-Specific Targeting of Tissue-Resident Gamma Delta T Cells with Recombinant Butyrophilin Heterodimeric Fusion Suresh de Silva, Anne Lai, Arpita Patel, Kinsley Evans, Kellsey Johannes, Kyung Jin Yoo, Louis Gonzalez, George Fromm, Keith Wilson & Taylor H. Schreiber Shattuck Labs, Inc. Austin, TX & Durham, NC #1736

Abstract �� TCR (TRGV and TRDV) Composition in Select TCGA Tumor Types Correlation of BTN/L and �� TCR Expression in Colorectal Adenocarcinoma

COAD d TCR Usage COAD g TCR Usage DLBCL g TCR Usage DLBCL d TCR Usage Figure 2. TCGA data 100 A. 100 B. 100 100 A. BTN3A1 B. BTN2A1 C. BTNL3 D. BTNL8 A major mechanism of acquired resistance to immune checkpoint inhibition involves downregulation of Colorectal DLBCL sets for (A) COAD BTN3A1 - aacr-2 COAD BTN2A1-aacr-2 COAD BTNL3-aacr-2 COAD BTNL8-aacr-2 1.0 1.0 1.0 1.0 80 antigen presentation, including the major histocompatibility complex (MHC I) complex itself. 80 80 80 Colorectal adeno- 0.8 0.8 0.8 0.8 carcinoma, (B) 0.6 Downregulation of antigen presentation on MHC I can render tumor cells invisible to alpha/beta 0.6 0.6 0.6 60 60 60 60 Diffuse large 0.4 (⍺�T) T cell directed therapies. Gamma/delta T cells (��T) are a small subset of the overall T cell 0.4 0.4 0.4 0.2 40 40 (DLBCL), 40 40 0.2 0.2 0.2 % TCR Usage

% TCR Usage 0.0

compartment but are characterized by increased cytolytic capacity relative to ⍺�T cells. Rather than via Spearman Correlation Spearman Correlation % TCR Usage % TCR Usage Spearman Correlation (C) Sarcoma and (D) Spearman Correlation 0.0 -0.2 0.0 0.0 MHC I, ��T recognize target cells via a complex of heterodimerized butyrophilin (BTN) proteins. Thus, 20 20 20 20 Stomach adeno- BTNL8 TRGV2TRGV3TRGV4TRGV5TRGV8TRGV9TRDV1TRDV2TRDV3 BTNL3BTNL8 TRGV2TRGV3TRGV4TRGV5TRGV8TRGV9TRDV1TRDV2TRDV3 BTNL3BTNL8 TRGV2TRGV3TRGV4TRGV5TRGV8TRGV9TRDV1TRDV2TRDV3 TRGV2TRGV3TRGV4TRGV5TRGV8TRGV9TRDV1TRDV2TRDV3 BTNL3 display of BTN heterodimers on the surface of tumor cells may enhance immunity to tumors that have BTN3A2BTN3A3BTN2A1 BTN3A1BTN3A2BTN3A3 BTN3A1BTN3A2BTN3A3BTN2A1 BTN3A1BTN3A2BTN3A3BTN2A1 0 0 0 0 carcinoma were downregulated MHC I, or which express low abundance or low affinity . analyzed for �� TCR Figure 4. Spearman correlation analysis was performed on TRGV2 TRGV3 TRGV4 TRGV5 TRGV8 TRGV9 TRDV1 TRDV2 TRDV3 the expression of select BTN’L’ proteins [(A) BTN3A1, (B) TRGV2 TRGV3 TRGV4 TRGV5 TRGV8 TRGV9 TRDV1 TRDV2 TRDV3 usage using the We have previously reported the generation of a heterodimeric BTN targeting the CD19 COAD dataset: STAD d TCR Usage SARC g TCR Usage SARC d TCR Usage DLBCL dataset:STAD g TCR Usage E. TRGV9 F. TRDV1 BTN2A1, (C) BTNL3 and (D) BTNL8], (E) gamma chain N=311 tumor (with or without matched normal) N=48 tumor (with or without matched normal) MiXCR framework. COAD vg9 - aacr -3 C. N=140 above with a TPM >0 at 2 or more delta TCRs 100 COAD TRDV1 - aacr-2 antigen, referred to as BTN2A1/3A1-Fc-CD19scFv. While the BTN2A1/3A1 complex is a potent activator 100 100 D. 100N=46 above with a TPM >0 at 2 or more delta TCRs 1.0 1.0 (TRGV9) and (F) delta chain (TRDV1) detected in the All patientSarcoma plotted have TPM>0 at 3 or more gamma TCRsCOAD dataset: All patientStomach plotted have TPM>0 at 3 or more gamma TCRsDLBCL dataset: Results revealed that of ��T expressing the �9�2 T cell (TCR), which is the major ��T population in human peripheral N=311 tumor (with or without matched normal) N=48 tumor (with or without matched normal) 0.8 0.8 TCGA data set for COAD tumor samples. While BTN3A1 80 N=14080 above with a TPM >0 at 2 or more 80delta TCRs N=4680 above with a TPM >0 at 2 or more deltaall TCRstumors expressed blood, �9�2 are not the predominant ��T in many tissues. The murine equivalent of BTN2A1/3A1-Fc- All patient plotted have TPM>0 at 3 or more gamma TCRs All patient plotted have TPM>0 at 3 or more gammamultiple TCRs V� chains 0.6 0.6 expression partially correlated with TRGV9 and TRDV1, 60 60 60 60 0.4 0.4 there was a distinct lack of correlation with BTN2A1 CD19scFv, BTNL1/6-Fc-CD19scFv, stimulated specific proliferation and increased the cytolytic capacity with broad usage 0.2 0.2

