Inhibition of human immunodeficiency virus type 1 gp120 presentation to CD4 T cells by antibodies specific for the CD4 binding domain of gp120.

Human immunodeficiency virus (HIV)-specific CD4 T-cell responses, particularly to the envelope glycoproteins of the virus, are weak or absent in most HIV-infected patients. Although these poor responses can be attributed simply to the destruction of the specific CD4 T cells by the virus, other factors also appear to contribute to the suppression of these virus-specific responses. We previously showed that human monoclonal antibodies (MAbs) specific for the CD4 binding domain of gp120 (gp120(CD4BD)), when complexed with gp120, inhibited the proliferative responses of gp120-specific CD4 T-cells. MAbs to other gp120 epitopes did not exhibit this activity. The present study investigated the inhibitory mechanisms of the anti-gp120(CD4BD) MAbs. The anti-gp120(CD4BD) MAbs complexed with gp120 suppressed gamma interferon production as well as proliferation of gp120-specific CD4 T cells. Notably, the T-cell responses to gp120 were inhibited only when the MAbs were added to antigen-presenting cells (APCs) during antigen pulse; the addition of the MAbs after pulsing caused no inhibition. However, the anti-gp120(CD4BD) MAbs by themselves, or as MAb/gp120 complexes, did not affect the presentation of gp120-derived peptides by the APCs to T cells. These MAb/gp120 complexes also did not inhibit the ability of APCs to process and present unrelated antigens. To test whether the suppressive effect of anti-gp120(CD4BD) antibodies is caused by the antibodies' ability to block gp120-CD4 interaction, APCs were treated during antigen pulse with anti-CD4 MAbs. These treated APCs remained capable of presenting gp120 to the T cells. These results suggest that anti-gp120(CD4BD) Abs inhibit gp120 presentation by altering the uptake and/or processing of gp120 by the APCs but their inhibitory activity is not due to blocking of gp120 attachment to CD4 on the surface of APCs.

LFA-1 expression on target cells promotes human immunodeficiency virus type 1 infection and transmission.

While CD4 and the chemokine receptors are the principal receptors for human immunodeficiency virus (HIV), other cellular proteins, such as LFA-1, are also involved in HIV infection. LFA-1 and its ligands, ICAM-1, ICAM-2, and ICAM-3, can be expressed on the cells infected by HIV, as well as on the HIV virions themselves. To examine the role of LFA-1 expressed on target cells in HIV infection, Jurkat-derived Jbeta2.7 T-cell lines that express either wild-type LFA-1, a constitutively active mutant LFA-1, or no LFA-1 were used. The presence of wild-type LFA-1 enhanced the initial processes of HIV infection, as well as the subsequent replication and transmission from cell to cell. In contrast, the constitutively active LFA-1 mutant failed to promote virus replication and spread, even though this mutant could help HIV enter cells and establish the initial infection. This study clearly demonstrates the contribution of LFA-1 in the different stages of HIV infection. Moreover, not only is LFA-1 expression important for initial HIV-cell interaction, subsequent replication, and transmission, but its activity must also be properly regulated.

Anti-CD4-binding domain antibodies complexed with HIV type 1 glycoprotein 120 inhibit CD4+ T cell-proliferative responses to glycoprotein 120.

HIV-specific CD4+ helper T cell responses, particularly to the envelope glycoproteins, are usually weak or absent in the majority of HIV-seropositive individuals. Since antibodies, by their capacity to alter antigen uptake and processing, are known to have modulatory effects on CD4+ T cell responses, we investigated the effect of antibodies produced by HIV-infected individuals on the CD4+ T cell response to HIV-1 gp120. Proliferative responses of gp120-specific CD4+ T cells were inhibited in the presence of either serum immunoglobulin from HIV-infected individuals or human monoclonal antibodies specific for the CD4-binding domain (CD4bd) of gp120. Human monoclonal antibodies to other gp120 epitopes did not have the same effect. The anti-CD4bd antibodies complexed with gp120 suppressed T cell lines specific for varying gp120 epitopes but did not affect T cell proliferation to non-HIV antigens. Moreover, inhibition by the anti-CD4bd/gp120 complexes was observed regardless of the types of antigen-presenting cells used to stimulate the T cells. These results indicate that the presence of anti-CD4bd antibodies complexed with gp120 can strongly suppress CD4+ helper T responses to gp120.

