brief communications

nature genetics • volume 33 • february 2003 125

Numerous physiological processes dependon the maintenance of systemic pHwithin a narrow range. Fundamental topH homeostasis are α-intercalated cellsin the collecting duct of the nephron,which are highly polarized such thatprotons are secreted across the apicalmembrane into the urine, in parallelwith reclamation of bicarbonate acrossthe basolateral membrane by theCl–/HCO3

– exchanger AE1. Functionalfailure of α-intercalated cells leads todRTA, characterized by metabolic acido-sis in conjunction with renal tract calci-fication, osteopenia and, in some cases,rickets and growth retardation. In con-trast with the heterogeneity of recessivedRTA, all reported cases of ddRTA areassociated with mutations in SLC4A1,encoding AE1 (ref. 1). Most of theseinvolve a missense alteration at Arg589.An exception is a mutation that trun-cates AE1 by 11 residues at the C termi-nus, resulting in an indistinguishablephenotype2.

We explored defective trafficking ofmutant AE1 as a putative molecular basisfor ddRTA for several reasons. First,haplo-insufficiency probably cannotexplain ddRTA, because heterozygousloss-of-function mutations in the longerisoform of SLC4A1 expressed in erythro-cytes (which cause hereditary spherocyto-sis) are usually not associated with a defectin renal acidification. Second, mutant AE1proteins associated with ddRTA haveshown almost normal anion-transportfunction and dimerization in heterolo-gous expression studies3–6, suggesting analternative dominant negative mecha-nism. Third, the cytoplasmic C terminusof AE1 (CLDADDAKATFDEEEGRDEYDEVAMPV) contains a putative tyrosine-based sorting signal, and these have beenimplicated in basolateral targeting inpolarized epithelia7,8.

We examined the distribution of epi-tope-tagged full-length wild-type AE1and a similar construct into which the

disease-associated truncation had beenintroduced (AE1∆11) in polarized kid-ney-derived cells. In the absence of anideal cell model for α-intercalated cells,we used MDCK cells (principal cell ori-gin) and rat IMCD cells (which expressAE1; ref. 9). In both cell types, wild-typeAE1 localized to the basolateral plasmamembrane domain (Fig. 1a,c). In markedcontrast, AE1∆11 was found at bothbasolateral and apical cell surfaces andsometimes also intracellularly (Fig. 1b,d).

The amount of intracellular AE1 washigher where cells were not fully polar-ized, consistent with evidence suggestingthat loss of epithelial polarity can causeintracellular retention of some proteinsthat normally reside at the plasma mem-brane10. This emphasizes the importanceof polarized cell models for studiesaddressing the fate of membrane proteinswith asymmetric distributions. Previousreports in non-polarized systems sug-gesting that mutant AE1 moleculesR589H and AE1∆11 cannot reach the cellsurface may not, therefore, adequatelyreflect the in vivo situation5,6. ForAE1∆11, our results suggest that ddRTAis not caused by the absence of protein atthe basolateral surface but rather by itsinappropriate presence at the apical sur-face. We predict that any apicalCl–/HCO3

– flux would alter electrochem-ical balance across the cell, impairingboth essential unidirectional basolateralbicarbonate reclamation and apical pro-ton secretion.

The non-polarized distribution ofmutant AE1 could be explained by theloss of a basolateral targeting signal pre-

Non-polarized targeting of AE1causes autosomal dominant distalrenal tubular acidosis

Published online 21 January 2003; doi:10.1038/ng1082

Autosomal dominant distal renal tubular acidosis (ddRTA) is caused by mutations inSLC4A1, which encodes the polytopic chloride–bicarbonate exchanger AE1 that isnormally expressed at the basolateral surface of α-intercalated cells in the distalnephron. Here we report that, in contrast with many disorders in which mutant mem-brane proteins are retained intracellularly and degraded, ddRTA can result from aber-rant targeting of AE1 to the apical surface.

Fig. 1 Steady-state localization of full-length and truncated AE1 in polarized epithelial cells by confocalmicroscopy. Filter-grown MDCK (a,b) or IMCD cells (c,d) transiently expressing wild-type AE1 (a,c) orAE1∆11 (b,d) tagged with hemagglutinin were labeled with antibodies to hemagglutinin (left columns)together with the lateral marker E-cadherin or the tight junctional marker ZO-1 (middle columns). Upperpanels, side views (xz); middle and lower panels, xy views through apical and middle sections of the cells,respectively, at the levels indicated by arrows in the upper panels. Wild-type AE1 localized basolaterally(a,c) whereas AE1∆11 was non-polarized (b,d). Scale bars = 10 µm.

