The down regulation of megalin/LRP2 by transforming growth factor beta (TGF-ß1) is mediated by the SMAD2/3 signalling pathway

Megalin/LRP2 is a receptor that plays important roles in the physiology of several organs, such as kidney, lung, intestine, and gallbladder; and also in the physiology of the nervous system. Megalin expression is reduced in diseases associated with fibrosis, including diabetic nephropathy, hepatic fibrosis and cholelithiasis, as well as in some breast and prostate cancers. One of the hallmarks of these conditions is the presence of the cytokine transforming growth factor beta (TGF-ß). Although TGF-ß has been implicated in the reduction of megalin levels, the molecular mechanism underlying this regulation is not well understood. Here, we show that treatment of two epithelial cell lines (from kidney and gallbladder) with TGF-ß1 is associated with decreased megalin mRNA and protein levels, and that these effects are reversed by inhibiting the TGF-ß1 type I receptor (TGF-ßRI). Based on in silico analyses, the two SMAD-binding elements (SBEs) in the megalin promoter are located at positions −57 and −605. Site-directed mutagenesis of the SBEs and chromatin immunoprecipitation (ChIP) experiments revealed that SMAD2/3 transcription factors interact with SBEs to repress the megalin promoter and that they are also required for the repressing role of TGF-ß1. In addition, high concentration of albumin reduced megalin expression and promoter activation that depend on the expression of SMAD2/3. Interestingly, the histone deacetylase inhibitor Trichostatin A (TSA), which induces megalin expression, reduced the effects of TGF-ß1on megalin mRNA levels. These data show the significance of TGF-ß and the SMAD2/3 signalling pathway in the regulation of megalin and explain the decreased megalin levels observed under conditions in which TGF-ß is upregulated, including fibrosis-associated diseases and cancer.


