Developing the next-generation of adenoviral vector vaccines

ABSTRACT

The COVID-19 pandemic saw the first extensive use of adenoviral vector vaccines, with over 3 billion doses produced during the first year of the pandemic alone and an estimated 6 million lives saved. These vaccines were safe and effective, and could be produced at low cost in several continents allowing widespread use in low- and middle-income countries (LMICs). Despite their successful deployment against SARS-CoV-2, their impact has been overshadowed by relatively lower immunogenicity in contrast to mRNA vaccine technologies and very rare but serious adverse events such as vaccine-induced thrombotic thrombocytopaenia (VITT). The next-generation of adenoviral vector vaccines must address these challenges: here, we explore strategies to improve immunogenicity and safety by novel serotype selection, vector engineering, capsid modification and new delivery technologies, and discuss opportunities for next-generation adenoviral vectors against infectious disease and cancer.

Adenovirus biology and their use as vaccine vectors

Adenoviruses are a family of medium-sized non-enveloped double-stranded DNA viruses, with over 65 serotypes from seven species (A-G) in a single genus (Mastadenoviridae) known to infect humans and primates.¹ Most people are exposed to at least one serotype during childhood, with typically asymptomatic mild respiratory, gastro-intestinal or ocular signs after infection.² After their discovery in the 1950s, adenoviruses were studied extensively as a model pathogen, leading to the discovery of mRNA splicing and contributing to major advances in cell and molecular biology such as understanding of gene regulation and protein folding.¹,^3–5 This was followed by considerable interest in adenoviruses as vectors for gene therapy owing to their wide cellular tropism, lack of integration into host genomes and high packaging capacity relative to other vectors.⁶ Although efficient vehicles for gene transfer, adenoviral vectors were associated with significant innate immune activation and robust immunogenicity against encoded gene products.^7,8 This discovery led to the use of adenoviral vectors as vaccines for infectious diseases and cancer: since the first adenoviral vector vaccine was developed in 1996, they have been used in over 200 clinical trials and a total of 8 vaccines have received licensure in at least one country (Table 1).^7,9,10 During the COVID-19 pandemic, several adenoviral vector vaccines were approved – the most widely used of these was the ChAdOx1-nCoV-19/AZD1222 vaccine developed by Oxford/AstraZeneca, followed by the Johnson & Johnson Ad26.COV2.S vaccine, the Gamaleya Institute’s Sputnik V vaccine combination, and the CanSino Ad5-nCoV-S vaccine.^11–16

Table 1.

Vaccination strategies using an adenovirus vector which have moved to clinical trials.

Strategy	Vaccine	Vector serotype	Indication	Clinical trial phase	Publications
Alternative vector usage	Ad26.ENVA.01	Ad26	HIV	I	191,192
	Ad26.ZEBOV	Ad26	Ebola Zaire	Licensed product	193–203
	Ad26.Mos4.HIV	Ad26	HIV	III	204
	Ad26.RSV.preF	Ad26	RSV	I, II, III	205–213
	Ad26.ZIKV.001	Ad26	Zika	I	214
	GamCovVac	HAd5, HAd26	SARS-CoV-2	Licensed product	215–218
	Ad35.ENVA	Ad35	HIV	I	192
	Ad35.CS.01	Ad35	Malaria	I, II	32,219,220
	AERAS-402	Ad35	TB	I	221
	Ad6NSMut	HAd6	HCV	I	20
	BBV154	ChAd36	SARS-CoV-2	Licensed product	166,222
	ChAd63-ME-TRAP	ChAd63	Malaria	I, II	28,223
	ChAd63-CS	ChAd63	Malaria	I	224
	ChAd63.HIVconsv	ChAd63	HIV	I	225
	ChAd63 PvDBP	ChAd63	Malaria	I	226
	ChAd63 Rh5	ChAd63	Malaria	I	227
	ChAd63-CA/ChAd63-CAT	ChAd63	Malaria	I	228
	ChAdOx1-MERS	ChAdOx1	MERS	I	19,229
	ChAdOx1 Chik	ChAdOx1	Chikungunya	I	29
	ChAdOx185A	ChAdOx1	M. tuberculosis	I	230
	AZD1222	ChAdOx1	SARS-CoV-2	Licensed product	11–13,231–236
	AZD2816	ChAdOx1	SARS-CoV-2	II/III	233,237
	ChAdOx1-5T4	ChAdOx1	Cancer	I, I/II	238
	ChAdOx1 biEBOV	ChAdOx1	Ebolavirus (Zaire + Sudan)	I	31
	ChAdOx1 NP+M1	ChAdOx1	Influenza A	I	239,240
	ChAdOx1 RVF	ChAdOx1	RVF	I, II	23
	ChAdOx1 hTI	ChAdOx1	HIV	I, II	241–243
	ChAdOx1.tHIVconsvX	ChAdOx1	HIV	I, II	244
	VTP-300	ChAdOx1	HBV	I/II, II	245
	ChAdOx2-HAV	ChAdOx2	HAV/Crohn’s	I	246
	ChAdOx2-RabG	ChAdOx2	Rabies	I, I/II	22
	ChAd155-RSV	ChAd155	RSV	I/II	85,247,248
	ChAd155-Rabies	ChAd155	Rabies	I	249,250
	cAd3 EBO Z	ChAd3	Ebolavirus (Zaire)	I, I/II	27,251,252
	cAd3 EBO S	ChAd3	Ebolavirus (Sudan)	I, II	253
	cAd3 Marburg	ChAd3	Marburg	I, II	30
	ChAd3-hIiNSmut	ChAd3	HCV	I	254
	PanAd3-RSV	ChAd3	RSV	I	24,255
	PRGN-2009	GC46	Cancer	I, II	256
	GRAd-CoV2	GrAd32	SARS-CoV-2	I, II/III	83,84,257
Vector Backbone design	rcAd26.MOS1.HIV	RC-HAd26	HIV	I	159
	Ad4-H5-VTN	RC-HAd4	Influenza A	I	187
	AdCLD-CoV19	HAd5/HAd35	SARS-CoV-2	I, I/II	258
Capsid engineering	Ad5HVR48.ENVA.01	HAd5/HAd48	HIV	I	116
	BBV154	ChAd36	SARS-CoV-2	Licensed product	166,222
Intranasal delivery	Nasovax	HAd5	Influenza A	II	259
	Ad4-H5-VTN	RC-HAd4	Influenza A	I	187
Aerosol delivery	Ad5-S	HAd5	SARS-CoV-2	Licensed product	164,260–264
Oral delivery	ND1.1	HAd5	Influenza A	I	265
	VXA-A1.1	HAd5	Influenza A	I	171
	Ad4-H5-VTN	RC-HAd4	Influenza A	I	187
	rcAd26.MOS1.HIV	RC-HAd26	HIV	I	159
	VXA-G1.1-NN	HAd5	Norovirus	I, II	43,169
	VXA-CoV2–1	HAd5	SARS-CoV-2	I, II	44

