Mapping the GDF15 Arm of the Integrated Stress Response in Human Cells and Tissues

Mapping the GDF15 Arm of the Integrated Stress Response in Human Cells and Tissues | 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 Mapping the GDF15 Arm of the Integrated Stress Response in Human Cells and Tissues View ORCID Profile Janell L.M. Smith , Kamaryn Tanner , Jack Devine , Anna S. Monzel , Taivan Batjargal , Maxwell Z. Wilson , Alan A. Cohen , Martin Picard doi: https://doi.org/10.1101/2025.01.31.635929 Janell L.M. Smith 1 Department of Psychiatry, Division of Behavioral Medicine, Columbia University Irving Medical Center , New York, NY 10032, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Janell L.M. Smith Kamaryn Tanner 2 Robert N Butler Columbia Aging Center, Mailman School of Public Health , New York, NY 10032, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jack Devine 1 Department of Psychiatry, Division of Behavioral Medicine, Columbia University Irving Medical Center , New York, NY 10032, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site Anna S. Monzel 1 Department of Psychiatry, Division of Behavioral Medicine, Columbia University Irving Medical Center , New York, NY 10032, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site Taivan Batjargal 3 Biomolecular Science and Engineering Program, University of California Santa Barbara , Santa Barbara, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Maxwell Z. Wilson 3 Biomolecular Science and Engineering Program, University of California Santa Barbara , Santa Barbara, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Alan A. Cohen 2 Robert N Butler Columbia Aging Center, Mailman School of Public Health , New York, NY 10032, United States 4 Department of Environmental Health Sciences, Mailman School of Public Health , New York, NY 10032, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site Martin Picard 1 Department of Psychiatry, Division of Behavioral Medicine, Columbia University Irving Medical Center , New York, NY 10032, United States 2 Robert N Butler Columbia Aging Center, Mailman School of Public Health , New York, NY 10032, United States 5 Department of Neurology, H. Houston Merritt Center for Neurological and Mitochondrial Disorders, Columbia University Irving Medical Center , New York, New York 10032, United States 6 New York State Psychiatric Institute , New York, NY 10032, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: martin.picard{at}columbia.edu Abstract Full Text Info/History Metrics Data/Code Preview PDF Abstract Mitochondrial stress activates the integrated stress response (ISR) and triggers cell–cell communication through the secretion of the metabokine growth differentiation factor 15 (GDF15). However, the gene network underlying the ISR remains poorly defined, particularly across metabolically diverse cellular states and tissues. Using RNAseq data from fibroblasts subjected to eleven metabolic perturbations, including genetic and pharmacological mitochondrial OxPhos defects, we showed that the ISR has multiple arms and developed an ISR GDF15 index quantifying the GDF15 arm of ISR activation in human cells. The ISR GDF15 index was validated using optogenetic activation of the ISR protein kinase R (PKR) in a stable cell line, demonstrating its rapid kinetics preceding to GDF15 gene expression. We then deployed the ISR GDF15 index across 44 postmortem human tissues, reporting that the ISR GDF15 was upregulated in the heart of individuals who died of an acute cause in the emergency room, whereas it was preferentially upregulated in the brain of individuals who died as inpatients after protracted hospital stays. ISR GDF15 was also moderately, positively correlated with age across all tissues. These data highlight multiple distinct ISR pathways and clarify which genes are related to the GDF15 arm of the ISR, yielding an ISR GDF15 index that can be used to investigate tissue-specific and age-related ISR activation in both in vitro cultures and human tissues. Ultimately, we expand our knowledge of the GDF15 arm of the ISR in humans, show its inducibility by the ISR kinase PKR, and demonstrate its applicability to detect ISR activation in specific human tissues. Main Text Cells sense reductive stress via the integrated stress response (ISR) 1 . Among many targets, the ISR activates the gene GDF15 , which encodes a secreted cytokine/metabokine known to be elevated with aging 2 and in several chronic diseases 3 , 4 . This elevation is particularly strong in primary mitochondrial diseases caused by defects in the oxidative phosphorylation (OxPhos) system 5 , 6 . However, the ISR is pleiotropic and triggers multiple cellular processes 7 . As a result, there is no consensus on the gene network that constitutes the ISR in human cells, or whether the same set of genes operate in metabolically diverse organs or tissues. Here, we seek to understand the dimensionality of the ISR, with a particular focus on the pathway or “arm” leading to GDF15 signaling, given its status as an ISR biomarker 8 and its role in mediating the ISR 9 . To identify ISR-related genes linked to inter-cellular GDF15 signaling, we queried multiple ISR gene lists (119 genes in total) 10 – 12 ( Figure 1A ), and examined their co-expression in a cellular-lifespan-system dataset generated in fibroblasts 13 . Sensitivity analyses were performed on each gene list separately, as well as a random selection of 119 selected genes out of all mapped genes in the RNAseq dataset. This process demonstrated the superiority of including all 119 genes from the combined lists, plus GDF15 , in the fibroblast dataset, over using the individual ISR gene lists, equally loaded genes of the 119 combined genes, or 119 randomly chosen genes ( Extended Data Figure 1 ). Download figure Open in new tab Extended Data Figure 1. Sensitivity