FormalPara Key Summary Points

Why carry out this study?

This study aimed to verify the potential applicability of KNP-301 to patients with neovascular and nonexudative age-related macular degeneration (AMD).

Complement inhibitors have been developed for patients with geographic atrophy. However, neovascular exudation frequently occurs as a side effect in patients treated with complement inhibitors. In such cases, additional injection of anti-vascular endothelial growth factor (VEGF) therapy is required.

In order to reduce the patient burden on the number of intravitreal injections, we developed KNP-301, a bispecific fragment crystallizable region (Fc)-fusion molecule to target complement pathway and angiogenesis simultaneously.

What was learned from this study?

KNP-301 exhibits therapeutic potential for both neovascular and nonexudative AMD. Featuring a dual-blockade targeting both C3b and VEGF, KNP-301 expands its applicability to a wider range of patients with late-stage AMD.

Introduction

Age-related macular degeneration (AMD) is a chronic and progressive eye condition significantly contributing to irreversible vision loss in the elderly. It is commonly differentiated into three stages: early, intermediate, and late [1, 2]. In the late stage, it is categorized into nonexudative and neovascular types. Nonexudative AMD advances slower than neovascular AMD and involves imperfections in the retinal pigment epithelium (RPE) due to drusen formation, a buildup of protein and lipids near the macula, leading to geographic atrophy (GA) [3, 4]. Initially, drusen formation does not impact vision, however, accumulation of larger drusen elevates the risk of AMD as the condition progresses.

Overexpression of complement component C3 in RPE cells and Bruch’s membrane plays a crucial role in the development of nonexudative AMD [2, 4,5,6]. The complement system, comprising classical, mannose-binding lectin, and alternative pathways, centers around C3 activation [7,8,9]. C3 convertases cleave C3 into C3a and C3b then, C3b binds to Factor B (FB) and properdin, which initiates the signal propagation through an amplification loop [10]. Specifically, C3b is part of the C5 convertase complex, generating a membrane attack complex (MAC) that causes cell lysis. The activation of the alternative pathway in the complement system contributes to immune responses and inflammation causing many human diseases. This is the main reason we are interested in blocking the alternative pathway in this study.

Patients with GA have been treated with complement inhibitors such as Syfovre® and Izervay™. Syfovre® (pegcetacoplan, Apellis Pharmaceuticals, Waltham, MA, USA) is a C3 blocker inhibiting classical and alternative pathways, and it was approved by the U.S. Food and Drug Administration (FDA) for GA in February 2023. Also, Izervay™ (avacincaptad pegol, Iveric Bio, Inc., Parsippany-Troy Hills, NJ, USA) is a PEGylated aptamer impeding C5 cleavage, which was also approved by the FDA for GA in August 2023 [4, 11,12,13].

In contrast to the slow progression of nonexudative AMD, neovascular AMD leads to rapid central vision loss [2]. In neovascular AMD, macular degeneration triggers the release of vascular endothelial growth factor (VEGF), leading to the formation of abnormal blood vessels that are prone to rupture, bleeding, and fluid leakage. This process damages the macula, resulting in scars and severe central vision loss. Standard treatments for choroidal neovascularization (CNV) include anti-VEGF therapies such as Avastin® (bevacizumab, Genentech, South San Francisco, CA, USA), Lucentis® (ranibizumab, Genentech), and Eylea® (aflibercept, Regeneron Pharmaceuticals, Inc.) [14, 15].

