Prevalence, tropism, and activity of cutavirus in circulating blood lymphocytes, stool, and skin biopsy specimens of patients with cutaneous T‐cell lymphoma and parapsoriasis en plaques

A significant association has been established between a newly emerging human parvovirus, cutavirus (CuV), and cutaneous T‐cell lymphoma/mycosis fungoides (CTCL/MF) and its precursor parapsoriasis en plaques (PP). CTCL is a heterogeneous group of skin malignancies of T cells, the cause of which remains unknown. This study aimed to determine the activity, spread, and cell tropism of the skin‐persistent CuV. CuV DNA was detected in both skin biopsies (6/20, 30%) and peripheral blood mononuclear cells (PBMCs) (4/29, 13.8%) from 49 CTCL/MF or PP patients, while none from 33 patients with any other type of skin disease or healthy subjects harbored CuV DNA. CuV DNA persisted in the skin or PBMCs for up to 15 years, despite circulating CuV‐specific IgG. Spliced CuV mRNA was expressed in skin, indicating viral activity. Also, both of two available stool samples contained encapsidated CuV genomes, suggesting that the patients excrete infectious virus into the environment. Finally, CuV was observed to target circulating and skin‐resident CD4 + T cells and some skin keratinocytes and macrophages. This is especially intriguing as malignant T cells in CTCL develop from CD4 + T cells. Hence, CuV should be further investigated for the overall role it plays in the complex tumor microenvironment of CTCL/MF.

4][5] Furthermore, we have observed that CuV DNA can be detected in the skin for years after acute infection, despite circulating antibodies. 3e exact cause of CTCL remains unknown, but it has been suggested that chronic antigen stimulation could lead to CTCL carcinogenesis. 6Accordingly, many patients with CTCL/MF also first present with long-standing reactive inflammatory conditions such as parapsoriasis en plaques (PP), which manifests in two forms, small plaque (SPP) and large plaque (LPP) parapsoriasis. 7Recently, we have observed a significant association of CuV also with PP, with the hitherto highest CuV-DNA skin prevalence of 67%, suggesting that CuV may possibly be involved in the progression of PP to CTCL/MF. 8nce, our current study aims to find evidence of viral activity and spread to the bloodstream and stool, and further pinpoint the host cells of CuV in CTCL/MF and PP skin.Our results will cast further light on the possible role of CuV in CTCL.

| Patients and samples
Group 1 comprised 26 patients (14 males and 12 females) providing 27 fresh-frozen skin-biopsy specimens, collected during follow-up visits in 2017-2022 with informed consent (ERB approvals HUS/ 192/2016, HUS/3392/2021).Nineteen had CTCL, and seven had PP, as diagnosed at the Department of Dermatology, Helsinki University Hospital (HUS), based on detailed dermatopathological and immunohistological analyses (Table 1).Fresh skin biopsies were taken as 5-mm punch biopsies into RNAlater solution and kept at +4°C overnight, before storage at −20°C until analysis.
We further obtained corresponding serum samples from 12, whole blood from three, stool samples from two, and serial formalinfixed paraffin-embedded (FFPE) skin-biopsy samples from three patients (Table 1).
Group 2 comprised 56 individuals, from the same HUS clinic, with CTCL, PP, peripheral T-cell lymphoma (PTCL), and some nonmalignant dermatological manifestations (Table 2), who provided 103 peripheral blood mononuclear cell (PBMC) samples, and three patients provided 12 sequential FFPE samples (Table 1).Cell pellets from Ficoll-isolated PBMCs were frozen and stored at −80°C until analysis in a blinded fashion.The study was approved by the Medical Ethics Review Board of HUS Helsinki University Hospital 2013 and HUS/2004/2018, and all patients provided written informed consent.

| DNA and RNA extraction, PCR, and EIA
Total DNA and RNA were extracted with QIAamp DNA or RNA kits, respectively (Qiagen), and stored at −20°C until analyses (Supporting Information S1: Table S1).CuV, and related bufa-(BuV) and tusavirus (TuV), DNAs were amplified with a multiplex real-time qPCR 3 and IgG was measured by enzyme immunoassay (EIA) 9 (Supporting Information S1: Table S1).

