Analysis of a Mouse Skin Model of Tuberous Sclerosis Complex

Tuberous Sclerosis Complex (TSC) is an autosomal dominant tumor suppressor gene syndrome in which patients develop several types of tumors, including facial angiofibroma, subungual fibroma, Shagreen patch, angiomyolipomas, and lymphangioleiomyomatosis. It is due to inactivating mutations in TSC1 or TSC2. We sought to generate a mouse model of one or more of these tumor types by targeting deletion of the Tsc1 gene to fibroblasts using the Fsp-Cre allele. Mutant, Tsc1ccFsp-Cre+ mice survived a median of nearly a year, and developed tumors in multiple sites but did not develop angiomyolipoma or lymphangioleiomyomatosis. They did develop a prominent skin phenotype with marked thickening of the dermis with accumulation of mast cells, that was minimally responsive to systemic rapamycin therapy, and was quite different from the pathology seen in human TSC skin lesions. Recombination and loss of Tsc1 was demonstrated in skin fibroblasts in vivo and in cultured skin fibroblasts. Loss of Tsc1 in fibroblasts in mice does not lead to a model of angiomyolipoma or lymphangioleiomyomatosis.


Introduction
Tuberous Sclerosis Complex (TSC) is an autosomal dominant genetic disorder due to inactivating mutations in either TSC1 or TSC2. There are well-established consensus diagnostic criteria for TSC; the major clinical criteria include angiofibromas or fibrous cephalic plaque, ungual fibromas, shagreen patch, angiomyolipomas, and lymphangioleiomyomatosis [1]. Tumors that develop in TSC are known to develop through a two hit mechanism, in which there is uniform germ-line or mosaic inactivating mutation in TSC1 or TSC2, and second hit inactivating mutation in the other allele of the same gene to lead to loss of function of either TSC1 or TSC2 in the cells comprising the tumor [2][3][4][5][6][7][8]. Similar findings have been made in the majority of sporadic angiomyolipoma and lymphangioleiomyomatosis lesions [3][4][5]8].
Heretofore no mouse model has been generated which replicates the core features of any of these TSC tumors, including angiofibromas, ungual fibromas, shagreen patch, angiomyolipomas, or lymphangioleiomyomatosis. Here we sought to generate such a model by targeting biallelic deletion of Tsc1 to fibroblasts. Fibroblasts or a closely related cell type are thought to be the cell which gives rise to angiofibromas, ungual fibromas, and shagreen patch [2], and might be the cell which gives rise to angiomyolipoma and lymphangioleiomyomatosis.

Mice and drug treatments
All procedures were carried out in accordance with the Guide for the Humane Use and Care of Laboratory Animals, and the study was approved by the Animal Care and Use Committee of Children's Hospital, Boston. The approved protocol contained the following procedures. Euthanasia was performed using CO2 narcosis. Mice were monitored 3 times per week. If any of the following conditions occurred mice were subject to euthanasia: skin irritation or breakdown, tumors > 10% body weight, tumors that limit mobility, lethargy, loss of appetite, weight loss > 15-20%, Rodent Body Condition Score of 1 or 2. Some of the Tsc1 cc Fsp-cre+ mice were subject to euthanasia for these reasons. None died spontaneously. Some mice were treated with Baytril cream for dermatitis to alleviate suffering, but anesthetics, analgesics, or other treatments were not used. Mice were housed under standard conditions meeting the standards of the Guide for the Humane Use and Care of Laboratory Animals, with strict cage crowding limits, a temperature of 70˚F and a 12 hour light/dark cycle.
Tsc1 conditional allele mice, denoted Tsc1 cc were originally derived in this laboratory [9] ( c denotes conditional, floxed allele; w denotes wild type allele). Mice with the Fsp-Cre allele were a generous gift of Gustavo Leone, Ohio State University, and were described in [10]. In these mice, Cre expression is driven by a 3.1-kb gene promoter sequence extending from 1.9 kb upstream of exon 1 to the end of intron 1 of the S100A4 gene (formerly known as Fibroblast specific protein1, or Fsp1). Mice were derived by standard breeding strategies. Tsc1 cc Fsp-cre+ mice were compared to Tsc1 cw Fsp-cre+, Tsc1 cc , and Tsc1 cw mice.
DNA was prepared from mouse toes/tails by standard procedures for genotyping. PCR genotyping for Tsc1 was performed using a three-primer system that allows simultaneous analysis of wildtype (w), conditional (c), and knockout (k) alleles, followed by agarose gel electrophoresis [9]. Primers that amplify a 300 bp portion of the cre recombinase were used to assess the presence of the Fsp-cre allele.
Rapamycin was purchased from LC laboratories (Woburn, MA). A 20 mg/ml stock was made using ethanol, and mixed daily for injection with sterile vehicle (0.25% PEG-200, 0.25% Tween-80). Mice were treated with rapamycin by intraperitoneal (IP) injection at 3 mg/kg three times per week for four weeks. Control mice received the vehicle solution IP on the same schedule.

