INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE BODY SPACES of multicellular animals are lined by continuous sheets of epithelial cells. Movement of solutes, ions, and water across these barriers occurs through both the transcellular pathway, via a large number of transmembrane channels and carriers, and the paracellular pathway, via intercellular tight junctions (TJs). Together these complementary pathways establish the selective and regulated barriers required for absorption and secretion. Our understanding of transcellular transport mechanisms is quite advanced, whereas the molecular basis for the fundamental property of charge selectivity in the paracellular pathway remains less well defined.

The barrier properties of TJs vary widely among tissues in both magnitude, typically quantified as electrical resistance, and charge selectivity. In so-called "tight" epithelia, where transcellular mechanisms generate steep electrochemical gradients, the paracellular permeability is small and the issue of charge selectivity is of minor importance. However, in "leaky" epithelia, like the small intestine and proximal tubules of the kidney, the paracellular pathway is a major component of overall transport and variations in charge selectivity have a significant impact on the composition of transported fluids (17, 23).

TJs are positioned as continuous circumferential cell-cell contacts at the border of the apical and lateral cell membranes. In freeze-fracture electron micrographs, the intercellular contacts correspond to rows, or fibrils, of transmembrane proteins that make extracellular adhesive contacts with rows from adjacent cells to seal the intercellular space. Claudins, small (~23 kDa) tetraspan proteins with two extracellular domains, are likely the critical structural proteins in the fibrils (5, 7).

Circumstantial evidence supports the idea that claudins create the variable properties of the TJs. The 20 claudin genes in mammals show developmental and tissue-specific expression patterns (18, 21).

Experimentally increasing the levels of different claudins in cultured epithelial models has been reported to either increase or decrease electrical resistance (6, 10, 12). This was interpreted to result from increasing or decreasing cell-to-cell adhesion. However, our recent studies suggest an alternative mechanism. We demonstrated that expression of claudin-4 in cultured Madin-Darby canine kidney (MDCK) cell monolayers increased transmonolayer electrical resistance by selectively decreasing the paracellular permeability for Na⁺ (P_Na), whereas P_Cl and the flux of a noncharged solute were unchanged (22). This finding implies that claudin-4 altered charge discrimination without altering the diffusion of nonelectrolyte solutes. However, these studies did not prove that claudin-4 was directly responsible. This led us, in the present study, to test whether claudins might behave like paracellular channels and directly determine ion permeability through the electrostatic charges on their extracellular amino acid residues. The extracellular domains of claudins contain regions of highly conserved residues and intervening positions where the charge can be positive, negative, or neutral. In this study we have focused on variable residues in the larger first extracellular domain (see Fig. 1, A and B).