Introduction
Progressive fibrosis is the final stage of several chronic diseases, including diabetes, obesity, gallstone disease, liver cirrhosis, pulmonary fibrosis, and cardiovascular and muscular diseases [1,2]. The hallmark of this condition is increased deposition of extracellular matrix (ECM) and alteration of its composition, resulting in loss of function of the affected organs and tissues. Along with abnormal ECM deposition, fibrosis also alters the function of several cell types, including epithelial cells that undergo the epithelialmesenchymal transition (EMT), which is characterized by loss of cell adhesion, suppression of E-cadherin expression [3,4] and expression of genes such as vimentin and alpha-smooth muscle actin (SMA) [3,5,6]. The EMT is thus associated with cell migration, tumour invasion and cancer. In kidney fibrosis for example, an important population of myofibroblasts originates from renal tubular epithelial cells via EMT [7].
The pro-inflammatory cytokine TGF-1 is a notorious inducer of fibrosis and cancer [8][9][10]. For example, in gallbladder cancer cells TGF- induces EMT in vitro [11] and in vivo [12,13]. In general, TGF-1 increases ECM production by stimulating collagen and fibronectin synthesis, and in epithelial cells it is involved in EMT [4]. The signal transduction pathway initiated by this cytokine requires activation of the serine/threonine kinase receptors TGF-ßRI and TGF-ßRII on the cell surface, resulting in their auto phosphorylation and subsequent phosphorylation of SMAD transcription factors [14]. The SMAD2 and SMAD3 proteins mediate the canonical TGF-1 pathway [15,16]. Under steady-state conditions, SMAD proteins exist as homo-oligomers that reside in the cytoplasm. Ligand activation causes SMAD2/3 phosphorylation by the TGF-ßRI kinase 4 (ALK5), leading to the formation of a complex with SMAD4. The SMAD complex can then translocate to the nucleus and regulate target genes by directly binding to the promoters with SMAD-binding elements (SBEs) or by associating with transcriptional coactivators or co-repressors [14,[17][18][19][20][21]. In addition to activating the canonical SMADdependent pathway, TGF-ß1 can also activate other signalling pathways that involve c-Jun N-terminal kinases (JNK), extracellular signal-regulated kinases (ERK) and p38 [22].
Megalin/LRP2 is a 600 kDa membrane glycoprotein that plays important physiological roles during embryonic development [23][24][25][26] as well as in adulthood [27]. Megalin expression is specifically reduced in epithelial cells in different fibrosisassociated pathologies that compromise the function of organs such as kidney, liver and gallbladder [27,47,49,[51][52][53][54][55][56]. However, the molecular mechanisms underlying the decreased expression of this receptor remain unclear. Diabetic rats that were administered streptozotocin exhibit increased albuminuria and type IV collagen deposition, and decreased megalin levels. These effects were reversed by injection of a soluble fragment of TGF-1 receptor type II (sTßRIIFc) [47]. TGF-1 is involved in the regulation of megalin receptor levels in the opossum kidney cell line (OK), derived from the renal proximal tubule. This ligand causes a reduction of megalin protein levels and reduces albumin endocytosis [52]. The inhibition of MAPK/ERK, p38, protein kinase C (PKC), protein kinase A (PKA) and Phosphoinositide 3-kinase (PI3K) showed that none of these signalling pathways were involved in lowering albumin endocytosis through TGF-1. In contrast, the use of dominant negative forms of the transcriptional factors SMAD2 and SMAD3 reduced the effect of TGF-1 on albumin endocytosis, suggesting that these factors could be important for the regulation of megalin levels [52]. In accordance with the profibrotic effect of albumin, high concentration of this protein significantly reduced megalin expression in PTC cells [57]. Similarly, Ang II also decreased megalin expression [58], and the in vivo inhibition of the angiotensin receptor (AT 1) with losartan protected against the reduction of 6 megalin observed in the kidneys of rats subjected to bovine serum albumin (BSA)-induced tubule-interstitial damage and proteinuria [27]. Despite efforts to identify the pathways that lead to decreased megalin levels in fibrosis-associated diseases, little is known about the molecular factors involved in these processes.
Additionally, megalin expression on the cell surface can be modified by ectodomain shedding through a mechanism involving the activity of PKC and matrix metalloproteases [59]. On the other hand, it has also been shown that in renal fibrosis the reduction of Ecadherin is in part explained by TGF- induced proteolysis [4]. In the same way, TGF-1 signalling has been associated with the activation of metalloproteinases such as tumor necrosis factor-alpha converting enzyme (TACE) and the subsequent shedding process of transmembrane proteins [60,61].
In this work, we explored two mechanisms to investigate how TGF-1 induces the reduction of megalin levels. The first one involves proteolytic processing. For the second mechanism we tested if TGF-1 transcriptionally regulates megalin expression through the transcription factors SMAD2/3. Our results indicate that the effects of TGF-1 are mostly transcriptional. We showed that transcription factors SMAD2/3 decrease megalin mRNA expression by binding directly to SBEs in the human megalin promoter and are required for the repression effect of TGF-1 and also of albumin. Our results suggest that TGF-1 and SMAD2/3 could be critical regulators of megalin expression in various pathophysiological conditions, including chronic renal disease, cholelithiasis and pulmonary diseases.

Cell culture and treatment
We used the LLC-PK1 epithelial cell line derived from porcine kidneys. Cells were cultured in alfa-MEM supplemented with 7.5% FBS containing 100 U/ml P/S, 2 mM glutamine and 2.5 g/ml Plasmocin TM . These cells have been previously used in studies on proximal tubule function and megalin expression and function [27,35]. The dog gallbladder epithelial cell line dGBECs are polarized cells that express several proteins present in gallbladder, including megalin and cubilin [35]. These cells were cultured in low-glucose DMEM supplemented with 2 mM glutamine, 1% MEM non-essential amino acids solution, 1% MEM vitamin solution, 7.5% FSB and P/S. The cells were seeded at approximately 70-80% confluence in complete media for 24 h and maintained at 37 °C in 5% CO 2 . HEK293 cells were grown and maintained in high-glucose DMEM supplemented with 10% FBS containing 100 U/ml P/S and 2.5 g/ml plasmocin. All the cultures were maintained in humid chambers. For albumin (BSA) treatments, LLC-PK1 cells were plated 9 at 3 x 10 4 cells/cm 2 for 48 h and were subsequently grown overnight in low-serum medium (alfa-MEM supplemented with 0.5% FBS containing 100 U/ml P/S, 2 mM glutamine and 2.5 mg/ml Plasmocin TM ) and treated with fatty acid-free BSA or vehicle (water) in serumfree medium at different time points (0.5, 1, 6 or 12 h) and concentrations (0.01, 0.1, 10, or 20 ng/ml).

