nature immunology https://doi.org/10.1038/s41590-025-02241-4Resource Multimodal profiling reveals tissue-directed signatures of human immune cells altered with age In the format provided by the authors and unedited Supplementary information https://doi.org/10.1038/s41590-025-02241-4 Classifier node Endpoint subset Internal subset Lineage Myeloid Mono/Mac cMono ncMono Macrophage DC pDC MoMac Lymphocytes TCR T cell B-like γδ T cellαβ T cell Neutrophil MyelocyteMast CD4/8 CD4+ TN/TCM CD4+ EffectorCD4+ TREG CD4+ TN CD4+ TCM CD4+ TEMRA CD4 Mem CD4 N/CM CD4 Res NK/ILCs B cell-like B cell B Mem NK/ILC CD56dim NK ILC1 ILC3 CD56bright NK Atypical B Naive B Memory B GC B PlasmapDC DCs cDC2cDC1 migDC Plasmablasts Plasma cell Plasmablast CD8+ T cell CD8+ TN/TCM CD8+ Effector CD8+ Not MAIT CD8+ Mem CD8+ N/CM CD8+ Res CD8+ MAIT CD8+ TEMRA MAIT pre-NK/ILC Lymphocyte CD4+ T cell CD4+ TRMCD4+ TEM CD8+ TEM CD8+ TRM CD8+ TN CD8+ TCM Supplementary Fig. 1: MMoCHi classifier hierarchy. Summary of the cell types classified during our automated annotation process with Multi-Modal Classifier Hierarchy (MMoCHi)12. Cell subsets are organized hierarchically. MMoCHi classifiers are applied at each decision point (orange nodes), in the hierarchical order in which they appear, to reach an “end point” annotation (indicated by green nodes). TCR 208,483 647C 689C 694B 759B 768B 778C 582C 583B 591C 621B 637C 640C D528D529 D512 D520 D523 D533 D534 D543 D563D570 21 19 19 20 622n = 19 19 20 173 8 17 11 17 11 12 1720n = 8 1 16 3 12 13 179 Non-naive CD8+ T cellsNon-naive CD4+ T cellsCD8+ T cellsCD4+ T cells CD4+ TRM CD8+ TRM CD8+ TEMRACD4+ TEMRA C lo na lit y sc or e C lo na lit y sc or e C lo na lit y sc or e C lo na lit y sc or e CD4+ T N CD4+ T CM CD4+ T REG CD4+ T EM CD4+ T RM CD4+ T EMRA CD8+ T N CD8+ T CM CD8+ T EM CD8+ T RM CD8+ T EMRA ILN ILN LLN LLN MLN MLN BLO BLOSPL SPLBM BM LNG LNGJE J JE J Other JE J Other Other Other BM SPL LNG BM JE J 0.3 0.6 0.2 0.4 0.3 0.2 0.1 0.0 0.1 0.0 0.4 0.2 0.0 0.2 0.1 0.0 0.0 0.0 0.2 0.4 0.6 0.1 0.2 0.3 CD4+ TN 10000 1000 100 10# of T C R s 1 1000 100 10# of T C R s BL O BL O SP L SP L IL N IL N LL N LL N M LN M LN BA L BA L PA R PA R JE L JE L JL P JL P BM BM BL O BL O SP L SP L IL N IL N LL N LL N M LN M LN BA L BA L PA R PA R JE L JE L JL P JL P BM BM BL O BL O SP L SP L IL N IL N LL N LL N M LN M LN BA L BA L PA R PA R JE L JE L JL P JL P BM BM BL O BL O SP L SP L IL N IL N LL N LL N M LN M LN BA L BA L PA R PA R JE L JE L JL P JL P BM BM BL O BL O SP L SP L IL N IL N LL N LL N M LN M LN BA L BA L PA R PA R JE L JE L JL P JL P BM BM BL O SP L IL N LL N M LN BA L PA R JE L JL P BM 1 CD4+ TCM CD8+ TN CD8+ TCM CD4+ TREG CD4+ TRM CD8+ TRM CD8+ TEM CD4+ TEM CD4+ TEMRA CD8+ TEMRA TCR numbers per donor by tissue and cell type a c e b d f Supplementary Fig. 2: T cell clonality is cell type and tissue specific. Summary of single-cell TCR sequences obtained from 208,483 T cells. a) Circle map showing clonal overlap within and across donors. Each dot represents one sample