Effectorless Fc-fusion improves FLT3L drug-like properties for cancer immunotherapy combinations

Summary

Background

Conventional dendritic cells (cDCs), are central to antitumour immunity, but their low prevalence in tumours limits the efficacy of immunotherapies. FLT3L is a key growth factor regulating cDCs development in the bone marrow. It expands cDCs when administered exogenously, favouring antitumour T cell priming and tumour control. Currently, FLT3L pharmacokinetic (PK) and pharmacodynamic (PD) properties require daily dosing for up to 14 days, which may limit its clinical use. In the present study, we developed and characterised a therapeutic modality named FLT3L-Fc NG2LH.

Methods

We improved human FLT3L PK properties by fusing it with a modified fragment crystallisable (Fc) domain of IgG1. To prevent Fc gamma receptor (FcγR) mediated effector function, we engineered an effectorless Fc format called NG2LH, consisting of the aglycosylation substitution N297G, combined with a graft of the lower hinge region of IgG2 onto an otherwise IgG1 Fc.

Findings

FLT3L-Fc NG2LH had limited binding to FcγRs and failed to elicit antibody dependent cellular cytotoxicity (ADCC) and cellular phagocytosis (ADCP). PK/PD studies using a mouse effectorless equivalent, mFLT3L-Fc, showed that a single injection of mFLT3L-Fc leads to sustained expansion of cDCs in blood, spleen, and B16F10 tumours. When combined with polyI:C and anti-PD-L1, a single mFLT3L-Fc injection delays the growth of B16F10 tumours and reinvigorates CD8+ T cell immunity.

Interpretation

The improved properties of FLT3L-Fc NG2LH are expected to mitigate the practical limitations of FLT3L usage in the clinic, and constitute an asset for future cancer immunotherapy combination regimens leveraging cDC biology in situ.

Funding

This work was performed at, and funded by Genentech Inc. South San Francisco, CA 94080, USA.

Research in context.

Evidence before this study

Dendritic cells are considered to be a limiting factor to immune checkpoint blockade efficacy in patients with cancer. These cells can be expanded in situ by the growth factor FLT3L, fostering antitumour immunity. However, in its unmodified form, FLT3L treatment requires daily dosing up to 14 days, hampering its broader use in the clinic.

Added value of this study

We developed a FLT3L-Fc fusion with improved drug-like properties that allows for sustained expansion of dendritic cells upon a single injection, and stimulation of antitumour immunity when combined with polyI:C and anti-PD-L1.

Implications of all the available evidence

By easing dosing constraints, FLT3L-Fc NG2LH sets the stage for a broader exploration of FLT3L based immunotherapies in patients with cancer.

Introduction

Discovered in 1973,¹ dendritic cells (DCs) are considered to be the cornerstone of adaptive immunity.² High resolution expression analysis studies have identified four major DC populations conserved between mouse and human: two main subsets of conventional DCs (cDC1 and cDC2), plasmacytoid DCs (pDCs) and monocyte-derived DCs (moDCs).^3,4 While the specific functions of these subsets are still unravelling, an array of evidence supports that cDC1 and cDC2 play a critical role in initiating and maintaining antitumour T cell immunity.^5,6

cDC1s are exceptionally good at capturing tumour-derived material and directing tumour antigens to the MHC-I cross-presentation machinery for priming of tumour-specific CD8+ T cells.7, 8, 9 Preclinical studies have shown increased tumour burden in Batf3−/− (cDC1 deficient) mice, caused by a defect in tumour-specific T cell immunity.¹⁰ Interestingly, while these processes rely on the ability of cDC1 to migrate from tumours to tumour-draining lymphoid organs,⁹ tumour-resident cDC1 are also essential to sustain intratumoral CD8+ T cell cytolytic activity.¹¹ The role for cDC2s in antitumour immunity is less clear. cDC2s have been shown to promote antitumour CD4+ T cell responses,¹² but also sustain CD8+ T cells in the tumour-microenvironment (TME).¹³ The recent development of cDC2 deficient mice¹⁴ should help clarify their contribution in antitumour immunity. In humans, single cell RNA sequencing analyses have recently shown that cDCs, and particularly cDC1s, represent a minor fraction of intratumoral myeloid cells across many indications.¹⁵ Despite this scarcity, cDC1 gene signatures have been associated with better prognosis in patients with cancer.16, 17, 18 Preclinical experiments have also established an important role for cDC1s in response to immunotherapy.¹⁹ cDC gene signatures positively correlate with response to anti-PD-L1 treatment in patients with cancer,²⁰ an observation further highlighted by our analysis of the non–small cell lung cancer (NSCLC) phase 3 OAK study (NCT02008227, Supplementary Figure S1). Overall, cDCs are rare but crucial populations in tumours that facilitate antitumour immunity at multiple points in space and time. A paucity in DCs at the tumour site could limit the efficacy of checkpoint blockade therapies such as anti-PD-L1, justifying the current need to develop new therapeutic modalities to expand the DC compartment to improve response to checkpoint blockade therapy.

cDCs develop in the bone marrow through a series of tightly regulated differentiation steps originating from haematopoietic stem cells (HSCs).² Briefly, HSCs differentiate into monocytes and dendritic cell progenitors (MDP) that further differentiate into common dendritic cell progenitors (CDP). CDPs ultimately give rise to pre-cDCs, which egress from the bone marrow and populate peripheral tissues as cDC1s and cDC2s. While there are a multitude of soluble cues and transcription factors involved in cDCs ontogenesis, FMS-like tyrosine kinase 3 ligand (FLT3L) has been identified as an essential growth factor regulating their development.^21,22 Indeed, Flt3l−/− mice harbour systemic cDC deficiencies,^22,23 and exogenous FLT3L treatment promotes cDC expansion in mice and humans.^24,25

Based on these observations, FLT3L has been explored as an in situ cDC modulator to promote antitumour immunity,²⁶ however, the sole expansion of cDCs is not sufficient and an activation/maturation agent of cDCs is also required for their ability to prime T cells and induce an antitumour response. This has led to combination strategies aiming to complement FLT3L activity. The current model focuses on cDC1s to optimise their antitumour functions via an expand-load-activate approach, where: 1) FLT3L expands cDCs, 2) cDC1s populate the TME and load tumour-associated antigens, and 3) innate agonists activate cDC1s, which migrate to tumour draining lymphoid organs and prime tumour-specific CD8+ T cells. This combination approach has been shown to be effective in multiple preclinical tumour models that are usually refractory to immunotherapy,27, 28, 29, 30, 31 and similar strategies are now translating into clinical applications.³² While clinical data are still limited, FLT3L combination therapies have shown promising results in patients with indolent non-Hodgkin's lymphoma.²⁸

FLT3L is a type I transmembrane protein that can be shed from the cell surface by the proteases ADAM10 and ADAM17.^33,34 It has been established that both membrane-tethered and soluble FLT3L are biologically active,²¹ though their respective contribution in vivo remains to be clarified. One limitation for using FLT3L in the clinic relates to the relatively small molecular size of its soluble form. The active soluble FLT3L is a non-covalent, 44 kDa homodimer³⁵ that is rapidly cleared from the organism when given exogenously.³⁶ In both mice and humans, PK property requires daily FLT3L dosing up to 14 days to observe optimal cDCs expansion.^25,27 This limitation has logistical implications that may hamper the ability to expand the usage of FLT3L in the clinic.

Here we describe the development of a FLT3L drug candidate with improved PK and PD properties. Using an engineered effectorless Fc-fusion technology, we generated a human molecule, termed FLT3L-Fc NG2LH, with in vitro potencies comparable to FLT3L. In preclinical settings, a single injection of the effectorless mouse surrogate molecule FLT3L-Fc (mFLT3L-Fc) results in stable intratumoral cDC1 expansion. mFLT3L-Fc combination therapies reinvigorate CD8+ T cells and show antitumour efficacy. The development of FLT3L-Fc NG2LH mitigates the above-mentioned limitations of FLT3L usage and constitutes a valuable asset for cancer immunotherapy combination regimens to leverage the antitumour cDC/T cell axis in situ.

Methods

Recombinant proteins

Unless otherwise stated, all recombinant antibodies and Fc-fusion proteins were produced in transiently transfected Chinese hamster ovary (CHO) cells and purified by affinity chromatography (MabSelect SuRe, Cytiva), followed by size exclusion chromatography (Superdex 200, Cytiva). FLT3L (residues T27-P179) and mFLT3L (residues T28-P183), without fusion to Fc, were produced in CHO stable cell lines and purified by anion exchange chromatography (HiTrap Q FF, Cytiva), followed by hydrophobic interaction chromatography (HiTrap Phenyl HP, Cytiva), followed by size exclusion chromatography (Superdex 200, Cytiva). The amino acid sequences of mature FLT3L-Fc NG2LH and mFLT3L-Fc are listed in Supplementary Table S1.

FcγR binding of IgG variants

Binding of the IgG Fc variants to Fcγ receptors and FcRn was evaluated by surface plasmon Resonance (SPR) on a Biacore T200 (Cytiva). The IgG variants were captured at a concentration of 1 μg/mL, on a Protein L Sensor chip (Cytiva). For FcRn binding, FcRn was serially diluted 3-fold, 8 times, at a starting concentration of 10 μM in 100 mM MES, 150 mM NaCl, 0.05% Tween-20, pH 6.0 and injected for 3 min, followed by 3 min of dissociation. The affinities were determined by fitting the data to a steady state model. For FcγRIIa H131, FcγRIIb, and FcγRIIIa V158, the receptors were diluted in HBS-P at pH 7.4 (Cytiva) to 10 μM, or 111 nM for FcγRI, and injected for 3 min. The measured maximum responses (Rmax) are presented as a percentage of the maximum response of IgG1.

FcγR binding ELISAs

The binding interactions with various human FcγRs (IA, IIA-H131, IIA-R131, IIB, IIIA-F158, and IIIA-V158) were assessed in a panel of ELISA-based ligand binding assays. ELISA plates were coated with anti-human FLT3L-Fab, derived from an anti-FLT3L antibody (R&D Systems, Clone # 40416), in 0.05 M sodium carbonate buffer (pH 9.6) overnight at 4 °C. After blocking in assay buffer (PBS, 0.5% BSA, 0.05% Tween 20, 10% Blocker Casein (Thermo Fisher Scientific)), plates were incubated with serial dilutions of the test materials at RT for 2 h. Plates were then incubated with the Gly-6xHis-glutathione S-transferase (GST) polypeptide tag FcγRs at RT for 2 h and with rabbit polyclonal horseradish peroxidase (HRP)-anti-GST (Jackson ImmunoResearch) for 1 h. The plates were washed 5 times in PBS + 0.05% Tween 20 after each incubation step. Tetramethylbenzidine (Kirkegaard & Perry Laboratories) was added, as the chromogenic substrate and plates were incubated at RT for 5–20 min depending on the FcγR tested. The reaction was stopped with 1 M phosphoric acid and the 450 nm absorbance measured with a SpectraMax i3 microplate reader (Molecular Devices).

