Portal de Periódicos da CAPES

The Retinal Rod Na+/Ca2+,K+Exchanger Contains a Noncleaved Signal Sequence Required for Translocation of the N Terminus

1999; Elsevier BV; Volume: 274; Issue: 53 Linguagem: Inglês

10.1074/jbc.274.53.38177

1083-351X

Colleen J. McKiernan, Martin Friedlander,

Connexins and lens biology

The retinal rod Na+/Ca2+,K+ exchanger (RodX) is a polytopic membrane protein found in photoreceptor outer segments where it is the principal extruder of Ca2+ ions during light adaptation. We have examined the role of the N-terminal 65 amino acids in targeting, translocation, and integration of the RodX using anin vitro translation/translocation system. cDNAs encoding human RodX and bovine RodX through the first transmembrane domain were correctly targeted and integrated into microsomal membranes; deletion of the N-terminal 65 amino acids (aa) resulted in a translation product that was not targeted or integrated. Deletion of the first 65 aa had no effect on membrane targeting of full-length RodX, but the N-terminal hydrophilic domain no longer translocated. Chimeric constructs encoding the first 65 aa of bovine RodX fused to globin were translocated across microsomal membranes, demonstrating that the sequence could function heterologously. Studies of fresh bovine retinal extracts demonstrated that the first 65 aa are present in the native protein. These data demonstrate that the first 65 aa of RodX constitute an uncleaved signal sequence required for the efficient membrane targeting and proper membrane integration of RodX. The retinal rod Na+/Ca2+,K+ exchanger (RodX) is a polytopic membrane protein found in photoreceptor outer segments where it is the principal extruder of Ca2+ ions during light adaptation. We have examined the role of the N-terminal 65 amino acids in targeting, translocation, and integration of the RodX using anin vitro translation/translocation system. cDNAs encoding human RodX and bovine RodX through the first transmembrane domain were correctly targeted and integrated into microsomal membranes; deletion of the N-terminal 65 amino acids (aa) resulted in a translation product that was not targeted or integrated. Deletion of the first 65 aa had no effect on membrane targeting of full-length RodX, but the N-terminal hydrophilic domain no longer translocated. Chimeric constructs encoding the first 65 aa of bovine RodX fused to globin were translocated across microsomal membranes, demonstrating that the sequence could function heterologously. Studies of fresh bovine retinal extracts demonstrated that the first 65 aa are present in the native protein. These data demonstrate that the first 65 aa of RodX constitute an uncleaved signal sequence required for the efficient membrane targeting and proper membrane integration of RodX. bovine human amino acid(s) polymerase chain reaction nucleotide(s) endoglycosidase H polyacrylamide gel electrophoresis Tris-buffered saline with Tween The retinal rod Na+/Ca2+,K+exchanger (RodX) is a member of a large family of cation exchangers that are expressed in a variety of tissues including heart, brain, kidney, lung, large intestine, pancreas, and spleen (1Philipson K.D. Nicoll D.A. Curr. Opin. Cell Biol. 1992; 4: 678-683Crossref PubMed Scopus (45) Google Scholar). Hydropathy analysis of these proteins predicts 10 to 12 membrane-spanning α-helices organized into two clusters with a large intracellular loop present between the sixth and seventh hydrophobic domains (see Fig. 1 A). RodX has been cloned and sequenced from a number of species including cow (2Reilaender H. Achilles A. Friedel U. Maul G. Lottspeich F. Cook N.J. EMBO J. 1992; 11: 1689-1695Crossref PubMed Scopus (159) Google Scholar), human, (3Tucker J.E. Winkfein R.J. Cooper C.B. Schnetkamp P.M.M. Investig. Ophthamol. Visual Sci. 1998; 39: 435-440PubMed Google Scholar) and dolphin (3Tucker J.E. Winkfein R.J. Cooper C.B. Schnetkamp P.M.M. Investig. Ophthamol. Visual Sci. 1998; 39: 435-440PubMed Google Scholar). These retina-specific exchangers share extensive sequence homology in the hydrophobic domains and are less homologous in the two large hydrophilic domains. The large, N-terminal hydrophilic domain is extracellular and immediately follows the N-terminal signal sequence, distinguishing this class of cationic exchangers from others that do not contain this large hydrophilic loop. Recent data suggest that the second, cytosolic hydrophilic domain controls functional activity of the RodX when expressed in heterologous systems (4Cooper C.B. Winkfein R.J. Szerencsei R.T. Schnetkamp P.P.M. Biochemistry. 