The Targeting of Cystinosin to the Lysosomal Membrane Requires aTyrosine-based Signal and a Novel Sorting Motif*

Received for publication, November 22, 2000, and in revised form, January 6, 2001Published, JBC Papers in Press, January 9, 2001, DOI 10.1074/jbc.M010562200

Stephanie Cherqui, Vasiliki Kalatzis, Germain Trugnan‡, and Corinne Antignac§

From the INSERM U423, Rein en Developpement et Nephropathies Hereditaires, Hopital Necker-Enfants Malades, 149rue de Sevres, 75015 Paris, France and ‡ INSERM U538, Trafic Membranaire et Signalisation dans les CellulesEpitheliales, Centre Hospitalier Universitaire-St Antoine, 27 rue de Chaligny, 75012 Paris, France

Cystinosis is a lysosomal transport disorder charac-terized by an accumulation of intra-lysosomal cystine.Biochemical studies showed that the lysosomal cystinetransporter was distinct from the plasma membranecystine transporters and that it exclusively transportedcystine. The gene underlying cystinosis, CTNS, encodesa predicted seven-transmembrane domain proteincalled cystinosin, which is highly glycosylated at theN-terminal end and carries a GY-XX-F (where F is ahydrophobic residue) lysosomal-targeting motif in itscarboxyl tail. We constructed cystinosin-green fluores-cent protein fusion proteins to determine the subcellu-lar localization of cystinosin in transfected cell lines andshowed that cystinosin-green fluorescent protein colo-calizes with lysosomal-associated membrane protein 2(LAMP-2) to lysosomes. Deletion of the GY-XX-F motifresulted in a partial redirection to the plasma mem-brane as well as sorting to lysosomes, demonstratingthat this motif is only partially responsible for the lyso-somal targeting of cystinosin and suggesting the exist-ence of a second sorting signal. A complete relocaliza-tion of cystinosin to the plasma membrane was obtainedafter deletion of half of the third cytoplasmic loop (ami-no acids 280–288) coupled with the deletion of the GY-DQ-L motif, demonstrating the presence of the secondsignal within this loop. Using site-directed mutagenesisstudies we identified a novel conformational lysosomal-sorting motif, the core of which was delineated to YF-PQA (amino acids 281–285).

Lysosomes are a principal site of intracellular digestion inmammalian cells. Extracellular biological materials destinedfor degradation in the lysosomes are endocytosed and shuttledto the lysosomes via early and late endosomes (1). In contrast,endogenous cellular structures which are to be degraded in thelysosome are either selectively (chaperon-mediated) or nonse-lectively (macro- or micro-) transported independently of theendosomes via a process called autophagy (2, 3). The lysosomalmembrane, due to the presence of a H1 pump, is responsible foracidification of the interior and for sequestration of the lysoso-mal enzymes responsible for the degradation process (4). Fur-thermore, the lysosomal membrane contains transport proteinsthat mediate the transport of degradation products such as

amino acids, sugars, and nucleotides from the lysosomal lumento the cytosol, where they are excreted or reused by the cell. Adefect in either the degradation or transport process can resultin an accumulation of the undegraded substrate or the degra-dation product within the lysosomes, impairing the physiologyof the cell and leading to a lysosomal storage disorder.

Several such disorders have been described and are usuallyclassified according to the molecule that has accumulated in-tra-lysosomally (5). Most of these disorders are due to a defectin one of the lysosomal hydrolases; however, a subset existsthat arises from a defect in one of the membrane transporters(6). One such example is the autosomal recessive disordercystinosis (MIM 21980), which is characterized by an accumu-lation of intra-lysosomal cystine (7). Various biochemical stud-ies over the years show that cystine transport across the lyso-somal membrane is carrier-mediated, that the carrier is locatedin the membrane itself, and that it is distinct from the plasmamembrane cystine transporters (8), which in contrast do notexclusively transport cystine (9, 10). Moreover, the lysosomalcystine transporter is stimulated by the acid pH of the lyso-some, which is ATP-dependent (11). Cystine is the disulfide ofthe amino acid cysteine, and cysteine has been found to exit thelysosome freely in cystinotic cells, implying that the cysteinetransporter is independent of the cystine transporter (12). Thisfinding has been exploited for the treatment of cystinosis as thecurrently used drug, cysteamine, is an aminothiol that reactswith cystine to form cysteine and a cystein-cysteamine mixeddisulfide that can also readily exit the cystinotic lysosome (13).Thus, cysteamine, if used early and in high doses, delays theprogression of cystinosis in affected individuals by reducingintra-lysosomal cystine levels (14).

We recently identified the gene underlying cystinosis, CTNS,(15) which is composed of 12 exons and extends over 23 kilo-bases of genomic sequence. The identification of mutations inthe CTNS-coding sequence of affected individuals, the mostcommon of which is a 57-kilobase deletion (15, 16), confirmedthis gene as underlying cystinosis. Three allelic clinical formshave been defined based on the severity of symptoms and age ofonset (11); the infantile form is the most severe and is charac-terized initially by the appearance of a proximal renal tubuledysfunction at 6–8 months, leading to end-stage renal failureby 10 years of age, and, eventually, by damage to most tissuesdue to widespread cystine accumulation. The juvenile form isless severe and characterized by a glomerular renal impair-ment appearing at a later age. Finally, the ocular nonnephro-pathic form is characterized by the presence of corneal cystine-crystal deposits without renal anomalies.

The 2.6-kilobase CTNS transcript encodes a novel 367 aminoacid protein named cystinosin, which shows homology with a55.5-kDa Caenorhabditis elegans protein (C41C4.7) and a yeasttransmembrane protein, ERS1 (15). Computer-aided sequence

* This work was supported by Vaincre les Maladies Lysosomales, theAssociation Francaise contre les Myopathies, and the Association pourl’Utilisation du Rein Artificiel. The costs of publication of this articlewere defrayed in part by the payment of page charges. This article musttherefore be hereby marked “advertisement” in accordance with 18U.S.C. Section 1734 solely to indicate this fact.

