Cancer DNA vaccines: current preclinical and clinical developments and future perspectives

Chimeric DNA vaccines are vaccines that encode xenogeneic antigens. They are proteins or peptides derived from different species in which the sequence is significantly homologous with the self-ortholog [43]. The subtle differences between the epitopes of the orthologue and the native protein elicit T and B cell responses against the xenoantigen [13, 43]. Hence, xenogeneic antigens are recognized as “non-self-antigens”, thus circumventing immune tolerance while preserving an optimal homology to allow T cell recognition [13, 44]. During recent years, different studies have demonstrated the higher efficacy of xenogeneic antigens compared to autologous antigens [44, 45]. A complex DNA vaccine construct that delivers several xenogeneic epitopes dramatically increased the CTL antitumor activity [46]. The efficacy of DNA xenovaccines was also tested in dogs [47], leading to the approval of the first xenogeneic DNA vaccine against human tyrosinase, Oncept, for the treatment of oral malignant melanoma in dogs [43].

It is also possible to design hybrid plasmids, which code for chimeric proteins that include both xenogeneic and homologous antigen domains [13]. In this type of plasmid, the xenogeneic moiety can circumvent immune tolerance and induce a more potent cellular response, while the homologous sequence can stimulate the activation of a broader immune response [43]. Indeed, the chimeric protein produced by transfected cells can be taken up by DCs, thus activating the T cell immune response but it can also be recognized and internalized by B cells [43]. Quaglino et al. found that the plasmid encoding the chimeric neu-Her-2 antigen was superior to both the fully autologous and the fully xenogeneic vaccines in inducing a protective antitumor immune response against ErbB2⁺ tumors [48]. Starting from these results, other DNA vaccines were constructed by shuffling genes from mouse, rat, human and other species, improving the antigen immunogenicity and vaccine efficacy [49‐52]. DNA xenovaccination has also been tested in the clinic in melanoma patients, with encouraging results [53, 54], and one clinical study (NCT00096629) using the human and murine prostate-specific membrane antigen is ongoing (Table 2).

Most anticancer DNA vaccines, both past and present, immunize using nonmutated TAs. However, these antigens are often present in normal or germline tissues, which can prevent a strong immune activation because of immune tolerance [55]. Several clinical trials using nonmutated TAs have failed to demonstrate beneficial effects compared with the standard of care treatment [14]. In contrast, neoantigens are the result of tumor-specific DNA alterations that create new epitopes. Due to their specific expression in cancer tissue and the potential lack of side effects, they represent ideal targets against cancer and can be used in the design of cancer vaccines [56, 57]. They can also turn “cold” tumors into “hot” ones and mediate the upregulation of PD-L1 in the TME, thus extending the applicability of the anti-PD-1/PD-L1 immunotherapy [58]. Neoantigens are presented by APCs to CD4+ and CD8+ T cells to activate an immune response. They are highly tumor-specific and, therefore, they represent an attractive immunotherapy target. It is expected that they are not affected by T cell tolerance, as they may be recognized as non-self by the host immune system and, thus, generate a specific anti-tumor response [59, 60]. Their identification starts with exon sequencing from a tumor biopsy. Then, mutations are identified compared to whole exome data from normal tissue. Prediction algorithms select those antigens that are recognized by MHC class I or II. Finally, in vitro and in vivo studies validate their ability to stimulate the CD8+ immune response, especially a CD4 response [61, 62]. However, not all peptides are immunogenic, and identifying which mutations are targeted by the immune system is currently a subject of great interest. Hence, the prediction of the immune response to neoantigens needs to be optimized. Assessing the immunogenicity of each neoepitope is not reasonably applicable on a large scale. Current computational approaches are being refined to improve the accuracy of neoantigen identification and are discussed in detail in [63]. Integrated pipelines will need to be developed beginning with tumor genomic characterization, variant analysis, and the accurate prediction of which mutations are likely to give rise to tumor-specific neoantigens [64]. Other hurdles are associated with the use of personalized neoantigens for cancer immunotherapy, such as the manufacturing time. The median period for the discovery and production of a personalized vaccine is approximately 4.5 months [65]. In particular, the time from the selection of mutations to vaccine release ranges from approximately 89–160 days [66]. This amount of time has to be reduced to cure patients with metastatic disease. Another issue concerns the genetic heterogeneity of tumors [67]. Thus, targeting a unique neoantigen would probably lead to the selection of antigen non-expressing tumor cells. It has been demonstrated that the use of a poly-epitope neoantigen RNA vaccine encoding up to 10 neoantigens was effective in 8/13 melanoma patients who were completely tumor-free after one year [66]. Compared to RNA and peptide vaccines, DNA vaccines seem to elicit a more potent CD8 response against the encoded neoantigens, making them more attractive for cancer vaccination [60, 68]. Hence, once identified, the neoantigen can be cloned into a DNA vaccine. This personalization permits the design of cancer vaccines tailored to each patient.

