Responses to homocitrullinated peptides can be stimulated in mice and are CD4 mediated

Proteins containing homocitrulline modifications are known to be immunogenic. We have previously shown that homocitrullinated (Hcit) peptides derived from aldolase A (Aldo), immunoglobulin heavy chain-binding protein (BiP), cytokeratin 8 (Cyk8) and vimentin (Vim) can stimulate CD4 T cell responses through multiple MHCII alleles in standard and HLA transgenic mice including HLA-HHDII/DR1, HLA-HHDII/DP4, HLA-DR4 and Balb/c mice. We have also observed that among several peptides screened, Aldo74-93Hcit, BiP562-579Hcit, Cyk8 371-388Hcit and Vim116-135Hcit peptides induced T cell responses in the different mouse strains mentioned above5,11. These epitopes are homologous in humans and mice. In this study we first aimed to confirm whether vaccination of Aldo74-93Hcit, BiP562-579Hcit, Cyk8 371-388Hcit and Vim116-135Hcit in combination (known as Modi-2) mixed with CpG and MPLA adjuvants would induce T cell responses in different mice strains. A schematic of the immunisation regime is shown in Fig. 1A. HLA-HHDII/DP4 and Balb/c mice immunised with the Modi-2 peptide combination were screened for peptide specific responses in IFNγ ELISpot assay. This screening showed significant responses to Aldo74-93Hcit (p = 0.0005), Bip562-579Hcit (p < 0.0001) and Vim116-135Hcit (p = 0.0472) in HLA-HHDII-DP4 mice (Fig. 1B). However no immune responses to Cyk8 371-388Hcit were observed in HLA-HHDII-DP4 mice. Balb/c mice induced significant immune responses to Aldo74-93Hcit (p = 0.0042), Cyk8 371-388Hcit (p = 0.0021) and Vim116-135Hcit (p = 0.0014). However, no immune responses to Bip562-579Hcit were identified in Balb/c mice (Fig. 1E). Mice immunised with the CpG and MPLA adjuvants alone, irrelevant peptide in CpG and MPLA or naïve mice failed to show any reactivity to the Hcit peptides (Supplementary Fig. 1).

Fig. 1: Modi-2 peptides elicit Hcit specific CD4 mediated responses in mice.

Schematic of immunisation regime (A). T cell responses to the Modi-2 peptides and counterpart wild type peptides in HLA-HHDII/DP4 (B–D) and Balb/c mice (E–H). Mice were vaccinated with three doses of homocitrullinated peptides mixed with the adjuvants CpG and MPLA and responses assessed against Hcit peptides (B, E), Hcit and wt peptides (C, F) or Hcit peptides in presence of CD4 or CD8 blocking antibodies (D, G) by IFNγ ELISpot assay. Representative ELISpot images (H). n ≥ 3 in each study and data is collated from at least two independent studies. Mean, standard deviations and significant p values are shown. Image in (A) created with BioRender.com.

To confirm whether responses to the Modi-2 peptides were modification specific without cross reactivity to the wild type (wt) peptides, responses to Aldo74-93wt, Bip562-579wt, Cyk8 371-388 wt and Vim116-135wt peptides were assessed in HLA-HHDII/DP4 and Balb/c mice. IFNγ responses to Aldo74-93Hcit, Bip562-579Hcit and Vim116-135Hcit were significantly higher than the responses to Aldo74-93wt, Bip562-579wt and Vim116-135wt, p < 0.0001, p < 0.0001 and p = 0.0286 respectively, showing minimal cross reactivity to the wt peptides in HLA-HHDII/DP4 mice (Fig. 1C). Similar findings were observed in Balb/c mice where IFNγ responses to Aldo74-93Hcit, Cyk8 371-388Hcit and Vim116-135Hcit were significantly higher than the responses to Aldo74-93wt, Cyk8 371-388 wt and Vim116-135wt, p = 0.0030, p = 0.0130 and p = 0.0011 respectively, with minimal cross reactivity to the wt peptides (Fig. 1F). Vaccination with Modi-2 peptides, that are 18–20 amino acids in length, may stimulate both CD4 and CD8 IFNγ responses. Therefore, we further investigated whether responses to the combination of Modi-2 peptides were mediated by CD4 or CD8 T cells. In HLA-HHDII/DP4 mice, blocking CD4 T cells in the ELISpot assay significantly reduced IFNγ responses to Aldo74-93Hcit (p = 0.0322) and Bip562-579Hcit (p = 0.0466) but blocking CD8 T cells had no significant effect on responses to those peptides (Fig. 1D). Responses to Vim116-135Hcit were reduced by the anti-CD4 blocking antibody but this did not reach significance and were also partially blocked by the anti-CD8 blocking antibody in HLA-HHDII/DP4 mice (Fig. 1D). In Balb/c mice, IFNγ responses to Aldo74-93Hcit, Cyk8 371-388Hcit and Vim116-135Hcit were significantly reduced after blocking CD4 T cells, while blocking CD8 T cells had no significant impact on the IFNγ responses (Fig. 1G). Example ELISpot images are shown in Fig. 1H.