and is colored by the donor of origin. The vast majority of TCR clones are shared within a donor. b) Boxplots displaying the number of TCR clones identified from each MMoCHi-defined subset and tissue, with each dot representing a donor. The dotted horizontal line indicates 100 cells—the minimum cut off for inclusion in subsequent analyses. c-d) Comparison of clonality scores (1−Pielou’s evenness index) across cell types (c) or tissues (d). Low clonality scores correspond to high clonal diversity. Number of samples within each cell-type or tissue displayed above each plot. e-f) Clonality scores compared across tissues as in (d), limited to either only TRM cells (e) or only TEMRA cells (f) to control for compositional shifts across tissues. 0 1 2 3 4 5 CD45RA 0.0 0.2 0.4 0.6 0.8 1.0 D en si ty 0 1 2 3 4 5 6 CD69 0.0 0.2 0.4 0.6 0.8 1.0 D en si ty 0 01 12 23 34 45 56 6 CD103 CD103 0.0 0.5 1.0 1.5 2.0 2.5 3.0 D en si ty 0 01 12 23 34 45 56 6 CD62L CD62L 0.0 0.5 1.0 1.5 2.0 2.5 3.0 D en si ty CD4+ TCM CD4+ TREG CD4+ TRM CD4+ TEM CD4+ TEMRA CD4+ TN MMoCHi CD4+ T cells 0 1 2 3 4 5 CD45RA 0.0 0.2 0.4 0.6 0.8 1.0 D en si ty 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 D en si ty 0.0 0.5 1.0 1.5 2.0 2.5 D en si ty 0 1 2 3 4 5 6 CD69 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 D en si ty TCM/TN helper T cells TREG TEM/Effector helper T cells TH1 TH17 CellTypist CD4+ T cells TFH cells 0 01 12 23 34 45 56 67 7 CD69 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 D en si ty CD103 0.0 0.5 1.0 1.5 2.0 2.5 D en si ty CD62L 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 D en si ty 5 50 01 12 23 34 4 CD45RA 0.0 0.2 0.4 0.6 0.8 1.0 D en si ty CD8+ T N CD8+ T CM CD8+ T RM CD8+ T EM CD8+ T EMRA γδ T 0.0 0.2 0.4 0.6 0.8 1.0 T CM /T N cytotoxic T T RM cytotoxic T γδ T T EM /T RM cytotoxic T T EM /T EMRA cytotoxic T CRTAM + γδ T 0.0 0.2 0.4 0.6 0.8 1.0 CD45RA 0.0 0.2 0.4 0.6 0.8 1.0 D en si ty CD69 0.0 0.2 0.4 0.6 0.8 1.0 D en si ty CD103 0.0 0.5 1.0 1.5 2.0 2.5 D en si ty 00 22 44 66 00 11 22 33 44 55 66 CD62L 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 D en si ty CD8+ TCM CD8+ TRM CD8+ TEM CD8+ TEMRA CD8+ TN MMoCHi CD8+ T cells TCM/TN cytotoxic T cells TRM cytotoxic T cells TEM/TEMRA cytotoxic T cells TEM/TRM cytotoxic T cells CRTAM+ γδ T cells γδ T cells CellTypist CD8+ T cells MMoCHi CellTypist majority voting Pr op or tio n Pr op or tio n No contiguous α- or β-chain sequence detected Multiple α- or β-chains detected (Likely doublet) Only an α- or a β-chain detected (Orphan chain) Single pair of α- and β-chain detected M M oC H i CellTypist majority voting Cells classified as CD4+ by MMoCHi T CM /T N he lpe r T T FH T REG T EM /Effe cto r h elp er T T H 1 T H 17 T RM cy tot ox ic T T EM /T RM cy tot ox ic T T EM /T EMRA cy tot ox ic T CD4+ TN CD4+ TCM CD4+ TREG CD4+ TRM CD4+ TEM CD4+ TEMRA 84963 99.21 369 0.43 271 0.32 22 0.03 0 0.00 0 0.00 8 0.01 7 0.01 1 0.00 66188 55.39 21411 17.92 14812 12.40 16069 13.45 302 0.25 134 0.11 121 0.10 409 0.34 40 