Purpose:

Clonal malignant T lymphocytes constitute only a fraction of T cells in mycosis fungoides skin tumors and in the leukemic blood of Sézary syndrome, the classic types of cutaneous T-cell lymphomas. However, lack of markers specific for malignant lymphocytes prevents distinguishing them from benign T cells, thus delaying diagnosis and the development of targeted treatments. Here we applied single-cell methods to assess the transcriptional profiles of both malignant T-cell clones and reactive T lymphocytes directly in mycosis fungoides/Sézary syndrome patient samples.

Experimental Design:

Single-cell RNA sequencing was used to profile the T-cell immune repertoire simultaneously with gene expression in CD3+ lymphocytes from mycosis fungoides and healthy skin biopsies as well as from Sézary syndrome and control blood samples. Transcriptional data were validated in additional advanced-stage mycosis fungoides/Sézary syndrome skin and blood samples by immunofluorescence microscopy.

Results:

Several nonoverlapping clonotypes are expanded in the skin and blood of individual advanced-stage mycosis fungoides/Sézary syndrome patient samples, including a dominant malignant clone as well as additional minor malignant and reactive clones. While we detected upregulation of patient-specific as well as mycosis fungoides– and Sézary syndrome–specific oncogenic pathways within individual malignant clones, we also detected upregulation of several common pathways that included genes associated with cancer cell metabolism, cell-cycle regulation, de novo nucleotide biosynthesis, and invasion.

Conclusions:

Our analysis unveils new insights into mycosis fungoides/Sézary syndrome pathogenesis by providing an unprecedented report of the transcriptional profile of malignant T-cell clones in the skin and blood of individual patients and offers novel prospective targets for personalized therapy.

Translational Relevance

Clonal malignant T lymphocytes constitute only a fraction of T cells in the most common types of cutaneous T-cell lymphomas, mycosis fungoides and Sézary syndrome. However, lack of specific markers prevents distinguishing them from benign T cells, thus delaying diagnosis and development of targeted treatment, resulting in poor clinical outcomes. Here we demonstrate the application of single-cell RNA sequencing for high-resolution profiling of the T-cell immune repertoire simultaneously with gene expression to assess the transcriptional profiles of expanded malignant clones and reactive T lymphocytes directly in mycosis fungoides/Sézary syndrome samples. As there is no cure for mycosis fungoides/Sézary syndrome at any stage of diagnosis, the translational implication of this study is for identifying novel prospective targets for therapy as well as indicating strategies for tailoring therapy to specific patients.

Cutaneous T-cell lymphomas (CTCL) are rare clonal disorders of skin-homing memory T cells (1, 2). The most common subtypes are the skin-limited form, mycosis fungoides, where malignant T cells reside primarily in the skin dermo-epidermal junction, and the leukemic variant, Sézary syndrome, characterized by circulating malignant T cells (1, 2). Previous studies indicate that mycosis fungoides originates from the clonal expansion of a transformed skin-resident effector-memory T cell (1). In comparison, clonal Sézary syndrome cells present a cell surface phenotype of skin-homing central memory T cells, which recirculate between the lymph nodes, blood, and skin where they are primarily localized to the dermis (1–3). However, recent data suggests that there is plasticity in the putative cell of origin in mycosis fungoides and Sézary syndrome, with leukemic cells of advanced mycosis fungoides found to express a central memory phenotype in blood and lymph nodes (4). The pathophysiologic distinction between mycosis fungoides and Sézary syndrome therefore remains a topic of active inquiry. Clinically, the most recent WHO/EORTC 2018 classification defines Sézary syndrome clinically as pruritic erythroderma, generalized lymphadenopathy, and clonally related neoplastic T cells with cerebriform nuclei (Sézary cells) in the skin, lymph nodes, and peripheral blood. Cases that otherwise meet the staging criteria for Sézary syndrome but arise in patients with a prior history of mycosis fungoides are noted by the EORTC/ISCL guidelines to represent “Sézary syndrome preceded by mycosis fungoides” or secondary “Sézary syndrome”.

Although many T lymphocytes are found infiltrating mycosis fungoides skin tumors and in the blood of Sézary syndrome patients, distinguishing malignant from benign lymphocytes is impaired by a lack of specific markers for identifying and isolating the malignant cells, thus delaying diagnosis and development of specific treatments, which results in poor clinical outcomes (5). Detection of the expanded T-cell clonotype has been critical for diagnosing mycosis fungoides/Sézary syndrome (6). However, the most commonly used clinical tests for establishing T-cell clonality are only partially sensitive and specific (7–10). Recent work has shown that high-throughput sequencing of the TCRB complementarity-determining region 3 can identify the expansion of malignant T-cell clones in mycosis fungoides skin lesions and precisely determine the fraction of tumor clone per skin volume (6). Although reliable for the early diagnosis of mycosis fungoides (6), this approach does not reveal the transcriptional profiles of the malignant clones that may be used to develop novel diagnostic and prognostic markers as well as specific targets for drug development. Further, T-cell clonal expansion can also result from inflammatory benign conditions (11–15), thus understanding the transcriptome of expanded clones is important for distinguishing malignancy.

Characterizing single-cell transcriptomes addresses these problems by providing an unbiased and comprehensive map of heterogeneous and rare lymphocyte populations as well as cell states within each tumor sample. We previously employed droplet-based single-cell RNA sequencing (scRNA-seq) to profile the transcriptomes of thousands of individual cells from advanced-stage mycosis fungoides/Sézary syndrome skin tumors (16). Our analysis provided an unprecedented view of lymphocyte heterogeneity in individual tumors (16), but identification of the expanded malignant clones was not feasible with the technology available.

Here, we employed recent advances in scRNA-seq to simultaneously establish the T-cell immune repertoire and the transcriptome of expanded T-cell clones directly in mycosis fungoides skin tumors and in the leukemic blood of patients with Sézary syndrome. We identified a heterogeneous clonotype expansion within individual advanced-stage mycosis fungoides/Sézary syndrome samples in terms of number of expanded clones, their TCR V, D, J gene use and CDR3 sequences as well as their gene expression. All samples presented the expansion of a dominant malignant clone as well as of minor (less abundant) neoplastic and reactive clones. While we found upregulation of several patient-specific processes regulating cancer cell growth, proliferation, survival, and migration by individual malignant clones from mycosis fungoides/Sézary syndrome samples, we also detected activation of mycosis fungoides– or Sézary syndrome–specific gene expression signatures as well as pathways common among mycosis fungoides and Sézary syndrome malignant clones, suggesting shared tumorigenic mechanisms across mycosis fungoides/Sézary syndrome subtypes. Thus, advances in scRNA-seq provide new insights into mycosis fungoides/Sézary syndrome pathogenesis by identifying the transcriptional profile of malignant and reactive T lymphocytes as well as revealing novel prospective therapeutic targets tailored to individual patients.

Patient samples

Samples from 19 patients with confirmed diagnoses of advanced-stage (Stage IIB–IV) mycosis fungoides or Sézary syndrome (2) were obtained from the Comprehensive Skin Cancer Center, Columbia University Medical Center (Supplementary Table S1). For scRNA-seq studies, we employed 4 mycosis fungoides skin and 4 Sézary syndrome blood samples, including two Sézary syndrome patients in which we analyzed both blood and skin. Validation of transcriptional data was performed by immunofluorescence microscopy on 8 skin samples (mycosis fungoides n = 5 and Sézary syndrome n = 3) as well as on 3 Sézary syndrome blood samples. Skin samples were obtained by 4-mm punch biopsy and transported in phosphate-buffered saline prior to analysis. Controls included healthy control skin (HC, n = 8: n = 4 scRNA-seq and n = 4 immunofluorescence microscopy) and atopic dermatitis skin samples (AD; n = 4, immunofluorescence microscopy) obtained from The Health Sciences Tissue Bank, University of Pittsburgh as well as blood from HC donors (n = 3) obtained from the Central Blood Bank of Pittsburgh. Human research protocols were approved by the Institutional Review Boards of Columbia University and the University of Pittsburgh. All participants gave written informed consent in accordance with the Declaration of Helsinki.

scRNA-seq and immune repertoire analysis

Experimental procedures (Supplementary Methods) followed established techniques (16) using the Chromium Single Cell 5′ Library V1 kit (10x Genomics). RNA-seq was performed using the Illumina NovaSeq6000 system. Cell-gene unique molecular identifier counting matrices generated were analyzed using Seurat 3.1 to identify distinct cell populations using Louvain clustering (17–19). Immune repertoire matrices were added to the gene expression dataset and analyzed with Seurat to determine major clones and differential gene expression of clonal cells. For gene differential tests within clonotypes and within non-clonal cells, we used the “FindMarkers” function in Seurat, which uses the Wilcoxon rank sum test to show differential genes with a minimum percentage of cells of 10% per identity requested, log-fold change >0.1, and P value < 0.05 for significance. This was used for each differential gene list throughout the analysis (20–22).

Pathway analysis

The differential gene lists were filtered for P value < 0.05 for significance and then run in Ingenuity Pathway Analysis (IPA; Qiagen) for significant upregulated pathways. Pathways were selected by enrichment scores (–log P values) and absolute z-scores over 2 (23).

