Alkaline Ceramidase 1 Protects Mice from Premature Hair Loss by Maintaining the Homeostasis of Hair Follicle Stem Cells.

Ceramides and their metabolites are important for the homeostasis of the epidermis, but much remains unknown about the roles of specific pathways of ceramide metabolism in skin biology. With a mouse model deficient in the alkaline ceramidase (Acer1) gene, we demonstrate that ACER1 plays a key role in the homeostasis of the epidermis and its appendages by controlling the metabolism of ceramides. Loss of Acer1 elevated the levels of various ceramides and sphingoid bases in the skin and caused progressive hair loss in mice. Mechanistic studies revealed that loss of Acer1 widened follicular infundibulum and caused progressive loss of hair follicle stem cells (HFSCs) due to reduced survival and stemness. These results suggest that ACER1 plays a key role in maintaining the homeostasis of HFSCs, and thereby the hair follicle structure and function, by regulating the metabolism of ceramides in the epidermis.

Introduction

The mammalian epidermis is the largest organ, consisting of the interfollicular epidermis (IFE) with the pilosebaceous unit attached, including the hair follicle (HF) and sebaceous gland (SG). The homeostasis of the different compartments of the epidermis is maintained by distinct adult stem cell populations residing in specialized niches (Hsu et al., 2014). Stem cells in the basal layer of the IFE generate transit-amplifying cells, which, upon ceasing proliferation, migrate toward the skin surface to differentiate sequentially into the spinous layer, granular layer, and cornified layer (also known as stratum corneum) (Fuchs, 2007). The epidermal permeability barrier (EPB) is established in the stratum corneum to protect the body of a mammal from outside-in threatening allergens or pathogens while preventing inside-out water or body fluid loss. Growing hair from the epidermis is another important and protective function of the skin (Sotiropoulou and Blanpain, 2012). Hair grows out from and is maintained in the HF, which cycles through three distinct stages: the growing stage (anagen), regression stage (catagen), and resting stage (telogen) (Alonso and Fuchs, 2006). Hair growth and reactivation of the HF cycle are executed by stem cells residing in the specific bulge region in HFs. Stem cell homeostasis should be tightly regulated for proper HF cycling and hair growth (Sotiropoulou and Blanpain, 2012), because mutations of genes involved in stem cell homeostasis can cause hair loss, namely alopecia (Nakamura et al., 2013).

Ceramides, a group of sphingolipids, play a key role in maintaining the structure and function of the epidermis (Pappas, 2009). In addition, ceramides and their metabolites sphingosine and sphingosine-1-phosphate (S1P) can act as bioactive lipids to mediate various biological processes (Hannun and Obeid, 2008, Mao and Obeid, 2008). Ceramides are synthesized by the action of ceramide synthases encoded by six distinct genes (CERS1–6) in mammals (Levy and Futerman, 2010). Different CERS isoforms use acyl-CoAs of different chain lengths as substrates, so ceramide species differing in acyl-chain length are synthesized in mammalian cells (Grösch et al., 2012). Jennemann et al. (2012) demonstrate that knocking out the ceramide synthase 3 responsible for the biosynthesis of ultra-long-chain ceramides impairs the EPB, resulting in the death of newborns. Knocking out ceramide synthase 4 responsible for the biosynthesis of long-chain (LC) ceramides reduces the hair follicle stem cell (HFSC) population and results in hair loss in mice without affecting the EPB (Ebel et al., 2014, Peters et al., 2015). These animal studies strongly suggest that different ceramides play important and distinct roles in maintaining the homeostasis of different compartments of the epidermis.

Once synthesized, ceramides can be hydrolyzed into sphingosine (SPH) by the action of ceramidases encoded by five distinct genes, the acid ceramidase (ASAH1/Asah1), neutral ceramidase (ASAH2/Asah2), alkaline ceramidase 1 (ACER1/Acer1), alkaline ceramidase 2 (ACER2/Acer2), and alkaline ceramidase 3 (ACER3/Acer3) (Mao and Obeid, 2008). ACER1 is a member of the alkaline ceramidase family that was identified by our group initially from the yeast Saccharomyces cerevisiae and then from mammals. Acer1 is predominantly expressed in the skin (Houben et al., 2006, Mao et al., 2003, Sun et al., 2008). Our previous studies found that Acer1 is highly expressed in differentiating (spinous layer) and differentiated epidermal keratinocytes (granular layer), but its expression is low in proliferative keratinocytes in the basal layer (Houben et al., 2006), indicating that ACER1 may have a role in epidermal differentiation. Indeed, with in vitro cell-culture systems, we demonstrate that ACER1 may mediate growth arrest and differentiation of human epidermal keratinocytes by regulating the hydrolysis of cellular ceramides (Sun et al., 2008). However, its physiological roles in regulating the metabolism of epidermal ceramides and the homeostasis of the epidermis remain unclear.

In this study, using mouse genetic approaches in combination with biochemical analyses, we demonstrate that ACER1, the mouse ortholog of human ACER1, is expressed in the IFE, SG, and HF and plays a key role in maintaining the homeostasis of these epidermal compartments. We reveal that Acer1 deficiency increases the levels of ceramides and SPH in the skin and progressively depletes HFSCs, thereby resulting in progressive alopecia. These results suggest that ACER1 plays a key role in regulating the homeostasis of the epidermis by controlling the catabolism of ceramides in epidermal compartments.

