The laminin-binding integrins regulate nuclear factor κB-dependent epithelial cell polarity and inflammation

The kidney collecting system develops from the ureteric bud by undergoing iterative branching events after interacting with the metanephric mesenchyme (MM), which gives rise to the nephrons (Carroll and Das, 2013). This process requires growth factor-mediated cell signaling and integrin-dependent cell-extracellular matrix (ECM) interactions (Mathew et al., 2012).

Integrins, the principal ECM-binding receptors, consist of non-covalently bound α and β transmembrane subunits. The 24 integrins found in mammals are made up from 18 α and eight β subunits that combine in a restricted manner to form specific αβ heterodimers with different ligand binding properties (Pozzi and Zent, 2003, 2011). Integrins link the ECM with the intracellular actin cytoskeleton and function as cell signaling molecules. Traditionally, they are thought to promote cell adhesion, migration and proliferation; however, more recently, integrins were shown to regulate epithelial cell inflammation (De Arcangelis et al., 2017; Niculescu et al., 2011; Plosa et al., 2020, 2014) and terminal differentiation (Vijayakumar et al., 2008). Integrins α3β1, α6β1 and α6β4 act as the major receptors for laminins (Yamada and Sekiguchi, 2015). The principal laminins found in the developing and adult collecting system of the kidneys are laminin-111, -511, -521 and -332 (Yazlovitskaya et al., 2015). Preferred ligands for integrin α3β1 include laminin-511, -521 and -332. Integrin α6β1 is the most promiscuous laminin receptor and binds laminin-111, -511, -521 and -322, and integrin α6β4 preferentially binds to laminin-332; however, it also interacts with laminin-511 and -521 (Delwel et al., 1994; Niessen et al., 1994; Viquez et al., 2017; Yazlovitskaya et al., 2015; Yazlovitskaya et al., 2019).

The contribution of the integrins in the developing kidney collecting system is incompletely understood. Kidney collecting duct epithelial cells grow on a basement membrane that consists of both collagen IV and laminins. Deletion of the integrin β1 subunit, which results in loss of expression of all αβ1 integrin heterodimers in the developing ureteric bud, resulted in a severe branching morphogenesis phenotype due to decreased epithelial cell adhesion, migration and proliferation (Zhang et al., 2009). Identification of the specific α subunits that regulate this process is inconclusive. The collagen binding integrins α1β1 and α2β1 do not contribute to kidney collecting system development (Mathew et al., 2012). Surprisingly, deletion of the integrin α3 subunit in the kidney collecting system only caused a mild branching morphogenesis phenotype (Kreidberg et al., 1996; Liu et al., 2009; Yazlovitskaya et al., 2015), and deletion of the α6 subunit (resulting in no expression of α6β1 and α6β4) did not result in any branching morphogenesis phenotype (Viquez et al., 2017).

To determine whether integrin α3 and the α6 integrins cooperate in kidney collecting duct development, we deleted both the α3 and α6 integrin subunits in the developing mouse ureteric bud. Surprisingly, the mice showed a mild renal developmental phenotype, but they all died by 14 months of age from severe inflammation around the collecting ducts and tubulointerstitial fibrosis mediated by epithelial cells that have acquired a more mesenchymal phenotype. We used α3/α6-null (Itgα3^−/−α6^−/−) collecting duct cells to show that the inflammatory and mesenchymal phenotypes were mediated by increased activation of the nuclear factor kappa-B (NF-κB) transcription factor pathway that induced pro-inflammatory cytokine production and cadherin switching. Thus, laminin-binding integrins play a key homeostatic role in kidney collecting duct cells by suppressing NF-κB-dependent signaling.

We previously deleted Itgα3 in the developing ureteric bud, which resulted in a mild branching morphogenesis defect that became evident at embryonic day (E)15.5 and persisted into adulthood (Yazlovitskaya et al., 2015). By contrast, deleting Itgα6 in the ureteric bud did not cause a developmental phenotype (Viquez et al., 2017). In this study, we deleted the integrin α3 and α6 subunits in the developing ureteric bud by crossing the Itgα3/6^flox/flox (also referred to here as α3^f/fα6^f/f) mice with Hoxb7Cre mice. This strategy successfully deleted these genes, resulting in decreased expression of the integrin α3 and α6 subunits in the pelvis of the kidney of 1-week-old mice, as assessed by immunoblotting. There was also a slight decrease in β1 integrin subunit expression; however, it was still robustly expressed (Fig. 1A). We confirmed β1 integrin expression in the aquaporin 2-expressing collecting ducts by immunofluorescence (Fig. 1B,C). When we performed a morphological examination of the kidneys from E15.5 and newborn mice, there appeared to be a mild branching morphogenesis defect in the Hoxb7cre:Itgα3/6^flox/flox mice (Fig. 1D-G). We verified there were significantly less aquaporin 2-expressing collecting ducts in the kidneys of newborn Hoxb7cre:Itgα3/6^flox/flox mice by immunostaining followed by quantification. (Fig. 1H-N). Thus, a mild branching morphogenesis defect was confirmed in Hoxb7cre:Itgα3/6^flox/flox mice at birth.