However, clinical needs persist for both nonexudative and neovascular AMD treatments. It has been disclosed that patients with GA who received Syfovre® treatment often developed neovascular AMD, requiring separate anti-VEGF therapy, and treating concurrent CNV and GA with anti-VEGF therapy demonstrated limited efficacy [16]. To address these challenges and unmet needs, we developed a novel bi-specific fusion protein with the dual blockade of C3b and VEGF (called KNP-301) for treating patients with late-stage AMD. We designed this fusion protein consisting of a C3b blocker and VEGF trap to address CNV and GA in one molecule (Fig. 1a). The C3b blocker comprised the extracellular domain of the CRIg protein encoded by the Vsig4 gene. The CRIg protein exists in two forms: the L form, consisting of both Ig-V and Ig-C2 domains, and the S form, containing only the Ig-V domain in the extracellular region [17]. The structure of the Ig-V (variable immunoglobulin domain) and Ig-C2 (constant immunoglobulin domain) within the extracellular domain of CRIg(L) was utilized for the binding to C3b, thereby selectively inhibiting the disease-associated alternative pathway. The VEGF trap is comprised of domain 2 of vascular endothelial growth factor receptor 1 (VEGFR-1) and domain 3 of vascular endothelial growth factor receptor 2 (VEGFR-2), which is fused to the C-terminal of fragment crystallizable region (Fc) via a G4S linker. The procedure for KNP-301 production is described in the Supplementary Materials. We expect our designed fusion protein would resolve unmet needs in patients with CNV and GA.

Fig. 1
figure 1

Structure and binding affinity of KNP-301. a KNP-301 and d mKNP-301 structure. C3b blocker consisted of Ig-V and Ig-C2 domain of CRIg protein. VEGF blocker was originated from aflibercept, which consisted of VEGFR1-D2 and VEGFR2-D3. Those two parts were connected by the Fc of human IgG. Binding affinity against C3b and VEGF proteins were assessed using SPR method; b and c KNP-301 and e and f mKNP-301. VEGF vascular endothelial growth factor, VEGFR vascular endothelial growth factor receptor, Fc fragment crystallizable region, D2 domain 2, D3 domain 3

Full size image

Methods

Surface Plasmon Resonance (SPR)

To measure the binding force of the fusion protein, the physical properties of the prepared fusion protein were analyzed using Biacore 8 K (GE Healthcare, 29129951) instrument. A C3b stock solution was diluted to 50 nM using a 1 × HBS-EP + buffer. Human VEGF165 was diluted to 5 nM using a 1 × HBS-EP + buffer. In the sample step, the diluted antigens were injected into the flow cells. Binding and separation steps are performed for 180 s and 400 s, respectively. After the separation time, a stabilization step for 60 s was performed. In the regeneration step, 10 mM glycine (pH 1.5) was injected into the flow cells for 30 s, followed by a stabilization step for 60 s. After subtracting the reference and 0 nM values from the sample values, binding kinetics were calculated using Biacore Insight Evaluation Software (Version 2.0.15.12933) and 1:1 binding model for curved fitting.

Enzyme-Linked Immunosorbent Assay (ELISA)

ELISA was performed to measure the affinity of KNP-301 to complement proteins and VEGF. Complement factors were purchased from Complement Technology, Inc. (human, rat, and mouse C3b) and Creative biolabs (cynomolgus C3b). The rabbit C3b was produced by Chempartner (Shanghai). Briefly, each target protein was coated to Nunc™ MaxiSorp™ ELISA Plates (BioLegend) for overnight at 4 °C. The next day, the antigen-coated plate was washed three times with Quantikine ELISA wash buffer 1 (R&D Systems, Minneapolis, MN, USA) using microplate washer (BioTek). Then, the BSA-based blocking solution (R&D Systems) was added in each well for 1–3 h at room temperature (RT). After the wash, the KNP-301 was added and incubated for each 1 h at RT. To detect a signal from horseradish peroxidase (HRP)-conjugated antibody, the substrate reagent (R&D Systems) and the stop solution (R&D Systems) were used following the manufacturer’s instructions. OD450 value was measured by Varioskan LUX.

Inhibition of the Classical Complement Pathway

Sheep red blood cells were centrifuged in Tris-buffered saline (TBS) twice, 1 ml of 20% sheep erythrocytes, and hemolysins were incubated at 4 °C for 30 min. Subsequently, the sheep erythrocytes were washed with TBS twice, before undergoing an additional wash with a GVB +  + buffer using a centrifuge. Next, the concentration of the sensitized sheep erythrocytes was adjusted to 1 × 109 cells/ml using a GVB +  + buffer. To analyze total hemolytic complement (CH50) through the sensitized erythrocytes, 3% or 4.5% human or mouse factor B-depleted serum was added to a 96-well plate (50 μl/well), and the KNP-301 were then treated at various concentrations. After culturing at 4 °C for 30 min, the sensitized sheep erythrocytes were added (2.5 × 106 cells/50 μl/well). Culturing was performed at 37 °C for 30 min. Centrifugation was performed at 600 × g for 10 min. The supernatant was collected and then the OD415 value thereof was measured. To induce the lysis of erythrocytes as a positive control, the lysis buffer containing Triton X-100 was used, and erythrocytes under serum-free conditions were defined as a negative control.

