2586 Research Article

IntroductionMiz1 is a zinc-finger transcription factor, which has beenidentified as a Myc-binding protein by a yeast two-hybridapproach (Peukert et al., 1997). The N-terminus of Miz1 bearsa POZ/BTB domain (Bardwell and Treisman, 1994), which isessential for both transactivation and Miz1-mediated cell cyclearrest (Peukert et al., 1997; Herold et al., 2002; Adhikary etal., 2005). Several proteins bind to and regulate Miz1 activity,including Bcl-6 (Phan et al., 2005), the topoisomerase IIbinding protein TopBP1 (Herold et al., 2002), the ubiquitinligase HectH9 (Adhikary et al., 2005), Smad protein complexescontaining Smad3 and Smad4 (Seoane et al., 2001) and Myc(Peukert et al., 1997).

A number of observations suggest that Miz1 is involved inrestricting proliferation in response to antimitogenic signals;for example, Miz1 is required for upregulation of the cellcycle inhibitor p21cip1 in response to DNA damage and inmodels of cellular differentiation (Herold et al., 2002; Phanet al., 2005; van de Wetering et al., 2002; Wu et al., 2003;Herold et al., 2002). A second target of Miz1-dependenttransactivation is the promoter of the gene encoding thecyclin-dependent kinase inhibitor p15Ink4b. Since p15ink4bis regulated by transforming growth factor beta (TGF�), thesefindings suggested that Miz1 might have a role in TGF�-dependent gene regulation (Seoane et al., 2001; Staller et al.,2001).

Inhibition of epithelial cell proliferation by TGF� isparalleled by a decrease in Myc expression (Pietenpol et al.,1990) that is required for the activation of the p15ink4b geneand exit from the cell cycle (Warner et al., 1999). Mycrepresses transactivation by Miz1, suggesting thatdownregulation of Myc allows activation of p15ink4b by Miz1,Smad3 and Smad4 (Seoane et al., 2001; Staller et al., 2001).This model is supported by the finding that Myc but not amutant, MycV394D, which cannot bind to Miz1 (Herold et al.,2002), represses p15ink4b in primary murine keratinocytes(Gebhardt et al., 2006). In addition, several other genes areregulated in response to TGF� in the same manner asp15ink4b, suggesting that TGF� induces a complex geneticprogram through the relief of Myc-mediated gene repression(Gebhardt et al., 2006). Most of the identified genes areinvolved in cell-cell or cell-matrix adhesion and some of them,such as integrin receptors, play a role in the maintenance of theepidermal stem cell compartment (Watt, 2002; Alonso andFuchs, 2003; Moore and Lemischka, 2006).

In previous experiments, we showed that constitutive Miz1(Zbtb17)-knockout mice die at E7.5, demonstrating that Miz1is essential for early embryonic development (Adhikary et al.,2003). Here we have explored the function of Miz1 in the skinwith the help of mice that carry a conditional allele of Miz1(C. Kosan, N. Fang, M. Godmann et al., unpublished). Weremoved the Miz1 POZ/BTB transactivation domain in

Previous work has implicated the Myc-bindingtranscription factor Miz1 in the control of keratinocyteproliferation and in the cellular response to TGF�. Miz1 isexpressed in basal keratinocytes of the interfollicularepidermis and in hair follicles. Here we have conditionallyknocked out the POZ/BTB transactivation domain of Miz1in keratinocytes using a keratin 14 (K14)-Cre mouse deleterstrain. K14Cre+/Miz1lox/lox mice have rough fur as a resultof altered hair follicle orientation, irregular hairpigmentation and disturbed hair fiber structure. A regionalthickening of the epidermis at the hair funnel orifice wasaccompanied by suprabasal proliferation, indicating adelayed exit of keratinocytes from the cell cycle. Inaddition, the catagen of the hair cycle was delayed in

K14Cre+/Miz1lox/lox mice and intrafollicular keratinocyteproliferation was increased. In aged K14Cre+/Miz1lox/lox

animals, the number of hair follicles remained unchangedbut the number of visible hairs, especially of zigzag hairs,was reduced and a pigmentary incontinence into the dermisdeveloped. Our data show that Miz1 is involved incontrolling proliferation and differentiation in hair folliclesand in hair fiber morphogenesis.

Supplementary material available online athttp://jcs.biologists.org/cgi/content/full/120/15/2586/DC1

Key words: Myc, Miz1, Transforming growth factor �1, Epidermis,Hair follicle, Catagen

Summary

Miz1 is required for hair follicle structure and hairmorphogenesisAnneli Gebhardt1, Christian Kosan2, Barbara Herkert3, Tarik Möröy2,4, Werner Lutz3, Martin Eilers3 andHans-Peter Elsässer1,*1Institute for Cell Biology, University of Marburg, Robert-Koch-Str. 6, 35033 Marburg, Germany2Institut de recherches cliniques de Montreal, 110 Avenue des Pins West, Montreal, QC, H2W 1R7, Canada3Institute for Molecular Biology and Tumor Research (IMT), University of Marburg, Emil-Mannkopff-Str.2, 35033 Marburg, Germany4Institut de recherches cliniques de Montreal, Department de Microbiologie et Immunologie, Université de Montréal, 110 Avenue des Pins West,Montreal, QC, H2W 1R7, Canada*Author for correspondence (e-mail: elsaesse@mailer.uni-marburg.de)

Accepted 26 May 2007Journal of Cell Science 120, 000-000 Published by The Company of Biologists 2007doi:10.1242/jcs.007104

Jour

nal o

f Cel

l Sci

ence

2587Miz1 POZ/BTB domain knockout in the epidermis

keratinocytes, using a mouse strain that expresses Crerecombinase under the control of the keratin 14 (K14)promoter (Huelsken et al., 2001). Our data show that Miz1 hasan essential role in controlling exit from the cell cycle duringthe hair cycle and during epithelial morphogenesis, as well asin hair morphogenesis.