We next assessed the long-term consequences of this mild branching defect on Hoxb7cre:Itgα3/6^flox/flox mice. The mice lost weight as they aged, started to die at about 200 days old and were all dead or euthanized by 480 days (Fig. 2A). Their median survival was 234 days, whereas there was no mortality in littermate controls over the same period. To identify the etiology for the increased mortality, we euthanized the mice at 3 months of age. The Hoxb7cre:Itgα3/6^flox/flox kidneys displayed a hypoplastic dysplastic papilla and there was marked inflammation surrounding the cortical collecting ducts (Fig. 2B-G, arrows). Under higher power, severe abnormalities of the tubular architecture, characterized by increased luminal cellularity, abnormal epithelial cell morphology, tubular obstruction and tubulointerstitial fibrosis were noted (Fig. 2H,I). As we aged the mice to over 1 year, the kidney phenotype progressed and large areas within the kidney tubulointerstitium were filled with closely packed inflammatory cells (Fig. 2J,K). The tubular cells were severely dysmorphic and appeared mesenchymal, and there was complete disorganization of the tubules (Fig. 2L,M). The mutant kidneys also displayed increased fibrosis, as verified by positive Picrosirius Red staining (Fig. 2N,O). Despite the major morphological abnormalities, there was normal expression of the laminin α3 and α5 chains in the papilla of the kidney (Fig. 2P-S). There were also no differences in laminin staining when we used a pan-laminin antibody (data not shown). Thus, the defects were not due to diminished secretion of the major laminins (111, 511, 521 and 332) found in the kidney collecting system. Taken together, these data show that although the Hoxb7cre:Itgα3/6^flox/flox mice only had a mild ureteric bud branching defect at birth, they developed progressive tubular pathology, tubulointerstitial fibrosis and severe inflammation.

To investigate whether the collecting duct cells were more mesenchymal in the Hoxb7cre:Itgα3/6^flox/flox mice compared to Itgα3/6^flox/flox mice, we performed immunostaining of the collecting ducts and immunoblotting of the dissected pelvis of the kidney for epithelial and mesenchymal cell markers. There was a marked decrease in the expression of E-cadherin in the collecting ducts [depicted by positive staining for Dolichos biflorus agglutinin (DBA)] of 6-month-old Hoxb7cre:Itgα3/6^flox/flox mice, and by 1 year there was virtually no expression in these animals. This loss of E-cadherin expression in aged mice was verified by immunoblotting (Fig. 3A,D). Consistent with the decreased E-cadherin expression there was increased expression of N-cadherin and α-smooth muscle actin (α-SMA, also known as ACTA2) in Hoxb7cre:Itgα3/6^flox/flox mice collecting ducts, which was confirmed by immunoblotting (Fig. 3B-E). Thus, deleting the laminin-binding integrins in the kidney collecting system caused collecting duct cells to lose epithelial markers and acquire a mesenchymal phenotype.

As histological examination of aged Hoxb7cre:Itgα3/6^flox/flox mouse kidneys suggested increased inflammation, we stained for macrophages (CD68) and lymphocytes (CD3) on kidney slides of 1-year-old mice. There were approximately five and 15 times more macrophages and lymphocytes, respectively, in the Hoxb7cre:Itgα3/6^flox/flox mice compared to controls (Fig. 4A-F). We next examined the immune cell population in the whole kidney using flow cytometry by gating on cells that express leukocyte common antigen (CD45). We found that the CD45⁺ population accounted for 7% of the cells in the kidneys of Hoxb7cre:Itgα3/6^flox/flox mice but for only 4.2% in the Itgα3/6^flox/flox mice (Fig. 4G,H), suggesting there were excessive leukocytes in the Hoxb7cre:Itgα3/6^flox/flox kidneys. When we assessed the percentages of neutrophils (GR1⁺), macrophages (F4/80⁺) or lymphocytes (CD3⁺) in the CD45⁺ leukocytes in the Hoxb7cre:Itgα3/6^flox/flox and Itgα3/6^flox/flox mice, no differences were found (Fig. 4I). Thus, there are excessive inflammatory cells within the Hoxb7cre:Itgα3/6^flox/flox kidney with no alteration in the cellular lineages.

