Enhanced cell surface expression, immunogenicity and genetic stability resulting from a spontaneous truncation of HIV Env expressed by a recombinant MVA

Abstract

During propagation of modified vaccinia virus Ankara (MVA) encoding HIV 89.6 Env, a few viral foci stained very prominently. Virus cloned from such foci replicated to higher titers than the parent and displayed enhanced genetic stability on passage. Sequence analysis showed a single nucleotide deletion in the 89.6 env gene of the mutant that caused a frame shift and truncation of 115 amino acids from the cytoplasmic domain. The truncated Env was more highly expressed on the cell surface, induced higher antibody responses than the full-length Env, reacted with HIV neutralizing monoclonal antibodies and mediated CD4/co-receptor-dependent fusion. Intramuscular (IM), intradermal (ID) needleless, and intrarectal (IR) catheter inoculations gave comparable serum IgG responses. However, intraoral (IO) needleless injector route gave the highest IgA in lung washings and IR gave the highest IgA and IgG responses in fecal extracts. Induction of CTL responses in the spleens of individual mice as assayed by intracellular cytokine staining was similar with both the full length and truncated Env constructs. Induction of acute and memory CTL in the spleens of mice immunized with the truncated Env construct by ID, IO, and IR routes were comparable and higher than by the IM route, but only the IR route induced CTL in the gut-associated lymphoid tissue. Thus, truncation of Env enhanced genetic stability as well as serum and mucosal antibody responses, suggesting the desirability of a similar modification in MVA-based candidate HIV vaccines.

Introduction

Recombinant vaccinia viruses (VACV), derived from the replication-deficient modified VACV Ankara (MVA) strain previously used as an attenuated smallpox vaccine (Mayr et al., 1975; Stickl et al., 1974), express proteins at a high level (Sutter and Moss, 1992), stimulate humoral and cellular immunity (Sutter et al., 1994), are safe even in immunodeficient animals (Stittelaar et al., 2001) and are unlikely to revert to greater virulence because the host range defect involves multiple gene deletions (Antoine et al., 1998; Meyer et al., 1991; Wyatt et al., 1998). Non-human primate studies have shown that recombinant MVA alone (Barouch et al., 2001; Durbin et al., 1998; Earl et al., 2002; Hirsch et al., 1996; Men et al., 2000; Negri et al., 2001; Nilsson et al., 2002; Ourmanov et al., 2000; Stittelaar et al., 2000; Zhu et al., 2000) or following DNA priming (Amara et al., 2002a; Amara et al., 2001) can prevent or reduce disease caused by numerous viruses including those related to HIV. Several clinical trials to evaluate recombinant MVA as a vaccine for HIV prevention and treatment have been initiated (Cebere et al., 2006; Harrer et al., 2005). Nevertheless, MVA vectors may need to be further optimized and additional routes of immunization explored to achieve effective protection in human trials. In particular, natural transmission of HIV is through a mucosal surface, and targeting immune responses to the gastrointestinal entry site before viral dissemination could protect and more effectively clear virus from the major site of HIV replication in the intestinal mucosa (Belyakov et al., 2001a; Veazey et al., 1998; Zhang et al., 1999).

We encountered problems when attempting to express certain integral membrane proteins at high levels by MVA vectors. For example, we were forced to express the measles virus F protein under a relatively weak promoter in order to prevent the selection of non-expressing mutants (Stittelaar et al., 2000). Similarly, during the course of experiments with a MVA/HIV recombinant expressing the HIV envelope glycoprotein (Env), we encountered a genetic instability problem. As described here, we isolated a spontaneous mutant with enhanced genetic stability, that had a 115-amino acid truncation of the cytoplasmic domain of Env, induced more intense surface staining of infected cells with HIV antiserum, grew to a higher titer, and induced enhanced serum and mucosal antibody responses compared to the parent virus.

Results

Isolation and characterization of recombinant MVA expressing a truncated HIV Env

Full-length HIV Env gp160 is processed into a transmembrane subunit gp41 with cytoplasmic and extracellular domains and a non-covalently linked gp120 subunit that binds the cellular receptors. The entire env open reading frame of HIV clade B strain 89.6, modified only by silent mutations that eliminated poxvirus transcription termination signals, was inserted into the MVA genome by homologous recombination. Live immunostaining with clade B anti-gp140 rabbit serum was used to identify the recombinant virus, MVA/89.6, which was clonally purified by repeated plaque isolations. During subsequent passages of MVA/89.6, however, we noted that some foci did not stain with Env antiserum, whereas others were larger and stained more intensely than the majority. The difference in size and intense immunostaining of the latter was retained after another clonal purification (Fig. 1A), and one such isolate was called MVA/89.6T. MVA/89.6T was more stable than MVA/89.6 as no non Env-staining foci were detected after propagation of the virus into a working stock. These data suggested that 89.6 Env expression was deleterious for MVA replication and that there was a growth selection for spontaneous mutations that relieved this adverse effect.

Figure 1.