Adenoviral vector vaccines are typically rendered replication-incompetent by deletion of the E1 gene. In the wild-type virus, this encodes a set of differentially spliced products which trans-activate expression of early proteins involved in DNA replication and are critical for the viral life cycle.¹⁷ Most vaccine vectors also harbor additional deletions in the E3 region, which is required for viral immune evasion in vivo but expendable during production in cell culture.¹⁷ Although the virus can replicate in specialized E1-complementing producer cell lines, E1/E3 deleted vectors are incapable of replication in the host – instead, the deleted genes are replaced by a transgene expressed after host cell entry, leading to adaptive immune responses to the encoded antigen Figure 1.^17,18 There are several key advantages of adenoviral vectors that justify the continued development of this platform: in comparison to DNA or protein subunit vaccines, adenoviral vectors have consistently elicited robust immunogenicity against a wide range of antigenic transgenes in both humans and animals across a number of clinical trials.^19–33 Adenovirus entry into host cells stimulates several innate immune pattern recognition receptor (PRR) signaling pathways – in contrast to protein-subunit or virus-like particle (VLP) platforms, this intrinsic immunogenicity means adenoviral vector vaccines do not require adjuvants, which can be expensive and complicate vaccine supply and manufacture.^34,35 Notably, strong immune responses are still seen in elderly individuals and other immunocompromised groups that are often refractory to other types of vaccine.^36–38 Even with the rise of mRNA vaccines and advances in adjuvant systems for protein-based vaccines (for example, CpG, AS01 and saponin-based formulations), adenoviral vectors are the most effective vaccine modality for eliciting potent CD8+ T cell responses, making them a promising platform against cancer and chronic infections.^39–41 Adenoviral vectors have also been licensed for delivery by aerosolised or intranasal routes, with a demonstrated ability to induce local tissue-resident memory responses and secretory IgA that may prove critical for protection against respiratory and gastrointestinal pathogens.^42–45 Other platforms such as mRNA or VLPs currently lack approved mucosal applications and will likely require further optimization to address challenges in mucosal delivery.⁴⁶ Finally, adenoviral vectors have proven to be a valuable tool against outbreak pathogens, especially in LMICs. The adenoviral genome is well-suited to manipulation and can be rapidly modified to introduce new transgenes against different pathogens.^47,48 The overall cost of production is also relatively low (approximately $3–4 per dose) and manufacture is easily scalable from research grade to industrial level production.^49,50 In contrast to mRNA or live vaccines, adenoviral vectors have long-term stability at standard refrigerator temperatures enabling low-cost distribution without specialist storage facilities.⁵¹ For example, the BNT162b2 and mRNA-1273 vaccines were initially priced at approximately $19.50 and $25–37 per dose and required storage at −70°C or −20°C respectively, creating substantial barriers to equitable deployment.⁴⁹ Overall, these features have made adenoviral vectors a leading platform technology: alongside mRNA vaccines, they were rapidly deployed against COVID-19 and were critical to the global response to the pandemic. The AZD1222/ChAdOx1-nCoV-19 vaccine received its first approval before the end of 2020, less than a year after the SARS-CoV-2 spike sequence was released. Within the next 12 months, over two billion doses had been released, estimated to have saved over 6 million lives.^52–54

Figure 1.

