INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

INTERMEDIATE FILAMENT (IF) proteins are one of the three major cytoskeletal components of mammalian cells, which also include microfilaments and microtubules. In simple epithelia, as in hepatocytes and exocrine pancreatic acinar cells, keratin polypeptides 8 (K8; a type II keratin) and 18 (K18; a type I keratin) are the major IF proteins (21, 33). Biliary and pancreatic ductal cells also express, in addition to K8/18, other type II (K7) and type I (K19) keratins. Keratins form obligate noncovalent heteropolymers that consist of at least one type I and one type II keratin. In pancreatic acinar cells, K8/18 form a cytoplasmic (C) filamentous network and a distinct apicolateral (AL) band of filaments (8, 36, 43). One established keratin function in the skin, cornea, and liver is to maintain cellular integrity (13, 14, 30, 38, 41), but little if anything is known regarding keratin function in the pancreas. Keratins in simple epithelia may also regulate the availability of other abundant cellular proteins such as heat stress and 14-3-3 proteins (21).

It is well established that hepatocyte K8/18 play an important role in hepatocyte cell resilience and in protecting the liver from damage induced by a variety of toxins. This conclusion is based on 1) the extensive spontaneous liver hemorrhage noted in K8 null mice (6), 2) the significant hepatocyte fragility when keratins are disrupted in transgenic mouse livers on overexpression of dominant-negative K18 Arg⁸⁹Cys (called K18C mice; Ref. 17), and 3) the increased susceptibility of K18C (18, 44) and K8-null (/) mice (26, 44) to toxin-induced liver injury. In the case of K18C mice, the above findings are associated with keratin C filament disruption in the liver. However, despite similar C filament disruption in hepatocytes and pancreatic acinar cells of K18C mice, the K18C mouse pancreas is histologically normal (17). In addition, pancreata of K18C mice maintain their AL keratin filaments and do not suffer significant consequences in terms of pancreatic stimulated secretion or susceptibility to pancreatic injury that is induced via two established pancreatitis mouse models (43). Similarly, although K18C acinar cells had lower viability compared with cells isolated from wild-type (WT) mice (68 vs. 93%, respectively), this difference was significantly less dramatic than hepatocyte viability differences on isolation from K18C (22%) vs. WT (95%) mice (43). This raised the hypothesis that the remaining AL filaments of acinar cells, which form more prominent bundles in acini than in hepatocytes, may compensate functionally for the lack of cytoplasmic filaments. Alternatively, acinar cells may use nonkeratin proteins to serve keratin-like protective roles.

Here we tested the above hypotheses using K8/ (5, 6) and K18/ mice (29) that reportedly had no pancreatic histological abnormalities and, although not studied in detail, had absent or normal pancreatic keratin filaments, respectively. We challenged these mice, and their heterozygous and WT littermates, using the two well-established pancreatitis models of caerulein (an analog to CCK), which induces pancreatic secretion (16), and choline/methionine-deficient diet supplemented with 0.5% ethionine (CD diet) (25). We examined pancreatic histology and serology, keratin and F-actin organization, and keratin phosphorylation under basal conditions and on induction of pancreatitis. We also examined the viability and secretagogue response of acini.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals and primary antibodies. Heterozygous (+/) and homozygous (/) K8 or K18 mice and their WT littermates were expanded and genotyped as described (5, 29). The targeted / mutation of K8 causes midgestational lethality in homozygous mice, with bleeding into the embryonic liver in ~50% of the mice. The remaining K8/ mice have a normal life span but do suffer from colorectal hyperplasia, chronic colitis, mild serum transaminase elevation, and female sterility (in an FVB/N background strain; Ref. 5). The only reported pathological phenotype of the K18/ mice is spontaneous accumulation of Mallory bodies in the liver as the mice age (29). The K8 and K18 +/ mice do not have a phenotype (5, 29).