Comparison of HIV Env of MVA/89.6 and MVA/89.6T. A. Foci of MVA/89.6 and MVA/89.6T infected CEF cells were immunostained with T8 mouse MAb against clade B gp120 at 3 days post-infection. B. Comparison of MVA/89.6 and MVA/89.6T Env proteins by SDS-polyacrylamide gel electrophoresis. MVA/89.6 or MVA/89.6T-infected BS-C-1 cells were metabolically labeled with [³⁵S]methionine and lysates immunoprecipitated with either rabbit R2144 polyclonal antiserum made against HIV IIIBgp140 or mouse D61 MAb against gp41. MWM lane contains standard protein markers with masses in kDa indicated on the side. C. Derived amino acid sequence of the C-terminal region of MVA/89.6 and MVA/89.6T gp41. Bold type indicates location of previously reported plasma membrane retrieval signals (YXXΦ, is1 and is2, and LL_852/853)

The expression of HIV Env in metabolically labeled BS-C-1 cells that were infected with MVA/89.6 or MVA/89.6T was compared. Infected cell lysates were prepared and the recombinant proteins were immunoprecipitated, with either polyclonal rabbit antiserum (R2144) prepared against the gp140 form of Env or a monoclonal antibody (MAb) directed against gp41 (D61), and analyzed by SDS-PAGE. Using the polyclonal antiserum, proteins of the sizes expected for gp160, gp120 and gp41 were immunoprecipitated from lysates of cells infected with MVA/89.6 (Fig. 1B). Although the gp120 expressed by MVA/89.6T co-migrated with that of MVA/89.6, the MVA/89.6T gp160 and gp41 bands migrated more rapidly and there was slightly less proteolytic processing (Fig. 1B). The rapid migration of gp160 and gp41 was confirmed by immunoprecipitation with a gp41-specific MAb (Fig. 1B). Thus, these data indicated that the gp41 of MVA/89.6T was smaller than that of 89.6. Quantitative Western blotting (described in Materials and Methods) confirmed that total expression of Env by both constructs was almost identical, but that the amount of cleaved gp120 was greater in MVA/89.6. Cleaved gp120 represented 54% of the total gp120 and gp160 expressed by the MVA/89.6 construct at 24 h, whereas cleaved gp120 represented 40% of the total gp120 and gp160 by MVA/89.6T.

DNA sequencing indicated that a single nucleotide deletion had occurred at position 2215 within the open reading frame of MVA/89.6T, resulting in a frame shift after aspartic acid 738 and a premature termination six codons downstream (Fig. 1C). Therefore, the cytoplasmic tail of the gp41 of MVA/89.6T is 115 amino acids shorter (not counting the 6 non-homologous residues) than that of MVA/89.6 gp41 (Fig. 1C). This spontaneous truncation of Env, which occurred in MVA, is reminiscent of similar mutations during in vitro passage of SIV and HIV-2 that enhance infectivity (Chakrabarti et al., 1989; Hirsch et al., 1989; Johnston et al., 1993; Kodama et al., 1989; Mulligan et al., 1992; Tsujimoto et al., 1988).

In order to confirm that the large focus phenotype was due to the truncated env sequence and not to mutations within the MVA genome in these MVA recombinants, we isolated an additional virus producing large foci from early passage of MVA/89.6. Comparison of the env sequence in this isolate to MVA/89.6 revealed a single nucleotide deletion within a 4 C run at position 1295 in gp120 of the Env causing a frame shift and truncation of the Env. Thus, two independent virus isolates that were picked because of their formation of large foci, had Env truncations.

Growth and genetic stability

BS-C-1 cells were infected at an multiplicity of infection (MOI) of 1 pfu/cell to compare the growth of MVA/89.6 and MVA/89.6T. When the virus titers were determined by measuring virus production by T8 mAb immunostaining, which would detect only plaques expressing Env, the MVA/89.6T produced slightly more infectious virus (Fig. 2A). Greater than 99% of the MVA/89.6T virus produced was expressing Env by T8 staining. Input MVA/89.6 virus contained about 20% non-staining plaques which increased to over 40% at the end of the growth cycle. When virus titers were determined using anti-VACV serum, viral titers were similar, suggesting that the non-staining MVA recombinants contribute to the measured MVA titers in these cases (Fig.2 B).

Figure 2.

Replication of MVA/89.6 and MVA/89.6T in CEF cells. CEF cells were infected with MVA/89.6 and MVA/89.6T at MOI of 1pfu/cell and harvested at designated time. Titers were determined in CEF by immunostaining with T8 mouse mAb reactive with gp120 Env (A) or with anti-VACV rabbit serum (B).

Further genetic stability was assessed by picking plaques from MVA/89.6 and MVA/89.6T and passaging them independently through 7 passages at MOI of 1 (Table 1). This analysis confirms that MVA 89.6 is much less stable than MVA/89.6T. At passage 7 individual plaques of the MVA/89.6T harvests were also assessed by immunostaining with T33 and T40 mAbs directed against Env gp41 and found to include viruses expressing further truncated Env.

Table 1.

Comparison of genetic stability of HIV Env expression from independently passaged MVA/89.6 and MVA/89.6T plaques

Virus	Passagea	Percentage of nonstaining focib
		#1	#2	#3	#4	#5
MVA/89.6	0c	<1	<1	<1	NDd	ND
	1	38	3	9
	3	89	55	100
	5	ND	ND	ND
	7
MVA/89.6T	1	<1	<1	<1	<1	<1
	3	<1	<1	<1	<1	<1
	5	2	3	<1	3	2
	7	13 (27)e	9 (16)	7 (23)	10 (18)	15 (30)

^aEach plaque was passaged independently in CEF at an MOI of 1

^bDetermined by immunostaining fixed plaques at 3 days post-infection with mAb T-8 which binds gp120

^cStability determined directly plating out picked plaque, before passage

^dNot done

^eDetermined by immunostaining fixed plaques at 3 days post-infection with mAbs T-33 and T-40 which bind gp41

Surface Env expression and antibody reactivity

MVA/89.6T had been isolated because of the bright live immunostaining of infected permissive chick embryo cells. However, MVA has a host-range defect for most mammalian cells. As our plan was to use the recombinant MVA for immunization of mammals, we quantitated surface expression of infected human HeLa cells by flow cytometry using mouse MAbs directed against gp120 or gp41 (Fig. 3A). The mean fluorescence intensities of cells infected with MVA/89.6T ranged from 2.0- to 2.7-fold higher than those of MVA/89.6-infected cells. These differences were specific for HIV Env, since the mean fluorescence intensities obtained with polyclonal VACV serum were similar for cells infected with MVA/89.6 and MVA/89.6T (not shown). In addition, the numbers of expressing cells as measured with antibodies to VACV were similar after infection with either recombinant virus (data not shown).

