How the antibody ADU-1805 blocks the SIRPα-CD47 immune checkpoint across all gene variants

Full-length cDNA of human SIRPα variant 1 (hSIRPαV1) (GenBank accession: NM_001040022.1↗) and hSIRPαV2 (GenBank accession: D86043.1↗) were synthesized (GeneArt, Thermo Fisher Scientific), subcloned into the pCI-neo vector (Promega) and used to immunize mice. Hybridomas were generated as described previously [28]. Selected stable hybridomas were cultured in serum-free media for 7 days, supernatants were harvested, and antibodies were purified using MabSelect Sure Protein A resin (GE Healthcare). Antibody concentrations were quantified using spectrophotometry. The isotype of antibodies was established using mouse a monoclonal antibody isotyping kit (Bio-Rad Laboratories).

Antibody sequences were identified by DNA sequencing of the selected hybridomas (LakePharma). Antibody VH and VL genes were synthesized by GeneArt (Thermo Fisher Scientific), subcloned into the pcDNA3.1(+) vector (Thermo Fisher Scientific) and expressed in FreeStyle 293-F (Thermo Fisher Scientific) or ExpiCHO-S cells (Thermo Fisher Scientific). Transfected cells were cultured in serum-free media for 7 days and mAbs were purified using MabSelect Sure Protein A resin (GE Healthcare).

Humanization of the mouse anti-human SIRPα.40A mAb (hSIRPα.40A) was performed by grafting complementarity-determining region (CDR) residues onto a human germline framework [29]. Differences between mouse hSIRPα.40A and the human framework residues were individually modeled by a homology model on basis of PDB ID 3UMT (light chain), PDB ID 1EHL (heavy chain) and PDB ID 3BGF (Fv) using Discovery Studio 4.5 (BIOVIA). Post-translational modification (PTM) motifs were removed where possible.

The BJAB (DSMZ), Raji (ECACC), THP-1 (ATCC), U-937 (ATCC) and NK-92MI (ATCC) human cell lines, the IC-21 (ATCC) mouse cell line, and the CHO-K1 (ATCC) hamster cell line were cultured as recommended by the vendor. Cell lines were validated as Mycoplasma negative by Baseclear B.V. (Leiden) using a validated PCR test.

A recombinant human SIRPα/His fusion protein (Sino Biological) was used for measuring monomeric binding affinity to hSIRPα.40A and derivatives thereof. Binding was assessed by bio-light interferometry (BLI) using amine coupling of mAbs to an AR2G biosensor (using standard NHS/EDC activation) followed by association/dissociation of recombinant hSIRPα/His and detection with the Octet RED96 (ForteBio).

For binding ELISA, CHO-K1 cells were transiently transfected with pCI-neo vectors encoding human, mouse or cynomolgus monkey (Macaca fascicularis) SIRP genes. Transfected cells were incubated with indicated mAbs, bound antibodies were detected using goat-anti-mouse IgG-HRP conjugate (Southern Biotech) or goat-anti-rat IgG-HRP conjugate (Jackson ImmunoResearch), visualized with TMB Stabilized Chromogen (Invitrogen), and detected using an EnVision (PerkinElmer).

Binding of anti-human SIRPα mAbs to human SIRPγ was assessed by flow cytometry using NK-92MI cells. Antibodies were incubated at 4 °C, stained with AF647-labeled donkey anti-human IgG conjugate (Jackson ImmunoResearch), and analyzed by flow cytometry (FACSVerse, BD Biosciences).

SIRPα blocking ability was studied using THP-1 and U-937 AML cell lines, where after incubation with FcR Blocking Reagent (Miltenyi Biotec) and indicated mAbs, DyLight 488-labeled recombinant human CD47/Fc-protein (R&D Systems) was allowed to bind at 4 °C and analyzed by flow cytometry (FACSCanto II, BD Biosciences). SIRPα blocking ability on IC-21 cells was studied following incubation with indicated mAbs and recombinant mouse CD47/Fc-protein (R&D Systems) at 37 °C, detection of bound CD47 protein using anti-human IgG-HRP conjugate (Jackson ImmunoResearch), which was visualized with TMB Stabilized Chromogen (Invitrogen) and detected using an EnVision (PerkinElmer).

