CR Lindsay et al. Role of Rac1 and P-Rex1 in Melanoblast Development

REFERENCES inflammatory profiles. Cytometry B Clin and is an endothelial adhesion receptor. Cytom 78:88–95 JImmunol179:6673–85 Bardin N, Anfosso F, Masse´ JM, et al. (2001) Identification of CD146 as a component Elshal MF, Khan SS, Raghavachari N, et al. (2007) Kagami S, Rizzo HL, Lee JJ et al. (2010) Circulating of the endothelial junction involved in the A unique population of effector memory Th17, Th22, and Th1 cells are increased in control of cell-cell cohesion. Blood 98: lymphocytes identified by CD146 having a psoriasis. J Invest Dermatol 130:1373–83 3677–84 distinct immunophenotypic and genomic pro- Larochelle C, Cayrol R, Kebir H et al. (2012) file. BMC Immunol 8:29 Dagur PK, Biancotto A, Wei L, et al. (2011) cell adhesion molecule identifies MCAM-expressing CD4( þ ) T cells in Elshal MF, Khan SS, Takahashi Y et al. (2005) encephalitogenic T lymphocytes and pro- peripheral blood secrete IL-17 A and CD146 (Mel-CAM), an adhesion marker of motes their recruitment to the central nervous are significantly elevated in inflammatory endothelial cells, is a novel marker of lym- system. Brain 135(Part 10):2906–24 autoimmune diseases. J Autoimmun 37: phocyte subset activation in normal periph- Mehta NN, Yu Y, Saboury B et al. (2011) Systemic 319–27 eral blood. Blood 106:2923–4 and vascular inflammation in patients with Dagur PK, Tatlici G, Gourley M, et al. (2010) Guezguez B, Vigneron P, Lamerant N et al. (2007) moderate to severe psoriasis as measured CD146 þ T lymphocytes are increased in Dual role of melanoma cell adhesion mole- by [18 F]-fluorodeoxyglucose positron emis- both the peripheral circulation and in cule (MCAM)/CD146 in lymphocyte endothe- sion tomography-computed tomography the synovial effusions of patients with various lium interaction: MCAM/CD146 promotes (FDG-PET/CT): a pilot study. Arch Dermatol musculoskeletal diseases and display pro- rolling via microvilli induction in lymphocyte 147:1031–9

A Rac1-Independent Role for P-Rex1 in Melanoblasts

Journal of Investigative Dermatology (2015) 135, 314–318; doi:10.1038/jid.2014.323; published online 11 September 2014

TO THE EDITOR melanoblasts lacking at the most distal melanoblast lineage, and DCT::b- Given the recent discovery of RAC1- points of migration (belly and paws). galactosidase (otherwise referred to as activating mutations in melanoma, and Constitutive deletion of Rac1 is DCT-lacZ). our finding that PIP3-dependent Rac- embryonically lethal, but a coat color First, we hypothesized that, if the exchanger 1 (PREX1) is overexpressed defect of mice with melanocyte-specific effects of P-Rex1 were mediated exclu- and drives metastasis in this cancer, an RAC1 abrogation (Tyr::Cre Rac1fl/fl)has sively via Rac1, double mutant Tyr::Cre important question is to establish whe- also been described; these mice have a Rac1fl/fl;P-Rex1 / mice would exhibit ther the functions of P-Rex1 are medi- larger belly spot on their ventral side, the same coat color phenotype as ated specifically by Rac alone (Lindsay suggesting that alternative Rho-GTPases Tyr::Cre Racfl/fl mice alone. However, et al., 2011; Berger et al., 2012). Here can be activated to enable melanoblast Tyr::Cre Racfl/fl; P-Rex1 / mice dis- we describe a Rac1-independent in vivo migration to the perimeter of the play a dramatic alteration in coat color role for P-Rex1 through identification Tyr::Cre Racfl/fl white belly (Sugihara phenotype from Tyr::Cre Racfl/fl mice and characterization of a mouse coat et al., 1998; Li et al., 2011). A role for (n ¼ 7; Figure 1a). The ventral and dorsal color phenotype. P-Rex1 is a guanine- Rac1 in proliferation was also observed, coats of these mice are almost entirely nucleotide exchange factor (GEF) for as there was a marked reduction white, with hypo-pigmented limbs and Rac, whose primary cell function is of melanoblast numbers in this tail. Graying pigmented areas were only induction of actin-mediated membrane phenotype. In line with these previous observed in the head coat. We con- ruffling and lamellipodia formation at studies, and because mice with melano- cluded from this experiment that P-Rex1 the leading edge of cell migration cyte-specific RAC1 abrogation require and Rac1 together constitute fundamen- (Welch et al., 2002; Hill et al., 2005; euthanization shortly after birth because tal signaling components of the mouse Barber et al., 2007). of neurological problems, we used the coat color phenotype, with minimal To investigate this question we same embryonic melanoblast reporter rescue of melanoblast development decided to examine the role of Rac1 models to assess the downstream conferred by other GEFs or Rho- and P-Rex1 in melanoblast develop- effects of P-Rex1 in vivo (Mackenzie GTPases. It was also clear that P-Rex1 ment. Previously, we reported a ‘‘white et al., 1997; Mort et al., 2010; Li must be able to exert phenotypic effects belly’’ phenotype of mice with Prex1 et al., 2011). Melanocyte-specific other than via Rac1. deletion (Lindsay et al., 2011). Impaired reporter mouse strains employed were To explore the Rac1-independent melanoblast migration was mostly Tyr::Cre Z/EG, which drives green effects of P-Rex1 further, we crossed responsible for this phenotype, with fluorescent expression in the Tyr::Cre Racfl/fl;P-Rex1 / mice with mice carrying the melanoblast reporter DCT-lacZ transgene (methods detailed in

