Neutralizing Antibodies Have Limited Effects on the Control of Established HIV-1 Infection In Vivo

Transcription

1 Immunity, Vol. 10, , April, 1999, Copyright 1999 by Cell Press Neutralizing Antibodies Have Limited Effects on the Control of Established HIV-1 Infection In Vivo Pascal Poignard,* Rebecca Sabbe,* Gaston R. Picchio,* Meng Wang,* Richard J. Gulizia,* Hermann Katinger, Paul W.H.I. Parren,* Donald E. Mosier,* and Dennis R. Burton * * Departments of Immunology and Molecular Biology concentration, the Ab might have little effect in controlling infection if the establishment of infection was not blocked. To directly address the question of whether Ab can impact the course of an ongoing HIV-1 infection, we decided to study in detail the consequences of passive administration of neutralizing Abs in an established The Scripps Research Institute infection using the hu-pbl-scid mouse model. We used La Jolla, California the only three human mabs, b12, 2G12, and 2F5 (Muster Institute of Applied Microbiology et al., 1993; Burton et al., 1994; Trkola et al., 1996), University of Agriculture known to have a potent neutralizing activity against a 1190 Vienna broad range of primary isolates (Burton and Montefiori, Austria 1997; D Souza et al., 1997). We reasoned that if high serum concentrations of these neutralizing Abs do not influence the course of HIV infection in the mouse model, Summary then it is unlikely that neutralizing Abs at the levels found in natural infection would impact the course of infection Neutralizing antibodies can protect against challenge in humans. This would help clarify the still unclear role with HIV-1 in vivo if present at appropriate concentraindividuals. of humoral immunity in the control of HIV-1 in infected tions at the time of viral challenge, but any role in the control of established infection is unclear. Here, we show that high serum concentrations of neutralizing monoclonal antibodies, either singly or as a cocktail, Results have little sustained effect on viral load in established HIV-1 infection in hu-pbl-scid mice. In some in- The Administration of a Single Neutralizing mab stances, virus replication of neutralization-sensitive Does Not Control HIV-1 Replication In Vivo virus continues even in the presence of high levels of Initial experiments focused on the activity of the human neutralizing antibody. In most instances, neutraliza- mab b12 on an established HIV-1 infection. Fifteen retion escape occurs in a few days, even from a cocktail constituted hu-pbl-scid mice were infected intraperiof three antibodies that recognize distinct epitopes. toneally (i.p.) with HIV-1 JR-CSF, a molecularly cloned HIV-1 The results imply that humoral immunity is unlikely to primary R5 virus that is sensitive to b12 neutralization play a significant role in the control of established in vitro (90% neutralization titer 5 g/ml). After 5 days, HIV-1 infection in humans. viral RNA could be detected in the mouse plasma using RT-PCR. At day 6, five mice were treated i.p. with 50 mg/kg of b12. This dose of b12 has been shown to Introduction completely protect hu-pbl-scid mice against challenge with HIV-1 JR-CSF if given up to 8 hr post viral chal- The classical and most definitive approach to establish- lenge (Gauduin et al., 1997). The injection results in an ing antiviral activities of antibody (Ab) is through passive initial serum concentration of about 500 g/ml that deimmunization studies. Such studies have shown that cays with a half-life of a week (Parren et al., 1995). There- Abs capable of in vitro neutralizing activity can protect fore, the initial Ab concentration is about 100-fold higher against challenge with neutralization-sensitive T cell line than the 90% in vitro neutralization titer and is still about adapted strains of HIV-1 in chimpanzees and in hu-pbl- 10-fold higher after a period of 3 weeks. Five control SCID mice (Safrit et al., 1993; Gauduin et al., 1995; Parren mice were given no treatment and another 5 mice were et al., 1995; Andrus et al., 1998). Protection against prigiven 50 mg/kg of a normal polyclonal human IgG prepamary isolates has been less studied and is more difficult ration. Plasma viral RNA levels were assessed over time to demonstrate, correlating with the relative resistance by quantitative RT-PCR (Roche HIV Amplicor assay). of these isolates to in vitro neutralization (Moore and In most mice, including controls, viral loads fell before Ho, 1995). Nevertheless, recently a potent human monoleveling off between days 7 and 13 after virus inoculaclonal Ab (mab) (b12) has been shown to protect against tion. Five mice with detectable viral loads at day 7 were HIV-1 primary isolate challenge in the hu-pbl-scid mouse followed further. In these mice, viral load levels were model (Gauduin et al., 1997). Protection was observed maintained for at least 5 weeks (data not shown). Most even when Ab was given several hours post viral expostrikingly, there were no significant differences in either sure. However, when breakthrough occurred, the titer the initial rate of decrease or the plateau levels of viral of the infectious virus was as high in treated animals as in control animals, indicating that, despite a high serum RNA between the b12-treated and the control mice (Fig- ure 1A). Next, the experiment was repeated with a second R5 To whom correspondence should be addressed ( burton@ primary virus, HIV-1 SF162. The kinetics of replication of scripps.edu). this virus in the hu-pbl-scid mouse model differ in that