§ To whom correspondence should be addressed. Tel.: 33 1 44 49 5098; Fax: 33 1 44 49 02 90; E-mail: antignac@necker.fr.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 276, No. 16, Issue of April 20, pp. 13314–13321, 2001© 2001 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org13314

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

analysis of cystinosin predicts that it is composed of seventransmembrane domains, suggesting that it also is a mem-brane protein. These domains are preceded by seven potentialglycosylation sites at the N-terminal end, characteristic of ly-sosomal membrane proteins that are highly glycosylated tohelp protect them from the lysosomal proteases in the lumen(17). Finally, following the seventh transmembrane domain is a10-residue C-terminal tail. Within this tail lies a 5-residuesequence, GY-DQ-L. This sequence is reminiscent of the ty-rosine-based sorting motif, GY-XX-F (where F is a hydrophobicresidue), which was identified in the 10–20 amino acid car-boxyl tails of lysosomal-associated membrane protein (LAMP)1

-1, -2, and -3 and in lysosomal acid phosphatase and shown tobe responsible for their correct targeting to the lysosome (forreview see Ref. 18). Taken together these results suggest thatcystinosin is a lysosomal membrane protein. We constructedcystinosin-green fluorescent protein (GFP) fusion proteins andcystinosin-c-Myc epitope-tagged proteins to study the subcellu-lar localization of cystinosin in vitro by transient and stabletransfections of various cell lines. This study allowed the de-termination of the subcellular localization of cystinosin andresulted in the identification of a novel motif required forlysosomal sorting.

EXPERIMENTAL PROCEDURES

Protein Fusion Constructs—To create a fusion construct with theCTNS sequence at the 59 end of and in-frame with the GFP sequence,the CTNS cDNA sequence was polymerase chain reaction-amplified(Advantage cDNA polymerase mix, CLONTECH) using a forwardprimer situated in the 59-non-coding region (59-AGT CTA GCC GGGCAG GGG AA-39) and a reverse primer (59-GGG GTA CCC CGT TCAGCT GGT CAT ACC C-39) situated at the end of the coding region,which deleted the stop codon and added a KpnI restriction site. Theamplified product was digested by XhoI (situated 250 base pairs up-stream of the ATG initiator codon) and KpnI and inserted into theXhoI/KpnI-digested pEGFP-N1 (CLONTECH) (creating the constructpCTNS-EGFP). Clones containing the CTNS insert were checked bysequencing (Applied Biosystems 373XL). To generate a fusion constructcontaining the CTNS sequence at the 39 end of the GFP sequence, theCTNS-coding sequence was polymerase chain reaction-amplified usinga forward primer (59-CCG CTC GAG GGA TAA GGA ATT GGC TGACTA T-39), which added an XhoI site and deleted the ATG start codon,and a reverse primer situated in the arm of the lgt10 phage clonecontaining the CTNS cDNA. The amplified cDNA insert was digestedby XhoI and HindIII (situated 303 base pairs downstream of the stopcodon) and inserted into the XhoI/HindIII-digested pEGFP-C1 (creatingpEGFP-CTNS). As a first step toward fusion of the c-Myc tag (EQKLI-SEEDL) to the 39 end of the CTNS sequence, the CTNS cDNA-codingregion was amplified using the BamHI-containing forward primer 59-GGG GAT CCT GAG CTC TGC CTC TTC C-39 and the XbaI-containingreverse primer 59-GCT CTA GAG CCA GCC CTG TGT GCC AG-39,restriction enzyme-digested, and subcloned into the expression vectorpcDNA 3.1 Zeo1 (Invitrogen). Site-directed mutagenesis (QuikChangeSite-Directed Mutagenesis Kit, Stratagene) to introduce the c-Myc tagwas then carried out on this construct using the primer 59-TAT GACCAG CTG AAC GAA CAA AAA CTT ATT TCT GAA GAA GAT CTGTAG CAC CCA GGG ACC-39 coupled with its reverse complement(creating pCTNS-cMyc).

Mutagenesis of the pCTNS-EGFP Construct—All the modifications ofpCTNS-EGFP listed under “Results” were carried out using the Strat-agene QuikChange Site-Directed Mutagenesis Kit according to themanufacturer’s recommendations and verified by sequencing the regionsurrounding the introduced mutation. Primers used are not listed herebut are available upon request.

Cell Culture and Transfection—MDCK and HeLa cells were grown inminimal essential medium (MEM; Life Technologies, Inc.) supple-mented with 10% fetal calf serum (Vabliotek), 100 units/ml penicillin/streptomycin (Life Technologies, Inc.), and 2 mM L-glutamine (Life

Technologies, Inc.). Confluent cells were passaged the day before tran-sient transfection, and 2 3 105 cells were distributed into 35-mm wellscontaining 3 sterile glass coverslips. Once cells reached 60% confluency,the supplemented MEM media was replaced by 1 ml of OPTIMEM (LifeTechnologies, Inc.), and cells were incubated at 37 °C for 30 min. Trans-fections were carried out with 5 mg of DNA using 12.5 mg of DAC 30(Eurogentec) in 1 ml of OPTIMEM, and the cells were incubated withthe DNA-liposome complex for a minimum of 6 h. Transfected cells werethen washed with phosphate-buffered saline (PBS), overlaid with sup-plemented MEM, and left to incubate for 48 h. For the stable transfec-tions, cells were distributed into 35-mm wells without coverslips andtreated as above. After the 48-h incubation, the culture medium wassupplemented with 2 mg/ml G418 (Invitrogen).