Abbreviations: GEF, guanine-nucleotide exchange factor; OHT, 4-hydroxytamoxifen; PREX1, PIP3- Lindsay et al., 2011). Relative to Tyr::Cre fl/fl / dependent Rac-exchanger 1 Rac mice or P-Rex1 embryos fl/fl / Accepted article preview online 30 July 2014; published online 11 September 2014 alone, Tyr::Cre Rac ;P-Rex1

314 Journal of Investigative Dermatology (2015), Volume 135 CR Lindsay et al. Role of Rac1 and P-Rex1 in Melanoblast Development

P-Rex1–/– Tyr::Cre Rac1fl/fl P-Rex1–/–; Tyr::Cre Rac1fl/fl

15,000 Control P-Rex1–/– ** ** 10,000 * ** 5,000 **

Tyr::Cre Rac1fl/fl P-Rex1–/– Number of melanoblasts ** Tyr::Cre Rac1fl/fl 0

Ctr

Rac1 f/f P-Rex1–/–

Rac1 f/f P-Rex1–/–

Ctr Ctr siPrex1–2 siPrex1–2 OHT siPrex1–4 DMSO siPrex1–4

80

Staurosporin Ctr siP-Rex1–2 siP-Rex1–4 Ctr siP-Rex1–2 siP-Rex1–4 8 ) ** 4 ** ) P-Rex1 4 60 6 Rac 1/2/3 40 4 Actin ** * DCT 20 2 ** Number of cells (10

CC3 Number of cells (10 0 0 DMSO OHT 0 246 0246 Days Days

Figure 1. P-Rex1 and Rac1 are fundamental components of a mouse coat color phenotype. (a) Ventral coats of P-Rex1 / , Tyr::Cre Rac1fl/fl and P-Rex1 / ; Tyr::Cre Rac1fl/flmice. Final photomicrograph shows dorsal and head coat of Tyr::Cre Rac1fl/fl;P-Rex1 / mice. All mice were bred from a pure genetic background (C57Bl6). A total of 7 Tyr::Cre Rac1fl/fl;P-Rex1 / mice were bred; as Tyr::Cre Rac1fl/fl mice have a limited life span and no breeding potential, Tyr::Cre Rac1fl/ þ mice were crossed with P-Rex1 / mice to produce approximately one in four Tyr::Cre Rac1fl/fl;P-Rex1 / pups. (b) Representative photos of DCT-lacZ melanoblast distribution observed in wild-type (control), P-Rex1 / , Tyr::Cre Rac1fl/fl,andTyr::Cre Rac1fl/fl;P-Rex1 / embryos at E15.5. Black interrupted lines represent the area used for melanoblast quantification shown in c. White demarcated lines represent typical areas of melanoblast sparing visualized for each phenotype (n ¼ 3). (c) Number of melanoblasts in wild-type (Ctr), P-Rex1 / (‘P-Rex1 / ’), Tyr::Cre Rac1fl/fl (‘Rac1 f/f’), and Tyr::Cre Rac1fl/fl;P-Rex1 / (‘Rac1 f/f P-rex1 / ’) embryos at E15.5 (X5 embryos). Lower quartile, median, and upper quartile are shown. **Po0.01 by t-test; *Po0.05 by t-test. (d) Western blots of the primary melanocyte cell line, small interfering RNA (siRNA) treated as indicated, then treated with DMSO or 4-hydroxytamoxifen (OHT) for 5 days were probed with antibodies as indicated; Ctr, control siRNA–treated cells. (e) Growth curve of the primary melanocyte cell line, siRNA-treated as indicated, then treated with DMSO or OHT. Each point (mean±SEM) is from three replicates of three independent experiments. Asterisks represent difference between control siRNA–treated cells and P-Rex1 siRNA–treated cells at the same time point; Ctr, control siRNA-treated cells.

