Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

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

An Author Correction to this article was published on 11 October 2021

This article has been updated

Abstract

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.

Introduction

Basal cell carcinomas (BCCs) of the skin represent nearly one-half of all cancers diagnosed in the United States1,2. Although majority of these tumors do not metastasize and death caused by these tumors is uncommon3, they are frequently associated with clinically significant morbidity caused by local invasion and tissue destruction. The basal cell nevus syndrome (BCNS), also known as Gorlin syndrome (GS) or the nevoid basal cell carcinoma syndrome (NBCCS), is an autosomal dominant disorder characterized by early onset of multiple BCCs, occasional childhood malignancies, and developmental defects4,5. The majority of GS cases are driven by mutations in the tumor suppressor gene (TSG) Patched (Ptch) and/or the G-protein coupled receptor smoothened (SMO)6, leading to activation of Gli transcription factors (Gli1, 2, 3), and thereby uncontrolled proliferation of basal cells and ultimately BCCs growth in the skin of affected individuals7,8,9,10.

In order to further define the pathogenesis of BCCs growth, we developed Ptch1+/−/ODCt/C57BL/6 transgenic mice wherein a single copy loss of Ptch1 causes tumor initiation, and whereupon ornithine decarboxylase (ODC) transgene triggers tumor promotion11,12. Importantly, we have demonstrated that the histologic features of BCCs in Ptch1+/−/ODCt/C57BL/6 mice are similar to those in GS patients, including palisading at the edge of tumor nests, retraction from surrounding stroma, the lack of keratin pearl, and the characteristic expression pattern of marker biomolecules11.

Here, we show that Ptch1 haploinsufficiency affects the metabolic landscape in phenotypically normal skin of Ptch1+/−/ODCt/C57BL/6 mice, wherein loss of a single copy of Ptch1 in these mice caused a marked reduction of glucose utilization, coupled with reversal of the tricarboxylic acid (TCA) cycle through reductive carboxylation of α-ketoglutarate via glutamate13,14. This metabolic shift was associated with an accelerated metabolism of ketogenic/lipid metabolites through 3-hydroxybutyrate, and cholesterol derivatives. Incidentally, high lipid biosynthesis, in particular, was found to be associated with enhanced susceptibility to BCCs in GS patients15. Importantly, increased acetylation of specific amino acids/proteins, specifically increased content of glutamate/glutamine, and reduced cellular energy charge were demonstrated herein. Thus, targeting metabolites in the skin of Ptch1+/−/ODCt/C57BL/6 mice may enable potential interventional approaches of BCCs development16,17,18.

Results

Ptch1 heterozygosity alters the metabolic landscape in phenotypically normal skin of Ptch1 +/−/ODC t/C57BL/6 mice

Metabolites identified by the “metabolon platformTM” in the phenotypically normal skin of Ptch1+/+/ODCt/C57BL/6 and Ptch1+/−/ODCt/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 platformTM”, 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+/−/ODCt/C57BL/6 mice were clearly distinguished from their Ptch1+/+/ODCt/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:

Figure 1
figure 1

Effects of Ptch1+/− heterozygosity on cutaneous metabolic landscape. (A) Clustering analysis of the abundance of named metabolites in the skin of Ptch1+/+/ODCt/C57BL/6 and Ptch1+/−/ODCt/C57BL/6 mice. Metabolites that were not significantly changed were not included in this analysis. Each row represents a single metabolite while each column represents an independent skin sample. The color scale indicates relative level of the abundance of each metabolite with red indicating levels greater than mean and blue indicating levels below the mean. (B) Pie chart showing numbers of biochemicals, which are increased or decreased in the skin; (C) Graph showing principle component analysis of global metabolomics data. (D) Graph showing the data of pathway set enrichment analysis (PSEA). Pathways containing greater than three identified metabolites are included in the analysis. The intensity of red color is associated with the PSEA score of each metabolite. PSEA score greater than one suggests the indicated pathway have above average number of metabolites that are significantly altered.