FLT3L-Fc NG2LH binding to recombinant human FLT3-Fc

Binding interactions between FLT3L-Fc NG2LH and recombinant human FLT3-Fc proteins were evaluated using SPR technology on a Biacore T200 instrument (Cytiva). FLT3L-Fc NG2LH was indirectly captured onto different flow cells (FCs) on a CM5 sensor chip using the method as described below. An anti-FLT3L antibody that recognises an epitope on the FLT3L domain of FLT3L-Fc NG2LH with minimal interference on FLT3L-Fc NG2LH binding to FLT3 was immobilised onto all FCs using an amine coupling procedure recommended by the manufacturer, resulting in an immobilisation level of approximately 13,000 response units (RU). FLT3L-Fc NG2LH was then injected over FC2, and the resulting capture level for FLT3L-Fc NG2LH was ∼20 RU. Various concentrations of recombinant human FLT3-Fc were diluted in running buffer (10 mM HEPES, 150 mM sodium chloride, 0.05% polysorbate 20, pH 7.4) and flowed over all FCs for 30–40 s at a flow rate of 50–100 μL/min; the dissociation was allowed to proceed for 1 min (for recombinant human FLT3-monoFc) or 15 min (for recombinant human dimeric FLT3-Fc). The surfaces were regenerated by applying 10 mM Glycine-HCl pH 1.5 for 30 s at 50 μL/min. This experiment was performed at 37 °C. Binding affinity and kinetic parameters (association and dissociation rate constants, and equilibrium dissociation constant) were calculated with the Biacore T200 Evaluation Software (version 3.1; Cytiva) using a 1:1 binding model. All sensorgrams were generated after in-line reference cell correction followed by buffer sample subtraction. Monomeric and dimeric forms of recombinant human FLT3-Fc fusion proteins were expressed in CHO cells with a monomeric Fc variant (monoFc)³⁷ or the WT Fc region of human IgG1 at the C-terminus respectively. An antibody targeting FLT3L-Fc NG2LH with minimal interference on FLT3L-Fc NG2LH binding to FLT3 was developed in-house using a rabbit single B cell-sorting culture and cloning technology.³⁸

Antibody dependent cellular cytotoxicity (ADCC) reporter assay [IgGs]

The NK-92 CD-16-FcεRlγ_NFAT-Luc cell line was generated for use in the reporter-based ADCC assay. The engineered NK-92 cells stably express an FCGR3A (high affinity variant)-FCER1G chimaera and an NFAT (nuclear factor of activated T cells) response element driving expression of firefly luciferase. The human FCGR3A-FCER1G chimeric cDNA was first chemically synthesised by the GeneArt™ Gene Synthesis service from Thermo Fisher Scientific. Amino acids 1–206 from FCGR3A (protein id = NP_001121065.1; coded by = NM_001127593.1) and amino acids 22–86 from FCER1G (protein id = NP_004097.1; coded by = NM_004106.1) were used for the chimeric receptor. A substitution of T to G at nucleotide 538 for FCGR3A was made to obtain the high affinity valine variant at codon 176. Restriction sites (EcoRI, 5′ end; BamHI, 3′ end) were added to the cDNA template as well as a Kozak sequence (GCCACC) immediately 5′ of the ATG start codon. The cDNA was subcloned into a lentiviral vector. A single clone having the correct ∼0.8 kb insert was chosen and used for further studies. The resulting construct was sequenced to confirm the entire cDNA insert. After completion of DNA sequencing, lentivirus particles were generated using this construct along with the NFAT-Luc reporter construct. These were used to transduce NK-92 cells to generate the reporter cell line. Dilutions of reference standard and samples were prepared in warm assay medium. Effector reporter NK-92 CD-16-FcεRlγ_NFAT-Luc cells and target WIL2-S cells (ATCC Cat. No. CRL 8885) were harvested and centrifuged at 300 g for 5 min and re-suspended in Assay Diluent. Effector/target cells suspensions were combined and 50 uL were added to the plate + 50 uL of prepared samples. Plates were incubated at 37 °C, 5% CO₂ for 2.5 h 50 uL/well of ONE-Glo™ Luciferase Reagent was added and plates incubated at 25 °C for 20 min and transferred to a Paradigm luminescence plate reader (Molecular Devices, Paradigm or i3x equipped with a LUM96 cartridge). Data were analysed using 4-parameter logistic curve fitting software (Molecular Devices, SoftMax Pro).

Antibody dependent cellular phagocytosis (ADCP) assay [IgGs]

This method was adapted from previous publications.^39,40 Human PBMCs were isolated from leukopaks (HemaCare) from two donors with FcγR genotype polymorphism: FcγRIIIA-158F/F, FcγRIIa 131H/R using Ficoll gradient centrifugation. CD14 positive monocytes were purified from PBMCs by negative depletion with a CD14 isolation kit (Stem Cell Technologies). Monocytes were plated at 0.1 × 10⁶ cells/cm² in X-VIVO-10 medium (Lonza) + 10% heat inactivated FBS. Macrophages were differentiated by the addition of 25 ng/mL of M-CSF (R&D Systems) for 7 days. 50 ng/mL of IFNγ (R&D Systems) was added for the final 24 h of differentiation. Target cells for the assay were GFP-expressing SKBR-3 cells. On the day of ADCP experiment, target cells and isolated macrophages were dislodged from culturing flasks with Accutase (Sigma–Aldrich). Cells were then incubated at 37 °C at a ratio of four macrophages (0.1 × 10⁶ cells/well) to one SKBR-3 cell (25,000 cells/well) for 24 h with treatment antibodies in ultra-low attachment 96-well U-bottom microplates (Corning) under a final volume of 200 μL of DMEM + 10% HI FBS. After 24 h, plates were centrifuged at 400 g for 4 min. 150 μL of supernatants were collected and transferred to another 96-well U-bottom plate. Pelleted cells were dislodged from the 96-well plates using Accutase and washed once with BD Stain Buffer (BD Biosciences). Macrophages were identified with anti-CD11b (clone ICRF44) and anti-CD14 (clone M5E2) antibodies (BD Biosciences) coupled to Alexa Fluor 647 (Invitrogen). Cells were acquired on an LSR Fortessa flow cytometer (BD Biosciences) and data were analysed with FlowJo (BD Biosciences). The percentage of cell killing was determined by measuring the reduction in GFP fluorescence resulting from its degradation in the lysosomes after internalisation using the following equation: Percentage of tumour cells killed = ((Percentage of GFP^pos, CD11b^neg, CD14^neg cells in no mAb control) − (Percentage of GFP^pos, CD11b^neg, CD14^neg cells in the presence of mAb))/(Percentage of GFP^pos, CD11b^neg, CD14^neg cells in no mAb control) × 100.

Antibody dependent cellular phagocytosis (ADCP) assay [FLT3L-Fc NG2LH]

ADCP assays were carried out using human monocyte-derived macrophages as effector cells and SEM (human acute lymphoblastic leukaemia line) as target cells. Briefly, macrophages were generated by isolating CD14+ cells from PBMCs (Miltenyi Biotec) and cultured in differentiation media (RPMI 1640, 10% FBS, 1% Glutamax, 1% Penicillin/Streptomycin, and 20 ng/mL M-CSF (R&D Systems)) at 37 °C, 5% CO₂. On day 3, cells were stimulated with 50 ng/mL of M-CSF and were cultured for 4 more days. On day 7, macrophages were stained with 10 μM Cell Trace Violet (Thermo Fisher Scientific), diluted to 1 × 10^ˆ6 cells/mL in ADCP assay media (IMDM, 10% FBS, 1% Glutamax, 1% Penicillin/Streptomycin), and added at 50 μL/well to a 96-well low adherent U-bottom plate (Corning). Target SEM cells were diluted in ADCP assay media at 2 × 10^ˆ6 cells/mL, pre-labelled with pHrodo (Thermo Fisher Scientific), and added (50 μL/well) to the assay plate containing macrophages. 100 μL of serial dilutions of FLT3L-Fc were added to each well and incubated for 4.5 h. Cells were centrifuged at 1200 rpm for 5 min and washed in PBS before fixation in 4% PFA for 10 min at 4 °C. Cells were acquired on a FACSCanto (BD Biosciences) and phagocytosis was analysed on FlowJo (BD Biosciences). Cell Trace Violet fluorescence was used to identify macrophages and phagocytosis was determined as the percentage of pHrodo green positive macrophages. The degree of phagocytosis was normalised by subtracting the percent pHrodo green positive macrophages from the control condition. Percent phagocytosis (%ADCP) fitted with a four-parameter model using Prism (GraphPad).

OCI-AML5 proliferation assay

Costar 96-well flat bottom plates (Thermo Fisher Scientific) were coated overnight at RT with 100 μL Poly-l-ornithine (Millipore), washed 3 times with PBS and dried. OCI-AML5 cells were seeded at 3000 cells/well in 100 μL assay media (RPMI 1640, 5% hFBS, 1 × Glutamax). 100 μL of FLT3L-Fc solutions were added to achieve final concentrations ranging from 10 μg/mL to 0.1 pg/mL in triplicates. Maximum and minimum control wells included 10 μg/mL FLT3L-Fc NG2LH or assay media alone. Assay plates were incubated at 37 °C, 5% CO₂ for 7 days. 100 μL of supernatant was removed from each well, and 100 μL CellTiter-Glo (Thermo Fisher Scientific) added. Plates were incubated for 10 min and luminescence was measured with a SpectraMax i3 plate reader (Molecular Device). Raw values were normalised to the maximum and minimum control wells. Dose response curves were plotted with XLfit (IDBS) and 50% effective concentration (EC₅₀) values were determined using a four-parameter logistic fit.

