Early Elevated IFNα Identified as the Key Mediator of HIV Pathogenesis and its low level a Hallmark of Elite Controllers

Advances in HIV therapy came from understanding its replication. Further progress toward “functional cure” -no therapy needed as found in Elite Controllers (EC)- may come from insights in pathogenesis and avoidance by EC. Here we show that all immune cells from HIV-infected persons are impaired in non-EC, but not in EC. Since HIV infects few cell types, these results suggest an additional mediator of pathogenesis. We identify that mediator as elevated pathogenic IFNα, controlled by EC likely by their preserved potent NK-cells and later by other killer cells. Since the earliest days of infection predict outcome genetic or chance events must be key to EC, and since we found no unique immune parameter at the onset, we suggest a chance infection with a lower HIV inoculum. These results offer an additional approach toward functional cure: a judicious targeting of IFNα for all non-EC patients.


Introduction
Potent anti-retroviral therapy (ART) was chiefly developed through advances in understanding stages and molecular events involved in HIV replication. However, these advances have not led to a "functional cure" in which no further therapy is needed to suppress virus resurgence and immune decline. One approach is development of longer lasting drugs and better delivery. An additional approach is to attempt a greater understanding of HIV pathogenesis. A small group of HIV-infected patients, with undetectable viral load without therapy, known as elite controllers (EC) (< 0.5%), were found by several clinical investigators and championed by Steven Deeks and Bruce Walker (1) who suggested that perhaps the EC status could be mimicked to reach a "functional cure" (2). Individuals with a genotype including HLAB57+ (ECB57+), which share the Bw4 serotype with the KIR3DL1 allele were found more prone to become EC (3). However, not all EC are HLAB57+ (4). Consequently, other mechanisms must contribute to the EC status. 4 Given our long-term studies on the pathogenic (high levels) effects of IFNα in AIDS (5,6), we questioned whether control of the pathogenic effects of IFNα could be a major immune mechanism that is controlled by EC. Indeed, as described in the extended data Table 1, our initial and other reported studies showed that elevated IFNα exerts several pathogenic effects.
These occur first at the innate phase of an IR by inhibiting IL-7-induced T-cell proliferation controlling T-cell homeostasis (7); second, at the initiation phase of the adaptive IR by inhibiting proliferation of CD4 + T helper cells (8) ; third, by differentiating T-cells to suppressive IL-10-Tr1 cells (9,10). Systemic IFNα (type I IFN) and mucosal IFN-lambda (IFNλ) (type III IFN) both exert antiviral activities via IFN-stimulated genes (ISGs) (11), albeit with different kinetics of ISGs induction and distinct cell targets expressing specific receptors.
IFNΑR1 and IFNΑR2 are constitutively expressed on virtually all nucleated cells such as CD4 + T-cells, whereas at homeostasis IFN receptors are constitutively expressed on epithelial cells and only in a selected pool of immune cells (12). Consequently, the CD4 + T-cells are functionally hampered by elevated IFNα at the initiation of the IR adaptive phase, but not by IFNλ even at high levels (extended data Fig. 1). interestingly, a 2023 report shows the efficacy of therapeutic administration of IFN in early covid patients (13).
The conclusion that elevated IFN is a central mediator of HIV pathogenesis is supported by the literature and by results throughout this paper. First, genetic studies have identified polymorphisms in IFN type 1 and type 3 pathways as contributing to variable responses to HIV infection (14). Second, SIV-infected African green monkeys do not express elevated levels of IFNα and avoid development of AIDS (15). By contrast SIV-infected rhesus macaques develop AIDS and have high level of serum IFNα (16). Third, clinical trials results from the EURIS phase 2B placebo control, carried out in 5 hospital centers from 1996-1998 on 240 patients, who had not received combined ART, were based on the induction of anti-IFNα antibodies by an active vaccine approach to lower IFNα. Vaccine responders achieved all study endpoints, 5 including a lower occurrence of HIV-related events, improved CD4 + T-cell counts and reduced viral load correlating with the rise of anti-IFN Abs (17). Fourth, we show that IFNα induces a series of immune cell abnormalities in untreated non-EC patients (extended data Table 1 and extended data Fig. 1).
We found that HIV untreated patients (non-EC) have abnormally high concentrations of serum IFNα and a high frequency of phenotypic alterations in all immune cell types. These alterations are known effects of IFNα (extended Fig. 1-2 and extended Table 1). In contrast EC show minimal abnormal immune cell alterations. We further identified potential immune mechanisms enabling each EC to control viral replication by its own immune capacity. Finally, we hypothesized that EC control of IFNα is at least partially circumstantial, namely a fortuitous infection with a low dose of HIV. Since IFN levels at the primary phase of the acute infection correlate with virus titer, we surmise a vicious circle of HIV → IFN → more HIV → elevated IFN → more HIV etc may be the critical factor contributing to loss of virus replication control by non-EC, and in the accompanying manuscript we described the mechanism leading to this vicious circle.

