endothelial = 0.643, GABA = 1.676, GLU = 9.597, microglia = 3.701, oligodendrocyte = 16.071, stromal cells = 0.157, indicating moderate-to-high errors. Rather than functioning equally for all predicted brain cell types, we found some nuanced, and potentially physiologically meaningful patterns when applying our model to the test dataset. Specifically, Ridge-predicted brain cell type proportions from blood DNAm significantly correlated with estimated brain immune cells: astrocytes (Pearson’s r = -0.607, p = 0.0127) and microglia (Pearson’s r = 0.525, p = 0.0366) ( Figure 4 ). In contrast, even though we identified significant blood DNAm-endothelial and DNAm-stromal cell associations in the EWAS meta-analysis, and the RMSE of the Ridge regression was low [RMSE=0.157], all proportions were estimated to be 0 for endothelial and stromal cell types in the testing dataset. Download figure Open in new tab Figure 4 Ridge regression performance for brain cell type proportion estimation based on peripheral blood DNA methylation (DNAm). Estimated brain cell type proportions based on epigenomic deconvolution in DNAm of frontal cortex samples is plotted on the y-axis. Brain cell type proportions estimated based on the optimal models from Ridge regression are plotted the x-axis. Every dot represents a single sample from the renormalized test datasets 4 ( Table S1 ). The Pearson’s correlation (r) and p-value (p) are given for every brain cell type. While the performance of our initial model was particularly good in predicting immune cell types, we considered potential improvements by reducing redundancy and increasing statistical power through the creation of variable methylated regions (VMRs), taking both the pairwise DNAm probe correlation and variability of DNAm probes into account. In addition, we included a Random Forest model for the VMRs to compare model performance. Given that the use of VMRs did not improve model performance for either Ridge regression ( Figure S7 ) or Random Forest models ( Figure S8 ), we performed a sensitivity analysis to determine whether the models and the data are suitable for cell type predictions. We used the same data and model to predict brain cell type proportions based on DNAm obtained from frontal brain tissue. Here, we were able to predict brain cell type composition in the independent testing dataset ( Figure S9 ), concluding that the combination of all DNAm sites in blood hold limited information on brain cell types, except for the immune brain cell types. 3.4 Blood-brain correlating DNAm sites were associated with brain cell type proportions in both brain and blood The previous analyses focused either on single sites (EWAS) or on a combination of sites without biological or statistical preselection of features (i.e., machine learning approaches). Next, we assessed whether a biologically relevant preselected combination of DNAm sites in peripheral blood can inform brain cell types. Out of the 429 probes from the brain cell type reference panel – which are selected to best characterize cell type proportion – that were present in all four frontal brain and blood datasets, we used 48 DNAm sites that showed a significant correlation (Pearson’s r range: |0.198 - 0.537|) between frontal brain and peripheral blood (p<0.05) ( Table S5 ). First, to confirm that these 48 DNAm sites out of the 429 sites in brain were still predictive of brain cell type proportions, we conducted a PCA on these sites and found that the first PC derived from them in frontal brain, grouped by brain cell type as indicated by the brain reference panel, explained 98-100% of the variance for each cell type proportion. All first PCs were significantly correlated with their corresponding brain cell type proportions (Pearson’s r: |0.571 - 0.969|, mean absolute correlation = 0.793, SD = 0.142). This indicated that using 11.2% (48 out of 429) of probes in the brain cell type reference panel was still informative for brain cell type composition in the frontal brain ( Figure S10 ). Then, we conducted the same analysis using blood DNAm and revealed that the first PC derived from the 48 blood-brain correlating probes in blood also explained variance of 98 to 100% in brain cell type proportions. This first PC was correlated with astrocytes (Pearson’s r = -0.793, p = 0.0340), endothelial cells (Pearson’s r = 0.400, p = 2.540*10 -5 ), microglia (Pearson’s r = 0.247, p = 0.0115), and stromal cells (Pearson’s r = 0.341, p = 4.00*10 -5 ) ( Figure S11 ). This is in line with the earlier findings that blood DNAm patterns are most predictive for immune-related cell types in the brain. 3.5 Correlations between blood immune cell types and brain cell type proportions are dependent on chronological age Given the biological cross-talk between the peripheral immune system and brain, we next added an even more coarse, yet perhaps most physiologically relevant, layer to our suite of analyses - that being the direct comparison of predicted cell type proportions between blood and brain. With epigenomic deconvolution based on an extended blood cell type panel, we estimated twelve different immune cell types in blood ( Figure S12 ). Eosinophiles in blood were positively correlated with oligodendrocytes in brain (Pearson’s r=0.216, p-value=0.02712; Figure S13 ). Furthermore, NK cells showed the strongest correlation with multiple cell types: Endothelial cells (Pearson’s r=-0.303, p-value=0.00176), stromal cells (Pearson’s r=- 0.334, p-value=0.000520), astrocytes (Pearson’s r=-0.242, p-value=0.0133), oligodendrocytes (Pearson’s r=0.290, p-value=0.00283), GABAergic neurons (Pearson’s r=-0.223, p-value=0.0230). After multiple test correction, the negative correlation between proportions of NK cells and stromal cells remained significant (FDR=0.0437), however, since proportions of both cell types are relatively low, caution should be taken with the biological interpretation of this correlation. Next, we investigated the effect of chronological age on the correlation between blood and brain cell type proportions ( Figure 5 ). Most previously reported correlations disappeared, except for NK - endothelial cells (Pearson’s r=-0.232, p-value=0.0180) and NK – stromal cells (Pearson’s r=-0.265, p-value=0.00660). While correcting for chronological age, regulatory T cells were significantly associated with glutamatergic neurons (Pearson’s r=-0.256, p-value=0.00875). None of these associations survived multiple test correction (FDR>0.05). Download figure Open in new tab Figure 5 Blood and brain cell type proportions were correlated. Pearson’s correlations were calculated between brain cell type proportions (y-axis) and blood cell types (x-axis) both residualized on age from matched individuals, based on epigenomic deconvolution estimates. Size and colour of the circles indicate size of the Pearson’s correlation (bright red: Pearson’s r = 1, white: Pearson’s r=0, dark blue: Pearson’s r=-1). *p<0.05, **p<0.01, ***p<0.001 3.6 Blood type proportions could serve as a peripheral biomarker for biological brain ageing Correlations between blood and brain cell type proportions were partially sensitive to age. However, age measures