Cellular origins and genetic landscape of cutaneous gamma delta T cell lymphomas

Abstract

Primary cutaneous γδ T cell lymphomas (PCGDTLs) represent a heterogeneous group of uncommon but aggressive cancers. Herein, we perform genome-wide DNA, RNA, and T cell receptor (TCR) sequencing on 29 cutaneous γδ lymphomas. We find that PCGDTLs are not uniformly derived from Vδ2 cells. Instead, the cell-of-origin depends on the tissue compartment from which the lymphomas are derived. Lymphomas arising from the outer layer of skin are derived from Vδ1 cells, the predominant γδ cell in the epidermis and dermis. In contrast, panniculitic lymphomas arise from Vδ2 cells, the predominant γδ T cell in the fat. We also show that TCR chain usage is non-random, suggesting common antigens for Vδ1 and Vδ2 lymphomas respectively. In addition, Vδ1 and Vδ2 PCGDTLs harbor similar genomic landscapes with potentially targetable oncogenic mutations in the JAK/STAT, MAPK, MYC, and chromatin modification pathways. Collectively, these findings suggest a paradigm for classifying, staging, and treating these diseases.

Fig. 1. Epidermal/dermal and panniculitic CGDTLs derived from distinct cells of origin.

a Schematic highlighting distinct clinical and histological presentations of disease involving epidermis, dermis, or subcutaneous tissue. Clinical photographs of disease lesions, hematoxylin and eosin staining of biopsies, and γδ T cell receptor immunostaining (see “Methods” section) for representative patients with epidermal, dermal, and panniculitic disease are shown. Skin schematic created with BioRender. Scale bar represents 100 μm in bottom right epidermal panel, bottom left dermal panel, and bottom right panniculitic panel; 200 μm in top right epidermal panel, bottom right dermal panel, and bottom left panniculitic panel; 500 μm in top right dermal panel and top right panniculitic panel. b Frequency of δ chain usage by skin compartment in CGDTL as assessed by RNA-seq and high-throughput TCR-Seq. Lymphomas involving epidermis and/or dermis (n = 8) or subcutaneous tissue (panniculitic) (n = 7). *** Indicates P value = 0.0002, two-sided Fisher’s exact test. c, d Flow cytometry analysis showing percentage of Vδ1 and Vδ2 T cells in normal human epidermis, dermis, and subcutaneous tissue (n = 5). Dots represent individual values, horizontal line represents mean, and error bars represent standard deviation. **** Indicates P value < 0.0001, one-way ANOVA followed by Tukey’s multiple comparison test. Source data are provided as a source data file.

Fig. 2. Vδ1 and Vδ2 CGDTL lymphomas are transcriptionally distinct.

a Principal component analysis of transcriptomes of Vδ1 and Vδ2 PCGDTL samples. * Indicates Vδ1 γδ MF with PCGDTL-like progression sample. b Volcano plot of differentially expressed genes significantly upregulated in Vδ1 (blue) and Vδ2 (red) with an adjusted P value < 0.05 (DESeq2). Select immune-related genes are labeled. c–e Top five pathways upregulated in Vδ2 PCGDTL lymphomas via GO pathway, MSigDB Hallmark, and CheA transcription factor-binding analysis respectively.

Fig. 3. Cell of origin and histologic subtype influences clinical phenotypes in CGDTL.

