Inhibition of Histone Deacetylation Promotes Abnormal Epidermal Differentiation and Specifically Suppresses the Expression of the Late Differentiation Marker Profilaggrin

ORIGINAL ARTICLE

Inhibition of Histone Deacetylation Promotes Abnormal Epidermal Differentiation and Specifically Suppresses the Expression of the Late Differentiation Marker Profilaggrin Nelli G. Markova1, Nevena Karaman-Jurukovska1, Adriana Pinkas-Sarafova1, Liuben N. Marekov2 and Marcia Simon1 Reversible protein acetylation modulates higher-order chromatin structure and transcription activity of the genome. The reversible acetylation is executed by the intrinsic acetylase and deacetylase activities of coregulators associated with the regulatory regions. Compounds capable of inhibiting deacetylase activity are a powerful tool for dissecting the role of protein acetylation in gene function. The ability of the deacetylase inhibitors to preferentially affect the homeostasis of transformed cells has also prompted studies for their clinical application. We present evidence that deacetylase inhibition with trichostatin A (TSA) affects the normal epidermal tissue architecture and pattern of expression by a mechanism(s) that does not correlate directly with the hyperacetylated histone status. While promoting abnormal differentiation, TSA specifically represses transcription initiation of the differentiation marker profilaggrin. Multiple factors, among which we have identified decreased Sp1 binding, a local decrease in acetylation activity, and enhanced synthesis and recruitment of a repressor histone demethylase, alter the chromatin configuration over the promoter, ultimately blocking its activation by c-jun. As compromised profilaggrin production leads to epidermal and consequently allergic disorders, our findings emphasize the need for a detailed investigation of the role deacetylase inhibitors may play in the maintenance of epidermal homeostasis in order to optimize their clinical applicability. Journal of Investigative Dermatology (2007) 127, 1126–1139. doi:10.1038/sj.jid.5700684; published online 28 December 2006

INTRODUCTION The eukaryotic chromatin is generally divided into two cytologically and functionally distinct fractions: the tightly compacted heterochromatin associated with transcriptionally silent genome regions and the less condensed euchromatin, which tends to be transcriptionally permissive (Wolffe and Kurumizaka, 1998). Among the factors that mediate chromatin transition is the deposition of covalent post-translational modifications on histone and non-histone chromatin proteins. The specific protein marks and the regulatory proteins 1

Department of Oral Biology and Pathology, The Living Skin Bank, School of Dental Medicine, SUNY Stony Brook, Stony Brook, New York, USA and 2 Laboratory of Structural Biology Research, NIAMS, NIH, Bethesda, Maryland, USA Correspondence: Drs Nelli G. Markova or Marcia Simon, Department of Oral Biology and Pathology, The Living Skin Bank, SUNY at Stony Brook, Stony Brook, New York 11794-8702, USA. E-mails: [email protected] or [email protected] Abbreviations: AP1, activator protein-1; GAPDH, glyceraldehyde-3phosphate dehydrogenase; HDAC, histone deacetylase; HDACi, histone deacetylase inhibitors; TSA, trichostatin A; NHEK, normal human epidermal keratinocytes grown in submerged culture; CAT, chloramphenicol acetyl transferase; LSD1, Lysine-specific demethylase 1 Received 12 July 2006; revised 18 October 2006; accepted 31 October 2006; published online 28 December 2006

1126 Journal of Investigative Dermatology (2007), Volume 127

that make them, remove them or act upon their presence epigenetically determine the transcription activity and thereby represent key regulators of cellular homeostasis (e.g., see Ura et al., 1997; Workman and Kingston, 1998; Spencer and Davie, 1999; Hake et al., 2004; Rosenfeld et al., 2006). The reversible acetylation is one of the best-studied posttranslational modifications modulating gene expression. Acetylation is generally correlated with active transcription, whereas deacetylation is more often linked to transcriptional repression (Ura et al., 1997; Workman and Kingston, 1998). The pattern of acetylation is controlled by the opposing activities of histone acetylases and deacetylases (HDACs) (Kuo and Allis, 1998). As the actively transcribed chromatin associates with dynamically acetylated proteins, the recruitment of both histone acetylases and HDACs to the regulatory regions through DNA-binding transcription factors serves to switch on and off gene transcription in response to cellular needs and extracellular cues (see Rosenfeld et al., 2006). However, mistargeting or perturbed balance between acetylating and deacetylating activities leads to aberrant expression of key regulators of cell growth, differentiation, and apoptosis and is often associated with malignancy (see Marks et al., 2004; Peart et al., 2005). The elucidation of the role of protein (de)acetylation in norm and disease has been facilitated by the discovery of & 2006 The Society for Investigative Dermatology

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

chemically diverse agents capable of inhibiting the protein deacetylases. In transformed cells and solid tumors and to a lesser extent in normal cells HDAC inhibitors (HDACi) induce cell growth and cell cycle arrest, differentiation and, in certain cases, apoptosis (see Yoshida et al., 2003; Marks et al., 2004; Dokmanovic and Marks, 2005; Peart et al., 2005). The ability of HDACi to preferentially affect the homeostasis of the transformed cells has prompted studies of their clinical application as anticancer agents either alone or in combination with other chemotherapeutics (see Johnstone and Licht, 2003; Marks et al., 2004; Garcia-Manero and Issa, 2005). Pretreatment of cancer cells with HDACi appears also to enhance their sensitivity to ionizing radiation (Zhang et al., 2004, see Karagiannis and El-Osta, 2006). Therefore, there is an ongoing effort to design HDACi specifically optimized for their radiation sensitizing properties (Jung et al., 2005). The therapeutic benefit of the radiotherapy, however, is limited by acute and long-term side effects that are especially pronounced in skin, which is always affected by external radiotherapy. Unexpectedly, it was recently observed that HDACi might also serve as radioprotectors of normal rat skin (Chung et al., 2004). These findings open the possibility of using HDACi to not only enhance the anticancer potential of the radiotherapy but also to selectively reduce skin morbidity. Earlier studies indicate that when the activity of HDACs is inhibited, cultures of both transformed and normal human epidermal keratinocytes cease to proliferate and rapidly undergo differentiation (Schmidt et al., 1989; Staiano-Coico et al., 1990a; Saunders et al., 1999; Brinkmann et al., 2001). Whether inhibition of protein deacetylation affects the homeostasis of the normal epidermal tissue, however, is unknown. We initiated the present studies with the goal to (i) examine the effect of inhibition of protein deacetylation on epidermal structure and (ii) determine whether HDACi can coordinately regulate the transcription of epidermal genes expressed during distinct stages of epidermal differentiation. We observed that inhibition of protein deacetylation in organ epidermal cultures with the classic HDACi trichostatin A (TSA) markedly affects the epidermal architecture and the expression pattern of epidermal marker genes by a mechanism(s) that does not correlate directly with their status of histone acetylation. Of the three genes residing in the Epidermal Differentiation Complex (EDC) (Mischke, 1998), TSA induces involucrin, has little effect on loricrin but specifically inhibits the initiation of profilaggrin transcription. The TSA-dependent repression of the endogenous profilaggrin transcription depends on protein synthesis and can be reproduced on minichromosomes carrying a reporter gene controlled by the profilaggrin promoter. Although it remains associated with acetylated histones, decreased levels of endogenous Sp1 binding, a local decrease in the acetylation activity, and enhanced synthesis and recruitment of a repressor histone H3 demethylating enzyme most likely contribute to altering the chromatin configuration over the promoter, thereby blocking its activation by c-jun. As aberrant profilaggrin/filaggrin production and function leads to epidermal and consequently allergic disorders (Palmer et al., 2006; Smith et al., 2006), our findings

emphasize the need for a detailed investigation of the mechanisms through which HDACi impact the epidermal homeostasis in order to optimize their clinical applicability. RESULTS Inhibition of protein deacetylation promotes abnormal epidermal differentiation but does not elicit a coordinated response from epidermal marker genes

To assess the role of protein acetylation in the maintenance of normal epidermal architecture, we incubated organ cultures of human foreskin for 48 hours in keratinocyte growth medium in the absence or presence of 100 nM TSA – an inhibitor of class I/II protein deacetylases (Yoshida et al., 1990). As revealed by the hematoxylin and eosin staining of the frozen sections, deacetylase inhibition led to drastic changes in the epidermal architecture (Figure 1a, e). All epidermal layers appeared affected. The basal layer lost its columnar organization and showed a reduced cellularity. Fewer spinous layers were observed. There was an increase in the thickness of the cornified layers. Similarly, after 2 days in TSA-containing medium, both subconfluent and confluent submerged cultures of normal human epidermal keratinocytes (NHEK) contained many enlarged cells with a flattened appearance and uneven periphery (data not shown). On the protein level, in the TSA-treated NHEK we observed decreased expression of the proliferation marker c-myc and induction of the differentiation marker cyclin-dependent kinase inhibitor p21WAF1/CIP1; the expression of cyclin D was not affected (Figure 1i, compare lanes 1 and 2). Both control (Figure 1i, lane 3) and NHEK treated with 100 nM (lane 4) and 200 nM TSA (lane 5) showed equal amounts of the poly(ADP-ribose) polymerase 1 – a specific target of chemically induced apoptosis-mediated proteolysis (Kaufmann et al., 1993). Significantly, incubation for 48 hours even with 200 nM TSA did not increase the intensity of the signature 85 kDa apoptotic fragment (arrow). Expanding the published data discussed above (Schmidt et al., 1989; Staiano-Coico et al., 1990a; Saunders et al., 1999; Brinkmann et al., 2001), these findings suggested that without inducing apoptosis TSA was blocking the proliferation and concomitantly accelerating the terminal differentiation of the epidermal keratinocytes. Therefore, we hypothesized that TSA may coordinately regulate the expression of the epidermal markers genes and/or gene clusters during the keratinocyte differentiation. We focused on the involucrin, loricrin, and profilaggrin genes. All three reside in the EDC on 1q21 (Mischke, 1998), encode proteins expressed in the later stages of epidermal differentiation and by both criteria were expected to be upregulated by TSA. As seen in Figure 1, treatment of the organ cultures with 100 nM TSA for 48 hours appeared to increase the amount of involucrin (b, f) and to restore the upper spinous layer specificity of expression, which is routinely lost in organ cultures (Simon M., unpublished results). The expression of loricrin remained practically unchanged (c, g), whereas the amount of profilaggrin was markedly reduced (d, h). These trends were confirmed on both RNA and protein level in epidermal preparations derived from control and TSA-treated foreskin www.jidonline.org 1127

