Species-specific differences in antagonism of APOBEC3 proteins by HIV-2 and SIVsmm Vif proteins

SIVsmm infecting sooty mangabeys has been transmitted to humans on at least nine independent occasions, giving rise to HIV-2 groups A to I. SIVsmm isolates replicate in human T cells and seem capable of overcoming major human restriction factors without adaptation. However, only groups A and B are responsible for the HIV-2 epidemic in Sub-Saharan Africa and it is largely unclear whether adaptive changes were associated with significant spread in humans. To address this, we examined the sensitivity of infectious molecular clones (IMCs) of five HIV-2 strains (4 group A and one AB recombinant) and representatives of five different SIVsmm lineages to inhibition by type I interferon (IFN) and various APOBEC3 proteins. We confirmed that SIVsmm strains replicate in primary human CD4+ T cells. However, SIVsmm replication was highly variable, typically lower relative to HIV-2 isolates and almost entirely prevented by type I IFN treatment. Viral propagation was generally dependent on intact vif genes, highlighting the need for efficient counteraction of APOBEC3 proteins. On average, SIVsmm strains were significantly more susceptible to inhibition by human APOBEC3D, F, G and H than HIV-2 IMCs. For example, human APOBEC3F reduced infectious virus yield of SIVsmm by ∼80% but achieved only ∼40% in the case of HIV-2. Functional and mutational analyses of human, sooty mangabey and rhesus macaque derived alleles revealed that an R128T polymorphism in APOBEC3F is important for species-specific counteraction by HIV-2 and SIVsmm Vif proteins. In addition, we found that changes of Y45H and T84S in SIVsmm Vif increase its ability to antagonize human APOBEC3F. Altogether, our results show that SIVsmm Vifs show some intrinsic activity against human ABOBEC3 proteins, but HIV-2 Vifs acquired adaptive changes to efficiently clear this barrier in the human host. AUTHOR SUMMARY SIVs infecting African monkey species do not infect humans, with one notable exception. SIVsmm from sooty mangabeys managed to cross the species barrier to humans on at least nine independent occasions. This is because SIVsmm strains seem capable of overcoming many innate defense mechanisms without adaptation and that their Vif proteins are active against human APOBEC3 proteins. Here, we show that replication of SIVsmm is highly variable in human CD4 T cells and more sensitive to interferon inhibition compared to HIV-2. While different lineages of SIVsmm were capable of counteracting human APOBEC3 proteins in a Vif-dependent manner, they were significantly more susceptible to inhibition by APOBEC3D/F/G/H compared to HIV-2. Mutational analyses revealed an R128T substitution in APOBEC3F and a T84S change in Vif are relevant for species-specific counteraction by HIV-2 and SIVsmm. Altogether, our results support that HIV-2 group A adapted to humans prior to or during epidemic spread.


