Differential Expression of AE1 in Renal HCO,-secreting and -reabsorbing Intercalated Cells*

The cortical collecting duct of the kidney contains two types of intercalated cells that transport HCO, in opposite directions. HCO, reabsorption takes place in the a-type intercalated cells, which express a CVHCO, exchanger on the basolateral membrane. This exchanger is the product of the anion exchanger 1 (AE1) or band 3 gene. HCO, secretion occurs in the @-intercalated cells, which have a CVHCO, exchanger on the apical membrane. Based on studies in an immortalized cell line, recently it was proposed that the apical anion ex- changer of @-intercalated cells is also AE1 (van Adels-berg, J. S., Edwards, J. C., and Al-Awqati, Q. (1993) J. Biol. Chem. 268, 11283-11289). In the present study we reinvestigated this issue by determining the distribu- tion of AE1 mRNA and protein in the two intercalated cell types using cells freshly isolated from the native epithelium. Using quantitative reverse transcriptase polymerase chain reaction, we found that a-intercalated cells, isolated from rabbit kidney by fluorescence-activated cell sorting, have high levels of AE1 mRNA, whereas @-inter- calated cells express very low levels. The ratio of AE1 mRNA levels in a- uersus @-intercalated cells averaged 10.1 * 2.6.

The cortical collecting duct of the kidney contains two types of intercalated cells that transport HCO, in opposite directions. HCO, reabsorption takes place in the a-type intercalated cells, which express a CVHCO, exchanger on the basolateral membrane. This exchanger is the product of the anion exchanger 1 (AE1) or band 3 gene. HCO, secretion occurs in the @-intercalated cells, which have a CVHCO, exchanger on the apical membrane. Based on studies in an immortalized cell line, recently it was proposed that the apical anion exchanger of @-intercalated cells is also AE1 (van Adelsberg, J. S., Edwards, J. C., and Al-Awqati, Q. (1993) J.
Using quantitative reverse transcriptase polymerase chain reaction, we found that a-intercalated cells, isolated from rabbit kidney by fluorescence-activated cell sorting, have high levels of AE1 mRNA, whereas @-intercalated cells express very low levels. The ratio of A E 1 mRNA levels in a-uersus @-intercalated cells averaged 10.1 * 2.6. In addition, metabolic acidosis increased the levels of A E 1 mRNA by --fold in cortical collecting duct cells. This difference was confirmed by Northern blotting. Western blotting using an antibody against rabbit A E 1 revealed a major immunoreactive product with a molecular weight of "110 kDa in cortical collecting duct cells. Deglycosylation reduced the size of the immunoreactive product to -90 kDa, which is compatible with the presence of a truncated form of AE1. Metabolic acidosis increased the intensity of the AE1 immunoreactive band. The level of AEl-immunoreactive protein was significantly higher in a-intercalated cells than in P-intercalated cells.
In aggregate, these data provide evidence for the differential expression of AE1 in HC0,-reabsorbing uersus HC0,-secreting renal intercalated cells both at the mRNA and at the protein level. These results give no support to the concept that AE1 functions both as a basolateral and an apical anion exchanger in cortical collecting duct cells.
The fine regulation of acidhase balance by the kidney takes place in the collecting ducts. The cortical collecting duct (CCD)' DK 39523, DK 45647, and DK 41841. The costs of publication of this * This work was supported by National Institutes of Health Grants article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  The abbreviations used are: CCD, cortical collecting duct; ICC, in-contains two types of intercalated cells (ICC), a-and ptype, which transport HCO, in opposite directions (for review, see Ref. 1). HCO, reabsorption occurs in a-ICCs, which are endowed with an apical H-ATPase and a basolateral CVHCO, exchanger, which is the product of the AE1 (band 3) gene (2). The cell responsible for HCO, secretion is the 0-ICC, which functionally is a mirror image of the a-cell and is modeled with a basolateral H-pump and an apical CVHCO, exchanger (1). Changes in acidhase balance are thought to result in an interconversion of the two ICC types by retargeting the transporters to opposite poles of the cells (3).