analysis for various ISR gene lists ( A ) Factor analysis performed on individual gene sets (GO Consortium which contained 37 genes, Any Genes which had 84 genes, and Jackson Laboratory which had 23 genes), as well as equal loadings for all the 120 genes (not shown), and the average of 20 different sets of 119 randomly chosen genes plus GDF15 (not shown). The factor that was most correlated with GDF15 was chosen for further sensitivity analyses. ( B ) The index scores of the chosen factors from individual gene sets depicted in A . ( C ) Comparison of the index scores of only control, SURF1 mutant, or Oligo treated cells for each gene set, including equal loadings of the 120 ISR genes; Kruskal-Wallis test performed with Dunn’s test for multiple testing. ( D ) The effect sizes of each treatments comparisons (Hedge’s g) from panel C . GO = GO Consortium, AG = Any Genes, JL = Jackson Laboratory Download figure Open in new tab Figure 1. Development of an ISR index using primary human fibroblasts ( A ) ISR related genes from three reference ISR pathway lists revealing minimal overlap. We use all 119 genes plus GDF15 (n=120) to compile an unbiased ISR gene list. ( B ) Bulk RNA seq data in primary human fibroblasts from three healthy donors and three individuals with mitochondrial disease (SURF1 gene mutations) were cultured across their replicative lifespans (up to 260 days) and treated with pharmacological and metabolic perturbations, including the OxPhos inhibitor oligomycin (1nM). ( C ) Overview of the analytical strategy to generate an ISR index using GDF15 as a bait. ( D ) Principal component analysis (PCA) of data from the lifespan study depicted in panel B including 339 samples 120 genes. The red arrow denotes the loading vector of GDF15 ; all other genes shown in Extended Data Figure 2 . ( E ) Factor analysis revealed one factor where GDF15 had the strongest positive loading (Factor 1, left ), yielding weights that reflect the co-regulation of all other genes with GDF15 ( right ). ( F ) Correlation of Factor 1 with standardized GDF15 expression. ( G ) Boxplots of ISR GDF15 scores for each condition, sorted by ascending median score. P-values from Kruskal-Wallis test and Dunn’s test for multiple comparisons. Oligo, Oligomycin; DEX = dexamethasone, mitoNUITs = mitochondrial nutrient uptake inhibitors, 2DG = 2-Deoxy-D-glucose, BHB = β-hydroxybutyrate, contact inhib = contact inhibition. Download figure Open in new tab Extended Data Figure 2. Principal component gene vectors and comparisons with age ( A ) Biplot with gene vector loadings of PC1 and PC2. ( B ) First 10 PC components vs fibroblast time in culture (days). ( C ) Effect sizes (Spearman’s rank correlations) of PCs vs time in culture (days). We used a longitudinal transcriptomic dataset including 339 fibroblast samples collected from either healthy donors or mitochondrial-disease donors with mutations in the OxPhos complex IV assembly-factor gene SURF1 ( Figure 1B ). To create a vast spectrum of bioenergetic conditions, each cell line was exposed to up to 11 metabolic perturbations including an inhibitor of the OxPhos system, oligomycin (Oligo), glucocorticoid agonist dexamethasone (DEX), mitochondrial-nutrient-uptake inhibitors (mitoNUITs), glucose deprivation plus β-hydroxybutyrate (BHB), low oxygen (3%), and contact inhibition 13 . OxPhos defects are potent triggers of the ISR in other datasets 1 and in this dataset 14 , making this model and the variety of conditions tested an ideal testbed to examine gene expression signatures of the GDF15 arm, and other dimensions of the ISR, in human cells ( Figure 1C ). To computationally isolate co-regulated ISR genes, we queried bulk gene expression RNAseq data using GDF15 as a bait; we included GDF15 alongside the 119 ISR genes from the three ISR gene lists and performed dimensionality-reduction analyses (the 120 genes can be found in Extended Data Table 1 ). This allowed us to explore gene expression patterns within our samples, with a focus on patterns most strongly associated with GDF15 . Using principal component analysis (PCA) to explore this gene-expression landscape across all treatments simultaneously ( Figure 1D ), we identified a major axis of genes co-regulated with GDF15 (principal component 2, explains 15.5% of variance). Extended Data Figure 2A shows the loadings for each gene on each of the first two principal components. Next, we used factor analysis to more quantitatively isolate the transcriptional signature of this GDF15 -related ISR axis. This approach converged on an optimal solution containing 12 factors (i.e., sets of loadings or “weights” applied to each of the 119 genes plus GDF15 for a total of 120 genes), which collectively account for a maximal portion (73.1%) of the shared variance within this dataset ( Figure 1E ). The relationships between the principal components and each factor are quantified in Extended Data Figure 5A , with specific correlation plots depicted in Extended Data Figure 5B-C . Factor 1—our index hereafter referred to as the ISR GDF15 score—explained 12.9% of the variance, and was more strongly correlated with GDF15 expression (Spearman’s rank correlation coefficient (ρ) = 0.87, p <1×10 -15 ) than any other factor, establishing construct validity for the calculated ISR GDF15 score ( Figure 1F ). The 11 remaining factors and their gene loadings, which are not explored in this work but available to readers in Extended Data Table 2 , may represent additional pathways indicative of other ISR dimensions. Download figure Open in new tab Download figure Open in new tab Extended Data Figure 4. Factor analysis comparisons with age ( A ) Plot of first two factors from the factor analysis. ( B ) Correlation of each fibroblast sample vs time (in days) of culture. ( C ) Remaining 11 factors scores vs time in culture. Download figure Open in