Enhanced HIV type 1 neutralization by human anti-glycoprotein 120 monoclonal antibodies in the presence of monoclonal antibodies to lymphocyte function-associated molecule 1.

Cellular adhesion receptor LFA-1 and its ICAM ligands are known to play a role in HIV infection. The presence of these molecules on virions and target cells promotes virus infectivity and has previously been shown to hinder virus neutralization by anti-HIV antibodies. To delineate the effect of these molecules on neutralization of HIV-1, human monoclonal antibodies (MAbs) to V3 and the CD4-binding domain (CD4bd) of gp120 were examined in the presence of anti-LFA-1 MAbs. When either of two anti-LFA-1 MAbs was present, higher levels of virus neutralization were achieved by both anti-V3 and anti-CD4bd MAbs. This effect was observed with primary HIV-1 isolates as well as with a laboratory-adapted strain. However, this activity was seen only when an anti-LFA-1 MAb was combined with anti-gp120 MAbs that exhibited virus-specific neutralizing activities, demonstrating the specificity of both the anti-LFA-1 and anti-gp120 MAbs. Enhanced neutralization by anti-gp120 MAbs was observed if the anti-LFA-1 MAb was present during the initial 24 hr only, if added 24 hr after infection, or if present throughout the culture period. These data suggest that the anti-LFA-1 MAbs could act at different stages of HIV-1 infection, including the initial virus-cell interaction as well as during the amplification and spread of virus from cell to cell. These findings demonstrate the significant role of LFA-1 in HIV-1 infection and have important implications for evaluating the neutralizing activity of anti-HIV antibodies.

Role of cellular adhesion molecules in HIV type 1 infection and their impact on virus neutralization.

While CD4 and several chemokine receptors are the principal receptors for human immunodeficiency virus type 1 (HIV-1) viruses, other cell membrane proteins also play a role in HIV-1 infection. A large array of host cell-derived membrane proteins, including adhesion molecules, are incorporated into the envelope of HIV-1 virions, and the profile of host cell proteins acquired by the virus depends on the cells used to propagate the virus. The major leukocyte adhesion molecules, such as leukocyte-function associated antigen-1 (LFA-1), intercellular adhesion molecule-1 (ICAM-1), and CD44, retain their biological functions when expressed on the virion surface, and have been shown to increase virus-cell interaction, enhance virus infectivity, and extend the host cell range of the virus. LFA-1 and its ICAM ligands are also necessary for syncytium formation and cell-to-cell transmission of HIV-1. Furthermore, several studies demonstrate that the presence and level of cell-derived adhesion molecules on the surface of HIV-1 virions affect the process by which antibody-mediated virus neutralization occurs and is measured: the level of virus neutralization is influenced by the host cell-derived adhesion molecules present on the virus, and thus, by the type of host cells in which the virus was produced. Adhesion molecules expressed on the target cells used in neutralization assays similarly affect HIV-1 neutralization by virus-specific antibodies. Consistent with these observations is the finding that neutralizing activities of both HIV+ plasma and human anti-gp120 monoclonal antibodies (Mabs) are enhanced by an anti-LFA-1 Mab capable of blocking LFA-1 functions. Hence, LFA-1, ICAM-1, and other cellular adhesion molecules are involved in different stages of HIV-1 infection and profoundly affect HIV-1 neutralization by virus-specific antibodies. These findings illuminate the biology of virus-cell interactions and have significant implications for evaluating candidate HIV vaccines.

A human monoclonal antibody specific for the V3 loop of HIV type 1 clade E cross-reacts with other HIV type 1 clades.

