Modification of Adenovirus vaccine vector-induced immune responses by expression of a signalling molecule

Adenoviral vectors are being developed as vaccines against infectious agents and tumour-associated antigens, because of their ability to induce cellular immunity. However, the protection afforded in animal models has not easily translated into primates and clinical trials, underlying the need for improving adenoviral vaccines-induced immunogenicity. A Toll-like receptor signalling molecule, TRAM, was assessed for its ability to modify the immune responses induced by an adenovirus-based vaccine. Different adenovirus vectors either expressing TRAM alone or co-expressing TRAM along with a model antigen were constructed. The modification of T-cell and antibody responses induced by TRAM was assessed in vivo in mice and in primates. Co-expression of TRAM and an antigen from adenoviruses increased the transgene-specific CD8+ T cell responses in mice. Similar effects were seen when a TRAM expressing virus was co-administered with the antigen-expressing adenovirus. However, in primate studies, co-administration of a TRAM expressing adenovirus with a vaccine expressing the ME-TRAP malaria antigen had no significant effect on the immune responses. While these results support the idea that modification of innate immune signalling by genetic vectors modifies immunogenicity, they also emphasise the difficulty in generalising results from rodents into primates, where the regulatory pathway may be different to that in mice.

www.nature.com/scientificreports www.nature.com/scientificreports/ We have shown in vitro in mammalian cells that expression of TRAM from plasmids in transient transfection assays activates the cxcl2, lcn and ifnb (interferon b) promoters, which are canonical early and late pro-inflammatory pathway and interferon signalling pathways target genes, respectively 10 . When co-expressed with the TAK1 (MAP3K7) serine/threonine kinase, TRAM expression increases T-cell responses induced by DNA vaccination in vivo in mouse 11 . We hypothesized that the expression of TRAM from an adenovirus vector could alter the virus-dependant signalling cascade in vivo, modifying viral immunogenicity. We investigated immune responses elicited by adenovirus vaccine immunization, with and without TRAM expression, in both murine and non-human primate models.

Materials and Methods
Plasmids, recombinant adenoviruses and recombinant MVA vaccines. The synthetic multiepitope string (TIP, Tuberculosis-Immunodeficiency virus-Plasmodium epitope), containing CD8+ and CD4+ T cell epitopes from Plasmodium berghei, Mycobacterium tuberculosis and SIV, was described previously 11 . To generate recombinant adenoviruses expressing both the TIP antigen and an immune activator, a Gateway-compatible bicistronic entry vector was generated, derived from pENTR4 (Fisher) and the bicistronic expression vector previously described 11 by a series of conventional cloning steps. To permit monitoring of antigen expression, the TIP antigen with a TIP antigen was fused in-frame to the N-terminus of GFP by PCR amplification and restriction cloning. One of the CMV IE94 promoters was replaced with the powerful human elongation factor 1α promoter (phEF1) from pEF-BOS 12 to minimise repetitive sequences in the vector. phEF1 was used to drive antigen expression, while the CMV IE94 promoter was used to drive expression of mTRAM, and other inflammatory activator ORFs (Fig. 1a). In the experiments described here, two almost identical designs of this bicistronic expression cassette were used. One, designated the SE cassette, is described above. The SETO cassette differs from the SE cassette by addition of a Tetracycline-operator containing sequence flanking the CMV promoter TATA box (Fig. 1a) to limit CMV-driven expression in Tetracycline repressor containing cells 13 .
SE and SETO expression cassettes were transferred from their plasmids of origin into the E1 locus of pAd/ PL-DEST (Invitrogen), or the ChAd63 (chimpanzee adenovirus type 63) vector 14 by in vitro recombination using LR clonase (Invitrogen). Control ChAd63 adenoviruses expressing blood-stage malaria antigens Pfm128 and PfAMA-1 or an irrelevant antigen (secreted alkaline phosphatase SEAP) were also used (hAd5-Pfm128, ChAd63-PfAMA-1 and ChAd63-SEAP). A monocistronic chimpanzee adenovirus serotypes serotype 63 (ChAd63) vector in which the complete human TRAM open reading frame was expressed by a CMV promoter was also created (ChAd63-TRAM), as previously described 15 . All adenoviruses were grown in human embryonic kidney 293 (HEK293) cells, purified by CsCl gradient ultracentrifugation, and dialyzed against 10 mM Tris (pH 7.4). Plasmid sequences were confirmed by Sanger sequencing, and are included in Supp. Data 1. In studies of both mouse and human TRAM, the entire unmodified open reading frame was expressed, including the myristoylation signal. The viruses were titered as described previously 13 on HEK293 cells grown under standard conditions. Wells were scored positive or negative 10 days after infection using hexon immunostaining (Cell Biolabs Inc, San Diego, CA, USA), and infectious unit titers per milliliter were calculated with the Spearman-Karber method. A recombinant modified vaccinia virus Ankara (MVA) vector encoding ME-TRAP was also used, and was described previously 16 .

