APOBEC3 proteins, in particular APOBEC3G/F, are important innate host factors that contribute to protection from HIV-1 infection by inducing high levels of guanine to adenine nucleotide substitutions (termed hypermutation) during HIV-1 viral replication. These nucleotide substitutions occur at different rates and locations across the HIV-1 genome and are thought to be particularly more frequent in the pol region. The virus has evolved ways to counteract these host factors by inducing degradation of APOBEC3G/F proteins through protein interactions with HIV-1 Vif. The aim of this thesis is to characterize and investigate the role of APOBEC3G/F-mediated hypermutation in the HIV-1 genome. We identified a subset of women from the Pumwani Commercial Sex Worker (CSW) cohort with significantly higher rates of hypermutated proviruses in pol. This degree of hypermutation correlated to less severe HIV disease progression as measured by CD4+ T cell count. This was in agreement with previous studies that evidence of APOBEC-mediated hypermutation correlate with reduced disease progression, confirming APOBEC3G/F proteins role in HIV-1 disease. Furthermore, we investigated the in vitro and ex vivo interaction between HIV-1 Vif and APOBEC3G from subjects infected with hypermutated and non-hypermutated proviruses. In vitro studies indicated that HIV-1 Vif protein expression in subjects with hypermutated proviruses were quite divergent and levels of APOBEC3G also differed between subjects. Ex vivo studies in subjects with hypermutated proviruses indicated that endogenous APOBEC3G expression was greater than in subjects with hypermutated proviruses. Both studies suggest that host and viral factors such as APOBEC3G and HIV-1 Vif are playing an influential role in HIV-1 pathogenesis. Further investigations into these interactions may lead to novel strategies into the development of therapeutic drugs for the fight against HIV-1.

APOBEC3 proteins, in particular APOBEC3G/F, are important innate host factors that contribute to protection from HIV-1 infection by inducing high levels of guanine to adenine nucleotide substitutions (termed hypermutation) during HIV-1 viral replication. These nucleotide substitutions occur at different rates and locations across the HIV-1 genome and are thought to be particularly more frequent in the pol region. The virus has evolved ways to counteract these host factors by inducing degradation of APOBEC3G/F proteins through protein interactions with HIV-1 Vif. The aim of this thesis is to characterize and investigate the role of APOBEC3G/F-mediated hypermutation in the HIV-1 genome. We identified a subset of women from the Pumwani Commercial Sex Worker (CSW) cohort with significantly higher rates of hypermutated proviruses in pol. This degree of hypermutation correlated to less severe HIV disease progression as measured by CD4+ T cell count. This was in agreement with previous studies that evidence of APOBEC-mediated hypermutation correlate with reduced disease progression, confirming APOBEC3G/F proteins role in HIV-1 disease. Furthermore, we investigated the in vitro and ex vivo interaction between HIV-1 Vif and APOBEC3G from subjects infected with hypermutated and non-hypermutated proviruses. In vitro studies indicated that HIV-1 Vif protein expression in subjects with hypermutated proviruses were quite divergent and levels of APOBEC3G also differed between subjects. Ex vivo studies in subjects with hypermutated proviruses indicated that endogenous APOBEC3G expression was greater than in subjects with hypermutated proviruses. Both studies suggest that host and viral factors such as APOBEC3G and HIV-1 Vif are playing an influential role in HIV-1 pathogenesis. Further investigations into these interactions may lead to novel strategies into the development of therapeutic drugs for the fight against HIV-1.

