Simultaneous RNA quantification of human and retroviral genomes reveals intact interferon signaling in HTLV-1-infected CD4+ T cell lines

Background IFN-α contributes extensively to host immune response upon viral infection through antiviral, pro-apoptotic, antiproliferative and immunomodulatory activities. Although extensively documented in various types of human cancers and viral infections, controversy exists in the exact mechanism of action of IFN-α in human immunodeficiency virus type 1 (HIV-1) and human T-lymphotropic virus type 1 (HTLV-1) retroviral infections. Results IFN-α displayed strong anti-HIV-1 effects in HIV-1/HTLV-1 co-infected MT-4 cells in vitro, demonstrated by the dose-dependent inhibition of the HIV-1-induced cytopathic effect (IC50 = 83.5 IU/ml, p < 0.0001) and p24 levels in cell-free supernatant (IC50 = 1.2 IU/ml, p < 0.0001). In contrast, IFN-α treatment did not affect cell viability or HTLV-1 viral mRNA levels in HTLV-1 mono-infected cell lines, based on flow cytometry and nCounter analysis, respectively. However, we were able to confirm the previously described post-transcriptional inhibition of HTLV-1 p19 secretion by IFN-α in cell lines (p = 0.0045), and extend this finding to primary Adult T cell Leukemia patient samples (p = 0.031). In addition, through microarray and nCounter analysis, we performed the first genome-wide simultaneous quantification of complete human and retroviral transciptomes, demonstrating significant transcriptional activation of interferon-stimulated genes without concomitant decrease of HTLV-1 mRNA levels. Conclusions Taken together, our results indicate that both the absence of in vitro antiproliferative and pro-apoptotic activity as well as the modest post-transcriptional antiviral activity of IFN-α against HTLV-1, were not due to a cell-intrinsic defect in IFN-α signalisation, but rather represents a retrovirus-specific phenomenon, considering the strong HIV-1 inhibition in co-infected cells.


