E4F1-mediated control of pyruvate dehydrogenase activity is essential for skin homeostasis

Significance

We found that the multifunctional protein E4 transcription factor 1 (E4F1) transcriptionally regulates a metabolic program involved in pyruvate metabolism that is required to maintain skin homeostasis. E4F1 deficiency in skin resulted in deregulated expression of dihydrolipoamide acetlytransferase (Dlat), a gene encoding the E2 subunit of the mitochondrial pyruvate dehydrogenase (PDH) complex. Accordingly, E4f1 knock-out (KO) keratinocytes exhibited impaired PDH activity and a metabolic reprogramming associated with remodeling of their microenvironment and alterations of the basement membrane, leading to epidermal stem cell mislocalization and exhaustion of the epidermal stem cell pool. Our data reveal a central role for Dlat in the metabolic program regulated by E4F1 in skin and illustrate the importance of PDH activity in skin homeostasis.

Abstract

The multifunctional protein E4 transcription factor 1 (E4F1) is an essential regulator of epidermal stem cell (ESC) maintenance. Here, we found that E4F1 transcriptionally regulates a metabolic program involved in pyruvate metabolism that is required to maintain skin homeostasis. E4F1 deficiency in basal keratinocytes resulted in deregulated expression of dihydrolipoamide acetyltransferase (Dlat), a gene encoding the E2 subunit of the mitochondrial pyruvate dehydrogenase (PDH) complex. Accordingly, E4f1 knock-out (KO) keratinocytes exhibited impaired PDH activity and a redirection of the glycolytic flux toward lactate production. The metabolic reprogramming of E4f1 KO keratinocytes associated with remodeling of their microenvironment and alterations of the basement membrane, led to ESC mislocalization and exhaustion of the ESC pool. ShRNA-mediated depletion of Dlat in primary keratinocytes recapitulated defects observed upon E4f1 inactivation, including increased lactate secretion, enhanced activity of extracellular matrix remodeling enzymes, and impaired clonogenic potential. Altogether, our data reveal a central role for Dlat in the metabolic program regulated by E4F1 in basal keratinocytes and illustrate the importance of PDH activity in skin homeostasis.

Renewal and wound healing of the epidermis rely on a pool of epidermal stem cells (ESC) located in the basal layer of the interfollicular epithelium (IFE) and in the bulge region of the hair follicle (HF). In the IFE, these long-lived ESC give rise to progenitors with increased proliferative capacities that differentiate into keratinocytes as they migrate upward into suprabasal layers. Numerous studies have addressed the role of several key signaling pathways, such as those implicating bone morphogenetic proteins, TGF-β, Notch, Sonic Hedgehog, or Wnt in skin homeostasis, and how they control ESC maintenance (1–3). The role of these pathways in regulating stemness has been attributed to the regulation of cell proliferation, cell death, cellular senescence, cell adhesion, or differentiation. Although previous data indicate that some of these stem cell regulators also control energy metabolism in the hematopoietic or neuronal lineages (4), very few studies have yet addressed their metabolic functions in keratinocytes. In addition, the potential role of specific metabolic regulators in the control of skin homeostasis remains poorly documented. Nevertheless, previous observations indicate that deregulation of the nutrient-sensing mammalian target of rapamycin pathway in basal keratinocytes occurs as a consequence of prolonged Wnt signaling, leading to the progressive exhaustion of HF bulge stem cells (5). Recent data also indicate that genetic inactivation in mouse epidermis of mitochondrial transcription factor A (Tfam), a gene involved in mitochondrial DNA replication and transcription, impinges on keratinocyte differentiation but does not impair maintenance of basal keratinocytes (6). Although these results suggest that basal keratinocytes display a metabolic status that is different from their differentiated counterparts, further studies are warranted to decipher the poorly understood role of metabolism in the regulation of epidermal cell fate.

We previously identified the multifunctional protein E4 transcription factor 1 (E4F1) as an essential regulator of skin homeostasis and ESC maintenance (7). E4F1 was originally identified as a cellular target of the E1A viral oncoprotein (8, 9). Since then, several laboratories have shown that E4F1 directly interacts with several oncogenes and tumor suppressors, including p53, BMI1, RB, RASSF1A, SMAD4, or HMGA2 proteins (10–16). Consistent with its implication in different oncogenic pathways, E4F1 acts as a survival factor in cancer cells (17, 18). Moreover, characterization of E4f1 knock-out (KO) mice showed that E4f1 is an essential gene in embryonic stem cells and during early embryogenesis (19). Using E4f1 conditional KO mice, we previously reported that E4f1 inactivation in the epidermis results in ESC defects through a mechanism that involves, at least partly, the deregulation of the Bmi1–Arf–p53 pathway (7). Here, we show evidence supporting a major role for E4F1 in pyruvate metabolism that governs ESC maintenance and skin homeostasis.

Results

E4f1 Inactivation in Basal but Not Suprabasal Adult Keratinocytes Leads to Epidermal Defects and Exhaustion of the ESC Pool.

Using E4f1 whole-body conditional KO mice (E4f1KO; RERT), we previously identified an essential role for E4f1 in adult skin homeostasis (7). In this genetically engineered mouse model, E4f1 inactivation was achieved in the entire skin of adult animals, including the dermal compartment. To assess the cell of origin of these skin defects, we generated new mouse strains by crossing E4f1 conditional KO mice to transgenic animals expressing the tamoxifen (tam) -inducible CreER recombinase under the control of the keratin 14 (K14) or keratin 10 (K10) promoters [hereafter referred to as E4f1(K14)KO and E4f1(K10)KO strains], allowing acute inactivation of E4f1 in adult keratinocytes of the basal or spinous layers, respectively (20).

Molecular and histological analyses of adult back skin of 8- to 12-wk-old E4f1(K10)KO animals confirmed that topical skin applications of tam activated the Cre recombinase in suprabasal but not in basal keratinocytes (Fig. S1). Neither histological alterations nor differences in the expression pattern of the basal-cell specific K14 marker and of the differentiation markers K10 and involucrin were identified in skin samples harvested up to 4 mo upon regular tam administration (Fig. S2). In sharp contrast, inactivation of E4f1 in adult basal keratinocytes of E4f1(K14)KO mice resulted in skin phenotypes that recapitulated those originally observed in tam-treated E4f1^-/flox; RERT adult mice. Thus, 2 wk after tam administration, E4f1(K14)KO mice displayed epidermal hyperplasia (acanthosis), altered differentiation (dyskeratosis), and a thicker and parakeratotic cornified layer (Fig. 1A and Fig. S3). Aberrant hyperproliferation and mislocalisation of basal keratinocytes was evidenced in E4f1-deficient epidermis by increased Ki67 staining, the presence of K14⁺ cells in suprabasal layers, and the expression of keratin 6 (K6) in the IFE (Fig. S4). At later time points (4–5 wk after tam administration), E4f1-deficient epidermis then became hypocellular, whereas the hyperkeratosis remained evident (Fig. 1A). In addition, impaired keratinocyte differentiation was illustrated by aberrant expression of K10 and involucrin (Fig. S4).

Fig. S1.

