Innate immune receptor C5aR1 regulates cancer cell fate and can be targeted to improve radiotherapy in tumours with immunosuppressive microenvironments

An immunosuppressive microenvironment causes poor tumour T-cell infiltration and is associated with reduced patient overall survival in colorectal cancer. How to improve treatment responses in these tumours is still a challenge. Using an integrated screening approach to identify cancer-specific vulnerabilities, we identify complement receptor C5aR1 as a druggable target which when inhibited improves radiotherapy even in tumours displaying immunosuppressive features and poor CD8+ T-cell infiltration. While C5aR1 is well-known for its role in the immune compartment, we find that C5aR1 is also robustly expressed on malignant epithelial cells, highlighting potential tumour-cell specific functions. C5aR1 targeting results in increased NF-κB-dependent apoptosis specifically in tumours and not normal tissues; indicating that in malignant cells, C5aR1 primarily regulates cell fate. Collectively, these data reveal that increased complement gene expression is part of the stress response mounted by irradiated tumours and that targeting C5aR1 can improve radiotherapy even in tumours displaying immunosuppressive features.


3
The composition of the tumour microenvironment (TME) impacts treatment responses in cancer (Bindea et al., 2013;Llosa et al., 2015;Tauriello et al., 2018). Immunosuppressive TME features, which act as a barrier to extensive CD8+ T cell infiltration typically characterise immune cold tumours which are associated with poor prognosis (Galon et al., 2006;Guinney et al., 2015). Indeed, low density of total T lymphocytes (CD3+) at the centre or invasive margins of tumours is associated with reduced overall survival in colorectal cancer (CRC) (Galon et al., 2006). Improving treatment responses for those patients with poor tumour lymphocyte infiltration remains a challenge.
Approximately one third of all colorectal cancers arise in the rectum. Locally advanced rectal cancers are typically treated with neoadjuvant chemoradiotherapy (nCRT) prior to surgery. Unfortunately, despite nCRT leading to complete pathological regression in 20-30% of these patients, 70-80% will fail to achieve complete responses. There is therefore a need to further improve treatment responses in a significant portion of patients receiving nCRT (Cercek et al., 2018;Li et al., 2022;van der Sluis et al., 2019). Identifying targets that modulate radiosensitivity, particularly in tumours displaying immunosuppressive features could improve treatment outcomes for the most difficult to treat tumours.
High expression of complement system components is part of the inflammatory environment of colon and rectal tumours displaying the worst survival outcomes (Becht et al., 2016;Domingo et al., 2021;Guinney et al., 2015;Krieg et al., 4 2022). The complement system is an ancient component of innate immunity and both canonical and non-canonical functions are increasingly being recognised as important for infection control, autoimmunity and cancer (Daugan et al., 2021a;Gros et al., 2008;Pio et al., 2013;Ricklin et al., 2010;Roumenina et al., 2019).
Whether in the context of cancer treatment, complement proteins are expressed and function independently of their role in the inflammatory environment remains to be fully understood.
In this study we find that in murine models which recapitulate an immunosuppressive TME, the complement system is the first immune response pathway to be upregulated at early timepoints following irradiation. Through an integrated screening approach, we identify complement receptor C5aR1 as a druggable target, which inhibits radiation-induced cell death/apoptosis through regulation of tumour cell fate. Interestingly, these effects are not observed in untransformed intestinal organoids or normal intestinal tissues in vivo.
Consequently, targeting C5aR1 with a clinical grade and orally active C5aR1 antagonist, PMX205, results in improved tumour radiation responses in vivo.
Importantly, PMX205 improves tumour radiation response in several murine models, including those displaying high radiation-induced complement expression and immunosuppressive features associated with CD8+ T cell exclusion.