ResultsKeratinocyte-specific ablation of the Miz1 POZ/BTBdomain in miceMiz1 is expressed in basal keratinocytes of the interfollicularepidermis (Gebhardt et al., 2006) and of the hair follicle,during all stages of the hair cycle (supplementary material Fig.S1A-C). We generated mice that carry a conditional allele ofMiz1 in which the first two coding exons, which encode aPOZ/BTB domain (Bardwell and Treisman, 1994), are flankedby loxP sites (C. Kosan, N. Fang, M. Godmann et al.,unpublished) (supplementary material Fig. S1D). Mice thatare homozygous for the Miz1 lox allele and that express Cre-recombinase under control of the keratin 14 promoter(K14Cre) (Huelsken et al., 2001; Wang et al., 1997), lack thePOZ/BTB domain of Miz1 in keratinocytes of theinterfollicular epidermis, the hair follicle and the sebaceousglands. Recombination starts at about E15 in K14Cre+ mice(Huelsken et al., 2001). Allele-specific PCR using genomicDNA of keratinocytes that were isolated from back skin of 1-day-old mice, without subsequent culture, showed thatrecombination was essentially complete in these cells (Fig.1A; for primer position see supplementary material Fig. S1D).This result was confirmed by RT-PCR analysis using RNAfrom keratinocytes taken from 1-day-old mice and cultured for3-4 days (Fig. 1B). In western blots of immunoprecipitatedproteins from keratinocyte lysates a truncated Miz1 translationproduct was detected using a monoclonal antibody raisedagainst the C-terminal part of Miz1, showing that therecombined allele results in the production of an N-terminallytruncated version of Miz1 lacking specifically the POZ/BTBdomain (Fig. 1C). We conclude that the Miz1 POZ/BTBdomain was specifically ablated by K14Cre in keratinocytes

from day E15 onwards, when the K14 promoter is usuallyactivated (Huelsken et al., 2001).

Deletion of the Miz1 POZ/BTB domain in keratinocytesimpairs late hair follicle and hair morphogenesisBoth K14Cre+/Miz1lox/lox and K14Cre+/Miz1lox/�POZ miceexhibited rough fur with 100 % penetrance (Fig. 2A; 115 miceanalyzed). This became obvious as soon as the first hairs werevisible (at about P5; supplementary material Fig. S2). Initially,the rough areas occurred in a striped pattern, which may reflectthe initial expression pattern of the Cre recombinase (Huelskenet al., 2001). The striped pattern disappeared when mice grewolder. By contrast, there were no obvious morphologicalabnormalities in heterozygous K14Cre+/Miz1lox/+ andK14Cre–/Miz1lox/�POZ mice (36 animals), or inK14Cre–/Miz1lox/lox and K14Cre–/Miz1lox/+ control mice (106animals).

Histology revealed that the length of hair follicles showed ahigher variation in K14Cre+/Miz1lox/lox mice compared withcontrol littermates (Fig. 2B-K), and that the orientation of somefollicles in respect to the skin surface was altered (Fig. 2K). Inaddition, in some areas, cyst-like hair follicles were observedcontaining hair remnants (Fig. 2H). These cystic alterationsoccurred predominantly in the upper part of the follicles andthe epithelium of the cyst-like structures was continuous withthe interfollicular epidermis as proved by serial sections (datanot shown). Similar structures called pseudocysts have beendescribed in nude mice (Meier et al., 1999). In addition to avariation in follicle orientation and length, the expression ofkeratin 14 (K14) was altered compared with wild-type skin.Usually, cells from the basal cell layer of the outer root sheetexpress K14 with a decreasing expression level towards thebottom of the follicle. In K14Cre+/Miz1lox/lox mice, we alsoobserved a strong expression of K14 in the lower part of somehair follicles (Fig. 2D,E,I-K). This ectopic expression wasespecially abundant in follicles delayed in catagen (Fig. 2I-K).

In about 30-40% of the hair follicles fromK14Cre+/Miz1lox/lox mice, the number of interfollicularepidermal cell layers was increased and cornified layers were

Fig. 1. Conditional knockout of Miz1POZ/BTB domain in keratinocytes. (A) K14cremediates efficient recombination of the Miz1lox allele in primary keratinocytes. Forgenotyping, PCR was performed on genomicDNA from keratinocytes that were isolatedfrom back skin of 1-day-old mice withoutsubsequent culture, using primers indicated insupplementary material Fig. S1. Note that inK14Cre+/Miz1+/lox keratinocytes, the wild-typefragment using primers 1 and 5 is inefficientlyamplified because of competition in the PCRby the shorter fragment. The wild-type andloxP allele fragment amplified with primers 1and 5 (about 1400 bp) differed only in 70 bpand could not be distinguished. (B) Levels ofMiz1 mRNA in keratinocytes K14Cre+/Miz1lox/lox and control mice. Miz1 mRNA was almost undetectable in K14Cre+/Miz1lox/lox keratinocytesusing primers 3 and 4, which are located in the region coding for the POZ/BTB domain. By contrast, when primers 6 and 7, which are locatedoutside the POZ/BTB domain coding region, were used, a signal was obtained as in control mice, indicating the presence of a truncated Miz1transcript lacking the POZ/BTB coding region. (C) Western blot analysis, using the anti-Miz1 antibody 10E2, of proteins immunoprecipitatedfrom keratinocyte lysates using the anti-Miz1 antibody H-190. A truncated Miz1 protein could be detected in K14Cre+/Miz1lox/lox andK14Cre+/Miz1+/lox keratinocytes (arrow). Arrowhead indicates Miz1.

