- Split View
-
Views
-
Cite
Cite
Roland Kabuß, Angel Ashikov, Stefan Oelmann, Rita Gerardy-Schahn, Hans Bakker, Endoplasmic reticulum retention of the large splice variant of the UDP-galactose transporter is caused by a dilysine motif, Glycobiology, Volume 15, Issue 10, October 2005, Pages 905–911, https://doi.org/10.1093/glycob/cwi085
- Share Icon Share
Abstract
Nucleotide-sugar transporters supply mainly the Golgi glycosyltransferases with substrates. Some glycosyltransferases in the endoplasmic reticulum (ER), however, also use activated sugars. Recent studies have demonstrated that UDP-galactose (UDP-Gal) is the substrate for the ER resident ceramide-galactosyltransferase (cer-GalT) and cells expressing cer-GalT are able to retain the UDP-Gal transporter (UGT) by physical contacts formed between the two proteins. Here, we describe a second active mechanism for ER localization of the UGT. The UGT is produced in two splice forms UGT1 and UGT2. The proteins vary only at their extreme C-termini but show strikingly different intracellular distribution. Although N-terminally epitope tagged forms of UGT1 localize exclusively to the Golgi, similar constructs of UGT2 show both ER and Golgi localization. The dilysine motif KVKGS contained in UGT2 can be demonstrated to be responsible for the dual localization because: (1) disturbance of the signal via site specific mutation or C-terminal extension completely shifts the transporter to the Golgi, (2) transfer of the dilysine motif is sufficient to redistribute the Golgi CMP-sialic acid transporter to the ER, and (3) replacement of KVKGS by the strong ER retention signal KKNT is sufficient to completely retain UGT2 in the ER.
Introduction
Nucleotide-sugar transporters (NSTs) supply the glycosyltransferases involved in the biosynthesis of glycoconjugates along the secretory pathway of eukaryotic cells with activated sugars that are synthesized in the cytoplasm or cell nucleus (Hirschberg et al., 1998). Most glycosyltransferases that use nucleotide activated sugars are active in the Golgi, whereas the majority of ER resident glycosyltransferases depends on the presentation of activated sugars by dolichol phosphate (Burda and Aebi, 1999). Accordingly, nucleotide sugar transport activities could be measured in Golgi membrane preparations (Sommers and Hirschberg, 1982; Capasso and Hirschberg, 1984; Hirschberg et al., 1998) and after molecular cloning, the corresponding NSTs could be localized to the Golgi membranes (Eckhardt et al., 1996; Dean et al., 1997; Yoshioka et al., 1997; Ishida et al., 1999; Lühn et al., 2001). Transport of some specific nucleotide sugars into ER vesicles must, however, exist, because a panel of nucleotide sugar requiring glycosylation reactions takes place in the ER. The UDP-glucose : glycoprotein glucosyltransferase (Zuber et al., 2001) uses a nucleotide sugar as substrate, evidence has been provided that proteoglycan biosynthesis is initiated in the ER (Kearns et al., 1993), and some members of the UDP-glycosyltransferase gene superfamily (Mackenzie et al., 1997) involved in glycosylation of lipophilic compounds (Meech and Mackenzie, 1997; de Wildt et al., 1999) are located in the ER lumen. Indeed, nucleotide sugar transport activities corresponding to these reactions could be measured in ER vesicles (Bossuyt and Blanckaert, 1997; Hirschberg et al., 1998; Castro et al., 1999), and, by indirect immunostaining methods, the yeast UDP-N-acetylglucosamine transporter (Roy et al., 2000) and the human UDP-glucuronic acid/UDP-N-acetylgalactosamine transporter (Muraoka et al., 2001) have been localized to the ER. Interestingly, the yeast UDP-N-acetylglucosamine transporter has a C-terminal dilysine motif (Nilsson et al., 1989; Zerangue et al., 2001), potentially responsible for ER retention.