To ascertain the antigenic relationship between HIV-1 viruses belonging to various genetically defined subgroups (clades), shared epitopes need to be defined. Human monoclonal antibodies (MAbs) are particularly useful for this purpose because they can detect complex regions of viral proteins that may be missed by sequence analysis and because, by definition, they react with epitopes that stimulate the human immune system. Monoclonal antibodies derived from the cells of HIV-1 clade B-infected subjects have been used extensively for this purpose. Here we describe the first human MAb derived from a clade E-infected individual; the MAb is specific for the V3 loop, recognizing a core epitope represented by the amino acids TRTSVR on the N-terminal side of the crown of the V3 loop. The IgG1(kappa) MAb, designated 1324E, binds to the clade E consensus V3 loop, to rgp120 proteins from clade E and to peripheral blood mononuclear cells infected in vitro with the virus that infected the subject from whose cells the MAb-producing heterohybridoma was derived. Strong cross-reactivity of the MAb to the V3 peptides, rgp120 proteins, and native monomeric gp120s representing clades A and C, as well as to cells infected with a clade C primary isolate, revealed a shared V3 epitope between these clades. When tested for its neutralizing ability, MAb 1324E neutralized a clade E isolate that had been adapted for growth in H9 cells but failed to neutralize five clade E primary isolates.

Human monoclonal antibodies to the V3 loop of HIV-1 with intra- and interclade cross-reactivity.

Five human anti-V3 mAbs were generated from Ab-producing cells derived from the blood of HIV-1-infected individuals from North America and selected using the V3 peptide of a divergent clade B isolate, HIV(RF). The anti-V3(RF) mAbs were mapped to a cluster of three overlapping epitopes present in the KSITKGP sequence located in the hypervariable region on the N-terminal side of the V3 loop. Broad immunochemical cross-reactivity was noted when the mAbs were tested for binding to V3 peptides derived from four clade A viruses, nine clade B viruses, and two clade C viruses. These results demonstrate antigenic relatedness in the V3 regions of these three HIV-1 clades. Affinities determined by surface plasmon resonance were higher for recombinant gp120 than for V3 peptides, suggesting that these mAbs recognize both linear and conformationally dependent epitopes of the V3 loop. Two of the mAbs neutralized four clade B T cell line-adapted and primary isolates with varying degrees of potency. The two neutralizing mAbs were the most cross-reactive with V3 peptides from several clades, had the highest affinity for V3(RF) and V3(MN), and stained HIV-infected cells. The data suggest that cross-reactivity, affinity, cell surface staining, and neutralizing activity are characteristics that describe an optimal fit between Ag and Ab. The results also demonstrate that the V3 peptides representing the sequence of several clade A, B, and C viruses share antigenic features that are recognized by the human immune response, a finding that suggests that cross-clade immunity to HIV-1 may be inducible by HIV-1 vaccines.

CTL induction using synthetic peptides delivered in emulsions--critical role of the formulation procedure.

Emulsions have been used with variable degrees of success to deliver antigen to stimulate immune responses. We have investigated three different ways of incorporating peptide antigen into soybean emulsions to induce CTL responses in mice. Two of these emulsions (oil-in-water, o/w, and water-in-oil-in-water, w/o/w) had peptide incorporated at the formulation stage, while the third had peptide added to a pre-formed o/w emulsion. High levels of CTL activity were induced when peptide was dispersed into the o/w or w/o/w emulsions, in contrast to addition of peptide to the pre-formed o/w emulsion, which did not stimulate a CTL response. Induction of CTL activity was independent of emulsion globule size but was correlated with a negative zeta potential and dispersion of peptide in the oil phase. The ability of peptide in soybean oil emulsion to induce CTL is critically dependent on dispersion of peptide at the time of emulsion formation.

Neutralization of HIV-1 primary isolates by polyclonal and monoclonal human antibodies.