Jour

nal o

f Cel

l Sci

ence

2588

thickened (Fig. 3A,B). These lesions were often located at thetransition between interfollicular and intrafollicular epidermisat the edge of the hair funnel. In these regions, the expressionof keratin 1 was seen throughout the suprabasal layers (Fig.3F) and the expression of filaggrin was seen throughout thecornified layer (Fig. 3H; red). Although the expression patternof K1 and filaggrin matched the normal distribution insuprabasal cells and in cells of the cornified layer, respectively,the expression of these proteins in the epidermis was increasedowing to the increased thickness of those layers (Fig. 3F,H). Inaddition, it is remarkable that K1 expression extended from theinterfollicular epidermis to the cyst-like structures, whichseemed to be derived from the hair follicles. Furthermore,expression of K14 (Fig. 3G,H; green) and of Ki67 (Fig. 3C,D;red), a marker of cell proliferation, was not restricted to thebasal cell layer in K14Cre+/Miz1lox/lox mice, but extended to thesuprabasal cells. These observations show that keratinocyteslocated at the orifice of the hair funnel have a delayed exit fromthe cell cycle. They may reflect an increased proliferativecompartment, because proliferating cells are exclusivelylocated in the basal epidermal layer of normal skin (Fuchs,1995). Furthermore, the thickened epidermis and cornifiedlayer indicate an enlarged compartment of differentiation.

Sebaceous glands were larger in mice overexpressing Myc

Journal of Cell Science 120 (15)

(Arnold and Watt, 2001), and showed a 129-fold upregulationof the sebaceous cell marker brain and skin serine protease(BSSP) in the skin (Frye et al., 2003). In order to test whetherthe structure of sebaceous glands is affected inK14Cre+/Miz1lox/lox mice we morphometrically measured thearea of sebaceous glands per hair follicle. We found a 1.5-foldincrease of sebaceous glands in K14Cre+/Miz1lox/lox mice(742±238 �m2/follicle in control mice versus 1093±206�m2/follicle in K14Cre+/Miz1lox/lox mice), but this differencewas not statistically significant. RT-PCR using primers againstBSSP did not reveal a different expression of this proteinbetween control and K14Cre+/Miz1lox/lox mice (data notshown).

Deletion of the Miz1 POZ/BTB domain in keratinocytesimpairs cycling of hair folliclesOn about day 15 after birth (P15) murine hair folliclessynchronously undergo a phase of regression called catagen(Botchkarev and Paus, 2003; Foitzik et al., 2000). During thisprocess, which is finished by about P19, the length of the hairfollicle is shortened and the remnant follicle lies completely inthe dermis. Catagen is delayed in TGF�-null mice (Foitzik etal., 2000) and almost completely blocked in Smad4-knockoutmice (Yang et al., 2005). To assess whether the hair cycle in

Fig. 2. Phenotype and histological analysis of K14Cre+/Miz1lox/lox mice. (A) Eight-day-old (P8) K14Cre+/Miz1 lox/lox mice (foreground) exhibitrough fur compared with control littermates. (B,C) Histological analysis of hematoxylin-stained sections revealed an impaired arrangement andwavy structure of hair follicles (C), when compared with sections taken from control mouse skin (B). (D) Sections from control mice (anagen)stained with an antibody against K14 (red). Expression of K14 was observed in the basal layer of the epidermis in the entire hair follicle. Thiswas strong in the upper part of the hair funnel and decreased towards the hair bulb. (E) K14Cre+/Miz1lox/lox mice showing strong K14expression extended into the lower part of some hair follicles (arrows). (F,G) Boxed areas of E at a higher magnification. (H) In addition, cyst-like hair follicles occurred with hair remnants. (I-K) Ectopic strong expression of K14 in hair follicles of K14Cre+/Miz1lox/lox mice which weredelayed in the catagen stage. Note the extreme misorientation of one follicle in K (arrow). Bars, 400 �m (B,C,D,E,I,J); 100 �m (H,K); 50 �m(F,G).

Jour

nal o

f Cel

l Sci

ence

2589Miz1 POZ/BTB domain knockout in the epidermis

K14Cre+/Miz1lox/lox mice is altered, we counted the number ofhair follicles extending into the subcutis at P19 (Fig. 4A,B).The percentage of such follicles was significantly (P<0.0001)increased in K14Cre+/Miz1lox/lox mice (up to 58%; medianvalue 35.7%) compared with control mice (up to 8%; medianvalue 3.4%) (Fig. 4I; P18/19). Follicular keratinocytesproliferate in anagen, but not in catagen of the hair cycle.Consistent with this notion, the BrdU-labeling index waselevated twofold in the hair follicle epidermis at P18-19 inK14Cre+/Miz1lox/lox mice (Fig. 4J; P<0.0001). By contrast, nodifference in proliferation was seen in the interfollicularepidermis (Fig. 4J).

After a short period of resting (3-4 days; telogen), hairfollicles again start to grow out into the subcutis (anagen)(Botchkarev and Paus, 2003). To test further whether the haircycle is delayed, we counted the number of hair follicles in the

subcutis at P22-23 (Fig. 4C,D), which approximately matchesthe time point of the telogen-anagen transition. Even at this latetime point, there was a considerable number of follicles inK14Cre+/Miz1lox/lox mice reaching the subcutis without beingin the anagen of the next hair cycle, whereas in control mice,only anagenic follicles were detectable in the subcutis (Fig. 4I;P22/23). At P26-27 most hair follicles are expected to be inanagen, extending into the subcutis (Fig. 4E,F). Quantificationof the percentage of late anagen hair follicles did not show adifference between control and K14Cre+/Miz1lox/lox mice,indicating that anagen is not impaired (Fig. 4I; P26/27). Inaddition, K14Cre+/Miz1lox/lox mice at 1 year of age showed asignificantly elevated number of follicles extending into thesubcutis, arguing that the delay in catagen was not restricted tothe first hair cycle (Fig. 4G-I). We conclude that the ablationof the POZ/BTB domain in the Miz1 protein results in a delayof catagen during the hair cycle.