| Benzonase endonuclease pretreatment for qPCR
To investigate whether the stool-derived CuV DNAs were encapsidated in viral particles (virions) or not ("naked DNA"), the T A B L E 1 Characteristics of cohorts used in the current study.S1).
To analyze two protein markers on the same tissue section after RISH, a dual fluorescent IHC technique was implemented (Supplement), as described.studied for CuV DNA and mRNA in a blinded analysis (Table 1).The total prevalence of CuV DNA in CTCL/MF and PP fresh-frozen skin was 6/20 (30%).Among the PP patients, CuV DNA was detected in 4/7 (57%) individuals, and all four also had mRNA.Among the CTCL/ MF patients, CuV DNA was detected in 2/13 (15.4%) and CuV mRNA in one of these two (Supporting Information S1: Table S1), whereas none of the six non-PP, non-MF patients had detectable CuV DNA in their skin.The CuV-DNA loads varied between 3.5 × 10 2 and 7.4 × 10 7 copies/million cells (cpm).All samples were negative for BuV and TuV DNA.One of the six CuV-positive patients (FFT6 in Supporting Information S1: Table S1) was later discovered to be the same patient (case C53), of whom we had previously analyzed multiple FFPE samples by qPCR, with up to 7 × 10 8 CuV DNA cpm. 3ditionally, all FFPE skins of FFT3, FFT6, and FFT21, were individual were very similar, whereas those from distinct patients varied (Figure 1).The shorter 233-nt CuV-DNA sequences, obtained from seven skins of patient FFT21, spanning 15 years, were similar but not identical (Supporting Information S1: Figure S1).However, the two longer over 500-nt sequences that we were able to amplify, taken in 2008 (PP358461) and 2022 (PP001446), did cluster together in the phylogenetic tree (Figure 1).
Spliced CuV mRNA was detected in 5/6 CuV DNA-positive fresh-frozen skin biopsies, and the mRNA copies ranged from 8 × 10

| Presence of encapsidated CuV DNA in stool samples
When control CuV DNA-negative stool samples from a healthy adult, 12 spiked with high amounts of plasmids containing BuV, CuV, or TuV DNA, were subjected to Benzonase treatment, there was a significant ( > 99%) reduction in plasmid copy numbers (Figure 2A).On the contrary, stool samples, obtained from two patients with CuV PCR-positive skin, showed no significant reduction in CuV DNA copies before and after Benzonase treatment (Figure 2B).

| DISCUSSION
An infectious cause for CTCL has long been speculated.Previously, CTCL skin samples have been screened for some viruses, such as retroviruses and herpesviruses. 13,148,15 In this study, we confirmed these associations and pinpointed the host cells of CuV in CTCL and PP mainly to skin CD4 + T lymphocytes.We also detected spliced viral mRNA in skin, indicating that CuV is actively transcribing in these CTCL and PP tissues.Furthermore, we observed CuV genomes both in circulating PBMCs and in stools, but not in serum.By nuclease treatment, we further showed that the stool-excreted viral DNA was encapsidated as virions, possibly enabling fecal-oral transmission.The persisting CuV DNA in CTCL is thereby not a dormant genome confined to skin, but actively expressing and spreading in the body.
CuV DNA has previously been detected and associated with CTCL/MF patients in France, Finland, Germany, and Japan, with prevalences of 8.5%-38%, 1,3-5 but was not found among 43 CTCL/ MF patient skin samples in Italy. 15However, regional differences in both virus and disease prevalences have been shown to occur, perhaps coupled with variable technical protocols and diagnostic challenges of the heterogenous CTCL/MF, which may further complicate any prevalence comparisons. 9,16,17 in their skin swabs. 8Moreover, up to 67% (8/12) of the PP patients had CuV DNA also in their corresponding skin biopsies, and during follow-up, two such patients developed CTCL/MF.These findings prompted us to investigate this relationship in more detail.
