Anti-CD19/CD20 bispecific antibody with dual Fc domains mediates enhanced effector functions and durable depletion of memory B cells in vivo

Abstract

Recent clinical studies suggest that more potent B cell depleting therapies and targeting more than one B cell antigen may result in improved clinical responses in autoimmune diseases and hematological malignancies. Here we describe an anti-CD19/CD20 bispecific antibody, HB2198, generated using GEM-DIMER™ technology. HB2198 incorporates Fab domains from rituximab and humanized FMC63 (huFMC63) for bivalent binding of both CD19 and CD20 and comprises two enhanced Fc domains to enable powerful effector functions via bivalent binding of Fcγ receptors (FcγR). Enhanced bivalent binding of HB2198 to FcγR was confirmed in vitro. HB2198 demonstrated robust depletion of human B cells that exceeded the levels observed with comparator anti-CD19 or anti-CD20 IgG1 antibodies in vitro. The mechanism of action of HB2198 included enhanced antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis (ADCP), as well as complement-dependent cytotoxicity (CDC) and direct cell killing activity. In cynomolgus monkeys, HB2198 administration resulted in > 99% depletion of circulating B cells within 1–3 days and mediated a durable shift in proportions of naïve and memory B cells in vivo. These data support the conclusion that HB2198 may provide an improved treatment option when potent and broad depletion of both CD19⁺ and CD20⁺ cells is required.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-16461-z.

Introduction

B lineage cells are a key part of the adaptive immune system and comprise cells of diverse maturation stages and functions. While antibody production is the hallmark of differentiated plasma cells, it is now clear that B lineage cells also play a key role in regulating other immune cells, for example, through antigen presentation, costimulation, and cytokine production^1,2. The immune-modulating functions of B lymphocytes have been corroborated in translational and preclinical studies. B cells were shown to modulate anti-tumor responses, and dysfunction of regulatory B cells has been demonstrated in autoimmune diseases, such as SLE^3,4. B lymphocytes infiltrate multiple tumor types and demonstrate predictive and prognostic significance in the context of both chemotherapy and checkpoint blockade further pointing to the important immunoregulatory function of these cells².

Several approved B cell-depleting therapies are available, including antibody-drug conjugates, T cell engagers (TCE), CAR-T cells, and antibody variants with enhanced effector functions through Fc engineering^5,6. The clinical benefit of anti-CD20 antibodies has been demonstrated in B cell malignancies and several autoimmune diseases, such as multiple sclerosis, rheumatoid arthritis, neuromyelitis optica spectrum disorder (NMOSD), ANCA associated vasculitis, and pemphigus vulgaris^7,8. Interestingly, CD20-targeting agents have demonstrated clinical benefits in these autoimmune diseases, although the primary autoantibody-producing cells, plasma cells or plasmablasts, do not express CD20⁹. While CD20 has been the predominant B cell target of interest since the first approval of rituximab in 1997, antibodies targeting CD19, inebilizumab and tafasitamab, have more recently been developed for autoimmune diseases and hematological cancers with demonstrated efficacy in NMOSD and diffuse large B cell lymphoma, respectively^9,10.

A significant unmet need for more efficacious B cell-depleting therapies remains in both oncology and autoimmune diseases. While encouraging results with CAR-T cells targeting multiple B cell antigens have been observed in B cell malignancies in clinical trials¹¹and several antibodies and TCE targeting multiple antigens are in development^5,6all currently approved B cell-depleting agents bind to a single B cell antigen, increasing the potential for incomplete depletion and risk of antigen escape mutants. In autoimmunity, depletion of CD19⁺, CD20⁻ plasmablasts and pro-B cells may provide additional clinical benefits over anti-CD20 antibodies. In line with this hypothesis, recent studies using CD19-targeting CAR-T cells and TCE demonstrated large reductions in autoantibody levels. Disease activity was also improved; for example, SLE Disease Activity Index-2 K (SLEDAI-2 K) in patients with SLE and Disease Activity Score 28 (DAS28) in patients with rheumatoid arthritis^12–14. CAR-T cells have been proposed to be more efficacious than conventional antibodies in B cell depletion in lymphoid tissues, which may explain the therapeutic benefits^12,13. Several engineering approaches have also been introduced aiming to enhance effector functions and, thereby, the efficacy of antibodies. The fragment crystallizable (Fc) region of antibodies plays a critical role in triggering ADCC, ADCP, and CDC and, therefore, has been a key focus for such modifications. For example, afucosylation and introduction of mutations facilitating FcγR binding have been successful in enhancing effector functions of antibodies, including those targeting B cells^10,15–17.

Here, we describe a novel CD19- and CD20-targeting antibody, named HB2198, that was generated by GEM-DIMER^™ technology¹⁸ via the incorporation of ACE2 collectrin-like domains¹⁹ into the hinge regions of the parental CD19- and CD20-targeting antibodies. HB2198 comprises two Fc domains, both of which incorporate FcγR binding enhancing mutations S239D/I332E (SD/IE) for potent effector functions, as well as Fab domains derived from humanized FMC63 and rituximab for bivalent binding to CD19 and CD20, respectively. HB2198 demonstrated potent and robust B cell depletion, ADCC and ADCP in vitro when compared to rituximab or tafasitamab. In contrast to tafasitamab, HB2198 also exhibited CDC and direct cell-killing activity. In cynomolgus monkeys, a rapid and potent depletion of circulating B cells of over 99% was followed by durable remodeling of the repopulating B cell compartment. HB2198 demonstrates significant potential for depletion of CD19⁺ and CD20⁺ cells for the treatment of autoimmune diseases and B-cell malignancies.

Results

Structure and expression of HB2198

HB2198 was generated using GEM-DIMER™ technology¹⁸. The molecule incorporates two Fab domains each from humanized FMC63 (huFMC63) and rituximab for bivalent binding of both CD19 and CD20, respectively, and comprises two enhanced Fc domains to enable powerful effector functions via bivalent binding of FcγR (Fig. 1A). Collectrin-like domains derived from ACE2 receptor mediate spontaneous “superdimerization” of the heavy chains, forming a strong noncovalent bond^18,19. The constant regions of the heavy and light chains are derived from human IgG1 and human Ig kappa, respectively. Each of the heavy chains incorporate the SD/IE double mutation reported to enhance FcγR binding¹⁵. The heavy chain 1 (H1) incorporates the human IgG1 hinge region and a twenty-three amino acid sequence of the human TNF receptor (TNFRSF1B) stem region. The heavy chain 2 (H2) incorporates the complete human IgG1 hinge region. Correct pairing of the heavy and light V regions was facilitated by electrostatic steering²⁰. The HB2198 protein was produced as a single secreted protein product by Chinese hamster ovary (CHO) cells stably transfected with four monocistronic DNA expression vectors, each encoding one of the four polypeptide chain types (H1, H2, L1, and L2). An ELISA assay detecting both CD19 and CD20 Fab domains demonstrated comparable and stable levels of HB2198 in PBS containing 1% BSA and in human serum during a two-week incubation at 37 C, illustrating the stability of HB2198 in both PBS and serum (Fig. S1).

