Poxviral promoters for improving the immunogenicity of MVA delivered vaccines

ABSTRACT

Modified vaccinia virus Ankara (MVA) is a replication-deficient poxvirus, attenuated in chick embryo fibroblast primary cells. It has been utilised as a viral vector to develop many vaccines against cancer and infectious diseases such as malaria, HIV/AIDS, influenza, and tuberculosis, MERS-CoV, and Ebola virus infection. There is accumulating data from many preclinical and clinical studies that highlights the excellent safety and immunogenicity of MVA. However, due to the complex nature of many pathogens and their pathogenicity, MVA vectored vaccine candidates need to be optimised to improve their immunogenicity. One of the main approaches to improve MVA immunogenicity focuses on optimising poxviral promoters that drive recombinant vaccine antigens, encoded within recombinant MVA vector genome. A number of promoters were described or optimised to improve the development of MVA based vaccines such as p7.5, pF11, and mH5 promoters. This review focuses on poxviral promoters, their optimisation, genetic stability, and clinical use.

KEYWORDS:

• poxviral vector
• promoter
• vaccine vector
• immunogenicity
• MVA

Introduction

Modified vaccinia virus Ankara (MVA) is a poxvirus that was attenuated during the 1960s via passaging a Vaccinia Virus (VACV) strain more than 570 times in chick embryo fibroblast primary cells (CEF).^1,2 Therefore, MVA has been used as a viral vector to develop many vaccines against cancer and infectious diseases such as malaria, HIV/AIDS, influenza, and tuberculosis^3-6 and recently MERS-CoV⁷ and Ebola virus infection.⁸ MVA is a replication-deficient virus with an excellent safety and immunogenicity profiles and it has been evaluated in multiple clinical trials.⁵ This vector has been used mainly as a boosting agent in heterologous prime-boost vaccination regimens, to expand primed T cells specific to recombinant antigens; the priming vaccination in these regimens is often achieved with DNA or adenoviral vectors.^8–10 Apart from vaccines against some malignancies,¹¹ recombinant MVA vectors (rMVA) have replaced recombinant vaccinia virus vectors (rVACV) in preclinical and clinical testing.⁵ Improving of MVA immunogenicity is an on-going research through various approaches to allow dose-sparing with concomitant cost reduction, or to potentially allow for a single dose rather than prime-boost vaccination.¹² One approach focuses on utilising stronger promoters to enhance recombinant antigen (rAg) expression, which would be associated with improved rAg-specific immune responses.^13–16 These must be poxviral promoters because poxviruses have evolved as cytoplasmic viruses, using virally-encoded transcription machinery¹⁷ that recognises their unique promoter sequences.¹⁸ When poxviral promoters were used to drive the expression of cellular genes, no transcription was detected, supporting their specificity to poxviral transcriptional machinery.¹⁹ Poxviral genes, promoters, and transcription factors are divided into early, intermediate, and late classes, depending on their expression timing during poxvirus infections.^20–23 MVA replication in most mammalian cells (non-permissive cells) ceases during the assembly of progeny virions after all stages of expression occur. This supports the utility of all promoter classes, including late promoters, for controlling transgene expression.^24,25 Some poxviral promoters have both early and late elements, allowing their open-reading frames (ORFs) or recombinant antigens to be expressed early in the virus infection and late after the viral genome replication,¹⁸ respectively.

Poxviral promoters can be utilised cross-strain; for example the vaccinia virus p7.5 promoter was used in fowlpox virus (FPV) to drive transgene expression. The transcription in FPV was initiated, at an early and late element, and terminated at a termination sequence similar to that of p7.5 in VACV, driving the same transgene.^26,27 This showed a genetic conservation and established the use of exogenous poxviral promoters in recombinant poxviral vectors. Similarly, the 4b late promoter from FPV (FP4b) shares genetic similarity to its VACV counterpart, utilises similar late element to that of VACV, and is often used cross-strain in recombinant VACV and MVA vectors.²⁸ For instance, the FP4b was used in MVA vectors derived from the MVA-BAC constructs.¹⁶ Another example is the late ATI promoter (driving the A25L ORF), found in both VACV and cowpox virus.²⁹ The cowpox ATI promoter was utilised successfully in recombinant FPV³⁰ and MVA.³¹ However, if the level of expression is very critical, a comparison should be performed to assess the activity of cross-stain exogenous promoters. A comparative study on the late VACV p11 promoter or the late FVP PFL1 promoter, driving the same transgene in either FPV or VACV revealed that transgene expression occurred in all recombinants, but with variable level of protein expression.^26,32 For developing poxviral vectored vaccines, a number of reports have shown a correlation between early expression of recombinant antigens in rMVA, and their induced cellular immunogenicity.^{13,16,33–35} In this report, I review and present the improvement of MVA vector immunogenicity that is achieved by searching for, or optimising, stronger poxviral promoters. It focuses on the frequently used promoters in MVA as a vaccine vector, but other poxviral promoters could be described elsewhere.

