Tumour-bearing mice were topically treated with the gel containing either caerin 1.1 and 1.9 peptides (molar ratio 1:1; “caerin”) or a control peptide P3 (“control”) ( Figure 1A , top). The tumour weights were significantly reduced in the caerin group ( Figure 1A , bottom). To investigate the mechanism underlying this antitumor activity, the penetration of caerin 1.9 through the skin was assessed using confocal microscopy. At 5 min post the topical application, FITC-labelled P3 was mostly distributed on the epidermal layer ( Figure 1B ), while FITC-labelled caerin 1.9 was largely present beneath the basal cell layer with a significantly in-depth distribution ( Figure 1C and Figure S1 ). Additionally, the dynamic trace of green fluorescence in the cross section of the tissues showed that FITC-labelled caerin 1.9 penetrated to more areas under the basal cell layer ( Supplementary V1, V2 ). These results confirm our previous observation that caerin 1.1 and 1.9 are able to penetrate intact skin and lead to the growth inhibition of subcutaneously transplanted TC-1 cells (21).

Total viable CD45⁺ leukocytes were isolated from the untreated, caerin and control groups ( Figure 1D and Supplementary Data 1 ). The unsupervised graph-based clustering method detected a total of 23 distinct cell clusters (cluster “0” to “22”) ( Figures 1E, F and Figures S3, S4 ), and their identities were annotated based on the marker genes ( Supplementary File 1 ). The gene expression data from extracted CD45⁺ cells were aligned and projected in a two-dimensional space through t-stochastic neighbour embedding (t-SNE) for the identification of tumour-associated immune cell populations and the differentially expressed genes associated with different groups ( Figure S2 ). The presence of established canonical marker genes, such as Nkg7, Cd19, Fcmr, Cd8b1 and Cd79a, indicated the identities of lymphocyte lineages. Myeloid cells were supported by the identification of Cd209a, Adgre1, Itgax, Csf1r, Lgals3, Cd11c, Cd14, Cd68, Ccr2 and Ly6c2 (24, 25) ( Supplementary Data 2 ). The expressions of the top 5 marker genes of each cluster were compared, showing a relatively high overlap between clusters 0 and 5 with other clusters, respectively ( Figure S4 ).

Non-macrophage cells included high populations of monocytes (cluster 1; marker genes: Ly6a, Ly6c2, Fcgr1 and Dpep2) and natural killer (NK) cells (cluster 5; Nkg7, Gzmk, Klrc1 and Cxcr6) ( Figure 1F ). Two clusters were detected as neutrophils, i.e., cluster 9 (Stfa2l1, Retnlg, S100a9 and Asprv1) and cluster 16 (Serpinb2, Ace, Cd177 and Ifitm6). B cells (cluster 18; Cd19, Cd79a, Fcmr and Vperb3) had a small population. Three populations showed the signature of dendritic cells, including conventional DC type 2 (cDC2) (cluster 7; Plet1, Cd209a, Ctnnd2 and Epcam) (26), migratory DCs (migDC) (cluster 20; Ccl22, Bcl2l14, Fscn1 and Cacnb3) (27) and plasmacytoid dendritic cells (pDCs) (cluster 21; Siglech, Ccr9, Ly6d and Pacsin1) (26). Moreover, there were three clusters with gene signatures characterising the phenotypes of T cells (clusters 14, 15 and 22). Cluster 14 was assigned to CD4⁺CD25⁺ T cells, represented by Foxp3, Ctla4, Ikzf2 and Tnfrsf4, while cluster 15 corresponded to CD8⁺ T cells with the signatures of Lef1, Tcf7, Satb1 and Cd8b1. CD4⁺CD8⁺ T cells had the lowest number of cells, with the marker genes such as Cd226, Klrg1 and Cxcr6. Fibroblast (cluster 6; Nusap1, Top2a, Pclaf and Mki67) (28), adipogenic stem and precursor cells (ASPCs) (cluster 11; Col11a1, Plpp3, Col6a1 and Gas1) (23, 29), basal cells (cluster 12; Ccnb2, Cdkn3, Hmmr and Birc5) (30) and osteoclast (Cluster 17; Oscar, Ctsk and Mmp9) (31) were detected as possible contaminants (also see Supplementary Data 2 ).

