healthmedicinet

Identification of dual tropic HIV-1 isolates from patient samples

Two methods were employed to isolate HIV-1 from patient samples and then screen for
viral tropism. The first method involved initial co-cultivation of PBMCs from an HIV-infected
patient with PBMCs from an HIV-negative donor (Fig. 1a). These standard co-cultivations were performed on a set of patient samples from
Zimbabwe as previously described 40]. This initial analysis screened over 28 HIV-1 isolates and found 10 that were syncytium
inducing (SI) and capable of replicating on MT2 cells 46]. These primary SI HIV-1 were then equalized for RT activity (see “Methods”) and added
to U87.CD4 cells expressing either CCR5 or CXCR4 40] (Fig. 1b). Two (C19 and C27) of 10 primary HIV-1 isolates were dual tropic, i.e. capable
of infecting and replicating in U87.CD4 cells expressing either CXCR4 or CCR5 as a
co-receptor. HIV-1 C23 showed CCR5 (r5) just above the limit of detection whereas
the remaining strains in this cohort were CXCR4 (x4) as previously described.

Fig. 1. Identification of dual/mixed isolates following HIV-1 propagation or construction
of env chimeric viruses. a HIV-1 subtype C isolates from Zimbabwe were propagated by PBMC co-cultivation as
described 46] and in the “Methods”. c The HIV-1 env coding region of HIV-1 subtype B infected patients was PCR amplified
and then cloned into pREC_nfl_HIV-1
_l4-3
?gp120/URA3 via yeast recombination/gap repair 47]. The propagated HIV-1 isolates (a) and cloned env genes expressed following 293T transfection (c) were screened for co-receptor usage by infection(1) (b) or cell-to-cell fusion (d) (Veritrop; 48]) on U87.CD4 cells expressing either CXCR4 or CCR5(2). In the virus infection system
(b), virus production was monitored by RT activity in supernatant 87]. In the cell-to-cell fusion assay (d), the ability of pREC_nfl_env
_ptX
to modulate receptor binding and cell fusion was monitored by firefly luciferase activity,
i.e. Rev/Tat in the effector cell controlling Luc expression under the control of
the HIV-1 LTR and rescued via the RRE housed in an intron with Luc 80]. Two primary HIV-1 isolates and two env chimeric viruses were propagated to measure
infectious titers on PBMCs (Additional file 1: Table S1), U87.CD4.CCR5 and U87.CD4.CXCR4 cells (e) using the classical Reed–Munch approach 79].

As part of another study that screened for co-receptor usage, HIV-1 env chimeric viruses
were produced by homologous recombination in yeast rather than propagating HIV-1 from
patient samples (Fig. 1c). The yeast-based recombination/gap repair cloning method can efficiently replace
the yeast URA3 gene with the gp120 coding region of the env gene in our pREC_nfl_HIV-1
_NL4-3
?gp120/URA3 vector (see “Methods”) 44], 47]. The efficiency of this cloning method limits the genetic bottleneck introduced by
prolonged PBMC co-cultivation and subsequent virus propagations. Following env cloning,
the pREC_nfl_env
_Ptx
vectors from each patient sample (Ptx—patient x) were purified from pooled yeast colonies
(1,000) and transfected into 293T cells. We have recently described an efficient
cell-to-cell fusion assay to screen for co-receptor usage using the U87.CD4.CXCR4
or U87.CD4.CCR5 as target cells 48]. In our preliminary screen of 11 patient-derived env chimeric viruses, we identified
three dual tropic HIV Envs (B9, B12, and B19), five r5s, and two x4s (Fig. 1d). The pREC_nfl_env vectors of B9, B12, and B19 were co-transfected with the complementing
vector (pCMV_cplt) to produce replication-competent chimeric HIV as described 47]. This virus was then used to infect U87.CD4.CXCR4 and U87.CD4.CCR5 cells to confirm
co-receptor usage. HIV-1 env_B9 did not replicate in either cell line.