Background
The two pathogenic human retroviruses, human immunodeficiency virus type 1 (HIV-1) and human Tlymphotropic virus type 1 (HTLV-1), remain responsible for significant morbidity and mortality worldwide. At present, 33 million people are estimated to be infected with HIV-1 and 10 to 30 million people with HTLV-1 [1,2]. Although both retroviruses share similarities in routes of transmission and in vivo tropism for CD4+ T cells, HIV-1 and HTLV-1 significantly differ in pathogenicity, disease progression and treatment outcome. Current treatment of HIV-1-infected individuals is based on highly active antiretroviral therapy (HAART), combining HIV-1 reverse transcriptase, protease, integrase and/or entry inhibitors [3]. HAART reduces HIV-1 plasma viral loads to undetectable levels in the majority of patients, hereby slowing down clinical progression to the acquired immunodeficiency syndrome (AIDS). HTLV-1, on the other hand, is able to cause adult T cell leukemia/lymphoma (ATLL), as well as HTLV-1 associated myelopathy/tropical spastic paraparesis (HAM/ TSP), albeit in a minority of infected individuals [1,4,5]. ATLL is an aggressive T cell malignancy of mainly CD4 + CD25+ T cells, affecting up to 6% of HTLV-1-infected individuals [5]. The diversity of patient symptoms together with their prognosis has led to the classification of ATLL into four clinical subtypes: acute, lymphoma, smouldering and chronic ATLL. Recently, consensus guidelines have been drafted for ATLL treatment with first line therapy comprising combination chemotherapy regimens for lymphoma patients and zidovudine (AZT) and interferon-α (IFN-α) combination therapy for treatment of acute, smouldering and chronic ATLL patients [6,7]. Up to 5% of HTLV-1-infected individuals may develop HAM/TSP, a neuro-inflammatory disease characterized by demyelinating lesions and lymphocytic activation and infiltration into the central nervous system [8,9]. Treatment of HAM/TSP patients consists of mainly symptomatic (antispasmodics) and empirical strategies such as the use of immunosuppressive drugs (corticosteroids), IFN-α or vitamin C [10]. However, consensus guidelines for HAM/TSP treatment are missing due to the lack of randomized controlled clinical trials.
In HTLV-1-infected cells, in vitro antiviral activity has been reported for IFN-α by a post-transcriptional mechanism impairing viral assembly. Hereby, IFN-α had no effect on viral protein synthesis, but inhibited virion release through the inhibition of Gag protein association with lipid rafts [47]. In addition, IFN-α stimulation induced transcriptional activation of ISGs in HTLV-1infected cells, reflecting intact IFN-α signaling pathways [48]. However, HTLV-1 expression has been reported to blunt IFN-α signaling pathways in vitro via HTLV-1 Tax and suppressor of cytokine signaling 1 (SOCS1) expression [48][49][50]. Such inhibition could impair antiviral, immunomodulatory and antiproliferative activities of IFN-α in vivo. Nevertheless, significant clinical benefit has been demonstrated for ATLL patients treated with AZT and IFN-α combination therapy and HAM/TSP patients treated with IFN-α monotherapy [7,51,52]. In conclusion, although IFN-α signaling pathways contribute extensively to the host immune and antiviral responses and therefore significantly determine the therapeutic success of IFN-α treatment, controversy exists in the exact mechanism of action of IFN-α in retroviral infections.
In the present study, we address the possible antiviral, antiproliferative and/or pro-apoptotic effects of IFN-α in HTLV-1 mono-infected and HIV-1/HTLV-1 co-infected CD4+ T cell lines in vitro. We demonstrate that IFN-α exerts pronounced anti-HIV-1 effects in HTLV-1 coinfected cells, but has a limited, post-transcriptional, antiviral effect upon HTLV-1 in mono-infected cells. As demonstrated by gene expression profiling, both STAT1/ STAT2-mediated antiviral signaling pathways and broad ISGs induction were fully activated by IFN-α in HTLV-1-infected cells. However, IFN-α failed to inhibit mRNA levels of any of the known HTLV-1 genes or to induce significant pro-apoptotic or antiproliferative activity in HTLV-1-infected cells. In conclusion, the apparent absence of IFN-α antiretroviral activity is selective towards HTLV-1, since a significant anti-HIV-1 effect can be observed in co-infected cells.

IFN-α selectively affects cell viability of HIV-1/HTLV-1 coinfected but not HTLV-1 mono-infected cells
To assess the effects of IFN-α on cell viability of HTLV-1-infected cell lines, we quantified pro-apoptotic (activecaspase 3) and proliferation-associated (PCNA) markers via flow cytometry after IFN-α treatment (10 -10 3 IU/ml) at different time points. As illustrated in Figure  Quantification of additional cell death-associated markers (DNA fragmentation, surface Fas) revealed no proapoptotic activity of IFN-α in MT-4 cells after 48 hours of treatment ( Figure 1D). Nevertheless, both MT-2 and MT-4 cells were resistant to IFN-α after 72 hours ( Figure 1B-C) and even in long-term culture (15 days, data not shown), with similar cell viability in comparison to untreated cells.
To evaluate the potential clinical relevance of this observation, we quantified HTLV-1 p19 levels in cell-free supernatant of HAM/TSP (n = 5) and ATLL (n = 6) PBMCs. IFN-α exerted no significant effect upon p19 levels of HAM/TSP PBMCs (t-test, p = 0.63), in agreement to our recent demonstration of a lack of ex vivo antiviral, pro-apoptotic and antiproliferative effects of IFN-α in HAM/TSP patients [54]. However, IFN-α significantly decreased p19 levels of ATLL PBMCs with 26% (t-test, p = 0.031) ( Figure 2E). This modest reduction of p19 levels observed in ATLL PBMCs was highly similar to the IFN-α-induced reduction observed in MT-2 cells ( Figure 2D) and to recent findings in larger cohorts of HTLV-1-infected individuals [55] as well as ATL patients (Khouri et al., unpublished).