Immunofluorescence—Transfected cells were washed with PBS andfixed with 2% paraformaldehyde (Merck) for 20 min at room tempera-ture. After washing with PBS, cells were blocked with 10 mM NH4Cl for10 min, washed, and permeabilized using 0.5% Triton X-100 (Sigma) for10 min. Coverslips were then either RNase-treated for 10 min and thecell nuclei stained with 0.5 mg/ml propidium iodide (Sigma) for 3 min orincubated with the first antibody for 1 h, washed in PBS, and incubatedwith the second antibody for 1 h. Finally, coverslips were washed inPBS, incubated with 100 mg/ml diazabicyclo[2.2.2]octane (Sigma) for 10min to maintain the fluorescence, and mounted using Glycergel(DAKO). Stained coverslips were visualized using a Zeiss Axiovert100M confocal microscope equipped with LSM 510 version 2.5 softwareand Ar/Kr (458 and 488 nm) and 23 He/Ne (543 and 633 nm) laserswith a 1003 oil immersion objective. All images correspond to a 1-mmcell section. Photographs were processed with Adobe Photoshop 5.5. Forthe transferrin internalization studies, transiently transfected HeLacells were washed three times in PBS and incubated for 30 min at 37 °Cwith 4 mg/ml Texas Red-conjugated human transferrin (MolecularProbes) in Dulbecco’s modified Eagle’s medium (Life Technologies, Inc.)supplemented with 20 mM HEPES (pH 7.4) and 2 mg/ml bovine serumalbumin. Cells were washed three times with PBS and fixed, mounted,and visualized as above.

Antibodies—The mouse monoclonal antibodies used were as follows:AC17, directed against canine LAMP-2 (19), was kindly provided by A.Le Bivic (Faculte des Sciences, Luminy, Marseille, France) and used ata dilution of 1:500; H4B4, directed against human LAMP-2, was pur-chased from the Developmental Studies Hybridoma Bank and used at adilution of 1:100; anti-GFP was purchased from Roche Molecular Bio-chemicals and used at a dilution of 1:250; 9E10, directed against thec-Myc epitope, was obtained from Santa Cruz Biotechnology and used ata dilution of 1:100; and EMA, directed against epithelial membraneantigen, was obtained from DAKO and used at a dilution of 1:50.Secondary donkey anti-mouse antibodies for the immunofluorescencestudies conjugated to either fluorescein isothiocyanate or tetramethyl-rhodamine B isothiocyanate were purchased from Jackson Laboratoriesand used at a dilution of 1:200.

Western Blot Analysis—Western blots were prepared from whole celllysates. 293 cells, non-transfected or transiently transfected with theconstruct pCTNS-EGFP, were scraped in PBS, centrifuged for 5 min at2500 rpm at 4 °C, and resuspended in lysis buffer (150 mM NaCl, 1%Nonidet P-40, 100 mM Tris HCl (pH 8), 0.02% sodium azide, 1 mM

phenylmethylsufonyl fluoride, 10 mg/ml pepstatin A, 10 mg/ml aproti-nin, and 10 mg/ml leupeptine). Cells were lysed for 20 min on ice andcentrifuged for 2 min at 12,000 rpm at 4 °C. Cell lysates were mixedwith Laemmli’s sample buffer and loaded directly without boiling ontoan 8% SDS-polyacrylamide electrophoresis gel. The separated proteinswere electro-transferred to nitrocellulose filters (Bio-Rad) and, afterblocking for 1 h in 0.05% Tween/PBS 1 1% polyvinylpyrolidane (Sig-ma), the membrane was incubated for 1 h at room temperature with a1/1000 dilution of the anti-GFP antibody. After three washes in 0.05%Tween/PBS, the filter was incubated for 1 h at room temperature with1/12000 horseradish peroxidase-conjugated sheep anti-mouse antibody(Amersham Pharmacia Biotech). The final detection was performedusing the Amersham Pharmacia Biotech ECL reagents according to themanufacturer’s recommendations.

RESULTS

Subcellular Localization of Wild-type Cystinosin—TheCTNS-coding region was cloned 59 to the GFP cDNA in theplasmid pEGFP-N1, generating pCTNS-EGFP. In this orienta-tion, the GFP end of the fusion protein was extra-lysosomal(Fig. 1). pCTNS-EGFP was transiently transfected into MDCKand HeLa cells, and the GFP fluorescence signal was observed48 h post-transfection. In both cell types, cystinosin-GFP was

1 The abbreviations used are: LAMP, lysosomal-associated mem-brane protein; MDCK, Madin-Darby canine kidney; MEM, minimalessential medium; PBS, phosphate-buffered saline; GFP, green fluores-cent protein; EMA, epithelial membrane antigen.

Lysosomal-targeting Signals of Cystinosin 13315

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

localized to small, discrete intracellular vesicles and occasion-ally to large vesicular structures (Fig. 2A). No signal could bedetected at the plasma membrane or in the nucleus. The sameresults were obtained upon stable transfection of MDCK cells(data not shown). Immunofluorescence studies using the anti-GFP antibody confirmed that the fluorescent signal did indeedcorrespond to cystinosin-GFP (data not shown). Moreover,western blot analysis was used to confirm the production of thecystinosin-GFP fusion protein. A major band with an apparentsize of ;80 kDa (Fig. 3) was detected that was higher than theexpected size of 67 kDa (cystinosin was predicted to be 40 kDa)and could be explained by the glycosylation of cystinosin via the7 putative N-terminal glycosylation sites; a smaller band withan apparent size of ;70 kDa could also be seen that couldpresumably correspond to the unglycosylated form of the pro-tein. Alternatively, this band could arise from proteolytic cleav-age of cystinosin, although no known cleavage sites were de-tected. These same bands were not detected in nontransfected293 cells (Fig. 3).

To verify that the fusion of the GFP to the C terminus ofcystinosin did not interfere with its native localization, twoadditional constructs were made. First, the CTNS-coding re-gion was subcloned downstream of the GFP gene in the plasmidpEGFP-C1 (pEGFP-CTNS) such that the GFP of the fusionprotein would be predicted to lie intra-lysosomal. However, nofluorescence could be observed upon transfection of this con-struct presumably due to its quenching by the acidic lysosomalpH (20). Second, by site-directed mutagenesis, a c-Myc tag wasadded to the 39 end of CTNS in the plasmid pcDNA 3.1 Zeo1(pCTNS-cMyc), and immunofluorescence analysis with the an-ti-epitope c-Myc antibody, 9E10, showed the same labeling ofintracellular vesicles (Fig. 2B) as the construct pCTNS-EGFP.