www.jidonline.org 315 CR Lindsay et al. Role of Rac1 and P-Rex1 in Melanoblast Development

Control P-Rex1–/– Tyr::Cre Rac1fl/fl P-Rex1–/– Tyr::Cre Rac1fl/fl

1 1 1 1 3 2 2 2 3 4 3 5 5 4 4 5 2 6 7 6 7 6 3

0′ 5′ 10′ 15′ 20′ 25′

Control

P-Rex1–/–

Tyr::Cre Rac1fl/fl

P-Rex1–/– Tyr::Cre Rac1fl/fl

80 Control 80 P-Rex1–/– 60 60 40 40 ) 20 20 –1 0.8 80 0 0 –20 –20 y -Axis ( μ m) y -Axis ( μ m) 0.6 60 –40 –40 ** ** * –60 –60 ** NS –80 –80 0.4 ** NS 40 ** –50 50 –50 50 ** ** x-Axis (μm) x-Axis (μm) 0.2 ** ** 20

80 fl/fl 80 –/– Tyr::Cre Rac1 P-Rex1 0.0 Euclidean distance ( μ m) 0 60 60 Tyr::Cre Rac1fl/fl speed ( μ m min Migration 40 40 Ctr –/– –/– Ctr –/– –/– 20 20 Rac1 f/f Rac1 f/f 0 0 P-Rex1 P-Rex1 –20 –20 y -Axis ( μ m) y -Axis ( μ m) –40 –40 Rac1 f/f P-Rex1 Rac1 f/f P-Rex1 –60 –60 –80 –80 –50 50 –50 50 x-Axis (μm) x-Axis (μm)

Figure 2. P-Rex1 has no additional effect on migration compared with Rac1 alone. All experiments show embryo skin explants at E15.5 (a) Combined Z-stack confocal images of Z/EG melanoblasts from wild-type (control), P-Rex1 / , Tyr::Cre Rac1fl/fl,andTyr::Cre Rac1fl/fl;P-Rex1 / embryos. Scale bars ¼ 10 mm. (b) Representative photomicrographs showing live imaging of Z/EG melanoblasts in the skin of wild-type, P-Rex1 / , Tyr::Cre Rac1fl/fl,andTyr::Cre Rac1fl/fl; P-Rex1 / embryos at E15.5 (n ¼ 3). Standard image measured at 20 magnification was 635 635 mm. (c) Three-hour tracks of melanoblasts from the same genotypes. (d) Migration speed of Z/EG melanoblasts in wild-type (Ctr), P-Rex1 / (‘P-Rex1 / ’), Tyr::Cre Rac1fl/fl (‘Rac1 f/f’), and Tyr::Cre Rac1fl/fl;P-Rex1 / (‘Rac1 f/f P-rex1 / ’) embryos (X5 embryos). **Po0.01 by t-test; *Po0.05 by t-test. (e) Z/EG melanoblast persistence, same genotypes, and annotation to that noted in d. Error bars indicate mean ±SEM. NS, nonsignificant.

