HIV-1 Subtype CRF01_AE and B differ in utilization of low levels of CCR5, Maraviroc susceptibility and Potential N-glycosylation sites

Abstract

HIV subtypes not only predominate in different geographical regions but also differ in key phenotypic characteristics. To determine if genotypic and/or phenotypic differences in the Envelope (Env) glycoprotein can explain subtype related differences, we cloned 37 full length Envs from Subtype B and AE HIV infected individuals from Singapore. Our data demonstrates that CRF01_AE Envs have lower Potential N Glycosylation Sites and higher risk of X4 development. Phenotypically, CRF01_AE were less infectious than subtype B Envs in cells expressing low levels of CCR5. Moreover, the Maraviroc IC₅₀ was higher for subtype B Envs and correlated with infectivity in low CCR5 expressing cells as well as PNGS. Specifically, the glycosylation site N301 in the V3 loop was seen less frequently in AE subtype and CXCR4 topic viruses. CRF01_AE differs from B subtype in terms of CCR5 usage and Maraviroc susceptibility which may have implications for HIV pathogenesis and virus evolution.

INTRODUCTION

The genetic diversity associated with HIV has been a major hurdle in developing an efficacious vaccine and emergence of drug resistance. Since its introduction into the human population, HIV-1 group M has been constantly evolving, that has led to designation of genetically related viruses into several clades/subtypes (Geretti, 2006; Worobey et al., 2008). Subtype B is the most common type found in North America and other developed countries including Western Europe, Australia and UK (Castro-Nallar et al., 2012). On the other hand, the non-B clades (mostly clade C) are known to predominate in developing countries with limited healthcare access like South Africa, Asia and the Middle East (UNAIDS, 2006). Other clades like A, D, J–K are mostly prevalent in sub-Saharan Africa (Castro-Nallar et al., 2012; Hemelaar, 2012; Hemelaar et al., 2006). More recently, recombination events between subtypes have given rise to Circulating Recombinant Forms (CRFs) of which up to 72 have been identified so far (Yan et al., 2015). The CRFs originated in Africa (CRF01_AE and CRF02_AG) but have been spreading rapidly including Southeast Asia, Russia, Brazil, Spain and other parts of the world (Hemelaar et al., 2011).

While different HIV subtypes circulate in different parts of the world, differences in subtypes with respect to disease progression have also been noted. Studies have shown that HIV subtypes vary in terms of plasma viral load (Kivela et al., 2005); rate of CD4+ T cell decline (Easterbrook et al., 2010; Ng et al., 2011a), prevalence of CXCR4 tropism (Ng et al., 2013), enhanced biological fitness including replicative advantage (Lau et al., 2010; Njai et al., 2006), increased virulence and pathogenicity (Hemelaar, 2013), and accumulation of anti-retroviral resistance mutations (Tang and Shafer, 2012). Moreover, studies have also documented differences in CD4 recovery after initiation of anti-retroviral therapy in patients infected with AE versus B subtype (Chow et al., 2015; Oyomopito et al., 2013). Interestingly, the mechanism behind the above phenomenon and why certain recombinant forms acquire the above fitness advantages, especially in certain populations, remains unclear.

The HIV Envelope (Env) glycoprotein is the most dynamic viral protein and is known to constantly evolve both at the population level and within a patient (Stalmeijer et al., 2004). The best documented phenomenon associated with HIV Env evolution is co-receptor switch from CCR5 usage during early infection to CXCR4 usage during late stages of the disease and is associated with a rapid CD4 decline (Jekle et al., 2003; Spijkerman et al., 1998). Besides co-receptor switch, the glycosylation pattern of the HIV Env also varies throughout the course of disease and is known to influence immune evasion (McMichael et al., 2010; Rong et al., 2009; Wei et al., 2003). In this regard, viruses isolated during early infection have fewer glycosylation sites along with shorter variable loops (Sagar et al., 2006). This confers the virus enhanced replicative potential with the glycosylation pattern increasing during chronic stages as the host mounts an effective antibody mediated humoral response (Albert et al., 1990; Wei et al., 2003). Interestingly, this pattern of glycosylation varies considerably between different subtypes (Utachee et al., 2010) and acquisition of N glycosylation sites has been associated with enhanced replication of certain viral quasi species compared to the inoculating virus (Derdeyn et al., 2004; Edwards et al., 2006).

The role of CCR5 in HIV disease is more complex than merely acting as a co-receptor for viral entry. It is clear that CCR5 levels in the host are regulated by CCR5 gene and promoter polymorphisms (Chalmet et al., 2008; Ometto et al., 2001; Samson et al., 1996) and affect HIV disease progression rate (Edo-Matas et al., 2011; Ometto et al., 2001). This is evident by the slow disease progression rate in CCR5Δ32 heterozygous individuals who have lower CCR5 expression on the cell surface (Liu et al., 1996). We recently demonstrated that CCR5 cell surface levels affect the process of Env mediated bystander apoptosis (Joshi et al., 2011) along with evolution of the viral Env repertoire and Maraviroc (MVC) susceptibility profile (Garg et al., 2016). Hence, both CCR5 binding by the virus and CCR5 levels in the host play a significant role in HIV biology.

To better understand the subtype based differences in disease progression we cloned 37 full length Env glycoproteins from CRF01_AE (AE) (27 Envs) and B subtype (10 Envs) infected HIV+ patients from Singapore. The differential rate of progression in this population correlates with viral subtype, with AE showing faster progression than B subtype (Ng et al., 2011a). We characterized the Envs for infectivity in cells expressing different levels of CCR5, pattern of potential N-linked glycosylation sites (PNGS) and MVC resistance profile along with other correlative analyses. Our data demonstrates that subtype B Envs have more PNGS and higher infectivity in cells expressing low levels of CCR5 when compared to AE Envs. Moreover, the MVC IC₅₀ was higher for subtype B Envs and correlated with infectivity in low CCR5 expressing cells and PNGS. Analysis of PNGS between subtypes based on sequences from the Los Alamos database showed that PNGS in gp160, gp120, gp41 and V3 loop for subtype AE are significantly lower than other subtypes. Based on our study an interesting pattern of differences between AE and B is observed; AE subtype Envs are characterized by reduced infectivity in low CCR5 expressing cells and better inhibition by MVC. The implications of these differences to disease progression remains to be seen.