To determine whether the labeled vesicles were indeed lyso-somes, transfected cells were immunofluorescently labeled us-ing antibodies directed against the late endosomal and lysoso-mal membrane protein LAMP-2. As seen in Fig. 4, in bothMDCK (Fig. 4, A–C) and HeLa (Fig. 4, D–F) cells, the cystino-sin-GFP signal colocalized with that of AC17, directed againstcanine LAMP-2, and H4B4, directed against human LAMP-2,respectively. The exact overlap between the LAMP-2 and cysti-

nosin localization patterns demonstrates that cystinosin is alysosomal protein. Interestingly, the above-mentioned largevesicular structures were also labeled by antibodies directed toLAMP-2 in cystinosin-overexpressing cells (Fig. 4, A and B),indicating that these large vesicles are also lysosomal struc-tures; immunostaining of nontransfected cells with the sameantibodies did not reveal the presence of these structures (datanot shown). Finally, colocalization studies using fluorescentlylabeled transferrin uptake ruled out the possibility that cysti-nosin also localizes to the early endosomes (Fig. 4, G–I).

Deletion and Mutagenesis of the Lysosomal-targeting Sig-nal—Previous studies on other lysosomal membrane glycopro-teins have delineated one of the lysosomal-targeting signals asGY-XX-F (21). Such a five-residue targeting sequence, GY-DQ-L, exists in the C-terminal cytoplasmic tail of cystinosin(amino acids 362–366), and it is followed by an N residue. Toconfirm that the lysosomal targeting of cystinosin is indeedencoded by this signal, we deleted this motif (DGY-DQ-L) fromthe pCTNS-EGFP construct. Although transfection of this con-struct still resulted in the sorting of cystinosin-GFP to thelysosomes, both to small discrete vesicles as well as to abun-dant large vesicular structures, as shown by colocalizationstudies using AC-17 (Fig. 5, A–C), a significant signal was also

FIG. 1. Predicted structure of cystinosin and representation ofthe deletions/mutations concerning the third cytoplasmic loop.A, schematic representation of the predicted cystinosin structure. Blueindicates the N-terminal uncleavable signal peptide, orange indicatesthe seven potential N-glycosylation sites, black indicates the thirdcytoplasmic loop, and red indicates the C-terminal GY-DQ-L lysosomal-targeting signal. The yellow and green boxes indicate the positions of thec-Myc and GFP tags, respectively. B, the amino acid sequence of thethird cytoplasmic loop is indicated. The two large deletions (DA and DB)and the smaller deletions (Da and Db) are represented by dashes. Theamino acid residues mutated to alanine residues (ma and mb) areunderlined.

FIG. 2. Subcellular localization of wild-type cystinosin. A, 48 hafter transient transfection of MDCK cells with the pCTNS-EGFPconstruct, cells were fixed, permeabilized, and incubated with pro-pidium iodide. The fluorescent signal from the cystinosin-GFP fusionprotein can be seen to be localized to small discrete intracellular vesi-cles and to large vesicular structures. B, MDCK cells transiently trans-fected with pCTNS-cMyc were treated as above and then incubatedwith mouse monoclonal antibody directed against the c-Myc epitopefollowed by a fluorescein isothiocyanate-conjugated donkey anti-mousesecondary antibody. The same fluorescence labeling of intracellularvesicles can be seen. Scale bar, 10 mm for all figures.

FIG. 3. Western blot analysis. Non-transfected 293 cells (lane 1)and 293 cells transiently transfected with the pCTNS-EGFP construct(lane 2) were lysed, and the cleared lysate was loaded on a 8% SDS-polyacrylamide electrophoresis gel. After electro-transfer, the filter wasincubated with a mouse monoclonal antibody directed against the GFPtag and subsequently with a horseradish peroxidase-conjugated sheepanti-mouse antibody. After immunodetection of the filter and a 10-sexposure to autoradiographic film, two bands could be observed in lane2, a major band with an apparent size of ;80 kDa (large arrow),probably corresponding to the glycosylated form of cystinosin, and asmaller band with an apparent size of ;70 kDa (small arrow). Neitherof these bands is detected in lane 1.

Lysosomal-targeting Signals of Cystinosin13316

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

found at the plasma membrane that colocalized with that of theepithelial membrane antigen antibody, EMA (Fig. 5, D–F).These results confirm that the GY-DQ-L motif is involved intargeting cystinosin to the lysosomes.

By site-directed mutagenesis, each codon of the lysosomal-targeting sequence was individually mutated to an A residue todetermine the critical residues of this sequence. Substitution ofthe Gly, Asp, or Gln residues for an Ala residue did not alter thelysosomal localization of cystinosin (Fig. 6, A, C, and D). Incontrast, mutation of the Tyr or Leu residues resulted in thesame localization profile as that obtained when the entire ly-sosomal-targeting sequence was deleted (Fig. 6, B and E). TheAsn residue after the lysosomal-targeting motif was also mu-tated to Ala to ensure that this amino acid was not part of thetargeting sequence. This substitution resulted in the samelocalization pattern as that seen with the wild type (Fig. 6F).Taken together these results demonstrate that it is the Tyr andLeu residues of the tyrosine-based sorting motif that are criti-cal for the targeting of cystinosin to the lysosomes.

Existence of a Second Lysosomal-targeting Signal—Our mu-tation analysis of the GY-DQ-L motif demonstrated that it isonly partially responsible for the targeting of cystinosin. Thesedata strongly suggested the presence of a second lysosomal-targeting signal. In the first instance, to determine whether theremainder of the cytoplasmic tail, which is not included in thedefined targeting signal, may also play a role in lysosomaltargeting, the entire cytoplasmic tail was deleted (amino acids358–367) from the pCTNS-EGFP construct. This resulted inthe same subcellular localization of cystinosin-GFP as obtainedafter deletion of the lysosomal-targeting signal (data notshown).

