Differential impact of APOBEC3-driven mutagenesis on HIV evolution in diverse anatomical compartments

Abstract

Objective: 

Previous studies on HIV quasispecies have revealed HIV compartmentalization in various tissues within an infected individual. Such HIV variation is a result of a combination of factors including high replication and mutation rates, recombination, and APOBEC3-host selective pressure.

Methods: 

To evaluate the differential impact of APOBEC3 editing in HIV-1 compartments, we analyzed the level of G-to-A hypermutation in HIV-1 protease and reverse transcriptase sequences among 30 HAART-treated patients for whom peripheral blood mononuclear cells and body tissues or fluids [cerebral spinal fluid (CSF), rectal tissue, or renal tissue] were collected on the same day.

Results: 

APOBEC3-mediated hypermutation was identified in 36% (11/30) of participants in at least one viral reservoir. HIV hypermutated sequences were often observed in viral sanctuaries (total n = 10; CSF, n = 6; renal tissue, n = 1; rectal tissue n = 3) compared with peripheral blood (total n = 4). Accordingly, APOBEC3 editing generated more G-to-A drug resistance mutations in sanctuaries: three patients’ CSF (i.e. G73S in protease; M184I, M230I in reverse transcriptase) and two other patients’ rectal tissues (M184I, M230I in reverse transcriptase) while such mutations were absent from paired peripheral blood mononuclear cells.

Conclusion: 

APOBEC3-induced mutations observed in peripheral blood underestimate the overall proportion of hypermutated viruses in anatomical compartments. The resulting mutations may favor escape to antiretrovirals in these compartments in conjunction with a lower penetration of drugs in some sanctuaries. On the other side, because hypermutated sequences often harbor inactivating mutations, our results suggest that accumulation of defective viruses may be more dominant in sanctuaries than in peripheral blood of patients on effective HAART.

Introduction

Within an individual, HIV exists as a population of related but distinct viral variants termed viral quasispecies ^[1]. These variants can be present in distinct anatomical locations in the same individual and have the properties to evolve independently from HIV found in peripheral blood. Such HIV compartmentalization in conjunction with extensive viral diversity is a key feature of HIV infection. Many viral factors contribute to such diversification: the error proness of HIV reverse transcriptase, recombination, and rapid rates of viral replication ^[2–4]. In addition, recent findings suggest that viral restriction factors APOBEC3 can provide a mechanism of viral diversity ^[5–8]. APOBEC3 proteins are polynucleotide cytidine deaminases ^[9] that mutate and profoundly inhibit the replication of HIV-1. Despite being counteracted by HIV-1 Vif protein, APOBEC3 proteins are incompletely neutralized in vivo^[5]. When fixed in viral DNA, APOBEC3-induced mutations register as guanosine-to-adenosine (G-to-A) changes in the viral plus strand and are termed hypermutations when occurring at excessive levels. These substitutions predominantly occur in a GG-to-GA or GA-to-AA dinucleotide context ^[10].

Knowledge of organ/tissue-specific impact of APOBEC3 in HIV evolution is important for investigating viral compartmentalization in humans. The inhibitory effects of the APOBEC3 proteins could lead to differential accumulation of defective viruses between reservoirs. In addition, it is possible that low levels of activity of these cellular enzymes could be beneficial to HIV-1 in compartments and facilitate immune evasion or accelerate the development of drug resistance. We attempted here to determine the differential contribution of APOBEC3 editing in HIV-1 evolution in different anatomical compartments.

Methods

In this study, 30 HIV-1-infected individuals for whom peripheral blood mononuclear cells (PBMCs) and body tissues or fluids were collected on the same day were investigated. Specifically, we analyzed 14 paired PBMCs/cerebral spinal fluid (CSF) (group 1), eight paired PBMCs/renal tissues (group 2), and eight paired PBMCs/rectal tissues (group 3).

