Rapid identification and evaluation of neoantigen-reactive T cell receptors from single cells.

Abstract

Engineered T cells expressing tumor-specific T cell receptors (TCRs) are emerging as a mode of personalized cancer immunotherapy that requires identification of TCRs against the products of known driver mutations and novel mutations in a timely fashion. We present a non-viral and non-next generation sequencing (NGS) platform for rapid, and efficient neoantigen-specific TCR identification and evaluation that doesn’t require the use of recombinant cloning techniques. The platform includes an innovative method of TCR alpha (TCRα) detection using Sanger sequencing, TCR pairings and the use of TCRα/β gene fragments for putative TCR evaluation. Using patients’ samples, we validated and compared our new methods head-to-head with conventional approaches used for TCR discovery. Development of a unique de-multiplexing method for identification of TCRα, adaptation of synthetic TCRs for gene transfer, and a reliable reporter system significantly shortens TCR discovery time over conventional methods and increases throughput to facilitate testing prospective personalized TCRs for adoptive cell therapy (ACT).

INTRODUCTION

Tumor infiltrated lymphocytes (TILs), composed predominantly of CD4 and CD8 cells, can recognize neoantigens arising from somatic mutations that are presented by tumor cells in a major histocompatibility (MHC)-restricted manner^{1, 2}. Adoptive cell transfer (ACT) immunotherapy using autologous TILs can mediate durable cancer regression in patients with metastatic melanoma^{3, 4} and selected epithelial cancers^5–8. However, the success of ACT using TIL for the widespread treatment of epithelial cancers has been limited by the low frequency of mutation reactive cells (usually 0.01 to 0.001 % of TIL) and the predominance of differentiated effector memory cells with limited proliferative potential. Redirecting T cell specificity by transferring tumor-specific TCRs into autologous PBL provides an attractive alternative source of cells for therapy to overcome those problems^{1, 5, 9}. A rapid method of identifying neoantigen-TCRs would thus be of value in developing this approach to cancer treatment.

Exomic and RNA-seq approaches have been developed to identify the relatively rare T cells that recognize neoantigens expressed in tumors^{1, 2, 10, 11}. One approach that has been used to identify neoantigen-reactive T cells involves the screening of TIL cultures from different resected tumors from the same patient or different parts of the same metastatic tumor¹². Whole exome sequences using those tumor specimens and normal tissue are performed to identify the somatic mutations present in the cancer. In this approach, all the mutant gene products are expressed on the patient’s own antigen presenting cells using either tandem mini genes (TMGs) or peptides and cocultured with TIL specimens (grown in 24-well plate) to identify tumor reactive lymphocytes. The reactivities to mutations are assessed by immunological assays combining functional marker such as Interferon-γ (IFNγ secretion measured by ELISPOT) and T cell activation markers (e.g. 4–1BB upregulation measured by flow cytometry)¹².

Multiple single-cell(sc)-sorting approaches have been used to identify neoantigen-specific TCRs at the single-cell level^13–16. Following the sorting of reactive T cells, several approaches have been employed to sequence TCRα and TCRβ chains, including the use of high throughput single cell analysis systems involving the barcoding of individual cells, preparation of single cell cDNA libraries, and next generation sequencing (NGS) of the resulting libraries^{13, 17}. Han et al¹⁸ combined multiplex PCR from single cell and PCR-based single cell barcoding to pool all samples to sequence them by NGS. Because of massive parallel sequencing abilities, NGS methods are well suited to process single cell transcriptomes from large number of single cell libraries including TCR sequences from rare clonotypes that may be missed in a low resolution sequencing approach such as 5’RACE or 96 well sorting followed by Sanger sequencing.

Single-cell PCR platforms followed by Sanger sequencing represent relatively rapid and cost-effective methods, particularly when multiplexing PCR primers can amplify a single TCR CDR3 region of a specific subfamily^19–21 of native TCRα /TCRβ pairs of sorted T cells. The typical scPCR protocol includes sorting of neoantigen reactive T cells in a 96 well plates²⁰, and PCR amplification of TCRα and TCRβ chains separately followed by Sanger sequencing (Fig. 1A). A problem with this approach has been its failure to sequence PCRα amplicons since up to 30 % of T cells express two distinctly different alpha chains¹⁸. The fact that T cells may express two different alpha chains owes itself to the process of TCR recombination. The TCRα locus on both chromosomes can simultaneously undergo recombination producing either two productive or one productive plus one non-productive TCR^{22, 23}. Mechanisms of allelic exclusion usually result in only one chain of TCRα and TCRβ in each T cell, the leakiness of the process can result in two functional TCRαs^{24, 25}. Sanger sequencing methods for those T cells will produce uninterpretable TCRα data (Supplementary Fig. 1).

