Differential Inhibition of HIV Replication by the 12 Interferon Alpha Subtypes

ABSTRACT

Type I interferons (IFNs) are a family of cytokines that represent a first line of defense against virus infections. The 12 different IFN-α subtypes share a receptor on target cells and trigger similar signaling cascades. Several studies have collectively shown that this apparent redundancy conceals qualitatively different responses induced by individual subtypes, which display different efficacies of inhibition of HIV replication. Some studies, however, provided evidence that the disparities are quantitative rather than qualitative. Since RNA expression analyses show a large but incomplete overlap of the genes induced, they may support both models. To explore if the IFN-α subtypes induce functionally relevant different anti-HIV activities, we have compared the efficacies of inhibition of all 12 subtypes on HIV spread and on specific steps of the viral replication cycle, including viral entry, reverse transcription, protein synthesis, and virus release. Finding different hierarchies of inhibition would validate the induction of qualitatively different responses. We found that while most subtypes similarly inhibit virus entry, they display distinctive potencies on other early steps of HIV replication. In addition, only some subtypes were able to target effectively the late steps. The extent of induction of known anti-HIV factors helps to explain some, but not all differences observed, confirming the participation of additional IFN-induced anti-HIV effectors. Our findings support the notion that different IFN-α subtypes can induce the expression of qualitatively different antiviral activities.

IMPORTANCE The initial response against viruses relies in large part on type I interferons, which include 12 subtypes of IFN-α. These cytokines bind to a common receptor on the cell surface and trigger the expression of incompletely overlapping sets of genes. Whether the anti-HIV responses induced by IFN-α subtypes differ in the extent of expression or in the nature of the genes involved remains debated. Also, RNA expression profiles led to opposite conclusions, depending on the importance attributed to the induction of common or distinctive genes. To explore if relevant anti-HIV activities can be differently induced by the IFN-α subtypes, we compared their relative efficacies on specific steps of the replication cycle. We show that the hierarchy of IFN potencies depends on the step analyzed, supporting qualitatively different responses. This work will also prompt the search for novel IFN-induced anti-HIV factors acting on specific steps of the replication cycle.

INTRODUCTION

Type I interferons (IFNs) are crucial cytokines of innate immunity. Among type I IFNs, the human genome encodes 13 subtypes of IFN-α, two of which have identical amino acid sequences. The similarities and differences of the IFN-α subtypes were recently reviewed (1). They emerged by gene duplication events (2, 3), and in humans they are all located on chromosome 9. Their apparent redundancy was questioned by several studies, which demonstrated that the different subtypes display differential capacities to inhibit virus replication. This was shown for several viruses, including HIV, influenza virus, herpesvirus, hepatitis B virus (HBV), hepatitis C virus (HCV), and hepatitis E virus (HEV) (4–11). Interestingly, the hierarchies of potency of the subtypes are different for different viruses. For instance, subtype α14 was the most effective IFN against HIV (4, 5), while α5 exhibited highest effectiveness against influenza A virus H3N2 (11). These findings on the one hand question the choice to use the same subtype to treat different infections and, on the other hand, strongly suggest that different subtypes induce different factors, which are active against some, but not other viruses.

The 12 different IFN-α proteins share 75 to 99% identity, and all bind to the same heterodimeric receptor (IFNAR) expressed on the cell surface (12–15). Engagement of the receptor triggers a cascade of events through the JAK-STAT pathway, leading to the expression of a large number of interferon-stimulated genes (ISGs) (16). The complex cascade of JAK-STAT activation may help explain how the binding to a common receptor may drive the expression of different gene subsets (17, 18). Indeed, depending on the binding affinity, itself driven by the primary sequence of the IFN subtypes (19), and on the stability of the complex formed with the receptor (20), different forms of phosphorylated STAT proteins and mitogen-activated protein (MAP) kinases can be recruited (21, 22). The homo- or heterodimerization of differently phosphorylated forms of STAT proteins will then drive the formation of different complexes, able to stimulate IFN-stimulated response elements (ISREs) and gamma-activated sequences (GASs) located upstream of different ISGs (16, 19, 23).

Among the antiviral genes induced, some are able to inhibit different steps of the HIV replication cycle, as reviewed in references 24, to ,27. A relationship between the previously characterized binding affinity of the IFN subtypes (19) and anti-HIV efficacy was shown in a mucosal tissue model (4). In that system, IFN-α14, followed by α6, α8, and α17, was a significantly more potent inhibitor of HIV than IFN-α1 and -α2. Also, anti-HIV activity was associated with increased Mx2 and BST-2 RNA expression levels. In a follow-up study, Lavender et al. measured the potency of IFN subtypes in a humanized mouse model (5). The authors reported that the IFN-α14 subtype administered during acute infection was the most potent to inhibit HIV replication, followed by subtypes α6, α17, and α21, while IFN-α2 was again relatively inefficient. IFN-α14 was also associated with reduced immune activation and, again, higher induction of Mx2 and BST-2 RNA expression. The conclusion of these studies agrees with previous literature showing differential induction of ISGs by the different IFN subtypes, as reviewed in reference 13.

Interestingly, a recent report suggested that the observed differences are quantitative rather than qualitative (28). The authors propose that the different subtypes induce similar responses, but with different efficacies. Accordingly, they report that the differences in the anti-HIV potencies observed at low and intermediate doses for four representative IFN-α subtypes vanish when using therapeutic doses that approach saturation of the response. Also, the use of a high concentration of the four subtypes resulted in an overlap of 72 to 80% in induction for a selection of 25 ISGs. A previous comparison of the more distantly related IFN-α2 and IFN-β also showed quantitatively and qualitatively similar ISG induction patterns at saturating doses (29). A very recent analysis of the genes induced by IFN-β and five IFN-α subtypes, however, reached the opposite conclusion, showing that a broader and specific pattern of ISGs was expressed by gut-derived CD4⁺ cells upon exposure to IFN-β (30). Thus, the issue remains debated (31, 32).