new tab Extended Data Figure 5. Correlations of factor scores and principal components ( A ) Correlation matrix of Spearman’s rank correlations of the factor scores vs the first 12 principal components from the PCA. ( B ) Spearman’s correlation of ISRGDF15 index (Factor 1) vs PC2 score. ( C ) Spearman’s correlations of ISRGDF15 vs remaining principal components up to 12. To understand the biology underlying the ISR GDF15 index, we compared the genes most strongly associated (positively or negatively) with GDF15 . The genes contributing most strongly to the ISR GDF15 index were positively enriched for regulation of endoplasmic reticulum–unfolded protein response [top 10 positive : DDIT3 , PPP1R15A , ERN1 , GADD45A , CBX4 , BBC3 , SLC3A2 , SIAH2 , CEBPB , TRIB3 ]. In contrast, negatively weighted genes for the ISR GDF15 index were enriched for the regulation of translation and translation initiation [top 10 negative : PPP1CA , APAF1 , SLC35A4 , FOS , IMPACT , NAIP , BIRC5 , EIF2B1 , NARS1 , GCN2 ] (see Extended Data Table 3 annotations and known functions of each gene), a well-described consequence of ISR activation in most cell types. These results are also consistent with the notion that cellular processes operate under energy constraints: that is, while the activation of some energy-dependent processes are upregulated, other processes must be downregulated 15 , 16 . As such, cells activating the GDF15-signaling ISR arm may generally downregulate protein synthesis, as previously observed 17 , despite known increases in specific proteins, including GDF15 14 , 18 . Functionally, ISR GDF15 was most significantly induced in oligomycin-treated cells (large effect size, g = 2.13, p < 0.001) and in SURF1-mutant cells ( g = 1.91, p <0.001, Figure 1G ) relative to untreated control cells, as expected. The conditions of no glucose with β-hydroxybutyrate ( g = 1.33, p <0.001) and the glycolysis inhibitor 2-deoxyglucose ( g = 0.97, p <0.05) also resulted in a higher ISR GDF15 , which is consistent with glucose deprivation being a known ISR activator 19 . The higher ISR GDF15 observed in the contact-inhibition condition ( g = 1.23, p <0.001) was surprising at first, as it was expected to be more physiologically similar to what cells experience in vivo (i.e., a more physiological skin cell culture). However, the culture protocol used in this condition called for only one media change per week (matching the passaging of the non-contact-inhibited cells). This protocol led to a higher density of cells throughout the week, which, in turn, may have depleted nutrients at a faster rate. Thus, the ISR was likely activated in the contact-inhibition group as a result of nutrient starvation. To our initial surprise, the low oxygen level (3%, not anoxia) treatment was not an activator of the ISR, as would be logically expected of a hypoxic treatment. However, upon reviewing the literature, we noted that the experimental conditions previously considered hypoxic treatments for these fibroblast cells was, in fact, likely a physiological oxygenation level 20 . In vivo skin has oxygenation levels between 1.1-4.6% 20 , and thus is likely to be a better control than our control group, which is cultured at 21% oxygen. Together, these data demonstrate the existence of a relatively small subset of positively and negatively inducible GDF15 -related ISR genes induced by mitochondrial OxPhos perturbations. To validate the ISR GDF15 index, we then tested how targeted and time-resolved ISR activation influenced GDF15 itself, and the ISR GDF15 index. To achieve this, we used data from a light-activatable protein kinase R (PKR) kinase in the H4 cancer line ( Figure 2A ) 21 . This optogenetic model triggers robust activation of the ISR (marked by p-eIF2a, ATF4, and CHOP induction) without other confounds including the simultaneous activation of other pathways, such as damage response pathways, for example. Download figure Open in new tab Figure 2. Validation of the ISR GDF15 index with time-resolved ISR activation ( A ) Optogenetic ISR activation strategy targeting an engineered light-sensitive PKR, an ISR kinase, which phosphorylates eIF2alpha to activate the ISR, as per Batjargal et al. ( B ) Experimental and data processing scheme to compute the ISR GDF15 scores across 8 timepoints spanning 12 hours of continuous PKR activation in H4 cells. ( C ) ISR GDF15 score and GDF15 RNA transcript levels for the H4 cells under increasing hours of ISR stimulation. The light sensitive–PKR H4 cells were stimulated with light for up to 12 hours, and cells harvested at eight timepoints for bulk RNAseq ( Figure 2B ). We deployed the ISR GDF15 index to the RNAseq data, quantifying its expression at each timepoint (See Figure 2C ). The ISR GDF15 score increased by 1 hour and peaked at 6-8 hours, followed by a plateau over the remaining hours of light stimulation. The trajectories of induction for other factor scores are shown in Extended Data Figure 6A . Download figure Open in new tab Extended Data Figure 6. All factors applied to opto-PKR H4 cells ( A ) Each factor score for the H4 cells, containing the ISR sensor kinase PKR engineered to be activate in response to light (opto-PKR), stimulated with 0–12 hours of light. Comparatively, PKR activation did not induce GDF15 expression over the first 4 hours. However, between hours 4 and 6 we observed a ∼5-fold increase in GDF15 mRNA at 6 hours ( Figure 2C ). Together, these findings support the accuracy and sensitivity of the ISR GDF15 index to PKR-mediated ISR activation, even in the absence of changes in GDF15 expression from 0 to 4 hours. This suggests that the ISR GDF15 index captures ISR biology more broadly and sensitively than GDF15 alone. To examine the extent to which the ISR GDF15 index it is conserved in human tissues, we used the Genome-Tissue Expression (GTEx) project, which