Viral bioactivity and antigen expression.
To assess bioactivity of the viruses generated, HeLa cells were grown in DMEM containing 10% heat inactivated fetal bovine serum in 96 well plates to 80% confluency (~1 × 10 5 cells per well). Three treatments were compared: no stimulus, infection with various recombinant viruses, application of IL-1β. In all cases, media was aspirated and 100μl Optimem medium (Gibco) containing the stimulus was applied for one hour. For viral infection, medium contained virus at 10 and 100 MOI. For cytokine stimulation, media contained 30 pg/ml IL-1β (Calbiochem). GFP production was scored visually using an inverted fluorescence microscope 35 hours post infection. Thirty six hours post infection, supernatants were aspirated and IL-8 quantified in the supernatants (DY208, RnDSystem DuoSet).
To assess transgene expression quantitatively, Raw246.7 or NIH-3t3 cells, grown in DMEM containing 10% heat inactivated fetal bovine serum, were seeded (5 × 10 4 Raw246.7 and 2 × 10 4 NIH-3T3 cells per well) into 96-well flat bottom plates and left to adhere overnight. The following day, cells were infected with AdHu5-TIPEGFP (human adenovirus type 5, expressing fused to EGFP: enhanced green fluorescent protein) and ChAd63-TIPEGFP vectors, with and without TRAM, at increasing MOIs. 24 hours later, culture media was removed from each well and 30 µL of trypsin (TrypLE Express Enzyme, Life Technologies) added and incubated for 10-15 minutes, prior to resuspension of cells in 10% FCS in PBS and acquisition.
Immunizations. Mice were used in accordance with the UK Animals (Scientific Procedures) Act under project license number 30/2414 granted by the UK Home Office. The experimental protocols were approved by the Oxford University institutional Animal Welfare and Ethics Review Committee (AWERB), prior to getting licensing from the UK home Office. For induction of short-term anaesthesia, animals were either injected intraperitoneally (i.p.) with ketamine and Domitor ® or anaesthetised using vaporised IsoFlo ® . Female BALB/cOldHsd (BALB/c) or C57BL/6OldHsd (C57BL/6) mice (Harlan UK Limited) were immunised intradermally (i.d., into both ears) or intramuscularly (i.m., into each of the tibialis muscles of the hind limbs) with 50μl of vaccine in endotoxin-free phosphate-buffered saline (PBS). Groups of a minimum of four mice were used. The number was based around power calculations in which the variable under study (such as IFN-gamma secreting cell numbers) was log normally distributed with standard deviation of 10% of the mean, as based on previous data, and a 80% power to detect a 2.5 fold difference.
Male rhesus macaques (Macaca mulatta) aged between 4 ½ to 8 ½ years were housed at the National Primate Research Centre and cared for according to the regulations and guidelines of the University of Wisconsin Institutional Animal Care and Use Committee. The experimental protocol was reviewed and approved by the Wisconsin Institutional Animal Care and Use Committee (IACUC). Animals were anesthetized for all procedures and were returned after completion of the study into the colony for potential use in other studies or breeding programs. Male rhesus macaques were initially screened for neutralising antibodies against ChAd63 and background ChAd63 adenoviruses were constructed by incorporating expression cassettes (illustrated) into the deleted E1 region. Two cassettes were studied, SE and SETO, the latter featuring repression of CMV promoter function during viral growth. All viruses expressed a model antigen (TIPEGFP) driven by the human EF1α promoter. (b) TRAM expression induces IL-8 response in vitro HeLa cells were either exposed to adenoviruses at an MOI of 10 or 100, or stimulated with IL1β, or exposed to vehicle (Optimem medium) for one hour. 36 hours after infection, antigen expression was assessed by microscopy and IL-8 secretion measured by EIA. (c-f) Coexpression of mouse TRAM does not increase the antigen transgene expression in mouse cell lines. Raw246.7 (c,d) or NIH-3t3 (e,f) cells were infected with the AdHu5 (a,e) or ChAd63 (d,f) at an MOI of 100, 10 or 1. Cells were harvested 24 hours later and the level of GFP expression measured by flow cytometry. Graphs represent the percentage of infected cells (GFP+) or the geometric mean fluorescence intensity (gMFI) of the GFP signal, each dot represents a replicate well with lines representing the median response.
Ex vivo IFNγ ELISPOT. Mouse blood or mashed and filtered spleens were used for enumeration of antigen-specific IFNγ-producing T cells as described 11 . Nonhuman primate specific IFN-γ ELISPOT assays were performed as previously described 17 using precoated ELISpot PLUS kits according to the manufacturer's recommendation (Mabtech USA, Mariemont, OH, USA). All tests were performed in duplicate using pools of 20-mers at a final concentration of 5 μg/ml. Background values (mean of spots without peptide) were subtracted. A response was considered positive if the mean number of SFCs (spot forming cells) of duplicate sample wells exceeded background plus two standard deviations (STD) and was >50 SFC per 1 × 10 6 cells.
Antibody responses to TRAP were measured using a luciferase immunoprecipitation system (LIPS) as previously described 18 . Briefly, serum samples were incubated for 1 hour with a cell lysate from 293 cells transfected with a TRAP-rLuc expression plasmid, prior to incubation with Protein A/G UltraLink Resin beads (ThermoScientific) in MultiScreen HTS membrane Barex plates (Millipore) for 1 hour. Unbound lysate and antibodies were removed by washing the plates prior to quantification of bound rLuc activity using Renilla luciferase assay system (Promega) and a Varioskan Flash luminometer (Thermo). Antibody levels are expressed as a fold increase from the pre-vaccination timepoint.
Statistical analysis. GraphPad Prism version 5.0 was used to perform two-way ANOVA testing when multiple doses or routes were tested.

