JNJ-78306358, a first-in-class bispecific T cell engaging antibody targeting CD3 and HLA-G

Summary

T cell-redirecting bispecific antibodies (bsAbs) to treat advanced stage solid tumors are gaining interest after recent clinical successes. The immune checkpoint human leukocyte antigen G (HLA-G) is expressed in several tumor types while in normal tissues expression is limited. Here, we describe JNJ-78306358, a T cell-redirecting bispecific antibody (bsAb) to treat advanced stage solid tumors. JNJ-78306358 binds with high affinity to the α3 subunit of HLA-G on cancer cells and with purposely engineered weaker affinity to CD3ε on T cells. JNJ-78306358 induced potent T cell-mediated cytotoxicity of HLA-G-expressing solid tumors in vitro and in vivo. JNJ-78306358 also blocked the interaction of HLA-G with its receptors in vitro, indicating that immune checkpoint blocking may contribute to its anti-tumor activity. These results suggest that T cell-redirection against HLA-G could be a potent and effective treatment for a wide range of solid tumor indications.

Graphical abstract

Highlights

• •JNJ-78306358 is a bispecific T cell engager that targets CD3 and the oncofetal protein HLA-G
• •JNJ-78306358 binds a unique epitope on HLA-G and competes with ILT2/4
• •JNJ-78306358 effectively kills HLA-G-expressing tumors in vitro and in vivo
• •The pre-clinical activity of JNJ-78306358 supported clinical evaluation

Health sciences; Medicine; Medical specialty; Immunology; Oncology; Pharmacology; Natural sciences; Biological sciences; Cancer systems biology

Introduction

HLA-G belongs to the non-classical human leukocyte antigen (HLA) Class I family. While the structure of HLA-G resembles those of other classical HLA Class I molecules and is composed of an α chain (consisting of α1, α2, and α3 domains) non-covalently bound to a β2 microglobulin (β2m) chain, HLA-G differs from the classical HLA Class Ia molecules in two critical attributes: the peptide-binding groove of HLA-G exhibits limited gene polymorphism, and it exerts an immune checkpoint function through binding to inhibitory receptors on immune cells.¹ The known receptors for HLA-G include inhibitory receptors immunoglobulin-like transcript (ILT)2 (CD85j/LILRB1) and ILT4 (CD85d/LILRB2), expressed on peripheral immune cells.² Interaction of HLA-G with ILT2 and ILT4 on immune cells occurs via its α3 domain and leads to inhibition of innate and adaptive immunity. Dimerization of HLA-G may also play a role in enhancing its interactions with ILT2/4 via avidity and to thus enhance its tolerogenic effects.^3,4 Alternative splicing of the HLA-G gene may lead to expression of seven isoforms, with HLA-G1, -G2, and -G5 representing the primary expressed HLA-G isoforms (in GTEx and TCGA cohorts), and all these isoforms include α3 domain.

HLA-G is primarily involved in maintaining maternal-fetal tolerance during pregnancy.⁵ Aberrant HLA-G expression on cancer cells may exploit this immune tolerance mechanism. Due to its limited expression in normal tissues, HLA-G expressed on cancer cells can serve as an address to specifically target cancer cell killing via immune cell redirection.⁶ Additionally targeting the α3 domain on HLA-G may prevent immune suppression by cancer cells by blocking interaction between HLA-G and ILT2/4.

Here, we present the development and preclinical activity of JNJ-78306358, an HLA-G x CD3ε T cell engaging bispecific antibody (bsAb). JNJ-78306358 mediates T cell-dependent cytotoxicity of HLA-G-expressing cancer cells as its primary mechanism of action with potential immune checkpoint blocking function as an additional activity. JNJ-78306358 specifically binds to the α3 domain of HLA-G, preventing HLA-G binding to ILT2/4 and to the cluster of differentiation 3 (CD3) epsilon (ε) on T cells, leading to T cell activation and specific cancer cell death. JNJ-78306358 showed specific and potent T cell-dependent killing activity against HLA-G-expressing cancer cell lines in vitro and potent anti-tumor activity in in vivo mouse xenograft models. In vitro, JNJ-78306358 also prevented HLA-G binding to ILT2/ILT4 on immune cells. These results supported clinical evaluation of JNJ-78306358 in a Phase 1 study in patients with advanced cancers with a high prevalence of HLA-G expression (NCT04991740).⁷

Results

HLA-G protein expression in tumors

HLA-G expression has been reported in many types of solid tumors using the monoclonal antibody (mAb) 4H84.^8,9,10 4H84 binds to a unique linear epitope of HLA-G α1 domain present on all HLA-G isoforms (HLA-G1-7; Figure S1). We confirmed binding of 4H84 to HLA-G1, HLA-G2, HLA-G3, and HLA-G5 isoforms (Table S1). Although in these studies, commercial bound to recombinant HLA-G isoforms, response magnitudes for MEM-G1 and 5A6G7 were lower than for JNJ-78306358 (Rmax ∼5–10 compared to Rmax ∼50).¹¹ Although the α3-domain of HLA-G (HLA-Gα3) is not present on HLA-G3 G4, and G7, immunohistochemistry (IHC) staining using 4H84 was considered representative of the expression profile of JNJ-78306358-targetable HLA-G1, -G2, -G5 and -G6 isoforms, because of the low prevalence of isoforms lacking α3 domain.

IHC analysis using 4H84 demonstrated high incidence of HLA-G expression (positivity of ≥1% at any intensity) in 75% of renal cell carcinoma (RCC; 58 of 77 whole tissue [WT] samples), 61% of ovarian carcinomas (53/87 WT; and 37% [14/38] tissue microarray (TMA)), 64% of colon cancer (18/28 WT; and 8% [4/51] TMA) (Figures 1A and 1B), 40% of rectal cancer (4/10 WT; and 8% [3/39] TMA), 35% of endometrial cancer (6/17 WT; and 11% [4/38] TMA), 37% of lung (106/290 WT; and 12% [30/244] TMA), 33% of head and neck carcinoma (3/11 WT; and 35% [11/31] TMA), 29% of esophageal cancer (6/21 WT; and 11% [2/18] TMA), 22% of breast cancer (4/18 WT; and 3% [1/37] TMA), and 22% of pancreatic cancer (29/129 TMA) (Table S2). Among lung cancer subtypes, the highest incidence of HLA-G positivity was observed in lung adenocarcinoma (LUAD) (39%, 97/248 TMA and WT). Additional solid tumors demonstrated positive HLA-G staining: prostate cancer 24% (8/34 WT; and 8% [2/26] TMA), stomach cancer 17% (9/52 TMA), hepatocellular carcinoma 15% (7/47 TMA), bladder cancer 10% (14/140 TMA; with 22% incidence in squamous cell carcinoma, 4/18), and small intestine cancer 8% (5/65 TMA) (Table S2). Differences in HLA-G prevalence between WT and TMA may be attributed to heterogeneity in HLA-G expression and small tumor size sampling with TMA leading to lower detected HLA-G expression in TMA.

Figure 1.

Level of HLA-G expression in human cancer and normal tissue samples by IHC

IHC was performed on FFPE tumor (A and B) and normal (C–F) tissues using 4H84 at 1 μg/mL and 2 μg/mL dilution, respectively. Detection and counterstaining of tumor tissues was achieved as per manufacturer’s instructions using Bond Polymer Refine Detection kit (Leica), while secondary HRP-conjugated OmniMap anti-mouse antibody, followed by chromogenic detection with DAB was performed for normal tissues.

(A) Quantification of HLA-G expression on large tissue samples from renal (n = 77), ovarian (n = 87) and colon (n = 28) cancers demonstrated overall positivity of 75%, 61% and 64%, respectively. Tumor HLA-G levels are represented as staining intensity (0, 1+, 2+ and −3+) and percent positivity for H-score >0.

(B) Representative IHC images (20x magnification) for each indication under respective graph.

(C and D) HLA-G expression by IHC in human placenta (C, bar = 1 mm) with positive staining of extravillous trophoblast (EVT) (D, bar = 100 μm).

(E and F) HLA-G expression by IHC in pituitary gland (E, bar = 1 mm) with positive staining of neuroendocrine cells (F, bar = 100 μm) in anterior pituitary.

HLA-G expression in healthy human normal tissues is restricted to placenta and pituitary gland

We evaluated the normal tissue expression of HLA-G by IHC using 4H84 on FFPE TMA and large format samples (Figures 1C–1F; Table S3). HLA-G immunoreactivity was most consistently observed in extravillous trophoblasts (EVTs) at the fetal-maternal interface of the placenta (Figures 1C and 1D), and in a subset of adenohypophyseal cells of the pituitary gland (Figures 1E and 1F). Limited and incidental HLA-G staining was also observed in splenic vascular/sinusoidal endothelial cells in red pulp and subsets of mononuclear cells in white pulp, and rare thymic epithelial cells in the medulla (cells in Hassall’s corpuscles).

HLA-G expression can be induced by cytokine exposure, hormone exposure, inflammation, hypoxia, or disease states. Disease skin tissue from psoriasis patients (n = 8), as well as healthy lung tissue from human donors clinically characterized as normal (n = 4, 1 sample per donor; Figure S2A), with asthma (n = 9, 2 samples per donor; Figure S2B) and chronic obstructive pulmonary disease (n = 2, 1 sample per donor), demonstrated that 95% of lung area was negative in normal donors lacking severe inflammation, with HLA-G positivity in type I pneumocytes of random rare foci comprising <5% of lung area. HLA-G positive pneumocytes were increased in asthma donors with HLA-G positive macrophages in areas of severe chronic inflammation. Analysis of psoriasis samples showed positive HLA-G staining of mononuclear cells in the dermis.

Positive HLA-G staining of type I and type II pneumocytes was also observed in regions adjacent to lung tumor in 35/60 samples (Figures S2C and S2D). HLA-G IHC positive pneumocytes were detected in benign lung parenchyma adjacent to HLA-G positive (15/35) and HLA-G negative (20/35) tumors. HLA-G positive pneumocytes were detected in areas with (22/35) and in areas devoid of (13/35) cellular infiltrates/inflammation in benign lung parenchyma adjacent to tumors.

A tissue cross reactivity (TCR) study was performed to determine potential target organs of toxicity, based upon the binding of JNJ-78306358 (and its bivalent mIgG2a mAb analog JNJ-78655317) in histologically prepared cryo-sections from human tissues. In this analysis, cell membrane staining was limited to placenta (extra-villous trophoblasts in the decidua) and pituitary (neuroendocrine cells in the anterior pituitary particularly within the pars distalis) and, as expected, extended to lymphocytes, consistent with CD3⁺ T cells for JNJ-78306358 (HLA-GxCD3) staining.

Together, these data indicated that normal human tissue expression of membrane associated HLA-G is likely restricted to the pituitary (cells in the pars distalis) and placenta (EVTs) and that there is a potential for HLA-G upregulation in normal tissue in response to inflammatory stimuli and in diseased tissue.

JNJ-78306358 specifically binds to the membrane proximal α3 domain of HLA-G, blocking the interaction with ILT2/4 immune checkpoint receptors

Hydrogen deuterium exchange (HDX)-based mass spectrometry (LC-MS) analysis of tryptic peptides was used to identify the epitopes bound by JNJ-78306358, ILT2 and ILT4 on HLA-G. The assay was performed with JNJ-78980577, an IgG1 mAb with the same anti-HLA-G variable region as JNJ-78306358. The epitopes on HLA-G for each ligand were defined as residues with HD exchange rates >10% of control. Recombinant HLA-G featured HLA-G1_C42S fused to β2m and the cognate peptide: KLPAQFYIL.¹² JNJ-78980577 bound to the same site on HLA-G as ILT2/ILT4 (Figure 2A). HDX-MS identified HLA-G peptides bound by ILT2 as residues 22–24 (QRT) and 107–109 (LSQ) of β-2-microgloblin, as well as residues 215–220 (HHPVFD), 221–222 (YE), 252–254 (EL), and 270–272 (AVV) of HLA-G (numbering based on Uniprot ID P61769 & P17693 for β2m and HLA-G respectively). HDX-MS identified peptides bound by JNJ-78980577 as residues 215–220 (HHPVFD), 221–222 (YE), and 273–275 (VPS) of HLA-G. The epitopes bound by JNJ-78980577 were overlapping with those of ILT2/ILT4. The HDX-MS identified epitopes of HLA-G bound by hILT2 were consistent with X-ray crystal structure of HLA-G-hILT2 complex (PDB: 6AEE)¹³ and extended to both the β2-microblobulin and HLA-G subunits. HDX-MS identified epitopes of HLA-G bound by hILT4 as residues 22–24 (QRT), 107–109 (LSQ), and 112 (I) of β-2-microgloblin, as well as residues 221–222 (YE) of HLA-G. The HLA-G epitopes bound by ILT4 are similar to those of ILT2. JNJ-78306358 binds to the α3 domain in the membrane proximal region of HLA-G where ILT2/ILT4 binding epitopes reside and can thus compete with the receptors for HLA-G binding.