Figure 3.

Measurement of HIV Env MAb binding to MVA/89.6- and MVA/89.6T-infected HeLa cells. A. Binding of mouse MAbs (indicated on x axis) directed against gp120 and gp41 to surface of unpermeabilized infected cells determined as mean fluorescence intensity by FACS analysis (y axis). Ratios of mean fluorescence intensities for MVA/89.6 and MVA/89.6T shown below. B. Binding of mouse MAbs directed against gp120 and gp41 to permeabilized infected HeLa cells. Annotated as in panel A. C. Binding of human neutralizing MAbs to surface of unpermeabilized infected cells. Annotated as in panel A.

In contrast to the difference in surface expression of Env, when permeabilized HeLa cells infected with recombinant viruses expressing truncated or full length Env were stained with the mouse MAbs used above, mean fluorescence intensities were comparable, suggesting that similar total amounts of Env were made (Fig. 3B). This implies that truncated Env was preferentially transported to the surface.

We also compared the binding of broadly neutralizing human MAbs to ascertain whether the epitopes on the extracellular domain of Env were adversely affected by truncation of the cytoplasmic domain. However, there was also greater binding of the neutralizing MAbs to the truncated Env compared to full length (Fig. 3C).

Effect of the truncation on Env function

When the HIV Env is cleaved into gp120 and gp41 subunits, the fusion domain located in gp41 is exposed allowing CD4/co-receptor dependent fusion of virus-infected cells. The 89.6 Env is dual tropic and can use either CXCR4 or CCR5 as the co-receptor. To assess whether the Env truncation adversely affected fusion activity, HeLa cells infected with MVA/89.6T or MVA/89.6 were compared in a fusion assay quantified by β-galactosidase synthesis (Nussbaum et al., 1994). Both the parental MVA/89.6 and the mutant MVA/89.6T induced fusion of cells expressing CXCR4 or CCR5, whereas recombinant VACVs expressing other Env genes were specific for CXCR4 or CCR5 (Fig. 4).

Figure 4.

MVA/89.6T expresses fusion competent HIV Env. HeLa cells were infected with recombinant VACV expressing the bacteriophage T7 RNA polymerase and one of the following HIV Env expressing viruses: vSC60 control expressing Env from CXCR4-tropic HXB2 strain, vCD43 control expressing Env from CCR5 tropic Ba-L strain and the two MVA/89.6 viruses expressing dual tropic Env. The infected HeLa cells expressing Env and T7 RNA polymerase were mixed with infected HeLa cells expressing β-galactosidase regulated by a bacteriophage T7 promoter, either CXCR4 or CCR5, and with or without CD4. Fusion of the two HeLa populations resulted in the expression of β-galactosidase indicated at the left.

Effects of Env truncation and route of inoculation on serum antibody responses

Groups of BALB/c mice were immunized with 10⁷ pfu of either MVA/89.6 or MVA/89.6T by the intramuscular (IM) needle route at 4-week intervals. Serum anti-HIV Env antibody was detected by ELISA 4 weeks after the first inoculation but was increased after the second (Fig.5A, and data not shown). No further increase was obtained after a third immunization (data not shown). Sera collected prior to immunization as well as after immunization with control MVA were negative (data not shown). In four independent experiments, including the ones depicted in Fig. 5A and B, the ELISA titers were 3.8, 3.7, 2.9, and 1.9 fold higher after inoculation with MVA/89.6T than after MVA/89.6 (p= .016, p=.021, .01 and .13, respectively).

Figure 5.

Serum IgG to HIV Env induced by MVA/89.6 and MVA/89.6T and effect of different routes of inoculation. (A) Groups of 8-10 mice were inoculated with 10⁷ pfu of virus and boosted at one month by IM inoculations. At 2 weeks after the second immunization, mice were bled and serum IgG response to HIV Env determined. (B) Comparison of MVA/89.6T IM route of inoculation with ID, IO, and IR needleless inoculations. Same protocol was used as in A, but only the response after 2^nd immunization is presented. Titers are the averages from individual mice and bars represent the standard error of the mean.

Additional experiments were carried out to compare the IM route of inoculation with MVA/89.6T with intradermal (ID), intraoral (IO), and intrarectal (IR) needleless inoculations (Fig. 5B). The IM, ID and IR routes consistently gave similar IgG titers whereas IO was more variable (Fig. 5B and data not shown).

Mucosal IgG and IgA elicited by different inoculation routes

The efficacy of an HIV vaccine may depend on the ability to induce mucosal immunity. We therefore characterized mucosal antibody responses after immunization by IM, ID, IO and IR routes. 89.6 Env-specific IgG and IgA antibody titers were determined in lung washes, fecal extracts, and rectal washes at 3 weeks after the third immunization. Following IM inoculations, MVA/89.6T elicited higher antibody titers from each source than did MVA/89.6 (Fig.6). We also compared different routes of inoculation of MVA/89.6T. IM, ID, and IR routes gave high IgG in lung washes (Fig. 6A), presumably representing systemic antibody perfused through lung capillaries, but low or barely detectable IgA. Although the IO route gave the lowest IgG response (Fig. 6A), it elicited the highest IgA in lung washings. As shown in Fig. 6B, immunization by the IR route gave the highest IgG and the highest IgA in the fecal extracts. Antibody titers in the rectal washes were low and not enhanced by the IR route of immunization (data not shown). All washings and extracts from mice immunized with control MVA were negative for Env antibody (data not shown). Thus, our data demonstrated that MVA/89.6T induces a higher titer of mucosal antibody compared to MVA/89.6. In addition mucosal HIV 89.6-specific antibody responses depend on routes of MVA/89.6T administration.