Pathology studies
Six micrometer sections were prepared from tissues fixed with 10% formalin overnight followed by 70% ethanol and paraffin embedding, and stained with hematoxylin and eosin by standard techniques at the Rodent Pathology Core at Harvard Medical School.
Skin samples were collected from both ventral and dorsal mouse skin. Skin thickness was quantified by a dermatopathologist (SG) who was blinded to the genotype of the mice. Multiple regions across a single long biopsy of skin were examined and measured at three distinct sites, and averaged to determine the skin epidermal and dermal thickness for each mouse. Chloroacetate esterase staining of skin sections was performed to identify mast cells. Sections were deparaffinized with xylene, put through an alcohol series, treated with Napthol AS-D chloroacetate esterase (Sigma) and counterstained with hematoxylin. Mast cells per high powered field (400x) were quantified in a blinded manner by a dermatopathologist (SG).
For immunofluorescence studies, ventral skin samples were flash frozen in OCT, cut in 5 micron sections, blocked in PBS with 10% milk, and incubated with anti-pS6 (S235/S236) (CST, at 1:100 dilution) and anti-Cre (CST, at 1:800 dilution) antibodies. Secondary antibodies were also obtained from CST, and were conjugated with Alexa-488 and Alexa-555 fluorophore (used at 1:500 dilution). Slides were mounted using ProLong Gold Antifade with DAPI (cat# P36941, Thermo). Sections were viewed using a Fluoview FV10i (Olympus) confocal microscope and analyzed using ImageJ (National Institutes of Health) software.

Primary fibroblast cell culture and immunoblotting
Fibroblast cell lines were derived from the skin of mouse pups of age P0-P7 from a breeding of Tsc1 cc Fsp-cre+ x Tsc1 cc mice. After euthanasia, and 70% ethanol sterilization, skin samples (1cm x 1cm) were collected from the underarm or abdominal areas. Skin samples were washed again in 70% ethanol and then washed in PBS. They were cut into small pieces, treated with 0.2% collagenase (Sigma) in serum-free Ham's F-12 media (Invitrogen) at 37˚C for 1hr, with agitation every 20 min. The cells were washed with PBS 3 times, plated in fibroblast culture medium, and maintained in an incubator at 37˚C with 5% CO 2 . Fibroblast culture medium is a 1:1 mix of DMEM and F-12 (Invitrogen) containing 15% FBS (Invitrogen), 1% HEPES buffer solution (GIBCO), 1% nonessential amino acid mixture (GIBCO), 1% L-glutamine (GIBCO), 1% penicillin/streptomycin (Cellgro) and 1% sodium pyruvate (GIBCO). For serum starvation, cells were cultured in the absence of serum for 24 h.

Statistical analysis
All data points are shown in relevant figures as dot plots. Average values are indicated by a horizontal line. P values were calculated using GraphPad Prism v7.0a using the Mann-Whitney test.

