Tubulointerstitial De Novo Expression of the α8 Integrin Chain in a Rodent Model of Renal Fibrosis – A Potential Target for Anti-Fibrotic Therapy?

In the normal kidney, the α8 integrin chain is expressed only on mesangial cells and vascular smooth muscle cells. α8 integrin ligates several matrix molecules including fibronectin, osteopontin and fibrillin-1. Recently, we detected de novo expression of α8 integrin on epithelial cells in renal cysts. We hypothesized that the α8 integrin chain is induced in tubular epithelia undergoing dedifferentiation and contributes to the fibrotic response in the tubulointerstitium (TI) after unilateral ureteral obstruction (UUO). After induction of UUO in rats by ligation of the right ureter, increased expression of the α8 integrin chain and its ligands was observed. In the TI, α8 integrin was localized to cytokeratin-positive epithelial cells and to interstitial fibroblasts; and colocalized with its ligands. In mice underexpressing α8 integrin UUO led to collagen deposition and fibroblast activation comparable to wild types. Mice lacking α8 integrin showed even more TI damage, fibroblast activation and collagen deposition after UUO compared to wild type mice. We conclude that the expression of the α8 integrin chain and its ligands is strongly induced in the TI after UUO, but underexpression of α8 integrin does not attenuate TI fibrosis. Mice lacking the α8 integrin chain are even more susceptible to TI damage than wild type mice. Thus, interactions of α8 integrin with its ligands do not seem to contribute to the development or progression of TI fibrosis in UUO. Targeting α8 integrin might not be a useful approach for anti-fibrotic therapy.


Introduction
Integrins, serving as receptors for extracellular matrix molecules, contribute to the maintenance of the structural and functional integrity of tissues and organs [1]. On the other hand, a dysregulated expression of integrins can also add to pathologic changes like tissue fibrosis: In a rat model of crescentic glomerulonephritis a1b1 integrin is overexpressed in glomeruli and the tubulointerstitium and treatment with an inhibitory antibody to a1b1 integrin reduced renal scarring and ameliorated renal survival [2]. Similarly, an inhibition of avb3 using specific RGD peptides resulted in a reduction of glomerulosclerosis during experimental Habu glomerulonephritis [3]. In Alport mice, the expression of avb6 integrin correlates with renal fibrosis and blocking of this integrin results in reduced deposition of collagen matrix [4]. The expression of the a5 integrin chain was induced in tubular epithelial cells after unilateral ureteral obstruction and facilitated epithelial to mesenchymal transition in these cells [5]. Thus, integrin signaling pathways seem to contribute to fibrotic changes in different renal diseases. For this reason integrins appear to be promising targets for more specific therapeutic approaches to prevent renal fibrosis [6]. On the other hand, integrins also serve to maintain tissue integrity. A complete lack of a1b1 integrin even leads to more severe glomerulosclerosis after experimental glomerular injury [7].
The a8 integrin chain dimerizes solely with the b1 chain to form a8b1 integrin, which is a receptor for fibronectin, vitronectin, tenascin C, osteopontin, nephronectin and fibrillin-1 [8][9][10][11][12]. In the normal kidney, the a8 integrin chain is expressed exclusively in vascular smooth muscle cells and mesangial cells [13]. The expression of a8 integrin is induced in various rat models of tissue fibrosis: in the scarring liver, lung and heart [14,15]. Own studies revealed an overexpression of a8 integrin in glomerulosclerosis [13]. Recently de novo expression of the a8 integrin chain in epithelial cells in the wall of renal cysts was observed [16]. We speculate that this de novo expression may be a more widespread feature of renal epithelial cells undergoing dedifferentiation. We therefore tested the hypothesis that the a8 integrin chain is expressed in dedifferentiating tubular epithelial cells during tubulointerstitial injury and that it contributes to the fibrotic response of the kidney. As a model of tubulointerstitial fibrotic injury, we performed unilateral ureteral obstruction (UUO) and studied a8 integrin expression patterns in this context. To investigate the functional role of a8 integrin for renal fibrosis, we induced UUO in mice with a deficiency for the a8 integrin chain.

