ABSTRACT
Natural killer (NK) cells during chronic viral infection have been well studied in the past. We performed an unbiased next-generation RNA-sequencing approach to identify commonalities or differences of the effect of HIV, HCV, and HBV viremia on NK cell transcriptomes. Using cell sorting, we obtained CD3− CD56+ NK cells from blood of 6 HIV-, 8 HCV-, and 32 HBV-infected patients without treatment. After library preparation and sequencing, we used an in-house analytic pipeline to compare expression levels with matched healthy controls. In NK cells from HIV-, HCV-, and HBV-infected patients, transcriptome analysis identified 272, 53, and 56 differentially expressed genes, respectively (fold change, >1.5; q-value, 0.2). Interferon-stimulated genes were induced in NK cells from HIV/HCV patients, but not during HBV infection. HIV viremia downregulated ribosome assembly genes in NK cells. In HBV-infected patients, viral load and alanine aminotransferase (ALT) variation had little effect on genes related to NK effector function. In conclusion, we compare, for the first time, NK cell transcripts of viremic HIV, HCV, and HBV patients. We clearly demonstrate distinctive NK cell gene signatures in three different populations, suggestive for a different degree of functional alterations of the NK cell compartment compared to healthy individuals.
IMPORTANCE Three viruses exist that can result in persistently high viral loads in immunocompetent humans: human immunodeficiency virus (HIV), hepatitis C virus, and hepatitis B virus. In the last decades, by using flow cytometry and in vitro assays on NK cells from patients with these types of infections, several impairments have been established, particularly during and possibly contributing to HIV viremia. However, the background of NK cell impairments in viremic patients is not well understood. In this study, we describe the NK cell transcriptomes of patients with high viral loads of different etiologies. We clearly demonstrate distinctive NK cell gene signatures with regard to interferon-stimulated gene induction and the expression of genes coding for activation markers or proteins involved in cytotoxic action, as well immunological genes. This study provides important details necessary to uncover the origin of functional and phenotypical differences between viremic patients and healthy subjects and provides many leads that can be confirmed using future in vitro manipulation experiments.
INTRODUCTION
Acute viral infection is usually accompanied by early production of infectious virions and rapid eradication of the virus by the host immune system. However, ineffective clearance of the virus can result in latent and even persistent infection within the host. In contrast to viruses such as Epstein-Barr virus, cytomegalovirus (CMV), or herpesviruses that may reinitiate replication after long periods of latency (1), three viruses—human immunodeficiency virus (HIV), hepatitis C virus (HCV), and hepatitis B virus (HBV)—can persistently replicate, resulting in considerable viral loads in the blood of immunocompetent individuals. Untreated patients infected with these viruses may harbor high viral loads for decades, which can be associated with variable degrees of immune activation or inflammation, but their antiviral host immune responses are insufficient to eradicate the virus (2).
The natural killer (NK) cell is an important antiviral effector cell (3). NK cells are capable of lysing virus-infected cells following major histocompatibility complex class I downregulation, not necessitating prior licensing. In addition, NK cells can produce inflammatory cytokines, such as gamma interferon (IFN-γ) and tumor necrosis factor (TNF), as well as support and regulate virus-specific immunity (4). Despite the fact that HIV, HCV, and HBV do not directly infect NK cells (HIV mainly infects CD4+ T cells, and hepatitis viruses infect hepatocytes), chronic exposure to replicating virus may affect NK cell phenotype and function indirectly. These changes may be the consequence of either alterations in the cytokine milieu, exposure to circulating viral components, or abnormal interaction with other leukocytes. HIV patient-derived NK cells have been shown to produce less IFN-γ following stimulation with cytokines like interleukin-12 (IL-12) and IFN-α (5, 6). Importantly, the capacity of NK cells to lyse target cells in vitro, such as K562 cells, is lower in HIV patients (7–9). This is in contrast to NK cells from chronic HCV- and HBV-infected patients that display an unaltered or mildly augmented cytotoxic potential compared to healthy individuals. Our group has previously shown that NK cell-mediated cytokine production in viral hepatitis patients is not or only slightly impaired, but results on this topic are conflicting since other researchers previously measured lower IFN-γ production by NK cells from hepatitis patients (4, 10–12). NK cells from HBV and HCV patients are described to have an activated phenotype (a high percentage of NK cells positive for NKp30, NKp46, and NKG2C) compared to healthy controls (HCs) (4). It should be noted here that study results often vary in this field (13) and that all studies performed to date on this topic examine a limited number of functions and phenotypical markers of NK cells, mainly by flow cytometry (3, 4, 14, 15). More unbiased approaches examining gene expression profiles of NK cells from patients with chronic HIV, HCV, and HBV have not been performed. These approaches may identify specific genes and signaling pathways affected in NK cells from patients with ongoing viral replication in vivo that were not recognized using conventional flow cytometry.