UDP-galactose (UDP-Gal) transport has been exclusively assigned to the Golgi (Perez and Hirschberg, 1985; Hirschberg et al., 1998), and, after it was cloned, the UDP-Gal transporter (UGT) has been localized to the Golgi (Ishida et al., 1996; Yoshioka et al., 1997; Oelmann et al., 2001). However, in a more recent study, Sprong et al. (1998) have shown that the ER-resident ceramide-galactosyltransferase (cer-GalT) is inactive, if expressed in chinese hamster ovary (CHO) cells of the complementation group Lec8, which exhibit a genetic defect in the UDP-Gal transport protein (Deutscher and Hirschberg, 1986). The lack of functionality could be complemented by cotransfecting the UGT cDNA (Sprong et al., 1998). The presented data clearly indicate that the lack of substrate in the ER prevented enzymatic activity. The same group has meanwhile shown that cer-GalT is able to make physical contacts to the UGT and thus can retain the protein in the ER (Sprong et al., 2003).
The existence of an alternative splice site in the human UGT gene enables the translation of two protein variants (Ishida et al., 1996). As shown in Figure 1, translation of the longer message results in the expression of UGT1 having the C-terminal sequence SVLVK, whereas translation of the shorter mRNA generates UGT2, in which the sequence SVLVK is replaced by LLTKVKGS. UGT2 contains a dilysine motif, shown in other proteins to limit exit from the ER (Nilsson et al., 1989; Zerangue et al., 2001).
In this study, we present a set of experiments, which conclusively demonstrates that the dilysine motif present in UGT2 is an autonomously active ER retention signal. We, therefore, suggest that a second, cer-GalT independent mechanism exists, by which the UGT can be retained in the ER.
Results
Differences in the subcellular localization of human UDP-Gal transporters 1 and 2
The first indication that the dilysine motif in UGT2 functions as an ER retention signal came from the observation that a C-terminally tagged hamster UGT2 localized exclusively to the Golgi, whereas addition of an N-terminal tag changed the staining pattern. Golgi staining remained and an additional signal, suggesting ER retention of the N-terminally tagged protein, became visible.
To directly show the difference in localization between human UGT1 and UGT2, we transiently expressed N-terminally FLAG-tagged constructs of both splice variants in CHO Lec8 cells and analyzed by intracellular immunofluorescence. Although the FLAG-tagged UGT1 clearly colocalized with the Golgi marker α-mannosidase II and showed no overlap with the ER marker calnexin (Figure 2), FLAG-tagged UGT2 displayed, in addition to the Golgi signal, a signal that could be addressed to the ER by colocalization with calnexin. Therefore, the difference in localization pattern between N-terminally FLAG-tagged UGT1 and UGT2 seems to be exclusively determined by the respective 5 or 8 C-terminal amino acids of the transport proteins.
The KxKxx motif is responsible for ER retention
As UGT2 contains the putative ER retaining dilysine motif KVKGS, we examined whether these five amino acids autonomously function as ER retention signal when added to a related protein like the CMP-sialic acid transporter (CST), which is solely Golgi localized (Eckhardt et al., 1999). Results shown in Figure 3 clearly demonstrate that the natural CST stained as a Golgi resident protein, whereas the KVKGS-extended form of the protein was distributed over ER and Golgi (Figure 3B and C).
In contrast, mutation of the lysine at the –3 position to alanine (KVKGS to KVAGS) in UGT2 resulted in transport of the protein to the Golgi (Figure 3F and G). These experiments clearly demonstrate that the dilysine motif contained in UGT2 acts as an autonomous ER-retention signal and confirm earlier studies, in which the crucial role of the lysine-residue in position –3 has been demonstrated (Zerangue et al., 2001). Obviously, the observation that N- and C-terminally tagged hamster UGT2 behaved different could also be explained by the fact that a C-terminal tag is masking the dilysine motif.
The dilysine motif in UGT2 is a weak ER retention signal
It is known that not all C-terminal KKXX and KXKXX sequences efficiently retain proteins in the ER (Itin et al., 1995; Andersson et al., 1999; Zerangue et al., 2001). The fact that also UGT2 and CST carrying the KVKGS motif localized to ER and Golgi leaves open the question if this is a property of the nucleotide sugar transporters or the dilysine motif. To investigate this, we replaced the KVKGS sequence in UGT2 by KKNT, a strong ER retention signal (Zerangue et al., 2001). Similarly, we tagged CST with the KKNT motif. Both transporters now showed exclusive ER localization (Figure 4). In combination with the above results these data demonstrate that the KVKGS sequence in UGT2 functions as an ER retention signal, but the destination signal is weak.