1999; 38: 6276-6283Crossref PubMed Scopus (32) Google Scholar). There is no information regarding the functional role of the large extracellular domain of the RodX. Most proteins that span the membrane multiple times (referred to as polytopic membrane proteins), such as the RodX, do not contain cleaved signal sequences (5Friedlander M. Blobel G. Nature. 1985; 318: 338-343Crossref PubMed Scopus (104) Google Scholar) (although there is no strict consensus, signal sequences tend to be 20 to 30 amino acids in length with a small amino acid at the −1 position (6von Heinje G. Eur. J. Biochem. 1983; 133: 17-21Crossref PubMed Scopus (1596) Google Scholar)). These polytopic proteins contain internal, uncleaved signal sequences that serve to direct the proper targeting, translocation, and integration of membrane-spanning segments. These internal topogenic sequences may also function autonomously, directing the translocation or integration of a single transmembrane segment in the absence of the remainder of the protein. Several polytopic membrane proteins with cleaved signal sequences have been identified and include the acetylcholine receptor (7Anderson D.J. Walter P. Blobel G. J. Cell Biol. 1982; 93: 501-506Crossref PubMed Scopus (57) Google Scholar), lactose permease (8Kaback H.R. Friedlander M. Mueckler M. Molecular Biology of Receptors and Transporters. Academic Press, Inc., San Diego, CA1992: 98-125Google Scholar), the 5-hydroxytryptamine 2c receptor (9Abramowski D. Staufenbiel M. J. Neurochem. 1995; 65: 782-790Crossref PubMed Scopus (33) Google Scholar), and the cardiac, renal and brain Na+/Ca2+ exchangers (1Philipson K.D. Nicoll D.A. Curr. Opin. Cell Biol. 1992; 4: 678-683Crossref PubMed Scopus (45) Google Scholar). Cloning of bovine RodX (bRodX)1 was reported by Reilaender et al. (2Reilaender H. Achilles A. Friedel U. Maul G. Lottspeich F. Cook N.J. EMBO J. 1992; 11: 1689-1695Crossref PubMed Scopus (159) Google Scholar), and the authors suggested that, similar to the cardiac exchanger, the N terminus may function as a cleaved signal sequence. If, in fact, the first 65 aa constitute a cleaved signal sequence, the large N-terminal hydrophilic domain would not be anchored to the membrane by a preceding transmembrane segment. If, on the other hand, the N terminal hydrophobic domain is not cleaved, the topology of this region of the protein would be significantly different, the large hydrophilic loop anchored to the membrane. Although the role of the N-terminal signal sequence in targeting and function of the RodX has not been studied, several studies of the sodium calcium exchanger from heart (10Sahin-Toth M. Nicoll D.A. Frank J.S. Philipson K.D. Friedlander M. Biochem. Biophys. Res. Commun. 1995; 212: 968-974Crossref PubMed Scopus (25) Google Scholar), brain (11Furman I. Cook O. Kasir J. Low W. Rahaminoff H. J. Biol. Chem. 1995; 270: 19120-19127Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar), and kidney (12Loo T.W. Ho C. Clarke D.M. J. Biol. Chem. 1995; 270: 19345-19350Abstract Full Text Full Text PDF PubMed Scopus (91) Google Scholar) demonstrate that disruption of the cleavage site or deletion of the signal sequence from the cardiac Na+/Ca2+ exchanger did not prevent membrane insertion or functional expression of the protein (10Sahin-Toth M. Nicoll D.A. Frank J.S. Philipson K.D. Friedlander M. Biochem. Biophys. Res. Commun. 1995; 212: 968-974Crossref PubMed Scopus (25) Google Scholar). These studies indicate that a cleaved N-terminal signal sequence is not required for folding and insertion of these exchangers into the membrane and support the concept that functional expression may be directed by internal topogenic signal sequences (5Friedlander M. Blobel G. Nature. 