bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1 Immune Profiling of Atherosclerotic Plaques Identifies Innate and Adaptive 2 Dysregulations Associated with Ischemic Cerebrovascular Events
3
4 Dawn M. Fernandez1, Adeeb H. Rahman2,3,4, Nicolas Fernandez3, Aleksey Chudnovskiy2, 5 El-ad David Amir3, Letizia Amadori4, Nayaab S. Khan1, Christine Wong1, Roza 6 Shamailova1, Christopher Hill1, Zichen Wang5, Romain Remark2,10, Jennifer R. Li6, 7 Christian Pina6, Christopher Faries6, Ahmed J. Awad7,11, Noah Moss1, Johan L.M. 8 Bjorkegren4, Seunghee Kim-Schulze2,3,8, Sacha Gnjatic2,3,9, Avi Ma’ayan5, J. Mocco7, 9 Peter Faries6, Miriam Merad2,3,9, and Chiara Giannarelli1,2,4*
10
11 1Department of Medicine, Cardiovascular Research Center 12 2The Precision Immunology Institute 13 3Human Immune Monitoring Center 14 4Department of Genetics and Genomic Sciences 15 5Department of Pharmacological Sciences, Mount Sinai Center for Bioinformatics 16 6Department of Surgery, Vascular Division 17 7Cerebrovascular Center, Department of Medicine 18 8Hematology and Medical Oncology Division, Department of Neurosurgery 19 9Department of Oncological Sciences, The Tisch Cancer Institute, Icahn School of 20 Medicine at Mount Sinai, New York, NY 10029, USA 21 10Present address: Innate Pharma, 117 Avenue de Luminy, 13009, Marseille, France 22 11Present address: Department of Neurosurgery, Medical College of Wisconsin, WI, USA 23
24 *Correspondence: [email protected]
25 bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
26 SUMMARY
27 Atherosclerosis is driven by multifaceted contributions of the immune system within the 28 circulation and at vascular focal sites. Yet the specific immune dysregulations within the 29 atherosclerotic lesions that lead to clinical cerebro- and cardiovascular complications (i.e. 30 ischemic stroke and myocardial infarction) are poorly understood. Here, using single-cell 31 mass cytometry with Cellular Indexing of Transcriptomes and Epitopes by Sequencing 32 (CITE-seq) we found that atherosclerotic plaques were enriched in activated, 33 differentiated, and exhausted subsets of T cells vs. blood. Next, using single-cell 34 proteomic, transcriptomic, and cell-to-cell interaction analyses we found unique functional 35 dysregulations of both T cells and macrophages in plaques of patients with clinically 36 symptomatic (SYM; recent stroke of TIA) or asymptomatic (ASYM, no recent stroke) 37 carotid artery disease. SYM plaques were enriched with a distinct CD4+ T cell subset, 38 and T cells were activated, differentiated and presented subset specific exhaustion. SYM 39 macrophages presented alternatively activated phenotypes including subsets associated 40 with plaque vulnerability. In ASYM plaques, T cells and macrophages were activated and 41 displayed a strong IL-1β signaling across cell types, that was absent in SYM plaques. 42 The identification of plaque-specific innate and adaptive immune dysregulations 43 associated with cerebrovascular events provides the basis for the design of precisely 44 tailored cardiovascular immunotherapies.
45
46 KEYWORDS
47 Atherosclerotic plaque, T cells, macrophages, cerebrovascular events, Stroke, CyTOF, 48 scRNA-seq, CITE-seq, cell-cell interactions, IL-1β, PD-1, T cell exhaustion.
49
2
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
50 INTRODUCTION
51 Vascular inflammation is a key component of atherosclerosis that contributes to plaque 52 instability and leads to clinical cardiovascular (CV) events including ischemic stroke and 53 myocardial infarction1. Despite decades of intensive research, the immune mechanisms 54 contributing to these processes remain largely unresolved. Most studies in the field rely 55 on the use of animal models that fail to adequately recapitulate the human immune 56 environment2,3 and lack spontaneous plaque rupture2,4. Pioneering histological studies 57 from plaques of acute coronary death patients have established that culprit lesions 58 present a large necrotic core, a thin fibrous cap and a high ratio of macrophages to 59 vascular smooth muscle cells5-12. Consequently, most research has focused on 60 macrophage infiltrates and their functionally diverse subsets as key drivers of plaque 61 instability13-18. However, the inherent immune cell diversity in atherosclerotic plaques 62 suggests a critical yet obscure role for other immune cell populations at the 63 atherosclerotic vascular site1,19-21. For example, murine studies demonstrate that plaque- 64 derived T cell subsets can be either pro- or anti-atherogenic 21-23, but the phenotypic and 65 functional diversity of these and other cell types in human atherosclerosis and their 66 contribution to human plaque pathology remain undefined. Several studies support the 67 hypothesis that circulating immune cells influence the clinical course of atherosclerosis, 68 as patients with acute CV events present increased monocyte and CD4+ T cell subtypes 69 in blood24-31. However, the interplay of systemic immune responses with those occurring 70 the atherosclerotic vascular site, where rupture occurs, are vastly under investigated. 71 Thus, immune cell phenotypes and functional relationships within and between the lesion 72 site and the blood of the same patient are important yet unexplored concepts that could 73 guide the design of conceptually innovative molecularly targeted immunotherapies.
74 Targeting inflammation to reduce CV risk in patients has long been proposed32, and the 75 recent outcomes of the Canakinumab Anti‐inflammatory Thrombosis Outcomes Study 76 (CANTOS)33 support the clinical efficacy of this approach. However, the inhibition of the 77 interleukin‐1β (IL-1β) pathway of the innate system only moderately (-15%) reduced CV 78 events in high risk post-MI patients, leaving a high residual CV risk. Furthermore, the 79 failure of the Cardiovascular Inflammation Reduction Trial (CIRT)34 which tested low‐dose 80 methotrexate, has shown that broad anti-inflammatory treatments are ineffective in 81 reducing CV events. Taken together these observations underscore that 82 immunomodulatory treatments must be tailored to specific immune defects.
83 Therefore, the identification of the specific immune dysregulations at the plaque site could 84 aid new viable immunotherapies for CV disease, beyond the traditional management of 85 risk factors and use of standard of care lipid lowering drugs35-37. The recent use of 86 innovative high-dimensional single-cell analyses to study murine atherosclerosis38-40 87 highlights the potential of exploiting these unbiased approaches to systematically resolve 88 the immune diversity in human atherosclerotic plaques and decipher the molecular 3
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
89 alterations of individual immune cells that contribute to human disease and its CV 90 complications.
91
92 RESULTS
93 Single-Cell Immunophenotyping of Human Atherosclerosis: Study Design
94 To map the immune microenvironment of atherosclerotic lesions, identify mirroring 95 immune changes in blood and pinpoint cell-specific alterations associated with clinical CV 96 events (i.e. stoke and TIA), we performed a large-scale CyTOF mass-cytometry 97 analysis41 combined with Cellular Indexing of Transcriptomes and Epitopes by 98 Sequencing (CITE-seq)42 and single-cell scRNA-seq analysis of plaques from a total of 99 46 prospectively enrolled patients undergoing carotid endarterectomy (Fig. 1, 100 Supplementary Table 1a). A first single-cell proteomic CyTOF analysis combined with 101 an unbiased MetaClustering computational approach (cohort 1, discovery cohort; n=15) 102 revealed an unexpected predominance of heterogeneous T cell subsets in atherosclerotic 103 plaques compared to blood. We also identified discrete adaptive dysregulations within 104 plaques of patients with recent cerebrovascular events (i.e. stroke, TIA) (symptomatic, 105 SYM; n=7) vs. asymptomatic patients (ASYM; n=8) (Supplementary Table 1b). Based 106 on these initial observations, we performed an in-depth CyTOF analysis of T cells in blood 107 and plaque of the same patient (n=23) (Fig. 1b) in prospectively enrolled ASYM (n=14) 108 and SYM (n=9) patients (cohort 2, validation cohort) (Supplementary Table 1c). We 109 complemented these studies with Cellular Indexing of Transcriptomes and Epitopes by 110 Sequencing (CITE-seq) analysis of blood and paired atherosclerotic plaque of the same 111 patient. Finally, to identify plaque specific immune alterations associated with clinical 112 cerebrovascular ischemic events, we investigated alterations of T cells and macrophages 113 in atherosclerotic plaques of SYM and ASYM patients combining CyTOF, scRNA-seq and 114 cell-cell interaction analyses (Fig. 1c).
115 Single-Cell Mass-Cytometry Immunophenotyping of Atherosclerotic Plaque and 116 Paired Blood
117 The phenotypic distribution of immune cells in atherosclerotic tissue and blood by CyTOF 118 were analyzed using viSNE43, an unbiased computational tool that distills high- 119 dimensional single-cell data into two dimensions (Fig. 2a). Unbiased MetaLouvain 120 clustering analysis44,45 was then used to first cluster single cells by shared proteins in 121 each tissue and blood for individual patient. Then, secondary clustering identified 122 MetaClusters (MCs) common across tissues and patients. MCs discovered in cohort 1 123 (discovery cohort, n=15) largely corresponded to known immune cell populations with 124 distinct distributions between blood and plaque of patients (Fig. 2a-d, Extended Fig. 1a 125 and b). All major immune cell lineages were identified in atherosclerotic lesions, with
4
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
126 CD4+ and CD8+ T cells combined being the most abundant (≈65%), and with CD8+ T cells 127 being the most enriched vs. blood (Fig. 2d). Signatures stratified by tissue type, identified 128 immune cells specifically enriched in atherosclerotic lesions that included macrophages 129 (MC 4), CD8+ EM T cells (MC 12), CD4+ EM T cells (MC 7) and CD4–CD8– T cells (MC 130 18) (Fig. 2e, f, Extended Fig. 1b; Supplementary Table 2). As expected, CD14+ 131 monocytes (MC 13) and natural killer (NK) cells (MC 6) were more abundant in blood. 132 Similarly, plasmacytoid dendritic cells (pDCs) (MC 16), B cells (MC 3), and CD4+CD8+ T 133 cells (MC 10) were enriched in blood, although the latter accounted for < 3% of the overall 134 immune compartment (Extended Fig. 1b and c).
135 Single-Cell Mass-Cytometry of Immune Cells in Atherosclerotic Plaque.
136 An independent MC analysis comparing immune cells derived from ASYM (n=8) and SYM 137 (n=7) plaques from patients of cohort 1 identified 15 plaque tissue-specific MCs 138 (Extended Fig. 2a-d). These included 2 MCs of macrophages (MC 1 and MC 4). In 139 particular, macrophages of MC 1 showed high expression of CD206 and CD163 140 (Extended Fig. 2a), markers of alternative M2-like macrophages in atherosclerotic tissue 141 that have been associated with either atheroprotective or proatherogenic functions46,47. 142 This analysis also revealed 9 MCs of T cells in plaques which comprised 3 subsets of 143 CD8+ T cells (MCs 3, 7 and 10) and 5 of CD4+ T cells (MCs 5, 8, 9, 13 and 15) (Extended 144 Fig. 2a, c, d, see Supplementary Information and Supplementary Fig. 1). 145 Interestingly, we found an exclusive enrichment of CD8+ T cells (MC 7) and of CD4+CD56+ 146 cells (MC 5) in SYM plaques (Extended Fig. 2e). Remarkably, these differences were 147 not driven by plaque pathology—as both prospectively enrolled ASYM and SYM patients 148 of cohort 1 presented the same type VI plaque according to the American Heart 149 Association (AHA) classification47—nor by their clinical characteristics (Supplementary 150 Information and Supplementary Fig. 2a-e).
151 Single-Cell Mass-Cytometry Analysis of T Cells in Atherosclerotic Plaque and 152 Paired Blood
153 The findings from our discovery cohort (cohort 1) suggested that in human atherosclerotic 154 plaques T cells were the most abundant type of immune cell, displayed a greater 155 heterogeneity than macrophages and presented unique alterations associated with 156 clinical CV events. To validate these data, further resolve the heterogeneity of the T cell 157 compartment in human atherosclerosis, and identify new specific adaptive dysregulations 158 in plaques, we analyzed a second independent cohort (cohort 2, validation cohort, n=23) 159 of prospectively enrolled ASYM and SYM patients (Supplementary Table 1) with an 160 extended mass-cytometry antibody panel (Fig. 1a; See Methods and Supplementary 161 Table. 3). This analysis confirmed that T cells were dominant in plaques compared to 162 blood, and discovered an unprecedented diversity of the adaptive immune compartment 163 in atherosclerosis (Fig. 3a-d, Extended Fig. 3a-b). We identified thirteen MCs of CD4+ T
5
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
164 cells that included central memory (CM) (MCs 7, 14, 17, 23 and 24), effector memory 165 (EM) (MCs 1, 4, 12, 18, and 22), terminally differentiated effector memory (EMRA) (MC 166 2), and regulatory T cells (MCs 6 and 19). The CD8+ T-cell compartment comprised 6 167 MCs, including naïve (MC 13), EM (MCs 20, 11 and 10), and EMRA (MCs 9 and 21) (Fig. 168 3e, Extended Fig. 3a-c; See Supplementary Information and Supplementary Fig. 3).
169 Atherosclerotic lesions had a higher frequency of CD8+ T cells (≈39% vs. ≈26% in blood), 170 and a lower frequency of CD4+ T cells (≈50% vs. ≈65% in blood) (Fig. 3f), confirming our 171 initial finding of a CD8+ T cell enrichment in plaque relative to blood. A cross-tissue 172 analysis revealed 4 MCs enriched in atherosclerotic plaques (Fig. 3g, Extended Fig.3b) 173 that were all EM T cells (CCR7low and CD45RAlow) and shared high expression of CD69, 174 an early activation and tissue residency marker48 (Fig. 3e). Three of these MCs were 175 CD8+ subsets (MCs 10, 11, 20), and one was a CD4+ subset (MC 12) (Fig. 3e-h, 176 Extended Fig. 3d). Despite belonging to different T cell lineages, MCs 10, 12, and 20 177 displayed the most similar marker expression pattern and unbiasedly clustered together 178 (Fig. 3e, Extended Fig. 3e), suggesting similar functional states in plaque. Conversely, 179 MC 11 unbiasedly clustered with EMRA subsets, suggesting either functional similarities 180 or a transition toward a terminally differentiating phenotype of this MC (Fig. 3e, Extended 181 Fig. 3d). A key observation was that CD8+ T cells from MC 10 were CD103+ and detected 182 exclusively in plaque (Fig. 3e and h, Extended Fig. 3b), and corresponded to a classical 183 tissue-resident memory (TRM) T cell that has been reported in human lymphoid and non- 184 lymphoid tissues48-50. Our study is the first to identify their presence in the atherosclerotic 185 arterial wall in humans.
186 Single-Cell Mass-Cytometry Analysis of T Cells Enriched in Atherosclerotic 187 Plaques
188 Plaque-derived CD4+ and CD8+ EM T-cells of MCs 11, 12, and 20 expressed markedly 189 higher levels of CD69 compared to blood (Extended Fig. 4a-c). Because CD69 can 190 define TRM subsets regardless of CD103 expression48, these observations suggest that 191 plaque-enriched MCs might represent TRM subsets. CD69 is also a marker of early 192 activation; therefore, we assessed the activation markers CD38, CCR5, and HLA-DR48,51. 193 CD38 was uniformly upregulated on cells of MCs 11 and 12, and MC 20 compared to 194 blood. In contrast, CCR5 and HLA-DR showed different patterns of expression. CCR5 195 was upregulated in plaque-derived cells from MCs 11 and 12, while MC 20 had similarly 196 high levels of CCR5 regardless of origin. HLA-DR was only upregulated in tissue cells of 197 MC20 compared to blood (Extended Fig. 4a-c). These data suggest that CD4+ and CD8+ 198 T cell subsets that are enriched in atherosclerotic plaque tissue are more activated than 199 their blood counterparts and present a heterogeneous spectrum of activation.
200 Surprisingly, T cells from MCs 11, 12, and 20 expressed high levels of PD-1, a negative 201 regulator of T cell activation and marker of T cell exhaustion52-54, in atherosclerotic tissue
6
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
202 compared to blood. Additionally, T cells in tissue generally expressed lower levels of the 203 co-stimulatory molecules CD28, CD27, and CD127 compared to blood (Extended Fig. 204 4a-c). Overall, plaque T cell subsets exhibited a chronically activated and differentiated 205 phenotype, and because they presented high PD-1 levels, these subsets may 206 simultaneously initiate a regulatory feedback and exhaustion reprogramming in response 207 to plaque chronic inflammation52-54. Consistent with this observation, we found lower 208 levels of perforin, but not of granzyme B, in plaques for CD8+ T cells of MC 11, 12 and 209 20, which suggests only partially reduced cytotoxic functions, typically seen in early 210 phases of T cell exhaustion55-57 (Extended Fig. 4d). Because MC 10 was exclusive to 211 atherosclerotic plaque, we compared it to the other CD8+ EM subsets (MCs 11 and 20) 212 in tissue (Extended Fig. 4c). CCR5 expression was similar between MCs 10 and 20, and 213 higher than in MC11, suggesting higher activation of both MC 10 and MC 20 than MC 11 214 in tissue. MC 10 also displayed lower levels of the co-stimulatory molecules CD27 and 215 CD28 than MCs 11 and 20, suggesting that the cells of MC 10 were more differentiated 216 than other EM CD8+ T cells in tissue52. Comparable levels of PD-1 between all CD8+ from 217 MCs 10, 11, and 20 suggested a similar degree of exhaustion (Extended Fig. 4c). An 218 independent MC analysis on plaque-derived T cells from the same cohort (cohort 2) 219 identified a MC of CD69+CCR5+PD1intCD127- CD8+ T cells whose frequencies correlated 220 with TCR clonality in tissue, suggesting clonal expansion of these cells in atherosclerotic 221 plaques (see Supplementary Information and Supplementary Fig. 4).
222 Simultaneous Surface Marker and Gene Expression Immune Profiling Across 223 Plaque and Blood: CITE-seq Analysis
224 In order to further investigate the identified functional differences and decipher key 225 differences in immune molecular mechanisms and signaling pathways in plaque tissue 226 and in blood, we analyzed the T cell compartments using Cellular Indexing of 227 Transcriptional Epitope Sequencing (CITE-seq)42, an innovative method that integrates 228 proteomic (surface marker expression) and transcriptional (gene expression, GEX) data 229 into a single readout. We also focused on plaque macrophages, given their undisputed 230 role in the pathogenesis of the disease17,21,46,58,59. We analyzed 5,232 total cells (1,643 231 from plaque and 3,589 from blood) and assigned cell types according to the canonical 232 surface marker expression from the Antibody Derived Tag (ADT) data. Consistent with 233 our CyTOF results T cells accounted for the majority of all immune cells in plaque (Fig. 234 4a-c) and CD8+ T cells were enriched in plaque (≈46%) compared to blood (≈10%). (Fig. 235 4c). CD8+ and CD4+ T cell subsets confirmed distinct functional states based on tissue 236 type, with blood having higher levels of naïve and central memory markers (CD62L and 237 CD27), and plaque having higher levels of the activation markers HLA-DR in both 238 lineages and of CD38 exclusively in CD8+ T cells (Extended Fig. 5a-e). Plaque 239 macrophages accounted for about 16% of CD45+ cells (Extended Fig. 5f, g). Similar to 240 our CYTOF results, we identified 2 macrophage subsets (CD64+HLADR+CD206hi and
7
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
241 CD64+HLADR+CD206lo) based on the varied expression of CD206, a known marker of 242 alternatively activated M2 phenotype, which accounted for ~47% of plaque macrophages 243 (Extended Fig. 5h, i).
244 GEX analysis of T cells in blood and atherosclerotic plaques
245 Next, we analyzed the single-cell transcriptional data of T cells in plaque and blood (Fig. 246 4d). Plaque T cells displayed gene expression signatures associated with T cell activation 247 (NFATC2, FYN, ZAP70), cytotoxicity (GZMA, GZMK), and T cell exhaustion (EOMES, 248 PDCD1, LAG3). These processes were reflected by signaling pathway analysis that 249 highlighted T cell effector functions (Interferon and DAP12 Signaling60,61), proliferation 250 and survival (IL-2 Signaling62), metabolic changes (Leptin Signaling63,64), and Cytokine 251 Signaling pathways (Fig. 4e). In contrast, T cells in blood presented genes associated 252 with cytokine inhibition (KLF2)65, RNA synthesis (EEF1A1, EIF3L, TCEA3) and metabolic 253 reprogramming of circulating T cells (TXNIP) (Fig. 4d). These cells upregulated TGF-β 254 signaling, RNA processing and splicing (HNRNPM, SRSF5, RBMX), and metabolism 255 (EIF4A2, PABC1, UBA52) pathways, processes generally inhibiting T cell functions66 256 (Fig. 4e). Collectively, suggesting a resting phenotype in blood.