Pseudo-temporal trajectory analysis

Pseudo-time analysis was performed using the Monocle 3.0 R package (24–26). Genes differentially expressed across PhenoGraph-identified clusters were used as an input for the Monocle analysis. For the heat map representation of pseudo-time genes, a time trace of each gene was taken using the “plot_genes_in_pseudotime” function and dividing time into 100 equally sized bins. Time was measured by selecting the longest path through the trajectory plot going from t = 0 to t = max. The Monocle results show the sequence of gene expression changes that each cell must go through (their trajectories) as part of a dynamic biological process such as differentiation, relating each cell to other cells; larger distance between cells correlates to a larger difference in gene expression.

Multicolor immunofluorescence microscopy

Immunofluorescence staining (Supplementary Methods) was performed on formalin-fixed paraffin-embedded skin samples and on cytospin smears of CD4+ T cells from peripheral blood mononuclear cell samples using the tyramide signal amplification (Thermo Fisher) as previously described (16).

Data access

All scRNA-seq data have been deposited in the Gene Expression Omnibus: GSE182861.

Patient characteristics

The mean age of patients in the study group was 65.7 years. There were 12 males and 7 females. Nine mycosis fungoides skin samples were analyzed in this study. Of these 9 patients, 7 had received one or more therapies prior to biopsy. The most common prior treatments for mycosis fungoides patients included bexarotene (n = 6), mechlorethamine gel (n = 5), and PUVA (n = 4). Ten Sézary syndrome patient samples were analyzed in this study. Of these 10 patients, 7 had received prior therapies. The most frequent prior therapeutics observed in this cohort included bexarotene (n = 3) and romidepsin (n = 3; See Supplementary Table S1 for clinical characteristics).

Individual mycosis fungoides tumors exhibit heterogeneous T-cell clonotype expansion

We employed scRNA-seq to assess the transcriptional profile of expanded T-cell clonotypes within skin tumors from 4 patients with advanced-stage mycosis fungoides (ref. 2; Supplementary Table S1). ScRNA-seq was performed on enzymatically digested skin as previously described (16). Cell transcriptomes were analyzed by Seurat (16–19) and visualized by t-distributed stochastic neighbor embedding (t-SNE), capturing major cell-type populations across samples by expression of canonical marker genes of human skin (Supplementary Fig. S1) as previously described (16, 27, 28). We focused on the CD3+ T lymphocyte subset (see circled population in Supplementary Fig. S1C). Their transcriptional profile from each tumor sample showed limited overlap with the 4 HC skin samples (Fig. 1A). Simultaneous analysis of the T-cell immune repertoire allowed identification of the expanded clonotypes within individual tumors (Fig. 1B). Significantly, we identified the expansion of several nonoverlapping clonotypes in individual mycosis fungoides samples that included a dominant clonotype (clonotype 1, Ct1) as well as less abundant clonotypes. In contrast, no clonal expansion was detected in HC skin samples. In parallel, reciprocal principal component analysis demonstrated that the samples did not suffer from batch effects (Supplementary Fig. S2A and S2B). Supplementary Table S2A reports the most abundant expanded clonotypes from each mycosis fungoides tumor, their frequency within each sample, V, D, J gene usage and CDR3 sequences. We detected the expansion of two clonotypes in MF17: a major clonotype comprised the largest part of T cells in the sample (clonotype 1, 89.5%) while the less abundant clonotype 2 corresponded to 0.4% of total T cells. Ten expanded clonotypes were detected in MF18, including a dominant (clonotype 1, 21%) and 9 less frequent (1.1%–0.4%) clonotypes. MF21 comprised three expanded clonotypes with a frequency of 8%, 4%, and 1.1%, respectively. Finally, we detected five expanded clonotypes in MF24, ranging from 10% to 0.3% of total T cells. Strikingly, none of these expanded clonotypes showed the same V, D, J gene use and CDR3 sequences (Supplementary Table S2A). Cells from the dominant clonotype 1 of each mycosis fungoides sample were CD4+, while minor clones were either CD4+ or CD8+ (Supplementary Fig. S3; Supplementary Table S2A).

Figure 1.

Transcriptional profile of T-cell clones within individual mycosis fungoides tumors. Transcriptomes of 11,932 CD3+ cells (856 cells from 4 HC and 11,076 cells from 4 mycosis fungoides skin samples) were analyzed. A, Two-dimensional t-SNE shows dimensional reduction of reads from single cells, revealing grouping in each mycosis fungoides sample compared with all HC skin samples. Cells from each subject are indicated by different colors. B, T-cell immune repertoire analysis of CD3+ cells shows the expansion of specific alphabetaTCR clonotypes (Ct) within each mycosis fungoides sample (see Supplementary Table S2A). Only clonotypes with clonal size ≥10 cells are considered expanded; these are color coded, whereas gray color indicates no clonal expansion. C, Dot-plots showing the proportion of cells and the scaled average gene expression of signature genes from the expanded clonotypes versus HC cells or non-clonal T cells from the same tumor. Gene differential tests are described in Materials and Methods. D, The differential gene lists were filtered by P value <0.05 for significance and then run in IPA (Qiagen; ref. 23) for significant upregulated pathways. Highly significant examples of distinct pathways activated by the clones are shown. Pathways are represented by enrichment scores (–log P values) and selected by absolute z-scores over 2 (23).

Figure 1.

Transcriptional profile of T-cell clones within individual mycosis fungoides tumors. Transcriptomes of 11,932 CD3+ cells (856 cells from 4 HC and 11,076 cells from 4 mycosis fungoides skin samples) were analyzed. A, Two-dimensional t-SNE shows dimensional reduction of reads from single cells, revealing grouping in each mycosis fungoides sample compared with all HC skin samples. Cells from each subject are indicated by different colors. B, T-cell immune repertoire analysis of CD3+ cells shows the expansion of specific alphabetaTCR clonotypes (Ct) within each mycosis fungoides sample (see Supplementary Table S2A). Only clonotypes with clonal size ≥10 cells are considered expanded; these are color coded, whereas gray color indicates no clonal expansion. C, Dot-plots showing the proportion of cells and the scaled average gene expression of signature genes from the expanded clonotypes versus HC cells or non-clonal T cells from the same tumor. Gene differential tests are described in Materials and Methods. D, The differential gene lists were filtered by P value <0.05 for significance and then run in IPA (Qiagen; ref. 23) for significant upregulated pathways. Highly significant examples of distinct pathways activated by the clones are shown. Pathways are represented by enrichment scores (–log P values) and selected by absolute z-scores over 2 (23).

Close modal

To establish the transcriptional profile of each clone, we determined the differentially expressed genes (DEG) of each clone compared to all other T cells from the same tumor (see Experimental Design and Supplementary Table S3A). The gene signatures (Fig. 1C) and upregulated pathways (Fig. 1D) from individual clones within each tumor show that cells from all the dominant clones are enriched by pro-tumorigenic pathways, while cells from minor clones exhibit a transcriptional profile of either reactive or malignant cells. In MF17, the dominant CD4+ clonotype 1 (NUSAP1, MIK67, PASK, TNFSF11, NME1) upregulated eIF2 signaling, regulation of eIF4 and p70S6K, and mTOR signaling. MF17 clonotype 2 exhibited a transcriptional pattern (CD8, TIGIT, CST7, LAG3, PDCD1, GZMA) similar to reactive CD8+ T cells expressing inhibitory receptors and up-regulating pathways such as Th1/Th2, T-cell exhaustion and PD1/PDL1 signaling. MF18 clonotype 1 (LIF, PAGE5, IL9R, TMPRSS3, CD33) identified CD4+ cells activating pathways such as oxidative phosphorylation (OXPHOS), sirtuin, and HER-2 signaling in cancer. Most of the less dominant MF18 clonotypes were CD8+ (clonotypes 2–6) and upregulated several pathways associated with T-cell exhaustion, including PD1/PDL1 and CTLA4 signaling, and Th1/Th2 activation. They were distinguished by expression of specific checkpoint receptors, such as LAG3 and PDCD1, as well as by various levels of effector molecules (PRF1, GZMA, GZMB, IFNG). Some clonotypes (#3, 4, 5) upregulated iCOS/iCOSL and CD28 signaling and others (#3, 9, 10) the CDC42 signaling pathway. We also detected clone-specific pathways including eIF2 signaling and regulation of eIF4 and p70S6K (clonotypes 2, 6, 8), RhoA and sirtuin signaling (clonotype 10). Finally, clonotype 7 expressed CD4+ Treg genes (FOXP3, IL2RA, LRRC32, LAYN, IKZF2) and upregulated immunosuppressive pathways. The largest MF21 clone (RAB32, KIR3DL2, CDCA7, MCM7, CCDC50) activated OXPHOS, sirtuin and kinetochore metaphase signaling, indicative of proliferation. Clonotype 2 (HAVCR2, PDCD1, LAG3, IFNG, PRF1) upregulated Th1/Th2 and immunosuppression pathways; while clonotype 3 (KRAS, RRAS2, SOX4, EGR2, IL13) activated IL17, HMGB1, and Th2 signaling. Major upregulated pathways by MF24 CD4+ clonotype 1 (AUTS2, CCND1, HDAC9, TNFSF11, RORB) included OXPHOS, BAG2 signaling and cell-cycle control of chromosomal replication. Examples of significant processes activated by MF24 minor clonotypes included Th1/Th2 and T-cell exhaustion (clonotype 2: KIR2DL3, KIR2DL4, VCAM1, IFIH1, NBL1); kinetochore metaphase and FAT10 signaling, polyamine regulation in cancer (clonotype 3: CENPU, CDKN2B, PCLAF, CKS1B TYMS); and OXPHOS, Th1, and PD1/PDL1 pathways in clonotype 4 (CXCR3, CXCR6, FASLG, GCSAM, SEMA4A). Finally, clonotype 5 (CD38, SART3, SNAPC5, IKBIP, IL10) induced processes such as the role of Tissue factor in cancer, FLT3 in hematopoietic progenitor cells, and chronic myeloid leukemia.