includes RNAseq data on up to 17 different organs from ∼1,000 donors 22 ( Figure 3A left ). We first asked whether the ISR GDF15 score was correlated with its bait GDF15 gene, when applied to individual human organs and tissues. If the internal correlation structure and loadings of the genes composing the ISR GDF15 index were similar across datasets, we expected positive correlations between the index and the gene GDF15 . Focusing on tissues with a sample size >20, we observed significant positive correlations between the ISR GDF15 and GDF15 in 43 out of 44 (97.7%) human tissues (ρ = 0.41-0.82, all p ’s <0.05 corrected for multiple testing, Figure 3B ). These findings suggest that the genetic architecture of the GDF15 -related arm of the ISR is at least partially conserved between fibroblasts in vitro and human tissues in vivo. Download figure Open in new tab Figure 3. ISR GDF15 index applied to human GTEx dataset correlated with age at death ( A ) Schematic of human GTEx dataset containing a variety of human organ samples, and applying the fibroblast generated index to the human transcriptomic data to get the human ISR GDF15 scores for each tissue. ( B ) ISR GDF15 correlations with GDF15 across all the tissue samples expressing all 120 genes in the ISR GDF15 index (Spearman’s ρ, +/- 95% confidence intervals, Bonferroni corrected for multiple comparisons). ( C ) Spearman’s rank correlations of each tissue’s ISR GDF15 score compared to donor age (red points denoting a Bonferroni corrected p<0.05). ( D ) Average Spearman’s correlations (+/- SEM) across all 44 tissues comparing the donor age with the chosen gene(s) or ISR GDF15 score or factor scores (left) and fibroblasts’ age correlated with the chosen genes or ISR GDF15 score or factor score (+/- 95% confidence interval) (right). ISR GDF15 average tissue correlation denoted with a red triangle. Wilcoxon signed-rank test used compared against 0. Given the well-known upregulation of GDF15 with age 2 , but not necessarily of the ISR per se, we examined the association of ISR GDF15 score with age in GTEx (range 20-70, average 53 years). The correlations were positive as expected in all human tissues, and 33 of 44 tissues were significant (average ρ = 0.23, p <0.001, Figure 3D ). Seeking a generalizable test to examine how different gene expression-based indices perform across fibroblasts and human tissues, we average expression across all tissues and individuals of different health status included in the dataset. GDF15 alone also was correlated with age, and in some tissues more strongly so than the ISR GDF15 score ( Figure 3D ). To pressure test this analysis further, we confirmed that senescence-associated genes (averaged tissue expressions of CDKN1A , CDKN2A , CCND2 ) were upregulated, whereas proliferative or replication-related genes (averaged tissue expressions of TOP2A , RRM2 , MKI67 ) were downregulated with advancing age, validating the use of cellular lifespan system 13 to probe time- or age-related changes. Other single ISR genes ATF4 and ATF5 were unrelated to age, whereas CHOP was mildly related to age (ρ = 0.13, p <0.001) with about half the effect size of GDF15 . In sum, ISR GDF15 was positively associated with age (ρ = 0.19, p <0.001), indicating that the GDF15 -related arm of the ISR is upregulated with age, and more strongly than individual genes classically referenced alone (or in conjunction with others) to denote activation of the ISR ( Figure 3D ). We then systematically examined the other 11 potential arms of the ISR isolated by factor analysis, ranked from negative to positive age-related effect sizes across all GTEx tissues ( Figure 3D ). Four other ISR arms were significantly negatively associated with age (Factors 12, 6, 2, 4), and one was positively associated with age (Factor 11). Applying the same analysis to aging human fibroblasts similarly showed both GDF15 alone (ρ = 0.51, p <0.001) and the ISR GDF15 score (ρ = 0.61, p <0.001) were positively associated with age, though these effect sizes were smaller than the senescence associated markers (ρ = 0.77, p <0.001), as expected. A direct comparison of effect sizes across each gene, the ISR GDF15 score , and the 11 factors revealed a moderate correlation between the in vivo and in vitro datasets (ρ = 0.69, p <0.01, Extended Data Figure 7A left ). The correlation was stronger without the 11 non- ISR GDF15 factors (ρ = 0.93, p <0.01, Extended Data Figure 7A right ). Given the coherence between the two systems (in vivo and in vitro), this finding supports the notion that the ISR GDF15 signaling arm has an age-related association similar to that of GDF15 in both strength and direction. In other words, the ISR GDF15 index accounts for most of the age-related signal of the ISR. The other factors may be important processes, but their age associations are considerably weaker than the identified ISR GDF15 arm. Download figure Open in new tab Extended Data Figure 7. Correlations between human and fibroblast age associations with genes and factors ( A ) Spearman’s correlations of the average Spearman’s ρs of all tissues comparing donor age to chosen genes, ISRGDF15 score, or other 11 factor scores (left). The panel on the right shows the same plot without the 11 non- ISRGDF15 factors. In the GTEx dataset, some donors died suddenly in the intensive care unit (76.4% of a cardiac arrest or related cause), while other individuals were hospital inpatients that endured more protracted death processes (involving coma, or other conditions). To test whether ISR activation was associated with organ-specific cause of death, we calculated the ISR GDF15 score for each tissue for each place of death: emergency room (ER) and hospital inpatient (HI). We focused on these groups for two reasons: i) each group included a sufficient number of samples to allow robust comparisons