Fig 1.
Construction of a HIV-1 proviral library with a soft-randomized Env.
(A) A representation of the HIV-1 env gene. Regions containing CD4 binding sites (colored boxes)—Loop D, CD4 binding loop, bridging sheet and V5—are expanded with NIH45-46 epitopes indicated by filled red ovals. Residues of Loop D and V5 regions of the HIV-1 isolate ADA, soft randomized in this study, are shown below. (B) Sequences of primers used to soft-randomize the Loop D and V5 regions of ADA Env are shown. Amino-acid numbers are based on HXB2 numbering. Soft randomizing primers were synthesized by hand mixing 88% of the original nucleotide and 4% each of the other three nucleotides (88:4:4:4) for the first two codon positions (nucleotides in lower case) of Loop D residues, and 91:3:3:3 ratio for the first two codon positions of V5 residues. For wobble positions of Loop D residues, an equimolar mix (50:50) of G and T (indicated as K) was used for 4- and 6-codon amino acids. For V5 primer, because it is anti-sense, C and A (indicated as M) were used. Nucleotides subject to soft randomization are colored in red. Both the 5’ and 3′ of the primers outside the randomized regions were extended (shown in green) to have a melting temperature matching the pairing primers (shown in blue).
Fig 2.
Soft randomization introduces a controlled number of amino-acid substitutions into a specific target region.
(A) The distributions of the number of amino-acid substitutions in the Loop D and V5 regions are shown. Pie charts are generated based on the data from approximately 600,000 next generation sequence reads. (B) Amino acid substitutions in Loop D and V5 regions of the parental virus library are shown in 2-dimensional heat maps. The relative frequency of the substitutions of each residue is represented by a red gradient. The original amino acids of wild-type ADA Env are indicated as a square with a diagonal line.
Fig 3.
Only a few passages of library viruses in the presence of a CD4bs bNAb is sufficient for escape viruses to emerge.
(A) Neutralization assays of the passage 0–5 swarms selected against NIH45-46. Assays were performed in TZM-bl reporter cells. ‘Parental library’ is the virus library without any passage, and ‘Control passage 5’ is the parental virus library passaged 5 times in the absence of NIH45-46. (B) Neutralization assays with the NIH45-46 passage 5 swarm against the indicated CD4bs bNAbs. VRC01 and VRC07 were isolated from the same donor as NIH45-46 (donor: NIH45), but 3BNC117, VRC-PG04, VRC-CH31 and VRC-PG20 were isolated from Patient 3, Donor 74, CH0219 and IAVI 23, respectively. Averages ± SD of three independent experiments performed in duplicates are shown. Statistical significance was calculated by two-way ANOVA and p value was <0.001 for all differences with the relevant control swarm.
Fig 4.
Resistance to one set of VRC01-class antibodies renders the viruses resistant to other antibodies in the same class.
(A) NIH45-46 passage 5 swarm was further passaged 5 times in the presence of NIH45-46 and 3BNC117 at the concentrations shown in S2A Fig. The resulting 3BNC117 passage 5 swarm was assessed for its sensitivity against the indicated antibodies. ‘Control passage 10’ is the parental library swarm passaged 10 times in total in the absence of any antibody. (B) Neutralization assays of the VRC07 passage 5 swarm using the indicated VRC01-class bNAbs. VRC07 passage 5 swarm was obtained by passaging the NIH45-46 passage 5 swarm 5 times each against 3BNC117 and VRC07. As a control (‘Control passage 15’), the parental virus library was passaged the same number of times (15) in the absence of any antibody. Statistical significance was calculated by two-way ANOVA and p value was <0.001 for each antibody. (C) Neutralizing assays of the same swarm as shown in (B) using non-VRC01 class CD4bs antibody b12 and non-CD4bs bNAbs 10–1074, 10E8 and PGDM1400. 10–1074 recognizes a V3-loop proximal glycan, 10E8 binds the gp41 MPER region, and PGDM1400 recognizes the apex of the Env oligomer. The table below the graphs shows the IC50 values of the indicated antibodies, calculated from the neutralization assays, for the control passage 15 and VRC07 passage 5 swarms. Dotted lines in (A-C) represent average IC80 values of the antibodies for ADA isolate, and were obtained from the LANL database CATNAP (http://hiv.lanl.gov/catnap). In the case of the N6 and 10–1074, only IC50 values are available: 0.074 μg/ml and 0.003 μg/ml, respectively. Averages ± SD of three independent experiments performed in duplicates are shown.
Fig 5.