To examine antibody-mediated neutralization of HIV-1 primary isolates in vitro, we tested sera and plasma from infected individuals against four clade B primary isolates. These isolates were analyzed further for neutralization by a panel of several human anti-HIV-1 mAb in order to identify the neutralizing epitopes of these viruses. Each of the HIV-1+ serum and plasma specimens tested had neutralizing activities against one or more of the four primary isolates. Of the three individual sera, one (FDA-2) neutralized all of the four isolates, while the other two sera were effective against only one virus. The pooled plasma and serum samples reacted broadly with these isolates. Based on the neutralizing activities of the mAb panel, each virus isolate exhibited a distinct pattern of reactivity, suggesting antigenic diversity among clade B viruses. Neutralizing epitopes were found in the V3 loop and CD4-binding domain of gp120, as well as near the transmembrane region (cluster II epitope) of gp41. A mAb directed to the cluster I epitope of gp41 near the immunodominant disulfide loop weakly neutralized one primary isolate. None of the mAb in the panel affected one primary isolate, US4, although this virus was sensitive to neutralization by some of the polyclonal antibody specimens. This isolate was also resistant to neutralization by a cocktail of 10 mAb, most of which individually inhibited at least one of the other three viruses tested. These results suggest that neutralizing activity for this latter virus is present in certain HIV-1+ sera/plasma, but is not exhibited by the mAb in the panel. Thus, effective neutralizing antibodies against primary isolates can be generated by humans upon exposure to HIV-1, but not all of these antigenic specificities are represented in a large panel of human anti-HIV-1 mAb.

Virus-specific, CD8+ major histocompatibility complex class I-restricted cytotoxic T lymphocytes in lymphocytic choriomeningitis virus-infected beta2-microglobulin-deficient mice.

Following infection with lymphocytic choriomeningitis virus (LCMV), normal adult mice generate virus-specific, major histocompatibility complex (MHC) class I-restricted cytotoxic T lymphocytes (CTL) which clear the virus after intraperitoneal infection or cause death following intracranial (i.c.) infection. We have investigated the response of beta2-microglobulin-deficient (beta2m-) mice of the H-2d haplotype (KOD mice) to LCMV infection. Unlike H-2b beta2m- mice, which generate CD4+ MHC class II-restricted CTL in response to LCMV, KOD mice generate high levels of CD8+ MHC class I-restricted, virus-specific CTL. These CTL are specific for the LCMV nucleoprotein epitope (residues 118 to 126) in association with the Ld class I molecule, analogous to the CTL response in wild-type mice. KOD mice are also susceptible to lethal LCM disease, with 75 to 80% of the mice dying 7 to 9 days following i.c. infection with virus. Similar to results with normal mice, lethal LCM disease in KOD mice is prevented by in vivo depletion of CD8+ T cells prior to i.c. infection. In contrast to wild-type mice, however, KOD mice cannot control LCMV and become persistently infected. Overall, these results demonstrate that beta2m is not an absolute requirement for presentation of endogenous antigen on Ld or for induction of virus-specific Ld-restricted CTL in vivo.

Resting cell neutralization assay for HIV-1 primary isolates.

A technique is described for detecting the activity of neutralizing polyclonal or monoclonal antibodies against HIV-1 primary isolates. Most commonly, neutralizing antibody activity for HIV-1 is assessed by quantifying the ability of antibodies to inhibit virus infection in mitogen-activated peripheral blood mononuclear cells or transformed lymphocytes. Because the target of HIV infection in vivo is neither a mitogen-activated nor a transformed cell, an assay using unstimulated peripheral blood mononuclear cells as a more physiologic target cell was developed. This "resting cell assay" mainly utilizes primary HIV-1 isolates that have been carried for only a few passages in vitro. The result is an assay that is more efficient to perform and that detects neutralizing activity with comparable or greater sensitivity than that previously described for assays of primary HIV-1 isolates.

Neutralization of a clade B primary isolate by sera from human immunodeficiency virus-uninfected recipients of candidate AIDS vaccines.