1985; 318: 338-343Crossref PubMed Scopus (104) Google Scholar). To study the role of the N-terminal hydrophobic domain in targeting and translocation of RodX, we generated three classes of modified RodX: 1) truncated constructs (bRodN) containing the coding sequence of RodX through the first putative transmembrane (or second hydrophobic) domain to study targeting and translocation in the absence of any internal signal sequences, 2) full-length exchanger with (bRodX) or without (bRodXΔ65) the first 65 aa to study the targeting of RodX without the native N terminus, and 3) chimeric proteins containing native bRodX N terminus or known cleaved signal sequences from β-lactamase. A more limited set of analogous constructs was generated with hRodX. Under no conditions did we observe cleavage of the N-terminal 65 aa of RodXin vitro. Furthermore, Western blot analysis of fresh bovine retinal membrane fractions using antipeptide antibodies directed against aa 10 to 19 or 179 to 194 of the bRod X resulted in the observation of an identically sized 210-kilodalton protein with either antibody. This indicates that the N terminus is present in at least a portion of RodX in vivo in bovine retina. The full-length cDNA of bRodX was a generous gift from Dr. Helmut Reilaender (Max-Planck-Institut fur Biophysik, Frankfurt/M, Germany). A series of full-length (bRodX) and truncated mutants (bRodN) were generated by a combination of PCR and subcloning methods. All constructs were sequenced to ensure that no mutations were introduced by PCR during amplification. bRodN constructs were generated (see Fig. 2 A) in which 1) the coding region for the first 65 aa was deleted (bRodNΔ65), 2) the putative cleavage site at Arg-65 was mutated to code for Ala (bRodN/R65A), or 3) the first 65 aa were replaced by the cleavable signal sequence of the canine cardiac sodium-calcium exchanger (bRodN/CXSS). The Δ65 mutant was generated by PCR using a 5′-oligonucleotide that contained the coding sequence for an initiator methionine. The R65A mutant was generated by two-stage PCR mutagenesis (overlap extension (13Ho S.N. Hunt H.D. Horton R.M. Pullen J.K. Pease L.R. Gene. 1989; 77: 51-59Crossref PubMed Scopus (6771) Google Scholar)) to create the single amino acid change of Arg to Ala (nt CGG to GCG). The bRodN/CXSS fusion construct was generated by introducing a PvuII site by PCR silent mutagenesis into both the 3′ end of the coding sequence for the canine cardiac Na+/Ca2+ exchanger signal sequence and the 5′ end of the bRodX at aa 66. The fragments were digested with the appropriate enzymes and subcloned sequentially into the appropriate vector. All bRodN constructs were subcloned into the EcoRV and PstI sites of pSP72 (Promega). Termination codons were added to the bRodN constructs by PCR. A 3′ primer was designed to anneal to nt 1443–1449 was followed by the coding sequence for two stop codons and an XbaI site. This oligonucleotide and an SP6 primer were used to generate PCR fragments that were subsequently gel-purified, digested, and subcloned into pSP72. pSPSLGSTP (14Lingappa V.R. Katz F.N. Lodish H.F. Blobel G. J. Biol. Chem. 1978; 253: 8667-8670Abstract Full Text PDF PubMed Google Scholar) was digested withBglII and NcoI to excise the β-lactamase signal sequence. The first 65 aa of RodX were amplified by PCR with oligonucleotide primers containing the appropriate restriction sites. The PCR product was digested and subcloned into the pSPSLGSTP vector. Plasmid DNA was prepared using the Qiaprep spin minikit (Qiagen). DNAs were digested with the appropriate restriction enzyme. Run-off transcripts were prepared per manufacturer's instructions using SP6 or T7 Message Machine kits (Ambion). RNA was quantified by gel electrophoresis or spectrophotometry. In vitro translations were performed using rabbit reticulocyte lysate (Promega) in the presence of [35S]methionine (ICN Biochemicals). Proteins were translated in the presence or absence of dog pancreatic microsomal membranes (Promega) per manufacturer's instructions. A portion of the reactions carried out in the presence of membranes were subsequently treated with endoglycosidase H (Endo H) (New England Biolabs) to removeN-linked sugars. Samples were either precipitated with 2 volumes of saturated ammonium sulfate (Sigma) or fractionated. Proteins were resolved by SDS-PAGE and visualized by autoradiography. Glycosylation and cleavage were assayed by mobility shift on SDS-PAGE. Fractionations were carried out as described by Borel and Simon (15Borel A.C. Simon S.M. Cell. 1996; 85: 379-389Abstract Full Text Full Text PDF PubMed Scopus (118) Google Scholar). Briefly, 80% of the reaction was loaded onto a 115-μl 0.5 m/2.0 m sucrose cushion in 10 mm HEPES, pH 7.4, and centrifuged at 186,000 ×g for 15 min in a Beckman TLA 100.1 rotor. The 0.5m supernatant (S1) containing proteins not associated with