Fig.1:

Fast and accurate identification of neoantigen reactive TCR pair from single T cell.

Another bottleneck in the process of TCR discovery occurs at the validation phase. Current methodologies²⁶ of TCR validation in T cells include 1) RNA mediated TCR delivery, 2) retroviral or lentiviral mediated TCR delivery, 3) transcriptionally active plasmid (TAP) DNA mediated TCR delivery²⁷, and 4) non-viral sleeping beauty transposon/transposase TCR delivary²⁸. The conventional method of producing retroviral constructs and supernatants is limited by the necessity to make TCR encoding retroviruses that requires at least 4 weeks to test TCR reactivity. We adopted non-cloning, and non-viral route by combining TCRα/β gene syntheses, IVT mRNA, to validate prospective TCRs in NFAT-driven luciferase reporter system. This approach, combined with Sanger Sequencing, and demultiplexing TCRα, permits the discovery of neoantigen-specific TCR in a timely fashion for ACT.

MATERIALS AND METHODS

Antibodies, flow cytometry, and cell sorting

The anti-mouse TCRβ-fluorescein isothiocyanate (FITC) was purchased from BioLegend. CD4-PE (RPA-T4), and CD8-PE-Cy7 (SK1) were purchased from BD Biosciences. 4–1BB-PE were purchased from eBioscience. Flow cytometry data acquisition was performed on FACSCanto II (BD Biosciences) followed by the analysis of acquired data by FlowJo software (TreeStar). Co-cultures were sorted based on 4–1BB⁺ separately for CD3⁺CD4⁺CD8⁻ (CD4) and CD3⁺CD4⁻CD8⁺ (CD8) by SH800S instrument (Sony Biotechnology).

Antigen-presenting cells

Immature dendritic cells (DC) were generated in the laboratory using a standard protocol of adherence method²⁹. Co-culture experiments were performed with Jurkat cells transfected with TCRs (equal concentration of TCRα and TCRβ) by electroporation. Transfected Jurkat cells were rested overnight for recovery and to allow optimal TCR expression prior to use in co-culture luciferase bioassays.

Single-cell sorting and Single-cell RT-PCR

Co-culture experiments were performed with either DCs or COS7 expressing A2 or A11 and reactive TIL cultures a day before single-cell sorting. Single-cell sorting was performed using a Sony SH800S instrument to isolate either CD8⁺4–1BB⁺ or CD4⁺4–1BB⁺ cells from a patient’s TIL. Cells were sorted directly into RT-PCR buffer (CellsDirect One-Step qRT-PCR kit, ThermoFisher) that contains one primer Cα and Cβ regions (1.2 μM for these gene-specific RT primers) as well as multiple Vα and Vβ primers (0.6 μM) in 10 μl volume in each well. Cycling conditions were as follows: 50°C for 15 min, 95°C for 2 min, 18 cycles of (95°C for 15s, 55°C for 30s, 72°C for 1min), 4°C. The second PCR was performed separately for TCRα and TCRβ chains. Platinum II Hot-Start PCR master mix (2X) was used for both PCR in a total of 25 μl volume that includes 2.5 μl of RT-PCR mix with several nested primers (0.6 μM each) for either TCRα or TCRβ targeting extended CDR3 regions of both chains. The PCR programs were as follows: 95°C for 7 min, 5 cycles of (95°C for 15s, 65°C for 15s, 72°C for 30s), 5 cycles of (95°C for 15s, 60°C for 15s, 72°C for 30s), 40 cycles of (95°C for 15s, 65°C for 15s, 72°C for 30s), 72°C for 7 min, and final 4°C incubation. The PCR products were sequenced by Sanger methods using internally nested Cα and Cβ region primers.

