SC134-TCB Targeting Fucosyl-GM1, a T Cell–Engaging Antibody with Potent Antitumor Activity in Preclinical Small Cell Lung Cancer Models

Small cell lung cancer (SCLC) is an aggressive neuroendocrine tumor associated with tobacco smoking, featuring rapid chemotherapy-resistant recurrence. Consequently, relapse rates are high, and 5-year survival rates are extremely low. Up to 15% of lung cancers fall within the SCLC subtype. First-line treatment for limited-stage SCLC (stages I–II), consists of surgery followed by chemotherapy, with or without radiotherapy (1). Extensive-stage (ES) SCLC (stages III and IV), often the stage at which patients present, has a much worse prognosis and is treated with a combination of platinum-based chemotherapy with or without immunotherapy (PDL1 inhibition). First-line PDL1 inhibition has somewhat improved outcomes [median overall survival (OS) improved by about 2 months], but patients still relapse; hence, there is an urgent need for more effective therapies.

T cell–redirecting bispecific (TCB) antibodies are a clinically validated treatment modality that are showing great promise, initially in hematologic cancers with CD19- or CD20-targeting TCB antibodies, but more recently also in solid tumors, in which encouraging response rates have been observed in uveal melanoma (targeting gp100) and multiple myeloma (targeting BCMA and GPRC5D; refs. 2, 3). Coengagement of tumor target and CD3 on T cells by TCB antibodies results in T-cell activation and target-dependent tumor cell killing, independent of MHC expression. This attractive mode of tumor killing necessitates the identification of tumor-specific targets to avoid off-tumor T cell–mediated toxicities.

Carbohydrate antigens are the most abundantly expressed cancer cell surface antigens (4). One such antigen is fucosyl-GM1 (FucGM1), a tumor-associated glycolipid. FucGM1 is composed of a ceramide lipid membrane component and a carbohydrate head group that is cell surface–exposed. It is expressed in 75% to 90% of SCLC cases (5) with little or no expression in normal tissues (6–12). The biosynthesis of FucGM1 involves several glycosyl transferases, some of which are overexpressed in SCLC, contributing to its progression (13–15). FucGM1 has been detected in culture media from SCLC cell lines, tumor extracts, serum of mouse xenografts, and in the serum of patients with ES-SCLC (16, 17). These reports provide convincing evidence for FucGM1 as a highly specific tumor antigen suitable for immunotherapeutic development. The SC134 mAb, generated through mixed cell and glycan conjugate immunizations, demonstrated specificity for FucGM1 and strong SCLC cell surface reactivity.

Here, we describe the first humanized anti–FucGM1 TCB, designed in an IgG-scFv “2 + 2” format, SC134-TCB. This was achieved by attaching the huOKT3-scFv to the carboxyl end of the humanized SC134 (h134) IgG1 light chain. The IgG1 backbone was further Fc-silenced to prevent Fc receptor–mediated adverse events. In vitro, SC134-TCB bound to FucGM1 and CD3 with nanomolar affinity. Ex vivo, SC134-TCB mediated picomolar target-dependent T-cell activation, proliferation, and potent SCLC cell killing. This culminated in significant in vivo DMS79 tumor control in NOD/SCIDγ (NSG) mice, resulting in 30% OS that further increased in combination with the PDL1 inhibitor atezolizumab. Critically, SC134-TCB presented little evidence for off-tumor T-cell activation. The tumor-selective nature of FucGM1 expression combined with the functional antitumor attributes of SC134-TCB suggests potential as a FucGM1-targeted TCB therapy for SCLC.

SC134 was created through classical hybridoma technology after BALB/c mice immunizations with DMS79 cells, followed by repeat FucGM1–HSA conjugates. Titers were assessed via ELISA for reactivity against purified FucGM1 (Matreya LLC) as well as for DMS79 cell surface binding. SC134 was obtained via limiting dilution cloning and isotyped as a murine IgG1 with the κ light chain. Following humanization of SC134 (h134), the SC134-TCB was created through fusion of the huOKT3-scFv (18) to the C-terminus of the light chain of h134, via a C-terminal (G4S)3 linker (19, 20). The Fc was silenced via introduction of three mutations at positions 234 to 236, the so-called “STR” methodology, recently shown to abolish all effector functions (21). Cloning, expression, and purification of the constructs were performed by evitria SA (details included in Supplementary Data).

