A Candidate HIV/AIDS Vaccine (MVA-B) Lacking Vaccinia Virus Gene C6L Enhances Memory HIV-1-Specific T-Cell Responses

The vaccinia virus (VACV) C6 protein has sequence similarities with the poxvirus family Pox_A46, involved in regulation of host immune responses, but its role is unknown. Here, we have characterized the C6 protein and its effects in virus replication, innate immune sensing and immunogenicity in vivo. C6 is a 18.2 kDa protein, which is expressed early during virus infection and localizes to the cytoplasm of infected cells. Deletion of the C6L gene from the poxvirus vector MVA-B expressing HIV-1 Env, Gag, Pol and Nef antigens from clade B (MVA-B ΔC6L) had no effect on virus growth kinetics; therefore C6 protein is not essential for virus replication. The innate immune signals elicited by MVA-B ΔC6L in human macrophages and monocyte-derived dendritic cells (moDCs) are characterized by the up-regulation of the expression of IFN-β and IFN-α/β-inducible genes. In a DNA prime/MVA boost immunization protocol in mice, flow cytometry analysis revealed that MVA-B ΔC6L enhanced the magnitude and polyfunctionality of the HIV-1-specific CD4+ and CD8+ T-cell memory immune responses, with most of the HIV-1 responses mediated by the CD8+ T-cell compartment with an effector phenotype. Significantly, while MVA-B induced preferentially Env- and Gag-specific CD8+ T-cell responses, MVA-B ΔC6L induced more Gag-Pol-Nef-specific CD8+ T-cell responses. Furthermore, MVA-B ΔC6L enhanced the levels of antibodies against Env in comparison with MVA-B. These findings revealed that C6 can be considered as an immunomodulator and that deleting C6L gene in MVA-B confers an immunological benefit by enhancing IFN-β-dependent responses and increasing the magnitude and quality of the T-cell memory immune responses to HIV-1 antigens. Our observations are relevant for the improvement of MVA vectors as HIV-1 vaccines.