Long-term skin alterations after deletion of the Miz1POZ/BTB domainWe observed K14Cre+/Miz1lox/lox mice up to an age of 18months. At this advanced age the fur was still rough, but notin a striped pattern (data not shown). In addition,K14Cre+/Miz1lox/lox mice exhibited a significant loss of hairson several areas of the back skin, compared with age-matchedcontrol mice (Fig. 5K), although the number of hair folliclesdid not change (data not shown). No other lesions such asulcerations, complete alopecia or spontaneous tumors haveoccurred in these mice.

The fur and skin color of 1-year-old K14Cre+/Miz1lox/lox

mice were usually darker than that of age-matched wild-typemice. Histology of the back skin showed numerous thickenedfollicles with hair remnants and abnormal content distributionof melanin (Fig. 5A-H). In areas of substantial decrease of hairdensity, stereo microscopy revealed black dot-like areas (Fig.5D), which are probably the cyst-like structures (Fig. 2H), andhyperpigmentation of the upper third of the hair follicle (Fig.5G). Compared with wild-type animals, K14Cre+/Miz1lox/lox

mice exhibited focal areas in the dermis and subcutis aroundhair follicles with a high content of melanin, which was, at leastpartly, phagocytosed by macrophage-like cells (Fig. 5E-H).

Pigmentation of the different hair types (zigzag, awl, guard)(Dry, 1926) was characterized by larger melanin packages andan irregular arrangement of melanin in K14Cre+/Miz1lox/lox

mice (Fig. 5J). Usually the zigzag hairs are the most abundanthair type, making up about 70% of all hairs (Sundberg andHogan, 1994). In K14Cre+/Miz1lox/lox mice, only 24-54%(mean value 41.2%) of all hairs counted, were zigzag hairsindicating that this hair type is mostly affected (Fig. 5I).Moreover, the zigzag hairs found lacked the characteristicbending (Fig. 5J, arrow). The significant reduction in zigzaghairs, whose tips usually show lighter pigmentation than otherhair types (Fig. 5K), together with pigment accumulation in thedermis, account for the darker appearance of the fur and theskin of K14Cre+/Miz1lox/lox mice.

DiscussionThe morphological phenotype of K14Cre+/Miz1lox/lox micecomprises an irregular order of hair follicles, alterations of hairfollicle and hair fiber morphology, a decrease of hair densityin aged animals and a delay in the catagen of the hair cycle. A

Fig. 3. (A,B) Histological analysis of the epidermis in the hairfollicle and around the hair funnel orifice. (C) Using an antibodyagainst the proliferation marker Ki67 only basal keratinocytes werestained in control mice. (D) In K14Cre+/Miz1lox/lox mice, suprabasalcells were also positive for Ki67, correlating with an increase in thenumber of cell layers in the epidermis. (E-H) In these areas somesuprabasal cells were not only positive for keratin 1 (E,F) but also forkeratin 14 (H; green), which was restricted to basal keratinocytes incontrol mice (G, green). Furthermore, K14Cre+/Miz1lox/lox mice had athicker cornified layer positive for filaggrin (G,H; red). Note that thesuprabasal cells of the epithelium from cyst-like structures (asterisks)also express K1, which is usually found in the interfollicular ratherthan in the intrafollicular epidermis (compare with Fig. 2H). Bar, 100�m (A,B,E-H); 20 �m (C,D).

Jour

nal o

f Cel

l Sci

ence

2590

1.5-fold increase in the number of sebaceous glands wasstatistically not significant and sebaceous cell markers, such asBSSP, were not elevated.

In mice, hairs undergo a synchronized hair cycle during thefirst 25 days after birth. Completely developed hairs (anagen)enter a phase of regression (catagen) at about P15. Catagen iscomplete by about P19, and after a short resting stage (telogen)of 2-3 days, hairs again enter anagen (Botchkarev and Paus,2003). In skin of K14Cre+/Miz1lox/lox mice, the onset andprogression of catagen between P15 and P19 was delayed in ahigh proportion of the hair follicles (Fig. 4). TGF�1 is a mainplayer in the induction and progression of catagen. This wasoriginally surmised by the observation that the expression ofTGF� receptors and TGF� correlated with the onset of catagenand that the local application of TGF� induced hair follicleregression (Welker et al., 1997; Paus et al., 1997). Strongerevidence came from TGF�-null mice, which revealed a delayof catagen (Foitzik et al., 2000). Moreover, an almost completeblock of catagen occurs in mice deficient in Smad4, which is

Journal of Cell Science 120 (15)

an essential component of the TGF�-signaling pathway (Yanget al., 2005).

