Human or mouse CD47-IgV-Domain proteins were labeled with NT-647 dye following the microscale thermophoresis (MST) assay protocol, and the labeled protein used for each assay was about 50 to 200 nM. Unlabeled peptides or ligand protein solution was diluted for appropriate concentration serially in buffer (50 mM Tris, 150 mM NaCl, 10 mM MgCl₂, 0.05% Tween-20, 0.5% DMSO and pH 7.5). The labeled proteins were then added to peptides or ligand protein solutions by 1∶1 (v/v). The samples were loaded into silica capillaries after incubation at room temperature for 30 min. Binding constants were determined by the MST assay according to manufacturer’s instructions with 20% or 40% light emitting diode (LED) power, 40% MST power at 25°C (Nano Temper, Monolith NT.115, Germany). Data analysis were performed by using a Nanotemper Analysis software.

Human or mouse CD47-IgV-Domain-hIg fusion proteins (20 nM, Sino Biological, China), or biotinylated human CD47 protein (20 nM, ACROBiosystems, USA) was incubated with titrating concentrations of peptides or anti-human CD47 antibody (B6H12) in phosphate-buffered saline (PBS, pH 7.4) for 1 hour at 4°C. The mixture was incubated with human or mice SIRPα overexpressed Chinese hamster ovary cells for 30 min at 4°C, washed to remove unbound mixtures and phycoerythrin (PE)-conjugated goat anti-human IgG or PE-conjugated streptavidin was incubated (eBioscience, USA). Cells were then analyzed by flow cytometry (BD Biosciences, USA). The IC₅₀ was determined from sigmoidal dose response curves illustrated by GraphPad Prism.

Murine bone marrow cells were isolated from 7 to 11 weeks old C57BL/6 or BALB/c mice. Human peripheral blood mononuclear cells were collected from venous blood of healthy volunteers, diluted with 2×PBS (pH 7.4) and separated with Ficoll density gradient. The cells were cultured in DMEM (GIBCO, USA) supplemented with 10% fetal bovine serum and 20 ng/mL granulocyte macrophage colony-stimulating factor (GM-CSF) or macrophage colony-stimulating factor (M-CSF, Peprotech, USA) for 7 days. Meanwhile, medium was replaced with fresh medium containing cytokine, and then the adherent cells were harvested. Phagocytosis assays were performed by co-culture of macrophages with carboxyfluorescein succinimidyl ester (CFSE⁺) or green fluorescent protein (GFP⁺) tumor cells at a 1:4 ratio in serum-free medium at 37°C for 4 hours in low-attachment 96-well tissue culture plates (Corning, USA). The cells were harvested, and primary macrophages were identified by flow cytometry using anti-F4/80 or anti-CD14 antibody (eBioscience, USA). Then, 7-AAD (eBioscience, USA) was added to exclude dead cells in some experiments. Phagocytosis rate was determined as the percentage of CFSE⁺ or GFP⁺ macrophages.

C57BL/6 mice were subcutaneously (s.c.) injected with 1×10⁶ MC38 cells or 2×10⁵ B16-OVA into the flank, and BALB/c mice were s.c. injected with 2×10⁵ CT26 cells into the flank. Tumor volumes were measured every other day by length (a), width (b) and height (c), and calculated as tumor volume=abc/2. Tumors were grown for approximately a week, the mice were then injected s.c. with 2 mg/kg pep-20 or normal saline as the negative control at the peritumoral site every day. Treatment was continued for 2 weeks, then the mice were euthanized, tumors, spleens and draining lymph nodes were dissected. For macrophages depletion, C57BL/6 mice were intraperitoneally (i.p.) injected with 150 µL clodronate liposome or control liposome (FormuMax Scientific, USA) on day 6, and every 4 days after injection of 1×10⁶ MC38 cells until finishing the experiment. The efficiency of macrophages depletion was determined by flow cytometry analysis of the CD45⁺ CD11b⁺ F4/80⁺ cells. For pep-20-D12 treatment, 1×10⁶ MC38 cells were injected s.c. into the flank of mice, and 2 mg/kg pep-20-D12 every day or 400 µg anti-mouse CD47 antibody (miap301, Bio-XCell, USA) every 3 days for a total of five times as the positive control was injected i.p. into the mice, with normal saline and ratIg as the negative controls. For radiotherapy (RT), 1×10⁶ MC38 cells were injected s.c. into the flanks of mice, tumors were allowed to grow to reach 80 to 100 mm³ before being treated with RT. Tumors locally received one 20 Gy dose IR and then 2 mg/kg pep-20-D12 was injected i.p. into the mice every day for 2 weeks.