Fig. 1.

Structure and target binding of HB2198 in primary cells and CD19⁺ or CD20⁺ cell lines. (A) Schematic representation of HB2198 structure. Anti-CD19 and anti-CD20 Fab domains are shown in maroon and blue, respectively. The H1 chains incorporate the human IgG1 hinge region and a 23 amino acid sequence of the TNFRSF1B stem region. The H2 chains incorporate the complete human IgG1 hinge region. FcγR binding enhancing mutations (SD/IE) are illustrated with green stars. White circles represent ACE2 domains mediating “superdimerization”. (B) Binding to human B cells was evaluated in PBMCs by incubating the cells in the presence of HB2198 and comparator antibodies as indicated. The cells were stained with secondary anti-human Fcγ-PE and B cell staining cocktail. B cells were identified as viable, CD3^-, CD40⁺, CD21⁺ by flow cytometry. (C–E) HB2198 and comparator antibody binding to Raji WT (C), Raji CD19KO (D) and Raji CD20KO cells (E) was evaluated by flow cytometry. (F–H) HB2198 and comparator antibody binding to wild-type HEK293 (F) and HEK293 transfected with human CD20 (G) or human CD19 cDNA (H). Triplicate mean±SD of geometric mean fluorescence intensity (gMFI) from a representative of 2-3 independent experiments (B–E), and mean±SD gMFI from a 2 independent experiments, each in duplicate (F–H) are shown.

HB2198 demonstrates potent binding to both CD19 and CD20

Flow cytometry and SPR were used to examine the binding of HB2198 to CD19 and CD20. Pharmaceutical grade rituximab (MabThera^®, Roche) and a research grade biosimilar of tafasitamab were used as comparators. Comparable binding and functional activities of the tafasitamab biosimilar and pharmaceutical grade tafasitamab (Minjuvi^®, Incyte Biosciences) were confirmed in cell-based flow cytometry, and ADCC-reporter assays (Fig. S2). HB2198 exhibited dose-dependent binding to human CD3⁻ CD40⁺ CD21⁺ B cells (Fig. 1B), with an EC₅₀ of 3.9 ± 0.27 nM. HB2198 showed somewhat higher maximal fluorescent signals than rituximab, and approximately 3-fold higher maximal fluorescence than huFMC63 or tafasitamab. It should be noted that an anti-human Fcγ antibody was used as a detection antibody, which may result in higher signal when detecting the dual Fc domains of HB2198.

SPR was used to measure monovalent affinities with immobilized antibody test articles and CD19 and CD20 reagents in the mobile phase. HB2198 bound to human CD19 and human CD20 with K_D values of 1.5 nM and 0.20 nM, respectively, which was comparable to the target binding affinities of rituximab, tafasitamab and huFMC63-SD/IE (Table 1, Fig. S3). Thus, HB2198 preserves the binding affinities of the two parental antibodies which it is derived from.

Table 1.

Binding of HB2198, rituximab, huFMC63 and tafasitamab to human CD19 and CD20.

	KD (mean ± SD, nM)
	Human CD19	Human CD20
HB2198	1.5 ± 0.92	0.20 ± 0.09
Rituximab	NB	0.46 ± 0.13
huFMC63	2.7 ± 0.85	NB
Tafasitamab	1.4 ± 0.50	NB

HB2198 and huFMC63 n = 11; rituximab n = 10; tafasitamab n = 4 determinations.

NB = no measurable binding.

To further illustrate potent and specific binding to both CD19 and CD20 on cells, binding of HB2198 to CD19 and CD20 on the surface of B lymphoma cells lines Raji wild-type (WT) (CD19⁺ CD20⁺), Raji CD19KO (CD19^- CD20⁺), Raji CD20KO (CD19⁺ CD20^-) cells was evaluated. The results show potent binding of HB2198 to all the cell types, while rituximab and tafasitamab only bound to cells expressing the relevant target antigen (Fig. 1C-E). Moreover, transfected HEK293 cells were examined using flow cytometry (Fig. 1F–H). On non-transfected HEK293 cells, no binding was observed for HB2198, rituximab, huFMC63, or tafasitamab (Fig. 1F). HB2198 showed binding similar to that of rituximab and tafasitamab on human CD20 and human CD19 transfected HEK293 cells, respectively (Fig. 1G, H).

HB2198 demonstrates strongly enhanced bivalent binding to Fcγ receptors

Both HB2198 and tafasitamab contain SD/IE mutations to increase binding affinity to CD16a, CD32a, and other FcγR, while HB2198 has dual Fc domains as opposed to the one conventional Fc domain of tafasitamab. The monovalent affinities of “superdimeric” antibodies with dual Fc domains to human FcγR are very similar to conventional antibodies, such as rituximab, when analyzed by immobilizing the antibodies¹⁸. To support planning for in vivo studies, we also measured the binding of HB2198 to cynomolgus monkey FcγR. As expected, the measured monovalent binding affinities of HB2198 to monkey FcγR were also comparable to conventional antibodies with SD/IE mutations, within 2- and 4-fold when compared to huFMC63-SD/IE and tafasitamab, respectively (Table S1).

However, when bivalent binding can occur, as in the BLI experiments with FcγR immobilized, HB2198 binding was strongly enhanced over the conventional comparator antibodies (Table 2, (Table S2). In fact, the apparent K_Ds of HB2198 for human FcγR, with the exception of the CD32a alleles and CD32b/c, could not be precisely determined due to the very slow dissociation (k_off). Nevertheless, the results show that the binding avidities of HB2198 for CD16a_V158, CD16a_F158, CD16b, CD32a_H131, CD32a_R131, CD32b/c and CD64 were increased by > 1400-fold, > 3800-fold, > 10,000-fold, > 2400-fold, > 5000-fold, 5900-fold, and > 28-fold when compared to rituximab, respectively. In contrast to HB2198, huFMC63-SD/IE and tafasitamab only showed increased affinities to this set of FcγR ranging from 2.0- to 97-fold over rituximab (Table 2). The binding avidities of HB2198 for CD16a_V158, CD16a_F158, CD16b, CD32a_H131, CD32a_R131, CD32b/c and CD64 were increased by > 53-fold, > 88-fold, > 160-fold, > 330-fold, > 440-fold, 200-fold, and > 14-fold compared to tafasitamab, respectively. Thus, the dual Fc domains of HB2198 result in considerably higher avidity for the immobilized FcγR than a conventional antibody containing a single Fc. We also evaluated the binding of HB2198 to human FcRn. HB2198 demonstrated 20.3-fold enhanced binding at pH 6.0 when compared to huFMC63 and rituximab, respectively (Table S5). The other parental antibody of HB2198, huFMC63, had a 5.1-fold tighter affinity for FcRn. In addition, we addressed the potential for non-specific FcγR clustering using an ADCC reporter assay, Jurkat-FcyRIIIa-158 V expressing Luciferase reporter cells (Fig. S5). No target-cell independent FcγR clustering was observed in these experiments suggesting low risk of non-specific activation of FcγR expressing cells.