INTRODUCTION
not shown) indicating that the gp41 truncation increases viral infectivity for TZM-bl reporter cells 150 as previously reported for SIVmac239 [36] and HIV-2 ST in SupT1 cells [37]. 151 The replicative capacity of the SIV constructs in primary human cells was more variable 152 compared to HIV-1 and HIV-2. The SIVmac239 and SIVsmm PG strains replicated as efficiently 153 as HIV-1 strains ( Figure 1A). Notably, SIVmac239 was passaged in human HUT-78 cells [34], 154 and SIVsmm PG was passaged in human PBMCs and CEMx174 cells after isolation from lymph  (Table S1) [6]. Like HIV-2, the SIVsmm lineage 1 (L1) strain Type I IFN induces numerous antiviral factors that are frequently counteracted by primate 165 lentiviruses in a species-specific manner [40][41][42]. Previous studies showed that SIVsmm strains 166 are capable of antagonizing several major human restriction factors [7]. It is poorly understood, 167 however, whether they counteract the antiviral effects of IFN in primary human T cells as 168 effectively as HIV-2. To address this, we infected activated human PBMCs in the presence of IFN-169 α. We found that treatment with IFN-α (500 U/ml) reduced infectious yield of HIV-1 about 2-fold 170 and resulted in moderately delayed replication kinetics ( Figure 1C). In comparison, only the HIV- 171 2 GH123 and ROD10 strains replicated efficiently in the presence of IFN-α, while infectious virus 10 To determine possible differences in the susceptibility of HIV-1, HIV-2 and SIVsmm to various 196 human APOBEC3 proteins, we measured infectious virus yield from HEK293T cells following 197 cotransfection of the proviral constructs with expression vectors for various APOBEC3s or an 198 empty control vector ( Figures 2B, S1). This system was chosen due to the reported lack of or only 199 low expression of endogenous APOBEC3 genes in this cell line. In agreement with these reports, 200 we observed no infectivity defects of vif* mutant IMCs in the absence of APOBEC3 201 overexpression ( Figure S1). All five APOBEC3 proteins analyzed (C, D, F, G and H haplotype II) 202 inhibited infectious HIV-1 production to some extent with average efficiencies ranging from 20% 203 (D) to 60% (H) ( Figure 2B). As expected, vif-defective HIV and SIV IMCs were usually more 204 susceptible to APOBEC3 inhibition than the parental WT viruses. However, NL4-3 Vif was less 205 effective in counteracting A3G than the Vif proteins of the primary HIV-1 CH058 strain. In 206 addition, the effects of intact vif genes on infectious virus yield were usually modest in the case of 207 A3C, A3D and A3F, most likely due to both relatively low antiviral activity and ineffective 208 counteraction by Vif. Surprisingly, the vif-defective HIV-2 7312A construct was largely resistant 209 to all APOBEC3 proteins. In contrast, lack of Vif function generally increased the susceptibility 210 of SIVsmm L5 and SIVmac239 especially to human A3F, A3G and A3H ( Figure 2B). Altogether, 211 the susceptibility of HIV-2 strains did not differ significantly from those of HIV-1 for APOBEC 212 3C, 3D, 3F and 3G ( Figure 2B). Unexpectedly, HIV-2 IMCs were less sensitive to inhibition by 213 A3H than HIV-1 ( Figure 2B, bottom). Most notably, SIVsmm IMCs were on average significantly 214 more susceptible to inhibition by human A3D, A3F, A3G and A3H than HIV-2 ( Figure 2B). In 215 contrast, SIVmac239 was largely resistant to all human APOBEC3 proteins investigated. 216 Altogether, these results suggest that HIV-2 strains acquired changes increasing their ability to 217 counteract human APOBEC3D, F, G and H proteins after zoonotic transmission of SIVsmm from 218 sooty mangabeys to humans. Unexpectedly, vif-defective HIV and SIV IMCs differ substantially 219 in their susceptibility to human APOBEC3 proteins. Notably, these differences were not just due 220 to differences in the absolute levels of infectious virus production ( Figure S1). Thus, although Vif 221 function clearly plays a key role in primate lentiviral susceptibility to APOBEC3 proteins, 222 alternative mechanisms also seem to be involved.

223
HIV-2 counteracts human APOBEC3 proteins more efficiently than SIVsmm 224 To determine whether the reduced susceptibility of HIV-2 to human A3D, A3F, A3G and A3H 225 relative to SIVsmm is associated with differences in the ability of the respective Vif proteins to 226 induce APOBEC3 degradation, we performed immunoblot analyses. We cotransfected HEK293T surprise, however, that the ability of HIV-1 IMCs to degrade A3D, A3F, A3G and A3H proteins 238 varied substantially and was usually less effective compared to HIV-2 IMCs ( Figure S2A). 239 Altogether, the levels of A3F and A3G expression relative to GAPDH correlated with the 240 infectious virus yields in the cell culture supernatant albeit with substantial variation ( Figure 3A, 12 3B, right panels). In agreement with previous data [43], the levels of APOBEC3 proteins in cellular 242 extract did not always match those detectable in the viral supernatants ( Figures 3A, 3B, S2A). 243 Thus, while Vif-mediated degradation and cellular expression levels of APOBEC3 proteins are a 244 major determinant of infectious virus production, not all HIV-1, HIV-2 and SIV constructs seem 245 to be equally susceptible to the inhibitory effect of these APOBEC3 proteins. In addition, our 246 results add to the evidence that Vif might also counteract APOBEC3 proteins by degradation- and HIV-2 7312A Vifs degraded A3F more efficiently than the SIVsmm Vifs ( Figure 3C).