Results

Loss of Acer1 Decreases Alkaline Ceramidase Activity in Epidermis and Leads to Progressive Alopecia

To study the role of ACER1 in regulating the homeostasis of the epidermis, we generated an Acer1 knockout mouse line by replacing the entire coding region of the Acer1 gene with the reporter/selection cassette containing the LacZ and Neo genes (Figure 1A). Interbreeding mice heterozygous for the Acer1 null allele, Acer1 mice, gave rise to wild-type (WT) mice (Acer1), heterozygous Acer1, and mice homozygous for the Acer1 null allele (Acer1) at the expected Mendelian ratio (+/+:+/−:−/−, 46:107:47), suggesting the absence of significant embryonic lethality. RT-qPCR showed that no coding sequence of the Acer1 gene is transcribed in the skin tissues of Acer1 mice on postnatal day 4 (P4) (Figure 1B). Consistently, alkaline ceramidase activity on the very-long-acyl-chain (VLC) ceramide C24:1 ceramide (C24:1-Cer), a preferred substrate of ACER1 (Sun et al., 2008), was markedly decreased in the skin tissues of P4 Acer1 mice, compared with Acer1 littermates (Figure 1C). In contrast, Acer1 knockout only moderately decreased skin alkaline ceramidase activity on the LC ceramide C18:1 ceramide (Figure 1C). These results suggest that Acer1 encodes major and minor skin alkaline ceramidase activity on VLC ceramide and LC ceramide, respectively.

Figure 1

Knockout of Acer1 in C57BL/6 Mice Induces Progressive Alopecia

(A) The Acer1 gene targeting strategy. The Acer1 ORF was replaced by the gene cassette containing the LacZ and Neo genes. ATG, the translation start codon; TGA, the translation stop codon; P1, the primer pair for the LacZ gene; P2, the PCR primer pair for Exon3/4 of the Acer1 gene.

(B) Acer1 mRNA levels in Acer1 and Acer1 mouse skin tissues. n = 3 individual mice.

(C) ACER1 enzymatic activity on C24:1 ceramide or C18:1 ceramide in Acer1 and Acer1 mouse skin tissues. n = 3 individual mice.

(D) Female mice of different ages and genotypes.

(E) The length (in days) of the first anagen (gray/black) of the hair growth cycle among Acer1, Acer1, and Acer1 mice. N.s., not significant.

(F) The images of dorsal skins taken from P122 Acer1 and Acer1 mice from day 1 (D1) to day 60 (D60) post depilation by waxing.

(G) The length of telogen (pink) and anagen (dark gray/black) following depilation on P122 mice.

The error bars indicate the SD.

To assess the role of ACER1 in skin homeostasis, we followed the gross morphological features of Acer1, Acer1, and Acer1 C57BL/6J mice up to 18 months of age. Acer1 mice showed no difference in body size (Figure 1D) and body weight (Figure S1A) from age- and sex-matched Acer1 or Acer1 littermates at young ages (1–7 months old; Figure S1A). However, at middle ages (13 and 18 months old; Figure S1A), Acer1 mice had lower body weight than Acer1 and Acer1 mice, which showed no difference in body weight. Necropsy showed that Acer1 mice were similar with regard to anatomy and size of major organs, including heart, brain, liver, lungs, kidneys, and spleens, compared with Acer1 or Acer1 mice (data not shown). Intercrossing of Acer1 mice gave rise to a similar litter number and size to the intercross of Acer1 mice (Figure S1B). These results suggest that Acer1 knockout does not affect mouse development, fertility, and general health. However, in contrast to Acer1 and Acer1 mice, Acer1 mice developed progressive hair loss starting at 2–3 months of age (Figure 1D), suggesting that ACER1 is essential for the hair growth and/or maintenance.

Loss of Acer1 Disturbs Hair Follicle Cycling

Since a disruption of HF cycling has been implicated in progressive hair loss (Courtois et al., 1994), we examined if Acer1 deficiency affects HF cycling. By tracing the cyclic changes of the dorsal skin color in mice, we found that there was no difference in hair cycling among Acer1, Acer1, and Acer1 mice up to 2 months of age (Figure 1E). Since the hair loss phenotype becomes more discernible in Acer1 mice at 3–4 months of age or older (Figure 1D), we investigated if loss of Acer1 affects HF cycling in mice at these ages. Although in rodents, synchronized HF growth occurs in waves that sweep dorsally, as the mouse ages, the waves occur less frequently and only in small patches (Stenn and Paus, 2001). This makes it difficult to evaluate the effect of Acer1 deficiency on HF cycling in mice older than 2 months. To this end, we depilated hair from the dorsal skin in P122 mice using a waxing technique (Muller-Rover et al., 2001) to synchronize hair cycling. After waxing-induced synchronization, we found that the transition of the first telogen to anagen was significantly delayed in Acer1 mice compared with Acer1 mice (Figures 1F, 1G, and S2). However, after the first cycle, both telogen and anagen were much shorter in Acer1 mice compared with Acer1 mice (Figures 1F, 1G, and S2), suggesting that loss of Acer1 accelerates HF cycling.

Acer1 Is Expressed in Multiple Skin Compartments

In Acer1 or Acer1 mice, the promoter-less lacZ reporter gene is placed downstream of the Acer1 promoter, so the expression of the LacZ gene is under the control of the endogenous Acer1 promoter. This allowed us to examine the expression of Acer1 by following lacZ gene-encoded β-galactosidase activity, which can be detected by in situ X-gal staining. We found X-gal stain in the IFE, HF, and SG in both P6 (Figure 2A) and P13 Acer1 mice (Figure 2B). P56 Acer1 mice exhibited a similar X-gal staining pattern to P13 Acer1 mice (Figure S3A). These compartment-specific expression patterns were confirmed by immunofluorescent staining (IF) with antibodies against compartment-specific markers, including the IFE basal layer marker K14 and the SG marker fatty acid synthase (FAS) (Figures 2C and 2D). X-gal stain was absent in the bladder, brain, colon, heart, kidney, liver, lung, pancreas, small intestine, spleen, stomach, seminal vesicle, and testis from P56 Acer1 mice (Figure S3A) and P56 Acer1 skin (Figure S3B). These results suggest that ACER1 is a skin-specific ceramidase that is highly expressed in all the major compartments of the epidermis.