The inability of antibodies induced by experimental human immunodeficiency virus type 1 (HIV-1) vaccines to neutralize HIV-1 primary isolates may be due to a failure to elicit such antibodies, antigenic differences between the vaccine and the strains tested, insensitivity of the assays used, or to a combination of factors. New neutralization assays were used to determine the ability of candidate AIDS vaccines to generate neutralizing antibodies for clade B primary isolate BZ167, which is closely related in portions of its envelope to the immunizing strains. Sera from HIV-uninfected volunteers in vaccine trials were tested, and neutralizing activity was found in recipients of recombinant (r) gp120MN or of rgp160MN-containing canarypox boosted with rgp120SF-2. Detection of antibodies that neutralize primary isolate BZ167 correlated with neutralizing activity for homologous vaccine strains. These data demonstrate that certain candidate AIDS vaccines can elicit antibodies that neutralize a primary isolate of HIV-1.

Synthetic peptides entrapped in microparticles can elicit cytotoxic T cell activity.

Peptides from Plasmodium berghei circumsporozoite protein (CS) and influenza A virus nucleoprotein (NP) were entrapped in microparticles prepared from poly (lactide-co-glycolide) polymers, and the microparticles were administered parenterally to mice. After immunization with single or multiple doses, splenocytes were tested for a cytotoxic T cell (CTL) response and high levels of CTL activity were detected. The CTL induced were CD8+, MHC class I restricted, and could recognize virus infected cells. Peptide entrapped in microparticles of mean size < 500 nm were better inducers of CTL than larger microparticles (mean > 2 microns and above). Microparticles could also be used to deliver lipid modified peptides (lipopeptides) and elicited higher levels of cytolytic activity than either free peptide in microparticles or lipopeptide alone. Microparticles provide a novel way of inducing a CTL response using synthetic peptides.

Comparison of numerous delivery systems for the induction of cytotoxic T lymphocytes by immunization.

A variety of vaccine delivery systems including peptides with various adjuvants, recombinant particles, live recombinant viruses and bacteria and plasmid DNA were tested for their ability to induce CD8+ cytotoxic T lymphocytes (CTL) against a well-defined epitope (amino acids 252-260) from the circumsporozoite (CS) protein of Plasmodium berghei. We compared routes of immunization that would be applicable for the administration of a malaria vaccine in humans. The majority of these vaccines did not induce high CTL responses in the spleens of immunized mice. However, both a yeast-derived Ty virus-like particle expressing the optimal nine-amino acid epitope SYIPSAEKI from the CS protein (CSP-VLP) and a lipid-tailed peptide of this same sequence induced high levels of the major histocompatibility complex (MHC) class I-restricted CTL with one and three subcutaneous immunizations, respectively. Moreover, these CTL were able to recognize naturally processed antigen expressed by a recombinant vaccinia virus. The levels of CTL induced by CSP-VLP could be augmented by co-immunization with certain cytokines. Target cells pulsed with CSP-VLP were recognized and lysed, showing that the particles were effectively processed and presented through MHC class I presentation pathway. The levels of CTL induced using CSP-VLP and lipopeptides are comparable to those observed after immunization with multiple doses of irradiated sporozoites.

Comparison of adjuvant formulations for cytotoxic T cell induction using synthetic peptides.