Figure 2.

JNJ-78306358 binds to a membrane-proximal region of the α3-subunit of HLA-G and competes with hILT2 and hILT4 binding to HLA-G

(A) Hydrogen-deuterium exchange-based mass spectrometry (HDX-MS) was used to identify HLA-G regions bound by JNJ-78980577, human ILT2, or human ILT4. Binding epitopes for each agent (JNJ-78980577, ILT2 and ILT4) are highlighted in red on the crystal structure of HLA-G/β2m (PDB ID 1YDP). JNJ-78980577, ILT2, and ILT4 share a partial binding epitope comprising the AB loop within the HLA-G α3 subunit.

(B) Sequence alignment of the α3 domain AB loop of HLA-G and related homologs highlight F219 as unique to human HLA-G (Mafa-AG: cynomolgus monkey homolog, Mamu-AG: rhesus macaque homolog, Qa-2 putative mouse homolog).

(C) Dose dependent inhibition of HLA-G binding to hILT2/4-expressing HEK cells was determined by monitoring the changes in fluorescence. The inverse of the normalized fluorescence is reported as % inhibition in HLA-G:ILT2/hILT4 binding. Antibodies and target cell lines tested are indicated in the legend.

We tested the ability of JNJ-78306358 to block binding between HLA-G and ILT2/4 using a cell-based inhibition assay. Briefly, recombinant HLA-G bound to a fluorescent dextramer was incubated with HEK293T cells expressing ILT2 or ILT4. Fluorescence was monitored in the absence or presence of JNJ-78306358 and JNJ-78980577. Interference of binding of fluorescent dextramer-HLA-G to HEK293T cells expressing ILT2 (EC₅₀ = 0.81 nM) or hILT4 (EC₅₀ = 1.3 nM) by JNJ-78306358 demonstrated competition with ILT2/4 (Figure 2C).

The HDX-identified epitope for JNJ-78980577 (and that of JNJ-78306358) is unique to HLA-G and diverges by 2–5 amino acids to other HLA class I molecules (HLA-A, HLA-B, HLA-C and HLA-E; Figure 2B). The amino acid identity within the HDX-identified peptides compared to human HLA-G is 92%, 92%, and 52% for mafa-AG (cynomolgus monkey), mamu-AG (rhesus macaque), and Qa-2 (mouse), respectively. Within the identified epitope, both mafa-AG and mamu-AG diverge from HLA-G at a single amino acid residue, whereas Qa-2 diverges at 6 of 13 amino acid residues. F219 is unique to human HLA-G (Figure 2B), and this residue may contribute to the lack of cross reactivity of JNJ-78306358 with mafa-AG, mamu-AG and Qa-2 (Figure S3A).

JNJ-78306358 induces potent and specific T cell-mediated killing of HLA-G expressing cells

JNJ-78306358 simultaneously binds to CD3ε on T lymphocytes and redirects T cells to cancer cells via binding to the α3 domain on HLA-G on tumor cells. JNJ-78306358 features mutations of L234A, L235A, and D265S (AAS) in the constant region (Fc) to abolish interactions with Fc receptors while heterodimerization was enhanced using the Zymeworks Azymetric mutations on each heavy chain (chain 1: T350V, L351Y, F405A, Y407V (chain 1) and chain 2: T350V, T366L, K392L, T394W).¹⁴ The molecule features an anti-CD3ε Fab on chain 1 and an anti-HLA-Gα3 scFv on chain 2 (Figure 3A). The binding affinity of JNJ-78306358 is 7 pM for recombinant HLA-G and 22 nM for recombinant CD3ε (Figures S3B and S3C).¹⁵

Figure 3.

JNJ-78306358 binds specifically to K562 cells expressing HLA-G and Jurkat cells expressing CD3, and induces T cell mediated killing of HLA-G-expressing cell lines

(A) JNJ-78306358 was designed as a heterodimeric “bipod” type bsAb.¹⁶ The antibody features mutations in the Fc to abolish interaction with Fcγ receptors. The anti-CD3 binding region formatted as a Fab binds to CD3ε with K_D = 22 nM. The anti-HLA-G binding region is formatted as a single-chain fragment variable (scFv) binding to HLA-G with K_D = 13 pM.

(B and C) Dose dependent binding of JNJ-78306358 to cells (K562 and CHO) expressing HLA-G or other class I MHC molecules (B), and Jurkat cells negative (CD3^Neg) or positive (CD3^Pos) for CD3 (C). The y axis represents fold-increase in mean fluorescence intensity over background. MHC Class I-transfected cells expressed HLA-A/B/C/E/G from Uniprot IDs P01892/P18464/P30508/P13747/P17693, respectively.

(D) JNJ-78306358-induced T cell activation and T cell-mediated killing of K562 cells expressing HLA-G. JNJ-78306358 was incubated for 3 days with K562 and human T cells (E:T ratio 8:1). Percent cytotoxicity and % CD25 expression on T cells were measured by flow cytometry. Error bars represent mean ± SEM.

On-target specificity of JNJ-78306358 was confirmed by concentration-dependent binding to cells expressing HLA-G and CD3 and by T cell-mediated cytotoxicity (Figures 3B–3D). The EC₅₀ for binding of JNJ-78306358 and JNJ-78980577 to K562-HLA-G cells, determined by flow cytometry, were 21.5 nM and 8.8 to ∼14 pM, respectively. JNJ-78306358 failed to bind HLA-G^neg cells, cells overexpressing other HLA Class I molecules (K562-HLA-A, HLA-B, HLA-C), or to parental K562, parental CHO, and HLA-E-overexpressing CHO cells, although binding by JNJ-783-6358 is isoform specific (Figure 3B). JNJ-78306358 also did not induce T cell-mediated cytotoxicity of these cells (Figure 3D), confirming the specificity of JNJ-78306358 to HLA-G. JNJ-78306358 additionally demonstrated dose-dependent binding to CD3-expressing Jurkat cells (Figure 3C). Control antibodies (HLA-G x null and null x CD3) did not induce cytotoxicity of HLA-G expressing cells (Figure S4).

JNJ-78306358 induces T cell-mediated killing of cancer cells endogenously expressing HLA-G

Several cancer cell lines (RERF-LC-Ad1, HuP-T3, NCI-H2009, BICR 6, SH-4, HCC1806) that endogenously express HLA-G at the RNA (RNAseq levels of 108.1, 63.0, 197.3, 14.7, 29.1 and 16.7 transcripts per million (TPM), respectively), and protein level (Wes analysis) were characterized for HLA-G expression in vitro by flow cytometry (Figures 4A and S5).^17,18 RERF-LC-Ad1 cells demonstrated a median receptor density value of 53,000 (21,000–56,000; n = 10), corresponding to geo-MFI index 111.3, consistent with high relative level of mRNA expression. For comparison, K562 cells engineered to express HLA-G showed a geo-MFI index of 120.0. HuP-T3 cells displayed a biphasic pattern of HLA-G staining, with 51%–75% HLA-G⁺ population with a median HLA-G receptor density per cells of 59,000 (range of 57,000–71,000; n = 9), corresponding to geo-MFI index values 42.2 [for 71.2% of HLA-G⁺ cell population]. NCI-H2009 showed no detectable membrane HLA-G expression (geoMFI index value 0.6), attributed to lack of β2m expression. Engineering NCI-H2009 to express β2m enabled HLA-G trafficking and its detection at the cell surface (geoMFI index 8.0; Figure 4A). NCI-H1975 and K562 cells do not express HLA-G (RNAseq levels of 0.15 and 0 TPM, respectively) and served as negative controls (Figure S5).

Figure 4.

JNJ-78306358 demonstrates potent T cell mediated killing of HLA-G expressing cancer cells

(A) Level of HLA-G expression, ratio of geoMFI, on indicated cell lines was assessed by flow cytometry with anti-HLA-G JNJ-78980577-PE antibody and compared to that of isotype control IgG1-PE antibody. Note, JNJ-78980577 is a mAb consisting of the same anti-HLA-G variable region as JNJ-78306358. BICR6 and HuP-T3 cell lines demonstrated biphasic staining for membrane HLA-G expression. The percentage of subpopulation with high HLA-G expression is indicated for these cell lines in parentheses. Graphs are normalized to mode.

(B) Dose response of JNJ-78306358 in healthy donor T cell-mediated killing of indicated cancer cell lines was assessed at 72 h with E:T ratio of 3:1 and 1:1. Duplicate samples were tested for each condition, and data are graphed as the mean ± SD. The experiment was performed with several donors for each cell line; results for a single donor are shown. Error bars represent mean ± SEM.

(C) JNJ-78306358 50% effective concentrations (EC₅₀) for T cell activation (CD25), cytotoxicity (at 96 h) and secretion of measured cytokines (at 48 h). Cytotoxicity (% killed RERF-LC-Ad1 target cells) was measured by xCELLigence real-time cell analyzer and percentage was calculated as described in STAR Methods. T cell activation (% CD25⁺ T cells) was measured on T cells by FACS analysis (CD25-BV421, Biolegend 302630, 1:125 dilution). Each donor (n = 6) is presented with a different color. Individual curves are reported in Figure S6.

(D) Correlation between level of HLA-G expression by cancer cells with secretion of INF-γ by T-cells and sensitivity to JNJ-78306358 mediated T cell killing. Quantification by MSD of in vitro IFN-γ release from 6 renal cell carcinoma DTC samples upon JNJ-78306358 treatment. HLA-G expression levels were determined by flow cytometry (receptor density, RD and % of HLA-G⁺ cells) and Wes (% area of HLA-G of total protein). IFN-γ arbitrary threshold based on dichotomic maximum induction (<100 pg/mL and >200 pg/mL) is indicated by red dotted line and correlates with HLA-G expression, where higher HLA-G expression leads to increased JNJ-78306358 induced INF-γ production.

(E) Extrapolation of sensitivity to JNJ-78306358 mediated T cell killing based on HLA-G quantification and IFN-γ secretion threshold (defined in D). Level of HLA-G expression by capillary Wes for indicated normal and tumor tissue samples, including renal DTC samples (analyzed in D). Values represent total area under the HLA-G^Pos peak normalized to area of total protein peak. HLA-G levels detected in HuP-T3 and RER-FL-CAd1 cell lines are shown as positive controls and correlate with their sensitivity to JNJ-78306358 mediated T cell killing.

JNJ-78306358 T cell-mediated cytotoxicity was characterized at commonly used effector to target cell (E:T) ratios using RERF-LC-Ad1, HuP-T3, BICR 6 and NCI-H2009-β2m and HLA-G negative NCI-H2009 cell lines, both with isolated T cells (Figure 4B) and PBMCs (Table S4). JNJ-78306358 induced complete T cell mediated killing of cell lines expressing high HLA-G levels: RERF-LC-Ad1 and HuP-T3. For NCI-H2009-β2m cell line expressing lower HLA-G levels, complete T cell mediated killing was achieved only at the highest JNJ-78306358 concentration. Cytotoxicity against BICR 6 cells was further reduced compared to these above cell lines and demonstrated donor dependency, reaching up to 80%–90% maximal cytotoxicity at 3:1 ratio, and lower cytotoxicity of ∼45% at 1:3 E:T ratio. JNJ-78306358 did not induce cytotoxicity of the negative control HLA-G^neg NCI-H1975 cells. These data show that the level of JNJ-78306358 cytotoxicity is dependent on membrane HLA-G expression levels.

JNJ-78306358 mediated T cell killing against RERF-LC-Ad1 was further measured using T cells isolated from 6 different donors at E:T ratio of 1:1 (Figure 4C). JNJ-78306358-mediated T cell activation, as measured by CD25⁺ T cells expression and showed similar concentration-dependent response as the cytotoxicity, with average EC₅₀ value of 0.028 nM (48 h, 1:1 E:T, n = 6 donors). This EC₅₀ value was slightly greater than for cytotoxicity (Figures 4C and S6). The maximal percentage of CD25⁺ T cells ranged from 60% to 80% at the 1:1 E:T ratio JNJ-78306358 also induced a concentration-dependent increase in: IFN-γ, IL-1β, TNF-α, IL-2, IL-4, IL-10, IL-6, IL-8 and IL-13, with IFN-γ and IL-8 showing highest response (Figure S7A). In one donor, IL-12p70 also showed a response. However, non-linear mixed-effects (NLME) model-fitting for effective concentration estimates was generally poor for tested cytokines and did not converge for cytokines IL-12p70, IL-1β, and IL-8. T cell activation was the most sensitive and consistent readout tested and was chosen to determine the minimal anticipated biological effect level (MABEL) concentration (Figure 4C).