To better determine how the lack of integrin α3 and α6 subunits leads to increased inflammation, we used previously generated Itgα3^−/−α6^−/− collecting duct cells, which displayed severe deficiencies in adhesion, migration and proliferation on laminins (Yazlovitskaya et al., 2019). We initially investigated whether there was a defect in cell polarization and cell-cell tightness. Both Itgα3^f/fα6^f/f and Itgα3^−/−α6^−/− collecting duct cells cultured on Transwells formed monolayers. When we examined the expression levels and localization of two key components of epithelial cell-cell interaction, the expression of E-cadherin, an adherens junction component, was markedly decreased in the Itgα3^−/−α6^−/− collecting duct cells and was primarily present in the cytoplasm rather than on the cell membrane (Fig. 5A-D,I,J). Levels of ZO-1 (also known as TJP1), a tight junction-associated protein, were significantly higher in the Itgα3^−/−α6^−/− collecting duct cells relative to the Itgα3^f/fα6^f/f collecting duct cells; however, like E-cadherin, this tight junction marker was not localized at the cell membrane (Fig. 5E-J). Consistent with the alterations in expression and/or localization of these epithelial markers, there was a significant increase in expression of the well-known marker of myofibroblast differentiation α-SMA, as well as N-cadherin and vimentin (Fig. 5I,J), which are more typically expressed by mesenchymal rather than polarized epithelial cells. Epithelial cells with a more mesenchymal phenotype produce profibrotic collagen I, whereas differentiated polarized epithelial cells do not (Hosper et al., 2013). Consistent with this finding, Itgα3^−/−α6^−/− collecting duct cells express significantly more collagen I than Itgα3^f/fα6^f/f collecting duct cells (Fig. 5I,J). Thus, deleting the α3 and α6 integrin subunits in collecting duct cells causes a major polarity defect, with the cells acquiring mesenchymal features and expressing more collagen I.

We next investigated the functional consequences of the lack of well-formed adherens junctions in the Itgα3^−/−α6^−/− collecting duct cells in order to explain the increased inflammation observed in the Hoxb7cre:Itgα3/6^flox/flox mice. We measured the transepithelial electrical resistance (TEER) (Fig. 6A) and inulin leak, which characterize the tightness of the confluent monolayer, (Fig. 6B) in collecting duct cells grown on Transwell filters that were not coated with extracellular matrix for 96 h. TEER was significantly decreased, whereas the inulin leak was significantly increased in the Itgα3^−/−α6^−/− collecting duct cells compared to the Itgα3^f/fα6^f/f collecting duct cells, consistent with impaired epithelial cell-cell interactions. We next investigated the ability of a murine macrophage cell line (RAW) to transmigrate through a confluent monolayer of Itgα3^f/fα6^f/f versus Itgα3^−/−α6^−/− collecting duct cells grown on Transwell filters. Approximately 60% more RAW cells transmigrated towards regular Dulbecco's minimum Eagle medium (DMEM) when plated on top of confluent Itgα3^−/−α6^−/− collecting duct cells compared to those on Itgα3^f/fα6^f/f collecting duct cells (Fig. 6C). This suggests that the confluent monolayer of Itgα3^−/−α6^−/− collecting duct cells was less tight and more permeable for macrophage transmigration.

In addition to increased permeability of the renal epithelium, another possible mechanism for the severe inflammation in the Hoxb7cre:Itgα3/6^flox/flox mouse kidney is that the collecting duct cells lacking α3 and α6 integrin subunits secrete excessive pro-inflammatory cytokines. Therefore, we placed RAW cells on top of either Itgα3^−/−α6^−/− or Itgα3^f/fα6^f/f collecting duct cells grown on Transwell filters and placed 24-h-conditioned medium from either Itgα3^−/−α6^−/− or Itgα3^f/fα6^f/f collecting duct cells in the lower chamber. The RAW cell migration was significantly higher through the Itgα3^−/−α6^−/− collecting duct than the Itgα3^f/fα6^f/f collecting duct cell monolayer, irrespective of the conditioned medium (Fig. 6D). Furthermore, there was significantly more migration towards the conditioned medium of the Itgα3^−/−α6^−/− collecting duct cells in both situations. Thus, Itgα3^−/−α6^−/− collecting duct cells have barrier function defects, and they may produce chemokines that promote RAW cell transmigration.