Table 2.

Binding of HB2198, rituximab, huFMC63, rituximab-SD/IE, huFMC63-SD/IE and tafasitamab to immobilized human FcγR.

	CD16a- 158 V		CD16a- 158 F		CD16b		CD32a- 131 H		CD32a- 131R		CD32b/c		CD64
	KD (nM)	Fold	KD (nM)	Fold	KD (nM)	Fold	KD (nM)	Fold	KD (nM)	Fold	KD (nM)	Fold	KD (nM)	Fold
Rituximab	595.4	1.0	1534.6	1.0	4085.3	1.0	1738.9	1.0	2332.7	1.0	6676.6	1.0	11.3	1.0
huFMC63	431.3	1.4	968.3	1.6	2254.0	1.8	1106.5	1.6	1229.5	1.9	2182.3	3.1	9.0	1.3
huFMC63 S239D/I332E	14.7	40.6	24.5	62.6	42.2	96.8	250.9	6.9	237.9	9.8	275.4	24.2	4.4	2.6
Tafasitamab	21.2	28.1	35.5	43.2	64.2	63.7	231.0	7.5	203.0	11.5	234.2	28.5	5.6	2.0
HB2198	< 0.4	> 1488	< 0.4	> 3836	< 0.4	> 10,213	0.70	2492	0.46	5095	1.13	5893	< 0.4	> 28.2

*Fold represents the K_D of rituximab divided by the K_D of the comparator.

Combined with the CD19 binding data, these results illustrate comparable binding properties of huFMC63 and tafasitamab to both CD19 and FcγR. Due to the clinical validation and FDA approval of tafasitamab comprising the SD/IE mutations, tafasitamab was used in subsequent experiments as a comparator anti-CD19 antibody in the functional characterization of HB2198.

HB2198 is superior to rituximab or tafasitamab in depleting human B cells in vitro

We next evaluated the ability of HB2198 to deplete B cells in overnight cultures of fresh human whole blood or PBMCs derived from healthy donors or SLE patients (Fig. 2). Samples from SLE patients were included due to promising clinical outcomes of B cell depleting therapies in clinical trials^12,21,22 and the reported resistance of SLE patient B cells to depletion when compared to B cells from healthy volunteers²³. Whole blood cultures were of particular interest since these cultures closely represent the environment of circulating human B cells in vivo. Due to potential interference of HB2198 with the binding of anti-CD19 and anti-CD20 detection antibodies, B cells were identified as CD3⁻ CD56⁻ CD21⁺ CD40⁺ cells by flow cytometry as described by Ryan et al.²⁴.

Fig. 2.

HB2198 efficiently depletes B cells from human whole blood and PBMC from healthy and SLE patient donors. Human fresh whole blood from 14 healthy donors (A– C) or five SLE patients (D–F) was incubated 18–24 h with HB2198 and comparator antibodies as indicated. Triplicate or duplicate mean of live CD3⁻ CD21⁺ CD40⁺ B cells were measured by flow cytometry. The composite dose-response curves from all experiments at 0.006–100 nM (range shared for all experiments) are graphed, with means from individual experiments plotted (A,D). Log EC50 and maximal depletion values for B cell depletion are illustrated for all concentration ranges used (ranging from up to 100 to up to 2,500 nM) for healthy (B,C) and SLE patient cultures (E,F). (G–L) Human PBMCs from nine healthy (G–I) or 13 SLE patient (J–L) donors were cultured in the presence of HB2198 or comparator antibodies (four SLE samples were tested in two independent experiments; all other donors were tested once). The composite dose-response curves from all experiments are graphed at 0.00128–100 nM (range shared for all experiments), with means from individual experiments plotted (G,J). Log EC50 and maximal depletion values for B cell depletion are illustrated for all concentration ranges used (ranging from up to 100 to up to 500 nM) for healthy (H,I) and SLE patient cultures (K,L). Dashed lines connect data obtained within the same experiment with same donor cells. Human IgG1 isotype was included as an assay control in D, G, and J. Donor matched values in the presence of HB2198 were compared to rituximab and tafasitamab by ANOVA using Holm-Sidak multiple comparison test (*^,# p < 0.05, **^{, ##} p < 0.01).

HB2198 resulted in enhanced B cell depletion activity when compared to either rituximab or tafasitamab (Fig. 2). HB2198 demonstrated improved potency (lower EC50 values) and maximal depletion in both healthy donor and SLE patient whole blood when compared to rituximab or tafasitamab, respectively (Fig. 2D–F). HB2198 also demonstrated enhanced maximal depletion of healthy donor PBMC B cells when compared to both rituximab and tafasitamab (Fig. 2G–I). In SLE patient PBMCs, HB2198 exhibited higher maximal depletion when compared to tafasitamab (Fig. 2G–L). Consistent with the expected mechanisms of action, activation of NK cells measured by elevated expression of CD69 was also observed in these cultures, with enhanced activity of HB2198 when compared to rituximab or tafasitamab in healthy donors (Fig. S6A). In SLE patients, the background NK cell activation was higher and the further activation induced by HB2198, rituximab or tafasitamab was comparable (Fig. S6B).

HB2198 is superior to rituximab and tafasitamab in ADCC activity in vitro

Several mechanisms of action have been reported for B cell depleting antibodies targeting CD19 or CD20, including ADCC, ADCP, CDC, and direct cell killing activity^25–27. We evaluated the ability of HB2198 to activate ADCC using cocultures of human PBMCs and Raji human B cell lymphoma cells expressing both CD19 and CD20. HB2198 induced potent, dose-dependent ADCC in Raji cells in three separate donors (Fig. 3). The overall cytotoxicity mediated by HB2198 was enhanced over both rituximab and tafasitamab, independently whether the donor was 158 V/V or 158 V/F for FcγRIIIa genotype. In addition, HB2198 induced potent ADCC in Raji CD19 KO and Raji CD20 KO cell lines, indicating that the activity is not dependent on expression of both targets. As expected, rituximab induced ADCC in Raji WT and Raji CD19 KO, but not in Raji CD20 KO cells, while tafasitamab induced ADCC in Raji WT and Raji CD20 KO, but not Raji CD19 KO cells. These data demonstrate superior ADCC activity of HB2198 on Raji lymphoma cell line and indicate that one target antigen is sufficient to trigger ADCC activity by HB2198.

Fig. 3.

HB2198 exhibits potent ADCC activity in vitro. Raji cells (A–C), Raji CD19 KO (D–F) and Raji CD20 KO cells (G - I) were incubated with PBMC that had been treated overnight with IL-2 in 25:1 ratio. HB2198 and comparator antibodies were added as indicated and ADCC activity was analyzed after a 4 h incubation. PBMC donors 1 and 2 were 158 V/V, and Donor 3 was 158 V/F for FcγRIIIa genotype. Triplicate mean ± SD of percent dead (propidium iodide⁺) target cells by flow cytometry is graphed for 3 independent PBMC donors, Donor 1 (A,D,G), Donor 2 (B,E,H) and Donor 3 (C,F,I) as indicated. 4-parameter non-linear regression curves are shown. When compared by ANOVA using Dunnett’s multiple comparison test, in each of the three donors, cytotoxicity by HB2198 was higher than that by rituximab or tafasitamab in Raji WT and Raji CD19 KO cells at four highest concentrations (p < 0.01). In Raji CD20KO cells, cytotoxicity by HB2198 was higher than rituximab (p < 0.01) but not when compared to tafasitamab.

