Retinoid Metabolism Is Altered in Human and Mouse ... · Retinoid Metabolism Is Altered in Human and Mouse Cicatricial Alopecia Helen B. Everts1, Kathleen A. Silva2,ShaliseMontgomery1,

Retinoid Metabolism Is Altered in Human and MouseCicatricial AlopeciaHelen B. Everts1, Kathleen A. Silva2, Shalise Montgomery1, Liye Suo1, Monica Menser1, Amy S. Valet3,Lloyd E. King3, David E. Ong4 and John P. Sundberg2,3

C57BL/6 mice develop dermatitis and scarring alopecia resembling human cicatricial alopecias (CAs), particularlythe central centrifugal CA (CCCA) type. To evaluate the role of retinoids in CA, the expression of retinoidmetabolism components were examined in these mice with mild, moderate, or severe CA compared with haircycle-matched mice with no disease. Two feeding studies were conducted with dams fed either NIH 31 diet(study 1) or AIN93G diet (study 2). Adult mice were fed AIN93M diet with 4 (recommended), 28, or 56 IU vitaminA g� 1 diet. Feeding the AIN93M diet to adults increased CA frequency over NIH 31 fed mice. Increased folliculardystrophy was seen in study 1 and increased dermal scars in study 2 in mice fed the 28 IU diet. These resultsindicate that retinoid metabolism is altered in CA in C57BL/6J mice that require precise levels of dietary vitamin A.Human patients with CCCA, pseudopelade (end-stage scarring), and controls with no alopecia were also studied.Many retinoid metabolism proteins were increased in mild CCCA, but were undetectable in pseudopelade.Studies to determine whether these dietary alterations in retinoid metabolism seen in C57BL/6J mice are alsoinvolved in different types of human CA are needed.

Journal of Investigative Dermatology (2013) 133, 325–333; doi:10.1038/jid.2012.393; published online 25 October 2012

INTRODUCTIONVitamin A and its derivatives (retinoids) are important fordevelopment and maintenance of multiple epithelial tissuesincluding skin, hair, and sebaceous glands, as severe, adverseeffects are seen in vitamin A deficiency or excess (Zouboulis,2006; Smith and Thiboutot, 2008; Everts, 2012). Vitamin A isstored in the liver and extrahepatic tissues as retinyl esters(O’Byrne et al., 2005). Retinol bound to retinol-bindingprotein 4 (RBP4) is the main circulating form of vitamin A(Blaner et al., 2002) and maintained at a constant blood level(Blomhoff et al., 1991). After a meal, some retinyl esters enternonhepatic tissues from circulating chylomicron remnants(Blaner et al., 2002). Retinoic acid (RA) synthesis occurs

locally in or near cells that utilize it. Precise spatial andtemporal levels of RA in skin are achieved by regulating keysteps in cellular vitamin A metabolism: storage as retinylesters, RA synthesis, and RA degradation (Everts, 2012).Retinol is transported into the cell through the protein calledstimulated by RA6 and binds cellular retinol-binding protein 1(RBP1, aka CRBP). This bound retinol is either esterified bylecithin:retinol acyltransferase for storage or reversibly oxi-dized to retinal via retinol dehydrogenases (dehydrogenasereductase (SDR family) member 9 (DHRS9), RDH10). Retinalis further oxidized to RA by retinal dehydrogenases 1–3(ALDH1A1–3). RA then moves to the nucleus bound tocellular RA-binding protein 2 (CRABP2), binds to RA receptorsalpha, beta, and gamma (RARA, B, C) where it activates thetranscription of over 500 genes (Dong et al., 1999; Balmer andBlomhoff, 2002). Retinyl ester hydrolases convert storedretinyl esters back to retinol when needed (Blomhoff et al.,1991; Gao and Simon, 2005).

When RBP1/CRBP is saturated or absent, retinol interactswith different enzymes. Acyl CoA:diacylglycerol acyltransfer-ase 1 (DGAT1) esterifies retinol for storage. Alopecia withaccelerated telogen to anagen transition was seen in basal epi-dermis and outer root sheath specific (Krt14 cre) Dgat1tm2Far

null mice (Shih et al., 2009). This increased anagen inductionand alopecia was reduced in Dgat1tm2Far null mice byseverely reducing dietary vitamin A intake, indicating thatalopecia was due to excess retinol and RA. Free retinol canalso be oxidized by medium chain alcohol dehydrogenases(ADH1, 3, and 4; current gene names Adh1, 5, and 7,Duester, 2001). It has been argued by some (Kumar et al.,

See related commentary on pg 285 ORIGINAL ARTICLE

1Department of Nutrition, The Ohio State University, Columbus, Ohio, USA;2The Jackson Laboratory, Bar Harbor, Maine, USA; 3Department of Medicine(Dermatology), Vanderbilt University Medical Center, Nashville, Tennessee,USA and 4Department of Biochemistry, Vanderbilt University Medical Center,Nashville, Tennessee, USA

Correspondence: Helen B. Everts, Department of Human Nutrition, The OhioState University, Columbus, Ohio 43210, USA. E-mail: [email protected]

Received 8 February 2012; revised 13 August 2012; accepted 21 August2012; published online 25 October 2012

Abbreviations: ADH1, 3, and 4, alcohol dehydrogenase 1, 3, and 4; AIN,American Institute of Nutrition; ALDH1A1, 2, and 3, retinal dehydrogenase 1,2, and 3; B6, C57BL/6J inbred mouse strain; CA, cicatricial alopecia; CCCA,central centrifugal CA; CRABP2, cellular retinoic acid-binding protein II;CYP26A1, B1, C1, cytochrome P450 26 family members; DGAT1,diacylglycerol acyltransferase 1; DHRS9, dehydrogenase reductase (SDRfamily) member 9; IHC, immunohistochemistry; IR, immunoreactivity; RA,retinoic acid; RARA, B, and G, retinoic acid receptor alpha, beta, and gamma;RBP1, (formerly CRBP) cellular retinol-binding protein 1

& 2013 The Society for Investigative Dermatology www.jidonline.org 325

http://dx.doi.org/10.1038/jid.2012.393

mailto:[email protected]

http://www.jidonline.org

2012) and refuted by others (Napoli, 2012) that excess retinolis oxidized to retinal by ADH1, and then to RA by ALDH1A1.When given a toxic retinol dose, mice that lack Adh7tm1Gdu

(class IV, previously Adh4) have decreased retinol oxidationin the kidney (Deltour et al., 1999a, b), but not in the liver(Molotkov et al., 2002). These Adh7tm1Gdu null mice are alsomore sensitive to vitamin A deficiency. ADH1 and ADH7(class IV) are present in skin (unpublished observation,Haselbeck et al., 1997), but their roles in skin have not beenstudied.