of ISR activation; and ii) acute ER deaths are more likely to have been associated with acute cardiocirculatory death within hours, while HI deaths are more likely to have endured metabolic disturbances and an protracted agonal period that tends to affect brain function (e.g., coma). Because the ISR is activated in a time-dependent manner 7 , we hypothesized that longer hospital stays and pathology in HI would produce higher ISR GDF15 activation compared to ER donors, who on average had less time to activate the ISR before death. The expected ISR activation pattern was only observed in a specific subset of tissues: brain, salivary gland, skeletal muscle, skin, tibial nerve and artery, and subcutaneous adipose tissue (no digestive tissues). We also found that donors from the ER had significantly higher ISR GDF15 scores in a different subset of tissues: whole blood, mammary breast, prostate, thyroid, esophageal, lung, transverse colon, stomach and the atrial appendage, left ventricle, aorta, and coronary artery of the heart; not in CNS tissues ( Figure 4C ). Download figure Open in new tab Figure 4. ISR GDF15 index applied to human GTEx dataset ( A ) Select tissue correlations of ISR GDF15 and GDF15 expression, Bonferroni corrected. ( B ) ISR GDF15 or GDF15 expression for each place of death, in brain frontal cortex or heart left ventricle; significance determined via Wilcoxon rank-sum test. ( C ) Hedge’s g effect sizes comparing each tissue’s ISR GDF15 activation score for each place of death, or GDF15 expression comparing each place of death—Hedge’s g values of hospital inpatient scores being greater than ER seen on the right, ER greater than hospital inpatient towards the left. Circles denote ISR GDF15 score, squares GDF15 expression, red as significant, gray non-significant, with BH correction for multiple comparisons. The grey brain and heart indicate the ISR GDF15 score and GDF15 expression of the brain frontal cortex and the heart left ventricle comparisons from B . See methods for specific tissues within each tissue grouping. Wilcoxon rank-sum test was used to compare either the ISR GDF15 score or GDF15 expression in ER or HI, with BH correction. ( D ) Proliferation scores of each tissue split into two groups based on whether the tissue had a higher ISR score in ER or HI from Figure 2C . ( E ) Spearman’s correlation of ISR GDF15 scores with proliferation scores. ER = Emergency Room, HI = hospital inpatient. To investigate this unexpected pattern, we drew upon previous research by Mick et al., which demonstrated that the same stressor triggered differential ISR activation between myoblasts (actively proliferating) compared to myotubes (non-proliferative) 1 . To explore whether proliferation status could contribute to the differential ISR activation across human tissues, we calculated the proliferation score of each tissue 23 (see methods for details) and split the tissues into two groups: proliferative vs non-proliferative, using the median proliferation score. We then compared the ISR GDF15 scores between the two groups based on their proliferation status. Proliferative tissues exhibited significantly higher ISR activation scores than the non-proliferative tissues ( Extended Data Figure 8A ). This was validated by the significant correlation between ISR activation scores and proliferation scores across tissues (ρ = 0.78, p <0.001, Figure 4D ). Finally, we categorized the tissues according to whether the donors had significantly higher ISR GDF15 scores based on deathplace (either HI to ER or ER to HI), and found that the two groups had different proliferative indexes ( g = 1.17, p <0.01, Figure 4E ). The tissues with higher ISR GDF15 in the ER group were more proliferative than those that were higher in the HI group. Ultimately, these analyses support the notion that proliferation status contributes to differential ISR activation in a tissue-dependent manner, and help explain the distribution of ISR activation observed across tissues according to place of death seen in Figure 4C . Download figure Open in new tab Extended Data Figure 8. Human tissues’ ISRGDF15 scores split based on proliferation status ( A ) ISRGDF15 score of each tissue split into two groups based on their proliferation score, using the median proliferation score to split the groups. Only the 31 tissues that were significant in Figure 4C are included, p-value determined by Mann-Whitney test. Interestingly, in these post-mortem tissues, ISR GDF15 was more sensitive than GDF15 expression to sudden death vs chronic illness (or vice versa), as reflected by 81.6% of tissues showing significantly different ISR GDF15 scores under these two scenarios compared to only 52.6% of tissues showing a significant difference in GDF15 expression under the same scenarios (Chi square = 5.96, p = 0.015, Figure 4C ). This was exemplified in the frontal cortex of the brain, which displayed a large effect size ( g = 0.91) for ISR GDF15 when comparing sudden death to chronic illness, but only a moderate effect size for GDF15 ( g = 0.57, Figure 4B ). These results are meaningful, particularly in the context of a robust correlation between ISR GDF15 and GDF15 expression in this tissue (ρ = 0.74, Figure 4A ). This supports the use of the ISR GDF15 score rather than GDF15 alone as an indicator of ISR activation. Discussion Together, this work challenges the monolithic understanding of the ISR in humans, showing instead that the ISR—as defined by popular gene lists 10 – 12 —contains distinct sub-responses or sub-pathways. In particular, our results highlight potential tissue-specificity in ISR activation with disease and conditions leading up to death (see Figure 4C-E ). These findings call for the consideration of tissue-specific regulatory mechanisms or downstream effects that have yet to be defined. We also identify a conserved