Activation of NFAT Signaling Establishes a Tumorigenic Microenvironment Through Cell Autonomous and Non-Cell Autonomous Mechanisms

ORIGINAL ARTICLE Activation of NFAT signaling establishes a tumorigenic microenvironment through cell autonomous and non-cell autonomous mechanisms

P Tripathi1, Y Wang1, M Coussens1, KR Manda1, AM Casey1, C Lin1, E Poyo2, JD Pfeifer3, N Basappa4, CM Bates5,LMa1, H Zhang6, MPan7, L Ding8 and F Chen1

NFAT (the nuclear factor of activated T cells) upregulation has been linked to cellular transformation intrinsically, but it is unclear whether and how tissue cells with NFAT activation change the local environment for tumor initiation and progression. Direct evidence showing NFAT activation initiates primary tumor formation in vivo is also lacking. Using inducible transgenic mouse systems, we show that tumors form in a subset of, but not all, tissues with NFATc1 activation, indicating that NFAT oncogenic effects depend on cell types and tissue contexts. In NFATc1-induced skin and ovarian tumors, both cells with NFATc1 activation and neighboring cells without NFATc1 activation have significant upregulation of c-Myc and activation of Stat3. Besides known and suspected NFATc1 targets, such as Spp1 and Osm, we have revealed the early upregulation of a number of cytokines and cytokine receptors, as key molecular components of an inflammatory microenvironment that promotes both NFATc1 þ and NFATc1 À cells to participate in tumor formation. Cultured cells derived from NFATc1-induced tumors were able to establish a tumorigenic microenvironment, similar to that of the primary tumors, in an NFATc1-dependent manner in nude mice with T-cell deficiency, revealing an addiction of these tumors to NFATc1 activation and downplaying a role for T cells in the NFATc1-induced tumorigenic microenvironment. These findings collectively suggest that beyond the cell autonomous effects on the upregulation of oncogenic proteins, NFATc1 activation has non-cell autonomous effects through the establishment of a promitogenic microenvironment for tumor growth. This study provides direct evidence for the ability of NFATc1 in inducing primary tumor formation in vivo and supports targeting NFAT signaling in anti-tumor therapy.

Oncogene (2014) 33, 1840–1849; doi:10.1038/onc.2013.132; published online 29 April 2013 Keywords: NFAT; tumorigenesis; tumor; microenvironment

INTRODUCTION and other cancer types suggests that NFAT activation may be The NFAT (the nuclear factor of activated T cells) family of used in a much wider set of tumors originated from cells 7–9 transcription factors is important for organ development and with no normal expression of these proteins. Although NFAT cellular homeostasis. The four NFATc (c1–c4) proteins reside in the genes have been linked to multiple aspects of various tumor cytoplasm in quiescent cells. When intracellular Ca2 þ increases, types, direct evidence showing NFAT activation initiates primary the serine/threonine phosphatase calcineurin dephosphorylates tumor formation in situ is lacking. In addition, the cellular function the NFATc proteins and exposes the concealed nuclear localiza- of NFAT signaling appears to be multifaceted and context- tion signals, leading to their cytoplasm to nuclear translocation. In dependent.10 Thus, the biological consequences of NFAT the nucleus, NFATc proteins complex with their partners to control activation in different tissues may be very different and the the transcription of target genes. Nuclear import and activation of mechanism by which NFAT affects tumorigenesis needs to be NFATc are opposed by phosphorylation of NFATc by GSK3 and further investigated. other NFAT kinases.1 However, when the serines targeted for In this study, we generated a transgenic system in which phosphorylation by NFATc kinases and for dephosphorylation by NFATc1 activation can be controlled by the administration of calcineurin are changed to alanines, the modiﬁed NFAT proteins doxycycline (Dox) in targeted tissues. We have discovered that become constitutively nuclear and active.2 NFATc1 activation induces tumor formation in situ by promoting Constitutively active NFATc1 was shown to induce transforma- local cytokine production to create an inﬂammatory microenvir- tion in preadipocytes.3 NFATc1 activation, as evidenced by its onment for cells with NFAT activation and their neighbors without nuclear presence and/or dephosphorylation state, has been NFAT activation to participate in tumor formation. Between two observed in multiple types of human lymphomas.4–6 The models with overlapping NFATc1 activation domains in the skin, observation of ectopic activation of NFATc1 in pancreatic cancer only the one with NFATc1 expression in follicle stem/progenitor

1Department of Medicine, Department of Cell Biology and Physiology, Washington University School of Medicine, St. Louis, MO, USA; 2Department of Bioengineering, Washington University School of Medicine, St. Louis, MO, USA; 3Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, USA; 4Department of Biology, University of Maryland, Baltimore, MD, USA; 5Department of Pediatrics, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA; 6Department of Cell Biology, University of Massachusetts Medical School, Worcester, MA, USA; 7Division of Oncology, Kaiser Permanente Medical Center, Santa Clara, CA, USA and 8Department of Medicine, The Genome Institute, Washington University School of Medicine, St. Louis, MO, USA. Correspondence: Dr F Chen, Department of Medicine, Department of Cell Biology and Physiology, Campus Box 8126, Washington University School of Medicine, St. Louis, MO 63110, USA. E-mail: [email protected] Received 17 October 2012; revised 27 February 2013; accepted 11 March 2013; published online 29 April 2013 NFAT signaling and tumorigenesis P Tripathi et al 1841 cells produced skin tumors, suggesting progenitor cell involve- in the ovary. In addition to urogenital tumors, Dox-treated ment. These and other findings reported here provide mechanistic mutants developed occasional skin tumors among numerous insights into the tumorigenic effects of NFAT activation beyond its precancerous lesions (Figures 1e and f). As early as 1 week after reported and suspected roles in direct transcriptional regulation of Dox treatment at P21, small lumps appeared randomly in skin oncogenes. throughout the body in B98% of mutants (n ¼ 150). While most lumps stayed small, some of these apparent precancerous lesions continued to grow into tumors of substantial size under RESULTS continuous Dox treatment. No control mice developed tumor at Conditional activation of NFAT signaling results in tumors in any sites. specific sites To study the role of NFAT signaling in urogenital organs, we created a transgenic model for inducible NFATc1 activation in cells NFATc1-induced tumors resemble ovarian sarcoma and skin targeted by the Hoxb7-Cre transgene11,12 that has known expres- squamous cell carcinoma in humans sion in the Wolffian duct, an embryonic structure providing Cytologically, the ovarian tumors were comprised of spindle- progenitors for multiple urogenital organs (Figure 1). In this shaped cells that bear morphologic resemblance to undifferen- system, Cre expression induces the removal of the transcriptional tiated mesenchymal cells (Figures 2a–f). Many tumor cells showed stop cassette in a ROSArtTA allele and the production of rtTA mild-to-moderate atypia. Regions of coagulative tumor cell (reverse tetracycline-controlled transactivator). When Dox is necrosis were also present (Figure 2b). These tumors have present, the Dox–rtTA complex binds to TetO of the TetO- scattered mitotic figures on hematoxylin and eosin-stained NFATc1Nuc transgene2 to induce the transcription of NFATc1Nuc sections and high mitotic index as revealed by Ki67 immunostain- (an activated form of NFATc1) (Figure 1a). We refer to mice ing (Figure 2c). The tumors showed destructive stromal invasion carrying all three alleles (Hoxb7-Cre, ROSArtTA and TetO-NFATc1Nuc) within the ovary and are associated with desmoplastic responses as mutants. Their littermates without the full set of three alleles (Figure 2d). The ovarian tumor cells strongly expressed mesench- cannot have NFATc1 activation even when treated with Dox and ymal markers, including a-smooth muscle actin (aSMA). There was thus are regarded as controls. NFATc1Nuc transcripts were detected insignificant expression of epithelial markers, such as cytokeratin 8 in Dox-treated mutants, but not in controls (Figure 1b, E13.5 (Figures 2e and f) and E-cadherin (not shown), in the tumor. Taken embryos, E: embryonic day). Dox-induced NFATc1 activation in together, the constellation of findings (mesenchymal cellular Wolffian duct derivatives during embryogenesis results in morphology and marker expression, high mitotic index, atypia, congenital renal defects and reduced viability with incomplete necrosis, desmoplasia, destructive stromal invasion and others) penetrance in mutants (the renal defects will be described indicate that the ovarian tumors in these mice resemble malignant separately). We examined mutants that survived past weaning ovarian stromal tumors,13 rather than benign ovarian fibroma, in and found tumors in the urogenital systems of both genders and humans. The ovarian tumors came as a surprise to us at first in the skin. In females, the tumors were in the ovary (Figures 1c because there was no previous report of Hoxb7-Cre expression in and d) while the male tumors were in the epididymis (data not the ovary. By crossing the Hoxb7-Cre transgene to the ROSALacZ shown). As epididymal tumors are very rare in humans, we chose reporter mice, we found b-galactosidase activity, reflecting Cre to perform most of the subsequent experiments in ovarian and expression, in the ovarian stroma containing the theca cells and skin tumors. The ovarian tumors can be noticed as early as at 3 other SMA þ stromal cells surrounding the ovarian follicles weeks of age. Hundred percentage of the female mutants (n ¼ 8) (Figures 2g and h and data not shown). These tumors were with Dox treatment since E0 (Embryonic day 0) developed tumors CD10 À , unlike some of the ovarian sarcomas with an endometrial