The highly attenuated VACV strain MVA is one of the most promising vectors to be used as an effective vaccine against HIV-1 [16]. MVA has an excellent safety profile, and MVA recombinants expressing HIV-1 antigens induce protection after simian/human immunodeficiency virus (SHIV) challenge, and elicit strong, broad, polyfunctional and durable immune responses to HIV-1 antigens in different animal models and humans trials [ [17,18,19,20,21,22], for a review [23]].
We have previously constructed a recombinant MVA expressing codon-optimized Env as monomeric gp120 and the poly-protein Gag-Pol-Nef of HIV-1 from clade B (referred as MVA-B), that in DNA prime/MVA boost protocols in mice induced strong immune responses to HIV-1 antigens [17,18,20]. In macaques, a similar MVA construct expressing Env (gp120 from SHIV 89.6P ) and Gag-Pol-Nef (from SIV mac239 ) showed strong specific CD4 + and CD8 + T-cell immune responses with a bias for CD8 + , and high protection after challenge with SHIV 89.6P [22]. Furthermore, the expression of HIV-1 antigens from MVA-B selectively induced in human dendritic cells (DCs) the expression of different cellular genes that might act as regulators of immune responses to HIV-1 antigens [24] and MVA-B-infected DCs co-cultured with autologous T lymphocytes induced a highly functional HIV-1-specific CD8 + Tcell responses including proliferation, secretion of IFN-c, IL-2, TNF-a, MIP1b, MIP1a, RANTES and IL-6, and strong cytotoxic activity against autologous HIV-1-infected CD4 + T lymphocytes [25]. Based on these previous results, MVA-B has recently entered a phase I clinical trial in healthy volunteers in Spain. However, more efficient poxvirus MVA-B vectors that enhance the magnitude, breath, polyfunctionality and durability of the immune responses to HIV-1 antigens are desirable. This is particularly relevant when a single immunogen is desirable for mass vaccination purposes to simplify the immunization protocols and reduce manufacturing cost. Deletion in the vector backbone of MVA-B of known or suggested immunomodulatory VACV genes, which antagonize host specific immune responses, is a general strategy that could enhance immunogenicity of the vector against HIV-1 antigens.
In this study, we have generated a new HIV-1 vaccine candidate, termed MVA-B DC6L, which contains a deletion in the vector backbone of MVA-B of the VACV C6L gene. We have analyzed the expression of C6, the innate immune responses elicited by MVA-B DC6L in human THP-1 cells and monocyte-derived DCs (moDCs) and we have examined if the deletion of C6L gene in the candidate HIV/AIDS vaccine vector MVA-B could improve the humoral and T-cell memory immune responses to HIV-1 antigens in mice. We showed that C6 is expressed early during viral infection and localizes in the cytoplasm of infected cells. MVA-B DC6L replicates in cell culture at the same level as parental MVA-B, indicating that C6 is not essential for virus replication. Furthermore, MVA-B DC6L up-regulated the expression of IFN-b and IFN-a/b-inducible genes (IFIT1 and IFIT2) in human THP-1 cells and moDCs, suggesting that C6 inhibits the IFN-b signalling pathway by blocking some unknown component involved in the induction of IFN-b. In DNA prime/MVA boost immunization protocols in mice comparing MVA-B DC6L and MVA-B, MVA-B DC6L significantly enhanced the magnitude and polyfuncionality of the HIV-1specific CD4 + and CD8 + T-cell memory immune responses, which are mostly mediated by CD8 + T cells of effector phenotype in both immunization groups. HIV-1-specific CD4 + T-cell memory responses induced by MVA-B and MVA-B DC6L were preferentially Env-specific. However, while MVA-B elicited more Env-and Gag-specific CD8 + T-cell memory responses, MVA-B DC6L induced preferentially Gag-Pol-Nef (GPN)-specific CD8 + T-cell memory responses. Moreover, MVA-B DC6L enhanced the levels of antibodies against HIV-1 Env in comparison to MVA-B.
Altogether, our findings revealed that C6 is a new VACV immunomodulatory protein, and that deletion of C6L in MVA represents an attractive alternative to increase the immunogenicity of MVA vaccine candidates.

Generation and in vitro characterization of MVA-B DC6L
The function of the VACV C6L is unknown, although it is predicted that have an immunomodulatory role [3]. To analyze the possible immunomodulatory function of C6L, we have constructed an MVA-B deletion mutant lacking VACV gene C6L (MVA-B DC6L, see Materials and Methods) from the previously described recombinant MVA-B (expressing HIV-1 Env, Gag, Pol and Nef antigens from clade B) [18]. The diagram of the MVA-B DC6L deletion mutant is shown in Figure 1A. PCR using primers for the C6L locus confirmed the deletion of the C6L gene in MVA-B DC6L ( Figure 1B). Deletion of C6L from MVA-B DC6L was also confirmed by DNA sequencing (data not shown). In addition, analysis by Western blot confirmed that MVA-B DC6L expresses HIV-1 antigens BX08 gp120 and IIIB GPN at the same level as their parental virus MVA-B ( Figure 1C).

C6 is non-essential in cell culture
The mere isolation of MVA-B DC6L deletion mutant demonstrated that the C6 protein is not essential for MVA replication. To determine whether deletion of C6L altered virus multiplication, we compared the growth of MVA-B DC6L and MVA-B in DF-1 cells. Kinetics studies revealed that deletion of C6L in the MVA-B genome did not affect virus replication. Hence, C6L is not essential for virus propagation in cultured cells ( Figure 1D). Furthermore, similar to the parental virus MVA-B or MVA, MVA-B DC6L is an attenuated virus which does not replicate in mammalian cells ( Figure S1).

