ORIGINAL ARTICLE

Loss of K2 Expression Causes Aberrant Aggregation of K10, Hyperkeratosis, and Inflammation Heinz Fischer1, Lutz Langbein2, Julia Reichelt3, Silke Praetzel-Wunder2, Maria Buchberger1, Minoo Ghannadan1, Erwin Tschachler1 and Leopold Eckhart1

Keratin K2 is one of the most abundant structural of the epidermis; however, its biological significance has remained elusive. Here we show that suprabasal type II , K1 and K2, are expressed in a mutually exclusive manner at different body sites of the mouse, with K2 being confined to the ear, sole, and tail skin. Deletion of K2 caused acanthosis and hyperkeratosis of the ear and the tail epidermis, corneocyte fragility, increased transepidermal water loss, and local inflammation in the ear skin. The loss of K2 was partially compensated by upregulation of K1 expression. However, a significant portion of K2-deficient suprabasal keratinocytes lacked a regular and developed massive aggregates of the , K10. Aggregate formation, but not hyperkeratosis, was suppressed by the deletion of both K2 and K10, whereas deletion of K10 alone caused clumping of K2 in ear skin. Taken together, this study demonstrates that K2 is a necessary and sufficient binding partner of K10 at distinct body sites of the mouse and that unbalanced expression of these keratins results in aggregate formation. Journal of Investigative Dermatology (2014) 134, 2579–2588; doi:10.1038/jid.2014.197; published online 29 May 2014

INTRODUCTION keratins lead to epidermal fragility (Omary et al., 2004). The Keratins are the main cytoskeletal proteins of epithelial cells deletion of K10 causes epidermal thickening and an including epidermal keratinocytes (Morley and Lane, 1994; enlargement of sebaceous glands (Reichelt and Magin, 2002; Candi et al., 2005; Moll et al., 2008). Distinct pairs of type I Reichelt et al., 2004). This mild phenotype was attributed to and type II keratins heterodimerize to form keratin interme- compensatory upregulation of the basal epidermal keratin K14 diate filaments that increase the mechanical resilience of (Reichelt et al., 2001). By contrast, the deletion of K1 results in epithelial cells besides having additional nonstructural roles a defect in skin barrier formation, inflammation, and perinatal (Morley and Lane, 1994; Magin et al., 2007; Pan et al., 2013; lethality (Roth et al., 2012). Ramms et al., 2013; Seltmann et al., 2013). Different muta- The expression of the type II keratin K2 (formerly K2e; Moll tions in keratin genes cause distinct epidermolytic bullous et al., 2008) is restricted to the uppermost suprabasal layers of diseases and certain forms of ichthyoses (Uitto et al., 2007; the epidermis (Rentrop et al., 1987; Collin et al.,1992;Herzog Arin, 2009; McLean and Moore, 2011). et al., 1994) where it colocalizes with the type I keratin K10 Epidermal keratinocytes express keratins in a strictly differ- (Smith et al., 1999). In humans, K2 is expressed throughout the entiation stage–dependent manner. K5 and K14 are expressed interfollicular epidermis but strongly increased in palmo- in the basal layer, whereas K1, K2, and K10 are confined to plantar skin and in the epidermis on other mechanically the suprabasal layers (Fuchs, 1990; Collin et al., 1992; Morley stressed body sites (Rothnagel et al., 1994; Swensson et al. and Lane, 1994; for a review, see Moll et al.,2008).The 1998). This expression pattern has led to the suggestion that K2 primary role of suprabasal keratins is to enhance the provides additional mechanical resilience to the epidermis. mechanical strength of the epidermis, and mutations in these KRT2 mutations that lead to the substitution of amino acid residues critical for filament formation cause ichthyosis bullosa 1Department of Dermatology, Medical University of Vienna, Vienna, Austria; of Siemens (Kremer et al., 1994; McLean et al.,1994; 2Department of Genetics of Skin Carcinogenesis, German Cancer Research Rothnagel et al., 1994). In this disease cell lysis occurs in 3 Center, Heidelberg, Germany and Institute of Cellular Medicine and North the spinous and granular epidermal layers upon mild stress, East England Stem Cell Institute, Newcastle University, Newcastle upon Tyne, UK which manifests particularly on soles, palms, knees, and Correspondence: Leopold Eckhart, Department of Dermatology, Medical elbows. Clinically, these lesions are characterized by acan- University of Vienna, Waehringer Guertel 18-20, 1090 Vienna, Austria. thosis, vacuolization of the granular layer, and hyperkeratosis. E-mail: [email protected] Skin inflammation has been observed in this condition Abbreviations: DTT, dithiothreitol; IL, interleukin; PAGE, polyacrylamide gel (Basarab et al., 1999). As ichthyosis bullosa of Siemens is eletrophoresis; SDS, sodium dodecyl sulfate; TEWL, transepidermal water loss; caused by mutated K2 proteins that act in a dominant-negative TNF-a, tumor necrosis factor alpha; TSLP, thymic stromal lymphopoietin manner, the symptoms of ichthyosis bullosa of Siemens Received 27 November 2013; revised 24 March 2014; accepted 2 April 2014; accepted article preview online 21 April 2014; published online 29 are not informative about the physiological function of May 2014 normal K2.

