The tight junction protein claudin-10 is known to exist in two isoforms, resulting from two alternative exons, 1a and 1b (Cldn10a, Cldn10b). Here, we identified and characterized another four claudin-10 splice variants in mouse and human. One (Cldn10a_v1) results from an alternative splice donor site, causing a deletion of the last 57 nucleotides of exon 1a. For each of these three variants one further splice variant was identified (Cldn10a_v2, Cldn10a_v3, Cldn10b_v1), lacking exon 4. When transfected into MDCK cells, Cldn10a, Cldn10a_v1 and Cldn10b were inserted into the tight junction, whereas isoforms of splice variants lacking exon 4 were retained in the endoplasmic reticulum. Cldn10a transfection into MDCK cells confirmed the previously described increase in paracellular anion permeability. Cldn10a_v1 transfection had no direct effect, but modulated Cldn10a-induced organic anion permeability. At variance with previous reports in MDCK-II cells, transfection of high-resistance MDCK-C7 cells with Cldn10b dramatically decreased transepithelial resistance, increased cation permeability, and changed monovalent cation selectivity from Eisenman sequence IV to X, indicating the presence of a high field-strength binding site that almost completely removes the hydration shell of the permeating cations. The extent of all these effects strongly depended on the endogenous claudins of the transfected cells.

Claudins are a family of proteins located in the tight junction (TJ) of epithelial cells. They are predicted to span the membrane four times (Fig. 1). As both the N-terminus and C-terminus reside in the cytoplasm, there are two extracellular loops through which claudins of neighbouring cells are assumed to interact, as they have been demonstrated to exhibit a barrier function and to regulate paracellular permeability (Wen et al., 2004; Hou et al., 2005). In tubular epithelia, a segment-specific distribution of different claudins is observed (Kiuchi-Saishin et al., 2002; Hashizume et al., 2004; Guan et al., 2005; Inai et al., 2005; Holmes et al., 2006; Ohta et al., 2006). Amongst these is claudin-10 which is differentially expressed along the length of many tubular epithelia including the nephron in the kidney.

In mouse submandibular gland, immunohistochemical staining for claudin-10 demonstrated localization within the TJ mainly in the terminal tubules, whereas it was hardly detectable in the ducts (Hashizume et al., 2004). In the epididymis of 10-week-old rats, claudin-10 was found in the TJs of the efferent duct and the initial segment, but not in the caput, corpus or cauda, although in cells of the two latter segments some extra-junctional staining was observed [basal and cytoplasmic, respectively (Guan et al., 2005)].

In mouse and rat gut, claudin-10 was found in all segments (Inai et al., 2005), but with highest expression levels near the ileocecal region (Holmes et al., 2006). Claudin-10 was present in the small intestine in all TJs along the villus-crypt axis, whereas it was restricted to the crypts and absent from the surface epithelium in the large intestine (Inai et al., 2005; Holmes et al., 2006).

In the kidney, claudin-10 distribution has been reported to differ between mouse and cattle (Kiuchi-Saishin et al., 2002; Ohta et al., 2006). In cattle, immunohistochemical staining revealed the presence of claudin-10 in the proximal tubule and the thick ascending limb of Henle's loop (Ohta et al., 2006), whereas, in mouse kidney, staining was originally reported to be restricted to the thin descending limb and the thick ascending limb of Henle's loop (Kiuchi-Saishin et al., 2002). In contrast to this finding, Van Itallie et al. (Van Itallie et al., 2006) found claudin-10 to be present in all parts of the nephron, except for the glomerulus.

Furthermore, Van Itallie et al. (Van Itallie et al., 2006) reported the existence of two claudin-10 splice variants (Cldn10a and Cldn10b), giving rise to two claudin-10 isoforms, claudin-10a and claudin-10b, respectively. Claudin-10a appeared to be restricted to the kidney, whereas claudin-10b was found in all tissues investigated. Within the kidney, they found claudin-10a to be preferentially expressed in the cortex and claudin-10b in the medulla. The two variants differ only in their first exon. In mouse and human claudin-10, the coding sequence of exon 1a encompasses 214 bp, whereas exon 1b consists of 220 bp, rendering claudin-10b two amino acids longer than claudin-10a. Although very similar in length, the two isoforms differ greatly with respect to the electric charge within their first extracellular loop (Fig. 1). Claudin-10a contains seven positively charged and two (mouse) or one (human) negatively charged amino acids within the first extracellular loop. By contrast, there are four positive and five negative charges in the first extracellular loop of claudin-10b. This observation gave rise to the assumption that expression of these two isoforms might have different effects on the ion permeability of TJs. Subsequent electrophysiological measurements revealed that, when expressed in MDCK-II cells, claudin-10a increased Cl permeability whereas Na+ permeability decreased. Expressing claudin-10b had no effect on the permeability of either ion. When expressed in LLC-PK1 cells, expression of claudin-10a again caused an increase in Cl permeability whereas expression of claudin-10b caused an increase in Na+ permeability. By specific site-directed mutation of single charged amino acids, the observed increase in Cl permeability caused by claudin-10a was linked to the positive charges of amino acid R33 [referred to as R32 by Van Itallie et al. (Van Itallie et al., 2006)] and R62 [referred to as either R59 or R61 by Van Itallie et al. (Van Itallie et al., 2006)].

In the present study, four further alternative Cldn10 transcripts were identified in mouse and human. When transfected into MDCK-II and MDCK-C7 cells (sub-clones of Madin-Darby canine kidney cells differing in their claudin-2 expression) claudin-10a, claudin-10b and one isoform (claudin-10a_i1) encoded by the additional splice variants were localized in the tight junction, whereas the other three were retained in the endoplasmic reticulum. Physiological function of the three tight junctional claudin-10 isoforms was further investigated in MDCK-C7 cells. Because of its high transepithelial resistance (≥1000 Ω·cm2), this cell line has previously been found advantageous for the detection of paracellular pore formation (Amasheh et al., 2002). Furthermore, MDCK-C7 cell layers are only moderately cation selective (PNa/PCl ∼1.5) and therefore may allow disclosure of increases in selectivity for cations as well as anions. Functional analyses demonstrate that claudin-10a and claudin-10b change the interaction of the paracellular pore with the hydration shell of the permeating anions and cations, respectively, rather than conveying a general increase in anion or cation permeability. Claudin-10a_i1 appears to further modulate claudin-10a function.

Comparison of claudin-10 effects observed in MDCK-II and MDCK-C7 cells indicates that the extent of these effects strongly depends on the endogenous claudins of the cells used as expression system.

Identification of Cldn10a, Cldn10a_v1 and Cldn10b in mouse and human

Two Cldn10 splice variants were found in the human and mouse NCBI database (Cldn10a, Cldn10b), only differing in their first exon. To investigate the expression of the two Cldn10 variants in mouse kidney, a PCR with variant specific forward primers was performed on mouse kidney cDNA. Expression of both variants was found, but also an additional lower band for Cldn10a was detected (Fig. 2A). After cloning and sequencing, it was found that the lower band had a deletion of the last 57 nucleotides of exon 1a, resulting from an alternative splice donor site in this exon, which causes an in frame deletion of 19 amino acids (Cldn10a_v1). Interestingly, this deletion is located in the predicted first extracellular loop and would cause a deletion of the second cysteine of the highly conserved W-GLW-C-C motif (Van Itallie and Anderson, 2006) and, in mouse, a deletion of six of the nine charged amino acids of this loop (in human: five of the eight charged amino acids; Fig. 1A, Fig. 3) including R62 investigated by Van Itallie et al. (Van Itallie et al., 2006).

