Worldwide genetic features of HIV-1 Env α4β7 binding motif: the Local Dissemination Impact of the LDI tripeptide

Abstract

Background

HIV-1 gp120 binds to integrin α4β7, a homing receptor of lymphocytes to gut-associated lymphoid tissues. This interaction is mediated by the LDI/V tripeptide encoded in the V2-loop. This tripeptide mimics similar motifs in MAdCAM and VCAM, the natural ligands of α4β7. In this study we explored the association of V2-loop LDI/V mimotopes with transmission routes and patterns of disease progression in HIV-infected adult and pediatric patients. HIV-1 env sequences available in the Los Alamos HIV Sequence Database were included in the analyses.

Methodology

HIV-1 V2-loop sequences generated from infected adults and infants from South and Southeast Brazil, and also retrieved from the Los Alamos Database, were assessed for α4β7 binding tripeptide composition. Chi-Square/Fisher’s Exact test and Mann Whitney U-test were used for tripeptide comparisons. Shannon entropy was assessed for conservancy of the α4β7 tripeptide mimotope.

Results

We observed no association between the tripeptide composition or conservation and virus transmission route or disease progression. However, LDI was linked to successful epidemic dissemination of HIV-1 subtype C in South America, and further to other expanding non-B subtypes in Europe and Asia. In Africa, subtypes showing increased LDV prevalence evidenced an ongoing process of selection towards LDI expansion, an observation also extended to subtype B in the Americas and Western Europe.

Conclusions

The V2-loop LDI mimotope was conserved in HIV-1C from South America and other expanding subtypes across the globe, which suggests that LDI may promote successful dissemination of HIV at local geographic levels by means of increased transmission fitness.

INTRODUCTION

Integrins comprise a superfamily of twenty-four heterodimeric transmembrane proteins which are essential for cell homing to and retention within specific sites, contributing significantly to the architecture of most tissues.^1,2 Integrin heterodimers are formed by the specific association of one of 18 α and one of eight β subunits.³ Stimulation of lymphocytes followed by their migration to lymphoid tissues, a mechanism required for most immune responses, relies on integrin activation and recognition.⁴

The α4β7 integrin (α4β7) plays a pivotal role in lymphocyte homing and retention to the gut-associated lymphoid tissue (GALT). Homing is mediated by VCAM and MAdCAM-1, the natural ligands of α4β7, which share a related tripeptide in which a central aspartic acid (Asp) residue is invariable. In both cases, the Asp is flanked by an aliphatic residue; in VCAM the tripetide appears as IDS,⁵ while in MAdCAM it appears as LDT.⁶ The LDI/V motif is not a direct contact site with its natural ligands. Rather, the Asp in each of these three ligands mediates metal cation coordination between the ligand and α4β7. HIV-1, which replicates at high levels in GALT during the acute phase of infection, encodes a similar tripeptide motif in the V2 loop of the gp120 envelope protein (HXB2 env gene codons 179–181),⁷ which also mediates binding to α4β7. Although α4β7 is not required for infection of CD4⁺ T cells,⁸ it has been suggested that the interaction between lentiviral envelope proteins and α4β7 facilitates the migration of virus-infected α4β7⁺/CD4⁺ T cells to GALT and may play an important role during acute infection.^7,9

While the core Asp at position 180 is among the most highly conserved residues in HIV-1 gp120,¹⁰ variation at position 179 and in particular at position 181 has been reported. For example, subtype C gp120s frequently encode LDI, while subtypes B and D gp120s frequently encode LDV.⁷ Recently, He et al.¹¹ have suggested that the conservation of the LDI/LDV tripeptide may influence the epidemic spread of distinct HIV-1 genetic forms in China. It should be noted however that it is not known whether and how the variation at position 181 impacts gp120 binding to α4β7. Position 181 is noteworthy in one other respect. In the RV144 vaccine trial, which showed a protective efficacy of 31.2 %, a sieve analysis of vaccinees who became infected revealed that viruses encoding position 181 residues other than the isoleucine were preferentially transmitted and thus escaped protection responses elicited by the vaccine.^12,13 Variation at position 181 is thus worth further characterization.