SCLC cell lines, DMS79, DMS153, DMS53, H526, H146, and H740, were obtained and authenticated by ATCC and were cultured in RPMI 1640 (Sigma) or Waymouth MB 752/1 (Gibco) medium (DMS153) supplemented with 10% FBS (Sigma). FucGM1 expression levels were determined using the QIFIKIT (Agilent). The colorectal cell line COLO205 was purchased from ATCC and cultured in RPMI 1640 supplemented with 10% FBS. Cell lines were tested monthly for the presence of Mycoplasma and were used at low passage numbers for experiments. Human gastric fibroblast cells, obtained from ScienCell (#2830), were cultured in fibroblast medium (ScienCell, #2301). Human cardiac fibroblasts (#C-12375), human cardiac myocytes (#C-12810), normal human epithelial keratinocyte (#C-12013), normal human dermal fibroblasts (#C-12302), normal human epithelial melanocyte (#C-12413), and human aortic smooth muscle (#12533) were obtained from PromoCell. These were cultured in the respective medium recommended by the supplier. Human thymic fibroblasts (#P10490) and human thymic epithelial cells (#P10491) were purchased from Innoprot and cultured in fibroblast medium (Innoprot #P60108) and epithelial cell medium (Innoprot #P60108), respectively.

Patient-derived tumor material was sourced from Crown Bioscience International and was acquired after approval by the IntegReview Ethical Review Board. The lipid nature of FucGM1 necessitated the use of frozen tissues for IHC target distribution analysis. The staining method was validated using smeared or optimal cutting temperature (OCT)–-embedded target-positive cells (DMS79). Frozen tissue sections were air-dried and rehydrated with cold PBS. The Human-on-Human kit (HOH-3000, VectorLabs) was used for staining human tissues with huIgG1 antibodies (h134 and isotype control; #31154, Thermo Fisher Scientific) according to the manufacturer’s protocol. Thyroid transcription factor 1 was also included as a marker for pulmonary small cell carcinoma (Supplementary Fig. S1E). Briefly, after blocking of endogenous peroxidase activity with hydrogen peroxide and a subsequent protein block, primary antibodies were applied using a kit-specific diluent mixture, followed by detection with horse radish peroxidase (HRPO)–linked anti–goat IgG polymer and DAB. Hematoxylin-modified Mayer’s solution was used for counterstaining. Slides were dehydrated, mounted, and scanned on NanoZoomer-SQ scanner. QuPath software V.3 was used for biomarker scoring. Summary scores were calculated, and each tissue core or section was assigned a H-score ranging from 0 to 300, based on staining intensity and percentage positive cells.

Human peripheral blood mononuclear cells (huPBMC) were obtained from Cambridge Bioscience or isolated from healthy donors or leukopak (BioIVT) through density gradient centrifugation. Pan–T cells were enriched using the Pan T Cell Immunoaffinity Isolation Kit (Miltenyi Biotec). All donors provided written informed consent.

TCB target binding was analyzed by flow cytometry and ELISA. For flow cytometry, 1 × 10⁵ effector or SCLC cells were incubated with serially diluted TCB antibodies for 1 hour at 4°C. Following washing, bound TCB antibodies was detected using FITC-labeled anti–human IgG Fc (Sigma). Stained samples were washed and resuspended in Intracellular Fixation buffer (Invitrogen), followed by analysis on a MACSQuant 16 flow cytometer, equipped with MACSQuant software version 2.13.3, using secondary antibody–stained controls to determine suitable gates. The ability of TCB antibodies to bind CD3ED (ACROBiosystems Inc.) and FucGM1 lipid (Matreya LLC) was determined by direct ELISA. Briefly, high-binding ELISA plates were coated with 200 ng/well FucGM1 or CD3ED, followed by blocking with protein-free blocking buffer (Thermo Fisher Scientific). Primary TCB dilutions were detected using anti–human IgG (γ-chain specific)–biotin antibody (Sigma), followed by a streptavidin–HRPO conjugate.

Effector cells (1 × 10⁵, huPBMCs or pan–T cells) were cocultured with target cells and serially diluted TCB for 48 hours, at an effector to target ratio (E:T) of 5:1, after which two complementary cytotoxicity and activation assays were performed. Lactate dehydrogenase (LDH) release in the culture supernatant was measured using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega), in which maximum lysis was established by using the kit-specific cell lysis buffer. The percentage cytotoxicity was calculated using the formula (experimental – effector spontaneous – target spontaneous)/(target maximum – target spontaneous) × 100, and the target and effector cells were further analyzed by flow cytometry. The percentage of dead target cells was determined using anti-EpCAM-Alexa Fluor 647 (BioLegend) to enable target cell gating, in combination with Live/Dead Yellow stain (MACSQuant 16). Additionally, T-cell activation was determined using labeled antibodies directed to CD69, CD4, and CD8 (BioLegend). T-cell proliferation was analyzed using intracellular staining of Ki67 using directly labeled antibody (Invitrogen #12-5698-82). In some assays, catumaxomab, an EpCAM-targeting TCB was included (MedChemtronica AB, HY-P9954).

