Structural analysis and binding sites of inhibitors targeting the CD47/SIRPα interaction in anticancer therapy

Graphical abstract

Abstract

Cluster of differentiation 47 (CD47)/signal regulatory protein alpha (SIRPα) is a negative innate immune checkpoint signaling pathway that restrains immunosurveillance and immune clearance, and thus has aroused wide interest in cancer immunotherapy. Blockade of the CD47/SIRPα signaling pathway shows remarkable antitumor effects in clinical trials. Currently, all inhibitors targeting CD47/SIRPα in clinical trials are biomacromolecules. The poor permeability and undesirable oral bioavailability of biomacromolecules have caused researchers to develop small-molecule CD47/SIRPα pathway inhibitors. This review will summarize the recent advances in CD47/SIRPα interactions, including crystal structures, peptides and small molecule inhibitors. In particular, we have employed computer-aided drug discovery (CADD) approaches to analyze all the published crystal structures and docking results of small molecule inhibitors of CD47/SIRPα, providing insight into the key interaction information to facilitate future development of small molecule CD47/SIRPα inhibitors.

1. Introduction

Currently, the high morbidity and mortality of tumors remain a major challenge to the effectiveness of cancer treatment even though a series of anticancer drugs have been developed based on different treatment strategies [1]. Cancer immunotherapy, which aims to improve antitumor immune responses with fewer off-target effects than chemotherapies and other agents that directly kill cancer cells, provides an alternative strategy to treat cancer through the immune system rather than the tumor itself [2] and is an exciting area in current cancer research [3], [4]. This novel therapy, including immune checkpoint blockade, adoptive cellular therapy and cancer vaccinology [5], has led to a growing number of immunotherapy drug approvals, with numerous treatments in clinical and preclinical development [3].

Herein, the immune checkpoint plays a central role in tissue homeostasis self-reactivity and autoimmunity, which targets the innate and adaptive immune systems [6], and is the signal recognition factor for immunosurveillance in immune cells [7]. However, some tumor cells evade immune clearance by modulating signal recognition between immune cells and tumor cells; thus, immune checkpoint blockade is the main promising system in immunotherapy [8], [9]. To date, the blockade of two well-known adaptive immune system checkpoints, cytotoxic T-lymphocyte-associated antigen 4 (CTLA4) and programmed cell death protein 1 (PD1), has led to the generation of several immune checkpoint inhibitors that have been used for treating multiple cancers [10], [11], [12], [13]; however, an unsatisfactory overall response rate, tumor heterogeneity, drug resistance and rapid progression after therapy in patients are some of the challenges of inhibiting adaptive immune checkpoints [14], [15], [16], [17].