ABSTRACT APOBEC3G (Apo3G) is a host cell restriction factor of viruses that produce a singlestranded (ss) DNA replication intermediate (Sheehy et al., 2002; Suspene et al., 2004). Apo3G is studied primarily for its ability to restrict propagation of the retrovirus, HIV. In cell culture, Apo3G can only inhibit HIV if it lacks its virion infectivity factor (Vif). The host-pathogen interface between Apo3G and HIV has become a new target of study for the development of novel HIV therapeutics (Prochnow et al., 2009; Sheehy et al., 2003). Apo3G induces mutagenesis of the HIV proviral DNA (Mangeat et al., 2003; Zhang et al., 2003). Apo3G has the ability to induce transition mutations, i.e. cytosine to thymine, through deamination of cytosine to form uracil. Deamination activity induces numerous mutations that causes gene inactivation of the HIV provirus thus restricting the HIV lifecycle. Apo3G attenuates HIV virion infectivity in the absence of the virion infectivity factor (Vif) by inducing genome mutations through deamination of cytosine to uracil in HIV minus strand DNA. Independent from deaminase activity, Apo3G may also interfere with HIV reverse transcription by preventing full length cDNA from forming (Iwatani et al., 2007), nucleocapsid (NC) mediated strand annealing (Guo et al., 2007; Guo et al., 2009; Li et al., 2007), and RNaseH activity of the reverse transcriptase (Li et al., 2007). Whether Apo3G is able to restrict HIV by a deamination-independent mode remains controversial. In particular, the existence of the deamination independent mode was challenged since the Apo3G deamination null mutant E259Q was shown to have limited or no ability to inhibit HIV-1 replication (Schumacher et al., 2008). This research assesses the ability of Apo3G to inhibit reverse transcription of HIV genomic RNA. It is hypothesised that based on the ability of Apo3G to bind and oligomerize on single stranded nucleic acids (Chelico et al., 2008), and its high affinity for RNA (Chelico et al., 2010), that Apo3G can inhibit RT mediated primer extension as well as nucleocapsid mediated strand annealing. Additionally, it is hypothesized that Apo3G cannot inhibit RT RNAseH activity, as Apo3G has been shown to have a low affinity for DNA/RNA hybrids (Iwatani et al., 2006). We will test these hypotheses by using in vitro assays that mimic in vivo reverse transcription events. Here we have shown that Apo3G is able to decrease the efficiency with which HIV-1 reverse transcriptase synthesizes DNA from an RNA primer annealed to an RNA template. Apo3G had a minimal affect on primer initiation and primarily inhibits primer elongation. Using iii the monomeric mutant, F126A/W127A, we show that the deamination independent mode of inhibiting reverse transcriptase is impaired without oligomerization on template RNA. We also provide evidence that the Apo3G mutant E259Q should not be considered a deamination null proxy for native Apo3G since it exhibits decrease in RNA binding affinity compared to the native form. We did not find that Apo3G inhibited HIV NC-mediated strand annealing activity or RNaseH activity of HIV-1 reverse transcriptase. The data suggest a two-tiered mechanism for inhibition of reverse transcriptase-mediated DNA synthesis that is dependent upon 1) the ability of Apo3G to oligomerize on RNA substrates and 2) bind RNA with high affinity. Ascribing a mechanism to the deamination independent mode of HIV-1 restriction by Apo3G suggests that the enzyme may use this mechanism in vivo to delay completion of proviral DNA synthesis which, may negatively impact the HIV-1 lifecycle.

Currently, the HIV pandemic remains a major global health challenge. In order to effectively control and cure HIV-1 infection, it is necessary to perform greater research on host-HIV interactions and develop novel preventive and therapeutic approaches. The human cytidine deaminase APOBEC3G (A3G) is the first identified host restriction factor, which can serve as an initial line of defense against HIV-1 by inducing lethal mutations on proviral DNA and disrupting viral reverse transcription and integration. In order to better understand the action of A3G on HIV-1 replication, my study was focused on characterizing the interplay between A3G and HIV-1 reverse transcriptase (RT). The results indicated that A3G directly bound to RT, which contributed to A3G-mediated inhibition of viral reverse transcription. Overexpression of the RT-binding polypeptide A3G65-132 was able to disrupt wild-type A3G and RT interaction, consequently attenuating the anti-HIV effect of A3G on HIV replication. While the potent antiviral activities of A3G make it an attractive candidate for gene therapy, the actions of A3G can be counteracted by HIV-1 Vif during wild-type HIV infection. In order to overcome Vif-mediated blockage and maximize the antiviral activity of A3G, this protein was fused with a virus-targeting polypeptide (R88) derived from HIV-1 Vpr, and various mutations were then introduced into R88-A3G fusion protein. Results showed that Vif binding mutants R88-A3GD128K and R88-A3GP129A exhibited very potent antiviral activity, and blocked HIV-1 replication in a CD4+ T lymphocyte cell line as well as human primary cells. In an attempt to further determine their potential against drug resistant viruses and viruses produced from latently infected cells, R88-A3GD128K was chosen and delivered by an inducible lentiviral vector system. Expression of R88-A3GD128K in actively and latently HIV-1 infected cells was shown to be able to inhibit the replication of both drug sensitive and resistant strains of HIV-1. In conclusion, this thesis has demonstrated one of the mechanisms that how A3G can disrupt HIV-1 reverse transcription. Meanwhile, an A3G-based anti-HIV-1 strategy has been developed, which provides a proof-of-principle for a new gene therapy approach against this deadly virus.