Adenovirus replication cycle. Adenoviruses (Ad) attach to primary cellular receptors such as CAR or CD46 and secondary integrins in order to enter host cells (1). After endocytosis (2), the acidic environment facilitates uncoating (3) and subsequent release of the virion into the cytoplasm (4). The virion is transported to the nuclear pore complex, where viral DNA is released to enter the nucleus (5). The early (E) genes are transcribed (6) followed by extensive mRNA splicing and translation of early gene products (7). These proteins reenter the nucleus to facilitate viral genome replication (8), transcription of late viral genes (9) and export of mRNA to the cytoplasm for translation (10). The newly synthesized L proteins return to the nucleus to enhance further replication, transcription, and viral assembly (8–11). Progeny virions are assembled in the nucleus (11) and released by cell lysis (12). In replication-deficient adenoviral vaccine vectors, steps 1 to 5 proceed similarly to replication-competent Ad. However, replacement of the E1 gene with an antigenic transgene prevents viral replication, instead leading to transgene expression (13, 14). Single-cycle adenoviral vaccine vectors keep the E1 gene but lack late genes involved in assembly or maturation; as a result, they retain the ability to replicate the viral genome leading to enhanced transgene expression (15, 16). Figure created with Biorender.

Despite these advantages, there are several limitations that need to be addressed in next-generation adenoviral vectors. Although they offered similar protection from hospitalization and death from COVID-19, mRNA vaccines were associated with improved efficacy against mild disease in clinical trials and better immunogenicity, with viral vectors typically eliciting more modest antibody responses in homologous prime-boost regimens.^38,55 The immunogenicity of adenoviral vector vaccines can be limited by host responses to the adenovirus itself – these anti-vector responses can block viral entry and expedite clearance of transduced cells.⁵⁶ Antibodies targeting the fiber knob and surface-exposed ‘hypervariable regions’ (HVRs) in the hexon can block entry, either through neutralizing interactions with cell-surface receptors, aggregating virions, blocking endosomal escape or intracellular trafficking, or directing viral degradation by TRIM21.^57–61 In addition, CD8+ T cell responses to adenoviral proteins (including structural capsid proteins such as the fiber, hexon or penton and viral gene products generated by ‘leaky’ expression even in replication-deficient vectors) can enhance clearance of adenovirus-transduced cells leading to reduced transgene expression.^7,8,62,63 These anti-vector responses can preclude the use of adenoviral serotypes commonly found in the human population and can limit the re-usability of the same vectors in prime-boost regimens or against different antigens.⁶⁴ In a trial performed during the pandemic, peak responses to a ChAdOx1-HBV vaccine were reduced in individuals vaccinated with the AZD1222/ChAdOx1-nCoV-19 vaccine approximately 3 months before, suggesting recent prior exposure may impact re-usability of rare serotype vectors in the future.⁶⁵ More concerningly, the widespread deployment of adenoviral vector vaccines was followed by reports of a very rare but serious complication after vaccination.^50–52 Vaccine induced thrombocytopaenia and thrombosis, or VITT, is characterized by systemic thrombosis (classically central venous sinous thromboses) with low platelet counts arising 5–30 days after vaccination.⁵³ VITT was observed in approximately 1 in every 27,000–100,000 first dose adenoviral vector vaccinations, with lower rates after booster doses.^66,67 Early investigations in affected patients identified high levels of auto-antibodies targeting platelet factor 4 (PF4), a platelet derived cationic protein, resulting in Fc-mediated platelet activation and aggregation. This leads to hypercoagulability followed by thrombocytopaenia due to platelet consumption and clearance.^{50–52,54,55} The mechanism behind auto-antibody induction is unclear: many studies have since confirmed that PF4 binds the surface of ChAdOx1, Human Adenovirus 26 (HAd26) and Human Adenovirus 5 (HAd5), likely due to low-affinity electrostatic interactions with the hexon protein in the viral capsid.^56–58 This could plausibly escape self-tolerance, either due to the combination of PF4 binding in the context of adenovirus-induced innate immune stimulation, the repetitive array of bound PF4 on the vector surface, or through cross-priming of anti-PF4 B cells after co-endocytosis of PF4:viral particle complexes. Other theories have implicated impurities such as residual producer cell protein or DNA, effects of the spike transgene, or platelet activation due to the pro-inflammatory cytokine milieu elicited by adenoviral vectors.^57–59 The affinity of PF4 to the HAd5 capsid was lower than to ChAdOx1 or HAd26 - however, HAd5 uniquely has high affinity to another plasma protein (Factor X) which enables liver transduction leading to hepatotoxicity at high doses.⁶⁸ More concerningly, an increased rate of HIV acquisition in men (but not women) was seen after vaccination with HAd5 vectors encoding HIV antigens in two clinical trials, an association not observed in more recent trials with other serotypes.^69–71 Future adenoviral vector vaccines will have to address these limitations: this review focuses on strategies to improve vaccine immunogenicity and safety, evade preexisting responses, and utilize new technologies to improve vaccine delivery.