Inhibition of the Alternative Complement Pathway

Rabbit erythrocytes were washed with TBS twice before undergoing an additional wash with a GVB ethylene-bis(oxyethylenenitrilo)tetraacetic acid (EGTA) buffer using a centrifuge. Next, the concentration of the rabbit erythrocytes was adjusted to 1 × 109 cells/ml using a GVB EGTA buffer. To analyze alternative pathway hemolytic complement (AH50) through the sensitized erythrocytes, 9% human or mouse C1q-depleted serum was added to a 96-well plate. Moreover, the fusion proteins were treated at various concentrations. After culturing at 4 °C for 30 min, rabbit erythrocytes were added (2 × 106 cells/50 μl/well). Culturing was performed at 37 °C for 1.5 h, and centrifugation was performed. After collecting the supernatant, the OD415 value thereof was measured. The alternative complement pathway inhibiting effects of fusion proteins were confirmed through AH50 hemolysis analysis. To induce lysis of the erythrocytes as a positive control, the lysis buffer containing Triton X-100 was used, and erythrocytes under serum-free conditions were defined as a negative control.

VEGF Bioassay

VEGF blocking effect of each construct was analyzed by VEGFR2/NFAT (nuclear factor of activated T-cells) reporter cell line (BPS Bioscience: 79,387) and Bio-Glo luciferase assay system (Promega) as noted in the manufacturer’s instructions. The cells were seeded into 96-well plates and then each protein and VEGF165 (Sino Biological: 11,066-HNAH) were added at each concentration. The plates were incubated for 4 h. Then, those were removed from the incubator to equilibrate to room temperature for 10 min. Afterwards, Bio-Glo reagent was added to each well. The plate was incubated with shaking for 10 min at room temperature and the luminescence was measured by Varioskan LUX.

Mouse Laser-Induced CNV Model

This study was conducted at Experimentica Ltd. Inbred male C57BL/6JRj mice aged 8 weeks at induction (Janvier Labs, France) were housed individually in ventilated cages. All animals were treated in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research; the EC Directive 2010/63/EU of the European Parliament and of the Council on the Protection of animals used for scientific purposes and using protocols approved and monitored by the Animal Experiment Board of Finland (Experimentica Ltd. animal license number ESAVI-9520–2020). In the CNV model, neovascularization was induced by perforating Bruch’s membrane using a diode laser. The success of the model induction was confirmed with spectral-domain optical coherence tomography (SD-OCT) and fluorescein angiography (FA). Mouse surrogate of KNP-301 and Aflibercept® were administered unilaterally by intravitreal injection immediately after lasering. In vivo imaging, including SD-OCT and FA, was performed after CNV induction and at day 7 post-lasering.

Mouse Nonexudative AMD Model

This study was conducted at Naason Science Inc. Male C57BL/6n (18 ~ 20 g) mice at 7 weeks of age were obtained from Orient Bio. All animal experiments were conducted according to Korea Food & Drug Administration (KFDA) guidelines for the care and use of laboratory animals and approved by the Institutional Animal Care and Use Committee (IACUC). Nonexudative AMD was induced in 8-week-old C57BL/6 mice by a tail vein injection of NaIO3 (20 mg/kg) in all groups. Mouse surrogate of KNP-301 was injected (intravitreal, IVT) once a week for 2 weeks (days 0 and 7). Two weeks after model generation (day 14), the ONL (outer nuclear layer) was isolated and analyzed.