The opposing functional polarity of a-and p-ICC could be achieved either by expressing functionally similar but structurally different transporters for apical and basolateral acid or base extrusion or by targeting the same proteins to opposite poles in the two cells. There is immunohistochemical evidence that in the rat (but not in the rabbit) at least two subunits of the H-ATPase can occur with either apical or basolateral polarity in a-and p-ICCs, respectively (4). On the other hand, the question whether the same anion exchanger can occur in opposite membranes of a-and p-ICCs is still unresolved. The basolateral and apical anion exchangers differ in inhibitor sensitivity and kinetic properties (2), and, most importantly, antibodies directed against the basolateral exchanger fail to stain the apical membrane of p-ICCs (5, 6). Nevertheless, recently the possibility was raised that the apical exchanger of HC0,-secreting p-ICCs is the same protein as the basolateral exchanger of a-ICCs, Le. A E 1 (7). These authors reported that apical membrane preparations from cultured peanut lectin-positive cells (presumably 0-ICC) contain a protein that reacts with an AI31 antibody on Western blots and has a size similar to that ofAEl (7).
Since the maximal rate of HCO, secretion in the CCD of alkalotic rabbits is comparable to the rate of HCO, reabsorption in the CCD of acidotic animals (8) (if indeed the same protein functions both as the basolateral CVHCO, exchanger in a-ICCs and the apical exchanger in p-ICCs), one can assume that A E 1 mRNA and protein are expressed at comparable levels in a-and P-type ICCs. Also, if the apical and basolateral CVHCO, exchangers are products of the same gene, one would expect that acidosis increases A E 1 expression in a-ICCs to achieve maximal HCO, reabsorption, whereas alkalosis would increase A E 1 expression in p-ICC concurrent with enhanced HCO, secretion. The findings of this study, however, do not conform with these predictions. Here we report that A E 1 mRNA and protein are predominantly expressed in a-type ICCs, and there is a significant increase in the levels of both the mRNA and the protein in acidotic uersus alkalotic animals.
EWERIMENTAL PROCEDURES Animals-Male New Zealand white rabbits, weighing 1.5-2.0 kg, were used. The animals were kept on standard diet and had access to tercalated cells; AE1, anion exchanger 1; PCR, polymerase chain reaction; bp, base pair(s). water ad libitum. Metabolic alkalosis was induced by intravenous infusion of 15 mmoVkg NaHCO,, 16-20 h before sacrifice. Metabolic acidosis was achieved by an intragastric load of 15 mmolkg of NH,CI. To keep the amount of Na' load constant, this latter group also received 15 mmoVkg NaCI, intravenously. For the last 12 h before the experiments, the rabbits were on restricted food intake (3 02). Urine samples were taken from the bladder immediately before sacrifice for the determination of urinary pH.
Cell Isolation-CCD cells were isolated from the renal cortex by solid phase immunoadsorption, using a monoclonal antibody against an ectoantigen on these cells, as described previously (9, 10). The immunodissected CCD cells were further fractionated into the three collecting duct cell types (i.e. a-and p-ICCs and principal cells) by fluorescenceactivated cell sorting using cell-specific markers as described (11, 12). Principal cells were identified with an fluorescein isothiocyanate-conjugated antibody that reacts specifically with these cells (DT.17; Ref, 11) p-ICCs with peanut lectin agglutinin coupled to phycoerythrin as described in detail elsewhere (11, 12), whereas a-ICCs were operationally defined as the DT.17-and peanut lectin agglutinin-negative population. To aid in the discrimination between live and dead cells, CCD cell preparations were also stained with DAPI (0.1 pglml), which is excluded from viable cells. The purity of the sorted cells was determined by immunocytochemistry using other cell-specific markers as described (11, 12) and was -92% for p-ICC, -91% for principal cells, and -82% for a-ICC.