A sequence analysis of the other three cytoplasmic domains

was then performed to identify a possible motif resembling oneof the documented lysosomal-targeting signals. The third cyto-plasmic loop, located between the fifth and sixth potentialtransmembrane domains, is tyrosine-rich and, hence, seemedto be a good candidate for containing a lysosomal-targetingsignal. In the first instance, we deleted the first nine aminoacids (amino acids 280–288) of this 19-residue cytoplasmicdomain (DA) from the construct pCTNS-EGFP and observedthe same cystinosin-GFP localization pattern as when the GY-DQ-L motif was deleted alone, with the exception that no largelabeled vesicles were seen (data not shown). Subsequently, theremaining 10 amino acids (amino acids 289–298) were deleted(DB) from pCTNS-EGFP, resulting in a localization of cystino-sin-GFP to the lysosomes but, again, not to the large lysosomalstructures without an obvious signal at the plasma membrane(data not shown). Each of these deletions, DA and DB, was thenincluded with DGY-DQ-L in pCTNS-EGFP. Transfection of theDA/DGY-DQ-L construct resulted in the exclusive localizationof cystinosin-GFP to the plasma membrane (Fig. 7A). In con-trast, transfection of the DB/DGY-DQ-L construct gave a simi-lar pattern (Fig. 7B) to that obtained using DGY-DQ-L alone,with the exception of the large labeled vesicles. In MDCK andHeLa cells transfected with the construct containing DA/DGY-DQ-L, there was no longer a colocalization between cystinosin-GFP and LAMP-2 (Fig. 8, A–C), whereas there was a colocal-ization between cystinosin-GFP and epithelial membraneantigen (Fig. 8, D–F). In contrast, in cells transfected with theconstruct containing DB/DGY-DQ-L, the cystinosin-GFP signalcolocalized with both that of AC17 and epithelial membraneantigen (EMA), demonstrating that this double deletion stilldirected the fusion protein to the lysosomes (data not shown).Taken together, these results suggest that the third cytoplas-

FIG. 4. Colocalization studies of wild-type cystinosin. A and B, MDCK cells transiently transfected with the pCTNS-EGFP construct(visualized in green) were fixed, permeabilized, and incubated with the mouse monoclonal antibody directed against canine LAMP-2 (AC17)followed by a tetramethylrhodamine B isothiocyanate-conjugated donkey anti-mouse secondary antibody (visualized in red). The large labeledvesicles and their lumen are visible (arrowhead). C, a superimposition of A and B shows a colocalization between cystinosin and LAMP-2 in caninecells. D and E, HeLa cells transiently transfected with the pCTNS-EGFP construct (visualized in green) were incubated with the mouse monoclonalantibody directed against human LAMP-2 (H4B4) followed by a tetramethylrhodamine B isothiocyanate-conjugated donkey anti-mouse secondaryantibody (visualized in red). F, a superimposition of D and E shows that cystinosin also localizes to lysosomes in human cells. G and H, HeLa cellstransiently transfected with the pCTNS-EGFP construct (visualized in green) were incubated 30 min at 37 °C with Texas Red-conjugated humantransferrin (visualized in red) and then fixed. I, a superimposition of G and H shows that cystinosin is not localized to early endosomes.

Lysosomal-targeting Signals of Cystinosin 13317

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

mic loop, in particular the first half, plays a role in the target-ing of cystinosin to the lysosome.

As seen here and reported by others (22, 23) that a Tyr playsa critical role in lysosomal targeting, we individually replacedthe two Tyr residues in the first half of the third cytoplasmicdomain by an Ala residue using site-directed mutagenesis ofpCTNS-EGFP. Each of these mutations coupled with DGY-

DQ-L resulted in the same localization pattern (data notshown) as that observed with DGY-DQ-L alone (Fig. 2A), indi-cating that this loop does not play a role in lysosomal targetingvia a tyrosine-based motif. Subsequently, blocks of four andfive amino acids beginning at each of the aforementioned tyro-sine residues were either deleted (designated Da and Db, re-spectively, see Fig. 1) or mutated (ma and mb, respectively) and

FIG. 5. Subcellular localization of cystinosin following deletion of the C-terminal GY-DQ-L signal. A, when MDCK cells weretransiently transfected with the DGY-DQ-L construct, a signal from the cystinosin-GFP fusion protein could be seen at the plasma membrane aswell as in small and abundant large intracellular vesicles (visualized in green). B, after incubation of these cells with AC17, a signal could be seenonly in intracellular vesicles (visualized in red). C, a superimposition of A and B identifies the intracellular vesicles seen in A as the lysosomes.D, upon transient transfection of HeLa cells with the DGY-DQ-L construct, the same fluorescent pattern as that in panel A could be seen (visualizedin green). E, after incubation of these cells with a mouse monoclonal antibody directed against EMA followed by a tetramethylrhodamine Bisothiocyanate-conjugated donkey anti-mouse secondary antibody, a labeling of the plasma membrane can be seen (visualized in red). F, asuperimposition of D and E confirms that the cystinosin-GFP fusion protein, upon deletion of the GY-DQ-L signal, is directed to the plasmamembrane as well as to lysosomes.

FIG. 6. Subcellular localization of cystinosin after site-directed mutagenesis of the GY-DQ-L-targeting signal. By site-directedmutagenesis, each residue of the GY-DQ-L-sorting motif in the construct pCTNS-EGFP was sequentially changed to an alanine residue. MDCKcells transiently transfected with the construct containing a mutation of the Gly (A), Asp (C), or Gln (D) residue results in a GFP fluorescencepattern identical to that of the wild-type cystinosin-GFP fusion protein. Transient transfection with the construct containing a mutation of the Tyr(B) or Leu (E) residue results in a fluorescent signal in intracellular vesicles but also in a redirection of the mutated cystinosin-GFP fusion proteinto the plasma membrane. Mutation of the Asn residue (F) following the targeting motif resulted in the same subcellular localization pattern as thewild-type cystinosin-GFP protein.