Interestingly, in addition to adaptive immunity, innate immune processes may also provide a good direction for cancer immunotherapy. Recently, CD47, which is one of the most attractive innate immune checkpoint regulators because of its inhibitory effect on the activation of macrophages and other myeloid cells against tumor, brings new hope to cancer patients. It belongs to the immunoglobulin (Ig) superfamily and is ubiquitously-expressed on all normal cells but overexpressed on hematological and solid malignancy cancer cell membranes [18], [19], [20], [21]. As a highly glycosylated transmembrane protein, CD47 consists of an extracellular amino terminal IgV-like domain, five membrane-spanning segments that are highly hydrophobic and a hydrophilic cytoplasmic C-terminus that is 3–6 amino acids long [22], and it interacts with the NH₂-terminal V-set domain of its ligand, SIRPα [23]. SIRPα is also known as CD172a or SHPS-1 and is highly expressed on the membrane of myeloid cells, such as monocytes, macrophages, neutrophils, and dendritic cells (DCs) [10], [24], [25]. CD47 interacts with SIRPα, tagging it with a “self” or “do not eat” signal, to trigger an inhibitory signaling cascade through the ITIM and ITSM motifs of SIRPα, inhibiting macrophage phagocytosis (Fig. 1) [26], [27], [28]. Unfortunately, when overexpressed CD47 on the surface of solid malignancies binds to SIRPα on macrophages, this can suppress the phagocytic responses of macrophages [29]. Therefore, as a dominant macrophage checkpoint, disruption of the CD47/SIRPα pathway can induce macrophage-mediated phagocytosis of tumor cells and thus is employed when generating next-generation immunoregulatory drugs [30], [31], [32]. Currently, the gradual increase in patents with treatments targeting the CD47/SIRPα pathway clearly demonstrates the great effort being made to discover new inhibitors of this target. Consequently, some CD47/SIRPα pathway-targeting antibodies against various cancers (Table 1), including acute myeloid leukemia (AML) [33], [34], anaplastic thyroid carcinoma (ATC) [35], lymphoma [36], [37], [38], lung cancer [39], [40] and breast cancer [41], have showed attractive results. For example, magrolimab (Hu5F9-G4) is an anti-CD47 monoclonal antibody that is currently undergoing phase III clinical trials, with an objective response rate of 75% in phase Ib clinical results during the treatment of myelodysplastic syndromes. ALX-148 is also a CD47 targeting antibody in phase II clinical trials whose objective responses could be observed in phase I clinical results for the treatment of patients with head and neck squamous cell carcinoma. It is undeniable that some adverse effects occur when using these antibodies therapeutically, such as the rapid target-mediated clearance, transient anemia, erythrocyte toxicity and infusion-related reactions [42], which limit their rapid development. In addition, there are certain limitations of using these antibodies, including a long half-life, poor permeability and lacking oral availability. Thus, the development of low molecular-weight inhibitors with superb pharmacokinetics and druggability is an effective strategy and research focus to overcome the limitations of therapeutic antibodies [43].

Fig. 1.

CD47/SIRPα pathway. Cell surface calreticulin (CRT) binds to low-density lipoprotein–related protein (LRP) to promote phagocytosis. CD47 can downregulate phagocytic activity by interacting with SIRPα on macrophages. Blockade of the CD47/SIRPα pathway through Anti-CD47 mAb can recover phagocytic activity of macrophage.

Table 1.

CD47/SIRPα targeting mAbs in clinical trial.

Code Name	Target	Organization	Therapeutic Groups	Highest
(Generic Name)				Phase
Hu5F9-G4 (Magrolimab)	CD47	Gilead	Bladder Cancer, Breast Cancer, Colorectal Cancer, Hematological Cancer, Lymphoma, Myeloid Leukemia, Non-Hodgkin's Lymphoma, Ovarian Cancer.	III
ALX-148	CD47	Alexo Therapeutics	Gastric Cancer, Head and Neck Cancer, Non-Hodgkin's Lymphoma.	II
TJC-4 (Lemzoparlimab)	CD47	AbbVie	Lymphoma, Myeloid Leukemia.	II
DSP-107	CD47	KAHR Medical	Non-Small Cell Lung Cancer.	II
IBI-188 (Letaplimab)	CD47	Innovent Biologics	Myeloid Leukemia, Non-Hodgkin's Lymphoma, Ovarian Cancer.	II
AO-176	CD47	Arch Oncology	Lymphoma, Multiple Myeloma.	II
TTI-622	CD47	Trillium Therapeutics	Lymphoma, Multiple Myeloma, Myeloid Leukemia, Non-Hodgkin's Lymphoma, Ovarian Cancer.	II
ZL-1201	CD47	ZAI Lab	Lymphoma Therapy.	I
AK-117	CD47	Akeso Biopharma	Non-Hodgkin's Lymphoma Therapy.	I
IMC-002	CD47	ImmuneOncia Therapeutics	Lymphoma.	I
SRF-231	CD47	Surface Oncology	Hematological Cancer Therapy, Lymphocytic Leukemia Therapy, Lymphoma Therapy, Multiple Myeloma Therapy.	I
CC-90002	CD47	Celgene	Myeloid Leukemia Therapy, Non-Hodgkin's Lymphoma Therapy.	I
TTI-621	CD47	Trillium Therapeutics	Hematological Cancer Therapy, Lymphocytic Leukemia Therapy, Myeloid Leukemia Therapy, Non-Hodgkin's Lymphoma Therapy.	I
GS-0189	SIRPα	Gilead	Oncolytic Drug.	I
CC-95251	SIRPα	Celgene	Solid Tumors.	I
FSI-189	SIRPα	Gilead	Non-Hodgkin's Lymphoma	I
BI-765063	SIRPα	Boehringer Ingelheim	Solid Tumors Therapy	I