To elucidate whether IFN-α subtypes induce measurable and relevant qualitatively different anti-HIV responses, we have explored not only the global anti-HIV activities of the subtypes, but also their relative efficacies at individual steps of the virus replication cycle. If the hierarchy of potency differs for different steps of the virus cycle, this would lend support to the hypothesis of qualitatively different responses. Also, identification of IFN subtypes that potently inhibit specific steps of the virus cycle will facilitate the identification of novel IFN-induced anti-HIV factors. In contrast to some previous studies, we have explored the full panel of subtypes throughout our study. This strategy allowed us to disclose specificities in the anti-HIV activities. While most subtypes potently inhibited the early steps of HIV replication, they display subtype-specific patterns on individual steps, and only for a subset of subtypes does the inhibition of late steps appear to participate in their global anti-HIV activity. Thus, our data support the induction of qualitatively different responses by at least some IFN subtypes.

RESULTS

Activity of the IFN-α subtypes on a generic ISRE.

Before exploring the activity of the 12 IFN-α subtypes on HIV infection, we measured their capacity to induce the ISRE of a gene that does not have direct antiviral activity (gene 6-16), to determine the range of concentrations that triggers an IFN-I response and the extent of it. As in previous studies, we express the IFN concentration in picograms per milliliter rather than in units, because the units are defined based on the potency to a chosen virus, and this would bias the comparison of their antiviral potency on HIV.

We used the previously described reporter cell line HL116 (33), which carries the luciferase gene under the control of the gene 6-16 ISRE. We exposed the reporter cells to increasing concentrations of each IFN-α subtype for 6 h, after which cells were lysed and the luciferase activity measured. As shown in Fig. 1A, the IFN-α subtypes differed both in the minimal concentration capable to induce measurable expression of the luciferase (from 16 pg/ml for IFN-α14 to 80 pg/ml for IFN-α1) and in the extent of induction at each concentration. A sigmoid dose-effect curve was observed, with evidence of saturation of the effect for a high concentration of IFN. For some of the subtypes, however, the plateau was 2 times lower than that for others (e.g., compare α17 and α21 to α8 and α14), indicating differential induction even at saturating concentrations. To quantify the differences of potencies, Fig. 1B shows the half-maximal effective concentrations (EC₅₀). The most potent effect was measured for IFN-α14, followed by α8, α6, and α17. These subtypes were previously shown to have high affinity for the receptor (19). IFN-α1 has the weakest affinity for the receptor, and accordingly, it is the least effective, followed by α4. The other subtypes, including α2, displayed intermediate efficacies. Overall, the potency of induction of this ISRE tends to correlate with its previously described affinity (19) for IFNAR1/IFNAR2 receptor (Fig. 1C) (r = 0.72, P = 0.008).

FIG 1.

Potency of induction of an ISRE by the different subtypes of IFN-α. HL116 cells, which carry the luciferase gene under the control of the ISRE of gene 6-16, were treated with increasing concentrations of IFN-α subtypes (from 3.2 to 10,000 pg/ml). Six hours after treatment, cells were lysed and a luciferase substrate was added. Luciferase activity was then measured (n = 3) and plotted as a function of IFN concentrations (A). Panel B shows the EC₅₀ values for each IFN-α subtype, ordered from the least effective to most effective one. Panel C shows the correlation between the previously determined affinity for the IFNAR1/2 receptor (19) and the efficacy of induction of the expression of luciferase (Spearman two-tailed test).

Inhibition of virus spread by the IFN-α subtypes.

To measure anti-HIV activities of the individual IFN-α subtypes, we initially used the MT4C5 cell line (Fig. 2A and B), which is susceptible to both R5 and X4 HIV strains. MT4-derived cells are among the few cell lines in which anti-HIV activities induced by IFN correspond to those observed in primary lymphocytes (34, 35). Conveniently, following HIV infection, MT4C5 cells do not produce measurable amounts of IFNs in the culture supernatants (36). No significant effect on cell death and proliferation was measured in the MT4C5 cell line for concentrations up to 2 ng/ml (Fig. 2D and E), and this dose was used in all of the experiments.

FIG 2.

Antiviral effect of IFNs on virus spread and on single-cycle infection. MT4C5 cells were pretreated with 2 ng/ml of the 12 IFN-α subtypes 24 h before infection with the R5-tropic virus NL-AD8. (A) Virus spread in culture was monitored over time by the percentage of infected (Gag⁺) cells in culture by flow cytometry. Curves were stopped on the last day for which cell death did not exceed 40%, after which massive cell death prevented precise quantification. One representative experiment is shown. Clustering of potent, intermediate, and weak interferons was confirmed in one independent experiment. (B) In single-cycle infection, nevirapine was added 16 h postinfection to prevent virus spread. Forty-eight hours postinfection, the percentage of infected (Gag⁺) cells was measured. The graph represents the efficacy of infection compared to its value for the condition without IFN (n = 3; ****, P < 0.0001 by analysis of variance [ANOVA] compared to cells not treated [NT]). (C) Gating strategy to measure cell viability and cell proliferation. Uninfected cell viability (D) and proliferation (E) were assessed every day after IFN treatment. The results shown correspond to day 3 posttreatment.

We determined the hierarchy of potencies of the 12 IFN-α subtypes in MT4C5 cells, using a multiple-cycle replication assay. Cells were pretreated for 24 h with each IFN subtype and then exposed to the R5-tropic virus NL-AD8. As shown in Fig. 2A, virus spread in culture was delayed to various extents in cultures treated with the different IFN-α subtypes. IFN-α1 and -α4 were relatively inefficient: they induced a minor delay in viral propagation, and 6 days postinfection, the vast majority of the cells in these cultures were infected, as under the condition without IFN. On the other hand, the subtypes α14 and α17 induced the most marked delay in virus spread, and less than 40% of cells were infected 11 days post-virus exposure. Subtypes α6, α8, α10, α16, and α21 also displayed marked HIV inhibition in our system, while the other subtypes induced only a moderate effect.

Inhibition of HIV infection in a single-cycle assay.

We then measured the inhibition in a single-cycle infection assay that covers the effects on entry, reverse transcription, integration, and early viral protein expression. Cells were pretreated with IFNs for 24 h before infection by the NL-AD8 virus. Nevirapine was added 16 h postinfection to prevent virus spread. The percentage of infected cells (Gag⁺) was measured 48 h postinfection by flow cytometry.

The extent of inhibition in MT4C5 cells was very potent for all subtypes, with the exception of IFN-α1, ranging from 60% to 90% (Fig. 2B). A similar, but not fully overlapping, pattern of inhibition was observed between the multiple-cycle and single-cycle assays in MT4C5 cells (Fig. 2A and B). IFN-α14 and -α10 were powerful in both assays, and subtypes α1 and α4 were the least effective. However, subtypes α8, α16, α17, and α21 did not markedly differ from α5 and α7 in the single-cycle assay, suggesting that inhibition of the late steps may play a role in their inhibition of HIV spread. Alternatively, differences in efficacy in the two systems may be related to the duration of the anti-HIV effect induced by IFNs.