Animal procedures
Animals were housed in a room maintained at 2262uC, exposed to a 12 hour dark/light cycle with free access to standard chow (#V1534, Sniff, Soest, Germany) and tap water. All procedures performed on animals were done in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the local government authorities (Regierung von Mittelfranken, approval number AZ # 54-2532.1-17/08) after evaluation of the local government's review board for animal research ethics. All surgery was performed under isoflurane or sodium pentobarbital anesthesia, and all efforts were made to minimize suffering. If judged necessary by a veterinarian, buprenorphine hydrochloride was injected to prevent of relief suspected pain or discomfort.
Male Sprague-Dawley rats of 180-220 g body weight were purchased from Charles River (Sulzfeld, Germany). For histological evaluations, 15 rats underwent unilateral ureteral obstruction. The right ureter was ligated with 4.0 silk at two separate locations. Rats were sacrificed by exsanguination under deep anesthesia at day 2 (n = 5), day 7 (n = 5) and day 14 (n = 5) after unilateral ureteral obstruction. The left kidneys were used as controls. At day 2 after ligation fibrotic changes were not significantly different from controls, while at day 7 after ligation fibrotic changes were clearly observed. At day 14 after ligation fibrotic changes observed were not essentially different from the changes detected at day 7. Thus, we chose the day 7 time point for further investigations.
One day before sacrifice, animals were kept in metabolic cages for 24 hours to collect urine. Blood samples were obtained at sacrifice. At the day of sacrifice, animals were equipped with a carotid artery catheter under general anesthesia and intraarterial blood pressure was measured in conscious animals 2 hours after anesthesia. Animals were exsanguinated in deep general anesthesia and the left ventricle and the kidneys were removed. Cortical tissue from each kidney was frozen in liquid nitrogen for RNA analyses. One part of the kidney was frozen in Cryoblock (medite Medizintechnik, Burgdorf, Germany) to prepare cryosections. Other parts of the kidney were fixed in methyl Carnoys solution or in paraformaldehyde for embedding in paraffin.

Measurement of physiologic parameters
Urinary albumin excretion was measured using an ELISA (CellTrend, Luckenwalde, Germany). Serum and urinary creatinine and serum urea were assessed using an automated analyzer (Integra 800, Roche Diagnostics, Mannheim, Germany).

Real time RT-PCR
Renal cortical tissue (10 mg) was homogenized in 500 ml of RLT buffer reagent (Qiagen, Hilden, Germany) with an ultraturrax for 30 seconds, and total RNA was extracted with RNeasy Mini columns (Qiagen) according to the standard protocol. First-strand cDNA was synthesized with TaqMan reverse transcription reagents (Applied Biosystems, Darmstadt, Germany) using random hexamers as primers. Final RNA concentration in the reaction mixture was adjusted to 0.1 ng/ mL. Reactions without Multiscribe reverse transcriptase were used as negative controls for genomic DNA contamination. RTproducts were diluted 1:1 with dH 2 O before PCR procedure. PCR was performed with an ABI PRISM 7000 Sequence Detector System and SYBR Green reagents (Applied Biosystems) according to the manufacturer's instructions. Primers used in this study are listed in table 1. Primer pairs were designed using the Primer Express software (Perkin Elmer, Foster City, CA, USA) except for TGFb [17], collagen I [18], vitronectin [19], nephronectin [20], fibronectin [21] and osteopontin [22]. The relative amount of the specific mRNA was normalized with respect to 18S rRNA. All samples were run in triplicates.

Immunohistochemical analyses
Staining of cryostat and paraffin sections was performed as previously described [23]. The primary rabbit antibody to a8 integrin (Ulrich Muller, Scripps Institute, LaJolla, USA) and the primary rabbit antibody to fibrillin-1 (Dieter Reinhardt, McGill University, Montreal, Canada) were used on cryopreserved tissue at a dilution 1:500. CY3 labeled anti-rabbit IgG (DAKO Diagnostica, Hamburg, Germany) was used as secondary antibody. Double immunofluorescence for a8 integrin and the cell markers smooth muscle actin (DAKO) and cytokeratin (DAKO) as well as for the a8 integrin ligands fibronectin (Life Technologies, Darmstadt, Germany), osteopontin (Jack Kleinman, Medical College of Wisconsin, Milwaukee, USA) and fibrillin-1 was performed as described [23]. Detection and quantification of a-smooth muscle actin, vimentin (Progen, Heidelberg, Germany), collagen I (Biogenesis, Poole, England), fibronectin, osteopontin was carried out in paraffin embedded renal tissue as described [23].
Evaluation of the expansion of various cell markers and matrix molecules was done with a Leitz Aristoplan microscope (Leica Instruments) in 5 (mice) to 10 (rats) non-overlapping medium power views using Metavue software (Visitron Systems, Puchheim, Germany). The stained area was expressed as percentage of the total area of renal cross section.