RESULTS

Env clones from selected AE and B subtype samples represent the larger HIV patient cohort

We obtained a total of 89 (65=Subtype AE, 24=Subtype B) archival plasma samples from a cross-sectional cohort of HIV infected patients from Singapore (Ng et al., 2013; Ng et al., 2011b). From the available samples we were able to amplify and clone 37 (27=Subtype AE, 10=Subtype B) full length functional Envs. The selected AE and B subtype samples from which full length Envs were cloned were fairly representative of the larger cohort (Table 1). Furthermore, as shown in Figure 1A, samples from subtype AE HIV infected individuals showed significantly lower CD4 counts compared to the B subtype in both the entire cohort (p=0.0213) and in the samples from which Envs were cloned (p=0.0270). However, with regards to plasma viremia, subtype B samples showed higher log viremia compared to AE subtype in the cloned Envs group (p=0.0299) although this was not statistically significant for the entire cohort (Figure 1B). This data demonstrates that while AE subtype is associated with lower CD4 counts, this phenomenon is likely independent of basal viremia.

Table 1.

Baseline Demographics and Clinical parameters, by subtype.

Patient Characteristics	Subtype AE All samples (n=65)*	Subtype B All samples (n=24)	P-value†	Subtype AE Cloned Envs (n=26)*	Subtype B Cloned Envs (n=10)	P-value†
Demographic
Age (Mean ±SEM)	39.06 ± 1.60	32.85 ± 1.88	0.0327	37.12 ± 2.73	32.71 ± 2.99	0.3591
Sex, number (%)
Female	9 (14.1)	2 (8.3)	0.47	0 (0)	0 (0)	1.00
Male	55 (85.9)	22 (91.7)		25 (100)	10 (100)
Ethnicity (%)
Chinese	54 (84.37)	15 (62.5)	0.040	24 (96)	6 (60)	0.017
Non-Chinese	10 (15.62)	9 (37.5)		1 (4)	4 (40)
Transmission risk (%)
Heterosexual	38 (59.37)	8 (33.33)	0.020	14 (56)	2 (20)	0.007
MSM	24 (37.5)	12 (50)		11 (44)	5 (50)
Others	2 (3.12)	4 (16.66)		0	3 (30)
Laboratory values
CD4 counts (Mean ± SEM)	287.3 ± 28.01	403.6 ± 30.84	0.0213	226.8 ± 36.30	383.3 ± 56.52	0.0270
Basal log Viremia (Mean ± SEM)	4.78 ± 0.104	5.028 ± 0.171	0.2199	4.77 ± 0.16	5.49 ± 0.27	0.0299

Abbreviations: MSM, men who have sex with men. SEM, standard error mean

SEM=standard error mean

^†From a Fisher’s exact test for the sex, ethnicity and transmission risk variable, otherwise from two-sample t-tests.

^*Data for one sample was not available.

Figure 1. Fair representation of the larger cohort by the selected AE and B subtype samples from which full length Env clones were available.

(A) CD4 counts from the total AE (N=65) and B (N=24) subtype samples available for the study versus the 37 samples from which full length Envs were cloned (subtype AE=27 and B=10). (B) Log viremia from the total AE (N=65) and B (N=24) subtype samples available for the study versus the 37 samples from which full length Envs were cloned (subtype AE=27 and B=10).

Envelope cloning and genotypic and phenotypic analysis of the cloned envelopes

We cloned 37 full length Envs directly from plasma samples derived from HIV infected patients from Singapore (Table 2). We determined the relatedness/uniqueness of the individual cloned Envs via maximum likelihood phylogenetic analysis constructed using the GTR model with a gamma distributed evolutionary rate difference among sites. As shown in Figure 2, the clones comprised of full length Envs that were phylogenetically distinct from each other. Interestingly, two Env clones obtained from patient HM153 were phenotypically distinct (one dual tropic and one R5 tropic), yet as expected, were closely related to each other in the phylogenetic tree. These data suggest that the cloned sequences correspond to authentic Envs, and are not likely to be products of PCR recombination. The Env length varied from 844 to 877 amino acids and was not significantly different between subtype B and AE (Figure 3A). Analysis of the V3 loop sequence revealed that 10 Envs belonged to subtype B while 27 were subtype AE (Table 1, 2 and Supplementary Table 1). Based on the Geno2pheno analysis and placing the False Positive Rate (FPR) cut off value as 2.5, two Envs were predicted to be X4R5 tropic and 35 as R5 tropic. A lower FPR value is predictive of CXCR4 co-receptor usage. However, tropism assay in U87 cells demonstrated that 34 Envs were R5 tropic and 3 (2 subtype AE and 1 subtype B) were X4R5 (Table 2), emphasizing the limitation of algorithms in predicting HIV-1 tropism.

Table 2.

Characteristics of the cloned Envelopes used in the study (N=37).

