Hemophilia A is an X-linked bleeding disorder caused by a deficiency in functional Factor VIII (FVIII) protein levels. Individuals with severe hemophilia A (plasma FVIII levels < 1%, which represent approximately 60% of patients), require replacement of FVIII to levels sufficient for maintaining hemostasis, and reducing the likelihood of joint bleeding and damage (1). Factor VIII replacement therapy typically includes routine prophylactic dosing levels of Factor VIII products at 20-50 International Units (IU) per kilogram of body weight. Currently, a wide variety of therapeutic products are available for the treatment of hemophilia A patients. These biological products include plasma-derived FVIII concentrates, recombinant FVIII analogs, a bispecific FVIII mimetic, and an adeno-associated virus vector-based gene therapy (2, 3). However, a major shortcoming of many plasma-derived and recombinant FVIII products is the relatively short circulating half-life (~12-15 hours), thus requiring frequent dosing (4). In the past two decades, drug manufacturers have designed novel protein products to overcome FVIII’s intrinsic short plasma FVIII half-life. Such so-called extended half-life (EHL) engineered FVIII products include PEGylated and Fc-fusion protein versions (3). Fc-fusion is a protein engineering strategy used to extend the circulating half-life of a bioactive fusion partner (peptide or protein) via genetic fusion with the immunoglobulin G (IgG) Fc domain which mediates interactions with the neonatal Fc receptor (FcRn) (5–7). Fc binding to FcRn within acidified endosomes avoids intracellular protein catabolism and shuttles the Fc-containing protein into a naturally occurring protein recycling pathway (8). The FcRn-mediated recycling pathway is responsible for the relatively long serum half-lives (~ 21 days) of IgG and albumin.

Antihemophilic Factor (Recombinant), Fc Fusion Protein is a marketed FVIII product for the treatment of hemophilia A. Prior in vitro studies have demonstrated that the recombinant Factor VIII-Fc fusion protein (rFVIIIFc) interacts with high-affinity and low-affinity Fc gamma receptors (FcγR) expressed by different immune cell populations (9–13). The immunomodulatory effects of the rFVIIIFc-FcγR interactions differ by the receptor and cell population, but include both activating and inhibitory pathways, similar to canonical IgG1 Fc-FcγR interactions.

We previously demonstrated that rFVIIIFc, unlike rFIXFc, activates natural killer (NK) cells via Fc-mediated interactions with FcγRIIIA (CD16) (12). Furthermore, we demonstrated rFVIIIFc-activated CD16⁺ NK cells lyse an anti-FVIII B cell clone (12). Here, we used human NK cell lines and primary NK cells enriched from peripheral blood leukocytes to further characterize rFVIIIFc-mediated NK cell activation. We identified the FVIII light chain (LC) as playing a key role in the activation of CD16⁺ NK cells by rFVIIIFc. Using antibodies that target different FVIII domains, we demonstrated that blocking C1 or C2 domain-mediated membrane binding. potently inhibited rFVIIIFc-mediated CD16⁺ NK cell activation, while targeting FVIII heavy chain (HC) domains did not. Our results suggest FVIII LC-mediated membrane binding possibly increases binding avidity of Fc to CD16, resulting in NK cellular activation. Moreover, we show that rFVIIIFc binds to CD16 with higher affinity than rFIXFc, and with similar affinity as obinutuzumab, a glyco-engineered (afucosylated) therapeutic IgG1 monoclonal antibody. Typically, low affinity FcγR (i.e., FcγRIIA, FcγRIIB, and FcγRIIIA) activation and functional response requires receptor crosslinking and increased avidity via binding of multiple IgG1 Fc domains with cell surface receptors (14, 15). We propose a working model in which we hypothesize that the LC C1 and C2 domains of the FVIII component of rFVIIIFc bind the NK cell plasma membrane in addition to Fc domain engagement with CD16. Additionally, the FVIII moiety of rFVIIIFc, unlike the FIX moiety of rFIXFc, increases the affinity of the Fc binding to CD16. Together, the FVIII LC- and Fc-mediated interactions result in increased avidity and NK cell activation. Our previous findings and working model provide a plausible explanation for retrospective case reports (16–18) and a prospective clinical study [NCT03093480] (19) suggesting that rFVIIIFc may rapidly tolerize hemophilia A patients with inhibitors (neutralizing anti-drug antibodies) during immune tolerance induction (ITI) therapy. In light of our findings, we also speculate on broader implications for design of novel Fc-fusion proteins to serve as antigen-specific B cell depletion agents.