membranes was removed. The 2.0 m sucrose cushion was extracted with urea (7 m final concentration) for 30 min on ice. The samples were then centrifuged at 280,000 × gfor 30 min. The supernatant (S2) containing membrane-associated, nonintegrated proteins was removed. Supernatant fractions were precipitated with two volumes of saturated ammonium sulfate. The remaining pellet (P) containing the proteins stably integrated into the membrane was resuspended in 1% sodium deoxycholate (Sigma), 0.1 n NaOH (Sigma). Proteins were resolved by SDS-PAGE and visualized by autoradiography and quantified using a Bio-Rad Molecular Imager. Oligonucleotide primers were synthesized corresponding to the coding sequence for the eighth and ninth hydrophobic domains of bRodX. A series of nested PCRs were performed using an adult human retinal library generously provided by Dr. Jeremy Nathans of Johns Hopkins University. Four DNA segments were amplified corresponding to nt 3362–3732, 3362–3693, 3382–3703, and 3464–3693 of the bovine sequence. The sequences of the fragments were 89–94% identical to the corresponding bovine sequence. The fragments were labeled by random priming (16Feinberg A.P. Anal. Biochem. 1984; 137: 266-267Crossref PubMed Scopus (5169) Google Scholar) and used as probes to screen the library (17Dobner P.R. Barber D.L. Villa-Komaroff L. McKiernan C. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 3516-3520Crossref PubMed Scopus (163) Google Scholar). Positive clones were plaque-purified and sequenced. Four clones corresponding to the coding sequence of the C-terminal half of the protein were identified. Primers were made to amplify the most 5′ portion of the most 5′ clone. Again, the PCR product was labeled by random priming and was used to rescreen the cDNA library. Two additional clones were identified. Using published sequence information (18Nicoll D.A. Applebury M.L. J. Biol. Chem. 1989; 264: 16207-16213Abstract Full Text PDF PubMed Google Scholar) oligonucleotide primers were synthesized to correspond to nt 1–22 and antisense 1500–1480. The 5′-end of the cDNA was amplified from the library by PCR. Three clones (nt 1–1500, 1162–2336, and 1975 to the end) were assembled into the full-length cDNA at the PstI (nt 1462) andEcoNI (nt 2058) sites. The full-length clone was inserted into the HindIII and XbaI sites of pSP73 (Promega). The full-length clone was sequenced by cycle sequencing using an automated sequencer (Applied Biosystems, Inc.) and found to be identical to the clone reported by Tucker et al. (3Tucker J.E. Winkfein R.J. Cooper C.B. Schnetkamp P.M.M. Investig. Ophthamol. Visual Sci. 1998; 39: 435-440PubMed Google Scholar). Constructs of the N-terminal portion of the human Rod X were generated in a similar fashion to the analogous bovine constructs. The 1–1500-nt fragment amplified from the library was subcloned into the pCR2.1 vector (Invitrogen). Termination codons and an XbaI site were added after the sequence coding for aa 487 to generate a construct analogous to bRodN. The hRodN Δ65 mutant was generated by PCR using a 5′-oligonucleotide primer containing a HindIII site and an ATG for an initiator methionine fused to nt 198 to 208 and the 3′-primer with the termination codons used above. The amplified products were digested with HindIII and XbaI and subcloned into pSP73 (Promega). The Δ65 mutation was generated in full-length hRodX by amplifying the nt 1–1500 fragment from the library with a 5′-oligonucleotide described above and an oligonucleotide corresponding to antisense nt 1480–1500. The resulting PCR product was digested with PstI and HindIII and subcloned into the full-length construct digested with the same enzymes. All constructs were sequenced to determine that no mutations were introduced during amplification. Two peptides were synthesized corresponding to amino acids 10–19 and 179–194 of the bovine rod exchanger. Peptides were linked to keyhole lymphocyte hemocyanin and sent to Cocalico Biologicals, Inc. for antibody production in rabbits. Antibodies were tested for specificity by Western blots against recombinant GST fusion proteins fused to the peptide sequence. A crude retinal membrane extract was prepared as described by Nicoll and Applebury (18Nicoll D.A. Applebury M.L. J. Biol. Chem. 1989; 264: 16207-16213Abstract Full Text PDF PubMed Google Scholar). Briefly, retinas were dissected on ice in a cold