HB2198 is superior to rituximab or tafasitamab in ADCP activity in vitro and retains potent activity in the presence of competing human IgG

ADCP by macrophages is a predominant mechanism of action of anti-CD20 and Fc-enhanced anti-CD19 antibodies in vivo, particularly in tissues^25,26,28. We hypothesized that the enhanced avidity of HB2198 to FcγR may facilitate ADCP activity of HB2198. Competing human IgG downregulates ADCP activity of antibodies, including both wild-type and afucosylated anti-CD20 mAbs²⁹hence, we also performed ADCP assay in the presence of 10 mg/mL human IgG, which is approximately the median serum IgG concentration in the general adult population³⁰.

Primary CD14⁺ cells were differentiated into macrophages with M-CSF and cocultured with target cells. Addition of HB2198 led to a dose-dependent increase in human macrophage uptake of fluorescently labeled Raji WT cells both in the absence and presence of competing human IgG. In the presence of 10 mg/mL human IgG, HB2198 resulted in a higher maximum uptake than either rituximab or tafasitamab (Fig. 4). The potencies of HB2198, rituximab and tafasitamab were not statistically different (EC₅₀ values 2.62 ± 2.7 nM, 7.8 ± 8.7 nM and 1.6 ± 1.9 nM, respectively). These data indicate that HB2198 has increased ADCP activity on Raji WT cells expressing both CD19 and CD20 antigens.

Fig. 4.

HB2198 induces potent ADCP activity in vitro. Raji WT cells (A, B), Raji CD19 KO cells (C, D) and Raji CD20 KO cells (E, F) were incubated with human M2 macrophages for 2 h in 1:2 ratio with HB2198 and comparator antibodies as indicated. In each graph, triplicate mean ± SD of percent of macrophages positive for target cells by flow cytometry is shown with full titration in a representative experiment combined with a summary of AUC values in three independent PBMC CD14⁺ donors. Similar cultures were performed using the same PBMC donors in the absence (A,C,E) and presence (B,D,F) of 10 mg/mL human IgG. Dashed lines connect data obtained within the same experiment with same donor cells. Values were compared by ANOVA using Dunnett’s multiple comparison test (*^,# p < 0.05, **^,## p < 0.01).

HB2198 also resulted in potent phagocytosis of both CD19⁻ CD20⁺ and CD19⁺ CD20⁻ target cells, in the presence and absence of competing human IgG (Fig. 4C-F). As expected, rituximab and tafasitamab had no activity in CD19⁺ CD20⁻ and CD19⁻ CD20⁺ target cells, respectively. These data indicate that HB2198 exhibits ADCP activity on both CD19⁺ or CD20⁺ target cells, further illustrating that one target antigen is sufficient to mediate the activity. Importantly, HB2198 substantially outperformed both comparator molecules, including tafasitamab, in phagocytosis of CD19⁺ CD20⁻ target cells, which is the phenotype of autoantibody-producing plasmablasts³¹.

HB2198 exhibits CDC activity and direct cell killing in vitro

CDC and direct cell death have also been reported to contribute to antibody-mediated B cell depletion, while relative roles in vivo are still under investigation^27,32,33. Direct cell death has been described for both rituximab and Fc-enhanced anti-CD19 antibodies, while anti-CD19 antibodies do not mediate CDC^27,32,33. Given that HB2198 merges an anti-CD19 and an anti-CD20 antibody, it was of interest to evaluate the CDC and direct killing potential of the molecule.

HB2198 induced dose-dependent CDC in Raji and SU-DHL-4 cells with comparable maximum activity between HB2198 and rituximab, while the potency of HB2198 was up to 10-fold reduced from rituximab in both cell lines (Fig. 5). Tafasitamab consistently showed no CDC activity as previously reported³². HB2198 also mediated a direct dose-dependent effect on SU-DHL-4 cell viability and apoptosis, as measured by ATP metabolism and the dual detection of Annexin V and TO-PRO-3 in SU-DHL-4 cells. HB2198 was roughly 10-fold more potent than rituximab and achieved comparable maximal efficacy (Fig. 5C, D). The maximal activity of tafasitamab was lower when compared to either HB2198 or rituximab (Fig. 5C, D. These data suggest that the mechanisms of action of HB2198 include CDC and direct, effector cell-independent, killing.

Fig. 5.

HB2198 induces CDC and direct cytotoxicity in vitro. To evaluate CDC activity, (A) SU-DHL-4 cells or (B) Raji cells were incubated with 20% human serum as a source of complement and with HB2198 and comparator antibodies as indicated for 2 h and cell number determined by ATP luminescence. To evaluate effector cell-independent direct cytotoxic activity, SU-DHL-4 cells were cultured for 48 (C) or 72 h (D) with test articles alone and cell viability determined by ATP luminescence (C) or stained with Annexin V and TO-PRO-3 and analyzed by flow cytometry (D). Triplicate mean ± SD of percentage of dead cells is shown across titration of antibody and 4-parameter non-linear regression curve shown. A representative of 2–3 independent experiments each is shown.

Binding of HB2198 to cynomolgus monkey CD19, CD20 and FcγR and depletion of monkey B cells

To evaluate the relevance of non-human primates as species for in vivo studies, we analyzed the binding of HB2198 to cynomolgus monkey CD19, CD20 and FcγR and its ability to deplete B cells in cynomolgus monkey whole blood. The affinities of HB2198 and rituximab to cynomolgus monkey CD20 were 1.1 ± 0.16 and 1.4 ± 0.11 nM, respectively (Table S3), which is comparable to that observed for human CD20 (Table 1). The binding of HB2198, huFMC63 and tafasitamab to cynomolgus CD19 was considerably weaker than that to human CD19, with average K_D values of 410 nM (n = 11), 470 nM (n = 11) and 1800 (n = 4), respectively^34,35. As expected based on the data with human FcγR (Table 2), HB2198 demonstrated enhanced avidity on cynomolgus monkey FcγR when bivalent binding could occur, with 5- to > 90-fold enhancement over tafasitamab and 25- to > 1000-fold enhancement over rituximab, depending on the receptor (Table S2).