A total of eight MΦ populations were present. The marker genes of cluster 0, including Pf4, Arg1, Pdpn and F13a1, were used to characterise the Arg1B MΦ in a previous study (32) ( Supplementary Data 2 ). Cluster 2 corresponded to a resident-like MΦ due to the high expression of Gas6, Stab1, Mrc1 and Folr2 (33). Cluster 3 exhibited mixed cell phenotypes, including proinflammatory MΦ (Cxcl10, Gbp2 and Thbs1), Ly6c^hi infiltrating MΦ (Chil3 and Plac8) and dendritic cells (Rsad2, Ifit1, Ifit2 and Ifi205). Thus, this cluster was labelled as MΦ/DCs. The significantly high expressions of MHCII members, such as H2-Ab1, H2-Aa, H2-Eb1 and H2-DMb1I, together with Cd74, suggested cluster 4 with an MHCII^hi MΦ phenotype (27). Cluster 8 showed marker genes depicting M1 MΦ-like phenotypes, including Ccl12 (34) and Cx3cr1 (35), as well as many genes playing roles in cell growth. Arg1A MΦ was assigned to cluster 10, with the signatures Arg1, Mmp12, Mmp13 and Lgals3 (32). The marker gene Ear2 was exclusive to cluster 13; it was thus assigned as Ear2^hi MΦ (36, 37). Cluster 19 was characterised by MΦ markers such as Lyve1, Cd209f and Cd163, while the marker gene Ccl8 was previously identified as the signature for TAM (38).

The sum of these MΦs represented the largest cell population, constituting 55.14% of the total cells in the untreated TC-1 tumour, and a similar fraction in mice treated with caerin (56.39%) or control gel (55.86%) ( Supplementary Data 2 ). It appeared that, in comparison to the control group, the caerin gel largely increased the proportions of Res-like MΦ (by 12.1%) and Arg1A MΦ (25.1%) ( Supplementary Data 1 and Figure S3C ). Notably, the population of Arg1A MΦ increased by 23.1% in the caerin group compared to the untreated group. The enrichment of TNFα signalling via NF-κB and IL-6/JAK/STAT3 signalling in the Arg1^hi MΦs of the control or untreated group, relative to the caerin group, was identified. Both pathways have been found to suppress the anti-tumour immune response in the TME, via enhancing proliferation, survival, invasiveness and metastasis of the tumour cells (39, 40). This suggested that the caerin gel reduced the immunosuppression of Arg1^hi MΦs in the TME of TC-1 tumour.

The average expression of the top five marker genes of each MΦ population was compared with respect to other genes across all MΦ cell populations ( Figure 2A ). The signature of Ear2^hi MΦ and TAMs appeared more exclusive, while the marker genes of Arg1B and MHCII^hi MΦ showed certain expressions in other MΦs. The correlation among these MΦs based on the expression of significantly upregulated genes demonstrated that TAMs correlated least with Arg1B, Ear2^hi and MHCII^hi MΦs ( Figure 2B ). Res-like MΦ was highly correlated with M1 MΦ, as were Ear2^hi with MHCII^hi MΦs. The proportions of the eight MΦs appeared similar in different groups ( Figure 2C ). However, the comparison of normalised cell numbers found significant elevations in Arg1B (fold change, FC = 1.29), Res-like (FC = 1.44), MHCII^hi (FC = 1.78) and Arg1A MΦs (FC = 2.11) in the caerin group, while TAMs were largely reduced to 0.63 and 0.47 with reference to the untreated or control group, respectively ( Figure 2D ). The top 20 enriched biological processes of these MΦ populations were compared ( Figure 2E and Supplementary Data 3 ). Immune response relevant processes, such as immune system process, cellular response to chemical stimulus and response to external stimulus, were highly detected in all MΦs, except Res-like and M1 MΦs. MHCII^hi MΦ played a considerable role in professional antigen processing and presenting.