room from freshly enucleated bovine eyes and homogenized with 10 strokes using a motorized homogenizer in buffer (30% sucrose, 10 mm Tris-Cl, pH 7.4, 65 mm NaCl, 2 mm MgCl2) with a complete mixture of protease inhibitors (aprotinin, leupeptin, pepstatin, phenylmethylsulfonyl fluoride). Particulate matter was removed by centrifugation at 2000 × g. The supernatant was diluted with 2 volumes of 10 mm Tris-Cl, pH 7.4 with fresh protease inhibitors. The membranes were recovered from the supernatant by centrifugation at 10,000 × g. The pellet was resuspended in the same buffer as above with fresh protease inhibitors, separated into aliquots, and flash frozen in liquid nitrogen. Proteins were resolved by SDS-PAGE and transferred to nitrocellulose (Micron Separations, Inc.) using a Bio-Rad Protean II. Blots were stained with Ponceau S (Sigma) to evaluate protein transfer. Stain was removed with TBST (20 mm Tris-Cl, pH 7.4, 150 mm NaCl, 0.1% Tween 20). Blots were blocked in 5% nonfat dry milk in TBST for 60 min, rinsed twice in TBST, and incubated with immune or preimmune sera in 3% bovine serum albumin in TBST for 2 h. Blots were washed 6 times for 5 min in TBST and incubated in peroxidase-conjugated anti-rabbit IgG (whole molecule) F(ab′)2 goat antibody (Sigma) in 3% bovine serum albumin in TBST. The blots were then washed 8 times for 5 min in TBST. Bands were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech). bRodX is a polytopic membrane protein predicted to contain 12 transmembrane domains by Kyte-Dolittle hydropathy analysis (Fig. 1 A) (19Kyte J. Doolittle R.F. J. Mol. Biol. 1982; 157: 105-132Crossref PubMed Scopus (16899) Google Scholar). In preliminary experiments, full-length bRodX was translated in the presence and absence of microsomal membranes. A decrease in the mobility of the translation product was observed in the presence of microsomal membranes due to glycosylation by glycosyltransferases present in the lumen of the microsomal vesicles (Fig. 1 B). A portion of the reaction carried out in the presence of membranes was treated with EndoH to remove carbohydrate moieties in order to detect mobility shifts on SDS-PAGE indicative of protein cleavage. Treatment of the reaction mixture with Endo H returned the product to the mobility of the primary translation product. No cleavage could be detected. Since the change in molecular weight upon processing of the full-length protein is small relative to the total molecular mass of the protein, we prepared a truncated RodX consisting of the first 483 aa only (Fig. 2 A) to 1) make mobility shifts more readily detectable and 2) better evaluate the role of the N terminus as a signal sequence in the absence of any internal signal sequences. Experiments using further truncated cDNAs lacking the coding region for the second hydrophobic domain were performed. They demonstrated that the second hydrophobic domain was not sufficient to direct membrane translocation. In the presence of membranes, bRodN was translocated across the membrane as evidenced by a decrease in mobility due to glycosylation (Fig. 2 B, compare lanes 2 and 3). After treatment with EndoH, the protein returned to the size of the primary translation product, indicating that the protein was not cleaved. Experiments with the N-terminal-deleted RodNΔ65 mutant demonstrated that the first 65 aa were required to direct the translocation of the N terminus across the membrane. The translation product was the same size whether or not microsomal membranes were present (Fig. 2 B,lanes 5 to 7). Cleaved signal sequences generally have a small aa in the −1 position relative to the cleavage site (6von Heinje G. Eur. J. Biochem. 1983; 133: 17-21Crossref PubMed Scopus (1596) Google Scholar). To determine if the Arg in the −1 position inhibited cleavage, we mutated it to an Ala. This mutation of the previously reported signal sequence cleavage site (R65A) did not affect the targeting or translocation of RodN (Fig. 2 B, compare lanes 2to 4 with lanes 8 to 10). Having determined that the presence but not the cleavage of the first 65 aa of RodX were required for targeting and translocation of the N terminus, we wanted to determine if a known cleaved signal sequence could substitute for the first 65 aa of RodX. A chimeric construct was prepared in which the cleaved signal sequence from the canine cardiac Na+/Ca2+ exchanger (10Sahin-Toth M. Nicoll D.A. Frank J.S. Philipson K.D. Friedlander M. Biochem. Biophys. Res. Commun. 