The x4 and r5 titers of each dual tropic primary and dual tropic env chimeric HIV-1
isolate along with a set of reference HIV-1 isolates were measured by standard TCID
₅₀
assays on U87.CD4.CXCR4 and U87.CD4.CCR5 cells, respectively (Fig. 1e). The dual tropic C19 and C27 viruses had equal infectious r5 and x4 titers whereas
the dual tropic env_B12 and env_B19 viruses had higher r5 than x4 titers. Interestingly,
the infectious titers derived from TCID
₅₀
assays using HIV-negative PBMCs was the same as the highest infectious titers from
either U87.CD4.CXCR4 or U87.CD4.CCR5 cells. For at least the C19 and C27 viruses,
the IU/ml on PBMCs did not reflect the simple addition of r5 and x4 infectious units
suggesting that the clones within these dual tropic HIV-1 isolates may be primarily
dual tropic rather than a mix of r5- and x4-only HIV-1 clones (Additional file 1: Table S1).

Phylogenetic analyses of the dm HIV-1 isolates and env chimeric viruses

Alignment of the env sequences confirmed that the bulk populations of the dual tropic
C19 and C27 were subtype C whereas dual tropic env_B12 and env_B19 aligned with other
subtype B HIV-1 sequences (Fig. 2a). In addition, the B12, B19, C19, and C27 viruses were not clonal but rather a population
of related env sequences, based on approximately 10–15 sequenced clones from each
(Fig. 2b–e).

Fig. 2. Neighbor joining phylogenic trees of two dual tropic HIV-1 isolates, two HIV-1 env
chimeric viruses, and of the sampled clones in their virus population. Alignments
and phylogenetic trees were constructed for the C2-V3 region (350 nt) of the four
reference strains, r5 B2, r5 C3, x4 A8, and x4 E6, the two dual tropic HIV-1 isolates
(C19 and C27), the two HIV-1 env chimeric viruses (B12 and B19) and set of reference
strains (a). As described in the “Methods”, the Env coding region of two dual/mixed (dm) HIV-1
isolates were PCR amplified and cloned into pREC env via yeast-based recombination.
The C2-V3 sequences from 21 B19 (b), 18 B19 (c), 26 C19 (d), 21 © clones (e) were aligned using MUSCLE and then schematic represented in phylogenetic trees.

Contribution of x4 and r5 variants to replicative fitness in dual tropic and mixed
r5x4 primary HIV-1 isolate

In a previous study, we performed two pairwise competition experiments in PBMC cultures
with 14 primary r5 and 15 x4 HIV-1 isolates 40]. The phylogenetic relationships of these are shown in Fig. 2. To summarize, both the r5 and x4 HIV-1 subtype C isolates were less fit than the
r5 and x4 HIV-1 isolates (respectively) of different group M subtypes. The poor replicative
fitness of subtype C isolates has now been observed in over 2,000 head-to-head competitions
with over 40 primary subtype C HIV-1 isolates 38]–40], 49]. Although ten x4 HIV-1 subtype C isolates were dramatically less fit than any other
x4 HIV-1 isolate, we also noted that two isolates, C19 and C27 were of a dual tropic
phenotype and capable of infecting both CD4+/CCR5+ and CD4+/CXCR4+ susceptible cells
40]. When competing r5 against x4 HIV-1 isolates in PBMCs, the x4 viruses, regardless
of subtype, had higher replicative fitness than an r5 HIV-1 isolate 40]. Thus, it is possible that reduced fitness of dual tropic C19 and C27 HIV-1 isolates
may have been related to high proportion of r5 HIV-1 clones within these viruses.

To determine the relative contribution of the CCR5-using and CXCR4-using phenotypes
in these dual tropic r5x4 HIV-1 isolates, the r5x4 C19 and C27 HIV-1 isolates were
competed against four reference HIV-1 isolates, the r5-using B2 and C3 primary isolates
and the x4-using A8 and E6 viruses (Fig. 1e). The control x4 A8 virus is characterized as having a high replicative fitness
whereas the x4 E6 is of lower replicative fitness when compared in direct competition
with other x4 isolates 39], 40]. Our control r5 B2 virus was also selected based on a higher replicative fitness
as compared to other r5 HIV-1 isolates whereas r5 C3 has low replicative fitness 39], 40]. First, the titers of dual tropic C19 and C27 viruses as well as control x4 A8 and
E6 HIV-1 were calculated using both PBMCs (“full” titer) and U87.CD4.CXCR4 (“x4” titer)
for TCID
₅₀
assays (Additional file 1: Table S1).