IFN is elevated in HIV-infected non-EC and modifies the distribution of blood immune cell types.
Blood immune cell types distribution, from non-EC compared to EC and HD, analyzed by Principal Component Analysis (PCA) shows that non-EC and HD cluster separately, while EC overlap both (Fig. 1A). Fig. 1B and 1C document the distinct proportion of circulating immune cell types in non-EC, EC and HD. We found that IFNα, but not IFNλ2, serum concentration was abnormally increased in non-EC compared to HD and the large majority of EC (Fig. 1D).
We also found a positive correlation between between IFNλ2 and IFNα serum levels in EC but not in non-EC patients (Fig. 1E1). Moreover, serum IFNλ2 concentration was negatively associated with CD4 + T-cells, and positively with CD8 + T-cells frequency in non-EC ( Fig. 1E2-3).

Elevated IFN induces an abnormal frequency and phenotypic alterations of NK-cells and other innate cell types in non-EC but not in EC.
NK-cells are a heterogeneous population of immune cells (24), as shown by spade analysis (Fig.   2A1) and dot plots (Fig. 2A2). As previously reported (25), non-EC had less early and mature NK-cells than HD (2.58% vs 4.61% and 48.1% vs 83.2%), and much more terminal NK-cells (17.91% vs 2.84%) (Fig. 2A3). EC and HD displayed similar proportions of NK-cell subsets.
In the mature NK subset, activating markers Helios, NCR and GrzB/perf, were less frequent in non-EC than in EC and HD ( Fig. 2B1-2). In contrast, inhibitory Killer-cell immunoglobulinlike receptors (iKIRs), which bind HIV-peptides presented on HLA-1b in activated CD4 + Tcells as well as inhibitory checkpoint receptors PD1 or CD39, leading cells to an exhaustionlike status (18), were overexpressed in non-EC compared with EC and HD (Fig. 2B3-4).
Noteworthy, CD38 and HLA-DR were prevalent in non-EC mature NK-cells. In addition, there was positive correlations between IFN levels and the frequency of iKIR + mature NK-cells, terminal NK-cells and iKIR + terminal NK-cells (Fig. 2C). This is consistent with IFN enhancing the development of non-functional NK-cells. Further, indicative of this loss of function among NK-cells in non-EC was the presence of the activation marker NKG2D, only weakly expressed on early NK-cells (Fig. 2D1) and the apoptotic marker CD95 highly expressed on mature NK-cells (Fig.2D2) thereby reducing their cytotoxic activity and increasing their propensity to apoptosis respectively.
To investigate whether elevated IFN induced these altered NK-cell subset distribution and surface phenotypes, we treated in vitro purified NK-cells with increasing dose of IFN. We found that elevated IFN inhibited NK-cells viability and proliferation capability in a dosedependent manner ( Fig. 2E1-2). Furthermore elevated IFN reduced CD56 expression in CD56 dim/neg NK-cells (Fig. 2E3), leading to a decrease of mature NK-cells frequency and a concomitant increase of terminal NK-cells percentage (Fig. 2E4). Addition of increasing amounts of IFN in NK-cell culture induced a dose-dependent decrease of NKG2D expression in early (Fig. 2E5) and increase of CD95 expression in mature NK-cells (Fig. 2E6). In summary the NK-cell abnormalities observed in non-EC can be in large part induced by elevated IFN levels. 8 As to other innate immune cells, CD11c − /CD123 + pDC frequency is lower in non-EC than in EC (13.4% versus 32.5 % respectively) (extended data Fig. 4A). This observation was previously made by others (26).  T-cells show an abnormally high expression of inhibitory checkpoint receptors in non-EC (extended data Fig. 4B). The expression of these checkpoint receptors is triggered directly by IFN including CD38 (27), HLA-DR (27) and indirectly by loss of CD26 (28).