Because the development of small molecule inhibitors of the CD47/SIRPα interaction has been slower than the development of antibody treatments, research on the CD47/SIRPα pathway is crucial. Currently, high-throughput screening and computer-aided drug design (CADD) are common approaches in the discovery of small molecule inhibitors [44], [45], [46], [47], [48]. CADD is a molecular design method based on computational chemistry [49], [50] and can help to produce valuable information on target proteins, lead compounds and protein–ligand interactions for rational drug design. Thus, CADD is used both to analyze hot spots of target proteins and protein–ligand interactions model for further drug discovery. In this review, we have employed the “View Interactions” tool in CADD to analyze the hot spots of CD47 and SIRPα based on their cocrystal structures. Subsequently, “LibDock” in CADD was further applied to predict the key interacting amino acids of CD47 and SIRPα from a small molecule inhibitor docking experiment. In addition, some of the peptides and small molecule inhibitors that have been reported to block the CD47/SIRPα interaction in fundamental research studies have been summarized, and their structure–activity relationships have been analyzed and compared to guide the discovery and design of new inhibitors blocking the CD47/SIRPα interaction with superb pharmacokinetics and druggability.

2. Structures of CD47/SIRPα complexes and cocrystal structures of monoclonal antibodies

The first high-resolution crystallographic structure of the CD47/SIRPα d1 complex (the ligand-binding domain) was published by Hatherley et al. in 2008 and the identifier of this complex in Protein Data Bank is 2JJT (PDB ID: 2JJT) [51]. Within the crystal, CD47 and SIRPα d1 form a 1:1 stoichiometry complex. The CD47 and SIRPα d1 molecules are interdigitated to each other so that their interaction is mainly mediated by loops at the intracellular side, which is consistent with what had been proposed by other authors based on their analyses [52]. The CD47/SIRPα d1 interaction interface is mainly formed of four N-terminal loops of the SIRPα d1 domain and the FG loop of CD47, which embeds into the cavity on the surface of SIRPα d1. Thr¹⁰² of the FG loop inserts deep into SIRPα d1 (Fig. 2, produced by Discovery Studio2019). The hot spot residue-mediated polar interactions on CD47 comprise Glu⁹⁷, Thr⁹⁹, Glu¹⁰⁰, Arg¹⁰³, Glu¹⁰⁴, and Glu ¹⁰⁶. Among them, Glu¹⁰⁴ and Glu¹⁰⁶ of CD47 form hydrogen bonds with SIRPα (Table. 2), and the BC loop surrounding the FG loop interacts with the wide edge of the SIRPα groove. Comparative analysis of the CD47/SIRPα d1 complex structures with isolated structures of CD47 and SIRPα demonstrates that complex formation slightly impacts the backbone of CD47. In contrast, the complex formation rearranges the CD47-interacting loops in SIRPα d1. In addition, Weiskopf et al. reported the high-affinity SIRPα variant FD6/CD47 complex in 2013 [53]. The root mean square deviation (RMSD) between the FD6/CD47 complex and the wild-type SIRPα/CD47 complex is 0.61 Å. FD6 and wild-type SIRPα interact with the overlapping CD47 epitope. However, key mutations in the C'D loop of FD6 may promote the interaction between Ala⁵³ and glutamic acids on CD47 [53].

Fig. 2.

Structure of the CD47/SIRPα d1 complex and the interaction between the FG loop and SIRPα d1. (A) Overview of the CD47/SIRPα d1 complex. CD47 is colored cyan, the FG loop is colored yellow, and SIRPα d1 is colored purple. (B) The FG loop/ SIRPα d1 interaction. The residues of the FG loop are colored green, and the interacting residues of SIRPα d1 are colored navy. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2.

Contact residues of CD47 and SIRPα d1 (distances <5.0 Å).

