Supporting Information Horwitz et al. 73/pnas.35295 A Copies ml - C 3NC7 7 697 698 7 7 73 76-2 2 Days Gp2 residue G458D G459D T278A 7/36 N28 K D 28 459 A28T ID# 697 ID# 698 ID# 7 ID# 7 ID# 73 ID# 76 ID# D human IgG ( g/ml).5mg 3NC7 i.v. n=3. 2 3 4 5 6 7 8 9 T /2 = 2. Fig. S. Treatment of HIV- infected humanized mice (hu-mice) with 3NC7 transiently reduces viral load and rapidly selects for antibody resistance mutations. (A) Viral loads from seven mice treated with the antibody 3NC7 (.5 mg twice per week s.c.). Antibody treatment began at day. Each line represents a single mouse. ( and C) Gp2 sequences were cloned from each mouse after 5 d of treatment with 3NC7 (C). Red and green dots indicate nonsynonymous and synonymous mutations, respectively, relative to HIV- YU2. Resistance mutations were identified in the CD4 binding site (CD4bs) at residues YU2 (28 282) and YU2 (458 459). The representation of each mutation among the viral sequences obtained in C is shown in, where the numbers in the center reflect the number of mice (7) and the total number of gp2 sequences obtained (36). (D) The serum half-life of 3NC7 in NOD/Rag / /IL-2Rγ null (NRG) mice was 2. d. Three NRG mice were injected with.5 mg 3NC7 i.v. at day and followed for 9 d. 3NC7 concentration was measured by human IgG-specific ELISA and the half-life was calculated using GraphPad Prism 5.. A C Copies ml - 7 3 4 Log change Average - -2-3 -4 3 4 Copies ml - 7 6 5 4 3 2 *** Wk Wk 6 Fig. S2. Combination immunotherapy with 3NC7, PG6, and -74 is sufficient to suppress viremia in HIV- YU2 infected hu-mice. (A) Plasma viral loads for hu-mice treated with combination immunotherapy were monitored weekly using the highly sensitive Abbot HIV-RealTime Assay. Each line represents a single animal. Treatment began at day. Antibodies were administered s.c. at. mg per antibody, twice per week, for a total of 6 wk. () Fold change in viral load (shown in A) from the baseline measurement for each animal. Red line shows the geometric mean viral load change at each time point. (C) Combination immunotherapy resulted in a highly significant drop in viral load in all animals (n = 5) after 6 wk of continuous treatment (***P <., Wilcoxon signed rank test, two-tailed). The average viral load drop was 2.5 log HIV- RNA copies per milliliter. Horwitz et al. www.pnas.org/cgi/content/short/35295 of7
Copies ( cells) - Intracellular HIV- DNA n.s. PMC Spleen P.LN Copies ( cells) - Intracellular HIV- RNA 7 n.s. PMC Spleen P.LN Fig. S3. Cell-associated HIV- DNA and RNA levels in immunotherapy-treated mice (Fig. ) are similar among lymphoid tissue compartments. (A and ) Cellassociated HIV- DNA (A) and RNA () levels per million human lymphoid cells isolated from the indicated tissue compartment (P.LN, peripheral lymph nodes). Each line represents a single mouse. Gray bars reflect geometric averages for each tissue compartment. n.s., not statistically significant. A Copies ml - Untreated 7 ART (3 wks) 7 ART Untreated -2 2 4 6 8 2 Days -2 2 4 6 8 2 ID# ID# 26 ID# 222 ID# 336 ID# 224 ID# 27 ART ID# 326 ID# 329 ID# 335 Gp2 residue 2 3 4 5 ID# 344 ID# 64 Fig. S4. Treatment of HIV- YU2 infected hu-mice with antiretroviral therapy (ART) rapidly lowers viral load, but viremia returns to pretreatment levels after stopping ART. (A) Untreated mice exhibit sustained viremia for over 2 d after infection (green line, geometric average). Three weeks of treatment with ART [tenofovir disproxil-fumarate/emtricitabine/raltegravir (TDF/FTC/RAL), daily by oral gavage] reduces viral load by an average of 2. log RNA copies per milliliter plasma. Twelve to 23 d after stopping ART, viral load returns to pretreatment levels. () gp2 sequences obtained from untreated mice (Upper) and ART-treated mice after stopping ART. Red and green dots indicate nonsynonymous and synonymous mutations, respectively, relative to HIV- YU2. ART-treated mice did not carry recurring mutations in gp2 that differed from those identified in untreated mice. Horwitz et al. www.pnas.org/cgi/content/short/35295 2of7