Elevated IFN alters immune cell homeostasis in non-EC compared to HD and a majority of EC.
Homeostatic disturbances of T-cell subsets after HIV infection are hallmarks of early disease.
We investigated the maintenance of CCR7 + CD4 + and CD8 + T-cells in the 3 groups. As anticipated, a lower frequency of both T-cell subsets was observed in non-EC compared to EC and HD (extended data Fig. 4C1-2). In addition, negative correlations were observed between CD4 + and CD8 + CM T-cell frequency and serum IFN levels in HIV-patients (extended data T-cell proliferation (7) occurring in CD4 + and CD8 + CCR7 + T-cells homing to lymph nodes (29).
Interestingly, whereas the frequency of ECB57+ CD8 + CM is similar to HD, the frequency of ECB57− CD8 + CM is reduced, albeit at a lesser degree than in non-EC (extended data Fig.4C5 and Table 1). The maintenance of T-cell homeostasis observed in ECB57+ but not ECB57− (Table   1) may be accounted for by the homology between HLAB57+ and NK iKIR (specifically 3DL1 allele) sharing the bw4 serotype. Therefore, iKIR + NK do not recognize HLAB57 peptide epitopes, and NK-cells killing is not inhibited. This enables the unaltered ECB57+ iKIR + NK cells lysis of infected CD4 + T-cells resulting in reduction of the viral load and its correlated IFN production level (30). Consequently, upon TCR stimulation, we observed a reduced 9 percentage of CM T-cells which differentiate into CD4 + T helper and cytotoxic CD8 + T-cells in non-EC but not in ECB57+.
Besides direct killing of some CD4 + T-cells by HIV, during the acute phase of infection, early HIV regulatory proteins Nef and Tat are released. Nef reduces cellular HLA-1a expression ( 31 ), and Tat enhances IFN production by macrophages (5), thus contributing along with high HIV to the elevated IFN. This in turn leads to NK-cells phenotypic and functionnal alterations. The outcome is a lack of early lysis of infected CD4 + T-cells in most HIV-infected patients, as shown here. These immune cell alterations are spared in EC.

Elevated IFN in non-EC is associated with a high frequency of IFN-induced phenotypic abnormalities of CD4 + and CD8 + T-cells.
We first identified the major circulating CD4 + Tconv (T helper Foxp3 -) cell subsets at different maturation stages based on their surface expression of CCR7 and CD45RA (32). SPADE analysis done on Tconv (Fig. 3A1) and histograms (Fig. 3A2) show that the frequency of their subtypes vary in both non-EC and EC compared to healthy donors. Both EC and non-EC have significantly lower frequencies of naive CD4 + T-cells compared with HD. While EC have higher CM and TEMRA, non-EC have higher EM and TEMRA. (Fig. 3A2). We then explored changes in activation/exhaustion phenotypes in CD4 + T-cell subsets in non-EC compared to EC and HD.
The absence (CD26, CD25, CD28) or increase (CD38, HLA-DR,CD39, PD1 and CTLA-4) of these markers are linked to an exhaustion-like state associated with lack of function (18). Each Tconv subset display an altered distinct phenotypic profile in non-EC compared to EC and HD ( Fig. 3A3 and extended data Fig.5). CM, EM and TEMRA from non-EC exhibited significantly higher phenotypic alteration scores (see methods) compared to those from EC (Fig. 3A4). These phenotypic abnormalities correspond in large part to the IFNα effects on CD4 + T-cells in culture (extended data Table 1 and extended data Fig.1). This is consistent with one key HIV pathogenic mediator, namely the effect of elevated IFN in non-EC. Furthermore, positive correlations are observed between the expression levels of different abnormally expressed markers studied in the CD4 + CM in non-EC but not in EC (extended data Fig.5e). We then investigated the circulating CXCR5 + CD45RO + CD4 + TFH (cTFH) that govern the B-cells hypersomatic mutation and Ig isotype commutation in follicular lymph nodes. Compared with HD, non-EC had lower frequency of cTfh-cells (Fig.3A5a), and, on these cells, they exhibited lower CXCR5 expression level (Fig. 3A5b) and higher coexpression of CD38 and HLA-DR ( Fig. 3A5c). In addition, while similar frequency of CD19 + B-cells was observed between the two studied groups (Fig. 3A6a), non-EC had a reduced percentage of CXCR5 + B-cells ( Fig.   3A6b) and on these cells CXCR5 expression level was lower (Fig. 3A6c). As to the regulatory CD4 + T-cell subset (CD4 + Foxp3 + ), their frequency and function were altered in non-EC but not in EC (Fig. 3B). Their proportion was increased ( Fig. 3B1-2) and the percentage of nonfunctional CD25 − Treg variant (28) was enhanced ( Fig. 3B3-4). Once again, the serum IFN level significantly correlated with this alteration (Fig. 3B5). In addition, memory CD4 + Treg phenotypic abnormalities associated with IFN were multiple in non-EC but minimal in EC. We next performed similar analysis on CD8 + T-cells in non-EC compared to EC and HD. We examined four well-defined CD8 + T-cell subsets as above for the Tconv. Distinct distributions of CD8 + T-cell subsets in the three groups were found, as shown by viSNE analysis (Fig. 4A1) and histograms (Fig. 4A2). Both EC and non-EC had a significantly lower frequency of naive CD4 + T-cells compared with HD. Non-EC also had lower CM and higher EM and TEMRA than HD and EC ( Fig. 4A1-2). As for the Tconv subsets, each CD8 + T-cell subset had a distinct pattern of phenotypic alterations in non-EC compared to EC and HD. Non-EC CM and TEMRA had a higher phenotypic alteration score than those of EC ( Fig. 4A3-5). A large majority of 11 these inhibitory checkpoints receptors are induced by IFN (extended data Table 1).
Furthermore, as described in Fig. 4A6, in non-EC, positive correlations were observed between the expression levels of the various studied markers in CD8 + CM. CD8 + HIV-specific cytotoxic T-cells are comprised of effector HLA-1a-restricted CTL and HLA-E-restricted CD8 + supp carrying iKIRs, the human analog of Ly49, which characterize murine CD8 + supp (33,34).