Common promoters in MVA vectors

A number of poxviral promoters have been tested and used in rVACV and rMVA, aimed at increasing rAg expression and potentially inducing more immune responses. For example, the P11 promoter³⁶ was used in rVACV, which is a strong late promoter that naturally drives the 11K protein³⁷ encoded by the F17R ORF.^36,38 Another described promoter was the I1L promoter, a strong late promoter³⁹ that could rival other poxviral intermediate and late promoters, such as the promoter of A1L, J3R, and E4L ORFs.⁴⁰ The I1L promoter is short (29 nucleotides) and can be stronger than the bacteriophage T7 promoter, in the hybrid T7/VACV expression system, the T7 promoter is often a prominent very strong promoter. However, the I1L promoter has to include a G residue at the 3ʹ of the ATG initiation codon, which can be a disadvantage.⁴⁰ The early and late p7.5 promoter is commonly used in MVA vectored vaccines for its ability to strongly express antigens during both early and late stages of VACV infection. It was originally described as two promoters, because it has two different transcriptional start sites, 55nt apart.⁴¹ The native version of p7.5 promoter is found in two copies driving the expression of C23L and B29R ORFs in the ITR (Inverted Terminal Repeats) of poxvirus genomes. Other early and late promoters have also been characterised such as the promoter of the TK gene,⁴² the F7L promoter,⁴³ or the H5R promoter.^44,45 However, the p7.5 promoter is often used as a conventional comparator with any newly optimised or described promoter.^{13,16,33,46,47}

New and optimised promoters for MVA vectors

The search for new promoters has continued despite the availability of strong promoters such as the p7.5, which has been used in MVA-based vaccines that have been tested in clinical trials with excellent immunogenicity outcomes. First, some studies reported the insertion of a transgene in a replacement of non-essential ORFs using the adjacent upstream endogenous promoters to control the transgene expression.^48,49 Using this method, a rVACV with three antigens from three different pathogens were constructed, and was able to induce reactive antibodies in rabbits. However, this method could result in unpredicted protein translation. Recently, the utility of endogenous promoters of non-essential ORFs with a more precise method was reported. In this method, a transgene is expressed exactly from the authentic start codon of the non-essential, replaced, ORFs.^16,50 Second, another development was achieved by searching for tandem promoters. A naturally tandem promoter (with two early elements) was also studied with Ovalbumin (OVA), and showed increased OVA expression but none of the conventional early promoters (p7.5 or mH5) was used in the comparison.⁵¹ The OVA-specific CD8⁺ T cell responses were also enhanced in C57BL/6 mice, although this improvement was observed only after three homologous rMVA immunisations,⁵¹ making this unlikely to be of use in clinical vaccine development.

Third, other studies have, instead, focused on genetically optimising existing promoters. The active core sequence of the p7.5 was dissected in a very comprehensive study, by Davison and Moss,⁵² in which they generated 99 rVACV mutants by substituting the 33-nucleotides in the early element of p7.5 promoter. Each mutant with a different mutated promoter was tested for the expression of Beta-galactosidase (β-gal), as a recombinant antigen. This study showed some absolutely crucial residues that prevented the β-gal gene expression when they are substituted. This study was then used to establish the consensus core sequence of VACV early promoters. The same virologists dissected the late promoter of the L4R ORF (Named the 28kDa promoter in early studies) in the same strategy and described the consensus core sequence of VACV late promoters.⁵³ As some VACV mutants increased the β-gal expression, this strategy was used later to optimise different early or late promoters. Therefore, by combining the optimal nucleotide at each residues of both early and late core sequences, a short synthetic promoter (SSP, Sometimes referred to by different acronyms like pSyn) was then produced, by two research groups.^14,15 The approach of mutating promoter core sequences was applied to manipulate the expression timing of other promoters; for example, the early/intermediate I3L promoter was mutated to shift its activity after the DNA replication (i.e. becoming early/late).⁵⁴ Another application of this approach is the eminent example in this field where the early/late H5R promoter (H5) was optimised by two nucleotide substitutions in its early and late core sequences,¹³ which was based on the early and late consensus sequences, resulting in the strong modified (mH5) promoter. Although the mH5 was not compared to the original H5 promoter, its activity was superior to that of p7.5 and SSP, in driving β-gal expression.

