A skin liquid biopsy method to assess the immune environment of cutaneous T-cell lymphoma

22 Cells infiltrating lesional skin cannot be analyzed in detail through immunohistochemistry or cell 23 separation techniques using skin biopsy specimens. Here, we describe a skin liquid biopsy method 24 that enables T cell isolation from small amounts of lesional whole blood from mycosis fungoides 25 (MF) patients. Gene expression analysis was performed on isolated cells which revealed that 26 CD4 + CD45RO + T cells contribute to the pathogenesis of MF, and CD8 + CD45RO + T cells serve as the 27 effector cells of CD4 + malignant MF. CD8 + CD45RO + T cells in lesional blood negatively correlate 28 with the modified severity-weighted assessment tool (mSWAT) and negatively regulate cellular 29 responses. CD4 + CD45RO + T cells highly express TNC, C1QB, PLK4, SGK1, RGS1 and CD69, and 30 have a lower diversity of the T-cell receptor repertoire. Our results provide new insights into the 31 pathogenesis of MF. The technique described here may be applicable to the evaluation of treatment 32 efficacy for other skin inflammatory diseases. 33


Introduction 34
Within lesional areas, the cells and surrounding environment play important roles in skin diseases. 35 Although skin lesions are visible, harvesting cells from the lesions is time-consuming and there can 36 prove challenging due to considerable cell and protein loss caused by tissue degradation. Various 37 techniques have been adopted by researchers to analyze the cells and environments of skin lesions, 38 including multiphoton excitation microscopy 1 , dermal open-flow microperfusion 2 , immersion of skin 39 samples in a medium to extract cells 3 , etc. However, critical information can be lost during the 40 processes of isolating cells using these currently available techniques. For instance, while 41 lymphocytes can be isolated from skin tissue using a punch biopsy, too few cells are collected to from 42 the few millimeters of tissue providing only limited information regarding the infiltrating cells. 43 Alternatively, blood samples from the lesion sight can be analyzed in a short period of time without 44 requiring enzyme treatment, and can provide information regarding the surrounding environment, 45 including the levels and types of cytokines and inflammatory cells. In fact, a previous study 46 successfully used sera from peripheral blood and from blood obtained from the site of psoriasis 47 lesions to assess the skin lesion environment 4 . Therefore, liquid biopsy from skin lesions is expected 48 to be effective for the isolation and analysis of cellular components and serum. 49 Skin biopsies are regularly obtained for diagnosis and assessment of treatment efficacy for cutaneous 50 T-cell lymphoma (CTCL). However, the diagnosis of this disease can be relatively difficult 5 , and 51 effective treatments are not yet clearly defined. It is, therefore, necessary to isolate and analyze the 52

