An essential role for ClC-4 in transferrin receptor function revealed in studies of fibroblasts derived from Clcn4-null mice

The nine members of the mammalian ClC family of Cl^– channels and transporters can be divided into three subfamilies based on functional properties, sequence similarity and subcellular localization. ClC-1, ClC-2 and ClC-Ka/b, members of the most studied subfamily, share 54% sequence identity and are Cl^– channels that are predominately expressed in the plasma membrane (Jentsch et al., 1999). The subfamily of Cl^–/H⁺ transporters, ClC-3, ClC-4 and ClC-5, share ∼78% of their sequence and are primarily expressed in endosomal membranes (Picollo and Pusch, 2005; Scheel et al., 2005). The third subfamily, encompassing ClC-6 and ClC-7, are probably Cl^–/H⁺ transporters that are 46% identical and are predominately expressed in late endosomal and lysosomal membranes, respectively (Graves et al., 2008; Ignoul et al., 2007). ClC-7 is also found in the plasma membrane of the extracellular lysosome of the osteoclast (Kornak et al., 2001).

The physiological significance of two out of the three members of the second subfamily, ClC-3 and ClC-5, have been highlighted through the discovery of mutations within the genes that encode them that lead to human disease or through the study of ClC-specific knockout mice. Mutations in the gene encoding ClC-5 (CLCN5) lead to Dent's Disease, a renal disease characterized by proteinuria and thus implying a role for ClC-5 in re-absoprtion of proteins within the proximal tubule. This role was subsequently confirmed in the studies of Clcn5 knockout (KO) mice (Piwon et al., 2000). Although mutations in humans have yet to be found for ClC-3, studies on Clcn3 KO mice have been pivotal in identifying the functional role of this ClC family member. For example, the study of the Clcn3 KO mouse have identified a role for ClC-3 in synaptic vesicle acidification, a process that may be crucial for retinal and hippocampal health, at least in the mouse (Stobrawa et al., 2001). Although subsequent human disease and studies on knockout mice have been pivotal to the understanding of the physiological role of ClC-3 and ClC-5, mutations within the CLCN4 gene have yet to be found in humans and a study on the Clcn4 KO mice revealed no overt phenotype, thus leaving the physiological role of ClC-4 unclear (Jentsch, 2008).

As ClC-4 shares a high sequence conservation of 78% with ClC-5, a transporter known to be involved in endosomal acidification and trafficking in the proximal tubule, it has been proposed that ClC-4 may play similar roles within the kidney (Gunther et al., 2003; Hara-Chikuma et al., 2005a; Mohammad-Panah et al., 2003). Colocalization of ClC-4 with an endosomal marker, FITC-dextran, in tissue sections of the rat proximal tubule, and co-localization of endogenous ClC-4 to Rab-5a-positive endosomes of Cos7 cells indicated that ClC-4 was appropriately localized subcellularly to modulate renal endosomes directly (Mohammad-Panah et al., 2003). In support of this finding, antisense experiments in LLPKC cells that reduced expression levels of endogenous ClC-4 revealed endosomal alkalinzation and impaired endocytic uptake of transferrin (Tfn), implying a role for ClC-4 in endosomal acidification and trafficking of the Tfn receptor within the kidney (Mohammad-Panah et al., 2003). However, it remained unclear whether this antisense-dependent cellular phenotype reported the specific depletion of ClC-4 protein.

The goal of the present study was to elucidate the physiological role of ClC-4 on Tfn receptor function through the generation of primary renal fibroblasts from ClC-4-null mice, a cell line completely devoid of ClC-4. Generations of these ClC-4-null and wild-type renal fibroblast isolated from sibling pairs, permitted detailed immunofluorescent and biochemical assessment of the specific functional role of ClC-4 in endosomes. In support of our previous studies, accumulation of Tfn by receptor-mediated endocytosis was reduced, and acidification of early and sorting endosomes was significantly impaired in ClC-4-null fibroblasts. The defect in Tfn uptake occurred in the absence of ClC-4, despite a minor upregulation of the Tfn receptor at the cell surface. We found that surface binding of Tfn to its receptor was impaired in ClC-4-null fibroblasts but could be rescued to wild-type levels by the addition of the iron chelator desoxiferramine, a reagent that mimicks iron removal in the endosome. Together, these results provide direct evidence that the role of ClC-4 in endosomal acidification is crucial for the diferric-Tfn:Tfn receptor dissociation events occurring within the endosome, and suggests that ClC-4 contributes to Tfn receptor function by directly modulating its ligand accessibility at the cell surface.