Functional characterization of the rFVIIIFc fusion protein through in vitro studies, as described here and previously by our group (12) and others (9–11, 13), have shown diverse Fc-mediated outcomes with respect to the immune system. These studies have demonstrated that rFVIIIFc interactions result in a wide range of Fc effector functions including activation of CD16⁺ NK cells (12), induction of regulatory macrophage polarization (9), inhibition of B-cell activation (10), activation of dendritic cells (DCs) (11), and inhibition of CD14⁺ blood monocyte differentiation into osteoclasts (13). We previously demonstrated the NK cell lines used in this study (NK-92 and PTA-6967) do not express high-affinity FcγRI/CD64 or low-affinity FcγRII/CD32 receptors (12), enabling us to elucidate the role of rFVIIIFc on FcγRIIIA/CD16-mediated cellular activation. We observed that rFIXFc, a fusion protein sharing an identical IgG1 Fc domain amino acid sequence as rFVIIIFc (26–28), did not activate CD16⁺ NK cells ( Figure 3A ). Thus, we hypothesized that the FVIII moiety of rFVIIIFc may play a role in CD16⁺ NK cell activation. Herein we have demonstrated that the FVIII LC domains influence the rFVIIIFc Fc effector function.

We observed that rFVIIIFc functions differently, with respect to the Fc effector function, when compared to other IgG1 Fc-containing therapeutic proteins (29). We have demonstrated the rFVIIIFc fusion protein potently activated CD16⁺ NK cells in the absence of other cell types (i.e., target cells), which, in the case of many therapeutic IgG1 mAbs, is generally a prerequisite for CD16⁺ NK cell activation and the antibody-dependent cellular cytotoxicity (ADCC) mechanism of action due to FcγR crosslinking and avidity effects. In the context of mAbs, there is evidence that distal regions/domains of the same molecule, such as the antigen-binding Fab region, contribute to stabilizing Fc-FcγRIII interactions (30–32) and impact Fc effector function. Furthermore, antigen binding by mAbs can also impact Fc-FcγR interactions (33, 34). In the case of Fc-fusion proteins, the Fc domain is covalently linked to a bioactive fusion partner, therefore, it would not be unreasonable to speculate that the bioactive fusion partner’s biology which includes binding to other proteins, lipids, or carbohydrate molecules may influence Fc binding and functional properties.

FVIII and FIX are plasma proteins with distinct structural and functional properties that are integral to maintaining hemostasis as part of the blood coagulation cascade. The FVIII zymogen circulates as a heterodimeric protein consisting of a HC made up of A1 and A2 domains, a B domain, and the LC made up of A3, C1, and C2 domains. Following proteolytic cleavage and activation by thrombin or activated Factor X (FXa), activated FVIII (FVIIIa) acts as an enzyme cofactor by forming a complex with activated Factor IX (FIXa) in the presence of Ca²⁺ and phospholipids. The FIX zymogen circulates as a multi-domain protein consisting of a gamma-carboxyglutamic acid (Gla) domain, two epidermal growth factor domains, and a trypsin-like peptidase domain. FIXa, a serine protease in complex with FVIIIa, cleaves and activates Factor X. Importantly, the FVIII LC C1 and C2 moieties contain discoidin-like domains (35). The FVIII C1 and C2 domains belong to a family of discoidin protein domains, which share similarity to the discoidin lectin of Dictyostelium discoideum (a slime mold) (36, 37). The D. discoideum discoidin lectin binds to galactose and promotes binding to anionic phospholipids (36). Mammalian proteins including coagulation factor V, factor VIII, and lactadherin each contain discoidin-like domains that form membrane-binding β-barrel structures (38–40). Biochemical studies have elucidated the role of the FVIII C1 and C2 domains in membrane phospholipid binding (41–43).

Here ( Figure 3A ) and previously (12), we have observed differences between rFVIIIFc and rFIXFc with respect to activation of NK cells despite shared human IgG1 Fc domain amino acid sequences (26–28). In sharp contrast to rFVIIIFc, rFIXFc did not activate primary human NK cells ( Figure 3A ). We hypothesized differences in the structure and function of FVIII and FIX likely account for differences in Fc effector function and CD16⁺ NK cell activation potential. Specifically, we demonstrated FVIII bound the NK cell surface ( Figure 1B ) but was not sufficient to activate NK cells ( Figures 1A , 2A ). Furthermore, pre-incubation of NK cells with rFVIII inhibited rFVIIIFc-mediated cellular activation while pre-incubation with rFIX did not ( Figure 3C ), suggesting direct competition of NK cell surface binding sites by rFVIII. However, it is important to note that in our in vitro experiments we have used inactivated forms of rFVIII, rFVIIIFc, rFIX, and rFIXFc, thus the results may not fully reflect the functional properties of the activated forms. With this limitation, we investigated the role of the FVIII moiety with respect to the rFVIIIFc-mediated activation of NK cells. Adding antibodies specific to different domains of FVIII as inhibitors of NK cell activation revealed that the C1 and C2 domains of the FVIII LC were involved in NK cell activation, but the A1 and A2 HC domains were not.