RA is further metabolized by cytochrome P450 26 familymembers (CYP26A1, B1, C1) with the assistance of cellularRA-binding protein 1. Both Cyp26a1 and Cyp26b1 messagelevels are RA inducible in both organotypic cultures andhuman epidermis (Lorie et al., 2009a, b).

Primary cicatricial alopecias (CAs) are scarring hair lossdiseases with permanent hair loss because of destruction offollicles and their stem cells by lymphocytic- and neutrophilic-mediated inflammation (Bergfeld and Elston, 2003). Centralcentrifugal CA (CCCA) is a progressive permanent alopeciaseen primarily in African-American women usually beginningin their mid-30s, in an anterior vertex and mid-crown pattern.C57BL/6J (B6) mice develop CA with similar characteristics aspatients with CCCA, including premature desquamation of theinner root sheath and follicular dystrophy (Sperling and Sau,1992; Sperling et al., 1994, 2000; Sperling, 2001; Sperling andCowper, 2006; Sundberg et al., 2011). In B6 mice, as the hairshafts twist, they penetrate through the outer root sheath andinduce an inflammatory response. Studies in asebia-mutantmice (ABJ/Le-Scd1ab-J/Scd1ab-J and DBA/1lacJ-Scd1ab-2J/Scd1ab-2J) also suggest that the inflammatory response in CAfollows hair shaft penetration through the follicle (Sundberget al., 2000). Other theories for the cause of the various CAs(McElwee, 2008) include: altered sebaceous gland functionwith reduced sebum (Stenn et al., 1999; Sundberg et al., 2000;Al-Zaid et al., 2011) and sebaceous gland duct inflammation(Al-Zaid et al., 2011); reduced peroxisome proliferator–activated receptor gamma and lipid metabolism (Karniket al., 2009); and hair follicle loss of immune privilege(Harries et al., 2009).

Few studies have examined the role of diet in CA. Retinoidsupplementation improved CA in four patients with low serumretinol levels (Kalz, 1958). As a result of this study and theunique localization pattern of RA synthesis within the hairfollicle and sebaceous gland during the hair cycle (Evertset al., 2007), the expression of retinoid metabolism proteinswere examined in the B6 mouse model of CA and humanpatients with CA. The initial study reported that all substrainsof B6 mice had a polymorphism in Adh7, which reduced itsfunction (Dolney et al., 2001a, b), and only B6 substrains witha higher frequency of disease had higher levels of DHRS9 intelogen follicles (Sundberg et al., 2011). As CA in B6 micespared telogen follicles and only late anagen and catagenfollicles were affected, studies examined retinoid metabolismin anagen hair follicles. The current report identifies RAsynthesis and signaling protein changes in both mouse andhuman CA, and that modifying vitamin A intake modifies CAseverity in B6 mice.

RESULTSRA synthesis, signaling, and degradation proteins were increasedin early/mild CA, but became undetectable in severe disease

The expression of retinoid synthesis components was previouslymeasured in telogen follicles of C57BL/6 mouse substrains(Sundberg et al., 2011). The current study evaluated anagenfollicles, as it is the stage when B6 CA is most severe and RAsynthesis components peaked (Everts et al., 2007; Sundberget al., 2011). Biopsies of B6 mice that spontaneously devel-oped CA in The Jackson Laboratory production facility wereevaluated by immunohistochemistry (IHC) using antibodiesagainst RBP1, DHRS9, ALDH1A1, ALDH1A2, ALDH1A3,CRABP2, CYP26A1, RARA, and RARB. These mice weregrouped by severity of disease into no disease, mild, mode-rate, or severe disease with hair follicles in mid-late anagenusing standard staging criteria (Muller-Rover et al., 2001).Biopsies from human patients with CCCA, end-stage CA(pseudopelade), and controls with no alopecia (tinea capitisand follicles adjacent to pilar cysts) were also examined.DHRS9 was increased in mild disease in mice and humanswith CCCA yet undetectable in severe disease (Figure 1,Supplementary Figures S1 and S2 online). A similar patternof expression was also seen for ALDH1A1, ALDH1A2, andRARB in mice and humans (Figure 2, Supplementary Figure S3online, data not shown). CYP26A1 showed the samepattern as well, but was only tested in mice (Figure 2;Supplementary Figure S3 online). In moderate mouse CA,levels of these proteins were similar to controls (data notshown). In areas of dermal scars in mice, DHRS9 andALDH1A2 were high, but ALDH1A1 localized only to theouter root sheath of telogen follicles and ALDH1A3 was notaltered (data not shown). CRABP2 IR also increased in milddisease, but remained high within dystrophic hair folliclesin all disease states (Figure 3). RARA localization changedwith more RARA-positive cells seen in the dermis, espe-cially surrounding the bulge in mild, moderate, and severemouse disease (Figure 3).

Dietary vitamin A altered the frequency and severity of CA inB6 mice

To test the hypothesis that abnormal vitamin A metabolism isone of the primary mechanisms causing CA in B6 mice, 12-week-old female B6 mice (n¼ 9–10) were fed purified dietscontaining one of the three levels of vitamin A (ResearchDiets, New Brunswick, NJ). Diets contained either the Amer-ican Institute of Nutrition (AIN; Reeves et al., 1993)recommended level (4 IU g� 1 diet), the highest level foundin a commercial diet used at The Jackson Laboratory (high;28 IU g diet), or twice the highest level (excess; 56 IU g�1).Mice fed excess vitamin A had less alopecia than either of theother groups, and all of these mice had significantly fewer hairfollicles in anagen, indicating that follicles were in theirnormal telogen phase (Figure 4a and b). Histological analysisshowed that more hair follicle dystrophy occurred when hairfollicles were in anagen or catagen with an inverse correlationbetween percent of hair follicle dystrophy and telogen follicles(Figure 4c). Mice fed high vitamin A had more folliculardystrophy and granulomas (Figure 4d and e).