a Hoxb7 promoter Cre c d

Ovarian Cre Ovary tumor

Uterus ROSA promoter STOP rtTA Uterus

loxP loxP Control Mutant

Doxycycline (Dox) e Outer surface f Outer surface

rtTA b E13.5

Inner surface Inner surface CT MT

GAPDH

TetO NFATc1Nuc NFATc1Nuc Control Mutant Figure 1. The Hoxb7-Cre transgene-mediated inducible activation of NFATc1 causes tumor formation. (a) The Hoxb7-Cre transgene directs the production of rtTA, which in turn, induces the transcription of NFATc1Nuc in the presence of Dox. (b) RT–PCR-detected transcripts of the NFATc1Nuc transgene only in the Dox-treated mutants. NFATc1 activation induced tumors in the ovary (c, d) and skin (e, f). Arrows in (f) point to the tumor mass.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1840 – 1849 NFAT signaling and tumorigenesis P Tripathi et al 1842 a H&E bcH&E Ki67 d H&E e SMA/CK8/DAPI f SMA/CK8/DAPI a

Normal Ovary Ovarian tumor b

Normal Ovary

Ovarian Ovarian Control Mutant tumor Mutant tumor Control Mutant g -Gal h -Gal/CYP11A1 i -Gal j H&E k H&E l H&E

Ovary Normal Skin Keratin pearls Oviduct Ovarian Theca follicle Cells Hair follicle Skin tumor

Cre/+;ROSALacZ/+ Cre/+;ROSALacZ/+ Cre/+;ROSALacZ/+ Control Mutant Mutant m H&E n PCNA/DAPI o SMA/E-Cad/DAPI p SMA/E-Cad/DAPI q VIM/CK5/DAPI r VIM/CK5/DAPI

Skin tumor Adipose tissue Skin tumor Skin tumor

Muscle layer

Mutant Mutant Control Mutant Control Mutant

Figure 2. NFATc1-induced tumors resemble ovarian sarcoma and skin SCC in humans. Unlike the normal ovarian structures (a), NFATc1- induced ovarian tumors are ﬁlled with spindle-shaped cells (b–f). These tumors have cellular atypia (b, inset-a) and necrotic zones (b, inset-b). Rectangular area in (b) are enlarged in the insets (a, b). These tumors have high proliferation rates as indicated by extensive Ki67 immunostaining (c). These tumors have desmoplastic zones (arrowheads in (d)), where adjacent tissues react to the rapid expansion of the tumor. The ovarian tumors have abundant mesenchymal cells expressing aSMA with few epithelial cells expressing cytokeratin 8 (CK8) (e, f). Arrows in (e) indicate epithelial expression of CK8 in the normal ovary. As revealed by b-galactosidase (b-Gal) assays, Hoxb7-Cre is expressed in ureter, ovary, oviduct (g, h) and hair follicles (i). Inside the ovary, Hoxb7-Cre is expressed in the stroma containing the theca cells (h). The skin tumors have keratin pearls similar to those seen in human SCC (j–n). Arrows in (k) point to areas where the hyperplastic hair follicles invade the dermis layer of the skin (k). Arrows in (l) indicate presence of keratin pearls. These tumors can invade the subcutaneous adipose layer beneath the dermis and sometimes reach the muscle layer (m). Arrowhead in (m) indicates an area of necrosis. High proliferation rate in the skin tumor is evident by the extensive presence of PCNA þ cells (n). Immunostaining with epithelial markers E-cadherin (E-Cad) and cytokeratin 5 (CK5) reveals the expansion and dermal invasion of the skin tumor cells (o–r). Unlike the ovarian tumors, most cells in the skin tumors have epithelial characteristics with only a small number of cells expressing mesenchymal markers SMA and vimentin (VIM) (o–r).

