*Review* **The Integral Role of Tight Junction Proteins in the Repair of Injured Intestinal Epithelium**

**Zachary M. Slifer and Anthony T. Blikslager \***

Department of Clinical Sciences, College of Veterinary Medicine, North Carolina State University, Raleigh, NC 27607, USA; zmslifer@ncsu.edu

**\*** Correspondence: Anthony\_Blikslager@ncsu.edu

Received: 30 August 2019; Accepted: 14 October 2019; Published: 1 February 2020

**Abstract:** The intestinal epithelial monolayer forms a transcellular and paracellular barrier that separates luminal contents from the interstitium. The paracellular barrier consists of a highly organized complex of intercellular junctions that is primarily regulated by apical tight junction proteins and tight junction-associated proteins. This homeostatic barrier can be lost through a multitude of injurious events that cause the disruption of the tight junction complex. Acute repair after injury leading to the reestablishment of the tight junction barrier is crucial for the return of both barrier function as well as other cellular functions, including water regulation and nutrient absorption. This review provides an overview of the tight junction complex components and how they link to other plasmalemmal proteins, such as ion channels and transporters, to induce tight junction closure during repair of acute injury. Understanding the components of interepithelial tight junctions and the mechanisms of tight junction regulation after injury is crucial for developing future therapeutic targets for patients experiencing dysregulated intestinal permeability.

**Keywords:** barrier function; tight junction; repair; occludin; claudin; NHE2; ClC-2

#### **1. Intestinal Epithelium as a Selective Barrier**

The intestine is lined with a monolayer of columnar epithelium that maintains two critical functions: (1) selectively filtering luminal contents, including nutrients, water and electrolytes, to allow for their translocation into the circulation and (2) forming a barrier to prevent the translocation of luminal toxins, commensal or pathogenic microorganisms, and foreign antigens into the circulation [1]. Under homeostatic conditions, these functions are regulated by both transcellular and paracellular pathways, the latter of which are primarily maintained by apical tight junction proteins through paracellular pore and leak permeability pathways [2,3]. The paracellular pathway is associated with the charge and size selective transport of materials through the space between intestinal epithelial cells.

Intestinal barrier homeostasis is disrupted through tight junction protein dysregulation, which occurs via a variety of injurious events, including microbial degradation and bacterial toxin exposure, exposure to cytotoxic agents, exposure to pro-inflammatory cytokines such as IFNγ and TNFα, intestinal autoimmune disease such as Celiac disease, and intestinal ischemia [4–9]. The loss of tight junction integrity results in the formation of a third pathway, known as the high-capacity and nonselective unrestricted permeability pathway, which can allow for the unrestricted movement of microorganisms and large proteins across the paracellular space [10]. An inability to rapidly repair the tight junctions in order to restore epithelial barrier function is detrimental to the patient, as it can result in various pathologies, including sepsis and multiple organ dysfunction [11,12]. Therefore, understanding factors that can regulate the tight junction complex during repair of injured intestinal epithelium is crucial for developing future therapeutic targets.

#### **2. Tight Junction Protein Structure**

Tight junctions are made up of a number of protein elements, including transmembrane claudins (total of 27 mammalian claudins), as well as myelin and lymphocyte (MAL) and related proteins for vesicle trafficking and membrane link (MARVEL) [13–18]. MARVEL domain-containing proteins are a component of a larger group of tight junction-associated MARVEL proteins (TAMPS) that include transmembrane proteins such as occludin and tricellulin [18–20]. Other tight junction-associated transmembrane molecules include junctional adhesion molecules (JAM-1, -2, and -3) that can regulate the formation of tight junctions and migration of neutrophils [21–26]. Additionally, intracellular scaffold proteins such as zonula occludens (ZO) -1, -2, and -3 play an integral role in tight junction protein assembly and link tight junction transmembrane proteins with the actin cytoskeleton [27–30].

Intestinal claudins exist in two different classes: sealing claudins and pore-forming claudins [31]. Increased membrane expression of sealing proteins results in a 'tighter' epithelial barrier, further restricting the movement of luminal contents through the paracellular space. Sealing tight junction proteins include claudins-1, -3, -4, -5, -8, -11, -14, 18, and -19 [17]. Alternatively, increased membrane expression of pore-forming proteins (including claudins-2, -10a/-10b, -15, -16, and -17) reduces the selectivity for luminal contents that can pass between epithelial cells, thereby increasing paracellular permeability [17]. Each pore-forming claudin has an ionic charge specificity for cations or anions as well as ionic size selectivity, thereby increasing the permeability for ions based on claudin-specific permeability characteristics. An interesting interaction between the two classes of claudins has been observed with the displacement of pore-forming claudins by sealing claudins. For example, in a claudin-8 transfected Madin–Darby canine kidney II (MDCK II) cell line in which claudin-8 expression occurred in the absence of doxycycline, claudin-8 displacement of claudin-2 was visualized upon immunofluorescent imaging [32,33]. Specifically, claudin-8 served to replace claudin-2 in tight junction strands in this model, which reduced the number of functional paracellular cation pores [32].

Tight junction protein expression in the intestinal tract is tissue- and age-specific. For example, claudin-2 is normally expressed in the human small intestine but is reported to be expressed only in the colonic crypt of fetal humans and absent in the adult colon under homeostatic conditions [34]. Overall, claudin-2 generally exhibits higher expression in leaky epithelial tissues, such as colonic tissues from a subset of patients with inflammatory bowel disease [35]. Additionally, its protein expression was detected throughout the crypt–villus axis of human small intestines but was only detected in undifferentiated crypt cells of human colonic tissue [36]. Other claudins follow suit regarding crypt–luminal axis expression with some pore-forming claudins (-2, -10, -13, -15) being restricted to the intestinal crypt base in murine tissue while other sealing claudins (-3, -4, -7, -8) are expressed in luminal epithelial cells [37–40].