Env	Subtype	U87 Tropism	Env length	FPR	X4 Risk	PNGS	MVC IC50	Viral load	CD4 count	TriMab IC50	TriMab IC90	Accession #
HM130	AE	R5	859	13.2	35.8	31	0.916	235043	242	0.3922	13.8672	KY213715
HM086	AE	R5	845	29.8	40	27	1.451	678942	20	0.3953	8.5938	KY213716
HM097	AE	R5	848	11.7	35.1	29	2.44	59323	20	3.825	>25	KY213717
HM123D	AE	R5	858	8.2	42.2	26	0.2351	98044	20	0.784	8.007	KY213718
HM073	AE	R5	850	8.9	28.8	28	0.861	213113	20	>25	>25	KY213719
FREE019	AE	R5	877	6.3	41.4	29	2.021	165540	22	0.3986	7.185	KY213720
HM153I*	AE	R5	868	26.9	33.3	34	0.612	13049	181	0.6133	4.7684	KY213721
HM033	AE	R5	865	3.2	32.8	28	1.524	95240	42	1.1719	14.0625	KY213722
HM051	AE	R5	853	27.3	43.9	32	1.08	3567880	75	2.741	>25	KY213723
HM089	AE	R5	862	71.7	36.9	30	1.835	74900	88	0.5037	9.7656	KY213724
HM043	AE	R5	850	63.2	27.9	32	1.7524	61745	187	0.3327	1.0742	KY213725
HM067	AE	R5	857	67.6	29.6	31	1.17	5722	191	0.498	5.1758	KY213726
HM040	AE	R5	854	73.1	30.4	29	0.7429	10527	200	0.5975	6.7383	KY213727
HM044	AE	R5	859	66.9	38.2	30	4.77	4650	203	0.4418	2.7849	KY213728
HM132F	AE	R5	865	2.7	30.2	30	4.3509	7635	285	0.3991	4.8828	KY213729
HM018	AE	R5	854	88.8	33.1	29	0.548	5674	338	0.5927	5.9331	KY213730
HM026	AE	R5	856	73.1	30.4	30	0.82	4240	348	1.2747	20.935	KY213731
HM108	AE	R5	847	22	30.8	31	1.33	197301	371	0.3906	5.7617	KY213732
HM032	AE	R5	860	4.7	36.9	30	1.449	78616	378	3.125	>25	KY213733
HM149	AE	R5	867	54.5	29.8	31	5.1774	768454	384	0.0977	1.1384	KY213734
HM019	AE	R5	858	19.4	30.5	29	2.584	11879	400	0.6115	11.4691	KY213735
HM034	AE	R5	851	2.8	40	29	ND	4922	456	ND	ND	KY213736
HM045	AE	R5	856	76.9	30.7	29	0.731	239846	488	0.3077	6.0547	KY213737
HM038	AE	R5	860	41.6	40.2	28	1.5472	25586	681	11.4258	>25	KY213738
HM069	AE	R5	846	5.3	34.3	26	0.259	NA	NA	1.3926	16.3086	KY213739
HM039	B	R5	858	71.7	30.5	33	3.44	1680160	207	0.4339	2.0508	KY213740
HM024	B	R5	844	5.3	26.4	31	1.18	1765830	275	1.7962	17.2745	KY213741
HM027	B	R5	862	83	30.6	33	5.04	454595	283	9.4727	>25	KY213742
HM080	B	R5	857	34.6	31	31	3.125	174861	309	0.1966	1.43	KY213743
HM151	B	R5	859	4.6	25	35	1.4203	16237	385	1.2938	12.5	KY213744
HM141	B	R5	856	50.2	31.1	32	1.4514	7708969	493	0.5318	3.2227	KY213745
HM053	B	R5	850	9.6	35	34	5.114	79255	557	2.7344	>25	KY213746
HM075	B	R5	850	22	31.3	30	2.08	23180	590	1.1938	14.8438	KY213747
HM122	B	R5	845	22	31.3	30	2.92	127270	635	>25	>25	KY213748
HM140	AE	R5X4	852	0.5	41.4	28	0.5429	1215259	30	0.9015	9.9609	KY213749
HM153B*	AE	R5X4	870	1.7	62.8	31	ND	13049	181	0.9178	7.5195	KY213750
HM106	B	R5X4	844	20.8	40.7	33	ND	1255410	99	4.6875	>25	KY213751

^*Clones HM153B and HM153I were from the same patient.

Figure 2. Maximum likelihood phylogenies based on HIV-1 Env sequences constructed using the GTR model with a gamma distributed evolutionary rate differences among sites.

(A) 27 subtype AE sequences and 10 subtype B sequences from Singapore used in this study. (B) 37 sequences from this study together with 26 other reference subtype AE and B from the HIV sequence compendium (http://www.hiv.lanl.gov/).

Figure 3. Genotypic and phenotypic analysis of cloned HIV Envelopes.

37 full length functional Envelopes were cloned from plasma samples of HIV infected patients and fully sequenced. Comparison of subtype AE and B Envs in terms of (A) Env length (B) Geno2pheno FPR (C) Risk of X4 development based on Geno2Pheno sequence analysis. Significantly greater risk for 4 development seen in subtype AE Envs (p=0.0167) compared to B subtype. (D) The sequence of envelopes cloned in this study was analyzed for number of PNGS. Significant differences in PNGS (p=0.0003) between subtype AE and B Envs. Comparison of CD4 counts between Envs with (E) Geno2Pheno FPR <21 vs ≥ 21 or (F) X4 risk >35 vs ≤35. (G) Log viremia between Env groups based on X4 risk >35 or X4 risk <35. (H) Fewer PNGS (p=0.0293) seen in Envs isolated from patients with CD4<200 vs CD4≥200.

Interestingly, no statistically significant difference in Env Length (Figure 3A) or Geno2pheno FPR (Figure 3B) was found between subtype B and AE Envs. However, the risk of X4 emergence based on Geno2pheno sequence analysis was significantly greater for subtype AE Envs (p=0.0167) compared to B subtype (Figure 3C). We next looked at differences in PNGS in Envs cloned in this study and found that subtype B had significantly more PNGS (p<0.0003) compared to AE Envs (Figure 3D). We also looked at correlates of X4 prediction and disease markers like CD4 count and plasma viremia. Although not statistically significant, CD4 counts were lower in patients with Envs that had a Geno2pheno FPR of <21 (Figure 3E). Interestingly, there was a significant difference (p=0.0219) in CD4 counts when comparing Envs with higher risk of X4 development (Figure 3F). Furthermore, no significant difference was found between risk of X4 development and plasma viremia (Figure 3G). Interestingly, patients with CD4 counts <200 had Envs with fewer PNGS (p=0.0293) than those with CD4 counts ≥ 200 (Figure 3H) consistent with other studies showing correlation between PNGS and stage of disease (Borggren et al., 2011). Thus, we have generated a set of full length functional AE and B subtype Envs from the same geographical region that have been characterized genotypically and can be used for phenotypic assays and to better understand the differences between subtypes in terms of disease progression.