Effector cell activation after overnight incubation in the presence of SC134-TCB (E:T, 5:1 or 2:1 for primary cells) was also analyzed through ELISpot assays (ELISpot Flex) using human IFNγ (ALP) capture and detection reagents according to the manufacturer’s instructions (Mabtech). Spots were analyzed using an automated plate reader (Cellular Technologies).

Proinflammatory cytokines released in coculture medium or whole blood were detected using the Meso Scale Discovery platform (MSD) in combination with the human proinflammatory panel 1 kit. For TCB-induced cytokine detection in whole blood, serially diluted SC134-TCB and huOKT3 (10 nmol/L) positive control were incubated in whole blood of four healthy human donors for 5 or 24 hours at 37°C and 5% CO₂. Following incubation, the plate was spun at 1,000 g for 10 minutes and serum harvested. Sera taken from whole blood assays and in vivo studies were diluted 1:2 in assay buffer, and the cytokine detection was performed according to the manufacturer’s protocol (MSD). Results were analyzed using Methodical Mind version 1.2.3 and MSD Discovery Workbench software version LSR_4_0_13.

Female NSG mice (Charles River Laboratories), ages between 8 to 20 weeks old, were used for in vivo studies. Animal experiments were carried out with ethical approval from the University of Nottingham and Nottingham Trent University and under a Home Office–approved project license. Mice were randomized into treatment groups, not blinded to the investigators. For admixing experiments, 2 × 10⁶ DMS79 and huPBMCs were premixed at a ratio of 1:1 and inoculated subcutaneously in the right flank in 2.5:1 PBS and Matrigel (Corning Life Sciences), and PBMC (2 × 10⁶) were intravenously re-administered on days 8 and 15. SC134-TCB (100 μg) was dosed intraperitoneally for three consecutive days starting on day 2, and this cycle was repeated on days 9 and 16.

In studies in which tumor and PBMCs were dosed separately, mice were inoculated (subcutaneously) in the right flank, on day 1 with 2 × 10⁶ DMS79 cells, in 1:1 PBS and Matrigel. huPBMCs were administered (iv) on days 3 and 18. SC134-TCB was dosed (100 μg, i.p.) biweekly for 3 weeks, starting on day 4. A tail bleed was taken 24 hours after the initial SC134-TCB dose. Tumor growth was monitored biweekly, and following euthanasia, the terminal blood, tumor, and spleen were taken for further analysis.

For tumor-infiltrated T cells (Tils) analysis, tumors were disaggregated and effector cells were labeled with CD4 efluor 450 (RPA-T4) and CD8 efluor 450 (RPA-T8; Miltenyi Biotec), anti–PD1 APC-Cy7 (BioLegend), and anti–TIM3 PE-Cy7 (BioLegend). For FucGM1 expression, tumor cells were labeled with anti–EpCAM AFG47 (BioLegend) and h134 plus antihuman IgG FITC secondary antibodies (Sigma). Spleens were harvested, and splenocytes were isolated and stained with relevant T-cell antibodies, as above. The presence of TCB was evaluated (tumor and spleen) using an anti–human IgG (Fc-specific) FITC secondary antibody (Sigma). Sections from formalin-fixed, paraffin-embedded–processed tumor samples were stained for the presence of CD4 and CD8 T cells using rabbit anti–human CD8 (05937248001, Roche) and rabbit anti–human CD4 (AB213215, Abcam) and for tumur cells using rabbit anti–human thyroid transcription factor 1 (Abcam, ab133737) and rabbit isotype control (172730, Abcam).

Female NSG mice (Charles River Laboratories) were injected with 100 µg of SC134-TCB and serially bled over 1 to 144 hours. Sandwich ELISA was used to determine TCB serum concentration. Briefly, wells were coated with anti–human IgG capture antibody (Sigma), followed by blocking and incubation with serum samples diluted 1:100 and 1:1,000. Captured TCB was detected via incubation with anti–light chain HRPO-labeled antibody (Sigma). Following incubation, plates were developed with TMB Core+ (Bio-Rad) and read at 490 nm using Tecan Infinite M Plex plate reader. The absorbance of known SC134-TCB concentrations was used to generate the standard curve against which measured values were interpolated (GraphPad Prism software V.10).

Statistical analyses were carried out using GraphPad Prism software V.10. For all statistical tests, a P value of <0.05 was used to ascribe statistical significance. All error bars denote the SD, unless otherwise noted.

