Plastin 1 Binds to Keratin and Is Required for Terminal Web Assembly in the Intestinal Epithelium

Coprecipitation from Cell Lysates with Glutathione Transferase (GST)-Plastin 1

Intestinal epithelial cells were isolated as described previously (Weiser, 1973), with modifications. The lumen of freshly removed small intestine was washed with 154 mM NaCl und 1 mM dithiothreitol (DTT) and cut into three pieces. The pieces were end-ligated, and the tubes were filled with phosphate-buffered saline (PBS). The tubes were incubated in PBS for 15 min in a water bath at 37°C and then opened on one end and emptied. They were then refilled with PBS containing 1.5 mM EDTA and 0.5 mM DTT and incubated for 30 min at 37°C. The resulting suspensions were collected by gently pressing the intestine. The epithelial cells were sedimented by centrifugation (900 × g for 5 min), washed with PBS, and disrupted with a Dounce homogenizer.

Plastin 1 was expressed in Escherichia coli as GST fusion and coupled to glutathione-Sepharose beads (GE Healthcare, Chalfont St. Giles, Buckinghamshire, United Kingdom) by using standard methods. Intestinal epithelial cells were lysed in lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 20 mM MgCl₂, 5 mM EGTA, 1% Triton X-100, and 1% Nonidet P-40) supplemented with protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). The cell lysate was incubated with the beads for 2 h or overnight at 4°C. After several washings with lysis buffer (eventually supplemented with 300 mM NaCl), protein complexes were eluted with SDS sample buffer. Alternatively, to maintain actin in monomeric form during the pulldown, the cells were lysed in the presence of 10 μM latrunculin A (Sigma-Aldrich), and the drug was allowed to act for 20 min before incubation with the beads. A similar procedure was used to assay coprecipitation of recombinant keratins. The GST-plastin 1–coupled glutathione-Sepharose beads were incubated with 0.09 nmol of K8 and/or K19 (Fitzgerald Industries International, Concord, CT) in a slightly different buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 20 mM MgCl₂, 2 mM EDTA, 0.1% Triton, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 mM benzamidine, and protease inhibitor cocktail).

Generation of a Plastin 1 Knockout Mouse Strain

The mouse genomic DNA library RPCI21 from RZPD (Deutsches Resourcezentrum für Genomforschung, Berlin, Germany) was screened using Southern blot analysis with a Pls1 cDNA probe encompassing the first 430 base pairs of the coding region. Bacterial artificial chromosome (BAC) clone RPCIP711O22404Q2 was retrieved and used for further Southern blot mapping and subcloning. Targeting vector and the strategy for genotyping are depicted in Figure 2. The vector was linearized before electroporation into 129SV embryonic stem cells. Positive clones were identified by Southern blot analysis. The 5′ probe (565 base pairs) was amplified with following primers: forward (fwd), cttcatagtagggagtcatggtg; and reverse (rev), gatataattggaaagatctaatg. The 3′ probe (1201 base pairs) was amplified with fwd, ggctacagagggcaggcaaaacc and rev, gagacactttagtgacatgaacg. A Neo probe (493 base pairs) was amplified with fwd, agggatctcctgtcatctcaccttgctcctcc and rev, gaagaactcgtcaagaaggcgatagaaggcga. An internal probe (462 base pairs) was amplified from the 3′ arm of the targeting vector using a Pls1-specific primer (gttgttgtgaactcaggtgc) and a vector-specific primer. Three clones were selected for injection into C57BL6 blastocysts. Chimeric males were then crossed with C57BL6 females to obtain heterozygous plastin 1-deficient mice. The plastin 1-deficient strain was backcrossed into C57BL6 for six generations. The following primers were used for reverse transcription-polymerase chain reaction (RT-PCR) on RNA from whole gut: E2-5′, atataaagacttgaagtagcccttccagt and E4-3′, ctgagtacgaatgctgggtgccct. Unless otherwise indicated, 8- to 12-wk-old mice were used for experiments.

Tissue Preparation and Histological Analyses

Mice were killed by cervical dislocation. The small intestine was isolated and divided in three identical parts corresponding to the duodenum, jejunum, and ileum. The large intestine was isolated near the caecum and close to the rectum. Samples were washed with ice cold PBS. For preparation of whole gut lysates, pieces of intestine were sonicated in hypotonic NaHCO₃ buffer (1 mM NaHCO₃ and 1 mM PMSF, pH 7.5). After 30-min incubation on ice, SDS buffer was added, and the samples were boiled for 10 min at 95°C, centrifuged, and the supernatants were resolved by SDS-polyacrylamide gel electrophoresis. For paraffin sectioning, short pieces of intestine were fixed in 3% paraformaldehyde (PFA), Carnoy solution (60% ethanol, 30% chloroform, 10% acetic acid), methanol or Afa (5% acetic acid, 75% ethanol, 2% formaldehyde), dehydrated in ethanol and embedded in paraffin. For cryosections, tissues were fixed for 2 h in PFA, incubated overnight in a 30% glucose solution, embedded in optimal cutting temperature media, and frozen at −80°C. Sections (5 μm) were stained with hematoxylin and eosin (HE) or processed for immunohistochemistry using standard methods. All primary antibodies used in this study are listed in Supplemental Table S1. The appropriate Alexa 488- or 568-coupled secondary antibodies were used. Nuclei were stained with propidium iodide or 4,6-diamidino-2-phenylindole (DAPI). For detection of apoptotic cells a DeadEnd fluorometric terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay kit (Promega, Madison, WI) was used according to the manufacturer's instructions. Images were acquired with a TCS-SP confocal laser scanning microscope (Leica, Wetzlar, Germany) or with DMR or DM 6000B conventional epifluorescence microscopes (Leica) equipped with DC 350 FX (Leica), HV-C20A (Hitachi, Tokyo, Japan), or CoolSNAP HQ (Roper Scientific, Tucson, AZ) cameras.