Further investigation of the database revealed the same 57 nucleotide deletion in spliced ESTs from Trichosurus vulpecula (brushtail possum) kidney (DY598763) and dog kidney (CO688322), stressing the potential relevance of this new splice variant. In addition, one human spliced EST was found with the same deletion (N41613), but this EST came from a placenta cDNA library. To prove the existence of this splice variant in human kidney, a PCR was successfully performed using a forward primer that bridges over this deletion and extends for six nucleotides into exon 2 (this primer did not give a product under the same conditions when using pure Cldn10a as template).

Three Cldn10 isoforms lack exon 4

Upon detailed analysis of the Cldn10 ESTs in the database, another putative messenger RNA from mouse kidney was identified that lacks exon 4 (BC021770), which, if translated, would result in an in frame deletion of 36 amino acids. In an attempt to identify this variant as well, a reverse primer was designed that bridges exon 4 and extends for four nucleotides into exon 3 (Figs 1 and 3). It was confirmed by PCR on mouse kidney cDNA that such a variant exists and it was also noticed that a lower band appeared. Upon cloning and sequencing this shorter product, it was found that, in addition to exon 4, it lacked the same 57 nucleotides from the first exon as mentioned before.

The three splice variants lacking exon 4 will be referred to as Cldn10a_v2, Cldn10a_v3 and Cldn10b_v1, respectively (see also Table 1 for nomenclature). As exon 4 has been predicted to contain the complete fourth transmembrane region, the C-termini of the resulting proteins may come to lie on the extracellular side of the membrane (note, however, that TMHMM Server v 2.0 assigns an alternative transmembrane region to these isoforms).

Tissue distribution of Cldn10 expression

To study tissue-specific expression of the identified Cldn10 variants, a semi-quantitative PCR was performed on two mouse tissue cDNA panels. It was found that, whereas Cldn10a expression was only observed in kidney and uterus, Cldn10b was expressed in all tissues tested, with low expression in liver and highest expression in kidney. A similar expression pattern for this isoform was also observed by Van Itallie et al. (Van Itallie et al., 2006). Cldn10a_v1, which was not described by Van Itallie et al. (Van Itallie et al., 2006), is also expressed in kidney and uterus (Fig. 2B).

The approximately 100 base pair shorter band that was also amplified in addition to Cldn10b (see arrows in Fig. 2B), although too scarce to be sequenced or cloned directly, could point to Cldn10b lacking exon 4. In fact an RT-PCR on mouse total kidney RNA with a forward primer in exon 1b and the reverse primer bridging exon 4, with subsequent sequencing, confirmed the existence of such a splice variant (c.f. Fig. 3).

Cldn10 expression along the nephron

Since no splice-variant-specific antibodies to claudin-10 exist, we carried out RT-PCRs for Cldn10a and Cldn10b on dissected tubule segments to look for differential expression along the nephron. It was found that Cldn10a is expressed in the proximal convoluted tubule (PCT), medullary thick ascending limb of Henle's loop (mTAL) and cortical collecting duct (CCD), and possibly in the outer (OMCD) but not inner medullary collecting duct (IMCD), whereas Cldn10b is expressed in mTAL, OMCD and IMCD, but not in PCT or CCD. Cldn10a_v1 expression was always found in conjunction with that of Cldn10a (Fig. 2C).

Transient Cldn10 expression in MDCK-II cells: subcellular localization

To determine the subcellular localization of each Cldn10 isoform, N-terminal HA epitope-tagged constructs were generated and transiently expressed in MDCK-II cells. Cells were coimmunolabeled with antibodies to HA to detect the different Cldn10 isoforms and either the TJ marker ZO-1 or the ER marker calreticulin, and analyzed by confocal laser scanning microscopy. For all six mouse Cldn10 isoforms, all isoforms translated from exon-4-containing constructs were found to be localized in the plasma membrane, showing partial colocalization with ZO-1, indicating integration into TJs (Fig. 4A). By contrast, the isoforms translated from constructs lacking exon 4 colocalized with calreticulin and/or were present in vesicular structures (Fig. 4B), suggesting that they failed to reach or be retained at the plasma membrane.

Cldn10 binds ZO-1

Claudins have been demonstrated to be anchored in the TJ through interaction of their C-terminal PDZ-binding motif to PDZ-domain-containing proteins such as ZO-1 (Itoh et al., 1999; Müller et al., 2003). To test if Cldn10 binds ZO-1, MDCK-II cells were transfected with the three Cldn10 variants containing exon 4, or with mutants in which the C-terminal PDZ-binding motif was mutated (ΔP). Although all three mutant variants still appeared to be located in the plasma membrane, the colocalization with ZO-1 was strongly reduced (yellow in Fig. 4C; supplementary material Fig. S1). However, at increased detector gain it could be demonstrated that some of the mutated protein still resided within the TJ (yellow in Fig. 4D).

Pull-down experiments confirmed an interaction between the ZO-1 PDZ domains and the C-terminus of Cldn10, which is identical in the three isoforms. Mutation of the PDZ-binding motif abolished the interaction of the PDZ domain with the ΔP C-terminus (Fig. 4E).

Stable expression of Cldn10a, Cldn10a_v1 and Cldn10b in MDCK-C7 and MDCK-II cells

High resistance MDCK-C7 cells were stably transfected with mouse and human Cldn10a and Cldn10b and with mouse Cldn10a_v1. For comparison with published data (Van Itallie et al., 2006), mouse Cldn10a and Cldn10a_v1 were additionally transfected into low resistance MDCK-II cells.

In agreement with the transiently expressed proteins (see above), all isoforms were located in the TJ as judged from colocalization with the TJ marker occludin identified by confocal laser-scanning microscopy (see supplementary material Fig. S1). Since isoforms translated from constructs lacking exon 4 were not detected on the cell surface, they were not analyzed further.

Dilution potentials and biionic potentials for cell monolayers expressing Cldn10 variants

Ussing chamber experiments were carried out to analyze transepithelial resistance and ion permeabilities of cell layers and these results are summarized in Fig. 5A (see also supplementary material Fig. S2 for alternative clones). These experiments demonstrated that MDCK-C7 cell layers transfected with mouse or human Cldn10b had a strongly reduced transepithelial resistance accompanied by an increased ratio in paracellular permeability for Na+ over Cl (PNa/PCl). The increase in PNa/PCl was predominantly caused by an increase in PNa. Transfection with mouse or human Cldn10a or mouse Cldn10a_v1 did not affect PNa/PCl (Fig. 5B; supplementary material Fig. S2).