were based on chronological age, rather than epigenetic age in the brain, which is associated with multiple brain-related conditions. We leveraged the cortical epigenetic clock to estimate epigenetic brain age ( Figure S14 ). We assessed which cell type proportions could be driving cortical EAA. Cortical EAA was significantly positively associated with endothelial cell type proportions (Pearson’s r = 0.215, p = 0.0282), and negatively with microglial proportions (Pearson’s r = -0.221, p = 0.0241) ( Figure 6 ). Importantly, none of the DNAm sites used to predict frontal brain age were present in the brain cell type reference panel, but results could be driven by one potential outlier with cortical EAA of 50 (without outlier: Endothelial Pearson’s r = -0.383, p = 0.701; Microglia Pearson’s r = -0.152, p = 0.125). Download figure Open in new tab Figure 6 Biological ageing was correlated with brain and blood cell type proportions. Brain and blood cell type proportions as estimated with epigenomic deconvolution are plotted on the y-axis. Cortical epigenetic age acceleration (EAA), defined as epigenetic age residualized on chronological age is plotted on the x-axis. Every dot represents a single sample in the combined dataset. The Pearson’s correlation (r) and p-value (p) are given for every brain cell type. Previous research has shown that peripheral epigenetic signatures, including immune-related signatures, are associated with structural variation in brain and brain age acceleration ( 81 , 82 ). We tested whether blood cell type proportions estimated based on peripheral epigenetic profiles could also serve as peripheral markers of brain ageing by assessing the association between epigenetic cortical ageing (EAA) and peripheral immune cell types in blood. Cortical EAA was negatively associated with NK cell type proportions (Pearson’s r = -0.243, p = 0.0130) and positively associated with neutrophil proportions (Pearson’s r = 0.208, p = 0.0344) ( Figure 6 ). These associations were not driven by the potential outlier (without outlier: NK Pearson’s r = -0.231, p = 0.0190; Neutrophil Pearson’s r = 0.200, p = 0.0433). 4. Discussion In the current study, we investigated whether DNAm profiles in peripheral blood could be informative for biological mechanisms underlying brain health-related outcomes, and whether DNAm profiles in peripheral blood were correlated with estimated brain cell type proportions, given their implications in neurodevelopment, neurodegeneration, and psychiatric disorders. This study showed that based on single DNAm sites and multivariate models, peripheral blood DNAm patterns displayed the strongest association with brain immune cell types, specifically astrocytes, microglia, and stromal cells, which also were most strongly linked to brain ageing. We identified DNAm sites in frontal brain tissue that were associated with estimated brain cell type proportions. DNAm levels were consistently negatively associated with astrocytes, microglia, and stromal cell proportions. Associations between GABAergic neuron proportions and DNAm were mainly positive in the frontal brain. In line with our findings, it has been previously shown that GABAergic neurons have higher global DNAm levels than glutamatergic neurons in the prefrontal cortex ( 83 ), and that glial cells (including astrocytes and microglia) have the lowest DNAm profiles compared to neuronal cell populations in both the human ( 83 , 84 ) and mouse brain ( 85 ). Furthermore, we observed that considerable DNAm sites were associated with both oligodendrocytes and glutamatergic neurons. This overlap can best be explained by the fact that both cell type proportions are most common and negatively correlated with each other. In other words, these considerably overlapping DNAm sites for oligodendrocytes and glutamatergic neurons have opposite effect sizes for each cell type. Next, we identified that ∼7% (3,764 sites) of the DNAm sites that were associated with brain cell type proportions had significantly correlated DNAm levels across brain and blood. Previous studies show that 6-11% of all measured probes from either array exhibit peripheral blood-brain DNAm correlation ( 33 – 36 ). This indicates that DNAm sites in the frontal brain associated with estimated brain cell type proportions exhibited comparable levels of peripheral blood-brain correlation with what can be expected from the current array platforms available. We used both single site and multivariate modelling to investigate whether DNAm sites in peripheral blood could predict brain cell type proportions. These approaches revealed that peripheral blood DNAm sites were predictive of immune cell type proportions in the brain, specifically astrocytes, microglia, and stromal cells. It has been shown that peripheral immune cells are in close contact with nerves in the central nervous system, indicating the biological plausibility for the central and peripheral immune system to communicate by paracrine signalling ( 86 ). Peripheral immune cells have been shown to be responsive to neurotransmitters and hormones released from the brain into the periphery, and the peripheral immune system also influences these neuroendocrine functions ( 87 , 88 ). Therefore, it is well possible that the peripheral immune system may be involved in brain health. Indeed, the peripheral immune system is linked to different brain-related conditions such as schizophrenia ( 89 ), autism spectrum disorder ( 90 ), and AD ( 91 ). The fact that peripheral blood DNAm is informative for estimating immune brain cell type proportions is interesting as potential means of using peripheral blood DNAm to study the brain’s immune system, which is difficult to sample. In peripheral blood, we identified a total of six statistically significant single DNAm sites associated with brain endothelial and stromal cell proportions, of which three are annotated to genes with neural-related functions. CTNNA2 enables actin filament binding activity and regulates neuron migration as well as neuron projection ( 92 ); LRRTM2 is predicted to be involved in regulation of postsynaptic density assembly ( 93 ), and ALG10 is involved in the formation of N-linked glycosylation of proteins, which is important for neuronal signalling and resting glial cell homeostasis in the brain ( 94 ). Indeed, all three neural-related genes show high expression levels in the brain for both neuronal and glial cells, according to the Human Protein Atlas ( proteinatlas.org , accessed on 17/07/2024). Two out of the three genes ( LRRTM2 and ALG10 ) are expressed in peripheral blood immune cells and lymphoid endothelial cells under baseline conditions. These single DNAm sites also showed statistically significant peripheral blood-brain DNAm correlation. However, the significant associations with endothelial and stromal cell proportions found in blood DNAm at these sites were not found in brain DNAm. It is worth noting that the HiBED deconvolution tool for endothelial and stromal cells specifically was constructed based on cells isolated from infant cord peripheral blood ( 75 ). Because of the shared developmental origins of stromal and