Figure 2

Acer1 Is Expressed in Multiple Compartments of the Mouse Skin

(A and B) X-gal staining of cryosections of the skin of P6 (A) and P13 (B) Acer1 mice. The X-gal signals are shown in blue, and tissue sections were counterstained with eosin (A) or nuclear fast red (B).

(C and D) Two consecutive skin cryosections from a P13 Acer1 mouse were stained with X-gal (the left panel) or antibody specific for K14 (C) or FAS (D), and counterstained for nuclei with DAPI (blue).

Scale bar, 100 μm.

Loss of Acer1 Alters Skin Sphingolipids

To determine the role of ACER1 in regulating the metabolism of sphingolipids in the skin, we compared the levels of various types of sphingolipids in skin tissues in adult Acer1 and Acer1 mice. In P33 adult mouse skin, loss of Acer1 increased the levels of ceramides carrying saturated LC or VLC (C16:0, C20:0, C22:0, C24:0, and C26:0) and total ceramides but not ceramides carrying unsaturated VLCs (C24:1 or C26:1) (Figure 3A) or dihydroceramides (Figure 3B). Loss of Acer1 also substantially increased the levels of various phytoceramides (Figure 3C) and α-hydroxyceramide species (Figure 3D) in the adult skin. Interestingly, loss of Acer1 did not affect the levels of the complex sphingolipids, mono-hexosylceramides (Figure 3E) or sphingomyelins (Figure 3F) in the adult skin. Unexpectedly, loss of Acer1 markedly increased the levels of sphingoid bases, including sphingosine (SPH), dihydrosphingosine (DHS), and phytosphingosine (PHS) (Figure 3G) without affecting their phosphates in the adult skin (Figure 3H).

Figure 3

Loss of Acer1 Alters Skin Sphingolipids in Adult Mice

(A–H) Skin tissues isolated from P33 Acer1 and Acer1 mice were analyzed by LC-MS/MS (n = 4 individual mice) for (A) ceramides, (B) dihydroceramides, (C) phytoceramides, (D) α-hydroxy-ceramides, (E) monohexosyl-ceramides, (F) sphingomyelins, (G) sphingoid bases, and (H) their phosphates.

(I–K) qPCR results showing mRNA levels of other ceramidases (I), Sgpp2 (J), and Sphk1 and SphK2 (K) and between P33 Acer1 and Acer1 mice. n = 3 individual mice.

The error bars indicate the SD.

To investigate whether the unexpected increases in sphingoid bases resulted from compensatory changes in the expression of other sphingolipid-metabolizing enzymes, we performed a qPCR array that simultaneously quantifies mRNA levels of major enzymes involved in the metabolism of sphingolipids. The results demonstrated that Acer1 loss altered mRNA levels of multiple enzymes (Figure S4). With qPCR for analyzing individual genes, we confirmed that Acer1 loss indeed increased the mRNA levels of Acer2, Acer3 (Figure 3I), and S1P phosphatase 2 (Sgpp2) (Figure 3J) while decreasing the mRNA levels of acid ceramidase (Asah1), neutral ceramidase (Asah2) (Figure 3I), sphingosine kinase 1 (Sphk1), and sphingosine kinase 2 (Sphk2) (Figure 3K). These results may explain why Acer1 loss increases rather than decreases its products. Taken together, these results suggest that ACER1 regulates the levels of various types of ceramides and sphingoid bases in the mouse skin.

Loss of Acer1 Promotes Both Cell Proliferation and Apoptosis in the Epidermis

Increasing numbers of studies demonstrate that ceramides with different acyl chains might have distinct roles in mediating biological responses. VLC ceramides, such as C24 and C24:1 ceramides, have been shown to promote cell proliferation and cell survival whereas LC ceramides, such as C16 and C18 ceramides, exert the opposite effects (Grösch et al., 2012). In addition, SPH has been shown to inhibit cell proliferation and to induce cell differentiation or apoptosis (Cuvillier, 2002, Hu et al., 2010, Xu et al., 2006, Xu et al., 2016). Since loss of Acer1 altered the levels of various ceramides and SPH in mouse skin, we investigated whether loss of Acer1 affected cell proliferation, differentiation, and apoptosis in different epidermal compartments. Immunohistochemistry (IHC) revealed that compared with P33 Acer1 mice, their Acer1 littermates had more cells stained positively for Ki-67, a proliferation marker, in the IFE basal layer, HF, and SG (Figures 4A and 4B). Consistently, western blot analyses showed that compared with Acer1 mice, Acer1 mice had increased expression of PCNA (another proliferation marker), cytokeratin 14 (K14; a marker of basal layer keratinocytes in the IFE, HF, and SG), and FAS (an SG marker) and decreased expression of K10 and involucrin, differentiation markers for the granular layer and spinous layer, respectively, in the epidermis (Figure S5A). qPCR analyses found that compared with their Acer1 littermates, P33 Acer1 mice had decreased mRNA levels of K10 and involucrin in the epidermis (Figure S5B). Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays found more apoptotic cells in the Acer1 epidermis, especially in the HF infundibulum region (Figures 4C and 4D). Taken together, these results suggest that Acer1 deficiency increases both cell proliferation and apoptosis while decreasing cell differentiation in the epidermis, suggesting a role in regulating the homeostasis of the major epidermal compartments.

Figure 4

Loss of Acer1 Increases the Proliferation and Apoptosis of Epidermal Keratinocytes while Dilating Pilary Canal

(A) IHC of skin tissue sections with anti-Ki-67 antibody.

(B) Quantification of Ki-67-positive cells in different compartments in the tissue sections from (A). n = 4 individual mice.

(C) Skin tissue sections from P64 Acer1 and Acer1 mice were stained with a TUNEL (green) assay kit and counterstained for nuclei with DAPI (blue). The yellowed dash lines separate the epidermis (upper) and dermis (lower) and outline the hair follicle.

(D) Quantification of apoptotic EKs from (C) by the value of TUNEL-positive cell versus area (mm2). n = 3 individual mice.