IFN-α/β/γ, cytokines (IL-1β, IL-2, IL-10, and transforming growth factor [TGF]-β),^19,20,21 immunomodulatory steroid hormones (hydrocortisone, β-estradiol, and progesterone)^22,23,24 and stress conditions (hypoxia)^25,26 may increase HLA-G expression, which could lead to an increase in JNJ-78306358 T cell mediated killing in HLA-G-expressing cells. To understand this potential effect, we performed several analyses. First, changes in HLA-G expression level was measured following exposure of cells to selected JNJ-78306358 induced cytokines (100 IU/mL IFN-γ,^27,28,29 20 ng/mL TNF-α, 50 ng/mL IL-6, or 100 ng/mL IL-10). Among the tested agents, IFN-γ, IL-6 and TNF-α treatments statistically upregulated cell surface HLA-G expression in several HLA-G expressing tumor cell lines tested (BICR 6, HuP-T3, and RERF-LC-Ad1), but did not cause de novo HLA-G expression in HLA-G^neg cells (NCI-H1975 and normal fibroblast cell line WI-38 VA-13 Subline 2RA) (Figure S7B). Second, JNJ-78306358-mediated cytotoxicity against RERF-LC-Ad1 was assessed before and following exposure to IFN-γ, one of the cytokines that can increase HLA-G levels on tumor cells. Similar levels of cytotoxicity (EC₅₀) were observed under all conditions tested, suggesting that the limited increase in HLA-G levels by IFN-γ treatment on these cancer cells is insufficient to affect JNJ-78306358 potency (Figure S7C).

JNJ-78306358 induces autologous T cell activation in HLA-G-expressing dissociated tumor cells (DTCs)

To assess clinical relevance of T cell activation and T cell-mediated killing induced by JNJ-78306358, we quantified the ability of JNJ-78306358 to induce activation of autologous tumor-infiltrating T cell (TILs) from renal cell carcinoma patients. First, HLA-G receptor density and percentage of HLA-G expression was measured on DTCs by flow cytometry and Wes, respectively (Figure S5). Flow cytometry analysis of DTCs showed heterogeneous (0%–86%) expression of HLA-G, consistent with IHC data from renal cancer tissues (Figure 1B). These results provided independent confirmation that HLA-G protein was expressed at the surface of DTCs with receptor density levels comparable to those on cancer tumor tissues.

To better understand the correlation between HLA-G expression levels and JNJ-78306358-induced T cell activation, DTCs were further evaluated by Wes for HLA-G expression (Figure 4D) then incubated in the presence of JNJ-78306358, to assess the induction of IFN-γ production by autologous TILs. JNJ-78306358 induced IFN-γ release by DTC T cells in three of five HLA-G⁺ DTC samples (Figure 4D), and the level of induced IFN-γ in the other two HLA-G-expressing DTCs were similar to the DTC sample lacking HLA-G expression (Figures 4D and 4E). The IFN-γ response appeared to be dichotomous with an arbitrary threshold of HLA-G expression level required to elicit in vitro activity of JNJ-78306358. HLA-G expression, evaluated by Wes was sufficient for production (>200 pg/mL) of IFN-γ. When comparing the levels of HLA-G expression by these DTCs with normal and tumor tissues, the data indicated that HLA-G expression threshold by Wes (>2.2) would not be reached in lung tumor-adjacent and normal tissue, while expression levels in placenta and pituitary tissues are higher than levels detected in DTCs (Figure 4E). Although sample numbers were limited, these data suggest that HLA-G expression level is not sufficient to lead to JNJ-78306358-mediated T cell killing on normal and tumor adjacent (normal) lung tissues.

Pharmacokinetics of JNJ-78306358 in mouse

The pharmacokinetics (PK) of JNJ-78306358 was determined in serum and non-perfused tumors from a single intraperitoneal (IP) dose of JNJ-78306358 of 0.3 mg/kg or 1 mg/kg in T cell-humanized mice (Table 1). JNJ-78306358 demonstrated linear PK with an average half-life of 6.1–8.4 days for doses tested. In tumor tissue, the maximum JNJ-78306358 concentrations were observed at 72 h for 0.3 mg/kg dose group, and 48 h for 1 mg/kg dose group with tumor-to-serum JNJ-78306358 concentration ratios for AUC_last of 45.20% and 29.54%, respectively.

Table 1.

PK parameters of JNJ-78306358 in serum and tumors following a single 0.3 or 1 mg/kg IP administration in mice bearing HuP-T3 tumors

Dose (mg/kg)		Cmaxa μg/mL	AUClasta μg.day/mL	AUCinfa μg.day/mL	CLa mL/day/kg	T1/2a day
0.3	Serum	3.54	21.15	35.33	8.69	8.43
	Tumor	1.45	9.56	11.11
	%Tumor/Serum AUClast ratio		45.20
1.0	Serum	14.29	75.40	103.87	9.84	6.06
	Tumor	3.33	22.27	26.14
	%Tumor/Serum AUClast ratio		29.54

AUC_inf, AUC from time 0 to infinity with extrapolation of the terminal phase; AUC_last, area under the plasma or tumor concentration-time curve from time zero to the time of the last observed quantifiable concentration; CL, total clearance of drug; C_max, maximum observed plasma or tumor concentration; PK, pharmacokinetic; T_1/2, half-life. N = 3/group.

^aIndividuals were unique for each time point; values do not represent serial collections from a single individual.

Soluble HLA-G is not significantly elevated in serum of cancer patients

Antibody PK profile can be affected by circulating soluble target, reducing therapeutically available systemic antibody. To assess the potential soluble target sink effect PK, we measured the levels of soluble HLA-G (sHLA-G) in serum from healthy donors, pregnant women (as a positive control) and cancer patients using an MSD assay. While sHLA-G could only be detected in 1/9 healthy normal individuals (333.4 pg/mL), sHLA-G was present in sera of all pregnant women samples with mean value 316.4 ± 204.6 pg/mL. Different levels of sHLA-G were measured in cancer patient sera with no statistically significant difference compare to these of healthy donors; (Figure 5): Of these, 3/31 lung (%) and 2/42 (%) ovarian cancer patient samples showed sHLA-G levels exceeding 1 ng/mL, with the highest level noted in one lung cancer sample at 3.67 ng/mL. Soluble HLA-G levels measured in this study were estimated to have limited impact on the free JNJ-78306358 concentration based on modeling, although this estimate may be lower due to use of 4H84 as detection reagent.

Figure 5.

Levels of human serum HLA-G

Soluble HLA-G (sHLA-G) levels in serum from healthy (normal, pregnant) donors and cancer (lung, kidney, ovary and colorectal) patients were measured by MSD assay with JNJ-78306358 as a capture antibody and 4H84 as a detection antibody. Dash line indicates the limit of quantification (LLOQ) of 39 ng/mL. Statistically significant differences were determined using t-test. Asterisks indicate P-value <0.001. Bars indicate median values.

JNJ-78306358 in vivo anti-tumor activity correlates with increased infiltration of T cells

The dose-dependent effects of JNJ-78306358 were evaluated in female NSG mice with human pancreatic HuP-T3 xenografts that exhibited a consistent median HLA-G expression of approximately 8,000 receptors per cell. Significant inhibition of tumor growth (ΔTGI) was observed at all tested doses of JNJ-78306358 (0.005, 0.01, and 0.1 mg/kg), with %ΔTGI values of 96%, 97%, and 160% respectively, compared to control mice treated with Null x CD3 on Day 41 after tumor implantation (p ≤ 0.05, n = 10/group; see Figure 6A). Although the two lower doses resulted in similar efficacy, the 0.1 mg/kg treatment displayed superior efficacy, reflecting non-linearity of efficacy. On Day 14, 24 h after the second treatment, representative micrographs were subjected to IHC processing and showed a dose-dependent increase in infiltration of CD8⁺ and CD4⁺ T cells by JNJ-78306358 (Figure 6B).

Figure 6.

HLA-G expression and JNJ-78306358 activity against human cell line- and patient-derived xenografts

(A) Efficacy of JNJ-78306358 on SC HuP-T3 pancreatic tumor cell line-derived xenografts (CDX) in T cell-humanized NSG mice. Group tumor volumes are graphed as the mean ± SEM (n = 10/group). Tumor cells were implanted on Day 0, T cells were administered IP on Day 8, and antibody treatment occurred on Days 9, 13, 16, 20, 23, 27, 30, 34, 37, and 41 (dosing period indicated by the bar below the X axis). ∗ Denotes significant difference (p ≤ 0.05) of treatment groups versus control group on Day 41. Error bars represent mean ± SEM.

(B) JNJ-78306358 induced T cell infiltration in HuP-T3 CDX in T cell-humanized mice. Tumor cells were implanted on Day 0, T cells were implanted on Day 8, followed by treatment with JNJ-78306358 or nullxCD3 antibodies at indicated doses on Days 9 and 13. Representative IHC micrographs of HuP-T3 CDX stained for the presence of intratumoral CD8⁺ and CD4⁺ T-cells are shown. Magnification bar denotes 200 μm.

(C) Study scheme of JNJ-78306358 in vivo anti-tumor efficacy against PDX tumor models from different cancer types in T cell humanized NSG mice. Tumors were implanted SC and after randomization (group mean tumor volumes 131–182 mm³) mice received 2 × 10⁷ human CD3⁺ T cells IP (allogeneic T cell transfer, ACT). JNJ-78306358 (0.03 and 0.3 mg/kg) treatment was initiated one day after ACT on Days 1, 5, 8,12,15 and 19. Tumors were collected from an additional untreated cohort (4 mice) on Day 14 and assessed for HLA-G expression by IHC (Table S5; Figure S7).

(D) Association between HLA-G expression in PDX (n = 10) and their sensitivity to JNJ-78306358. HLA-G IHC staining was performed with 4H84 and JNJ-78655317. Level of HLA-G expression is presented as H-score. Antitumor efficacy of JNJ-78306358 at 0.3 mg/kg was assessed at the first tumor measurement after the end of therapy, Day 20 or Day 21 depending on the tumor model, using the vehicle control group as a reference. Antitumor efficacy for respective tumor models is annotated on either as complete responses (CR), as the %ΔTGI>0, or %ΔTGI ≤0. HLA-G expression is denoted by different colors; high (blue), low-medium (gray) and no expression (red).

(E) Representative anti-tumor efficacy of JNJ-78306358 against tumor models presented in panel D expressing high (RXF 488; blue), low-medium (BXF 439; gray), and no HLA-G (PAXF 546; red), plotted as mean tumor volumes over time.

Correlation between JNJ-78306358 efficacy and levels of HLA-G expression of patient derived tumor xenografts (PDX)

We evaluated the correlation between HLA-G expression and the anti-tumor efficacy of JNJ-78306358, as well as the minimum HLA-G expression level required for tumor growth inhibition using 10 patient-derived xenograft (PDX) models with varying HLA-G expression levels (Figures 6C–6E). The PDX models were selected based on their HLA-G expression levels. HLA-G protein expression was initially screened using western blot, and HLA-G receptor densities were measured using flow cytometry on fresh DTCs derived from PDX tissues. The PDX models exhibited heterogeneous HLA-G expression, covering a wide range of HLA-G positivity (Figure S8; Table S5).

For the efficacy study, the PDX models were used in female NSG mice reconstituted with in vitro activated and expanded human CD3⁺ pan T cells from a single donor, which allowed direct comparison of the anti-tumor activity between models. JNJ-78306358 was administered intravenously twice per week at 0.3 mg/kg and 0.03 mg/kg and compared to vehicle control group (Figure 6C). PDX models were selected based on the HLA-G expression determined using 2 different IHC assays. On day 14 post implantation, an untreated group of tumors were split in two, where one-half was FFPE processed and stained using 4H84 antibody and the other half frozen and stained using JNJ-78655317 (an IgG1-based scFv-fusion having the identical anti-HLA-G variable region as JNJ-78306358). Staining of the samples with the two antibodies demonstrated comparable level of HLA-G expression independent of antibody used (correlation coefficient = 0.8972, adjusted R² = 0.7919, p < 0.001; Figures 6D and S8; Table S5), demonstrating that 4H84 is a relevant IHC antibody to clinically evaluate HLA-G expression before JNJ-78306358 treatment.