C6 is expressed early in infection
Western blot analyses using rabbit polyclonal antibodies raised against C6 (produced as described in Material and Methods) identified a 18.2 kDa protein in DF-1 cells infected with MVA-B, but not with MVA-B DC6L (Figure 2A). We then determined when C6 is expressed in DF-1 cells infected with WR and MVA ( Figure 2B). C6 expression was detected at 3 hours post-infection, and increased with time for at least 22 hours. Treatment of DF-1 cells at the onset of infection with cytosine arabinoside (AraC), an inhibitor of viral DNA replication and therefore late protein expression, did not abrogate C6 expression, indicating that C6L is an early gene. The increased amount of early C6 protein at 22 h post-infection compared to what is produced when viral DNA replication is inhibited by AraC ( Figure 2B), is probably due to enhanced stability of the protein after virus DNA synthesis and to the contribution of progeny virus.
The intracellular localization of C6 was examined by immunofluorescence in DF-1 cells infected with different VACV strains ( Figure 2C). C6 was detected in the cytoplasm, presumably in viral factories, of DF-1 cells infected with WR, MVA and MVA-B, but not in MVA-B DC6L. The reduced fluorescence intensity of C6 (in green) indicates low levels of protein expression in comparison with the late protein A27 (in red).

MVA-B DC6L up-regulates IFN-b expression in human macrophages and dendritic cells
As a first step to determine whether C6 impairs the response of innate immune cells to MVA-B, we examined by real time PCR the expression of IFN-b, IFN-b-induced genes (IFIT1, IFIT2) and chemokines by human THP-1 macrophages infected for 1, 3 and 6 hours with MVA, MVA-B and MVA-B DC6L ( Figure 3A). Compared to MVA and MVA-B, MVA-B DC6L (5 PFU/cell, Figure 3A and 1 PFU/cell, data not shown) markedly upregulated IFN-b as well as IFIT1 and IFIT2 expression in THP-1 cells. MVA-B DC6L also increased the expression of MIP-1a and RANTES, but not that of IL-8 and IP-10 ( Figure 3A and data not shown).
To confirm that C6 impaired IFN-b and IFN-b-dependent gene expression in innate immune cells, we infected human moDCs  [72] and [73]. The different regions are indicated by capital letters. The right and left terminal regions are shown. Below the map, the deleted or fragmented genes are depicted as black boxes. The deleted C6L gene is indicated. The HIV-1 Gag-Pol-Nef (from isolate IIIB) and gp120 (from isolate BX08) clade B sequences driven by the synthetic early/late (sE/L) virus promoter inserted within the TK viral locus (J2R) are indicated (adapted from [18]).  Figure 3B). We used low virus doses, since MVA induces apoptosis of human moDCs [26]. Similarly to the results obtained with human THP-1 cells, MVA-B DC6L strongly increased IFN-b expression compared to MVA and MVA-B in moDCs. Whereas the three viruses used at 0.2 PFU/ml similarly stimulated IFIT1 and IFIT2 mRNA expression in moDCs, MVA-B DC6L was a much more potent inducer than MVA and MVA-B at lower infective doses (0.002 PFU/ml, Figure 3B). Furthermore, MVA-B DC6L stimulated the release by moDCs of much higher levels of IFN-b ( Figure 3C) and bioactive type I IFNs than MVA and MVA-B ( Figure 3D).
Thus, deletion of C6L in the MVA-B genome promotes IFN-b production, suggesting that C6 interferes with the signalling pathway controlling IFN-b gene expression in innate immune cells.
The overall HIV-1-specific immune response at 53 days postboost was mainly mediated by CD8 + T cells (70%-85%) ( Figure 4C Figure 4B and 4C). In both immunization groups, HIV-1-specific CD8 + T cells were mainly of the EM (60.5-63%) and TEMRA (37%-39.5%) phenotypes ( Figure 4D). All HIV-1-specific CD4 + T cells were of the EM phenotype in the DNA-B/MVA-B group. Although most of HIV-1-specific CD4 + T cells (82.8%) were of the EM phenotype in the DNA-B/MVA-B DC6L group, a substantial proportion (17.2%) of cells expressed the TEMRA phenotype ( Figure 4D). No CM T cells producing IFN-c and/or IL-2 were detected in both immunization groups ( Figure 4D). To have a detailed assessment of the quality of T-cell memory responses, we next evaluated the production of IFN-c and/or IL-2 by HIV-1-specific CD4 + and CD8 + T-cell memory cells ( Figure 5