RNA Isolation and cDNA Synthesis-Total RNA was isolated using TRI ReagentTM (Molecular Research Center, Inc.), and poly(A)+ RNA was isolated by SDS lysis and oligo(dT) selection (13). RNA concentrations were calculated from the ODs a t 260 nm. cDNA was synthesized using 0.5-2 pg of total RNA and 200 units1pg RNA of Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.). The reaction mixture also contained 50 mM Tris-HC1 (pH 8.31.75 mM KCl, 10 mM dithiothreitol, 3 mM MgCI,, 0.5 mM each dGTP, dATP, dTTP, and dCTP, 20 units of RNasin (Promega), and 3.3 PM random pd(N), primers (Boehringer Mannheim). Following denaturation of the RNA a t 75 "C for 3 min and annealing of primers at room temperature for 10 min, reverse transcription was carried out a t 42 "C for 60 min, and then the reaction was terminated by heating the tubes to 75 "C for 10 min.
Primer Selection for A E l S e n s e and antisense PCR primers were designed based on the cDNA sequences of the published sequences of mouseAEl(14, 15). The oligonucleotide sequences used as AEl primers After an initial 2-min denaturation at 96 "C, PCR was carried out for 30 cycles with denaturation a t 95 "C for 1 min, annealing a t 54 "C for 1 min, and primer extension at 72 "C for 1 min. The reaction mixtures were incubated for a final extension a t 72 "C for 8 min. The relative abundance of p-actin mRNA in each CCD cell sample was determined using primers and conditions as described (16). cDNA samples derived from pairs of rabbits (one acidotic, the other alkalotic) were always amplified simultaneously in the same PCR.
After amplification, 4 pl of tracking dye was added to each sample, and 20 pl was run on a 6% polyacrylamide gel. Gels were dried, and the amount of radioactivity in the PCR products was determined using a model 425 PhosphorImagerTA' (Molecular Dynamics).
DNA Sequencing-The 580-bp AE1 PCR product generated with rabbit CCD cDNA as template was extracted by phenolkhloroform, and unincorporated primers and nucleotides were removed by centrifugal diafiltration using a Microcon-100 ultrafiltration unit (Amicon). Direct sequencing was performed using primer 5 or primer 2 by the dye deoxy terminator chemistry on an A B 1 373A automated sequencer.
Northern Blot Analysis-Northern blotting was carried out using standard protocols (17). In brief, 2.5 pg of poly(A)+ RNAoriginating from immunodissected CCD cells from acidotic and control rabbits was fractionated on a 1.2% agarose gel containing 1.1% formaldehyde. RNA was transferred to a nylon membrane (0.45 pm; MSI Inc.) and probed with a gel-purified PCR fragment generated with primers 5 and 2 and labeled with "P during the PCR. Prehybridization was performed a t 42 "C for 10-12 h in 5 x SSC, 5 x Denhardt's solution, 50% formamide, 100 pg1ml salmon sperm DNA, and 0.5% SDS. Hybridization was done using the same conditions as for prehybridization for 12 h. Two washes were carried out a t room temperature for 15 min each, with 1 x SSC, 0.1% SDS, followed by two washes with 0.25 x SSC, 0.1% SDS. After a final wash at 45 "C for 15 min with 0.1 x SSC, 0.1% SDS, the blot was exposed to x-ray film.