258
Altogether, these results show that HIV-2 acquired an increased ability to degrade various human 259 APOBEC3 proteins and further suggest that especially A3F might represent a barrier to successful 260 spread of SIVsmm after zoonotic transmission.

262
To further examine the species-specific evolution of APOBEC3 antagonism in the HIV-263 2/SIVsmm/mac lineage, we compared the susceptibility of HIV-2 and SIVsmm to inhibition by 264 13 human and monkey-derived A3F, A3G and A3H orthologues. Two A3F alleles, each, were 265 available from sooty mangabeys and macaques. The two SMM A3F alleles differed in four amino 266 acid residues from one another, in 14 residues from the macaque homologue and in ~50 residues 267 from the human version ( Figure 4A). Most variations did not fall within previously proposed active 268 site residues, and the two Zinc-coordinating Cys residues as well as the catalytic Glu residues are 269 generally conserved. Functional analyses showed that HIV-1 CH077 efficiently antagonizes 270 human and sooty mangabey A3Fs in a Vif-dependent manner but is inhibited by ~50% by the  Figure 5B). The three macaque A3G orthologues showed moderate activity against HIV-1 CH077 283 and were not efficiently counteracted by its Vif protein ( Figure 5B, 5C). On average, the SIVsmm 284 strains were slightly more susceptible to the human than to the macaque orthologues, while the 285 opposite was observed for the HIV-2 IMCs ( Figure 5C). However, these differences were modest, 286 which agrees with previous data suggesting that A3G does not represent an effective barrier against zoonotic transmission of SIVsmm [29]. SIVsmm L2 was susceptible to the MAC A3G(LR) 288 orthologue but resistant to the remaining two macaque variants ( Figure 5B). This agrees with 289 previous data showing that SIVsmm is sensitive to A3G(LR) because its Vif protein contains a 290 Gly at amino acid position 17 and that a G17E substitution renders SIVmac resistant to this A3G 291 variant [30]. The SIVsmm IMCs analyzed in the present study were obtained after passage in 292 macaques (Table S1), which may explain why most of their Vif proteins contained E17 and were 293 resistant to A3G(LR). Notably, HIV-2 strains were less sensitive to inhibition by the macaque 294 A3G(LR) variant than SIVsmm L2 although their Vif proteins generally contain a Gly at position 295 17.

296
Finally, we examined the susceptibility of the IMCs to inhibition by human and sooty mangabey 297 A3H, which differ in 28 amino acid positions and the presence of the last exon ( Figure 6A). In 298 agreement with published data [45], HIV-1 counteracted human but not sooty mangabey A3H 299 ( Figure 6B). In contrast, HIV-2, SIVsmm and SIVmac were generally resistant against both human 300 and sooty mangabey A3H ( Figure 6B, 6C). Altogether, our results support that A3G and A3H do 301 not represent significant barriers to successful cross-species transmission of SIVsmm to humans.

302
In contrast, human A3F displayed substantially higher activity against SIVsmm than against HIV-303 2, suggesting specific adaptation to this antiviral factor during viral adaptation to humans. respectively. Notably, human A3F differs by a T128R change from sooty mangabey and macaque 313 A3Fs ( Figure 4A). To determine whether this loop 7 amino acid substitution contributes to the 314 species-specificity of Vif antagonism, we introduced a R128T mutation in human A3F and the 315 reverse T128R change in sooty mangabey A3F. We found that the R128T change increased the 316 antiviral activity of human A3F against HIV-1 CH077 ( Figure 7A) and HIV-2 ( Figure 7B), while 317 the reverse change in sooty mangabey A3F had no significant effect on its antiviral activity. Thus,

318
R128T renders human A3F less sensitive to antagonism by HIV-1 and HIV-2 Vif proteins. Human 319 A3F and the T128R mutant sooty mangabey A3F were equally effective against SIVsmm L2 320 ( Figure 7C). In comparison, wildtype sooty mangabey A3F showed the lowest and the R128T 321 human A3F variant the highest activity against SIVsmm L2. Thus, the T128R substitution in sooty 322 mangabey A3F seems to reduce its susceptibility to counteraction by SIVsmm Vif although the

