ORIGINAL ARTICLE

Major Locus on Mouse Chromosome 17 and Minor Locuson Chromosome 9 are Linked with Alopecia Areata inC3H/HeJ Mice

John P. Sundberg,nw Dawnalyn Boggess,n Kathleen A. Silva,n Kevin J. McElwee,nw Lloyd E. King,z Renhua Li,n

Gary Churchill,n and GregoryA. CoxnnThe Jackson Laboratory, Bar Harbor, Maine, USA; wDepartment of Dermatology, Philipp University, Marburg, Germany; and zDivision of Dermatology,Vanderbilt University, and Department of Veterans A¡airs, Nashville,Tennessee, USA

Alopecia areata is an autoimmune disease that targetsactively growing (anagen) hair follicles in humans,mice, rats, dogs, horses, and cattle. C3H/HeJ mice spon-taneously develop alopecia areata from 5 mo of age andolder in females and later in males. Frequency of diseaseapproached 20% in a colony by 18 mo of age. C57BL/6Jmice do not develop alopecia areata. A segregating F2population of female mice (n¼1096) was generatedfrom crossing these two strains. Alopecia areata(n¼138) and clinically normal (n¼ 214) mice were gen-otyped at 12 mo of age using 211 microsatellite probes.The peak logarithm of odds ratio score on mouse chro-mosome 17 (10.9) was around marker D17Mit134 at 16.9cM from the centromere.The mouse histocompatibility

locus, H2, the mouse equivalent of human leukocyteantigen in humans, was a likely candidate. Twelve-month-old C3H.SW-H2b/SnJ mice (C3H/HeJ congenicmice in which the H2k purported susceptibility locuswas replaced with the H2b purported resistance locus)did not develop alopecia areata, supporting this locusas being important in alopecia areata. A suggestivelinkage was also found on mouse Chromosome 9 (loga-rithm of odds ratio score 2.0) around D9Mit206, 20 cMfrom the centromere. The interval on mouse Chromo-some 17 contains several orthologous genes potentiallyassociated with human alopecia areata. Key words: auto-immune disease/H2/HLA/mouse models. J Invest Dermatol120:771 ^775, 2003

Alopecia areata (AA) in humans is a suspected auto-immune disease that targets hair follicles in the ac-tively growing anagen stage. AA is associated with aheterogeneous phenotype. Patients with AA oftendevelop alopecic lesions of various sizes on any site

of haired skin on the body. These foci of alopecia wax and waneeither resolving completely or extending to involve the entirescalp (alopecia totalis) or the entire body (alopecia universalis).Nearly 2% of the general population will develop some form ofAA during their lives (Safavi et al, 1995). The pathogenesis of AAinvolves activation of the cell-mediated immune system by unde-¢ned antigen(s) through the lymphocyte costimulatory cascadewith activated CD4þ and CD8þ T cells targeting anagen stagehair follicles (Carroll et al, 2002). AA is most likely a disorder ofmultifactorial etiologies comprising immunologic, environmen-tal, and genetic components. A strong genetic in£uence is sup-ported by increased concordance in identical twins (Hendren,1949; Weidman et al, 1956; Barsky and Gigli, 1961; Bonjean et al,1968; Cole and Herzlinger, 1984;Werth et al, 1992) or siblings (Ins-ler and Helm, 1989) as well as identi¢cation of families with sev-eral generations of members with AA (Hordinsky et al, 1984; Van

