Letters https://doi.org/10.1038/s41591-019-0673-2

Undulating changes in human plasma proteome profiles across the lifespan

Benoit Lehallier 1,2,3*, David Gate1,2,3,4, Nicholas Schaum5, Tibor Nanasi1,2,3,6, Song Eun Lee1,2,3,4, Hanadie Yousef1,2,3,4, Patricia Moran Losada1,2,3, Daniela Berdnik1,2,3,4, Andreas Keller 7, Joe Verghese8,9, Sanish Sathyan8,9, Claudio Franceschi10,11, Sofiya Milman8,12, Nir Barzilai8,12 and Tony Wyss-Coray 1,2,3,4*

Aging is a predominant risk factor for several chronic diseases Perhaps the strongest evidence that blood can be used to study that limit healthspan1. Mechanisms of aging are thus increas- aging comes from experiments employing heterochronic parabio- ingly recognized as potential therapeutic targets. Blood from sis, which is a surgically induced state that connects the circulatory young mice reverses aspects of aging and disease across systems of young and old mice. These studies show that multiple multiple tissues2–10, which supports a hypothesis that age- tissues, including muscle, liver, heart, pancreas, kidney, bone and related molecular changes in blood could provide new insights brain, can be rejuvenated in old mice2–10. Plasma (the soluble fraction into age-related disease biology. We measured 2,925 plasma of blood) from old mice is sufficient to accelerate brain aging after proteins from 4,263 young adults to nonagenarians (18–95 infusion into young mice9, and young plasma can reverse aspects years old) and developed a new bioinformatics approach that of brain aging10,14. Together, these studies support the notion that uncovered marked non-linear alterations in the human plasma the plasma proteome harbors key regulators of aging. Identifying proteome with age. Waves of changes in the proteome in the such protein signatures may help in understanding mechanisms fourth, seventh and eighth decades of life reflected distinct of organismal aging. However, plasma proteomic changes with age biological pathways and revealed differential associations have not been thoroughly exploited and require new tools to derive with the genome and proteome of age-related diseases and insights into the biology of aging. In this study, we carried out a phenotypic traits. This new approach to the study of aging led deep proteomic analysis of plasma from young adults to nonage- to the identification of unexpected signatures and pathways narians. Using new analysis tools, we discovered changes in protein that might offer potential targets for age-related diseases. expression across the lifespan and linked these changes to biological Aging underlies declining organ function and is the primary pathways and disease. risk factor for several diseases1. Thus, a deeper understanding of aging is likely to provide insights into mechanisms of disease and Results to facilitate the development of new antiaging therapeutics. A grow- Linear modeling links the plasma proteome to functional aging ing number of investigators have applied genomic, transcriptomic and identifies a conserved aging signature. We analyzed plasma and proteomic assays (collectively referred to as ‘omics’) to stud- isolated from EDTA-treated blood acquired by venipuncture from ies of aging11. Human genetic studies have uncovered relatively 4,263 healthy individuals aged 18–95 years from the INTERVAL15 few modifiers of aging, yet other omics modalities, which measure and LonGenity16 cohorts (Fig. 1a and Extended Data Fig. 1). more dynamic gene modifications or products, have provided valu- Currently, one of the most advanced tools for the measurement of able insights. For example, the transcriptome varies greatly during plasma proteins is the single‐stranded oligonucleotides known as aging across tissues and organisms12, pointing to evolutionarily aptamers17,18, which bind to targets with high affinity and specific- conserved, fundamental roles of developmental and inflammatory ity. To generate a proteomic dataset of the human lifespan, we used pathways13. The protein composition of cells, bodily fluids and tis- the SomaScan aptamer technology, which is capable of quantifying sues changes similarly with age and provides insights into complex thousands of proteins (Supplementary Tables 1 and 2) with high biological processes, as proteins are often direct regulators of cel- precision within and between runs19 (Supplementary Table 3). The lular pathways. In particular, blood, which contains proteins from INTERVAL and LonGenity datasets analyzed here can be interro- nearly every cell and tissue, has been analyzed to discover biomark- gated with an interactive web interface (https://twc-stanford.shin- ers and gain insights into disease biology. Accordingly, organismal yapps.io/aging_plasma_proteome/). aging results in proteomic changes in blood that reflect aspects of Because females have a longer average lifespan than males20, aging of different cell types and tissues. we assessed whether sex and aging proteomes are interconnected

1Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA. 2Wu Tsai Neurosciences Institute, Stanford University, Stanford, CA, USA. 3Paul F. Glenn Center for the Biology of Aging, Stanford University, Stanford, CA, USA. 4Department of Veterans Affairs, VA Palo Alto Health Care System, Palo Alto, CA, USA. 5Institute for Stem Cell Biology and Regenerative Medicine, Stanford University, Stanford, CA, USA. 6Institute of Cognitive Neuroscience and Psychology, Hungarian Academy of Sciences Research Centre for Natural Sciences, Budapest, Hungary. 7Clinical Bioinformatics, Saarland University, Saarbrücken, Germany. 8Institute for Aging Research, Department of Medicine, Albert Einstein College of Medicine, Bronx, NY, USA. 9Department of Neurology, Albert Einstein College of Medicine, Bronx, NY, USA. 10Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna, Bologna, Italy. 11Department of Applied Mathematics, National Research Lobachevsky State University of Nizhny Novgorod, Nizhny Novgorod, Russia. 12Department of Genetics, Albert Einstein College of Medicine, Bronx, NY, USA. *e-mail: [email protected]; [email protected]

Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine 1843 Letters NATurE MEDicinE

(Fig. 1b–d). The proteins most strongly changed with sex included sets of proteins may be required to model changes in a large set of the well-known follicle-stimulating hormone (CGA FSHB), clinical and functional parameters (Extended Data Fig. 3d). human chorionic gonadotropin (CGA CGB) and prostate-specific As most biological pathways that change with age are evolution- antigen (KLK3). With age, the most prominent overall changes arily conserved23, we next identified aging-related proteins conserved with respect to fold change and statistical significance included between mice and humans. We analyzed mouse plasma (n = 110; sclerostin (SOST), ADP ribosylation factor interacting protein 2 age, 1–30 months) using SomaScan, which reliably measures hun- (ARFIP2) and growth differentiation factor 15 (GDF15), in addi- dreds of non-human proteins and has proven useful in mouse stud- tion to several proteins that differed with sex, such as CGA FSHB. ies6,24 (Fig. 1i and Supplementary Table 9). In mice, 172 proteins The proteins most strongly associated with age also changed sig- changed with age (out of 1,305 proteins measured; Supplementary nificantly with sex (Fig. 1d): 895 of the 1,379 proteins altered Table 10; q < 0.05), and 46 proteins overlapped with human aging- with age were significantly different between the sexes (q < 0.05; related proteins (Fig. 1j). Remarkably, many of these proteins were Supplementary Table 4). These results are aligned with several stud- also modulated by heterochronic parabiosis: young mice exposed ies that demonstrated that males and females age differently21. To to old plasma (young heterochronic mice) showed a relatively older determine whether these findings are representative of the general plasma signature, whereas aged mice exposed to young plasma population, we compared changes identified in this study with (old heterochronic mice) showed a younger signature (Fig. 1k). findings from four independent cohorts from the US and Europe Altogether, standard linear modeling of the plasma proteome dur- (n = 171; age range, 21–107 years; Extended Data Fig. 1d) and with ing the human lifespan revealed established aging pathways, possibly an independent study22. Although these independent cohorts used indicating accelerated and decelerated aging in humans and mice. an older version of the SomaScan assay measuring only a subset of Intriguingly, changes to the conserved aging-related proteins did not the current proteins (1,305 proteins; Supplementary Table 2), we occur simultaneously (Fig. 1l). Thus, the chronology of aging in the observed high consistency of the aging and sex proteomes across plasma proteome requires further investigation. cohorts (Extended Data Fig. 2). To establish the biological relevance of these changes, we que- Clustering protein trajectories reveals undulation of the aging ried the Gene Ontology (GO), Kyoto Encyclopedia of Genes and plasma proteome. Although standard linear modeling showed Genomes (KEGG) and Reactome databases and measured enrich- prominent changes in plasma protein composition, the undulating ment of proteins in pathways using sliding enrichment pathway behavior of the 46 conserved proteins (Fig. 1l), and, more globally, analysis (SEPA) (Supplementary Tables 5 and 6). The heat maps the 2,925 plasma proteins as a group when they were visualized as produced by SEPA first illustrated the relationship between the top z-scored changes across the lifespan, was striking (Fig. 2a,b). These 100 proteins and the biological pathways they represent; second, undulating patterns were detected in independent human cohorts the heat maps emphasized how a restricted list of top aging-related and in mice (Extended Data Fig. 4), suggesting that they are robust proteins revealed biological pathways that would have escaped and conserved. common pathway mining modalities (Fig. 1e). SEPA indicated that To reduce the complexity of the proteome, we grouped proteins incremental lists of proteins are needed to determine the biologi- with similar trajectories using unsupervised hierarchical cluster- cal functions of sex-related proteins and pointed to expected dif- ing (Fig. 2c) and identified eight clusters of protein trajectories ferences in hormonal metabolism and activity. Conversely, an changing with age, which ranged in size from 8 to 1,415 proteins extensive list of aging proteins contained enrichment for blood- (Supplementary Table 11). In addition to linear patterns (clusters 1 related pathways, such as heparin and glycosaminoglycan binding, and 5), several non-linear trajectories were evident, including step- as recently reported22. wise, logarithmic and exponential trajectories (clusters 2, 3, 4, 6, 7 To determine whether the plasma proteome can predict biologi- and 8) (Fig. 2d). Notably, these cluster trajectories were similarly cal age and serve as a ‘proteomic clock’, we used 2,817 randomly detectable in independent cohorts (Extended Data Fig. 5). Of the selected individuals to fine-tune a predictive model that was tested eight clusters analyzed, six were enriched for specific biological path- on the remaining 1,446 individuals (Fig. 1f). We identified a sex- ways (q < 0.05; Extended Data Fig. 6 and Supplementary Table 12), independent plasma proteomic clock consisting of 373 proteins suggesting distinct, yet orchestrated, changes in biological pro- (Supplementary Table 7), which was highly accurate in predict- cesses during the lifespan. For example, proteins present in blood ing age in the discovery, validation and four independent cohorts microparticles consistently decreased with age (cluster 5), and other (r = 0.93–0.97; Fig. 1g and Extended Data Fig. 3a,b). Remarkably, blood-related pathways, such as heparin and glycosaminoglycan individuals who were predicted to be younger than their chrono- binding, increased in a two-step manner (cluster 4), whereas levels logical age performed better on cognitive and physical tests (Fig. 1h of proteins involved in axon guidance and EPH–ephrin signaling and Supplementary Table 8). Although a reduced model compris- remained constant until age 60 before rising exponentially (cluster 6) ing only nine proteins predicted age with good accuracy (Extended (Fig. 2d). Altogether, most plasma proteome changes across the Data Fig. 3c and Supplementary Table 7), a combination of different lifespan were non-linear.

Fig. 1 | Linear modeling links the plasma proteome to functional aging and identifies a conserved aging signature. a, Schematic representation of analysis of the plasma proteome. b,c, Volcano plots representing changes of the plasma proteome (n = 4,263) with sex (b) and age (c). Linear models (l.m.), adjusted for age, sex and subcohort, were tested using the F-test. d, Relative percentage of variance explained by age and sex. Values for each plasma protein are connected by edges. e, Pathways associated with sex and age identified by SEPA (n = 4,263). Proteins upregulated and downregulated were analyzed separately. The top ten pathways per condition are represented. Enrichment was tested using Fisher’s exact test (GO) and the hypergeometric test (Reactome and KEGG). f, Schematic representation of biological age modeling using the plasma proteome. g, Prediction of age in the validation cohort (n = 1,446) using 373 plasma proteins. The Pearson correlation coefficient between chronological and predicted age is given. h, Association between delta age (difference between predicted age and chronological age) and functional readouts in old. Top associations in both the discovery and validation datasets are represented. i, Schematic representation of the comparison between the human and mouse aging proteomes. j, Conserved markers of aging. Both human and mouse aging effects are signed by the beta age of the corresponding linear analysis. Forty-six plasma proteins change in the same direction in mice and humans (red dots) and define a conserved aging signature. k, Alteration of the conserved aging signature by parabiosis. Normed principal-component analysis was used to characterize changes of the conserved aging signature when mice were exposed to young or old blood. l, Age-related changes of the conserved aging signature. Plasma protein levels were z scored, and aging trajectories were estimated by LOESS.

1844 Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine NATurE MEDicinE Letters

a b Sex proteome c Aging proteome

4,263 participants (18–95 years) CGA.FSHB SPINT3 KLK3 350 DEFB104A 350 SOST CGA.FSHB LEP CGA.LHB 300 IGFBP6 300 CGA.CGB ARFIP2 GDF15 250 CRISP2 250 MLN IGDCC4 200 VIT CLIC5 200 PTN NUDT16L1 RET CGA.LHB LHB SLITRK4 IL1RL1 150 IGFBP3 MB 150 MSMP SCARF2 CGA.CGB ( P value sex) ( P value age) 10 100 10 100 2,925 Male –log plasma 50 –log 50 proteins 0 0 Female –0.4 –0.2 0.0 0.2 0.40.6 –0.005 0.000 0.005 0.010 0.015 Age Up in Up Down Up female Sex effect (beta lm) in male with age Age effect (beta lm) with age

d e Aging vs sex pathways Aging vs sex effects Aging Sex 10 10010 100 R-HSA-373076-Class A/1 (rhodopsin-like receptors) R-HSA-500792-GPCR ligand binding 0.6 hsa04080-neuroactive ligand–receptor interaction R-HSA-388396-GPCR downstream signaling GO:0005179-hormone activity GO:1901681-sulfur compound binding 0.4 GO:0008201-heparin binding GO:0005539-glycosaminoglycan binding GO:0044449-contractile fiber part GO:0030016-myofibril

(partial Eta-2) 0.2 GO:0043292-contractile fiber GO:0030017-sarcomere hsa01100-metabolic pathways Percentage of variance GO:0005578-proteinaceous extracellular matrix 0.0 GO:0031012-extracellular matrix R-HSA-392499-metabolism of proteins Age Sex R-HSA-948021-transport to the Golgi and subsequent modification R-HSA-975576-n-glycan antennae elongation in the medial/trans-Golgi GO:0005179-hormone activity Signed –log (FDR) R-HSA-2980736-peptide hormone metabolism 10 –3 +3

fg h Prediction of biological age Associations with ∆age 100 Validation dataset (n = 1,446) 4 Trail making test B time Pearson: 0.97 80 2 Discovery Validation ∆ age (n = 2,817) (n = 1,446) 60 y 0 Aging age clock ∆ 40 –2 Predicted age (years) Association with ∆ age 20 validation (signed effect) –4 Physical grip Clinical and functional 20 40 60 80 100 –4 –2 024 x readouts Chronological age (years) Association with ∆age discovery (signed effect)

ijConserved aging proteome k 10 INSR GDF15 Alteration of the conserved aging CGA TSHB signature by parabiosis IL18BP 5 CCL7 MRC1 CHRDL1 Human aging proteome 26 proteins Young–iso 0 Young–het

Conserved KLK7 PC1 (33%) –5 GDF11.MSTN aging markers MSTN Old–het MIA –10 EPHB6 Mouse aging proteome Old–iso Mouse aging (signed effect) 20 proteins –200 –100 0 100 200 300 PC2 (15%) Human aging (signed effect)

l Non-uniform changes of the conserved aging signature Young

Old Young

Old B MB MI A PYY UB B ED A BO C IGF 1 GCG INS R KLK7 TYK2 CD3 6 CCL 5 CCL 7 SNX4 CAST PPBP SGTA BMP7 GHRL STIP1 ISLR 2 MSTN MRC1

z score WNK3 KRT18 GDF15 PLCG1 EPHB6 IL18BP RTN4 R EDA2R LSAM P CHI3L1 ALCAM HMOX 2 CYP3A4 HAPLN 1 CHRDL 1 SPOCK 2 TNFAIP6

–0.5 +0.5 TNFSF15 CGA.TSH B TNFRSF1A GDF11.MSTN INHBA.INHB

Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine 1845 Letters NATurE MEDicinE

