Permanent control of HIV-1 pathogenesis in exceptional elite controllers: a model of spontaneous cure

Elite controllers (EC) represent a small subset of HIV-1-infected people that spontaneously control viral replication. However, natural virological suppression and absence of immune dysfunction are not always long-term sustained. We define exceptional EC (EEC) as HIV-1 subjects who maintain the EC characteristics without disease progression for more than 25 years. We analyzed three EEC, diagnosed between 1988 and 1992, who never showed signs of clinical disease progression in absence of any antiretroviral treatment. A comprehensive clinical, virological, and immunological study was performed. The individuals simultaneously exhibited ≥3 described host protective alleles, low levels of total HIV-1 DNA (<20 copies/106 CD4+ T-cells) without evidence of replication-competent viruses (<0.025 IUPM), consistent with high levels of defective genomes, strong cellular HIV-1-specific immune response, and a high poly-functionality index (>0.50). Inflammation levels of EEC were similar to HIV-1 negative donors. Remarkably, they showed an exceptional lack of viral evolution and 8-fold lower genetic diversity (<0.01 s/n) in env gene than other EC. We postulate that these EEC represent cases of spontaneous functional HIV-1 cure. A non-functional and non-genetically evolving viral reservoir along with an HIV-1-specific immune response seems to be key for the spontaneous functional cure.

more than 10 years of infection and transient EC 10 defined as individuals that lost spontaneous viral control during the follow-up, for comparison with the study subjects. We estimated the average diversity over sequence pairs for each time point in every individual and calculated the average number of base differences per site (p-distance) with 500 bootstrap replicates using MEGA6. All positions containing gaps and missing data were eliminated. Differences in the average p-distance per individual between groups were tested for significance using pairwise Wilcoxon Rank-Sum Test in R v3.5.0, applying the Benjamini-Hochberg procedure for multiple comparisons correction (cut off for significance False Discovery Rate 0.05). immune responses. Cell stimulation. Peripheral blood mononuclear cells (PBMCs) were thawed, washed and in vitro stimulated with overlapped HIV-1 (Gag)-specific peptide pool (NIH AIDS Research and Referenced Reagent Program) as described 10,20 .
Susceptibility and viral inhibition assay. CD4 + T cells alone or in combination with autologous CD8 + T cells from each patient were infected by spinoculation with 40 ng of p24 Gag of the laboratory viral strain HIV-1 NFN-SX (CCR5 tropic), and HIV-1 NL4-3 (CXCR4 tropic). CD4 + T cells were activated for 3 days with PHA and IL-2 before infection. Supernatants were sampled at day 7 and quantified with HIV-1 p24 Gag ELISA (Perkin-Elmer life Sciences, USA).
Assay of soluble biomarkers. Plasma samples were collected in EDTA-lined tubes, and aliquoted and stored at −20 °C. The levels of high sensitivity C-reactive protein (hsCRP) and β2-microglobulin (β2M), D-dimer, IL-6 and sCD163 were performed as described 22 .
HIV-1 antibody determination. Plasma antibody responses were measured in two samples, the 2017 one and another 14 to 19 years before for comparison with two long term viremic controllers as controls (#12 and #60) with a follow-up of >10 years with > 500 cells/µl and VL >2,000 cp/µl. End-point plasma titers of HIV-1 antibodies were measured in an enzyme immunoassay (Genscreen ™ HIV-1/2 Version 2, Bio-Rad). Antibodies against the HIV-1 proteins were determined by Western Blot (Biokit, Werfen, Spain) and Line Immuno Assay (InnoLia, Fujirebio Europe N.V, Belgium) and quantified with the LiRAS software. Neutralizing HIV-1 antibody titers in purified IgGs were tested against a mini panel of isolates 23 .

Statistical analysis.
Changes over time in total HIV-1 DNA were evaluated using the Wilcoxon signed rank test. Gag-specific T-cell response poly-functionality was quantified with the poly-functionality index algorithm 24 employing 0.1.2 beta version of the "FunkyCells Boolean Dataminer" software (www.FunkyCells.com) provided by Dr. Martin Larson (INSERM U1135, France). Analyses were performed with Prism 7 (GraphPad, USA) and the statistical significance was set at 5% for all tests.