#### *Special Functions of Select Tight Junction Proteins*

As discussed throughout the remainder of this review, regulation of tight junction proteins is a vital component of epithelial barrier repair after injury. However, specific tight junction proteins can have additional special functions that are species and age dependent. In the case of claudin-4, there is an age-dependent disparity between cellular localization of tight junction proteins [41]. Intestinal porcine claudin-4 is localized to the apical surface of porcine jejunal enterocytes within the first two days of age and only localizes to the lateral surface between adjacent epithelial cells beyond two days of age [41]. This age-specificity of claudin-4 localization in piglet jejunum may be due to immunological naivety that newborn piglets experience. Piglets must be able to acquire and absorb immunoglobulins and other immune-related molecules, including cytokines and antimicrobial peptides, from colostrum within the first day of life to prevent death via bacterial sepsis [42]. It is reported that this age-specific, apical surface localization of claudin-4 occurs in jejunal enterocytes at the same period of time in which the vital immune macromolecules are absorbed into the bloodstream [41]. Therefore, this special function of claudin-4 localization is necessary to allow for the proper uptake of immune molecules by the piglet, and subsequent relocalization to the lateral surface may aid in sealing of the paracellular space between adjacent epithelial cells once the immune constituents are absorbed.

Another tight junction protein, tricellulin, serves a special function in the tight junction barrier where more than two epithelial cells meet. Tight junction strands between two adjacent epithelial cells typically associated laterally to pair with another tight junction strand, forming bicellular tight junctions between the two cells [43]. However, these bicellular tight junctions are not continuous at tricellular epithelial cell contacts and have therefore been described as tricellular tight junction proteins. While traditional tight junction proteins, such as occludin and claudins, are found in both bicellular and tricellular tight junctions, tricellulin is concentrated to the tricellular tight junction and its knockdown in the EpH4 cell line of immortalized mouse mammary gland epithelium resulted in altered organization of bicellular tight junction proteins [20]. Additionally, tricellulin has been shown to not affect the permeability for ions while forming a barrier to macromolecules in tricellulin-transfected MDCK II cells overexpressing tricellulin in the tricellular tight junctions [44]. Overall, it is crucial to consider all components and special functions of tight junction proteins when studying tight junction structure. Furthermore, the understanding of these tight junction special functions may be crucial to restoring barrier function following injury.

#### **3. Acute Mechanisms of Repair in Injured Intestinal Epithelium**

When the monolayer of intestinal epithelium is injured, such as that which occurs during ischemia/reperfusion injury or exposure to pathogenic microbes such as rotavirus [45,46], detachment of the epithelium from the basement membrane and separation of adjacent epithelial cells from one another due to dysregulation and loss of tight junctional proteins occurs. Furthermore, the loss of polarity-establishing tight junctional proteins results in the loss of cell polarity, which abolishes apical and basolateral positioning of localized molecules such as ion channels/transporters, resulting in their mislocalization [47]. When homeostatic positioning of ion channels and transporters is lost, this can subsequently lead to the dysregulation of a multitude of cellular functions including water absorption/secretion, intracellular and organelle pH, and nutrient absorption.

Once the cause of intestinal injury is resolved, such as restoration of blood flow in ischemic injury, rapid mechanisms of intestinal mucosal repair take place in a well-orchestrated series of reparative events. Initially, small intestinal villi contract via the contraction of myofibroblasts adjacent to the epithelial basement membrane and centrally along the central lacteal. Villus contraction is characterized histologically by a quantitatively diminished villus height [48] and occurs in response to mediators such as PGE2 [49]. Villus contraction results in reduction of the denuded surface area that remains to be covered by epithelial cells. Simultaneously, restitution of epithelial cells shouldering the site of injury occurs to cover the denuded area [50]. These cells depolarize to disassemble microvilli, allowing for subsequent lamellipodia-driven movement via actin–myosin treadmilling, while maintaining transient attachment to the basement membrane through integrins [11]. Although the underlying intestinal layers may not appear exposed to luminal contents since the mucosa is no longer denuded, the unrestricted permeability pathway via poorly formed tight junctions allows for microorganisms and macromolecules to cross the epithelial barrier. In order for the tight junction barrier and cell polarity to be restored, tight junction proteins internalized during injury, such as the endocytosis of occludin that accompanies anoxic injury in Caco-2 cells, must be reinserted back into the membrane via recycling endosomes [51–53]. Ultimately, following these acute repair mechanisms, crypt cells can proliferate and differentiate to restore the proper number of epithelial cells to the monolayer in order to regain full homeostatic function.

#### **4. Regulation of Tight Junctions via Ion Channels**/**Transporters**

Closure of the tight junction after acute intestinal injury is paramount in restoring barrier function and returning to homeostatic functioning. Tight junction proteins can be regulated by many factors, including cytokines, growth factors, and nutrients. For example, transport of glucose by SGLT1 has been shown to result in the physiological opening of tight junctions in an NHE3-dependent mechanism [54]. Alternatively, ion channel/transporters, including proteins from the Na+/H<sup>+</sup> exchanger (NHE) family as well as chloride channel protein 2 (ClC-2) have also been shown to regulate tight junction proteins,

specifically after intestinal ischemic injury [55–57]. This review will examine the reparative role of these transport proteins specifically related to restoration of junctions.