Human cDC1 differentiation assay

This assay was adapted from.⁴¹ CD34+ human cord blood stem cells (StemCell Technologies) were seeded at 5000 cells per well in 96-well U-bottom plates (Costar) in 100 μL expansion media (StemSpan media–StemCell Technologies, 10% FBS, 40 ng/mL IL-3, 200 ng/mL SCF, 100 ng/mL TPO–cytokines from Peprotech). 100 μL of FLT3L-Fc NG2LH was added to achieve final concentrations ranging from 100 nM to 0.01 pM. Assay plates were incubated at 37 °C, 5% CO₂ for 7 days. Next, 10 μL of the cell culture was transferred to a new 96-well U-bottom plate with 100 μL differentiation media (RPMI 1640, 10% FBS, 2 mM Glutamax, 10 mM HEPES, 5 nM 2-Mercaptoethanol, 1 mM sodium pyruvate, 100 U/mL Penicillin/Streptomycin, 40 ng/mL SCF, 5 ng/mL GM-CSF, 5 ng/mL IL-4). 100 μL of FLT3L-Fc NG2LH was added to achieve final concentrations ranging from 100 nM to 0.01 pM. After 6 days, half of the medium (100 μL) was removed from each well and replenished with fresh differentiation media + FLT3L-Fc NG2LH. Cultures were incubated for an additional 6 days. Cells were then harvested and stained with Human TruStain FcX (Biolegend) followed by anti-CD11c BV421 (clone B-ly6, BD Biosciences) and anti-CD141 PE (clone AD5-14H12, Miltenyi Biotec). Cells were washed and stained with Fixable Viability Dye eFluor780 (Thermo Fisher Scientific) and fixed in PBS + 1% PFA. Samples were acquired on an Attune NxT flow cytometer (Thermo Fisher Scientific) and data analysed with FlowJo (BD Biosciences). The frequency of CD11c^lowCD141^high cDC1 cells was calculated as a percentage of all viable, FSC/SSC gated cells. This population was previously demonstrated to resemble endogenous human cDC1.⁴¹

Mouse bone marrow proliferation assay

Femurs and tibias were collected from 6 to 10 week-old C57BL/6 mice (Charles River Lab). Bone marrow was flushed and single cell suspensions obtained after red blood cell lysis with ACK buffer and filtration through 70 μM cell strainers. Cells were resuspended at 3 × 10^ˆ5/mL in assay media (IMDM, 10% Hyclone FBS, 50 μM β-Mercaptoethanol, 1% Non-essential amino acids, 1 mM Sodium pyruvate, 1% Glutamax, 100 U/mL Penicillin/Streptomycin). 100 μL of cell suspension was plated in clear 96-well flat bottom plates with dark walls (Thermo Fisher Scientific) and 100 μL of FLT3L was added at indicated concentrations. On day 3, 50 μL of FLT3L at similar concentration was added to the culture. On day 7, 150 μL of supernatant was removed and 100 μL of CellTiter-Glo (Promega) added and mixed gently. After 15 min incubation at RT, luminescence was acquired on a SpectraMax i3 plate reader (Molecular Device).

Mouse studies and ethics

All animal studies were carried out in compliance with NIH guidelines for the care and use of laboratory animals and were approved by the Institutional Animal Care and Use Committee (IACUC) at Genentech, Inc. Female naive C57BL/6 mice (6–8 weeks old) were obtained from Charles River Laboratories Inc. (Hollister, CA). The mice were housed at Genentech in standard rodent micro-isolator cages and were acclimated to study conditions at least 3 days before tumour cell implantation. Mice were housed in individually ventilated cages within animal rooms maintained on a 14:10-h, light:dark cycle. Animal rooms were temperature and humidity-controlled, between 68–79 °F (20.0–26.1 °C) and 30–70% respectively, with 10–15 room air exchanges per hour. Female mice have been chosen to ensure results are comparable to data generated previously. Females are the most commonly used sex due to the fact that health issues related to aggression between cage mates (e.g., fight wounds) are minimal. As tumour immunity is shared between males and females, we believe the conclusions of this study should apply to both females and males. For PD studies, N = 4–5 mice were used in each experimental group. For tumour growth studies, N = 10 mice were used in each experimental group. These numbers were selected based on standard practices in the field for data accuracy and reproducibility. Mice were distributed to treatment groups according to a sequential algorithm which minimises the difference in standard deviation of tumour volumes across groups. Animals were maintained in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council 2011). Genentech is an AAALAC-accredited facility and all animal activities in this research study were conducted under protocols approved by the Genentech Institutional Animal Care and Use Committee (IACUC). For studies with tumour-bearing mice treatment information was not blinded during tumour measurement. Study data is collected using an in-house Excel-based tool that facilitates electronic collection of tumour volumes and/or body weights. Maximum total tumour volume before euthanasia is 2000 mm³, defined according to institutional policy concerning tumour endpoints in rodents. In addition, to prevent excessive pain or distress, the mice were euthanised following tumours ulceration or any signs of ill health. The ethical protocols regarding animal experiments have been reviewed and approved by the IACUC at Genentech with the numbers below: 22-1801K, 22-1801L (Fig. 3), 17-0591T (Fig. 3e), 17-0591V (Fig. 4a), 17-0591Y, 22-1801K, 22-1801L (Fig. 4b), 23-0871E (Fig. 5), 17-0591AM, 17-0591AY (Fig. 6), 20-0079 (Supplementary Figure S6), 19-1805 (Supplementary Figure S7), 17-0591A, 17-0591D (Supplementary Figure S8), 17-0591AB, 22-1801W (Supplementary Figure S9), 22-1801K, 22-1801L (Supplementary Figure S11).

Fig. 3.

Improved PK/PD properties of mFLT3L-Fc in C57BL/6 mice. (a) Experimental design to compare PK/PD properties of mFLT3L or mFLT3L-Fc dosed IV at 2 mg/kg. Timing of B16F10 tumour inoculation, mFLT3L or mFLT3L-Fc dosing, and sample collections are indicated. N = 5 per group were analysed in data presented in panels b–d. Data are representative of two independent experiments. (b) Average (SD) plasma concentration-time profiles following IV administration of mFLT3L (blue lines) or mFLT3L-Fc (red lines) to C57BL/6 mice. (c) Quantification of circulating cDC1, cDC2, pDCs, and monocytes 4 and 9 days post mFLT3L (blue bars) or mFLT3L-Fc (red bars). Median values and range are shown. (d) Quantification of intratumoral cDC1, cDC2, pDCs, and monocytes 9 days post mFLT3L (blue bars) or mFLT3L-Fc (red bars). Median values and range are shown. (e) Experimental design comparing dose range activity of mFLT3L-Fc dosed IV in naive C57BL/6 mice. N = 4 per group were analysed in data presented in panels f–g. (f) Average (SD) plasma concentration-time profiles following IV administration of mFLT3L-Fc. (g) Enumeration of cDC1s, cDC2s, pDCs, and monocytes in blood 4, 9, and 13 days post-mFLT3L-Fc treatment. Median values and range are shown. Statistical tests compare 2 vs 10 mg/kg and 10 vs 30 mg/kg groups at day 13.

Fig. 4.

mFLT3L-Fc combination therapy shows antitumour efficacy in B16F10 tumours. (a) Evaluation of antitumour efficacy of escalating dose of mFLT3L-Fc. Experimental design indicates timing of B16F10 tumours inoculation and mFLT3L-Fc, polyI:C and anti-PD-L1 treatments. B16F10 tumour growth since implantation is shown. Red dotted lines represent the spline of each experimental group. Grey dotted lines represent the spline of the control group (anti-gp120), added on each graphic for reference. N = 8–10 per group. (b) Evaluation of antitumour efficacy of 1 vs 2 cycles of mFLT3L-Fc, polyI:C and anti-PD-L1 combination therapy. Experimental design indicates timing of treatment and B16F10 tumours growth since implantation. Red dotted lines represent the spline of each experimental group. Grey dotted lines represent the spline of the control group (anti-PD-L1 + polyI:C), added on each graphic for reference. N = 10 per group in both (a, b). Time to progression (TTP) indicates the time required for tumours to reach 500 mm³. Data are representative of two independent experiments.

Fig. 5.

Activity of mFLT3L-Fc + polyI:C in B16F10 tumours and draining lymph nodes. (a) Experimental design indicating timing of B16F10 tumour inoculation, mFLT3L-Fc, polyI:C dosing, and samples collection. N = 5–7 per group. Data in panels b–n are representative of two independent experiments, median values and range are shown. Gating strategy used to identify immune populations is shown in Supplementary Figure S10. (b) IL6, IL1ɑ, TNFɑ, and CXCL10 concentration in plasma 6 h after polyI:C treatment. (c–n) Tumours and dLNs analysis 24 h after polyI:C treatment. (c) Quantification of cDC1 in tumours. (d) Frequency of MHC-II^HI, CD86^HI cDC1 in tumours, representative FACS plots are shown. (e) Quantification of cDC1 in dLNs. (f) Frequency of MHC-II^HI, CD86^HI cDC1 in dLNs, representative FACS plots are shown. (g) Quantification of cDC2 in tumours. (h) Frequency of MHC-II^HI, CD86^HI cDC2 in tumours, representative FACS plots are shown. (i) Quantification of cDC2 in dLNs. (j) Frequency of MHC-II^HI, CD86^HI cDC2 in dLNs, representative FACS plots are shown. (k) Quantification of monocytes in tumours. (l) Viperin expression in monocytes from tumours, representative FACS plots are shown. (m) Quantification of monocytes in dLNs. (n) Viperin expression in monocytes from dLNs, representative FACS plots are shown.

Fig. 6.

mFLT3L-Fc combination therapy reinvigorates T cell immunity. (a) Experimental design indicating timing of B16F10 tumour inoculation, mFLT3L-Fc, polyI:C and anti-PD-L1 treatment and tissue collection (tumours and dLNs). N = 7 per group were analysed in data presented in panels b–h. Median values and range are shown. Data are representative of two independent experiments. (b) B16F10 tumour weights 14 days post implantation. (c) Gating strategy to identify subpopulations of CD8+ T cells in the dLNs. Representative phenotypic profiles are presented in Supplementary Figure S12. (d) Quantification of CD8+ T cells subsets in dLNs. (e) Frequency of CD8+ T cells subsets in dLNs, percent of CD8+ T cells are shown. (f) Gating strategy to identify subpopulations of CD8+ T cells in tumours. Representative phenotypic profiles are presented in Supplementary Figure S12. (g) Quantification of CD8+ T cells subsets in tumours. (h) Frequency of CD8+ T cells subsets in tumours, percent of CD8+ T cells are shown.

Pharmacokinetic measurements of mFLT3L and mFLT3L-Fc in mouse studies

mFLT3L or mFLT3L-Fc was dosed via tail vein injection and blood samples for PK analysis were collected at indicated time points (Fig. 3a and e). Sample collection was done via retro-orbital bleed or cardiac puncture and processed to collect plasma. For PK data shown in Fig. 3b, concentrations of mFLT3L were measured with a mFLT3L ELISA (R&D, following manufacturer's instructions, Limit of detection = 0.00016 μg/mL) and concentrations of mFLT3L-Fc were measured with a mFLT3L_mFc ELISA (capture with anti-FLT3L and detection with anti-mIgG2a(aa), Limit of detection = 0.023 μg/mL). For PK data shown in Fig. 3f, concentrations of mFLT3L-Fc were measured with a mFLT3L ELISA (R&D). The plasma concentration vs time data was analysed by non-compartmental analysis to provide an estimation of PK parameters. The estimated parameters included the maximum concentration (C_max), half-life (HL), and area under the plasma concentration-time curve from time = 0 to the time of the last measurable concentration (AUClast) among others.