Together these results confirm that vaccination with the Modi-2 peptides can stimulate IFNγ responses in transgenic mice expressing human HLA-DP4 and Balb/c mice which express murine H-2d haplotype. However, responses vary between different mice strains. These results also support previous data5,10 and confirm that responses to the Modi-2 peptides are homocitrulline-specific and mediated by CD4 T cells.

Modi-2 peptides can be formulated with SNAPvax technology to enable codelivery with immunomodulatory adjuvant and facilitate manufacturing

Peptide based vaccines require delivery with an immunostimulatory adjuvant that enhances the magnitude and quality of immune responses. In addition, manufacturing peptide-based vaccines provides many challenges. Most immunogenic peptides are hydrophobic in nature and therefore they have high aggregation potential and low solubility in aqueous or organic solvents. These factors make peptide-based vaccines difficult to manufacture to clinical GMP scale. The mixture of Modi-2 peptides with CpG/MPLA adjuvant does indeed show very large particle sizes with an aggregation of particulate matter which is highly polydisperse (Fig. 2A and Supplementary Table 6). Similar moderate size particles are seen for the Modi-2 peptides mixed with TLR7/8 adjuvant (Supplementary Fig. 2 and Supplementary Table 6). These large particle sizes result in significant material loss upon sterile filtration through a 0.2 µm filter (Supplementary Table 7).

Fig. 2: Formulation of Modi-2 peptides with SNAPvax technology.

A DLS histograms showing the size of particles of Modi-2 peptides in CpG/MPLA or TLR7/8 formulation. Each colour represents a different repeat measurement. B A schematic figure of the SNAPvax (Self-assembling Nanoparticles based on Amphiphilic Peptides) platform. C DLS histograms showing the size of nanoparticles of Modi-2 SNAPvax formulations in different buffers. The particle size is shown in nanometer22 ± standard deviation (SD). Each colour represents a different repeat measurement. At least 3 replicate assays were performed. D Transmission electron micrograph images of Modi-2 SNAPvax formulations in Histidine or Tris buffers. E UPLC-PDA chromatogram showing distinct peaks for each of Modi-2 peptides in the SNAPvax formulation. Images in (B) from Barinthus Biotherapeutics.

Here we describe the use of the SNAPvax technology (Barinthus Biotherapeutics plc) which incorporates peptide antigens in amphiphiles linked to combined TLR-7 & -8 (TLR-7/8) ligands that self-assemble into ~20 nm nanoparticle micelles (Fig. 2B)21. Covalently linking peptide antigens and TLR-7/8 ligands ensures codelivery to the same APC, while use of amphiphiles that self-assemble to uniformly small nanoparticles ( ~ 20 nm) well below the size cut-off for sterile filters ( ~ 200 nm) improves manufacturability by providing both improved solubility in aqueous buffers and compatibility with sterile filtration. We first assessed the impact of the buffer on the size of nanoparticles formed with the Modi-2 SNAPvax formulations. Modi-2 SNAPvax was formulated in different aqueous buffers including PBS, Histidine (His) and Tris that enables sterile filtration through a 0.2 µm filter. The DLS analysis showed that the nanoparticle size of Modi-2 SNAPvax in histidine formulation was 12.6 nm ± 0.9 SD, whereas it was 14.3 nm ± 0.2 SD and 15.5 nm ± 0.1 SD in Tris and PBS formulation, respectively (Fig. 2C) with low polydispersity index values (Supplementary Table 6).Transmission electron microscopy analysis confirmed the spherical shape and similar size of the particles compared to DLS measurement. Example transmission electron micrograph images of nanoparticle formulations are shown in Fig. 2D. Zeta potential at respective formulation pH is shown in Supplementary Table 6. Modi-2 SNAPvax in histidine, with a zeta potential value of +35 mV, indicates that the particles are not prone to forming aggregates; hence, the formulation should remain stable. In addition, minimal material loss is reported upon sterile filtration from the nanoparticle formulations (Supplementary Table 7). As Modi-2 SNAPvax were formulated as “mosaic formulation” where each nanoparticle contains all four homocitrullinated peptide antigens, we used UPLC-PDAMS analysis to confirm that the four Modi-2 peptides with distinct peaks were detected (Fig. 2E). These results confirm that Modi-2 SNAPvax forms nanoparticles with similar size and mosaic composition of the four homocitrullinated peptides when formulated in different aqueous buffers. To further assess the stability of the Modi-2 SNAPvax formulations particle size was assessed over time with storage at −20 °C (Supplementary Figure 3) and with repeat freeze thaw cycles (Supplementary Fig. 4) and shown to retain initial particle size characteristics with multiple freeze thaws and over a 12-month period stored at −20 °C.