Inhibition of virus entry by the IFN-α subtypes.

The inhibition observed in the single-cycle assay may result from sequential restriction at different steps of the infection process. We thus measured the inhibition on individual steps, starting with virus entry, the efficiency of which can be evaluated by the Vpr–β-lactamase (BLaM) assay (37). MT4C5 cells were pretreated for 24 h before being exposed to NL-AD8 particles carrying the fusion protein Vpr–BLaM. Every subtype reduced virus entry by 35 to 50%, including IFN-α1 and -α4 (Fig. 3A), whose overall impacts on HIV infection were limited.

FIG 3.

Inhibition of early steps of the HIV replication cycle by the different IFN-α subtypes. MT4C5 cells were pretreated with 2 ng/ml of the indicated IFN-α subtypes 24 h before infection with the R5-tropic virus NL-AD8 (A) or the X4-tropic virus NL4-3 (B). Viral particles were loaded with the fusion protein Vpr-BlaM. Virus entry efficacy was measured by the percentage of BlaM⁺ cells 16 h postexposure, relative to the untreated control (n = 2 in panel A and n = 4 in panel B). Expression of HIV receptors CCR5 (C), CXCR4 (D), and CD4 (E) on the surface of uninfected IFN-treated cells was measured by the binding of fluorescently labeled antibodies and fluorescence-activated cell sorter (FACS) analysis. (F) The efficiency of reverse transcription was measured by a qPCR assay targeting the gag gene in IFN-treated cells 7 h postexposure to the NL-AD8 virus. qPCR values are normalized by actin and expressed as percentages of the value for untreated cells (n = 3; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 by ANOVA compared to NT).

In parallel, we measured the inhibition of entry for the CXCR4-tropic virus NL4-3 (Fig. 3B). The potency of the IFN-inducible IFITM proteins (38) was shown to be influenced by the tropism of the virus (39), and IFN could also modulate the expression of HIV receptors. An overall more effective inhibition of viral entry was observed with NL4-3 (except for IFN-α1), together with more marked differences in the individual inhibition potencies.

The modulation of HIV receptors by IFN subtypes explains at least in part these findings. Indeed, CCR5 expression on the surface of IFN-treated cells was homogeneously reduced by 20 to 25% (Fig. 3C), while CXCR4 expression was generally more markedly and differentially affected (Fig. 3D). In contrast, expression of CD4 was not perturbed by IFN treatment (Fig. 3E).

Inhibition of viral DNA synthesis by the IFN-α subtypes.

Since the levels of inhibition of NL-AD8 virus entry were similar among different subtypes, we could quantify the effect of treatment with IFN subtypes on the synthesis of proviral DNA. This part of the virus cycle was previously shown to be effectively targeted by different IFN-induced effectors, the characterization of which is only partial (35, 40–42). We used a quantitative PCR (qPCR) assay targeting the gag gene to amplify the reverse-transcribed viral DNA 7 h postexposure to the virus inoculum. In contrast to their common effect on entry, IFN-α subtypes displayed very different impacts on DNA synthesis (Fig. 3F). IFN-α10, -α14, -α16, and α17 were particularly efficient, reducing DNA synthesis by 80 to 85%, while subtypes α1, α2, and α21 were the least effective, reducing the signal by 50 to 60%. These findings show that postentry steps up to reverse transcription are differentially inhibited by the different subtypes.

Inhibition of late steps of the virus cycle by the IFN-α subtypes.

Because of the potent reduction of the DNA synthesis, and given the differential effects observed, the analysis of the impact of IFN subtypes on subsequent steps of the virus cycle could not be reliably performed in this experimental setting. To measure the anti-HIV activities of IFN-α subtypes on viral protein synthesis and virus release, untreated cells were infected during 16 h, washed, and separated into subcultures—each of which was treated by one IFN subtype for 24 h. The amount of the viral p24 protein was then measured both in cell lysates and in the culture medium. The total amount of p24 (cell lysate plus supernatant, as shown in Fig. 4A) informs on the efficiency of viral protein synthesis, a part of which is released in the supernatant (Fig. 4B). The efficacy of virus budding can be calculated by the ratio of p24 released in the supernatant divided by the total amount of p24 (Fig. 4C).

FIG 4.

Inhibition of late steps of the HIV replication cycle by the different IFN-α subtypes. Untreated MT4C5 cells were bulk infected with NL-AD8 virus. Sixteen hours later, cells were washed and split into parallel wells to be treated with the different IFN-α subtypes (2 ng/ml). Twenty-four hours after treatment, the supernatants and cell pellets were recovered and lysed, and the amount of p24 was measured by ELISA. Panel A shows the total (supernatant plus cell lysate) p24 produced per well. Panel B shows the amount of p24 released in the supernatants (n = 5). Panel C shows the virus budding efficacy, as measured by the ratio of p24 in the culture supernatant divided by the total p24 and expressed as a percentage of the amount for untreated cells. Panel D shows the residual infectivity of virions produced in the presence of the different IFN-α subtypes. The supernatants of cells infected and treated as described above were collected and used to infect the reporter TZM-bl cells, using p24-normalized amounts of virus. Two days postinfection, the β-galactosidase activity was measured (n = 3; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 by ANOVA compared to NT).

When considering the total amount of p24, half of the IFN subtypes led to a significant decrease (Fig. 4A). IFN-α6, -α8, -α16, and -α17 caused the strongest effect (approximately 50%), and this may participate in the relative potency of subtypes α16 and α17. IFN-α2, -α5, and -α14 also had a significant effect. Of note, the reduced amounts of p24 detected in this experiment may result from inhibition of viral transcription and/or translation. The amount of p24 release in the supernatant (Fig. 4B) mirrors production, and the efficacy of virus budding was not significantly affected by IFN treatment (Fig. 4C). This result can be explained by the fact that we used a virus encoding Vpu protein, which allows it to circumvent the potent restriction factor BST-2. Only some subtypes showed a measurable but not significant effect on virus budding, in particular, IFN-α14 and -α21. Such inhibition may contribute to the potency displayed on virus spread by subtype α21, which was not matched by inhibition of early steps, as measured in the single-cycle assay, or viral protein synthesis.