HIV-1 genetic forms are heterogeneously disseminated worldwide. Nine pure genetic subtypes and over 70 circulating recombinant forms (CRFs) have been identified to date.¹⁴ Despite such diversity, specific HIV-1 subtypes/CRFs are successfully disseminated in distinct geographic areas relative to others. In South America (SA), HIV-1 subtype B predominates in the AIDS epidemic, but other subtypes such as F1 and C, and BF1 and BC recombinant forms are on the rise.^15–18 In the present study, we have assessed the HIV-1 env V2 α4β7 binding LDI/V tripeptide composition of viruses circulating in south and southeastern Brazil (where over 80% of HIV infections in the country are located). We provide evidence that the LDI tripeptide is expanding across the continent, suggesting a preference for the transmission of viruses encoding an isoleucine at position 181. Analysis of HIV-1 env V2 sequences from other parts of the world retrieved from the Los Alamos Database further corroborates the hypothesis of local hegemony of distinct HIV-1 subtypes carrying the LDI variant.

MATERIALS AND METHODS

Patients and Samples

Samples from 27 HIV-infected infants followed-up since birth at the Pediatric Unit of Universidade Federal do Rio de Janeiro, RJ (southeast Brazil) and from 68 HIV-infected adults followed-up at the HIV/AIDS outpatient clinic of Hospital de Clínicas de Porto Alegre, RS (southern Brazil) were included in this study. Clinical and epidemiological data from patients of both cohorts were available and patients were classified according to likely infection route and disease progression profiles (Supplementary Tables 1 and 2). Vertically-infected children (VIC) were part of a previous study and disease progression patterns were defined using HIV viral load measurements and CD4⁺ T-cell counts according to established criteria.^19–21 Among VIC, six samples were identified as long-term non-progressors (LTNP), with ≥ 8 years of infection and presenting CD4⁺ T-cell counts ≥ 25% without ARV therapy, whereas the remaining were classified as progressors. Among infected adults (IA), 18 subjects have been identified as intravenous drug users (IDU), thus categorized as parenterally-exposed/infected. In this group, nine individuals were classified as LTNP or did not progress to AIDS before 8 years of follow-up (probable LTNP) according to established definitions.²² IAs naïve to treatment at sampling time with less than 10 years of follow-up were categorized as undetermined progressors and were excluded from progression analyses. All patients or child guardians have signed a consent form and agreed to participate to this research, and the study has been approved by the Institutional Review Boards of both University centers involved. The use of previously collected infant samples has been granted by the IRB of UFRJ.

PCR, DNA sequencing and sequence analysis

Blood samples had their peripheral blood mononuclear cells (PBMC) isolated using Ficoll density gradient. PBMC were used for genomic DNA extraction using the QIAGEN Blood Mini Kit (QIAGEN, Chatsworth, CA) according to the manufacturer’s specifications. HIV proviral sequences comprising the HIV-1 gp120 V1–V2 hypervariable regions were PCR-amplified using a semi-nested protocol as described elsewhere.²³ Cycling conditions were the same for the first and second rounds, except for varying the annealing temperature according to the primers used (Supplementary Table 3). DNA sequencing was carried out in an automated ABI 3130XL Genetic Analyzer (Life Technologies) and manually edited with SeqMan v.7.0 (DNASTAR Inc, Madison, U.S.A.). Sequences were then aligned using BioEdit v.7.0.²⁴ Deduced Env amino acid sequences were compared regarding positions 179–181 in HXB2, corresponding to the putative integrin α4β7 binding site at the V2 loop.

Characterization of HIV-1 recombinant forms in samples obtained in this study was limited, since sequencing was performed only for the V1–V2 loops. Therefore, sequences were classified only according to the subtype of V1–V2 hypervariable loop using the HIV Blast online tool (http://www.hiv.lanl.gov/content/sequence/BASIC_BLAST/basic_blast.html). For subtype confirmation, sequences were codon-aligned with HIV-1 subtypes B, C and F reference sequences retrieved from the Los Alamos Database. A Neighbor-Joining tree was constructed using MEGA v.5.0²⁵ and the Kimura-2-parameter model was used. Recombinants remained classified as BC or BF when Blast and phylogenies could not show accurate resolution for V2 subtype and sequences were highly similar to the recombinant forms as suggested by the HIV Blast searches (identity ≥ 80%). HIV-1 subtype definition for the retrieved sequences was the one provided at the database website.