Arg1^hi MΦs are known to deplete L-arginine locally, to assist wound healing and tissue fibrosis, and have been considered immunosuppressive and tumorigenic in certain tumours (41, 42). Arg1 was highly expressed in four MΦ clusters, including MΦ/DCs, Arg1A, Arg1B and Ear2^hi MΦs ( Figure 3A ). Projected to a two-dimensional tSNE space, Ear2^hi MΦ was distributed more separately from the other three MΦs, indicating possible functional specificity ( Figure 3B ). MΦ/DCs had the highest number of total (704) and unique (447) genes significantly upregulated ( Figure 3C ). The KEGG pathway analysis revealed that apoptosis was highly enriched in MΦ/DCs, Arg1B and Ear2^hi MΦs (q-value < 0.0017) ( Figure 3D and Supplementary Data 3 ). MΦ/DCs were more enriched with immune response relevant pathways than other MΦs, such as chemokine signalling, B cell receptor and TNF signalling. Ferroptosis, autophagy and mitophagy pathways were detected with significance only in Arg1A MΦ, while Ear2^hi MΦ was comparatively more enriched with HIF-1 signalling and glycolysis/gluconeogenesis.

The GSEA analysis found that interferon α and γ responses, G2M checkpoint and E2F targets were most enriched in Arg1B MΦs of the caerin group, whereas glycolysis, oxidative phosphorylation, TNFα signalling via NF-κB and epithelial–mesenchymal transition were most enriched in that of the control group ( Figure 3E and Supplementary Data 4 ). Five macrophage-associated genes, Fcrls, Lars2, Ccl4, Sgk1 and Nfkbia, were upregulated (adj. p < 0.05) in the caerin group. MYC targets v2 and INFα response became more pronounced in Arg1A MΦs ( Figure 3F ), while the latter was also enriched in MΦ/DCs ( Figure S5A ). The EC value of inflammatory response was similar for the caerin and control groups. INFα response was the only pathway enriched in Ear2^hi MΦ by the caerin gel ( Figure S5B ). Notably, the expression of Cebpb was remarkably upregulated in MΦ/DCs and Ear2^hi MΦ of the caerin group.

Three DC populations were identified, including cDC2, migDC and pDC. The normalised expressions of the top five marker genes revealed exclusive signatures of migDCs and pDCs, implying distinct phenotypic and functional properties of these two clusters ( Figure 4A ). Since MΦ/DCs also showed dendritic cell signatures, such as the high expression of Thbs1, Adam8 and Ccr2, it was included in the comparative analysis with the DCs. cDC2s expressed Tlr1, Tlr3, Tlr5 and Tlr6, while pDCs preferentially expressed Tlr1, Tlr7, Tlr9 and Tlr12. MΦ/DCs had higher expressions of Tlr2, Tlr4, Tlr5, Tlr6, Tlr8 and Tlr13 ( Figure 4B ). The expression of many TLRs was absent or reduced in migDCs, except for Tlr13 in the untreated and control groups. Thus, the cDC2s and pDCs should sense and respond to different innate immune stimuli. Genes encoding chemokines, chemokine receptors and cytokines exhibited distinct expression patterns in the three DCs with different treatments ( Figure 4B ). cDC2 exclusively expressed Il1r, Il18rap and Ifngr1, whereas migDCs had comparatively increased the expressions of Mmp25, Il15, Ccl22, Ccl17, Ccl5, Cx3cl1, Ccr7, Cxcr5 and Il15ra. With respect to pDCs, Ccl25, Ccl4 and Cxcr3 displayed elevation in the control group. In contrast, caerin gel significantly lowered the expression of Ccl4. The expression of several other chemokines and receptors was shared between cDC2s and migDCs, including Cxcl9, Cxcl16 and Ccr6, which were absent from pDCs. The unique repertoires of chemokines and chemokine receptors expressed by these three DCs indicated divergent functions modulated by the treatments. The expressions of genes playing roles in antigen processing, such as Fcer1g, Ctss and Cd86, were increased more in cDC2 cells. The enrichment analysis of the KEGG pathway identified Th1, Th2 and Th17 cell differentiation, and C-type lectin receptor signalling pathways significantly represented in cDC2s and migDCs ( Figure 4C ).

