2.1. PD-L1 & PD-1

PD-1 is expressed on activated T cells while PD-L1 is expressed primarily on antigen presenting cells (APCs) (such as dendritic cells, DC). However, it is also over expressed in some cancers as an immune escape mechanism. PD-1, PD-L1 interactions serve to dampen the T cell response resulting in T cell anergy [19]. The use of siRNA to silence PD-L1 is the most widely reported nucleic acid based immune checkpoint targeting immunotherapy.

In early studies it was demonstrated that a relatively simple formulation of siPD-L1 combined with cationic transfection reagent (polyethylenimine, PEI) could result in T cell expansion and a high degree of survival following tumor challenge [20]. Furthermore, that the combination of non-specific siRNA and PEI displayed potent antitumor effects via DC activation and stimulation of multiple toll like receptors (TLRs) [20]. However, this effect was not as pronounced as the siPD-L1 construct which was able to induce tumor-specific memory CD8⁺ T cells. Since, PEI has been utilised in many siPD-L1 formulations and the backbone has been improved upon through the addition of targeting moieties such as folic acid and dermatan sulphate [21], [22], [23]. A study by Li et al. likewise demonstrated that PEI formulated siPD-L1 could prolong survival in a lung cancer model, however, they also showed that a rationally formulated albumin based nanocomplex (FX/HP) comprising paclitaxel, a fluorinated CXCR4 antagonist and siPD-L1 were significantly superior [24]. In addition to delivering siRNA, the FX/HP nanoparticles were immunogenically active, able to induce DC maturation and antagonise CXCR4 thus improving T cell infiltration [24]. This is particularly useful as tumors may be resistant to PD-L1 blockade as the fibrotic tumor microenvironment prevents T cell infiltration.

This study highlights the potential of siPD-L1 and carrier to be formulated to obtain synergistic or additive effects which would otherwise be hard to achieve with a traditional antibody-based formulation. Combination of siPD-L1 with chemotherapy has been attempted by several groups. The particulate nature of the siRNA carriers lends itself to co-formulation with drugs, as many of the carriers were originally developed, or have been extensively used, in the drug delivery field. A good example of this can be seen in co-formulation of siPD-L1 with anthracycline drug doxorubicin (Dox). A liposomal formulation of Dox (Doxil) is currently utilised clinically and therefore represents a suitable candidate for particulate formulation [25]. In addition, Dox has also been demonstrated to induce immunogenic cell death (ICD), a form of apoptosis resulting in the release of immunostimulatory factors (including acellular ATP and HMGB1) and translocation of calreticulin to the cell surface, therefore co-formulation with siPD-L1 may be logically justified [26]. A number of recent studies have formulated Dox with siPD-L1 utilising polymeric or lipid polymer hybrid systems [27], [28], [29], [30], [31]. In each case it has been demonstrated that the combination of Dox with siPD-L1 results in reduced tumor growth compared to monoformulated drug. Notably, Wang et al. formulated siPD-L1 with Dox in a self-assembling lipid polymer hybrid nanoparticle and demonstrated a 30% clearance of tumors following treatment [28]. To improve tumor targeting of siPD-L1/Dox formulations, particles have been developed which are sensitive to various aspects of the tumor microenviroment including pH, reactive oxygen species or reduction [30,31]. Particles may also be targeted via surface ligands such as T7 peptide which bind to the transferrin receptor overexpressed on tumor cells [31].

Drug/siPD-L1 synergy is, however, not limited to ICD inducers, for example Yoo et al. combined siPD-L1 with gemcitabine, a drug not typically associated with ICD, in a magnetic nanoparticle system [32]. In a pancreatic cancer model, the drug combination treatment reduced cancer cell proliferation, leading to decreased tumor growth and increased survival rates, this was particularly apparent when using high doses, with 67% of mice surviving to the end of the study (all control mice were euthanised on week 6). The prognosis for pancreatic cancer is extremely poor with a 1% 5-year survival rate and tumors are typically refractory to conventional immune checkpoint blockade [32]. Novel, rationally formulated therapies such as nucleic acid-based immune checkpoint blockade may provide some hope in the development of future therapeutics. In a similar vein, Li et al. combined siPD-L1 with imatinib in a liposomal system, this formulation significantly downregulated PD-L1 expression which in turn was linked to tumorigenic mTOR pathway inhibition both in vitro and in vivo [33]. In their model, PD-L1 silencing sensitised cancer cells to imatinib, resulting in higher apoptosis and therefore decreased tumor volume and growth rate. Immunogenically, the combination of imatinib cytotoxicity, mTOR pathway inhibition and PD-L1 silencing synergistically enhanced IFN-γ production [33]. The interplay between PD-L1 silencing and mTOR autophagy has been utilised in a number of studies [34].