a b Protein trajectories c Protein profiles Protein clusters 2

15 1

10 0 Height

5 –1 2,925 plasma proteins Protein levels ( z score) 2 3 2 PTN.3045.72. 3 3220.40. 3 T. RE ARFIP2.12630.8. 3 OMD.5358.3. 3 CHRDL2.6086.15. IGDCC4.9793.145. GDF15.4374.45.2 LRRC15.6557.50. 3 2 3 3 3 CHAD.13460.4. 3 1 2 0 3 2 2 3 3 1 3 3 2 3 .13101.60. CDON.4541.49. 3 3 3 COL11A2.11278.4. MLN.5631.83. TM2.6904.14. 3 3 2 3 3488.64.2 3 R 3 SOST CST6.14711.27. 3 3 EGFR.2677.1. 2 3 T. 2 3 3 3 LR 3 3 CA 3 3 3 3 2 3 3 3 RP.2813.11.2 3 3 C1QTNF3.7251.64. NELL1.6544.33. 3 CCDC80.3234.23. CHRDL1.3362.61. NBL1.2944.66.2 SELL.4831.4. 1 3 3 3 MMP12.4496.60.2 3 3 1 CGA.FSHB.3032.11. 1 2 2 3 PRSS2.5034.79. AG 3 3 SCARF2.8956.96. 3 3 AIMP1.2714.78.2 3 5 3 3 3 3 3 MSMP.8080.24. 3 APOL1.11510.31. 3 3 3 AP2.9294.45. 3 1 QPCTL.8866.53. 3 1 3 2 3 3 3 3 2 3 2 2 2 3 3 3 3 SCARA5.10419.1. 3 3 3 NPPB.7655.11. 3 1 3 1 1 3 3 2 3 1 2475.1. 3 3 3 3 4 3 COL6A3.11196.31. 3 MF 3 3 6 3 3 2 1 2 3 3 2 2 3 SUMF1.6941.11. 3 3 3 3 3 3 3 3 3 3 3 T. 3 2 3 2 COLEC12.5457.5.2 2 3 FSTL3.3438.10. HMGN1.9187.2. 1.6375.75. 3 3 LST1.9531.24. 3 1 3 3 CTRB1.5671.1. 3 3 3 3 2 3 3 3 3 2 3 3 2 3 AS.9459.7. 3 2 KLK7.3378.49. 3 3 3 3 3 3 1.13381.49. CXCL13.3487.32.2 3 3 3 2 2 3 CR1.9021.1. 3 3 2 1 KI ENPP5.6556.5. 1 2 2 3 2 3 SCG3.7957.2.3 2 3 3 MUSK.11547.84.3 4 3 F 3 3 1 2 3 3 2 2 WFDC2.11388.75. LT 3 3 5 3 2 3 3 LHB.8376.25.4 EPHB2.8225.86. 3 2 3 3 3 3 3 3 3 3 3 3 3 3 1 3 3 3 3 1 3 AP.5029.3.1 3 TN4.7083.74. 3 3 2 3 3 3 2 3 3 3 3 3 3 6 3 5 3 3 3 4588.1. 3 2 3 3 3 3 3 LT 2 3 3 3 3 3 3 3 ASC.7179.69.3 3 3 3 2 R1.2590.69.4 2 3 2 2 1 3 3 3 3 1 3 3 3 3 3 .3364.76.2 MAP2K4.5242.37.3 3 3 1 3 THBS3.8982.65. 3 3 3 3 3 F AV 3 3 2 2 1 3 5 3 3 3 3 3 3 2 3 3 3 2 3 3 PTPRD.9296.15. 3 3 3 3 3 3 3 3 3 3 3 2 3 FCRL4.8973.23. 2 3 3 1 3 3 2 TXNDC5.11212.7.3 3 3 3 3 3 3 TFF2.9191.8.3 2 3 3 4 3 UNC5B.7776.20. 3 1 2 2 3 3 3 3 3 3 3 3 3 1 3 3 3 3 1 Y. 3 3 3 2 1 3 3 3 3 2 2 3 3 3 3 2 3 MRC2.3041.55. 1 3 3 3 2 3 3 2 3 2 CDSN.7085.81.3 3 3 SIGLEC7.2742.68. 2 3 3 3 2 3 3 2 3 H 2 3 3 3 3 3 3 3 3 1 3 3 3 3 3 3 3 3 3 2 1 3 3 4 3 3 3 3 3 1 3 FBP1.7206.20.3 3 3 3 3 2 2 3 2 MA 3 3 3 3 2 3 2 2 2 TPD2.7948.129.3 3 2 3 1 3 3 2 3 3 3 1 3 3 3 3 3 3 3 3 2 1 3 3 3 3 1 2 TNFRSF11B.8304.50. 3 3 3 3 NF ESM1.3805.16.2 3 2 3 3 3 4 3 3 3 2 3 3 3 2 3 3 3 3 2 3 3 3 3 1 3 3 3 3 3 XX Y 3 3 2 3 3 3 2 3 3 3 3 3 2 A.3169.70. 2 1 3 2 3 2 3 ADPGK.6221.1. 2 IL19.3035.80. 3 RO 3 3 2 3 3 1 3 3 3 3 1 3 2 3 2 1 3 .2421.7. 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 OBP2B.5680.54. 3 3 2 1 1 3 TFPI.3336.50. 3 3 3 2 3 3 3 2 4979.34. 3 3 3 3 2 3 3 PIANP.9599.6. 2 3 3 2 2 SSR1.8106.15.3 1 3 3 3 3 3 2 3 3 2 3 3 3 3 3 2 3 3 3 3 3 3 3 2 4 1 CA6.3352.80.3 3 2 2 3 3 2 .14006.36. 3 3 3 3 3 3 3 .8337.65.3 3 3 3 3 3 PTGDS.10514.5. 3 3 3 3 2 4 3 3 3 3 3 3 3 3 3 3 2 3 3 3 1 3 1 3 3 3 FJX1.7921.65.3 3 3 3 3 3 3 3 3 3 3 2 3 1 3 3 3 3 3 3 3 1 3 COL15A1.8974.172. 3 3 3 3 3 3 2 3 3 3 3 3 3 3 1 3 3 3 3 3 2 2 CCL11.5301.7. 3 3 3 3 3 3 1 3 3 3 2 SCUBE1.8989.40.3 2 3 3 3 PP 1 3 3 3 3 3 3 3 3 2 3 3 3 1 3 3 3 3 1 3 3 3 3 3 3 2 3 3 3 2 3 3 3 2620.4. 2 3 3 3 3 3 3 3 2 3 3 3 2 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 3 3 3 3 .12370.30. 3 2 3 3 3 3 3 1 2 2 3 3 3 3 3 3 3 3 3 3 3 3 3 1 3 3 3 1 3 3 3 3 3 3 3 3 3 3 3 1 3 3 3 3 KDR.3651.50. 3 3 3 3 CGB2.6213.10. 2 3 MXRA8.10521.10. 3 PTH.5954.62. 3 3 3 3 2 3 3 2 3 3 3 3 3 3 3 3 2 3 1 2 3 3 3 CELA2A.7140.1. 3 3 3 2 TCH1.5107.7.2 3 2 3 2 3 3 3 3 3 3 3 3 3 3 1 2 3 3 3 3 3 .13464.8. 3 3 4 GL 2 3 P1B2.7218.87. 3 3 3 3 3 3 3 C9.3060.43. 3 3 3 T. 3 3 3 3 3 2 2 3 3 3 3 1 3 3 3 2 3 2 3 1 3 3 3 3 3 3 3 3 3 1 3 AIP6.5036.50.1 2 3 3 2 3 1 2 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 3 3 3 4162.54.2 3 EPHB6.5078.82. 3 3 3 3 3 3 3 3 3 3 3 3 3 3 1 2 3 3 2 2 1 3 3 3 1 3 3 3 2 3 3 2 1 3 3 COL1A1.11140.56. 3 3 3 PCOLCE.11237.49. 3 3 3 3 3 3 3 3 3 3 3 GK.3894.15. 3 1 3 3 3 CARS.14098.28.3 3 3 3 2 3 3 3 3 3 3 1 3 2 3 3 3 3 SCO1.7853.19. 3 3 2 3 3 3 3 3 3 3 3 2 3 3 3 B4GA 3 3 3 3 4 3 3 2 3 3 3 3 3 3 3 3 2 3 1 3 3 3 3 3 3 3 1 3 3 3 CHIT1.3600.2. 3 3 3 3 3 2 3 1 2 3 4 3 3 3 3 INHBC.6408.2. 3 3 2 3 2 TN2.3325.2.2 3 3 3 3 CTSV 2 3 3 3 3 3 3 3 3 3 3 3 3 3 H.11424.4. 2 1 3 3 3 3 3 3 3 1 3 3 3 3 2 3 3 3 3 3 3 2 2 4 3 2 3 3 3 3 3 3 3 3 3 3 2 3 BP4.13133.73.3 4 3 3 3 3 3 3 3 2 3 3 3 3 3 .9986.14. 3 3 1 3 ALDOC.9876.20. 3 3 2 3 3 1 3 3 1 3 3 3 3 3 3 3 3 3 3 3 2 1 3 3 3 3 6414.8. 3 1 3 1 3 3 TUM.8252.2. 3 3 3 3 3 3 3 3 3 3 1 T. 3 3 3 3 2 3 3 3 2 3 3 2 3 3 3 3 2 3 1 3 3 3 3 3 3 3 2 3 3 3 2 3 3 3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 3 3 3 4 3 3 3 3 3 2 2 1 3 3 3 3 3 3 3 3 3 3 1 3 3 3 1 1 3 3 3 3 2 3 1 3 3 F 3 3 2 2 3 BOC.4328.2.2 3 3 2 3 3 3 1 –2 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 2 3 3 3 3 3 3 3 M2.2997.8.1 3 3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 3 3 3 3 3 3 2 2 O MXRA7.8005.1. LILRA4.8299.66. 3 3 3 3 3 3 2 2 2 3 3 3 3 3 3 3 1 I2.14066.49. TGFBI.3283.21.1 3 2 2 3 4 3 3 IDU 3 3 3 3 2 3 3 3 3 3 3 3 3 3 2 3 3 3 3 1 3 3 .8250.2. 3 3 3 1 3 3 1 3 3 3 3 3 3 3 3 3 3 3 3 1 3 3 3 3 1 RU 3 3 2 2 A 3 F. FCER2.3291.30. 2 3 3 3 3 3 2 3 EFNA5.2615.60.2 3 3 2 3 REN.3396.54. 3 3 3 .14094.29.3 3 3 3 1 3 3 3 3 3 3 3 3 3 3 3 PSD2.9118.7. 3 3 3 ASL.11241.8. 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2 3 GDF5.2752.62.2 2 1 3 OR2.7861.9.3 1 3 2 3 3 4 5 3 3 3 3 BCAN.3461.58. 3 3 5 SEMA3C.6448.36. 2 3 3 3 2 3 2 3 3 PLA2G12B.9380.2.3 3 NDST1.6927.7. EPYC.9278.9. CD2.7100.31. 3 AT 3 3 2 3 1 F. 3 2 O .11208.15. 3 GFRA2.2515.14. J CDH5.2819.23.2 NUDT9.9482.110. W 13992.12.3 F AC RGMB.3331.8. DP N AP5.6440.31. .5509.7. 2 T LT GP1BA.4990.87. DLK2.9359.9. CCL21.2516.57. NRCAM.5109.24. P AS.5392.73.2 GAS1.5463.22.3 CGA.LHB.2953.31. GGT2.6334.9. FA N HFE2.3332.57. YP2.12812.25. BO2.5116.62. DEFB1.6629.3. NGFR.8374.5.3 MBL2.3000.66. CCL14.2900.53. 2780.35.2 THBS2.3339.33.1 MA RIPK2.8993.151. R CGA.CGB.4914.10. BDNF MASP1.3605.77. 4 POFUT2.6042.52. 3 7849.3.3 PVRL2.6245.4.3 IL16.2774.10. P2.4.14615.46. 3 JA EFNA2.14124.6. PRSS1.3049.61. TCN2.5584.21.3 TWSG1.9234.8. TNFRSF17.2665.26.2 6234.74. IGLON5.6478.2. MB.3042.7. GPC6.5350.14.2 .9212.22.3 F .4962.52. .13098.93.3 TN DCN.2666.53. 2 N .6295.67. TNFRSF1B.3152.57. .3727.35.1 LMAN2.9468.8. 3 F CA3.3799.11.2 T1.8229.1. F. AK1.5012.67. Y1.7097.8. STC1.4930.21. .2681.23.2 GHR.2948.58.2 AG BPGM.12020.39. 3 GNMT 3195.50.2 FCRL1.5728.60. SRSF6.11573.3. FCN2.3313.21.2 PI3.4982.54.1 CSMD2.9971.5.3 F CNTFR.14101.2. F. CTBS.6115.40.3 APOF C1.9337.43. 3 .5744.12. F TNC.4155.3. PCBD2.6899.37.3 H6PD.7161.25. PA GFRA1.3314.74. U.4158.54. Y KCNE5.8908.14. L CDH3.2643.57.2 GER.4125.52. T. IL6S CECR1.6077.63. Y MRC1.2637.77.2 FNDC5.8041.5. A.2597.8. F A2.9276.7. TREM1.9266.1. OA A.3868.8. T. NP PCDHB10.9963.19. POR.2731.29. FURIN.6276.16.3 PBP.12378.71.3 FGF20.2763.66. SIGLEC12.8352.26. ERLEC1.8957.72.3 TTN.11352.42. ERBB3.2617.56.35 PT 14112.40. AMTS5.3168.8. MF NR RNASE1.7211.2. IL18BP.3073.51. M151A.7856.51.3 C6.7228.2.3 CFI.2567.5.6 LSAMP.2999.6. TLL1.6383.90.3 CXCL10.4141.79. ITIH1.7955.195.3 NTN4.3327.27.1 TA AC TESC.12831.21.3 CA1.4969.2. 1 2837.3.2 T M RGMA.5483.1. 3 TMEM132B.8890.9. CXCL16.2436.49. 86.3623.84.4 A.4148.49.2 CAN.3280.49.2 IBSP.3415.61. A RC.3043.49. S PRL.2585.2. KIRREL2.9917.16. RO LILRB2.5091.28. 12423.38.3 SHISA3.7057.18.3 CD209.3029.52. A 5801.72.3 ELF5.13457.33. 3 F SPINK6.5731.1.3 T2.5030.52. 1 IL1R2.14133.93. A C14orf93.6439.59.3 Y RSPO4.8464.31. .12700.9. .8381.18.4 PSG5.9314.9. AMP8.7064.2. GCG.4891.50. 1 G CHGB.8235.48. Y96.3647.49. 4 9017.58.3 TFF3.8323.163. CD55.5069.9.3 C X2.11462.8.3 AIFM1.13424.51. 3 REG4.11102.22. IM3.6574.11.3 TNNI3.5441.67. G3.9950.229. SPOCK2.5491.12. MAN1B1.7071.23. 4 T3.6897.38.3 M19A4.6511.17.3 SYT7.7121.2.3 N4R.5105.2. POMC.9204.33.3 CCL23.2913.1. EG AB1.5259.2. 3 TN3.13720.95. TFF1.9185.15. PRCP.5722.78. 1.6070.11.3 AD1.9231.23. .3026.5. GZMA.3440.7. 2 GREM2.5598.3. L VCR2.5134.52.2 AMBN.6522.57.3 CD JTB.9038.12. .9358.3. TNXB.5698.60.3 NS D LMAN2.7638.30.3 LY. .8864.59.3 A .12706.2. 3078.1. 2 8364.74. 3 T CTSA.3179.51.2 KLK3.8468.19. T3.7970.315. CB.12900.29. CCL23.3028.36. G1.5092.51.3 SELE.3470.1. VI AM174A.6597.24.3 SFN.4829.43. IP2.10630.5. 3 L SEMA4D.5737.61. MANSC1.9557.5. A 5957.30. A3.4993.16. 1 A4H.3204.2. VL2.13575.40. 3 A1.6936.7. .5011.11. M150B.6284.7. Y 5617.41.3 PENK.9076.25. F2.5316.54. 3 CST8.10572.65. 3 A.13434.172. 7790.21.3 NPTX2.6521.35.3 6.7981.230. G3.10078.5.3 SEC13.14689.3. 3 ARHGAP30.12807.89. V F. GHRL.8447.11.3 R IGFBP6.14088.38. CCL25.2705.5. CPZ.6493.9.3 F. A.3466.8. AM11.6586.19. PIANP.14114.18. T A BMX.3414.40. A VC CP TA UFY1.11425.31.3 CEL.9796.4. PY A.5443.62.2 .9353.8. 3 MCFD2.10476.23.3 M4.6572.10. ASAH2.3212.30. P.9028.5.3 HS6ST2.13524.25. COL18A1.2201.17. APCS.2474.54. PTPR . FGF8.4394.71.2 MANEA.8014.359. P CTSZ.4971.1. 14705.1. 5822.22. FKBP6.12529.32. 12663.1. 1.8369.102. A.7167.102.3 LRIG3.3322.52.2 IGFLR1.7244.16. O X1L.5599.88. 3 T. LT. .ITGB5.4917.62. CDH15.11215.6. FKBP7.9288.7. F8.13499.30. COCH.7227.75.3 FA 2.9595.11.3 M3B.9177.6. 3 UBE2J2.8802.24.3 1.11300.32.3 F10.4878.3.1 FIG A2R.3083.71.1 DLL1.5349.69. R BDN.8613.97. M2.13534.20. IGF1R.4232.19. HG LY TB.3506.49. CNTN2.3296.92.2 UBB.6641.60.3 RPN1.6458.6. .3489.9. C2.9361.7. A BPI.4126.22.1 E2.2805.6. APOM.10445.20. . T1.4912.17. 1 NAB1.13933.276. T F 1.4547.59.2 AHSG.3581.53. EPHA5.3806.55. .7092.7. CRHBP.6039.24. HBEG LCN2.2836.68.2 A.4499.21.1 BID.5798.3. GDF11.MSTN.2765.4. YN.2635.61.2 A CCL16.4913.78. TIMP4.6462.12. CLEC4M.3030.3.2 N Q.7625.27. PP F. A1.8909.77.3 AT 2.5756.66.3 8388.24.3 PA CADM3.3630.27. 4 HAPLN1.3196.6. .12748.6.3 UQCC3.8228.10. 10671.67.3 CCL17.3519.3.2 P1.3640.14. 13387.55.3 CD8B.9310.2.3 8.2859.69. 2 GPC3.4842.62.2 AFP.5792.8. T. PSMD11.13572.43.3 HPGDS.12549.33. KR AKT2.14685.17.3 INS.4883.56. IGFBP1.2771.35. ICS.9841.197.3 .10416.79.3 CBLN1.9313.27. 3 PL CXCL11.3038.9. D SLC5A5.12826.5. MIA.2687.2.1 13497.34. RNASE6.5646.20. A3.7888.58.3 CLIC5.12475.48. L1.11192.168. W ABO.9253.52. GEM.12817.1. T1.7224.11.3 ICOS.14084.191.3 CT SURF1.8009.121. 3 PI15.5745.64. PRSS22.4534.10. A4.9267.2. T1.5638.23.3 MPO.2580.83. RNF165.11561.32. M.5620.13.3 EMC4.13516.46.3 L1.4905.63. T. P EVI2B.13028.2. CA9.3798.71. .12686.15.3 6604.59. 3 GE3.6442.6. CANT1.6480.1. 3 FCGR2A.3309.2. TMED10.6506.54. INA.11436.6.3 F. FA T3.3437.80.3 CD3E.8069.85. 5.10425.3.3 C3.2755.8.2 LA 5936.53. F. NCR1.8360.169. KIR2DS2.10428.1. 3 HINT2.5612.16. 4306.4.2 P PLG.3710.49. T7.7806.33.3 KLK15.6491.59. CYTH4.12746.4. KAL1.6603.18.3 DSG2.9484.75. SCGB1C1.5960.49. FA C1QC.14100.63.3 C EDN2.12574.36. ESD.4984.83. 9.3324.51. FUT5.4549.78. L CBLN4.5688.65. 3 CDNF SGT R .2975.19.2 AMTS13.3175.51.5 EFNB1.13104.32. S100A6.13090.17. SCIN.12684.5.3 N M3B.5618.50.3 NMB.9321.400. TN.2939.10.2 F3.4931.59. 1 ZL1.9207.60.3 MMP10.8479.4.3 TM2.8906.60.3 NEGR1.7050.5. EMILIN3.8773.172.3 3.2611.72.2 LT F10.3077.66. SELPLG.11266.8. 3 GX .4132.27. 2 FA TP.12513.8. G RT MSMB.10620.21. 3 NELL2.6022.57.3 CFP.2960.66. .10551.7. A.4703.87.2 A6.13672.3. 8843.34. R.2652.15. 1 5623.11.3 T SIRPB2.5669.26. PRLH.6543.182. ISL1.11549.6.3 F. PPT1.9244.27.3 J F. ERAP2.8960.3. .8759.29.3 CDH12.10701.30. 3 PA 4B.7141.21. 3 IL2.3070.1. QP VTN.13125.45.3 A4.9744.139. SEMG1.7115.5. T. DNAJC10.8297.8.3 P4.5636.10.3 IGFBP7.3320.49. Z.4920.10. P CILP.5717.2.3 EIF5.2612.5.2 NPC2.6259.60. MSTN.14583.49. T LCK.4560.34.2 4C.7208.60. 3 NTN1.6649.51.3 AP.5739.75.3 G7.12627.97. SMOC1.13118.5. HA A CD34.9023.9.3 A.2826.53. OX RNASE4.5644.60.3 SI TIE1.2844.53. INSR.3448.13.2 IL22.2778.10. CD4.3143.3. 1.5716.49.3 P5.3232.28. AT DMPK.9010.3. CGA.14056.4. LPO.4801.13. GPC1.8697.38.3 FGR.3810.50. DRGX.8034.6. CD93.14136.234. 3 DCBLD2.9338.2.3 O.5861.78. MSRA.7137.8.3 SERPINF2.3024.18.2 ANGPTL1.11142.11. LT .2737.22.2 SDF2L1.6990.44.3 . B1.14599.18. BCL2A1.3413.50.2 M1L1.13652.2. CD36.2973.15. PIGR.3216.2. 1.12436.84.3 C1.8985.13.3 CL POSTN.3457.57. CCL3.3040.59. A .5743.82.3 .9855.10.3 VTI1B.8963.8.3 AR CA NADK.13624.17.3 FA CISD1.7745.3.3 V2.9457.3. 3 CD74.6974.6.3 ADGRF5.6409.57.3 A GRID2.12758.47. S COL28A1.10702.1. AL1.8309.12. 3 WNK3.5493.17. IGFBP3.2571.12. TP53I11.13022.20. T IL22RA2.5087.5.3 APP.3171.57. A9.7809.22.3 SIAE.9263.57. MMP9.2579.17. LAMP3.9355.26. 1.5639.49. 5.11177.16. 3 TMUB2.9226.6.3 SPON1.4297.62. .12033.3.3 L AP.9910.9. KLK11.2831.29. GN ZP4.7766.25.3 R2.4959.2. 1 PEAR1.8275.31.3 DCK.9836.20. .9219.70. A1.4209.60.2 GRPEL1.7113.1. GLCE.7808.5. ABL2.5261.13. F. RV CHGA.8476.11. IGF1.2952.75.2 ALCAM.5451.1.3 T. MFN1.11364.18. 3 LGALS9.9197.4. 3 MDK.2911.27.2 CUZD1.7943.16. NPS.6390.18. P1.2876.74. IM T3.10346.5. 3 AR.2977.7. 2.13122.19.3 Y. SECTM1.13093.6.3 MPZ.10615.18.3 A1.6473.55.3 C3.7847.66. 3 MAZ.13436.54.3 J. BA 2.2622.18. SLC35G2.13501.10.3 ASPH.6998.106. STX18.9037.1.3 LT A.9389.12.3 AF2.9738.7.3 ENPP7.4435.66. SLC9B2.9088.20. MENT C.13711.10.3 A.12667.2. CNP.6609.22. SGCB.7034.4. SPINK9.8042.88.3 EPO.5813.58. .9826.135. 3 IL2RB.9343.16. 3 PA CBR1.12381.26. 3 T AZGP1.9312.8. .1.12531.5. 3 NSDHL.8038.41. AP1.12510.3. 3 ENO2.10339.48. RA.11382.5. BMP7.2972.57. T EHMT2.5843.60. T AV SH2D1A.4567.82.2 FKBP14.9340.17.3 DIABLO.3122.6. RP1.3329.14.2 CD63.9190.7.3 PTPN4.14254.27. 1.9767.22.3 NDC80.12730.3. IST1.12434.25.3 ISM1.8355.80. EFNB2.14131.37. RFESD.13603.7. F. LIFR.5837.49. B.8894.80.3 J AD RT AC P IRF6.9999.1. .13929.27. 3 CCL28.2890.59. L1.11592.1. IL17RA.2992.59. F. V1.13023.8. .12847.27. CLPS.5749.53. VEG FA T5.12461.8.3 CSRP3.9171.11.3 SUSD1.9582.93.3 CP1.3858.5.1 D ME T. CAPN3.12385.4. HMGCS1.13496.19.3 T3.6921.24.3 LAMC2.9580.5.3 CCDC126.6388.21. NAAA.3173.49. 2 LC KCNF1.12834.3. 3 T. CTSS.3181.50.2 MDM1.7898.29. TGFB3.3520.58. PRR16.10621.26. SEC61G.8872.1. 3 PRELP.5675.6.3 AP.3034.1. GNS.3616.3.5 RB1.5024.67. TBCA.12501.10.3 18.5354.11. 3 STIP1.5489.18. C5orf38.6378.2. YO 1.9512.24. 3 3.9128.34. 12400.25.3 MMP2.4160.49. T2A1.9829.91. LCN1.11708.2. EFNA3.14153.8. 12355.223. LA F2.8761.7. LI APOB.2797.56. SFRP1.3221.54.1 1.12471.47.3 LRP1.10699.52. 3 GZMK.9545.156. ISLR2.13124.20. NG 3.10705.14. RT AG PSG4.5649.83.3 IDO1.9759.13. PCDHA4.7157.22. YNAP.10692.48.3 IL5RA.13686.2. BNIP3.7045.4. RT CYB5R3.7215.18.3 PBPL.6364.7. V GFRAL.6920.1.3 CD59.11514.196. 3 HP.3054.3. T4.6576.1. K.5583.62.3 CRP.4337.49. CRISPLD2.5691.2. RFNG.7203.125. RE DLG3.7897.75.3 GNRH2.10708.3.3 ED F. ARSB.3172.28. F. N AL2.9246.1. 3 PMEL.6472.40. IL1RN.5353.89. K7.3388.58. 2 P45.6553.68. 3 CD84.8770.136. NAALAD2.7986.98.3 BDP1.8929.7.3 1.7867.154.3 HS6ST1.5465.32. YO T1.10370.21. 3 ENAH.9757.29. D IP3.14009.65. 3 D BCAM.2816.50. PA RT A PA GRN.4992.49.1 G K3.3387.1. 2 IL15RA.14054.17. CD200.5112.73. XG.8044.90. TM1.9368.64.3 RCL1.4467.49.2 L CFB.4129.72. X5B.7887.57.3 IL25.4137.57. PA 2.12970.35. 3 Y A M3.2998.53. ASPN.6451.64. KLK8.2834.54. AD1.6625.31. GCK.12960.9. BAD.5870.23.2 X2.8397.147. A TFF3.4721.54. AC A P1.13580.2. IP1.12430.78. T20.12975.11.3 EEA1.14043.12.3 RAB26.6997.32.3 C2.6434.18.3 DLL4.3305.6. SEL1L2.9082.25.3 1.13954.9. AMY1A.7918.114. PTK6.3832.51.1 ADH4.8325.37. PVR.8064.125.3 CRH.5614.44. 3 A.5430.66. 3 A PA SPR.9287.6.3 ANP32B.4194.26. FL IL37.2723.9.2 RNMTL1.9584.105. 3 A.5864.10. 3 T1.12632.14. AA.10087.10. PNLIPRP1.6627.25. BMP6.8459.10. CT2.10772.21. GP5.7185.29. SIGLEC9.3007.7. MGA.11587.5. ULBP2.3082.9. CST7.3302.58. GALNT1.7090.17.3 GUCA2B.6223.5. PC.13990.1. 3 WFIKKN2.3235.50.2 TPST1.7928.183.3 ASIP.5676.54. PTS.12014.19.3 H A1.9937.7. G6B.8659.68. C6.4127.75. PRG3.9015.1.3 L PDCD1LG2.3004.67.2 AM.7968.15. 3 GS NID1.3213.65.2 CKB.3800.71. S T JPH1.6940.18. ENG.4908.6. X6C.8903.1.3 HERC5.12934.1.3 X2.13422.66. PA GDF9.3067.67.1 LY F LT C5orf63.13378.80.3 T3.13123.3.3 TEK.3773.15.4 AC FGF10.2441.2. NTNG1.5637.81. P SYNE2.9789.52. YPD3.13107.9. EXTL2.6528.95. G1.9243.10.3 SKIL.14670.1. STK16.3471.49. NEGR1.13109.82.3 GO CP IGSF8.6984.6. ERP44.6064.4. L.5727.35.3 PG STX7.8274.64.3 PDE3A.5254.69. US SOD2.5008.51. 1 BP1.11516.7.3 CCL22.3508.78. A4.14645.253. FCAR.4987.17. 6.10832.24. RANBP3.14037.18.3 2.14045.12.3 CDKN1B.3719.2.2 WFDC12.6037.6.3 FN1.3435.53. 2 T C1QBP.4967.1. 1 NUP210L.9606.4.3 TK.3823.9. LRRC3.6917.49. 2.10562.42.3 L CLSTN3.6291.55. 3 T 20.11117.2. AX1BP3.12498.12. 3 PSAPL1.8814.33. CBFB.10048.7. RETN.3046.31. IL36G.9117.4.3 OS.11248.43. L PCDH10.9018.38. A1 X4I2.7850.1. IL32.9051.13. SS 1.5245.40. TDGF1.5810.25. BA AG UBE2G2.9199.6.3 VA AMBPL1.12401.3.3 OT A1.12557.18. SLAMF6.5128.53.3 MMP13.4925.54. NEO1.8900.28. C5.7927.16. 3 RCN2.10645.72.3 IL3.4717.55. PUF60.10575.31.3 IL11.4493.92. TE1.8386.11. TNFSF15.2968.61. FA C17orf67.9070.1. 3 PLXNB2.9216.100. C1RL.9348.1. ARL3.12571.14. PIN4.12718.43. 3 CTRL.9229.9. 3 LY TGFBR3.3009.3. CD3G.7924.7. 3 HLA.DMA.10639.1.3 CSF2.4697.59. EPHA2.4834.61.2 C4.6468.37. 3 T1.10571.14. CO KIR3DL2.5096.51. PTPRS.6049.64.3 VIMP.11286.78. K4.13719.19.3 C2.7823.22. EVL.11656.110. F SF1.12777.11. .14624.51.3 HBZ.6919.3.3 NP X7A1.8390.25. CHST15.4469.78. 2 HMBS.11530.37.3 NR1H2.9016.12. IP2.7753.21. ETNK1.8328.9.3 1L.12041.33. L4.12864.9. T7.11383.41. LR CXCL9.9188.119. 3 PRKCSH.5687.5. CCL5.5480.49. LIPF SYT5.9099.19.3 SMIM9.8888.33. NK3.12658.72. 