Development and preclinical characterization of AMG 329: a human antibody neutralizing FLT3 ligand

ABSTRACT

The feline McDonough sarcoma-like tyrosine kinase 3 (FLT3)/FLT3 ligand (FLT3L) signaling pathway regulates the development and activity of dendritic cells (DCs) and other myeloid cells, including monocytes. FLT3L, DCs, and monocytes have been implicated in several autoimmune diseases. Here, we describe the development and characterization of a human immunoglobulin G1λ monoclonal antibody (AMG 329; formerly MEDI1116/VIB-1116/HZN-1116) targeting human FLT3L. AMG 329 was derived from a large human combined antibody display library; it was optimized to enhance affinity for FLT3L and reduce antibody dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity. Binding affinity was determined by surface plasmon resonance interaction analysis. Specificity of FLT3L was measured using cell-based flow cytometry and an in vitro functional neutralization assay. ADCC activity was measured using an in vitro cell culture system. Toxicity and toxicokinetics were evaluated in cynomolgus monkeys during AMG 329 dosing (5–100 mg/kg; ≤ 27 weeks) and recovery (≤32 weeks). The AMG 329 antigen-binding region selectively bound to human and cynomolgus monkey FLT3L with affinities of 170 and 63 pM, respectively. AMG 329 specifically bound to and neutralized soluble and cell-bound human FLT3L and did not induce ADCC. AMG 329 administration generally reduced circulating plasmacytoid, conventional DC, and classical monocyte relative proportions in cynomolgus monkeys in a non–dose-dependent manner. Disruption of the FLT3/FLT3L signaling pathway presents a new potential therapeutic approach to treat autoimmune and inflammatory diseases. AMG 329 is a selective human monoclonal antibody antagonist of FLT3L that is currently being investigated in clinical studies.

Introduction

Feline McDonough sarcoma-like tyrosine kinase 3 (FLT3) and its ligand, FLT3L, play integral roles in the proliferation, differentiation, and survival of hematopoietic cells.¹ Increased FLT3/FLT3L signaling and elevated circulating FLT3L levels have been associated with multiple autoimmune and inflammatory diseases, including rheumatoid arthritis, Sjögren’s disease, and systemic lupus erythematosus.^2–5 More recently, an FLT3 variant (rs76428106-C) that leads to the partial loss of wild-type FLT3 expression but a two-fold increase in circulating FLT3L levels has been identified; this variant is associated with an increased risk of several autoimmune diseases and hemopoietic malignancy.⁵ FLT3, a class III receptor tyrosine kinase expressed constitutively by hematopoietic progenitor cells and some dendritic cell (DC) subsets, plays a critical role in the development and survival of DCs.^1,2 FLT3L is a type I transmembrane protein that can be cleaved by tumor necrosis factor-alpha converting enzyme (TACE/ADAM17) to generate a soluble form; both the soluble and membrane-bound isoforms are biologically active.^1,2,6 FLT3L is derived from both nonhematopoietic cells and T cells,^1,7 with derivation from T cells particularly important in the inflammatory setting because it is driven by the production of gamma-chain cytokines.⁸

DCs are antigen-presenting immune cells that bridge innate and adaptive immune responses and play a central role in autoimmune disease pathogenesis.^9,10 Multiple distinct DC subtypes, including plasmacytoid DCs (pDCs) and conventional DCs (cDCs), contribute to immune responses via distinct mechanisms; pDCs can produce high levels of type I interferons, and cDCs can activate and stimulate the proliferation of CD4⁺/CD8⁺ T cells.^9–12 Aberrant DC activity has been implicated in many inflammatory and autoimmune diseases, such as psoriasis, systemic sclerosis, systemic lupus erythematosus, type 1 diabetes, and arthritis.^10,11,13 Similarly, aberrant activity of cDCs and their downstream inflammatory mediators (e.g., interleukin [IL]-23, IL15, IL-12, IL-17) has been associated with multiple inflammatory and autoimmune conditions, including rheumatoid arthritis, systemic lupus erythematosus, and Sjögren’s disease.^14–18 Monocyte infiltration and activation have also been observed in several autoimmune diseases, such as Sjögren’s disease, rheumatoid arthritis, systemic sclerosis, and inflammatory bowel disease.^19,20

FLT3 is constitutively expressed on bone marrow multipotent progenitor cells, but is maintained only on pDCs, DC1s, and a subset of DC2s; these cells are dependent on FLT3L for development and survival.^2,21,22 Mice genetically deficient in FLT3L lack pDCs, cDC1s, and cDC2s in lymphoid and nonlymphoid tissues.²¹ Similarly, DCs were depleted in cynomolgus monkeys 1 week after treatment with an anti–FLT3-CD3 immunoglobulin (Ig) G antibody.²³ FLT3 inhibition induces apoptosis in mouse and human DCs in vitro and inhibits immune activation in mice using in vitro and in vivo models.²² Although FLT3L is known for its critical role in DC development,²⁴ there is also evidence for activation-induced expression of FLT3 on monocytes, B cells, and T cells, whereas FLT3L appears to promote the proliferation and survival of these cells.^1,3,25–27 While FLT3-mediated proinflammatory survival via DCs or other cell types provides an advantageous physiological response in healthy individuals, it likely has deleterious effects in autoimmune diseases. In patients with autoimmune or inflammatory conditions, inhibition of the FLT3/FLT3L signaling pathway may decrease the proinflammatory effects of pDCs, cDCs, and monocytes, thereby reducing disease progression.