Lysosomal-targeting Signals of Cystinosin13318

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

assayed with DGY-DQ-L. Transfection of the Da/DGY-DQ-Lconstruct resulted in the same plasma membrane localizationof cystinosin-GFP as seen with the DA/DGY-DQ-L construct(Fig. 9A), whereas transfection of the ma/DGY-DQ-L constructresulted in the appearance of a distinct labeling of the lyso-somes with no large lysosomal structures seen (Fig. 9B). Incontrast, transfection of both the Db/DGY-DQ-L and mb/DGY-DQ-L constructs resulted in the same localization pattern asobtained with ma/DGY-DQ-L (data not shown). These resultsindicate that the beginning of the third cytoplasmic loop, fromthe Tyr to the Ala residue, is crucial for the targeting of cysti-nosin to lysosomes and that the rest of the loop also plays aminor role.

DISCUSSION

Cystinosin is a novel integral membrane protein that, whendefective, is responsible for the lysosomal transport disordercystinosis, characterized by an accumulation of intralysosomalcystine (15). The disease phenotype led to the prediction thatcystinosin is a cystine transporter localized in the lysosomalmembrane (8). The construction of a cystinosin-GFP fusionprotein has now allowed us to confirm the predicted subcellularlocalization due to the colocalization of this fusion protein withan antibody directed against the lysosomal membrane protein,LAMP-2. The targeting of lysosomal membrane proteins hasbeen largely studied, and two main classes of signals have sofar been identified, a tyrosine-based GY-XX-F or Y-XX-F(where F is a hydrophobic residue), and a di-leucine (or isole-ucine)-sorting motif, both typically found in the short C-termi-nal cystoplasmic tails of glycoproteins that cross the membraneone or more times (18). Sequence analysis of the 10-amino acidcytoplasmic tail of cystinosin permitted the identification of atyrosine-based targeting signal, GY-DQ-L, which when de-leted, results in a redirection of cystinosin to the plasma mem-brane. By site-directed mutagenesis we have determined thatthe critical elements for this redirection are the Tyr and Leuresidues, consistent with the results obtained for LAMP-1 (22,23). Indeed, lysosomal membrane proteins can be delivered tothe lysosome directly, from the trans-Golgi network or indi-rectly, by first appearing at the cell surface and subsequentlybeing internalized via endocytosis, and the Tyr and F residueshave been shown to be involved in both of these pathways (21).In contrast, mutation of the Gly residue did not alter thelysosomal localization of cystinosin, again consistent with thatseen for LAMP-1, although it has been shown that this residuedoes have a role in the direct targeting of this latter proteinfrom the trans-Golgi network to the lysosome (21).

Interestingly, mutagenesis of the Tyr or Leu residues of

GY-DQ-L or deletion of this entire pentapeptide still results ina localization of cystinosin-GFP to the lysosomes. This is incontrast to the situation for LAMP-1, where it has been shownthat mutagenesis of the Tyr or F residues from its GY-QT-Isignal results in the exclusive detection of LAMP-1 on the cellsurface and not in intracellular vesicles (21). These resultsimply that there may be a second sorting signal at play forcystinosin. Such a situation has been described for tyrosinase,which contains both a di-leucine motif as well as a tyrosine-based motif in its cytoplasmic tail, and mutagenesis studiesshow that it is the di-leucine motif that is necessary for theefficient sorting of this protein to late endosomes and lyso-somes, whereas the Y-XX-F motif seems to be part of a weaksecondary sorting signal (24). Along this line, we deleted theentire cytoplasmic tail from the cystinosin-GFP fusion proteinto determine whether, as is the situation for tyrosinase, asecond targeting signal resided in the tail even though a di-leucine motif was not detected. However, transfection of thisconstruct resulted in the same localization pattern as obtainedwhen just the GY-DQ-L signal was deleted. Because a secondsignal was not detected in the cytoplasmic tail of cystinosin, wefocused our attention on the three predicted cytoplasmic loops,which would be the most susceptible to an interaction withadaptor proteins involved in the sorting of proteins to theirfated location. None of these loops contain a di-leucine motif,but the third cytoplasmic loop, composed of 19 amino acids, wastyrosine-rich. Deletion of the first nine amino acids of this loop(amino acids 280–288) coupled with the deletion of the definedcarboxyl GY-DQ-L motif, resulted in the complete relocaliza-tion of cystinosin-GFP to the plasma membrane without asignal remaining in the lysosomes. This redirection was due toa novel sorting motif, the core of which was defined as YFPQA(amino acids 281–285) by mutagenesis studies.

This novel sorting motif does not resemble any of the lyso-somal sorting motifs so far defined. In addition, serial deletionsof the remainder of the third cytoplasmic loop, although they donot provoke as drastic an effect, do seem to indicate that thisregion also has a minor role in lysosomal targeting. This isreminiscent of the situation involving the epidermal growthfactor receptor. Epidermal growth factor receptor is a trans-membrane domain protein with three lysosomal sorting signalsin its carboxyl tail: a tyrosine-based motif (YLVI), a tyrosinekinase domain, and an uncharacterized region situated be-tween amino acids 1022 and 1123 (25–27). Within this third1022–1123 amino acid-targeting domain of epidermal growthfactor receptor, it has been shown that the region 1022–1063appears to contribute more significantly to sorting than resi-dues 1063–1123 (27). Moreover, it is noteworthy, that mutatingthe core of the second sorting motif in cystinosin to alanineresidues does not have as a dramatic effect on the relocalizationof cystinosin as compared with the complete deletion of thispentapeptide. Thus, the YFPQA region of cystinosin seems toact as a conformational motif that is recognized as part of aparticular secondary structure formed by the rest of the thirdcytoplasmic loop.