CD47 contact residue	CD47 residue location	SIPRα contact residue	SIPRα residue location
Gln1, Lys6	A loop	Ile30, Gly34, Pro35	B2C
Asn27, Glu29, Glu35	BC loop	Gln52, Lys53, Glu54	C'D loop
Tyr37, Lys39	C strand	Ser66, Thr67, Arg69	DE loop
Asp46	C' strand	Lys93	F strand
Glu97, Glu100	F strand	Lys96, Gly97, Ser98, Asp100	FG1 strand
Leu101, Thr102	FG loop
Arg103, Glu104, Glu106	G strand

To date, four complex crystal structures of CD47/mAbs have been reported. They are the CD47/magrolimab complex (PDB ID: 5IWL) [30], CD47/B6H12.2 complex (PDB ID: 5TZU) [54], CD47/C47B222 complex (PDB ID: 5TZ2) [54], and CD47/C47B161 complex (PDB ID: 5TZT) [54]. Magrolimab, also known as Hu5F9-G4, is an IgG4 antibody and can form a Hu5F9-G4 diabody/CD47-ECD complex with the CD47 extracellular domain (CD47-ECD). Hu5F9-G4 employs VH and VL to cover the CD47-ECD surface area with 365 Ų and 310 Ų, respectively. Superposition of the CD47/magrolimab complex and CD47/SIRPα complex reveals that their surface interaction involves similar epitopes on CD47, including the BC and FG loops. Specifically, the CDR loops of magrolimab form 11 hydrogen bonds with the CD47-ECD surface (Fig. 3, produced by Discovery Studio2019). Moreover, the crystal structures of the CD47-B6H12.2, CD47-C47B222, and CD47-C47B161 complexes demonstrate that the interaction interface of all three antibodies overlaps with the SIRPα binding epitope regions on the FG loop of CD47. Thus, these complex structures demonstrate that the FG loop of CD47 is a key component of the interaction, indicating that the FG loop may become a potential target for further structure-based drug design.

Fig. 3.

Structure of the Magrolimab/CD47 complex. The FG and BC loops of CD47 are colored yellow and brown, respectively. The residues in the VH of magrolimab are colored cyan, and the residues in the VL of magrolimab are colored purple. The green dashed lines indicate hydrogen bond formation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Magrolimab mainly binds to N-terminal pyroglutamate of CD47 which is critical for CD47/SIRPα interaction and magrolimab binds to the BC and FG loops, which are highly overlapping epitopes with SIRPα [55], [56]. Analogously, the common binding area of B6H12.2, C47B161 and C47B222 is the CC′ and FG loops on CD47. According to these findings, the N-terminal pyroglutamate and the BC, CC′ and FG loops can be viewed as potential binding areas for subsequent structure-based drug design, and Tyr³⁷, Asp⁴⁶, Glu⁹⁷, Glu¹⁰⁰ and Glu¹⁰⁶ may be developed into binding sites for inhibitors targeting CD47. Additionally, the loops of SIRPα undergo structural changes in the interaction with CD47, which indicates that C′D, DE, and FG loops may be possible targets. Among these loops, Glu⁵⁴, Gly⁵⁵, Ser⁶⁶ and Ser⁹⁸ undergo considerable movement, implying that these residues play crucial roles in the interaction and may act as binding sites for designing future SIRPα inhibitors.

3. Bioactive peptide inhibitors blocking the CD47/SIRPα interaction

The development of non-antibodies with succinct synthesis and lower modification cost has led to the discovery of RS-17, Pep-20, D4-2, and SP5. All four peptides, whose sequences are listed in Table 3, can directly block the CD47/SIRPα interaction. RS-17 and Pep-20 block the CD47/SIRPα interaction by binding to CD47, while D4-2 and SP5 can bind to SIRPα to disrupt the CD47/SIRPα interaction.

Table 3.

Sequences of bioactive peptides inhibitors blocking the CD47/SIRPα interaction.