257 Unbiased hierarchical clustering of individual T cells identified 16 clusters with distinct 258 enrichments in plaque or blood (Fig. 4f). Plaque-enriched clusters mostly corresponded 259 to CD8+ T cells (1, 3, 4, and 12), and one CD4+ T cell cluster (9), confirming our CyTOF 260 observations that plaques are enriched in CD8+ T cells and are composed of functionally 261 heterogeneous subsets. Blood-enriched clusters mostly corresponded to CD4+ T cells (6, 262 11,15, 16). The clusters fell into two main functional categories, either activated or resting 263 subsets (Fig. 4g). Blood-enriched (11, 15, 16) and plaque-enriched (4, 12) clusters 264 presented a resting phenotype consistent with upregulated Rho GDP-dissociation 265 inhibitor (RhoDGI) signaling, which antagonizes processes involved in T cell activation67. 266 Blood-enriched Cluster 15 also displayed genes associated with RNA processing and 267 Oxidative Phosphorylation (ATP5MG, COX4I1, UQCRB), suggesting their involvement in 268 regulating T cell functions66. Conversely, plaque-enriched clusters 1, 3, 9 and blood- 269 enriched cluster 6 presented distinct T cell activation states based on the differential 270 representation of several canonical pathways. For example, RhoA signaling68 was 271 upregulated in clusters 3, 6, and 9, and PCKθ signaling69,70 was upregulated in clusters 272 1, 3 and 9. Plaque-specific clusters 1, 3, and 9 uniquely upregulated signaling associated 273 with activation-induced T cell exhaustion, including PKCθ signaling, required for TCR- 274 induced T cell activation71, and NFAT signaling72,73. Overall, these data confirm our 275 CyTOF observations that in blood, T cells display a quiescent state, while in plaques, they 276 exhibit distinct degrees of activation which overlaps with exhaustion, suggesting the 277 stepwise and progressive loss of T-cell functions in response to chronic, persistent 278 inflammation55-57,74. A sub analysis of individual CD4+ and CD8+ T cells across blood and 279 plaque is available as Supplementary Information (Supplementary Fig. 5)
8
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
280 GEX analysis of macrophages in atherosclerotic plaques
281 The transcriptional profile of lesional macrophages was analyzed using hierarchical 282 clustering, that identifed 5 distinct clusters. This method revealed a greater functional 283 heterogeneity compared to the two subsets detected in our CyTOF and CITE-seq 284 proteomic analysis (Fig. 4h). Signaling pathway analysis revealed that clusters 1, 2 and 285 3, were more activated and pro-inflammatory than cluster 5, which presented a foam cell 286 transcriptional signature (Fig. 4h and i). Cluster 1 expressed genes involved in 287 macrophage migration75 (MHCII genes [i.e. HLA-DRA] and CD74) and demonstrated 288 upregulation of NFAT signaling, which dampens innate inflammatory responses through 289 cytokine inhibition76. Cluster 2 was highly inflammatory, expressing genes involved in 290 inflammatory responses (CYBA, LYZ, S100A9, AIF1, S100A8), toll-like receptor (TLR) 291 binding (S100A9, S100A8), and oxidoreductase activities (CYBA). Interestingly, cluster 2 292 upregulated genes involved in matrix metalloprotease (MMP) suppression (TIMP1), a 293 process that regulates extracellular matrix (ECM) degradation77. This data suggests that 294 these cells might contribute to plaque stabilization. Cluster 3 showed reduced gene 295 expression compared to other clusters. However, these cells uniquely upregulated genes 296 involved in pro-inflammatory responses (JUNB, NFKBIA)78 and highly expressed 297 MALAT1, a long non-coding gene implicated in M2 macrophage polarization46,79 and foam 298 cell formation80. Cluster 3 upregulated liver X receptor (LXR) and retinoid X receptor 299 (RXR) activation signaling, processes that modulate reverse cholesterol transport and 300 efflux81, suggesting these were precursory foam cells. Finally, cluster 5 upregulated 301 genes involved in cholesterol uptake and metabolism (PSAP, APOC1, APOE) and lipid 302 accumulation (PLIN2)82, and may represent a foam cell subset functionally distinct from 303 cluster 3. These macrophages showed increased RhoGDI and Apelin signaling, and 304 reduced NO/ROS production and pro-inflammatory cytokine signaling (e.g. IL-1, IFN). 305 With the exception of the pro-atherogenic chemokine CCL283, this data suggested an 306 anti-inflammatory nature for macrophages of cluster 5.
307 Discrete Immune Dysregulations Linked to Clinical Cerebrovascular Outcomes
308 Single-Cell Proteomic Analysis of ASYM and SYM plaques
309 To identify the plaque-specific immune perturbations associated with ipsilateral acute 310 ischemic cerebrovascular events, we stratified the profiles of SYM and ASYM patients 311 from the T cell MC analysis (Cohort 2, Fig. 3, 5a). We first analyzed MC frequency and 312 found that SYM patients had a greater frequency of MC12 (CD4+ EM T cells) in the culprit 313 plaque (≈32%, SYM vs. ≈20%, ASYM) (Fig. 5b-d, and Supplementary Fig. 6a). 314 Remarkably, these changes were unrelated to the plaque type V or VI seen in cohort 2 315 based on the pathological criteria of the AHA84, the current gold standard to identify 316 vulnerable/complicated atherosclerotic lesions (Fig. 5e-g, See Supplementary 317 Information, Supplementary Fig. 2f-h). Surface marker expression of MC 12 did not
9
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
318 identify significant phenotypic differences in MC 12 of SYM and ASYM patients (see 319 Supplementary Information, Supplementary Fig. 6b).
320 Single-Cell Transcriptomic Analysis of ASYM and SYM plaques
321 To further investigate discrete adaptive immune alterations and identify functional 322 dysregulations of macrophages associated with clinical cerebrovascular outcomes, we 323 performed a scRNA-seq analysis of T cells and macrophages in plaques of SYM and 324 ASYM patients.
325 CD4+ T cells Transcriptional Alterations in ASYM and SYM Plaques
326 Gene set enrichment analysis of differentially expressed genes (DEGs) in CD4+ T cells 327 from SYM vs. ASYM plaques (Fig. 5h-i) highlighted a transcriptional profile consistent 328 with stronger activation and effector functions in ASYM patients. ASYM plaques 329 upregulated genes involved in IL-12 signaling, type I interferon production, and increased 330 IFNG expression, which collectively indicate activated effector functions of T cells61. 331 Interestingly, these cells also exclusively upregulated IL-1 and IL-6 signaling pathways 332 (Fig. 5i). Because IL-1 acts as a licensing signal that stabilizes cytokine transcripts and 333 enables effector functions85 and IL-6 promotes T cell survival86, these pathways may 334 enhance effector functions of CD4+ T cells in ASYM plaques. Conversely, SYM CD4+ T 335 cells explicitly lacked IL-1 and IL-6 signaling pathways, which may indicate that 336 therapeutics that target these pathways33 may be less effective in targeting plaque 337 specific immune responses following acute ischemic events. Furthermore, CD4+ T cells 338 from SYM plaques displayed distinct signaling pathways involved in T cell migration 339 (RhoGTPase)68, T cell activation (PDGFR-)87, T cell differentiation (Wnt88, IL-262) and T 340 cell homeostasis (CD28).
341 Unbiased hierarchical clustering of individual CD4+ T cells across all plaques identified 4 342 main clusters with distinct enrichments between ASYM (clusters 15 and 58) and SYM 343 (clusters 11 and 35) plaques (Fig. 5j). ASYM-enriched clusters shared many of the same 344 pathways highlighting a similar activated cell phenotype (Fig. 5k). Conversely, SYM- 345 enriched cluster 11 was activated and proinflammatory, as indicated by Th1 pathway and 346 RhoA signaling, while SYM-enriched cluster 35 presented an overall downregulation of 347 gene expression.
348 CD8+ T cells Transcriptional Alterations in ASYM and SYM Plaques
349 Gene expression analysis of CD8+ T cells in ASYM and SYM plaques identified a 350 transcriptional profile (Fig. 6a) which partially overlapped with that of CD4+ T cells (Fig. 351 5h). CD8+ T cells in ASYM plaques upregulated activation genes (CD69, NFKBIA, IFNG, 352 and TNF) and signaling pathways (type I interferon signaling, TNF signaling, IL-6 353 signaling) (Fig. 6b). These patterns of upregulation were shared between T cell lineages,
10
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
354 suggesting a similar activation state. However, upregulation of unique genes (CXCR4, 355 GZMB, CCL4L2) and signaling pathways (granzyme, TGF) indicated specific 356 chemotactic and cytotoxic functions of these CD8+ T cells in ASYM plaques. In contrast, 357 SYM CD8+ T cells were largely characterized by a distinctive set of genes (RAC2, CCL5, 358 CYBA, GZMK, LIMD2) (Fig. 6a) and signaling pathways (TCR, CXCR2, cellular 359 senescence, and IFN) (Fig. 6b, c). Hierarchical clustering of all CD8+ T cells identified 5 360 clusters with distinct enrichments between ASYM (clusters 3 and 4) and SYM (clusters 1, 361 5 and 7) plaques (Fig. 6d). Signaling pathway analysis (Fig. 6e) revealed that ASYM- 362 enriched cluster 4 was characterized by inflammatory (IL-6, chemokine signaling) and 363 regulatory pathways related to T cell activation and migration (i.e. RAC signaling, 364 TNFR1/2 signaling, TGF signaling). This suggests that activation signaling is tightly 365 regulated in this cluster. SYM-enriched cluster 1 and ASYM-enriched cluster 3 shared a 366 similar, activated functional state characterized by RhoA, Integrin, IL-8 and NFAT 367 signaling pathways (Fig. 6e). However, cluster 1 was distinctive in the co-activation of 368 Interferon and T cell exhaustion signaling (Fig. 6e), suggesting that these highly activated 369 cells initiate a parallel exhaustion reprogramming. Because T cell exhaustion is 370 characterized by the gradual and continuous depletion of effector functions in response 371 to chronic, non-resolving inflammation55-57,74,89, the identification of concomitant activation 372 and exhaustion signaling suggests that these activated cells initiate exhaustion 373 reprogramming.
374 Macrophage Transcriptional Alterations in ASYM and SYM Plaques
375 Similar to the T cell compartment, ASYM macrophages were more activated, pro- 376 inflammatory, and displayed enhanced foam cell functions compared to SYM 377 macrophages. ASYM macrophages highly expressed a series of pro-inflammatory 378 chemokines (Fig. 6f) corresponding to signaling pathways including IL-1 and IL-6 (Fig 379 6g). In fact, IL-1 regulation was highly represented in ASYM plaques, which upregulated 380 genes IL1B and IL1RAP, a component of the IL-1 receptor complex, and NFKBIA and 381 NLRP3 mediators of IL-1 production90 (Fig 6f). Interestingly, inhibitory IL-1 decoy 382 receptor (IL1RN), was co-expressed, indicating tight self-regulatory mechanisms for IL- 383 1 signaling in macrophages of ASYM plaques (Fig 6f). Additionally, these macrophages 384 upregulated genes involved in lipid metabolism and lipoproteins implicated in foam cell 385 formation (e.g. LPL), a hallmark of all stages of atherosclerosis1,21,59.
386 In SYM plaques, macrophages upregulated cytolytic genes GZMA, GZMK, GZMM and 387 LYZ, as well as the proatherogenic chemokine CCL5 that has been linked to an unstable 388 plaque phenotype91,92(Fig. 6f). Signaling pathway analysis revealed heterogeneous 389 populations of inflammatory, reparative, pro-angiogenic, and anti-fibrotic macrophages 390 (Fig. 6h). Top pathways included CXCR4 signaling, involved in cell migration and TLR2 391 signaling93,94, which indicates pro-inflammatory functions of these cells. We also
11
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
392 identified pathways involved in podosome formation (RAC195,96), and phagocytosis 393 (RAC197 and IL-2198 signaling), both processes that promote reparative functions in 394 macrophages. We also identified pathways regulating macrophage survival and wound 395 healing (IL-35 signaling99) (Fig. 6h). Furthermore, we identified pro-angiogenic IL-8 and 396 CXCR2 signaling, which have been implicated in plaque instability100-103, and the 397 Hedgehog and Wnt signaling, both involved in anti-fibrotic responses which may 398 contribute to plaque destabilization104-106 in SYM macrophages.
399 Unbiased hierarchical clustering revealed heterogeneous and highly specialized 400 macrophage subsets enriched in either ASYM plaques (clusters 5, 6 and 8) or SYM 401 plaques (clusters 2, 9 and 10) (Fig. 6i). Consistent with our initial analysis, ASYM 402 macrophages were either phenotypically activated and pro-inflammatory (cluster 6 and 8) 403 or demonstrated a foam cell profile (cluster 5) (Fig. 6i, j). Macrophages from cluster 8 404 expressed proinflammatory genes IL1B, CXCL2, CXCL3 and CXCL8, and upregulated 405 activation pathways (inflammasome pathway, IL-1, IL-6, iNOS, MIF and TREM1 406 signaling) (Fig. 6i and j). Cluster 5 uniquely upregulated genes involved in lipid uptake 407 and metabolism (APOC1, APOE, FABP5) and cholesterol transportation (NPC2) 408 suggesting that these macrophages were foam cells (Fig. 6i). Cells in this cluster also 409 upregulated apoptosis signaling. In contrast, clusters from SYM plaques 2, 9 and 10 410 displayed fewer proinflammatory functions (i.e. NFAT signaling) and stronger signaling 411 associated with alternatively activated phenotypes linked to plaque instability (Fig. 6j). 412 Clusters 9 and 10 shared similar phenotypes including PPAR/RXR signaling that has 413 been reported to drive macrophage polarization towards the reparative M2 phenotype107- 414 111. Cluster 9 was distinct based on the activation of NFAT signaling, which is a shared 415 signaling to that of macrophages in rheumatoid arthritis112 and inflammatory bowel 416 disease113, conditions that are at high risk of CV disease114,115. Cluster 2 was uniquely 417 characterized by the upregulation of genes responsible for iron metabolism (FTH1) and 418 iron storage (FTL). This cluster may represent alternatively activated pro-inflammatory 419 non-foamy macrophages with proatherogenic potential, which are known to be involved 420 in the clearance of iron derived from hemoglobin deposited in areas of intraplaque 421 hemorrhage47. This phenomenon has been associated with plaque progression and 422 features of plaque instability and is typically seen in AHA type VI complicated 423 plaques84,116,117, the same type that is exhibited by all SYM patients in our cohort.
424 Cell-to-Cell Communications in Atherosclerotic Plaques Associated with 425 Cerebrovascular Events
426 We next utilized an established computational approach118 to infer cellular communication 427 mechanisms119,120 by quantifying the potential ligand-receptor interactions between cell 428 types based on their single-cell gene expression profiles. Using this approach118, we 429 identified cell-to-cell communication mechanisms that may contribute to the distinct 430 functional state of T cell and macrophage in atherosclerotic plaques of ASYM and SYM 12
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
431 patients (Fig. 7 and Extended Fig. 6, see Supplementary Information, and 432 Supplementary Figs. 7-8).
433 T cell-T cell Interactions in ASYM and SYM plaques
434 In ASYM plaques, we identified potential ligand-receptor interactions between CD4+ and 435 CD8+ T cells that could sustain the activated, pro-inflammatory phenotype, seen in both 436 T cell lineages, for example through TNF and IFNγ signaling (Fig. 7a; Extended Fig.6a, 437 b; Supplementary Figs. 7a,b, 8a,b). Additionally, we found ligand-receptor interactions 438 involved lymphocyte chemotaxis and migration121, including cytokines and chemokine 439 ligands that interact with receptors on both CD4+ and CD8+ T cells. In SYM plaques, we 440 identified distinct interactions known to be involved in key functional states like T cell 441 activation, differentiation, and exhaustion. Interestingly, both CD4+ and CD8+ T cells 442 expressed PSEN1, a gamma secretase component known to cleave Notch receptors 443 (NOTCH1/2/3)122 that were expressed by both T cell linages in plaque. This signaling is 444 involved in T cell function and may contribute to the observed T cell activation123, 445 differentiation124, PD-1 expression and regulation of cytolytic functions125 in our plaques. 446 Our analysis also revealed that T cells expressing the same ligands could interact through 447 distinct receptor pairs on CD4+ and CD8+ T cells in SYM and ASYM plaques. For instance, 448 IL-15 expressed by CD8+ T cells could modulate both CD4+ and CD8+ T cell functions, 449 but through different receptors expressed by these cells in ASYM and SYM plaques. In 450 ASYM plaques, we identified IL-15-IL15RA on CD4+ T cells, an interaction involved in cell 451 proliferation, cell migration, and inhibition of apoptosis126. On the other hand, in SYM 452 plaques IL-15-IL2RB on both CD4+ and CD8+ T cells could regulate Th1 453 reprogramming127 and differentiation128, two functional phenotypes that emerged in SYM 454 plaques by from our CyTOF and scRNA-seq analyses.
455 T Cell-Macrophage Interactions in ASYM and SYM plaques
456 Identified ligand-receptor interactions between T cells and macrophages in ASYM 457 plaques highlighted unique adaptive signals involved in macrophage activation, 458 migration, regulation of lipids and foam cell formation, as well as innate 459 immunomodulatory signaling (Fig. 7b; Extended Fig. 6a, b, Supplementary Fig. 7a-c). 460 Specifically, T cell-macrophage communications involved the activating and pro- 461 inflammatory interactions TNF-TNFRSF1A/1B and IFNG-IFNGR1/2. T cells expressed 462 ligands (HSP90B1, BGN, HRAS) that were predicted to stimulate TLR2 signaling on 463 macrophages, a key pathway that regulates the pro-inflammatory status of macrophages 464 in human atherosclerosis129. Consistent with the specific identification of foam cells in 465 ASYM plaque from our scRNA-seq analysis, we found T cell-macrophage interactions in 466 ASYM plaques known to regulate lipid accumulation and foam cell formation though the 467 expression of multiple ligands that bind to the proatherogenic LDL Receptor-Related 468 Protein (LRP1) on macrophages130. In SYM plaques, T cells might also stimulate
13
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
469 macrophage activation through unique TLR signaling cascades by which ligands (VCAN, 470 HSP90B1) bind to TLRs (TLR1/2/7). Additionally, in SYM plaques we identified ligand- 471 receptor interactions between T cells and macrophages involved in cell survival (HLA- 472 A/C- LILRB1 and B2M-LILRB1)131 and pro-atherogenic functions linked to plaque 473 vulnerability (CCL5-CCR5)91,92,132. Interestingly, T cell expressed ligands that may 474 promote the reparative macrophage polarization detected in these plaques by scRNA- 475 seq analysis, through the expression of GDF11, a TGF superfamily member which 476 bound to ACVR2A/1B on macrophages133.
477 In ASYM plaques, we identified macrophage ligands interacting with CD4+ and CD8+ T 478 cell receptors that could promote T cell activation (JAG1-NOTCH1/3)134, and T cell 479 recruitment through the expression of distinct T cell chemokines. Importantly, ASYM 480 macrophages exclusively expressed IL1B that was predicted to bind to the IL-1 receptor 481 accessory protein IL1RAP expressed by CD4+ and CD8+ T cells, an interaction that may 482 enhance effector functions of T cells infiltrating ASYM plaques85. This data are in line with 483 our findings from scRNA-seq analysis that IL-1 signaling is exclusive to ASYM plaques. 484 On the other hand, in SYM plaques, we identified fewer ligand-receptor interactions 485 between macrophage and T cells. Among those, macrophages modulated T cell 486 activation through the interactions between IL15-IL2RB/G and BTLA-CD247, the latter 487 of which is a component of the TCR-CD3 complex required for T cell activation135. 488 Macrophages could also promote T cell recruitment136 into SYM plaques through their of 489 chemokines-receptor interactions (Fig. 7b; Extended Fig. 6a, b; Supplementary Fig. 490 7a-c).