stromal origin. These findings are consistent with the hypothesis invasion, no metastasis was observed in the mutants we have that the ovarian tumors found in the mutants originated from the examined. As most mutants died from tumor burden and other ovarian stromal cells with NFATc1 activation. comorbidities associated with NFATc1 activation in other tissues, it Hoxb7-Cre expression, as reflected by b-galactosidase activity, is currently unknown if any of the NFATc1-induced tumors will was also detected in the skin, primarily in hair follicle cells become metastatic if given sufficient time to progress in these (Figure 2i). Histopathological analyses showed that the skin tumors mice. Although epididymal tumors were not extensively character- diffusely invaded the dermis, the subcutaneous fat, and super- ized in this study, Hoxb7-Cre expression in the epididymal epithelia, ficially into the subadjacent connective tissues and skeletal muscle a Wolffian duct derivative, is well established,11,12 making these (Figures 2k–m). Geographic necrosis was present (arrowhead in tumors most likely primary tumors induced by NFATc1 activation Figure 2m). These skin tumors have keratin pearls typical of human but not metastasis from other sites. Besides the three sites cutaneous squamous cell carcinoma (SCC), though vague rudi- mentioned, tumors have not been found in other places with mentary structures resembling skin adnexal structures were known Hoxb7-Cre expression, including neuronal tissues, renal present (Figure 2l). Similar to the ovarian tumors, these skin collecting duct and ureteral epithelia. It appears that the oncogenic tumors were highly proliferative, as revealed by proliferating cell effects of NFATc1 activation are tissue/cell type-specific. nuclear antigen (PCNA) staining (Figure 2n). Many cells in the skin tumors expressed the epithelial marker E-cadherin (Figures 2o and Tissue stem cells/progenitor cells participate in NFATc1-induced p). Immunostaining with cytokeratin 5, a diagnostic marker for tumorigenesis human SCC, further confirms the extensive presence of cells of In the skin, stem cells are known to reside in the hair follicle bulge epithelial origin in the tumor proper and its resemblance to human and express CD34 and NFATc1 ((Figure 3a) and Horsley et al.14). SCC (Figures 2q and r). Although both the skin tumor and ovarian Although NFATc1 expression appeared stronger in the CD34 þ tumors showed histological features of malignancy and local bulge stem cells in the Dox-treated mutants, the NFATc1 antibody

Oncogene (2014) 1840 – 1849 & 2014 Macmillan Publishers Limited NFAT signaling and tumorigenesis P Tripathi et al 1843 a NFATc1/ b NFATc1/ c CD34/ d CD34/ e NFATc1/ f HA CD34/DAPI CD34/DAPI HA/DAPI HA/DAPI HA/DAPI

Control Mutant Control Mutant Mutant Mutant g NFATc1/ h NFATc1/CD34/DAPI i NFATc1/CD34/DAPI j K14/CD34/DAPI CD34/DAPI

Keratin Skin tumor Skin tumor pearl Mutant

Control Mutant Mutant Skin tumor k NFATc1/ l NFATc1/ m NFATc1/ n NFATc1/ o K14/ CD44/DAPI CD44/DAPI CD44/DAPI CD44/DAPI CD44/DAPI

Skin tumor

Keratin Skin tumor pearl Keratin pearl

Control Mutant Mutant Mutant Mutant p -Gal q H&E r H&E s NFATc1/ t NFATc1/ E-Cad/DAPI E-Cad/DAPI

Msx2-rtTA/+; Msx2-rtTA/+; LacZ/+ Nuc TetO-Cre/+;ROSA Control TetO-NFATc1 Normal human skin Human SCC Figure 3. Tissue stem cells/progenitor cells may participate in NFATc1-induced tumorigenesis. The CD34 þ hair follicle stem cells in the bulge expressed NFATc1 in both the control and mutant (a, b). Although the expression of NFATc1 appeared more intense in some mutant cells, definitive identification of transgene-produced NFATc1Nuc relies on the detection of the transgene-specific HA tag (c–f). There was no HA staining in the controls (c). In the mutants, most CD34 þ bulge stem cells were HA þ (d) and most NFATc1 þ cells in the hair follicles were HA þ (e, f). NFATc1 and CD34 had extensive co-expression in the skin tumor but were both absent in cells surrounding the keratin pearls (g–i). Keratin-14 (K14) had extensive expression in the skin tumor and was present in the NFATc1 À , CD34 À cells surrounding the keratin pearls (j). CD44 þ , NFATc1 þ cells were present in hair follicle bulge (k). Many CD44 þ , NFATc1 þ cells were present in the skin tumor (l–n). Keratin pearls were exclusively CD44 þ , K14 þ (o). The Msx2-rtTA transgene was expressed in hair follicle matrix cells and hair shaft excluding the bulge where stem cells reside (p).The Msx2-rtTA-driven NFATc1 activation did not elicit tumorigenesis in the skin of mutants treated for 3 months since weaning (q, r). Immunostaining of human SCC sections shows the upregulation of NFATc1, compared with normal human skin sections (s, t). Inset in (t) shows the NFATc1 immunostaining in higher magnification. could not distinguish between the endogenous NFATc1 and the follicles as shown by b-galactosidase activity in Figure 2i. Co- transgene-derived NFATc1Nuc proteins (Figure 3b). We thus immunostaining of mutant hair follicles with NFATc1 and HA examined the presence of the hemagglutinin (HA) tag fused to antibodies showed that most NFATc1 þ cells were also HA þ the NFATc1Nuc protein that specifically recognizes the transgene (Figures 3e and f). In addition, the skin tumors contained a large product. HA signal was absent in the control without transgene number of cells that co-express NFATc1 and CD34, indicating a expression (Figure 3c) but present in the CD34 þ bulge stem cells potential abnormal expansion of the skin stem cell population in the Dox-treated mutant (Figure 3d), further confirming that the within the tumor (Figures 3g–i), likely as a direct effect of Cre expression targets CD34 þ stem cell populations in the hair transgene-induced NFATc1 activation in the CD34 þ hair follicle