Characterization of E4f1(K10)KO model. (A) β-Galactosidase staining of skin section prepared from K10CreER^T2 animals crossed with Cre-dependent β-Gal reporter mice. (Magnification: 20×.) (B) mRNA level of the Cre recombinase in different tissues prepared from K10CreER^T2 mice determined by RT-qPCR. (C, Upper) schematic representation of the E4f1^flox allele before/after Cre-mediated recombination. Recombination of the E4f1^flox allele in E4f1(K10)KO was verified by semiquantitative PCR performed on genomic DNA prepared from total skin samples (containing epidermis, dermis, and hypodermis) using the indicated primers. E4f1KO mouse embryonic fibroblasts (Mefs) were used as a positive control.

Fig. S2.

E4f1 inactivation in suprabasal keratinocytes. Histological analyses of skin sections prepared from E4f1(K10)KO mice or E4f1(K10)CTR control littermates, 6 wk after tam administration. Shown are representative microphotographs of HES-stained skin sections or IHC analyses of cytokeratin 14 (K14), cytokeratin 10 (K10), or involucrin expression, as indicated. (Scale bars, 100 μm.)

Fig. 1.

E4F1 deficiency in basal keratinocytes leads to skin defects and exhaustion of the ESC pool. (A) Microphotographs of (HES)-stained skin sections prepared from E4F1(K14)KO mice or E4F1(K14)CTR littermates, 1, 2, or 5 wk after tam administration. Dashed lines indicate the separation between the epidermis and the dermis. (Scale bars, 100 μm.) (B) Whole mounts of tail epidermis prepared from adult E4F1(K14)KO and CTR mice, 5 wk after tam application, stained with K15 antibody and DAPI. Brackets: bulge area (BG) of the HF. (Scale bar, 100 μm.) (C) Number of follicular stem cells (FSC) in back skin epidermis prepared from the same mice as in B. FACS-analysis of α6/CD34 CD34^high FSC in back skin epidermis prepared from the same mice as in B (mean ± SEM; n = 10). (D) Number of label-retaining (EdU⁺) interfollicular stem cells (LRCs) detected by immunofluorescence (IF) on back-skin sections prepared from adult E4F1(K14)KO mice or E4F1(K14)CTR littemates, 5 wk after tam application. Histobars represent the mean value ± SEM of EdU⁺ cells per millimeter of epidermis (n = 5 animals per group). (E) Clonogenic assays performed with E4F1KO and CTRL primary murine keratinocytes cultured in presence or absence of 4OHT, as indicated (n = 5). Histobars represent the total number of clones per well relative to control cells (expressed as percentages) determined after rhodamine B staining. ***P < 0.001; **P < 0.01; ns, not significant.

Fig. S3.

(A) Microphotographs of HES-stained skin sections prepared from E4f1(K14)KO mice or E4f1(K14)CTR control littermates, 2 wk after tam administration. Note the increased thickness of the epidermis (= epidermal hyperplasia or acanthosis) and increased thickness of the cornified layer (= hyperkeratosis). Epi, epidermis; SC, stratum corneum or cornified layer. The dashed line indicates the separation between the epidermis and the dermis. (B) Microphotographs at higher magnification of skin sections prepared from the same animals as in A showing the presence of nuclei in the cornified layer (parakeratosis) and abnormal keratinocytes (dyskeratosis). (Scale bars, 100 μm.)

Fig. S4.

IHC analyses of K14, K10, involucrin, cytokeratin 6 (K6), and Ki67 expression in skin sections prepared from representative E4f1(K14)KO mice or E4f1(K14)CTR control littermates, 2 wk after tam administration. (Scale bars, 100 μm.)

Consistent with our previous observations, ablation of E4f1 in adult basal keratinocytes altered ESC function and resulted in the definitive exhaustion of the ESC pool. Indeed, tam administration to E4f1(K14)KO mice led to the loss of expression of the bulge HF stem cell marker keratin 15 (K15) (Fig. 1B). Flow cytometry analysis of HF stem cells identified by the coexpression of CD34 and high levels of α6-integrin (CD34⁺/α6^high) confirmed that tam-treated E4f1(K14)KO adult epidermis contained fewer HF stem cells compared with control epidermis (0.65% ± 0.15 vs. 2.25% ± 0.4) (Fig. 1C). We also tracked ESCs in the IFE by analyzing the number of label retaining cells (LRCs) using an adaptation of an in vivo labeling protocol of multipotent ESC based on the utilization of the nucleotide analog ethynyl-2′-deoxyribose (EdU) (21). These analyses showed that E4f1 inactivation resulted in a significant decrease in the number of LRCs 5 wk after tam administration, indicating that E4f1 deficiency led to the definitive loss of ESC in vivo (Fig. 1D). Finally, E4f1KO ESC defects were also illustrated ex-vivo by their impaired clonogenic potential (Fig. 1E).

Thus, these data demonstrate that the epidermal hyperplasia, hyperkeratosis, differentiation defects, and exhaustion of the ESC pool of E4F1-deficient epidermis originate from alterations in basal rather than suprabasal keratinocytes.

E4F1 Controls Pyruvate Metabolism in Keratinocytes Through Transcriptional Regulation of the E2 Subunit of the Pyruvate Dehydrogenase Complex Dlat.

Using a pan-genome ChIP approach combined with next-generation sequencing (ChIP-seq), we identified E4F1 binding sites at the whole-genome level in primary mouse embryonic fibroblasts and embryonic stem cells (18, 22). Functional annotation of E4F1 direct target genes indicated a significant enrichment in genes implicated in metabolism, including a set of five genes encoding core components or regulators of the mitochondrial pyruvate dehydrogenase (PDH) complex (PDC), a multimeric complex that converts pyruvate into Acetyl-CoA (AcCoA). In embryonic stem cells, this set of E4F1-controlled genes includes the E2 and E3 subunits of the PDC, dihydrolipoamide acetyltransferase (Dlat) and dihydrolipoyl dehydrogenase (Dld), the regulatory subunit of the PDH phosphatase complex (Pdpr), the pyruvate transporter of the inner mitochondrial membrane Brp44l/Mpc1 (23, 24), and the mitochondrial transporter Slc25a19 that transports the PDH cofactor thiamine pyrophosphate (25). These results prompted us to evaluate whether E4F1 also controlled this set of PDH-related genes in primary keratinocytes. First, we confirmed by quantitative ChIP that endogenous E4F1 was recruited to the promoter of Dlat, Dld, Slc25a19, and Brp44l in cultured primary murine keratinocytes (Fig. 2A). Similarly to E4f1KO fibroblasts and muscle cells, E4F1-deficient keratinocytes displayed a marked decrease of Dlat mRNA level. However, the expression of Dld, Pdpr, Slc25a19, and Brp44l/Mpc1 remained unchanged upon E4f1 inactivation, suggesting that other E4F1-independent mechanisms contribute to their expression in keratinocytes (Fig. 2B). The mRNA levels of other PDH-related genes, including Pdha1, which encodes the E1 subunit of the PDC, the PDH -kinases 1/4 (Pdk1, Pdk4) and -phosphatases 1/2 (Pdp1, Pdp2) remained unchanged in E4f1-deficient keratinocytes (Fig. S5). Decreased Dlat expression was confirmed at the protein level, as shown by immunostaining of skin samples prepared from E4f1(K14)KO animals, 1 wk after tam-administration. This decrease was further confirmed by immunoblotting both in tam-treated E4f1(K14)KO epidermis and in cultured E4f1KO primary keratinocytes (Fig. 2 C–E). Consistent with DLAT deregulation, altered PDH enzymatic activity was detected in these cells (Fig. 2 F and G and Fig. S6A). Taken together, these data indicate that the E2 subunit of the PDC Dlat is a major direct transcriptional target of E4F1 in basal keratinocytes.