#### *4.1. NHE2 and Intestinal Repair*

The gastrointestinal epithelium is home to many ion transporters that are collectively responsible for regulating homeostatic cell functions, including the regulation of nutrient absorption, cytosolic and organelle pH, water absorption and secretion, and cell volume [58]. One major family of ion transporters in the human GI tract is the *SLC9* gene family, also known as the NHE family. NHE isoforms belonging to the *SLC9A* gene subgroup (*SLC9A1-9*) can be either plasmalemmal or intracellular, depending on the isoform and tissue location within the gut [59]. Additionally, the Na+/H<sup>+</sup> exchanger 5 (NHE5) is the only isoform for which expression has not been shown in the gastrointestinal tract [60]. These proteins are responsible for the electroneutral antiport of Na<sup>+</sup> into intestinal epithelium in exchange for H<sup>+</sup> secreted from the cell to maintain cellular pH and volume.

An additional function of NHEs that continues to be explored is the link between NHEs and the tight junction. One mechanism that links NHEs to the tight junction is through binding to the actin cytoskeleton. Specifically, Na+/H<sup>+</sup> exchanger 3 (NHE3) has been shown to bind directly to the actin cytoskeleton and indirectly through various binding partners, including ezrin [61,62]. The ezrin protein is known to link the plasma membrane to the cytoskeleton in its active, phosphorylated conformation through binding to actin with its C-terminal region [63–65]. This interaction with the cytoskeleton has been shown to regulate plasma membrane tension, which is involved in motility and endocytosis [66]. Ezrin links the cytoskeleton to the plasma membrane through binding of its N-terminal region to either membrane lipids or cytoplasmic regions of transmembrane proteins, including NHE3 [66,67]. Thus, by linking transmembrane proteins such as NHE3 to the cytoskeleton, there is an indirect link between transmembrane proteins and tight junction proteins.

Of the NHE isoforms that have been described in the gut, Na+/H<sup>+</sup> exchanger 2 (NHE2) is one of the least described NHEs in regards to its homeostatic and pathophysiologic functionality. However, NHE2 has been linked to paracellular barrier function and tight junction regulation during the recovery of injured intestinal epithelium [55,56]. In both porcine and murine models of intestinal ischemic injury, NHE2, rather than NHE1 or NHE3, appears to be the primary NHE responsible for regulating tight junction proteins during the recovery of ischemia-injured intestines [55,56]. During ex vivo recovery of porcine intestinal ischemia, selective pharmacologic inhibition of NHE2 enhanced epithelial recovery, as evidenced by significant elevations in transepithelial electrical resistance (TER) while inhibition of NHE1 or NHE3 did not elicit a recovery response [56]. In the same study, this NHE2-specific inhibitory effect on recovery was independent of epithelial restitution, and NHE2 was shown to co-immunoprecipitate with ezrin/radixin/moesin (ERM)-binding phosphoprotein 50 (EBP50), also known as NHE regulatory factor 1 (NHERF1), in ischemia-injured porcine ileum. These data suggest that NHE2 regulates restoration of the tight junction barrier during recovery of intestinal ischemia and is potentially linked to the actin cytoskeleton through binding partners (Figure 1). Although NHE2 is also implicated in the in vivo recovery of murine intestinal ischemia, the genetic knockout of NHE2 in the murine model has the inverse effect when compared to pharmacologic inhibition of NHE2 in the porcine model of intestinal ischemia [55]. Specifically, NHE2 null mice exhibit increased blood-to-lumen 3H-mannitol flux at 1.5 and 3 hours after ischemic injury as well as a change in localization of occludin and claudin-1 from the membrane to the cytosol when compared to wild-type mice [55]. Additionally, epithelial restitution after intestinal ischemia was unaffected by the absence of NHE2 in this model. It is important to note that pharmacologic inhibition or genetic knockout of NHE2 may affect intracellular pH (pHi) since NHEs are known to contribute to pHi changes, and these potential pHi changes can affect charge selectivity of the paracellular pathway [68,69]. However, further studies will be required to determine if NHE-mediated changes in pHi are linked to alterations in the tight junction. Together, this information suggests that NHE2 regulates acute recovery after intestinal ischemic injury in a tight junction-dependent manner, whereas its absence delays restoration of tight junction barrier function.

**Figure 1.** Schematic representation of the potential link of NHE2 to the actin cytoskeleton and subsequently the tight junction through binding partners. The primary candidate protein linking NHE2 to the actin cytoskeleton is phosphorylated ezrin. Based on information known about NHE3 and data from NHE2 in vivo studies, NHE2 may bind directly to ezrin or indirectly through additional binding partners, including NHERF1/EBP50.

#### *4.2. ClC-2 and Intestinal Repair*

Chloride secretion from intestinal epithelium into the lumen is crucial for homeostatic water absorption/secretion via maintaining an osmotic balance with luminal accumulation of both chloride and sodium ions. This subsequently allows for proper mucosal hydration of the epithelial layer, which protects the lumen as food passes through the intestine [70]. The primary protein responsible for chloride transport into the intestinal lumen is the apically located cystic fibrosis transmembrane receptor (CFTR) [71]. However, another contributor to transepithelial chloride transport within intestinal epithelium is the voltage-gated ClC-2 protein, one of nine mammalian proteins belonging to the chloride channel (ClC) protein family [72]. ClC-2 has been shown to localize in the plasma membrane at tight junction complexes within mouse intestinal epithelium [73] or has plasmalemmal basolateral localization within guinea pig colons [74], suggesting species- or tissue-specific localization of ClC-2.