The study reported in Supplementary Table S3 is a sparse PK-sampling study, where individual data points were measured from different animals (i.e., not all the time points are available for all the animals). In contrast, the two studies reported in Supplementary Tables S4 and S5 are serial PK-sampling studies, where a full PK profile for each individual animal is available. WinNonlin Phoenix (version 8.2, Pharsight Corp., Mountain View, CA) was used to perform NCA analyses and report statistical results. For sparse PK-sampling studies, PK samples are naive-pooled to assume data is coming from one animal. In this case, exposure parameters (i.e., C_max and AUCall) are reported as average (SE). For sparse studies, Phoenix does not report SE for AUCinf or HL. For serial studies, PK data for each animal is available for each time point, yielding individual PK curves, and all parameters will be reported as average (SD) (N > 3 to calculate SD).

Mouse tumour monitoring and mFLT3L-Fc combination therapy

Female C57BL/6 mice (Charles River Lab) were inoculated by subcutaneous injection of 3 × 10^ˆ5 B16F10 mouse melanoma cells suspended in 100 μL of Hank's Balanced Salt Solution and matrigel into the right hind flank. Animals were distributed into treatment groups of 10 mice per group with similar distribution of tumour volumes. Mouse selection and treatment were initiated three days after inoculation before tumours were fully established. mFLT3L-Fc was administered at indicated doses intravenously (IV) through the tail vein. Anti-PD-L1 and control anti-gp120 (control antibody specific for the human immunodeficiency virus envelope glycoprotein gp120, generated internally), were administered at indicated times at 10 mg/kg (IV) for the first dose, followed by 5 mg/kg intraperitoneally (IP) for the remainder of the doses. PolyI:C high molecular weight (Invivogen) was administered IP at indicated time and doses at a concentration of 100 μg/100 μL. Tumour sizes and body weights were recorded twice weekly over the course of the study. Tumours were measured in two dimensions (length and width) using digital callipers, and the tumour volume (mm³) was calculated using the formula (length × width^ˆ2)/2. Mice whose tumour volume exceeded 2000 mm³ or whose body weight loss was 20% of their starting weight were euthanised per IACUC guidelines.

Analysis of immune cells in blood and tissues

Blood samples were collected via retro-orbital bleed or cardiac puncture. For isolation of circulating cells, blood was collected in 50 μL EDTA-coated capillaries. Red blood cells were lysed using ACK buffer and remaining cells resuspended in FACS buffer (PBS + 0.5% BSA + 0.05% Na Azide). For serum analysis, blood samples were centrifuged at 10 min at 10,000 g and serum collected and frozen at −80 °C until analysis. Spleens were mashed, filtered through 70 μm filters and red blood cells were lysed using ACK buffer. Remaining cells were resuspended in 1 mL FACS buffer. For FACS analysis, tumours were collected in cold PBS and minced with scissors in 2 mL of digestion buffer (PBS + Collagenase P 200 μg/mL–Roche 11249002001 + DNase I 100 μg/mL–Roche 10104159001). After 30 min incubation at 37 °C, tumour digests were filtered through 70 μm filters and resuspended in 500 μL FACS buffer. Draining lymph nodes were collected and mechanically dissociated in cold PBS, filtered through 70 μm, counted and resuspended in FACS buffer. For stainings, isolated cells were incubated with anti-mouse CD16/32 (Fc-block, clone 2.4G2, BD Biosciences) and stained using the antibody panels detailed in Supplementary Table S2. For intracellular staining following the surface staining, cells were fixed and permeabilized using the Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) following manufacturer's instructions. Cells were enumerated directly on the flow cytometer using CountBright Counting Beads (Thermo Fisher Scientific) and absolute numbers were calculated based on the fraction of the samples used in staining. Samples were acquired on a LSR-Fortessa (BD Biosciences) and data were analysed with FlowJo (BD Biosciences).

Luminex analysis

Serum cytokines were quantified using the Cytokine & Chemokine Convenience 26-Plex Mouse Panel (Thermo Fisher Scientific). Briefly, serum samples were thawed, centrifuged 10 min at 10,000 g, diluted 1/5 in assay buffer and analysed following manufacturer's instructions. Samples were acquired on a Luminex 200 (Thermo Fisher Scientific).

Patient recruitment and ethics statement

Details on the OAK trial, study plan and results have been published elsewhere.⁴² An independent data monitoring committee reviewed safety. Protocol approval was obtained from independent ethics committees for each site. This study is registered with ClinicalTrials.gov, number NCT02008227. All patients gave written informed consent.

Antibodies

Antibodies used in this study were commercially available and have been listed in the Supplementary Table S2 together with the catalogue reference numbers. Validation for commercially available antibodies can be found at manufacturers websites.

Cell lines

NK-92 CD-16-FcεRlγ_NFAT-Luc were generated internally from NK-92 (ATCC CRL-2407, RRID: CVCL_2142). SkBr3 Her2+ were obtained from ATCC (HTB-30, RRID: CVCL_0033). This line is solely used for investigational purposes. CHO-1 and CHO-2 cell lines were derived from CHO K1 (ATCC Cat # CCL-6, RRID: CVCL_1907). B16F10 were obtained from ATCC (CRL-6475, RRID:CVCL_0159). OCI-AML5 cells were obtained from DSMZ (ACC-247, RRID: CVCL_1620). SEM cells were obtained from DSMZ (ACC-546, RRID: CVCL_0095). Cell lines purchased externally are centralised in Genentech's cell bank (gCELL) to support the needs of cell-based research within Genentech. gCELL is tasked to bank verified, quality assured cell lines for distribution throughout the organisation. This provides a consistent source of cell lines for all levels of research to enable experimental reproducibility and access to baseline information such as morphology, growth conditions and in vitro properties. gCELL also provides a mechanism to ensure cell lines are used in accordance with all terms and conditions. All stocks are tested for mycoplasma prior to and after cells is cryopreserved. Two methods are used to avoid false positive/negative results: Lonza Mycoalert and Stratagene Mycosensor. All cell lines tested negative for mycoplasma.

Statistics

For graphics displaying multiple comparisons, a sequential approach was taken to compare statistically results from multiple groups in order to reduce the impact of multiple testing. Initially, all groups were compared using a Kruskal–Wallis test. Only if that test showed statistical significance at the 5% level, were pairs of groups compared using the Mann–Whitney test. Two-tailed tests were used for comparisons of pairs of groups using GraphPad Prism 9 (San Diego, CA, US). For graphics displaying standard deviation, no visible bars are indicative of no discernible differences between replicates. For tumour growth experiments, growth curves were fitted with 4 knots smoothing spline using GraphPad Prism 9 (San Diego, CA, US). For survival analyses, the log rank test was used to compare Kaplan–Meier survival curves. Cox proportional hazards regression models were used to generate hazard ratios and 95% confidence intervals as stratified by distribution based median DC score cut-points. Multivariable Cox proportional hazards regression models were used to investigate dependence between DC score for prediction of OS benefit. Analysis was done with R (R Core Team, 2014).

Role of funders

This work was performed at Genentech Inc. South San Francisco, CA 94080 by Genentech employees, including study design, data collection, data analyses, interpretation, or writing of reports. The study has been funded by Roche/Genentech laboratories.

Results

Design and molecular properties of FLT3L-Fc NG2LH

The molecular design of FLT3L-Fc NG2LH consists of three main components: an effectorless Fc domain, the growth factor domain of FLT3L and a native spacer region to connect these two domains. We first focused on engineering the Fc domain (Fig. 1a–d). In order to improve the PK properties of the molecule, we utilised the fragment crystallisable (Fc) domain of IgG1. The Fc domain of IgGs is responsible for the prolonged half-life of antibodies and Fc-fusion therapeutics, primarily due to its pH-dependent interaction with the neonatal Fc receptor (FcRn), which promotes recycling over degradation during pinocytosis by endothelial cells.⁴³ The additional molecular weight provided by the Fc can also reduce renal elimination. However, an Fc domain may mediate some effector functions such as ADCC and ADCP, through interactions with Fc gamma receptors on immune cells. In the context of a FLT3L-Fc therapeutic relying on a systemic expansion of phagocytes, the Fc effector function may be a liability, potentially leading to the elimination of the newly expanded cells and/or the release of proinflammatory cytokines. In order to remove the potential risks of residual effector function on an aglycosylated IgG1 Fc, a new effectorless Fc format was developed, termed NG2LH. NG2LH consists of the aglycosylation substitution, N297G (NG),⁴⁴ in combination with a graft of the lower hinge region of IgG2 (2LH) onto an IgG1 Fc (Fig. 1a).⁴⁵ Relative to other IgG formats, IgG1 NG2LH had equivalent or reduced binding to individual Fc-gamma receptors, but retained binding to FcRn (Fig. 1b). As an example, anti-Her2 antibodies in the NG2LH format had no detectable ADCC activity (Fig. 1c), nor ADCP activity (Fig. 1d), against a Her2+ cell line. However, all other formats tested, including IgG4 SP and N297G demonstrated moderate to strong ADCP, highlighting the improvement of the effectorless properties of the Fc NG2LH over other commonly used effector-attenuated Fc formats.

Fig. 1.

Design of FLT3L-Fc NG2LH. (a) Alignment of selected human IgG Fc hinge regions. IgG4 SP, IgG4 Fc variant containing the S228P∗ substitution for preventing Fab arm exchange. IgG1 NG, IgG1 Fc variant containing the N297G aglycosylation substitution. IgG1 2LH, IgG1 Fc variant substituted with the lower hinge residues of IgG2. IgG1 NG2LH, IgG1 Fc variant containing both the N297G and lower hinge substitutions. (b) Relative binding of Fc variants to Fc receptors. (c) ADCC with engineered NK reporter cells against SkBr3 Her2+ cell line using 4D5 anti-Her2 antibodies containing indicated Fc variants. (d) ADCP with primary macrophages against SkBr3 Her2+ cell line using 4D5 anti-Her2 antibodies containing indicated Fc variants. (e) Domain structure and boundaries of FLT3L. SP: signal peptide, TM: transmembrane domain, ICD: intracellular domain. (f) Diagram of FLT3L-Fc NG2LH and alignment of a portion of the FLT3L stalk region with the spacer regions of FLT3L-Fc NG2LH (S162) and FLT3L-Fc NG2LH (P167). Blue residues belong to FLT3L, green to Fc. (g) Non-reducing SDS-PAGE of FLT3L-Fc NG2LH (P167) and FLT3L-Fc NG2LH (S162). Red asterisk identifies the band corresponding to the mass of a monomer.