resident and systemic pathogenetic T cells and effector T cells. Mycosis fungoides (MF) is the most 53
common cutaneous lymphoma and is considered to be a low-grade T-cell lymphoma 6 . The 54 premycotic and mycotic phases can last for several years, except in some cases that develop very 55 rapidly 7 . Due to the relatively low awareness, and difficulties in the diagnosis of MF, many patients 56 seek dermatologic consultation for the first time after their condition has already progressed to the 57 mycotic or tumor phases. This condition can lead to death within a few months, following mass 58 7 in the epidermis and serum of patients with MF 15 . Malignant T cells expressing CCR4 are recruited 129 by CCL17 16 . CCL19 is a ligand of CCR7 and can migrate to the lymph nodes 16 . We found that CCL17, 130 as well as the other CCR4 ligand, CCL22, strongly correlated with mSWAT, with a more pronounced 131 effect observed in lesional blood than in peripheral blood (Figure 3d). The serum level of CCL22 132 reportedly decreases following treatment with bexarotene 17 . In this study, bexarotene combined with 133 phototherapy caused a decrease in CCL22 levels in lesional blood, but not in peripheral blood ( Figure  134 6d, e). 135 CXCL1 is involved in neutrophil migration and is associated with the inflammatory disease, 136 psoriasis 18 . In this study, mass cytometry analysis showed that the number of granulocytes increased 137 in lesional blood ( Figure 2e). Although reports on the relationship between CXCL1 and MF are scarce, 138 CXCL1 reportedly promotes invasion, metastasis, proliferation, and survival in many cancers 19 . 139 Moreover, CXCL11 is strongly correlated with CD8 + CD45RO + T cells in lesional blood (r = 0.47; 140 Figure 3d). Meanwhile, CXCL11 is not involved in CD4 + , but rather CD8 + cells migration 14 . The 141 number of CD4 + CD45RO + T cells in skin lesion was correlated with CXCL10, CXCL5, and CCL22 142 in lesional blood. CXCL10 and CCL22 are keratinocyte-derived chemokines. CXCL5 is reported to 143 be downregulated in the advanced stage of MF 20 ; however, there have been no reports on the 144 relationship between CXCL5 levels and the number of CD4 + CD45RO + T cells in a skin lesion. The 145 number of CD8 + CD45RO + T cells infiltrating tissues was not correlated with that of CD4 + CD45RO + 146 T cells in lesional blood. These results indicate that the chemokine profile of lesional blood differs 147 from that of peripheral blood demonstrating the effectiveness of the skin liquid biopsy technique for 148 obtaining a detailed chemokine profile of lesional blood. 149 Effective photo(chemo)therapy decreases levels of CTCL-related chemokines in lesional and 151 peripheral blood 152 Photo(chemo)therapy is a standard treatment for MF 6 . Bath-PUVA therapy is highly effective for the 153 treatment of the early Ia/Ib stage of MF 21 . However, the underlying mechanisms of 154 photo(chemo)therapy have not been fully elucidated. Therefore, we investigated how the chemokine 155 profiles change before and after photo(chemo) therapy with or without bexarotene. Effective 156 photo(chemo)therapy significantly decreased the serum levels of CCR4 ligands, such as CCL17 and 157 CCL22 in lesional blood (Figure 4e). This reduction in CCL17 and CCL22 levels might suppress the 158 recruitment of malignant T cells and could be one of the modes of action of photo(chemo)therapy for 159 CTCL treatment. In a previous study, other phototherapy treatments, including the use of narrow-160 band UVB, did not affect the serum levels of CCL17 and CCL22 in peripheral blood 22 . In patients 161 with atopic dermatitis, narrow-band UVB phototherapy decreases CCL17 and CCL22 levels 23 . 162 Meanwhile, in the current study the levels of CCL5, CCL17, CCL21, and CXCL1, but not CCL22, 163 were decreased in peripheral blood (Figure 4d). Based on these chemokine profiles, the number of 164 CD8 + CD45RO + T cells in lesional blood was lower than that in peripheral blood after treatment 165