In addition to its role in transepithelial chloride transport, ClC-2 has been shown to regulate intestinal tight junction barrier function in various injury models. After porcine intestinal ischemic injury, stimulation of ClC-2 with the ClC-2 agonist lubiprostone during ex vivo recovery on Ussing chambers resulted in marked increases in TER and reduced mucosal-to-serosal mannitol flux [57]. Contrasting the effect of ClC-2 stimulation with lubiprostone, the genetic absence of ClC-2 in a murine model of intestinal ischemia resulted in significant increases in blood-to-lumen mannitol clearance while also reducing expression of membrane-bound occludin and claudin-1 after up to 3 hours of in vivo recovery [75]. In this murine model, occludin co-localized with ClC-2 after co-immunoprecipitation studies, and its localization to the tight junction region was diffuse in ClC-2 null mice after up to 3 hours of recovery [75]. Additionally, in a murine model of dextran sulfate sodium (DSS)-induced colitis, the absence of ClC-2 increased disease severity, as measured through significant losses in body weight and significant increases in disease activity index [76]. ClC-2 null mice treated with DSS also demonstrated significantly increased expression of claudin-2 and reduced occludin expression in the same study. Interestingly, a recent in vitro study established Caco-2 cells overexpressing ClC-2 (Caco-2ClCN2), and this ClC-2 overexpression resulted in a decrease of the pore-forming claudin-2 protein while maintaining claudin-1 and claudin-4 protein levels to that of control cells [77]. As an aside, although cell volume and pHi is partially regulated by ClC-2 and thus the genetic knockout

of ClC-2 can affect these intracellular factors, studies will be needed to determine if ClC-2-mediated changes in these intracellular factors have an effect on the tight junction [78,79]. Based on these studies, there appears to be a mechanistic link between ClC-2 and the regulation of membrane claudin expression, but further studies will need to be carried out to determine how ClC-2 plays a role in claudin expression patterns. Nonetheless, current data suggest the critical role of ClC-2 in barrier function during recovery from epithelial injury while also reinforcing the link between ClC-2 and the tight junction barrier.

The link between ClC-2 and the tight junction was initially shown to exist through intracellular caveolar trafficking of occludin via interaction with both caveolin-1 and the small GTPase Rab5 in a cell line derived from human intestinal Caco-2 cells (Figure 2) [80]. This connection between ClC-2, occludin, and caveolin-1 was further supported in vivo with a model of DSS-induced colitis. ClC-2 null mice treated with DSS had significantly displaced occludin/caveolin-1 densitometry readings toward high-density, detergent-soluble fractions of sucrose density gradient-based fractions when compared to wild-type mice treated with DSS [76]. These data suggest that after DSS-induced colitis, occludin and caveolin-1 are strongly associated in the cytosol of mice lacking ClC-2 but not in mice normally expressing ClC-2. In tandem, overexpression of ClC-2 in Caco-2ClCN2 cells was reported to not only exhibit enhanced tight junction barrier function through significant increases in TER and reductions in apical-to-basal inulin flux, but this ClC-2 overexpression further connected ClC-2 to caveolin-1 and caveolar trafficking of occludin [77]. Specifically, ClC-2 overexpression in Caco-2ClCN2 cells exhibited both significantly increased occludin protein and reduced endocytosis of occludin when compared to control cells while simultaneously diminishing both caveolin-1 protein and caveolae assembly [77]. Furthermore, this study reported that selective inhibition of ClC-2 lead to both reduced occludin protein and increased caveolin-1 protein. Taken together, there is strong evidence from both in vitro and in vivo models that links ClC-2 to the tight junction protein occludin and its regulation by caveolar trafficking. Based on the presented evidence, it is believed that ClC-2 facilitates the shuttling of endocytosed tight junction proteins back to the apical–lateral membrane to repair injured tight junctions. However, further mechanistic studies are required out to determine the precise mechanisms of these events.

**Figure 2.** Schematic representation of the link between ClC-2 and caveolar trafficking of occludin. This schematic is a slight modification from a previously published figure [81] to more closely associate ClC-2 to both the tight junction complex and to caveolin-1-associated endocytosis and recycling of tight junction proteins such as occludin. Note that the representation of ClC-2 at the tight junction complex is not exclusively apical in localization, which leaves room for ClC-2 to be more closely associated with occludin in the membrane.

#### **5. Conclusions**

Injury of intestinal epithelium affects both the epithelial cells and the junctional structures that link them. A great deal of attention has been focused on mechanisms of epithelial restitution, but a lesser level of attention has been paid to the reassembly of tight junctions within repairing epithelium. This intriguing process appears to be intimately associated with ion channels, which in the case of NHE2 and ClC-2, is facilitated by a close association with tight junction regulatory proteins. With ClC-2 in particular, the mechanism of ion channel-facilitated tight junction reassembly has been linked to endosomal recycling of tight junction proteins, with evidence of restoration of the positioning of tight junction integral membrane proteins during the reparative process, and increased membrane expression of sealing claudins with cellular over-expression of ClC-2. However, how precisely ion channels interact with structures such as endosomes, and how this facilitates insertion of sealing tight junction proteins at the repairing tight junction will require further study. Nonetheless, it does appear that ion channels such as NHE2 and ClC-2 have a greater cellular function than ion transport alone. It is conceivable that the transport of select ions accompanies a structural change that sets off a series of signaling events associated with tight junction reassembly, but this will require additional mechanistic work. Ultimately, further studies to uncover the relationship between ion channels and reassembly of tight junctions has the potential to lead to novel therapeutic targets for patients with increased intestinal paracellular permeability.