TCRα deconvolution PCR plate design

The sequencing data for TCRα often results in ambiguous uninterpretable sequences due to the presence of two distinctly different copies of TCRα (Supplementary Fig.3). In this even, TCRα PCR amplification was repeated with 12 sub-multiplex PCR primer groups. The 12 sub-multiplex PCR primer groups were added in each row of 96-well plates (Fig.1C) to generate deconvolution plates, which could be premade and frozen for at least 6 months at −80°C. The diluted RT-PCR mix from an ambiguous TCRα well was added in each well of the deconvolution (Fig. 1C). In the event that sequenced T cell populations appeared oligoclonal, such as when co-culture is performed with TIL fragments that are reactive towards specific peptide pools (PPs) or TMGs, pooled TCRα wells served as a template if all those ambiguous wells had a common productive TCRβ. The PCR reagents except primers/templates and programs were the same as the first TCRα PCR amplification. After the completion of the 2^nd PCR for TCRα, plates were purified and sequenced with the Sanger method using the same nested Cα primer used in the first sequencing. Sanger sequencing was re-analyzed for productive TCR pairings and used to reconstitute double stranded TCRα/TCRβ gene synthesis.

TCR reconstruction, subcloning into a retroviral vector and TCR transduction

Sanger sequencing produced sequencing data that contained the 3’ end of the variable region and the full CDR3 region of matching TCRα and TCRβ genes. The data were analyzed using IMGT/V-Quest tool (http://www.imgt.org/IMGT). The pairing methods are described elsewhere²⁰. Briefly, the productive TCRα and TCRβ data were aligned based on RT-PCR well number and the IMGT data was utilized to reconstruct the full-length TRAV and TRBV regions of each pair. The constant regions were replaced with modified murine TRAC and TRBC sequences to avoid mismatches with endogenous human TCR when expressed in human PBL. The full-length TCR coding sequence was placed in line to be expressed as a single mRNA with 2A peptide separating TCRβ and TCRα and subcloned into the pMSGV1 vector.

Retroviral supernatants were generated by transfecting TCR constructs into 293GP cells. Retroviral supernatants were collected at 48 hours of post-transfection. Meanwhile, PBLs from a healthy donor were activated in T cell 50/50 medium supplemented with 50 ng/mL anti-CD3 and 300 IU/mL IL-2 for 3 days. Retroviral particles were transduced in activated lymphocytes by using the spinoculation techniques, described earlier³⁰. The transduction of TCR was confirmed by using mouse TCRβ antibody at day 15 and used for reactivity and avidity experiment.

Somatic mutation screening reagents, TMG construction, and peptide synthesis

Whole exome sequencing (WES) and RNA-seq were performed in the Surgery Branch as described elsewhere²⁹. Non-synonymous mutations of each tumor biopsy sample were called based on a bioinformatics analysis and TMGs were constructed by concatenating up to 17 minigenes. Each minigene consisted of mutant amino acids flanked by 12 amino acids of the wild type protein sequence. TMGs were composed of coding regions of minigenes flanked by a signal peptide at 5’ end and DC-LAMP trafficking sequence at the 3’ -end to enhance processing and presentation. TMGs were subcloned into pcRNA2SL plasmid, a modified version of pCDNA3.1 plasmid using EcoRI and BamHI. The TMG plasmids were linearized with NotI (New England Biolabs) for IVT templates. Purified linear plasmid DNA (1 μg) was used to synthesize mRNA by in vitro transcription (mMessage mMachine T7 Ultra kit, Life Technologies). The mRNAs were purified by precipitating with LiCl₂ and diluted with DEPC H₂O. After checking the quality of RNA and measuring concentration, RNA concentration was adjusted to 1 μg/μl and stored at −80°C. Crude peptides were purchased from GenScript for in vitro screening of T cells. HPLC-purified mutant peptides were used for confirmatory validation purposes. For screening, APCs (DCs, or HLA-restricted COS7 cells) were pulsed with peptide pools (each peptide’s final concentration was 10 μg/mL for 25-mers). APCs were washed after 2 hours and immediately used in co-culture assays.