Western Blotting-Immunoselected CCD cells were lysed in 1% SDS, which was followed by centrifugation at 12,000 x g for 10 min. From sorted a-and p-intercalated cells, protein fractions were isolated using TRI ReagentT". Proteins were solubilized in 1% SDS, and protein concentrations were determined using the BCA protein assay reagents (Pierce). Proteins were separated by SDS-polyacrylamide gel electrophoresis on a 7.513% separatinghtacking gel, and electrotransferred to methanol-activated polyvinylidine difluoride membranes (Immobilon, Millipore). The membranes were stained with 0.05% Ponceau red, and the positions of the molecular weight standards were marked on the membrane. After 60 min of blocking with 2.5% Carnation nonfat dry milk and 50% fetal calf serum in phosphate-buffered saline a t room temperature, the blots were probed with the anti-rabbit AE1 monoclonal antibody VB4d3 (kindly provided by Drs. L. s. Ostedgaard and V. L. Schuster) diluted to 5 pg1ml in 10 mM Tris-HC1,150 mM NaCI, 0.5% Tween 20 (TBST), for 60 min a t room temperature. The blots were then washed 4 times with TBST and incubated with a horseradish peroxidase-coupled secondary antibody (anti-mouse IgG2a; Zymed) diluted to 0.5 pg/ml in phosphate-buffered saline containing 50% StabilZyme HRP (conjugate stabilizer from BSI Corp.) for 60 min a t room temperature and then washed 4 times with TBST. The reaction products were visualized using the enhanced chemiluminescence detection system (Amersham Corp.).
Deglycosylation-In several experiments, the SDS-lysate was treated with 40,000 unitdml N-glycosidase (PNGase F, from New England Biolabs) at 37 "C for 60 min, in the presence of 1% Nonidet P-40, before electrophoresis.

RESULTS
Expression of AEl mRNA in CCD Cells-Using oligonucleotide primers 5 and 2 and cDNA derived from immunoselected rabbit CCD cells as template, a -580-bp PCR product was amplified (Fig. 1). This PCR product is of the predicted size based on the published sequence of the mouse A E 1 (18). The identity of this PCR product was verified by two independent methods. First, nested PCR was performed using sense and antisense primers with start positions corresponding to nucle- otides 2038 and 2112 (primers 3 and 2) and to nucleotides 1531 and 1608 (primers 5 and 4 on the mouse AE1). The nested primer pairs yielded PCR products with the expected sizes (74 and 77 bp, respectively), indicating that the 580-bp product is indeed amplified from AE1 cDNA. Second, the identity of the 580-bp PCR product as AE1 was verified by direct sequencing, which revealed 86% sequence homology to the corresponding regions of both the rat (19) and the mouse (18) kidneyAE1. The predicted amino acid sequence of the rabbit CCD AE1 PCR product shows 87 and 85% homology to the mouse and rat AE1, respectively (14, 19). On the other hand, the AE1 PCR product shows only 62% homology to the corresponding region of the rabbit ileal AE2 (20), 61% to the mouse AE2 (211, and 56% to the mouse AE3 (22).
The level of p-actin mRNA was determined with a similar PCR technique in each CCD cDNA sample. When the relative level of p-actin mRNA in samples containing increasing amounts of cDNA was compared with that of AE1 mRNA, we found that AE1 mRNA is abundant in CCD cells, as the slope of AE1 mRNA uersus increasing amounts of starting cDNA was only about an order of magnitude lower than that of p-actin.