Other proteins that have a nonclassical lysosomal-sortingsignal have been described. For example, P-selectin does notcontain a classic lysosomal-targeting signal but rather a novelsequence, KCPL, in its C-terminal tail that has been shown tobe responsible for its lysosomal targeting (28). The one strikingdifference between cystinosin and other proteins targeted tothe lysosome is that, for the latter, the lysosomal-targetingsignal(s) has been found to be situated in the carboxyl tail,whereas the second putative lysosomal-targeting signal forcystinosin is localized in one of its cytoplasmic loops. Thisseems to be a unique situation, although a recent study has

FIG. 7. Subcellular localization of cystinosin after deletion ofthe third cytoplasmic loop and the GY-DQ-L motif. A, in MDCKcells transiently transfected with the pCTNS-EGFP construct carryingthe deletion DA and DGY-DQ-L, the deleted cystinosin-GFP is localizedentirely to the plasma membrane. B, after transient transfection withthe pCTNS-EGFP construct carrying the deletion DB and DGY-DQ-L,the GFP signal is divided between the plasma membrane and smallintracellular vesicles.

Lysosomal-targeting Signals of Cystinosin 13319

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

shown that the late endosomal/lysosomal targeting of MLN64,a four transmembrane domain cholesterol-binding protein, ismediated via a region situated between amino acids 47 and171, which comprises the transmembrane domains (29). How-ever, the sorting motif and its exact localization within thisregion have not yet been defined.

The targeting of proteins carrying a Y-XX-F to the lysosomehas been extensively studied and found to require a directinteraction with the adaptor proteins AP-1, AP-2, and AP-3(30). AP-1 is responsible for the delivery of proteins from thetrans-Golgi network to the endosomal-lysosomal system (31),AP-2 mediates the internalization at the plasma membraneand subsequent delivery to the lysosome (32), and AP-3 isinvolved in an alternative pathway of transport from the trans-Golgi network or endosomes to the lysosome (33). Because wehave shown that the GY-XX-F motif is active in cystinosin, oneof the pathways by which this protein is sorted to the lysosomemust involve one or more of these three adaptor proteins. Onthe other hand, we do not know which pathway the secondlysosomal sorting motif uses or with which adaptor proteins it

interacts to bring cystinosin to the lysosome. It cannot beignored, however, that when the third cytoplasmic loop is al-tered we no longer see the labeling of any large lysosomalstructures but just small discrete lysosomes. Thus, it is tempt-ing to speculate that this second sorting motif may act to bringcystinosin to the lysosome via the intermediate of these largestructures. Additional support for this hypothesis arises fromthe fact that in cells transfected with DGY-DQ-L, these largevesicles seem to become more abundant than in cells trans-fected with wild-type cystinosin-GFP. These results could becorrelated with the fact that removal of the tyrosine-basedsorting motif would require an increase in activity of the secondsorting motif and, thereby, in the number of large vesicles.Hence, the pathway used by this second and novel sorting motifmay represent a previously uncharacterized pathway for thesorting of lyosomal proteins to their final destination.

Cystinosin has been shown to be homologous to a 55.5-kDaprotein of C. elegans (C41C4.7) as well as to a yeast transmem-brane protein ERS1 (15). Surprisingly the C-terminal lysoso-mal-targeting signal of cystinosin, GY-DQ-L, is not found ineither of these two other organisms, whereas the YFPQA motifis conserved in both C. elegans (YFPQV) and yeast (YIPQV),suggesting an important role for this motif. Moreover, therehas been a splice site mutation (IVS1112T.C) reported asassociated with the less severe juvenile cystinosis, which hasbeen predicted to lead to a truncated protein missing exon 11.This splice site mutation alters the reading frame, producing amissense mutation after amino acid residue 284 and introduc-ing a new termination codon at position 289 (34). Although theencoded product lacks the GY-DQ-L motif, a CTNS-GFP fusionconstruct carrying the IVS1112T-.C mutation transfectedinto CTNS2/2 cells results in a partial rescue of the cellularphenotype, implying the truncated cystinosin protein is stilltargeted to lysosomes (34). These observations are consistentwith our results demonstrating the existence of a second lyso-somal sorting signal for cystinosin. Infantile cystinosis is amultisystemic disease that can lead to death at 10 years of agedue to the widespread accumulation of intralysosomal cystine,

FIG. 8. Colocalization studies of cystinosin carrying the deletion of the first nine amino acids of the third cytoplasmic loop andof the GY-DQ-L motif. A and B, MDCK cells transiently transfected with the pCTNS-EGFP construct carrying the deletions DA and DGY-DQ-L(visualized in green) were fixed, permeabilized, and incubated with AC17 (visualized in red). C, a superimposition of A and B shows that the deletedcystinosin-GFP fusion protein and LAMP-2 do not colocalize in canine cells. D and E, HeLa cells transiently transfected with the same construct(visualized in green) were incubated with EMA (visualized in red). F, a superimposition of D and E shows that the deleted cystinosin-GFP fusionprotein colocalizes with the epithelial membrane antigen, demonstrating that this double deletion entirely directs the fusion protein to the plasmamembrane.

FIG. 9. Subcellular localization of cystinosin after deletionand mutagenesis of the YFPQA motif coupled with the deletionof the GY-DQ-L motif. A, in MDCK cells transiently transfected withthe pCTNS-EGFP construct carrying the deletion Da and DGY-DQ-L,the deleted cystinosin-GFP is localized entirely to the plasma mem-brane. B, after transient transfection with the pCTNS-EGFP constructcarrying the mutation ma and DGY-DQ-L, the GFP signal is dividedbetween the plasma membrane and small intracellular vesicles.

Lysosomal-targeting Signals of Cystinosin13320

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

highlighting the crucial role of cystinosin in the normal meta-bolic functions of the body. Due to this indispensable role ofcystinosin in the lysosomes of all cells, it is thus not surprisingto find that this protein has a rescue pathway by which it canbe sorted to the lysosome. In conclusion, the identification ofthis novel lysosomal-targeting motif not only provides an in-sight into the way cystinosin functions but also represents anaddition to the ever-growing list of targeting motifs required forthe correct sorting of proteins to their fated cellular location.

Acknowledgments—We are extremely grateful to Jean-Louis Delau-nay for invaluable help and advice. We thank Tounsia Aıt Slimane forhelp with the initial immunofluorescence experiments, Andre LeBivicfor the kind and indispensable gift of the AC-17 antibody, Fabien Alpyfor providing the Western blot protocol, Yann Goureau for technicalassistance with the confocal microscope, Yves Deris for photographicassistance, and George Patterson for help with computer programming.Finally, we thank Patrice Codogno and Fred Dice for helpfuldiscussions.