2 Immunity 432 Figure 1. Plasma Levels of HIV-1 RNA Measured by Quantitative RT-PCR in HIV-1- Infected hu-pbl-scid Mice after Administration of the Neutralizing Ab b12 Hu-PBL-SCID mice with a HIV-1 JR-CSF (squares) or HIV-1 SF162 (circles) isolate infection were injected i.p. with 1 mg of b12 (solid symbols) or control Ab (open symbols). Mice received a single injection at day 0 (A and B) or multiple injections at days 0, 5, and 10 (C and D). In (A) and (B), mice were treated 6 days after viral infection. In (C) and (D), mice were treated when viral steady state was reached, 8 and 10 days after viral infection for HIV-1 SF162 and HIV-1 JR-CSF, respectively. The vertical arrows indicate days of Ab injection. Each curve represents the values for a single animal. Blood was obtained by bleeding from the orbital sinus of the mice. Plasma HIV-1 RNA levels were measured by the quantitative Roche RT PCR assay (Amplicor HIV Monitor, Roche Molecular Systems), following the manufacturer s recommended protocol, except that plasma volume was 100 l. The limit of HIV-1 RNA detection, as indicated by the horizontal arrow, was 400 copies/ml. the viral RNA levels tend to plateau earlier and at higher b12. Out of seven HIV-1 JR-CSF isolates rescued from neutralizing levels than following HIV-1 JR-CSF infection. Another difference Ab-treated mice, four displayed a complete relevels is that HIV-1 SF162 is more sensitive to neutralization sistance to b12 neutralization at concentrations as high by b12 than HIV-1 JR-CSF (90% neutralization titer 1 g/ as 50 g/ml. In contrast, isolates rescued from normal ml). A 50 mg/kg dose of Ab now results in a serum human IgG-treated mice remained sensitive to b12 at concentration about 500-fold higher than the in vitro levels similar to the original HIV-1 JR-CSF (Figure 2A). Of 90% neutralization titer. Although the viral load dropped note, three isolates rescued from b12-treated mice displayed to undetectable levels in one b12-treated mouse, again a neutralization sensitivity similar to control isoto there was no overall significant difference between b12- lates. From six HIV-1 SF162 isolates that were rescued from treated and control mice (Figure 1B). b12-treated mice, five displayed a marked increase in To verify that the lack of effect of b12 administration resistance to b12 neutralization (Figure 2B). The remaining was not due to decrease of Ab concentration under a isolate was as sensitive to b12 neutralization neutralization threshold, we studied the effect of multi- as control isolates. ple injections of b12. In order to better make evident an eventual impact of Ab administration, we decided to administer b12 when the viral load had stabilized. Two The Neutralization Escape Mutants Display Envelope mice were injected starting 10 days after HIV-1 JR-CSF in- Mutations Consistent with Their Phenotype fection and two additional mice were injected starting To determine if changes in b12 neutralization phenotype 8 days after HIV-1 SF162 infection. Infected hu-pbl-scid were associated with genotypic changes, we derived mice received three injections of 50 mg/kg of b12, one sequences corresponding to the entire V1 to C4 regions every 5 days, and were then followed over 15 days. of gp120 from isolates rescued from treated and control Control animals received injections of purified normal mice. All the isolates that displayed in vitro resistance human IgG at the same dose. As shown in Figures 1C to b12 neutralization had 1 3 mutations compared to and 1D, in all b12-treated animals the viral load remained the wild-type sequence, including mutations in each stable or increased over time, despite Ab plasma levels case in a region known to be involved in the CD4 binding maintained at above 250 g/ml for 15 days. site and in the b12 epitope (these were P365Q, P365T, M369R, P365Q and M369R, and V368A). Detailed analysis The Viruses Rescued from Neutralizing Ab-Treated of b12 escape mutation patterns will be given in Mice Are Mostly, but Not Exclusively, a further report. Therefore, envelope sequences were Neutralization Escape Mutants consistent with the observation that these isolates were The lack of effect of b12 could reflect the rapid develop- neutralization escape mutants. Of note, out of four ment of Ab neutralization escape mutants. To investisequenced, HIV-1 JR-CSF and three HIV-1 SF162 escape mutant env genes gate this possibility, viruses were rescued from b12- none showed the same pattern of mutation, treated mice by in vitro coculture with fresh activated suggesting that multiple genotypes can lead to b12 human PBL and assayed for in vitro neutralization with escape. The escape viruses were rescued from mice

3 Neutralizing Abs in Established HIV Infection 433 Figure 2. In Vitro Neutralization of HIV-1 Isolates Rescued from b12-treated hu-pbl SCID Mice In vitro neutralization by mab b12 of isolates rescued from HIV-1 JR-CSF (squares)- or HIV-1 SF162 (circles)-infected hu-pbl-scid mice, treated previously with mab b12 (solid symbols) or control Ab (open symbols, inset graphs). sacrificed as early as 7 days after b12 injection, sug- of plasma virus decay following anti-cd4 treatment. Decay gesting that neutralization escape mutants were selected was exponential in all mice, and a simple linear regresgesting within a week after b12 injection. Sequencing of sion analysis as in Ho et al. (1995) gave viral decay proviral DNA from tissues of sacrificed animals gave slopes in the range of 0.20 to 0.54 per day, very identical results, confirming the rapid in vivo selection comparable to those observed in humans treated with of escape mutants. protease inhibitors. Therefore, viral turnover in humans and the mouse model appears to be similar. The Kinetics of Plasma HIV-1 Turnover in the hu-pbl-scid Mouse Model The Injection of a Cocktail of Three Neutralizing To calculate the decay rate of plasma virus in the hu- mabs Does Not Control HIV-1 Replication In Vivo PBL SCID mouse and to verify that i.p. administration We then reasoned that a cocktail of neutralizing Abs of an Ab can lead to detectable effects on viral load directed against distinct epitopes would likely be more in this model, we treated mice with an anti-cd4 mab effective at controlling HIV replication and at preventing capable of blocking HIV infection. Administration of the the development of escape mutants. Three HIV-1 JR-CSF - humanized anti-cd4 Ab OKTcdr4a (Pulito et al., 1996) infected hu-pbl-scid mice were treated with one i.p. to mice with an established HIV-1 infection dramatically injection of 1 mg each