The trajectory analysis revealed five developmental states in these four DC clusters, where MΦ/DCs emerged at an early pseudotime with a 100% proportion of State-5, as well as on the branch of State-4 ( Figure 4D ). It joined cDC2s (8%), migDCs (1%) and pDCs (0.5%) at State-3, which developed into two states with cDC2 contributed to the entire State-1. The migDCs and pDCs only appeared in a branch at a late developmental stage of DCs represented by similar expression patterns of cytokines/chemokines and their receptors (incl. Ccl3, Tlr4, Mmp13, CCr5 and Cxcl2). The three groups showed cell distributions on the entire trajectory ( Figure 4D ). The caerin gel induced an early appearance of State-1 and 4 and slightly expanded State-2 and 3. The marker genes with similar expression patterns along pseudotime were thus clustered to unravel the function of these states and their function ( Figure 4E ). States-1 and 3 were closely correlated and then hierarchically clustered with State-4, with the high expression of genes related to TNF signalling at late pseudotime, such as Tnfsf4, 9, 10 and Tnfrsf18. The genes with antigen-presenting and processing functions were expressed at an early time of State-2 (Clec4a2, Clec4e and Clec4d) and 5 (Fcgr1 and Fcgr4), yet late for State-1 (Cd86) ( Supplementary Data 5 ).

The caerin group had the highest number of differentially expressed genes in migDCs, compared to the other two groups ( Supplementary Data 2 ). The FC and p-values of the top 40 significantly regulated genes were summarised in Figure S6A . The upregulation of four genes relevant to the activation of NF-κB signalling was detected, including Bcl3 (43), Tradd (44), Mtdh (45) and Ncoa6 (46), indicating more pro-inflammatory migDCs in the caerin group. The overall gene expression of the four DCs was compared, which showed that Ly6a, Tmed7, H2-Oa, Bag1 and Cebpb were significantly elevated by the caerin gel ( Figure S6B ). The pathways more enriched in the DCs of the caerin group included antigen processing and presentation IL-17, CAMs and MAPK signalling pathways ( Figure S6C and Supplementary Data 5 ).

The two treatments regulated the expression of genes in NK cells differently compared to the untreated group; the top five maker genes of the caerin group included Tnfaip3, Phf20l1, Hnrnpab, Cdkn1B and Gnas ( Figure S6A ). A total of seven NK subpopulations (C0 to C6) were identified ( Figure 5A ). The subpopulation C0 and C5 of the caerin group showed higher proportions and were enriched in immune response-relevant pathways, such as T cell receptor signalling, cytokine–cytokine receptor interaction and Th17 cell differentiation ( Figure S6B ). C0 was characterised by the marker genes Cd4 (47), Cd40lg (48) and Icos (49) ( Figure 5B ), which represented the phenotype of active NK cells (C0_ActiNK). The C1 displayed significantly higher expressions of Gzma, Gzmk, Klrc1 and Nkg7, indicating the identity of mature NK cells (C1_MatNK). Many marker genes associated with C2 were involved in interferon induction and activation (Ifit1, Ifit3b, Ifit3, Isg20, Igtp, Ifi204 and chemokine encoding gene Cxcl10), indicating the stimulation of NK cells by interferons ( Supplementary Data 6 ). In addition, C2 appeared to associate with positive regulation of inflammatory responses, as indicated by the high expression of Cd69 and the enrichment of IFN-γ signalling, toll-like receptor signalling and the pathways related to viral infections. Thus, this subpopulation was referred to as C2_InflamNK.