Code Name(Generic Name)	Target	Organization	Sequence
RS-17 [58]	CD47	China Pharmaceutical University	RRYKQDGGWSHWSPWSS-NH2
Pep-20 [55]	CD47	Sun Yat-sen University and Zhengzhou University	AWSATWSNYWRH
D4-2 [59], [60]	SIRPα	Kobe University	Ac-yRYSAVYSIHPSWCG-NH2
SP5 [62]	SIRPα	Zhengzhou University	CTQDAWHIC

3.1. CD47-targeted peptides

3.1.1. Pep-20 and its derivatives

In 2020, Pep-20 and its derivatives that have comparable affinity to the CD47/SIRPα interaction were identified by Wang et al. using a subtractive phage biopanning strategy [57]. The K_d values of pep-20 binding to human and mouse CD47 are 2.91 ± 1.04 µM and 3.63 ± 1.71 µM, respectively, which are close to that of cognate SIRPα [58], [59]. In addition, a human CD47/SIRPα blocking assay also revealed that pep-20 exhibited an IC₅₀ of 24.56 µM with the anti-CD47 antibody (B6H12), which served as a positive control. Pep-20 remarkably enhances the phagocytosis of MCF7 (human breast tumor cell lines), HT29 (human colon tumor cell lines) and Jurkat (human leukemia cell lines) and exhibits an enhancement of phagocytosis similar to that of the positive control (B6H12). Excitingly, the injection of pep-20 at a dose of 2 mg/kg daily in mice had no obvious influence on the reduction in the number of red blood cells, which is a common toxicity effect of CD47/SIRPα blockade [57]. Furthermore, after replacing three terminal residues of pep-20 with D-amino acids, the obtained peptide pep-20-D12 significantly improved stability without a functional decrease compared with pep-20, accompanied by an intravenous elimination T1/2 that increased by tenfold compared with pep-20. Pep-20-D12 remarkably slows tumor progression, and the combination treatment of pep-20-D12 and IR shows tumor growth regression in colon tumor (MC38 cells)-bearing mice [57]. A subsequent docking model and alanine substitution experiment of pep-20/CD47 revealed that Phe⁴, Glu¹⁰⁴ and Glu¹⁰⁶ of CD47 are key positions for inhibitors targeting CD47. These findings provide valuable information of CD47 binding sites and the key structure of pep-20 for small-molecule inhibitor design.

3.1.2. Rs-17

Additionally, Xu et al. from China Pharmaceutical University discovered RS-17 in 2020 [60]. The K_d value of RS-17 binding to the CD47 protein was 3.85 ± 0.79 nM. At a concentration of 20 μg/ml, RS-17 effectively binds to CD47 of SCC-13 (human epidermal squamous tumor cells) and HepG2 (human liver tumor cells) with corresponding binding rates of 55.5% and 71.2%, respectively; thus, the phagocytic efficiency of macrophages against HepG2 cells was greatly improved, showing a half phagocytic index of B6H12. Moreover, an in vivo assay demonstrated that the weight loss and tumor volume increase in liver tumor-bearing mice were similar between RS-17 and B6H12 and that RS-17 effectively inhibited tumor growth in liver tumor-bearing mice.

3.2. SIRPα-targeted peptides

3.2.1. d4-2

Hazama et al. utilized random nonstandard peptides integrated discovery (RaPID) system, which combines flexizyme-assisted genetic code reprogramming and mRNA display to obtain macrocyclic peptides of interest, to design and gain anti-SIRPα peptides L4-4, D4-1, D4-2 and D4-4 [61], [62]. Among these peptides, D4-2 shows comprehensive high binding affinity to SIRPα of C57BL/6 and NOD mouse strains with corresponding K_d values of 10 nM and 8.22 nM, respectively. D4-2 evidently blocked the mCD47-Fc/NOD SIRPα interaction in a dose-dependent manner in HEK293A (human embryonic kidney cells) cells with an IC₅₀ value of 0.180 mM. The crystal structure of the D4-2/NOD SIRPα complex shows that the interaction area of D4-2/NOD SIRPα occupies 976.5 Ų. Arg², Ser⁴, Ala⁵, Val⁶, IIe⁹, His¹⁰, Pro¹¹, Ser¹², Trp¹³ and Gly¹⁵ of D4-2 form hydrogen bonds with IgV-NOD SIRPα, and Arg² of D4-2 forms a salt bridge with Asp⁸⁴ of IgV-NOD SIRPα. Ala⁵ and Pro¹¹ of D4-2 form hydrophobic interactions with Phe⁵¹ and Phe⁵⁶ of IgV-NOD SIRPα. All these residues stabilize the cyclic structure of D4-2 and mediate the binding of D4-2 to IgV-NOD SIRPα. Further crystal structure comparison shows that the binding of D4-2 to Phe⁵⁶ and Ala⁶⁵ in the C’E loop of IgV-NOD SIRPα, which are key residues controlling the interaction with CD47, changes the conformation and induces the inhibition of the CD47/IgV-NOD SIRPα interaction [63].