View larger version (60K): [in this window] [in a new window]   Fig. 1.   Replacing a basic residue, Lys65, with aspartic acid in the first extracellular domain of claudin-4, (K65D), eliminates its ability to discriminate against the paracellular movement of Na+ ions. A: amino acid sequence alignment of the first extracellular domain of wild-type claudin-4 (CLDN-4 WT) and the charge-reversal mutant K65D. Acidic (red) and basic (blue) residues are noted, and charges are depicted above and below the sequences. B: schematic of predicted membrane topology of claudins, noting the cytosolic NH2 (N) and COOH (C) termini and the first (1) and second (2) extracellular domains. C: upon induction, claudin-4 (not shown) and K65D localize to the lateral plasma membrane. Stable MDCK II Tet-Off cell lines transfected with claudin-4 and K65D were grown on Snapwell filters for 4 days, induced or noninduced for transgene expression. Immunofluorescence microscopy was performed for the tight junction (TJ) proteins ZO-1 (a, c) and claudin-4 (b, d) on noninduced (a, b) and induced (c, d) monolayers. Localization of ZO-1 is unchanged (a, c). Endogenous claudin-4 is shown in b and induced K65D in d, detected by a monoclonal antibody directed against an epitope on the cytoplasmic COOH-terminal tail of both proteins. D: expression of wild-type claudin-4 and K65D can be tightly regulated and does not alter the levels of other TJ proteins. Inducible transfected cell lines were plated and grown on filters noninduced () or induced (+) for transgene expression as in C. Whole cell lysates were analyzed for claudin-4, ZO-1, occludin, and claudin-2 by immunoblotting (22). Endogenous claudin-4 in the noninduced cells is not visible at this exposure. E: K65D mutation eliminates the paracellular discrimination against Na+ by claudin-4. Dilution potentials were compared between monolayers that were noninduced (open bars) and induced (solid bars) for the transgene as in C. Induction of claudin-4 decreased the dilution potential from 8.4 ± 0.4 to 4.2 ± 0.9 mV (P = 0.007), revealing an increased discrimination against Na+ relative to the control monolayer (n = 4 clones). The dilution potentials before and after induction of K65D were 7.5 ± 0.5 to 6.2 ± 0.8 mV (P = not significant; n = 6 clones). The transmonolayer resistance increased significantly upon expression of both claudin-4 (40.9 ± 3.2 to 114.0 ± 16.7  · cm2, P = 0.0007) and K65D (35.1 ± 0.7 to 52.3 ± 5.2  · cm2, P = 0.0034). * P < 0.05. Significance was calculated using Student's t-test. Error bars represent SE.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Plasmid constructs and cell lines. Claudin-4^K65D (K65D) was generated using the Stratagene (La Jolla, CA) Quik-Change site-directed mutagenesis kit. Full-length human claudin-15 was amplified by PCR with the use of human kidney cDNA (Quick-clone cDNA; Clontech Laboratories, Palo Alto, CA) as a template and primers 53988 (5'-CCCACCATGTCGATGGCTG-3') and 53989 (5'-CAGAGCTGCTACACGTAGGC-3'). The amplified product was cloned into the TOPO-TA vector (InVitrogen, San Diego, CA) and subcloned into the pTRE vector (Clontech Laboratories). Claudin-15 mutations were made using the Quik-Change site-directed mutagenesis kit (Stratagene). All wild-type and mutant claudin constructs were verified by DNA sequencing in both directions. Clonal cell lines of MDCK II Tet-Off cells (Clontech Laboratories) were derived by standard transfection and selection techniques; regulated expression of the transgene products was accomplished by varying doxycycline levels in the culture media as previously described (22).

Immunoblots and immunofluorescence. Anti-human claudin-4 mouse monoclonal antibody (Zymed, South San Francisco, CA) was used for immunoblots at a dilution of 1:4,000. Immunofluorescence for claudin-4 and K65D was performed using established protocols (22). Anti-human claudin-15 mouse monoclonal antibody (Zymed) was used for immunoblots at a dilution of 1:2,000 and for immunofluorescence on 1.0% paraformaldehyde-fixed cells at a 1:100 dilution. Immunoblots and immunofluorescence for ZO-1, occludin, and claudin-2 were performed using established protocols (22). Autofluorograms were scanned (Afga Duoscan T1200; Agfa-Gevaert, Mortsel, Belgium), and the integrated optical densities of constant areas at a specific molecular mass range were determined using Gel-Pro Analyzer (Media Cybernetics, Silver Spring, MD). Local area background-integrated optical density was subtracted from each band.

Freeze-fracture electron microscopy. Freeze-fracture electron microscopy was carried out using established protocols (13).