As a final step of the replication cycle, we measured the infectivity of virions released by IFN-treated cells. For this purpose, TZM-bl cells carrying the Lac-Z gene under the control of the long terminal repeat (LTR) promoter were exposed to p24-normalized virus inputs (Fig. 4D). A decrease in infectivity of 25 to 35% was observed for IFN-α14, -α10, -α8, and -α2, although the effect was statistically significant only for IFN-α14. We verified that under these experimental conditions, IFN carryover does not affect HIV infection (data not shown).

Overall, our results indicate that late steps of the HIV replication cycle are less effectively restrained by IFN treatment than early steps, and only a subset of IFN-α subtypes exert significant inhibition.

Inhibition of HIV primary viruses by the IFN-α subtypes.

Differences in IFN susceptibility were previously described for patient-derived viruses in association with their exposure to endogenous or exogenous IFN, with transmitted/founder (T/F) viruses described as less susceptible in some reports but not in others (43–49). To verify that different IFN-α subtypes exerted differential effects also on primary viruses, we measured the susceptibility of primary viruses, including T/F and chronic infection isolates, using the single-cycle assay, which covers all the early steps of replication (Fig. 5). In this system, three T/F viruses and a primary chronic infection isolate displayed subtype-specific susceptibilities. Similar to the observation with NL-AD8, subtypes α1 and α4 were the least effective, and α14 was the most effective, with the exception of CH040, for which α8 exerted a stronger inhibition. Differences in susceptibilities depending on the interferon subtypes were thus shared by a series of primary viruses.

FIG 5.

Inhibition of primary HIV isolates by IFN subtypes. MT4C5 cells were treated with the different IFN-α subtypes for 24 h before being exposed to primary viruses produced by transfection of 293T cells. Nevirapine was added 16 h postinfection, and the percentage of infected (Gag⁺) cells was determined 48 h postinfection by FACS analysis. CH040 (A), CH077 (B), and AD17 (C) are transmitted/founder viruses, while StCOR1 (D) was isolated from the chronic phase of infection (n = 3; *, P < 0.05; **, P < 0.01; ***, P < 0.001, by ANOVA compared to NT).

The three T/F viruses displayed quite different overall susceptibilities, with CH040 showing relative resistance (inhibition ranging from 26% to 53% [Fig 5A]), CH077 being the most susceptible isolate (inhibition ranging from 59% to 78% [Fig. 5B]), and AD17 being intermediate (inhibition ranging from 42% to 62% [Fig. 5C]). By comparison, the chronic infection isolate StCOR1 displayed intermediate resistance (inhibition ranging from 44% to 70% [Fig. 5D]), very similar to the T/F isolate AD17. In this experimental setting, we did not observe a trend of lower susceptibility for T/F viruses.

Inhibition of HIV in primary cells.

To validate the main observations of this study in primary cells, we have separately measured inhibition at the early steps (Fig. 6A) and late steps of virus replication (Fig. 6B, C, and D) using peripheral blood mononuclear cells (PBMCs). IFN treatment of PBMCs had a measurable effect on cell death and cell proliferation at the concentration of 2 ng/ml, which became negligible when the IFN concentration was reduced to 0.4 ng/ml (Fig. 6E and F). Inhibition of HIV was thus measured at this IFN concentration, which represents a 5-fold reduction compared to the concentration used in the MT4C5 cell line. As observed in the cell line, pretreatment of PBMCs for 24 h before exposure to HIV resulted in differential inhibition of single-cycle infection, depending on the IFN subtype. As a consequence of the reduced IFN concentration, the extent of inhibition was lower than that observed in the cell line, ranging from 40% to 75%. As in MT4C5 cells, IFN-α1 and -α4 were weak inhibitors and subtypes α8, α10, and α14 were potent inhibitors. Interestingly, IFN-α5 was the most potent subtype in PBMCs, while it was slightly less effective than subtypes α8 and α10 in the cell line, indicating a general but incomplete overlap of responses induced in the two systems.

FIG 6.

Inhibition of early and late phases of HIV replication by IFNs in primary cells. (A) Inhibition of early steps. PBMCs from a healthy donor were treated with the full panel of IFN-α subtypes at the concentration of 0.4 ng/ml, 24 h before being exposed to a luciferase-encoding NL-AD8 HIV strain, in a single-cycle assay. Luciferase activity was measured 48 h postinfection and expressed as percentage of that of untreated cells. (B) Inhibition by IFNs of total p24 production (late step) was measured after bulk infection of untreated PBMCs for 16 h, followed by splitting the culture into parallel wells, where cells were treated by the different IFN subtypes (0.4 ng/ml). The impact of IFN subtypes on the amount of p24 in the supernatant (C) and on virus budding (D) was measured in the same cultures. (n = 3; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001, by ANOVA compared to NT). Uninfected cell viability (E), and proliferation (F) were assessed 3 days after IFN treatment.

To test the effect of IFNs on the late steps of virus replication, untreated PBMCs were bulk infected and then split into independent cultures in the presence of different IFN subtypes. The effect of treatment on the total amount of p24 produced (cell lysate and supernatant) was measured 24 h later (Fig. 6B). Despite the relatively low concentration used, most IFN subtypes significantly reduced viral protein synthesis, with α14 being the most potent, while α5 was among the weakest, in striking contrast with what was observed in the early steps of replication. As observed in the cell line, the amount of p24 in the supernatant reflected the total p24 produced (Fig. 6C), and consequently no significant effect on budding was observed (Fig. 6D).

Overall, the different hierarchies of potency observed in PBMCs between early and late steps of virus replication strongly suggest that treatment of PBMCs with different IFN subtypes results in the induction of subtype-specific anti-HIV activities.

Induction of known anti-HIV factors by the IFN-α subtypes.