To understand clonotype expansion in the context of T-cell differentiation in mycosis fungoides tumors, we performed a trajectory analysis using Monocle 3 (24). All T cells from mycosis fungoides and HC samples were placed on these trajectories on the basis of changes in their transcriptome (Fig. 2A). Interestingly, T cells from HCs were distributed to early pseudotimes, whereas mycosis fungoides T cells were enriched in late pseudotimes, showing a clear temporal separation (Fig. 2B). Strikingly, we observed a major trajectory branch from each tumor sample that corresponded to cells from the dominant clonotype, indicating distinct gene expression by these cells (Fig. 2C). We next determined expression of some of the signature genes from the dominant clones along the pseudotime. In MF17, we established that cells from the major differentiation branch corresponded to a proliferating subset (MIK67, PCLAF, TYMS) within clonotype 1 (GCSAM, TNFSF11, NET1; Fig. 2D; Supplementary Fig. S4A). Likewise, cells from the major differentiation branch in MF18 (PAGE5, LIF, IL1RL1), MF21 (HACD1, KIR3DL2, RAB32), and MF24 (TNFSF11, KIR3DL2, RORB) expressed genes specific to the dominant clone and markers of proliferation (PCLAF, TYMS, CCNA2, CCND1, RRM2, MIK67, CDCA7; Fig. 2DG; Supplementary Fig. S4A). Notably, apart from MF17, the main clonotype branch from each mycosis fungoides sample appeared to be developmentally connected to cells expressing the minor clonotypes. A selection of gene markers identifying the less frequent clones within each tumor along the pseudotime is shown in Supplementary Fig. S4B. Interestingly, we detected a second differentiation branch in MF24 that appeared to diverge from the dominant malignant clone. This branch comprised non-proliferating CD3+CD4CD8IL7RA+ cells that differentially expressed markers of cytotoxicity (GZMB, GZMA, PRF1, IFNG, FASLG), natural killer (NK) receptors (KLRC1, KLRD1, TYROBP, FCGR3A, DNAM1) as well as TRDC and TRGC1, likely representing gamma/delta T cells (ref. 29; Supplementary Fig. S5). TCR gamma IHC was performed on this patient's tumor samples at the time of biopsy and a monoclonal TCR gamma chain rearrangement was detected. The patient was initially biopsied when his previously indolent disease had become aggressive, following the abrupt development of new and worsening tumors. Therefore, a potential phenotypic switch was suggested clinically that corresponded with the gamma/delta phenotype observed in the second differentiation branch. Thus, our transcriptome analysis identified the specific gene expression of expanded clonotypes in the mycosis fungoides tumor microenvironment (TME).

Figure 2.

Gene expression dynamics of mycosis fungoides clonotype expansion along the pseudotime. A, Single-cell pseudotime trajectories of T lymphocytes derived from mycosis fungoides skin tumors and HC skin estimated using Monocle 3 (24). A continuous value from 0 to 20 was assigned to each cell as a pseudotime. B, Sample identity and (C) clonotype expression along the pseudotime. D–G, Expression dynamics of clonotype-specific or proliferation markers along the pseudotime of individual mycosis fungoides and HC T cells.

Figure 2.

Gene expression dynamics of mycosis fungoides clonotype expansion along the pseudotime. A, Single-cell pseudotime trajectories of T lymphocytes derived from mycosis fungoides skin tumors and HC skin estimated using Monocle 3 (24). A continuous value from 0 to 20 was assigned to each cell as a pseudotime. B, Sample identity and (C) clonotype expression along the pseudotime. D–G, Expression dynamics of clonotype-specific or proliferation markers along the pseudotime of individual mycosis fungoides and HC T cells.

Close modal

The dominant clonotypes from mycosis fungoides tumors express a common gene expression signature

While the transcriptomes of all HC samples coincided, the transcriptional profile of T cells from patient samples only partially overlapped with each other (Fig. 3A), and in particular, the transcriptional profiles of the dominant clones from each mycosis fungoides tumor showed no overlap (Fig. 3B). However, transcriptome comparison identified 258 genes in common (Fig. 3C; Supplementary Table S4A), including those involved in OXPHOS (ATP5F1, COX5A, NDUFA4), pyrimidine deoxyribonuclease (NME1, RRM2, TYMS), and dTMP (DHFR, SHMT2, TYMS) de novo biosynthesis, remodeling of epithelial adherens junctions (ACTB, ARPC1B, TUBB) as well as genes associated with BAG2, sirtuin, and FAT10 signaling cascades. We focused on 4 of these common genes, PCLAF, OCIAID, DHFR, and TFDP1, whose expression was high in the dominant clonotype but low in the other T cells within the tumors and HCs (Fig. 3D). We validated these transcriptional data by multicolor immunofluorescence microscopy in advanced mycosis fungoides tumor samples using TOX and Ki67 to identify tumor lymphocytes (16). Representative examples from 5 mycosis fungoides tumor samples confirmed transcriptional results by showing coexpression of either PCLAF, OCIAID2, TFDP1, or DHFR with TOX and Ki67 (Fig. 3EH). In comparison, HC or AD skin was negative for expression of these markers. Thus, simultaneous analysis of T-cell clonality and gene expression identifies common signature genes among the major expanded clonotypes of mycosis fungoides tumors.

Figure 3.

Major T-cell clones from individual mycosis fungoides tumors express a common gene expression signature. A, Transcriptomes of T lymphocytes of all 4 mycosis fungoides and all 4 HC skin samples combined. B, Gene expression of dominant T-cell expanded clonotypes from all mycosis fungoides tumors combined. C, Venn diagram showing overlap of expressed genes by T lymphocytes from the dominant expanded clone from each mycosis fungoides tumor. D, Dot-plot showing the proportion of cells and the scaled average gene expression of 4 examples of common genes (C) identified in cells from the dominant clonotype from each patient sample. E–H, Multicolor immunofluorescence microscopy staining for DHFR, PCLAF, TFDP1, or OCIAID2 in advanced mycosis fungoides samples (n = 5), HC (n = 4), and AD (n = 3) skin. Tumor lymphocytes were identified by coexpression of TOX and Ki67, as indicated. Representative examples are shown (1,000×). DAPI stains nuclei.

Figure 3.

Major T-cell clones from individual mycosis fungoides tumors express a common gene expression signature. A, Transcriptomes of T lymphocytes of all 4 mycosis fungoides and all 4 HC skin samples combined. B, Gene expression of dominant T-cell expanded clonotypes from all mycosis fungoides tumors combined. C, Venn diagram showing overlap of expressed genes by T lymphocytes from the dominant expanded clone from each mycosis fungoides tumor. D, Dot-plot showing the proportion of cells and the scaled average gene expression of 4 examples of common genes (C) identified in cells from the dominant clonotype from each patient sample. E–H, Multicolor immunofluorescence microscopy staining for DHFR, PCLAF, TFDP1, or OCIAID2 in advanced mycosis fungoides samples (n = 5), HC (n = 4), and AD (n = 3) skin. Tumor lymphocytes were identified by coexpression of TOX and Ki67, as indicated. Representative examples are shown (1,000×). DAPI stains nuclei.