der Steen et al, 1992; Dawn and Kumar, 1996). Human AA geneassociation studies focus on the major histocompatibility locus,HLA, as the most likely region for genes that regulate susceptibil-ity or resistance to AA, as AA is considered to be an autoimmunedisease (Duvic et al, 1991; Morling et al, 1991; Welsh et al, 1994;Colombe et al, 1999; deAndrade et al, 1999; Kavak et al, 2000).Increased association of human AA with other autoimmunediseases in the same individual and/or in blood relatives of af-fected people also suggests that AA patients have an inherited sus-ceptibility to autoimmune diseases (Muller and Winkelmann,1963; Gollinck and Orfanos, 1990; Shellow et al, 1992). AA is nota disease unique to humans. Mice, rats, dogs, horses, cattle, andpossibly other species develop various forms of AA (Sundberget al, 1995; in press; McElwee et al, 1998b). At least eight inbredand one congenic strains of mice develop AA-like diseases, oneof which consistently develops abnormalities of nails as well asalopecia, similar to a subpopulation of human patients with AA(McElwee et al, 1999).Within inbred strains of mice spontaneousAA can approach a frequency of 20% (Sundberg et al, 1994). Thebest characterized and most studied of these mouse strains withAA is the C3H/HeJ inbred strain (Sundberg et al, in press). In thisstudy, genome-wide screens were done, using mapping methodsfor single and multiple loci, in a segregating F2 population fromC3H/HeJ female mice with AA that were crossed with normalC57BL/6J mice. Their progeny were intercrossed to produce F2o¡spring in order to identify loci responsible for susceptibilityto AA. Two genetic intervals were identi¢ed. One interval onmouse chromosome (m. Chr.) 17 includes the homologous HLAinterval associated with human AA. Twelve-month-old congenic

Reprint requests to: John P. Sundberg, D.V.M., Ph.D.,The Jackson Labora-tory, 600 Main Street, Bar Harbor, ME 04609-1500; Email: jps@jax.orgAbbreviations: AA, alopecia areata; H2, mouse major histocompatibility

locus; HLA, human leukocyte antigen; LOD, logarithm of odds ratio;m. Chr., mouse chromosome; QTL, quantitative trait locus.

Manuscript received September 9, 2002; revised November 1, 2002;accepted for publication December 20, 2002

0022-202X/03/$15.00 . Copyright r 2003 by The Society for Investigative Dermatology, Inc.

771

C3H mice that had this interval exchanged with a resistant inter-val did not develop AA, providing support for the mapping re-sults. A candidate interval on m. Chr. 9 was also identi¢ed.

MATERIALS AND METHODS

Mice Female C3H/HeJ mice that spontaneously developed AA andclinically normal C57BL/6J males that never develop AA were obtainedfrom The Jackson Laboratory (Bar Harbor, ME). These mice wereintercrossed to generate F2 mice that were subsequently followed until 1 yof age. Initial evaluation of the ¢rst 462 F2 mice revealed that 10% offemales and 2% of males developed AA. Because of the late onset of AAin male mice (12^15 mo of age for males compared to 6^8 mo for females)and a lower prevalence in males, only females were aged in subsequent F2populations. A total of 1096 F2 female mice were observed for 12 mo todetermine which developed clinical and histologic evidence of AA. Ofthese 138 developed disease. These 138 mice with AA, along with 214clinically normal littermates, were subjected to genome-wide screens. Allwork was performed with approval of The Jackson Laboratory IACUC.Congenic C3H.SW-H2b/SnJ mice, in which the C3H/HeJ-H2k locus was

replaced with H2b, were also obtained fromThe Jackson Laboratory. A totalof 85 retired female mice were aged to 12 mo or older and observed weeklyfor any evidence of alopecia.Mice were maintained in a humidity-, temperature-, and light cycle

(12:12) controlled vivarium under speci¢c pathogen-free conditions(Sundberg et al, 1994; McElwee et al, 1998a). Mice were caged in double-pen polycarbonate cages (330 cm2 £oor area) at a maximum capacity offour mice per pen. Mice were allowed free access to autoclaved food(NIH diet 31, 6% fat) and acidi¢ed water (pH 2.8^3.2). Only female F2were maintained, evaluated weekly, and necropsied at 12 mo of age.