In an attempt to associate the observed differences in virus inhibition with the induction of anti-HIV genes, we measured the RNA expression levels of several known anti-HIV factors following treatment of MT4C5 cells with each IFN subtype (Table 1). We then investigated the correlation between the RNA expression level and the extent of inhibition of steps of the virus cycle on which these effectors are known to act. The efficiency of HIV entry into target cells can be inhibited by the expression of the IFN-inducible IFITM proteins (50) and by the IFN-inducible cholesterol-25 hydrolase (CH25H), which reduces the fluidity of the target cell’s membrane, thereby inhibiting the entry of different viruses, including HIV (51, 52). The expression of CH25H in MT4C5 cells, however, was not measurably induced by IFN (Table 1). The IFITM1 gene, instead, was induced by all IFN subtypes (Table 1), and a significant inverse correlation was found between the RNA expression level of IFITM1 and the efficiency of entry for both R5 (r = 0.80, P = 0.003 [Fig. 7A]) and X4 (r = 0.66, P = 0.024 [Fig. 7B]) viruses. Although expression of IFITM2 gene was also modulated by most IFN-α subtypes, some subtypes induced its expression, while others decreased it compared to untreated cells (Table 1). Expression of the IFITM3 gene was also induced by all subtypes, with the notable exception of IFN-α1 and -α10 (Table 1). No correlation was observed, however, between the expression level of IFITM2 or -3 and the extent of inhibition of HIV entry. Thus, the inhibition of HIV entry may be in part due to the induction of IFITM1.

TABLE 1.

Quantification of RNA expression for known anti-HIV factors induced by the different subtypes of IFN-α

Factor	Fold change in factor RNA expression vs untreateda
	IFN-α1	IFN-α2	IFN-α4	IFN-α5	IFN-α6	IFN-α7	IFN-α8	IFN-α10	IFN-α14	IFN-α16	IFN-α17	IFN-α21
IFITM1	6.4	9.4	7.0	8.0	8.0	8.0	7.4	8.0	10.6	9.4	9.3	6.7
IFITM2	0.4	4.5	4.3	0.4	5.6	2.3	27.1	3.0	0.6	0.6	0.3	0.3
IFITM3	1.9	2.8	2.2	3.0	3.6	3.2	4.2	1.9	3.4	3.1	2.4	2.1
CH25H	ND	ND	ND	ND	ND	ND	ND	ND	ND	ND	ND	ND
TRIM5α	2.3	2.6	2.2	2.6	2.8	2.6	3.1	2.6	3.5	2.8	2.5	2.2
SAMHD1	1.2	1.4	1.4	1.5	1.5	1.6	1.9	1.7	1.8	1.5	1.3	1.5
Mx2	1.3	3.3	2.5	2.0	2.4	1.7	3.7	2.1	5.1	2.4	2.0	1.2
TRIM22	3.4	4.9	4.0	4.6	4.5	3.9	5.8	4.2	5.9	4.7	4.2	3.5
SLFN11	1.4	1.3	1.3	1.1	1.3	1.0	1.1	1.1	1.3	1.4	1.0	1.2
ISG15	29.3	49.9	40.0	42.5	48.8	45.4	83.7	48.1	83.0	55.6	59.5	35.2
BST2	2.0	2.5	2.2	2.3	2.5	2.2	3.0	2.7	3.2	2.9	2.5	2.3
APOBEC3G	1.4	1.5	1.3	1.2	1.2	1.2	1.2	1.1	1.4	1.4	1.3	1.1
GBP2	1.3	1.6	1.0	1.4	1.6	1.5	1.8	2.4	2.3	1.4	1.3	1.2
GBP5	1.1	1.0	1.1	1.5	1.4	1.9	1.8	1.0	1.7	1.1	1.1	1.0
90K	9.7	14.2	4.5	14.5	11.3	26.1	17.5	4.4	8.4	20.0	14.3	6.0
MARCH8	0.9	0.8	0.8	1.1	1.2	0.7	1.0	1.0	0.9	1.0	1.0	0.9
SERINC3	1.0	0.8	0.9	0.8	0.8	0.8	0.7	0.9	0.8	0.8	0.7	0.8
SERINC5	0.8	1.2	1.2	1.3	1.5	1.2	1.9	0.8	1.3	0.9	0.7	1.0

^aND, not determined.

FIG 7.

Correlation between the level of expression of specific RNAs and the efficacy of individual steps of the HIV replication cycle. The levels of expression of known anti-HIV factors in cells treated by different IFN-α subtypes (as reported in Table 1) were plotted against the efficacy of completion of specific steps of the HIV replication cycle. A significant correlation was found for 3 genes (Spearman two-tailed test). Expression of IFITM1 inversely correlated with the efficacy of virus entry, as measured by the Vpr-BLaM assay for both R5 (A) and X4 (B) viruses. Expression of TRIM5α inversely correlated with the efficacy of reverse transcription (C), and expression of GBP2 inversely correlated with the infectivity of virions produced by cells treated with the different IFN-α subtypes (D).

The restriction factors TRIM5α and SAMHD1 are known to reduce the efficiency of reverse transcription (53, 54). TRIM5α was induced by all subtypes (Table 1), and the level of induction was inversely correlated with the efficiency of DNA synthesis in IFN-treated target cells (r = 0.66, P = 0.020) (Fig. 7C), suggesting that it may participate in the differential effect measured here. On the other hand, expression of SAMHD1 was not modulated by any IFN subtype (Table 1), as previously reported (4, 5). IFN-induced posttranslational modifications, however, are responsible for its antiviral effect (55), and thus the participation of SAMHD1 in the inhibition of viral DNA synthesis should not be excluded.

Because of the potent and differential effects on reverse transcription induced by IFN-α subtypes, the efficacy of downstream events in the early phase of HIV infection could not be reliably measured. Thus, we could not investigate the correlation between the expression level of the anti-HIV factor Mx2, which inhibits the nuclear import of the preintegration complex (56, 57) and the efficacy of this event in IFN-treated cultures. Expression of Mx2, however, was potently induced by IFN-α14, -α2, and -α8, while subtypes α1, α7, and α21 did not induce its expression in our system (Table 1).

To explore the participation of known anti-HIV factors in the inhibition of late steps of virus replication, we next compared the expression level of TRIM22, an antiviral factor that acts on viral RNA transcription (58, 59). The RNA expression level was increased by all subtypes (Table 1), but we did not find a correlation with the extent of inhibition of p24 synthesized in IFN-treated cultures. The p24 level also did not correlated with the expression of Schlafen factor 11 (SLFN11), a factor that is capable of inhibiting viral RNA translation (60), but which was not induced by IFN in our system (Table 1).

On the contrary, ISG15 RNA expression was potently upregulated by all IFN subtypes (Table 1), and it may participate in the reduction of virus particle production (61), together with BST-2 (62), which was induced by all subtypes except α1 (Table 1), but whose effect in our system would be largely neutralized by the expression of Vpu. We did not, however, find a correlation between the induction of either of these factors and the extent of budding inhibition.