ClC-4-null animals were produced as previously described (Rugarli et al., 1995). Real-time PCR analysis on isolated kidney tissues from ClC-4-null mice confirmed there was no significant change in the amount of ClC-5, ClC-3A or ClC-3B mRNA in the absence of mRNA encoding for ClC-4 relative to wild-type levels, as shown in Table 1 (P=0.14, P=0.35 and P=0.84, respectively). The lack of ClC-4 protein expression was confirmed by immunoblotting brain lysates using a ClC-4 specific antibody generated as previously described (Mohammad-Panah et al., 2003). As expected, tissues from ClC-4-null mice lacked the single 85-90 kDa band corresponding to ClC-4 that was readily detected in tissues from wild-type mice (Fig. 1A). Immunoblotting tissue lysates using antibodies generated against ClC-3 or ClC-5 confirmed that similar to the mRNA levels, the protein expression levels of both ClC-3 and ClC-5 did not differ from that in tissues obtained from wild-type animals (n=4, P=0.5 for both ClC-3 and ClC-5) (Fig. 1B,C). As neither mRNA nor protein expression of the closely related ClC-3 and ClC-5 was significantly altered in ClC-4-null mice, our subsequent analysis on endosomal trafficking and acidification in primary kidney fibroblasts isolated from these mice will specifically reveal the consequences of complete ClC-4 depletion.

As we have shown previously that ClC-4 colocalizes with dextran-containing endosomes within the proximal tubule, we examined the possible consequences of complete ClC-4 depletion on fluid-phase endocytosis (Mohammad-Panah et al., 2003). Primary renal fibroblasts derived from wild-type and ClC-4-null mice were incubated with BODIPY-dextran for 30 minutes at 37°C to allow for sufficient accumulation of dextran in early endosomes (Cabezas et al., 2005). Confocal microscopy revealed that ClC-4 appears as punctuate structures that partially colocalize with BODIPY-dextran in wild-type renal fibroblasts (merge image, Fig. 2A). As expected, ClC-4 specific staining in ClC-4-null fibroblasts was absent. BODIPY-dextran distribution was similar in both ClC-4-null and wild-type cells and quantification of this uptake revealed very similar levels of accumulation between these cell groups (33.05±2.57 and 28.58±2.56, respectively; P=0.305) (Fig. 2B). This finding indicates that ClC-4 expression is not required for effective internalization via fluid-phase endocytosis.

Our previous antisense studies on LLCPK1, proximal tubule cells that endogenously express ClC-4, revealed defective transferrin (Tfn) accumulation upon reduction of ClC-4 protein expression, implying a role for ClC-4 in this pathway (Mohammad-Panah et al., 2003). To determine whether this phenotype is observed in the primary ClC-4-null fibroblasts, receptor-mediated endocytosis was assessed by allowing the uptake of diferric Tfn conjugated to a fluorophore rhodamine (Tfn-Rhd). After a 20 minutes of uptake, confocal microscopy of wild-type fibroblasts revealed a significant overlap between perinuclear punctuate structures labeled using ClC-4 specific antibody described in Fig. 1 and vesicular structures Tfn-Rhd, suggesting that ClC-4 is expressed in Tfn-bearing endosomes (Fig. 3A). In fully confluent monolayers (but not in subconfluent monolayers), staining at the cell periphery is also observed. As expected, the same anti-ClC-4 antibody failed to label specific structures in fibroblasts obtained from ClC-4-null mice. Furthermore, the uptake of Tfn-Rhd into these structures was markedly reduced in the ClC-4-null fibroblasts (Fig. 3B). Quantification of Tfn-Rhd uptake revealed a significant impairment of Tfn accumulation in ClC-4-null fibroblasts relative to wild-type fibroblasts, suggesting ClC-4 is an essential participant in receptor-mediated endocytic uptake (Fig. 3D) (P=0.0012).