Subtype AE Envs are characterized by lower PNGS than Subtype B

The HIV Env is heavily glycosylated and is known to affect many aspects of HIV biology including Env folding, generation of host immune response, virus transmission etc. (Clark, 2014; Huang et al., 1997; Mathys and Balzarini, 2014). Although, the location of potential N and O linked glycosylation sites in the Env are encoded by the virus genome, the glycosylation pattern is affected by the interactions between the Env and the cell type the virus infects (Clark et al., 1997). Interestingly, compared to the high degree of variability in the viral Env, the positions of N linked glycosylation sites are relatively well conserved between clades and isolates (Zhang et al., 2004). Moreover, a direct link between Env glycosylation pattern and Env phenotypes like switch to CXCR4 co-receptor usage along with subtype differences has been observed (Ng et al., 2013; Utachee et al., 2010). Analysis of 1268 subtype B and 423 subtype AE sequences showed a higher prevalence of PNGS in the B subtype Envs (Figure 4A) consistent with findings seen with the Envs cloned in this study. Overall AE subtype had lower PNGS than all other subtypes including A, B, C and D, suggesting a more universal applicability of our findings. We next focused on different regions of the Env to determine whether a specific region of the Env contributed towards the observed differences in PNGS between AE and other subtypes. Using sequences from the Los Alamos database we found higher PNGS both in the gp120 (Figure 4B) and gp41 region in subtype B (Figure 4C). On examination of the variable loop regions (Figure 4D and E), the V3 loop stood out because the number of PNGS was lower for X4-tropic viruses especially for subtype AE (Figure 4E) when compared to subtypes A, B, C and D. We next looked at the target sequence motif for N linked glycosylation, the NX(S/T) motif where X represents any amino acid other than proline (Gavel and von Heijne, 1990). When all PNG sites in V3 loop were compared for subtypes AE and B, we found that the NX(S/T) motif at N-glycosylation site N301 (position 6 in V3 loop) was more prevalent in subtype B viruses than subtype AE viruses (Table 3). Moreover, number of PNGS for 3 categories of subtype AE viruses (R5-only, dual tropic and X4-only) showed a significant step-wise reduction, indicating a possible role of N301 glycan in co-receptor tropism (Table 3). We also looked at differences in N332/N334 glycan site and found that the N334 glycan site for subtype AE viruses tends to be more glycosylated in X4 isolates compared to R5 tropic viruses (Table 3). Although, it might be possible that the stronger presence of glycosylation at the N334 site in subtype AE could work in tandem with the lack of glycosylation at site N301 for the X4 tropic subtype AE viruses, we note that glycosylation at this site, unlike the case for N301, does not show a stepwise trend from R5-tropic to dual-tropic and then to X4-tropic viruses. Moreover, these N332/N334 glycan sites also do not seem to play a significant role in co-receptor tropism in the other subtypes examined. Hence, even if the glycan at the N332/N334 sites do play some role in tropism for some subtypes, its effect would be much less significant compared to the glycan site at N301. Thus, combined findings from our data above demonstrate that PNGS in gp160, gp120, gp41 and V3 loop for subtype AE are significantly lower than CCR5-tropic subtype B viruses.

Figure 4. Analysis of PNGS in different regions of the Env in subtype B versus AE viruses.

(A) Bar graph representing average number of PNGS in the gp160 region of subtype B and AE sequences available at the Los Alamos database. Other subtypes Envs including A, B, C and D were also analyzed. (B) Bar graph representing average number of PNGS in various HIV subtypes in the gp120 region from sequences available at the Los Alamos database. (C) Bar graph representing average number of PNGS in various HIV-1 subtypes in the gp41 region from sequences available at the Los Alamos database. (D) Bar graph representing average number of PNGS in various HIV-1 subtypes in the V1/V2 loop region from sequences available at the Los Alamos database. (E) Bar graph representing average number of PNGS in various HIV-1 subtypes in the V3 loop region from sequences available at the Los Alamos database. The number of sequences analyzed for each Env region in parts A–E is indicated in part A. Dual-tropic viruses were excluded in the “X4” analysis.

Table 3.

Subtype B or AE sequences with PNG at position 301.

	% PNG at Pos 301 (Total)	p-value (relative to R5)	% PNG at Pos 332/334 (Total)	p-value (relative to R5)
Subtype A (R5)	95.76% (283)		95.51% (245)
Subtype A (dual)	97.37% (38)	0.64	92.11% (38)	0.40
Subtype A (X4)	12.50% (8)	1.63E-24	100% (8)	0.54
Subtype AE (R5)	93.78% (563)		94.27% (558)
Subtype AE (dual)	52.0% (50)	2.0E-23	93.88% (49)	0.26
Subtype AE (X4)	34.62% (104)	7.1E-64	100% (103)	0.0003
Subtype B (R5)	98.91% (3594)		95.04% (3366)
Subtype B (dual)	85.63% (668)	2.9E-75	92.22% (630)	6.33E-07
Subtype B (X4)	71.16% (215)	2.5E-142	97.10% (207)	0.02
Subtype C (R5)	94.39% (1498)		88.8% (1357)
Subtype C (dual)	94.32% (88)	0.98	94.87% (78)	0.09
Subtype C (X4)	78.38% (74)	3.16E-08	100% (69)	0.003
Subtype D (R5)	96.51% (401)		93.67% (380)
Subtype D (dual)	65.43% (162)	2.83E-26	83.75% (160)	0.14
Subtype D (X4)	42.38% (151)	6.53E-61	93.29% (149)	0.87

Structural analysis of the V3 loop bound to CXCR4/CCR5 and the N301 glycosylation site

Our data above suggested that the AE and B subtypes differ in their Env glycosylation pattern especially at position 301. We hence proceeded with structural analysis to determine if the above difference in Env glycosylation may have an effect on CXCR4 or CCR5 binding. A closer look at this position in the structure of V3 loop bound to CCR5 and CXCR4 indicated that N-glycosylation at position 301 may sterically hinder binding to the CXCR4 receptor but would not affect binding of V3 loop to CCR5. This hypothesis is corroborated by the fact that while GlyProt found that it was sterically not feasible to create a basic glycan structure at N301 for the V3-loop bound CXCR4 structure, it nonetheless successfully added a basic glycan to N301 in the V3-loop bound CCR5 structure (Figure 5A and B). Hence, viruses without N-glycosylation at position 301 could potentially bind to CXCR4 more easily than viruses with N-glycosylation at that position. As a greater proportion of subtype AE viruses compared to subtype B viruses did not have the N-glycosylation motif at this position, it is possible that this may play a role in enabling more subtype AE viruses to switch to CXCR4 co-receptor usage. The higher risk of X4 development seen in our Geno2pheno analysis (Figure 3C) is consistent with this observation.

Figure 5.

(A and B) Structural representation of glycosylation site N301 (Position 6 in V3 loop, yellow tube). GlyProt was used to add glycan (magenta balls) to the protein structures. GlyProt could not find any feasible orientation to glycosylate N301 in the (A) CXCR4 structure (red tube), but successfully glycosylated this position in the (B) CCR5 structure (blue tube). Glycan was also added to positions 11 and 176 in the CXCR4 structure, and positions 268 in the CCR5 structure. The residues on the host receptors that are closest to N301 in the V3 loop are highlighted in cyan (position Ser23 in CXCR4 with a distance of 2.96 Angstrom from N301 and position Glu18 in CCR5 with a distance of 3.87 Angstrom from N301). (C) Structural overview of PNG sites that were found to be more prevalent in subtype B (in pink) and PNG sites that were found to be more prevalent in subtype AE (in blue). Env trimer (PDB:4TVP) was structurally aligned to Env trimer (PDB:3J70) bound to CD4 (red balls and surfaces). The orange ball represents gp41 and the green ball and ribbon structures represent gp120. In the ribbon portion of the structure, V1 loop is colored in orange, V2 in magenta, V3 loop in yellow (and red, representing its position after CD4 binding), V4 loop in cyan and V5 loop in purple. The directions of the arrows indicate the movement of the V1V2 and V3 loops right after CD4 binding.