Fig. 2.

E4f1 inactivation in basal keratinocytes results in decreased Dlat expression and impaired PDH activity. (A) ChIP-qPCR assays performed with anti-E4F1 or control (CTR) antibodies in cultured primary keratinocytes on the promoter region of Dlat, Dld, Mpc1/brp44l, Pdpr, and Slc25a19. A gene-poor noncoding region of chromosome 6 (NC2) and the Pdha1 promoter region (TSS, transcription start site) were used as controls. Enrichments are represented as percentages of input (mean value ± SEM; n = 3). (B) mRNA levels of E4f1, Dlat, Dld, Brp44L/Mpc1, Pdpr and Slc25a19 in the epidermis of E4f1(K14)KO mice or E4f1(K14)CTR littermates, 1 wk after tam administration. Histobars represent the mean value ± SEM (n = 5) determined by RT-qPCR. (C and D) Protein levels of E4F1, DLAT, PDHA1, and β-actin (loading control) determined by immunoblotting in (C) total protein extracts prepared from epidermis of E4f1(K14)KO mice and control littermates, 1 wk after tam administration, or (D) E4f1KO or CTRL cultured primary keratinocytes. ^#Nonspecific band. (E) Immunostaining of DLAT in skin sections prepared from the same mice as in C. Sections were counterstained with DAPI. (Scale bar, 50 μm.) (F and G) PDH activity measured in protein extracts prepared (F) from epidermis of E4f1(K14)KO and CTR mice, 1 wk after tam administration, or (G) from cultured primary keratinocytes of the indicated genotype using Colorimetric Assay kit (Biovision) (mean ± SEM, n = 5). All analyses in cultured primary keratinocytes were performed after 4 d of culture in the presence of 4OHT. ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant.

Fig. S5.

(A) mRNA levels of E4f1, Dlat, Dld, Brp44l, Pdpr, Slc25a19 in E4f1KO or CTR cultured primary keratinocytes, 4 d after 4OHT addition to the culture medium. Histobars represent the mean value ± SEM (n = 4) determined by RT-qPCR. (B) mRNA levels of Pdha1, Pdk1, Pdk4, Pdp1, Pdp2 in the epidermis of E4f1(K14)KO mice or E4f1(K14)CTR control littermates, 1 wk after tam administration. Histobars represent the mean value ± SEM (n = 5 animals per group) determined by RT-qPCR. (C) mRNA levels of Pdha1, Pdk1, Pdk4, Pdp1, Pdp2 in E4f1KO or CTR cultured primary keratinocytes, 4 d after 4OHT addition to the culture medium. Histobars represent the mean value ± SEM (n = 4) determined by RT-qPCR. *P < 0.05; **P < 0.01; ns, not significant.

Fig. S6.

E4F1 depletion leads to decreased PDH activity. (A) PDH activity in E4f1KO or CTR cultured primary keratinocytes, using a different method, DipStick Assay Kit (ab109882, Abcam), than the one used in Fig. 2. (Histobars represent the mean value ± SEM (n = 5). (B) GLUT1 levels determined by flow cytometry in E4f1KO or CTR cultured primary keratinocytes. (C) IHC analysis of BASIGIN (Upper) and of the glucose transporter GLUT1 (Lower) expression in skin sections prepared from E4f1(K14)KO mice and control littermates, 2 wk after 4OHT administration. (Scale bar, 50 μm.) (D) OCR measured using a Seahorse Bioanalyzer in E4f1KO or CTR cultured primary keratinocytes. All analyses in cultured primary keratinocytes were performed after 4 d of culture in presence of 4OHT. **P < 0.01.

E4f1KO Results in Metabolic Reprogramming of Keratinocytes.

Next, we characterized the metabolic consequences of impaired PDH activity in E4F1-deficient keratinocytes. We postulated that decreased PDH activity in E4F1-deficient keratinocytes triggered a decrease of glucose-derived AcCoA production and the redirection of the glycolytic flux toward lactate production (Fig. 3A). Consistent with this hypothesis and the role of AcCoA as a donor substrate for acetylation reactions, E4f1KO keratinocytes exhibited decreased histone H4 acetylation, as shown by immunoblotting using an anti-pan-acetyl lysine histone H4 antibody (Fig. 3B). Moreover, in line with increased pyruvate metabolism by the NADH-dependent lactate dehydrogenase (LDH), E4f1KO keratinocytes displayed an increased NAD⁺/NADH ratio (Fig. 3C). Increased expression of the glucose transporter GLUT1 suggested that glucose uptake increased upon E4f1 inactivation in keratinocytes (Fig. S6B). These cells also exhibited increased expression of the monocarboxylate transporter MCT4 that favors the efflux of lactate outside the cell (Fig. 3 D and E). Accordingly, increased lactate secretion by E4f1KO keratinocytes was evidenced by a change in their extracellular acidification rate (ECAR) (Fig. 3F). Other metabolic changes were observed in E4F1-deficient keratinocytes, as illustrated by increased fatty acid oxidation (FAO) (Fig. 3G). This adaptative metabolic response was likely sufficient to sustain mitochondrial respiration because no significant difference was observed in oxygen consumption upon E4f1 inactivation in keratinocytes cultured in complete medium (Fig. S6D). Analyses of tam-treated E4f1(K14)KO mice and control littermates confirmed that E4F1-deficient keratinocytes underwent the same metabolic reprogramming in vivo. Thus, immunohistochemistry (IHC) analyses of skin samples prepared from these animals indicated that E4f1 inactivation in basal keratinocytes resulted in increased expression of GLUT1, MCT4, and of CD147/BASIGIN, a chaperone required for MCT4 relocalization at the cytoplasmic membrane (Fig. 3H and Fig. S6C). Strikingly, E4f1(K14)KO mice exhibited lactic acidemia and increased level of circulating ketone bodies, a by-product of FAO (Fig. 3 I and J). Moreover, the clonogenic potential of E4f1KO keratinocytes was partly rescued by addition of exogenous acetate that can replenish AcCoA pools (Fig. 3K), confirming that the profound metabolic reprogramming of E4F1-deficient keratinocytes impinged on their biological functions.

Fig. 3.