Among the proteins induced by IFN-α and described to be able to reduce the infectivity of virions produced by treated cells, we found a significant inverse correlation between the level of expression of GBP2 (guanylate binding protein 2) (63) and the infectivity of virions (r = 0.75, P = 0.007) (Fig. 7D), but not for other factors, including GBP5, 90k (64), and MARCH-8 (65). The decrease in virion infectivity observed here may also be due to the incorporation IFITM proteins into the viral membrane (66, 67), and a trend was observed with the level of IFITM1 expression (P = 0.056). SERINC3/5 proteins may also be incorporated in the viral membrane and reduce virus infectivity (68, 69). The absence of induction in our setting (Table 1) does not exclude their possible participation, because their subcellular localization (and not the expression) is modulated by IFNs in producer cells (70). Finally, APOBEC3G, whose incorporation into budding virions potently impairs HIV infectivity (71, 72), was not induced in our system (Table 1)—an observation that may be due to the previously described cell-type-dependent inducibility (73).

DISCUSSION

Most of the experimental work on the anti-HIV activities of IFN-I has been performed with IFN-α2, which is also used to treat some viral infections in the clinical settings. Only in the last few years have researchers approached the complexity of the IFN-α subtypes for the induction of an anti-HIV response. The majority of the reports (4, 5, 14, 30, 74) concluded on a qualitatively different response induced by the different subtypes, which accounts for the ranking of IFN in terms of potency. It was also established that at least some IFN-α subtypes have evolved under strong purifying selection, demonstrating their nonredundant role (3). A few reports, however, questioned this conclusion, based on the large overlap in the sets of genes that are induced by different type I IFNs (28, 29). In these reports, the qualitative differences observed in the ISG induction profiles are attenuated when clinically relevant, saturating concentrations of IFNs are used. Under those conditions, relatively divergent IFNs, such as IFN-β and IFN-α2, induce similar sets of genes (29), and when primary cells were used, differences in the donors dominated over differences of subtypes when analyzing the expression pattern of selected ISGs (28). Thus, the demonstration that subtypes intrinsically differ in their functional roles was still missing (31, 32).

Two caveats are in large part responsible for the controversy. First, in most reports after an initial characterization of a variable number of subtypes, the authors focused on the comparison of a few subtypes, representative of “potent” and “weak” groups. This approach intrinsically limits the number of differences that may be observed. Second, analysis of the antiviral effect on the full cycle of HIV replication may result in similar inhibitory efficacy for IFNs whose effect is mediated by different antiviral factors.

In our study, we explored the anti-HIV potencies of the full panel of IFN-α subtypes, by characterizing their individual efficacy to inhibit specific steps of the virus replication cycle. By this approach, if IFN subtypes differ only quantitatively in their capacity to induce antiviral factors, the hierarchy of potency should be the same on all steps, while if qualitative differences exist, the orders of potency may be different for different steps.

Our results show that the treatment of both a T-cell line and PBMCs with the different IFN subtypes results in distinct HIV inhibition profiles. We confirm that IFN-α14 is the most potent subtype in inhibiting HIV spread in culture, and we show that its overall potency is due to the ability to potently affect multiple steps of the viral cycle. Conversely, subtypes α1 and α4 are weak inhibitors of each event studied here. Importantly, a group of IFNs displayed relatively potent inhibition of HIV. Among them, IFN-α6, -α8, and -α10 actively restrained the early steps leading to the synthesis of viral DNA in MT4C5 cells, while the efficacy of subtypes α16 and α17 was associated with reduced viral protein synthesis, and that of IFN-α21 was associated with inhibition of virus budding. In PBMCs, inhibition of early steps was very potent following treatment with IFN-α5, while this subtype was among the weakest to inhibit late steps of HIV replication. These results support the hypothesis that different IFN-α subtypes induce the expression of only partially overlapping sets of proteins.

The only known anti-HIV ISG products whose expression level correlated with the antiviral effect observed in our study were IFITM1, TRIM5α, and GBP2, suggesting their potential implication in the inhibition of precise steps. IFITM proteins differ in their subcellular localization, with IFITM1 mostly at the plasma membrane, while the other members of the family locate in endosomes (75). The Env glycoprotein complex of the HIV-1 strain NL-AD8 was shown to be relatively resistant to IFITM restriction (76), and this may explain the stronger inhibition and broader diversity among IFN subtypes observed when we used the X4-tropic NL4-3 virus. Like other IFN-induced antiviral factors, IFITM proteins have been shown to be effective against a large number of enveloped viruses, including filoviruses (Ebola and Marburg viruses), flaviviruses (dengue and West Nile viruses), and a coronavirus (severe acute respiratory syndrome coronavirus 1 [SARS-CoV-1]), as well as influenza A virus (77).

TRIM5α can intercept the incoming virus particle and induce premature uncoating (53). The human ortholog of TRIM5α was initially considered to be inefficient against HIV-1 (53, 78). A few studies, however, measured a significant inhibition of HIV replication and showed that it participates in the selective pressure exerted in vivo on HIV (79–82). TRIM5α was shown to exert significant inhibition of HIV replication in IFN-treated cells (40, 41). Importantly, a recent study demonstrated that the anti-HIV activity of human TRIM5α is mediated by the immunoproteasome (83). Here, we found a significant correlation between its expression level induced by different IFN subtypes and the reduction in the amounts of reverse-transcribed DNA. There is clear evidence in the literature (35, 40, 41, 84) that other factors, not fully characterized, participate in inhibiting early steps of the HIV replication cycle. The comparison of the ISGs induced by different IFN subtypes, in parallel with the measurement of their relative effects on DNA synthesis, may help in identifying such effectors.

Concerning GBP2/5, a series of studies allowed elucidation of their anti-HIV activity, collectively showing that they prevent the maturation of the cellular protease furin, which is responsible for the cleavage of the HIV Env precursor glycoprotein, and affect their glycosylation, thus reducing the incorporation of functional glycoproteins into virions (63, 85–87). Their interference with a cellular helper factor explains the broad pattern of virus restriction, which extends to Zika virus (ZIKV) and measles virus (63).