We also surveyed all PNG sites in subtype AE and B where the difference in the proportion of PNGs at the given site was greater than 20%. For the variable loop regions, due to the flexibility of PNGs occurring in the vicinity of any given site, the same threshold of 20% was applied over a distance of 4 amino acid residues from the given PNG site. Only 3 sites (positions 289, 362 and 816) met this criterion (Table 4) and these sites could potentially represent differential PNG sites between subtype AE and B viruses (Figure 5C). When we compared these differential PNG sites to the viruses used in this study, we observed a similar glycosylation pattern (Table 4). The non-glycosylated amino acid residue at position 362 is within 8 angstrom from CD4 and N-glycosylation at this site could bring the virus closer to CD4 and potentially play a role in CD4 binding.

Table 4.

Positions in the Env where the difference in proportion of PNG at a given site is greater than 20% between subtypes AE and B.

Positions	289	362	816
HIV Region	C2	C3	gp41 (C-terminal)
Subtype AE (Los Alamos)	93.1%	0.0%	11.8%
Subtype B (Los Alamos)	67.7%	62.5%	75.2%
Difference (%B - %AE, Los Alamos)	−25.4%	62.5%	63.4%
Subtype AE (this study)	96.4%	0.0%	21.4%
Subtype B (this study)	44.4%	66.7%	100.0%
Difference (%B - %AE, this study)	−52.0%	66.7%	78.6%

Prevalence of residues (layer mutations) that differ significantly between HIV-1 subtype B and AE viruses

The close correlation between the subtype differences seen in our set of Envs versus those from the Los Alamos database suggests that some of these genotypic differences could be subtype specific and a consequence of founder effect. Recently, Zoubchenok et al analyzed residues in AE Envs important for CD4 binding (Zoubchenok et al., 2017) especially at position 375 and its interaction with other residues in the gp120 inner domain layers as HIV evolved. They determined that AE strains possess a Histidine at position 375 instead of the highly conserved Serine in group M. This His375 residue is part of the Phe43 cavity where Phe at position 43 in CD4 engages with gp120 and is required for efficient AE Env binding to CD4. Moreover, residues in the inner domain layers of gp120 have co-evolved with His375 to facilitate viral entry. We hence looked at subtype specific differences in residues at positions 375, 61 (layer 1), 105 and 108 (layer 2), 474 to 476 (layer 3) from Envs cloned in this study as well as sequences available from the Los Alamos database. Analysis of Env sequences confirmed the presence of His at position 375 in AE subtype and prevalence of Ser in the B subtype (Figure 6). Overall, percentages of the major residues based on the isolates in this study matches well with those from the Los Alamos database. Moreover, our data concur with findings of Zoubchenok et al except for the lower prevalence of His at position 105 in subtype B which is likely a bias caused by small sample size (Figure 6). These data suggest that genotypic differences between AE and B subtype including PNGS and His 375 are consistent with subtype specificity. Whether these changes affect phenotype of the virus remains to be seen.

Figure 6. Prevalence of residues that differ significantly (also called “layer mutations”) between HIV-1 subtype B and subtype AE viruses.

Analysis of subtype B and subtype AE sequences at positions 375, 61 (layer 1), 105 and 108 (layer 2), 474 to 476 (layer 3) from Envs cloned in this study and Env sequences available from the Los Alamos database. The percentage of the major residues based on the isolates in this study is depicted in light blue and light red. The percentages match well with sequences from the Los Alamos database (db) depicted in dark blue and dark red.

Subtype B Envs show higher infectivity in cells expressing low levels of CCR5

Recent studies have documented differences in disease progression rate in HIV subtype B versus AE infected individuals. Subtype differences have been shown to influence CD4 decline rates, (Touloumi et al., 2013) leading to faster time to anti-retroviral therapy (Ng et al., 2011a). However, the reason behind these differences remains unknown. Our genotypic analysis showed conserved differences between AE and B subtype in terms PNGS, CD4 binding region and V3 loop. This raises the possibility that phenotypic differences between AE and B subtype may also be conserved and correlate with the genotypic differences. We hence conducted a phenotypic analysis of the Envs in terms of virus infectivity and CCR5 engagement. We tested the infectivity of the Envs in TZM-bl cells which are adherent cells expressing high levels of CXCR4 and CCR5 or T cell lines L23 and H6 that express low or high levels of CCR5 respectively (Joshi et al., 2011). We did not find significant differences in infectivity between the AE and B subtypes in cells expressing high levels of CCR5 (TZM-bl and H6) (Figure 7A and B). Interestingly, significant difference was found between AE and B subtypes in terms of infectivity in cells expressing low levels of CCR5 (L23) (Figure 7C), with B subtype Envs showing significantly higher infectivity (p=0.042) in L23 cells. These data suggest phenotypic differences in viral subtype in terms of CCR5 usage may translate to differences in viral tropism.

Figure 7. Subtype B Envelopes have higher infectivity in cells expressing low levels of CCR5.

Virus stocks were prepared in 293T cells by transfection with pNLLuc-R-/E- HIV backbone along with different full length Env constructs. The virus stocks were used to conduct infectivity assays in (A) TZM (B) H6 and (C) L23 cells. Luciferase activity was determined 48 hrs post infection using the BriteLite Plus Luciferase assay substrate (PerkinElmer). Significantly higher (p=0.0421) infectivity of B subtype envelopes in low CCR5 expressing cells L23.