All of the study participants were receiving HAART and had undetectable plasma viremia (viral load <50 copies/ml) at the study time point. Patients were enrolled in previous studies carried out in our hospital under ANRS/ORVACS (agence nationale de recherche sur les SIDA et les hépatites virales/Objectif Recherche VACcin Sida) projects. They were selected on the basis of sample availability.

PBMCs were isolated by centrifugation on Ficoll Hypaque. Rectal and renal biopsies were digested and collected in Roswell Park Memorial Institute medium supplemented with a cocktail of antibiotic and antimycotic. Total DNA was extracted from PBMCs or from 200 ml of resuspended digested biopsies using a QIAamp DNA mini kit (Qiagen, Courtaboeuf, France), according to the manufacturer's instructions, to obtain 100 μl of elute. DNA extracts were stored at −20°C.

For each patient and in each compartment, we performed population-based sequencing of protease and reverse transcriptase using an in-house nested PCR assay. Briefly, 5 μl aliquots of DNA were used for the first round of PCR using Expand High Fidelity PCR System (Roche) to minimize PCR-induced sequence errors following the manufacturer's instructions and using outer primers described at www.hivfrenchresistance.org. The second-round PCR using Expand High Fidelity PCR System (Roche) was performed with inner primers described at www.hivfrenchresistance.org. Each PCR involved 40 repeat cycles (94°C for 30 s, 55°C for 30 s, and 72°C for 1 min) followed by incubation at 72°C for 7 min. The PCR product was purified with an Amicon Microcon-100 centrifugal filter device (Millipore) for sequencing.

To investigate G-to-A substitutions at the population level, we aligned each HIV-1 sequence from each sample to a reference sequence. The reference used in each patient with hypermutated sequences was the pretherapeutic plasma (nonhypermutated) sequence whenever available (all individuals with paired hypermutated PBMCs and/or CSF and PBMCs-renal tissue). However, this information was lacking for group 3 (paired PBMCs/rectal biopsies). In this case, HXB2 was used as the reference sequence.

Differences in the G-to-A mutation frequencies were analyzed using the Hypermut 2.0 program (http://www.hiv.lanl.gov/content/sequence/HYPERMUT/hypermut.html). A sequence was considered hypermutated if it registered a P value of less than 0.05 on the Fisher's exact test that compared the number of G-to-A changes in APOBEC3 versus control contexts. HIV-1 protease and reverse transcriptase sequences recovered from PBMCs and tissues related to this work have been submitted to GenBank and were given accession numbers KF741230 through KF741273.

Results

We first sought to ascertain the population distribution of HIV-1 G-to-A substitutions (GG or GA dinucleotide context/GGG trinucleotide context for APOBEC3G hotspot) in viral sequences from each sample in order to identify APOBEC3-induced footprint within the context of natural in-vivo sequence variation (Fig. 1a,b/supplementary file 1, https://links.lww.com/QAD/A464). Overall, hypermutated sequences were identified in 36% (11/30) of participants in at least one viral compartment. As can be seen in Fig. 1a, APOBEC3-induced hypermutation is more frequent (P = 0.047) in viral compartments (total n = 10; CSF, n = 6; renal tissue, n = 1; rectal tissue n = 3) compared with peripheral blood (total n = 4). Specifically, in the first group of patients (P1-P14, for whom CSF samples were available), hypermutation was detected both in PBMCs and CSF in two patients, whereas four other patients exhibited hypermutated sequences only in CSF. In the second group (P15-P22, for whom renal samples were available), only one patient exhibited hypermutation detected both in PBMCs and in renal tissue. In the third group (P23-P30, for whom rectal samples were available), hypermutation was detected more frequently in rectal tissue (three cases) than in PBMCs (one patient). Interestingly, when hypermutation is detected in peripheral blood as well as in another compartment (P11, P12, P22), the rate of APOBEC3-induced variation at protease-reverse transcriptase sites was different between compartments (Fig. 1b). For example, PBMC viral sequences of patient 22 (P22) showed clear APOBEC3-induced mutagenesis in the protease region (but not in reverse transcriptase), whereas in contrast, analyzing renal tissue in the same patient, viral population showed clear APOBEC3-induced genetic variation in the reverse transcriptase region (but not in protease). Such differential G-to-A hypermutation footprints were also observed between CSF and PBMCs when APOBEC3-mediated editing was detected in both compartments (P11, P12).