TGF�1 is thought to activate the promoter of several genesthrough two pathways: activation of a Smad3/4 proteincomplex and downregulation of Myc with a subsequent reliefof Miz1 repression, allowing a synergistic activation of thepromoter by Miz1 and the Smad complex (Seoane et al., 2001;Gebhardt et al., 2006). This dual-input model has beenconfirmed for the p15ink4b gene, which is involved in theinhibition of keratinocyte proliferation by TGF� (Pietenpol etal., 1990; Warner et al., 1999). Consistent with a role for Miz1in TGF�1-mediated gene regulation, we found that theinduction of a number of TGF� target genes was reduced inkeratinocytes isolated from K14Cre+/Miz1lox/lox mice relativeto control keratinocytes (supplementary material Fig. S3). Invivo, K14Cre+/Miz1lox/lox follicular keratinocytes show atwofold increase in the BrdU-labeling index, demonstrating arequirement for Miz1 in controlling cell cycle exit during theanagen-catagen transition of the hair cycle. By contrast,

Fig. 4. Hair cycle alteration in K14Cre+/Miz1lox/lox mice. (A-H) Representative sections of comparable areas of back skin from P19 (A,B), P23(C,D), P26 (E,F) and 1-year-old (G,H) control (A,C,E,G) and K14Cre+/Miz1lox/lox mice (B,D,F,H). Note the sharp boundary between dermisand subcutis (arrow). Such sections were used to count the number of all hair follicle and those which are extending into the subcutis,representing hair follicle either in catagen or anagen, depending on the chosen time point and the hair follicle morphology. (I) Quantitativeanalysis of the percentage of subcutaneous follicles at different ages as indicated. The bars represent the medians. The differences of themedians for P18/19 and for one year were statistically significant (for both P<0.0001). (J) Proliferation of keratinocytes measured by the BrdU-labeling index. Mice received BrdU 1.5 hours before being sacrificed and sections were stained with an anti-BrdU antibody to count thepercentage of S-phase cells, either in the interfollicular skin or in the hair follicle. The twofold difference found in the hair follicle epitheliumwas statistically significant (P<0.0001). The bars represent the medians. n, number of animals analyzed.

Jour

nal o

f Cel

l Sci

ence

2591Miz1 POZ/BTB domain knockout in the epidermis

p15ink4b-knockout mice had no obvious skin phenotype(Latres et al., 2000), indicating that the catagen delay inK14Cre+/Miz1lox/lox mice is not mediated by p15ink4b.

During catagen, epidermal cells of the follicles are removedby apoptosis (Botchkareva et al., 2006) and several genes havebeen suggested to cause involution (Botchkarev and Paus,2003; Randall et al., 2003). The expression of the proapoptoticprotein clusterin (Reddy et al., 1996; Yang et al., 2000; Leskovet al., 2003), which is a target of TGF �1 and Miz1 generegulation in keratinocytes (Gebhardt et al., 2006), was reducedin response to TGF� in K14Cre+/Miz1lox/lox mice(supplementary material Fig. S3B,D). An impaired inductionof clusterin during hair follicle regression could complementthe delay of cell cycle exit and differentiation by asimultaneous decrease in apoptosis. Taken together the data areconsistent with the view that the deletion of the Miz1 POZdomain interferes with the TGF� pathway in this stage of thehair cycle.

In addition to the alterations in the hair cycle, we observedK14- and Ki67-positive cells in suprabasal layers of the hairfunnel, indicating a delayed differentiation in K14Cre+/Miz1lox/lox keratinocytes. This phenomenon has not been

reported in TGF�1-deficient mice. Previous work has shownthat the switch between proliferation and differentiation inepidermal keratinocytes is accompanied by a loss of Mycexpression and an increase in expression of the Max-networkproteins Mxd1 and Mxd3, suggesting that Myc-activated targetgenes are repressed during differentiation (Hurlin et al., 1995).Our findings suggest that activation of Miz1-dependenttransactivation causally contributes to the transition fromproliferation to differentiation in K14Cre+/Miz1lox/lox

keratinocytes.In addition to the described morphological changes of the

hair follicle, the hair fiber structure was altered. The melanininclusions were coarse and irregularly arranged in the hairfibers of K14Cre+/Miz1lox/lox mice compared with hair fibersfrom control mice, indicating impairment of the hairpigmentation process in the hair bulb. This is confirmed by theobservation that melanin accumulates around hair follicles inthe dermis with aging (Fig. 5F), which is probably due to apoorly defined process known as pigmentary incontinencedescribed in inflammatory skin diseases (Schaumberg-Leverand Lever, 1973; Nagao and Iljima, 1974; Masu and Seiji,1983) and in senile lentigo (Ünver et al., 2006). In addition,

Fig. 5. Histological analysis of skin from 1-year-old control (A,E) and K14Cre+/Miz1lox/lox mice (B,C,F-H). Hematoxylin and eosin staining (A-C) revealed aberrant hair follicle morphology with thickened epithelium and distorted hair remnants in the follicle (B,C). In addition,pigmentary incontinence was observed in K14Cre+/Miz1lox/lox (F,H) but not in control (E) mice with Fontana-Masson staining. (G) Melanindeposits in the upper third of the hair follicle epithelium. (D) Hair remnants and ectopic melanin deposition appear as black spots instereomicroscopic analysis of back skin (arrows). (I-K) K14Cre+/Miz1lox/lox mice showed a decrease in hair density (K) while specificallyzigzag hairs were reduced (I; P<0.0001; bars represent mean values). In the upper panel in K, the light area is due to the unpigmented tips ofthe zigzag hairs which are missing in K14Cre+/Miz1lox/lox mice (lower panel). The pigmentation of the hairs was irregular in K14Cre+/Miz1lox/lox

mice with coarse melanin inclusions, and the bends of zigzag hairs were diminished (J; arrow). Bars, 100 �m (A-C,E-H); 200 �m (D); 50 �m(J); 5 mm (K).

Jour

nal o

f Cel

l Sci

ence

2592

aberrant pigmentation of the epidermis in the upper third ofhair follicles was occasionally observed in K14Cre+/Miz1lox/lox

mice. These ectopic melanin locations suggest that the transferof melanosomes from melanocytes to keratinocytes isimpaired.