AMG 329 is a first-in-class human IgG1λ monoclonal antibody designed to selectively bind to human FLT3L and disrupt FLT3/FLT3L interactions, which may have the potential benefit of reducing pDC, cDC, and classical monocyte numbers in various autoimmune and inflammatory conditions. The efficacy and safety of AMG 329 is currently being investigated in a Phase 2 study in two patient populations with Sjögren’s disease: Population 1 includes patients with moderate to high systemic disease activity, and Population 2 includes patients with moderate-to-severe symptoms, but limited systemic organ involvement (NCT06312020).²⁸ Here, we describe the derivation and in vitro characterization of AMG 329 and pharmacodynamic effect of AMG 329 on circulating DCs in cynomolgus monkeys.

Results

Derivation, optimization, and in vitro characterization of AMG 329

The parental clone for AMG 329 was identified from a naïve phage display library of human antibody fragments.^29–33 Four naïve human antibody phage libraries were panned using a solution-phase panning strategy consisting of three iterative rounds with decreasing concentrations of human or mouse FLT3L (100–25 nM). These methodologies have been previously described.³³ Outputs enriched for human and mouse FLT3L binding were then converted to ScFv-Fc or single-chain Fabs, and high-throughput screening was performed using Human FLT3/FLT3L competition homogeneous time-resolved fluorescence (HTRF). Approximately 4,000 colonies were selected for this HP-HTRF screening to assess FLT3/FLT3L competition. Next, HTRF hits were converted to IgG (containing the Fc silencing triple mutation [TM]: L234F/L235E/P331S)³⁴ and screened for FLT3L downregulation.³⁵ After lead selection in the IgG-TM format, mouse and cynomolgus monkey cross-reactivity was assessed via enzyme-linked immunosorbent assay (ELISA). The lack of binding to other family members, such as stem cell factor (SCF), was also confirmed via ELISA.³³ Lead antibodies were further assessed using FLT3 signaling assays in RS4;11 cell lines. Binding to endogenous FLT3L was confirmed using primary human T cells. The relative epitope was assessed by Octet and the affinity by BIAcore. The resulting lead antibody clone (5D9) was selected for further optimization, as described below.

Anti-FLT3L IgG clone 5D9 was identified as a highly specific but low-affinity inhibitor of FLT3L-induced FLT3 downregulation on RS4;11 B-cell myeloma cells compared with other candidate clones identified by an HTRF FLT3-Ig binding assay. A rapid off-rate³³ made it an ideal candidate for affinity maturation (Table 1, Figure S1). Two rounds of affinity maturation resulted in a final candidate clone (AM40) with approximately 7,000-fold improved affinity for human FLT3L over 5D9 and the ability to bind both human and cynomolgus monkey FLT3L. The final clone containing the IgG-TM Fc silenced backbone was renamed AMG 329 (formerly named MED1116, VIB-1116, or HZN-1116). Biolayer interferometry (BLI) analysis confirmed direct competition with the FLT3-binding epitope (Figure S2).

Table 1.

Summary of kinetic rate constant and K_D data for the binding of human and cynomolgus FLT3L protein to lead clones from optimization of 5D9.³²

Interaction	Stage	kon (× 105 M−1s−1)	koff (× 10−2 s−1)	KD (pM)
Human FLT3L/human FLT3 receptor		3.6	81.2	22,340
Human FLT3L/5D9 parent	Parental	0.6	707.2	1,157,000
Human FLT3L/5D9-C06	Round 1	5.5	8.8	1,610
Human FLT3L/AM40-FAB	Round 2	10.9	1.8	170
Cynomolgus FLT3L/AM40FAB	Round 2	26.6	1.7	63

5D9, parental antibody for AMG 329; AM40-FAB, antigen-binding region of AMG 329; FLT3, feline McDonough sarcoma-like tyrosine kinase 3; FLT3L, feline McDonough sarcoma-like tyrosine kinase 3 ligand; KD, equilibrium binding constant; Kon, association rate constant; Koff, dissociation rate constant.

Species cross-reactivity

The sequence homology of the FLT3L extracellular domain is 95% for humans and cynomolgus monkeys and 73% for humans and mice. 5D9, the parental antibody for AMG 329, demonstrated binding to both human and cynomolgus monkey FLT3L, but not to mouse FLT3L (Figure 1).

Figure 1.

Cross-reactivity of 5D9 for human, cynomolgus monkey, and mouse FLT3L. 5D9 demonstrated binding to both human and cynomolgus monkey FLT3L but not mouse FLT3L.³³ 5D9, parental antibody for AMG 329; Ab, antibody; FLT3-Fc, FLT3-Fc chimera protein; FLT3L, feline McDonough sarcoma-like tyrosine kinase 3 receptor ligand; IgG, immunoglobulin G; MFI, mean fluorescence intensity.