For CD8⁺ T cells detection in tumors, single cell suspensions from tumor tissues were prepared by gentle mechanical disruption, the tumor masses were then digested with collagenase IV (Invitrogen, USA) and Dnase I (Sigma, USA) for 30 min at 37°C, and cells were passed through a 70 µm nylon cell strainer. Tumor cells were stained with anti-mouse CD45, anti-mouse CD3, anti-mouse CD8α or isotype control, and then analyzed by flow cytometry.

For intracellular cytokine staining assay, tumor-infiltrating lymphocytes (TILs) were purified using Ficoll-gradient centrifugation (GE, USA). Single cell suspensions from TILs, draining lymph nodes or spleens were re-stimulated with 20 ng/mL phorbol 12-myristate 13-acetate (PMA, Sigma, USA) and 1 µM ionomycin (Sigma, USA) for 4 hours to CT26 and MC38 tumor-bearing mice, or 10 µg/mL OVA_257-264 peptide for 6 hours to B16-OVA tumor-bearing mice, in the presence of protein transport inhibitor cocktail (eBioscience, USA). Interferon (IFN)-γ production from CD8⁺ T cells was detected with anti-mouse CD3, anti-mouse CD8α, and anti-mouse IFN-γ or isotype control, and then analyzed by flow cytometry.

For ELISA assay, spleen or draining lymph node cells were re-stimulated with 0.5 µg/mL anti-CD3 and 0.5 µg/mL anti-CD28 antibodies (eBioscience, USA) for 3 days to CT26 and MC38 tumor-bearing mice, or 10 µg/mL OVA_257-264 peptide for 5 days to B16-OVA tumor-bearing mice. Cellular supernatant IFN-γ secretion was measured by using an ELISA kit (eBioscience, USA).

Statistical analysis were performed with unpaired Student’s t-test for analyzing differences between groups. The Kaplan-Meier curve and the log-rank test were used for analyzing overall survival rate of the mice. Data were represented as means±SEM unless otherwise indicated. *p<0.05, **p<0.01 and ***p<0.001 were considered statistically significant.

The CD47 expression in different tumor tissues from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) data sets was evaluated. It was found that CD47 expression was markedly higher in various kinds of tumor tissues compared with normal counterparts (online supplemental figure S1A). Consistently, CD47 was expressed at high levels on various solid and hematologic tumor cell lines shown by flow cytometry analysis (online supplemental figure S1B). As previous studies have shown,21 the high expression of CD47 in tumors suggested the potential of targeting CD47 for antitumor immunotherapy.

To identify the peptides binding to human CD47, the solution-phase biopanning was performed with affinity beads capture strategy, which improved accessibility of the protein-binding site according to the standard protocol of the Ph.D.-12 phage display peptide library using human CD47-IgV-Domain protein as the target.22 Phages that are not specifically binding with CD47 can be avoided by employing a negative selection strategy. After five rounds of bio-panning, phages of enrichment were obtained and phage plaques were randomly selected for DNA sequencing. Subsequently, the sequences were aligned using Clustal Omega,23 and several consensus sequences were apparently different, such as ‘YKEHYLY’ and ‘TXSNY’ (online supplemental figure S2). This indicated that the binding regions of these phages to CD47 might not be the same cluster. Therefore, in order to reduce the ‘off-rate’, peptides with different common sequences were accepted. Based on the selection strategy, 12 candidate peptides were further screened to identify which could bind to human CD47 (online supplemental table S1).

To assess the affinity of these candidate peptides toward human CD47, the preliminary binding assay was performed. Several peptides were confirmed to bind human CD47 with a relatively strong affinity comparable to the native CD47/SIRPα interaction, such as pep-3 and pep-20. In addition, due to the presence of consensus sequences, it can be hypothesized that the peptides pep-5, pep-9 and pep-19 may have higher affinities. But due to the poor solubility, it’s not available to use the higher titrating concentrations to determine the saturation binding concentration (online supplemental table S1). Subsequently, pep-20 was examined to bind to human and mouse CD47 with the K_D values of 2.91±1.04 µM and 3.63±1.71 µM, respectively (figure 1A,B).