Phenotyping mice All mice were examined for evidence of AA bytrained personnel on a weekly basis. A¡ected mice initially developedfocal alopecia on their ventral abdomen near the rear legs that spread withapproximately lateral symmetry to involve the entire ventral skin.Concurrently, focal areas of alopecia developed randomly on the dorsalskin. These waxed and waned. Over periods of months, the majority ofa¡ected female mice developed generalized alopecia similar to alopeciauniversalis in humans. Mice were euthanized by CO2 asphyxiation andnecropsies were performed. Representative skin was collected from dorsaland ventral trunk, placed on aluminum foil to lay the tissue £at and orientit in a cranial to caudal manner, then ¢xed by immersion in Fekete’s acid^alcohol^formalin, transferred after overnight ¢xation into 70% ethanol,trimmed, embedded in para⁄n, sectioned at 6 mm, stained withhematoxylin and eosin, and examined by an experienced pathologist(JPS). A minimum of three strips of skin from both the dorsal and ventralthoracic regions (six minimum) 41.0 cm in length were examinedhistologically. In mice with AA, anagen stage truncal hair follicles werein¢ltrated or surrounded to various degrees by in£ammatory cells thatwere predominantly lymphocytes (Sundberg et al, 1994). Mice were scoredusing both gross and/or histologic criteria as having AA or being clinicallynormal. Peripheral blood £uorescence-activated cell sorter analyses or otherhormonal and biochemical assays had not identi¢ed a scoreable dichotomyto permit quantitative trait analyses (Sundberg et al, 1994).

Genotyping DNAwas isolated from tail tips amputated at the time ofnecropsy. DNAwas extracted using a DNeasy kit (Qiagen,Velencia, CA).DNAs from all F2 segregants were genotyped by polymerase chainreaction using microsatellite markers purchased from Research Genetics(Huntsville, AL). Thermal cyclers (MJ Research, Cambridge, MA) wereused, and products were separated on 4% Metaphor/LE (3:1) agarose gels(BMA Biomedicals). The Jackson Laboratory’s allele typing service alsoprovided multiplex genotyping data by using ABI 370 instrumentation(http://jaxmice.jax.org). To distinguish between C3H and B6, 211informative microsatellite markers were used, initially providing coverageof at least every 20 cM on each autosome and m. Chr. X (Mahler et al, 1999;Farmer et al, 2001). The list of markers is available on request. Linkage mapsand marker positions reported are based onThe Jackson Laboratory’s onlineMouse Genome Informatics resource (www.informatics.jax.org).

Statistical analysis Genome scans were carried out by using bothmarker regression and interval mapping analysis. Because similar resultswere obtained by using both methods, only the results of intervalmapping are presented. Logarithm of odds ratio (LOD) scores for maine¡ects were computed at 2 cM intervals across the autosomal genome,and LOD scores for pairwise scans were done at 5 cM resolution.Interpretation of results was simpli¢ed by focusing on the individual trait

of AA versus clinical normality. Pairwise interactions among loci wereassessed by using genome-wide scans (Sen and Churchill, 2001; Sugiyamaet al, 2001). Brie£y, the signi¢cance of a pair of loci was assessed ¢rst byusing an overall F test (with 8 and 399 degrees of freedom) for a two-wayANOVA model of genotypic e¡ects.When this test exceeded the genome-wide threshold as established by permutation analysis, secondary tests werecarried out to test for an interaction e¡ect. If an interaction was indicated,testing was stopped and an interacting pair was declared. If no signi¢cantevidence for interaction was uncovered, each of the single models wascompared with the two quantitative trait locus (QTL) additive model.Both tests must be signi¢cant to indicate an additive pair. Thesesecondary tests were carried out at a nominal 0.005 level to account fortesting of more than one but fewer than 10 potential pairs. All softwareused in this analysis are available at www.jax.org/research/churchill or byrequest (GAC).

RESULTS

Frequency of AA Initially we used an intercross betweenC3H/HeJ females with AA and CAST/EiJ. This strategy wouldhave yielded numerous polymorphisms in regions of potentialinterest. None of the 295 F2 mice from this cross developed AA,however. These studies were terminated and we reverted to anintercross strategy between a¡ected C3H/HeJ females andC57BL/6J males, a strain that does not develop AA. The ¢rstgroup of F2 mice generated 23 out of 229 (10%) females and¢ve out of 233 (2%) males with AA (total 28 of 462, 6%).Subsequently only F2 females were aged, yielding 138 mice(12.6%) with AA from a total of 1096 mice. A genome-widescreen used 211 microsatellite markers on 138 AA and 214clinically normal mice. At the time of necropsy therewas concordance between clinically evident alopecia andhistologically con¢rmed AA.