The fur of mice comprises at least three different hair types:zigzag hairs (about 70%), guard hairs (about 28%) and awlhairs (about 2%) (Sundberg and Hogan, 1994). The relativenumber of zigzag hairs was reduced to about 40-50% and thecharacteristic bending of this hair type, usually leading to threekinks per hair, were absent in most hairs of this type. Recently,a similar hair phenotype has been described in lymphotoxin-�(LT-�) but not in LT-�-knockout mice (Cui et al., 2006). Itremains to be seen whether Miz1 is also linked to the LT-�signaling pathway regulating gene expression independently ofTGF�.

Materials and MethodsTransgenic miceMiz1lox/lox mice were generated according to standard procedures (Nagy et al., 2003)(C. Kosan, N. Fang, M. Godmann et al., unpublished). In brief, exon 3 and exon 4of Miz1, encoding the POZ/BTB domain, were flanked by loxP sites. A furthermouse strain was generated by transient expression of Cre recombinase in Miz1+/lox

ES cells leading to a Miz1+/�POZ genotype. Miz1lox/�POZ mice were generated bycrossing Miz1+/�POZ with Miz1lox/lox mice. Miz1lox/lox or Miz1lox/�POZ mice werecrossed with K14Creneo mice (Huelsken et al., 2001), here designated K14Cremice. These mice express Cre recombinase under the control of the keratin 14promoter, inducing Miz1 ablation in basal keratinocytes of the epidermal layer ofthe skin (Fuchs, 1995; Wang et al., 1997). Mice were genotyped by standard PCR,using the primers 1, 2 and 5 listed in supplementary material Table S1, with anannealing temperature of 60°C and an extension time of 45 seconds.

Cell culture1-day post partum mice were sacrificed by decapitation, keratinocytes were isolatedand cells were cultivated in Eagle’s minimum essential medium with Earle’s BSSand without CaCl2 (Cambrex) as described (Gebhardt et al., 2006).

RNA isolation and preamplificationTotal cellular RNA was isolated with the RNeasy reagent (Qiagen) and a subsequentDNase digestion was included. RNA was amplified with MessageAmp RNAamplification kit (Ambion). All procedures were performed according to themanufacturer’s instructions. RNA quality was controlled by agarose gelelectrophoresis.

Reverse-transcriptase PCR (RT-PCR)First-strand cDNAs of amplified RNA were synthesized with M-MLV ReverseTranscriptase (Invitrogen) and random hexamer primers (Promega). For each PCRamplification, aliquots were taken after different cycles to determine the linear rangeof the amplification. Primer sequences are available in supplementary material TableS1.

Real-time quantitative PCR (RQ-PCR)cDNA was synthesized as described above. Real-time PCRs were performed withthe light cycler ABIprism 7000, using the qPCRTM Core Kit for Sybr Green I(Eurogentec) according to the manufacturer’s protocol. rps16 was used as aninternal control for normalization. Primer sequences are shown in supplementarymaterial Table S1.

Immunoprecipitation and western blot analysisLysates from keratinocytes (300 �g protein) were immunoprecipitated using themonoclonal anti-Miz1 antibody H-190 (Santa Cruz). Precipitated proteins wereseparated by 10% reducing SDS polyacrylamide gel electrophoresis and transferredto Immobilon-P transfer membrane (Millipore) by standard procedures. The westernblot was incubated with the monoclonal anti-Miz1 antibody 10E2 (1:400).Peroxidase-conjugated anti-mouse immunoglobulin was used as secondary antibody(1:3000; Amersham Biosciences). Proteins were visualized by using ECL plusWestern Blotting Detection System (Amersham Biosciences).

ImmunohistochemistryTissue samples were fixed in 4% PBS-buffered formalin and 5 �m paraffin sectionswere stained with hematoxylin and eosin according to standard procedures. Inaddition, the Fontana-Masson silver method for melanin staining was applied

Journal of Cell Science 120 (15)

(Ünver et al., 2006). For immunofluorescence, sections were preincubated in 3%bovine serum albumin (Roth), 0.1% Tween 20 in PBS for 30 minutes and the firstantibody was applied overnight at 4°C. After incubation with fluorescence-labeledsecondary antibodies for 45 minutes at room temperature (Alexa Fluor 546 goat-anti-rabbit IgG, Alexa Fluor 488 goat-anti-mouse IgG or Alexa Fluor 488 goat-anti-rabbit IgG; 1:1000; Molecular Probes), nuclei were counterstained with DAPI (1mg/ml; 1:2000) and slides were mounted in Mowiol (Hoechst). For BrdU staining,mice received 200 �l of a BrdU solution (10 mg/ml in PBS) by intraperitonealinjection 1.5 hours before being sacrificed. Slides were pretreated in 2 M HCl, 0.1%Triton X-100 for 30 minutes at room temperature. The following antibodies wereused at the dilutions indicated: anti-Keratin 1 (1:1000; Covance), anti-Keratin 14(1:1000; Covance), anti-Filaggrin (1:500; Covance) and anti-BrdU (1:50;DakoCytomation).

Pictures were taken either with the BX61 Olympus microscope equipped with aF-View digital camera (Soft Imaging System, Münster, Germany) or with theAxiovert100 microscope (Zeiss) equipped with a MC 100 SPOT camera.

Hair analysisHairs were pulled out of the skin or were collected by shaving from comparablesites of the back. Hairs were roughly oriented and stuck to a strip of tape that wasthen stuck to a slide. More than 300 hairs per animal were counted under themicroscope, distinguishing between zigzag hairs and hairs exhibiting more than onerow of melanin inclusions.

Morphometric analysisThe area of sebaceous glands were measured in three control and threeK14Cre+/Miz1lox/lox mice using a 20� lens and the program analysis® (Soft ImagingSystem, Münster, Germany). 56-60 hair follicles per animal were analyzed. Thesebaceous gland area from all glands measured was divided by the number offollicles analyzed and expressed in �m2/follicle.