Identification of radiation-responsive targets in immunosuppressive tumours
When grown subcutaneously, tumour organoids originally derived from villinCreER Apc fl/fl Kras G12D/+ Trp53fl/fl TrgfbrI fl/fl (AKPT) mice display TME features resembling those of CRCs associated with poor outcome (Figures 1A-C). These features include stromal-rich regions with high numbers of fibroblasts and macrophages but relatively few CD8 + T-cells ( Figures 1A-C). Interestingly, we found that although irradiation was able to enhance infiltration of Tregs, macrophages, neutrophils, CD4 + and CD8 + T-cells into these tumours, such infiltration was still limited to stromal regions and did not increase intra-epithelial suggest that complement gene expression is globally, although transiently, induced following irradiation in this model of an immunosuppressive TME.

C5aR1 is a radiation-responsive druggable target
To identify potential targets within the complement cascade that could be therapeutically inhibited, we queried the CanSAR database (querying complement system components, receptors, proteases and regulators). A gene was only considered a hit if it was "druggable" based on structural and ligandbased assessment as shown in green (Figure 2A and Supplementary Table 2).
We also interrogated the DepMap database (which combines data from CRISPR-Cas9 and RNAi screens in more than 700 cell lines), to identify cancer-derived complement proteins that may have autocrine functions specifically impacting cell fate under stress conditions. We reasoned that looking for non-essential hits would allow the identification of genes providing stress-specific dependencies and, therefore, potential therapeutic targets less likely to mediate toxicity in normal tissues. Following the combined CanSAR and Depmap analysis, we found three hits: C5, C5AR1, C4BPA (Figure 2A and Supplementary Table 2).
ATR was included in the screen as a positive control for essential genes since its deletion is lethal in several cell lines due to its role in replication and the DNA damage response (Cimprich and Cortez, 2008;Flynn and Zou, 2011).
Interestingly, C1QBP, a complement gene which was recently shown to play a role in the DNA damage response by modulating DNA resection, was essential in a number of cell lines, further validating our screening approach (Bai et al., 2019) (Figure 2A) Table 2).

(Supplementary
C5 encodes a complement component, which when cleaved will form C5a. C5aR1 is the main signalling receptor for the C5a ligand. C4BPA encodes the alpha chain of complement regulator C4BP (Hofmeyer et al., 2013) and there are currently no known pharmacological approaches for targeting C4BPA. To further narrow down which hit would be the best therapeutic target, we assessed the association of C5, C5aR1 and C4BPA mRNA expression with prognostic outcomes and found that only high C5aR1 mRNA expression was associated with significantly poor disease-free survival in colorectal cancer ( Figure 2B-D).
We confirmed that high C5aR1 mRNA expression was correlated with decreased overall survival in a further independent dataset (Supplementary Figure 2A).
In vivo we found that C5aR1 was robustly expressed in AKPT tumours at baseline. A transient induction in C5aR1 expression was also observed after RT  Figure 2C, for negative control C5aR1 staining). Interestingly, we found that at earlier timepoints post RT C5aR1 expression is more prominent in the epithelium, while stromal C5aR1 expression appears to dominate at later timepoints (3 and 7 days)( Figure 2G). The dominance of stromal C5aR1 8 expression at these later timepoints may reflect increased infiltration of C5aR1expressing immune cells following irradiation, although this remains to be formally assessed. AKPT tumours harbour stromal infiltration features comparable to those patients with tumours classified as consensus molecular subtype (CMS) 4. We therefore asked whether C5aR1 expression might be differentially expressed across subtypes (CMS1-4). Analysis of pre-treatment rectal tumour biopsies identified that those classified as CMS4 had the highest RNA levels of C5aR1 expression compared to the other subtypes ( Figure 2H).
To directly assess whether radiation could impact tumour cell intrinsic expression of C5aR1 or C5, we turned to an in vitro system. Following treatment of mouse and human colorectal tumour cells with irradiation, we found a modest but