For depicting temporal trends in the prevalence of LDI and LDV-containing HIV-1 sequences, these latter were stratified into 5-year groups (≤ 1980, 1981–1985, 1986–1990, 1991–1995, 1996–2000, 2001–2005 and 2006–2010, or into a small number of groups, depending on sequence availability) according to the collection date depicted in the Los Alamos HIV Sequence Database.

All HIV-1 env V2 nucleotide sequences herein generated have been deposited at the GenBank database and have been assigned the accession numbers KM102250 to KM102339.

Statistical Analyses

Shannon diversity indexes²⁶ were estimated to quantify the degree of diversity within the 179–181 tripeptide domain, using the Shannon Entropy-One online tool available at the Los Alamos Database website (http://www.hiv.lanl.gov/content/sequence/ENTROPY/entropy_one.html). LDI/V and other variant tripeptide distribution and prevalence were compared using Chi-squared and Fisher’s Exact tests in SPSS.

RESULTS

α4β7 mimotope tripeptide variability among HIV forms circulating in South America

We generated V2 sequences of 23 subtype B, three subtype F1 and one BF1 recombinant virus from the VIC group, a ratio consistent with the HIV molecular epidemiology scenario in southeastern Brazil.^27,28 Of those, 23 subjects could be classified with respect to their disease progression profiles. For the IA group, 34 subtype B and an equal number of subtype C sequences has been determined, reflecting the high occurrence of HIV-1 C in southern Brazil compared to the rest of the country.^18,29–31 All subjects in this group were classified according to transmission route and disease progression pattern. There was no specific association between the α4β7-binding LDV/I tripeptide and transmission or progression status. The LDV/I tripeptide composition was also evaluated within individual subtypes (subtype B, n = 57; C, n = 34 and F1, n = 3). Additional SA V2 sequences have been retrieved from Los Alamos (171 subtype B, 12 C, 16 F1, 130 BF and 12 BC) for analysis. All α4β7 binding tripeptides experimentally obtained in our study groups harbored the aspartic acid at position 180, as well a striking majority of SA sequences retrieved from the Los Alamos database, regardless of the viral subtype or transmission route. However, clear differences in the variability of positions 179 and 181 were observed between SA variants. An analysis comprising all SA HIV-1 V2 sequences showed increased variation in subtype B, whilst subtypes C and F1 presented fewer non-LDI/V tripeptide variants (Table 1). BF recombinants also revealed a greater number of tripeptide variants when compared to BC (Table 1). Env amino acid positions 179–181 were assessed for conservancy on the tripeptide mimotope among HIV-1 genetic forms. We observed a higher heterogeneity at position 179 in subtypes B and BFs, whilst subtypes C, F1 and BCs presented lower variability at that position (Supplementary Figure S1A). Valine and isoleucine were the most prevalent residues observed at position 181 (59% and 35%, respectively) in subtype B, while all subtype F1 sequences presented isoleucine at position 181. Subtype C, BC and BF recombinants showed varied prevalence of Ile181 (Supplementary Figure S1B), but all higher than subtype B.

Table 1.

α4β7 binding motif variants found in South American HIV-1 subtypes B, C and F1 and BF and BC recombinants.

Tripeptide variants	Subtype B	%	Tripeptide variants	Subtype C	%	Tripeptide variants	Subtype F1	%	Tripeptide variants	BC	%	Tripeptide variants	BF	%
LDV	122	53.5	LDI	36	78.2	LDI	16	84.1	LDI	7	58.4	LDI	75	57.8
LDI	69	30.3	LDV	4	8.7	PDI	1	5.3	LDV	3	25	LDV	32	24.6
LDL	10	4.4	undefined	2	4.3	VDI	1	5.3	IDV	1	8.3	LDL	7	5.4
SDV	3	1.3	IDI	1	2.2	LGI	1	5.3	PDI	1	8.3	IDI	2	1.5
TDV	3	1.3	VDI	1	2.2	Others*	0	0	Others*	0	0	IDM	2	1.5
IDI	3	1.3	LDM	1	2.2							Others*	12	9.2
Others*	18	7.9	TDV	1	2.2

^*Others include all variants with a frequency below 1% for a given subtype/recombinant

The Shannon diversity of positions 179–181 was calculated for all HIV-1 subtypes/recombinants. A higher heterogeneity degree was observed at positions 179 and 181 in subtype B and BF recombinants, whereas subtypes C and F1 showed higher conservation at those positions (Table 2).

Table 2.