**Funding:** The APC was funded by USDA-NIFA, 2017-67015-26804 (to ATB).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Claudins in the Renal Collecting Duct**

**Janna Leiz 1,2 and Kai M. Schmidt-Ott 1,2,3,\***


Received: 22 October 2019; Accepted: 20 December 2019; Published: 28 December 2019

**Abstract:** The renal collecting duct fine-tunes urinary composition, and thereby, coordinates key physiological processes, such as volume/blood pressure regulation, electrolyte-free water reabsorption, and acid-base homeostasis. The collecting duct epithelium is comprised of a tight epithelial barrier resulting in a strict separation of intraluminal urine and the interstitium. Tight junctions are key players in enforcing this barrier and in regulating paracellular transport of solutes across the epithelium. The features of tight junctions across different epithelia are strongly determined by their molecular composition. Claudins are particularly important structural components of tight junctions because they confer barrier and transport properties. In the collecting duct, a specific set of claudins (Cldn-3, Cldn-4, Cldn-7, Cldn-8) is expressed, and each of these claudins has been implicated in mediating aspects of the specific properties of its tight junction. The functional disruption of individual claudins or of the overall barrier function results in defects of blood pressure and water homeostasis. In this concise review, we provide an overview of the current knowledge on the role of the collecting duct epithelial barrier and of claudins in collecting duct function and pathophysiology.

**Keywords:** epithelial barrier; barrier formation; collecting duct; tight junction; claudin

#### **1. Introduction**

Selective barriers formed by epithelial monolayers are vital for many critical physiological processes [1]. These barriers are composed of intercellular multiprotein complexes along the apical-basal axis of epithelial cells creating direct cell-cell interactions and forming apical junctional complexes [2]. These complexes include tight junctions (TJs), adherens junctions, and desmosomes.

TJs are the most apical and diverse of these junctional complexes. They are formed by transmembrane as well as cytoplasmic proteins and are linked to the cytoskeleton. The protein composition of TJs varies greatly among different epithelia and determines the barrier properties and permeability [3].

Overall, TJs serve a dual function in epithelial layers acting as (a) a fence, separating membrane proteins between the apical and basolateral membrane and (b) as a gate by regulating size- and charge-selective movements of ions, solutes, and small molecules via the paracellular route [4–7]. Consequently, TJs are not only critical to establish and maintain cell adhesion and epithelial polarity but are also necessary to create selective paracellular permeability between compartments.

Several studies show that the main protein family mediating TJ characteristics and functions are the membrane-spanning claudins. So far, more than 25 different claudins have been identified, and at least 10 of them are expressed in spatiotemporal patterns along the renal nephron (for a detailed review see [8]).

Structurally, the transmembrane proteins connect in both *cis* and *trans* to neighboring claudins of the same, as well as the opposing, cell membranes forming TJ strands [9]. Claudins are sufficient to arrange those strands when expressed in cells lacking endogenous TJ formation [10].

Here, we review the role of claudins in epithelial barrier formation with emphasis on the renal collecting duct as well as their implications in collecting duct physiology and function.

#### **2. The Renal Collecting Duct**

The renal collecting duct is the most distal part of the renal tubules. It connects renal nephrons with the renal pelvis and—together with the distal convoluted tubule and the connecting tubule—it contributes to the aldosterone-sensitive distal nephron. In general, the renal collecting duct plays important roles in fine-tuning urinary composition, extracellular fluid volume, electrolyte balance, blood pressure regulation, water homeostasis, and acid-base regulation [11,12].

Although most of the water and solute reabsorption in the kidney occurs in the more upstream segments of the nephron, transport variability in the renal collecting duct is significantly higher, and ion and water transport are under strict hormonal control [11]. This permits to adjust reabsorption and secretion to prevalent physiological conditions and to control the body's water and electrolyte balance closely.

The main solutes reabsorbed in the collecting duct are sodium (Na+) and chloride (Cl−), while potassium (K+) is secreted into the urine [13] (Figure 1). Transport of these ions occurs via the transcellular route—mediated by channels and membrane transporters—and via the paracellular route—mediated by TJs.

**Figure 1.** Transport in the renal collecting duct. In the renal collecting duct Na+ and Cl− are reabsorbed, while K+ is secreted into the urine. A steep osmolality gradient between the interstitium and lumen across the epithelial barrier is the driving force for water reabsorption.

#### **3. Claudins and Their Roles in Collecting Duct Epithelial Barrier Formation and Paracellular Ion Transport**

In general, claudins can be pore- or barrier-forming [14]. As in other segments of the renal nephron, it is believed that the expressed claudins are the main mediators of TJ permeability characteristics, and thus, control the paracellular pathway in the collecting duct [9,15]. Among the predominately expressed claudins in the renal collecting duct are claudin-3 (Cldn-3), -4 (Cldn-4), -7 (Cldn-7) and -8 (Cldn-8) [16,17]. Furthermore, a recently published study proposes a role for claudin-19 (Cldn-19) in the renal collecting duct [18]. Overall, the present knowledge strongly supports the theory that these claudins either enforce the epithelial barrier in general or build a paracellular Cl− channel, while other aspects of their functions are less clear (Table 1).

**Table 1.** Claudins expressed in the renal collecting duct and their proposed function in barrier formation and collecting duct tight junctions.