TRAM expression from a bi-cistronic adenovirus vector increases IL-8 production in vitro.
Bi-cistronic AdHu5 vectors and ChAd63 vectors encoding a model transgene of interest (containing malaria and TB epitopes, fused to EGFP), driven by a human EF1α (hEF1α) promoter, were constructed including cassettes allowing inflammatory signalling components to be expressed under the control of a CMV promoter (Fig. 1a), while the antigen was driven by an hEF1α promoter, a strong mammalian promoter 12 .
Infection of HeLa cells by these vectors showed that TRAM-expressing vectors consistently caused the release of IL-8, a chemokine chosen for bioactivity monitoring (Fig. 1b). Significant increased in IL-8 concentrations were not observed with control viruses that did not express TRAM relative to mock infection (Fig. 1b). The increase with TRAM expressing viruses was much larger than that observed with vectors expressing another inflammatory activator, TAK1 11 . That the pro-inflammatory response was not observed in cells infected with control viruses, in which the only difference is the absence of the TRAM gene, or with a vector containing a different transgene (TAK1), provides evidence of the transgene-specific modification of the signalling response in the cells. A GFP was used in this system to confirm that the insertion of the TRAM gene in the adenovirus vector did not modify the amount of antigen expressed, an important control to confirm that the pro-inflammatory effect was due to the transgene itself rather than to an indirect effect. Antigen expression, assessed using GFP fluorescence, was similar in three infected cell lines: HeLa (Fig. 1b), NIH-3T3 and RAW264.7 (Fig. 2) irrespective of the presence of TRAM in the virus. We have however not tagged the TRAM molecule in the transgene design out of concern that any epitope tag might alter the intra-cellular activity of TRAM, thus preventing the possibility to measure directly the increase of TRAM molecules within the cells with this method. www.nature.com/scientificreports www.nature.com/scientificreports/

