Characterization of the bispecific VHH antibody tarperprumig (ALXN1820) specific for properdin and designed for low-volume administration

ABSTRACT

The bispecific antibody tarperprumig (ALXN1820) was developed as a treatment option for diseases involving dysregulated complement alternative pathway (AP) activity that could be administered in small volumes, either subcutaneously or intravenously. Tarperprumig incorporates a C-terminal variable domain of a heavy chain only antibody (VHH) that binds properdin (FP) connected via a flexible linker to an N-terminal VHH that binds human serum albumin (HSA). The purified bispecific VHH antibody exhibits an experimental molecular weight average of 27.4 kDa and can be formulated at > 100 mg/mL. Tarperprumig binds tightly to FP and HSA with sub-nanomolar affinity at pH 7.4 and can associate simultaneously with FP and HSA to form a ternary complex. Tarperprumig potently and dose-dependently inhibits to completion in vitro AP-dependent complement C5b-9 formation, AP-dependent hemolysis, and the AP deposition of C3, FP and C9. X-ray crystallography revealed that the isolated FP-binding VHH recognizes the thrombospondin repeat 5 domain of FP, thereby preventing FP from binding to the AP convertase owing to severe steric hindrance. Tarperprumig cross-reacts with cynomolgus monkey FP and serum albumin. In summary, tarperprumig exhibits properties tailored for subcutaneous administration and is currently in clinical development for the treatment of complement AP-related disorders.

Introduction

The complement system, comprising the alternative pathway (AP), lectin pathway (LP) (also known as the mannose-binding lectin pathway), and classical pathway (CP), is involved in a variety of functions, including the opsonization of pathogens and host cells for clearance and the killing of target cells by lysis. The system activates complement C3, which is followed by a common terminal pathway in which C5 is cleaved to form the anaphylatoxins C5a and C5b, the initiator of membrane attack complex formation.¹

Properdin (FP) is a ~ 50 kDa plasma glycoprotein that stabilizes the AP C3 and C5 convertases 10fold against the actions of the regulators of complement activity,^2–4 and inhibits the cleavage of C3b by factor I.⁵ Monomeric FP subunits self-associate in a head-to-tail configuration to form dimers, trimers, and tetramers.^6,7 FP is implicated in a range of complement-mediated disorders, including sickle cell disease,⁸ atypical hemolytic uremic syndrome,⁹ hemolytic anemias,¹⁰ arthritis,¹¹ ischemia reperfusion injury,¹² allergic airway inflammation,¹³ abdominal aortic aneurysms,¹⁴ and neonatal hypoxic ischemic brain injury.¹⁵

Tarperprumig (ALXN1820; CAS registry number: 2630946-41-5) was developed to provide patients with the option of an APblocking agent with an extended half-life that can be formulated at high concentrations, principally for subcutaneous self-administration. Heavy chain variable region antigen binding domains from camelid heavy-chain only antibodies (VHHs) were used as building blocks because these can be combined easily,¹⁶ provide high potency and selectivity, and are much smaller than immunoglobulin G (IgG) antibodies, enabling the achievement of higher molar formulation concentrations. To achieve an extended circulatory half-life with a molecule less than the size of the renal cutoff, a human serum albumin (HSA)binding VHH was also incorporated. Such an approach has been successfully used for other bispecific VHH antibodies.^17–24 A design featuring monovalent FP binding was required to avoid the theoretical risk of forming multivalent complexes with FP oligomers. This was achieved through the incorporation of a single C-terminal FP-binding VHH. Here we report the derivation and in vitro properties of the resulting FP- and HSA-binding bispecific VHH antibody tarperprumig and reveal its mechanism of action.

Results

Bivalent design and physical properties of tarperprumig

Tarperprumig is a bispecific VHH antibody incorporating an Nterminal HSA-binding VHH, to enable recycling, linked to a C-terminal FP-blocking VHH via a flexible linker (Figure 1a). A repeating G₄E rather than the conventional G₄S was incorporated to mitigate against potential serine o-glycosylation during expression in Chinese hamster ovary (CHO) cells while maintaining hydrophilicity. Linker length was kept the same as was found to be optimal in the anti-C5 bispecific VHH antibody gefurulimab.¹⁷ The highly purified protein migrates in non-reducing, denaturing sodium dodecyl sulfate polyacrylamide gel electrophoresis at the expected size of approximately 28 kDa (Figure 1b) and exhibits an actual mass by mass spectrometry of 27,349.1 Da, which is very close to the theoretical molecular weight of 27,350.2 Da for the mature protein in which the N-terminal glutamine has cyclized to form pyroglutamate. Tarperprumig formulated at concentrations well above 100 mg/mL was stable, yielding a single peak corresponding to protein monomer in nonreducing capillary electrophoresis sodium dodecyl sulfate electropherogram profiles (Figure 1c), with no evidence of scrambling of the two intradomain disulfide bonds.

Figure 1.

Sequence and physical properties of tarperprumig. (a) Amino acid sequence of tarperprumig. The final purified protein is recovered with an N-terminal pyroglutamate from cyclization of the N-terminal glutamine. The three CDRs within each VHH domain are boxed. The HSAbinding VHH domain is followed by the (Gly4-Glu)2-Gly4 linker which is highlighted in black and is in white font. This is followed by the fp-binding VHH domain. (b) Tarperprumig is smaller than an IgG, migrating at the expected size of around 28 kDa gauged against size markers in a non-reducing, denaturing 4–12% NuPAGE BisTris gel (10 µg applied). (c) Non-reducing CE-SDS electropherogram profile of tarperprumig formulated at >100 mg/mL.

