http://orcid.org/0000-0001-9732-3321
Figures
Abstract
Increase of peripheral blood CD4 lymphocyte counts is a key goal of combined antiretroviral therapy (cART); most, but not all, recipients respond adequately and promptly. A small number of studies have examined specific genetic factors associated with the extent of CD4 recovery. We report a genome-wide examination of factors that predict CD4 recovery in HIV-infected women. We identified women in in a cohort study who were on cART with viral load below 400 copies, and drew racially and ethnically matched samples of those with good CD4 response over 2 years or poor response. We analyzed the exomes of those women employing next generation sequencing for genes associated with CD4 recovery after controlling for non-genetic factors identified through forward stepwise selection as important. We studied 48 women with good CD4 recovery and 42 with poor CD4 recovery during virologically-suppressive cART. Stepwise logistic regression selected only age as a statistically significant (p<0.05) non-genetic predictor of response type (each additional year of age reduced the odds of good recovery by 11% (OR = 0.89, CI = 0.84–0.96, p = 0.0009). After adjustment for age and genomic estimates of race and ethnicity, 41 genes harbored variations associated with CD4 recovery group (p≤0.001); 5 of these have been previously reported to be associated with HIV infection, 4 genes would likely influence CD4 homeostasis, and 13 genes either had known functions or were members of product families that had functions for which interactions with HIV or effects on lymphocyte homeostasis were biologically plausible. Greater age was the strongest acquired factor that predicted poor CD4 cell recovery. Sequence variations spanning 41 genes were independently predictive of CD4 recovery. Many of these genes have functions that impact the cell cycle, apoptosis, lymphocyte migration, or have known interactions with HIV. These findings may help inform new hypotheses related to responses to HIV therapy and CD4 lymphocyte homeostasis.
Citation: Greenblatt R, Bacchetti P, Boylan R, Kober K, Springer G, Anastos K, et al. (2019) Genetic and clinical predictors of CD4 lymphocyte recovery during suppressive antiretroviral therapy: Whole exome sequencing and antiretroviral therapy response phenotypes. PLoS ONE 14(8): e0219201. https://doi.org/10.1371/journal.pone.0219201
Editor: Cristian Apetrei, University of Pittsburgh Centre for Vaccine Research, UNITED STATES
Received: February 22, 2019; Accepted: June 18, 2019; Published: August 15, 2019
Copyright: © 2019 Greenblatt et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data underlying the results presented in the study are available from https://statepi.jhsph.edu/wihs/wordpress/. Additionally, the accession number for the WIHS in dbGaP genomic data is now provided in the manuscript (phs001503). The WIHS cohort operates under an alternative data sharing plan registered with the National Institutes of Health and access to phenotypic and genomic data can be requested by submitting a Concept Sheet which can be found along with instructions for Concept Sheet submission can be found at https://mwccs.org.
Funding: Data in this manuscript were collected by the Women’s Interagency HIV Study (WIHS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The contents of this publication are solely the responsibility of the authors and do not represent the official views of the National Institutes of Health (NIH). The National Institute of Allergy and Infectious Diseases (NIAID; https://www.niaid.nih.gov) provided grant support to the following co-authors: U01-AI-034989 (RMG, PB, RB, BA); U01-AI-042590 (GS); U01-AI-035004 (KA); U01-AI-034993 (MC); U01-AI-034994 (SK); and U01-AI-031834 (DG).
Competing interests: The authors have read the journal's policy and the authors of this manuscript have the following competing interests: National Institute of Allergy and Infectious Diseases (NIAID; https://www.niaid.nih.gov) provided grant support to the following co-authors: U01-AI-034989 (RMG, PB, RB, BA); U01-AI-042590 (GS); U01-AI-035004 (KA); U01-AI-034993 (MC); U01-AI-103390 (SK); and U01-AI-031834 (DG).
Introduction
Numeric recovery of CD4 T cells (referred to as CD4 cells from here on) in peripheral blood is an important indicator of response to combination antiretroviral treatment (cART). The number of CD4 cells predicts the occurrence of HIV-associated opportunistic infections and cancers.[1–5] Suppression of HIV replication is a major determinant of immunologic recovery during cART. However, even when viremia is optimally controlled, 15–30% of cART recipients have very slow or minimal gains in circulating CD4 cells.[6–8]. Persons with poor CD4 cell recovery after cART-mediated suppression of HIV replication (so-called immunologic non-responders or INRs) may experience continued immune impairment and an increased risk of clinical complications [9–11].
The definition of INR after cART suppression varies in terms of time period and number of cells that is attained. Peripheral blood CD4 cell counts may increase for more than 5 years after initiation of suppressive therapy.[12] However, a CD4 cell gain of 50–150 cells/mm3 in the first year of suppressive therapy is a common standard for a good response, as are subsequent increases of 50–100 cells/mm3 per year. A short term gain followed by a sustained increase defines a treatment response that is considered adequate in the current Guidelines for use of Antiretroviral Agents.[13] Other definitions of CD4 cell count responses to treatment are based on expansion from pre-treatment CD4 counts and/or functional recovery. The varied definitions of immunologic recovery influence the reported proportion of optimal and INR cases; studies applying the most stringent recovery definition tend to report a larger proportion of suboptimal responses.
Poor CD4 cell recovery has been associated with age and telomere length, male sex, hepatitis C co-infection, pre-treatment CD4 counts, pre-treatment HIV RNA viral load (VL), specifics of the cART regimen, and duration of infection prior to cART initiation.[7, 10, 11, 14–20]
Previous studies demonstrated that host genetics influence expansion of CD4 populations during cART. The AIDS Clinical Trials Group examined 137 single nucleotide polymorphisms (SNPs) among 17 genes and found that SNPs of TNFSF10, TNF, BCL2L11, IL15RA, and IL15 were associated with differential CD4 cell recovery at 12 months of cART virologic suppression.[21] IL7R gene polymorphisms were associated with rate of CD4 cell recovery in a large cohort of cART recipients, a finding consistent with the known role of IL-7 in establishing CD4 lymphocyte homeostasis.[22] CCR5 genotype/CCL3L1 copy number was associated with extent and rate of CD4 cell recovery in another cohort.[23] cART recipients who were homozygous for HLA-Bw4 were significantly more likely to be INRs than other individuals.[24, 25] Mitochondrial DNA genotype also influences rate and extent of CD4 cell recovery among cART recipients.[26, 27] The participants of these studies were either predominantly male or the proportion of females was not specified; determination of factors associated with CD4 lymphocyte recovery in females is important because sex dimorphisms are recognized for immune responses, circulating numbers of T cells, and intensity of HIV viremia. In the study of IL-7R haplotype 2 among Ugandan cART recipients[25]; female sex was associated with significantly longer time until reaching a threshold value of 500 cells/mm3, but this association was not independent of other factors such as pre-treatment CD4 cell count and age.
We report herein the associations of demographic and clinical characteristics and exome sequence variations with CD4 cell recovery of women (participants of the Women’s Interagency HIV Study, WIHS), limited to women who experienced rapid expansion of peripheral blood CD4 cell count or women who had slow or minimal expansion of these cells while receiving cART with suppression of plasma VL to below 40 copies/ml.
Materials and methods
Study population
WIHS is a longitudinal, multisite, observational cohort study of HIV infection among U.S. women.[28, 29] Use of cART was defined according to expert HIV treatment guidelines.[13] Our analysis used a dichotomized measure of self-reported adherence of the proportion of doses of antiretroviral medications taken as being either ≥95% or <95% of what the provider instructed over the last six months. Peripheral blood CD4 counts and VL quantitation in plasma were determined via standard assays in laboratories that participated in the NIAID laboratory quality assurance program. The CD4 nadir is defined as the lowest point to which the CD4 count has dropped and is predictive of slower immune recovery and long-term morbidity. Nadir CD4 count was the lowest value measured by WIHS, or pre-WIHS ascertained via medical record review. Anti-Müllerian hormone (AMH) levels in plasma were measured using a commercially available ELISA with a lower limit of detection of 0.08 ng/ml. AMH is an indicator of ovarian follicular reserve, and falls below assay detection 3–5 years prior to menopause, a time that corresponds to the onset of consistent estradiol depletion. [30] Informed consent was provided by all participants via protocols approved by institutional review committees or boards (IRB) at each affiliated institution with consent to studies of host genetics was specifically obtained for women who contributed to this study. The IRBs at the Chicago site were Cook County CORE Center IRB, Cook County Health & Hospitals System IRB, Northwestern University IRB, Northwestern Memorial Hospital IRB, Rush University IRB, Rush-Presbyterian-St. Lukes IRB, University of Illinois at Chicago IRB. The IRBs at the Los Angeles/Southern California site were University of California at Los Angeles Medical Center IRB, Western IRB, University of Hawaii/Kapi’olani Health Research Institute IRB, University of Southern California IRB. The IRB at the New York Brooklyn site was SUNY Downstate Medical Center Institutional Review Board. The IRBs at the New York Bronx site were Albert Einstein College of Medicine IRB, Beth Israel IRB, Mt. Sinai Medical Center IRB. The IRBs at the San Francisco Bay Area site were Alameda Health System IRB, Sutter Health IRB, University of California, San Francisco IRB, California Committee for the Protection of Human Subjects (CPHS). The IRBs at the Washington District of Columbia site were Georgetown University Medical Center IRB, Howard University IRB, Inova Health Systems IRB, Montgomery County Health Department IRB, Whitman Walker Clinic IRB.