Results
clinical characteristics of the individuals. The three individuals of the study (EEC-3, EEC-9, and EEC-56) maintained undetectable viral loads (except two early blips in EEC-9) and stable CD4 + and CD8 + T-cell counts in absence of ART for more than 25 years of infection (Table 1 and Fig. 1A-C). Hepatitis C Virus (HCV) infection was diagnosed in the three individuals because of intravenous drug use (IVDU) and all reported viral clearance, either spontaneously or after treatment ( Table 1). The individuals reported no further exposure to HIV-1 since their diagnosis. The two females of the study gave birth uninfected children in absence of any ART. All individuals were classified as clinical status A, with no symptoms of HIV-1 clinical progression or AIDS defining events.
Participants showed several host protective haplotypes, including HLA-B*57:01, HLA-B*58:01, HLA-B*27:05 or HLA-B*14:02 9,25 . In addition, all individuals presented the T > C mutation in the HLA-C polymorphism (rs9264942) in homozygosis (Table 1) associated with lower plasma viremia and increased levels of HLA-C expression 25 . All subjects displayed three to four host protective haplotypes and a high genetic score (Table 1)

Viral persistence.
To evaluate the extent of viral persistence in the study subjects we used the most sensitive assays available, and compared with values obtained from samples from individuals on ART 26 . The quantification of HIV-1 DNA in infected CD4 + T cells, which are believed to constitute most of the latent reservoir in peripheral blood, showed detectable, albeit extremely low (median of 10.4 [9.1-23.0] copies/10 6 CD4 + T cells) total HIV-1 DNA in all subjects (Fig. 1D). Similar numbers of provirus-containing cells were detected in frozen samples obtained 11 to 13 years earlier (median of 12.5 [IQR 2.5-31.3] copies/10 6 CD4 + T cells) (data not shown) suggesting the stability of the viral reservoir for long periods of time.
As total HIV-1 DNA indistinctly measures both defective and replication-competent forms, we quantified the frequency of IUPM in CD4 + T cells in a co-cultured quantitative viral outgrowth assay (qVOA). We did not detect any replication competent virus in a large amount (28-63 million cells) of highly activated CD4 + T cells from the individuals, indicating values of IUPM below 0.025 (Fig. 1E).
The ultrasensitive plasma viremia assay only detected 2 copies of HIV-1 RNA per mL of plasma in EEC-9 and in the other two subjects it was undetectable in 9 ml of plasma (Fig. 1F). Quantification of cell-associated HIV-1 www.nature.com/scientificreports www.nature.com/scientificreports/ RNA (ca-HIV-1 RNA) showed absence of viral transcription in peripheral CD4 + T cells in the three subjects in more than 1 million tested cells (Fig. 1G).

Sequence analysis, genetic variability and evolutionary dynamics of viral populations. HIV-1
viral replication is inevitably linked with viral diversity and evolution 27 . Therefore, we retrospectively explored genetic evolution in HIV-1 DNA samples from the study subjects during the last 15 years. A total of 86 env sequences from seven different time points over 14 years were obtained from EEC-3, and 94 env sequences from nine different time points over 19 years were recovered from EEC-56. Only two env sequences from one out of five time points were obtained from EEC-9 over 15 years. All env sequences corresponded to R5-tropic viruses according to PSSM genotypic test. Of all the 182 recovered sequences, only two from one sample in EEC-3 were hypermutated, compared to greater numbers found in other HIV-1-infected subjects or individuals on ART 28,29 which support the very low viral replication level in these EEC.
We inferred the HIV-1 infection time of the individuals based on the correlation between Spanish env sequence divergence and sampling time 17 , generating estimations from 1983 to 1986 (Table 1). These estimations suggest that primary infections occurred five to six years before diagnosis and thus these subjects may have been infected for more than 30 years.