Therefore, in the present study, we performed an unbiased RNA sequencing approach on purified NK cells from viremic patients. We show for the first time that HIV and HCV viremia induced numerous interferon-stimulated genes (ISG) in NK cells, but this was not observed in NK cells from chronic HBV patients. In addition, HIV viremia affected more genes and signaling pathways in NK cells than viral hepatitis, including immune-related genes and genes involved in cytotoxic action. This study provides important details necessary to uncover the origin of the functional and phenotypical differences between viremic patients and healthy subjects and clearly demonstrates intrinsic differences in NK cells obtained from patients with three distinct chronic infections.
RESULTS
Characteristics of three untreated patient cohorts chronically infected with HIV, HCV and HBV.To identify the effects of chronic viremia on NK cell gene transcripts, we isolated NK cells from cohorts representing the three causes of chronic viremia in humans: HIV-, HCV-, and HBV-infected individuals (Table 1). None of the patients were on anti(retro)viral therapy, and all had relatively high serum viral loads. The average viral load of the HIV-infected cohort was 1.1 × 105 copies/ml and the average CD4 count was 374. The average viral loads of the HCV and HBV cohorts were 3.9 × 106 and 2.5 × 108 IU/ml, respectively. We matched the HBV-infected cohort—based on age, gender, and ethnicity—to an Asian healthy control group (Table 1, healthy control 1) and matched the HIV- and HCV-infected cohorts to a second, predominantly Caucasian control group (Table 1, healthy control 2).
Patient characteristics of HBV, HCV, and HIV patients and matched healthy control groupsa
NK cell ISG are not induced by HBV, but both HIV and HCV infection have parallel patterns of upregulated ISG.To determine whether the NK cell transcriptional signature in blood was modulated as a consequence of chronic infection, gene expression profiles of sorted NK cells from patients were compared to their matched controls. In NK cells from HIV patients, 272 differentially expressed genes (DEGs) with ≥1.5-fold-higher expression compared to healthy controls were identified (q-value, 0.2). A total of 241 of the 272 DEGs were downregulated, while 31 were upregulated in HIV NK cells compared to their controls (see Table S1 in the supplemental material). The numbers of DEGs were lower in NK cells from HCV and HBV patients (53 and 56, respectively), and the downregulated versus upregulated genes were more balanced compared to their respective controls (30 versus 23 in HCV and 27 versus 29 in HBV, respectively). In NK cells from HIV and HCV patients, 22 genes were differentially expressed in both groups, but there was no overlap with the HBV-infected cohort. In HCV patients, 2 ISGs were among the top 10 upregulated genes compared to 5 during HIV: ISG15, MX1, IRF7, STAT1, and OAS1 (Fig. 1D). Despite the observation that, during HCV and HIV infection, these individual genes are among the genes with the highest fold change (Fig. 2A) compared to healthy controls, we did not detect large clusters common to HIV- or HCV-derived NK cells (Fig. 2B). Apart from the therapeutic application of IFN-α (21–23), it has been shown that chronic exposure to type I IFN during HIV infection leads to desensitization to its signaling, as well as ongoing immune activation (24, 25). To investigate the specific ISGs that are induced during viremia in humans, we evaluated the expression levels of 139 ISGs in NK cells (Table S2 in the supplemental material; Fig. 2). In NK cells from the HIV-, HCV-, and HBV-infected cohorts, differential expression of 9, 14, and 1 ISG, respectively, was observed. In sharp contrast, NK cells from HBV-infected patients did not differentially express any ISG, except for IRF4 (log2-fold change of 1.75), encoding a protein that may compete with IRF-5 for binding to MyD88, thereby inhibiting downstream TLR-mediated induction of proinflammatory cytokines, including IFN-γ (26).