In the following sections, we will briefly review the published literature on the functions of these claudins in the collecting duct. As a general note, it should be acknowledged that gain-of-function studies utilizing overexpression of claudins in cell lines have a general limitation due to the influence of the endogenous components of the TJ in each cell line. The expressed claudins in the respective chosen model make it difficult to determine the exact impact of single claudins on permeability characteristics. In contrast, loss-of-function studies, if carried out in the appropriate system (i.e., the collecting duct in vivo or cell lines that accurately represent collecting duct epithelia), are more appropriate to outline the molecular functions of each claudin.

#### *3.1. Cldn-3*

Several studies have found that Cldn-3 enforces the paracellular barrier [19,20]. Its overexpression in otherwise leaky Madin-Darby Canine Kidney (MDCK) II cells leads to an altered TJ structure with increased transepithelial resistance (TER) and decreased permeability for ions as well as molecules of 332 Da and 4 kDa, indicating a general role for Cldn-3 in barrier formation and enforcement [19]. Furthermore, Cldn-3 has been described to promote tubule formation in vitro and might thus play a role during tubule morphogenesis in the developing kidney [30]. The conditional knockout of Cldn-3 resulted in increased urinary pH without any effect on urine or plasma electrolytes [21]. However, the collecting duct-specific role of Cldn-3 has not been studied so far.

#### *3.2. Cldn-7*

Cldn-7-deficient mice die shortly after birth due to severe renal salt wasting and dehydration [27]. Interestingly, collecting duct cells isolated from Cldn-7-deficient mice demonstrate an increase in TER and show decreased paracellular permeability for both Na+ and Cl− [31]. This suggests that Cldn-7 may form a non-selective paracellular channel facilitating Cl− and Na+ reabsorption in collecting ducts.

Different and partially conflicting roles of Cldn-7 have been suggested in other experimental systems, complicating the interpretation of Cldn-7 biology. Overexpression of Cldn-7 in LLC-PK1 cells induces an increase in TER accompanied by a reduced Cl− and elevated Na+ conductance [24]. In contrast, the knockdown of Cldn-7 in those cells leads to loss of anion selectivity and decreased Cl− permeation, whereas in cation-selective MDCK cells, knockdown of Cldn-7 leads to increased Na+ permeability [25]. Mutation experiments in vitro demonstrated the importance of specific negatively charged amino acids in the extracellular loop of Cldn-7 for paracellular permeability and charge

selectivity [32]. These contrasting results could indicate different roles for Cldn-7, depending on the cellular background, and endogenously expressed TJ components.

#### *3.3. Cldn-4 and Cldn-8*

Both, Cldn-4 and Cldn-8, are thought to serve as a cation barrier [33–35] and an anion channel [22]. This hypothesis is supported by data from in vivo experiments. Collecting duct-specific knockout of either Cldn-4 or Cldn-8, causes hypotension, hypochloremia, metabolic alkalosis and renal wasting of Na+ and Cl− [23,29]. These phenotypes would be consistent with Cldn-4 and Cldn-8 acting as paracellular Cl− channels, which are necessary for a paracellular "chloride shunt" required for effective transcellular Na+ reabsorption (via the epithelial sodium channel).

Different effects of Cldn-4 overexpression on charge selectivity, depending on the used cell model, have been described. Overexpression of Cldn-4 leads to an increase in TER in cation-selective MDCK II and anion-selective LLC-PK1 cells, but a decrease in Na+ permeability is only observed in MDCK II cells [34].

Cldn-8 has been shown to be necessary to recruit Cldn-4 to the TJ and to implement the protein into the junctional complex. In the absence of Cldn-8, Cldn-4 is mainly found in the endoplasmic reticulum and the Golgi complex, but not in the apical cell membrane where TJs are located [22]. Thus, Cldn-8 knockout causes a functional double knockout on the TJ level due to its requirement for correct Cldn-4 localization [22].

#### *3.4. Cldn-19*

In addition to the longer known claudins of the renal collecting duct described above, a recently published study indicates a role for Cldn-19, which had previously been linked to thick ascending limb functions [36–38]. Cldn-19 is associated with tightness and cation selectivity of the epithelial barrier. Interestingly, the TJ localization of Cldn-19 is promoted by a low osmolality, whereas high osmolality favors an intracellular localization, suggesting that it may contribute to tonicity-induced changes in paracellular ion selectivity [18]. This role of a claudin in epithelial adaption to the changing osmolality along the corticomedullary axis provides an interesting aspect of TJ physiology but collecting duct-specific knockout models of Cldn-19 have not yet been generated.

#### **4. The Collecting Duct Epithelial Barrier in Electrolyte-Free Water Reabsorption**

Although water transport in the renal collecting duct is not directly facilitated by water channel-forming claudins (contrasting with observations of Cldn-2 in the proximal tubule [39]), TJs in the collecting duct contribute to water reabsorption indirectly.

The driving force for water reabsorption in the renal collecting duct is a steep osmolality gradient formed by high concentrations of osmolytes in the interstitium, which increases towards the renal medulla [40]. The tight collecting duct epithelial barrier is crucial to maintain the osmolality gradient between the tubular lumen and the interstitium.

Water transport in the renal collecting duct is mainly controlled by arginine vasopressin (AVP), also called antidiuretic hormone. If the water content in the body is low, AVP binds to its type 2 receptor (V2R) localized in the basal cell membrane of collecting duct principal cells and stimulates the expression of aquaporin-2 (AQP2) water channels. Furthermore, AVP triggers a signaling cascade leading to the accumulation of AQP2 in the apical membrane (for a detailed review see [11,41]). This mechanism enables reabsorption of electrolyte-free water from the intraluminal urine and forms the basis of urinary concentrating ability. Inactivation of either V2R or AQP2 leads to polyuria with massive excretion of electrolyte-free water, a condition called nephrogenic diabetes insipidus [42].