0.03 5137 15.22 2310 6.85 24626 72.99 929 2.75 130 0.39 223 0.66 48 0.14 333 0.99 5 0.01 3180 4.31 27879 37.82 4058 5.50 1360 1.84 21261 28.84 10728 14.55 3727 5.06 1484 2.01 40 0.05 3686 11.30 5831 17.88 7580 23.24 7920 24.28 1243 3.81 89 0.27 87 0.27 5023 15.40 1159 3.55 3 0.04 0 0.00 11 0.13 25 0.31 0 0.00 0 0.00 6 0.07 492 6.04 7614 93.41 M M oC H i CellTypist majority voting Cells classified as CD8+ by MMoCHi T CM /T N cy tot ox ic T T RM cy tot ox ic T γδ T T EM /T RM cy tot ox ic T T EM /T EMRA cy tot ox ic T CRTA M + γδ T CD8+ TN CD8+ TCM CD8+ TRM CD8+ TEM CD8+ TEMRA 37394 96.84 150 0.39 2 0.01 660 1.71 290 0.75 119 0.31 822 36.86 52 2.33 3 0.13 1019 45.70 324 14.53 10 0.45 2350 1.90 96297 77.94 11280 9.13 11975 9.69 124 0.10 1528 1.24 838 3.46 65 0.27 4 0.02 13698 56.49 9085 37.47 558 2.30 105 0.33 2124 6.58 40 0.12 9531 29.53 13920 43.13 6558 20.32 Supplementary Fig. 3: Multimodal classification improves discrimination of CD4+ and CD8+ T cell subsets. a) Heatmap displaying concordance of MMoCHi (multimodal classification) and CellTypist (classification on gene-expression alone) annotations for all cells classified as a CD4+T cell by MMoCHi. The upper value shows the number of cells and the lower the percent of CellTypist annotations that are shared with MMoCHi classification. CellTypist has joint labels for some T cell subsets which can be further resolved by MMoCHi by using surface protein expression (boxed in red). Only CellTypist labels detected with greater than 1% frequency are shown, and rows sum to 100%. b-c) Density plots showing expression of CD45RA, CD62L, CD69 and CD103 across MMoCHi defined (b) or CellTypist defined (c) CD4+ T cell subset. d-f) Same as (a-c), but for CD8+T cells. g) Proportion of events within each subset with an αβ TCR detected by V(D)J sequencing. A proportion of MMoCHi- labeled CD8+TRM and CD8+TEMRA cells were annotated as γδ T cells by CellTypist (boxed in green), but reveal high frequencies of detected αβ TCRs. b ca ee fd g αβ TCR clonotype recovery from V(D)J sequencing BCR 133,179 D533 D534 D543 D563 D570 D528 D529 D512 D520 D523 647C 689C 694B 759B 768B 778C 582C 583B 591C 621B 637C 640C 10000 1000 100 10 1 # of B C R s BL O SP L IL N LL N M LN BA L PA R JE L JL P BM BL O SP L IL N LL N M LN BA L PA R JE L JL P BM BL O SP L IL N LL N M LN BA L PA R JE L JL P BM BL O SP L IL N LL N M LN BA L PA R JE L JL P BM Naive B Memory B Plasmablast Plasma cell Naive B Naive B Plasm ablast 19 1021 18n = 8 420 19 3 1110 19 2 52 4 1 910 6n = n = Non-naive B cell lineage C lo na lit y sc or e C lo na lit y sc or e C lo na lit y sc or e 0.04 0.06 0.02 0.02 0.03 0.04 0.01 0.00 Plasma cellPlasmablast 0.01 0.00 0.02 0.00 Memory B Memory B Plasm a Plasm a Plasm ablast Plasm ablast ILN LLN MLN BLOSPL SPL SPLBM BM BM JE JLN LN LNG LNG JE J 0 BLO BM SPLLN LN G JE J 0 20 40 60 80 100 Pe rc en t i so ty pe u sa ge IgM IgD IgG IgA Clonality score 0 IgM IgD IgG IgA IgE IgM IgD IgG IgA IgE IgM IgD IgG IgA IgE IgM IgD IgG IgA IgE 25 25 50 75 0 25 50 75 