3 A B3 GL YBX2.6372.7. DPY30.13943.38.3 QDPR.11257.1. 3 PSMA6.3860.7. SFRP2.7751.121. SLAMF1.7953.20. 3 DKK4.3365.7.2 A NCAM1.4498.62. VCL.8750.46.3 PPIB.4718.5. T2.6909.40. OM.8261.51.3 I PIWIL1.12793.4. TPM1.5033.27.1 GZMM.5704.74. 3 AT A3.12657.2. SLURP1.6401.73.3 NTF3.4145.58. AT N A O6.9894.13. NXPH1.4562.1. TREML4.11139.4.3 FA VCAM1.2967.8. 1 F1.13583.19. ST3GAL6.6947.4.3 RNF8.14663.44. T1.13482.14.3 IFNA8.6214.84. R PB.7181.17. RY AKT2.5360.9. CTF1.2889.37. OBO1.5740.17. 3 CAST VSIG2.8018.43.3 NSG2.13409.9.3 AMTS3.8845.2. AR A3.8076.4. 3 FBLN1.6470.19. 3 SYT11.7089.42.3 PIR.13634.209.3 FUT10.7156.2.3 SPINK5.8028.22.3 ERH.11614.29. OC.3758.68. WBP1L.9532.5.3 DSC2.13126.52. 3 CDH7.7959.34.3 GPC5.4991.12. XIP1.12894.3.3 NRXN1.8971.9.3 CD48.3292.75. O1L.7060.2.3 RBF D UNC5C.5139.32. 3 GPI.4272.46. SEMA6A.7945.10. TA TS AT MFNG.5605.77.3 FGF7.4487.1.1 ABL1.3341.33. TA UCMA.10977.55. AU CA7.6466.7. 3 EFNA4.2614.28.2 1C1.5735.54. HT CXCL5.2979.8.2 PCSK1.13388.57. IL4.2906.55. RSPO3.13094.75.3 RIC3.7125.4. MICB.5102.55.3 M3D.13102.1. KPNA2.2860.19. CAPN2.14684.17.3 SLC4A8.12798.46. TG NDUFV2.7748.11.3 SVEP1.11109.56. OLFML3.8660.5.3 YZL2.6377.54. SVEP1.11178.21. X15B.12422.143. PT SFTPC.5738.25. ENTPD5.4437.56. RBP7.14208.3. XC2.14051.54.3 PTX3.6447.73.3 SLITRK4.7139.14. GGH.9370.69. 3 D AZ1.9893.27.3 6G6C.6256.9. 3 TMEM70.8074.32.3 IL7R.5089.11. DEAF1.6369.82.3 NT5C.12560.9.3 SLG.3052.8.2 PTPN11.3397.7. PLD5.8081.55.3 L TPSB2.3403.1. F. IFIT2.9853.3.3 LRP1.9182.3. CLPB.11860.3.3 ISCU.7201.5. T. P ST8SIA2.7920.30.3 SIRPB1.6247.9. KITLG.9377.25.3 R1AIP2.10553.8. FSTL1.13112.179.3 TFRC.6895.1.3 RPN1.10490.3.3 TA A IL27RA.5132.71. IDE.3197.70. Y T2.12524.18. GSTP1.4911.49.2 IGBP1.12358.6. VPS29.14318.1.3 M12.4420.7. SCG2.8377.87. 4 AN1.5031.10.1 VI1.8255.34. E X3.12717.65.3 C1D.7821.6. APOE.2937.10. NSG1.7802.53. DCP1A.14008.22.3 L1.6530.63. 17.12923.51.3 RK3.4359.87.1 R GPNMB.8289.8. GZMB.4133.54. 2 ICAM2.5486.73. NBR1.7696.3. ASIC4.6951.26.3 MOB1A.12426.19. BP5.4985.11. IL9.5834.18. OSM.2693.20. CLU.4542.24.2 TMIE.7992.3.3 TPT1.3872.2. AC PCBD1.11313.100. PLCG1.4563.61. 2 SPINT1.2828.82. VT T2.5613.75.3 LILRA5.7787.25. XDH.11264.33.3 LT NAB2.13680.3. NDUFB11.7747.47. R T FGF8.2443.10. FRZB.13740.51. NTRK2.4866.59. A CA10.13666.222. 3 V3.9830.109. R IL5.11071.1. AG UBE2I.2877.3. AAH2.8396.42. G0S2.8931.124. TO TSN.12477.42. XL2.6504.65. 3 H TO C TA IL1RL2.2994.71. DGKB.12895.28. T6.10372.18.3 O QS AM23.7049.2. 3 CLMP.10440.26.3 T AR.10085.25. THPO.5947.90. PA MYBPC1.7648.9.3 ASGR1.5452.71. RPIA.12333.87.3 MPP6.13490.1. TP2A3.13510.7. ELL2.11494.4. FGFR4.4988.49. TR CRABP2.11696.7. LY HBG1.10665.30. 3 WISP1.13692.154.3 IL21.7124.18. LA CD79B.6351.55. 3 TF6.11277.23. S100A13.7223.60.3 FCRL6.6617.12. PTHLH.2962.50.2 G4.12844.10. PYGL.11441.11. 3 A YCD.11538.216. PEX14.8300.82.3 IL23R.5088.175.3 BCL2.3412.7. ENTPD6.8932.1. LY LV RT T M IL13.3072.4.2 DNAJB1.3852.19. 2 TXN.8813.160. PHF3.11544.39. 3 LRP8.3323.37.1 IL6.4673.13. NQO2.9754.33.3 SRSF7.12987.12. PLA2G2A.2692.74. CD300A.5630.48. PPBP.2790.54. RS1.6497.10.3 UBXN4.9970.7. NPDC1.10424.31. OA ST ADH1B.9834.62. MT O LILRB3.11334.7.3 CYP3A4.2943.5. TBP.2875.15.2 ITM2C.9523.34. 3 TO DNAJB12.8006.12.3 RPE.12646.2.3 CYR61.6264.9.3 ELL.11459.81.3 G2.5093.47.3 CALR.5264.65. EFHA2.6912.6.3 A33.6419.75. 3 TPST2.8024.64. UBB.6651.74. HK1.13131.5. A1C.8877.22.3 R CGA.TSHB.3521.16. PPM1A.12619.14.3 LEP.8484.24. NXT1.9942.2.3 SEMA3E.5363.51. VV BPC4.12603.87.3 CD96.9735.44. LCN8.5643.2. O EPHA1.3431.54. LRIT2.11716.28. 3 FI CUL9.12991.49. TF3.8858.21.3 DKK1.3535.84. STX2.7738.299. N.3381.24. AS 4A1.8357.43. 3 FGF6.4130.71. DHH.4389.2. 1 RT LRRK2.10990.21.3 TMPRSS11A.7231.37.3 V RHOG.12540.25. 3 PLS1.12737.12. CELF2.7245.2. 3 TLR1.11149.3. TLR4 FRK.10034.16. A ING1.3888.8. CPL2.6079.59. 11B.5762.35. 3 VCAN.9561.21. MLL2.13623.4. PSMB5.12580.7. 3 GMFG.13062.4. T2.7234.12.3 TMEM2.8992.1. PMEP AS1.10361.25.3 TO AMN.4322.28.3 ED IL17RC.5468.67. HRK.8008.28. GPX7.8345.27.3 FCGR2B.3310.62.1 DLK1.6496.60.3 UXS1.8258.22.3 AG EH T MP3.7903.18.3 IL2RG.2634.2. CDC42.9840.2. CLPSL2.7767.1. 3 S.11247.20. FS PDGFB.4149.8. PA NPPB.3723.1. EIF5A.5888.29. 3 CFC1.3294.55.2 PXDNL.11324.3. 3 CFL1.4203.50. 2 LIPK.6413.79.3 PRRG4.9579.59.3 PILRA.8825.4. A RK2.13013.41. UNC5A.7975.97.3 TGIF2.9847.21.3 ALIN.7736.28. IL10.2773.50. CHI3L1.11104.13. PGRMC2.10506.53. CD5.14065.11. BPC3.10447.18.3 NTS.7857.22. CHST1.7803.4. FIBCD1.9378.6. CAM1.8031.11.3 O STX6.10945.11. AB2.3399.31.2 TNS2.11667.29. A G1.5183.53.3 L ABCC6.8935.22.3 VTI1A.7952.2.3 CBX5.4540.11. L TE1.7740.33. MMRN2.11895.21. M134B.8556.5. CCL7.4886.3. OCRL.10011.65.3 GOSR1.7805.52. 3 AM21.7204.1. OX T CHST9.11646.4. GAL.13389.8. ESAM.7841.84. NPF MPL.3473.78.2 NRG2.8060.7. 3 APOE.5312.49. TA EA10.13610.9.3 LR A8.12711.19. IL2RA.3151.6. 1 IL34.4556.10. FUT9.6991.24.3 YES1.2878.66.2 NRXN1.5110.84. PRDX1.3855.56. L2.11375.49.3 SYT3.10452.24. N C ICAM4.6550.4.3 PGRMC1.7863.50. DNAJB9.11214.40. 3 TMED2.10761.5.3 RELB.11464.9. CPVL.8891.7.3 NGFR.8949.3.3 SYT2.9577.26.3 AP1.5606.24. PDE2A.5246.64. THSD1.5621.64. LP LGI3.8003.57.3 AP1.8285.64.3 AT CAMK1.3592.4. FA RRM1.11360.39. TA IFI16.13940.19. 3 PTEN.3831.21.1 ADM.14115.34. BST1.4535.50.2 CP2.9237.54. PLXNC1.4564.2. 2 ITM2A.7765.15. HAMP.3504.58. T2.11175.45. EA3.12576.21. CYTIP.13612.7.3 IL20.4138.25. B3GNT2.7980.72.3 IL4R.3055.54.2 SNX8.6925.26. IL1B.3037.62.1 TK SIT1.11194.6. 3 TN MT POLI.13536.56. MSR1.11207.3. LMNB1.3889.64.2 AMP5.8290.1. GD AH1B2.2642.4.1 NXPH3.6054.6. KLK6.3450.4. CA2.4970.55.1 TTC17.8800.14. SERPINA10.13119.26.3 FUT3.4548.4.2 2B1.5351.52. 3 MDH1.3853.56. H2AFZ.4163.5. UBE2L3.3874.8. TREML2.5736.1.3 APOL1.9506.10. YGD.12366.16.3 AM163A.6260.14.3 HCAR2.13495.48. WISP3.5927.4. MLEC.6285.71.3 PLOD2.6923.1.3 NUDC.8887.21. HS G RT RT CHI3L2.9383.24. T CLN5.8874.53. RAB7B.12409.90. TGFB1.2333.72.1 BGN.3284.75. CBS.10086.39. DES.12030.82. ELK1.10006.25. IL17C.9255.5. A.12678.66. RBL1.12879.5. FL ZNF382.14616.16. FA D CNTF JA SELP.4154.57.2 CTRC.5626.20. EIF4G3.13991.47.3 MBD4.3891.56. CHST11.7779.86. 3 TBXAS1.8098.37. FA R PRPSAP1.9478.69. RY PA PIM1.5359.65.3 NME2.4249.64.2 ADSS.12644.63. PCDHA4.10533.1.3 A.14024.196. IGLL1.6485.59.3 PQLC2L.7939.1. 3 TFPI2.9233.71.3 MXI1.9035.2. 3 TNFRSF21.5404.53. GCA.12594.5. ITIH5.8233.2.3 A SDHAF2.6420.4.3 C PTPN1.3005.5.2 IL24.3321.2. S FGF1.3486.58.2 PDE7A.5178.5. BR.2636.10. 2 TPM2.4472.5. CD86.5337.64. CD22.2891.1.3 O AT THPO.8059.1. CXCL1.2985.35. 1 IL1A.4851.25.1 CD163.5028.59. KLK5.3201.49. IFNW1.7196.21. AU GPNMB.8606.39.3 PRLR.5114.65. GALP.9398.30. 3 ARA.12954.71.3 PA CCL24.4128.27. BIN1.9574.11. C1QTNF5.7810.20. 3 PC EZR.9753.17. PA T1.9859.180.3 GAN.10054.3. EFNB3.7785.1.3 TNR.11302.237. TTC9B.9386.42. SRXN1.14268.4. CDHR3.9463.26. 3 WBP1.10943.36.3 PSME1.5918.5. AC IRF2.12801.33. USE1.8057.78.3 JUN.10356.21. TPPP2.12800.5. LMCD1.9530.6.3 AP2.12343.14. CH2.12756.3. CTSH.8465.52.3 PA TEX29.10557.6. RY RNF122.11160.56. SPOP.13727.44. PTPN9.12633.3.3 MUC1.9176.3. CDH2.3797.1. 1 LT AM9.3795.6. 2 TNS4.9927.96. L FGFR3.3809.1.2 F ARL1.12392.30.3 ICAM5.8245.27. LY DSC3.4981.6. RT ETHE1.3847.56.2 AR PPIE.5238.26. MPG.12438.127.3 UFC1.3405.6. 2 LT TMPO.8265.225. 3 CREBBP.13614.6.3 S O TTL.13973.62.3 SUN3.8852.10. M NoneX.13231.90.3 LR PSD.13055.53.3 RNASE2.8394.56.3 T3.12867.40.3 MTFR1.9095.5.3 T8D1.8955.60.3 TM4.8646.61. STMN4.6267.51. CNTN4.3298.52. GH1.8462.18. HPGD.4995.16. 1 MT1F A TB1.13511.29. R CD47.6653.58. CEP57.9905.8. DUSP6.12341.8. MP4.7732.45.3 PSG7.5651.50.3 J2.14204.55. PDIL RT R CD177.13116.25. PTPRH.11222.62. 3 KEL.7070.25.4 K1.2247.20.11 TXNRD1.13967.14. AMH.4923.79.1 PLG.4150.75. SPHK2.4468.21. SNAP25.13105.7. R AP AC LDHB.3890.8.2 CTXN3.10467.58. 3 PSMD7.3898.5. VRK1.12553.5. RCAN1.13465.5.3 O A SO CDHR3.8222.49.3 PUS1.11201.19.3 COPS2.14029.42.3 ITGA5.6932.42.3 FUT8.8244.16. OR3.12490.92.3 THSD7A.11174.8. NO AU ITG UBC.6647.55. SBSN.5724.58. N SOCS3.11440.58.3 CLEC11A.2966.65. 2 CFH.4159.130. R1AIP1.9039.47. MST1R.2640.3.2 TYK2.5260.80.3 NHLRC3.9087.8. PRDM1.14197.2.3 CADM1.3326.58. DCTN2.5879.51. 3 HA UA GJ PDK1.5227.60.3 LCK.3452.17.2 PTK2B.8918.64. GPLD1.9500.5. BR D FGFR3.13669.6. FG1.11681.8.3 CA LECT1.9928.125. 3 EWSR1.12988.49. K HDGFRP2.4553.65. CPXM1.6255.74.3 CCPG1.8889.5. 3 AM19.8948.13. SLAMF7.5487.7. SI IL31.10455.196. 3 C PDGF PA CRK.4976.57. FA NTF4.4146.58. EDC4.13066.42. 3 CCL1.2770.51. NCR3.3003.29.2 RIC3.11228.37.3 RELL2.6600.70.3 FNDC4.13451.2. 3 BRSK2.9790.28. A L1CAM.4246.40.2 T14.11120.49. FCGR1A.3312.64.1 EID3.8079.39. D RETS ZAP70.4476.22. TBX22.11146.4. CXCL8.3447.64. 2 GD PRH1.10502.15. 3 BO3.5117.14. AIFM1.9522.3.3 V DBNL.4978.54. TNFRSF1B.8368.102. DPP7.8346.9. SPOCK3.9906.21. CO PSMA1.7880.9.3 UBE2N.3905.62.1 CCNH.9848.22. PRPF6.9581.4. MED1.3892.21. B HTN1.10608.9.3 AO CRKL.9877.28. AP4M1.10076.1. T K IL36B.14149.9. DHFR.9823.2. S C1orf185.10667.78. CHP1.12458.79. 3 TK1.4301.58.2 ADSSL1.13998.26. PSG3.6444.15.3 SLC22A16.9969.8. TA PSG8.6238.55.3 TPI1.4309.59. RGS19.12713.365. PCDH9.10558.26. 3 TRH.5659.11.3 J CD44.9283.8.3 LCORL.4304.18.2 IMPDH1.5229.90.3 XCL1.4143.74.2 CBL.12016.60.3 RPS7.3864.5. XDC2.9074.6. P TP5B.4965.27. X8A.10497.242. OSMR.10892.8. CTSC.3178.5.2 BIRC5.3472.40.2 IL1R1.2991.9.2 PPIL2.14314.6. AX4.12398.15.3 SERPINA6.4785.30.3 TLR2.3835.11. SNX4.3903.49.2 STX1B.6966.144.3 NCR2.2734.49.4 RBM39.4284.18.3 ST8SIA6.6930.95.3 SEMA3A.3222.11. DEPP.7178.59. LAMA4.6577.64. GLP1R.13085.18. RFK.13059.33. AT QRFP.6463.59.3 GALNT16.8923.94. PLK1.3394.81.2 NEFH.9900.36. GRP.5897.58. R KIN.13476.16. A YN.3453.87.2 AIF1.2849.49. ULBP3.2747.3.2 LEMD1.8040.9. 3 NRXN3.5111.15. 3 LDHC.9828.86.3 D GE2.6294.11. ULBP1.3081.70. 2 OD1.13653.335.3 CELA3B.6357.83. SERPINA4.3449.58.2 HD PA SSB.13526.5. NCAM2.6507.16. EFNB2.8772.5. JPH3.9089.77. YSD1.6594.64. NMT1.5196.7. CBR3.14091.42. 3 TMEM57.12367.52. APK2.4355.13.1 PA RAB14.14283.12.3 CO DNER.11442.1.3 PA TO NAMP T KIF1C.9899.28. CCL26.9168.31. HGS.13644.30. LGMN.3622.33. LT GGF1.8051.10. GRP.8400.74. RP1.10534.40. A A SNCA.8458.111.3 NFU1.7770.25.3 LRFN2.7200.4.3 IARS.12815.9. CAND1.13937.75. TP63.10040.63. 3 IL17D.4136.40. PA PPM1L.7187.3. GY CD79A.7796.10. 3 FIG4.6948.82.3 ODC1.13689.2. AREG.2970.60. INSL5.10462.14. BNIP3L.7835.2.3 CAM4.10910.6. 3 O TLN2.14082.56. M GNPD RPS4X.9758.17. FBP2.9867.23.3 NET1.14260.112.3 SPINT2.2843.13. 2 REXO2.13590.1. 3 PSG6.6456.17.3 IFI16.12893.159. 3 MUL1.9315.16.3 F PDGFD.9341.1. TB2.10081.17. CLMP.9585.80. 3 MP EPHB3.9220.7. HRSP12.14636.25. VPS24.12508.9.3 ARHGAP1.11955.1. PGM1.9173.21.3 HIF1A.13089.6. OA CENP USB1.11328.9.3 CSMD1.9598.23. SV2A.12880.1. GZMH.3373.5. F INSL4.6410.26. LY LILRB4.6453.70. S DNER.9769.48.3 ILF3.12759.47.3 MFGE8.4455.89.2 EMID1.13021.12. RPS27A.2846.24. IGFL3.6961.14. CTGF GLO1.9883.29. PA DCTPP1.4314.12. TE4.8065.245. N S O CCM2.12347.29. SU IRF9.12439.67. GOLM1.8983.7.3 SLITRK5.4568.17. PPIL1.9884.8. KCNE2.10427.2. DHX38.13645.14. ITM2A.10516.53. 3 M19A2.6430.36. 3 M20B.7198.197. 3 TBCE.11211.7. CD33.3166.92. RPS3.5026.66.1 V LIPN.8097.77. P MOCS3.14229.5. EIF2B1.10080.9. HNF4A.10041.3. SLC26A11.13502.2. 3 IER3.9786.310. RAB27A.9504.19. INSL3.5723.4. 3 FKBP2.9339.204.3 CSK.3363.31.4 INSL6.5754.76. 3 POFUT1.5634.39. 3 FGF4.4123.60. A HSD17B1.4708.3. PSS1.14007.22. UR ZNF334.12763.69. NCF2.10047.12. CLEC5A.8282.15.3 T3LG.14093.10. 3 AG AARS.12340.17. CRLF2.13694.24. TBK1.3400.49.2 SULF2.8305.18. 3 IFNGR1.5825.49.3 ICAM3.2649.77. EIF3G.11454.87. 3 VA PA LRFN1.7910.41. EIF3 ZAP70.3837.6. CH1.12451.62.3 CNA2D3.8885.6. GRAP.12820.1. YW OBP2A.6526.77. DPYSL3.12707.26. 3 PA PVRL4.5734.13.3 CO ICOSLG.5061.27.3 KR LDOC1.8378.3. SUN5.11260.47. 3 NQO1.9837.60.3 DCLK3.7826.1.3 PURA.12537.88.3 APEX1.9849.13. SP ARS.13083.18. STC2.6231.46.3 INSIG1.10663.42.3 BPIF RHOD.12442.4.3 VEGFC.3132.1. CYGB.11546.7. DMKN.8535.102. CHD7.14005.2. HGD.9832.33.3 PL FA CETN2.13078.3. 3 FH SMAP1.11649.3. P TMOD1.12595.11.3 PSG9.9335.28.3 TMEM9.9249.17.3 IFNA14.7180.114.3 AP70.13552.7. LAD1.6407.63.3 CHST6.4429.51. 2 EREG.4956.2. GRB2.5464.52. O1LB.7994.41. GCNT1.7016.12. SERPINE1.2925.9. FGF5.3065.65.1 TA LGALS1.8046.9. FA PDE6D.13491.40.3 CMA1.3423.59.2 CX3CL1.2827.23. CSF2RA.10438.19. SMD3.10563.13.3 LILRB1.5090.49. LA FOLH1.5478.50.2 CYB5A.11287.14. HMG20A.12712.9. PR K K TGFB2.4156.74.1 CHRD.13438.115. STX8.10903.50. ORB.12483.62. NR1D2.12885.42.3 AM171B.8786.6. UGT1A8.8899.75.3 ARSK.8269.327. CO PGK1.5020.50.1 A APLP2.10627.87. NRN1.8806.18.3 DHX9.10527.22. 3 HCK.3374.49. TSSK2.7873.32. DDC.3538.26. 1 HTRA2.3317.33. ABLIM3.13578.98.3 MAN1A2.9077.10.3 PLN.11254.13. AC IFNL1.4396.54. 1 MAPKAPK5.8382.47.3 TRA2B.12373.73. ZNF415.12811.55. LRP1.8601.167. 3 MZF1.14662.6. DN LLIP.13963.7. YNLL1.3881.49. 2 NDUFB4.10677.9.3 MINPP1.5586.66.3 GCH1.11185.145.3 PSG2.6405.74.3 IQCF1.7991.54. RAP2A.9885.41. S ZNF18.11372.2.3 PDE4D.5255.22. GNAI3.12650.43. IL31RA.8273.84.3 D RT KDELC2.8296.117. IGFL4.6353.60. 3 MY ORC6L.12389.4. RND1.13574.50. SCN2B.8353.15.3 P FA P4HA1.11645.9. 3 PR AMIGO1.9979.13. GALNT3.6593.5. FXR1.13076.4.3 RLBP1.12936.38.3 C EMC1.11989.35.3 IFNA7.14129.1. NPTX1.9256.78. MILR1.11173.29. FSTL5.7099.33.3 ER DGCR2.8055.33. IL13RA1.2633.52. CUL4B.13743.56. S A ND UTS2R.13530.5. 3 RS2.13941.82. 3 SUSD3.5752.63.3 ISG15.14151.4. FCAMR.9568.289.3 XRCC1.12535.2. MRPL33.13453.2. M20A.6433.57.3 FO TY RO FSTL4.9350.3. VA CD70.5807.77. AM107A.2760.2. 2 ST EIF4B.14675.20. TI RBBP6.9887.40. APOO.9373.405.3 RFFL.14186.13. PSMA2.4280.47. CT55.9363.11. FGFR2.3808.76. RMI1.13926.1.3 DTX1.11430.49. KLK5.14039.33. NDUFB8.9800.20.3 ASTL.7993.23.3 KLRC3.7795.14. ITGAL.11617.1. SEMG2.6373.54. SERPINC1.3344.60. SOCS7.11657.86.3 NX3.11138.16. CD300C.5066.134. 3 LRRN3.10471.25. 3 C2orf66.5677.15.3 GPD1L.12420.10.3 K IL1RL1.4234.8. TO SNX1.8807.13. SSB.13625.19.3 CDHR5.9962.1. NRG1.9178.30. KIF23.5228.25.2 H1FX.12709.63. NTMT1.12454.105. CCBL2.12682.5. PRRG1.9008.6. 3 ARHGEF10.9061.3. TPO.3873.51. DEFB118.9306.7. DY SKP1.3902.21. UBXN4.9997.12.3 CNDP2.3192.3. 2 ITGB7.11205.10.3 CHIC2.10914.4. 3 DUSP4.10035.6. FLNA.12906.137.3 RT L TWF1.12871.10. TO AD2.13043.157. 3 N CREG1.9357.4. 3 LY PLAL1.12428.2. DIMT1.12694.28. 3 S100A12.5852.6.3 HERC1.12705.9. PSMG3.12729.12.3 TMEM27.7816.23.3 TSR3.7018.10. A PITPNA.9934.29.3 CALB1.9918.23. 3 JA PCDH10.8780.2.3 CSF3R.2719.3. GEB10.11456.2.3 KYNU.4559.64.2 IFIT3.13642.90.3 ULK3.12437.18.3 AMTS15.4533.76.2 RIPK2.8970.9. R DA TP1A1.11993.227.3 DPEP2.8327.26. ZNRF4.9467.24. MG AT NR3C2.12931.16.3 GRB7.11281.6.3 LO HINT1.5900.11. TJP1.12001.7. HEG1.8947.268. FBXL4.11416.23.3 GPR101.11371.1.3 EFNB3.2514.65. OCK2.13654.1. CLK2.11327.56. 3 PPIH.12449.16.3 MZT1.8072.19.3 SF3B4.13449.25.3 SIRPG.9241.40. GFRA3.2505.49. KIAA1161.8068.43. SCMH1.12604.16. TENM3.11107.25. 3 FGF23.3807.1. BMP10.3587.53. C CYBP.12432.23. LILRA5.8766.29. AMTS4.2809.25. 2 B3GA FL C DEFB104A.5763.67.3 SHC1.5272.55. SEMA7A.7019.13.3 DEFB128.6360.7. 3 ML FTH1.FTL.5934.1. FGF16.4393.3. ERAP1.4964.67. KCT2.6365.62. RQCD1.8975.26. CISD2.8094.20. NME1.5909.51. SLITRK3.10565.19. MG EIF4E2.9745.20. NCL.13655.34. A MTHFS.14107.1. N RGS3.12827.37. PDXK.11098.1. CTSO.9264.11.3 P ARL11.12433.8. KCNG4.13525.17. GBP6.7818.101. IFNGR2.9180.6. FL GALNT7.10888.4.3 FA HLA.DQA2.7757.5. CD300LG.9909.4. LRIT3.11534.6. 3 GJD2.11678.105.3 HMGB2.6913.189.3 SLC5A8.13691.10. SYT8.7932.23.3 SHC2.14074.2.3 CLEC4D.7752.31. LDLR.13129.40. 3 CD74.8748.45. PFDN5.4271.75.2 SE R SEMA5A.13132.14. KLRC3.11571.75. TNFSF18.2708.54. USP21.12681.63.3 PTPN2.3401.8.2 SYT17.9110.2. RNF114.9957.9.3 G11A.5726.49. 3 CBD7.13563.259. AM177A1.8039.41.3 BA PA TFIP11.9043.7. B ANGPT2.2602.2. BUB1 A SP SEC61B.7878.2. NENF NUDT16L1.12497.29.3 CVR1B.2806.49.2 IFNL2.4397.26. 2 TXNIP.11682.7.3 RBM28.11927.3. CNTNAP1.14617.7. ENDOU.5656.53. DPEP1.8794.13. MAP2K1.2864.2. ITM2B.8086.49. HIST3H2A.14144.3. HMGB3.12775.6. YBB2.10000.28.3 E PCDH PRB4.12590.67. 3 POMC.2558.51.3 SAP18.12830.4.3 CSTD2.2619.72. CCL19.4922.13. 1 PNKP.13657.2. LEPR.5400.52.3 CS C7orf69.9544.24.3 ERCC1.12585.39.3 CDH15.5410.53.3 CD207.3361.26. ECE1.8767.44.3 GCNT4.10842.7. CCNB1.5347.59. NTRK3.2658.27. 1 S AMTS6.6441.62. ANK2.7624.19.3 CTSG.2431.17.3 POLM.12851.5.3 RO RPS27A.4474.19. SMC3.14324.52.3 DLK1.8380.244. CAM2.9116.28.3 C12orf49.9387.13.3 PLBD1.6315.58.3 A RNF215.7758.217. 3 CANX.8834.58. GS C1QTNF1.6304.8. MYC.10362.35.3 M ST3GAL1.5657.28. PDLIM4.12387.7.3 SYTL4.11563.51.3 SHC4.11692.21. 3 MNX1.9576.58.3 BCL10.8768.4. AM29.13549.15.3 PTPN7.14688.6. L MID2.14249.68. 3 CKAP2.5345.51. RASA1.5481.16. MY HMGCR.5230.99.3 RAD23B.12522.6. ARRB1.12643.4. 3 C2orf40.6362.6. UFM1.3836.51. CXCL9.11593.21. B4GA PA PA NUDT8.7872.5. PSME3.5204.13.3 CYCS.2942.50. PVRIG.6464.40. TDP1.10073.22. MPP7.12732.13. LRIT3.11919.84. ANXA2.4961.17. ASAH1.5748.20. GS GP M172A.5615.62.3 ITFG1.9347.13.3 FA PSPN.2696.87. QS S100A9.5339.49.3 VA LMOD1.12504.26.3 FCRL3.4440.15. PTH2.7257.18. ITM2C.10560.1. GTF2I.13609.11. AKR1A1.4192.10. DNAJB11.7110.2. SIGIRR.8326.63. TMEM87B.13485.20. URKA.3091.70. GALNT2.5764.4. A NR1D1.5236.2. RGS8.11666.72. GHDC.9730.22.3 PDIA5.5593.11.3 TS E SEMA6C.14597.5. P2RX6.7233.73. APBB1.14206.28. PDXP.3897.61. MSLN.3893.64. 