MVC IC₅₀ are higher for subtype B and correlate with L23 infectivity and PNGS

Our data above suggests subtype B Envs are more efficient than AE subtype in infecting cells expressing low levels of CCR5. Moreover, we have recently shown that CCR5 levels also affect HIV evolution and MVC sensitivity (Garg et al., 2016). We hence investigated whether Envs with higher infectivity in L23 cells also showed reduced sensitivity to the CCR5 antagonist Maraviroc. As none of the patients were on Maraviroc therapy these findings would have implications for Maraviroc treatment based on virus subtype. We determined MVC sensitivity for the cloned Envs by conducting pseudotyped virus based infectivity assays in TZM-bl cells in the presence of varying concentrations of MVC. Interestingly, we found that subtype B Envs were inherently less sensitive to MVC compared to subtype AE Envs (Figure 8A). Moreover, the MVC IC₅₀ correlated with infectivity in L23 cells (Figure 8B) as well as number of PNGS in the Env (Figure 8C). Although relatively infectious in L23 cells, the infectivity of HM034 was too poor in TZM cells to determine MVC IC₅₀ (Table 2). Overall, these data suggest that compared to subtype AE, subtype B Envs are less susceptible to MVC that correlates with better infectivity in L23 cells and higher PNGS. Collectively these data support the notion that AE subtype Envs might engage CCR5 differently than subtype B Envs.

Figure 8. MVC IC50s are higher for subtype B Envelopes and correlates with L23 infectivity and PNGS.

(A) pNLLuc-R-/E- virus stocks pseudotyped with different Envs were prepared as described above. Virus stocks were added to TZM-bl cells in the presence of different concentrations of MVC. Infection was determined 48 hrs post infection by measuring luciferase activity in the cultures. Infectivity curves of each Env were fitted using Sigma Plot software using the logistic 4-parameter regression method and IC50 values calculated from the curves. B subtype Envelopes have significantly higher MVC IC50s (p=0.0430) than AE subtypes. (B) Significant correlation of MVC IC50s with L23 infectivity (p=0.0123) and (C) PNGS (p=0.0058).

Neutralization sensitivity of AE and B subtype Envs to TriMab antibody mix

Sensitivity of HIV Env to neutralization by the TriMab antibody mix has been shown to correlate with PNGS (Borggren et al., 2011). As a striking difference between AE and B subtypes was seen in terms of PNGS, we determined the neutralization sensitivity of the cloned AE and B Envs using the TriMab antibody mix (Table 2). As shown in Figure 9A and B, there was no significant difference in the mean TriMab Inhibitory Concentration 50 (IC₅₀) or Inhibitory Concentration 90 (IC₉₀) between AE and B subtype Envs. Interestingly, the AE subtype Envs that were neutralization resistant (IC₉₀>25μg/ml) showed significantly lower (p=0.0297) PNGS than the resistant B Envs (Figure 9C). This suggests that AE Envs may acquire resistance to TriMab antibody mix in the absence of PNGS upregulation, further highlighting the genotypic and phenotypic differences between these subtypes.

Figure 9. Determination of neutralizing antibody sensitivity of AE and B subtype Envs.

Virus stocks pseudotyped with various Envs were used to infect TZM-bl cells in the presence of different concentrations of the TriMab antibody mix as described in the methods. Infection was determined 48h post infection by measuring luciferase activity. (A) TriMab IC50 for AE and B subtype Envs. (B) TriMab IC90 for AE and B subtype Envs. (C) PNGS in Envs that were resistant to TriMab inhibition defined as IC90 >25μg/ml.

DISCUSSION

The marked variability in the course and manifestation of HIV disease can be attributed to several host factors as well as viral determinants (Garg and Joshi, 2017). One of the key factors contributing to this variability is the remarkable genetic heterogeneity of HIV-1 (Tebit et al., 2007) which has led to the evolution of several subtypes and CRFs since its introduction into the human population. Interestingly, certain strains/subtypes circulate in defined populations which emphasizes the relevance of host genetic makeup in governing the course of HIV disease (Kawashima et al., 2009). In this study, we cloned 37 full length Envs from a cohort from Singapore with documented differences in disease progression (Ng et al., 2013; Ng et al., 2011b) and determined the genotypic and phenotypic differences in the Env glycoproteins derived from AE and B subtype infected individuals. While genetic analysis alone is of value, its credibility can be further improved when combined with phenotypic analysis, which was the objective of this study.

Of the 37 cloned Envs, 10 Envs belonged to subtype B while 27 were subtype AE and a total of 3 were dual tropic (2 AE and 1 B subtype). The rapid spread of the AE subtype throughout Southeast Asia makes these viral variants both interesting and important to study (Hemelaar et al., 2011). This is likely a result of complex multifactorial processes with both the replicative capacity of the transmitted virus (Claiborne et al., 2015) and host genetics (Joshi et al., 2017) playing important roles in the outcome of this phenomenon. With regards to co-receptor usage we saw interesting results. It is generally accepted that during early asymptomatic HIV infection, CCR5 tropic viruses predominate that are highly transmissible but poorly fusogenic (van’t Wout et al., 1994). With disease progression, there is a shift in co-receptor usage to CXCR4 or CXCR4/CCR5 (dual tropic) accompanied by increase in Env fusogenic potential leading to rapid CD4 decline and progression to AIDS (Bjorndal et al., 1997; Karlsson et al., 1994; Koot et al., 1993). However, our U87 tropism assay and Geno2pheno analysis both predicted majority of our Envs as CCR5 tropic and only 3 Envs as dual tropic. This is surprising as these patients were at various stages of HIV disease and in line with other findings, suggesting that in many cases infected individuals progress to AIDS without co-receptor switching and maintain R5 tropism (Jansson et al., 1999; Karlsson et al., 1994; Koot et al., 1993).

While several host factors have been implicated in HIV pathogenesis, the role of CCR5 in this process is undeniable. CCR5 levels vary considerably in the human population due to CCR5 gene and promoter polymorphisms (Chalmet et al., 2008; Gornalusse et al., 2015; Ometto et al., 2001; Samson et al., 1996) and this variability in CCR5 levels translates to differences HIV disease progression (Joshi et al., 2017). This is best documented by slower disease progression in CCR5Δ32 heterozygous individuals who show reduced CCR5 expression on the cell surface (Liu et al., 1996). To extend the physiological relevance of our findings, we conducted infectivity assays for our panel of cloned Envs in cells expressing high or low levels of CCR5 (Joshi et al., 2011). While there was no difference in infectivity in high CCR5 expressing cells; subtype B Envs were more efficient at infecting low CCR5 expressing cells compared to AE Envs. This phenomenon could be an adaptation acquired by subtype B Envs to overcome neutralization by antibodies, as these Envs are known to be highly susceptible to antibody mediated neutralization (Binley et al., 2004).