Discussion
The MVA vector, despite of its attenuated phenotype, still contains genes that encode proteins that can interfere with host immune responses to viral infection [34], and it is described that deletion of immunomodulatory proteins in orthopoxviruses can enhance immune responses ( [17,34,35,36,37,38,39]). The function of some of these genes, like the VACV gene C6L, is unknown. We report here the immunomodulatory role of C6L, showing the effects of the C6 protein on virus replication, innate immune sensing and immunogenicity in vivo.
MVA-B, the attenuated VACV vector MVA expressing the clade B HIV-1 antigens Env, as monomeric gp120, and Gag, Pol and Nef, as a polyprotein of about 160 kDa is considered a vaccine candidate against HIV/AIDS [18] based on preclinical studies in different animal models [17,18,20,22] and on gene signatures triggered in human DCs infected with MVA-B, where the expression of HIV-1 proteins induced the expression of immunomodulatory molecules such as cytokines, cytokine receptors, chemokines, chemokine receptors and molecules involved in antigen uptake and processing [24]. Moreover, human DCs exposed to MVA-B induced highly functional HIV-1-specific CD8 + T-cell responses in HIV-1 infected individuals [25]. Thus, due to the good immunogenicity behavior of MVA-B, a prophylactic phase I clinical trial was initiated in Spain.
To improve the immunogenicity elicited by MVA-B and to investigate the possible immunomodulatory role of C6L we have removed from the MVA-B viral genome the C6L gene, generating the deletion mutant termed MVA-B DC6L. First, we showed in cultured cells that MVA-B DC6L does not express the C6 protein, but efficiently produced the four HIV-1 antigens (Env, Gag, Pol and Nef) in a stable manner and at the same level as MVA-B during the course of virus infection. Also, MVA-B DC6L replicates similarly to MVA-B in cultured cells, indicating that deletion of C6L has no effect on virus propagation. Therefore, C6L is not essential for viral replication in cell culture. Moreover, similar to MVA-B, MVA-B DC6L maintains an attenuated phenotype and does not replicate in mammalian cells. Western blot analyses demonstrated that C6 is expressed early in cells infected with the VACV strains WR and MVA. This early expression profile is consistent with genome-wide transcriptome analyses that detected C6 mRNA 30 minutes post-infection [2]. Most VACV immunomodulatory proteins are expressed early during infection, and the early expression pattern of C6 suggests that it is involved in immune evasion as we confirmed in experiments using human macrophages and DCs. In addition, C6 localizes to the cytoplasm of infected cells, opening the possibility that C6 modulates, directly or indirectly, intracellular signalling pathways controlling immune responses.
Yeast two-hybrid and pull-down assays revealed that VACV C6 protein binds to three host human cell proteins [14]. However, none of these proteins seems to be directly related with the host immune response. One of the C6 binding partners is programmed cell death 6 interacting protein (PDCD6IP/ALIX), which has been involved in the regulation of apoptosis, cytokinesis and HIV-1 budding. VACV C6 also interacts with keratin 4 (KRT4), present in intermediate filaments, and which also binds IMV surface protein A27. C6 protein has also been detected in a low proportion in intracellular mature virions [13], similar to other proteins of the poxvirus family Pox_A46 (A46). One possible reason for presence of C6 in the virion could be that C6 is necessary for viral cycle early after virus entry or that C6 have a function in IMV-cell attachment, fusion, and/or microtubule transport through their interaction with KRT4. Finally, C6 also binds to troponin I, skeletal, fast (TNNI2), a co-activator of estrogen receptor-related receptor a (ERRa), suggesting that C6 could have a role in ERRa-mediated transcriptional activity. Additional experiments will be required to decipher the relationship between the C6 interaction with binding partners and C6 immunomodulatory function.
To examine the immunomodulatory role of C6L, we charac-   post-infection and at low virus doses, probably because long-term viral exposure and high viral doses induce extensive apoptosis [26]. Previous experiments revealed in macrophages infected with MVA a critical role for TLR2-TLR6-MyD88 in the production of IFN-b-independent chemokines and of MDA-5-IPS-1 in the production of IFN-b-dependent chemokines [40]. Phosphorylation of IRF3, IRF7 and STAT-1, which are essential for the transcription of the IFN-b gene (IRF3) and critical targets of the IFN-b signalling pathway required for the transcriptional activation of IFN-b-dependent genes (IRF7 and STAT-1), were detected in THP-1 cells infected with MVA and associated with the induction of IFN-b [40]. Whether C6 blocks directly or indirectly IRF3, IRF7 or STAT-1 expression or activation by phosphorylation is under investigation.