Pharmacokinetics (PK) Study in New Zealand White Rabbits

This study was conducted at WuXi AppTec Co., Ltd., and in accordance with the WuXi IACUC standard animal procedures along with the IACUC guidelines that was in compliance with the Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals. Male New Zealand white rabbits weighing 25–30 kg were supplied by Pizhou Dongfang Breading Co., Ltd. Eighteen rabbits were given a single IVT injection of 2.5 mg KNP-301. The rabbits were killed and their eyes were enucleated at the following time points: 1, 24, 72, 168, 336, and 672 h after dosing (n = 3 eyes for each time point). KNP-301 concentration in plasma and ocular tissues were measured by ELISA method. The concentration-versus-time data were analyzed by non-compartmental approaches using the WinNonlin software program (version 6.3, Pharsight, Mountain View, CA, USA).

Viscosity Test

Viscosity was measured using a m-VROC® viscometer (RheoSense) employing a viscometer-rheometer-on-a-chip. A B05 chip capable of measuring viscosity in the range of 0.2–2000 cP was used; 250 μl of the sample was carefully loaded into a 0.25-ml syringe and placed in a viscometer and room temperature was maintained using a ThermoCube (Solid State Cooling Systems, Wappingers Falls, NY, USA). The shear rate applied was 1500–8000 s−1.

Dynamic Light Scattering (DLS) Analysis

Samples were centrifuged at 12,000 × g for 10 min. The supernatant was transferred in the 96-well plate and analyzed using DLS instrument of Zetasizer APS (Marlvern). The assay temperature was 25 °C and the equilibration time was 120 s. The measurement duration was chosen as automatic. The size–intensity and size–volume profiles were recorded. Samples were analyzed three times independently. The average hydrodynamic radius was calculated based on triplicate analysis.

Statistics

Statistical analysis was performed using SPSS™ and GraphPad PRISM™ statistical software. A p value ≤ 0.05 was used as a threshold for statistical significance.

Results

Biochemical Binding Assessment of KNP-301

KNP-301 is a fusion protein consisting of C3b blocker (CRIg-Fc) and VEGF trap (Fig. 1a). The binding affinities (KD) of KNP-301 against human C3b and VEGF were 4.34 nM and 62.8 pM, respectively (Fig. 1b and c). We then addressed the species cross-reactivity of KNP-301 as it was reported that the VEGF trap has cross-reactivity with various species [18]. In this study, the VEGF trap exhibited cross-reactivity with VEGF from mouse, rat, rabbit, cynomolgus monkey, and human as well. The C3b blocker of KNP-301 showed binding to C3b from human and cynomolgus monkey, whereas it showed limited binding to C3b from mouse, rat, and rabbit (Table 1). Therefore, it was necessary to generate a mouse surrogate to evaluate the efficacy in mouse models. This was designed by changing only the C3b binding domain sequence from KNP-301 (named as mKNP-301) (Fig. 1d). The mouse CRIg has a single Ig-V domain unlike humans that have two domains, Ig-V and Ig-C [19]. The binding affinity of mKNP-301 (mouse surrogate of KNP-301) was measured using SPR analysis, which showed binding affinities of 7.97 nM for mouse C3b and 0.904 pM for mouse VEGF. (Fig. 1e and f). Both KNP-301 and mKNP-301 exhibited tight binding to C3b and VEGF proteins.

Table 1 Species cross reactivity of KNP-301
Full size table

The complement pathway is composed of more than 30 proteins [2]. We confirmed the selectivity of KNP-301's binding ability to the components of the complement pathway, with a specific focus on the alternative pathway (Fig. 2a). The assessment of selective binding between C3 and C3b was particularly crucial due to the process where C3 splits into C3a and C3b. The similar binding abilities between C3 and C3b could hinder the overall signal transduction within the complement pathway. Upon assessing the binding abilities of C3 and C3b by ELISA, KNP-301 exhibited approximately a tenfold higher binding affinity against C3b compared to C3 (Fig. 2b). Additionally, KNP-301 showed over 150-fold lower binding affinities to other complement factors (FB, Factor H (FH), and C5) in the alternative pathway than to C3b. Besides, a difference of over 150-fold in binding affinity was observed for the complement proteins C2 and C4, positioned in the classical and lectin pathways (Fig. 2c). This signified that KNP-301 had a selective binding capability to C3b.