491 Macrophage-Macrophage Interactions in ASYM and SYM Plaques
492 We detected more diverse and distinctive communication patterns in autocrine 493 macrophage interactions. Specifically, identified macrophage-macrophage interactions in 494 ASYM plaques could induce self-activation and pro-inflammatory interactions via ICAM1 495 binding to ITGAX, ITGB2 and ITGAM, self-recruitment via CCL7 and CCL13 binding to 496 CCR1, CCR2, CCR5 and CXCR3, and activation of inflammatory IL-1 signaling via IL1B 497 binding to IL1R2 and IL1RAP. Also macrophages could stimulate MMP activity and 498 angiogenesis as seen with MMP9, ADAM9, ADAM12, ADAM15 and VEGFA signaling 499 through receptors ITGB1, ITGAV, ITGA3, ITGA5, ITGA6, and NRP1, indicating a strong 500 influence of macrophages in driving ASYM plaque development and progression. In SYM 501 plaques, lesional macrophages could activate antigen processing and presentation 502 through HLA-C upregulation and might promote a proinflammatory state via IL18 503 signaling through IL18R1, and FLT3 to FLT3LG. Interactions were also observed between 504 chemoattractants CCL2, CCL5, CXCL9, CXCL10 and IL16 that bind CXCR3, CCR2 and 505 CCR5 receptors. Additionally, our analysis showed that SYM macrophages may be 506 involved self-maintaining the reparative M2-like phenotype seen in these plaques through
14
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
507 the expression of the M2 marker F13A1137 (Fig. 7c; Extended Fig. 6a, b; 508 Supplementary Fig. 8c).
509
510 DISCUSSION
511 In this study, we provide a first in depth single-cell CyTOF and CITE-seq immune mapping 512 of plaques and paired blood in patients with atherosclerosis. Importantly, unbiased high- 513 dimensional single-cell CyTOF and scRNA-seq analyses uncovered new specific innate 514 and adaptive immune dysregulations at the atherosclerotic site associated with clinical 515 cerebrovascular events.
516 Understanding the contribution of the immune system to the development of human 517 atherosclerosis has been a daunting task. Recently, human plaque immune composition 518 of has been inferred using deconvolution methods on bulk RNA-seq data from human 519 atherosclerotic tissue and proof-of-concept CyTOF analysis of an individual plaque38. A 520 main limitation of this approach is that bulk RNA-seq averages of gene expression 521 patterns in the whole tissue limiting the resolution to identify biologically relevant 522 functional variations of individual cells. Single-cell technological advances allow precise 523 measurements of individual cell phenotypical and functional variations138-141 that have the 524 potential of changing our conceptual understanding of human vascular biology and 525 disease state.
526 Our study is the first and the largest to integrate single-cell CyTOF, CITE-seq and scRNA- 527 seq to study the immune composition of human plaques and to identify the plaque specific 528 immune dysregulations associated with clinical cerebrovascular events at unprecedented 529 scale and resolution. We designed unbiased CyTOF and CITE-seq studies to directly 530 compare immune cell repertoires from paired plaque and blood from the same patient. 531 These analyses underscore unique differences and highly specialized functions of plaque 532 T cells. Not only were T cells prominent in atherosclerotic lesions, but they were overall 533 more activated, differentiated, and exhausted compared to their blood counterparts. In 534 particular, we identified an enrichment of 4 specific effector memory T cell subsets in 535 plaques that presented distinct activation and differentiation patterns indicating a 536 functional heterogeneity of the T cell immune compartment. Interestingly, one of these 537 subsets corresponded to a CD103+ tissue resident memory (TRM) CD8+ T cells, that have 538 previously been described in other non-lymphoid human tissues48, but never in the 539 atherosclerotic arterial wall. A key finding of our CyTOF analysis was that plaque T cells 540 expressed high levels of PD-1, a marker of T cell exhaustion reprogramming that is 541 upregulated upon T cell activation68-71, that was confirmed at the transcriptional level by 542 our CITE-seq analysis. We found that plaque T cells expressed a correspondent 543 exhaustion transcriptional signature (i.e. EOMES, LAG3) that parallel that of exhausted T
15
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
544 cells in the tumor microenvironment56. CITE-seq analysis also revealed the co-existence 545 of exhausted and activated T cell subsets within the same plaque, suggesting that highly 546 activated T cells may initiate an exhaustion reprogramming, the stepwise and progressive 547 loss of T cell functions, possibly sustained by chronic unresolving plaque inflammation55- 548 57,74,89. CyTOF analysis on plaque-derived immune cells also identified two main subsets 549 of macrophages, resembling the classically-activated M1 and alternatively-activated M2 550 phenotypes46,47. Interestingly, plaque macrophages were more resolved at the single-cell 551 transcriptional level in our CITE-seq analysis, indicating a diverse functional 552 heterogeneity of these cells in plaque. We found distinct macrophage subsets displaying 553 activated and pro-inflammatory functional states, inflammatory macrophages expressing 554 genes involved in lipid metabolism, a subset distinctively resembling foam cell functions, 555 and macrophages with unique anti-inflammatory transcriptional signatures. These data 556 highlight the functional specialization and adaptation of lesional macrophages to the pro- 557 inflammatory, lipid rich human plaque microenvironment that is not accurately reflected 558 by the M1 and M2-phenotypes resolved by canonical surface protein marker 559 analysis17,21,46.
560 Our combined analyses of CyTOF and scRNA-seq data underscored new innate and 561 adaptive immune dysregulations and alterations of key cell-cell communications in the 562 plaque microenvironment associated with clinical cardiovascular outcomes. We found 563 that ipsilateral plaques of symptomatic patients had significant adaptive dysregulations 564 consistent with an expansion of activated effector memory CD4+ T cell subset. 565 Interestingly, this change was not related to either plaque pathology according to the AHA 566 classification84 or clinical characteristics.
567 Single-cell transcriptomic analysis of plaque T cells highlighted more heterogeneous 568 distinctive alterations of both CD4+ and CD8+ T cells associated with recent clinical 569 events. Specifically, in ipsilateral plaques of symptomatic patients with recent stroke or 570 TIA, both T cell lineages presented gene expression signatures largely consistent with 571 activation, differentiation, and exhaustion. Conversely, T cells were overall activated in 572 plaques of asymptomatic patients. Cell-cell interaction analysis showed that these 573 functional states were tightly regulated by specific interactions with other T cells and 574 macrophages. Previous murine studies have shown that PD-1 signaling, a major regulator 575 of T-cell exhaustion, is required to modulate atherogenic responses of activated T cells 576 in the arterial wall and its inhibition results in aggravated atherosclerosis142. Our data that 577 T cell exhaustion is subset-specific and co-exists with activated T cells in ipsilateral 578 human plaques from symptomatic patients, suggests an unforeseen and more complex 579 role of exhaustion in human disease that may have important clinical implications. The 580 fundamental discovery that PD-1 inhibitors reverse anergy of tumor T cells has 581 fundamentally changed the therapeutic options and prognosis for patients with advanced 582 malignancy53,54,143. Our data that exhausted T cells expressing high levels of PD-1 exists
16
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
583 in atherosclerotic plaques suggest that off-target effects of PD-1 inhibitors may increase 584 T cell activation. Considering that T cell activation aggravates atherosclerosis21, treatment 585 with PD-1 inhibitors may have unforeseen consequences in cancer patients with 586 underlying atherosclerotic CV disease. Additionally, the higher exhaustion seen in 587 subsets of plaques of patients with recent cerebrovascular events, suggest that any 588 possible off-target effect may have higher consequences in patients with recent history of 589 stroke or TIA which are already at higher risk of recurrent ischemic events within 5 590 years144. Future cardio-oncology monitoring studies designed to investigate unrecognized 591 cardiovascular side effects of increasingly used immune checkpoint inhibitors143 would be 592 useful to test this hypothesis and help improve patient selection and cardiovascular 593 monitoring in the clinical setting.
594 Similar to the plaque T cells, macrophages infiltrating plaques of asymptomatic patients 595 presented an activated and pro-inflammatory phenotype. Conversely, lesional 596 macrophages from ipsilateral plaques of patients with recent CV events displayed a 597 transcriptional signatures more consistent with distinct pro-inflammatory and reparative 598 functions, including a proatherogenic subsets of cell involved in iron metabolisms derived 599 from the hemoglobin deposited in areas of intraplaque hemorrhage that characterize AHA 600 type VI complicated plaques47,84,116,117, the same plaque type that is exhibited by all SYM 601 patients in our cohort. We found that the specialized functions of lesional macrophages 602 in atherosclerotic plaques from symptomatic and asymptomatic patients were self- 603 regulated but also dependent on T cell ligand signals, indicating unique and highly 604 coordinated cross-communications between innate and adaptive immune responses at 605 the lesional site.
606 A key and unexpected finding of our study was the discovery that IL-1 signaling, which 607 was targeted in the CANTOS to reduce the CV risk of high risk post-MI patients 33, was 608 differentially regulated in plaques from symptomatic and asymptomatic patients. Plaque 609 macrophages of asymptomatic patients expressed higher levels of IL1B, and associated 610 IL-1 signaling compared to symptomatic patients. Additionally, we found that IL-1 did not 611 exclusively signal innate pro-inflammatory functions in macrophages but also in T cells, 612 suggesting a key role at the intersection of innate and adaptive immune responses by 613 sustaining effector functions of T cells in ASYM plaques. These data suggest that post- 614 MI patients in the CANTOS may have been less susceptible in their response to IL-1β 615 inhibition at the culprit plaque site, than high-risk patients without recent CV event.
616 In conclusion, by combining proteomics and transcriptomics single-cell analyses of 617 atherosclerotic plaques and blood of the same patients provides a first immune atlas of 618 human atherosclerosis and identified new specific innate and adaptive immune 619 dysregulations at the atherosclerotic site associated with clinical cerebrovascular events.
17
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
620 Our results strongly support the concept that an accurate selection of patients is required 621 to optimize treatment efficacy, and that new immunotherapies must be precisely tailored 622 to discrete immune molecular and cellular defects in subsets of patients32,33,145,146.
623
624 ACKNOWLEDGMENTS
625 We thank the Human Immune Monitoring Center (HIMC), in particular Oksana Mayovska, 626 Victor Guo, Xiaochen Ivy Qin, Hui Emily Xie, Manishkumar Patel, Melanie Davila, Brian 627 Lee, Shermineh Bradford, Laura Walker, and Kevin Tuballes. We thank Swathy Sajja and 628 Peik Sean Chong for their coordination efforts. We are grateful to Dr. Alice Kamphorst for 629 her critical review of this manuscript. We thank the Biorepository and Pathology Core of 630 the Icahn School of Medicine at Mount Sinai. This work utilized mass cytometry 631 instrumentation supported by NIH grant S10OD023547-01. This work was funded by NIH 632 grants K23HL111339, R03HL135289. C.G. was also funded by NIH grants 633 R21TR001739, and UH2TR002067 and partially supported by the American Heart 634 Association (14SFRN20490315). D.F. is supported by the NIH grant 5T32HL007824-20. 635 A.M. and Z.W. are supported by NIH grants U54-HL127624 (LINCS-DCIC) and U24- 636 CA224260 (IDG-KMC).
637
638 AUTHOR CONTRIBUTIONS
639 Conceptualization: C.G., M.M., and A.H.R.; methodology: C.G., A.H.R., E.D.A., N.F., 640 A.M., S.G., D.M.F. J.L.M.B.; software: E.D.A., N.F., Z.W., A.M.; formal analysis: A.H.R., 641 D.M.F., E.D.A., N.F., Z.W., A.M. investigations: D.M.F., A.H.R., A.C., L.A., N.K., R.R., 642 R.S., S.K-S. C.W., R.S. C.H.; Resources: J.R.L., C.P., N.M., C.F., A.J.A, J.M., P.F. and 643 A.M.; data curation: A.H.R., D.M.F, N.F., E.D.A., S.K-S, J.L.M.B.; Writing: C.G; revision 644 and editing: C.G., A.H.R., D.M.F., M.M., S.K-S, C.W., C. H., R.S., N.F.; data visualization: 645 C.G., D.M.F., N.F., A.H.R., A.C., and A.M; supervision: C.G., A.H.R., S.K-S and M.M.; 646 project administration: C.G.; funding acquisition: C.G.
647 DECLARATIONS OF INTERESTS
648 J.L.M.B is the founder and former CEO of Clinical Gene Networks (CGN) and receives 649 financial compensation as a consultant for CGN.
18
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
650 FIGURE LEGENDS
651 Figure 1. Study Overview. (a) Collection and processing of atherosclerotic tissue 652 (plaque) and paired blood from the same patient. (b) Comparison of immune cells in 653 atherosclerotic tissue and paired blood using CyTOF or CITE-seq analyses. (c) Single- 654 cell analysis of immune dysregulations associated with cerebrovascular events (i.e. 655 Stroke, TIA) in symptomatic (recent stroke) and asymptomatic (no recent stroke) patients 656 with carotid artery disease: a three tier approach. Analysis of: 1. single-cell proteomic 657 data from CYTOF, 2. Gene expression data from scRNA-seq, and 3. Cell-Cell 658 communications based on ligand and receptor interactions. ADT: Antibody-tag data from 659 CITE-seq; GEX: gene expression data from CITE-seq.
660 Figure 2. T Cells and Macrophages Dominate the Atherosclerotic Immune 661 Landscape. (a, b) MetaLouvain Clustering of CD45+ cells derived from blood and 662 atherosclerotic plaque tissue (n=15). Representative ViSNE plots of immune cells 663 displaying tissue type (a), or overlaid with color-coded immune cell populations (b). (c) 664 Heatmap of normalized, Z-scored protein expression of MetaCluster (MC) data, relating 665 the MC communities (left) to the expression of protein markers (top) and the annotated 666 cell types (right). (d) Bar chart of the relative frequency of immune cell types derived from 667 aggregated MC data. pDC, plasmacytoid dendritic cell; DC, dendritic cell. (e) Volcano plot 668 of the MC frequency fold change atherosclerotic plaque, color coded by the tissue 669 enrichment (plaque: blue, blood: pink). P values were calculated using a paired, two-sided 670 Student's t-Test and FDR calculation in R. (f) Scatter bar plots of population frequency 671 from MCs enriched in plaque tissue. Data were analyzed with the Wilcoxon test. Values 672 are mean ± SD **p<0.01, ***p<0.001.
673 Extended Figure 1. Definition of MetaClusters from the Discovery Cohort. (a) viSNE 674 plots displaying the protein marker overlaid in spectral colors. (b) Bar plot of MC 675 frequencies in blood (black) and plaque (white). Annotated MCs are shown on the right. 676 Statistical significance was determined using unpaired t-test with multiple comparisons 677 corrected by FDR=0.5%. *q<0.05, ***q<0.001, and ****q<0.0001, (c) Comparison of 678 population frequency of MCs enriched in blood. ***p<0.001, ****p<0.0001 by Wilcoxon 679 test. Values of all plots are mean ± SD.
680 Extended Figure 2. Tissue–specific Metaclustering Identifies Dysregulations in T 681 Cells. (a) MetaLouvain Clustering of CD45+ cells from atherosclerotic plaque tissue 682 (n=15). (b) Average population frequency of plaque-derived MCs. (c) Representative 683 viSNE plot depicting individual MC distribution. (d) viSNE plots of selected myeloid MC 684 surface markers. (e) Volcano plot of the fold change of MC frequencies in ASYM (left) 685 and SYM (right) atherosclerotic plaques. Statistical significance was determined using 686 unpaired t-test with multiple comparisons corrected by FDR=0.5%.
19
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
687 Figure 3. Diversity of the T-Cell Compartment in Human Atherosclerosis. (a-d) 688 MetaLouvain clustering of T cells from blood and plaque tissues (n=23). Representative 689 viSNE plot shows the distribution of blood and plaque (a), the expression of the T-cell 690 markers CD8 (b) and CD4 (c), and the distribution of the T cell MC communities (d). (e) 691 Clustergrammer heatmap showing protein marker expression (top) in each MC (left) and 692 the canonical annotation of these communities (right). Light gray dendrogram bars 693 indicate the clustering of MCs based on the cosine distance method in Clustergrammer. 694 (f) Bar chart of the relative frequency of T-cell MCs. (g) Volcano plot of the fold change of 695 MC frequency in blood (left) and atherosclerotic plaque (right). P values were calculated 696 using a paired, two-sided Student's t-Test and correspondent FDR calculated in R. (h) 697 Scatter bar plots of population frequency from MCs enriched in plaque tissue. (**p<0.01, 698 ***p<0.001, ****p<0.0001) Data were analyzed with the Wilcoxon test. Values are mean 699 ± SD.
700 Extended Figure 3. Definition of the T Cell Compartment from the Validation Cohort 701 (cohort 2). (a) viSNE plots of T-cell MC data displaying the protein marker overlaid in 702 spectral colors. (b) Bar plot of T-cell MC frequencies in blood and plaque. (c) Bar plot of 703 MC frequencies per patient (n=23). (d) Bar charts of normalized surface marker 704 expression of plaque-enriched MCs in blood and plaque. (e) Similarity matrix of T cell 705 MetaClusters based on the cosine distance method. Statistical significance for (b) and (d) 706 were determined using multiple unpaired Student’s t-tests corrected by the FDR=0.5%. 707 *q<0.05, **q<0.01, ***q<0.001, and ****q<0.0001.
708 Extended Figure 4. Unique T-Cell Signature in Atherosclerotic Plaques. Scatter bar 709 plots of normalized marker expression between blood and plaque in the CD8+ and CD4+ 710 T-cell MCs enriched in plaque (a-d), and comparison of MCs 10, 11, and 20 in plaque 711 tissue (e). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by paired two tailed Student’s t- 712 test (a-d), and One-Way ANOVA test with Bonferroni’s post-hoc correction (e). Values 713 are mean ± SD.
714 Figure 4. Simultaneous Epitope and Transcriptomic Analysis of Paired 715 Atherosclerotic Plaque and Blood: CITE-seq Analysis. (a, b) viSNE plots of cells 716 clustered from the Antibody Derived Tag (ADT) expression overlaid by (a) tissue, or (b) 717 immune cell type. (c) Frequency of immune cells in blood and plaque. (d-i) Analysis of the 718 gene expression data (GEX) from T cells (plaque n=921; blood n=1,652) and 719 macrophages (n=265). (d) Volcano plot of the top 5000 DEGs in T cells in plaque (purple) 720 or blood (pink). (e) Pathway analysis of T cell gene upregulated in blood or plaque. Bars 721 indicate the combined score [p value (Fisher’s exact test) multiplied by the Z-score of the 722 deviation from the expected rank] from Enrichr. (f) Heatmap of hierarchically clustered top 723 50 variable genes across all T cells in plaque and blood. Rows: z-scored gene expression 724 values; columns: individual cell barcode. Identified cell clusters are numbered and color- 725 coded. Bottom: cluster enrichment in blood (pink) or plaque (purple) and correspondent
20
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
726 p values. Boxes list key genes in correspondent clusters. p values,**p<0.01, ***p<0.0001, 727 ****p<0.0001. (g, i) Canonical signaling pathway analysis of the top 5000 DEGs per 728 cluster corresponding to T cells (g) and Macrophages (i).
729 Extended Figure 5. Analysis of the CITE-seq ADT Expression. viSNE plots of T cells 730 clustered using the ADT data, and overlaid by (a) tissue type, or expression of canonical 731 (c) or functional (d) T cell markers. (b) Frequency of T cell subtypes per tissue. (e) 732 Functional surface marker expression in individual cells per tissue type. P values were 733 determined using the Mann-Whitney test, lines indicate average expression values. (f, g) 734 viSNE plots of plaque cells overlaid to indicate macrophage population (f), or expression 735 of macrophage markers (g). (h) viSNE plot of clustered macrophages overlaid with CD206 736 expression. (i) macrophage population gated for expression of CD206 and CD64.