Improving serotype choice and vector backbone design

The choice of adenovirus serotype offers an opportunity to exploit natural variation in vector biology. The earliest adenoviral vector vaccines were developed using HAd5, a species C human adenovirus; although highly immunogenic in unexposed individuals, preexisting immunity to HAd5 is common in most human populations and was associated with reduced immunogenicity in human clinical trials.^2,64 This prompted the development of viral vectors from rare human serotypes (eg HAd26) and simian adenoviruses (eg ChAdOx1, derived from Chimpanzee Adenovirus Y25), in addition to capsid engineering approaches discussed below.^72,73 Despite this, there are important differences in host-adenovirus interactions between different serotypes, with alternative host cell receptors used for cell entry, different routes of intracellular trafficking and variable induction of innate immune signaling pathways.^74,75 In naïve preclinical models, HAd5 and related group C adenoviruses consistently elicit more robust immune responses than rare serotype or simian adenoviruses from species B, D and E such as HAd26, Human Adenovirus 35 (HAd35), Chimpanzee Adenovirus 63 (ChAd63) and ChAdOx1.^76–79 HAd5-based vectors were used in the COVID-19 pandemic and often elicited comparable immune responses to rare serotype vectors, perhaps surprisingly given the high seroprevalence of anti-vector responses.⁸⁰ Investigation of natural adenovirus diversity to identify novel group C adenoviruses with low seropositivity may therefore represent a strategy to bypass preexisting immunity and utilize the inherent immunogenicity of HAd5-like vectors. Pre-clinical data from group C primate adenoviruses supports this – a vector based on a group C gorilla adenovirus (GC45) matched HAd5 immunogenicity in naïve animals using a malaria antigen and has substantially lower preexisting responses in a human sero-survey.⁸¹ Several non-human group C adenovirus derived vectors have now entered clinical trials, including Chimpanzee Adenovirus 155 (ChAd155), Gorilla Adenovirus 32 (GrAd32) and Chimpanzee Adenovirus 3 (ChAd3), showing promising immunogenicity and tolerability.^82–85 Alternatively, understanding the mechanism of improved immunogenicity may inform engineering of existing vectors – a comparison of several human adenoviral vectors identified that HAd5 and ChAd3 vectors elicit weaker IFN signaling and drive higher transgene expression than rare serotype vectors, suggesting approaches to evade IFN-mediated translational arrest may improve transgene-directed immune responses.⁷⁷

There are other potentially useful differences in vector biology that could be exploited for vaccine development. For example, Human Adenovirus 40 and 41 (HAd40 and HAd41) have natural tropism for gastrointestinal epithelia, prompting their use in development of orally delivered vaccines.⁸⁶ Identifying other serotypes with improved entry to the respiratory mucosa or tropism for antigen-presenting cells may help elicit durable responses at immunological barrier sites, a major hurdle in development of infection/transmission-blocking vaccines for respiratory pathogens. Differences in adenoviral biology may also be exploited to improve vaccine safety: one group showed PF4 binding varied considerably across a panel of adenovirus serotypes, with Human Adenovirus 34 (HAd34) showing no detectable binding.⁸⁷ Although the effect on VITT has not been demonstrated, the use of novel serotypes screened for lack of undesirable capsid interactions could be used to develop safer next-generation viral vectors. Beyond the seven species of mastadenovirus that infects humans and primates, a large diversity of other adenovirus families with potentially exploitable features circulate in other vertebrates. A snake barthadenovirus (previously atadenovirus) was recently shown to have acquired an additional capsid protein, leading to hyper-stable virions which retained remarkable infectivity after heating or long-term storage at ambient temperatures; this improved thermostability could improve delivery without need for a cold-chain distribution network.⁸⁸ Other distant adenoviruses have already been used as vectors with promising results – for example, a Bovine Adenovirus 3 (BAd3) vector encoding the influenza A H5 hemagglutinin matched HAd5 immunogenicity in mice after intramuscular vaccination, but showed significantly better responses after intranasal vaccination enabling equivalent protection even at a 30-fold lower dose.⁸⁹ This may be a result of improved mucosal tropism: BAd3 uses sialic acid as an entry receptor which is abundant in mucosal epithelia, rather than the Coxsackievirus and Adenovirus Receptor (CAR) protein used by most human and primate vectors. However, another group using a BAd3-vectored TB vaccine showed improved dendritic cell (DC) activation and antigen presentation than HAd5 vectors, suggesting this could result from inherent differences in immunogenicity between different adenovirus genera.^89,90 Overall, expanding the repertoire of vectorized adenoviral serotypes, including by sampling more diverse non-human adenoviruses, could be used to exploit natural differences in vector biology that improve immunogenicity, safety and deployability in addition to escaping preexisting immunity.

A different approach to improving vector immunogenicity harnesses the adenovirus replication cycle to improve transgene responses. Historically, live-attenuated viral vaccines are associated with durable immune responses at low doses.⁹¹ Replication-competent adenoviruses (RC-Ads) with the native E1 region restored can replicate their genome up to 10,000-fold in infected cells, resulting in substantially higher transgene expression and more robust immune responses.⁹² Although the production of infectious progeny is clearly undesirable and poses safety risks in immunocompromised individuals, several groups have improved on this strategy by developing single-cycle adenoviruses (SC-Ads) capable of just a single round of replication in the host.^92,93 By retaining the native E1 gene and instead deleting structural/assembly genes expressed late in the adenovirus replication cycle such as the capsid cement protein IIIa or the viral protease, SC-Ads are unable to produce infectious progeny but preserve the ability to amplify their genomes within host cells.^94,95 This results in substantially improved transgene expression, with 33-fold greater influenza A hemagglutinin expression by a single-cycle Human Adenovirus 6 (SC-HAd6) compared to a replication-deficient vector, leading to significantly higher and longer-lasting antibody responses in preclinical models.⁹² Similarly, a SC-HAd6 vectored vaccine elicited durable and robust responses to the Ebola glycoprotein in mice, Syrian hamsters and rhesus macaques, including strong CD8+ T cell activation, which were superior to those elicited by a replication-deficient vector.⁹⁶ SC-Ad vectors have also been developed against SARS-CoV-2 – after intramuscular or intranasal delivery in Syrian hamsters, SC-Ads encoding the spike protein consistently outperformed replication-deficient adenoviruses. Notably, antibody responses to an SC-Ad vector seemed far more durable with very little reduction over 10 months following a single vaccination, leading to significantly better protection after challenge at this late timepoint.⁹⁷ The first SC-Ad vector has now entered clinical trials, a major milestone in translating this technology into human use.⁹⁸