endothelial cells in blood and brain, the tissue differences can potentially drive a similar DNAm profile in the cell library and between peripheral and stromal cells in our combined dataset. These cells are responsible for lining and supporting the peripheral blood vessels that make up the blood-brain-barrier (BBB) and the statistically significant association of these brain-annotated DNAm sites in peripheral blood with brain cell types may reflect true biological correlations. Hence, DNAm levels in both brain and peripheral blood may reveal different relevant factors, such as BBB dynamics, for studying brain-related phenotypes, especially on a cell type-specific level. Summarized, the DNAm associations highlight that some DNAm from both the frontal brain as well as peripheral blood are informative for estimated brain cell type proportions, potentially reflecting brain function. In addition to the three neural-related DNAm sites, we also identified three peripheral blood-based DNAm sites annotated to immune-related proteins to be associated with endothelial and stromal cell proportions: ALOX5 is mainly expressed in bone-marrow derived cells, and plays a crucial role in leukotriene synthesis, which is an important regulator of inflammation ( 95 ). Collecting Subfamily Member 1 ( COLEC11 ) has been shown to play an important part in facilitating recognition and removal of potential pathogens ( 96 ). HLA-J represents a pseudogene, i.e., a gene structurally representing a gene, but not coding for a protein, possibly derived from HLA-A, which plays a central role in the CD8T cell immune response to pathogens ( 97 ). The association between blood DNAm of immune-related genes with estimated brain cell type proportions could be reflecting the shared immune function of peripheral and central nervous system tissues. The presence of gene expression levels in both peripheral immune system cells and brain tissue - specifically endothelial cells - support this notion (Human Protein Atlas). Systemic inflammation could alter stromal and endothelial cell proportions: endothelial cells line the blood vessels of the BBB, and stromal cells support the integrity of the BBB through lymphatic drainage of molecules and cells away from the brain. Systemic peripheral inflammation can result in disruptive BBB changes, often on the histological level such as cell damage or tight junction changes, ultimately breaking down the barrier ( 98 , 99 ). Therefore, these findings reflect how peripheral DNAm profiles might indicate a biological and physical interaction of the immune system with the brain. We not only identified an association between peripheral DNAm profiles and immune brain cell type proportions, but we also identified associations between estimated peripheral immune cell proportions and different estimated brain cell types. We identified a negative association between NK cells in peripheral blood and brain endothelial and stromal cells after accounting for the subjects’ age. The association between NK cells and endothelial cells is of interest, as endothelial cells lining the BBB can express chemokines that are ligands for receptors on NK cells, thereby attracting peripheral NK cells to the brain ( 100 , 101 ). NK cells have the capacity to disrupt the BBB and subsequently endothelial cells in case of brain inflammation ( 101 , 102 ). We also identified a negative association for regulatory T cells in peripheral blood and glutamatergic neurons in the frontal brain. Glutamate has proliferative and activating effects on regulatory T cells ( 103 , 104 ). The proliferative effects of glutamate, released by glutamatergic neurons, on regulatory T cells would, however, not explain the negative correlation between cell type proportions. One should also consider receptor activity and abundance as well as cell activity and negative feedback loops when studying neurotransmitter release and its effect on its downstream ligand (i.e., glutamate receptors on T cells). Different peripheral immune cell types, specifically the T- and NK cells discussed above, have been found to underlie epigenetic ageing measured with peripheral blood DNAm ( 72 ). Therefore, we investigated whether a similar association between epigenetic ageing and cell types, especially immune cells, also exists in the frontal brain. There was an association between epigenetic ageing of the brain and different estimated brain cell types. We identified that in line with previous research, NK cells and neutrophils are negatively and positively associated with EAA and other measures of epigenetic age such PhenoAge and Pace of Ageing ( 72 , 105 , 106 ). Furthermore, the negative association with microglia proportions and EAA is in line with previous literature reporting that lower levels of microglia are regulating cognitive decline ( 107 , 108 ) and that microglial EAA has been observed in different neurological conditions ( 109 ). In summary, both brain cells and peripheral immune cells are involved in brain health in complex ways, and EAA measures in both tissues could be used to further interrogate molecular mechanisms underlying brain health. The results from the current study should be interpreted with their own limitations. First of all, we made use of publicly available databases, and all studies included different types of brain tissue from demographically different populations. For example, in the combined dataset we created, we included both postmortem and live brain tissue. It has been shown that postmortem and live brain tissue exhibit different gene expression patterns ( 110 ) and post-transcriptional RNA modifications ( 111 ), and it could thus be argued that epigenetic modifications, including DNAm might also be different between the two types of brain tissues. However, we did not observe substantial differences in cell type proportions between the individual cohorts, suggesting that the two types of tissues were similar enough for comparisons, especially because the normalization and batch correction steps of our preprocessing pipeline should have removed these effects. This is further supported by a previous study showing that the percentage of DNAm was stable up to 72 hours postmortem interval ( 112 ). Second, we applied a machine learning algorithm to train a predictor of brain cell type proportions using peripheral DNAm profiles. We used three datasets with a total n=88 for training, and validated our findings in an independent dataset (n=16). The estimates obtained based on a sample size of N=104 distinct individuals with diverse demographics in the current study may be improved by an increase in sample size. The importance of statistical power is also underlined by our findings that the highest number of significantly associated DNAm sites were identified in oligodendrocytes, the most variable cell type in the frontal brain. Increases in sample size would also increase statistical power. Third, we analyzed samples from the frontal brain only, but different brain regions show different brain cell type proportions ( 19 , 113 , 114 ), the findings presented in the current study are thus only representative of the frontal