Gene expression profiling of IFN-α-treated HTLV-1infected cell lines
While HIV-1 replication was extremely sensitive to IFNα antiviral activity in HIV-1/HTLV-1 co-infected MT-4 cells, anti-HTLV-1 activity remains questionable since we did not detect an effect of IFN-α on cell viability of HTLV-1 mono-infected cells. However, we could measure a modest inhibitory effect of IFN-α on HTLV-1 p19 protein levels in MT-2 cells. To clarify whether IFN-α signaling is blunted in HTLV-1-infected cell lines, possibly explaining the lack of IFN-α responsiveness, we performed gene expression profiling of MT-2 and MT-4 cells after 6 hours of IFN-α stimulation. Agilent Whole Humane Genome microarray analysis was performed in two separate microarray experiments with duplicate samples for both cell lines. Gene expression profiling of MT-2 revealed 77 genes significantly regulated by IFN-α, of which 64 were up-and 13 down-regulated. Gene expression profiling of MT-4 revealed 284 genes significantly regulated by IFN-α, of which 262 were up-and 22 down-regulated. The top 20 of the most significant upregulated genes are shown in Table 1 for MT-2 and Table 2 for MT-4 cells, and included classical ISGs such as Mx, OAS and IRFs genes. We found no common down-regulated genes for MT-2 and MT-4, while 69% (44/64) of up-regulated genes of MT-2 were also upregulated in MT-4 cells. Altogether, gene expression profiling demonstrated significant activation of ISGs and thus intact IFN-α signaling in HTLV-1-infected MT-2 and MT-4 cells.

Pathway analysis confirms intact IFN-α antiviral signaling pathways in HTLV-1-infected cell lines
In order to characterize possible antiviral, pro-apoptotic and/or antiproliferative pathways represented by the respective IFN-α up-and down-regulated genes, which were identified via microarray analysis described above, Ingenuity Pathway Analysis (IPA) was performed for both cell lines. Genes were sorted into molecular gene networks and canonical pathways, of which significantly overrepresented networks/pathways were identified. In the case of MT-2, seven molecular networks were identified as significantly modulated by IFN-α treatment. The principal network contained 25 genes, all up-regulated, representing antimicrobial and inflammatory responses. In the case of MT-4, 12 molecular networks could be identified as significantly modulated by IFN-α treatment. The principal network contained 26 genes, all up-regulated, representing infection mechanism. The principal IFN-α-modulated canonical pathway was identical for both cell lines, i.e. interferon signaling (Figure 3), although subtle differences were observed (IFN-γ, MX1, up-regulated only in MT-2 and JAK2, SOCS1, IRF1, TAP1, OAS1, up-regulated only in MT-4). Thus, IPA illustrated IFN-α-induced modulation of molecular networks and pathways implicated in viral infection and/or IFN signaling in both cell lines. Moreover, no apoptotic, cell cycle-related or antiproliferative pathways were identified, corroborating our flow cytometry findings that IFN-α had no effect on cell viability of HTLV-1infected cell lines.