set of GDF15 -related canonical ISR genes upregulated by experimental mitochondrial perturbations, as well as other groups of genes (factors 2-12) that reflect the multidimensional nature of the ISR. The identified ISR GDF15 index and GDF15 expression were co-regulated across most human tissue, and some aspects of the ISR genetic architecture ( ISR GDF15 in particular) were found to be conserved across three distinct datasets, increasing our confidence in the robustness of these findings. Finally, this work illustrates that ISR GDF15 is activated in a time-dependent manner that precedes GDF15 expression, making it an indicator of early (pre-GDF15 induction) ISR detection. The limitations of this study include i) the use of linear statistical methods, while genes likely interact in non-linear ways 24 , ii) analyses from post-mortem human tissues, where the post-mortem period could alter gene expression profiles, at least in the brain 25 , iii) the optogenetic experiments were only performed with a single ISR kinase, which, while sufficient to robustly induce the ISR 21 may not paint the full picture of ISR biology. We also note that the genes from the literature-based consensus lists could be missing other important genes. Further work to demonstrate the activation or inactivation of the index could be performed with the use of ISR inhibitors such as ISRIB, which would provide a negative control that was not available in these experiments. This would be especially useful for opto-PKR, given the relatively high baseline expression of GDF15 observed here, which is consistent with high GDF15 expression in several cancer cell lines (see Human Protein Atlas 26 ). Overall, we present a general approach to develop multi-gene indices, and the ISR GDF15 index in particular provide a foundation for future research aiming to dissect the multifaceted biology of the human ISR in relation to health, disease, and aging. Methods ISR Gene List Compilation Gene lists for the ISR were gathered by inputting the phrase “integrated stress response” into the search bars of the main websites for three sources: GO Consortium, The Jackson Laboratory, and Any Genes 10 – 12 . The genes from each lists were merged to make one combined ISR gene list. Some gene names were altered to fit the nomenclature of the specific transcriptomic dataset being used, as different gene aliases were used in each dataset for some genes (i.e., NARS1/NARS, WARS1/WARS, Igtp/IRGM). The genes found in each gene list are reported in Extended Data Table 1 . Information about the top 10 positively and negatively associated genes in the ISR GDF15 index was found by searching each gene in URL: https://www.genecards.org/ . Fibroblast Lifespan Study The lifespan study is explained in full detail in Sturm et al 13 . RNAseq data was downloaded from the shiny app URL: https://columbia-picard.shinyapps.io/shinyapp-Lifespan_Study/ . Fibroblast transcriptomic data was previously normalized via DEseq2, v1.30.1, and was log2 transformed. A total of 339 samples (excluding control line technical replicate data) was used for further analysis. Five healthy control lines and three patient-derived SURF1 mutant lines were cultured, with or without various metabolic treatments, as described in Sturm et al. 13 , 14 . PCA R version 4.4.1 with R studio was used to perform PCA (and all subsequent statistical analyses). The ‘prcomp’ function was used to generate PCA, ( stats package v4.4.1) using 339 fibroblast samples and the master list of ISR genes plus GDF15 . Principal component biplots of components beyond PC1 vs PC2 are plotted in Extended Data Figure 3 , as well as their correlations against fibroblast days in culture in Extended Data Figure 2B-C . Factor Analysis The same fibroblast dataset of 339 samples and 120 genes was used in the factor analysis. The ‘factanal’ function ( stats package) was used, with a varimax rotation. The scree method 27 was used to determine the number of factors, as no predetermined number was hypothesized given that the number of causes and outputs of the ISR have yet to be determined. Of the 119 ISR genes, the ISR GDF15 index included 77 positive and 42 negative gene loadings, reflecting ISR genes positively and negatively related to GDF15 (see Extended Data Table 4 for individual gene loadings). Individual genes, such as GDF15 , were centered and scaled by using the scale() function in R before comparing against the ISR GDF15 score or age. A biplot of ISR GDF15 (Factor 1) vs Factor 2, as well as all factor scores compared to age are depicted in Extended Data Figure 4 . Opto-PKR ISR Score Raw FPKM values were normalized to TPM from Batjargal et al. and was log2 transformed. Then, the values were centered and scaled to the initial fibroblast dataset, and finally the ISR GDF15 index was applied, as depicted in the bottom half of Figure 2B . GTEx RNA sequencing data was downloaded as gene expression read counts from GTEx Analysis V8. RNA integrity number (RIN) values of less than 6 were filtered out of the data. Trimmed Mean of M-values (TMM) values were calculated using DGE method (BiocGenerics package, version 0.50.0). To convert the human TMM values to the ISR GDF15 score, human data were scaled to the initial fibroblast dataset, as depicted in Figure 2B (by subtracting the human TMM values by fibroblast means and dividing by the fibroblast standard deviation), and applied to the correlation matrix of the initial fibroblast data to the human data. Tissue samples of less than 20 were excluded from the analysis, leaving the kidney cortex with the minimum number of tissue samples of 86. Four tissues were excluded from the application of the ISR GDF15 index due to missing one or more genes from the ISR list: bladder, brain cerebellar hemisphere, brain cerebellum, and adrenal gland. A total of 44 tissues were included in