HB Everts et al.Retinoid Metabolism in Cicatricial Alopecia

326 Journal of Investigative Dermatology (2013), Volume 133

A second feeding study was conducted to test the hypoth-eses that significantly lowering dietary vitamin A wouldreduce the frequency and severity of CA. Typical mousecommercial diets (chow) contain up to seven times the AINrecommended level of dietary vitamin A (4 IU g�1 diet,Reeves et al., 1993). The chow diet fed to mice at TheJackson Laboratory was analyzed and contained 6–28 IUvitamin A g�1 depending on the lot. This variability isimportant as vitamin A is stored in the liver and dams fedchow during breeding produce pups with large amounts ofvitamin A in their liver. Feeding dams a vitamin A-deficientdiet during breeding reduced liver stores of vitamin A in theirpups (Smith et al., 1987) and reduced alopecia in Dgat1tm2Far

null mice (Shih et al., 2009). To significantly lower vitamin Alevels in this study, dams were fed AIN93G diet for twogenerations. This study also evaluated whether inducinganagen by wax stripping would lead to CA but these hypo-theses were incorrect. Alopecia was consistently high in all themice not wax stripped. However, wax stripping reduced the

frequency of alopecia in mice fed recommended and highlevels of vitamin A (Figure 4a). The one wax-stripped mousewith alopecia fed the recommended diet developed signifi-cantly more ulcers, dermal scars, and epidermal hyperplasiathan any other mouse in the study. The only other mice withepidermal hyperplasia and dermal scars were the mice fedhigh vitamin A in study 2 (Figure 4f). They also had signifi-cantly less follicular dystrophy and granulomas than mice fedthe same diet in study 1 (Figure 4d and e, Po0.05). Withinstudy 2, mice fed high vitamin A had significantly lowernumber of granulomas than mice fed recommended levels ofvitamin A (Figure 4e, Po0.05). This suggests that these micefed high vitamin A in study 2 had a previous bout of severedisease.

Liver retinoid analysis revealed that dams fed the 4 IU dietdid not lower liver retinyl palmitate levels in their pups aspredicted (Figure 5b, Smith et al., 1987). Both liver retinol andretinyl palmitate increased with dietary vitamin A as expected(Figure 5a and b, Po0.001). Liver retinol was significantlygreater in the second study (Figure 5a, Po0.01). In addition,wax stripping significantly reduced liver retinyl palmitatelevels (Figure 5b, Po0.01). Skin retinol increased with dietary

Figure 1. Immunoreactivity of dehydrogenase reductase (SDR family)

member 9 (DHRS9) increased in C57BL/6J mice with mild dermatitis and

biopsies from human patients with central centrifugal cicatricial alopecia

(CCCA), but was reduced in mice with severe disease and biopsies from

human patients with pseudopelade. Immunohistochemistry was performed

with an antibody against DHRS9 in dorsal skin from chow fed C57BL/6J inbred

mouse strain (B6) mice with no disease (a, n¼ 18), mild disease (c, n¼6),

severe disease (e, n¼ 38), biopsies from patients with no alopecia (b, normal

Caucasian skin adjacent to pilar cyst removed by excision n¼ 2, tinea capitus

from African-American skin n¼ 1), CCCA (d, n¼ 14), or pseudopelade

(f, n¼3). b and f are from Caucasian patients, whereas d is from an African-

American patient. Bar¼101mM for main picture and 10.1mM for insets.

Retinol-CRBP

Control Mild Mod. Severe

Retinal

Free retinol

ADH7a DHRS9

Retinal

ALDH1A1ALDH1A1

ALDH1A2

ALDH1A3PMCL

Retinoicacid-CRABP2

Active RARA

CYP26A1

RARB

Activate transcription

Degradation

Free retinoicacid

Figure 2. Overview of retinoid metabolism and how these proteins changed

during the progression of cicatricial alopecia. The left shows a schematic of

RA synthesis, degradation, and signaling pathways in B6 mice. The right shows

a summary of IHC results in chow fed B6 mice with no (control), mild,

moderate, or severe disease. Immunoreactivity intensity ranges from dark to

light; very strong (maroon), strong (red), moderate (hot pink), mild (pink), or

undetectable (white). ALDH1A1, 2, and 3, retinal dehydrogenase 1, 2, and 3;

B6, C57BL/6J inbred mouse strain; CL, companion layer; CRABP2, cellular

retinoic acid-binding protein II; CRBP, cellular retinol-binding protein;

CYP26A1, cytochrome P450 26 family member A1; DHRS9, dehydrogenase

reductase (SDR family) member 9; IHC, immunohistochemistry; Mod.,

moderate; PM, premedulla; RA, retinoic acid; RARA and B, retinoic acid

receptor alpha and beta.


www.jidonline.org 327


vitamin A in both the studies (Figure 5c, Po0.05). In mice fedhigh vitamin A, skin retinol was greater in study 1 than instudy 2. This trend was seen when mice wax stripped and notwax stripped were compared separately, but was significantwhen all mice in study 2 fed high vitamin A were combined(Figure 5c, Po0.05). There were no significant effects of pupor dam diets on skin retinyl ester levels (Figure 5d). Skin retinylester levels were significantly reduced in wax-stripped mice(Figure 5c and d, Po0.05).

IHC analysis was performed to compare results from thesediet studies to those performed in NIH31 fed mice. RARB wassignificantly reduced as the percentage of follicular dystrophyincreased (Supplementary Figures S4b, d and S5 online).RARB was also low in dermal scars (Supplementary FiguresS4f and S5 online). In contrast, DHRS9 remained high indystrophic hair follicles, but dropped slightly in dermal scars(Supplementary Figure S4a, c, e online). None of the mice inthese diet controlled studies had severe follicular dystrophy asseen in NIH31 fed mice.