Metabolic reprogramming of E4F1-deficient keratinocytes. (A) Schematic representation of the metabolic reprogramming in E4f1KO keratinocytes including the redirection of the glycolytic flux toward lactate production. (B) Protein levels of acetylated-lysine histone H4 and β-actin (loading control) determined by immunoblotting in total protein extracts prepared from epidermis of E4f1(K14)KO mice (Upper) or from E4f1KO cultured primary keratinocytes (Lower) and match control (CTR) samples. (C) NAD⁺/NADH ratio in E4f1KO and CTR cultured primary keratinocytes (mean ± SEM; n = 9). (D and E) MCT4 expression was determined by IF (D) and by immunoblotting (E) in E4f1KO and CTR cultured primary keratinocytes. (Magnification: 40×.) (F) ECAR of E4f1KO and CTR primary keratinocytes in basal conditions or after addition of glucose (mean ± SEM; n = 5). (G) Relative level of FAO measured upon incubation of E4f1KO and CTR cultured keratinocytes with ³H-palmitate (mean ± SEM of n = 5). (H) Immunohistochemistry analysis of MCT4 expression in skin sections prepared from E4f1(K14)KO and CTR mice, 2 wk after 4OHT administration. (Scale bar, 50 μm.) (I and J) Lactate (I) and ketone bodies (J) levels in the serum of E4f1(K14)KO mice and CTR littermates, 1 wk after tam administration (n = 5 animals per group). (K) Clonogenic assays performed with E4f1KO and CTR primary keratinocytes in presence or absence or acetate, as indicated (n = 3). ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant. TCA, tricarboxylic acid cycle.

Metabolic Reprogramming of E4F1-Deficient Keratinocytes Associates with Remodeling of the Microenvironment and Loss of Adhesion of the ESC with the Basement Membrane.

In many tumors, increased lactate secretion has been linked to the remodeling of the extracellular matrix (ECM) and degradation of the basement membrane (BM) by ECM-remodeling enzymes (26). To further characterize the consequences of the metabolic reprogramming of E4F1-deficient keratinocytes, we performed histological analyses of E4f1(K14)KO skin. Electron microscopy analyses indicated that E4f1 inactivation in basal keratinocytes resulted in disorganization of the BM, which appeared either diffused with thinner lamina densa or focally disrupted (Fig. S7A). Alterations of the BM in E4f1KO skin was confirmed upon staining of skin sections by the Gomori reticulin method, which stains the argyrophilic (silver staining) fibrous structures present in the BM (Fig. S7B). Immunostaining of skin samples prepared from E4f1(K14)KO mice with anti-laminin V antibody showed that the expression pattern of this essential component of the BM was diffused and focally discontinuous in E4f1KO skin sections compared with its defined and continuous pattern in control samples (Fig. 4A). This defect also correlated with an abnormal expression pattern of integrin β4 (Itgβ4). In areas showing broad disruption of the BM, Itgβ4 expression was not restricted to the basal pole of keratinocytes but was also detected at the apical or lateral sides of both basal and suprabasal keratinocytes (Fig. 4A). Remodeling of the ECM within the dermal compartment was also evidenced by picro-Sirius red staining of collagen fibers on skin sections prepared from tam-treated E4f1(K14)KO mice (Fig. S7C). These results led us to investigate whether the massive remodeling of the ECM and alterations of the BM observed upon E4f1 inactivation resulted from increased activity of ECM-remodeling enzymes. Increased matrix metallopeptidase 9 (MMP9) and cathepsin activities were detected by gelatin-zymography in protein extracts prepared from total skin samples of tam-treated E4f1(K14)KO mice (Fig. 4B and Fig. S7D). Moreover, increased MMP2, MMP9, and cathepsin activities were also evidenced in the culture medium of E4f1KO primary keratinocytes (Fig. 4 C and D). Addition of the LDH-inhibitor oxamate in the culture medium decreased cathepsin activities, confirming that their induction resulted from the metabolic reprogramming of these cells (Fig. 4D). Moreover, stable expression of ectopic TIMP1, a broad MMP inhibitor, in feeder cells partly rescued the clonogenic potential of E4f1KO ESC (Fig. 4E). Improved clonogenicity of E4F1-deficient ESC was also observed upon incubation with GM6001, a pharmacological MMP inhibitor with broad spectrum (Fig. 4F). Taken together, these data indicate that the induction of ECM remodeling enzymes in E4F1-deficient keratinocytes is a consequence of their metabolic reprogramming and impinges on their clonogenic potential.

Fig. S7.

E4F1 deficiency results in remodeling of the extracellular matrix and alterations of the basement membrane. (A) Transmission electron microscopy analysis of skin samples prepared from E4f1(K14)KO mice and control littermates, 1 wk after tam administration. (B and C) Staining of skin sections prepared from E4f1(K14)KO mice and E4f1(K14)CTR littermates with reticulin (B) and with Sirius red (C). Arrows indicate structural alterations of the basement membrane. (D) Cathepsins activities measured by zymography in total skin extracts prepared from E4f1(K14)KO mice and E4f1(K14)CTR control littermates. *P < 0.05. (Scale bars, 50 μm.)

Fig. 4.

E4f1 inactivation in basal keratinocytes results in alterations of the BM, remodeling of the ECM and ESC mislocalization. (A) IF analysis of Laminin V (LamV, red) and Intβ4 (green) expression in skin sections prepared from E4f1(K14)KO and CTR mice, 2 wk after tam administration. Sections were counterstained with DAPI (blue). (Scale bars, 50 μm.) (B, Left) Representative zymogel analysis of MMP activities in protein extracts prepared from skin samples of E4f1(K14)KO and CTR mice, 2 wk after tam administration. For normalization, β-actin immunoblotting was performed on the same protein extracts. (Right) Quantification of MMP2 and MMP9 activities (mean ± SEM, n = 5 animals per group). (C, Left) MMP activities measured by zymography in conditioned media (CM) from E4f1KO or CTR cultured primary keratinocytes. For normalization, β-actin expression was analyzed by immunoblotting in protein extracts prepared from the same cells. (Right) Quantification of MMP2 and MMP9 activities (mean ± SEM, n = 4). (D, Left) Cathepsin activities measured by zymogel in CM from E4f1KO or CTR primary keratinocytes cultured in the presence or absence of oxamate. (Right) Quantification of cathepsin activities (mean ± SEM, n = 3). (E) Clonogenic assays performed with E4f1KO and CTR primary keratinocytes in presence of feeders transduced with empty or TIMP1-encoding retroviruses (mean ± SEM; n = 9). (F) Clonogenic assays performed with E4f1KO and CTR primary keratinocytes in presence or absence or the MMP-inhibitor GM6001 (mean ± SEM; n = 6). (G) Mislocalization of E4f1KO ESC. LRCs were identified by IF on skin sections prepared from E4f1(K14)KO and CTR mice, 2 wk after tam administration. MCT4 costaining was performed to identify metabolically reprogrammed cells. Sections were counterstained with DAPI. (Scale bar, 30 μm.) Arrows indicate EdU⁺ LRCs. Histobars represent the total number of EdU⁺ LRCs per millimeter of epidermis and their respective localization in the basal or suprabasal (SB) layers (mean ± SEM, n = 4 animals per group). ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant.

