Type 1 diabetes (T1D) is a T cell-mediated disease with a strong immunogenetic human leukocyte antigen (HLA) dependence. HLA allelic influence on the T cell receptor (TCR) repertoire shapes thymic selection and controls activation of diabetogenic clones yet remains largely unresolved in T1D. We sequenced the circulating TCRβ chain repertoire from 2250 HLA-typed participants across three cross-sectional cohorts, including individuals with T1D and healthy related and unrelated controls. We found that HLA risk alleles show higher restriction of TCR repertoires in individuals with T1D. We leveraged deep learning to identify T1D-associated TCR subsequence motifs that were also observed in independent TCR cohorts residing in pancreas-draining lymph nodes of individuals with T1D. Collectively, our data demonstrate T1D-related TCR motif enrichment based on genetic risk, offering a potential metric for autoreactivity and groundwork for TCR-based diagnostics and therapeutics.

Aims Characterise T-cell receptor gene (TR) repertoires of small intestinal T cells of patients with newly diagnosed (active) coeliac disease (ACD), refractory CD type I (RCD I) and patients with CD on a gluten-free diet (GFD). Methods Next-generation sequencing of complementarity-determining region 3 (CDR3) of rearranged T cell receptor β (TRB) and γ (TRG) genes was performed using DNA extracted from intraepithelial cell (IEC) and lamina propria cell (LPC) fractions and a small subset of peripheral blood mononuclear cell (PBMC) samples obtained from CD and non-CD (control) patients. Several parameters were assessed, including relative abundance and enrichment. Results TRB and TRG repertoires of CD IEC and LPC samples demonstrated lower clonality but higher frequency of rearranged TRs compared with controls. No CD-related differences were detected in the limited number of PBMC samples. Previously published LP gliadin-specific TRB sequences were more frequently detected in LPC samples from patients with CD compared with non-CD controls. TRG repertoires of IECs from both ACD and GFD patients demonstrated increased abundance of certain CDR3 amino acid (AA) motifs compared with controls, which were encoded by multiple nucleotide variants, including one motif that was enriched in duodenal IECs versus the PBMCs of CD patients. Conclusions Small intestinal TRB and TRG repertoires of patients with CD are more diverse than individuals without CD, likely due to mucosal recruitment and accumulation of T cells because of protracted inflammation. Enrichment of the unique TRG CDR3 AA sequence in the mucosa of patients with CD may suggest disease-associated changes in the TCRγδ IE lymphocyte (IEL) landscape.

TCR Sequencing data of each patient

Supplementary Figure 5. Example of evolved clonotypes in one patient. To illustrate the dynamics of evolved clonotypes, a patient with MRD at day 29 (index clone persists at 0.76% with a large number of VH-replaced variants of the index clone) is shown. For each time point (Pre-treatment, x-axis and Day 29 post-treatment, y-axis), this plot shows the fraction of the total clone size (major clone plus evolved clonotypes) accounted for by each clonotype. Each data point represents a different clonotype. 1,607 evolved versions of the index clone were seen only in the pre-treatment sample, and 112 versions were seen only in the post-treatment sample; these are shown in red. 24 evolved clonotypes were seen at both time points, and these are shown in blue. The index clone at diagnosis is shown as a black data point (top right).

Supplementary Figure 2 has the statistical analysis of TCR sequencing from new samples of control patients (no immunotherapy) compared with DC vaccine patients.

The relative frequency of the top 25 clones from tumor 1 tracked in pre-therapy blood, post-therapy blood, and in tumor 2 post-therapy

Supplementary Figure 1 shows representative IHC staining of TIL to confirm that these T cells are found in the parenchyma of the tumor and not tumor blood vessels.

Out-of-field tumor regression is a rare event following local single dose or fractionated RT.

PDF file - 65KB, Table S1 Summary for TCR V-beta CDR3 sequence analysis. Table S2 Number of unique Productive Sequences in Top 25% abundance at four different timepoints. Table S3 Number of unique productive Sequences in Top 25% clones at baseline and at Day 30-60.

