Patched1 haploinsufficiency severely impacts intermediary metabolism in the skin of Ptch1+/−/ODC transgenic mice

The study of dominantly heritable cancers has provided insights about tumor development. Gorlin syndrome (GS) is an autosomal dominant disorder wherein affected individuals develop multiple basal cell carcinomas (BCCs) of the skin. We developed a murine model of Ptch1 haploinsufficiency on an ornithine decarboxylase (ODC) transgenic background (Ptch1+/−/ODCt/C57BL/6) that is more sensitive to BCCs growth as compared with Ptch1+/+/ODCt/C57BL/6 littermates. Ptch1+/−/ODCt/C57BL/6 mice show an altered metabolic landscape in the phenotypically normal skin, including restricted glucose availability, restricted ribose/deoxyribose flow and NADPH production, an accumulation of α-ketoglutarate, aconitate, and citrate that is associated with reversal of the tricarboxylic acid cycle, coupled with increased ketogenic/lipogenic activity via acetyl-CoA, 3-hydroybutyrate, and cholesterol metabolites. Also apparent was an increased content/acetylation of amino-acids, glutamine and glutamate, in particular. Accordingly, metabolic alterations due to a single copy loss of Ptch1 in Ptch1+/−/ODCt/C57BL/6 heterozygous mice may provide insights about the cancer prone phenotype of BCCs in GS patients, including biomarkers/targets for early intervention.


Results
Ptch1 heterozygosity alters the metabolic landscape in phenotypically normal skin of Ptch1 +/− /ODC t /C57BL/6 mice. Metabolites identified by the "metabolon platform TM " in the phenotypically normal skin of Ptch1 +/+ /ODC t /C57BL/6 and Ptch1 +/− /ODC t /C57BL/6 mice comprised a total of 859 biochemicals, 727 of which were of known identity (designated as "named" biochemicals) and 132 compounds were of unknown structural/functional identity (designated as "unnamed" biochemicals). Out of the 859 biochemicals, expression of 249 was increased while the expression of 267 was decreased significantly (p < 0.05). Out of the 132 unidentified metabolites, the expression of 37 was increased and expression of 24 was decreased significantly (p < 0.05) (Fig. 1A,B). Based on the metabolites' pathway classification network by the "metabolon platform TM ", identified biochemicals were grouped into 8 recognized metabolic pathways (Supplementary Table 1). A principle component analysis (PCA) showed that the phenotypically normal skin specimens from Ptch1 +/− /ODC t / C57BL/6 mice were clearly distinguished from their Ptch1 +/+ /ODC t /C57BL/6 littermates, indicating that introduction of Ptch1 heterozygosity significantly alters the skin metabolome of these mice (Fig. 1C). A PCA that did not include "unnamed" metabolites has shown essentially a similar pattern (not shown). The metabolic profile of the major pathways was as follows: In the carbohydrate metabolic pathway, 7 biochemicals were found to be significantly elevated in Ptch1 +/− / ODC t /C57BL/6 mice, including lactose, raffinose, sucrose, ribitol, galactitol, and fucose (Supplementary Table 1). Yet, the top 10 metabolites that were found to be significantly decreased were 2-phosphglycerate, 6-phosphogluconate, N-acetylglucosamine 6-phosphate, galactonate, lactate, phosphoenolpyruvate (PEP), 3-phosphoglycerate, fructose 1,6-diphosphate/glucose 1,6-diphosphate/myo-inositol diphosphates, sedoheptulose, and glucuronate (Supplementary Table 1). The majority of these biochemicals represent key intermediates of the glycolytic and pentose phosphate pathways (PPP), suggesting a major reduction in cellular functions that regulate glucose utilization through glycolysis, including reduced ribose/deoxyribose for nucleic acids biosynthesis, as well as NADPH levels that is largely produced through the PPP. NADPH is a major contributor for fatty acid bio-synthesis 19 . Significantly, tightly linked downstream metabolites which comprise the "entry/exist" portion of the TCA cycle, including citrate, aconitate and α-ketoglutarate were considerably elevated ( Fig. 2A), while other downstream TCA cycle intermediates were significantly reduced.