of b12, 2G12, and 2F5 (50 mg/ dropped the viral load to undetectable levels in 1 to 5 kg each). These three mabs are capable of neutralizing days in all of the treated mice (Figure 3), showing that HIV-1 JR-CSF in vitro (the in vitro 90% neutralization titers an i.p. injected Ab can efficiently target HIV-producing for the virus used are 5, 100, and 30 g/ml for b12, 2G12, cells and that viral turnover in the hu-pbl-scid mouse and 2F5, respectively). Therefore, injection of 50 mg/kg model can allow the detection of a decrease in virus of each Ab resulted in an initial plasma concentration production following treatment. To verify that the rapid approximately 100-, 5-, and 16-fold higher than the in appearance of neutralization escape mutants following vitro 90% neutralization titer for b12, 2G12, and 2F5, Ab treatment was relevant to human infection and not respectively. In one mouse, the viral load dropped to the consequence of a much faster viral turnover in the undetectable levels and remained undetectable (Figure mouse model than in humans, we analyzed the kinetics 4A). In the other two mice, the viral load rapidly dropped

4 Immunity 434 mutations (N388D) and (P365Q) consistent with escape from 2G12 and b12, respectively (Trkola et al., 1996; Mo et al., 1997). The ELDKWA epitope, that strongly correlates with 2F5 neutralization (D Souza et al., 1997), was unchanged in the isolate with a 2F5 resistance phenotype. However, a lys glu mutation was found close to the ELDKWA epitope that may be associated with resistance (Figure 4C). The Kinetics of Appearance of the Triple Neutralization Escape Variant To study the kinetics of appearance of the triple escape variant T1, we analyzed plasma viral RNA from the corresponding Ab cocktail-treated mouse. The C4 to V4 region of gp120 that includes mutations consistent with b12 and 2G12 escape was amplified and sequenced from plasma viral RNA taken at different time points (Figure 5). On the day of Ab injection, all clones analyzed displayed wild-type sequence. By day 5, 25% of the clones displayed the mutant genotype, and by day 7 all clones analyzed displayed the mutant genotype. The appearance of the escape mutant genotype is coincident with the rebound in viral load that begins at day 5 (Figure 4A). By day 13, the wild-type genotype has reemerged to once again become the predominant species. It is presumed that the reemergence of the wildtype genotype reflects a decline in Ab levels and there- Figure 3. Plasma Levels of HIV-1 RNA in HIV-1-Infected hu-pbl- fore that a high Ab concentration must be maintained SCID Mice after Anti-CD4 Ab Administration to favor selection of the 2G12/b12 escape mutants. The horizontal arrow indicates the limit of HIV-1 RNA detection (400 copies/ml). Each curve represents the values for a single animal. (A) Hu-PBL-SCID mice with an established HIV-1 JR-CSF isolate infec- Discussion tion were injected intraperitoneally at day 0 with 1 mg of OKTcdr4A (solid symbols) or control Ab (open symbols). (B) Hu-PBL-SCID mice In order to study the impact of neutralizing Abs on an with an established HIV-1 SF162 isolate infection were injected intraongoing HIV-1 infection, we have administered potent peritoneally at day 0 with 1 mg of OKTcdr4A (solid symbols) or were left untreated (open symbols). neutralizing Abs to HIV-1-infected hu-pbl-scid mice after the infection had been established for some days, in most cases when the viral load had reached steady but increased again at day 5, to reach initial levels at state levels. Our results show that passive administra- day 7. This shows that HIV can replicate in vivo despite tion of either a single neutralizing human mab or of a the presence of a combination of neutralizing Abs at cocktail of three such Abs has minimal effect on the concentrations higher than the in vitro 90% neutraliza- control of an ongoing HIV-1 infection in the hu-pbltion concentrations. SCID mouse model. Following Ab administration, in some instances the virus remains sensitive to Ab neutralization, but in most instances a neutralization escape A Virus that Escapes a Cocktail of Three Neutralizing Abs Is Selected In Vivo mutant is rapidly selected. This suggests, together with Viruses were rescued from both mice and tested in vitro the reemergence of wild-type virus as Ab concentration for neutralization by 2G12, b12, and 2F5. As shown in wanes as described above, that the Ab dose used in Figure 4B, the two isolates T1 and T2 rescued from the the experiments results in an in vivo concentration close cocktail-treated mice displayed a variable resistance to to the threshold required to exert sufficient pressure on neutralization by all three Abs. One isolate (T2) displayed the virus in order to induce neutralization escape. The a phenotype resistant to neutralization by 2G12, while selection of Ab escape mutants depends on the strength remaining sensitive to neutralization by b12 and 2F5. of Ab-mediated pressure, the rate of viral turnover, and Most strikingly, the other isolate (T1) displayed a resis- on the potential escape mutant fitness (Coffin, 1995). tant phenotype to all three Abs. This demonstrates that For a minority of Ab-treated mice, the Ab-mediated pres- a variant able to escape the three most potent anti- sure in combination with the other two factors is not HIV-1 Abs described to date can be rapidly selected in sufficient to result in neutralization escape. These mice vivo. To verify whether the phenotypic changes were have the highest viral loads and presumably this increases the threshold of Ab concentration required for associated with genotypic modifications, the whole env gene of both isolates was sequenced. As shown in Fig- neutralization escape. ure 4C, sequencing revealed that one isolate (T2) displayed mutations (N336D and T338I) compatible with pressure, in combination with the other factors, is suffi- For the majority of mice in the study, the Ab-mediated 2G12 escape, whereas the other isolate (T1) displayed cient to rapidly lead to the selection of escape mutants.