Direct competitions showed that the dual tropic C19 and C27 HIV-1 were completely
outcompeted by A8 but still could compete with the x4 E6 HIV-1 (Fig. 3a), when virus inoculums were equalized based on PBMC titers. When inoculums were
equalized based on x4 titers, the fitness of dual tropic C19 and C27 further decreased
in competition with x4 E6 (Fig. 3a). The dual tropic C19 and C27 were then competed in PBMCs against the r5 B2 and
C3 control viruses using the equalized “full” titers determined on PBMCs or the equalized
“r5” titers determined on U87.CD4.CCR5. As observed with nearly all x4-using viruses,
the dual tropic C19 and C27 HIV-1 isolates were more fit than the r5 C3 virus when
inoculating with the PBMC titers (Fig. 3b). However, dual tropic C19 and C27 were only slightly more fit than the r5 B2 HIV-1.
The “CXCR4”-using component of the dual tropic HIV-1 isolates are not necessarily
in competition with these r5 HIV-1 in PBMC since they likely infect different susceptible
cells in the PBMC population. Competitions with the control r5 B2 HIV-1 suggest a
low frequency of an r5 phenotype in the dual tropic C19 and C27 viruses. When using
U87.CD4.CCR5-derived titers, it was not surprising that C19 and C27 could further
out replicate the r5 B2 and C3 viruses (Fig. 3b) in competition considering the fivefold increase in C19 and C27 inocula to equalize
their “r5” titer with the titer of pure r5 B2 and C3 viruses.

Fig. 3. Relative replicative fitness of the dual tropic C19 and C27 HIV-1 when competed against
reference strains on PBMCs. a The dual tropic HIV-1 isolates, C19 and C27 were competed against the x4 reference
strains, A8 and E6 on PBMC using equal MOIs (0.004) of each virus tittered on U87.CD4.CXCR4
cells or PBMCs (same donor and blood draw as those used in competition). b A similar set of competitions on PBMCs involved the same dm viruses competed against
the r5 reference strains, B2 and C3 using equal MOIs tittered on U87.CD4.CCR5 cells
or PBMCs. Fitness difference (or ratio) is shown with all competitions where the relative
fitness of the dual tropic HIV-1 isolate is plotted as a ratio to the relative fitness
of the reference virus in the competition. The titers of all dual tropic and reference
HIV-1 isolates are shown in Additional file 1: Table S1.

In the competitions performed in PBMC cultures with PBMC-derived titers, C19 and C27
could replicate and compete with either the r5 (B2 and C3) or the x4 (A8 and E6) viruses
(Fig. 3). In Fig. 4, we performed the same competitions as in Fig. 3 but prevented replication of either the “CXCR4”-using component (panel A) or the
“CCR5”-using component (panel C) of the dual tropic C19 and C27 viruses. First, the
C19+A8, C19+E6, C27+A8, and C27+E6 dual infections were added in equal “x4” titers
(determined on U87.CD4.CXCR4 cells) to PBMC cultures in the presence of high maraviroc
(MVC) concentrations. As described in Fig. 4b, the IC
₉₉
concentration of MVC, a CCR5 antagonist 50], 51], resulted in complete inhibition of the r5 B2 and C3 viruses but no apparent inhibition
of C19 and C27 in PBMCs. Lack of C19 and C27 inhibition by MVC suggests that the majority
of the HIV-1 clones in these quasispecies may be dual tropic (r5/x4) and fully capable
of infecting CXCR4+ cells in the PBMC cultures such that MVC does not inhibit any
virus replication. The composition of the quaispecies in terms of co-receptor usage
is described below (Fig. 1). Interestingly, there was no significant difference in the replicative fitness of
C19 or C27 (versus control x4 A8 or E6 HIV-1) when comparing results of competitions
performed in PBMCs, PBMCs + MVC, or in U87.CD4.CXCR4 cells (Fig. 4a). The latter condition would only support replication of the control x4 viruses
and the “CXCR4”-using component of the dual tropic viruses. The reference viruses
A8 and E6 were significantly more fit than the “CXCR4”-using component of the dual
tropic C19 and C27 viruses regardless of the above conditions (Fig. 3a).