Shannon entropy indexes of HIV-1 subtypes and recombinant forms circulating in South America. Data includes sequences from our study groups and from the Los Alamos HIV Sequence Database. Diversity was significantly higher in HIV-1B and BF recombinant forms for positions 179 and 181 (indicated in bold). Lower variability was observed in position 180, explained by the aspartic core maintenance.

Subtype/Recomb.Form	Shannon Entropy Index
	AA 179	AA 180	AA 181
B	0.595	0.056	0.908
C	0.313	0.000	0.619
F1	0.409	0.206	0.000
BC	0.566	0.000	0.637
BF	0.595	0.000	0.939

LDI favors dissemination of HIV-1 non-subtype B forms in South America

As shown above, the proportion of LDV/I was equivalent for subtypes B, C, F1 and BC and BF recombinants, even when data retrieved from Los Alamos Databases was included. No statistical significance in LDV/I distribution was observed for any particular HIV-1 genetic form, indicating that both LDV and LDI are maintained as potential α4β7 integrin putative binding mimotopes (Supplementary Figure S2). Notwithstanding, there was a clear difference in LDI prevalence over LDV when both tripeptides were analyzed separately. LDI variants were markedly prevalent in the V2 loop of subtypes C and F1, but not in subtype B, which showed a higher prevalence of LDV (Figure 1A). The addition of V2 sequences retrieved from the Los Alamos data base corroborated our results, evidencing a high prevalence of the LDI in subtypes C, F, BF and BC, but not in subtype B (Figure 1B). Of note, epidemiological data suggests that all those HIV-1 non-B forms are on the rise in SA (except for pure subtype F1, which is thought to be extinct on the continent), suggesting that the LDI tripeptide composition may promote HIV dissemination of those variants in the continent.

Figure 1.

α4β7 mimotope tripeptide distribution among HIV-1 V2 sequences from South America. (A) HIV-1 subtypes B and C in pediatric and adult samples (study groups). Because the number of V2 F1 subtypes was limited in our dataset (n = 3), it is not shown. (B) HIV-1 subtypes B, C, F1, BC and BF from South America (including sequences retrieved from the Los Alamos HIV Sequence Database).

Prevalence of LDI and dissemination of HIV-1 subtypes across the globe

Our data from SA suggested that the LDI tripeptide variation could be linked to a more successful HIV epidemic dissemination. We wanted to further test this hypothesis by analyzing HIV-1 V2 tripeptide from several other parts of the world. Viral env V2 loop sequences across continents were retrieved from the Los Alamos Database, and we assessed conservation and prevalence of the LDI α4β7-binding mimotope in HIV-1 forms undergoing expansion is various locales, compared to sequences from subtypes previously established at those locations.

Based on data reported by He et al.¹¹ in China, we verified that LDI accounts for 94% of α4β7 binding domains in the fastest disseminating HIV-1 CRF07_BC viruses. We also verified a high prevalence of LDI in the spreading CRF01_AE and CRF08_BC forms (69.5% and 84%, respectively; Table 3). In contrast, only 45% of subtypes B viruses carried LDI, while the majority encoded LDV or other variants.

Table 3.

α4β7 tripeptide mimotope distribution among worldwide HIV-1 V2 sequences retrieved from the Los Alamos HIV Sequence Database

SUBTYPE	TRIPEPTIDE	TOTAL SEQS	%
FORMER USSR
A/A1 (145)*	LDI	141	97.2
	LDV	1	0.7
	others	3	2.1
B (24)	LDI	7	29.2
	LDV	7	292
	others	10	41.7
SOUTHEAST ASIA
CRF01_AE (465)	LDI	357	76.8
	LDV	8	1.7
	others	100	21.5
B (43)	LDI	6	14
	LDV	24	55.8
	others	13	30.2
CHINA
CRF01_AE (282)	LDI	196	69.5
	LDV	8	2.8
	others	78	27.7
CRF07_BC (341)	LDI	321	94.1
	LDV	13	3.8
	others	7	2.1
CRF08_BC (62)	LDI	52	83.9
	LDV	1	1.6
	others	9	14.5
B (229)	LDI	103	45
	LDV	79	34.4
	others	47	20.5
SOUTH AFRICA
C (744)	LDI	311	41.8
	LDV	48	6.4
	others	385	51.8
A/A1 (4)	LDI	0	0
	LDV	3	75
	others	1	25
B (13)	LDI	3	23.1
	LDV	8	61.5
	others	2	15.4
WESTERN AFRICA
CRF02_AG (80)	LDI	11	13.8
	LDV	39	48.8
	others	30	37.5
EAST AFRICA
A/A1 (303)	LDI	125	41.2
	LDV	146	48.2
	others	32	10.6
C (93)	LDI	62	66.7
	LDV	15	16.1
	others	16	17.2
D (40)	LDI	3	7.5
	LDV	31	77.5
	others	6	15
CENTRAL AFRICA
CRF02_AG (56)	LDI	7	12.5
	LDV	21	37.5
	others	28	50
CRF11_cpx (30)	LDI	13	43.3
	LDV	14	46.7
	others	3	10