ART + 45-46 G54W 45-46 G54W ART + 45-46 G54W 4RAL 7 ART Ab 7 ART Ab RAL Copies ml -2 2 4-2 2 4 Fig. S5. Mice treated with ART plus 45-46 G54W combination therapy exhibit rapid viral rebound after both treatments are stopped. (A and ) Mice were treated with ART (TDF/FTC/RAL) for a total of 3 wk. Five into ART, combination therapy was initiated with 45-46 G54W. Combination therapy was terminated after 3 wk on ART and mice were placed on either 45-46 G54W (A, copied from Fig. 3) or raltegravir () only. (For clarity, only mice below the limit of viral load detection at the time combination therapy was stopped are shown.) Each line represents a single mouse. In contrast to mice continuing on 45-46 G54W monotherapy, viremia in mice continuing on raltegravir monotherapy quickly rebounded to pretreatment levels, indicating that antibody monotherapy sustained viremic suppression after combination therapy was stopped. Amino acid changes during bnab therapy PG6-74 45-46 G54W ID# 686 ID# 67 ID# 667 ID# 664 ID# 396 ID# 348 ID# 342 ID# 582 3NC7 ID# 58 ID# 578 ID# 566 Fig. S6. Viruses in mice that escaped antibody therapy (Figs. 3 and 5A) harbor amino acid changes in gp2 in regions that are targeted by the antibodies (). Amino acid changes in mice treated with either 45-46 G54W or 3NC7 carried mutations at either (or both) YU2 (279 28) or YU2 (458 459). The one mouse escaping -74 had the mutation YU2 N332K, and mice escaping PG6 had mutations at YU2 6/62.. Klein F, et al. (22) HIV therapy by a combination of broadly neutralizing antibodies in humanized mice. Nature 492(7427):8 22. Horwitz et al. www.pnas.org/cgi/content/short/35295 3of7
Escape during therapy AAV Passive Mouse IC 5 bnab ID# 342 5.67 4546W ID# 348 >5 4546W ID# 396 >5-74 ID# 664 >5 PG6 ID# 566 >5 3NC7 ID# 582 >5 3NC7 ID# 683 >5-74 ID# 5 9.78 3NC7 ID# 52 >5 3NC7 Post-therapy rebound ID#.66 4546W ID# 7.5 4546W ID# 33.8 4546W ID# 346.36 4546W ID# 333 73-74 Fig. S7. Viruses emerging during antibody therapy are resistant to antibody neutralization, whereas viruses emerging after therapy remain sensitive to neutralization. Pseudoviruses expressing envelope from representative viral clones isolated from hu-mice during and after antibody therapy were assayed for neutralization in a TZM-bl assay (antibody IC 5 s in micrograms per milliliter are shown). In all cases, viruses cloned from mice during antibody therapy were highly resistant to the respective antibody used in treatment (escape during therapy). This was true also for mice with reemergent virus during adenoassociated virus (AAV)-directed antibody therapy. y comparison, viruses cloned from mice after antibody therapy, when antibody concentrations in serum were low, remained highly sensitive to the respective antibodies used in therapy (posttherapy rebound). This suggests that reemergent virus after antibody therapy was a product of latently infected cells. Horwitz et al. www.pnas.org/cgi/content/short/35295 4of7
mino acid changes after bnab therapy PG6 ID# 669-74 ID# 399 ID# 333 ID# 346 45-46 G54W ID# 33 ID# 7 ID# 3NC7 ID# 585 ID# 572 Fig. S8. Gp2 sequences from virus that rebounded after antibody concentrations in plasma decreased to subtherapeutic levels (Figs. 4 and 5) generally did not harbor mutations at sites known to confer antibody resistance. Only one of nine mice for which sequences could be obtained carried mutations at resistance sites, signifying that the antibodies sustained viral suppression until rebound virus emerged from latently infected cells after antibody concentrations decayed. One -74 treated mouse, ID number 399, had viral clones bearing mutations in the -74 binding site at either YU2 D3G or YU2 N332T. Of note, viremia in this mouse rebounded while antibody was still present (34 μg/ml) and this mutant virus was likely resistant to -74 at this concentration. Horwitz et al. www.pnas.org/cgi/content/short/35295 5of7