Figure 6.

Mucosal IgG and IgA to HIV Env induced by MVA/89.6 and MVA/89.6T and effect of different routes of inoculations. Groups of 8-10 mice were inoculated with 10⁷ pfu of virus and boosted at one and two months by IM, and compared to ID, IO, and IR needleless inoculations. At 3 weeks after the third inoculation, lung washes (A) and fecal extracts (B) from 5 individual mice were processed as in Materials and Methods, and assayed for mucosal IgG and IgA by ELISA. Results are the averages of individual mouse titers; bars represent the standard error of the mean.

Effects of Env truncation and route of inoculation on cytolytic T lymphocyte (CTL) responses

To compare the CTL responses elicited by the different forms of Env, we immunized BALB/c mice with 2 immunizations of 10⁷ pfu of MVA/89.6 or MVA/89.6T one month apart by IM route. Two weeks after the first and second immunization, mice were sacrificed and splenocytes from individual mice were assayed directly with Env or vaccinia peptide-pulsed P815 cells. CD8+ IFN-γ+, CD8+ IFN-γ+ TNF+, and CD8+ IFN-γ+ IL-2+ cells were enumerated by flow cytometry. In addition the same splenocytes were stimulated in vitro with the Env peptide and tested in a ⁵¹Cr release assay. MVA/89.6T and MVA/89.6 induced similar CD8+IFN-γ+, CD8+ IFN-γ+ TNF+, and CD8+ IFN-γ+ IL-2+ cells by ICS staining (Fig.7A) and similar specific lytic activities as measured by chromium release (Fig.7B). This is consistent with the similar amount of Env expressed by each construct.

Figure 7.

Env specific CTL responses of spleen cells. Mice received two immunizations one month apart with 10⁷ pfu of MVA/89.6 or MVA/89.6T by the IM route. At two weeks after 1^st and 2^nd immunization, splenocytes from individual mice were assayed directly ex vivo with Env or VACV peptide-pulsed P815 cells and cytokine-expressing cells were quantitated by flow cytometry (A) or stimulated in vitro with Env peptide and assayed in ⁵¹Cr release assay (2 weeks after 2^nd immunization only) (B). Open symbols in B represent the % lysis of unpulsed P815 cells by Env stimulated splenocytes.

Additional studies were carried out to determine the optimal routes for inducing effector CTL in mucosal (Peyer’s patches) and systemic (spleen) lymphoid tissues by MVA/89.6T. BALB/c mice were immunized IM, ID, IO or IR with 10⁷ pfu of MVA/89.6T and boosted twice at 4-week intervals. The CTL responses were measured at three weeks after the last immunization. The splenic CTL responses induced by MVA/89.6T were higher by ID, IO and IR administration than by the IM route (Fig. 8A). A significant CD8+ CTL lytic activity in the Peyer’s patches, however, was elicited only after IR immunization (Fig. 8A). Negligible HIV-specific lytic activity was measured using unstimulated spleen cells (without peptide) from immunized mice (data not shown).

Figure 8.

Detection of HIV Env-specific CTL response in spleen cells or Peyer’s patches in acutely (A) or long term (B) infected mice immunized with MVA/89.6T. Mice were immunized three times with 10⁷ pfu of or MVA/89.6T by IM, ID, IO, or IR routes. Mice were sacrificed at 3 weeks (A) or 13 months (B) after the last immunization and pooled splenocytes from 5 mice were stimulated in vitro with P18-89.6R10 HIV Env peptide.

Mucosal CD8+CTL immune responses after immunization with MVA/89.6T are long lasting

Thirteen months after the last immunization of mice with MVA/89.6T, we determined the long-term memory P18-89.6A9-specific CD8+ CTL responses in spleen and in Peyer’s patches. Spleen cells from immunized mice were activated with P18-89.6A9 peptide for 7 days. The data recapitulated the functional activity of precursors of CD8+ CTL in vivo. Thus, the specific lytic responses were higher using spleen cells from mice immunized with MVA/89.6T by IR or ID routes than by the IM route (Fig. 8B). In addition, long term memory CTL were detected in Peyer’s patches of mice immunized with MVA/89.6T by the IR route (Fig. 8B).

To quantify the memory CD8+ CTL elicited by the different routes of immunizations with MVA/89.6T, a P18-89.6A9/H-2Dd tetramer that can bind to the T-cell receptor of antigen-specific CTL was used. After 3 or 7 days of in vitro stimulation of spleen cells with the P18-89.6A9 peptide, we double-stained cells with P18-89.6A9/H-2Dd tetramer plus anti-CD8 antibody and compared the number of double positive cells in each group. The number of tetramer-positive cells after 7 days stimulation with peptide was increased dramatically compared to 3 days of stimulation (data not shown). The number of activated memory CD8+ CTL after IR or ID immunizations were slightly higher than after IO or IM immunizations (Fig. 9). The high intensity tetramer positive cells from mice immunized by IR or ID routes correlated with the high memory CTL response at 13 months in the functional ⁵¹Cr release assay (Fig. 8B).

Figure 9.