Raw data for this study were generated at Scancell Ltd. Derived data supporting the findings of this study are available from the corresponding author upon request.

Hybridoma-derived SC134 is a murine IgG1 isotype with the κ light chain, with variable regions belonging to mouse heavy-chain IGHV3-1*02 F (IGHJ4*01 F and IGHD5-1*01 P) and light-chain IGKV5-39*01 (FIGKJ1*01 F). FucGM1 binding specificity of SC134 was ascertained via glycolipid ELISA across a panel of commercially available glycolipids (Fig. 1A; Supplementary Fig. S1A) as well as using thin-layer chromatography of DMS79 glycolipid extract and commercially purified FucGM1, with detection by immunoblotting using SC134 (Fig. 1B). Across both techniques, SC134 specifically targeted FucGM1 with no cross-reactivity to GM1, nor to closely related gangliosides. SC134 bound FucGM1 with nanomolar efficiency on ELISA (Fig. 1C), as well as for cell surface binding, across a range of SCLC cell lines, including DMS79, DMS153, and DMS53 (Fig. 1D). Additionally, preincubation of SC134 with FucGM1 blocked DMS79 cell surface binding (Supplementary Fig. S1B), further supporting the binding specificity of SC134. Expression levels of FucGM1 in a subset of SCLC cell lines ranged from ∼500,000 specific antibody–binding capacity for DMS79 and DMS153 to ∼100,000 for DMS53 (Supplementary Fig. S1C).

In order to facilitate clinical translation, a humanized lead candidate, h134, a human IgG1 based on SC134, was created. h134, similarly to parental chimeric SC134 bound FucGM1 with an EC₅₀ of 0.07 nmol/L, with no reactivity toward GM1, thus maintaining FucGM1 specificity (Fig. 1E). Crucially, h134 bound a range of SCLC cell lines, with DMS79 showing the highest binding levels, followed by H740 and DMS153 (Fig. 1F) and with EC₅₀ values ranging from 14 nmol/L (H740) to 41 nmol/L (DMS79) and 42 nmol/L (DMS153). SCLC is rarely resected, and consequently, we relied on frozen patient-derived xenograft tissues of SCLC (Crown Bioscience UK), in which immunostaining with h134 was used to evaluate tumor FucGM1 expression. FucGM1 expression was evident in ∼85% (11/13) of the cases with 23% strong, 31% moderate, and 31% weak staining observed, with >60% of neoplastic cells staining positive for FucGM1 in the positive tissues (Fig. 1G). Laboratory-grown DMS79 (Fig. 1H; H-score of 87) and in vivo grown DMS79 xenograft tissues (Fig. 1I; H-scores of 131 ± 16) approximated or matched the H-scores of a large fraction of the patient-derived xenograft tissues, indicating that DMS79 is a valid model for SC134-TCB efficacy evaluation. In sharp contrast, across a FDA-approved multihealthy human tissue array, only three tissues exhibited isolated focal staining in a donor-dependent manner: pituitary (1/3), skin (2/3), and thymus (3/3; Fig. 1J), confirming the tumor specificity of FucGM1 expression. This prevalent SCLC tumor FucGM1 expression, together with its absence in most normal tissues (Supplementary Fig. S1D), supported the advancement of SC134 as a bispecific antibody for T-cell redirection.

SC134-TCB was created on the basis of a native human IgG1 antibody with the CD3-targeting scFv fused via a flexible linker to the light chain, thus enabling bivalent FucGM1 binding (Fig. 2A). Additionally, the Fc was rendered nonfunctional via the introduction of the “short tandem repeat” mutations at positions 234 to 236 that abolish all effector functions (21). In cell-based assays, SC134-TCB, like h134, exhibited avid DMS79 cell surface binding with an EC₅₀ of 25.1 and 9.7 nmol/L, respectively, whereas the nontargeting B12-TCB did not show any DMS79 binding (Fig. 2B; Supplementary Table S1). FucGM1 binding was also observed on DMS153 and H740, with an EC₅₀ of 33.9 and 17.8 nmol/L, respectively (Fig. 2,C; Supplementary Table S1). No SC134-TCB binding was observed on target-negative cell lines (Supplementary Fig. S2). The CD3-targeting scFv was derived from the humanized OKT3 sequence (18) and fused to the C-terminus of the h134 light chain through a 15-residue flexible linker, an approach that previously has been shown to render the anti-CD3 portion “functionally monovalent” (19). This was evident from the pan–T-cell binding in which the parental huOKT3 bound T cells with 0.3 nmol/L EC₅₀, compared with the EC₅₀ of 29.2 and 32.5 nmol/L for SC134-TCB and B12-TCB, respectively (Fig. 2D; Supplementary Table S1).