0 0.00 0.01 0.02 0.03 0.04 0.05 0.00 0.01 0.02 0.03 0.04 0.05 0.00 0.01 0.02 0.03 0.04 0.05 25 50 75 50 75 100 Pe rc en t i so ty pe u sa ge M ea n m ut at io n co un t M ea n m ut at io n co un t M ea n m ut at io n co un t M ea n m ut at io n co un t Naive B Naive B Memory B Memory B BLO BM JEJ LNG LN SPL Plasmablast Plasmablast Plasma cell Plasma 0 25 50 75 BL O SP L BM LN G JE J LN a c f h i j b d e g Supplementary Fig. 4: Somatic hypermutation and clonal expansion in plasma cells and plasmablasts. Summary of single-cell BCR sequences obtained from 133,179 B cells. a) Circle map showing clonal overlap within and across donors. Each dot represents one sample and is colored by the donor of origin. The vast majority of BCR clones are shared within a donor. b) Boxplots displaying the number of BCR clones identified from each MMoCHi-defined subset and tissue, with each dot representing a donor. The dotted horizontal line indicates 100 cells—the minimum cut off for inclusion in subsequent analyses. c-d) Comparison of clonality scores across cell types (c) or tissues (d). Low clonality scores correspond to high clonal diversity. Number of samples within each cell-type or tissue displayed above each plot. e) Clonality scores compared across tissues as in (d), limited to either only plasmablasts or only plasma cells to control for compositional shifts across tissues. f-g) Comparison of isotype usage (as determined from the BCR sequence) in each cell type (f) or tissue (g). The IgE isotype was excluded from tissue-analysis as the frequency was less than 0.5% in each tissue. h-i) Mean number of somatic hypermutations detected within each cell type (h) or tissue (i). j) Visualization of the relationship between clonality score and mean mutation count, score. Manual annotation of CD4+ T cell population CD4+ TN CD4+ TN/TREG CD4+ TCM Cycling CD4+ TCM CD4+ GC TFH CD4+ TRM/TFH-like CD4+ TREG CD4+ TEM GZMB+ TH1 CD4+ TEM GZMK+ TH1 CD4+ TEM TH17 CD4+ TEM GZMB+ CD4+ TEMRA GZMK+ CD4+ TEMRA JEJ CD4+ TRM LNG CD4+ TRM Cycling CD4+ TEM Cycling GZMK+ CD4+ TEMRA Cycling CD4+ TREG KLRD1+ CD4+ TRM a c b d 20 4 Avg. ADT expression 10080604020 Percent expressing gene (%) 0 .5 1 Row-normalized Gene expression GZMK CD3 CD4 CD8 CD45RA CD45RO CD62L CD27 ICOS PD-1 CD25 CD69 CD71 CD57 CD49a CD103 CD101 CD161 KLRG1 GPR56 CD11a CD4 CD8A CCR7 SELL MKI67 STMN1 FOXP3 IKZF2 B3GAT1 PDCD1 ICOS IL21 CXCR5 ITGAE ITGA1 IL2 KLRD1 KLRK1 GZMB GZMH GNLY CTSH RORC IL17A FGFBP2 CX3CR1 TBX21 IFNG CD4+ T N Cyc lin g C D4+ T CM CD4+ T N /T REG CD4+ T CM CD4+ T REG Cyc lin g C D4+ T REG CD4+ G C T FH CD4+ T RM /T FH -lik e JE J C D4+ T RM KLR D1+ C D4+ T RM LN G C D4+ T RM CD4+ T EM Cyc lin g C D4+ T EM GZMK + T H 1 C D4+ T EM GZMB + T H 1 C D4+ T EM T H 17 C D4+ T EM GZMK + C D4+ T EMRA Cyc . G ZMK + C D4+ T EMRA GZMB + C D4+ T EMRA 2 4 Mean protein expression in group CD3 CD4 CD8 TCRαβ TCR Vα7.2 TCR Vδ2 CD45RA CD45RO CD62L CD69 CD103 CD49a KLRG1 CD244 GPR56 CD57 CD161 CD11a CD101 CD146 CD8+ T N CD8+ T CM CD8+ MAIT T C 17 