In order to confirm that this cellular defect was directly related to the lack of ClC-4 expression and not due to a non-specific defect in the endocytic pathway, we performed a reconstitution experiment wherein ClC-4-deficient fibroblasts were transfected with recombinant ClC-4 (Fig. 3B). Reconstitution of ClC-4 expression was sufficient in rescuing endosomal Rhd-Tfn uptake to wild-type levels (bar graph, Fig. 3D), indicating that ClC-4-null fibroblasts report the specific consequences of ClC-4 removal.

Impairment of endosomal acidification is thought to be a major contributor to the defect in receptor-mediated endocytosis in cells depleted of ClC-5 (Gunther et al., 2003; Hara-Chikuma et al., 2005a; Mohammad-Panah et al., 2003). An impairment of endosomal acidification in the absence of ClC-4 may result in the reduced endocytic uptake of Tfn observed in ClC-4-null fibroblasts. To test whether ClC-4 contributes to endosomal acidification, we used fluorescence ratio imaging to measure the pH of early and sorting endosomes of wild-type and ClC-4-null fibroblasts in situ. Briefly, cells were loaded with a pH-sensitive and endosomal specific fluorophore either Tfn or dextran conjugated to fluorescein (i.e. FITC-Tfn or FITC-dextran). Endocytic uptake of FITC-Tfn by the ClC-4-null cells was too low to permit accurate measurements pH of transferrin bearing endosomes. As fluid-phase endocytosis is not affected in these cells, the pH of FITC-dextran bearing endosomes could be quantified. Wild-type and ClC-4-null fibroblasts were loaded with FITC-dextran for 5 minutes to allow for sufficient labeling of early and sorting endosomes. Following dextran uptake, a pH calibration curve was developed by changing the medium with calibrated buffers of decreasing pH increments in the presence of ionophores (Fig. 4A). Interpolation of the standard curve (see Materials and Methods) indicated that the pH of early and sorting endosomes of ClC-4-null fibroblasts were significantly less acidic in comparison with the endosomes of wild-type fibroblasts (5.60±0.01 and 6.08±0.10, respectively; ^*P=0.0063) (Fig. 4B), suggesting that ClC-4 contributes to the regulation of pH within these compartments. As the relative acidity of this compartment in wild-type cells is consistent with that previously determined for dextran-accessible early and sorting endosomes in fibroblasts, we are confident this acidification defect (∼0.42 pH units) reflects a true phenotype produced by the absence of ClC-4 (Yamashiro and Maxfield, 1984; Yamashiro and Maxfield, 1987a). Furthermore, our finding that the 5-minute uptake of dextran was not impaired, despite the significant alkalinization of this compartment, supports previous studies that suggest the initial stages of internalization by fluid phase endocytosis is not impaired by endosomal alkalinization (Hurtado-Lorenzo et al., 2006).

In addition to impaired endosomal acidification, mis-trafficking of endocytic receptors away from the cell surface also contributes to the endocytic defect observed in cells devoid of ClC-5 (Christensen et al., 2003). To determine whether mis-trafficking of the Tfn receptor away from the cell surface contributes to the reduced endocytic uptake observed in the absence of ClC-4, Tfn receptor levels at the cell surface were assessed by cell surface biotinylation. Surprisingly, biotinylation revealed a modest but significant enhancement of the surface expression of the Tfn receptor in ClC-4-null fibroblast relative to wild-type cells (1.07±0.03 and 1.00±0.01, ^*P=0.015) (Fig. 5A,B). These results indicate that the lack of cell surface receptors does not contribute to the impaired endocytic phenotype observed and suggests that there may be a defect in Tfn binding to these receptors. To determine whether the Tfn binding at the cell surface is impaired both wild-type and ClC-4-null cells, samples were cooled to 0°C for 30 minutes to stop endocytosis and then incubated with biotinylated-Tfn (Tfn-Btn) to allow for binding to surface Tfn receptors only. Cells were then washed and their lysates were analyzed by SDS-PAGE, transferred to nitrocellulose and probed with extravidin. Relative to β-actin, the amount of Tfn-Btn binding to the cell surface was greatly reduced in ClC-4-null fibroblasts (0.68±0.036 and 1.00±0.039, respectively; ^*P=0.0040) (Fig. 5C,D), indicating that the number of receptors accessible for binding is dramatically reduced in the complete absence of ClC-4.