Spearman Correlation expression with these select gamma and delta chains while V� chains were Spearman Correlation of peripheral blood ��T in mice, but not other tissue-restricted ��T populations. These data suggested 40 40 40 40 0.0 0.0 % TCR Usage % TCR Usage % TCR Usage % TCR Usage which were selected based on their abundance in single that distinct BTN heterodimers may preferentially activate tissue-restricted subsets of ��T. restricted to TRDV1, TRGV2TRGV3TRGV4TRGV5TRGV8TRDV1TRDV2TRDV3 BTNL3BTNL8 TRGV2TRGV3TRGV4TRGV5TRGV8TRGV9TRDV2TRDV3 BTNL3BTNL8 20 20 20 20 BTN3A1BTN3A2BTN3A3BTN2A1 BTN3A1BTN3A2BTN2A1 cell RNA sequencing data from COAD tumor samples. To characterize potential differences between peripheral blood and tissue-restricted ��T, we performed 2 and 3 with V�1 being the most 0 0 0 0 a multi-layered analysis, including single-cell sequencing of ��T TCR and CDR3 analysis from paired prevalent across the Development of a Jurkat-76 Cell Line-Based In Vitro Assay to Screen BTN/L Pairs peripheral blood and tumor tissues from human patients with melanoma, prostate, and colon TRGV2 TRGV3 TRGV4 TRGV5 TRGV8 TRGV9 TRDV1 TRDV2 TRDV3 TRGV2 TRGV3 TRGV4 TRGV5 TRGV8 TRGV9 TRDV1 TRDV2 TRDV3 four tumor types. that Activate Specific �� TCRs Detected Using Single Cell RNA Sequencing SARC dataset: STAD dataset: cancer. These data identified tissue-specific enrichment of individual ��T TCRs, with corresponding N=254 tumor (with or without matched normal) N=383 tumor (with or without matched normal) N=126 above with a TPM >0 at 2 or more delta TCRs N=282 above with a TPM >0 at 2 or more delta TCRs Jurkat-76 �� AllTCR patient plotted Usage have TPM>0 at 3 or moreand gamma CDR3TCRsSARC dataset: Diversity inAll patient Matched plotted have TPM>0 at 3Tumor or more gamma TCRs and Peripheral Blood in tissue-specific preferences for individual BTN proteins. Based on this information, we generated a panel STAD dataset: A. B. Figure 5. (A) Jurkat-76 cell line- N=254 tumor (with or without matched normal) + N=383 tumor (with or without matched normal) Jurkat-76-Vγ9δ2 DTC_Colon_iPairN=126 above + with Mini-bulk a TPM >0 at 2 or more delta TCRs DTC_Colon_iPair + Mini-bulk 15 based reporter assays have been of distinct heterodimeric BTN proteins, and show that specific ��T populations are preferentially activated Colorectal AdenocarcinomaAll patient plotted have TPM>0 Patients at 3 or more gamma TCRs Detected by SingleN=282 above Cell with a TPM >0RNA at 2 or more deltaSequencing TCRs ** by specific BTN heterodimers in a lock-and-key fashion. These data are a prerequisite for designing ��T All patient plotted have TPM>0 at 3 or more gamma TCRs * developed to screen various A. BTN/L combinations that can specific therapeutics that may target both immune neglected and acquired resistant tumors that have 100 100 10 Colorectal Tumor Colorectal Tumor activate specific �� TCRs that are limited response to ⍺�T cell directed approaches. 80 80 5 uncovered by our single cell RNA %CD69+ cells sequencing pipeline. CD69 expression is used as a marker of 60 60 0 Methodology for Identifying �� TCR Sequences in Patient Samples �� T cell activation. (B) Representative data from one 40 no stim AgentX 40 of the Jurkat-76 reporter systems demonstrates plate-bound Percentage Usage BTN2A1+BTN3A1 I. Analysis of �� TCRs in TCGA tumors using the MiXCR framework Percentage Usage 20 20 anti-CD3+anti-CD28 BTN2A1 plus BTN3A1 can activate the V�9�2 TCR in the TCR identification was assessed using MiXCR version 2.1.6 with custom R scripts (Bolotin 2015). IMGT BTN2A1+BTN3A1BTN2A1/3A1-Fc-CD19scFv + AgentX 0 TCR repertoire sequence reference imgt.201918-4.sv5.json from http://www.imgt.org/IMGTrepertoire 0 presence of an auxiliary factor BTN2A1/3A1-Fc-CD19scFv + AgentX (AgentX). t-test - *p<0.05; **p<0.005 was used to facilitate alignment of germline TCR g and TCR d and assembly of repertoires PBMC_Colon_iPair + Mini-bulk PBMC_Colon_iPair + Mini-bulk TRDV1 TRDV2 TRDV3 TRDV4 TRDV5 TRDV6 TRDV7 TRDV8 (Lefranc 2011). Bulk RNA-seq fastq files for four TCGA tumor types (COAD, DLBC, SARC and STAD) B. TRGV2 TRGV3 TRGV4 TRGV5 TRGV8 TRGV9 The GADLEN™ Platform Allows for the Generation of Therapeutics that Target were used as input. To maximize sensitivity of capturing TCR g and TCR d alignments, the align function 100 100 in MiXCR was run with the assemblePartial and extend options. Quantification of expression was Peripheral Blood Peripheral Blood Multiple Tumor Antigens (Matched) generated using Salmon’s quasi-alignment method (Patro 2017). Gencode GrCH38 v27 CHR transcripts 80 80 (Matched) A. Gamma Delta T cell Engager B. GADLEN C. GADLEN Targets for Heme were used to build the index used for Salmon (Frankish 2018). (GADLEN) Mechanism of Action (MOA) Indications 60 60 II. Single cell sorting and sequencing of �� TCRs from dissociated tumors 40 40 Percentage Usage Percentage Usage 20 20