This thesis addresses two important topics in HIV-1 medicine; (i) the clinical relevance of pre-treatment G-A hypermutation and the contribution of host and viral genetics to its development and; (ii) the influence of genetic variation in host enzymes responsible for antiretroviral drug metabolism on response to therapy. These themes are outlined below. HIV-1 Hypermutation At present, limited data exists regarding the relative roles of host encoded cytidine deaminases APOBEC3G and APOBEC3F in promoting G-A hypermutation of HIV-1 proviral DNA in vivo, nor the clinical relevance of hypermutation or the influence of genetic diversity of the APOBEC3G locus and of the viral encoded vif protein that counteracts the action of APOBEC3G. The analyses contained within this thesis demonstrate that within the WA HIV cohort, clinically relevant hypermutation is restricted to a minority of individuals and is mediated predominantly by APOBEC3G. In this study, the presence of HIV-1 hypermutation had a substantially greater effect on plasma viremia than other known host antiviral factors such as CCR5D32 or specific HLA-B alleles. Furthermore, the considerable genetic diversity of the vif gene is likely to make a greater contribution to the development of hypermutation than the limited genetic diversity of the APOBEC3G gene in Caucasians. These data indicate that G-A hypermutation is a clinically relevant phenomenon and may provide a fresh perspective to the area of HIV/AIDS therapies. Genetic Determinant of HIV-1 Treatment Response Thymidine kinase 2 (TK2) and thymidylate kinase (dTMPK) are rate limiting enzymes for the metabolism of the antiretrovirals d4T and AZT, respectively, and are thus central to the antiviral efficacy and toxicity of these agents. However, the genetic diversity of TK2 and dTMPK and their influence on toxicities associated with their use is largely unknown. The results discussed in this thesis indicate that in contrast to the highly conserved TK2 locus, the dTMPK locus of Caucasian individuals, including regulatory regions potentially influencing transcription and translation, is considerably polymorphic and organised into five common haplotypes. The results regarding the contribution of dTMPK genetic variation to toxicities associated with AZT therapy are encouraging. A common dTMPK haplotype had significant, albeit modest, effect on haematological parameters (haemoglobin and mean corpuscular volume) in HIV-infected patients, although no AZT-specific treatment effect was observed in this relatively haematologically stable cohort. In addition, another common dTMPK haplotype provided significant protection against AZT-induced adipocyte mtDNA depletion in a pilot study of AZT- and d4T-treated individuals. The dTMPK haplotypes characterised in this thesis should facilitate further studies regarding dTMPK genetic variation in HIV-1 infection and response to treatment, which are warranted from the clinical results presented herein.

Vif (Viral Infectivity Factor) est une protéine auxiliaire qui augmente le «fitness» viral dans l’hôte infecté. Vif est essentielle à la formation de particules virales infectieuses dans les cellules dites «non-permissives», alors que des virus ΔVif se répliquent efficacement dans des lignées cellulaires T dites « permissives ». Les cellules non-permissives expriment les facteurs de restriction APOBEC3G (APOlipoprotein B mRNA-Editing enzyme Catalytic polypeptide 3G ou A3G) et A3F, deux cytidine désaminases dont l’action hypermutatrice est létale pour le virus. Vif réduit de façon considérable le taux d’expression des protéines A3G/3F par deux mécanismes principaux : (1) en recrutant une E3 ubiquitine ligase, Vif induit la dégradation d’A3G par le protéasome et (2) en se fixant sur l’ARNm, Vif régulant négativement la traduction d’A3G par un mécanisme dépendent de la région 5'-UTR. L'objectif de mon projet a été de déterminer le rôle et le mécanisme de la régulation traductionnelle du facteur de restriction A3G par la protéine Vif du VIH-1 ex vivo. En parallèle, nous nous sommes intéressés à déterminer les domaines de la protéine Vif impliqués dans l’activité chaperonne d'ARN. Par l’analyse des « westerns blots » issus de co-transfections de différents vecteurs d’expression