Effect of Metabolic Acidosis and Alkalosis on the Expression of A E l mRNA in CCD Cells-To examine whether changes in
acidhase balance result in an altered expression ofAE1 mRNA, we induced metabolic acidosis or alkalosis in pairs of rabbits. Urinary pH averaged 5.4 2 0.4 in acidotic uersus 8.3 2 0.2 in alkalotic rabbits ( p < 0.001). A E 1 mRNA levels were determined in cDNAs derived from isolated CCD cells, and the results are summarized in Fig. 2. The relative abundance ofAEl mRNA, calculated from the ratio of [32]dCTP incorporated into the 580-bp PCR product and into the 350-bp p-actin PCR product was significantly higher in acidotic than in alkalotic animals. The average increase in AE1 mRNA levels in acidosis uersus alkalosis was 4.53-fold. This difference was confirmed by Northern blotting of CCD mRNAs obtained from pooled samples from three acidotic and three control rabbits using the PCR product as probe (Fig. 3). In both cases a single hybridizing mRNA species was detected with a size of approximately 4.5 kilobases, and its intensity was significantly higher in mRNA obtained from acidotic than than from control animals (Fig. 3). mRNA in a given cell type originating from acidotic uersus alkalotic rabbits, we found that both a-ICC and p-ICC expressed significantly higher levels in acidotic rabbits (average increase following acidosis was 13.8and 6.5-fold in a-ICC and p-ICC, respectively; Fig. 4B). At the same time, the relative abundance ofAEl mRNAin a-uersus p-cells were unaltered by changes in acidhase balance; AE1 mRNA in a-ICC uersus AE1 mRNA in p-ICC was 10.69 * 2.5 in acidotic and 10.77 * 5.3 in alkalotic rabbits ( n = 5 for each group), suggesting that acidosis and alkalosis are not accompanied by opposite changes in AE1 mRNA expression in HC0,-secreting uersus HC0,-reabsorbing intercalated cells.  2) and rabbit CCD cell lysates (lanes 3 and 4 ) before (lanes I and 4 ) or following treatment with 40,000 unitslml N-glycosidase (PNGase F) at 37 "C for 60 min. Immunoblotting was performed as described under "Experimental Procedures," using antibody vB4&. B, Western blots of CCD cell lysates obtained from normal ( N ) and acidotic (A) rabbits.

Distribution o f A E l mRNA in Different CCD Cell
Effect of Metabolic Acidosis and Alkalosis on the Level ofAEl Protein in CCD Cells-To determine whether the acidosisinduced increase in AE1 mRNA levels in CCD cells is accompanied by similar changes at the protein level, we performed Western blot analysis using a monoclonal antibody against rabbit AE1 (VB4&, a generous gift of Drs. L. S. Ostedgaard and V. L. Schuster). Western analysis revealed the presence of a major immunoreactive band in both rabbit red blood cells (Fig.  5A, lunes 1 and 2 ) and CCD cell membranes (Fig. 5A, lanes 3  and 4 ) . The molecular mass of the main product in CCD cells is around 110 kDa, which is slightly larger than the size of the erythroid AE1. This was a surprising finding since the kidney form ofAEl mRNApredicts a truncated protein (19). Therefore, we tested the possibility that this large size is due to excessive glycosylation. Deglycosylation resulted in a significant decrease in molecular mass of the main band, yielding in a product that is approximately 8 kDa smaller than the erythroid form (Fig. 5A, lanes 2 and 3), confirming at the protein level for the first time that the kidney AE1 is indeed a truncated form. The intensity of the -110-kDa protein was significantly stronger in membrane preparations derived from acidotic than from normal rabbits (Fig. 5B) indicating that the increased expression of AE1 mRNA observed in acidotic rabbits is accompanied by higher AE1 protein levels.
Distribution of AE1 Protein in the lluo Intercalated Cell Q p e s S i n c e t h e PCR and Northern experiments indicated that AE1 mRNA is predominantly expressed in a-ICCs, we examined whether a similar pattern of expression exists at the protein level. Western blots of lysates from sorted a-and p-ICCs using the AE1 antibody VB4a3 are shown on Fig. 6. The band with molecular mass -110 kDa corresponds to AE1, and the bands with lower molecular masses probably correspond to degradation products generated during the protease treatment used to obtain single cell preparations for cell sorting. It is clear that a-ICCs contain much higher levels of the -110-kDa AEl-immunoreactive protein as well as its putative degradation products than p-ICCs do, suggesting that the significant differences in AE1 mRNA expression between the two intercalated cell types is accompanied by similar differences at the protein level.

P-intercalated cells (lane 2), and a-intercalated cells (lane 3)
using the anti-AEl antibody V J 3 4 A . Cells were isolated by immunoselection and fluorescence-activated cell sorting as described under "Experimental Procedures." The band with molecular weight -110 kDa corresponds to AE1, and the bands with lower molecular weights probably correspond to degradation products generated during the protease treatment used to obtain single cell preparations for cell sorting.