In the carbohydrate metabolic pathway, 7 biochemicals were found to be significantly elevated in Ptch1+/−/ODCt/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-synthesis19. 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.

Figure 2
figure 2

Effects of Ptch1+/− heterozygosity on cutaneous glucose metabolism and pentose phosphate pathway. (A) Graphs showing the levels of various metabolites in glycolysis/TCA cycle/PPP; (B) Graph showing impact on ketone body, 3-hydroxybutyrate level. *p < 0.05; **p < 0.01; ***p < 0.001.

The 10 most elevated biochemicals in the lipid class were associated with bile acids and steroid metabolism, including taurodeoxycholate, tauro-beta-muricholate, taurocholate, taurochenodeoxycholate, tauroursodeoxycholate, taurohyodeoxycholic acid, chenodeoxycholate, 18-methylnonadecanoate (i20:0), corticosterone and dicarboxylate fatty acid sebacate which levels were increased between 3.9 to 130.9 -fold in the skin of Ptch1+/−/ODCt/C57BL/6 mice, as compared to Ptch1+/+/ODCt/C57BL/6 mice (Supplementary Table 1). The 10 most diminished biochemicals in the lipid class were linolenoylcarnitine (C18:3), 1-palmitoyl-2-oleoyl-GPG (16:0/18:1), decanoylcarnitine (C10), octanoylcarnitine (C8), glycerophosphoethanolamine, (S)-3-hydroxybutyrylcarnitine, (R)-3-hydroxybutyrylcarnitine, mead acid (20:3n9), N-palmitoleoyltaurine, and eicosapentaenoate (EPA; 20:5n3) which levels were decreased between 0.15 to 0.36- fold relative to their levels in Ptch1+/+/ODCt/C57BL/6 mice (Supplementary Table 1). Interestingly, the 10 most diminished lipid metabolites were essentially all carnitine derivatives that are associated with fatty acids transport across the mitochondrial membrane for energy utilization, including phospholipids20. The decrease in fatty acids and complex phospholipids is presumably due to the sharp decrease of 3-carbon intermediates (glycerol based) in the glycolytic pathway that, in general, provide the backbone for the synthesis of these lipids. These results reinforce the selective utility of acetyl CoA in the augmentation of 3-hydroxybutyrate and cholesterol metabolites, but not fatty acids or complex phospholipids.

Changes in the amino acids metabolic pathways included the top 10 increased metabolites, comprising N2,N6-diacetyllysine, phenol glucuronide, N-acetylcitrulline, N-acetylglutamine, formiminoglutamate, phenol sulfate, p-cresol glucuronide, N2, N5-diacetylornithine, 4-hydroxyphenylacetate sulfate, and N-acetylglutamate which were increased between 6.5 to 34.2 fold in Ptch1+/−/ODCt/C57BL/6 mice as compared to their levels in Ptch1+/+/ODCt/C57BL/6 littermates (Supplementary Table 1). The amino acid, N2, N6-diacetyllysine was increased 34.2- fold. In addition, 4 out of these 10 metabolites were enriched in the tyrosine metabolic pathway, including phenol glucuronide, phenol sulfate, p-cresol glucuronide, and 4-hydroxyphenylacetate sulfate. The other 5 metabolites are known to be important intermediates of the urea cycle, including arginine and proline metabolites: N-acetylcitrulline, and N2,N5-diacetylornithine; glutamate metabolites: N-acetylglutamine, N-acetylglutamate, formiminoglutamate, and histidine metabolites (Supplementary Table 1). In this regard, the enrichment of glutamine/glutamate metabolites, in particular, is highly relevant to the reversal of the TCA cycle via reductive carboxylation of α-ketoglutarate13,14. Also intriguing, are the high levels we found of n-acetyl-lysines which might reflect the increased content of the acetyl-CoA pool as well as their potential role as potent epigenetic regulators21.