Close modal

Peripheral blood of patients with Sézary syndrome exhibits heterogeneous T-cell clonotype expansion

We next assessed clonal expansion in the leukemic blood of patients with advanced-stage Sézary syndrome (ref. 2; Supplementary Table S1; Supplementary Fig. S6A and S6B). The T-cell transcriptomes of individual Sézary syndrome samples minimally overlapped with those of HC T cells as well as with each other (Fig. 4A and B). As for mycosis fungoides tumors, we found a heterogeneous, nonoverlapping T-cell clonotype expansion among Sézary syndrome samples (Supplementary Table S2B; Fig. 4C and D; Supplementary Fig. S2C and S2D). As observed in mycosis fungoides tumors, each Sézary syndrome sample exhibited the expansion of a dominant clone (clonotype 1, Ct) as well as of less abundant clonotypes. Moreover, we established that cells from the dominant clone were CD4+ in all patient samples, while minor clones were either CD4+ or CD8+ (Supplementary Fig. S6C and S6D). By comparing the average gene expression of each clone to that of all other T cells in the same blood sample, we established a gene expression signature for each expanded clonotype in individual Sézary syndrome tumors (Supplementary Table S3B; Fig. 4E; Supplementary Fig. S7A). The SS1 CD4+ dominant clone 1 (14.8%, BIRC3, MYC, PAGE5, TOX, HACD1) upregulated several pathways associated with TCR and HER-2 oncogene signaling. Of the three minor CD8+ T-cell clones (1.4%–0.5%), clonotype 2 (TNFSF12, CLCF1, TNK2, IFNG, NCOA4) represented exhausted benign CD8+ T cells, while clonotypes 3 (CD226, HAVCR2, NBPF3, SESN2, SH3RF2) and 4 (MCM7, CDK5, CDKN3, PCNX2, MTIF2) upregulated oncogenic pathways including eIF protein and mTOR signaling as well as cell-cycle control of chromosomal replication. Two CD4+ T-cell malignant clonotypes were expanded in SS2, including a larger clonotype 1 (79.6%, MAL, NELL2, HACD1, PCSK1N, PIM2) and a minor clone 2 (1.05%, FGFBP2, GNLY, GZMB, GZMH, SPON2), which upregulated eIF and HER-2 pathways (clone 1) as well as polyamine regulation in cancer and FAT10 signaling (clone 2). The two SS3 CD4+ T-cell clones activated OXPHOS, estrogen receptor signaling and hereditary breast cancer pathways (clone 1, 4.5%, KIR3DL2, TOP2A, CDCA7, IGFBP4, PRSS21) as well as kinetochore metaphase and HER-2 signaling (clone 2, 0.8%: CRNDE, RARG, MAP2K2, CCR10, THEMIS). Finally, in addition to the eIF pathways, the malignant SS4 clonotype 1 (21.3%, FOXP3, TOX, CUL9, PLS3, HACD1) upregulated mTOR and BAG2 signaling. The three additional CD8+ T-cell clones (14.8–0.5%) expressed genes associated with oncogenic pathways (clonotype 2: KIR3DL2, CDKN2A, FAM111, TTC16, TPRG1; clonotype 3: CDC6, CEMPW, PERP, DAP, CENPN; clonotype 4: CDCA4, MCM4, POLR2M, KLRC3, DTHD1).

Figure 4.

Transcriptional profile of T-cell clones within individual Sézary syndrome blood samples. Transcriptomes of T cells from the leukemic blood of 16,809 CD3+ cells from 3 HC (3,383 cells) and 4 Sézary syndrome (13,426 cells) blood samples were analyzed. A, t-SNE reveals grouping in each Sézary syndrome sample compared with all HC skin samples. All samples combined (B). Cells from each subject are indicated by different colors. C, T-cell immune repertoire analysis of CD3+ cells shows the expansion of specific alphabetaTCR clonotypes within each Sézary syndrome sample or in all combined samples (D). E, Dot-plots showing the proportion of cells and the scaled average gene expression of signature genes from the expanded clonotypes versus HC cells or non-clonal T cells from the same Sézary syndrome blood sample. F, Venn diagram showing overlap of expressed genes by T lymphocytes from the dominant expanded clone from each Sézary syndrome sample. G, Dot-plot showing the proportion of cells and the scaled average gene expression of 4 examples of common genes identified in 4F. Multicolor immunofluorescence microscopy staining for MFSD1, TOX, and YBX1 (H) and ID3, TOX, and PGK1 (I) in the peripheral blood (n = 3) and skin lesions (n = 3) of advanced Sézary syndrome samples, as well as HC blood and skin (n = 3 each). Tumor lymphocytes were identified by coexpression of TOX, as indicated. Representative examples are shown (1,000×). DAPI stains nuclei.

Figure 4.

Transcriptional profile of T-cell clones within individual Sézary syndrome blood samples. Transcriptomes of T cells from the leukemic blood of 16,809 CD3+ cells from 3 HC (3,383 cells) and 4 Sézary syndrome (13,426 cells) blood samples were analyzed. A, t-SNE reveals grouping in each Sézary syndrome sample compared with all HC skin samples. All samples combined (B). Cells from each subject are indicated by different colors. C, T-cell immune repertoire analysis of CD3+ cells shows the expansion of specific alphabetaTCR clonotypes within each Sézary syndrome sample or in all combined samples (D). E, Dot-plots showing the proportion of cells and the scaled average gene expression of signature genes from the expanded clonotypes versus HC cells or non-clonal T cells from the same Sézary syndrome blood sample. F, Venn diagram showing overlap of expressed genes by T lymphocytes from the dominant expanded clone from each Sézary syndrome sample. G, Dot-plot showing the proportion of cells and the scaled average gene expression of 4 examples of common genes identified in 4F. Multicolor immunofluorescence microscopy staining for MFSD1, TOX, and YBX1 (H) and ID3, TOX, and PGK1 (I) in the peripheral blood (n = 3) and skin lesions (n = 3) of advanced Sézary syndrome samples, as well as HC blood and skin (n = 3 each). Tumor lymphocytes were identified by coexpression of TOX, as indicated. Representative examples are shown (1,000×). DAPI stains nuclei.

Close modal

Although we found that the transcriptional profile of the major clones did not overlap with each other (Fig. 4D), they exhibited 59 genes in common (Fig. 4F; Supplementary Table S4B), including genes associated with RhO family GTPases (EZR, MYL12A, SEPTIN9, VIM) and sirtuin (ATP5F1C, PGK1, SLC25A5, SLC25A6) signaling as well as those regulating the epithelial–mesenchymal transition pathways (LTB, TNFSF10, VIM). We focused on MFSD10, PKG1, ID3, YBX1, which show little to no expression by other T cells within the same tumor (Fig. 4G), and we validated this expression by multicolor immunofluorescence microscopy in several samples from patients with Stage IV Sézary syndrome, both in the leukemic blood as well as in the affected skin lesions (Fig. 4H and I). Thus, scRNA-seq identifies tumor-specific and common gene expression signatures of expanded malignant clonotypes from the blood of patients with advanced-stage Sézary syndrome.

Comparing T-cell clonotype expansion between the blood and skin of patients with Sézary syndrome

We analyzed the relationship between clonotype expansion in blood and in the erythrodermic skin of patients with advanced Sézary syndrome. While the same dominant clonotype (ct1) was expanded in the blood and the skin of each patient (SS1: 15% blood, 40% skin and SS2: 79.6% blood, 25% skin of total TCR+ cells), only blood exhibited the expansion of additional clones (Supplementary Table S2B; Fig. 5A and B). Although the transcriptional profiles partially overlapped between the blood and skin from each patient (Fig. 5A and B), we observed that a considerable number of genes from clonotype 1-positive cells were commonly expressed between the two compartments (SS1 n = 688, SS2 n = 463). Examples of common genes between blood and skin in SS1 included CCND3, PAGE5, MAL, HACD1, PIM3, and in SS2: NELL2, MAL, NME2, PLK2, MYC (Fig. 5C). Besides detecting activation of patient-specific pathways (SS1: Th1/Th2 activation, signaling by RhO family GTPases, FAT10 signaling; SS2: eIF2 and mTOR signaling, regulation of eIF4 and p70S6K), we also found upregulation of common pathways between patients, including OXPHOS, sirtuin, and HER-2 signaling (Supplementary Fig. S7B).

Figure 5.

scRNA-seq profiles T cells from Sézary syndrome peripheral blood and lesional skin. A and B, Transcriptional profile and clonotype expansion of T cells from the leukemic blood and the lesional skin of advanced-stage Sézary syndrome patients (n = 2; SS1 and SS2). C, Dot-plots showing the proportion of cells and the scaled average gene expression of a selection of common genes from the clonotype 1–positive cells of blood and skin in SS1 and SS2. D, Single-cell pseudotime trajectories of T lymphocytes derived from Sézary syndrome blood samples and skin lesions, estimated using Monocle 3. A continuous value from 0 to 20 was assigned to each cell as a pseudotime. E, Sample identity and (F) clonotype expression along the pseudotime. G and H, Expression dynamics of genes associated with clonotype 1 expansion or skin homing receptors in SS1 (G) and SS2 (H).

Figure 5.

scRNA-seq profiles T cells from Sézary syndrome peripheral blood and lesional skin. A and B, Transcriptional profile and clonotype expansion of T cells from the leukemic blood and the lesional skin of advanced-stage Sézary syndrome patients (n = 2; SS1 and SS2). C, Dot-plots showing the proportion of cells and the scaled average gene expression of a selection of common genes from the clonotype 1–positive cells of blood and skin in SS1 and SS2. D, Single-cell pseudotime trajectories of T lymphocytes derived from Sézary syndrome blood samples and skin lesions, estimated using Monocle 3. A continuous value from 0 to 20 was assigned to each cell as a pseudotime. E, Sample identity and (F) clonotype expression along the pseudotime. G and H, Expression dynamics of genes associated with clonotype 1 expansion or skin homing receptors in SS1 (G) and SS2 (H).