embryos at E15.5 displayed a substantial melanocyte cell line with short interfering (Figure 1d). There was no increase in reduction in melanoblast numbers across RNAtoP-Rex1(methodsdetailedinLi cleaved caspase-3 evident in the absence their entire body (Figure 1b and c). To et al., 2012). The use of this model of P-Rex1 and/or Rac, suggesting that the assess whether the cause of this reduction system also allowed us to delete Rac1 reduced cell numbers observed in in melanoblast numbers could be function when these cells were treated Tyr::Cre Racfl/fl;P-Rex1 / embryos accounted for by decreased proliferation with 4-hydroxytamoxifen (OHT). We first were not accounted for by increased cell ± increased cell death, we next treated used western blotting to confirm that death (Figure 1d). Growth curves and our previously described primary immor- efficient P-Rex1 knockdown and OHT- anti-BrdU immunofluorescence of the talized Tyr::CrER2 INK4a / Racfl/fl induced Rac deletion were achieved same cell lines confirmed that there was

316 Journal of Investigative Dermatology (2015), Volume 135 CR Lindsay et al. Role of Rac1 and P-Rex1 in Melanoblast Development

a reduced proliferation in P-Rex1- Movies S1–3 online; methods detailed O’Prey, as well as the biological services, histol- ogy, and imaging staff at the CRUK Beatson depleted cells, both in the presence and in Lindsay et al., 2011). Institute for Cancer Research in general. We also absence of OHT (Figure 1e; Supple- To conclude, we have elucidated a thank Catherine Winchester for her assistance with mentary Figure S1a and b online). Taken proliferative role of P-Rex1 when Rac1 checking the manuscript. together, these results suggest, in addition is deleted. As E15.5 migratory charac- to our previously reported effects of Rac1 teristics are not altered in the Tyr::Cre Colin R. Lindsay1, Ang Li1, deletion on mouse coat color, that the Racfl/fl; P-Rex1 / double mutant William Faller1,BradOzanne1, loss of a Rac1-independent proliferative embryos, this suggests that the role of Heidi Welch2, Laura M. Machesky1 and effect of P-Rex1 also contributes to the P-Rex1 in migration is almost exclu- Owen J. Sansom1 coat color phenotype observed in sively mediated via Rac1. Here we 1Cancer Research UK Beatson Institute, Tyr::Cre Racfl/fl;P-Rex1 / mice (Li focused on E15.5 embryos, a useful time Glasgow, UK and 2The Babraham Institute, et al., 2011). point to observe the late migratory Cambridge, UK E-mail: [email protected] In line with our previous character- effects observed with previously ization of the P-Rex1 knockout pheno- described Prex and Rac phenotypes (Li type alone, we next decided to et al., 2011; Lindsay et al., 2011). With SUPPLEMENTARY MATERIAL delineate whether the coat color pheno- a greater number of embryos, we would Supplementary material is linked to the online fl/fl / type of Tyr::Cre Rac ;P-Rex1 have performed further embryo time- version of the paper at http://www.nature.com/jid mice could also be a consequence of point analyses at E13.5 to ensure there reduced melanoblast migration (Lindsay was no earlier melanoblast migratory REFERENCES et al., 2011). To assess this, mice with deficit that could contribute to this Barber M, Donald S, Thelen S et al. (2007) the Z/EG double reporter transgene phenotype, although even this Membrane translocation of P-Rex1 is medi- were crossed with Tyr::Cre Racfl/fl; experiment could not completely ated by G protein betagamma subunits and / phosphoinositide 3-kinase. JBiolChem282: P-Rex1 mice, driving green fluore- exclude such a possibility. 