Fig. 4. Fitness of the dual tropic HIV-1 isolate when blocking the virus component using CCR5
or CXCR4. a The primary dual tropic HIV-1 isolates were competed against the x4 50] titereds (all titered on U87.CD4.CXCR4 cells) in U87.CD4.CXCR4 cultures or in PHA-stimulated,
IL-2 treated PBMC cultures with or without maraviroc. The fitness differences are
plotted as described in Fig. 2 and in the “Methods”. b The level of maraviroc inhibition of the monoinfections or dual infections was measured
by relative RT activity in the cell free supernatant and plotted as a percentage of
the no drug control. c The primary dual tropic HIV-1 isolates and r5 reference strains were also titered
on U87.CD4.CCR5 cells and then competed together in PHA-stimulated, IL-2 treated PBMC
cultures with or without AMD3100. The same dual infection with the same MOI (0.004)
was repeated in U87.CD4.CCR5 cultures. The fitness differences are plotted as described
in Fig. 2 and “Methods”. d The level of AMD3100 inhibition of the monoinfections or dual infections was measured
by relative RT activity in the cell free supernatant and plotted as a percentage of
the no drug control. All experiments were performed in triplicate with the exception
of c, d (performed in duplicate) due to limited supply of AMD3100.

To limit replication to only the “CCR5”-using component of the dual tropic viruses,
we performed competitions in PBMCs + AMD3100 or in U87.CD4.CCR5 cell cultures with
the “r5” titers of these viruses. AMD3100, a CXCR4 antagonist 52], 53], showed strong inhibition of x4 A8 and E6 viruses as well as the dual tropic C19
and C27 viruses whereas the r5 B2 and C3 viruses were largely unaffected (Fig. 4d). Reduction in B2 and C3 replication, as compared to no-drug controls, relates to
the cellular toxicity from high AMD3100 concentrations, i.e. required for complete
inhibition of x4 virus replication but this did not have an effect on the competition
experiments as later shown with the B12 and B19 viruses. Whereas the dual tropic C19
and C27 could easily out replicate the r5 B2 and C3 viruses in PBMC cultures, the
“CCR5”-using component of these viruses were significantly less fit than B2 and C3
(Fig. 4c). This data is consistent with the relatively weak fitness of r5 subtype C HIV-1
isolates 38]–40], 49]. Again these were based only on the r5 phenotype present as r5 or r5/x4 clones in
the quasispecies in C19 and C27. Thus, use of AMD3100 with PBMCs or U87.CD4.CCR5 cells
would limit replication to only the CCR5-using and r5/x4 components on CCR5+ cells.
In PBMCs (without drug) where the “CXCR4” component was not inhibited, the C19 and
C27 were 50- to 100-fold more fit than the r5 B2 and C3 controls. After eliminating
replication of the pure CXCR4-component with AMD3100, the r5 competitors, B2 and C3
were now significantly more fit than the “CCR5”-using phenotype of the dual tropic
C19 and C27 viruses (Fig. 4c). The 10,000-fold shift in fitness when inhibiting the “CXCR4”-using component of
the primary C19 and C27 isolates (Fig. 4c) would suggest that the majority of these dual mixed virus populations were comprised
of dual tropic clones (e.g. clones that use either CXCR4 or CCR5 for entry, thus r5x4)
and that the “CXCR4”-using phenotype (in absence of inhibition) was dominant in determining
fitness.