The minimal anti-HIV immune cells phenotypic defects and the control of HIV replication
in EC is linked to control of IFN. To further identify which anti-HIV immune function, uncommonly found in humans, could account for the control of HIV replication in each EC, we investigated which anti-HIV factor(s) known to neutralize circulating virions or to lyse infected CD4 + T-cells might contribute to the maintenance of the EC status .
Considering the frequency of elevated IFN -induced immune cell types phenotypic anomalies in non-EC, we examined serum IFN levels in all patients. The median in non-EC is 95 fg/mL and in EC 18 fg/mL. In our American HIV patient's cohort, 16 out of 20 non-EC compared to 3 out of 18 EC, showed a higher IFN level (Table 1). Interestingly, among the 3 EC expressing elevated serum IFN, EC11 and EC52 compensate their high IFN level by the expression of NK-cells activating NKG2C receptor contributing to early lysis of infected CD4 + T-cells. EC68 cell sample was not available (Table 1). Regarding IFN it is notable that only EC51 had high level compared to other EC (32000 fg/mL vs 1027 fg/mL). pDC frequency were higher in EC compared to other HIV-infected persons (32% versus 14%, Table 1). It is unlikely that greater numbers of pDC be a key immune parameter causing the EC state and in control of elevated IFNα. Rather it is likely that their increase is a consequence of the better control of IFNα in the early days of infection, thereby avoiding its negative impact.
Given that NK-cells act at the onset of infection, and before the virus setpoint, we focused on them in both ECB57+ and ECB57− . Mature ECB57+ NK-cells (9 out of 10 tested) are predominantly iKIR − , whereas most mature ECB57− NK-cells are iKIR + ( Fig. 5A2 and 5B1). Therefore, these ECB57+ NK-cells have the functional capacity to lyse infected CD4 + T-cells, releasing circulating virions as early as the beginning of the IR. It is notable that one ECB57− (EC13) behaved as an ECB57+, but its mechanism is as yet not defined (Table 1). Interestingly, in ECB57− mature NKcells the expression of NKG2C counterbalance the inhibitory iKIR signals (Fig. 5B3) and allow 13 these NK-cells to kill infected CD4 + T-cells expressing HIV-peptides in an HLA-E presentation, thus reducing the viral load in ECB57-.
The percentage of circulating CD8 + CM was relatively high in ECB57+ (identical to HD median: 10% vs 10%) but reduced in ECB57− albeit higher than in non-EC (median: 6% vs 4%, Table 1 Table 1 and extended data Fig. 1). These few but irreversible alterations might occur during the early period of HIV infection in the innate phase of the IR and be augmented by HIV cytopathic effects induced on thymic stromal cells, known to occur during early infection in the innate phase of the IR, and, as reported, also induced by IFN (35).