An advantage of DNA vaccines is the possibility of delivering several antigen genes in the same construct, at the same time and with the same delivery method. The presence of immunodominant and unconventional epitopes simultaneously delivered by a polyepitope DNA vaccine can induce a broad CTL response specific to multiple antigens [69]. In this way, it is possible to overcome the antigen mutation or deletion by tumor cells, the variation or absence of the appropriate T cell repertoire and the MHC haplotype in patients [69].

When designing a poly-epitope DNA vaccine, many parameters should be considered. First, the competition for antigen recognition at the surface of the APC and the affinity of the selected epitopes for MHC molecules should be considered [70, 71]. Palmowski et al. demonstrated that the use of an MHC class I polyepitope vaccine leads to the preferential expansion of CTLs with a single immunodominant specificity [72, 73]. In addition, the affinity of the selected epitopes for MHC molecules and transporters could influence the CTL immunodominance and the consequent immune response [70].

Second, although the CD8 T cell response has been considered to be the main protagonist in the antitumor immune response resulting from vaccination, the insertion of an epitope/antigen recognized by CD4 T cells into a DNA vaccine could activate a broader and stronger immune response. Several studies suggest the importance of the CD4 T cell population for cancer immunotherapy [74, 75]. Recently, it has been demonstrated that CD4 T cells recognize a higher number of neoantigens than previously known and can generate potent antitumor responses [62, 76]. Hence, a coordinated CD4 and CD8 response is necessary for the complete eradication of a tumor [76]. T helper (Th) peptides have already been used in combination with DNA vaccines to increase the activation of Th cells, thus further eliciting the CTL immune response [77‐82]. An example of a Th epitope is the pan DR epitope (PADRE). This synthetic Th epitope, encoded in a DNA vaccine and administered with an antigen-encoding plasmid, increased the number of antigen-specific CD8 T cells, resulting in potent protective and therapeutic antitumor effects [83]. Other studies demonstrated that a PADRE-encoding DNA generated CD4 Th1 cells that play an important role in maintaining long-term memory responses, helping the activity of CD8 T cells [84].

Many techniques have been developed to find new epitopes. These studies led to the identification of NY-ESO-1, MelanA/MART-1, SSX4, MELOE-1 and TRAG-3 in melanoma, EphA2 and MAGE-6 in renal cell carcinoma, CEA, MAGE-3 and telomerase in lung carcinoma, TRAG-3 in breast carcinoma, and NY-ESO-1, p53 and SSX4 in ovarian cancer, among others [85]. Some of these tumor antigens recognized by CD4 T cells belong to the same categories as those recognized by cytotoxic CD8 T cells [75].