DISCUSSIONThis study emphasizes that the effects of an essential nutrient,vitamin A, must be tightly regulated to prevent adverse effectsbecause of excess or deficiency. In this study, both low andhigh levels of dietary vitamin A given to B6 mice affected thedevelopment and degree of follicular scarring, indicating thatprecise control of retinoid metabolism is required for normal

function of hair follicles. Initial studies localized RA synthesisand signaling proteins in normal mouse skin (Everts et al.,2004, 2007). A subsequent study found that only B6 substrainswith a higher frequency of inflamed and follicular scarring hadhigher levels of DHRS9 in telogen follicles (Sundberg et al.,2011), suggesting a possible basis for their tendency todevelop CA. As CA occurs primarily during mid-late anagen,the current study analyzed mid-late anagen hair follicles andfound that RA synthesis, degradation, and binding proteinswere altered during the progression of CA in B6 mice. Theseproteins were increased in mild disease and reduced in severedisease. A similar pattern of expression was also seen inbiopsies from patients with CCCA, although the full progres-sion of disease was not obtained.

This study characterized changes in retinoid metabolismthroughout the entire progression of CA in a mouse model ofprimary human CA. Others (Flowers et al., 2011) found at onlyone time point Rbp1 (Crbp) and Crabp2 increased in skinspecific (Krt14-cre recombinase) Scd1tm2Ntam-null male B6mice with mild CA. They also reported that several otherretinoid metabolism genes were increased in their microarrayanalyses, but changes in expression of these genes were notfurther validated at the message or protein level. They alsofound that retinol, retinyl esters, and RA levels were increasedin the skin of these null mice. Feeding a vitamin A-deficientdiet did not reduce skin RA levels or change the phenotype,arguing that the lack of Scd1 led to the increased skin RAlevels. In the current study, RA metabolism was altered in theabsence of a Scd1 mutation and significantly reducing dietaryvitamin A made the disease worse. These current studiesexpand on the observations of Flowers et al. (2011) byshowing that as follicular dystrophy increases, the expressionof proteins responsible for RA synthesis and degradation beginto decrease and reach undetectable levels in severe CA. Thisobservation was further confirmed by the reduction of RARB,a classic RA target gene, as follicular dystrophy increased inthe two diet studies preceding the reduction of DHRS9. Inaddition, these alterations in RA synthesis proteins were alsoseen in biopsies from patients with CA. It is unknown at thistime whether altered RA synthesis and signaling caused thealopecia or this alteration in retinoid metabolism is secondaryto the disease.

CYP26A1 also localized within the hair follicle and wasaltered during CA. CYP26A1 was previously reported only inbasal keratinocytes, eccrine sweat glands, and sebaceousglands (Heise et al., 2006). CYP26A1 has a key role in main-taining RA levels through a feedback inhibitory loop (Reijntjeset al., 2005). High levels of RA directly induce CYP26A1expression (Loudig et al., 2000). Then CYP26A1 degrades thisexcess RA. In addition, when RA levels are low, CYP26A1levels are also low. In retinol dehydrogenase 1 (Rdh1tm1.1Jln)null mice reduction in CYP26A1 expression compensated forthe reduced RA synthesis and maintained normal RA levels(Zhang et al., 2007). These observations are consistentwith the concurrent alterations of CYP26A1 with DHRS9 inthis study.

Epigenetic silencing may explain the expression resultsseen. Mutations in either CRABP2 or RARA resulted in

Figure 3. CRABP2 and retinoic acid receptor alpha (RARA) levels were

increased in C57BL/6J mice with cicatricial alopecia. Immunohistochemistry

was performed with antibodies against CRABP2 (a, c, e) and RARA (b, d, f),

in mice with no disease (a, b), mild disease (c, d), or severe disease (e, f).

Bar¼101mM for main picture and 10.1mM for insets.



hypermethylation and epigenetic silencing of RARA targetgenes such as RARB, CRBP, and CYP26A1 (Ren et al., 2005;Corlazzoli et al., 2009; Laursen et al., 2012). As CRABP2functions to deliver RA to RARA (Dong et al., 1999) and RARArepresses transcription in the absence of RA (Chen and Evans,1995; Horlein et al., 1995), the above studies suggest that anydefect resulting in reduced RA binding to its receptor couldcause hypermethylation and gene silencing of RARA targetgenes. Such defects would include reduced levels of RAprecursors retinol or retinyl esters, reduced uptake of retinolor retinyl esters into the skin, reduced RA synthesis, reduceddelivery of RA to RARA via CRABP2, increased RAdegradation, or any defect in RARA itself. It is possible thatthe reduction in RA synthesis enzymes, RARB, and CYP26A1in moderate and severe CA, was because of this mechanism ofaberrant RA signaling induced hypermethylation. Alternatively,a primary effect of increased dietary vitamin A could bebecause of toxic effects during mild disease on specific hair

follicle cells, leading to their death or decreased biochemicalfunction such that the levels of RA synthesis enzymes andreceptors decrease as CA disease progresses. The currentstudies indicate the balance between RA synthesis andretinol degradation is altered in a mouse model of humanCA and imply that altered retinoid metabolism is involved inthe pathogenesis of CA. Future studies will determine whetherCA is because of altered retinoid metabolism or is because ofloss of retinoid related enzymes present in skin because ofscarring induced by another cause.

The current study implies that precise levels of dietaryvitamin A are required to compensate for the defects inretinoid metabolism found in substrains of C57BL/6 mice.A polymorphism in Adh7 (class IV, previously reported asAdh4) was previously found in C57BL/6 mice that reduced itsfunction (Dolney et al., 2001a, b; Sundberg et al., 2011).C57BL/6 mice have arginine120, whereas C3H/HeJ and otherstrains of mice have cysteine120. Human ADH7 (class IV) also