Design of TCRα/TCRβ1

The relatively short length of reconstituted TCRα (about 925 bp) and TCRβ (approximately 1 kb) allowed template DNA to be synthesized for in vitro transcription. Several factors were considered to design a universal TCR gene synthesis model, such as the ability to enrich template DNA by universal primer pair, synthesizing each chain separately to increase the efficiency of IVT, and adding 3’ and 5’ untranslated region (UTR) to enhance stability IVT-mRNA. The 5’ end of the gene fragment included the F8 primer binding site, consensus T7 promoter, mini-CMV promoter sequence, Kozak sequence containing start codon. The 3’ -end included stop codon with universal primer sites that serve as PCR priming site as well as short 3’UTR (Supplementary Table 2). The universal primer binding sites were included to quickly enrich the TCR template to obtain enough (about 2–4 μg) of DNA template for in vitro transcription (IVT). The CDS of TCRα/TCRβ was designed by adding variable region, leader sequence, and constant region of murine modified constant region for both chains. The murine constant regions eliminate cross dimerization with endogenous TCRα and TCRβ in Jurkat cell.

Synthesis of TCRα/TCRβ and IVT

Once putative neoantigen-specific TCRs for a patient are determined, TCRA and TCRB are reconstituted to be synthesized as gBlock double stranded DNA fragments (IDT). PCR amplification of all gBlocks was performed using either (8F + AD-reverse) or (T7 forward + AD-reverse) primer pair to enrich gBlock concentration. The PCR program for template enrichment was as follows: 95°C for 7 min, 30 cycles of (95°C for 15s, 55°C for 15s, 72°C for 1 min), 72°C for 7 min, and 4°C. The PCR products were column purified by using the QIAquick PCR Purification kit (Qiagen, Cat No. 28104). The purified PCR DNAs (1 μg/μl) were used as templates for in vitro transcription (IVT). The IVT was performed using T7 mScript Standard mRNA Production System (Cellscript, Cat. No. C-MSC11610).

CD4⁺ or CD8⁺ Jurkat-NFAT-Luc reporter cell production

The Jurkat-NFAT-luciferase cell line was obtained from BPS Bioscience. To generate CD4⁺ and CD8⁺-Jurkat NFAT-Luc cells, we constructed retroviral plasmids by subcloning human CD4 and CD8α/CD8β coding regions into the pMSGV1 vector. These plasmids were then transfected into 293GP cells. Retroviral supernatants were collected at 48 hours post-transfection, centrifuged to discard cell debris, diluted 1:1 with T cell medium and used to transduce Jurkat-NFAT-Luc cell using the spinoculation method, as described previously²⁹. Transduced Jurkat-NFAT-Luc were tested for CD4 or CD8 expression by FACS.

Transfection of TCR mRNA

Healthy, viable CD8⁺Jurkat-NFAT-Luc cells were transfected with TCR mRNAs. The reporter cells were centrifuged at 90 × g for 10 min and cell pellets were resuspended in Opti-MEM media. The TCR mRNA mix (5μg of each chain) were added to 1 × 10⁶ reporter cells resuspended in 90 μl OPTI-MEM medium. The electroporation protocol for Jurkat cells was 400 V with one ms pulse (ECM 830, BTX). Transfected cells were then transferred into 1 ml T cell media in 24-well plate and rested overnight before use in co-culture experiments.

Luciferase reporter assay

TCRα and TCRβ mRNAs (5 μg each) were co-transfected in reporter Jurkat cells by electroporation in a volume of 100 μl Opti-MEM (Invitrogen). Electroporation was 400V, over a 1 ms pulse (BTX), and cells were rested overnight in a 24 -well plate with T cell 50/50 medium. Transfected Jurkat reporter cells were then used for co-culture assay for 5 hours in a round bottom 96- well plate. Cells were centrifuged and cell pellets were used to perform luciferase assay using Steady-Glo luciferase kit (Promega). Luminescence was measured with FLUOstar Omega (BMG LABTECH).

