The association between poverty and gene expression within peripheral blood mononuclear cells in a diverse Baltimore City cohort

Socioeconomic status (SES), living in poverty, and other social determinants of health contribute to health disparities in the United States. African American (AA) men living below poverty in Baltimore City have a higher incidence of mortality when compared to either white males or AA females living below poverty. Previous studies in our laboratory and elsewhere suggest that environmental conditions are associated with differential gene expression (DGE) patterns in peripheral blood mononuclear cells (PBMCs). DGE have also been associated with hypertension and cardiovascular disease (CVD) and correlate with race and sex. However, no studies have investigated how poverty status associates with DGE between male and female AAs and whites living in Baltimore City. We examined DGE in 52 AA and white participants of the Healthy Aging in Neighborhoods of Diversity across the Life Span (HANDLS) cohort, who were living above or below 125% of the 2004 federal poverty line at time of sample collection. We performed a microarray to assess DGE patterns in PBMCs from these participants. AA males and females living in poverty had the most genes differentially-expressed compared with above poverty controls. Gene ontology (GO) analysis identified unique and overlapping pathways related to the endosome, single-stranded RNA binding, long-chain fatty-acyl-CoA biosynthesis, toll-like receptor signaling, and others within AA males and females living in poverty and compared with their above poverty controls. We performed RT-qPCR to validate top differentially-expressed genes in AA males. We found that KLF6, DUSP2, RBM34, and CD19 are expressed at significantly lower levels in AA males in poverty and KCTD12 is higher compared to above poverty controls. This study serves as an additional link to better understand the gene expression response in peripheral blood mononuclear cells in those living in poverty.


Introduction 46
In 2018, 38.1 million people living in the United States (11.8% of the US population) were living in poverty. African American (AAs) in the US had the highest poverty rate of any twice as likely to have worse cardiovascular health compared with white women (18). 69 Additionally, there is a disproportionately high incidence and prevalence of CVD and poverty-70 related health complications in AA men and women compared with non-Hispanic whites and 71 other ethnicities (24)(25)(26)(27)(28). 72 While considerable work has been spent identifying the environmental risk factors 73 associated with living in poverty, there still remains the challenge of fully understanding how 74 these risk factors translate biologically into disease and health disparities. Biomarkers have 75 brought some clarity into physiological changes associated with SES. For example, adults with 76 early lifetime exposure to poverty and low SES had increased production of cortisol and greater 77 stress-response production of the pro-inflammatory cytokine interleukin-6 (IL-6) and longer 78 duration of IL-6 elevation in circulation (29)(30)(31). 79 Social determinants of health, including racial discrimination (32) and experiencing 80 traumatic stress (33), can transcriptionally activate inflammatory pathways in blood cells. 81 Transcriptional activation or repression of inflammatory genes and related pathways has also 82 been linked with low SES. The whole-blood transcriptional profiles of low SES AAs showed 83 over-expressed inflammatory pathways including those related to interleukin-8 signaling and 84 transcription factor NF-kB signaling (34). AAs experiencing racial discrimination also had 85 elevated expression of NF-kB, AP-1, CREB, and the glucocorticoid receptor compared with 86 European Americans, suggesting stress-response pathways may influence the predisposition to 87 inflammation and related disparate health conditions (32). Several of these genes are part of the 88 conserved transcriptional response to adversity (CTRA) gene network. The CRTA is a stress 89 response pathway mediated by the sympathetic nervous system and beta-adrenergic signaling to 90 induce transcriptional activation of inflammatory genes in the immune system (32,35). In a 91 recent study of over 1,000 diverse and young adults, family poverty status was associated with 92 transcription of interferon-response factors and immune cell activation of dendritic cells and may 93 contribute to lifespan-related predisposition to inflammatory diseases and conditions (36). 94 In 2018, nearly 26.1% of black or AAs were living below the federal poverty line in 95 Baltimore City in Maryland, with individuals earning less than $12,140, and 50% of all 96 individuals living in poverty in Baltimore City had incomes below 50% of the federal poverty 97 line, earning approximately $6,000 per individual (37). Those living in poverty in Baltimore 98 City, particularly AA men, have the highest risk for mortality when compared with other groups 99 influenced by poverty, and this was linked with neighborhood income inequality (24,25). 