& 2014 Macmillan Publishers Limited Oncogene (2014) 1840 – 1849 NFAT signaling and tumorigenesis P Tripathi et al 1844 stem cells. Interestingly, while a high percentage of cells within As NFATc1 is a transcription factor, the observed upregulation of the tumor expressed NFATc1, the cells surrounding the keratin c-Myc and Stat3 activation in NFATc1 À cells reflects indirect non- pearls were exclusively devoid of NFATc1 and CD34 (Figure 3h). cell autonomous effects of NFATc1 activation, likely initiated by Keratin-14, a known diagnostic marker for SCC, had extensive the secretion of factors that have paracrine functions. expression in NFATc1-induced SCC-like tumors, especially in cells As NFATc proteins are known regulators of a number of cytokines surrounding the keratin pearls (Figure 3j). A potential explanation and interleukin (IL)-6, in particular, has been shown to regulate Stat for these data is that cells with NFATc1 activation did not activation in tumors,17 we sought to investigate the potential role of participate in, but created the environment for, the formation of IL-6 and other cytokines in mediating the effects of NFATc1 the keratin pearls and that multiple lineages of cells participated in activation. By reverse transcriptase PCR (RT–PCR), we showed the development of the SCC-like tumor with NFATc1 þ , CD34 þ that NFATc1 activation was accompanied by an increased IL-6 cells expanding and acting as tumor-inducing cells for NFATc1 À transcription in skin tumors (Figure 4h). Areas of high levels of cells at the same time. IL-6 are found in both skin and ovarian tumors (Figures 4I and j). CD44 is expressed in multiple areas in normal skin, including bulge However, many areas with high IL-6 levels were populated primarily cells with NFATc1 expression (Figure 3k). Co-expression of CD44 and by NFATc1 À cells (Figures 4i and j), suggesting that NFATc1 did not NFATc1 was widely seen in mutant hair follicles and NFATc1-induced directly upregulate IL-6 transcription in these cells. Although some skin tumors (Figures 3l and m), likely due to the expansion of the IL-6 may be secreted by infiltrating immune cells, high IL-6 levels NFATc1 þ ,CD44þ hair follicle cells. However, CD44 was also present can be seen in local-resident non-immune cells (Figures 4i–k and in cells within the keratin pearls that are NFATc1 À (Figures 3n and data not shown). Thus, NFATc1 þ cells appeared to signal their o). As CD44 has expression in other skin cells, such as keratinocytes neighboring cells, likely through a paracrine action, to secrete IL-6. (Yasaka et al.15 and data not shown), NFATc1 is probably acting The upregulation of IL-6, and possibly other cytokines, provides an through cell autonomous effects on CD44 þ , NFATc1 þ cells for their explanation to the observed widespread nuclear presence of Stat3 expansion and through non-cell autonomous effects on CD44 þ , in these tumors (Figures 4c, f and g). These findings suggest that a NFATc1 À cells (likely including keratinocytes) for their participation major effect of NFATc1 activation is the creation of a promitogenic in the keratin pearls. In a separate model (Msx2-rtTA; TetO-NFATc1Nuc) local environment for both the tumor-initiating cells and their where NFATc1Nuc expression was Dox-induced in hair follicle matrix neighbors to contribute in tumor formation. cells and hair shaft excluding the bulge, no skin tumor was observed 16 (Figures 3p–r and Lin et al. ). As the key difference in the targeted Allografts of NFATc1-induced tumors showed continuous cell population between the two models was the presence of dependency on NFATc1 for tumor progression and survival but NFATc1Nuc in bulge stem cells in the Hoxb7-Cre; ROSArtTA; TetO- Nuc not on T-cell functions NFATc1 model, these results suggest that the abnormal activation To directly test the essential role of NFATc1 activation in of bulge stem cells could have an important role in the tumorigenesis and to examine the involvement of T cells in tumorigenesis of the SCC-like skin tumor. establishing the tumorigenic microenvironment, we studied the To assess the prevalence of NFATc1 expression in the human ability of the cells with NFATc1 activation to initiate tumorigenesis SCC, we performed NFATc1 immunostaining on human SCC tissue in athymic nude mice with severe T-cell deficiency. We first arrays that consisted of human SCC specimens histopathologically cultured the cells from an NFATc1-induced ovarian tumor (Figures confirmed by board-certified pathologists. Among 42 human SCC 5a and b). These cells were then injected to the rear flanks of nude specimens investigated, we observed NFATc1-expressing cells in mice. Tumor growth was detected at about 6 weeks after the 24 (57% of the total) specimens (Figures 3s and t). Twelve control injection in the Dox-treated recipient mice, but not in untreated samples analyzed show no NFATc1 expression beyond that in the ones (Figure 5c). To further test the dependency of tumor growth hair follicle bulge (Figures 3s and t). Thus, NFATc1 expression and progression on NFATc1 activation, we stopped Dox treatment appears to be associated with a substantial percentage of the in a subgroup of these mice. Existing tumors started to shrink human SCCs we examined. within days after Dox withdrawal (Figure 5d). Such trend was Activation of oncogenic proteins in NFATc1 À cells appears to result reversed when NFATc1 activation was restored (Figures 5d and e), indicating a continuous dependency of the tumors on NFATc1 from non-cell autonomous mechanisms 18,19 þ activation, similar to that seen in cases of oncogene addiction. Although NFATc1 cells were extensively present in the NFATc1- The tumor allografts retain some key characteristics of the primary induced tumors, not all NFATc1 þ cells were proliferating while À tumors in many aspects, including the presence of a large number many neighboring NFATc1 cells were undergoing active of NFATc1 þ cells (Figure 5f). These allografts were also primarily proliferation. In the example of a skin tumor given in Figures 4a þ ± þ composed of aSMA mesenchymal cells (Figure 5g). Similar to and d, 53.67 14.22% of NFATc1 cells in the tumor were the primary tumor, NFATc1 þ and NFATc1 À cells intermingled and proliferating (PCNA þ ). Within the same tumor, a smaller but still ± À were both actively proliferating (Figure 5h). NFATc1 activation also substantial percentage (26.92 7.43%) of the NFATc1 cells were stimulated Stat3 activation in both NFATc1 þ and NFATc1 À cells also proliferating (Figure 4d). NFATc1 activation in these (Figure 5i). While T cells are seen in the primary tumor, the grafts tumors had induced extensive c-Myc expression (Figure 4b). grown in nude mice are completely void of T cells (Figures 5j and While about 83.74±3.17% NFATc1 þ cells were expressing c-Myc, ± À k). As inflammatory cytokines, such as IL-6, are similarly 37.07 6.42% NFATc1 cells also had c-Myc expression upregulated in the grafts, the establishment of the inflammatory (Figure 4e). Although earlier reports have suggested that NFATc1 tumorigenic microenvironment is thus independent of T cells can directly regulate c-Myc expression,7 our finding of c-Myc À (Figures 4j and k, 5l, and data not shown) but likely dependent on activation in a significant percentage of NFATc1 cells raised the local tissue-resident cells and other immune cells in this model. possibility that NFATc1 activation may also regulate c-Myc expression in a non-cell autonomous fashion through paracrine actions. By using an antibody targeting phospho-Stat3 (Tyr705), Key molecular components of a proinflammatory we found that 53.14±16.23% NFATc1 þ cells and 26.18±9.42% microenvironment induced by NFATc1 activation NFATc1 À cells had nuclear phosphorylated Stat3 in NFATc1- To capture the broader picture of molecular changes related to induced tumors (Figures 4c and f). Immunoblotting of lysates from the oncogenic effects of NFATc1 activation and to further normal skin and NFATc1-induced skin tumors showed that the characterize the molecular components in the microenvironment level of phospho-Stat3 was greatly increased in NFATc1-induced of NFATc1-induced tumors, we performed whole-genome tran- skin tumors, though total Stat3 level was unchanged (Figure 4g). script profiling on three sets of samples: hair follicles and skin

Oncogene (2014) 1840 – 1849 & 2014 Macmillan Publishers Limited NFAT signaling and tumorigenesis P Tripathi et al 1845 a PCNA/NFATc1/DAPI b c-Myc/NFATc1/DAPI c p-Stat3/NFATc1/DAPI

Skin tumor Skin tumor Skin tumor d e f 80 100 P<0.0001 80 70 P=0.0004 90 (n=4) 70 P=0.001 (n=4) 80 (n=4) 60 60 70 50 60 50 40 50 40

30 40 30 30 20 20 20 10 10 10 0 0 0 % of NFATc1+ % of NFATc1– % of NFATc1+ % of NFATc1– % of NFATc1+ % of NFATc1– cells being cells being cells being cells being cells being cells being PCNA+ PCNA+ c-Myc+ c-Myc+ p-Stat3+ p-Stat3+ g CT CT MT MT i NFATc1/ j NFATc1/ k NFATc1/ p-Stat3 IL6/DAPI IL6/DAPI IL6/DAPI