a, b Prevalence in patients with Vδ1 or Vδ2 disease of lymph node involvement (Vδ1 n = 24; Vδ2 n = 10) and tumor stage skin lesions (Vδ1 n = 25; Vδ2 n = 13) at diagnosis, respectively. c, d Prevalence of ulcerated skin lesions (Vδ1 n = 26; Vδ2 n = 12) and cytotoxic marker staining (Vδ1 n = 22; Vδ2 n = 9), respectively, based on cell of origin. e Kaplan–Meier curves of patients with Vδ1 (n = 27) or Vδ2 (n = 12) lymphomas. f Kaplan–Meier curve based on depth of disease in the diagnostic biopsy specimen (epidermal n = 9; dermal n = 13; panniculitic n = 12). g Swimmer’s plot highlighting disease course of Mycosis-fungoides like cases that acquire an ulcerative phenotype (n = 6). HSCT, hematopoietic stem cell transplant. h Kaplan–Meier survival curve for patients with dermal Vδ1 lymphomas (n = 11; excluding γδ MF cases) and Vδ2 lymphomas (n = 12). i–k Frequency of B symptoms (Vδ1 n = 26; Vδ2 n = 12), hemophagocytic lymphohistiocytosis (HLH) (Vδ1 n = 25; Vδ2 n = 11), and visceral spread (Vδ1 n = 28, Vδ2 n = 12), respectively, among patients with Vδ1 or Vδ2 disease. l Images of positron emission tomography–computed tomography from two patients with Vδ1 lymphomas with spread to the gastrointestinal tract. White arrows highlight areas of increased signal; stomach (GD4) and large intestine (GD5). For e, f ** indicates P value <0.005, log-rank test. For b, i, j *indicates P value <0.05, and **indicates P value <0.005, two-sided Fisher’s exact test. Source data are provided as a source data file.

Fig. 4. CD1d-lipid binding of Vδ1 CGDTL TCR.

a Frequency of γ chain usage among PCGDTL and γδ MF by disease subtype. b Frequency of Vγ9+ and Vγ9− cells among Vδ2 cells from subcutaneous adipose tissue in normal donors (n = 5). c Frequency of CD1d-PBS-57 tetramer staining, gated on Vδ1 cells. n = 5 donors in each condition. d Flow cytometry histogram of retrovirally transduced CD8+ T cells with fluorescently labeled CD1d-PBS-57 tetramers. e Structural model of GD8 TCR in complex with CD1d-lipid. On right, CDR3γ residues Arg 103 and Tyr 107 are shown as sticks, with predicted polar interactions with glycolipid head group shown as dotted lines. f Representative histogram of flow cytometry analysis of CD1d-PBS-57 tetramer staining in HEK293 cells co-transfected with CD3 and TCR. Amino acid sequence of GD8 and GD8-Mutant CDR3γ are indicated. g Percent of CD1d-PBS-57 tetramer positive cells in HEK293 cells co-transfected with TCR and CD3. n = 3 independent experiments, **** indicates P value <0.0001, one way ANOVA followed by Tukey’s multiple comparisons test. Dots represent individual values, horizontal line represents mean, and error bars represent standard deviation. Source data are provided as a source data file.

Fig. 5. The genomic landscape of CGDTL.

a Putative driver mutations in CGDTL. Rows represent genes, and each column corresponds to a CGDTL sample. Top bar plot indicates number of non-synonymous mutations detected in each sample, and bar plot on the right indicates total number of mutations in each putative driver gene. CNLhom and CNLhet; homozygous and heterozygous copy number loss, respectively. b Median contribution of COSMIC mutational signatures in CGDTL subtypes, Vδ1 n = 11, Vδ2 n = 7. DSBR double-strand break repair defect. c Frequency of copy number alterations. Driver genes within significant GISTIC peaks are listed. d Frequencies of arm level events in CGDTL subtypes. e Kaplan–Meier curve depicting percent survival with (n = 6) or without (n = 9) recurrent hotspot mutations in MAPK pathway genes (KRAS, MAPK1, and NRAS). ** Indicates P value < 0.005, log-rank test.

Fig. 6. Proposed model of CGDTL cell of origin and disease pathogenesis.

Epidermal/dermal disease arises from the Vδ1 cell and can potentially bind to lipid antigens presented by CD1d. Panniculitic disease arises from the Vδ2/Vγ3 cell in the subcutaneous tissue. Both cells acquire similar genetic alterations. Vδ1 γδ MFs clinically can have either a non-cytotoxic, indolent course or can switch after prolonged indolence to a cytotoxic phenotype that is more aggressive. Vδ2 lymphomas are more aggressive, associated with inflammatory gene signatures and the development of cytokine-driven paraneoplastic syndromes including hemophagocytic lymphohistiocytosis. Scale bar represents 100 μm (top) and 200 μm (bottom). Created with BioRender.