Modi-2 vaccine can stimulate efficient immune responses as a SNAPvax formulation

Modi-2 SNAPvax formulations were assessed for the capacity to induce antigen-specific T cell responses in mice as confirmed by IFNγ ELISpot analysis following vaccination. HLA-HHDII/DP4 and Balb/c mice were vaccinated with three doses of Modi-2 SNAPvax formulations and responses to homocitrullinated and wild type peptides were screened ex vivo. This screening showed significant responses to Aldo74-93Hcit (p < 0.0001), Bip562-579Hcit (p < 0.0001) and Vim116-135Hcit (p = 0.0101) in HLA-HHDII/DP4 mice (Fig. 3A). Balb/c mice showed significant responses to Aldo74-93Hcit (p < 0.0001), Cyk8 371-388Hcit (p < 0.0001) and Vim116-135Hcit (p < 0.0001) (Fig. 3C). Mice immunised with TLR7/8 adjuvant only or an irrelevant peptide SNAPvax formulation demonstrate no Hcit peptide specific responses (Supplementary Fig. 5) This data is consistent with the responses seen from the peptides mixed with CpG and MPLA adjuvants. Direct comparisons of SNAPvax formulation with peptide mixed with CpG and MPLA or TLR7/8 adjuvants is shown in Supplementary Fig. 5. Since Modi-2 SNAPvax can be formulated with similar particle size in different buffers the formulations of Modi-2 SNAPvax were assessed in PBS, histidine or tris containing buffers for immunogenicity and all stimulated similar immune responses (Supplementary Fig. 6).

Fig. 3: Modi-2 SNAPvax stimulates Hcit specific CD4 responses.

T cell responses to the Modi- 2 homocitrullinated and matched wild type peptides in HLA-HHDII/DP4 (A, B) and Balb/c mice (B–E). Mice were vaccinated with three doses of homocitrullinated peptides conjugated to SNAPvax and responses assessed against Hcit peptides (A, C), Hcit and wt peptides (C, D) or Hcit peptides in presence of CD4 or CD8 blocking antibodies (E). n ≥ 3 in each study and data is collated from at least two independent studies. Mean, standard deviations and significant p-values are shown.

Responses to Modi-2 SNAPvax formulations were also homocitrulline-specific showing no cross reaction to the wild-type peptides. IFNγ responses to Aldo74-93Hcit, Bip562-579Hcit and Vim116-135Hcit were significantly higher than the responses to Aldo74-93wt, Bip562-579wt and Vim116-135wt, p < 0.0001, p < 0.0001 and p = 0.0283 respectively, confirming minimal cross reactivity to the wt peptides in HLA-HHDII/DP4 mice (Fig. 3B). Similar findings were seen in Balb/c mice where IFNγ responses to Aldo74-93Hcit, Cyk8 371-388Hcit and Vim116-135Hcit were significantly higher than the responses to Aldo74-93wt, Cyk8 371-388 wt and Vim116-135wt, p < 0.0001, with minimal cross reactivity to the wt peptides (Fig. 3D).