Renal histology
Tubulointerstitial injury was evaluated in kidney sections stained with period acid-Schiff's (PAS) reagent. Tubulointerstitial injury was assessed in 5 non-overlapping medium power randomly sampled views per kidney section using a tubulointerstitial damage score grading tubular atrophy, tubular dilatation, interstitial fibrosis and interstitial inflammation as follows: grade 0, no change; grade 1, lesions involving less than 25% of the area; grade 2, lesions affecting 25-50%; grade 3, lesions involving more than 50% and grade 4 involving (almost) the entire area.

Analysis of data
A student's t-test was used to test significance of difference between two groups. Otherwise, analysis of variance (ANOVA), followed by posthoc Bonferroni test, was used to test significance of differences between groups. A p-value of less than 0.05 was considered significant. The procedures were carried out using the PASW Statistics 18 software (SPSS Inc., Chicago, USA). Values are displayed as means 6 SEM.

Results
After UUO, a8 integrin is de novo expressed in tubular epithelial cells and fibroblasts and colocalizes with its ligands fibronectin, osteopontin and fibrillin-1 In response to UUO, an increase in the expression of the profibrotic cytokines connective tissue growth factor (CTGF) and TGFb as well as of the TGFb-associated matrix molecule latent TGFb binding protein (LTBP-1) was observed (table 2). Concomitantly, the expression of collagen I was induced and immunoreactivity of a-smooth muscle actin and vimentin as markers of activated myofibroblasts was more abundant (table 2).
The mRNA expression and immunoreactivity of a8 integrin was significantly increased after UUO (figure 1). While confined to mesangial cells and vascular smooth muscle cells in the control kidney (white arrowheads), tubulointerstitial staining for a8 integrin was observed after UUO (figure 1B). The expression of the a8 integrin chain dimerization partner b1 integrin chain was also induced (table 2). As binding of integrins to their ligands is essential for integrin signaling, we also investigated the expression of the ligands of a8 integrin: The expression of vitronectin and nephronectin was low (at detection limits) and was not increased after UUO (table 2). In contrast, the expression and immunoreactivity for fibronectin, osteopontin and fibrillin-1 was significantly induced in kidneys of UUO animals (table 2). Fibronectin and fibrillin-1 were detected in the tubulointerstitium (figure 2), while osteopontin was detected in tubular epithelial cells (figure 2).
To determine the cell types expressing the a8 integrin chain, double immunofluorescent studies were performed for a8 integrin and the cell markers cytokeratin for epithelial cells and a-smooth   muscle actin for activated interstitial fibroblasts. There was a partial colocalization observed with both markers (figure 3A and B) arguing for a de novo expression of a8 integrin in tubular epithelial cells and interstitial fibroblasts in this model in addition to the established expression of a8 integrin in mesangial cells and vascular smooth muscle cells. Moreover, colocalization of a8 integrin with its ligands fibronectin and fibrillin-1 was observed in the tubular interstitium ( figure 3C and E). Colocalization of a8 integrin with osteopontin was observed in tubular epithelial cells ( figure 3D).
The expression of a8 integrin is reduced in heterozygous a8 integrin-deficient (a8+/2) mice and is barely detectable in homozygous a8 integrin-deficient (a82/2) mice after UUO To assess the contribution of de novo expressed a8 integrin to tubulointerstitial fibrosis, UUO was induced in a8 integrindeficient mice. As our studies in rats had revealed overt fibrotic changes in the obstructed kidney after 7 days of UUO, this time point was chosen for our further experiments. The expression of a8 integrin was compared in the genotypes after UUO. Kidneys of a8+/2 mice expressed reduced amounts of a8 integrin (0.4860.08 fold induction versus 1.0060.14 in wild types, p,0.05), while expression levels of a8 integrin in a82/2 mice was at detection limits (0.001260.0003 fold induction). Expression of the a8 integrin ligands fibronectin, osteopontin and fibrillin-1 was increased after UUO, without any significant differences between wild types and a8+/2 mice or a82/2 mice (table 3).
Renal fibrosis after UUO is more severe in a82/2 mice but unchanged in a8+/2 mice compared to wild type mice After 7 days of UUO, body weights, left ventricular weights and blood pressure were comparable between all experimental groups (table 4). Relative kidney weights tended to be higher after UUO, reaching statistical significance only in a82/2 mice (table 4). Albumin excretion was not significantly different between groups (table 4). Creatinine clearance tended to be reduced in all genotypes after UUO, while serum creatinine values tended to be increased after UUO in wild type and a82/2 mice and increased significantly in a8+/2 mice (table 4). In a82/2 mice, serum urea    (table 4) and UUO led to higher serum urea levels in a82/2 mice compared to wild types (table 4).
Tubulointerstitial damage as assessed by a tubulointerstitial lesion score was hardly detected and not different when control groups were compared, but increased significantly after UUO (figure 4A). No differences in tubulointerstitial damage were seen in wild type and a8+/2 mice, but more prominent damage was detected in a82/2 mice compared to wild type or a8+/2 mice after UUO (figure 4A). Consistent with this finding, tubulointerstitial collagen I expansion after UUO was the same in wild type and a8+/2 mice, but more severe in a82/2 mice compared to wild type or a8+/2 mice (figure 4B). Moreover, tubulointerstitial fibroblast activation to myofibroblasts was seen in all genotypes after UUO, but in a82/2 mice fibroblast activation was detected more frequently compared to wild type and a8+/2 mice, which did not differ in the load of myofibroblasts (figure 4C). Exemplary photomicrographs showing the increase in collagen I and asmooth muscle actin staining in kidney tissue of obstructed wild type and a82/2 mice are presented in figure 5. Supporting these findings, a-smooth muscle actin mRNA expression was significantly higher in obstructed a82/2 mice compared to wild types (table 3). Expression levels of collagen I mRNA, however, were induced in response to induction of UUO to a similar degree in all genotypes (table 3).