Fig. 2
figure 2

C3b selective binding affinity of KNP-301. a Complement factors involved in alternative pathway were presented. Relative binding affinity of KNP-301 against b C3 and c complement factors. The binding affinity was assessed by the ELISA method. Relative binding affinity was calculated using the IC50 value as follows; IC50 to C3b/IC50 to other complement factors. FB factor B, FH factor H

Full size image

Effects of KNP-301 on Complement Pathway-Mediated Hemolysis

Generally, the complement system functions primarily by disintegrating target cells through the formation of MAC [20]. To validate the C3b blocking effect of KNP-301, hemolysis assays using erythrocytes were utilized to assess the activity of the complement system. Antibodies to red blood cells induce antibody opsonization and hemolysis at the red blood cells in the presence of serum through complement-dependent pathways [21]. In order to monitor alternative pathways in the complement system, we utilized C1q-depleted serum in the assay because C1q binding to the Fc region of antibody in opsonization mediates the classical and lectin pathways. Treatment with KNP-301 or CRIg-Fc fusion protein hindered the alternative pathway with 1.7 μg/ml of AH50 (Fig. 3a, b). However, these treatments were unable to block the classical pathway up to 250 μg/ml of concentration (Fig. 3d, e). A similar trend was observed with mKNP-301 as well (Fig. 3c, f). Taken together, fusion proteins containing the CRIg domain selectively inhibited the alternative pathway.

Fig. 3
figure 3

Functional test of C3b and VEGF blocking activities of KNP-301. To evaluate the C3b blocker function, hemolysis assay was performed. Inhibitory effect of KNP-301 on a–c alternative and d–f classical pathway-mediated hemolysis in vitro. The concentration of the KNP-301 (b, e) and mKNP-301 (c, f) were 60 μg/ml and 300 μg/ml, respectively. d At high concentrations, the trend of increasing hemolytic activity appeared to be within the margin of error for the results. P values were calculated by using one-way ANOVA followed by Tukey’s post hoc analysis: **** conditions compared to Fc only, p < 0.0001. g To test the functional activity of VEGF blocker part, VEGFR2/NFAT reporter assay was conducted. HEK293 cells expressing VEGFR2/NFAT reporter system were lysed after 25 ng/ml VEGF165 and test substances treatment. mKNP-301 mouse surrogate KNP-301, Fc fragment crystallizable region, AH alternative pathway hemolytic complement, CH total hemolytic complement

Full size image

Effects of KNP-301 on VEGF-Mediated Bioassay

To assess the VEGF blocking activity of KNP-301, VEGF-mediated cellular assays were conducted. The VEGF/NFAT reporter construct was transfected into HEK293 cells, and then VEGF and test article including KNP-301 or aflibercept were treated. The luminescence signal was evaluated in the presence of VEGF. The treatments of KNP-301 and aflibercept reduced the signal in a dose-dependent manner, showing similar levels of inhibiting potency with single digit nM of IC50 (half maximal inhibitory concentration) (Fig. 3g).

Efficacy of mKNP-301 on Mouse Neovascular AMD Model

The potential for dual inhibition of VEGF and the alternative pathway by KNP-301 was investigated in a preclinical model involving laser-induced CNV formation, a common model to study neovascular AMD [22, 23]. All treatments were intravitreally injected into the right eye immediately after lasering, followed by in vivo imaging on day 0. The success of perforating Bruch's membrane was confirmed using FA and SD-OCT in vivo imaging on day 0, and the lesions were monitored for 7 days (Fig. 4a). The body weight of the mice was monitored for 7 days (Fig. 4b) and the changes in body weight following administration were confirmed to be negligible. As shown in Fig. 4c, the irradiated lesions showed recovery in both the aflibercept and mKNP-301 treatment groups, compared to those in the vehicle group. The decrease in CNV grade observed in the control group was a natural tendency in the laser-induced CNV model. This phenomenon occurred as the induced puncture in the eye artificially heals. Therefore, we compared the effects of the vehicle and the treatment groups at the same time point (day 7). CNV grading of the vehicle, mKNP-301 and aflibercept groups were 63.6, 38.1, and 26.7%, respectively (Fig. 4d). The CNV grade of the mKNP-301 group was not significantly different from that of the aflibercept treatment group. Moreover, on day 7, mice treated with aflibercept and mKNP-301 exhibited lower CNV volumes, which were calculated using AI-based SD-OCT scan analysis (Fig. 4e). In this model, the therapeutic effect of mKNP-301 was similar to that of aflibercept, as VEGF primarily acted in neovascular formation, while the involvement of the complement system was relatively minor.