& 2014 The Society for Investigative Dermatology www.jidonline.org 2579 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

K2 K2

K1 K1

* * K2 + K1 K2 + K1

##* Nail

K2 + K1 **

Nail # #

K2 + K1

Figure 1. Keratins K1 and K2 are differentially expressed in the epidermis of the mouse. (a–d) Immunofluorescence labeling of K1 (red) and K2 (green) in the skin of the ears (a), back (b), tail (c), and paw (d) of C57BL/6 wild-type mice. Nuclei were counterstained (blue) with Hoechst 33258. Asterisks mark infundibula of hair follicles. **The image of the paw is a compilation of pictures. (e, f) X-gal staining (blue) of beta-galactosidase expressed under the control of the Krt2 promoter in the tail (e) and toe (f)ofKrt2 þ / mice. Arrows and arrowheads mark corresponding positions of the toe pad in d and f, whereas # symbols mark scales in c and e. Bars ¼ 40 mm(a, b), 100 mm(c), 200 mm(d).

Here we investigate the role of K2 in epidermal homeostasis (Figure 1a–c). By contrast and surprisingly different from the by defining the expression patterns of suprabasal keratins in situation in human skin, K1 was absent from the interfolli- the mouse and by generating and characterizing K2-deficient cular epidermis of the mouse ear, whereas it was present in mice. We show that suprabasal epidermal keratinocytes of the infundibular regions (Figure 1a). In the back skin, which several murine body sites lack expression of K1 while expre- is densely covered by pelage hair follicles, K2 was absent, ssing K2 together with K10. The corresponding sites of K2 whereas K1 was expressed throughout the interfollicular knockout mice contain prominent K10 aggregates and epidermis (Figure 1b). In the tail skin, K1 and K2 were develop epidermal acanthosis and hyperkeratosis as well as confined to the interscale regions within which the infundi- skin inflammation. Together with our finding that K2 forms bulum regions of the sparse hair follicles contained K1 and aggregates in the absence of K10, our study establishes that the region more distant from the hair follicles contained K2 K2/K10 intermediate filaments are essential for normal epi- (Figure 1c). On the soles and footpads, K1 and K2 were also dermal morphology at specific body sites. expressed in an alternating pattern (Figure 1d, arrows). All skin sites that contained either K1 or K2 expressed K10 RESULTS (Supplementary Figure S1 online). Essentially the same Keratins K1 and K2 are differentially expressed in the suprabasal results were obtained by immunofluorescence analysis of ear and plantar epidermis the mouse strains C57BL/6, mixed 129/C57BL/6, mixed The expression patterns of K2 and K1 were determined by C57BL/6/Balb/c, and SKH-1. Together, these expression immunofluorescence labeling of the skin from various body patterns suggested that the suprabasal epidermis of the sites of wild-type (WT) mice (Figure 1). K2 was expressed in mouse has two mutually exclusive schemes of keratiniza- the suprabasal layers of the interfollicular epidermis of the tion, one involving the formation of intermediate filaments ear with the exception of the infundibulum of hair follicles made of K1/K10 heterodimers and the other involving the

2580 Journal of Investigative Dermatology (2014), Volume 134 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

formation of intermediate filaments made of K2/K10 a Krt2 +/+ Krt2 –/– heterodimers. In an independent experimental approach, the activity of the murine Krt2 promoter was assessed in mice carrying a beta-galactosidase reporter gene inserted into one of the two Krt2 alleles (Supplementary Figure S2 online). Visualization of beta-galactosidase activity in situ revealed reporter gene expression in the ear skin (Supplementary Figure S3 online), in the interscale regions of tail skin (Figure 1e), and in those regions of the sole and toe pads that lacked eccrine sweat glands (Figure 1f, for overview and negative control see Supplementary Figure S3 online). No expression of the reporter gene was detected in the back skin of heterozygous mice (not shown) and in any tissue of WT mice that were used as negative controls (Supplementary Figure S3 online). Thus, +/+ –/– b Krt2 Krt2 the beta-galactosidase activity staining confirmed the expres- sion sites of K2 as detected by immunofluorescence analysis of tissue sections and revealed a highly patterned organization of Krt2 promoter activity on the tail as well as on the soles and toe pads. Further investigations were focused on the epidermis of the ears because of the uniform expression of K2 in the interfollicular epidermis at this body site.