RESULTS AND DISCUSSION

Here we describe a novel method to eliminate the dual TCRα ambiguity by de-multiplexing original nested PCR primers (Supplementary Table 1) into 12 subgroups of primer pools to amplify both TCRα amplicons in different subgroups (Fig. 1A, B, C). Each primer pool contains either 3 or 4 nested primers that was added in each column of a 96-well plate (Fig. 1B, C). Since each row of a 96-well plate contains all 12 sub-multiplexing primer pools to deconvolute one ambiguous TCRα well, a 96-well convolution plate can deconvolute 8 independent ambiguous TCRα wells from the original RT-PCR plates. Using this approach, we can identify two different TCRα species, at least one of them being the productive TCRα, that can be matched with its counterpart productive TCRβ to create native TCR pairs (Fig. 1C). The probability of having both TCRα chains in the same primer subgroup is less than 2%. On such a rare occasion, in-group de-multiplexing PCRs can resolve the TCRα ambiguity.

The deconvolution of TCRα was performed in four different patients to accurately identify TCRαs to match their counterpart productive TCRβs and to form neoantigen reactive TCRs (Fig. 1D). We identified two productive TCRα chains (TRAV6*01 and TRAV19) for patient #1 and two productive TCRα chains (TRAV14 and TRAV39) for patient #4. The sub-multiplexing method identified one productive (TRAV19) and an unproductive (TRAV26–2*01) for patient # 2. Similarly, patient #3 had one productive (TRAV12*01) and one unproductive (TRAV19) TCRα in a well that had a productive TCRβ. This new approach allowed us to identify TCRα chains accurately and efficiently using Sanger sequencing within 1–2 days without compromising efficiency.

To rapidly evaluate putative TCRs from a patient, we synthesized TCR genes encoding alpha or beta chains and used in vitro transcribed (IVT)-TCR-mRNAs to transfect them into the Jurkat luciferase cells. We created a universal TCR gene fragment design for testing purposes and developed a protocol for TCR testing that can be carried out in 7 days or less time in ideal conditions (Fig. 2A). Functional endpoint measures of TCR stimulation are determined by measuring either cytokine productions (IFNγ or IL2) or measuring activation of a reporter construct driven by a Nuclear factor of activated T cells (NFAT) promoter. Since Jurkat cells express very low levels of CD4, and no CD8, the CD4 and CD8 were retrovirally introduced to provide co-stimulation. Transduction efficiencies for both CD4 and CD8 in the Jurkat NFAT-Luc cells were over 85% as detected by FACS staining of CD4 and CD8 (Fig. 2B). The modified Jurkat cells provide consistency, and scalability for the increasing demand of validating neoantigen reactive TCR testing.

Fig.2.

TCR design for streamlined mRNA synthesis and transfection into Reporter cells for functional evaluation.

The method was then utilized for the analysis of two prospective TCRs identified by screening patient-4347 TIL against products of a mutated KRAS^G12D expressed by the patient’s tumor cells. We designed and synthesized gene fragments of 2 TCRαs and 2 TCRβs that form two prospective TCRs. The IVT mRNAs (TCR1 and TCR2) were tested for functional reactivity towards KRAS^G12D. Since HLA restriction was known to be HLA-A11 (unpublished data), we used COS7-A2 (HLA-A11plasmid stably transfected) cells for co-culture experiments by pulsing with either WT or minimal epitope peptide mixes of mutated KRAS^G12D for 2 hours. The reporter bioassay showed that Jurkat cells transfected with TCR2 specifically reacted to KRAS^G12D presented by COS7-A11 (Supplementary Fig. 2).

To determine the minimum amount of TCR-mRNA required to generate a positive luciferase signal, we measured transfection efficiency of various concentrations of TCR2-IVT mRNA into Jurkat cells and measured its expression by FACS using an antibody targeting the mouse constant region of TCRβ. Transfection plateaued around 20% with increasing concentration between 1.25 and 10 μg for each chain of TCR2 mRNA (patient-4373) (Supplementary Fig. 3). To determine sensitivity of luciferase assay, co-culture of these transfected cells with COS7-A11 pulsed with KRAS^G12D were performed and resulted in comparable luciferase activity based on the transfection efficiency (Fig. 2C). To determine optimal reporter activation, we also performed a time- course co-culture experiment using COS7-A2 and COS7-A11 and TCR2 (patient- 4373)-transfected CD8⁺-Jurkat-NFAT-Luc cells. Luciferase activity was higher at 5 hours compared to that of overnight co-culture experiments (Fig. 2D).