In addition, we intended to obtain near full-length genome sequences from proviral DNA. After 337 amplification attempts with 12.4 × 10 6 CD4 + cells and 6.4 × 10 6 PBMCs, only 11 were positive, 9 from patient EEC-3 and 1 for each of EEC-9 and EEC-56. Of these 11 positives, only one complete genome was recovered in EEC-3 (GeneBank MN055643). This sequence showed a G > A deleterious mutation in the major splice donor site in position 290 that suppose a very limited replicative capacity of the virus 30,31 . The remaining sequences (92%) showed important deletions throughout the genome mostly in pol to env genes. These data supported the notion that most provirus in these individuals were defective. Altogether, these results show that this group of EEC has very limited levels of viral persistence in blood, with an extremely low HIV-1 DNA reservoir in peripheral CD4 + T cells with no capacity to generate viral mRNA, viral particles or replication competent virions.
The phylogenetic trees showed very short branches with the major presence of viral sequences with no distance to the most common recent ancestor ( Fig. 2A). This pattern was already evident in the first sample and it was maintained in different samples over the years. There were isolated small clusters of sequences showing longer branches in EEC-3 and EEC-56, which were not hypermutated. However, these viral populations were evolutionary dead ends and did not appear in further samples.
Moreover, the genetic variability of the viral quasispecies showed a very restricted genetic diversity, estimated to be around 0.010 ± 0.003 substitutions per nucleotide (s/n) throughout the entire follow-up (Fig. 2B). The degree of env-based heterogeneity was up to eight times lower in EEC than in other HIV-1-infected groups with controlled viral replication, including subjects on ART, viremic LTNP or transient EC one year before the loss of control (FDR < 0.05) (Fig. 2B). Altogether, the practically null viral genetic evolution and extremely low complexity of the viral populations support the absence of viral replication in our EEC group over 30 years. Year www.nature.com/scientificreports www.nature.com/scientificreports/ one intracellular cytokine (TNFα, IL-2 and/or IFNγ). EEC presented higher levels of Gag-specific total CD8 + T-cells when compared with HIV-1-infected individuals on suppressive ART (Fig. 3A). Such differences were also detected in Central Memory (CM), Effector Memory (EM) and Terminally Differentiated (TD) CD8 + T-cell levels (Supplementary Information. Figure 2A,C). Simultaneous release of multiple cytokines was determined using the poly-functionality index (pINDEX). The pINDEX in Gag-specific total CD8 + T-cells in EEC was higher in three-, four-and five-functions (p = 0.017, p = 0.022, and p = 0.008 respectively) when compared with subjects on suppressive ART ( Fig. 3B and Supplementary Information. Figure 2D,E). These values were clearly higher than those of previously published with transient EC one year before the loss of control 10 .
CD4 + susceptibility and CD8 + viral inhibition. To analyze the susceptibility of the target cells to HIV-1 infection, we pulsed autologous CD4 + T cells from each individual with laboratory-adapted R5-and X4-tropic viral strains for 7 days (Fig. 3C,D). This assay shows that CD4 + T cells from the individuals were susceptible to HIV-1 infection. When we added autologous CD8 + T cells to these co-cultures, we observed a reduction of viral replication from 11 to 74% with the X4-tropic virus and up to 63% with the R5 tropic virus (Fig. 3C,D). Thus, host CD4 + T cells are not intrinsically refractory to HIV-1 infection with R5 or X4-tropic viruses, and that host CD8 + T cells are effective in suppressing viral replication ex vivo.
Myeloid cell quantification. We also compared innate immune cellular subsets from EEC with HIV-1-infected individuals on ART and with non-HIV-1-infected healthy donors (HD) who were matched with EEC by sex and age. EEC had increased levels of total peripheral mDCs compared with subjects on ART (Fig. 3E), while no statistical differences were observed comparing EEC with HD. Similar results were obtained comparing the levels of myeloid subsets CD1c + and CD141 + (Suppl. Figure 2F,G). No differences were observed in pDCs (data not shown). Thus, EEC showed more preserved levels of CD1c + and CD141 + cells than subjects on ART, and similar levels than sex/age-matched HD and other EC 32 .