Stat3

NFATc1

-Tubulin

CT CT MT MT h IL6

NFATc1

Gapdh Skin tumor Ovarian tumor Ovarian tumor Figure 4. High level of heterogeneity in the tumors and the non-cell autonomous promitogenic effects of NFATc1 activation. Immunostaining using NFATc1 and PCNA as well as the subsequent statistical analyses (two-tailed t-test) indicated that a high percentage of NFATc1 þ cells were actively proliferating (53.67±14.22%). The percentage of proliferating NFATc1 À cells (26.92±7.43%, P ¼ 0.0004, n ¼ 4), albeit smaller, was very substantial (a, d). NFATc1 activation led to significant upregulation of c-Myc in the tumor tissue (b, e). 83.74±3.17% of the NFATc1 þ cells expressed c-Myc, whereas only 37.07±6.42% NFATc1 À cells were c-Myc þ (Po0.0001, n ¼ 4) (e). Double immunostaining with NFATc1 and phospho-Stat3 antibodies showed higher percentage of NFATc1 þ cells with nuclear Stat3 (53.14±16.23%) than NFATc1 À cells with nuclear Stat3 (26.18±9.42%, P ¼ 0.001, n ¼ 4) (c, f). Western blot analysis of normal skin lysates from control mice and skin tumor from mutant mice was done by using an antibody against Stat3 (revealing the level of Stat3 proteins regardless of phosphorylation status) and an antibody against phospho-Stat3 (Tyr705, revealing the level of activated Stat3). The results showed activation of Stat3 specifically in NFATc1-induced skin tumors while the level of total Stat3 was unchanged (g). NFATc1 activation specifically upregulated IL-6 in the tumors as revealed by RT–PCR (h). IL-6 upregulation was also detected by immunostaining in skin (i) and ovarian tumors (j, k). CT, control; MT, mutant. collected from mutant and control mice treated with Dox for 2 Besides cytokines, a number of genes with the largest fold days as well as NFATc1-induced skin tumors and normal skin from changes are related to acute inflammation, innate immune similarly treated littermate controls. In each sample sets, gene response and recruitment of lymphocytes (SAA3, SAA1, LCN2 ontology) analysis for biological process indicated that most of the and RETNLG), as well as to keratinization of the skin (SPRRs, significant changes (fold change 42, Po0.05) were in genes EG433016, CHI3Ls and KRTs) (Supplementary Tables 1–3). These related to cell proliferation, growth and metabolic processes, all of results are consistent with the histological findings of keratin which are important for tumor initiation and progression pearls and the extensive presence of cells with keratin-14 (Figure 6a). One hundred and fifty-six significantly changed genes expression in skin tumors. We found that the skin epidermis were present across all three data sets (Figure 6b). Many became hyperproliferative after 2 days of NFATc1 activation in chemokines/cytokines-related genes were among the most mutants (Figures 6e and f). Increased levels of proliferation, as significantly changed genes. These and other selected genes with evidenced by PCNA staining, was observed in the basal layer of high fold changes were subjected to RT–PCR analysis (Figure 6c) the epidermis and in the outer root sheath cells of the hair follicle with a subset further confirmed by quantitative PCR (Q-PCR) in the mutants 2 days after NFATc1 activation (Figures 6g and h). (Figure 6d). All genes examined by Q-PCR showed increased As transgene expression was not detected in some of these cells, expression at levels similar to those seen in microarray analyses. the hyperproliferation in these NFATc1 À cells again pointed to

& 2014 Macmillan Publishers Limited Oncogene (2014) 1840 – 1849 NFAT signaling and tumorigenesis P Tripathi et al 1846 a b NFATc1/DAPI c

Without Dox

With Dox

Ovarian tumor culture Ovarian tumor culture With Dox

d Dox stopped/Dox restarted e 3000 P=0.001 P=0.001 3500 (n=4) (n=4) 3 Sacrificed 3 P=0.04 3000 2500 (n=4) Dox stopped 2500 2000

2000 Sacrificed 1500 1500

1000 1000

500 500 Nude mice tumor volume in mm Nude mice tumor volume in mm

0 0 30 40 50 60 70 80 90 100 110 120 No Dox Dox Dox Dox Days after ovarian tumor cell injection in nude mice Provided Stopped Restarted upto d64 d76-99 d99-120 f NFATc1/-Gal/DAPI g SMA/CK8 h NFATc1/PCNA

Nude mice+Ovarian tumor cells Nude mice+Ovarian tumor cells Nude mice+Ovarian tumor cells

i NFATc1/p-Stat3/DAPI j CD3/DAPI k CD3/DAPI l IL6/DAPI

Nude mice + Nude mice + Nude mice + Ovarian tumor cells Primary Ovarian tumor Ovarian tumor cells Ovarian tumor cells

Figure 5. Allografts of NFATc1-induced tumors showed continuous dependency on NFATc1 for tumor progression and survival but not on T-cell functions. Ovarian tumors from mutants were isolated and cultured (a). Most of the cultured cells expressed NFATc1 (b). Cultured tumor cells were injected subcutaneously into the lower flanks of nude mice. Hundred percentage of the Dox-treated recipients developed tumors, whereas none of the untreated mice did (c). Termination of Dox treatment resulted in significant decrease of tumor size. Such decrease was reverted again if Dox treatment was restarted (d). Statistical analysis (two-tailed t-test) showed a significant increase in tumor volume when Dox was continuously provided up to 64 days (1940.54±714.34 mm3, P ¼ 0.001, n ¼ 4). Conversely, a significant decrease in tumor volume (40.80±81.60 mm3) was observed when Dox was interrupted from day 76 to day 99 (P ¼ 0.001, n ¼ 4). After Dox was restarted on day 99, tumor volume increased again and reached 702.34±523.30 mm3 (P ¼ 0.04, n ¼ 4) on day 120 (e). Tumor size was measured at the end of each interval. Immunostaining of tumors from nude mice indicated that the majority of tumor cells were b-galactosidase (b-Gal) þ (f). The allografts in the nude mice resembled the original tumors in overall morphology, cellular composition (g), and the substantial levels of proliferation and Stat3 activation in NFATc1 À cells (h, i). By immunostaining with an anti-CD3 antibody, T cells were seen in primary ovarian tumor but not in the grafts grown in nude mice (j, k). Extensive IL-6 expression was noticed in the grafts in the nude mice (l).