CDR, complementarity determining region; CE-SDS, capillary electrophoresis sodium dodecyl sulfate; FP, properdin; HSA, human serum albumin; IgG, immunoglobulin G; LM, low molecular weight marker.

Tarperprumig binds specifically and with high affinity to its cognate ligands

The rate and equilibrium binding constants determined by surface plasmon resonance for the association of purified tarperprumig with human and cynomolgus monkey FP, HSA, and cynomolgus serum albumin (CSA) are shown in Table 1. Tarperprumig binds tightly to human FP (hFP) with a dissociation constant (K_D) at pH 7.4 of 323 pM, although binding was approximately 20-fold weaker at pH 6.0. Tarperprumig binding to cynomolgus monkey FP (cFP) was roughly 9- and 50-fold weaker than to hFP at pH 7.4 and pH 6.0, respectively. Tarperprumig binds to HSA at pH 7.4 with a K_D value of 439 pM, which was relatively unchanged at pH 6.0 or when in the presence of a saturating (25 nM) mobile-phase concentration of hFP at pH 7.4 in which a K_D of 223 pM was obtained (Table 1). Binding to CSA was between four and five times weaker than binding to HSA at both pH 7.4 and pH 6.0.

Table 1.

Binding constants for tarperprumig with antigens.

Immobilized	Mobile phase	pH	ka(1/Ms)	kd(1/s)	KD(M)	χ2
hFP	Tarperprumig	7.4	1.69 × 106	5.47 × 10−4	3.23 × 10−10	0.08
		6.0	9.30 × 105	5.98 × 10−3	6.44 × 10−9	0.57
cFP	Tarperprumig	7.4	1.56 × 106	4.54 × 10−3	2.92 × 10−9	0.22
		6.0	NA	NA	3.31 × 10−7	0.27
HSA	Tarperprumig	7.4	5.53 × 106	2.43 × 10−3	4.39 × 10−10	0.86
		6.0	5.71 × 106	1.82 × 10−3	3.19 × 10−10	0.63
CSA	Tarperprumig	7.4	5.14 × 106	1.08 × 10−2	2.11 × 10−9	0.47
		6.0	5.42 × 106	7.29 × 10−3	1.35 × 10−9	0.27
HSA	Tarperprumig +25 nM hFP	7.4	2.51 × 106	3.37 × 10−4	2.23 × 10−10	1.20

cFP, cynomolgus monkey properdin; CSA, cynomolgus serum albumin; hFP, human properdin; HSA, human serum albumin.

NA, not applicable as the K_D was determined using the steady state method.

Tarperprumig can bind concurrently to its cognate ligands

Various combinations of tarperprumig, FP, and HSA were incubated together and then applied to size-exclusion chromatography to ascertain changes in macromolecular size (Figure 2). Individually applied tarperprumig and HSA migrated as single peaks, although FP eluted as three expected peaks corresponding to dimers, trimers, and tetramers.⁷ The individually applied proteins and binary combinations of tarperprumig with either HSA or FP yielded peaks that eluted later and were therefore smaller than the peaks corresponding to a mixture of tarperprumig, FP, and HSA (black chromatogram in Figure 2a). As expected, HSA and FP did not directly interact with each other (Figure 2b). Thus, a mixture of tarperprumig, FP, and HSA resulted in a triad of complexes that were larger than those observed with the corresponding individual proteins or their binary combinations, consistent with formation of a ternary complex in which tarperprumig simultaneously binds HSA and FP.

Figure 2.

Tarperprumig can bind hFP and HSA simultaneously. (a) A mixture of tarperprumig, FP and HSA (black trace) formed a larger earlier eluting peak from SEC than individual protein components or all binary protein combinations. (b) HSA and FP do not interact in the absence of tarperprumig.

FP, properdin; hFP, human properdin; HSA, human serum albumin; SEC, size-exclusion chromatography.

Tarperprumig: a potent and selective inhibitor of the AP

In a Wieslab^Ⓡ assay performed in 5.3% human serum,²⁵ tarperprumig potently and dosedependently inhibited the formation of C5b-9 by the AP (Figure 3a) with a half-maximal inhibitory concentration (IC50) of 6.5 nM. Tarperprumig had no effect on the terminal complement activities of the CP and LP in 1% human serum (Figure 3b,c), while the anti-C5 monoclonal antibody N19–8,²⁶ included as a positive control, was active in blocking C5b-9 formation by all three pathways. Tarperprumig also inhibited the AP hemolytic activity of cynomolgus and rhesus monkeys, but had no effect on the AP activities of guinea pig, beagle, mouse, rat, rabbit, and minipig (Figure S1). Correspondingly, tarperprumig potently and dose-dependently inhibited APdependent hemolysis of rabbit red blood cells (RBCs) in 20% (v/v) human serum (Figure 3d) with an IC50 of 29 nM. In 40%, 60% and 80% (v/v) human serum, (Figure 3e) tarperprumig was found to inhibit AP-dependent hemolysis with IC50 values of 24.9 nM, 42.9 nM, and 62.7 nM, respectively. Tarperprumig inhibited the deposition of C3, FP, and C9 onto a myeloperoxidase substrate by 20% human serum (Figure 3f-h) with comparable IC50 values ranging between 15 and 20 nM.