Anti-tumor efficacy of JNJ-78306358 was observed in all PDX models with detectable HLA-G expression levels, but not in the two tumor models that lacked HLA-G expression (H-value (JNJ-78655317) = 0), namely GXF281 and PAXF546 (Figures 6D and 6E; Table S5). The PDX efficacy results were annotated with H-score for each model using both IHC assays. Treatment with JNJ-78306358 resulted in complete tumor regression in all treated mice at both tested doses of JNJ-78306358 in five of the tumor models (RXF 488, PAXF 2175, RXF 2706, LXFA 2204, PRXF MRI-H 1579). Among these, three PDX models exhibited high expression of HLA-G (average H value ∼200; with ≥60–70% of tumor cells showing 2–3+ intensity level), but the HLA-G expression was highly heterogeneous and/or less extensive (average H value ≤ 105; with <20% HLA-G+ tumor cells and/or 1–2+ intensity level) in PAXF 2175 and RXF 2706 tumors. Among the tumor models that did not result in complete responses but demonstrated dose-dependent tumor growth inhibition, a lower percentage of HLA-G+ cells with lower expression intensity was measured (average H value = 60–176).

JNJ-78306358 induced T cell redirection and tumor infiltrated lymphocytes (TILs) lead to efficacy in CD34⁺ HSC humanized mice

The correlation between HLA-G expression in patients and poor clinical prognosis has been established,³⁰ however the direct role of HLA-G in tumor escape and targeted therapeutic benefit has only been pre-clinically studied in simplified or clinically irrelevant in vivo models.^31,32,33,34 JNJ-78306358 efficacy was evaluated against two HLA-G expressing PDX models (PAXF 1657 and LXFA 983) in CD34⁺ HSC humanized mice, with biologically relevant level and activation state of T cells, and additional improved engraftment of human innate cells (NSG-SGM3 and NOG-EXL mice). PAXF 1657 model was implanted in human CD34⁺ HSC engrafted NSG-SGM3 mice. ΔTGI was calculated on day 35 post tumor implantation, a time at which at least 8 animals of the 11 treated remained in each group. Significant antitumor efficacy was observed with JNJ-78306358 treatment at all doses tested compared to PBS control group (p ≤ 0.05; Figure 7A) or HLA-G x null antibody (1.0 mg/kg). HLA-G x null antibody treatment did not result in biologically significant %ΔTGI of 8% compared to PBS control. Treatment with JNJ-78306358 at 0.1, 0.3, and 1 mg/kg elicited ΔTGI of 99%, 104%, and 98%, respectively, and resulted in 8 of 11, 9 of 11, and 10 of 11 mice with CRs, respectively, on the last study day (Day 37). With potent efficacy and high CR rate, leading to absence of tumor for analysis, PAXF 1657 model was not suitable to evaluate pharmacodynamic effect of JNJ-78306358. Therefore, activation status of tumor-infiltrated CD8⁺ and CD4⁺ T cells in human CD34⁺ HSC engrafted NOG-EXL mice upon treatment with JNJ-78306358 was assessed in a separate efficacy/pharmacodynamics study in the SC LXFA 983 model, in which JNJ-78306358 did not elicit CRs (Figure 7B). Flow cytometry and IHC analysis of leukocytes from tumors collected on experiment days 5 and 12 from mice bearing the LXFA 983 tumor model showed strong increase in both CD4⁺ and CD8⁺ T cells, as well as CD25⁺ CD8⁺ cells (Figures 7C and 7D).

Figure 7.

JNJ-78306358 exerts T cell-mediated anti-tumor activity in CD34⁺ HSC humanized mice

(A) JNJ-78306358 efficacy against PAXF 1657 human pancreatic PDX tumors in CD34⁺-Human Stem Cells (HSC)-humanized NSG-SGM3 mice. HSCs were implanted 91 to 105 days prior to tumor implantation followed by treatment with JNJ-78306358 at indicated doses, HLA-G x null at 1 mg/kg or PBS control on Days 10, 13, 17, 20, 24, 27, 31, and 34 (dosing period indicated by the bar below the X axis). (n = 10–11/group). ∗∗∗Denotes significant difference (p ≤ 0.001) between each JNJ-78306358-treatment group versus PBS vehicle control group on Day 35. Error bars represent mean ± SEM.

(B) JNJ-78306358 efficacy against SC LXFA 983 lung PDX tumors in CD34⁺ HSC-humanized NSG-SGM3 mice. HSCs were implanted 84 to 91 days prior to tumor implantation, followed by treatment with JNJ-78306358 at 0.3 mg/kg or null x CD3 at 1 mg/kg on Days 18, 21, 25, 28, 32, 35, 39, and 42. (n = 8–9/group). ∗Denotes significant difference (p ≤ 0.05) between the JNJ-78306358 0.3 mg/kg group versus control (null x CD3) group at Day 46.

(C) Increased number of T cells (CD3⁺/CD56^neg, CD8⁺ and CD4⁺) and activation (increased CD25 expression) of CD8⁺ T cells upon treatment with JNJ-78306358 (0.3 mg/kg) of PDX LXFA 983 tumors in human CD34⁺ NOG-EXL mice. HSCs were implanted approximately 91 days prior to tumor implantation, followed by biweekly treatment with JNJ-78306358 at 0.3 mg/kg starting on Day 21. Flow cytometry analysis was performed on leukocytes from each collected tumor on experiment day 5 and total cell count for selected population (X axis) plotted. The horizontal bar in each dataset indicates the mean value.

(D) Intratumoral CD3⁺ T cell infiltration of experiment described in B at Day 12, after administration of 4 doses of JNJ-78306358 at 0.3 mg/kg. Tumors were FFPE fixed. Representative IHC micrographs are shown for LXFA 983 tumors treated on Days 1, 4, 8 and 11, followed by tumor harvest 24 h later on Day 12, and stained for CD3.

Discussion

Challenges in treating solid tumors with T cell engagers include lack of cancer-specific antigens leading to undesirable on-target/off-tumor toxicity, poor infiltration of immune cells into tumors, and immunosuppressive tumor micro-environment (TME). To address these challenges, JNJ-78306358 targets a highly restricted tumor antigen, HLA-G, and binds CD3 with weaker affinity. JNJ-78306358 induces efficient T cell redirection and robust T cell-mediated cytotoxicity in highly translational preclinical models. Additionally, JNJ-78306358 prevents HLA-G binding to its receptors, ILT2/4, suggesting a potential immune checkpoint blockade function.

Assessment of HLA-G expression in normal tissues by IHC confirmed HLA-G expression on placenta and identified lower expression in pituitary, spleen (vascular/sinusoidal cells of red pulp and subset of mononuclear cells in the white pulp), thymus (in a subset of epithelial cells in the medulla) and nail bed (proximal nail matrix). A TCR study detected HLA-G in placenta and pituitary, but not other tissues, indicating that potential on target–off tumor activity of JNJ-78306358 would be restricted to these two organs. HLA-G was detected using the commercial 4H84 antibody since it could detect HLA-G in FFPE tissue whereas JNJ-78306358 could only detect HLA-G in frozen tissue sections. HLA-G expression can be induced in disease and/or inflammatory conditions. HLA-G induction was confirmed in pneumocytes from lung tissue adjacent to lung tumors and in lung tissue associated with asthma. Nevertheless, in vitro experiments demonstrated that IFN-γ did not induce HLA-G de novo,^35,36 and upregulated expression in cells with baseline HLA-G expression to a limited extent. Importantly, the membrane HLA-G upregulation did not result in increased T lymphocyte mediated cytotoxicity in the presence of JNJ-78306358.

While HLA-G expression is highly restricted under normal conditions, its expression in cancer is upregulated and was detected across a variety of solid tumors.^9,37,38 IHC analysis confirmed recent data³⁴ showing highest prevalence of HLA-G in renal cell carcinoma, colon, ovarian and rectal cancers, followed by lung cancer with lung adenocarcinoma having the highest prevalence, and endometrial cancer. Other solid tumors including head and neck squamous cell carcinoma (HNSCC), esophageal and pancreatic cancer showed >20% incidence of HLA-G⁺ expression. IHC staining showed correlation of HLA-G detection in PDX samples between 4H84 and JNJ-78655317 indicating concordance of HLA-G detection by 4H84 and JNJ-78655317 and applicability of 4H84 as a potential marker for patient selection.

Target shedding from either tumor cells or healthy cells may act as a soluble sink for therapeutics, and could have a detrimental impact on drug exposure.³⁹ sHLA-G was undetectable or at low levels in serum from most healthy individuals and cancer patients. JNJ-78306358 reactive sHLA-G at > 1 ng/mL was present in only a small subset of cancer patients. Therefore, sHLA-G was not a critical factor impacting the PK of JNJ-78306358.

The expression profile of HLA-G within the tumor tissues and in preclinical models was heterogeneous. JNJ-78306358 showed potent activity against tumors expressing variable levels of HLA-G and showing heterogeneous intra-tumoral expression profile, eliciting significant tumor growth inhibition, and in several models complete tumor regression in the majority of mice at doses as low as 0.03 mg/kg in both T cell- and CD34⁺-HSC-engrafted humanized mouse models. In vitro evaluation of JNJ-78306358-mediated T cell activation showed a dichotomous response with a clear threshold of HLA-G expression necessary to induce IFN-γ response in tumor-infiltrating T cells. Further, the data from in vivo studies assessing anti-tumoral activity of JNJ-78306358 across 10 PDX models indicated that HLA-G expression, while required and in general correlating with anti-tumor activity of JNJ-78306358, is not sufficient to elicit complete responses.

Recent studies have evaluated cell-based therapies for targeting HLA-G.^{34,40,41,42,43} Indeed, these studies reinforce that HLA-G is a promising antigen in solid tumors. CAR-T, CAR-NK, or γδ CAR-T cells expressing anti-HLA-G antigen receptors showed promising pre-clinical efficacy and showed potential co-upregulation with EGFR. In some cases, these cell-based therapies could induce complete responses. In this study, we uniquely evaluate the ability of a T cell redirecting bispecific antibody to redirect T cells toward tumors. JNJ-78306358 features weaker CD3 binding to alleviate the potential for cytokine release syndrome associated with sharp therapeutic indices and T cell activation induced cytokine release. BsAbs having a high-affinity T cell arm can have weaker in vivo efficacy¹⁵ as a consequence of enrichment in secondary lymphoid organs,⁴⁴ and potentially due to T cell over-activation and exhaustion. JNJ-78306358 binds a membrane proximal epitope enabling formation of a more effective T cell synapse.^45,46 Binding to the membrane proximal α3 domain, and potentially engaging both monomeric and dimeric forms of HLA-G enables JNJ-78306358 to compete with ILT2/ILT4 receptors. Immunoregulatory HLA-G/ILT2/4 signaling affects a wide spectrum of immune cells compared to those modulated by cytotoxic T lymphocyte-associated antigen (CTLA)-4/B7 and programmed death (PD)-1; known targets of current IC blockade therapies. Neutralizing the ILT2/4 checkpoint pathway may serve as an additional mechanism of JNJ-78306358 and could also have an advantage over direct inhibition of either ILT2 or ILT4. However, whether JNJ-78306358 may also exert ILT2/4 blockage at clinically relevant doses estimated from T cell mediated killing-based models remains in question. Based on specific HLA-G expression on tumor cells and its regulatory role, particularly in non-inflamed tumors, JNJ-78306358 potentially may overcome the immunosuppressive solid tumor microenvironment and enhance inflammation of tumors.

JNJ-78306358 induced potent, dose-dependent cytotoxicity against cells expressing HLA-G. The anti-tumor activity of JNJ-78306358 was dependent on recruitment of T cells, as shown by its ability to activate T cells in the presence of tumor cells. The HLA-G binding variable region is exquisitely selective for HLA-G over other MHC class I ligands, and this was considered a critical element of the antibody design due to broad expression of these ligands on normal cells. Treatment of tumors with JNJ-78306358 resulted in infiltration of T cells and anti-tumor activity was highly dependent on expression level of HLA-G in PDX models. Overall, these preclinical results supported the clinical advancement of JNJ-78306358.

Limitations of the study

There are some limitations in our study, such as the inability of JNJ-78306358 to detect HLA-G from FFPE tissue and relatively small sample sizes. We also note the limitation of limited sampling in TMA analysis.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Adam Zwolak (azwolak1@its.jnj.com).

Materials availability

Reagents generated in this study are available from the lead contact upon request with a completed Materials Transfer Agreement.

Data and code availability

• •Data: All data supporting the findings of this study are available within the main manuscript and the supplementary files.
• •Code: This paper does not report original code. The software used in this study is described in the above section and the key resources table in details.
• •All other items: Any additional information required to re-analyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was financially supported by Janssen R&D, LLC.

Author contributions

N.O., A.Z., S.S., M.V.L., and S.L. conceptualized the study. N.O., A.Z., K.R., D.W., K.T., D.A.A., T.-W.S.Y., B.G., M.v.H., K.P., J.S., J.H., L.L., J.C., R.J.B., J.L., J.G.G., and D.B. designed the methodology. N.O., A.Z., K.v.d.V., S.V., K.M., K.R., T.P., D.W., J.A., J.P., K.T., L.A., F.Y., S.J., K.S., Y.H., K.B., V.T., A.B., M.v.H., G.C., B.V., and M.O. performed experiments. N.O., A.Z., K.M., K.R., J.A., K.T., L.L., and J.C. wrote the manuscript.