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

a

Control

TSA-treated

e

M

– +

i

Cyclin D H&E

PARP1 – + +

191

c-myc

97 64

p21

3

1 2

b

f

j +

–

–

+

Inv – +

4 5 Foreskin Lori + –

PF +

–

RNA

Inv

18s

Protein

c

g 1

2

3

4

5

6

k–

+

–

+

Inv – +

7

8

9

10

Lori

d

NHEK Lori – +

–

PF +

h

RNA 18s

PF

– +

– +

– + Protein

1 2

3

4

5

6

7

8

9 10

Figure 1. Effect of protein deacetylase inhibition in organ and submerged epidermal cultures. (a) Hematoxylin and eosin (a, e, bar ¼ 30 mm) and (b–h, bar ¼ 12 mm) indirect immunofluorescent staining of frozen sections prepared from control and TSA-treated organ cultures of foreskin epidermis. The green color shows the signals for involucrin (Inv), loricrin (Lori), and profilaggrin (PF) detected after incubation with the respective antibodies and Alexafluor 488conjugated secondary antibodies. The nuclei were visualized with DAPI. (i) The expression of the designated proteins in total NHEK extracts in the absence (lanes 1, 3), 100 nM (lanes 2, 4) and 200 nM (lane 5) TSA. The arrow points to the 85 kDa polypeptide generated upon apoptotic-specific cleavage of poly(ADPribose) polymerase 1. (j and k) Analyses of RNA and protein levels of the designated epidermal markers in control () and 100 nM TSA-treated ( þ ) (j) foreskin organ and (k) submerged NHEK cultures. The arrowhead points to the mature filaggrin peptide discernible only in the foreskin preparations. Lanes 1 and 2 are photographs of ethidium bromide stained total RNA gels typically used in the analyses; lanes 3 and 4 represent GelCode Blue stained gels of the total proteins extracted after the removal of the RNA. The RNA signals (top rows) are in reference to 18S rRNA (middle rows).

organ cultures (Figure 1j) and were also reproduced in NHEK (Figure 1k). TSA treatment of foreskins and confluent NHEK increased involucrin mRNA (top) and protein (bottom) production (Figure 1j, k, lanes 5, 6). It did not affect the loricrin mRNA levels (Figures 1j, k, lanes 7, 8, top). The amount of the protein did not change in the foreskin epidermis (Figure 1j, lanes 7, 8, bottom) and in NHEK remained barely detectable (Figure 1k, lanes 7, 8, bottom). Profilaggrin mRNA and protein levels were only 20–30% of the untreated controls (Figures 1j and k, lanes 9, 10). Apparently, the TSA-mediated profilaggrin suppression was not significantly influenced by differentiation, as it was observed with NHEK grown with 3T3 feeder cells in serumcontaining medium, as well as in NHEK grown in serum-free medium at low (0.05 mM) and high (1.2 mM) calcium concentrations (data not shown). TSA modulates the expression of the late differentiation markers at the level of transcription initiation

Establishment of a permissive chromatin configuration through histone acetylation is a prerequisite for transcriptional activation (Ura et al., 1997). As chromatin 1128 Journal of Investigative Dermatology (2007), Volume 127

immunoprecipitation with a pan-acetyl antibody revealed, in NHEK, where they are normally transcribed, profilaggrin, loricrin, and involucrin genes (Figure 2a) are associated predominantly with the acetylated, active chromatin. In contrast, in the control dermal fibroblasts the epidermalspecific sequences (shown for profilaggrin) were detected in the hypoacetylated chromatin fraction, which was not immunoprecipitated with the pan-acetyl antibody. In both cell types the gene for the housekeeping glyceraldehyde-3phosphate dehydrogenase (GAPDH), used as a control, was found predominantly within the acetylated chromatin. Significantly, treatment with TSA, which caused accumulation of acetylated histones in both NHEK and dermal fibroblasts (Figure 2b), did not lead to redistribution of the profilaggrin gene between the two chromatin fractions in either cell type. To determine whether the TSA-mediated alterations occurred at the level of transcription initiation or resulted from post-initiation events, we isolated nuclei from control and TSA-treated NHEK, incubated them with [a-32P]GTP and hybridized the labeled nascent transcripts to immobilized profilaggrin, involucrin, loricrin, or GAPDH cDNA sequences. As seen in Figure 2c, the nuclei isolated from

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

a

KC DF KC DF KC DF KC DF PF GAPDH PF GAPDH PG TSA – + – + – + – + – + – + – + – + – –

b

d TSA (hour) 0

DF KC Bound Unbound Anti-pan-acetyl

Bound NIS

c TSA – PF Inv

+

–

+

9 24

g

– – + – + –

+ +

CHX M TSA PF GAPDH

e TSA (hour) 0–

1 6 9 24 24

++ + – +

1

2 3 4

Ac-His

Lori Inv Lori Inv TSA – + – + – + – + – + – +

b ub b ub b b Anti-pan-acetyl NIS

6

PF c-myc 18s

–+ – +

Gelcode Blue

f

Hours TSA

0 – +

CHX ActD 9 9 9 – + – + – +

PF Lori GAPDH

GAPDH

Figure 2. TSA represses profilaggrin transcription initiation. (a) Chromatin immunoprecipitation analysis of profilaggrin (PF), loricrin (Lori), involucrin (Inv), and GAPDH DNA in chromatin isolated from control () and TSA-treated ( þ ) NHEK (KC) and dermal fibroblasts (DF). The crosslinked chromatin was immunoprecipitated with anti-pan-acetyl antibody or with nonspecific IgGs (NIS) and the designated DNA sequences in the bound and unbound fractions were detected by PCR. (b) Acetylated histone H3 detected by Western blotting in the nuclear extracts of control () and TSA-treated ( þ ) NHEK and dermal fibroblasts. (c) Nascent profilaggrin, loricrin, involucrin and GAPDH transcripts synthesized in nuclei isolated from control () and TSA-treated ( þ ) NHEK. (d) Levels of profilaggrin, c-myc, and 18S rRNA detected by RNase protection in NHEK treated with TSA for the indicated times. (e) Total histones were obtained from the same cells after RNA isolation and the level of acetylation detected with anti-pan-acetyl antibody. The GelCode staining of the protein gel is shown as a reference. (f) Effect of cycloheximide (CHX) and actinomycin D (ActD) on the TSA-mediated regulation of profilaggrin and GAPDH mRNA levels in NHEK, detected by RT-PCR. (g) Nascent profilaggrin and GAPDH transcripts synthesized in nuclei isolated from control (lane 1), TSA (lanes 2, 4) and cycloheximide (lanes 3, 4) treated NHEK, detected after RNase H protection. Lane M shows the mobility of the denatured 90, 76, and 67 bp fragments of the pBR322/MspI DNA marker (New England BioLabs, Beverly, MA).

TSA-treated NHEK initiated markedly less profilaggrin but more involucrin transcripts compared to the untreated controls; the loricrin and the control GAPDH transcripts were practically unaffected. Thus, while not affecting loricrin, inhibition of protein deacetylation exerts its opposing effects on profilaggrin and involucrin production already at the level of transcription initiation. TSA-mediated profilaggrin repression is sensitive to inhibition of protein synthesis

Treatment of NHEK with 100 nM TSA for up to 24 hours revealed that the decrease in profilaggrin mRNA levels occurred between 6 and 9 hours of incubation and was delayed compared to the effect on c-myc mRNA (Figure 2d). It was observed several hours after a hyperacetylated histone status was established (Figure 2e). Therefore, it was unlikely that the prevailing mechanism of the TSA-mediated repression involved immediate chromatin remodeling or modifying the activity of an already present transcriptional regulator(s) following hyperacetylation of the histones associated with the profilaggrin gene. Rather, the data suggested the synthesis of a factor in trans, which then interfered with the initiation of profilaggrin transcription. Indeed, inhibitors of both translation and transcription were able to reverse the TSA effect. When NHEK were treated with TSA for 24 hours and subsequently for 9 hours with either cycloheximide or actinomycin D in the presence of TSA (Figure 2f), the level of profilaggrin mRNA in the presence of TSA was higher than in its absence. Moreover, nuclei isolated from NHEK treated with TSA for 24 hours and then for additional 9 hours with TSA and cycloheximide, synthesized only about 3-fold less profilaggrin transcripts, compared to the cycloheximide-

untreated TSA-treated controls (Figure 2g, compare lanes 4 and 2). In comparison, in nuclei isolated from cells that had not been treated with TSA but subjected to treatment with cycloheximide, profilaggrin transcription initiation was inhibited more than 10-fold (compare lanes 3 and 1). TSA treatment did not alter the amount of nascent GAPDH transcripts. As previously reported (Nirunsuksiri et al., 1998), the initiation of GAPDH transcription was not sensitive to cycloheximide either. The inhibitory effect of TSA is transmitted through specific regulatory elements