737 Figure 5. CD4+ T Cell Dysregulations Associated with Cerebrovascular Events. (a) 738 Patients by clinical association. (b) T cell MC frequencies in plaque from the validation 739 cohort stratified by asymptomatic (ASYM, n=14) and symptomatic (SYM, n=9) patients. 740 (c) Volcano plot of T cell MCs enriched in SYM (red), or ASYM (blue) patients. P values 741 were calculated using multiple t-Test with an FDR=0.5% (Benjamini, Krieger and Yekuteli 742 method) (d) Frequency of MC12 per each patient. ****q<0.00001. Values are mean ± SD. 743 (e) AHA plaque type classification in SYM and ASYM plaques of cohort 2 (n=23). (f) 744 Assessment of T cell MCs per (f) clinical phenotype in type VI plaques from either SYM 745 and ASYM patients, or (g) per plaque type (type VI vs. type IV) in all 23 patients. (h-k) 746 Single-cell gene expression data of CD4+ T cells (n=1,200). (h) Top 5000 Differentially 747 expressed genes across plaques of SYM and ASYM patients. (i). Signaling pathway 748 analysis of the upregulated genes of CD4+ T cells from SYM or ASYM. (j) Unbiased 749 hierarchical clustering of the top 100 variable genes across all CD4 T Cells. Column 750 categories from top to bottom indicate cluster number, cell cluster enrichment in SYM/ 751 ASYM, and cells from SYM (red) and ASYM (blue) subjects. (k) Canonical signaling 752 pathway analysis of SYM and ASYM enriched clusters.
753 Figure 6. Transcriptional Dysregulations Associated with Cerebrovascular Events 754 in CD8+ T Cells and Macrophages. Differential gene expression of the top 5000 755 differentially expressed genes (DEGs) across SYM and ASYM plaques (a) CD8+ T Cells 756 (n=1,747) and (f) macrophages (n=747) was calculated using the Welch’s T-test using 757 UMI normalized gene expression data. Enrichr signaling pathway analysis of CD8+ T cells 758 (b, c) and macrophages (g, h) on the top 5000 DEGs between SYM or ASYM patients. 759 Hierarchically clustered cells based on the Z-scored expression of the top 100 variable
21
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
760 genes across all CD8+ T cells (d) and macrophages (i). Canonical pathway analysis of 761 SYM and ASYM enriched clusters in CD8+ T cells (e) or Macrophages (j).
762 Figure 7. Cell-Cell Interactions Associated with Cerebrovascular Events. Ligand- 763 Receptor interactions involving the top 30 SYM/ASYM differentially regulated 764 ligands based on absolute value sum of log2 fold change across ligands and visualized 765 using Clustergrammer. (a) T cell-T cell, (b) T cell-Macrophage and Macrophage-T cell, 766 and (c) Macrophage-Macrophage interactions are presented as (ligand-receptor). 767 Heatmaps represent the Log2 fold change of the interaction score between SYM and 768 ASYM. Log2 foldchange values were clipped at +/-10. (d) Visual summary of the key 769 innate and adaptive immune functional alterations and main differences in cell-cell 770 communications seen in plaques from ASYM and SYM patients.
771 Extended Figure 6. All Protein-Ligand Interactions. (a) Circos plots of the significant 772 ligand-receptor interactions between cell types, mediated by CD4+ T cells (top row), CD8+ 773 T Cells (middle row), or macrophages (bottom row). (b) Venn diagrams of ligand-receptor 774 pairs from the top 5000 genes (>0.5 Log2 fold change) show unique and overlapping 775 paired between cell-types. Gene Ontology terms were identified for each group using 776 Enrichr.
777
22
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
778 METHODS
779 Human Specimens and Patient Clinical Characteristics
780 Forty-six patients undergoing carotid endarterectomy at the Mount Sinai Hospital were 781 enrolled in an ongoing clinical study approved by the Institutional Review Board of the 782 Icahn School of Medicine at Mount Sinai (IRB 11-01427). Eligible subjects gave informed 783 written consent. The exclusion criteria were current infection, autoimmune disease, active 784 or recurrent cancer, severe renal failure requiring dialysis, peripheral arterial occlusive 785 disease causing pain at rest. Peripheral venous blood and atherosclerotic specimens 786 were obtained from each patient.
787 Supplementary Table 1 summarizes the demographic and clinical characteristics of the 788 patients. There were 29 men and 17 women with a mean age of 72.9±8.7 years (median 789 73.0 years). Symptomatic patients were defined as having had a cerebral ischemic event 790 (stroke, TIA, amaurosis fugax) ipsilateral to the collected plaque within 6 months 791 according to validated criteria157,158. Asymptomatic patients had no ischemic events within 792 6 months before surgery.
793 Sample Collection
794 Fasting blood was collected preoperatively into tubes containing ACD Solution A as anti- 795 coagulant (BD Vacutainer, 364606). Carotid atherosclerotic tissue specimens were 796 obtained at the bifurcation point and placed immediately in Dulbecco’s modified Eagle’s 797 medium (DMEM, Corning, 10-013-CV) supplemented with 10% fetal bovine serum (FBS, 798 Gibco, 10082-147) on ice.
799 Isolation of PBMCs
800 Blood was processed within 2 hr of collection and spun at 1500 rpm for 10 min using an 801 Eppendorf A-4-62 rotor. Plasma was collected, and the cell fraction was diluted 1:2 with 802 phosphate-buffered saline (PBS, Corning, 1-031-CV), placed on Ficoll-Paque Plus (ratio 803 7:3) (GE Healthcare Life Sciences, 17144003), and centrifuged for 20 min at 1800 rpm at 804 room temperature. After centrifugation, the PBMC layer was collected and washed twice 805 in PBS at room temperature.
806 Cell Dissociation from Atherosclerotic Plaques
807 Plaque specimens were processed within 1 hr after surgery. Each specimen was washed 808 extensively in DMEM, weighed, and digested as described 168 at 37° C for 1 hr in 10 ml 809 of DMEM containing 10% fetal bovine serum; collagenase type IV (Sigma, C5138) at a 810 final concentration of 1 mg/ml; and DNase (Sigma, DN25), hyaluronidase (Sigma, 811 H3506), collagenase type XI (Sigma, C7657), and collagenase type II (Sigma, C6885), 812 each at a final concentration of 0.3 mg/ml. The mixture was filtered consecutively through
23
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
813 70 μm and 40 μm cell strainers, washed twice in PBS, and centrifuged at 300 g for 8 min. 814 Dead cells were removed with a kit (Miltenyi Biotec, 130-090-101, or Stem Cell 815 Technologies, 17899) according to the manufacturer’s instructions. Cells were counted 816 with an automatic cell counter.
817 Preparation of Paired Samples for CyTOF
818 Cells isolated from plaque specimens and paired PBMCs from the same patient were 819 barcoded using a CD45-based approach and optimized for scarce clinical samples159, 820 pooled, and stained with antibody panels. All antibodies were validated at the Human 821 Immune Monitoring Center of the Icahn School of Medicine at Mount Sinai. Antibodies 822 were either purchased pre-conjugated from Fluidigm, or purchased purified and 823 conjugated in-house using MaxPar X8 Polymer Kits (Fluidigm) according to the 824 manufacturer’s instructions. The samples were then washed and stained with cisplatin- 825 195Pt or Intercalator Rh103 (Fluidigm, 201064 and 201103A) as a viability dye 160, 826 washed, fixed, and permeabilized with BD Perm/Wash and Cytofix/Cytoperm buffers (BD 827 Biosciences, 51-2090KZ and 51-2091KZ), and stored in freshly diluted 2% formaldehyde 828 (Electron Microscopy Sciences) in PBS containing 0.125 nM Iridium 191/193 intercalator 829 (Fluidigm, 201192) until acquisition. See Supplementary Table 3 for list of antibodies used 830 in this study. A limitation of our T cell focused panel consisted in the lack of anti-foxp-3 831 antibody, which limited the ability of dissecting the Tregs compartment in human 832 atherosclerosis.
833 Acquisition and Processing of CyTOF Data
834 CyTOF data were acquired with a CyTOF2 using a SuperSampler fluidics system 835 (Victorian Airships) at an event rate of <400 events per second and normalized with Helios 836 normalizer software (Fluidigm). Barcoded samples were deconvoluted and cross-sample 837 doublets were filtered using a Matlab-based debarcoding application44. Data were 838 uploaded to Cytobank (https://mtsinai.cytobank.org; Cytobank, Menlo Park, CA) for 839 analysis and visualization.
840 Preprocessing of CYTOF Data
841 CYTOF data were processed with Cytobank to sequentially remove calibration beads, 842 dead cells, debris, and barcodes for CD45+ plaque and CD45+ blood cells. For Cohort 1, 843 we analyzed 41,222 plaque cells with an average of 2,748 cells per sample. For Cohort 844 2, we analyzed 49,088 plaque cells with an average of 2,134 cells per sample. For the T- 845 cell MC analysis (Figure 3) and manual gating analysis (Supplementary Data to Figure 846 3), T cells were considered as CD3+CD14loCD16loCD56loCD19loCD64lo, and gated as 847 such. For the manual gating analysis, T cells were further stratified as CD8+ or CD4+. 848 Sub-classifications were defined as follows: naïve (CD45RA+CCR7+) CM (CD45RA– 849 CCR7+), EM (CD45RA–CCR7–), EMRA (CD45RA+CCR7–), T regulatory cells
24
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
850 (CD4+CD127loCD25+CCR4+), Th1 (CD4+CXCR3hiCCR6-), Th2 (CD4+CXCR3loCCR6+), 851 and Th17 (CD4+CXCR3loCCR6hi) (Supplementary Table 4)53.
852 Visualizing and Clustering CyTOF Data
853 CyTOF data were analyzed and visualized with Cytobank. Tissue-associated 854 inflammatory cells and matched PBMCs were analyzed with an R-based semi-automated 855 pipeline developed at the Human Immune Monitoring Center of the Icahn School of 856 Medicine at Mount Sinai. This pipeline uses viSNE43 as a dimensionality-reducing 857 visualization tool and Phenograph44 as an automated clustering algorithm. Stratifying 858 signatures between the sample groups were identified by calculating the fold change 859 between mean cluster abundances across groups. P values were calculated using a 860 paired, two-sided Student's t-Test using the FDR correction of the R package. Fold 861 change and p values were visualized using a volcano plot (See Figures 2 and 3, and 862 Supplementary Table 2). Clusters were manually annotated according to canonical 863 patterns of marker expression, and Marker expression values were Z-scored and 864 visualized with Clustergrammer (http://amp.pharm.mssm.edu/clustergrammer/). 865 Statistical analysis of cell population frequencies and marker expression using the 866 multiple t test with FDR correction, one-Way ANOVA test with Bonferroni’s post-hoc 867 correction, two tailed Student’s t-test, Wilcoxon test and Spearman rank correlation were 868 done with GraphPad Prism 8.0.
869 Characterization of Plaque Type
870 The carotid plaque specimens were fixed in formalin (Thermo Scientific, 9990244), 871 embedded in paraffin at the Pathology Biorepository at Mount Sinai Hospital, cut into 5- 872 μm-thick sections, placed on charged glass slides, and stained with Masson’s trichrome 873 stain as recommended by the manufacture (American Master Tech KTMTRPT). The 874 images were used to evaluate the plaque type according to the American Heart 875 Association Classifications84.
876 TCR Variable Beta Chain Sequencing and Statistical Analysis
877 The CDR3 variable regions of the TCR chain were sequenced with the ImmunoSEQ 878 Assay (Adaptive Biotechnologies, Seattle, WA). Genomic DNA was extracted from 879 formalin-fixed, paraffin-embedded tissue samples and amplified by bias-controlled 880 multiplex PCR, followed by high-throughput sequencing. Sequences were collapsed and 881 filtered to identify and quantitate the absolute abundance of each unique TCR CDR3 882 region161. Clonality was defined as 1 – Peilou’s eveness162 and calculated as described41. 883 Clonality values range from 0 to 1 and describe the shape of the frequency distribution: 884 clonality values approaching 0 indicate a very even distribution of frequencies, whereas 885 values approaching 1 indicate an increasingly asymmetric distribution in which a few 886 clones are present at high frequencies. Correlations between clonality and CyTOF data
25
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
887 were assessed with Spearman’s rank correlation test after running a Shapiro-Wilk test for 888 normality. Statistical analyses were done with GraphPad Prism 8.0.
889 Cellular Indexing of Transcriptomes and Epitopes by sequencing (CITE-seq) and 890 Single-Cell RNA sequencing (scRNA-seq) Sample Preparation
891 A protocol adapted from Stoeckius et al 201742 was used to perform CITE-seq . All 892 antibodies used for CITE-seq were conjugated at the Human Immune Monitoring Core 893 at Mount Sinai using ThunderLink Plus conjugation kits (Expedeon, Cat#425-0000), 894 following manufacturer’s instructions. Single cell suspensions of atherosclerotic plaque 895 and PBMCs were enriched for immune cells using a CD45+ enrichment kit (Miltenyi, 130- 896 045-801), and equal proportions of CD45+ enriched blood and plaque samples were 897 barcoded and immuno-stained on ice with the CITE-seq antibody panel (See 898 Supplemental Table 3).
899 The single-cell suspensions of CITE-seq and scRNA-seq samples were converted to 900 barcoded scRNA-seq libraries by using the Chromium Single Cell 3’ Library, Gel Bead & 901 Multiplex Kit, and Chip Kit (10x Genomics) 163,164. Chromium Single Cell 3’ v2 Reagent 902 (10X Genomics, 120237) kit was used to prepare single-cell RNA libraries according to 903 the manufacturer’s instructions. DNA quality was measured with an Agilent Bioanalyzer 904 and quantity was assessed with the Qubit fluorometric assay (Invitrogen). Both CITE-seq 905 and scRNA-seq used Cell Ranger Single-Cell Software Suite (version 2.1.1, 10X 906 Genomics) to quality control the single-cell expression data. For CITE-seq, the Cell 907 Ranger outputs estimated 3,573 cells with 50,701 mean reads per cell, and 3,879 median 908 UMI counts per cell. For the scRNA-seq data a total of 7,169 cells were analyzed and 909 averages of the Cell Ranger outputs from the n=6 patients an average of 111,670 mean 910 reads per cell, and 2,929 median UMI counts per cell. The filtered gene-barcode matrix 911 output from Cell Ranger, which removes barcodes with a low unique molecular identifier 912 (UMI) count, was analyzed in further analyses using Clustergrammer2 913 (https://github.com/ismms-himc/clustergrammer2)165.
914 CITE-seq Expression Analysis
915 PBMC and Plaque-derived cells were identified by de-hash tagging (following the cell 916 hashing technique from Stoeckius et al 42). The surface marker expression, from Antibody 917 derived tags (ADT) data from a 21 antibody panel (see Supplementary table 3) was 918 analyzed using Cytobank. First, doublets between PBMCs and plaque were removed by 919 exclusion of overlapping barcodes. Second, all individual cells were clustered using the 920 t-SNE algorithm based on expression of all 21 markers. Third, immune cell types were 921 determined based on the expression of canonical markers, and were thus gated directly 922 from the viSNE plot. Correlations of the gene and surface marker expression can be found 923 in Supplementary Figure 9.
26
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
924 Ribosomal and mitochondrial genes were dropped from the gene expression analysis. 925 Gene expression signatures of the atherosclerotic plaque and blood were derived from 926 the cell types obtained from the ADT manual gating (CD4 T cells, CD8 T cells, B Cells, 927 Macrophages, Monocytes, NK Cells, NKT Cells, Plasma Cells, CD1c DCs, pDCs), and 928 used to predict cell type across the cohort of single-cell RNA-seq of atherosclerotic 929 plaques. The gene expression data corresponding to the cells of gated populations was 930 used for further analysis by Clustergrammer2, Enrichr, and Ingenuity Pathway Analysis 931 (IPA).
932 Batch Correction of Single-Cell RNA-seq data
933 Plaque gene expression from six subjects was combined, and batch normalized using 934 mutual nearest neighbors (MNN) (https://github.com/chriscainx/mnnpy)166. First, gene 935 expression data from all subjects was combined into a single dataset and suspected red 936 blood cells were dropped (e.g. cells with high hemoglobin expression). Second, gene 937 expression data was Arcsinh normalized and filtered to keep the top 10,000 genes based 938 on variance across the aggregated dataset as described166. Third, MNN batch correction 939 was performed using the one of the samples as a reference (See Supplementary Figure 940 10).
941 Cell Type Prediction of Single-Cell RNA-seq using CITE-seq Gene Expression 942 Signatures
943 Cell type from single-cell gene expression data across six subjects was predicted using 944 a gene expression signature method. First, 1,641 single cells from the CITE-seq sample 945 were assigned a cell type based manual annotation of single cell t-SNE clustering in 946 surface marker space (cell types include: CD4+ T cells, CD8+ T cells, B Cells, 947 Macrophages, Monocytes, NK Cells, NKT Cells, Plasma Cells, CD1c DCs, pDCs). 948 Second, cell type gene expression signatures were generated from the MNN batch 949 corrected CITE-seq gene expression data. Cell type signatures were calculated by Z- 950 scoring gene expression levels across all 7,169 cells, averaging gene expression levels 951 across cell types, and keeping the top 200 differentially expressed genes for each cell 952 type (using the Welch’s T-test). Third, cell type is predicted for all 7,169 cells across all 953 samples by assigning each cell to the closest cell type signature based on cosine distance 954 (gene expression data is Z-scored before distance calculation). Cell type prediction of the 955 CITE-seq dataset using the MNN-batch corrected gene expression signatures matched 956 previous ADT-based assignments 81% of the time.
957 Single-Cell RNA-seq and CITE-seq Gene Expression and Pathway analysis
958 Single-cell gene expression data was analyzed for differential genes expression across 959 tissue types (CITE-seq), or across patient types (using aggregated scRNA-seq). 960 Differential gene expression was calculated using UMI normalized gene expression data.
27
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
961 When differential expression was calculated using data from multiple subject samples, 962 which had been subject to MNN batch correction, UMI normalized gene expression data 963 was used in place of MNN normalized data as described166. Differential expression was 964 calculated using the Welch’s T-test with multiple hypothesis correction using Benjamini- 965 Hochberg correction. Single-cells were hierarchically clustered using the top genes based 966 on variance to identify subclusters within specific cell types. The hierarchical clustering 967 was done using cosine distance with average linkage. Clusters were defined by assigning 968 groups at a manually chosen level in the dendrogram. Differentially expressed genes 969 across cell clusters were identified with the Welch t test with unequal variance; multiple 970 hypothesis correction was done with the Benjamini-Hochberg correction in the SciPy 971 Stats package, (https://docs.scipy.org/doc/scipy/reference/index.html). For our canonical 972 pathway analysis, we analyzed cell clusters containing at least 25 cells using Ingenuity 973 Pathway Analysis (QIAGEN Inc.). The top 5000 DEGs corresponding to the identified cell 974 clusters were investigated with the gene enrichment analysis tool Enrichr167,168. For CITE- 975 seq, the Enrichr pathway analysis terms were chosen from the Reactome library. 976 Reactome, Biocarta, NCI Nature, and GO Biological Functions libraries were used to 977 analyze the scRNA-seq data. Terms from these libraries were plotted using the combined 978 score, which represents the p value (Fisher’s exact test) multiplied by the Z-score of the 979 deviation from the expected rank.
980 Cell Type Ligand-Receptor Interaction
981 Potential ligand-receptor interaction between one set of ligand expressing cells and 982 another set of receptor expressing cells is calculated as the average of the product of 983 ligand and receptor expression (respectively from set one and two) across all single-cell 984 pairs:
1 985 퐼 = ∑푛 푙 ⋅ ∑푚 푟 ( ) 푖 푖 푗 푗 푚∗푛 986 퐼= interaction score between ligand expressing cells in set one and receptor expressing 987 cells in set two 988 푙푖= ligand expression of cell 푖 in cell set one 989 푟푗= receptor expression of cell 푗 in cell set two 990 푛= number of cells in set one 991 푚= number of cells in set two 992 993 Prior knowledge ligand and receptor interactions were obtained from Ramilowski et al169. 994 Cell types were obtained from the 7,169 cell MNN batch corrected dataset and gene 995 expression levels were calculated using UMI normalized data. Potential ligand-receptor 996 interactions were calculated for three cell types (CD4 T Cells, CD8 T Cells, and 997 Macrophages) resulting in nine cell-type/cell-type interaction scores. Visualization of the 998 interactions was performed using CIRCOS plots (http://mkweb.bcgsc.ca/tableviewer/),
28
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
999 and Clustergrammer. Comparison of receptor-ligand pairs between cell types were 1000 analyzed visualize using a Venn diagram generate from 1001 (http://bioinformatics.psb.ugent.be/webtools/Venn/).