Many marker genes of C3 appeared relevant to stress response (Hspa1a, Hspa1b, Bcl2 and a few mitochondrial enzymes), while few genes appeard directly relevant to the immune functions of NK cells, suggesting the likelihood of a transitional state, which was thus named transitional NK (C3_TransNK) ( Figure 5B and Supplementary Data 6 ). Xcl1 and Sell were previously identified as the signatures for human CD56^bright NK (50); in addition, the exclusive expression of Ncr1 (i.e., NKp46/Cd335) to C4 was detected, and thus an identity of CD335^bright NK cells could be assigned to the fifth subpopulation (C4_Cd355briNK). Blk was only detected as the marker for the C5, and it was found tumour suppressive in chronic myeloid leukaemia stem cells (51) and oncogenic in cutaneous T-cell lymphoma (52). Moreover, the high expressions of Cd163l1, Il17re, Rorc and Il7r implied the phenotype of Rorc⁺CD127⁺ NK cells, which exhibited adaptive immune features (53). Thus, this subpopulation was labelled as C5_AdapNK. The C6 exclusively expressed Trem1, Csf3r and Cxcr2, and highly expressed Alox5ap and Ccr1. This implied that these NK cells were associated with tissue residency (C6_ResNK).

The trajectory analysis showed a developmental progression with several states separated by four branch points (a, b, c and d) ( Figure 5C ). The percentage contribution of each subpopulation within each state was compared, and States-2, 3, 4, and 7 cells were dominated by C0_ActiNK. C5_AdapNK cells were found to mainly develop in States-2, 7 and 3, while its proportion decreased along the pseudotime. Caerin gel caused a more scattered cell distribution at State-1 and a higher State-7 population ( Supplementary File 1 ); in addition, there were more cells distributed in the transition from State-4 to State-6 ( Figure 5C ). Developing from branch point a, State-2 showed high expressions of the genes (Nfkbia, Jun, Tnfaip3, Fos, Cxcl2, Junb, Icos and Ltb) closely associated with immune response, such as the signalling of TNF, T cell receptor and NF-κB ( Figure S7 and Supplementary Data 6 ). Many Rps and Rpl genes were abundant in the cells of State-7 and State-3 originated from branch points b and c, respectively, suggesting an active state of mRNA processing and translation. Moreover, State-3 cells appeared more proinflammatory, due to the enrichment of TNF-α/NF-κB signalling supported by the upregulation of Rpl30, Rack1, Rpl8, Rps11 and Rps13. Divided by branch point d, State-5 cells had high expressions of Gzmb, Plac8, Ly6a, Ly6e and several Ifit genes, which were the signatures associated with interferon and cytokine-mediated signalling. In contrast, the elevation of the genes directly signifying functional NK cells, including Klrd1, Klrc1, Xcl1, Gzmk and Nkg7, was present in State-6.

C1_MatNK, C2_InflamNK and C3_TransNK were dominated by one cell state ( Figure S8A ). In the caerin group, the C1_MatNK and C2_InflamNK cells were visible at higher proportions in State-6 and State-5, respectively, which included more genes related to functional NK cells ( Figure 5D ). There was a substantial proportion of State-3 cells detected in C5_AdapNK of the caerin group, further indicating transcriptome alteration in adaptive immunity. The expression of the top five significantly upregulated genes in each subpopulation of the caerin group showed genes exclusively expressed in C4_Cd355briNK (Lrrc8c, Nol11, Asb8 and Sltm) and C6_ResNK (Lmbrd1, Gdi1, Dusp5, Taok3 and Tm9sf3), with respect to the untreated group ( Figure S8B ). The NK cell marker Gzmb was upregulated significantly in all subpopulations except C1_MatNK of the caerin group ( Figure S8C ).