In the nucleotides metabolism super pathway, the 10 most elevated metabolites (between 2.02 fold to 7.85 fold) were the following purines and pyrimidines 5,6-dihydrothymine, adenosine-2′3′-cyclic monophosphate, N-carbamoylaspartate, guanosine-2′3′-cyclic monophosphate, adenine, orotate, dihydroorotate, 5,6-dihydrouracil, cytosine, and 1-methylhypoxanthine whereas the 10 most decreased metabolites (between 0.13 to 0.55 fold) were N6-succinyladenosine, uric acid ribonucleoside, N1-methyladenosine, thymidine 3′-monophosphate, 2′-deoxycytidine 5′-monophosphate, N1-emthylguanosine, 2′-deoxyadenosine 5′-monophosphate, 2′deoxyadenosine, orotidine, 2′-deoxyguanosine 5′-monophosphate (dGMP) (Supplementary Table 1). It is interesting to note that orotate has been shown as a tumor promoter in hepatocarcinogenesis22.

In the vitamin and cofactors metabolism pathway, 4 biochemicals including nicotinamide riboside, nicotinate, 1-methylnicotinamide, and quinolinate are found to be increased between 2.05 to 5.77 fold relative to Ptch1+/+/ODCt/C57BL/6 littermates (Supplementary Table 1). Interestingly, these are all enriched in nicotinate and nicotinamide metabolism. On the other hand, 7 biochemicals which include dehydroascorbate, pyridoxate, N1-methyl-4-pyridone-3-carboxamide, biliverdin, N1-methyl-2-pyridone-5-carboxamide, oxalate, and pantothenate are decreased between 0.27 to 0.75 fold as compared to Ptch1+/+/ODCt/C57BL/6 littermates (Supplementary Table 1).

The top10 increased xenobiotics were 3-formylindole, formononetin, indolin-2-one, biochanin A, p-cresol sulfate, soyasaponin II, soyasaponin I, indole-3-acetamide, and soyasaponin III which were increased between 4.0 to 22.2 fold as compared to their levels in Ptch1+/+/ODCt/C57BL/6 (Fig. 1A & Supplementary Table 1). The top 10 diminished xenobiotics were 4-ethylphnylsuflate, ferulic acid 4-sulfate, 3-(3-hydroxyphenyl) propionate sulfate, N-acetylpyrraline, 2-hydroxyhippurate, 1,2,3-benzenetriol sulfate (2), catechol sulfate (Supplementary Table 1).

The Online Metabolon softwareTM 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 cells16. Here, however, we found that Ptch1 heterozygosity in Ptch1+/−/ODCt/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+/+/ODCt/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 glutamate13,14, 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-oxidation24. 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 transition25.

Ptch1 heterozygosity suppressed cutaneous pentose phosphate pathway (PPP) in Ptch1 +/−/ODCt/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 +/−/ODCt/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+/−/ODCt/C57BL6 specimens, which, however, were separated from of Ptch1+/+/ODCt/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 pathway26 (Fig. 2A).

Figure 3
figure 3

Effects of Ptch1+/− heterozygosity on lipid profile in the skin. (A) Pie chart depicting the numbers of lipids, which are increased or decreased in the skin; (B) Graph showing principle component analysis of lipidomics data; (C) Graphs showing impact on mono & diacylglycerols, phosphatidylcholines, lysophosphatidylethanolamines, and phosphatidylinositol. *p < 0.05; **p < 0.01; ***p < 0.001.

Ptch1 heterozygosity promotes cutaneous cholesterol metabolism in Ptch1 +/−/ODCt/C57BL/6 mice

Interestingly, the only lipid group which showed higher cutaneous accumulation is cholesteryl esters (CEs) (p = 0.0578) (Supplementary Table 3). This may be an expected finding confirming the role of PTCH1 protein in cholesterol transport8. Cholesterol metabolism is associated with the biosynthesis of corticosteroids, bile acids, vitamin D, and cell membrane components (Fig. 4A)27,28,29. As listed in Fig. 4C,D & Supplementary Table 5, various types of Cholesterol esters (CEs) were significantly upregulated in the phenotypically normal skin of Ptch1+/−/ODCt/C57BL/6 mice. Other important cholesterol metabolites are glucocorticoids (such as cortisol in humans and corticosterone in rodents). A 4.2-fold increase of corticosterone was noted in phenotypically normal skin of Ptch1+/−/ODCt/C57BL/6 as compared to Ptch1+/+/ODCt/C57BL/6 (Fig. 4B), suggesting a potential role in inflammation during the process of BCCs development30.