1995; 212: 968-974Crossref PubMed Scopus (25) Google Scholar) was fused in-frame to aa 66–483 of bRodN. Translation of this chimeric protein in the presence of microsomal membranes produced a slight decrease in the mobility of the translation product (Fig. 2 B, comparelanes 11 and 12); treatment with EndoH resulted in an increase in mobility when compared with the primary translation product. The translation product had a mobility similar to the primary translation product of the Δ65 mutant (Fig. 2 B, comparelanes 5 and 13. The minimal change in mobility in the presence of microsomal membranes is explained by the decrease of the mobility due to glycosylation being offset by the increase in mobility due to the cleavage of the N-terminal signal sequence. A similar effect is observed when the native cardiac Na+/Ca2+ exchanger is translated in the presence of the microsomal membranes in vitro (20Hryshko L.V. Nicoll D.A. Weiss J.N. Philipson K.D. Biochim. Biophys. Acta. 1993; 1151: 35-42Crossref PubMed Scopus (62) Google Scholar). Human RodX constructs analogous to bRodN and bRodNΔ65 were prepared to determine if the first 65 aa functioned similarly in the human isoform. Results from coupled translation/translocation experiments with the hRodN were identical to bRodN (Fig. 2 C) as were results from experiments using full-length RodX (data not shown). Our observation that RodXΔ65 was not glycosylated in the presence of microsomal membranes could be explained by 1) the protein not being targeted to the membrane or 2) the protein being inserted in reverse orientation such that the glycosylation sites were outside of the vesicle and not accessible to the glycosyltransferases. To distinguish between these two possibilities, translation reactions were fractionated over sucrose cushions and extracted with urea to distinguish stably integrated proteins (P) from ones peripherally associated with the membrane or in the translocation channel (S2) (21Gilmore R. Blobel G. Cell. 1985; 42: 497-505Abstract Full Text PDF PubMed Scopus (163) Google Scholar). Fractionations were carried out in the presence and absence of microsomal membranes. A small fraction of translation product was always found in the pellet (P) fraction in the absence of membranes due to nonspecific aggregation of hydrophobic stretches present in these proteins (5Friedlander M. Blobel G. Nature. 1985; 318: 338-343Crossref PubMed Scopus (104) Google Scholar). Constructs containing only the N terminus of the exchanger through the second hydrophobic domain were used to investigate the role of the first 65 aa independently of any internal signal sequences. In the absence of membranes, nearly 75% of bRodN was found in the S1 fraction (Fig. 3 A, comparelane 2 to lane 4; Fig. 3 B). Upon the addition of membranes, the protein was found almost exclusively in the P fraction (Fig. 3 A, lane 8, and B). Both in the presence and the absence of membranes, the Rod N Δ65 mutant was found predominantly in the S1 fraction (Fig. 3 A,lanes 10 and 14, and B). Quantification of fractionation experiments showed no difference in the proportion of the reaction product in the P fractions of bRodN without membranes and both bRodNΔ65 without and with membranes (Tukey test,p > 0.05) (Fig. 3 B). The portion of the reaction product in the P fraction of the bRodN fraction with membranes was significantly different (p < 0.001). These data indicate that the lack of glycosylation of the N terminus is due to a targeting defect rather than the N terminus being inserted in reverse orientation. Experiments performed with the hRodN constructs yielded similar results (data not shown). To determine if RodX contains internal signal sequences sufficient to target it to the membrane, we fractionated full-length RodX with and without the first 65 aa. Proteins were translated in vitro and fractionated as above. The full-length exchanger with and without the first 65 aa targets to the membrane and becomes stably integrated (Fig. 4 A, compare lanes 8and 16). We observed a portion of the translation product in the P fractions even in the absence of membranes (lanes 4and 12) as discussed above. Phosphoimager analysis revealed that there was no difference in the fractionation pattern of the full-length and N-terminal-deleted protein (Fig. 4 B). No