Neutralization of FLT3L

AMG 329 specifically neutralized both soluble and cell-bound human FLT3L, as demonstrated by the increased expression of FLT3 on the cell surface (Figure 2). As expected, based on sequence homology comparisons, neither mouse nor rat FLT3L was neutralized by AMG 329, demonstrating that neutralization of FLT3L was species-specific (Figure S3), and rodent models were not relevant for toxicology studies. AMG 329 specifically bound to FLT3L and did not bind to or neutralize other class III tyrosine kinase receptor ligands (SCF, macrophage colony stimulating factor [M-CSF], or platelet-derived growth factor [PDGF]) (Figure S4). BLI with AMG 329, CAT26, and MAB608 demonstrated that AMG 329 blocks the FLT3L-FLT3 interaction (Figure S2).

Figure 2.

AMG 329 neutralizes soluble and cell-bound human FLT3L. (a) Increasing concentrations of anti-FLT3L antibodies were preincubated with 96 pM human soluble FLT3L before being added to human FLT3-expressing RS4;11 cell cultures. FLT3 expression was quantified using a fluorescently labeled anti-FLT3 antibody and measured by flow cytometry. Increased fluorescence indicates a higher level of FLT3 on the cell surface, demonstrating neutralization of FLT3L activity. Each point represents the average of duplicate wells. (b) AM40/AMG 329 neutralizes the activity of cell-surface FLT3L on CD4+ T cells at concentrations greater than 1 nM. Increased FLT3 expression indicates a higher level of FLT3 on the cell surface, demonstrating the neutralization of FLT3L activity. Neutralization of cell-bound FLT3L was measured using FLT3L-expressing CD4+ T cells co-cultured with FLT3-expressing RS4;11 cells. The isotype control curve is at 0% expression. 5D9, parental antibody for AMG 329; CD, cluster of differentiation; FLT3, feline McDonough sarcoma-like tyrosine kinase 3 receptor; FLT3-Fc, FLT3-Fc chimera protein (positive control); FLT3L, FLT3 ligand; MFI, mean fluorescence intensity; pAb, polyclonal antibody.

FcR binding, antibody-dependent cellular cytotoxicity (ADCC), and C1q binding

AMG 329 showed low binding to human Fc-γ receptors (FcγR), no binding to FcγRI or FcγRIIIB, and low levels of binding to both forms of FcγRIIA, both forms of FcγRIIIA, and FcγRIIB (Table S1). Binding to the Fc neonatal receptor (FcRn) was unaffected by the triple mutation in the Fc of AMG 329 (Table S2). Cytotoxicity curves from a representative blood donor indicated that AMG 329 did not induce cell death in CD4⁺ T cells (treated with IL-7 to upregulate FLT3L expression) in the presence of allogeneic peripheral blood mononuclear cells (Figure 3). No binding to human C1q was observed with AMG 329, even when tested at concentrations up to 100 µg/mL (Figure 4).

Figure 3.

AMG 329 does not show antibody-dependent cellular cytotoxicity. IL7–treated CD4⁺ T cells and allogeneic PBMCs were incubated with anti-CD4, AMG 329, or hIgG-TM control. CD4⁺ T cells were treated with IL7 to upregulate FLT3L expression. CD4⁺ T cells were incubated with PBMCs at a 1:25 ratio and assayed with a range of antibody concentrations (anti-CD4, AMG 329, or hIgG-TM control) for 6 hours. Cytotoxicity was measured using a colorimetric assay to detect lactate dehydrogenase. Data are presented from one donor representative of six independent experiments of nonrepeating PBMC donors. Each point represents the average of triplicate wells. Error bars represent standard deviation. CD, cluster of differentiation; FLT3L, feline McDonough sarcoma-like tyrosine kinase 3 receptor ligand; hIgG, human immunoglobulin G; PBMC, peripheral blood mononuclear cells; TM, triple mutation.

Figure 4.

AMG 329 did not bind human C1q in vitro. Binding of AMG 329 to C1q was evaluated by ELISA. Serial dilutions of AMG 329, rituximab (human IgG1; positive control), and IgG4 S241P L248E (Abzena; hinge stabilized and Fc silenced; negative control) were incubated with purified human C1q. Anti-C1q-HRP was used to measure binding with the TMB microwell peroxidase assay. All samples were tested in duplicate and ran on two separate occasions. EC50, half maximal effective concentration; ELISA, enzyme-linked immunosorbent assay; IgG, immunoglobulin G; NA, not applicable; OD, optical density.

Primate studies

Pharmacodynamics and pharmacokinetics

Overall, peripheral blood immunophenotyping evaluations in cynomolgus monkeys (Figure S5) showed that AMG 329 administration reduced pDC, cDC1, cDC2, and classical monocyte relative proportions from baseline in a non–dose-dependent manner at the dose levels tested compared with concurrent vehicle control values. During the 5-week, 5-dose toxicity study, AMG 329-related decreases in pDC relative proportions were noted beginning on day 8 of the dosing phase with significant decreases vs control on days 15, 29, and 36 for all treatment groups (adjusted p < .05); maximal decreases were observed between days 25 and 81 of the recovery phase (Figure 5a). Similar trends were observed during the 27-week, 14-dose chronic toxicity study, with AMG 329-related decreases in pDC counts observed beginning on day 15 of the dosing phase (adjusted p < .001 vs vehicle control at days 29 and 190 for all treatment groups) that were maintained through day 137 of the recovery phase (Figure 5b).

Figure 5.