60a d

b e

c f

4 lU g–1 diet28 lU g–1 diet56 lU g–1 diet





% H

air

loss

% H

air

folli

cles

in a

nage

n%

Hai

r fo

llicl

e dy

stro

phy

% F

ollic

ular

dys

trop

hy%

Gra

nulo

mas

% D

erm

al s

cars

00

aab

b

b

b

ab ab

ab

ab ab

ab0

0 0 0 0

ab

abab

a

a

a

NA NA

NA

NA NA

NA

ab

ab ab ab ab

ab

0*

#

Study 1 Study 2 not waxed Study 2 waxed





40

20

60

0

40

20

60

0

40

20

20

20

10

10

8

6

4

2

0

0

10

00 20 40

40

30

60 80 100

% Telogen

120

30

40

50

Figure 4. Cicatrical alopecia was altered by dietary vitamin A and wax stripping. The percentage of mice with hair loss (a) was determined by visual inspection

in B6 mice fed 4 (open bars), 28 (closed bars), or 56 (hatched bars) IU vitamin A g� 1 diet. H&E slides were then scored for % of hair follicles in anagen

(b) and telogen (c), % of follicular dystrophy (d), % granulomas (e), and % of dermal scars (f). *Significant effect of wax stripping as analyzed by chi-square

(Po0.05), #significant diet effect in the wax-stripped mice (Po0.05). Results with different letters are significantly different from each other in mice with hair loss

(Po0.05). (c) r¼ � 0.86, Po0.001. n¼ 9–10 study 1, n¼ 8–19 study 2. B6, C57BL/6J inbred mouse strain; H&E, hematoxylin and eosin; NA, not applicable.




contains arginine120, similar to C57BL/6 mice (Kedishviliet al., 1995). Additional polymorphisms in ADH7 (class IV)were also found in humans linked to alcohol dependence,heroin addiction, and reduced risk of head and necksquamous cell carcinoma (Edenberg et al., 2006; Levranet al., 2009; Wei et al., 2010). Polymorphisms in ADH7(class IV) may predispose B6 mice and humans to CA. Studiesin transgenic mice suggest two roles for Adh7 (class IV).

Adh7tm1Gdu null mice have reduced fetal survival when damsare deficient in vitamin A (Deltour et al., 1999a). When theseAdh7tm1Gdu null mice are stressed with excess retinol therewas reduced retinol clearance in the kidney (Deltour et al.,1999a, b), but not the liver (Molotkov et al., 2002). These dataimply that reduced Adh7 (class IV) activity increasesusceptibility to both vitamin A deficiency and toxicity in B6mice and some humans. This phenomenon may explain whymore severe disease was seen in the second study when damswere fed lower vitamin A. The mice fed the AIN93G diet werealso slow to breed; the number of litters and the number ofpups per litter were decreased compared with chow fed mice.These pups appeared runted at first, but appeared to catch uptheir growth after several days. In addition, many of the dam’sfirst litters were stillborn. It is possible that C57BL/6 micerequire a narrower window of vitamin A for optimal functionthan other strains of mice. In the current study, this optimalwindow is above 4 and below 28 IU vitamin A g�1 of diet.Variations in vitamin A levels in commercial unpurified dietsrange from 6.2 to 28.7 IU in seven tests from two laboratories,11.8–24.3 IU within the same lot (Sundberg unpublished data)implying mixing is not uniform. In susceptible strains, thisvariation in vitamin A concentration in commercial diets canhave a profound impact on the frequency of alopecia anddermatitis, and explain the apparent randomness of CAincidence in B6 mice in The Jackson Laboratory facility.Humans with similar polymorphisms may also require moreprecise levels of dietary vitamin A. Anecdotal reports suggestreduced alopecia with retinyl palmitate supplementationat roughly 3.5 times the recommended daily allowance(8,000 IU day� 1) in patients with low serum vitamin A(Wilma Bergfeld, personal communication). Future studiesare needed to determine the optimal dose of vitamin A inhumans and mice susceptible to CA.

Retinoids are not the first treatment choice in CA patients,but they do help some patients and are listed as a second orthird line of treatment in some forms of CA (Harries et al.,2008). This variable benefit of retinoid therapy is similar to theinitial clinical trials in cancer (reviewed in, Freemantle et al.,2003). It is now well established that retinoid resistance occursin cancer because of epigenetic hypermethylation of RARBand other RA synthesis enzymes (Tang and Gudas, 2011).Current studies combining low-dose retinoid treatment withmethyl transferase inhibitors in mouse models of cancer weremore successful (Tang et al., 2009). A similar situation ofretinoid resistance may occur in CA. If this is the case then thisdual treatment strategy may be worth testing for the treatmentof CA. As high vitamin A worsened CA in this mouse study,this implies that pharmacological doses of retinoids may havebeen toxic. High vitamin A intake (12 IU g�1 diet) alsoaccelerated alopecia areata in graft-induced C3H/HeJ mice(Duncan et al., 2012). Telogen effluvium was also seen inpatients treated with retinoids (Ruzicka et al., 1992). Thus, abetter treatment option may be to restore retinoid signalingwith methyl transferase inhibitors while maintaining adequatedietary vitamin A levels. The exact ‘‘adequate’’ level of dietaryvitamin A is likely to be very different in mice and humanswith different genetic backgrounds.

350 Study 1Study 2 not waxedStudy 2 waxed

Study 1Study 2 not waxedStudy 2 waxed



b

b

a

a

a a aab ab

bc

*

abc

c

c

b

c

#

*

Live

r re

tinol

(nm

ol/g

tiss

ue)

Live

r re

tinyl

pal

mita

te(n

mol

g–1

tiss

ue)

Ski

n re

tinol

(nm

ol g

–1 ti

ssue

)S

kin

retin

yl e

ster

s(n

mol

g–1

tiss

ue)

300

250

10,000

a

b

c

d

8,000

6,000

4,000

2,000

0

4

3

2

1

0

8

6

4

2

0

4 28

Dietary vitamin A (IU g–1 diet)

56

4

#

28


56

4 28Dietary vitamin A (IU g–1 diet)

56

200

150

100

50

4 28 560


Figure 5. Dietary vitamin A altered liver and skin retinoids differently in the

various studies. Liver retinol (a), liver retinyl palmitate (b), skin retinol (c), and

skin retinyl esters (d) were analyzed by HPLC in study 1 (open bars, n¼ 9–10),

study 2 with no wax stripping (closed bars, n¼ 8–19), and study 2 with wax

stripping (hatched bars, n¼10–16). *Significant study effect (Po0.001 in a,

and Po0.05 in c), #significant effect of wax stripping (Po0.005), results with

different letters are significantly different from each other (Po0.05).