Figure 4
figure 4

Effects of Ptch1+/− heterozygosity on cholesterol metabolism. (A) Flow diagram of cholesterol dependent metabolic pathways in skin; (BF) Graphs showing the impact of Ptch1+/− heterozygosity on corticosterone (B), cholesteryl esters (C,D), primary bile acid metabolites  (E) and secondary bile acid metabolites (F). *p < 0.05; **p < 0.01; ***p < 0.001.

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-kB31. 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 tumorigenesis32,33. Incidently, heteromeric complexes of FXR with RXR play an important role in retinoic acid metabolism, including its clinical application to control inflammation and certain cancers34,35.

Cutaneous alterations of nucleic acid metabolism by Ptch1 heterozygosity in Ptch1 +/−/ODCt/C57BL/6 mice

ODC provides a pool of polyamines which are precursors of purines and pyrimidines36. Significantly, adenosine 2′,3′-monophosphate (cAMP) and guanosine 2′,3′-monophosphate (cGMP) were consistently elevated by 5.1 and 4.4- fold, respectively. Similarly, the levels of purine containing nucleotide/nucleosides catabolites such as inosine, guanosine and hypoxanthine were higher in Ptch1+/−/ODCt/C57BL/6 than in their control littermates, albeit these did not achieve statistical significance (Table 1). Additionally, guanine and adenine were significantly increased by 1.6 and 3.9- fold, respectively (Table 1). Interestingly, the level of both 2′-deoxy adenosine 5′-monophosphate (dAMP) and 2′-deoxy guanosine 5′-monophosphate (dGMP) were decreased, including adenine and guanine containing nucleosides, suggesting that shortage of deoxyribose due to severe reduction of the PPP (vide supra) might be implicated. These examples include N1-methyladenosine, N6-carbamoylthreonyladenosine, N6-succinyladenosine, and N1-methylguanosine. Urate, the end product of purine catabolism, was not substantially altered, but its metabolite allantoin was found to be increased significantly in the skin of Ptch1+/−/ODCt/C57BL/6 mice (Table 1).

Table 1 Effects of Ptch1 heterozygosity on nucleotide metabolism.

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 latter two metabolites have recently been demonstrated as second messenger and they play essential role in cell proliferation37. 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 intervention38,39,40. 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 allele41. 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 PTCH110,41,42,43,44,45, 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+/−/ODCt/C57BL/6 mice and their Ptch1+/+/ODCt/C57BL/6 littermates, leading to the following important findings.

First, we found a sharp decrease of key glycolytic intermediates in the skin of Ptch1+/−/ODCt/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+/−/ODCt/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+/−/ODCt/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+/−/ODCt/C57BL/6 mice. Incidentally, 3-hydroxybutyric acid levels were also found to promote tumorigenesis by modulating H3K9 acetylation46.

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+/−/ODCt/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 GS47.

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 pathways48 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+/−/ODCt/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 levels49. 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+/−/ODCt/C57BL6 mice during the initiation phase of BCCs development38 can identify novel molecular biomarkers/targets for chemoprevention of BCCs risk. Most importantly is apparently the glutamine pathway50 that under 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 synthesis50, followed by glutamine/glutamate contribution to reductive carboxylation of a-ketoglutarate14, citrate metabolism vis. ATP citrate lyase (citrate cleavage enzyme)23, and HMG-CoA reductase that is integral to the synthesis of cholesterol metabolites7,8. 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 humans51.

Figure 5
figure 5

Metabolic alterations in the skin during the initiation phase of BCCs development in the skin of Ptch1+/−/ODCt/C57BL/6 mice. Detailed description of this model has been included in discussion.