Pancreatitis models. All experiments included mice that were sex- and age matched. In the caerulein model, mice were injected intraperitoneally with 50 µg/kg caerulein (Sigma) or with carrier (0.9% NaCl) 7 times hourly and then killed 12 h after the first injection. The number of animals in each group was three for saline controls, six for caerulein (for +/ and / genotypes), and 12 for WT. Alternatively, mice were killed 1 h after a single injection, 1 h after six hourly injections, or after 48, 120, or 240 h of recovery after 7 hourly injections. All time points are given as hours after the first injection. Food was withheld from the mice for 16-18 h before the injections, but water was given ad libitum. In the CD diet model, two types of experiments, lethality and serology/histology, were performed. In both experiments, mice were fed (after a 12-h food starvation) a powdered CD diet (Teklad, Madison, WI) supplemented with 0.5% DL-ethionine (Sigma) or were fed normal mouse chow (Dean's Animal Feeds, Belmont, CA). For the serology/histology experiment, mice (n = 3 for control diet, n = 7 for CD diet per genotype group; age = 6-7 wk) were killed 60 h after initiating the diet, followed by serum and tissue isolation as detailed in Tissue processing, histology grading, enzyme assays, and fluorescence and electron microscopy. For the lethality experiment, mice (6 wk old, n = 10-26 per genotype) were fed normal or CD diet for 72 h and then were switched to normal diet and monitored for 7 additional days.

Tissue processing, histology grading, enzyme assays, and fluorescence and electron microscopy. Mice were killed by CO₂ inhalation, and blood was drawn by intracardiac puncture (0.3-1.0 ml) for subsequent serum collection. The pancreas was then excised and divided into three or more pieces for immediate fixation in 10% buffered formalin (pH 7.4; Columbia Diagnostics, Springfield, VA), embedding in optimum cutting temperature compound (Miles, Elkhart, IN), then freezing for cryosectioning or snap-freezing in liquid N₂ for subsequent biochemical analyses. Formalin-fixed tissues were processed for hematoxylin and eosin staining at Histo-tec Laboratory (Hayward, CA). Pancreatitis severity was assessed by assigning a relative score for vacuoles, cell death, edema, and inflammation on examination of the hematoxylin and eosin-stained sections (0 = none, 1 = mild, 2 = moderate, and 3 = severe) by a pathologist (S. A. Michie) who did not know the mouse genotypes from which the sections were obtained. Serum amylase and lipase were assayed using an Express Plus instrument (Bayer, Tarrytown, NY). Immunofluorescence staining and transmission electron microscopy (EM) were performed as described (43), except that nuclear staining was also done with YO-PRO-1 iodide or Toto-3 iodide (Molecular Probes, Eugene, OR; Refs. 19 and 24).

Protein preparation from whole tissue and by high-salt extraction. Pancreas tissue samples were homogenized with 600 µl of buffer [0.187 M Tris · HCl (pH 6.8), 3% SDS, and 5 mM EDTA] per 25 mg of tissue using a Teflon homogenizer. Samples were heated (98°C, 5 min) and sheared with a 22-gauge needle, then centrifuged (2 min, 14,000 rpm) to remove undissolved tissue. The protein content of the supernatant was determined using the BCA protein assay (Pierce, Rockford, IL), and samples (15 µg/lane) were analyzed by SDS-PAGE (22). Gels were stained with Coomassie blue or transferred to membranes for immunoblotting and visualization of Ab-bound species by enhanced chemiluminescence. High-salt extraction (HSE), an established method to obtain highly enriched keratin fractions (1), was done by homogenizing the pancreas with a Teflon homogenizer in 1% Triton X-100 and 5 mM EDTA in PBS (pH 7.4), containing 0.1 mM phenylmethylsulfonyl fluoride, 10 µM pepstatin, 10 µM leupeptin, and 25 µg/ml aprotinin (4°C, 2 min), followed by centrifugation (10 min, 14,000 rpm, 4°C). The pellet was further homogenized (4°C, 100 strokes) in high-salt buffer [10 mM Tris · HCl (pH 7.6), 140 mM NaCl, 1.5 M KCl, 5 mM EDTA, and 0.5% Triton X-100] containing the above protease inhibitors. Samples were incubated with rocking (4°C, 30 min) and centrifuged as above. Pellets were dissolved with sample buffer and then analyzed by SDS-PAGE.