Detailed characterization of gene expression in malignant T cells isolated from lesional blood 168
CD4 + CD45RO + T cells in lesional blood have a unique TCR repertoire with a reduced diversity as 169 compared with TCR repertoire in peripheral blood ( Figure 5a). To confirm the functions of T cells in 170 MF pathogenesis, gene ontology (GO) analysis using 48 genes was performed. The most enriched of 171 the eight biological processes in CD4 + CD45RO + T cells isolated from CD4 + malignant MF, was 172 "cellular response to vascular endothelial growth factor stimulus" (Figure 5d). CD4 + CD45RO + T 173 cells highly expressed TNC, CD69, C1QB, PLK4, SGK1, and RGS1. Tenascin-C (TNC) is a large 174 glycoprotein highly expressed in cancer and is an essential component in cancer metastasis 24,25 . 175 CD69 is a marker of resident memory T cells, and a recent report showed that CTCL malignant T 176 cells express a resident memory T-cell phenotype 26 . The C1QB gene expresses the complement C1q 177  chain, which is positively associated with breast cancer, renal cell cancer, and CTCL 27,28 . Polo-like 178 kinase4 (PLK4) is a serine/threonine kinase crucial for mitosis and DNA integrity and has emerged 179 as a therapeutic target for the treatment of multiple cancers 29,30 . The serum and glucocorticoid-180 regulated kinase 1 (SGK1) and the receptor of the G-protein signaling 1 (RGS1) are reportedly 181 upregulated in CTCL, as determined using genome wide and single cell analyses 31,32 . The principal 182 component analysis of full transcriptomes in this study revealed a high degree of individual variation 183 in gene expression (Figure 5e, j). 184 GO analysis using 461 genes was performed in CD8 + malignant MF. The most enriched biological 185 process was "pre-NOTCH expression and processing" among the eight most enriched biological 186 processes in CD8 + CD45RO + T cells ( Figure 5g) isolated from CD8 + malignant MF. CD8 + CD45RO + 187 T cells highly expressed exonuclease 1 (EXO1), E2F1, DNA polymerase theta (POLQ), chromatin 188 licensing and DNA replication factor 1 (CDT1), and WNT7A. EXO1 contributes to cell cycle 189 checkpoints and replication fork maintenance. Although the relationship between EXO1 and CTCL 190 is not well characterized, this gene plays a critical role in the tumorigenesis of several types of 191 tumors 19,33 . The transcription factor E2F1 is a proto-oncogene and is highly expressed in CTCL 192 cells 34,35 . POLQ has also been associated with tumorigenesis 36 . CDT1 is required for assembly of the 193 pre-replication complex and is expressed in CTCL skin tumors 32 . Meanwhile, WNT7A is a member 194 of the Wnt gene family, and aberrant Wnt signaling leads to blood cancer i.e., leukemia, lymphoma, We concluded that CD8 + CD45RO + T cells in lesional blood possess a unique TCR repertoire ( Figure  200 5b). Hence, these expanded CD8 + CD45RO + T cells might be tumor antigen-specific. Moreover, the 201 TCR repertoire was skewed in lesional blood compared with that in peripheral blood. CD8 + CD45RO + 202 T cells highly expressed the following representative genes in lesional blood samples: cyclin-203 dependent kinase 4 inhibitor (CDKN2B), protein tyrosine phosphatase receptor type T (PTPRT), 204 cAMP-responsive element-binding protein 3-like protein 1 (CREB3L1), cell adhesion molecule 4 205 (CADM4), and inositol polyphosphate-5-phosphatase F (INPP5F; Figure 5h). CDKN2B and PTPRT 206 are tumor suppressor proteins 38,39 , while CREB3L1 activates the expression of genes encoding cell 207 cycle inhibitors, including p21 40 ; these genes inhibit cell proliferation. CADM4 is also a tumor 208 suppressor protein and plays an important role in cancer progression 41,42 . INPP5F is a 209 polyphosphoinositide phosphatase that inhibits STAT3 signaling via inhibiting its phosphorylation 43 . 210 Genes involved in inflammation (Krüppel-like factor 4, KLF4; S100A8; S100A12; interleukin 211 1 IL1B; hematopoietic cell kinase, HCK) were highly expressed in peripheral blood ( Figure 5h). 212 KLF4 is a mediator of inflammation that activates NF-B signaling 44 . S100A8 and S100A12, Ca2 + -213 binding proteins belonging to the S100 family, are used as indicators of inflammation 45,46 . IL1 is a 214 potent proinflammatory cytokine implicated in the pathogenesis of many inflammatory diseases 45,46 . 215 HCK is a member of the Src family of tyrosine kinases reported to cause pulmonary inflammation 216 and stimulation of the innate immune response 47 . 217 GO analysis using 468 genes was performed in CD8 + CD45RO + T cells of CD4 + malignant MF. The 218 enriched biological processes were "negative regulation of intracellular signal transduction", 219 "negative regulation of STAT cascade", and "negative regulation for cellular response to growth 220 factor stimulation" among the eight most enriched biological processes in CD8 + CD45RO + T cells 221 ( Figure 5g) isolated from CD4 + malignant MF. While a previous study reported the effector cells 222 (CD8 + CD45RO + T cells) to express exhausted phenotypes 9 , here we found that CD8 + CD45RO + T 223 cells from lesional blood negatively regulated cellular responses. 224

Stage progression and skewed TCR repertoires in CD8 + CD45RO + T cells 226
Biopsy specimens and lesional blood were collected from erythema, plaque, and tumor lesions. While 227 the cell infiltration was observed to be more prominent in the tumor stage (Figure 6a, e), the 228 proportion of CD8 + CD45RO + T cells was highest in the plaque stage (Figure 6a, e). These results 229 were confirmed via cell separation from lesional blood (Figure 6b), in which a higher number of 230 CD8 + CD45RO + T cells were observed in the plaque stage, and fewer CD8 + CD45RO + T cells were 231 observed in the tumor stage. The CD8 + CD45RO + T cells in lesional blood also showed unique TCR 232 repertoires in the erythema and plaque lesions (Figure 6c, g). 233 The expression of TCR repertoires was similar among the stages. However, TCR repertoire was 234 skewed in the plaque lesion. These results support the previous finding that neoplastic T-cell clones In conclusion, a method of skin liquid biopsy was successfully developed in this study. The method 246 was applied to assess the immune environment of CTCL. Sufficient isolated cells and sera were 247 successfully obtained from a small amount of lesional blood collected via skin liquid biopsy. The 248 cells and sera isolated from lesional blood can also be used for further gene expression analysis in 249 target cells. Our results provide new insights into the pathogenesis of the rare cutaneous lymphoma, 250 MF. The technique described in this study could be further applied to other skin inflammatory 251 diseases, such as atopic dermatitis and psoriasis and could be used to evaluate the efficacy of 252 treatment for these diseases. 253