The increased transmigration of the RAW cells towards the conditioned medium of Itgα3^−/−α6^−/− collecting duct cells suggested that they secrete chemotactic factors. Therefore, we performed a multiplex analysis for chemokines on the conditioned medium of Itgα3^f/fα6^f/f and Itgα3^−/−α6^−/− collecting duct cells. Ten of 32 cytokines (31%), including mediators of macrophage chemotaxis, were significantly increased in the conditioned medium of Itgα3^−/−α6^−/− collecting duct cells compared to Itgα3^f/fα6^f/f collecting duct cells (Fig. 7A; Table S1), and many of them were recognizable gene products of NF-κB activation (including G-CSF, IL-6, KC and MIP-2). The NF-κB transcription factor complex is held in an inactive state in the cytoplasm by binding to its inhibitor IκB, which is ubiquitylated upon activation. This allows for phosphorylation of the NF-κB complex at several sites, among them serine 536 (Mattioli et al., 2004), which facilitates its shuttling into the nucleus where it can bind to promoter sites that regulate transcription of target genes (Hoffmann and Baltimore, 2006). We assessed the activity status of NF-κB in Itgα3^−/−α6^−/− and Itgα3^f/fα6^f/f collecting duct cells by performing immunoblotting for p65 subunit in the cytosolic and nuclear fractions, and found that the total amount (data not shown) and nuclear p65 levels were significantly higher in the Itgα3^−/−α6^−/− collecting duct cells compared to the Itgα3^f/fα6^f/f collecting duct cells (Fig. 7B,C). Consistent with this, there was increased serine 536 phosphorylation of p65 in the cytoplasmic fraction of Itgα3^−/−α6^−/− collecting duct cells (Fig. 7B,D). We then verified NF-κB activity using an NF-κB luciferase reporter that contains a firefly luciferase gene under the control of a multimerized NF-κB responsive element located upstream of a minimal promoter. There was an almost twofold increase in the basal NF-κB activity in the Itgα3^−/−α6^−/− collecting duct cells compared to the Itgα3^f/fα6^f/f collecting duct cells. (Fig. 7E). BMS-435541 is a selective inhibitor of the IκB kinase (IKK) catalytic subunits, which inactivates NF-κB by sequestering it inactive in the cytoplasm due to masking the NF-κB nuclear localization signals. Treatment of Itgα3^−/−α6^−/− collecting duct cells with BMS-435541 reduced NF-κB activity in these cells to the levels detected in α3^f/fα6^f/f collecting duct cells (Fig. 7F). We also determined NF-κB activity in vivo by staining kidneys of 12-month-old mice with DBA, to mark the collecting ducts, and an antibody directed against serine 276 of p65 (pp65), which indicates the active form of NF-κB. Approximately 75% of collecting duct cells of Hoxb7cre:Itgα3/6^flox/flox mice had positive staining for nuclear pp65, whereas this was rarely seen in the Itgα3/6^flox/flox mice (Fig. 7G,H). Thus, deleting the laminin receptors in the collecting ducts and collecting duct cells results in significant activation of NF-κB.

One of the key characteristics of the Itgα3^−/−α6^−/− collecting duct cells is that they exhibit more mesenchymal features than the Itgα3^f/fα6^f/f collecting duct cells. The major transcription factors regulated by NF-κB that control epithelial-to-mesenchymal transitions (EMTs) are Snail and Slug (also known as Snai1 and Snai2, respectively), which accumulate in the nucleus when activated. Therefore, we investigated their activation by determining their nuclear localization (Wu et al., 2009). Immunoblots of cytosolic fractions of Itgα3^−/−α6^−/− and Itgα3^f/fα6^f/f collecting duct cells did not show any difference in cytosolic Snail and Slug; however, higher levels of both transcription factors were present in the nuclear fraction of Itgα3^−/−α6^−/− collecting duct cells (Fig. 8A-C). The total levels of Snail and Slug were increased in the mutant cells (data not shown). We next defined whether downregulating the p65 subunit of NF-κB by small interfering (si)RNA affected the amount and localization of Snail or Slug in the Itgα3^−/−α6^−/− collecting duct cells. This manipulation decreased both p65 phosphorylation at serine 536 and total p65 in the nuclear and cytosolic fractions. Consistent with the decreased expression of p65, there was a significant decrease in both Snail and Slug in the nuclear fraction of the Itgα3^−/−α6^−/− collecting duct cells (Fig. 8D-H). These data suggest that increased NF-κB signaling in the Itgα3^−/−α6^−/− collecting duct cells caused the mesenchymal phenotype by increasing Snail and Slug transcriptional activity.

We next determined the effects of downregulating NF-κB activity on the mesenchymal phenotype observed in the Itgα3^−/−α6^−/− collecting duct cells by analyzing the levels of adherens junction markers, as well as collagen I. The levels of E-cadherin were increased, whereas the levels of N-cadherin were decreased in the Itgα3^−/−α6^−/− collecting duct cells treated with p65 siRNA (Fig. 8I-M). Consistent with these results, decreased collagen I production occurred in p65 siRNA-treated Itgα3^−/−α6^−/− collecting duct cells. These results suggest that the elevated NF-κB activity in the Itgα3^−/−α6^−/− collecting duct cells is a major contributor to their mesenchymal phenotype and excessive collagen I production.