Fig. 4
figure 4

Efficacy study in a laser-induced CNV mouse model. a Study scheme of a mouse CNV model. Bruch’s membrane was perforated by a diode laser on day 0, then intravitreal injection was conducted to administrate mKNP-301 or aflibercept. The mice were monitored for 7 days. Fluorescein angiography (FA) and spectral-domain optical coherence tomography (SD-OCT) in vivo imaging was conducted on day 0 and day 7. The number of mice were as follows: vehicle group; n = 11, aflibercept group; n = 10, and mKNP-301 group, n = 14. b Body weight change after mKNP-301 and aflibercept treatment in the neovascular AMD induced mice. c Representative FA scans from choroidal focus levels on day 0 and day 7. d Qualitative assessment of choroidal leak. The damage of brunch’s membrane was qualitatively graded as being leaky from FA and SD-OCT scans. Fisher’s exact test for follow-up day 7: **vehicle vs. aflibercept group, p = 0.005 and *vehicle vs. mKNP-301 group, p = 0.037. The difference between aflibercept and mKNP-301 groups was not significant. e Quantitative analysis of CNV volume in different treatment groups on day 7. The CNV volume was analyzed from SD-OCT scans using an AI-based method. P values were calculated by using one-way ANOVA followed by Tukey’s post hoc analysis: *** vehicle vs. aflibercept group, p = 0.0003 and ** vehicle vs. mKNP-301 group, p = 0.005. The difference between aflibercept and mKNP-301 groups was not significant. mKNP-301 mouse surrogate KNP-301, IVT intravitreal, CNV choroidal neovascularization

Full size image

Efficacy of mKNP-301 on Mouse Nonexudative AMD Model

Several in vivo models were employed for nonexudative age-related macular degeneration [24,25,26]. In this study, the sodium iodate-induced nonexudative AMD mouse model was utilized. Systemic injection of sodium iodate induced rapid RPE degeneration [24, 25]. Mice administered sodium iodate intravenously exhibited various pathological features, including RPE and photoreceptor degeneration, as well as RPE movement [26].

The sodium iodate-induced nonexudative AMD model was used to evaluate the efficacy of mKNP-301. Sodium iodate was injected on day 0, followed by the administration of mKNP-301 via the IVT route. Then, the second administration was conducted on day 7, and mice were monitored for 14 days (Fig. 5a) before it was sacrificed. The body weight of mice did not change with any treatment (Fig. 5b). The ONL was analyzed using DAPI (4′,6-diamidino-2-phenylindole) staining as thickness, area, and volume of ONL are closely associated with visual acuity in nonexudative AMD [27]. The cell number and ONL area were reduced in the nonexudative AMD model (Fig. 5c–e). However, the cell counts and ONL area were significantly recovered by mKNP-301 treatment compared to those of the vehicle group (nonexudative AMD control). Moreover, C3 levels were analyzed by immunofluorescence, which was used as a marker of repeated or chronic complement activation [28]. C3 intensity was significantly higher in the nonexudative AMD-induced mice than in normal mice and the mice that were administered mKNP-301 showed significantly lower C3 intensity than the vehicle group (AMD control) (Fig. 5f, g). Given the minor role of the VEGF pathway in this model, assessment of target engagement with VEGF was not conducted.