Co-expression of TRAM with antigen in AdHu5 bicistronic vector increases CD8+ and decreases antibody responses to the transgene in mice.
To assess whether the co-expression of TRAM was associated with a modification of AdHu5-TIP EGFP-induced immunogenicity in vivo, groups of BALB/c mice were immunized with the mTRAM encoding virus and compared the immune responses with the non-adjuvanted counterpart under different conditions. Insertion of the murine Tram gene increased the CD8+ T cell responses to the Pb9 epitope (Fig. 2a, two way ANOVA, p < 0.0001 for the TRAM effect), while it did only marginally increased CD4+ T cell responses (Fig. 2b, p = 0.0272 for the TRAM effect, observed at one dose only). This effect was not restricted to the intramuscular route: The increase of CD8+ T cell response was detected both i.m. and i.d. (Fig. 2c, two way ANOVA, p = 0.008 for the TRAM effect), and its magnitude was not route-dependant (p = 0.192 for the route effect). The CD4+ T cells responses did not differ significantly by TRAM or by the route of AdHu5 injection at that dose (Fig. 2d. p = 0.223 and p = 0.45 respectively). TRAM expression also markedly reduced the antibody response to the transgene, as measured by the reduction of GFP-specific antibody titres induced by the TRAM-expressing virus with both immunization routes (Fig. 2e, two-way ANOVA, p < 0.0001 for the TRAM effect, p = 0.89 for the route effect).  indicated (a,b), or at 4 × 10 8 iu/mouse (c-e). Specific CD8 + T cell responses to Pb9 peptide (a,c) or CD4+ responses to P15 peptide (b,d) were measured by IFN-γ ELIspot at week 2. Antibody titres against GFP protein were measured at the same time point (e) and the long dotted lines represent the detection limit of the assay. Each dot represents one mouse and the short lines represent the geometric means per group. P values are indicated (2-way ANOVA).

Adjuvant effect of human TRAM-encoding adenovirus in ChAd63 vector in mouse.
To confirm these findings, a monocistronic chimpanzee adenovirus serotypes serotype 63 vector (ChAd63) was generated, in which the complete human TRAM open reading frame was expressed by a CMV promoter (ChAd63-TRAM). No antigen was encoded in this 'adjuvant' virus. It was mixed with a ChAd63 vector encoding the liver-stage malaria antigen ME-TRAP (ChAd63-ME TRAP), and antigen chosen because, when delivered by ChAd63 as part of prime boost regimen, ME-TRAM induces modest protection against malaria in field trials 20 . It also includes the pb9 epitope included in the TIPEGFP constructs.  -injection (c,d). Groups of mice were immunized i.m. at day 0 with ChAd63-ME TRAP alone or was mixed with ChAd63-hTRAM, expressing human TRAM, at a 1:1 or 1:25 ratio. (c,d) A control group included ChAd63-ME TRAP mixed with similar ratios of an adenovirus encoding an irrelevant antigen (secreted alkaline phosphatase, ChAd63-SEAP). Specific CD8 + T cell responses (c) and CD4 + responses (d) were measured by IFN-γ ELIspot at week 2. Each dot represents one mouse and the short lines represent the geometric means per group. Only the group that received 25x TRAM had a response significantly superior to the group that received ChAd63-ME TRAP with no adjuvant (nil).

Adjuvant effect of human TRAM-encoding adenovirus in Rhesus macaques. The adjuvant effect
of human TRAM-expressing vector was investigated in a non-human primate using the mixture approach described above. To this end, three groups of six Rhesus macaques were immunized with ChAd63-ME TRAP alone, or mixed with 5x ChAd63-hTRAM, or with 5x ChAd63-PfAMA1, encoding a blood-stage malaria antigen used as a control.
After a first injection, the TRAP-specific IFNγ ELISPOT responses peaked at week 2 or 4 and ranged from 0 to 1825 SFC/million PBMCs, with 4/6 animals responding in the control and the hTRAM-adjuvanted groups, versus 5/6 in the PfAMA1 group (Fig. 4a-c). No significant difference of the TRAP-specific response was observed, and in all animals, responses had declined to below detection level at week 8 (Fig. 4a-c). A second ChAd63 injection failed to boost the primary responses, as the number of IFNγ-producing cells decreased to below 400 SFU/million PBMC in responding animals (2/6 in the control and hTRAM groups, 1/6 in the PfAMA1 group, Fig. 4a-c). All animals were boosted at week 16 with MVA-ME TRAP. Strong TRAP-specific IFNγ responses were elicited in all animals one week post MVA injection, ranging from 1675 to 21890 SFC/million PBMCs. No significant difference was observed between the groups (Fig. 4a-c). www.nature.com/scientificreports www.nature.com/scientificreports/ To determine the phenotype of the TRAP specific T cell response, PBMCs were restimulated overnight with TRAP peptides and IFN-γ production measured by intracellular cytokine staining. No significant difference between groups was observed for the low IFN-γ + CD4 + or IFN-γ + CD8 + T cells responses measured two weeks after adenovirus administration (data not shown). After the MVA boost, slightly higher frequencies of TRAP specific CD4 + T cells (Fig. 4d) were observed compared to CD8 + T cells (Fig. 4e), however there was again no significant difference in the frequency of IFN-γ + CD4 + or CD8 + T cells between the groups. In addition, there was no significant difference in either the frequency or proportion of T effector, T effector memory or T central memory cells between the three groups (data not shown). TRAP antibodies were measured pre-vaccination, at week 4 post first adenovirus injection and 3 weeks post-MVA boost. While increases in TRAP specific responses were observed after vaccination, mainly after MVA boost (indicated by a greater than 1 fold increase Fig. 4f,g), no significant difference between the three groups of macaques at these two time points was observed. In conclusion, strong TRAP specific immune responses were observed in all vaccinated animals with the peak response observed one week post-MVA boost. Vaccination induced TRAP specific IFN-γ + CD4 + and IFN-γ + CD8 + T cells and TRAP specific antibodies, yet neither enhancement of the CD8+ response nor decrease of the antibody response by co-vaccination with ChAd63 expressing TRAM was observed.

Discussion
In this study, the adjuvant effect of a signalling pathway molecule on a co-administered antigen previously observed with DNA immunization was investigated in the context of viral vector immunization. A modest (1.5 to 2 fold) increase of CD8 + T cell responses was observed in mice immunized with TRAM expressing adenovirus vaccines, concomitant with a marked decrease of antibody responses. In mice, the effect was observed in experiments where the murine tram gene was inserted in the same vector as the antigen, or when the antigen and TRAM (murine or human) were encoded by different vectors mixed together prior to injection. The effect was observed with the human serotype 5 as well as with the chimpanzee serotype 63 adenoviruses described above. In rhesus macaques however, there was no evidence of any adjuvant effect induced by a vector encoding human TRAM on the immune response elicited by a co-administered adenovirus, under the conditions studied. Difficulties translating murine adjuvants into primates was observed previously with 4-1BBL 21 .
The adjuvant effect was restricted to increase of CD8+ responses, while a decrease of antibody responses was observed in mouse. Such effect was previously observed with a Venezuelan equine encephalitis virus adenovirus vaccine: when IFN-alpha was co-expressed with the antigen, a reduction of transgene-specific antibody responses was observed 22 . Thus, in vivo interferon induction induced by TRAM 7-9 may explain some of the effects observed.
The immune modulation induced by TRAM seen in mice was not observed in primates. The regime tested in primates used an adjuvant virus administered in a molar excess relative to the antigen expressing virus. The ratio of adjuvant to antigen expressing virus was lower than that at which significantly enhanced immunogenicity was observed in mice, but was chosen because it was considered to be a more realistic candidate for manufacturing and human use, compared with regimes with higher ratios of adjuvant to antigen (such as that used in Fig. 3a). It is possible that the mixture used was not optimal in primates.
Induction of immune responses is dependent on complex pathways, and therefore immune modulation by modification of intracellular signalling pathway may differ between species. In particular, a splice variant of TRAM, TAG, acts as a negative regulator of the TLR4 signalling pathway in primates 23 . This molecule competes with TRAM for the recruitment of TRIF, and is not found in mice, and such regulation could explain the difference of TRAM-adjuvant effect between mice and non-human primates. Therefore, it is unlikely that any modification aiming at increasing the expression of TRAM from an adenovirus vector would result in a suitable adjuvant system in human, because of the negative regulation system. However, a combination with another stimulation may overcome the negative regulation: TRAM is a membrane-anchored molecule, containing a TIR-domain adaptor that induces the MyD88-dependent signaling pathway, and ultimately activates NF-kB. We hypothesized that over-expression of TRAM could mediate intracellular signal activation, that would result in enhanced immune responses. The intracellular signal activation may be enhanced in the presence of TLR stimulation. Therefore, combining a TRAM expressing vector with TLR stimulating small molecule may enhance the adjuvant properties and vaccination efficacy. This is an area for further research.
Altogether, these results support the concept that modification of viral vector immunogenicity by modifying virally induced innate immune signalling profiles, but emphasise the difficulty of doing so in a cross-species manner. Adjuvanting viral vector induced responses in primates can be achieved by altering antigenic structure: oligomerization of several antigens with molecules such as C4bp variants 19,24 . In addition, fusion of antigens to MHC class II invariant chain also induced increased T-cell responses in mice and in non-human primates 25 . Such structural modifications of antigens may have more consistent effects between species than modulation of signalling pathways.