100 Baltimore City is also the only county on the eastern seaboard north of Washington D.C. to have 101 an average life expectancy at birth below 72 year of age (38). 102 Understanding the biological mechanisms linked with poverty status will help us 103 elucidate how disparate health conditions arise and influence treatment outcomes, particularly in 104 Baltimore City. In this study, we analyzed gene expression within a diverse sub cohort of AA 105 and white, male and female individuals from the Healthy Aging in Neighborhoods of Diversity 106 Across the Lifespan (HANDLS) study living above or below the federal poverty line (28). The 107 objective of this study was to gain insight into the association between poverty status and 108 differential gene expression (DGE). 109 6 was comprised of 54 AAs and whites (W), who were either male (M) or female (F), and self-115 reported living either below (BL) or above (AB) 125% of the 2004 Federal poverty guideline in 116 Baltimore City at the time of PBMC collection (n=6-7/group; see S1 Appendix). There was no 117 significant difference in age, cholesterol or CRP levels, monocyte or white blood cell counts, or 118 history of diabetes, diagnosis of hypertension, or current smoking status between each of the 8 119 groups. After an initial data quality analysis of our microarray raw probe signals, we removed 120 one white male and one white female living below poverty from our analysis for a total N of 52. 121 We used iPathway Guide software (39-42) to identify significantly and differentially-122 expressed genes between groups in our cohort. Genes were considered to be significant if the 123 |logFC| > 0.60 and they had an adjusted false discovery rate (FDR) <0.30. We identified 1,058 124 unique mRNAs as significant in our analysis and dependent upon the comparison being 125 performed. There were 127 genes significantly different between AA men living below poverty 126 compared with AA men living above ( Figure 1A). Of these, seven mRNAs were up-regulated in 127 AA men living in poverty (DUSP2, HLA-DQB1, CD19, MCOLN2, RMBM38, DDX39, and 128 KLF6) and the remaining 120 were down-regulated (see. S2 Appendix for complete list). 129 130 Figure 1: Gene expression changes associated with poverty in the HANDLS cohort. Significantly 131 and differentially-expressed genes were identified in Baltimore City residents living below (BL) 132 vs above (AB) poverty via microarray. Volcano plots generated in iPathway Guide show 133 significantly (|LogFC| >0.6, FDR AdjPvalue <0.3) up-regulated (red) or down-regulated (blue) 134 genes in individuals living in poverty who were (A) African American males (AAM), (B) 135 African American females (AAF), (C) white males (WM), or (D) white females (WF). N= 6/7 136 per group; see S1 Appendix for completely demographics. 137 7 138 We identified 993 transcripts that were significantly different between AA females living 139 below poverty compared with AA females living above ( Figure 1B). Of these transcripts, 70 140 genes were up-regulated and 923 were down-regulated in those living in poverty. Within whites 141 in Baltimore, there were fewer significant genes when comparing those below poverty compared 142 with above. We identified only two mRNAs (MT1E and SSR4) significantly different between 143 white males living in poverty compared with white males living above, of which both transcripts 144 were down-regulated ( Figure 1C). When comparing white females living in poverty with those 145 above poverty, there were 10 transcripts significantly different, three of which were up-regulated 146 (CD68, FBP1, and FN1) and 7 down-regulated (EBF,TMEM16J,MCOLN2,AIM2,BLK,147 FCRL5, and HLA-DQB1; Figure 1D). identify common and unique mRNA profiles associated with poverty in each group. We found 154 that a majority of the 1,058 significant transcripts were unique to each demographic group 155 ( (B) or Race (C) to identify common and overlapping genes. N= 6/7 per group; see S1 Appendix 168 Table 1 for completely demographics. White females below poverty (WFBL) or above poverty 169 (WFAB); white males below poverty (WMBL) or above (WMAB); African American males 170 below poverty (AAMBL) or above (AAMAB); African American females below poverty 171 (AAFBL) or above (AAFAB). 172

173