Phenotype determination
Study visits during reported cART use were categorized according to whether VL was suppressed and according to whether CD4 counts demonstrated adequate gains over pretreatment and nadir values. Imputed values for CD4 counts or VL were applied to one or two visits with missing values for these measures if there was no change in cART status between at least two nearby visits with the measurements available. To mitigate volatility in CD4 counts, non-missing CD4 counts were smoothed using values from two or three nearby visits, when available. Detailed specification of the phenotyping algorithm is provided in online S1 Appendix, and the SAS program implementing it is in online S2 Appendix. These appendices include programming for phenotypes that were not used for this study.
Selection for this study
Initial selection for this study was completed using the phenotyping detailed in the preceding section, and was based on the initiation of cART, having plasma VL below assay detection (e.g., <80 copies) for ≥1 year, and having either the most rapid and durable gains in CD4 counts for at least 2 years or the poorest gains for at least one year during cART visits with viral suppression. Because race is associated with total leukocyte counts in peripheral blood, we matched the good and poor responder groups by race and ethnicity. Our goal was to select 50 women with rapid CD4 cell responses and 50 with slow CD4 cell responses, matched for race and ethnicity. Initial screening with the algorithm (online S1 appendix) identified 201 rapid responders and 78 slow responders. We reviewed longitudinal plots of viral load, CD4 cell count, and treatment status (examples in Fig 1) stratified by race (i.e., African-American, White) and Hispanic ethnicity, and identified 46 clearly slow responders. We identified 50 race-matched rapid responders with similarly clear and rapid CD4 cell expansion after initiation of suppressive cART.
Example plots of CD4 and VL changes overtime for a good CD4 responder (1A) and a poor responder (1B). Fig 1A legend. WIHS Study Visit denotes visit number (2 visits per year). The vertical black line at visit 16 denotes the timepoint at which the beginning of the phenotypic window begins, which is followed by a period of rapid CD4 T cell gain (orange bracket). The following graphics were used to display relevant clinical, disease, and treatment information: dashed blue line denotes CD4 T cell count, solid green line denotes nadir CD4 T cell count, dashed black line denotes HIV viral load, HIV antiretroviral therapy (ART) status was denoted using ◯ (no ART) or ▲ (combination ART (cART) visit, and HIV viremia status was denoted using ◯ designated in blue font (HIV RNA below the assay detection limit) or ✖ designated in blue font (HIV RNA greater than or equal to the assay detection limit. The visit phenotype was denoted using the following symbols: U (not determined due to the lack of sufficient observation), designated in red font (long term off cART and viremic with stable CD4), l designated in green font (long term off cART and viremic with declining CD4), H (first cART visit), H designated in red font (on cART with viremia), or H designated in green font (on cART with virologic suppression). Fig 1B legend. WIHS Study Visit denotes visit number (2 visits per year). The vertical black line a visit 12 denotes the timepoint which is followed by a period of minimal CD4 T cell gain (orange bracket). The graphics were used to display relevant clinical, disease, and treatment information are the same as for Fig 1A, except that the visit phenotype had the following additional symbols: s designated in red font (short term off cART with viremia and declining CD4), H designated in green font (reinitiated cART visit), or h designated in green font (virologic suppression with minimal CD4 gain).
Ninety-two of the 96 initially selected women had available samples for exome sequencing (examples provided in Fig 1). Two additional eliminations occurred following genetic data collection; one had an inadequate genomic material, and exome sequencing failed in the second, resulting in a final sample of 90 women.
Statistical analyses
We used forward stepwise logistic regression to select variables from a set of demographic, clinical, and treatment variables, for inclusion in a final model of acquired factors associated with good recovery. Predictors were required to have a p-value for association with CD4 response group of <0.05 in a multivariate regression. SAS version 9.4 was used for these analyses.
Whole exome sequencing.
Banked peripheral blood mononuclear cells underwent genomic DNA extraction using the Puregene DNA Isolation System (Invitrogen, Carlsbad, CA), quantitated, and normalized to a concentration of 45 ng/mL. Approximately 6.75 micrograms of DNA was subjected to sequencing at an average depth of read of 50X. Sequencing was performed by Expression Analysis Inc. (Durham, NC). Prior to library construction and sequencing, DNA was evaluated for evidence of degradation by electrophoresis.
Exome sequencing was completed with 100 base-pair paired-end sequencing. A PhiX control sample was included on all flow cells. DNA fragments were amplified using Clonal Single Molecule Array technology and Sequencing-by-Synthesis using Reversible Terminator Chemistry. Sequence data was stored as FASTQ files for each specimen. Sequences were required to pass Illumina (San Diego, CA) purity filter reads and was accompanied by a per-base quality score, as defined by Illumina-scaled Phred metric.
Variant calling file generation procedure.
Paired-end whole-exome reads were processed and aligned to the human genome. The alignments were used to call variants in order to create the Variant Calling Format (VCF) file using the Genome Analysis Toolkit from The Broad Institute (GATK) following best practices.[31, 32] A diagram of the VCF file generation procedure is provided in S1 Fig. In brief, paired-end reads were clipped from adapters using fastq-mcf from the EA-utils package (https://code.google.com/p/ea-utils/). Paired-end libraries were evaluated for quality using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Paired-end reads were aligned to the reference genome assembly GRCh37/hg19. Alternate loci and patches were excluded. Alignments were performed with bwa-mem (http://bio-bwa.sourceforge.net).[33] Variant calling was performed using GATK version 3.4–46 (https://www.broadinstitute.org/gatk).
Alignments were processed with samtools (http://www.htslib.org)[34] to remove PCR duplicates. Read Groups lines were defined in the BAM file using AddOrReplaceReadGroups in picard-tools (http://picard.sourceforge.net). Re-alignment was performed in regions around known indels with the RealignerTargetCreator and IndelRealigner in GATK using the gold standard indels from Mills, the 1000 Genomes Project (KGP), and the KGP phase 1 indels (GATK Resource Bundle, version 2.8, ftp.broadinstitute.org). Recalibration of the base quality scores was performed to assign an empirically accurate error model to the bases with BaseRecalibrator/PrintReads in GATK using the known variation from the gold standard indels from in the above listed sources and dbSNP 138. Variant calling was performed per sample with the GATK HaplotypeCaller in GCVF mode. All samples were called together to gain more confidence in weak calls with GenotypeGVCFs and dbSNP 138. Variants were recalibrated using VariantRecalibrator using known variation data from dbSNP 137, HapMap 3.3, and KGP Omni 2.5. A Variant Quality Score logarithm of odds was calculated for each variant using ApplyRecalibration (filer level = 99.0). Final annotations for variant filtration were performed with SelectVariants for approximate read depth (DP<8) and genotype quality (GQ<20) as previously suggested.[35] Only SNP variation was called. Genomic data for the WIHS is indexed in the database of Genotypes and Phenotypes (dbGaP); the accession number is phs001503.
Genetic association analyses.
Allele and genotype frequencies were determined by gene counting. Population substructure in good and poor CD4 response groups was estimated by principal component analysis (PCA). The first two principal components (PCs) were sufficient to estimate the substructure of the cohort due primarily to genetic racial and ethnic differences, which were employed to control for the potential confound of ancestry-based substructure in the multiple regression analyses.
The approaches used for association analysis in host genomic studies typically focus on common polymorphisms but are poorly suited to detect the impact of rare polymorphisms and cannot estimate their combined effect or “variant burden”. In order to leverage the availability of common and rare polymorphisms yielded from whole exome sequencing, two approaches to estimate the magnitude of variant burden with the CD4 recovery phenotype were considered: the Combined Multivariate and Collapsing (CMC) method[36] and the Kernel-Based Adaptive Collapsing (KBAC) method[37] Both methods collapse variants into a single covariate based on gene regions (i.e., a gene). CMC can detect the effects of both common and rare alleles, while KBAC detects effects of rare alleles only. Gene associations identified by both CMC and KBAC would suggest that both common and rare variants contribute to the observed association, while gene associations identified by CMC or KBAC alone, would suggest that the variant burden at said gene was due primarily to common or rare variants, respectively. Both unadjusted and adjusted (i.e., age, genomic estimates of population substructure) associations were calculated. For CMC and KBAC, gene regions were defined by RefSeq release 63. Statistical analyses were carried out using HelixTree software (Golden Helix, Bozeman, MT).
The CMC method groups variants according to the frequency of each polymorphism within a gene region to perform a multivariate test collapsed over the gene region to estimate the association with the phenotype. Both common (i.e., ≥5% minor allele frequency [MAF]) and rare (i.e., <5% MAF) SNPs are included in CMC analyses. In the current study we defined five potential groups, or bins, based on the frequency of the rare allele for each SNP as provided by KGP: <1%, 1%≤x<2.5%, 2.5%≤x<5%, 5%≤x<10%, ≥10%. SNPs identified by whole exome sequencing in WIHS not present in the KGP reference database were assigned to the bin defined as <1% because the variant was assumed to be too rare to be detected in KGP.
The KBAC method collapses the genotypic information across all polymorphisms in a gene region into compound, multi-marker genotype without the need for binning. However, common SNPs (i.e., ≥5% MAF) must be excluded from analysis. The counts of the multi-marker genotypes are used to perform a multivariate test to estimate their association with the outcome. The KBAC test weights multi-marker genotypes based on risk for the outcome with increased risk genotypes assigned higher weights to better distinguish between causal and non-causal genotypes. Accordingly, the KBAC test is a one-sided test.
Initially, both CMC and KBAC regression analyses were performed at the level of the gene with all transcripts consolidated into a single region. In order to evaluate the possibility that associations may be narrowed to specific transcripts, the CMC and KBAC regression analyses were repeated for every transcript only for those gene regions that met a priori significance thresholds to minimize the added multiple testing penalty. For candidate gene analyses, a significance threshold of p<0.05 was selected a priori for the evaluation of previously identified candidate genes for CD4 recovery. Based on the exploratory nature of the exome analyses using both CMC and KBAC covariate-adjusted models, we report here the genetic predictor terms that had a gene-wise permutation-based p<0.001 by either CMC or KBAC after one million permutations, a threshold that is more appropriate when conducting gene-wise association studies.[38] The CMC analyses evaluated 18,987 genes while KBAC analyses evaluated 19,653 genes; genes associated with CD4 recovery using either method where retained.
Genes that were associated with CD4 recovery group were crossed referenced via the HIV:host protein interaction database (https://www.ncbi.nlm.nih.gov/genome/viruses/retroviruses/hiv-1/interactions/). Both direct (i.e., direct host protein:viral protein) and indirect (HIV protein interaction with a host protein that interacts with the protein encoded for by an identified CD4 recovery gene) were considered.
Results
Clinical factors
Ninety WIHS participants contributed to analyses reported here; 42 met criteria for poor CD4 recovery on virologically-suppressive cART and 48 women demonstrated good CD4 expansion on virologically-suppressive cART. Fig 1 provides examples of CD4 and VL plots for women demonstrating (Fig 1A) good (n = 48) and (Fig 1B) poor (n = 42) CD4 gain.
Table 1 summarizes the characteristics of the two CD4 response groups (white columns), and the associations estimated by univariate logistic regression (gray columns). Even though we limited analysis to women who had HIV RNA levels below assay threshold during the CD4 response time period, ≤95% adherence to the prescribed regimen was common. CD4 nadir was associated with CD4 response phenotype (OR = 2.0 per 100 cell/ml, 95% CI 1.34–3.1, p = 0.0008). However, CD4 count and the amount of increase above CD4 nadir during the response period are used in the CD4 response group definition, so the meaning of this association is questionable.
Age at the onset of the response-defining treatment period was strongly associated with CD4 response group (OR for a good response = 0.89 per year, 95% CI = 0.838–0.955, p = 0.0009). Being a tobacco smoker at start of the CD4 response phenotype was also associated with CD4 response group: OR for a good response = 0.413, 95% CI = 0.175–0.976, p = 0.044). Plasma AMH level was available for 87 women. Undetectable AMH was associated with poor response group: OR for a good response = 0.221, 95% CI = 0.087–0.564, p = 0.0016.
All women reported current or past use of nucleoside antiretroviral drugs, though prior use of ddI or d4T was associated with poor CD4 response (OR for a good response = 0.333, 95% CI = 0.116–0.955, p = 0.041). Self-reported adherence of >95% to the cART regimen was associated with a good CD4 response (OR = 3.300, 95% CI = 1.288–8.5, p = 0.013).
Most of these covariates are associated with each other. For current smokers age and years smoked will line up exactly, and even for former smokers there will likely be a relation. Because ovarian follicular reserve declines with age, AMH detection also declines with age. ddI and d4T were used earlier in the epidemic, and so will also be associated with age. To sort these out we constructed a multivariate model with forward stepwise logistic regression, with p<0.05 for inclusion. Nadir CD4 count was not included as a candidate predictor in the model because it is linked with criteria for group assignment. Age had the smallest p-value on its own, and none of the other variables had p<0.05 after controlling for age. The confidence intervals for many of those other predictors include substantial effects, so one should not conclude this finding to mean they don’t matter. It just means we don’t know much about how or if they matter.
However, our decision rule generates a final model of non-genetic factors that includes only age. Nadir CD4 count was not included as a candidate predictor in the model because it is linked with criteria for group assignment.
Whole exome and candidate gene analyses
Table 2 details the 41 genes that had p≤0.001 for predicting response group by CMC (a regression analysis approach that evaluates the genetic burden of both common and rare alleles) and/or KBAC (a regression analysis approach that evaluates the genetic burden of rare alleles), along with genes previously reported to be associated with HIV-related outcomes (i.e., candidate genes). The reported values are from regression analyses controlling for age (based on the multivariate analysis of non-genetic variables discussed immediately above) and the first two ancestry principal components that were included to account for any residual variability in population substructure due to race and/or ethnicity after matching on self-reported race and ethnicity. Five of these genes were previously reported to be associated with HIV-related outcomes, including HECW2, CLCN3, ALG2, SRP14, and CCL25. Another four genes have functions that would likely influence CD4 lymphocyte homeostasis; these were: SLC12A2, USP12, GMIP, KLC3. Eleven of the genes associated with CD4 response had unknown functions. The remaining 21 genes either had known functions or were members of product families that had functions for which interactions with HIV or effects on lymphocyte homeostasis were biologically plausible. None of the 8 candidate genes previously reported to be linked with HIV treatment responses were statistically significantly associated (p<0.05) with CD4 response group in this study. A detailed description of the estimates of association for both CMC and KBAC is provided in online S1 Table and the polymorphisms incorporated into each gene level burden regression analysis is provided in online S2 Table.
Discussion
Several predictors of CD4 cell response group among suppressed (by clinical assay) cART recipients were identified in this study. Of the non-genetic factors, only age was statistically significantly associated with CD4 cell response group in multivariable analysis. Greater age was associated with lower likelihood of a good CD4 recovery response, a finding that has been previously reported.[86, 87] This is in keeping with the loss of immune function that is known to occur with aging
Some studies have reported that females have higher circulating numbers of CD4 lymphocytes than males[88, 89], but substantial sex differences in CD4 cells have not been consistently observed.[90] Some reports note that sex differences in CD4 cell counts is driven by higher counts among women less than 50 years of age.[89] We found that a biomarker of gonadal aging (ovarian follicular function) was associated with recovery group, but the association was not statistically significant when adjusted for chronological age.
Many of the genes identified in this study appear to participate in processes related to lymphocyte homeostasis and thymus activity. Some genes that are known to interact with HIV also were associated with CD4 lymphocyte response group.
The two approaches we employed to evaluate genetic burden are related, but provide distinct information. While each analysis estimates genetic burden, CMC evaluates both rare and common polymorphisms and identified 23 genes that suggests that the basis of these associations is due primarily to common and not rare variant burden, while KBAC evaluates rare polymorphisms and identified 17 genes that suggests that the basis of these associations is due primarily to rare variant burden. Only one gene was identified by both CMC and KBAC (i.e., CCL25), indicating that both rare and common variants of this gene may influence CD4 cell recovery pattern.
Further investigation of the 41 positional candidate genes revealed that 9 encoded for RNA products, while the remainder encoded for proteins. Twenty-five of the 32 protein-coding genes encoded for proteins that interacted with HIV proteins (Table 2, last column). The 9 RNA-encoding genes have not been evaluated for their potential HIV protein interactions. However, the expression of microRNAs often differs by HIV status and disease characteristics.
In a post-hoc analysis, we identified publically available gene expression data collected in a sample of South Asian men (n = 10) and women (n = 2) who exhibited expected (n = 5) and poor (n = 7) CD4 T-cell recovery after cART initiation (Gene Expression Omnibus [GEO] expression set GSE77939) which we interrogated to determine if a subset of the 41 genes that were enriched for sequence variations associated with poor CD4 cell recovery also exhibited differences in gene expression. The definitions of expected and poor CD4 cell recovery were loosely similar to ours, with HIV-positive individuals required to take ART for at least one year who did (expected responders) and did not (poor responders) experience a CD4 lymphocyte count gain of more than 150 cells/μL in one year and a viral load <400 copies/mL). Additionally, non-responders consistently had CD4 cell counts of less than 250 cells/uL, while expected responders showed a CD4 count that exceeded 250 cells/μL (personal communication, S. Aurora). Gene expression measures for 7 (i.e., CTAGE7P, MIR3662, MIR5010, MIR6769A, MIR6784, RP11-383C5.3) of the 41 candidate genes identified in our sample were not available. Of the 34 genes that we identified by whole exome sequencing for which gene expression data was also available, seven were differentially expressed between CD4 responder groups (i.e., ALG2, GIMP, GRHL2, LCE1D, SLC12A2, TMEM121, ZNF493; all p<0.05) based on GEO2R analysis. These gene expression findings provide complementary support that the identified candidate genes that differed in terms of enrichment of sequence variations by CD4 cell recovery group also show differences in gene expression. Whereas the number of genes that were differentially expressed exceed that expected by chance (i.e., 7 genes were differentially expressed where approximately 2 would be expected by chance alone, p<0.00022), it is also possible that a subset of the other candidate genes may also be differentially expressed and associated with CD4 cell response to ART but in other tissues (e.g., thymus) or at different times (e.g., viral load set-point). Future studies are warranted that examine gene expression and sequence variation (e.g., expression-quantitative trait locus) analyses in relation to CD4 lymphocyte response to treatment.
Further bioinformatic analyses to evaluate for a biological pattern(s) across the genes were identified using KEGG gene set enrichment analyses. One pathway, hsa00260 (glycine, serine, and threonine metabolism) featured two genes associated with CD4 recovery: CDCH and GCAT. Enrichment of genes in the hsa00260 pathway has been highlighted in several studies of HIV infection.[91, 92] Because we identified HIV-related genes that were associated with CD4 cell recovery pattern, it is possible that ongoing HIV replication at very low levels might have been a contributor to CD4 cellular expansion. Thus, when specimens were available, we conducted ultrasensitive HIV RNA assays to detect HIV RNA quantities below the threshold of clinical assays in use at the time of this study. These ultra-sensitive assays utilized Hologic (San Diego, CA) using the Hologic Aptima® HIV-1 Quant DX Assay on the Panther System with up to 9 replicates of each sample and can detect HIV RNA levels as low as 1 copy/ml. Thirty-eight of 43 good CD4 cell responders with available samples for testing had ≥1 replicate test that had detectable HIV RNA. All 10 of the available poor CD4 cell responder samples had detectable HIV RNA in ≥1 replicate. This finding indicates that persistent viral replication exists, even among individuals thought to have optimal viral suppression. However, despite the frequency of ongoing low-level replication, most cART recipients experience rapid expansion of CD4 lymphocyte populations. It is possible that individuals vary in terms of how on-going low-level viral replication impacts CD4 recovery. The fact that some of the variations in genes we found to be predictive of CD4 response group had known relationships with HIV also supports a continuing role of HIV replication in the setting of host differences in determining immunologic recovery.
Conclusions
This study confirms the finding that increasing age is an independent and strong predictor of poor CD4 recovery during cART. We found that age, in a group of mid-life women, was a better predictor of CD4 recovery than AMH, a biomarker of gonadal aging. This result suggests that loss of ovarian estrogen is not a leading determinant of CD4 recovery during cART. A major contribution of this study, which broadly searched for relationships between host genetics and CD4 cell recovery, is the identification of new genes that may influence immune reconstitution in treated HIV infection. These findings may lead to novel hypotheses and identification of new host factors which contribute to HIV pathogenesis and recovery.
Supporting information
S1 Fig. Flow diagram of variant calling file generation.
Object shapes indicate programs (ovals) or data files (rectangles). Arrows connect program inputs and outputs. Rectangle color indicates the type of data: study data (green), reference data (blue), and intermediate processed data (maroon). Abbreviations: 1000G = 1000 Genomes; dbSNP = Nartional Center for Biotechnolgy Institute SNP database; GATK = Genome Analysis Tookit from The Broad Institute; Hapmap = The haplotype map database (www.hapmap.org); hg19 = Human Genome UCSC assembly 19; SNP = single nucleotide polymorphism; VCF = Variant Calling Format.
https://doi.org/10.1371/journal.pone.0219201.s001
(TIFF)
S1 Table. Detailed estimates of gene-wise associations using CMC and KBAC.
The study design is listed in column A and is entitled “StudyDesign” which has one of two values: “Novel Loci” identified by exome sequencing and “Candidate Gene” because the given gene was specified a priori. The gene abbreviation is specified in column B. The chromosome location of the gene is listed in column C. The gene identifier is listed in column D. The gene name is defined in column E. The nucleotide start position for the gene is listed in column F. The nucleotide end position for the gene is listed in column G. The gene transcript identifier is listed in column H. The strand (i.e., “+”, “-“) encoding the gene is listed in column I. The total number of markers (SNPs) included in the CMC-based associated test is listed in column J. The number of markers included in each of the five possible bins for the CMC-based test of association are listed in columns K through O (i.e., CMC Bin0 Count, CMC Bin1 Count, CMC Bin2 Count, CMC Bin3 Count, CMC Bin4 Count). The unadjusted p-value for the omnibus CMC-based test of association for each gene is provided in column P. The unadjusted estimate of association (beta and associated standard error [SE]) for the omnibus CMC-based test of association is listed in column Q. The unadjusted estimate of association (beta and associated standard error [SE]) for the first bin (“0”) is listed in column R. The unadjusted estimate of association (beta and associated standard error [SE]) for the second bin (“1”) is listed in column S. The unadjusted estimate of association (beta and associated standard error [SE]) for the third bin (“2”) is listed in column T. The unadjusted estimate of association (beta and associated standard error [SE]) for the fourth bin (“3”) is listed in column U. The unadjusted estimate of association (beta and associated standard error [SE]) for the fifth bin (“4”) is listed in column V. The Chi-squared value statistic for the likelihood ratio test of the adjusted estimate of association as compared to the unadjusted estimate of association, when adjusting for genomic estimates of race and ethnicity (i.e., principal component 1 [PC1], principal component 2 [PC2]), is listed in column W. The adjusted estimate of association (beta and associated standard error [SE]), when adjusting for genomic estimates of race and ethnicity (i.e., principal component 1 [PC1], principal component 2 [PC2]) and age, for the omnibus CMC-based test of association is listed in column X. The adjusted estimate of association (beta and associated standard error [SE]) is listed in column Y. The adjusted estimate of association (beta and associated standard error [SE]) for PC1 is listed in column Z. The adjusted estimate of association (beta and associated standard error [SE]) for PC2 is listed in column AA. The adjusted estimate of association (beta and associated standard error [SE]) for age is listed in column AB. The adjusted estimate of association (beta and associated standard error [SE]), for the first bin (“0”) is listed in column AC. The adjusted estimate of association (beta and associated standard error [SE]), for the second bin (“1”) is listed in column AD. The adjusted estimate of association (beta and associated standard error [SE]), for the third bin (“2”) is listed in column AE. The adjusted estimate of association (beta and associated standard error [SE]), for the fourth bin (“3”) is listed in column AF. The adjusted estimate of association (beta and associated standard error [SE]), for the fifth bin (“4”) is listed in column AG. The Chi-squared value statistic for the likelihood ratio test of the adjusted estimate of association as compared to the unadjusted estimate of association, when adjusting for genomic estimates of race and ethnicity (i.e., principal component 1 [PC1], principal component 2 [PC2]), and age, and non-adherence, is listed in column AH. The adjusted estimate of association (beta and associated standard error [SE]), when adjusting for genomic estimates of race and ethnicity (i.e., principal component 1 [PC1], principal component 2 [PC2]) and age, for the omnibus CMC-based test of association is listed in column AI. The adjusted estimate of association (beta and associated standard error [SE]) is listed in column AJ. The adjusted estimate of association (beta and associated standard error [SE]) for PC1 is listed in column AK. The adjusted estimate of association (beta and associated standard error [SE]) for PC2 is listed in column AL. The adjusted estimate of association (beta and associated standard error [SE]) for age is listed in column AM. The adjusted estimate of association (beta and associated standard error [SE]) for non-adherence is listed in column AN. The adjusted estimate of association (beta and associated standard error [SE]) for adherence is listed in column AO. The adjusted estimate of association (beta and associated standard error [SE]), for the first bin (“0”) is listed in column AP. The adjusted estimate of association (beta and associated standard error [SE]), for the second bin (“1”) is listed in column AQ. The adjusted estimate of association (beta and associated standard error [SE]), for the third bin (“2”) is listed in column AR. The adjusted estimate of association (beta and associated standard error [SE]), for the fourth bin (“3”) is listed in column AS. The adjusted estimate of association (beta and associated standard error [SE]), for the fifth bin (“4”) is listed in column AT. The total number of markers (SNPs) included in the KBAC-based associated test is listed in column AU. The number of multi-marker genotypes estimated for the KBAC test is listed in column AV. The unadjusted p-value for the KBAC-based test of association for each gene is provided in column AW. The KBAC test statistics (i.e., the KBAC score) for the unadjusted estimate of association is listed in column AX. The adjusted (i.e., genomic estimates of race and ethnicity [PC1, PC2], age) p-value for the KBAC-based test of association for each gene is provided in column AY. The KBAC test statistics (i.e., the KBAC score) for the PC1-, PC2-, and age-adjusted estimate of association is listed in column AZ. The adjusted (i.e., genomic estimates of race and ethnicity [PC1, PC2], age, self-reported non-adherence) p-value for the KBAC-based test of association for each gene is provided in column BA. The KBAC test statistics (i.e., the KBAC score) for the PC1-, PC2-, age-, non-adherence-adjusted estimate of association is listed in column BB.
https://doi.org/10.1371/journal.pone.0219201.s002
(XLSX)
S2 Table. Single nucleotide polymorphisms incorporated into each gene burden analysis.
The study design is listed in column A and is entitled “StudyDesign” which has one of two values: “Novel Loci” identified by exome sequencing and “Candidate Gene” because the given gene was specified a priori. The chromosome location of the gene is listed in column B. The gene abbreviation is specified in column C. The gene identifier is listed in column D. The nucleotide position is listed in column D. The RefSeq identifier (rsID) is provided, if available, in column E. The reference allele is provided in column F. The excess heterozygosity ratio statistic, R, is listed in column G. The p-value for Fisher’s Exact test for the association of a given SNP with the CD4 Recovery Group outcome is listed in column H. The odds ratio (OR) and the 95% confidence interval (95% CI) of the association of a given SNP with the CD4 Recovery Group outcome is listed in column I. The sample size is listed in column J. The SNP call rate is listed in column K. The reference minor allele frequency (MAF) is provided in column L. The p-value for the Fisher’s Exact Test for the deviation of a given SNP from the Hardy-Weinberg expectation is listed in column M.
https://doi.org/10.1371/journal.pone.0219201.s003
(XLSX)
S2 Appendix. Detailed specification of the phenotyping algorithm.
https://doi.org/10.1371/journal.pone.0219201.s005
(PDF)
Acknowledgments
The authors thank Dr. Geralyn Messerlian for provision of the AMH measures. We also thank Andrew Worlock and Sangeetha Nair of Hologics for providing the ultrasensitive HIV RNA assays used in this study and Sheila Keating for her assistance in arranging this testing. We also thank Sunilkumar Arora for providing additional information on the gene expression data that he deposited in GEO. We particularly thank the participants of WIHS, whose diligence in attending study visits provides such valuable information. De-identified participant data (i.e., clinical, demographic, treatment, next generation DNA sequencing) is securely maintained by the WIHS and is available on request after securing an approved Concept Sheet (https://mwccs.org).
Data in this manuscript were collected by the Women’s Interagency HIV Study (WIHS). The contents of this publication are solely the responsibility of the authors and do not represent the official views of the National Institutes of Health (NIH). WIHS (Principal Investigators): UAB-MS WIHS (Mirjam-Colette Kempf and Deborah Konkle-Parker), Atlanta WIHS (Ighovwerha Ofotokun and Gina Wingood), Bronx WIHS (Kathryn Anastos and Anjali Sharma), Brooklyn WIHS (Howard Minkoff and Deborah Gustafson), Chicago WIHS (Mardge Cohen and Audrey French), Metropolitan Washington WIHS (Seble Kassaye), Miami WIHS (Margaret Fischl and Lisa Metsch), UNC WIHS (Adaora Adimora), Connie Wofsy Women’s HIV Study, Northern California (Ruth Greenblatt [current manuscript lead author], Bradley Aouizerat, and Phyllis Tien), WIHS Data Management and Analysis Center (Stephen Gange and Elizabeth Golub), Southern California WIHS (Joel Milam) (WIHS I–WIHS IV).
References
- 1. Yanik EL, Napravnik S, Cole SR, Achenbach CJ, Gopal S, Dittmer DP, et al. Relationship of immunologic response to antiretroviral therapy with non-AIDS defining cancer incidence. AIDS. 2014;28(7):979–87. pmid:24681415; PubMed Central PMCID: PMC4040952.
- 2. Yanik EL, Napravnik S, Cole SR, Achenbach CJ, Gopal S, Olshan A, et al. Incidence and timing of cancer in HIV-infected individuals following initiation of combination antiretroviral therapy. Clin Infect Dis. 2013;57(5):756–64. pmid:23735330; PubMed Central PMCID: PMC3739467.
- 3. Abraham AG, D'Souza G, Jing Y, Gange SJ, Sterling TR, Silverberg MJ, et al. Invasive cervical cancer risk among HIV-infected women: a North American multicohort collaboration prospective study. J Acquir Immune Defic Syndr. 2013;62(4):405–13. pmid:23254153; PubMed Central PMCID: PMC3633634.
- 4. d'Arminio Monforte A, Sabin CA, Phillips A, Sterne J, May M, Justice A, et al. The changing incidence of AIDS events in patients receiving highly active antiretroviral therapy. Arch Intern Med. 2005;165(4):416–23. pmid:15738371.
- 5. Ives NJ, Gazzard BG, Easterbrook PJ. The changing pattern of AIDS-defining illnesses with the introduction of highly active antiretroviral therapy (HAART)in a London clinic. J Infect. 2001;42(2):134–9. pmid:11531320.
- 6. Kaufmann GR, Perrin L, Pantaleo G, Opravil M, Furrer H, Telenti A, et al. CD4 T-Lymphocyte Recovery in Individuals With Advanced HIV-1 Infection Receiving Potent Antiretroviral Therapy for 4 Years: The Swiss HIV Cohort Study. Arch Intern Med. 2003;163(18):2187–95. PubMed 10.1001/archinte.163.18.2187 pmid:14557216.
- 7. Le T, Wright EJ, Smith DM, He W, Catano G, Okulicz JF, et al. Enhanced CD4+ T-cell recovery with earlier HIV-1 antiretroviral therapy. The New England journal of medicine. 2013;368(3):218–30. pmid:23323898; PubMed Central PMCID: PMC3657555.
- 8. Gazzola L, Tincati C, Bellistri GM, Monforte A, Marchetti G. The absence of CD4+ T cell count recovery despite receipt of virologically suppressive highly active antiretroviral therapy: clinical risk, immunological gaps, and therapeutic options. Clin Infect Dis. 2009;48(3):328–37. pmid:19123868.
- 9. Gutierrez F, Padilla S, Masia M, Iribarren JA, Moreno S, Viciana P, et al. Patients' characteristics and clinical implications of suboptimal CD4 T-cell gains after 1 year of successful antiretroviral therapy. Current HIV research. 2008;6(2):100–7. pmid:18336257.
- 10. van Lelyveld SF, Gras L, Kesselring A, Zhang S, De Wolf F, Wensing AM, et al. Long-term complications in patients with poor immunological recovery despite virological successful HAART in Dutch ATHENA cohort. AIDS. 2012;26(4):465–74. pmid:22112603.
- 11. Okulicz JF, Le TD, Agan BK, Camargo JF, Landrum ML, Wright E, et al. Influence of the timing of antiretroviral therapy on the potential for normalization of immune status in human immunodeficiency virus 1-infected individuals. JAMA internal medicine. 2015;175(1):88–99. pmid:25419650; PubMed Central PMCID: PMC4286496.
- 12. Hunt PW, Deeks SG, Rodriguez B, Valdez H, Shade SB, Abrams DI, et al. Continued CD4 cell count increases in HIV-infected adults experiencing 4 years of viral suppression on antiretroviral therapy. AIDS. 2003;17(13):1907–15. pmid:12960823.
- 13.
Adolescents PoAGfAa. Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. In: Services DoHaH, editor. April 2015 ed: US. DHHS; 2015.
- 14. Kaufmann G, Bloch M, Finlayson R, Zaunders J, Smith D, Cooper D. The extent of HIV-1-related immunodeficiency and age predict the long-term CD4 T lymphocyte response to potent antiretroviral therapy. AIDS. 2002;16:359–67. pmid:11834947
- 15. Kalayjian RC, Spritzler J, Matining RM, Fiscus SA, Gross BH, Francis IR, et al. Older HIV-infected patients on antiretroviral therapy have B-cell expansion and attenuated CD4 cell increases with immune activation reduction. AIDS. 2013;27(10):1563–71. pmid:24047762; PubMed Central PMCID: PMC3805132.
- 16. Vinikoor MJ, Joseph J, Mwale J, Marx MA, Goma FM, Mulenga LB, et al. Age at antiretroviral therapy initiation predicts immune recovery, death, and loss to follow-up among HIV-infected adults in urban Zambia. AIDS Res Hum Retroviruses. 2014;30(10):949–55. pmid:24998881; PubMed Central PMCID: PMC4179921.
- 17. Blanco JR, Jarrin I, Martinez A, Siles E, Larrayoz IM, Canuelo A, et al. Shorter telomere length predicts poorer immunological recovery in virologically suppressed HIV-1-infected patients treated with combined antiretroviral therapy. J Acquir Immune Defic Syndr. 2015;68(1):21–9. pmid:25321176.
- 18. Marcus JL, Leyden WA, Chao CR, Xu L, Quesenberry CP Jr., Tien PC, et al. Differences in Response to Antiretroviral Therapy by Sex and Hepatitis C Infection Status. AIDS Patient Care STDS. 2015;29(7):370–8. pmid:26061798.
- 19. Maman D, Pujades-Rodriguez M, Subtil F, Pinoges L, McGuire M, Ecochard R, et al. Gender differences in immune reconstitution: a multicentric cohort analysis in sub-Saharan Africa. PLoS One. 2012;7(2):e31078. pmid:22363550; PubMed Central PMCID: PMC3281917.
- 20. Finkel DG, John G, Holland B, Slim J, Smith SM. Women have a greater immunological response to effective virological HIV-1 therapy. AIDS. 2003;17(13):2009–11. pmid:12960846.
- 21. Haas DW, Geraghty DE, Andersen J, Mar J, Motsinger AA, D'Aquila RT, et al. Immunogenetics of CD4 lymphocyte count recovery during antiretroviral therapy: An AIDS Clinical Trials Group study. J Infect Dis. 2006;194(8):1098–107. pmid:16991084.
- 22. Hartling HJ, Thorner LW, Erikstrup C, Harritshoj LH, Kronborg G, Pedersen C, et al. Polymorphism in interleukin-7 receptor alpha gene is associated with faster CD4(+) T-cell recovery after initiation of combination antiretroviral therapy. AIDS. 2014;28(12):1739–48. pmid:24911352.
- 23. Ahuja SK, Kulkarni H, Catano G, Agan BK, Camargo JF, He W, et al. CCL3L1-CCR5 genotype influences durability of immune recovery during antiretroviral therapy of HIV-1-infected individuals. Nat Med. 2008;14(4):413–20. pmid:18376407; PubMed Central PMCID: PMC2630879.
- 24. Rauch A, Nolan D, Furrer H, McKinnon E, John M, Mallal S, et al. HLA-Bw4 homozygosity is associated with an impaired CD4 T cell recovery after initiation of antiretroviral therapy. Clin Infect Dis. 2008;46(12):1921–5. pmid:18466093.
- 25. Rajasuriar R, Booth DR, Gouillou M, Spelman T, James I, Solomon A, et al. The role of SNPs in the alpha-chain of the IL-7R gene in CD4+ T-cell recovery in HIV-infected African patients receiving suppressive cART. Genes Immun. 2012;13(1):83–93. pmid:21938017.
- 26. Guzman-Fulgencio M, Berenguer J, Micheloud D, Fernandez-Rodriguez A, Garcia-Alvarez M, Jimenez-Sousa MA, et al. European mitochondrial haplogroups are associated with CD4+ T cell recovery in HIV-infected patients on combination antiretroviral therapy. J Antimicrob Chemother. 2013;68(10):2349–57. pmid:23749950.
- 27. Grady BJ, Samuels DC, Robbins GK, Selph D, Canter JA, Pollard RB, et al. Mitochondrial genomics and CD4 T-cell count recovery after antiretroviral therapy initiation in AIDS clinical trials group study 384. J Acquir Immune Defic Syndr. 2011;58(4):363–70. pmid:21792066; PubMed Central PMCID: PMC3204178.
- 28. Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P, et al. The Women's Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology (Cambridge, Mass. 1998;9(2):117–25. pmid:9504278.
- 29. Bacon MC, von Wyl V, Alden C, Sharp G, Robison E, Hessol N, et al. The Women's Interagency HIV Study: an observational cohort brings clinical sciences to the bench. Clin Diagn Lab Immunol. 2005;12(9):1013–9. Epub 2005/09/09. pmid:16148165; PubMed Central PMCID: PMC1235804.
- 30. Scherzer R, Greenblatt RM, Merhi ZO, Kassaye S, Lambert-Messerlian G, Maki PM, et al. Use of antimullerian hormone to predict the menopausal transition in HIV-infected women. Am J Obstet Gynecol. 2016. pmid:27473002.
- 31. Van der Auwera GA, Carneiro MO, Hartl C, Poplin R, Del Angel G, Levy-Moonshine A, et al. From FastQ data to high confidence variant calls: the Genome Analysis Toolkit best practices pipeline. Curr Protoc Bioinformatics. 2013;43:11 0 1–33. pmid:25431634; PubMed Central PMCID: PMC4243306.
- 32. DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43(5):491–8. pmid:21478889; PubMed Central PMCID: PMC3083463.
- 33. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60. pmid:19451168; PubMed Central PMCID: PMC2705234.
- 34. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25(16):2078–9. pmid:19505943; PubMed Central PMCID: PMC2723002.
- 35. Carson AR, Smith EN, Matsui H, Braekkan SK, Jepsen K, Hansen JB, et al. Effective filtering strategies to improve data quality from population-based whole exome sequencing studies. BMC Bioinformatics. 2014;15:125. pmid:24884706; PubMed Central PMCID: PMC4098776.
- 36. Dering C, Schillert A, Konig IR, Ziegler A. A comparison of two collapsing methods in different approaches. BMC Proc. 2014;8(Suppl 1 Genetic Analysis Workshop 18Vanessa Olmo):S8. pmid:25519408; PubMed Central PMCID: PMC4143760.
- 37. Liu DJ, Leal SM. A novel adaptive method for the analysis of next-generation sequencing data to detect complex trait associations with rare variants due to gene main effects and interactions. PLoS Genet. 2010;6(10):e1001156. pmid:20976247; PubMed Central PMCID: PMC2954824.
- 38. Neale BM, Sham PC. The future of association studies: gene-based analysis and replication. Am J Hum Genet. 2004;75(3):353–62. Epub 2004/07/22. pmid:15272419; PubMed Central PMCID: PMC1182015.
- 39. Marenholz I, Bauerfeind A, Esparza-Gordillo J, Kerscher T, Granell R, Nickel R, et al. The eczema risk variant on chromosome 11q13 (rs7927894) in the population-based ALSPAC cohort: a novel susceptibility factor for asthma and hay fever. Hum Mol Genet. 2011;20(12):2443–9. pmid:21429916.
- 40. Hosono K, Noda S, Shimizu A, Nakanishi N, Ohtsubo M, Shimizu N, et al. YPEL5 protein of the YPEL gene family is involved in the cell cycle progression by interacting with two distinct proteins RanBPM and RanBP10. Genomics. 2010;96(2):102–11. pmid:20580816.
- 41. Ding J, Zhao J, Sun L, Mi Z, Cen S. Citron kinase enhances ubiquitination of HIV-1 Gag protein and intracellular HIV-1 budding. Arch Virol. 2016. pmid:27339686.
- 42. Wang CJ, Zhang ZZ, Xu J, Wang M, Zhao WY, Tu L, et al. SLITRK3 expression correlation to gastrointestinal stromal tumor risk rating and prognosis. World J Gastroenterol. 2015;21(27):8398–407. pmid:26217092; PubMed Central PMCID: PMC4507110.
- 43. Mattheisen M, Samuels JF, Wang Y, Greenberg BD, Fyer AJ, McCracken JT, et al. Genome-wide association study in obsessive-compulsive disorder: results from the OCGAS. Mol Psychiatry. 2015;20(3):337–44. pmid:24821223; PubMed Central PMCID: PMC4231023.
- 44. Gelman BB, Soukup VM, Schuenke KW, Keherly MJ, Holzer C 3rd, Richey FJ, et al. Acquired neuronal channelopathies in HIV-associated dementia. J Neuroimmunol. 2004;157(1–2):111–9. pmid:15579287.
- 45. Kochl R, Thelen F, Vanes L, Brazao TF, Fountain K, Xie J, et al. WNK1 kinase balances T cell adhesion versus migration in vivo. Nat Immunol. 2016. pmid:27400149.
- 46. Callicott JH, Feighery EL, Mattay VS, White MG, Chen Q, Baranger DA, et al. DISC1 and SLC12A2 interaction affects human hippocampal function and connectivity. J Clin Invest. 2013;123(7):2961–4. pmid:23921125; PubMed Central PMCID: PMC3999052.
- 47. Panichareon B, Nakayama K, Iwamoto S, Thurakitwannakarn W, Sukhumsirichart W. Association of CTXN3-SLC12A2 polymorphisms and schizophrenia in a Thai population. Behav Brain Funct. 2012;8:27. pmid:22643131; PubMed Central PMCID: PMC3433332.
- 48. Dahlin A, Wittwer M, de la Cruz M, Woo JM, Bam R, Scharen-Guivel V, et al. A pharmacogenetic candidate gene study of tenofovir-associated Fanconi syndrome. Pharmacogenet Genomics. 2015;25(2):82–92. pmid:25485598; PubMed Central PMCID: PMC4331349.
- 49. Mouaffak F, Kebir O, Bellon A, Gourevitch R, Tordjman S, Viala A, et al. Association of an UCP4 (SLC25A27) haplotype with ultra-resistant schizophrenia. Pharmacogenomics. 2011;12(2):185–93. pmid:21332312.
- 50. Chu TT, Liu Y. An integrated genomic analysis of gene-function correlation on schizophrenia susceptibility genes. J Hum Genet. 2010;55(5):285–92. pmid:20339380.
- 51. Gao X, Yang J, Wang M, Zhang J. TCF21 genetic polymorphisms and breast cancer risk in Chinese women. Oncotarget. 2016. pmid:27270650.
- 52. Gooskens SL, Gadd S, Guidry Auvil JM, Gerhard DS, Khan J, Patidar R, et al. TCF21 hypermethylation in genetically quiescent clear cell sarcoma of the kidney. Oncotarget. 2015;6(18):15828–41. pmid:26158413; PubMed Central PMCID: PMC4599240.
- 53. Fujimaki T, Oguri M, Horibe H, Kato K, Matsuoka R, Abe S, et al. Association of a transcription factor 21 gene polymorphism with hypertension. Biomed Rep. 2015;3(1):118–22. pmid:25469260; PubMed Central PMCID: PMC4251179.
- 54. Miller CL, Haas U, Diaz R, Leeper NJ, Kundu RK, Patlolla B, et al. Coronary heart disease-associated variation in TCF21 disrupts a miR-224 binding site and miRNA-mediated regulation. PLoS Genet. 2014;10(3):e1004263. pmid:24676100; PubMed Central PMCID: PMC3967965.
- 55. Egana-Gorrono L, Guardo AC, Bargallo ME, Planet E, Vilaplana E, Escriba T, et al. MicroRNA Profile in CD8+ T-Lymphocytes from HIV-Infected Individuals: Relationship with Antiviral Immune Response and Disease Progression. PLoS One. 2016;11(5):e0155245. pmid:27171002; PubMed Central PMCID: PMC4865051.
- 56. Yahyaei S, Biasin M, Saulle I, Gnudi F, De Luca M, Tasca KI, et al. Identification of a Specific miRNA Profile in HIV Exposed Seronegative Individuals. J Acquir Immune Defic Syndr. 2016. pmid:27171739.
- 57. Pincetic A, Kuang Z, Seo EJ, Leis J. The interferon-induced gene ISG15 blocks retrovirus release from cells late in the budding process. J Virol. 2010;84(9):4725–36. pmid:20164219; PubMed Central PMCID: PMC2863725.
- 58. Yang D, Hurley JH. Structural role of the Vps4-Vta1 interface in ESCRT-III recycling. Structure. 2010;18(8):976–84. pmid:20696398; PubMed Central PMCID: PMC3124813.
- 59. Ikram MK, Sim X, Jensen RA, Cotch MF, Hewitt AW, Ikram MA, et al. Four novel Loci (19q13, 6q24, 12q24, and 5q14) influence the microcirculation in vivo. PLoS Genet. 2010;6(10):e1001184. pmid:21060863; PubMed Central PMCID: PMC2965750.
- 60. Zhang X, Liu Y, Zhang L, Yang Z, Yang L, Wang X, et al. Associations of genetic variations in EYA4, GRHL2 and DFNA5 with noise-induced hearing loss in Chinese population: a case- control study. Environ Health. 2015;14:77. pmid:26400775; PubMed Central PMCID: PMC4581404.
- 61. Spadoni JL, Rucart P, Le Clerc S, van Manen D, Coulonges C, Ulveling D, et al. Identification of Genes Whose Expression Profile Is Associated with Non-Progression towards AIDS Using eQTLs. PLoS One. 2015;10(9):e0136989. pmid:26367535; PubMed Central PMCID: PMC4569262.
- 62. Imamura K, Maeda S, Kawamura I, Matsuyama K, Shinohara N, Yahiro Y, et al. Human immunodeficiency virus type 1 enhancer-binding protein 3 is essential for the expression of asparagine-linked glycosylation 2 in the regulation of osteoblast and chondrocyte differentiation. J Biol Chem. 2014;289(14):9865–79. pmid:24563464; PubMed Central PMCID: PMC3975031.
- 63. Arya R, Del Rincon I, Farook VS, Restrepo JF, Winnier DA, Fourcaudot MJ, et al. Genetic Variants Influencing Joint Damage in Mexican Americans and European Americans With Rheumatoid Arthritis. Genet Epidemiol. 2015;39(8):678–88. pmid:26498133.
- 64. Beaudoin M, Goyette P, Boucher G, Lo KS, Rivas MA, Stevens C, et al. Deep resequencing of GWAS loci identifies rare variants in CARD9, IL23R and RNF186 that are associated with ulcerative colitis. PLoS Genet. 2013;9(9):e1003723. pmid:24068945; PubMed Central PMCID: PMC3772057.
- 65. Jahan AS, Lestra M, Swee LK, Fan Y, Lamers MM, Tafesse FG, et al. Usp12 stabilizes the T-cell receptor complex at the cell surface during signaling. Proc Natl Acad Sci U S A. 2016;113(6):E705–14. pmid:26811477; PubMed Central PMCID: PMC4760780.
- 66. Ohashi M, Holthaus AM, Calderwood MA, Lai CY, Krastins B, Sarracino D, et al. The EBNA3 family of Epstein-Barr virus nuclear proteins associates with the USP46/USP12 deubiquitination complexes to regulate lymphoblastoid cell line growth. PLoS Pathog. 2015;11(4):e1004822. pmid:25855980; PubMed Central PMCID: PMC4391933.
- 67. Ivanova E, Berger A, Scherrer A, Alkalaeva E, Strub K. Alu RNA regulates the cellular pool of active ribosomes by targeted delivery of SRP9/14 to 40S subunits. Nucleic Acids Res. 2015;43(5):2874–87. pmid:25697503; PubMed Central PMCID: PMC4357698.
- 68. Mary C, Scherrer A, Huck L, Lakkaraju AK, Thomas Y, Johnson AE, et al. Residues in SRP9/14 essential for elongation arrest activity of the signal recognition particle define a positively charged functional domain on one side of the protein. RNA. 2010;16(5):969–79. pmid:20348448; PubMed Central PMCID: PMC2856890.
- 69. Keene SE, Telesnitsky A. cis-Acting determinants of 7SL RNA packaging by HIV-1. J Virol. 2012;86(15):7934–42. pmid:22593161; PubMed Central PMCID: PMC3421668.
- 70. Takeda E, Murakami T, Matsuda G, Murakami H, Zako T, Maeda M, et al. Nuclear exportin receptor CAS regulates the NPI-1-mediated nuclear import of HIV-1 Vpr. PLoS One. 2011;6(11):e27815. pmid:22110766; PubMed Central PMCID: PMC3218035.
- 71. Jia D, Huang L, Bischoff J, Moses MA. The endogenous zinc finger transcription factor, ZNF24, modulates the angiogenic potential of human microvascular endothelial cells. FASEB J. 2015;29(4):1371–82. pmid:25550468; PubMed Central PMCID: PMC4734806.
- 72. Li JZ, Chen X, Gong XL, Hu HY, Shi D, Lu YM, et al. Identification of a functional nuclear localization signal mediating nuclear import of the zinc finger transcription factor ZNF24. PLoS One. 2013;8(11):e79910. pmid:24224020; PubMed Central PMCID: PMC3815127.
- 73. Gauvin J, Chagnon-Choquet J, Poudrier J, Roger M, Montreal Primary HIVI, Slow Progressor C. Fluctuations in Blood Marginal Zone B-Cell Frequencies May Reflect Migratory Patterns Associated with HIV-1 Disease Progression Status. PLoS One. 2016;11(5):e0155868. pmid:27203285; PubMed Central PMCID: PMC4874627.
- 74. Mavigner M, Cazabat M, Dubois M, L'Faqihi FE, Requena M, Pasquier C, et al. Altered CD4+ T cell homing to the gut impairs mucosal immune reconstitution in treated HIV-infected individuals. J Clin Invest. 2012;122(1):62–9. pmid:22156200; PubMed Central PMCID: PMC3248296.
- 75. Trivedi PJ, Bruns T, Ward S, Mai M, Schmidt C, Hirschfield GM, et al. Intestinal CCL25 expression is increased in colitis and correlates with inflammatory activity. J Autoimmun. 2016;68:98–104. pmid:26873648; PubMed Central PMCID: PMC4803021.
- 76. Qin S, Sui Y, Murphey-Corb MA, Reinhart TA. Association between decreased CXCL12 and CCL25 expression and increased apoptosis in lymphoid tissues of cynomolgus macaques during SIV infection. J Med Primatol. 2008;37 Suppl 2:46–54. pmid:19187430; PubMed Central PMCID: PMC3581082.
- 77. Deng YJ, Huang ZX, Zhou CJ, Wang JW, You Y, Song ZQ, et al. Gene profiling involved in immature CD4+ T lymphocyte responsible for systemic lupus erythematosus. Mol Immunol. 2006;43(9):1497–507. pmid:16143398.
- 78. Tadokoro K, Hashimoto R, Tatsumi M, Kosuga A, Kamijima K, Kunugi H. The Gem interacting protein (GMIP) gene is associated with major depressive disorder. Neurogenetics. 2005;6(3):127–33. pmid:16086184.
- 79. Ng SY, Yoshida T, Zhang J, Georgopoulos K. Genome-wide lineage-specific transcriptional networks underscore Ikaros-dependent lymphoid priming in hematopoietic stem cells. Immunity. 2009;30(4):493–507. pmid:19345118; PubMed Central PMCID: PMC3012962.
- 80. Hendrickson SL, Lautenberger JA, Chinn LW, Malasky M, Sezgin E, Kingsley LA, et al. Genetic variants in nuclear-encoded mitochondrial genes influence AIDS progression. PLoS One. 2010;5(9):e12862. pmid:20877624; PubMed Central PMCID: PMC2943476.
- 81. Min-Oo G, Bezman NA, Madera S, Sun JC, Lanier LL. Proapoptotic Bim regulates antigen-specific NK cell contraction and the generation of the memory NK cell pool after cytomegalovirus infection. J Exp Med. 2014;211(7):1289–96. pmid:24958849; PubMed Central PMCID: PMC4076589.
- 82. Nieters A, Conde L, Slager SL, Brooks-Wilson A, Morton L, Skibola DR, et al. PRRC2A and BCL2L11 gene variants influence risk of non-Hodgkin lymphoma: results from the InterLymph consortium. Blood. 2012;120(23):4645–8. pmid:23047821; PubMed Central PMCID: PMC3512239.
- 83. Morton LM, Purdue MP, Zheng T, Wang SS, Armstrong B, Zhang Y, et al. Risk of non-Hodgkin lymphoma associated with germline variation in genes that regulate the cell cycle, apoptosis, and lymphocyte development. Cancer Epidemiol Biomarkers Prev. 2009;18(4):1259–70. pmid:19336552; PubMed Central PMCID: PMC2716008.
- 84. Tanaka-Nakanishi A, Yasunaga J, Takai K, Matsuoka M. HTLV-1 bZIP factor suppresses apoptosis by attenuating the function of FoxO3a and altering its localization. Cancer Res. 2014;74(1):188–200. pmid:24177179.
- 85. Limou S, Melica G, Coulonges C, Lelievre JD, Do H, McGinn S, et al. Identification of IL7RA risk alleles for rapid progression during HIV-1 infection: a comprehensive study in the GRIV cohort. Current HIV research. 2012;10(2):143–50. pmid:22329520.
- 86. Wood E, Hogg RS, Yip B, Harrigan PR, O'Shaughnessy MV, Montaner JS. The impact of adherence on CD4 cell count responses among HIV-infected patients. J Acquir Immune Defic Syndr. 2004;35(3):261–8. pmid:15076240.
- 87. Tanaskovic S, Fernandez S, Price P, French MA. Interleukin-7 signalling defects in naive CD4+ T cells of HIV patients with CD4+ T-cell deficiency on antiretroviral therapy are associated with T-cell activation and senescence. AIDS. 2014;28(6):821–30. pmid:24499954.
- 88. Al-Mawali A, Pinto AD, Al Busaidi R, Al-Zakwani I. Lymphocyte subsets: reference ranges in an age- and gender-balanced population of Omani healthy adults. Cytometry A. 2013;83(8):739–44. pmid:23839863.
- 89. Jentsch-Ullrich K, Koenigsmann M, Mohren M, Franke A. Lymphocyte subsets' reference ranges in an age- and gender-balanced population of 100 healthy adults—a monocentric German study. Clin Immunol. 2005;116(2):192–7. pmid:15993366.
- 90. Valiathan R, Deeb K, Diamante M, Ashman M, Sachdeva N, Asthana D. Reference ranges of lymphocyte subsets in healthy adults and adolescents with special mention of T cell maturation subsets in adults of South Florida. Immunobiology. 2014;219(7):487–96. pmid:24661720.
- 91. Devadas K, Biswas S, Haleyurgirisetty M, Wood O, Ragupathy V, Lee S, et al. Analysis of Host Gene Expression Profile in HIV-1 and HIV-2 Infected T-Cells. PLoS One. 2016;11(1):e0147421. pmid:26821323; PubMed Central PMCID: PMC4731573.
- 92. Raynor J, Sholl A, Plas DR, Bouillet P, Chougnet CA, Hildeman DA. IL-15 Fosters Age-Driven Regulatory T Cell Accrual in the Face of Declining IL-2 Levels. Front Immunol. 2013;4:161. pmid:23805138; PubMed Central PMCID: PMC3690359.