5 Neutralizing Abs in Established HIV Infection 435 Figure 4. Treatment of HIV-1 JR-CSF -Infected hu-pbl-scid Mice with a Cocktail of Neutralizing Abs (A) Plasma levels of HIV-1 in HIV-1 JR-CSF - infected hu-pbl-scid mice after administration of a cocktail of neutralizing Abs (b12, 2G12, and 2F5). Ab was administered 11 days after viral infection. The time of Ab injection is indicated by the arrow at day 0 in the figure. Each curve represents the values for a single animal. M1 and M2 indicate the mouse from which the isolates T1 and T2 were rescued, respectively. Plasma HIV-1 RNA levels were measured by the quantitative Roche RT PCR assay (Amplicor HIV Monitor, Roche Molecular Systems), using a plasma volume of 50 l. The limit of detection, as indicated by the horizontal arrow, was 800 copies/ml. (B) In vitro neutralization by (1) 2F5, (2) 2G12, and (3) b12 of the challenge virus HIV-1 JR-CSF (closed square) and the viral isolates T1 (closed triangle) and T2 (open triangle) isolated from infected hu-pbl-scid mice following treatment with a cocktail of the Abs 2F5, 2G12, and b12 (see [A]). (C) Deduced amino acid sequence of the C3, V4, and gp41 regions of gp160 from wild-type (WT) JR-CSF virus and escape variants T1 and T2 following triple Ab therapy (see [A] and [B]). glycoprotein every day. Escape from neutralizing Abs only requires 1 3 mutations, and several mutation pat- terns can lead to escape from the same Ab. Since variants with replication advantage can rapidly take over (Coffin, 1995), viral escape is then possible within a short time. In most cases, Ab escape seems to occur at limited cost to the virus as replicative capacities of escape mutants and wild-type viruses are similar. Some differences are apparent. In particular, the triple escape mu- tant described does not appear as fit as the wild-type virus, as shown by rapid reversion to wild-type as Ab concentration wanes. However, it should be noted that viruses resistant to all three mabs used have been characterized in natural infection (Parren et al., 1998b), suggesting that such viruses are sufficiently fit to replicate in man. The interplay between the wild-type and the mutant virus shows the remarkable flexibility of HIV-1 in face of selective pressure. Of further note, in our experiments the Ab-mediated pressure is never suffi- cient (with one statistically nonsignificant possible exception) to reduce viral loads below the detection level except transiently. It has now been well demonstrated that only antiviral drug treatments that have a profound effect on viral replication can prevent the appearance Two hypotheses can be envisaged: (1) the escape variants were present in the viral inoculum and then selected; (2) they were rapidly generated during the ongoing infection and then selected under Ab pressure. We favor the second hypothesis, since, in one set of experiments, we used the molecularly cloned virus HIV-1 JR-CSF, and because several different escape viral variants were obtained following treatment with the same Ab. The appearance of escape mutants in a few days in the mouse model is consistent with known features of viral mutation. In hu-pbl-scid mice, about 5% of CD4 T cells, i.e., about to CD4 T cells, are infected as shown by in situ hybridization studies (D. E. M., unpublished data). The half-life of infected CD4 T cells has been shown to be approximately 1.2 days (Perelson et al., 1996). A similar half-life of infected cells in the SCID mouse model has been suggested by the work of Pettoleo-Mantovani et al. (1998) and our own results (D. E. M., unpublished data). Therefore, it can be hypothesized that about cells are infected every day in a hu-pbl-scid mouse. The mutation rate of HIV-1 is approximately /bp. Therefore, in the env gene, about three mutations per bp should be generated every day in the mouse. This can be approximated to one mutation per amino acid of the envelope

6 Immunity 436 Figure 5. Interplay between Wild-Type and Neutralization Escape Viruses upon Ab Treatment The C4 to V4 region of gp120 that includes mutations consistent with b12 and 2G12 escape was amplified and sequenced from plasma viral RNA from the T1 variant (Figure 4) taken at different time points. The bars represent the percentage of clones displaying either the wild-type sequence (gray) or the escape mutant sequence (black). The number of clones sequenced was 5, 4, 11, and 12 at day 0, 5, 7, and 13, respectively. The line represents the approximate Ab concentration (b12 concentration 2G12 concentration) in the mouse plasma estimated from the known pharmacokinetics of the Abs and checked as total Ab concentration at initial and mid-point by ELISA reactivity with gp120. of resistance (Gunthard et al., 1998). The observed very rapid emergence of escape variants in Ab-treated mice is likely associated with the inability of neutralizing Abs to have an effect on viral replication