A recent study has demonstrated the importance of an intact epithelial barrier in the renal collecting duct for efficient water reabsorption [43]. Deletion of the transcriptional regulator Grainyhead-like 2 (Grhl2), an epithelial transcription factor that induces the expression of barrier-enforcing molecular TJ components including Cldn-4 [44], results in a leaky collecting duct epithelium and a decreased TER across the collecting duct epithelium. Leakage of interstitial osmolytes across the Grhl2-deficient collecting duct epithelium is associated with defective retention of osmolytes in the interstitium of the inner medulla. Grhl2-deficient mice show signs of diabetes insipidus and fail to concentrate their urine adequately, although AQP-mediated water transport across the apical and basolateral membranes of Grhl2-deficient collecting ducts is intact [43]. This indicates that a tight collecting duct epithelial barrier is crucial for the maintenance of osmolality gradients and for effective collecting duct water reabsorption. Interestingly, Grhl2 deficiency (unlike deficiencies of individual claudins) was not associated with abnormalities of Cl− and Na+ reabsorption. It needs to be acknowledged that Grhl2 functions are not restricted to the effects on the TJ, which may explain the difference in phenotypes. In addition, it is possible that the barrier defect of Grhl2-deficient collecting duct cells is more profound, leading to non-ion-selective leakage of interstitial osmolytes into hypotonic urine.

#### **5. Aldosterone and Its Role in Transcellular and Paracellular Transport Regulation**

In the renal collecting duct, Na+ transport is separated from Cl− transport [45]. Na+ reabsorption from the intraluminal urine occurs transcellularly via the epithelial sodium channel (ENaC) that locates to the apical membrane of collecting duct principal cells. This generates a lumen-negative potential, providing a driving force for K+ secretion via the renal outer medullary potassium channel (ROMK). In contrast, Cl− reabsorption in the collecting duct occurs predominantly via the paracellular route. This "chloride shunt" is important to limit the built-up of a lumen-negative potential and, thereby, facilitates continued Na+ reabsorption and prevents excessive K+ secretion.

The key regulator of ENaC is aldosterone, a hormone secreted from the adrenal gland in response to hyperkalemia and hypovolemia [11,46]. Overall, aldosterone plays a central role in blood pressure regulation by controlling plasma Na+ and K+ levels and thus indirectly influences water retention or loss. However, growing evidence supports the hypothesis that aldosterone controls Na+ reabsorption and K+ secretion not only by regulating the abundance of ENaC and increasing transcellular transport but also by adjusting paracellular Cl− permeability in multiple ways:

For instance, aldosterone triggers the expression of Cldn-8 when ENaC is active, presumably to seal the paracellular route for Na+ back flux. Thereby net flux of Na+ can be increased [28].

Additionally, aldosterone regulates channel-activating protease 1 (Cap1) [47]. Cap1, in turn, stimulates ENaC and inhibits the Cl− conductivity by directly regulating Cldn-4 *trans*-interactions [23,48]. Thereby K+ secretion is favored over Cl− reabsorption (Figure 2a).

Aldosterone also induces the phosphorylation and activation of the with no lysine kinases 4 (WNK4) [49]. Expression of WNK4 in MDCK II cells has been shown to reduce the TER and increase Cl− permeability, without affecting TJ structure [50]. Activated WNK4 phosphorylates Cldn-4 on threonine residues decreasing the cells' TER and increasing apical to basal anion passage [51].

In the absence of aldosterone, WNK4 inhibits ENaC and ROMK activity and thus directly opposes Cap1 [52,53]. WNK4 phosphorylation suspends this inhibition (Figure 2b). Taken together, this indicates the possibility that aldosterone might regulate claudins through Cap1 and WNK4, coordinating Cl− reabsorption or K+ secretion, respectively.

Kahle and colleagues proposed WNK4 as the functional switch regulating Na+ and Cl− reabsorption independently from K+ secretion depending on the physiological conditions [53]. However, the factors facilitating different functional states of WNK4 haven't been provided, and to our knowledge, it is still unknown how the reverse actions of Cap1 and WNK4, that are both mediated by aldosterone, are regulated to decide in favor of K+ secretion or Cl− reabsorption, respectively.

In diabetes, the role of aberrant aldosterone signaling in the progression of renal disease has long been established [54,55]. Mediated by the divergent aldosterone levels, Cldn-4 and Cldn-8 are overexpressed in the distal nephron from type 1 diabetic rats, and the expression of WNK4 and its co-localization with Cldn-4 and Cldn-8 is also increased [56,57]. This might result in increased activation of Cldn-4 and Cldn-8 by WNK4 under diabetic conditions and could implicate disturbed paracellular transport in renal disease progression. However, additional experimental evidence verifying this hypothesis is needed.