Peripheral blood immunophenotyping evaluations in cynomolgus monkeys: pDcs over time. Data were pooled for males and females and presented as the percent of baseline of relative proportions (percent of CD45^brightSSC^dim lymphocytes and CD45^moderateSSC^moderate monocytes) during (a) the 5-week, 5-dose toxicity study and (b) the 27-week, 14-dose chronic toxicity study. *adjusted p < .05; **adjusted p < .01; ***adjusted p < .001 vs control. IQR, interquartile range; IV, intravenous; pDC, plasmacytoid dendritic cell; SC, subcutaneous.

During the 5-week, 5-dose toxicity study at the dose levels tested, non–dose-dependent decreases in cDC1 relative proportions were noted during the dosing phase for animals administered AMG 329 vs vehicle controls (adjusted p < .05 for AMG 329 30 mg/kg on days 8, 15, 29, and 36, and intravenously [IV] administered AMG 329 100 mg/kg on days 15, 29, and 36); cDC1 levels in AMG 329 treated animals remained depleted through day 109 of the recovery phase (Figure 6a). During the 27-week, 14-dose chronic toxicity study, non–dose-dependent decreases in cDC1 relative proportions at the dose levels tested were observed between day 8 of the dosing phase through day 53 of the recovery phase (adjusted p < .05 at day 29 of the dosing phase for AMG 329 5 mg/kg and 100 mg/kg IV at day 29 and day 190 of the dosing phase for subcutaneously [SC] administered AMG 329 100 mg/kg) (Figure 6b).

Figure 6.

Peripheral blood immunophenotyping evaluations in cynomolgus monkeys: cDC1 counts over time. Data were pooled for males and females and presented as the percent of baseline of relative proportions (percent of CD45^brightSSC^dim lymphocytes and CD45^moderateSSC^moderate monocytes) during (a) the 5-week, 5-dose toxicity study and (b) the 27-week, 14-dose chronic toxicity study. *adjusted p < .05; **adjusted p < .01 vs control. cDC1, conventional dendritic cell 1; IV, intravenous; SC, subcutaneous.

During the 5-week, 5-dose toxicity study, non–dose-dependent decreases in relative proportions of cDC2 were observed at the dose levels tested on days 8, 29, and 36 of the dosing phase (for AMG 329 30 mg/kg, 100 mg/kg SC, and 100 mg/kg IV vs control, adjusted p < .05 on day 29 and p < .01 on day 36) and maintained through day 81 of the recovery phase (Figure S6a). During the 27-week, 14-dose chronic toxicity study, non–dose-proportional decreases in the relative proportions of cDC2 were noted from days 8 through 190 at the dose levels tested during the dosing phase in animals administered AMG 329 vs vehicle controls (adjusted p < .01 for all treatment groups at day 190) with some maintenance through the recovery phase (Figure S6b). Relative proportions of pDCs, cDC1s, and cDC2s generally returned to baseline levels by the end of the recovery phase for both studies.

Decreases in the relative proportions of classical monocytes during the 5-week, 5-dose toxicity study were observed on day 29 of the dosing phase in animals administered AMG 329 vs concurrent vehicle controls (adjusted p < .05 for AMG 329 30 mg/kg and p < .01 for AMG 329 100 mg/kg SC) (Figure 7a). Similarly, decreases in the relative proportions of classical monocytes were observed on days 29 and 190 of the dosing phase during the 27-week, 14-dose chronic toxicity study in animals administered AMG 329 vs vehicle controls (adjusted p < .05 at day 15 for AMG 329 30 mg/kg and 100 mg/kg IV and p < .01 for AMG 329 5 mg/kg and p < .05 for AMG 329 100 mg/kg IV at day 190); these decreases were maintained through day 137 of the recovery phase and returned to baseline levels by the end of the recovery phase (Figure 7b). No significant or meaningful changes in the relative proportions of nonclassical monocytes, intermediate monocytes, or B cells were observed in either study (Figures S7–S9).

Figure 7.

Peripheral blood immunophenotyping evaluations in cynomolgus monkeys: classical monocyte counts over time. Data were pooled for males and females and presented as the percent of baseline of relative proportions (percent of CD45^brightSSC^dim lymphocytes and CD45^moderateSSC^moderate monocytes) during (a) the 5-week, 5-dose toxicity study and (b) the 27-week, 14-dose chronic toxicity study. *adjusted p < .05; **adjusted p < .01 vs control. IV, intravenous; SC, subcutaneous.

Toxicokinetic exposure in cynomolgus monkeys, as evaluated by mean Cmax and AUC values of AMG 329, exhibited an approximately dose-proportional increase with escalating dose levels from 5 to 100 mg/kg SC. AMG 329 accumulation was observed following multiple SC and IV administrations. The half-life of AMG 329 was 7 days for SC and 11 days for IV administration at 100 mg/kg for 5 weekly doses. For 14 biweekly doses at the same concentration, the half-lives for SC and IV administration were 12 and 8.5 days, respectively.

The dosing regimen in the cynomolgus monkey toxicology study was designed to identify toxicity at dose levels significantly higher than the expected human exposure and to generate an acceptable safety margin for human clinical trials, rather than to investigate a dose dependent response in pharmacology.

T-cell – dependent antibody responses (TDAR) to keyhole limpet hemocyanin (KLH)

No AMG 329-related differences were observed in anti-KLH IgM or IgG titers in cynomolgus monkeys treated with KLH; the magnitude and duration of KLH responses were similar across dosing groups when compared with the vehicle control (Figure S10).