Finally, it is important to identify the most immunogenic epitopes derived from tumor antigens. New in silico techniques are being developed to improve the prediction of epitope immunogenicity to design a poly-epitope vaccine. They not only consider the binding affinity to the MHC and the different HLA subtypes but also the conformation and interaction with the HLA, immunodominance vs tolerance, etc. [86]

Many recent preclinical studies have investigated the use of polyepitope DNA vaccines to reach a broad immune response. As a result, an increased IFNg production, a higher Th and CTL response [86, 87], and a general decrease in the tumor growth rate and metastasis formation were observed in different types of cancer models [88, 89]. Some preclinical studies focus on the HPV model, using DNA vaccines encoding E6 and E7 molecules [90], or E7 with a helper epitope [88]. Another example is SCT-KDR2, which encodes the mouse β2microglobulin + KDR2 (VEGFR2 antigen peptide) + MHC class I H-2D^b, in a B16 melanoma tumor model [89]. A non-exhaustive list of the most recent preclinical trials (in the last 5 years) can be found in Table 3. Additionally, many clinical trials are testing the safety and efficacy of polyepitope DNA vaccines, such as NCT02348320 and NCT02157051 for breast cancer, NCT02172911 for cervical cancer, and NCT01322802 and NCT03029611 for ovarian cancer. In particular, in the clinical studies NCT02348320 and NCT03199040, a personalized polyepitope vaccine against breast cancer is being used, as well as in the NCT03122106 for pancreatic cancer, and the results will help to establish the relevance of this vaccine strategy. This would address tumor heterogeneity and the loss of immunogenicity associated with TAAs, which accounts for the failure of the current anticancer treatments [58]. A complete list of the ongoing clinical trials could be found in Table 2.

A good option to further optimize the efficacy of cancer DNA vaccination could be the combination of the 3 cited approaches, designing a poly-epitope chimeric vaccine containing specific neoantigens. In the clinic, this could reduce the number of nonresponding patients by developing a stronger and more complete immune response.

Immunostimulatory cytokines can increase the effect of the vaccine on effector T cells. They are generally encoded by the antigen-encoding vaccine, by another plasmid or injected as proteins in combination with the vaccine. In recent studies, the most commonly used cytokines include IL-2, IL-12 and GM-CSF. IL-2 is involved in the differentiation of immature T cells into both Tregs and effector T cells. Its great efficacy against metastatic melanoma and metastatic renal cell carcinoma led to its approval by the FDA [7, 93]. IL-12 is another important cytokine involved in T cell activation and effector function, and its combination with a vaccine increases the vaccine’s efficacy [94]. A plasmid encoding IL-12 combined with a DNA vaccine against cervical cancer promoted mouse survival and decreased the number of MDSCs in the TME [95]. GM-CSF is used in many clinical trials (Table 2) for its activity on DC maturation and T cell activation and proliferation. However, this molecule can also attract MDSCs, and it is not clear how this cytokine balances between immune activation and inhibition in vivo. Current clinical studies are seeking to answer this question [7]. Other cytokines could be used in combination with DNA vaccines, e.g., INFγ, IL-15, and IL-7 [7, 9].

The combination with other types of adjuvants could also be tested, such as TLR-activators. Recently, we demonstrated that the insertion of some CpG immunostimulatory motifs inside the antigen gene sequence through codon optimization could enhance cytokine production, thus increasing the efficacy of a DNA vaccine against P815 mastocytoma [37].