Subtype based differences in HIV disease progression, geographical spread, co-receptor usage and drug resistance have been documented (Hemelaar, 2012, 2013). Subtype AE is rapidly spreading in Southeast Asia and a faster disease progression with this subtype along with higher baseline viral load has been reported (Li et al., 2015; Ng et al., 2011a). However, in these studies the slightly higher baseline viral load with AE subtype, was not statistically significant when compared to subtype B, consistent with our study. Within our cloned Envs group, subtype B showed significantly higher viremia, which may be a result of the small sample size. Nevertheless, the lower CD4 counts in the AE subtype suggests that viremia per se may not be responsible for the differences in CD4 counts. Thus, the genotypic and phenotypic pattern of differences between subtype AE and B Envs warrant further investigation especially in the context of HIV pathogenesis.

In our study, subtype AE Envs showed significantly lower levels of PNGS compared to subtype B. This difference in PNGS correlated with better utilization of low levels of CCR5 by subtype B versus AE, as well as reduced sensitivity to CCR5 antagonist MVC in subtype B. This phenotypic relationship between the ability to utilize low levels of CCR5 for infection and higher MVC IC₅₀ is consistent with our in vitro study (Garg et al., 2016) where adaptation of HIV YU-2 to replicate in cells expressing low levels of CCR5 resulted in a similar phenotype. The differences in PNGS was also confirmed in a larger set of Envs from the Los Alamos database. While differences in PNGS with regards to stage of HIV diseases have been well studied (McMichael et al., 2010; Rong et al., 2009; Wei et al., 2003), the difference with regards to viral subtypes is interesting and could be relevant to immune evasion by certain subtypes or a compensation for replicative fitness. Interestingly, it has also been suggested that fewer PNGS may have a role in transmitted founder viruses as well as better sexual transmission (Shi et al., 2016). The occurrence of fewer PNGS in subtype AE may provide a transmission advantage that may explain the rapid spread of this subtype across Southeast Asia.

Recently, Zoubchenok et al analyzed residues in CRF01_AE Envs important for CD4 binding (Zoubchenok et al., 2016) and identified Histidine 375 as a critical residue in the process. Overall, our data concur with findings of Zoubchenok et al except for the lower prevalence of His at position 105 which is likely a bias caused by small sample size. In another study, Prevost et al showed that AE subtype expose the ADCC epitopes on the Env glycoprotein as a consequence of the H375 substitution (Prevost et al., 2017). Furthermore, the insensitivity of AE subtype to broadly neutralizing antibodies that bind the CD4 interacting region on Env has been well documented (Chen et al., 2016; Utachee et al., 2014). While these studies provide evidence that Envs from AE subtype engage CD4 differently and have specific changes within the CD4 binding region of the Env; our study shows that AE subtype also differs in CCR5 engagement compared to subtype B Envs from the same region.

The relevance of these phenotypic differences in Env glycoprotein between subtypes and how they affect disease outcome and or viral spread remains to be determined. The role of PNGS in viral tropism as well as sensitivity to neutralizing antibodies is well established (McMichael et al., 2010; Rong et al., 2009; Wei et al., 2003). Moreover, it is well studied that subtype B HIV seems to be highly sensitive to broadly neutralizing antibodies from our studies and others (Binley et al., 2004; Joshi et al., 2014). One could hence anticipate that increase in PNGS in subtype B is a consequence of selection against neutralizing antibodies. Increased infectivity in cells expressing low levels of CCR5 may also be an adaptive measure to circumvent antibody pressure by subtype B Envs. On the other hand, recombinant subtypes like AE may circumvent antibody pressure differently. It is documented that AE subtypes show higher viremia than B subtypes (Kivela et al., 2005) and is plausible that this subtype may have evolved a way to circumvent antibody pressure without shielding of the Env via glycosylation unlike subtype B viruses. This is further supported by our observation that AE Envs that were resistant to TriMab antibody mix had fewer PNGS compared to the B subtype resistant Envs. The need to utilize low levels of CCR5 for replication and survival may hence also be dispensable for subtype AE viruses. While our study focusses on the subtype based differences in the Env glycoprotein, the role of other viral proteins like Pol and Gag in this phenomenon cannot be overlooked. Ng et al described that polymorphisms in the pol gene were responsible for differences in replication capacity and disease progression in subtype A and D HIV infected individuals in Uganda (Ng et al., 2014). Similarly, Clairborne et al attributed the early T-cell activation and immune dysfunction in HIV infected Zambians to the replicative fitness of the transmitted virus with the gag gene contributing significantly to the phenomenon (Claiborne et al., 2015). It is worth noting here that while genotypic and phenotypic differences in AE and B subtype Envs correlates with CD4 decline, this correlation does not necessary mean causation. Although Env glycoprotein is a major determinant of HIV pathogenesis, (Garg et al., 2011; Garg et al., 2012; Joshi et al., 2014; Joshi et al., 2011) further studies will be needed including those with full length viruses derived from these Envs to determine the role of Env variants in AE versus B subtype pathogenesis.

Overall, our study established 37 full length Env clones from subtype B and AE HIV infected individuals from Singapore along with their genotypic and phenotypic characterization. Our data demonstrates that subtype B Envs have more PNGS, higher infectivity in cells expressing low levels of CCR5 and a higher MVC IC₅₀ when compared to AE Envs. Hence, AE Envs most likely engage CCR5 differently than subtype B. These data hold important implications for subtype based differences in HIV pathogenesis, virus evolution and MVC resistance.

MATERIALS AND METHODS

Ethical Statement

The study was reviewed and approved by the Texas Tech University Health Sciences Center’s regional Institutional Review Board and considered exempt (study number recorded as IRB# E11052). The Singapore National Healthcare Group ethics committee approved this study [DSRB: 2012/00438].

Study population

Details of study population have been described elsewhere (Ng et al., 2013; Ng et al., 2011b). In brief, treatment naïve HIV-1-infected patients presenting for care at the Singapore Communicable Disease Centre outpatient clinic from February 2008 to August 2009 were enrolled into the study. Inclusion criteria were based on (1) confirmed HIV-1 diagnosis by ELISA and Western Blot and (2) treatment-naïve at presentation for care. Clinical data and demographic information were obtained from case notes, self-reported surveys in medical interviews and electronic medical records. Follow-up was per routine clinical care, with patients seen at 3 to 6 month intervals.