Genetic linkage analysis Two intervals were identi¢ed thatwere statistically sign¢cant or suggestive of linkage (Figs 1^3,Table I). The m. Chr. 17 interval includes H2 (human HLAortholog, Fig 4), tumor necrosis factor a (Tnfa), lymphotoxin a

Figure 2. Fine mapping, at 2 cM intervals, of m. Chr. 9 revealedone, possibly two, suggestive intervals involved in mouse AA.LOD score is the solid line and posterior probability density score is thedotted line.

Figure1. A scan at 5 cM resolution revealed a signi¢cant interval onm. Chr. 17 and a suggestive one on m. Chr. 9. Based on a permutationtest, LOD scores of 1.8 or 2.2 correspond to a p-value of 0.10 and 0.05,respectively.

772 SUNDBERG ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

and b (Lfta, Lftb), and many other genes of interest (http://informatics.jax.org). This 12^20 cM interval is still too large toconsider candidate genes. The broad range of statisticalsigni¢cance on m. Chr. 17 suggests that multiple genes may beinvolved. On m. Chr. 9 there was an interval suggestive oflinkage (LOD score of 2.0). Genes within this interval code forproteins involved with T cell function (CD3D, CD3G, CD3E,THY1, and regulatory T cell molecule, CRTAM), cytokines thatregulate in£ammation (IL10RA), or neural cell adhesionmolecule (NCAM1). Markers for NCAM are commonly usedby some laboratories to study mouse hair cycles (Paus et al, 1999).The human homolog is 11q22^q24 (Fig 5). Pairwise analysesfailed to identify epistatic interactions with other genes.None of the C3H.SW-H2b congenic female mice aged and

observed weekly until they were over 12 mo of age developedAA, in contrast to C3H/HeJ mice that are H2k that routinelydevelop AA at rates of up to 20% in females. The congenicinterval encompasses the interval identi¢ed as signi¢cant on m.Chr. 17.

DISCUSSION

In previous research using DNA derived from humans, associa-tion studies were conducted to con¢rm or refute a hypothesis.The studies focused on a speci¢c gene to examine allele fre-quency as associated with AA phenotype expression. In thismouse genetic study, no hypotheses as to genes or loci involvedin AAwere predetermined. The entire genome was scanned withmicrosatellite markers at approximately 20 cM intervals and can-didate loci were identi¢ed by statistical analysis. This avoids anypotential bias in results that may be introduced in gene associa-tion studies through the preselection of genes and loci for exam-ination.Two intervals were identi¢ed in this mouse model linkedto spontaneous AA, one of which is in the H2 region homolo-gous to HLA in humans associated with AA, further validatingthis mouse model. The maximum LOD score was calculated by¢tting results into a one, two, or three QTL model for a binarytrait. If the di¡erence between LOD scores for these criteria is41.6, then it is interpreted that two or more QTLs are locatedin that interval. There was not statistical evidence, based on thenumbers of mice available for analysis, for multiple QTLs at theintervals on either m. Chr. 9 or 17 in this study.There was a clear bias in AA phenotype expression in females

in both the inbred C3H/HeJ and F2 hybrid mice compared to

males. Early analyses of small groups of F2 mice did identifyan interval on m. Chr. X around DXMit172 (48.7 cM), justdistal to the androgen receptor gene (Ar, 36.0 cM). Signi¢cancedisappeared when all mice in this study were analyzed. Similar

Figure 3. Fine mapping, at 2 cM intervals, of m. Chr. 17 revealed ahighly signi¢cant interval that encompasses the MHC 2 complex.LOD score is the solid line and posterior probability density score is thedotted line.

Figure 4. M. Chr. 9 map with location and symbols for homologoushuman genes within the AA susceptibility interval.The marker withthe peak LOD score is in the gray box. Potential candidate genes are in boldtype. Map modi¢ed from http://www.informatics.jax.org.