as profound as that exerted by drug combinations. One caveat to these considerations, as suggested by the work of Van Rom- pay et al. (1998), is that the Ab half-life could be shorter following administration to HIV-1-infected animals com- pared to uninfected that could potentially diminish their efficiency. However, our estimation of b12 and 2G12 concentration at days 3, 5, and 7 following Ab-cocktail administration does not indicate that the half-life of these two Abs was shortened by injection into HIV-1- infected animals (Figure 5; data not shown). Ab serum concentrations achieved in mice following i.p. injection are typically of one order of magnitude or more greater than the in vitro concentrations required to fully neutralize the HIV-1 isolates used in this study. Therefore, the free virus present in mice at the time of Ab administration is mostly if not completely neutralized. The limited impact of Ab administration on viral replication in the hu-pbl-scid mouse model then suggests that HIV-1 propagation in vivo must be mediated to varying degrees by routes other than free virus, most probably via cell-to-cell spread. Blocking of cell-to-cell transmission generally requires Ab concentrations considerably higher than those needed to block free virus entry (Pantaleo et al., 1995), explaining why the virus is able to replicate in the presence of relatively high Ab concentrations. In mice where Ab administration has no effect on virus replication and there is no evidence of neutralization escape, it would seem that infection is propagated mainly independently of free virus, which is then largely a reflection of viral production by infected cells. A number of considerations suggest that Abs are unlikely to be more effective in HIV-1 infection in humans than in the mouse model used here. Most importantly, the neutralization potency of Abs in infected humans is much less than following Ab administration in the mouse model. In sera from HIV-1-infected individuals, Ab neutralization titers are typically low (Pilgrim et al., 1997). Moreover, contemporaneous autologous isolates are rarely neutralized effectively (Pilgrim et al., 1997). In contrast, in the mouse model, serum neutralization titers following Ab administration are high and the contemporaneous virus is truly sensitive to neutralization by the Ab used. Further, we have previously shown that a pool of immune globulin from HIV-1-infected patients (HIVIG) is less potent at protecting hu-pbl-scid mice against HIV-1 primary isolate infection than the mab b12 (Gauduin et al., 1997) and is therefore likely to have even less impact on an established HIV-1 infection. All these arguments indicate that HIV-1-neutralizing Ab levels in humans will not reach the threshold needed to put signif- icant pressure on the virus. In agreement with this point of view, analysis of autologous neutralizing Ab re- sponses in HIV-1-infected individuals has recently led Schønning et al. (1998) to question the efficiency of neutralizing Abs in the control of HIV-1 infection, and to suggest that earlier observations of neutralization es- cape in patients may be epiphenomena (Albert et al., 1990; Tremblay and Wainberg, 1990; Arendrup et al., 1992). Rare HIV-1-infected individuals appear to have rela- tively high neutralizing Ab titers in their sera. In such individuals, neutralizing Ab level may reach the concen- tration threshold we have established here for exerting pressure on the virus. However, neutralization escape is likely to occur even more easily in these individuals than in the mouse model because viral diversity in human infection should be far greater than in the model permitting easier neutralization escape. In humans, di- versity is established over years during an ongoing infec- tion, whereas, in many of our experiments in the mouse model, infection was only recently established by a mo- lecular clone. Furthermore, in humans the number of infected cells is much higher than in hu-pbl-scid mice. Since viral turnover in the model and HIV-1-infected individuals is similar, the ongoing generation of viral diversity in humans should be far greater, facilitating neutralization escape. This remains valid even considering that most CD4 T cells are activated and susceptible to infection in the hu-pbl-scid mouse model. One caveat is that in the mouse model, Ab effector functions such as complement activation and Ab depen- dent cellular cytotoxicity (ADCC) are suboptimal. These functions may lead to some underestimation of the effi- cacy of Abs in human. However, our data also suggest that neutralization escape occurs mostly by mutations that reduce Ab binding to the viral envelope glyco- proteins, which would also allow the virus to escape Ab-mediated complement lysis and ADCC as well as neutralization. In conclusion, the results imply that anti- bodies at levels found in natural infection are unlikely to play a significant role in the control of HIV-1 infection in humans. Finally, the data presented here, using the currently most potent available HIV-1-neutralizing mabs (D Souza et al., 1997), suggest that antibody-based therapeutics face considerable hurdles. Abs of higher affin- ity for mature oligomeric envelope glycoproteins and therefore higher neutralizing potency (Sattentau and