1002 Differential regulation of ligand-receptor interactions between symptomatic and 1003 asymptomatic cells was identified by comparing single-cell/single-cell ligand-receptor 1004 interaction scores for symptomatic and asymptomatic cells. Fold change of symptomatic 1005 vs asymptomatic interaction scores were calculated using the average interaction scores 1006 from symptomatic and asymptomatic. The significance of differential regulation was 1007 assessed by comparing the distributions of cell-cell ligand-receptor interaction scores 1008 from symptomatic and asymptomatic cells using the Welch’s T-test. Symptomatic vs 1009 Asymptomatic differentially regulated interactions were identified based on the following 1010 4 criteria: (1) one percent of cells must express the ligand and receptor of interest in either 1011 symptomatic or asymptomatic cells, (2) the interaction score of symptomatic or 1012 asymptomatic cells must be at least half of the baseline interaction score (defined as the 1013 interaction score between all cells in the dataset), (3) the log2 fold change of the 1014 interaction score (symptomatic vs asymptomatic) must be greater than 0.5, (4) the P- 1015 value of the Welch’s T-test with multiple hypothesis correction (Benjamini-Hochberg) 1016 must be less than 0.05. For visualization, log2 Fold Change values were clipped at +/-10.
1017 Data and Software Availability
1018 The data discussed in this publication will be deposited using Kaggle Datasets and 1019 be publicly available.
1020
29
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1021 REFERENCES 1022 1. Lusis, A.J. Atherosclerosis. Nature 407, 233-241 (2000). 1023 2. Jawien, J. The role of an experimental model of atherosclerosis: apoE-knockout mice 1024 in developing new drugs against atherogenesis. Current pharmaceutical 1025 biotechnology 13, 2435-2439 (2012). 1026 3. von Scheidt, M., et al. Applications and Limitations of Mouse Models for 1027 Understanding Human Atherosclerosis. Cell Metab 25, 248-261 (2017). 1028 4. van der Heiden, K., Hoogendoorn, A., Daemen, M.J. & Gijsen, F.J. Animal models for 1029 plaque rupture: a biomechanical assessment. Thrombosis and haemostasis 1030 115(2015). 1031 5. Stone, G.W., Mintz, G.S. & Virmani, R. Vulnerable Plaques, Vulnerable Patients, and 1032 Intravascular Imaging. Journal of the American College of Cardiology 72, 2022-2026 1033 (2018). 1034 6. Narula, J., et al. Histopathologic characteristics of atherosclerotic coronary disease 1035 and implications of the findings for the invasive and noninvasive detection of 1036 vulnerable plaques. Journal of the American College of Cardiology 61, 1041-1051 1037 (2013). 1038 7. Finn, A.V., Nakano, M., Narula, J., Kolodgie, F.D. & Virmani, R. Concept of 1039 vulnerable/unstable plaque. Arteriosclerosis, thrombosis, and vascular biology 30, 1040 1282-1292 (2010). 1041 8. Virmani, R., Burke, A.P., Farb, A. & Kolodgie, F.D. Pathology of the vulnerable plaque. 1042 Journal of the American College of Cardiology 47, C13-18 (2006). 1043 9. Kolodgie, F.D., et al. Pathologic assessment of the vulnerable human coronary 1044 plaque. Heart 90, 1385-1391 (2004). 1045 10. Schaar, J.A., et al. Terminology for high-risk and vulnerable coronary artery plaques. 1046 Report of a meeting on the vulnerable plaque, June 17 and 18, 2003, Santorini, 1047 Greece. European heart journal 25, 1077-1082 (2004). 1048 11. Virmani, R., Burke, A.P., Kolodgie, F.D. & Farb, A. Vulnerable plaque: the pathology of 1049 unstable coronary lesions. J Interv Cardiol 15, 439-446 (2002). 1050 12. Kolodgie, F.D., et al. The thin-cap fibroatheroma: a type of vulnerable plaque: the 1051 major precursor lesion to acute coronary syndromes. Curr Opin Cardiol 16, 285-292 1052 (2001). 1053 13. Quillard, T., Franck, G., Mawson, T., Folco, E. & Libby, P. Mechanisms of erosion of 1054 atherosclerotic plaques. Curr Opin Lipidol 28, 434-441 (2017). 1055 14. Franck, G., et al. Flow Perturbation Mediates Neutrophil Recruitment and 1056 Potentiates Endothelial Injury via TLR2 in Mice: Implications for Superficial Erosion. 1057 Circulation research 121, 31-42 (2017). 1058 15. Libby, P. Superficial erosion and the precision management of acute coronary 1059 syndromes: not one-size-fits-all. European heart journal 38, 801-803 (2017). 1060 16. Moreno, P.R., et al. Macrophage infiltration in acute coronary syndromes. 1061 Implications for plaque rupture. Circulation 90, 775-778 (1994). 1062 17. Moore, K.J. & Tabas, I. Macrophages in the pathogenesis of atherosclerosis. Cell 145, 1063 341-355 (2011). 1064 18. Farb, A., et al. Coronary plaque erosion without rupture into a lipid core. A frequent 1065 cause of coronary thrombosis in sudden coronary death. Circulation 93, 1354-1363 1066 (1996). 30
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1067 19. Grivel, J.C., et al. Activation of T lymphocytes in atherosclerotic plaques. 1068 Arteriosclerosis, thrombosis, and vascular biology 31, 2929-2937 (2011). 1069 20. Jonasson, L., Holm, J., Skalli, O., Bondjers, G. & Hansson, G.K. Regional accumulations 1070 of T cells, macrophages, and smooth muscle cells in the human atherosclerotic 1071 plaque. Arteriosclerosis 6, 131-138 (1986). 1072 21. Tabas, I. & Lichtman, A.H. Monocyte-Macrophages and T Cells in Atherosclerosis. 1073 Immunity 47, 621-634 (2017). 1074 22. Kyaw, T., et al. Cytotoxic and proinflammatory CD8+ T lymphocytes promote 1075 development of vulnerable atherosclerotic plaques in apoE-deficient mice. 1076 Circulation 127, 1028-1039 (2013). 1077 23. Cochain, C. & Zernecke, A. Protective and pathogenic roles of CD8(+) T cells in 1078 atherosclerosis. Basic Res Cardiol 111, 71 (2016). 1079 24. Dumitriu, I.E., et al. High levels of costimulatory receptors OX40 and 4-1BB 1080 characterize CD4+CD28null T cells in patients with acute coronary syndrome. 1081 Circulation research 110, 857-869 (2012). 1082 25. Giubilato, S., et al. Expansion of CD4+CD28null T-lymphocytes in diabetic patients: 1083 exploring new pathogenetic mechanisms of increased cardiovascular risk in 1084 diabetes mellitus. European heart journal 32, 1214-1226 (2011). 1085 26. Guo, M., et al. miR-146a in PBMCs modulates Th1 function in patients with acute 1086 coronary syndrome. Immunology and cell biology 88, 555-564 (2010). 1087 27. Liuzzo, G., et al. Identification of unique adaptive immune system signature in acute 1088 coronary syndromes. International journal of cardiology 168, 564-567 (2013). 1089 28. Methe, H., et al. Enhanced T-helper-1 lymphocyte activation patterns in acute 1090 coronary syndromes. Journal of the American College of Cardiology 45, 1939-1945 1091 (2005). 1092 29. Shantsila, E. & Lip, G.Y. Monocytes in acute coronary syndromes. Arteriosclerosis, 1093 thrombosis, and vascular biology 29, 1433-1438 (2009). 1094 30. Ghattas, A., Griffiths, H.R., Devitt, A., Lip, G.Y. & Shantsila, E. Monocytes in coronary 1095 artery disease and atherosclerosis: where are we now? Journal of the American 1096 College of Cardiology 62, 1541-1551 (2013). 1097 31. Leers, M.P.G., et al. Intermediate and nonclassical monocytes show heterogeneity in 1098 patients with different types of acute coronary syndrome. Cytometry. Part A : the 1099 journal of the International Society for Analytical Cytology 91, 1059-1067 (2017). 1100 32. Ridker, P.M., Thuren, T., Zalewski, A. & Libby, P. Interleukin-1beta inhibition and the 1101 prevention of recurrent cardiovascular events: rationale and design of the 1102 Canakinumab Anti-inflammatory Thrombosis Outcomes Study (CANTOS). American 1103 heart journal 162, 597-605 (2011). 1104 33. Ridker, P.M., et al. Antiinflammatory Therapy with Canakinumab for Atherosclerotic 1105 Disease. The New England journal of medicine 377, 1119-1131 (2017). 1106 34. Ridker, P.M., et al. Low-Dose Methotrexate for the Prevention of Atherosclerotic 1107 Events. The New England journal of medicine 380, 752-762 (2019). 1108 35. Baigent, C., et al. Efficacy and safety of cholesterol-lowering treatment: prospective 1109 meta-analysis of data from 90,056 participants in 14 randomised trials of statins. 1110 Lancet 366, 1267-1278 (2005).
31
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1111 36. Cholesterol Treatment Trialists, C., et al. Lack of effect of lowering LDL cholesterol 1112 on cancer: meta-analysis of individual data from 175,000 people in 27 randomised 1113 trials of statin therapy. PloS one 7, e29849 (2012). 1114 37. Cholesterol Treatment Trialists, C., et al. The effects of lowering LDL cholesterol 1115 with statin therapy in people at low risk of vascular disease: meta-analysis of 1116 individual data from 27 randomised trials. Lancet 380, 581-590 (2012). 1117 38. Winkels, H., et al. Atlas of the Immune Cell Repertoire in Mouse Atherosclerosis 1118 Defined by Single-Cell RNA-Sequencing and Mass Cytometry. Circulation research 1119 122, 1675-1688 (2018). 1120 39. Cole, J.E., et al. Immune cell census in murine atherosclerosis: cytometry by time of 1121 flight illuminates vascular myeloid cell diversity. Cardiovasc Res 114, 1360-1371 1122 (2018). 1123 40. Cochain, C., et al. Single-Cell RNA-Seq Reveals the Transcriptional Landscape and 1124 Heterogeneity of Aortic Macrophages in Murine Atherosclerosis. Circulation 1125 research (2018). 1126 41. Lavin, Y., et al. Innate Immune Landscape in Early Lung Adenocarcinoma by Paired 1127 Single-Cell Analyses. Cell 169, 750-765 e717 (2017). 1128 42. Stoeckius, M., et al. Simultaneous epitope and transcriptome measurement in single 1129 cells. Nat Methods 14, 865-868 (2017). 1130 43. Amir el, A.D., et al. viSNE enables visualization of high dimensional single-cell data 1131 and reveals phenotypic heterogeneity of leukemia. Nature biotechnology 31, 545- 1132 552 (2013). 1133 44. Levine, J.H., et al. Data-Driven Phenotypic Dissection of AML Reveals Progenitor-like 1134 Cells that Correlate with Prognosis. Cell 162, 184-197 (2015). 1135 45. Blondel, V.D., Guillaume, J.L., Lambiotte, R. & Lefebvre, E. Fast unfolding of 1136 communities in large networks. J Stat Mech-Theory E (2008). 1137 46. Moore, K.J., Sheedy, F.J. & Fisher, E.A. Macrophages in atherosclerosis: a dynamic 1138 balance. Nature reviews. Immunology 13, 709-721 (2013). 1139 47. Guo, L., et al. CD163+ macrophages promote angiogenesis and vascular permeability 1140 accompanied by inflammation in atherosclerosis. J Clin Invest 128, 1106-1124 1141 (2018). 1142 48. Kumar, B.V., et al. Human Tissue-Resident Memory T Cells Are Defined by Core 1143 Transcriptional and Functional Signatures in Lymphoid and Mucosal Sites. Cell Rep 1144 20, 2921-2934 (2017). 1145 49. Cheuk, S., et al. CD49a Expression Defines Tissue-Resident CD8(+) T Cells Poised for 1146 Cytotoxic Function in Human Skin. Immunity 46, 287-300 (2017). 1147 50. Hombrink, P., et al. Programs for the persistence, vigilance and control of human 1148 CD8(+) lung-resident memory T cells. Nature immunology 17, 1467-1478 (2016). 1149 51. Contento, R.L., et al. CXCR4–CCR5: A couple modulating T cell functions. Proceedings 1150 of the National Academy of Sciences 105, 10101 (2008). 1151 52. Strioga, M., Pasukoniene, V. & Characiejus, D. CD8+ CD28- and CD8+ CD57+ T cells 1152 and their role in health and disease. Immunology 134, 17-32 (2011). 1153 53. Kamphorst, A.O., et al. Rescue of exhausted CD8 T cells by PD-1-targeted therapies is 1154 CD28-dependent. Science 355, 1423-1427 (2017).
32
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1155 54. Kamphorst, A.O., et al. Proliferation of PD-1+ CD8 T cells in peripheral blood after 1156 PD-1-targeted therapy in lung cancer patients. Proceedings of the National Academy 1157 of Sciences of the United States of America 114, 4993-4998 (2017). 1158 55. Wherry, E.J. T cell exhaustion. Nature immunology 12, 492-499 (2011). 1159 56. Wherry, E.J. & Kurachi, M. Molecular and cellular insights into T cell exhaustion. 1160 Nature reviews. Immunology 15, 486-499 (2015). 1161 57. Yi, J.S., Cox, M.A. & Zajac, A.J. T-cell exhaustion: characteristics, causes and 1162 conversion. Immunology 129, 474-481 (2010). 1163 58. Hansson, G.K., Libby, P. & Tabas, I. Inflammation and plaque vulnerability. Journal of 1164 internal medicine 278, 483-493 (2015). 1165 59. Tabas, I. Apoptosis and plaque destabilization in atherosclerosis: the role of 1166 macrophage apoptosis induced by cholesterol. Cell death and differentiation 11 1167 Suppl 1, S12-16 (2004). 1168 60. Teng, M.W., et al. T cells gene-engineered with DAP12 mediate effector function in 1169 an NKG2D-dependent and major histocompatibility complex-independent manner. J 1170 Biol Chem 280, 38235-38241 (2005). 1171 61. Heufler, C., et al. Interleukin-12 is produced by dendritic cells and mediates T helper 1172 1 development as well as interferon-gamma production by T helper 1 cells. Eur J 1173 Immunol 26, 659-668 (1996). 1174 62. Rollings, C.M., Sinclair, L.V., Brady, H.J.M., Cantrell, D.A. & Ross, S.H. Interleukin-2 1175 shapes the cytotoxic T cell proteome and immune environment-sensing programs. 1176 Sci Signal 11(2018). 1177 63. Gerriets, V.A., et al. Leptin directly promotes T-cell glycolytic metabolism to drive 1178 effector T-cell differentiation in a mouse model of autoimmunity. Eur J Immunol 46, 1179 1970-1983 (2016). 1180 64. De Rosa, V., et al. A key role of leptin in the control of regulatory T cell proliferation. 1181 Immunity 26, 241-255 (2007). 1182 65. Weinreich, M.A., et al. KLF2 transcription-factor deficiency in T cells results in 1183 unrestrained cytokine production and upregulation of bystander chemokine 1184 receptors. Immunity 31, 122-130 (2009). 1185 66. Lynch, K.W. Consequences of regulated pre-mRNA splicing in the immune system. 1186 Nature reviews. Immunology 4, 931-940 (2004). 1187 67. Krammer, P.H., Arnold, R. & Lavrik, I.N. Life and death in peripheral T cells. Nature 1188 reviews. Immunology 7, 532-542 (2007). 1189 68. Robertson, A.K., et al. Disruption of TGF-beta signaling in T cells accelerates 1190 atherosclerosis. J Clin Invest 112, 1342-1350 (2003). 1191 69. Baker, C.M., et al. Opposing roles for RhoH GTPase during T-cell migration and 1192 activation. Proceedings of the National Academy of Sciences of the United States of 1193 America 109, 10474-10479 (2012). 1194 70. Zhang, N. & Bevan, M.J. CD8(+) T cells: foot soldiers of the immune system. Immunity 1195 35, 161-168 (2011). 1196 71. Sun, Z., et al. Requirement for RORgamma in thymocyte survival and lymphoid 1197 organ development. Science 288, 2369-2373 (2000). 1198 72. Hogan, P.G. Calcium-NFAT transcriptional signalling in T cell activation and T cell 1199 exhaustion. Cell Calcium 63, 66-69 (2017).
33
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1200 73. Crabtree, G.R. & Olson, E.N. NFAT signaling: choreographing the social lives of cells. 1201 Cell 109 Suppl, S67-79 (2002). 1202 74. Stelekati, E., et al. Long-Term Persistence of Exhausted CD8 T Cells in Chronic 1203 Infection Is Regulated by MicroRNA-155. Cell Rep 23, 2142-2156 (2018). 1204 75. Wu, G., et al. Relationship between elevated soluble CD74 and severity of 1205 experimental and clinical ALI/ARDS. Scientific reports 6, 30067 (2016). 1206 76. Elloumi, H.Z., et al. A cell permeable peptide inhibitor of NFAT inhibits macrophage 1207 cytokine expression and ameliorates experimental colitis. PloS one 7, e34172 1208 (2012). 1209 77. Watanabe, N. & Ikeda, U. Matrix metalloproteinases and atherosclerosis. Curr 1210 Atheroscler Rep 6, 112-120 (2004). 1211 78. Fontana, M.F., et al. JUNB is a key transcriptional modulator of macrophage 1212 activation. Journal of immunology 194, 177-186 (2015). 1213 79. Huang, C., et al. Exosomal MALAT1 derived from oxidized low-density lipoprotein- 1214 treated endothelial cells promotes M2 macrophage polarization. Mol Med Rep 18, 1215 509-515 (2018). 1216 80. Huangfu, N., et al. LncRNA MALAT1 regulates oxLDL-induced CD36 expression via 1217 activating beta-catenin. Biochem Biophys Res Commun 495, 2111-2117 (2018). 1218 81. Hong, C., et al. Constitutive activation of LXR in macrophages regulates metabolic 1219 and inflammatory gene expression: identification of ARL7 as a direct target. J Lipid 1220 Res 52, 531-539 (2011). 1221 82. Son, S.H., Goo, Y.H., Chang, B.H. & Paul, A. Perilipin 2 (PLIN2)-deficiency does not 1222 increase cholesterol-induced toxicity in macrophages. PloS one 7, e33063 (2012). 1223 83. Lin, J., Kakkar, V. & Lu, X. Impact of MCP-1 in atherosclerosis. Curr Pharm Des 20, 1224 4580-4588 (2014). 1225 84. Stary, H.C., et al. A definition of advanced types of atherosclerotic lesions and a 1226 histological classification of atherosclerosis. A report from the Committee on 1227 Vascular Lesions of the Council on Arteriosclerosis, American Heart Association. 1228 Circulation 92, 1355-1374 (1995). 1229 85. Jain, A., Song, R., Wakeland, E.K. & Pasare, C. T cell-intrinsic IL-1R signaling licenses 1230 effector cytokine production by memory CD4 T cells. Nat Commun 9, 3185 (2018). 1231 86. Ayroldi, E., et al. Interleukin-6 (IL-6) prevents activation-induced cell death: IL-2- 1232 independent inhibition of Fas/fasL expression and cell death. Blood 92, 4212-4219 1233 (1998). 1234 87. Daynes, R.A., Dowell, T. & Araneo, B.A. Platelet-derived growth factor is a potent 1235 biologic response modifier of T cells. J Exp Med 174, 1323-1333 (1991). 1236 88. Ma, J., Wang, R., Fang, X. & Sun, Z. beta-catenin/TCF-1 pathway in T cell development 1237 and differentiation. J Neuroimmune Pharmacol 7, 750-762 (2012). 1238 89. Crawford, A., et al. Molecular and transcriptional basis of CD4(+) T cell dysfunction 1239 during chronic infection. Immunity 40, 289-302 (2014). 1240 90. Sutterwala, F.S., Haasken, S. & Cassel, S.L. Mechanism of NLRP3 inflammasome 1241 activation. Ann N Y Acad Sci 1319, 82-95 (2014). 1242 91. Braunersreuther, V., et al. A novel RANTES antagonist prevents progression of 1243 established atherosclerotic lesions in mice. Arteriosclerosis, thrombosis, and vascular 1244 biology 28, 1090-1096 (2008).