334
To identify variations that might affect the ability of Vif to counteract APOBEC3F in a species-335 specific manner, we aligned the amino acid sequences of the HIV-2, SIVsmm and SIVmac IMCs 336 analyzed in the present study. In order to focus on those that would be representative of human-  SIVsmm less susceptible to human A3F ( Figure 8D). The species-specificity of this effect came 360 as surprise since 45H dominates in both SIVsmm as well as HIV-2 Vif proteins ( Figure 8B).

361
To further determine the effect of the T85S and Y45H changes on the ability of SIVsmm Vif 362 to antagonize A3F, we performed Western blot and infectious virus production assays ( Figure 9). 363 We found that both changes increased infectious virus yield in the presence of human A3F from 364 ~60% to 80% but did not further enhance the already effective counteraction of sooty mangabey 365 A3F ( Figure 9A). Western blot analyses indicated that the T85S and Y45H changes do not 366 markedly affect the ability of SIVsmm Vif to induce SMM A3F degradation. For comparison, we 367 used the HIV-2 7312 construct and confirmed that it is hardly susceptible to A3F inhibition, even 368 in the absence of Vif ( Figure 9A). Thus, HIV-2 may also have evolved Vif-independent 369 mechanisms to avoid A3F restriction. the vif-defective derivatives did not yield infectious virus ( Figure 10A). All five mutant SIVsmm 377 constructs exhibited similar levels of virus production as the wildtype virus. However, the T84S 378 substitution in Vif generally resulted in accelerated replication kinetics ( Figure 10B) and slightly higher total virus production ( Figure 10C). In comparison, the reverse S84T change did not alter 380 the replicative fitness of HIV-2 7312 in human CD4+ T cells ( Figure 10A). Unexpectedly, the 381 H28Y, H28Y/Y45H and H28Y/N32R/Y45H changes that render SIVsmm Vif more similar to 382 HIV-2 Vifs all moderately but significantly reduced infectious virus production in SIVsmm-383 infected human PBMCs, while the individual Y45H change had no significant effect ( Figure 10C).

384
Altogether, these results show that the consistent amino acid differences between SIVsmm and 385 HIV-2 Vif proteins at positions 28 and 32 have a slightly negative effect on viral replication fitness 386 but the T84S change significantly accelerated replication kinetics in primary human PBMCs.

387
It has been shown that primary SIVsmm strains replicate in primary human cells [6] and are 389 capable of counteracting some major human restriction factors, including APOBEC3 proteins  Based on previous data using uncloned viral strains [51], we expected HIV-1 to replicate to 404 higher titers in human PBMC than HIV-2 ( Figure 1). However, we did not expect that HIV-1 IMCs 405 were on average less efficient in degrading human APOBEC3 proteins than HIV-2 ( Figure S2A).

406
Only overexpression of A3H reduced the infectious virus yield of HIV-1 more efficiently than that 407 of HIV-2 ( Figure 2B). Altogether, these results agree with previous findings that HIV-1 Vif SIVsmm, the data show that human APOBEC3F was better at resisting Vif-mediated degradation 463 than APOBEC3G so that more was left in cells and thus in virions to exert the restriction activity.  In the present study, we confirmed that SIVsmm is preadapted for growth in human cells and 469 capable of counteracting human APOBEC3 proteins. We also show, however, the SIVsmm strains 470 that were transmitted to humans were initially most likely highly sensitive to IFN and apparently 471 acquired adaptive changes for epidemic spread in humans, including increased capability to 472 counteract APOBEC3D, F, G and H. We found that a T84S variation in Vif is relevant for species-473 specific counteraction by HIV-2 and SIVsmm but altogether the determinants seem to be complex.

474
One open question is why HIV-2 A is still less fit for replication and spread in humans (compared 475 to HIV-1) although it crossed the species-barrier about a century ago and its Vif proteins seem to 476 be highly effective at degrading human APOBEC3 proteins.                 [16,17]