d’A3G en présence ou absence de Vif et d’un dominant négative de la Cullin 5, nous avons démontré ex vivo que Vif requiert les tiges boucles 2 et 3 de la région 5’UTR (de façon simultanée) pour inhiber la traduction d’A3G. La régulation traductionnelle d’A3G par Vif cause 50% de la réduction total d’A3G en présence de Vif. Ensuite, nous avons observé qu’une une petite uORF (Upstream Open Reading Frame), contenue entre ces deux tiges-boucles est nécessaire pour l’inhibition d’A3G par Vif. Les uORFs, présents dans 50 % des gènes eucaryotes, interviennent principalement dans des mécanismes de régulation traductionnelle. Ainsi, nous avons observé in vitro et ex vivo que l’uORF régule négativement la traduction de l’ORF majeure d’A3G. Par l’analyse de l’expression de différents mutants de l’uORF, nous avons démontré ex vivo que 60% des complexes d’initiation de la traduction synthétisent A3G par « leaky scanning » et 40 % sont recrutés dans la traduction de l’uORF, en inhibant ainsi l’expression d’A3G. A partir de ces résultats nous proposons un modèle de l’inhibition d’A3G par Vif. Dans ce modèle, Vif pourrais inhiber l’étape de terminaison de la traduction de l’uORF en causant un « stalling » des ribosomes en empêchant ainsi à des nouveaux complexes d’initiation d’attendre l’ORF majeur. Nous avons aussi déterminé l’impact de la régulation traductionnelle d’A3G par Vif sur l’incorporation d’A3G dans les particules virales et sur l’infectivité virale. Nous avons alors observé que l’inhibition traductionnel d’A3G par Vif réduit l’incorporation d’A3G dans les particules virales avec un profil de diminution d’A3G similaire à celui observé dans les cellules. En utilisant des cellules indicatrices TZM-bl, nous avons ensuite observé que l’inhibition traductionnelle d’A3G par Vif augmente l’infectivité virale de 50%. Finalement, nous avons déterminé les domaines de la protéine Vif impliqués dans l’activité chaperonnes d'ARN. En utilisant des essaies de dimérisation des fragments d’ARN du VIH-1, nous avons pu mettre en évidence que le domaine C-terminal de la protéine Vif était impliqué dans cette activité. Ces résultats nous ont permis de mieux comprendre ce phénomène de restriction cellulaire et pourraient être importants dans le développement de nouvelles stratégies d’inhibition de la réplication virale qui ciblerait spécifiquement l’interaction de Vif avec l’ARNm d’A3G. / The HIV-1 viral infectivity factor (Vif) is a small basic protein essential for viral fitness and pathogenicity. Vif allows productive infection of non-permissive cells (including most natural HIV-1 targets) by counteracting cellular cytosine deaminases APOBEC3G (A3G) and A3F by different mechanisms and thus preventing its incorporation into viral particles. The Vif-induced degradation of A3G through the proteasome pathway has been extensively studied, but little is known about the translational repression of A3G mRNA by Vif. After cellular co-transfection of A3G mRNA constructs mutated in their untranslated regions (UTRs) in presence or absence of Vif, and in conditions where the proteasome-induced degradation of A3G was inhibited, we show that the 5’-UTR of A3G mRNA is crucial for the translational inhibition by Vif. The core binding factor, CBF-, required to stabilize the Vif/A3G complex is dispensable for this specific repression. According to our previous secondary structural model of the 5’-UTR, the two distal stem-loop structures are sufficient for a complete translational inhibition of A3G. We show that residue K26 of Vif is critical for A3G neutralization, both for its proteasome-induced degradation and translation inhibition of itsmRNA. Interestingly, we observe a strict correlation between the cellular reduction of A3G through translation inhibition and the quantity of A3G incorporated into viral particles. Both mechanisms account for about 50% decrease of A3G in cell. Thus, we showed for the first time that A3G mRNA translational inhibition by Vif is a 5’-UTR mRNA-dependent mechanism, and that any of these two mechanisms, degradation or translation, is sufficient to restore viral infectivity. Regulating the translation of A3G could thus be considered as a new target to restore a functional expression of A3G and viral restriction.