C5aR1 regulates tumour cell survival under stress
PMX205 is a selective inhibitor of C5aR1 currently undergoing clinical testing for ALS where it is reportedly well-tolerated (Kumar et al., 2020). We assessed the effects of treating colorectal tumour cells with PMX205. As anticipated, given the fact that C5aR1 is a G-protein coupled receptor (GPCR), we found reduced ERK1/2 and RelA phosphorylation (as a readout for NFB signalling) in PMX205 treated cells ( Figure 3A). However, PMX205 had negligible effects on AKT phosphorylation (at Threonine 308) (Supplementary Figure 3A and B).
Changes in MAPK signalling could impact cell cycle distribution which in turn could impact cellular radiosensitivity. However, we did not note any significant differences in cell cycle profiles by flow cytometry in cells treated +/-PMX205 +/-RT (Supplementary Figure 3C). We also did not note significant changes in H2AX levels between treatment groups (at late timepoints) suggesting that DNA Targeting C5aR1 did not result in increased apoptosis in the absence of irradiation, in agreement with a stress-specific role in modulating cell death.
Supporting this, in HCT 116 xenografts, we observed increased apoptosis in PMX205 treated tumours following irradiation ( Figure 3E).
To investigate whether apoptosis of PMX205-treated tumour cells occurred downstream of attenuated GPCR-associated signalling, we first depleted NF-B inhibitor, IB as a means of interrogating the NF-B dependence of the effects observed. If PMX205-mediated apoptosis was occurring in an NF-B-dependent manner, IB depletion would be expected to result in decreased apoptosis in PMX205 treated cells. We indeed observed that, following irradiation, IB depletion in PMX205 treated cells attenuated the apoptotic response ( Figure 3F and Supplementary Figures 3I and J). IB depletion did not have a dramatic effect on apoptosis levels in the vehicle + RT treated cells. We hypothesize this is because IB levels are reduced by DNA damaging agents, and NF-B signalling is already high in these cells. We also depleted RelA and found that while RelA depletion increased apoptosis levels in RT + vehicle treated cells (as expected); there was no further increase in apoptosis in RT + PMX205 treated cells (presumably since these cells already have reduced "active" RelA Figures 3K and L). Furthermore, interrogation of the TCGA patient datasets where high C5aR1 mRNA expression was associated with poor outcome, identified that C5aR1 expression was positively and significantly correlated with pro-survival/anti-apoptosis genes including NF-B target genes in the BCL2 family ( Figures 3G and H). We confirmed these associations in two further independent patient datasets including rectal cancer patient biopsies collected prior to radiotherapy ( Figure 3I and Supplementary Figure 3M).

(Supplementary
Treatment of colorectal cancer cells with PMX205 also resulted in reduced mRNA expression of BCL2 (Supplementary Figure 3N).
We next assessed the effects of ERK inhibition on PMX205-induced apoptosis by treating colorectal cancer cells with PMX205 and ERK inhibitor Selumetinib (with or without irradiation). As expected, PMX205 and Selumetinib alone resulted in enhanced apoptosis following irradiation; with PMX205 displaying the most significant effects (Supplementary Figure 3O and P). A moderate (yet not significant) increase in apoptosis was also observed when both compounds were combined (Supplementary Figure 3O and P). These data suggest that although attenuated ERK may contribute to apoptosis following PMX205 treatment, it is unlikely to be a main driver of the apoptotic effect observed. Together, these data suggest that C5aR1 mediates tumour cell pro-survival signalling, with NF-B acting as a key regulator of this response. In support of this conclusion, increased cell death following PMX205 treatment occurs downstream of attenuated NF-B signalling.