Isolation and stimulated secretion of pancreatic acini. Pancreatic acini were isolated from mice by CLSPA-collagenase (Worthington, Lakewood, NJ) digestion as described (31, 43). The viability and stimulated secretion response were determined by measuring the extent of amylase release from isolated acini after incubation (10-60 min, 37°C) in the presence or absence of 0.1 nM [Tyr(SO₃H)²⁷]-CCK-8 amide (CCK8; Sigma). Samples were analyzed in duplicate, and experiments were repeated three times. Amylase release into the incubation buffer was measured using the Phadebas amylase test as described (45), and values were calculated as percentage of released/total amylase in the acini. The total amylase content was determined by solubilization of the acini in SDS.

Statistical analysis. Numerical data were compared using t-test or Wilcox test and the JMP 3.1 program (DataViz, Trumbull, CT). Data are given as means ± SE. For the lethality experiments, results were analyzed using Fisher's exact test.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Keratin expression in normal, K8/, and K18/ mice. We examined keratin expression in the mouse models by immunoblotting total pancreas extracts with Ab to K8, K18, or K19 and by visualizing the keratin profile obtained after HSE. As predicted, keratins were not detected in K8/ mice (Fig. 1) due to instability of type I keratins (K18 and K19) in the absence of type II (K8). In K18/ mice, K8 expression is significantly decreased, but not absent, likely because of relative stabilization by the low levels of K19 (Fig. 1). K19 levels, relative to K8, increase significantly in K18/ mice as determined by immunoblotting (Fig. 1A) and HSE (Fig. 1B). Assignment of the keratin bands in the HSE (Fig. 1B) was confirmed by immunoblotting using K8/18/19-specific Ab (not shown). The HSE profile (Fig. 1B) indicates that there is no obvious upregulation of other keratins. Hence, pancreata of WT mice express K8/18/19, but K8/ mice pancreata show no detectable keratins and K18/ mice pancreata express decreased K8 but increased K19 levels.

View larger version (54K): [in this window] [in a new window]   Fig. 1.   Keratin (K) expression in pancreata of mice with wild-type (WT), homozygous null (/), and heterozygous (+/) K8 and K18 genotypes. Pancreata were homogenized in SDS-containing sample buffer and analyzed by immunoblotting (A) with antibodies (Ab) to K8 (Troma I), K18 (anti-Endo B), and K19 (Troma III) B: extraction of keratins using a high-salt buffer as described in MATERIALS AND METHODS. Note the relative increase of K19 levels in K18/ mice (lane 5). Mr, relative molecular mass.

Keratin and F-actin distribution within the exocrine pancreas. Immunostaining of WT mouse pancreas revealed the unexpected result that mouse acinar cells manifest weak but distinct K19 AL staining (Fig. 2, d and g) in addition to the previously described ductal and centroacinar K19 localization (8). In contrast to mice, human acinar cells do not express K19, although K19 is present in ductal and centroacinar cells that appear more prominent in human compared with mouse pancreata (Fig. 3). Immunostaining of pancreata from K8/ and K18/ mice confirms the biochemical data shown in Fig. 1. For example, K8/ mice pancreata lack keratin filament staining in acinar cells (Fig. 2, b, e, and h), with scarce K19 staining in some ductal cells (Fig. 2e) that express small amounts of K7 (not shown), another type II keratin. Staining and blotting with anti-K7, K16, and K17 Ab showed no upregulation of these proteins in acinar cells of K8/ or K18/ mice (not shown). In contrast to WT acinar cells, K18 / acinar cells lack C keratins but show an intense AL K8 and K19 staining (Fig. 2f). F-actin localizes similarly in all mouse genotypes along the AL membranes (Fig. 2, g-i). Double staining of F-actin and K19 confirms the AL localization of K19 and shows that K19 is localized on the cytoplasmic side of the actin band [Fig. 2g, inset; note that actin (green) is more cortical than keratin (red) staining]. Staining with anti-mK18 Ab showed C and AL filaments in WT pancreata (i.e., similar staining to that in Fig. 2a; not shown) but absent staining in K8/ and K18/ mice (not shown).