Table I. Signi¢cant or suggestive intervals involved in mouse AA

Chr Max LODfor one QTL

Max LODfor two QTLs

Max LODfor three QTLs

95%CI (cM)

% Varianceexplaineda

Peak marker Markerlocation

17 10.9 11.5 12.4 12^20 13.5 D17Mit134 16.9 cM9b 2 2.7 3.2 0^35 2.6 D9Mit206 20.0 cMaThe phenotypic variance explained is based on a single QTL model.bSuggestive QTL.

GENETICS OFALOPECIA AREATA 773VOL. 120, NO. 5 MAY 2003

results have been reported in the human AA gene associationstudies when more patients were tested (Barahamani et al,2002). The bias in disease expression is probably due to gonadalhormones. A sex bias in the frequency and age of female miceat AA onset has been stated in previous published studies(Sundberg et al, 1994). A study involving estradiol or testosteronesupplementation indicated that testosterone exerts an inhibitorye¡ect on the development of AA (McElwee et al, 2001). Assuch, the low level of AA in male F2 mice is probably theresult of testosterone, possibly acting to depress immune systemactivity.In the intercross study, gene mapping initially identi¢ed two

intervals, one being very signi¢cant on m. Chr. 17 and a secondwith suggestive levels of signi¢cance on m. Chr. 9. The intervalof highest signi¢cance on m. Chr. 17 corresponds directly tothose found in human AA gene association studies. Human AAhas been shown to be associated with several MHC class II alleles,HLA-DQ3, DR4, and DR11. Alleles DQB1n03 and DRB1n04 aresigni¢cantly associated with all forms of AAwhereas DRB1n1104,DRB10401, and DQB1n0301 are speci¢cally associated with long-standing and/or extensive alopecia totalis and universalis (Co-lombe et al, 1995a; 1995b; Duvic et al, 1995). The highest signi¢-cance was identi¢ed for D17Mit134 at 16.9 cM, less than 2 cMfrom the homologous mouse genes for HLADQ (H2-Ab1, 18.64cM) and HLADR (H2-Eb2, 18.67 cM) (www.informatics.jax.org).Many other human autoimmune diseases have been shown to beassociated with a particular MHC class II haplotype (Theo¢lo-poulos, 1995). Certain haplotypes seem to be associated with pre-disposal of the individual towards autoimmunity (Campbelland Milner, 1993; Corzo et al, 1995; Tomlinson and Bodmer,1995; Blum and Miller, 2000). AA may have MHC class restric-tion (Kianto et al, 1977; Hacham-Zadeh et al, 1981; Averbakhand Pospelo, 1986; Frentz et al, 1986; Orecchia et al, 1987; Duvicet al, 1991; Kuntz et al, 1997). A variety of immunoregulatorygenes are also found within this genetic interval, which mayultimately prove to be important in this and other autoimmunediseases.The second interval of suggestive signi¢cance on m. Chr. 9

does not correspond to any other genes or intervals that are sug-gestive in human AA gene association studies such as IL1 (Coxet al, 1998; Tazi-Ahnini et al, 2001), IL1rn (Tarlow et al, 1994;Barahamani et al, 2002), IL1b (Galbraith et al, 1999; Tazi-Ahniniet al, 2001), Km1 (Galbraith and Pandey, 1989; Dugoujon et al,1992; Dugoujon and Cambon-Thomsen, 1995; Galbraith et al,1999), or Mx1 (Tazi-Ahnini et al, 2000). The identi¢cation of amouse m. Chr. 9 interval for an AA susceptibility allele is poten-tially an important discovery, although the low LOD score tem-pers enthusiasm. The locus de¢ned in this genome-wide screen isstill quite large and potentially encompasses numerous candidategenes. These results suggest that m. Chr. 9 is worthy of closer ex-amination, however, especially with CRTAM, CD3D, CD3E,CD3G, andTHY1, all genes associated withTcell function, beinglocated within this interval.Preliminary analyses of our data suggested more intervals of

signi¢cance, up to four, with one potential epistatic pair. As withsome of the human AA association studies, however (Barahamaniet al, 2002), signi¢cance diminished or disappeared as largergroups were analyzed.The C3H/HeJ AA mouse model has signi¢cant linkage with