The next component of the therapeutic is the FLT3L growth factor domain. Starting from the N-terminus, the domain structure of endogenous FLT3L consists of: 1) a signal peptide at the N-terminus for directing the protein to the cell surface, which is cleaved from the mature cytokine during translocation; 2) a four-helix bundle domain (FHBD) which forms a non-covalent homodimer and directly interacts with FLT3, inducing receptor dimerisation⁴⁶; 3) an unstructured stalk region between the transmembrane domain and the FHBD; and 4) a short intracellular domain (Fig. 1e). In order to take advantage of the benefit of extended half-life provided by the Fc, while avoiding effector function activities, we recombinantly fused the FHBD of FLT3L to Fc NG2LH.

Finally, we chose an appropriate spacer length and composition for connecting the two domains (Fig. 1f). Early iterations of prototype FLT3L-Fc fusions with a short linker between the FHBD and the core hinge of the Fc, such as FLT3L-Fc (S162), appeared to lack proper formation of core hinge intermolecular disulfides in a significant fraction of the molecules. This was evidenced by the appearance of an additional lower molecular weight band in non-reducing SDS-PAGE, corresponding to the mass of a monomer (Fig. 1f and g), despite observation of the molecule behaving as a dimer in solution via analytical size-exclusion chromatography (Supplementary Figure S2). To extend the spacer region between the structured FHBD and the core hinge disulfides without introducing an unnatural, potentially immunogenic linker sequence, we utilised a portion of the native stalk region of FLT3L. One concern with utilising the full stalk is the presence of a tryptophan residue at position 169. Being in an unstructured region, this tryptophan would likely be solvent exposed and susceptible to oxidation.⁴⁷ We found that an extension of the FHBD to the proline at position 167 in the stalk was sufficient to restore core hinge intermolecular disulfides of FLT3L-Fc NG2LH, allowing for covalent dimer formation (Fig. 1f and g), thus improving the stability and homogeneity of the molecule.

In vitro characterisation of FLT3L-Fc NG2LH

Having generated the stable fusion protein FLT3L-Fc NG2LH, we then measured its binding properties and functional activity in vitro. Kinetic parameters (k_a and k_d) and K_D values for the binding interactions between FLT3L-Fc NG2LH and recombinant human FLT3-Fc protein were determined by surface plasmon resonance (SPR) and analysed following a monovalent (1:1) analyte binding model. The average K_D values for FLT3L-Fc NG2LH binding to recombinant human monomeric FLT3-monoFc and dimeric FLT3-Fc were 272 nM and 86.4 pM, respectively (Fig. 2c, representative sensorgrams shown in Fig. 2a and b). Monovalent binding of FLT3L-Fc NG2LH to recombinant human FLT3-monoFc was as expected since one FLT3-monoFc molecule bears only one binding site for the captured FLT3L-Fc NG2LH molecule. Although one dimeric FLT3-Fc molecule has two binding sites to the captured FLT3L-Fc NG2LH, no significant or systematic discrepancy was observed between the experimental data and the fitted curves using the 1:1 binding model, suggesting that the simplest 1:1 analyte binding model is sufficient to describe the binding interactions (Fig. 2b). It is likely that one dimeric FLT3-Fc molecule interacts with both binding sites on the captured FLT3L-Fc NG2LH in a bivalent fashion, which explains the much higher binding affinity of FLT3L-Fc NG2LH to dimeric FLT3-Fc in comparison with FLT3-monoFc. The binding affinity of FLT3L-Fc NG2LH to human FLT3 receptor on cells was assessed by measuring the binding of a fixed concentration of iodine 125 (125I) FLT3L-Fc NG2LH in the presence of increasing concentrations of unlabelled FLT3L-Fc NG2LH at equilibrium on OCI-AML5 cells, which express human FLT3 (FLT3L receptor).48, 49, 50 The average inhibition constant (K_i) value measured was 70.9 (SD 15.8) pM (Supplementary Figure S3), consistent with the affinity measured for FLT3L-Fc NG2LH to recombinant FLT3 (86.4 pM, Fig. 2c).

Fig. 2.

In vitro characterisation of FLT3L-Fc NG2LH activity. Representative sensorgrams of FLT3L-Fc NG2LH binding to recombinant human monomeric FLT3-monoFc (a) and dimeric FLT3-Fc (b). Black solid lines are curves fit to the data using a 1:1 binding model, and coloured lines are raw data. The tested recombinant human FLT3-monoFc concentrations (from bottom to top) are 62.5, 125, 250, 500, 1000, and 2000 nM; recombinant human dimeric FLT3-Fc concentrations (from bottom to top) are 3.75, 7.5, 15, 30, 60, and 120 nM. The sensorgrams were generated after in-line reference cell correction followed by buffer sample subtraction. The experiments were conducted at 37 °C. (c) Binding affinities of recombinant human monomeric FLT3-monoFc and dimeric FLT3-Fc proteins to captured FLT3L-Fc NG2LH. k_a association rate constant; k_d dissociation rate constant; K_D equilibrium dissociation constant. Data for FLT3L-Fc NG2LH binding to recombinant human FLT3-monoFc were averaged from three independent experimental Biacore runs. Data for FLT3L-Fc NG2LH binding to recombinant human dimeric FLT3-Fc were averaged from two independent experimental Biacore runs. (d) ADCP for FLT3L-Fc NG2LH and its variants. Data shown are a representative set taken from one of three independent experiments. Average (SD) of duplicate values are shown. (e) Proliferation of OCI-AML5 cells in vitro. A representative dose response curve is shown, Average (SD) of triplicate values. Data were normalised to 10 μg/mL FLT3L-Fc NG2LH as the maximum (100%) response and the assay media alone control as the minimum (0%) response. (f) Human cDC1 differentiation from cord blood stem cells. Frequency of cDC1 at the end of culture is shown (percent of live cells). Results were obtained from 2 donors. (g) In vitro proliferation of mouse bone marrow cells in response to mFLT3L or mFLT3L-Fc. Average (SD) of duplicate values are shown.

The binding activities of FLT3L-Fc NG2LH to six different Fcγ receptors was examined in ELISA based ligand binding assays. In line with the results obtained with IgG1 NG2LH (Fig. 1b), FLT3L-Fc NG2LH showed no binding to human FcγRIA, two allotypes of FcγRIIA (H131 and R131), FcγRIIB, or two allotypes of FcγRIIIA (F158 and V158) (Supplementary Figure S4). We next evaluated the ability of FLT3L-Fc NG2LH to induce FcγR mediated ADCP activity (Fig. 2d). While dose dependent ADCP activity was observed with FLT3L fused to wild-type human IgG1 Fc (FLT3L-Fc IgG1), FLT3L fused to Fc containing the N297G effector attenuating mutation (FLT3L-Fc NG) showed reduced ADCP, and no activity was detected with FLT3L-Fc NG2LH.

The human OCI-AML5 cell line has been reported to proliferate in response to stimulation with human FLT3L.48, 49, 50 FLT3L-Fc NG2LH and wild-type human FLT3L induced similar dose dependent proliferation of OCI-AML5 cells in vitro with EC₅₀ potency values of 2.5 (SD 0.5) pM and 2.3 (SD 0.5) pM for FLT3L-Fc NG2LH and FLT3L, respectively (Fig. 2e). Furthermore, FLT3L-Fc NG2LH induced the differentiation of CD11c^low CD141^high cDC1 from primary CD34+ cord blood progenitors in vitro (Fig. 2f). Both human FLT3L-Fc NG2LH and its mouse effectorless surrogate FLT3L-Fc (mFLT3L-Fc) molecule induced dose dependent proliferation of bone marrow cells in vitro, with comparable potencies (EC50 ∼0.4 nM, Fig. 2g), validating mFLT3L-Fc as a good surrogate molecule for further in vivo explorations.

In vivo PK/PD properties of FLT3L-Fc fusion

Based on these biochemical and biological properties, we next compared the in vivo PK/PD properties of the mouse FLT3L (mFLT3L) and mFLT3L-Fc molecules evaluated at 2 mg/kg (Fig. 3a, Supplementary Table S3). Immune populations were identified as described in Supplementary Figure S5a. The PK data confirmed that mFLT3L-Fc showed a significantly higher exposure compared to mFLT3L as observed in Fig. 3b and estimated by AUC and half-life values reported in Supplementary Table S3. While we observed a significant expansion of blood cDC1, cDC2, and pDCs with both mFLT3L and mFLT3L-Fc 4 days post treatment, this expansion returned to baseline by day 9 with mFLT3L but not with mFLT3L-Fc (Fig. 3c), in concordance with our PK data. A parallel phenotype was observed in tumours (Fig. 3d), where intratumoral cDC1, cDC2, and pDC expansion was only observed with mFLT3L-Fc 9 days post treatment. A significant expansion was also observed for circulating monocytes, but no expansion of monocytes, macrophages or neutrophils was detected in tumours by day 9 with any of the molecules (Fig. 3d, Supplementary Figure S5c). Finally, mFLT3L-Fc had no impact on overall T cells and B cells numbers in both blood and tumour (Supplementary Figure S5b and c). Altogether, these results establish that the increased exposure observed with mFLT3L-Fc translates into increased PD, and overall improved activity over mFLT3L.

We then conducted a dose ranging study with mFLT3L-Fc. Following a single IV dose (Fig. 3e), mFLT3L-Fc demonstrated a dose-proportional increase in C_max and a greater than dose-proportional increase in AUC_last (Fig. 3f, Supplementary Table S4). mFLT3L-Fc binds to FLT3 and expands the total pool of its target in mice. Therefore, this greater than dose-proportional increase in AUC indicated target mediated drug disposition (TMDD), the expanded cells acting as a sink, removing the drug from the circulation. Non-specific clearance and half-life at 30 mg/kg were estimated to be 16.6 (SD 1.69) mL/day/kg and 1.77 (SD 0.654) days, respectively. Dose-dependent increases of cDC1s, cDC2s, pDCs, and myeloid cell numbers in blood and spleen were observed. While cDC1s, cDC2s, and pDCs expansion increased between 2 and 10 mg/kg doses 13 days post mFLT3L-Fc treatment, it plateaued between 10 and 30 mg/kg (Fig. 3g, Supplementary Figure S5e). A moderate dose dependant expansion of other myeloid cells was observed in blood and spleen, while T and B cell numbers did not increase in circulation and were moderately altered in the spleen (Supplementary Figure S5d and e). The observed correlation between PK profiles and dose-dependent expansion of cDCs supports on-target pharmacologic activity of mFLT3L-Fc. In naïve and tumour-bearing mice, FLT3L-Fc showed comparable PK including comparable C_max and AUC at the same dose level (Supplementary Figure S6b, Supplementary Table S5). And similar cDC1 and cDC2 expansion was observed in blood and spleen (Supplementary Figure S6c and d). These results suggest comparable PK/PD properties in naïve and tumour-bearing mice.