View larger version (151K): [in this window] [in a new window]   Fig. 2.   Keratin and F-actin distribution in the exocrine pancreas of WT, K8/, and K18/ mice. K8 (a-c, red), K19 (d-i, red), and F-actin (g-i, green) were stained with Troma I, Troma III, and FITC-phalloidin, respectively. Nuclei (a-f, blue) were stained with Toto-3. Frozen pancreas sections were fixed with acetone or with 0.5% formaldehyde for keratin and F-actin staining, respectively. Double labeling of F-actin and K19 was done after 0.5% formaldehyde fixation. Panels a-c, d-f, and g-i represent double labeling as indicated. A, acinar cells; D, ductal or centroacinar cells; arrowheads in g, F-actin staining in endothelial cells. Bar = 10 µm. Note that the cytoplasmic filaments in WT mice in a are composed of K18 (K18 staining is identical to K8; not shown) and K8, whereas K19 is found preferentially in apicolateral (AL) filaments. K18/ mice pancreata lack keratin cytoplasmic (C) filaments but retain the AL filaments, at least in part, due to K19 partnering with K8. Double staining of F-actin and K19 confirms the AL localization of K19 and shows that K19 is localized on the cytoplasmic side of the actin band (g, inset).

View larger version (141K): [in this window] [in a new window]   Fig. 3.   K19 distribution in normal mouse and human exocrine pancreas. Human pancreatic biopsies were triple-stained with YO-PRO-1 (nuclear stain in C) and Ab to human (h) K8/18 (A) and hK19 (B). Centroacinar cell cytoplasm (positive for anti-K19) and nuclei are indicated by arrowheads (b and c), and an outline of 3 acini is highlighted with a dotted line for orientation. Bar = 10 µm. The schematic in D highlights the species difference in human and mouse K19 expression. Gray shading, K19; flat cells, ductal and centroacinar cells.

View larger version (188K): [in this window] [in a new window]   Fig. 4.   Ultrastructural assessment of WT, K8/, and K18/ mice pancreata. Thin sections were processed for conventional transmission electron microscopy (EM; a, c, and e) and for keratin immunogold EM labeling (b, d, and f) using anti-K8/18 rabbit Ab 8592 and goat anti-rabbit Ab conjugated to 10-nm gold particles. Normal rabbit Ab was used as a negative control and manifested no gold particle labeling (not shown; see also K8/ mouse staining in d, which lacked keratin filaments). K, keratin filament bundles (note intermediate filament decoration with gold particles in b and f); D, desmosome; L, lumen; Z, zymogen granule; Bar in a = 1 µm (for a, c, and e); bar in b = 0.5 µm (for b, d, and f).

Pancreatic histology, acini viability, and acini response to stimulated secretion. We compared the histology and isolated acini viability of K8/ and K18/ mice with WT mice pancreata. K8/ and K18/ mice pancreata appeared histologically normal in young and older (up to 7 mo old) mice (see, e.g., Fig. 5, a-c). There were very few focal areas of abnormal acinar cell organization in K8/ and K18/ mice of unclear significance (not shown). Isolated acini from all mouse genotypes excluded trypan blue similarly, but basal amylase secretion, a more accurate reflection of cell leakiness and viability, showed some differences. Hence basal amylase release at 30 min (37°C) was similar in WT and K18/ mice, whereas K8/ mice showed a lower viability (Table 1). However, acini from K18/ and K8/ mice released amylase in response to CCK8 similarly, thereby indicating that the acinar cells from these mice possess normal functional signaling mechanisms for stimulated secretion. In addition, more stringent collagenase digestion to isolate acinar cells did not give any significant differences in cell viability when comparing WT with K8/ or K18/ mice (not shown).

View larger version (117K): [in this window] [in a new window]   Fig. 5.   Effects of CD diet and caerulein on the pancreatic histological appearance in WT, heterozygous, K8/, and K18/ mice. Mice were allowed to eat the CD diet (CDD; d-f) ad libitum for 60 h, whereas control mice were fed standard mouse chow. Mice receiving the caerulein treatment (g-i) were injected with caerulein (50 µg/kg ip) 7 times at hourly intervals and then were killed 12 h after the first injection. Control mice for the caerulein model were injected with saline (a-c). Indistinguishable results in control mice of the CD diet and caerulein experiments were observed (not shown). Tissues were stained with hematoxylin and eosin. D, dead cells; E, edema; V, vacuoles.