34
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1245 92. Herder, C., et al. RANTES/CCL5 and risk for coronary events: results from the 1246 MONICA/KORA Augsburg case-cohort, Athero-Express and CARDIoGRAM studies. 1247 PloS one 6, e25734 (2011). 1248 93. Nagashima, H., et al. CXCR4 signaling in macrophages contributes to periodontal 1249 mechanical hypersensitivity in Porphyromonas gingivalis-induced periodontitis in 1250 mice. Mol Pain 13, 1744806916689269 (2017). 1251 94. Angsana, J., Chen, J., Liu, L., Haller, C.A. & Chaikof, E.L. Efferocytosis as a regulator of 1252 macrophage chemokine receptor expression and polarization. Eur J Immunol 46, 1253 1592-1599 (2016). 1254 95. Wells, C.M., Walmsley, M., Ooi, S., Tybulewicz, V. & Ridley, A.J. Rac1-deficient 1255 macrophages exhibit defects in cell spreading and membrane ruffling but not 1256 migration. J Cell Sci 117, 1259-1268 (2004). 1257 96. Wheeler, A.P., et al. Rac1 and Rac2 regulate macrophage morphology but are not 1258 essential for migration. J Cell Sci 119, 2749-2757 (2006). 1259 97. Lee, D.J., Cox, D., Li, J. & Greenberg, S. Rac1 and Cdc42 are required for phagocytosis, 1260 but not NF-kappaB-dependent gene expression, in macrophages challenged with 1261 Pseudomonas aeruginosa. J Biol Chem 275, 141-146 (2000). 1262 98. Vallieres, F. & Girard, D. IL-21 enhances phagocytosis in mononuclear phagocyte 1263 cells: identification of spleen tyrosine kinase as a novel molecular target of IL-21. 1264 Journal of immunology 190, 2904-2912 (2013). 1265 99. Jia, D., et al. Interleukin-35 Promotes Macrophage Survival and Improves Wound 1266 Healing After Myocardial Infarction in Mice. Circulation research (2019). 1267 100. Martin, D., Galisteo, R. & Gutkind, J.S. CXCL8/IL8 stimulates vascular endothelial 1268 growth factor (VEGF) expression and the autocrine activation of VEGFR2 in 1269 endothelial cells by activating NFkappaB through the CBM (Carma3/Bcl10/Malt1) 1270 complex. J Biol Chem 284, 6038-6042 (2009). 1271 101. Apostolakis, S., Vogiatzi, K., Amanatidou, V. & Spandidos, D.A. Interleukin 8 and 1272 cardiovascular disease. Cardiovasc Res 84, 353-360 (2009). 1273 102. Lutgens, E., et al. Atherosclerotic plaque rupture: local or systemic process? 1274 Arteriosclerosis, thrombosis, and vascular biology 23, 2123-2130 (2003). 1275 103. Tziakas, D.N., et al. Interleukin-8 is increased in the membrane of circulating 1276 erythrocytes in patients with acute coronary syndrome. European heart journal 29, 1277 2713-2722 (2008). 1278 104. Xiao, H., et al. Anti-fibrotic effects of pirfenidone by interference with the hedgehog 1279 signalling pathway in patients with systemic sclerosis-associated interstitial lung 1280 disease. Int J Rheum Dis 21, 477-486 (2018). 1281 105. Cao, H., et al. Inhibition of Wnt/beta-catenin signaling suppresses myofibroblast 1282 differentiation of lung resident mesenchymal stem cells and pulmonary fibrosis. 1283 Scientific reports 8, 13644 (2018). 1284 106. Nishikawa, K., Osawa, Y. & Kimura, K. Wnt/beta-Catenin Signaling as a Potential 1285 Target for the Treatment of Liver Cirrhosis Using Antifibrotic Drugs. Int J Mol Sci 1286 19(2018). 1287 107. Nunez, V., et al. Retinoid X receptor alpha controls innate inflammatory responses 1288 through the up-regulation of chemokine expression. Proceedings of the National 1289 Academy of Sciences of the United States of America 107, 10626-10631 (2010).
35
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1290 108. Penas, F., et al. Treatment in vitro with PPARalpha and PPARgamma ligands drives 1291 M1-to-M2 polarization of macrophages from T. cruzi-infected mice. Biochim Biophys 1292 Acta 1852, 893-904 (2015). 1293 109. Rigamonti, E., Chinetti-Gbaguidi, G. & Staels, B. Regulation of macrophage functions 1294 by PPAR-alpha, PPAR-gamma, and LXRs in mice and men. Arteriosclerosis, 1295 thrombosis, and vascular biology 28, 1050-1059 (2008). 1296 110. Chawla, A. Control of macrophage activation and function by PPARs. Circulation 1297 research 106, 1559-1569 (2010). 1298 111. Pontis, S., Ribeiro, A., Sasso, O. & Piomelli, D. Macrophage-derived lipid agonists of 1299 PPAR-alpha as intrinsic controllers of inflammation. Crit Rev Biochem Mol Biol 51, 7- 1300 14 (2016). 1301 112. McInnes, I.B. & Schett, G. Cytokines in the pathogenesis of rheumatoid arthritis. 1302 Nature reviews. Immunology 7, 429-442 (2007). 1303 113. Pazianas, M., Rhim, A.D., Weinberg, A.M., Su, C. & Lichtenstein, G.R. The effect of anti- 1304 TNF-alpha therapy on spinal bone mineral density in patients with Crohn's disease. 1305 Ann N Y Acad Sci 1068, 543-556 (2006). 1306 114. Mackey, R.H., Kuller, L.H. & Moreland, L.W. Update on Cardiovascular Disease Risk in 1307 Patients with Rheumatic Diseases. Rheum Dis Clin North Am 44, 475-487 (2018). 1308 115. Singh, S., Kullo, I.J., Pardi, D.S. & Loftus, E.V., Jr. Epidemiology, risk factors and 1309 management of cardiovascular diseases in IBD. Nat Rev Gastroenterol Hepatol 12, 1310 26-35 (2015). 1311 116. Corna, G., et al. Polarization dictates iron handling by inflammatory and alternatively 1312 activated macrophages. Haematologica 95, 1814-1822 (2010). 1313 117. Fan, Y., et al. The effect of anti-inflammatory properties of ferritin light chain on 1314 lipopolysaccharide-induced inflammatory response in murine macrophages. 1315 Biochim Biophys Acta 1843, 2775-2783 (2014). 1316 118. Kumar, M.P., et al. Analysis of Single-Cell RNA-Seq Identifies Cell-Cell 1317 Communication Associated with Tumor Characteristics. Cell Rep 25, 1458-1468 1318 e1454 (2018). 1319 119. Choi, H., et al. Transcriptome analysis of individual stromal cell populations 1320 identifies stroma-tumor crosstalk in mouse lung cancer model. Cell Rep 10, 1187- 1321 1201 (2015). 1322 120. Skelly, D.A., et al. Single-Cell Transcriptional Profiling Reveals Cellular Diversity and 1323 Intercommunication in the Mouse Heart. Cell Rep 22, 600-610 (2018). 1324 121. Bromley, S.K., Mempel, T.R. & Luster, A.D. Orchestrating the orchestrators: 1325 chemokines in control of T cell traffic. Nature immunology 9, 970-980 (2008). 1326 122. De Strooper, B., Iwatsubo, T. & Wolfe, M.S. Presenilins and gamma-secretase: 1327 structure, function, and role in Alzheimer Disease. Cold Spring Harb Perspect Med 2, 1328 a006304 (2012). 1329 123. Eagar, T.N., et al. Notch 1 signaling regulates peripheral T cell activation. Immunity 1330 20, 407-415 (2004). 1331 124. Cummings, M., et al. Presenilin1 regulates Th1 and Th17 effector responses but is 1332 not required for experimental autoimmune encephalomyelitis. PloS one 13, 1333 e0200752 (2018).
36
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1334 125. Mathieu, M., Cotta-Grand, N., Daudelin, J.F., Thebault, P. & Labrecque, N. Notch 1335 signaling regulates PD-1 expression during CD8(+) T-cell activation. Immunology 1336 and cell biology 91, 82-88 (2013). 1337 126. Marra, P., et al. IL15RA drives antagonistic mechanisms of cancer development and 1338 immune control in lymphocyte-enriched triple-negative breast cancers. Cancer Res 1339 74, 4908-4921 (2014). 1340 127. Read, K.A., Powell, M.D., McDonald, P.W. & Oestreich, K.J. IL-2, IL-7, and IL-15: 1341 Multistage regulators of CD4(+) T helper cell differentiation. Exp Hematol 44, 799- 1342 808 (2016). 1343 128. Waickman, A.T., et al. CD4 effector T cell differentiation is controlled by IL-15 that is 1344 expressed and presented in trans. Cytokine 99, 266-274 (2017). 1345 129. Monaco, C., et al. Toll-like receptor-2 mediates inflammation and matrix degradation 1346 in human atherosclerosis. Circulation 120, 2462-2469 (2009). 1347 130. Mueller, P.A., et al. Deletion of Macrophage Low-Density Lipoprotein Receptor- 1348 Related Protein 1 (LRP1) Accelerates Atherosclerosis Regression and Increases C-C 1349 Chemokine Receptor Type 7 (CCR7) Expression in Plaque Macrophages. Circulation 1350 138, 1850-1863 (2018). 1351 131. Barkal, A.A., et al. Engagement of MHC class I by the inhibitory receptor LILRB1 1352 suppresses macrophages and is a target of cancer immunotherapy. Nature 1353 immunology 19, 76-84 (2018). 1354 132. Tyner, J.W., et al. CCL5-CCR5 interaction provides antiapoptotic signals for 1355 macrophage survival during viral infection. Nature medicine 11, 1180-1187 (2005). 1356 133. Li, W., et al. GDF11 antagonizes TNF-alpha-induced inflammation and protects 1357 against the development of inflammatory arthritis in mice. FASEB J 33, 3317-3329 1358 (2019). 1359 134. Osborne, B.A. & Minter, L.M. Notch signalling during peripheral T-cell activation and 1360 differentiation. Nature reviews. Immunology 7, 64-75 (2007). 1361 135. Lanzavecchia, A., Iezzi, G. & Viola, A. From TCR engagement to T cell activation: a 1362 kinetic view of T cell behavior. Cell 96, 1-4 (1999). 1363 136. Lynch, E.A., Heijens, C.A., Horst, N.F., Center, D.M. & Cruikshank, W.W. Cutting edge: 1364 IL-16/CD4 preferentially induces Th1 cell migration: requirement of CCR5. Journal 1365 of immunology 171, 4965-4968 (2003). 1366 137. Becker, M., et al. Integrated Transcriptomics Establish Macrophage Polarization 1367 Signatures and have Potential Applications for Clinical Health and Disease. Scientific 1368 reports 5, 13351 (2015). 1369 138. Papalexi, E. & Satija, R. Single-cell RNA sequencing to explore immune cell 1370 heterogeneity. Nature reviews. Immunology (2017). 1371 139. Satija, R. & Shalek, A.K. Heterogeneity in immune responses: from populations to 1372 single cells. Trends Immunol 35, 219-229 (2014). 1373 140. Tanay, A. & Regev, A. Scaling single-cell genomics from phenomenology to 1374 mechanism. Nature 541, 331-338 (2017). 1375 141. Tang, F., Lao, K. & Surani, M.A. Development and applications of single-cell 1376 transcriptome analysis. Nat Methods 8, S6-11 (2011). 1377 142. Bu, D.X., et al. Impairment of the programmed cell death-1 pathway increases 1378 atherosclerotic lesion development and inflammation. Arteriosclerosis, thrombosis, 1379 and vascular biology 31, 1100-1107 (2011). 37
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1380 143. Darvin, P., Toor, S.M., Sasidharan Nair, V. & Elkord, E. Immune checkpoint inhibitors: 1381 recent progress and potential biomarkers. Exp Mol Med 50, 165 (2018). 1382 144. Edwards, J.D., Kapral, M.K., Fang, J. & Swartz, R.H. Long-term morbidity and 1383 mortality in patients without early complications after stroke or transient ischemic 1384 attack. CMAJ 189, E954-E961 (2017). 1385 145. Ridker, P.M., et al. Effect of interleukin-1beta inhibition with canakinumab on 1386 incident lung cancer in patients with atherosclerosis: exploratory results from a 1387 randomised, double-blind, placebo-controlled trial. Lancet 390, 1833-1842 (2017). 1388 146. Everett, B.M., et al. Rationale and design of the Cardiovascular Inflammation 1389 Reduction Trial: a test of the inflammatory hypothesis of atherothrombosis. 1390 American heart journal 166, 199-207 e115 (2013). 1391 147. Fainboim, L., Navarrete, C. & Festenstein, H. Precursor and effector phenotypes of 1392 activated human T lymphocytes. Nature 288, 391-393 (1980). 1393 148. Brenchley, J.M., et al. Expression of CD57 defines replicative senescence and antigen- 1394 induced apoptotic death of CD8+ T cells. Blood 101, 2711-2720 (2003). 1395 149. Gattinoni, L., et al. Wnt signaling arrests effector T cell differentiation and generates 1396 CD8+ memory stem cells. Nature medicine 15, 808-813 (2009). 1397 150. Jiang, Y., Li, Y. & Zhu, B. T-cell exhaustion in the tumor microenvironment. Cell Death 1398 Dis 6, e1792 (2015). 1399 151. Sun, Z., et al. PKC-theta is required for TCR-induced NF-kappaB activation in mature 1400 but not immature T lymphocytes. Nature 404, 402-407 (2000). 1401 152. Mackay, L.K., et al. The developmental pathway for CD103(+)CD8+ tissue-resident 1402 memory T cells of skin. Nature immunology 14, 1294-1301 (2013). 1403 153. Hintzen, R.Q., et al. Regulation of CD27 expression on subsets of mature T- 1404 lymphocytes. Journal of immunology 151, 2426-2435 (1993). 1405 154. Schiott, A., Lindstedt, M., Johansson-Lindbom, B., Roggen, E. & Borrebaeck, C.A. 1406 CD27- CD4+ memory T cells define a differentiated memory population at both the 1407 functional and transcriptional levels. Immunology 113, 363-370 (2004). 1408 155. Wang, Z., et al. Clonally diverse CD38(+)HLA-DR(+)CD8(+) T cells persist during 1409 fatal H7N9 disease. Nat Commun 9, 824 (2018). 1410 156. Vanham, G., et al. Subset markers of CD8(+) cells and their relation to enhanced 1411 cytotoxic T-cell activity during human immunodeficiency virus infection. J Clin 1412 Immunol 11, 345-356 (1991). 1413 157. Herrmann, J., et al. Expression of lipoprotein-associated phospholipase A(2) in 1414 carotid artery plaques predicts long-term cardiac outcome. European heart journal 1415 30, 2930-2938 (2009). 1416 158. Versari, D., et al. Dysregulation of the ubiquitin-proteasome system in human 1417 carotid atherosclerosis. Arteriosclerosis, thrombosis, and vascular biology 26, 2132- 1418 2139 (2006). 1419 159. Lai, L., Ong, R., Li, J. & Albani, S. A CD45-based barcoding approach to multiplex 1420 mass-cytometry (CyTOF). Cytometry. Part A : the journal of the International Society 1421 for Analytical Cytology 87, 369-374 (2015). 1422 160. Fienberg, H.G., Simonds, E.F., Fantl, W.J., Nolan, G.P. & Bodenmiller, B. A platinum- 1423 based covalent viability reagent for single-cell mass cytometry. Cytom Part A 81A, 1424 467-475 (2012).