An acidic endosome with a pH of less than 6.0 is required for the effective elution of iron from internalized diferric-Tfn (Yamashiro and Maxfield, 1987b), thus allowing the recycling of apo-Tfn back to the cell surface where it dissociates. Endosomal alkanization has been shown to lead to ineffective iron removal from Tfn in the endosome, causing the recycling of diferric-Tfn back to the cell surface (Yamashiro and Maxfield, 1987b) where it remains tightly bound to its receptor, reducing receptor occupancy. Pre-occupancy of Tfn receptors with endogenous diferric-Tfn and thus reduced receptor availability for Tfn-Btn in ClC-4-null fibroblasts may account for the reduced surface binding observed above. To determine whether this is the case, the amount of Tfn-Btn bound to the cell surface with and without a treatment to remove pre-bound diferric-Tfn (see Materials and Methods) was quantified. Pre-treatment of ClC-4-null fibroblasts under conditions to remove surface bound diferric-Tfn prior to Tfn-Btn binding significantly increased surface Tfn-Btn binding relative to no treatment. In fact, the treatment fully recovered the Tfn-Btn binding defect to wild-type levels (1.09±0.02, P=0.1633) (Fig. 5C,E) confirming that pre-occupancy of Tfn receptors at the cell surface accounts for the reduced Tfn receptor accessibility at the cell surface observed in ClC-4-null fibroblasts. As endosomal acidification is the primary determinant of iron dissociation from transferrin and the subsequent release of transferrin from its receptor, we suggest that the reduced receptor accessibility measured in ClC-4-null fibroblasts reflects defective endosomal acidification.

To determine the extent to which impaired receptor accessibility contributes to the defect in endocytic uptake observed in ClC-4-null fibroblast, cells were incubated with Tfn-Btn for time intervals up to 20 minutes. Normalization of the amount of Tfn-Btn accumulated to actin revealed similar times for half-maximal uptake for wild-type and ClC-4-null fibroblasts (7.7±2.3 and 9.4±3.2 minutes, respectively; P=0.2028), but the amount of accumulated Tfn-Btn after 20 minutes in ClC-4-null fibroblasts relative to wild-type cells was significantly reduced in the absence of ClC-4 (0.75±0.07 and 1.25±0.07, ^*P=0.0053) (Fig. 6A). However, when the amount of Tfn-Btn accumulated at each time interval is normalized to the amount of Tfn receptors at the cell surface that are accessible for binding, there is no significant difference between the amount of Tfn-Btn accumulated at 20 minutes in ClC-4-null and wild-type renal fibroblasts (0.71±0.06 and 0.85±0.06, P=0.1802) (Fig. 6B).

To determine whether the recycling pathway in receptor-mediated endocytosis was impaired in the absence of ClC-4, wild-type and ClC-4-null fibroblasts were loaded with Tfn-Btn for 20 minutes. The Tfn-Btn was then chased with medium for the time intervals indicated in Fig. 6C. There was no significant difference in the amount of internalized Tfn-Btn recycled back to the cell surface after 20 minutes between wild-type or ClC-4-null fibroblasts (0.32±0.11 and 0.16±0.044, respectively; P=0.259) (Fig. 6C) and the times for half-maximal recycling were similar (t_1/2=6.62 minutes and t_1/2=6.13 minutes, respectively), indicating the absence of ClC-4 does not impair the rate of endocytic recycling of the Tfn receptor. Taken together, these results indicate that reduced binding of Tfn to its receptor at the cell surface contributes to the majority of the internalization defect observed in endocytosis in the absence of ClC-4 expression and indicates that ClC-4 plays a key role in modulating interactions between Tfn and its receptor within the endosome.

In order to determine whether the functional expression of ClC-4 contributes to the regulation of other receptor-mediated trafficking pathways, we examined the fate of the EGF receptor after ligand (EGF) binding. After EGF binding, the receptor is trafficked from early endosomes to late endosomes and then to lysosomes for degradation (Yarden and Sliwkowski, 2001). We compared the kinetics of EGF-induced EGFR degradation in ClC-4-null cells and their wild-type controls. These studies (shown in Fig. 7) reveal that there is no significant effect of depleting ClC-4 expression on delivery of the EGFR to lysosomes for degradation. The time for half-maximal degradation of the EGFR after EGF addition was not delayed in ClC-4-depleted cells relative to wild-type cells (t_1/2=68.3 minutes versus t_1/2=80.0 minutes, respectively). One and one half hours after addition of EGF, abundance of the EGF receptor had decreased to 26.3±8.2% of initial levels for ClC-4-null and 31.7±3.3% for wild-type cells (P=0.4256). Therefore, ClC-4 function is not important for sorting of the EGFR to lysosomes or in the degradative function of lysosomal compartments.