Detection of long term memory CTLs as quantitated by P18-89.6A9/H-2D^d tetramer staining thirteen months after MVA/89.6T immunization. Mice were immunized three times with 10⁷ pfu of MVA/89.6T by IM, ID, IO, or IR routes. After in vitro stimulation of spleen cells (pooled splenocytes from 5 mice/group) with P18-89.6R10 HIV env peptide, cells were double-stained cells with P18-89.6A9/H-2D^d tetramer plus anti-CD8 antibody and the number of double positive cells in each group compared. CON are spleen cells from unimmunized mice.

Discussion

The purpose of this study was to characterize MVA/89.6T, a spontaneous mutant derived from recombinant MVA/89.6 expressing a full-length HIV Env, and to compare the immunogenicities of the two viruses. Although MVA/89.6 had been successfully used to elicit CD8+ CTL responses in mice (Belyakov et al., 1999; Belyakov et al., 1998c), the recombinant virus was troublesome because viral foci that did not stain with HIV antiserum were noted after serial passages. The gradual accumulation of non-expressing mutants had been detected with other recombinant MVAs, suggesting that they can have a selective growth advantage when the expressed protein exhibits some toxicity. Serendipitously, however, we observed foci that were larger and stained more brightly with HIV antiserum than the majority. Moreover, the mutant was more genetically stable, as non-staining foci were not observed. Biochemical characterization revealed a frame-shift mutation after amino acid 738 in the cytoplasmic tail of Env. The more intense staining of the MVA/89.6T foci, compared to those of MVA/89.6, was consistent with greater surface expression of Env even in non-permissive cells as determined by flow cytometry. Another independent truncation mutant was isolated. However, the frame shift mutation was in gp120, making it undesirable for vaccine purposes.

The cytoplasmic tails of primate lentivirus Env proteins are about 150 amino acids, considerably longer than the 20 to 40 amino acids for most other retroviruses (Hunter and Swanstrom, 1990). The role of this domain is not understood and naturally occurring Env truncations have been observed for SIV and HIV-2 (Chakrabarti et al., 1989; Hirsch et al., 1989; Johnston et al., 1993; Kodama et al., 1989; Mulligan et al., 1992; Tsujimoto et al., 1988), apparently providing a selective advantage under certain conditions of in vitro propagation. When expressed by recombinant methods, these truncated Env proteins as well as those of HIV can provide higher surface expression, greater syncytium formation, and better exposure of conserved regions of the ectodomain than the full-length forms (Ashorn et al., 1990; Earl et al., 1991; Edwards et al., 2002; Haffar et al., 1990; Mulligan et al., 1992; Ritter et al., 1993; Rowell et al., 1995; Spies and Compans, 1994; Wyss et al., 2005). Consistent with the latter findings, the truncated Env described here exhibited increased surface expression, fusion-competence and reactivity with a panel of MAbs. Previous studies identified conserved sequences within the cytoplasmic domain of gp41 that act as retrieval signals, inducing endocytosis (Berlioz-Torrent et al., 1999; Bultmann et al., 2001; Byland et al., 2007; Rowell et al., 1995). These include a membrane proximal tyrosine-based motif and three distal motifs containing a hydrophobic sequence in one, dileucine motif in one, and a C terminal dileucine motif. Our finding of increased surface expression of the spontaneously truncated Env emphasizes the importance of the distal retrieval signals, since the tyrosine motif was still present. Although the HIV Env cytoplasmic domain has been reported to influence the incorporation of Env into virions, we found that the truncated Env was associated with virus-like particles when SIV gag-pol was co-expressed (Earl et al., 2002).

Comparisons of immune responses between mice that received DNA vaccines encoding truncated or full length SIV or HIV Env demonstrated higher levels of specific antibody elicited by the truncated forms, which was attributed to greater expression (Liu et al., 1996; Vzorov et al., 1999). In our study, we found that the antibody response to HIV-1 89.6 Env induced by MVA/89.6T in mice by IM route was greater than that achieved with MVA/89.6, and that MVA/89.6T could induce similar levels of antibody using several different routes of inoculation. The higher antibody response could be explained by the increased surface expression of the truncated Env compared to the full-length protein. However, other factors resulting from the lower toxicity of the truncated Env in MVA-infected cells may also have contributed to the beneficial effect. MVA/89.6 and MVA/89.6T induced similar systemic Env-specific CTL responses as assayed by ICS and by lytic assays. This is not surprising since both MVA/89.6 and MVA/89.6T express similar amounts of Env as determined by quantitative Western and FACS analysis.

We previously showed that MVA/89.6 induced both systemic and mucosal CTL responses to Env when administered IR to mice (Belyakov et al., 1998c). Moreover, the IR route overcame prior immunity to VACV that had been administered systemically (Belyakov et al., 1999). Because HIV infections commonly involve the mucosal route of entry, we compared the antibody responses as well as CTL induced by the MVA/89.6T recombinant given IO, IR, ID and IM. We found that the IO and IR routes of inoculation stimulated production of more IgA in the lung washes and rectal fecal extracts, respectively, than occurred by ID or IM routes. In contrast, the serum IgG responses were similar by IM, ID and IR but more variable by the IO route. The splenic CTL responses induced by MVA/89.6T were higher when inoculated by ID, IO and IR routes than by the IM route. Previously we demonstrated that long-lasting protection against mucosal viral transmission could be accomplished by CD8+ CTLs, which must be present in the mucosal site of exposure (Belyakov et al., 1998a). A significant P18-89.6A9-specific CD8+ CTL lysis in the Peyer’s patches, however, was elicited only after IR immunization with MVA/89.6T. Moreover, the memory responses were long lasting as they were detected at 13 months (last experimental time point) after the last vaccination in mice.