In summary, this study suggests that retinoid metabolism isaltered in CA and high dietary vitamin A may worsen thedisease. Future studies are needed to determine whether thechanges observed occur in all primary lymphocytic andneutrophilic forms of CA and the optimal level of dietaryvitamin A to prevent the occurrence and severity of humanCA. In B6 mice studies, the diet should start with 1.5–3 timesthe recommended levels of vitamin A (6–12 IU g�1 diet in B6mice). This optimal level maybe different in different geno-types, therefore future studies should analyze linkage betweenAdh7 (class IV) single-nucleotide polymorphisms with CA.As all substrains of B6 mice have Adh7 polymorphisms,but differ in disease frequency, future studies should alsoidentify other genetic mutations in RA synthesis or signal-ing genes. Epigenetic changes in retinoid metabolism inCA should also be explored. In addition, RARB expressionlevels may be a useful biomarker for severity of CA inspecific substrains of B6 mice and types of human CA. Thisstudy also found that the frequency of this rare diseasecould be increased by diet so that future studies can beconducted in B6 mice to define the pathogenesis of CA anddevelop specific treatments of human CA.

MATERIALS AND METHODSMice

The Jackson Laboratory (Bar Harbor, ME) Institutional Animal Care

and Use Committee approved all procedures and hair loss was

observed weekly. Hematoxylin and eosin slides of biopsies were

scored for specific follicle stage, follicular dystrophy, granulomas,

epidermal hyperplasia, ulcers, and dermal scars by a board certified

pathologist (JPS, Sundberg et al., 2008).

Initial expression study mice

Mice were fed the NIH 31 diet with 6% fat; (LabDiet 5K52, Purina

Mills, St Louis, MO) throughout the study. There were 51 female

controls. Cases were divided into mild (n¼ 6), moderate (n¼ 15),

severe (n¼ 38), and dermal scars (26). Hair follicles were staged as

anagen II–IIIc versus anagen IV–catagen I, and hair follicles in these

precise stages were compared between the groups (Muller-Rover

et al., 2001).

Diet study 1 mice

Dams and female pups were fed the NIH 31 diet until 12 weeks of

age, until switching to the AIN93 Maintenance (M) diet containing

4 (n¼ 9), 28 (n¼ 10), or 56 (n¼ 10) IU vitamin A g� 1 diet (Research

Diets) for 4 months.

Diet study 2 mice

Dams were fed the AIN93 growth (G) diet throughout gestation.

Female pups from these dams were maintained on this AIN93G

diet during their growth phase and when they became pregnant.

Mice from this second generation of dams fed the AIN93G diet

were used in this study. At 6 weeks of age, these mice were fed

the AIN93M diet containing 4 (n¼ 19), 28 (n¼ 35), or 56 (n¼ 22)

IU vitamin A g� 1 diet (Research Diets). At 16 weeks of age, mice

(n¼ 10, 4 IU; n¼ 16, 28 IU; n¼ 14, 56 IU) were depilated and then

euthanized 2 weeks later.

Human samplesParaffin blocks of biopsies from 14 human patients with CCCA, 3

human patients with pseudopelade, and 3 nonalopecic dermatoses

(tinea capitus and skin adjacent to pilar cysts) were obtained with

Institutional Review Board approval from archives at Vanderbilt

University (Nashville, TN; Supplementary Table S1 online).

Procedures

IHC was performed as described previously (Everts et al., 2007).

Antibodies were purchased for CYP26A1 (Alpha Diagnostic Intl., San

Antonio, TX), RARA, and RARB (Santa Cruz Biotech, Santa Cruz, CA).

Other antibodies used were produced and validated in Dr Ong’s

laboratory (Everts et al., 2007). A blinded IHC score of RARB

immunoreactivity (IR) was measured using a scale of 0–4 multiplied

by the percentage of cells/mid-late anagen hair follicles. Two blinded

researchers scored human samples for % of cells and strength on a

0–4 scale with 0¼ no IR; 1¼ 1–25% cells, weak IR; 2¼ 26–50%

cells, moderate IR; 3¼ 51–75% cells, intense IR; 4¼ 76–100% cells,

very intense IR. The IHC level was calculated as the sum of intensity

plus % cells as previously validated (Shamonki et al., 2006).

Retinoids were extracted (Pappas et al., 1993) and analyzed by

HPLC with a modification of Fleshman et al., (2012) using a C18

4.6� 100 mm, 3.5mM column. The solvents were (A) 80:20 MeOH/

0.1% formic acid and (B) MtBE. In all, 10–20ml of sample in 100% A

was injected and retinoids separated with initial conditions of

100%A, a linear gradient to 100%B over 13 minutes, held at

100%B for 3 minutes, then equilibrated back to 100%A over

2 minutes. Spectra were analyzed using a photodiode array (Agilent

Technologies, Santa Clara, CA) at 325 nm. Standards were purchased

from Sigma (St Louis, MO).

Chi-squared analysis was performed to determine differences in

hair loss using Prism 5 (Graph Pad; La Jolla, CA). All other data were

analyzed by analysis of variance, followed by Tukey post hoc tests

when appropriate using SPSS, v19 (IBM; Armonk, NY).

See Supplementary Data online for more detailed methods.

CONFLICT OF INTERESTThe authors state no conflict of interest.

ACKNOWLEDGMENTSThis work was supported by grants from the National Institutes of Health(AR052009 to HBE, AR052710 to JPS), the North American Hair ResearchSociety (HBE), the Cicatricial Alopecia Research Foundation (CARF to JPS), andthe Wauford Foundation (HBE).

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at http://www.nature.com/jid

REFERENCES

Al-Zaid T, Vanderweil S, Zembowicz A et al. (2011) Sebaceous gland loss and

inflammation in scarring alopecia: a potential role in pathogenesis. J Am

Acad Dermatol 65:597–603

Balmer JE, Blomhoff R (2002) Gene expression regulation by retinoic acid.