To explore whether the TSA-dependent inhibition of profilaggrin transcription affects the basal transcription machinery exclusively or involves additional regulatory elements upstream of the transcription initiation site, we transfected NHEK with chloramphenicol acetyl transferase (CAT) reporter plasmids controlled by decreasing lengths of the profilaggrin 50 -region 1532/ þ 9 (Jang et al., 1996). As seen in Figure 3a 100 nM TSA repressed the activity of all profilaggrin constructs. Under the same conditions the activity of CAT reporters regulated by involucrin (2473/ þ 9), loricrin (154/ þ 9), or keratin K5 (705/ þ 6) promoter regions remained unaltered. The minimal region of the profilaggrin promoter, which retained full responsiveness to TSA in a concentration- (Figure 3b) and time- (Figure 3c) dependent manner, extended 116 bp upstream of the transcription initiation site. Although we cannot exclude involvement of the basal transcription apparatus, the fact that constructs 59/ þ 9-CAT and 41/ þ 9-CAT gradually lost responsiveness to TSA indicated that upstream regulatory elements within the proximal promoter region were required www.jidonline.org 1129

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

0

c 0

b

150

24

100 100

48

50 200

Relative CAT activity (%)

a

0

–1532 –354 –116 –59 –41

Inv

Lori

K5

TSA (nM)

pCAT constructs

Time (hours)

–116/+9

d –150 S-100

ets

S-70

AP1

S-50

–100 S-50

S-35

Oct

–50

+1

Relative CAT activity (%)

e

HDAC inhibition alters the DNA/protein interactions over the promoter

f

150

400 300

100 200

50 100

0

0 wt S-100 ets AP1 S-70 S-50 S-35 OctA

constructs 2.5 to 3-fold. The exceptions were the reporters S-100 and S-50, which harbored mutations in the GT repeats at positions 101 to 96 and 57 to 45, respectively. The S-100 and S-50 mutant constructs retained only 30% of the wild-type activity but were not sensitive to TSA (Figure 3e). Mutation of the S-100 and S-50 motifs in the context of the entire 50 -region up to 1532 elicited the same response (data not shown), thus confirming their essential role in the transmission of the TSA effect. Similar to its impact on the endogenous profilaggrin expression, the TSA mediated downregulation of profilaggrin promoter activity in the minichromosomes was differentiation independent (data not shown). Notably however, it was keratinocyte specific. In HeLa cells, which do not transcribe the profilaggrin gene but in which the profilaggrin promoter shows a weak activity (Jang et al., 1996), TSA induced at least 2-fold higher reporter expression, which depended on an intact S-50 motif (Figure 3f).

wt S-100 ets

AP1 S-50 OctA

pCAT constructs Figure 3. The effect of TSA is transmitted through specific regulatory sequences. (a) Relative CAT activity detected in extracts of control (black bars) NHEK transfected with reporters regulated by decreasing lengths of the profilaggrin 50 -region between 1532 and þ 9, involucrin region 2473/ þ 9, loricrin region 154/ þ 9, and K5 region 705/ þ 6 and in NHEK treated with 100 nM TSA for 48 hours (striped bars). (b and c) The TSA dose and time responses, respectively of reporter 116/ þ 9-CAT. (d) Schematic presentation of the profilaggrin proximal promoter and the designated regulatory motifs. The arrow marks the transcription initiation site. (e, f) Relative CAT activity of the wild-type and the designated mutant reporters in control (black bars) and TSA-treated (striped bars) (e) NHEK and (f) HeLa cells. The activity of each of the wild type reporters in the control cells is set at 100%. The data (mean7SD) are an average of at least five experiments, each in duplicate.

for full suppression. In contrast, the regulatory elements responsible for the enhanced involucrin transcription were apparently not present or functional within the respective reporter. Mutation scanning and functional analyses (Jang et al., 1996, 2000; Andreoli et al., 1997; Markova et al., 2006 and unpublished results (Markova N.G., Jang S-I, Steinert P.M., unpublished results)) had previously identified proximal motifs indispensable for full profilaggrin promoter activity (Figure 3d). After transfection into NHEK, treatment with 100 nM TSA for 48 hours reduced the expression of the wild type 116/ þ 9-CAT and all but two of the indicated mutant 1130 Journal of Investigative Dermatology (2007), Volume 127

Comparison between DNase I digestion profiles of profilaggrin promoter DNA unspecifically associated with BSA (Figure 4a, lanes 1, 2) or specifically bound to nuclear proteins isolated from control (lanes 3, 4) and TSA-treated (lanes 5, 6) NHEK revealed that deacetylase inhibition did not grossly change the pattern of protection. Under both conditions the region encompassing nucleotides 30 to 99 was protected from DNase I digestion and showed DNase I hypersensitivity at the nucleotides marked with asterisks. Largely, this profile was determined by simultaneous binding of Sp1 and c-jun (compare lanes 1–6, 9 with lanes 10–12). However, with the TSA-treated samples we detected two new DNase I hypersensitive sites – at position 52 and 96 (arrowheads), which resided within the S-50 and S-100 motifs, respectively. The hypersensitivity at position 95 (marked with a dot) was no longer discernible. The TSA-specific DNase I digestion profile depended on intact S-50 and S-100. Mutation of S-50 abolished the footprint between 30 and 60, as well as the hypersensitivity at positions 35 and 55; the TSA-specific hypersensitivity at position 52 was no longer present (compare lanes 5–7). Similarly, mutation of S-100 reduced the protection between 90 and 99 and nucleotides 95 and 96 were no longer preferred targets for the DNase I attack (compare lanes 5, 6, and 8). These notable distinctions were indicative of changes in the chromatin structure. The bandshift profiles presented in Figure 4b, c indicated that the predominant interactions (complex A) of S-50 and S100 involve transcription factors Sp1 and Sp3. In addition, both motifs bind weakly to the Notch effector RBP-Jk (complex B). Compared to the untreated controls, the binding profiles of S-50 and S-100 with nuclear extracts of TSAtreated NHEK consistently showed a 40–50% decrease in the intensity of complex A, whereas the intensity of complex B remained unchanged (Figure 4d, lanes 1–4). The decrease in the Sp1/Sp3 binding was not caused by cytotoxicity of TSA, because the same extracts showed higher intensity of the

NG Markova et al.

a m

NHEK

S50 S- m 10 0

BSA

1 23 4 5 6 7 8 9

d

c-

A B

ju n Sp 1 an

1

1 2 3 4 5 6 7 8 9

RBP-J


c-

SP

A BS

c

Sp1, Sp3

ju n

GA

dI/dc self S-100m SP1c SP1cm SP1ab SP3ab RBP-J
SP2ab

b

dI/dc self S-50m SP1c SP1cm RBP-J
SP1ab SP3ab SP2ab

Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

S-50

S-100

– 30 – 40

d TSA

– 50

– + 1 2

– + 3 4

– + – – 7 8 9 10

– + 5 6

Sp1 c-jun

– 60 A B

A B

– 70

C

– 80 – 90 – 100

e – 110 TSA: – – + + 1 2 3 4 5 6

+ + 7 8

9 10 11 12

S-100 S-50 AP1-c – + – + – + SP1 SP3 c-jun junB

AP1/S-50 – + – + RBP-Jk junD c-fos

Gelcode blue

Figure 4. TSA modifies the DNA/protein interactions over profilaggrin promoter. (a) DNase I footprints of wild-type and indicated mutant profilaggrin proximal promoter sequences in the presence of BSA (lanes 1, 2, and 9), NHEK nuclear proteins (lanes 3–8), recombinant Sp1 (lane 10), c-jun (lane 11), or Sp1 and c-jun (lane 12). The black triangles show digestion with decreasing amounts of DNase I. The long double-headed arrows show the protected regions. The stars mark the DNase I hypersensitive sites common for both control () and TSA-treated ( þ ) NHEK samples, the arrowheads point to the hypersensitive sites unique for the TSA-treated preparations, the dot marks the hypersensitive site evident only upon interactions of the promoter with nuclear proteins derived from control NHEK. The GA lane represents the pyridine residues profile after Maxam and Gilbert chemical sequencing of the promoter region. The numbers on the left show the positions of the nucleotides relative to profilaggrin transcription initiation site. (b–d) Bandshift analysis of the indicated oligonucleotides after interactions with NHEK nuclear proteins. A, B, C represent the specifically retarded bands, the stars point to nonspecific complexes; the oligonucleotide competitors and the interfering antibodies are shown above the lanes. The positions of the supershifts (in italics) obtained upon incubation with the indicated antibodies are marked with arrowheads. (e) Western blot analyses with the indicated antibodies of the nuclear extracts (shown on the left) derived from control () and TSA-treated ( þ ) NHEK.

activator protein 1 (AP1)-containing complexes (lanes 5, 6) (Jang et al., 1996) and incubation of NHEK with 200 nM TSA for 48 hours did not induce apoptotic-specific cleavage of the poly(ADP-ribose) polymerase 1 enzyme (Figure 1i). Thus, the observed intensities of the bandshifts likely reflected the altered amounts of the respective transcription factors in response to TSA (Figure 4e). Surprisingly, therefore, in the presence of nuclear extracts from TSA-treated NHEK, oligonucleotides AP1/S-50 (Figure 4d) and AP1/S-100 (data not shown), which encompassed AP1 and S-50 or S-100 motifs, respectively, assembled not only fewer Sp1-specific complexes A, but also fewer AP1-containing complexes C; no change in the intensity of complex B was detected (compare lane 7, 8). As none of the AP1 proteins known to interact with the profilaggrin AP1 motif (Jang et al., 1996) were found downregulated by TSA (Figure 4e), these results indicated that, through interactions over S-50 and S-100, treatment with TSA could alter the protein occupancy at the AP1 motif. Forced expression of Sp1 alleviates the TSA-mediated repression and releases the block on c-jun transactivation