Several genes associated with neutrophil activation (Rap1B, Bri3, Rab24, Psmc2 and Pdap1) were significantly upregulated in NK cells of the caerin group, while Nlrc5 and Commd7, negatively regulating NF-κB transcription factor activity, were downregulated with respect to the control ( Figure 5E ). Fam49b was significantly upregulated in all the subpopulations of the caerin group, and its role as a tumour suppressor in pancreatic ductal adenocarcinoma cells via regulating mitochondrial fission was found (54). The upregulation of Ppp3ca in most subpopulations suggested a higher-level induction of NFAT signalling (55). The GSEA analysis revealed that gene sets associated with T cell receptor signalling and antigen processing and presentation were enriched in C0_ActiNK of the caerin group, while C3_TransNK was enriched with natural cell activation and cell killing ( Figure 5F ). Moreover, the regulation of cytokinesis and INFγ-mediated signalling pathway was respectively enhanced in C4_Cd355briNK and C5_AdapNK. Inactivation of the MAPK activity pathway was enriched in C6_ResNK. Two interferon stimulation-associated genes (Bst2 and Isg20) were largely expressed in the caerin group, especially in C2_InflamNK. In terms of KEGG pathways, natural killer cell-mediated cytotoxicity, IL-17 signalling pathway and Th1 and Th2 cell differentiation were enriched in C0_ActiNK, with the relevant genes significantly upregulated by caerin gel in comparison to the control ( Figure 5G ). More activated T cell receptor signalling and more pronounced antigen processing and presentation were present in C2_InflamNK and C6_ResNK, respectively ( Figure S9 ).

In the caerin group, the proportion of CD8⁺ T cells was 2.21%, which was remarkably higher than that of the untreated or control group ( Supplementary Data 1 ). CD8⁺ and CD4⁺CD25⁺ T cells were more closely correlated with NK cells than CD4⁺CD8⁺ T cells in a 2D tSNE space ( Figure S10A ). Taking these four cell populations together, CD8⁺ T cells had a much higher proportion of nearly 30% in the caerin group ( Figure 6A ). The expression of selected marker genes of CD8⁺ T cells was also detected mainly in other T cells and NK cells, including those positively regulating the activation of CD8⁺ T cells, such as Tcf7 (56), Lef1 (57), Satb1 (35), S1pr1 (58) and Txk (56) ( Figure 6B ). The expressions of Nsg2, Tcf7, Satb1, Lef1 and S1pr1 were almost exclusively and highly expressed in CD8⁺ T cells ( Figure 6C ). Ccr7 showed a higher expression in B cells and migDCs, while Itk exhibited an elevation in NK and B cells. The expression (in Log₂ value) of nine genes related to the regulation of T cell activation was statistically compared among the three groups ( Figure 6D ). Lef1, Sell, Txk, Tpt1, Lat, Itk and S1pr1 were upregulated, whereas Ccr7 and Ms4a4b were downregulated. Several Gimap genes, previously shown to be involved in lymphocyte development, or associated with inflammatory and autoimmune diseases (59), were detected as the marker genes for CD8⁺ T cells, including Gimap1, 3, 4, 5 and 6 ( Figure 6E ). It showed that more cells expressed higher levels of these Gimap genes, especially Gimap6, in the caerin group. The GSEA analysis revealed that the signalling of B cells, T cells, chemokine and toll-like receptor as well as natural killer cell-mediated cytotoxicity KEGG pathways were enriched in the caerin group with respect to the control group ( Figure 6F ). The Reactome pathways enriched by the unique marker genes of CD8⁺ T cells of the caerin group included signalling of several FGFR mutant-related pathways, interleukins and cytokines in the immune system ( Figure S10B ). This further suggested a potentially more activated state of CD8⁺ T cells induced by the caerin gel treatment.