C5aR1 deficiency does not result in increased apoptosis in healthy intestinal epithelium
To assess whether the cell-intrinsic effects of complement were specific to malignant cells we first considered if organoids derived from different genotypes expressed complement genes when grown in vitro (and therefore in the absence of systemic complement or TME-derived complement products). Interestingly, following RNA-seq we found that complement genes were significantly differentially expressed in both APKT tumour organoids as well as organoids derived from villinCreER Kras G12D/+; Trp53 fl/fl Rosa26 N1icd/+ (KPN) mice (compared to untransformed WT organoids). Over 74% of the complement genes queried were expressed at significantly higher levels in the AKPT organoids compared to the untransformed WT organoids (including C5aR1) suggesting that increased cell intrinsic complement expression is a malignant cell-associated phenomenon ( Figure 4A and B).
To investigate whether targeting C5aR1 might also alter pro-survival signalling in healthy tissues we assessed RelA and ERK phosphorylation in untransformed intestinal organoids. We noted that while ERK phosphorylation was reduced following PMX205 treatment, RelA phosphorylation was not affected  Figure 4C). In line with the lack of RelA phosphorylation changes observed by western blotting, we noted that transcriptional NF-B antiapoptosis target genes were not differentially expressed in PMX205 + RT vs RT alone treated organoids ( Figure 4C). Similarly, in vivo, small intestines did not show significant transcriptional changes in anti-apoptosis target genes in the BCL2 family with deletion of C5aR1 (following irradiation) ( Figure 4D). These data suggest that changes in NFB-anti-apoptotic signalling downstream of C5aR1 do not occur in the untransformed intestinal epithelium. In line with these signalling changes, we did not observe an increase in apoptosis in small intestines following PMX205 treatment or C5aR1 loss ( Figure 4E and F). In fact, small intestinal crypts in vivo had significantly reduced apoptosis following irradiation and PMX205 treatment ( Figure 4E). Similarly, small intestinal crypts of C5aR1 -/displayed significantly reduced apoptosis compared to WT mice following total abdominal irradiation ( Figure 4F). Together these data indicate that C5aR1 attenuates stress-induced apoptosis in malignant but not nontransformed epithelial cells.

C5aR1 inhibition improves tumour radiation response
To assess whether targeting C5aR1 could improve tumour radiation responses in vivo we treated MC38 subcutaneous tumours with PMX205 and either no irradiation, fractionated radiotherapy (3 x 4.45 Gy) or single dose irradiation (9Gy) (as shown in three different treatment schemes in Figure 5A). PMX205 treatment in the absence of irradiation did not have significant effects on tumour response, consistent with a non-essential role for C5aR1 in the absence of stress ( Figure 5B). PMX205 treatment also had no negative effects on mouse weight (Supplementary Figure 5A and B). However, following treatment with either single dose (9 Gy) or equivalent multiple fractionation (3 x 4.45 Gy) regimens (equivalent assuming an / ratio of 5.06) (Suwinski et al., 2007

Targeting C5aR1 does not increase the % of CD8 + T-cells in the tumour following irradiation
Targeting C5aR1 is currently undergoing clinical testing as a means of reinvigorating anti-tumour CD8 + T cell responses (Massard et al., 2019). Since we had observed stromal C5aR1 expression in irradiated tumours, we next investigated tumour immune infiltration changes in tumour draining lymph nodes and tumours following PMX205 and irradiation (tissues harvested 7 days postirradiation as shown in Figure 6A). Interestingly, no significant changes in CD3, CD4, NK, or B-cells were found in the tumour draining lymph nodes across treatment groups (Figures 6B-E and Supplementary Figure 6A). Mice treated with PMX205 alone, however, did display a higher % of CD8 + T cells compared to those treated with the PMX205 and irradiation combination (Figures 6F). Mice treated with PMX205 alone also had a reduced % of Tregs compared to vehicle treated mice ( Figure 6G). However, these changes did not correlate with altered functionality/effector functions of CD8 + T cells which showed comparable expression of IFN, GrzB and TNF across all treatment groups at this timepoint ( Figures 6H-L). In the tumour, we found that although 9 Gy irradiation, as expected, significantly increased levels of CD3 + and CD8 + T-cells; treatment with PMX205 did not further increase the % of these cells ( Figure 6M, 6N and Supplementary Figure 6B). In fact, the % of CD8 + T-cells was significantly reduced in irradiated tumours after PMX205 treatment when compared to irradiated vehicle-treated mice ( Figure 6N). This was surprising given the improved tumour regression previously observed in these models ( Figure 5B).