Le VIH-1, via sa protéine Vif, contrecarre l’activité des facteurs de restriction A3G et A3F de plusieurs manières dont l’inhibition traductionnelle. Nous avons démontré que cette répression traductionnelle d’A3G par Vif est spécifique de la région 5’UTR. De plus, une séquence uORF contenue dans cette région est essentielle pour la régulation de la traduction ainsi que pour la répression par Vif. Nous avons montré qu’A3G et A3F sont traduits par leaky-scanning et ré-initiation et que la distance entre l’uORF et l’ORF principal est importante pour l’inhibition traductionnelle d’A3G et d’A3F. Ensuite, nous avons montré que ce mécanisme est très conservé à travers différents sous-types de Vif et que les acides aminés de Vif en position 39, 48 et 127 sont impliqués dans cette répression. Enfin, nous avons observé que l’interaction entre Vif et A3G est nécessaire pour la régulation traductionnelle. Des expériences de transfections des différents mutants d’A3G et d’A3F dans un système infectieux tel que le clone pNL4.3 sont envisagées. Celles-ci permettront de voir l’effet de l’uORF sur l’infectivité virale mais aussi sur l’assemblage, la maturation et la libération des nouvelles particules virales. / The HIV-1, through its Vif protein, counteracts the restriction factors A3G and A3F activity in several ways including translational inhibition. We demonstrated that this translational repression of A3G by Vif is specific of the 5’UTR. Moreover, an uORF sequence contained in this region is essential for both the translational regulation and the repression by Vif. We showed that A3G and A3F are translated by leaky-scanning and re-initiation mechanisms and that the distance between the two ORFs is important for the translational inhibition. Then, we showed that this mechanism is conserved among several Vif subtypes and that the Vif amino acids 39, 48 and 127 are implicated in this repression. Finally, we observed that the Vif-A3G interaction is necessary for the translational inhibition. A3G and A3F mutants transfections experiments into an infectious system like the pNL4.3 clone are considered. It will permit to observe the uORF effect on the viral infectivity but also on the assembly, the maturation and the release of the viral particles.

The APOBEC3 (Apo3) family of proteins are single stranded (ss) DNA cytosine deaminases (C → U). They are grouped into two different structural groups, the single catalytic domain Apo3 enzymes (Apo3A, Apo3C, and Apo3H) and the double catalytic domain Apo3 enzymes (Apo3B, Apo3D, Apo3F, and Apo3G). Apo3G has been implicated in protection from HIV proliferation by becoming encapsidated into budding HIV virions and subsequently mutationally inactivating the synthesized provirus. This largely occurs in the absence of HIV viral infectivity factor (Vif) which mediates the ubiquitination and degradation of Apo3G. Apo3G is a processive enzyme, able to catalyze numerous deaminations in a 5'CCC motif in a single interaction with a substrate. There is a paucity of biochemical data on other Apo3 family members. We performed basic biochemical assays that determined the relative specific activities, processivity, cytosine motif preferences, and binding affinities for DNA, of Apo3A and Apo3H using synthetic DNA substrates in deamination assays. We found Apo3A to be an enzyme with low processivity and Apo3H to be a highly processive enzyme; both of which deaminate a 5'TC motif. Using a reconstituted HIV replication assay we assessed if processivity is needed for efficient restriction of HIV. We were able to demonstrate that each, Apo3G, Apo3A, and Apo3H were able to catalyze deaminations during in vitro reverse transcription. The mutation profile of both Apo3A and Apo3H showed that the 5'TC motif preference was less effective compared to Apo3G in triggering missense and nonsense mutations in the HIV protease active site coding sequence. Nuclear DNA can become deaminated by the related Apo3 family member activation-induced deaminase (AID), when it is present in the nucleus of activated B cells. Apo3A and Apo3H are located in the nucleus but the extent of the damage they cause has only recently been investigated. Here we used an in vitro transcription assay to determine the efficiency of Apo3A and Apo3H deamination during transcription and found that, like AID, they are highly capable of causing deaminations during transcription. Taken together, the results presented here demonstrate that processivity is not necessary for an Apo3 enzyme to catalyze deaminations during HIV reverse transcription and that Apo3A and Apo3H can catalyze deaminations during DNA transcription that could damage host genomic DNA. These results imply a potential cost for maintaining nuclear deaminases.

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22372号 / 医博第4613号 / 新制||医||1043(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 小柳 義夫, 教授 竹内 理, 教授 朝長 啓造 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(エネルギー科学) / 甲第19905号 / エネ博第337号 / 新制||エネ||67(附属図書館) / 32982 / 京都大学大学院エネルギー科学研究科エネルギー基礎科学専攻 / (主査)教授 片平 正人, 教授 森井 孝, 教授 木下 正弘 / 学位規則第4条第1項該当 / Doctor of Energy Science / Kyoto University / DGAM