Plasma higg (ug/ml) C Copies ml - D Gp2 residue 2 3 4 5 E 5 7 AAV expression (2d) AAV alone 2 4 6 8 AAV+ART 52 7 2 4 6 8 Copies ml - 498 7 7 ID# 498 ID# 5 ID# 52 ID# 55 ID# 59 Gp2 residue 2 3 4 5 ART AAV Escape Virus 28 459 28 459 ART+AAV 3NC7-2 2 4 6 8 2 4 2 4 6 8 55 7 Days 2 4 6 8 2 4 59 7 Escape Virus 2 4 6 8 2 4 µg ml - ID# 5 ID# 52 ID# 55 ID# 59 Fig. S9. Mice treated with ART and AAV-directed expression of 3NC7 antibody exhibit sustained viral suppression after ART cessation. (A) Mice were injected i.v. with 2 genomic copies of AAV 3NC7 in the presence or absence of ART. Mice treated with ART (TDF/FTC/RAL) had significantly lower antibody expression 2 d after AAV injection (P =.4, Mann Whitney test), possibly owing to inhibition of AAV transduction by nucleotide analogs. () The five ART plus AAV 3NC7 -treated mice with detectable antibody levels at 2 d after AAV injection (bracketed in A) were treated for a total of 28 d with ART. Each line represents a single mouse. ART was terminated and viral loads were monitored for an additional 8 d. Although two mice quickly escaped AAV 3NC7 therapy upon ART termination, the remaining three mice were aviremic for more than 8 d after ART was withdrawn. The proportion of mice that remained controlled by AAV 3NC7 was similar to results obtained for passive 3NC7 therapy (Fig. 3). The two surviving mice at day 8 were reinfected with virus from mice that escaped 3NC7 therapy and carried resistance mutations to that antibody. Upon reinfection virus reemerged in both mice, signifying that prolonged viral suppression was due to AAV 3NC7 and not to loss of the human graft. (C) Individual viral load plots for each mouse shown in, with yellow lines/symbols showing gp2-binding human IgG for each mouse. (D and E) Gp2 sequences obtained from (D) the two mice that escaped AAV 3NC7 after stopping ART carried mutations conferring resistance to 3NC7 and (E) virus that reemerged in AAV 3NC7 -controlled mice after reinfection at day 8 carried mutations conferring 3NC7 resistance that were present in the reinfection inoculum. Red and green dots indicate nonsynonymous and synonymous mutations, respectively, relative to HIV- YU2. Horwitz et al. www.pnas.org/cgi/content/short/35295 6of7
bnabs Primary T CD4+ target cells (FarRed+) + + Infected primary T CD4+ donor cells No infection Preincubation 2 of coculture 45-46 G54W 3NC7 % of Gag+ target cells No Antibody -74 PG6 Gag C Inhibition (%) 5 FarRed. 45-46 G54W 3NC7-74 PG6 5. 5 [Ab] µg/ml. 5. IC 5 (ug/ml) HIV- YU-2 Cellassociated Cell-free 45-46W.3.35 3NC7.4.47-74.2 3. PG6 7 X* Cell-free Cell-associated D * <.2 µg/ml.2-.5 µg/ml.5-7.5 µg/ml ~5% inhibition at 5 µg/ml Fig. S. Antibodies 45-46 G54W, 3NC7, and -74 block cell contact-mediated viral transmission in vitro. (A) Experimental design of an assay to measure antibody blockage of viral transmission under cell contact-permitting (cell-associated) conditions. Primary HIV- infected CD4+ T cells were preincubated for h with broadly neutralizing antibodies (bnabs) before coculture with autologous target cells stained with FarRed dye. After 48 h of coculture, the fraction of productively infected (Gag+) target cells was measured by flow cytometry (example shown in ). () Flow cytometric analysis of Gag+/FarRed+ cells after coincubation with HIV- YU2 infected donor cells (or uninfected donor cells, No infection). Donor cells were preincubated with the indicated bnab at 5 μg/ml (or no antibody) before coincubation with FarRed+ target cells. Plots show Gag+/FarRed+ cells after 48 h of coincubation with donor cells. (C) Antibodymediated inhibition of cell-free (dashed lines) and cell-associated (solid lines) viral transmission for the indicated bnab. Cell-free infections were performed on HeLa-derived P4C5 target cells and quantitated by measuring β-gal after 36 h. Results are mean ± SD from three to five independent experiments. Curves represent fitted results. (D) Summary of IC 5 values for cell-free and cell-associated viral transmission from C. Whereas all four antibodies potently inhibited cellfree virus transmission, 45-46 G54W, 3NC7, and -74 also potently inhibited viral transmission under cell contact-permitting conditions. PG6 reached only a partial IC 5 for cell-associated viral transmission at the maximum concentration tested (5 μg/ml), possibly accounting for its reduced efficacy in vivo relative to the other bnabs tested. Horwitz et al. www.pnas.org/cgi/content/short/35295 7of7