α3β1 and the α6 integrins functionally cooperate in cell adhesion, migration, spreading and signaling on laminin-511 and laminin-332 substrates (Yazlovitskaya et al., 2019). They are highly expressed in the developing ureteric bud and, based on in vitro studies, were thought to be critical for its development (Chen et al., 2004; Zent et al., 2001). However, deleting either integrin α3 or α6 subunits individually during ureteric development in vivo resulted in minor or no developmental phenotypes, suggesting redundancy of these integrins (Kreidberg et al., 1996; Liu et al., 2009; Viquez et al., 2017; Yazlovitskaya et al., 2015). Surprisingly, when we deleted the α3 and α6 subunits there was only a mild ureteric bud developmental phenotype that was indistinguishable from integrin α3-null mice (Kreidberg et al., 1996; Liu et al., 2009; Yazlovitskaya et al., 2015). However, these mice died at about 14 months of age due to severe inflammation, tubulointerstitial fibrosis and intratubular obstruction around the kidney collecting ducts. Mechanistically, Itgα3/α6-null collecting duct cells acquired a mesenchymal phenotype and produced excessive collagen and inflammatory cytokines, which was mediated by excessive NF-κB transcription factor pathway activation. Thus, although the laminin-binding integrins play a minor role in ureteric bud development, they are required to maintain epithelial cell polarity and downregulate epithelial cell inflammation by controlling NF-κB-dependent activity in kidney collecting ducts during normal homeostasis.

It was surprising that the deletion of both the integrin α6 and α3 subunits did not worsen ureteric bud branching morphogenesis relative to the deletion of the integrin α3 subunit only. This contradicts in vitro studies that show the additive effects of blocking these integrins in collecting duct epithelial cell adhesion, migration, tubulogenesis and ureteric bud development in organ culture (Chen et al., 2004; Zent et al., 2001). However, these data are consistent with observations in the global integrin α3/α6-null mouse in which the developing ureteric buds were almost normal at E16.5 when the mice died, despite having partial or complete absence of the ureters (De Arcangelis et al., 1999). The ureteric bud branching phenotype resulting from α3/α6 deletion was far less severe than the branching defect found in mice with ureteric bud deletion of the β1 integrin subunit using the same Hoxb7 cre mouse, which deletes β1-containing integrins that bind collagens, RGD-containing ligands and laminins (Zhang et al., 2009). Thus, in the context of the whole animal, the laminin-binding integrins play a minor role in ureteric bud development, which leaves open the question as to which integrin-ECM interactions are required for ureteric bud development. It is not integrin-collagen interactions as the integrin α1- and α2-null mice are normal (Mathew et al., 2012), and to the best of our knowledge there are no studies demonstrating a role for the RGD-binding integrins (α5β1 and αv containing integrins) in ureteric bud branching morphogenesis in vivo. Thus, deletion of more than one class of ligand-binding integrin is likely required to induce a major ureteric bud branching morphogenesis defect.

Despite the minimal ureteric bud branching morphogenesis defect, the Itgα3^−/−α6^−/− collecting ducts lost many of their epithelial characteristics and acquired a mesenchymal phenotype. These changes are like those found in the developing integrin α3/α6 subunit-null mouse mammary gland in which baso-apical epithelial cell polarization is disturbed and the number of myoepithelial cells is increased (Romagnoli et al., 2020). Laminins have been shown to be the critical component of the basement membrane that enables mammary gland epithelial cells to polarize and suppress EMT via the α6-containing integrins, whereas fibronectin induces EMT via integrin α5β1 (Chen et al., 2013; Park and Schwarzbauer, 2014; Williams et al., 2008). Thus, the interaction between laminins and laminin-binding integrins appears to be critical for normal tubular epithelial cell differentiation in different branched organs.

One of our major findings in the Hoxb7cre:Itgα3/6^flox/flox mice was the chronic inflammation with a macrophage and lymphocyte predominance around the kidney tubules that was associated with progressive fibrosis. These data are consistent with previous studies demonstrating that the deletion of the integrin α6 subunit in the gut results in colitis, which degenerates over time into infiltrating adenocarcinoma. The mechanism was that hemidesmosome disruption resulted in intestinal epithelial cells detaching from the basement membrane, which caused them to over produce IL-18, resulting in hyperplasia and enhanced intestinal permeability (De Arcangelis et al., 2017). In addition, postnatal deletion of the α6 integrin subunit in the skin resulted in chronic inflammation mediated by amphiregulin and epiregulin (Niculescu et al., 2011). A similar phenotype, characterized by chronic inflammation and emphysema, was present when the integrin β1 subunit was deleted in type 2 alveolar epithelial cells of the postnatal lung (Plosa et al., 2020). Likewise, when β1 integrin was deleted from the acinar, ductal and islet cells of the developing pancreas the mice survived but developed chronic pancreatitis with inflammatory infiltrates and mild polarity defects (Bombardelli et al., 2010). Thus, the deletion of integrins in epithelial structures that do not acquire fatal developmental phenotypes develop inflammation over time, which suggests that epithelial cell integrins play a role in suppressing this vital component of chronic injury.