Figure 3.

Tarperprumig potently and selectively inhibits AP activation of both C3 and C5. Wieslab® assay of the inhibition by tarperprumig (green traces) of membrane attack complex formed by AP (a), CP (b) and LP (c) pathways of human serum, in comparison with the IgG based C5-blocking mAb N19-8 (red traces). Inhibition by tarperprumig of APdependent hemolysis in 20% human serum (d), and in 40% (red trace), 60% (green trace) and 80% (black trace) human serum (e). Inhibition of deposition of C3 (f), FP (g), and C9 (h). Red lines in panels (d), (f), (g) and (h) are fits. IC50 values are located on the figure.

AP, alternative pathway, CP, classical pathway; FP, properdin; IC50, half-maximal inhibitory concentration; IgG, immunoglobulin G; LP, lectin pathway; mAb, monoclonal antibody; n/a, not applicable as no inhibition was observed.

Tarperprumig recognizes FP thrombospondin repeat 5

To clarify the mechanism of inhibition and to establish the epitope in FP and paratope of tarperprumig in detail, we determined the structure of the complex formed between the twochain monomeric FP deletion mutant FPΔ23 and a C-terminal His6-tagged version of the FP-binding VHH domain of tarperprumig, referred to as TPP-3077, comprising residues 138–256 of full-length tarperprumig. The deletion mutant FPΔ23 missing thrombospondin repeats (TSRs) 2 and 3 was chosen because it forms the smallest and most compact complex with TPP3077. The two TSRs deleted from FP (Figure 4a) are located in the periphery of the molecule and they are not essential for interaction with C3b and the AP (pro)convertase.⁶ In FP oligomers, TSR2 and TSR3, together with TSR4, form the extended arms that connect the convertase-binding, triangular-shaped body formed by the remaining domains.²⁷ Diffraction data obtained with crystals of the FPΔ23:TPP3077 complex extended to 1.9 Å (Table 2, Figure 4b,c). The crystals contained two complexes in the asymmetric unit that are virtually identical except for crystal packing interactions involving mainly TSR4 and the C-terminal end of TSR1. We also determined the structure of the unbound TPP-3077 based on diffraction data extending to 1.3 Å (Table 2, Figure S2a). This structure has eight TPP-3077 molecules in the asymmetric unit that differ mainly by minor differences in the complementary determining regions (CDRs) likely to arise from crystal packing interactions (Figure S2b).

Figure 4.

The fp-binding VHH of tarperprumig recognizes an epitope in TSR5. (a) Outline of the FP domain structure with two FB interacting regions indicated. (b, c) examples of the 2mF_o-DF_c electron density contoured at 1.75 σ with the final atomic model. (d) Cartoon representation of the complex revealing that the VHH epitope is located exclusively in FP TSR5. (e) Open book view of the intermolecular interface. The footprint of the VHH on FP is indicated by the orange area (left) whereas the footprint of FP on TPP3077 is outlined in light blue (right). The positions of selected residues in the interface are indicated. (f, g) magnified views of the interface with atoms colored as in panels b and c. Dotted lines indicate hydrogen bonds and salt bridges connecting FP and TPP-3077. Water molecules are shown as pink spheres.

FP, properdin; TSR, thrombospondin repeat; VHH, variable region antigen binding fragment.

Table 2.

Data collection and refinement statistics generated in phenix.Table_one.

Structure	FPΔ23:TPP-3077	TPP-3077
PDB entry	8Q6R	8Q78
Wavelength	0.9763	0.9763
Resolution range	42.85–1.9 (1.968–1.9)	27.62–1.269 (1.269–1.225)
Space group	P21	P1
Unit cell	79.21 63.701 119.249 90 103.752 90	45.727 53.611 100.77 90.029 90.022 64.792
Total reflections	610707 (45286)	1649310 (109664)
Unique reflections	89619 (8051)	245404 (9517)
Multiplicity	6.8 (5.6)	6.7 (4.9)
Completeness (%)	98.07 (88.69)	83.97 (37.24)
Mean I/sigma(I)	9.05 (1.26)	16.05 (3.34)
Wilson B-factor	32.98	13.09
R-merge	0.1127 (1.137)	0.06126 (0.3776)
CC1/2	0.998 (0.751)	0.999 (0.886)
Reflections used in refinement	89497 (8034)	213953 (9516)
Reflections used for R-free	1899 (171)	1949 (83)
R-work	0.2029 (0.3460)	0.1678 (0.2449)
R-free	0.2382 (0.3666)	0.1924 (0.2537)
Number of nonhydrogen atoms	7492	8778
Macromolecules	6716	7448
Ligands	412	88
Solvent	364	3530
Protein residues	875	968
RMS(bonds)	0.023	0.015
RMS(angles)	1.72	1.82
Ramachandran favored, allowed, outliers (%)	96.87, 3.13, 0.00	98.32, 1.68, 0.00
Rotamer outliers (%)	0.68	0.38
Clashscore	2.52	1.96
Average B-factor	59.96	21.04

Note. Statistics for the highest-resolution shell are shown in parentheses.