RNA isolation, reverse transcription and qPCR
Total RNA was extracted using an RNA-Solv isolation system following the manufacturer's

Immunoblot analysis
Cells were lysed in lysis buffer (phosphate-buffered saline (PBS) containing 1% Triton X-100, 2 mM PMSF, 1 mM pepstatin, 2 M antipain, 1 M leupeptin, and 0.3 M aprotinin) for 30 min at 4 °C. The cells were then removed using a cell scraper, and the nuclei eliminated by centrifugation at 20,000x g for 5 min at 4 °C. Protein concentrations were measured using a BCA protein assay kit. Equal amounts of protein from each extract were subjected to reducing conditions, and megalin, E-cadherin or ß-tubulin were immunodetected as previously described [35,62]. The immunoblots were visualized using the Pierce enhanced chemiluminescence (ECL) system, and densitometric analysis was

Nuclear Extracts
Cells were washed twice with PBS and centrifuged for 10 min at 300x g. The pellet was resuspended in 1 ml of hypotonic lysis buffer and incubated for 30 min at 4 °C. Nuclei were centrifuged at 10,000xg for 10 min at 4 °C, and histones were extracted with 400 l of 0.2 M sulphuric acid (H 2 SO 4 ) and incubated in a rotator for 30 min at 4 °C. Eluted immunoprecipitated samples were precipitated for 30 min on ice in 25% (v/v) trichloroacetic acid (TCA) followed by centrifugation for 10 min at 16,000xg and two washes with ice-cold acetone. Samples were air dried and dissolved in sample buffer, and the protein concentration was then measured by BCA assay. The nuclear proteins were frozen in liquid nitrogen and stored in aliquots at -80 °C.

In silico analysis
To locate the transcription factor binding sites in the human megalin promoter, tools from MatInspector software (Genomatix) were used as described above [66,67].

Statistical analysis
Immunoblots were quantified using ImageJ 1.45s software, and qPCR analysis was performed using a relative quantification mathematical model, as previously described [68].
Data are expressed as the means ± SD from at least three independent experiments.
Comparisons were performed using analysis of variance (ANOVA) followed by Bonferroni correction. All of the statistical analyses were performed using GraphPad Prism 5. 18