We previously showed responses to Modi-2 peptides in the CpG/MPLA adjuvant formulation were CD4-mediated. However, the SNAPvax formulation has been shown to enhance the potency of peptide-based vaccines that mainly induced CD8 T cells responses21,24. Therefore, we reinvestigated whether responses to Modi-2 SNAPvax formulation were CD4 or CD8 mediated. In Balb/c mice, responses to Aldo74-93Hcit were significantly reduced by anti-CD4 blocking antibody (p = 0.0018) and significantly, albeit partially, blocked by anti-CD8 blocking antibody (p = 0.0032) (Fig. 3E). Responses to Cyk8 371-388Hcit and Vim116-135Hcit were also significantly lost after blocking CD4 T cells, p = 0.0041 and p = 0.0048, respectively. Whereas there was no significant loss in these responses after blocking CD8 T cells in Balb/c (Fig. 3E). These results were consistent with results shown in the previous section where Modi-2 peptides were mixed with the adjuvants CpG/MPLA (Fig. 1G), confirming that IFNγ responses to Modi-2 SNAPvax formulations were also mediated by CD4 T cells. Immunogenicity analysis therefore suggests similar immune responses stimulated in mouse models from the large particulate peptide in CpG and MPLA formulation and the nanoparticle SNAPvax formulation.

Homocitrulline-specific CD4 responses mediate tumour therapy in several mouse models

In the tumour environment MPO secreted by infiltrating MDSCs can mediate homocitrullination of proteins in the tumour microenvironment making homocitrullinated epitopes a promising target for tumour therapy. However, it is possible that not all lysines are naturally homocitrullinated and presented by MHCII to be recognised by CD4 T cells. We therefore aimed to investigate if the Hcit peptides in Modi2 vaccine are naturally presented by examining if the CD4 mediated IFNγ responses to Aldo74-93Hcit, BiP562-579Hcit, Cyk8 371-388Hcit and Vim116-135Hcit were sufficient to provide potent tumour therapy in vivo.

We first assessed the expression of Aldo A, Bip, Cyk8 and Vim native proteins in in vitro grown B16 HHDII iDP4, 4T1 and CT26 tumour cell lines by western blot. This assessment confirmed that Aldo A, Bip and Vim proteins were expressed in B16F1 HHDII iDP4, 4T1 and CT26 tumour cell lines. However, Cyk8 was only expressed in 4T1 and was not detected in in vitro grown CT26 and B16F1 HHDII iDP4 tumour cell lines. Py8119 tumour cells were used as a negative control for Cyk8 protein and a positive control for Vim protein. Py230 cells were used as a positive control for Cyk8 and a negative or low positive control for Vim. B16 HHDII iDP4 cells were used as a positive control for Aldo A. B16F1 cells treated with thapsigargin, an endoplasmic reticulum (ER) stress inducer that enhances the Bip expression25, were used as a positive control for Bip protein (Fig. 4A, B). Raw images for western blots are shown in Supplementary Fig. 16.

Fig. 4: Murine tumours express Modi-2 target antigens and show evidence of carbamylation/homocitrullination.

A, B The relative expression of Aldo, BiP, Cyk8 and Vim in different murine tumour cell lines measured by western blot. C The percentage of carbamylation in murine tumour cell lines ex vivo assessed by flow cytometry as representative staining histogram (i) and averages (ii). Anti-carbamyl lysine antibody staining is gated on live CD45- cells. Values show the percentage of gated cells staining positive. Mean of at least 3 independent samples, standard deviations and significant p values are shown.

As Modi-2 vaccine targets homocitrullinated proteins, we assessed the level of carbamylation (homocitrullination) in 4T1, CT26 and B16 HHDII tumours ex vivo. In the absence of homocitrullinated protein specific detection reagents a general anti-carbamylation antibody was used. Tumour cells isolated ex vivo from 4T1, CT26 and B16 HHDII tumours implanted subcutaneously were stained with anti-carbamylation antibody when reaching 10–15 mm in diameter. Representative staining is shown in Fig. 4Ci and average expression in Figure 4Cii. Gating strategy is shown in Supplementary Figure 7. Although homocitrullination/carbamylation is detected, this staining with the pan anti-carbamylation antibody does not detect if Aldo A, Bip, Cyk8 and Vim proteins expressed in tumour cells are specifically carbamylated.