To explore the functional relevance of the above observations, we co-transfected wild type and mutant profilaggrin

354/ þ 9-CAT reporters in NHEK together with Sp1 or/and c-jun-loaded expression vectors (Figure 5). Sp1 enhanced the expression of the wild-type (a) and the AP1-mutant (e) reporters in a dose-dependent manner in both TSA-untreated and treated NHEK. Moreover, forcing higher levels of Sp1 into the cells interfered with the repressive effect of TSA at both 100 and 200 nM concentrations (b). The S-100 (c) and S-50 (d) mutant reporters remained Sp1-responsive, probably because interactions of Sp1 at the remaining intact S-motifs were still able to activate the promoter, albeit to a lesser extent and at higher Sp1 concentrations (see also Markova et al., 2006). As TSA only marginally repressed the activity of S-100 and S-50, it is unclear to what extent Sp1 contributed to the alleviation of the TSA effect for these reporters. Underscoring the cell type specificity, forced expression of Sp1 did not increase the reporter expression in the control HeLa cells but prevented the TSA-mediated activation in S-100 and S-50-dependent mode (data not shown). In contrast to Sp1, the dose-dependent activation of the wild type promoter enacted by forced expression of c-jun (Figure 5f) in the control NHEK (black bars) was completely prevented by the deacetylase inhibitor (striped bars). The inability of c-jun to transactivate the promoter in the presence www.jidonline.org 1131

NG Markova et al.

Relative CAT activity (%)

Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

a

b

300

Sp1

Sp1

200 100 0

0.2 g 0.5 g

Mock

Mock

c 300

d

e Sp1

Sp1

Sp1

200 100 0 Mock

0.2  g 0.5 g

Relative CAT activity (%)

0.2 g 0.5 g

Mock

AP1-CAT

S-50-CAT

f

300

0.2 g 0.5 g

Mock

S-100-CAT

g c-jun

h

c-jun

c-jun

200 100 0 Mock

0.2 g 0.5 g

–354/+9-CAT

Mock

0.2 g 0.5 g

0.2 g 0.5  g

Mock

S-100-CAT

i

S-50-CAT

j

1000

c-jun/Sp1

Sp1/Sp3 300 200

500

m

Sp1/Sp3

Sp3

0

0

0.2  g ea

Sp1/Sp3

Sp1

0.5  g

0.5 g

M

0.2  g

AP1

0

0.2 g

0.5 g

S-50

0.2  g

0.5  g

0.2  g

wt

0.5  g

0

0

100 0.2 g

Relative CAT activity (%)

Sp1

–354/+9-CAT

0

Relative CAT activity (%)

–354/+9-CAT

Figure 5. Inhibition of protein deacetylation interferes with c-jun activation of profilaggrin promoter. Relative activities of wild-type and mutant profilaggrin CAT reporters in control (black bars) and TSA-treated (striped and shaded bars) NHEK co-transfected with (a–e) 0.2 or 0.5 mg of Sp1, (f–h) c-jun, or both (i) Sp1 and c-jun expression vectors. The cells in (b) were cotransfected with 0.5 mg Sp1 and treated with 100 nM (striped bars) or 200 nM (shaded bars) TSA. (j) The reporter 354/ þ 9 was co-transfected with 0.2 or 0.5 mg Sp1 and Sp3, Sp1 þ Sp3, or Sp1 þ Sp3-mutant expression vectors and treated with 200 nM TSA (shaded bars). For each set, the activity of CAT is presented as a percentage of the activity of the mock-transfected reporter. The data (mean7SD) are an average of at least four experiments, each in duplicate.

of TSA was consistent with the lack of a parallel between the TSA-induced levels of c-jun in the NHEK extracts (Figure 4e) and its binding efficiency over AP1/S-50 oligonucleotide (Figure 4d). Notably, the S-100 and S-50 mutations, which did not interfere with c-jun transactivation in the control cells, were able to release the TSA-imposed block (Figure 5g, h). Simultaneous co-transfection of Sp1 and c-jun led to a dose-dependent synergistic activation of the wild-type promoter in both TSA-treated and control NHEK, indicating that increasing the amount of Sp1 was sufficient to release the 1132 Journal of Investigative Dermatology (2007), Volume 127

TSA-mediated block in c-jun transactivation. The synergism was evident with both S-100 (data not shown) and S-50 mutants and was lost only upon mutation of the AP1 recognition motif (Figure 5i). About 30–50% of the complexes A assembled over S-100 and S-50 (Figure 4b, c) contained Sp3 – a transcription factor of the Sp1 family often found to competitively antagonize Sp1 activation of natural promoters in various cell types, including epidermal keratinocytes (Jang and Steinert, 2002). However, as Figure 5j shows, in the absence of TSA (black bars), forced expression of Sp3 activated the wild-type reporter just like Sp1. Rather than compete with each other, Sp1 and Sp3 appeared to additively upregulate the activity. When Sp3 was substituted with a mutant lacking DNAbinding domain, the transactivation was comparable to that exerted by the respective amount of Sp1 alone. When the transfected NHEK were grown in 200 nM TSA (shaded bars), Sp3 was less efficient than Sp1 in activating the reporters. This result was consistent with the reported sensitivity of Sp3 activator function to acetylation within its inhibitory domain (Braun et al., 2001). Nevertheless, Sp3 was still able to counteract the silencing caused by TSA. Combined with the lower levels of nuclear Sp3 in the TSA-treated NHEK (Figure 4e), these results argue against Sp3 being a major factor whose action Sp1 counteracts in transmitting the TSA effect. Combined with the lack of response in levels and binding activity (Figures 4b–e), our preliminary data (not shown) indicate that significant involvement of RBP-Jk in the repression mechanism exposed by the inhibition of protein deacetylation is also unlikely. Deacetylase inhibition affects the recruitment of Sp1 and c-jun and decreases the acetylation activity over profilaggrin promoter in an S-motif dependent mode

We explored the extent to which TSA treatment affected the recruitment of Sp1 and c-jun over the promoter by incubating streptavidin magnetic beads coated with biotinylated profilaggrin fragment 147/ þ 9 with nuclear extracts from TSAtreated and control NHEK. Invariably, the input (52.6713.9), the bound (54.1715.4) and the unbound (52.7711.4) material from the TSA-treated NHEK contained about 50% less Sp1 protein, compared to the untreated controls (Figure 6a). Inclusion of recombinant Sp1 in the binding reactions increased the amount of the recruited Sp1 and eliminated the difference between the TSA-treated and untreated samples (Figure 6a, lanes 11, 12), whereas inclusion of recombinant c-jun could not apparently affect the recruitment of Sp1 (lanes 13, 14). The retention of Sp1 on the promoter-coupled beads was competed by S-50 (lanes 7, 8), S-100 (data not shown) and consensus Sp1 (lanes 19, 20) but was not sensitive to competition with either OctA (lanes 15, 16) or AP1 (lanes 17, 18) oligomers. Thus, under both conditions the recruitment of Sp1 over the promoter reflected the amount of Sp1 protein in the respective nuclear extracts and was not influenced by the AP1 interactions. Despite the induction of c-jun expression, the bound fractions from the TSA-treated NHEK retained 1.5 to 2-fold less c-jun compared to the untreated controls (compare lanes

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

a

– – – – S-50 rSp1 rc-jun OctA AP1 Sp1c Pretreatment R T R T R T R T R T R M i i b b ub ub b b ub ub b Tb Rb Tb Rb Tb Rb Tb Rb Tb

98

Sp1

64 50 36

c-jun

b

Acetylated histones (dpm/ g)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

c

150

– – – – S-50 Pretreatment Ri Ti Rb Tb Rub Tub Rb Tb

100

CBP 50

p300 12 3 4 5 6 7 8

0

Pretreatment

Bound –

–

–

RbTb RbTb Samples

Figure 6. TSA treatment reduces the acetylation activity over profilaggrin promoter. (a and c) Western blot analysis of bound (b), unbound (ub) and input (i) nuclear proteins from control (R) and TSA-treated (T) NHEK. The competitor oligonucleotides or recombinant proteins with which the extracts were preincubated are shown above the lanes. The arrowheads point to the detected Sp1 and c-jun (in a) and CBP and p300 (in c). The mobility of the protein markers (SeeBlue, Invitrogen) are shown in lane M. (b) Acetylation activity detected in the input, bound and unbound fractions derived from control (black and shaded bars) and TSA-treated (striped bars) NHEK. Where indicated, before capturing, the extracts were incubated with 100-fold molar excess of S-50 or AP1 oligonucleotides, or with 1.5 mg recombinant Sp1, or c-jun. The samples marked with asterisks represent the acetylation activity detected in the bound fractions after preincubation of 500 mg nuclear extract from control NHEK with the nuclear proteins from control (Rb) and TSA-treated (Tb) NHEK, which were captured over the promoter in the absence (black bars) or in the presence (shaded bars) of S-50 competitor. The data represent four independent acetylation experiments.