1 GAP1.13587.10. MANSC4.9578.263. 3 SMR3A.7838.27.3 PG MYZAP.9055.81. HERC4.7860.9. ST8SIA4.7038.45.3 O USP8.13450.49.3 ALKBH5.8793.13. ANKRD46.7851.30. EL AV SEPT10.12690.33.3 HDLBP.13570.43.3 NE CCL20.2468.62. 3 PTPRR.6361.49. KLHL13.12463.7. 3 GNGT2.10917.40. IL12RB1.2632.5. A DPP10.7890.68.3 MAP3K3.12990.39. KLRC1.5629.58. 3 TSLP.3010.53.2 IL12RB2.3815.14. CALCA.10494.48. IL17RD.3376.49. CBX7.13027.20. 3 ARF3.12578.13. AM24B.8775.61.3 IL17A.9170.24. HSD17B2.8979.1. EIF4A3.4997.19. LT HMGB1.2524.56. ELK3.5707.55. GCNT2.7143.9.3 RFXAP.12726.3. MMP16.5268.49.3 DNAJC4.8016.19. 3 A POMC.4890.10.1 EN KCNA10.13042.7.3 IL1RAP.14048.7.3 AMTS1.3174.2. 1 SI R TUB2.12493.42.3 ASAP2.13518.5.3 BRPF1.11671.19. RH CA5A.8791.151.3 D TNFRSF4.3730.81. RPS3A.5484.63. ZNF23.11565.58. 3 PSG11.7846.44. 3 G MCTS1.12488.9. CLEC2D.7054.87. TP4B.9994.217. 3 BECN1.13032.1. ANXA2.13700.10. RDH16.12881.17.3 INPP5E.11370.20.3 O SCARF2.9925.56. 3 PES1.4267.81.3 SPINT3.7926.13. 3 K PRNP.6545.58. NMES1.6406.3. 3 KLRF1.5098.79. 3 GNPTG.10666.7. 3 NEURL4.13468.5. ARMC5.8785.1.3 UBTD2.12875.28. 3 COL26A1.9539.25. PLA2G5.2449.1. 4 B4GA SYTL1.12892.10.3 NPIPB3.9954.2. PEX5.5915.58.3 TIRAP.9839.148. PMS2.11312.40. 3 LMAN2L.8013.9. 3 U2AF2.13577.25.3 POU2F1.11715.1. LCMT1.4237.70. FL PDK2.12651.21.3 ZPLD1.5590.11. 3 GMEB2.12804.5.3 SPOCK1.5490.53. TNFSF8.3421.54. 2 RRBP1.9492.5. BARD1.13977.28.3 IFNAR1.9183.7. ISOC1.9816.37.3 NGRN.7153.66. DNAJB2.11438.6. 3 TPK1.12018.84. DERL1.13393.46. TMEM59L.9959.60. AG CASP3.3593.72. SSU72.13566.2. PCBP1.11458.30. CCDC90B.7792.58. PA V HMGCS2.13704.5.3 DOK2.14246.50. 3 ASMTL.10089.7.3 CCNC.7817.36.3 PF4V1.5663.18. RGS10.11634.32. COL9A2.9804.11. MCF2L.13934.3. 3 CASP2.4904.7. F MANBA.6382.17. A FBXL4.9951.36. COG8.9543.131. 3 UGP2.13939.14. 3 MMP14.5002.76.1 KLK4.2833.20. CSDE1.12735.39.3 OSTN.7265.32.3 URKB.3346.72. SEMA3B.5667.3. APBB2.12753.6. PSMD1.8824.2. 3 MED4.14021.81.3 TTMP.7963.36. AC PRC1.11591.43. 3 PRDM4.12779.30. ERLIN1.8776.10. NDUFS4.10584.7.3 CXCL14.5730.60.3 LRPPRC.9046.46. N MYZAP.9964.10. TEX29.13500.9. ARP11.12882.7.3 FEN1.12577.100.3 GLIPR1.9265.10.3 KCNE5.8756.41.3 RRM2B.8925.25. AG TNMD.6578.29. YIPF6.9984.12. PLXNA1.9005.16.3 PSMA1.3859.50.2 A CDC42BPB.3629.60.4 SCGB1D2.6508.68. 3 KLRK1.14095.1. C1orf198.8035.6. 3 AMELX.8578.45. TREML1.9329.28. 3 TNFRSF19.5131.15.3 CD109.3290.50. 2 ERMAP.8631.13. ST CREB3L4.11308.8.3 KIAA1324.10637.50.3 PRKCI.3379.29. PGRMC2.8681.93.3 MAP2K3.6151.18.3 ALDH3A1.11480.1. 3 MCEE.9388.18. FCRL5.6103.70. ENTPD1.7999.23. 3 S SMD4.9106.87.3 CCL13.4144.13. MAPK8.3825.18.2 CAPSL.13688.2. SAMHD1.11303.7. 3 CR C1orf226.7989.5. 3 ABHD12.7825.7. ANGPTL7.6371.50. SENP7.12626.6.3 R OPBP1.9947.22.3 DHX58.14012.17. PVRL3.13557.3. 3 HK2.13130.150. 3 ARPP19.4963.19. MF OX TFRC.8795.48. LRMP.10704.91.3 SSRP1.5032.64. VCR1.9073.35.3 6G6D.6469.62.3 ARFIP1.13488.3.3 CASC4.10613.33.3 ASPR P RBM3.12747.89.3 ZNF410.12843.6. M107B.7801.30.3 AG IL15RA.3445.53. WDR1.10723.41. CLEC4E.7077.9. IGSF9.10518.14.3 SA CXADR.11204.80. C17orf89.9030.56.3 F PSMD4.13568.30. MINOS1.7956.11.3 CAMK1D.3418.12.2 CD40LG.3534.14.2 TRAF4.13041.47.3 RNF128.6510.56. S PRKCZ.2645.54. IL3RA.13744.37. SMDT1.9497.3.3 EHD4.11421.10. 3 HOGA1.10024.44.3 KIF16B.11672.17. GNRH1.5627.53. LARGE.7935.26. WSCD2.6274.15.3 CHCHD10.11270.17. ADCK4.9794.17. PPP1R3B.12768.3.3 SERF1A.7842.52.3 STIM1.9271.101.3 NRXN2.8876.51.3 TNFRSF1A.2654.19. NL CCDC167.7797.11. SERPINI2.7117.21. UGT2A1.8907.11.3 SLC27A2.9045.3.3 THRA.12527.50. LILRB5.7015.8. TMEM132A.7871.16.3 VSIG1.8367.142. SP CHST5.7020.13.3 CRELD1.7628.40. D XRCC4.9886.28. DSG1.2976.58. 2 CERS5.13494.6. R GCKR.5223.59. RBBP5.13631.1. RNF43.14120.2. LCTL.10890.135.3 TEAD3.12854.3. PCSK7.4459.68. C8orf33.9613.16.3 DCUN1D3.13553.4. 3 ZFP91.13651.54. TNFSF9.2599.51. DEFB107A.6399.52.3 RNF13.8087.250. NCMAP.14614.41.3 AG LINGO3.10827.67. TN PATA GRAMD1C.8842.16. 3 IFNA6.5714.88. UGT1A6.7891.45.3 RAET1E.7800.85.3 SP AL ST HSP90B1.6393.63. ERCC4.9895.77.3 PTCHD3.13517.3.3 OLA1.12659.13.3 RAD1.12670.15. 3 R MG COL20A1.8804.39. ERP29.13728.19.3 CD226.5062.60. ZNF843.8321.27. 3 NLGN4X.5357.60.3 GL TGFB1I1.8777.5.3 DPYSL5.12683.156.3 E SA SORBS3.12976.49. KMT2C.11402.17. 3 CR RMDN3.9290.8. ZNF175.12716.3. CASS4.12855.16. TBX5.11202.70.3 CHST3.7189.55. 3 PLD3.10948.14. 3 NCR3LG1.7854.38. TBX3.13978.122.3 DGCR14.11356.19. ASNA1.13620.10. 3 UCK2.12515.45. 3 LGALS4.2982.82. TLR4.11101.18.3 EIF4EBP2.4184.43.3 TIGAR.12476.50.3 GPR135.11465.4.3 A TUFT1.5690.49.3 HM RAD51.2871.73. 2 CLEC2B.7786.83.3 SEZ6L2.3867.49.1 LGALS7.9196.8. 3 B ADGRG5.4551.72.3 BRD1.11607.15. 3 PLBD2.6536.54. 3 NRG4.14139.16. MRPL14.8021.59.3 ITSN1.14070.56. MEPE.3209.69. VA ASGR2.9474.22. SLC30A5.9594.30.3 S100A4.14116.129.3 MUC4.10659.49. NAPB.12655.30. ANXA1.4960.72. AXIN2.12925.105. 3 CD274.5060.62. ST6GAL C4A.C4B.4481.34. BIRC3.4973.18.1 GRB14.13628.58. LRP1B.11275.94. MG PDE9A.5201.50.4 SPINK8.7962.11.3 PRRG1.8306.54.3 CSRP2.12968.2. LONP1.6398.12. NI STMN3.8019.73. NRXN3.9799.3. RELL1.13399.33. ST8SIA3.6929.10.3 PODXL.8792.17. SPTLC2.7267.2. PELI2.12702.13.3 DHX8.11601.26. TYMSOS.8263.64.3 AD BTNL9.7950.142.3 TNFSF4.2839.2.1 MMP17.2838.53.1 RPL30.12478.15. PGK2.13936.24. BGLAP.11067.13. DUSP13.6525.17.3 CDHR1.8372.29. 3 NFE2L1.11154.3.3 MAPK13.5006.71.1 PRKCB.5475.10.3 SUSD2.10021.1. FBXO3.14628.72.3 SPSB1.13942.140. AC DDX25.13984.23. KIAA1161.8093.13.3 AM171B.8061.102. 3 MAP3K7 CXCL12.3516.60. SMCO2.7925.18. F NLRP1.11661.11. SLC6A14.13053.6. NDRG4.7934.11. SIGLEC8.7864.3. ENTPD3.4436.1. AM175B.8945.7. 3 CRIM1.8699.43.3 CRISP2.9282.12. MSRB3.7824.88. K CBD6.10075.75.3 RNF150.9774.59. 3 SMR3B.8595.75.3 CLC.11094.104. 3 CD58.10938.13. LGALS2.3033.57. RNF149.9773.15. 3 UBE2J1.6900.30.3 CN CSPG4.8951.162. AXIN2.8429.16.3 C1QL4.7132.55. SAP30.12888.18. GRAP2.5265.12. IL10RB.2631.50. S100A2.11969.5. PMVK.12450.42.3 ADRBK1.3347.9. AM96A.8787.21. 3 PKD2.13745.10. 3 OSM.14063.17.3 DGCR6.9444.70. CD38.11513.92. 3 ANCL.10063.10.3 CPNE1.5346.24.3 APBB2.12761.12. IMPDH2.5250.53.3 TMED4.9319.59. A KIR3DL3.7944.1. CSNK2A1.5224.20.3 W HNRNPC.11429.80. SETD2.12647.52. SMAD2.10364.6.3 RMDN1.7096.30. ITPKA.13473.55. MTMR1.11167.6. TXNDC11.8769.30. UPK3B.8286.44. WFDC3.6384.19. C6orf89.10885.36.3 CIRBP.12724.81. FA ZHX3.10036.201.3 N ANP32A.13073.14. ANGPT4.2500.2. CLEC2L.8242.9. OX RNASE10.5602.62.3 TMCO5A.6907.17.3 TO CRB1.12012.33. 3 FBXL5.12846.3. NTRK1.3477.63. 2 A SCARB2.5100.53.3 R ARP16.7881.244. A B3GNT6.7082.2. M209B.7066.199. LRRC74A.8389.8. MAPK1.3115.64. B4GA LT ASH1L.12622.96. TTC17.9550.153.3 LEPR.9566.103.3 A ETS2.12350.86.3 CASC4.8838.10.3 HT SLC6A16.13056.18.3 RXFP1.14135.3.3 MMP16.9719.145.3 PEAR1.8892.14. CDKN2B.9874.28. GPR26.13540.1. TNFSF12.5939.42. ARPC3.13573.5. RCVRN.14334.3. CAMK2B.3351.1. A AD NLRP4.12794.6. MUTYH.10030.8.3 SERPINF1.9211.19. SIGLEC10.6048.4. GLRX2.12486.8. BCAP29.11570.94. ANKRD27.12445.50. ANGPTL1.9092.33. KCTD5.12473.48.3 PRKCA.2644.11.2 RNF148.9955.40. 3 RAD51L3.12554.10.3 CSDC2.12754.14. PRIM1.13591.31. 3 F NoneX.13230.174. SLC25A18.5280.68.5 AM213A.13423.94.3 PA GPHA2.6395.58. AL ARID3A.3875.62. R F NLRP10.12821.6. UNC93B1.13487.24.3 DEFB135.6411.58. 3 ERBB2.2616.23.18 CLEC2A.10953.14. GORAB.7247.1. IL18R1.14079.14.3 P IGFBPL1.7815.49.3 AHSA1.11633.89. ARRDC3.12352.70.3 RO SCGB2A1.5001.6.2 RGS18.13982.33. AG ST3GAL2.6281.51. EMILIN3.8964.14. 3 KCNE3.8067.21.3 KLK12.3199.54. WFDC5.6969.14. S A YWHAZ.5858.6. PA UBL4A.11490.42.3 C22orf15.7073.69.4 ENTHD2.7947.19.3 TNFSF14.5355.69.3 AX1BP1.12721.4.3 IGDCC3.11952.1.3 MEF2C.8933.84. SNAI2.10014.31. A KPNA4.12698.72. 3 PRDX4.7789.182. ZNF774.12760.34. NCAM1.7746.230. EVPL.14019.73.3 MICALL2.12891.1. 3 MRAP.7895.108.3 M163B.10880.38.3 R1AIP2.10640.9. A4GA SLAMF8.8994.65.3 GSTK1.13474.40. GPLD1.11939.11. PRKG1.13067.5. 3 SA AM189A2.5719.66.3 SERPINA9.7266.4. PA EIF1AD.13545.97.3 WNT7A.4889.82.1 ZNF10.11567.23. MMGT1.7225.51.3 PDGFRL.9713.67.3 ZNF358.10529.19. CUBN.12904.180.3 B RSRP1.9050.170. LY C18orf32.8236.8. 3 VSTM4.7242.14.3 P HNF1A.11193.27. EPHA10.6036.78. MR B3GNT1.8259.25.3 IGDCC3.7118.24.3 COLGAL TEAD4.12516.13.3 SNX17.12845.18.3 LRRC37A2.8897.3. PCDHA7.8758.2.3 CTCF SPTLC1.7886.26. FA FA SARS2.12348.46. MYCT1.13541.1. RAB35.13514.121.3 NISCH.12738.43.3 C6orf226.8078.15.3 KIAA2013.6538.90. DDX58.12382.2.3 VSTM2L.6549.60.3 AM173A.7148.42.3 PLA2G2C.8850.5. 3 BOLA2.8404.102.3 VWC2L.7995.16. CYP3A4.7879.12. GGA3.11683.19. AM174B.9054.8. 3 RNPC3.12814.17. P HRASLS2.7822.11.3 USP25.9215.117. ANAPC7.11690.47.3 B4GA APBB1.12822.34. IFNA5.6210.100. APMAP.10605.22. C10orf35.8220.15.3 AKMIP3.9068.17.3 B3GNT8.9297.12.3 APPL1.12825.18. M6PR.10491.21.3 CNTN5.3299.29. 2 RO POMGNT2.6359.50.3 SLC41A2.12511.83.3 NPLOC4.12993.21.3 MYSM1.11536.9. PA NUDT12.13947.371.3 RNASE13.6424.2. 3 UMODL1.9114.84.3 GGA1.13594.158. PTGFRN.12727.7.3 C D PRSS37.5653.23. AM159B.10504.6. DNAJC18.8033.1. VPS4B.12668.7. LCN10.13007.66. KIAA1467.7892.132.3 A MRAP2.10889.2.3 MAPK14.5007.1. 1 HS CCL3L1.2783.18. CADM4.8619.12. LILRA6.7059.14. CUL3.10045.47. T NE CR HS3ST4.8998.15.3 ITPRIPL1.9221.6. CE LAIR1.11284.24.3 ZNF276.14692.3. NMRAL1.13988.67. GSKIP.12849.25. M19A3.7839.99.3 CCL27.2192.63.10 O ADGRF1.11243.90. AT COL6A5.11155.16. SNX7.14245.195. IGSF11.10700.10.3 MCEMP1.6283.60.3 YTHDC1.8878.48.3 EDF1.12415.122.3 N6AMT1.11096.57.3 AM19A5.5609.92.3 PDZD11.6219.14. GALNT10.7003.4.3 NXF1.12453.161.3 NEURL1.13604.27.3 NR4A1.8089.173.3 RNF219.8822.163. 3 F UBQLN4.12720.71.3 BA LRRC19.7014.27.3 CSF2RB.10512.13. OSCAR.7116.31. ANXA9.13588.11. PTP4A2.11699.16. WFDC10A.13429.3.3 WFDC13.9345.436. 3 PSMA4.14099.20.3 C2orf82.9439.454. 3 LGALS8.4909.68. LY PPP2R1A.12621.55. LINGO1.6620.82. PIK3CG.3391.10. CEP57.9067.152.3 PA TMEM132D.13416.8. 3 SP ST6GAL1.6035.2.3 SMAD4.12022.12. PLA2G2E.2447.7.4 RAD23A.10058.1. CGREF1.6257.56.3 C21orf33.5981.6. 3 C3orf18.6994.19.3 REPIN1.13554.78.3 LRCH4.11252.30. 3 N RNGT RNF148.7742.11. 3 CTNNA2.9872.23.3 BCL2L2.13097.11. C RBM24.11380.84.3 ARHGEF1.13976.9. GDF11.14587.16. FA PLA2G10.2949.6. ZNF264.9993.11. 3 IER3IP1.6412.26.3 DSCAM.9175.48.3 FA NC HECW1.12669.30.3 GUCA1A.10008.43. ZNF75D.11596.47.3 ASH2L.12832.10. CD1D.8749.194. CHEK1.2853.68.2 KCNIP1.13650.11. VPS4A.11476.43. THYN1.12424.107.3 LRRN1.11293.14.3 CASP10.5340.24. 3 POMGNT1.8253.2. 3 GALNT11.8700.325.3 HHLA2.14132.21. CYTH2.12533.135. 3 RIPP SPINK13.10624.45.3 APOLG.8343.224. EPS15L1.4212.5.3 MAMDC2.6119.14.3 TNFRSF10D.14121.24. 3 RAB27B.13596.3. SPG21.11122.97. EXOSC3.12605.1.3 INHBA.INHBB.8467.9.3 TMEM119.11110.4.3 RBM23.11590.5. FCGR3B.3311.27.1 E KIAA1324.9097.5. CRISP3.3187.52. BRICD5.6612.90.3 IFNGR2.8818.13.3 POGLUT1.6467.65.3 F PTH1R.13470.43.3 DEFB112.5689.1.3 F COL8A1.4807.13. UBE2B.9865.40. NR5A2.12444.39. IFNA10.14128.121. S EXOSC1.7930.3. IL27.EBI3.2829.19.2 PIK3C2A.14028.22. F MRPL34.6933.20.3 GPR107.12963.1.3 HDHD2.13472.35. 3 CDC42B DEFB116.11144.10. STK17B.5249.31. F S100A11.14011.17.3 ARF6.12425.104. 3 TO IFNLR1.7192.37.3 SH3BP2.7769.29. MMEL1.3627.71. DEFB110.8340.9. 3 LRRC4C.9369.174. RPRD1A.13515.8. PDCD5.12517.52. 3 PITPNB.12484.67. TSG101.13044.5. EFCAB14.10830.5. DTX3L.11643.73.3 SLC14A1.13430.50.3 NHP2L1.13602.6. UBE2 ANGPT1.2811.27. KIR3DS1.5097.14. ANGPTL4.3796.79. EIF1AX.9850.38. PDRG1.12593.33. UBE2V1.11626.7.3 MEGF10.11168.3. COLEC11.4430.44.3 DNAJC17.14655.1. T GRIA4.10760.107.3 DDX39B.9742.59.3 CA DEFB121.5765.53. 3 APCDD1.6599.5.3 STX10.12842.43.3 PDGFRB.3459.49.2 SSMEM1.8298.8. ZNF566.12795.2. M LR RBM19.11468.15.3 ZNF174.11486.26. KIR2DL2.7773.20. HEPHL1.6354.13.3 A RECQL.11431.235.3 LRRC4B.11911.13. P NRSN1.11654.77. CD300E.10798.4. AMICA1.5094.62.3 SYNCRIP.4224.7.2 DCUN1D5.8760.10.3 IL12B.IL23A.10365.132. PECAM1.2695.25. EEF1B2.5882.34. ELMO1.12764.3. CNTNAP2.6965.19.3 DEFB136.9332.6. F PDCL2.14192.31. NHEJ1.11351.233. 3 UQCRB.12957.62.3 F MAPK3.2855.49. T SCAMP5.13509.5.3 PCDHB1.9941.70. TNFRSF8.2605.49.2 PA SUMO3.14623.26. M S SU LT BCL2L1.4423.77. RU EHBP1.12813.18. DEFB119.8315.5. 3 DUSP16.14631.22. AD DEFB134.10610.8. 3 CSTD2.14034.22. EDDM3B.7782.34.3 RAB39B.12403.30. ARPP21.12860.7. WFIKKN1.3191.50.2 BCAR3.12634.79.3 DUSP15.12838.28. ENTPD1.3182.38. MRPL32.7982.10.3 TMEM190.10442.1.3 PRKAR1B.12479.50. CDC25B.12427.8. AD F POLR1C.12939.1.3 FA SPINK2.13405.61.3 CA SLC35B3.10660.33.3 J PCDHB2.10748.216. ARHGEF2.12848.9. SLC26A5.9127.4. ASCC1.10647.18.3 SERPING1.4479.14. SMIM10.9379.248. DEFB119.13455.10. 3 NRBP1.12616.45.3 NAP1L2.13529.39.3 COMMD7.2823.7. DOCK9.14002.18. PIP4K2A.12697.30. F C17orf78.6571.75.3 F CSF2RB.11137.43.3 SNAPIN.6975.52.3 DEFB115.8391.12.3 SMPDL3A.4771.10. 3 UBE2D4.13475.10. DNAJB14.8053.16. SEPHS1.14083.25.3 P4HA2.11348.132.3 SPRED1.8318.13.3 MA SNAP29.12357.41. ARHGAP36.6289.78.3 P CHCHD2.8015.144. HSD17B14.13972.4.3 SEMA4C.7923.41. FL AD LY GABBR2.9930.48.3 CLINT1.11659.31.3 ZNF134.12787.47. UNC13A.14052.26. F SEMA3G.5628.21. SLC30A3.9081.39.3 CCDC101.12788.6.3 LY TSNARE1.7069.9. DCAF5.11283.13. 3 MPDZ.14036.116. TNFRSF14.5352.11.3 AD ZNF329.12803.9. C MAN2B2.9251.28.3 DNAJC15.7197.2.3 CHST14.7262.191. METTL1.12514.16.3 PLOD3.10612.18. PLA2G16.7865.126. CDKN3.14178.18. DNAJC19.4545.53. S ZNF180.12771.19. C1orf210.8088.56.3 HSP90AB1.5467.15. PSMD5.10716.35. KCNIP3.10513.13. POLH.10022.207. 3 F KREMEN2.3202.28.2 PYDC1.12835.101. DEFB123.9362.11. MRPL52.7123.25.3 APBB3.13589.10. TMEM234.8107.12. OPHN1.12591.27. 3 ABHD14A.5715.4. LPCA SLC6A9.11653.69. HSD17B10.4217.49. 3 AC TGFBR2.5133.17. AKR1C1.12618.50. UPK3BL.6633.43. KLHL12.12695.62. ITGA1.ITGB1.3503.4. SEC22A.13713.164.3 HS3ST3B1.6986.17.3 CD300L C8A.C8B.C8G.2429.27.4 TMPRSS15.3189.61.2 A CLECL1.14308.192.3 CDK2AP1.9450.18.3 C9orf89.7778.104. 3 SCARA3.11292.13.3 PDZK1IP1.8260.13. C10orf54.14123.34. 3 FA TENM4.11365.17. 3 TNFSF13B.3059.50.2 CCDC51.12790.10.3 DNAJC30.7866.11.3 TMEM41B.10661.4.3 RABEPK.13599.15. RHPN2.11439.88. CSNK1D.11289.31. CAMK2A.3350.53. C1orf115.8366.19.3 CE KIAA0040.14603.51. SEC11C.10658.28. P RAB22A.12408.333.3 C11orf68.8307.47.3 CALN1.10933.107.3 PTPRH.11988.24. PPCDC.13996.16. OSBPL11.12878.60. CREB3L1.11198.37. RARRES3.10961.15.3 PCDH15.14228.1. CLEC6A.6911.103. ADCY SCP2D1.14175.78. ZFYVE27.9102.28.3 HS3ST3A1.8268.98.3 STMN2.10900.272. 3 C19orf18.9445.44.3 CE ALDH3B1.12940.35. 3 DEFB125.9486.13. 3 DY COLEC10.6558.5. 3 MMRN2.9723.105. GSTT2B.11273.176.3 RNF34.13386.248. ZNRF3.14122.132.3 KIR2DL5A.8000.17.3 SN IL1RAPL2.5082.51. 3 RPLP2.10949.59.3 HNRN ARHGEF25.13519.112. CD200R1L.8980.19. 3 MADCAM1.11258.41.3 TMEM52.11186.12. YWHAE.14157.21. SLC14A2.12818.159. 3 PCDHGC5.7983.1.3 ZBTB33.12785.49. PDGFRA.10366.11.3 CD200R1.5103.30. C19orf80.7183.102. DEFB113.13374.4. 3 DEFB103A.5679.16.3 KIRREL3.4557.61. CLEC10A.10955.4. 3 XPNPEP1.3481.87. TMEM230.11542.11. B3GA FA MAN1A1.11291.73.3 SP PLA2R1.10916.44. ANAPC10.12345.4. ARFGAP2.11664.32. 3 B4GA GABBR2.13948.50. CSNK2A1.3427.63.2 ARHGDIA.6454.38. TMEM132C.11128.29. CLEC14A.10461.57. LAMT DIRAS3.12406.119. CRABP1.11967.23.3 ARHGAP5.14748.31.3 BET1L.10959.125. CCNB1IP1.9728.4. FA PCOLCE2.6081.52. 3 MAD1L1.13618.15. COMMD1.12509.115. KIAA0319.10013.34.3 ANXA10.13605.16. GNPN KIAA1024.9075.121.3 TMEM167A.11112.18.3 DNAJC27.12799.65.3 DNASE2B.6533.20. PRMT3.12696.166. 3 APDC1A.13548.53.3 SIGLEC14.8248.222.3 MAPKAPK5.3821.28.1 RAPGEF5.9059.14.3 KCNN1.13539.131. MTRF1L.11134.30. RAP1GAP.13735.1.3 HIST1H1C.2987.37.3 C16orf54.12394.53. 3 PLEKHA7.12731.12. 3 TO TMEM132C.7173.141. FA ARHGDIB.9846.32. SIGLEC12.10037.98. 3 ZDHHC14.11677.17.3 HMG20B.11551.16. DEFB106A.5664.57.3 ARHGAP25.11333.82.3 HNRNPM.12783.29. 3 HLA.DPB1.8755.202.3 TMPRSS5.8002.27.3 LOC652493.6561.77.3 SMYD2.12636.113.3 CD300LB.10713.151.3 R TMPRSS11D.6547.83. 3 GAPDHS.8004.15.3 MAPKAPK3.3822.54.2 C1GA GPCPD1.12786.61. SETMAR.12462.20.3 GADD45GIP1.9302.90. PLEKHA1.12459.13. TMEM40.12941.24. KIAA0319L.13698.28.3 CNEP1R1.13532.25. SIGLEC11.8913.22. RNASEH1.8470.213. DEFB105A.10962.46.3 ARHGEF7.13932.45.3 GRAMD1C.8336.267.3 APOBEC3G.13930.3.3 MA SERPINA12.6551.94.3 TA KCNAB2.10015.119. RARRES1.8398.277. TNFRSF12A.5138.50. PPP2R3A.13665.35. MAPKAPK2.3820.68.2 HNRNPDL.10852.114. CLEC12A.11187.11. 3 TNFRSF18.5526.53.3 TMEM167B.13421.17.3 TMEM237.12856.14. KIAA1324L.8363.18.3 TNFRSF6B.5070.76. CDK8.CCNC.3359.11.2 P DEFB108B.5611.56.3 HE TNFRSF10A.4832.75. TMEM8B.12742.160. TNFRSF11A.8316.36. PCDHB4.10522.167. SLCO5A1.11669.39. 3 GABARAPL1.12661.44. CDC2.CCNB1.3422.4.2 CSGALN MRE11A.11319.106. CARHSP1.12808.103. COLGAL CSNK1G2.12653.13. 3 RAP1GDS1.14106.46. RPS6KA6.12688.115. GABARAPL2.12494.99. CSH1.CSH2.13103.125. SUV420H2.12452.32. IGHE.IGK.IGL.4135.84. HNRN CALCOCO2.12534.10. TMPRSS11B.10895.28.3 CDK2.CCNA2.3357.67. P TNFRSF13B.2704.74. CRLF1.CLCF1.2607.54.2 ST6GAL FGA.FGB.FGG.4907.56. ST6GALNA ST6GAL LAMA1.LAMB1.LAMC1.2728.62. 2 ST6GAL IL12A.IL12B.10367.62.3 CXCL3.CXCL2.3148.49. 1 Q.YWHAZ.SFN.4707.50.2 PIK3CA.PIK3R1.3390.72. CDK5.CDK5R1.3358.51.2 CSNK2A1.CSNK2B.5225.50.3 CSNK2A2.CSNK2B.5226.36.3 AKT1.AKT2.AKT3.3392.68. PRKAA1.PRKAB1.PR PRKAA2.PRKAB2.PRK G.YWHAH.YWHA A H