Env cloning

Viral RNA was isolated from the plasma samples using the QiaAmp viral RNA mini kit (Qiagen) following the manufacturer’s protocol followed by cDNA synthesis using the ProtoScript II first strand cDNA synthesis kit (New England Biolabs). The full length Env region (containing the open reading frames for the Env and Rev genes) was then amplified with subtype B or AE specific primers (Nie et al., 2010) and a nested PCR reaction using the Phusion High Fidelity PCR kit (New England Biolabs). The amplified Env region was cloned into the pCDNA3.1+ vector using the pCDNA3.1 directional TOPO® expression kit (Invitrogen) followed by full length sequencing analysis to verify the authenticity of the inserts. Contigs were established using DNA star software (DNASTAR Inc., Madison, WI) and Env Open reading frames for all constructs established. The GenBank accession numbers for the cloned Envs are listed in Table 2. Functionality of each Env was determined by generating pseudotyped HIV particles using respective Env constructs and pNLLuc R-/E- backbone as described below.

Cell lines and Transfections

SupT1 cells expressing low (L23) medium (M10) or high (H6) CCR5 have been described previously (Joshi et al., 2011) and were maintained in RPMI medium supplemented with 10% FBS, penicillin streptomycin (5000U/ml) and Blasticidin (3μg/ml). 293T, HeLa and TZM-bl cells (kindly provided by the NIH AIDS research and reference reagent program) were maintained in Dulbecco’s Modified Eagles medium supplemented with 10% FBS and penicillin streptomycin (5000U/ml). TZM-bl are HeLa derived cells that express the HIV receptor CD4 and the co-receptors CXCR4 and CCR5 along with the luciferase and beta galactosidase genes under the control of HIV LTR (Platt et al., 2009). These cells readily support HIV infection/replication. U87 cells expressing CD4 and either CXCR4 or CCR5 were obtained from the NIH AIDS reagent program and were maintained in DMEM-10 supplemented with 1 μg/ml Puromycin and 300 μg/ml G418. All transfections were conducted using the Turbofect transfection reagent (Fisher Scientific) following the manufacturer’s instructions.

Generation of virus stocks and infectivity assays

293T cells were transfected with the pNLLuc-R-/E- (Connor et al., 1995) HIV backbone along with different Env constructs. Virus stocks were harvested 48 hrs post transfection and titrated in TZM-bl cells using 2-fold serial dilutions of the stocks. Infectivity assays were conducted in TZM cells in the presence of 20μg/ml DEAE dextran (Sigma) and in H6/L23 cells in the presence of 10μg/ml Polybrene (Sigma). Luciferase activity was determined 48 hrs post infection using the BriteLite Plus Luciferase assay substrate (PerkinElmer). Infectivity for each Env was calculated as percent of YU-2 Env control. For some experiments, the CXCR4 antagonist AMD-3100 (2μM) or the CCR5 antagonist MVC (1μM) was added at the time of infection.

MVC and neutralizing antibody sensitivity

pNLLuc-R-/E- virus stocks pseudotyped with different Envs were prepared as described above. Virus stocks were added to TZM-bl cells in the presence of 20μg/ml DEAE dextran and different concentrations of MVC. For determining neutralization sensitivity, virus stocks were added to different concentrations of the TriMab antibody mix (IgGb12, 2F5, and 2G12) starting with a concentration of 25 μg/ml. The virus antibody mix was incubated at 37°C 1 hr and subsequently added to TZM-bl cells. Infection was determined 48 hrs post infection by measuring luciferase activity in the cultures as above. Infectivity curves for each Env were fitted using the Sigma Plot software (Systat software Inc.) and logistic 4-parameter regression method. Subsequently, IC₅₀ and IC₉₀ values were calculated from the curves.

PNG analysis between subtypes based on sequences from the Los Alamos database

In order to examine the distribution of PNGS across Env and their possible role in co-receptor tropism, full length Env protein sequences that are either X4-tropic or R5-tropic but not dual-tropic were downloaded from Los Alamos HIV sequence database (http://www.hiv.lanl.gov/), with the default settings of omitting problematic sequences, for subtypes A, AE, B, C and D. As our data indicated that PNGS in the V3 loop region could play a significant role in co-receptor tropism, a separate larger set of sequences that contains the V3 loop region and has tropism data (R5-tropic, dual-tropic and X4-tropic) were downloaded from the Los Alamos database for the analysis of the effects of glycosylation at positions 301, 332 and 334 have on tropism. The sequence motif NX(S/T), where X represents a non-proline amino acid residue, is used to predict PNGS. The Env sequences were segregated into various regions (gp120, gp41, V1V2 loop, V3 loop, V4 loop and V5 loop) for further PNG analyses. When comparing differential PNG sites between subtypes B and AE, sequences were aligned to the reference HXB2 sequence using MAFFT (Katoh et al., 2005) and HXB2 amino acid numbering was used. Positions between subtypes AE and B where there were greater than 20% differences in their proportion of PNGS were highlighted in the PDB structure.

Structural visualization of PNG sites

To visualize the differential PNG sites between subtypes AE and B on the Env, the asymmetric units of 4TVP (Pancera et al., 2014) representing the Env trimer were structurally aligned to the Env trimer bound to CD4 structure 3J70 (Rasheed et al., 2015). As there are no experimentally determined native structures of the Env V3 loop binding to either of the co-receptors, we used the published homology model of V3 loop binding to both co-receptors for this analysis. N-glycosylation site N301 in the V3 loop was highlighted in the V3 loop-CXCR4 complex and the V3 loop-CCR5 complex using structural models 0095767.s012 and v3loop_cxcr4.1 built by Tamamis and Floudas (Tamamis and Floudas, 2013, 2014). GlyProt (Bohne-Lang and von der Lieth, 2005) was used to add basic glycan structures onto all feasible glycan sites in these 2 structures where the V3 loop is bound to the co-receptors. Structural alignments were conducted using MUSTANG (Konagurthu et al., 2006) and structures were rendered and visualized using YASARA (Krieger and Vriend, 2014).

Data analysis

Data were analyzed using GraphPad Prism (GraphPad Software, Inc, La Jolla, CA) and SAS 9.3 (SAS Institute, Inc., Cary, North Carolina). Differences between groups were assessed using two-sample t-test. All p values were two-sided and data were considered significant at p<0.05. Pearson’s correlation with linear regression was used for all correlation determination using the GraphPad Prism Software.

HIGHLIGHTS.

• HIV subtype CRF01_AE is associated with lower CD4 counts.

• CRF01_AE Envs have lower PNGS than other subtypes

• Subtype B are better at utilizing lower levels of CCR5 for viral entry.

• N301 glycosylation site is seen less frequently in AE subtype and CXCR4 topic viruses.