CC, Cross correlation coefficient; PDB, protein data bank; R, residual; RMS, root mean square.

According to our analysis of the X-ray diffraction data using the Proteins, Interfaces, Structures and Assemblies software,²⁸ the VHH – FP interface is relatively polar with a buried surface area of 1310 Å,² which is smaller than observed in the majority of VHH – antigen complexes.²⁹ The epitope is located almost entirely within FP TSR5 residues Asp356-Gly362 (Figure 4d) whereas the paratope is formed mainly by TPP-3077 CDR2 and CDR3, but with a minor contribution from CDR1 and framework residues located in the β-strand preceding CDR2 (Figure 4e-g and Figure S2c). Central in the interface is FP Arg359; its guanidinium group is recognized by a double salt bridge with VHH Glu187 in CDR2 and hydrogen bonds formed with the side chain of Asn233 in CDR3 (Figure 4g). These polar interactions are enforced by the nearby side chains of Leu235 in CDR3 and the framework residue Leu184. Another important residue in the interface is FP His358, which forms hydrogen bonds with the side chains of Glu187 and Thr193 in the VHH CDR2 (Figure 4f). From CDR1, the side chain of His170 engages in a hydrogen bond with the FP Asp356 main chain carbonyl (Figure 4f). In CDR3, the main chain of Glu238 interacts through a hydrogen bond with the side chain of FP Asp356 (Figure 4g). Besides direct VHH – FP interactions, water molecules also bridge the antigen and the antibody. Hydrogen bonds formed by two water molecules bridge CDR2 main chain atoms with the mannosyl group attached to FP Trp318 and the main chain of Ala361 (Figure 4f). Four water molecules form a network that stabilizes the direct interaction of CDR3 with FP Asp356 (Figure 4g).

The fp-binding VHH of tarperprumig, TPP-3077, binds to FP TSR5, the region of FP that interacts with C3b

The structure of the FPΔ23:TPP-3077 complex offers an explanation for the experimental data concerning the properties of tarperprumig. First, the structure rationalizes the observed pattern of species cross-reactivity of tarperprumig against FP from various mammals (Figure 5a). The epitope defined in the structure is almost conserved between human and cynomolgus FP explaining the cross-reactivity of tarperprumig (Table 1, Figure S1). In contrast, FP from mice, guinea pig, rat, rabbit, and minipig are poor antigens for tarperprumig, which can also be understood with the structure of the complex. Only human and cynomolgus FP have a histidine at position 358 and, in addition, rat and mouse FP have a proline at position 359 instead of the arginine present in hFP. Both side chains are extensively recognized by the TPP-3077 through polar interactions (Figure 4f,g).

Figure 5.

Structural basis for tarperprumig inhibition of the AP. (a) The sequence of the epitope explains the selective specificity of TPP-3077 for hFP and cFP. The triangles indicate the epitope on FP that is bound by TPP-3077. (b) The crystal structure of the fp-stabilized proconvertase from PDB entry 7NOZ used for the comparison. (c) The superposition of FPΔ23 from the TPP-3077 complex (domains colored as in Figure 4a) with FP from the proconvertase complex in panel b (colored gray) illustrates that interaction with TPP-3077 does not induce significant conformational changes in the convertase-binding TSR5 and the TSR6 index finger loop. A small difference at the N-terminal end of TSR4 is due to crystal packing interactions and flexibility in this end of TSR4 is a common feature.³⁰ (d) Model illustrating the major steric clash that would occur if the proconvertase attempted to interact with TPP-3077 bound FP. In panel d, carbon atoms in TPP-3077 are shown in orange, whereas carbon atoms in FP are colored as in panel b.

AP, alternative pathway; cFP, cynomolgus monkey properdin; FP, properdin; hFP, human properdin; TSR, thrombospondin repeat; MACFA, Macaca fascicularis.

In order to understand the mechanism of inhibition exerted by tarperprumig, we compared structures of the FPΔ23:TPP-3077 complex and the FP-stabilized proconvertase C3bBP.³¹As demonstrated in Figure 5b,d, both TPP-3077 and the C345c domain of C3b interact almost exclusively with TSR5 in FP. Because binding of the VHH does not induce significant conformational changes in FP TSR5 (Figure 5c), an allosteric mechanism can be excluded. Instead, the structural comparison reveals that inhibition is most likely due to direct competition for the epitope (Figure 5d). Within the FP-stabilized proconvertase, the TPP-3077 epitope Asp356Gly362 engages in van der Waals interactions with C3b residues 1634–1638 organized in a small α-helix. Equally important, should TPP-3077-bound FP associate with the proconvertase C3bB, a major steric clash would occur between the VHH and the C345c domain of C3b (Figure 5d) involving approximately one in six of the atoms in TPP-3077.