Results:
The decrease in megalin mRNA and protein levels induced by TGF-β1 in epithelial cell lines is dependent on the activation of TGF-ß type I receptor kinase.
Although it has been established that TGF-ß1 reduces megalin protein levels [47,52], it is not known whether TGF-ß1 directly regulates megalin transcription. Therefore, megalin cells showed decreased megalin mRNA levels. However, megalin mRNA levels show the largest decrease (as percentage) in dGBECs when exposed to higher TGF-ß1 concentrations. Next, we studied the effects of TGF-ß1 exposure over time in LLC-PK1 cells and dGBECs by incubating them with 2.5 ng/ml and 10 ng/ml TGF-ß1 respectively.
In both cell types, megalin mRNA levels significantly decreased after 6h of TGF-β1 treatment (Figs 1B and 1D). mRNA expression of plasminogen-activator inhibitor (PAI-1), a known TGF-ß1 target gene, was measured as positive control (Figs 1E and 1F) to show that the cells had an active, operative TGF-ß1 signalling pathway. performed in OK cells [52]. Interestingly, the distribution of megalin in LLC-PK1 and dGBECs differs in confluent cells. In dGBECs (Fig 2D) it is clearly perinuclear and vesicular, whereas in LLC-PK1 ( Fig 2B) besides some vesicular localization, megalin is 20 also found in clusters close to the cell periphery. As control we measured the expression of the epithelial marker E-cadherin (Figs 2A, and 2C), a known TGF-ß1 target that is downregulated by the cytokine [69]. Taken together, these results indicate that TGF-ß1 decreases megalin mRNA levels, causing a concomitant reduction of megalin protein levels. The canonical TGF-ß1 pathway includes the SMAD proteins, which are transcriptional modulators [19]. Therefore, we tested whether the SMAD2/3 signalling pathway is directly involved in the down-regulation of megalin expression. To inhibit SMAD2/3 phosphorylation (and therefore activation of the pathway), we used the specific inhibitor SB-431542 (SB), which blocks the activity of TGF-ßRI (ALK5) serine/threonine kinase.
SB does not interfere with the enzymatic function of other receptors of the same family such as ALK 2,3 and 6 and does not affect the activation of other kinases (ERK, JNK or p38) associated with TGF- pathways [70,71]. As a result of SB treatment, both megalin and E-cadherin levels recovered from the effects of TGF-β1, as shown in the lower immunofluorescence panels (Figs 2B and 2D). When LLC-PK1 cells were treated with TGF-β1 (2.5 ng/ml) and 10 µM SB for 20 h, the effect of TGF-ß1 on megalin mRNA levels was completely reversed, and even increased compared to the control condition ( Fig 3A, fourth bar). Similarly, western blot analyses (Fig 3C and 3D) show that SB completely counteracted the effect of TGF-ß1 restoring megalin protein levels. Furthermore, the addition of 10 µM SB alone resulted in a nearly 2-fold increase in basal megalin mRNA and protein levels compared with the control (Figs 3A, 3C and 3D). PAI-1 mRNA levels were measured as a response-control ( Fig 3B). Interestingly, treatment with SB alone induced a significant decrease in PAI-1 mRNA levels by 30% (Fig 3B, third bar), suggesting that under steady-state conditions, LLC-PK1 epithelial cells have a basal activity of the TGF-ßRI/SMAD2/3 signal transduction pathway. In addition, to determine whether SMAD2/3 signalling responds to treatments, we used immunofluorescence to assess nuclear localization of pSMAD2/3. TGF-ß1 induced nuclear translocation of pSMAD2/3, and this effect was inhibited by SB ( Fig 3E). As is shown in Fig S1,    In order to complement this result, we also determined the levels of the megalin carboxyterminal fragment (CTF), which is generated after ectodomain shedding. We found that CTF levels were not increased after a 10 h treatment with TGF-ß1 2.5 or 20 ng/ml .
However, treatment with the -secretase inhibitor DAPT stabilized CTF levels (Figs 5A and   5B). The product of CTF processing is the intracellular cytoplasmic domain of megalin (MICD) which, in the case of this receptor, has been suggested to negatively control its own transcription [73]. We were able to detect the CTF but the MICD is extremely unstable and its detection erratic even in the presence of proteasome inhibitors. However, since TGF-ß1 did not increase CTF levels, we discarded an increased production of the MICD.
According to the results shown in Fig 4, total megalin levels were not modified after 10 h of treatment with TGF-ß1 at the two concentrations tested (Figs 5C and 5D). Overall, these results do not support a role of TGF-ß1 in the degradation or shedding of megalin, at least in the first 10-12 h of treatment.

SMAD2/3 proteins interact with two SBEs (SMAD-binding elements) identified in the megalin promoter.
The SMAD transcription factors act by forming complexes and they simultaneously activate and repress target genes in the same cell, depending on the context of the SBEs and the presence of proteins that induce the formation of the protein complexes that bind to these sites [14,18]. amplified from the complex by qPCR using two primer pairs designed for this purpose ( Fig   6B). The results showed that SMAD2/3 interact with the SBEs localized at positions -605 (SBE1) and -57 (SBE2) in the human megalin promoter (Fig 6B). The relative occupancy for the factors binding to the two SBE is shown in Fig 6C (black bars). As a negative control for the immunoprecipitation experiments we used an unrelated antibody (anti-HA).
The -280 and -680 SBEs from the PAI-1 promoter were analysed as positive controls for ChIP (Figs 6B and 6C, white bars). These results support the idea that SMAD2/3 could directly inhibit megalin expression at the transcriptional level by binding to the SBEs in the megalin promoter.