Isolation of Brush Borders

Two procedures were used. In the first procedure (Matsudaira and Burgess, 1979; Ferrary et al., 1999), the whole small intestine was washed with ice cold PBS. The mucosa from longitudinally opened segments was scraped at 4°C with a square glass coverslip, diluted in 10 volumes/mg buffer A (10 mM imidazole, 5 mM EDTA, 1 mM EGTA, pH 7.4, and 0.2 mM DTT) supplemented with protease inhibitor cocktail, and stirred at 4°C for 1 h. Mechanical cellular disruption was then achieved with 5 strokes in a Dounce homogenizer. After centrifugation (1000 × g for 10 min at 4°C), the pellet was washed three times with buffer A, resuspended in buffer B (75 mM KCl, 5 mM MgCl₂, 1 mM EGTA, and 10 mM imidazole, pH 7.4) with protease inhibitor cocktail, and mixed with a sucrose solution in buffer B to a final 40% concentration. This sample was overlaid on the same volume of 65% sucrose solution in buffer B and centrifuged at 15,000 × g for 30 min at 4°C. The purified BBs were recovered at the interface of the 40%:65% sucrose gradient.

In the second procedure (adapted from Mooseker and Tilney, 1975; McConnell and Tyska, 2007), the intestine was cut into three pieces and washed with ice cold saline (150 mM NaCl and 2 mM imidazole, pH 7.2). Segments were opened longitudinally, cut into small pieces (∼2 cm), and stirred in a beaker containing ice-cold sucrose dissociation buffer (SDB; 200 mM sucrose, 12 mM EDTA, 19 mM KH₂PO₄, and 78 mM Na₂HPO₄) for 30 min in a cold room. The isolated enterocytes were collected by centrifugation (300 × g for 8 min) in SDB, resuspended in 1 volume of homogenization buffer (10 mM imidazole, pH 7.2, 4 mM EDTA, 1 mM EGTA, and 1 mM DTT, pH 7.2) supplemented with protease inhibitor cocktail, and homogenized in a blender with two 15-s bursts at high speed. The blender was rinsed with 1 volume of buffer B′ (buffer B from the first procedure supplemented with 1 mM DTT), which was pooled with the cell homogenate. The BBs were pelleted by centrifugation (1000 × g for 8 min), washed with buffer B′, and resuspended in the same buffer.

Brush-Border Contraction

A flow chamber was made of a polylysine-coated coverslip fixed with double-sided tape on a microscope slide. Two parallel stripes of tape were placed on opposite sides of the coverslip, leaving a 5-mm long slit on the two other sides. Isolated BBs kept on ice were injected into the slit and allowed to adhere before washing extensively with buffer B′. The flow chamber was then placed on a microscope stage, and the adherent isolated BBs were observed by differential interference contrast microscopy. Movies were recorded at one frame every 5 s for 15 min. At ∼3 min of recording, buffer B′ supplemented with 200 μM ATP was injected with a pipette into the flow chamber to replace the former solution free of ATP. The movies were acquired for the remaining time without any further manipulation. The acquisitions were made with a DM 6000B epifluorescence microscope (Leica) coupled to a CoolSNAP HQ charge-coupled device camera (Roper Scientific), and driven by MetaMorph software (MDS Analytical Technologies, Toronto, ON, Canada).