(E) H&E staining of skins from P35 Acer, Acer1, and Acer1 mice.

Note abnormally dilated pilary canals marked with asterisks and lipid accumulation in pilary canals marked with stars in Acer1 mice. The error bars indicate the SD. Scale bar, 100 μm.

ACER1 Alters the Morphology of the Hair Follicle Infundibulum and Sebaceous Gland but Not the Number of Hair Follicles

Acer1 mice exhibited a progressive alopecia phenotype (Figure 1D), suggesting a role for ACER1 in regulating hair growth or maintenance. As hair growth and maintenance depend on the homeostasis of the HF, we investigated whether Acer1 deficiency reduced the number of HFs and/or altered their morphology. H&E staining showed that the pilary canal in the HF infundibulum region was wider and morphologically irregular in Acer1 mice compared with that in Acer1 mice (Figure 4E), although the HF numbers were similar in these mice (Figure S5C).

Because Acer1 is highly expressed in the SG and an abnormality in the SGs could also lead to hair loss, we investigated if loss of Acer1 affected the function or morphology of the SG. We performed oil red O (ORO) staining to monitor lipid secretion from SG. As shown in Figure S5D, loss of Acer1 reduced ORO staining on the surface of the skin while enhancing ORO staining in the HF pilary canal, suggesting that Acer1 deficiency may impair lipid secretion from the SG to the skin surface. The accumulation of lipids in the opening of the pilary canal may be responsible for the dilation of the HF infundibulum in Acer1 mice. Taken together, these results suggest that loss of Acer1 alters the morphology of the HF and disrupts the lipid secretory function of the SG.

Loss of Acer1 Progressively Depletes the Population of Hair Follicle Stem Cells

The initiation of HF cycling involves HFSC reactivation in transition between telogen and anagen (Greco et al., 2009). Since we observed a delay in the telogen-to-anagen transition after depilation in Acer1 mice (Figures 1F and 1G), we postulated that Acer1 loss might affect the homeostasis of HFSCs. To this end, we determined the HFSC population by measuring the number of label-retaining cells (LRC) or cells expressing CD34, a mouse quiescent HFSC marker. P3 newborns were injected with 5-ethynyl-2′-deoxyuridine (EdU) and chased for 4–6 weeks. EdU-retaining cells (EdU+ LRC) were visualized and quantified using a Click-iT EdU immunofluorescence staining kit (Life Technologies). As shown in Figure 5A, we found that the number of LRCs was significantly reduced in Acer1 mice compared with Acer1 mice. Consistently, IHC showed that Acer1 mice had fewer CD34-positive cells in the bulge region of the HF compared with Acer1 mice (Figure 5B). Furthermore, IHC staining of skin tissues collected from mice at different ages showed an age-dependent reduction of CD34-postive cells (Figure 5C). To quantify the number of CD34-positive HFSCs more accurately, we performed flow cytometric analyses of HFSCs co-stained with antibodies against CD34 and integrin alpha 6 (α6). As shown in Figure 5D, the number of CD34/α6 double-positive cells were decreased in Acer1 mice in an age-dependent manner. Taken together, these results suggest that Acer1 deficiency progressively reduces the HFSC population.

Figure 5

Loss of Acer1 Progressively Depletes the Population of Quiescent HFSCs

(A) Left: LRCs in the HF from P50 Acer1 and Acer1 mice. The left panels, EdU; the middle panel, DAPI; and the right panel, EdU/DAPI merged. White dashed lines outline the HF and IFE. Right: Quantification of EdU-positive LRCs in the HF region. n = 3 individual mice. Scale bar, 50 μm.

(B) IHC of skin sections from P92 Acer1 and Acer1 mice with an antibody against the bulge stem cell marker CD34. Scale bar, 50 μm.

(C) Quantification of percentages of CD34-positive HF from different ages (P52, P92, and P333). n = 3 individual mice.

(D) Left: FACS sorting of FITC-CD34 (x axis) and PE-α6 (y axis) double-positive cells from P25 and P122 Acer1 and Acer1 mice. Right: Quantification of percentages of CD34/α6 double-positive cells. n = 3 individual mice.

(E) qPCR analysis of Acer1 on primary cultured P25 Acer1 and Acer1 HFSCs.

(F) Left: Colony formation assays on primary Acer1 and Acer1 HFSCs cultured for 2 weeks. Scale bar, 1 mm. Right: Quantification of (F) by counting the number of colonies per 100 mm2 and the average size of colonies (mm2). n = 3 individual mice.

(G) FACS of CD34/α6 double-positive cells from HFSCs cultured in 3C media-Matrigel.

(H) qPCR analysis of HFSC quiescence markers (CD34 and Lhx2) and HFSC activation signature gene Wnt10b on primary cultured Acer1 and Acer1 HFSCs. n = 3 individual mice.

The error bars indicate the SD.

Loss of Acer1 Reduces Hair Follicle Stem Cell Survival and Quiescence

To test if loss of Acer1 affects the characteristics of HFSCs such as self-renewal and quiescence in a cell-autonomous mechanism, we performed colony formation assays with HFSCs isolated from Acer1 or Acer1 mice. HFSCs were isolated from P25 Acer1 and Acer1 mice in telogen, labeled with anti-CD34 and anti-a6 integrin antibodies, and sorted by fluorescence-activated cell sorting (FACS) as described in the Experimental Procedures. qPCR analyses confirmed that Acer1 mRNA is expressed in Acer1 HFSCs but not in Acer1 HFSCs (Figure 5E). When the same number of viable Acer1 or Acer1 HFSCs were cultured over a monolayer of Swiss3T3 feeder cells, we found that Acer1 HFSCs formed fewer but larger colonies compared with Acer1 HFSCs (Figure 5F), indicating that loss of Acer1 reduces HFSC survival or the colony formation ability (stemness) while enhancing their proliferation at least in part through a cell-autonomous mechanism.