Discussion

AMG 329 is a first-in-class human monoclonal antibody that binds to and neutralizes FLT3L with high affinity and specificity. Our results demonstrate that AMG 329 does not induce ADCC and is unlikely to induce complement-dependent cytotoxicity, suggesting a low possibility of Fc-related off-target effects. Importantly, the AMG 329-related reduction in peripheral blood pDCs, cDC1s, cDC2s, and classical monocytes in cynomolgus monkeys aligns with the expected pharmacology of AMG 329, with durable responses during the recovery phase observed for many of these cellular subsets; recovery toward control values 224 days postdosing suggests reversibility. In addition, TDAR responses to KLH in cynomolgus monkeys indicated intact immune competence³⁶ following AMG 329 administration, despite the substantial reduction in pDCs, cDCs, and classical monocytes. Together, these findings support the advancement of AMG 329 in clinical development for the treatment of inflammatory or autoimmune conditions.

DCs play important roles in activating both innate and adaptive immunity, initiating immune responses by interacting with T cells, and regulating the activation of immunity vs tolerance; thus, DCs have been implicated as potential therapeutic targets for various pathological conditions, including autoimmune diseases.³⁷ pDCs, in particular, have generated attention as potential therapeutic targets due to their involvement in multiple inflammatory and autoimmune diseases and their ability to produce large amounts of type I interferons and proinflammatory cytokines/chemokines.^13,18,38 Several molecules aimed at depleting or inhibiting pDCs or blocking signaling downstream of pDCs are currently under investigation.^13,18 For example, compounds directed against pDC surface receptors (e.g., anti-BDAC2 and anti-ILT7 monoclonal antibodies), inhibiting endosomal activation of TLR7/TLR9, or targeting the type I interferon pathway (e.g., anti-interferon AR1/interferon α monoclonal antibodies such as anifrolumab, sifalimumab, and rontalizumab) have been evaluated in clinical trials for immune-mediated conditions, including systemic lupus erythematosus, lupus nephritis, and rheumatoid arthritis.^13,18

Given that cDC1 and cDC2 have also been implicated in the pathogenesis of multiple autoimmune diseases^14–18 and therapies targeting pDCs/type I interferon do not affect cDC-driven pathways (e.g., IL-23, IL-15, Th17, cytotoxic T lymphocytes),¹⁰ targeting cDC1s, cDC2s, and pDCs with AMG 329 may provide a greater therapeutic benefit than targeting pDCs alone. While pDCs specialize in type I interferon production and cDCs specialize in antigen presentation, both cell types are capable of both functions. In vivo, pDCs and cDCs occupy discrete niches, and the depletion of both DC subsets may be required to inhibit established, auto-reactive, tissue-resident adaptive immune cells. To our knowledge, no other compounds targeting a combination of three DC subtypes (pDC, cDC1, and cDC2) or FLT3L have been evaluated in clinical studies. The ability of AMG 329 to simultaneously inhibit pDC- and cDC-driven pathology in autoimmune diseases is unique not only because of the dual pharmacology with a single agent, but also because target cell depletion occurs largely via a noninflammatory mechanism (in contrast to agents acting through ADCC, T-cell engagement, or antibody-drug conjugates). Additionally, target cell depletion via neutralization of a survival factor does not require cell contact with a natural killer cell, T cell, or other effector cell. These features differentiate AMG 329 from other therapies and are reflected by the robust pharmacodynamic effects described herein.

Monocytes have also been implicated in the development and progression of autoimmune diseases; monocyte infiltration is observed in multiple autoimmune conditions, including systemic sclerosis, rheumatoid arthritis, and Sjögren’s disease.^19,20 Monocytes contribute to autoimmune disease pathogenesis by promoting inflammation, over-activation of lymphocytes, and fibrogenesis, which can lead to tissue damage;^19,20 therefore, by reducing monocytes, in addition to pDCs and cDCs, the mechanism of action for AMG 329 has a potential advantage over therapies targeting individual cell types.

Pharmacologic or genetic inhibition of FLT3/FLT3L signaling has been shown to reduce DC numbers in multiple studies.^21–23 Bone marrow toxicity following the neutralization of FLT3L has not been reported nor observed in the current study; this suggests there may be redundancy with c-kit for bone marrow progenitors, whereas DC populations that constitutively express FLT3 are rapidly depleted by neutralization of FLT3L. In an experimental autoimmune encephalomyelitis model, symptomatic mice treated with a FLT3 inhibitor displayed significant improvements in disease progression compared with mice treated with vehicle controls, demonstrating the therapeutic potential of FLT3/FLT3L inhibition in autoimmune diseases. Targeting FLT3L with AMG 329 may also have applications beyond autoimmunity.

The results presented are limited to cell-based assays and healthy cynomolgus monkeys. These findings establish AMG 329 as a promising therapeutic candidate for multiple autoimmune or inflammatory diseases and AMG 329 is currently being investigated in a Phase 2 clinical study (NCT06312020).