We then explored how the glycosylation profile of our therapeutic affects its PK/PD properties, a particularly important consideration for the scale up in manufacturing (Supplementary Figure S7). The glycan composition outside of the Fc domain of Fc-fusion proteins has been previously reported to influence PK, likely through binding and uptake via lectins such as asialoglycoprotein receptors expressed in the liver and mannose receptors expressed on immune cells.⁵¹ In order to evaluate the potential impact of different glycosylation patterns of the therapeutic on its PK/PD profile, FLT3L-Fc was generated in CHO cells under different production conditions (CHO-1 and CHO-2), which resulted in different glycoprofiles (Supplementary Figure S7a). Additionally, a mutant version, FLT3L-Fc N149Q, in which one of the N-linked consensus sites was removed, was also generated under CHO-2 conditions. Despite significant differences in mannose content and terminal sialylation between FLT3L-Fc produced in CHO-1 and CHO-2, very little difference was observed with respect to PK in mice (Supplementary Figure S7b). Furthermore, FLT3L-Fc N149Q CHO-2, which has half the number of N-linked glycosylation sites as FLT3L-Fc, also showed similar PK/PD profiles in mice (Supplementary Figure S7b and c), highlighting a negligible contribution of the FLT3L glycosylation profile to the behaviour of the molecule in vivo.

mFLT3L-Fc combination therapy shows antitumour activity in a tumour model refractory to immunotherapy

Our PK/PD experiments established a dose-dependent expansion of cDCs upon mFLT3L-Fc treatment, and we next explored how the magnitude of cDC expansion would correlate with tumour control. Mice bearing checkpoint blockade resistant B16F10 tumours were treated with a mFLT3L-Fc + polyI:C (a synthetic double-stranded RNA analogue and TLR3/RIG-I/MDA5 agonist) + anti-PD-L1 combination regimen (Fig. 4a). While mFLT3L-Fc alone or in combination with anti-PD-L1 showed limited efficacy on B16F10 tumour growth, the antitumour activity of polyI:C + anti-PD-L1 was increased by mFLT3L-Fc combination in a dose-dependent manner (Fig. 4a). To note, anti-PD-L1 also increased the efficacy of FLT3L + polyI:C combination (Supplementary Figure S8). mFLT3L-Fc combination therapy was also tested in a more inflamed Hepa1-6 tumour model (Supplementary Figure S9a), and while polyI:C + anti-PD-L1 was effective, we observed a trend of improved efficacy in combination with mFLT3L-Fc (Supplementary Figure S9b). These data establish the potential of using a combination of mFLT3L-Fc with an innate immune adjuvant to potentiate the efficacy of ICB therapies. Finally, as mFLT3L-Fc induced cDC expansion starts to contract 10–14 days after treatment (Fig. 3g), we explored if the efficacy of this triple combo therapy could be increased by repeating the treatment cycle (Fig. 4b). An additional round of mFLT3L-Fc combination therapy was able to further delay the growth of B16F10 tumours (Fig. 4b), suggesting that sustained cDC expansion and activation can further improve the benefit of mFLT3L-Fc based combination therapies and further increase ICB efficacy. To note, following the third injection of high-dose poly-IC, signs of toxicity were observed in the group receiving repeated doses of mFLT3L-Fc. This underscores the importance of further refining the dosing strategy of the innate adjuvant in the combination setting.

PolyI:C leads to systemic cDCs and myeloid cells activation, and cDCs accumulation in draining lymph nodes

To investigate the contribution of polyI:C to mFLT3L-Fc combination therapy, we monitored the activation of cDCs and myeloid cells in B16F10 tumours and draining lymph nodes (dLNs) of mice treated with mFLT3L-Fc ± polyI:C (Fig. 5a) as well as plasma cytokines (Fig. 5b). While mFLT3L-Fc did not lead to changes in plasma cytokines, the addition of polyI:C increased plasma concentrations of the inflammatory cytokines IL6, IL1ɑ, TNFɑ, and CXCL10. An expansion of cDC1 and cDC2 in both the tumour and the dLNs was noted with mFLT3L-Fc treatment alone. However, the addition of polyI:C (Fig. 5c and g) induced a contraction of intra-tumoral cDCs associated to an activated MHC-II^HI CD86^HI cDC phenotype (Fig. 5d and h). Concomitantly, one injection of polyI:C (Fig. 5e and i) induced an increase of cDCs in the dLNs associated with an activated MHC-II^HI CD86^HI DC phenotype (Fig. 5f and j). This apparent decrease of cDCs in tumours and increase in the dLNs data suggest that polyI:C may have prompted the migration of mature MHC-II^HI CD86^HI cDCs from the tumour to the dLNs, where they may accumulate. Intratumoral macrophages also showed reduced numbers in mice treated with mFLT3L-Fc + polyI:C (Fig. 5k), which was associated with active IFNAR signalling as supported by the increased expression of Viperin (Rsad2, a type I IFN stimulated gene). Monocytes were expanded in the dLNs upon mFLT3L-Fc treatment and their number remained unchanged in presence of polyI:C (Fig. 5m), while displaying increased Viperin expression (Fig. 5n). A moderate effect of polyI:C on the abundance of other immune populations was observed in both B16F10 tumours and dLNs (Supplementary Figure S10). To note, the analysis performed after the second dose of polyI:C (Supplementary Figure S11) showed a similar decrease in cDCs number in tumours of mice treated with mFLT3L-Fc + polyI:C, and concomitant broad myeloid cells activation, but did not show a significant increased number of cDCs in the dLNs. It is plausible that polyI:C induces a widespread activation of myeloid cells, and a release of inflammatory cytokines driving activation-induced cell death at a later timepoint. This hypothesis gains support from the observed upregulation of Viperin in various myeloid cells of mice treated with polyI:C (Supplementary Figure S11), along with increased plasma concentrations of IL6, TNFɑ and IL1ɑ (Fig. 5b). Therefore, the reduction in the number of cDCs and myeloid cells in the tumour, coupled with their moderate accumulation in the dLNs at the latest time point is likely explained by a combination of cDCs migration from the tumour to the dLNs and activation-induced cell death. However, determining the relative contributions of each of these factors remains challenging. Moreover, the later time point analysed, 9 days post mFLT3L-Fc, at peak of mFLT3L-Fc activity (Fig. 3g), leads to a large mobilisation of newly differentiated cDCs from the bone marrow to the dLNs, which may further mask the differential due to cDCs migrating from the tumours. Overall, these findings highlight the systemic effects of polyI:C in combination with mFLT3L-Fc, on cytokine release and myeloid cells activation, and the specific ability of mature cDCs to accumulate in the dLNs upon activation in this therapeutic setting.

mFLT3L-Fc combination therapy reinvigorates CD8+ T cell immunity

We next determined how the antitumour efficacy of the mFLT3L-Fc combination therapy modulates CD8+ T cell responses (Fig. 6a) by monitoring CD8+ T cell phenotypes and numbers in tumours and dLNs 14 days after tumour implantation (7 days after onset of anti-PD-L1 treatment). Confirming observations made from tumour growth experiments (Fig. 4a), tumour weights were decreased in mice treated with the mFLT3L-Fc + polyI:C + anti-PD-L1 regimen (Fig. 6b). We observed populations of PD1+ and PD1− CD8+ T cells in the dLNs (Fig. 6c, Supplementary Figure S12). Within the PD1+ CD8+ T cells, we detected sub-populations of TCF1+ GzmB− cells and TCF1− GzmB+ cells (Fig. 6c). PD1+ TCF1+ GzmB− CD8+ T cells have been described as proliferative precursors of exhausted T cells (T_PEX) and are associated with response to immune checkpoint blockade (ICB), while PD1+ TCF1− GzmB+ CD8+ T cells are considered to be effector/exhausted T cells (T_EFF).^52,53 While we observed a moderate expansion of T_PEX and T_EFF in dLNs upon mFLT3L-Fc treatment (Fig. 6d), this was further increased in presence of anti-PD-L1, supporting a specific activity of anti-PD-L1 in favouring the priming of CD8+ effector T cells. Addition of polyI:C to the mFLT3L-Fc + anti-PD-L1 combo further increased the number of T_PEX, and comparable levels of T_EFF in the dLNs. In contrast, PD1− CD8+ T cells numbers showed limited changes across treatment groups (Fig. 6d). Monitoring the relative frequency of CD8+ T cells subsets showed a decreased in PD1− cells and a respective increase in PD1+ T cells upon mFLT3L-Fc + anti-PD-L1 or mFLT3L-Fc + anti-PD-L1 + polyI:C treatment (Fig. 6e), supporting that mFLT3L-Fc combination therapies stimulates CD8+ T cell immunity.

In addition, we monitored subsets of PD1+ and PD1− CD8+ T cells in the tumours (Fig. 6f, Supplementary Figure S12). Within the PD1+ CD8+ T cells, we detected sub-populations of exhausted PD1+ TCF1− GzmB− (T_EX) and PD1+ TCF1− GzmB+ (T_EFF) cells. In contrast to the dLNs, TCF1 expression in the tumour was low in all CD8+ T cell subsets and remained unchanged across treatment groups (data not shown). In line with the observations made in the dLNs, mFLT3L-Fc alone had a moderate effect on CD8+ T cells subsets numbers in tumours (Fig. 6g), but combination with anti-PD-L1 increased the number of intratumoral T_EX and T_EFF subsets. Interestingly, these numbers decreased in presence of polyI:C, in line with the overall drop in immune cell content observed with mFLT3L-Fc + polyI:C combination (Fig. 5c, g, and k). CD8+ T cells subsets frequencies showed a significant shift from a high prevalence of PD1− CD8+ T cells, to a high prevalence of T_EFF cells in mice treated with mFLT3L-Fc + anti-PD-L1 ± polyI:C (Fig. 6h), while the proportion of T_EX remained unchanged.