The signaling mechanism mediated by costimulatory/inhibitory molecules plays an important role in T cell-mediated immunity. Many cells in the TME can express ligands for inhibitory receptors on T cells, leading to their inactivation [96]. Inhibitory receptors include CTLA-4, PD-1, TIM-3, LAG-3, etc. [96] In several studies, the in vivo blockade of CTLA-4 delayed tumor growth in animal models and resulted in tumor rejection in patients affected by melanoma [97, 98]. This effect was mainly due to the inhibition of TGF-β- and IL-10-secreting Tregs and an increased T effector cell activation [99]. Interestingly, this also resulted in immunity against the secondary exposure to tumor cells, suggesting that the memory component of the immune response can be evoked by anti CTLA-4 antibodies [100]. Antagonist antibodies that target PD-1 and its ligand PD-L1 have also achieved impressive and durable results in many solid tumors, leading to their FDA approval for different cancer types [7]. Recently, a relationship between ICB administration and the neoantigen burden has been demonstrated [101]. Snyder et al. sequenced 64 patients with advanced melanoma and showed that the somatic mutation burden was strongly associated with the clinical response to anti-CTLA4 [102]. Similarly, Rizvi et al. demonstrated that the mutation burden was a strong predictor of clinical response in non-small cell lung cancer (NSCLC) patients treated with anti-PD1 therapy, and that this therapy enhances neoantigen-specific T cell reactivity [103]. The higher prevalence of somatic mutations in cancer cell genomes was a common feature among cancers with a higher probability of responding to ICB. Thereafter, the link between the mutation burden and the clinical benefit following ICB immunotherapy was validated multiple times and in multiple tumor types [63]. This is related to the concept that with an increased tumor mutation burden, the probability of a cognate T cell clonally expanding against a specific tumor antigen will increase. In other words, high tumor mutation burden tumors often have more neoantigens that could be recognized by the processes involved in antitumor immunity, making such cancers more likely to respond to ICB therapy [59, 104].

To increase T cell activity in the TME and to broaden the number of patients responding to ICB, combinations of ICB with different strategies were tested for a variety of malignancies in preclinical and clinical studies [96]. Some examples include combination with radiation therapy [105], other antibodies [106], photodynamic therapy [107], and cancer vaccines [108]. Combination with cancer DNA vaccination seems to be promising in coupling the benefits of ICB with the ability of vaccines to prime the antigen-specific CTL response [88, 109]. A potent cancer vaccine that induces a T cell response against tumor-specific antigens could also increase the number of responders to ICB [63]. However, only a minority of patients respond to ICB therapy, suggesting the need for a rational use of ICB based on biomarkers predictive of the immune response to avoid nonresponsiveness to therapy and undesired side effects [110, 111].

In the last few years, it has been reported that anticancer chemotherapy can play a double role in tumor eradication. Many chemotherapeutic drugs, such as gemcitabine [112], paclitaxel [113], cyclophosphamide [114] and others, applied in ultralow (metronomic) noncytotoxic doses, not only target tumor cells inducing TA release but also enhance T cell infiltration/activity in the TME and remove immunosuppressive cells. In a preclinical study, the combination of cyclophosphamide with DNA vaccines enhanced mouse survival and decreased the expression of immunosuppressive cytokines, such as IL-10 and VEGF [115]. Based on preclinical and clinical studies, the combination of the appropriate chemotherapeutic drug and vaccine therapy may play a substantial role in future cancer treatments, especially when patients do not respond to ICB [116]. Indeed, it has been demonstrated that treatment with chemotherapy restored sensitivity to checkpoint blockade through TLR4 simulation [116]. Further clinical studies are necessary to better define the optimal agents and schedule of administration.

DNA vaccines could also be combined with targeted therapies that are able to mediate tumor cell antigen release and enhance T cell priming. Sunitinib, a multitargeted receptor tyrosine kinase inhibitor, was found to decrease Tregs and MDSCs and increase INFγ-producing T cells in renal cell carcinoma patients [117]. The combination of sunitinib with a viral vaccine encoding CEA decreased the tumor volume in a mouse model [118]. Although not already tested with DNA vaccines, other tyrosine kinase inhibitors already approved by FDA, such as pazopanib, axitinib, and cabozantinib, could improve the patient response to vaccination.

Other strategies that can be used in combination with DNA vaccines include endocrine therapy and radiotherapy (RT).

In hormonally driven tumors such as prostate cancer and breast cancer, endocrine therapy is part of the standard of care, and the effect of letrozole in decreasing the Tregs in the TME has already been demonstrated [119]. Furthermore, androgen deprivation in prostate cancer induces thymic regeneration and increases the number of effector T cells [7]. In an ongoing clinical trial (NCT02204098, phase I), the effect of Mam-A vaccine administration in combination with anastrozole, letrozole, tamoxifen, exemestane, and goserelin is under investigation.