We used cell culture to explore the mechanisms whereby expression of the laminin-binding integrins regulate inflammation around the collecting ducts. The Itgα3^−/−α6^−/− collecting duct cells lost epithelial polarity, which allowed for significantly more macrophage transmigration across the monolayer. Tight and adherens junctions have been shown to be critical for leukocyte restriction in multiple organs, including the lung (Burns et al., 2003), gastrointestinal tract (Brazil et al., 2019; Ginzberg et al., 2001; Luissint et al., 2016) and uroepithelium (Parkos, 2016). There is also evidence that the transmigrating leukocytes induce disruption of barrier function (Brazil et al., 2019; Parkos, 2016). Thus, it is likely that a vicious cycle is set up in Hoxb7cre:Itgα3/6^flox/flox mice, whereby the polarity defect increases leukocyte transmigration, which in turn causes epithelial cell injury and further increases tubule permeability. In addition to exhibiting abnormal permeability, Itgα3^−/−α6^−/− collecting duct cells secreted high levels of inflammatory cytokines. Although this played a direct role in the increased macrophage transmigration, its effects were less profound than those mediated by the permeability defect. Increased cell permeability and excessive cytokine production was also present when the β1 integrin subunit was deleted from type 2 alveolar epithelial cells (Plosa et al., 2020).

Another key finding in our study is that loss of the integrin α3 and α6 subunits resulted in excessive NF-κB activity in the collecting duct and collecting duct cells. The role of NF-κB in kidney tubules is poorly studied and is primarily described in the setting of acute kidney injury (Henke et al., 2007; Kusch et al., 2013). In this context, downregulation of NF-κB by expressing the human NF-κB super-repressor IκBαΔN in the renal proximal tubule resulted in amelioration of injury and decreased inflammation; and targeting IKK with siRNA, which resulted in decreased NF-κB activation, attenuated renal ischemia perfusion injury in rats (Markó et al., 2016; Wan et al., 2011). These data are consistent with our finding about where excessive NF-κB activation resulted in increased injury, inflammation and secretion of inflammatory cytokines. We also showed that the mesenchymal phenotype of the Itgα3^−/−α6^−/− collecting duct cells is due to NF-κB-dependent activation of Slug and Snail, leading to aberrant expression of adherens junctions proteins E-cadherin and N-cadherin, and accumulation of collagen I. NF-κB is an essential component of EMT and metastasis in cancer by inducing Snail stabilization by both transcriptional and posttranslational mechanisms that are highly conserved from fly to mammal (Huber et al., 2004; Skrzypek and Majka, 2020; Wu et al., 2009). The Snail genes in the kidney are highly regulated and their expression is decreased in different developmental stages so that essential epithelialization can occur (Grande et al., 2015). In addition, Snail genes are kept silent in the mature kidney as its aberrant activation is sufficient for the development of renal fibrosis (Grande et al., 2015). Thus, in Hoxb7cre:Itgα3/6^flox/flox mice, it is likely that the usually silenced Snail family of transcription factors are activated by NF-κB-dependent signaling, which causes fibrosis and EMT.

Although we show that lack of α3 and α6 integrin subunits leads to increased NF-κB activity, an outstanding question is how this activation is achieved. We speculate that it might be in part because of the loss of integrin α3β1 binding to the ubiquitin-modifying enzyme TRAF6, which plays a role in integrin α3β1-dependent Akt activation by regulating Akt K63-linked polyubiquitylation (Yazlovitskaya et al., 2015). TRAF6 is also critical for recruiting the ubiquitin-dependent kinase transforming growth factor beta-activated kinase 1 (TAK1) and its downstream kinase IKK to assemble a signaling complex that activates NF-κB (Li et al., 2005; Li and Verma, 2002; Shi and Sun, 2018). It is therefore possible that the deletion of integrin α3β1 increases the amount of TRAF6 available to assemble TAK1/IKK complexes, resulting in excess NF-κB signaling. Clearly, integrin α3β1 binding to TRAF6 is not the only mechanism in which NF-κB signaling is regulated, as the phenotype is only present when both integrin α3 and α6 subunits are deleted and not in mice deficient of only a single integrin subunit.

In conclusion, we have demonstrated that the laminin-binding integrins only play a minor role in regulating ureteric bud development in vivo. However, these integrins are critical for maintaining collecting duct homeostasis because they are required for the terminal differentiation of the collecting duct cells into a tight epithelium, and they play a crucial role in downregulating inflammatory cytokine production by these cells. These functions are mediated by modulating the NF-κB signaling pathway; a previously unrecognized critical function of integrins in polarized kidney tubule epithelial cells.