We have investigated the capacity of synthetic peptides delivered in different adjuvant formulations to induce cytotoxic T lymphocyte (CTL) responses to a class I H-2Kd-restricted Plasmodium berghei circumsporozoite epitope, CS 252-260. Using three immunogen formulations: soybean emulsion; Montanide ISA720; and lipopeptide (P3-CS), we first evaluated the effects of immunization routes on CTL induction. No CTL response was induced in mice immunized s.c. or i.p. with CS peptide formulated in soybean emulsion. In contrast, immunization with lipopeptide P3-CS either s.c. or i.p. effectively primed for CTL. Interestingly, CS peptide emulsified in Montanide ISA720 induced a CTL response only when delivered s.c. and not i.p., indicating the critical influence of immunization routes on CTL induction. We then compared the effectiveness of eight adjuvant formulations to induce CTL response following a single s.c. immunization. Notably, lipopeptide P3-CS and CS peptide admixed with P3 or POE lipid molecules stimulated a vigorous CTL response. However, only mice immunized with P3-CS and CS peptide admixed with P3 molecule generated long-lived CTL which persisted in vivo for 5 months. Thus, based on a simultaneous comparison of the different adjuvant formulations, we demonstrated that the conjugated and unconjugated P3 lipopeptides were the most effective immunogens for eliciting primary and memory CTL in mice.

Alterations of a dominant epitope of lymphocytic choriomeningitis virus which affect class I binding and cytotoxic T cell recognition.

We have investigated mutation of a dominant cytotoxic T cell (CTL) epitope from the nucleoprotein (NP) of lymphocytic choriomeningitis virus (LCMV). Five NP peptide analogs with single substitutions at the predicted anchor residues (designated by the wild type amino acid, the position number and the new amino acid: P2A, P2R, M9L and M9K) and at a non-anchor position (S5N) were examined for binding to class I, H-2 Ld molecules. Each of the substitutions decreased or abolished the capacity of the NP peptide to increase cell surface Ld expression and to induce Ld stabilization in the cell lysates, indicating that these substitutions significantly affected peptide binding to Ld. We tested the peptide analogs for recognition by bulk primary CTL specific for LCMV, and for their ability to stimulate in vitro the CTL originally induced by wild type LCMV. Except for the M9L change, all mutations reduced CTL recognition by at least 100-fold, and the analogs failed to stimulate the CTL in vitro. The M9L peptide was recognized by the CTL and stimulated CTL in vitro almost as well as wild type; however, this peptide induced Ld stabilization in the cell lysates to a much lesser extent than wild type. Overall, this study demonstrates that mutations in the NP epitope affected peptide binding to the Ld molecule and CTL recognition.

Mutations inside but not outside the peptide binding cleft of the H-2 Ld molecule affect CTL recognition and binding of the nucleoprotein peptide from the lymphocytic choriomeningitis virus.

In order to investigate the role of residues inside and outside the peptide binding cleft of the Ld molecule in peptide presentation to cytotoxic T lymphocytes (CTL), we constructed a series of point mutations in the Ld gene. We determined the effects of the mutations in the Ld molecule on the binding and recognition of an Ld-restricted CTL epitope derived from the nucleoprotein (NP) of the lymphocytic choriomeningitis virus (LCMV). Each of the mutations within the Ld peptide binding cleft resulted in a complete loss of CTL recognition. Addition of the LCMV NP peptide to cells expressing these mutants did not increase surface Ld expression, suggesting that the mutations altered peptide binding. Mutations involving pockets D and E within the cleft affected LCMV peptide binding and recognition as drastically as those in pocket B, which was predicted to interact with a main anchor residue of the peptide. In striking contrast, the mutations located outside the cleft did not change either recognition or binding. These results demonstrate that the Ld residues in the peptide binding cleft are the main determinants dictating LCMV NP peptide binding, and that the residues in each of the pockets within the cleft play a role in this interaction. Surprisingly, one mutation outside the peptide binding cleft, T92S, abrogated CTL lysis of target cells treated with the LCMV NP peptide, but not virus-infected cells. These data show that this mutation selectively altered the presentation of the LCMV NP peptide introduced to the cell exogenously, but not endogenously. This implies that the pathway by which peptides associate with class I molecules within the cell differs from that of exogenous peptide binding.

Roles of the six peptide-binding pockets of the HLA-A2 molecule in allorecognition by human cytotoxic T-cell clones.