Electrophysiology. Electrophysiological characterization of MDCK II monolayers was carried out according to published methods (22). Stable MDCK II Tet-Off cell lines (Clontech Laboratories) transfected with wild-type or mutant claudins were grown on Snapwell filters (Costar; Corning Life Sciences, Acton, MA) for 4 days, induced without doxycycline or noninduced with doxycycline (50 ng/ml). Protein expression is maximal by day 2. Transmonolayer resistance was measured by using a modified Ussing chamber with a microcomputer-controlled voltage-current clamp (Harvard Apparatus, Holliston, MA) with buffer A (120 mM NaCl, 10 mM HEPES, pH 7.4, 5 mM KCl, 10 mM NaHCO₃, 1.2 mM CaCl₂, and 1 mM MgSO₄) in the apical and basolateral chambers. Transmonolayer dilution potentials were measured upon dilution of the apical chamber (buffer B: 60 mM NaCl, 120 mM mannitol, 10 mM HEPES, pH 7.4, 5 mM KCl, 10 mM NaHCO₃, 1.2 mM CaCl₂, and 1 mM MgSO₄) relative to the basolateral chamber (buffer A) (22). Dilution potentials were immediately stable and repeatedly measured (every 6 s) for at least 30 s after buffer A had been replaced with buffer B in the apical chamber. Voltage and current electrodes consisted of a Ag-AgCl wire in 3 M KCl saturated with AgCl housed in a glass barrel with a microporous ceramic tip (Harvard Apparatus). Liquid junction potentials were calculated by using the Henderson diffusion equation for univalent ions (11) under dilution potential conditions and were factored into the reported dilution potentials.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A charge-reversing mutation of claudin-4, K65D, eliminates its ability to discriminate against Na+. Claudin-4 contains a single basic residue, Lys⁶⁵, in a charge-variable region of the first extracellular domain (Fig. 1, A and B). To test whether electrostatic repulsion from Lys⁶⁵ of claudin-4 discriminates against the paracellular movement of Na⁺ (22), we reversed the charge at this position by replacement with aspartic acid (K65D) using site-specific mutagenesis (Fig. 1A). The K65D mutant was expressed in stably transfected clones of MDCK cells under a tightly regulated doxycycline-repressible promoter (1). Properties of monolayers induced to express claudin-4 or the K65D mutant were compared with noninduced controls of the same clone; at least three clones were examined for each example.

Claudin-15 creates Na+-selective paracellular channels. To test the generality of the hypothesis that charged residues on the extracellular domains of claudins establish paracellular charge selectivity, we performed charge-reversing mutations in the opposite direction, from acidic to basic residues, and on a different claudin. Claudin-15 was selected for study because it has three acidic residues at positions that are typically variable among the claudin family members. Furthermore, their combined reversal creates a first extracellular domain with only positive charges (Fig. 2A). We reasoned that if the electrostatic channel model were correct, this extreme case might reverse the junction selectivity from a preference for Na⁺ to a preference for Cl ions.

View larger version (91K): [in this window] [in a new window]   Fig. 2.   Replacing acidic (red) with basic (blue) residues on the first extracellular domain of claudin-15 reverses the paracellular selectivity from Na+ to Cl ions. A: amino acid sequence alignment of the first extracellular domain of claudin-15 and charge-reversal mutants: m1 (E46K); m2 (D55R); m3 (E64K); m1,2 (E46K, D55R); and m1,2,3 (E46K, D55R, E64K). The charges on wild-type claudin-15 and m1,2,3 are noted above and below the sequences. B: expression of wild-type (WT) claudin-15 and mutants is inducible. Densitometric analysis revealed a decrease in the level of occludin between 45 and 65% upon induction of claudin-15 and all mutants. Immunoblots were performed as in Fig. 1D. C: claudin-15 and mutants localize to the apicolateral plasma membrane upon expression (m1,2,3 shown). Inducible transfected cell lines were plated and grown on filters as in Fig 1C. Immunofluorescence microscopy for ZO-1 (a, c) and claudin-15 m1,2,3 (b, d) was performed on cells noninduced (a, b) or induced (c, d) for the expression of m1,2,3. D: claudin-15 and mutants increase the number, complexity, and depth of TJ fibrils. Freeze-fracture electron microscopy was performed on cells noninduced (a) or induced (b) for the expression of m1,2,3. Inducible transfected cell lines were prepared as in Fig. 1C. Freeze-fracture electron microscopy was performed as previously reported (13). Bar, 0.2 µm. E: charge-reversal mutants of claudin-15 reverse paracellular charge selectivity. Dilution potential experiments were performed as in Fig. 1E. Clonal lines are compared with themselves in the noninduced (open bars) and induced state (solid bars); results from 3-5 clonal cell lines are averaged for wild type and each mutant. The dilution potentials before and after induction of claudin-15 and mutants were as follows: claudin-15, 7.8 ± 0.6 to 7.1 ± 0.6 mV; m1, 7.8 ± 0.9 to 6.4 ± 1.1 mV; * m2, 8.1 ± 0.2 to 1.2 ± 0.6 mV; * m3, 8.5 ± 0.4 to 0.9 ± 1.3 mV; * m1,2, 6.4 ± 0.6 to 2.5 ± 0.7 mV; and * m1,2,3, 7.6 ± 0.6 to 7.1 ± 0.7 mV. Potentials of equal magnitude but opposite charge were generated following basal dilutions, confirming that the charge selectivity is paracellular. Transmonolayer resistances for noninduced clones were similar, with a range of 27.2-34.5  · cm2. A significant increase in transmonolayer resistance was observed upon induction of claudin-15 and all mutants as follows: * claudin-15, 60.6 ± 4.4  · cm2; * m1, 53.8 ± 12.0  · cm2; * m2, 105.5 ± 21.3  · cm2; * m3, 56.7 ± 1.6  · cm2; * m1,2, 50.0 ± 3.9  · cm2; and * m1,2,3, 41.0 ± 3.8  · cm2. * P < 0.05 (Student's t-test). Error bars represent SE.