SC134-TCB induced T cell–redirected lysis in a donor-dependent manner with subnanomolar to picomolar efficiency, using pan–T cells as effectors and DMS79 as target at an E:T ratio of 5:1. The median EC₅₀ was 190 and 70 pmol/L for LDH and flow-based killing, respectively (Fig. 3A). The percentage maximum killing varied depending on the donor and assay used, with up to 50% (LDH) and 60% (flow-based killing) observed (Fig. 3B). Increasing the E:T ratio further increased maximum killing up to 100% (LDH release, Fig. 3C), coinciding with a slightly raised background from the nontargeting control B12-TCB. Additionally, SC134-TCB demonstrated superior killing efficiency compared with an EpCAM-targeting control TCB (Supplementary Fig. S3C). Overall, both assays showed good concordance, with flow-based killing being more sensitive than LDH release. SC134-TCB–mediated SCLC killing was target-dependent as the nontargeting control B12-TCB showed no killing using LDH release or much reduced, donor-dependent killing (flow-based analysis), likely reflecting the more sensitive nature of the latter assay (Supplementary Fig. S3A). Additionally, SC134-TCB exhibited no killing of the target-negative H146 cell line (Supplementary Fig. S3B). SC134-TCB also induced killing on the more moderately FucGM1-expressing DMS153, with comparable efficiency but reduced maximal killing (Fig. 3D). SC134-TCB–mediated cytotoxicity by pan–T cell and huPBMC was largely matched, with subtle differences depending on the assay (Supplementary Fig. S4).

SC134-TCB exhibited picomolar T-cell activation (based on CD69 expression) in a donor-dependent manner on coincubation with DMS79 or DMS53 (Fig. 3E) with maximum activation ranging from 20% to nearly 90% for coincubation with DMS79. T-cell activation was strictly target-dependent as no activation was evident with the nontargeting B12-TCB (Fig. 3F) nor in cocultures with target-negative human umbilical vein endothelial cells (Supplementary Fig. S5). Activation of both CD4 and CD8 T cells was achieved with picomolar efficiency (Fig. 3G). Target-dependent T-cell proliferation, based on Ki67 staining, was also evident for both CD4 and CD8 T cells, with picomolar efficiency over a 3-day coincubation with DMS79, with EC₅₀ values for CD8 T-cell proliferation being lower than that of CD4 T cells, at 13.4 and 95.7 pmol/L, respectively (Fig. 3H). SC134-TCB–mediated T-cell activation triggered multifunctional cytokine production that was comparable between pan–T cells and huPBMCs (Fig. 3I; Supplementary Fig. S6A). IFNγ was the dominant cytokine across both effector types and was produced with picomolar efficiency (Supplementary Fig. S6B). Cytokine production by the nontargeting B12-TCB remained low in comparison, indicating the strict target dependency of the T-cell activation (Supplementary Fig. S6B).

The efficient SC134-TCB in vitro SCLC cell killing encouraged us to validate its activity on in vivo tumor growth. The dosing schedule was selected on the basis of the observed plasma clearance profile in a single-dose study (100 μg, i.v.) in non–tumor-bearing NSG mice, in which plasma concentration of SC134-TCB remained stable for up to 6 days (Supplementary Fig. S7). The in vivo impact of SC134-TCB was first tested in NSG mice inoculated with admixed huPBMCs and DMS79 cells at a 1:1 ratio. In this setting, effector cells alone had a significant impact on DMS79 growth (Supplementary Fig. S8), resulting in 60% tumor-free survival (Fig. 4A). Biweekly administration of SC134-TCB significantly improved this outcome, resulting in 100% tumor-free survival. Importantly, SC134-TCB maintained its antitumor effect when huPBMCs were administered, after inoculation of tumor cells, on day 3. Biweekly administration of SC134-TCB in this setting led to a 45% (P = 0.013) tumor growth inhibition compared with the tumor plus effector control group (Fig. 4B), including one strongly regressing tumor by day 36 (Supplementary Fig. S9A). This coincided with a significant improvement in OS (P = 0.0042) in the SC134-TCB–treated group in which 40% OS was seen compared with 10% in the control group (Fig. 4C). Terminal tumor analysis revealed the presence of tumor-infiltrated T cells (Tils) in both the control and the SC134-TCB–treated group, ranging from 0.4% to 3.3% in the control group and 0.2% to 11.9% in the SC134-TCB–treated group. Increased levels of exhaustion, based on PD1 and Tim3 coexpression, were apparent in the SC134-TCB treatment group compared with the control group; this however, did not reach significance (Fig. 4D). Critically, in the SC134-TCB–treated mice, significantly more SC134-TCB could be detected in the terminal tumor samples compared with the near absence of SC134-TCB in the spleens (Fig. 4E), indicating that the majority of dosed SC134-TCB successfully targets the tumors. A condensed regimen follow-up study was performed with a single administration of huPBMCs and earlier tumor harvest, 3 days after the last TCB dose, when tumors showed active regression, to capture the mode of action of SC134-TCB. SC134-TCB led to a 64% tumor growth inhibition, P = 0.0249 (Fig. 4F), concomitant with a significantly increased CD4⁺ as well as CD8⁺ T-cell content (Fig. 4E; Supplementary Fig. S9D). The infiltrated T cells showed an activated, effector phenotype (PD1⁺ and CXCR3⁺, respectively; Fig. 4H and I).