4 and 3). Preincubation of both extracts with excess of AP1 oligonucleotides prevented efficient recruitment of c-jun but had no effect on the recruitment of Sp1 (lanes 17, 18). In contrast, not only was more c-jun recruited upon preincubation with Sp1-binding oligonucleotides (lanes 7, 8, 19, and 20), but also there was no longer a difference between the TSA-treated and untreated samples. (Preincubation with control OctA oligomers had little effect on the recruitment of c-jun (lanes 15, 16)). Significantly, inclusion of recombinant c-jun in the binding reactions increased the amount deposited over the promoter by the control but not by the TSA þ extract (lanes 13, 14). Inclusion of recombinant Sp1, on the other hand, markedly increased c-jun binding in both types of extracts (lanes 11, 12). Thus, in contrast to Sp1, the recruitment of c-jun was not proportional to the amount of the protein in the nuclear extracts. Apparently, it was modulated by the relative amounts of Sp1 and another, negative regulator(s) recruited through S-50 (and S-100). Often, the activity of the DNA-bound transcription factors is modulated by co-regulatory complexes with intrinsic acetylase and/or deacetylase activities (see Rosenfeld et al., 2006). To assess whether the TSA treatment affected the acetylation activity over the promoter, we subjected the input, bound and unbound samples to an in vitro acetylation assay. The amount of the histones, acetylated under the designated experimental conditions, is shown in Figure 6b. The TSA treatment had no effect on the acetylation activity of

the input extracts per se and did not markedly change the acetylation activity that remained in the unbound fractions (compare black and striped bars). However, the acetylation activity deposited by the TSA þ extract over the promoter was almost 4-fold less than the untreated bound control. Although the overall acetylation activity in the bound fractions was lower, we observed a similar ratio between the TSA-untreated and treated bound samples when the magnetic beads were coated not with the entire proximal promoter region but only with a biotinylated S-50 oligonucleotide (Figure 6b, compare black and striped bars for (Bound) and (S-Bound). Our previous work has documented that c-jun is essential for recruitment and/or activation of acetylating enzymes at the profilaggrin promoter (Markova et al., 2006). Accordingly, preincubations of the nuclear extracts with AP1 oligomers decreased the overall acetylation activity, especially in the TSA-treated samples. However, increasing the amount of c-jun available for binding by preincubation of the nuclear extracts with recombinant c-jun, whereas increasing the acetylation activity in the control bound fraction by 2-fold, in fact enhanced the TSA effect. In contrast, preincubation of the extracts with excess of S-50 oligomers or with recombinant Sp1 eliminated the difference between the TSA-treated and untreated bound fractions and resulted in acetylation activities which were about 2-fold higher than the activity of the control bound fraction. There was no change in the acetylation activity in the respective unbound fractions (data not shown). www.jidonline.org 1133

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

TSA treatment enhances the recruitment of the transcriptional repressor LSD1

In search for the co-regulator(s) involved in the S-50mediated response to TSA, we resolved electrophoretically the nuclear proteins retained over S-50-coupled streptavidinmagnetic beads. The bands marked with dots clearly showed higher intensity upon TSA treatment (Figure 7a, compare lanes 1 and 2) and were sensitive to competition with S-50 oligomers (lanes 5, 6). These bands were digested in-gel with trypsin and subjected to mass spectroscopy. The tryptic peptide profile for each band was used to identify the respective proteins through online database search. P1 and P2 were identified as the polypeptides KIAA0166 and KIAA0185, respectively (Nagase et al., 1996). They encompass domains characteristic of transcription factors and their homologs have been found to act as transcriptional coregulators (Scaerou et al., 1999; Sweet et al., 2003). P3 and P5 were identified as nuclear myosin I and b-actin, respectively. Increased cellular levels of actin have been previously observed in NHEK in which protein deacetylation 1134 Journal of Investigative Dermatology (2007), Volume 127

a

– Bound + –

TSA

– Unbound + –

S-50 Bound + –

Pretreatment

250

P1 P2

148 P3 P4 98

64 50 P5 36 1

2

3

4

5

6 S-50

b TSA

Beads

UnInput Bound bound

– +

– + – + – + – +

Pretreatment

UnBound bound

– +

-LSD1 -H3 -di-Me-K4 1 2 3 4 5 6 7 8 9 10 11 12

c

300 % of Bound control

Because the suppression of profilaggrin transcription initiation by TSA was sensitive to cycloheximide, we hypothesized that TSA promotes the synthesis of a repressor(s), whose recruitment over the promoter via S-50 (and S-100) modifies in turn the interactions of Sp1 and c-jun and causes a decrease in the acetylation activity. If we were correct, we might expect that providing this factor(s) to the binding reactions of the TSA extract would lower the acetylation activity of the bound fraction. To test this, we incubated the promoter-coated beads with a nuclear extract from NHEK that had not been treated with TSA and to which we added either the TSA (Rb) or the TSA þ (Tb)-bound fractions. Whereas combining the control extract with the Rb fraction resulted in 3-fold more acetylated histones, the addition of the Tb fraction drastically decreased the acetylation activity (Figure 6b, the black bars marked with asterisks). When we repeated the experiment this time adding the bound fractions eluted after preincubation of the extracts with and TS50 ), the TS50 fraction was no S-50 oligomers (RS50 b b b longer able to diminish the acetylation activity of the bound fraction from the control NHEK (Figure 6b, shaded bars marked with asterisks). Under the experimental conditions, the observed acetylation activity reflected mostly the presence of acetylating enzymes, as class I/II deacetylases were inhibited by sodium butyrate and class III enzymes were inactive in the absence of the co-substrate NAD. Therefore, we assessed the protein fractions described above for the presence of three coregulators, namely CBP, p300, and pCAF, with proven acetyl transferase activity (Nakatani, 2001) and known to associate with and acetylate Sp1 in response to TSA (Huang et al., 2005). We found that both CBP and p300 were recruited over the promoter (Figure 6c), whereas pCAF was practically undetectable (data not shown). Without much affecting their levels within the cells, treatment with TSA reduced the retention over the promoter of both acetylases in a mode dependent on S-50.

200 100

0

I

B UB B UB

S-50 LSD1/H3

I

B UB B UB

S-50 Pretreatment di-Me-K4/H3

Figure 7. TSA treatment enhances the recruitment of repressor histone demethylase LSD1. (a) Coomassie R250-stained SDS protein gel showing the nuclear proteins from control () and TSA-treated ( þ ) NHEK in the bound and unbound fractions eluted from S-50-coated streptavidin-magnetic beads. The dots designate the proteins P1–P5, identified by mass spectrometry, whose retention was diminished upon preincubation of the extracts with excess S-50. The mobility of the marker proteins (SeeBlue, Invitrogen) is shown on the right. (b) The amount of LSD1 (top) and total histone H3 (middle) in the input, bound, and unbound samples described above, as well as in the samples eluted from streptavidin-magnetic beads, which have not been coated with DNA (Beads), were estimated by Western blotting. The same samples were subjected to a demethylation assay and the amount of histone H3 bearing di-methylated lysine K4 residues was estimated with anti-di-methyl-K4 specific antibody (bottom). (c) The intensities of LSD1 and di-methyl-K4 in lanes 3–12 were quantitated after scanning and normalized to the intensities of the respective histone H3 signals. The results are presented as a percentage of the LSD1/H3 and di-methyl-K4/H3 ratios, respectively, in the control bound fractions and are derived from three independent experiments.

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

has been inhibited by sodium butyrate and may be in part responsible for the cell shape and motility changes in these cultures (Staiano-Coico et al., 1989, 1990b) and in the epidermis (Figure 1h). The association of nuclear myosin and especially actin and actin-related proteins with chromatin remodeling, histone acetyl transferase and deacetylase complexes, and their importance in transcription by RNA polymerase II has now been firmly established (e.g., see Blessing et al., 2004). The role these proteins may play in regulating profilaggrin transcription remains to be investigated. P4 was identified as KIAA0106. It represents the recently characterized flavin adenine dinucleotide-dependent histone demethylase Lysine-Specific Demethylase (LSD1). LSD1 functions as a transcriptional co-repressor by demethylating lysine 4 (K4) of histone H3, where the methylated status is associated with active transcription (Shi et al., 2004). The protein has been found as a component of a number of corepressor complexes (Humphrey et al., 2001; Shi et al., 2005). Perhaps significantly, one of these complexes, immunoprecipitated with an antibody against HDAC1, also contained actin (Humphrey et al., 2001). We confirmed the increased retention of LSD1 over profilaggrin promoter in response to TSA by probing the input, bound, and unbound fractions captured over the promoter with an antibody specific for LSD1 (Figure 7b, top panel). The amount of histone H3 retained from the respective samples (middle panel) was used for normalization. Quantitation of the intensities of the scanned bands revealed that the extract from the TSA-treated NHEK deposited over the promoter about 3-fold more LSD1 (lanes 5, 6). The difference reflected the increase in LSD1 in TSA-treated keratinocytes, detected in the input (lanes 3, 4) and unbound (lanes 7, 8) samples. Competition with S-50 oligomers (lanes 9, 10) reduced the retention to only 10% of the control and eliminated the difference between the TSA-treated and untreated samples (Figure 7c). To assess the extent to which the presence of LSD1 correlated with the capacity of the protein fractions to demethylate lysine K4 of histone H3, bulk histones were subjected to in vitro demethylation. By using an antibody specifically recognizing di-methylated lysine K4 of histone H3 (Figure 7b, lower panel), we observed that the capacity for demethylation in the TSA-treated samples (lanes 4, 6, and 8) was 2 to 3-fold higher than the respective controls (lanes 3, 5, and 7). Preincubation of the extracts with S-50 did not alter this ratio in the unbound fractions (lanes 11, 12) but increased the amount of di-methyl-K4 remaining in the bound fractions and eliminated the TSA-induced difference (lanes 9, 10) (see also Figure 7c). DISCUSSION Herein, we have provided evidence that when applied at submicromolar concentrations the class I/II deacetylase inhibitor TSA causes morphological changes in all layers of foreskin epidermis. The decreased cellularity and the disorganization of the basal layer, the reduction in the number of spinous layers and the thickening of the stratum corneum are all in agreement with the previously reported