C5aR1 inhibition can improve radiotherapy in tumours with an immunosuppressive microenvironment
The improved tumour responses observed in the context of reduced tumour CD8 + T cell numbers, made us question whether PMX205 could be used to improve radiation responses in other models displaying low CD8 + T-cell infiltration. We, therefore, next asked whether PMX205 could improve tumour response in the AKPT models where we had previously observed robust C5aR1 expression and low CD8 + T cell tumour infiltration. In support of C5aR1 having cell intrinsic effects we observed that, in vitro, PMX205 reduced survival of AKPT organoids upon treatment with increasing doses of irradiation ( Figure 7A). When these organoids were grown as subcutaneous tumours, PMX205 alone had no significant effect on tumour growth (as previously observed in the MC38 model) ( Figure 7B and C). However, combination treatment with PMX205 and RT, resulted in a significant tumour growth delay, increased apoptosis and dramatic improvement in tumour-free survival with 20% of mice having unpalpable tumours at the end of the experiment ( Figure 7D-F). Together our data indicate that targeting C5aR1 can improve tumour radiation responses even in models with low tumour CD8 + T-cell infiltration. Importantly, improved responses are associated with increased tumour cell apoptosis and without concomitant increases in healthy intestinal epithelial cell apoptosis ( Figure 7G).

Discussion
Identifying tumour-promoting components of the TME presents therapeutic opportunities. However, expression of these components can be extremely dynamic, and may be governed by selective pressures such as those posed by treatment-induced stress responses. The effects of these selective pressures on dysregulation of complement components and their evolving functions in the TME is unclear. Here we report that complement gene expression is induced after irradiation of murine tumour models which recapitulate features of human tumours displaying the worse outcomes. Amongst these genes, we note that C5aR1 expression is transiently induced following radiotherapy, likely as a stress response mounted to promote tumour cell survival. Interestingly, in AKPT subcutaneous tumour models, C5aR1 is expressed by both the tumour epithelium and stroma. We find that epithelial expression is more prominent at baseline and early timepoints post irradiation while stromal expression dominates at later timepoints. Increased C5aR1 stromal expression at later timepoints may reflect recruitment of C5aR1 expressing immune cells into the stroma following irradiation (as indicated in Figure 1C). Interestingly, in rectal cancer patients we find that those classified as CMS4 have the highest levels of C5aR1 expression compared to the other subtypes. Together, our data indicate that patients with tumours displaying immunosuppressive features and high C5aR1 expression could represent patient populations most likely to benefit from C5aR1 targeting therapy.
Previous reports investigating the effects of complement inhibition on radiotherapy response have focused on the role of complement in modulating anti-tumour immune responses, albeit with conflicting results (Elvington et al., 2014;Surace et al., 2015). Furthermore, none of these reports considered the effect of complement-targeting therapies on cancer-cell intrinsic functions (Bai et al., 2019;Block et al., 2019;Cho et al., 2014). Whether targeting complement would have the same effects in tumour and normal tissues had, until this study, remained completely unexplored. Using a combination of RNA-sequencing and in silico mining of depmap, CanSAR, and patient datasets, we identify C5aR1 as a druggable target for enhancing stress-specific cell death in tumours. Our data indicate that C5aR1 negatively regulates apoptosis in cancer cells by modulating cell survival pathways such as NF-B, and that attenuating such pro-survival signalling can render cancer cells more susceptible to cell death following RT.