To test whether loss of Acer1 reduces the long-term maintenance of HFSCs, we cultured Acer1 or Acer1 HFSCs for three passages in a feeder-free 3C-Matrigel medium that maintains HFSC stemness much longer than the traditional 2D-feeder cell culture (Chacón Martínez et al., 2017). This feeder-free culture condition allowed us to assess the percentage of cultured cells that retain the stem cell markers without contamination of feeder cells. As shown in Figure 5G, flow cytometry revealed that the CD34/integrin α6 double-positive cell population was much lower in the Acer1 HFSC culture than in the Acer1 HFSC culture, suggesting that loss of Acer1 may reduce HFSC stemness in addition to HFSC survival.

As we observed that the colony size of Acer1 HFSCs is larger than that of Acer1 HFSCs (Figure 5F), we wanted to test whether loss of Acer1 accelerates HSFC activation. To this end, we measured the mRNA levels of HFSC quiescent and activation signature genes in Acer1 and Acer1 HFSCs. With qPCR, we demonstrated that loss of Acer1 reduced the mRNA levels of the quiescent markers CD34 and Lhx2 (Figure 5H) and elevated the mRNA levels of the activation signature gene Wnt10b (Figure 5H), suggesting that loss of Acer1 reduces the ability of HFSCs to maintain quiescence.

Discussion

Our group previously demonstrated that both the mouse ACER1 and its human counterpart are predominantly expressed in the skin (Mao et al., 2003, Sun et al., 2008), but neither the skin compartment-specific expression of ACER1 nor its physiological role in skin homeostasis is clear. In this study, we generated a mouse strain in which the endogenous mouse Acer1 gene was replaced with the bacterial LacZ gene so that we could study the tissue-specific expression of Acer1 and its physiological role in the skin simultaneously. We confirmed that Acer1 is predominantly expressed in the skin and further revealed that Acer1 is highly expressed in different epidermal compartments (the IFE, HF, and SG) (Figures 2 and S3). We further demonstrated that ACER1 plays a key role in regulating the homeostasis of the epidermis, the HF in particular, by controlling the metabolism of ceramides and their metabolites, sphingoid bases, in major epidermal compartments.

Our group previously demonstrated that ACER1 catalyzes the hydrolysis of VLC ceramides in vitro (Sun et al., 2008). In this study, we demonstrated that knocking out Acer1 decreased skin alkaline ceramidase activity on the VLC ceramide, C24:1 ceramide, more significantly than that on the LC ceramide, C18:1 ceramide, indicating that similar to human ACER1, mouse ACER1 may prefer VLC ceramides to LC ceramides as substrates. Loss of Acer1 caused a marked increase in the levels of VLC and LC ceramides, phytoceramides, and α-hydroxyceramides in the skin, indicating that ACER1 catalyzes the hydrolysis of various ceramide species in the skin although its preferred substrates are VLC ceramides. Interestingly, loss of Acer1 also markedly increased the levels of free sphingoid bases (SPH, DHS, and PHS) without affecting sphingoid base phosphates. This result was not expected because SPH, as a product of ACER1, should be increased in Acer1 mice. With the sphingolipid pathway-specific qPCR array that simultaneously analyzes mRNA levels of enzymes known to be involved in the metabolism of sphingolipids, we revealed that loss of Acer1 upregulated the SPH-generating enzymes ACER2, ACER3, and SGPP2 (Figures 3I and 3J) while downregulating the enzymes SPHK1 and SPHK2, which convert SPH to its phosphate (Figure 3K). The changes in the expression of these enzymes may be responsible for the unexpected increase in the sphingoid base, but why and how these changes occur remain unclear. Taken together, these results suggest that ACER1 plays a key role, directly and by altering the expression of other enzymes, in regulating the homeostasis of sphingolipids in the skin.

Ceramides have been shown to act as bioactive lipids to regulate various biological processes (Hannun and Obeid, 2008). It was thought that all ceramides had the same role in cellular responses. However, emerging evidence indicates that different ceramide species may have different or even opposing roles in cellular processes. LC ceramides (C16 ceramide and C18 ceramide) have anti-proliferative effects such as cell-cycle arrest, senescence, and apoptosis (Saddoughi and Ogretmen, 2013), whereas VLC ceramides (C22, C24, and C24:1 ceramide) have been shown to promote cell proliferation and cell survival (Grösch et al., 2012). Like ceramides, SPH, and other free sphingoid bases have been shown to act as signaling molecules to mediate cell growth arrest, autophagy, and apoptosis (Cuvillier, 2002, Cuvillier et al., 2001, Hu et al., 2005, Kim et al., 2004, Nava et al., 2000, Vogler et al., 2003, Xu et al., 2006, Xu et al., 2016, Yi et al., 2016). These results indicate that as a key enzyme in regulating ceramides and sphingoid bases in the epidermis, loss of Acer1 is expected to trigger profound effects on the biology of the epidermis.