All animal experiments were approved by the Vanderbilt University Medical Center Institutional Animal Care and Use Committee. To generate mice in which the integrin α3 and α6 subunits were deleted in the developing ureteric bud, we intercrossed Itgα3^flox/flox (Sachs et al., 2006) and Itgα6^flox/flox (Marchetti et al., 2013). The homozygote double mutants were then crossed with a mouse that expressed Cre recombinase under the control of the Hoxb7 promoter (Hoxb7cre) (Kobayashi et al., 2005; a gift from Dr A. McMahon, University of Southern California, USA). Littermate Itgα3/α6^flox/flox mice that lacked Cre were used as controls. All mice are on the C57Bl/6 background. We aged Hoxb7cre:Itgα3/6^flox/flox mice and their littermate Itgα3/6^flox/flox controls until they either died or developed criteria requiring euthanasia. The survival differences were highly significant (P<0.002) according to both the Mantel–Cox and Gehan–Breslow–Wilcoxon tests.

For histological examination, paraffin kidney sections were stained with Hematoxylin and Eosin (H&E) or Picrosirius Red, using standard protocols. For immunohistochemistry stains, paraffin sections were incubated with the primary and secondary antibodies indicated below, followed by colorimetric detection by Vector Red (Vector Laboratories). Immunofluorescence staining was performed on frozen kidney sections that were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 and incubated with the antibodies described below. Images were collected by confocal microscopy (Leica TCS SPE or Zeiss LSM710, 780 or 880, objective 63×/1.4 NA Plan Apochromat oil). The following primary antibodies were used: anti-DBA (1:100; Vector Laboratories, FL-1031); anti-CD68 (1:500; Abcam, ab53444); anti-CD3 (1:500; Abcam, ab5690); anti-Gr1 (1:500; Abcam ab238132); anti-E-cadherin (1:20; Invitrogen, 13-1900); anti-phospho-p65 (S276) (1:500; Abcam, ab106129); anti-aquaporin 2 (1:50; Cell Signaling Technology, 3487); anti-β1 integrin (1:20 in Fig. 1; BD Pharmingen, 555003 Ha2/5); anti-laminin α5 (1:250; Thermo Fisher Scientific, BS1086R); and anti-laminin α3 (1:200; Thermo Fisher Scientific, PA38937). The following secondary antibodies were used at 1:250 dilution: anti-rabbit IgG Alexa Fluor 488 (Life Technologies, A21206); anti-rabbit IgG Alexa Fluor 594 (Life Technologies, A21207); anti-mouse IgG Alexa Fluor 594 (Life Technologies, A21203); and anti-rat IgG Alexa Flour 488 (Life Technologies A21208). Quantification of immunostained sections was performed on ten non-overlapping images per kidney from a total of three kidneys obtained with a 40× objective.

The kidney papillae from individual 3-day-old pups were isolated and lysed in T-PER reagent (Thermo Scientific, Waltham, MA, USA) with protease inhibitors (Sigma-Aldrich, P8340, St. Louis, MO, USA) and phosphatase inhibitor cocktails 1 and 2 (Sigma-Aldrich, P5726 and P0044, respectively), using a Polytron homogenizer. Protein extracts (20 µg) were subjected to western blot analysis and membranes were developed using the Western Lightning Chemiluminescence Plus detection system (PerkinElmer Cetus, Wellesley, MA, USA). The following primary antibodies were used: anti-α3 integrin (1:1000; R&D Systems, AF2787); anti-α6 integrin (1:1000; Cell Signaling Technology, 3750); anti-E-cadherin (1:1000; BD Transduction Labs, 61081); anti-N-cadherin (1:1000; Cell Signaling Technology, 14215); anti-α-SMA (1:4000; Cell Signaling Technology, 19245); anti-ZO-1 (1:4000; Invitrogen, 617300); anti-vimentin (1:1000; Cell Signaling Technology, 5741); anti-phospho-NFκB p65 (S536) (1:1000; Cell Signaling Technology, 3033); anti-NFκB p65 (1:1000; Cell Signaling Technology, 8242); anti-Snail (1:500; Abcam, 53519); anti-Slug (1:500; Abcam, 27568); anti-collagen I (1:1000; Millipore, AB765P); anti-Lamin A/C (1:2000; Cell Signaling Technology, 4777); and anti-β-actin (1:10,000; Proteintech, 66009).