There is ample evidence that the PDL1:PD1 checkpoint limits T-cell activity (22, 23). Several observations in this study prompted us to evaluate the impact of PDL1 inhibition in our NSG DMS79 tumor model. First, up to 50% PDL1-positive DMS79 cells were observed after 48-hour ex vivo SC134-TCB treatment in huPBMC cocultures (Fig. 5A). Additionally, in vivo, SC134-TCB treatment induced a significant increase in PDL1-positive DMS79 (Fig. 5B). Finally, Tils recovered at termination from SC134-TCB treatment groups displayed increased levels of exhaustion (Fig. 4D). Importantly, atezolizumab in combination with chemotherapy is now the standard of care for the treatment of ES-SCLC (24). Accordingly, it was important to evaluate whether atezolizumab-driven PDL1 inhibition could further improve the SC134-TCB antitumor activity. SC134-TCB, in the absence of PDL1 inhibition, exerted a significant tumor growth inhibition compared with the control group, confirming its impact across several donor effector cells (Fig. 5C; Supplementary Fig. S10). Biweekly administration of SC134-TCB led to a 66% (P < 0.0001) tumor growth inhibition compared with the tumor plus effector control group (Fig. 5C), resulting in 30% OS (P = 0.0054, Fig. 5D). Atezolizumab (200 μg, dosed biweekly) resulted in an initial slowing of the DMS79 growth in combination with the administered huPBMC, which recovered after day 25, resulting in a small but significant survival benefit (Fig. 5C and D). Notably, the addition of atezolizumab to the SC134-TCB regimen further improved the OS to 50%, a significant enhancement compared with the atezolizumab treatment group (P = 0.0031). Terminal tumor analysis revealed a trend toward increased Til recovery in the SC134-TCB–treated and atezolizumab combination groups, not reaching significance (Fig. 5E). Til exhaustion levels based on PD1 and Tim3 coexpression, however, were significantly reduced in the atezolizumab combination group (Fig. 5F), suggesting that PDL1 blockade may have reduced terminal exhaustion in the Tils.

In order to further refine the in vivo SC134-TCB dose response, we next evaluated a lower dose (10 μg) biweekly administered as well as a weekly dosing of 100 μg (Fig. 5G; Supplementary Fig. S11). Reducing the dosing frequency to weekly dosing maintained significant tumor growth inhibition (Fig. 5G) but led to a reduction in OS to 30% compared with 70% for biweekly dosing (Fig. 5H). Maintaining the biweekly regimen, but lowering the dose to 10 μg, seemed to initially produce superior tumor control, but this did not reach significance, compared with 100 μg, with overall similar survival being 70% for both groups.

Cytokine release syndrome (CRS) is a potentially life-threatening complication that can develop following the administration of an immunotherapy agent, causing a rapid release of proinflammatory cytokines into the blood from activated immune cells. The ability of SC134-TCB to induce the release of proinflammatory cytokines in the absence of target engagement was tested in whole blood from four healthy donors. Representative data for one donor are shown in Fig. 6A. SC134-TCB, over the concentration range tested, triggered minimal release (up to 24 hours) of IL2, IL6, TNFα, and IFNγ, all remaining below 10 pg/mL. In contrast, huOKT3 control showed robust cytokine production across all four cytokines. Cytokine production was also analyzed in mice sera, 24 hours after pan–T adoptive transfer and the first SC134-TCB dose, in an antitumor study (Fig. 6B). A small SC134-TCB–induced increase in IFNγ and IL2 was detectable, potentially already reflecting tumor target–mediated cytokine production. The other cytokines remained near the limit of detection.