3.2.2. Sp5

In addition to D4-2, a series of macrocyclic peptides binding to SIRPα, including SP1 to SP6, were also developed in 2020 by Xu et al from Zhengzhou University [64]. Among these, SP4 and SP5 display higher affinity to SIRPα with K_d values of 0.85 μM and 0.38 μM and block the SIRPɑ/CD47 interaction in a dose-dependent manner in CHO-K1-hSirpɑ cells. SP5 (200 μM) not only effectively promotes the phagocytosis of HT29 (human colon tumor cells) by macrophages but also exhibits desirable in vivo efficacy by inhibiting tumor growth in colon tumor MC38 mouse model and melanoma B16-OVA mouse model.

3.3. CADD guides the design of peptide inhibitors

Analysis of the previously described peptides can lead to conclusions that these peptides share similar interaction areas that overlap with the epitopes in the CD47/SIRPα interaction area. For example, Pep-20 occupies Phe⁴, Glu¹⁰⁴ and Glu¹⁰⁶ of CD47 to block the D47-SIRPα interaction, and D4-2 binds to Phe⁵⁶ and Ala⁶⁵ of SIRPα to block the D47-SIRPα interaction. Moreover, these peptides mainly form hydrogen bonds with their receptor. Collectively, residues Phe⁴, Glu¹⁰⁴ and Glu¹⁰⁶ of CD47 and residues Phe⁵⁶ and Ala⁶⁵ of SIRPα may be developed into binding sites for structure-based CD47/SIRPα small molecule inhibitor design.

4. Small molecule inhibitors blocking the CD47/SIRPα interaction

4.1. NCGC00138783 and its derivatives

Miller et al. utilized quantitative high-throughput screening (qHTS) assays to screen NCATS chemical libraries based on time-resolved Fo¨rster resonance energy transfer (TR-FRET) and bead-based luminescent oxygen channeling assay formats (AlphaScreen), resulting in the discovery of the parent compound NCGC00138783 [65], [66] whose scaffold is 2-((2-(2-(3,5-dimethyl-1H-pyrazol-4-yl) ethyl)-5,6-dihydro-[1,2,4] triazolo [1,5-c] quinazolin-5-yl) thio) butanal. This compound selectively blocks the CD47/SIRPα interaction without disrupting its binding to other receptors [67], [68], [69]. A novel laser scanning cytometry assay (LSC) was established to measure the cell surface binding of these compounds, and the results showed that NCG00138783 has an IC₅₀ value of 40 μM. Further medicinal chemistry work attempting to optimize the potency and drug-like properties of NCGC00138783 led to the discovery of its derivatives (Fig. 4, produced by ChemDraw). The acyl group of NCGC00138783 is linked with a monocyclic substituted amino group and hydroxyl group to obtain a range of compounds displaying great inhibitory activity toward the CD47/SIRPα interaction. Among these small molecule compounds, NCGC00138783, NCG00538430 and NCG00538419 showed antagonistic activity in both the ALPHA screening assay and LSC assay.

Fig. 4.

The core scaffold structure and representative compounds of NCGC00138783 and its derivatives are shown.