c d 3 Targeted deletion of K2 leads to hyperkeratosis, increased water 100 * * loss, and localized inflammation 80 ) To investigate the role of K2 in vivo, we generated mice in –1 h / 2 which both alleles of the Krt2 gene were disrupted (Krt2 ) 60 –2 (Supplementary Figure S2 online). Krt2 / , Krt2 þ / ,andWT þ / þ 40 (Krt2 ) mice were born at Mendelian ratios without 1 macroscopic differences. However, within the first 6 weeks 20 TEWL (g m / of life, all Krt2 mice developed scaly skin on the ears and Intact corneocytes (%) 0 0 hyperkeratotic calluses on the soles and toe pads +/+ –/– +/+ –/– (Supplementary Figure S4a and b online), whereas none of their Krt2 þ / and Krt2 þ / þ littermates showed this phenotype Figure 2. Deletion of K2 leads to hyperkeratosis, increased water loss, and decreased corneocyte stability. (a) Hematoxylin and eosin staining of ear skin. (Supplementary Figures S3 and S4 online). Furthermore, the Magnified details are shown in the lower panels. Arrows point to enlarged / ears of Krt2 mice were more pigmented than those of keratohyalin granules and an arrowhead indicates cytolysis. (b, c) Corneocytes control mice (Supplementary Figure S4a online). Microscopic were prepared from the ear skin of wild-type ( þ / þ ) and K2-deficient ( / ) investigation of hematoxylin and eosin–stained skin sections mice (n ¼ 4 per group) and scored under the microscope for being intact or revealed prominent acanthosis, orthokeratotic hyperkeratosis damaged (arrowheads). (d) Transepidermal water loss (TEWL) was measured þ / þ / in the epidermis of the ear (Figure 2a), and to a lesser extent, on the ears of wild-type ( ) and K2-deficient ( )mice(n ¼ 5 per group). Error bars show standard deviations. *P 0.005, Student’s t-test. Bar ¼ 40 mm. also in the epidermis of the tail and the palm skin of Krt2 / o mice (Supplementary Figure S4c and d online). Cell prolifera- tion was dramatically enhanced in Krt2 / ear epidermis as misshapen or fragmented corneocytes were obtained from virtually every basal layer cell and also some parabasal K2-deficient mice (Figure 2b and c). Transepidermal water loss keratinocytes were immunopositive for Ki67 (Supplementary was twice as high on the ears of Krt2 / mice as on the Figure S4e online). The number of nucleated suprabasal layers ears of Krt2 þ / þ mice (Figure 2d). Together, these results was increased from 1–2 in Krt2 þ / þ (Figure 2a) and Krt2 þ / suggested that K2 is required for the normal stress resilience mice (Supplementary Figure S5 online) to 7–8 in Krt2 / of corneocytes and for barrier function of the epidermis on mice (Figure 2a) in which the granular layer was markedly mouse ears. thickened. Krt2 / keratinocytes differentiated in a partly Histological investigations suggested the presence of an disorganized manner and accumulated large coalescent gran- inflammatory infiltrate in the ears of Krt2 / mice (Figure 2a), ules (Figure 2a, arrows). Occasionally, focal suprabasal which was neither observed in the skin of the back (not cytolysis was observed (Figure 2a, arrowhead). shown), tail, or soles (Supplementary Figure S4 online) of these To determine the functional consequences of K2 deficiency, mice nor in the ears of control mice (Figure 2a). The ears of we isolated corneocytes and assessed their morphology. Krt2 / mice showed massively elevated expression levels of Ear skin of WT mice yielded approximately 80% normally thymic stromal lymphopoietin and IL-18 mRNAs but normal shaped and 20% defective corneocytes, whereas only abundance of TNF-a mRNA (Figure 3a). Sections of K2- 60% apparently normal and approximately 40% severely deficient ears contained significantly elevated numbers of T

www.jidonline.org 2581 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

cells (Figure 3b and d) and mast cells (Figure 3c and e). Taken a TSLP IL-18 TNFα together, these results suggested that the absence of K2 led to 50 * 10 * 2 NS the weakening of the epidermal barrier and in consequence to 40 8 skin inflammation. 30 6 1 20 4 Loss of type II keratin K2 leads to alterations in the expression levels of other cytoskeletal proteins and to aggregation of type I 10 2

keratin K10 in the ear epidermis (a.u.) mRNA expression 0 0 0 +/+ –/– +/+ –/– +/+ –/– To investigate the molecular basis for the disturbed epidermal homeostasis in the ears of Krt2 / mice, we determined the b Krt2 +/+ Krt2 –/– expression of keratins and other structural components of keratinocytes. Western blot analysis showed a strong upregu- lation of the suprabasal keratins K1 and K10, and of the inflammation-associated keratin K16, whereas the expre- ssion level of the basal layer type II keratin K5 was not altered in the ears of Krt2 / mice compared with that of Krt2 þ / þ mice (Figure 4a). and loricrin levels were markedly / increased in Krt2 ears (Figure 4a). The pattern of filaggrin c Krt2 +/+ Krt2 –/– bands suggested a normal proteolytic maturation of profilag- grin in Krt2 / ear skin. Western blot analysis with an anti- body against the N-terminus of K2 confirmed the absence of both full-length and truncated K2 proteins in Krt2 / ear skin (Figure 4a). mRNA quantification by reverse transcription– quantitative PCRs suggested that differences in amounts of keratins, filaggrin, and loricrin were caused by changes in the corresponding mRNAs (Supplementary Figure d 400 * 80 * S6 online). 300 Immunofluorescence analysis confirmed the changes in protein abundance and showed altered distribution patterns 200 40

of all antigens tested. K5 was redistributed from a strictly basal 100 þ / þ

layer–specific localization in the Krt2 epidermis to the Mast cells per mm basal and first suprabasal layers in the Krt2 / epidermis 0 0 CD3-positive cells per mm CD3-positive +/+ –/– +/+ –/– (Figure 4b). Expression of K1 was restricted to the infundibu- lum of hair follicles in Krt2 þ / þ ears but extended to the Figure 3. Deletion of K2 leads to inflammation in ear skin. (a) mRNA obtained þ / þ / interfollicular epidermis in Krt2 / ears (Figure 4b and c). from the ears of wild-type ( ) and K2-deficient ( ) mice was analyzed by quantitative PCR for the expression of proinflammatory cytokines. Expression K10 was abundant in suprabasal keratinocytes of both geno- levels (arbitrary units, a.u.) were calculated relative to that of the housekeeping types, however, with an aberrant intracellular distribution in a gene, beta-2-microglobulin. Representative results from at least three / significant portion of cells in the Krt2 epidermis (Figure 4b independent experiments are shown. (b, c) Ear sections were immunostained and c). In the latter cells K10 was not evenly distributed in the for CD3 (b) and stained with toluidine blue to visualize mast cells (arrows) (c). cytoplasm but aggregated into clumps. Double labeling of K1 The border between the epidermis and the dermis is indicated by a dashed and K10 showed that K10-positive clumps were present in line. The densities of T cells (CD3-positive) (d) and mast cells (e)were þ / þ / cells that contained little or no K1 (Figure 4c). Furthermore, determined in the ears of wild-type ( ) and K2-deficient ( ) mice and compared by Student’s t-test (three mice per genotype). Error bars show the the deletion of Krt2 induced the interfollicular expression of standard deviation. *Po0.05; NS, not significant (Student’s t-test). Bar ¼ 40 mm. K16 (Figure 4b) and K6 (not shown). Filaggrin and loricrin, TNF-a, tumor necrosis factor alpha; TSLP, thymic stromal lymphopoietin. which are both involved in the cornification of keratinocytes, were highly expressed in the thickened granular layer where filaggrin was located in enlarged keratohyalin granules Krt10 / double knockout mice. Double knockout mice (Figure 4b, see also Figure 5b and Supplementary Figure S8b were viable and developed scaly skin on their ears, similar online). These data suggest that the lack of K2 leads to to Krt2 / and Krt10 / mice (not shown). The ultrastruc- abnormal differentiation of keratinocytes and that K1 does tural histology of WT (Krt2 þ / þ Krt10 þ / þ ), double knockout not fully substitute K2 as binding partner for K10 in ear skin, (Krt2 / Krt10 / ), and single knockout mice (Krt2 / resulting in the formation of K10 aggregates. Krt10 þ / þ and Krt2 þ / þ Krt10 / )wasinvestigatedby electron microscopy. Compared with WT mice (Figure 5a; Co-deletion of Krt2 and Krt10 abrogates aggregate formation but Supplementary Figure S7a and b online), all the keratin not hyperkeratosis knockout mice (Figure 5b–d; Supplementary Figures S8 and To further define the nature and impact of the keratin S9 online) showed an increase in the number of epidermal cell aggregates in K2-deficient mice, we crossed these mice with layers and a decrease in the abundance of cytoplasmic keratin K10-deficient mice (Reichelt et al., 2001) to generate Krt2 / filament bundles in suprabasal cells, which were regularly