Patients 255
We recruited 20 patients with MF (mean age: 66.85 y ± 13.60, 9 women, 11 men) from the 256 Department of Dermatology at the Nagoya City University. The exclusion criteria were: 1) age below 257 20 years, 2) HTLV-1 positive status, 3) pregnancy. The institutional review board of the Nagoya City 258 University Graduate School of Medical Sciences approved the study (Approval number: #60-18-259 0101). The written informed consent was obtained from the patients. The patients' profiles are 260 described in Table S1. Samples were obtained from the first 14 patients who visited our department 261 from Dec 2018 to Dec 2019 and used for tissue staining, flow cytometry analyses, and chemokine 262 assays. The other samples from patients 15 to 20 were used for RNA-seq and TCR repertoire analyses 263 (Table S1). The RNA-seq of case 4, the TCR repertoire of cases 4, 10, 11, and 14 were analyzed at 264 the time of readmission. 265 266

Skin biopsy and lesional blood 267
Skin biopsies were regularly obtained from patients for the diagnosis and assessment of treatment 268 efficacy. After local anesthesia on the skin lesion with lidocaine without epinephrine, we performed 269 a punch biopsy of 4-or 5-mm depth. Oozing blood from the wounded area was collected into an 270 Eppendorf tube with anticoagulant as quickly as possible to avoid clotting. Next, 5 L of 100 U mL -271 1 heparin sodium (TERUMO) for flow cytometry analysis, and 3 L of 0.5 mol L -1 EDTA (Invitrogen) 272 for RNA-seq and TCR repertoire analyses, were used as anticoagulants. Approximately 200-300 L 273 of blood was collected from each wounded area. We also collected 30-50 L of lesional blood 274 without anticoagulant to obtain serum. After collecting lesional blood, we performed the punch 275 biopsy at the same wounded area again at a depth that was sufficient to obtain a skin sample. Serums 276 were collected and stored at -80 °C until analysis. We collected peripheral blood from the patient's 277 arm and treated the peripheral blood in the same manner as for lesional blood. 278

Mass cytometric immunoassay 297
The blood cells were stained for mass cytometry after hemolysis using the Maxpar Human Peripheral 298 Blood Phenotyping Panel kit (Fluidigm). Hemolyzed peripheral blood and lesional blood were 299 resuspended in 1 mL PBS and incubated for 5 min at room temperature with 1 mL of Cisplatin-108Pt 300 (Fluidigm). The cells were then washed using Maxpar Cell Staining Buffer (Fluidigm), centrifuged, 301 supernatant were discarded, and the pellets resuspended in 50 L of the same buffer. We then added 302 50 L of a prepared cocktail of titrated Maxpar metal-conjugated antibodies (Fluidigm). After 303 incubation for 15 min at room temperature, we washed the cells twice and fixed them with 2% 304 paraformaldehyde. The stained cells were analyzed in the St. Luke's MBL Corp using CyTOF. Mass 305 cytometry data were analyzed using the Cytobank (https://www.cytobank.org/). 306 307 Cell sorting 308 CD4 + CD45RO + and CD8 + CD45RO + T cells were sorted using the FACS melody (BD Bioscience). 309 The sorted T cells were collected for TCR repertoire and RNA-seq analyses as described below. 310 311