Oncogene (2014) 1840 – 1849 & 2014 Macmillan Publishers Limited NFAT signaling and tumorigenesis P Tripathi et al 1847 Cell Proliferation a 20.06% b 22.42% Metabolic Process 15.43% 7.20% 29.94% 7.61% Growth 30.29% 23.61% Locomotion Hair follicle with 30.23% 11.73% NFAT activation Death NFAT-induced for 2 days skin tumor (1200) Hair Follicle 14.01% 10.20% Skin Pigmentation (1800) 8.25% Cellular Process Common 21.29% 2.97% Genes Viral Reproduction (156) 40.04% Rhythemmic Process Skin with NFAT activation for 2 days (460) Skin Tumor c d CT CT MT MT CT CT MT MT 10000 Ccl3 Lcn2

Ccl4 Osm Fold Change by Microarray Ccr5 Osmr 1000 Fold Change by Q-PCR Cd14 Retnlg Chi3l3 S100a8 100 Csf2 S100a9 Csf2rb2 Saa1 IL1a Saa3 10 IL1b Spp1 IL6 Sprr2d 1 IL19 Sprr2e Fold change difference in transcript

IL24 Gapdh level in skin after NFAT activation (log10)

e H&E f H&E g PCNA/DAPI h PCNA/DAPI

Control Mutant Control Mutant i IL1/DAPI j IL1/DAPI k OSM/DAPI l OSM/DAPI

Control Mutant Control Mutant

Figure 6. Transcript profiling revealed that NFATc1 activation creates a proinflammatory microenvironment for tumor development. Gene ontology analysis revealed that the top categories for genes with significant transcriptional changes include cell proliferation, growth and metabolic processes (a). Venn diagram showed 156 genes commonly changed across all three sets (b). A number of the secretory factors were among the genes with the most significant changes. These and other significantly changed genes with potential links to NFAT signaling were subjected to further RT–PCR using skin from mutant and control mice treated with Dox for 48 h (c). CT, skin from control mice; MT, skin from mutant mice. Some of these genes were further validated by Q-PCR, which shows similar trend of expression change in mutant skin as suggested by microarray (d). hematoxylin and eosin and PCNA staining showed hyperproliferative mutant skin epidermis after 2 days of NFATc1 activation compared with control samples (e–h). Immunostaining showed more extensive expression of IL-1b and oncostatin M in skin from mutant mice compared with their littermate controls (i–l). non-cell autonomous effects of NFATc1 activation, most likely the tumorigenic microenvironment and the progression of these through alterations in microenvironment. By immunostaining, we tumors. have identified more extensive expression of IL-1b in skin from mutant mice compared with controls 2 days after NFATc1 activation (Figures 6I and j). Similar increases in oncostatin M DISCUSSION levels were observed in mutant skin (Figures 6k and l). The We have previously shown that calcineurin–NFAT signaling is extensive presence of proinflammatory cytokines at very early involved in the development of multiple organ systems and in stage of NFATc1 activation suggests that tissue-resident cells the cellular responses to extracellular environment.1,20–23 As we are capable of initiating the establishment of a promitogenic continue these investigations, we serendipitously discovered that environment before significant infiltration of immune cells. The the activation of NFAT signaling in vivo can cause tumor formation cytokine cascades within these local cells and subsequently in in a subset of tissues, including the ovary and skin. Our studies on infiltrating immune cells may support the continuous evolution of the primary and the transplanted tumors provide direct in vivo

& 2014 Macmillan Publishers Limited Oncogene (2014) 1840 – 1849 NFAT signaling and tumorigenesis P Tripathi et al 1848 evidence that NFATc1 activation is sufficient for initiating redundancy.27,28 In the tumor model, the key for stem cell expansion tumorigenesis and is necessary for the progression of the may be in the alteration of the stem cell microenvironment/niche. tumors formed. At least in the systems we tested, NFATc1 can This is consistent with the finding that escape of hair follicle stem behave as a pleiotropic oncogene and tumors caused by its cells from growth-restrictive niches was a key process in wounding- activation remain addicted to it. induced tumorigenesis in skin.29 The expansion of the CD34 þ stem Although the effects of NFATc1 activation may differ in different cells can only partially explain the NFATc1-induced skin tumor tissues, it is common for the NFATc1-induced tumors we studied, formation that may involve a complex set of sequential or parallel as well as the tumor allografts in the nude mice, that NFATc1 processes, including the induction of other cell types to build activation has effects beyond the cells where the activation structures like the keratin pearls and the possible differentiation of occurs. Although c-Myc upregulation and Stat3 activation have the CD34 þ stem cells into other cell types. been observed in some, but not all, NFATc1 þ cells in NFATc1- Upregulation of NFATc1 expression has been reported pre- induced tumors, many NFATc1 À cells within the tumor also show viously in a number of human malignancies.10 Our finding in upregulation of c-Myc and activation of Stat3 (Figures 4b, c, e and human SCC specimens further expands the tumor types in which f). These findings suggest that the role of NFAT goes beyond cell significant NFATc1 expression can be found in tumor cells and autonomous effects in ways of direct regulation of c-Myc may contribute to the initiation and progression of these tumors. transcription and Stat3 activation. Combined with our analyses As our study indicates that NFATc1 activation has pleiotropic of the molecular and cellular components of the microenviron- effects on the cells expressing it and on the neighboring cells in ment in these tumors, a picture emerges that factors secreted by the context of alterations in a number of signaling pathways and the NFATc1 þ cells initiate the establishment of an inflammatory in the microenvironment, inhibiting NFATc1 activation could be and promitogenic tissue environment for both NFATc1 þ and