the genetic interval that encompasses the major histocompatibil-ity loci, as also seen with human AA gene association studies,thus providing additional support for the validity of this model.C3H.SW-H2b/SnJ congenic mice have had the C3H/HeJ-H2k re-placed by H2b (as found in resistant C57BL/6J mice). NormalC3H/HeJ mice with H2k develop AA by 12 mo of age to variousdegrees based on sex. None of these congenic mice developedAA, supporting this interval as being important in AA. Crosses,currently in progress, between C3H/HeJ mice with clinically evi-dent AA and C3H.SW-H2b/SnJ congenic mice will help dissectout the genes involved in this region, something that cannot beeasily done with human families.

Figure 5. M. Chr. 17 map with location and symbols for homolo-gous human genes within the AA susceptibility interval.The markerwith the peak LOD score is in the gray box. Potential candidate genes are inbold type. Bracketed interval on the left indicates the approximate mapped lim-its for the C3H.SW-H2b/SnJ congenic interval. Map modi¢ed from http://www.informatics.jax.org.

774 SUNDBERG ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

This work was supported by grants fromThe National Alopecia Areata Foundation(JPS), National Institutes of Health (AR43801 and RR173 to JPS andP30AR41443 to LEK), Council for Nail Research (JPS), Dermatology Founda-tion/Glaxo Dermatology Fellowship (KJM), and Department of Veterans A¡airs(LEK). Institutional services (The Jackson Laboratory) were supported by a NationalCancer Institute Cancer Center Support Grant (CA34196).

REFERENCES

deAndrade M, Jackow CM, Dahm N, Hordinsky M, Reveille JD, Duvic M: Alope-cia areata in families: Association with the HLA locus. J Invest Dermatol SympProc 4:220^223, 1999

Averbakh EV, Pospelo LE: HLA antigens in patients with alopecia areata.Vestn Der-matol Venereol 1:24^26, 1986

Barahamani N, deAndrade M, Slusser J, Zhang Q, Duvic M: Interleukin-1 receptorantagonist allele 2 and familial alopecia areata. J Invest Dermatol 118:335^337,2002

Barsky S, Gigli I: Alopecia areata in twins. Arch Dermatol 83:224^225, 1961Blum A, Miller H: The major histocompatibility complex and in£ammation. South-

ern Med J 93:169^171, 2000Bonjean M, Prime A, Avon P: Pelade chez deux jumeaux homozygotes (sic). Lyon

Med 219:1852^1853, 1968Campbell RD, Milner CM: MHC genes in autoimmunity. Curr Opin Immunol

5:887^893, 1993Carroll J, McElwee KJ, King LE, Byrne M, Sundberg JP: Gene array pro¢ling and

immunomodulation studies de¢ne a cell mediated immune response underly-ing the pathogenesis of alopecia areata in a mouse model and humans.J Invest Dermatol 119:392^402, 2002

Cole GW, Herzlinger D: Alopecia universalis in identical twins. Int J Dermatol 23:283,1984

Colombe BW, Price VH, Khoury EL, Garovoy MR, Lou CD: HLA class II antigenassociations help to de¢ne two types of alopecia areata. J Am Acad Dermatol33:757^764, 1995a

Colombe BW, Price VH, Khoury EL, Lou CD: HLA class II alleles in long-standingalopecia totalis/alopecia universalis and long-standing patchy alopecia areatadi¡erentiate these two clinical groups. J Invest Dermatol 104 (Suppl):4s^5s, 1995b

Colombe BW, Lou CD, Price VH:The genetic basis of alopecia areata: HLA associa-tions with patchy alopecia areata versus alopecia totalis and alopecia universalis.J Invest Dermatol Symp Proc 4:216^219, 1999

Corzo D, Salazar M, Granja CB, Yunis EJ: Advances in HLA genetics. Exp ClinImmunogenet 12:156^170, 1995

Cox A, Camp NJ, Nicklin MJH, diGiovine FS, Du¡ GW: An analysis of linkagedisequilibrium in the interleukin-1 gene cluster, using a novel grouping meth-od for multiallelic markers. AmJ Hum Genet 62:1180^1188, 1998