^*Numbers in parentheses refer to the number of sequences analyzed for each given HIV subtype and location.

In Southeast Asia, we observed a striking prevalence of the recombinant form CRF01_AE, consistent with recent literature.³² Of 601 HIV-1 V2 sequences originated from those countries, 465 (77%) were CRF01_AE and 77% of them presented LDI (Table 3). In contrast, LDI prevalence was only 14% among subtype B sequences, lower than that of LDV (56%).

In Western Europe, mounting evidence of the rise of non-B HIV-1 subtypes, including CRF01_AE and subtypes C, F1 and G has been published.^33–39 The estimated distribution of LDI in these HIV-1 forms reaches 59–89%. Taking together, LDI prevails among emerging HIV subtypes in Europe in areas previously marked by subtype B epidemics.

In the former Soviet Union (FSU), introduction of HIV started during the 80’s among MSM infected with subtype B, determining the hegemony of this subtype for over a decade.^40,41 Currently, subtype A/A1 accounts for 71% of the viruses in the region and its LDI prevalence is 97%, while LDV accounts for only 0.7% (Table 3). Subtype B sequences from FSU, although very limited, showed equivalent rates of LDI and LDV (approximately 30% each).

Since sub-Saharan Africa presents an extremely diverse distribution of HIV variants,⁴² we focused our evaluation of the LDI distribution in HIV-1 forms according to their predominance in the South/East and West/Central regions. In the former, HIV-1C accounts for over 95% of infections.⁴³ In South Africa, LDI accounted for 42% of HIV-1C sequences, whereas LDV prevalence was only 6% (Table 3). In East Africa, recent surveys showed dissemination of subtype A/A1 in Kenya, Uganda and Rwanda. In Rwanda, subtype A/A1 already accounts for 95% of HIV epidemics.^44–46 Unexpectedly, HIV-1A LDI prevalence was slightly below that of LDV (41% versus 48%). HIV-1C showed a striking prevalence of LDI over LDV, while HIV-1D evidenced distinctly larger distribution of LDV (78%) compared to LDI (7%) (Table 3).

In West/Central Africa, all HIV-1 forms co-circulate.⁴⁷ Despite such diversity, CRF02_AG has recently presented evident dissemination, spreading faster than other HIV variants in Cameroon, Senegal and Ghana.^48,49 V2 sequences from CRF02_AG, which belong to subtype A, were separately analyzed in West and Central Africa. For western African countries, CRF02_AG accounted for 58%. LDI prevalence in CRF02_AG from this region was low (14%) compared to LDV (49%). In Cameroon and Central African Republic, CRF02_AG contributed to 19% of all infections, and the LDI frequency was similar to that of Western Africa. In contrast, the LDI frequency in CRF11_cpx, currently expanding in Cameroon and in the Central African Republic, revealed a higher frequency of LDI (Table 3).

LDI outgrows LDV over time in the epidemics

The low prevalence of LDI among HIV-1A/A1 sequences in East Africa and CRF02_AG sequences from West/Africa, where these respective genetic forms predominate, was unexpected. Moreover, even with the increasing prevalence of non-B subtypes in North and South America, as well as in Western Europe, HIV-1B still predominates in the epidemic, with lower LDV prevalence. One alternative explanation for those observations could be that a shift of LDI over LDV over time is still going on, while the retrieved sequences from the Los Alamos database represent only a snapshot that includes older sequences. To evaluate this scenario, we analyzed LDI/LDV prevalence over time in those geographic areas using the sample collection dates available in the database. HIV-1A/A1 sequences of East Africa, from 1990 to 2000, evidenced a higher prevalence of LDV (71%) over LDI (29%). In contrast, LDI sequences showed elevated prevalence (69%) over LDV forms in this location from 2000 to 2010 (Table 4). In West and Central Africa, although LDV maintained high prevalence over LDI HIV CRF02_AG, increased frequencies of LDI were observed after 2000 (24% and 27%, respectively) compared to sequences sampled from 1990–2000 (19 and 17%, respectively) (Supplementary Table 4). Most strikingly, when analyzing data from the Americas and Western Europe, from where a great number of sequences are available, a very clear trend for an increase in time of LDI over LDV in HIV-1B is observed (Figure 2). Taken altogether, our data suggest that although LDV is still predominant in some areas, it is being replaced by LDI over time likely due to increased transmission potential of viruses encoding the latter.