Lack of antiviral activity of IFN-α in HTLV-infected cell lines
To assess the possible antiviral activity of IFN-α against HTLV-1 and to validate cellular genes identified through microarray analysis, nCounter analysis was performed [56]. All known HTLV-1 viral genes, including the antisense HBZ, and a selection of specific cellular genes, i.e. STAT1, STAT2, STAT3 and CD69, an IFN-induced early activation marker of T lymphocytes, were included for nCounter analysis. mRNA levels of selected genes were quantified and normalized to a cellular housekeeping gene (hypoxanthine phosphoribosyl-transferase 1). IFN-α significantly increased mRNA levels of STAT1 and STAT2 genes in MT-2 and MT-4 cells (n = 4, t-test, p = 0.0024, p = 0.010 and p = 0.0023, p = 0.032, respectively, Figure 4), confirming functional activation of JAK/STAT pathways and IFN-α antiviral signaling. Moreover, mRNA levels of cellular STAT3 and CD69 genes were also significantly increased in MT-4 cells, confirming our microarray results (n = 4, t-test, p = 0.0041, p = 0.0018, respectively, data not shown). As indicated by pathway analysis (Figure 5A-B), IFN-α predominantly activated STAT1 and STAT2 signaling and downstream genes, representing main antiviral pathways, rather than STAT3 signaling, representing inflammatory and carcinogenic pathways. HTLV-1 viral mRNA levels were ± 10-fold higher in MT-2 cells, in comparison to MT-4 cells, in agreement with the superior virion production by MT-2 cells (Figure 2C-D). Surprisingly, after 6 hours, IFN-α treatment had no effect on mRNA levels of any of the HTLV-1 viral genes in both cell lines (n = 4, t-test, Figure 6A-B). Due to the surprising absence of an antiviral effect at 6 hours, we also performed nCounter analysis after 2 and 48 hours of IFN-α treatment of MT-2 and MT-4 cells to verify whether "early" or "late" inhibition of HTLV-1 viral gene transcription might occur. At 2 hours, IFN-α again had no effect on viral gene mRNA levels in both cell lines (n = 4, t-test, data not shown). However, IFN-α significantly increased mRNA levels of STAT1 and STAT2 genes in MT-4 cells after 2 hours (n = 4, t-test, p = 0.023 and p = 0.029, respectively, Additional file 1), illustrating early activation of JAK/STAT pathways and IFN-α signaling. At 48 hours, IFN-α again failed to considerably down-regulate HTLV-1 viral gene transcription (Additional file 1) although IFN-α signaling was sustained in both MT-2 and MT-4 cells, demonstrated by increased STAT1 and STAT2 mRNA   Because of the intact transcriptional activation of ISGs, the absence of distinct antiviral activity of IFN-α against HTLV-1 was not due to a general defect in IFN-α signaling pathways in MT-2 or MT-4 cells. In agreement with our findings, over-expression of a subset of ISGs in chronic HTLV-1 infection has recently been shown to fail to constitute an efficient antiviral host response, but might instead contribute to HAM/TSP pathogenesis [55]. We speculate that IFN-α fails to decrease HTLV-1 mRNA levels due to highly virus-specific retroviral restriction factors, as IFN-α exerted strong anti-HIV-1 yet weak anti-HTLV-1 effects in HIV-1/HTLV-1 co-infected MT-4 cells. Since all known HTLV-1 mRNAs, including antisense HBZ, remain unchanged upon IFN treatment, defective RNAseL activity, downstream of OAS gene activation (reviewed in [57]), might be hypothesized as a possible HTLV-1 escape mechanism,. Although several OAS family members are IFN-inducible in both HTLV-1 infection (this study, Table I and results not shown) and HIV-1 infection [40], little is known of downstream RNAse L activation, which occurs at the protein level [57]. In addition, blunting of IFN-α biological activity has been mainly addressed in HTLV-1 mono-infection, but not in HIV-1/HTLV-1 coinfection. HTLV-1 expression has been reported to up-regulate SOCS1 expression, inducing ubiquitination and proteasomal degradation of IRF3, leading to the inhibition of type I interferon production and thus inhibiting activation of IFN-α signaling pathways [50]. Still, up-regulation of SOCS1 mRNA levels was shown in CD4+ T cells isolated from HAM/TSP patients and asymptomatic carriers, but not from ATLL patients [50]. Furthermore, HTLV-1 Tax has been shown to induce SOCS1 expression, leading to the inhibition of RIG-I-dependent antiviral signaling, but not the JAK/STAT signaling pathways [58]. Inhibition