Results mentioned above confirmed that 4T1, B16 and CT26 tumour models were suitable to assess the anti-tumour potency of the Modi-2 vaccine. Mice were therefore challenged with B16 HHDII iDP4 cells and vaccinated four days later with 4nmol of Modi-2 mixed with CpG/MPLA or 4nmol Modi-2 SNAPvax and compared to unimmunised or adjuvants alone. At the point of vaccination tumours were < 1 mm diameter. No difference was noted with adjuvants alone compared to unimmunised mice. Those vaccinated with 4nmol Modi-2 SNAPvax Histidine buffer formulation showed a significantly enhanced overall survival compared to unvaccinated mice (p = 0.0009) (Fig. 5A, B). Modi-2 CpG/MPLA formulation also showed 50% overall survival at day 50 which was significantly higher than the unvaccinated B16 HHDII iDP4-bearing mice p = 0.0044 (Fig. 5A, B). Similar findings were also seen in the CT26 tumour model with Modi-2 SNAPvax formulation. Adjuvants only had no impact on tumour therapy. The vaccination of CT26-bearing mice with 4nmol Modi-2 SNAPvax Tris buffer formulation or Modi-2 CpG/MPLA induced potent anti-tumour responses with 100% overall survival which was significantly higher than the unimmunised or adjuvant only immunised CT26-bearing mice, p < 0.0001 and p = 0.0007 respectively (Fig. 5C, D). The vaccination of 4T1-bearing mice with 4nmol of Modi-2 SNAPvax Histidine buffer formulation also showed significantly potent anti-tumour therapy with 70% overall survival respectively compared with the unvaccinated 4T1-bearing mice, p = 0.0090 respectively (Fig. 5E, F). Mice in all tumour therapy studies showed no clinical signs associated with autoimmunity and weights remained stable throughout the duration of the studies (Supplementary Fig. 8).

Fig. 5: Modi-2 SNAPvax vaccination mediates tumour therapy.

HHDII/DP4 transgenic (A, B) or Balb/c (C–G) mice were injected with 1 x 105 B16 HHDII/iDP4 (A, B), 1.75 x 104 CT26 (C and D) or 5 × 103 4T1 (E–G) tumour cells respectively on day 1 followed by immunisation with 4nmol Modi-2 peptides in CpG/MPLA or 4nmol of Modi-2 SNAPvax on days 4, 11 and 18 (A, B) or days 4, 8 and 11 (C–G). Overall survival (OS) and tumour growth monitored (n ≥ 5). Numbers on tumour growth curves represent the number of tumour free (TF) mice out of the total. Survival analysed using LogRank test. (G), Tumours from 4T1 model disaggregated and stained ex vivo for CD45, CD8, CD4 infiltrate and MHCII expression. Representative tumour examples and histograms of averages (n > 3) with error bars of standard deviations are shown.

We further examined the tumour infiltrating lymphocytes (TILs) from tumours that developed in the Modi-2 SNAPvax vaccinated and unvaccinated 4T1-bearing mice. 4T1 tumours show a degree of cell death upon disaggregation but this is similar in both immunised and control animals and samples are gated to focus on the live TIL population (Supplementary Figure 9). The subsequent flow cytometry gating strategy is shown in Supplementary Figure 10. A representative staining example is shown in Fig. 5G. The analysis of tumours from Modi-2 SNAPvax immunised mice were those that failed therapy but despite this they showed evidence of a significant increase in infiltration of CD45+ cells into tumours. Within the CD45+ population there was a significant increase in CD4+ cells in vaccinated compared to control tumours (Fig. 5G). Supplementary Fig. 10 confirms that CD4+ cells within the CD45+ population are T cells. Modi-2 SNAPvax immunised mice also showed a significant increase in the expression of MHCII on both CD45+ and CD45- cells within tumours (Fig. 5G). The percentage of CD45-MHCII+ in the tumours of unimmunised mice was significantly lower than seen in tumours from the vaccinated mice (Fig. 5G). Data across multiple mice is shown in Fig. 5H where a significant increase is demonstrated for CD4 T cell infiltration and MHCII expression. This data suggests the vaccine increases the leukocyte infiltration into tumours, particularly for CD4 T cells and that both lymphocytes and non-lymphocytes show an increase in MHCII reflecting a more proinflammatory environment but the analysis of tumours that failed therapy may be underestimating the effect of the Modi-2 vaccine. These results show that Modi-2 SNAPvax induced homocitrulline specific T cell responses that were sufficient to mediate tumour therapy in HLA-HHDII/DP4 and Balb/c mouse tumour models.