To evaluate the ability of the peptide candidates to block the CD47/SIRPα interaction, the blocking assay was performed in vitro. Three peptides (pep-15, pep-16 and pep-20) could block the interaction of CD47/SIRPα (online supplemental figure S3), among which pep-20 exhibited the greatest dose-dependent effect with an IC₅₀ of 24.56 µM (figure 1C). The anti-CD47 antibody (B6H12) served as a positive control with an IC₅₀ of 6.94 µg/mL (online supplemental figure S4). The same experiment was also performed in a mouse blocking system, and the blockade of pep-20 to mouse CD47/SIRPα interaction was determined with an IC₅₀ of 12.03 µM approximately (figure 1D). These results clearly demonstrated that pep-20 could bind to CD47 and block the interaction of CD47/SIRPα.

Previous studies showed that phagocytosis of macrophages was the major mechanism of CD47 targeted therapies.8 Thus, the in vitro capability of pep-20 to facilitate macrophages-mediated phagocytosis of tumor cells was investigated. Three different human tumor cell lines of MCF7, HT29 and Jurkat were offered to macrophages as target cells, and the results showed that pep-20 significantly increased macrophages-mediated phagocytosis of tumor cells, which was slightly inferior to the anti-CD47 antibody (B6H12) positive control (figure 1E–G). Subsequently, whether the mouse macrophages possessed efficacy of phagocytosis towards murine tumor cells in the presence of pep-20 was also investigated. As a result, similar phagocytosis occurrences were observed from CT26, MC38 and B16-OVA tumor cells treated with pep-20 (figure 1H–J). To confirm the specificity of pep-20 targeting CD47, the CD47 knockdown HT29 and MC38 tumor cells were established. CD47 knockdown impairs the efficacy of pep-20 to induce macrophage-mediated phagocytosis of tumor cells (online supplemental figure S5). These results demonstrated that pep-20 might have potential therapeutic activity in tumor-bearing mice.

To further verify the ability of pep-20 to block CD47/SIRPα interaction, the SIRPα tyrosine residues phosphorylation with pep-20 intervention was evaluated.24 As a negative regulation signal of CD47/SIRPα axis, the ligation of SIRPα by CD47 could lead to tyrosine phosphorylation of immunoreceptor tyrosine-based inhibitory motifs (ITIMs) in its cytoplasmic domain.25 For the validity of the experimental results, SIRPα expression on several murine tumor cells was evaluated and it was found that SIRPα was not detected on CT26 cells. Following incubation of mouse bone marrow-derived macrophages with CT26 cells, it was found that pep-20 could reduce the tyrosine residues phosphorylation of SIRPα, which was consistent with that of anti-CD47 antibody (miap301) (online supplemental figure S6). The results clearly demonstrated that pep-20 could directly block the CD47/SIRPα interaction.

The antitumor efficacy of pep-20 in tumor-bearing mice was subsequently investigated. MC38 cells were engrafted s.c. on the flank of mice. Tumor growth was significantly suppressed by pep-20 treatment (2 mg/kg), and mice displayed prolonged overall survival compared with the negative control of normal saline (figure 2A,B). To confirm these results, the antitumor efficacy of pep-20 was also evaluated using similar treatment program in CT26 tumor-bearing mice. The delay of tumor growth and prolonged overall survival were also observed (online supplemental figure S7A,B).

To determine whether pep-20 treatment engages antitumor immune response, the cells from the tumor tissues, tumor-draining lymph nodes and spleen were obtained after treatment. CD47 blockade with pep-20 remarkably increased the intratumoral CD8⁺ T cell population (figure 2C). In addition, IFN-γ expressing CD8⁺ T cells from tumor-draining lymph nodes and spleen were significantly increased in pep-20 treated MC38 tumor-bearing mice (figure 2D–G). These results were also confirmed in CT26 model (online supplemental figure S7C-G).

Based on these results, we wondered whether pep-20 could enhance tumor antigen specific T-cell immune response. To verify this hypothesis, B16-OVA tumor-bearing mice was established. Consistently, pep-20 treatment could significantly suppress tumor growth (figure 3A). The cells from the tumor-draining lymph nodes and spleen were stimulated with OVA_257-264 peptide ex vivo after treatment. Pep-20 significantly increased the CD8⁺ T cell population in tumor tissue (figure 3B). Identically, IFN-γ expressing CD8⁺ T cells (stimulated by antigenic peptide OVA_257-264) in major peripheral lymphoid organ from tumor-bearing mice were also increased after pep-20 treatment, especially in draining lymph nodes (figure 3C–F). Collectively, these results demonstrated that blockade of CD47 with pep-20 could delay the tumor growth and induce antigen-specific T-cell immune response.