Discussion
Our data show that a8 integrin was de novo expressed in tubular epithelial cells and interstitial fibroblasts in response to UUO. The a8 integrin ligands fibronectin, osteopontin and fibrillin-1 were upregulated and colocalized with a8 integrin in the tubulointerstitium. However, underexpression of a8 integrin did not attenuate renal fibrosis after UUO. On the other hand, the ''extreme'' genotype of mice with a homozygous deletion of a8 integrin, leading to the complete absence of the a8 integrin chain, induced more pronounced fibrotic changes in response to UUO. Thus, our results demonstrate that the a8 integrin chain is induced de novo in tubular epithelial cells and interstitial fibroblasts but this does not appear to add to interstitial fibrosis. Rather, the de novo expression of a8 integrin seems to play a protective role.
De novo expression of a8 integrin in tubulointerstitial fibrosis together with an increased expression of b1 integrin and a8b1 ligands, which also colocalize with a8 integrin to allow for a8 integrin-dependent signaling, seems to support the concept that a8 integrin may have a functional relevance for fibrotic alterations in the tubulointerstitium of the kidney. An increase in a8 integrin expression was detected in fibrotic events of other organs, like heart, liver and lung [14,15] and was induced by profibrotic TGFb and angiotensin II [13,24]. Therefore, it was suggested to have profibrotic properties. The data of our present study, however, do not bear out this notion.
In accordance with our present data, hypertension-induced glomerulosclerosis was not attenuated in mice deficient for a8 integrin [25]. Overexpression of a8 integrin in the glomerulus, however, was exclusively due to an increased expression in mesangial cells which normally express a8 integrin, and did not involve de novo expression of this integrin in other glomerular cells [13]. This is in contrast to our findings in the tubulointerstitium where epithelial cells and fibroblasts, which are normaly negative for a8 integrin, de novo express this integrin. The a8 integrin chain is mainly found on mesenchymal cells [26], but some reports also exist describing epithelial a8 integrin expression [16,27,28] mainly in less differentiated epithelium. De novo expression of a8 integrin in tubular epithelial cells during fibrotic changes of the tubulointerstitium seems to support these previous findings. But a  change in the degree of epithelial differentiation alone might not be sufficient to trigger an increase in fibrotic events. Nevertheless, the lack of a clear profibrotic tubulointerstitial effect of a8 integrin signaling in our study was surprising in view of the many observations previously reported on related a integrin chains, the b1 integrin chain which dimerizes with the a8 integrin chain, and at least one ligand of a8 integrin in the UUO model: The a5 integrin chain which is closely related to a8 integrin is de novo expressed in tubular epithelial cells after UUO [5]. In UUO, tubulointerstitial expression of the b1 integrin chain increases and inhibition of b1 integrin signaling has antifibrotic effects [29]. The a8 integrin ligand osteopontin was also upregulated and colocalized with a8 integrin in tubular epithelial cells after UUO. Osteopontin was reported to contribute to tubulointerstitial fibrotic changes in the obstructed kidney [30]. Thus it seemed conceivable that these functions of osteopontin could occur via a8 integrin-dependent signaling. The data of our study, however, cast doubt on such an interrelation and argue for a signaling via other osteopontin receptors, like CD44 or av integrins, to convey profibrotic effects.