2582 Journal of Investigative Dermatology (2014), Volume 134 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

+/+ –/– +/+ –/– +/+ –/–kDa +/+ –/– +/+ –/– kDa 100 170 70 130 55 100 43 70 35 K2 K1 K10 55 26 43 70 35 55 * * 26 43 K5 K16 lor flg Protein 15

Krt2 +/+ Krt2 –/– Krt2 +/+ Krt2 –/–

inf

K1 K1 K5 K5

hf

K10 K10 K16 K16

inf inf

flg hf flg lor lor

K10 K1 K10 + K1

Figure 4. Deletion of K2 causes changes in epidermal differentiation and aggregation of K10. (a) Proteins were prepared from the ear skin of wild-type ( þ / þ )and K2-deficient ( / ) mice and subjected to western blot analysis using antibodies against the indicated keratins, loricrin (lor), and filaggrin (flg). Proportional but reduced amounts of proteins were electrophoresed on a separate gel and stained with Coomassie blue. Positions of molecular mass markers are indicated on the right. Arrowheads indicate the bands corresponding to the predicted size of the respective western blot antigen. An open arrowhead marks the K2 band. Asterisks mark the position of filaggrin monomers. (b) Immunofluorescence analysis of K1, K5, K10, K16, loricrin (lor), and filaggrin (flg) in the mouse ear skin. hf, hair follicle; inf, infundibulum. (c) Immunofluorescence double labeling of K10 and K1 in the ear skin of Krt2 / mice. The last panel shows a detail of the merged fluorescence image. Arrowheads indicate a cluster of cells containing K10 clumps. Bar ¼ 40 mm. confined to the cell periphery in association with desmosomes and e), suggesting that K10 was an essential and most probably (WTinFigure5eandKrt2 / Krt10 þ / þ in Figure 5f). the main component of these structures. Moreover, the upper suprabasal cells of all of our keratin Immunogold labeling showed that K2 and K10 formed the knockout variants contained characteristic large keratohyalin keratin filaments in suprabasal but not basal cells of the ear granules of various shapes (Figure 5b–d, Supplementary Figure epidermis of WT mice (Supplementary Figure S7c–f online). S8b and d online, Supplementary Figure S9c and d online), By contrast, K10 was strongly labeled in the electron-dense which often were also connected to desmosomes via filament parts of the spongy clumps of Krt2 / Krt10 þ / þ mice and bundles (Supplementary Figure S8b online, arrows). Suprabasal was absent from the cytoplasm of the same cells (Figure 5h–l; keratinocytes of Krt2 / Krt10 þ / þ mice contained distinct Supplementary Figure S8d and e online). The K10-positive spongy clumps of intermediate electron density (Figure 5b and clumps were often attached to desmosomes via filamentous g; Supplementary Figure S8d online), which were absent from bundles (Figure 5i and j; Supplementary Figure S8f and g Krt2 / Krt10 / mice (Figure 5d; Supplementary Fig. S9c online) or enclosed mitochondria (Figure 5l; Supplementary

www.jidonline.org 2583 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

a SC bcdSC SC

SC

KH KH KH KH KH

BM KA KH WT KH

e KH KA

KA

KA

WT KA KA

f

Krt2 –/– g

BM Krt2 –/–

h BM Krt10 –/– Krt2 –/– Krt10 –/–

BM mn

Krt2 –/– K10 Krt2 –/– k KA i KA

Mi K10 Krt2 –/–

–/– Krt2 l Mi j Mi

KA K10 Krt2 –/– K10 Krt2 –/– K2 Krt10 –/– K2 Krt10 –/–

Figure 5. Epidermal ultrastructure in the absence or presence of keratins K2 and K10. Electron microscopy (EM) (a–g, i) and immuno-EM (K10: h, j–l;K2:m, n) analysis of the ear skin of wild-type (WT) (a, e), K2-deficient (b, f–l), K10-deficient (c, m, n), and K2/K10 double-deficient (d) mice. In Krt2 / mice, the cytoplasm of suprabasal cells is rather electron-lucent with short, compact keratin filament bundles (f, arrows) closely associated with desmosomes (f; arrowheads). Large poriferous keratin aggregates (b, g; KA) are strongly K10 positive (h) and often associated with desmosmes via filaments (i, j; arrows) and regularly embedding mitochondria (l;Mi).Krt10 / mice show numerous small KA (c) labeled for K2, often close to desmosomes (m, n; arrows). Krt2 / Krt10 / keratinocytes lack KAs (d). Granular cells of all mutants possess large keratohyalin granules of various sizes (b–d, KH). BM, basement membrane; SC, stratum corneum. Bars ¼ 2 mm(a–d), 500 nm (e, g, h, k–n), 250 nm (i, j).