Determining that different subtypes induce divergent responses has relevant implications. The engagement of different pathogen recognition receptors by different virus families induces the expression of different IFN subtypes (88–90) and of different ISGs (91). Interestingly, Harper et al. found an inverse relationship between the induction of the different subtypes in plasmacytoid dendritic cells (pDCs) exposed to HIV and their relative potencies in preventing HIV replication (4). In their system, IFN-α1 and -α2 were potently induced, while IFN-α14 and then subtypes α6, α8, and α17 were significantly more potent inhibitors of HIV. Independent work has shown that in patients, HIV infection induces the expression of IFN-α2 and to a lesser extent subtypes α6 and α16 (92, 93), and thus it is not surprising that HIV, being exposed to the factors induced by IFN-α2, became relatively resistant to this IFN subtype. Conversely, its susceptibility to some other subtypes is most likely the consequence of low exposure to them. This suggest that we may profit from the broad arsenal of defense mechanisms that can be induced by IFNs by choosing the appropriate subtype and by identifying novel effectors able to interfere with HIV replication. IFNs were recently shown in an in vivo model to be able to control viremia early after infection, if the decay of pDCs (the major IFN producer cells) was prevented (94). Also, their inclusion in strategies aiming the reduction of the latent viral reservoir is being considered (14, 95).

MATERIALS AND METHODS

Cells.

HEK293T cells, HL116 cells, and TZM-bl cells were maintained in Dulbecco’s modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum (FCS) and antibiotics (100 IU/ml penicillin and 100 μg/ml streptomycin). HL116 medium was supplemented with HAT (20 μg/ml of hypoxanthine, 0.2 μg/ml of aminopterin, and 20 μg/ml of thymidine). MT4C5 cells were maintained in RPMI 1640 medium supplemented with 10% heat-inactivated FCS and antibiotics (100 IU/ml penicillin and 100 μg/ml streptomycin). Peripheral blood samples were obtained from healthy donors through the Etablissement Français du Sang in accordance with the approved guidelines. PBMCs were isolated from fresh whole blood by density gradient centrifugation using SepMate tubes (StemCell Technologies) and were stimulated with 100 U/ml of phytohemagglutinin for 24 h. PBMCs were then cultured in RPMI medium containing 10% FCS, antibiotics (100 IU/ml penicillin and 100 μg/ml streptomycin), and 100 U/ml of interleukin-2 (rhIL-2; Immuno Tools). All cultures were maintained at 37°C in a humidified atmosphere with 5% CO₂.

Preparation of virus stocks.

Virus stocks were prepared by transfecting subconfluent 293T cells (in T75 flasks) with JetPei (DNA transfection reagent; Polyplus, reference no. 101-10N), following the manufacturer’s instructions. NL-AD8, CH040, CH077, StCOR1, and AD17 viruses were produced by transfecting 8 μg of pNLAD8, pCH040, CH077, pSTCOR1, or pAD17, respectively. Chimeric Vpr-BLaM NL-AD8, or NL4-3 viruses were produced by cotransfecting 12 μg of pNLAD8 or pNL4-3 with 4 μg of pVpr-BLaM. NL-AD8–luciferase viruses were produced by cotransfecting 7 μg of pLN4-3ΔenvLuc plus 1 μg of pEnvAD8. Medium was changed 16 h later, and the virus-containing supernatant was collected 40 h posttransfection, centrifuged at 500 × g for 5 min to remove cell debris, separated into several aliquots, and frozen at −80°C until used.

IFNs.

The IFN-α subtypes were part of the Human Interferon-α Sampler set (PBL Assay Science, reference no. 11002-1). The kit was stored at −80°C. Prior to experiments, IFNs were diluted in RPMI at a concentration of 1,000 ng/ml, aliquoted in 10 μl, and stored at –20°C. Each aliquot was used for a single experiment.

Measurement of IFN activity on HL116 cells.

HL116 cells were plated on 96-well, black, flat-bottom plates. The next day, the cells were treated with each IFN subtype at concentrations ranging from 10 ng/ml to 0.0032 ng/ml (5-fold dilutions). Six hours after treatment, the cells were washed with phosphate-buffered saline (PBS) and then lysed for 10 min at room temperature.

Luciferase activity was measured with the Luciferase Assay System kit (Promega, reference no. E4550), following the manufacturer’s instructions, with a Varioscan device programmed to inject the luciferase substrate into each well, with a 2-s delay between injection and reading and 10 s of integration.

Multiple-cycle assay.

Twenty-four-well plates were seeded with MT4C5 cells and treated with the different subtypes of IFN-α at a final concentration of 2 ng/ml. Twenty-four hours later, the cells were exposed to a replicative NL-AD8 virus for 2 h. Then, the supernatant was removed and replaced with fresh medium containing each of the IFN-α subtypes at a final concentration of 2 ng/ml. Every 2 or 3 days, the medium of infected cells was replaced with fresh medium containing a 2-ng/ml final concentration of each IFN-α. Each day for 11 days, a fraction of the cells was removed and fixed with 1% paraformaldehyde (PFA), and an intracellular Gag staining with KC57-fluorescein isothiocyanate (FITC; Beckman) was performed. The fluorescence intensity of the antibody was then measured by flow cytometry.

Single-cycle infectivity measurement with MT4C5 cells.

Ninety-six-well round-bottom plates were seeded with MT4C5 cells and treated with 2 ng/ml of each IFN subtype, respectively. Twenty-four hours later, cells were infected with NL-AD8, CH040, CH077, AD17, or StCOR1 virus. Sixteen hours after infection, cells were treated with nevirapine (10 μM) to prevent multiple cycles of replication. Forty-eight hours after infection, the medium was removed and the cells were fixed with 1% PFA. Then, an intracellular Gag staining (KC57-FITC; Beckman) was performed. The fluorescence intensity of the antibody was then measured by flow cytometry.

Single-cycle infectivity measurement with PBMCs.

Ninety-six-well round-bottom plates were seeded with PBMCs and treated with 0.4 ng/ml of each IFN subtype. The cells were infected 24 h later with the NL-AD8–luciferase virus. Forty-eight hours after infection, the medium was removed, and the cells were lysed. The luciferase activity was measured with a Renilla luciferase assay system (Promega, reference no. E2820), according to the manufacturer’s instructions, with a Varioskan Flash device (Thermo Fisher Scientific) programmed to inject the luciferase substrate into each well, with a 2-s delay between injection and reading and 10 s of integration.