Close modal

To understand T-cell clonotype expansion in blood and skin in the context of Sézary syndrome, we performed single-cell trajectory analysis. All T cells were placed on these trajectories based on changes of their transcriptomes, and a clear temporal separation was noted between blood and skin (Fig. 5D and E). In both samples, clonotype 1–positive cells from blood and skin were shifted towards a more differentiated status compared to non-clonal T cells or cells from minor clones (Fig. 5F). We next examined the transition of average expression values along pseudotime for a panel of signature genes associated with clonotype 1 expansion in SS1 and SS2 (Fig. 5G and H; Supplementary Fig. S8). Genes such as HACD1, PAGE5, PLS3, TSHZ2, IGFBP4 were specifically expressed by cells from SS1 in both blood and skin, while HACD1, PLK2, ID, GEM, PDCD1 were specific to SS2. In both samples, expression of genes such as TCF7, MAL, MYC, BIRC3, CTLA4, ITM2C, and PIM2, appeared to be associated with the development of the dominant clone. Clonotype 1–positive cells also expressed CCR7, CCR4, CXCL13, and SELL indicating a skin-homing recirculating phenotype and expression of TOX confirmed their malignant nature. Thus, we identified a gene expression signature associated with the development of the major malignant clone in Sézary syndrome blood, which recirculates between blood, skin, and lymph nodes.

Transcriptional profile of non-clonal T cells in advanced mycosis fungoides/Sézary syndrome tumors

We next assessed the transcriptional profile of non-clonal T cells within each mycosis fungoides/Sézary syndrome sample. We found that profiles of either non-clonal CD4+ or CD8+ T cells do not overlap between mycosis fungoides/Sézary syndrome samples and HCs (Fig. 6AD). Seurat analysis of combined patient and HC samples identified distinct Louvin clusters, of which we depicted a selection of the top DEGs as well as pathways that were significantly upregulated in each of these clusters (Fig. 6E and F; Supplementary Fig. S9). In mycosis fungoides, we identified a large non-clonal CD4+ T-cell subset (cluster 0) containing cells from all patients that upregulated Th1/Th2 and TCR/CD28 signaling as well as patient-specific CD4+ clusters (MF18, cluster 3; MF24, clusters 1, 2; Fig. 6A). The MF18–specific cluster 3 differentially expressed markers associated with T follicular helper (TFH)-like cells (CXCL13, CD200, PDCD1), while the MF24-specific CD4+ clusters upregulated NRF2-mediated stress response, ferroptosis, and TNFR2 signaling (cluster 1), as well as markers of gamma/delta T cells (KLRC1, KLRD1, CLEC2B, TRDC; cluster 3). Similarly, we found a large cluster of CD8+ T cells derived from all mycosis fungoides tumors (cluster 0) as well as patient-specific CD8+ clusters (MF18, cluster 1; MF24, clusters 2, 4; Fig. 6B). Cluster 0 (CD27, LAG3, PARK7) activated OXPHOS, PD1/PDL1 and glucocorticoid receptor signaling. Examples of patient-specific pathways included TNFR2 and apoptosis signaling (MF18, cluster 1); as well as ferroptosis, autophagy (MF24, cluster 2), mTOR and NK signaling (MF24, cluster 4). In contrast to the mycosis fungoides skin samples, each of the Sézary syndrome blood samples exhibited distinct non-clonal CD4+ (Fig. 6C and E) and CD8+ (Fig. 6D and F) T-cell clusters, with little to no overlap in gene expression among samples and with HCs. In SS1, CD4+ cells (cluster 2: SESN1, IL4R, KLF10) upregulated TNFR2, IL10, and CD40 signaling while the two SS1-specific CD8+ clusters, cluster 3 (NKG7, ITGAM, KLRC1) and 5 (WHAMM, NFKBIZ, SERINC5), respectively, activated cytotoxicity, integrin, and NAD pathways as well as TNFR2, IL9 and IL10 signaling. SS2 CD4+ cells (cluster 4: DUSP4, PTGER4, CSRNP1) upregulated TNFR2, TWEAK, and APRIL signaling, whereas CD8+ cells (cluster 2: IFNG, RGS1, NR4A3) upregulated PD1/PDL1 and Th2 pathways, including IL4 signaling. While SS3-specific CD4+ cells represented Tregs (cluster 1: TIGIT, CTLA4, FOXP3), both CD4+ and CD8+ cells exhibited Th1/Th2 activation and PD1/PDL1 signaling. Interestingly, SS4 CD4+ (cluster 5: DPP4, CDKN1A, IER3) and CD8+ (cluster 1: XCL1, EOMES, CCL3) cells both upregulated signaling through the estrogen receptor and specifically upregulated IFN and NFR2-mediated oxidative stress pathways (CD4+) as well as cytotoxicity and RhO family GTPase signaling (CD8+).

Figure 6.

Transcriptional profile of non-clonal T cells in the TME of mycosis fungoides/Sézary syndrome tumors. Transcriptomes of non-clonal CD4+ (A) and CD8+ (B) T lymphocytes from individual mycosis fungoides tumor or non-clonal CD4+ (C) and CD8+ (D) T cells from Sézary syndrome blood samples (color coded by subject), revealing discrete Louvain clusters. E and F, Pathway analysis by Ingenuity for significant upregulated pathway of the non-clonal CD4+ and CD8+ T-cell clusters that are mycosis fungoides–specific (E) or Sézary syndrome–specific (F). Highly significant examples are shown. Pathways are represented by enrichment scores (–log P values) and selected by absolute z-scores over 2 (23).

Figure 6.

Transcriptional profile of non-clonal T cells in the TME of mycosis fungoides/Sézary syndrome tumors. Transcriptomes of non-clonal CD4+ (A) and CD8+ (B) T lymphocytes from individual mycosis fungoides tumor or non-clonal CD4+ (C) and CD8+ (D) T cells from Sézary syndrome blood samples (color coded by subject), revealing discrete Louvain clusters. E and F, Pathway analysis by Ingenuity for significant upregulated pathway of the non-clonal CD4+ and CD8+ T-cell clusters that are mycosis fungoides–specific (E) or Sézary syndrome–specific (F). Highly significant examples are shown. Pathways are represented by enrichment scores (–log P values) and selected by absolute z-scores over 2 (23).

Close modal

Altogether, non-clonal reactive CD4+ and CD8+ mycosis fungoides/Sézary syndrome lymphocytes are characterized by a predominant anti-inflammatory and exhausted phenotype; however, tumor-specific reactive subsets are also detected.

Clonal malignant T lymphocytes make up only a fraction of T cells infiltrating mycosis fungoides skin lesions and the peripheral blood of patients with Sézary syndrome (30, 31), but determining their transcriptional profiles has only recently been possible by scRNA-seq methods that allow high resolution profiling of the T-cell immune repertoire simultaneously with gene expression. With this approach, we showed here that several nonoverlapping clonotypes are expanded in the skin and blood of individual advanced-stage mycosis fungoides/Sézary syndrome patient samples, including a dominant malignant clonotype as well as additional minor reactive and malignant clones. These results are consistent with previous work assessing clonal heterogeneity by 4-color TCRgamma PCR assay (32) and by whole-exome sequencing but contrast with other studies indicating that mycosis fungoides/Sézary syndrome are monoclonal (6, 31, 33). We propose that multiple clonotype expansion in mycosis fungoides/Sézary syndrome results from antigen-driven processes rather than by secondary somatic mutations of assembled CDR3 regions. Genetic damage, genomic destabilization (34, 35), and/or therapeutic intervention might promote the neoplastic transformation from a pool of benign reactive lymphocytes, thus facilitating the generation of one or more malignant clones. In this context, a recent study employing a multimodal single-cell analysis showed that environmental cues along with genetic aberrations likely contribute to transcriptional profiles of malignant T cells of individual patients with mycosis fungoides/Sézary syndrome (36). Our observations that the rearranged clonal TCRs comprised nonoverlapping combinations of various germline genes supports our hypothesis and, moreover, we did not detect expression of RAG1/2 genes in the expanded clonotypes, indicating that TCRB post-thymic recombination in lymphocytes was unlikely (37). Finally, we observed that the dominant clones appear to be developmentally connected to cells from the minor benign clones within most tumors, suggesting a potential transformation of benign reactive lymphocytes. We note that several nonmalignant clonotypes are expanded in the skin lesions of the mycosis fungoides samples as well as in the Sézary syndrome blood samples, particularly within CD8+ T cells. Clonal inflammatory T cells may result from several mechanisms including chronic dermatitis or autoimmune conditions (11, 12, 14, 15), age-dependent clonal expansion of cytomegalovirus-specific CD8+ T cells (38) or from activation of skin-resident T cells by superantigens (39). Bacterial infections, frequently by Staphylococcus aureus, are a major source of morbidity and mortality in mycosis fungoides/Sézary syndrome and promote mycosis fungoides/Sézary syndrome progression (39, 40). While most patients tested presented a history of bacterial infection and antibiotic treatment, MF18 exhibited expression of superantigen-reactive TRVB segments (41) such as TRBV5–1, TRBV5–4, and TRBV5–6 by cells from minor T-cell clonotypes (#5, 10, 6). Likewise, clonotype 1 in SS1 blood and skin expressed TRBV5–4 and TRAV9–2 and in SS2 expressed TRBV20–1 (3). Therapeutic intervention may also provide potent selective pressure for the expansion of resistant cancer clones (42); however, we also detected multiple clones in two untreated patients (MF18 and SS1). Although most studies using PCR-based methods found identical clones in different lesions from the same patients (43, 44), clonal heterogeneity was found in multiple concurrent skin samples (32). Thus, larger studies using scRNA-seq methods are needed to establish the relationship between disease progression and clonal expansion in different mycosis fungoides lesions.