AE1

xz

xy

xy

E-cadherin merge AE1∆11

xz

xy

xy

E-cadherin merge

AE1 ZO-1 merge AE1∆11 ZO-1 merge

xzxz

xy

xy

xy

xy

a b

c d

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reg

enet

ics

brief communications

126 nature genetics • volume 33 • february 2003

sent in the deleted portion of its C termi-nus. We therefore replaced the C-terminaltail of a reporter protein, the type I mem-brane protein CD8, with the 27 C-termi-nal amino acids of AE1. Both thewild-type CD8–AE1 and CD8–AE1∆11chimeras could be stably expressed, incontrast with whole AE1∆11. CD8 itselflocalizes preferentially to the apicaldomain of MDCK cells11. The completeC-terminal tail of AE1 in the wild-typeCD8–AE1 chimera redirected a consider-able fraction of CD8 to the basolateralsurface in MDCK cells (Fig. 2a), whereasthe mutant AE1 tail in CD8–AE1∆11 didnot (Fig. 2b). This supports the presenceof a sorting signal within the 11 C-termi-nal residues of AE1.

Various membrane proteins are sortedto the basolateral domain through inter-action of tyrosine-based targeting motifsin their cytoplasmic domains with adap-tor-protein complexes. The motif YDEV(904–907) in the tail of AE1 conforms

with one subset of these, YXXØ (whereØ is a hydrophobic residue). To deter-mine whether basolateral sorting of AE1is dependent on Tyr904, we expressedfull-length AE1 containing the mutationY904A. The distribution of this con-struct was similar to that of AE1∆11(Fig. 2c).

YXXØ motifs interact with µ subunitsof adaptor-protein complexes7,12, one ofwhich, AP-1B, is specific to polarizedepithelial cells13,14. Because AP-1B medi-ates sorting of certain basolateral mem-brane proteins such as the LDLreceptor14, we investigated whether sort-ing of AE1 depended on AP-1B. Wetransfected wild-type CD8–AE1 orCD8–AE1∆11 both transiently and sta-bly into the proximal renal tubular cellline LLC-PK1, which is reported to lackthe µ1B subunit and thus cannot formAP-1B complexes13,14. We confirmedthat LLC-PK1 cells did not expressdetectable µ1B (data not shown). Assess-

ing distribution of the constructs, weobserved that CD8–AE1∆11 was againpreferentially localized to the apical sur-face (Fig. 2e), whereas the wild-typechimera was predominantly localized tothe basolateral surface (Fig. 2d), suggest-ing that an adaptor other than AP-1B isinvolved in basolateral sorting of AE1.Future studies will be required toaddress this observation.

Taken together, these data indicate thata tyrosine-based motif in the C terminusof AE1, presumably of the YXXØ type,has an important role in the polarizeddistribution of AE1 at the cell surface, andthat its absence is associated with the clin-ical appearance of ddRTA. This does notexclude the possibility that additionalsequences present in the C-terminal cyto-plasmic domain or other parts of thismultiple membrane-spanning proteinmay also contribute to fidelity of thesteady-state distribution of AE1. Suchmotifs might interact with other basolat-eral proteins, the cytoskeleton or extracel-lular matrix and thereby selectivelystabilize the basolateral localization ofAE1. Candidate sequences have yet to beidentified, and potential binding partnersare poorly characterized.

Our data implicate loss of polarizedtrafficking of AE1 as the primary causeof dysfunction of α-intercalated cells inddRTA associated with AE1∆11. Despitethe importance of polarized traffickingin asymmetrically functioning cells,there are, to date, few other examples ofsimilar inherited targeting diseases inhumans. In one variant of familialhypercholesterolemia, the mutant LDLreceptor is mis-targeted to the apical cellsurface owing to loss of basolateral sort-ing information15. ddRTA is the firstexample of a primary transporter disor-der of epithelial polarity.