Declaration of interests

All authors were employees of Johnson & Johnson Innovative Medicine at the time of this work. Some authors are listed as inventors on US-20220033505-A1.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Alexa 680 goat anti-mouse IgG Ab (secondary for 4H84)	ThermoFisher Scientific	A21057; RRID: AB_2535723
Alexa 680 goat anti-Rabbit IgG Ab	ThermoFisher Scientific	A21076; RRID: AB_2535736
Alexa Fluor 647 goat anti-Human IgG1 (H+L)	Jackson ImmunoResearch	109-605-003; RRID: AB_2337880
Anti-AlexaFluor488	ThermoFisher Scientific	A-11094; RRID: AB_221544
Anti-IgG1+IgG2a+IgG3 antibody	Abcam	ab133469; RRID: AB_2910607
Anti-Mouse HQ	Roche	7017782001
Anti-Rabbit HQ	Roche	7017812001
Anti-rabbit IgG (H&L) (secondary for anti-vinculin)	Rockland	611-732-127; RRID: AB_220158
Biotin-C36B40 (anti-CD3B376 mAb)	Janssen R&D	Lot #MOS00289506
Goat anti-hIgG1-AF647 (Fab’)2	Jackson ImmunoResearch	109-606-098; RRID: AB_2337899
Human immune globulin	Gammagard	00944-2700-05
Rabbit anti-human CD4 antibody (EPR6855)	Abcam	ab133616; RRID: AB_2750883
Rabbit anti-human CD8 (SP57)	Roche	790-4460
Mouse anti-human cytokeratin (AE1/AE3)	Leica Biosystems	PA0094; RRID: AB_1873731
SulfoTag-R10 (anti-human IgG Fc mAb)	Janssen R&D	Lot #MOS00227411
Vinculin (E1E9V) XP® Rabbit mAb	Cell Signaling Technologies	13901; RRID: AB_2728768
anti-ILT-2-APC	Biolegend	333709; RRID: AB_2136384
anti-ILT-4-APC	Biolegend	338706; RRID: AB_2136524
Chemicals, peptides, and recombinant proteins
10% Formalin	VWR	16004-112
100X NEAA	Sigma	M7145
140-Proof ethanol	Koptec	V1401TP
4% Paraformaldehyde	VWR/Alfa Aesar	J61899
ACK lysis buffer	Lonza	BP10-548E
ACK lysis buffer	Invitrogen	A10492-01
Annexin V green reagent	Incucyte Essen Bioscience	4642 Lot 18A1025-120318
Antibody diluent	Agilent Dako	S0809
APC dextramer	Immudex	DX01
BD Brilliant Stain buffer	BD	563794
Benzonase nuclease	Novagen	2927812
β-estradiol	Sigma	E2758
Biotin Free Fc block	Accurate Chemical	NB309
Blasticidin	Sigma	9/03/3513
Bluing reagent	Roche	760-2037
Bond Epitope Retrieval Solution 1	Leica Biosystems	AR9961
Bond Epitope Retrieval solution 2	Leica Biosystems	AR9640
Caspase 3/7 Apoptosis reagent green	IncuCyte Essen Bioscience	4440
CC1 buffer	Roche	6414575001
CFSE	Invitrogen	C34554
Cultrex reduced growth factor matrix	R&D Systems	3433-005-01
Dako Protein block	Agilent	X0909
Dimethyl sulfoxide	Sigma	D2650
DISCOVERY Antibody Block	Roche	760-4204
DISCOVERY CC1	Roche	950-500
DISCOVERY ULTRA Reaction buffer	Roche	950-300
DMEM	Gibco or Sigma	41965-039 or 11995-065
DPBS, Ca++, Mg++ free	Gibco	C14190-144
DPBS, no calcium, no magnesium	Sigma or Gibco	D8537 or 14190-250
Dulbecco’s phosphate-buffered saline (DPBS), 10x	Gibco	14200-075
EDTA	Invitrogen	15575-038
EMEM	Gibco	11095-080
Enzyme free dissociation buffer	Gibco	13151-014
Eosin	Roche	6544304001
Fc block (In VivoMAb anti-mouse Fc receptor) Clone 2.4G2	Bio X Cell	CUSTOM 2.4 G2
Fetal bovine serum (FBS)/Fetal calf serum (FCS)	Sigma or Biowest	A7979 or S1810-500
Formalin	VWR	VWRS9713.9937
Freestyle medium	Sigma	14571C
Full Range Rainbow Molecular Weight Marker	GE Healthcare	RPN800E
Gentamicin	Sigma or Gibco	G1397-100ML or 15750-037
Haematoxylin II	Roche	5277965001
Halt™ Protease and Phosphatase Inhibitor Cocktail	ThermoFisher Scientific	1861281
Ham’s F12	Gibco	21765-029
Heat-inactivated fetal bovine serum	Gibco	10082-147
Heat-inactivated fetal bovine serum	BioWest	S1810-500
Hematoxylin	Roche	760-2021
HEPES	Sigma	H0887
Hoechst 33342	ThermoFisher Scientific	H3570
Hydrocortisone	Sigma	H4001-1G
IMDM	Gibco	21980-032
L-glutamine	Sigma	G7513
LIVE/DEAD Fixable Near IR stain	ThermoFisher Scientific	L10119/L34976
MagicMark™ XP Western Protein Standard	Invitrogen	LC5602
MEM Alpha (1X) + GlutaMAX-1	Gibco	32561-037
MEM non-essential amino acids	Gibco	11140076
MEM with Earl’s salts and L-glutamine	Gibco	31095-029
MEM-GlutaMAX	ThermoFisher Scientific	41090101
Methanol	Sigma	32213
Moxicyte viability reagent	Orflo	MXA055
M-PER buffer	ThermoFisher Scientific	78501
MSD Read buffer T	Meso Scale Discovery	R92TC-1
MSD Wash Buffer	Meso Scale Discovery	R61AA-1
NEAA	Sigma	M7145
Neutral-buffered formalin	Surgipath	380075432AS
NuPage antioxidant	Novex ThermoFisher Scientific	NP0005
NuPage sample buffer (4×)	Invitrogen	NP0007
NuPAGE® 4-12% Bis-Tris Mini gel	Novex/Invitrogen	NP0322BOX/ WG1402BX10/ NP00335/NP00336
NuPAGE® MOPS SDS Running Buffer	Novex	NP0001
Odyssey blocking buffer	Li-COR	927-40000
PBS	Sigma	D8537
PBS (binding)	ThermoFisher Scientific	10010023
Penicillin/streptomycin	Sigma	P4458
Pierce 660nm Protein Assay Reagent	ThermoFisher Scientific	22660
Pluronic acid, 10% F-68	Gibco	24040-032
Protease Inhibitor Mini Tablets	ThermoFisher Scientific	A32955
Protein block, serum free	Agilent Dako	X0909
Puromycin	Gibco	A1113803
Quantum™ Simply Cellular® beads	Bangs Lab	anti-mouse, #815B; anti-human, #816A
Reaction buffer (10X)	Roche	950-300
Read buffer 4x	Meso Scale Discovery	R92TC-3
RIPA Buffer	ThermoFisher Scientific	89901
RPMI 1640	Gibco	61870-036
RPMI 1640	Sigma	R0883
RPMI 1640 without phenol red	Gibco	32404014
Sodium pyruvate	Sigma	S8636
Sodium pyruvate	Gibco	11360070
Sodium selenite	Sigma	S5261
Stain buffer (with FBS or BSA)	BD Pharmingen	554657 or 554656
Surfactant	Meso Scale Discovery	R92TC-2
Sytox Blue	ThermoFisher Scientific	S11348
Sytox Green Dead Cell Stain	ThermoFisher Scientific	S34860
TBST (TBS with 0.1% Tween20)	Bio-Rad	170-6531
Total protein assay module	ProteinSimple	DM-TP-01
T-PER buffer	Pierce	78510
Transfer Buffer	Novex	NP0006-1
V-PLEX® and V-PLEX Plus Proinflammatory Panel 1 (human) Kit	Meso Scale Discovery	K15049D
Wash buffer (1XDPBS, 0.05% Tween-20)	Janssen R&D	N/A
XCell II Blot Module	ThermoFisher Scientific	EI9051
Xylene	VWR	28975-325
Critical commercial assays
Anti-HQ-HRP detection kit	Roche	701793601
BCA Protein Assay Kit	ThermoFisher Scientific	23227
Bond Polymer Refine Detection Kit	Leica Biosystems	DS9800
DISCOVERY ChromoMap DAB kit	Roche	760-159
MACS Human Tumor/Xenograft Dissociation kit	Miltenyi Biotec	130-095-929
MACS Mouse Tumor Dissociation kit	Miltenyi Biotec	130-096-730
Mouse cell Depletion Kit	Miltenyi Biotec	130-104-694
OmniMap DAB anti-Rb Detection Kit	Roche	760-149
sHLA-G ELISA kit	ExBio	RD194070100R
T Cell Activation / Expansion Kit, human	Miltenyi Biotec	130-091-441
Trypan blue 0.4%	Gibco	15250
TrypLE	Gibco	12563-029
TrypLE Express	Gibco	12605036
TrypLE select	Invitrogen	12563-011
TrypLE ™ Select Enzyme (1×), no phenol red	Thermo Fisher Scientific	12563011
UltraPure LMP agarose	Invitrogen	16520-050
IL-2	Miltenyi Biotec	130-097-743
Insulin	Gibco	12585-014
Recombinant human IFN-y	R&D Systems	285-IF-100
Recombinant human IL-6	R&D Systems	206-IL-200
Recombinant human TNF-a	R&D Systems	210-TA-100
Transferrin	Sigma	T8158
Biotinylated sHLA-G (MHGW8.ECO.PP.DB.002)	Janssen R&D	NA
0.5 mL Polypropylene assay blocks	Bio-one	786261
1.2 mL Deepwell 96-Well plate Assay, polypropylene	ThermoFisher Scientific, Abgene	AB1127
10 mL Syringe	BD Biosciences	301304
15 mL Conical tube	Falcon	352097
6-well Clear flat bottom cell culture plate	Corning	353046
70-μm Strainer	Falcon	352350
96-Well Polypropylene plate	Greiner	650201 or 786261
96-Well V bottom plate	Corning	3894
96-Well Microplate, Polypropylene, U – Bottom, Chimney Style, Clear, No Lid	Greiner Bio-One	650261
96-Well E-Plates	ACEA Biosciences/Agilent	5232368001
Abgene™ 96-Well Polypropylene DeepWell Storage Plates	ThermoFisher Scientific	AB1127
Albumin standard	Pierce	23209
BD Microtainer tube (serum separator)	VWR	VT365967
Breath-easy membrane	Sigma	Z380059
Cell Strainer, 70 μm, white	Falcon	352350
Deep well storage plates, 96-well, 2.2 ml, Mark II	VWR, Abgene	732-4910
Falcon 50 ml conical tubes	ThermoFisher Scientific	12716688
Falcon 96-well U-bottom plate	Corning	353077
Falcon® Centrifuge Tubes, Polypropylene, Sterile, 50 mL Conical Tube	Corning	352070
Flat bottom tissue culture plate 96-well	Corning	3598
GentleMACS C-Tubes	Miltenyi	130-093-237
iBlot Gel Transfer Stacks, PVDF regular	Invitrogen	IB401001
Immun-Blot® PVDF membrane	Bio-Rad	162-0177/ 162-0174
Liquid coverslip	Roche	650-10
Low attachment 96-well U-bottom plate	Costar	7007
Lysing Matrix D tube	MP Biomedicals	6913-100
MSD 10-plex plates	Meso Scale Discovery	K15235N
MSD 96-Well Streptavidin Gold Plate	Meso Scale Discovery	L15SA-1
MSD Small Spot Plate	Meso Scale Discovery	L45SA-1
Nunc seal	Nunc	236366
Sterile plate covers/ Lid Universal Clear Sterile PS	ThermoFisher Scientific	250002
Sterile U-bottom 96-wells plates for flow	Falcon	351177
TC 5-layer flask, 875 mm2	Falcon	353144
V-bottom 96-well plate	Nunc	NUNC249952
V-bottom polypropylene 96 well plates	Greiner	65261
VWR Micro slides, Superfrost Plus	VWR	48311-703
Experimental models: Cell lines
HEK	ATCC	CRL-1573
K562	ATCC	CCL-243
CHO	ATCC	CCL-61
RERF-LD-Ad1	Radiation Effects Research Foundation	JCRB1020
BICR6	Sigma-Aldrich	5070501
NCI-H2009	ATCC	CRL-5911
NCI-H1975	ATCC	CRL-5908
HuP-T3	Sigma-Aldrich	93121055

AF, Alexa Fluor; APC, allophycocyanin; BCA, bicinchonic acid; BSA, bovine serum albumin; CD, cluster of differentiation; CFSE, carboxyfluorescein succinimidyl ester; DAB, diaminobenzidine; DMEM, Dulbecco’s modified Eagle medium; DPBS, Dulbecco’s phosphate-buffered saline; EMEM, Eagle’s minimal essential medium; Fab, fragment antigen binding; FBS, fetal bovine serum; Fc, fragment crystallizable; FCS, fetal calf serum; h, human; H+L, heavy plus light; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; HLA, human leukocyte antigen; Ig, immunoglobulin; IHC, immunohistochemistry; IL, interleukin; IMDM, Iscove’s modified Dulbecco’s medium; IR, infrared; mAb, monoclonal antibody; MEM, modified Eagle medium; MSD, Meso Scale Discovery; N/A, not applicable; NEAA, non-essential amino acids; PBS, phosphate-buffered saline; PK, pharmacokinetics; RIPA, radio immunoprecipitation assay; RPMI, Roswell Park Memorial Institute; s, soluble.