TGF-ß1 represses megalin promoter activity via SMAD2/3.
To directly test the repressive effect of SMAD2/3 on the human megalin promoter, LLC-PK1 cells were co-transfected with the human megalin promoter coupled to luciferase (-1500PromBasic) in the presence or absence of overexpressed SMAD2/3. The results showed that the expression of SMAD2/3 triggered a significant decrease (50%) in megalin promoter activity compared with the control without SMAD2/3 transfection (Fig 7A). The positive control was the p800Luc plasmid, containing the PAI-1 promoter fragment (-800 to +71) with two consensus SBEs located at positions -280 and -680 bp of the transcription start site [64,65]. In this case, the overexpression of SMAD2/3 increased the activation of the PAI-1 promoter (Fig 7B). Interestingly, in a complementary experiment we found that megalin protein levels were significantly decreased by the overexpression of SMAD2/3.

Immunoblots showed that megalin levels decreased almost 40% in cells overexpressing
SMADs (Figs 7E and 7F), and we obtained similar results when megalin levels were detected by FACS. The cells expressing SMAD2/3 with the FLAG epitope expressed significantly lower levels of megalin than the controls (Fig 7G). On the other hand, the direct effect of TGF-ß1 was also tested with the same promoters coupled to luciferase and, as expected, the cytokine decreased luciferase expression when the luciferase gene was under the control of the megalin promoter ( Fig 7C). Also, as expected, TGF-ß1 increased luciferase activity in cells where the luciferase gene was under the control of p800Luc ( Fig   7D).

29
To directly address the role of SMAD2/3 in the repressive effect of TGF-ß1 on megalin, we silenced the expression of SMAD2/3 in LLC-PK1 cells and then we transfected them with the different promoter constructs coupled to luciferase. The silencing was very efficient (84% decrease of SMAD2 and 72% of SMAD3), as seen in the immunoblot for SMAD 2 and SMAD3 (Fig 8A and 8B). Under conditions in which the expression of SMAD2/3 was significantly downregulated, the repressive role of TGF-ß1 on the activation of the megalin promoter was not evident, indicating that the cytokine affects megalin expression through the canonical SMAD2/3 pathway ( Fig 8C). As control, the TGF-ß1 induced activation of p800Luc was blocked in the SMAD2/3 knock-down cells (Fig 8D). At the protein level, the downregulation of megalin induced by the cytokine was also dependent on the expression of SMAD2/3 (Fig 8E and 8F). The reduction of SMAD2/3 expression was associated with a significant decrease in the activation of the p800Luc promoter under non-stimulation conditions (Fig 8D), indicating a basal activation of the system in this cell type, similarly to what was found when TGF-RI was inhibited by SB ( Fig 3B). . These data indicate that the two SMAD response elements independently repress the megalin promoter. In addition, in the SBE mutant promoters, the basal activity was significantly increased compared with the wild type promoter in the absence of overexpressed SMAD factors (Fig 9B, all white bars), suggesting that endogenous SMAD factors were basally active in LLC-PK1 cells (as suggested in Fig 3A, 3C, 3D and FigS1).
Finally, the relevance of these two SBEs in the repressor effect of TGF-ß1 was determined by the same type of assays using the wild-type and mutant promoters (Fig 9C). In the SBE1 mutant and in the double SBE1/2 mutant, the repressor role of TGF-ß1 was abolished. The mutant SBE2 responded to the cytokine but to a lesser extent than the wild-type promoter.
Overall, these results strongly suggest that both SBEs identified in the human megalin promoter could function as binding sites for a transcriptional repressor complex that is linked by SMAD2/3 and have a role in the repressive effect of TGF-ß1.