Figure 2.

Line trends of LDI increase and LDV decrease over time in the HIV/AIDS epidemics among HIV-1 subtype B sequences retrieved from the Los Alamos HIV Sequence Database isolated and characterized in North America (A), South America (B) and Western Europe (C). The asterisks in all three panels refer to merging of periods 2006-2010 and >2010, due to low numbers of sequences deposited in the Los Alamos Database after 2010.

DISCUSSION

In the present study, we assessed the HIV-1 env V2 region of viral subtypes circulating in South America, focusing on the recently described α4β7 integrin binding LDI/V tripeptide. We show that LDI either already predominates or is expanding over LDV. We provide evidence that similar trends may be occurring across the globe in various HIV-1 subtypes/CRFs.

It has been consistently shown that the HIV envelope undergoes different types of selective pressures at mucosal transmission versus during the course of infection. Mucosal transmission is inefficient, and in order to reach target cells, the virus is required to break through the physical barriers of the mucosa and enter target cells that can support productive infection. This represents the first selective pressure in sexual transmission.^50–54 As a consequence, the productive infection that follows primary contact generally reflects an expansion from a single founder virus with a particular phenotype.⁵⁵ Although env selection has been largely investigated in different infection routes in an isolated manner, little is known about the way in which viral signatures of transmission are related to different routes of transmission. Nor is it understood why in many parts of the world HIV-1 subtype C appears to be replacing other subtypes. HIV-1C, which typically encodes LDI, is the prevalent subtype in the global AIDS epidemic and is increasing in South America. Moreover, it has been suggested that the interaction between gp120 and α4β7 may facilitate mucosal transmission, which is consistent with the observation that α4β7high/CCR5high activated CD4⁺ T-cells are the lymphocyte subset preferentially infected by HIV.⁵⁶ We speculate that some early transmitting viral gp120s bear structural features that exploit α4β7 interactions allowing them to gain access to the GALT, where cells that are highly susceptible to productive infection are readily available. We wonder whether the LDI tripeptide confers greater ability to interact with the α4β7 integrin, and this might explain the high prevalence of LDI in SA HIV-1C V2 sequences and also in other disseminating non-B HIV-1 forms in SA despite the observation that V2-α4β7 interaction does not appear essential for viral infectivity per se.⁵⁷

To further test our hypothesis of an increased epidemic potential provided by the LDI tripeptide observed in SA, we expanded our analysis to gp120 sequences from other areas of the world. We found evidence suggesting that HIV forms carrying the LDI tripeptide variant are expanding elsewhere. For instance, CRF01_AE, which originated in Central Africa and predominantly encodes LDI, has disseminated to countries such as Thailand, Vietnam, Malaysia, Cambodia, Indonesia and Singapore, where previously subtype B, which typically encodes LDV, predominated in the early years of the AIDS epidemic.^58,59 Similarly, subtype B predominated in Western Europe,^38,60,61 but in recent years, there has been a significant expansion of non-B subtypes in that area.^38,61 We should note however that this expansion may be explained in part by the arrival of African immigrants.⁶² Recent epidemiological studies revealed CRF01_AE expansion in countries such as Cyprus, France, Italy, Greece and Spain,^33,34 subtype F1 in Romania³⁶ and subtype G in the Iberian Peninsula,³⁴ all showing noticeable prevalence of the LDI α4β7 binding tripeptide. Likewise, the AIDS epidemic in FSU turned to be represented by LDI prevalent subtypes A/A1, showing greater distribution than the previously dominant form CRF03_AB. The HIV epidemics in Central/East Africa reveals a more diverse scenario, with the introduction of subtypes A/A and D.⁴² While the first showed a great prevalence of the LDI V2 tripeptide, the second displays a majority of LDV-carrying strains. HIV-1D revealed to be dominant and basically restricted to Uganda and Kenya. Similar findings were observed for CRF02_AG in West/Central Africa. Interestingly, however, a temporal analysis of sequences showed that in most of those cases, LDV is being outgrown by LDI, consistent with a scenario of epidemic spread advantage conferred by the latter tripeptide. Even more strikingly, the same scenario of outgrowth of LDI over LDV across the AIDS epidemic chronology was corroborated among HIV-1B sequences for the Americas and Western Europe, from where a large number of HIV-1 sequences were available.