Within the gene therapy field, several well-established strategies have been used to expand the insert capacity for transgenes within the adenoviral genome: these could also be used to improve vector immunogenicity. The adenoviral vector vaccines licensed during the COVID-19 pandemic harbored deletions of the E1 and E3 genes only, leaving approximately 8 kilobases (kb) for foreign sequence incorporation Figure 2.⁹⁹ Notably, there is evidence of background ‘leaky’ expression of other adenoviral proteins by vectors even in the absence of E1 transactivation, which may contribute to clearance of transduced cells by anti-vector CD8+ T cells after recognition of conserved vector-derived antigens.^8,100 In contrast, high-capacity adenoviruses (also referred to as gutless or helper-dependent adenoviruses) can be produced by eliminating nearly all the adenoviral coding sequence, retaining only the inverted terminal repeats (ITRs) and packaging signal (ψ) necessary for genome replication and encapsidation.^99,101–104 Although these require the deleted viral functions to be complimented in trans during production, high-capacity adenoviruses can accommodate inserts up to 105% of the wild-type viral genome, potentially enabling the insertion of multiple antigens simultaneously.^{99,101,102,104} Helper-dependent adenoviruses may therefore provide a useful framework for multi-pathogen or multi-antigen vaccines, with potential utility against outbreak pathogen families such as filoviruses.¹⁰⁵ From a public health perspective, the ability to encode multiple antigens within the same construct is far more economically viable than independent production of multiple monovalent vaccines mixed prior to administration, as would be required for protein or mRNA vaccines.^106,107 The elimination of the remaining wild-type vector genes from high-capacity adenoviruses may also impact immunogenicity – in preclinical models, high-capacity adenoviruses were found to elicit broader multi-specific cellular responses against a wider range of antigen-derived peptides than traditional viral vectors, which was suggested to result from reduced antigenic competition from vector-derived products.^108–111

Figure 2.

Adenovirus genome organization. Adenovirus (Ad) genomes vary among the different adenoviral serotypes but are typically about 36 kb in size. The adenoviral wild-type (WT Ad) genome contain inverted terminal repeats (ITRs) at both ends and a packaging signal (ψ) following the 5’ ITR. The genome consists of early (E), late (L), and major late promoter (MLP) genetic sequences (see upper panel). In first-generation adenoviral (ΔE1/ΔE3 Ad) vaccine vectors, the E1 and E3 coding sequences are deleted, allowing for the insertion of foreign genes (transgenes) of up to 8 kb in size (middle panel). Helper-dependent adenoviruses (HD-Ad), with most of the viral genome removed, enable the insertion of larger genetic payloads, including multiple antigens (lower panel). The theoretical maximum capacity for transgene insertion is approximately 105% of the wild-type adenoviral genome. Figure created with Biorender.

Finally, the ability to incorporate multiple transgenes into the genomes of E1/E3 deleted viral vectors or high-capacity adenoviruses could be used to encode additional immunomodulatory factors to enhance vaccine responses. This has been attempted before with variable success – for example, direct expression of the human TRAM protein (an adaptor in TLR2 and TLR4 signaling transduction) enhanced CXCL8 secretion from virally transduced cells and led to stronger CD8+ T cell responses (but weaker antibody responses) in mice, but this failed to translate to non-human primates.¹¹² A similar approach used a vector encoding the Eimeria tenella 3-1E protein, shown to induce dendritic cell activation and IL-12 secretion following TLR11 agonism, which enhanced CD8+ T cell responses to GFP or an HIV-gag transgene.¹¹³ More recently, co-delivery of a vector carrying a monocistronic construct encoding an anti-CTLA4 monoclonal antibody (mAb) alongside a vector encoding the SARS-CoV-2 spike protein led to significantly enhanced T cell and total IgG responses.¹¹⁴ The same approach also improved cytotoxic T cell responses to a cancer neo-epitope vaccine.¹¹⁴ Interestingly, this strategy required co-delivery of the antigen and mAb-encoding vectors at the same injection site, suggesting local blockade of the inhibitory CTLA4 receptor at the same time as transgene expression was needed.¹¹⁴ The improvement in immunogenicity achievable by co-expression of antigens and immunomodulatory factors must be balanced with potential risks of nonspecific immune stimulation – however, especially in cancer vaccinology, the large packaging capacity of viral vectors may be a useful tool for tailoring vaccine responses.