Although detailed electrophysiological studies have confirmed that ClC-4 exhibits a comparable function with the related protein ClC-5, and acts as a Cl^–/H⁺ antiporter (Picollo and Pusch, 2005; Scheel et al., 2005), it was unclear previously whether ClC-4 is functionally expressed in the same organelle as ClC-5, i.e. the endosome. As previously mentioned, this uncertainty was due in part to the lack of disease-causing mutations in CLCN4 pointing to a defect in endosomal function, whereas, in the case of CLCN5, mutations are associated with Dent's disease and proteinuria caused by a defect in endocytosis of this protein via the megalin receptor (Christensen et al., 2003; Piwon et al., 2000). Furthermore, there was no clear-cut renal phenotype in mice in which Clcn4 expression is disrupted, possibly reflecting the functional compensation by other closely related ClC family members (Jentsch, 2008) (R.M.-P., L.W., B.E.S., Y.W., L.J.H., X.-D.L. and C.E.B., unpublished). Our previous studies on LLPCK cells, in which we reduced endogenous ClC-4 using an antisense strategy and measured a decrease in the uptake of transferrin (Tfn), suggested that ClC-4 is functionally expressed in Tfn-bearing endosomes (Mohammad-Panah et al., 2003). However, it remained unclear whether this cellular phenotype reported the specific depletion of ClC-4 protein and the mechanism by which Tfn uptake was reduced was not defined. The current studies, which employed primary cultures of renal fibroblasts obtained from ClC-4-null and their wild-type siblings, provides direct evidence that ClC-4 is functionally expressed in endosomal membranes and, further, that its function in this organelle is essential for pH-dependent processing of the Tfn receptor.

As ClC-4 depletion had no impact on the kinetics of EGFR downregulation through lysosomal degradation, ClC-4 function is specifically involved in regulating the functional expression of recycling receptors (i.e. Tfn receptor) rather than receptors that are targeted for degradation after internalization (EGFR). These new findings support a growing literature, which suggests that these two cargo-specified endosomal pathways may be distinguished on the basis of their associated proteins (Leonard et al., 2008).

It is well documented that the Tfn receptor must enter an acidic but non-lysosomal compartment with a pH of less than 6.0 to permit the dissociation of iron from receptor bound diferric-Tfn (Dautry-Varsat et al., 1983). After iron removal, apo-Tfn is recycled back to the cell surface where it rapidly dissociates from its receptor and becomes accessible for subsequent diferric-Tfn binding and internalization. Impaired endosomal acidification is known to inhibit iron removal from Tfn in the endosome, causing diferric-Tfn to be recycled back to the cell surface where it remains tightly bound to its receptor (Yamashiro and Maxfield, 1987b), thereby reducing Tfn receptor accessibility for continued uptake. In the absence of ClC-4, we detected enhanced diferric-Tfn bound to the Tfn receptor at the cell surface, providing support for our proposal that the defective diferric-Tfn uptake measured in ClC-4-null fibroblasts is due to a primary defect in endosomal acidification. Although we have yet to determine whether ClC-4-null mice exhibit iron deficiency, the upregulation of Tfn receptor expression as we observed in ClC-4-null fibroblasts, is a known response to iron deficiency in cultured cell lines (Muller-Eberhard et al., 1986).