Importantly, in the normal intestine targeting C5aR1 does not result in increased apoptosis and, in fact, C5aR1 deficiency appears to confer a protective phenotype (which might be explained by the lack of attenuated NF-B signalling).
Why or how C5aR1 may differentially regulate signalling in normal and malignant cells remains to be elucidated. However, divergent consequences of autocrine complement signalling between cell lines have been previously reported (Daugan et al., 2021b). To therapeutically target C5aR1, we used the specific antagonist PMX205 which has FDA and EMA "orphan drug" approval for ALS; allowing accelerated progression to clinical trials (Ricklin and Lambris, 2016). This class of cyclic peptides have minimal penetrance into the cell (Niyonzima et al., 2021) and should therefore primarily impact cell-surface C5aR1 (and its effects on GPCR signalling). This is important since recent reports indicate that C5aR1 intracellular pools are present in tumour cells where they contribute to tumourigenesis through -catenin stabilisation (Ding et al., 2022).
We have compared, for the first time, single dose, and equivalent fractionation doses in colorectal cancer models to specifically assess the effect of targeting complement at the level of C5aR1. In both settings we observe that PMX205 improves tumour radiation response. We propose that increased PMX205mediated cell death is a new mechanism that can be exploited to improve radiation responses even in tumours displaying immunosuppressive features.
The importance of modulating anti-tumour immune host responses following C5aR1 targeting has been highlighted in previous studies and ourselves and others have observed reduced tumour burden in C5aR1 -/mice (data not shown) (Ding et al., 2022;Markiewski et al., 2008;Surace et al., 2015). The defect in tumour uptake displayed by C5aR1 -/mice complicates the investigation of tumour radiation responses in this model and might have contributed to some of the previous conflicting reports on the effects of targeting the complement system on tumour radiation responses. The reduced tumour CD8 + T-cell infiltration observed here, however, is in line with a previous study indicating that local production of anaphylatoxins C3a and C5a and signalling through their respective receptors is required for dendritic cell maturation and CD8 + T-cell activation (Surace et al., 2015).
Overall, this study indicates that increased complement gene expression is part of the stress response mounted by irradiated tumours to sustain tumour cell survival via C5aR1 signalling. These data therefore indicate that, beyond its previously described functions, C5aR1 can also sustain tumour cell survival in a cell intrinsic manner. Consequently, targeting complement receptor C5aR1 can improve radiotherapy responses even in tumours displaying reduced CD8+ T-cell infiltration. Importantly, C5aR1's pro-survival functions appear to be malignant cell specific since increased apoptosis is not observed in the normal intestinal epithelium following C5aR1 loss. This work is relevant since identifying targets that specifically modulate cancer cell radiosensitivity and can do so even in the absence of robust tumour CD8 + T cell infiltration could lead the way to improving treatment outcomes for the most difficult to treat tumours.