Based on these results, we hypothesized that the observed disruption of the BM impacted on the maintenance of ESC within their normal microenvironment, leading to the definitive exhaustion of the ESC pool. To test this hypothesis, we analyzed EdU⁺ LRCs on skin sections prepared from E4f1(K14)KO mice or control littermates 2 wk after tam administration and evaluated their localization within the epidermis. The same skin sections were also processed to assess MCT4 expression as a surrogate marker of the metabolic reprogramming of E4f1KO keratinocytes. Two weeks after tam administration, E4f1KO epidermis displayed approximately the same number of EdU⁺ LRCs than control epidermis (0.25 ± 0.05 vs. 0.23 ± 0.02 per millimeter, respectively). However, the number of E4f1KO LRCs in a suprabasal position was significantly increased compared with LRCs of control epidermis that remained, as expected, in the basal layer (suprabasal: 0.1 ± 0.04 vs. 0.01 ± 0.01; basal: 0.14 ± 0.03 vs. 0.21 ± 0.003 per millimeter, respectively) (Fig. 4G). Interestingly, this atypical feature of E4f1KO LRCs was particularly evident within focal epidermal lesions exhibiting MCT4 positivity, whereas LRCs remained in contact with the BM in the adjacent MCT4⁻ areas of the same epidermis (Fig. 4G). Five weeks after tam administration, the number of LRC diminished in E4f1KO epidermis, confirming that E4f1 inactivation finally ended in the exhaustion of the ESC pool (Fig. 1D). Thus, these data show that the metabolic reprogramming triggered by E4f1 inactivation in basal keratinocytes associates with the remodeling of their microenvironment and alterations of the BM, leading to the loss of attachment of the ESC within their normal niche and their definitive loss.

DLAT Is a Central Component of the E4F1-Regulated Metabolic Program in Basal Keratinocytes.

Because Dlat appeared as one of the most deregulated E4F1 target gene in the epidermis of tam-treated E4f1(K14)KO animals, we evaluated the consequences of shRNA-mediated depletion of Dlat in primary keratinocytes. Lentiviral-mediated delivery of shRNAs targeting Dlat in populations of primary murine keratinocytes led to an expected decrease of DLAT protein level and PDH activity (Fig. 5A and Fig. S8A) and induced phenotypes that were reminiscent of those observed in E4f1KO keratinocytes. Thus, DLAT depletion in primary keratinocytes resulted in their metabolic reprogramming, as illustrated by increased MCT4 expression, and increased MMP2, MMP9, and cathepsin activities (Fig. 5 B and C and Fig. S8 B and C). Furthermore, similarly to E4F1-deficient ESC, DLAT-depleted keratinocytes also displayed an alteration of their clonogenic potential (Fig. 5D). Taken together, these data strengthen the role of DLAT as a central component of E4F1-regulated metabolic program in primary keratinocytes.

Fig. 5.

DLAT depletion in primary murine keratinocytes results in their metabolic reprogramming, increased MMP activity and ESC defects. (A) DLAT and ACTIN (loading control) expression determined by immunoblotting in total protein extracts prepared from murine primary keratinocytes transduced with lentiviruses encoding control or Dlat ShRNAs. (B) IF analysis of MCT4 expression in murine primary keratinocytes expressing control or Dlat ShRNAs. (Magnification: 40×.) (C, Left) MMP activities measured by zymogel in CM from primary keratinocytes expressing control or Dlat ShRNAs. (Right) Quantification of MMP2 and MMP9 activities (mean ± SEM, n = 3). (D) Clonogenic assays performed with primary murine keratinocytes expressing control or Dlat ShRNAs (mean value ± SEM; n = 5).

Fig. S8.

DLAT depletion leads to decreased PDH activity, induction of MCT4 expression, and increased cathepsin activities. (A) PDH activity was measured in cultured primary keratinocytes transduced with lentiviruses encoding control or Dlat ShRNAs. Histobars represent the mean value ± SEM (n = 3). (B) MCT4 protein expression in DLAT-depleted primary keratinocytes analyzed by immunoblotting. (C) Cathepsins activities measured by zymogel in conditioned media from primary keratinocytes expressing control or Dlat shRNAs. **P < 0.01.

Discussion

Our analyses performed in different mouse models where E4f1 was genetically inactivated in the basal or the spinous layers of the epidermis show that the complex skin phenotypes observed upon E4f1 inactivation originate from defects in basal keratinocytes. Our results indicate that E4F1 deficiency in these cells leads to a metabolic reprogramming of keratinocytes that affects skin homeostasis and ended in the definitive exhaustion of the ESC pool. We found that this metabolic shift, which includes the redirection of the glycolytic flux toward lactate production, is a direct consequence of PDH deficiency. Moreover, our data identify DLAT, the E2 subunit of the PDC, as an essential component of this metabolic program regulated by E4F1 in keratinocytes.

Whether E4F1-mediated control of the PDC in keratinocytes is clinically relevant remains to be determined. It is worth noting, however, that a homozygous nonsynonymous mutation in the coding region of the E4f1 gene has been recently identified in a patient presenting clinical symptoms resembling those of Leigh syndrome patients (27). Although skin abnormalities have been reported only in some Leigh syndrome patients (28), they are part of the broad spectrum of clinical manifestations that are commonly observed in several mitochondrial disorders (29). Further investigations are necessary to evalutate whether E4F1-mediated control of mitochondrial activities, which likely extend beyond the control of the PDC, contribute to the skin manifestations observed in these patients.

Another pathological situation that has been associated with changes in PDH activity is cancer. Interestingly, the metabolic rewiring of E4f1KO keratinocytes is reminiscent of the one observed in many cancer cells that display increased aerobic glycolysis, even in high oxygen conditions, an effect known as the Warburg effect. It is well established that PDH deregulation in cancer cells can result from posttranslational modifications of PDC subunits by inhibitory kinases (PDKs), activating phosphatases (PDPs), or the lipoamidase SIRT4 (30, 31). We failed to detect deregulation of Pdks and Pdps mRNA levels in E4f1KO keratinocytes, and our data rather support the notion that transcriptional control of Dlat is the main mechanism by which E4F1 controls PDH activity in normal epidermal cells. It remains to be seen whether E4F1-mediated control of Dlat is an alternative regulatory mechanism of the PDC in skin cancer cells. Nevertheless, our data clearly show that the control of PDH activity by E4F1 in basal keratinocytes is essential for normal skin homeostasis.

Interestingly, as with cancer cells, we show that the metabolic reprogramming of E4f1KO basal keratinocytes results in increased activity of ECM-remodeling enzymes, including MMPs and cathepsins. The exact molecular mechanism by which increased glycolysis activates MMP activity in cancer cells remains controversial. Previous studies have suggested that the MCT-chaperone CD147/BASIGIN increases MMP activity through a yet undefined mechanism (32). However, recent data contradict this working model (33). Whatever the mechanism, the high glycolytic profile and increased activity of tissue-remodeling enzymes of fully transformed cells have been associated with their increased migratory and invasive properties that contribute to metastatic dissemination. Our data show that the metabolic reprogramming of normal E4f1KO keratinocytes recapitulates some features of cancer cells, including their ability to induce the focal degradation of the basement membrane and to remodel their microenvironment. Here, we show that these alterations impact on ESC maintenance within their niche, leading to their mechanical elimination and ending in the complete exhaustion of the ESC pool. Interestingly, we previously reported that the ability of E4F1 to control the Bmi1–ARF–p53 pathway partly contributes to ESC self-renewal (7). These data raise interesting questions regarding the connection between the metabolic reprogramming of E4F1-deficient keratinocytes and the deregulation of the p53 pathway in these cells. The potential cross-talk between PDH activity and the control of the p53 pathway is a promising hypothesis that warrants further investigation.