Co-receptor usage of the C19 and C27 quasispecies

The predicted amino acid sequences of the V3 loop of each C19 and C27 clone (Fig. 2) were aligned and then used in various algorithms to predict co-receptor usage. Based
on the C27 consensus sequence, five “groups” of C19 clones were identified based on
a unique V3 loop amino acid profile (Additional file 1: Fig. S1). There were three “groups” of clones in the C27 isolate. The five C19 groups
all had Ser and Lys at position 11 and 25 and were predicted x4 by the 11/25 rule
(positive charge at either or both sites) whereas two neutral amino acids, G11 and
A25 predicted a pure r5 phenotype in all of the C27 clones (Additional file 1: Fig. S1). Using the PSSM 54], Geno2Pheno algorithms 25], and PhenoSeq 55], all C19 and C27 clones were predicted to use at least CXCR4 for entry and that CXCR4-using
clones may dominate the virus population (PSSM predicted a possible dual tropism)
(Additional file 1: Fig. S1).

The predicted co-receptor usage was then compared to actual co-receptor usage of the
HIV-1 clones from C19 and C27. The HIV-1 Env genes sequenced in Fig. 2a and used for co-receptor prediction in Additional file 1: Fig. S1, were introduced into pREC_nfl_HIV-1
_l4-3
?env/URA3 via yeast-based recombination and described above 42]. This Env expression vector was then used to pseudotype virus produced from 293T
cells co-transfected with the pNL luc-AM vector 56]. Equal virus titers (based on RT activity) were used to infect U87.CD4 cells expressing
either CCR5 or CXCR4. Three (C19-14, C19-30, and C19-29) of 26 C19 clones and 2 of
20 C27 env clones were non-functional. Of the functional clones, none of the C19 or
C27 clones were purely CCR5 tropic whereas several clones were “pure” CXCR4-using,
i.e. C19-17, -28, -10, -25 (Fig. 5c) and C27-7, -21, -8, -26 (Fig. 5d).

Fig. 5. Determining the co-receptor usage of C19 and C27 Env clones introduced into HIV-1
NL4-3 backbone via yeast-based recombination. The 26 C19 and 21 C27 env regions (same
as those sequenced for Fig. 1) were cloned into pREC_env/URA3 via yeast-based recombination/gap repair 81], 82]. A 132 and 118 amino acid sequence of the C2-V3 region was used to group the C19
(a) and C27 (b) based on sequence identity/difference. The pREC_env expression vector was then used
to pseudotype virus produced from 293T cells co-transfected with the pNL luc-AM vector
56]. Equal virus titers (based on RT activity) were used to infect U87.CD4 cells expressing
either CCR5 or CXCR4. The infecting virus is reverse transcribed and integrated to
express luciferase but is incapable of subsequent rounds of infection 56]. Luciferase activity (relative light units) was measured from lysed cells collected
72 h post infection with both the C19 (c) and C27 (d) Env pseudotyped virus infections. All assays were performed in triplicate. Background
was subtracted from results and dotted lines on c, d represent the value three times the standard deviation of the background.

For these analyses, we sequenced approximately 500 nt for each clone and identified
15 unique C19 Env sequence patterns, five of these were described as “groups” I–V
(Fig. 5a). Eleven unique sequences and four “groups” (I–IV) were identified in the C27 HIV-1
isolates (Fig. 5b). In general, the clones within the same group (sharing the same C2-V3 nt sequence)
showed similar co-receptor usage profiles (compare Fig. 5a with c; b with d). However, this was not always the case. For example in group I,
C19-17 and C19-28 appeared to be pure x4 tropic whereas clones C19-11, -20, and -24
could infect both CCR5 and CXCR4 expressing cells (r5x4). It is important to note
that clones that share sequence identity in the C2-V3 region may still differ in the
remaining ~2,200 nt of Env (not sequenced), specifically in the V1V2 region which
has been shown to influence co-receptor usage. Nonetheless, similar co-receptor usage
was observed for most clones that share at least the C2-V3 sequences.