Pattern Recognition Proteases
Receptors Regulators

(B) Graph shows CD3 + T-cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(C) Graph shows CD4 + T-cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(D)Graph shows total B cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(E) Graph shows total NK cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(F) Graph shows CD8 + T-cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant; * = p<0.05 by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(G) Graph shows Tregs-cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant; ** = p<0.01 by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(H) Graph shows IFN + CD8 T-cells (as a % of live cells) in tumour
draining lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(I) Graph shows GrzB + CD8 T-cells (as a % of live cells) in tumour draining
lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(J)Graph shows GrzB + IFN + CD8 T-cells (as a % of live cells) in tumour
draining lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(K) Graph shows TNF + IFN + CD8 T-cells (as a % of live cells) in tumour
draining lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test. (L) Graph shows TNF + CD8 T-cells (as a % of live cells) in tumour draining lymph nodes of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant by ordinary one-way ANOVA with Tukey's multiple comparisons test. (M) Graph shows CD3 + T-cells (as a % of live cells) in tumours of mice receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant; * = p<0.05; **** = p < 0.0001 by ordinary one-way ANOVA with Tukey's multiple comparisons test.

(N) Graph shows CD8 + T-cells (as a % of live cells) in tumours of mice
receiving 0 or 9 Gy and either vehicle or PMX205 treatment following the same dosing scheme as shown in A. Tumours were harvested 7 days after irradiation. ns = not significant; * = p<0.05; ** = p < 0.01 by ordinary one-way ANOVA with Tukey's multiple comparisons test. For survival fraction studies, organoids were used 24 hours after passaging and plated into a 24-well plate. Vehicle or PMX205 treatment was carried out as previously described and added into the surrounding media 1 hour prior to irradiation. Organoid irradiation was carried out using an x-ray irradiator with lead shielding allowing for cumulative doses across a single plate (within a twenty-four well plate, two columns of four received 0Gy, two received 4Gy, and two received 9Gy). Organoids were imaged over the course of 3 days using a JuLi stage Real-Time Cell History recorder (NanoEnTek Inc., Korea) by using a 4x objective.
Using the images, organoids were manually counted at day 0 and 3 with a surviving fraction subsequently calculated. RT_Image software package, version 3.13.1, running on IDL version 8.5.1 was used to visualize CT images and perform treatment planning (Graves et al., 2007). Therapeutic irradiations were performed using an x-ray energy of 225kVp, a current of 13mA, a power of 3000 Watts, and a beam filter of 0.3 mm Cu, producing a dose rate of ~300cGy/min at the isocenter. Treatment x-ray beams were shaped using a 10 or 15 mm collimator to selectively irradiate the target while sparing adjacent tissue when performing xenograft irradiation. For total abdominal irradiations the 20 mm collimator with two beams; 0 and 180 degrees was used. After radiation delivery mice were removed and placed in a recovery box prior to being returned to their cage.
Subcutaneous tumours were measured with the use of callipers and volumes calculated by the ellipsoid estimation method as previously described (Taniguchi et al., 2014). Mice were euthanised as per APLAC guidelines in a CO2 chamber and by cervical dislocation.
For heterotopic AKPT organoid-derived tumour models, female C57BL/6 mice were purchased from Charles River at 5-6 weeks of age. Mice were housed within a pathogen-free environment with a 12-hour light cycle at the RRI Biomedical Science Animal Unit at the University of Oxford. AKPT organoids were suspended in a PBS, Matrigel mixture (1:1) prior to subcutaneous injection.
Resulting tumours were monitored once a day and their volume determined from: Length x Width x Height x 0.52 using calliper measurements. For experiments shown in Figures 1 and 2, once the mean tumour volume across the mice reached 150-200 mm 3 , mice received irradiation using the Gulmay Medical RS320 irradiator (300kV, 10mA, 1.81Gy/min).
For the experiments shown in Figure 7, four groups of mice (n=10) were established: control, PMX205 treatment, Irradiation, and a combination of irradiation plus PMX205 treatment. Once mean tumour volume for all mice reached 100 mm 3 , treatment was then initiated. PMX205 was delivered by oral gavage once a day for three days. For groups receiving radiation, 9 Gy was delivered using the Gulmay Medical RS320 irradiator (300kV, 10mA, 1.81Gy/min). Once tumour size reached a length by width 1.2 cm geometric mean mice were euthanised by a Schedule 1 method.
For normal tissue studies, total abdominal irradiation of either C57/BL6 (JAXX)(treated with vehicle or PMX205) or WT and C5aR1 -/-BALBc/J (JAXX) mice was performed on anaesthetised animals (with the use of ketamine 100mg/kg/xylazine 20mg/kg) and using a 225 kVp cabinet X-ray system filtered with 0.5 mm Cu (at Comparative Medicine Unit, Stanford, CA). Mice were euthanised as per APLAC guidelines in a CO2 chamber and by cervical dislocation.
For in vivo experiments involving PMX205 treatment, 10 mg/kg PMX205 (Tocris #5196, or synthesized and purified as previously described (Kumar et al., 2020)) was administered to mice orally flanking the irradiation doses. Vehicle control in these experiments refers to 20% ethanol/water. Apoptosis assessment by morphology in cell lines was carried out as previously described (Olcina et al., 2015;Olcina et al., 2020). Adherent and detached cells   were not included in the heatmap due to lack of CRISPR-Cas9 or RNAi data.
C4A, C4B, VSIG4, C8A, C8B, CD93 and CR1 were not included due to their very low expression in the majority of tissues.

46
CMS in rectal tumours. The Grampian cohort profiled by the S:CORT consortium was used. Pre-treatment rectal tumour biopsies from patients treated with radiotherapy and capecitabine were selected (N=129). Xcell array data was normalised and CMS called using the R tool CMScaller (Eide et al., 2017).
Patients gave consent to biobank their samples in the Grampian Biorepository