We observed that few genes overall were significantly different when stratifying by sex 174 and within the same poverty status ( Figure 2B). We observed that 128 genes were unique when 175 comparing WMs and WFs living in poverty, compared with only 6 that were significantly 176 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101 different between AAMs and AAFs. There were 7 genes significantly different between males 177 and females regardless of race or poverty status (PRKY,EIF1AY,RPS4Y1,RPS4Y2,SMCY,178 USP9Y,and CYorf15B), of which nearly all are X-or Y-linked genes ( Figure 2B). When 179 comparing between races and within the same sex and poverty status, we identified 140 genes 180 different between AAMs and WMs and 123 genes different between AAFs and WFs ( Figure  181 2C). Together, we observed the most significant gene changes with respect to poverty status and 182 this remained the primary focus of our analysis. monocyte chemotactic protein-1 production (GO: 0071639). We observed similar patterns for 199 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. including viral myocarditis (H00295), fructose-1, 6-biphosphate deficiency (H00114), and 218 asthma (H00079; Figure 3C). We also conducted a GO and KEGG pathway analysis when 219 comparing our sub cohort groups by sex and race (S1 Appendix Figure 1). A complete list of all 220 pathway rankings for each subgroup and comparisons in Figures 3 and S1 Appendix Figure 1  221 can be found in S1 Appendix. 222 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10. 1101 We decided to focus the remainder of our pathway and gene expression analysis on AA 223 males and females as both groups had the most significant gene expression differences ( Figure 1  224 and 2) when comparing those living below the poverty line with those above in each group. GO 225 analysis identified 1 shared biological process between AA males and females living in poverty 226 (endosome organization; GO: 0007032; Figure 3A and 4A). There were 11 out of 22 genes 227 within the endosome organization process that exhibited significant down-regulation in either 228 AA males and females living in poverty ( Figure 4B Top 10 most significant were long chain fatty-acyl-CoA biosynthetic processes, B-cell 242 differentiation, genetic imprinting, and interleukin-12-mediating signaling ( Figure 4A). AA 243 males living in poverty had 119 biological processes identified in iPathway Guide compared 244 with AA males above poverty. Top significant processes included neutrophil degranulation, toll-245 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.13.20100818 doi: medRxiv preprint like receptor 9 signaling, MyD88-dependent toll-like receptor signaling, and positive regulation 246 of toll-like receptor signaling ( Figure 4A). 247 We also identified unique and overlapping molecular functions and cellular components 248 in AA males and females. Two molecular functions were identified in both AA males and 249 females living in poverty: GDP Binding (GO: 0019003) and single-stranded RNA binding (GO: 250 0003727; S1 Appendix Figure 2A). For both of those GO terms, seven genes were significantly 251 down-regulated in AAs living in poverty compared with above poverty (S1 Appendix Figure  252 2B). Twenty-six molecular functions were unique within AA females, including top significant 253 functions including GTPase activity and binding, nuclear receptor transcription coactivator 254 activity, and hormone receptor binding (S1 Appendix Figure 2A). There were 35 molecular 255 functions unique to AA males living in poverty and top pathways included Double-stranded 256 RNA binding, insulin-like growth factor receptor binding, and toll-like receptor binding (S1 257 Appendix Figure 2A). 258 There was one cellular component GO pathway related to the early endosome (GO: 259 0005769) shared between AA males and females (S1 Appendix Figure 3A). Genes in this 260 pathway were predominantly down-regulated in AAs living in poverty, particularly in AA 261 women, where 32 of the 34 genes in this pathway were significantly down-regulated (S1 262 Appendix Figure 3B). Similar to our GO analysis of biological processes and molecular 263 functions, there were more unique pathways for AA males and females. There were 21 cellular 264 components identified as significant in AA females living in poverty, including the SWI/SNF 265 complex, nuclear heterochromatin, and membrane rafts. Of the 26 pathways significant in AA 266 males living in poverty, some of the most significant included phagocytic vesicle membranes, 267 recycling of the endosome membrane, and SNARE complex (S1 Appendix Figure 3A). 268 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Gene Expression Validation 270