In addtion to siPD-L1 drug combinations, siRNA may also be combined with other nucleic acid constructs such as other siRNAs, plasmid DNA (pDNA) or mRNA [35], [36], [37]. In one notable example, siPD-L1 was combined with a pDNA construct expressing IL-2 in a single lipid-dendrimer-calcium phosphate formulation [37]. IL-2 promotes T cell proliferation and enhances effector T cell activity, but requires repetitive administration which can cause negative side effects [37]. In this study, the combination of the two nucleic acid constructs enhanced antitumor responses through increased cytotoxic T cell proliferation and CD8⁺ T cell infiltration resulting in reduced tumor growth and increased survival rates [37]. It is worth considering that siRNA and pDNA have distinct spatial requirements, the siRNA being active in the cytosol and the pDNA needing to reach the nucleus. The developed formulation looks to be a promising candidate for potent co-delivery of pDNA and siRNA, though the exact mechanisms and intracellular trafficing of pDNA/siRNA molecules was not established. In contrast to pDNA, mRNA is active in the cytosol and therefore may represent a more logical choice for co-formulation with siRNA. Indeed, the inclusion of siPD-L1 into an mRNA cancer vaccine delivered via lipid coated calcium phosphate nanoparticles was shown to reduce tumor growth and enhance IFNγ responses [35]. Interestingly, this study demonstrated the inclusion of siPD-L1 was more effective than the co-delivery of anti PD-L1 mAbs [35]. While there have been few studies comparing siRNA to mAb, those which have typically reported siPD-L1 to be more effective, whether this is reproduced clinically has yet to be demonstrated [20].

In contrast to its ligand, the silencing of PD-1 as a therapeutic intervention has been relatively understudied. This may be because there are additional difficulties targeting PD-1 as it is expressed on T cells which are generally considered a challenging target to transfect with non-viral vectors [38]. However, there are an increasing number of particulate formulations which have been developed to transfect T cells, including lipid and polymeric systems [39], [40], [41]. In many cases these technologies have been developed for the in situ delivery of chimeric antigen receptor expressing constructs (reviewed [42]). Platforms using anti CD3/CD4 antibodies or ScFv as a targeting moiety have achieved high in vivo transfection however this may undermine the aims of nucleic acid mediated immune checkpoint blockade as a means of reducing the use of monoclonal antibodies [43,44]. The use of aptamers could resolve this issue, indeed anti CTLA4 and anti 4-1BB aptamers fused to siRNA constructs have been used to target T cells in vivo [45,46]. These chimeric constructs may represent the future of nucleic acid immune checkpoint blockade (reviewed [47]). Alternately, platforms have been developed based on ‘constrained lipid nanoparticles’ which have demonstrated the ability to delivery siRNA to T cells in vivo [48].

The use of cells ex vivo affords the opportunity to study the relative contribution of PD-1/PD-L1 and their potential as targets for nucleic acid-based checkpoint blockade. In a candidate cell-based DC vaccine it was observed that PD-L1 silencing in DCs increased T cell proliferation and DC activation [6]. The PD-1 silencing on T cells proved even more influential in improving T cell priming, and skewing towards a Th1 phenotype, but the greatest efficacy in T cell proliferation was observed using dual silencing of both PD-1 and PD-L1 [6]. In accordance with this, it was found that silencing either PD-1 or PD-L1 increased killing of tumor cells, however co-silencing resulted in the most potent cytotoxicity [49]. In in vivo models, Kwak et al. developed a PLGA based system to deliver both siPD-L1 and siPD-1 [34]. In keeping with the in vitro observations, a reduced tumor growth rate was observed in both the monoformulated and co-formulated siRNAs, however the co-formulation of siPD-1 and siPD-L1 was shown to be superior with tumor growth comparable to anti PD-L1 mAb [34].

In summary, studies most commonly suppressed PD-L1 as opposed to PD-1. Combinatory treatment with other drugs or dual suppression of other checkpoint targets produced enhanced antitumor effects. However, difficulty in T cell transfection highlights the importance of a suitable carrier.

2.2. Cytotoxic T-lymphocyte-associated protein 4 (CTLA4)

CTLA4 was amongst the earliest immune checkpoints to be described, and it is expressed on activated T cells and Tregs. During the interaction of a T cell with an APC, CTLA4 competes with CD28 for CD80/CD86 with a higher affinity, the lack of ‘signal 2’ co-stimulation provided by this engagement results in T cell anergy [50].