Fig. 5
figure 5

Efficacy study in a NaIO3-induced non-exudative aged macular degeneration model in C57BL/6 mice. a Study scheme of a mouse nonexudative AMD model. Nonexudative AMD was induced by an intravenous injection of NaIO3 (20 mg/kg). Then, intravitreal injection was conducted to administrate mKNP-301 on day 0 and day 7 (n = 15/each group). The mice were monitored for 14 days. b Body weight change after mKNP-301 treatment in the mice with nonexudative AMD. c Representative outer nuclear layer (ONL) scans by treatment (size of bar = 50 μm). d ONL cell count and e ONL area were assessed using DAPI staining. f, g C3 level (stained red) in retina. C3 was used as an indicator of disease progression in retina (size of bar = 50 μm). p values were calculated by one-way ANOVA followed by Tukey’s post hoc analysis: d **** non-AMD vs. vehicle group, p < 0.0001 and # vehicle vs. mKNP-301 group, p = 0.03; e **** non-AMD vs. vehicle group, p < 0.00001 and ###vehicle vs. mKNP-301 group, p = 0.0002; g ** non-AMD vs. vehicle group, p = 0.008, * non-AMD vs. mKNP-301 group, p = 0.028 and ####vehicle vs. mKNP-301, p < 0.0001. mKNP-301 mouse surrogate KNP-301, IVT intravitreal, AMD age-related macular degeneration, ONL outer nuclear layer

Full size image

Ocular Pharmacokinetic Profiles of KNP-301 After IVT Administration

The ocular pharmacokinetic profiles of KNP-301 are presented in Fig. 6 and Table 2. We utilized New Zealand white rabbits for our ocular pharmacokinetic study due to the similarity of retinal structure to higher animals and human [29, 30]. Although the CRIg moiety has a low tendency in binding to rabbit C3b, a high dosage of IVT injection will allow for the assessment of drug exposure as C3b is not acting as a sink due to its negligible levels in rabbits as well as in humans. The amount of KNP-301 in aqueous and vitreous humor was measured, while amounts in serum samples were too low to detect. KNP-301 was predominantly maintained in vitreous humor rather than aqueous humor. The half-life (T1/2) of KNP-301 in aqueous humor was calculated to be 114 h, while in vitreous humor, it was 226 h. A study reported the vitreous half-life of aflibercept in rabbits to be 94.1 h (3.92 days) [31]. Compared to aflibercept, KNP-301 exhibited a longer half-life, suggesting that KNP-301 would offer an advantage in extending the dosing cycle.

Fig. 6
figure 6

Pharmacokinetics of KNP-301. Mean KNP-301 concentration–time profiles for ocular tissues; vitreous humor and aqueous humor after 2.5 mg/eye intravitreal administration (n = 3 animals)

Full size image
Table 2 KNP-301 ocular tissue pharmacokinetic (PK) parameters after IVT
Full size table

Physicochemical Properties of KNP-301

Typically, the hydrodynamic radius (Rh) of monoclonal antibody and Fc are about 5.41 and 3.19 nm, respectively [32]. Studies suggest that there is a correlation of the hydrodynamic radius of molecules to the intravitreal retention time [33, 34]. Hence, expanding the protein’s size and hydrodynamic radius could lead to an increase in retention time of therapeutic molecules at the intravitreal chamber. The hydrodynamic radius of CRIg-Fc and KNP-301 were 5.99 nm and 8.63 nm, respectively (Table 3). The elongated shape of KNP-301 might be associated with its longer half-life at the intravitreal chamber compared to typical antibodies in rabbits.

Table 3 Hydrodynamic radius of KNP-301
Full size table

For patients with AMD, current therapies such as anti-VEGF and complement inhibitors are administered via the IVT route. Intravitreal injection can be uncomfortable due to its invasive nature, yet it remains the most effective treatment for patients with AMD [35]. Similarly, KNP-301 is administered via intravitreal injection, akin to current AMD treatments. However, the limited injection volume for IVT administration raised concerns regarding the necessity for high-concentration formulations. The higher protein concentration could potentially cause an exponential increase in viscosity, which may affect the accuracy of the delivered dose [36]. Therefore, we evaluated the viscosity of KNP-301 at concentrations of 30, 60, and 120 mg/ml (Fig. 7a and b) to ensure its injectability. The viscosity was measured as low as 13.0 cP at 120 mg/ml, indicating that KNP-301 exhibited suitable properties for injection into eyes with low-gauge syringes even at high concentrations.