The expression levels of mRNAs that were identified in our microarray analysis were 271 quantified and validated by RT-qPCR in an expanded cohort of AA males living below (n=27) or 272 above (n=29) poverty (S1 Appendix). We chose to validate genes in AA males because this 273 demographic group is at the greatest risk for mortality in Baltimore City (24). We selected genes 274 identified in our microarray analysis within the top 10 up-regulated or down-regulated between 275 AA males living in poverty compared with those living above poverty (Table 1, red), as well as 276 non-significant control genes and genes of interest related to immune signaling or inflammation. 277 We validated that there is a significant difference in the expression of KLF6, DUSP2, 278 RBM38, CD19, and KCTD12 between AA males living in poverty versus those living above 279 poverty ( Figure 5). With the exception of KCTD12, the rest of the genes had lower levels of 280 expression in AA males living in poverty. We were unable to validate the expression levels of 281 CD36, HLA-DQB1, or GIMAP1, which were significant in our microarray but not our expanded 282 cohort ( Figure 5). Collectively, these data indicate that our microarray analysis and validation 283 approach can identify genes differentially-expressed in populations living in poverty. 284 285 Figure 5: mRNA validation in African American males living in poverty. 11 mRNAs identified 286 from our microarray analysis were validated using real-time quantitative PCR in African 287 American males living above (AB) or below (BL) poverty. Genes were quantified using mRNA-288 specific primers and outliers were excluded. All genes were normalized to the combined average 289 of ACTB and GAPDH. Violin plots of the eleven genes compared BL (n=27) vs AB (n=29). 290 *P<0.05, **P<0.01; Student's t-test. 291 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Discussion 293
Our analysis identified several significantly and differentially-expressed mRNAs in 294 PBMCs associated with poverty status in Baltimore City residents. AA males and females living 295 in poverty had more genes differentially-expressed than whites when compared with their above 296 poverty controls. In AA males, we used RT-qPCR to validate the expression of several of the 297 transcripts identified in our microarray within an expanded cohort ( Figure 5). Unexpectedly, 298 several significant genes, including KLF6, DUSP2, RBM38, and CD19, were found to be down-299 regulated in AA males living in poverty, opposite of our initial findings in our microarray. This 300 may be due to the fact that our microarray was performed in six or 7 individuals, as well as the 301 fact that RT-qPCR is much more accurate in single-gene expression validation compared with 302 global discovery analysis (44). We did not identify any transcripts to be significant in our RT-303 qPCR analysis that were not identified in our microarray. 304 The low levels of expression of these transcripts in AA males living in poverty offers 305 several intriguing paths to pursue during follow-up analyses. Kruppel-like factor 6 (KLF6) has 306 well-defined roles as a transcriptional activator in endothelial cells and regulates the repair of 307 vascular damage after tissue injury (45). KLF6 also represses the expression of B-cell leukemia 6 308 (BLC6) in macrophages and monocytes to promote pro-inflammatory gene expression patterns 309 necessary for alleviating tissue damage. When KLF6 was repressed in mouse models, the 310 required inflammatory response from macrophages to promote healing was attenuated (46,47). 311 This response network is also mediated through miR-223 and hypoxia inducible factor 1 alpha 312 (HIF1A), both of which are important node regulators of vascular inflammatory response (48, 313 49). RNA-binding motif protein 38 (RBM38) has identified roles in endothelial repair following 314 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.13.20100818 doi: medRxiv preprint arterial injury (50) and deficient mouse models develop an accelerated aging phenotype and have 315 impaired hematopoiesis (51). 316 We also identified low levels of dual specificity phosphatase 2 (DUSP2) in AA males 317 living in poverty. DUSP2 negatively regulates ERK1 and ERK2 signaling within the MAP 318 kinase signaling pathway and is important for T-cell gene expression regulation (52) We found TLRs to be down-regulated with poverty in our cohort, highlighting the need to further 335 study these genes and related pathways in response to social and environmental stress within 336 those living in impoverished conditions over a long period of time. It is possible adverse 337 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101 conditions affect different communities in different ways on the level of a single gene, yet have 338 similar consequences when looking across multiple genes within the same pathway. 