4][5] Remarkably, the highest CuV DNA prevalence, 57%, was found in the skin biopsies of PP patients, being similar to the 67% in our previous study. 8This could indicate that CuV has a role in the early phases of CTCL/MF development.All the CTCL and PP tissue samples analyzed in this, or our previous studies, were negative for BuV and TuV DNA, confirming the specificity of our CuV PCR and lack of contaminations. 3,8Besides viral DNA, we also detected spliced viral mRNA in five of the six CuV DNA-positive skin samples from CTCL and PP patients.We designed the primers to amplify a region covering an intron, to rule out amplification of contaminating DNA.Detection of CuV mRNA in PP skin, even if non-spliced, was also confirmed by a Japanese group during the review process of our study, 18 with a CuV-DNA prevalence of 38%, same as they had in CTCL/MF. 5Expression of mRNA means that the persistent virus is active in the skin, expressing its genes in the nucleus, whereby it may be more likely to affect the cell and tumor microenvironment.3][24][25] However, reports on B19V and HBoV1 activity in the form of spliced mRNA or replication in tissues have furthermore been very scarce. 23,24This makes CuV especially intriguing compared to its relatives.recirculate between skin and blood, and since neoplastic T-cell clones are also found in the circulation, 26,27 it is logical that we found CuV in the CD4+ cells of both skin and PBMCs.In CTCL/MF, elevated ratios ( > 4:1) of CD4+ to CD8 + T lymphocytes have been reported. 28This explains the very scarce CD8 + T cells observed in our CTCL/MF tissue compared to early CTCL/MF tissue of the same patient or in PP patients.
Based on our previous study 3  CuV was initially discovered in diarrheic stool samples of children, with a low prevalence of 1.4%. 1 We have further observed a CuV-DNA prevalence of 2.5% (7/284) in stools of adults of the same age range as Group 1, all seven had gastroenteritis. 12In a recent Chinese study, CuV DNA in stools associated with rheumatoid diseases, however, at a prevalence ( < 6%) similar to what we found for CuV IgG in adults. 9,29Nevertheless, stool samples from CTCL or PP patients have not been studied previously, since they are not routinely collected from this patient group.Previous studies on parvovirus B19 have shown that the parvovirus capsid protects the genome against nucleases like Benzonase. 30,31In the current study, we were able to show, with the Benzonase-PCR assay, the presence possess the skin-tropic memory phenotype. 32Furthermore, it is known that also malignant T cells in MF show great intra-tumoral genetic heterogeneity, 33 the implications of which in relation to CuV would also need to be examined.
We clearly detected CuV DNA also in a few keratinocytes and macrophages.It has been shown that keratinocytes can function as accessory antigen-presenting cells in the epidermis, 34 and in CTCL, there is an apparent crosstalk between keratinocytes, fibroblasts, stromal cells, and malignant T cells. 35This complex signaling loop eventually leads to the activation of STAT proteins and induction of expression of Th2 molecules sustaining tumorigenesis in CTCL. 36ich of these target cells truly are permissive for CuV replication
CuV-PCR positive, including the six sequential FFPE skins from patient FFT21, obtained 2007-2019, with viral loads between 4.30 × 10 6 and 1.69 × 10 8 cpm.In addition to skin, PBMCs of patients FFT4, FFT5, and FFT23 and stools of patients FFT5 and FFT23 were also CuV-PCR positive, with viral loads of 1.78 × 10 1 -3.46 × 10 3 cpm in PBMCs, and of 1.05 × 10 3 -1.72 × 10 6 CuV copies/ml in stool supernatants (Supporting Information S1: Table S1).Furthermore, of altogether 16 CuV-DNA positive skin or PBMC samples, we could amplify and sequence longer CuV-DNA sequences of 233nt, and of 10 samples even over 500 nt, in the VP1 region (Figure 1).All CuV DNA sequences of over 500 nt of the same F I G U R E 1 Maximum likelihood phylogenetic tree made with 514-nt amplicons in the VP1 region (NC_039050.1:nt 1980-2497) of cutavirus isolated from this study and those available worldwide.Bufavirus 1-3 have been included as outgroups.Sequences annotated with bold circles are from this study.The tree was constructed with MEGA11, using the gamma distributed general time reverse model (GTR + G) with 1000 bootstrap resampling.The bootstrap values are shown at nodes.