Finally, we compared five assays to determine co-receptor usage of these primary isolates:
(1) the relative infection by primary C19 and C27 HIV-1 isolate on U87.CD4.CCR5 cells
and U87.CD4.CXCR4 cells (Fig. 1c, d), (2) the TCID
₅₀
values derived on CCR5+ and CXCR4+ cells, (3) the relative inhibition by AMD3100 and
MVC on PBMCs (Fig. 4b, d), (4) predicted co-receptor usage from the clones (Additional file 1: Fig. S1), and (5) the actual co-receptor usage of Env clones in the quasispecies
(Fig. 5c, d). For these two dual tropic HIV-1 isolates, the vast majority of clones in the
quasispecies were used both R5 and X4 co-receptors. As discussed below, the use of
co-receptor antagonist coupled with the TCID
₅₀
measurements on U87.CD4.CCR5 cells and U87.CD4.CXCR4 cells provides the best prediction
of the co-receptor usage within the HIV-1 quasispecies. The C19 and C27 primary HIV-1
isolates achieved a dual/mixed phenotype through similar quasispecies compositions.
Both had more clones using both co-receptors, very few using only CXCR4, and no pure
r5 clones (Fig. 5c, d). As described in our fitness analyses, the CXCR4 usage phenotype of a dual tropic
virus is largely dominant in replicative fitness such that when it is inhibited, there
is a total loss in fitness of these isolates.

Contribution of the CXCR4 and CCR5 using variants to replicative fitness in env chimeric
viruses

Predominance of the x4 and r5x4 clones in the C19 and C27 primary HIV-1 isolates may
be related to inability of r5 clones to compete during propagation of these primary
HIV-1 isolates (as previously described; 57]). As described above, we also identified two patients infected with dual tropic HIV-1
through env chimeric virus construction 42]. By this method, there is no propagation to result in selection. Bias in the HIV-1
env quasispecies composition could be introduced during PCR and cloning. However,
the targets for primer annealing and yeast recombination on the HIV-1 genome were
conserved for all HIV-1 strains 47]. In contrast to the C19 and C27 primary isolates, the B12 and B29 env chimeric viruses
had higher r5 than x4 titers (Fig. 1e). Thus, there were minimal adjustments for equal “r5” titers and relative fitness
of B12 and B19 env chimeric viruses were similar to B2 and C3. AMD3100 had minimal
inhibition of B12 and B19 (adjusted for equal r5 titers) (Fig. 6d), which was expected based on the low x4-using component in the viral populations.
The inhibition of “CXCR4”-using component (with AMD3100) (Fig. 6c) resulted in only a slight reduction in replicative fitness of the dual tropic B12
and B19 env chimeric viruses. This 2- to 10-fold decreased fitness in B12 and B19
with AMD3100 treatment was minimal compared to 10,000-fold loss in C19 and C27 fitness
under the same conditions.

Fig. 6. Fitness of the dual tropic env chimeric virus when blocking the virus component using
CCR5 or CXCR4. a The dm env chimeric viruses (B12 and B19) were competed against the x4 reference viruses (all
titered on U87.CD4.CXCR4 cells) in U87.CD4.CXCR4 cultures or in PHA-stimulated, IL-2
treated PBMC cultures with or without maraviroc. The fitness differences are plotted
as described in Fig. 2 and “Methods”. b The level of maraviroc inhibition of the monoinfections or dual infections was measured
by relative RT activity in the cell free supernatant and plotted as a percentage of
the no drug control. c The env chimeric HIV-1 and r5 reference strains (B2 and C3) were titered on U87.CD4.CCR5
cells and then competed together in PHA-stimulated, IL-2 treated PBMC cultures with
or without AMD3100. The same dual infection with the same MOI (0.004) was repeated
in U87.CD4.CCR5 cultures. The fitness differences are plotted as described in Fig. 2 and “Methods”. d The level of AMD3100 inhibition of the monoinfections or dual infections was measured
by relative RT activity in the cell free supernatant and plotted as a percentage of
the no drug control. All experiments were performed in triplicate with the exception
of c, d (performed in duplicate) due to limited supply Of AMD3100.