Similar to the current studies, the role of ClC-3 and ClC-5 in endosomal acidification had been previously established in cell lines derived from ClC-3 or ClC-5-null animals (Gunther et al., 2003; Hara-Chikuma et al., 2005a; Hara-Chikuma et al., 2005b; Mohammad-Panah et al., 2003; Stobrawa et al., 2001). For example, Hara-Chikuma and colleagues demonstrated the crucial role for ClC-3 in acidifying the late endosomal compartment in comparative, ratio-imaging of late endosomes labeled with FITC-α2-microglobulin from wild-type and Clcn3 KO animals, a result they also supported by direct analysis of endosomes in vitro (Hara-Chikuma et al., 2005b). A role for ClC-5 in endosomal acidification has been demonstrated through the study of endosomes and cells isolated from Clcn5 KO mice (Gunther et al., 2003; Hara-Chikuma et al., 2005a; Mohammad-Panah et al., 2003). For example, Hara-Chikuma and colleagues were able to establish an acidification defect of 0.6 pH units in the early endosomes of Clcn5 KO cells through ratio-imaging of FITC-Tfn in these compartments (Hara-Chikuma et al., 2005a). The localization of ClC-3 in late endosomes, instead of early and recycling endosomes where ClC-4 is most abundant, is likely to be one reason why endogenous ClC-3 is incapable of compensating for ClC-4 depletion in primary renal fibroblasts. However, it remains unclear why ClC-5, which is thought to colocalize with ClC-4 in early and recycling endosomes, cannot fully compensate in its absence in these cells.

Clues to why ClC-5 is incapable of compensating in endosomal acidification in the face of ClC-4 depletion may lie in the finding that ClC-4 and ClC-5 exhibit Cl^– currents that are differentially regulated by extracellular pH. Cl^– currents mediated by both ClC-4 and ClC-5 are inhibited by external acidity; however, ClC-5 is more sensitive than ClC-4, exhibiting a half maximal inhibition at pH 6.0 rather than at pH 5.5 for ClC-4 (Friedrich et al., 1999). These studies were confirmed in our own laboratory (R.M.-P., unpublished). In support of this, a higher sensitivity of ClC-5 to acidity was also observed in the analysis of its full transporter activity where ClC-4 exhibited more activity at an external pH of 5.8 than did ClC-5 (Picollo and Pusch, 2005). Therefore, although the two proteins may co-exist in overlapping compartments, their contribution to the function of these compartments may not overlap because of their differential regulatory properties.

In addition to the differential regulation, ClC-5 may not be able to compensate for the depletion of ClC-4 if these two proteins form a heterodimeric complex with a distinctive function and/or localization. This hypothesis was supported by biochemical studies wherein both proteins could be co-immunoprecipitated from rodent kidney tissue (Mohammad-Panah et al., 2003). Recently, this idea gained support from another laboratory using heterologous expression systems wherein ClC-4 could be co-immunoprecipitated with ClC-3 or ClC-5 (Suzuki et al., 2006). However, until we understand the differential properties of each individual ClC protein, the functional consequences of complex formation cannot be rigorously tested.

At present, it is not clear how ClC-4 or the related proteins ClC-3 and ClC-5 may actually function to acidify intracellular compartments. Initially, it was thought these proteins exist as channels to mediate the inward influx of Cl^– ions that serve to dissipate the large intraluminal positive potentials generated by the vacuolar ATPase during acidification (Gunther et al., 1998). However, it has recently been reported that ClC-4 and ClC-5 are not Cl^– channels at all but are Cl^–/H⁺ anti-porters (Picollo and Pusch, 2005), similar to the prokaryotic ClC protein ecClC (Jentsch et al., 2005; Scheel et al., 2005). The direction of Cl^–/H⁺ exchange was convincingly demonstrated by Verkman and colleagues who showed both ClC-3 and ClC-5 mediate the accumulation of Cl^– ions into endosomes (Hara-Chikuma et al., 2005a; Hara-Chikuma et al., 2005b) and thus in light of recent findings, presumably the exit of luminal H⁺ ions. Although at first it was not clear how this anti-porter activity could facilitate acidification of the endosome, if one assumes the stoichiometry of Cl^– and H⁺ exchange is 2:1, as demonstrated for their prokaryotic homolog ecClC (Accardi and Miller, 2004; Accardi et al., 2005) is conserved for ClC transporters, the activity would remain electrogenic and would be able to dissipiate positive luminal potentials. However, the stoichiometry of exchange by ClC-4 and ClC-5 transporters could not be assigned unambiguously (Picollo and Pusch, 2005) and thus requires further characterization.