While correlates of immunity in HIV infection are still not delineated, both neutralizing antibodies and CTL likely play important roles in suppressing HIV infection (McMichael, 2006). Our data showing binding of conformational and neutralizing MAbs to the truncated Env suggested that the removal of a portion of the cytoplasmic domain did not have an adverse affect on the antigenicity of the extracellular domain. However, the HIV neutralizing antibody titers in the mice were not high enough to determine the effect of truncation on immunogenicity (D. Montefiori, personal communication). Based on the studies described here, we constructed recombinant MVA containing the truncated HIV 89.6 Env as well as the SIV gag-pol for vaccine studies in monkeys. Strain-specific neutralizing antibody was detected after three or four immunizations (Earl et al., 2002). Furthermore, intradermal and intramuscular immunizations with the recombinant MVA alone or preceded by a DNA vaccine protected against an IR challenge with the pathogenic SHIV 89.6P (Amara et al., 2001; Amara et al., 2002b; Earl et al., 2002). Moreover, poorer protection was obtained when the MVA contained only the gag-pol component of the SHIV vaccine (Amara et al., 2002a). These data have encouraged us to construct additional stable recombinant MVA vaccines (L. Wyatt and P. Earl, unpublished) with truncated HIV Env that have and will soon enter clinical trials.

Materials and Methods

Cell cultures and viruses

The protocols used for generation of recombinant viruses and their identification by immunostaining have been described in detail and were carried out in secondary chicken embryo fibroblasts (Earl et al., 1998). BS-C-1 cells (ATCC CCl-26) were grown in modified Eagle’s minimal essential medium (EMEM) supplemented with 10% heat inactivated fetal bovine serum (FBS) (Hyclone, Logan, UT), 2 mM L-glutamine (Invitrogen, Carlsbad, CA), 100 U/ml penicillin, and 100 μg/ml streptomycin sulfate (Invitrogen). Suspension cultures of HeLa S3 cells (ATCC-CCL-2.2) were grown in MEM for spinner cells (Quality Biologicals, Inc., Gaithersburg, MD) containing 5% heat-inactivated equine serum (Hyclone).

Construction of recombinant MVA expressing the full-length Env of HIV strain 89.6 (referred to as MVA/89.6) was previously described (Belyakov et al., 1998c). Briefly, the 89.6 env reading frame was modified by making silent mutations within potential poxvirus transcriptional termination sequences (Earl et al., 1990), inserted into deletion II of MVA, and regulated by the VACV modified early/late H5 promoter (Wyatt et al., 1999). Unless otherwise mentioned, MVA/89.6T and MVA/89.6 stocks used for comparative purposes contained >99% and >90% Env-expressing virus, respectively.

Antibodies

Hybridomas that secrete T8, T43, and D19 MAbs against linear epitopes in gp120 were derived from mice that had been immunized with clade B HIV isolate BH8 oligomeric Env protein (Earl et al., 1994). Similarly derived D61, T32, and T30 MAbs recognize linear epitopes in gp41 while D43, T-33 and T-40 are directed against conformational epitopes in gp41(Earl et al., 1997). Polyclonal antibody R2144 was made in a rabbit immunized with gel purified gp140 IIIB Env (P. Earl, unpublished). Anti-VACV polyclonal antiserum was made in rabbits. Human neutralizing MAbs derived from HIV infected individuals directed against gp120 are: 2G12 (Trkola et al., 1996), b12 (Burton et al., 1994) and 447-52D (Gorny et al., 1992), while 2F5 (Purtscher et al., 1994) is directed against gp41.

Expression of HIV Env

Radioimmunoprecipitation of virus infected cell lysates has been described in detail previously (Wyatt et al., 1996). Briefly, virus infected BS-C-1 cells were labeled with 100 μCi/ml of [³⁵S]methionine (New England Nuclear, Boston, MA) in methionine-free EMEM containing 5% dialyzed FBS overnight and then lysed. The cell lysates were incubated overnight with either antibody R2144 or D61, followed by 20% protein A-Sepharose suspension. Immunoprecipitated proteins were resolved by electrophoresis in a 10% SDS-polyacrylamide gel, and visualized by autoradiography.

For quantitative Western blotting, BS-C-1 cells were infected with MVA/89.6 and MVA/89.6T at a multiplicity of 10 and harvested at 24 h. Two-fold dilutions of lysates were subjected to SDS-polyacrylamide gel electrophoresis (PAGE), transferred to a nitrocellulose membrane, and incubated with mouse T8 mAb and secondary R-Dye-800CW conjugated donkey anti-mouse IgG (H+L) (Rockland Immunochemicals, Inc., Gilbertsville, PA). Reactive bands were quantitated using the Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, Nebraska). Band intensities from four consecutive lysate dilutions were subjected to linear regression analysis using Prism software (Graph Pad, San Diego, CA) and the slope determined.

Genetic stability of the HIV env genes in MVA/89.6 and MVA/89.6T was assessed by repeated passages in CEF. Virus was passaged at an MOI of approximately 1 pfu/cell in CEF, grown for three days, harvested, and reinoculated at similar MOI onto freshly made CEF for each passage. Yields of passaged viruses were titrated in CEF, using medium containing 0.5% methylcellulose. After 3 days incubation, the infected cells were fixed and stained with mAbs reactive to Env (Earl et al., 1998).