Overall, these data support a mode of action of mFLT3L-Fc combination therapies that expands intratumoral cDC1 and favours the activation of CD8+ T cells in dLNs. Particularly, this combination regimen expands the pool of T_PEX (PD1+ TCF1+) in dLNs and increases the abundance of T_EFF (PD1+ TCF1− GzmB+), which migrate to the tumour, mediate antitumour activity and further differentiate into T_EX (PD1+ TCF1− GzmB−).

Discussion

In the present study, we describe the development of FLT3L-Fc NG2LH, an effectorless recombinant human FLT3L, and validate this approach with a mouse surrogate molecule. The development of FLT3L-Fc NG2LH offers the potential to significantly expand the pool of intratumoral cDCs after a single injection, obviating the need for daily dosing required with FLT3L.

To prevent the risk of depleting FLT3-expressing cells via Fc-mediated effector function of the molecule and inflammatory response, the NG2LH format was designed to be highly attenuated, especially with respect to ADCP, in comparison to common “effector-reduced” Fc isotypes and formats such as IgG4 and aglycosylated IgG1. Additionally, because the target cells for FLT3L are professional antigen presenting cells, the potential for immunogenicity of the therapeutic was a strong consideration for retaining as much native sequence as possible in the FHBD and Fc domain, as well as the spacer region between them. Therefore, FLT3L-Fc NG2LH retains a fully WT FHBD and the spacer region utilises a minimal portion of the native FLT3L stalk. With respect to the Fc, the IgG2 lower hinge graft boosts the attenuation of effector function, while maximising native epitopes relative to the natural human IgG isotype repertoire. In vitro experiments established that FLT3L-Fc NG2LH potency is comparable to recombinant FLT3L, but shows improved PK/PD properties in vivo. Preclinical experiments showed that a single injection of mFLT3L-Fc leads to a sustained expansion of cDCs in lymphoid tissues and tumours. Moreover mFLT3L-Fc promotes antitumour CD8+ T cell immunity and increases tumour control when combined with polyI:C + anti-PD-L1 in the immune cold B16F10 tumour model. These observations pave the way for using FLT3L-Fc NG2LH as a therapeutic agent in patients with cancer in order to leverage cDCs' immune potential in situ and maximise response to ICB.

DCs play opposite roles in cancer, by mounting antigen specific T cell responses against tumour antigens, and by maintaining peripheral tolerance to self-antigens preventing autoimmunity.^54,55 Tolerogenic vs immunogenic function of DCs is controlled by their maturation status as, among other mechanisms, immature DCs lack the expression of costimulatory molecules required for T cell activation, favouring T cell anergy or deletion.⁵⁶ Thus, expanding cDCs with FLT3L-Fc NG2LH for antitumour therapy needs to take into account their maturation status.

In line with previous observations,^27,29 mFLT3L-Fc shows limited antitumour efficacy as a single agent in our preclinical experiments, despite robustly expanding the pool of intratumoral cDCs. Shifting cDCs from tolerogenic to immunogenic activity relies on the integration of multiple signals such as damage/pathogen associated molecular patterns, as well as cytokines and cellular interactions.⁵⁷ In our study, we used the synthetic double-stranded RNA analogue and TLR3/RIG-I/MDA5 agonist, polyI:C high molecular weight to trigger the maturation of mFLT3L-Fc expanded cDCs.⁵⁸ Addition of polyI:C to mFLT3L-Fc + anti-PD-L1 was required to achieve strongest antitumour efficacy, highlighting the therapeutic benefit of including innate immune adjuvant in FLT3L based therapies. Interestingly, the homoeostatic maturation leading to mature cDCs enriched in immunoregulatory molecules (mregDCs), which have tolerogenic activity,⁵⁹ can be overridden by TLR3 engagement to promote immunogenic properties of mature cDC1.⁶⁰ Other studies have successfully leveraged alternate pathways to achieve DCs maturation, such as STING agonists, oncolytic viruses or anti-CD40 agonist antibodies.^29,31,61 It is worth mentioning that in addition to their cytotoxic effect, chemo and radiotherapies can have adjuvant activity on cDCs,^62,63 which could be of interest for building FLT3L combinations together with approved therapies.

While polyI:C was used here to mature cDC1, which can promote ongoing T cell responses in the TME,⁶⁴ our data show that polyI:C also leads to a broad activation of the intratumoral myeloid compartment, which is associated with a decrease in myeloid cells cellularity in tumours but not dLNs. Interestingly, our results suggest that upon activation, mFLT3L-Fc expanded cDCs can migrate from the tumour to the dLN where they are able to accumulate, a phenomenon observed for both cDC1 and cDC2. While TLR3 is restricted to cDC1 in the TME, RIG-I and MDA5 expression are detected in cDC2, macrophages and monocytes (data not shown). Thus, polyI:C may trigger a broad myeloid cells driven inflammatory response in the TME, which could favour overall cDCs activation and migration, but may also trigger activation related cell death.⁶⁵ While our preclinical experiments show a selective intratumoral cDC expansion by mFLT3L-Fc, the systemic expansion of myeloid cells observed in our PD studies suggests that FLT3L treatment could potentially affect intratumoral myeloid cells beyond cDCs. FLT3L mainly acts in the bone marrow by favouring differentiation of cDCs, which then populate peripheral tissues.^27,66 As FLT3 is expressed on HSCs, FLT3L also expands other myeloid progenitors that give rise to monocyte and macrophage populations.⁶⁶ Systemic myeloid cell expansion has been observed in healthy individuals treated with FLT3L,²⁵ although the magnitude of this expansion is lower than for cDCs. This aligns with our own results and confirms a preferential activity of FLT3L on the cDC differentiation axis in the bone marrow. That said, considering the diverse phenotypes and functions of myeloid cells in cancer,^15,67 a careful characterisation of myeloid cell populations in tumours upon FLT3L treatment will be of interest. Beside their modulation in the TME, systemic expansion of monocytic populations by FLT3L-Fc could also potentiate a pathologic inflammatory response. Indeed, we observed increased systemic levels of pro-inflammatory cytokines (TNFɑ, IL6, IL1ɑ) as well as monocyte and macrophage activation in mFLT3L-Fc + polyI:C treated mice. These results can be put in perspective with the repeated cycles efficacy experiments (Fig. 4b), in which sustained myeloid cells expansion by mFLT3L-Fc and combination with high doses of polyI:C led to toxicity, which may potentially be caused by exacerbated inflammation due to activation of myeloid cells.^68,69 However, causality was not investigated as this study was designed to investigate efficacy and not safety. Thus, our experiments show the potential in unleashing antitumour immunity by combining mFLT3L-Fc with polyI:C, and highlight the need to closely monitor the dosing of the innate adjuvant. Importantly, clinical studies combining FLT3L and polyI:C injected i.t or s.c demonstrated manageable flu-like symptoms in patients, associated with promising signs of efficacy and/or immune activation.^28,70

In line with other FLT3L preclinical studies,^27,28,31,71 mFLT3L-Fc + polyI:C unlocks tumour responsiveness to ICB, in our case anti-PD-L1. Additionally, the magnitude of cDC expansion could be further extended using multiple rounds of treatments, as suggested by the increased efficacy observed when two cycles of mFLT3L-Fc + polyI:C were administered. Recent studies aiming to understand anti-PD-1/PD-L1 mechanism of action have unveiled the essential role of T_PEX in response to therapy,^52,53 and the conceptual vision behind these therapeutic modalities has shifted from reversing exhaustion in the tumours to promoting T cell immunity in the lymph nodes.⁷² Our experiments demonstrate that mFLT3L-Fc is able to leverage this process by expanding the pool of T_PEX in the dLNs, which is associated with increased frequency of effector T cells in the tumours and decreased tumour mass. This is in line with observations showing constant egress of TCF1+ CD8+ T cells from dLNs to tumours, where they rapidly develop an exhausted phenotype.⁷³ Interestingly, our T cell analysis showed that the combination of mFLT3L-Fc + anti-PD-L1 is effective at promoting CD8+ T cell immunity, but translates in limited antitumour efficacy in the B16F10 tumour model, which could be improved by polyI:C combination. Polarisation of tumour associated monocytes and macrophages towards a type I IFN response have been associated with antitumour efficacy of multiple treatment. In addition to promote cDCs maturation, polyI:C may play an essential role in remodelling the TME in the context of FLT3L-Fc combination therapy to favour T cell activity.74, 75, 76

Confirming previous observations,²⁰ bulk RNAseq analysis of tumour biopsies obtained from a cohort of patients with NSCLC treated with anti-PD-L1 suggests that the presence of cDCs in tumours is associated with increased response to anti-PD-L1 therapy, while showing no statistical benefit to Docetaxel treatment. Several studies have recently shown that PD-L1 expression on cDCs is crucial in suppressing CD8+ T cells activity.^20,77,78 One could speculate that anti-PD-L1 therapy may be able to modulate PD-L1 expressing mregDCs favouring effector CD8+ T cells function in tumours and therefore increasing clinical benefit.

Overall, FLT3L-Fc NG2LH may facilitate the development of combination therapies leveraging cDCs in situ in order to improve response to ICB and immunotherapies in general.⁷⁰ Many questions remain to be explored related to the impact of FLT3L on cDCs and myeloid cells in tumours and lymphoid tissues. Particularly, our study does not decipher the precise contribution of innate immune adjuvants and ICB therapies (in our case polyI:C and anti-PD-L1) on the different immune compartments. Answering these questions will be essential to optimise FLT3L-Fc combination regimens and maximise therapeutic success.

Contributors

JD, ET, DH, RK, AMB, AGG, DW, MK, YK, YS, SM, XW, AE, AA, YY, HK, NT, DB, EM, MM, MB, GKR, GS, WP, PD, WS, AMO, PC, LC, SC, IH, CCK, JMS, and TWB designed, performed, analysed and/or interpreted experiments and reviewed the underlying data. PD and SS performed clinical data analysis. JD wrote the manuscript with the contribution of DH, XW, AE, PD, SS, IH, CCK, JMS, TWB, and CM. JAE, GAL, YL, TWB, and CM conceptualised and led the study. JD and CM have accessed and verified the underlying data. In addition, all the data have been screened and cleared by an external mandated expert using RESIS (for more information: https://www.resis-srl.com) as required by Genentech's internal data review policy. CM and TWB were responsible for the decision to submit the manuscript. All authors read and approved the final article.

Data sharing statement

Further information and requests for resources should be directed to Christine Moussion, moussion.christine@gene.com. Reagents and data collected for the study will be available upon reasonable request.