Similar to human whole blood cultures, HB2198 showed potent binding to cynomolgus B cells with comparable EC50 and maximum binding when compared to human B cells (Fig. 6A, Fig. S4, Table S3, Table S5), while both HB2198 and tafasitamab bound weaker to HEK293 cells expressing cynomolgus CD19 compared to HEK293 expressing human CD19, by 67-fold and 15-fold, respectively (Figs. 1H and 6B). HuFMC63 and tafasitamab also had reduced and less consistent binding to and activity on cynomolgus monkey B cells compared to human B cells (Figs. 1B and 6A). These results are consistent with the weak binding of huFMC63 and tafasitamab to cynomolgus monkey CD19 determined by SPR (Table S3). Importantly, HB2198 induced potent (2 ± 0.2 nM) depletion on cynomolgus monkey B cells in overnight cultures, with maximum activity comparable to that observed in cultures of human whole blood (Figs. 2A and B and 6C). These data indicate that cynomolgus monkey is a relevant species for pharmacological evaluation of HB2198.

Fig. 6.

HB2198 binds to cynomolgus monkey B cells and is highly active in depleting B cells in vivo. (A-B) HB2198 binding in cynomolgus monkey PBMCs (A) or HEK293 stably transfected with cynomolgus monkey CD19 cDNA (B) Representative binding profiles of 2–3 experiments are shown. All three cynomolgus monkey PBMC binding experiments are shown in (Fig. S3). (C) HB2198 depletion of B cells (CD3⁻ CD40⁺) in cynomolgus whole blood after overnight culture. (D-J) Pharmacodynamic effect of HB2198 on circulating B cells in cynomolgus monkeys treated with a single dose of 0.2 mg/kg (purple symbols), 1 mg/kg (red), 5 (green), 25 mg/kg (blue) or repeated dose of 25 mg/kg on Days 0 and 14 as indicated (blue). (D–F) Average percent of baseline B cells (live, CD45⁺ lymphocytes that are CD40⁺ CD3⁻ CD14⁻ CD56⁻ CD159a⁻) following a single infusion (D,E) and repeated infusion (F) is graphed over time. (G-I) The proportion of CD27-negative naïve and CD27⁺ memory B cells before and after 3 months (Day 84) post infusion in individual animals following a single infusion (G) and throughout 6 months following repeated infusion (H). Percent reduction in memory cells from baseline in single and two-dose cohorts (I). Time course of naïve and memory B cells over a follow-up period of 6 months for two-dose animals is shown (J). Values were compared by ANOVA using Dunnett’s multiple comparison test (* p < 0.05, ** p < 0.01).

Infusion of HB2198 in cynomolgus monkeys results in rapid and potent B cell depletion

To evaluate the efficacy of HB2198 in inducing B cell depletion in vivo, HB2198 was administered via IV infusion to cynomolgus monkeys at 0.2, 1, 5, and 25 mg/kg as a single dose or at 25 mg/kg twice on Days 0 and 14. Similar to previously published studies²⁴CD40 was used as a CD19- and CD20-independent marker of peripheral blood B cells, because HB2198 may mask the binding of anti-CD19 and anti-CD20 antibodies. No HB2198-related adverse effects were observed at any dose level.

Potent and rapid dose-dependent B cell depletion was observed one day following administration of HB2198 (Fig. 6D, E). A single infusion of 1, 5 or 25 mg/kg led to approximately 99% depletion of CD40⁺ B cells as early as 1 day post infusion and B cells remained approximately 99% depleted at Day 3 post-treatment as measured by flow cytometry. Up to 99.3% and 99.9% depletion of peripheral B cells was observed in 5 and 25 mg/kg cohorts on Day 3, respectively. Peripheral B cells also remained > 95% and > 99% depleted at Day 7 in the 5 mg/kg and 25 mg/kg groups, respectively. B cell counts slowly returned toward baseline with higher dose groups taking longer to repopulate their B cell compartments. At the highest dose examined, the 25 mg/kg infused animals had the slowest recovery, remaining below 50% repletion (average 39% of baseline B cell numbers) at Day 84 (Fig. 6D).

A cohort of four animals received two doses of HB2198 at 25 mg/kg HB2198, 14 days apart, matching the initial clinical dosing schedule used for both anti-CD19 and anti-CD20 antibodies in the treatment of autoimmune diseases^10,21. Approximately 99% depletion of peripheral B cells was observed by Day 1 post infusion and peripheral B cells remained at < 90% until the second infusion 14 days later (Fig. 6F). After the second infusion, CD40⁺ B cell numbers dropped again to > 99% depletion at 24 h post-dose. B cells recovered to approximately 50% of baseline at Day 84 post first infusion and recovered to 87% of baseline by Day 168 (6 months). The number of T cells and NK cells in the lymphocyte compartment were only transiently affected by HB2198, with NK cells transiently depleted on Days 1 and 3, returning to baseline by Day 7 post infusion ((Fig. S7)).

We further tracked the proportion and number of CD27⁺ memory B cells during depletion and repletion of cynomolgus monkeys treated with HB2198. Memory B cells were effectively depleted at Day 1 post-infusion, similar to naïve B cells. Interestingly, the recovery of CD27⁺ memory cells was significantly slower than that of naïve B cells. Memory B cells remained significantly lower than baseline after 3 months for all single-dose groups as well as through 6 months in the two-dose cohort (Fig. 6G to J). A more significant reduction in memory B cells was observed in the two-dose group when compared to the single dose cohorts (Fig. 6I), suggesting a benefit of repeat-dosing for tissue-depletion of memory B cells. Naïve B cells fully recovered during the 6-month follow-up, while memory B cells remained 50% depleted through the entire observation period in the two-dose cohort (Fig. 6J). Overall, these data indicate potent in vivo depletion of B cells in cynomolgus monkeys and suggest durable remodeling of the B cell compartment with an increased ratio of less activated CD27^- naïve B cells to memory B cells.

Discussion

This study demonstrates potent targeting of CD19, CD20 and FcγR with HB2198, an antibody with dual Fc-domains. “Superdimerization” of antibodies via collectrin-like domain from ACE2 receptor allows for the generation of multispecific, multivalent molecules from various combinations of parental antibodies or Fc-fusion proteins¹⁸. HB2198 was generated via dimerization of CD19- and CD20-targeting antibodies derived from huFMC63 and rituximab, respectively. The two Fc domains of HB2198 are functionally bivalent for FcγR binding resulting in enhanced avidity. Consistently, HB2198 mediated superior effector functions in vitro when compared to the parental antibodies and mediated rapid and potent B cell depletion in cynomolgus monkeys.

Antibody-like molecules with two Fc domains in tandem configuration have been described before, with enhanced avidity to FcγR^36–38. However, further development of these constructs has not been reported, partly due to challenges in protein expression and stability. Specifically, anti-Klebsiella pneumonia antibodies with two Fc domains in tandem exhibited relatively poor stability and low production levels and were not advanced into rodent studies for these reasons³⁶. In vitro characterization of an anti-CD20 antibody with dual tandem Fc domains was also reported with enhanced ADCC and ADCP function, while no protein scale-up or in vivo studies were performed^37,38. GEM-DIMER technology enables joining of two full-length antibodies, allowing for parallel orientation of the two Fc domains and for bivalent targeting of antigens of interest, while retaining the many beneficial characteristics of conventional antibodies. In the case of HB2198, initial developability evaluation is encouraging with demonstrated serum stability and successful protein scale-up for studies in non-human primates.