A profibrotic function of a8 integrin in the development of tubulointerstitial fibrosis altogether seems unlikely, as in mice expressing reduced amounts of a8 integrin tubulointerstitial fibrosis is not attenuated. On the contrary, mice completely lacking a8 integrin developed a somewhat more severe fibrosis compared to mice with conserved renal expression of a8 integrin. Thus, a8 integrin could play a protective role in the tubulointerstitium. This might be due to an increased sensitivity of cells lacking a8 integrin for mechanical strain. In mice with a deficiency for a8 integrin more severe glomerular damage, i.e. disruption of the glomerular tuft, was detected in glomeruli after induction of glomerular hypertension [25] arguing for an important contribution of a8 integrin to the maintenance of tissue integrity. This might hold true not only for the glomerulus, but also for the tubulointerstitium, where a8 integrin is not regularly expressed. Similar protective functions were described for other integrin chains as well: Loss of a1 integrin results in more severe glomerulosclerosis after injury [7] and a selective deletion of the a3 integrin chain in podocytes also leads to increased accumula- Figure 4. Tubulointerstitial changes in a8 integrin-deficient mice after unilateral ureteral obstruction (UUO). A: Tubulointerstitial lesion score. B: Expansion of tubulointerstitial collagen I deposition. C: Abundance of activated interstitial myofibroblasts as assessed after a-smooth muscle actin staining (sma). WT, wild type; a8+/2, mice with a heterozygous deficiency for a8 integrin; a82/2, mice with a homozygous deficiency for a8 integrin. Data are means6sem. * p,0.05 UUO versus sham, # p,0.05 a82/2 versus wild type UUO. doi:10.1371/journal.pone.0048362.g004 Figure 5. Representative examples of collagen I and a-smooth muscle actin stained renal sections of WT and a8 integrindeficient mice. UUO, unilateral ureteral obstruction; co, control animal; WT, wild type; a82/2, mice with a homozygous deficiency for a8 integrin; sma, a-smooth muscle actin; coll I, collagen I. doi:10.1371/journal.pone.0048362.g005 tion of extracellular matrix in the glomerulus [31]. In this context one limitation has to be mentioned. We cannot completely rule out an effect of a somewhat reduced renal mass in a8 integrindeficient mice. A reduced renal mass itself is known to aggravate the sequelae of renal injury, especially in the glomerulus [32]. It can further not be excluded that a more prominent role for a8 integrin exists in milder and more chronic forms of renal fibrosis, as UUO is a model of severe tubulointerstitial changes, which might override the effects of a8 integrin-mediated interactions. Finally, we cannot rule out that de novo expression of a8 integrin in tubular epithelial cells and activated interstitial fibroblasts is merely a marker of dedifferentiation in these cells not serving a special pathological function in tubulointerstitial fibrosis after UUO.
Taken together, our data do not support the hypothesis that de novo tubulointerstitial expression of a8 integrin contributes to fibrotic processes after tubulointerstitial injury. Thus, targeting a8 integrin does not seem to be a useful approach to attenuate renal fibrosis. More likely, the de novo expression of a8 integrin might serve protective functions as a complete loss of a8 integrin somewhat aggravates tubulointerstitial injury.