Materials and methods

Derivation, optimization, and in vitro characterization of AMG 329

Derivation and optimization

A combined set of human antibody phage display libraries with a Screening in Product Format approach³⁵ was used to identify antibodies targeting human FLT3L (membrane bound and soluble) with cross-reactivity to cynomolgus monkey FLT3L. This format included the L234F/L235E/P331S TM in the heavy chain-constant FcR to minimize ADCC and complement-dependent cytotoxicity.³⁴ A solution-phase panning strategy was used over three iterative rounds with decreasing concentrations of human FLT3L (100–25 nM). Isolates from the third round were assessed using a competition panning assay. After selection, isolates were expressed in HEK293 cells; expression supernatants were assessed via homogenous time-resolved fluorescence for inhibition of the human FLT3/FLT3L interaction to identify the parental clone, 5D9. Two rounds of affinity optimization were performed using parsimonious and block mutagenesis to enhance the affinity of the parental clone. The final clone AM40 was renamed MEDI1116; over the course of clinical development, the final clone was renamed VIB-1116, then HZN-1116, and currently AMG 329.

Phage panning and screening

A detailed methodology of the naïve phage library panning approach leading to the isolation of the AMG 329 precursor 5D9 has been previously described.³³ Four phage display libraries were used: the Bone Marrow Vaughan (BMV) library, a nonimmunized library of 1.4 × 1,010 scFv fragments,³⁰ a combined spleen (CS) library of 1.3 × 1,011 scFv fragments,³¹ the DP47 library, a naïve human library of 1 × 1,010 scFv fragments,²⁹ and the Dyax library with > 1,010 Fabs.³² These libraries were panned by employing two different parallel strategies: alternative panning, using alternating rounds of human and mouse FLT3L panning, and competition panning, using human FLT3-Fc in > 100-fold excess as the phage elution agent. Isolates from the second or third round were assessed using a competition panning assay. After selection, isolates were expressed as Scfv-Fcs in HEK293 cells; expression supernatants were assessed via HTRF for inhibition of the human FLT3/FLT3L interaction. Further screening in IgG TM format resulted in the identification of 5D9, the parental clone to AMG 329.

Affinity optimization of clone 5D9

The parental clone 5D9 bound FLT3L with low affinity. Therefore, affinity maturation methods were used to achieve a target KD of approximately 300 pM.³³ Prior to the start of affinity maturation, antibody germlining was used to change framework residues to the nearest antibody germlines. For affinity maturation, block and parsimonious mutagenesis strategies were used in parallel. Block mutagenesis involved generating libraries of 5D9 containing random “blocks” or stretches of 5–6 positions in the complementary-determining regions (CDRs) in an overlapping pattern. Resulting libraries of about 1x10⁶ to 1x10⁷ clones were first enriched using phage display panning techniques and later screened using HTRF. Parsimonious mutagenesis consisted of mutating every position in all six CDRs to all 20 amino acids and then combining any beneficial mutations.

After panning and screening, 24 clones from block mutagenesis and 30 clones from parsimonious mutagenesis were converted to an IgG format and assessed for affinity improvement. The best of these was designated as clone 5D9-Clone 6 (C06), which achieved a KD of 1,610 pM (measured by Biacore). This was approximately a 700-fold improvement in affinity over the parental 5D9. A further round of both block and parsimonious mutagenesis was used to achieve sub 300 pM KD values. One clone meeting these criteria, AM40, had an affinity of ~ 170 pM to human FLT3L and ~ 63 pM to cyno FLT3L (Table 1). AM40 bound and neutralized endogenous cell-surface FLT3L and bound endogenous soluble FLT3L in human serum. The AM40 binder was combined with the IgG-TM silenced backbone and renamed MEDI1116; over the course of clinical development, the final clone was renamed VIB-1116, then HZN-1116, and currently AMG 329.

Species cross-reactivity

Human and cynomolgus monkey sequence homology of FLT3 and the extracellular domain of FLT3L (corresponding to amino acids 27–184 of human FLT3L) were examined using the Clustal O (1.2.4) multiple sequence alignment algorithm.³⁹ Species cross-reactivity was measured using Chinese hamster ovary cells expressing FLT3L derived from humans, cynomolgus monkeys, or mice. Cells were incubated with the parental antibody, 5D9, or isotype control for one hour at 4°C followed by detection with phycoerythrin-labeled antihuman IgG secondary antibody; FLT3-Fc (a human FLT3-Fc chimera protein that binds to and inhibits FLT3L function; R&D Biosystems #368-ST) was used as a positive control.

Kinetic binding

Kinetic binding of AMG 329 to human and cynomolgus monkey FLT3L protein was measured using surface plasmon resonance interaction analysis with a streptavidin-capture assay format (Biacore T200; Cytiva). Biotin-labeled human FLT3L was immobilized on a streptavidin sensor chip with a final surface density of ~ 135 resonance units. AM40-FAB, the antigen-binding region of AMG 329, was injected over human FLT3L and reference surfaces at a flow rate of 75 μL/min; a sequential approach was utilized for kinetic measurements using three-fold serial dilutions of AM40-FAB (0.023–50 nM). The resulting binding response curves yielded the association phase data. Binding data corrected for injection artifacts and/or nonspecific binding interactions were globally fitted to a 1:1 binding model (BIAevaluation 4.1 software; Cytiva) to determine the kinetic rate constants k_on and k_off, from which the apparent dissociation constant was calculated as k_off/k_on. The same procedure was repeated to determine the affinity of AM40-FAB for cynomolgus monkey FLT3L.