Since our goal is to develop modified MVA-B with enhanced immunogenicity to HIV-1 antigens, we carried out a detail characterization of the HIV-1-specific memory immune responses induced in mice using DNA prime/MVA boost approach, and compared parental MVA-B with the deletion mutant MVA-B DC6L. We used ICS, which allows a more extensive characterization of T-cell effector functions at the single-cell level [41]. Our findings revealed that at 53 days post-boost, DNA-B/MVA-B DC6L triggered higher magnitude and polyfunctionality of total HIV-1-specific CD4 + and CD8 + T-cell memory immune responses (specific for Env, Gag and GPN) than DNA-B/MVA-B. The vaccine-induced T-cell memory responses were predominantly mediated by CD8 + T cells in both immunization groups, with most of the response mediated by CD8 + EM and TEMRA T cells, which have been described to have a powerful and direct antiviral capacity [30,31,32,33] and have been associated with HIV-1 viral control in early and chronic infection [42,43,44]. HIV-1-specific CD4 + T-cell memory responses were preferentially Env-specific in both immunization groups, similar to what it is obtained with other MVA HIV-1 vaccines as MVA-CMDR [45] or other NYVAC HIV-1 vaccines [21,46,47]. Furthermore, immunization with DNA-B/MVA-B DC6L induced preferentially GPN-specific CD8 + T-cell memory responses compared with DNA-B/MVA-B. The shift towards GPN-response triggered by MVA-B DC6L might result from the activation of the intrinsic pathway of antigen presentation by the Gag-Pol-Nef intracellular polyprotein due to the increased IFN-b production promoted by the deletion of C6L. However, it is not clear yet whether the enhanced CD8 + T-cell response to GPN is due to a greater breadth of response or could reflect an enhanced response to single epitopes. The functional relevance of Env-specific CD4 + T-cell responses or GPN-specific CD8 + T-cell responses in setting of vaccination for prevention of HIV-1 infection needs to be further explored. Further experiments need to be done in other animal models such as non-human primates to determine the possible benefits of the response elicited by MVA-B DC6L. Interestingly, enhanced Gag response has been associated with better control of virus in macaques infected with SIV and in HIV-1-infected individuals [48,49].
How significant are our in vivo findings with regard to immune requirements for HIV-1 protection? While definition of correlates of protection to HIV-1 remains to be firmly established, there are a number of markers that can be used as potential indicators for an effective HIV-1 vaccine, such as: 1) activation of HIV-1-specific CD4 + and CD8 + T cells; 2) triggering polyfunctional responses; 3) enhanced magnitude and breath of the immune response; 4) induction of long-term memory cells of effector phenotype; 5) production of neutralizing antibodies with broad specificities. Several features of the T-cell response to HIV-1 are correlated with control of viral replication [50,51], and a correlation of the CD8 + T-cell response with a lowering of peak viremia in acute HIV-1 infection has been described [52,53]. Also, in non-human primates there is a good correlation between vaccine-induced HIV-1-specific cellular immunogenicity and protection after a challenge with a pathogenic SHIV [22,27,28], where CD8 + T cells play an important role in immunity to HIV-1 [54]. Several studies have demonstrated that T-cell polyfunctionality is associated with protective antiviral immunity [ [50,51,55,56,57], for a review [33]]. In HIV-1-infected patients that are nonprogressors, HIV-1specific CD8 + T cells were polyfunctional [50]. Furthermore, the generation of memory CD8 + T cells of EM and TEMRA phenotypes have been associated with the control of HIV-1 infection in patients [42,43,44] and in non-human primates [58]. Moreover, it was described recently that effector memory T-cell responses elicited after vaccination in non-human primates could control highly pathogenic SIVmac239 infection early after mucosal challenge, showing the important role of memory T-cell responses [59]. These observations suggest that polyfunctional CD8 + T cells and effector memory T cells are important components of a protective immune response [33]. Importantly, both MVA-B and MVA-B DC6L triggered immune responses that fulfill several of the characteristics of a promising candidate HIV-1 vaccine. Indeed immunization with DNA-B/MVA-B DC6L induced activation of HIV-1-specific CD4 + and CD8 + T cells, enhanced magnitude and polyfunctionality of the immune response, triggered long-term memory T cells of effector phenotype, and increased the levels of antibodies directed against Env.
However, the immunological parameters required for protection against HIV-1 infection in humans remain unknown. The phase III Thai clinical trial reported a modest protection of about 31% against HIV-1 infection in vaccinees with the combination of recombinant vaccines canarypox and gp120, in spite of poor neutralizing antibodies and of reduced T-cell responses against HIV-1 [60]. The phase III Thai trial has pointed out that further developments of poxvirus vectors is needed. Among MVA vectors, several phase I clinical studies for HIV/AIDS have been performed with DNA prime/MVA boost protocols or with MVA administered alone [61,62,63,64,65] and revealed promising findings. These studies proved the safety and immunogenicity of the MVA vectors and reported an important proportion of responders with multigenic responses that persisted up to one year post-vaccination. In light of our results, one might postulate that MVA-B with deletion of the C6L gene could improve the immunogenicity of the vaccines by enhancing the magnitude, polyfunctional and memory responses T cell responses.
In conclusion, we show that deletion of C6L on MVA-B upregulates IFN-b expression in human macrophages and DCs, and improves MVA-B immunogenicity in vivo, increasing the magnitude, polyfunctionality and memory T-cell responses against HIV-1 and the generation of Env-specific antibodies. Thus, VACV C6 protein interferes with host immune responses by at least in part, blocking some component(s) of the IFN-b signalling pathway. Understanding the mechanism of action of C6 will provide new insights in virus-host cell interactions and viral immunomodulation. Further work should be devoted to explore the relevance of our findings in a non-human primate model, as MVA-B DC6L represents a promising vector for developing HIV-1 vaccines.