7 Neutralizing Abs in Established HIV Infection 437 Antibodies The production of human mabs b12, 2G12, and 2F5 has been de- scribed previously (Burton et al., 1991, 1994; Muster et al., 1993; Trkola et al., 1996). The human mab b12 recognizes a discontinuous epitope overlapping the CD4 binding site on gp120 (Roben et al., 1994). The human mab 2G12 recognizes a conformationally sensitive gp120 epitope unrelated to the V1, V2, V3, and CD4 binding site (Trkola et al., 1996) 2F5 is an anti-gp41 human mab that has been mapped to the ELDKWA sequence (Muster et al., 1993). Moore, 1995; Parren et al., 1998a) would be desirable especially if they could interrupt cell-to-cell spread. These data also do not support the use of therapeutic vaccines solely based on neutralizing Abs, at least until neutralizing Ab titers higher than those elicited by natural infection can be obtained by vaccination. The data further suggest that a vaccine based solely on Ab would need to provide sterilizing immunity since even a low level of HIV-1 replication could lead to escape from Ab neutralization and the spread of a vaccine-resistant virus. Experimental Procedures Env Sequences Proviral DNA was extracted from infected PBMCs or hu-pbl-scid mouse tissue by standard procedures. To analyze gp120 sequence from the virus isolates present, PCR was performed to amplify the entire V1 to C4 regions of gp120 using the following primers: 5 -TATGGGATCAAAGCTTAAAGCC-3 (nt 6564 to 6585, sense) and 5 -GCCTTGGTGGGTGCTACTC-3 (nt 7695 to 7713, anti-sense). To analyze the genotype of isolates from Ab cocktail-treated mice, the entire HIV-1 JR-CSF gp160 was amplified by PCR using four primers with overlaps in sequence. The C1 to C3 regions were amplified using the following primers: 5 -ATCAGCACTTGTGGAAAGGG-3 (nt 6264 to 6283, sense) and 5 -TCAGTGCCACTTGACTTTTCA-3 (nt 7417 to 7437, anti-sense). The gp120 V4 to gp41 regions were ampli- fied using the following primers: 5 -TGAAAAGTCAAGTGGCACTGA-3 (nt 7417 to 7437, sense) and 5 -CCTTACAGTAGACCATCCAGGC-3 (nt 8813 to 8834, anti-sense). To study the genotype of circulating virus, RNA was extracted from mouse plasma using the Qiagen viral RNA kit (Qiagen). RNA was reverse transcribed to cdna (1st strand cdna synthesis kit, Boehringer Mannheim). cdna was amplified from C4 to the beginning of V4 gp120 regions using the following primers: 5 -AGGACCAG GGAGAGCATTTT-3 (sense, nt 7156 to 7175) and 5 -TCAGTGCCAC TTGACTTTTCA-3 (nt 7417 to 7437, anti-sense). PCR products were cloned using the TOPO TA Cloning Kit (In- vitrogen). Double stranded plasmid DNA was isolated from 5 to 10 clones using a Qiagen Quickspin kit (Qiagen). Sequences were obtained by automated sequencing (ABI, Perkins-Elmer). Generation of hu-pbl-scid Mice A breeding colony of CB-17 SCID mice, maintained under specific pathogen-free conditions at The Scripps Research Institute, was used as source of animals for these studies. Mice of the non-leaky phenotype (mouse IgM serum level 5 g/ml) were reconstituted by i.p. injection of normal human PBMC, as described previously (Mosier et al., 1993). Two weeks after reconstitution, mouse sera were tested for the presence of human immunoglobulin. Only human immunoglobulin-positive mice were used for HIV-1 infection. Each experiment used mice reconstituted with cells from a single donor. All procedures for maintenance and injection of the hu-pbl-scid mice were performed in a biosafety level 3 facility, in accordance with The Scripps Research Institute guidelines. HIV-1 Virus Stocks Virus stocks were made by infection of PHA-stimulated PBMCs. Virus was recovered after 5 7 days of culture and the median tissue culture infectious dose per milliliter (TCID 50 /ml) was determined by end-point dilution. Mice were infected with 1000 TCID 50. The following HIV-1 primary isolates were used in this study: HIV-1 SF162 (Cheng- Mayer and Levy, 1988), a R5 isolate, and HIV-1 JR-CSF, a molecularly cloned R5 virus (Koyanagi et al., 1987), obtained from the National Institute of Health AIDS Research and Reference Reagent Program, contributed by I. Chen. Rescue of Viral Isolates from Infected hu-pbl-scid Mice Hu-PBL-SCID mice were sacrificed, and cells were recovered from peritoneal lavage, spleen, and lymph nodes as previously described (Mosier et al., 1991). For virus isolation, cells were cocultured with PHA-stimulated PBMC. Twice weekly, culture supernatant was removed and fresh culture medium (RPMI 1640 supplemented with 10% FCS, 40 U/ml IL-2, penicillin, and streptomycin) was