In contrast to the findings of Van Itallie et al. (Van Itallie et al., 2006) obtained in low resistance MDCK-II cells, no effect of Cldn10a on PCl was observed in high resistance MDCK-C7 cells, indicating that the positive charges in position 33 and 62 may be necessary but not sufficient for the reported increase in Cl permeability. Therefore, the experiments were repeated in MDCK-II cells and PCl/PNa determined in dilution potential measurements. In agreement with the findings of Van Itallie et al. (Van Itallie et al., 2006), Cldn10a induced an increase in PCl/PNa, an effect that was predominantly due to an increase in PCl and that was not observed in cells transfected with Cldn10a_v1. As permeability to NO3 was even greater than to Cl, NO3 permeability was also investigated in MDCK-C7 cells. In these cells, a small increase in PNO3/PNa was observed upon transfection with Cldn10a, but not with Cldn10a_v1. Conversely, permeability to the larger pyruvate was reduced in Cldn10a-transfected cells, compared with control cells and cells transfected with Cldn10a_v1. Transfection with Cldn10b caused a decrease in PNO3/PNa, PCl/PNa and PPyr/PNa, which is fully explained by the increase in PNa demonstrated above (Fig. 6).

When comparing relative permeabilities for various monovalent cations in MDCK-C7 cells, these permeabilities followed Eisenman sequence IV for alkali metals: K+>Rb+>Cs+=Na+>Li+ (Fig. 7A, inset) in control cell monolayers (Eisenman, 1962). Values found in Cldn10a (human or mouse)-transfected or Cldn10a_v1 (mouse)-transfected cell monolayers did not differ significantly from the values found in control cells.

In sharp contrast, Cldn10b (mouse or human)-transfected cell monolayers showed a preference for Na+>Li+>K+>Rb+>Cs+, corresponding to Eisenman sequence X. Interestingly, not only mutation of the PDZ-binding motif but also the N-terminal fusion of YFP to the human Cldn10b abolished the decrease in transepithelial resistance, the increase in PNa (a slight increase in PNa was still observed in the ΔP mutant) and the change in Eisenman sequence (Fig. 7A), although YFP-CLDN10b appeared to be located in the TJ (supplementary material Fig. S1).

Permeability to divalent cations was also generally increased in Cldn10b (mouse or human)-transfected cell monolayers, but unchanged in Cldn10a- and Cldn10a_v-transfected cell monolayers (Fig. 7B). Differences between permeabilities of the divalent cations investigated (Mg2+, Ca2+, Sr2+, Ba2+) were too small to attribute them to specific `Sherry sequences' for divalent ions (Diamond and Wright, 1969).

Co-culture experiments

As shown above, Cldn10a and Cldn10a_v1 are always expressed together. It can therefore be postulated that both isoforms, claudin-10a and claudin-10a_i1, are integrated into the same TJ, where they will interact both within the same membrane (cis-interaction) and with the extracellular loops of the corresponding isoforms from neighbouring cells (trans-interaction). Furthermore, Cldn10a and Cldn10b were both detected in micro-dissected mTAL material, although isoform-specific antibodies would be required to determine their expression at the cellular level to determine if they may be in direct contact. To explore the effects of such interactions in trans, co-culture experiments were carried out. Equal amounts of suspensions of the respective transfected MDCK-C7 cells were mixed and seeded onto filter supports to be used in Ussing chamber experiments.

When claudin-10a- and claudin-10a_i1-expressing cells were co-cultured, electrical resistance was increased, beyond control values (Fig. 5A). Permeability to monovalent cations was unchanged (compared with controls, claudin-10a and claudin-10a_i1), whereas permeability to divalent cations was decreased (Fig. 7). Cell layers retained the increased permeability for NO 3 observed in claudin-10a-expressing cell layers, however, the decrease in pyruvate permeability was lost (Fig. 6B). Thus, claudin10a_i1 appears to modulate the permeability of the TJ to large organic anions. In additions, these finding support trans-interactions between the respective isoforms.

Co-cultures of cells expressing claudin-10a and claudin-10b still exhibited the increase in cation permeability observed in claudin-10b-expressing cells, however, the permeability sequence almost reversed to Cs+>Rb+>K+>Na+>Li+, which corresponds to Eisenman sequence I (Fig. 7A). The combined effect of these two claudins therefore reduces interaction between ion and pore below that observed in control cells.

Claudin-10 isoforms

Two major isoforms of claudin-10, Cldn10a and Cldn10b, were previously described by Van Itallie et al. (Van Itallie et al., 2006). These isoforms differ in the amino acids encoded by the first of the five claudin-10 exons. This results in differences in the first extracellular loops of the protein isoforms and, most remarkably in the net charge of the first extracellular loop (CLDN10a, +5 in mouse, +6 in human; CLDN10b, –1 in mouse and human; see Fig. 1). CLDN10a_i1, a third isoform identified in this study, also appears to reside within the TJ. This variant lacks 19 amino acids within the first extracellular loop, including four of the positively and, in mouse, one of the negatively charged amino acids of the first loop of CLDN10a, resulting in a net charge of +2 in mouse and human. One of the positive charges lacking (R62) is one of the two positive charges found to be essential for CLDN10a function as a paracellular anion channel (Van Itallie et al., 2006). Furthermore, CLDN10a_i1 lacks the second of two cysteines (C61 in CLDN10a) highly conserved within the claudin family. Mutation of this cysteine in Cldn5 completely abolished its effect on paracellular tightness (Wen et al., 2004). However, it cannot be ruled out that an additional cysteine (C77) present in the first loop could substitute for C61.

For each of the messenger RNAs encoding these three isoforms there also exists an alternatively spliced transcript lacking exon 4 which, when overexpressed in MDCK-II cells, was not incorporated into the TJ and was therefore not investigated electrophysiologically. Whether the isoforms lacking exon 4 are of any functional significance remains to be determined, paricularly given their apparently low expression levels as compared with Cldn10a and Cldn10a_v1. One possibility is that cis-oligomerization between Cldn10a and Cldn10a_i1 and the ER-retained variants could serve to regulate surface expression levels of Cldn10a and Cldn10a_i1.

Tissue distribution

With respect to expression of the six Cldn10 splice variants in various mouse tissues (brain, eye, spleen, lymph node, thymus, lung, liver, stomach, kidney, testis, prostate, placenta, uterus, skeletal muscle, heart, smooth muscle), we observed ubiquitous expression of the two transcripts containing exon 1b (Cldn10b, Cldn10b_v1), thus confirming the findings for Cldn10b already described (Van Itallie et al., 2006). All four transcripts containing exon 1a (Cldn10a, Cldn10a_v1, Cldn10a_v2, Cldn10a_v3) were found in the kidney and the uterus. Within the kidney, PCT and CCD expressed only transcripts containing exon 1a, OMCD and IMCD predominantly (OMCD) or only (IMCD) transcripts containing exon 1b, whereas mTAL expressed all variants.

The expression pattern of claudin-10 described by Kiuchi-Saishin et al. (Kiuchi-Saishin et al., 2002) is at variance with that found by Van Itallie et al. (Van Itallie et al., 2006). Kiuchi-Saishin et al. found claudin-10 exclusively in the proximal tubule and the TAL (Kiuchi-Saishin et al., 2002). Similar to our data, Van Itallie et al. (Van Itallie et al., 2006) found weak claudin-10 expression in PCT and DCT, and high expression in the macula densa, the cortical and medullary TAL and both CCD and IMCD, with a Cldn10a preferentially expressed in the cortex whereas Cldn10b was more abundant in the medulla. Thus, the findings in the present study seem to be most similar to those of Van Itallie et al. (Van Itallie et al., 2006).