Subsequently, we demonstrated that LRP1 may not be the major cell surface receptor for FVIII on NK cells ( Figures 5A, B ). However, we did show that lactadherin, a phospholipid-binding protein containing discoidin domains similar to FVIII C1 and C2 domains, did compete with rFVIIIFc binding to the plasma membrane and prevent NK cell activation ( Figure 5C ) suggesting that FVIII-mediated phospholipid-binding may promote Fc-CD16 interactions.

Our findings implicating FVIII LC domains in rFVIIIFc Fc effector function ( Figure 4 ) are consistent with recent findings by Duan et al. (13), demonstrating that rFVIIIFc inhibits CD14⁺ blood monocyte differentiation into osteoclasts via Fc-FcγRII interactions on the monocyte surface. Moreover, Duan et al., demonstrated that rFVIIIFc-mediated inhibition of monocyte-derived osteoclast formation is dependent on FVIII C1 and C2 domain binding on the monocyte surface, described as a “double touchpoint” model. Interestingly, FVIII C1 and C2 domains have also been linked to poor transplacental delivery of rFVIIIFc in mouse models (44) suggesting the rFVIIIFc FVIII LC domains may also impact Fc interactions with the neonatal Fc receptor (FcRn), responsible for transcytosis of IgG across syncytiotrophoblasts and transplacental delivery of maternal IgG to the fetus (45). While the exact nature of rFVIIIFc and FcRn interactions as related to transplacental transfer are still unknown, rFVIIIFc-FcRn interactions do confer an enhancement in circulating half-life as compared to non-Fc-fused FVIII products, demonstrated in animal models (46) and in clinical trials (47, 48).

Interestingly, while we have demonstrated that the FVIII LC domains play a role in rFVIIIFc-mediated CD16⁺ NK cell activation, our measurements of binding kinetics suggest that the FVIII moiety impacts Fc binding properties even in the absence of a cell surface. Using a cell-free soluble binding system, which would negate FVIII C1 and C2-mediated cell surface binding, we observed differences in binding kinetics between rFVIIIFc and rFIXFc ( Table 1 ), with rFVIIIFc having similar FcγRIIIA binding kinetics as obinutuzumab, an afucosylated IgG1 mAb glyco-engineered for enhanced engagement with CD16 (49). These data suggest that the FVIII moiety may not only provide a means for attachment to the cell surface providing avidity and allowing monomeric Fc to overcome the low-affinity FcγR activation threshold, but the presence of the FVIII moiety also likely affects the Fc domain’s affinity for CD16. This finding is consistent with our earlier finding that Fc-FcγR and Fc-complement C1q binding and signaling properties are influenced by the active molecular entity covalently attached to the Fc in different fusion proteins (29). This may help explain rFVIIIFc’s pleotropic effects on many different FcγR-expressing cell types in in vitro studies. Interestingly we observed that rFVIIIFc binds CD16 with a ~5-fold higher affinity than rFIXFc, attributable to differences in both the association and dissociation rate constants ( Table 1 ). These results could be explained by our previous findings (29) demonstrating that Fc-Fc receptor interactions are influenced by the molecular entity that is fused to the Fc fragment. This is also consistent with observations that different Fc-fusion proteins have different serum half-lives (6). Similarly, in general, the half-life of Fc-fusion proteins is considerably lower than that of antibodies. Taken together these in vitro data and preclinical and clinical observations suggest that allosteric effects attributable to the ‘fusion-partner’ affect the Fc binding kinetics.