To facilitate the understanding of NCGC00138783 binding to CD47/SIRPα, we conducted docking experiments of NCGC00138783 docking to CD47 and SIRPα and employed the CD47/SIRPα complex (PDB ID: 2JJT) as a receptor. Consequently, we found that NCGC00138783 is more prone to bind to SIRPα than CD47 with the corresponding highest LibDock Score of 134 and 85. Furthermore, the 3,5-dimethyl-1H-pyrazolyl group, central [1,2,4] triazolo[1,5-c] quinazoline group and amide group of NCGC00138783 are predicted to form hydrogen bonds and T-stacking interactions with SIRPα, including Leu³⁰, Gly³⁴, Pro³⁵, Gln⁵², Lys⁵³ and Lys⁹³, which are key residues in the CD47/SIRPα interaction. The central [1,2,4] triazolo[1,5-c] quinazoline scaffold is predicted to form pi-pi stacking with Phe⁷⁴ and hydrogen bonding with Gly³⁴, which makes it insert into the hydrophobic cavity. The amide group of 3,5-dimethyl-1H-pyrazolyl is predicted to form hydrogen bonding with Gln⁵², which lies in the high polarity area (Fig. 5, produced by Discovery Studio2019). Above all, NCGC00138783 binds to SIRPα and occupies the key binding positions of the CD47/SIRPα interaction which is essential information for future small molecule inhibitor design and helpful for the discovery of novel inhibitors blocking the CD47/SIRPα interaction.

Fig. 5.

Docking analysis of NCGC00138783 to SIRPα. The interaction area of CD47/SIRPα is colored brown, and the predicted interaction area of NCGC00138783/SIRPα is colored yellow. The interacting amino acids in SIRPα are colored purple, and the predicted interacting amino acids in NCGC00138783 is colored cyan. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4.2. 1,2,4-oxadiazole compounds

In a patent application, inventors from Aurigene Discovery Technologies Limited reported a series of small molecules blocking the CD47/SIRPα interaction [70]. The scaffold of these compounds is oxadiazole which can enhance macrophage-mediated phagocytosis of human lymphoma and myeloma cells, with corresponding normalized phagocytosis rates in the range of 20%–66% and 17%–77% at 10 μM. Among these small molecules, compounds 1 to 14 display comprehensive effects on both luciferase-based and FACS-based phagocytosis assays, and compound 6 has normalized phagocytosis rates of 66% and 74%, respectively (Fig. 6, produced by ChemDraw). Moreover, compound 6 inhibited tumor growth in a dose-dependent manner with inhibition rates of 53%, 64% and 67% at doses of 3, 10 and 30 mg/kg, respectively, in an A20 mouse model without body weight loss.

Fig. 6.

The core scaffold structure and representative 1,2,4-oxadiazole compounds are shown.

Further docking analyses of compound 6 to CD47 (PDB ID: 2JJT) by us revealed that the core 1,2,4-oxadiazol and butyramide groups insert into a hydrophobic pocket containing Trp⁴⁰, Thr¹⁰⁷ and Lys⁶ of CD47, which is the key residue in the CD47/SIRPα interaction. The carbonyl group of butyramide is predicted to form hydrogen bonds with Thr⁷ and Thr¹⁰⁷, which is near the core CD47/SIRPα interaction area. Two carbonyl groups of carbamoyl proline are predicted to interact with Asn⁵, which is close to Lys⁶ of CD47 (Fig. 7, produced by Discovery Studio2019). Overall, promising compound 6 could inspire the design of follow-up lead compound scaffolds, and the binding model of compound 6 to CD47 may provide information for further discovery of small molecules blocking the CD47/SIRPα interaction.

Fig. 7.

Docking analysis of compound 6 binding to CD47. The interaction area of CD47/SIRPα is colored purple, and the predicted interaction area of compound 6/CD47 is colored yellow. The interacting amino acids in CD47 are colored purple, and the predicted interacting amino acids compound 6 is colored cyan. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4.3. CADD guides the design of small molecule inhibitors