To predict any potential off-tumor toxicity in the clinic, the activity of SC134-TCB was measured in the presence of normal primary cells derived from different human tissues, including two tissues where low amounts of FucGM1 were detected: thymus (3/3) and skin (2/3; pituitary cells were not available; Fig. 1J). Responses to thymic epithelial cells and thymic fibroblasts were very low (<10 spots) at all SC134-TCB concentrations tested, and the same was true for epithelial keratinocytes, dermal fibroblasts, and epidermal melanocytes with <20 spots on ELISpot (Fig. 6C). In contrast, SC134-TCB in combination with the target-positive DMS79 showed robust IFNγ induction. Next, we assessed the reactivity of SC134-TCB against a small panel of high-risk tissues, in which any reactivity to these cells could lead to fatal toxicity in the clinic. No responses were observed against cardiac myocytes, cardiac fibroblasts, aortic smooth muscle, and gastric fibroblasts at any SC134-TCB concentration (Fig. 6D).

Effective treatment options for patients with ES-SCLC remain limited despite the addition of PDL1 inhibitors to chemotherapy as a first-line strategy. Consequently, SCLC selective targets carry great value for the development of targeted therapies. The virtual absence of the FucGM1 glycolipid expression in normal healthy tissues, along with its low donor-dependent expression observed in the skin, pituitary, and thymus, combined with its extensive tumor expression, with more than 80% of positive patient-derived SCLC tissues, renders it an extremely attractive target for the development of a T cell–redirecting modality. Early-phase clinical results have already demonstrated that tumor selectivity of the target antigen is key to ensure a therapeutic window with T cell–engaging modalities (25).

Hybridoma-derived SC134 specifically targeted FucGM1 with no cross-reactivity to GM1, nor closely related gangliosides. SC134-TCB, based on h134, showed avid FucGM1 binding, due to its native antibody format, combined with nanomolar T-cell targeting, substantially weaker than the parental huOKT3 mAb, confirming the functional monovalency of the anti-huCD3 linkage (19). Coengagement of cellular FucGM1 and CD3 resulted in potent (subnanomolar to picomolar) ex vivo SCLC cell killing through the activation of pan–T cells or huPBMC from a range of donors and ensuing redirected cytotoxicity. SC134-TCB induced target-dependent activation and proliferation of both CD4 and CD8 T cells, coinciding with multifunctional cytokine production, encompassing IFNγ, IL2, and TNFα. Fc silencing of our full antibody-based format ensured that in the absence of target engagement, limited or no T-cell activation nor cytokine production was observed.

In vivo, using huPBMC-administered NSG mice, SC134-TCB treatment led to significant tumor growth inhibition, concomitant with a significant survival benefit. Previously, it has been postulated that CD3 affinity tuning in TCB formats aids biodistribution in preventing diversion of the TCB to the secondary lymphoid organs such as the spleen (26–28). Critically, SC134-TCB was robustly detected in tumors of treated mice, in contrast to the spleen where negligible TCB was detected, indicating effective tumor targeting. This was also evident from the shorter-duration antitumor study in which SC134-TCB induced a pronounced tumor T-cell infiltration, displaying an activated, effector phenotype. Although SC134-TCB induced 40% OS, terminal tumor analysis revealed increased levels of PD1/Tim3 double-positive Tils, markers previously identified in chronic viral infection models and representing T-cell exhaustion (29, 30). PDL1 inhibition is now standard of care in combination with chemotherapy for ES-SCLC. Additionally, terminal sample analysis revealed TCB-induced effector cell PD1 expression, combined with increased PDL1 expression by the tumor cells, suggesting an adaptive immune resistance mechanism, also observed in other preclinical TCB studies (31–33), and prompting us to evaluate the impact of blocking PDL1 in vivo. Combining the PDL1 inhibitor atezolizumab with SC134-TCB led to a significant increased OS, compared with the control group, in addition to a significant reduction in PD1/Tim3 double-positive Tils, indicating that PDL1 blockade improved Til exhaustion and the overall outcome of the study. The more efficacious TCB antitumor activity in combination with blockade of the PD1/PDL1 axis has also been observed across a number of other solid tumor-targeting TCB such as for the B7H6 and carcinoembryonic antigen T-cell engagers (34–36). Finally, reduced SC134-TCB dosing maintained the antitumor impact, with seemingly stronger early tumor growth inhibition, indicating that the reduced TCB-mediated T-cell activation retains effective tumor killing. Notwithstanding, in order to maintain long-term tumor control, strategies to impart immune memory formation or the in situ generation of additional effector cells are warranted.