Preclinical data have demonstrated the additive effect of RT and vaccines with an enhanced destruction of tumor cells, the release of TAs, an increase in IFNg production, and a global decrease of the tumor volume. T cells specific for other antigens not included in the vaccine were also generated [120, 121]. Since RT is a part of the standard of care, many trials using vaccines after radiation try to evaluate the mutual effects from the 2 therapies (Table 2).

In Fig. 3, the main mechanisms of action of therapies postulated to mediate synergistic effects in combination with DNA vaccines are shown.

The first question concerns the antigen type: TAAs or neoantigens? Relevant TAAs have been identified for most tumors, but immune tolerance can limit their efficiency. On the other hand, neoantigen identification is time consuming and expensive, and neoantigens do not reflect the tumor heterogeneity in the individual patient (e.g., in metastasis). Second, how many antigens should be encoded in the same plasmid? It is still not clear whether one antigen is superior to another in terms of frequency of immune response or clinical effect [133]. In one study, it has been shown that neoantigens with a predicted high affinity are more immunogenic and that a poly-specific and poly-functional DNA vaccine encoding neoantigens was the most efficient solution to prevent tumor growth in mice [134]. Further studies are needed to generalize these findings. Most of the clinical trials use a relevant antigen for the vaccination (e.g., PAP, E6/E7), but none compare the combination of this antigen with others or with neoantigens to evaluate their effect on immune activation. This point arouses a last question: is it worth mixing TAAs and neoantigens in the same vaccine? Furthermore, should the different antigens be administered in the same plasmid or is it better to encode them in different plasmids administered together? To our knowledge, none of the clinical studies address this question, which could be a further area of investigation in the future.

An important parameter in the combinatorial approach is the choice of the right therapies to be combined. This aspect depends on the tumor type, its presence, and the possibility of detecting patient-specific biomarkers, among others. In addition, the administration schedule of multiple therapies is crucial. Until now, vaccines have been used as a last-line therapy. To integrate them as a standard of care therapy, it is crucial to know when to administer them. This decision should consider the time for the immune system to generate a specific immune response against the delivered antigen, the need for multiple doses of administration and the interaction with the combined therapy. For example, some studies revealed that using ICB after the peptide/RNA vaccine treatment induced a sustained remission with no sign of disease recurrence [66]. However, this should be adapted to the specific therapy and patient. Furthermore, the doses should be consequently adapted to the combination regimen.

Most of the critical points aroused in cancer DNA vaccination can ultimately be addressed only in clinical practice because translation from animals is extremely difficult. This is due to the different tumor characteristics and the differences in the immune system between humans and animals [135]. To try to overcome this problem, many orthotopic injection models, genetically engineered mice, xenograph and humanized models have been developed. However, they all fail to recapitulate the chaotic way in which malignant transformation occurs during cancer development in human patients. Mouse models provide valuable insight into the mechanisms of action and provide important proof of concept for human studies, but there remains a need for larger animal models encompassing a fully competent immune system. Some researchers suggest the use of canine and porcine models, especially for skin cancers [135]. However, housing, ethical regulation, and breeding difficulties limit the use of big animal models. Furthermore, even these models have limitations, and the idea of a universal model for oncoimmunology currently seems unrealistic.

Immunotherapies have a variable response rate from one patient to another and are often associated with side effects. For this reason, it is important to identify biomarkers that could predict the patient response to a specific therapy and to standardize the therapy according to the predicted biomarker. Furthermore, biomarkers may be useful for monitoring treatment response. However, the translation of biological data into predictive or prognostic biomarkers is complicated by the complex interactions between tumors and the immune system and by host and tumor variability. Many studies are using bioinformatics tools and new genomic and proteomic technologies to predict specific tumor signatures, generating complex datasets that give rise to analytical challenges. Currently, we can rely on imperfect biomarkers, such as PD-L1 expression in the tumor or the tumor mutation burden. This gap in knowledge leaves room for further studies that will help treatment selection and design the best combination therapy for each patient.