The quantitative proteomic analysis showed that more proteins were significantly regulated by the caerin gel, with a total of 48 proteins upregulated uniquely, including those related to immune response (including Stat1, Gzma, Ifit1, Ifit3, Lfih1, Tap1 and Cd72) ( Figure 7A ; Supplementary Data 7 and Figure S11 ). Several upregulated proteins were exclusively correlated with the genes (Ifih1, Tgtp2, Gbp2 and Ifit3) highly expressed in monocytes identified by the scRNA-seq analysis ( Figure 7B ). The high quantities of several proteins (H2-Q7, Tap1, Tapbpl, Gzma and Stat1) were closely correlated with their gene expression in T cells, DCs and NK cells. The exclusive expression of Hp and Serpinb2 to neutrophils was detected. The gene expression of H2-D1 was significantly upregulated in all cells, as well as the diverse distribution of Tapbp, Stat1 and Zbp1, implying that they were largely regulated by the caerin gel.

The protein–protein interaction (PPI) analysis of upregulated proteins identified intensive interactions only present in the caerin group ( Figure 7C and Figure S13 ). Stat1 was found as the node with the highest degree, while it was a marker gene for monocytes, NK cells (also C2_InflamNK), CD4⁺CD25⁺ T cells and migDCs. The gene ontology enrichment analysis of this PPI network revealed biological processes related to the regulation of immune response, such as defence response (FDR = 1.52E-13), immune system process (FDR = 5.12E-13) and innate immune response (FDR = 6.72E-12) ( Supplementary Data 8 ). More specifically, the positive regulations of INFα/β secretion (supported by Gbp5, Iigp1, Stat1, Ifit1, Ifit3, Ifi204 and Gbp2), response to cytokine stimulus (Ifih1, Zbp1, Parp14, Ddx58 and Parp9) and antigen processing and presentation of peptide antigen via MHC class I (Tapbp, H2-Q7, Tap1, and H2-D1) were largely enhanced by the caerin gel ( Figure 7D ). In terms of molecular function, the binding of TAP, immunoglobulin/receptor and T cell receptor and antigen processing and presentation were enriched in the caerin group relative to the other two groups.

The FC values of proteins differentially regulated by caerin and control were compared ( Figure 7E ); many upregulated proteins (such as C9, C8b, C8g, Igj, vtn and Serpina3n) appeared to play roles in the processes related to immune response and complement activation. This was in accordance with the identification of more immune activity of MΦs and DCs, as well as the recruitment of activated CD8⁺ T cells in the TME of the caerin group, as suggested by scRNA-seq. The KEGG pathways enriched in upregulated proteins were thus highly associated with signalling in immune response and process ( Supplementary Data 8 ), such as antigen processing and presentation (FDR = 6.0E-4), the signalling of NOD-like receptor (FDR=6.0E-4) and natural killer cell-mediated cytotoxicity (FDR = 7.7E-3).

Six-to-eight-week-old, specific pathogen-free adult female C57BL/6 (H-2b) mice were ordered from the Animal Resource Centre of Guangdong Province and kept at the Animal Facility of the First Affiliated Hospital of Guangdong Pharmaceutical University. Experiments were approved by Animal Experimentation Ethics Committee (Ethics Approval Number: FAHGPU20160316). Five mice were kept in each cage on a 12-h light/darkness cycle (22°C and 75% humidity), provided with sterilised standard mouse food and water. TC-1 tumour-bearing mice were i.p. injected with 1% sodium pentobarbital. Mice were euthanised by CO₂ inhalation at the end of the experiment.

A murine TC-1 cell line transformed with HPV16 E6/E7 was obtained from Shanghai Institute for Cell Resources Centre and cultured following the protocol described elsewhere (21). Caerin 1.1 (GLLSVLGSVAKHVLPHVVPVIAEHL-NH₂), caerin 1.9 (GLFGVLGSIAKHVLPHVVPVIAEKL-NH₂) and the control peptide P3 (GTELPSPPSVWFEAEFK-OH), were synthesised (purity>99%) (Mimotopes Proprietary Limited, Wuxi, China). The lipopolysaccharide concentrations of caerin 1.1, caerin 1.9 and P3 were 0.03, 0.03 and 0.44 EU/ml respectively, as measured by Kinetic Turbidimetric Assay (Xiamen Bioendo Technology Co., Ltd).