Materials and Methods

Animals

We developed Ptch1+/−/ODCt/C57BL/6 murine model in which Ptch1 mutations drive aberrant activation of Sonic hedgehog (SHH) signaling11. 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 reside52. The male ODCt breeders (6–7 weeks old) were purchased from Taconic (Germantown, NY, USA). The detailed breeding protocols and genotyping of Ptch1+/−/ODCt/C57BL/6 animals are described earlier11. 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+/−/ODCt/C57BL/6 and Ptch1+/+/ODCt/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 Materials53.

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 acids54.

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.

Change history

References

  1. Rogers, H. W., Weinstock, M. A., Feldman, S. R. & Coldiron, B. M. Incidence Estimate of Nonmelanoma Skin Cancer (Keratinocyte Carcinomas) in the U.S. Population, 2012. JAMA Dermatol 151, 1081–1086, https://doi.org/10.1001/jamadermatol.2015.1187 (2015).

    Article  PubMed  Google Scholar 

  2. Siegel, R. L., Miller, K. D. & Jemal, A. Cancer statistics, 2018. CA Cancer J Clin 68, 7–30, https://doi.org/10.3322/caac.21442 (2018).

    Article  PubMed  Google Scholar 

  3. Rees, J. R. et al. Survival after squamous cell and basal cell carcinoma of the skin: A retrospective cohort analysis. Int J Cancer 137, 878–884, https://doi.org/10.1002/ijc.29436 (2015).

    ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

  4. Gorlin, R. J. Nevoid basal-cell carcinoma syndrome. Medicine (Baltimore) 66, 98–113 (1987).

    CAS  Article  Google Scholar 

  5. Lo Muzio, L. Nevoid basal cell carcinoma syndrome (Gorlin syndrome). Orphanet J Rare Dis 3, 32, https://doi.org/10.1186/1750-1172-3-32 (2008).

    Article  PubMed  PubMed Central  Google Scholar 

  6. Pasca di Magliano, M. & Hebrok, M. Hedgehog signalling in cancer formation and maintenance. Nat Rev Cancer 3, 903–911, https://doi.org/10.1038/nrc1229 (2003).

    Article  PubMed  Google Scholar 

  7. Huang, P. et al. Cellular Cholesterol Directly Activates Smoothened in Hedgehog Signaling. Cell 166, 1176–1187 e1114, https://doi.org/10.1016/j.cell.2016.08.003 (2016).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  8. Zhang, Y. et al. Structural Basis for Cholesterol Transport-like Activity of the Hedgehog Receptor Patched. Cell 175, 1352–1364 e1314, https://doi.org/10.1016/j.cell.2018.10.026 (2018).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  9. Gong, X. et al. Structural basis for the recognition of Sonic Hedgehog by human Patched1. Science 361, https://doi.org/10.1126/science.aas8935 (2018).

    Article  Google Scholar 

  10. Epstein, E. H. Basal cell carcinomas: attack of the hedgehog. Nat Rev Cancer 8, 743–754, https://doi.org/10.1038/nrc2503 (2008).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  11. Tang, X. et al. Ornithine decarboxylase is a target for chemoprevention of basal and squamous cell carcinomas in Ptch1+/− mice. J Clin Invest 113, 867–875, https://doi.org/10.1172/JCI20732 (2004).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  12. Witherspoon, M., Chen, Q., Kopelovich, L., Gross, S. S. & Lipkin, S. M. Unbiased metabolite profiling indicates that a diminished thymidine pool is the underlying mechanism of colon cancer chemoprevention by alpha-difluoromethylornithine. Cancer Discov 3, 1072–1081, https://doi.org/10.1158/2159-8290.CD-12-0305 (2013).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  13. Madsen, J., Abraham, S. & Chaikoff, I. L. Conversion of Glutamate Carbon to Fatty Acid Carbon Via Citrate in Rat Epididymal Fat Pads. J Lipid Res 5, 548–553 (1964).