Indeed, in this study, we demonstrated that loss of Acer1 exerts effects on cell proliferation, differentiation, and survival in epidermal compartments. We demonstrated that loss of Acer1 inhibited the expression of various keratinocyte differentiation markers, such as cytokeratin 1 (K1) and cytokeratin 10 (K10), suggesting a role for ACER1 in keratinocyte differentiation. Consistently, in our previous study, we found that knocking down human ACER1 also inhibited Ca2+-induced differentiation of epidermal keratinocytes in culture (Sun et al., 2008). As Acer1 knockout and Acer1 knockdown have a similar effect on ceramides but an opposing effect on SPH, we speculated that ACER1 mediates keratinocyte differentiation likely through ceramides rather than SPH and other sphingoid bases. Although loss of Acer1 significantly decreases K1 and K10, Acer1 mice appear to have a normal structure of the IFE as Acer1 mice, suggesting that loss of Acer1 may have a mild effect on epidermal differentiation. In this study, we also found that loss of Acer1 enhanced the proliferation of cells in different compartments of the epidermis. The proliferative effect could be caused by a marked increase in the levels of various VLC ceramides, phyotoceramides, and/or α-hydroxyceramides. Unexpectedly, loss of Acer1 also increased apoptosis of cells in different compartments of the epidermis (Figures 4C and 4D), although our previous study found that knocking down the human Acer1 did not lead to apoptosis of epidermal keratinocytes in culture. As loss of Acer1 accumulated both LC ceramides and free sphingoid bases in the epidermis, whereas Acer1 knockdown only accumulated VLC ceramides, we postulated that loss of Acer1 led to apoptosis of cells in the epidermis likely due to accumulation of LC ceramides and/or free sphingoid bases. In agreement with this notion, Uchida et al. (2010) demonstrated that blocking the acid ceramidase (ASAH1) and neutral ceramidase (ASAH2) enhances both the generation of LC ceramides and the apoptosis of epidermal keratinocytes in response to UV radiation (UVR), and Tolleson et al. (1999) found that treatment with the ceramide synthase inhibitor fumonisin B1 (FB1) accumulated free DHS in epidermal keratinocytes and led to the death of these cells.

Emerging evidence suggests that the metabolism of ceramides in the epidermis plays a key role in hair growth and maintenance (Ebel et al., 2014, Peters et al., 2015). Loss of Cers4, which is responsible for the synthesis of LC ceramides (C18-20 ceramides), results in age-related hair loss in mice. Interestingly, in this study, we demonstrated that loss of Acer1, responsible for the hydrolysis of epidermal ceramides, also led to progressive hair loss in mice. Although Cers4 catalyzes a different reaction from ACER1, loss of Cers4 causes similar effects on sphingolipids as loss of Acer1, accumulating various ceramide species (C16:0-, C24:0-Cer, C22:0-, and C24:0-hydroxy-ceramide) while reducing the levels of its substrates C18-20 ceramides in the skin (Peters et al., 2015). The similar changes in these ceramide species may explain why mice deficient in Acer1 and Cers4 have a similar hair loss phenotype and suggest that accumulated ceramides are indeed responsible for hair loss in mice.

We demonstrated that loss of Acer1 caused the hypertrophy of SG and accumulation of lipids in the HF infundibulum while decreasing lipids on the surface of the epidermis. Lipid accumulation in the HF infundibulum may cause abnormality in the structure of the HF. It has been proposed that age-related accumulation of lipids in the HF infundibulum impairs the maintenance of the hair shaft and thereby leads to progressive alopecia (Ebel et al., 2014). These results suggest that loss of Acer1 induces progressive hair loss in part due to impairments in both the structure and function of the SG. We found that loss of Acer1 also progressively decreased the HFSC population in mice, indicating that ACER1 is critical for the homeostasis of the HFSC. Taken together, these results suggest that loss of Acer1 causes alopecia in mice by breaching both the lipid secretory function of the SG and the homeostasis of HFSCs.

Our results indicate that loss of Acer1 leads to HFSC depletion probably due to a decrease in stem cell survival and/or quiescence. We found that loss of Acer1 altered HF cycling by increasing the length of the resetting phase while shortening the following phases (Figures 1F and 1G), indicating that loss of Acer1 accelerates hair cycling. Similar to loss of Acer1, Ebel et al. (2014) showed that loss of Cers4 also accelerates HF cycling. These results suggest that increased ceramides in the epidermis may accelerate HF cycling. There are many other mouse genes whose inactivation accelerates HF cycling and HFSC depletion, resulting in hair loss (Nakajima et al., 2013, Oda et al., 2012). These results suggest that loss of Acer1 causes HFSC depletion in part by accelerating hair cycling. Accelerated hair cycling in Acer mice may be caused by loss of quiescence or increased activation of HFSCs. In line with this notion, we demonstrated that when cultured with feeder cells in vitro, Acer1 HFSCs formed large colonies and lost HSFC markers more quickly than Acer1 HFSCs (Figure 5E). Cell counting showed that Acer1 HFSCs form larger colonies because of an increase in cell proliferation (Figure S5A). These results suggest that loss of Acer1 indeed reduces the ability of HFSCs to main quiescence. In addition to the difference in colony size, we also found that fewer colonies were formed from Acer1 HFSCs than from Acer1 HFSCs when the same number of these cells was plated on a feeder layer (Figure 5E), suggesting that loss of Acer1 may reduce HFSC survival as well. In addition to the cell-autonomous effects on HFSCs, Acer1 deficiency may deplete HFSCs also through a non-cell-autonomous mechanism because Acer1 is highly expressed in different cell types in the hair follicle. Taken together, these results suggest that loss of Acer1 impairs the homeostasis of HFSCs by enhancing their proliferation and inhibiting their survival and quiescence through both cell and non-cell-autonomous mechanisms.

Loss of Acer1 reduces HFSC quiescence or increases HFSC activation probably by regulating the Lhx2 and/or Wnt signaling pathway. With qPCR analyses, we found that loss of Acer1 reduced the mRNA levels of the transcription factor Lhx2 (LIM homeobox 2), which has been implicated in the maintenance of quiescence of HFSCs (Rhee et al., 2006). Similar to loss of Acer1, loss of Lhx2 significantly reduces CD34 (Folgueras et al., 2013) and the LRC population, increases cell proliferation in the epidermis (Rhee et al., 2006), disrupts HF architecture, and causes baldness (Folgueras et al., 2013). Interestingly, loss of Acer1 did not reduce the expression of the Sox9, Nfatc1, and Lgr5 in HFSCs (Figure S6B), although these genes are known downstream targets of Lhx2 (Kim and Kim, 2014, Kim et al., 2014, Mardaryev et al., 2011), suggesting that loss of Acer1 reduces HFSC quiescence through an unknown target(s) of Lhx2. In contrast to Lhx2, loss of Acer1 increased the mRNA levels of Wnt10b in HFSCs (Figure 5G). Wnt10b serves a signature gene of transit-amplifying cells and plays an important role in stem cell quiescence (Lim et al., 2016) and anagen activation (Li et al., 2013). These results suggest that loss of Acer1 may reduce the quiescence of HFSCs in part by inhibiting Lhx2 and/or activating the Wnt pathway.