the analyses. Three genes required changing aliases to match the fibroblast genes: DELE1 was changed to KIAA0141 , WARS1 was changed to WARS , and NARS1 was changed to NARS . When comparing either the ISR GDF15 score of ER vs HI or GDF15 of ER vs HI, tissues with less than 20 samples per group (ER or HI), were filtered out. Six tissues were filtered out: kidney cortex, pancreas, small intestines of the ileum, spleen, uterus, and vagina for having less than 20 samples in the ER group. For distinguishing cardiac-related deaths within the HI or ER groups, we included the following terms as cardiac-related under the cause of death columns either DTHCOD or DTHFUCOD: “cardiac”, “acute mi”, “myocardial”, “myocardiac”, “arrest”, “cardio”, “bradycardia”, “heart”, “aortic dissection”, “probable mi”, and “chf”. Causes of death that did not include any of those terms were deemed non-cardiac. The HI group was composed of 25.2% cardiac-related deaths. Various tissues in the GTEx dataset were noted in the pathology notes column to contain other tissues (often fat) these samples were not filtered out. Tissue Groups As described in Devine et al. 23 , tissues were split into the following groups: liver into Anabolic; all brain tissues into central nervous system (CNS); skeletal muscle, heart atrial appendage and heart left ventricle into Contractile; colon tissues, esophageal tissues, and stomach into Digestive; testis, ovary, and prostate into Reproductive; salivary gland, adipose subcutaneous and visceral omentum, thyroid, and pituitary into Secretory; skin lower leg and suprapubic, artery tibial, breast mammary, whole blood, artery coronary, lung, and artery aorta into Other. Senescence and Proliferation Index Scores Genes for senescence ( CDKN1A , CDKN2A , CCND2 ) 28 or proliferation ( TOP2 , RRM2 , MKI67 ) 23 were selected to create either a tissue senescence score or proliferation score, by averaging the three senescent or proliferative genes respectively. Selection criteria for the proliferation index genes are explained in the methods of Devine et al. 23 . When splitting the tissues into two groups, proliferative vs non-proliferative, the median score of 0.63 was used. Statistics All correlations performed were Spearman’s rank correlations. When comparing group differences of two groups, Wilcoxon rank-sum tests were used. When noted, Bonferroni correction for multiple comparisons was used, with a p-value threshold of 0.05. Otherwise, we used the Benjamini-Hochberg (BH) method, a less stringent form of multiple testing corrections, for more exploratory questions. Effect sizes were determined via Hedge’s g, an unbiased analog of Cohen’s d 29 . Hedge’s g was calculated via the R function ‘cohen.d’ from the package effsize , with the Hedge’s g correction applied. When comparing multiple independent treatment groups simultaneously, Prism version 10.2.3 (347) was used to perform a Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. Wilcoxon signed-rank test was performed using Prism, when testing if a group significantly differed from 0. Author contributions J.L.M.S and M.P. conceived the project. J.L.M.S. performed the analyses with K.T. J.L.M.S. generated the figures. J.L.M.S., M.P., A.A.C., K.T., A.S.M., and J.D. interpreted the data. T.B. and M.Z.W provided data for the optogenetic ISR activation. J.L.M.S., and M.P. drafted the manuscript with A.A.C. All authors reviewed and edited the final manuscript. Funding This work was supported by NIH grants R35GM119793, R01AG066828, R01AG086764, the Wharton Fund, and Baszucki Group to M.P. Data and code availability GTEx v8 RNAseq data is available for download at https://gtexportal.org/home/downloads/adult-gtex/bulk_tissue_expression.Cellular . The Lifespan Study fibroblast dataset can be downloaded at https://columbia-picard.shinyapps.io/shinyapp-Lifespan_Study/ . R code for the analyses will be made available at www.github.com/mitopsychobio . Competing Interests The authors declare no competing interests in regard to this work. Extended Data Material This article contains 8 Extended Data Figures and 4 Extended Data Tables. View this table: View inline View popup Acknowledgements Some figure elements were generated with Biorender.com. We are grateful to the GTEx consortium and to Gabriel Sturm for creating the valuable Cellular Lifespan Study data resources that made this work possible. Footnotes Added in an external dataset of opto-PKR validation of the ISRGDF15 index (now new Figure 2). Added in more extended data figures. Manuscript revised to include new data. https://gtexportal.org/home/downloads/adult-gtex/bulk_tissue_expression.Cellular References 1. ↵ Mick , E. et al. Distinct mitochondrial defects trigger the integrated stress response depending on the metabolic state of the cell . Elife 9 ( 2020 ). doi: 10.7554/eLife.49178 OpenUrl CrossRef PubMed 2. ↵ Conte , M. et al. GDF15, an emerging key player in human aging . Ageing Res Rev 75 , 101569 ( 2022 ). doi: 10.1016/j.arr.2022.101569 OpenUrl CrossRef PubMed 3. ↵ Deng , Y. T. et al. Atlas of the plasma proteome in health and disease in 53,026 adults . Cell 188 , 253 – 271 e257 ( 2025 ). doi: 10.1016/j.cell.2024.10.045 OpenUrl CrossRef 4. ↵ You , J. et al. Plasma proteomic profiles predict individual future health risk . Nat Commun 14 , 7817 ( 2023 ). doi: 10.1038/s41467-023-43575-7 OpenUrl CrossRef PubMed 5. ↵ Sharma , R. et al. Circulating markers of NADH-reductive stress correlate with mitochondrial disease severity . J Clin Invest 131 ( 2021 ). doi: 10.1172/JCI136055 OpenUrl CrossRef 6. ↵ Hubens , W. H. G. et al. Blood biomarkers for assessment of mitochondrial dysfunction: An expert review . Mitochondrion 62 , 187 – 204 ( 2022 ). doi: 10.1016/j.mito.2021.10.008 OpenUrl CrossRef PubMed 7. ↵ Labbe , K. et al. Specific activation of the integrated stress response uncovers regulation of central carbon metabolism and lipid droplet biogenesis . Nat Commun 15 , 8301 ( 2024 ). doi: 10.1038/s41467-024-52538-5 OpenUrl CrossRef PubMed 8. ↵ Asundi , J. et al. GDF15 is a dynamic biomarker of the integrated stress response in the central nervous system . CNS Neurosci Ther 30 , e14600 ( 2024 ). doi: 10.1111/cns.14600 OpenUrl CrossRef 9. ↵ Kim , K. H. & Lee , M. S . GDF15 as a central mediator for integrated stress response and a promising therapeutic molecule for metabolic disorders and NASH . Biochim Biophys Acta Gen Subj 1865 , 129834 ( 2021 ). doi: 10.1016/j.bbagen.2020.129834 OpenUrl CrossRef PubMed 10. ↵ AnyGenes . UNDERSTANDING THE INTEGRATED STRESS RESPONSE (ISR) . ( 2025 ). 