Dawn G, Kumar B: Pro¢le of alopecia areata in northern India. Int J Dermatol 35:22^27, 1996

Dugoujon JM, Cambon-Thomsen A: Immunoglobulin allotypes (GM and KM) andtheir interactions with HLA antigens in autoimmune diseases: a review. Auto-immunity 22:245^260, 1995

Dugoujon JM, Guitard E, Senegas MT: Gm and Km allotypes in autoimmune dis-eases. G Ital Cardiol 22:85^95, 1992

Duvic M, Hordinsky MK, Fiedler VC, O’BrienWR,Young R, Reveille JD: HLA-Dlocus associations in alopecia areata: DRw52a may confer disease resistance.Arch Dermatol 127:64^68, 1991

Duvic M,Welsh EA, Jackow C, Papadopoulos E, Reveille JD, Amos C: Analysis ofHLA-D locus alleles in alopecia areata patients and families. J Invest Dermatol104 (Suppl.):5s^6s, 1995

Farmer MA, Sundberg JP, Bristol IJ, Churchill GA, Li R, Elson CO, Leiter EH:A major quantitative trait locus on chromosome 3 controls colitis severity inIL-10-de¢cient mice. Proc Natl Acad Sci USA 98:13820^13825, 2001

Frentz G,Thomsen K, Jakobsen BK, Svejgaard A: HLA-DR4 in alopecia areata. JAmAcad Dermatol 14:129^130, 1986

Galbraith GMP, Pandey JP: Km1 allotype associated with one subgroup of alopeciaareata. AmJ Hum Genet 44:426^428, 1989

Galbraith GM, Palesch Y, Gore EA, Pandey JP: Contribution of interleukin 1b andKM loci to alopecia areata. Hum Hered 49:85^89, 1999

Gollinck H, Orfanos CE: Alopecia areata: Pathogenesis and clinical picture. In: Or-fanos CE, Happle R (eds) Hair and Hair Diseases. Berlin: Springer-Verlag,1990:p 529^569

Hacham-Zadeh S, Brautbar C, Cohen C, Cohen T: HLA and alopecia areata inJerusalem.TissueAntigens 18:71^74, 1981

Hendren S: Identical alopecia areata in identical twins. Arch Dermatol 60:793^795, 1949

Hordinsky MK, Hallgren H, Nelsen D, Filipovich AH: Familial alopecia areata:HLA antigens and autoimmunity formation in an American family. ArchDermatol 120:464^468, 1984

Insler MS, Helm CJ: Alopecia areata including the cilia and eyebrows of two sisters.Ann Ophthalmol 21:451^453, 1989

Kavak A, Baykal C, Ozarmagan G, Akar U: HLA and alopecia areata. Int J Dermatol39:589^592, 2000

Kianto U, ReunalaT, Karvonen J, Lassus A,Tiilikainen A: HLA-B12 in alopecia area-ta. Arch Dermatol 113:1716, 1977

Kuntz BM, Selzle D, Braun-Falco O, Scholz S, Albert ED: HLA antigens in alopeciaareata. Arch Dermatol 113:1717, 1997

Mahler M, Sundberg JP, Birkenmeier EH, Bristol JJ, Elson CO, Leiter EH: Geneticanalysis of susceptibility to dextran sulfate sodium-induced colitis in mice.Genomics 55:147^156, 1999

McElwee KJ, Boggess D, Burgett B, Bates R, Bedigian HG, Sundberg JP, King LE:Murine cytomegalovirus is not associated with alopecia areata in C3H/HeJmice. J Invest Dermatol 110:986^987, 1998a

McElwee KJ, Boggess D, Olivry T, et al: Comparison of alopecia areata in humanand nonhuman mammalian species. Pathobiology 66:90^107, 1998b

McElwee K, Boggess D, Miller J, King LE, Sundberg J: Spontaneous alopeciaareata-like hair loss in one congenic and seven inbred laboratory mouse strains.J Invest Dermatol Symp Proc 4:202^206, 1999

McElwee KJ, Silva K, Beamer WG, King LE, Sundberg JP: Melanocyte and gonadactivity as potential severity modifying factors in C3H/HeJ mouse alopeciaareata. Exp Dermatol 10:420^429, 2001