of cytoplasmatic pattern recognition receptors such as RIG-I, has been associated with IRF3 inhibition and thus subsequent inhibition of type I interferon production. Consequently, HTLV-1-induced SOCS1 expression could counteract activation of IFN-α signaling via reduced type I interferon production. However,  whereas exogenous IFN-α has been shown to increase SOCS1 expression in HeLa-cells, HTLV-1 expressed from an infectious molecular clone reduced IFN-αinduced up-regulation of SOCS1 mRNA levels [48]. Taken together, the precise correlation between HTLV-1 and SOCS1 expression and its effect on IFN-α signaling, remains unclear. Our microarray analysis revealed significant IFN-α-induced up-regulation of suppressors of cytokine signaling SOCS1, SOCS2 and SOCS3 levels in MT-4 cells, but not in MT-2 cells, although IPA pathway analysis revealed strikingly similar IFN signaling in both cell lines. Therefore, an IFN-α-induced increased SOCS1 level is not a generalized finding in HTLV-1 infection in vitro and is not by itself sufficient to define the blunted biological activity of IFN-α. HTLV-1 expression has also been reported to upregulate IRF4 levels in HTLV-1-transformed cell lines and PBMCs of ATLL patients [59,60]. IRF4 was shown to negatively regulate type I interferon production and appeared to be associated with AZT + IFN-α antiviral resistance in ATLL patients [59][60][61]. Our microarray results showed no effect of IFN-α on IRF4 expression in MT-2 or MT-4 cells, although more sensitive nCounter analysis revealed slight IFN-α-induced upregulation of IFR4 levels, which was significant in MT-4 cells, but not in MT-2 cells (p = 0.049 and p = 0.65, respectively, data not shown). Furthermore, treatment with exogenous IFN-α was able to activate IFN-α signalling to a similar extent in both cell lines ( Figure 6A-B and Tables 1-2), although SOCS1 and IRF4 were significantly up-regulated in MT-4 cells only. Nevertheless, our study was limited to the broad antiviral, pro-apoptotic and antiproliferative activities of IFN-α, as well as IFN-α signaling in HTLV-1-infected cells, whereas the precise contribution of cellular factors such as SOCS1 or IRF4 has been investigated in detail in previous studies [50,[58][59][60][61]. However, it should be stated that some of these studies [48,50] investigate de novo infection with HTLV-1 molecular clones, in contrast to stable HTLV-1 infection (this study). The latter might be closer to the in vivo situation, considering the latency of the virus in vivo and its slow molecular evolution, pointing at limited de novo infection [1]. In addition, there is increasing evidence that activation of multiple IFN-α signaling pathways is required to generate the antiviral, pro-apoptotic and immunomodulatory effects of IFN-α [16]. Antiviral and antiproliferative activities of IFN-α have been reported to depend on both STAT-and p38signaling pathways [62]. Although we observed transcriptional activation of STAT1, STAT2 and downstream ISGs in both MT-2 and MT-4 cells, other important IFN-α signaling pathways, such as p38-signaling, could be affected by HTLV-1 replication, possibly explaining the absence of explicit antiviral or antiproliferative activity against HTLV-1 in vitro.
Altogether, one can assume that the in vitro antiretroviral activity and, possibly, the in vivo therapeutic success of IFN-α for both HIV-1 and HTLV-1 is determined by virus-specific factors including viral life cycle-related factors (replication, virion production) and the balance between factors blunting or stimulating IFN-α signaling pathways. For example, inhibition of HIV-1 assembly and release of virions by IFN-α has been described through the induction of ISG15, an ubiquitin-like protein [34]. Through ISG15 up-regulation, IFN-α could affect HTLV-1 assembly, which could explain the posttranscriptional inhibition of p19 secretion. On the other hand, a recent systematic screen for antiviral activity of 389 ISGs, revealed the ability of ADAR to enhance HIV-1 replication [31]. Unfortunately, no data are available on IFN signaling and/or specific ISGs levels in HIV-1/ HTLV-1 co-infection, which is of significant clinical importance since several cohort studies have revealed accelerated clinical progression to AIDS and/or increased mortality in HIV-1/HTLV-1 co-infected versus HIV-1 mono-infected individuals [63][64][65].