Fourth, promoters were optimised through combining of core sequences. Five copies of the p7.5 early element and one copy of its late element were combined, resulting in a hybrid tandem promoter called pHyb.²⁴ This promoter enhanced the early expression of OVA as compared to the p7.5 promoter. OVA-specific CD8⁺ T cell responses in C57BL/6 mice were only improved after four homologous rMVA immunisations using the pHyb as compared to the p7.5;²⁴ and that makes it difficult to be applicable for human vaccine development. Different combinations were made to produce novel tandem promoters, for instance, by combining four late elements from four different late promoters with the late element of the SSP or generating five copies of the p7.5 early element, but none has rivalled the pHyb, in terms of cellular immunogenicity.⁵¹

Fifth, swapping the core elements of promoters was utilised to optimise poxviral promoters. For example, the early and late elements of the SSP promoter were swapped with an adjustment for the spacer region between the early element and the transcriptional start site; this promoter was named LEO promoter. The LEO promoter enhanced GFP (green fluorescent protein) expression and the magnitudes of GFP-specific CD8⁺ T-cells in mice.³⁴ Further improvement of the LEO promoter was achieved, where the spacer sequence between the promoter elements and the transgene transcriptional start site was elongated from 38 to 160 nucleotides, which improved antigen-specific memory CD8⁺ and CD4⁺ T cell responses, as tested with GFP or a Leishmania antigen (a.k.a LACK).⁵⁵ However, in another study, the insertion of an IRES sequence of 550 nucleotides between the early B8 promoter (named pB8) and a reporter antigen, in which the IRES indirectly served as a spacer, did not improve the transgene expression nor the immunogenicity of rMVA.⁵⁶ This suggests that the spacer should have an optimal length.

Sixth, as mentioned above, MVA endogenous early promoters, which naturally control the expression of non-essential or fragmented ORFs, were used to drive a transgene that is inserted exactly at the ATG start codons of the ORFs. For example, utilising the endogenous pB8 and pF11 promoters, which control the expression of B8R and F11L ORFs, respectively, enhanced transgenic luciferase expression (in vitro) with improved murine malaria epitope-specific CD8⁺ T cell responses (in vivo) that both were higher than that of the p7.5¹⁶ or the mH5 promoters.⁵⁰ Utilising endogenous promoters is linked to using their respective ORFs as insertion sites, which is important for designing multivalent MVA-based vaccines. Having said that, the pB8 and pF11 promoters were used as ectopic promoters, at the Thymidine Kinase (TK) insertion site, out of their natural context, and continued to improve cellular immunogenicity as compared to the mH5 promoter.⁵⁰ The pF11 was then utilised to control the influenza MVA-NP+ M1 and the malaria MVA-METRAP vaccine candidates, both vaccines are being tested in clinical trials, and both are now controlled by the conventional p7.5 promoter. The pF11 enhanced significant levels of vaccine-specific CD8⁺ T cell magnitudes for the two candidates.⁵⁰ Therefore, replacing p7.5 with the pF11 promoter would enhance the potential of these vaccine candidates.

Immunogenicity of MVA with different promoter

MVA is a replication-deficient virus that is able to infect the same tissues as the replication-efficient VACV WR in vivo (mouse models), and expresses recombinant antigens to a higher level than VACV WR, up to 6 hours post-infection.⁵⁷ However, in spite of its strong immunogenicity, MVA gene expression declines to a very low level just after 24 hours post-infection, with no expression by 48 hours post infection, likely because the virus does not replicate in vivo.^25,57 Early MVA transgene expression, which occurs immediately after the uncoating of virions, was reported to enhance cellular immune responses to recombinant antigens as discussed above (also see Table 1). The level of early transgene expression in vitro was shown to correlate with the level of cellular immunogenicity in vivo,^33,34 especially when the early/late p7.5 and mH5 promoters were utilised. However, for early promoters, such as the endogenous pF11 promoter, this correlation was only observed when both expression and immunogenicity were determined in vivo, in mouse models.⁵⁰

Table 1. Comparison of the p7.5, mH5, and SSP promoter activities In vitro and in vivo by different studies.