Cells were phenotypically characterized using a FACSCanto II or FACSVerse flow cytometer with fluorochrome-conjugated mAbs (Additional file: Table S1). Further details can be found in Additional file: Extended methods. 1 2

Human blood was obtained from healthy volunteers who provided informed consent (Sanquin Bloodbank, Nijmegen, The Netherlands) and PBMCs were isolated by Ficoll-paque density gradient centrifugation. CD14+ monocytes were enriched (> 70% purity) using RosetteSep Human Monocyte Enrichment Cocktail (Stemcell Technologies). NK cells were enriched (> 90% purity) using the untouched human NK Cell Isolation Kit (Miltenyi Biotec).

Granulocytes were isolated from erythrocyte-depleted whole blood upon incubation with 10 ng/mL recombinant human interferon-γ (Immunotools) for 1 h at 37 °C in 5% CO₂. Non-adherent blood cells were collected, and the percentage of granulocytes was determined by flow cytometry on the FACSCanto II (based on high FSC and SSC).

Similar procedures where applicable were also applied to EDTA whole blood obtained from healthy cynomolgus monkeys (Biomedical Primate Research Centre (BPRC), Rijswijk, The Netherlands).

Human Burkitt’s lymphoma BJAB cells were labeled with cell proliferation dye eFluor450 (Thermo Fisher Scientific). Labeled cells were mixed with assay medium (RPMI 1640 [Gibco], 10% fetal bovine serum [Gibco] and 100 IU/mL penicillin-streptomycin [Gibco]), indicated mAbs and 0.1 μg/mL rituximab (anti-hCD20), and were then added to human granulocytes (at a ratio of 1:1 tumor cell per phagocyte) and incubated at 37 °C in 5% CO₂ for 2 h. Thereafter, 0.1 μg/mL propidium iodide was added to the mixture and trogocytosis (e.g. visualized as the appearance of eFluor450+ granulocytes) was assessed using the FACSVerse flow cytometer.

Human macrophages were generated from CD14-enriched monocytes cultured in CellCarrier 96-well flat-bottom microplates (PerkinElmer) in medium (IMDM [Gibco], 8.5% fetal bovine serum [Gibco] and 100 IU/mL penicillin-streptomycin [Gibco]) containing 50 ng/mL human monocyte colony stimulating factor (M-CSF) for 7 days at 37 °C in 5% CO₂. Raji cells labeled with eFluor450 were mixed with assay medium, indicated mAbs (anti-CD47 antibodies were used at 66.7 nM (10 μg/mL) and anti-SIRPα antibodies were titrated ranging from 66.7 nM (10 μg/mL) to 6.67 nM (1 μg/mL) and 0.67 nM (0.1 μg/mL)) and 1 μg/mL rituximab, and were then simultaneously added to human macrophages (at a ratio of 2.5:1 tumor cells per phagocyte) and incubated at 37 °C in 5% CO₂ for 2 h. After washing and fixation with 2% formaldehyde, cells were stained with biotin-conjugated anti-human CD19 (eBioscience) for 1 h at room temperature (RT) and Alexa Fluor 488-conjugated streptavidin (Thermo Fisher Scientific). Nuclei were stained with DRAQ5 (Thermo Fisher Scientific) and phagocytosis was analyzed with the Operetta High-Content Imaging System (PerkinElmer). Data were processed and analyzed with Columbus V2.6 software (PerkinElmer). The phagocytosis of tumor cells was quantified counting at least 200 macrophages per sample and using an Uptake Index, as follows: (number of tumor cells inside macrophages/number of macrophages) × 100.

CHO-K1.hSIRPαV1 cells were seeded in CellCarrier 384-well flat-bottom microplates (PerkinElmer) and indicated mAbs were added in assay medium together with human NK cells (at an effector:target cell ratio of 1:5). After overnight incubation at 37 °C in 5% CO₂ cells were washed, stained with Fixable Viability Dye eFluor660 (Thermo Fisher Scientific) and fixed in 5% formalin for 10 min at RT. Fixed cells were washed and nuclei were stained with 1 μg/mL Hoechst 33342 (Life Technologies). Viable target cells were measured with the Operetta High-Content Imaging System and data were processed and analyzed with Columbus V2.6 software.

Human U-937 leukemia cells were labeled with CellTrace CFSE dye (Thermo Fisher Scientific). Labeled U-937 cells were seeded in U-bottom 96-well plates, mixed with indicated mAbs and 20% human complement serum (Sigma-Aldrich) in assay medium, and incubated for 4 h at 37 °C in 5% CO₂. Thereafter, 0.1 μg/mL DAPI was added to the mixture and CDC was assessed using the FACSVerse flow cytometer.