Functional characterization of claudin-10a, claudin-10a_v1 and claudin-10b

When expressed in low resistance MDCK-II or high resistance MDCK-C7 cells, CLDN10a, CLDN10a_i1 and CLDN10b were located in the TJ of the transfected cell layers, as judged from colocalization with the TJ marker protein occludin by confocal laser scanning microscopy. Therefore these cell layers were further investigated electrophysiologically. In agreement with the data published by Van Itallie et al. (Van Itallie et al., 2006), Cldn10a-transfected low resistance MDCK-II cell layers showed an increase in paracellular anion (Cl, NO 3) permeability, an effect not observed in Cldn10a_v1-transfected cells. By contrast, when high-resistance MDCK-C7 cells were transfected with Cldn10a, no effect on PCl was observed. Cldn10a transfected cell layers, however, showed an increased PNO3 and a decreased PPyr, indicating a stronger interaction between the TJ pore and the permeating anion. Neither effect was present in Cldn10a_v1- or in Cldn10b-transfected cell layers.

Of the three claudin-10 isoforms investigated in the present study, only claudin-10a_i1 appeared to lack a specific function with respect to paracellular ion permeabilities in MDCK-C7 cells. However, when the extracellular loops of claudin-10a and claudin-10a_i1 were brought into contact in trans by co-culturing the respective transfected cells, the decrease in paracellular pyruvate permeability was lost whereas the increase in NO 3 permeability remained unchanged. We therefore conclude that claudin-10a_i1 may specifically modulate the function of claudin-10a with respect to organic anions. As both isoforms are always coexpressed (see Fig. 2), it is tempting to speculate that organic anion permeability is regulated by relative amounts of claudin-10a and claudin-10a_i1 within the TJ, especially in the PCT and possibly also in the uterus. Validation of this hypothesis will require isoform-specific antibodies to analyze expression at the single cell level.

In Cldn10b-transfected MDCK-C7 cell layers, electrical resistance was greatly reduced, because of a dramatic increase in cation permeability. Furthermore, the permeability sequence for monovalent cations changed from Eisenman sequence IV in control, Cldn10a- or Cldn10a_v1-transfected MDCK-C7 cells to Eisenman sequence X in Cldn10b-transfected MDCK-C7 cells, indicating a change in pore–ion interaction from a low field strength to a high field strength binding site in the presence of Cldn10b. This increase in field strength causes partial removal of the hydration shell of the permeating ion and thus sorts ions according to their unhydrated diameter. A low transepithelial resistance (in the order of 30 Ω·cm2) and a high Eisenman sequence (VIII or IX) has previously been determined by Greger (Greger, 1981) in isolated TAL segments of the rabbit kidney.

In contrast to all other nephron segments investigated, Cldn10a and Cldn10b were both detected in microdissected mTAL material. As no isoform-specific antibodies are available, it is not clear whether cells expressing these two isoforms are present in distinct regions of the TAL, or whether they are coexpressed throughout the TAL. To mimic the latter possibility, cells expressing the respective isoforms were co-cultured and subjected to Ussing chamber experiments. Strikingly, monovalent cation permeability, although still high, completely reversed its permeability sequence to Eisenman sequence I. This indicates that removal of the hydration shell of these ions is no longer possible in paracellular pores consisting of trans-interacting claudin-10a and claudin-10b. This is in contrast to the conditions found in isolated TAL tubules (Greger, 1981), making it unlikely that such an interaction occurs under physiological conditions, at least in the TAL.

Thus, claudin-10b is most probably the major contributor to the low resistance and strong cation selectivity of the TAL in vivo.

Alterations of the C- and N-terminus

Mutation of the C-terminal PDZ-binding motif greatly reduced the amount of claudin-10 residing in the TJ, as judged from immunocytochemical stainings. This, as expected from analogous experiments on claudin-16 (Kausalya et al., 2006), reduced (CLDN10b) or abolished (CLDN10a) the contribution of the claudins to the electrophysiological properties of the transfected cell monolayer. It is noteworthy that alteration of the N-terminus by the addition of YFP to CLDN10b also completely abolished functionality of the protein, although, as judged from confocal laser scanning microscopy, YFP-CLDN10b was still located in the TJ. One possibility is that the YFP interferes with cis-interactions of neighbouring claudins within the TJ.

Even more dramatic effects were observed for the three splice variants lacking exon 4, which, if translated, would carry an inframe deletion of 36 amino acids. These 36 amino acids contain the complete fourth transmembrane region of the proteins (topology according to UniProt database), which means that the C-termini of these proteins should come to lie on the extracellular side of the membrane. Alternatively, another database (TMHMM Server v 2.0) assigns an alternative transmembrane region for these isoforms. In both cases, the resulting structural changes are envisaged to prevent the C-terminal interactions with PDZ domains.

None of these isoforms appeared to reach the TJ but colocalized with the ER marker calreticulin. Given that these three variants are expressed in vivo, it will be of interest to determine in future experiments, if they possess any regulatory function. For example, cis-dimerization with functional isoforms could regulate surface expression levels of functional claudin-10 isoforms.

Implications for renal physiology

The high PNa caused by the presence of claudin-10b contributes directly to the lumen positive potential within the TAL, because of the formation of a lumen positive dilution potential, especially in the cortical segment of the TAL (Greger, 1981). This increases the driving force for the resorption of other cations, especially Ca2+ and Mg2+. As both PCa and PMg are increased in the presence of claudin-10b, this isoform appears to be of major importance to the resorption of divalent cations. Potential interactions between claudin-10b and the two claudins that have already been linked to Mg2+ resorption, claudin-16 and 19 (Müller et al., 2003; Kausalya et al., 2006; Hou et al., 2008) (for a review, see van de Graaf et al., 2007) are therefore highly interesting and will be subject of future experiments.

When extrapolating Eisenman sequence X observed in Cldn10b-transfected cell layers to estimate the H+ permeability (Eisenman, 1962), it seems very probable that Cldn10b-transfected cell layers are considerably more permeable to H+ than control or Cldn10a-transfected cell layers. Such a paracellular H+ permeability in the TAL has for example been postulated as part of a model for ammonium uptake (Kikeri et al., 1989). Unfortunately H+ permeability could not be investigated directly, as this would involve the application of a pH gradient across the cell layer. Changing the extracellular pH, however, is known to affect the paracellular cation selectivity by titrating the electrical charge (Cereijido et al., 1978) and adhesion (Lim et al., 2008) of TJ proteins, and thus would affect the parameter to be determined in the present case.

Comparison of the findings in MDCK-C7 cells with those in MDCK-II cells underlines the strong influence of the presence of endogenous claudins on the permeability induced by transfected exogenous claudins previously demonstrated by Van Itallie et al. (Van Itallie et al., 2003). The major difference between the endogenous claudins of the two MDCK cell lines is the presence of claudin-2 in MDCK-II but not in MDCK-C7 cells (western blot analyses on claudins-1 to 5 and claudin-7; S.M.K., unpublished results). Claudin-2 is known to act as a paracellular cation pore (Amasheh et al., 2002).