C D8+ MAIT CD14 6+ JE J T C 17 C D8+ T RM JE J C D8+ T RM CD8+ T RM Cyc lin g C D8+ T RM GZMK + C D8+ T EM Cyc lin g C D8+ T EM JE J C D8+ T EMRA G ZMK + C D8+ T EMRA Cyc . G ZMK + C D8+ T EMRA G ZMB + C D8+ T EMRA Cyc . G ZMB + C D8+ T EMRA ITGAD + γδ T JE J γ δ T RM GZMK + γδ T RM GZMK + Vγ9 Vδ2 T GZMB + γδ T EMRA GZMB + Vγ9 Vδ2 T EMRA γδ pr oli fer ati ng ITGB1 TRAV1-2 SLC4A10 KLRB1 ITGA1 CCL3 TRDV2 TRGV9 TRDC TYROBP GZMK GNLY FGFBP2 CX3CR1 GZMB GZMH ITGAD AREG KLRC2 KIR2DL4 CD4 CD8A CD8B CCR7 SELL MAL RORC IL17A MCAM CCL20 CD40LG STMN1 MKI67 20 4 Avg. ADT expression 10080604020 Percent expressing gene (%) 0 .5 1 Row-normalized Gene expression Manual annotation of CD8+ and γδ T cell population CD8+ TN CD8+ TCM TC17 CD8+ MAIT CD146+ JEJ Tc17 CD8+ TRM JEJ γδ TRM JEJ CD8+ TEMRA JEJ CD8+ TRM Cycling γδ T GZMB+ Vγ9Vδ2 TEMRA GZMB+ γδ TEMRA GZMB+ CD8+ TEMRA ITGAD+ γδT GZMK+ Vγ9Vδ2 T CD8+ MAIT GZMK+ CD8+ TEM GZMK+ CD8+ TEMRA GZMK+ γδ TRM CD8+ TRM Cycling CD8+ TRM Cycling CD8+ TEM Cycling GZMK+ CD8+ TEMRACycling GZMB+ CD8+ TEMRA Supplementary Fig. 5: Manual annotation of T cell populations identified by the MMoCHi classifier. a-d) Manual annotation of unsupervised clustering was performed to identify underlying heterogeneity within MMoCHi-classified CD4+T cells (a, b) and CD8+T cells (c, d). Lineage-specific UMAP embeddings are colored by annotated subset. (b, d) Heatmaps of immune subsets showing landmark-registered surface protein expression (top) and dot plots of row-normalized gene expression (log(CP10k+1); bottom) for each subset. Dot size corresponds to the percentage of cells within a population that has any expression of the transcript. Manual annotation of B-lineage cells Pro B Naive B Memory BAtypical B Cycling GC B Cycling B Plasma cell Plasmablast GC B Manual annotation of myeloid cells cMono ncMono IFITM3+ Mac. MARCO+ Mac. Cycling MARCO+ Mac. MMP9+ Mac. SDS+ Mac. Cycling SDS+ Mac. SPIC+ Mac. Cycling SPIC+ Mac. cDC1 cDC2 Migratory DC Cycling DC pDC Manual annotation of NK cells and ILCs CD56dim NK Supplementary Fig. 6: Manual annotation of immune subsets identified by the MMoCHi classifier. a-f) Manual annotation of unsupervised clustering was performed to identify underlying heterogeneity within MMoCHi-classified B cells (a, b), Myeloid cells (c, d), or NK/ILCs (e, f). Lineage-specific UMAP embeddings (a, c, e) are colored by annotated subset. Heatmaps (b, d, f) of immune subsets showing landmark-registered surface protein expression (left) and dot plots of row-normalized gene expression (log(CP10k+1); right) for each subset. Dot size corresponds to the percentage of cells within a population that has any expression of the transcript. Cycling CD56dim NK CD56bright NK Cycling CD56bright NK pre-NK/ILC ILC1 Cycling ILC1 Transitional ILC1-3 ILC3 FC N 1 S 10 0A 9 S 10 0A 12 FC G R 3A C 1Q A LI LR B 2 C C L2 C X C L1 0 M A R C O TR E M 2 S TM N 1 M K I6 7 M M P 9 S E LE N O P C D 5L S D S A D A M D E C 1 H E S 1 S P IC S LC 40 A 1 C X C L1 2 C LE C 9A X C R 1 LG A LS 