2584 Journal of Investigative Dermatology (2014), Volume 134 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

Figure S8d online). Frequently, short compact K10-positive heterozygous carriers of the mutant Krt10 allele (Porter et al., filament bundles connected adjacent desmosomes leading to 1996; Reichelt et al., 1997), the mutation of the Krt2 gene did peculiar protrusions of the cytoplasm (Supplementary Figure not lead to the expression of a stable truncated K2 protein, as S8e and f). In contrast, keratinocytes of Krt2 þ / þ Krt10 / evidenced by western blot (Figure 4a; Supplementary Figure mice did not contain such large spongy keratin aggregates S2b online) and immunofluorescence analyses (Supplemen- as observed in Krt2 / Krt10 þ / þ mice but, instead, small tary Figure S2c online) using an antibody against the N-term- clumps with uniform electron density (Figure 5c and m) that inal portion of K2, nor to aberrations in heterozygous mice were strongly positive for K2 by immunogold labeling (Supplementary Figures S3 and S5). Therefore, the phenotype (Figure 5m and n). Similar to K10 aggregates in Krt2 / of the mouse model reported here can be attributed to the Krt10 þ / þ mice, the K2-positive aggregates were regularly ablation of K2 expression. found near mitochondria and associated with desmosomes The development of severely disturbed epidermal home- (Figure 5n). In summary, our genetic experiments and ultra- ostasis on the ears, soles, and tail of K2 knockout mice structural investigations suggest that each one of the two demonstrated that K2 has an essential but body-site–restricted highly expressed suprabasal keratins, K2 and K10, depends on role in murine keratinocyte differentiation. The deletion of one the presence of the other at an equivalent level of abundance allele of Krt2, like hemizygous deletions of Krt10 (Reichelt and otherwise undergoes aberrant aggregation. However, the et al., 2001 and unpublished observations) or Krt1 (Roth et al., concomitant suppression of both K2 and K10 expression 2012), is well tolerated, at least under conditions of normal causes effects that are independent of any keratin aggregation mechanical stress to the skin. By contrast, homozygous and suffice to induce hyperkeratosis. deletion of type II Krt2 caused aberrant aggregation of the type I dimerization partner, K10, and defects in the DISCUSSION cytoskeleton by widely preventing cytoplasmic distribution The results of this study shed new light into the role of K2 of filaments and association of short dense bundles to in vivo and the interdependencies of K1, K2, and K10, which desmosomes. Interestingly, the absence of K2 induced the together form the keratin intermediate filaments of the differ- expression of K1 in the epidermis of the ear, and those cells, entiated keratinocytes of the suprabasal epidermis. Our results which contained high levels of K1 according to demonstrate that the epidermis of the mouse ear but not that of immunofluorescence, formed little if any K10 aggregates the back expresses the Krt2 gene. In addition to these data, (Figure 4c). This suggested that upregulation of K1 can which essentially confirm previous studies of K2 expression in partially compensate the loss of K2 in the ear skin. The the mouse (Rentrop et al., 1987; Herzog et al., 1994), the signaling pathway leading to upregulation of Krt1 is elusive, present study reveals the absence, and thus at least local and it remains to be investigated whether the induction of K1 dispensability, of K1 in the interfollicular epidermis of the expression depends on signals different from those that induce mouse ear, tail, and paw. The mutually exclusive expression hyperkeratosis-associated keratins (K6, K16). Notably, the pattern of K1 and K2 in the mouse notably differs from that in genes encoding K1 and K2 are located next to each other, the human epidermis, which contains K1 and K2 in the flanked by Krt77, which codes for another suprabasal keratin interfollicular skin of all body sites as well as in the ridged (Langbein et al., 2013). Krt77 is expressed during embryonic skin of palms and soles (Herzog et al., 1994, Swensson et al., and early postnatal development of the mouse (Langbein 1998). Although it has been suggested previously that the et al., 2013) but is not upregulated in K2-deficient ear skin expression of K1 is downregulated while that of K2 is (Supplementary Figure S6 online). Interestingly, the deletion of upregulated during terminal differentiation of human K2 led to elevated expression of the late differentiation interfollicular keratinocytes (Rothnagel et al., 1994), the markers, loricrin, and filaggrin, including the formation of continued presence of K1 in the granular layer has large keratohyalin granules, possibly representing a compen- prevented a clear conclusion about the role of K2 in the satory barrier maintenance program of suprabasal keratino- human epidermis. The determination of the mutually cytes. The interdependencies in the transcriptional control of exclusive expression patterns K1 and K2 in the mouse suprabasal type II keratins and cornification proteins remain to implies that both type II keratins are sufficient on their own be determined in future studies. to complement the type I keratin K10 in the formation of the The generation of K2, K10, and K2/K10 knockout mice intermediate filaments within the suprabasal epidermal and the ultrastructural characterization of their epidermis cytoskeleton. reveal that K2 forms aggregates when K10 is missing and In extending the list of mouse models carrying targeted that K10 forms aggregates when K2 is missing in the mutations of keratin genes, we report the phenotype of mice in suprabasal epidermis of mouse ears. These results provide which the Krt2 gene is disrupted by the insertion of a beta- experimental support for the hypothesis that K2 and K10 are galactosidase reporter gene followed by a poly-adenylation type I and II heterodimerization partners in the normal site. This construct leads to the effective suppression of suprabasal epidermis. To the best of our knowledge, this splicing to the 30-terminal exons 4–9 of the Krt2 gene but hypothesis has not been tested previously. The expression of allows the formation of a Krt2 mRNA containing an open a mutated K10 protein was reported to cause the disappear- reading frame for a truncated K2 protein. Unlike a mutation of ance of K2 from mouse sole skin, indicating that K2 depends the Krt10 gene, which resulted in the expression of a highly on K10 (Porter et al., 1996). However, deletion of another abundant truncated K10 protein and a marked phenotype in type I keratin, K9, also diminishes the abundance of K2