Electron Microscopy

For transmission electron microscopy (TEM), small pieces of tissue (∼1–2 mm) were cut open and fixed for 2 h at room temperature in 2.5% glutaraldehyde and 2% PFA in 80 mM cacodylate buffer, pH 7.2, and 0.05% CaCl₂. After washing with cacodylate buffer, the tissue was postfixed for 30 min at 4°C with 1% OsO₄ and 1.5% potassium ferrocyanide in cacodylate buffer and at room temperature for 1 h with 2% uranyl acetate in 40% ethanol. The samples were dehydrated in a series of graded ethanol solutions and embedded in Epon before ultrathin sectioning. For TEM on isolated BBs, samples were incubated in buffer B′ (see Isolation of Brush Borders) containing 15 mM MgCl₂ and let adhere for 1 h at 4°C on glass coverslips coated with polylysine. The coverslips were washed once before fixation in 0.1 M sodium phosphate buffer, pH 7.0, containing 2% glutaraldehyde and 2 mg/ml tannic acid for 1 h at 4°C. After washing in PBS, they were postfixed in 1% OsO₄ in 0.1 M phosphate buffer at pH 6.0 for 45 min and then with 0.5% uranyl acetate for 2 h. The samples were dehydrated in a series of graded ethanol solutions and then embedded in Epon before ultrathin sectioning. The observations were made with a Philips CM120 electron microscope (FEI, Eindhoven, The Netherlands). Image acquisition and measurements were made with the iTEM software (Olympus France SA, Rungis, France).

Transepithelial Resistance Measurements

Jejunum of fasted mice was isolated, washed in isotonic Ringer's solution (115 mM NaCl, 25 mM NaHCO₃, 1.2 mM MgCl₂, 1.2 mM CaCl₂, 2.4 mM K₂HPO₄, and 0.4 mM KH₂PO₄, pH 7.4), longitudinally opened, and mounted into a voltage-clamp system (surface, 0.15 cm²; Titis Business Corporation, Paris, France). Tissue was bathed on each side with isotonic Ringer's solution at 37°C. The resistance (ohms·per square centimeter) was measured after a 30-min equilibration period.

In Vivo Experiments

All studies involving animals were approved by the corresponding institutional boards for experimental animal welfare. To induce colonic epithelial injury (Mashimo et al., 1996) DSS (mol. wt. 40,000; ICN Biomedicals, Aurora, OH) was administered in the drinking water (2.5%, wt/vol) for 13 consecutive days, and body weight was recorded. When mice reached extremis they were removed from the experiment and scored as dead. The survival curves were analyzed by Kaplan–Meier transform of probability versus days of DSS treatment. A p value of 0.05 was considered as significant. Three to five animals of each genotype and sex were killed after 8 d, and the large intestine was processed for histological analysis (HE staining). Scoring of damage and inflammation induced by DSS treatment was done in a blinded manner. Six sections, each 100 μm apart, from two or three pieces of the colon were scored according to three parameters, severity of inflammation, extent of inflammation, and crypt damage (Dieleman et al., 1998).

For 5-bromo-2′-deoxyuridine (BrdU) in vivo labeling, 5-bromo-2′-deoxyuridine (100 mg/kg body weight) was administered by intraperitoneal injection. Mice were killed, and the intestines were processed for histological analysis. Incorporated BrdU was visualized with an anti-BrdU antibody and the Super Sensitive Detection kit (BioGenex, San Ramon, CA) after incubation with prewarmed 2 N HCl followed by 0.2% trypsin at 37°C.

Molecular Biology and Other Methods

Standard molecular biology methods were essentially as described by Sambrook et al. (2001). Genomic DNA was extracted from mouse tails by a modified method of Laird et al., 1991. RNA was isolated from intestinal mucosa with an RNeasy kit (QIAGEN, Hilden, Germany). Radioactive labeling was performed with a Random Primed DNA labeling kit (Roche Diagnostics, Mannheim, Germany). PCR fragments were cloned into the pGEM-T Easy vector system (Promega) and sequenced. DNA sequencing was done at the service laboratory of the Center for Molecular Medicine (Cologne, Germany) by using an automated sequencer (ABI 377 Prism; PerkinElmer Life and Analytical Sciences, Boston, MA). For biochemical determinations in serum, a Cobas Integra 700 Instrument (Roche Diagnostics) was used at the Department of Clinical Chemistry of the Cologne University Hospital. Blood glucose concentration was measured with test strips on an Accutrend GCT apparatus (Roche Diagnostics). Unless indicated otherwise, statistical analysis was performed using Student's t test, and data are shown as mean ± SD. A p value of 0.05 was considered significant.

Plastin 1 Accumulates in the Terminal Web and Interacts with Keratin 19

Although actin filaments at the rootlets are bundled by villin and plastin 1, villin-deficient mice do not show alterations of the rootlets and TW (Ferrary et al., 1999). Therefore, we analyzed the immunolocalization of these two proteins in more detail. Although villin is homogeneously distributed in the apical pole of the enterocytes (Figure 5A and Supplemental Figure S3A), plastin 1 is enriched in the TW compared with the microvillar part, as shown by costaining with tropomyosin, and, coincidentally, β-actin (AC-74 antibody) (Figure 1A). We have observed that antibodies raised against actin, besides displaying a punctate cytoplasmic labeling, stain the TW strongly but the MV only faintly (also see Figure 5A), probably because the densely packed microvillar actin is not as easily accessible to antibodies as in the rootlets. We thus hypothesized that apart from its role as an actin-bundling protein, plastin 1 may play a specific role in the TW and perhaps establish links to components of the apical cytoskeleton other than actin filaments. We performed pulldown experiments on epithelial cell lysates by using GST-fused recombinant plastin 1 (Figure 1B) under conditions of high ionic strength (300 mM NaCl) or in the presence of 10 μM latrunculin A, which provokes depolymerization of F-actin, to rule out any possible cosedimentation being bridged by actin filaments. We found that neither condition resulted in the association of actin to the GST-plastin 1–loaded beads. We then used a panel of antibodies recognizing BB components to probe the eluates in Western blots and found a positive signal only for K19 but not for K8. To check whether this interaction is direct, we used a similar approach with recombinant keratins. As for the pulldowns on cell lysates, purified K19 cosedimented with plastin 1–loaded beads (Figure 1C). Under the same conditions, K8 was observed to bind unspecifically to the beads or to GST, yielding inconclusive results for this keratin. We can thus conclude that plastin 1 and K19 interact directly. These experiments indicate that plastin 1 could establish a direct link to the keratin network, with which it partially colocalizes in the TW region (Figure 1D).

Plastin-deficient Mice Show No Overt Phenotype

To investigate the role of plastin 1, we designed a targeting vector in which exon 2 of the Pls1 locus was replaced by a neomycin resistance cassette (Figure 2). Immunoblot and immunostaining analysis of intestine samples ruled out any compensation of the plastin 1 deficiency by plastin 2 or 3 (Supplemental Figure S1). Plastin 1-deficient mice are fertile and display a normal reproductive rate. Sex distribution of pups, inheritance of the recombined allele, and life span were not significantly changed. No alterations in body size and weight are apparent in Pls1^−/− versus Pls1^+/+ mice (Supplemental Figure S2A), and no obvious differences in behavior have been observed. We determined the serum concentration of several metabolic parameters that depend on intestinal absorption, such as glucose, cholesterol, and triglycerides. Because plastin 1 is also expressed in the kidney, creatinine levels were also determined. None of these parameters changed in the Pls1^−/− mice (Supplemental Table S2). The lack of metabolic defects linked to potential intestinal absorption alterations in the Pls1^−/− mice could be because of compensation by increased food intake, decreased stool production, or increased length of the intestine (Supplemental Figure S2, B and C). However, these were not significantly different in the Pls1^−/− versus Pls1^+/+ animals. To investigate whether plastin 1 deficiency results in morphological alterations of the intestinal epithelium, we analyzed HE-stained sections of all intestinal segments. No obvious changes in the organization of villi or crypts were observed in the Pls1^−/− mice (Supplemental Figure S2D). The distribution of the major cell types (absorptive and goblet cells) of the intestinal epithelium was also unaltered (unpublished data).

In the Absence of Plastin 1, Microvilli Are Shorter and Lack Rootlets

To study the ultrastructure of the apical pole of enterocytes in Pls1^−/− animals, we performed TEM analysis of intestine sections. The enterocytes of Pls1^−/− animals present the characteristic apico-basal polarity and clearly develop apical MV (Figure 3A). At higher magnification, the striated pattern of actin filaments is observed in MV of both Pls1^+/+ and Pls1^−/− mice (Figure 3B). However, in Pls1^+/+ sections, the rootlets are easily detected as elongated electron dense structures, whereas they are never observable with this technique in Pls1^−/− samples (Figure 3B, insets). Moreover, Pls1^−/− MV are 20% shorter than their Pls1^+/+ counterparts (1.25 ± 0.26 μm, n = 86 in Pls1^+/+and 1.01 ± 0.16 μm, n = 194 in Pls1^−/−; p < 0.0001) and show a slight basal constriction (width at the base, 115 ± 14.5 nm, n = 142 in Pls1^+/+and 101.3 ± 11.9 nm, n = 164 in Pls1^−/−; p < 0.0001) (Figure 3B). Finally, in Pls1^+/+ mice, organelles are excluded from the very apical part of the cytoplasm corresponding to the TW region. In Pls1^−/− mice, this zone of organelle exclusion is strongly narrowed to ∼40% of its width in Pls1^+/+ samples (731.4 ± 206.5 nm, n = 17 in Pls1^+/+ and 304.9 ± 83.8 nm, n = 26 in Pls1^−/−; p < 0.0001) (Figure 3B). To analyze their degree of organization, we measured diameter, perimeter, area, and density of transversal sections through the MV, which did not differ significantly (unpublished data). The packing angle γ formed by three adjacent MV, which is expected to be of 60° for tightly packed perfect cylinders, was not modified in the absence of plastin 1 (65.9 ± 9.7°, n = 115 in Pls1^+/+; 62.0 ± 12.1°, n = 89 in Pls1^−/−) (Figure 2C). We were not able to accurately evaluate the number of actin filaments per microvillus in the Pls1^−/− sections because the quality and contrast of the dotted filament pattern is poor compared with the Pls1^+/+ counterparts (Figure 3C, sections at higher magnification). Presumably, this reflects a different organization of the actin bundles in the absence of plastin 1.

Brush Borders of Plastin 1-deficient Mice Are Highly Fragile

As indicated by the striated (Figure 3B) or punctate (Figure 3C) patterns observed, respectively, in longitudinal and transversal TEM sections, actin filaments are still present in Pls1^−/− MV. To better understand the organization of the actin cytoskeleton inside MV, we performed TEM on isolated BBs. This technique preserves the plasma membrane and the cytoskeleton of the apical pole and washes out the soluble components, greatly increasing the contrast and allowing the observation of individual actin filaments. The first isolation procedure we used (Matsudaira and Burgess, 1979; Ferrary et al., 1999) gave very good results for the Pls1^+/+ BBs, whereas only a very few damaged BBs were obtained from the Pls1^−/− mouse intestine (Figure 4A). We modified this method by using an F-actin–stabilizing buffer (Mooseker and Tilney, 1975; McConnell and Tyska, 2007). This greatly increased the stability of the Pls1^−/− BBs, which were nicely preserved and obtained with a similar yield as Pls1^+/+ BBs (Figure 4B). Phalloidin labeling confirmed the preservation of the actin cytoskeleton in both samples (Figure 4B, insets). This preparation was used for TEM, where F-actin filaments aligned in parallel array are clearly observed in both Pls1^−/− and Pls1^+/+ samples (Figure 4C). The myosin 1 bridges between actin filaments and the plasma membrane are also detected in both preparations (Figure 4C, insets, arrowheads). The constriction at the base of the MV is clearly seen in isolated Pls1^−/− BBs. The better contrast achieved by TEM on isolated BBs showed highly rudimentary rootlets composed of only few short filaments going beyond the level of the plasma membrane in Pls1^−/− BBs (Figure 4C, insets). Nevertheless, these clearly differ from the long and dense rootlets observed in the Pls1^+/+ situation, emphasizing the previous results (Figure 3).

The Composition of the Terminal Web, but Not of the Microvilli, Is Altered in the Plastin 1-deficient Mouse

We next investigated whether the amounts and localization of microvillar components were altered as a consequence of plastin deficiency (Supplemental Figure S3A). As expected, staining of F-actin with fluorescein isothiocyanate (FITC)-phalloidin revealed no alterations in cryosections of the small intestine of Pls1^−/− animals. Staining for villin and espin yielded the typical pattern of MV distribution both in Pls1^−/− and Pls1^+/+ animals. No alterations were also found in the distribution of ezrin and two class I myosins, myosin 1a and myosin 1e, that link the actin core bundles to the plasma membrane. In addition, Western blot analysis revealed no changes in the expression levels of any of these proteins in isolated BBs, except for a slight reduction in the amount of β-actin (Figure 5B). We conclude that the absence of plastin 1 does not result in compensatory overexpression or altered distribution of other actin-binding proteins in the enterocyte MV.

The loss of the actin rootlets observed in TEM pictures (Figures 3B and 4C) prompted us to examine the TW region in more detail. Spectrin and myosin II form thin filaments that connect adjacent rootlets. Myosin II additionally associates with the actin filaments of the adhesion belt (Glenney et al., 1982; Hirokawa et al., 1982). Tropomyosins show a polarized distribution (Percival et al., 2000), with isoforms 5a and 5b displaying a predominantly apical localization (Dalby-Payne et al., 2003). They stabilize the microfilaments of the actin core only at the rootlets and are not found within the MV (Bretscher and Weber, 1978). Loss of plastin 1 leads to major alterations in the localization of these proteins (Figure 5A). Stainings for α-II-spectrin and myosin IIa showed that neither of them localizes to the TW in Pls1^−/− mice, whereas their basolateral localization is not affected. Staining for tropomyosin revealed a strongly decreased signal in the TW region of the Pls1^−/− mice. In contrast to the unaltered F-actin staining observed with phalloidin (Supplemental Figure S3A), we found an altered localization pattern with three different antibodies against actin: two antibodies (AC-74 and a polyclonal antibody) specific for β-actin and the wide-spectrum anti-actin antibody AC-40. Although in the Pls1^−/− mouse the TW staining was lost, the fainter microvillar and the punctate cytoplasmic distributions were unaffected. The decreased actin staining in the TW of the Pls1^−/− sections is thus consistent with the lack of rootlets.

Because the TW region is connected to the underlying intermediate filament and microtubule networks and because plastin 1 is able to bind K19, we also examined those in intestinal sections. No alterations were found in the subcellular distribution of either α- or γ-tubulin or of the keratin network as assessed with a pancytokeratin antibody (Supplemental Figure S3B). Specific stainings for K8 and K19 did not reveal noticeable changes in their respective distribution patterns in the enterocytes of Pls1^−/− mice (Supplemental Figure S3C). Finally, Western blot analysis on lysates of whole epithelial cells did not reveal alterations in the amounts of any of the components mentioned above. However, consistent with the immunohistochemistry data, dramatically reduced levels of myosin IIa, α-II-spectrin, and tropomyosin 5a/5b were observed in isolated BBs from Pls1^−/− mice. Notably, although the amount of K8 was unaffected, the amount of K19 associated to BBs was substantially reduced in Pls1^−/− samples (Figure 5B).

In view of these strong TW defects, in particular of myosin II redistribution, we challenged the functionality of the actomyosin networks of the BBs by performing ATP stimulation assays. The stimulation of isolated BBs by ATP provokes the contraction of the TW region and the shedding of membrane at MV tips (McConnell and Tyska, 2007). The first effect is because of the contraction of the actin/myosin II belt linked to adherens junctions and the membrane shedding is the consequence of myosin 1a movement toward the plus ends of the actin bundles. These effects were, however, preserved in the Pls1^−/− BBs (data not shown), and the contraction could be inhibited in Pls1^−/− and Pls1^+/+ samples by the addition of the myosin II inhibitor blebbistatin. This striking result in view of the strong delocalization of myosin II can be explained by the fact that only a tiny pool of the TW myosin II is required for BB contraction (Keller et al., 1985). This is consistent with myosin II being present all along the lateral surface of the Pls1^−/− enterocytes up to the adhesion belt (Figure 5A).

The Transepithelial Resistance Is Reduced in Plastin 1-deficient Mice but the Structure of the Junctional Complexes Is Apparently Preserved

Because the TW is linked to the junctional complex that seals the intestinal epithelium, we investigated whether the alterations observed in the TW of the Pls1^−/− mice have any impact on the morphology of this complex. No noticeable morphological changes in the junctional complex were observed in transmission electron micrographs of Pls1^−/− sections (Figure 6A). Stainings for the transmembrane elements of the tight junctions claudin-1 and occludin (Figure 6, B and C) or for the zonula occludens 1 (ZO-1) protein that connects the tight junctions with the actin cytoskeleton (Figure 6D) did not reveal any alteration in the Pls1^−/− mouse. Similarly, no alterations were noticed in the localization of the transmembrane component of the adherens junctions E-cadherin (Figure 6D) or of the proteins that link the adherens junction to the actin belt, α-catenin (Figure 6E) and β-catenin (data not shown). The absence of noticeable morphological alterations does not rule out a difference in functionality. Indeed, measurements of the transepithelial resistance showed a significantly reduced resistance in the jejunum of Pls1^−/− (21.93 ± 2.03 ohm cm², n = 6) compared with Pls1^+/+ mice (29.90 ± 2.70 ohm cm², n = 7; p < 0.05) (Figure 6F).

Plastin 1-deficient Mice Are Sensitive to DSS-induced Colitis

The observations of the fragility of the isolated BBs and the decreased transepithelial resistance of Pls1^−/− mice prompted us to investigate the stability of the intestinal epithelium by using a model of colitis induced by DSS, an agent that causes mild epithelial injury when administered orally (Mashimo et al., 1996). In general, Pls1^−/− mice seemed more sensitive to the effects of DSS. Pls1^−/− animals of both sexes exhibited evidence of a colitis, with anal bleeding or prolapse occurring at day 5, whereas in Pls1^+/+ animals symptoms were apparent 2 d later. Pls1^−/− females were significantly more sensitive, with a 50% survival rate at 9 d (log rank, p < 0.0001) compared with 12 d for Pls1^+/+ females (Figure 7A). Survival rates of Pls1^−/− males were not significantly different from those of Pls1^+/+ males, despite their earlier onset of symptoms. Inspection of the intestine after 8 d of treatment revealed in all cases ulcerations and bleeding confined to the large intestine.

Histological examination of the colon showed lesions of variable severity, with inflammatory cell infiltrates in the mucosa, destruction of the crypt morphology (Figure 7B, empty arrowheads), loss of the epithelium (Figure 7B, arrows), and distension of the submucosa (Figure 7B, asterisks) in both strains. Abscesses in the submucosa (Figure 7B, e′) were more frequent in Pls1^−/− females than in the other populations. When severity and extension of inflammation were scored, we observed a significantly greater response in the Pls1^−/− than in the Pls1^+/+ females (15.8 ± 1.7 vs. 9.8 ± 1.6; p < 0.0428), whereas males behaved similarly (11.4 ± 1.5 in Pls1^−/− vs. 9.6 ± 2.2 in Pls1^+/+) (Figure 7C). Although we observed a trend toward higher scores of crypt damage in Pls1^−/− mice of both sexes, the differences were not statistically significant.

Increased Epithelium Renewal in the Intestine of Plastin 1-deficient Mice

The increased fragility of the BBs and the higher susceptibility to induced colitis are in contrast to the mild phenotype displayed by the Pls1^−/− mice. The question thus arises whether an increased epithelium renewal rate compensates for the plastin 1 deficiency. Newly generated cells in the crypts migrate toward the top of the villi within 2 to 7 d and then they undergo apoptosis and are finally shed into the lumen (Crosnier et al., 2006). We used a TUNEL assay to examine whether the loss of plastin 1 results in an increased apoptosis rate in the intestinal epithelium. Although occasional apoptotic enterocytes were observed in the Pls1^+/+ mice, the number of TUNEL-positive epithelial cells was increased fourfold in the Pls1^−/− mice (p < 0.0001) (Figure 8, A and B). To analyze whether the increased cell loss is balanced by an increased proliferation rate, we stained the proliferation zone in the crypts for Ki67, a nuclear protein expressed during all active phases of the cell cycle, and for phosphohistone 3, a marker of cells in M phase of the cell cycle. These stainings revealed no alterations in the position and width of the proliferating zone in the Pls1^−/− mouse (Supplemental Figure S4). A potential increase in cell renewal in Pls1^−/− mice was analyzed in a BrdU incorporation assay (Figure 8, C and D). Two hours after BrdU administration, proliferating cells within the crypts showed a pattern similar to Ki67 staining in both Pls1^+/+ and Pls1^−/− mice. However, after 24 h, epithelial cells of Pls1^−/− mice occupied 52.8% of the villus length compared with 36.8% in Pls1^+/+ (p < 0.0001). Similarly, whereas BrdU-positive cells had nearly reached the villus tip of Pls1^−/− mice (91.2% migration distance) at 48 h, the distance migrated by Pls1^+/+ cells was only 79.2% (p < 0.0001). The increased cell migration had no effect on epithelial cell differentiation, as shown by staining for I-Fabp (Supplemental Figure S4).

Plastin 1 Is Important for Brush Border Integrity

Unlike villin, plastin 1 gives rise to actin bundles in vitro that closely resemble those of the MV core bundle (Glenney et al., 1981; Matsudaira et al., 1983). It is therefore surprising that in Pls1^−/− mice MV assemble with apparently normal actin cores, although with the remarkable abnormalities discussed below. Plastin 1 deficiency is not compensated by persistent expression of either of the two other plastins that can be detected in intestinal cells and in sensory cells of the inner ear during assembly of MV and stereocilia in embryonic stages (Tilney et al., 1992; Daudet and Lebart, 2002). It has been proposed that villin and plastin 3 are responsible for the formation of rudimentary MV, but these reach their normal length when expression of plastin 1 starts and the protein localizes apically (Ezzell et al., 1989; Chafel et al., 1995). The existence of MV both in villin and in plastin 1 knockout mice suggests that the apical localization of plastin 3 during embryonic development could be sufficient for the formation of rudimentary MV.

However, MV also form in the adult epithelium, where plastin 3 is no longer expressed, which highlights a likely redundancy among the three actin-bundling proteins (plastin 1, villin, and espin) of the MV to warrant their morphogenesis. Villin and espin, albeit not overexpressed, must thus provide sufficient bundling activity, although TEM sections across MV suggest that the organization of the bundles is altered in the absence of plastin 1. In vitro studies assigned an essential role to villin for MV assembly (Friederich et al., 1990; Friederich et al., 1992; Costa de Beauregard et al., 1995); however, no alterations in the morphology and composition of the MV were found in villin deficient mice (Pinson et al., 1998; Ferrary et al., 1999; Athman et al., 2002). Espin-deficient mice have no apparent morphological alterations in the intestinal epithelial cells (Revenu and Robine, unpublished), presumably reflecting its low abundance relative to plastin 1 and villin.

This leaves plastin 1 as the only actin-bundling protein that by itself plays an important role for the proper formation and maintenance of intestinal MV. Our data are in agreement with the proposal that plastin 1 stabilizes and elongates existing MV but is not required for their de novo assembly (Arpin et al., 1994). Plastin 1 accumulates in the TW region of the BB and constitutes the first BB protein for which a deficiency results in dramatic alterations in that region, as demonstrated by the absence or reduced amount of typical components such myosin II, spectrin, and tropomyosin. This points at a general alteration that presumably results from the missing or rudimentary rootlets observed in TEM images, because these components stabilize or interconnect the rootlets, or link them to the adhesion complex (Figure 9).

Plastin 1, a Likely Linker between the Microvillar Actin and the Apical Keratin Network and a Key Factor Stabilizing the Microvillar Rootlet

That absence of plastin 1 results in clear alterations of the TW indicates that this protein fulfills a specific function that cannot be compensated by other BB proteins. The biochemical interaction of plastin 1 with the keratin network might well be at the base of the ultrastructural and functional defects observed in the BB of the Pls1^−/− mice. Early TEM studies on intestinal epithelial cells showed that the intermediate filament network is linked to the microfilament network (Hirokawa et al., 1982, 1983). We hypothesize that plastin 1 is responsible for anchoring the rootlets to the underlying keratin network through an interaction with K19, perhaps contributing to the mechanical strength of the apical pole. Additionally, plastin 1 could have a stabilizing function for the filaments solely because of its bundling activity, as has been proposed for all bundlers from biological and biophysical studies (Zigmond et al., 1992; Tilney et al., 2003) because depolymerization beyond a cross-link requires the unbinding of the cross-linking protein (Prost et al., 2007). Moreover, plastin 1 could contribute to stabilizing the rootlets by preventing the action of the severing and depolymerizing protein cofilin. Such mechanism has been described for plastin 3 (Giganti et al., 2005) and for yeast fimbrin (Nakano et al., 2001) and is therefore likely to work for other plastins. It is also possible that absence of plastin 1 results in the destabilization of the rootlets indirectly through the mislocalization of tropomyosin, a protein that stabilizes actin filaments directly by slowing depolymerization from the pointed end (Broschat et al., 1989). The absence of plastin 1 would result in detached and unprotected rootlets, leading to the faster depolymerization of actin filaments up to the region where the actin core is attached to the plasma membrane. Consequently, the underlying keratin network becomes displaced apically, leading to a narrowed organelle free zone underneath the MV (Figure 9).

That absence of plastin 1 causes no obvious defects in the keratin network, at least with the resolution that can be achieved with the optical microscope, is not surprising. Studies in Caco2 cells have shown that the polarity of the keratin network establishes 2 d before that of the microtubules and actin filaments and the formation of the BB (Wald et al., 2005), therefore independently of plastin 1. The reverse, however, may not hold true. Although K19 knockout mice do not present any overt phenotype, presumably because the defects are compensated by other class I keratins, no studies have been published addressing specifically the intestinal epithelium (Tamai et al., 2000). Interestingly, silencing of the K19 gene in Caco2 cells results in alterations of the apical actin network that are apparently not compensated by K18 or another keratin (Salas et al., 1997). This sheds light on a particular interaction between K19 and the actin network that could be linked with the interaction that we demonstrate between plastin 1 and K19. The interaction of calponin homology (CH) domain-containing proteins with intermediate filaments is emerging as a common theme, enlightening a role for this CH family in the establishment of boundaries between intermediate filaments and actin (Chang and Goldman, 2004). In particular, plastin 2 forms a complex with vimentin tetramers that is important for the assembly of the vimentin cytoskeleton during cell adhesion. The first CH domain of plastin 2 has been identified as the domain participating in the interaction with vimentin (Correia et al., 1999; Delanote et al., 2005). For dystrophin, another CH domain protein, an interaction of the actin-binding domain with K19 has been reported in vitro (Stone et al., 2005).

Loss of Plastin 1 Results in Fragility of the Intestinal Epithelium

The profound alterations of the TW must explain the sensitivity of the BB of Pls1^−/− mice to the mechanical stress applied during biochemical isolation. A decreased stability of the BB has been reported in myosin 1a-deficient mice and in myosin 1B-deficient Drosophila melanogaster that has been related to a decreased calcium-buffering activity (Tyska et al., 2005; Hegan et al., 2007). The loss of this buffering activity would result in increased actin filament severing by villin. Although plastin 1 is capable of binding calcium ions through its N-terminal EF hands (Lin et al., 1994), it is not as abundant as class I myosins; therefore, no major influence on the calcium concentration is to be expected in the BBs of the Pls1^−/− mouse. Carbachol treatment and fasting/refeeding, two situations that provoke elevated intracellular calcium levels, have also revealed that the plastin 1 deficiency does not affect the severing activity of villin (data not shown).

A major consequence of the altered structural integrity of the BB in the Pls1^−/− mice is the decreased transepithelial resistance of the intestine despite apparently normal structure and protein composition of the adhesion complex. Pls1^−/− mice are also more sensitive to DSS-induced colitis, in agreement with the increased fragility of the apical pole. Higher sensitivity to DSS treatment occurs in villin-deficient mice that recent studies attribute to the loss of the antiapoptotic activity of this protein leading to an imbalance between cell renewal and apoptosis (Ferrary et al., 1999; Wang et al., 2008). Although it is not clear why Pls1^−/− female mice die significantly earlier than the males upon DSS treatment, a possible explanation may be the different cell renewal rates between males and females. Indeed, BALB/c females are more resistant to parasitic infections because of a higher cell renewal rate allowing a faster expulsion of the parasites (Bancroft et al., 2000; Cliffe et al., 2005). We speculate that in the females, contrary to the males, the cell renewal rate cannot increase further under stress situations, making them more sensitive. An alternative explanation could be the well documented differences in the immunological response between males and females (Schuurs and Verheul, 1990), which would fit well with the increased scores of inflammatory response observed in the Pls1^−/− females.

Because Pls1^−/− mice do not display an overt phenotype, the increased fragility of the BB and the increased rate of cell death could be balanced by the elevated rate of cell migration. A possible explanation for the lack of evidence of an increased number of proliferating cells in the crypts of the Pls1^−/− mice is a shortening of the cell cycle of the stem cell population as reported for example after gamma or beta irradiation of the intestine (Tsubouchi and Potten, 1985).

In summary, the absence of plastin 1 strongly affects the cohesion of the apical pole probably by the loss of an anchorage between the different cytoskeletal networks. This destabilization must be responsible for the enhanced fragility of the whole epithelium when challenged. Plastin 1 thus is a major player in the integrity of the intestinal barrier.