Flow cytometry

Surface and total expression of HIV Env by recombinant MVA/HIV viruses were quantitated by fluorescent activated cell sorting (FACS). HeLa S3 spinner cells were infected with MVA or recombinant MVA at a multiplicity of 5 and incubated at 37°C in 5% CO₂. At 15-18 h post-infection, the cells were washed with phosphate buffered saline (PBS) containing 3% FBS (PBS-FBS) and 5×10⁵ cells were dispensed into each well of a 96 well plate (Corning #3799). Cells were incubated for 1 h at 4°C with a gp120 or gp41 antibody (as described in Results) diluted in PBS-FBS, and followed by PE conjugated goat anti-mouse antibody (BD Biosciences, San Jose, CA) or FITC conjugated anti-human IgG (BD BioSciences) for 30 min at 4°C. Following two PBS-FBS washes, the cells were resuspended in 2% paraformaldehyde in PBS. For permeabilized cell studies, 0.1% saponin was added to PBS-FBS at all steps. Utilizing a FACSCalibur with CellQuest software, 10,000 cells were acquired and analyzed using FlowJo software (Tree Star, Inc., Ashland, OR). In each experiment anti-VACV rabbit IgG, followed by FITC conjugated goat anti-rabbit Ig (BD BioSciences) was used as a control to assure that cells were equally infected.

HIV-1 Env-mediated cell fusion assay

Env-mediated fusion was quantified as previously described (Nussbaum et al., 1994) with modifications. Effector HeLa cells were infected with MVA/89.6 or MVA/89.6 T, and vTF7.3, expressing bacteriophage T7 RNA polymerase at a multiplicity of 10. Similarly, target HeLa cells were infected with vCB3 expressing membrane-associated CD4, vCB21R expressing lac Z reporter gene and one of two coreceptors (vCBYFI expressing CXCR4 or vCCR5 expressing CCR5). Following incubation at 31°C overnight and washing, cells were adjusted to 1×10⁶ cells/ml and 100 μl of infected effector and target cells were mixed together in duplicate wells. After incubation of plates for 3 h at 37°C, fusion was quantified by measurement of β-galactosidase activity using substrate CPRG (Roche Diagnostic, Indianapolis, IN) in NP-40 treated cell lysates with a spectrophotometer. Cell fusion activity was expressed as units of β-galactosidase/min.

Mouse immunizations

Groups of 6-10 BALB/c female mice (approximately 6-10 weeks old) were inoculated with 1×10⁷ pfu of MVA or recombinant MVA by the following routes: IM, 0.1ml by needle; ID, 0.1ml by a Biojector 2000 needleless injector (Bioject Medical Technologies, Inc., Portland, OR); IO, 50 μl by a Syrijet Mark II needleless injector (Mizzy, Inc., Cherry Hill, NJ) into buccal mucosa; and IR, 150 μl through an umbilical vein catheter inserted about 4 cm (Belyakov et al., 1998a). For ID and IO inoculations, mice were anesthetized with Avertin (stock solution of 10 g of 2,2,2 tribromethanol in 10 ml of tertiary amyl alcohol). Avertin was diluted 1:80 and used at a dose of 0.6ml/25g mouse. For ID inoculations using Biojector, the mice were shaved and given 2 injections, one on either side of the base of the tail, lateral to the spine. For IO injections, the Syrijet was aimed at the buccal mucosa. For IR inoculations, the inhalation anesthesia methoxyflurane, (Pitman-Moore, Inc., Mundelein, IL) was used. Mice were immunized on day 0 and boosted at 1 month and 2 months. Mice were bled on day -2, and again two weeks after each booster inoculation.

Mouse mucosal washes collections

Samples of lung lavage, rectal washes, and fecal extracts were taken from individual mice for antigen-specific IgG and IgA antibody assays 3 weeks after final immunizations. First, lung lavage samples were taken under methoxyflurane inhalation anesthesia. The lungs were flushed with 1 ml of PBS; this procedure was repeated two times. The lung lavage fluid collected was then centrifuged at 1,200 rpm for 5 min at 4°C to remove any cellular debris. After collecting lung lavage samples mice were sacrificed. Feces were collected into a microcentrifuge tube from the intestine and extracted in ratio of 100 mg feces/1ml PBS; the mixture was vortexed vigorously until the solution was homogenous. Fecal samples were then spun in a microcentrifuge at 5000xg for 15 min, and supernatant fluid from centrifuged fecal homogenates was used for ELISA assay. For rectal washes, 1ml of PBS was gently run though intestinal tissue from rectum to caecum. The rectal lavage fluid collected was then centrifuged at 1,200 rpm for 15 min to remove any tissue or fecal matter, and supernatants were collected. All samples were stored for a short time at -20°C until assayed and individually assayed by ELISA.

ELISA assays

To measure the levels of IgG and IgA in the sera of mice and mucosal IgG and IgA in lung and rectal washings and fecal pellets, 96 well U-bottom plates (Immulon-2 (Dynex Technologies, Chantilly, VA) were coated overnight with sheep antibody to the C-terminus of gp120 (International Enzymes, Inc, Fallbrook, CA) at 0.5 μg/ml in bicarbonate buffer (Roche Molecular Biochemicals, Indianapolis, IN), followed by purified 89.6 gp140 as described (Earl et al., 2002). Two-fold serial dilutions of sera or sample suspensions were incubated for 2 h at room temperature followed by horseradish peroxidase-conjugated anti-mouse IgG (Roche Molecular Biochemicals) or horseradish peroxidase-conjugated anti mouse IgA (Southern Biotechnology Associates, Inc., Birmingham, AL). BM Blue substrate (Roche Molecular Biochemicals) was added for 30 minutes and absorbency read at 370 and 492nm.