It was recently proposed that basal keratinocytes rely more on glycolysis to sustain their energetic demand than their differentiated progeny in which mitochondrial-reactive oxygen species trigger epidermal differentiation through Notch and β-catenin signaling (6). Our data do not necessarily contradict this model, but provide clear evidence that when glycolysis is further increased in basal keratinocytes, such as in E4F1-deficient cells, this profoundly alters epidermal homeostasis and ESC maintenance. Our results also question the mechanisms leading to the inhibition of keratinocyte differentiation observed in E4f1KO epidermis.

Altogether our results identify E4F1 as an essential regulator of the metabolic status of basal keratinocytes and stress the importance of a tight control of the PDH activity for epidermal homeostasis.

Materials and Methods

Generation of Mouse Models and Experimental Treatment.

Generation of E4f1 KO and E4f1 cKO mice was previously described (7, 19). These mice were intercrossed with K14CreER (20) or K10CreER^T2 mice to generate experimental groups (E4f1^+/flox; K14CreER, E4f1^−/flox; K14CreER, E4f1^+/flox; K10CreER^T2, and E4f1^−/flox; K14CreER^T2 [referred as to E4F1(K14)CTR, E4F1(K14)KO, E4F1(K10)CTR, E4F1(K10)KO, respectively]. Compound mice were maintained on a mixed genetic background (129Sv/J/DBA/C57BL/6) and housed in a pathogen-free barrier facility. Cre-mediated recombination of the E4f1^flox allele was induced by topical applications of tamoxifen (Sigma; diluted in ethanol, 2 mg/d for 4 consecutive days) on shaved back or tail skin of 8- to 12-wk-old animals. Experiments were approved by the regional ethics committee for animal welfare (Comité éthique pour l'expérimentation animale du Languedoc Roussillon, protocol 12068). Oligonucleotides used for genotyping these animals are provided as in SI Materials and Methods.

Histology, IHC, and Immunolabeling of Skin Sections.

IHC and immunolabeling of skin sections were performed as previously described (7) using the following antibodies: anti-DLAT (sc-32925 Santa Cruz), MCT4 (sc-50329 Santa Cruz), BASIGIN (G-19 sc-9757, Santa Cruz), Laminin V (generous gift from C. Feral’s laboratory, University of Nice, Nice, France), Intβ4 (553745 BD Pharmingen), K14 (PRB-155P (Covance), K10 (PRB-159P Covance), Involucrin (sc-15230 Santa Cruz).

Culture of Primary Keratinocytes.

Murine primary keratinocytes were isolated from newborn skin as previously described (7) and grown in calcium-free Eagle’s MEM (Bio-Whittaker; Lonza) supplemented with 8% (vol/vol) calcium-free FBS (Sigma). Cre-mediated recombination was achieved by adding 4-hydroxy Tamoxifen (4OHT, Sigma; 1 μM final) to the culture medium.

Lactate, Ketone Bodies, and PDH Activity Measurement.

Lactate and ketone bodies concentration were measured from tail blood samples using a lactometer (EKF Diagnostics) and β-ketone strips (Optium, Abbott). PDH activity was measured with PDH Enzyme Activity Dipstick Assay Kit (Abcam) and PDH Activity Colorimetric Assay kit (Biovision) according to the manufacturers’ recommendations.

Statistic Analyses.

The unpaired Student’s t test was used in all analyses. Statistical significance was expressed as follows: *P < 0.05, **P < 0.01, ***P < 0.001.

SI Materials and Methods

Generation of K10CreERT2 Mice and Genotyping of Animal Models.

The Tg(Krt10-creER^T2)39.28.ICS model expressing the CreER^T2 in the spinous layer was generated by pronuclear injection in FVB/N embryos of the modified BAC RP23-336D20 that express a tam-dependent Cre recombinase under the control of the Krt10 promoter. The line was backcrossed into the C57B/6J genetic background for 10 generations before being intercrossed with E4f1 flox animals. To assess Cre-mediated recombination in suprabasal keratinocytes, these mice were crossed with a Cre-dependent β-galactosidase reporter strain Gt(ROSA)26Sor^tm1Sor (MGI:1861932).

E4f1(K14)KO and E4f1(K10)KO animals were genotyped using the following primers:

E4f1 KO allele: Fwd: 5′-CACTGCCTTGGAGGACTTTG-3′; Rev: 5′-CCTCTGTTCCACATACACTTCATTC-3′.

E4f1 flox allele: Fwd: 5′-CCCCAAGAAGCCCAAGTTCC-3′; Rev: 5′-GGCTGCTGCGTGGATTTC-3′.

K14CreER allele: Fwd: 5′-GCCAAGGGGAATGGAAAGTGCC’-3; Rev: 5′-CTCGTTGCATCGACCGGTAA-3′

K10CreER^T2 allele: Fwd: 5′-CCATGGTGATACAAGGGACATCTTCC-3′; Rev: 5′-GCAAAGCCTAGCACCTGTGAGACACG-3′.

Recombination of the E4f1 flox allele in E4F1(K10)KO mice was determined by semiquantitative PCR on genomic DNA prepared from total skin and the following primers:

E4f1 recombined allele (floxed allele): Fwd: 5′- CCTGGGTGCAGATTGGATC-3′; Rev: 5′- GCTAGGTAGGGTAGGAGGCTGTCT-3′.

Internal control (NC2): Fwd: 5′- ACTGGGATCTTCGAACTCTTTGGAC-3′; Rev: 5′- GATGTTGGGGCACTGCCTCATTCACC-3′.

Clonogenic Assays.

For clonogenic assays, primary keratinocytes were seeded in collagen-I–coated plates (BD Bioscience; 50 µg/mL) on a feeder layer of Mitomycin C-treated 3T3-J1 fibroblasts, as previously described (7). Briefly, cells were cultured in DMEM:HamF12 (3:1; Invitrogen) medium supplemented with 10% (vol/vol) calcium-free FBS (Sigma), 4 mM l-glutamine (Gibco), 110 mg/L sodium pyruvate (Gibco), 0.16 ng/mL cholera toxin (Sigma), 0.4 μg/mL hydrocortisone (Sigma), 5 μg/mL insulin (Sigma), and 10 μg/mL EGF (Roche). Cells were grown for 15 d at 32 °C in 8% CO₂ and fixed with 3.7% (vol/vol) PFA (Electron Microscopy Sciences) for 10 min at room temperature, then stained with 1% Rhodamine B. For rescue experiments, Na-Acetate (0.25 mM; Sigma) and GM6001 MMP inhibitor (0.5 μg/mL; MedChem Express) were added in the culture medium and maintained until the end of the assay. Fresh medium was replaced every other day. For TIMP1-based rescue experiments, ESCs were seeded on 3T3-J1 fibroblasts that were previously transduced with empty or TIMP1-encoding retroviruses.

Identification of Label-Retaining ESC.