29967–76 scent protein expression in the melano- One potential Rho-GTPase, RhoG, is Berger MF, Hodis E, Heffernan TP et al. (2012) blast lineage (methods detailed in a likely candidate for P-Rex1 interac- Melanoma genome sequencing reveals fre- Lindsay et al., 2011; Figure 2a). Live tion: it is the most structurally similar quent PREX2 mutations. Nature 485:502–6 imaging of melanoblasts was performed Rho-GTPase to Rac and has been shown Helmy IM, Azim AMA (2012) Efficacy of ImageJ using E15.5 embryo skin from each to cooperate with Rac for induction of in the assessment of apoptosis. Diagn Pathol 7:15 genotype (Figure 2b; Supplementary cell transformation (Roux et al., 1997). Hill K, Krugmann S, Andrews DR et al. (2005) Movies S1–3 online). Consistent with Moreover, there are distinct regulatory Regulation of P-Rex1 by phosphatidylinositol our previous work, significant reduc- and functional similarities between (3,4,5)-trisphosphate and Gbetagamma subu- tions in migration speed were observed P-Rex1 and Vav , which have nits. JBiolChem280:4166–73 between wild-type, P-Rex1 / and been characterized as the predominant Lawson CD, Donald S, Anderson KE et al. (2011) Tyr::Cre Racfl/fl melanoblasts (Figure 2c GEFs required for RhoG activation P-Rex1 and Vav1 cooperate in the regulation of formyl-methionyl-leucyl-phenylalanine- and d). However, there was no signifi- (Samson et al., 2010; Lawson et al., dependent neutrophil responses. JImmunol cant difference in speed between 2011). Further studies are underway to 186:1467–76 fl/fl / Tyr::Cre Rac ; P-Rex1 and investigate the phenotypes of these and Li A, Ma Y, Yu X et al. (2011) Rac1 drives Tyr::Cre Racfl/fl melanoblasts alone, sug- other potential Rho-GTPases in melanoblast organization during mouse gesting that there was no change in melanoblast migration and melano- development by orchestrating pseudopod-dri- ven motility and cell-cycle progression. Dev migratory characteristics to account for magenesis. Given that there is now Cell 21:722–34 fl/fl / the Tyr::Cre Rac ; P-Rex1 pheno- considerable effort to generate Rac1 Li A, Ma Y, Jin M et al. (2012) Activated mutant type (Figure 2c and d). These results inhibitors, our data would suggest NRas(Q61K) drives aberrant melanocyte sig- were matched by similar differences functions for proteins upstream of Rac1 naling, survival, and invasiveness via a Rac1- between the same genotypes when that may become further therapeutic dependent mechanism. J Invest Dermatol 132:2610–21 Euclidean distance was measured targets in melanoma. All experiments Lindsay CR, Lawn S, Campbell AD et al. (2011) (Figure 2e), as well as no observable were conducted and approved in P-Rex1 is required for efficient melanoblast change in cell morphology evident in accordance with institutional and UK migration and melanoma metastasis. Nat Tyr::Cre Racfl/fl;P-Rex1 / compared guidelines, and all animal studies were Commun 2:555 with Tyr::Cre Racfl/fl melanoblasts alone performedinaccordancewithlocal Mackenzie MA, Jordan SA, Budd PS et al. (1997) (Supplementary Figure S1c online; regulatory guidelines. Activation of the receptor tyrosine kinase Kit is required for the proliferation of melano- methods detailed in Helmy and Azim, blasts in the mouse embryo. Dev Biol 192: 2012). Finally, no cell death was seen in CONFLICT OF INTEREST 99–107 our melanoblast time-lapse movies of The authors state no conflict of interest. Mort RL, Hay L, Jackson IJ (2010) Ex vivo live any genotype, again suggesting that imaging of melanoblast migration in embryo- ACKNOWLEDGMENTS nic mouse skin. Pigment Cell Melanoma Res P-Rex1 contributes to coat color 23:299–301 phenotype by promoting cell prolife- This research was supported by a Medical Research Council clinical fellowship (CRL), CRUK, Roux P, Gauthier-Rouvie`re C, Doucet-Brutin S ration using a Rac1-independent mech- and a European Research Council consolidator et al. (1997) The small GTPases Cdc42Hs, anism (Li et al., 2011; Supplementary grant (OJS). We thank Colin Nixon and Margaret Rac1 and RhoG delineate Raf-independent