Overall, our findings suggest that the FVIII moiety plays a role in rFVIIIFc CD16 signaling, but FVIII alone is not sufficient for pro-inflammatory NK cell activation. In Figure 6 , we propose a model for rFVIIIFc interactions with CD16⁺ NK cells. In our proposed model, we hypothesize (i) the rFVIIIFc FVIII C1 and C2 domains bind the plasma membrane; (ii) with the FVIII moiety tethered, the flexible hinge region promotes enhanced engagement of a high affinity Fc domain with CD16; (iii) upon crosslinking of multiple CD16 receptors, the intracellular signaling cascade is initiated; and (iv) resulting in NK cell activation, including release of pre-formed cytolytic granules (degranulation) and pro-inflammatory cytokine secretion (IFNγ). Our proposed model is supported by a detailed rFVIIIFc structural study demonstrating the flexibility and rotational freedom at the fusion site (Fc hinge region) and functional independence between the rFVIII and Fc moieties (50).

While the pleotropic effects of rFVIIIFc have been demonstrated in multiple in vitro studies and associated with multiple FcγR-expressing leukocyte populations including monocytes, macrophages, DCs, B cells, and NK cells (9–13), our data may provide a plausible molecular mechanism underlying the potential of using rFVIIIFc in ITI therapy to eradicate inhibitors as described in retrospective case reports (16–18) and a prospective ITI study (verITI-8) (19). However, we acknowledge that in vitro studies have limitations, and do not claim the observations reported herein fully capture the complexity of in vivo systems or the clinical experience of patients with hemophilia A treated with rFVIIIFc. Consistent with other in vitro studies characterizing the Fc effector function of rFVIIIFc, we measured NK cell activation using hyper-physiological FVIII concentrations, which do not correlate with circulating rFVIIIFc levels during standard prophylaxis treatment. However, the concentrations we used in vitro may be more relevant to the use of rFVIIIFc during high-dose ITI regimens (Bonn protocol). In the ITI regimen patients with hemophilia A and high titer inhibitors are administered daily doses of rFVIIIFc at 200 IU/kg. Additionally, in circulation, there is likely to be limited accessibility to the FVIII C1 and C2 domains, as FVIII forms a noncovalent complex with vWF (see Figure 3C for vWF-mediated inhibition). Based on the in vitro data, in vivo vWF could considerably inhibit rFVIIIFc from activating CD16⁺ NK cells at the physiological vWF concentration (~35 nM) (21), however, higher rFVIIIFc dosing levels during ITI therapy may exceed the vWF inhibitory threshold and allow for rFVIIIFc Fc-mediated immunomodulatory properties. Of course, in vitro studies are unable to model the biological outcomes in circulation where rFVIIIFc can interact with multiple FcγR-expressing immune cell types at the same time.

More broadly, Fc-fusion proteins have previously been designed to act specifically as antigen-specific B cell depletion agents. For example, Fc fusion proteins with human myelin oligodendrocyte glycoprotein (MOG) or a myelin basic protein immunodominant peptide have both been shown to specifically deplete myelin-reactive B cells using in vivo animal models (51, 52). These studies provide proof-of-concept for the use of Fc-fusion proteins as antigen-specific B cell depletion agents. Perhaps novel Fc fusion proteins can be designed using Fc domains engineered with enhanced affinity to CD16 to create potent antigen-specific B cell depletion agents. One could speculate that such a fusion protein may have clinical utility in the case of autoimmune disorders with a well-characterized antigen or in the case of immunogenic therapeutic protein product (i.e., FVIII for hemophilia A or recombinant acid-alpha glucosidase (rGAA) for Pompe’s disease) as the target. Such a Fc-fusion B cell depletion agent would target B cell receptor (BCR)-expressing memory B cells and would provide a more targeted B cell depletion therapy strategy than anti-CD20 targeting monoclonal antibodies (including rituximab, obinutuzumab, ocrelizumab, ofatumumab, and ublituximab), which deplete the entire CD20⁺ B cell population. While these mAbs have successfully been used for the treatment of autoimmune diseases (53) and B-cell malignancies (54), these global B-cell depletion therapy strategies put the patient at increased risk of infection. We previously demonstrated that rFVIIIFc promotes CD16⁺ NK cell-mediated lysis of a FVIII-specific B cell clone BO2C11 (12). Therefore, an antigen-specific B cell depletion strategy, such as a CD16⁺ NK cell activating Fc-fusion protein, could provide a favorable safety benefit.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

Ethical approval was not required for the studies on humans in accordance with the local legislation and institutional requirements because only commercially available cryopreserved peripheral blood mononuclear cells and established cell lines were used.

HL: Conceptualization, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. JO: Formal analysis, Investigation, Writing – original draft. ZS: Conceptualization, Formal analysis, Funding acquisition, Supervision, Writing – review & editing. BG: Conceptualization, Formal analysis, Funding acquisition, Supervision, Writing – review & editing.