Historically, stand-alone solid tumor–targeting antibodies have had limited impact. BMS-986012 is a clinical-stage, non-fucosylated, FucGM1-targeting antibody currently trialed in combination with nivolumab for the treatment of ES-SCLC (37). Two additional SCLC targets in clinical development are delta-like ligand 3 (DLL3), an inhibitor of Notch 1 ligand and B7H3 (CD276), an immunoregulatory protein member of the B7 family. Targeted drug delivery (antibody–drug conjugate) is currently being evaluated for the latter target (38), although the increase and decrease in the DLL3-targeted antibody–drug conjugate rovalpituzumab tesirine serve as a reminder that clinical success is not guaranteed. More recently, a B7H3-targeting TCB has been developed, equally comprising a “2 + 2” format, but with the anti-CD3 arm fused to the C-terminus of the antibody heavy chain (39). DLL3, on the other hand, has been at the forefront of many targeted therapies for SCLC, largely due to its tumor specificity, in which its cytoplasmic and cell surface expression contrast with its largely intracellular and restricted normal tissue expression (40, 41) Targeting DLL3, however, may pose a subtype-specific vulnerability as its expression may be preferentially associated with the SCLC A subtype because it is a direct transcriptional target of ASCL1 (42). Nevertheless, a plethora of DLL3-targeting TCB are in clinical development, notably AMG757 (tarlatamab; ref. 43), BI 764532 (31), and HPN238 (44). Response rates from DeLLphi-301, in which tarlatamab was evaluated in patients with ES-SCLC that had progressed on therapy, were significantly better than those from standard care, suggesting that SCLC is a setting in which T-cell engaging modalities can offer benefit (45). However, toxicities, such as CRS, were evident in about half of patients, combined with less frequent, immune effector cell–associated neurotoxicity syndrome. CRS is commonly observed with TCB both in solid and liquid tumors and usually effectively managed with supportive care (46). Nevertheless, whole-blood assays in which SC134-TCB was incubated for up to 24 hours with whole blood from four healthy donors presented little evidence for CRS, likely a result from rendering the Fc portion of our format inert (19) combined with the absence of target. The potential for off-tumor toxicity was evaluated using primary cells derived from low-level FucGM1-expressing tissues (detected via IHC) as well as from critical organs, in combination with SC134-TCB and PBMCs. The absence of significant reactivity to the thymus and skin where low-level FucGM1 expression had been detected, in combination with the absence of reactivity to the small panel of high-risk tissues, suggests that SC134-TCB may not carry inherent toxicity liabilities, but a more extensive analysis is warranted.

In hematologic tumors, T-cell bispecific modalities have proven their merit, with several recent approvals (47). Solid tumors, however, are a more challenging setting, not least due to the need for tumor-specific targets to ensure a reasonable therapeutic window, in spite of TCB not relying on MHC expression for T cell–mediated target killing (48). Whether TCB impact is driven by the baseline T-cell infiltrate or whether external T-cell recruitment is necessary and sufficient for antitumor activity seems to be governed by a number of factors, such as the amount and quality of the resident Tils, the TCB target density and homogeneity, and the TCB dose (49, 50). Nonetheless, as higher levels of effectors are instrumental to better outcomes, strategies to increase their frequency, for instance, through tumor-specific vaccination and/or tumor vasculature targeting are actively being explored (39, 51). Additionally, combination strategies to further enhance effector T-cell expansion such as through CD28-driven or 4-1BB–driven costimulation, modulation of the regulatory T-cell compartment via anti-CTLA4, or checkpoint inhibition through PD1/PDL1 blockade have all preclinically demonstrated further improvements in antitumor effects of TCB (50). Validation of these TCB combination strategies in the clinic is eagerly awaited.

SC134-TCB, targeting a supremely SCLC-selective glycolipid, FucGM1, represents an attractive development candidate as an additional T cell–redirecting therapy option for patients with SCLC.

All authors except Z. Dunster report employment with Scancell. L. Durrant reports other support from Scancell during the conduct of the study, as well as a patent for WO2021/043810A1 pending to Scancell. No disclosures were reported by Z. Dunster.

F. Dave: Formal analysis, investigation, methodology, writing–review and editing. P. Vaghela: Formal analysis, investigation, methodology, writing–original draft. B. Heath: Formal analysis, investigation, methodology. Z. Dunster: Investigation, methodology. E. Dubinina: Methodology. D. Thakker: Formal analysis, methodology. K. Mann: Resources. J. Chadwick: Formal analysis, methodology. G. Cane: Formal analysis, methodology. B.G. Kaira: Formal analysis, methodology. O.J. Mohammed: Formal analysis, methodology. R. Choudhury: Formal analysis, methodology. S. Paston: Supervision, validation. T. Parsons: Resources, validation, writing–review and editing. M. Vankemmelbeke: Conceptualization, supervision, validation, writing–original draft, writing–review and editing. L. Durrant: Conceptualization, supervision, funding acquisition, validation, writing–review and editing.