Although the clinical and histopathologic presentations of mycosis fungoides and Sézary syndrome are nonspecific, they are classified separately (2) on the basis of peripheral blood involvement and distinct biological and molecular features (1, 45, 46). However, we found that the gene expression by malignant clones could be grouped in three ways according to: (i) patient-specific processes regulating cancer cell growth, proliferation, survival, and cell migration; (ii) processes distinguished by the mycosis fungoides/Sézary syndrome classification; and (iii) those common to mycosis fungoides/Sézary syndrome. The sharing of common oncogenic pathways suggests that these two diseases are related at some deeper level and indeed some of the patients with Sézary syndrome had a preceding mycosis fungoides diagnosis. However, larger patient cohorts will be needed to test this hypothesis.

Consistent with the mycosis fungoides/Sézary syndrome classification, we detected gene signatures specific to mycosis fungoides or Sézary syndrome that likely represent major drivers for their development and progression. In mycosis fungoides these signatures included markers of malignancy identified in other human cancers such as OCIAD2, ARMC10, OLA1, MT2A, HMGA1 as well as genes associated with cell-cycle progression (RRM2, RAN, CDK4, TFDP1, TYMS, UBEC2), proliferation (PCLAF, PCNA, MIK67, NME1), or genes associated with DNA replication, repair, and chromosomal stability (SEM1, H2AFY, MCM7, BAZ1B, H3F3A, PPP4C). In addition to the expression of already known Sézary syndrome markers such as PLS3, KIR3DL2, TOX, GATA3, MYC, and tissue homing receptors (CCR7, SELL, CXCL13, CCR4), recurrent genes in Sézary syndrome clonal cells included those associated with resistance to apoptosis (PAGE5, MFSD10), cell proliferation and division (IGFBP4, ID3, EMP3, CDC25B, PPP1CC), as well as DNA damage and repair (FKBP1A, UBC, TEX264, ROMO1). Interestingly, the dominant clone from all Sézary syndrome samples expressed HACD1, a protein tyrosine phosphatase not previously associated with Sézary syndrome that by facilitating fatty acid metabolism may support tumorigenesis and tumor progression through membrane biosynthesis, generation of signaling intermediates, and production and storage of energy (47).

Finally, pathways common to mycosis fungoides and Sézary syndrome included HER-2 signaling that is involved in proliferative, survival, metabolic, and invasive functions of many human cancers (48), and sirtuin signaling that is linked to various tumor processes including epithelial–mesenchymal transition, invasion, and metastasis (49). Significantly, we found that the epithelial adherens remodeling pathway is also upregulated, suggesting major interconnected processes occurring in malignant cells. Most cancer cells derive energy from rapid glycolysis (50), which although much less efficient at generating ATP than OXPHOS nonetheless provides the metabolites necessary for fast proliferation of cancer cells. In agreement with recent studies showing that OXPHOS is also upregulated in certain cancers (51), we found that malignant clones from all mycosis fungoides/Sézary syndrome samples upregulated OXPHOS. In parallel, to compensate for increased energy usage and their need to synthesize large amounts of nucleotides to support DNA replication and RNA production, malignant T-cell clones also upregulated de novo nucleotide biosynthesis (52). Lastly, dominant and minor malignant clones upregulated signaling through the eIF proteins, particularly eIF2 as well as eIF4 and p70S6K and their master regulator mTOR (53). Notably, aberrant expression and/or phosphorylation of eIF proteins are involved in malignant transformation, tumor initiation, progression, and metastasis by stimulating the expression of antiapoptotic and pro-invasion proteins (53). Thus, multiple malignant clones and several oncogenic processes within each clone all contribute to tumor development, progression, and heterogeneity. Consequently, lack of long-term efficacy of treatments targeting specific oncogenic pathways may be explained by the presence of other non-targeted but coexisting tumorigenic pathways. In addition, the successful eradication of a fast-growing clone may nonetheless result in the subsequent emergence of a minor clone. In both cases, this may lead to disease relapse.

Large numbers of non-clonal CD4+ and CD8+ T cells infiltrate the skin lesions of patients with mycosis fungoides as well as the leukemic blood of patients with Sézary syndrome. We observed that the majority of non-clonal T cells from all mycosis fungoides tumors exhibited similar gene expression within the CD4+ and CD8+ lineages, and both signatures were characterized by a memory phenotype and antitumor Th1 and cytotoxic pathway activation while also upregulating expression of checkpoint receptors as well as several anti-inflammatory and immunosuppressive mechanisms such as glucocorticoids and sex hormone signaling. Conversely, non-clonal CD4+ and CD8+ T cells from Sézary syndrome blood exhibited mainly activation of patient-specific pathways. These included signaling of distinct anti-inflammatory cytokines such as IL9, IL10, IL4, and IFN, pro-tumorigenic, and antiapoptotic processes induced by TNF family members such as TWEAK (54) and APRIL (55) as well as immunosuppressive pathways by PD1/PDL1 signaling. Patient-specific immune responses in mycosis fungoides included the MF18-specific CD4+ cluster that expressed markers of CXCL13+ TFH cells, which were previously identified in the TME of breast cancer and were associated with a good prognosis (56). Interestingly, we found that several patient-specific non-clonal CD4+ and CD8+ T-cell clusters activated the TNFR2 signaling cascade, which plays an import role in CD4+FOXP3+Tregs as well as in CD8+ T-cell effectors (CD8eff) and CD8+FOXP3+Tregs (57, 58). TNFR2+CD4+Tregs were detected in SS1 and SS2 blood samples as well as in the TME of MF24. Significantly, genomic analysis of mycosis fungoides and Sézary syndrome identified recurrent alterations in TNFR2 (59) and additionally TNFR2 antagonistic antibodies induced a marked responsiveness of Sézary cells and Tregs in vitro (57), thus TNFR2 represents an important prospective target for mycosis fungoides/Sézary syndrome therapy. However, we detected only negligible expression of FOXP3 by MF24 CD8+ cells, indicating that the TNF/TNFR2 signaling pathway is likely involved in reinforcing the cytotoxicity of CD8effs via activation of PI3K/AKT and/or by inducing apoptotic signals to terminate the immune response (57, 58). Both MF24-specific CD4+ and CD8+ tumor infiltrating lymphocytes also upregulated NRF2-mediated oxidative stress, which governs reprogramming of cancer cell metabolism (60), and ferroptosis, an iron-dependent form of cell death that can cause inflammation-associated immunosuppression in the TME (61). Interestingly, we detected a subset of CD3+CD4CD8IL7R+ lymphocytes in MF24 that expressed NK cytotoxicity receptors and markers of cytotoxicity. This subset, which diverged from the dominant malignant clone in pseudo-temporal maturation trajectories, likely identifies gamma/delta T lymphocytes (29). Although gamma/delta T cells are difficult to distinguish from CD8+ and NK cells due to their overlapping transcriptomes, and further the scRNA-seq methodology used does not detect rearrangement at the TRD and TRG loci, nonetheless cells from this cluster differentially expressed the TRDC and TRGC1 gene-segments that encode for the constant regions of the TCRdelta and gamma chains (29). Although we cannot exclude the malignant nature of these cells, histopathologic and clinical features did not support the differential diagnosis of either NK/T-cell lymphoma or gamma/delta CTCL in patient MF24 (62, 63). An alternative hypothesis is that they may represent a reactive gamma/delta T-cell subset that infiltrates the TME of MF24. Indeed, gamma/delta T cells possess powerful cytotoxic and pro-inflammatory activities that allow them to kill a broad range of tumor cells; however, immunosuppressive signals originating from the TME can limit their effectiveness in several solid cancers (64).

In summary, our analysis reveals new insights into mycosis fungoides/Sézary syndrome pathogenesis by providing an unprecedented report on the transcriptional profile of malignant and benign cells within the TME of each tumor, demonstrating the power of scRNA-seq to distinguish between tumor heterogeneity and clonality in mycosis fungoides/Sézary syndrome tumorigenesis. This complexity reflects genetic and environmental influences that may drive development of malignant clones. In particular, our observations of polyclonality as well as upregulation of patient-specific and common oncogenic pathways may be critical for developing appropriate therapy tailored to specific patients. Further studies would validate and extend these results by expanding the number of patient samples in following candidate biomarkers for uncovering mechanisms of pathogenesis and progression as well as for therapeutic leads.

No disclosures were reported.