Discussion
We conclude that besides circulating HIV, the key mediator of AIDS development is elevated circulating IFN To be a central major mediator of pathogenesis the criterion should be: active at the earliest days of infection ,ie., in the innate immune phase, be elevated and circulating, and have broad pathogenic effects. IFN fits these criteria. In non-EC, it is circulating at elevated levels inducing immune cell type abnormalities linked to poor immune cell function  Table 1). Correspondingly, in EC the absence of circulating high levels IFN is associated with minimal immune cell types anomalies.We propose that EC avoid the pathogenic vicious circle of high HIV → high IFN → more HIV → pathogenic IFN. These assumptions were based on documented experimental data, as follows: 2) The IFN-induced impairment of IL-7-induced immune cell homeostasis found in non-EC results in reduction of NK-cells and CCR7 + CD4 + and CD8 + T-cells. As a consequence, populations of helper and cytotoxic T-cells are reduced during the later adaptive IR. 15 3) The hampered IFN-induced initiation of adaptive IR in non-EC results in diminished CD4 + T-cells activation and proliferation (extended data Table 1).

4) In non-EC, inhibition of CTL and CD8 + supp function linked to IFNα-induced inhibitory
checkpoints results in loss of control of viral replication ( Fig.4 and extended data Table 1).
Concerning the possible immune mechanism enabling control of HIV replication possessed by EC postulated by the AAC (2) (Fig. 5), we found that in addition to our hypothesis of a putative lower infectious inoculum, each EC possessed one or more distinct immune mechanisms (Table   1) which could contribute to their avoidance of elevated IFN levels. These mechanisms include early lysis of infected CD4 + T-cells by phenotypically-unaltered NK-cells and additionally during the adaptive IR by cytotoxic CTLs and CD8 + supp. These data prompted us to consider that the EC status does not solely depend on one specific genetic and/or immune cell profile but also on a distinct immune capacity, peculiar to each EC and likely additional circumstantial factors such as an inoculum low size. No doubt that the HLA presentation of HIV-peptides by infected cells, including the known HLAB57+ but also HLA-E are early contributing factors for NK lysis and, following the adaptive IR for CTL or CD8 + supp T-cells.
It is also clear that HLAB57 genotype is insufficient to account for the EC status, since not all EC are HLAB57+ and further HLAB57+ is also present in up to 11 percentage of patients with progressive disease, a percentage similar to the uninfected Caucasian population (4). As reported by Altfled et al (36), we also suggest that evasion from cytotoxic NK-cell mediated IR by HIV is far more deleterious for early control of viral replication than CTL and CD8 + supp, effective only during the adaptive IR. Considering that many immune mechanisms disrupted by HIV due to elevated IFN in non-EC are avoided by EC (Table 1), we propose that a fortuitous low infectious inoculum is an important contributor to the EC status. This hypothesis has several predictions now under analysis. 16 This study showed by a comprehensive characterization of the pathogenic levels of IFN during both anti-HIV innate and adaptive immune reaction : 1) how our data identified elevated IFN as the key mediator used by HIV to block immune cell attack in non-EC, and 2) why EC, by avoiding this production of elevated IFN, naturally block HIV replication. Thus, these studies suggest that reducing IFN production by any acceptable and efficient means could convert non-EC to EC. This proposal is further supported by the experiment carried out on HIVinfected humanized mice showing that anti-IFNα treatment reduced HIV proviral DNA by 14-fold in spleen cells and by 7-fold in bone marrow cells (27). In an accompanying manuscript, we will describe the mechanism for the IFN inefficient anti-HIV IR promoting its progressive elevation.

Online Methods
Human samples. HD were obtained through Etablissement Français du Sang (EFS, Paris, France). 67 people living with HIV were recruited and subdivided into three group: EC (n=18) were obtained from the NVS cohort (Baltimore), non-EC (n=27) were obtained from NIH (Bethesda n=19) and from the Laboratoire de Référence SIDA (Liège n=11). Patient groups did not significantly differ in terms of age, gender, disease status. All participants or their surrogates provided informed consent in accordance with protocols approved by the regional ethical research boards and the Declaration of Helsinki. Clinical data are indicated in extended data Table 2.  Table 3a).

Flow cytometry analysis.
MAbs panels, staining. Immunophenotypic studies were performed on frozen samples, using up to 23-colours flow cytometry panels. See extender data Table 3  Statistical analyses. Statistical significance of differences between groups was assessed using the unpaired nonparametric Mann-Whitney. Non-parametric, paired Wilcoxon tests were used for paired data. Correlations were assessed by the nonparametric Spearman test. Analyses were performed with GraphPad-Prism, and R. Two-sided P value less than .05 was considered statistically significant (ns: nonsignificant; *P < .05; **P < .01; ***P < .001; ****P < .0001).

Figure 5
Extended data Tables and Figures: Extended data Table 1: IFN type I effects on immune cell subtypes and its correlation  with suppressive cytokines. A-Known potential pathologic effects during chronic viral infection of IFN type I  non-EC B57 + B57 -HD non-EC B57 + B57 -