20 40 60 80 20 40 60 80 95 YWHAB.YWHAE.YW Age (years) Age (years) z score 2,925 plasma proteins –0.5 +0.5

d Cluster trajectories Cluster 7 Cluster 6 Cluster 3 Cluster 4 2 2 2 2 n = 12 n = 60 n = 117 n = 107

1 1 1 1

0 0 0 0

–1 –1 Top pathway(s) –1 Top pathway(s) –1 Top pathway(s) Axon guidance/ Protein binding (heparin, Top pathway(s) Protein levels ( z score) Protein levels ( z score) Protein levels ( z score) Extracellular space Protein levels ( z score) EPH–ephrin signaling glycosaminoglycan) NA –2 –2 –2 –2 20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 Age (years) Age (years) Age (years) Age (years)

Cluster 2 Cluster 1 Cluster 5 Cluster 8 2 2 2 2 n = 1,002 n = 1,415 n = 204 n = 8

1 1 1 1

0 0 0 0

–1 Top pathway(s) –1 Top pathway(s) –1 Top pathway(s) –1 Top pathway(s) Cellular localization

Protein levels ( z scor e NA Blood microparticle Receptor activity Protein levels ( z score) Protein levels ( z score) Protein levels ( z score) 2 –2 and binding functions –2 –2 20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 Age (years) Age (years) Age (years) Age (years)

Fig. 2 | Clustering of protein trajectories identifies linear and non-linear changes during aging. a, Protein trajectories during aging. Plasma protein levels were z scored, and trajectories of the 2,925 plasma proteins were estimated by LOESS. b, Trajectories are represented in two dimensions by a heat map, and unsupervised hierarchical clustering was used to group plasma proteins with similar trajectories. c, Hierarchical clustering dendrogram. The eight clusters identified are represented by orange boxes. d, Protein trajectories of the eight identified clusters. Clusters are grouped by the similarity of global trajectories, with the thicker lines representing the average trajectory for each cluster. The number of proteins and the most significant enriched pathways are presented for each cluster. Pathway enrichment was tested using the GO, Reactome and KEGG databases. The top 20 pathways for each cluster are listed in Supplementary Table 12.