A.M. Gaydosik: Formal analysis, validation, investigation, writing–review and editing. C.J. Stonesifer: Data curation, writing–review and editing. A.E. Khaleel: Data curation. L.J. Geskin: Data curation, writing–review and editing. P. Fuschiotti: Conceptualization, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.

This work was supported by NIH/NCI grant R21 CA209107-02 (to P. Fuschiotti), and by the Cutaneous Lymphoma Foundation Catalyst Research Award (to P. Fuschiotti). We thank Dr. Robert Lafyatis and Tracy Tabib for discussions and assistance with the scRNA-seq.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Campbell
JJ
,
Clark
RA
,
Watanabe
R
,
Kupper
TS
.
Sézary syndrome and mycosis fungoides arise from distinct T-cell subsets: a biologic rationale for their distinct clinical behaviors
.
Blood
2010
;
116
:
767
71
.
2.
Willemze
R
,
Cerroni
L
,
Kempf
W
,
Berti
E
,
Facchetti
F
,
Swerdlow
SH
, et al
.
The 2018 update of the WHO-EORTC classification for primary cutaneous lymphomas
.
Blood
2019
;
133
:
1703
14
.
3.
Wang
L
,
Ni
X
,
Covington
KR
,
Yang
BY
,
Shiu
J
,
Zhang
X
, et al
.
Genomic profiling of Sézary syndrome identifies alterations of key T-cell signaling and differentiation genes
.
Nat Genet
2015
;
47
:
1426
34
.
4.
Rindler
K
,
Bauer
WM
,
Jonak
C
,
Wielscher
M
,
Shaw
LE
,
Rojahn
TB
, et al
.
Single-cell RNA sequencing reveals tissue compartment-specific plasticity of mycosis fungoides tumor cells
.
Front Immunol
2021
;
12
:
666935
.
5.
Oka
T
,
Miyagaki
T
.
Novel and future therapeutic drugs for advanced mycosis fungoides and Sézary syndrome
.
Front Med (Lausanne)
2019
;
6
:
116
.
6.
de Masson
A
,
O'Malley
JT
,
Elco
CP
,
Garcia
SS
,
Divito
SJ
,
Lowry
EL
, et al
.
High-throughput sequencing of the T-cell receptor β gene identifies aggressive early-stage mycosis fungoides
.
Sci Transl Med
2018
;
10
:
eaar5894
.
7.
van Dongen
JJ
,
Langerak
AW
,
Bruggemann
M
,
Evans
PA
,
Hummel
M
,
Lavender
FL
, et al
.
Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4-CT98-3936
.
Leukemia
2003
;
17
:
2257
317
.
8.
Ponti
R
,
Fierro
MT
,
Quaglino
P
,
Lisa
B
,
Paola
F
,
Michela
O
, et al
.
TCRgamma-chain gene rearrangement by PCR-based GeneScan: diagnostic accuracy improvement and clonal heterogeneity analysis in multiple cutaneous T-cell lymphoma samples
.
J Invest Dermatol
2008
;
128
:
1030
8
.
9.
Sufficool
KE
,
Lockwood
CM
,
Abel
HJ
,
Hagemann
IS
,
Schumacher
JA
,
Kelley
TW
, et al
.
T-cell clonality assessment by next-generation sequencing improves detection sensitivity in mycosis fungoides
.
J Am Acad Dermatol
2015
;
73
:
228
36
.
10.
Gibson
JF
,
Huang
J
,
Liu
KJ
,
Carlson
KR
,
Foss
F
,
Choi
J
, et al
.
Cutaneous T-cell lymphoma (CTCL): current practices in blood assessment and the utility of T-cell receptor (TCR)-Vβ chain restriction
.
J Am Acad Dermatol
2016
;
74
:
870
7
.
11.
Robins
HS
,
Campregher
PV
,
Srivastava
SK
,
Wacher
A
,
Turtle
CJ
,
Kahsai
O
, et al
.
Comprehensive assessment of T-cell receptor beta-chain diversity in alphabeta T cells
.
Blood
2009
;
114
:
4099
107
.
12.
Mato
T
,
Masuko
K
,
Misaki
Y
,
Hirose
N
,
Ito
K
,
Takemoto
Y
, et al
.
Correlation of clonal T-cell expansion with disease activity in systemic lupus erythematosus
.
Int Immunol
1997
;
9
:
547
54
.
13.
Jackow
CM
,
Cather
JC
,
Hearne
V
,
Asano
AT
,
Musser
JM
,
Duvic
M
.
Association of erythrodermic cutaneous T-cell lymphoma, superantigen-positive Staphylococcus aureus, and oligoclonal T-cell receptor V beta gene expansion
.
Blood
1997
;
89
:
32
40
.
14.
Rosetti
F
,
Madera-Salcedo
IK
,
Rodriguez-Rodriguez
N
,
Crispin
JC
.
Regulation of activated T-cell survival in rheumatic autoimmune diseases
.
Nat Rev Rheumatol
2022
;
18
:
232
44
.
15.
Brunner
PM
,
Emerson
RO
,
Tipton
C
,
Garcet
S
,
Khattri
S
,
Coats
I
, et al
.
Nonlesional atopic dermatitis skin shares similar T-cell clones with lesional tissues
.
Allergy
2017
;
72
:
2017
25
.
16.
Gaydosik
AM
,
Tabib
T
,
Geskin
LJ
,
Bayan
CA
,
Conway
JF
,
Lafyatis
R
, et al
.
Single-cell lymphocyte heterogeneity in advanced cutaneous T-cell lymphoma skin tumors
.
Clin Cancer Res
2019
;
25
:
4443
54
.
17.
Macosko
EZ
,
Basu
A
,
Satija
R
,
Nemesh
J
,
Shekhar
K
,
Goldman
M
, et al
.
Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets
.
Cell
2015
;
161
:
1202
14
.
18.
Stuart
T
,
Butler
A
,
Hoffman
P
,
Hafemeister
C
,
Papalexi
E
,
Mauck
WM
3rd
, et al
.
Comprehensive integration of single-cell data
.
Cell
2019
;
177
:
1888
902
.
19.
Satija
R
,
Farrell
JA
,
Gennert
D
,
Schier
AF
,
Regev
A
.
Spatial reconstruction of single-cell gene expression data
.
Nat Biotechnol
2015
;
33
:
495
502
.
20.
Butler
A
,
Hoffman
P
,
Smibert
P
,
Papalexi
E
,
Satija
R
.
Integrating single-cell transcriptomic data across different conditions, technologies, and species
.
Nat Biotechnol
2018
;
36
:
411
20
.
21.
Soneson
C
,
Robinson
MD
.
Bias, robustness, and scalability in single-cell differential expression analysis
.
Nat Methods
2018
;
15
:
255
61
.
22.
Hao
Y
,
Hao
S
,
Andersen-Nissen
E
,
Mauck
WM
, 3rd,
Zheng
S
,
Butler
A
, et al
.
Integrated analysis of multimodal single-cell data
.
Cell
2021
;
184
:
3573
87
.
23.
Kramer
A
,
Green
J
,
Pollard
J
Jr
,
Tugendreich
S
.
Causal analysis approaches in ingenuity pathway analysis
.
Bioinformatics
2014
;
30
:
523
30
.
24.
Trapnell
C
,
Cacchiarelli
D
,
Grimsby
J
,
Pokharel
P
,
Li
S
,
Morse
M
, et al
.
The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells
.
Nat Biotechnol
2014
;
32
:
381
6
.
25.
Cao
J
,
Spielmann
M
,
Qiu
X
,
Huang
X
,
Ibrahim
DM
,
Hill
AJ
, et al
.
The single-cell transcriptional landscape of mammalian organogenesis
.
Nature
2019
;
566
:
496
502
.
26.
Qiu
X
,
Mao
Q
,
Tang
Y
,
Wang
L
,
Chawla
R
,
Pliner
HA
, et al
.
Reversed graph embedding resolves complex single-cell trajectories
.
Nat Methods
2017
;
14
:
979
82
.
27.
Kim
D
,
Kobayashi
T
,
Voisin
B
,
Jo
JH
,
Sakamoto
K
,
Jin
SP
, et al
.
Targeted therapy guided by single-cell transcriptomic analysis in drug-induced hypersensitivity syndrome: a case report
.
Nat Med
2020
;
26
:
236
43
.
28.
Tabib
T
,
Huang
M
,
Morse
N
,
Papazoglou
A
,
Behera
R
,
Jia
M
, et al
.
Myofibroblast transcriptome indicates SFRP2(hi) fibroblast progenitors in systemic sclerosis skin
.
Nat Commun
2021
;
12
:
4384
.
29.
Pizzolato
G
,
Kaminski
H
,
Tosolini
M
,
Franchini
DM
,
Pont
F
,
Martins
F
, et al
.
Single-cell RNA sequencing unveils the shared and the distinct cytotoxic hallmarks of human TCRVδ1 and TCRVδ2 γδ T lymphocytes