In summary, this comparative study on renal fibroblasts isolated from ClC-4-null mice supports a view that ClC-4 contributes to acidification of endosomes in vivo, a phenomenon that is crucial for regulating Tfn receptor occupancy in this cell line. This study resolved the molecular mechanism for defective Tfn uptake observed in ClC-4 depleted cell lines and points to a possible role of ClC-4 in iron uptake via the Tfn:Tfn receptor complex. The upregulation of the Tfn receptor at the cell surface in the absence of ClC-4, a known response to iron deficiency in cultured cell lines supports this putative role for ClC-4 in iron absorption. It remains unknown why mice do not exhibit an overt renal phenotype in the face of ClC-4 depletion, but the possibility exists that endogenous ClC-5 may partially compensate for defective Tfn receptor function in proximal tubule epithelia through its supportive role in trafficking of the megalin/cubulin complex, which also binds and internalizes Tfn (Christensen et al., 2003). Future work will focus on uncovering the physiological significance of ClC-4 mediated iron release from Tfn in the endosome and will serve to elucidate the role of ClC-4 in mediating other ligand:receptor dissociation events dependent upon an acidic endosome, such as interactions with the megalin/cubulin receptor complex (Czekay, et al., 1997).

ClC-4-null mice were generated by breeding two strains of mice, Mus spretus and C57BL/6J, as described previously (Rugarli et al., 1995). Clcn5 and Clcn3 knockout mice were kindly provided by S. Guggino (Johns Hopkins University, Maryland, MD (Silva et al., 2003) and F. Lamb (Iowa University, Iowa City, IA) (Arreola et al., 2002), respectively.

Kidney tissues from wild-type and ClC-4-null mice were homogenized in Trizol reagent (Invitrogen). RNA was extracted and was further purified with RNAeasy MinElute Cleanup kit (Qiagen) and was reverse-transcribed with SuperScript first-strand synthesis system (Invitrogen). The cDNA was amplified by PCR with 0.2 μM of gene-specific primers (mClC4, mClC-3A, mClC-3B and GADPH) or 0.1 μM of mClC-5. With an initial incubation of 10 minutes at 95°C, the amplification was preceded at 94°C for 15 seconds and 56°C for 1 minute for 50 cycles. The following pairs of primers were used: mClC-4F, gcgtctcatcgggtttgc; mClC-4R, ttgccacaatgccctcttg; mClC-3AF, tgtaactcacaacggacgcctc; mClC-3AR, tattgaagcggggtcttggttt; mClC-3BF, gaaagcttggtctgaggcagt; mClC-3BR, ctccacgtgctgctttagttg; mClC-5F, aatcatcaccaaaaaggatgtgttaa; mClC-5R, ccatggtccgcaatgtcc.

cDNA coding for human ClC-4 was generously provided by Al George (Vanderbilt University, Nashville, TN). The anti-ClC-4 polyclonal antibody was generated as previously described (Mohammad-Panah et al., 2003). Rabbit polyclonal anti-ClC-3 antibody was obtained from Chemicon and monoclonal anti-HA, anti-transferrin receptor, anti-β-actin and anti-epidermal growth factor receptor antibodies were obtained from Covance, Zymed, Sigma and Santa Cruz, respectively.

Expression of ClC-4 and ClC-3 proteins in wild-type and ClC-4-null mice brains and ClC-5 protein in wild-type and ClC-4-null mice kidney was determined by immunoblotting as described previously (Mohammad-Panah et al., 2002). Tissue lysates were analyzed by SDS-PAGE and immunoblotted with anti-ClC-4, anti-ClC-3 or anti-ClC-5.

Fluid-phase endocytosis was measured by assaying the internalization of BODIPY-labeled dextran (MW=10,000 Da) (Molecular Probes, Leiden, The Netherlands). Wild-type and ClC-4-null cells were incubated with 1 mg/ml BODIPY-dextran at 37°C for 30 minutes and washed with PBS. After fixing with 4 % (v/v) paraformaldehyde cells were washed and immunolabeled for ClC-4.

Primary cells were transiently transfected with appropriate ClC-4 cDNA using FuGENE 6 Transfection Reagent (Roche Diagnostic, Mannheim, Germany) according to the manufacturer's protocol.

Receptor-mediated endocytosis in primary fibroblasts was measured using confocal microscopy by assaying the uptake of diferric transferrin conjugated to tetramethylrhodamine (Rhd-Tfn, Molecular Probes, Leiden, The Netherlands). Briefly, serum starved cells were loaded with 5 μg/ml of Rhd-Tfn for 60 minutes. Following fixation, cells were labeled with ClC-4 antibody for subsequent immunolocalization.