Similarly to siPD-1, there are few studies on siCTLA4 potentially due to difficulty in transfecting T cells in vivo [38]. However, Li et al. developed PEG-PLA nanoparticles to deliver siCTLA4 to T cells [51]. Successful transfection was demonstrated in vitro and in vivo and, following treatment, reduced tumor volume was observed. This was marked by an increased T cell tumor infiltration and a dose-dependent increase in serum IFNγ [51]. As both PD-1 and CTLA4 are expressed by T cells, the logical combination of siPD-1 and siCTLA4 was performed by Liang et al. Although each siRNA construct used in isolation resulted in reduced tumor weight and volume, the largest decrease was observed using co-inhibition [52]. Co-inhibition also resulted in increased IFN-γ secretion and decreased IL-10, IL-6 and survivin [52].

2.3. Ecto-5’-nucleotidase (CD73) and Ectonucleoside triphosphate diphosphohydrolase-1 (CD39)

CD73 is an enzyme expressed on the surface of a number of tissues; in cancer it has been associated with increased cell proliferation, neovascularisation and tumor invasiveness [53]. As a molecule (in concert with CD39), it is responsible for the breakdown of immunostimulatory ATP to immunosuppressive adenosine [54].

There has been particular interest in silencing CD73 for the treatment of glioblastoma. It has been shown that following treatment with siCD73, there was reduced cell proliferation and glioblastoma cells cultured in the presence of adenosine had increased viability [55]. Therefore to silence CD73, Azambuja et al. developed cationic nanoemulsions to carry siRNA, which, upon treatment of cells, produced a large decrease in AMPase activity and cell viability [56,57]. To target glioblastoma, the optimised formulation was delivered intra-nasally to capitalise on the nose to brain-pathway. In this study it was found that nanoparticles were present in the brain six hours following adminstration and peaked at 18h before returning to undetectable levels by 32h [57]. When used in a therapy study, they observed a 60% tumor volume reduction with the siCD73 formulation with no shrinkage being observed in control goups [57].

In addition to glioblastoma, siCD73 has been trialled in breast cancer models, in a series of studies Jadidi-Niaragh et al. developed chitosan lactate nanoparticles which were safe, highly stable and were efficient in transfecting 4T1 triple negative breast cancer cells [58]. The base chitosan particles have subsequently been improved by the addition of a magnetic SPION core, cell penetrating peptide (TAT), drug co-formulation and/or a folate targeting moiety [59,60]. When tested in vivo, Ghalamfarsa et al. observed that silencing of CD73 resulted in decreased blood vessel formation and a reduction in angiogenic factors e.g. VEGF, which have been linked to hypoxic conditions within the tumor [59]. These treatments resulted in reduced tumor growth, cancer cell migration and colony formation [59]. Most interestingly, the anti-angiogenic effects of CD73 silencing were enhanced with co-silencing of HIF-1α which further reduces CD73 expression by reducing binding to certain cells within the hypoxic TME [59]. Together these studies illustrate the potential of siRNA when used in rationally formulated combinatory regimes.

In the only study utilising an ASO to knock down an immune checkpoint we identified, CD39 was selected as the target [61]. In this study it was reported that ASO CD39 treatment significantly increased the ratio of CD8 to Tregs, and that combination of ASO CD39 with anti PD-1 mAb resulted in reduced tumor volume [61].

2.4. CD47

In contrast to the other molecules described in this review which focus on engaging T cells, the exploitation of the CD47 axis primarily relies on the activity of macrophages. Though macrophage polarisation and activation will also result in augmented T cell responses. CD47 is ubiquitously expressed on several cell types but is over expressed on the cancer cell surface, it engages with SIRPα expressed on phagocytes to deliver a ‘don't eat me’ signal which prevents cell clearance through phagocytosis. As the effect of CD47 is due to macrophages, it is possible to speculate that siCD47 therapy may be most suited to cancers of organs with high resident macrophage populations.

To deliver siCD47, Wang et al. developed liposome-protamine-hyaluronic acid nanoparticles and treatment resulted in increased phagocytosis by macrophages, leading to reduced formation and proliferation of lung metastases [62]. A striking 93% tumor reduction was attributed to macrophages, as macrophage depletion by liposomal clondronate significantly ablated this therapeutic effect [62]. To further target siCD47 to the tumor bed, Wu et al. used a glutamine modified polyplex to deliver the siRNA. Cancer cells require high levels of glutamine to support their mitosis, therefore nanoparticles accumulate due to the ‘glutamine trap’ effect [63]. As a measure of this, the glutamine polyplex uptake was low in healthy cells but was triggered by glutamine deprivation as a result of cancer cells depleting the local glutamine supply. Tumor growth significantly decreased following siCD47 delivery, but more so when administered alongside the chemotherapy drug cisplatin, indicating suitability for combination treatment [63].