Asking whether the functionality of UDP-Gal and CMP-sialic acid transport proteins is altered by changes in their subcellular localization, we used constructs made in this study (listed in Table I) to test complementation activity in CHO Lec8 (Deutscher and Hirschberg, 1986) or Lec2 (Stanley and Siminovitch, 1977) cells. Cells were transiently transformed with the respective cDNAs, and reconstitution of wild type was determined in flow cytometry analyses using the polysialic acid specific monoclonal antibody 735 (Frosch et al., 1985) in the case of Lec2 cells (Eckhardt et al., 1996) and the HNK-1 antibody L2-412 (Kruse et al., 1984) in the case of Lec8 cells (Bakker et al., 2005). To induce the formation of the HNK-1 epitope in Lec8 cells, we cotransfected the UGT cDNAs with a cDNA encoding the β1,3glucuronyltransferase (Terayama et al., 1997). The results shown in Figure 5 indicate that all constructs, independent on their subcellular localization, are able to restore the wild type glycosylation patterns.
Construct . | C-terminus . |
---|---|
hUGT1 | PKSVLVK |
hUGT2 | PKLLTKVKGS |
hamUGT2 | PKLLTKVKGS |
hamUGT2-K396A | PKLLTKVAGS |
hamUGT2-BamHI | PKLGSKVKGS |
hamUGT2-KKTN | PKLGSKKTN |
mCST | IIGV |
mCST-KVKGS | IIGVGSKVKGS |
mCST-KKTN | IIGVGSKKTN |
Construct . | C-terminus . |
---|---|
hUGT1 | PKSVLVK |
hUGT2 | PKLLTKVKGS |
hamUGT2 | PKLLTKVKGS |
hamUGT2-K396A | PKLLTKVAGS |
hamUGT2-BamHI | PKLGSKVKGS |
hamUGT2-KKTN | PKLGSKKTN |
mCST | IIGV |
mCST-KVKGS | IIGVGSKVKGS |
mCST-KKTN | IIGVGSKKTN |
All constructs were made in pcDNA3, and the expressed proteins have a FLAG tag (MDYKDDDDK) at the N-terminus and the indicated sequence at the C-terminus. Amino acid changes form the original sequences are indicated in boldface. h, human; ham, hamster; m, mouse.
Construct . | C-terminus . |
---|---|
hUGT1 | PKSVLVK |
hUGT2 | PKLLTKVKGS |
hamUGT2 | PKLLTKVKGS |
hamUGT2-K396A | PKLLTKVAGS |
hamUGT2-BamHI | PKLGSKVKGS |
hamUGT2-KKTN | PKLGSKKTN |
mCST | IIGV |
mCST-KVKGS | IIGVGSKVKGS |
mCST-KKTN | IIGVGSKKTN |
Construct . | C-terminus . |
---|---|
hUGT1 | PKSVLVK |
hUGT2 | PKLLTKVKGS |
hamUGT2 | PKLLTKVKGS |
hamUGT2-K396A | PKLLTKVAGS |
hamUGT2-BamHI | PKLGSKVKGS |
hamUGT2-KKTN | PKLGSKKTN |
mCST | IIGV |
mCST-KVKGS | IIGVGSKVKGS |
mCST-KKTN | IIGVGSKKTN |
All constructs were made in pcDNA3, and the expressed proteins have a FLAG tag (MDYKDDDDK) at the N-terminus and the indicated sequence at the C-terminus. Amino acid changes form the original sequences are indicated in boldface. h, human; ham, hamster; m, mouse.