Furthermore, 12
sequential FFPE skin-biopsy samples were available from three patients with CuV DNA-positive PBMC samples (Table3).One of these patients, PBL1, had, within 3 years, repeatedly CuV DNA in skin samples with lymphocytic cell infiltrates, and also in the last skin sample exhibiting parapsoriasis.One PBMC sample of this same patient had the highest CuV-DNA load compared to all the CuV DNA-positive PBMC samples.F I G U R E 2 Benzonase assay on patient stool and control samples.(A) Control stool samples spiked with bufavirus (BuV), cutavirus (CuV), and tusavirus (TuV) plasmids with and without benzonase treatment.(B) Stool samples from FFT5 and FFT23 with and without benzonase treatment.Note: Y-axis represented in logarithmic scale; the reductions in viral loads were for BuV 99.9992%, CuV, 99.901%, and TuV, 99.9961%.Repeated two times.p-Values represented as nonsignificant, ns, p > 0.05, or significant: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

3. 4 |
RISH and IHC reveal CuV mainly in T lymphocytes, but also in some keratinocytes and a few macrophagesCuV-specific RISH signals, indicated as red dots (Figure3A-E), were observed in 11 FFPE skins of four PP or CTCL/MF individuals with CuV-DNA loads above 10 6 cpm.The CuV signals were observed both in the nucleus and in the cytoplasm of the cells, and both in the dermis and epidermis.The signals were not evenly dispersed but remained in the same regions in sequential sections (Supporting Information S1: Figure S3).We also observed CuV-specific RISH signals in CuV PCR-positive FFPE non-plaque lesions (nevi and eczema) from the parapsoriasis patient FFT21 (Supporting Information S1: Figure S4).CuV is a linear single-stranded DNA virus packaging mainly the negative-sense strand in the capsid, but the genome is made doublestranded inside the nucleus.With the two CuV probes detecting the sense or antisense NS gene, we got equally intense RISH signals, indicating that the cells harbored both the virion-bound negativesense CuV-DNA strand and the complementary intracellular positive-sense DNA strand (and maybe also viral mRNA), (Figure 3A-C, respectively).The CuV probes showed no signals on four CuV PCRnegative skin tissue controls or on six low-copy CuV PCR-positive skins (one shown in Figure 3F).The positive (human PPIB) and negative (bacterial dapB) mRNA technical control probes worked as expected (representative RISH images shown in Figure 3G-H, Supporting Information S1: Figure S5).On comparison, the CuV NS1 antisense probe RISH signals persisted after RNase treatment, whereas the antisense housekeeping gene (PPIB) control mRNA signals, as expected, did not (Supporting Information S1: Figure S5).With chromogenic IHC on the same RISH slides, we observed CuV DNA in some keratinocytes and a few macrophages, but mostly in CD3 + T lymphocytes (Figure4A-C).Further, very few to no B cells were observed in these skin tissues (Data not shown).With fluorescent dual IHC, we further confirmed the presence of CuV nucleic acids specifically in CD4+ but not in CD8+ cells in CTCL/MF (Figure4D) and in early CTCL/MF (Figure4E) from patient FFT6 as well as in PP tissues (Figure4F,G) from patients FFT3 and FFT21.In addition, CuV nucleic acids were also observed in CD4-positive PBMCs from PP patient FFT4 (Figure4H).There were much fewer CD8+ cells in CTCL/MF tissues than in the early CTCL/MF or PP tissues (Figure4D-G).