Charge-reversing mutations on the first extracellular domain of claudin-15 create Cl-selective paracellular channels. To determine whether all three acidic residues in the first extracellular domain of claudin-15 influence paracellular charge selectivity, we replaced them with basic residues, singly and in combination (Fig. 2A). Immunoblot analysis demonstrated that expression of all mutant proteins could be tightly regulated and that, except for occludin, their induction did not alter levels of other TJ proteins (Fig. 2B). Freeze-fracture electron microscopy performed on a subset of the mutants revealed an increase in TJ fibril complexity (Fig. 2D) in a pattern that was indistinguishable from that of cells expressing exogenous wild-type claudin-15. Immunofluorescence microscopy of all mutants revealed localization similar to that seen with wild-type claudin-15 (Fig. 2C). Dilution potentials revealed that reversing the charge at the first acidic position, m1 (E46K), had no effect on paracellular charge selectivity. In contrast, mutation of either position m2 (D55R) or m3 (E64K) resulted in a significant decrease in the ratio of P_Na to P_Cl (Fig. 2E). Introducing a positive charge at position m2 actually reversed the overall permeability ratio to favor Cl ions. We conclude that some, but not all, charged positions in claudin-15 influence paracellular charge selectivity.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The results presented here provide the first conclusive evidence that claudins directly regulate charge selectivity of the paracellular pathway. Their extracellular sequences share a core of conserved residues and presumably a similar tertiary structure. We speculate that the extracellular adhesive contacts between claudins bring them together in a manner that creates aqueous channels lined by the variably charged positions. This charged-channel model is consistent with the observation that TJs in all epithelia have a relatively similar size discrimination for uncharged solutes but have variable charge selectivity and overall electrical resistance (17). A similar model was proposed almost four decades ago based on elegant physiological experiments (25), but the physical basis has remained unknown.

The paracellular pathway has been characterized as having large aqueous spaces and unique charge selectivities. Unlike transmembrane ion channels, the paracellular pathway shows selectivity for net charge rather than specific ions (3, 14). Parallels to the paracellular pathway can be found in the properties of gap junctions, where connexons provide large aqueous cell-to-cell channels and show moderate charge and solute selectivity (16). Similar to our work on claudins, an extracellular domain of connexins with charged residues has been characterized to influence gap junction charge selectivity (20).

We chose to study the larger first extracellular domain of claudins because it has greater numbers and variability of charges than the second domain. Furthermore, consistent with a role for charge in the first extracellular domain influencing charge selectivity, indirect published evidence suggests that claudin-16, with 10 negative residues on its first extracellular domain, forms paracellular channels selective for divalent cations (19). Studies are currently in progress to characterize any contribution of the second extracellular domain in paracellular charge selectivity.

We speculate that paracellular ionic selectivities of different epithelia are determined by varying combinations and levels of different claudins. It is already known that claudins show unique expression patterns and that some TJs contain several different claudins (4, 8, 15, 18, 19). Physiological or pathological alterations in claudins are expected to have a significant impact on epithelial barrier properties. These could be manifest as changes in transport, antigen and pathogen entry, or, conceivably, tissue morphogenesis. Indeed, two recent reports of human diseases resulting from mutations in claudin-14 and 16 and the phenotype of a mouse knockout of claudin-11 (9, 19, 24) can be rationalized as defects in paracellular ionic charge selectivity. Understanding and manipulating the molecular basis of selectivity should also have therapeutic applications for enhancing the level and location of drug delivery.