View this table: [in this window] [in a new window]   Table 1.   Comparison of WT, K8/ and K18/ mouse pancreatic acini viability and response to stimulated secretion

Acute pancreatitis phenotypes in WT, K8/, and K18/ mice. To study the physiological effect of cytoplasmic or total acinar cell keratin filament absence, we subjected WT, K8/, and K18/ mice and their heterozygous littermates to two established acute pancreatitis models, namely the CD diet and caerulein injections. Feeding mice the CD diet for 72 h and then switching to normal chow for 7 days showed no significant lethality differences (Table 2). Histologically, the CD diet and caerulein injections induced acute pancreatitis in all mouse types, with vacuole formation and cell death in the CD diet model and coupled with edema and mild inflammation in the caerulein model (Fig. 5 and Table 3). However, the histological edema and inflammation scores of the caerulein-induced damage in K8/ mice were less dramatic than WT mice (Fig. 5 and Table 3), despite more prominent vacuolization for both null mouse types (Table 3). K8+/ mice also had a significantly lower inflammation score than WT mice (Table 3). No hemorrhage was noted in the caerulein-treated mice. The CD diet mice had similar increases in serum amylase and lipase (which are markers of pancreatic injury) as the WT mice, with a trend toward more prominent serum amylase elevations in K18/ and less prominent amylase rises in K8/ mice (Table 4). There was a statistically significant amylase difference in untreated K18/ versus WT sera (P = 0.01) and a similar trend in untreated K8/ sera, which likely reflect mild underlying hepatitis and/or necrosis, given that mouse hepatocytes also produce amylase but not lipase (28). Therefore, absence of pancreatic cytoplasmic or total keratins appears inconsequential in terms of susceptibility to pancreatic injury and, if anything, may even be somewhat protective. This is in sharp contrast to the marked propensity for liver injury in mice that have cytoplasmic disruption or total absence of keratin filaments (18, 26, 44).

View this table: [in this window] [in a new window]   Table 2.   Effect of the pancreatitis-inducing CD diet on WT, K8/ and K18/ mouse lethality

Keratin dynamics during acute pancreatitis. Caerulein injections induce a rapid (within 1 h) breakdown of WT and K18C pancreatic keratin and actin filaments (43). Both C and AL keratin filaments almost completely disappear in WT acinar cells (43), and a similar breakdown of K18/ AL filaments also occurs (not shown). We previously showed that acinar cells of K18C and WT mice begin to recover 7 h after stopping caerulein injections (i.e., the 12-h time point) and form bona fide cytoplasmic filaments (43). Although C filaments are basally absent in K18/ acinar cells, they surprisingly do form on recovery (Fig. 6, b and f). The newly formed keratin C filaments are composed of K8 and K19 in K18/ mice (Fig. 6, b and f) and of K8, K18, and K19 in WT mice (Fig. 6, a and e; K18 staining was confirmed by Troma II Ab staining, not shown). The K19 and K8 staining colocalizes, as determined by double labeling, and recovery from caerulein injury in K8/ mice is not associated with induction of any acinar keratin filaments (not shown). After 5 days, keratin C filaments in K18/ mice pancreata decrease, with a concomitant increase of AL filaments, whereas WT mice pancreata show widespread keratin C filaments (not shown). After 10 days (240-h time point), keratin organization returns in WT and K18/ mice to the preinjury state (Fig. 6, c, d, g, and h), with normal-appearing histological architecture (not shown) and normal F-actin organization (Fig. 6d, inset). Antibodies to K16, K17, or K7 did not manifest any altered staining in caerulein-treated mice (not shown).

View larger version (86K): [in this window] [in a new window]   Fig. 6.   Keratin filament reorganization on caerulein-induced pancreatitis in WT and K18/ mice. Mice were killed 48 or 240 h (10 days) after caerulein injections, and frozen sections of the isolated pancreata were fixed in acetone and then stained with anti-K8/18 (Ab 8592; a-d), anti-K19 (Ab Troma III; e-h), anti-K8 phospho(p)S436 (Ab 5B3; i-l), and FITC-phalloidin (inset in d). Initially, caerulein induces a dramatic acinar keratin filament reorganization with loss of AL + C filament staining in WT mice (43) and loss of AL filament staining in K18/ mice (not shown). Note that K18/ mice, which normally lack C filaments (see Fig. 2), express after 48 h de novo keratin C filaments that stain with K8 and K19 Ab (b and f, respectively). These C filaments disappear by 10 days in K18/ mice, with reversion to basal state staining (d and h). Also note keratin hyperphosphorylation during early recovery (i and j), which reverts to baseline (near-absent) levels on complete recovery (k and l).