www.jidonline.org 2585 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

(Fu et al., 2014), although K2 and K9 are expressed in knockout mice on a Balb/c background (Reichelt et al., 2001; Wallace different regions of the mouse sole and footpads et al., 2012) were crossed to generate Krt2 / Krt10 / mice. All (Supplementary Figure S10 online) and, in contrast to a experiments were performed with age-matched animals. previous suggestion (Bragulla and Homberger, 2009), are therefore unlikely to heterodimerize in vivo.Our Antibodies and antisera ultrastructural characterization of K10-deficient ear skin All antisera and antibodies used are listed in Supplementary Table S1 confirms that K2 requires K10 for intermediate filament online. formation. By contrast, K10 has two optional heterodimeri- zation partners, i.e., K1 and K2, as suggested by our Histological and immunofluorescence analysis demonstration of coexpression of K10 with either K1 alone Tissues were fixed in phosphate-buffered 4.5% formaldehyde for 24 (on back skin) or K2 alone (in the interfollicular epidermis of hours and embedded in paraffin. Sections were stained with the ears). Similar to K10 clumping in the ear skin of hematoxylin–eosin or processed for immunofluorescence analysis K2-deficient mice, K10 forms aggregates in the back skin according to a protocol published previously (Fischer et al.,2007). of K1-deficient mice (Roth et al., 2012). The specificity of the immunolabelings was confirmed by replacing Although the clumping of the partner keratin appears to be a the primary antibodies with the respective isotype antibodies, which direct and, at the subcellular level, marked effect of K2 abolished the specific labeling. Primary and secondary antibodies deficiency in mouse ear epidermis, this phenomenon is used for immunofluorescence analysis are listed in Supplementary dispensable for the establishment of hyperkeratosis as evi- Table S1 online. Counterstaining was performed with Hoechst 33258 denced by the phenotype of K2/K10 double knockout mice. (Sigma Aldrich, St Louis, MO). Mast cell infiltrates in the dermis were Perhaps the lack of the normal intermediate filament cytoske- identified with 1% toluidine blue staining. Visualization and doc- leton leads to increased mechanical fragility of suprabasal umentation were performed using the AX 70 microscope (Olympus, keratinocytes and corneocytes and a defective epidermal Hamburg, Germany) equipped with a Spot slider camera (SPOT barrier that triggers epidermal thickening as an attempt to Imaging Solutions, Sterling Heights, MI) and the imaging software rescue the barrier. It is conceivable that the lack of the normal Metamorph (Visitron Systems, Puchheim, Germany). Whole mount keratin cytoskeleton directly induces the upregulation of K6 X-Gal staining of mouse tissues expressing beta-galactosidase reporter and K16, which may establish a proinflammatory milieu gene was performed as previously described (Lessard and Coulombe, similar to that found in several human skin diseases (Pan 2012). Pictures were recorded using the Leica MZ16A Stereo et al., 2013). Even distinct keratinopathies that are caused by Microscope (Leica Microsystems GmbH, Wetzlar, Germany). cell-autonomous defects in the cytoskeleton involve skin inflammation (Gu et al., 2003; Yoneda et al., 2004; Lu Western blot analysis and protein staining et al., 2007). Thus, it is interesting that Krt2 deletion causes Tissue samples were homogenized using a bead mill (Precellys24, the upregulation of thymic stromal lymphopoietin and IL-18. PEQLAB Biotechnology, Erlangen, Germany) containing 10 ceramic The latter has been reported to have a critical role in the beads of a diameter of 1.4 mm and lysis buffer containing 50 mM Tris pathological effects of K1 deficiency (Roth et al., 2012). (pH 7.4), 2% SDS, and 200 mM DTT (Sigma Aldrich, Vienna Austria) Unlike the deletion of Krt1, the deletion of Krt2 leads to a for 2 50 seconds at 6,500 r.p.m. The lysates were heated to 95 1C nonlethal, persistent, and locally restricted phenotype, which for 10 minutes and cleared by centrifugation. SDS–PAGE was makes the Krt2 knockout mouse a model suitable for testing conducted on 8–18% gradient gels (GE Amersham, Pharmacia therapeutic strategies of suppressing keratin-associated Biotech, Uppsala, Sweden). The proteins were then transferred by hyperkeratosis and inflammation in future studies. semidry electroblotting onto nitrocellulose membranes (GE), immu- In summary, the results of this study suggest that the high nolabeled with the primary and secondary antibodies listed in expression levels of K2 and K10 are associated with the Supplementary Table S1 online and detected by chemiluminescence requirement for sufficiently high levels of their respective with the ImmunStar Western C Substrate Kit (Bio-Rad, Hercules, CA) heterodimerization partner. The strikingly different contribu- according to the manufacturer’s instructions. On separate gels, tions of K2 and K1 to keratinization in the ear and back skin of proteins (amounts proportional to those used for western blot) were the mouse offer special opportunities to dissect the roles of separated by electrophoresis and stained with Coomassie Blue (Page individual keratins in the suprabasal epidermis and further Blue staining solution, Fisher Scientific, Vienna, Austria). characterization of body-site–specific features of the epidermis in the main model system of skin research. Transepidermal water loss Transepidermal water loss was measured on the outer side of ears MATERIALS AND METHODS immediately after euthanizing mice using a Tewameter from Courage Mice and Khazaka, model TM300 (Cologne, Germany), which was Mice were maintained according to the animal welfare guidelines of equipped with an adaptor having a diameter of 2 mm. the Medical University of Vienna, Austria. The Krt2 gene was targeted by the International Knockout Mouse Consortium KOMP-CSD (ID: Test of corneocyte stability 22728) (Skarnes et al., 2011) from which heterozygous mice were Cornified envelopes were prepared according to the method of Koch obtained for breeding to homozygosity. Genotyping and validation of et al. (2000).Imageswererecordedbyaninvertedmicroscopesetto the mice was performed as described in the Supplementary Information phase contrast mode (Olympus, model IMT-2, Hamburg, Germany), online. Krt2 knockout mice on a C57BL/6 background and Krt10 equipped with a digital camera (Olympus, E-520). Images were