We then utilized this approach to evaluate responses against the KRAS^G12V neoepitope, expressed by patient-4385’s tumor, resulting in the identification of 4 reactive TCR candidates from a population of CD8⁺ T cells that up-regulated 4–1BB⁺ in response to stimulation with full length (FL) KRAS^G12V and TMG2. To evaluate the reduction in testing time by using our IVT-mRNA-Jurkat reporter system, we simultaneously synthesized reconstituted TCRα and TCRβ gene fragments and generated IVT mRNAs for each TCR. First, co-culture experiment was performed with retrovirally transduced PBL and autologous DC transfected either with full length mutated KRAS^G12V or a tandem minigenes encoding the KRAS^G12V mutation and evaluated for reactivity using ELISPOT assays. The ELISPOT assay identified TCR4 to be KRAS^G12V reactive and was also positive in TILfragment#11 (Supplementary Fig. 4). Second, TCR-mRNAs transfected Jurkat cells were cocultured with autologous DC transfected either with full length mutated KRAS^G12V or a tandem minigenes encoding the KRAS^G12V mutation. The modified Jurkat NFAT-Luc bioassay also identified TCR4 to be reactive to KRAS^G12V (Fig. 3A). The IVT-mRNA-luciferase system independently confirmed TCR4 reactivity much faster than the identification of TCR4 reactivity through the conventional method.

Fig.3:

Comparison of graded peptide titrations between ELISPOT and reporter bioassays

Finally, we compared the functional avidity of transduced T cells with transfected Jurkat T cells, since this represents an important determinant in selecting TCRs for use for gene therapy. The results of the ELISPOT assay demonstrated that PBMC transduced with TCR2 from patient-4373 could recognize targets pulsed with a peptide corresponding to the minimal KRAS^G12D epitope at a minimum concentration of 0.1 ng/ml (Fig. 3B). Between 1 and 10 ng/ml of minimal KRAS^G12D was required for stimulation of Jurkat T cells, but these levels were similar when compared responses of retrovirally-transduced (Fig. 3C) and mRNA-transfected Jurkat reporter T cells (Fig. 3D). The conventional ELISPOT or ELISA method is sensitive and reliable, although several steps requiring an hour or more incubation time are required. On the other hand, the Jurkat luciferase system requires less than 30 min to measure luciferase activity after the completion of co-culture incubation with an option to freeze cells for future analysis. TCR evaluation method by using gene fragments followed by luciferase bioassay described in the 7-day streamlined protocol (Fig. 2A) provide a reliable TCR screening platform compared to the traditional retroviral-ELISPOT protocols that takes much longer time.

The streamlined TCR identification and testing technique (9 days) proposed in this paper requires minimal hands-on labor compared to several alternate techniques. Our TCR screening technique is easily adoptable to both class I and class II restricted neoantigens. When restriction HLA information was not available, we used Dendritic cells (DC) to present neoantigens in coculture experiments. However, due to difficulties in culturing DC, we use HLA expressing COS7 cells to shorten our overall timeframe of 9 days. Widely diverse technologies are currently employed to identify scTCR in a similar timeframe.

Hu¹⁴ et al described an elegant streamlined approach to test the specificity and avidity of antigen specific TCRs using Golden gate cloning techniques. Another novel method, termed Repertoire and Gene Expression by sequencing (RAGE-Seq)³¹ identifies full length TCR sequences by combining targeted capture and long-read sequencing of TCR mRNA transcripts with short-read transcriptome profiling of barcoded single cell libraries generated by droplet-based partitioning. This technique reduces the higher error rate of long read sequencing by coupling short-read technologies. This method accurately identifies CD3 regions of TCR pairs from large number of lymphocytes. The cost per cell basis appears less expensive than our method, but the upfront infrastructure costs are significantly higher to execute this platform. Our method identifies neoantigen specific TCRs in similar timeframe with cost comparable to the Golden gate method or RAGE-Seq technology.