Poloxamer 407 (WPAK592B) and poloxamer 188 (WPAK539B) were purchased from Badische Anilin-und-Soda-Fabrik (Ludwigshafen, Germany). The gel was prepared as previously described (21). Briefly, 46 g of poloxamer 407 and 10 g of poloxamer 188 were dissolved in 200 ml of distilled water, and caerin 1.1 and caerin 1.9 were then added. The solution was mixed thoroughly and filtered through a 0.22-μm membrane filter to prepare a 20-mg/ml gel and stored at 4°C.

TC-1 cells, at approximately 70% confluency, were harvested with 0.25% trypsin-EDTA solution and washed with PBS. 5 × 10⁵ cells/mouse in 0.2 ml of PBS were injected subcutaneously into the left flank. TC-1 tumour-bearing mice were either treated with caerin gel or control gel or left untreated for 7 consecutive days by applying 20 μl of each gel onto the shaved skin surface above the tumour. Two days after the final treatment, mice were sacrificed, and the tumours were isolated and weighed.

Six-week-old female C57BL/6 mice were anesthetised by intraperitoneal injection of pentobarbital before hairs on the dorsal side of the ear were shaved and gently wiped with normal saline and dried naturally. The mice were treated with thermosensitive gels containing FITC-labelled caerin 1.9 or P3, by evenly applying 100 μl of the gels to the shaved areas for 5 min. The ear skin surface was then gently rinsed with distilled water three times to remove any residual gels and then cut, and sections of 6-μm thickness were stained with DAPI. The samples were observed using 405 and 488 nm by a Zeiss LSM 880 Airyscan confocal microscope (Zeiss, Germany). Dual-channel combined imaging videos were recorded.

TC-1 tumours were cut into 2 × 2 mm pieces; digested in 2.35 ml of RPMI 1640, 100 µl of Enzyme D, 50 µl of Enzyme R and 12.5 µl of Enzyme A into a gentleMACS C Tube; and disassociated using gentle MACS Dissociator from Miltenyi (Gladbach, Germany). After the removal of dead cells and cell debris, the remaining cells were labelled with CD45 microbeads (130–110–618). The viability of the CD45⁺ cells were more than 80% of total cells confirmed by flow cytometry and trypan blue staining. Cells were washed once with ice-cold PBS containing 10% foetal bovine serum post sorting and counted using a hemocytometer. After that, the cells were loaded to a 10x chromium machine (10x Genomics, San Francisco, CA) and run through the library preparation procedures following guidance from the Chromium Single Cell 3′ Reagent Kits v2 (more details were provided in the Supplementary Methods ).

The tumour samples were the same biological triplicates from which the CD45⁺ cells were extracted for scRNA-seq. Certain amounts of samples containing 500 µg of protein were subjected to trypsin digestion by the filter-aided proteome preparation (FASP) described elsewhere (115) followed by LC-MS/MS analysis using a Q Exactive hybrid quadrupole-orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) (see Supplementary Materials for detailed method).

Interactions among significantly regulated proteins were predicted using STRING (116). A required interaction score of 0.700 was selected for all PPI, to highlight the most confident interactions. Neither the first nor second shell of the PPI was included. Protein without any interaction was excluded.

The gene ontology terms, including biological process, molecular function and cellular component, were annotated using STRING, and the enrichment of was analysed accordingly. The enrichment of KEGG pathways (117) and Reactome pathways (118) was assessed based on the significantly upregulated genes (p < 0.05) of different cell populations/subpopulations. The genes differentially expressed in three groups were analysed by Gene Set Enrichment Analysis (GSEA) with p < 0.05 using GSEA v4.1.0 (119).