The effect of claudin-10a on anion permeability was enhanced by the presence of claudin-2 in MDCK-II cells, whereas the effect of claudin-10b on cation permeability was masked by claudin-2. As claudin-2 is not present in nephron segments distal to the thin descending limb of Henle's loop (Kiuchi-Saishin et al., 2002), such an interaction should be of no relevance to the function in these distal nephron segments, but may be of importance in the PCT. Future studies will have to address how other endogenous claudins present in MDCK cells may contribute to the observed effects.

Conclusions

In conclusion, six claudin-10 variants have been found in human and mouse. In these variants, the presence (claudin-10a, claudin-10a_v1, claudin-10b) or absence (claudin-10a_v2, claudin-10a_v3, claudin-10b_v1) of exon 4 determines, whether the claudin is localized in the TJ or retained in the endoplasmic reticulum, respectively. The presence of either exon 1a or 1b determines, whether expression of the resulting claudin-10 variants is restricted to kidney and uterus (claudin-10a, claudin-10a_v1, claudin-10a_v2, claudin-10a_v3) or whether expression is ubiquitous (claudin-10b, claudin-10b_v1). Within the kidney, the segments of the nephron investigated in the present study express either exclusively exon 1a-containing variants (PCT, CCD), or exon 1b-containing variants (IMCD), or both (mTAL, OMCD).

Of the three claudin-10 variants localized in the TJ, two (claudin-10a, claudin-10b) had direct effects on the ion selectivity of the paracellular pathway, when overexpressed in MDCK cells. Both claudin-10 variants appear to increase the field-strength of the ion binding site within the paracellular pore, thus favouring the passage of ions with small dehydrated ion radius. In the case of claudin-10a, this effect relates to anions, in claudin-10b, to cations. Claudin-10a_v1 has no direct effect, but modulates claudin-10a effects on organic anion permeability.

Comparison of results in different expression systems shows that the magnitude of claudin-10 effects on paracellular ion permeability depends strongly on the endogenous claudins present in these cells. This is of great importance for the ubiquitously distributed claudin-10b, which thus qualifies as a potent cation channel in tissues not expressing claudin-2.

Cldn10 variant and isoform nomenclatures are given in Table 1.

RT-PCR and cloning of cldn10 variants

Total RNA was extracted from mouse and human kidney using TRIzol reagent (Invitrogen, Carlsbad, CA), according to the protocol described by the manufacturer. Kidney cDNA was synthesized from 2 μg total RNA using Omniscript Reverse Transcriptase (Qiagen, Hilden, Germany) in a 20 μl reaction volume.

PCRs were performed with Phusion DNA polymerase (Finnzymes, Espoo, Finland) unless otherwise specified. PCR products were cloned into pGEM-T easy according to the manufacturer's instructions (Promega, Madison, WI) and sequenced.

There are some amino acid differences between published mouse Cldn10 sequences. Compared with the mouse genomic database (NCBI build 37, July 2007), our clones have one non-silent nucleotide difference, resulting in one amino acid difference T217A or T219A for variants a and b, respectively. The sequence resulting in the Thr to Ala amino acid change was found in the majority of Cldn10 messengers in the NCBI nucleotide database. Therefore, we have assumed this to be the most prevalent sequence.

Molecular cloning of human CLDN10a and CLDN10b cDNA was performed by PCR with respective forward primers (5′-GCGGCGCGACATGTCCAGG-3′, 5′-CCGGCGGCATGGCTAGCA-3′) and reverse primer (5′-CGAGCTCTTTTAGACATAAGC-3′) using human kidney cDNA as template. Cycling conditions were 30 seconds at 96°C, 30 seconds at 58°C (CLDN10a) or 60°C (CLDN10b) and 90 seconds at 72°C for 35 cycles with an initial denaturation of 1 minute and final extension of 10 minutes. The resulting PCR products encompassing the complete CLDN10a and CLDN10b coding sequences were cloned into pTOPO (Invitrogen).

The human CLDN10 variant with a deletion of the last 57 nucleotides of exon 1a (CLDN10a_v1) was amplified using forward primer 5′-GATGAACTGCGCAGGTTATATA-3′ and reverse primer 5′-AGATGTGGCCCCGTTGTATG-3′ under the following cycling conditions: 10 seconds at 98°C, 30 seconds at 65°C and 15 seconds at 72°C for 35 cycles with an initial denaturation of 30 seconds and final extension of 10 minutes.

Generation of expression constructs

To obtain expression constructs for each mouse Cldn10 variant, primers were designed to introduce a 5′ KpnI restriction site, a Kozak sequence (underlined in the following sequence) and an in frame N-terminal HA epitope tag (ACCATGGCTTACCCATACGATGTTCCAGATTACGCT) to replace the start ATG and a 3′ XbaI site. After PCR on mouse kidney cDNA, the amplicons were cloned into pGEM-T easy and subcloned into the mammalian expression vector pcDNA3 (Invitrogen, Carlsbad, CA) using the KpnI and XbaI restriction sites.

Expression clones that lack exon 4 were created using a PCR-based strategy: both the region 5′ and 3′ of exon 4, both with 12 base pairs overlap, were amplified separately using the full length constructs as template. These PCR products were then used together as a template in a second PCR. Subsequent subcloning into pcDNA3 was carried out as described above.

To obtain expression constructs for all Cldn10 variants with an inactivating mutation of the PDZ-binding motif, a PCR was performed on the existing expression constructs, using the reverse primer 5′-CGTCTAGATTACGCCGCGGCATTTTTATCAAACTGTTTTG-3′ that changed TyrVal(stop) into AlaAla(stop).

Human CLDN10a and b were similarly subcloned into eukaryotic expression vectors using the 5′ HindIII and 3′ XbaI restriction sites for pFLAG-CMV4 (Sigma Aldrich, St Louis, MO) and for a modified pcDNA3.1 vector (a generous gift from Otlmar Huber, Charité, Berlin), tagging the N-terminus of claudin-10 with yellow fluorescent protein (YFP).

All expression constructs were sequence verified and primer sequences that are not mentioned are available upon request.

Semi-quantitative PCR on multiple tissue mouse cDNA panels

PCRs were performed on MTC mouse Panels I and III from Clontech (Mountain View, CA) using forward primer 5′-TCCAACGAATGGAAAGTGACC-3′ and either reverse primer 5′-TCTCCTTCTCCGCCTTGATAC-3′ to obtain Cldn10a/Cldn10a_v1 or reverse primer 5′-CGTTGTATGTGTAGCCCATTTTTT-3′ to obtain Cldn10a_v2/Cldn10a_v3 and forward primer 5′-TCGCCTTCGTAGTCTCCATC-3′ and reverse primer 5′-TCTCCTTCTCCGCCTTGATAC-3′ to amplify Cldn10b.

Cycling conditions for Cldn10a/Cldn10a_v1 were: 10 seconds at 98°C, 30 seconds at 62°C and 20 seconds at 72°C for 35 cycles, for Cldn10a_v2/Cldn10a_v3: 10 seconds at 98°C, 30 seconds at 68°C and 16 seconds at 72°C and for Cldn10b: 10 seconds at 98°C, 30 seconds at 62°C and 20 seconds at 72°C. Initial denaturations of 30 seconds and final extensions of 10 minutes were included. PCR volumes were 50 μl and 5 μl samples were taken after 25 and 30 cycles.