    CAS  Article  Google Scholar 

  14. Kopelovich, L. Pathways of fatty acid biosynthesis: conversion of glutamate carbon to fatty acid carbon via citrate by lactating mouse mammary gland and adenocarcinoma of mouse mammary gland. Biochim Biophys Acta 210, 241–249 (1970).

    CAS  Article  Google Scholar 

  15. Teperino, R. et al. Hedgehog partial agonism drives Warburg-like metabolism in muscle and brown fat. Cell 151, 414–426, https://doi.org/10.1016/j.cell.2012.09.021 (2012).

    CAS  Article  PubMed  Google Scholar 

  16. Boroughs, L. K. & DeBerardinis, R. J. Metabolic pathways promoting cancer cell survival and growth. Nat Cell Biol 17, 351–359, https://doi.org/10.1038/ncb3124 (2015).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  17. Qiu, B. & Simon, M. C. Oncogenes strike a balance between cellular growth and homeostasis. Semin Cell Dev Biol 43, 3–10, https://doi.org/10.1016/j.semcdb.2015.08.005 (2015).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  18. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144, 646–674, https://doi.org/10.1016/j.cell.2011.02.013 (2011).

    CAS  Article  Google Scholar 

  19. Lunt, S. Y. & Vander Heiden, M. G. Aerobic glycolysis: meeting the metabolic requirements of cell proliferation. Annu Rev Cell Dev Biol 27, 441–464, https://doi.org/10.1146/annurev-cellbio-092910-154237 (2011).

    CAS  Article  PubMed  Google Scholar 

  20. Longo, N., Frigeni, M. & Pasquali, M. Carnitine transport and fatty acid oxidation. Biochim Biophys Acta 1863, 2422–2435, https://doi.org/10.1016/j.bbamcr.2016.01.023 (2016).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  21. James, A. M. et al. Non-enzymatic N-acetylation of Lysine Residues by AcetylCoA Often Occurs via a Proximal S-acetylated Thiol Intermediate Sensitive to Glyoxalase II. Cell Rep 18, 2105–2112, https://doi.org/10.1016/j.celrep.2017.02.018 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  22. Laconi, E. et al. Studies on liver tumor promotion in the rat by orotic acid: dose and minimum exposure time required for dietary orotic acid to promote hepatocarcinogenesis. Carcinogenesis 14, 1771–1775 (1993).

    CAS  Article  Google Scholar 

  23. Abraham, S., Kopelovich, L. & Chaikoff, I. L. Dietary and Hormonal Regulation of the Hepatic Citrate-Cleavage Enzyme. Biochim Biophys Acta 93, 185–187 (1964).

    CAS  Article  Google Scholar 

  24. Fernie, A. R., Carrari, F. & Sweetlove, L. J. Respiratory metabolism: glycolysis, the TCA cycle and mitochondrial electron transport. Curr Opin Plant Biol 7, 254–261, https://doi.org/10.1016/j.pbi.2004.03.007 (2004).

    CAS  Article  PubMed  Google Scholar 

  25. Amaral, A. U., Cecatto, C., Castilho, R. F. & Wajner, M. 2-Methylcitric acid impairs glutamate metabolism and induces permeability transition in brain mitochondria. J Neurochem 137, 62–75, https://doi.org/10.1111/jnc.13544 (2016).

    CAS  Article  PubMed  Google Scholar 

  26. Rui, L. Energy metabolism in the liver. Compr Physiol 4, 177–197, https://doi.org/10.1002/cphy.c130024 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

  27. Payne, A. H. & Hales, D. B. Overview of steroidogenic enzymes in the pathway from cholesterol to active steroid hormones. Endocr Rev 25, 947–970, https://doi.org/10.1210/er.2003-0030 (2004).

    CAS  Article  PubMed  Google Scholar 

  28. Chiang, J. Y. Regulation of bile acid synthesis. Front Biosci 3, d176–193 (1998).

    CAS  Article  Google Scholar 

  29. Holick, M. F. Vitamin D: A millenium perspective. J Cell Biochem 88, 296–307, https://doi.org/10.1002/jcb.10338 (2003).