Discussion

Here, we describe the characterization of tarperprumig, a new therapeutic candidate with potential to treat the many alternative pathway driven diseases. Tarperprumig is a bispecific VHH antibody comprising a C-terminal VHH that binds and blocks FP, connected via a flexible linker to an N-terminal anti-HSA VHH designed to confer an extended half-life. The relatively small size of tarperprumig in comparison with conventional IgG antibodies and achievement of formulation concentrations >100 mg/mL allow for the administration of higher molar concentrations of tarperprumig than can be achieved with IgG-based antibodies formulated at that concentration, potentially enabling the delivery of therapeutic amounts of tarperprumig in volumes small enough to be administered subcutaneously. Tarperprumig binds to FP with subnanomolar affinity at blood pH and potently and dose-dependently blocks AP-dependent C5b-9 formation, AP hemolysis and AP-dependent C3, FP, and C9 deposition. Inhibition was, in each case, to completion in contrast to the expectations of neutralizing a purely stabilizing target. The completeness of inhibition is possibly explained as the summation effects, including blocking FP from serving as an initiator of complement activation on permissive surfaces and from stabilizing the AP C3 and C5 convertases. As expected, tarperprumig did not block CP- or LP-dependent C5b-9 formation because FP is not involved in either of these pathways. This potent, complete and selective blockade of AP activity portends favorably for corresponding in vivo activity.

The molecular explanation for this inhibitory effect of tarperprumig on FP function was evident from our X-ray studies, which revealed how the isolated FP-binding VHH domain of tarperprumig interacts with the TSR5 region of FP, thus preventing sterically the association of FP with C3b present in the AP proconvertase C3bB. Because FP interacts with C3b in the AP convertase in an almost identical manner (i.e., TSR5),^6,32 the mechanism of inhibition outlined for the proconvertase also applies to the convertase. Integrative structural studies of FP oligomers revealed that these are relatively rigid, planar molecules with their C3b and TPP-3077 binding sites separated by an average of 200 Å.²⁷ Hence, as our structure of the FP complex with TPP3077 does not reveal conformational changes in the FP monomer that could propagate to other subunits in a FP oligomer, we do not foresee allosteric effects induced by tarperprumig binding to the TSR5 domain in one subunit of an FP oligomer that could hinder the binding of the inhibitor to a second subunit in the oligomer. In contrast, even partial saturation of FP trimers or tetramers with tarperprumig could reduce the activity of FP considerably because the biological activity of FP dimers and monomers is significantly lower than for the trimers and tetramers.^7,27,32

Our structure of the interface between FP and the FP-binding VHH of tarperprumig together with the sequence diversity within the TSR5 epitope across species, provides insights into why tarperprumig cross-reacts with non-human primate FP and thereby interferes with AP activity in the serum of these species, but not with the AP activity of serum from other species. Further studies of engineered FP variants will test the contribution to binding of residues within the epitope, including the role of H358 (as present in FP from primates) in supporting tarperprumig binding and P359 (as present in mouse and rat FP) in preventing the interaction with tarperprumig. Notwithstanding the underlying molecular explanation, the observed primate cross reactivity of tarperprumig provides a tractable path to assess preclinically the safety of pharmacologic blockade of FP.

The interaction between tarperprumig and HSA is an important prerequisite for an extended duration of action in vivo. First, tarperprumig incorporates essentially the same albumin binding VHH that is present in the anti-C5/anti-HSA bispecific VHH antibody gefurulimab (ALXN1720, CAS registry number: 2456407-94-4), with the exception that the N-terminal amino acid residue of the anti-HSA VHH of tarperprumig is glutamine while that of gefurulimab is glutamic acid. Gefurulimab is currently being evaluated in a Phase 3 clinical study as a subcutaneously administered treatment for adults with generalized myasthenia gravis (NCT05556096).¹⁹ Binding of this VHH to HSA, either as an isolated domain or when incorporated into the bispecific VHH antibody gefurulimab, did not interfere with the binding of HSA to the neonatal Fc receptor (FcRn)^17,18 that is required for the recycling of HSA³³ and provided half-life extension when incorporated into bispecific VHH antibodies together with anti-C5 VHHs.¹⁷ Second, tarperprumig binds to HSA with similar high affinity at both pH 7.4 and pH 6.0, ensuring that tarperprumig remains tightly bound to HSA as the complex encounters the lower pH conditions while traversing from the blood through the endocytic recycling pathway. Third, binding to HSA is unaffected by the presence of saturating concentrations of FP and tarperprumig could simultaneously bind both HSA and FP based on the observation of a stable ternary complex comprising tarperprumig, HSA and hFP by size-exclusion chromatography. This suggests that tarperprumig could maintain pharmacologic suppression of circulating FP in vivo while circulating bound to the vast reservoir of circulating HSA.

In therapeutic development, it is important to anticipate and mitigate the downstream consequences of target engagement. FP is oligomeric and circulates at significant levels, with 4–25 µg/mL (75–472 nM) reported for serum,¹⁰ which increases the likelihood of polyvalent complex (pIC) formation if bivalent FP-specific IgG-based monoclonal antibodies are used pharmacologically. Such pICs may cause inflammation and, upon cellular uptake, increase trafficking to sites of antigen processing, which may lead to immunogenicity. Indeed, pICs were observed elsewhere upon addition of bivalent anti-TNF IgG to the trimeric target TNFα³⁴ but mitigated with monovalent antibodies.³⁵ Similarly, tarperprumig was designed to avoid the potential for formation of pICs in complexation with FP by incorporating monovalent FP binding.