Capsid engineering

The adenoviral vectors used in the COVID-19 pandemic all retained the unmodified capsid proteins of their original serotypes. However, substantial progress has been made using modifications of viral capsid proteins (Figure 3) in order to alter viral tropism, improve immunogenicity and prevent capsid interactions with serum antibodies and plasma proteins. The ability to engineer the viral capsid was first explored in order to escape anti-vector responses by modification of the exposed HVRs in the hexon protein – these unstructured protein loops are a major target of neutralizing antibodies after natural adenovirus infection or immunization.^59,61 HAd5 chimeric vectors, created by replacing the HVR loops of HAd5 with those from rarer serotypes such as human adenovirus 48 (HAd48), have been used to evade preexisting neutralizing responses.^115–117 More recently, a machine-learning model based on a variational autoencoder framework has been used to generate diverse synthetic hexon sequences predicted to fold into native hexon-like structures, potentially simplifying generation of novel or chimeric vectors with low preexisting immunity.¹¹⁸ However, even synthetic serotypes would still be expected to elicit de novo responses that could limit re-use of the same vector. Instead, several groups have attempted to shield the adenovirus surface to prevent exposure of neutralizing epitopes – for example, by chemical PEGylation of the virion or encapsulation in liposomes or sodium alginate microspheres.^119–123 These approaches have been successfully used to evade preexisting responses and can reduce de novo responses to the virion leading to improved responses after boosting – however, chemical modifications can reduce vector infectivity and need to be compatible with low-cost GMP grade manufacture.¹¹⁹ More promisingly, other groups have modified capsid proteins to allow spontaneous shielding of the vector – for example, through incorporation of an albumin-binding peptide to allow decoration with serum albumin after administration, or incorporating a biomineralization-initiating peptide to drive spontaneous hydroxyapatite shell formation.^124,125 In both cases, these approaches shielded from anti-vector neutralization with mild or no effect on virus infectivity.^124,125 Capsid modification may also provide an opportunity to develop safer vectors with reduced risk of VITT. HAd5 PEGylation and HVR1 modification have been demonstrated to reduce PF4 binding in a qPCR-based assay, suggesting capsid modification or shielding may open a route toward vectors with lower toxicity or fewer safety concerns.⁸⁷ Further work, including developing reproducible animal models, is needed to confirm whether these approaches could reduce incidence of VITT after vaccination.

Figure 3.

Adenovirus structure. The structure of the adenovirus vector. The adenovirus virion has an icosahedral capsid composed of 240 hexon trimers, with 12 vertices occupied by a pentamer of the penton protein and trimeric fiber proteins that extend from each penton base. The fiber proteins consist of a shaft with variable numbers of repeats between serotypes and a fiber knob critical for receptor binding during cell entry. The capsid also contains protein IX (pIX), protein VI (pIX), protein VIII (pVIII) and protein IIIa (pIiia) – these cement proteins which contribute to capsid stability, cell entry and intracellular trafficking. Within the core, the linear double-stranded DNA viral genome is capped with the terminal protein (TP); other core proteins include protein V (pV), protein mu (pMu), protein IVa2, protein VII (pVII) and the adenoviral protease.

Rather than attempting to reduce host responses to the capsid, immunogenic epitopes from other pathogens have been displayed on the surface of the adenoviral vector in order to enhance vaccine immunogenicity. The hexon HVR loops of HAd5, ChAd68 and HAd3 have been shown to tolerate insertion of foreign sequences leading to their high-density display on the virion surface.^126–138 This has been associated with potent humoral responses to displayed epitopes, as elicited by other virus-like particle (VLP) platforms, where the repetitive nanoscale organization of antigens can optimally activate B cells via B cell receptor (BCR) cross-linking.¹³⁹ This strategy has been used to direct responses to linear epitopes from Plasmodium falciparum, influenza A, human immunodeficiency virus (HIV) and several other pathogens, with antibody responses to hexon-incorporated epitopes often greatly exceeding responses to the same antigens encoded as a transgene.^126–138 In addition to the hexon, short linear epitopes have also been inserted into the fiber and penton proteins – although these proteins are less abundant per virion than the hexon, these have still proven effective at eliciting responses to displayed epitopes.¹³⁸ Insertions into most capsid proteins are limited to small peptides as larger insertions reduce viral infectivity or completely abolish rescue.¹⁴⁰ In contrast, the C-terminus of the hexon-associated protein IX can tolerate fusion of larger protein domains, allowing whole antigens from Plasmodium falciparum, Human papillomavirus and Trypanasoma cruzi (as well as model antigens such as GFP) to be displayed.^{138,140–143} This has also been associated with improved T cell responses to fused antigens, potentially due to leaky expression of protein IX downstream of E1 inserted antigen cassettes with promiscuous promoters.^141,144 Although these approaches have promising immunogenicity, capsid modifications can reduce virion stability or yield and may be lost during viral passage.¹⁴⁵ There are also fundamental limitations on which antigens can be displayed – genetically-encoded antigens must assemble within the steric constraints of the virion capsid and in the same cellular sub-compartment, precluding native oligomerisation and acquisition of post-translational modifications such as glycosylation or disulfide bond formation.¹⁴⁶ A promising approach to bypass these obstacles uses post-translational coupling of antigens to the virion. The incorporation of a peptide DogTag into an HAd5 hVR loop allowed coupling of DogCatcher-antigen fusions to the virion surface by isopeptide bond technology.¹⁴⁷ Display of the SARS-CoV-2 Receptor Binding Domain (RBD) led to over 10-fold better humoral responses than a vector encoding the full-length spike as a transgene, while maintaining potent T cell responses to encoded antigens.¹⁴⁷ A similar strategy used unnatural amino acids incorporation to facilitate click-chemistry mediated addition of exogenous ligands without affecting viral yield or infectivity.¹⁴⁸ Although these strategies require separate purification of recombinant proteins, this allows coupling of ligands in their native confirmations retaining antigenicity or functionality without requiring large potentially genetically unstable modifications of the viral genome. These approaches are therefore a promising way to leverage the potent humoral immunogenicity of VLP-based vaccines with the ability of adenoviral vector platforms to elicit cellular responses to an encoded antigenic transgene.