Based on the results described above, it can be presumed that macrophages could mobilize the T-cell response after phagocytosis of tumor cells by pep-20. To further verify the effects of pep-20-mediated phagocytosis by macrophages to CD8⁺ T-cell response, the proliferation of OT-I T cells was evaluated in vitro. As predicted, macrophages could significantly enhance the activity of CD8⁺ T cells after engulfing B16-OVA cells in the presence of pep-20, which was consistent with the anti-CD47 antibody (miap301) (online supplemental figure S8A,B). In addition, more secretion of IFN-γ from CD8⁺ T cells in the culture supernatants was detected in pep-20-treated group (online supplemental figure S8C). These results demonstrated that the phagocytosis of macrophages enhanced by pep-20 was able to stimulate antitumor CD8⁺ T-cell immune response.

To further demonstrate whether macrophages directly contribute to the antitumor effects of pep-20, macrophages were depleted with clodronate liposome before injection of pep-20 in MC38 tumor-bearing mice.26 Prior to the experiment, the macrophages were confirmed to be removed effectively in peripheral blood and spleen compared with control liposome (online supplemental figure S9). The detailed description of the treatment program is presented in figure 4A. By monitoring tumor sizes of MC38-bearing mice, no obvious difference was observed between pep-20 and normal saline groups in the presence of clodronate liposomes. In contrast, tumor growth was markedly suppressed by pep-20 compared with normal saline group in the presence of control liposomes (figure 4B). Overall, these findings indicated that macrophage depletion was able to impair the therapeutic efficacy of pep-20.

To further verify whether depletion of macrophages could attenuate the effects of pep-20 on antitumor immune response, the activity of CD8⁺ T cells after clodronate liposomes treatment was studied. When the macrophages were depleted, a significant reduction in the percentages of intratumoral CD8⁺ T cells was observed with pep-20 treatment, compared with the mice without macrophages depletion (figure 4C). Additionally, the IFN-γ-expressing CD8⁺ T cells from the draining lymph nodes and spleen following stimulation with PMA and ionomycin ex vivo were examined. The secretion of IFN-γ from CD8⁺ T cells showed no significant differences with or without pep-20 treatment after macrophages depletion despite the increase of CD8⁺ T cells population in the tumor tissue, indicating the critical role of macrophages in mediating the antigen cross-presentation and thus activating the CD8⁺ T cells (figure 4C–E). Therefore, these results indicated that the antitumor effects of pep-20 were dependent on CD8⁺ T-cell activation mediated by macrophages.

Previous studies have demonstrated that CD47 is expressed on a variety of cells, including hematopoietic cells.6 Therefore, the blood toxicity caused by CD47/SIRPα blockade has received more attention. To preliminarily assess the toxicity of pep-20 in vivo, C57BL/6 naïve mice were injected s.c. with pep-20 at a dose of 2 mg/kg daily over 14 days. Blood samples were collected, and hematological parameters were analyzed either before or after pep-20 treatment. It was found that the red blood cell count and hemoglobin level had no significant differences (online supplemental figure S10A,B). Subsequently, the organ coefficients and hepatic damage analysis also revealed no significant differences after pep-20 treatment (online supplemental table S2 and S3). Consistently, histopathology analysis illustrated no abnormalities by H&E staining of tissue sections from major organs (online supplemental figure S10C). Collectively, the results demonstrated that pep-20 could not induce anemia and other obvious side effects at a dose of 2 mg/kg in mice.

To determine the binding detail of pep-20 to CD47, the 3D conformation of pep-20 was predicted by PEP-FOLD3, as shown by using the MOE software (online supplemental figure S11A). Next, the possible interaction of pep-20 with human CD47 protein (PDB ID: 2JJS)27 was predicted by ZDOCK. In order to confirm the binding model, pep-20-derived alanine substitution analogs were tested by cell-based blocking assay. As shown in online supplemental figure S11B, the predicted W2, W6, Y9 and W10 residues may be the major binding sites. Subsequently, a possible binding pattern complex 796 was obtained by correlation analysis of the top 1000 predictions (online supplemental figure S11C), and the docking model and interaction residues of CD47/pep-20 were displayed (online supplemental figure S11A). It was found that the binding sites of CD47/pep-20 were located exactly within the CD47/SIRPα interaction interface (online supplemental figure S11D). These findings revealed the docking model of pep-20, which lays down a potential platform to generate more potent derivates in the future.