Alteration of Loop D residues is preferred to that of V5 residues for viral escape of CD4bs bNAbs.
(A) The frequency of mutation at each amino acid position in the Loop D and V5 regions of the VRC07 selected passage 5 virus swarm. The top 150 sequences determined by their copy numbers detected through deep sequencing, was analyzed. Soft-randomized residues are shown on X-axis, and bold “N” indicates a putative N-glycosylation site. ‘Control passage 15’ virus is the library virus passaged in the absence of an antibody but the same number of times as the VRC07-selected viruses. The residue alterations in the control virus likely represent those advantageous for replication in GHOST-R3/X4/R5 cells. Mutations in Loop D appear to be more frequently selected than those in V5, and loss of glycosylation is one of the most frequently selected changes in both regions. (B) Amino acid substitutions in the Loop D and V5 regions of VRC07 selected passage 5 viruses are shown in 2-dimensional heat maps. The relative frequency of the substitutions of a residue is represented by a color gradient. (C) Three Loop D (N276, D279 and A281) and two V5 (N461 and S465) residues found most frequently substituted in escape variants are indicated in orange in a space-filling model of an Env trimer (PDB 5V8M). In this structure, Env is derived from a clade A isolate, BG505, and its residue 279 is N, and both 461 and 465 are T. Residues involved in the association of VRC07 are derived from the VRC07-bound Env monomer structure (PDB 4OLU) and are indicated in blue. One monomer of gp120 is shown in light green, 2nd gp120 monomer in grey, and gp41 molecules in pink. N-glycans are represented as balls and sticks in dark green.
Fig 6.
HIV-1 can escape from CD4bs bNAbs through a wide range of diverse pathways.
(A) Upper panel: Loop D and V5 sequences of the six clones randomly chosen from top 25 of the VRC07 passage 5 swarm. Sequence ranking is based on copy numbers in the VRC07 passage 5 swarm, normalized by the copy number of individual sequence in the control swarm passaged the same number of times but without antibody (Control passage 15 swarm). Lower panel: Neutralization assays of these virus clones against the indicated antibodies. WT is the parental NL-ADA virus. (B) The sequences of Loop D and V5 (upper panel) and neutralization assays (lower panel) of four additional clones chosen from top 150 for the small number of mutations (2 or 3) they contain. (C) Neutralization assays of clonal viruses against additional CD4bs bNAbs N6, VRC-CH31, VRC-PG04 and VRC01. The dotted lines in (A)—(C) represent average IC80 values of antibodies for the ADA isolate. Averages ± SD of three independent experiments performed in duplicates are shown. Statistical significance compared with WT NL-ADA was calculated by two-way ANOVA and p values were <0.001 for all in (A)—(C).
Fig 7.
A subset of escape mutations selected from soft-randomized library are found in natural isolates and previously-identified in vivo escape variants.
(A) Sequence analysis of the Loop D and V5 regions of our top 150 escape variants (top panel) and 5471 naturally-occurring HIV-1 isolates (bottom panel) was performed using the AnalizeAlign tool available at LANL (www.hiv.lanl.gov), and presented in logo plots. (B) Similar sequence analyses of these 150 escape variants (top panel) and the escape mutations detected in the clinical trials evaluating VRC01 or 3BNC117 is shown. Only four trials [6, 7, 11, 12], from which complete sequence information is available, are included in the analyses. For clarity, parental sequences and insertion mutations have been excluded. Clinical trial IDs, antibodies, and references are indicated for each analysis. Amino-acid numbers are based on HXB2 numbering.
Fig 8.
Escape variants exhibit slower growth than WT virus but outgrow WT virus in the presence of bNAbs.
CD4-Ig neutralization assays of the virus clones 1, 5, 10, 15, 20 and 25 (A) or clones 14, 73, 75 and 142 (B) performed in TZM-bl cells. WT is the parental virus, NL-ADA. Averages ± SD of three independent experiments performed in duplicates are shown for (A) and (B). (C) IC50 values of CD4-Ig calculated from the inhibition assays shown in (A) and (B) for WT virus and each of indicated virus clones. The clones indicated by bold letters were characterized for their replication efficiency in the experiments shown in (D). (D) Replication efficiency of WT and indicated virus clones in CD4+ T cells in the presence and absence of 10 μg/ml VRC07. Progeny viruses were harvested every 2 days up until 8 days post infection and quantified by RT-qPCR. Averages ± SD of three independent experiments performed in duplicates are shown. Statistical significance was calculated using two-way ANOVA, compared WT to each clone and indicated with corresponding colors.