change in the mobility of the Δ65 mutant was ever observed upon translation in the presence of membranes. Thus, in the absence of the N-terminal 65 aa, RodX was stably inserted into the membrane, but the appropriate final topology was never attained. Experiments performed with the hRodX constructs produced similar results (data not shown). The coding sequence for the first 65 aa RodX were fused in-frame to a cDNA encoding a chimeric protein containing a portion of the globin gene with an engineered glycosylation site, a stop transfer sequence from immunoglobulin M heavy chain, and a portion of prolactin. In the presence of microsomal membranes, chimeric protein was targeted and translocated across the membrane as evidenced by a decrease in mobility in the presence of membranes (Fig. 5 B, comparelanes 2 and 3). Treatment with EndoH resulted in an increased mobility when compared with the primary translation product (Fig. 5 B, compare lanes 2 and4), indicating that the protein was cleaved. When the N-terminal 65 aa of RodX were used in the place of the β-lactamase signal sequence, no cleavage was observed (Fig. 5 B, comparelanes 5 and 7); however the N terminus of the protein was translocated as evidenced by its glycosylation in the presence of microsomal membranes (compare lanes 5 and 6). Translocation was not due to the immunoglobulin M stop transfer sequence. Identical results were observed with a construct in which truncation occurred at the end of the globin coding sequence (data not shown). Thus, the first 65 aa of RodX can function as an uncleaved N-terminal signal sequence in both heterologous, as well as native, contexts. To determine if the first 65 aa are cleaved from bovine RodX in vivo, we prepared crude bovine retinal extracts for analysis by Western blot using antipeptide antibodies raised to both the N terminus (aa 10–19) and an internal portion of the protein (aa 179–194). Conditions were selected to minimize proteolysis by preparing a crude extract to reduce preparation time, conducting the entire procedure in the cold room, and using a broad range of protease inhibitors in the buffers. Using an antibody directed against either the N terminus (aa 10–19) or an internal sequence (aa 179–194), an identical, single band of approximately 210 kDa was detected by Western blots (Fig. 6). A lower molecular weight species was not detected. Since the wild type, full-length native RodX has an apparent molecular mass of 210 kDa by SDS-PAGE and the cleaved form, a predicted molecular mass of 203 kDa, we interpret these data to indicate that the majority of the native RodX purified from fresh photoreceptors is not cleaved at aa 65. These in vivo data support our in vitro observation that the N-terminal 65 aa of RodX serve as an uncleaved signal sequence. Multiple attempts to obtain N-terminal sequence of RodX freshly purified from bovine retina in our laboratory revealed that the N terminus is blocked and not available for sequencing (data not shown). In this study we report that the first 65 aa of RodX constitute an uncleaved signal sequence sufficient to target the protein to microsomal vesicles; in addition, the first 65 aa of RodX are required to translocate the N-terminal hydrophilic domain across the membrane. Translocation of the N terminus could also be directed by the cleaved N-terminal signal sequence from the cardiac Na+/Ca2+ exchanger signal sequence. It is also of interest to note that although the native N-terminal signal sequence is uncleaved, the cardiac Na+/Ca2+ signal sequence, when fused to Rod X Δ65, could be cleaved in vitro, consistent with its behavior in the native cardiac Na+/Ca2+ exchanger. Last, the first 65 aa of bRodX also function as an uncleaved signal sequence in a heterologous context, targeting globin, a soluble cytosolic protein, to microsomal membranes. Deletion of the first 65 aa from a truncated form of both hRodN and bRodN results in a protein that does not associate with microsomal vesicles. In the context of the full-length protein though, the N-terminal 65 aa are not required for targeting and insertion of the protein into the membrane. RodXΔ65 is targeted to the membrane as efficiently as the full-length protein, but the N-terminal hydrophilic domain is never translocated as evidenced by a lack of glycosylation. Hence, the protein is stably integrated into the membrane, yet it never attains