Conclusion
We found that IFN-α treatment of retrovirus-infected CD4+ T cells revealed strong anti-HIV-1 but limited anti-HTLV-1 effects. We demonstrate intact early as well as sustained IFN-α signaling in HTLV-1-infected cells, despite various reports of SOCS1-and HTLV-1 Taxinduced blunting of IFN-α signaling. We speculate that qualitative rather than quantitative differences in IFN-α signaling and downstream ISGs could determine the therapeutic success of IFN-α in vivo, depending on both host genetics and disease context. Therefore, large cohorts with long-term clinical follow-up of ATLL and HAM/TSP patients, as well as HIV-1/HTLV-1 coinfected individuals are required to identify viral and host factors specifically limiting antiretroviral activity of IFN-α against HTLV-1.

Patients samples and cell lines
Diagnosis of HTLV-1 infection, HAM/TSP and ATLL was made according to published criteria [67], combining ELISA (Murex), Western blot, INNO-LIA (Innogenetics) and clinical data. Written informed consent was obtained from all participants and this study was approved by the Ethics Committees of CpqGM-FIOCRUZ and HUPES/UFBA (Salvador-Bahia, Brazil) and the Universidad Peruana Cayetano Heredia (Lima, Peru). PBMCs of 5 HAM/TSP and 6 ATLL patients (4 acute, 2 chronic) were isolated by Ficoll-Hypaque density gradient centrifugation (Sigma-Aldrich).

Microarray analysis
Total RNA was extracted from MT-2 and MT-4 cells treated in the absence or presence of IFN-α (1000 IU/ ml) at 6 hours, using RNeasy kit according to the manufacturer's protocol (QIAgen Benelux B.V., Venlo, the Netherlands). Agilent Whole Human Genome microarray analysis was performed by the VIB MicroArray Facility (Leuven, Belgium). Data were analysed using the Agilent Feature Extraction Software version 10.1.1.1 and 10.5.1.1. Briefly, the intensities of the fluorescent probes Cy3 and Cy5, representing the transcription values, were log 2 -transformed and normalized by quantile normalization using the R package preprocessCore [74]. The contrasts in expression between IFN-α-treated and untreated cells at 6 hours of stimulation were estimated using the Limma package from Bioconductor (www. bioconductor.org). For the selection of differentially transcribed genes, a fold-change cut-off of two (i.e. an absolute log 2 -ratio larger than 1) was combined with a p-value cut-off of p < 0.05 when corrected for multiple testing. GEO submission of microarray data is detailed in Additional file 2.

Ingenuity pathway analysis
The Ingenuity Pathway Analysis (IPA) program was used to perform a pathway/function level analysis on genes resulting from the microarray analysis for both MT-2 and MT-4 (IPA version 9.0, Build 116623, Content version 3211, Ingenuity Systems, Red Wood City, CA). Hereby, uncorrected p-values and absolute fold-changes were used with cut-offs of p < 0.001 and fold-change value of 2 (up or down), respectively. Based on a scientific literature database, genes were sorted into gene networks and canonical pathways and significantly overrepresented pathways were identified. The maximum number of networks to be generated was set to 25, with a maximum number of 35 molecules per network. The signaling pathways and networks in Figure 3 were generated through the use of IPA (Ingenuity Systems, www.ingenuity.com).

Statistical analysis
Statistical analysis was performed with GraphPad Prism 5 software. Parametric t-tests were used and p < 0.05 was considered significant. In microarray analysis, a moderated t-test was used, as implemented in the Limma package, to test whether a contrast was significantly different from zero. To verify the false discovery rate, p-values were corrected for multiple testing with Benjamini-Hochberg.