REFERENCES

1. Schekman, R., and Orci, L. (1996) Science 271, 1526–15332. Dice, J. F. (1990) Trends Biochem. Sci. 15, 305–3093. Seglen, P. O., and Bohley, P. (1992) Experientia (Basel) 48, 158–1724. Mellman, I., Fuchs, R., and Helenius, A. (1986) Annu. Rev. Biochem. 55,

663–7005. Neufeld, E. F. (1991) Annu. Rev. Biochem. 60, 257–2806. Mancini, G. M., Havelaar, A. C., and Verheijen, F. W. (2000) J. Inherited

Metab. Dis. 23, 278–2927. Gahl, W. A., Bashan, N., Tietze, F., Bernardini, I., and Schulman, J. D. (1982)

Science 217, 1263–12658. Gahl, W. A., Tietze, F., Bashan, N., Bernardini, I., Raiford, D., and Schulman,

J. D. (1983) Biochem. J. 216, 393–4009. Bannai, S., and Kitamura, E. (1980) J. Biol. Chem. 255, 2372–2376

10. Tate, S. S., Yan, N., and Udenfriend, S. (1992) Proc. Natl. Acad. Sci. U. S. A.89, 1–5

11. Gahl, W. A., Schneider, J. A., and Aula, P. P. (1995) in The Metabolic andMolecular Basis of Inherited Disease (Scriver, C. R., Beaudet, A. L., Sly,

W. S., and Valle, D., eds) pp. 3763–3797, McGraw-Hill Inc., New York12. Pisoni, R. L., Acker, T. L., Lisowski, K. M., Lemons, R. M., and Thoene, J. G.

(1990) J. Cell Biol. 110, 327–33513. Thoene, J. G., Oshima, R. G., Crawhall, J. C., Olson, D. L., and Schneider, J. A.

(1976) J. Clin. Invest. 58, 180–18914. Markello, T. C., Bernardini, I. M., and Gahl, W. A. (1993) N. Engl. J. Med. 328,

1157–116215. Town, M., Jean, G., Cherqui, S., Attard, M., Forestier, L., Whitmore, S. A.,

Callen, D. F., Gribouval, O., Broyer, M., Bates, G. P., van’t Hoff, W., andAntignac, C. (1998) Nat. Genet. 18, 319–324

16. Touchman, J. W., Anikster, Y., Dietrich, N. L., Maduro, V. V., McDowell, G.,Shotelersuk, V., Bouffard, G. G., Beckstrom-Sternberg, S. M., Gahl, W. A.,and Green, E. D. (2000) Genome Res. 10, 165–173

17. Fukuda, M. (1991) J. Biol. Chem. 266, 21327–2133018. Hunziker, W., and Geuze, H. J. (1996) Bioessays 18, 379–38919. Nabi, I. R., Le Bivic, A., Fambrough, D., and Rodriguez-Boulan, E. (1991)

J. Cell Biol. 115, 1573–158420. Patterson, G. H., Knobel, S. M., Sharif, W. D., Kain, S. R., and Piston, D. W.

(1997) Biophys. J. 73, 2782–279021. Honing, S., and Hunziker, W. (1995) J. Cell Biol. 128, 321–33222. Williams, M. A., and Fukuda, M. (1990) J. Cell Biol. 111, 955–96623. Guarnieri, F. G., Arterburn, L. M., Penno, M. B., Cha, Y., and August, J. T.

(1993) J. Biol. Chem. 268, 1941–194624. Calvo, P. A., Frank, D. W., Bieler, B. M., Berson, J. F., and Marks, M. S. (1999)

J. Biol. Chem. 274, 12780–1278925. Felder, S., Miller, K., Moehren, G., Ullrich, A., Schlessinger, J., and Hopkins,

C. R. (1990) Cell 61, 623–63426. Opresko, L. K., Chang, C. P., Will, B. H., Burke, P. M., Gill, G. N., and Wiley,

H. S. (1995) J. Biol. Chem. 270, 4325–433327. Kornilova, E., Sorkina, T., Beguinot, L., and Sorkin, A. (1996) J. Biol. Chem.

271, 30340–3034628. Blagoveshchenskaya, A. D., Norcott, J. P., and Cutler, D. F. (1998) J. Biol.

Chem. 273, 2729–273729. Alpy, F., Stoeckel, M. E., Dierich, A., Escola, J. M., Wendling, C., Chenard,

M. P., Vanier, M. T., Gruenberg, J., Tomasetto, C., and Rio, M. C. (2001)J. Biol. Chem. 276, 4261–4269

30. Bonifacino, J. S., and Dell’Angelica, E. C. (1999) J. Cell Biol. 145, 923–92631. Pearse, B. M., and Robinson, M. S. (1990) Annu. Rev. Cell Biol. 6, 151–17132. Robinson, M. S. (1994) Curr. Opin. Cell Biol. 6, 538–54433. Odorizzi, G., Cowles, C. R., and Emr, S. D. (1998) Trends Cell Biol. 8, 282–28834. Thoene, J., Lemons, R., Anikster, Y., Mullet, J., Paelicke, K., Lucero, C., Gahl,

W., Schneider, J., Shu, S. G., and Campbell, H. T. (1999) Mol. Genet. Metab.67, 283–293

Lysosomal-targeting Signals of Cystinosin 13321

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Stéphanie Cherqui, Vasiliki Kalatzis, Germain Trugnan and Corinne AntignacSignal and a Novel Sorting Motif

The Targeting of Cystinosin to the Lysosomal Membrane Requires a Tyrosine-based

doi: 10.1074/jbc.M010562200 originally published online January 9, 20012001, 276:13314-13321.J. Biol. Chem. 

  10.1074/jbc.M010562200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/276/16/13314.full.html#ref-list-1

This article cites 33 references, 18 of which can be accessed free at

by guest on April 10, 2018

http://ww

w.jbc.org/

Dow

nloaded from