2 C LE C 10 A C D 1C C D 83 LA M P 3 C C L1 9 C C R 7 LM N B 1 IG FB P 7 LI LR A 4 P LD 4 JC H A IN C D 33 C D 62 L C D 14 C D 16 C D 12 3 C D 1c cMono ncMono IFITM3+ Mac. MARCO+ Mac. Cyc. MARCO+ Mac. MMP9+ Mac. SDS+ Mac. Cyc. SDS+ Mac. SPIC+ Mac. Cyc. SPIC+ Mac. cDC1 cDC2 Migratory DC Cycling DC pDC 0 .5 1 Row-normalized gene expression 0 .5 1 Row-normalized gene expression 10080604020 Percent expressing gene (%) 10080604020 Percent expressing gene (%) 1 2 30 Avg. ADT expression 1 2 30 Avg. ADT expression C D 19 C D 20 C D 38 Ig M Ig D C D 27 C D 11 c C D 71 Pro B Naive B Memory B Atypical B GC B Cycling GC B Cycling B Plasma Plasmablast M M E S O X 4 P C D H 9 Y B X 3 TC L1 A IG H D IG H M FC E R 2 C D 27 S C IM P G P R 18 3 M A R C K S TN FR S F1 3B TN FS F9 TB X 21 FG R FC R L3 TN FR S F1 B IT G A X IL 4R S U G C T LM O 2 R G S 13 A IC D A S TM N 1 M K I6 7 M ZB 1 JC H A IN TN FR S F1 8 D E R L3 P R D X 4 S D C 1 ba dc fe 0 .5 1 Row-normalized gene expression 10080604020 Percent expressing gene (%) K LR F1 E O M E S FC G R 3A G ZM H FG FB P 2 TB X 21 S TM N 1 TY M S M K I6 7 N C A M 1 G ZM K X C L1 X C L2 S E LL G P R 18 3 N C R 2 LD LR A D 4 TN FR S F1 8 K IR 2D L4 IL 7R K IT TN FR S F2 5 R O R C IL 4I 1 1 2 30 Avg. ADT expression G PR 56 C D 16 C D 33 5 C D 56 C D 62 L C D 44 C D 39 C D 49 a C D 10 3 C D 69 C D 12 7 C D 71 CD56dim NK Cycling CD56dim NK CD56bright NK Cycling CD56bright NK pre-NK/ILC ILC1 Cycling ILC1 Transitional ILC1-3 ILC3 HLCA annotation level 1 po pV c on fid en ce s co re 1 2 3 4 5 6 Endothelial Stroma Epithelial Immune Proliferating cells Reference T cell proliferating Migratory DC cDC1 AT2 pDC Alveolar macrophage proliferating Alveolar Macrophage MT+ Interstitial macrophage perivascular Plasma B cells CD4+ T cells ncMono cDC2 CD8+ T cells Mast cells Alveolar macrophage CCL3+ NK cells cMono Mono-derived macrophage Alveolar macrophages M ig ra to ry D C cD C 1 pD C Pl as m ab la st Pl as m a M em or y B N aï ve B C D 4+ T C M C D 4+ T N C D 4+ T EM C D 4+ T R EG C D 4+ T R M C D 8+ M AI T C D 8+ T N nc M on o cD C 2 C D 8+ T R M C D 8+ T EM R A C D 8+ T EM IL C 1 IL C 3 C D 4+ T EM R A γδ T M as t c el l pr e- N K/ IL C N K C D 56 di m N K C D 56 br ig ht cM on o Pr og en ito r Er yt hr oi d M ac ro ph ag e 1.0 0.8 0.6 0.4 0.2 0.0 R ow -n or m al iz ed P ro po rti on ba c H LC A an no ta tio n popV annotation Supplementary Fig. 7: Leveraging our annotation to re-annotate the Human Lung Cell Atlas using popV. a) We trained popV using our data (using the MMoCHi annotations) as the reference data set and the Human Lung Cell Atlas (HLCA)90 as the query dataset. A joint scVI embedding (calculated as part of the popV pipeline) is presented as a UMAP with the reference cells in grey and the query cells colored by their classification into major cell types (as provided by HLCA; note that these original annotations were not used in our analysis). b) UMAP embedding of the joint data colored by popV confidence levels (1: low confidence; 6: high). popV