Due to the low “CXCR4”-using component levels in the B19 virus, equalizing x4 titers
required the addition of 1,000-fold more volume of B19 (compared to B12) in competition
with x4 control strains A8 and E6. Even with this high concentration of B19 virus
over the A8 and E6 viruses, the B19 had only slightly greater fitness than A8 and
equal fitness to E6 viruses whereas the B12 virus was slightly more fit than both
A8 (by tenfold) and E6 (by threefold) in PBMCs (Fig. 6a). Inhibition by MVC appeared to be masked by very high levels of virus needed to
reach equalized titers of the “CXCR4”-using component (Fig. 6b). This r5 HIV-1 concentration likely saturated the susceptible CCR5+/CD4+ cells
in the PBMC culture. Nonetheless, some inhibition of the more abundant “CCR5”-using
component in the B12 and B19 viruses did reduce the overall fitness of B12 and B19
viruses compared to the A8 and E6 viruses (Fig. 6a). By reducing the MVC concentration, the B12 and B19 viruses regained fitness which
again suggests that the “CCR5”-using component is contributing to the relative production
of B12 and B19 in these competitions.

Prediction of low level x4 usage

With the B12 and B19 primary HIV-1 isolates, use of the co-receptor antagonists coupled
with the TCID
₅₀
measurements on U87.CD4.CCR5 cells and U87.CD4.CXCR4 cells provided the best prediction
of co-receptor usage within the HIV-1 quasispecies. Thus, the x4-using or r5x4 component
of the B12 and B19 chimeric env viruses was less than 1 and 0.1 % of the r5-using
component, respectively. Based on the estimated low frequency of x4 or r5x4 clones,
phenotypic analyses of co-receptor usage would require the cloning and analysis of
100 clones for B12 and 1,000 B19 clones (Table 1a, b) to identify r5 clones. Instead, we compared the predicted co-receptor usage
for the B19 HIV-1 population found in: (1) env RT-PCR product from the original patient
plasma sample, (2) in the pREC_nfl_envB12 plasmid population following yeast-based
cloning of the B19 env PCR product, and (3) in the HIV-1 envB12 chimeric virus populations
produced from 293T cells co-transfected with the pREC_nfl_envB12 + pCMV_cplt (see
42] for methods). The viral load in the B12 and B19 plasma samples was 50,000 copies/ml.
Over 5,000 copies of RNA (from plasma and env chimeric virus) was reverse transcribed
and then PCR amplified for subsequent 454 pyrosequencing of the V3 loop as described
in the “Methods”. Over 100,000 copies of DNA plasmid was sampled for 454. Unfortunately,
the RT-PCR products of the B19 plasma sample (58,000 copies/ml) were unavailable for
these analyses. An average depth of ~1,500–4,000 reads were obtained per sample. Only
105 nt of the V3 loop were analyzed and used to predict co-receptor usage based on
11/25 rule, subtype specific PSSM, Geno2Pheno (G2P, with a 5 % FPR), and PhenoSeq
(Table 1a, b). For B19 plasma sample, 95 % of the 2,363 V3 reads were identical and predicted
to be r5 by all three algorithms. Approximately 2 % of the reads (N = 51) had Lys
at amino acid position 25 in the V3 loop and were predicted by PSSM and G2P to be
x4 (Table 1a). Consistent with the low titers of x4 versus r5 in this B19 env chimeric virus
(0.1 %), we could not detect an x4 clone in ~4,100 V3 reads from the plasmid or the
resulting virus. For B12 env chimeric virus, 10 clones (0.3 %) were predicted as x4
using by G2P but only 6 of these carried Lys at position 25 and were x4 using based
on PSSM (Table 1b). Again, the percentage of predicted x4 or dual tropic clones in the B12 chimeric
virus population was consistent with the relative x4 and r5 titers (Fig. 1e).

Table 1. Determination of predicted co-receptor usage in the HIV-1 isolate by 454 deep sequencing
analyses