Ethics Statement
The animal studies were approved by the Ethical Committee of Animal Experimentation (CEEA-CNB) of Centro Nacional de Biotecnologia (CNB-CSIC, Madrid, Spain) in accordance with national and international guidelines and with the Royal Decree (RD 1201/2005). Permit numbers: 152/07 and 080030.
Studies with peripheral blood mononuclear cells (PBMCs) from healthy blood donors recruited by the Blood Center of Lausanne (Switzerland) were approved by the ethics commission for clinical research from the Faculty of Biology and Medicine of Lausanne. Written informed consent was obtained from donors. All information were kept confidential by the Blood Center.
The poxvirus strains used in this work included: Western Reserve (WR), modified vaccinia virus Ankara (MVA) and the recombinant MVA-B expressing the HIV-1 BX08 gp120 and HIV-1 IIIB Gag-Pol-Nef proteins [18]. Viruses were grown in CEF cells, purified through two 36% (w/v) sucrose cushions, and titrated by plaque immunostaining assay [66]. Cell lines were infected with viruses as previously described [18,40].

Construction of plasmid transfer vector pGem-RG-C6L wm
The plasmid transfer vector pGem-RG-C6L wm was used for the construction of the recombinant virus MVA-B DC6L, with C6L (C6L in Copenhagen strain of VACV is equivalent to MVA 019L in MVA) gene deleted (for simplicity, we used throughout the work the ORF nomenclature of Copenhagen strain to refer the MVA genes). pGem-RG-C6L wm was obtained by sequential cloning of five DNA fragments containing dsRed2 and rsGFP genes and C6L recombination flanking sequences into the plasmid pGem-7Zf(2) (Promega). The construction of the plasmid pGem-Red-GFP wm (4540 bp), containing dsRed2 and rsGFP genes under the control of the synthetic early/late (E/L) promoter was previously described [17]. MVA-B genome was used as the template to amplify the right flank of