Antibody-mediated activation of FcγRIIA-131H was established using CHO-K1.hSIRPαV1 cells and Jurkat FcγRIIA-131H cells (Promega) at an effector:target cell ratio of 1:2, following the manufacturer’s instructions.

To assess the allogeneic immune reaction, PBMCs from two human donors (referred to as the responder and stimulator (30 Gray (Gy) irradiated) were added together at a R:S ratio of 1.5:1) in presence of mAbs and incubated at 37 °C in 5% CO₂ for 5 days. Supernatants were collected to quantify IFNγ levels by ELISA (Thermo Fisher Scientific). The remaining cells were stained with fluorescent mAbs against CD3, CD4, CD8, CD19 and CD56 for 30 min at 4 °C and analyzed by flow cytometry.

Human PBMCs were seeded in U-bottom 96-well plates, treated with 100 μg/mL of indicated mAbs and 1 μg/mL SEB (Sigma-Aldrich), and incubated for 3 days at 37 °C in 5% CO₂. CD3 blast formation was assessed using the FACSVerse flow cytometer.

EDTA-treated human whole blood collected from healthy donor volunteers was washed with PBS, a 1% erythrocyte suspension (v/v) was prepared in PBS, and 50 μL of the serially (2-fold) diluted mAbs or phytohemagglutinin (PHA-P; Sigma-Aldrich) were incubated with 50 μL of the 1% erythrocyte suspension for 2 h at RT in clear 96-well U-bottom plates. Hemagglutination (visible as loss of RBC “button” formation) was quantitated using the ChemiDoc Touch Imaging System and analyzed with Image Lab 5.2.1 software (Bio-Rad Laboratories).

Blood was collected from healthy donor volunteers who provided informed consent (HaemoScan BV, Groningen, The Netherlands) and buffered with sodium citrate. To assess platelet aggregation (impedance method), blood was diluted with 0.9% NaCl continuously mixed with a stir bar, impedance electrodes were fixed into the blood-containing tubes, and indicated mAbs, adenosine 5′-diphosphate sodium salt (ADP; Sigma-Aldrich) or vehicle (10 mM L-Histidine pH 5.5 containing 0.1 M sodium chloride) were added to the blood suspension. Aggregation was measured for 6 min. The maximum slope of the aggregation curve for the first 3 min was determined from the recordings by “R: A language and environment for statistical computing” (R Foundation for Statistical Computing). To assess platelet activation, the blood suspension was incubated with indicated mAbs, arachidonic acid (Sigma-Aldrich) or vehicle for 1 h at 37 °C. Samples were centrifuged and plasma was collected to execute the thromboxane B2 enzyme immunoassay (Cayman Chemical).

Cytokine release was assessed on sodium heparin-preserved whole blood obtained from 24 healthy donor volunteers who provided informed consent (Sanquin Bloodbank, Nijmegen, The Netherlands). Indicated mAbs were added to polystyrene U-bottom 96-well plates, whole blood was added, and plates were incubated overnight at 37 °C in 5% CO₂. The cytokines IL-6, IL-8, TNF-α, MCP-1, MIP-1α, and MIP-1β in the supernatants were detected using a custom human 6-plex assay kit (Thermo Fisher Scientific) and analyzed on the Bio-Plex MAGPIX multiplex reader (Bio-Rad Laboratories) equipped with Bio-Plex Manager 6.1 software (Bio-Rad Laboratories).

For tumor cell engraftment, 0.75 × 10⁶ Daudi cells (diluted 1:1 with Matrigel) were injected subcutaneously into the left flank of approximately 11-week-old NOD.Cg-Prkdc^scidIL2rg^tm1Wjl/SzJ (NSG) mice, purchased from Charles River Laboratories (France). The animals were observed for tumor growth three times per week, starting 7 days after tumor cell injection. Treatment was initiated once tumors reached a size of 233 mm³ ± 78 mm³. Mice were given intravenous injections of 50 μg rituximab (anti-hCD20, human IgG1) or vehicle (0.9% NaCl) three times per week. In two of the groups that received rituximab, mice were given intraperitoneal injections of 500 μg anti-mSIRPα (clone .20A, mouse IgG1) three times per week, or alternatively mice were given daily intraperitoneal injections of 500 μg anti-hCD47 (clone B6H12, mouse IgG1) for a duration of 4 weeks. Mice were monitored for morbidity and mortality daily. Tumor size was measured three times per week and mice were sacrificed when tumor size reached 2000 mm³. Tumor sizes were measured using a digital caliper and tumor volumes in mm³ calculated with a modified ellipsoid formula: V = (length x width²) × 0.28. Animals were sacrificed when they reached humane endpoint or if they survived till day 34 after start of treatment.