.
Proc Natl Acad Sci U S A
2019
;
116
:
11906
15
.
30.
Fivenson
DP
,
Hanson
CA
,
Nickoloff
BJ
.
Localization of clonal T cells to the epidermis in cutaneous T-cell lymphoma
.
J Am Acad Dermatol
1994
;
31
(
5 Pt 1
):
717
23
.
31.
Ortonne
N
,
Huet
D
,
Gaudez
C
,
Marie-Cardine
A
,
Schiavon
V
,
Bagot
M
, et al
.
Significance of circulating T-cell clones in Sézary syndrome
.
Blood
2006
;
107
:
4030
8
.
32.
Vega
F
,
Luthra
R
,
Medeiros
LJ
,
Dunmire
V
,
Lee
SJ
,
Duvic
M
, et al
.
Clonal heterogeneity in mycosis fungoides and its relationship to clinical course
.
Blood
2002
;
100
:
3369
73
.
33.
Buus
TB
,
Willerslev-Olsen
A
,
Fredholm
S
,
Blumel
E
,
Nastasi
C
,
Gluud
M
, et al
.
Single-cell heterogeneity in Sézary syndrome
.
Blood Adv
2018
;
2
:
2115
26
.
34.
Padilla-Nash
HM
,
Wu
K
,
Just
H
,
Ried
T
,
Thestrup-Pedersen
K
.
Spectral karyotyping demonstrates genetically unstable skin-homing T lymphocytes in cutaneous T-cell lymphoma
.
Exp Dermatol
2007
;
16
:
98
103
.
35.
Tubbs
A
,
Nussenzweig
A
.
Endogenous DNA damage as a source of genomic instability in cancer
.
Cell
2017
;
168
:
644
56
.
36.
Herrera
A
,
Cheng
A
,
Mimitou
EP
,
Seffens
A
,
George
D
,
Bar-Natan
M
, et al
.
Multimodal single-cell analysis of cutaneous T-cell lymphoma reveals distinct subclonal tissue-dependent signatures
.
Blood
2021
;
138
:
1456
64
.
37.
Dobbeling
U
,
Dummer
R
,
Hess Schmid
M
,
Burg
G
.
Lack of expression of the recombination activating genes RAG-1 and RAG-2 in cutaneous T-cell lymphoma: pathogenic implications
.
Clin Exp Dermatol
1997
;
22
:
230
3
.
38.
Herne
KL
,
Talpur
R
,
Breuer-McHam
J
,
Champlin
R
,
Duvic
M
.
Cytomegalovirus seropositivity is significantly associated with mycosis fungoides and Sézary syndrome
.
Blood
2003
;
101
:
2132
6
.
39.
Krejsgaard
T
,
Willerslev-Olsen
A
,
Lindahl
LM
,
Bonefeld
CM
,
Koralov
SB
,
Geisler
C
, et al
.
Staphylococcal enterotoxins stimulate lymphoma-associated immune dysregulation
.
Blood
2014
;
124
:
761
70
.
40.
Blaizot
R
,
Ouattara
E
,
Fauconneau
A
,
Beylot-Barry
M
,
Pham-Ledard
A
.
Infectious events and associated risk factors in mycosis fungoides/Sézary syndrome: a retrospective cohort study
.
Br J Dermatol
2018
;
179
:
1322
8
.
41.
Linnemann
T
,
Gellrich
S
,
Lukowsky
A
,
Mielke
A
,
Audring
H
,
Sterry
W
, et al
.
Polyclonal expansion of T cells with the TCR V beta type of the tumor cell in lesions of cutaneous T-cell lymphoma: evidence for possible superantigen involvement
.
Br J Dermatol
2004
;
150
:
1013
7
.
42.
Greaves
M
,
Maley
CC
.
Clonal evolution in cancer
.
Nature
2012
;
481
:
306
13
.
43.
Delfau-Larue
MH
,
Petrella
T
,
Lahet
C
,
Lebozec
C
,
Bagot
M
,
Roudot-Thoraval
F
, et al
.
Value of clonality studies of cutaneous T lymphocytes in the diagnosis and follow-up of patients with mycosis fungoides
.
J Pathol
1998
;
184
:
185
90
.
44.
Murphy
M
,
Signoretti
S
,
Kadin
ME
,
Loda
M
.
Detection of TCR-gamma gene rearrangements in early mycosis fungoides by non-radioactive PCR-SSCP
.
J Cutan Pathol
2000
;
27
:
228
34
.
45.
van Doorn
R
,
van Kester
MS
,
Dijkman
R
,
Vermeer
MH
,
Mulder
AA
,
Szuhai
K
, et al
.
Oncogenomic analysis of mycosis fungoides reveals major differences with Sézary syndrome
.
Blood
2009
;
113
:
127
36
.
46.
Park
J
,
Daniels
J
,
Wartewig
T
,
Ringbloom
KG
,
Martinez-Escala
ME
,
Choi
S
, et al
.
Integrated genomic analyses of cutaneous T-cell lymphomas reveal the molecular bases for disease heterogeneity
.
Blood
2021
;
138
:
1225
36
.
47.
Rohrig
F
,
Schulze
A
.
The multifaceted roles of fatty acid synthesis in cancer
.
Nat Rev Cancer
2016
;
16
:
732
49
.
48.
Moasser
MM
.
The oncogene HER2: its signaling and transforming functions and its role in human cancer pathogenesis
.
Oncogene
2007
;
26
:
6469
87
.
49.
Zhao
E
,
Hou
J
,
Ke
X
,
Abbas
MN
,
Kausar
S
,
Zhang
L
, et al
.
The roles of sirtuin family proteins in cancer progression
.
Cancers
2019
;
11
.
50.
Vander Heiden
MG
,
Cantley
LC
,
Thompson
CB
.
Understanding the Warburg effect: the metabolic requirements of cell proliferation
.
Science
2009
;
324
:
1029
33
.
51.
Gaude
E
,
Frezza
C
.
Tissue-specific and convergent metabolic transformation of cancer correlates with metastatic potential and patient survival
.
Nat Commun
2016
;
7
:
13041
.
52.
Villa
E
,
Ali
ES
,
Sahu
U
,
Ben-Sahra
I
.
Cancer cells tune the signaling pathways to empower de novo synthesis of nucleotides
.
Cancers
2019
;
11
:
688
.
53.
de la Parra
C
,
Walters
BA
,
Geter
P
,
Schneider
RJ
.
Translation initiation factors and their relevance in cancer
.
Curr Opin Genet Dev
2018
;
48
:
82
8
.
54.
Vince
JE
,
Silke
J
.
TWEAK shall inherit the earth
.
Cell Death Differ
2006
;
13
:
1842
4
.
55.
Tai
YT
,
Lin
L
,
Xing
L
,
Cho
SF
,
Yu
T
,
Acharya
C
, et al
.
APRIL signaling via TACI mediates immunosuppression by T-regulatory cells in multiple myeloma: therapeutic implications
.
Leukemia
2019
;
33
:
426
38
.
56.
Gu-Trantien
C
,
Migliori
E
,
Buisseret
L
,
de Wind
A
,
Brohee
S
,
Garaud
S
, et al
.
CXCL13-producing TFH cells link immune suppression and adaptive memory in human breast cancer
.
JCI Insight
2017
;
2
:
e91487
.
57.
Ye
LL
,
Wei
XS
,
Zhang
M
,
Niu
YR
,
Zhou
Q
.
The significance of tumor necrosis factor receptor type II in CD8(+) regulatory T cells and CD8(+) effector T cells
.
Front Immunol
2018
;
9
:
583
.
58.
Yang
Y
,
Islam
MS
,
Hu
Y
,
Chen
X
.
TNFR2: Role in cancer immunology and immunotherapy
.
Immunotargets Ther
2021
;
10
:
103
22
.
59.
Ungewickell
A
,
Bhaduri
A
,
Rios
E
,
Reuter
J
,
Lee
CS
,
Mah
A
, et al
.
Genomic analysis of mycosis fungoides and Sézary syndrome identifies recurrent alterations in TNFR2
.
Nat Genet
2015
;
47
:
1056
60
.
60.
He
F
,
Antonucci
L
,
Karin
M
.
NRF2 as a regulator of cell metabolism and inflammation in cancer
.
Carcinogenesis
2020
;
41
:
405
16
.
61.
Chen
X
,
Kang
R
,
Kroemer
G
,
Tang
D
.
Broadening horizons: the role of ferroptosis in cancer
.
Nat Rev Clin Oncol
2021
;
18
:
280
96
.
62.
Wang
H
,
Fu
BB
,
Gale
RP
,
Liang
Y
.
NK-/T-cell lymphomas
.
Leukemia
2021
;
35
:
2460
68
.
63.
Daniels
J
,
Doukas
PG
,
Escala
MEM
,
Ringbloom
KG
,
Shih
DJH
,
Yang
J
, et al
.
Cellular origins and genetic landscape of cutaneous gamma delta T cell lymphomas
.
Nat Commun
2020
;
11
:
1806
.
64.
Chabab
G
,
Barjon
C
,
Bonnefoy
N
,
Lafont
V
.
Pro-tumor γδ T cells in human cancer: polarization, mechanisms of action, and implications for therapy
.
Front Immunol
2020
;
11
:
2186
.

Supplementary data