339 AA women living below poverty had the most genes differentially-expressed compared 340 with above poverty controls (Figure 1). There is significant down-regulation of genes related to 341 fatty-acid biosynthesis and metabolism, including PTPLB, ACSL3, and ACSL5, and participants. For example, profiling of young and old AAs and white males living above or 358 below poverty identified hundreds of mRNAs and long, non-coding RNAs associated with 359 poverty in white males. This included enrichment for genes related to stress and immune 360 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.13.20100818 doi: medRxiv preprint signaling in whites living above poverty (62). Poverty status and social adversity are also 361 associated with differential expression of stress-and inflammatory-related genes within the 362 conserved transcriptional response to adversity (CRTA) network (32,(34)(35)(36). There are 53 genes 363 in the CRTA network including cytokines, inflammatory transcriptional regulators (including 364 NF-kB), and members of the Type I interferon response. Their expression is susceptible to 365 environmental stressors (63). When comparing the CRTA gene set with our microarray analysis, 366 we did find some overlap in significant genes. AA males living in poverty have significantly 367 lower levels of interferon induced with helicase C domain 1 (IFIH1), a member of the Type I 368 interferon response pathway. AA females in poverty also expressed IFIH1 at significantly lower 369 levels, in addition to lower levels of inflammatory transcript factor NFKB1, and higher levels of 370 preB-cell receptor immunoglobulin lambda like polypeptide 1 (IGLL1) compared with AA 371 females living above poverty (data not shown). We did not observe any overlap with the CRTA 372 network and expression differences in whites in our study. Together, this suggests additional 373 studies are needed for these over-lapping transcripts, as these CRTA-related genes have been 374 identified in several independent cohorts. 375 Our and other related analyses examining gene expression profiles related to poverty 376 status highlight an important question: how does environmental stress mechanistically induce 377 these changes? Prior analysis of DGE in AA and white women with hypertension identified 378 differentially-expressed microRNA (miRNA) profiles associated with differentially-expressed 379 mRNAs in AA hypertensives (64, 65). miRNAs post-transcriptionally regulate gene expression 380 levels and aberrant expression of these regulatory RNAs can lead to more aggressive diseases, 381 including cancer in AAs (66, 67). We used TargetScan's default settings (68) to predict and 382 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.13.20100818 doi: medRxiv preprint identify possible miRNA regulators of each of the top genes identified in our study in AA males 383 and females (miRNAs listed in S2 Appendix). In vitro confirmation of these results is necessary. 384 Allelic variation and methylation can also influence gene expression levels. Variants 385 likely to promote up-regulation of proinflammatory cytokine genes IL-1 and IL-6 and down-386 regulation of anti-inflammatory IL-10 are more likely found in AA women (69). Ancestral 387 alleles can strongly influence expression quantitative trait loci (eQTLs) governing stress-and 388 pathogen-response gene expression in immune cells (70). Recent studies also suggest that 389 neighborhood disadvantage and SES at birth are correlated with differential CpG methylation 390 states (71, 72). Increased methylation states of inflammatory genes in individuals coming from 391 SES disadvantage have been linked with differential expression of FKBP5, CD1D,F8,KLRG1,392 and NLRP12 (73). Understanding environmental stress response by untangling the connections 393 between allelic variation, epigenetic change, and pre-and post-transcriptional regulation will 394 pinpoint the most important pathways that contribute to gene expression influenced by living in 395 poverty. This may then offer new strategies to prevent these changes from causing downstream 396 health complications in AA males and females and others who live in poverty. 397 There are a few limitations to this study. While the HANDLS study is a representative 398 sampling of residents in Baltimore City, gene expression and pathway validation in additional 399 independent diverse cohorts is warranted to identify if the pathways we have identified here are 400 representative within AAs living in other parts of the United States. Our study was not able to 401 discern which factor of living within poverty is primarily associated with DGE in our cohort, for 402 example food insecurity or nutritional factors, the influence of psychosocial stress or perceived 403 violence, or many of the other complex environmental influences resulting from living in an 404 urban impoverished area. This initial study also does not identify a specific biological 405 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101 mechanism through which poverty transduces physiological responses. It's possible these 406 predispositions are channeled through just one or several key genes unique to each AA and white 407 males and females or that the few overlapping genes identified here are the most relevant. This 408 analysis was a cross-sectional study of gene expression in HANDLS. A longitudinal analysis 409 may provide deeper insight into the most important pathways and genes affected by living in 410 poverty. 