AcknowledgmentsWe thank N. Rungroj and Y. Su for technicalassistance, P. Luzio, M. Robinson and R. Sandfordfor invaluable discussion and J. Schwartz forIMCD cells. This work was supported by theWellcome Trust, the UK National KidneyResearch Fund, Addenbrooke’s Charities andChildren’s Kidney Care Fund and the Renal Careand Research Association. M.A.J.D. received aSackler Fellowship.

Competing interests statementThe authors declare that they have no competingfinancial interests.

Mark A.J. Devonald1,2, Annabel N.Smith1, Jenny P. Poon1, Gudrun Ihrke3* &Fiona E. Karet1,2*

*These authors contributed equally to this work.

xz

xy

CD8–AE1 E-cadherin

xz

xy

xy

CD8–AE1∆11 E-cadherin

xz

xy

xy

AE1(Y904A) ZO-1

xy

xz

xy

xy

xz

xy

xy

xz

xy

xy

xz

xy

xy

CD8–AE1 lectin merge mergeCD8–AE1∆11 lectin

Fig. 2 Steady-state localization of CD8–AE1 chimeras and AE1(Y904A) in filter-grown polarized MDCK(a–c) or LLC-PK1 cells (d–e). Panel composition and scale bars are as in Fig. 1. MDCK cells stably express-ing wild-type CD8–AE1 (a) or CD8–AE1∆11 (b) or transiently expressing AE1(Y904A) (c) were labeledwith antibodies to CD8 and E-cadherin (a,b) or hemagglutinin and ZO-1 (c). a,b, The wild-type tail ofAE1 was able to redirect CD8 to the basolateral surface whereas the truncated tail was not. c, TheY904A mutation reproduced the non-polarized distribution of AE1∆11. LLC-PK1 cells stably expressingwild-type CD8–AE1 (d) or CD8–AE1∆11 (e) were labeled with antibody to CD8 and then with secondaryantibody conjugated to Texas Red and counter-stained with peanut lectin (apical surface marker) con-jugated to fluorescein isothiocyanate. As in MDCK cells, relative distribution of the two constructs dif-fered considerably, with the wild-type chimera predominantly basolateral and the mutant chimerapredominantly apical, indicating that the tyrosine-containing C terminus was still able to direct baso-lateral sorting in the absence of AP-1B.

a b c

d e

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reg

enet

ics

brief communications

nature genetics • volume 33 • february 2003 127

1Department of Medical Genetics, 2Division ofNephrology and 3Department of ClinicalBiochemistry, University of Cambridge,Cambridge Institute for Medical Research,Addenbrooke’s Hospital Box 139, CambridgeCB2 2XY, UK. Correspondence should beaddressed to F.E.K. (e-mail:fek1000@cam.ac.uk).

Received 4 September; accepted 18 December 2002.

1. Karet, F.E. J. Am. Soc. Nephrol. 13, 2178–2184 (2002).2. Karet, F.E. et al. Proc. Natl. Acad. Sci. USA 95,

6337–6342 (1998).3. Bruce, L.J. et al. J. Clin. Invest. 100, 1693–1707

(1997).4. Jarolim, P et al. J. Biol. Chem. 273, 6380–6388

(1998).5. Quilty, J.A., Li, J. & Reithmeier, R.A. Am. J. Physiol.

Renal Physiol. 282, F810–F820 (2002).6. Toye, A.M., Bruce, L.J., Unwin, R.J., Wrong, O. &

Tanner, M.J. Blood 99, 342–347 (2002).7. Bonifacino, J.S. & Dell’Angelica, E.C. J. Cell Biol.

145, 923–926 (1999).

8. Mostov, K.E., Verges, M. & Altschuler, Y. Curr. Opin.Cell Biol. 12, 483–490 (2000).

9. Obrador, G. et al. J. Am. Soc. Nephrol. 9, 746–754(1998).

10. Low, S.H. et al. Mol. Biol. Cell 11, 3045–3060 (2000).11. Ihrke, G., Bruns, J.R., Luzio, J.P. & Weisz, O.A. EMBO

J. 20, 6256–6264 (2001).12. Ohno, H. et al. Science 269, 1872–1875 (1995).13. Ohno, H. et al. FEBS Lett. 449, 215–220 (1999).14. Fölsch, H., Ohno, H., Bonifacino, J.S. & Mellman, I.

Cell 99, 189–198 (1999).15. Koivisto, U.M., Hubbard, A.L. & Mellman, I. Cell

105, 575–585 (2001).

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reg

enet

ics