Single-cell whole kidney suspensions were made using collagenase XI (Sigma-Aldrich, C7657, 0.7 mg/ml) and type IV DNAse (Sigma-Aldrich, D5025, 30 µg/ml) digestion, and were subsequently passed through a 40 µm filter. Briefly, cells were blocked with anti-CD32 antibody (1:50; BD Biosciences, 553142), incubated with conjugated primary antibody and analyzed using a 5-laser BD LSRII analytical flow cytometer (BD Biosciences) and FlowJo analysis software (BD). The following primary conjugated antibodies were used in flow cytometry experiments: CD45-FITC (1:50; Invitrogen, 11-0451-82); CD3-PE-Cy7 (1:50; BioLegend, 100220); CD11b-PerCP (1:50; BioLegend, 101230); F4/80-PE (1:25; Invitrogen, MF48004); and Gr1-APC (1:50; BD Biosciences, 557661).

Collecting duct cells were isolated from 5-6-week-old Itgα3/6^flox/flox mice and Itgα3/6^−/− collecting duct cells were generated as described previously (Yazlovitskaya et al., 2019). In some experiments, the collecting duct cells were treated with IκB kinase inhibitor BMS-34554 (100 µM in water/3% Tween 80) or the vehicle alone for 1 h. To silence NFκB p65, collecting duct cells were transfected with non-silencing siRNA (20 nM, transfection control) or p65-silencing siRNA [NFκB p65 siRNA (m), 29411, Santa Cruz Biotechnology] using Lipofectamine RNAiMAX according to manufacturer's instructions. Transfected cells were used 48 h later.

Total cell lysates were prepared using M-PER reagent with protease inhibitors and phosphatase inhibitor cocktails 1 and 2. The isolation of cytoplasmic and nuclear fractions was performed using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific, 78833) according to the manufacturer's instructions. Protein extracts (40 µg) were subjected to western blot analysis and developed using the Western Lightning Chemiluminescence Plus detection system according to the manufacturer's protocol. Densitometry was performed using ImageJ; protein levels were normalized to β-actin.

Tight/adherens junction tightness of collecting duct cells was measured as described previously (Basuroy et al., 2003). Resistance was measured (data presented as ohms/cm²) on confluent monolayers of collecting duct cells using an EVOM voltmeter (World Precision Instruments, Sarasota, FL, USA). Experiments were performed three times in quadruplicates.

Inulin leak was measured on monolayers of collecting duct cells grown on Transwell filters with 8 μm pores as described previously (Basuroy et al., 2003; Elias et al., 2014). After fluorescein isothiocyanate-conjugated inulin was applied to the upper chamber, medium from the upper and lower chambers was sampled at the indicated time points. Fluorescence was detected using a Fluoroscan plate reader, and inulin leak was calculated as flux/h/cm². Experiments were performed three times in triplicates.

Monolayers of Itgα3^f/f/6^f/f or Itgα3^−/−/α6^−/− collecting duct cells were grown on Transwell filters with 3 µm pores (upper chamber). RAW 264.7 macrophages were placed on top of the collecting duct cells monolayers and they migrated through the collecting duct cells towards the medium in the lower chamber. The medium in the lower chamber was either regular growth medium [DMEM with 10% fetal bovine serum (FBS)] or conditioned medium collected from either Itgα3^f/f/6^f/f or Itgα3^−/−/α6^−/− collecting duct cells grown in confluence for 24 h. Macrophages that migrated to the underside of the Transwell filter after 6 h were fixed, stained, imaged in six non-overlapping sections and quantified as the number of migrated cells per high power field (HPF).

The cytokine/chemokine Magnetic Bead 32-Multiplex Panel (Millipore, MCYTMAG-70K-PX32) assay was performed on regular collecting duct cell growth medium conditioned for 24 h in triplicate according to the manufacturer's instructions. The assay was read on the Luminex MAGPIX platform in the Vanderbilt Hormone and Analytical Services Core.

Itgα3^f/f/6^f/f or Itgα3^−/−/α6^−/− collecting duct cells (50,000 cells/well) were cultured in a 96-well plate for 24 h until 90% confluence. Cells were transfected with a NanoLuc Reporter Vector with the NF-κB Response Element (pNL3.2NFκB-RE[NlucP/NFκB-RE/Hygro, Promega, N1111) using FuGENE HD transfection reagent (Promega E2311). In experiments with the NF-κB inhibitor, cells were treated with solvent (3% Tween 80 in regular growth medium) or 10 µM BMS-345541 in solvent for 1 h. NF-kB activity was detected using the Nano-Glo Luciferase Assay System (Promega, N1110) according to the manufacturer's instructions and is presented as fold change of NF-kB activity of Itgα3^f/fα6^f/f cells.

A two-tailed Student's t-test was used for comparisons between two groups with results representing mean±s.e.m. For comparisons between more than two groups, one-way ANOVA was used, with secondary analysis by Tukey's test for multiple comparisons as indicated. For both statistical analyses, P<0.05 was considered statistically significant. All statistical analyses were performed with the use of SigmaStat software (Systat Software, San Jose, CA, USA).

We dedicate this article to our colleague Elisabeth Georges-Labouesse.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259161.