A single-dose toxicity study was conducted at Covance Preclinical Services GmbH (Münster, Germany) according to a written study protocol and facility standard operating procedures in compliance with Institutional Animal Care and Use Committee (IACUC) criteria, national legal regulations on animal welfare, and accepted animal welfare standards. All animals were experimentally naive, purpose-bred cynomolgus monkeys originating from Asia. For the single-dose study, male (n = 4) and female (n = 4) animals were administered a single 15-min intravenous (i.v.) infusion of ADU-1805 (0.3, 3 or 30 mg/kg) or vehicle control (10 mM L-Histidine pH 5.5 containing 0.1 M NaCl). In-life evaluations included clinical observations, body weight, food consumption, standard neurologic and cardiovascular safety pharmacology evaluations, clinical pathology (serum chemistry, hematology, and coagulation), and toxicokinetics. To assess the pharmacokinetic properties of ADU-1805 in cynomolgus monkey serum, blood was drawn on 0, 1, 8, and 24 h, and 3, 8, 11, 15, 22, 29, 36, 43, 59 days post single-dose ADU-1805. Fifty-nine days following the initial dose, the animals were necropsied and examined for gross observations, organ weights, and routine histopathologic evaluation was conducted on formalin-fixed paraffin embedded tissues collected at necropsy.

Data are reported as mean ± standard deviation (SD) as specified. Statistical significance was determined by Student’s t-test or one-way analysis of variance (ANOVA) as indicated, using GraphPad Prism version 8 (CA, USA). All Student’s t-tests were two-sided under the assumption of equal variance between samples. All one-way ANOVA tests were corrected for multiple comparisons using statistical hypothesis testing. Differences were considered statistically significant if p < 0.05.

An unbiased single-nucleotide polymorphism (SNP) analysis of human SIRPα, based on data available at EnsEMBL (https://www.ensembl.org↗), revealed that SIRPαV1, SIRPαV2 and SIRPαV8 are the most prominent haplotypes present among the human population (Fig. 1b). Of these, SIRPαV1 and SIRPαV2 differ the most in their IgV domain sequence (Fig. 1c). While SIRPαV1 is the most abundant allele among European, Admixed American and African populations, the SIRPαV2 allele is the most commonly found allele in East Asian population.

hSIRPα.40A was generated and identified as an antibody that demonstrated potent pan-allele SIRPα binding (i.e. binding human SIRPαV1, SIRPαV2 and SIRPαV8) and lacked appreciable SIRPβ1 binding (Fig. 2a). In contrast, the KWAR23 antibody binds to all SIRPα alleles and also the SIRPβ1 activating receptor. hSIRPα.40A and KWAR23 both bind human SIRPβL [31] and SIRPγ. hSIRPα.40A showed potent antagonism of the two most prevalent SIRPα alleles (e.g. SIRPαV1 and SIRPαV2), as determined by blockade of CD47 binding to U-937 and THP-1 AML cell lines that express SIRPαV1 (data not shown) and SIRPαV2 [32], respectively (Fig. 2b).

The functional activity of hSIRPα.40A was assessed in vitro using a macrophage-based phagocytosis assay (. 3a, b). In this assay human peripheral blood-derived macrophages that endogenously express SIRPα are co-incubated with Burkitt’s lymphoma Raji cells (expressing both CD20 and CD47 (Additional file 3: Figure S1A, B)). In presence of rituximab, hSIRPα.40A augmented tumor-cell uptake (calculated using the uptake index) of Raji cells by macrophages obtained from both SIRPA homozygous (SIRPαV1/SIRPαV1 and SIRPαV2/SIRPαV2) and heterozygous (SIRPαV1/SIRPαV2) individuals (Fig. 3c). The relevance of the unique binding profile of hSIRPα.40A was illustrated by the anti-hSIRPαV1 allele-specific mAb that only enhanced tumor cell phagocytosis by SIRPαV1/SIRPαV1 homozygous-derived macrophages while showing moderate or no phagocytosis by macrophages obtained from SIRPαV1/SIRPαV2 or SIRPαV2/SIRPαV2 individuals, respectively. Overall, this demonstrates the advantages of a pan-allele SIRPα antibody to all homozygous and heterozygous SIRPA individuals.