DISCUSSION
The exact mechanisms by which the opposing functional polarity of HCO, secreting p-ICCs and HCO, reabsorbing a-ICCs is achieved are unclear. Whereas there is evidence that the same H-ATPase subunits can occur with either basolateral or apical polarity (4,231, the idea that the two membranes harbor the same anion exchanger remains controversial. It has been recently proposed that anion exchange in both a-and p-ICC is mediated by the same AE1 gene product (7).
The major findings of this study speak against this hypothesis. If the apical anion exchanger of p-ICCs were the product of the same gene as the basolateral exchanger of a-ICCs, the mRNA levels in the two cell types should be comparable. Our results demonstrate that the levels of AE1 mRNA are significantly higher in a-ICC than in p-ICC. In fact, mRNA levels of AE1 detected in p-ICCs are comparable with those seen in principal cells, a cell type that is probably not involved in HCO, transport (cf Ref. 24). In addition, the observed ratio of AE1 mRNA expression in a-versus p-ICCs is likely to be an underestimate for two reasons. First, for technical reasons, the a-ICC population isolated in our experiments is not 100% homogenous, and therefore not all cells express AE1. Second, as we reported earlier (6, ll), the p-ICC population (isolated as peanut lectin positive cells) includes about 10% a l p hybrid cells. Thus, the expression ofAEl mRNAin "true" 0-ICC is probably even lower than the above results would suggest.
The low levels of AE1 mRNA in p-ICC make it unlikely that the apical exchanger of p-ICC is encoded by this gene, unless there is a significant discrepancy between mRNA and protein levels in the two cell types. Such discrepancy might arise if the translation ofAEl mRNAis much more efficient in p-cells than in a-cells and/or the protein is significantly more stable in the apical than in the basolateral membrane. In this case, the low levels of AE1 mRNA found in p-cells might suffice to maintain AE1 protein levels comparable with those seen in a-cells. Our Western blot data, however, refute this possibility, as the pattern of expression of AE1 protein was similar to that observed at the mRNA levels, i.e. sorted a-cells contained significantly more AE1 protein than p-ICCs did.
Another line of indirect evidence arguing against AE1 being the apical exchanger of p-ICCs is our observation that AE1 mRNA and protein levels in CCD cells (containing both a-and p-ICCs) are increased by metabolic acidosis and/or decreased by metabolic alkalosis. If the apical and basolateral exchangers were the product of the same gene, one would not expect to see significant changes in the mRNA levels in acidotic versus alkalotic animals, since in acidosis the basolateral exchanger of a-ICC, which mediates HCO, reabsorption, should be expressed at increased levels, whereas in alkalosis, the apical exchanger of p-ICCs, which mediates the extrusion of HCO, from the cell into the urine, should increase. This, however, is not the case; both in acidosis and in alkalosis, a-ICCs expressed significantly higher (about 10-fold) levels of A E 1 mRNA than did p-ICCs.
Our results agree well with those obtained by Da Silva et al.
L. Schuster for providing monocIonaI antibody W34& against rabbit band 3.
isolated ICCs, whereas their study examined AE1 protein in cultured ICCs and a transformed cell line. Since our earlier studies revealed a remarkable plasticity of p-ICCs in culture, which spontaneously differentiate into a-ICC and principal cells (27, 28), it is possible that the cultured cells in Van Adelsberg's study were also inhomogeneous, and contained both pand a-ICCs, which might be responsible for the expression of AE1.
In summary, the results of this study suggest that HCO, secretion and reabsorption in the CCD are probably mediated by structurally different anion exchangers. Our recent observations that another member of the anion exchanger family, AE2 (21, is expressed at high levels in CCD cells and changes in acidibase balance regulate AE1 and AE2 expression in opposite directions (29) reinforce the notion that multiple anion exchangers participate in the regulation of HCO, transport in these cells.