ability of HDACi to induce growth arrest (Brinkmann et al., 2001), to cause rearrangement of the cytoskeleton (StaianoCoico et al., 1989, 1990b) and to promote the formation of cornified envelopes (Schmidt et al., 1989) in cultured epidermal keratinocytes. To our knowledge, ours is the first study examining the effects of altered protein (de)acetylation on the homeostasis of normal human epidermal tissue. Recently Chung et al. (2004) reported that topical application of HDACi, including TSA, after g-irradiation of rat skin leads to thickening of the epidermis. There are several reasons that may explain the apparent contradiction between these observations and our findings. The effect may depend on the length of the treatment. It may reflect species specificity. The response to TSA may be influenced by the presence of hair follicles in the epidermis or by continuous blood supply through the dermis. The reported rat epidermal hyperplasticity may reflect a combined effect of irradiation and HDAC inhibition. Our data also illustrate the complexity of the acetylationrelated regulation in the skin cells. As evident from the chromatin immunoprecipitation experiments, in the dermal fibroblasts the epidermal genes are buried in transcriptionally silent hypoacetylated chromatin domains. In NHEK, these genes reside in hyperacetylated chromatin regions and are thus open for transcription. However, the transcriptional competence conferred to the epidermal genes by their preferential association with acetylated histones is clearly modulated by downstream acetylation-related events. Despite residing in the same chromosome locus, the EDC genes on 1q21 respond to TSA in a gene-specific rather than clusterspecific mode. Such gene-specific response differs from the response of the major histocompatibility class II and the adjacent histone gene clusters, for example, both of which are coordinately upregulated upon hyperacetylation in Blymphoid cells (Gialitakis et al., 2006). Apparently, the effect of TSA on the epidermal marker genes is not coordinated with respect to keratinocyte differentiation either. For the EDC genes, which are expressed in advanced stages of epidermal differentiation, superimposed on the transcriptional competence is the modulation of transcription initiation, which in the case of profilaggrin is suppressed, in the case of loricrin is not affected and in the case of involucrin is enhanced. Whether the normalization of the involucrin levels in the lower spinous layers of the TSA-treated organ cultures is due to inhibited transcription initiation in the less differentiated cells or to post-transcriptional events is at present unclear. Largely, TSA represses profilaggrin transcription initiation by inhibiting the promoter activity. The integrity of the Sp1/ Sp3-interacting S-100 and S-50 motifs proximal to the transcription initiation site is required for the keratinocytespecific TSA-mediated repression of the promoter to occur. As treatment of NHEK with TSA lowers the amounts of Sp1 and Sp3, TSA may be causing repression by simply diminishing the transactivation provided directly by these transcription factors. Several of our observations, however, argue against it being the decisive mechanism. First, the nuclear extracts derived from TSA-treated cells still contain significant amounts of Sp1 and Sp3. Second, the effect of TSA www.jidonline.org 1135

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

is sensitive to inhibition of protein synthesis and the nuclear extracts from TSA-treated keratinocytes are enriched in a protein(s), which, when added to extracts from untreated NHEK mimics the TSA effect in vitro. Third, despite the protein binding similarities between the four profilaggrin Smotifs (Markova et al., 2006), only S-100 and S-50, which share a marked homology to the regulatory elements controlling the response of several genes to the class I/II HDACi sodium butyrate (see Davie, 2003), represent profilaggrin TSA-response elements (Figure 3e). Fourth, treatment of NHEK with TSA not only diminishes the binding of Sp1(Sp3) at S-100 and S-50 but also enhances interactions, which result in the recruitment of fewer co-activator histone acetylases such as CBP and p300 and more co-repressors such as LSD1 (and possibly actin or actin-related proteins). The net activity of the Sp1(Sp3) complexes to promote or hinder transcription depends on the abundance, affinity, and duration of residence of diverse groups of both DNA-binding and non-binding proteins on the Sp1(Sp3) multimers (Mastrangelo et al., 1991; Davie, 2003). We propose that the Sp1(Sp3) occupancy at S-100 and S-50 may dictate the balance between the associated co-activators and corepressors. Higher levels of Sp1(Sp3) within the keratinocyte would ensure that the profilaggrin S-motifs are predominantly occupied by Sp1(Sp3) and the associated co-activator histone acetylases CBP and p300. This might contribute to the formation of a chromatin configuration that favors synergistic transactivation by c-jun. On the contrary, reduced levels of Sp1(Sp3) simultaneously with increased synthesis of the repressive demethylase LSD1 in response to TSA might favor the assembly of predominantly repressive complexes, which would replace the acetylase co-activators. The result, reflected in the appearance of new sites of DNase I hypersensitivity and reduced binding at the AP1 motif, might be remodeling of the chromatin into a configuration, which prevents efficient transactivation of the promoter by c-jun. To date little is known about the effect of protein deacetylase inhibition on the expression and activity of LSD1. Treatment with TSA did not affect the expression of one LSD1 target gene (SCN2A) in Rat-1 fibroblast cells (Lunyak et al., 2002) but resulted in derepression of the same gene in HeLa cells (Shi et al., 2005). Possibly, these differences reflect species and/or cell specificity. Recently; Shi et al. (2005) provided evidence that the LSD1 enzyme may favor hypoacetylated histone substrate. Consistent with this, the induction of major histocompatibility complex class II genes by TSA in B-lymphoid cell background involves an increase in H3 acetylation and lysine K4 methylation (Gialitakis et al. 2006). The fact that we observed lower acetylation activity over the profilaggrin promoter tempts us to speculate that in addition to modulating the acetylation of specific regulatory proteins (e.g., through diminished CBP/ p300 recruitment), the TSA-mediated reconfiguration may promote the demethylation of lysine K4 of histone H3 via local depletion of acetylated histones. This could also be achieved through activation of TSA-independent class III deacetylases, which would function upstream of LSD1. At least one such deacetylase – SIRT1, is recruited over 1136 Journal of Investigative Dermatology (2007), Volume 127

profilaggrin promoter in an S-motif-dependent mode (Markova et al., 2006). Significantly, AP1 and Sp1 interactions have been determined as critical for the activity of involucrin (Crish et al., 2006) and loricrin (Jang and Steinert, 2002) promoters, as well. The differential response of the three EDC genes and their promoters to TSA in the same cellular milieu suggests that when deacetylation is inhibited the AP1 and Sp1 transcription factors can assemble distinct combinations of co-regulators, depending on promoter- and/or DNA-binding site specificity thus providing an additional level of combinatorial control of gene expression. It is also conceivable that inhibition of deacetylation might promote specific posttranslational modifications of Sp1, AP1, and/or other regulatory proteins to which the three promoter regions are differentially sensitive. Although HDACi invariably cause accumulation of acetylated histones, only a relatively small number of genes (2–17%) have been shown so far to change their expression patterns (Van Lint et al., 1996; Lindemann et al., 2004). The data presented herein suggest that profilaggrin may be included in this short list. The disruption of normal epidermal morphology and the paradoxical downregulation of profilaggrin expression also suggest that inhibition of protein deacetylation may cause alterations in the epidermal function that could lead to skin morbidity and thus pose a challenge to the clinical applicability of deacetylase inhibitors as either radiation sensitizers or protectors. Although certainly not resolving the complexities of the HDACi-mediated regulation of the epidermal homeostasis, our studies provide a basis for further investigation, which may facilitate the design of clinical protocols to maximize the therapeutic effectiveness and limit the detrimental side effects of HDACi on the epidermis. MATERIALS AND METHODS The committees on research compliance for use of recombinant DNA and radioactive materials and the ethics committee of the State University of New York at Stony Brook approved all described experiments. Parental consent for use of the foreskin samples was granted. The study was conducted according to the Declaration of Helsinki Principles.

Plasmids and antibodies The wild-type and mutant CAT reporters have been described (Jang et al., 1996). The expression vectors for Sp1, Sp3, and mutant Sp3 were gifts from R. Tjan; the expression vector for c-jun was a gift from M. Karin. Recombinant Sp1 and c-jun proteins were purchased from Promega Corp. (Madison, WI). The antibodies against profilaggrin and loricrin were from Covance (Berkeley, CA); the anti-involucrin antibody has been described (Simon and Green, 1989). The AlexaFluor-conjugated secondary antibodies were from Molecular Probes (Eugene, OR). The antibodies against Sp1, Sp3, cjun, c-myc, p21, cyclin D and poly(ADP-ribose) polymerase 1, and the horseradish peroxidase-conjugated secondary antibodies were purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA); the anti-RBP-Jk antibody was from Chemicon (Temecula, CA). The antidi-methyl-K4-histone H3 and anti-acetyl-histone H3 antibodies were

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

from Upstate Group (Lake Placid, NY). The rabbit anti-LSD1 and pan-acetyl antibodies were a gift from V. Russanova.