Statistical analysisTo test the significance of the quantitative data, the unpaired, two-tailed Student’st-test was applied using the Graph Pad Prism program.

We thank Waltraud Ackermann, Sigrid Bischofsberger andAngelika Filmer for expert technical assistance and Birgit Samans forprofiling the data from the microarray experiment. K14Cre mice werea generous gift from W. Birchmeier, Max Delbrück Center, Berlin.This study was supported by the Deutsche Forschungsgemeinschaft(DFG) through the grant Ei 222-1 to Martin Eilers, Hans-PeterElsässer and Werner Lutz, and through the grant MO 435/15-1 andMO 435/15-2 to Tarik Möröy and by the InternationalGraduiertenkolleg 767 ‘Transcriptional Control in DevelopmentalProcesses’.

ReferencesAdhikary, S., Peukert, K., Karsunky, H., Beuger, V., Lutz, W., Elsasser, H. P., Moroy,

T. and Eilers, M. (2003). Miz1 is required for early embryonic development duringgastrulation. Mol. Cell. Biol. 23, 7648-7657.

Adhikary, S., Marinoni, F., Hock, A., Hulleman, E., Popov, N., Beier, R., Bernard,S., Quarto, M., Capra, M., Goettig, S. et al. (2005). The ubiquitin ligase HectH9regulates transcriptional activation by Myc and is essential for tumor cell proliferation.Cell 123, 409-421.

Alonso, L. and Fuchs, E. (2003). Stem cells of the skin epithelium. Proc. Natl. Acad.Sci. USA 100 Suppl. 1, 11830-11835.

Arnold, I. and Watt, F. M. (2001). c-Myc activation in transgenic mouse epidermisresults in mobilization of stem cells and differentiation of their progeny. Curr. Biol. 11,558-568.

Bardwell, V. J. and Treisman, R. (1994). The POZ domain: a conserved protein-proteininteraction motif. Genes Dev. 8, 1664-1677.

Botchkarev, V. A. and Paus, R. (2003). Molecular biology of hair morphogenesis:development and cycling. J. Exp. Zool. B Mol. Dev. Evol. 298, 164-180.

Botchkareva, N. V., Ahluwalia, G. and Shander, D. (2006). Apoptosis in the hairfollicle. J. Invest. Dermatol. 126, 258-264.

Cui, C. Y., Hashimoto, T., Grivennikov, S. I., Piao, Y., Nedospasov, S. A. andSchlessinger, D. (2006). Ectodysplasin regulates the lymphotoxin-beta pathway forhair differentiation. Proc. Natl. Acad. Sci. USA 103, 9142-9147.

Dry, F. W. (1926). The coat of the mouse (Mus Musculus). J. Genet. 16, 287-340.Foitzik, K., Lindner, G., Mueller-Roever, S., Maurer, M., Botchkareva, N.,

Botchkarev, V., Handjiski, B., Metz, M., Hibino, T., Soma, T. et al. (2000). Controlof murine hair follicle regression (catagen) by TGF-beta1 in vivo. FASEB J. 14, 752-760.

Frye, M., Gardner, C., Li, E. R., Arnold, I. and Watt, F. M. (2003). Evidence that Mycactivation depletes the epidermal stem cell compartment by modulating adhesiveinteractions with the local microenvironment. Development 130, 2793-2808.

Fuchs, E. (1995). Keratins and the skin. Annu. Rev. Cell Dev. Biol. 11, 123-153.Gebhardt, A., Frye, M., Herold, S., Benitah, S. A., Braun, K., Samans, B., Watt, F.

Jour

nal o

f Cel

l Sci

ence

2593Miz1 POZ/BTB domain knockout in the epidermis

M., Elsasser, H. P. and Eilers, M. (2006). Myc regulates keratinocyte adhesion anddifferentiation via complex formation with Miz1. J. Cell Biol. 172, 139-149.

Herold, S., Wanzel, M., Beuger, V., Frohme, C., Beul, D., Hillukkala, T., Syvaoja, J.,Saluz, H. P., Haenel, F. and Eilers, M. (2002). Negative regulation of the mammalianUV response by Myc through association with Miz-1. Mol. Cell 10, 509-521.

Huelsken, J., Vogel, R., Erdmann, B., Cotsarelis, G. and Birchmeier, W. (2001). beta-Catenin controls hair follicle morphogenesis and stem cell differentiation in the skin.Cell 105, 533-545.

Hurlin, P. J., Queva, C., Koskinen, P. J., Steingrimsson, E., Ayer, D. E., Copeland, N.G., Jenkins, N. A. and Eisenman, R. N. (1995). Mad3 and Mad4: novel Max-interacting transcriptional repressors that suppress c-myc dependent transformation andare expressed during neural and epidermal differentiation. EMBO J. 14, 5646-5659.

Latres, E., Malumbres, M., Sotillo, R., Martin, J., Ortega, S., Martin-Caballero, J.,Flores, J. M., Cordon-Cardo, C. and Barbacid, M. (2000). Limited overlapping rolesof P15(INK4b) and P18(INK4c) cell cycle inhibitors in proliferation andtumorigenesis. EMBO J. 19, 3496-3506.

Leskov, K. S., Klokov, D. Y., Li, J., Kinsella, T. J. and Boothman, D. A. (2003).Synthesis and functional analyses of nuclear clusterin, a cell death protein. J. Biol.Chem. 278, 11590-11600.

Masu, S. and Seiji, M. (1983). Pigmentary incontinence in fixed drug eruptions.Histologic and electron microscopic findings. J. Am. Acad. Dermatol. 8, 528-532.

Meier, N., Dear, T. N. and Boehm, T. (1999). A novel serine protease overexpressed inthe hair follicles of nude mice. Biochem. Biophys. Res. Commun. 258, 374-378.