more effective than targeting one or more of the downstream NFATc1 À cells to proliferate and become part of the tumor. From pathways and factors in treating tumors with NFATc1 activation. our comparison of two models with or without hair follicle stem The oncogene addiction revealed in the tumor allograft study cell involvement, it appears that both the cell autonomous effects further suggests that NFAT signaling may be an attractive target of NFATc1 activation and the alterations in tissue microenviron- for anti-tumor therapy. ment may affect the stem cells/progenitor cells and their niches, leading to their participation in tumor formation. One intriguing question is the nature of this secreted factor(s) MATERIALS AND METHODS that has the autocrine/paracrine function. This secreted factor is Mouse (Mus musculus) strains and Dox treatment most likely a major initiating factor for, and a component of, the All animal studies were approved by the Washington University Animal proinflammatory and promitogenic microenvironment. Our find- Studies Committee and have been conducted according to relevant NIH ing of elevated IL-6 transcript level in the tumors led to the guidelines. The Hoxb7-Cre, ROSArtTA, TetO-NFATc1Nuc strains and the 2,11,23,30 suspicion of IL-6 fulfilling the role of such a secreted factor relevant genotyping methods were described previously. Nude (Figure 4h). However, further immunostaining of the tumor mice (Foxn1nu, Harlan, Indianapolis, IN, USA) were used for tumor grafting tissues, especially tissues from early-stage tumors, revealed at 6–8 weeks old. Dox was given through drinking water provided ad libitum at 2 mg/ml. Dox treatment was started at conception and at birth distinct populations of cells expressing NFATc1 and IL-6, through water intake by the mother or at 21 days postnatally by the respectively (Figures 4i–k). Thus, it appears that the upregulation targeted mice. of IL-6 is itself induced by this secreted factor(s). Transcript profiling suggested the presence of proinflammatory microenvironment in the very early stages after NFATc1 activation (Figure 6). Primary tumor cell culture For primary tumor cell culture, ovarian tumors from 5-week-old Hoxb7-Cre; There is growing evidence supporting the role of proinflammatory rtTA Nuc cytokines in tumor progression.24 Upregulation of a number of ROSA ; TetO-NFATc1 mice were harvested and cut into small pieces of 1 mm and cultured in Dulbecco’s Modified Eagle’s Medium-F12 (10% proinflammatory cytokines (Figures 4 and 6), as a result of NFATc1 o þ fetal bovine serum, 5% penicillin/streptomycin and 2 mg/ml Dox). Cells activation, can provide the mitogenic signals for both NFATc1 grew out of the tumor tissue chunks were fed with fresh media every 2–3 À þ and NFATc1 cells to proliferate. The ability of the NFATc1 cells days, and subcultured before confluence. Cultured cells were tested for to recruit neighboring NFATc1 À cells into the formation of tumors nuclear NFATc1 expression in the presence of Dox. (beyond the vasculature) also makes the NFATc1 þ cells behave like tumor-initiating cells. Whether the cytokine genes are direct or Tumor growth in nude mice indirect targets of NFATc1 is still under investigation. It is also For allografts, aforementioned cultured tumor cells (o15 passages) were possible that one of the cytokine genes is directly regulated by grown to 70–80% confluence, trypsinized and resuspended in phosphate- NFATc1 and its upregulation initiates a cascade of cytokine buffered saline. 3 Â 106 cells were injected subcutaneously into lower production, leading to the establishment of the proinflammatory flanks of eight nude mice. Group 1 mice were untreated. Groups 2–4 were and promitogenic environment. Dox-treated from the day of injection. Group 2 was killed on the 64th day. NFATc1 is normally expressed in the CD34 þ hair follicle bulge Dox was stopped for group 3 on the 76th day and restarted on the 99th stem cells. Inhibition of calcineurin/NFATc1 signaling has been day. Group 3 mice were finally killed on the 127th day. Dox was stopped for group 4 on the 99th day and the mice were killed on the 113th day. shown to relieve the NFATc1-mediated repression of CDK4 and 31 leads to stem cell activation and hair growth.14 In the Hoxb7-Cre, Tumor volume was determined using a previously described formula. ROSArtTA, TetO-NFATc1Nuc mutants, NFATc1 activation in hair follicle stem cells led to apparent expansion of the CD34 þ stem cells and Histology, immunostaining and western blotting the development of SCC-like skin tumor (Figures 3h and i). The Tissues were fixed with 4% paraformaldehyde and embedded in paraffin. involvement of the stem cells in this model is further supported by Seven micrometer sections were collected and stained with hematoxylin the absence of skin tumors in the Msx2-rtTA, TetO-NFATc1Nuc and eosin. Human tissue microarrays were obtained from Biomax (Biomax Inc, Rockville, MD, USA). X-Gal staining and immunostaining on cryostat mutants where NFATc1 activation is in an overlapping set of skin 21,32 cells excluding the bulge stem cells. These results seem contra- sections were performed as described. Antibodies used were: rabbit anti-CYP11a1 (Chemicon, Billerica, MA, USA, AB1294, 1:200), mouse dictory to the finding by Horsley et al. on surface. However, it is well monoclonal anti-aSMA (Sigma, St Louis, MO, USA, A2547, 1:200), rat anti- known that the calcineurin/NFAT pathway operates through cytokeratin 8 (TROMA-1, Developmental Studies Hybridoma Bank, Iowa, IA, establishing a balanced signaling with additional inputs from other USA, 1:200), mouse monoclonal anti-NFATc1 (BD Pharmingen, La Jolla, CA, 25,26 pathways in different cellular contexts. CDK2 and CDK4 have USA, 556602, 1:100), mouse monoclonal anti-CD3 (DAKO, Carpinteria, CA, been found to be dispensable for cell cycle progression, likely due to USA, A0452, 1:200), rat monoclonal anti-CD34 (Abcam, Cambridge, MA,