Quantification of proteomic changes across the lifespan uncov- (Extended Data Fig. 7b) but were still detectable using different sta- ers waves of aging-related proteins. To quantitatively understand tistical models (e.g., smaller or larger sliding windows) (Extended the proteomic changes occurring throughout life, we developed Data Fig. 7c, Supplementary Fig. 1 and Supplementary Table 13), the software tool differential expression-sliding window analysis indicating the robustness of these age-related waves. (DE-SWAN) (Fig. 3a). This algorithm analyzes protein levels within Intriguingly, the three age-related crests were largely com- a window of 20 years and compares two groups in parcels of 10 years posed of different proteins (Fig. 3d and Supplementary Table 14), (e.g., 35–45 years compared to 45–55 years), while sliding the win- but a few proteins were among the top ten differentially expressed dow in increments of 1 year from young to old. Using DE-SWAN, in each crest, such as GDF15, which was consistent with its pro- we detected changes at particular stages of life and determined nounced increase across the lifespan (Fig. 3a). Other proteins, such the sequential effects of aging on the plasma proteome (while also as chordin-like protein 1 (CHRDL1) or matrix metallopeptidase 12 controlling for the effect of confounding factors). This approach (MMP12), were significantly changed only at the last two crests, identified hundreds of proteins changing in waves throughout reflecting their exponential increase with age. Overlap between pro- aging (Fig. 3b). Summing the number of differentially expressed teins changing at age 34, 60 and 78 years was statistically significant proteins at each age uncovered three crests at ages 34, 60 and 78 (P < 0.05) but limited (Fig. 3e), and most proteins changing in old (Fig. 3c, Extended Data Fig. 7a and Supplementary Table 13). These age were not identified by linear modeling (Fig. 3f). This prompted crests disappeared when the ages of individuals were permutated us to use SEPA to determine whether these waves reflected distinct

1846 Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine NATurE MEDicinE Letters biological processes. Strikingly, we observed a prominent shift CVD-associated proteins were strongly enriched in waves of pro- in multiple biological pathways with age (Fig. 3g). At young age teins defining middle and old age compared to young age (Fig. 4g). (34 years), we observed a downregulation of proteins involved in This enrichment corresponded to an increased incidence of CVD structural pathways, such as the extracellular matrix. These changes after 55 years of age31. Finally, AD- and DS-associated proteins were reversed in middle and old age (60 and 78 years, respectively). overlapped with the top proteins defining middle and old age but At age 60 years, we found a prominent role of hormonal activity, not those defining young age (Fig. 4h,i). The fact that the proteome binding functions and blood pathways. At age 78 years, key pro- defining these two diseases also changed in old individuals of a sep- cesses still included blood pathways but also bone morphogenetic arate disease-free cohort supports the notion of accelerated aging protein signaling, which is involved in numerous cellular func- in DS and AD32,33. Altogether, these results show that waves of pro- tions25. Pathways changing with age by linear modeling overlapped teomic aging are differentially linked to the genomic and proteomic most strongly with the crests at age 34 and 60 years (Fig. 3g), indi- traits of various diseases. cating that dramatic changes occurring in the elderly might be masked in linear modeling by more subtle changes at earlier ages. Discussion Altogether, these results showed that aging is a dynamic, non-linear Our analysis of the plasma proteome reveals complex, non- process characterized by waves of changes in plasma proteins that linear changes over the human lifespan. Although linear analysis reflect complex shifts in biological processes. provides information about the aging plasma proteome, modeling non-linear protein trajectories is necessary to fully appreciate these Proteins linked to age-related diseases are enriched in distinct undulating changes. waves of aging. The plasma proteome is sensitive to the physi- It is well known that men and women age differently21, yet we ological state of an individual but is also genetically influenced26. were surprised that two-thirds of proteins that changed with age To deconvolute complexity between the genome, proteome and also changed with sex (895 of 1,379 proteins). This supports the physiology, we asked whether the top aging-related proteins change National Institutes of Health (NIH) policy on the inclusion of owing to genetic polymorphisms or whether they are among the women in clinical research and using sex as a biological variable top predictors of disease or phenotypic traits. More specifically, in experiments. Nevertheless, a unique proteomic clock can be we sought to determine whether proteins that comprised the three used to predict age in men and women, and deviations from this waves of aging were uniquely linked to the genome or proteome of plasma proteomic clock are correlated with changes in clinical and age-related diseases and traits (Fig. 4a). We used the ranked lists of functional parameters (Fig. 1f–h). This panel of 373 proteins can be the top proteins identified by DE-SWAN at each of the three crests used to assess the relative health of an individual and to measure (Fig. 3c and Supplementary Table 14) and summed the number of healthspan, analogous to epigenetic clocks based on DNA meth- proteins linked to the genome and proteome of specific diseases and ylation patterns34. More large-scale plasma proteomic studies traits separately for each wave (i.e., the cumulative sum) (Fig. 4b–i are required to establish the validity and utility of this clock and and Extended Data Fig. 8). First, we mined the genomic atlas of the whether specific protein subsets are more appropriate to reflect human plasma proteome26 (Fig. 4b) and discovered that the aging particular clinical and functional parameters. proteome is also genetically determined (Fig. 4c and Extended Blood is a sensitive marker of functional aging that also plays Data Fig. 8). However, the rank of proteins determined by trans- an active role in aging. Several studies have shown that soluble association appeared more random with aging (Fig. 4c), suggesting factors from young mouse blood reverse aspects of aging2–10,35. that other sources drive the aging plasma proteome. We then tested Here we describe a 46-protein aging signature that was conserved whether the waves of aging proteins were differentially linked with in humans and mice, containing known aging-related proteins changes in cognitive and physical functions identified in Fig. 1h. such as GDF1536 and IGF1–INSR37 but also less investigated ones Interestingly, the proteome associated with these traits overlapped (Fig. 1l). This conserved signature may allow deeper investigation with the proteome defining middle and old age, when these func- of translational aging interventions in mice, such as heterochronic tions decline the most (Fig. 4d,e). Finally, we used public datasets parabiosis, which partially reverses age-related changes of these and summary statistics from SomaScan proteomic studies focused proteins (Fig. 1k). on age-related diseases, including Alzheimer’s disease (AD)27, Down By deep mining the aging plasma proteome, we identified undu- syndrome (DS)28 and cardiovascular disease (CVD)29. A plasma lating changes during the human lifespan. These changes were proteomic study predicting body mass index (BMI)30 was used as a the result of clusters of proteins moving in distinct patterns, cul- control because weight gain varies widely with age (Supplementary minating in the emergence of three waves of aging. Unexpectedly, Fig. 2). As expected, the proteome linked to BMI was not selectively we found that these clusters were often part of shared biological enriched for proteins defining waves of aging (Fig. 4f). Conversely, pathways, particularly cellular signaling (Fig. 2). In addition,

Fig. 3 | Sliding window analysis distinguishes waves of aging plasma factors. a, DE-SWAN. DE-SWAN compares protein levels between groups of individuals in parcels of 10 years (e.g., 30–40 years compared to 40–50 years). DE-SWAN identifies linear and non-linear changes during aging. Examples are shown of DE-SWAN for three proteins and five age windows. Red and blue rectangles show the two parcels, and the red and blue lines represent the mean within each parcel. DE-SWAN provides statistics for each age window and each plasma protein, allowing detailed analysis of plasma proteomic changes during aging. b, Waves of aging plasma proteins characterized by DE-SWAN (n = 4,263). Within each window, –log10 (P values) and –log10 (q values) were estimated by linear modeling adjusted for age and sex, and significance was tested using the F-test. Local changes attributable to age were signed on the basis of the corresponding beta age. c, Number of plasma proteins differentially expressed during aging. Three local peaks at the ages of 34, 60 and 78 years were identified by DE-SWAN. d, Top ten plasma proteins identified by DE-SWAN at age 34, 60 and 78 years (n = 4,263). Linear models adjusted for age, sex and subcohort were used, and significance was tested using the F-test. Blue and yellow represent local decrease and increase, respectively. # and $ indicate different SOMAmers targeting the same protein. *q < 0.05, **q < 0.01, ***q < 0.001. e, Intersections between waves of aging proteins (n = 4,263; q < 0.05). Linear models adjusted for age, sex and subcohort were used, and significance was tested using the F-test. f, Intersections between linear modeling and the aging waves (n = 4,263; q < 0.05). Linear models adjusted for age, sex and subcohort were used, and significance was tested using the F-test. g, Visualization of pathways significantly enriched for aging-related proteins identified by linear modeling and DE-SWAN at age 34, 60 and 78 years (n = 4,263). Proteins upregulated and downregulated were analyzed separately. The top ten pathways per condition are represented. Enrichment was tested using Fisher’s exact test (GO) and the hypergeometric test (Reactome and KEGG).

Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine 1847 Letters NATurE MEDicinE

we provided biological relevance for the three main waves of aging- life (Fig. 3g). We conclude that linear modeling of aging based on related proteins (Fig. 3), which are characterized by key biologi- omics data does not capture the complexity of biological aging cal pathways with little overlap. In comparison, linear modeling across the lifespan. Thus, DE-SWAN will be invaluable for analyz- failed to identify changes occurring late in the eighth decade of ing longitudinal datasets with linear and non-linear quantifiable

a Differential expression-sliding window analysis (DE-SWAN)

Detailed statistics 1 Protein trajectories Fold Plasma NME1 s 2 change proteins

1 2,925 pasma 20 40 60 80 0 1

–1 Plasma GDF15 P /q values Individual protein trajectorie Protein levels ( z score) –2 proteins 2,925 pasma 20 40 60 80 ARFIP2 Age (years) 20 40 60 80

Plasma Age (years)

Age (years) Age (years) Age (years) Age (years) Age (years)

bc d Waves of aging proteins Top aging proteins Peaks of aging proteins 1,400 * *** *** SVEP1$ ** *** *** SVEP1# 1,200 * *** *** WFDC2 *** *** CHRDL1 ** *** *** CCDC80 1,000 *** *** *** SMOC1 *** *** PTGDS ** *** *** FSTL3 800 *** *** *** RSPO4 ** *** *** SCARF2 * *** *** PTN 600 *** *** MMP12 *** *** *** GDF15 400 NPPB * *** *** SOST *** ** *** ARFIP2 2,925 plasma proteins *** 200 *** *** EPHB6 *** *** SCG3

# of significant proteins ( q < 0.05) SERPINF2 0 * *** *** OMD *** ** MSMP 20 40 60 80 *** *** CHAD 20 40 60 80 95 *** CDON Age (years) Beta age *** *** *** COL11A2 Age (years) Signed –log10 (FDR) *** *** *** *** RET –3 3 –0.03 0.03 Age 34 Age 60 Age 78

e g Overlap between waves of Aging pathways Signed –log10 aging proteins Lm Age 34 Age 60 Age 78 (false-discovery rate) –3 +3 Young age R-HSA-375276-peptide ligand-binding receptors

R-HSA-418594- Gαi signalling events R-HSA-388396-GPCR downstream signaling GO:0005179-hormone activity GO:0008201-heparin binding GO:0005539-glycosaminoglycan binding GO:1901681-sulfur compound binding R-HSA-373076-class A/1 (Rhodopsin-like receptors) R-HSA-500792-GPCR ligand binding hsa04080-neuroactive ligand–receptor interaction Middle age GO:0044420-extracellular matrix component GO:0005604-basement membrane GO:0072562-blood microparticle Old age GO:0071772-response to BMP GO:0030509-BMP signaling pathway GO:0071773-cellular response to BMP stimulus f Overlap between aging waves GO:0030514-negative regulation of BMP signaling pathway and linear modeling GO:0070848-response to growth factor 600 GO:0003723-RNA binding GO:0017076-purine nucleotide binding GO:0019199-transmembrane receptor protein kinase activity 400 GO:0004714-transmembrane receptor protein tyrosine kinase activity GO:0009617-response to bacterium R-HSA-3000178-ECM proteoglycans Overla p 200 GO:0005578-proteinaceous extracellular matrix GO:0031012-extracellular matrix R-HSA-1474244-extracellular matrix organization 0 R-HSA-3000171-non-integrin membrane–ECM interactions Age 60 R-HSA-2022090-assembly of collagen fibrils & other multimeric structures Age 34 Age 78 R-HSA-1474228-degradation of the extracellular matrix Lm adj GO:0005578-proteinaceous extracellular matrix 1,000 500 0 GO:0031012-extracellular matrix # of proteins (q < 0.05)

1848 Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine NATurE MEDicinE Letters

a Relevance of the aging waves b Genome–proteome associations c Trans-associations and aging waves CFH NLRP12 1,200 VTN 1,000

HRG 800 Proteins ranked by 600 P value at 34, 60 and 78 years 400 ABO NS P < 0.05 200 Age 34

Number of trans-associations Age 60 ZFPM2 Age 78 0 Random Link with genome and proteome Gene of disease and traits Protein 0 500 1,000 1,500 2,000 2,500 3,000 Proteins ranked in each wave

d e f Body mass index Cognition and aging waves Handgrip and aging waves and aging waves 12 NS P < 0.05 20 NS P < 0.05 35 Age 34 Age 34 Age 60 Age 60 30 10 Age 78 Age 78 Random 15 Random 25 8 20 6 10 15 4 10 5 NS P < 0.05 with handgrip ( q < 0.05) 2 5 Age 34 Number of proteins associated Number of proteins associated Number of proteins associated Age 60 with body mass index ( q < 0.05) Age 78

with trail making test B time ( q < 0.05) 0 0 0 Random 020406080 100 020406080 100 020406080 100 Proteins rank in each wave Proteins rank in each wave Proteins rank in each wave

g Cardiovascular disease hiAlzheimer’s disease Down syndrome and aging waves and aging waves and aging waves

12 NS P < 0.05 25 Chr. 21 NS P < 0.05 Age 34 Age 34 50 Age 60 Age 60 10 Age 78 Age 78 Random 20 Random 40 8 15 30 6 10 20 4

NS P < 0.05 5 10 Age 34 2 Number of proteins associated Number of proteins associated

Age 60 Number of proteins associated Age 78 with down syndrome ( q < 0.05) Random with Alzheimer’s disease ( q < 0.05) with cardiovascular disease ( q < 0.05) 0 0 0 020406080 100 020406080 100 020406080 100 Proteins rank in each wave Proteins rank in each wave Proteins rank in each wave

Fig. 4 | Waves of aging-related proteins are differentially linked to the genomes and proteomes of diseases and traits. a, Relevance of the aging waves. Schematic representation of analysis. The proteins changing at 34, 60 and 78 years were ranked by P value and were associated with the genome and proteome of diseases and traits. b, Association between the genome and the proteome. The network was created using the protein quantitative trait locus associations identified by Sun et al.26. c, Enrichment for trans-association in the waves of aging-related proteins identified by DE-SWAN. Aging- related proteins at age 34, 60 and 78 years were ranked on the basis of P value, and the cumulative number of trans-associations was enumerated. One- sided permutation tests (1 × 105 permutations) were used to assess significance. d,e, Enrichment for proteins involved in cognitive (d) and physical (e) performance in the waves of aging-related proteins. f–i, Enrichment for disease-associated proteins in the waves of aging proteins, including for BMI (f), CVD (g), AD (h) and DS (i). changes and for integrating non-linear changes in the analysis of with cognitive and physical impairments. These proteins also dis- omics datasets. criminated patients from age-matched controls in AD, DS and CVD Sources of variation of the plasma proteome can be diverse and (Fig. 4g–i), suggesting that the characteristic plasma proteins of under genetic control26,38. Intriguingly, we observed that the rela- aging are amplified in these age-related diseases. Using an AD- and tive importance of trans-associations decreased with aging (Fig. 4c), DS-free aging cohort, we provided evidence of accelerated aging for which led us to investigate sources of variance with a focus on dis- these two diseases32,33. Further investigation of these proteins is war- ease-associated proteomes. Proteins comprising the waves in mid- ranted to determine whether these associations indicate aging bio- dle and old age differentially overlapped with proteins associated markers and/or causal mechanisms of disease. Nonetheless, these

Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine 1849 Letters NATurE MEDicinE results suggest that variance within the aging plasma proteome 13. Aramillo Irizar, P. et al. Transcriptomic alterations during ageing refect the slowly transitions from hard coding factors (i.e., genomic) to soft shif from cancer to degenerative diseases in the elderly. Nat. Commun. 9, 327 (2018). coding factors (e.g., diseases, environmental factors and resulting 14. Castellano, J. M. et al. Human umbilical cord plasma proteins revitalize changes in cognitive and physiological functions). hippocampal function in aged mice. Nature 544, 488–492 (2017). The undulating nature of the aging plasma proteome and its 15. Di Angelantonio, E. et al. Efciency and safety of varying the frequency of interactions with diseases should be considered when develop- whole blood donation (INTERVAL): a randomised trial of 45 000 donors. ing proteomic signatures for diagnostic purposes. Indeed, disease Lancet 390, 2360–2371 (2017). 16. Gubbi, S. et al. Efect of exceptional parental longevity and lifestyle factors on proteomes overlap significantly with the waves of aging proteins prevalence of cardiovascular disease in ofspring. Am. J. Cardiol. 120, (Supplementary Table 15). Accounting for heterogeneous and com- 2170–2175 (2017). plex changes to the plasma proteome during life will likely improve 17. Zhou, J. & Rossi, J. Aptamers as targeted therapeutics: current potential and the sensitivity and specificity of prognostic and diagnostic tests. challenges. Nat. Rev. Drug Discov. 16, 440 (2017). Moreover, these results are pertinent when considering the use of 18. Emilsson, V. et al. Co-regulatory networks of human serum proteins link 39 genetics to disease. Science 361, 769–773 (2018). blood or blood products to treat aging and age-related diseases . 19. Gold, L. et al. Aptamer-based multiplexed proteomic technology for Specifically, identifying plasma proteins that promote or antagonize biomarker discovery. PLoS One 5, e15004 (2010). aging at different stages of life could lead to more targeted therapeu- 20. Austad, S. N. & Fischer, K. E. Sex diferences in lifespan. Cell Metab. 23, tics and/or preventative therapies. Such reliable tests and treatments 1022–1033 (2016). 21. Ostan, R. et al. Gender, aging and longevity in humans: an update of an are urgently needed for several diseases, and, in the future, we hope intriguing/neglected scenario paving the way to a gender-specifc medicine. to describe plasma proteome changes that predict subjects transi- Clin. Sci. 130, 1711–1725 (2016). tioning to disease. Of particular interest are studies of AD, for which 22. Tanaka, T. et al. Plasma proteomic signature of age in healthy humans. Aging blood-based biomarkers are unavailable, and clinical symptoms are Cell 17, e12799 (2018). believed to occur up to two decades after disease onset. 23. Cohen, A. A. Aging across the tree of life: the importance of a comparative perspective for the use of animal models in aging. Biochim. Biophys. Acta. Mol. Basis Dis. 1864, 2680–2689 (2018). Online content 24. Guiraud, S. et al. Identifcation of serum protein biomarkers for utrophin Any methods, additional references, Nature Research reporting based DMD therapy. Sci. Rep. 7, 43697 (2017). summaries, source data, extended data, supplementary informa- 25. Wang, R. N. et al. Bone morphogenetic protein (BMP) signaling in tion, acknowledgements, peer review information; details of author development and human diseases. Genes Dis. 1, 87–105 (2014). 26. Sun, B. B. et al. Genomic atlas of the human plasma proteome. Nature 558, contributions and competing interests; and statements of data and 73–79 (2018). code availability are available at https://doi.org/10.1038/s41591- 27. Sattlecker, M. et al. Alzheimer’s disease biomarker discovery using SOMAscan 019-0673-2. multiplexed protein technology. Alzheimers Dement. 10, 724–734 (2014). 28. Sullivan, K. D. et al. Trisomy 21 causes changes in the circulating proteome Received: 18 January 2019; Accepted: 30 October 2019; indicative of chronic autoinfammation. Sci. Rep. 7, 14818 (2017). 29. Ganz, P. et al. Development and validation of a protein-based risk score for Published online: 5 December 2019 cardiovascular outcomes among patients with stable coronary heart disease. JAMA 315, 2532–2541 (2016). References 30. Carayol, J. et al. Protein quantitative trait locus study in obesity during 1. Harman, D. Te aging process: major risk factor for disease and death. weight-loss identifes a leptin regulator. Nat. Commun. 8, 2084 (2017). Proc. Natl Acad. Sci. USA 88, 5360–5363 (1991). 31. Go, A. S. et al. Heart disease and stroke statistics—2013 update: a report from 2. Baht, G. S. et al. Exposure to a youthful circulation rejuvenates bone repair the American Heart Association. Circulation 127, e6–e245 (2013). through modulation of β-catenin. Nat. Commun. 6, 7131 (2015). 32. Franceschi, C., Garagnani, P., Parini, P., Giuliani, C. & Santoro, A. 3. Conboy, I. M. et al. Rejuvenation of aged progenitor cells by exposure to a Infammaging: a new immune-metabolic viewpoint for age-related diseases. young systemic environment. Nature 433, 760–764 (2005). Nat. Rev. Endocrinol. 14, 576–590 (2018). 4. Huang, Q. et al. A young blood environment decreases aging of senile mice 33. Franceschi, C. et al. Te continuum of aging and age-related diseases: kidneys. J. Gerontol. A Biol. Sci. Med. Sci. 73, 421–428 (2018). common mechanisms but diferent rates. Front. Med. 5, 61 (2018). 5. Katsimpardi, L. et al. Vascular and neurogenic rejuvenation of the aging 34. Horvath, S. & Raj, K. DNA methylation-based biomarkers and the epigenetic mouse brain by young systemic factors. Science 344, 630–634 (2014). clock theory of ageing. Nat. Rev. Genet. 19, 371–384 (2018). 6. Lofredo, F. S. et al. Growth diferentiation factor 11 is a circulating factor 35. Castellano, J. M., Kirby, E. D. & Wyss-Coray, T. Blood-borne revitalization of that reverses age-related cardiac hypertrophy. Cell 153, 828–839 (2013). the aged brain. JAMA Neurol. 72, 1191–1194 (2015). 7. Salpeter, S. J. et al. Systemic regulation of the age-related decline of pancreatic 36. Wiklund, F. E. et al. Macrophage inhibitory cytokine-1 (MIC-1/GDF15): beta-cell replication. Diabetes 62, 2843–2848 (2013). a new marker of all-cause mortality. Aging Cell 9, 1057–1064 (2010). 8. Sinha, M. et al. Restoring systemic GDF11 levels reverses age-related 37. Cohen, E. & Dillin, A. Te insulin paradox: aging, proteotoxicity and dysfunction in mouse skeletal muscle. Science 344, 649–652 (2014). neurodegeneration. Nat. Rev. Neurosci. 9, 759–767 (2008). 9. Villeda, S. A. et al. Te ageing systemic milieu negatively regulates 38. Suhre, K. et al. Connecting genetic risk to disease end points through the neurogenesis and cognitive function. Nature 477, 90–94 (2011). human blood plasma proteome. Nat. Commun. 8, 14357 (2017). 10. Villeda, S. A. et al. Young blood reverses age-related impairments in 39. Sha, S. J., et al. Safety, tolerability, and feasibility of young plasma infusion in cognitive function and synaptic plasticity in mice. Nat. Med. 20, the plasma for Alzheimer symptom amelioration study: a randomized clinical 659–663 (2014). trial. JAMA Neurol. 76, 35–40 (2018). 11. Valdes, A. M., Glass, D. & Spector, T. D. Omics technologies and the study of human ageing. Nat. Rev. Genet. 14, 601–607 (2013). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 12. Stegeman, R. & Weake, V. M. Transcriptional signatures of aging. J. Mol. Biol. published maps and institutional affiliations. 429, 2427–2437 (2017). © The Author(s), under exclusive licence to Springer Nature America, Inc. 2019

1850 Nature Medicine | VOL 25 | December 2019 | 1843–1850 | www.nature.com/naturemedicine NATurE MEDicinE Letters

Methods files provided by SomaLogic were bridged to data from the first batch of samples Plasma proteomics measurements. Te SomaScan platform was used to quantify using calibrators. relative levels of protein involved in several processes, such as intercellular signaling, extracellular proteolysis and metabolism. Tis platform was Normalization of INTERVAL and LonGenity datasets. Relative fluorescence established to identify biomarker signatures of diseases and conditions, including units (RFUs) of each plasma protein were log10transformed. We normalized the cardiovascular risk29, cancer40 and neurodegenerative diseases27. Te SomaScan levels of each protein within each subcohort on the basis of the average of the platform is based on modifed single-stranded DNA aptamers (SOMAmer subjects in the 60- to 70-year range. Supplementary Figure 3 shows representative reagents) binding to specifc protein targets. Assay details were previously normalization examples. Note that this normalization is needed when fitting aging described19. Diferent versions of the SomaScan assay were used in the LonGenity, trajectories (Fig. 2) but does not affect the results when ‘subcohort’ is included as a INTERVAL and four independent human cohorts. Tese versions contained 5,284, covariate in the modeling. 4,034 and 1,305 aptamers, respectively. The data from the four independent cohorts were log10 transformed and Of the 4,034 aptamers measured in the INTERVAL cohort, 3,283 were bridged together using the SomaLogic procedure on the basis of calibrators. contained in the publicly available dataset (European Genoma–Phenome Archive However, the number of samples in the 60- to 70-year range was too small to EGAS00001002555). Our study focused on 2,925 aptamers with identical SeqId reliably bridge these data to the INTERVAL and LonGenity cohorts. and SeqIdVersion in both INTERVAL and LonGenity cohorts (Supplementary Table 1). Of the 2,925 aptamers, 888 were measured in the four independent Linear changes in the aging plasma proteome. To determine the effect of age and cohorts and in mice (Supplementary Table 2). sex at the protein level, we used the following linear model: Protein level α β1 age β2 sex β3 subcohort ε Human cohort characteristics. INTERVAL cohort. Participants in the INTERVAL  þ þ þ þ randomized controlled trial (ISRCTN24760606) were recruited with the active The type II sum of squares was calculated using the ANOVA function of the R collaboration of the National Health Service (NHS) Blood and Transplant (http:// car package43. This sum of squares type tests for each main effect after the other www.nhsbt.nhs.uk), which has supported feld work and other elements of the main effects. q values were estimated using the Benjamini–Hochberg approach44. trial. DNA extraction and genotyping were co-funded by the National Institute It should be noted that the age range differed between cohorts. If the adjustment for Health Research (NIHR), the NIHR BioResource (http://bioresource.nihr. for cohort effect decreases the number of false positives, it could also alter the ac.uk/) and the NIHR Cambridge Biomedical Research Centre at the Cambridge true-positive rate. In the four independent cohorts, the ‘subcohort’ covariate also University Hospitals NHS Foundation Trust. Te INTERVAL study was funded by accounted for batch effect, as samples from different cohorts were measured in NHS Blood and Transplant (11-01-GEN). Te academic coordinating center for different batches (except for PRIN06 and GEHA, which were measured together). INTERVAL was supported by core funding from the NIHR Blood and Transplant To determine the relative proportion of variance explained by age and sex, we Research Unit in Donor Health and Genomics (NIHR BTRU-2014-10024), the calculated the partial Eta2 as follows: UK Medical Research Council (MR/L003120/1), the British Heart Foundation (RG/13/13/30194) and the NIHR Cambridge Biomedical Research Centre at the Sum of squares Partial Eta2 effect Cambridge University Hospitals NHS Foundation Trust. Proteomic assays were ¼ Sum of squareseffect Sum of squareserror funded by the academic coordinating center for INTERVAL and Merck Research ð þ Þ Laboratories (Merck & Co.). A complete list of the investigators and contributors to the INTERVAL trial was previously reported15. Te academic coordinating Validation of the aging-related proteins. To provide confidence in the center would like to thank blood donor center staf and blood donors for reproducibility of the protein assays, we compared our findings with the associations participating in the INTERVAL trial. For more information, see the Nature with age reported by Tanaka et al.22. To this end, we merged our results with those Research Reporting Summary. from Tanaka et al. using ‘SomaId’. Of note, Tanaka et al. used the same version of the Proteomics measurements from 3,301 human plasma samples (1,685 males SomaScan platform that we used for the four independent cohorts (1,305 proteins). and 1,616 females) from two different subcohorts were used for this study. Age ranged from 18 to 76 years with a median age of 45 years (first quartile = 31; SEPA. To determine the biological meaning of a group of plasma proteins, we third quartile = 55). Sample selection, processing and preparation were ranked the top 100 proteins on the basis of the product of –log10(P values) and detailed previously26. beta age (or beta sex) and queried three of the most comprehensive biological annotation and pathway databases: GO45, KEGG46 and Reactome47. Using these LonGenity cohort. LonGenity is an ongoing longitudinal study initiated in 2008 databases, we tested enrichment for pathways in the top 10 to top 100 proteins and designed to identify biological factors that contribute to healthy aging16. The in increments of 1 protein. The 2,925 proteins measured in this study cover 90% LonGenity study enrolls older adults of Ashkenazi Jewish descent with age 65–94 of the human GO, Reactome and KEGG terms containing more than eight genes years at baseline. Approximately 50% of the cohort consists of offspring of parents (Supplementary Figure 4). with exceptional longevity, defined as having at least one parent who survived To analyze each incremental list of proteins, we used the R topGO package48 for to 95 years of age. The other half of the cohort includes offspring of parents with GO analysis and the R clusterProfiler package49 for KEGG and Reactome analyses. usual survival, defined as not having a parental history of exceptional longevity. As input for SEPA, we used gene symbols provided by SomaLogic (Supplementary Proteomics measurements from 1,030 human plasma samples (457 males and 573 Table 1). The 2,925 proteins measured by SomaScan served as the background set females) collected at baseline in LonGenity participants were used for this study. of proteins against which to test for over-representation. Because several individual Age ranged from 61 to 95 years with a median age of 74 years (first quartile = 69; proteins (33 of 2,925) were mapped to multiple gene symbols, we kept only the third quartile = 80). LonGenity participants are thoroughly characterized first gene symbol provided by SomaLogic to prevent false-positive enrichment. demographically and phenotypically at annual visits that include collection For KEGG and Reactome analysis, clusterProfiler requires EntrezID as input. of medical history and physical and neurocognitive assessments. Sixty-eight Therefore, we mapped gene symbols to EntrezID using the org.Hs.eg.db package50. individuals without clinical and functional data were excluded from the analysis. Again, to avoid false-positive enrichment, only the first EntrezID was used when The LonGenity study was approved by the institutional review board (IRB) at the gene symbols were mapped to multiple EntrezID. q values were estimated using Albert Einstein College of Medicine. the Benjamini–Hochberg approach44 for the different databases taken separately. For GO analysis, q values were calculated for the three GO classes (molecular Four additional independent cohorts. One hundred seventy-one human plasma function, cellular component and biological process) independently. To identify samples (84 males and 87 females) were obtained from four different cohorts from the most biologically meaningful terms and pathways, we reported only those the US and Europe (VASeattle, PRIN06, PRIN09 and GEHA). Sample selection, with 20–500 proteins measured by the SomaScan assay. In addition, we focused on processing and preparation of the VASeattle cohort were detailed previously41. pathways that were consistently highly significant (q < 0.05 for at least 20 different Participants from the PRIN06, PRIN09 and GEHA42 cohorts were enrolled by incremental lists of proteins) and kept the top ten pathways per condition (e.g., for multiple Italian study centers. Participants were mainly of European ancestry. Age each wave of aging proteins). Ranking was performed on the basis of the minimum ranged from 21 to 107 years with a median age of 70 years (first quartile = 58; third false-discovery rate across the incremental lists of proteins. SEPA can be viewed as quartile = 89). Written informed consent was obtained from each subject. The an extension of the gene set enrichment analysis approach51, with more control for IRB determined that our research did not meet the definition of human subject true and false positives. research per Stanford’s Human Research Protection Program policy, and no IRB approval was required for this study. Validation of the aging signature in mice. Male and virgin female C57BL/6JN For these cohorts, all samples were stored at −80 °C, and 150 µl aliquots of mice were shipped from the National Institute on Aging colony at Charles River plasma were sent on dry ice to SomaLogic. Plasma samples were analyzed in (housed at 67–73 °F) to the Veterinary Medical Unit (VMU; housed at 68–76 °F) three different batches: 24 samples in 2015, 70 samples in 2016 and 77 samples in at the VA Palo Alto (VA). At both locations, mice were housed on a 12-h light/ 2017. In addition to these 171 plasma samples, 12 additional aliquots from 4 of dark cycle and provided food and water ad libitum. The diet at Charles River was these samples were measured in the different batches to estimate intra- and inter- NIH-31; the diet at the VA VMU was Teklad 2918. Littermates were not recorded assay variability (Supplementary Table 3). Data for 1,305 SOMAmer probes were or tracked. Mice that were 18 months old and younger were housed at the VA obtained. No sample or probe data were excluded. HybNorm.plateScale.medNorm VMU for no longer than 2 weeks before being killed; mice older than 18 months

Nature Medicine | www.nature.com/naturemedicine Letters NATurE MEDicinE

were housed at the VA VMU until they reached the experimental age. After window and compared protein levels of individuals in parcels below and above lk anaesthetization with 2.5% vol/vol Avertin, blood was drawn by cardiac puncture. (i.e., lk 10y; lk vs lk; lk 10y ). To test for differential expression, we used the ½� ½ þ All animal care and procedures were carried out in accordance with institutional followingI linear model: guidelines approved by the VA Palo Alto Committee on Animal Research. Protein level α β1 age β2 sex ε Heterochronic parabiosis was conducted as previously described3,10,52 with 3- and  þ Low=High þ þ 18-month-old mice. Briefly, incisions in the flank were made through the skin and peritoneal cavity of both sets of mice, and adjacent peritoneal cavities were sutured with age binarized according to the parcels. For each lk, q values were estimated together. Adjacent knee and elbow joints were then sutured together to facilitate using the Benjamini–Hochberg correction. The type II sum of squares was 43 coordinated locomotion. Skin was then stapled together using surgical autoclips calculated using the ANOVA function of the R car package . (9 mm; Clay Adams), and mice were placed under heat lamps to recover from To assess the robustness and relevance of DE-SWAN results, we tested multiple anesthesia. Each individual mouse was injected subcutaneously with Baytril antibiotic parcel widths (5, 10, 15 and 20 years). In addition, we used multiple q-value (5 μg g−1) and buprenorphine (0.05–0.1 μg ml−1 in phosphate-buffered saline) for thresholds and compared these results with those obtained by chance. To this end, pain management and 0.9% (wt/vol) NaCl for hydration. Mice were monitored and we randomly permutated the phenotypes of the individuals and applied DE-SWAN administrated drugs and saline over the next week as previously described. to this new dataset. To keep the data structure, age and sex were permuted EDTA-treated plasma was isolated by centrifugation at 1,000 g for 10 min at together. In addition, we analyzed the INTERVAL and LonGenity cohorts 4 °C before aliquotting and storing at −80 °C. A total of 110 plasma samples were separately (Supplementary Fig. 1). Finally, we tested the same linear model when analyzed, and aliquots of 150 µl of plasma were sent on dry ice to SomaLogic. adjusting for subcohort. This led to a loss of statistical power when the age range of Samples were sent in two different batches: 29 samples in 2016 and 81 samples in the INTERVAL and LonGenity cohorts overlapped, but the three waves of aging- 2018. Data for 1,305 SOMAmer probes were obtained, and no sample or probe related proteins remained and the ranks of the top proteins were nearly identical (Supplementary Fig. 1). We used the model adjusted for subcohort when trying data were excluded. RFUs of each plasma protein were log10 transformed. The SomaScan assay was developed and validated for human fluids but has to understand the waves of aging proteins (Figs. 3d–g and 4). The significance been successfully used in mouse research6,24. To understand how similar mouse and levels of the intersections between aging plasma protein signatures identified by human sequences are, we downloaded all homologies between mouse and human linear modeling and DE-SWAN at different ages were determined using the R 57 along with sequence identifiers for each species (HOM_MouseHumanSequence. SuperExactTest package . rpt) from Mouse Genome Informatics (http://www.informatics.jax.org/) as plain text files. Then, the protein reference sequences for both organisms were extracted Relationships between the aging waves and the genome and the proteome of from UniProt (https://www.uniprot.org/). On these matched sequence pairs, for diseases and traits. To quantify the overlap between proteins changing with age each protein we computed a global pairwise sequence alignment. The alignments at different stages of life and the genome and the proteome of diseases and traits, were calculated by using the R ‘Biostrings’ library53. The average identity was 0.85, we ranked DE-SWAN results on the basis of P values and created a k-ranked list of supporting the use of the SomaScan assay with mouse plasma. aging-related proteins, Lk. To reflect the degree to which the genome or proteome To determine the effect of age and sex at the protein level, we used the 81 samples is linked to the waves of aging plasma proteins, we walked down Lk and counted from 1 month to 30 months. To this end, we fitted the following linear model: the number of proteins associated with the genome or specific proteome. When different versions of the SomaScan platform were used, we walked down Lk until Protein level α β1 age β2 sex ε  þ þ þ reaching the top 100 proteins measured in both studies. To identify specific genetic variants associated with the aging plasma proteome, The type II sum of squares was calculated, and q values were estimated using the we mined the summary statistics generated by Sun et al.26., who found 1,927 Benjamini–Hochberg approach. associations with 1,104 plasma proteins. The Qgraph58 R package was used to To characterize the effects of young and old blood on the aging plasma create a network between the genome and the 2,925 proteins analyzed in this study. proteome, normed scaled principal-component analysis was performed using the To determine whether the aging proteome was associated with the proteome R ade4 package54. of clinical and functional variables, we used the individuals from the LonGenity cohort and tested the following linear model for the top variables identified in Prediction of human biological age using the plasma proteome. To determine Supplementary Table 8: whether the plasma proteome could predict biological age, we used glmnet55 and Protein level α β1 age β2 sex β3 variable of interest ε fitted a LASSO model (alpha = 1; 100 lambda tested; ‘lamda.min’ as the shrinkage  þ þ þ þ variable was estimated after tenfold cross-validation). Input variables consisted of The type II sum of squares was calculated using the ANOVA function of the z-scaled log10–transformed RFUs and sex information. Two-thirds (n 2,817) of = 43 the INTERVAL and LonGenity samples were used for training the model, and the R car package . remaining 1,446 samples were used as a validation. In addition, the 171 samples To determine whether the aging proteome was associated with disease from the four independent cohorts were used to further assess the robustness of proteomes, we integrated data and results from previous proteomic studies using the predictive model. the SomaScan platform. We re-analyzed one AD dataset publicly available by 27 To estimate whether a subset of proteins in the the aging clock could provide AddNeuroMed and used summary statistics from published studies focused on 29 28 30 similar predictive results, we used a two-step approach that we described previously56. CVD , DS and BMI . One hundred models (100 lambda) including 0–373 proteins were created in step 1, AddNeuroMed is a European multi-center study in which the AD proteome and we estimated the accuracy of each of these models on the discovery and was quantified in plasma samples from 681 control individuals and individuals validation datasets, separately. Broken-stick regression was used to determine the best with mild cognitive impairment (MCI) and AD using a previous version of the compromise between the number of variables and prediction accuracy. SomaScan assay. Files used were downloaded from the Synapse portal in March 2016 (syn5367752) and included measurements of 1,016 plasma proteins from 931 Associations between delta age and clinical and functional variables in old age. samples. We limited our analysis to the 645 samples available at visit 1 (191 control, We used the individuals from the LonGenity cohort to identify associations 165 MCI and 289 AD). Raw data were log10 transformed. Four samples (two between deviations from the proteomic clock (delta age = predicted age – control and two AD) were considered as outliers on the basis of visual inspection chronological age) and 334 clinical and functional variables (Supplementary of the results of a principal-component analysis and filtered out. Table 8). To this end, we tested the following linear model: To identify plasma proteins associated with AD, we used linear models with diagnosis, age, sex and center as covariates: Variable of interest α β1 age β2 age β3 sex ε  þ þ þ þ Protein level α β1 diagnosis β2 age β3 sex β4 center ε  þ þ þ þ þ For binary outcomes, logistic regression was used. This analysis was separately performed in the discovery (n = 2,817) and validation (n = 1,446) cohorts. Type II The type II sum of squares was calculated using the ANOVA function of the R sum of squares were calculated using the ANOVA function of the R car package43. car package43. To determine whether aging plasma proteins were involved in other disease Clustering of protein trajectories. To estimate protein trajectories during aging, signatures, we identified three studies using the SomaScan platform in large plasma protein levels were z scored, and locally estimated scatterplot smoothing human cohorts providing detailed summary statistics. Carayol et al. mined the (LOESS) regression was fitted for each plasma factor. To group proteins with plasma proteome to obtain new insights into the molecular mechanisms of obesity. similar trajectories, pairwise differences between LOESS estimates were calculated Of 1,129 proteins measured, they identified 192 plasma proteins significantly on the basis of the Euclidian distance, and hierarchical clustering was performed associated with BMI (P < 0.05 after Bonferroni correction). Summary statistics we using the complete method. To understand the biological functions of each cluster, used were obtained from Supplementary Data 1 of their publication30. Ganz et al. we queried the Reactome, KEGG and GO databases, as described above. derived and validated a nine-protein risk score to predict risk of cardiovascular outcomes29. In addition to these 9 proteins, 191 other proteins were significantly DE-SWAN. To identify and quantify linear and non-linear changes of the plasma associated with cardiovascular risk (P < 0.05 after Bonferroni correction). Summary proteome during aging, we developed the DE-SWAN approach. Considering statistics for these 200 proteins are available in their eTable 4, and the 1,130 proteins a vector l of k unique ages, we iteratively used lk as the center of a 20-year measured in this study are listed in their eTable 1. Finally, Sullivan et al. used an