Organ and cell cultures, transfections, and CAT reporter assays For the organ cultures, fresh neonatal foreskin specimens obtained from circumcisions were immediately placed in keratinocyte medium. The medium consists of a 3:1 (v/v) mixture of DMEM and Ham’s F-12 medium supplemented with adenine (1.8  104 M), penicillin (1000 U/ml), streptomycin (1 mg/ml), hydrocortisone (0.4 mg/ml), insulin (5 mg/ml), epidermal growth factor (10 ng/ml), cholera toxin (1.2  1010 M), CaCl2 (1.2 mM), and 5% fetal bovine serum (Randolph and Simon, 1993). The samples were incubated with continuous rotation in the absence or presence of 100 nM TSA for 48 hours at 371C. NHEK derived from neonatal foreskin were cultured in the same medium over 3T3 feeder layers as described (Randolph and Simon, 1993). HeLa cells were purchased from ATCC (Bethesda, MD) and grown in DMEM according to supplier’s recommendations. Where indicated, the cells were treated for the designated periods of time with 100 or 200 nM TSA, 10 mg/ml cycloheximide, or 5 mg/ml actinomycin D. Under the experimental conditions over 90% of the TSA-treated cells remained viable and non-apoptotic. The transfections and the measurement of the CAT activity 48–60-hour post-transfection were carried out as published by Jang et al. (1996).

Indirect immunofluorescence and Western blot analyses The indirect immunofluorescence was carried out on frozen sections of the organ cultures as previously published (Markova et al., 2006). Western blot analyses were performed with nitrocellulose replicas of proteins resolved on pre-cast gels (Invitrogen, Carlsbad, CA), developed with SuperSignal West Pico system (Pierce, Rockford, IL), detected on Kodak BioMax Light films and quantitated after scanning with Un-scan-it Gel software (Silk Scientific Co., Orem, UT). The loading and transfer were monitored by irreversible staining of the protein gels after transfer with GelCode Blue Stain Reagent (Pierce, Rockford, IL) and by reversible staining of the nitrocellulose filters with Ponceau S (Sigma, St Louis, MO).

RNA and nuclear run-on analyses Total RNA was extracted from heat separated foreskin epidermis (Steinert et al., 1985) or from NHEK with Trizol (Life Technologies, Gaithersburg, MD), resolved on methylmercury denaturing gels, transferred to nitrocellulose and hybridized to the indicated [a-32P]dCTP-labeled cDNA probes as described previously (Andreoli et al., 1997). Conventional RNase protection was carried out with total NHEK RNA and labeled antisense RNA probes using HybSpeed RPA kit from Ambion (Austin, TX). The protected RNA fragments were resolved on 6% denaturing polyacrylamide gels (Invitrogen, Carlsbad, CA) and detected on Kodak BioMax MS or MR films. The RT-PCR was performed on 0.5 mg total RNA with profilaggrinspecific primers (0.3 mM) or GAPDH-specific primers (0.06 mM) using OneStep RT-PCR kit (Qiagen Inc., Valencia, CA). To increase the sensitivity, 1 mCi [a-33P]dCTP was included in the reactions and the amplified region of both mRNAs was shorter than 200 nt. The labeled cDNA fragments were resolved on 6% denaturing DNA gels and detected on Kodak BioMax MR films. Nuclei were isolated from cultured NHEK as described by Dlugosz and Yuspa (1993). The in-vitro elongation of the initiated transcripts and their detection by

hybridization with specific cDNA fragments was performed as previously published (Jang et al., 1996). For the RNase H protection, the labeled nascent transcripts were hybridized to biotinylated antisense oligonucleotides corresponding to the first 64 bases of profilaggrin or GAPDH mRNA. The duplexes were captured on streptavidin-magnetic beads and digested with RNase H as described by Russanova et al. (1995). The protected RNA fragments were eluted from the beads, resolved, and detected as for the conventional RNase protection. The relative intensities of the signals obtained in the RNA analyses were quantitated after scanning of the films using Un-scan-it Gel software.

Chromatin immunoprecipitation The chromatin immunoprecipitation experiments were carried out as described by Markova et al. (2006). The in vivo crosslinked chromatin was incubated overnight with 50 mg rabbit anti panacetyl antibody or nonspecific IgGs, followed by an overnight incubation with magnetic beads-coupled anti-rabbit IgG. The bound material was eluted, the crosslinks were reversed and the immunoprecipitated DNA was amplified by PCR with primers specific for profilaggrin, involucrin, loricrin, and GAPDH genes. The PCR fragments were resolved electrophoretically and detected on Kodak BioMax MR films. The histone proteins were precipitated according to Lennox and Cohen (1989) and analyzed by Western blotting.

Preparation of nuclear extracts, bandshift, and DNase I footprinting analysis The NHEK nuclear extracts were prepared and assayed in bandshift experiments as described previously (Jang et al., 1996; Markova et al., 2006). The sequences of all oligonucleotides are available upon request. The DNase I footprinting of the profilaggrin promoter fragment 147/ þ 9 was carried out with SureTrack DNase I footprinting system (Pharmacia Biotech, Uppsala, Sweden) as suggested by the manufacturer.

Promoter-capture, acetyl transferase, and demethylation assays Biotinylated profilaggrin promoter fragment 147/ þ 9 or doublestranded S-50 oligonucleotides (2–4 pmoles) were captured over 200 mg M-280 streptavidin-magnetic beads (Dynal, Carlsbad, CA) in 10 mM Tris, pH 8.0, 1 mM EDTA. The unconjugated DNA (less than 90%) was removed with a magnetic particle concentrator. The DNAconjugated beads were blocked by 0.5% BSA in TGED buffer (20 mM HEPES, pH 7.9, 1 mM EDTA, 10% glycerol, 0.01% Triton X100, 10 mM sodium butyrate) and then incubated overnight at 41C with 500 mg NHEK nuclear extracts in the same buffer supplemented with 100 mM NaCl. The beads were extensively washed with TGED buffer and the specifically bound proteins were eluted in 30 ml TGED buffer supplemented with 1 M NaCl (Masumi et al., 1999). The acetylation activity of the input, unbound and bound samples was assessed with 1 mg unfractionated bulk histones (Sigma-Aldrich, St Louis, MO) and [H3]-coenzyme A, as described by Markova et al. (2006). The demethylation assay with the same samples was carried out with 30 mg unfractionated bulk histones in demethylation buffer, according to Shi et al. (2004), after desalting of the protein samples.

Identification of TSA-induced S-50-captured proteins Proteins captured over S-50-coated streptavidin-magnetic beads were separated on SDS gels and identified as described by Marekov www.jidonline.org 1137

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

and Steinert (2003). Mass spectra were acquired on a Voyager DEPro time-of-flight mass spectrometer. Proteins were identified by an online database search (http://129.85.19.192/profound_bin/WebProFound.exe). CONFLICT OF INTEREST

factors of the Sp1, CREB, AP1, and AP2 families. J Biol Chem 277:42268–79 Jang SI, Karaman-Jurukovska N, Morasso MI, Steinert PM, Markova NG (2000) Complex interactions between epidermal POU domain and activator protein 1 transcription factors regulate the expression of the profilaggrin gene in normal human epidermal keratinocytes. J Biol Chem 275:15295–304

The authors state no conflict of interest.

Johnstone RW, Licht JD (2003) Histone deacetylase inhibitors in cancer therapy: is transcription the primary target? Cancer Cell 4:13–8

ACKNOWLEDGMENTS

Jung M, Velena A, Chen B, Petukhov PA, Kozikowski AP, Dritschilo A (2005) Novel HDAC inhibitors with radiosensitizing properties. Radiat Res 163:488–93

We thank Gabriele Hatch and Alice Shih for growing large amounts of NHEK and Dr S.-I. Jang for his involvement in the construction of the CAT reporters. We are grateful to Dr R. Tjan for Sp1 and Sp3 expression vectors, and Dr M. Karin for c-jun expression vector, and to Dr V. Russanova for anti LSD1 and pan-acetyl antibodies.

REFERENCES Andreoli JM, Jang SI, Chung E, Coticchia CM, Steinert PM, Markova NG (1997) The expression of a novel, epithelium-specific ets transcription factor is restricted to the most differentiated layers in the epidermis. Nucleic Acids Res 25:4287–95 Blessing CA, Ugrinova GT, Goodson HV (2004) Actin and ARPs: action in the nucleus. Trends Cell Biol 14:435–42 Braun H, Koop R, Ertmer A, Nacht S, Suske G (2001) Transcription factor Sp3 is regulated by acetylation. Nucleic Acids Res 24:4994–5000 Brinkmann H, Dahler AL, Popa C, Serewko MM, Parsons PG, Gabrielli BG et al. (2001) Histone hyperacetylation induced by histone deacetylase inhibitors is not sufficient to cause growth inhibition in human dermal fibroblasts. J Biol Chem 276:22491–9 Chung YL, Wang AJ, Yao LF (2004) Antitumor histone deacetylase inhibitors suppress cutaneous radiation syndrome: Implications for increasing therapeutic gain in cancer radiotherapy. Mol Cancer Ther 3:317–25 Crish JF, Gopalakrishnan R, Bone F, Gilliam AC, Eckert RL (2006) The distal and proximal regulatory regions of the involucrin gene promoter have distinct functions and are required for in vivo involucrin expression. J Invest Dermatol 126:305–14 Davie JR (2003) Inhibition of histone deacetylase activity by butyrate. J Nutr 133:2485S–93S Dlugosz AA, Yuspa SH (1993) Coordinate changes in gene expression which mark the spinous to granular cell transition in epidermis are regulated by protein kinase C. J Cell Biol 120:217–25 Dokmanovic M, Marks PA (2005) Prospects: histone deacetylase inhibitors. J Cell Biochem 96:293–304 Garcia-Manero G, Issa JP (2005) Histone deacetylase inhibitors: a review of their clinical status as antineoplastic agents. Cancer Invest 23:635–42 Gialitakis M, Kretsovali A, Spilianakis C, Kravariti L, Mages J, Hoffmann R et al. (2006) Coordinated changes of histone modifications and HDAC mobilization regulate the induction of MHC class II genes by trichostatin A. Nucleic Acids Res 34:765–72 Hake SB, Xiao A, Allis CD (2004) Linking the epigenetic ‘‘language’’ of covalent histone modifications to cancer. Br J Cancer 90:761–9 Huang W, Zhao S, Ammanamanchi S, Brattain M, Venkatasubbarao K, Freeman JW (2005) Trichostatin A induces transforming growth factor beta type II receptor promoter activity and acetylation of Sp1 by recruitment of PCAF/p300 to a Sp1. NF-Y complex. J Biol Chem 280:10047–54 Humphrey GW, Wang Y, Russanova VR, Hirai T, Qin J, Nakatani Y et al. (2001) Stable histone deacetylase complexes distinguished by the presence of SANT domain proteins CoREST/kiaa0071 and Mta-L1. J Biol Chem 276:6817–24 Jang S-I, Steinert PM, Markova NG (1996) Activator protein 1 activity is involved in the regulation of the cell type-specific expression from the proximal promoter of the human profilaggrin gene. J Biol Chem 271:24105–14 Jang SI, Steinert PM (2002) Loricrin expression in cultured human keratinocytes is controlled by a complex interplay between transcription