Experimental model and study participant details

Cell lines and primary cells

Cell lines were obtained from American Type Culture Collection, Sigma-Aldrich, Radiation Effects Research Foundation, or Deutsche SammLung von Mikroorganismen und Zellkulturen, and cultured at 37°C, 5% CO₂ (list of cell lines and culture conditions is available in the supplemental information). They were confirmed to be free of mycoplasma. Cell lines were selected based on HLA-G transcriptomics data (RNAseq) originating from the Cancer Cell Line Encyclopedia (CCLE) collection.

Primary T cells were obtained from Biological Specialty Company Hemacare. PBMCs were provided by Biobank Belgian Red Cross.

PDX tissues were obtained from Charles River Discovery Services, and PDX models for in vivo studies were established and maintained at Charles River Discovery Services. All reagents are listed in the key resources table and Tables S6–S8.

In vivo studies: Mice

Female NOD-scid IL2Rgnull (NSG), NOD.Cg-Prkdcscid Il2rgtm1Sug Tg (SV40/HTLV-IL3,CSF2)10-7Jic/JicTac (NOG-EXL), and NOD.Cg-Prkdc^scid Il2rg^tm1Wjl Tg (CMV-IL3,CSF2,KITLG)1Eav/MloySzJ (NSG-SGM3) (Jackson Laboratories, Taconic, and Charles River) mice were used when they were approximately 6-17 weeks of age and weighed approximately 25 g. PDX tissues were obtained from Charles River Discovery Services, and PDX models for in vivo studies were established and maintained at Charles River Discovery Services. All reagents are listed in the key resources table and Tables S6–S8.

In vivo studies: Tumor models and human immune cells

The human pancreatic cancer cell line HuP-T3 was obtained from Sigma-Aldrich. HuP-T3 tumor cells were grown in complete medium, harvested during exponential growth using 1× TrypLE and resuspended in 50% Cultrex / 50% cold (4°C) serum-free medium at a concentration of 5×10⁷ cells/mL. NSG mice were implanted SC with 1×10⁷ cells in 0.2 mL in the right flank. The PDX models were established and maintained at Charles River Discovery Services, Freiburg, Germany.

IHC established HLA-G expression for each tumor model. HLA-G receptor density was measured by flow cytometry for dissociated tumor cells (DTCs) obtained ex vivo from tumors grown in immunocompromised mice. HuP-T3 DTCs showed a single broad peak of HLA-G expression, with a mean of ∼8,000 receptors/cell. For PDX DTCs showing biphasic staining, positively staining peak comprising different percentage of the DTCs, was assessed for the mean of receptors/cell.

To humanize the immune system of the mice, human pan-T cells or CD34⁺ HSCs were implanted. For adoptive T cell transfer, human pan-T cells were activated and expanded using the T cell activation and expansion kit as directed (Miltenyi Biotec, # 130-091-441, with the exception that only ¼ of the recommended bead concentration was utilized. Beginning 3 days after thaw, cells were cultured with medium containing interleukin (IL)-2 at a concentration of 0.1 μg/μL. On the day of engraftment into mice, CD3 beads were removed from the T cells using magnets, and cells were resuspended in RPMI-1640 serum-free medium at a concentration of 1×10⁸ cells/mL, for an IP injection of 2×10⁷ cells in 0.2 mL per mouse. T cell-humanized mice were given fragment crystallizable (Fc) block at 0.2 mg/mouse IP and human immunoglobulin at 10 mg/mouse IP at least half an hour prior to antibody dosing, to compensate for the low immunoglobulin (Ig) environment in the immune-deficient mice. For CD34⁺ HSC humanization, NOG-EXL or NSG-SGM3 engraftment was evaluated prior to tumor implantation, and the animals were randomized into groups such that all groups had matching mean tumor volumes, while ensuring a comparable distribution among the groups of mice humanized with HSCs from each of the donors. Mice engrafted with human CD34⁺ HSCs develop multi-lineage human immune cells, including functional CD4⁺ and CD8⁺ T cells.⁴⁷

In all studies, Day 0 is the day of tumor cell or tumor fragment implantation. JNJ-78306358 was administered IV or IP twice a week, starting 1-2 days post randomization . The doses selected for JNJ-78306358 were based on in vitro potency. PK analysis demonstrated this dosing schedule provided adequate exposure (Table S9).

HuP-T3 tumors collected for IHC analysis staining (3 mice/time point/treatment) were fixed in 10% neutral buffered formalin for approximately 24 hours and then stored in 70% ethanol, sectioned, and stained for tumor and immune effector cell markers. Whole tumor tissue blocks and/or tumor TMAs were obtained for PDX. TMAs were supplied as slides.

The PK exposure and dose-response relationship of JNJ-78306358 were evaluated in serum and SC HuP-T3 tumors in the dedicated PK Study. HuP-T3 tumor bearing NSG female mice were randomized into groups of 18 on Day 8 post tumor cell implantation according to tumor volume, such that all groups had mean values of 195 mm3. Mice were humanized with T cells on the day following randomization (Day 9) and treatment was initiated on Day 10. JNJ-78306358 at 0.3 or 1 mg/kg was administered IP for a single dose. Survival bleeds via the retro-orbital route (n=3 animals per time point per group) were performed at 2 hours post dose. Mice were euthanized at 6, 24, 48, 72, 168, and 264 hours post dose for serum and tumor collection. Snap-frozen tumor samples and serum were stored at 80°C until analyzed.

HLA-G immunohistochemistry (IHC) of solid tumor tissues

Whole tumor tissue blocks and/or tumor TMAs were obtained for 15 different solid tumor types. TMAs were supplied as slides. Formalin-fixed, paraffin-embedded (FFPE) whole tumor blocks were sectioned (at 4 μm) and placed on positively charged glass slides. All slides were deparaffinized before antigen retrieval and staining.

For HLA-G IHC staining, antigen retrieval was carried out using Bond Epitope Retrieval Solution 2 for 10 minutes at 100°C. Sections were treated with 3% to 4% hydrogen peroxide to block endogenous peroxidases. Sections were then blocked with serum-free protein block and incubated with the primary antibody, ie, mouse anti-human HLA-G (4H84, diluted to 1 μg/mL), for 40 minutes at RT. Detection and counterstaining were achieved using Bond Polymer Refine Detection kit as per manufacturer’s instructions. Staining procedures for HLA-G were performed on a BOND RX autostainer (Leica Biosystems). Representative images of tumors stained for HLA G were taken at 40× magnification.

Evaluation of HLA-G IHC positivity was performed by a Janssen board-certified pathologist using conventional light microscopy. All viable tumor cells present on the tissue section were evaluated and scored. Positivity for HLA-G was defined as tumor cells showing partial or complete membranous and/or cytoplasmic staining. The proportion of tumor cells with any IHC positivity was noted as a percent tumor staining score. The most prevalent (modal) staining intensity of all positive staining tumor cells was scored on a scale of 1+ to 3+. The simplified/modal H value was calculated by multiplying the percent tumor staining score with the modal staining intensity.

HLA-G IHC of human normal tissues

FFPE human normal tissue microarray (TMA) FDA999j (US Biomax), contained 99 cores from 76 cases with core biopsy samples of 32 types of human organs. FFPE pituitary normal tissue microarray PIT502 (US Biomax), contained 50 cores from 25 cases. FFPE non-diseased human eye, pituitary and other tissues were purchased from multiple vendors. Lungs from normal and diseased (COPD or asthma) donors were purchased from the International Institute for the Advancement of Medicine (IIAM). All tissues were evaluated for quality (H&E stained section) and protein preservation (IHC for CD31, see below).

Benign tumor adjacent lung tissue was evaluated in FFPE samples (n=144) of non-small cell lung carcinoma (NSCLC). Benign tumor adjacent lung was present in 60 samples and was highly variable extending to approximately 5.5 mm from tumor margins.

4H84, an all isoform-HLA-G antibody binding the α-1 domain, was used as the IHC detection reagent for staining normal tissues and was used from various vendors due to availability. The concentration of 4H84 used in IHC staining experiments varied over time depending upon lot and source of monoclonal antibody 4H84 (Abcam/ab52455, Novus Bioscience/NB11055297). The concentrations of primary antibody used in these experiments were as follows: ab52455 was diluted at 1 μg/ml or 2 μg/ml; NB11055297 was diluted at 0.25 μg/ml, 1 μg/ml, or 2 μg/ml. Different concentrations were used to maintain standard levels of reactivity in IHC experiments across samples. Mouse IgG1 (Abcam; ab91353 or from Cell Signaling Technology; Cat#5415) was used as an isotype control and was diluted to the same concentration as the primary antibody. For IHC staining, normal tissue samples were stained using either a Ventana Discovery Ultra (Roche Diagnostics) or Bond RX (Leica Biosystems) autostainer. As IHC staining using different lots of 4H84 resulted in minor variations in staining intensity, bridging studies on control samples were performed with each lot of 4H84 antibody and on each staining run to ensure that staining intensity and instrument variation were normalized.

FFPE tissues stained on the Ventana Discovery Ultra were stained in a fully automated process, including deparaffinization and retrieval. Antigen retrieval was performed using Discovery CC1 at 95°C for 32 minutes. Samples were incubated with Discovery Inhibitor (Roche Diagnostics) to quench endogenous peroxidases for 8 minutes at room temperature. Samples were next incubated with Protein Block (Dako; Agilent) for 8 minutes at room temperature. 4H84 antibody (2 μg/ml) was applied to human tissue samples and incubated for 60 minutes at room temperature. After primary antibody incubation, the secondary antibody (OmniMap anti-Ms HRP; Roche Diagnostics) was added for 20 minutes at room temperature. Further signal amplification was achieved with the HQ Amplification Kit (Roche Diagnostics) using 8 minutes of incubation time. ChromoMap DAB (Roche Diagnostics) was used to detect the protein signal. Hematoxylin staining and bluing were performed, followed by slide dehydration in alcohols and xylene. The slides were then coverslipped and reviewed by a Janssen pathologist.

FFPE tissues stained on the Leica Bond RX were stained in a fully automated process, including deparaffinization and retrieval. Antigen retrieval was conducted using ER2 at 100°C for 10 minutes. Samples were incubated with Protein Block (Dako; Agilent) for 15 minutes at room temperature followed by treatment with Bond Peroxide Block for 5 minutes at room temperature. 4H84 antibody (0.25 μg/ml or 1 μg/ml) was applied to tissue samples and incubated for 40 minutes at room temperature. After primary antibody incubation, Bond Post Primary Antibody was added for 10 minutes at room temperature and was followed by staining with Bond Polymer for 10 minutes at room temperature. Bond Polymer Refine DAB was used to detect the protein signal. Hematoxylin staining was performed, followed by slide dehydration in alcohols and xylene. The slides were then coverslipped and reviewed by a Janssen pathologist.

CD31 IHC on normal tissues for tissue quality assessment

CD31, or PECAM-1, is an effective protein marker for examining tissue quality from formalin-fixed, paraffin-embedded (FFPE) slides due to its specificity for endothelial cells, enabling the assessment of vascular integrity and density within tissues⁵³. As CD31 is known to be preserved in archival FFPE samples in both human and non-human tissues, this marker is also useful to help ensure tissue quality in FFPE samples intended for IHC.