brain. To draw robust conclusions about other brain regions, these brain regions should be studied in depth. Fourth, there was a variety in age range (e.g., only elderly individuals or a mix of adolescents and adults) across the sub-cohorts included in the current study. Our downstream data analyses were corrected for age, and therefore represents general effects across a 50-year age span. Thus, when considering research cohorts with a narrower age-range, it should be noted that correlations in one age group might not exist in another age group. Lastly, it is well-known that the commercial arrays do not distinguish between DNAm and DNA hydroxymethylation (DNAhm) ( 115 ). Given the importance of DNAhm in the brain ( 116 ), important brain cell type-associations could have been masked, as both epigenetic modifications show cell type-specific patterns in the brain ( 117 ). Furthermore, the difference between DNAm and DNAhm levels in both brain and blood would be important to take into account when investigating epigenetic correlations between tissues. Based on the results presented in our study, we provide future directions for using DNAm as a tool to create cross-tissue biomarkers for unmeasured biological variables ( 118 ), such as brain cell type proportions. A larger sample size and the combination of both DNAm and DNAhm is recommended when aiming to develop such biomarkers. We also provided a short-list of DNAm sites which are overlapping on the Illumina 450K and EPICv1 arrays showing high peripheral blood-brain correlation and associations with brain cell type proportions ( Supplementary Table 5 ). These DNAm sites can be used to identify the potential role of brain cell type proportions involved in the phenotype of interest when performing association analyses in peripheral blood. Our study also showcased an example workflow to analyze diverse sets of cohorts. All in all, these results and recommendations open up the avenue for developing predictors of brain cell type proportions with larger sample size without the need to access brain tissue, allowing for more in-depth studies of biological pathways that underpin brain health and other brain-related phenotypes such as psychiatric/neurodevelopmental disorders and neurodegenerative diseases. Conclusions In summary, the results presented here support the hypothesis that the peripheral immune system shows biological interactions with the central nervous system, specifically in immune cell types, as reflected by DNAm. We showed that peripheral DNAm can be informative for immune brain cell type proportion estimates. Given the existing DNAm correlation between peripheral blood and brain, the usage of a stable cell identity measure, i.e., DNAm, in peripheral blood as biomarker of some brain cell type proportions could be plausible. Lastly, we showed that brain ageing was associated with specific brain and blood cell type proportions. Altogether, these results and recommendations could add to the interpretation of peripheral blood-derived DNAm-associations in the context of brain-related outcomes. Declarations Ethics approval and consent to participate All samples used in the current manuscript were publicly available (datasets). Each individual study has obtained ethics approval and consent from patients and/or family to participate. The following information is obtained from the original publications. GSE214901: “The study protocol was approved by the Ethics Committee of the University of Fukui, Japan (Assurance no. 20200028), Yamaguchi University School of Medicine, Japan (Assurance no. 2020–202), and Sugita Genpaku Memorial Obama Municipal Hospital (Assurance no. 2–7). Moreover, this study was carried out in accordance with the Declaration of Helsinki and the Ethical Guidelines for Clinical Studies of the Ministry of Health, Labour and Welfare of Japan. All participants provided either written informed consent or both informed consent and assent.” (Nishitani, Translation Psychiatry, 2023). GSE95049: “The Research Ethics Board at the Douglas Mental Health University Institute approved the project. Signed informed consent was obtained for each subject from next of kin.” (Farré, Epigenetics & Chromatin, 2015). GSE111165: “This study was approved by the University of Iowa’s Human Subjects Research Institution Review Board. Written informed consent was obtained.” (Braun, Translation psychiatry, 2019). GSE59685: “Ethical approval for the study was provided by the NHS South East London REC 3. Matched blood samples collected before death were available for a subset of individuals (Supplementary Tables 1 and 2) as part of the Alzheimer’s Research UK funded study “Biomarkers of AD Neurodegeneration”, with informed consent according to the Declaration of Helsinki (1991). For validation purposes STG and PFC tissue was obtained from 144 individuals archived in the Mount Sinai Alzheimer’s Disease and Schizophrenia Brain Bank ( http://icahn.mssm.edu/research/labs/neuropathology-and-brain-banking ) and EC, STG and PFC samples from an additional 62 individuals archived in the Thomas Willis Oxford Brain Collection ( http://www.medsci.ox.ac.uk/optima/information-for-patients-and-the-public/the-thomas-willis-oxford-brain-collection ).” (Lunnon, Nature Neuroscience, 2014). Availability of data and materials The datasets analysed during the current study are available in the Gene Expression Omnibus repository, https://www.ncbi.nlm.nih.gov/geo/ . We used the datasets with the following GEO Accession codes: GSE214901; GSE95049; GSE111165; GSE59685. Competing interests The authors declare that they have no competing interest. Funding MM was supported by a personal grant from the Dutch Research Council (NWO/ZonMW): Rubicon (grant no. 04520232320009). MM, EIND, and MHC were supported by personal grants from the Social Exposome Cluster (“Society to Cell” Clyde Hertzman Memorial Fellowship). Authors’ contributions MM conceptualized and designed the study, performed analyses, visualized, interpret, wrote the manuscript, and acquired funding. MPF, EIND, and HRE supported analyses related to cell type proportions, machine learning approaches, and brain ageing, respectively. MPF, EIND, HRE, and MHC interpret results and revised the manuscript. MSK provided supervision and computational resources. All authors read and approved the final manuscript. Acknowledgements We would like to thank Keegan Korthauer for her helpful insights on performing machine learning analyses. This research utilized the FlowSorted.BloodExtended.EPIC software packages developed at Dartmouth College, which are governed by the licensing terms provided by Dartmouth Technology Transfer. Funder Information Declared Dutch Research Council, https://ror.org/04jsz6e67 , 04520232320009 Abbreviations AD Alzheimer’s disease ADHD Attention-deficit/hyperactivity disorder ALG10 Alpha-1,2-Glucosyltransferase ALOX5 Arachidonate 5-Lipoxygenase BBB Blood-brain-barrier COLLEC11 Collectin Subfamily Member 11 CTNNA2 Catenin Alpha-2 DNAm DNA methylation DNAhm DNA hydroxymethylation EAA Epigenetic age acceleration EWAS Epigenome-wide association study FDR False Discovery Rate HiBED Hierarchical Brain Extended Deconvolution HLA-J Major Histocompatibility Complex, Class I, J IQR Interquartile range LRRTM2 Leucine Rich Repeat Transmembrane Neuronal 2 NK cells Natural Killer cells PCA Principal Component analysis PD Parkinson’s disease RF Random Forest RMSE Root Mean Squared Error SD Standard deviation References 1. ↵ IHME IoHMaE . Global Health Data Exchange (GHDx) 2022 [ 2. ↵ Chene G , Beiser A , Au R , Preis SR , Wolf PA , Dufouil C , et al. Gender and incidence of dementia in the Framingham Heart Study from mid-adult life . Alzheimers Dement . 