Oncogene (2014) 1840 – 1849 & 2014 Macmillan Publishers Limited NFAT signaling and tumorigenesis P Tripathi et al 1849 USA, ab8158, 1:50), rat monoclonal anti-CD44 (BD Pharmingen 558739, 6 Pham LV, Tamayo AT, Yoshimura LC, Lin-Lee YC, Ford RJ. Constitutive 1:50), rabbit monoclonal anti-E-cadherin (Cell Signaling Technology, NF-kappaB and NFAT activation in aggressive B-cell lymphomas synergistically Danvers, MA, USA, 3195, 1:200), rabbit polyclonal anti-IL-6 (Abcam activates the CD154 gene and maintains lymphoma cell survival. Blood 2005; 106: ab6672, 1:600), rabbit polyclonal anti-PCNA (Bethyl Labs, Montgometry, 3940–3947. TX, USA, IHC-00012, 1:200), rabbit polyclonal anti-c-Myc (Abcam ab39688, 7 Buchholz M, Schatz A, Wagner M, Michl P, Linhart T, Adler G et al. Overexpression 1:200), rabbit monoclonal anti-Stat3 (Cell Signaling 4904, 1:50), rabbit of c-myc in pancreatic cancer caused by ectopic activation of NFATc1 and the polyclonal anti-phospho-Stat3 (Cell Signaling 9131, 1:100), chicken Ca2 þ /calcineurin signaling pathway. EMBO J. 2006; 25: 3714–3724. polyclonal anti-b-galactosidase (Abcam 9361, 1:100), rabbit anti-b- 8 Jauliac S, Lopez-Rodriguez C, Shaw LM, Brown LF, Rao A, Toker A. The role of Catenin (Sigma C2206, 1:200), rat monoclonal anti-PECAM (BD NFAT transcription factors in integrin-mediated carcinoma invasion. Nat Cell Biol Pharmingen 550274, 1:100), rabbit polyclonal anti-IL-1b (Santa Cruz 2002; 4: 540–544. Biotechnology, Santa Cruz, CA, USA, sc7884, 1:50) and goat polyclonal 9 Muller MR, Rao A. NFAT, immunity and cancer: a transcription factor comes of anti-oncostatin M (R&D Systems, Minneapolis, MN, USA, af495na, 1:50). age. Nat Rev Immunol 2010; 10: 645–656. Appropriate Alexa Flour488 or 555-conjugated secondary antibodies 10 Pan MG, Xiong Y, Chen F. NFAT gene family in inflammation and cancer. Curr Mol (Molecular Probes, Invitrogen, Carlsbad, CA, USA, 1:1000) were used for Med (in press). detection. For western blotting, protein extracts were separated by SDS– 11 Zhao H, Kegg H, Grady S, Truong HT, Robinson ML, Baum M et al. Role of PAGE, transferred to Immobilon-P membranes (Millipore, Bedford, MA, fibroblast growth factor receptors 1 and 2 in the ureteric bud. Dev Biol 2004; 276: USA) and probed by the appropriate primary antibodies. After incubation 403–415. with Alexa680-conjugated goat anti-rabbit IgG (Molecular Probes) and 12 Guo Q, Tripathi P, Poyo E, Wang Y, Austin PF, Bates CM et al. Cell death serves as a IRDye800CW-conjugated goat anti-mouse IgG (Rockland, Gilbertsville, PA, single etiological cause of a wide spectrum of congenital urinary tract defects. USA) secondary antibodies, antibody complexes were visualized by an J Urol 2011; 185: 2320–2328. 32 Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA). 13 Anderson B, Turner DA, Benda J. Ovarian sarcoma. Gynecol Oncol 1987; 26: 183–192. RT–PCR, microarray and Q-PCR 14 Horsley V, Aliprantis AO, Polak L, Glimcher LH, Fuchs E. NFATc1 balances quies- cence and proliferation of skin stem cells. Cell 2008; 132: 299–310. Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA) 15 Yasaka N, Furue M, Tamaki K. CD44 expression in normal human skin and skin and purified with RNeasy Mini Kit (Qiagen, Valencia, CA, USA). Comple- tumors. J Dermatol 1995; 22: 88–94. mentary DNA was prepared from RNA using Invitrogen ThermoScript RT– 16 Lin C, Yin Y, Chen H, Fisher AV, Chen F, Rauchman M et al. Construction and PCR System (Invitrogen). The detailed information about the primer characterization of a doxycycline-inducible transgenic system in Msx2 expressing sequences and PCR conditions are provided as Supplementary Table 4. For cells. Genesis 2009; 47: 352–359. transcript profiling, three pairs of control and mutant mice were treated 17 Lesina M, Kurkowski MU, Ludes K, Rose-John S, Treiber M, Kloppel G et al. Stat3/ with Dox for 48 h before hair follicles and dorsal skin patches were Socs3 activation by IL-6 transsignaling promotes progression of pancreatic harvested. In addition, three skin tumors from mutants treated with Dox intraepithelial neoplasia and development of pancreatic cancer. Cancer Cell 2011; from P21 for 3–4 weeks and three normal skin samples from similarly 19: 456–469. treated control mice were also collected. Microarray was done on 18 Torti D, Trusolino L. Oncogene addiction as a foundational rationale for targeted MouseWG-6 v2.0 Expression BeadChips (Illumina, San Diego, CA, USA) at anti-cancer therapy: promises and perils. EMBO Mol Med 2011; 3: 623–636. the Genome Institute at Washington University in St Louis. Results were 19 McCormick F. Cancer therapy based on oncogene addiction. J Surg Oncol 2011; analyzed using the Partek Genomics Suite (Partek Inc., St Louis, MO, USA). 103: 464–467. Genes with significant changes (X2-fold and P 0.05) were further o 20 Graef IA, Chen F, Chen L, Kuo A, Crabtree GR. Signals transduced by Ca(2 )/ validated by Q-PCR. Complementary DNA was generated with oligo-dT þ calcineurin and NFATc3/c4 pattern the developing vasculature. Cell 2001; 105: primers using the Superscript III Kit (Invitrogen). Real-time quantification 863–875. was done using a SYBRGreen PCR kit (Life Technologies, Carlsbad, CA, 21 Chang CP, McDill BW, Neilson JR, Joist HE, Epstein JA, Crabtree GR et al. Calci- USA). Relative gene expression was deduced using the DDCt method. neurin is required in urinary tract mesenchyme for the development of the pyeloureteral peristaltic machinery. J Clin Invest 2004; 113: 1051–1058. CONFLICT OF INTEREST 22 Li SZ, McDill BW, Kovach PA, Ding L, Go WY, Ho SN et al. Calcineurin-NFATc The authors declare no conflict of interest. signaling pathway regulates AQP2 expression in response to calcium signals and osmotic stress. Am J Physiol Cell Physiol 2007; 292: C1606–C1616. 23 Wang Y, Jarad G, Tripathi P, Pan M, Cunningham J, Martin DR et al. Activation of ACKNOWLEDGEMENTS NFAT signaling in podocytes causes glomerulosclerosis. J Am Soc Nephrol 2010; We thank Dr Crabtree for providing the TetO-NFATc1Nuc mice and Dr Yuan Zhu for 21: 1657–1666. helpful discussion of the study. FC is supported in part by institutional funds from the 24 Sansone P, Bromberg J. Environment, inflammation, and cancer. Curr Opin Genet Department of Medicine at Washington University School of Medicine and NIH grants Dev 2011; 21: 80–85. (DK81592 and DK67386). We also thank the George M. O’Brien Center for Kidney 25 Wu H, Peisley A, Graef IA, Crabtree GR. NFAT signaling and the invention of Disease Research at Washington University (P30DK079333) for core services. We vertebrates. Trends Cell Biol 2007; 17: 251–260. thank Michael Heinz from the Genome Technology Access Center (P30 CA91842, 26 Rao A. Signaling to gene expression: calcium, calcineurin and NFAT. Nat Immunol. UL1RR024992) in the Department of Genetics at Washington University School of 2009; 10: 3–5. Medicine for help with gene expression analysis. 27 Barriere C, Santamaria D, Cerqueira A, Galan J, Martin A, Ortega S et al. Mice thrive without Cdk4 and Cdk2. Mol Oncol 2007; 1: 72–83. 28 Li W, Kotoshiba S, Berthet C, Hilton MB, Kaldis P. Rb/Cdk2/Cdk4 triple mutant mice REFERENCES elicit an alternative mechanism for regulation of the G1/S transition. Proc Natl 1 Graef IA, Chen F, Crabtree GR. NFAT signaling in vertebrate development. Curr Acad Sci USA 2009; 106: 486–491. Opin Genet Dev 2001; 11: 505–512. 29 Wong SY, Reiter JF. Wounding mobilizes hair follicle stem cells to form tumors. 2 Pan M, Winslow MM, Chen L, Kuo A, Felsher D, Crabtree GR. Enhanced NFATc1 Proc Natl Acad Sci USA 2011; 108: 4093–4098. nuclear occupancy causes T cell activation independent of CD28 costimulation. 30 Belteki G, Haigh J, Kabacs N, Haigh K, Sison K, Costantini F et al. Conditional and J Immunol 2007; 178: 4315–4321. inducible transgene expression in mice through the combinatorial use of 3 Neal JW, Clipstone NA. A constitutively active NFATc1 mutant induces a trans- Cre-mediated recombination and tetracycline induction. Nucleic Acids Res 2005; formed phenotype in 3T3-L1 fibroblasts. J Biol Chem 2003; 278: 17246–17254. 33:e51. 4 Marafioti T, Pozzobon M, Hansmann ML, Ventura R, Pileri SA, Roberton H et al. The 31 Janik P, Briand P, Hartmann NR. The effect of estrone-progesterone treatment on NFATc1 transcription factor is widely expressed in white cells and translocates cell proliferation kinetics of hormone-dependent GR mouse mammary tumors. from the cytoplasm to the nucleus in a subset of human lymphomas. Br J Cancer Res 1975; 35: 3698–3704. Haematol 2005; 128: 333–342. 32 McDill BW, Li SZ, Kovach PA, Ding L, Chen F. Congenital progressive hydrone- 5 Medyouf H, Alcalde H, Berthier C, Guillemin MC, dos Santos NR, Janin A et al. phrosis (cph) is caused by an S256L mutation in aquaporin-2 that affects its Targeting calcineurin activation as a therapeutic strategy for T-cell acute phosphorylation and apical membrane accumulation. Proc Natl Acad Sci USA lymphoblastic leukemia. Nat Med 2007; 13: 736–741. 2006; 103: 6952–6957.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

& 2014 Macmillan Publishers Limited Oncogene (2014) 1840 – 1849