Intracellular cytokine staining

Intracellular cytokine staining was performed two weeks after the first and second immunization using splenocytes prepared from individual animals. 2×10⁶ splenocytes were mixed with 1×10⁵ P815 cells that had been pulsed with either the Env V3 peptide IGPGRAFYTT or the VACV E3L peptide VGPSNSPTF. After 2 h at 37°C, brefeldin A (Sigma, St. Louis, MO) was added to a concentration of 10 μg/ml. Four h later, cells were placed at 4°C overnight. Cells were incubated with anti-CD16/32, clone 2.4G2 (gift of K. Grebe), then stained with peridinin chlorophyll-a protein (PerCP) conjugated anti-CD8 (clone 53-6.7, BD Pharmingen). After fixation and permeabilization, cells were stained with allophycocyanin (APC) conjugated anti-interferon gamma (IFN-γ) (clone XMG1.2) and either fluorescein isothiocyanate conjugated anti-interleukin 2 (IL2) (clone JES6-5H4) or anti-tumor necrosis factor (TNF) (clone MP6-XT22) (BD Pharmingen), washed, and resuspended in 2% paraformaldehyde. About 100,000 cells were acquired on a FACSCalibur (San Jose, CA) using Cell Quest and analyzed with FlowJo software (Tree Star, Inc, Ashland, OR) to determine the percentage of CD8+ splenocytes that expressed IFN-γ, IL2, and TNF.

Cytotoxic T lymphocyte assay

For the data in Fig. 7, spleens from individual mice were disrupted by dounce homogenization and subsequent passage through sterile screens. Erythrocytes were lysed with ACK lysis buffer and then washed with RPMI 1640 containing 10% FBS. For the data in Fig. 8, spleens from groups of 5 mice were aseptically removed and pooled and single-cell suspensions prepared by gently teasing them through sterile screens. The erythrocytes were lysed in Tris-buffered ammonium chloride and the remaining cells washed extensively in RPMI 1640 containing 2% fetal bovine serum.

Peyer’s patches were carefully excised from the wall of the small intestine, pooled, and dissociated into single cell suspensions by enzymatic digestion with collagenase type VIII (300 U/ml; Sigma, St Louis, Mo.) and DNAse I (3 U/ml; Sigma) for 60 min. Cells were collected, washed, resuspended in complete medium (CM; RPMI-1640 supplemented with 10% FBS, 10 U penicillin (Invitrogen, Carlsbad, CA), 100 μg/ml streptomycin (Invitrogen), 2 mM glutamine (Invitrogen), 50 μM ß-mercaptoethanol (Invitrogen), and 20 mM Hepes (BioWhittaker), layered onto a discontinuous density gradient containing 75% and 40% Percoll (Pharmacia Inc., Uppsala, Sweden), and centrifuged at 600 X g for 25 min. The interface band containing cells between the 75% and 40% Percoll was carefully removed and washed with RP-2. The resulting population was > 90% viable lymphocytes with a cell yield of 1 × 10⁷ lymphocytes/mouse. Most Peyers’s patches CD3⁺ T cells isolated from normal mice were CD4⁺, while CD3⁺CD8⁺ T cells were less frequent. Collagenase VIII digestion does not alter the expression of CD3, CD4, or CD8 on splenic T cells treated with this enzyme (Belyakov et al., 1998b)

Immune cells from the spleen or Peyer’s patches (5 × 10⁶ per ml) in 12-well culture plates were incubated with 1 μM synthetic CTL epitope peptide (Env IGPGRAFYTT for data in Fig.7) or P18-89.6R10 (IGPGRAFYAR) or P18-89.6A9 (IGPGRAFYA) (for data in Fig.8), the minimal epitope of P18-89.6 (Belyakov et al., 1998c), in complete T cell medium (CTM): RPMI 1640 containing 10% fetal bovine serum, 2 mM L-glutamine, penicillin (100 U/ml),streptomycin (100 mg/ml), and 5 × 10^-5 M 2-mercaptoethanol. On day 3, we added 10% concanavalin A supernatant-containing medium (“T-stim”, Collaborative Biomedical Products, Bedford, MA) as a source of IL-2. Cytolytic activity of CTL lines was measured by a 4-h assay with ⁵¹Cr labeled P815 targets tested in the presence or absence of Env peptide (1μM). For testing the peptide specificity of CTL, ⁵¹Cr-labeled P815 targets were pulsed for 2 h with peptide at the beginning of the assay. The percent specific ⁵¹Cr release was calculated as 100 X (experimental release-spontaneous release)/(maximum release - spontaneous release). Maximum release was determined from supernatants of cells that were lysed by addition of 5% Triton-X 100. Spontaneous release was determined from target cells incubated without added effector cells (Belyakov et al., 2000; Belyakov et al., 2001b). Because our target cells (p815) express only MHC class I, not class II, we expected that lysis was mediated by CD8+ class MHC-restricted T cells.

Fluorescent MHC/peptide/avidin tetramer staining of CD8+ CTL phycoerythrin-conjugated H-2Dd tetramers were prepared as previously described (Belyakov et al., 2000; Dzutsev et al., 2007), either folded around P18-89.6A9 peptide or around a “motif” peptide (AGPARAAAL) that bound tightly to the MHC molecule but did not interact with T cell receptor specific for the H-2Dd/P18-89.6A9 peptide complex. Phycoerythrin-conjugated H-2Ld tetramers presenting an analogous “motif” peptide were prepared similarly. T cells were washed in FACS buffer, blocked for 20 min on ice with Fc block (anti-CD16/CD32), and then stained with either anti-CD8 (Pharmingen, SanDiego, CA), or 1 ml tetramer PE for 60 min in FACS buffer on ice. After washing with cold FACS buffer, the cells were analyzed by flow cytometry.

Statistical Analysis

Statistical differences between responses elicited by full length and truncated recombinant immunizations of mice were assessed by Mann-Whitney using Prism software (Graph Pad, San Diego, CA).