Several plate-based multiplex methods^{18, 32} have been described to identify TCR pairings using NGS platforms. Our method employs similar multiplex techniques to amplify single cell TCRs without use of NGS based sequencing and analysis. Another group³² described a multiplex method called the rhTCRseq method³²(RNase H-dependent PCR-enabled T cell receptor sequencing) using sorting into 384 well plates followed by production of a full-length cDNA library, and addition of dual index barcodes for sequencing using the MiSeq platform. The major advantages of these techniques are elimination of TCRα ambiguity and sequencing of pooled barcoded TCRs thereby, simplifying the workflow and reducing the cost. Compared to these NGS methods, our method achieves TCR pairings in a comparable time frame with fewer overall steps and expense.

Our novel scTCR amplification enhanced the power of the Sanger sequencing method in accurately identifying native TCRα to form neoantigen reactive TCR pairs. Our use of premade frozen RT-PCR plates, PCR plates, TCRα deconvolution plates, and frozen reporter cells (CD8 or CD4-specific), combined with the use of IVT mRNA for screening, makes this TCR discovery platform scalable and for use in carrying out relatively rapid and high-throughput screening of tens of hundreds of candidate TCRs in conventional laboratory and clinical setting in a time-efficient manner. In addition, currently we are evaluating a 384 well plate format to further increase throughput, scalability and TCRα deconvolution efficiency.

The neoantigen-specific TCR discovery platform described here enhances the sequential process of identifying neoantigens and identifying TCRs reactive to both common driver mutations and unique private neoantigens that are frequently encountered in the screening system. For these screening experiments, single-cell PCR followed by Sanger sequencing is ideal for identifying TCRα/β pairs associated with reactive T cells. Our method overcomes the limitations of Sanger sequencing of TCRα and provides a TCR discovery platform distinct from the growing family of multiplexing single-cell technologies. The technology described in this paper can reduce the substantial time required for functional TCR discovery and expedite translation to ACT for cancer treatment.

Supplementary Material

Supplemental Data File_1

Supplemental Figure 1: Typical data obtained from the PCR product of TCRα by Sanger method. A, Sanger sequencing method provided clean sequence and TCR alignment is productive and unambiguous. B, Sequencing data of TCRα were contaminated with two DNA species in the PCR amplicon mix, TCR alignment may be productive, but clearly ambiguous as indicated by a double peak in the chromatographic data. C, Sanger sequencing produced a TCRα data that is uninterpretable.

Supplemental Data File_2

Supplemental Figure 2: Reporter assay of two putative TCRs (patient-4373) introduced in CD8⁺-jurkat-NFAT-Luc cells by mRNA TCR. Co-culture of COS7-A11 pulsed with KRAS^G12D minimal epitope and CD8⁺-jurkat-NFAT-Luc transfected with either TCR-mRNA or purified PCR products (equal amount of TCRα and TCRβ) were performed for overnight incubation. The PCR mix of TCR1 and TCR2 did not show any reactivity towards KRAS^G12D ME, but TCR2-mRNA was reactive towards KRAS^G12D ME.

Supplemental Data File_3

Supplemental Figure 3: The transfection efficiency of in vitro transcribed TCR mRNA was tested using mouse TCRβ fluorescence antibody. Transfection was performed by electroporation of CD8⁺-jurkat-NFAT-Luc cells with decreasing amount of TCRα/TCRβ -mRNA (10 μg each to 1.25 μg each) for TCR2 (patient-4373). Transfected cells were rested overnight before staining for FACS analysis. The TCR2 FACS data demonstrate that TCR transfection did not increase with the use of a higher amount of TCR-mRNA. The TCR2 transfection efficiency was about 20% in all reporter cells transfected with various amounts of TCR2-mRNA.

Supplemental Data File_4

Supplemental Figure 4: Autologous DC (patient-4385) was transfected with WT-FL KRAS gene, KRAS^G12V-FL, and TMG carrying mutated KRAS^G12Vand co-cultured with four putative TCRs transduced into donor PBL along with original positive TIL fragment 11 (F11, patient-4385). The ELISpot data confirmed that TCR4 (HLA-A2 specific TCR) was reactive to both TMG2 and mutated KRAS^G12V gene.

Supplementary Table_1

Supplemental Table 1: Subgroup of TRAV primers for deconvoluting TCRα ambiguity in the Sanger sequencing method.

Supplementary Table_2

Supplemental Table 2: List of primers and linkers for designing T7 RNA polymerase transcribable double stranded TCRα/β gene fragments.