While we were preparing this manuscript, Liakath Ali et al. (2016) reported similar findings to ours regarding the hair loss phenotype in a different knockout strain. They reported several SG abnormalities in Acer1 mice, such as a different adipocyte arrangement, lipid droplet structure, and nuclear shape (Liakath Ali et al., 2016), thus explaining why loss of Acer1 impairs the secretion of neutral lipids to the skin surface via the HF infundibulum. In addition, Liakath Ali et al. (2016) showed increased trans-epidermal water loss and hyper-metabolism in Acer1 mice, suggesting alterations in energy balance and thermal insulation. This may explain why loss of Acer1 causes body weight loss in middle-aged or older mice, another important phenotypic change observed by both studies. However, there is one major discrepancy between our studies and theirs. We demonstrated that loss of Acer1 increased the levels of free sphingoid bases without affecting the levels of sphingoid base phosphates or complex sphingolipids (monohexosylceramides and sphingomyelins) in the skin, while Liakath Ali et al. (2016) showed that there were significant decreases in the levels of sphingoid bases and their phosphates and increases in the levels of complex sphingolipids in the skins of Acer1 mice compared with Acer1 mice (Figure 3). These differences do not appear to be caused by the use of different mouse strains because we showed that loss of Acer1 caused the same changes in the sphingolipid profile in mice with an FVB background (our unpublished data). These differences could be due to the ways in which skin tissues were prepared for lipid analyses. We analyzed sphingolipids in the whole skin, which encompasses different epidermal compartments where Acer1 is expressed, whereas Liakath Ali et al. (2016) did so only in the stratum corneum where no Acer1 is expressed. In addition, we but not Liakath Ali et al. (2016) revealed the role of ACER1 in regulating the homeostasis of HFSCs and made significant progress in understanding the mechanism by which ACER1 maintains the homeostasis of HFSCs.

In conclusion, we demonstrated that the skin-specific ceramidase ACER1 regulates the metabolism of ceramides and sphingoid bases in the skin as well as homeostasis of the SG and HF so loss of Acer1 results in progressive HFSC depletion and alopecia (Figure 6). Along with other previous studies, our study suggests that tight regulation of the metabolism of ceramides and free sphingoid bases is critical for the homeostasis of HFSCs, the epidermis, and its appendages.

Figure 6

Model Describing the Role of ACER1 in the Homeostasis of the Different Compartments of the Epidermis

This figure depicts the structural and functional differences between the epidermis from (A) Acer1/Acer1 and (B) Acer1 mice. Acer1 is expressed in the IFE, SG, and bulge regions in Acer1/Acer1 mice (A), and upon loss of Acer1, both ceramides (Cer) and sphingoid bases (SPH) are accumulated in these epidermal compartments in Acer1 mice. Accumulated sphingolipids lead to the hypertrophy of IFE and SG, widening of the pilary canal (gray dashed double-headed arrows), the accumulation of lipids (red), and the depletion of HFSCs. Purple circles represent quiescent stem cells and cyan color represents Acer1 expression.

Experimental Procedures

Animals

All mice used in this study were housed under conventional laboratory conditions at a constant room temperature (RT, 22°C), humidity level (55%), 12-hr light/12-hr dark cycle, and food (W.F. Fisher & Son) and water available ad libitum. The generation of mice heterozygous for the Acer1 null allele, Acer1 mice, was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of California at Davis (UC Davis). All other animal studies were performed following procedures approved by the IACUC at Stony Brook University (Stony Brook, NY).

Acer1 Null C57BL/6J Mice

A gene-trap embryonic stem cell clone (18475A-H9) in which one copy of the whole open-reading frame (ORF) of the mouse Acer1 gene was replaced by a gene cassette containing the reporter gene LacZ and the neomycin resistant gene (Neo) (Figure 1A) was created by Regeneron Pharmaceuticals using VelociGene technology as described (Valenzuela et al., 2003). This embryonic stem cell clone was made into live mice by the Repository Knockout Mouse Project (KOMP) at the University of California at Davis. Briefly, embryonic stem cells expanded from the 18475A-H9 clone were injected into C57BL/6J blastocysts, which were transplanted into the uteri of pseudopregnant C57BL/6J mice. The resulting chimeric mice were crossed to WT C57BL/6J mice to generate offspring heterozygous for the Acer1 null allele, Acer1 mice, which were inbred to generate Acer1 null (Acer1) mice and their Acer1 and Acer1 littermates.

Statistics

Statistical analyses were performed using an unpaired two-tailed Student's t test. The p values for significance are stated in the figures (∗p > 0.05; ∗∗p > 0.01; ∗∗∗p > 0.001). The error bars indicate the SD.

qPCR

Skin RNAs were extracted, reverse transcribed into cDNAs (Wang et al., 2015), and subjected to qPCR analyses using the primer pairs listed in Table S1. qPCR was performed and analyzed with the ABI Prism 7000 system (Applied Biosystems) using the ΔΔCt method with β-Actin as an internal control as described (Mao et al., 2003).

Alkaline Ceramidase Activity Assay

Skin tissues were harvested from P0 newborn mice, and alkaline ceramidase activity was measured using D-e-C24:1 ceramide or D-e-C18:1 ceramide as a substrate according to the protocol that we developed previously (Wang et al., 2015).