    CAS  Article  PubMed  Google Scholar 

  30. Terao, M. & Katayama, I. Local cortisol/corticosterone activation in skin physiology and pathology. J Dermatol Sci 84, 11–16, https://doi.org/10.1016/j.jdermsci.2016.06.014 (2016).

    CAS  Article  PubMed  Google Scholar 

  31. Toledo, A. et al. Taurodeoxycholate stimulates intestinal cell proliferation and protects against apoptotic cell death through activation of NF-kappaB. Dig Dis Sci 49, 1664–1671 (2004).

    CAS  Article  Google Scholar 

  32. Yang, F. et al. Spontaneous development of liver tumors in the absence of the bile acid receptor farnesoid X receptor. Cancer Res 67, 863–867, https://doi.org/10.1158/0008-5472.CAN-06-1078 (2007).

    CAS  Article  PubMed  Google Scholar 

  33. Gadaleta, R. M., Cariello, M., Sabba, C. & Moschetta, A. Tissue-specific actions of FXR in metabolism and cancer. Biochim Biophys Acta 1851, 30–39, https://doi.org/10.1016/j.bbalip.2014.08.005 (2015).

    CAS  Article  PubMed  Google Scholar 

  34. Mukherjee, S. et al. Retinoids in the treatment of skin aging: an overview of clinical efficacy and safety. Clin Interv Aging 1, 327–348 (2006).

    CAS  Article  Google Scholar 

  35. Athar, M., Walsh, S. B., Kopelovich, L. & Elmets, C. A. Pathogenesis of nonmelanoma skin cancers in organ transplant recipients. Arch Biochem Biophys 508, 159–163, https://doi.org/10.1016/j.abb.2011.01.004 (2011).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  36. Elmets, C. A. & Athar, M. Milestones in photocarcinogenesis. J Invest Dermatol 133, E13–17, https://doi.org/10.1038/skinbio.2013.179 (2013).

    Article  PubMed  PubMed Central  Google Scholar 

  37. Seifert, Rc. C. M. P. and cUMP: emerging second messengers. Trends Biochem Sci 40, 8–15, https://doi.org/10.1016/j.tibs.2014.10.008 (2015).

    CAS  Article  PubMed  Google Scholar 

  38. Kopelovich, L. & Shea-Herbert, B. Heritable one-hit events defining cancer prevention? Cell Cycle 12, 2553–2557, https://doi.org/10.4161/cc.25690 (2013).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  39. Fan, M. et al. Altered transcriptome signature of phenotypically normal skin fibroblasts heterozygous for CDKN2A in familial melanoma: relevance to early intervention. Oncotarget 4, 128–141, https://doi.org/10.18632/oncotarget.786 (2013).

    Article  PubMed  PubMed Central  Google Scholar 

  40. Phatak, A. et al. Global gene expression of histologically normal primary skin cells from BCNS subjects reveals “single-hit” effects that are influenced by rapamycin. Oncotarget 10, 1360–1387, https://doi.org/10.18632/oncotarget.26640 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  41. Levanat, S. et al. A two-hit model for developmental defects in Gorlin syndrome. Nat Genet 12, 85–87, https://doi.org/10.1038/ng0196-85 (1996).

    CAS  Article  PubMed  Google Scholar 

  42. Johnson, R. L. et al. Human homolog of patched, a candidate gene for the basal cell nevus syndrome. Science 272, 1668–1671 (1996).

    ADS  CAS  Article  Google Scholar 

  43. Gailani, M. R. et al. The role of the human homologue of Drosophila patched in sporadic basal cell carcinomas. Nat Genet 14, 78–81, https://doi.org/10.1038/ng0996-78 (1996).

    CAS  Article  PubMed  Google Scholar 

  44. Xie, J. et al. Activating Smoothened mutations in sporadic basal-cell carcinoma. Nature 391, 90–92, https://doi.org/10.1038/34201 (1998).