Results
Fsp-cre mediated loss of Tsc1 leads to multiple phenotypes including premature mortality To attempt to develop a mouse model of Shagreen patch, cephalic patch, facial angiofibroma, ungual fibroma, renal angiomyolipoma, and/or lymphangioleiomyomatosis, we used an Fspcre transgene [10] to drive loss of Tsc1 in fibroblasts in mice in vivo. Through initial breeding experiments we found that Tsc1 cc Fsp-cre+ mice were generated at Mendelian ratios, displayed no unusual mortality through the age of 6 months, and that both sexes were fertile. However, Tsc1 cc Fsp-cre+ mice began to die at 6 months and had a median survival of 338 days in contrast to Tsc1 cw Fsp-cre+ and Tsc1 cc littermates (Fig 1A, p < 0.0001). Male and female Tsc1 cc Fsp-cre+ mice showed a similar reduction in survival (Fig 1A right). However, female Tsc1 cc Fsp-cre+ mice all died due to the development of chylous ascites, while male Tsc1 cc Fsp-cre+ mice died of sudden death. Tsc1 cc Fsp-cre+ mice (n = 12; 4 males, 8 females; age 9-14 months) displayed multiple findings at necropsy including osteopetrosis (100%), liver hemangioma (67%), hyperplastic vessels (33%) and uterine rhabdomyosarcoma (88% in females). Other findings included stomach adenoma, lung adenoma, lung thrombus, thymic atrophy, ear angioma, and aortic angioma (intraluminal). However, there was no evidence of angiomyolipoma in the kidneys or other organs of these mice. Nor were any pulmonary lesions or cystic lung disease identified.
Hence we focused on the skin manifestations seen in the Tsc1 cc Fsp-cre+ mice.
Tsc1 cc FspCre+ mice have a major skin phenotype A clear and universal phenotype observed in the Tsc1 cc Fsp-cre+ mice was the development of loose, apparently excessive skin on both the dorsal and ventral surfaces (Fig 1B and 1C). This appeared by one month of age, and was a reliable marker of genotype in all mice. Tsc1 cc Fsp-cre + mice also had a blunt nose, reduced eye opening, paw enlargement (Fig 1C), and dermatitis, as reflected in repetitive scratching behavior. Repetitive scratching often led to dermatitis, and a need to sacrifice the mouse on a humane basis. Interestingly, this was seen only in male (19 of 25, 76%) and not in female (0 of 13) mice (p<0.0001, Fisher's exact test). Normal mouse skin consists of a very thin surface epithelium (epidermis), a thicker connective tissue layer (dermis), and a much larger adipose tissue layer (hypodermis) (Fig 2A left). A thin layer of striated muscle separates the skin from other structures. Tsc1 cc Fsp-cre+ mice displayed an approximate two-fold increase in the thickness of the dermis (Fig 2C, p < 0.0001 for ventral dermis, p = 0.0005 for dorsal dermis) coupled with a more modest increase in ventral and dorsal epidermis thickness (Fig 2B, p = 0.0009 for ventral epidermis, p = 0.09 for dorsal epidermis). Many sections also showed a marked reduction in the hypodermis or adipose layer. Inspection of H & E sections suggested that there was an increased number of mast cells in the dorsal dermal layer, and chloroacetate esterase staining was performed to enable enumeration of mast cells (Fig 3A). There was an average 3.6-fold increase in mast cells per HPF in the mutants (Fig 3B) in both dorsal (p = 0.0038) and ventral (p = 0.036) skin. This was not due to repetitive scratching behavior, since that occurred only on the dorsal skin.
Given these abnormal features of the skin in the Tsc1 cc Fsp-cre+ mice, we hypothesized that there would be a concordant increase in blood vessel density, as seen in both facial angiofibromas and angiomyolipomas in TSC. However, using endomucin immunohistochemistry to identify blood vessels, we found that blood vessel density was not significantly different between normal and mutant skin on either dorsal or ventral sites, though there was a trend toward more blood vessels in the mutant mice ( Fig 3C).
As noted above, the hypodermis layer was markedly thinned in the Tsc1 cc Fsp-cre+ mice (Fig 2A). Fat necrosis was seen consistently in the mutant mice in this region, and was likely a major contributor to hypodermal atrophy (Fig 3A bottom row).