Albumin reduces megalin expression by a mechanism that involves the TGF-βRI/SMAD2/3 pathway
Although high concentrations of albumin are known to decrease megalin expression [57], it is not known if this effect is dependent on the activation of the TGF- signalling pathway.
Albumin induces TGF-ß1 secretion [74] in human proximal tubule HK-2 cells and in cells from the collecting duct that lack megalin [75]. Therefore, the pro-inflammatory effects of albumin could be megalin independent. To determine whether the albumin-induced megalin reduction is connected to the canonical TGF-ß1 pathway, LLC-PK1 cells were incubated with increasing concentrations of albumin for 24 h and the mRNA expression of megalin and PAI-I was measured. As previously reported, albumin significantly reduced megalin expression at a concentration of 10 mg/ml [57] (Fig 10A). A lower concentration was able to induce PAI-1 expression (Fig 10B). Moreover, the albumin-induced decrease in megalin expression and the increase in PAI-1 expression were significantly overcome by SB (P ≤ 0.05 in both cases), suggesting that the effects of albumin involves, directly or indirectly, activation of the TGF-ß1 receptor (Figs 10C and 10D). Again, we observed that the inhibition of the TGF-ßRI by SB significantly increased megalin mRNA levels and reduced PAI-1 expression as shown before (Fig 3A and B). To complement these results, the effects of albumin were determined directly in the activation of megalin human promoter as described for TGF- (Fig 7 and Fig 8) Similarly to the effect of the cytokine, albumin significantly decreased megalin expression (Fig 10E). Moreover, the repressing effect of albumin on megalin was dependent on the presence of SMAD2/3 ( Figure 10F) since it was abolished in SMAD2/3 knockdown cells.

The repression of megalin mRNA expression by TGF-ß1 is reverted by inhibition of histone deacetylases
Transcriptional repression involves, among other mechanisms, histone deacetylase (HDAC) activity [76]. We wanted to evaluate the effects of the HDAC inhibitor TSA on megalin mRNA expression in LLC-PK1 cells by qPCR. TSA treatment, starting at a 50 nM concentration, elicited a significant increase in megalin mRNA levels compared with the vehicle (control) (Fig 11A and 11B, third bar). These results are concordant with previous studies that analysed the regulation of megalin expression by TSA in NRK and Caco-2 cells [77]. We next examined the effect of TSA on TGF-ß1 -induced reduction of megalin expression that had been previously observed in LLC-PK1 cells (Figs 1A, 1B and 2A,2B).
The cells were treated with TGF-ß1 and/or TSA for 6 h and mRNA levels were then evaluated by qPCR. Whereas TGF-ß1 decreased mRNA as expected, TSA significantly counteracted TGF-ß1 response resulting in megalin mRNA levels similar to the control ( Fig 11B). Interestingly, TSA alone also increases PAI-I expression, however, the induction of the PAI-1 gene by TGF-ß1 was blocked by TSA and was not additive, as has been described in other cell types [78,79] (Fig 11C). In contrast to megalin mRNA levels, even though TSA alone increases megalin protein levels after 48 h of treatment, its effect over the reduction induced by TGF-ß1 was not evident under this experimental condition ( Fig   11D). As a control for TSA treatment, the acetylated form of histone H3 was detected by immunoblotting after a 6 h treatment (Fig 11E). These results indicate that TSA counteracts the repressive effect of TGF-ß1 on megalin mRNA expression, via modulation of HDAC activity. However long term TGF-ß1 treatment may reduce megalin protein levels involving other pathways in addition or independently of the HDACs.  Using MatInspector software (www.genomatix.de) we analysed the human megalin promoter and found two SMAD protein response elements (SBEs) within the first 1,500 base pairs upstream of the transcription start site. Other SMAD-independent response elements, including Ap1, Sp1 and FAST1, are also present. Sp1 activity has been previously described for the human and rat megalin promoters [63,86]. SMAD transcription factors form complexes that can simultaneously activate or repress hundreds of target genes in the same cell, depending on the context of the DNA-binding sites and on the presence of proteins that induce the formation of the protein complexes that bind to these sites [18]. In our study, immunoprecipitation assays revealed that SMAD2/3 bind to the megalin promoter at the two identified SBEs.