    ADS  CAS  Article  PubMed  Google Scholar 

  45. Aszterbaum, M. et al. Ultraviolet and ionizing radiation enhance the growth of BCCs and trichoblastomas in patched heterozygous knockout mice. Nat Med 5, 1285–1291, https://doi.org/10.1038/15242 (1999).

    CAS  Article  PubMed  Google Scholar 

  46. Sangalli, J. R. et al. Metabolic gene expression and epigenetic effects of the ketone body beta-hydroxybutyrate on H3K9ac in bovine cells, oocytes and embryos. Sci Rep 8, 13766, https://doi.org/10.1038/s41598-018-31822-7 (2018).

    ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

  47. Preidis, G. A., Kim, K. H. & Moore, D. D. Nutrient-sensing nuclear receptors PPARalpha and FXR control liver energy balance. J Clin Invest 127, 1193–1201, https://doi.org/10.1172/JCI88893 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  48. Majumdar, R. et al. Glutamate, ornithine, arginine, proline, and polyamine metabolic interactions: the pathway is regulated at the post-transcriptional level. Frontiers in Plant Science 7 (2016).

  49. Banerjee, S., Schmidt, T., Fang, J., Stanley, C. A. & Smith, T. J. Structural studies on ADP activation of mammalian glutamate dehydrogenase and the evolution of regulation. Biochemistry 42, 3446–3456, https://doi.org/10.1021/bi0206917 (2003).

    CAS  Article  PubMed  Google Scholar 

  50. Jizhang, N., Pavlova, N. & Thompson, C. B. Cancer cell metabolism: The essential role of the non essential amino acid, glutamine. The EMBO J 36, 1302–1305, https://doi.org/10.15252/emboj/20196/51 (2017).

    Article  Google Scholar 

  51. Athar, M. et al. Inhibition of Smoothened Signaling Prevents Ultraviolet B-Induced Basal Cell Carcinomas through Regulation of Fas Expression and Apoptosis. Cancer Res 64, 7545–7552, https://doi.org/10.1158/0008-5472CAN-04-1393 (2004).

    CAS  Article  PubMed  Google Scholar 

  52. Arumugam, A. et al. Keratin-6 driven ODC expression to hair follicle keratinocytes enhances stemness and tumorigenesis by negatively regulating Notch. Biochem Biophys Res Commun 451, 394–401, https://doi.org/10.1016/j.bbrc.2014.07.129 (2014).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  53. Heske, C. M. et al. Matrix Screen Identifies Synergistic Combination of PARP Inhibitors and Nicotinamide Phosphoribosyltransferase (NAMPT) Inhibitors in Ewing Sarcoma. Clin Cancer Res 23, 7301–7311, https://doi.org/10.1158/1078-0432 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  54. Zhang, M. et al. Adipocyte-derived lipids mediate melanoma progression via FATP proteins. Cancer Discov 8, 1006–1025, https://doi.org/10.1158/2159-8290 (2018).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

This work was supported by the NIH grant RO1 ES026219 to M.A.

Author information

Affiliations

Authors

Contributions

C.L. conducted the experiments. C.L., B.M. and S.M.M. performed the data analysis. M.K. and Z.W. contributed to the experiment implementation. M.A., D.R.B. and L.K. supervised the project and drafted the manuscript. S.A. and A.K. contributed to data interpretation. All authors approved the final version of the manuscript.

Corresponding authors

Correspondence to Levy Kopelovich or Mohammad Athar.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The original online version of this Article was revised: The original version of this Article contained errors in Figure 3c, where the graph showing “PI” was a duplication of “LPE”.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Li, C., Mishra, B., Kashyap, M. et al. Patched1 haploinsufficiency severely impacts intermediary metabolism in the skin of Ptch1+/−/ODC transgenic mice. Sci Rep 9, 13072 (2019). https://doi.org/10.1038/s41598-019-49470-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1038/s41598-019-49470-w

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing