Abstract
Using monolayers of intestinal Caco-2 cells, we discovered that the θ isoform of protein kinase C (PKC), a member of the “novel” subfamily of PKC isoforms, is required for monolayer barrier function. However, the mechanisms underlying this novel effect remain largely unknown. Here, we sought to determine whether the mechanism by which PKC-θ disrupts monolayer permeability and dynamics in intestinal epithelium involves PKC-θ-induced alterations in claudin isotypes. We used cell clones that we recently developed, clones that were transfected with varying levels of plasmid to either stably suppress endogenous PKC-θ activity (antisense, dominant-negative constructs) or to ectopically express PKC-θ activity (sense constructs). We then determined barrier function, claudin isotype integrity, PKC-θ subcellular activity, claudin isotype subcellular pools, and claudin phosphorylation. Antisense transfection to underexpress the PKC-θ led to monolayer instability as shown by reduced 1) endogenous PKC-θ activity, 2) claudin isotypes in the membrane and cytoskeletal pools (↓claud-1, ↓claud-4 assembly), 3) claudin isotype phosphorylation (↓ phospho-serine, ↓ phospho-threonine), 4) architectural stability of the claudin-1 and claudin-4 rings, and 5) monolayer barrier function. In these antisense clones, PKC-θ activity was also substantially reduced in the membrane and cytoskeletal cell fractions. In wild-type (WT) cells, PKC-θ (82 kDa) was both constitutively active and coassociated with claudin-1 (22 kDa) and claudin-4 (25 kDa), forming endogenous PKC-θ/claudin complexes. In a second series of studies, dominant-negative inhibition of the endogenous PKC-θ caused similar destabilizing effects on monolayer barrier dynamics, including claudin-1 and -4 hypophosphorylation, disassembly, and architectural instability as well as monolayer disruption. In a third series of studies, sense overexpression of the PKC-θ caused not only a mostly cytosolic distribution of this isoform (i.e., <12% in the membrane + cytoskeletal fractions, indicating PKC-θ inactivity) but also led to disruption of claudin assembly and barrier function of the monolayer. The conclusions of this study are that PKC-θ activity is required for normal claudin assembly and the integrity of the intestinal epithelial barrier. These effects of PKC-θ are mediated at the molecular level by changes in phosphorylation, membrane assembly, and/or organization of the subunit components of two barrier function proteins: claudin-1 and claudin-4 isotypes. The ability of PKC-θ to alter the dynamics of permeability protein claudins is a new function not previously ascribed to the novel subfamily of PKC isoforms.
The epithelium of the intestinal mucosa is the largest interface between the body and the external environment. An important characteristic of this interface is its ability to maintain a highly selective permeability barrier that protects the internal milieu from hostile factors in the lumenal environment. Barrier permeability, which in general is maintained by epithelial tight-junctional proteins, permits the absorption from the lumen of needed water and electrolytes but prevents the passage of inflammatory and infectious agents (e.g., pathogenic bacteria) into the mucosa and systemic circulation. Loss of barrier function (i.e., increased permeability) is thought to lead to the penetration of bacterial derivatives and other harmful immunoreactive antigens into the mucosa and to cause the initiation or perpetuation of inflammatory processes and tissue injury (Hollander, 1992, 1998; Hermiston and Gordon, 1995; Keshavarzian et al., 1999; Banan et al., 2000a). Indeed, disruption of barrier function (“leaky gut”) has been implicated in the pathogenesis of several gastrointestinal and systemic disorders, including inflammatory bowel disease (IBD) (Hollander, 1992, 1998; Keshavarzian et al., 1992, 1999, 2003; Peeters et al., 1994; Hermiston and Gordon, 1995; Soderholm et al., 1999). Not surprisingly, one of the key difficulties in managing IBD patients stems from our incomplete understanding of the processes regulating intestinal barrier function.
A key discovery in recent years in IBD pathophysiology was the understanding that a leaky intestinal barrier, the so-called hyperpermeable gut, can lead to intestinal inflammation and tissue injury and that maintaining a normal mucosal barrier function is required for intestinal health. For example, transgenic rodents with a hyperpermeable-intestinal-barrier exhibit intestinal mucosal inflammation (Hermiston and Gordon, 1995). Similarly, loss of intestinal barrier function induced by the injection of bacterial derivatives into the mucosa of rodents elicits IBD-like conditions (Yamada et al., 1993). Accordingly, investigating the molecular events underlying the alterations of gut barrier function is of fundamental biological and clinical value.
In our efforts to better understand mechanisms underlying barrier disruption and gut inflammation, we have been investigating molecular mechanisms underlying loss of barrier function in the epithelium of the intestinal mucosa. Using monolayers of intestinal Caco-2 cells, as a well established and widely used model for barrier function, we have reported (Banan et al., 1998a, 1999; 2000a,b, 2001a,b,c) that cytoskeletal assembly and stability is required for epithelial integrity. Moreover, we have shown that changes in protein kinase C (PKC) affect epithelial barrier function (Banan et al., 2001a,b). Although PKC, in general, is known to alter epithelial barrier, the role of specific PKC isoforms in barrier regulation and their mechanisms have remained largely unknown.
PKC consists of a family of serine- and threonine-specific kinases, which includes at least 12 known isoenzymes that can be classified into three subfamilies (Housey et al., 1988; Ponzoni et al., 1993; Maruvada and Levine, 1999; Banan et al., 2001a,b, 2002a, 2003a,b, 2004): conventional (or “classical”) PKC isoforms (α, β1, β2, and γ), “novel” PKC isoenzymes (δ, ϵ, θ, η, and μ), and “atypical” PKC isoforms (λ, τ, and ζ). Intestinal cells (including Caco-2 cells) express at least 10 isoforms of PKC, including PKC-α, PKC-β1, PKC-β2, PKC-δ, PKC-ϵ, PKC-θ, PKC-η, PKC-ζ, PKC-λ, and PKC-τ (McKenna et al., 1995; Maruvada and Levine, 1999; Banan et al., 2001a,b, 2002a, 2003a,b, 2004). These isozymes differ in their mechanism of activation, subcellular distribution, substrate type, and expression, suggesting that each of these PKC isoforms can perform unique biological tasks (Persons et al., 1988; Melloni et al., 1990; Mischak et al., 1993; Ponzoni et al., 1993; Banan et al., 2002a, 2003a, 2004). In our previous studies, we demonstrated both damaging and protective mechanisms/pathways affecting gut barrier integrity and permeability. For example, we showed that growth factors (epidermal growth factor or transforming growth factor-α) prevent oxidant-induced barrier dysfunction via activation of the classical PKC-β1 isoform as well as by the atypical PKC-ζ isoform, leading to monolayer protection (Banan et al., 2001a, 2003a,b). More recently, we reported novel findings that PKC-θ seems to affect epithelial barrier function (Banan et al., 2004). We showed that activation of the 82-kDa PKC-θ, a novel PKC isoform, is involved in epithelial barrier alterations. Despite the importance of the θ isoform of PKC to intestinal barrier permeability, the fundamental mechanisms underlying PKC-θ-mediated alterations in monolayer barrier function still remain largely unexplored.
The ability of the intestinal epithelium to maintain selective barrier permeability depends on a complex array of protein filaments that includes the cytoskeleton and barrier function proteins such as claudins (Banan et al., 1996, 1998a,b, 1999, 2000a,b, 2001a,c; Unno et al., 1997; Furuse et al., 1999; Yoo et al., 2003a). For example, claudins, a crucial family of barrier function proteins, constitute the major components of the junctional strands that delineate apical-lateral membranes as well as form the so-called seal (barrier) in the intercellular space (Furuse et al., 1999; Ishizaki et al., 2003; Yoo et al., 2003a; Fujibe et al., 2004). The claudin family consists of at least 23 different isotypes that are essential in regulation of permeability function of cell monolayers (Furuse et al., 1999). In particular, claudin isotypes such as claudin-1 to -5 are thought not only to form aqueous pores within the cellular junctional strands but also are important to paracellular permeability of the intestinal epithelium (Yoo et al., 2003a). The mechanisms underlying regulation of claudins and intestinal barrier function remain poorly understood.
In the current study, we determined the role of the θ isoform of PKC in the underlying mechanisms of barrier regulation and investigated possible alterations of key claudin proteins in regulation of the permeability of the intestinal epithelium. Exploring the role of the PKC-θ in the mechanism of epithelial barrier function is essential because 1) it is of significant biological importance to establish the idea that “specific” isoforms of PKC play fundamental roles in endogenous modulatory mechanisms of cellular proteins required for the maintenance of intestinal barrier function, and 2) a better understanding of effectively modulating (e.g., by PKC-θ) the permeability function of the intestinal mucosal epithelium could lead to the development of novel therapeutic strategies for inflammatory diseases of the gastrointestinal tract that are related to disruption of barrier function.
To this end, we studied the effects of PKC-θ on claudin proteins using three complementary molecular approaches. First, using an antisense approach, native (82-kDa) PKC-θ isoform was reliably underexpressed. Second, using a dominant-negative approach, endogenous PKC-θ isoform was inactivated. Third, using a sense expression approach, native PKC-θ was reliably increased. We tested the hypothesis that PKC-θ isoform is required for changes in the dynamics of claudin assembly and cytoarchitecture and monolayer permeability function in intestinal epithelium. We now report novel mechanisms dependent on the PKC-θ isoform activity, namely, alterations of the claudin-1 and claudin-4 isotypes phosphorylation, membrane assembly, and distribution as well as permeability function in cell monolayers. The biological ability to change the dynamics of barrier function protein claudins is a new mechanism not previously attributed to the novel subfamily of PKC isoforms in cells.
Materials and Methods
Cell Culture. Caco-2 cells were obtained from American Type Culture Collection (Manassas, VA) at passage 15. This widely used colonic cell line was chosen for our studies because they form monolayers that morphologically resemble intestinal cells, with defined apical brush borders and tight junctions, and a highly organized claudin-ring network upon differentiation (Gilbert et al., 1991; Meunier et al., 1995; Banan et al., 1998b). Caco-2 cells are a transformed cell line, and monolayers of tumor cells may respond differently than do nontransformed cells, including enterocytes in native tissue. Nonetheless, there is a wide body of evidence collected on these colonic cells over the past decade that shows their “normal” intestinal epithelial characteristics, which makes them near ideal for gut barrier function studies. For example, Caco-2 cells form monolayers that can be studied for weeks, rather than just days, as is typical of most fresh in vitro preparations (Hurani et al., 1993; Meunier et al., 1995; Unno et al., 1997). This allowed us to measure alterations in intestinal barrier function. In addition, Caco-2 cells closely resemble normal intestinal cells in that they express intestinal hydrolases such as sucrase-isomaltase and alkaline phosphatase. Furthermore, these cells are similar to native intestinal epithelial cells in that they have receptors for prostaglandins, growth factors, vasoactive intestinal peptide, low-density lipoprotein, insulin, and specific substrates such as dipeptides, fructose, glucose, hexoses, and vitamin B12 (Gilbert et al., 1991; Meunier et al., 1995; Banan et al., 1998b). Accordingly, this cell line provides a suitable in vitro model for our studies. To show the general validity of our findings, in select experiments another colonic cell line, namely, HT-29, was used (Balogh et al., 1995; Banan et al., 2002b). Wild-type cells or stably transfected cells (see below) were split at a ratio of 1:6 upon reaching confluence and set up in either six- or 24-well plates for experiments or in T-75 flasks for propagation. Cells were used 7 to 10 days postconfluence. The utility and characterization of these cell lines have been reported previously (Balogh et al., 1995; Meunier et al., 1995; Banan et al., 1996, 1998a,b, 1999, 2000a,b, 2001b,c, 2002ab, 2003a,b, 2004).
Plasmids and Transfection. The sense, antisense, and dominant-negative plasmids of PKC-θ were constructed as we described previously (Banan et al., 2001a, 2002a, 2004). A unique tetracycline-responsive expression (TRE) system was used to overexpress the native PKC-θ. cDNA encoding the entire reading frame of PKC-θ plasmid was subcloned into the TRE vector, creating TRE PKC-θ. The antisense or dominant-negative PKC-θ plasmid was also subcloned to create AS-PKC-θ and dominant-negative PKC-θ.
Cultures of intestinal cells grown to 50 to 60% confluence were cotransfected with hygromycin resistance plasmid and expression plasmids encoding either sense PKC-θ, antisense PKC-θ, or dominant-negative PKC-θ by Lipofectin (Invitrogen, Carlsbad, CA) as we described previously (Banan et al., 2001a; 2004). Control conditions included vector alone. Briefly, cells were incubated for 16 h at 37°C with the plasmid DNA in serum-free media in the presence of LipofectAMINE (25 μl/25-cm2 flask; Invitrogen). Subsequently, the DNA-containing solution was removed and replaced by fresh media containing 10% fetal bovine serum to relieve cells from the shock of exposure to serum-free media. After transfection, cells were subjected to hygromycin selection (1 mg/ml). Resistant cells were maintained in DMEM/fetal bovine serum and 0.2 mg/ml hygromycin (selection medium). For inducible expression of PKC-θ, cells were transfected with a plasmid expressing the tetracycline-responsive transactivator (tTA) along with a second plasmid conferring resistance to G418 (Geneticin). After selection in 0.6 mg/ml G418 (selection media), one such clone (i.e., parental tTA) was then itself transfected with the TRE PKC-θ system. Hygromycin resistance plasmid was included to confer resistance to hygromycin (selection marker, 1 mg/ml). Control conditions included vector alone (TRE-z). Multiple clones expressing PKC-θ or lacking PKC-θ activity were assessed by immunoblotting and activity assay (Banan et al., 2004) and then used for experiments.
Experimental Design. In the first series of experiments, post-confluent monolayers of wild (naive)-type cells were incubated with vehicle (isotonic saline) for 30 min and then assessed for baseline conditions. Experiments were then repeated using multiple clones of stably transfected cells. In all experiments, we assessed claudin skeletal stability (e.g., subapical ring cytoarchitecture), claudin subcellular distribution and assembly (claudin isotype distribution in membrane/cytoskeletal/cytosolic pools), claudin isotype phosphorylation (phospho-serine and phospho-threonine), PKC-θ intracellular distribution (membrane, cytoskeletal, and cytosolic fractions), PKC-θ activity (immunoprecipitation and in vitro kinase assay), and monolayer barrier permeability (clearance of two different sized fluorescent dextran probes, FD70 kDa versus FD4 kDa).
In a second series of experiments, sense-transfected cell clones overexpressing PKC-θ were used. To this end, multiple sense clones that were stably overexpressing PKC-θ (i.e., TRE PKC-θ) were grown as monolayers and then exposed to vehicle. Outcomes measured were as described above. In these overexpression studies, cells were grown for 48 h in the absence of tetracycline before experiments.
In a third series of experiments, monolayers of antisense-transfected cells lacking PKC-θ activity were treated with vehicle. In all experiments, PKC-θ activity was determined in immunoprecipitated samples (see below). In a corollary series of experiments, we investigated the effects of a PKC-θ dominant-negative mutant on the state of monolayer barrier function, claudin isotype subcellular pool/assembly, and claudin isotype phosphorylation and cytoarchitecture. To this end, claudin isotypes (the structural protein subunit of permeability junctions) were isolated and then analyzed by immunoblotting. Claudin integrity was assessed by 1) immunofluorescent labeling and fluorescence microscopy to determine the percentage of cells with normal claudin isotypes, 2) detailed analysis by high-resolution laser scanning confocal microscopy (LSCM), and 3) immunoprecipitation and PAGE analysis of claudins.
Fractionation and Immunoblotting of PKC. Cell monolayers grown in large 75-cm2 flasks were processed for the isolation of the cytosolic, membrane and cytoskeletal fractions as we described previously (Banan et al., 2001a, 2004). Protein content of these various fractions was assessed by the Bradford method (Bradford, 1976). Samples (5 μg of protein/lane) were separated by PAGE. The immunoblotted proteins were visualized by ECL (Amersham Biosciences, Inc., Piscataway, NJ) and autoradiography (e.g., 1 h at –20°C). The exposure times were adjusted to ensure linear responses. Under these conditions, the chemiluminescence assay was linear between 1 and 10 μg of total protein. Standard (purified PKC-θ) loading controls (1 μg/lane) were also run concurrently with each run. To further verify equal loading, blots were routinely stained with 0.1% India ink in TBST buffer.
For the isolation of the cell fractions, after treatments, postconfluent monolayers were scraped and ultrasonically homogenized (GE130, GE Ultrasonic Processor, amplitude 50, 6 pulses/s, duration 20 s) in Tris-HCl buffer (20 mM Tris-HCl, pH 7.5, 0.25 mM sucrose, 2 mM EDTA, 10 mM EGTA, 2 μg/ml aprotinin, 2 μg/ml pepstatin, 2 μg/ml leupeptin, and 2 μg/ml phenylmethylsulfonyl fluoride). The homogenates were then ultracentrifuged (100,000g for 40 min at 4°C), and the supernatant was removed and used as a source of the cytosolic fraction. Next, pellets were washed with 0.2 ml of Tris-HCl buffer and resuspended in 0.8 ml of buffer containing 0.3% Triton X-100 and maintained on ice for 1 h. The samples were then centrifuged (100,000g for 1 h at 4°C), and the supernatant was used as the source of the membrane fraction. To this remaining pellet, we added 0.3 ml of cold (4°C) lysis buffer (150 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA, 1 mM EGTA, 1% Nonidet P-40, 0.1% sodium deoxycholate, 0.1% SDS, 2 μg/ml aprotinin, 2 μg/ml pepstatin, 2 μg/ml leupeptin, and 2 μg/ml phenylmethylsulfonyl fluoride). The samples were then placed on ice for 1 h and ultracentrifuged as described above. The remainder of the lysate or Triton-insoluble cytoskeletal fraction was then removed. For total extraction, which provides the fraction used to confirm total PKC-θ, scraped monolayers were placed directly into 1.5 ml of cold lysis buffer and subsequently ultracentrifuged as described above. The supernatant was used for bulk protein determination.
For immunoblotting, samples (5 μg protein/lane) were added to a standard SDS buffer, boiled, and then separated on 7.5% SDS-PAGE (Banan et al., 2004). The immunoblotted proteins were incubated with a primary monoclonal antibody to PKC-θ [nPKC-θ (E7), sc-1680 (human reactive), Santa Cruz Biotechnology, Inc., CA] at 1:3000 dilution. A horseradish peroxidase-conjugated antibody (Molecular Probes, Eugene, OR) was used as a secondary antibody at 1:4000 dilution. Proteins were visualized by enhanced chemiluminescence and autoradiography and subsequently analyzed by densitometry. The identity of the PKC-θ band was assessed by a procedure that we described previously (Banan et al., 2004). 1) Using a PKC-θ blocking peptide (sc-1680 P; Santa Cruz Biotechnology, Inc.) in combination with the anti-PKC-θ antibody that prevents the appearance of the corresponding “major” band in Western blots. Additionally, 2) in the absence of the primary antibody to PKC-θ, no corresponding band for PKC-θ was observed. 3) The PKC-θ band ran at the expected molecular mass of 82 kDa as confirmed by a known positive control for PKC-θ (from rat brain lysates). 4) Prestained molecular weight markers (Mr 67,000 and 93,000) were run in adjacent lanes. In other studies using total PKC extracts, we confirmed our previous findings (Banan et al., 2004) that expression of PKC-θ or inhibition of PKC-θ did not affect the relative expression levels of other PKC isoforms.
Immunoprecipitation and PKC-θ Activity Assay. Immunoprecipitated PKC-θ was collected and processed for its ability to phosphorylate a synthetic peptide (Banan et al., 2004). Briefly, after treatments, confluent cell monolayers were lysed by incubation for 20 min in 500 μl of cold lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 μg/ml antiprotease cocktail, 10% glycerol, 1 mM sodium orthovanadate, 5 mM NaF, and 1% Triton X-100). The lysates were clarified by centrifugation at 14,000g for 10 min at 4°C. For immunoprecipitation, the lysates were incubated for 90 min at 4°C with anti-PKC-θ (1:2000 dilution, in excess). The extracts were then incubated with protein A/G plus agarose for 1 h at 4°C. The immunocomplexes were collected by centrifugation (2500g for 5 min) in a microcentrifuge, tube, and washed three times with immunoprecipitation buffer containing 5 mM Tris-HCl, pH 7.4, and 0.2% Triton X-100. They were then washed one time with kinase buffer (20 mM HEPES, pH 7.5, 10 mM MgCl2, 2 mM MnCl2, and 20 μM ATP) and resuspended in 20 μl of kinase buffer and 5 μl of 5× reaction buffer (1 mg/ml histone H1 and 0.25 mg/ml l-α-phosphotidyl-l-serine) plus 5 μCi of [γ-32P]ATP and subsequently incubated for 5 min at 30°C. Reactions were then stopped by the addition of 8 μl of 5× sample buffer, and the samples were boiled at 95°C for 5 min before separation by 7.5% PAGE. The extent of histone H1 phosphorylation was determined by scintillation counting of excised Coomassie Blue histone polypeptide bands. Counts for blanks were subtracted from the sample activity. Sample activity was also corrected for protein concentration (Bradford method; Bradford, 1976), and PKC-θ activity was reported as picomoles per minute per milligram of protein.
Immunofluorescent Staining and High-Resolution Laser Scanning Confocal Microscopy of Claudin Isotypes. Cell monolayers were fixed in a standard cytoskeletal stabilization buffer and then postfixed in 95% ethanol at –20°C as we described previously (Banan et al., 1998a,b, 1999, 2000a,b, 2001a,b,c). Cells were subsequently processed for incubation with an isotype-specific primary antibody, monoclonal anti-claudin isotypes (claudin-1, -2, -3, -4, or -5; Zymed Laboratories, South San Francisco, CA), and then with a secondary antibody (fluorescein isothiocyanate- and/or Texas Red-conjugated; Sigma-Aldrich, St. Louis, MO). After staining of claudins, cells were observed using a 63× oil immersion plan-apochromat objective, numerical aperture 1.4 (Carl Zeiss, Jena, Germany). The claudin elements were examined in a blinded manner for their overall morphology, orientation, and disruption as we have described for other cytoskeletal components previously (Banan et al., 1998a,b, 1999, 2000a,b, 2001a,b,c). At least 1200 cells per group (200 × 6 slides) were examined by a blinded observer in four different subapical fields by laser scanning confocal microscopy, and the percentage of cells displaying normal claudins was determined, reported as mean ± S.E.M. The identity of the treatment groups for all slides was decoded only after examination was complete.
In other studies, immunofluorescent analysis of the subcellular distribution of claudins was performed using monolayers that were 10 days postconfluence. Using laser scanning confocal microscopy, Z-scans of the entire height of cell monolayers were obtained with scans performed in 200-nm steps over a range of 10 μm (25–50 Z-stacks per cell monolayer area). Because claudins such as claudin-1 and claudin-4 exhibited the highest levels of expression at the “subapical” membrane (e.g., 1.8-μm subapical junctional) areas of intestinal cells, we focused on these areas in subsequent immunofluorescent studies.
Fractionation and Immunoblotting of Claudins. Subcellular fractions were isolated using the methods we described for fractionation and immunoblotting of PKC and cytoskeleton (Banan et al., 2003b, 2004). After fractionation, the “claudin skeleton” was recovered by separately incubating (at 37°C for 30 min) the subcellular fractions (membrane, cytoskeletal, and cytosolic) with stabilizing agents, 1 mM MgSO4 and 1 mM ATP in cytoskeletal stabilization buffer (0.1 M Pipes, pH 6.9, 30% glycerol, 5% dimethyl sulfoxide, 10 μg/ml antiprotease cocktail, 1 mM EGTA, 1 mM MgCl2, and 1 mM ATP). Claudins were then recovered by centrifugation and resuspended in the above-mentioned stabilization buffer (cytoskeletal stabilization buffer). Fractionated samples were then flash frozen in liquid N2 and stored at –70°C until immunoblotting. For immunoblotting, samples (5 μg of protein/lane) were placed in a standard SDS sample buffer, boiled, and then subjected to PAGE on 7.5% gels using appropriate (isotype-specific) antibodies. Standard (purified) claudin loading controls (5 μg/lane) were run concurrently with each run. To additionally verify equal loading, blots were routinely stained with 0.1% India ink in TBST buffer. Furthermore, after the blots were stripped, actin (∼43 kDa) immunoblotting was performed as an internal control for equal loading.
Analysis of Claudin Phosphorylation. Claudins were collected and assessed for phosphorylation by PAGE (Banan et al., 2004). For immunoprecipitation, cell lysates were incubated for 4 h at 4°C with isotype-specific monoclonal anti-claudin antibodies (claudin-1, -2, -3, -4, or -5; 1:50 dilution, in excess). The extracts were then incubated with protein G plus Sepharose 4B (Zymed Laboratories) for 2 h at 4°C. The immunocomplexes were collected by centrifugation (2500g for 5 min) in a microcentrifuge tube, washed three times with immunoprecipitation buffer containing 5 mM Tris-HCl, pH 7.4, and 0.2% Triton X-100. The resultant pellets were resuspended in a standard SDS sample buffer and boiled at 95°C for 5 min before separation by PAGE. Gels were transferred to nitrocellulose membranes, blocked with 1% bovine serum albumin and 0.01% Tween 20 in phosphate-buffered saline for blotting by either anti-phosphoserine or anti-phosphothreonine (1:3000 dilution; BD Transduction Laboratories, Lexington, KY), and for detection of immune complexes by horseradish peroxidase-conjugated secondary antibody, and then incubated with chemiluminescent (ECL) reagents and autoradiographed.
To assess the specificity of claudin phosphorylation, in corollary experiments, blots used to detect the phospho-serine and phosphothreonine states of claudin immunoprecipitates were placed in protein stripping buffer and then reprobed with the corresponding isotype-specific monoclonal claudin antibodies. These protocols were done to confirm that changes in the serine- and threonine-phosphorylation were in fact specifically associated with claudin-1 and claudin-4 proteins (and not due to simple reactivity against nonspecific targets). To further assess phosphorylation specificity, we concurrently coincubated the immunoprecipitated PKC-θ with the immunoprecipitated claudin (claudin-1 or -4) under in vitro (test tube) conditions. For these protocols, the 1 mg/ml PKC-θ immunocomplexes were resuspended in 20 μl of kinase buffer (20 mM HEPES, pH 7.5, 10 mM MgCl2, 2 mM MnCl2, and 20 μM ATP) and 5 μlof5× reaction buffer (containing 1 mg/ml claudin-1 or -4 and 0.25 mg/ml l-α-phosphotidyl-l-serine) plus 5 μCi of [γ-32P]ATP and subsequently incubated for 5 min at 30°C. Reactions were then stopped by the addition of 8 μlof5× sample buffer, and the samples were boiled at 95°C for 5 min before separation by PAGE. Identical patterns of phosphorylation were seen as that of the protocols using immunoprecipitated claudins alone after PAGE.
In all experiments, positive loading controls were run concurrently with each run. Specifically, for claudin phosphorylation blots, known positive controls (5 μg of corresponding phosphorylated claudins) were run with each gel (purified claudin isotype was phosphorylated by an in vitro kinase reaction via addition of 1 mg/ml PKC-θ in 20 μl of kinase buffer containing 20 mM HEPES, pH 7.5, 10 mM MgCl2, 2 mM MnCl2, and 20 μM ATP). These positive controls also served as additional loading controls (5 μg of loading control per lane) that allowed for the comparison of data derived from different blots (or different days). Moreover, to ensure accuracy, loading controls were routinely run in duplicates or in some cases in triplicates. To further verify equal loading of lanes, blots were routinely stained with 0.1% India ink in TBST buffer (containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.05% Tween 20).
Determination of Monolayer Barrier Permeability by Fluorometry. Status of monolayer barrier function was assessed by a widely used and validated technique that measures the apical to basolateral paracellular clearance of fluorescent markers such as fluorescein dextrans (FD70 kDa, 1 mg/ml) as we (Banan et al., 1999, 2000a,b, 20001a,b,c, 2002a) and others (Hurani et al., 1993; Sanders et al., 1995; Unno et al., 1997) described previously. In select experiments, a lower molecular weight fluorescein dextran (FD4 kDa, 1 mg/ml) was used. Briefly, fresh phenol-free DMEM (800 μl) was placed into the lower (basolateral) chamber and phenol-free DMEM (300 μl) containing probe (FD70 or FD4) was placed in the upper (apical) chamber. Aliquots (50 μl) were obtained from the upper and lower chambers at zero time and at subsequent time points and transferred into clear 96-well plates (clear bottom; Costar, Cambridge, MA). Fluorescent signals from samples were quantitated using a fluorescence multiplate reader (FL 600; BIO-TEK Instruments). The excitation and emission spectra for probes were excitation = 485 nm and emission = 530 nm. In exploratory studies assessing barrier function in our PKC-θ-transfected models, we consistently observed picomolar to nanomolar fluxes of FD in a range of sizes (4–70 kDa) across Caco-2 monolayers. Such FD flux was dependent on the transfected clones used (e.g., 1 μg of antisense versus 5 μg of antisense clone) and thus flux (clearance) increased with increasing amounts of DNA transfected. Also, when adding FD at concentrations in the range 0.001 to 1 mg/ml, we observed a graded increase in FD flux. We were also able to consistently observe fluxes of picogram quantities of FD across monolayers. Clearance was then calculated as flux divided by surface area of the monolayer (Hurani et al., 1993; Sanders et al., 1995; Unno et al., 1997). Thus, flux (nanoliters per hour) was calculated and subsequently normalized to surface area of monolayers (0.3 cm2) and reported as a clearance (nanoliters per hour per square centimeter). More specifically, clearance (Cl) was calculated using the following formula: Cl (nL/h/cm2) = Fab/([FD70 or FD4]a × S), where Fab is the apical to basolateral flux of FD70 or FD4 (light units per hour, which is dependent on the fluorescent signals of samples obtained from the lower and upper chambers of Transwell culture inserts), [FD70 or FD4]a is the concentration at baseline (light units per nanoliter), and S is the surface area (i.e., 0.3 cm2). Simultaneous (concurrent) controls were run with each experiment.
Statistical Analysis. Data are presented as mean ± S.E.M. All experiments were carried out with a sample size of at least six observations per treatment group. Statistical analysis comparing treatment groups was performed using analysis of variance followed by Dunnett's multiple range test (Harter, 1960). Correlational analyses were done using the Pearson test for parametric analysis or, when applicable, the Spearman test for nonparametric analysis. p values <0.05 were deemed statistically significant.
Results
Our previous findings (Banan et al., 2004) showed that intestinal cells cotransfected with complementary DNA (cDNA) encoding the hygromycin resistance (for selection) and the PKC-θ anti-sense (AS-PKC-θ) stably underexpress the novel θ, (82-kDa) isoform of PKC (nPKC-θ). Multiple clones of intestinal cells transfected with 1 to 5 μg of PKC-θ antisense cDNA demonstrated a dose-dependent underexpression of the PKC-θ protein (Fig. 1A, immunoblot). The clone transfected with 4 μg of PKC-θ antisense (-θ4) led to the largest reduction (∼99%) in the levels of native PKC-θ protein while inducing maximum monolayer barrier disruption (increasing clearance of fluorescein sulfonic acid, 0.478 kDa). Transfection of only the empty vector (control vector alone) did not do so. In the present investigation, we have studied the underlying mechanisms by which nPKC-θ affects barrier function.
Stable Underexpression of the nPKC-θ Isoform by Antisense Leads to Instability of Claudin Isotypes Assembly and Architecture. Using the aforementioned unique clones of intestinal Caco-2 cells, we first assessed the possible role of the PKC-θ isoform in the molecular dynamics of key claudin proteins involved in barrier function. Underexpression of native PKC-θ caused by transfection of anti-sense (AS-PKC-θ; 1, 2, 3, 4, or 5 μg of cDNA) dose-dependently injured the claudin-1 ring-like skeleton (Fig. 1A, graph). For example, for the 4-μg clone underexpressing PKC-θ (and exposed to vehicle) claudin-1 in the monolayer was disrupted as demonstrated by a low percentage of intestinal Caco-2 cells displaying the normal claudin-1 isotype. Indeed, the clone transfected with 4 μg of PKC-θ antisense (-θ4) provided maximum increases in claudin-1 instability. WT cells (those expressing native PKC-θ levels), in contrast, showed normal claudin-1 as demonstrated by high percentage of cells with intact claudin-1 isotype. As expected, transfection of only the empty vector by itself (vector alone) did not affect the claudin-1 isotype (% normal claudin-1 = 98 ± 2% for empty vector-transfected cells exposed to vehicle and 99 ± 1% for wild-type cells exposed to vehicle). In fact, both empty vector clones and wild-type cells responded in a similar manner to vehicle, exhibiting normal claudin-1.
Representative laser scanning confocal microscopy of immunofluorescently labeled claudin-1 corroborates (Fig. 1B, a–c) that the 4-μg clones underexpressing PKC-θ exhibited a loss of the normal claudin-1 skeletal-ring architecture (b; see arrows). This instability is seen in the intracellular appearance of a torn, beaded, and fragmented claudin-1 “ring” (from the “1.8-μm subapical” area of cell-cell contact). Wild-type cells (a), on the other hand, show an intact caludin-1 ring at these same areas of cell-cell contact. This normal architecture is indistinguishable from the empty vector-transfected clones (c), which were also exposed to vehicle.
We then assessed other claudin isotypes in our PKC-θ clones and native counterparts (Fig. 2, A and B). Native PKC-θ underexpression in a dose-dependent manner also injured the claudin-4 isotype ring-like skeleton (Fig. 2A). As for claudin-1 isotype, the 4-μg PKC-θ antisense clone (-θ4) led to largest increases in claudin-4 isotype instability. In wild-type intestinal cells, claudin-4 was normal. Furthermore, transfection of vector alone, as might be expected, did not affect the claudin-4 isotype (% normal claudin-4 = 100 ± 1% for empty vector-transfected cells exposed to vehicle and 100 ± 1% for wild-type cells exposed to vehicle).
Laser confocal microscopy further revealed (Fig. 2B, a–c) that in wild-type cells (a) claudin-4 isotype, similar to claudin-1, looks like an intact ring on the inner side of the plasma membrane, i.e., 1.8-μm subapical areas of cell-cell contact (see arrows). This is demonstrated by a continuous and smooth distribution of the claudin-4 ring-like architecture at these areas. Moreover, in PKC-θ underexpressing (4-μg) clones (b), the claudin-4 ring shows a clear fragmentation and disorganization, whereas for the empty vector clone (c) claudin-4 architecture was highly maintained (resembling wild-type cells).
In comparison, other intestinal isotypes of claudin, including claudin-2, claudin-3, and claudin-5, did not seem to be affected by antisense to PKC-θ. For instance, representative “two-color” (double stain) studies of claudin-1 or -4 compared, for example, with claudin-2 in the same intestinal cells (Fig. 3, A and B, respectively) showed that Caco-2 cells express little or no claudin-2 isotype (compared with either claudin-1 or -4). Similarly, other representative two-color studies of claudin isotypes, including claudin-4 and claudin-2 (Fig. 3C) corroborate the above-mentioned findings. Studies of other claudin isotypes such as claudin-3 and claudin-5 are shown in Fig. 3D.
Underexpression of Endogenous nPKC-θ Isoform Causes a Dynamic Instability of Monolayer Barrier Function: Increased Permeability to Large Fluorescein Dextrans (FD70 kDa, FD4 kDa). Underexpression of endogenous PKC-θ by antisense (AS-PKC-θ) led to loss of Caco-2 monolayer barrier function (Table 1). In particular, barrier function in multiple clones of intestinal cells transfected with 1 to 5 μg of PKC-θ antisense plasmid showed a dose-dependent instability of monolayer permeability function as demonstrated by large increases in 70-kDa fluorescein dextran (FD70) clearance. WT cells (expressing native PKC-θ levels) showed normal permeability function. Interestingly, the clone transfected with 4 μg of PKC-θ antisense (-θ4) provided maximum increases in barrier permeability, paralleling findings on the instability of claudin isotypes -1 and -4. Transfection of the empty vector, as expected, did not disrupt barrier function (FD70 kDa clearance = 0 ± 0 nl/h/cm2 for vector-transfected cells exposed to vehicle and 0 ± 0 for wild-type cells exposed to vehicle). Both empty vector clone and wild-type cells responded similarly to vehicle, showing normal barrier function.
Furthermore, there was a dose-dependent instability of Caco-2 monolayer barrier function in the above-noted AS-PKC-θ clones when we used a smaller size permeability probe, namely, 4-kDa fluorescein dextran (FD4) (Table 1). Not surprisingly, the 4-μg clone also exhibited maximum increases in monolayer permeability to FD4 kDa. Indeed, there was a size-dependent increase in monolayer permeability in order of decreasing size for these probes (70 kDa FD < 4 kDa FD), suggesting the dynamic nature of barrier alterations. We observed a similar trend of alterations for disruption of both barrier function and claudin stability in another intestinal epithelial cell line, HT-29 (Table 2). For example, 4 μg of AS-PKC-θ clone also led to largest instability in claudin (i.e., reduced percentage of normal claudin-1 and claudin-4 isotypes) in HT-29 cells. Because the clone transfected with 4 μg of AS-PKC-θ cDNA led to maximum levels of monolayer barrier disruption and claudin instability, we used this clone for subsequent mechanistic studies.
Antisense Reduction of Endogenous nPKC-θ Leads to Abnormal Alterations in the Membrane and Cytoskeletal Assembly of Claudin Isotypes. We determined the effects of native PKC-θ underexpression on the cytosolic, membrane, and cytoskeletal pools of claudin-1 and -4 by assessing the 22- and 25-kDa (respectively) structural protein of these isotypes. The various subcellular claudin pools were isolated from the inhibitory (antisense) clones and analyzed after SDS-PAGE fractionation. Tables 3 and 4 show results of analysis for the subcellular distribution of claudin-1 and claudin-4 in the cytosolic, membrane, and cytoskeletal fractions of Caco-2 cells (data presented as a fraction of total claudin isotype). These findings indicate that underexpression of PKC-θ abnormally decreases claudin-1 (Table 3) and claudin-4 (Table 4) in the membrane and cytoskeletal pools while it concomitantly increases their proportions in cytosolic pools.
For example, PKC-θ-underexpressing clone (4 μg of AS-PKC-θ) exhibited an abnormal reduction in the particulate (particulate = membrane + cytoskeletal) pool of claudin-1 or claudin-4 as demonstrated by a decrease in its band density (not shown), indicating instability of claudin-1 or -4 assembly. In wild-type cells, we did not find any decreases in the particulate pool of claudin, indicating normal assembly of claudin skeleton. In these wild-type cells, as might be expected, claudin-1 or -4 particulate distribution was comparable with that found in the empty vector clone. Transfection of empty vector, as for its lack of effects on permeability and claudin architecture, had no affect on the particulate pool of claudin. These findings on the subcellular assembly/distribution of claudins parallel the destabilizing effects of antisense, underexpression of native PKC-θ on intestinal claudin architecture and monolayer barrier function.
Antisense Suppression of nPKC-θ Causes Alterations in Serine- and Threonine-Phosphorylation States of Claudin Isotypes in Intestinal Epithelium. We subsequently probed molecular processes underlying the observed effects of PKC-θ on the monolayer claudins and barrier permeability. Accordingly, claudin-1 and claudin-4 isotypes were immunoprecipitated (by isotype-specific antibodies) from the transfected and wild-type cells, and subsequently subjected to PAGE to determine their phosphorylation (Fig. 4, A to D). Immunoprecipitated claudin-1 isotype was largely serine- and threonine-phosphorylated in wild-type cells (Fig. 4, A and B, corresponding lane a), but not in antisense-transfected cells where suppression of native PKC-θ markedly decreased claudin-1 serine/threonine-phosphorylation (corresponding lane c). We found a similar trend of alterations for the phosphorylation state of claudin-4 isotype in wild-type and antisense clones (Fig. 4, C and D). In the empty vector clones, both claudin isotypes were phosphorylated (corresponding lane b in all blots). The state of claudin serine/threonine-phosphorylation in these vector clones resembled that of the wild-type cells. When the same phosphorylation blots were stripped and reprobed with the corresponding isotype specific monoclonal claudin antibody, we confirmed that changes in phosphorylation were in fact specifically associated with the claudin-1 and -4 proteins (not shown).
Native nPKC-θ Isoform Is Complexed with Claudin. To further study the mechanism underlying the unique stabilizing affects of native PKC-θ on claudins, we used immunoprecipitation analysis (Fig. 5, A–D). In a first series of approaches, cells were initially immunoprecipitated with a monoclonal PKC-θ antibody and then the immune complexes were analyzed for the presence of claudin-1, assessing whether this PKC isoform physically coprecipitates with claudin-1 isotype. Antisense clones underexpressing endogenous PKC-θ did not show any complex formation between these two proteins (Fig. 5A, lane c). In contrast, transfection of the empty vector control did not do so, exhibiting coprecipitation of these two proteins (lane d). Similar to empty vector clone, the amount of claudin-1 coprecipitation was markedly enhanced in wild-type (resting) vehicle-treated cells (lane b), indicating likely presence of a PKC-θ/claudin-1 complex under native conditions. As expected, an irrelevant primary antibody (normal rabbit serum) did not lead to coprecipitation (i.e., no complex formation bands were seen), further suggesting specificity of the coprecipitation seen.
In a second (reverse) series of approaches (Fig. 5B), we further corroborated the aforementioned coassociation findings. Here, anti-claudin-1 antibody was used and immunoprecipitates were then analyzed for the presence of PKC-θ. Not surprisingly, PKC-θ was not seen in the complex in antisense clones, i.e., no coprecipitation with claudin-1 (lane c). On the other hand, wild-type (vehicle)-treated cells showed an accumulation of native claudin-1/PKC-θ complexes (lane b), paralleling findings in Fig. 5A. Similarly, empty vector cells exhibited the formation of the endogenous claudin-1/PKC-θ complexes (lane d). As before, an irrelevant antibody was ineffective (no coprecipitation).
In a third series of approaches, we further assessed the specificity of inhibition of formation of the PKC-θ/claudin-1 complexes in our PKC-θ antisense clones. Here, we probed lysates from two other PKC isoform antisense clones, the classical PKC-β2 and the PKC-α AS. As expected, in these other clones, we could not inhibit the formation of PKC-θ/claudin-1 complexes (not shown).
Using the aforementioned series of approaches, we found a similar trend of coprecipitation and complex formation between PKC-θ and claudin-4 isotype in intestinal cells (Fig. 5, C and D). As for claudin-1, probing lysates from other PKC isoform antisense clones—the classical PKC-β2 AS and the PKC-α AS—did not suppress the formation of PKC-θ/claudin-4 complexes. This again indicates the specificity of PKC-θ/claudin coprecipitation we observed. Also, incubation with an irrelevant antibody was ineffective (not shown).
Subcellular Activity and Distribution of nPKC-θ: Native PKC-θ Isoform Is Constitutively Active and Present Mostly in the Membrane and Cytoskeletal Fractions of Intestinal Cells. Findings from in vitro kinase activity assay of cytosolic, membrane, and cytoskeletal subcellular fractions (Fig. 6A) confirm that the native θ isoform of PKC is mostly active in the particulate cell fractions (particulate includes membrane and cytoskeletal fractions) with only a small activity in the cytosolic fractions, indicating the constitutive activity of the θ isoform of PKC under native conditions. Empty vector control was indistinguishable from the wild-type cells. In antisense-transfected clones underexpressing PKC-θ, on the other hand, there was an almost complete lack of native θ isoform activity compared with the wild-type cells. Immunoblotting assessment of the subcellular distribution of the PKC-θ protein (Fig. 6B) additionally shows that the native (82-kDa) θ isoform of PKC is distributed mostly in the membrane and cytoskeletal fractions of wild-type intestinal cells. Not surprisingly, in antisense clones, we found a near complete absence of PKC-θ protein in these same subcellular fractions, further indicating PKC-θ inactivity. Overall, these findings confirm that the endogenous PKC-θ isoform is “constitutively active” in the membrane and cytoskeletal (particulate) subcellular fractions because it is found to be most active in these fractions under native conditions. These findings on PKC-θ subcellular activity parallel our findings on the subcellular distribution and coprecipitation of claudin isotypes.
Native nPKC-θ Activity Robustly Correlates with Multiple Indices of Claudin Isotype Stability and Barrier Permeability Function. Using data across all-experimental conditions, we report significant (p < 0.05) correlations (r = 0.93 and 0.95, respectively) between PKC-θ activity (in vitro kinase activity assay from the membrane and cytoskeletal fractions) and enhanced monolayer claudin-1 and claudin-4 integrity/stability (i.e., % normal). We found other robust correlations when two other markers of monolayer barrier stability, either FD70 permeability (decreased FD70-kDa clearance) or FD4 permeability (decreased FD4-kDa clearance) were correlated with PKC-θ activity (r = 0.96 and 0.98, respectively; p < 0.05 for each). Additional robust correlations were found between other markers of monolayer barrier stability, including enhanced claudin-1 or claudin-4 membrane/cytoskeletal pools (i.e., claudin-1 or claudin-4 assembly) and PKC-θ activity (r = 0.90 and 0.92, respectively; p < 0.05 for each). We found still other consistent correlations such as those between claudin-1 phosphorylation or claudin-1 stability and PKC-θ activation (r = 0.88 and 0.97, respectively; p < 0.05 for each) as well as between claudin-4 phosphorylation or claudin-4 stability and PKC-θ activation (r = 0.89 and 0.94, respectively; p < 0.05 for each).
Inactivation of Endogenous nPKC-θ Isoform in the Membrane and Cytoskeletal Fractions by Targeted Dominant-Negative Approach and Its Consequent Destabilization of Claudin Membrane Assembly and Monolayer Barrier Function. The aforementioned findings together indicate that PKC-θ could play a pivotal intracellular function in claudin and barrier stability. We then further assessed the role of PKC-θ by using an “independent” dominant-negative approach to stably decrease the activity of endogenous PKC-θ isoform. Here, we used mutant clones we recently developed by transfecting cells with a PKC-θ dominant-negative fragment and a plasmid encoding hygromycin resistance (Banan et al., 2004). Using this targeted approach, we are capable of markedly reducing the steady-state activity levels for native PKC-θ isoform in various subcellular fractions of these dominant-negative mutants (Fig. 7, 3-μg clone). In wild-type cells, similar to empty vector clone, native PKC-θ isoform activity is high in the particulate (membrane and cytoskeletal) cell fractions, showing steady-state constitutive activity. As expected, dominant-negative fragment transfection did not affect PKC-θ protein expression (not shown).
Table 5 shows the dose-dependent effects of varying levels of PKC-θ dominant-negative plasmid (1–5 μg of mutant cDNA) on Caco-2 monolayer barrier function assessed by FD70 clearance. We observed similar effects by PKC-θ dominant-negative plasmid in another intestinal cell line, HT-29 (Table 6). Because transfection of 3 μg of dominant-negative plasmid to PKC-θ led to maximum barrier disruption (increased FD70-kDa clearance), this mutant clone was used in subsequent studies.
Assessment of the percentage of mutant cells with a normal claudin-1 and claudin-4 architecture in Caco-2 monolayers demonstrates (Fig. 8, A and B, respectively) that dominant-negative suppression of endogenous PKC-θ activity results in instability of both claudin isotypes as demonstrated by low percentage (approximately 25%) of mutants displaying normal claudins. Claudin instability was comparable with that seen in HT-29 mutant cells (Table 6). In wild-type cells almost 100% of cells exhibit normal claudins.
Additionally, analysis of the subcellular distribution of claudin-1 (Table 3) and claudin-4 (Table 4) from the same mutant clones showed that dominant-negative inactivation of endogenous PKC-θ largely leads to attenuation (reduction) of claudin isotypes in the membrane and cytoskeletal pools. Representative immunoblot of the key particulate (membrane + cytoskeletal) pool from the same mutant clones corroborates that inactivation of PKC-θ reduces claudin isotypes in the particulate pool (Fig. 9, claudin-1 shown), indicating an abnormal claudin assembly.
Assessment of claudin phosphorylation from these mutant clones further demonstrates (Fig. 10, A and B) that inhibition of native PKC-θ activity markedly suppresses serine- and threonine-phosphorylation of claudin-1. Wild-type cells show substantial, steady-state phosphorylation levels for claudin-1 isotype. Claudin-4 isotype followed an almost identical trend of changes in both wild-type and dominant mutant cells (not shown).
Stable Overexpression of nPKC-θ Isoform Protein in Intestinal Cells. The above-mentioned findings using two independent molecular biological approaches (antisense and dominant negative) indicate that native PKC-θ plays a unique role in claudin and monolayer permeability alterations. Using a third molecular approach, we further studied the role of PKC-θ in regulation of claudin assembly and barrier permeability by using intestinal cell clones that stably overexpress the PKC-θ isoform (approximately 2-fold increase in 4-μg sense clones compared with parental type cells) (Fig. 11A, see immunoblot). To this end, parental Caco-2 cells (tTA Parental) were cotransfected with plasmids encoding both hygromycin resistance and a TRE system for full-length endogenous PKC-θ, i.e., TRE PKC-θ. In this TRE system, overexpression (∼2-fold elevation) of native PKC-θ is reached in the absence of tetracycline (TTX), whereas its presence decreases PKC-θ expression to the levels seen in the parental cell line. Using such an approach, we investigated claudin isotypes and permeability alterations in both overexpressing and parental cell lines.
Overexpression of nPKC-θ Isoform Alters Intestinal Monolayer Claudin Assembly, Phosphorylation, and Cytoarchitecture and Barrier Permeability.Table 5 depicts results of transfection experiments in Caco-2 cells that overexpression of PKC-θ (i.e., TRE PKC-θ) by itself disrupts monolayer barrier function as demonstrated by increased FD70 probe clearance. Multiple clones of intestinal cells transfected with 1, 2, 3, 4, or 5 μg of TRE PKC-θ plasmid show a dose-dependent increase in monolayer barrier permeability. HT-29 cells behaved in a similar manner, showing barrier disruption (Table 6). Transfection of only the empty vector (TRE-z) was ineffective (not shown). The clone transfected with 4 μg of TRE PKC-θ led to the highest increase in monolayer permeability, and it was subsequently used for other experiments.
PKC-θ overexpression deleteriously affected monolayer claudins as demonstrated by low percentage of Caco-2 cells with normal claudins (Fig. 11, A and B, claudin-1). As before, HT-29 cells behaved in a comparable manner (Table 6). This overexpression-induced instability of claudin was suppressed when tetracycline was present (i.e., TRE PKC-θ + TTX). In parental cells, in contrast, claudin structure was normal, paralleling findings on barrier permeability. As before, transfection of the empty vector control was ineffective.
Laser confocal microscopy of the intracellular architecture of the monolayer claudin-1 (Fig. 11B, a–d) further corroborates that clones overexpressing PKC-θ (i.e., TRE PKC-θ exposed to vehicle) exhibit an abnormal rearrangement of the claudin-1 ring at areas of cell-cell contact (c). This abnormality is shown by the appearance of a beaded, fragmented and disorganized claudin-1 ring (at the 1.8-μm subapical cell-to-cell junctions). In the presence of tetracycline (which as noted prevents overexpression of PKC-θ; d) these same clones exhibit a highly maintained and smooth claudin-1 ring. This normal architecture is indistinguishable from the parental cells exposed to vehicle (a and b, with or without tetracycline, respectively).
Changes in claudin-4 isotype architecture followed a similar pattern as demonstrated by confocal microscopy (Fig. 11C, a–d). For example, in clones overexpressing PKC-θ (TRE PKC-θ exposed to vehicle; c), the claudin-4 isotype seems fragmented and disrupted. In the presence of tetracycline (d), normal claudin-4 ring is seen. Similar to claudin-1, in parental cells (with or without tetracycline) (a and b, respectively) claudin-4 ring architecture is intact.
Tables 3 and 4 show results of analysis for the subcellular distribution of claudin-1 and claudin-4 isotypes in the cytosolic, membrane, and cytoskeletal fractions in clones overexpressing PKC-θ. In these TRE (overexpressing) PKC-θ cells both claudin-1 (Table 3) and claudin-4 (Table 4) were reduced in the membrane and cytoskeletal fractions (largely shifted into cytosolic fractions), further indicating claudin instability.
Representative immunoblot of the particulate pool (Fig. 12) further corroborates that in cells overexpressing PKC-θ (TRE PKC-θ exposed to vehicle) the membrane + cytoskeletal pool of monolayer claudin-1 were markedly reduced (c), indicating an unstable claudin-1 assembly. Incubation of these clones with tetracycline (i.e., TRE PKC-θ + TTX, where tetracycline prevents overexpression of PKC-θ), as might be expected, maintained claudin-1 particulate pool at near normal. Similarly, parental type cells exposed to vehicle (with or without tetracycline) displayed a normal particulate pool of claudin-1. Transfection of the empty vector by itself was ineffective (i.e., both empty vector-transfected and parental cells responded in a similar manner to vehicle, exhibiting a normal, steady-state claudin particulate pool). Changes in claudin-4 isotype membrane/cytoskeletal pool followed an almost identical trend (not shown).
Assessment of claudin phosphorylation further shows that PKC-θ ectopic expression also reduced claudin phosphorylation as determined by immunoblotting of immunoprecipitated claudin isotypes (Fig. 13, A and B, claudin-1 shown). PKC-θ overexpressing cells show an abnormally low levels of serine- and threonine-phosphorylation for claudin-1 (see corresponding lane c in A and B), which was prevented when tetracycline was present (corresponding lane d). In parental cells, claudin-1 serine/threonine-phosphorylation was normal (lanes a and b). As expected, claudin-4 isotype phosphorylation followed a similar trend of alterations (not shown). The noted findings on barrier function claudin isotypes membrane assembly and cytoarchitecture as well as phosphorylation parallel the destabilizing effects of PKC-θ overexpression on intestinal monolayer barrier permeability.
Overexpression of Native nPKC-θ Causes a Reduction in the Membrane and Cytoskeletal Associated Activity of This Isoform: PKC-θ Activity Correlates with Several Indices of Monolayer Claudin Function.Figure 14 shows results from in vitro kinase activity assay that overexpression of the PKC-θ (TRE PKC-θ) results in a substantial increase in the amount of its activity in the cytosolic fractions (while reducing membrane and cytoskeletal activities). When tetracycline is present, as expected, transfected clones (i.e., TRE PKC-θ + TTX) show near native, constitutive activity levels for PKC-θ in the membrane and cytoskeletal fractions. Moreover, parental cells exposed to vehicle (with or without tetracycline) (i.e., tTA parental ± TTX) show a normal steady-state, constitutive activity for PKC-θ. This PKC isoform is constitutively active under these endogenous conditions because it is natively most active in the membrane and cytoskeletal cell fractions.
We now report other robust (p < 0.05) correlations (e.g., r = –0.94 and –0.91, respectively) such as between reduced PKC-θ activity (decreases in constitutive activity in the membrane/cytoskeletal fractions) and increased monolayer claudin-1 or claudin-4 isotypes instability. When additional markers of monolayer barrier instability, reduced claudin particulate pool (e.g., decreased claudin-1 membrane assembly) or decreased claudin-1 serine/threonine-phosphorylation or even increased barrier (FD) permeability were used against PKC-θ other robust correlations were found (r = –0.92, –0.87, and –0.90, respectively; p < 0.05 for each). Similar supporting correlations are reached when either monolayer barrier stability (e.g., FD70 permeability clearance) or claudin instability/disassembly (e.g., decreased claudin-1 membrane pool) were correlated (r = –0.95). Collectively, these findings using targeted expression of PKC-θ (TRE PKC-θ) parallel our other findings on the destabilizing effects of both PKC-θ underexpression (AS-PKC-θ) and PKC-θ inactivation (dominant-negative PKC-θ) on monolayer claudins and permeability function.
Discussion
In the present investigation, we have demonstrated that the PKC-θ isoform is required for dynamic changes in claudin-1 and claudin-4 isotypes and for permeability function in the intestinal epithelium. The mechanism underlying this unique biological effect of the -θ isoform of PKC seems to be alterations in the organization, assembly, and/or phosphorylation of the claudin isotypes. This is the first time that PKC-θ-dependent mechanisms have been ascribed to the dynamics of permeability protein claudins in cells. We have thus discovered a novel biological mechanism among the novel subfamily of PKC isoforms. These conclusions are based on several independent lines of evidence as discussed below.
First, antisense to native PKC-θ, which leads to underexpression of endogenous PKC-θ, induces unstable-like conditions to the monolayer claudins and barrier function. This instability requires inactivation of the PKC-θ, which is due to the decreased subcellular activity of this isoform in the critical membrane and cytoskeletal fractions. Second, suppression of PKC-θ by the same antisense causes instability of the molecular dynamics of the barrier protein claudins. Here, inactivation of native PKC-θ 1) reduces both serine- and threonine-phosphorylation of claudins, 2) decreases the membrane/cytoskeletal distribution of claudins, 3) increases the instability of the molecular assembly of the claudins, and 4) reduces the percentage of intestinal cells displaying normal claudin-ring architecture. In contrast, in wild-type cells the native (82-kDa) PKC-θ isoform is constitutively active in the key membrane and cytoskeletal pools as well as complexed with claudin isotypes. Third, dominant mutant inactivation of native PKC-θ causes an antisense-like instability in the molecular dynamics of claudins and in permeability function in the intestinal epithelium. In these PKC-θ mutants, barrier dysfunction (i.e., increased permeability) was not only present but also claudin isotypes exhibited decreased phosphorylation as well as reduced particulate levels and reduced cytoarchitectural integrity. Fourth, ectopic expression of native PKC-θ, which also induces a decrease in the subcellular activity of PKC-θ in the key membrane/cytoskeletal pools, evokes a sequence of destabilizing alterations in these sense-transfected clones. Indeed, in these sense clones, inducible expression of PKC-θ results in an almost identical and consistent cascade of disruptive changes to both cellular claudin isotypes and monolayer permeability. Collectively our findings strongly support a unique model for intestinal monolayer barrier dysfunction: ↓PKC-θ isoform activity ⇒ ↓claudin-1/claudin-4 isotype phosphorylation ⇒ ↑claudin-1/claudin-4 membrane/cytoskeletal pool instability (↓C1 and C4 isotypes assembly) ⇒; ↑claudin isotype ring architectural instability ⇒ ↑monolayer permeability.
In the current study, we mainly used Caco-2 cells because it has been extensively studied by us and is a reliable model for studying physiology (such as barrier function) and pathology (activation of inflammatory pathways), and the relevance of these findings to human diseases has been shown by several investigators (Keshavarzian et al., 1992, 2003; Hurani et al., 1993; Meunier et al., 1995; Banan et al., 1996, 1998a,b, 1999, 2000a,b, 2001b,c, 2002ab, 2003a,b, 2004; Unno et al., 1997). More specifically, we and others have shown that monolayers of Caco-2 cells are a reliable and relevant model for 1) studying intestinal barrier function, 2) evaluating mechanisms of barrier disruption (a key mechanism for IBD), and 3) assessing efficacy and mechanisms of PKC modulation of barrier (a key goal of the current study) (Keshavarzian et al., 1992, 2003; Hurani et al., 1993; Meunier et al., 1995; Unno et al., 1997; Banan et al., 2000a,b, 2001a,b,c, 2002a, 2003a,b, 2004). We also showed that findings derived from this in vitro monolayer model are similar to those observed in in vivo studies such as those seen in tissues from patients with IBD (Keshavarzian et al., 1992, 2003). The similarities we previously reported between in vitro and in vivo results support the appropriateness of our Caco-2 model and its use for studying cellular and molecular mechanisms of barrier function/dysfunction.
Although PKC-mediated signal transduction is widely acknowledged to be important in epithelial cell function, its mechanisms have remained only poorly understood. In resting cells, some PKC isoforms are inactive due to their predominant localization to cytosolic pools, whereas other PKC isoforms are active due to their predominant localization to particulate (membrane and cytoskeletal) pools (Mullin et al., 1998; Gopalakrishna and Jaken, 2000; Banan et al., 2002a, 2003a). Different cell types not only express multiple isozymes of PKC, isozymes belonging to various subfamilies (classical, novel, and atypical), but also most PKC isotypes have unique modes of activation, substrate specificity, tissue expression, and subcellular distribution, suggesting that each isoform of PKC mediates distinct biological functions (Housey et al., 1988; Melloni et al., 1990; Mischak et al., 1993; Babich et al., 1997; Banan et al., 2002a, 2003a,b, 2004). It is thus not surprising that the effects of PKC signaling in cells are highly complex and seem to vary widely with respect to experimental conditions as well as tissue and cell types. For example, pharmacological studies suggest that PKC-ϵ, a member of novel PKC subfamily, mediates cytokine-induced disruption of the intestinal cells (Yoo et al., 2003b). We previously showed that attenuation of the novel PKC isoform, PKC-δ, via molecular interventions protects against oxidant-induced damage to intestinal cells (Banan et al., 2002a). We also discovered that the 78-kDa classical β1 isoform of PKC (PKC-β1) is key in protection of colonocytes against oxidant-induced injury (Banan et al., 2003b). Moreover, the 72-kDa atypical ζ isoform of PKC (PKC-ζ) is also required for protection of colonocytes and acts by suppressing oxidative reactions, especially reactions mediated by inducible NOS and NO (Banan et al., 2003a). We have thus shown that both PKC-β1 and PKC-ζ isoforms perform unique tasks in mediating signaling cascades initiated by growth factors (epidermal growth factor and transforming growth factor-α), which leads to protection of the intestinal epithelium (Banan et al., 2003a,b, 2004). Other PKC isoforms such as PKC-δ seem to have opposite effects, allowing or potentiating signaling processes induced by oxidants (H2O2), which results in injury to the epithelium (Banan et al., 2002a). Thus, activating, inhibiting, or mimicking different isotypes of PKC leads to modification of distinct biological tasks and processes (e.g., protection, damage) in the intestinal epithelium. Our present findings on the 82-kDa PKC-θ, we believe, suggest a new function among the novel subfamily of PKC isoforms—“regulation” of epithelial barrier function through changes in phosphorylation, assembly, and organization of the crucial claudin components of tight junctions.
Using monolayers of intestinal cells, we reported (Banan et al., 1998a, 1999, 2000a,b, 2001a,b,c, 2003b) that tubulin and/or actin assembly is important to the maintenance of normal epithelial barrier permeability. A question that remains to be answered is: How do microtubule and/or even actin dynamics relate to claudins or other proteins involved in maintaining barrier function? Based on the known organization of tight-junctions, we propose a mechanism in which microtubules (tubulin) might affect barrier function proteins (claudins) and thereby affect barrier stability. In this view, tubulin, a known structural or adapter protein, is a critical regulatory protein linking cytoskeletal components (e.g., actin, microtubules) and tight-junctions (e.g., claudins, occludin, and ZO). This mechanism requires that adapter proteins with a structural function exist and provide interactions between cytoskeletal components and barrier function proteins as was suggested by recent studies (Tsukita et al., 2001; Jin et al., 2002). For example, ZO-1 (a protein localized to the tight-junction complex) is known to bind to claudins and occludin as well as to actin, and more importantly, it seems to function as an adapter (and as a structural protein) at the cytoplasmic surface of tight-junctions to stabilize or recruit other proteins, especially cytoskeletal proteins (Tsukita et al., 2001). Occludin (another barrier function protein) has a binding domain that binds to ZO, which in turn binds to the cytoskeleton. Indeed, such protein components form a huge macromolecular complex at the cytoplasmic surface of tight-junctions where actin and microtubules are also found (e.g., inner side of subapical plasma membrane as we and others showed previously; Banan et al., 1999, 2000a, 2003b). Thus, adapter proteins can act as critical regulatory and structural proteins, allowing interactions between barrier function (tight-junctional) proteins (e.g., claudins) and the cytoskeleton. Accordingly, loss of any key structural and/or adapter proteins (e.g., tubulin, actin, or ZO) may lead to breakdown of this macromolecular complex and the consequent injurious changes in epithelial barrier function. Consistent with this mechanism, we previously showed that stability of actin and microtubule cytoskeletons are critical to the maintenance of barrier function (Banan et al., 1999, 2000a,b, 2001a,b,c, 2003b). Not surprisingly, both microtubules and actin play a central role in maintaining cellular integrity, structure, and shape (MacRae, 1992; Rodinov and Gelfand, 1993; Mandelkow and Mandelkow, 1995; Banan et al., 2000a). As such, they also govern cell membrane morphology and polarity as well as intracellular transport, other functions essential to the maintenance of barrier function (Gilbert et al., 1991; Banan et al., 1998b, 2000a).
Our series of reports on various PKC isoforms to date suggest a possible mechanism (target) for PKC isoform-induced affects on intestinal barrier function—protein phosphorylation of cytoskeletal components through indirect or direct interaction (e.g., complex formation) with these proteins. For example, PKC-β1 overexpression causes an increase in serine-phosphorylation of the tubulin subunits of microtubules while complexing with tubulin (Banan et al., 2001a, 2003b). This increase was prevented by antisense to PKC-β1 or when complex formation was suppressed. Similarly, in the current study, the θ isoform of PKC seems to complex with claudins and alter claudin phosphorylation (and barrier function) in intestinal cells, suggesting that this PKC isoform may be acting, directly or indirectly, on these tight junction proteins to alter barrier function. This mechanism is consistent with previous studies. PKC has been implicated in rearrangement of the cytoskeleton (Persons et al., 1988; Mischak et al., 1993; Goodnight et al., 1995; Gopalakrishna and Jaken, 2000) and barrier function proteins (Yoo et al., 2003a,b), although it is not completely known which PKC isoforms are key in these processes. Alternatively, PKC isoforms may target, directly or indirectly, phosphorylation of other tight junction proteins (e.g., ZO-1 and ZO-2). Further studies are needed to explore the nature of the interactions (complex formation) between PKC isoforms and their targets, especially tight-junction proteins, in intestinal epithelial cells.
The ability of PKC-θ to affect the architectural organization and permeability of intestinal cells in monolayers could lead to development of new therapeutic modalities for a variety of intestinal disorders, especially IBD, where loss of mucosal permeability function has been found (Hollander, 1992, 1998; Keshavarzian et al., 1992, 1999; Peeters et al., 1994; Hermiston and Gordon, 1995, 2003; Soderholm et al., 1999). The pathophysiology of IBD, which includes ulcerative colitis and Crohn's disease, oscillates between active (symptomatic) phases of disorder when mucosal permeability is increased, and inactive (asymptomatic) phases when mucosal permeability is low. This waxing and waning in barrier permeability is consistent with the disrupted nature of barrier regulatory (structural proteins) in IBD mucosa. For example, we recently showed that the degree of mucosal structural protein instability (an index of barrier function) is closely correlated with the degree of inflammatory response and disease severity in IBD flare (Keshavarzian et al., 2003). The presence of unstable structural components, which indicates an abnormal, leaky gut barrier, has also been shown in intestinal epithelium under IBD-like conditions (Banan et al., 2000a, 2001a, 2002a, 2003a,b, 2004). These findings are further consistent with the current characterization of the pathophysiological development of IBD and its strong link to disruption of mucosal barrier function. For example, disruption of gut barrier permeability function elicits symptoms of inflammation and IBD-like responses in animals (Yamada et al., 1993; Hermiston and Gordon, 1995). Accordingly, a leaky gut is likely to be key in initiation and continuation of the IBD flare where mucosal inflammation leads to a vicious cycle of oxidative processes, barrier structural protein instability, and eventually mucosal injury. The regulatory effects of PKC-θ on barrier function, as we have observed in monolayers of intestinal epithelium, could be crucial in attenuating the perpetuation of barrier instability and inflammatory events in IBD. It is thus possible that changes in PKC-θ such as inactivation (decreased expression) or even hyperactivation might occur during mucosal inflammatory states such as in IBD. Accordingly, up- or down-regulation of PKC-θ isoform activity could lead to barrier instability seen in IBD. Future studies will be needed to explore this venue in IBD patients.
Although PKCs modulate almost all vital functions of cell, our knowledge of their mechanism is limited. This is mainly due to lack of potent and specific activators and inhibitors of PKC isoforms for assessing end points, including barrier regulation. Pharmacological models, which currently are a major tool in investigation of these PKC isoforms depend mainly on “selectivity” of activators and inhibitors—a nonspecific choice at best. Also, the fact that PKC isoenzymes are involved in so many biological functions of cells makes it very hard, if not impossible, to isolate the desired pathway in the cells without interfering with others. Finally, compartment specific PKC isoforms (membrane or cytosolic) or subisotypes (e.g., β1 and δ) might have different or even opposite functions, and this makes interpretation of experimental outcomes difficult. Fortunately, new techniques and agents, especially “targeted molecular biology”, are being developed (as we have done) and hold promise for helping to develop knowledge in this important area, especially to resolve issues concerning barrier function and disease processes.
Footnotes
-
This work was supported in part by a grant from Rush University Medical Center, Department of Internal Medicine, and by National Institutes of Health Grants NIDDK 60511 and NCCAM 01581 (to A.B.) and National Institute on Alcohol Abuse and Alcoholism Grant 13745 (to A.K.). Portions of this work were presented in the abstract form during the annual meeting of the American Gastroenterological Association (Digestive Disease Week), May 2005.
-
doi:10.1124/jpet.105.083428.
-
ABBREVIATIONS: IBD, inflammatory bowel disease; PKC, protein kinase C; TRE, tetracycline-responsive expression; DMEM, Dulbecco's modified Eagle's medium; tTA, tetracycline-responsive transactivator; PAGE, polyacrylamide gel electrophoresis; Pipes, 1,4-piperazinediethanesulfonic acid; LSCM, laser scanning confocal microscopy; ECL, enhanced chemiluminescence; TBST, Tris-buffered saline/Tween 20; AS, antisense; FD, fluorescein dextran; TTX, tetracycline.
- Received January 12, 2005.
- Accepted March 2, 2005.
- The American Society for Pharmacology and Experimental Therapeutics