Acinar cell keratin phosphorylation changes in caerulein-induced pancreatitis. We used phosphoepitope-specific keratin Ab to ask whether keratin phosphorylation is altered upon pancreatic injury in WT and K18/ mice. This is based on the strong correlation of keratin hyperphosphorylation with liver injury in vivo (21) and the protective effect it has from hepatotoxic injury (19). The cytoplasmic filaments that form on recovery from caerulein-induced injury contain K8 that is newly phosphorylated on S436 (Fig. 6, i and j) and S79 (not shown but very similar to i and j). The K8 pS436 filaments make up a distinct subpopulation of the overall keratin pool (not shown), and K8 pS436 becomes dephosphorylated by 240 h when acini are nearly fully recovered (Fig. 6, k and l; i.e., similar to the staining pattern before injury, with faint staining particularly of some ductal cells; not shown). Immunoblot analysis of whole pancreatic extracts from caerulein-treated WT and K18/ mice (Fig. 7) confirmed the immunofluorescence data shown in Fig. 6. Initially, keratins degrade in WT and K18/ mice pancreata starting a few hours after caerulein injection (e.g., see multiple K8 reactive bands; Fig. 7). This is followed by increased keratin synthesis, particularly K19 in WT and K18/ mice (Fig. 7; e.g., 48-h time point), during recovery.

View larger version (73K): [in this window] [in a new window]   Fig. 7.   Keratin expression during acute caerulein-induced pancreatitis in WT and K18/ mice. Mice were given hourly injections of 50 µg/kg ip caerulein and killed 5 (5 injections during the first 5 h), 48, 120, or 240 h (7 injections during the first 6 h each) after the first injection. Pancreata were then removed and homogenized, followed by immunoblotting of the homogenates (15 µg/lane) with antibodies to K8 (Troma I), K18 (anti-Endo B), K19 (Troma III), K8 pS436 (5B3), and K8 pS79 (LJ4). Note the disappearance (Coomassie-stained bands indicated by arrows) of some unknown proteins (likely due to degradation) and the appearance of other proteins (indicated by asterisks) during the early 5- and 48-h time points that revert to baseline by 240 h.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The major findings of this study are the following: 1) Absence of keratins is well tolerated in the pancreas, not only under basal conditions but more surprisingly in the context of two pancreatitis injury models (Fig. 5 and Tables 1-4). This finding is unexpected given that the same keratins (K8/18) in hepatocytes play a major cytoprotective role, in that their absence or disruption results in markedly fragile hepatocytes and in dramatically increased susceptibility to stress- or toxin-induced liver injury. This includes liver exposure to griseofulvin, acetaminophen, pentobarbital, partial hepatectomy, and microcystin (17, 18, 44). 2) Contrary to previous reports (Ref. 8 and references therein), K19 is expressed in mouse acinar cells, where it is preferentially localized in the AL compartment and is not present in C filaments that contain K8/18 (Figs. 1, 2, and 4). K19 acinar cell expression provides one potential mechanism for stabilization of keratin AL filaments in K18/ and in K18 dominant-negative mice that overexpress K18 Arg⁸⁹Cys. Mouse K19 acinar cell expression is not conserved in human acinar cells (Fig. 3). 3) Findings of this study, coupled with earlier findings (43), help define two keratin filament compartments in mouse pancreatic acinar cells (summarized in Fig. 8): a C compartment that contains K8/18 and an AL compartment that contains K8/18/19. Both compartments are absent in K8/ mice, whereas the AL compartment remains intact in conjunction with an absent or disrupted C compartment in K18/ or K18C mice, respectively. 4) Keratin filaments undergo a dramatic and reversible reorganization in response to pancreatic injury in vivo, in association with reversible keratin hyperphosphorylation (Figs. 6 and 7). Most notably, pancreatic K19 becomes overexpressed in K18/ mice, particularly after injury, and acquires the ability to form C filaments that reorganize into their basal state AL distribution on recovery.