1138 Journal of Investigative Dermatology (2007), Volume 127

Karagiannis TC, El-Osta A (2006) The paradox of histone deacetylase inhibitor-mediated modulation of cellular responses to radiation. Cell Cycle 5:288–95 Kaufmann SH, Desnoyers S, Ottaviano Y, Davidson NE, Poirier GG (1993) Specific proteolytic cleavage of poly(ADP-ribose) polymerase: an early marker of chemotherapy-induced apoptosis. Cancer Res 53: 3976–85 Kuo MH, Allis CD (1998) Roles of histone acetyltransferases and deacetylases in gene regulation. Bioessays 20:615–26 Lennox RW, Cohen LH (1989) Analysis of histone subtypes and their modified forms by polyacrylamide gel electrophoresis. Methods Enzymol 170: 532–49 Lindemann RK, Gabrielli B, Johnstone RW (2004) Histone-deacetylase inhibitors for the treatment of cancer. Cell Cycle 6:779–88 Lunyak VV, Burgess R, Prefontaine GG, Nelson C, Sze SH, Chenoweth J et al. (2002) Corepressor-dependent silencing of chromosomal regions encoding neuronal genes. Science 298:1747–52 Marekov LN, Steinert PM (2003) Charge derivatization by 4-sulfophenyl isothiocyanate enhances peptide sequencing by post-source decay matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. J Mass Spectrom 38:373–7 Markova NG, Pinkas-Sarafova A, Simon M (2006) A metabolic enzyme of the short-chain dehydrogenase/reductase superfamily may moonlight in the nucleus as a repressor of promoter activity. J Invest Dermatol 126: 2019–31 Marks PA, Richon VM, Miller T, Kelly WK (2004) Histone deacetylase inhibitors. Adv Cancer Res 91:137–68 Mastrangelo IA, Courey AJ, Wall JS, Jackson SP, Hough PV (1991) DNA looping and Sp1 multimer links: a mechanism for transcriptional synergism and enhancement. Proc Natl Acad Sci USA 88:5670–4 Masumi A, Wang I-M, Lefebre B, Yang X-J, Nakatani Y, Ozato K (1999) The histone acetylase pCAF is a phorbol-ester-inducible coactivator of the IRF family that confers enhanced interferon responsiveness. Mol Cell Biol 19:1810–20 Mischke D (1998) The complexity of gene families involved in epithelial differentiation. Keratin genes and the epidermal differentiation complex. Subcell Biochem 31:71–104 Nagase T, Seki N, Ishikawa K, Tanaka A, Nomura N (1996) Prediction of the coding sequences of unidentified human genes. V. The coding sequences of 40 new genes (KIAA0161-KIAA0200) deduced by analysis of cDNA clones from human cell line KG-1. DNA Res 3:17–24 Nakatani Y (2001) Histone acetylases – versatile players. Genes Cells 6: 79–86 Nirunsuksiri W, Zhang SH, Fleckman P (1998) Reduced stability and biallelic, coequal expression of profilaggrin mRNA in keratinocytes cultured from subjects with ichthyosis vulgaris. J Invest Dermatol 110:854–61 Palmer CN, Irvine AD, Terron-Kwiatkowski A, Zhao Y, Liao H, Lee SP et al. (2006) Common loss-of-function variants of the epidermal barrier protein filaggrin are a major predisposing factor for atopic dermatitis. Nat Genet 38:441–6 Peart MJ, Smyth GK, van Laar RK, Bowtell DD, Richon VM, Marks PA et al. (2005) Identification and functional significance of genes regulated by structurally different histone deacetylase inhibitors. Proc Natl Acad Sci USA 102:3697–702

NG Markova et al. Inhibition of Protein Deacetylation Represses Profilaggrin Transcription

Rosenfeld MG, Lunyak VV, Glass CK (2006) Sensors and signals: a coactivator/corepressor/epigenetic code for integrating signal-dependent programs of transcriptional response. Genes Dev 20:1405–28

Staiano-Coico L, Khandke L, Krane JF, Sharif S, Gottlieb AB, Krueger JG et al. (1990a) TGF-alpha and TGF-beta expression during sodium-N-butyrateinduced differentiation of human keratinocytes: evidence for subpopulation-specific up-regulation of TGF-beta mRNA in suprabasal cells. Exp Cell Res 191:286–91

Russanova VR, Driscoll CT, Howard BH (1995) Adenovirus type 2 preferentially stimulates polymerase III transcription of Alu elements by relieving repression: a potential role for chromatin. Mol Cell Biol 15:4282–90

Staiano-Coico L, Steinberg M, Higgins PJ (1990b) Epidermal cell-shape regulation and subpopulation kinetics during butyrate-induced terminal maturation of normal and SV40-transformed human keratinocytes: epithelial models of differentiation therapy. Int J Cancer 46:733–8

Saunders N, Dicker A, Popa C, Jones S, Dahler A (1999) Histone deacety lase inhibitors as potential anti-skin cancer agents. Cancer Res 59: 399–404

Steinert PM, Parry DA, Idler WW, Johnson LD, Steven AC, Roop DR (1985) Amino acid sequences of mouse and human epidermal type II keratins of Mr 67, 000 provide a systematic basis for the structural and functional diversity of the end domains of keratin intermediate filament subunits. J Biol Chem 260:7142–9

Randolph RK, Simon M (1993) Characterization of retinol metabolism in cultured human epidermal keratinocytes. J Biol Chem 268:9198–205

Scaerou F, Aguilera I, Saunders R, Kane N, Blottiere L, Karess R (1999) The rough deal protein is a new kinetochore component required for accurate chromosome segregation in Drosophila. J Cell Sci 112: 3757–68 Schmidt R, Cathelineau C, Cavey MT, Dionisius V, Michel S, Shroot B et al. (1989) Sodium butyrate selectively antagonizes the inhibitory effect of retinoids on cornified envelope formation in cultured human keratinocytes. J Cell Physiol 140:281–7 Shi Y, Lan F, Matson C, Mulligan P, Whetstine JR, Cole PA et al. (2004) Histone demethylation mediated by the nuclear amine oxidase homolog LSD1. Cell 119:941–53 Shi YJ, Matson C, Lan F, Iwase S, Baba T, Shi Y (2005) Regulation of LSD1 histone demethylase activity by its associated factors. Mol Cell 19: 857–64

Sweet T, Khalili K, Sawaya BE, Amini S (2003) Identification of a novel protein from glial cells based on its ability to interact with NF-kappaB subunits. J Cell Biochem 90:884–91 Ura K, Kurumizaka H, Dimitrov S, Almouzni G, Wolffe A (1997) Histone acetylation: influence on transcription, nucleosome mobility and positioning, and linker histone-dependent transcriptional repression. EMBO J 16:2096–107 Van Lint C, Emiliani S, Verdin E (1996) The expression of a small fraction of cellular genes is changed in response to histone hyperacetylation. Gene Expr 5:245–53 Wolffe AP, Kurumizaka H (1998) The nucleosome: a powerful regulator of transcription. Prog Nucl Acid Res Mol Biol 61:379–422

Simon M, Green H (1989) Involucrin in the epidermal cells of subprimates. J Invest Dermatol 92:721–4

Workman JL, Kingston RE (1998) Alteration of nucleosome structure as a mechanism of transcriptional regulation. Ann Rev Biochem 67:545–79

Smith F, Irvine AD, Terron-Kwiatkowski A, Sandilands A, Campbell LE, Zhao Y et al. (2006) Loss-of-function mutations in the gene encoding filaggrin cause ichthyosis vulgaris. Nat Genet 38:337–42

Yoshida M, Kijima M, Akita M, Beppu T (1990) Potent and specific inhibition of mammalian histone deacetylase both in vivo and in vitro by trichostatin A. J Biol Chem 265:17174–9

Spencer VA, Davie JR (1999) Role of covalent modifications of histones in regulating gene expression. Gene 240:1–12

Yoshida M, Matsuyama A, Komatsu Y, Nishino N (2003) From discovery to the coming generation of histone deacetylase inhibitors. Curr Med Chem 10:2351–8

Staiano-Coico L, Helm RE, McMahon CK, Pagan-Charry I, LaBruna A, Piraino V et al. (1989) Sodium-N-butyrate induces cytoskeletal rearrangements and formation of cornified envelopes in cultured adult human keratinocytes. Cell Tissue Kinet 22:361–75

Zhang Y, Jung M, Dritschilo A, Jung M (2004) Enhancement of radiation sensitivity of human squamous carcinoma cells by histone deacetylase inhibitors. Radiat Res 161:667–74

www.jidonline.org 1139