Moore, K. A. and Lemischka, I. R. (2006). Stem cells and their niches. Science 311,1880-1885.

Nagao, S. and Iljima, S. (1974). Light and electron microscopic study of Riehl’smelanosis. Possible mode of its pigmentary incontinence. J. Cutan. Pathol. 1, 165-175.

Nagy, A., Gertsenstein, M., Vintersten, K. and Behringer, R. (2003). Manipulating themouse embryo, 3rd edn. New York: Cold Spring Harbor Laboratory Press.

Paus, R., Foitzik, K., Welker, P., Bulfone-Paus, S. and Eichmuller, S. (1997).Transforming growth factor-beta receptor type I and type II expression during murinehair follicle development and cycling. J. Invest. Dermatol. 109, 518-526.

Peukert, K., Staller, P., Schneider, A., Carmichael, G., Hanel, F. and Eilers, M. (1997).An alternative pathway for gene regulation by Myc. EMBO J. 16, 5672-5686.

Phan, R. T., Saito, M., Basso, K., Niu, H. and Dalla-Favera, R. (2005). BCL6interacts with the transcription factor Miz-1 to suppress the cyclin-dependent kinaseinhibitor p21 and cell cycle arrest in germinal center B cells. Nat. Immunol. 6, 1054-1060.

Pietenpol, J. A., Holt, J. T., Stein, R. W. and Moses, H. L. (1990). Transforming growthfactor beta 1 suppression of c-myc gene transcription: role in inhibition of keratinocyteproliferation. Proc. Natl. Acad. Sci. USA 87, 3758-3762.

Randall, V. A., Sundberg, J. P. and Philpott, M. P. (2003). Animal and in vitro modelsfor the study of hair follicles. J. Invest. Dermatol. Symp. Proc. 8, 39-45.

Reddy, K. B., Jin, G., Karode, M. C., Harmony, J. A. and Howe, P. H. (1996).Transforming growth factor beta (TGF beta)-induced nuclear localization ofapolipoprotein J/clusterin in epithelial cells. Biochemistry 35, 6157-6163.

Schaumberg-Lever, G. and Lever, W. F. (1973). Electron microscopy of incontinentiapigmenti. J. Invest. Dermatol. 61, 151-158.

Seoane, J., Pouponnot, C., Staller, P., Schader, M., Eilers, M. and Massague, J. (2001).TGFbeta influences Myc, Miz-1 and Smad to control the CDK inhibitor p15INK4b.Nat. Cell Biol. 3, 400-408.

Staller, P., Peukert, K., Kiermaier, A., Seoane, J., Lukas, J., Karsunky, H., Moroy,T., Bartek, J., Massague, J., Hanel, F. et al. (2001). Repression of p15INK4bexpression by Myc through association with Miz-1. Nat. Cell Biol. 3, 392-399.

Sundberg, J. P. and Hogan, M. E. (1994). Hair types and subtypes in the laboratorymouse. In Handbook of Mouse Mutations with Skin and Hair Abnormalities (ed. J. P.Sundberg), pp. 57-68. Boca Raton: CRC Press.

Ünver, N., Freyschmidt-Paul, P., Horster, S., Wenck, H., Stab, F., Blatt, T. andElsasser, H. P. (2006). Alterations in the epidermal-dermal melanin axis and factorXIIIa melanophages in senile lentigo and ageing skin. Br. J. Dermatol. 155, 119-128.

van de Wetering, M., Sancho, E., Verweij, C., de Lau, W., Oving, I., Hurlstone, A.,van der Horn, K., Batlle, E., Coudreuse, D., Haramis, A. P. et al. (2002). The beta-catenin/TCF-4 complex imposes a crypt progenitor phenotype on colorectal cancercells. Cell 111, 241-250.

Wang, X., Zinkel, S., Polonsky, K. and Fuchs, E. (1997). Transgenic studies with akeratin promoter-driven growth hormone transgene: prospects for gene therapy. Proc.Natl. Acad. Sci. USA 94, 219-226.

Warner, B. J., Blain, S. W., Seoane, J. and Massague, J. (1999). Myc downregulationby transforming growth factor beta required for activation of the p15(Ink4b) G(1) arrestpathway. Mol. Cell. Biol. 19, 5913-5922.

Watt, F. M. (2002). Role of integrins in regulating epidermal adhesion, growth anddifferentiation. EMBO J. 21, 3919-3926.

Welker, P., Foitzik, K., Bulfone-Paus, S., Henz, B. M. and Paus, R. (1997). Hair cycle-dependent changes in the gene expression and protein content of transforming factorbeta 1 and beta 3 in murine skin. Arch. Dermatol. Res. 289, 554-557.

Wu, S., Cetinkaya, C., Munoz-Alonso, M. J., von der Lehr, N., Bahram, F., Beuger,V., Eilers, M., Leon, J. and Larsson, L. G. (2003). Myc represses differentiation-induced p21CIP1 expression via Miz-1-dependent interaction with the p21 corepromoter. Oncogene 22, 351-360.

Yang, C. R., Leskov, K., Hosley-Eberlein, K., Criswell, T., Pink, J. J., Kinsella, T. J.and Boothman, D. A. (2000). Nuclear clusterin/XIP8, an x-ray-induced Ku70-bindingprotein that signals cell death. Proc. Natl. Acad. Sci. USA 97, 5907-5912.

Yang, L., Mao, C., Teng, Y., Li, W., Zhang, J., Cheng, X., Li, X., Han, X., Xia, Z.,Deng, H. et al. (2005). Targeted disruption of Smad4 in mouse epidermis results infailure of hair follicle cycling and formation of skin tumors. Cancer Res. 65, 8671-8678.

Jour

nal o

f Cel

l Sci

ence