Interfollicular ESC were identified on skin sections by IHC using an adaptation of an in vivo labeling protocol of multipotent ESC (21) based on the utilization of the nucleotide precursor EdU. Briefly, 10-d-old mice were injected every 12 h intraperitoneally for a total of four injections with 5 mg/mL EdU (CarboSynth) diluted in 7 mM NaOH. After a 2 mo-chase, EdU⁺ ESC were identified on skin sections using a detection method of EdU following the manufacturer’s instructions (Life Technologies). Follicular bulge stem cells were identified in tail or back-skin whole mounts upon staining with an anti-keratin 15 (LHK15, Vector Laboratories) antibody, or by FACS analysis upon staining with anti-CD34 (BD Biosciences) and anti-integrin α6 (BD Biosciences) antibodies, as previously described (7).

FACS Analysis.

For analysis of GLUT1 expression, cultured primary keratinocytes were resuspended in PBA buffer [PBS containing 1 mM EDTA, 2% (vol/vol) FBS (Sigma)] and were incubated at 37 °C for 30 min with the GFP-tagged receptor binding domain of the human T-cell leukemia virus that binds to the extracellular domain of the GLUT1 receptor (34). Cells were then washed with PBA buffer and incubated with 3 ng/μL of propidium iodide to discriminate dead cells before analysis on a FACSCALIBUR flow cytometer (BD Biosciences). For FACS analysis of ESC, freshly isolated primary keratinocytes from adult skin were stained for 30 min on ice with PE-conjugated anti–α6-integrin (BD Pharmingen) and anti-CD34 biotinylated (RAM34; BD Biosciences) antibodies and then with Alexa 647-conjugated streptavidin (BD Pharmingen). FACS-analysis was performed on a FC500 (Beckman Coulter) and data were analyzed with FlowJo software.

Immunofluorescence Analysis of Cultured Primary Keratinocytes.

Cells were fixed with methanol (Sigma) at −20 °C for 5 min before overnight incubation at 4 °C with an anti-MCT4 rabbit polyclonal antibody (sc-50329, Santa Cruz). Revelation was performed using an Alexa 488-conjugated anti rabbit IgG antibody (ThermoFischer) for 2 h at room temperature. Cover glasses were mounted with Mowiol (Biovalley) before analysis on a Zeiss apotome.

Lentiviral Particle Production and Transduction.

Lentiviral particles were produced in 293T packaging cells by transient transfection using Jet-PEI reagent (Ozyme) of pLKO vectors encoding either a control ShRNA (Sigma, SHC002) or ShRNAs targeting mouse Dlat (Sigma-Mission, TRCN0000041608 and TRCN0000041609). Seventy-two hours after transfection, viral supernatants were harvested and added on primary keratinocytes overnight in presence of polybrene (5 μg/mL, Sigma). Antibiotic selection of transduced keratinocytes was performed 48 h after transduction with puromycin (1.17 μg/mL; HyClone).

Transmission Electron Microscopy and Ultrastructural Evaluation.

Adult skin samples were fixed in 3.5% (wt/vol) glutaraldehyde in 0.1 M Sorensen’s phosphate buffer, pH 7.4, overnight at 4 °C. The tissues were then rinsed in Sorensen’s buffer and postfixed in 1% osmic acid for 2 h in the dark at room temperature. After two rinses, the tissues were dehydrated in a graded series of ethanol solutions (30–100%) and embedded in EmBed 812 resin. Sections (60-nm thickness; Leica-Reichert Ultracut E) were counterstained with uranyl acetate and observed using a Hitachi 7100 transmission electron microscope (Centre de Ressources en Imagerie Cellulaire de Montpellier, France).

Immunoblotting.

Total cell extracts were lysed in 0.08 M Tris-HCI (pH 6.8), 2% (wt/vol) SDS, 12% (wt/vol) sucrose, 2% β-mercaptoethanol, bromophenol blue traces and separated with 10% (wt/vol) SDS/PAGE, transferred on a nitrocellulose membrane before probing with the following antibodies recognizing: E4F1 (18), DLAT (sc-271534, Santa Cruz), MCT4 (H-90, sc-503329, Santa Cruz), PDHA1 (459400, Life technologies), BASIGIN (Santa Cruz), panacetylated lysines of histone H4 (Cell signaling) and ACTIN (A3854, Sigma).

Quantitative RT-PCR.

Primary keratinocytes were lysed in TRIZOL reagent (Invitrogen), and total RNAs were isolated according to the manufacturer’s recommendations. cDNAs were synthesized from 500 ng of total RNA using random hexamers and SuperScript III Reverse transcriptase (Invitrogen). Quantitative real-time PCR (RT-qPCR) was performed on a LightCycler 480 SW 1.5 apparatus (Roche). 18s transcripts were used for normalization. Primers sequences were as follows:

E4f1: Fwd: 5′-CCAAAGCCTACCTGCTCAAG-3′;Rev: 5′-CTGGGCATTCTTGGTTTTGT-3.

Dlat: Fwd: 5′-TCCCTGCGCATCAGAAGGTT-3′; Rev: 5′-CCAACTGGAACATCTCTGGTC-3′.

Dld: Fwd: 5′-CCTTGTAGCTACGGGCTCAG-3′; Rev: 5′-CCCACATGACCCAAAAATTC-3′.

Brp44l/Mpc1: Fwd: 5′-TCCAGAGATTATCAGTGGGCGGAT-3′; Rev: 5′-GCCAGTTTCGAGGTTGTACCTTGT-3′.

Pdpr: Fwd: 5′-AAGACAAAGGACTAGCCCAGC-3′; Rev: 5′-GATAGGCCACGGATGTACCC-3′.

Pdha1: Fwd: 5′-GCTGGTTGCTTCCCGTAAT-3′; Rev: 5′-TAGTACTTGAGCCCATGCTCTC-3′.

Slc25a19: Fwd: 5′-TCAGTGTCAGGATTTGTCACCCGT-3′; Rev: 5′-AGAATGCTCTTGGGCCTTCCTCTT-3′.

Pdk1: Fwd: 5′-GCCAATCTAACAGGCAACTCT-3′; Rev: 5′-CATGAAGCAGTTCCTGGACTT-3′.

Pdk4: Fwd: 5′-CGCTTAGTGAACACTCCTTCG-3′; Rev: 5′-CGAACTTTGACCAGCGTGT-3′.

Pdp1: Fwd: 5′-CTGCTGTTCACCACCATACA-3′; Rev: 5′-GAAGCGTATCTCCTTCCTTGAG-3′.

Pdp2: Fwd: 5′-ACTGTGTCCTACTGGATCTTCAA-3′; Rev: 5′-CAGGTTCCTACTCGTGGCA-3′.

18s: Fwd: 5′-GTAACCCGT TGAACCCCATT-3′; Rev: 5′-CCATCCAATCGGTAGTAGCG-3′.

Zymography.