In addition to display of epitopes, modification of capsid proteins can also be used to direct entry into specific cell types. The adenovirus fiber protein mediates entry into host cells – for most human and primate serotypes this occurs after binding to the CAR receptor, with some serotypes binding CD49⁷⁴. Several groups have introduced targeting motifs into the variable loops in the fiber shaft or replaced the fiber knob with alternative ligands in order to modify viral tropism. For example, introducing the integrin-binding RGD motif from the adenovirus penton into fiber HI loops improves transduction of CAR-deficient cells, including potentially useful antigen presenting cell subsets.^149–151 Alternatively, fiber knob replacement from serotypes that use non-CAR receptors has been shown to enhance transgene delivery to mucosal sites such as the lung and intestine.^152,153 More recently, post-translational coupling techniques have allowed attachment of high-affinity targeting proteins such as nanobodies, which require internal disulphides preventing correct folding after genetic incorporation into capsid proteins. This has been achieved by incorporating the SpyTag or DogTag into the fiber HI loop, allowing SpyCatcher/DogCatcher-fused nanobody coupling by isopeptide bond formation, and has successfully been used to direct adenovirus entry into B cells.^154,155 Other approaches for post-translational fiber modification include reactive cysteine incorporation into the HI loop – this led to virion aggregation under traditional purification techniques, but was successfully used for thiol-mediated attachment of exogenous proteins.¹⁵⁶ Many of these tropism-modification approaches have primarily been used in oncolytic adenoviruses in order to enhance uptake into tumour cells or in gene therapy – however, these strategies could also be adapted for infectious disease vaccines, potentially improving viral vector delivery to specific antigen-presenting cell subsets or eliciting localized immunity in physiologically-relevant sites such as the nasal mucosa.

Novel delivery technologies and immunization routes

Innovations in delivery technologies may also present opportunities for the adenoviral vector platform. The natural stability of adenoviruses in the respiratory or gastrointestinal tract means vectors retain infectivity across a wide range of temperatures and are relatively resistant to low pH or mucus.¹⁵⁷ In contrast, mRNA-LNPs have struggled in clinical trials after mucosal delivery, with previous studies demonstrating LNP disintegration due to shear forces and aggregation during aerosolization.^46,158 These features have positioned viral vectors as a leading platform for mucosal delivery of antigens – for example, by intranasal vaccination, inhalation of nebulized viral particles or oral delivery.^33,159,160 These approaches have been successful in a number of preclinical studies, with mucosally-administered viral-vectored vaccines providing protection from infection and transmission of SARS-CoV-2 in small animal models and non-human primates.^44,161,162 Although results from clinical trials in humans have been mixed, some studies have had very encouraging results leading to licensure of intranasal and aerosolised adenoviral vector vaccines against COVID-19 in India and China.^{42,45,163-166} For example, in individuals already vaccinated with two doses of an inactivated vaccine, a single aerosolised dose of 1e10 viral particles (VP) of HAd5 encoding the spike protein elicited superior systemic T cell responses and equivalent systemic IgG as an intramuscular immunization with 5e10 VP, in addition to inducing secreted IgA detectable in saliva samples.⁴² In addition to respiratory delivery, other groups have used orally delivered adenoviral vectors to prime mucosal responses.^33,44,86 An oral tablet vaccine formulation using HAd5-vectors with a TLR3 agonist is stable at room temperature and has shown tolerability and immunogenicity in early clinical trials using SARS-CoV-2, influenza or norovirus antigens, including induction of mucosal IgA as well as systemic IgG^44,167–172 Notably, participants were challenged with H1N1/Cal09 three months after vaccination in a comparison with intramuscular injection of a quadrivalent inactivated influenza vaccine: the oral HAd5 formulation reduced both infection and viral shedding relative to the inactivated vaccine or placebo groups.¹⁷² An orally delivered HAd5 norovirus vaccine was also able to elicit durable systemic and mucosal responses, which is particularly promising given the recent clinical failures of parentally delivered norovirus subunit vaccines.^169,173 For respiratory pathogens like SARS-CoV-2, where intramuscular vaccination provides robust protection from morbidity but typically fails to block shedding or infection, the mucosal administration of adenoviral vectors may therefore be a promising approach to improve vaccine efficacy.¹⁷⁴