2015 ; 11 ( 3 ): 310 – 20 . OpenUrl CrossRef PubMed 3. Seshadri S , Wolf PA , Beiser A , Au R , McNulty K , White R , et al. Lifetime risk of dementia and Alzheimer’s disease. The impact of mortality on risk estimates in the Framingham Study . Neurology . 1997 ; 49 ( 6 ): 1498 – 504 . OpenUrl CrossRef PubMed 4. ↵ Driver JA , Logroscino G , Gaziano JM , Kurth T . Incidence and remaining lifetime risk of Parkinson disease in advanced age . Neurology . 2009 ; 72 ( 5 ): 432 – 8 . OpenUrl CrossRef PubMed 5. ↵ Zubenko GS , Zubenko WN , McPherson S , Spoor E , Marin DB , Farlow MR , et al. A collaborative study of the emergence and clinical features of the major depressive syndrome of Alzheimer’s disease . Am J Psychiatry . 2003 ; 160 ( 5 ): 857 – 66 . OpenUrl CrossRef PubMed Web of Science 6. Starkstein SE , Mizrahi R , Power BD . Depression in Alzheimer’s disease: phenomenology, clinical correlates and treatment . Int Rev Psychiatry . 2008 ; 20 ( 4 ): 382 – 8 . OpenUrl CrossRef PubMed Web of Science 7. Burchill E , Watson CJ , Fanshawe JB , Badenoch JB , Rengasamy E , Ghanem DA , et al. The impact of psychiatric comorbidity on Parkinson’s disease outcomes: a systematic review and meta-analysis . Lancet Reg Health Eur . 2024 ; 39 : 100870 . OpenUrl PubMed 8. ↵ McGrath JJ , Lim CCW , Plana-Ripoll O , Holtz Y , Agerbo E , Momen NC , et al. Comorbidity within mental disorders: a comprehensive analysis based on 145 990 survey respondents from 27 countries . Epidemiol Psychiatr Sci . 2020 ; 29 : e153 . OpenUrl CrossRef PubMed 9. ↵ Schmitt A , Malchow B , Hasan A , Falkai P . The impact of environmental factors in severe psychiatric disorders . Front Neurosci . 2014 ; 8 : 19 . OpenUrl CrossRef PubMed 10. Tsankova N , Renthal W , Kumar A , Nestler EJ . Epigenetic regulation in psychiatric disorders . Nat Rev Neurosci . 2007 ; 8 ( 5 ): 355 – 67 . OpenUrl CrossRef PubMed Web of Science 11. Andreassen OA , Hindley GFL , Frei O , Smeland OB . New insights from the last decade of research in psychiatric genetics: discoveries, challenges and clinical implications . World Psychiatry . 2023 ; 22 ( 1 ): 4 – 24 . OpenUrl CrossRef PubMed 12. Pihlstrom L , Wiethoff S , Houlden H . Genetics of neurodegenerative diseases: an overview . Handb Clin Neurol . 2017 ; 145 : 309 – 23 . OpenUrl CrossRef PubMed 13. ↵ Cannon JR , Greenamyre JT . The role of environmental exposures in neurodegeneration and neurodegenerative diseases . Toxicol Sci . 2011 ; 124 ( 2 ): 225 – 50 . OpenUrl CrossRef PubMed Web of Science 14. ↵ Greenhalgh AD , David S , Bennett FC . Immune cell regulation of glia during CNS injury and disease . Nat Rev Neurosci . 2020 ; 21 ( 3 ): 139 – 52 . OpenUrl PubMed 15. ↵ Bradl M , Lassmann H . Oligodendrocytes: biology and pathology . Acta Neuropathol . 2010 ; 119 ( 1 ): 37 – 53 . OpenUrl CrossRef PubMed Web of Science 16. ↵ Yap CX , Vo DD , Heffel MG , Bhattacharya A , Wen C , Yang Y , et al. Brain cell-type shifts in Alzheimer’s disease, autism, and schizophrenia interrogated using methylomics and genetics . Sci Adv . 2024 ; 10 ( 21 ): eadn7655 . OpenUrl CrossRef PubMed 17. ↵ Maitra M , Mitsuhashi H , Rahimian R , Chawla A , Yang J , Fiori LM , et al. Cell type specific transcriptomic differences in depression show similar patterns between males and females but implicate distinct cell types and genes . Nat Commun . 2023 ; 14 ( 1 ): 2912 . OpenUrl CrossRef PubMed 18. Saavedra J , Nascimento M , Liz MA , Cardoso I . Key brain cell interactions and contributions to the pathogenesis of Alzheimer’s disease . Front Cell Dev Biol . 2022 ; 10 : 1036123 . OpenUrl PubMed 19. ↵ Hannon E , Dempster EL , Davies JP , Chioza B , Blake GET , Burrage J , et al. Quantifying the proportion of different cell types in the human cortex using DNA methylation profiles . BMC Biol . 2024 ; 22 ( 1 ): 17 . OpenUrl CrossRef PubMed 20. ↵ Pak V , Adewale Q , Bzdok D , Dadar M , Zeighami Y , Iturria-Medina Y . Distinctive whole-brain cell types predict tissue damage patterns in thirteen neurodegenerative conditions . Elife . 2024 ; 12 . 21. ↵ Stockmeier CA , Rajkowska G . Cellular abnormalities in depression: evidence from postmortem brain tissue . Dialogues Clin Neurosci . 2004 ; 6 ( 2 ): 185 – 97 . OpenUrl PubMed 22. Nestler EJ , Barrot M , DiLeone RJ , Eisch AJ , Gold SJ , Monteggia LM . Neurobiology of depression . Neuron . 2002 ; 34 ( 1 ): 13 – 25 . OpenUrl CrossRef PubMed Web of Science 23. ↵ Wang Q , Wang M , Choi I , Sarrafha L , Liang M , Ho L , et al. Molecular profiling of human substantia nigra identifies diverse neuron types associated with vulnerability in Parkinson’s disease . Sci Adv . 2024 ; 10 ( 2 ): eadi8287 . OpenUrl CrossRef PubMed 24. ↵ Skene NG , Bryois J , Bakken TE , Breen G , Crowley JJ , Gaspar HA , et al. Genetic identification of brain cell types underlying schizophrenia . Nat Genet . 2018 ; 50 ( 6 ): 825 – 33 . OpenUrl CrossRef PubMed 25. ↵ Bryois J , Skene NG , Hansen TF , Kogelman LJA , Watson HJ , Liu Z , et al. Genetic identification of cell types underlying brain complex traits yields insights into the etiology of Parkinson’s disease . Nat Genet . 2020 ; 52 ( 5 ): 482 – 93 . OpenUrl CrossRef PubMed 26. ↵ Price KM , Wigg KG , Nigam A , Feng Y , Blokland K , Wilkinson M , et al. Identification of brain cell types underlying genetic association with word reading and correlated traits . Mol Psychiatry . 2023 ; 28 ( 4 ): 1719 – 30 . OpenUrl CrossRef PubMed 27. ↵ Meijer M , Franke B , Sandi C , Klein M . Epigenome-wide DNA methylation in externalizing behaviours: A review and combined analysis . Neurosci Biobehav Rev . 2023 ; 145 : 104997 . OpenUrl PubMed 28. ↵ Li E . Chromatin modification and epigenetic reprogramming in mammalian development . Nat Rev Genet . 2002 ; 3 ( 9 ): 662 – 73 . OpenUrl CrossRef PubMed Web of Science 29. ↵ Bird A . DNA methylation patterns and epigenetic memory . Genes Dev . 2002 ; 16 ( 1 ): 6 – 21 . OpenUrl FREE Full Text 30. Feinberg AP , Irizarry RA , Fradin D , Aryee MJ , Murakami P , Aspelund T , et al. Personalized epigenomic signatures that are stable over time and covary with body mass index . Sci Transl Med . 2010 ; 2 ( 49 ): 49ra67 . OpenUrl Abstract / FREE Full Text 31. ↵ Pascual-Ahuir A , Fita-Torro J , Proft M . Capturing and Understanding the Dynamics and Heterogeneity of Gene Expression in the Living Cell . Int J Mol Sci . 2020 ; 21 ( 21 ). 32. ↵ Nigg JT . Considerations toward an epigenetic and common pathways theory of mental disorder . J Psychopathol Clin Sci . 2023 ; 132 ( 3 ): 297 – 313 . OpenUrl PubMed 33. ↵ Edgar RD , Jones MJ , Meaney MJ , Turecki G , Kobor MS . BECon: a tool for interpreting DNA methylation findings from blood in the context of brain . Transl Psychiatry . 