**Figure 2.** (**a**) Aldosterone-mediated induction of channel-activating protease 1 (Cap1) activates Na+ reabsorption, but simultaneously inhibits the paracellular "chloride shunt", resulting in excessive K+ excretion. When aldosterone is secreted (1), it triggers the expression of Cap1 in principal cells of the renal collecting duct (2). Cap1 inhibits Cl− reabsorption directly by disrupting *trans*-interactions of Cldn-4. Simultaneously, Cap1 activates apical Na+ channels (ENaC), and thus, increases the transcellular reabsorption of Na+ (3). Consequently, the growing luminal negative potential drives K+ secretion into the urine via renal outer medullary potassium (ROMK) channels (4). (**b**) Aldosterone activates ENaC, ROMK, and the paracellular "chloride shunt" via with no lysine kinases 4 (WNK4) phosphorylation. When aldosterone is secreted (1), it leads to the phosphorylation of WNK4. This suspends the tonic inhibition of ROMK and ENaC by WNK4 (2). Furthermore, WNK4 phosphorylates claudins located to the tight junctions (TJs) of the renal collecting duct and thereby increases Cl− reabsorption (3). Green arrows indicate positive regulation, red lines indicate inhibition.

#### **6. Chloride Reabsorption in Renal Collecting Ducts and Potential Involvement in Disease**

The aldosterone-sensitive distal nephron is the main side of Cl− reabsorption. It occurs via multiple ways, including paracellular transport via TJs as well as transcellular pathways in intercalated cells [12,23,29] and is driven by the lumen-negative transepithelial potential generated by the unilateral Na+ reabsorption to maintain electroneutrality (Figure 3). As in aldosterone signaling, it becomes increasingly evident that crosstalk between paracellular and transcellular transport occurs.

The importance of efficient Cl− reabsorption becomes obvious in claudin-deficient mouse models. For example, Cldn-7 deficiency in mice is lethal within 12 days after birth due to severe salt-wasting and subsequent chronic dehydration (see above) [27]. Cldn-7−/<sup>−</sup> mice show reduced ROMK and increased ENaC, Aqp2, and Na+ Cl− cotransporter mRNA. These changes in channel expressions are probably due to a compensatory mechanism to inhibit further urinary loss of salt and water [27].

It has long been established that accumulation of luminal Cl− depolarizes the membrane, and thereby, inhibits the apical Na+ channel ENaC [58]. Several studies also have demonstrated that pathological increases in Cl− reabsorption are associated with diseases, such as pseudo-hypoaldosteronism type II (PHA-II) or Gordon's syndrome.

**Figure 3.** Chloride transport in the renal collecting duct. Chloride is transported via the paracellular route mediated by tight junctions (TJs), as well as, via the transcellular route using transporters and channels in α- and β-intercalated cells (depicted in bold).

PHA-II is a rare Mendelian syndrome leading to hypertension, hyperkalemia, and metabolic acidosis [59]. These symptoms are the exact opposite of the Cldn-4 and Cldn-8 knockout phenotypes in mice [23,29].

Interestingly, a long-known causative gene for PHA-II is WNK4 [60], and WNK4 can regulate Cl− conductance presumably by phosphorylation of Cldn-4 and Cldn-8 [50]. The PHA-II-causing mutation of WNK4 increases paracellular Cl− permeability in vitro [49,50]. Consistent with the proposed Cl− pore activity of Cldn-4 and Cldn-8, this is an additional indicator of an activating effect of claudin phosphorylation by WNK4. Furthermore, a recent study demonstrates that WNK4 is overexpressed in Cldn-7-deficient cultured, collecting duct cells [31]. Conversely, the PHA-II mutant of WNK4 is associated with increased Cldn-7 phosphorylation and enhanced paracellular Cl− conductivity [26]. This adds a new aspect and raises the possibility that claudins might act both up- and down-stream of WNK4 to regulate paracellular transport.

Another gene associated with PHA-II is KLHL3, encoding Kelch-like protein 3 [61,62]. Interestingly, KLHL3 normally induces ubiquitination and degradation of Cldn-8, while disease-associated mutations of KLHL3 abolish the interaction of KLHL3 with Cldn-8 [29]. In addition, KLHL3 leads to the ubiquitination of WNK4. In line with PHA-II symptoms, loss of KLHL3 increases Cl- permeability in vivo [29], which may contribute to the disease phenotype.

#### **7. Conclusions**

Collecting duct epithelia express claudins enforcing a high TER such as Cldn-3, -4, -7, -8, and -19, consistent with a demand for strong epithelial barrier function in the presence of steep transepithelial gradients. Based on our current knowledge, these claudins are thought to either support a tight barrier in general or to act as a cation barrier and/or an anion channel.

Consequently, the renal collecting duct is comprised of an especially tight epithelial barrier compared to other, more upstream segments of the nephron. As a result, the collecting duct lumen and interstitium are strictly separated. Nevertheless, the TJs of the collecting duct exhibit a regulated paracellular permeability for ions such as Cl-. Hence, the collecting duct TJ participates in a range of physiological functions:

It allows for controlled paracellular transport of Cl- to the interstitium, which is important in the setting of Na+ reabsorption via ENaC and secretion of K+ via ROMK.

It prevents paracellular diffusion and back flux of osmolytes into the urine and thus promotes the formation of steep gradients across the epithelium, which are necessary to drive electrolyte-free water reabsorption.

Future research will be needed to elucidate further the precise mechanisms that regulate TJ properties and that mediate crosstalk between paracellular and transcellular transport processes and how these processes relate to renal pathophysiology.

**Author Contributions:** Writing—original draft preparation: J.L., K.M.S.-O.; Writing—review and editing: J.L., K.M.S.-O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the German Research Foundation (DFG) Research Training Group GRK 2318, by Collaborative Research Grant 1365, and by Research Unit FOR 2841. We acknowledge support from the German Research Foundation (DFG) and the Open Access Publication Fund of Charité-Universitätsmedizin Berlin.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