F I G U R E 3
RNAscope ISH RED assay with controls.(A-C) CuV PCR-positive skin from patient FFT6, early CTCL/MF (A) and CTCL/MF skin (B, C), hybridized with CuV NS antisense probe (A, B), or with CuV NS sense probe (C).(D-E) CuV PCR-positive PP skin from patient FFT3 (D) and FFT21 (E), hybridized with CuV NS antisense probe.(F) CuV NS antisense probe on a CuV PCR-negative CTCL/MF skin, as a negative tissue control.(G, H) CuV PCR-negative CTCL/MF-skin sample hybridized with the human PPIB mRNA probe (G), bacterial dapB probe (H), as positive and negative technical controls, respectively.Representative positive probe signals (red punctuated dots) indicated with arrows in the images and in the zoomed-in areas in the bottom-right corners.Cell nuclei stained with hematoxylin.Scale bars, 50 μm.CTCL, cutaneous T-cell lymphoma; CuV, cutavirus; MF, mycosis fungoides; PCR, polymerase chain reaction.CTCL/MF is still unclear.Both SPP (10% of cases) and LPP (35% of cases) have previously been shown to develop into CTCL/MF during a median of 10 and 6 years, respectively, 7 while suitable biomarkers are unknown.It remains to be seen whether CuV could be such a marker.
Besides skin, CuV DNA was also detected in the circulating blood cells exclusively of patients with CTCL/MF or PP, correlating with our observation in the skin biopsy samples.A higher viral load in PBMCs corresponded to finding CuV DNA also in skin.Nevertheless, none of the available serum samples contained CuV DNA, indicating lack of viremia.From our RISH and IHC analysis of PBMCs, CuV-specific RISH signals were observed mostly in CD4 + T cells, whereas CD8 + T cells were negative.Since the majority of cutaneous CD4 + T cells F I G U R E 4 Representative images from RNAscope ISH RED assay on CTCL/MF skin (A-D) or early CTCL/MF skin (E) from patient FFT6, parapsoriasis skin from patient FFT3 (F) and FFT21 (G) and PBMCs from parapsoriasis patient FFT4 (H), hybridized with CuV NS antisense probe detecting the viral mRNA or the sense single-stranded DNA genome.(A-C) RISH, where CuV DNA/mRNA is seen as red dots, coupled with chromogenic IHC, with cell markers in green showing keratinocytes labeled with K10 antibody (A), macrophages labeled with CD68 antibody (B) and T cells labeled with CD3 antibody (C).(D-H) RISH with CuV shown as green dots, coupled with fluorescent dual IHC, with cells labeled with CD4 antibody in red and CD8 antibody in blue.Cell nuclei stained with hematoxylin (A-C) or DAPI (D-H).Representative CuV-positive cells indicated with arrows in the zoomed-in areas in the bottom-right corners.Scale bars, 50 μm.CTCL, cutaneous T-cell lymphoma; CuV, cutavirus; DAPI, 4′,6-diamidino-2-phenylindole; MF, mycosis fungoides; PBMCs, peripheral blood mononuclear cells; RISH, RNAscope in situ hybridization.
and current study, patient FFT6 harbored CuV DNA in skin, from 2012 until his death in 2019(8 years), despite the presence of circulating CuV IgG.Interestingly, individuals with CuV DNA-positive skin biopsies have been found to have had CuV-specific IgG antibodies for up to 20 years before the biopsies.3Moreover, in the current study, we observed CuV DNA to persist in circulating lymphocytes for at least 5 years (PBL3), in skin for over 15 years (FFT21), and in skin and PBMC samples taken 10 years apart (PBL2).Overall, such long-term persistence of CuV is in good agreement with the fact that MF is often preceded by the longterm inflammatory condition of parapsoriasis.7 Furthermore, in CTCL/MF or PP patients, biopsies taken from both MF/PP plaque and nonplaque areas seem to carry CuV DNA, as shown for both FFT21 in this study and C53 in our previous study,3 concurring with virus spreading in the body.A comparison of intrapatient CuV DNA sequences from sequential samples, showed some minor evolutionary differences even if they clustered together in the phylogenetic tree, distinct from those of other patients.