Discussion
The functional role of these SBEs and of the SMADs in the repression of megalin expression induced by TGF-ß1, was deciphered using luciferase activation assays. The on the confluence level of the cells [89]. The presence of E-cadherin at cell-adhesions is associated with an inhibition of SMAD2/3 phosphorylation (and reduction of the signaling induced by TGF-ß1). In contrast, the reduction of E-cadherin localization at the membrane increases TGF-ß1 mediated SMAD phosphorylation [90]. In general, under the conditions in which we performed the activation experiments (mRNA expression, promoter activation and measurements of luciferase activity) the cells were not completely confluent as when we did the immunofluorescence studies (Fig 2B and 2D). If cells are not "superconfluent", E-cadherin localizes both at the cell membrane and intracellularly. One possibility is that in our system these two elements (autocrine cytokine secretion and a partial E-cadherin mislocalization) are operating under basal conditions and therefore, we cannot discard that SB-421542 treatment is uncovering these mechanisms.

39
SMAD2/3 have the ability to induce gene activation because they interact with transcriptional co-activators such as p300 and CBP, which have histone acetyltransferase (HAT) activity [91]. Moreover, the SMAD proteins can also interact with transcriptional co-repressors such as TGIF [20], c-SKI [92], SIP-1 [93], SRF [94] and NF-1 [95] to form complexes that can bind to DNA sequences such as FAST-2, TGF-ß inhibitory element (TIE), and SBEs, and can induce gene repression via HDAC recruitment. Regarding the possibility that these proteins might recruit HDACs, our work also shows that the repressive effect triggered by TGF-ß1 on megalin mRNA levels was counteracted by inhibition of HDAC activity. The inhibition of HDACs in non-stimulated cells increased megalin mRNA and protein expression; whereas long term TGF-ß1 treatment reduced megalin protein levels involving pathways not associated with HDACs. We cannot discard the possibility that other repressive mechanisms could be operating in the regulation of megalin by TGF-ß1. The human megalin promoter contains a CpG island that when methylated downregulates megalin expression [96]. Additionally, the inhibition of DNA methylation by 5Azacytidine increases megalin expression [77]. Moreover, recent evidence shows that TGF-ß1 regulates ECM production and differentiation controlling gene expression through DNA and histone methylation [97]. TGF-ß1 also regulates EMT in cancer cells through an increase in DNA methylation and histone 3 lysine methylation at the E-cadherin promoter, depending on DNA methyltransferase 3A [98].
Another system that decreases megalin expression is induced by exposure to a high concentration (10-20 mg/ml) of albumin [57], as we showed here and as has been shown previously [82]. Although the mechanisms involved in the reduction of megalin expression by albumin are not entirely clear, other studies have demonstrated that albumin can induce 40 the release of TGF-ß1 in PTCs [99]. Furthermore, albumin can directly activate the TGF-β1 type II receptor in endothelial cells [100] and TGF-β secretion in astrocytes [101] via SMAD2/3 pathway activation [102]. Moreover, albumin has been shown to induce TGF-ß1 expression through the EGF receptor and ERK-MAPK signalling activation after a 72 h treatment [74]. In the present study, we found that albumin reduces megalin mRNA expression and activation of its promoter. The inhibition of TGF-ß1R activation as well the decrease in SMAD2/3 cellular levels significantly counteracted the effect of albumin on megalin expression, reinforcing the idea that the SMAD2/3-TGF-ßRI canonical signalling pathway mediates part of the effect of albumin on megalin levels.
In addition to albumin, other megalin ligands such as Ang II, leptin and CTGF induce fibrotic responses through their signalling receptors [55,[103][104][105][106]. Ang II, through its signalling receptor, AT1, may also stimulate the SMAD-dependent pathway during the EMT, causing renal fibrosis [103,106]; and leptin also induces TGF-ß1 expression [51]. As is the case for high albumin concentrations, both Ang II and leptin have been associated with reduced megalin expression [51,58]. For leptin, the reduction in megalin was not observed at the transcriptional level; thus, the authors suggested that this reduction was an acute effect that could likely be explained by megalin degradation. However, considering our results, chronic effects could include the regulation of megalin transcription because the expression of TGF-β is increased by leptin.