RT-PCR on dissected mouse nephron segments

Kidneys used for microdissection were taken from 12-week-old male C57BL/6J mice. Tubular segments were isolated and first strand cDNA was made as described previously (Stehberger et al., 2003; Smith et al., 2005). Although PCT (proximal convoluted tubule, mTAL medullary thick ascending limb of Henle's loop), CCD, OMCD and IMCD (cortical, outer and inner medullary collecting duct) were unequivocally identified, glomeruli appeared to be contaminated with macula densa and/or DCT material (judged from the presence of NKCC2 and NCC mRNA) and were therefore omitted from the present analyses.

Samples of the cDNA preparations were checked by TaqMan real-time PCR for their Hprt1 Ct values to estimate relative concentrations. Individual cDNA amounts used in the actual PCR were adjusted accordingly. Used primer pairs and cycling conditions were identical to those described above, except that the number of cycles was now 40. The β-actin controls were amplified using Taq polymerase with forward primer 5′-GGGCTGTATTCCCCTCCATC-3′ and reverse primer 5′-CTCCGGAGTCCATCACAATG-3′ under the following cycling conditions: 10 seconds at 94°C, 30 seconds at 57°C and 25 seconds at 72°C for 36 cycles with an initial denaturation of 10 seconds and final extension of 7 minutes.

Cell culture and transfection

Monolayers of cells of the high resistance MDCK-C7 (Gekle et al., 1994) strain and the low resistance MDCK-II strain were grown in 25 cm2 culture flasks containing MEM-EARLE (PAA, Pasching, Austria), supplemented with 10% (v/v) fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin (PAA). Cells were cultured at 37°C in a humidified 5% CO2 atmosphere.

MDCK-II and MDCK-C7 cells were either transiently or stably transfected with mouse Cldn10 variant constructs using Lipofectamine 2000 (Invitrogen, Gaithersburg, MD) or with human CLDN10 variant constructs employing the Lipofectamine plus method (Gibco BRL, Invitrogen) as previously described by Kausalya et al. (Kausalya et al., 2006), Müller et al. (Müller et al., 2003) and Amasheh et al. (Amasheh et al., 2002), respectively. G418-resistant cell clones were screened for claudin-10 expression by western blot analysis which was carried out as described in detail by Amasheh et al. (Amasheh et al., 2002). Briefly, membrane protein was extracted (membrane lysis buffer: 20 mM Tris, 5 mM MgCl2, 1 mM EDTA, 0.3 mM EGTA, protease inhibitors; Complete EDTA-free, Boehringer, Mannheim, Germany) and separated by SDS-PAGE (12.5%, 10 μg protein/lane). Claudin-10 was detected by immunoblotting, employing antibodies raised against human claudin-10 (Zymed Laboratories, Invitrogen Immunodetection, San Francisco, CA), against the FLAG-tag or against the HA-tag (Sigma Aldrich, St Louis, MO), as appropriate. To exclude secondary effects through changes in the expression of endogenous claudins, western blots were also probed for claudins 1-5, 7 and 8 (see supplementary material Fig. S3).

Positive clones were screened further by immunocytochemistry as described below, to ensure appropriate localization of claudin-10. MDCK-C7 cells mock-transfected with respective empty vectors served as controls.

Immunocytochemistry

Transiently transfected MDCK-II cells were grown on coverslips and processed for confocal microscopy as previously described (Müller et al., 2003; Kausalya et al., 2006) using rat anti-HA (Roche; 1:100), rabbit anti-calreticulin (ABR; 1:300), and rabbit anti-ZO-1 (Zymed; 1:200) antibodies and suitable fluorescently labeled secondary antibodies (Invitrogen).

Stably transfected MDCK-C7 and MDCK-II cells were grown on transparent culture plate inserts (pore size 0.4 mm, effective area 0.6 cm2; Millicell-PCF, Millipore, Bedford, MA). Cells were rinsed with PBS, fixed with methanol or 2% paraformaldehyde in PBS, and permeabilized with PBS containing 0.5% Triton X-100. No differences resulting from the two fixation methods were observed. Primary antibodies employed were mouse anti-occludin and rabbit anti-claudin 10 (Zymed). Concentrations of primary antibody were 10 mg/ml. Secondary antibodies Alexa Fluor 488 goat anti-mouse and Alexa Fluor 594 goat anti-rabbit (both used at concentrations of 2 mg/ml) were purchased from Molecular Probes (Eugene, OR). DAPI (4′,6-diamidino-2-phenylindole dihydrochloride; 1 mM) was used to stain cell nuclei. Fluorescence images were obtained with a confocal microscope (Zeiss LSM510, Jena, Germany) using excitation wavelengths of 543 nm, 488 nm and 405 nm.

Pull-down experiments

Peptides corresponding to the 11 C-terminal amino acids of wild-type or mutant Cldn10 (last two amino acids mutated to alanine) were custom synthesized and coupled to Dynabeads (Dynal) by Biogenes, Germany. The pull-down experiment was performed essentially as described by Müller et al. (Müller et al., 2003) but with some modifications. Briefly, to construct Myc-tagged human ZO-1 PDZ domain cDNA, PDZ domains (amino acids 1-507 of ZO-1) were fused in frame to an N-terminal Myc tag in pGBKT7 vector (Clontech) as described by Kausalya et al. (Kausalya et al., 2001). Subsequently, Myc-tagged PDZ domains were generated by in vitro transcription and translation (Quick coupled T7 TNT: Promega). 10 μl of the product was incubated with peptide-coupled Dynabeads (2-10 μg peptide) for 2 hours at 4°C in binding buffer (25 mM Tris, pH 7.5, 50 mM NaCl and 0.1% Tween 20; or 0.1% Triton X-100, 20 mM MgCl2 and 1 mM DTT). Beads were then rinsed with washing buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl and 0.1% Tween 20, or 0.1% Triton X-100, 20 mM MgCl2, and 1 mM DTT) and were resuspended in SDS sample buffer. Bound ZO-1 PDZ domains were analyzed by SDS-PAGE and western blotting and detected using mouse anti Myc antibodies (Roche).

Electrophysiology

For electrophysiological measurements and molecular analyses, epithelial cell monolayers were grown on culture plate inserts (pore size 0.45 μm, effective area 0.6 cm2; Millicell-HA). Confluent cell layers were used on days 5-7 after seeding. Inserts were mounted in Ussing chambers specially designed for insertion of Millicell filters (Kreusel et al., 1991), and water-jacketed gas lifts were filled with 10 ml circulating fluid on each side. The standard bath solution contained: 119 mM NaCl, 21 mM NaHCO3, 5.4 mM KCl, 1.2 mM CaCl2, 1 mM MgSO4, 3 mM Hepes, and 10 mM D(+)-glucose. The solution was constantly bubbled with 95% O2 and 5% CO2, to ensure a pH value of 7.4 at 37°C. Resistance of bath solution and filter support (Rfilter) was measured before each single experiment and subtracted.

To determine ion permeabilities, 5 ml of the standard bath solution on either the apical or the basolateral side of the cell layer was replaced by one of the solutions listed in Table 2. Resulting potentials were corrected for liquid junction potentials which were determined in analogous experiments using empty filter supports. Liquid junction potentials may be greatly affected by the freshness of agar bridges used, so that errors cannot be completely excluded. Although this would affect total permeabilities, it would not change the relative differences between the clones investigated and would thus be irrelevant to the general conclusions.