View larger version (25K): [in this window] [in a new window]   Fig. 8.   Pancreatic acinar cell keratin filament phenotypes vary depending on genotype. A schematic representation of keratin filament organization highlights the normal situation of AL and C filaments in WT mouse acini, absent keratin filaments in K8/ acini, absent C but prominent AL bundles in K18/ acini (this study), and disrupted C filaments with prominent AL bundles in transgenic mice that overexpress a K18 dominant-negative mutation (Arg89Cys) (43). The equivalent Arg in epidermal keratins is commonly mutated in human epidermal skin diseases (13, 14, 30). The AL bundles likely remain intact in K18/ mice acinar cells due, at least in part, to the preferential localization of K19 in that compartment.

K19 is "tailless" but not idle. K19 is unique among the keratin family members since it has a very short (13 amino acid) COOH-terminal non--helical segment (called the "tail" in IF proteins) that is significantly longer in all other keratins (4, 11, 40). It is expressed in stratified and simple epithelia and in hair follicles and may serve as a skin stem cell marker (32, 34). In simple epithelia, K19 is expressed in several organs, including the intestine, biliary, and pancreatic ducts (21). In human pancreas, K19 is expressed in all epithelial cells in utero but is progressively lost from acinar cells and remains expressed only in ductal cells (8). Our data indicate that K19 is located not only in mouse pancreas ductal and centroacinar cells as previously described (8) but also in the AL compartment of acinar cells. This is supported by specific Troma III Ab staining, complete colocalization with F-actin in terms of ductal and acinar cell decoration (Fig. 2; Refs. 36, 37, and 43), and the retention of AL filaments in acinar cells of K18/ mice due to the presence of K19 (Figs. 1 and 4). The significance of the human and mouse species differences in K19 expression is unknown, and it remains to be determined if K19 is present in acinar cells of other species, such as rats and pigs.

Do identical keratins in pancreatic acinar cells and hepatocytes have different cytoprotective roles despite their similar hyperphosphorylation during injury? A surprising finding of this study is that keratin absence in acinar cells manifests a markedly different phenotype compared with hepatocytes. For example, hepatocyte viability on isolation and susceptibility to liver injury are significantly impacted by the lack of keratins or disruption of cytoplasmic filaments (21). In addition, baseline liver histology is slightly abnormal in transgenic mice that lack K8/18 (i.e., in K8/ mice; Ref. 26) or in transgenic mice with disrupted hepatocyte keratins (2, 17). In contrast, baseline K8/ mice pancreatic histology is normal (Fig. 5), and K8/ acini viability is only marginally abnormal and their response to CCK-stimulated secretion is intact (Table 1). In this context, the exocrine pancreas is unique, compared with the intestine and liver, given that lack of keratin filaments in all three organs results in an apparently normal pancreatic phenotype but abnormal liver and intestinal phenotypes (5, 6). Similarly, subjecting the pancreas of K18/ or K8/ mice to two well-established injury models still did not unmask any significant cytoprotective role of keratins, as contrasted with liver exposure to a variety of stresses. One potential explanation, which will require further testing using other pancreatic injury models, is that the type of injury we deployed does not involve K8/18 in the same fashion as the tested liver injury models. Although this scenario is possible, the response to injury in terms of keratin hyperphosphorylation was similar in the liver (21, 38, 42) and pancreas (Figs. 6 and 7), at least in the context of K8 S436 and S79 phosphorylation. This suggests that similar signaling mechanisms are involved in the injuries subjected to both cell types, albeit site-specific phosphorylation (aside from what we tested) or other signaling cascades may be different when comparing the pancreatitis models with the liver stress models. This raises the possibility that acinar cell K8/18 may not have the same cytoprotective function as hepatocyte K8/18 or that acinar cells have a keratin-like cytoprotective function that compensates for the disruption of cytoplasmic keratins in K18C mice (Ref. 43 and Fig. 8) or for the lack of cytoplasmic or total keratins (this study). The only keratin-related mouse model that shows clear damage in the exocrine pancreas is the transgenic mouse model that overexpresses human K8 (9). These mice develop dysplasia, loss of acinar cell architecture, acinar to ductal cell differentiation, inflammation, fibrosis, and fat formation. Since these mice do not have a liver-associated phenotype, it remains unclear at this stage whether the observed pancreas changes reflect overexpression levels and/or expression of a human (vs. mouse) protein. Ongoing experiments that entail overexpression of mouse K8 should help address this question.