The idea that an enhanced binding of LDI to α4β7 compared to LDV came from the original observation by Arthos et al.⁷ that HIV-1C gp120 had a better binding when compared to gp120 derived from HIV-1A and B in an in vitro binding assay (see Fig. 1 of that paper) and that HIV-1C had an increased prevalence of LDI when compared to other subtypes (see Suppl. Fig 8 of the paper). However, one should note that the currently available assay to measure the binding of gp120 to α4β7 suffers from several limitations, which would make the conclusion that LDI should bind stronger to α4β7 just too simplistic. Firstly, the assay requires the α4β7 to be in a high state of activation through the use of Mn²⁺, a state unlikely to be predominantly present physiologically. Secondly, HIV-1 gp120 binds to the target cell as a trimer, a condition not reproduced in the assay, which is made with monomeric proteins. Finally, more recent work by the group of Dr. Arthos has pointed out that additional, underlying sequences in gp120 V2 and in V1 influence binding to α4β7,⁶³ and it is unlikely that the LDI/V tripeptide is the unique determinant to this binding. Other unexplored features of the gp120/α4β7 binding, such as binding kinetics itself and the signal transduction that follows that interaction, are also not measured in this simple assay. Overall, the currently available assay may not reflect the net impact of each individual tripeptide variant in HIV transmission, and most importantly its dissemination at an epidemic level.

To further test a possible predominance of LDI among recent HIV infections, we have assessed the V2 tripeptide composition among more than 5,000 transmitted/founder viruses published recently^64–66 (Supplementary Fig. S3). Rather than such predominance, we found compositions that resemble the HIV-1 subtypes involved (e.g., a predominance of LDI in HIV-1C, and of LDV in HIV-1B), further supporting the hypothesis that additional determinants in Env are important for HIV transmission.

Among the putative explanations for a transition from LDV to LDI is the possibility that as various regions of gp120 have evolved, immune pressure has led to this transition. Such an explanation must however take into account the fact that recent structural analyses of near-native trimmers indicate that, in the fully unliganded state, the LDV/I is not exposed⁶⁷ and thus unlikely to be subject to immune pressure. However, the envelope structure is dynamic and variable, and therefore we cannot rule this possibility out entirely. A second consideration arguing against immune pressure directed at LDV/I is the potential for autoreactivity, insofar as either motif is likely to be a structural mimic of epitopes presented by human MAdCAM, VCAM and fibronectin. In this respect, the human immune system may recognize LDV/I as self, which would lead to the deletion of immune responses targeting this motif.

In conclusion, we provide in this report evidence that the LDI epitope is expanding over time and across the globe. This is noteworthy because this tripeptide is involved in the interaction with α4β7, which may be linked to increased transmission fitness. Moreover, the LDI epitope has been associated with the molecular sieving in the RV114 trial.¹² Although much remains to be learned about the role of the LDI/LDV epitope in HIV transmission and pathogenesis, the data presented in this report further reinforces the necessity of further investigating this important this HIV-1 env V2 tripeptide.

Supplementary Material

Supplemental Digital Content

Figure S1. Amino acid composition at positions 179 (A) and 181 (B) of HIV-1 gp120 sequences of different subtypes and recombinant forms found in South America retrieved from the Los Alamos HIV Sequence Database.

Figure S2. Distribution of the LDI/LDV tripeptides present in HIV-1 gp120 sequences of different subtypes and recombinant forms found in South America and analyzed in the current study.

Figure S3. Distribution of the LDI/V tripeptides among transmitted/founder HIV-1 sequences of distinct subtypes retrieved from the large studies: (A) Keele et al.⁶⁴, (B) Sterrett et al.⁶⁵ and (C) Sanchez et al.⁶⁶