Adenoviral vectors have also been shown to be compatible with novel technologies for subcutaneous delivery. Delivery by jet injector or after coating onto poly(L-lactic acid) polymer microneedles have both been shown to elicit comparable immunogenicity to intramuscular vaccination.^175,176 More recently, several studies have produced silicon microneedles containing viral vector liquid formulations: these have elicited improved antibody responses to P.falciparum antigens than intradermal immunization with the same dose, and can achieve equivalent responses at a lower dose than intramuscular immunization.^177,178 Interestingly the adenoviruses incorporated into microneedle formulations showed remarkable stability, with retained vector infectivity at 40°C for 2 months and room temperature for up to 6 months.¹⁷⁸ These approaches may bypass some of the limitations of the adenoviral vector platform: subcutaneous delivery via microneedles was shown to reduce de novo anti-vector responses, and both oral and intranasal delivery can improve vaccine immunogenicity in animals pre-exposed to the same serotype.^177,179 Furthermore, intramuscular immunization has been shown to cause local microtrauma and capillary damage, allowing occasional leakage of viral vector into the bloodstream.^180,181 Anti-platelet antibodies can be elicited in animal models after direct intravascular administration of adenoviruses but not intramuscular vaccination, suggesting rare inadvertent entry into the bloodstream may be important in VITT pathogenesis.^181,182 Alternative routes of administration, such as by sub-cutaneous, inhalational or oral delivery, may therefore provide an opportunity to improve vector safety as well as eliciting crucial responses at barrier sites such as the respiratory mucosa or gastrointestinal tract.

Conclusions and future perspectives

Adenoviral vectors have been in development as vaccines for over 20 years and saw widespread deployment for the first time during the COVID-19 pandemic. Their rapid licensure and deployment were possible due to the modular nature of the vaccine platform and straightforward manufacture, and had a huge impact on reducing morbidity and mortality globally.^52,53 Many other candidates entered clinical trials but were insufficiently immunogenic or scalable for further development, including newer platforms such as DNA vaccines and self-amplifying RNA vaccines in additional to traditional platforms like protein-subunit vaccines.^183–186 While adenoviral vector vaccines proved effective, their first large scale deployment identified critical areas for improvement. One key challenge is immunogenicity – in particular, improving antibody responses in the context of homologous prime-boost regimens. Strategies to enhance immune responses include use of capsid display or single-cycle adenoviruses, or identifying group C adenoviruses that evade preexisting responses. More concerning are complications such as VITT – further investigation is needed to understand the mechanism of anti-PF4 antibody formation and develop preclinical models to allow evaluation of updated vectors. Approaches that modify or shield the capsid, or delivery techniques that limit exposure of the viral capsid to platelets or platelet-derived factors, may help mitigate these risks.

Within the infectious disease and pandemic preparedness context, approaches that can block transmission of respiratory, gastrointestinal or genitourinary pathogens by establishing robust responses at barrier sites would be particularly valuable. Mucosally delivered adenoviral vaccines have already demonstrated some clinical success – for example, the intranasal ChAd36-vectored COVID-19 vaccine BBV154, the aerosolised CanSino Ad5-nCoV and an orally administered Ad4-vectored influenza vaccine.^42,166,187 While the stability and natural broad tropism of adenoviruses already facilitates their use as mucosal vaccines, vector engineering to further modify viral tropism or direct entry to specific antigen-presenting cells may be a useful strategy to enhance tissue-resident responses. High-capacity adenoviruses may offer a scalable solution for multi-pathogen vaccine development, allowing stockpiling of vaccines with broad pan-family reactivity without requiring production and mixing of multiple components. Equally, approaches that improve thermostability or enable lower doses without compromising immunogenicity could improve vaccine access and deployment globally, a critical failure during the COVID-19 pandemic.¹⁸⁸ There could also be an increasing role of adenoviral vector vaccines within the cancer vaccine landscape: adenoviral vectors are already recognized as strong inducers of CD8+ T cell responses, and their modular nature could support development of bespoke vaccines against patient-specific neoantigens transgenes in addition to off-the-shelf vaccines against shared tumor-associated antigens.^189,190 In this context, rare adverse events may be more acceptable: approaches such as single-cycle or even replication-competent adenoviruses, or use of high-capacity adenoviruses that encode multiple immunomodulatory factors in addition to antigenic transgenes, may therefore provide a flexible and powerful tool.

Adenoviral vectors played a crucial role in the global response to the COVID-19 pandemic, credited with saving over 6 million lives in the first year of use.⁵² By addressing key challenges identified during their widespread use, a future generation of adenoviral vaccines has the potential to be at the forefront of pandemic preparedness and to support the development of safe, effective and affordable vaccines against emerging pathogens, endemic infectious diseases and cancer.

Biographies

Alexander Sampson Alex is a final year DPhil student at the Pandemic Sciences Institute (PSI) and Oxford Vaccine Group (OVG) under the supervision of Professor Teresa Lambe OBE and Dame Professor Sarah Gilbert. His research focuses on adenoviral vector platform development and the development of vaccines with broader reactivity across betacoronaviruses.

Matěj Hlaváč Matěj is a final-year DPhil student at the Pandemic Sciences Institute (PSI), working under the supervision of Professor Dame Sarah Gilbert and Dr. Susan Morris. His research focuses on adenoviral-vector vaccine technologies and platform modifications and development, with a particular emphasis on multivalent vaccine candidates.

Funding Statement

This work was supported by the VaxHub Global project under grant [EP/Y530542/1] and the UK Engineering and Physical Sciences Research Council (EPSRC) for the Manufacturing Research Hub for a Sustainable Future (VaxHub sustainable) under grant [EP/X038181/1].

Disclosure statement

The authors declare the following competing financial interests: SCG is a co-founder of Barinthus Biotherapeutics, formerly Vaccitech, and Alazid, and is named as an inventor on patents covering the use of adenoviral vectored vaccines. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be constructed as a potential conflict of interest.