The Online Metabolon software TM Pathway Set Enrichment Analysis (PSEA), showed that a majority of significantly modulated pathways comprised carbohydrate/TCA cycle, lipids, amino acids, and nucleotide metabolism ( Fig. 1D & Supplementary Table 2).
Ptch1 heterozygosity alters cutaneous energy metabolism. Increased glucose utilization via glycolysis is prevalent in cancer cells 16 . Here, however, we found that Ptch1 heterozygosity in Ptch1 +/− /ODC t / C57BL/6 mice suppressed glucose utilization as is evidenced by a significant decrease in multiple intermediates of the glycolytic cascade ( Fig. 2A,B), including a profound decrease of more than 50% in the levels of fructose 1,6-diphosphate, dihydroxyacetone phosphate (DHAP), 3-phosphoglycerate, 2-phosphoglycerate, phosphoenolpyruvate, and lactate as compared to their Ptch1 +/+ /ODC t /C57BL/6 littermates ( Fig. 2A). As for the TCA cycle, while citrate, cis-aconitate, and α-ketoglutarate were increased significantly ( Fig. 2A), there was a significant decrease in the level of fumarate (48%) and malate (40%). This metabolic pattern is remarkable, suggesting reversal of the TCA cycle due to reductive carboxylation of α-ketoglutarate via glutamate 13,14 , leading back through citrate to increased acetyl-CoA by the citrate cleavage enzyme (ATP citrate lyase) 23 (Fig. 2A). Thus, Ptch1 heterozygosity alters skin energy metabolism, potentially leading to a metabolic state that appears to prefer production of acetyl-CoA from TCA cycle reversal, including presumably a limited production of acetyl-CoA through fatty acid β-oxidation 24 . The latter is consistent with a significant increase in 3-hydroxybutyrate (Fig. 2B), supporting a switch towards ketone body-dependent energy metabolism. It is also of interest that 2-methylcitrate was significantly enhanced. This metabolite was shown to alter mitochondrial membrane permeability pore transition 25 .
Ptch1 heterozygosity suppressed cutaneous pentose phosphate pathway (PPP) in Ptch1 +/− / oDc t /C57BL/6 mice. Since we observed decreased glycolysis in Ptch1 heterozygous mice, such as reduction of about 40% in glucose 6-phosphate, we further assessed the impact of Ptch1 gene dose on the PPP. This pathway involves a irreversible conversion of glucose 6-phosphate to ribulose 5-phosphate, wherein the latter was found to be significantly decreased. Moreover, all other intermediates of this pathway including 6-phosphogluconate, ribulose 5-phosphate, xylulose 5-phosphate, and ribose 5-phosphate were significantly reduced. Among these, 6-phosphogluconate was decreased by 72% ( Fig. 2A). Obviously, a severe reduction of the PPP would, in turn, affect ribose/deoxyribose levels, and thereby nucleosides/nucleotides for energy production and DNA synthesis. Furthermore, the PPP also serves as a major resource for the production of NADPH that is largely used for de-novo fatty acids synthesis.
Ptch1 heterozygosity alters cutaneous lipid metabolism in Ptch1 +/− /oDc t /C57BL/6 mice. In the lipidomics profiling we employed a "complex lipids platform" in order to gain further insight into the effects of Ptch1 heterozygosity on cutaneous lipid metabolism. As illustrated in Fig. 3A, out of 948 identified biochemical lipid metabolites, only 25 were found to be increased, 342 decreased and the remaining 581 showed no significant changes (Supplementary Table 3). A PCA of the complex lipids data suggest intra-group variations among the Ptch1 +/− /ODC t /C57BL6 specimens, which, however, were separated from of Ptch1 +/+ /ODC t /C57BL/6 littermates specimens, falling into two distinct clusters (Fig. 3B). The decreased lipids consisted primarily of fatty acids, phospholipids as well as diacylglycerols (DAG) and monoacylglycerols (MAG), with no significant changes in sphingolipids (Fig. 3C & Supplementary Table 4). Yet, phosphatidylcholines (PC), phosphatidylinositols (PI) and lysophophatidylethanolamines (LPE) represented the three major groups showing significant reduction in this category of phospholipid (Fig. 3C & Supplementary Table 4). The decrease of complex phospholipids may be due to a severe reduction of 3-carbon intermediates via the glycolytic pathway 26 (Fig. 2A).