Inflammation biomarkers. EEC had lower levels of hsCRP, β2-microglobulin and D-dimer levels than subjects on ART, but comparable to HD (Fig. 4A-E). IL-6 and sCD163 were similar between EEC and ART groups, despite the levels of IL-6 and sCD163 were higher than in HD, probably due to previous history of HCV and residual expression of HIV antigens.  HIV-1-specific antibody levels. All EEC included in this study had detectable antibodies against all viral proteins in the LIA-scan analysis and western blot analysis. However, HIV-1-specific antibody levels, determined by ELISA over 14 to 19 years of follow-up, were consistently lower (p = 0.0040) than in viremic LTNP controls (Fig. 4F). Participants displayed a low, albeit persistent, neutralizing response against a mini panel of isolates (data not shown).

Discussion
Herein we characterize three exceptional elite controllers who spontaneously controlled HIV-1 replication and its associated clinical pathogenesis for more than 25 years. According to our observations, the almost total absence of viral diversity and evolution analyzed during at least 15 years, probably due to a replication-impaired viral reservoir, adds to previously described protective host genetic factors and effective immune responses.
All three EEC individuals included in this study showed a consistently low viral DNA reservoir in peripheral CD4 + T cells that was 50 times lower than in individuals on ART using the most sensitive techniques 26 . Total HIV DNA, cell-associated HIV RNA, residual plasma viremia, and IUPM were also lower than other values independently reported for EC 33,34 . Moreover, HIV-1 env sequences had nearly null viral genetic evolution, extremely low complex populations and ancestral dating supporting the absence of viral replication over three decades. This replication impairment is further supported by the variation in only two nucleotides between the only complete genomic sequence of EEC-3 and another one obtained 12 years earlier 15 . The lack of sequence heterogeneity despite a long-term untreated infection might reflect a combination of a slow-evolving virus and the clonal expansion of infected immune cells 35 as suggested by the presence of identical viral clusters in the phylogenetic trees. Finally, the low levels of proviral DNA seemed to be largely defective (>92%), since they had no capacity to generate viral mRNA, viral particles or replication competent virions, implying limited viral persistence in these EECs. The lack of viral replication, as shown by different markers (Fig. 1D-G) and low hypermutation, despite detectable HIV-1 DNA has been mainly associated with the presence of proviral defective genomes and the stability of the viral reservoir 15,16,36 .
Low population size and viral diversity are associated with low replication and viral fitness 37 . These viral characteristics favour important fitness losses, because of the irreversible accumulation of deleterious mutations, known as the Muller´s ratchet effect as observed "in vitro" 38 . The combination of these factors is able to prevent viral fitness recovery, and even if some remaining virus is present in the body, it will not generate replication competent viruses because viral populations do not have enough variability to recover fitness 39 . The low diversity and www.nature.com/scientificreports www.nature.com/scientificreports/ population size could have contributed to the maintenance of the control of viral replication in the participants during the follow-up. All study subjects showed the lowest viral diversity values among comparative groups of subjects on ART or LTNP with viral load control. In addition, in a previous study we showed that low diversity and lack of viral evolution was highly correlated with permanent viral control in persistent EC 10 . On the contrary, transient EC who lost control during the follow-up showed higher levels of genetic diversity 10 (Fig. 2B). Thus, extremely low viral diverstiy could be a convenient prognostic marker for the identification of these EEC.
The functional characterization of viruses from EEC has been limited by the impossibility of complete virus recovery. In a previous study, EEC-56 showed the major presence of 228 nucleotide deletions in the 5′ LTR-gag region; in EEC-3 a 247 nucleotide deletion was positioned in pol gene up to the vif orf 15 . These deleted genomes became dominant during follow-up 15 . In other studies, we could clone env sequences from EEC-3 in recombinant viruses that displayed a very limited and retarded replication capacity 40,41 , similarly to what has been observed in env recombinant viruses from some EC 42 . Pseudoviruses with env clones from EEC-3 exhibited a very low CD4 binding, transfer and fusion capacity, and a very inefficient cell signaling resulting in a very restricted replicative capacity 41 .