411 We analyzed gene expression within a mixed population of peripheral blood 412 mononuclear cells and did not discern whether DGE profiles were associated with unique cell 413 types. Follow-up analyses are warranted to experimentally validate these initial findings and 414 identify if DGE profiles in specific cell types serve as better biological biomarkers for the effect 415 of poverty status on gene expression. Results from our initial analysis can be validated with such 416 an approach as well as inform targeted gene arrays or RT-qPCR validation studies. 417 In conclusion, our study provides additional evidence that gene expression patterns in 418 PBMCs are associated with poverty status, particularly in African Americans. We observed that 419 genes related to endosomal function and toll-like receptor signaling are differentially-expressed 420 in AA males living in poverty. Impoverished environments have long been associated with an 421 elevation in prevalance of health disparities. This work helps elucidate the genetic pathways 422 potentially suspectible to environmental stress in AA males and females. While these findings 423 are promising, a more nuanced examination of what is it about poverty status that influences 424 governs these gene expression differences is needed. Disentangling other factors that are often 425 associated with poverty status such as segregation, high crime, low quality housing, and poor 426 education will be useful for successful interventions. Together, this deeper understanding of the 427 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. was prepared according to the manufacturer's protocol and raw signal data was filtered as 450 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.13.20100818 doi: medRxiv preprint previously described (64, 65). Briefly, probe signal data were subjected to Z-score and FDR 451 normalization to identify outliers. Individual genes with pair-wise z-test P-values <0.05, average 452 intensity >0, |Z-ratios| >1.5-fold, and FDR ≤ 0.3 were considered significant and were based on 453 the maximum score identified from multiple probes of the same gene. The logFC and FDR 454 values were next imported into iPathway for further analysis (see below). The microarray data 455 has been submitted to the Gene Expression Omnibus (GSE149256). 456 457 Gene and Pathway Analysis: We uploaded our microarray gene sets into Advaita iPathwayGuide 458 (Ann Arbor, MI) to identify the top up-and down-regulated genes that were poverty-related (39-459 42). Each gene set for AA males, AA females, white males, and white females compared those 460 living below 125% of the 2004 federal poverty line at the time of sample collection with those 461 living above. Genes were considered significant if |LogFC| >0.6 and FDR AdjPvalue <0.3 to 462 reduce false positives and plotted on volcano plots using iPathway. Our pathway enrichment 463 analysis was complied with the gene ontology (GO) tool of iPathway Guide Program. Gene 464 ontology terms were biological process, molecular function, and cellular components associated 465 with significantly differentially-expressed genes. A P-value correction was applied with an 466 elimination pruning (43) to remove potential false positives for the GO analysis. False positives 467 for the biological and disease pathways tools in iPathwayGuiede were removed with a p-value 468 correction using FDR. 469 470 RT-qPCR: Total RNA from AA males living above or below the poverty line was reverse 471 transcribed into cDNA using SuperScript II reverse transcriptase (Invitrogen) and random 472 hexamers. Levels of each target transcript were quantified by RT-qPCR using 2x SYBR green 473 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.13.20100818 doi: medRxiv preprint PCR master mix (ThermoFisher) and gene-specific primers (see S1 Appendix for forward and 474 reverse sequences). All reactions were run in duplicate using an Applied Biosystems 475 QuantStudio 6 Flex System Real-Time using the 384-well block (Thermofisher). Transcript 476 expression levels were normalized to the combined average of GAPDH and ACTB and compared 477 between AA men living below or above the poverty line using the 2 -ΔΔCt method (74). 478

479
Statistical Analysis: A one-way ANOVA was used when comparing multiple group means for 480 each physiological metric in our cohort and a Fisher's Exact Test was used to determine the the 481 association between poverty status and disease history or smoking status. mRNA expression 482 levels in the validation cohort were examined for Gaussian distribution by measuring skewness, 483 kurtosis, and using a visual inspection of histograms. Outliers from each group for each validated 484 were removed using Grubb's Test with an α = 0.05. Violin plots were generated using GraphPad 485 Prism 8 software. A Student's t-test was used when comparing two groups, unless indicated 486 otherwise. A P-value of <0.05 was considered statistically significant unless otherwise specified. 487 488