J Lipid Res 43:1773–808

Bergfeld WF, Elston DM (2003) Cicatricial alopecia. In: Olsen EA ed Disorders

of Hair Growth: Diagnosis and Treatment. 2nd edn McGraw-Hill Medical

Publishing Division: New York, 363–98



http://www.nature.com/jid

http://www.nature.com/jid


Blaner WS, Gamble MV, Vogel S et al. (2002) Retinol-binding protein (RBP):

essential physiologic functions. J Nutr 132:2979S

Blomhoff R, Green MH, Green JB et al. (1991) Vitamin-A metabolism—new

perspectives on absorption, transport, and storage. Physiol Rev 71:951–90

Chen JD, Evans RM (1995) A transcriptional co-repressor that interacts with

nuclear hormone receptors. Nature 377:454–7

Corlazzoli F, Rossetti S, Bistulfi G et al. (2009) Derangement of a factor

upstream of RAR alpha triggers the repression of a pleiotropic epigenetic

network. PLoS One 4:e4305

Deltour L, Foglio MH, Duester G (1999a) Impaired retinol utilization in Adh4

alcohol dehydrogenase mutant mice. Dev Genet 25:1–10

Deltour L, Foglio MH, Duester G (1999b) Metabolic deficiencies in alcohol

dehydrogenase Adh1, Adh3, and Adh4 null mutant mice. Overlapping

roles of Adh1 and Adh4 in ethanol clearance and metabolism of retinol to

retinoic acid. J Biol Chem 274:16796–801

Dolney DE, Szalai G, Duester G et al. (2001a) Molecular analysis of genetic

differences among inbred mouse strains controlling tissue expression

pattern of alcohol dehydrogenase 4. Gene 267:145–56

Dolney DE, Szalai G, Felder MR (2001b) Differences in charge and kinetic

properties of alcohol dehydrogenase 4 from C57BL/6 mice compared to

other inbred strains are associated with a cysteine120 to arginine120

substitution. Biochem Genet 39:239–50

Dong D, Ruuska SE, Levinthal DJ et al. (1999) Distinct roles for cellular retinoic

acid-binding proteins I and II in regulating signaling by retinoic acid.

J Biol Chem 274:23695–8

Duester G (2001) Genetic dissection of retinoid dehydrogenases. Chem-Biol

Interact 130:469–80

Duncan FJ, Silva KA, Johnson C et al. (2012) Endogenous retinoids in the patho-

genesis of alopecia areata. J Invest Dermatol (doi:10.1038/jid.2012.344)

Edenberg HJ, Xuei XL, Chen HJ et al. (2006) Association of alcohol

dehydrogenase genes with alcohol dependence: a comprehensive analy-

sis. Hum Mol Genet 15:1539–49

Everts HB (2012) Endogenous retinoids in the hair follicle and sebaceous gland.

Biochim Biophys Acta 1821:222–9

Everts HB, King LE Jr, Sundberg JP et al. (2004) Hair cycle-specific immuno-

localization of retinoic acid synthesizing enzymes Aldh1a2 and Aldh1a3

indicate complex regulation. J Invest Dermatol 123:258–63

Everts HB, Sundberg JP, King LE Jr. et al. (2007) Immunolocalization of

enzymes, binding proteins, and receptors sufficient for retinoic acid

synthesis and signaling during the hair cycle. J Invest Dermatol

127:1593–604

Fleshman MK, Riedl KM, Novotny JA et al. (2012) An LC/MS method for d8-

beta-carotene and d4-retinyl esters: beta-carotene absorption and its

conversion to vitamin A in humans. J Lipid Res 53:820–7

Flowers MT, Paton CM, O’Byrne SM et al. (2011) Metabolic changes in skin

caused by Scd1 deficiency: a focus on retinol metabolism. PLoS One

6:e19734

Freemantle SJ, Spinella MJ, Dmitrovsky E (2003) Retinoids in cancer therapy

and chemoprevention: promise meets resistance. Oncogene 22:7305–15

Gao J, Simon M (2005) Identification of a novel keratinocyte retinyl ester

hydrolase as a transacylase and lipase. J Invest Dermatol 124:1259–66

Harries MJ, Meyer KC, Paus R (2009) Hair loss as a result of cutaneous

autoimmunity: frontiers in the immunopathogenesis of primary cicatricial

alopecia. Autoimmun Rev 8:478–83

Harries MJ, Sinclair RD, MacDonald-Hull S et al. (2008) Management

of primary cicatricial alopecias: options for treatment. Br J Dermatol

159:1–22

Haselbeck RJ, Ang HL, Duester G (1997) Class IV alcohol/retinol dehydrogen-

ase localization in epidermal basal layer: potential site of retinoic acid

synthesis during skin development. Dev Dyn 208:447–53

Heise R, Mey J, Neis MM et al. (2006) Skin retinoid concentrations are

modulated by CYP26AI expression restricted to basal keratinocytes in

normal human skin and differentiated 3D skin models. J Invest Dermatol

126:2473–80

Horlein AJ, Naar AM, Heinzel T et al. (1995) Ligand-independent repression bythe thyroid-hormone receptor-mediated by a nuclear receptor co-repres-sor. Nature 377:397–404

Kalz F (1958) Cicatricial alopecia and vitamin A. AMA Arch Derm 78:740–3

Karnik P, Tekeste Z, McCormick TS et al. (2009) Hair follicle stem cell-specificPPARg deletion causes scarring alopecia. J Invest Dermatol129:1243–57

Kedishvili NY, Bosron WF, Stone CL et al. (1995) Expression and kineticcharaterization of recombinant human stocach alcohol-dehydrogenase-active-site amino-acid-sequece explains substrate specificity comparedwith liver isozymes. J Biol Chem 270:3625–30

Kumar S, Sandell LL, Trainer PA et al. (2012) Alcohol and aldehydedehydrogenases: retinoid metabolic effects in mouse knockout models.Biochim Biophys Acta, Mol Cell Biol Lipids 1821:198–205

Laursen KB, Wong PM, Gudas LJ (2012) Epigenetic regulation by RAR alphamaintains ligand-independent transcriptional activity. Nucleic Acids Res40:102–15

Levran O, Londono D, O’Hara K et al. (2009) Heroin addiction in AfricanAmericans: a hypothesis-driven association study. Genes Brain Behav8:531–40