To evaluate the therapeutic effect of SIRPα blockade in vivo we generated the anti-mouse SIRPα surrogate mAb mSIRPα.20A that specifically bound mouse SIRPα, lacked cross-reactivity to SIRPβ, and blocked CD47 binding, similar to anti-mSIRPα clone p84 (Additional file 4: Figure S2A, B) [33]. Surrogate mAb mSIRPα.20A bound to all mouse SIRPA alleles, including NOD SIRPα which is able to bind to human CD47 (Additional file 5: Table S2) [26]. The ability of mSIRPα.20A to eliminate subcutaneously engrafted Daudi Burkitt’s lymphoma cells in NSG mice (that express the NOD SIRPA allele) was tested in combination with rituximab, analogous to the xenograft model described previously (Additional file 6: Figure S3A) [8]. Mice treated with the combination of mSIRPα.20A and rituximab showed decreased lymphoma burden and significantly prolonged survival compared to rituximab alone, confirming earlier observations (Additional file 6: Figure S3B, C) [25]. The blocking anti-hCD47 mAb B6H12 was taken along for comparison and showed complete inhibition of lymphoma engraftment when combined with rituximab. These results should be compared with caution as NSG mice lack an antigen sink for the anti-hCD47 antibody (e.g. anti-human CD47 does not bind to CD47 expressed on mouse cells).

To allow for human use, the mouse parental hSIRPα.40A antibody was humanized. First, a chimeric version of hSIRPα.40A was generated by grafting the VH and VL sequences of hSIRPα.40A onto the human constant domains of an IgG1, IgG2 or IgG4 heavy chain and human kappa light chain, respectively (Fig. 3d). Although the parental hSIRPα.40A mAb augmented rituximab-induced phagocytosis of Raji cells by human macrophages similar to the anti-CD47 blocking mAb (AB6.12-IgG4PE), the activity of hSIRPα.40A was completely abrogated when its VH and VL sequences were grafted onto a human IgG1 or IgG4 Fc backbone. In contrast, the human IgG2 chimeric variant of hSIRPα.40A retained the activity of the mouse parental mAb. We hypothesized that the mAb Fc of the chimeric hSIRPα.40A interacted with FcγRs present on macrophages, which include at least the high affinity human IgG receptor FcγRI (CD64) and FcγRII (CD32) [34]. Indeed, human IgG1 and IgG4 variants of chimeric hSIRPα.40A bound to FcγRI while the human IgG2 variant did not (data not shown) [35]. In addition, human IgG1 and IgG4 Fc variants that minimize antibody Fc-FcγR interactions restored the enhancement of rituximab-mediated phagocytosis as compared to their wild-type counterparts (Additional file 7: Figure S4A), while similar mutations of the human IgG2 Fc did not further alter macrophage-dependent phagocytosis. Together, these data imply that an anti-SIRPα mAb should be grafted on a human IgG2 backbone to prevent engagement of FcγR on myeloid cells when bound to the antigen (creating a heterotrimeric interaction referred to as the ‘scorpion effect’ [36]) (Additional file 7: Figure S4B).

Subsequently, the mouse variable domains of hSIRPα.40A antibody were humanized by CDR grafting technology using matching human VH and VL frameworks [29], designated ADU-1805. ADU-1805 was confirmed to bind to monomeric human SIRPα antigen with a dissociation constant (KD) of 11 × 10^{− 9} M, similar to the parental and chimeric hSIRPα.40A mAb (Table 1). Moreover, ADU-1805 bound to SIRPα expressed on human monocytes (EC50 = 0.23–1.57 nM) and neutrophils (EC50 = 0.27–1.29 nM) but minimally bound to human lymphocytes (EC50 = 0.94–7.33 nM), which are known to express SIRPγ but not SIRPα [27] (Fig. 4a). Next, ADU-1805 was shown to enhance rituximab-induced phagocytosis, in a concentration-dependent manner, by human macrophages obtained from different human individuals (Fig. 4b). Also, ADU-1805 was shown to enhance rituximab-mediated cell killing by neutrophils in a concentration-dependent manner, through a process called trogocytosis [37] (Fig. 4c, d).