Fig. 1: Population distribution of APOBEC3-induced hypermutation in viral compartments in each patient.(a) Graphic representation of HIV hypermutation. Gray boxes represent hypermutated sequences in a specific anatomical compartment; blank boxes mean absence of hypermutation. (b) Differential impact of APOBEC3 editing on viral diversification and emergence of drug resistance mutations in viral anatomical compartments. The protease and reverse transcriptase regions from HIV isolates from each compartment are shown. G-to-A changes in each sample sequence were compared to a reference sequence (see methods) and are indicated by bars. G-to-A changes were classified according to the dinucleotide context in which they appear: APOBEC3 context [GG-to-AG (red bars), GA-to-AA (cyan bars)]; other contexts [GC-to-AC (green bars), GT-to-AT (magenta bars)]. CSF, cerebral spinal fluid; PBMCs, peripheral blood mononuclear cells.

We also investigated the possibility that APOBEC3-induced mutations may affect HIV evolution in anatomical compartments, leading to enhanced rate of variation at sites involved in drug resistance (Fig. 1b). First, focusing on patients with APOBEC3-footprint viruses in anatomical compartments while viruses from paired PBMCs were not hypermutated, we found evidence that some of them (P13, P27, P29) harbored one or several G-to-A drug resistance mutations (G73S in protease/M184I, M230I in reverse transcriptase) in sanctuaries (CSF or rectal tissue) while these mutations were absent from paired nonhypermutated viruses in PBMCs, suggesting that such resistance mutations resulted from APOBEC3 editing. However, these viruses also harbored in frame stop codons at tryptophan positions (as a result of TGG-to-TAG mutations), resulting in viral inactivation. In patients harboring APOBEC3 footprint in PBMCs as well as in anatomical compartments (P11, P12, P22), we noted that most of G-to-A drug resistance mutations vary between compartments likely as a result of variable levels and sites of APOBEC3-induced hypermutation between these compartments. For example, for patient 11 (P11), proviral DNA in PBMCs harbored G-to-A mutations E138K and M184I in reverse transcriptase, whereas viruses in the CSF harbored G-to-A mutations M184I and M230I. Finally, we identified hypermutated sequences in reservoirs (P9, P30) with no evidence of G-to-A drug resistance mutations either in protease or in reverse transcriptase.

Discussion

In the context of the development of novel therapeutic strategies aimed at eliminating all possible sources of HIV replication within an infected individual, careful consideration must be given to the identification and characterization of the viruses residing within a particular cell tissue, and/or organ. One key feature of HIV infection is the extensive diversity of the viral population within an infected individual. This diversity results from a variety of mechanisms including APOBEC3 editing. While the inhibitory effects of the APOBEC3 proteins are well established, it is also possible that low levels of activity could be beneficial to HIV-1 by providing an additional mechanism for acquiring sequence variation through sublethal levels of editing ^[6–8,11]. However, this hypothesis is still controversial in the literature as supported by Armitage et al.^[12] who show that even the incorporation of very few A3G molecules in virions is sufficient to cause inactivating levels of HIV hypermutation. It is unlikely, in these conditions, that APOBEC3 contributes to viral diversification. However, its impact in vivo may be relevant as recombination between hypermutated and nonhypermutated viruses (that may lead to viruses with advantageous mutations) could be important.