In this study, we showed that TGF-ß1 at low concentrations (2.5 to 10 ng/ml) reduced total megalin protein levels only as a result of decreased transcription and not of increased receptor degradation. Megalin half-life showed no changes in cells treated with TGF-ß1 at two different concentrations (2.5 and 20 ng/ml) and in the window of time 41 considered in our study (up to 12 h). Importantly, the concentrations of TGF-ß1 used in most of our work (2.5 to 10 ng/ml) are in the physiological range but the concentration of 20 ng/ml is rather high and in the range of diabetic patients [107,108] In addition, the cytokine did not modify receptor levels at the cell surface as could be expected by the shedding of the ectodomain. Therefore, TGF-ß signalling did not increase receptor degradation within the first 12 h of treatment. Recently, Mazzocchi et al [73] showed that in alveolar cells, TGF-ß1 is able to reduce surface megalin levels by inducing shedding of the megalin ectodomain, specifically by TGF-ß-induced secretion of matrix metalloproteases (MMPs)-2, -9 and -14. This process is followed by -secretase regulated intramembrane proteolysis (RIP) that induces the generation of the MICD as described previously [59,109]. The authors indicate that only high TGF-ß1 concentrations (20 ng/ml) decreased megalin mRNA levels after 24 and 48 h treatments, suggesting that this effect was the result of the auto inhibitory effect of MICD as repressor of megalin transcription.
In contrast with the report of Mazzocchi et al, our study did not find an increase in the levels of megalin CTF, the MICD precursor, after 10 h of TGF-ß1 treatment. Moreover, we found a significant decrease in megalin mRNA expression already detectable after 5 h and with a cytokine concentration almost 10 times lower than the one used on the alveolar cells.
We cannot discard that, in addition to our observation that TGF-ß1 directly represses megalin transcription via SMAD2/3, the MICD-dependent mechanism could also be operating. However, this mechanism could only function with a 10-fold increase in the concentration of the cytokine and in a different time frame. TGF-ß1 also induces the proteolytic shedding of E-cadherin [4], an epithelial membrane protein that was significantly decreased by TGF-ß1 in our study. The E-cadherin reduction induced by TGF-ß1 could also be caused by gene repression mediated by SMAD-interacting protein (SIP-1) via co-repressor C-terminal binding protein (CtBP) recruitment [110]; as well as by the activation of different transcription factors such as Snail, Twist, Slug, and Zeb that bind to specific sequences (called E-boxes) within the promoter [111].Therefore, more than one mechanism could be activated in response to TGF-ß1 exposure that results in functional reduction of megalin.
Considering the role of megalin in regulating the availability of fibrotic molecules, the decrease of megalin expression induced by its own ligands (and by TGF-ß could be associated with the evolution and/or the severity of some pathological conditions, such as fibrosis and cancer, in different megalin-expressing epithelial tissues. In other words, megalin itself could be considered a protective agent for those conditions by acting, for example, as an anti-fibrotic protein. Similarly, in some cancer cells, megalin expression is also decreased [34,50]. Because megalin is required for the internalization and activation of Vitamin D to 1,25-OH Vitamin D [45, 46], its decrease would directly impact the activation of its nuclear receptor, VDR, which plays an important anti-proliferative role in the control of some cancer types, especially breast, prostate and colon cancer [112].

CONCLUSION
The knowledge generated in this work establishes the molecular mechanism involved in the reduction of megalin expression by TGF-ß and will contribute to understanding in more detail the regulation of megalin in diseases in which this proinflammatory cytokine plays a central role.