Fig. 7
figure 7

Viscosity of KNP-301. a The viscosity on KNP-301 was measured at 33, 65, and 130 mg/ml using viscometer. b Viscosity of KNP-301 under conditions with or without Tween-20. The KNP-301 concentration was 120 mg/ml

Full size image

Discussion

In this study, KNP-301, a novel bispecific fusion protein blocking C3b and VEGF, was explored in pre-clinical neovascular AMD and GA animal models. In clinical practice, anti-VEGF therapy and anti-complement pathway therapies have been applied to patients with neovascular AMD and GA, respectively. However, clinical unmet needs are not fully addressed yet as single therapies targeting neovascular AMD or GA generate unwanted side effects of the other late-stage AMD. KNP-301 is a therapeutic drug designed to overcome these limitations of existing medications which only target single late-stage AMD. KNP-301 demonstrated potent efficacy in both neovascular and nonexudative AMD animal models, with a relatively long retention time at the intravitreal chamber, enabling further development towards clinical applications.

KNP-301 has several advantages as follows: First, in the process of designing the substance, we took into account safety associated with the full inhibition of the complement pathway. The approved drugs (C3 or C5 inhibitor) imped both the classical and alternative pathways. However, there is a safety concern of bacterial infection as reported previously showing that C3- and C5-deficient patients are at an increased risk for pyogenic infections and have an evaluated susceptibility to Neisseria meningitides, respectively [37]. Thus, KNP-301 has potential in lowering the risk of bacterial infection after subsequent injections. Second, it can block both VEGF and alternative pathways with a single substance. In the clinic, complement inhibitor treatment for patients with GA induces neovascular AMD-like neovascular exudation. According to released data from Apellis Pharmaceuticals, the main side effect of anti-C3 therapy was neovascular exudation among 12% of the patients with monthly Syfovre® treatment. Those patients should be monitored and require additional anti-VEGF treatment. KNP-301 could potentially provide the convenience of single-agent treatment in patients requiring both anti-VEGF and complement inhibitor therapies. Third, KNP-301 exhibits a large molecular size and hydrodynamic radius, which contributes to half-life extension. When applied clinically, we are designing treatment regimens with similar or longer dosing intervals than those of existing CNV treatments. Based on rabbit PK data (excluding the loading dose), we anticipate dosing intervals of approximately 3 months for both CNV and nonexudative AMD. Consequently, this could lower the patients’ burden by requiring fewer injections throughout the treatment cycle.

This study has some limitations. First, although the laser-induced CNV mouse model is widely used for preclinical studies, the laser-induced perforations differ from the human CNV phenomenon [39]. Human eyes function as closed systems, while the mouse model becomes an open system due to laser-induced puncturing. While CNV is a chronic disease in humans, we observed a natural reduction in the punctured regions over time in the mouse model. Therefore, the punctured regions of the vehicle and treatment groups have to be assessed simultaneously. Despite these differences from human CNV, the mouse model has been instrumental in comprehending CNV pathology and evaluating therapeutic approaches. Second, although there are various models for CNV and nonexudative AMD, there is no animal model for assessing both disease mechanisms. Thus, we conducted our experiments with one specific disease model at each time. Although no recognizable adverse events were recorded in pre-clinical studies, since mKNP-301 serves as a mouse surrogate of KNP-301, subsequent research in higher animals such as NHP with our KNP-301 would be necessary in CNV and nonexudative AMD models. To address this limitation, toxicity studies in NHP are currently underway. The efficacy and safety of KNP-301 will be addressed in subsequent studies. Third, further studies are needed to address the analysis of target engagement since our in vivo studies addressed the phenomenological outcomes. Regardless of these limitations, we believe KNP-301 has potential in treating patients with both neovascular and nonexudative AMD.

Conclusions

Herein, this study provides initial evidence that KNP-301 could be a potential next-generation therapy for patients with late-stage AMD involving either wet AMD or GA. KNP-301 demonstrated in vitro and in vivo efficacy in both neovascular and nonexudative mouse models, suggesting its potential as a treatment for both patients with late-stage AMD.