added. Fresh PHA-stimulated PBMC were added after 1 week of culture, and cultures were maintained for 2 weeks. Virus was then grown and titered as described. Acknowledgments We thank Robert Zivin for providing the OKTcdr4a Ab, Andrew Beernink, Michael Neal, and Ann Hessel for skilled technical assistance, and John P. Moore for manuscript review. This work was supported by National Institutes of Health grants AI33292 (D. R. B.), HL59727 (D. R. B. and D. E. M.), and AI (P. W. H. I. P.). P. W. H. I. P. acknowledges a scholar award from the Pediatric AIDS Foundation (PFR-77348). Received December 14, 1998; revised March 1, References Albert, J., Abrahamsson, B., Nagy, K., Aurelius, E., Gaines, H., Nystrom, In Vitro Neutralization Assay G., and Fenyo, E.M. (1990). Rapid development of isolate- In brief, 100 TCID specific neutralizing antibodies after primary HIV-1 infection and 50 of HIV-1 were incubated in triplicate with serial 2-fold dilutions of test Ab or control for 1 hr at 37 C in100 l of consequent emergence of virus variants which resist neutralization culture medium. Virus/Ab mixture was then incubated overnight with by autologous sera. AIDS 4, PHA-activated PBMCs in the culture medium (200 l final Andrus, L., Prince, A.M., Bernal, I., McCormack, P., Lee, D.H., Gorny, volume). On the following day, cells were washed extensively and M.K., and Zolla-Pazner, S. (1998). Passive immunization with a hucultured for 7 days. Culture supernatants were then harvested and man immunodeficiency virus type 1-neutralizing monoclonal anti- tested for viral production by p24 antigen measurement using an body in Hu-PBL-SCID mice: isolation of a neutralization escape ELISA kit (Moore et al., 1990). Neutralization was calculated as percent variant. J. Infect. Dis. 177, reduction in the average amount of p24 in supernatants from Arendrup, M., Nielsen, C., Hansen, J.E., Pedersen, C., Mathiesen, L., test culture related to that of untreated controls. and Nielsen, J.O. (1992). Autologous HIV-1 neutralizing antibodies: emergence of neutralization-resistant escape virus and subsequent development of escape virus neutralizing antibodies. J. Acquir. Im- Plasma Viral Load mune Defic. Syndr. 5, Blood was obtained by bleeding from the orbital sinus of the mice. Plasma HIV-1 RNA levels were measured by the quantitative Roche Burton, D.R., and Montefiori, D. (1997). The antibody response in RT PCR assay (Amplicor HIV Monitor, Roche Molecular Systems), HIV-1 infection. AIDS 11 Suppl A, S87 S98. following the manufacturer s recommended protocol, except that Burton, D.R., Barbas, C.F. III, Persson, M., Koenig, S., Chanock, R., plasma volume was 50 to 100 l. The limit of detection was 400 to and Lerner, R. (1991). A large array of human monoclonal antibodies 800 copies/ml. to type 1 human immunodeficiency virus from combinatorial libraries

8 Immunity 438 of asymptomatic seropositive individuals. Proc. Natl. Acad. Sci. USA Katinger, H., Moore, J.P., and Burton, D.R. (1998b). Antibody neutralization-resistant 88, primary isolates of Human Immunodefiency Virus Burton, D.R., Pyati, J., Koduri, R., Sharp, S.J., Thornton, G.B., Parren, type 1. J. Virol. 72, P.W.H.I., Sawyer, L.S., Hendry, R.M., Dunlop, N., Nara, P.L., et al. Perelson, A.S., Neumann, A.U., Markowitz, M., Leonard, J.M., and (1994). Efficient neutralization of primary isolates of HIV-1 by a recombinant Ho, D.D. (1996). HIV-1 dynamics in vivo: virion clearance rate, in- human monoclonal antibody. Science 266, fected cell life-span, and viral generation time. Science 271, 1582 Cheng-Mayer, C., and Levy, J.A. (1988). Distinct biological and serological properties of human immunodeficiency viruses from the Pettoello-Mantovani, M., Kollmann, T.R., Katopodis, N.F., Raker, C., brain. Ann. Neurol. 23, S58 S61. Kim, A., Yurasov, S., Wiltshire, H., and Goldstein, H. (1998). thy/ Coffin, J.M. (1995). HIV population dynamics in vivo: implications for liv-scid-hu mice: a system for investigating the in vivo effects of genetic variation, pathogenesis, and therapy. Science 267, multidrug therapy on plasma viremia and human immunodeficiency D Souza, M.P., Livnat, D., Bradac, J.A., and Bridges, S.H. (1997). virus replication in lymphoid tissues. J. Infect. Dis. 177, Evaluation of monoclonal antibodies to human immunodeficiency Pilgrim, A.K., Pantaleo, G., Cohen, O.J., Fink, L.M., Zhou, J.Y., Zhou, virus type 1 primary isolates by neutralization assays: performance J.T., Bolognesi, D.P., Fauci, A.S., and Montefiori, D.C. (1997). Neu- criteria for selecting candidate antibodies for clinical trials. AIDS tralizing antibody responses to human immunodeficiency virus type clinical trials group antibody selection working group. J. Infect. Dis. 1 in primary infection and long-term-nonprogressive infection. J. 175, Infect. Dis. 176, Gauduin, M.C., Safrit, J.T., Weir, R., Fung, M.S., and Koup, R.A. Pulito, V., Roberts, V., Adair, J., Rothermel, A., Collins, A., Varga, (1995). Pre- and postexposure protection against human immunodemanization S., Martocello, C., Bodmer, M., Jolliffe, L., and Zivin, R. (1996). Huficiency virus type 1 infection mediated by a monoclonal antibody. and molecular modeling of the anti-cd4 monoclonal J. Infect. Dis. 171, antibody, OKT4A. J. Immunol. 