the correct, final topology. In the absence of the first 65 aa of RodX, the N-terminal 382 aa preceding the first transmembrane domain would have to be translocated post-translationally, as opposed to co-translationally, in the presence of a signal sequence. This type of post-translational translocation occurs in type III proteins that contain no N-terminal signal sequence but instead contain a C-terminal membrane anchor. Unlike RodXΔ65 though, the protein segment preceding the first transmembrane domain is small, usually less than 100 aa in length (22Kutay U. Hartmann E. Rapoport T.A. Trends Cell Biol. 1993; 3: 72-75Abstract Full Text PDF PubMed Scopus (262) Google Scholar). Post-translational translocation through membranes requires chaperones in the case of mitochondria (23Hajek P. Koh J.Y. Jones L. Bedwell D.M. Mol. Cell. Biol. 1997; 17: 7169-7177Crossref PubMed Google Scholar) as well as in yeast (24McClellan A.J. Endres J.B. Vogel J.P. Palazzi D. Rose M.D. Brodsky J.L. Mol. Biol. Cell. 1998; 9: 3533-3545Crossref PubMed Scopus (73) Google Scholar) and bacteria (25Driessen A.J. Fekkes P. van der Wolk J.P. Curr. Opin. Microbiol. 1998; 1: 216-222Crossref PubMed Scopus (146) Google Scholar). These cytosolic chaperones maintain the protein in an unfolded, translocation competent structure. As the large hydrophilic loop of RodX would normally be translated with the ribosome attached to the translocation machinery, it would not interact with cytosolic proteins. It is unlikely that cytosolic chaperone binding sites would be present in the loop. In the absence of endoplasmic reticulum resident chaperones, proteins that normally fold in the endoplasmic reticulum aggregate; hence, they would be unavailable for translocation (26Haynes R.L. Sheng T. Nicchitta C.V. J. Biol. Chem. 1997; 272: 17126-17133Abstract Full Text Full Text PDF PubMed Scopus (13) Google Scholar). We did not observe cleavage of the N terminus of bRodX in vivo as evidenced by Western blot analysis of freshly prepared total retinal extracts. Earlier reports describing the cloning of RodX suggested that the cleavage at Asp-66 could have been the result of a nonspecific proteolysis during purification of the protein (2Reilaender H. Achilles A. Friedel U. Maul G. Lottspeich F. Cook N.J. EMBO J. 1992; 11: 1689-1695Crossref PubMed Scopus (159) Google Scholar); such a cleavage during purification would result in an unblocked N terminus, which would be available for N-terminal sequencing. These data do not preclude the possibility that a portion of RodX is not cleaved in vivo but suggest that any such cleavage would most likely not be due to a signal peptidase cleavage event. The requirement of an uncleaved signal sequence on Rod X in order for the protein to be correctly targeted and translocated across microsomal membranes distinguishes it from other cation exchangers that are able to functionally insert into the membrane in the absence of an N-terminal signal sequence (10Sahin-Toth M. Nicoll D.A. Frank J.S. Philipson K.D. Friedlander M. Biochem. Biophys. Res. Commun. 1995; 212: 968-974Crossref PubMed Scopus (25) Google Scholar, 11Furman I. Cook O. Kasir J. Low W. Rahaminoff H. J. Biol. Chem. 1995; 270: 19120-19127Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar, 12Loo T.W. Ho C. Clarke D.M. J. Biol. Chem. 1995; 270: 19345-19350Abstract Full Text Full Text PDF PubMed Scopus (91) Google Scholar). This difference may be due to the presence of the large N-terminal hydrophilic domain present between the first and second hydrophobic segments of RodX. The topological information in the first transmembrane domain may be sufficient to direct the translocation of a short N terminus across the membrane in the absence of an N-terminal signal sequence (as in the cardiac exchanger) but not adequate for a segment as long as that of the rod exchanger. Further studies will need to be performed to determine if the requirement for a noncleaved signal sequence is dependent upon the length of the protein segment that must be translocated across the membrane or if it is a peculiarity of retinal-specific proteins. We thank Stacey Hanekamp and Kelli Connaughton for assistance in the cloning of the human RodX. We are also grateful to Gunter Blobel, Sheila Fallon, Ken Philipson and Sandy Simon for helpful comments and suggestions on the manuscript and throughout the course of these studies.