The more restricted expression of SIRPα was hypothesized to allow SIRPα-targeting antibodies to differentiate from CD47-targeting agents. ADU-1805 lacked binding to human RBCs and platelets, and did not trigger hemagglutination, which is in line with its binding characteristics (Fig. 5a, b). Also, SIRPα-targeting with the chimeric hSIRPα.40A mAb did not induce platelet aggregation or activation (Additional file 8: Figure S5). The restricted expression of SIRPα was further demonstrated by comparing the reactivity of ADU-1805 and anti-CD47 towards human PBMCs. Anti-CD47 bound to all cell subsets present in the PBMC fraction (e.g. monocytes, B-cells, T-cells and NK cells), whereas ADU-1805 bound to monocytes and showed only minimal binding to T-cell subsets (Additional file 9: Figure S6). Altogether, based on the presented in vitro data, this confirms the hypothesis that ADU-1805 will show a biological activity profile differentiated from CD47-targeting agents by its more restrictive binding pattern (i.e. no antigen sink, minimal or no effect on RBCs and platelets).

A second potential differentiation was revealed by studying the effect of ADU-1805 on the role of CD47 in cell-cell adhesion through its interaction with SIRPγ on neighboring T-cells [27]. Piccio et al. have demonstrated that blocking the SIRPγ-CD47 interaction with specific antibodies against either CD47 or SIRPγ impaired T-cell activation by CD47+ antigen presenting cells [38]. Hence, we evaluated whether ADU-1805 affected T-cell activation in a PBMC-based allogeneic MLR. ADU-1805 did not alter the T-cell secretion of IFNγ triggered by the allogeneic MLR, whereas anti-CD47 mAb treatment inhibited IFNγ secretion (Fig. 5c). To understand the underlying cause of the reduced IFNγ secretion as seen for CD47-targeting mAbs, we characterized the immune cell subsets that were present at day 5. While the representation of the various cell types remained unchanged in the ADU-1805 and isotype control antibody conditions, anti-CD47 treatment reduced the number of CD4+ T-cells in comparison to its respective isotype control antibody (Fig. 5d). Similarly, we found that anti-CD47 also reduced activation and blast formation of CD4+ T-cells in a SEB-induced T-cell proliferation assay (Additional file 10: Figure S7A, B), whereas ADU-1805 did not appear to affect T-cell activation and proliferation.

To complement the nonclinical antibody development, we demonstrated that ADU-1805 did not engage FcγRIIA, nor did it induce ADCC via FcγR-bearing NK cells (Additional file 11: Figure S8A, B). In addition, ADU-1805 did not induce CDC of the SIRPα-expressing U-937 AML cell line, consistent with the observation that human IgG2 is a poor C1q binder [39] (Additional file 11: Figure S8C). Furthermore, ADU-1805 did not induce cytokine secretion in human whole blood, similar to the FDA-approved human IgG2 antibody panitumumab targeting epidermal growth factor receptor (EGFR) (Additional file 12: Figure S9).

To assess the differentiation of ADU-1805, safety and pharmacokinetics (PK) of ADU-1805 were established in vivo, in a single dose intravenous infusion in cynomolgus monkeys (Table 2). First, two putative variants, SIRPαV1 (NM_001284750.1↗) and SIRPαV2 (XP_015313155.1↗) were identified in cynomolgus monkey, that share 99.2% sequence identity. These variants share a sequence identity of > 91% with human SIRPαV1 and SIRPαV2 and ADU-1805 bound to both cynomolgus variants with an EC50 ≤ 1 nM, similar to its binding affinity for human SIRPα (Additional file 13: Figure S10A). Furthermore, the ADU-1805 binding profile was comparable for human and cynomolgus monkey leukocytes (Additional file 13: Figure S10B).

Upon single-dose administration ADU-1805 measurements in serum followed by PK modelling demonstrated a dose proportional increase in exposure for the two higher dose levels with an estimated half-life of 1.86–6.41 days (Fig. 6a; Table 3). The administration of and exposure to ADU-1805 was well tolerated at all dose levels and no test-article related changes were observed. In contrast to the anti-CD47 mAb Hu5F9-G4 treatment-induced acute anemia in cynomolgus monkeys [40], none of the ADU-1805 doses affected the hemoglobin levels after single-dose administration. This finding supports that targeting SIRPα via ADU-1805 may have a favorable safety profile compared to CD47-targeting agents (Fig. 6b).

All data generated that are relevant to the results presented in this article are included in this article and its supplementary files (Additional files). Other data that were not relevant for the results presented here are available from the corresponding author upon reasonable request.