The effects of inhibition of FP by treatment with tarperprumig can be anticipated from the characteristics associated with the FP deficiencies.³⁶ These deficiencies are associated with an increased risk of infection with Neisseria spp., including N. meningitidis, although recurrence of infection is not common,³⁷ pointing to the development of protective adaptive immunity.³⁸ Accordingly, vaccination of individuals with FP deficiency is currently recommended.³⁹ Similarly, the primary infectious risk associated with tarperprumig is expected to be that of N. meningitidis and because tarperprumig leaves the CP and LP intact, vaccination against Neisseria spp. is expected to provide effective protection.

Given the findings, tarperprumig represents a very promising new therapeutic candidate that could be contemplated for a wide range of AP-driven diseases. The molecule possesses the desired preclinical properties that supported the rationale for the recently completed Phase 1 study in healthy volunteers (NCT04631562).

Materials and methods

Generation of tarperprumig

The generation of bispecific anti-FP/anti-HSA bispecific VHH antibodies including tarperprumig is described in detail elsewhere.⁴⁰ Briefly, llamas were immunized with hFP (CompTech, catalog number: A139) at Abcore Inc. VHH libraries were constructed by Qoolabs using total RNA isolated from peripheral blood mononuclear cells and converted to copy DNA by reverse transcriptase polymerase chain reaction using primers specific for llama IgG heavy chain. VHH-encoding DNA fragments were isolated by gel electrophoresis and cloned into the pADL-10b phage display vector (Antibody Design Labs). Phage-VHH clones that were enriched from the VHH library for binding to hFP were isolated and tested for binding by enzyme-linked immunosorbent assay. The VHH sequences of FP binders were expressed as C-terminal His6-tagged proteins in HEK293 cells and binding to hFP was confirmed using biolayer interferometry Octet instrument analysis (Satorius) of harvested supernatants.

An FP-binding VHH termed LVP-058 blocked AP-mediated lysis of rabbit erythrocytes by human serum and bound recombinant cFP according to biolayer interferometry analysis. LVP058 was humanized by grafting CDRs 1, 2, and 3 into framework regions 1, 2, and 3 of the human IgHV3-23 × 01 germline gene segment.⁴¹ Within framework region 2, four residues were reverted to llama residues. These include Phe37 and Glu44 of the llama germline VHH subfamily 2, corresponding to residues Phe174 and Glu181 of mature tarperprumig. These substitutions were incorporated to increase overall stability and aqueous solubility while maintaining antigen binding in accordance with Vincke et al.⁴² Arg45 and Leu47, present in the parental llama VHH but not in the human germline corresponding to Arg182 and Leu184 of mature tarperprumig, were incorporated to increase antigen binding. Within framework region 3, Lys86 and Pro87 present in the parental llama VHH and corresponding to Lys223 and Pro224 of mature tarperprumig, were retained to increase antigen binding.

The humanized HSA-binding VHH of tarperprumig is essentially the same as that incorporated into the bispecific anti-C5 antibody gefurulimab (ALXN1720) as previously described, with the exception that the N-terminal residue of tarperprumig is glutamine and the N-terminal residue of gefurulimab is glutamate.¹⁷ The HSA-binding VHH of tarperprumig was fused via a (Gly4-Glu)2-Gly4 linker to the N-terminal end of humanized anti-FP VHH yielding tarperprumig. Tarperprumig used in these studies was stably expressed in CHO cells and was highly purified from conditioned medium by sequential protein A affinity chromatography, anion exchange chromatography, then cation exchange chromatography. The purified protein was buffer exchanged into phosphate-buffered saline (PBS).

Surface plasmon resonance binding studies

Binding of tarperprumig with FP at pH 7.4 and pH 6.0 was quantified by surface plasmon resonance on a BIAcore 3000 system. Purified hFP was sourced from CompTech (catalog number: A139). Recombinant cFP with C-terminal His6-tag was expressed in Expi293F and highly purified over a HisTrap HP column (GE Life Sciences). hFP and cFP were each biotinylated with Sulfo-NHS-LC-Biotin and captured on one flow cell of a CAP chip (GE Life Sciences, catalog number: 289202–33). This was followed by injection of varying concentrations of tarperprumig in running buffer (HBS-EP buffer [0.01 M HEPES, pH 7.4, 0.15 M NaCl, 3 mm ethylenediaminetetraacetic acid [EDTA], 0.005% [v/v] surfactant [P20], GE Life Sciences, catalog number: BR1001–88) over both the immobilized FP and a second reference flow cell. For the binding of tarperprumig to albumin, varying concentrations of tarperprumig in running buffer were injected over both a reference flow cell and a flow cell containing either human (Grifols, lot number: 13533-692-16) or cynomolgus monkey (Athens Research and Technology, product number 16-16 -011,202-CM,) serum albumin that had been immobilized on a CM5 Chip (GE Life Sciences, catalog number: BR-1000-12) via amine coupling (GE Life Sciences, catalog number: BR-1000-50). Binding of tarperprumig to immobilized HSA was also performed at pH 7.4 with various concentrations of tarperprumig diluted in HBS-EP containing 25 nM FP (a three-fold excess over the highest concentration of tarperprumig). FP and albumin binding data were processed with a 1:1 Langmuir model using BIA evaluation software with “double referencing”.