Construction of MVA-B DC6L deletion mutant
MVA-B DC6L deletion mutant was constructed by screening for transient Red2/GFP co-expression using dsRed2 and rsGFP genes as the transiently selectable markers, as previously described [17]. Briefly, 3610 6 DF-1 cells were infected with MVA-B at a multiplicity of 0.05 PFU/cell and then transfected 1 h later with 6 mg of DNA from plasmid pGem-RG-C6L wm using Lipofectamine (Invitrogen) according to the manufacturer's recommendations. After 72 hours, the cells were harvested, lysed by freezethaw cycling and sonicated. Following 6 consecutive rounds of plaque purification in DF-1 cells, MVA-B DC6L was obtained and the deletion of C6L gene was confirmed by PCR amplifying the C6L locus. MVA-B DC6L was grown in CEF cells, purified by centrifugation through two 36% (w/v) sucrose cushions in 10 mM Tris-HCl pH 9, and titrated in DF-1 cells by plaque immunostaining assay, using rabbit polyclonal antibody against VACV strain WR (Centro Nacional de Biotecnología; 1:1000) followed by anti-rabbit-HRP (Sigma; 1:1000), as previously described [66].
MVA-B DC6L deletion mutant was free of contamination with mycoplasma or bacteria.

PCR analysis of MVA-B DC6L deletion mutant
To test the purity of MVA-B DC6L deletion mutant, viral DNA was extracted from DF-1 cells mock-infected or infected at 2 PFU/cell with MVA, MVA-B or MVA-B DC6L. Primers RFC6L-AatII-F and LFC6L-BamHI-R (described above) spanning C6L flanking regions were used for PCR analysis of C6L locus. The amplification protocol was previously described [17]. PCR products were resolved in 1% agarose gel and visualized by ethidium bromide staining. The C6L deletion was also confirmed by DNA sequence analysis.

Analysis of virus growth
To determine virus-growth profiles, monolayers of DF-1 cells grown in 12-well tissue culture plates were infected in duplicate at 0.01 PFU/cell with MVA-B or MVA-B DC6L. Following virus adsorption for 60 min at 37uC, the inoculum was removed. The infected cells were washed once with DMEM without serum and incubated with fresh DMEM containing 2% FCS at 37uC in a 5% CO 2 atmosphere. At different times post-infection (0, 24, 48 and 72 hours), cells were collected, freeze-thawed three times and briefly sonicated. The intracellular viruses were titrated by immunostaining as described above.
Expression of C6 protein in E. coli and production of anti-C6 polyclonal antibodies The C6 ORF (MVA 019L, 157 aa, 18.2 kDa) was amplified by PCR using primers C6L-NheI-F (59-AGGCTAGCGTTTAGG-AAAAAAAAATATC-39) (NheI site underlined) and C6L-BamHI-R (59-AAGGATCCCATGAATGCGTATAATA-39) (BamHI site underlined), and VACV MVA DNA as template. The product (488 bp) was digested with NheI and BamHI and cloned into plasmid pET-27b(+) (Novagen). The ligation product was used to transform BL21 E.coli, and the plasmid of a kanamycin-resistant positive colony was sequenced to confirm that it contained the C6L sequence. The plasmid generated was termed pET-27b-C6L (5837 bp). Plasmid pET-27b(+) provided a tract of 6 histidines at the carboxyl terminus of the C6 protein, generating a recombinant C6 protein of about 24 kDa. Kanamycin-resistant colonies were grown in Luria broth medium until an OD of 0.5 at 595 nm. Isopropyl 1-thio-b-Dgalactopyranoside was added (0.5 mM) and the culture grown for 4 additional hours. Cells were pelleted by centrifugation. For lysis, cells were suspended in 50 mM Tris-HCl, pH 7.5, 0.3 M NaCl, 8 M Urea, and incubated with 1 mg/ml lysozyme for 30 min in the presence of 1 mM phenylmethylsulfonyl fluoride. The suspension was freezed-thawed twice. Cellular debris was removed by centrifugation, and the supernatant was incubated with Probound resin (Invitrogen). Elution was carried out with different concentrations of imidazol (100 to 500 mM) in 50 mM Tris-HCl, pH 7.5, 0.3 M NaCl. Eluted fractions were pooled and loaded on desalting columns following the manufacturer's protocol (GE-Healthcare, Freiburg, Germany), and fractions were collected. Protein was quantified using the Bradford assay, fractionated by 12% SDS-PAGE and analyzed by Western blot using anti-His tag antibody (1:5000) to detect the presence of VACV C6 protein. Fractions containing the C6 protein (with an estimated purity of 90%) were stored in aliquots at 220uC. The C6 protein (1150 mg) was injected into New Zealand White rabbits to produce anti-C6 serum and rabbit polyclonal antibody against C6 (Biomedal Laboratories, Sevilla, Spain).