Further analysis showed that primary and secondary bile acid metabolites appear to be more significantly upregulated; between 3.3 to 130.9 -fold (Fig. 4E,F). The most upregulated metabolite in bile acid metabolism was taurodeoxycholate of 130-fold, followed by tauro-beta-muricholate of 42.0-fold (Fig. 4E,F & Supplementary Table 1). Taurodeoxycholate has been demonstrated to increase proliferation and to inhibit apoptotic cell death through activation of NF-kB 31 . Other biological active bile acid metabolites were also highly upregulated, including taurocholate (23.3-fold), taurochenodeoxycholate (19.8-fold), tauroursodeoxycholate (9.5-fold) and taurohyodeoxycholate (8.5-fold) (Fig. 4E,F & Supplementary Table 1). It is important to note that bile acids are physiological ligands of farnesoid X receptor (FXR). FXR regulates the expression of genes controlling cholesterol homeostasis, including other lipids, and glucose metabolism which were shown to positively affect tumorigenesis 32,33 . Incidently, heteromeric complexes of FXR with RXR play an important role in retinoic acid metabolism, including its clinical application to control inflammation and certain cancers 34,35 . Cutaneous alterations of nucleic acid metabolism by Ptch1 heterozygosity in Ptch1 +/− /oDc t / C57BL/6 mice. ODC provides a pool of polyamines which are precursors of purines and pyrimidines 36 .
Pyrimidine metabolism, on the other hand, was significantly augmented. The precursor metabolites of the pyrimidine de novo synthetic pathway, including N-carbamoylaspartate, dihydroorotate, and orotate were increased significantly by 4.9, 3.4 and 3.4-fold respectively ( Table 1). The levels of uridine 5′-monophosphate (UMP), which is converted to cytidine or thymidine containing nucleotides for DNA synthesis, did not change appreciably, yet both cyclic UMP (cUMP) and cyclic CMP (cCMP) were increased significantly. The  www.nature.com/scientificreports www.nature.com/scientificreports/ latter two metabolites have recently been demonstrated as second messenger and they play essential role in cell proliferation 37 . In contrast, deoxynucleic acid thymidine 5′-monophosphate (TMP) and 2′-deoxycytidine 5′-monophosphate (dCMP) were decreased substantially by 41% and 56% respectively, consistent with increased accumulation of related catabolites including cytosine, dihydrouracil and dihydrothymine (2.7, 3.0 and 7.9-fold respectively) ( Table 1).

Discussion
Identification of TSGs and their effector pathways in several autosomal dominant hereditary cancer syndromes has provided insights about the underlying mechanisms of cancer initiation and potential therapeutic intervention [38][39][40] . GS is a dominantly heritable cancer syndrome in which BCCs arise through a two-hit mechanism wherein "one hit" is an inherited, inactivating mutation in PTCH1, and the second hit is a somatically derived mutation in the remaining PTCH1 allele 41 . In sporadic BCCs, the majority of the cases harbor two somatic mutations in PTCH1, or less often, an activating mutation in SMO mimicking loss of PTCH1 10,41-45 , although additional molecular changes are probably necessary for BCCs development.
In order to model the state of PTCH1 haploinsufficiency in GS patients, the present study focused on the global expression profile of metabolites in the phenotypically normal skin of Ptch1 +/− /ODC t /C57BL/6 mice and their Ptch1 +/+ /ODC t /C57BL/6 littermates, leading to the following important findings.
First, we found a sharp decrease of key glycolytic intermediates in the skin of Ptch1 +/− /ODC t /C57BL/6 mice, severely curtailing glucose flow/utilization and ATP production therefrom.
Second, we found a sharp decrease of key intermediates of the PPP in the skin of Ptch1 +/− /ODC t /C57BL/6 mice, affecting the flow of ribose/deoxyribose for nucleosides/nucleotides synthesis, as well as production of reduced NADP that is a major resource for de-novo fatty acids synthesis.