38
bioRxiv preprint doi: https://doi.org/10.1101/721688; this version posted August 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1425 161. Carlson, C.S., et al. Using synthetic templates to design an unbiased multiplex PCR 1426 assay. Nat Commun 4, 2680 (2013). 1427 162. Kirsch, I., Vignali, M. & Robins, H. T-cell receptor profiling in cancer. Mol Oncol 9, 1428 2063-2070 (2015). 1429 163. Azizi, E., et al. Single-Cell Map of Diverse Immune Phenotypes in the Breast Tumor 1430 Microenvironment. Cell 174, 1293-1308 e1236 (2018). 1431 164. Lambrechts, D., et al. Phenotype molding of stromal cells in the lung tumor 1432 microenvironment. Nature medicine 24, 1277-1289 (2018). 1433 165. Fernandez, N.F., et al. Clustergrammer, a web-based heatmap visualization and 1434 analysis tool for high-dimensional biological data. Sci Data 4, 170151 (2017). 1435 166. Haghverdi, L., Lun, A.T.L., Morgan, M.D. & Marioni, J.C. Batch effects in single-cell 1436 RNA-sequencing data are corrected by matching mutual nearest neighbors. Nature 1437 biotechnology 36, 421-427 (2018). 1438 167. Kuleshov, M.V., et al. Enrichr: a comprehensive gene set enrichment analysis web 1439 server 2016 update. Nucleic Acids Res 44, W90-97 (2016). 1440 168. Chen, E.Y., et al. Enrichr: interactive and collaborative HTML5 gene list enrichment 1441 analysis tool. BMC Bioinformatics 14, 128 (2013). 1442 169. Ramilowski, J.A., et al. A draft network of ligand-receptor-mediated multicellular 1443 signalling in human. Nat Commun 6, 7866 (2015). 1444
39
Figure 1 STUDY SUMMARY
a. Tissue and Blood Processing b. CyTOF and CITE-Seq Computational Pipeline
Surgical resection of Single cell Cell Multiplexing atherosclerotic plaques isolation barcoding
Plaque
Blood
c. Single Cell Analyses of Symptomatic and Asymptomatic Plaques
10 0% 25 16 15 14 80 % 8 13 9 21 60 % 20 11 10 2 40 % 22 1 19 4 20 % 12 18 23 7 0% 24 ASYM SYM
Asymptomatic Symptomatic (No Stroke/ TIA) (Stroke/TIA)
50 CD69
TNFAIP3 40 ASYM SYM CXCR4 ~60 ~40 Autocrine 30 DUSP1 % %
JUNB CD4 -Log10 qVal NR4A2 20 CTSW ZFP36 JUN CCL4 NFKBIA ACTBSH3BGRL3 CD55 ACTG1 10 CNOT6L DUSP2 NPM1 TRAC Paracrine IFNG GZMH KLF6 ZFP36L2 CORO1A MΦ CD8 B2MEMP3 Ligands FOS IL32 ZAP70 FOSB CFL1 TNF Rec eptors 0 -6 -4 -2 0 2 4 6 Log2 Fold Change Figure 2 Macrophages + (4) Blood CD1c DCs a b (1) Monocytes Plaque Neutrophils (13) (2) CD4-CD8- pDCs T Cells (16) (18) B Cells (3)
NK Cells (6)
CD4+ T Cells CD8+ T Cells (7, 9, 17) (11, 12) CD4+CD8+ T Cells (10) c Cell Markers d 100% Undefined CCR7 CD11b CD11c CD123 CD127 CD14 CD141 CD16 CD163 CD19 CD1c CD206 CD235ab CD24 CD25 CD27 CD3 CD33 CD4 CD45RA CD56 CD61 CD64 CD66b CD8 CD86 HLA_ABC HLADR pDCs 16 pDCs CD1c+ DCs 3 B Cells Neutrophils 8 Undefined 80% NK Cells 6 NK Cells B Cells 2 Neutrophils Monocytes 17 CD4+ CM CD27+CD25int Macrophages 9 CD4+ EM CD27+CD25hi 60% CD4- CD8- T Cells CD4+ CD8+ T Cells 10 + + CD4 CD8 CD8 Tcells + 7 CD4 EM CD4 T Cells 18 CD4-CD8- MetaClusters 12 CD8+ EM 40% 11 CD8+ EMRA 13 CD14+ Monocytes 15 Undefined 4 Macrophages 20% 1 CD1c+ DCs 5 Undefined 14 Undefined 0% Blood Plaque -3.8 3.8 Z- Score M e ta C lu s te r 4 M e ta C lu s te r 1 2 e f M a c ro p h a g e s C D 8 + E M T C e lls 4 0 C D 4 5 + M e ta C lu s te r in g A n a ly s is 2 5 * * * s s l l 2 0 l 3 0 l e
7 .5 e C C
+ * * + 1 5 5 Blood 5 4 2 0 4 D D 1 0 C Plaque 6 C
f f o 1 2 o
1 0 5 % % )
e 5 .0 u l 1 3 0 0 a B lo o d P la q u e B lo o d P la q u e v
- 1 6 p (
0 M e ta C lu s te r 7 M e ta C lu s te r 1 8 1 4
g + - -
o C D 4 E M T C e lls C D 4 C D 8 T C e lls
L 2 .5 1 0 - 7 1 5 5 0 2 0 3 1 8 * * s s
l 4 0 l l l 1 5 e e
C * * C
+ 3 0 + 5 5
4 4 1 0 D 2 0 D 0 .0 C C
f f o o
-1 0 -5 0 5 1 0 5 1 0 L o g 2 (F o ld C h a n g e ) % %
0 0 B lo o d P la q u e B lo o d P la q u e Extended Figure 1 a CD141 CD206 CD25 CD11b CD235ab CD8 CD19
HLA ABC CD127 CD61 CD11c CCR7 CD19 CD86
CD3 HLADR CD64 CD163 CD45RA CD1c CD24
CD56 CD33 CD16 CD66b CD14 CD123
MetaCluster Frequency b 1. CD1c+ DCs 11. CD8+ EMRA 40 + Blood 2. Neutrophils 12. CD8 EM +
s Plaque 3. B Cells 13. CD14 Monocytes l l 30 **** e 4. Macrophages 14. Undefined C
+ 5. Undefined 15. Undefined
5 * *** 4 20 **** 6. NK Cells 16. pDCs D
C + + + int * 7. CD4 EM 17. CD4 CM CD27 CD25 f
o * 10 8. Undefined 18. CD4-CD8-
% * * **** 9. CD4+ EM CD27+CD25hi 0 10. CD4+ CD8+ 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 1 1 1 1 1 1 1 1 1 MetaCluster c M e ta C lu s te r 1 3 M e ta C lu s te r 6 M e ta C lu s te r 1 6 M e ta C lu s te r 3 M e ta C lu s te r 1 0 C D 1 4 + M o n o c y te s N K C e lls p D C s B C e lls C D 4 + C D 8 + T C e lls
5 0 2 5 2 5 2 5 2 5 * * * * * * * * s s s s s l
2 0 l l l 2 0 2 0 l
4 0 l 2 0 l l l l e
e e N S e e C C C C C
+
1 5 + + + 3 0 1 5 1 5 + 1 5 5 5 5 5 5
4 * * * 4 4 4 * * * 4 D D D D 1 0 D C 1 0 1 0
2 0 C C 1 0 C C
f f f f f o o o o o
5 % 5 5
1 0 % % 5 % %
0 0 0 0 0 B lo o d P la q u e B lo o d P la q u e B lo o d P la q u e B lo o d P la q u e B lo o d P la q u e Extended Figure 2
MC15 a b MC14 4% MC2
CD127 CD4 CD3 CD25 CD8 HLA ABC CD61 CD86 CD14 CD64 CD141 CD206 CD235ab CD163 CD123 CD11c CD11b HLADR CD1c CD56 CD16 CD33 CD66b CD24 CD45RA CD19 6% 3% 9 CD4+ CD127+ T Cells MC1 15 Undefined MC12 7% 8 CD4+ CD127lowCD25+ T Cells 2% MC… 11 CD4- CD8- T Cells MC3 MC11 7 + low 14% CD8 CD127 T Cells 3% 3 CD8+ CD127lowCD45RAlnt T Cells 10 CD8+ CD127+ T Cells MC10 MC4 13 CD4+ CD127low T Cells 9% 10% 5 CD4+ CD56+ T Cells 6 CD33+ Dendritic Cells MC9 MC5 2 CD19+ B Cells 12% 10% 14 CD56+CD16+ NK Cells MC7 4 CD206lowCD163low Macrophages 7% 1 CD206hiCD163hi Macrophages MC8 MC6 12 CD66b+CD24+ Granulocytes 3% 3% -3.6 3.6 Z- Score c Metaclusters d CD86 CD163 CD206 CD141
1
4
15 2 6 CD14 CD33 HLADR CD64 8 14
9 3 13 7 10 e Figure 3 T Cells CD8 CD4 MetaClusters a b c d
Blood Plaque
Cell Markers e f
CD25 ICOS CD4 CCR7 CD27 CD28 CD26 CD127 CCR6 CCR4 CD103 HLADR CD38 CXCR3 CXCR5 CD69 CCR5 PD1 GZMB Perforin CD57 CD45RA CD8 100% 15 CD8- CD4- EM 20 CD8+ EM CD103-CD127int EM 12 CD4+ EM CD57- 15 10 CD8+ EM CD103+ 20
1 CD4+ EM PD1+ GZMB+CD57- EM/EMRA 12 80% 22 CD4+ EM PD1+ GZMB+CD57+ 10 9 CD8+ EMRA CD57+ 1 - - + 22 25 CD8 CD4 EMRA PD1 9 21 CD8+ EMRA CD57- 25 11 CD8+ EM PD1+ CD103-CD127low 21 60% 24 CD4+ CM CXCR3+ 11 14 CD4low CM CM 24 14 5 CD8+ CD4+ CD57- + + + 5 3 CD8 CD4 CD57 3 18 CD4+ EM PD1-GZMB- CD103- 18 + - - + EM 40% MetaClusters 4 CD4 EM PD1 GZMB CD103 4 6 CD4+ ICOS+ CD38+ 6 19 CD4+ ICOS+CD25+CD38- 19 CD8-CD4- Naive Naïve/CM 16 16 13 CD8+ Naïve 13 20% 17 17 CD4+ CM CXCR3- CD127high 7 7 CD4+ CM CXCR3- CD127low 23 23 CD4+ CM CXCR3- GZMB+ 2 + EM 8 2 CD4 EMRA CD8- CD4- CM 8 0%
-4.6 4.6 Blood Plaque Z- Score M e ta C lu s te r 1 0 M e ta C lu s te r 1 1 C D 8 + C D 1 0 3 + E M C D 8 + C D 1 0 3 - C D 1 2 7 lo E M g C D 3 + M e ta C lu s te rin g A n a ly s is h 2 0 * * 2 5 * * * 1 0 .0 s s 2 0 l l l 1 2 l 1 5 e Blood e C C
1 5 + + 3 Plaque 1 8 3 1 0 D D C C 1 0
f f o 7 .5 o
5 % % 5 )
e 1 7
u 2 4 l 0 0 a 2 0 B lo o d P la q u e v B lo o d P la q u e -
p 6
( 5 .0 M e ta C lu s te r 1 2 M e ta C lu s te r 2 0 0
1 C D 4 + E M C D 8 + C D 1 0 3 - C D 1 2 7 in t E M g 5 o
L 3 5 0 8 0 - 8 1 1 * * * * * * * *
2 .5 s 4 0 s
9 l l 1 0 l l 6 0 1 3 e e
2 2 C C
1 6 3 0 + +
3 3 4 0 D D
C 2 0 C
f f o o 0 .0 2 0 -6 -4 -2 0 2 4 6 % 1 0 %
L o g 2 (F o ld C h a n g e ) 0 0 B lo o d P la q u e B lo o d P la q u e Extended Figure 3 a CD103 CD57 CD45RA CD127 ICOS CD206 CD26 CD38
CCR6 CD25 CCR7 CCR4 CXCR5 CXCR3 HLADR
CD69 CCR5 PD-1 CD27 CD28 Perforin Granzyme B
b T Cell MetaCluster Frequency c 40 **** Blood **** Plaque s
l **** l 30 e C **** +
3 20 D C
** **** f o 10 * % **** * *** *** *** % Cells of T 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 MetaCluster
MetaCluster 11 d CD8+ CD127lo EM Patient 0.8 1 2 3 4 5 6 7 8 9 10
n * Blood * o * i * * * s * * Plaque s 0.6 * * e * * r
* 11 12 13 14 15 16 17 18 19 20 p * * x * * * E * *
* 0.4 * d * e * z * * i * l * 21 22 23 24 25 a 0.2 * m r o N 0.0 8 3 n 7 c 3 b 1 S 7 5 4 5 9 6 5 7 8 6 3 6 7 2 A 0 i 2 1 2 6 _ 2 6 2 3 B 0 R 2 5 R 1 r 1 1 6 O R R R R R M 2 R D T Cell MetaCluster Community D 5 fo D D C D C C C D C D C D Z C A D D e D D D D I D C 4 r C P C C C X C C C C C G X L C C D C e C C C C C C H C P MetaCluster 12 CD4+ EM Blood
0.8 * * * * n Plaque * * * o * i * * * * s * s * 0.6 * * e * r * * p * * * x * E * *
0.4 * * d * * * e * * z i l * a
0.2 * m * r * o * N 0.0 8 A 3 n 7 c 3 b 1 S 7 5 4 5 9 6 5 7 8 B 6 3 R 6 7 2 0 ri 2 1 2 6 _ 2 R R R 6 R 2 R 3 0 R 2 5 D R 1 o 1 D 1 6 D O D D D D M 2 D D D 5 D f D D D IC C C C C C Z D C A C 4 r C P C C C X C C C C C G X L C C D C e C C C C C C H C P MetaCluster 20 CD8+ CD127int EM *
0.8 * * * * n * Blood * * o * * * * i * * * s * * * Plaque s 0.6 * e * * r * * * p * * x * * T Cell MetaCluster Community CellMetaCluster T E *
*
0.4 * * d * * * e * * * z * * * * * i * * * l * * * * * a * * *
0.2 * m * r * o N 0.0 8 A 3 n 7 c 3 b 1 S 7 5 4 5 9 6 5 7 8 B 6 3 R 6 7 2 0 ri 2 1 2 6 _ 2 R R R 6 R 2 R 3 0 R 2 5 -4.6 4.6 D R 1 o 1 D 1 6 D O D D D D M 2 D D D 5 D f D D D IC C C C C C Z D C A C 4 r C P C C C X C C C C C G X L C C D C e C C C C C C H Z- Score C P Extended Figure 4
MetaCluster 11 a CD8+ CD127lo EM
CD69 CCR5 PD-1 CD27 CD28 0.8 0.8 0.8 0.8 0.8 **** ** **** *** ** 0.6 0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2 0.2 Normalized Expression Normalized Expression Normalized 0.0 0.0 Expression Normalized 0.0 Expression Normalized 0.0 Expression Normalized 0.0 Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque
MetaCluster 20 b CD8+ CD127int EM CD69 CCR5 PD-1 CD27 CD28 0.8 **** 0.8 N.S. 0.8 0.8 0.8 **** **** ** 0.6 0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2 0.2 Normalized Expression Normalized Expression Normalized Expression Normalized 0.0 0.0 0.0 Expression Normalized 0.0 Expression Normalized 0.0 Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque
MetaCluster 12 c CD4+ EM CD69 CCR5 PD-1 CD27 CD28 0.8 **** 0.8 0.8 0.8 0.8 *** **** ** ** 0.6 0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2 0.2
Normalized Expression Normalized 0.0 Expression Normalized 0.0 Expression Normalized 0.0 Expression Normalized 0.0 Expression Normalized 0.0 Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque
MetaCluster 11 MetaCluster 20 MetaCluster 12 d CD8+ CD127lo EM CD8+ CD127int EM CD4+ EM
Perforin Granzyme B Perforin Granzyme B Perforin Granzyme B 0.8 0.8 0.8 0.8 0.8 0.8 ** NS ** * 0.6 0.6 0.6 0.6 * 0.6 0.6 NS 0.4 0.4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2 0.2 0.2 Normalized Expression Normalized Expression Normalized Expression Normalized Expression Normalized Expression 0.0 0.0 Normalized Expression 0.0 0.0 0.0 0.0 Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque
e CD8+ T Cell MetaClusters in Tissue
CD69 CCR5 PD1 CD27 CD28 NS 1.0 1.0 NS 1.0 1.0 1.0 NS NS 0.8 0.8 0.8 0.8 **** 0.8 * **** NS **** * 0.6 0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2 0.2
Normalized Expression 0.0 Normalized Expression 0.0 Normalized Expression 0.0 Normalized Expression 0.0 Normalized Expression 0.0 0 1 0 0 1 0 0 1 0 0 1 0 0 1 0 1 1 2 1 1 2 1 1 2 1 1 2 1 1 2 ______C C C C C C C C C C C C C C C M M M M M M M M M M M M M M M Fig 4 CITEseq Analysis
a b c 100% Undefined pDCs 80% CD1c_DCs Plasma Cells 60% NKT Cells NK Cells 40% B Cells Monocytes Macrophages 20% CD8+ CD4+ T Cells Plaque T Cells Blood CD8 T Cells Plaque n=921 Macrophages 0% CD4 T Cells Blood n=1,652 Plaque n=265 PBMCBlood Plaque d e Upregulated Genes Plaque T cells Upregulated in Blood Upregulated in Plaque
**** mRNA Splicing **** Adaptive Immune System **** Gene Expression **** Cytokine Signaling in Immune system **** Translation **** Immuno-interactions b/n Lymph. and non-Lymph. Cells **** Metabolism of proteins **** Antigen processing-Cross presentation **** Respiratory electron transport **** Interferon Signaling ** Infectious disease **** DAP12 signaling ** Cell Cycle **** Prolonged ERK activation events ** Cellular responses to stress **** Signaling by Leptin *** Signaling by TGFβ Receptor Complex **** TCR signaling ** Immune System **** Interleukin-2 signaling 0 0 20 40 60 80 50 Combined Score 100 150 Combined Score
f T Cells h Macrophages 4 11 12 16 1 3 6 9 15 Cluster 4 Plaque/Blood 1 2 3 5 Cluster CD8/CD4 HLA-DQB1 HLA-DPA1 CD74 IL32 S100A6 ACTB HLA-DQA1 HLA-DRB5 HLA-DRB1 S100A4 CD52 PFN1 HLA-DPB1 HLA-DRA SH3BGRL3 ACTG1 SH3BGRL3 S100A8 TIMP1 S100A6 S100A9 SRGN GZMA CD74 TMSB4X S100A4 FCN1 ATP5e GZMK HLA-B B2M S100A11 CYBA PTMA NKG7 HLA-C S100A10 AIF1 UBA52 MYL6 TYROBP CCL5 HLA-A PFN1 LYZ H3F3B ZFP36L2 TSC22D3 JUNB NEAT1 FOS MALAT1 JUNB GADD45B MALAT1 C15orf48 NFKBIA SAT1 HSP90AA1 TPT1 EEF2 BTG1 GPX1 TMSB10 APOE EEF1A1 NACA TXNIP PFDN5 CD63 CSTB UBA52 LDHB PTMA TPT1 LGALS1 CTSB FAU GNB2L1 TMSB10 EEF1A1 FCER1G VIM EEF1D TOMM7 FTH1 ACTB NPC2 RNASEI NPM1 BTF3 FTL B2M SERF2 FABP5 EEF1B2 LTB E1F1 TMSB4X YBX1 PLIN2 PABPC1 PFDN5 FTH1 PSAP CCL2 FTL APOC1 **** **** **** **** Plaque/Blood **** *** **** ** -5 5 -5 5 Z-Score Z-Score **** g i 5 1 2 3 12 11 16 15 4 1 3 9 6 Interferon Signaling Role of NFAT in Regulation of the Immune Response T Cell Exhaustion Signaling Pathway Production of NO and ROS in Macrophages Cytotoxic T lymphocyte-mediated Apoptosis RhoGDI Signaling TGF-b Signaling Cyclins and Cell Cycle Regulation IL-1 Signaling Apelin Signaling Pathway PKCq Signaling in T Lymphocytes LXR/RXR Activation Role of NFAT Regulation of the Immune Response LPS/IL-1 Mediated Inhibition of RXR Function RhoA Signaling Inhibition of Matrix Metalloproteases Oxidative Phosphorylation Interferon Signaling RhoGDI Signaling Chemokine Signaling
-4.690 2.416 -4.359 3.000 Z-Score Z-Score Extended Figure 5 a b 100% c T Cells CD 8 T Ce lls CD4 CD8 269 1316 80% CD 4 T Ce lls 84.3 271 60% CD 8+ C D4+ T Cel ls 26.2 55.7 7.5 11 0.0 0.3 40%
20%
Plaque 0% Blood Blood Plaque
d CD62L 301.2 CD27 27.5 CD38 333.1 HLADR 286.5 90.1 14.5 96.4 115.4
26.7 7.3 27.7 46.4
7.3 2.9 7.3 18.4 0.2 0.3 0.3 6.75
e CD62L CD27 CD38 HLADR CD4+ CD8+ CD4+ CD8+ CD4+ CD8+ CD4+ CD8+
p<.0001p<.0001 p<.0001 p<.0001 p<.0001 p=0.7934 p<.0001 p<.0001 p<.0001 10001000 1000 100 100 1000 10000 1000 1000
100 1000 100 100 100 10 10 100 1010 10 100 100 10 10 11 1 1 1 1 1 0.10.1 0.1 10 10
Raw Intensity 0.1 0.1 0.1 0.1 0.010.01 0.01 0.01 0.0010.001 0.01 0.01 0.01 0.001 0.001 1 1 BloodBlood PlaquePlaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque Blood Plaque f Macrophages g CD64 HLADR
153.8 2192
53.8 537.9
Other Cells 19.6 131.9 Macrophages 5.7 32.2 0.4 7.164
h CD206 i Macrophages
2192
537.9 131.9 47% 32.2
7.16 53% Figure 5
100% 25 a b 16 Asymptomatic Symptomatic 15 c d 14 CD8-CD4- (No Stroke/ TIA) (Stroke/TIA) 80% 8 13 5 50 **** 9 21 CD8+ 20 4 12 40 ~60% 60% 11 10 2 22 3 30 1 40% 19 4 + 2 20 12 CD4 18 -log(q value) 14 ~40% 20% 23 7 1 9 % of CD3+ Cells 10 24 6 0% 17 0 0 -10 -5 0 5 10 15 ASYM SYM ASYM SYM Difference e f g Type V Type V TypCeo VhoI rPt 2la qTyupee VI Plaque Cohort 2 Type V vs Type VI Type VI 20% Type VI 10 Not Determined 10 ND ND 8 MC12 8 ) 30% ) e e u 26% u l 6 CD4 T Cells l 6 CD4 T Cells a a v v
q SYM q (