of protecting viral capsids in CuV PCR-positive stools of two patients (one CTCL, one PP), suggesting that the patients were excreting infectious virions in their stools-despite their infections had persisted for at least 5 years.From RISH experiments of the 11 high viral-load CTCL/MF and PP skin samples, CuV-specific signals were observed in both the nucleus and the cytoplasm of cells localized in both the epidermis and dermis.In our previous study, we also observed a similar pattern of localization in other PP skins. 8That our RISH assays revealed equal intensities of CuV-specific signals with both the sense and antisense probes targeting the respective strands of the linear ssDNA CuV genome, indicated that double-stranded genomes had been formed in the cells.Furthermore, when removing mRNAs in the tissue sections by RNase A before RISH, the CuV-specific signals remained whereas the human mRNA signals in the control slide disappeared (because RISH does not detect chromosomal DNA).These results confirmed both that most CuV RISH signals were from viral DNA and that at least not all of the CuV DNA was integrated in the human genome.Nevertheless, the more sensitive RT-PCR did clearly show the presence of spliced viral mRNA, which was confirmed by sequencing and DNase and no-RT controls, not to be amplified from DNA.Previously, Phan et al. 1 observed CuV-specific RISH signals in the upper spinous layer of the skin.However, it was unclear whether the CuV-specific signal was present within the infiltrating T lymphocytes or within epithelial cells.In the current study, we observed CuVspecific RISH signals mostly in CD3+ and CD4 + T lymphocytes, but also in a few keratinocytes and macrophages in CTCL/MF and PP skin.Malignant T cells in CTCL mostly develop from CD4 + T cells and remains unknown.Hence, in-depth multiplex RISH-IHC and transcriptome studies are needed to understand the CuV-host interaction in the inflammatory microenvironment and its possible role in disease progression.In conclusion, the presence of CuV mRNA in skin and CuV DNA in stools, PBMCs, and several sequential tissue samples of the same individuals with CTCL/MF or its precursor PP, indicates that CuV spreads to different body regions following primary infection, and remains actively persistent for years, excreting potentially infectious virions to the environment.It is thus intriguing to speculate that the decades-long persistent CuV infection could cause a chronic inflammatory condition, which for some subjects would lead to a slow transformation of T cells and perhaps alteration of the evolution of malignant T-cell subclones.CuV should thus be further investigated for the overall role it plays in the complex tumor microenvironment of CTCL/MF and whether it could be applied as a biomarker for MF or perhaps define a subtype of MF.AUTHOR CONTRIBUTIONS Conceptualization: Ushanandini Mohanraj and Maria Söderlund-Venermo.Data curation: Ushanandini Mohanraj.Formal analysis: Ushanandini Mohanraj.Funding acquisition: Ushanandini Mohanraj, Maria Söderlund-Venermo, and Annamari Ranki.Investigation: Ushanandini Mohanraj.Methodology: Ushanandini Mohanraj.Resources: Liisa Väkevä and Annamari Ranki.Software: Ushanandini Mohanraj.Supervision: Maria Söderlund-Venermo.Validation: Ushanandini Mohanraj.Visualization: Ushanandini Mohanraj.Writing-Original Draft Preparation: Ushanandini Mohanraj.Writing-Review and Editing: Ushanandini Mohanraj, Liisa Väkevä, Annamari Ranki, and Maria Söderlund-Venermo.
Patient group-wise distribution of CuV DNA among PBMC samples from Group 2.
Patients of Group 2 with CuV DNA-positive PBMC samples with corresponding FFPE skin biopsy samples.
Some countries include PP as early CTCL/MF whereas others determine it as a precursor of CTCL/MF, as we do here.Nonetheless, we previously observed a significant association of CuV also with PP, as 6/13 (47%) PP patients, but only 1/51 (2%) healthy controls, exhibited CuV DNA T A B L E 3