Analysis of our above docking results and the published phagocytosis assays revealed the common structural characteristics of anti-CD47 compounds: (i) hydrogen bond interactions are crucial for anti-CD47 compound activity; (ii) the 2-position side chain of oxadiazol inserted into the pocket consisting of Asn⁵, Thr⁷, Pro²², Phe²⁴ and Thr¹⁰⁷ has a crucial effect on phagocytic activity; (iii) the terminal group of the oxadiazol 2-position side chain, including the amino group, carboxyl group, amide group and guanidine group, which can form hydrogen bonds with Thr⁷ and Thr¹⁰⁷ in the pocket, exhibits higher phagocytic activity; and (iv) the oxadiazol 5-position side chain contains a terminal carboxyl group and ureido. Oxadiazol, ureido and carboxyl groups are separated by one carbon atom, which makes two carbonyl groups in suitable positions to form hydrogen bonds with Asn⁵; (v) the substituted short carbon chain and substituted ring alpha carbon of the terminal carboxyl group in the oxadiazol 5-position side chain can achieve better activity; (vi) central oxadiazole is needed for pi-lone pair conjugation with Thr⁷.

In addition, analysis of anti-SIRPα compounds also indicates several features: (i) hydrophobic interactions are essential for SIRPα binding; (ii) central quinazoline reaches a hydrophobic pocket containing Val²⁷, Leu³⁰, Ile³⁶, Phe⁷⁴ and Lys⁹³ to form two T-stacking interactions with Phe⁷⁴; and (iii) amide groups forming hydrogen bonds with Gln⁵², benzene, and four-membered ring- or three-membered ring- substituted amide groups show favorable antagonistic activity.

5. Summary and outlook

Recently, CD47/SIRPα inhibitors have aroused enormous interest among researchers and have been remarkably affective in cancer treatment. This field is progressing rapidly, and some mAbs targeting the CD47/SIRPα pathway have reached clinical phase II and phase III [71]. Notably, despite the excellent clinical performance shown by CD47/SIRPα antibodies, the limitations of antibody drugs including poor tumor permeability, undesirable oral bioavailability and poor stability, hinder their clinical application [72], [73], [74], [75]. Small-molecule inhibitors can eliminate the problems caused by antibody drugs and thus have attracted the attention of researchers and have become a promising research area.

Currently, several crystal structures of CD47/SIRPα and the structures of antibodies together with receptors have been published which provides guidance for the rational design of small molecule inhibitors blocking the CD47/SIRPα interaction. Moreover, there are two published small molecule compound categories: small molecules containing 3,5-dimethyl-1H-pyrazolyl and [1,2,4] triazolo[1,5-c] quinazoline scaffolds and small molecules containing oxadiazole scaffolds.

Unfortunately, no small molecules blocking the CD47/SIRPα interaction have reached clinical research yet. The shortage of target structure information limits the development of small molecule inhibitors. Excitingly, using CADD to analyze the CD47/SIRPα interaction will provide some crucial information for the design of small molecule inhibitors. First, CADD allows researchers to analyze the interaction between inhibitors and their receptors based on their crystal structures, which improves the understanding of the interaction process and helps to determine key information, including pocket atoms and hot spot residues. Next, CADD helps researchers to explore the interaction between inhibitors and receptors without crystal structure through docking. Here, based on the previous reports and our docking research, we conclude that Glu¹⁰⁴ and Glu¹⁰⁶ are hot spots on CD47, while Gln⁵², Lys⁵³ and Phe⁵⁶ are hot spots on SIRPα.

There is a limited number of small molecule inhibitors targeting the CD47/SIRPα pathway, which indicates the early stage of this research, but the favorable clinical results are promising. The CADD technologies will accelerate the discovery of novel inhibitors. These peptides and small molecule inhibitors will be the foundation for the design of new compounds. As it relates to drug design, the interaction area of CD47/SIRPα is broad. Therefore, it is crucial to identify the best binding positions for small molecules, and drug design based on these structural data will lead to the successful development of CD47/SIRPα inhibitors.

CRediT authorship contribution statement

Bo Huang: Conceptualization, Writing – original draft, Writing – review & editing, Visualization. Zhaoshi Bai: Supervision, Writing – review & editing. Xinyue Ye: Visualization. Chenyu Zhou: Resources. Xiaolin Xie: Visualization. Yuejiao Zhong: Resources. Kejiang Lin: Supervision, Conceptualization. Lingman Ma: Supervision, Writing – original draft, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.