Third, we identified a metabolic sequence that is associated with reversal of the TCA cycle through reductive carboxylation of α-ketoglutarate via glutamate, leading to an accumulation of citrate, and thereby an abundance of acetyl CoA in the skin of Ptch1 +/− /ODC t /C57BL/6 mice. Notably, reversal of the TCA cycle would largely curtail ATP production, as well.
Fourth, in support of these results we found a significant increase in the level of 3-hydroxybutyric acid, a ketone body, which is formed from acetyl-CoA through acetoacetate, and is used as an energy source when glucose supply is low or absent and when reversal of the TCA cycle occurs. Apparently, increased levels of 3-hydroxybutyrate as a major source of energy is associated with an activated HH pathway in the skin of Ptch1 +/− / ODC t /C57BL/6 mice. Incidentally, 3-hydroxybutyric acid levels were also found to promote tumorigenesis by modulating H3K9 acetylation 46 .
Fifth, we found a significant increase of cholesterol metabolites, including taurine esters of bile acids. This reflect, in part, the utility of acetyl CoA under conditions of low glucose where, in addition to 3-hdroxybutyrate for energy, the cholesterol biosynthetic pathway is selectively activated in the skin of Ptch1 +/− /ODC t /C57BL/6 mice. The fact that other lipid forms were not affected apparently due to severe reduction of glycolytic intermediates, specifically glycerol containing moieties, which are the prime source for the synthesis of complex phospholipids. Bile acids are known modulators of FXR signaling and they could play an important role in maintaining the high propensity for BCCs induction that is associated with GS 47 .
Sixth, the changes we observed in amino acids metabolites may reflect the increased acetylation pool that would affect the regulatory potential of proteins as well as the role of acetylated individual amino acids, including their metabolic interconversion pathways 48 during the early development of BCCs. Of particular interest is the apparent abundance of glutamine/glutamate (both acetylated and unacetylated) as compared with α-ketoglutarate levels (about 16 fold) which is a key causal factor to affect reversal of the TCA cycle in the skin of Ptch1 +/− /ODC t /C57BL/6 mice. Another important consideration is based on evidence that reversal of the TCA cycle via glutamate dehydrogenase (GDH) is specifically associated with a low cellular energy charge, i.e., decreased ATP levels 49 . Indeed, the general pattern that emerges in the present study regarding alterations that involve the purine/pyrimidine pools is one of reduced synthesis of di-tri-phosphonucleotides.
In summary, our experimental model (Fig. 5) specifies that metabolic profiling of the skin in Ptch1 +/− /ODC t / C57BL6 mice during the initiation phase of BCCs development 38 can identify novel molecular biomarkers/ targets for chemoprevention of BCCs risk. Most importantly is apparently the glutamine pathway 50 that under www.nature.com/scientificreports www.nature.com/scientificreports/ conditions of glucose shortage plays a key role in the overall metabolic pattern that emerges herein, including reduced glucose utilization, coupled with reversal of the TCA cycle, limited cellular energy charge, as well as increased reliance on a specific subset of lipid metabolism, i.e., 3-hydroxybutyrate and cholesterol metabolites, including bile acids. Specifically, therefore, relevant biomarkers/targets would presumably consist of the urea/ ammonia cycle vis. glutamine synthesis 50 , followed by glutamine/glutamate contribution to reductive carboxylation of a-ketoglutarate 14 , citrate metabolism vis. ATP citrate lyase (citrate cleavage enzyme) 23 , and HMG-CoA reductase that is integral to the synthesis of cholesterol metabolites 7,8 . Whether some or all of these sites might be due to Ptch1 haploinsufficiency or an activating mutation in SMO, remains to be established. We have previously shown that cyclopamine, and perhaps other SMO antagonists, are potent in vivo inhibitors of UVB-induced BCC in Ptch1 +/− mice and most likely in humans 51 .