				rAg expression(1)
Study	Technique	Transgene	Promoter	(AraC+) Early expression	(AraC-) Total expression	% Early activity (2)	Fold in early activity (3) (mH5 vs. others)	Cellular immunogenicity (in BALB/c) (4)
^58	Semi-quantitative W.B.	pp65 (CMV)	p7.5	Not done
			mH5	0.90	2.25	40%		NA
			SSP	0.13	2.23	6%	7-fold
^13	β-Gal activity	LacZ	p7.5	0.21	0.535	39%	5-fold	NA
			mH5	1.1	2.6	42%	NA
			SSP	0.32	8.24	4%	3-fold
^33	Immuno-fluorescence	HIVA-Pk tag	p7.5	Not done	~ 28		NA	mH5 enhanced statistically significant responses.
			mH5	Not done	~ 258		NA
			SSP	Not done	~ 270		NA
^50	Luciferase activity	pb9-rLuc	p7.5	22827225	73882325	30%	3.5-fold.	Significant 1.7-fold
			mH5	78702975	183750000	42%	NA	NA
			SSP	39475725	1160000000	03%	2-fold	Not significant 1.3-fold

Different research groups assessed poxviral promoter activity, using different transgenes (inserts) and techniques. (1) Early (AraC+ treated samples) and overall (AraC-, untreated samples) expression of different recombinant antigen (rAg) is presented. (2) Percentage of promoter early activity is presented by dividing values of AraC+ over AraC-. (3) A more reliable and frequently used calculation is the fold difference in promoter early activity. This shows the fold of early activity differences between two viruses, without normalising to the total amount of expression (used by⁵⁸ and¹³). (4) Cellular immunogenicity studies compared p7.5, SSP, and mH5 promoters.

The p7.5, mH5, and SSP are very frequently used in rMVA vectors. Several studies have reported the superiority of mH5 over the other two promoters in driving early expression of different antigens that associated with enhanced cellular immunogenicity (Table 1). One extensive study tested these three different promoters in rMVA, controlling the expression of an HIV-1 antigen.³³ Although this study did not report the early expression (i.e. by treating cells with a DNA replication blocking agent such as Cytosine Arabinoside (AraC)), the total expression of the SSP in vitro was higher than the mH5, but that did not correlate with in vivo cellular immunogenicity. The SSP had the strongest in vitro activity, but the mH5 promoter induced the highest magnitude of transgenic epitope-specific CD8⁺ T cells in vivo. The in vitro strong expression of the SSP promoter with the relatively low in vivo immunogenicity was interpreted as high expression toxicity on cells,³³ but it might only be due to its weak early expression as compared to the strong early activity of mH5.^13,58

In addition to reports in Table 1, which support the use of early promoters to improve cellular immunogenicity in rMVA, an early study by Coupar et al (1986) showed the importance of using an early promoter in VACV-based vaccine for priming and re-stimulating T cells.⁵⁹ Bronte et al also showed that early transgene expression in professional antigen presenting cells (dendritic cells) is critical for priming cytotoxic T cells.⁶⁰ Interestingly, when MVA was used to infect human monocyte derived dendritic cells, MVA continues to express its early genes with no late protein expression. The majority of CD8⁺ T cell-specific epitopes in VACV are products of early genes, controlled by early promoters, in both humans⁶¹ and mice.⁶² This supports the importance of early expression in poxviral vectors. In addition, the early transcription was also reported to occur very late in the infection at the viral assembly stage.⁴⁷ Therefore, promoters with strong early activity, in vivo rather than in tissue culture, should be chosen to improve antigen-specific cell-mediated immune responses in rMVA based vaccines.

Regarding antibody-mediated immunity, there was no clear study focusing on promoter influence,⁶³ Coupar et al did not find a difference in the induced antibody titres after using different promoters to drive Influenza haemagglutinin (HA). Observations made on vaccines developed by the Jenner institute, Oxford University, suggest that the promoter type (i.e. early vs. late) does not impact on the induced antibody titres (communication with scientists from the Jenner Institute). This is despite the fact that in VACV or MVA infections, antibodies are induced mainly against late poxviral antigens.⁶⁴ However, it could be because the antibody-mediated immunity is directed against pre-made MVA antigens, which are available upon vaccination, rather than the de novo synthesised proteins in vivo. In support of this, a study that blocked the late protein expression of MVA in vivo (in mouse and non-human primate models) showed that induced antibody responses against recombinant antigens were not affected.⁶⁵