NK cells of HIV-infected patients differentially express numerous genes encoding NK activating surface markers and cytotoxicity-related genes and display a downregulation of ribosome assembly genes. (A) Representative flow cytometry plots showing the gating strategy for peripheral CD56+ CD3− NK cells used for all included subjects. (B) Summary of the post-sort purity percentages of CD56+ CD3− cells in virally infected and healthy subjects. (C) Representative flow cytometry plots showing comparable pre- and post-sort percentages in healthy individuals in comparison with HIV-, HCV-, and HBV-infected patients. (D) Relative expression of the 10 most upregulated genes in HIV-, HCV-, and HBV-infected individuals compared to matched healthy controls (fold change, ≥1.5; q-value, 0.2), including several common ISGs in HIV/HCV infection. NK cells derived from HIV-infected individuals strikingly downregulated several genes related to ribosome biogenesis (RPL27, RPS7, RPL24, RPS13, and RPL10L), and during both HIV and HCV infection, ISGs (IFI6 and IFI27L2) were among the top 10 most upregulated genes.
NK cell ISG are not induced by HBV, but both HIV and HCV infection have parallel patterns of upregulated ISGs. (A) Expression relative to matched healthy subjects of several ISGs induced during chronic HCV and HIV infections that is indicative of parallel ISG induction in NK cells. Chronic infection with HCV and HIV induces the expression of several ISGs, but in HBV patients only a single NK cell ISG was upregulated compared to healthy controls. (B) Relative NK cell ISG expression of NK cells from HIV, HBV, and HCV patients and healthy subjects. Each row represents a single ISG. No large clusters of up- or downregulation are common to the viremic subjects, confirming that the expression patterns of individual subjects vary considerably. Patients are presented in the following order: healthy control and HBV, HCV, and HIV infections. The colors represent a range of fold differences from −1.5 (blue) to 1.5 (red).
NK cells of HIV-infected patients differentially express genes coding for NK activating surface markers and cytotoxicity-related genes and display downregulation of ribosome assembly genes.In NK cells derived from HIV-infected individuals, genes coding for surface markers impacting NK cell activation status and cytotoxicity were differentially expressed (see Table S1 in the supplemental material). Genes encoding NK cell-activating receptors, such as CD160 and TNFSFR1 (log2-fold changes of −1.8 and −0.9, respectively) were downregulated. Conversely, six genes coding for proteins whose ligation results in inhibitory signaling were also downregulated in HIV-derived NK cells only: KLRG1, KLRF1, CD244, PILRA, LAIR1, and CD300A/C (log2-fold changes of −0.6, −4.0, −1.5, −1.3, −1.1, and −1.3/−2.7, respectively). We did not observe this degree of modulation of NK cell receptor transcription in HCV or HBV. Importantly, two genes directly involved in the mechanism of cytotoxic action, CTSC (27) (coding cathepsin C, log2-fold change of −1.1) and HCST (28) (coding part of the activating complex KLRK1-HCST, log2-fold change of −3.8) were downregulated in NK cells from HIV-infected individuals only. Interestingly, NK cells derived from HIV-infected individuals also downregulated several genes related to ribosome formation (Table S1, Fig. 1D), including RPL27, RPS7, RPL24, RPS13, and RPL10L.
Fluctuations in serum HBV DNA and alanine aminotransferase (ALT) levels impact the expression by NK cells of immune genes but only minimally affect cytotoxicity or IFN signaling-related genes.As described above, when analyzing the complete HBV-infected cohort, to our surprise only a single ISG was detected to be differentially expressed (IRF4, Fig. 2A). However, since chronic HBV infection is a heterogeneous disease characterized by fluctuating ALT levels and viral loads, we investigated the consequence of these variable virological parameters on ISG expression in purified NK cells by stratifying the HBV cohort into four separate groups (as depicted in Fig. 3A). Compared to healthy controls, the IT (immune tolerant or HBeAg+ infection) phase, the IA (immune active or HBeAg+ hepatitis) phase, the IC (inactive carrier or HBeAg− infection) phase, and the ENEG (HBeAg− hepatitis) cohorts differentially expressed 37, 47, 64, and 45 genes, respectively (Fig. 3B; see also Table S3). Further examination of each subgroup revealed 5, 11, 8, and 8 immune-related genes differentially expressed in each phase (Table S3). Despite this limited variation of NK cell gene transcription, we found that among the DEGs identified, the majority of genes actively transcribed in the IA phase were immune related genes (Fig. 3C; see Table S3). IRF4, the single upregulated ISG in NK cells of the HBV cohort, was upregulated in all phases, except in the ENEG phase. Overall, the analysis of NK cell-derived transcripts of heterogeneous HBV patients show little correlation with fluctuations of clinical parameters (ALT and viral load), which is in concordance with the stable NK cell phenotype as measured by flow cytometry, previously described by our group (17).