In this study, analyzing APOBEC3-induced hypermutation in different reservoirs, we show that its impact is different between anatomical compartments within an infected individual, thus bringing a further mechanism of diversification in a given target tissue within a given microenvironment. Differential APOBEC3 activity might be explained by distinct expressions of APOBEC3 proteins in tissues. In this regard, based on an extensive survey of A3 expression in 20 tissues, Refsland et al.^[13] showed that the A3 mRNAs are expressed broadly, with different levels of APOBEC3 expression and not confined to cells of the immune compartment. Wang et al.^[14] provide experimental evidence that there is intracellular expression and regulation of functional APOBEC3G in human neuronal cells, astrocytes, and microglial cells, which might have a role in glial and neuronal cell-mediated innate defense against HIV infection in the central nervous system. Similarly, in another experimental study, Depboylu et al.^[15] observed an increased APOBEC3G expression in microglia/macrophage-derived cells and T lymphocytes in rhesus macaque brain after simian immunodeficiency virus infection. Induction of APOBEC3G was accompanied by G-to-A hypermutations in the gag and pol regions of retroviral DNA isolated from brain of monkeys. These data are consistent with our study showing a higher rate of APOBEC3-induced hypermutation in CSF compared with PBMCs.

Regarding drug resistance, one key question is whether APOBEC3 editing might participate to the compartmentalization of drug resistance in specific reservoirs. Our study evidenced that G-to-A drug resistance mutations can be detected in hypermutated sequences recovered from sanctuaries when absent from paired PBMCs, suggesting the implication of APOBEC3 editing in the emergence of such mutations in sanctuaries. In the present dataset, hypermutated sequences are clearly defective because of the presence of in frame stop codons (as a result of TGG to TAG mutations at tryptophan positions). However, we previously detected ^[16] in other clinical isolates (patients PBMCs) hypermutated sequences with drug resistance mutations (E138K and M184I), but with no inactivating mutations in reverse transcriptase or in protease. In addition, as discussed above, viral recombination between compartments has been shown to possibly generate viruses with advantageous phenotypic properties ^[17,18]; it is, thus, possible that cytidine deamination in combination with viral recombination allows for the selection of replication-competent viruses with beneficial mutations ^[11] in the context of the generation of a very large number of viruses produced within an infected individual. One other original information brought by our study is the impact of APOBEC3 in inducing differential defective viral genomes between reservoirs ^[19]. G-to-A hypermutation in HIV-1 is clearly capable of limiting virus proliferation due to the introduction of inactivating mutations ^[20–22], although the virus can integrate into cellular DNA and persist in resting CD4⁺ T cells. As APOBEC3-induced mutagenesis is one of the major actors of virus inactivation in vivo^[19], these results suggest that under suppressive HAART, remaining replication-competent HIV genomes are present at different rates between compartments. Our study has, however, some limitations: hypermutated sequences were analyzed at a population level as no individual cloning was performed. Therefore, it is difficult to estimate the level of hypermutated genomes per sample. In addition, considering the amount of template sequences between compartments, when there is a low number of templates, it is difficult to derive conclusions about the significance of hypermutated/nonhypermutated sequences.

In summary, our data provide support for a wide range of HIV mutational process induced by APOBEC3 proteins between anatomical compartments in vivo, which might play a role in emerging drug resistance. On the other side, it highlights the differential degree of accumulation of defective viruses between viral compartments in patients under suppressive HAART that should be taken into consideration in the context of future viral eradication strategies.

Acknowledgements

The authors gratefully thank G. Le Mallier and P. Grange for their technical assistance.

S.F., V.C., and A.-G.M. designed research, were involved in interpretation of the data, and wrote the article.

S.F., S.L.-N, C.S., I.M., M.W., and G.C. performed research.

M.A.V., R.T., A.S., and C.K. were responsible for patient management.

S.F., V.C., and A.-G.M. did the critical revision of the article.

All authors approved the final version of the article.

The research leading to these results has received funding from the Agence Nationale de Recherche sur le SIDA (ANRS), the Association de Recherche en Virologie et Dermatologie (ARVD), the European Community's Seventh framework Program (FP7/2007–2013) under the project ‘Collaborative HIV and Anti-HIV Drug Resistance Network (CHAIN)’.

Conflicts of interest

There are no conflicts of interest.