Neutralization of FLT3L

In vitro cell-based assays were used to measure the neutralization of soluble and cell-bound FLT3L by AMG 329. Soluble human FLT3L (96 pM) was preincubated with increasing concentrations of AMG 329, FLT3L-Fc, or anti-mouse FLT3L and then added to human FLT3-expressing RS4;11 cells. FLT3 expression was quantified by flow cytometry using an anti-FLT3-PECy7 antibody (BioLegend Clone BV10A4H2). Neutralization of cell-bound FLT3L was measured using FLT3L-expressing CD4⁺ T cells co-cultured with FLT3-expressing RS4;11 cells at a ratio of 15:1 (CD4⁺:RS4;11). Cells were preincubated with 5D9 (the parental antibody of AMG 329), AM40 (AMG 329), or isotype control at concentrations of 0.01–100 nM; FLT3 expression was analyzed by flow cytometry. When FLT3L binds to FLT3, FLT3 is downregulated/internalized, thus lower surface FLT3 expression is an indirect measure of FLT3L binding.^40–42

FcR binding, ADCC, and C1q binding

Binding of AMG 329 to FcγR and FcRn was assessed by steady-state affinity analysis (Biacore T200; Cytiva). His-tagged FcγR (Sino Biological) was loaded to a specific resonance unit level at the start of each cycle, and five three-fold serial dilutions of AMG 329 and trastuzumab (positive control antibody; Roche #N3022H01) without regeneration between each concentration was used for each receptor tested. Antibodies were passed over the chip at increasing concentrations, running at a flow rate of 30 μL/min, followed by a single dissociation step. Sensorgrams were analyzed using 1:1 kinetics for high-affinity FcγRI and steady-state binding for low-affinity FcγR. For FcRn analyses, human FcRn (Sino Biological #CT009-H08H) was coupled onto a Series S CM5 (carboxymethylated dextran) sensor chip (Cytiva) at 10 μg/mL in sodium acetate pH 5.5 using standard amine coupling. Purified antibodies were titrated in an eight-point serial dilution from 31.25–4,000 nM in phosphate-buffered saline (PBS) containing 0.05% polysorbate 20 pH 6.0 or a three-point serial dilution from 500–2000 nM in PBS containing 0.05% polysorbate 20 pH 7.4. Antibodies were passed over the chip at increasing concentrations at a flow rate of 30 μL/min at 25°C.

The ADCC activity of AMG 329 was measured using an in vitro cell culture system (IL7–treated CD4⁺ T cells and allogeneic peripheral blood mononuclear cells incubated with anti-CD4 [Absolute Antibody #AB02329–10.0], AMG 329, or human IgG-TM control). CD4⁺ T cells and peripheral blood mononuclear cells were isolated from fresh human blood from distinct donors; CD4⁺ T cells were incubated with peripheral blood mononuclear cells at a 1:25 ratio and assayed with a range of antibody concentrations (anti-CD4, AMG 329, or human IgG-triple mutation control) for six hours. Cytotoxicity was measured using a colorimetric assay to detect lactate dehydrogenase (Cytotox96 Non-radioactive Cytotoxicity Assay; Promega #G1782); plates were read on a spectrophotometer at 490 nm.

The complement-dependent cytotoxicity activity of AMG 329 was assessed through the binding of AMG 329 to C1q, which was evaluated by ELISA. Plates were coated with serial dilutions of AMG 329, IgG4 S241P L248E (negative control), or rituximab (Roche) (positive control). AMG 329 and the negative control were coated at a starting concentration of 100 μg/mL, with a seven-point, five-fold serial dilution performed (100–0.0064 μg/mL). Rituximab was used at a starting concentration of 10 μg/mL, and a seven-point, two-and-a-half-fold serial dilution (10–0.04096 μg/mL) was performed. After overnight incubation followed by blocking (1× PBS pH 7.4 + 2% [w/v] bovine serum albumin), 5 μg/mL of purified human C1q (Pathway Diagnostics) was added to the ELISA plates for binding. Anti-C1q-HRP (Abcam) diluted 1:100 in blocking buffer was added to each well for detection. Binding was detected using the TMB Microwell Peroxidase assay, and absorbance was read at 450 nm using a SpectraMax i3× (run 1) or SpectraMax M3 plate reader (run 2) (Molecular Devices). All samples were tested in duplicate and ran on two separate occasions. GraphPad Prism 9.0 (GraphPad Software) was used for data analysis, and data were fitted using a four-parameter nonlinear regression.

Primate studies

Pharmacodynamics

A 5-week, 5-dose toxicity study with a 22-week recovery phase and a 27-week, 14-dose chronic toxicity study with a 32-week recovery phase were conducted to evaluate the nonclinical safety of AMG 329 using the pharmacologically relevant species (cynomolgus monkey). In the 5-week, 5-dose toxicity study, male and female cynomolgus monkeys were administered with vehicle (20 mM histidine/histidine-HCl, 240 mM sucrose, 0.04% polysorbate 80, pH 6.0) or AMG 329 once weekly at dose of 5, 30, or 100 mg/kg/week SC or a dose of 100 mg/kg/week IV. In the 27-week chronic toxicity study, the same vehicle or AMG 329 was administered to male and female cynomolgus monkeys once every 2 weeks for a total of 14 doses. AMG 329 was administered at 5, 30, or 100 mg/kg/dose SC or 100 mg/kg/dose IV. All procedures involving cynomolgus monkeys were conducted in an Association for Assessment and Accreditation of Laboratory Animal Care-accredited facility and were reviewed and approved by the Labcorp Early Development Laboratories (Madison, WI) Institutional Animal Care and Use Committee.