Analysis of recombination and mTORC1 activation in Tsc1 cc Fsp-cre+ mice skin
The Fsp-cre transgene has been reported to lead to high level recombination in mesenchymal lineage cells including fibroblasts [10]. To define the extent of cre expression and recombination, we performed confocal microscopy of fresh-frozen OCT-embedded skin sections of the  mice, using anti-Cre and anti-pS6(S235/S236) antibodies. Although there was considerable background staining by the anti-Cre antibody, a population of Cre+ pS6+ fibroblast cells were clearly identified in the Dermis layer of skin of the Tsc1 cc Fsp-cre+ mice, and not seen in control skin sections (Fig 4). This indicates that dermal fibroblasts are showing cre expression and activation of mTORC1, as expected. Rapamycin treatment reduces dermal thickness and mast cell numbers but does not rescue skin abnormalities to a major extent Rapamycin is an allosteric inhibitor of mTORC1 that markedly reduces its kinase activity for many substrates, particularly S6Kinase. It has been shown to be remarkably effective in the treatment of a variety of TSC mouse tumor and brain models, and for a number of TSC-related tumors [11][12][13][14]. To examine the potential benefit of rapamycin for this model, we performed a pilot experiment, treating Tsc1 cc Fsp-cre+ mice with either vehicle (n = 2) or rapamycin (n = 2) at 3mg/kg given by intraperitoneal injection 3 days per week for 4 weeks, beginning at age 3 months. Analysis of this pilot data indicated that there was a trend toward reduction of ventral epidermal thickness in treated mice, and toward reduction in mast cell count in dorsal sections (Fig 5). However, other measures appeared to be unaffected by treatment in this limited number of samples.

Tsc1 cc Fsp-cre+ skin fibroblasts show Tsc1 loss and mTORC1 activation
To confirm that recombination and loss of Tsc1 expression was occurring in skin fibroblasts of these mice, we prepared primary skin fibroblast cultures from Tsc1 cc Fsp-cre+ and control Tsc1 cc mice. Genotyping confirmed the presence of the Fsp-cre allele in these cultures, and showed that there was near complete conversion of the conditional allele to the null allele ( Fig  6A). The Tsc1 cc Fsp-cre+ fibroblast cultures showed marked reduction of Tsc1 and Tsc2 expression, as expected. Tsc1 cc Fsp-cre+ fibroblasts also showed constitutive activation of mTORC1 during serum deprivation with high levels of pS6(S240/244) and some increase in pS6K(T389), and a dramatic reduction in pAKT(S473) levels in response to serum stimulation, consistent with previous data on signaling effects of mTORC1 activation in cells lacking Tsc1 (Fig 6B) [15][16][17].

Discussion
Consistent with multiple previous reports [10,18,19], the Fsp-cre allele used here led to targeting of Cre recombinase expression to fibroblasts leading to recombination and loss of Tsc1 expression in the skin fibroblasts of the Tsc1 cc Fsp-cre+ mice (Figs 5 and 6), and likely other tissues. The mice displayed a modest phenotype for several months until a variety of tumors developed in the male mice, and chylous ascites developed in female mice, leading to a median survival of a little less than one year. The major phenotype that was seen in these mice was in the development of the skin disorder described. All organs of these mice were examined pathologically, with special emphasis on the kidneys and lungs in a cohort of Tsc1 cc Fsp-cre+ mice at ages 9-12 months. No evidence of kidney angiomyolipoma or lymphangioleiomyomatosis was seen in any mouse. These observations strongly suggest that stromal fibroblasts are not the cell of origin of these two tumors, which are known to be closely related in terms of their genetics and pathologic features [4,5,[20][21][22]. Expression of melanocyte proteins, VEGFD, and Cathepsin K by these tumors suggests the possibility that the cell of origin is derived from the neural crest [23][24][25][26][27]. The lack of angiomyolipoma and lymphangioleiomyomatosis development in these mice are consistent with that hypothesis, as well as other possibilities.
The skin abnormalities seen in all Tsc1 cc Fsp-cre+ mice have are not a close match to the angiofibromas and ungual fibromas seen in TSC patients. Tsc1 cc Fsp-cre+ mice were shown to have thickening of the dermis with increased numbers of fibroblasts. However, there is not a prominent vascular component nor a marked increase in mast cells reported in human TSC patient lesions [28]. Furthermore, there was a minimal response to rapamycin treatment in contrast to what has been reported for TSC patients using topical rapamycin for skin treatment [29]. This may be due in part to the differences in human vs. mouse skin components, especially the abundant hair follicles present in mouse skin. In addition, however, although it can be predicted that dermal fibroblasts will undergo spontaneous loss of the second allele of TSC1/TSC2 at multiple sites, the distribution of angiofibromas in TSC subjects is highly circumscribed, suggesting that other factors influence the development of these lesions beyond loss of TSC1 or TSC2. Further investigation is required.