Declaration of interests

Jérémie Decalf, Evangeline Toy, Dongping He, Radhika Kenkre, Amy M Berkley, Alan G Gutierrez, Devon Wong, Mandy Kwong, Yee-Seir Kee, Yue Sun, Srividya Myneni, Xiangdan Wang, Ahmad Ebtikar, Anthony Ancheta, Yanli Yang, Hok Seon Kim, Nga Tang, Debarko Banerji, Elaine Mai, Pranay Dogra, Gautham K Rao, Geraldine Strasser, Wilson Phung, Peter Day, Wendy Sandoval, Ayse Meric Ovacik, Pamela Chan, Shomyseh Sanjabi, Laetitia Comps-Agrar, Sivan Cohen, James A Ernst, Greg A Lazar, Christopher C Kemball, Iraj Hosseini, Yichin Liu, Jill M Schartner, Travis W Bainbridge, Christine Moussion are or were Genentech Inc. employees when this work was performed. Meredith McLerie and Matt Betzenhauser are Curia Global Inc. employees.

Appendix A. Supplementary data

Supplementary Figure S1.

Higher DC scores predict favourable responses in patients with lung cancer. Kaplan–Meier curves of overall survival (OS) for Atezolizumab (a) and Docetaxel (b) treated patients with non-squamous NSCLC histology in the OAK trial (NCT02008227) dichotomised as DC high and DC low. Numbers at risk are reported in corresponding tables. DC signature included XCR1, BATF3, IRF8, and FLT3 genes.20

Supplementary Figure S2.

Analytical size exclusion chromatograms of FLT3L-Fc NG2LH (“P167”) and FLT3L-Fc NG2LH with a shorter spacer (“S162”).

Supplementary Figure S3.

Binding affinity of FLT3L-Fc NG2LH to human FLT3 receptor. Competitive homologous binding curve of radiolabelled Iodine-125 FLT3L-Fc NG2LH to OCI-AML5 cells. A representative binding curve from two independent experiments is shown. Average (SD) of triplicate values are shown. FLT3L-Fc NG2LH was indirectly radiolabelled using the Iodogen method⁷⁹ to a specific activity of 1.32 μCi/μg. OCI-AML5 cells were fixed with a solution of 4% paraformaldehyde (PFA)/4% sucrose for 5 min at room temperature (RT) and seeded in 96-well plates in cold binding buffer (Opti-MEM + 2% FBS + 50 mM HEPES, pH 7.2 + 0.1% Na Azide) at 654,000 cells per well. A fixed concentration of radiolabelled FLT3L-Fc NG2LH was mixed with 1:3 serially diluted non-labelled FLT3L-Fc NG2LH starting at 150 nM. The FLT3L-Fc NG2LH mixture was added to the cells and incubated at RT for 12 h under gentle agitation. Cells and antibodies were then transferred to Millipore multiscreen filter plates (Billerica), washed 4 times with 250 mL cold binding buffer and dried for 30 min and the filters were punched into 5 mL polystyrene tubes. The radioactivity was measured using a Wallac WIZARD 2470 Gamma Counter (PerkinElmer Life and Analytical Sciences) set at 1 count per minute with 0.8 counting efficiency. The data were fitted using the heterologous one site-fit Ki competitive binding model in Prism (GraphPad).

Supplementary Figure S4.

Binding of FLT3L-Fc NG2LH to human Fcγ receptors. The binding activities of FLT3L-Fc NG2LH (black circles) and FLT3L-Fc IgG1 (open squares) to FcγRIA, two allotypes of FcγRIIA [H131 and R131], FcγRIIB, and two allotypes of FcγRIIIA [F158 and V158] were measured in ELISA based ligand binding assays. Samples were tested, in duplicate, and a total of three independent experiments were performed for each FcγR binding assay. Binding curves from representative experiments are presented.

Supplementary Figure S5.

Extended data fromFig. 3. See Fig. 3a for experimental design. For data presented in panels b–c, N = 5 per group; median values and range are shown. Data are representative of two independent experiments. (a) Gating strategy used to identify immune populations across tissues. (b) Quantification of circulating Ly6C− monocytes, neutrophils, T and B cells 4 and 9 days post mFLT3L (blue bars) or mFLT3L-Fc (red bars). (c) Quantification of intratumoral macrophages, neutrophils, T and B cells 9 days post mFLT3L (blue) or mFLT3L-Fc (red). See Fig. 3e for experimental design of data presented in panels d–e. N = 4 per group; median values and range are shown. (d) Enumeration of Ly6C− monocytes, neutrophils, T and B cells in blood at 4, 9, and 13 days post-mFLT3L-Fc treatment. Statistical tests compare 2 vs 10 mg/kg, and 10 vs 30 mg/kg groups at day 13. (e) Enumeration of immune population in spleen 13 days post-mFLT3L-Fc dose range treatment.

Supplementary Figure S6.

Comparison of FLT3L-Fc NG2LH PK/PD properties in naive or B16F10 tumour bearing mice. (a) Female C57BL-6 mice (6–8 weeks) were distributed into two groups designated as either naive or tumour-bearing (N = 5 mice per group). Mice were inoculated with 0.3 million B16F10 cells in HBSS + Matrigel injected subcutaneously into the lateral flank with a maximum injection volume of 100 μL. Three days post inoculation, both naïve and tumour-bearing mice were administered 2 mg/kg FLT3L-Fc via IV injection. Plasma samples were collected at 30 min, 6 h, 1, 3, 7, and 10 days post-dose and FLT3L-Fc concentrations were measured using human FLT3L/huFc ELISA with a minimum quantifiable concentration of 0.00141 μg/mL. (b) Average (SD) plasma concentration-time profiles following IV administration of FLT3L-Fc to naive and tumour-bearing mice. (c) cDC1 and cDC2 enumeration in blood at pre- (day −5) and post- (days 3 & 10) FLT3L-Fc treatment in naive and tumour-bearing mice. (d) cDC1 and cDC2 enumeration in spleen 10 days post-FLT3L-Fc treatment in naive and tumour-bearing mice. Median and range are shown. For this experiment, cDC1s were identified as IRF8+ cells.

Supplementary Figure S7.

Negligible contribution of the FLT3L glycosylation profile to the behaviour of the molecule in vivo. (a) Global N-linked glycan composition was assessed using LC-MS analysis. 10 μg of protein were denatured with 8 M guanidine HCl at 1:1 volume ratio and reduced with 100 mM dithiothreitol for 10 min at 95 °C. Samples were diluted with 100 mM Tris HCl, pH 7.5, to a final concentration of 2 M guanidine HCl, followed by overnight N-linked deglycosylation at 37 °C with 2 μL of PNGase F (P0705S, New England BioLabs). After deglycosylation, 150 ng of each sample were injected onto a HPLC system (Agilent 1260) via an autosampler. Glycans were enriched and separated on a PGC-Chip (G4240-64010, Agilent) containing porous graphitised carbon columns. Separated glycans were analysed on-line via nanospray ionisation into a Q-TOF mass spectrometer (Agilent 6520). Acquired mass spectral data were searched against a glycan library in the Agilent MassHunter Qualitative Analysis software. The software algorithm utilised a combination of accurate mass with a mass tolerance of 10 ppm and expected retention time for glycan identification. Each N-linked glycan was label-free quantified relative to the sum of all identified N-linked glycans within each sample by integrating the AUC of each extracted glycan chromatogram. (b) Female SCID.bg mice (6–8 weeks) were distributed into three groups (N = 5 mice per group) and administered 1 mg/kg of FLT3L-Fc CHO-1, FLT3L-Fc CHO-2 or FLT3L-Fc N149Q CHO-2. Plasma samples were collected from each animal at −4, 0 (30 min, 6 h), 1, 3, 6, and 10 days post-dose and FLT3L-Fc concentrations were measured using human FLT3L/huFc ELISA with a minimum quantifiable concentration of 0.00047 μg/mL. (c) Blood was collected at −4, 3, 6, and 10 days. For this experiment, cDC1s were identified as IRF8+ cells.

Supplementary Figure S8.

Antitumour efficacy of FLT3L + polyI:C + anti-PD-L1 triple combo therapy. (a) Evaluation of antitumour efficacy of mFLT3L in different combination settings. Experimental design indicates timing of B16F10 tumours inoculation and mFLT3L, polyI:C and anti-PD-L1 treatments. To note, in this experiment, 30 μg mFLT3L was dosed daily under 100 μL by intraperitoneal injection. (b) B16F10 tumour growth since treatment onset. Red dotted lines represent the spline of each experimental group. Grey dotted lines represent the spline of the control group (anti-gp120), added on each graphic for reference. N = 8–10 per group. Time to progression (TTP) indicates the time required for tumours to reach 500 mm³. Data are representative of two independent experiments.

Supplementary Figure S9.

Antitumour efficacy of mFLT3L-Fc + polyI:C + anti-PD-L1 triple combo therapy in Hepa1-6 tumour model. (a) Experimental design indicates timing of Hepa1-6 tumours inoculation and mFLT3L-Fc, polyI:C and anti-PD-L1 treatments. (b) Hepa1-6 tumour growth since treatment onset. Red dotted lines represent the spline of each experimental group. Grey dotted lines represent the spline of the control group (anti-gp120), added on each graphic for reference. N = 8–9 per group. Enumeration of mice excluded from the study due tumour ulceration or tumours reaching 1500 mm³ and mice achieve complete tumour regression (CR) are indicated. Data are representative of two independent experiments.

Supplementary Figure S10.

Extended data fromFig. 5. (a) Gating strategy used to identify immune populations in B16F10 tumours. (b) Quantification of pDCs, neutrophils and T cells in B16F10 tumours. (c) Gating strategy used to identify immune populations in dLNs. (d) Quantification of pDCs, macrophages, neutrophils, T and B cells in dLNs. Median values and range are shown in panels c and d.

Supplementary Figure S11.

Activity of mFLT3L-Fc + polyI:C in B16F10 tumours and draining lymph nodes after two doses of polyI:C. (a) Experimental design indicating timing of B16F10 tumour inoculation, mFLT3L-Fc, polyI:C dosing, and samples collection. N = 4–5 per group. Data shown in panels b–g are representative of two independent experiments. Median values and range are shown in panels b–e. To note, one mouse was lost in the mFLT3L-Fc + polyI:C group of each experiment–one had to be euthanised due to severe dermatitis, the other one from an unclear cause. (b, d) Quantification of cDC1 and cDC2 in tumours (b) or dLNs (d) 24 h after the second dose of polyI:C. (c, e) CD86 and MHC-II expression on cDC1 and cDC2 in tumours (c) or dLNs (e), 24 h after the second dose of polyI:C. Representative FACS profiles are shown. (f, g) Median fold change of Viperin expression in myeloid populations present in tumours (f) or dLNs (g) compared to untreated mice.

Supplementary Figure S12.

Extended data fromFig. 6. Representative gating strategy for CD8+ T cells subsets identification in B16F10 tumours and dLNs.