2017 ; 7 ( 8 ): e1187 . OpenUrl 34. Braun PR , Han S , Hing B , Nagahama Y , Gaul LN , Heinzman JT , et al. Genome-wide DNA methylation comparison between live human brain and peripheral tissues within individuals . Transl Psychiatry . 2019 ; 9 ( 1 ): 47 . OpenUrl PubMed 35. Nishitani S , Isozaki M , Yao A , Higashino Y , Yamauchi T , Kidoguchi M , et al. Cross-tissue correlations of genome-wide DNA methylation in Japanese live human brain and blood, saliva, and buccal epithelial tissues . Transl Psychiatry . 2023 ; 13 ( 1 ): 72 . OpenUrl PubMed 36. ↵ Hannon E , Lunnon K , Schalkwyk L , Mill J . Interindividual methylomic variation across blood, cortex, and cerebellum: implications for epigenetic studies of neurological and neuropsychiatric phenotypes . Epigenetics . 2015 ; 10 ( 11 ): 1024 – 32 . OpenUrl CrossRef PubMed 37. ↵ Rovira P , Sanchez-Mora C , Pagerols M , Richarte V , Corrales M , Fadeuilhe C , et al. Epigenome-wide association study of attention-deficit/hyperactivity disorder in adults . Transl Psychiatry . 2020 ; 10 ( 1 ): 199 . OpenUrl PubMed 38. Hannon E , Dempster EL , Mansell G , Burrage J , Bass N , Bohlken MM , et al. DNA methylation meta-analysis reveals cellular alterations in psychosis and markers of treatment-resistant schizophrenia . Elife . 2021 ; 10 . 39. Aberg KA , McClay JL , Nerella S , Clark S , Kumar G , Chen W , et al. Methylome-wide association study of schizophrenia: identifying blood biomarker signatures of environmental insults . JAMA Psychiatry . 2014 ; 71 ( 3 ): 255 – 64 . OpenUrl PubMed 40. Cordova-Palomera A , Fatjo-Vilas M , Gasto C , Navarro V , Krebs MO , Fananas L . Genome-wide methylation study on depression: differential methylation and variable methylation in monozygotic twins . Transl Psychiatry . 2015 ; 5 ( 4 ): e557 . OpenUrl PubMed 41. Dempster EL , Pidsley R , Schalkwyk LC , Owens S , Georgiades A , Kane F , et al. Disease-associated epigenetic changes in monozygotic twins discordant for schizophrenia and bipolar disorder . Hum Mol Genet . 2011 ; 20 ( 24 ): 4786 – 96 . OpenUrl CrossRef PubMed Web of Science 42. Roubroeks JAY , Smith AR , Smith RG , Pishva E , Ibrahim Z , Sattlecker M , et al. An epigenome-wide association study of Alzheimer’s disease blood highlights robust DNA hypermethylation in the HOXB6 gene . Neurobiol Aging . 2020 ; 95 : 26 – 45 . OpenUrl CrossRef PubMed 43. ↵ Moore K , McKnight AJ , Craig D , O’Neill F . Epigenome-wide association study for Parkinson’s disease . Neuromolecular Med . 2014 ; 16 ( 4 ): 845 – 55 . OpenUrl PubMed 44. ↵ Dantzer R . Neuroimmune Interactions: From the Brain to the Immune System and Vice Versa . Physiol Rev . 2018 ; 98 ( 1 ): 477 – 504 . OpenUrl CrossRef PubMed 45. ↵ Nudel R , Allesoe RL , Werge T , Thompson WK , Rasmussen S , Benros ME . An immunogenetic investigation of 30 autoimmune and autoinflammatory diseases and their links to psychiatric disorders in a nationwide sample . Immunology . 2023 ; 168 ( 4 ): 622 – 39 . OpenUrl CrossRef 46. Berk M , Williams LJ , Jacka FN , O’Neil A , Pasco JA , Moylan S , et al. So depression is an inflammatory disease, but where does the inflammation come from? BMC Med . 2013 ; 11 : 200 . 47. Tylee DS , Sun J , Hess JL , Tahir MA , Sharma E , Malik R , et al. Genetic correlations among psychiatric and immune-related phenotypes based on genome-wide association data . Am J Med Genet B Neuropsychiatr Genet . 2018 ; 177 ( 7 ): 641 – 57 . OpenUrl CrossRef PubMed 48. Collins JM , Keane JM , Deady C , Khashan AS , McCarthy FP , O’Keeffe GW , et al. Prenatal stress impacts foetal neurodevelopment: Temporal windows of gestational vulnerability . Neurosci Biobehav Rev . 2024 ; 164 : 105793 . OpenUrl PubMed 49. Jyonouchi H . Autism spectrum disorder and a possible role of anti-inflammatory treatments: experience in the pediatric allergy/immunology clinic . Front Psychiatry . 2024 ; 15 : 1333717 . OpenUrl PubMed 50. ↵ Enduru N , Fernandes BS , Bahrami S , Dai Y , Andreassen OA , Zhao Z . Genetic overlap between Alzheimer’s disease and immune-mediated diseases: an atlas of shared genetic determinants and biological convergence . Mol Psychiatry . 2024 . 51. ↵ Ermakov EA , Melamud MM , Buneva VN , Ivanova SA . Immune System Abnormalities in Schizophrenia: An Integrative View and Translational Perspectives . Front Psychiatry . 2022 ; 13 : 880568 . OpenUrl PubMed 52. ↵ Saccaro LF , Crokaert J , Perroud N , Piguet C . Structural and functional MRI correlates of inflammation in bipolar disorder: A systematic review . J Affect Disord . 2023 ; 325 : 83 – 92 . OpenUrl PubMed 53. ↵ Vazquez-Gonzalez D , Carreon-Trujillo S , Alvarez-Arellano L , Abarca-Merlin DM , Dominguez-Lopez P , Salazar-Garcia M , et al. A Potential Role for Neuroinflammation in ADHD . Adv Exp Med Biol . 2023 ; 1411 : 327 – 56 . OpenUrl PubMed 54. ↵ Tansey MG , Wallings RL , Houser MC , Herrick MK , Keating CE , Joers V . Inflammation and immune dysfunction in Parkinson disease . Nat Rev Immunol . 2022 ; 22 ( 11 ): 657 – 73 . OpenUrl CrossRef PubMed 55. ↵ Zenebe Y , Akele B , M WS , Necho M . Prevalence and determinants of depression among old age: a systematic review and meta-analysis . Ann Gen Psychiatry . 2021 ; 20 ( 1 ): 55 . OpenUrl CrossRef PubMed 56. ↵ Schneider B , Prvulovic D , Oertel-Knochel V , Knochel C , Reinke B , Grexa M , et al. Biomarkers for major depression and its delineation from neurodegenerative disorders . Prog Neurobiol . 2011 ; 95 ( 4 ): 703 – 17 . OpenUrl CrossRef PubMed 57. ↵ Weiskopf D , Weinberger B , Grubeck-Loebenstein B . The aging of the immune system . Transpl Int . 2009 ; 22 ( 11 ): 1041 – 50 . OpenUrl CrossRef PubMed Web of Science 58. ↵ Horvath S . DNA methylation age of human tissues and cell types . Genome Biol . 2013 ; 14 ( 10 ): R115 . OpenUrl CrossRef PubMed 59. Horvath S , Raj K . DNA methylation-based biomarkers and the epigenetic clock theory of ageing . Nat Rev Genet . 2018 ; 19 ( 6 ): 371 – 84 . OpenUrl CrossRef PubMed 60. ↵ Shireby GL , Davies JP , Francis PT , Burrage J , Walker EM , Neilson GWA , et al. Recalibrating the epigenetic clock: implications for assessing biological age in the human cortex . Brain . 2020 ; 143 ( 12 ): 3763 – 75 . OpenUrl CrossRef PubMed 61. ↵ Graves AJ , Danoff JS , Kim M , Brindley SR , Skyberg AM , Giamberardino SN , et al. Accelerated epigenetic age is associated with whole-brain functional connectivity and impaired cognitive performance in older adults . Sci Rep . 2024 ; 14 ( 1 ): 9646 . OpenUrl CrossRef PubMed 62. Grodstein F , Lemos B , Yu L , Klein HU , Iatrou A , Buchman AS , et al. The association of epigenetic clocks in brain tissue with brain pathologies and common aging phenotypes . Neurobiol Dis . 2021 ; 157 : 105428 . OpenUrl CrossRef PubMed 63. ↵ Levine ME , Lu AT , Bennett DA , Horvath S . Epigenetic age of the pre-frontal cortex is associated with neuritic plaques, amyloid load, and Alzheimer’s disease related cognitive functioning . Aging (Albany NY ). 2015 ; 7 ( 12 ): 1198 – 211 . OpenUrl PubMed 64. Harvanek ZM , Boks MP , Vinkers CH , Higgins-Chen AT . The Cutting Edge of Epigenetic Clocks: In Search of Mechanisms Linking Aging and Mental Health . Biol Psychiatry . 