Assessment of the capacity of tarperprumig to simultaneously bind both hFP and HSA

Purified tarperprumig, hFP (CompTech, catalog number: A139, lot number: 23c), and HSA (Grifols, lot number: 13533-692-16) were mixed in various combinations. When tarperprumig was mixed with HSA alone, they were mixed at 10.9 nM and 10.6 nM, respectively. When tarperprumig was mixed with hFP alone, they were mixed at 12.5 nM and 12.4 nM, respectively. When tarperprumig, HSA and hFP were mixed together simultaneously, they were mixed at 6.9 nM, 6.7 nM and 6.9 nM, respectively. The mixtures were incubated for 30 minutes at room temperature and then applied to a TSKgel G3000SW size exclusion column (7.8 mm × 30.0 cm, 0.5 μM) (Tosoh Bioscience) with a TSKgel SWXL Type Guard Column (6.0 mm × 4.0 cm, 0.7 μM) on a Waters Alliance e2695 high-performance liquid chromatography system. Eluting proteins were detected by A280 nM in 1 × PBS buffer (pH 7.4; Gibco Life Technologies).

AP hemolysis

Reactions were initiated at 37°C with the addition of 30 µL of rabbit RBCs (Lampire Biological Laboratories, catalog number: 7206403) in gelatin veronal buffered saline without calcium and magnesium (GVB^O) (pH 7.4; CompTech, catalog number: B103) plus 10 mm MgCl₂ (Boston BioProducts, catalog number: BM-670) and 10 mm ethylene glycol tetraacetic acid (EGTA) (Boston BioProducts, catalog number: BM-721) to achieve the following initial component concentrations in a 130 µL reaction volume: rabbit RBCs (7.6 × 10⁶ cells/mL), 20%, 40%, 60% and 80% (v/v) of normal human serum (NHS) (CompTech, catalog number: NHS) or 20% (v/v) cynomolgus monkey serum (complement preserved from BioIVT, catalog number: NHP01SRMP25), and tarperprumig (0–500 nM) in GVB^O plus 10 mm Mg-EGTA. After a 12 minute reaction time, reactions were terminated by centrifugation at 1700 × g for 3 minutes and 85 µL of each supernatant were transferred to a flat-bottom 96-well plate for measurement of optical density (OD) at 415 nm on a FlexStation 3-plate reader (Molecular Devices). Hemolysis (%) values were calculated relative to the OD at 415 nm from reactions containing zero tarperprumig with or without 20 mm final EDTA (Boston BioProducts, catalog number: BM-711) as 0% and 100% hemolysis controls, respectively.

For species cross-reactivity studies, human, cynomolgus monkey, rhesus monkey, rat, minipig, beagle, mouse, and rabbit sera were sourced from BioIVT, and guinea pig serum was sourced from CompTech. Reactions were initiated by the addition of rabbit RBCs to achieve a final serum concentration of 20% (v/v). Absorbance values at 700 nm were read every 30 s for 45 minutes.

% inhibition = $\left(\frac{\mathrm{AU}{\mathrm{C}}_{\mathrm{sample}}\mathit{\hspace{0.17em}}-\mathrm{AU}{\mathrm{C}}_{\mathrm{PC}}}{\mathrm{AU}{\mathrm{C}}_{\mathrm{NC}}-\mathrm{AU}{\mathrm{C}}_{\mathrm{PC}}}\right)\times 100$

where AUC is area under the curve, NC is serum devoid of tarperprumig but containing 50 mm final of the inhibitor EDTA, and PC is serum devoid of tarperprumig.

Assessment of the blockade of C5b-9 formation by tarperprumig

Wieslab^Ⓡ complement system assay kits (COMPL CP310 RUO [CP], COMPL AP330 RUO [AP], and COMPL MP320 RUO [LP], Svar Life Science)²⁵ were used according to the manufacturer’s instructions to assess C5b-9 formed by NHS (CompTech, catalog number: NHS). Varying tarperprumig or antiC5 antibody N19–8,²⁶ concentrations were present in 100 µL incubations, comprising pathway assay buffer and either 1% (v/v) serum for CP and LP assays or 5.3% (v/v) serum for AP assay. Incubations were at 37°C for 60 minutes in microtiter wells that had been coated with pathway-specific activators. The formed C5b-9 was quantified according to the kit instructions.

Assessment of the blockade of C3, FP and C9 deposition by tarperprumig

Reactions were initiated by the application of 100 μL compositions comprising tarperprumig (0–400 nM), 20% (v/v) NHS (complement preserved from CompTech, catalog number: NHS) in GVB^O plus 20 mm Mg-EGTA, to 96-well plates coated with the AP-specific activator, myeloperoxidase. After incubation at 37°C for 60 minutes, plates were washed followed by addition of 100 μL of antibody (2 µg/mL final) specific for either human C3c (Quidel, catalog number: A205), hFP (Quidel, catalog numbers: A233 and A235), or human C9 (Assaypro, catalog number: 15451–05011). After incubation at 37°C for 30 minutes, plates were washed followed by addition of 100 μL of alkaline phosphate-conjugated secondary antibody (Fisher Scientific, catalog numbers: G21060 or G21079) and incubated at 37°C for 30 minutes. After further washing with 200 μL p-nitrophenylphosphate, substrate was applied at room temperature for 13 minutes. Product color development was stopped by the addition of 50 μL of NaOH solution. Levels of C3 fragment, FP or C9 formed on well surfaces were detected by recording the OD at 405 nm on a plate reader (FlexStation 3, Molecular Devices). The 0% and 100% values were derived from incubations without tarperprumig either in the presence or absence of 50 mm EDTA (Boston BioProducts, catalog number: BM-711), respectively.

