Single-cell RNA sequencing comparison of CD4+, CD8+ and T-cell receptor γδ+ cutaneous T-cell lymphomas reveals subset-specific molecular phenotypes

Abstract

Background: Malignant clones of primary cutaneous T-cell lymphomas (CTCL) can show a CD4+, CD8+ or T-cell receptor (TCR)-γδ+ phenotype, but their individual impact on tumour biology and skin lesion formation remains ill defined.

Objectives: To perform a comprehensive molecular characterization of CD4+ vs. CD8+ and TCR-γδ+ CTCL lesions.

Methods: We performed single-cell RNA sequencing (scRNAseq) of 18 CTCL skin biopsies to compare classic CD4+ advanced-stage mycosis fungoides (MF) with TCR-γ/δ+ MF and primary cutaneous CD8+ aggressive epidermotropic cytotoxic T-cell lymphoma (Berti lymphoma).

Results: Malignant clones of TCR-γ/δ+ MF and Bertilymphoma showed similar clustering patterns distinct from CD4+ MF, along with increased expression of cytotoxic markers such as NKG7, CTSW, GZMA and GZMM. Only advanced-stage CD4+ MF clones expressed central memory T-cell markers (SELL, CCR7, LEF1), alongside B1/B2 blood involvement, whereas TCR-γδ+ MF and Berti lymphoma harboured a more tissue-resident phenotype (CD69, CXCR4, NR4A1) without detectable cells in the blood. CD4+ MF and TCR-γδ+ MF skin lesions harboured strong type 2 immune activation across myeloid cells, while Berti lymphoma was more skewed toward type 1 immune responses. Both CD4+ MF and TCR-γδ+ MF lesions showed upregulation of keratinocyte hyperactivation markers such as S100A genes and KRT16. This increase was entirely absent in Berti lymphoma, possibly reflecting an aberrant keratinocyte response to invading tumour cells, which could contribute to the formation of the typical ulceronecrotic lesions within this entity.

Conclusions: Our scRNAseq profiling study reveals specific molecular patterns associated with distinct CTCL subtypes.

Figure 1

Single-cell RNA sequencing map. (a) Uniform manifold approximation and projection (UMAP) plot of unsupervised clustering of 145 817 cells integrated from CD4⁺ advanced-stage mycosis fungoides (MF; n = 7), T-cell receptor (TCR)-γ/δ⁺ MF (n = 7), Berti lymphoma (n = 4) and healthy control (HC) samples (n = 4) according to overlapping transcriptomic features, resulting in 34 distinct cell clusters. (b) Feature plot of the integrated single-cell dataset showing the proliferation marker MKI67. (c) Dot plot depicting canonical cell markers for each cluster. Dot size represents the percentage of cells expressing the marker within the respective cluster. Colour intensity represents average expression levels within clusters. (d) Bar plot representing proportions of cells per sample that contribute to each cluster in HCs (light green), and CD4⁺ MF (red), Berti lymphoma (Berti lym.; dark blue) and TCR-γ/δ⁺ MF (light blue) lymphoma subsets. The suffix ‘pro’ denotes proliferating subsets. BC, B cells; BEC, blood endothelial cells; FB, fibroblasts; KC, keratinocytes; LC, Langerhans cells; LEC, lymphatic endothelial cells; MEL, melanocytes; MP, macrophages; NK, natural killer cells; pDC, plasmacytoid dendritic cells; SG, sweat gland cells, SMC, smooth muscle cells; TC, T cells; Tpro, proliferating T cells; Treg, regulatory T cells.

Figure 2

Lymphocyte subpopulations paired with the T-cell receptor (TCR) clonality landscape. (a) Uniform manifold approximation and projection (UMAP) plot of the lymphoid cluster, consisting of T cells and natural killer (NK) cells, with feature plots of CD4, CD8A, FOXP3 and MKI67 gene expression. Expression levels for each cell are colour-coded by the intensity of red colouring and overlaid onto UMAP plots. (b) Dot plot showing canonical T and NK cell markers. Dot size represents the percentage of cells expressing the marker within the respective cluster. Colour intensity represents average expression levels within clusters. (c) Absolute cell counts per lymphoid cluster, depicting healthy control (HC; n = 4), CD4⁺ mycosis fungoides (MF; n = 7), TCR-γ/δ⁺ MF (n = 7) and Berti lymphoma (Berti lym.; n = 4) subsets; each dot represents a single sample. Statistical significance was calculated using a Kruskal–Wallis test for multiple comparisons with the Dunn’s post hoc test; median (interquartile range). (d) UMAP plots representing the top expanded TCR⁺ T-cell clone from each sample according to identical CDR3 sequences coloured in red (HC), light red (CD4⁺ MF), dark blue (Berti lymphoma) and light blue (TCR-γ/δ⁺ MF); polyclonal TCRs are depicted in green and cells without detectable TCR in grey. (e) Bar plot showing absolute cell counts per cluster and sample. (f) Dot plot showing selected subset-specific genes characteristically expressed by malignant clones of CD4⁺ MF, Berti lymphoma or TCR-γ/δ⁺ MF groups. DEG, differentially expressed gene; NK, natural killer cells; TC, T cells; Tpro, proliferating T cells; Treg, regulatory T cells.

Figure 3

Pseudotime trajectory analysis of malignant clones from CD4⁺ mycosis fungoides (MF), Berti lymphoma and T-cell receptor (TCR)-γ/δ⁺ MF. (a) Top expanded clones from each sample were ordered on a pseudotime plot using Monocle 2, resulting in a bifurcated trajectory; cells are coloured by pseudotime. (b) Trajectory plot with overlaid MKI67 expression; highest and lowest levels are shown in red and grey, respectively. (c) Trajectory plot coloured according to the lymphoma subset. (d) Trajectory plots split for each individual sample, coloured according to the lymphoma subset. (e) Trajectory plot coloured according to cell state calculated by Monocle 2. (f) Bar plot showing the relative distribution of top clones among cell states for each sample. (g) Dot plot of state-specific differences in normalized gene expression among the top clones of CD4⁺ MF, Berti lymphoma and TCR-γ/δ⁺ MF malignant clones. Dot size represents the percentage of cells expressing the marker within the respective cell state and colour denotes gene expression levels. DEG, differentially expressed gene.

Figure 4

Myeloid cell characteristics across conditions. (a) Uniform manifold approximation and projection (UMAP) plot of the myeloid cell cluster, comprised of dendritic cells (DCs), macrophages (MPs), plasmacytoid DCs (pDCs) and proliferating cells (suffix ‘pro’). (b) Dot plot showing selected canonical markers for each myeloid cluster. (c) Absolute cell counts per myeloid cluster, depicting healthy control (HC; n = 4), CD4⁺ MF (n = 7), Berti lymphoma (Berti lym.; n = 4) and T-cell receptor (TCR)-γ/δ⁺ MF (n = 7) subsets; each dot represents a single sample. Statistical significance was calculated with a Kruskal–Wallis test for multiple comparisons with Dunn’s post hoc test; median (interquartile range). (d) Feature plots depicting selected type 2-associated markers. Gene expression levels for each cell are colour-coded (red) and overlaid onto UMAP plots. (e, f) Violin plots showing normalized expression of selected genes in DC3 and MP populations, respectively.

Figure 5

Keratinocyte characteristics across conditions. (a) Uniform manifold approximation and projection (UMAP) plot comprising keratinocytes (KC), sweat gland cells (SG) and proliferating cells (suffix ‘pro’). (b) Dot plot representing canonical markers of each cluster. (c) Absolute cell counts per cluster, with each bar representing a different condition: healthy control (HC; n = 4); CD4⁺ mycosis fungoides (MF; n = 7); Berti lymphoma (Berti lym.; n = 4); and T-cell receptor (TCR)-γ/δ⁺ MF (n = 7). Each dot represents a single sample. Statistical significance was calculated using the Kruskal–Wallis test for multiple comparisons with Dunn’s post hoc test; median (interquartile range). (d) Dot plot showing selected subset-specific genes differentiating keratinocytes between HC, CD4⁺ MF, Berti lymphoma or TCR-γ/δ⁺ MF. (e) Gene set enrichment analysis dot plot showing top enriched terms from the Gene Ontology Biological Process (GO:BP) database when analysing all upregulated differentially expressed genes (DEGs) from comparing keratinocytes from the CD4⁺ MF, Berti lymphoma or TCR-γ/δ⁺ MF subsets with HC keratinocytes. Dot size represents the number of DEGs associated with each enriched term and GeneRatio represents the ratio between the number of DEGs associated with each significantly enriched pathway and the total number of genes included in that pathway. (f) ‘cnetplot’ of the top upregulated DEGs associated with each enriched pathway. Genes are coloured according to average log fold change.