RNA-seq analysis 312
CD4 + CD45RO + and CD8 + CD45RO + T cells were prepared as described above. The T cells were 313 lysed with TRIzol Reagent (Thermo Fisher Scientific) and stored at -80 °C . The lysates were sent to 314 Genewiz Inc. for RNA sequencing. RNA-seq and related analyses were completed by Genewiz Inc. 315 In brief, RNA was extracted with chloroform/isopropanol, and was recovered from the extracts using 316 RNA Clean and Concentrator -5 columns (ZymoResearch) following the manufacturer's instructions. 317 The RNA purity was assessed with an Agilent 2100 Bioanalyzer. RNA was subjected to library 318 preparation with the TaKaRa SmartSeq Stranded (Takara Bio) and sequenced with the Illumina Hiseq 319 (Illumina). Sequences were mapped to grch38 with HISAT2 (version 2.0.1). Differentially expressed 320 genes were counted using the DESeq2 (n ≥ 2) and EdgeR (n = 1) package in R (version 3.6.3). Up-321 and downregulated genes were defined as those that (i) were differentially expressed in peripheral 322 and lesional blood cells with p-value lower than 0.05, and (ii) have greater than two-fold change in 323 the average normalized count data of peripheral and lesional blood cells. GO analysis of differentially 324 expressed genes was performed using the Metascape webtool (www.metascape.org). 325 326 TCR repertoire analysis 327 CD4 + CD45RO + and CD8 + CD45RO + T cells were prepared as described above. The T cells were 328 lysed with Isogen-LS (NIPPON GENE) and stored at -80 °C . The lysates were sent to Repertoire 329 Genesis Inc. (Ibaraki, Japan) for next-generation sequencing, which was performed as described 330 previously 52 . Briefly, total RNA was converted to complementary DNA (cDNA) with the SuperScript 331 reverse transcriptase (Invitrogen). Double strand (ds)-cDNA was synthesized and ligated with a 5′ 332 adaptor oligonucleotide and cut with the SphI restriction enzyme. Next, ds-cDNA was amplified 333 through PCR using primers specific for the adaptor and TCR constant region. The sequencing was 334 performed with the Illumina MiSeq paired-end platform (2 × 300 bp). Data processing was performed 335 with the repertoire analysis software developed by Repertoire Genesis Inc. TCR sequences were 336 assigned with a data set of reference sequences from the international ImMunoGeneTics information 337 system database (http://www.imgt.org). 338 339

Statistics 349
Statistical analyses were performed using GraphPad Prism 7. All numerical data were summarized 350 using mean ± standard deviation (SD). P-values were determined using paired or unpaired Student's 351 t-test, unless otherwise indicated in the figure legend. P-values < 0.05 were considered statistically 352 significant. *p < 0.05, **p < 0.01. N.S., not significant.   Dermatol. 32, 448-453 (1995).    patient's arm and erythematous lesion, respectively, on the same day. In this patient, lesional blood (approximately 300 µL) was obtained from the abdominal erythematous lesion. The same amount of peripheral blood was obtained to match the heparin and EDTA concentrations used for lesional blood.
c) Flow cytometry of CD4 + and CD8 + T cells obtained from 5 L of peripheral and lesional blood.
The number of CD4 + and CD8 + T cells in 5 L of peripheral and lesional blood is shown (right, n = 4). d) Representative mass cytometry analysis of peripheral and lesional blood from a patient with MF through viSNE analysis. The color of each dot represents its immune cell subset. e) The proportion of each cell population in peripheral and lesional blood obtained from patients with MF (n = 5) quantified by mass cytometry. A paired t-test was used for statistical analysis. * p < 0.05. PB: peripheral blood, LB: lesional blood. using sera from peripheral and lesional blood (n = 14). Paired t-test. * p < 0.05, ** p < 0.01. d) Heatmap showing the correlation between mSWAT, CD4 + CD45RO + cells, and CD8 + CD45RO + cells in peripheral and lesional blood, cell counts in the lesional area, and the concentration of each chemokine in peripheral and lesional blood. tis-CD4 + CD45RO + and tis-CD8 + CD45RO + indicate the cell counts in the lesion tissue. Pearson correlation analysis was used for statistical analysis. * p < 0.05, ** p < 0.01. pre-and post-treatment. Wilcoxon signed rank test, * p < 0.05, **p < 0.01.     TRAV2   TRAJ10   TRAV5   TRAJ15   TRAJ29   TRAV21   TRAJ58   TRAV5   TRAJ15   TRAJ29   TRAV21  TRAJ58   TRAV1-2  TRAJ33   TRAV1