For CD31 staining, FFPE tissues were stained on the Ventana Discovery Ultra in a fully automated process, including deparaffinization and retrieval. Antigen retrieval was performed using Discovery CC1 at 95°C for 32 minutes. Samples were incubated with Inhibitor CM (Roche Diagnostics) to quench endogenous peroxidases for 12 minutes at room temperature. A mouse monoclonal anti-CD31 antibody (clone JC70; Roche Diagnostics; provided ready to use) was applied to human tissue samples and incubated for 32 minutes at 37°C. After primary antibody incubation, the secondary antibody (OmniMap anti-Ms HRP; Roche Diagnostics) was added for 20 minutes at room temperature. Further signal amplification was achieved with the HQ Amplification Kit (Roche Diagnostics) using 12 minutes of incubation time. ChromoMap DAB (Roche Diagnostics) was used to detect the protein signal. Hematoxylin staining and bluing were performed, followed by slide dehydration in alcohols and xylene. The slides were then coverslipped and reviewed by a Janssen pathologist.

Method details

Antibodies and antigens

Recombinant antibodies and HLA-G constructs were expressed in ExpiCHO cells (Thermo) according to manufacturer’s proctols. Antibodies were purified using protein A affinity capture (MabSelect SuRe, Cytiva) followed by ion-exchange chromatography. Antibody and antigen purity was > 95 % homogeneity as assayed by analytical size-exclusion chromatography and LC-MS.

HDX-MS epitope mapping

An HLA-G stock solution was prepared by mixing recombinant human HLA-G with DPBS, ligands JNJ-78980577, recombinant ILT2 or ILT4, respectively. The on-exchange reaction was initiated by mixing 10 μL of HLA-G stock solution with 30 μL of deuterated buffer (20 mM MES, 150 mM NaCl, pD_corr 6.4 in 95% D2O or 20 mM Tris, 150 mM NaCl, pDcorr 6.4 in 95% D2O). The reaction mixture was incubated for 15, 50, 150, 500, and 1,500 s at 3.2°C. The on-exchanged solution was quenched by the addition of 40 μL of chilled 8 M urea, 1 M TCEP, pH 3.0 (pH was adjusted with aqueous NaOH) and immediately analyzed.

A non-deuterated sample was prepared by mixing 10 μL of an HLA-G stock solution without mAbs and 30 μL of H2O. An aliquot of 40 μL of non-deuterated sample was mixed with 40-μL of 8 M urea, 1 M TCEP, pH 3.0 and immediately analyzed.

A fully deuterated sample was prepared by incubating a mixture of 22 μL of an HLA-G stock solution without mAbs and 66 μL of 100 mM TCEP in D2O at 55°C for 2 h. After the fully deuterated sample was cooled down to 23°C, a 40-μL aliquot of 8 M urea, 1 M TCEP, pH 3.0 was added to 40 uL of the fully deuterated sample and immediately analyzed.

HDX-MS analysis was performed using an automated HDx3 (LEAP Technologies, Morrisville, NC) system analogous to previously described^48,49 except the protease column was placed outside of the cold box. In the system, 75 μL of a quenched solution was passed over an immobilized pepsin/FPXIII column⁵⁰ at 600 μL/min with buffer A at room temperature. Peptic fragments were loaded onto a reverse phase trap column at 600 μL/min with buffer A and desalted for 1 min. The desalted fragments were separated by a C18 column with a linear gradient from 8% to 35% buffer B in buffer A at 100 μL/min over 20 min.

Mass spectrometric analyses were carried out using an LTQ™ Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) with the capillary temperature at 275°C, resolution 150,000, and mass range (m/z) 300 – 2,000.

BioPharma Finder 3.0 (Thermo Fisher Scientific) was used for the peptide identification of non-deuterated samples prior to the HDX experiments. HDExaminer version 2.1 (Sierra Analytics, Modesto, CA) was used to extract centroid values from the MS raw data files for the HDX experiments.

ILT-2/4 blocking

HEK cells were transduced with either lentivirus to express ILT-2-GFP or ILT-4- GFP tag using the manufacturer’s protocol (Dharmacon) and selected with blasticidin. Expression was confirmed by flow cytometry using anti-ILT-2-APC or anti-ILT-4-APC (Biolegend (cat. # 333709 or 338706). No binding to parental HEK cells was confirmed by flow cytometr. For blocking assays, APC-labeled streptavidin dextramer was bound up with biotinylated HLA-G extracellular domain according to manufacturer’s protocols. HLA-G-bound dextramer was mixed with HEK-ILT-2/4 cells and level of bound dextramer was assayed by flow cytometry in the presence of different concentrations of JNJ-78306358. Blocking % was estimated as the inverse of % bound dextramer at each concentration. Data were plotted using Graphpad Prism.

JNJ-78306358 antibody discovery and development

The anti-CD3 and anti-HLA-G variable regions were discovered by immunizing transgenic humanized rats [OmniRat (OMT™)] with recombinant CD3ε protein or recombinant HLA-G protein, respectively. The antibody features mutations of L234A, L235A, and D265S in the constant region (Fc) to abolish interaction with Fc receptors and heterodimerization is enhanced using the Zymeworks Azymetric platform mutations.^14,51 The molecule comprises an anti-CD3 Fab region on chain 1 and an anti-HLA-G scFv on chain 2.

Binding of JNJ-78306358 to cells expressing HLA proteins or CD3

K562 cells (parental, HLA-A, -B, -C and -G) were resuspended at 1×10⁶ cells/mL and CD3 KO Jurkat T cells (CD3^-) and chinese hamster ovary (CHO) and CHO-HLA E-GFP cells were resuspended at 1×10⁶ cells/mL in Roswell Park Memorial Institute (RPMI) 1640 with 10% fetal bovine serum (FBS), and left unstained. Jurkat T cells (CD3⁺) were stained with carboxyfluorescein succinimidyl ester (CFSE) according to manufacturer’s protocol. Equal numbers of stained CD3⁺ and unstained CD3^– Jurkat cells or Parental CHO and CHO-HLA-E-GFP cells were mixed for the experiment, while K562 cells were analyzed as individual cell suspensions. Cells (K562 [parental and HLA-expressing daughter cell lines], or equal volumes of CD3^+/– Jurkat cells and CHO/CHO-HLA-E-GFP) were plated at ∼50,000 cells/well in 50 μL RPMI 1640 / 10% FBS in a 96 well V-bottom plate. JNJ-78306358 was prepared at 2× final concentration, using half-log serial dilutions prepared in RPMI-1640 with 10% FBS, and resulting in a final concentration range of 60 to 0.0006 nM in the assay. An equal volume (50 μL) of antibody was added to the wells and the plates were incubated for 1 hour at 37°C. Cells were washed and resuspended in 50 μL of 2 μg/mL of secondary antibody (ie, goat anti-human Alexa Fluor [AF]647) and incubated for 30 minutes at 4°C. Cells were washed with Stain buffer and Running buffer (Stain buffer / 1 mM EDTA / 0.1% pluronic acid) and finally resuspended in 30 μL of Running buffer with 1:1000 Sytox Blue live/dead stain and stored at 4°C if not analyzed immediately. Triplicate samples were acquired on the IQue Screener Plus flow cytometer (Intellicyt). Samples were analyzed using Forecyt software (Intellicyt) to assess signal to background ratios and curves were generated using Genedata Screener (Genedata) and Prism (GraphPad, version 8.0).

Flow cytometry analysis of endogenous HLA-G expressing cell lines

Adherent cells (RERF-LC-Ad1, HuP-T3, BICR 6, NCI-H1975 and NCI-H2009 [parental and β2m-expressing NCI-H2009]) were dissociated with 5 mL TrypLE select, collected in corresponding culture medium, washed once with cold PBS and resuspended to 3×10⁶ cells/mL in cold Dulbecco’s phosphate buffered saline (DPBS). Cells were plated in 100 μL aliquots into a 96 well plate and labeled with LIVE/DEAD Near-IR stain. After incubation for 30 minutes at room temperature (RT) in the dark, cells were washed, blocked with 25 μL of Biotin Free Fc block for 15 minutes, and resuspended in 50 μL/well of stain buffer with 10 μg/mL JNJ-78980577-PE or isotype silent Fc IgG1-PE. Cells were incubated for 45 minutes at 4°C in the dark, then washed and resuspended in stain buffer. Samples were acquired on the BD LSRFortessa Cell Analyzer (equipped with a Blue 488 nm, Violet 405 nm, UV 355 nm, Red 640 nm & Yellow/Green laser 561 nm). The data was analyzed using FLOW Jo (BD Biosciences, Version 10.0.8). Geomean MFI (geoMFI) values for PE in live cell population were used to calculate the ratio of geoMFI HLA-G/geoMFI isotype control (geoMFI index).

Cytotoxicity assay with K562-HLA expressing cells

K562 cells (parental, HLA-A, -B, -C and -G) cells were prepared in Iscove’s modified Dulbecco’s medium (IMDM) / 10% FBS at 1×10⁵ cells/mL and 100 μL/well was added to 96 well U bottom polystyrene plates. JNJ-78306358 and control antibodies were prepared in 9 point, 1:3 serial dilutions in IMDM / 10% FBS as 4× stocks, ranging in final concentration from 10 nM to 1.5 pM. The antibodies were added to the assay plates at 50 μL/well. Cryopreserved pan T cells were thawed and resuspended to a concentration of 1.6×10⁶ cells/mL. Pan T cells were labeled with CD4/CD8 double APC stain allowing to identify tumor cells by CD4/CD8 gate exclusion. T cells were added at 50 μL/well (80,000 cells/well) for an effector:target (E:T) ratio of 8:1, a ratio where maximal cytotoxicity was achieved at the lowest possible E:T ratio in preliminary experiments. Tumor cell cytotoxicity was determined by flow cytometry following Sytox Green (diluted 1:1,000) viability staining and was calculated as the number of dead tumor cells per well divided by the number of tumor cells per well, expressed as a percentage. The antibody concentrations versus percent cytotoxicity were plotted in Prism (version 7.0), together with percentage of T-cell activation (CD25⁺).

In vitro cytotoxicity assay with endogenous HLA-G expressing cells

Kinetic Impedance-based, label-free T cell redirection experiments were performed with an xCELLigence Real-Time Cell Analyzer (RTCA; Agilent) to measure real-time changes in cell confluency, while T cell activation and cytokine release were evaluated in parallel assay plates by flow cytometry and Meso Scale Discovery (MSD) assay, respectively.

Seeding densities for the target cells allowed the tumor cells to remain sub-confluent and dividing within the approximate logarithmic growth phase when T cells were activated to begin killing the tumor cell targets (∼24 hours post addition of test agent). Target cells (RERF-LC-Ad1, HuP-T3, BICR 6, NCI-H1975, and NCI-H2009-β2m) were seeded on a 96 well e Plate in a volume of 100 μL/well, followed by attachment for 2 to 3 hours at 37°C. Test antibodies (JNJ-78306358, null x CD3, and HLA-G x null) were serially diluted at 4× concentration in cell culture medium and added to the wells (50 μL/well). The typical dilution scheme was a 9 point, 1:3 serial dilution, ranging in final concentration from 20 to 0.003 nM. Next, cryopreserved pan T cells were thawed and added to the plate (50 μL/well of T cells), for a total assay volume of 200 μL/well. T cell seeding density varied depending on the desired E:T ratio, which ranged from 1:1, 1:3 to 3:1 across all experiments. The e Plate was equilibrated with all components at RT for 30 minutes before returning to xCelligence RTCA for data collection. CI readings were collected at 15 minute intervals for up to 120 hours.

Identical assay plates were set up in parallel to measure T cell activation (after 72 hours) and cytokines (after 48 hours), after which the plates were centrifuged and assay supernatants (50 μL) transferred to V bottom 96 well plates, sealed, and stored at 80°C prior to cytokine measurement, while cells were washed with stain buffer containing 2 mM EDTA and resuspended in 100 μL/well of stain buffer containing anti-CD4, anti-CD8, and anti-CD25 antibodies (see supplemental information for flow cytometry antibody panel). Plates were sealed and incubated at 4°C for 1 hour, then centrifuged, washed once with stain buffer containing 2 mM EDTA, and resuspended in Sytox Green viability stain (diluted 1:1,000). Samples were incubated for 10 minutes at 4°C protected from light and then immediately analyzed on an iQue Screener Plus flow cytometer. T cell activation was measured as the number of CD25⁺ live T cells per well divided by the number of live T cells per well, expressed as a percentage. Harvested supernatants (50 μL, diluted 1:2 in assay diluent) were analyzed for 10 cytokines (interferon [IFN] γ, interleukin [IL]-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-13, tumor necrosis factor [TNF] α, IL-12p70) using the V-PLEX® and V-PLEX Plus Proinflammatory Panel 1 (human) Kit, according to the manufacturer’s protocol. Cytokine concentrations in pg/mL were calculated in Workbench software (MSD) based on values from the 7-point standard curve for each analyte. Values were graphed in Prism (version 7.0) using a 4-parameter variable slope model derived from the log10 of the molecule concentration versus the quantified cytokine level in pg/mL. Results of this experiment were modeled using a non-linear mixed effects (NLME) model.