2586 Journal of Investigative Dermatology (2014), Volume 134 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

analyzed in a blinded manner and the percentage of undamaged Fu DJ, Thomson C, Lunny DP et al. (2014) is required for the cornified envelopes was calculated relative to total number. structural integrity and terminal differentiation of the palmoplantar epidermis. J Invest Dermatol 134:754–63 Quantification of mRNA by reverse transcription real-time PCR Fuchs E (1990) Epidermal differentiation: the bare essentials. JCellBiol 111:2807–14 Reverse transcription and quantitative real-time PCRs were performed Gu LH, Kim SC, Ichiki Y et al. (2003) A usual frameshift and delayed according to published protocols (Fischer et al., 2011) using primers termination codon mutation in causes a novel type of listed in Supplementary Table S2 online. The expression levels epidermolysis bullosa simplex with migratory circinate erythema. J Invest of the target genes were normalized to that of the housekeeping Dermatol 121:482–5 gene beta-2-microglobulin. Herzog F, Winter H, Schweizer J (1994) The large type II 70-kDa keratin of mouse epidermis is the ortholog of human keratin K2e. J Invest Dermatol 102:165–70 Conventional and immunoelectron microscopy Koch PJ, de Viragh PA, Scharer E et al. (2000) Lessons from loricrin-deficient For electron microscopy, small pieces of ear skin were fixed in 2.5% mice: compensatory mechanisms maintaining skin barrier function in the glutaraldehyde in sodium cacodylate and postfixed in 2% osmium absence of a major cornified envelope protein. JCellBiol151:389–400 tetroxide. The specimens were stained overnight in 0.5% uranyl Kremer H, Zeeuwen P, McLean WHI et al. (1994) Ichthyosis bullosa of acetate, dehydrated, and embedded in Epon. Ultrathin sections were Siemens is caused by mutations in the keratin 2e gene. J Invest Dermatol inspected under an electron microscope EM910 (LEO, Oberkochen, 103:286–9 Germany) (Langbein, et al. 2004). Immunoelectron microscopy was Langbein L, Spring H, Rogers MA et al. (2004) Hair keratins and hair follicle- specific epithelial keratins. Methods Cell Biol 78:413–51 performed according to published protocols (Langbein et al., 2013). Langbein L, Reichelt J, Eckhart L et al. (2013) New facets of keratin K77: Details are described in the Supplementary Information online. Interspecies variations of expression and different intracellular location in embryonic and adult skin of man and mouse. Cell Tiss Res 354:793–812 CONFLICT OF INTEREST Lessard JC, Coulombe PA (2012) -null mice develop palmoplantar The authors state no conflict of interest. keratoderma, a hallmark feature of pachyonychia congenita and related disorders. J Invest Dermatol 132:1384–91 ACKNOWLEDGMENTS Lu H, Chen J, Planko L et al. (2007) Induction of inflammatory cytokines by a Mice carrying a hemizygous deletion of Krt2 were obtained from the keratin mutation and their repression by a small molecule in a mouse International Knockout Mouse Consortium (KOMP/EUCOMM). LE was sup- model for EBS. J Invest Dermatol 127:2781–9 ported by COST Action BM0903: SKINBAD (Skin Barrier in Atopic Diseases) Magin TM, Vijayaraj P, Leube RE (2007) Structural and regulatory functions of and by COST Action BM1002: Nanonet. ET and LE were supported by keratins. Exp Cell Res 313:2021–32 COST Action TD1206 StanDerm. This study was also supported in part McLean WH, Moore CB (2011) Keratin disorders: from gene to therapy. Hum by a research grant from CE.R.I.E.S., Neuilly, France, by the Wilhelm Mol Genet 20:R189–97 Sander-Stiftung, Munich, Germany (Grant 2007.133.2 to LL), and by Newcastle Health Care Charity and the Newcastle upon Tyne Hospitals McLean WHI, Morley SM, Lane EB et al. (1994) Ichthyosis bullosa of Siemens– NHS Charity (Grant PFC/ ML/0809 to JR). We thank Harald Hoeger for a disease involving keratin 2e. JInvestDerm103:277–81 maintaining the mice and Kathrin Sautner, Michael Mildner, and Christina Moll R, Divo M, Langbein L (2008) The human keratins: biology and Kukacka (Medical University of Vienna) for technical support and helpful pathology. Histochem Cell Biol 129:705–33 discussions. We are grateful to Ingrid Hausser-Siller (EM-lab Dermatology and Morley SM, Lane EB (1994) The keratinocyte cytoskeleton. In: Leigh IM, Lane EM Core facility, Heidelberg University, Germany) for invaluable discussions EB, Watt FM (eds.) The Keratinocyte Handbook. Cambridge University and the Core Facility Electron Microscopy for technical equipment (German Press: Cambridge, pp293–321 Cancer Research Center, Heidelberg, Germany). Omary MB, Coulombe PA, McLean WH (2004) Intermediate filament proteins and their associated diseases. N Engl J Med 351:2087–100 SUPPLEMENTARY MATERIAL Pan X, Hobbs RP, Coulombe PA (2013) The expanding significance of keratin Supplementary material is linked to the online version of the paper at http:// intermediate filaments in normal and diseased epithelia. Curr Opin Cell www.nature.com/jid Biol 25:47–56 Porter RM, Leitgeb S, Melton DW et al. (1996) Gene targeting at the mouse 10 locus: severe skin fragility and changes of cytokeratin REFERENCES expression in the epidermis. JCellBiol132:925–36 Arin MJ (2009) The molecular basis of human keratin disorders. Hum Genet Ramms L, Fabris G, Windoffer R et al. (2013) Keratins as the main component 125:355–73 for the mechanical integrity of keratinocytes. Proc Natl Acad Sci USA Basarab T, Smith FJ, Jolliffe VM et al. (1999) Ichthyosis bullosa of Siemens: 110:18513–8 report of a family with evidence of a keratin 2e mutation, and a review of Reichelt J, Bauer C, Porter R et al. (1997) Out of balance: consequences of a the literature. Br J Dermatol 140:689–95 partial knockout. JCellSci110:2175–86 Bragulla HH, Homberger DG (2009) Structure and functions of keratin Rentrop M, Nischt R, Knapp B et al. (1987) An unusual type-II 70-kilodalton proteins in simple, stratified, keratinized and cornified epithelia. JAnat keratin protein of mouse epidermis exhibiting postnatal body-site speci- 214:516–59 ficity and sensitivity to hyperproliferation. Differentiation 34:189–200 Candi E, Schmidt R, Melino G (2005) The cornified envelope: a model of cell Reichelt J, Bu¨ssow H, Grund C et al. (2001) Formation of a normal epidermis death in the skin. Nat Rev Mol Cell Biol 6:328–40 supported by increased stability of keratins 5 and 14 in keratin 10 null Collin C, Moll R, Kubicka S et al. (1992) Characterization of human cytokeratin mice. Mol Biol Cell 12:1557–68 2, an epidermal cytoskeletal protein synthesized late during differentia- Reichelt J, Magin TM (2002) Hyperproliferation, induction of c-Myc and tion. Exp Cell Res 202:132–41 14-3-3sigma, but no cell fragility in keratin-10-null mice. JCellSci115: Fischer H, Eckhart L, Mildner M et al. (2007) DNase1L2 degrades nuclear DNA 2639–50 during corneocyte formation. J Invest Dermatol 127:24–30 Reichelt J, Furstenberger G, Magin TM (2004) Loss of keratin 10 leads to Fischer H, Szabo S, Scherz J et al. (2011) Essential role of the keratinocyte- mitogen-activated protein kinase (MAPK) activation, increased keratino- specific endonuclease DNase1L2 in the removal of nuclear DNA from cyte turnover, and decreased tumor formation in mice. J Invest Dermatol hair and nails. J Invest Dermatol 131:1208–15 123:973–81