Figure S1. Relative composition of TCRs in unsorted tumors within each treatment group. Figure S2. TCRβ CDR3 sequence distance among AH1-pentamer+/tetramer+CD8+ T cells. Table S1. Summary of TCR sequencing of whole 4T1 tumors. Table S2. Summary of TCR sequencing of AH1-pentamer+CD8+ T cells. Table S3. Abundance of AH1 clone signature (amino acid sequences) in CD4+ or CD8+ sorted T cells. Table S4. Three AH1 reactive T cell clones were shared between 4T1 tumors and AH1 peptide-vaccinated animals.

Concurrent but not sequential blockade of PD-1 is required to augment the efficacy of fractionated RT

Figure legends for Supplemental Figures 1-6

Representative staining for PD-L1 on tumor cells (CD45-) and CD11b+Gr1+ cells

Supplementary Figure 1. Frequency of B cells with unique IGH rearrangements identified in bone marrow aspirates from 9 normal individuals. We performed multiplex PCR and sequencing to capture the IGH rearrangement repertoire present in the bone marrow samples of 9 healthy adults, and analyzed the frequency of the most common clones in each sample. The average frequency of the top clone was 0.8% of B cells, or ~0.08% of total nucleated cells. Above, the probability density of a normal distribution calculated using the geometric mean and standard deviation of top clone frequency is charted for this dataset.

Treating rapidly evolving pathogenic diseases such as COVID-19 requires a therapeutic approach that accommodates the emergence of viral variants over time. Our machine learning (ML)-guided sequence design platform combines high-throughput experiments with ML to generate highly diverse single-domain antibodies (VHHs) that bind and neutralize SARS-CoV-1 and SARS-CoV-2. Crucially, the model, trained using binding data against early SARS-CoV variants, accurately captures the relationship between VHH sequence and binding activity across a broad swathe of sequence space. We discover ML-designed VHHs that exhibit considerable cross-reactivity and successfully neutralize targets not seen during training, including the Delta and Omicron BA.1 variants of SARS-CoV-2. Our ML-designed VHHs include thousands of variants 4-15 mutations from the parent sequence with significantly improved activity, demonstrating that ML-guided sequence design can successfully navigate vast regions of sequence space to unlock and future-proof potential therapeutics against rapidly evolving pathogens. ### Competing Interest Statement The authors have declared no competing interest.

Supplementary Figure 3 has the top 25 TCR clones from tumor biopsies at pre-treatment and the frequency of these clones in the blood.

Supplementary Figure 1. Gemcitabine and CTLA-4 blockade synergize to create an inflammatory microenvironment and anti-tumor response. Mice with palpable tumors were treated with low dose gemcitabine followed by systemic CTLA-4 blockade. Survival was improved (A) and tumors with combination treatment had a significant influx in CD4 and CD8 Tcells (B). Elispot (C) demonstrated improved enhanced IFNγ in response to SPAS-1 peptide (SNC9H8). Supplemental Figure S2. Lack of significant change of TIM3 or LAG3 over time in peripheral blood mononuclear cells (PBMC) over time. Supplementary Table S1. Toxicities in the limb according to Wieberdink criteria Supplemental Table 2. Nanostring fold change results of all genes

Supplementary Table 1: TCRB sequencing sample IDs

Cox Proportional Hazard table of the Estimated TIL content and patient survival

Supplementary Figure 2. Amplification bias across iterative primer mix optimization. The PCR amplification bias (proportional representation, relative to mean) of each of our synthetic templates (1,116 for IGH) was calculated for an equimolar mix of forward and reverse primers, with a final optimized primer mix, and after computational normalization (these results are from a subsequent experiment, not the one used to derive computational normalization factors). The calculated dynamic range and interquartile range of PCR bias, and sum of squared log (PCR bias) values are shown for each primer mix. Modification of multiplex primer concentrations followed by computational normalization is an effective way to minimize PCR bias even in complex multiplexed assays.