The permeability ratio, PNa/PCl was calculated using Eqn 1, from `dilution potentials' resulting from the replacement of NaCl by mannitol (Table 2, solutions 1 and 2). Absolute permeabilities for Na+ and Cl (PNa, PCl) were calculated according to Eqn 2 (see Hou et al., 2005), using the PNa/PCl calculated from Eqn 1 and the transepithelial resistance determined during the same experiment.

Relative permeabilities for monovalent cations X+, PX/PNa, were calculated from `biionic potentials' resulting from the replacement of NaCl with the respective chloride salt of the cation (Table 2, solutions 3-6), using the average PNa/PCl calculated from dilution potentials of the appropriate cell type, employing Eqn 3.

To determine permeabilities for divalent cations, X2+, from analogous biionic experiments, standard conditions had to be changed to bicarbonate-free conditions (Table 2, solution 7), to avoid precipitation. 5 ml of this solution on the basolateral side was exchanged for 5 ml of a solution in which NaCl was iso-osmotically replaced by mannitol (Table 2, solution 8). The resulting dilution potential was used to determine PNa/PCl as described above. Subsequently, 5 ml of solution on the apical side was replaced with the appropriate solution (Table 2, solution 9-12). The potential difference between the original condition (140 mM Na+ on the apical and basolateral side) and the final condition (70 mM Na+ on both sides, mannitol on the basolateral side versus chloride salt of a divalent cation on the apical side) was used to calculate relative permeabilities, PX/PNa, from Eqn 4. Absolute permeabilities, PX were calculated, using the PNa determined within the same experiment. It proved impossible to do the reverse of this experiment (i.e. mannitol on the apical side, divalent cation on the basolateral side), because an increase in basolateral divalent cation concentration elicited transcellular Cl secretion by activating the basolateral calcium sensing receptor, thus distorting the recorded biionic potential signal.

Relative permeabilities for monovalent anions Y were also carried out under bicarbonate-free conditions, employing two different methods. Method a: relative permeabilities, PY/PNa, were calculated from `dilution potentials' resulting from the replacement of NaY (Table 2, solution 7, 13, or 14) with mannitol (Table 2, solution 8), using Eqn 5. Method b: relative permeabilities, PY/PCl, were calculated from `biionic potentials' resulting from the replacement of NaCl (Table 2, solution 7) with the respective sodium salt of the anion (Table 2, solution 13 or 14), using the average PNa/PCl calculated from dilution potentials, employing Eqn 6.

Equations

General

ΔE = EblEap (corrected for liquid junction potentials):
\[\ s=2.303{\cdot}\frac{R{\cdot}T}{F},\ \]
where R is the universal gas constant, T is absolute temperature, F, is the Faraday constant.

aIon-ap and aIon-bl, are apical and basolateral ion activities, respectively, calculated from the modified Debye-Hückel formalism and constants published by Meier (Meier, 1982).

Relative permeabilities of Na+ and Cl

\[\ \frac{P_{Na}}{P_{Cl}}=\frac{10^{({\Delta}E{/}s)}{\cdot}a_{Cl-ap}-a_{Cl-bl}}{a_{Na-ap}-10^{({\Delta}E{/}s}){\cdot}a_{Na-bl}}.\ \]
(1)

Absolute permeabilities of Na+ and Cl

\[\ P_{Cl}=\frac{G}{[NaCl]}{\cdot}\frac{R{\cdot}T}{F^{2}}{\cdot}\frac{1}{1+P_{Na}{/}P_{Cl}},P_{Na}=P_{Cl}{\cdot}\frac{P_{Na}}{P_{Cl}}=\frac{G}{[NaCl]}{\cdot}\frac{R{\cdot}T}{F^{2}}{\cdot}\frac{1}{1+P_{Na}{/}P_{Cl}}{\cdot}\frac{P_{Na}}{P_{Cl}},\ \]
(2)
where G is transepithelial conductance [1/(transepithelial resistance)].

Relative permeability of a monovalent cation X

\[\ \frac{P_{x}}{P_{Na}}=\frac{a_{Na-ap}+P_{Cl}{/}P_{Na}{\cdot}a_{Cl-bl}-10^{({\Delta}E{/}s)}{\cdot}(a_{Na-bl}+P_{Cl}{/}P_{Na}{\cdot}a_{Cl-ap})}{10^{({\Delta}E{/}s)}{\cdot}a_{X-bl}-a_{X-ap}}.\ \]
(3)

Relative permeability of a divalent cation X

\[\ \frac{P_{x}}{P_{Na}}=\frac{-n_{Na}-P_{Cl}{/}P_{Na}{\cdot}n_{Cl}{\cdot}z_{Cl}}{n_{x}{\cdot}z_{x}}\ \]
(4)
with
\[\ n=z{\cdot}\frac{{\Delta}E}{s}{\cdot}\frac{a_{bl}{\cdot}10^{z({\Delta}E{/}s)}-a_{ap}}{10^{z({\Delta}E{/}s)}-1},\ \]

where a is the respective activity and z, the respective charge number.

Relative permeability of a monovalent anion Y

Method a (constant, low [Cl], contribution assumed to be negligible):
\[\ \frac{P_{Y}}{P_{Na}}=\frac{a_{Na-ap}-10^{({\Delta}E{/}s)}{\cdot}a_{Na-bl}}{10^{({\Delta}E{/}s)}{\cdot}a_{Y-ap}-a_{Y-bl}}.\ \]
(5)
Method b (constant [Na+]):
\[\ \frac{P_{Y}}{P_{Cl}}=\frac{P_{Na}{/}P_{Cl}{\cdot}a_{Na-ap}+a_{Cl-bl}-10^{({\Delta}E{/}s)}{\cdot}(P_{Na}{/}P_{Cl}{\cdot}a_{Na-bl}+a_{Cl-ap})}{10^{({\Delta}E{/}s)}{\cdot}a_{Y-ap}-a_{Y-bl}}\ \]
(6)

We would like to thank Carsten Wagner and Ana Velic (Institute of Physiology, University of Zürich) for microdissection of nephron segments. Valuable technical assistance from Detlef Sorgenfrei and Christina Papadopoulos (Institute of Clinical Physiology, Charité, Berlin) is gratefully acknowledged. This study was financially supported by the Agency for Science and Technology (A*STAR), Singapore to W.H., by the European Union (European Renal Genome Project EuReGene, FP6 05085 and EUNEFRON, FP7) to D.M., by the German Research Foundation (DFG GU447/11-1) and by the Sonnenfeld-Stiftung to D.G.