RNA analysis by quantitative real-time polymerase chain reaction
Total RNA was isolated from THP-1 cells and moDCs infected with MVA, MVA-B and MVA-B DC6L, using the RNeasy kit (Qiagen, Hombrechtikon, Switzerland). Reverse transcription of 100 ng to 500 ng of RNA was performed using the ImProm II RT System kit (Promega). Quantitative PCR was performed with a 7500 Fast Real-Time PCR System (Applied Biosystems, Rotkreuz, Switzerland) using the Power SYBR Green PCR Master Mix (Applied Biosystems), as previously described [40]. Expression levels of IFN-b, IFIT1, IFIT2, RANTES, MIP-1a, IL-8, IP-10 and HPRT genes were analyzed by real-time PCR using specific oligonucleotides (sequence be provided upon request). Gene specific expression was expressed relative to the expression of HPRT in arbitrary units (A.U.). All samples were tested in duplicates.

IFN-b and type I IFN measurement
IFN-b concentrations in cell-culture supernatants were measured by ELISA (PBL Biomedical Laboratories, Picataway, NJ). Type I IFNs were quantified as previously described using the reporter cell line HL116 which stably expresses the luciferase gene under the control of the IFN-a/b-inducible 6-16 promoter [67,68,69]. HL116 cells were grown in DMEM supplemented with 10% FCS and 50 mM hypoxanthine-aminopterin-thymidine. HL116 cells (10 5 cells in a 96-well plate) were incubated for 6 h with cell culture supernatants or recombinant human IFN-b (Cell Sciences, Canton, MA) as a standard. Luciferase activity in HL116 lysates was quantified using the Luciferase assay reagent (Promega) and a luminometer (EG&E Berthold, Germany).

Mice immunization schedule
BALB/c mice (6-8 weeks old) were purchased from Harlan. A DNA prime/MVA boost immunization protocol was performed as previously described [17,18]. Groups of animals (n = 4) received 100 mg of DNA-B (50 mg of pCMV-BX08 gp120+50 mg of pCDNA-IIIB GPN) by intramuscular route (i.m.) and two weeks later received an intraperitoneal (i.p.) inoculation of 1610 7 PFU of the corresponding recombinant vaccinia viruses (MVA-B or MVA-B DC6L) in 200 ml of PBS. Mice immunized with sham DNA (DNA-w) followed by MVA booster were used as control group. At 53 days after the last immunization, mice were sacrificed and spleens processed for IFN-c ELISPOT and intracellular cytokine staining (ICS) assays. Two independent experiments have been performed.

Antibody measurements by ELISA
Antibodies anti-HIV-1 gp160LAV envelope protein were measured by ELISA as previously described [17,18].

Statistical procedures
The statistical analysis of ELISPOT and ICS assays was realized as previously described [17,71]. We have developed a novel approach that corrects measurements for the medium response (RPMI), calculating confidence intervals and p-values. Only antigen responses values significantly larger than the corresponding RPMI are represented.