There have been formulations devised to deliver siCD47 whilst also enhancing T cell responses [36,64]. Indeed, siCD47 has been delivered alongside siPD-L1 in an EpCAM-targeted liposomal nanoparticle. Co-silencing of siCD47 and siPD-L1 increased IFN-γ production and, although single silencing was sufficient to reduce tumor growth, the most significant tumor growth inhibition was seen with the dual silencing nanoparticle [36]. The relative contributions of macrophages and T cells was not established though T cell numbers were elevated in formulations containing siPD-L1. To increase the numbers of CD8 T cells in the tumor Chen et al. developed a core shell-corona nanoparticle capable of first releasing chemokine ligand 25 (CCL25) then transfecting siCD47 [64]. CCL25 serves to inhibit CD4⁺ T cell differentiation into Treg cells and promotes survival of CD8⁺ T cells [64]. The co-delivery had a range of effects including increased tumor infilitrating lymphocytes and altering CD8⁺/CD4⁺ T cell ratios and CD8⁺/Treg ratios [64]. Furthermore, tumor growth was delayed, and the size and number of metastases decreased. siCD47 monotherapy only slightly inhibited tumor growth and CCL25 monotherapy had no effect on tumor growth [64]. Interestingly, the antitumor effects were inhibited in CD8⁺ depleted mice, demonstrating that antitumor efficacy was CD8⁺ T cell dependent [64].

3.1. OX40/OX40L

OX40 is expressed on activated T cells whilst its ligand (OX40L) is expressed on antigen presenting cells. Engagement of OX40 with OX40L serves to prolong the survival of T cells and prevent the differentiation of CD4 T cells into Tregs [76]. It has been demonstrated that cancer cells transfected with OX40L either do not establish tumors or established very slow growing tumors [77].

When delivered as mRNA in a charge-altering releasable transporter, OX40L has been shown to be extremely potent, mice were able to eradicate 100% of tumors following intratumoral mOX40L administration, the growth of distal, non-treated tumors, was also reduced. In the same study, mOX40L was tested with mIL-12, mCD80 and mCD86 [78]. In these studies it was shown that therapy with mIL-12, mCD80 and mCD86 stimulation cleared almost all tumors with significant growth delay in distal tumors. The triple combination of mOX40L/mCD80/mCD86 demonstrated the highest efficacy in tumor regression and survival, followed by mOX40L/mIL-12, whilst mIFN-γ had no effect. The mOX40L/mCD80/mCD86 combinations led to natural killer, CD4⁺ and CD8⁺ T cell activation in draining lymph nodes, local tumors and distal tumors [78]. This was confirmed by upregulation of activation, cytotoxicity and proliferation markers. Foxp3 and CTLA4 were downregulated, further contributing to a pro-inflammatory shift in the tumor microenvironment [78]. Combinations of mOX40L with mCD70, another member of the TNF family involved in T cell stimulation, were demonstrated to have no effect [78].

In addition to co-stimulatoy molecules, the combination of mOX40L with cytokines mIL-23 and mIL-36γ was investigated by Hewitt et al. in a lipid nanoparticle [79]. Delivery of mOX40L resulted in total tumor regression in 50% of mice, with H22 tumors. Dual treatment with either mIL-23 or mIL-36γ prolonged survival which was further improved using triplet treatment [79]. Surprisingly, the delivery of the mRNA constructs was more potent than delivery of the recombinant proteins and resulted in >70% survival.

Immunologically, the triplet treatment increased Th1-related molecules and upregulated certain genes, indicating the cytokines IL-23 and IL-36γ were responsible for greater transcriptional changes [79]. IL-36γ was also responsible for increased CCL7 and IL-1α secretion, a later wave of IL-22, IFN-γ, TNF-α and IL-1β secretion and for the activation and proliferation of many lymphocyte types both as a single and combination treatment. Interestingly, tumors that were unresponsive to other immune checkpoint blockade treatments such as anti-PD-L1 or anti-CTLA4, were significantly reduced in size when triplet treatment was administered, results were further enhanced when used in combination with other immune checkpoint targets [79].

Combined, these studies reveal OX40L to be a highly promising candidate for co-stimulatory nucleic acid-based immune checkpoint blockade. Future work may comprise of formulating mOX40L with other drugs or immune interventions. The move away from intratumoral (I.T) delivery would also be clinically preferable and maybe achieved through the use of a targeted nanocarrier.