HB2198 resulted in enhanced human B cell depletion when compared to rituximab or tafasitamab in vitro. Multiple mechanisms of action of HB2198 appear to mediate depletion as it demonstrated enhanced ADCC and ADCP activities, and HB2198 also incorporated the CDC and direct cell killing activities of rituximab. In contrast, tafasitamab lacked CDC activity, consistent with published reports³². The direct killing activity of HB2198 also was enhanced over tafasitamab further illustrating the unique combination of mechanisms by which HB2198 can eliminate its CD19-expressing target cells. It is important that an antibody-based drug can outcompete the endogenous IgG for FcγR binding, particularly in diseases with reduced effector cell function, such as in SLE³⁹or when the target cell is surrounded by high levels of immunoglobulin, as in the case of autoantibody-producing cells. HB2198 demonstrated a significantly enhanced ADCP function even in the presence of physiological concentration of human IgG at 10 mg/mL. Given these results and given that macrophages are considered the key cells mediating the activity of therapeutic antibodies in tissues^25,26,28HB2198 has the potential to induce effector function not only by opsonizing target cells but also by efficiently outcompeting endogenous IgG to bind FcγR on effector cells.

Bispecific targeting of CD19 and CD20 may provide therapeutic benefits over therapies targeting either antigen alone. HB2198 exhibited potent binding to both CD19⁻ CD20⁺ and CD19⁺ CD20⁻ cells and it was active on both types of target cells indicating that one antigen binding domain, combined with the dual enhanced Fc domains, was sufficient for the potent activity of HB2198. Therefore, HB2198 has potential for depletion of both CD19⁺ plasmablasts and CD20-expressing pathogenic T cells, which have been reported to occur at least partly through the acquisition of CD20 from activated B cells via trogocytosis⁴⁰. These CD20⁺ T cells play a role in proinflammatory autoimmune responses, thus elimination of CD20⁺ CD19⁻ T cells appears to contribute to the beneficial effects of CD20-targeting B cell depleting therapies in autoimmunity^40–42. In addition to the benefits of dual CD19- and CD20-targeting in autoimmune diseases, bispecific antigen targeting is also expected to reduce the risk of antigen escape variants in the treatment of B cell malignancies, a known mechanism of treatment failure^43,44.

B cell depletion following administration of HB2198 into cynomolgus monkeys was rapid and robust. Depletion of approximately 99% of circulating CD40⁺ B cells was observed already on Day 1, and up to 99.9% depletion was observed on Day 3. These data compare favorably to previous studies with rituximab. Specifically, Ryan et al. reported an average of 87.3% and 88.3% depletion of CD40⁺ B cells three days following administration of rituximab at doses of 10 mg/kg and 20 mg/kg, respectively²⁴. These data are particularly noteworthy because, due to the low cross-reactivity of huFMC63 on cynomolgus monkey CD19, the in vivo activity of HB2198 is likely primarily attributable to CD20 binding combined with the dual Fc domains with enhanced FcγR binding. However, the contribution of CD19 binding in vivo cannot be ruled out, particularly at the higher dose levels. While the relative contribution of enhanced binding to the various FcγR including potential impact on biodistribution or activity via inhibitory CD32B/FcγRIIIB receptor was not evaluated in these studies, the results demonstrate a potent B cell depleting activity of HB2198 in vivo. In addition, immunogenicity of therapeutic antibodies, including B cell depleting antibodies is common in monkeys, and was not specifically addressed in these studies. However, any impact of potential immunogenicity, particularly after the second dose, cannot be ruled out. The preliminary safety assessment of HB2198 is encouraging, with no non-specific FcγR activation observed in a reported cell line in vitro and no HB2198-related adverse effects observed in vivo.

Importantly, the results also illustrate a durable HB2198-mediated remodeling of the B cell compartment as the proportion of naïve CD27⁻ B cells increased and that of CD27⁺ memory B cells decreased following administration. The reduction in the percentage of memory B cells was most significant in animals that received two doses of HB2198, suggesting that repeat dosing further enhanced activity in lymphoid tissues, such as spleen, bone marrow and lymph nodes, where memory B cells reside⁴⁵. An analogous shift in naïve and memory B cells has been observed in autoimmune patients treated with B cell depleting therapies and the rate of memory B cell recovery correlated with the likelihood of relapse^14,22,46,47. In patients treated with CD19-targeting CAR-T or TCE, these changes in B cell compartment are also associated with reductions in autoantibody levels through depletion of CD19⁺ plasmablasts^14,22. It should be noted that memory B cells are more abundant in patients with autoimmune disease than in healthy controls⁴⁷and hence, it is plausible that such changes in memory B cells are more readily observed in patients than in healthy individuals. To our knowledge, this is the first study to demonstrate a durable shift in memory B cell compartment in healthy non-human primates.

Our study has limitations. Low cross-reactivity of huFMC63 in monkeys may have reduced the biological activity of HB2198 in vivo compared to that expected in human CD19 transgenic animals or patients. In addition, our study was carried out in healthy cynomolgus monkeys and, therefore, potential therapeutic benefits could not be directly addressed. Future studies will consider human CD19 transgenic mice, while these models also have limitations due to the different FcγR structure and biology in mice. Because of the multi-specific nature of HB2198, such animal models would ideally use mice transgenic for human CD19, human CD20, and human FcγR. No tissues were collected in the current study. Therefore, tissue-resident B cells were not enumerated, while the significant reduction in the proportion memory B cells suggests potent activity of HB2198 in memory B cell-rich tissues. We plan to address this directly by immunophenotyping of spleen, bone marrow and lymph nodes in a future study.

Conclusion

HB2198 represents a novel class of therapeutic antibodies with dual enhanced Fc domains. HB2198 is a promising candidate to provide broad depletion of both CD19⁺ and CD20⁺ cells. While the dosing schedule of HB2198 in clinical settings remains to be determined, the molecule is likely to offer antibody-like, off-the-shelf convenience of administration, with the ease of readministration when necessary. The demonstrated potent activity in vitro and in vivo supports the advancement of HB2198 in multiple indications where depletion of CD19⁺ and CD20⁺ B cells is needed.

Materials and methods

Cell lines and blood samples

B cell lymphoma lines Raji (ATCC# CCL-86), Raji CD19KO (Sigma ATG001), Raji CD20KO (Abcam ab273871), SU-DHL-4 (ATCC# CRL-2957) were cultured in RPMI1640 + GlutaMAX (Gibco) supplemented with 10% fetal bovine serum (Gibco). Frozen human and cynomolgus monkey PBMCs were purchased from BioIVT and thawed just prior to use. Fresh healthy human (Stanford Blood Center), SLE patient (Cureline, Inc.), and cynomolgus monkey whole blood (AlphaGenesis) was collected in heparinized tubes and shipped overnight at ambient temperature. Human PBMCs, healthy human whole blood, and SLE patient whole blood were obtained with informed consent and approvals by Institutional Review Boards of BioIVT, Stanford Blood Center and Cureline, respectively.