According to the data reported in this work, viral antigens and/or truncated viral proteins could be generated in these individuals from defective genomes or from new alternative spliced HIV-RNA variants with translationally competent ORFs as reported in cART subjects 43 . The apparent limited amount of peripheral viral antigen might be associated with the observed low, albeit detectable, levels of HIV-1-specific antibodies, including neutralizing antibodies (data not shown). However, such levels were still superior to those of the two long-term described cases of HIV-1 cure after allogeneic stem cell transplant with a CCRΔ32 donor 4,5 . Likewise, both frequency and poly-functionality of Gag-specific CD8 + T-cell responses were higher in EEC when compared with subjects on ART (Fig. 3A) and similar or above the median of a group of persistent EC (with more than 18 years of controlled infection) and higher than those of transient EC 10 . The CTL responses supported the capacity of the immune system to mount effective adaptive immune responses as confirmed by the ex vivo viral inhibition assay. Inhitibion from autologous CD8 + T cells was similar to previous described EC and higher than comparisions with non-HIV infected donors 44 . Innate immune responses seem to be relatively normalized as for the quantification of inflammation biomarkers and frequency of myeloid cells when compared to HD individuals. Only IL-6 and sCD163 were slightly superior compared to HD to subjects on ART. Values shown in Fig. 4A-E were in the range of those found in indepentdent cohorts of EC 32,45 . This could be due to the previous long-term story of HCV infection, and in fact, sCD163 has been proposed as a marker of liver fibrosis 46 . This is a substantial difference with previously published EC profiles where despite the viral control, inflamation levels were maintained over www.nature.com/scientificreports www.nature.com/scientificreports/ time 10,47 . Whether this could be a distinctive marker easy to assay for EEC needs to be examined in extended populations, but previous works with persistent EC suggested this idea 10,48 . The interplay of T cells with the preserved number and functions of innate cells might have been a critical factor for the maintenance of the high HIV-1-specific T-cell responses in these subjects 49 . Although there is sligth decline in CD4 + T-cell levels in EEC-9, this decrease can be considered physiological with aging, during the 20 years of follow up and with the fact that the studied subjects are now over 60 years old, and moreover this diminution is comparable with European uninfected individuals 50 .
We can only speculate how these HIV-1-infected subjects have resulted in this EEC phenotype of potential functional viral suppression. Thus, primary infection might have occurred with a low fitness viral founder strain, or alternatively, initial innate immune responses might have shaped the selection of an unfit virus. Host genetic factors, including those related with the HLA function, and cellular-adaptive immune responses might have further contributed to this clinically unusual non-progressive profile. Although these are naturally occurring cases, and not the result of a clinical intervention, they provide knowledge on how to achieve a functional cure for HIV-1 infection. Based on all these evidencies we believe that these individuals do not require any ART. Perhaps other individuals, most probably within the persistent EC, LTNP-EC, and EEC groups with similar host (at least three to four alleles), inmune responses (with a polyfunctionality index > 0.50) and viral factors (with stable viral diversity (below 0.010 ± 0.003 (s/n) throughout the entire follow-up) could also have spontaneously achieved a HIV-1 functional cure 51 .
Four previous cases of durable control over HIV replication without ART have been reported 12 . While those cases also had extraordinarily low HIV burdens, they differed from ours in their shorter follow up (median of 9 years since HIV diagnosis) and their weak reactive Western blots. A more recent study in HIV-1 Controllers (HIC) followed up for a median of 21 years has shown a similar small HIV blood reservoir in the presence of weak T-cell activation levels 11 . Our study adds three new additional characteristics to this EEC phenotype: (i) even more time since HIV diagnosis without ART, (ii) low viral diversity and lack of viral evolution over the years, and (iii) inflammatory markers in peripheral blood similar to those in healthy donors.
The incidental accumulation of protective host factors, unfit and/or defective viruses that precluded viral evolution and diversification, effective innate and adaptive immune responses, might have take place simultaneously to achieve a HIV-1 functional spontaneous cure. This model of control suggests that new curative strategies would be successful combining approaches that achieve a reduction of the HIV-1 latent reservoirs, maintain extremely low viral diversity, an impaired viral fitness, and enhanced HIV-1 specific immune responses.