( SYM 4 g 4 g o ASYM o ASYM l l - - 2 2 14% 10% IFN-gammaMC 1pathway4 IFN-gamma pathway MC14 IL12-mediated signaling events IL12-mediated signaling events 0 0 ASYM SYM positive-20 regulation-10 0of type 1I interferon0 20 production30 40 positive-20 regulation-10 of0 type I1 interferon0 20 production30 40 TNF signaling TNF signaling Total=22Total=17 Asym vs Sym Type V vs Type VI Signaling by TGF-beta Receptor Complex Signaling by TGF-beta Receptor Complex Total=29 Toll-Like Receptors Cascades Toll-Like Receptors Cascades IL6-mediated signalingCD4 events T Cells IL6-mediated signaling events h i AP-1 transcription factor network AP-1 transcription factor network 50 CD69 Interferon gamma signaling SYM Interferon gamma signaling Signal transduction through IL1R ASYM Signal transduction through IL1R TNFAIP3 40 IFN-gamma pathway **** positive regulation of canonical Wnt signaling pathway positive regulation of canonical Wnt signaling pathway CXCR4 IL12-mediated signaling events **** **** PDGFR-beta signaling pathway PDGFR-beta signaling pathway 30 positive regulation of type I interferon production *** DUSP1 RHO GTPase Effectors_Homo sapiens RHO GTPase Effectors_Homo sapiens **** TNF signaling Interleukin-2 signaling_Homo sapiens **** Interleukin-2 signaling_Homo sapiens JUNB **** -Log10 qVal Signaling by TGF-beta Receptor Complex NR4A2 IRS-related events triggered by IGF1R **** IRS-related events triggered by IGF1R 20 CTSW ZFP36 *** Toll-Like Receptors Cascades JUN Signaling by Rho GTPases **** CCL4 Signaling by Rho GTPases NFKBIA *** IL6-mediated signaling events ACTBSH3BGRL3 positive regulation of CD4-positive, alpha-beta T cell activation**** positive regulation of CD4-positive, alpha-beta T cell activation CD55 *** ACTG1 AP-1 transcription factor network 10 CNOT6L DUSP2 Transcription factor CREB and its extracellular signals * Transcription factor CREB and its extracellular signals NPM1 TRAC IFNG GZMH KLF6 *** Interferon gamma signaling ZFP36L2 CORO1A Costimulation by the CD28 family * Costimulation by the CD28 family B2MEMP3 FOS IL32 ZAP70 ** Signal transduction through IL1R FOSB CFL1 IL8- and CXCR2-mediated signaling events **** IL8- and CXCR2-mediated signaling events TNF 0 * -6 -4 -2 0 2 4 6 0 ** 0 10 20 30 40 50 Log2 Fold Change positive regulation of canonical Wnt signaling pathway 10 20 30 40 50 Combined Score Combined Score PDGFR-beta signaling pathway RHO GTPase Effectors_Homo sapiens j Interleukin-2 signaling_Homo sapiens k IRS-related events triggered by IGF1R 11 15 35 Signaling56 58 by ClusterRho GTPases **** **** **** positive ****regulationASYM of/ SYMCD4-positive,Enrichment alpha-beta T cell activation ASYM / SYM TranscriptionH3F3B factor CREBTSC22D3 and itsJUN extracellularBTG1 signalsID2 EIF1 CXCR4 DDIT4 DNAJB1 KLRB1 CostimulationZFP36 by the CD28ZFP36L2 familySRGN HSP90AA1 NFKBIA DUSP1 RGS1 UBB IL8- and CXCR2-mediatedTNFAIP3 FOS signalingUBC eventsBIRC3 11 58 15 35 CD69 JUNB KLF6 DDX5 RhoA Signaling 0 10 20 30 40 50 Th1 Pathway TPT1 CombinedNAC ScoreEEF2 HNRNPA1 FTH1 EEF1A1 NPM1 EEF1D FTL PTMA PCKq Signaling in T Lymphocytes EEF1B2 BTF3 NBEAL COMMD6 ATP5L FAU GLTSCR2 TOMM7 UQCRB Chemokine Signaling GNB2L1 PABPC1 COX4I1 COX7C IL-6 Signaling UBA52 LDHB PFDN5 SARAF Th2 Pathway CD3D CFL1 ARHGDIB VIM HLA-C IL-1 Signaling ATP5E CORO1A LGALS1 S100A10 HLA-B CYBA MYL6 S100A6 ANXA1 HLA-A TNFR2 Signaling ACTG1 GAPDH CD52 B2M TNFR1 Signaling ACTB TRAC S100A4 TMSB4X PFN1 IL32 SH3BGRL3 TMSB10 -3.162 1.476 -4 4 Z-Score Z-Score CD8 T Cells CD8 T Cells
SYM SYM ASYM ASYM
type I interferon signaling pathway type I interferon signaling pathway AP-1 transcription factor network AP-1 transcription factor network granzyme-mediated apoptotic signaling pathway granzyme-mediated apoptotic signaling pathway FoxO family signaling FoxO family signaling Figure 6 Toll-Like Receptors Cascades Toll-Like Receptors Cascades MyD88-independent toll-like receptor signaling pathway MyD88-independent toll-like receptor signaling pathway TNF signaling TNF signaling CD8 T Cells a CCL4 b IL6-mediated signaling events c IL6-mediated signaling events 50 TGF-beta receptor signaling TGF-beta receptor signaling CD69 SYM HIF-1-alpha transcription factor network ASYM HIF-1-alpha transcription factor network 40 DUSP1 NR4A2 **** typenon-canonical I interferon Wnt signaling signaling pathway pathway non-canonical Wnt signaling pathway JUNB AP-1Beta-catenin transcription independent factor network WNT signaling **** Beta-catenin independent WNT signaling 30 **** GNLY ZFP36 granzyme-mediatedTCR signaling in naive apoptotic CD8+ T signaling cells pathway **** TCR signaling in naive CD8+ T cells IFNG ** JUN FoxOp53 pathway family signaling **** p53 pathway -Log10 qVal FGFBP2 CTSW **** Toll-Likepositive regulationReceptors of Cascades cytokine-mediated signaling pathway ** positive regulation of cytokine-mediated signaling pathway 20 CCL4L2 CXCR4 *** CCL3 MyD88-independentPAR1-mediated thrombin toll-like signaling receptor events signaling pathway * PAR1-mediated thrombin signaling events NFKBIA FOS GZMH * FOSB TNFIL8- andsignaling CXCR2-mediated signaling events ** IL8- and CXCR2-mediated signaling events TNF GZMB DUSP2 CYBA CORO1A *** 10 TNFAIP3 B2M IL6-mediatedRAC1 signaling signaling pathway events * RAC1 signaling pathway RAC2 GZMK *** LIMD2 TGF-betaCellular Senescence receptor signaling * Cellular Senescence RGS2 S100A9 HLA-B MYO1F S100A8 *** CCL5 ANXA7 HIF-1-alphaIFN-gamma transcriptionpathway factor network *** IFN-gamma pathway 0 **
-5 0 5 0 * 0 10 20 30 40 10 20 30 40 Log2 Fold Change non-canonical Wnt signaling pathway Combined Score Combined Score Beta-catenin independent WNT signaling d TCR signaling in naive CD8+ T cells e p53 pathway 1 2 3 4 5 7 Cluster * **** **** **** positive**** ASYM regulation/ SYM of /cytokine-mediated Both signaling pathway 3 1 5 7 4 PAR1-mediatedASYM / SYMthrombin signaling events RhoA Signaling Integrin Signaling IL8- andS100A10 CXCR2-mediated3 S100A4 signalingGZMH eventsGNLY A VIM TMSB10 S100A6 NKG7 IL-8 Signaling RAC1 signalingSH3BGRL CD52pathway LGALS1 GZMB Role of NFAT in Regulation of Immune System Cellular Senescence p53 Signaling DDIT4 ZFP36 TSC22D3 H3F3B CCL3 ERK / MAPK Signaling IFN-gammaID2 pathwayJUNB DUSP2 UBB DDX5 RGS1 DUSP1 EIF1 UBC KLF6 Interferon Signaling T Cell Exhaustion Signaling Pathway
0 CXCR4 FOS DNAJB1 SRGN ANXA1 BTG1 10 CD69MacrophageHSP90AA120 CCL4L2Terms30 MT2A 40 Rac SignalingMacrophage Terms ZFP36L2 NFKBIACombinedJUN ScoreCCL4 Hypoxia Signaling in the Cardiovascular System IL-6 Signaling CMC1ASYMCCL5 SYMCD74 GZMK TRBC1 ASYMChemokineSYM Signaling TGF-β Signaling Toll-LikeTPT1 ReceptorsGNB2L1 CascadesPFDN5 BTF3 FTH1 Toll-Like ReceptorsTNFR2 Signaling Cascades EEF1A1 TOMM7 EEF1D HNRNPA1 PTMA TNFR1 Signaling ActivatedUBA52 TLR4EEF1B2 signallingNACA FTL COX4I1 Activated TLR4 signalling MyD88-independent TLR3/TLR4 cascade -3.162MyD88-independent2.530 TLR3/TLR4 cascade -4 4 Z-Score IFN-gamma pathway ZIFN-gamma-Score pathway Interleukin-1 signaling Interleukin-1 signaling IL6-mediated signaling events IL6-mediated signaling events f 35 g extracellular matrix disassembly h extracellular matrix disassembly CXCL8 macrophage activationMacrophage Terms macrophage activation
30 CXCL2 Metabolism of lipids and lipoproteins Metabolism of lipids and lipoproteins CXCL3 Signal transduction through IL1R Signal transduction through IL1R ASYM SYM
25 TNFAIP3 NFKBIA Toll-LikeCXCR4-mediated Receptors signaling Cascades events IL1B **** CXCR4-mediated signaling events ActivatedRAC1 signaling TLR4 signallingpathway **** RAC1 signaling pathway 20 CD83 **** MyD88-independentinterleukin-21-mediated TLR3/TLR4 signaling cascade pathway **** EREG **** interleukin-21-mediated signaling pathway CCL3 IFN-gammaSignalling to pathway ERKs ** Signalling to ERKs -Log10 qVal CCL4 15 **** **** ATF3 **** Interleukin-1Signaling by Hedgehogsignaling Signaling by Hedgehog AREG LYZ IL6-mediatedSignaling by Wntsignaling events **** Signaling by Wnt 10 **** ADIRF extracellularinterleukin-35-mediated matrix disassembly signaling pathway **** interleukin-35-mediated signaling pathway AIF1 GZMA NLRP3 CORO1A GZMM ** macrophageinterleukin-15-mediated activation signaling pathway * MMP19TREM2 HLA-DPB1 interleukin-15-mediated signaling pathway 5 IFNGR2 CCL5 GZMK * LPL MMP14 HLA-E RAC2 MetabolismIL4-mediated of signaling lipids and events lipoproteins * IL4-mediated signaling events CXCL5 TNF JAK1 JAK3 CCL13 TNFAIP6 IL1RN STAT3 CCL7 IL1RAP ICAM3 ERAP2 ** SignalIL8- and transduction CXCR2-mediated through signaling IL1R events * IL8- and CXCR2-mediated signaling events FCER1G 0 *
-10 -5 0 5 0 ** 0 20 40 60 20 40 60 CXCR4-mediated signaling events Combined Score Combined Score RAC1 signaling pathway i interleukin-21-mediated signaling pathway j 2 3 5 6 8 9 10SignallingCluster to ERKs **** **** ** **** **** **SignalingASYM / bySYM Hedgehog / Both ASYM / SYM Signaling by Wnt H3F3B ZFP36 LYZ CXCL8 SAT1 TIMP1 interleukin-35-mediatedE1F1 JUNB IER3 signalingCXCL3 pathwayCCL3L3 H3F3A S100A8 NFKBIA SOD2 CCL3 8 6 9 10 5 2 interleukin-15-mediatedDUSP1 S100A9 IL1B signalingAREG pathwayCCL4L2 FOS FCN1 CXCL2 GOS2 SRGN Role of NFAT in Regulation of Immune Response IL4-mediated signaling events Inhibition of Matrix Metalloproteases IER3 NFKBIA IL1B CXCL2 CXCL8 CXCL3 PPARα / RXRα Activation IL8- and CXCR2-mediated signaling events CD9 APOC1 RNASE1 VIM EMP3 SH3BGRL3 Apoptosis Signaling CD63 APOE MMP9 S100A10 S100A4 GAPDH 0 Inflammasome pathway NPC2 FABP5 20 FN1 ANXA2 40 S100A6 SERF2 60 IL-6 Signaling CTSD CSTB CombinedPSAP LGALS1 Score PFN1 ANXA1 CTSB SPP1 CCL2 S100A11 MYL6 TREM1 Signaling IL-1 Signaling FTH1 FTL CCL18 TMSB10 TMSB4X ACTB iNOS Signaling
HLA-C UBA52 CYBA GPX1 HLA-DRA TNFR2 Signaling HLA-B TPT1 FCER1G HLA-DRB5 CD74 MIF Regulation of Innate Immune Response HLA-A EEF1A1 OAZ1 HLA-DRB1 HLA-DQA1 TNFR1 Signaling B2M TYROBP CD14 HLA-DPA1 HLA-DQB1 FAU AIF1 ATP5E HLA-DPB1 CST3 -4.700 5.477 -4 4 Z-Score Z-Score Figure 7
a -CD4-CD4-CD8-CD8 b -CD4-CD8-Mac-Mac c -Mac Function Interaction CD8CD4CD4CD8 Function Interaction MacMacCD4CD8 Function Interaction Mac INFG Signaling IFNG-IFNGR1 TNF Signaling TNF-TNFRSF1B Macrophage activation APP-LRP1 GO:0060334, GO:0045589 IFNG_IFNGR2 GO:0033209, TNF-TNFRSF1A GO:0043032, ICAM1-ITGAM TNF Signaling TNF-TRAF2 GO:0071356 TNF-TRAF2 GO:0042116, LRPAP1-LDLR GO:0033209, GO:0071356 TNF-TNFRSF1A TNF-LTBR GO:0043031 ICAM1-ITGAX TNF-TNFRSF1B TNF-TNFRSF1B ICAM1-ITGB2 T cell migration and CCL20_CCR6 IFNG signaling IFNG-IFNGR2 Macrophage migration and CCL3-CCR1 GO:0060334, GO:0071346 IFNG-IFNGR1 chemotaxis chemotaxis CCL20-CXCR3 CXCL12-SDC4 TLR signaling and VCAN-TLR2 GO:1905517, GO:0010818, CCL3_CCR4 CCL7-CCR1 macrophage activation VCAN-TLR1 GO:0030335, CCL7-CCR5 GO:0072678, CCL3-CCR5 GO:2000406 GO:0042116, HSP90B1-TLR2 GO:0048247, CCL7-CXCR3 CCL4-CCR5 GO:0043032, HRAS-TLR2 GO:0050918, CCL13-CCR1 CCL4-CCR5 GO:0034123 BGN-TLR2 GO:0010758 CCL13-CCR5 CCL5_CCR4 HSP90B1-TLR7 CCL13-CXCR3 CCL5_SDC4 HSP90B1-TLR1 CCL13-CCR2 CCL5-CCR1 Migration, chemotaxis, and CCL13-CCR2 VEGFB-NRP1 CCL5-CCR5 cytokine signaling CCL13-CCR5 CCL2-CCR5 CCL5-CXCR3 GO:0010759 CCL13-CXCR3 CCL2-CCR2 Notch signaling pathway PSEN1-NCSTN GO:0010818, CCL2-CCR2 CCL5-CCR1 GO:0007219, PSEN1-NOTCH1 GO:0072678 CCL2-CCR4 Inflammatory response and IL1B-IL1R2 GO:0035333 PSEN1-NOTCH2 GO:0019956, CCL20-CCR6 cytokine signaling IL1B-IL1RAP PSEN1-NOTCH3 GO:0019221, CCL20-CXCR3 GO:0006954, IFNG-IFNGR2 Inflam. Resp. and cytokine sig. IL15-IL2RB GO:1905517 CCL3-CCR1 GO:0050729, CXCL10-SDC4 GO:0002675, GO:0050729, IL15_IL15RA CCL3-CCR4 CCL4-CCR1 CCL3-CCR5 GO:0019221 GO:0019221 IL16-CCR5 IL10-IL10RA CCL3L3-CCR5 IL10-IL10RB T cell proliferation IL10-IL10RA CCL4-CCR1 IL18_IL18R1 GO:0042129 IL10-IL10RB CCL4-CCR5 FLT3LG-FLT3 IL10_SIRPG CCL5-CCR5 Macrophage differentiation MMP9-CD44 T cell cytotoxicity and B2M-KIR2DL1 CCL5-CXCR3 GO:0030225 MMP9-ITGAM inhibiting surface receptor sig. B2M-LILRB1 CCL5-SDC4 MMP9-ITGB2 GO:0002767, HLA-A-KIR2DL1 CCL7-CCR1 MMP9-LRP1 GO:0001914 HLA-A-LILRB1 CCL7-CCR2 VEGFA-ITGB1 HLA-C-KIR2DL1 CCL7-CCR5 VEGFA-NRP1 HLA-C-LILRB1 CCL7-CXCR3 VEGFA-SIRPA CXCL10-CXCR3 F13A1-ITGA4 Log2 Fold Change CXCL12-CXCR3 Extracellular matrix TIMP2-ITGA3 CXCL9-CXCR3 organization TIMP2-ITGB1 IL16-CCR5 GO:0030198 ADAM12-ITGB1 Lipid regulation and APP-LRP1 ADAM12-SDC4 ASYM SYM cholesterol efflux APOE-LRP1 ADAM15-ITGA5 GO:0010874, LRPAP1-LDLR ADAM15-ITGB1 GO:0033344 HSP90B1-LRP1 MMP2-SDC2 LRPAP1-LRP1 ADAM9-ITGA3 PSAP-LRP1 ADAM9-ITGA6 SERPINE2-LRP1 Cytokine and chemokine CCL20-CCR6 MMP9-LRP1 signaling CCL20-CXCR3 Immune response HLA-A-LILRB1 GO:0019221, CCL3L3-CCR5 GO:0050776 HLA-C-LILRB1 GO:0019956, IL1RN-IL1R2 Macrophage Differentiation GDF11-ACVR1B GO:0008009 CXCL10-CXCR3 GDF11-ACVR2A CXCL16-CXCR6 T cell mediated immunity JAG1-NOTCH1 CXCL9-CXCR3 GO:0002456 JAG1-NOTCH2 Foam cell differentiation LPL-CD44 JAG1-NOTCH3 GO:0010743, LPL-LRP1 T cell proliferation and IL1B-IL1RAP GO:0010744 ADAM9-ITGAV inflam. response IL15-IL15RA Lipid regulation APOE-LDLR GO:0050727, VEGFA-ITGAV GO:0045834, APOE-LRP1 GO:0042127, VEGFA-ITGB1 GO:0046889, ADAM9-ITGB1 GO:0050729 BTLA-TNFRSF14 GO:0033700, LRPAP1-LRP1 IL15-IL2RB GO:0032368 LRPAP1-SORL1 IL15-IL2RG IL10-IL10RA Log2 Fold Change IL10-IL10RB IL10-SIRPG TCR sig. GO:0050852 BTLA-CD247 ASYM SYM T cell differ. GO:0043382 IL23A-IL12RB1 Log2 Fold Change Extracellular matrix org. MMP9-CD44 GO:0030198 MMP9-ITGB2 d ASYM SYM ASYM SYM
CCL5 CXCL8 CXCL2 PD-1 CXCL5 CCL3 CCL4 MΦ MMP14 Thrombus
IL1RAP MΦ Reparative CD4/8 MMP19 GZMB↑ Pro-Inflammatory Cytotoxic CXCR3 Activated CCR5 IL1β IL1R1 MΦ MΦ TNF IFNγ IFNγ TNF TLR2 LYZ↑ MΦ IFNγ TNF TLR1/2/7
IL1RAP Hb/MΦ IL1RN IL1R1 IL1β IL1β MΦ Subset MΦ Iron clearance HSP90B1 VCAN TNF Pro-inflammatory IFNγ Alternatively Activated PD-1 CCL5 TNF IFNγ HSP90B1 GDF11 CD4 CD8 CD4 CD8 PD-1↑ IFNγ GZMB↑ ↑ T cell Activation GZMB↓ Notch1/2/3 ↑Effector Functions LILRB1 PSEN1 ↑Chemoattraction ↑ T cell Activation/ Differentiation ↑ T cell exhaustion ↑Cytotoxicity CCL20 CCL3 (EOMES, LAG3, PDCD1) CCL4 CCL5 Extended Figure 6 All Significant Interactions a CD4(ligand) - CD8(receptor) CD4(ligand) - CD4(receptor) CD4(ligand) - Mac(receptor)
CD8(ligand) - CD8(receptor) CD8(ligand) - CD4(receptor) CD8(ligand) - Mac(receptor)
Mac(ligand) - CD8(receptor) Mac(ligand) - CD4(receptor) Mac(ligand) - Mac(receptor)
b CD4+ T Cells CD8+ T Cells Macrophages CD4 (Ligand)-CD4 (Receptor) CD8 (Ligand)-CD4 (Receptor) Macrophage (Ligand)-CD4 (Receptor) Notch receptor processing GO:0035333 Lymphocyte migration GO:2000403 IL-15-med. signaling GO:0035723 T cell migration GO:2000404, GO:0072678 Inflammatory response GO:0006954 Response to IL-2 GO:0071352 IL-35-med. signaling GO:0070757 T cell chemotaxis GO:0010818 Response to IFNᵞ GO:0071346 IL-27-med. signaling GO:0070106 cAMP-mediated signaling GO:0019933 T cell activation GO:0002286 Acute inflam. response GO:0002675 CD8 (Ligand)-CD8 (Receptor) T cell chemotaxis GO:0010818 CD4 (Ligand)-CD8 (Receptor) Type I IFN signaling GO:0060337, Macrophage (Ligand)-CD8 (Receptor) T cell activation GO:0002286 T cell activation GO:0002286 Response to IFNᵞ GO:0071346 T cell med. cytotoxicity GO:0001914 T cell med. cytotoxicity GO:0001914 Type I IFN signaling GO:0060337 T cell proliferation GO:0042129 T cell proliferation GO:0042129 Notch receptor processing GO:0035333 Cytokine production GO:0001819 TNF production GO:0032680 Cell-matrix adhesion GO:0007160 T cell med. immunity GO:0002711 CD8 (Ligand)-Macrophage (Receptor) Cell-cell adhesion GO:0034113 CD4 (Ligand)-Macrophage (Receptor) Mᶲ activation GO:0043032 Macrophage (Ligand)-Macrophage (Receptor) ECM organization GO:0030198 Cholesterol efflux reg. GO:0010875 ECM organization GO:0030198 Mᶲ migration and chemotaxis GO:1905517, ECM disassembly GO:0010715 Foam cell differentiation GO:0010743 Cholesterol reg. GO:0010874, GO:0070723 IL-1 and IL-6 reg. GO:0032755, GO:0050716 Mᶲ activation GO:0042116, GO:0043032 Response to IL-1 GO:0070555 Foam cell differentiation GO:0010743 IL-1 and IL-6 reg. GO:0050702, GO:0032755 Foam cell differentiation GO:0010743 Mᶲ migration and chemotaxis GO:1905517 Reverse Cholesterol Transport GO:0043691 Lipid catabolism reg. GO:0050994