HEK293 cells expressing human or cynomolgus monkey CD19 were generated by transfection using Lipofectamine 3000 (Invitrogen). Stable clones were generated by hygromycin selection on 10 cm dishes and isolated using cloning cylinders. Clones were screened for CD19 expression and used for subsequent binding assays. Plasmids were designed using Vectorbuilder tools (Vectorbuilder, Inc.), with CAG promoter driving expression of human or cynomolgus monkey CD19 (Genbank Accession numbers NM_001178098.2 and XM 005591540.3, respectively). HEK293 cells expressing human CD20 were purchased from ACRO Biosystems (Catalog # CHEK-ATP034).

Expression and purification of HB2198

The HB2198 protein and control antibodies were produced as a single secreted protein product by CHO cells. In vitro studies were performed with HB2198 produced from transient transfection, while a stable cell line was generated for preclinical development, including the study in cynomolgus monkeys. CHO cells were transfected with four monocistronic DNA expression vectors, each encoding one of the four polypeptide chain types (H1, H2, L1, and L2). Generation of stable cell line expressing HB2198 was carried out by ExcellGene SA (Monthey, Switzerland).

The HB2198 and control antibodies were purified by Protein A affinity chromatography. Affinity chromatography was carried out by applying the supernatant to a column packed with the CaptivA Protein A Affinity Resin (Repligen, MA) pre-equilibrated with phosphate buffered saline (PBS pH 7.4). The column was washed with PBS until the OD280 value returned to baseline. Target protein was then eluted with 0.25% acetic acid at pH 3.5. Fractions were collected, buffered with 1 M HEPES, and fractions containing the target protein were pooled, buffer exchanged into 100 mM HEPES, 100 mM NaCl, 50 mM sodium acetate, pH 6.0, filtered through a 0.2 μm membrane filter and stored at 4 ℃ prior to use. The protein concentration was calculated from the OD280 value and calculated extinction coefficient.

The purity of HB2198 used in these studies was ≥95% based on SEC-UV analysis. Endotoxin levels in purified samples were determined using EndoChrome-K Kinetic LAL assay and/or EndoSafe-PTS cartridges (Charles River) according to manufacturer instructions (limit of detection 0.01 EU/mg). Host Cell DNA and protein (HCP) contaminants for HB2198 protein preparation used in in vivo studies were determined using qPCR and ELISA, respectively (both Cygnus Technologies), and were found to be below the limit of detection of the assay and 25.5 ng HCP/mg protein, respectively.

CD19, CD20 and FcγR binding

The binding of recombinant human or cynomolgus monkey CD19 and CD20 proteins to immobilized HB2198 and comparators was measured using SPR on a Carterra LSA platform. HB2198 and comparator antibodies were captured at densities of 588 Resonance Units (RU) and 330 respectively, on a CMDP chip containing covalently coupled anti-human Fc antibody. CD19 binding was conducted in HEPES buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, with 0.005% Tween-20, pH 7.4 with 0.05 mg/mL BSA). CD20 binding using recombinant full-length CD20 was conducted in HEPES buffer containing 0.05% n-Dodecyl-β-D-maltoside (DDM) and 0.01% cholesteryl hemisuccinate (CHS). Data were analyzed using a 1:1 Langmuir binding model with the Carterra Kinetics Software.

Binding to immobilized FcγR was measured by Biolayer Interferometry (BLI) using an Octet Red384 (Sartorius Stedim Biotech, France) with HEPES buffer at 26 °C with orbital shaking speed of 1000 rpm. Human FcγR were captured on anti-penta-histidine biosensors (Sartorius) at 5 µg/mL. Biosensors were then incubated in varying concentrations of test compounds for 15 s, followed by a 60 s dissociation incubation in HEPES buffer. Data was analyzed using a standard 1:1 binding model. Twofold serial dilutions were analyzed with the following top concentrations: CD16a_V158, 2 µM; CD16a_F158, 5 µM; CD16b, 12 µM; CD32a_H131, 10 µM; CD32a_R131, 10 µM; CD32b/c 10 µM; and CD64, 2.5 nM. Data were analyzed using a 1:1 Langmuir binding model with the Octet Analysis Studio 13.0 (Sartorious).

Cell binding

Target cells or PBMCs were incubated with blocking cocktail (5 µg/mL human BD Fc Block^™, BD Biosciences; and 10 µg/mL human AffiniPure^™ F(ab’) Fragment IgG, Fcγ specific, Jackson ImmunoResearch), washed with staining buffer (1xPBS/2%FCS/1 mM EDTA), and cells (2 × 10⁴, target cells; 1 × 10⁵, PBMCs) were plated in U-bottom non-coated assay plates with a titration of test antibodies for 30 min on ice. Cells were washed, then incubated with Live/Dead-NIR (Thermo Fisher) and 5 µg/mL PE (phycoerythrin) anti-human IgG, Fcγ specific polyclonal antibody (eBiosciences) for 30 min on ice. Cells were washed, fixed with 1% PFA, and returned to staining buffer then analyzed with flow cytometry (Novocyte, Agilent). Binding data was measured by PE⁺ live lymphocytes (target cells) or PE⁺ live B cells (PBMCs) determined by CD3^-CD40⁺ (cynomolgus monkey) and CD3^-CD21⁺CD40⁺ (human) using FlowJo software (BD Biosciences). 4-parameter non-linear regression curves are shown.

Direct cell killing

Target cells were plated in RPMI medium at a density of 5 × 10⁴ cells per well into 96-well plates and cocultured with titrated antibody for 48 h at 37 °C/5% CO₂. Cell viability was then assessed through luminescent ATP (CellTiter-Glo) and read on a SpectraMax iD3 plate reader. Specific cytotoxicity was calculated as follows: 100 × (1- (treated sample/untreated sample)). For apoptosis measurement, cells were cultured as above for 72 h and cell viability assessed by flow cytometry after 20 min stain with AnnexinV-alexa488 and live-dead dye TO-PRO-3 (Thermofisher) according to manufacturer’s instructions. The percentage of AnnexinV + TO-PRO-3 + cells was determined after gating on all cells by forward and side scatter.

B cell depletion in human and cynomolgus monkey whole blood

Heparinized human or cynomolgus monkey blood (90 µL) was incubated with 10 µl of test antibodies. After 18 h, human samples were stained with antibodies against CD3, CD56, CD69, CD21 and CD40 for 30 min at 25 °C. The same antibody cocktail was used for monkey samples except for CD159a, which was substituted for CD56. Samples were depleted of erythrocytes with RBC Lysis Buffer (EBiosciences) for 15 min at 25 °C before washing twice and resuspended in FACS Buffer (BD Biosciences) before being analyzed immediately with flow cytometry (Novocyte). Lymphocytes were gated and B cells were determined by the number of positive CD3^- CD56^- CD21⁺ CD40⁺ (human) and CD3^- CD159a^- CD40⁺ (monkey) cells. Percent B cell depletion was calculated as follows: 100 × (1-(treated sample/untreated sample)). NK cell activation was analyzed from the same samples by gating on lymphocytes and determining the number of CD69⁺ CD56⁺ (human) or CD159a⁺ (monkey) NK cells.