Cell viability and cell proliferation.

MT4C5 cells or PBMCs were incubated with carboxyfluorescein succinimidyl ester (CSFE) (CellTrace CFSE cell proliferation kit; Invitrogen, reference no. C34554) following the manufacturer’s instructions and then were split for treatment with 2 ng/ml (MT4C5 cells) or 0.4 ng/ml (PBMCs) of each IFN subtype. Then, a part of the culture was collected every day and incubated with the LIVE/DEAD Fixable Violet Dead Cell Stain kit (Invitrogen, reference no. L34964), and the fluorescence intensity of the CFSE and the ratio of live to dead cells were measured by flow cytometry.

Virus entry efficiency: Vpr–β-lactamase assay.

The measurement of virus-cell fusion efficiency was performed using chimeric viruses whose Vpr protein was fused with β-lactamase (Vpr-BLaM) (37). MT4C5 cells were treated with 2 ng/ml of each IFN subtype. Twenty-four hours later, cells were collected and infected with a Vpr-BLaM-expressing NL-AD8 or NL4-3 virus for 4 h at 37°C.

The cells were washed and incubated overnight with CO₂-independent medium supplemented with 10% FCS and antibiotics (100 IU/ml penicillin and 100 μg/ml streptomycin) and CCF2 (LiveBLAzer FRET-B/G loading kit with CCF2-AM; Invitrogen, reference no. K1032). The next day, cells were fixed with 1% PFA, and the fluorescence intensity of cleaved and uncleaved CCF2 was measured by flow cytometry.

Expression of the HIV-1 receptors: CD4, CCR5, and CXCR4 staining.

MT4C5 cells were treated with 2 ng/ml of each IFN subtype for 24 h. Then, the cells were stained with anti-CD4 antibodies (Abs) (mouse anti-human CD4, clone OKT4; BioLegend), anti-CCR5 monoclonal antibodies (MAbs) (mouse anti-human CCR5 fluorescein MAb, clone 39A; BD Biosciences), and anti-CXCR4 MAbs (phycoerythrin [PE]-conjugated mouse anti-human CD184 clone 12G5 [RUO]; BD Biosciences). The fluorescence intensity of every group of antibodies was then measured by flow cytometry.

Reverse transcription efficiency assay.

MT4C5 cells were treated with 2 ng/ml of each IFN subtype. Twenty-four hours later, cells were infected with an NL-AD8 virus. To prevent DNA contamination due to transfection, this virus was amplified for 1 week on MT4C5 cells, with several washes before infection. Seven hours postinfection, DNA extraction was performed. Reverse transcription efficacy was quantified by quantitative PCR (qPCR) targeting of the gag gene with the primers gagFw (5′-TAGAGGTAAAAGACACCAAGGA) and gagRv (5′-ATGTCCCCCCACTGTGTTTA) and with the gag probe (6-carboxyfluorescein [FAM]-AGAAGAGAAGGCTTTCAGCCCAGAAGTAATACCC–6-carboxymethylrhodamine [TAMRA]). Results were normalized by calculating the number of gag copies per cell after evaluating the amount of cells by qPCR of actin with the primers ActinFw (5′-AACACCCCAGCCATGTACGT) and ActinRv (5′-CGGTGAGGATCTTCATGAGGTAGT) and with with the probe ActinProbe (5′-FAM-CCAGCCAGGTCCAGACGCAGGA-TAMRA). qPCR assays were performed with FastStart Essential DNA Probes Master (Roche, reference no. 06402682001) using a LightCycler 96 (Roche).

Protein synthesis and budding efficiency measurement.

A T75 flask was seeded with MT4C5 cells or PBMCs, and the cells were infected with an NL-AD8 virus. Sixteen hours postinfection, the medium was changed, and the cells were plated and treated with the different IFN-α subtypes at a final concentration of 2 ng/ml for MT4C5 cells and 0.4 ng/ml for PBMCs. Twenty-four hours after treatment, a part of the supernatant was collected and stored at −80°C. The remaining supernatants and the cell pellets were lysed. The p24 concentrations were measured by enzyme-linked immunosorbent assay (ELISA) (INNOTEST HIV Antigen Mab; Fujirebio, reference no. 80564), and the budding efficiency was calculated as follows: budding = [supernatant p24]/[supernatant p24] + [pellet p24]).

Residual virus infectivity test.

Ninety-six-well flat-bottom plates were seeded with TZM-bl cells. The next day, the medium was changed, and the cells were infected in the presence of 30 μg/ml of DEAE-dextran with an equal concentration of nanograms of p24 of the viruses produced on MT4C5 cells treated with each IFN-α subtype at a final concentration of 2 ng/ml. Forty-eight hours postinfection, the medium was removed, and the cells were washed with PBS and then lysed. The β-galactosidase activity was measured with the βGal Reporter Gene Assay kit (Roche, reference no. 11758241001) on a Varioscan device.

RNA expression analysis.

RNA expression analysis was performed by RNA sequencing in untreated MT4C5 cells or cells treated with 2 ng/ml of each subtype of IFN-α at 24 h posttreatment. The NEBNext Ultra II Directional RNA Library Prep kit for Illumina (New Englands Biolabs, Inc., Ipswich, MA, USA) was used to prepare total RNA sequencing libraries as previously described (96). The quality of final amplified libraries was examined with a DNA screen-tape D1000 on a TapeStation 2200, and the quantification was done on the QBit 3.0 fluorometer (ThermoFisher Scientific, Canada). Subsequently, total transcriptome (RNA-seq) libraries with a unique index were pooled in an equimolar ratio and sequenced for paired-end 125-paired-base sequencing using one lane of a high-output flow cell on an HiSeq 2500 V4 system at the Next-Generation Sequencing Platform, Genomics Center, CHU de Québec-Université Laval Research Center, Québec City, Canada. Reads were trimmed using Trimmomatic v.0.36 with the options TRAILING:30, SLIDINGWINDOW:4:20, and MINLEN:30. All other options used the default values. The quality check was performed on raw and trimmed data to ensure the quality of the reads using FastQC v.0.11.5 and MultiQC v.1.5. The quantification was performed with Kallisto v.0.44. Differential expression analysis was performed in R v.3.5.0 using the DESeq2 package v.1.20.0.

Analysis.

The flow cytometry analyses were performed on the FlowJo software. Statistical analyses were performed with the Prism 7 software.