11. Laboratory , T. J . Gene Ontology Annotations . ( 2024 ). 12. ↵ Consortium , G. O. “Gene Ontology Resource.” , ( 2025 ). 13. ↵ Sturm , G. et al. A multi-omics longitudinal aging dataset in primary human fibroblasts with mitochondrial perturbations . Sci Data 9 , 751 ( 2022 ). doi: 10.1038/s41597-022-01852-y OpenUrl CrossRef 14. ↵ Sturm , G. et al. OxPhos defects cause hypermetabolism and reduce lifespan in cells and in patients with mitochondrial diseases . Commun Biol 6 , 22 ( 2023 ). doi: 10.1038/s42003-022-04303-x OpenUrl CrossRef 15. ↵ Yang , X. et al. Physical bioenergetics: Energy fluxes, budgets, and constraints in cells . Proc Natl Acad Sci U S A 118 ( 2021 ). doi: 10.1073/pnas.2026786118 OpenUrl Abstract / FREE Full Text 16. ↵ Wieser , W. & Krumschnabel , G . Hierarchies of ATP-consuming processes: direct compared with indirect measurements, and comparative aspects . Biochem J 355 , 389 – 395 ( 2001 ). doi: 10.1042/0264-6021:3550389 OpenUrl CrossRef PubMed Web of Science 17. ↵ Kaspar , S. et al. Adaptation to mitochondrial stress requires CHOP-directed tuning of ISR . Sci Adv 7 ( 2021 ). doi: 10.1126/sciadv.abf0971 OpenUrl FREE Full Text 18. ↵ Young , S. K. & Wek , R. C . Upstream Open Reading Frames Differentially Regulate Gene-specific Translation in the Integrated Stress Response . J Biol Chem 291 , 16927 – 16935 ( 2016 ). doi: 10.1074/jbc.R116.733899 OpenUrl Abstract / FREE Full Text 19. ↵ Pakos-Zebrucka , K. et al. The integrated stress response . EMBO Rep 17 , 1374 – 1395 ( 2016 ). doi: 10.15252/embr.201642195 OpenUrl Abstract / FREE Full Text 20. ↵ Carreau , A. , El Hafny-Rahbi , B. , Matejuk , A. , Grillon , C. & Kieda , C . Why is the partial oxygen pressure of human tissues a crucial parameter? Small molecules and hypoxia . J Cell Mol Med 15 , 1239 – 1253 ( 2011 ). doi: 10.1111/j.1582-4934.2011.01258.x OpenUrl CrossRef PubMed 21. ↵ Batjargal , T. et al. Optogenetic control of the integrated stress response reveals proportional encoding and the stress memory landscape . Cell Syst 14 , 551 – 562 e555 ( 2023 ). doi: 10.1016/j.cels.2023.06.001 OpenUrl CrossRef PubMed 22. ↵ Consortium , G. T . The Genotype-Tissue Expression (GTEx) project . Nat Genet 45 , 580 – 585 ( 2013 ). doi: 10.1038/ng.2653 OpenUrl CrossRef PubMed 23. ↵ Devine , J. et al. Brain-body mitochondrial distribution patterns lack coherence and point to tissue-specific and individualized regulatory mechanisms . bioRxiv ( 2024 ). doi: 10.1101/2024.09.20.614152 OpenUrl Abstract / FREE Full Text 24. ↵ Kontio , J. A. J. , Rinta-Aho , M. J. & Sillanpaa , M. J . Estimating Linear and Nonlinear Gene Coexpression Networks by Semiparametric Neighborhood Selection . Genetics 215 , 597 – 607 ( 2020 ). doi: 10.1534/genetics.120.303186 OpenUrl Abstract / FREE Full Text 25. ↵ Liharska , L. E. et al. A study of gene expression in the living human brain . medRxiv ( 2023 ). doi: 10.1101/2023.04.21.23288916 OpenUrl Abstract / FREE Full Text 26. ↵ Atlas , T. H. P. ( 2024 ). 27. ↵ Cattell , R. B . The Scree Test For The Number Of Factors . Multivariate Behav Res 1 , 245 – 276 ( 1966 ). doi: 10.1207/s15327906mbr0102_10 OpenUrl CrossRef PubMed Web of Science 28. ↵ Bertolo , A. , Baur , M. , Guerrero , J. , Potzel , T. & Stoyanov , J . Autofluorescence is a Reliable in vitro Marker of Cellular Senescence in Human Mesenchymal Stromal Cells . Sci Rep 9 , 2074 ( 2019 ). doi: 10.1038/s41598-019-38546-2 OpenUrl CrossRef PubMed 29. ↵ Hedges , L. V. & Olkin , I . Statistical methods for meta-analysis . ( Academic Press , 1985 ). View the discussion thread. Back to top Previous Next Posted July 01, 2025. Download PDF Data/Code 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 Mapping the GDF15 Arm of the Integrated Stress Response in Human Cells and Tissues 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 Mapping the GDF15 Arm of the Integrated Stress Response in Human Cells and Tissues Janell L.M. Smith , Kamaryn Tanner , Jack Devine , Anna S. Monzel , Taivan Batjargal , Maxwell Z. Wilson , Alan A. Cohen , Martin Picard bioRxiv 2025.01.31.635929; doi: https://doi.org/10.1101/2025.01.31.635929 Share This Article: Copy Citation Tools Mapping the GDF15 Arm of the Integrated Stress Response in Human Cells and Tissues Janell L.M. Smith , Kamaryn Tanner , Jack Devine , Anna S. Monzel , Taivan Batjargal , Maxwell Z. Wilson , Alan A. Cohen , Martin Picard bioRxiv 2025.01.31.635929; doi: https://doi.org/10.1101/2025.01.31.635929 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 Cell Biology Subject Areas All Articles Animal Behavior and Cognition (7622) Biochemistry (17648) Bioengineering (13870) Bioinformatics (41880) Biophysics (21423) Cancer Biology (18553) Cell Biology (25458) Clinical Trials (138) Developmental Biology (13364) Ecology (19866) Epidemiology (2067) Evolutionary Biology (24290) Genetics (15589) Genomics (22475) Immunology (17711) Microbiology (40327) Molecular Biology (17145) Neuroscience (88472) Paleontology (666) Pathology (2826) Pharmacology and Toxicology (4815) Physiology (7635) Plant Biology (15114) Scientific Communication and Education (2044) Synthetic Biology (4286) Systems Biology (9815) Zoology (2268)