MMP activity was determined in murine keratinocytes lacking E4F1 or upon DLAT depletion. Two days after tam administration or viral tranduction, cells were washed twice with PBS and incubated for an additional 48 h in the same serum-free medium as the one used for clonogenic assays. Conditioned media from equal number of cells were collected and mixed with sample buffer [0.25 M Tris pH 6.8, 8% (wt/vol) SDS, 40% (vol/vol) glycerol, 0.096 mg/mL Bromophenol blue]. Samples were separated by electrophoresis in TG-SDS buffer (Euromedex) in a 10% (wt/vol) polyacrylamide gel containing 1 mg/mL gelatin (Sigma) at 4 °C. The gel was treated with 2.5% (vol/vol) Triton X-100 (Sigma) for 1 h at 4 °C and was then incubated with activity buffer (50 mM Tris pH 7.6, 5 mM CaCl₂, 0.02% Triton) for 24 h at 37 °C. The gel was stained with 40% (vol/vol) methanol, 10% (vol/vol) acetic acid, 0.1% Coomassie blue. For cathepsin zymography, gels were washed in renaturing buffer [65 mM Tris pH 7.4, 20% (vol/vol) glycerol] three times for 10 min and then incubated with activity buffer (0.1 M Na-P buffer pH 6, 2 mM DTT, 1 mM EDTA) for 30 min at room temperature and then overnight at 37 °C, as described previously (35). Samples were treated with the Lactate dehydrogenase inhibitor sodium-oxamate (0.5 mM; Sigma) for 4 d. For normalization, β-ACTIN expression was analyzed by immunoblotting in protein extracts prepared from the same cells.

ChIP.

To perform CTRL and E4F1 ChIP qPCR, primary keratinocytes from 18 to 20 newborn skins were pooled and cultured for 2 d. Cells were then fixed with PFA 1% for 8 min, then washed with cold PBS-Glycine 125 mM and incubated 5 min on ice with PBS-glycine 125 mM. Cells were then scrapped in PBS and washed extensively. Cells were lysed for 1 h on ice using lysis buffer (10 mM Tris pH 8, 140 mM NaCl, 0.1 SDS, 0.5% Triton X-100, 0.05% NaDoc, 1 mM EDTA, 0.5 mM EGTA, 1 mM pepstatinA, 1 mM aprotinin, 1 mM Leupeptin, and 0.1 mM PMSF). Chromatin sonication was performed with an Epishear Sonicator (Active Motif). An aliquot was decross-linked and deproteinized for DNA fragmentation size control. E4F1 ChIP was performed with 1 µL of rabbit polyclonal antibody raised against full-length E4F1 protein incubated in presence of 25 mg of keratinocytes chromatin and 20 µL of Dynabeads protein G. After 16-h incubation, immunoprecipitates are washed with the five following buffers. W1: Tris pH 8 10 mM, KCl 150 mM, Nonidet P-40 0.5%, EDTA 1 mM. W2: Tris pH 8 10 mM, NaCl 100 mM, NaDoc 0.1%, Triton X-100 0.5%. W3a: Tris pH 8 10 mM, NaCl 400 mM, NaDoc 0.1%, Triton X-100 0.5%. W3b: Tris pH 8 10 mM, NaCl 500 mM, NaDoc 0.1%, Triton X-100 0.5%. W4: Tris pH 8 10 mM, LiCl 250 mM, NaDoc 0.5%, Nonidet P-40 0.5%, EDTA 1 mM. W5: Tris pH 8 10 mM, EDTA 1 mM. Input and immunoprecipitated DNA were decross-linked over night at 65 °C, diluted in TE and incubated first with RNAseA (37 °C, 45 min) and then proteinase K (55 °C, 45 min). Proteins were removed with phenol-chloroform-isomalylic-alcohol and DNA was recovered by chromatography (nucleospin extract II columns, Macherey-Nagel). Immunoprecipitated DNAs were analyzed by QPCR (Roche LC480 and SYBRGreen mix) with the following promoter-specific primers:

• Dlat: 5′-ACAGACGCGCCACATTACTGC and 5′-GCTGCTCTTGGAGAGGTCACT;

• Dld: 5′-TACACACGACTCCAGCTCTGCAT and 5′ATAAGTCTTACCAGGCGTTCAGCG;

• Brp44l: 5′-ACACTTCTGGAGACTGAGGCTCTT and 5′ ACAGAGACGGTGAGATCCTGCAAA;

• Pdpr: 5′-TCTGAGGCTCCAGTGAACAATGCT and 5′- TAAGGCCTTTCAGTGCTTGGCTTG;

• Slc25a19: 5′-TGAAGTCTGCGCGGCTATGGAATA and 5′- TAATACCAGGCTTCCCGCCATCTT;

• NC2: (gene-poor noncoding region of chromosome 6): 5′- CCCCTTTCTGAAGCACTCTG and 5′- TAAGGCGTCATTTCCCAAAG;

• Pdha1: 5′-AGGAACATGTGGCCGTCCATTA and 5′- TTCACCACTTCTTCGCTGGTCTGT.

ECAR and Oxygen Consumption Rate.

ECAR and oxygen consumption rate (OCR) were measured using a XF24 Extracellular Flux Bioanalyzer (Seahorse Bioscience). Briefly, 25 × 10⁴ primary keratinocytes per well were seeded on a XF24 V7 cell culture microplate. Measurement was done in XF Assay medium (Seahorse Bioscience) pH 7.4 supplemented with 100 μM Sodium Pyruvate (Gibco). For measurements of ECAR, glucose (25 mM final, Sigma) was injected in the culture medium during the analysis. After the analysis, cells were lysed in Laemmli buffer [0.08 M Tris⋅HCI (pH = 6.8), 2% (wt/vol) SDS, 12% (wt/vol) sucrose] and dosed by Bradford protein assay.

For measurements of OCR, cells were incubated for 1 h at 37 °C in XF Assay medium containng Glutamax (Seahorse Bioscience) pH 7.4 supplemented with 4.5 g/L glucose, 1 mM pyruvate, 0.2 μM BSA-conjugated palmitate, and 0.5 mM carnitine. OCR was measured over 4 min in three measurement intervals to assess basal metabolic rate, oligomycin C (1 μg/mL final)-sensitive OCR associated to ATP production, and maximal respiratory capacity [upon FCCP (0.6 μM final) followed by rotenone (0.1 μM final) administration]. The ECAR and OCR data were normalized by the amount of protein for each sample.

NAD⁺/NADH Ratio.

The NAD⁺/NADH ratio was determined using NAD/NADH-Glo-Assay kit (Promega) following the manufacturer’s recommendations.

FAO.

FAO was measured in triplicates by quantifying the production of ³H₂O from [9,10-³H]palmitate. Briefly, cells were plated at 5 × 10⁴ cells per well in 12-well culture plates, and treated for 4 d with 4OHT. Cultured cells were washed three times with Dulbecco’s PBS and incubated with 200 μL of [9,10(n)-³H]palmitic acid (60 Ci/mmol, NEN) bound to fatty-acid-free albumin (final concentration 125 μM) containing 1 mM carnitine. After 2-h incubation at 37 °C, the mixture was removed and added to a tube containing 200 μL of cold 10% (wt/vol) TCA. The tubes were centrifuged for 10 min at 2,200 × g at 4 °C, and aliquots of supernatants (350 μL) were removed, mixed with 55 μL of 6 M NaOH, and applied to ion-exchange resin (DOWEX). The columns were washed twice with 750 μL of water, and the eluates were counted. Palmitate oxidation rates were expressed as counts per minute per cell.