Altered activity of specific pathways in NK cells from HCV and HIV patients, but not from HBV patients.In addition to analysis of the expression of individual genes, additional analysis for the activity of specific networks or pathways was conducted using Ingenuity Pathway Analysis (IPA). Using stringent criteria (P < 10−4; overlap, >5%), nine pathways were identified to be modified in NK cells from HIV patients, four pathways were identified to be modified in chronic HCV patients, and none were identified to be modified in chronic HBV patients compared to NK cells from healthy individuals. The pathway of IFN signaling in NK cells of both HIV and HCV patients was upregulated compared to their respective healthy controls (Table 2). Also, the NK cell signaling pathway was modified in NK cells from both chronic HIV and HCV patients, as well as pathways active during DNA repair and stress (nucleotide excision repair, DNA double-strand break repair by NHEJ, and GADD45 signaling). Likely related is the downregulation of the eukaryotic initiation factor 2 (eIF2) pathway in NK cells from the HIV-infected cohort, which includes various ribosomal assembly genes and other eukaryotic initiation factors (pathway P value, 7.58 × 10−8; z-score, −2.4). Upon induction by stress signals, eIF2 has a strong negative impact on cellular mRNA translation (29). Also, modulation of various pathways related to cell-cell contact or actin-mediated processes (such as remodeling of epithelial adherence junctions, regulation of actin-based motility by Rho, and FcγR-mediated phagocytosis in macrophages and monocytes) was found and was exclusively observed for the NK cells of chronic HIV patients. Several of the genes in these pathways encode proteins involved in the exocytosis of secretory lysozymes, and low abundance of their protein products might directly affect NK cell effector function (29). Finally, pathways involved in glycolysis and oxidative phosphorylation were altered in the NK cells of HIV patients only compared to NK cells from healthy controls. These altered metabolic pathways may affect NK cell functionality and may be linked to dysfunctional NK cell responses in HIV patients, as has been observed for CD8+ T cells (30). However, little is known regarding how metabolic processes impact NK cell activity during chronic viral infections (31).
HIV viremia results in NK cell downregulation of eIF2 signaling and upregulation of IFN signaling determined using pathway analysis of transcripts obtained from NK cells of HIV- and HCV-infected patientsa
DISCUSSION
Using a highly sensitive RNA-sequencing-based approach, we were able to compare, for the first time, the gene expression profiles of NK cells of chronic HIV-, HCV-, and HBV-infected individuals. We clearly demonstrate distinctive gene signatures in the NK cells of the three different patient populations, suggesting different degrees of functional alterations of the NK cell compartment compared to that of healthy individuals. The induction of ISG was pronounced in HIV and HCV patient-derived NK cells compared to their healthy counterparts but was almost completely absent in chronic HBV patients. In addition, the downregulation of ribosome assembly genes, as well as inhibitory receptor-encoding and cytotoxicity-elated genes, was observed in NK cells from HIV-infected patients, but not or to a lesser extent in NK cells from chronic HBV- and HCV-infected patients. This study provides important details necessary to uncover the origin of the functional and phenotypical differences between viremic patients and healthy subjects and clearly demonstrate intrinsic differences in NK cells obtained from patients with three distinct chronic infections.
NK cells from HBV-infected individuals did not differentially express any ISG compared to healthy individuals, except for IRF4 which is a nonclassical ISG (32), inducible in a type I IFN-independent manner. In 2004, the Chisari group showed that innate responses in the liver are not triggered or are only weakly triggered by HBV, resulting in limited systemic effects during the chronic phase of infection (33). Supporting this, serum levels of type I and type III IFNs are (undetectably) low during HBV infection (18, 34). However, in the blood of HIV and HCV patients, we and others have shown that systemic IFN and elevated IP-10 levels can be readily detected, which corresponds to the degree of ISG induction, as shown in Fig. 2A (35–39). In line with a stronger IFN response during HCV infection, it has been shown that STAT1 is upregulated (40). Here, we show by direct ex vivo analysis that STAT1 and STAT2 are indeed upregulated in NK cells during HCV and HIV infection. STAT4, however, which is downstream of the IL-12 receptor, is significantly downregulated in HIV infection, further pointing toward a dominance for an IFN-induced over an IL-12-induced response in NK cells from these chronic infections.
Despite the fact that HIV does not replicate in NK cells, HIV patient-derived NK cells displayed downregulation of genes involved in ribosome assembly. The cause of this downregulation of ribosomal protein transcription remains uncertain. However, similar findings have been reported for primary CD4+ T cells and Jurkat T cell lines in which pre-rRNA processing was perturbed in HIV infection (41, 42). Several of these ribosomal genes are in the pathway downstream of eIF2. For eIF2 specifically, the phosphorylation of alpha subunit of eIF2 blocks protein translation (43), allowing cells to conserve resources, and subsequently switch to a new gene expression program to prevent stress damage (44). Interestingly, due to additional interactions of HIV with cellular translational machinery, the synthesis of viral structural proteins is reported to be sustained (45). The downregulation of ribosome assembly in both CD4+ T cells and NK cells may reflect a common effect of systemic HIV proteins on both lymphocyte populations. Further studies will be necessary to dissect the link between ribosome biogenesis alteration and viral infection, possibly unraveling a novel intricate network of interactions between HIV and host cells.
The analysis of NK cell-derived transcripts of heterogeneous HBV patients showed little correlation with fluctuations of clinical parameters (serum ALT level and HBV DNA). Using FFPE liver biopsy specimens from HBV patients, we have previously shown that liver resident NK cells have an increased transcriptional activity in the phases with increased ALT levels (18), which was recently confirmed by others (46). In the peripheral blood compartment, fluctuating serum HBV DNA and ALT levels did not lead to dramatic changes in the expression of NK cell surface markers or production of IFN-γ evaluated by flow cytometry (17). Our present study, using a completely different, unbiased approach, confirms that only few immune genes are modulated when comparing NK cells from healthy individuals with HBV patients in each of the clinical phases. This indicates that HBV DNA or serum HBsAg does not trigger NK cell activation, which corresponds to previous reports (14, 47). It is important to decipher in coming studies whether the conflicting findings reported in literature have a technical or biological reason.
Our study has a few limitations. First, we restricted this study to transcriptional analysis and did not include analysis at the protein level. This is important, since the abundance of a transcript does not necessarily reflect protein expression. Similarly, we have not taken the degree of protein phosphorylation into account using these methods, which may impact STAT1/4 function. Second, the study investigated baseline mRNA expression in the peripheral blood only, aiming to capture the effects of ongoing viremia on NK cell transcripts in the blood compartment. However, since leukocytes residing in the liver parenchyma (infected by HCV and HBV) are phenotypically and functionally not identical to their circulating counterpart (13), our findings may not necessarily be generalized to intrahepatic NK cells. Third, although we did not determine killer-cell immunoglobulin-like receptor (KIR) or human leukocyte antigen (HLA) polymorphisms in this study, these polymorphisms, as well as the CMV status of the included subjects, may affect NK cell differentiation, phenotype, and function. We matched the infected cohorts to healthy controls based on ethnicity (as well as gender and age), and we therefore estimate the effect of uneven KIR/HLA distribution and CMV status to be of minor influence. Fourth, we sorted bulk CD56+ lymphocytes and did not separately analyze the functionally distinct CD56bright and CD56dim NK cell subpopulations. However, this approach is sufficient to uncover major effects of viremia on bulk NK cell transcripts, and the frequencies of these subpopulations were similar in all infected cohorts. Despite these limitations, this is the first observational study in which gene expression profiles in sorted NK cells were compared, and it provides leads that may be the first step in uncovering the origin of the functional and phenotypical differences between viremic patients and healthy subjects.
MATERIALS AND METHODS
Patient selection and characteristics.All patients were recruited from the outpatient clinic of the Erasmus MC Rotterdam. The HBV cohort was of Asian, and the HCV/HIV cohorts were of Caucasian ethnicity (Table 1). All infected cohorts were matched by age, gender, and ethnicity to healthy individuals (HBV patients to healthy control group 1 and HIV/HCV to healthy control group 2 [Table 1]). Included patients did not have other liver disease, were not pregnant, and were not on anti(retro)viral therapy for 2 years. Liver fibrosis was determined by histology or transient elastography (maximum of F2 Metavir score/7.0 kPa). Patients were categorized into four clinical HBV phases according to the 2017 EASL Guidelines (16) and in correspondence with our previous work (17, 18) (Fig. 3A). In clinical practice, the HBV clinical phases have been used to guide the moment to start antiviral suppression with nucleotide analogs (recommended in the IA and ENEG phase). CD4 counts were measured using Canto II flow cytometer (BD). The serum ALT level was measured on an automated analyzer, and serum HBsAg and HBeAg levels were measured on an Abbott Architect analyzer. The serum HIV viral load, the HCV viral load, and the HBV DNA levels were measured using the COBAS AmpliPrep-COBAS TaqMan HIV, HCVv2, and HBVv2 (Roche Molecular Systems), respectively.
Variation of serum HBV DNA and ALT levels among four HBV cohorts is associated with variation of NK cell immunological genes but has minimal impact on the expression of individual genes related to cytotoxicity or IFN signaling. (A) Graphical representation of the patient characteristics of the HBV-infected cohort. Based on serum HBV DNA, ALT levels, and HBeAg presence at the time of sampling, patients were categorized into four clinical HBV phases according to international guidelines (European Association for the Study of the Liver, 2017). The HBsAg levels are presented to illustrate lower values among IC and ENEG patients which are characteristic for the natural history of HBV. (B) Total number of DEGs in each HBV clinical phase. The numbers of upregulated (rightward) or downregulated (leftward) genes per phase are shown as horizontal bars. (C) The top up- or downregulated genes are depicted per phase compared to healthy individuals. Regardless of serum ALT levels, no ISGs were induced, except for IRF4. IRF4 was highly expressed in all phases, except for the ENEG phase. Among the genes upregulated in the IA phase, a few are immune related, including FCRL2/5, PDLIM2, and SIT1.
Ethics statement.The study was performed in accordance with the Declaration of Helsinki as adopted by the 64th WMA General Assembly, Fortaleza, Brazil, in October 2013. The institutional ethical review board of the Erasmus Medical Center approved the protocols, and written informed consent was obtained from all study participants.
Sorting of CD3− CD56+ NK cells.PBMC were isolated from the peripheral blood of 8 chronic HCV patients, 6 HIV patients, 8 HBV patients in each HBV phase, and 20 healthy controls and stored at −150°C. After thawing, the cells were stained using anti-CD3-Alexa-Fluor 700 (OKT-3, Beckman) and anti-CD56-APC-eFluor 780 (CMSSB, Beckman). CD3− CD56+ lymphocytes were purified using a FACSARIA cell sorter (BD) as depicted in Fig. 1A. The purities of all samples included in the analysis were higher than 95% (Fig. 1B and C). Sorted NK cells were kept in RNAlater (Qiagen), and RNA was isolated by using a PicoPure RNA isolation kit (Arcturus). Although a trend toward higher percentage of CD56bright NK cells was observed in patients versus healthy individuals, we found no significant differences between the ratios of CD56bright versus CD56dim NK cells when comparing the HIV-infected (10.1% versus 5.3% CD56bright/total NK cells [P = 0.46]), HCV-infected (8.8% versus 5.3% of CD56bright/total NK cells, P = 0.17), and HBV-infected (3.6% versus 1.6% CD56bright/total NK cells, P = 0.41) cohorts to controls.
RNA sequencing of purified NK cells.Total RNA was isolated from purified NK cells using a PicoPure RNA isolation kit (Arcturus), and the RNA concentration was measured using the NanoDrop ND1000 spectrophotometer. The quality and integrity of the RNA samples was analyzed on a Bioanalyzer 2100 (Agilent Technologies). mRNA was isolated from total RNA using the poly-T-oligo-attached magnetic beads. After fragmentation of the mRNA, cDNA synthesis was performed. This was used for ligation with the sequencing adapters and PCR amplification of the resulting product. The sequence libraries were prepared using a TruSeq RNA kit (Illumina) according to the manufacturer's protocol. The quality and yield after sample preparation was measured with a DNA 1000 Lab-on-a-Chip. The size of the resulting products had a peak around 300 bp on a DNA 1000 chip. The cluster generation, hybridization to a flow cell, amplification, linearization, denaturation, and sequencing were performed on an Illumina HiSeq 2000, HiSeq 2500, or HiSeq 4000 platform at GenomeScan (Leiden, the Netherlands). The sequencing strategy and data quality control was performed as described previously (19) (∼92% reads; Phred score, >36).
Alignment of short reads to genome.The generated reads were demultiplexed and trimmed to remove low quality nucleotides as well as reads with low quality score (Phred score, <30) using a trimming tool based on the FASTX tool kit (http://usegalaxy.org). The clean reads in the FastQ files were independently aligned to the human reference genome (UCSC hg19) using Bowtie 2.0, with no mismatch allowed. About 70% (66 to 74%) of the reads were uniquely aligned to the reference genome. The aligned reads in SAM files were sorted in SAMTools by the genomic tracks using the default parameters and then exported into BAM files. Before annotating the aligned reads, eight individual BAM files of each sample were merged and indexed in SAMTools.
Differential gene expression profiling and pathways analysis.The sorted reads were assembled to the transcriptomes and quantified to get gene expression abundances by using human genome track on hg19 as the reference. The abundance of read counts for each selected transcript was quantified by assessing the total number of reads mapped to the transcript using the Cufflinks program. The abundance was normalized per kilobase of transcripts per million mapped reads (FPKM). The expression values of each sample summarized at the gene level (summed FPKM of transcripts sharing the same gene symbol) were first normalized to the overall median of that sample then used to perform differential expression testing between NK cells from two groups (significance was defined as >1.5-fold, q-values of <0.2). DEGs were analyzed using core expression analysis (IPA; Ingenuity Systems). The DEG symbols were mapped against the complete Ingenuity knowledge base (genes only) to identify affected canonical pathways. The P values and log2-fold change of the DEG were used by IPA to predict alterations in the activity of the network or pathway. The significance of the prediction made by IPA is based on an activation z-score (values of >2 suggests increased activity), which infers activation states of the pathway compared to a model that assigns predicted changes (20). We focused on canonical pathways that are significantly deregulated in one viral infection. The significance was defined based on P values calculated using the right-tailed Fisher exact test (P < 10−4) and a minimum of 5% deregulated pathway molecules in the disease condition compared to the controls.
Data availability.RNA sequencing data have been deposited in NCBI’s Gene Expression Omnibus (GEO) database under accession number GSE125686.
ACKNOWLEDGMENTS
We thank Heleen van Santen and Melek Polat for help with collecting patient material and Marieke van der Heide-Mulder for excellent assistance with NK cell sorting.
This study was supported by the Foundation for Liver and Gastrointestinal Research (SLO) and the Virgo consortium, funded by Dutch government project FES0908.
L.L.B., R.A.G., and A.V. contributed to acquisition of data. J.H. contributed to the statistical analysis, and L.L.B., R.A.G., and J.H. contributed to the analysis and interpretation of data. L.L.B. contributed to the writing of the manuscript. A.B. designed and supervised the study and acquired funding. L.L.B., A.B., A.V., and J.H. contributed to critical revisions of the manuscript.
A.B. was an advisory board member for Gilead Sciences and has received research support from Gilead Sciences, Fujirebio, BMS, Roche, and Janssen. R.A.G. is currently employed by Gilead Sciences. A.V., L.L.B., and J.H. have no conflicts to report.
FOOTNOTES
- Received 5 April 2018.
- Accepted 20 August 2018.
- Accepted manuscript posted online 5 September 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/JVI.00575-18.
- Copyright © 2019 American Society for Microbiology.