Human tumours express Modi-2 target antigens and expression correlated with low overall survival rates

To screen which cancer patients may benefit from the Modi-2 vaccine, we assessed the expression of Modi-2 target proteins on tumour tissues derived from patients with breast, colorectal, lung and prostate cancer. breast, colorectal, lung and prostate TMAs were stained with specific antibodies for Aldo A, Bip, Cyk8 and Vim. Representative staining is shown in Fig. 6A. Analysis showed that the H-score of the staining for Aldo A, Bip and Cyk8 were significantly higher in colorectal cancer tissues compared with colorectal normal adjacent tissues, p < 0.0001, p = 0.0017, p = 0.0005 respectively (Fig. 6B). Our IHC results also showed that normal colorectal adjacent tissues showed significantly higher H-score of Vim staining than colorectal cancer tissues, p < 0.0001 (Fig. 6B). The breast cancer tissues also showed medium to high H-score staining to Aldo, Bip and Cyk8 proteins (Fig. 6C). However, the H-score staining of Vim was modest on breast cancer tissues. Unfortunately, breast TMAs did not include normal or normal adjacent tissues, therefore we could not compare the expression of proteins and the level of carbamylation between breast cancer and normal or normal adjacent tissues.

Fig. 6: Human tumours express Modi-2 target antigens.

Colorectal (CRC) and breast (BR) tumour microarrays stained by immunohistochemistry for vimentin, cytokeratin 8, aldolase A, BiP and MPO using antigen specific antibodies and homocitrullination using an anti-carbamyl antibody. A representative staining images of low, moderate and high staining. B Data shown as H-score for normal or normal adjacent tissue (NAT) compared to tumour tissue on the CRC TMAs. C Data shown as H-score for tumour tissue on the BR TMAs.

The IHC analysis also showed that the H-score of Aldo A, Bip and Cyk8 proteins was significantly higher on lung cancer tissues compared with lung normal adjacent tissues (Supplementary Fig. 11A). There was no significant difference in the H-score of Vim, carbamylation and MPO staining between lung cancer and normal adjacent tissues. Similar results observed on prostate TMAs (Supplementary Fig. 11B). These results confirm that Aldo A, Bip, Cyk8 and Vim were expressed with different intensity in several human tumours including breast, colorectal, lung and prostate. Representative staining images of low, moderate and high staining of Aldo A, Bip, Cyk8 and Vim on lung and prostate TMAs are shown in Supplementary Figure 12.

Unfortunately, survival data was not available for the TMAs stained in this study therefore we further investigated if the expression of these proteins are correlated with the overall survival in cancer patients from available data sets. In silico analysis of integrated proteo-transcriptomics dataset of breast cancer from Tang et al.26 revealed that breast cancer patients with high expression of Aldo, Bip, Cyk8 and MPO, the enzyme responsible for carbamylation, had poor overall survival which was significantly decreased compared with patients who had low expression of those proteins, p = 0.044, p = 0.0055, p = 0.0236 and p = 0.0040, respectively (Supplementary Fig. 13A). In silico analysis of microarray gene expression colorectal cancer dataset (downloaded from www.kmplot.com) showed that colorectal cancer patients with high relative expression of Vim, Aldo, Cyk8 and MPO had poor overall survival which was significantly decreased compared with patients who had low relative expression of Vim, Aldo, Cyk8 and MPO, p < 0.0001, p = 0.0008, p = 0.0152 and p = 0.0127, respectively (Supplementary Fig. 13B).

In silico analysis of microarray gene expression lung cancer dataset (downloaded from www.kmplot.com) revealed that lung cancer patients with high relative RNA expression of Aldo A, BiP, Cyk8 and MPO were significantly associated with low overall survival rates in lung cancer patients, p < 0.0001, p = 0.0300, p < 0.0001 and p = 0.0004, respectively (Supplementary Fig. 14). In silico analysis of RNAseq advanced prostate cancer dataset from Abida et al.27 showed that patients with advanced prostate cancer with high relative expression of Cyk8 had poor overall survival which was significantly decreased compared with patients who had low expression of Cyk8, p = 0.0079 (Supplementary Fig. 15). This in silico data supports our IHC staining data and provides further evidence that expression of these proteins is altered in a variety of solid cancers and could be appropriate targets for immune therapy. Although there is no data on levels of carbamylation and overall survival, the enzyme MPO that is most often produced by a subset of MDSCs in tumours mediates carbamylation could be as an indication of carbamylation potential. The association of high MPO with lower survival and poorer prognosis suggests that carbamylation is also a candidate target for therapies.

Data presented in this study demonstrates the efficacy of the Modi-2 vaccine in murine tumour models and provides evidence that the Modi-2 vaccine has potential as tumour therapy to treat patients with solid tumours including breast, colorectal, lung and prostate that express Aldo A, Bip, Cyk8, Vim and MPO.