The antibody concentrations versus either percent cytotoxicity or percent T cell activation were plotted in Prism (version 7.0) and the data was fit to a 4-parameter variable slope curve-fitting model to calculate ECx values.

Cytotoxicity was additionally assessed upon cytokine stimulation of HLA-G^neg (NCI H1975) and HLA-G⁺ (RERF-LC-Ad1) cell line. Assays were conducted with additional IFN-γ treatments, either to pre-treat tumor cells with 100 IU/mL IFN-γ for 72 hours (IFN-γ pre-stim), or to add IFN-γ concurrent with the assay (IFN-γ stim), and included standard conditions in the absence of exogeneous IFN-γ. In this case, Caspase 3/7 signal for NCI-H1975 and RERF-LC-Ad1 cells was monitored to assess cell viability. Conditions were tested in triplicate; means ± SD were graphed.

Cytokine release assay with primary renal cancer patient DTC in autologous setting

Renal dissociated tissue cells (DTC) were purchased from Discovery Life Sciences (DLS) company and consisted of 10-30% EpCAM⁺ tumor cells and 30-90% CD45⁺ immune cells, which was assessed by the vendor using flow cytometry. Frozen DTC were thawed, resuspended in assay medium and seeded in 96-well plate (30,000 DTC /well in 50 μL) overnight at 37°C, 5% CO₂. Afterwards, JNJ-78306358 was added and volume adjusted with assay medium to 200 μL per well. Plates were sealed with a breath-easy membrane and left in the incubator for 72 hours, thereafter plates were centrifugated and 100 μL of supernatants was transferred to a U-bottom 96-wells plate and stored at -80°C until use.

Supernatants (50 μL, diluted 1:2 in assay diluent) were analyzed for IFN-γ using the V-PLEX® and V-PLEX Plus Proinflammatory Panel 1 (human) Kit, according to the manufacturer’s protocol. Plates were read on a Sector Imager 6000 (MSD). Cytokine concentrations in pg/mL were calculated in Workbench software (MSD) based on values from the 7-point standard curve for each analyte. Values were graphed in Prism (version 7.0) using a 4-parameter variable slope model derived from the log10 of the molecule concentration versus the quantified cytokine level in pg/mL. Results of this experiment were modeled using a non-linear mixed effects (NLME) model.

Capillary Wes methods

Various samples (cell lines, DTCs from cancer patients, PDX tissues) were incubated in M PER lysis buffer with protease/phosphatase inhibitor [1:100] and benzonase nuclease [1:1,000]) for 15 minutes at 4°C, with vortexing every 5 to 10 minutes. Fresh frozen lung tumors (n=20), tumor-adjacent lung tissue (n=3), benign lung tissue (n=2), and healthy lung tissue (n=3) were crushed with cryoPREP (Covaris) according to the manufacturer’s protocol and lysed in T PER buffer with Halt™ Protease and Phosphatase Inhibitor Cocktail and benzonase nuclease with intermittent 3× sonication on ice at 30% power for 15 seconds. Samples were then centrifuged at 13,000 rpm for 15 minutes at 4°C and supernatant was collected and stored at 80°C for later analysis by Capillary Wes.

Protein content of each sample was determined and lysates diluted to 0.25 mg/mL to obtain a final concentration of 0.2 mg/mL. Prepared samples were loaded on the plate together with Wes reagents and 1:50 dilution of primary antibody, 4H84 (Exbio cat# 11-499-C100), according to layout in manufacturer's protocol. Total protein assay was performed in a separate plate and data served for normalization of the HLA-G signal. The plates were loaded into the Wes machine (ProteinSimple) and Size 12-230kDa assay or Total protein 12-230kDa assay was run and analyzed with Compass Software (ProteinSimple; Compass for SW software, version 5.0.1).

Receptor density evaluation

To determine receptor density, flow cytometry analysis was performed with additional analysis of antibody binding to IgG (Quantum ™ Simply Cellular®) beads. IgG beads were incubated with antibodies (at saturating concentrations), according to the manufacturer’s protocol. Beads were run at specific photomultiplier tube voltages, gated in FlowJo software, and geoMean values determined for each bead separately. Using BD Relative Linear Scale Calibration Plot macro from the Bang’s Lab (www.bangslabs.com/products/quickcal), code for the specific bead lot was accessed to generate a standard curve, which was used to determine receptor density for each cell line or HLA-G⁺ population.

Flow cytometry and Capillary Wes profiling of renal DTCs

DTCs from renal cancer patients (Discovery Life Sciences) were analyzed by flow cytometry and Capillary Wes for HLA-G protein expression. For flow cytometry analysis, cells were thawed at 37°C and resuspended in medium. Approximately 1×10⁵ to 2×10⁵ cells from each patient sample were added per well of a low-attachment 96-well plate and analyzed by flow cytometry (see supplemental information for antibody panel). HLA-G receptor density was determined using geoMean PE values and the method described in above section.

For Capillary Wes analysis, approximately 4×10⁶ DTCs were centrifuged at 400×g for 5 minutes, supernatants were removed, and cells were washed once with 1 mL of cold DPBS. Cells were lysed, and samples were stored at -20°C until Capillary Wes analysis was performed.

In vivo studies: Sample collection and preservation

Samples were collected according to all relevant animal welfare guidelines. Tumors were collected immediately after euthanasia, and directly processed (for flow cytometry analysis), transferred to fixative (FFPE samples) or directly to liquid nitrogen (snap-frozen samples). The tumor section was weighed before processing for flow cytometry analysis or snap freezing. Snap-frozen samples were stored at −80°C until shipment.

For IHC HLA-G positivity fixation was performed in 10% neutral buffered formalin for approximately 24 hours. The fixative was then replaced by submerging the samples in 70% ethanol for up to seven days. Thereafter, samples were dehydrated by sequentially incubating them in the following solutions: 70% ethanol (1 h), 80% ethanol (2 h), 100% ethanol (1 h), 100% isopropanol (1.5 h), xylene (two times: 1 h; 1.5 h). Finally, samples were embedded in paraffin.

For flow cytometry analysis, tumors were cut into 2–4 mm pieces and treated with a human tumor dissociation kit (Miltenyi Biotec, # 130-095-929) following the manufacturer’s instructions. Cells were resuspended in 1× ACK lysis buffer (150 mM ammonium chloride, 10 mM potassium bicarbonate, 0.1 mM EDTA, pH 7.2–7.4) and incubated for 1–3 min at room temperature. The cells were washed with FC buffer (2% FBS in PBS), centrifuged, resuspended in FC buffer and processed for FC analysis with 5 × 10⁵ cells per well.

In vivo studies: Flow cytometry

Cells in 96-well plate were pelleted and the supernatant was removed. Fc-block antibody (purified rat anti-mouse CD16/CD32 (2.4G2), 0.5 mg/ml BD Biosciences catalog #553142) was added to each well at 10 μl/well of a 1:100 dilution in FC buffer (2% FBS in PBS). After 5 minutes of incubation, specific cell surface antibodies (see supplemental information for the antibodies used in the flow cytometry panel) were added in Zombie Aqua Fixable Viability stain (BioLegend, catalog #423101, diluted 1:100 in PBS buffer) to each well and the plates were incubated for 30 min at 4°C protected from light. Cells were washed with FC buffer and finally, the cells were resuspended in FC buffer for analysis with the Attune NXT Acoustic Focusing Cytometer (violet (405 nm)/blue (488 nm)/yellow (561 nm)/red (638 nm) laser configuration).

For the analysis of HLA-G expression on tumor cells of PDX tissues by flow cytometry, samples were further purified using the mouse cell depletion kit from Miltenyi according to the manufacturer’s protocol and an anti-murine major histocompatibility complex (MHC) Class I antibody was used to exclude the mouse cells from the analysis (see supplemental information for antibody panel used).

In vivo studies: Immunohistochemistry of xenograft tissues

Formalin-fixed, paraffin-embedded (FFPE) whole tumor blocks were sectioned (at 4 μm) and placed on positively charged glass slides. All slides were deparaffinized before antigen retrieval and staining. Antigen retrieval was carried out using Bond Epitope Retrieval Solution 2 for 10 minutes at 100°C. Sections were treated with 3% to 4% hydrogen peroxide to block endogenous peroxidases. Sections were then blocked with serum-free protein block and incubated with the primary antibody, ie, mouse anti-human HLA-G (4H84, diluted to 1 μg/mL), for 40 minutes at RT. Detection and counterstaining were achieved using Bond Polymer Refine Detection kit as per manufacturer’s instructions. Staining procedures for HLA-G were performed on a BOND RX autostainer (Leica Biosystems). Representative images of tumors stained for HLA-G were taken at 1× magnification. Evaluation of HLA-G IHC positivity was performed by a Janssen board-certified pathologist using conventional light microscopy. All viable tumor cells present on the tissue section were evaluated and scored. Positivity for HLA-G was defined as tumor cells showing partial or complete membranous and/or cytoplasmic staining. The proportion of tumor cells with any IHC positivity was noted as a percent tumor staining score. The most prevalent (modal) staining intensity of all positive staining tumor cells was scored on a scale of 1+ to 3+. The simplified/modal H value was calculated by multiplying the percent tumor staining score with the modal staining intensity.

In vivo studies: JNJ-78306358 bioanalysis

An electro chemiluminescent (ECL) immunoassay method was used, employing the Meso Scale Discovery (MSD) 96 well streptavidin plates coated with biotinylated anti-CD3 target for capture and SulfoTag-R10Z8E9 anti-human-IgG-Fc monoclonal antibodies for detection of JNJ-78306358. The lowest quantifiable concentration was determined to be 39.06 ng/mL in serum samples, and 19.53 ng/mL in tumor lysate samples (mouse HuP-T3 xenografts). The resulting ECL signals were measured with MESO SECTOR S 600 (MSD) plate reader, and the data was analyzed using Watson LIMS™ (ThermoFisher Scientific) with a 5 parameter logistic fit standard curve with 1/y2 weighting.

Whole blood was collected in Microtainer serum-separator tubes. Samples were centrifuged at 7,500 rcf for 5 minutes, and supernatants were collected and stored at 80°C. Tumors were collected and snap-frozen in liquid nitrogen prior to processing. To process the solid tumors into a liquid homogenate, each frozen tumor was transferred into a Lysing Matrix D tube. An appropriate amount of RIPA lysis buffer containing protease inhibitor was added to each tube based on the weight of the tumor sample. Tumor samples were homogenized 3 times by OMNI Bead Ruptor 24 (Omni International) at 4 m/second for 30 seconds. Tumor lysates were transferred into new tubes and centrifuged at 14,000 rpm for 30 minutes at 4°C. The supernatant was removed and transferred to a clean tube without disturbing the tissue pellet. The protein concentrations for the tumor tissue lysates were determined using the bicinchoninic acid assay method. The tumor tissue lysate samples were normalized in lysis buffer to 5 mg/mL protein, then aliquoted and stored at 70°C.

sHLA-G detection in serum of healthy donors and cancer patients

Uncoated MULTI-ARRAY Standard plates from MSD were used to analyze soluble HLA-G in the serum samples. JNJ-78306358 and 4H84 mAb were used as capture and detection Ab, respectively. Capture antibody JNJ-78306358 was coated on the plate at 4 μg/ml overnight at 4^oC without shaking. The plate was washed and blocked. Afterwards, 25 μl of serum samples and calibrators were added to the plates and incubated for an hour on a shaker. Plates were then washed before 1 hour incubation with 4H84 mAb at 2 μg/ml, followed by its detection with 1 μg/ml with anti-mouse secondary Ab. Electro-chemiluminescence signal was generated by adding 2X Read buffer as suggested by vendor. Final analysis was performed on MSD Discovery workbench Version 4.0. Stastically significant differences were determined using t-test.

Study approval

All experiments were carried out in accordance with The Guide for the Care and Use of Laboratory Animals,⁵² the European Communities Council Directives 2010/63/EU, and the USA Animal Welfare Act, and were approved by the local ethics committees of Janssen Pharmaceuticals, Spring House, PA, and Charles River Discovery Research Services, Germany. Tumor growth in mice was monitored twice weekly and measured by a caliper. Tumor volume was calculated using the formula (D×d2/2). Studies were performed under internal IACUC approval ONC516.

Quantification and statistical analysis

Experiments were performed in triplicate, unless specifically indicated. Statistical methods are indicated in figure legends. Data were presented as mean SEM and analyzed using one way ANOVA or T-test as indicated. Statistical significance was set at p < 0.05 or 0.001 as indicated.

Published: February 4, 2025