Materials and Methods
Animals. We developed Ptch1 +/− /ODC t /C57BL/6 murine model in which Ptch1 mutations drive aberrant activation of Sonic hedgehog (SHH) signaling 11 . These mice were generated by deletion of exons 1 and 2 of the Ptch1 gene. ODC transgene overexpression in the skin of these mice was driven by a K6 promoter near the hair follicles where basal skin cells presumably reside 52 . The male ODC t breeders (6-7 weeks old) were purchased from Taconic (Germantown, NY, USA). The detailed breeding protocols and genotyping of Ptch1 +/− /ODC t /C57BL/6 animals are described earlier 11 . Of note, none of the animals, whether Ptch1 wild type or Ptch1 mutated, studied herein were exposed to UVB or other pro-carcinogenic environmental stimuli. Phenotypically normal skin is defined here as untreated skin of animals with no visible neoplastic lesions. Hence, no difference in skin phenotype was observed between Ptch1 +/− /ODC t /C57BL/6 and Ptch1 +/+ /ODC t /C57BL/6 mice. All animal care and experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Alabama at Birmingham and Columbia University. All experiments were carried out in accordance with relevant guidelines and regulations.
Skin tissue samples preparation. After the termination of the experiment, skin tissues from each animal were harvested, snap frozen and stored in −80 °C freezer. At least 100 mg of skin tissue from each animal was used for the metabolon and lipid analysis. At least eight skin samples from eight animals were analyzed in each group.
Metabolon platform. Sample preparation, data collection and data analysis were performed by standard protocol of Metabolon (Morrisville, NC, USA). Briefly, samples were prepared using the automated MicroLab STAR ® system from Hamilton Company. Proteins were precipitated with methanol under vigorous shaking for 2 min (Glen Mills GenoGrinder 2000) followed by centrifugation. The resulting extract was divided into five fractions: two for analysis by two separate reverse phase (RP)/UPLC-MS/MS methods with positive ion mode electrospray ionization (ESI), one for analysis by RP/UPLC-MS/MS with negative ion mode ESI, one for analysis by HILIC/UPLC-MS/MS with negative ion mode ESI, and one sample was reserved for backup. Samples were placed briefly on a TurboVap ® (Zymark) to remove the organic solvent. The sample extracts were stored overnight under nitrogen before preparation for analysis. Several types of controls and standard were analyzed in concert with the experimental samples to ensure the data quality. The sample extracts were analyzed by ultrahigh performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS) followed by data extraction and compound identification, curation, metabolite quantification and data normalization. Detailed methods can be found in Supplementary Materials 53 .
Complex lipid platform. Lipids were extracted from samples in methanol:dichloromethane in the presence of internal standards. The extracts were concentrated under nitrogen and reconstituted in 0.25 mL of 10 mM ammonium acetate dichloromethane:methanol (50:50). The extracts were transferred to inserts and placed in vials for infusion-MS analysis, performed on a Shimazdu LC with nano PEEK tubing and the Sciex SelexIon-5500 QTRAP. The samples were analyzed via both positive and negative mode electrospray. The 5500 QTRAP scan was performed in MRM mode with the total of more than 1,100 MRMs. Individual lipid species were quantified by taking the peak area ratios of target compounds and their assigned internal standards, then multiplying by the concentration of internal standard added to the sample. Lipid class concentrations were calculated from the sum of all molecular species within a class, and fatty acid compositions were determined by calculating the proportion of each class comprised by individual fatty acids 54 .

Pathway set enrichment and principle component analyses. Pathway set enrichment analysis
(PSEA) was performed by Metabolync portal: https://retiredportal.metabolon.com. The pathway enrichment value is calculated by comparing the ratio of significantly changed compounds in a particular pathway to the ratio of significantly altered compounds relative to all named compounds in the study.
PCA is an unsupervised analysis that reduces the dimension of the data, and where the total variance is defined as the sum of the variances of the predicted values of each component and for each component, the proportion of the total variance was computed by Metabolon (Morrisville, NC, USA).
We should note that the Metabolon platform used herein cannot detect "high-energy" compounds, nor can it detect volatile metabolites, some of which are considered in the context of this study; for example, acetyl-CoA, di-and tri-phosphonucleotides, acetate, and oxaloacetate. However, these are being inferred based on the current understanding of metabolomics.
Statistical analysis. Student's t-test was used to compare the mean of the two groups. In all calculations, p < 0.05 was considered statistically significant.