HF Cycle Monitoring

To determine the length of the first anagen of the HF cycle, P21 mice were shaved on the back and the shaved areas were photographed daily to record color changes with time. To monitor HF cycling after the first anagen, the HF cycle was synchronized in telogen. Mice were depilated by applying bikini wax onto their backs, and the colors of the depilated areas were monitored daily as described above. The stage of the HF cycle was determined by the skin color as described (Muller-Rover et al., 2001).

In Situ β-Galactosidase Assay

Freshly obtained tissues were embedded in optimal cutting temperature compound (Tissue Tek). Cryosections were done at 5 μm on a cryostat set at −25°C. For in situ β-galactosidase activity assays, sections were fixed for 10 min at 4°C in 0.5% glutaraldehyde/PBS. Samples were then washed with PBS three times for 5 min each and stained for 6–24 hr at 37°C with X-gal working solution with 1 mg/mL X-gal in PBS, 5 mM K3Fe(CN)6, and 5 mM K4Fe(CN)6, and counterstained with nuclear fast red (Vector Labs) or eosin (Sigma) (Folgueras et al., 2013).

LC-MS/MS Sphingolipid Analysis

Skin tissues collected from Acer1 or Acer1 mice at different ages and skin sphingolipids were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) as described (Bielawski et al., 2006) in the Lipidomics Core Facility at Stony Brook University.

Histology

Mouse dorsal skin tissues were harvested, fixed in 10% formalin, embedded in paraffin, and sectioned at 5 μm. Deparaffined and rehydrated sections were stained with H&E.

Immunohistochemistry Staining

IHC staining was done as described in our previous study (Wang et al., 2015). Briefly, skin sections were deparaffined and rehydrated, antigens were retrieved, and endogenous peroxidase activity was quenched. Sections were stained by Thermo Fisher Histostain-Plus Kit (Thermo Fisher), incubated with 1.1 mM DAB (3,3′-diaminobenzidine tetrahydrochloride hydrate, D5637, Sigma), and counterstained with hematoxylin (Sigma). The stained tissue sections were imaged under a Zeiss fluorescent microscope.

Immunofluorescence Staining

Skin tissue cryosections were fixed with 4% paraformaldehyde for 10 min, permeabilized with 0.1% Triton X-100 in PBS for 15 min, and washed with PBS three times, 5 min each, at RT, blocked with 1% goat serum at RT for 30 min, incubated with a primary antibody at 4°C overnight, followed by a corresponding fluorophore-conjugated secondary antibody (1:400; Jackson ImmunoResearch) for 45 min, and counterstained with 1 μg/mL of DAPI/PBS for 10 min before being imaged under the Zeiss fluorescent microscope.

Analysis of Skin Label-Retaining Cells

For newborn LRC analysis, Acer1, Acer1 and Acer1 newborns (P0) were injected with EdU (100 mg/g body weight) daily for 4 consecutive days. At 1 month of age (P31), the mice were euthanized, the dorsal skins were isolated, paraffin-embedded, sectioned, and the resulting sections were subjected to EdU staining using a Click-iT EdU Alexa Fluor 562 Imaging Kit (Life Technologies).

Fluorescence-Activated Cell Sorting

Sorting was done as described (Nowak and Fuchs, 2009). Briefly, epidermis was separated from the whole skin by 0.25% EDTA-free trypsin treatment at 4°C overnight. Single cells were selected by passing through a 70 μm cell strainer followed by a 40 μm cell strainer. Cells were incubated with FITC-CD34 and PE-integrin α6 antibodies (Table S3) at RT for 30 min. CD34/α6 double-positive cells were sorted out by FACS on a BD FACSAria III system (BD Biosciences), compared with negative controls without adding antibodies, and collected to E media (Nowak and Fuchs, 2009) and seeded onto the mitomycin C-treated Swiss 3T3 feeder cells on 12-well or 24-well culture plates with 0.3 mM CaCl2.

Colony Formation Assay

CD34/α6 double-positive cells were cultured with feeder cells on a 12-well culture plate for 14–21 days until colonies became visible. Cell cultures were washed with PBS before being fixed with 4% paraformadehyde at RT for 10 min. The fixed cell cultures were washed with PBS and stained with 2% Rhodamine B (Sigma)/d2H2O at RT.

Feeder-free Culture of HFSC

HFSC were cultured in feeder-free 3C media in Matrigel as described (Chacón Martínez et al., 2017). Briefly, 3C media is modified from E media (Nowak and Fuchs, 2009) by adding 10 ng/mL EGF, 5 ng/mL transferrin, 5 μM Y27632, 20 ng/mL mouse recombinant VEGF-A, 20 ng/mL human recombinant FGF-2 (Sigma), and 0.3 mM CaCl2. 2.5 × 104 cells were suspended in 40 μL of an ice-cold 1:1 mixture of 3C media and growth-factor-reduced Matrigel (Corning) dispensed as a droplet in a 24-well culture plate.

Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling Assay

Apoptotic cells were assessed using a DeadEnd Fluorometric TUNEL System (Promega). Briefly, deparaffinized skin sections were permeabilized by immersing in 20 μg/mL of Proteinase K at RT for 10 min, and equilibrated in an equilibrium buffer at RT for 10 min before fragmented DNAs were labeled with a TdT reaction mix at 37°C for 1 hr and then stopped by immersing the tissue sections in 2× SCC buffer. The tissue sections were counterstained with 1 μg/mL of DAPI/PBS for 10 min before being imaged under a Zeiss fluorescent microscope.

Antibodies

Primary antibodies used for IHC, IF, FACS, and western blotting are listed in Table S3.

Author Contributions

C.-L.L. and C.M. conceived and designed experiments; C.-L.L. and R.X. generated knockout mice; C.-L.L., R.X., J.K.Y., and F.L. performed experiments; J. Cao, L.H., and B.R. contributed to reagents and equipment; J. Chen, E.C.J., J.B.S., X.Z., A.J.S., L.M.O., and Y.A.H. contributed to interpretation of data and manuscript editing. C.-L.L. and C.M. wrote the manuscript.