Genetic stability of different MVA promoters

Identifying a new promoter or insertion site within any viral vector should be accompanied with demonstrating the genetic stability of transgenes. Transgenes with endogenous pB8 or pF11 promoters, which are readily linked to new insertion loci, have proved genetically stable.¹⁶ Genetic stability data were absent in other reports such as LEO³⁵ or pHyb⁴⁶ promoters, although one could argue it might not be relevant if the transgene was inserted in a well-defined insertion locus or if a conventional promoter is being optimised, such as optimising the SSP into LEO promoter. The conventional use of the p7.5 promoter involves the insertion of a third copy driving transgenes at a different insertion site, in addition to the two native copies located at each terminus of poxviral genome. The homologous recombination between these three copies of the p7.5 promoter has never been a concern or addressed. A number of MVA based vaccines, utilising the p7.5 promoter, have been tested in Phase I, II, and IIb clinical trials, with no concern on genetic stability of the transgene.⁵ Moreover, recombination between any two copies of the p7.5 would result in a very small genome that is unable to produce infectious virions and would be selected against.⁶⁶ In contrast, two transgenes from Cytomegalovirus (p65 and IE1 antigens) controlled by two SSP promoters were unstable after ten passages of rMVA, as compared to utilising two mH5 promoters.⁵⁸ This instability was not a result of the encoded antigens per se, but was rather associated with the SSP promoter. However, these rMVAs were constructed differently, which might have resulted in less stable rAg with the SSP than with the mH5 promoter.⁵⁸ Both antigens with SSP promoters were inserted into the DelII (deletion II region in the MVA genome), in a head-to-head orientation, whereas the same antigens with mH5 promoters were inserted separately in DelII and DelIII regions, partially because the antigens with two mH5 promoters at the DelII failed to produce recombinant viruses. Besides, being located very closely, the SSP promoters were inserted into the DelII insertion site, which was reported to be less stable insertion site for foreign genes.⁶⁷ Therefore, the conclusion on the instability of transgenes with the SSP is still to be determined in a much finer comparison. However, the strong late SSP promoter has been associated with the disadvantage of expressing toxic rAg, in vivo.^33,68 Therefore, promoters with optimal strength such as the p7.5 or mH5 (not as strong as SSP) were preferred for the development of certain vaccines,^33,68 while other researchers re-engineered rAg to be less toxic by various approaches.⁶⁷

MVA promoters in clinical trials

Conventionally, MVA vectored vaccines were tested in clinical trials utilising the p7.5 or mH5 promoters (cited in this review⁵), and no clinical trial has compared the effect of different promoter on the immune responses in humans due to obvious cost and the availability of alternative preclinical models. However, as discussed above, various promoters have been described, optimised, or constructed, and their expression activity was evaluated in vitro and in vivo. Improved immune responses, T cell responses in particular, were observed with some new promoters, in comparison to the conventionally used ones, discussed above. This improvement might increase the immunogenicity of some current vaccines that are already in clinical testing. For instance, MVA85A is a vaccine against TB, which has been recently tested in humans (infants) as the first TB vaccine to enter clinical trial phase IIb since the BCG was widely introduced in the 1940s. The trial resulted in modest cellular immunogenicity and modest protection (17% efficacy, not significant).⁶⁹ Although a vaccine against a complex pathogen such as TB might require more than changing a promoter, it would benefit from enhancing the vaccine immunogenicity by using a strong promoter.

Concluding remarks

MVA is a highly immunogenic and safe viral vector, and there is accumulating evidence of its safety and immunogenicity gained from many clinical trials. This makes it a vector of choice, especially as a booster. There are not many vectors with both advantages and any improvement to MVA would add to its excellent safety and immunogenicity. The use of optimal promoters to drive transgene expression is a well-known approach to enhance the cellular immunogenicity of T cell-inducing vaccine vectors such as MVA. Therefore, researchers should continue to add to this area of poxvirology; and more promoters are expected to be described or optimised to further enhance the potential of many MVA vectored vaccine candidates. This review is mainly focused on the frequently utilised rMVA promoters for enhancing vaccine immunogenicity. Therefore this article could not be inclusive for all other poxviral promoters that are being described in the literature.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

The author thanks Prof. Sarah C. Gilbert and Prof. Simon J. Draper at the Jenner Institute, University of Oxford, for their scientific guidance and support during and after his DPhil study.