To evaluate the contribution of the major histocompatibility complex class I pockets to the binding of self-peptides recognized by alloreactive cytotoxic T-lymphocyte (CTL) clones, we have constructed an extensive library of HLA-A2 mutants with different amino acid substitutions in each of the six pockets. When these mutants were tested in cytotoxicity assays with a panel of HLA-A2-specific alloreactive CTL clones, each CTL clone showed a unique pattern of reactivity, implying the different contributions of each pocket to binding individual peptides. We noted that the majority of the mutants in pocket B significantly affect recognition by the CTL clones. Unexpectedly, the mutations influencing allorecognition are found in all other pockets as well. Overall, this study demonstrates that each of the six peptide-binding pockets plays an important and distinct role in binding of self-peptides required for recognition of the HLA-A2 molecule by alloreactive CTLs.

Overlapping cytotoxic T-lymphocyte and B-cell antigenic sites on the influenza virus H5 hemagglutinin.

To define the recognition site of cytotoxic T lymphocytes (CTLs) on influenza virus H5 hemagglutinin (HA), an H5 HA-specific CTL clone was examined for the ability to recognize monoclonal antibody-selected HA variants of influenza virus A/Turkey/Ontario/7732/66 (H5N9). On the basis of 51Cr release assays with the variants, a CTL epitope was located near residue 168 of H5 HA. To define the epitope more precisely, a series of overlapping peptides corresponding to this region was synthesized and tested for CTL recognition. The minimum peptide recognized by the CTL clone encompassed residues 158 to 169 of H5 HA. Relative to the H3 HA three-dimensional structure, this CTL epitope is located near the distal tip of the HA molecule, also known as a major B-cell epitope on H3 HA. A single mutation at residue 168 (Lys to Glu) in the H5 HA variants abolished CTL recognition; this same amino acid was shown previously to be critical for B-cell recognition (M. Philpott, C. Hioe, M. Sheerar, and V. S. Hinshaw, J. Virol. 64:2941-2947, 1990). Additionally, mutations within this region of the HA molecule were associated with attenuation of the highly virulent A/Turkey/Ontario/7732/66 (H5N9) (M. Philpott, B. C. Easterday, and V.S. Hinshaw, J. Virol. 63:3453-3458, 1989). When tested for recognition of other H5 viruses, the CTL clone recognized the HA of A/Turkey/Ireland/1378/83 (H5N8) but not that of A/Chicken/Pennsylvania/1370/83 (H5N2), even though these viruses contain identical HA amino acid 158-to-169 sequences. These results suggest that differences outside the CTL epitope affected CTL recognition of the intact HA molecule. The H5 HA site defined in these studies is, therefore, important in both CTL and B-cell recognition, as well as the pathogenesis of the virus.

Hemagglutinin mutations related to attenuation and altered cell tropism of a virulent avian influenza A virus.

The H5 hemagglutinin (HA) of a highly virulent avian influenza virus, A/Turkey Ontario/7732/66 (H5N9), was previously shown to have five neutralizing epitopes, and escape mutants within one epitope (group 1) were markedly attenuated (M. Philpott, B. C. Easterday, and V. S. Hinshaw, J. Virol. 63:3453-3458, 1989). To define the genetic changes related to these antigenic and biologic properties, the HA genes of mutants within each of the epitope groups were sequenced by using the polymerase chain reaction. The mutations in the attenuated group 1 mutants were located near the distal tip of the HA molecule in close proximity to the receptor-binding site, on the basis of alignment with the three-dimensional structure of the H3 HA. All group 1 mutations involved charged amino acids. The group 1 mutants, similar to the wild-type virus, spread systemically and were recovered from the spleens of infected chickens but, unlike the wild-type virus, failed to produce severe necrosis in the spleens. Viral replication in the spleens was investigated by in situ hybridization of spleen sections from chickens infected with the wild-type or attenuated mutants. Wild-type virus replication was demonstrated in large, mononuclear, macrophagelike cells; however, group 1 mutant virus was detected attached only to erythrocytes within the red pulp. These results suggest that the attenuated mutants differ in their cell tropism within the spleen.