Lymphocyte populations were identified using a non-GLP flow cytometry method on blood samples collected via the femoral vein; lymphocyte populations are defined in Table S3. Immunophenotyping results were pooled for males and females and reported as the median (interquartile range) percent of baseline of relative proportions for each cell type; relative proportions were defined as percent of CD45^brightSSC^dim lymphocytes for T-cell and B-cell populations or percent of CD45^brightSSC^dim lymphocyte and CD45^moderateSSC^moderate monocytes for DC and monocyte populations. Wilcoxon rank sum tests were used for statistical comparisons between each AMG 329 treatment group and vehicle controls with an FDR p value adjustment to correct for multiple testing. Animals with starting baseline values of zero were excluded from percent of baseline calculations.

TDAR to KLH

To evaluate TDAR (performed during the 27-week, 14-dose chronic toxicity study only), blood anti-KLH IgM and IgG titers were measured using a validated ELISA assay at multiple timepoints throughout the study from predose through recovery. KLH was administered SC into the right thigh at a volume of 1 mL/animal (10 mg/mL) 14 days before the primary administration of vehicle or AMG 329 (to evaluate primary immune response), on day 92 of the dosing phase (secondary recall response), and on day 169 of the recovery phase (to assess tertiary response once pDCs largely returned to baseline). Data were pooled for males and females and reported as the percent of baseline titers. Wilcoxon rank sum tests were used for statistical comparisons between each AMG 329 treatment group and vehicle controls with an FDR p value adjustment to correct for multiple testing.

Funding Statement

Provided by Horizon Therapeutics, PLC (now Amgen Inc.).

Disclosure statement

Annie Lau-Kilby, Agata Bartczak, Susan Chyou, James Hester, Qian Wang, M. Jack Borrok and Kamelia Zerrouki are current full-time employees of Horizon Therapeutics (now Amgen) and may hold Horizon/Amgen stock or stock options. Michele Gunsior is a former full-time employee of Horizon (now Amgen) and AstraZeneca and may hold Horizon/Amgen and/or AstraZeneca stock or stock options, and is a current employee of Astria Therapeutics. Anna M. Hansen, Dorothy Sims, and Yan Chen are full-time employees of AstraZeneca, and may hold AstraZeneca stock or stock options. Xiaodong Xiao is a former employee of AstraZeneca and may hold AstraZeneca stock or stock options. Peter Pavlik is a former employee of AstraZeneca and a current employee of Aptevo Therapeutics. Kerry A. Casey is a former employee and shareholder of AstraZeneca and a current employee and shareholder of Regeneron Pharmaceuticals. William A. Rees is a former full-time employee of Horizon (now Amgen) and may hold Horizon/Amgen stock or stock options.

Author contributions

Annie Lau-Kilby, Michele Gunsior, Anna M. Hansen, Dorothy Sims, Xiaodong Xiao, Peter Pavlik, Yan Chen, Kerry A. Casey, Kamelia Zerrouki, and William A. Rees contributed to the study concept and design. Annie Lau-Kilby, Agata Bartczak, Susan Chyou, Anna M. Hansen, Dorothy Sims, Xiaodong Xiao, Peter Pavlik, Yan Chen, Kerry A. Casey, Kamelia Zerrouki, Qian Wang and William A. Rees contributed to data acquisition. Agata Bartczak, Susan Chyou, and Anna M. Hansen performed statistical analyses. Annie Lau-Kilby, Michele Gunsior, Agata Bartczak, Susan Chyou, James Hester, Dorothy Sims, Xiaodong Xiao, Peter Pavlik, Yan Chen, Kerry A. Casey, Kamelia Zerrouki, Qian Wang, M. Jack Borrok, Anna M. Hansen, and William A. Rees contributed to data interpretation and provided their critical review and final approval of the manuscript.

Data availability statement

Access to data may be granted upon reasonable request to qualified researchers. Data can be requested by submitting a research proposal and statistical analysis plan to Horizon Therapeutics (now Amgen). Data will be provided following the review and approval of the plan and execution of a data sharing agreement. For more information or to submit a request, please e-mail medinfo@amgen.com. Qualified researchers may request data from Amgen clinical studies. Complete details are available at the following: https://wwwext.amgen.com/science/clinical-trials/clinical-data-transparency-practices/clinical-trial-data-sharing-request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2025.2527677

Abbreviations

ADCC

antibody-dependent cellular cytotoxicity

cDC

conventional dendritic cell

DC

dendritic cell

ELISA

enzyme-linked immunosorbent assay

FcR

Fc receptor

FcγR

Fc-γ receptor

FcRn

Fc neonatal receptor

FLT3L, FLT3

Feline McDonough sarcoma-like tyrosine kinase 3 ligand, Feline McDonough sarcoma-like tyrosine kinase 3

Ig

immunoglobulin

KLH

keyhole limpet hemocyanin;

IL

interleukin

IV

intravenous

M-CSF

macrophage colony-stimulating factor

PBS

phosphate-buffered saline

pDC

plasmacytoid dendritic cell

PDGF

platelet-derived growth factor

SC

subcutaneous

SCF

stem cell factor

TDAR

T-cell–dependent antibody responses.