Amasheh, S., Meiri, N., Gitter, A. H., Schöneberg, T., Mankertz, J., Schulzke, J. D. and Fromm, M. (
2002
). Claudin-2 expression induces cation-selective channels in tight junctions of epithelial cells.
J. Cell Sci.
115
,
4969
-4976.
Cereijido, M., Robbins, E. S., Dolan, W. J., Rotunno, C. A. and Sabatini, D. D. (
1978
). Polarized monolayers formed by epithelial cells on a permeable and translucent support.
J. Cell Biol.
77
,
853
-880.
Diamond, F. M. and Wright, E. M. (
1969
). Biological membranes: the physical basis of ion and nonelectrolyte selectivity.
Annu. Rev. Physiol.
31
,
581
-646.
Eisenman, G. (
1962
). Cation selective glass electrodes and their mode of operation.
Biophys. J.
2
,
259
-323.
Gekle, M., Wünsch, S., Oberleithner, H. and Silbernagl, S. (
1994
). Characterization of two MDCK-cell subtypes as a model system to study principal cell and intercalated cell properties
Pflügers Arch.
428
,
157
-162.
Greger, R. (
1981
). Cation selectivity of the isolated perfused cortical thick ascending limb of Henle's loop of rabbit kidney.
Pflügers Arch.
390
,
30
-37.
Guan, X., Inai, T. and Shibata, Y. (
2005
). Segment-specific expression of tight junction proteins, claudin-2 and -10, in the rat epididymal epithelium.
Arch. Histol. Cytol.
68
,
213
-225.
Hashizume, A., Ueno, T., Furuse, M., Tsukita, S., Nakanishi, Y. and Hieda, Y. (
2004
). Expression patterns of claudin family of tight junction membrane proteins in developing mouse submandibular gland.
Dev. Dyn.
231
,
425
-431.
Holmes, J. L., Van Itallie, C. M., Rasmussen, J. E. and Anderson, J. M. (
2006
). Claudin profiling in the mouse during postnatal intestinal development and along the gastrointestinal tract reveals complex expression patterns.
Gene Expr. Patterns
6
,
581
-588.
Hou, J., Paul, D. L. and Goodenough, D. A. (
2005
). Paracellin-1 and the modulation of ion selectivity of tight junctions.
J. Cell Sci.
118
,
5109
-5118.
Hou, J., Renigunta, A., Konrad, M., Gomes, A. S., Schneeberger, E. E., Paul, D. L., Waldegger, S. and Goodenough, D. A. (
2008
). Claudin-16 and claudin-19 interact and form a cation-selective tight junction complex.
J. Clin. Invest.
118
,
619
-628.
Inai, T., Sengoku, A., Guan, X., Hirose, E., Iida, H. and Shibata, Y. (
2005
). Heterogeneity in expression and subcellular localization of tight junction proteins, claudin-10 and -15, examined by RT-PCR and immunofluorescence microscopy.
Arch. Histol. Cytol.
68
,
349
-360.
Itoh, M., Furuse, M., Morita, K., Kubota, K., Saitou, M. and Tsukita, S. (
1999
). Direct binding of three tight junction-associated MAGUKs, ZO-1, ZO-2, and ZO-3, with the COOH termini of claudins.
J. Cell Biol.
147
,
1351
-1363.
Kausalya, P. J., Reichert, M. and Hunziker, W. (
2001
). Connexin45 directly binds to ZO-1 and localizes to the tight junction region in epithelial MDCK cells.
FEBS Lett.
505
,
92
-96.
Kausalya, P. J., Amasheh, S., Günzel, D., Wurps, H., Müller, D., Fromm, M. and Hunziker, W. (
2006
). Disease-associated mutations affect intracellular traffic and paracellular Mg2+ transport function of claudin-16.
J. Clin. Invest.
116
,
878
-891.
Kikeri, D., Sun, A., Zeidel, M. L. and Hebert, S. C. (
1989
). Cell membranes impermeable to NH3.
Nature
339
,
478
-480.
Kiuchi-Saishin, Y., Gotoh, S., Furuse, M., Takasuga, A., Tano, Y. and Tsukita, S. (
2002
). Differential expression patterns of claudins, tight junction membrane proteins, in mouse nephron segments.
J. Am. Soc. Nephrol.
13
,
875
-886.
Kreusel, K. M., Fromm, M., Schulzke, J. D. and Hegel, U. (
1991
). Cl secretion in epithelial monolayers of mucus-forming human colon cells (HT-29/B6).
Am. J. Physiol.
261
,
C574
-C582.
Lim, T. S., Vedula, S. R., Hui, S., Kausalya, P. J., Hunziker, W. and Lim, C. T. (
2008
). Probing effects of pH change on dynamic response of Claudin-2 mediated adhesion using single molecule force spectroscopy.
Exp. Cell Res.
314
,
2643
-2651.
Meier, P. C. (
1982
). Two parameter Debye-Hückel approximation for the evaluation of mean activity coefficients of 109 electrolytes.
Anal. Chim. Acta
136
,
363
-368.
Müller, D., Kausalya, P. J., Claverie-Martin, F., Meij, I. C., Eggert, P., Garcia-Nieto, V. and Hunziker, W. (
2003
). A novel claudin-16 mutation associated with childhood hypercalciuria abolishes binding to ZO-1 and results in lysosomal mistargeting.
Am. J. Hum. Genet.
73
,
1293
-1301.
Ohta, H., Adachi, H., Takiguchi, M. and Inaba, M. (
2006
). Restricted localization of claudin-16 at the tight junction in the thick ascending limb of Henle's loop together with claudins 3, 4, and 10 in bovine nephrons.
J. Vet. Med. Sci.
68
,
453
-463.
Smith, A. N., Jouret, F., Bord, S., Borthwick, K. J., Al-Lamki, R. S., Wagner, C. A., Ireland, D. C., Cormier-Daire, V., Frattini, A., Villa, A. et al. (
2005
). Vacuolar H+-ATPase d2 subunit: molecular characterization, developmental regulation, and localization to specialized proton pumps in kidney and bone.
J. Am. Soc. Nephrol.
16
,
1245
-1256.
Stehberger, P. A., Schulz, N., Finberg, K. E., Karet, F. E., Giebisch, G., Lifton, R. P., Geibel, J. P. and Wagner, C. A. (
2003
). Localization and regulation of the ATP6V0A4 (a4) vacuolar H+-ATPase subunit defective in an inherited form of distal renal tubular acidosis.
J. Am. Soc. Nephrol.
14
,
3027
-3038.
van de Graaf, S. F. J., Bindels, R. J. M. and Hoenderop, J. G. J. (
2007
). Physiology of epithelial Ca2+ and Mg2+ transport.
Rev. Physiol. Biochem. Pharmacol.
158
,
77
-160.
Van Itallie, C. M. and Anderson, J. M. (
2006
). Claudins and epithelial paracellular transport.
Annu. Rev. Physiol.
68
,
403
-429.
Van Itallie, C. M., Fanning, A. S. and Anderson, J. M. (
2003
). Reversal of charge selectivity in cation or anion-selective epithelial lines by expression of different claudins.
Am. J. Physiol. Renal Physiol.
285
,
F1078
-F1084.
Van Itallie, C. M., Rogan, S., Yu, A., Vidal, L. S., Holmes, J. and Anderson, J. M. (
2006
). Two splice variants of claudin-10 in the kidney create paracellular pores with different ion selectivities.
Am. J. Physiol. Renal Physiol.
291
,
F1288
-F1299.
Wen, H., Watry, D. D., Marcondes, M. C. and Fox, H. S. (
2004
). Selective decrease in paracellular conductance of tight junctions: role of the first extracellular domain of claudin-5.
Mol. Cell. Biol.
24
,
8408
-8417.

Supplementary information