156, Gauduin, M.C., Parren, P.W.H.I., Weir, R., Barbas, C.F., Burton, D.R., Roben, P., Moore, J.P., Thali, M., Sodroski, J., Barbas, C.F. III, and and Koup, R.A. (1997). Passive immunization with a human monorecombinant Burton, D.R. (1994). Recognition properties of a panel of human clonal antibody protects hu-pbl-scid mice against challenge by Fab fragments to the CD4 binding site of gp120 that primary isolates of HIV-1. Nat. Med. 3, show differing abilities to neutralize human immunodeficiency virus Gunthard, H.F., Wong, J.K., Ignacio, C.C., Guatelli, J.C., Riggs, N.L., type 1. J. Virol. 68, Havlir, D.V., and Richman, D.D. (1998). Human immunodeficiency Safrit, J.T., Fung, M.S., Andrews, C.A., Braun, D.G., Sun, W.N., virus replication and genotypic resistance in blood and lymph nodes Chang, T.W., and Koup, R.A. (1993). hu-pbl-scid mice can be after a year of potent antiretroviral therapy. J. Virol. 72, protected from HIV-1 infection by passive transfer of monoclonal Ho, D.D., Neumann, A.U., Perelson, A.S., Chen, W., Leonard, J.M., antibody to the principal neutralizing determinant of envelope and Markowitz, M. (1995). Rapid turnover of plasma virions and CD4 gp120. AIDS 7, lymphocytes in HIV-1 infection. Nature 373, Sattentau, Q.J., and Moore, J.P. (1995). Human immunodeficiency Koyanagi, Y., Miles, S., Mitsuyasu, R., Merrill, J., Vinters, H., and virus type 1 neutralization is determined by epitope exposure on Chen, I. (1987). Dual infection of the central nervous system by AIDS the gp120 oligomer. J. Exp. Med. 182, viruses with distinct cellular tropisms. Science 236, Schønning, K., Joost, M., Gram, G.J., Machuca, R., Nielsen, C., Mo, H., Stamatatos, L., Ip, J.E., Barbas, C.F., Parren, P.W.H.I., Burphism Nielsen, J.O., and Hansen, J.E. (1998). Chemokine receptor polymorton, D.R., Moore, J.P., and Ho, D.D. (1997). Human immunodeficiency and autologous neutralizing antibody response in long-term virus type 1 mutants that escape neutralization by human HIV-1 infection. J. Acquir. Immune Defic. Syndr. Hum. Retrovirol. monoclonal antibody IgG1b12. J. Virol. 71, , Moore, J.P., and Ho, D.D. (1995). HIV-1 neutralization: the conse- Tremblay, M., and Wainberg, M.A. (1990). Neutralization of multiple quences of viral adaptation to growth on transformed T cells. AIDS HIV-1 isolates from a single subject by autologous sequential sera. 9, S117 S136. J. Infect. Dis. 162, Moore, J.P., McKeating, J., Weiss, R., and Sattentau, Q.J. (1990). Trkola, A., Purtscher, M., Muster, T., Ballaun, C., Buchacher, A., Dissociation of gp120 from HIV-1 virions induced by soluble CD4. Sullivan, N., Srinivasan, K., Sodroski, J., Moore, J.P., and Katinger, Science 250, H. (1996). Human monoclonal antibody 2G12 defines a distinctive Mosier, D.E., Gulizia, R., Baird, S., Wilson, D., Spector, D., and Specdeficiency virus type 1. J. Virol. 70, neutralization epitope on the gp120 glycoprotein of human immunotor, S. (1991). Human immunodeficiency virus infection of human- PBL-SCID mice. Science 251, Van Rompay, K.K., Berardi, C.J., Dillard-Telm, S., Tarara, R.P., Can- Mosier, D.E., Gulizia, R.J., MacIsaac, P.D., Torbett, B.E., and Levy, field, D.R., Valverde, C.R., Montefiori, D.C., Cole, K.S., Montelaro, J.A. (1993). Rapid loss of CD4 T cells in human-pbl-scid mice R.C., Miller, C.J., et al. (1998). Passive immunization of newborn by noncytopathic HIV isolates. Science 260, rhesus macaques prevents oral simian immunodeficiency virus in- Muster, T., Steindl, F., Purtscher, M., Trkola, A., Klima, A., Himmler, G., Ruker, F., and Katinger, H. (1993). A conserved neutralizing epitope on gp41 of human immunodeficiency virus type 1. J. Virol. 67, Pantaleo, G., Demarest, J.F., Vaccarezza, M., Graziosi, C., Bansal, G.P., Koenig, S., and Fauci, A.S. (1995). Effect of anti-v3 antibodies on cell-free and cell-to-cell human immunodeficiency virus transmission. Eur. J. Immunol. 25, Parren, P.W.H.I., Ditzel, H.J., Gulizia, R.J., Binley, J.M., Barbas, C.F. III, Burton, D.R., and Mosier, D.E. (1995). Protection against HIV-1 infection in hu-pbl-scid mice by passive immunization with a neutralizing human monoclonal antibody against the gp120 CD4-binding site. AIDS 9, F1 F6. Parren, P.W.H.I., Mondor, I., Naniche, D., Ditzel, H.J., Klasse, P.J., Burton, D.R., and Sattentau, Q.J. (1998a). Neutralization of human immunodeficiency virus type 1 by antibody to gp120 is determined primarily by occupancy of sites on the virion irrespective of epitope specificity. J. Virol. 72, Parren, P.W.H.I., Wang, M., Trkola, A., Binley, J.M., Purtscher, M., fection. J. Infect. Dis. 177,