Endocytosis was also assessed through the uptake of biotin-labeled human transferrin (Tfn-Btn) (Molecular Probes) as previously described with minor modifications (Nieland et al., 2004). Serum-starved cells were incubated at 37°C with 35 μg/ml Tfn-Btn for indicated time points. Cells were washed with cold PBS (pH 4.5) and cold neutral pH buffer [150 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 20 mM HEPES (pH 7.4)] to strip the surface bound Tfn-Btn. Total surface-bound Tfn-Btn was assessed by incubating the cells with Tfn-Btn for 20 minutes on ice. Recycling of Tfn-Btn was followed after a 20 minute uptake at 37°C in medium supplemented with serum, 3 μg/ml unlabeled transferrin and 100 μM desferrioxamine mesylate (Nieland et al., 2004). All cells were lysed directly into SDS sample buffer [63 mM Tris-Cl (pH 6.8), 10% glycerol, 0.5 mM EDTA, 2% SDS, 2% β-mercaptoethanol, and 50 mM dithiothreitol] and analyzed by SDS-PAGE and immunoblotting using Extravidin (Sigma, USA).

All fluorescent images were acquired using the palette function of LSM510 confocal acquisition software. The rhodamine-specific immunofluorescence was converted to a grayscale image and the background subtracted prior to quantification. Mean pixel intensity of the grayscale image was measured (Image J software, USA) for the perinuclear region by superposing a box at eight locations around the nucleus.

Early endosomes were labeled at 37°C with 1 mg/ml of FITC-dextran (MW=10,000 Da) by a 5-minute pulse, followed by a 5-minute chase in primary cell lines. Cells were washed and placed in a thermostatted Leiden chamber holder on the stage of a Leica DM IRB microscope. A filter wheel (Sutter Instruments, Novato, CA) was used to alternate between two excitation filters: 440 nm (pH-independent) and 490 nm (pH-dependent). Fluorescent light was directed at the sample using a 505 nm dichroic mirror, with the emitted light filtered using a 510 nm filter placed in front of a Cascade II camera (Photometrics, Tucson, AZ). After acquisition of baseline data, an in situ calibration was performed by sequentially bathing the cells with isotonic K⁺-solution (145 mM KCl, 20 mM MES, 10 mM glucose, 1 mM MgCl₂ and 1 mM CaCl₂) of decreasing pH values (pH 7.15 to 4.58), containing 10 μg/ml nigericin. Calibration curves were obtained by plotting the fluorescence intensity ratio (490:440 nm) against pH and endosomal pH values were calculated by interpolation from the curve using the measured 490:440 nm ratios corrected for background.

Primary cells were subjected to cell surface biotinylation as described previously (Dhani et al., 2003). After biotinylation, eluants were analyzed for the presence of the Tfn receptor. Band densitometry using Image J (NIH Imaging) was used to quantify the level of biotinylated Tfn receptor (surface levels) relative to its total cellular expression level.

Serum-starved primary fibroblasts were placed on ice for 30 minutes to stop endocytosis. Half the cells of each genotype were stripped of surface-bound diferric transferrin by treatment with 50 μM desferrioxamine in PBS (pH 4.5) for 20 minutes. All cells were washed in neutral buffer and incubated with 35 μg/ml of Tfn-Btn for 20 minutes to allow for Tfn receptor binding at the cell surface. Cells were washed with cold neutral buffer to remove unbound ligand and lysed directly into SDS sample buffer and analyzed by western blotting as described above.

Degradation of the epidermal growth factor receptor (EGFR) was assessed as previously described with minor modifications (Progida et al., 2007). Briefly, wild-type and ClC-4-null fibroblast were serum starved for 1 hour in the presence of 10 μg/ml cycloheximide. Cells were then simulated with 100 ng/ml EGF ligand for times intervals indicated to promote EGFR degradation. The intensity of the EGFR remaining at each time point was calculated relative to the intensity measured without EGF treatment.

Data are shown as mean±s.e.m. Statistical analyses, ANOVA followed by Bonferroni's non-paired t-test or paired t-test as appropriate, were conducted using Prism software and P values of less than 0.05 were considered significant.