0 0 GADLEN scFv Binding affinity domain (recombinant protein) (KD) nM CD19scFv CD19 13.8 CD20scFv CD20 9.6 TRGV2 TRGV3 TRGV4 TRGV5 TRGV8 TRGV9 TRDV1 TRDV2 TRDV3 TRDV4 TRDV5 TRDV6 TRDV7 TRDV8 CD33scFv CD33 12.8 C. CLL1scFv CLL1 0.33 Colorectal Tumor Matched Peripheral Blood Figure 6. (A) The GADLEN design consists of a heterodimer of butyrophilin (BTN’L’) extracellular domains fused to a tumor V� CDR3 Diversity V� CDR3 Diversity V� CDR3 Diversity V� CDR3 Diversity antigen targeting scFv domain via an Fc linker. (B) The MOA of the GADLEN platform involves activation of the �� TCR through the BTN/L heterodimer domain and targeted killing of a tumor cell via the binding of a surface tumor antigen by the scFv domain. (C) Multiple GADLENs have been generated and are currently in preclinical development that target specific antigens expressed Figure 1. The above schematic describes the workflow for on lymphoma and cells (CD19, CD20, CD33, and CLL1). Binding of the select scFvs to their recombinant targets was obtaining GDT from tumor tissue for single cell RNA confirmed by MSD-based ELISA. sequencing. Freshly resected tumor tissue was mechanically dissociated using a proprietary protocol to create a single cell Summary of Findings suspension of tumor and tumor infiltrating (TIL). The ��T cells were stained with a pan �� TCR and • A combined approach of TCGA analysis using the MiXCR framework and single cell RNA sequencing was employed to better deposited into 96 well plates using fluorescence activated cell Figure 3. Single cell RNA sequencing of �� TCRs from (A) Colorectal tumors and (B) matched peripheral blood revealed a understand the V gamma chain (TRGV) and V delta chain (TRDV) usage, diversity, and the butyrophilin expression profile in sorting (FACS). The RNA was extracted from the cells and distinct difference in usage between the two compartments with tumor-derived �� T cells expressing multiple � chains with solid tumors and lymphoma. TCR g and TCR d chains were amplified using specific primer TRGV9 and TRGV4 being most abundant. Colorectal tumor derived �� T cells also expressed more TRDV1 compared to • This information is being used to screen specific BTN/L combinations and their ability to activate various �� TCR subtypes that peripheral blood �� T cells that predominantly express TRDV2. (C) CDR3 sequence analysis revealed a greater diversity in sets using a proprietary method. are resident in specific tumor tissues, which will ultimately guide the design of targeted GADLEN therapeutics. both the � and � chains in the peripheral blood compared to the tumor as depicted by the representative CDR3 tree maps • We are currently developing multiple GADLEN therapeutics that include scFv domains that target specific tumor antigens that from a single patient. Clonal expansion of �� T cells was observed in colorectal tumors when compared to peripheral blood. are highly expressed in hematologic malignancies and solid tumors.