Morling N, Frentz G, Fugger L, Georgsen J, Jakobsen B, Odum N, Svejgaard A:DNA polymorphism of HLA class II genes in alopecia areata. Dismarkers9:35^42, 1991

Muller SA, Winkelmann RK: Alopecia areata. An evaluation of 736 patients. ArchDermatol 88:290^297, 1963

Orecchia G, Belvedere MD, Martinetti M, Capelli E, Rabbiosi G: Human leukocyteantigen region involvement in the genetic predisposition to alopecia areata.Dermatologica 175:10^14, 1987

Paus R, Muller-Rover S, van der Veen C, et al: A comprehensive guide to the recog-nition and classi¢cation of distinct stages of hair follicle morphogenesis. J InvestDermatol 113:523^532, 1999

Safavi KH, Muller SA, SumanVJ, Moshell AN, Melton LJ: Incidence of alopecia areatain Olmstead County, Minnesota 1975^89. Mayo Clin Proc 70:628^633,1995

Sen S, Churchill GA: A statistical framework for quantitative trait mapping. Genetics159:371^387, 2001

ShellowWV, Edwards JE, Koo JY: Pro¢le of alopecia areata: A questionnaire analysisof patient and family. Int J Dermatol 31:186^189, 1992

Sugiyama F, Churchill GA, Higgins DC, Johns C, Makaritsis KP, Gavras H, PaigenB: Concordance of murine quantitative trait loci for salt-induced hypertensionwith rat and human loci. Genomics 71:70^77, 2001

Sundberg JP, CordyWR, King LE: Alopecia areata in aging C3H/HeJ mice. J InvestDermatol 102:847^856, 1994

Sundberg JP, Oliver RF, McElwee KJ, King LE: Alopecia areata in humans andother mammalian species. J Invest Dermatol 104:32s^33s, 1995

Sundberg JP, McElwee KJ,Whiting D, King LE: Alopecia areata: spontaneous andskin graft-induced mouse models. In: Chan LS (ed) Animal Models of HumanIn£ammatory Skin Diseases. Boca Raton, FL: CRC Press, in press

Tarlow JK, Clay FE, Cork MJ, Blakemore AI, McDonagh AJ, Messenger AG, Du¡GW: Severity of alopecia areata is associated with a polymorphism in the inter-leukin-1 receptor antagonist gene. J Invest Dermatol 103:387^390, 1994

Tazi-Ahnini R, McDonagh A, diGiovine F, et al: Structure and polymorphismof the human gene for the interferon-induced p78 protein (MX1): Alopeciaareata association with the Down syndrome region. J Invest Dermatol 114:827,2000

Tazi-Ahnini R, McDonagh AJ, Cox A, et al: Association analysis of IL1A and IL1Bvariants in alopecia areata. Heredity 87:215^219, 2001

Theo¢lopoulos AN: The basis of autoimmunity: Part II. Genetic predisposition.ImmunolToday 16:150^159, 1995

Tomlinson IP, Bodmer WF: The HLA system and the analysis of multifactorial ge-netic disease.Trends Genet 11:493^498, 1995

Van der Steen P,Traupe H, Happle R, Boezeman J, Strater R, Hamm H:The geneticrisk for alopecia areata in ¢rst degree relatives of severely a¡ected patients. Anestimate. Acta DermVenereol (Stockh) 72:373^375, 1992

Weidman AI, Zion LS, Mamelok AE: Alopecia areata occurring simultaneously inidentical twins. Arch Dermatol 74:424^426, 1956

Welsh EA, Clark HH, Epstein SZ, Reveille JD, Duvic M: Human leukocyte anti-gen-DQB01n03 alleles are signi¢cantly associated with alopecia areata. J InvestDermatol 103:758^763, 1994

Werth VP, White WL, Sanchez MR, Franks AG: Incidence of alopecia areata andlupus erythematosus. Arch Dermatol 128:368^371, 1992

GENETICS OFALOPECIA AREATA 775VOL. 120, NO. 5 MAY 2003