Lorie EP, Chamcheu JC, Vahlquist A et al. (2009a) Both all-trans retinoic acidand cytochrome P450 (CYP26) inhibitors affect the expression of vitaminA metabolizing enzymes and retinoid biomarkers in organotypic epider-mis. Arch Dermatol Res 301:475–85

Lorie EP, Cools M, Borgers M et al. (2009b) Topical treatment with CYP26inhibitor talarozole (R115866) dose dependently alters the expression ofretinoid-regulated genes in normal human epidermis. Br J Dermatol160:26–36

Loudig O, Babichuk C, White J et al. (2000) Cytochrome P450RAI(CYP26)promoter: a distinct composite retinoic acid response element underliesthe complex regulation of retinoic acid metabolism. Mol Endocrinol14:1483–97

McElwee KJ (2008) Etiology of cicatricial alopecias: a basic science point ofview. Dermatol Ther 21:212–20

Molotkov A, Fan X, Duester G (2002) Excessive vitamin A toxicity in micegenetically deficient in either alcohol dehydrogenase Adh1 or Adh3. Eur JBiochem 269:2607–12

Muller-Rover S, Handjiski B, van der Veen C et al. (2001) A comprehensiveguide for the accurate classification of murine hair follicles in distinct haircycle stages. J Invest Dermatol 117:3–15

Napoli JL (2012) Physiological insights into all-trans-retinoic acid biosynthesis.Biochim Biophys Acta, Mol Cell Biol Lipids 1821:152–67

O’Byrne SM, Wongsiriroj N, Libien J et al. (2005) Retinoid absorption andstorage is impaired in mice lacking lecithin: retinol acyltransferase (LRAT).J Biol Chem 280:35647–57

Pappas RS, Newcomer ME, Ong DE (1993) Endogenous retinoids in ratepididymal tissue and rat and human spermatozoa. Biol Reprod48:235–47

Reeves PG, Nielsen FH, Fahey GC Jr (1993) AIN-93 purified diets forlaboratory rodents: final report of the American Institute of Nutrition adhoc writing committee on the reformulation of the AIN-76A rodent diet.J Nutr 123:1939–51

Reijntjes S, Blentic A, Gale E et al. (2005) The control of morphogen signalling:regulation of the synthesis and catabolism of retinoic acid in thedeveloping embryo. Dev Biol 285:224–37

Ren MQ, Pozzi S, Bistulfi G et al. (2005) Impaired retinoic acid (RA) signalleads to RAR beta 2 epigenetic silencing and RA resistance. Mol Cell Biol25:10591–603

Ruzicka T, Sommerburg C, Goerz G et al. (1992) Treatment of cutaneouslupus-erythematosus with acitretin and hydroxychloroquine. Br J Derma-tol 127:513–8

Shamonki MI, Kligman I, Shamonki JM et al. (2006) Immunohistochemicalexpression of endometrial L-selectin ligand is higher in donor eggrecipients with embryonic implantation. Fertil Steril 86:1365–75

Shih MYS, Kane MA, Zhou P et al. (2009) Retinol esterification by DGAT1 isessential for retinoid homeostasis in murine skin. J Biol Chem 284:4292–9



Smith KR, Thiboutot DM (2008) Sebaceous gland lipids: friend or foe?. J Lipid

Res 49:271–81

Smith SM, Levy NS, Hayes CE (1987) Impaired immunity in vitamin A deficient

mice. J Nutr 117:857–65

Sperling LC (2001) Scarring alopecia and the dermatopathologist. J Cutan

Pathol 28:333–42

Sperling LC, Cowper SE (2006) The histoplathology of primary cicatricial

alopecia. Semin Cutan Med Surg 25:41–50

Sperling LC, Sau P (1992) The follicular degeneration syndrome in black patients-

‘‘hot comb alopecia’’ revisited and revised. Arch Dermatol 128:68–74

Sperling LC, Skelton HG, Smith KJ et al. (1994) Follicular degeneration

syndrome in men. Arch Dermatol 130:763–9

Sperling LC, Solomon AR, Whiting DA (2000) A new look at scarring alopecia.

Arch Dermatol 136:235–42

Stenn KS, Sundberg JP, Sperling LC (1999) Hair follicle biology, the sebaceous

gland, and scarring alopecias. Arch Dermatol 135:973–4

Sundberg JP, Boggess D, Sundberg BA et al. (2000) Asebia-2J (Scd1ab2J): a new

allele and a model for scarring alopecia. Am J Pathol 156:2067–75

Sundberg JP, Sundberg BA, Schofield P (2008) Integrating mouse anatomy andpathology ontologies into a phenotyping database: Tools for data captureand training. Mamm Genome 19:413–9

Sundberg JP, Taylor D, Lorch G et al. (2011) Primary follicular dystrophy withscarring dermatitis in C57BL/6 mouse substrains resembles centralcentrifugal cicatricial alopecia in humans. Vet Pathol 48:513–24

Tang XH, Albert M, Scognamiglio T et al. (2009) A DNA methyltransferaseinhibitor and all-trans retinoic acid reduce oral cavity carcinogenesisinduced by the carcinogen 4-nitroquinoline 1-oxide. Cancer PreventionRes 2:1100–10

Tang XH, Gudas LJ (2011) Retinoids, retinoic acid receptors, and cancer.Ann Rev Pathol 6:345–64

Wei S, Liu ZS, Zhao H et al. (2010) A single nucleotide polymorphism in thealcohol dehydrogenase 7 gene (alanine to glycine substitution at aminoacid 92) is associated with the risk of squamous cell carcinoma of thehead and neck. Cancer 116:2984–92

Zhang M, Hu PR, Krois CR et al. (2007) Altered vitamin A homeostasis andincreased size and adiposity in the rdh1-null mouse. FASEB J 21:2886–96

Zouboulis CC (2006) Isotretinoin revisited: pluripotent effects on humansebaceous gland cells. J Invest Dermatol 126:2154–6




Documents

Retinoid Metabolism Is Altered in Human and Mouse ... · Retinoid Metabolism Is Altered in Human and Mouse Cicatricial Alopecia Helen B. Everts1, Kathleen A. Silva2,ShaliseMontgomery1,