Nature Medicine | www.nature.com/naturemedicine NATurE MEDicinE Letters extended version of the SomaScan platform to study DS and identified a large 54. Dray, S. & Dufour, A. B. Te ade4 package: implementing the duality diagram number of dysregulated proteins28. We used the results for the discovery cohort for ecologists. J. Stat. Sofw 22, 1–20 (2007). (sheet A of Supplementary File 1) in which 258 of 3,586 proteins were reported to 55. Friedman, J., Hastie, T. & Tibshirani, R. Regularization paths for generalized be associated with DS (P < 0.05 after Bonferroni correction). linear models via coordinate descent. J. Stat. Sofw. 33, 1–22 (2010). Because detailed protein information was not available for all studies, we used 56. Lehallier, B. et al. Combined plasma and cerebrospinal fuid signature for the either gene symbols or unitprotID to merge disease proteomes characterized in prediction of midterm progression from mild cognitive impairment to published studies with the aging proteome identified in this study. When multiple Alzheimer disease. JAMA Neurol. 73, 203–212 (2016). P values were reported for the same gene symbol (or a combination of gene 57. Wang, M., Zhao, Y. & Zhang, B. Efcient test and visualization of multi-set symbols), only the most significant P value was retained. intersections. Sci. Rep. 5, 16923 (2015). 58. Epskamp, S., Cramer, A., Waldorp, L., Schmittmann, V. & Borsboom, D. Reporting Summary. Further information on research design is available in the qgraph: network visualizations of relationships in psychometric data. Nature Research Reporting Summary linked to this article. J. Stat. Sofw. 48, 1–18 (2012).

Data availability Acknowledgements We created a searchable web interface to mine the human INTERVAL and We thank the members of the Wyss-Coray laboratory for feedback and support. We LonGenity datasets: https://twc-stanford.shinyapps.io/aging_plasma_proteome/. thank the clinical staff for human blood and plasma collection/coordination. We The independent human cohorts and mouse protein data are available in thank A. Butterworth for his help in getting access to the INTERVAL proteomics data. Supplementary Tables 16 and 17. The INTERVAL data are available through the The AddNeuroMed data are from a public–private partnership supported by EFPIA European Genome–Phenome Archive under accession EGAS00001002555. companies and the European Union Sixth Framework program priority FP6-2004- LIFESCIHEALTH-5. Clinical leads responsible for data collection were I. Kłoszewska Code availability (Lodz), S. Lovestone (London), P. Mecocci (Perugia), H. Soininen (Kuopio), M. Tsolaki An R package for DE-SWAN is available in GitHub: http://lehallib.github.io/DEswan/. (Thessaloniki) and B. Vellas (Toulouse); imaging leads were A. Simmons (London), L.O. Wahlund (Stockholm) and C. Spenger (Zurich); and bioinformatics leads were References R. Dobson (London) and S. Newhouse (London). This work was supported by National Institutes of Health National Institute on Aging (NIA) F32 1F32AG055255 01A1 (D.G.), 40. Mehan, M. R. et al. Protein signature of lung cancer tissues. PLoS One 7, Hungarian Brain Research Program Grant No. 2017-1.2.1-NKP-2017-00002 (T.N.), the e35157 (2012). Fulbright Foreign Student Program (T.N.), the Cure Alzheimer’s Fund (T.W.-C.), 41. Britschgi, M. et al. Modeling of pathological traits in Alzheimer’s disease Nan Fung Life Sciences (T.W.-C.), the NOMIS Foundation (T.W.-C.), the Stanford Brain based on systemic extracellular signaling proteome. Mol. Cell Proteomics 10, Rejuvenation Project (an initiative of the Stanford Wu Tsai Neurosciences Institute), M111 008862 (2011). the Paul F. Glenn Center for Aging Research (T.W.-C.), NIA R01 AG04503 and DP1 42. Franceschi, C. et al. Genetics of healthy aging in Europe: the EU-integrated AG053015 (T.W.-C.) and the NIA-funded Stanford Alzheimer’s Disease Research project GEHA (GEnetics of Healthy Aging). Ann. NY Acad. Sci. 1100, Center P50AG047366, NIA K23AG051148 (S.M.), R01AG061155 (S.M.), the American 21–45 (2007). Federation for Aging Research (S.M.), R01AG044829 (J.V. and N.B.), NIA R01AG057909 43. Fox, J. & Weisberg, S. An R Companion to Applied Regression (N.B.), the Nathan Shock Center of Excellence for the Basic Biology of Aging (SAGE Publications, 2011). P30AG038072 (N.B.) and the Glenn Center for the Biology of Human Aging (N.B.). 44. Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Statis. Soc. B 57, 289–300 (1995). Author contributions 45. Ashburner, M. et al. Gene Ontology: tool for the unifcation of biology. B.L. and T.W.-C. planned the study. D.B., C.F., S.M., J.V., S.S. and N.B. provided human Nat. Genet. 25, 25–29 (2000). plasma samples. N.S., S.E.L. and H.Y. performed the mouse experiments. B.L. analyzed the 46. Kanehisa, M., Furumichi, M., Tanabe, M., Sato, Y. & Morishima, K. KEGG: data, with contributions from T.N. and A.K. P.M.L. developed the searchable web interface new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids (shiny app). B.L., D.G. and T.W.-C. wrote the manuscript. A.K., C.F., S.M., J.V., S.S., N.B. Res. 45, D353–D361 (2017). and T.W.-C. supervised the study. All authors edited and reviewed the manuscript. 47. Crof, D. et al. Te Reactome pathway knowledgebase. Nucleic Acids Res. 42, D472–D477 (2014). 48. Alexa, A. & Rahnenfuhrer, J. topGO: enrichment analysis for Gene Ontology. Competing interests https://doi.org/10.18129/B9.bioc.topGO (2016). The authors declare no competing interests. 49. Yu, G., Wang, L. G., Han, Y. & He, Q. Y. clusterProfler: an R package for comparing biological themes among gene clusters. OMICS 16, 284–287 (2012). 50. Carlson, M. org.Hs.eg.db: genome wide annotation for human. https://doi. Additional information org/10.18129/B9.bioc.org.Hs.eg.db (2017). Extended data is available for this paper at https://doi.org/10.1038/s41591-019-0673-2. 51. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based Supplementary information is available for this paper at https://doi.org/10.1038/ approach for interpreting genome-wide expression profles. Proc. Natl Acad. s41591-019-0673-2. Sci. USA 102, 15545–15550 (2005). Correspondence and requests for materials should be addressed to B.L. or T.W.-C. 52. Castellano, J. M. et al. In vivo assessment of behavioral recovery and circulatory exchange in the peritoneal parabiosis model. Sci. Rep. 6, 29015 (2016). Peer review information Brett Bennedetti and Jennifer Sargent were the primary editors 53. Pagès, H., Aboyoun, P., Gentleman, R. & DebRoy, S. Biostrings: efcient on this article and managed its editorial process and peer review in collaboration with manipulation of biological strings. https://doi.org/10.18129/B9.bioc.Biostrings the rest of the editorial team. (2019). Reprints and permissions information is available at www.nature.com/reprints.

Nature Medicine | www.nature.com/naturemedicine Letters NATurE MEDicinE

Extended Data Fig. 1 | Sample demographics. Age (a, b), cohort (a, b) and sex distributions (c) of the 4,263 subjects from the INTERVAL and LonGenity cohorts. (d) Age and cohort distributions of the 171 subjects from the 4 independent cohorts.

Nature Medicine | www.nature.com/naturemedicine NATurE MEDicinE Letters

Extended Data Fig. 2 | Comparing age and sex effects in independent cohorts. (a) Age and sex effects in the INTERVAL and LonGenity studies (n = 4,263) were compared to age and sex effects in 4 independent cohorts analyzed together (n = 171) and to age effect from Tanaka et al. (n = 240, 2018). The aging plasma proteome was measured with the SomaScan assay in these cohorts and 888 proteins were measured in all studies (b) Scatter plot representing the signed -log10(q value) of the sex effect in the INTERVAL/LonGenity cohorts (x axis, n = 4,263) vs the 4 independent cohorts (y-axis, n = 171). Similar analysis for the age effect in the 4 independent cohorts (c, n = 171) and in Tanaka et al study (d, n = 240).

Nature Medicine | www.nature.com/naturemedicine Letters NATurE MEDicinE

Extended Data Fig. 3 | Deeper investigation of the aging proteomic clock. (a) Prediction of age in the 4 independent cohorts (n = 171) using the proteomic clock. Only 141 proteins out of the 373 constituting the clock were measured in these samples. (b) Prediction of age in the discovery cohort (n = 2,817) using the 373 plasma markers. (c) Feature reduction of the aging model in the Discovery and Validation cohorts to estimate whether a subset of the aging signature can provide similar results to the 373 aging proteins. Dashed lines represent a broken stick model and indicate the best compromise between number of variables and prediction accuracy. (d) Heatmap representing the associations between delta age and 334 clinical and functional variables. For quantitative traits, linear models adjusted for delta age, age and sex were used and significance was tested using F-test. For binary outcomes, binomial generalized linear models adjusted for delta age, age and sex were used and significance was tested using likelihood ratio chi-square test. As in (c) the analysis was performed for the top 2 to top 373 variables predicting age. The non-uniformity in the heatmaps suggests that specific subsets of proteins may best predict certain clinical and functional parameters.

Nature Medicine | www.nature.com/naturemedicine NATurE MEDicinE Letters

Extended Data Fig. 4 | Proteins and proteome undulations in independent human cohorts and in mouse. (a) Trajectories of 5 selected proteins based on the INTERVAL and LonGenity cohorts (n = 4,263, left) and 4 independent human cohorts (n = 171, right). Trajectories were estimated using LOESS regression. Undulation of the 1,305 plasma proteins measured in 4 independent cohorts (b, n = 171) and in mouse (c, n = 81). Plasma proteins levels were z-scored and LOESS regression was fitted for each plasma factor.

Nature Medicine | www.nature.com/naturemedicine Letters NATurE MEDicinE

Extended Data Fig. 5 | Cluster trajectories in independent cohorts. Protein trajectories for the 8 clusters identified in the INTERVAL and LonGenity cohorts (left column). Thicker lines represent the average trajectory for each cluster. Cluster trajectories for the subset of proteins measured in the 4 independent cohorts (middle column). Corresponding cluster trajectories in 4 independent cohorts (right column).

Nature Medicine | www.nature.com/naturemedicine NATurE MEDicinE Letters

Extended Data Fig. 6 | Pathways in clusters. Pathway enrichment was tested using GO, Reactome and KEGG databases (n = 4,263). Enrichment was tested using Fisher’s exact test (GO) and hypergeometric test (Reactome and KEGG). The top 4 pathways for each cluster are shown. Pathway IDs and number of plasma proteins associated are represented in the table.

Nature Medicine | www.nature.com/naturemedicine Letters NATurE MEDicinE

Extended Data Fig. 7 | DE-SWAN age effect for multiple q-values cutoffs, windows size and after phenotypes permutations. Different Q-value cutoffs are represented in (a). Similar analysis with different after phenotype permutations (b) and different windows size in (c). The 3 local peaks identified at age 34, 60 and 78 are indicated by colored vertical lines.

Nature Medicine | www.nature.com/naturemedicine NATurE MEDicinE Letters

Extended Data Fig. 8 | Cis-associations and aging waves. Enrichment for cis-association in the waves of aging proteins identified by DE-SWAN. Aging proteins were ranked based on p-values at age 34, 60 and 78 and the cumulative number of cis-associations was counted. One-sided permutation tests (1e + 5 permutations) were used to assess significance.

Nature Medicine | www.nature.com/naturemedicine nature research | reporting summary

Tony Wyss-Coray ([email protected]) Corresponding author(s): Benoit Lehallier ([email protected])

Last updated by author(s): 10/22/2019 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

Statistics For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section. n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Our web collection on statistics for biologists contains articles on many of the points above. Software and code Policy information about availability of computer code Data collection No software was used for data collection

Data analysis R version 3.6.1 with packages: DEswan (0.0.0.9001), car (3.0-3), topGO (2.36), clusterProfiler(3.12.0) , org.Hs.eg.db (3.8.2) , NHANES (2.1.0), ade4 (1.7-13), SuperExactTest (1.0.7), qgraph (1.6.3). Details and references can be found within text in the relevant Methods sections. For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information. Data Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability October 2018 We created a searchable web interface to mine the human INTERVAL and LonGenity datasets (https://twc-stanford.shinyapps.io/aging_plasma_proteome/) The Human independent cohorts and mouse protein data are available in Supplementary Tables 16 and 17. The INTERVAL data is available through the European Genome-Phenome Archive (https://ega-archive.org/studies/EGAS00001002555).

1 nature research | reporting summary Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design All studies must disclose on these points even when the disclosure is negative. Sample size Methods, "Human cohorts characteristics" and "Validation of the aging signature in mice" subsections. Human cohorts. This is a discovery study, not focusing on specific effect-size. Power calculation are not applicable for our study but we are well-powered (4000+ subjects) and we identified hundreds of highly significant changes after adjustment for multiple comparisons. Sample sizes of the Human cohorts are listed in Extended Data 1) Mouse cohorts. No statistical methods were used to predetermine sample size. Sample size was determined based on the number of animals used in prior experiments conducted in the Wyss-Coray lab (Villeda et al., Nature 2011; Villeda et al., Nature Medicine 2014, Yousef et al., Nature Medicine 2019). Again, we are well powered (110 mice) and we identified hundreds of highly significant changes. Sample sizes of the mouse cohorts are listed in Supplementary Table 9.

Data exclusions Methods, "Human cohorts characteristics" and "Validation of the aging signature in mice" subsections. For the 4 independent cohorts and the mouse data. "Data for 1305 SOMAmer probes were obtained and no sample or probe data were excluded" For the Interval cohort, we used the dataset publicly available without excluding samples / proteins For the LonGenity cohort. "Sixty-eight subjects without clinical and functional data were excluded from the analysis." By excluding these samples, the same subjects (and not different subsets) were used in the whole paper, making the interpretation of the results more straightforward. For the Addneuromed data, no exclusion criteria were pre-established but "we filtered out 4 addneuromed samples based on visual inspection of the results of a Principal Component Analysis (PCA)".

Replication All attempts at replication were successful (See Extended Data 2, 3a, 4).

Randomization Human cohorts. Within each subcohort, the age was balanced and age distribution between cohorts overlapped to assess cohort and batch effect. Sex was balanced. Mouse cohorts. Number of male mice per group was balanced. Female mice were available only at 3m, 12, 18 and 21 months.

Blinding Somalogic measured proteins levels without any information about the samples.

Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. Materials & experimental systems Methods n/a Involved in the study n/a Involved in the study Antibodies ChIP-seq Eukaryotic cell lines Flow cytometry Palaeontology MRI-based neuroimaging Animals and other organisms Human research participants Clinical data

Animals and other organisms October 2018 Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research Laboratory animals Methods, "Validation of the aging signature in mice" subsection. A total of 110 male and virgin female C57BL/6JN mice were used. Mouse groups are summarized in ST9. In the aging cohort, 6 1 months old (mo), 10 3mo, 6 6mo, 6 9mo, 10 12mo, 6 15mo, 10 18mo, 10 21 mo, 5 24 mo, 6 27mo and 6 30mo were used. In the parabiosis cohort, 11 4mo and 18 19mo were used.

Wild animals This study did not involve wild animals

2 This study did not involve samples collected from the field.

Field-collected samples nature research | reporting summary

Ethics oversight Methods, "Validation of the aging signature in mice" subsection. All animal care and procedures were carried out in accordance with institutional guidelines approved by the VA Palo Alto Committee on Animal Research Note that full information on the approval of the study protocol must also be provided in the manuscript.

Human research participants Policy information about studies involving human research participants Population characteristics Participants were healthy blood donors from the Interval, LonGenity, VAseattle, GEHA, PRIN06 and PRIN09 cohorts. For Interval, age ranged from 18 to 76 years with a median age of 45 (1st Quartile=31, 3rd Quartile=55). For LonGenity, age ranged from 61 to 95 years with a median age of 74 (1st Quartile=69, 3rd Quartile=80). For the 4 independent cohorts (combined). Age ranged from 21 to 107 years with a median of 70 years (1st Quartile=58, 3rd Quartile=89).

Recruitment Cohorts characteristics are summarized in Extended data 1 Participants in the INTERVAL randomized controlled trial were recruited with the active collaboration of the National Health Service Blood and Transplant England (www.nhsbt.nhs.uk), which has supported field work and other elements of the trial. As described by Sun et al. (Nature, 558, pages73–79, 2018), 50,000 participants were enrolled in the randomized trial of varying blood donation intervals. People with a history of major diseases (e.g. myocardial infarction, stroke, cancer, HIV, and hepatitis B or C) or with recent illness or infection were excluded . For proteomics measurements, subjects were randomly selected. LonGenity is an ongoing longitudinal study initiated in 2008, designed to identify biological factors that contribute to healthy aging. The LonGenity study enrolls older adults of Ashkenazi Jewish descent, age 65-94 years at baseline. Approximately 50% of the cohort consists of offspring of parents with exceptional longevity (OPEL), defined by having at least one parent that survived to 95 years of age. Subjects were randomly selected within each cohort. Samples from the 4 independent cohort were obtained from cohorts in US (VAseattle) and Europe (GEHA, PRIN06 and PRIN09). GEHA cohort (Franceschi et al. Ann N Y Acad Sci 2007) and VAseattle samples (Britschgi, M., et al. Mol Cell Proteomics 2011) used were described previously. Subjects from the PRIN06 and PRIN09 cohorts were enrolled by multiple Italian study centers. Subjects were randomly selected within each cohort.

Ethics oversight All participants from the Interval cohort gave informed consent before joining the study and the National Research Ethics Service approved this study (11/EE/0538) The LonGenity study was approved by the Institutional Review Board (IRB) at the Albert Einstein College of Medicine. The IRB has determined that our research using VAseattle, GEHA, PRIN06 and PRIN09 cohorts does not meet the definition of human subject research per STANFORD's HRPP policy because 1) we are not obtaining or receiving private individually identifiable information 2) data or specimens were not collected specifically for this study 3) no direct intervention or interaction. For these reasons, this part of the study did not require approval from the IRB Note that full information on the approval of the study protocol must also be provided in the manuscript.

Clinical data Policy information about clinical studies All manuscripts should comply with the ICMJE guidelines for publication of clinical research and a completed CONSORT checklist must be included with all submissions. Clinical trial registration ISRCTN24760606

Study protocol Study protocol is described by Moore et al (Trials. 17;15:363, 2014, https://www.ncbi.nlm.nih.gov/pubmed/25230735)

Data collection According to Moore et al (Trials. 17;15:363, 2014), INTERVAL is a randomised trial of whole blood donors enrolled from all 25 static centres of NHS Blood and Transplant. Recruitment of about 50,000 male and female donors started in June 2012 and was completed in June 2014.

Outcomes According to Moore et al (Trials. 17;15:363, 2014), the primary outcome is the number of blood donations made. Multiple secondary outcome were investigated. The most important are (i) donor quality of life (assessed using the Short Form Health Survey) and (ii) the number of 'deferrals' due to low haemoglobin (and other factors), iron status, cognitive function, physical activity, and donor attitudes. Several papers published the results of this clinical trial 2016 results in: https://www.ncbi.nlm.nih.gov/pubmed/27645285

2017 results in: https://www.ncbi.nlm.nih.gov/pubmed/28941948 October 2018 2019 extension study results in: https://www.ncbi.nlm.nih.gov/pubmed/31383583

3