B cell depletion in PBMCs

Freshly thawed PBMCs were incubated with indicated test articles for 18–24 h. For flow cytometry analysis samples were washed with FACS buffer (1xPBS/2%FCS/1mM EDTA) and stained with Live/Dead NIR and antibodies to CD45, CD3, CD56, CD14, CD21, and CD40 for 20–30 min on ice. Cells were washed once and fixed with 1% PFA for 10 min, returned to FACS buffer and collected on ACEA Novocyte flow cytometer (Agilent). Results were analyzed in FlowJo software (Treestar) by gating live CD45 positive cells and determining the number of CD3- CD56- CD40 + CD21 + positive B-cell populations. Percent B cell depletion was calculated using untreated samples as a 100% control in Excel and graphed in Prism. Resulting mean ± SD are graphed in Prism with 4PL regression, EC50 and Emax reported.

ADCC

Target cells were first labeled with ViaFluor 405 (VF405) (Thermofisher) before plating 1 × 10⁴ cells per well in a 96-well round bottom plate. Labeled target cells were then incubated with test antibodies at 7 concentrations using 5-fold dilutions in a total volume of 100 µL. After 15 min in a 37 °C CO2 incubator, 2.5 × 10⁵ PBMCs were added at an E: T ratio of 25:1. After a 4 hour incubation at 37 °C, 5% CO₂, cells were washed, stained with propidium iodide (PI) and analyzed by flow cytometry. Cytotoxicity was expressed as the percentage of cells staining positive for both VF405⁺ and PI.

ADCP

Target cells were labeled with 1 µM pHodo GreenAM (Thermofisher) for 30 min per manufacturer’s instructions and 2 × 10⁴ cells plated in RPMI1640/10% FBS with the addition of test antibody titration in triplicate 10 min prior to the addition of effector cells. M2 macrophages were differentiated with M-CSF (peprotech) from primary human CD14⁺ cells (BioIVT) for 5–7 days and labeled with 1 µM CellTrace Violet (CTV, Thermo Fisher) and washed twice with RPMI/10% FBS prior to use. 1 × 10⁴ macrophage cells were added as effectors cells at E: T ratio of 1:2, with pHrodo-Green labeled target cells and cultured at 37 °C, 5% CO₂ for 2 h. Cells were stained with Live Dead (Near-IR, Thermo Fisher) and washed once in FACS Buffer (PBS/2%FCS/1mM EDTA). Uptake was measured as percent of live pHrodo⁺ macrophages by flow cytometry (Novocyte ACEA, Agilent). Data were analyzed in Flowjo software (TreeStar).

CDC

Raji target cells at a density of 2–3 × 10⁴ per well were plated in 96-well plates and titrated test antibodies were added 10 min before adding human serum (Sigma) to 25% final concentration. After a 2-hour culture at 37 °C 5% CO₂, cell viability was assessed with CellTiter-Glo (Promega) and bioluminescent ATP was read on a SpectraMax iD3 plate reader according to the manufacturer’s recommendations. Percent CDC activity was calculated as follows: (Sample-MIN/MAX-MIN) *100 using luciferase unit (RLU) of 1% Triton-X treated samples for 15 min as 100% lysis (MAX) and RLU of Untreated wells as 0% lysis (MIN).

In vivo study in cynomolgus monkeys

In vivo effects of HB2198 on B cell depletion were evaluated after a single IV infusion of 0.2, 1, 5 or 25 mg/kg and after a repeated dose of 25 mg/kg (on Days 0 and 14) in cynomolgus monkeys (Macaca fascicularis). Two or four age- and sex-matched animals were administered using one-hour infusions in the single and two-dose groups, respectively. Blood samples were collected for immunophenotyping for 3 months (84 days) for single infusion cohort and 6 months (168 days) for the two-dose cohort. For blood count, whole blood was collected in EDTA tube and complete blood count (CBC) run on Beckman Coulter DX H600 Hematology analyzer. Flow cytometry of whole blood was performed as previously described to monitor the number of CD40⁺ CD3⁻ NKG2A⁻ CD14⁻ CD56⁻ B lymphocytes. All work was performed under study protocols reviewed and approved by University of Louisiana Lafayette’s Animal Care and Use Committee (IACUC) and in accordance with regulation of USDA animal welfare Act (CFR 9 sections 1,2,3).

Statistical analysis

Statistical analyses were performed using GraphPad Prism (San Diego, California USA, www.graphpad.com). Statistical significance was determined by ordinary one-way ANOVA with Holm-Sidak or Dunnet’s multiple comparisons post-test.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ADCC

Antibody-dependent cellular cytotoxicity

ADCP

Antibody-dependent cellular phagocytosis

CDC

Complement-dependent cytotoxicity

CHS

Cholesteryl hemisuccinate

DAS28

Disease Activity Score 28

DDM

n-Dodecyl-β-D-maltoside

huFMC63

humanized FMC63

NMOSD

Neuromyelitis optica spectrum disorder

PBMC

Peripheral blood mononuclear cell

Fc

Fragment crystallizable

FcγR

Fcγ receptors

SLE

Systemic lupus erythematosus

SLEDAI-2K

SLE Disease Activity Index-2 K

SPR

Surface Plasmon Resonance

TCE

T cell engager

Author contributions

All authors contributed study design, analysis and interpretation of the results. GL, NLSC, BF, LAT, BL, UE, MEF, SJC executed experiments. JP, DJC provided project administration and supervision. All authors read and approved the final manuscript.

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Competing interests

All authors are employees of Hinge Bio, Inc. and each may hold shares in Hinge Bio, Inc. and/or have stock option agreements. D.J.C. is an inventor on patent applications related to this work (U.S. Application No. 17/373,751, “Tetrahedral Antibodies”; U.S. Application No. 18/153,840, “Tetrahedral Antibodies”; U.S. Application No. 18/188,412, Topologically Engineered Superdimeric Antibodies”; U.S. Application No. 18/949,940, “NK/Monocyte Engagers”; and “U.S. Application No. 19/050,008, “NK/Monocyte Engagers”) and a member of Biomolecular Holdings LLC, the assignee of the patent applications.

Ethics approval and consent to participate

All methods were carried out in accordance with the relevant local guidelines and regulations, and all experimental protocols were approved by the local institutional committees. All animal work was performed under study protocols reviewed and approved by University of Louisiana Lafayette’s Animal Care and Use Committee (IACUC) and in accordance with regulation of USDA animal welfare Act (CFR 9 Sects. 1,2,3). The studies were designed in accordance with ARRIVE guidelines. No human participants were directly involved with the study, while human samples were commercially purchased for in vitro experiments. Human PBMCs, healthy human whole blood, and SLE patient whole blood were obtained from BioIVT, Stanford Blood Center, and Cureline, respectively, with informed consent and approvals by Institutional Review Boards of BioIVT, Stanford Blood Center and Cureline, respectively.