www.jidonline.org 2587 HFischeret al. Role of Keratin Pair K2/K10 in Skin Barrier

Roth W, Kumar V, Beer HD et al. (2012) maintains skin integrity and spatial and temporal pattern of keratin expression in development. Br J participates in an inflammatory network in skin through interleukin-18. Dermatol 140:582–91 JCellSci125:5269–79 Swensson O, Langbein L, McMillan JR et al. (1998) Specialized keratin Rothnagel JA, Traupe H, Wojcik S et al. (1994) Mutations in the rod domain of expression pattern in human ridged skin as an adaptation to high physical keratin 2e in patients with ichthyosis bullosa of Siemens. Nature Genet stress. Br J Dermatol 139:767–75 7:485–90 Uitto J, Richard G, McGrath JA (2007) Diseases of epidermal keratins and their Seltmann K, Fritsch AW, Ka¨sJAet al. (2013) Keratins significantly contribute to linker proteins. Exp Cell Res 313:1995–2009 cell stiffness and impact invasive behavior. Proc Natl Acad Sci USA Wallace L, Roberts-Thompson L, Reichelt J (2012) Deletion of K1/K10 does not 110:18507–12 impair epidermal stratification but affects desmosomal structure and Skarnes WC, Rosen B, West AP et al. (2011) A conditional knockout nuclear integrity. J Cell Sci 2012 125:1750–8 resource for the genome-wide study of mouse gene function. Nature Yoneda K, Furukawa T, Zheng YJ et al. (2004) An autocrine/paracrine loop 474:337–42 linking aggregates to tumor necrosis factor alpha-mediated Smith LT, Underwood RA, McLean WH (1999) Ontogeny and regional cytotoxicity in a keratinocyte model of epidermolysis bullosa simplex. variability of keratin 2e (K2e) in developing human fetal skin: a unique JBiolChem279:7296–303

2588 Journal of Investigative Dermatology (2014), Volume 134