highlights non- immune cells as annotated at low confidence, which is expected as these cells are not present in our dataset. c) Heatmap comparing popV annotations (columns) to the original HLCA annotations in the study (rows) showing a high level of concordance. Specifically, our dendritic cell subset annotations (automated) overlap with the curated HLCA labels, highlighting the quality of label transfer. In other lineages, popV is able to resolve additional subtypes for B cells, NK cells, and T cells. Notably, in labelling T cells (bottom left cluster on the UMAP), popV has relatively low confidence (compared to other immune lineages) since distinguishing between some subtypes (e.g. TCM and TN) is difficult using gene expression alone (as available in the HLCA study). U M AP 2 U M AP 2 UMAP1 UMAP1 ba c M ac ro ph ag e cM on o nc M on o LNG * * * * * * * * * * * * * * * * * * * * * * * * BLO cM on o * * * * * * * * cM on o nc M on o BM * * * * * * * * * * * * * * * * TNFα signaling via NFκB [Hallmark] Inflammatory response [Hallmark] Cytokine-cytokine receptor interaction [KEGG] Complement [Hallmark] Cytokine activity [GO Molecular Function] TLR signaling pathway [KEGG] IFNγ response [Hallmark] IFNα response [Hallmark] Response to IFNα [GO Biological Process] Fatty acid deriv. biosynthetic process [GO Biological Process] MYC targets (v2) [Hallmark] -2 -1 0 1 2 NES (>40/<40) -2 -1 0 1 2 NES (>40/<40) -2 -1 0 1 2 NES (>40/<40) * * * * * ** ** * * * * * * * * * * * * * * * * * * * * * * * * ** * * * * * ** * * ** * * * * * * Apoptosis [Hallmark] IFNγ response [Hallmark] MTORC1 signaling [Hallmark] TCR and costimulatory signaling [WikiPathways] TCR signaling pathway [KEGG] Modulators of TCR signaling and T cell activation [WikiPathways] TCR pathway [Pathway Interaction Database] C D 4+ T C M C D 4+ T EM C D 4+ T R EG C D 4+ T R M C D 8+ T EM C D 8+ T EM R A C D 8+ T R M C D 4+ T R M C D 8+ T R M C D 4+ T C M C D 4+ T EM C D 4+ T R EG C D 4+ T R M C D 8+ T EM C D 8+ T EM R A C D 8+ T R M C D 4+ T C M C D 4+ T EM C D 8+ T EM C D 8+ T EM R A C D 8+ T EM R A C D 8+ T EM R A JEJ LNG BLO BM SPLLN Cytokine-cytokine receptor interaction [KEGG] Inflammatory response [Hallmark] Mitotic spindle [Hallmark] BCR signaling pathway [Signaling Gateway] Immunoglobulin complex [GO Cellular Component] M em or y B N aï ve B LN * * * * * * * * Supplementary Fig. 8: Signatures of immune aging across tissues and subsets. GSEA was used to perform pathway analysis on pseudobulk DE across age (>40 y.o./<40 y.o.). Pathways from various sources (in brackets) shown if significant in one or more subset/tissue combinations. a-c) Heatmaps displaying normalized enrichment score for pathways identified in (a) monocytes and macrophages, (b) T cells, and (c) B cells. * denotes an adj. p-val (FDR) < 0.05. b CD4+ T cells Human Immune Health Atlas (BLO) Age q=0.008 IL18BP N or m al iz ed c ou nt s (lo g 10 ) 0.01 0.03 0.005 <40 >40 a N or m al iz ed c ou nt s (lo g 10 ) Memory B cells Human Immune Health Atlas (BLO) 0.005 0.01 0.03 q=0.043 q=0.042 IL18R1 IL18RAP Age <40 >40 <40 >40 Supplementary Fig. 9: Interrogation of IL-18 Pathway genes in Human Immune Health Atlas. a) Expression of IL-18 pathway genes with sufficient detection in memory B cells, comparing younger (<40 years) and older (>40 years) donors. b) IL18BP expression in CD4+ T cells, comparing younger (<40 years) and older (>40 years) donors. Statistical significance by two-sided Wilcoxon rank sum followed by multiple comparisons correction (FDR). -2 -1 0 1 2 NES (CMV+/CMV-) BL O BM SP L LN LN G JE J CD8+ TN CD8+ TEM CD8+ TRM CD8+ TEMRA * * * * * * **** * * ** * * * fe q <0.05 q <0.05 0 2k 4k 6k 8k 10k Gene ranks 0 2 4 6 8 G N LY + s ig na tu re ge ne s co re s GNLY FGFBP2 CX3CR1 PRSS23 PTGDS LAIR2 PLEKHG3 GPR141 ITGAM ASCL2 CES1 ZNF683 RAP1GAP2 GZMH ADGRG1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 G N LY + s ig na tu re ce ll sc or es CMV Supplementary Fig. 10: Effect of CMV serostatus on cell composition and aging signatures. a) Heatmaps of immune subset composition changes by CMV serostatus across tissues. (LFC CMV-/CMV+) assessed by multivariate linear regression. b) Violin plots of GZMK+CD8+TEMRA frequencies across donors by age and CMV serostatus. c) Consensus scHPF factor for the GZMK+CD8+T cell signature. Dot plots show gene ranks and cell scores by CMV serostatus group, with linear mixed modeling used for the adjusted cell scores. d) Heatmap showing normalized enrichment scores (NES) from pre-ranked GSEA using top 200 genes from the GZMK+ signature as gene set in CD8+T cell subset-level DE across CMV serostatus. e) Consensus scHPF factor for the CD8+T cell signature #4. Dot plots show gene ranks and cell scores by CMV serostatus group, with linear mixed modeling used for the adjusted cell scores. f) Heatmap showing normalized enrichment scores (NES) from pre- ranked GSEA using top 200 genes from CD8+T cell signature #4 as gene set in CD8+T cell subset-level DE across CMV serostatus. * denotes an adj. p-val (FDR) < 0.05. - + -0.8 -0.4 0.0 0.4 0.8 Pa rti al re g. c el l s co re CMV - + BL O BM SP L LN LN G JE J CD8+ TN CD8+ TEM CD8+ TRM CD8+ TEMRA -2 -1 0 1 2 NES (CMV+/CMV-) * * * * *** * * ** * * **** dc 0 4k2k 6k 8k 10k G ZM K + s ig na tu re ge ne s co re s 8 6 4 2 0 GZMKVCAM1CMC1 EOMESHLA-DQA1 HLA-DQA2LGR6 HLA-DQB1 DTHD1 PDCD1 HLA-DRB5 HLA-DRB1 FAM49A FCRL3 RASGEF1A Gene ranks 0.0 0.5 1.0 1.5 2.0 2.5 3.0 G ZM K + s ig na tu re ce ll sc or es CMV - + -0.8 -0.4 0.0 0.4 0.8 Pa rti al re g. c el l s co re CMV - +<40 >40 Age 0 50 100 CMV- CMV+ % G ZM K + C D 8+ T EM R A b a BL O BM SP L LN LN G JE J CD4+ T cells CD4+ TCM CD4+ TEM CD4+ TRM CD4+ TREG CD4+ TN CD8+ T cells BL O BM SP L LN LN G JE J CD8+ TEM CD8+ TRM CD8+ TEMRA CD8+ TN BM SP L LN LN G JE J B cells BMEM Plasmablast Plasma BN Myeloid cells BL O BM SP L LN LN G JE J * cMono ncMono Macrophage * * NK/ILCs BL O BM SP L LN LN G JE J * *CD56dim NK CD56bright NK ILC1 ILC3 pre-NK/ILC -1 -.5 0 .5 1 Su bs et fr eq ue nc y Lo g 2(C M V+ /C M V- )