To develop a more stable peptide which could be administered systemically, modification of pep-20 with D-amino acid substitution was attempted. According to the docking model on the interaction of pep-20 and CD47, and to ensure the blockading effect and hydrolysis stability, both N-terminal and C-terminal residues were systematically substituted by up to three D-amino acids (online supplemental table S4). The blocking assay was performed as mentioned above. The results showed that pep-20-D12, with three D-amino acid residues substituted both in N-terminal and C-terminal of the parent pep-20, could retain the blockade activity toward human and mouse CD47/SIRPα (online supplemental figure S12A,B), indicating that the N-terminal and C-terminal residues of pep-20 could be substituted without significant functional decrease. Subsequently, the enzymatic degradation stability was examined in 10% human serum at 37°C. Different from parent pep-20, pep-20-D12 exhibited potent proteolytic resistance, which retained the same initial concentration for up to 36 hours (online supplemental figure S12C,D). In addition, pep-20-D12 showed more prolonged half-life (T_1/2) in vivo with the intravenous elimination T_1/2 of 6.36 hours and C_max of 54.45 µg/mL (40 mg/kg) in C57BL/6 mice than pep-20 with T_1/2 of 0.59 hours and C_max of 33.60 µg/mL (online supplemental figure S12E,F). These findings suggested that the stability of pep-20 could be greatly improved by terminal D-amino acid substitution.

In view of the evidence that pep-20-D12 has better proteolytic stability than parent pep-20, the antitumor efficacy and immune stimulatory activity of both peptides were performed in MC38-bearing mice via systemic treatment. Pep-20-D12 slowed the tumor progression significantly, whereas pep-20 had a slight impact on tumor growth (online supplemental figure S13A). Similarly, pep-20-D12 enhanced the infiltration of tumor-specific T cells and antitumor immunity (online supplemental figure S13B-D).

Consistently, the tumor growth inhibition efficacy by systemic treatment of pep-20-D12 was similar as the anti-CD47 antibody (figure 5A,B). To examine the effects of pep-20-D12 treatment on antitumor T-cell immune response, CD8⁺ T cells from tumor tissues, tumor-draining lymph nodes and spleen were analyzed. Blockade of CD47 with systemically pep-20-D12 treatment also remarkably increased the CD8⁺ T cell population in tumor tissues (figure 5C). Furthermore, IFN-γ production by CD8⁺ T cells from TILs, tumor-draining lymph nodes and spleen was also significantly increased, especially in TILs and draining lymph nodes (figure 5D–F). More importantly, pep-20-D12 showed less blood toxicity compared with anti-CD47 antibody by evaluating the hematological parameters of mice after treatment (online supplemental figure S14A,B). These results indicated that pep-20-D12 maximally extended antitumor effects with systemic administration.

It was reported that the proportion of monocytic myeloid-derived suppressor cells (M-MDSCs) in tumor tissues increased significantly after local IR in mice.28 Considering that these cells could be differentiated into macrophages subsequently,29 30 we attempted to evaluate whether combined pep-20-D12 treatment with IR could be more effective to inhibit tumor growth than either treatment alone. MC38 tumors locally received IR once (20 Gy) and were then treated by injection i.p. of pep-20-D12 daily for 2 weeks. Surprisingly, a combination of pep-20-D12 treatment with IR showed a significant delay or even complete regression in tumor growth (figure 6A,B). The mechanism study indicated that the proportion of tumor infiltrating macrophages and monocyte-derived MDSCs was significantly increased on day 3 and the end of treatment by IR (online supplemental figure S15A,B, figure 6C-F). More importantly, pep-20-D12 could promote the tumor infiltrating monocyte-derived MDSCs-mediated phagocytosis of tumor cells (online supplemental figure S15C). These results demonstrated that the combination of CD47 blockade pep-20-D12 peptide with IR could be a very promising strategy to synergistically eradicate the tumor.