www.jidonline.org 317 ADiCesareet al. Rationale for Speckled Hyperpigmentation

pathways that cooperate to transform NIH3T3 Sugihara K, Nakatsuji N, Nakamura K et al. (1998) This work is licensed under cells. Curr Biol 7:629–37 Rac1 is required for the formation of three germ a Creative Commons Attri- Samson T, Welch C, Monaghan-Benson E et al. layers during gastrulation. Oncogene 17:3427–33 bution-NonCommercial-NoDerivs 3.0 Un- (2010) Endogenous RhoG is rapidly activated Welch HC, Coadwell WJ, Ellson CD et al. (2002) ported License. To view a copy of this after epidermal growth factor stimulation P-Rex1, a PtdIns(3,4,5)P3- and Gbetagamma- through multiple guanine-nucleotide regulated guanine-nucleotide exchange factor license, visit http://creativecommons. exchange factors. Mol Biol Cell 21:1629–42 for Rac. Cell 108:809–21 org/licenses/by-nc-nd/3.0/

Rationale for the Development of Speckled Hyperpigmentation in the Areas of Psoriatic Plaques after Treatment with Biologic Agents

Journal of Investigative Dermatology (2015) 135, 318–320; doi:10.1038/jid.2014.297; published online 21 August 2014

TO THE EDITOR keratinocyte–derived antagonist (i.e., and other adhesion molecules (Wang We read with great interest the original beta defensin 3) in psoriatic plaques. et al., 2013). Therapeutic neutralization article by Wang et al. (2013) published Taken together, all these findings of IL-17 and TNF with biologic agents is in this journal. suggest that the psoriatic inflammatory able to recover pigmentation signaling, The authors investigated the effects milieu exerts a protective role against leading to postinflammatory hyper- of proinflammatory cytokines (i.e., TNF melanogenesis. Recently, we (Di Cesare pigmentation in the areas correspon- and IL-17) on epidermal melanocytes et al., 2013) and others (Balato et al., ding to the plaques that were enriched and their contribution to postinflamma- 2011) reported that psoriatic patients with melanocytes. In our opinion, this is tory hyperpigmentation in psoriatic have a low number of melanocytic a clue point to elucidate the mechanism patients. nevi as compared with healthy controls. of posttherapeutic development of lenti- They demonstrated that primary Indeed, we found that the probability of gines confined to the areas of pre- human melanocytes respond to IL-17 having melanocytic nevi in psoriatic existing psoriatic plaques that has been and/or TNF stimulation forming clusters patients was 82% lower compared with reported following phototherapy, topical and modulating the expression of a controls even independently of personal treatments, or biologic agents (Burrows broad panel of . In particular, and medical confounders such as socio- et al., 1994; Ana Costa et al., 2009; synergistic stimulation of melanocytes demographic factors, skin type and Santos-Juanes et al., 2008; Martı` et al., with IL-17 and TNF leads to increased acute and/or chronic sun exposure, 2009; Bardazzi et al., 2012). expression of mitogenic cytokines and family history of cancer, and psoriasis- We observed the appearance of growth factor genes (e.g., CXCL-1 and associated factors, such as disease family speckled hyperpigmentation within the IL-8) and downregulates the pigmenta- history, obesity, metabolic syndrome, psoriatic plaque areas during treatment tion signaling pathway and melanin cigarette smoking, and alcohol consump- with biologics in six patients (2 F, 4 M; production (e.g., c-Kit, MC1R, Mitf, tion (Di Cesare et al., 2013). median age: 64 years; range: 50–75 Dct). There is also a significant decrease In their paper, Wang et al. (2013) also years) during the treatment with adalimu- of tyrosinase levels and cellular melanin demonstrated that psoriatic skin has an mab (3/6) (Figure 1), infliximab (1/6), content. increased number of melanocytes, with ustekinumab (1/6), and etanercept (1/6) Moreover, given the pathogenic role a positive MelanA staining. They (Figure 2). A written consent for skin of TNF and IL-17 in psoriasis, Wang reported that psoriatic lesional skin, biopsies and photographic documenta- et al. investigated the pigmentation genes and not nonlesional or healthy control tion was obtained for each patient. All in psoriatic plaques and, as expected, skin, is Ki67 þ /MART-1 þ . This means patients were affected with moderate- they found a broad inhibition of genes that, despite IL-17/TNF brake on to-severe psoriasis (psoriasis area and involved in melanin synthesis, catalytic melanogenesis, melanocytes do not severity index410) and 5/6 had conco- enzymes, and lineage-specific melano- migrate from the inflamed skin to IL-17/ mitant psoriatic arthritis; 4/6 patients had cyte transcriptional factors. Indeed, TNF-free areas, but conversely they still o10 nevi, and 2/6 had a total number of they found overexpression of a MC1R infiltrate the plaques, expressing CCL20 nevi between 10 and 30. All patients had received several cycles of topical (corti- costeroids and/or vitamin D analogue, Abbreviation: TNF, tumor necrosis factor retinoids) and systemic treatments includ- Accepted article preview online 18 July 2014; published online 21 August 2014 ing ciclosporin (5/6), methotrexate (3/6),

318 Journal of Investigative Dermatology (2015), Volume 135