Structure determination of the complex formed between the anti-fp VHH of tarperprumig and hFP

The FP-binding VHH of tarperprumig designated as TPP-3077 and comprising residues 138–256 of tarperprumig with a C-terminal His6-tag was expressed in Expi293 cells. Batches (500 mL) of 4day conditioned medium were applied to a 5 mL HisTrap Excel column (GE Life Sciences), equilibrated and washed with 20 mm NaPO₄, 20 mm imidazole, 500 mm NaCl, pH 7.4. Highly pure fractions eluted by increasing the imidazole to 500 mm were dialyzed into 10 mm HEPES, 75 mm NaCl, pH 7.2. The crystal structure of TPP-3077 was solved in complex with FPΔ23 (also known as FPhtΔ2,3), a two-chain FP construct lacking TSRs 2 and 3. FPΔ23 was expressed and purified, as described in Pedersen et al. ⁴³ It was then incubated with TPP-3077 (lot number: 2018.12.03) in a 1:1.5 molar ratio, and the complex was separated from the free components using size-exclusion chromatography on a 24 mL Superdex 200 Increase column (Cytiva), equilibrated in 20 mm HEPES (pH 7.4) buffer containing 150 mm NaCl. The purified complex was concentrated to 3.3 mg/mL and drops containing 400 nL of protein and 400 nL of reservoir were set up using the Oryx crystallization robot in 96-well plates for vapor diffusion crystal trials. Well-diffracting crystals were obtained in a 95% Morpheus III screen, condition G1 (1.2% cholic acid derivatives mix, 0.1 M Buffer system 1, pH 6.5, 30% Precipitant Mix 1) and 5% Milli-Q water. Crystals were grown at 19°C for 2 weeks and frozen in loops without additional cryoprotectant. X-ray diffraction data were collected at the PETRA P14 beam line and processed with the X-ray Detector Software (XDS).⁴⁴ The structure was solved by molecular replacement using phenix.phaser,⁴⁵ utilizing coordinates of previously determined structures of FP⁶ and the structure of TPP-3077 (described below). The molecular replacement solution was manually rebuilt in COOT⁴⁶ and then refined with phenix.refine⁴⁷ in an iterative manner until convergence. The carbohydrate geometry was validated with the CCP4 program privateer.⁴⁸

TPP3077 was crystallized at 4.3 mg/mL in Morpheus I screen, condition A4 (0.06 M divalent cations, 0.1 M Buffer System 1, pH 6.5, 37.5% Precipitant Mix 4). Drops containing 200 nL of protein and 200 nL of reservoir were set up using the Oryx crystallization robot in 96-well plates for vapor diffusion at 19°C. Crystals were grown for 4 days and frozen without additional cryoprotectant. Diffraction data were collected at the PETRA P14 beam line and processed as described above. The structure was solved by molecular replacement using a library of nanobody structures extracted from the Protein Data Bank as search models.

Funding Statement

This study was funded by Alexion, AstraZeneca Rare Disease, Boston, MA, USA.

Disclosure Statement

DD was an employee of Alexion, AstraZeneca Rare Disease at the time of study, is a shareholder of Alexion, Astra Zeneca Rare Disease.

JC, RP, TH, FS and JH are employees of Alexion Pharmaceuticals, Inc, a wholly owned subsidiary of AstraZeneca PLC, and owns AstraZeneca stock.

PT was an employee of Alexion, AstraZeneca Rare Disease at the time of study, was a shareholder of Alexion, AstraZeneca Rare Disease and is the inventor on numerous patents and patent applications owned by Alexion, AstraZeneca, Rare Disease and/or Alexion Pharmaceuticals, Inc., including patents for tarperprumig.

Abbreviations

AP

alternative pathway

CDR

complementary determining regions

cFP

cynomolgus monkey properdin

CP

classical pathway

CSA

cynomolgus serum albumin

EDTA

ethylene diamine tetraacetic acid

EGTA

ethylene glycol tetraacetic acid

FcRn

neonatal Fc receptor

FP

properdin

GVB^O

gelatin veronal buffered saline without calcium and magnesium

hFP

human properdin

HSA

human serum albumin

IgG

Immunoglobulin G

LP

lectin pathway

NHS

normal human serum

OD

optical density

PBS

phosphate buffered saline

pIC

polyvalent complex

RBC

red blood cell

TSR

thrombospondin repeats

VHH

variable-region of a heavy chain from a heavy chain only antibody

Data availability statement

Alexion, AstraZeneca Rare Disease will consider requests for disclosure of clinical study participant-level data provided that participant privacy is assured through methods like data de-identification, pseudonymization, or anonymization (as required by applicable law), and if such disclosure was included in the relevant study informed consent form or similar documentation. Qualified academic investigators may request participant-level clinical data and supporting documents (statistical analysis plan and protocol) pertaining to Alexion-sponsored studies. Further details regarding data availability and instructions for requesting information are available in the Alexion Clinical Trials Disclosure and Transparency Policy at https://www.alexionclinicaltrialtransparency.com/data-requests/.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2024.2415060