2023 ; 94 ( 9 ): 694 – 705 . OpenUrl CrossRef PubMed 65. Oblak L , van der Zaag J , Higgins-Chen AT , Levine ME , Boks MP . A systematic review of biological, social and environmental factors associated with epigenetic clock acceleration . Ageing Res Rev . 2021 ; 69 : 101348 . OpenUrl CrossRef PubMed 66. Zhang ZZ , Moeckel C , Mustafa M , Pham H , Olson AE , Mehta D , et al. The association of epigenetic age acceleration and depressive and anxiety symptom severity among children recently exposed to substantiated maltreatment . J Psychiatr Res . 2023 ; 165 : 7 – 13 . OpenUrl PubMed 67. Han LKM , Aghajani M , Clark SL , Chan RF , Hattab MW , Shabalin AA , et al. Epigenetic Aging in Major Depressive Disorder . Am J Psychiatry . 2018 ; 175 ( 8 ): 774 – 82 . OpenUrl CrossRef PubMed 68. Jeremian R , Malinowski A , Chaudhary Z , Srivastava A , Qian J , Zai C , et al. Epigenetic age dysregulation in individuals with bipolar disorder and schizophrenia . Psychiatry Res . 2022 ; 315 : 114689 . OpenUrl CrossRef PubMed 69. Bourdon C , Etain B , Spano L , Belzeaux R , Leboyer M , Delahaye-Duriez A , et al. Accelerated aging in bipolar disorders: An exploratory study of six epigenetic clocks . Psychiatry Res . 2023 ; 327 : 115373 . OpenUrl PubMed 70. Salontaji K , Haftorn KL , Sanders F , Page CM , Walton E , Felix JF , et al. Gestational epigenetic age and ADHD symptoms in childhood: a prospective, multi-cohort study . Mol Psychiatry . 2024 . 71. ↵ Postberg J , Schubert MT , Nin V , Wagner L , Piefke M . A perspective on epigenomic aging processes in the human brain and their plasticity in patients with mental disorders - a systematic review . Neurogenetics . 2024 . 72. ↵ Jonkman TH , Dekkers KF , Slieker RC , Grant CD , Ikram MA , van Greevenbroek MMJ , et al. Functional genomics analysis identifies T and NK cell activation as a driver of epigenetic clock progression . Genome Biol . 2022 ; 23 ( 1 ): 24 . OpenUrl CrossRef PubMed 73. ↵ Murthy M , Shireby G , Miki Y , Vire E , Lashley T , Warner TT , et al. Epigenetic age acceleration is associated with oligodendrocyte proportions in MSA and control brain tissue . Neuropathol Appl Neurobiol . 2023 ; 49 ( 1 ): e12872 . OpenUrl CrossRef PubMed 74. ↵ Gamo NJ , Arnsten AF . Molecular modulation of prefrontal cortex: rational development of treatments for psychiatric disorders . Behav Neurosci . 2011 ; 125 ( 3 ): 282 – 96 . OpenUrl CrossRef PubMed 75. ↵ Zhang Z , Wiencke JK , Kelsey KT , Koestler DC , Molinaro AM , Pike SC , et al. Hierarchical deconvolution for extensive cell type resolution in the human brain using DNA methylation . Front Neurosci . 2023 ; 17 : 1198243 . OpenUrl CrossRef PubMed 76. ↵ Salas LA , Zhang Z , Koestler DC , Butler RA , Hansen HM , Molinaro AM , et al. Enhanced cell deconvolution of peripheral blood using DNA methylation for high-resolution immune profiling . Nat Commun . 2022 ; 13 ( 1 ): 761 . OpenUrl CrossRef PubMed 77. ↵ Houseman EA , Accomando WP , Koestler DC , Christensen BC , Marsit CJ , Nelson HH , et al. DNA methylation arrays as surrogate measures of cell mixture distribution . BMC Bioinformatics . 2012 ; 13 : 86 . OpenUrl CrossRef PubMed 78. ↵ Willer CJ , Li Y , Abecasis GR . METAL: fast and efficient meta-analysis of genomewide association scans . Bioinformatics . 2010 ; 26 ( 17 ): 2190 – 1 . OpenUrl CrossRef PubMed Web of Science 79. ↵ Navarro-Delgado E . RAMEN: Regioanl Association of Methylome variability with the Exposome and geNome . https://ericknavarrodgithubio/RAMEN/ 2024 ;R package version 0.19000. 80. ↵ Wright MN ZA . ranger: A Fast Implementation of Random Forests for High Dimensional Data in C++ and R . Journal of Statistical Software . 2017 ; 77 ( 1 ): 1 – 17 . OpenUrl 81. ↵ Freytag V , Carrillo-Roa T , Milnik A , Samann PG , Vukojevic V , Coynel D , et al. A peripheral epigenetic signature of immune system genes is linked to neocortical thickness and memory . Nat Commun . 2017 ; 8 : 15193 . OpenUrl PubMed 82. ↵ Sant’Anna Barbosa Ferreira P , van Dongen J , den Braber A , Boomsma DI , de Geus EJC , van ’t Ent D . Epigenetic age acceleration in peripheral blood correlates with brain-MRI age acceleration . Brain . 2025 . 83. ↵ Kozlenkov A , Wang M , Roussos P , Rudchenko S , Barbu M , Bibikova M , et al. Substantial DNA methylation differences between two major neuronal subtypes in human brain . Nucleic Acids Res . 2016 ; 44 ( 6 ): 2593 – 612 . OpenUrl CrossRef PubMed 84. ↵ Tian W , Zhou J , Bartlett A , Zeng Q , Liu H , Castanon RG , et al. Single-cell DNA methylation and 3D genome architecture in the human brain . Science . 2023 ; 382 ( 6667 ): eadf5357 . OpenUrl CrossRef PubMed 85. ↵ Liu H , Zhou J , Tian W , Luo C , Bartlett A , Aldridge A , et al. DNA methylation atlas of the mouse brain at single-cell resolution . Nature . 2021 ; 598 (7879): 120 – 8 . OpenUrl CrossRef PubMed 86. ↵ Felten DL , Felten SY , Carlson SL , Olschowka JA , Livnat S . Noradrenergic and peptidergic innervation of lymphoid tissue . J Immunol . 1985 ; 135 ( 2 Suppl ): 755s – 65s . OpenUrl CrossRef PubMed 87. ↵ Blalock JE . A molecular basis for bidirectional communication between the immune and neuroendocrine systems . Physiol Rev . 1989 ; 69 ( 1 ): 1 – 32 . OpenUrl CrossRef PubMed Web of Science 88. ↵ Kerage D , Sloan EK , Mattarollo SR , McCombe PA . Interaction of neurotransmitters and neurochemicals with lymphocytes . J Neuroimmunol . 2019 ; 332 : 99 – 111 . OpenUrl CrossRef PubMed 89. ↵ van Mierlo HC , Schot A , Boks MPM , de Witte LD . The association between schizophrenia and the immune system: Review of the evidence from unbiased ‘omic-studies’ . Schizophr Res . 2020 ; 217 : 114 – 23 . OpenUrl CrossRef PubMed 90. ↵ Robinson-Agramonte MLA , Noris Garcia E , Fraga Guerra J , Vega Hurtado Y , Antonucci N , Semprun-Hernandez N , et al. Immune Dysregulation in Autism Spectrum Disorder: What Do We Know about It? Int J Mol Sci . 2022 ; 23 ( 6 ). 91. ↵ Bettcher BM , Tansey MG , Dorothee G , Heneka MT . Peripheral and central immune system crosstalk in Alzheimer disease - a research prospectus . Nat Rev Neurol . 2021 ; 17 ( 11 ): 689 – 701 . OpenUrl CrossRef PubMed 92. ↵ Drees F , Pokutta S , Yamada S , Nelson WJ , Weis WI . Alpha-catenin is a molecular switch that binds E-cadherin-beta-catenin and regulates actin-filament assembly . Cell . 2005 ; 123 ( 5 ): 903 – 15 . OpenUrl CrossRef PubMed Web of Science 93. ↵ Ko J , Fuccillo MV , Malenka RC , Sudhof TC . LRRTM2 functions as a neurexin ligand in promoting excitatory synapse formation . Neuron . 2009 ; 64 ( 6 ): 791 – 8 . OpenUrl CrossRef PubMed Web of Science 94. ↵ Conroy LR , Hawkinson TR , Young LEA , Gentry MS , Sun RC . Emerging roles of N-linked glycosylation in brain physiology and disorders . Trends Endocrinol Metab . 2021 ; 32 ( 12 ): 980 – 93 . OpenUrl CrossRef PubMed 95. ↵ Nagashima T , Ichimiya S , Kikuchi T , Saito Y , Matsumiya H , Ara S , et al. Arachidonate 5-lipoxygenase establishes adaptive humoral immunity by controlling primary B cells and their cognate T-cell help . Am J Pathol . 2011 ; 178 ( 1 ): 222 – 32 . OpenUrl CrossRef PubMed 96. ↵ Dong X , Wu W , Ma L , Liu C , Bhuckory MB , Wang L , et al. Collectin-11 Is an Important Modulator of Retinal Pigment Epithelial Cell Phagocytosis and Cytokine Production . J Innate Immun . 2017 ; 9 ( 6 ): 529 – 45 . OpenUrl PubMed 97. ↵ Allard M , Oger R , Benlalam H , Florenceau L , Echasserieau K , Bernardeau K , et al. Soluble HLA-I/peptide monomers mediate antigen-specific CD8 T cell activation through passive peptide exchange with cell-bound HLA-I molecules . J Immunol . 2014 ; 192 ( 11 ): 5090 – 7 . OpenUrl Abstract / FREE Full Text 98. ↵ Varatharaj A , Galea I . The blood-brain barrier in systemic inflammation . Brain Behav Immun . 2017 ; 60 : 1 – 12 . OpenUrl CrossRef PubMed 99. ↵ Matsuoka RL , Buck LD , Vajrala KP , Quick RE , Card OA . Historical and current perspectives on blood endothelial cell heterogeneity in the brain . Cell Mol Life Sci . 2022 ; 79 ( 7 ): 372 . OpenUrl CrossRef PubMed 100. ↵ Wang S , de Fabritus L , Kumar PA , Werner Y , Ma M , Li D , et al. Brain endothelial CXCL12 attracts protective natural killer cells during ischemic stroke . J Neuroinflammation . 2023 ; 20 ( 1 ): 8 . OpenUrl PubMed 101. ↵ Ning Z , Liu Y , Guo D , Lin WJ , Tang Y . Natural killer cells in the central nervous system . Cell Commun Signal . 2023 ; 21 ( 1 ): 341 . OpenUrl PubMed 102. ↵ Li Z , Li M , Shi SX , Yao N , Cheng X , Guo A , et al. Brain transforms natural killer cells that exacerbate brain edema after intracerebral hemorrhage . J Exp Med . 2020 ; 217 ( 12 ). 103. ↵ Long Y , Tao H , Karachi A , Grippin AJ , Jin L , Chang YE , et al. Dysregulation of Glutamate Transport Enhances Treg Function That Promotes VEGF Blockade Resistance in Glioblastoma . Cancer Res . 2020 ; 80 ( 3 ): 499 – 509 . OpenUrl Abstract / FREE Full Text 104. ↵ Xu T , Stewart KM , Wang X , Liu K , Xie M , Ryu JK , et al. Metabolic control of T(H)17 and induced T(reg) cell balance by an epigenetic mechanism . Nature . 2017 ; 548 ( 7666 ): 228 – 33 . OpenUrl CrossRef PubMed 105. ↵ Zhang Z , Reynolds SR , Stolrow HG , Chen JQ , Christensen BC , Salas LA . Deciphering the role of immune cell composition in epigenetic age acceleration: Insights from cell-type deconvolution applied to human blood epigenetic clocks . Aging Cell . 2024 ; 23 ( 3 ): e14071 . OpenUrl CrossRef PubMed 106. ↵ Sun M , Yang H , Hu Y , Fan J , Duan M , Ruan J , et al. Differential white blood cell count and epigenetic clocks: a bidirectional Mendelian randomization study . Clin Epigenetics . 2024 ; 16 ( 1 ): 118 . OpenUrl PubMed 107. ↵ Li X , Li Y , Jin Y , Zhang Y , Wu J , Xu Z , et al. Transcriptional and epigenetic decoding of the microglial aging process . Nat Aging . 2023 ; 3 ( 10 ): 1288 – 311 . OpenUrl PubMed 108. ↵ Blank T , Prinz M . Microglia as modulators of cognition and neuropsychiatric disorders . Glia . 2013 ; 61 ( 1 ): 62 – 70 . OpenUrl CrossRef PubMed Web of Science 109. ↵ Kular L , Klose D , Urdanoz-Casado A , Ewing E , Planell N , Gomez-Cabrero D , et al. Epigenetic clock indicates accelerated aging in glial cells of progressive multiple sclerosis patients . Front Aging Neurosci . 2022 ; 14 : 926468 . OpenUrl CrossRef PubMed 110. ↵ Collado-Torres L , Klei L , Liu C , Kleinman JE , Hyde TM , Geschwind DH , et al. Comparison of gene expression in living and postmortem human brain . medRxiv . 2023 . 111. ↵ Rodriguez de Los Santos M , Kopell BH , Buxbaum Grice A , Ganesh G , Yang A , Amini P , et al. Divergent landscapes of A-to-I editing in postmortem and living human brain . Nat Commun . 2024 ; 15 ( 1 ): 5366 . OpenUrl PubMed 112. ↵ Rhein M , Hagemeier L , Klintschar M , Muschler M , Bleich S , Frieling H . DNA methylation results depend on DNA integrity-role of post mortem interval . Front Genet . 2015 ; 6 : 182 . OpenUrl PubMed 113. ↵ Siletti K , Hodge R , Mossi Albiach A , Lee KW , Ding SL , Hu L , et al. Transcriptomic diversity of cell types across the adult human brain . Science . 2023 ; 382 ( 6667 ): eadd7046 . OpenUrl CrossRef PubMed 114. ↵ Langlieb J , Sachdev NS , Balderrama KS , Nadaf NM , Raj M , Murray E , et al. The cell type composition of the adult mouse brain revealed by single cell and spatial genomics . bioRxiv . 2023 . 115. ↵ Huang Y , Pastor WA , Shen Y , Tahiliani M , Liu DR , Rao A . The behaviour of 5-hydroxymethylcytosine in bisulfite sequencing . PLoS One . 2010 ; 5 ( 1 ): e8888 . OpenUrl CrossRef PubMed 116. ↵ Kinney SM , Chin HG , Vaisvila R , Bitinaite J , Zheng Y , Esteve PO , et al. Tissue-specific distribution and dynamic changes of 5-hydroxymethylcytosine in mammalian genomes . J Biol Chem . 2011 ; 286 ( 28 ): 24685 – 93 . OpenUrl Abstract / FREE Full Text 117. ↵ Tooley KB , Chucair-Elliott AJ , Ocanas SR , Machalinski AH , Pham KD , Hoolehan W , et al. Differential usage of DNA modifications in neurons, astrocytes, and microglia . Epigenetics Chromatin . 2023 ; 16 ( 1 ): 45 . OpenUrl CrossRef PubMed 118. ↵ Lisa J. Schmunk TPC , Daniel L. McCartney , Hira Javaid , Waylon J. Hastings , Vanja Jovicevic , Dragoljub Kojadinović , Natacha Tomkinson , Eliska Zlamalova , Kirsty C. McGee , Jack Sullivan , Archie Campbell , View ORCID Profile Andrew M McIntosh , Veronika Óvári , Karl Wishart , Christian E. Behrens , Emma Stone , Miloš Gavrilov , Rob Thompson , Hurdle bio-infrastructure team , Thomas Jackson , Janet M. Lord , Thomas M. Stubbs , Riccardo E. Marioni , Daniel E. Martin-Herranz . A novel framework to build saliva-based DNA methylation biomarkers: quantifying systemic chronic inflammation as a case study . bioRxiv . 2023 . View the discussion thread. Back to top Previous Next Posted August 29, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Leaping over the blood-brain barrier: DNA methylation as a link between peripheral and central immune systems Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Leaping over the blood-brain barrier: DNA methylation as a link between peripheral and central immune systems Mandy Meijer , Maggie Po Yuan Fu , Erick Isaac Navarro-Delgado , Hannah-Ruth Engelbrecht , Gustavo Turecki , Meingold Hiu-ming Chan , Michael Steffen Kobor bioRxiv 2025.08.25.672204; doi: https://doi.org/10.1101/2025.08.25.672204 Share This Article: Copy Citation Tools Leaping over the blood-brain barrier: DNA methylation as a link between peripheral and central immune systems Mandy Meijer , Maggie Po Yuan Fu , Erick Isaac Navarro-Delgado , Hannah-Ruth Engelbrecht , Gustavo Turecki , Meingold Hiu-ming Chan , Michael Steffen Kobor bioRxiv 2025.08.25.672204; doi: https://doi.org/10.1101/2025.08.25.672204 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Neuroscience Subject Areas All Articles Animal Behavior and Cognition (7624) Biochemistry (17650) Bioengineering (13871) Bioinformatics (41882) Biophysics (21424) Cancer Biology (18566) Cell Biology (25461) Clinical Trials (138) Developmental Biology (13365) Ecology (19867) Epidemiology (2067) Evolutionary Biology (24290) Genetics (15590) Genomics (22476) Immunology (17713) Microbiology (40331) Molecular Biology (17148) Neuroscience (88477) Paleontology (666) Pathology (2828) Pharmacology and Toxicology (4816) Physiology (7635) Plant Biology (15114) Scientific Communication and Education (2044) Synthetic Biology (4286) Systems Biology (9815) Zoology (2268)