Discussion
We have demonstrated that the dilysine motif in the human and hamster UGT functions as a weak ER retention signal. In dilysine motifs, the lysine at the –3 position is essential, whereas the second lysine can be at the –4 or –5 position. But not all sequences that fulfil these criteria will target a protein to the ER, indicating that also surrounding amino acids play a role. Moreover, sequences with a lysine at –4 are more likely to function as a retention signal than those with a lysine at the –5 position (Teasdale and Jackson, 1996; Zerangue et al., 2001). Based on this information, the dilysine motif in UGT2 could be marked as a questionable retention signal. However, a clear difference in the localization of UGT1 and UGT2 has been observed in all our experiments, if proteins were expressed with N-terminal tags, whereas the use of C-terminally tagged proteins let to unequivocal Golgi staining. The putative ER retention signal KVKGS present in UGT2 was mutated to destroy the essential position –3. This experiment shown in Figure 3 resulted in a pure Golgi localized mutant. In a number of systematic analytical steps we than identified the KVKGS motif as a weak ER-retention signal, which retains activity also after transplantation to the C-terminus of CST.
Still it seems difficult to categorize dilysine motifs as retaining or nonretaining because earlier studies, using combinatorial variants of the dilysine motif, demonstrated that a cell surface protein, used as a model substance, could be targeted to virtually every compartment between ER and cell surface by the artificially made dilysine motifs (Zerangue et al., 2001). Our results, demonstrating that also a naturally occurring membrane protein can show dual destination as a result of the presence of an ER retention signal, provides important evidence for the physiological relevance of the motif.
Surprisingly, all constructs used in this study were able to complement the defect in Lec8 or Lec2 cells. No differences were visible between the complementation activities of mutant and natural proteins. Even NST variants with a strong ER retention signal, which did not show any Golgi localization (Figure 4), were able to fully restore the wild type phenotype of mutant cell lines (Figure 5). This observation suggests that activated sugars can be passively transported throughout ER and Golgi. On the other hand, one cannot rule out the possibility that small quantities of the recombinant proteins reach the Golgi apparatus and thus drive the glycosylation machinery. In transient expression systems recombinant proteins are drastically overexpressed and may overstress the systems that survey cellular targeting processes. In fact, the dilysine motifs are known to be involved in both ER retention and retrieval (Andersson et al., 1999). Thus, the system per se may, to a certain extend, allow cycling of proteins between ER and Golgi, and this process may be more pronounced in the transformed cells.
There are other members in the NST family that contain dilysine motifs (Martinez-Duncker et al., 2003). One of these is the yeast UDP-N-acetylglucosamine transporter, which has in fact been localized to the ER (Roy et al., 2000). In contrast to UGT2, the addition of a C-terminal tag was not sufficient to destroy ER-retention. Also the human UDP-glucuronic acid/UDP-N-acetylgalactosamine transporter (Muraoka et al., 2001) has been localized to the ER even so no dilysine or other signal is present that could explain ER retention. Together these data underline that other mechanisms exist for the retention of NSTs in the ER.
One of these alternative mechanisms may be protein complex formation as it has been demonstrated recently in the case of UGT1 (Sprong et al., 2003). Although UGT1 normally is localized exclusively in the Golgi, it can be detected in the ER when coexpressed with cer-GalT, a member of the UDP-glycosyltransferase gene superfamily (Mackenzie et al., 1997), which by itself harbors a C-terminal dilysine motif (Nilsson et al., 1989; Zerangue et al., 2001) and localizes to the ER (Sprong et al., 1998). Using coimmunoprecipitation experiments the authors found physical interaction between UGT1 and cer-GalT. Obviously, the cer-GalT, retained in the ER via the intrinsic dilysine motif, binds to the UDP-Gal transport protein and thus retains the protein in the ER. Because cer-GalT is not expressed in CHO cells (Sprong et al., 2003), and recombinant UGT1 fully migrates to the Golgi, it can be excluded that this second mechanism of retention exists in CHO cells.
So far, both splice variants of the UGT have only been cloned from man (Ishida et al., 1996), but EST sequences of both forms are also found in mouse, and the encoded proteins have identical C-termini as human UGT1 and 2. Moreover, northern blots of different mouse tissues and developmental stages always show two signals corresponding to the two splice variants (Ishida et al., 1999). Although human expression data have not been published, Kawakita and Ishida (2002) mentioned that northern blot analysis revealed that both forms are ubiquitously expressed in every human tissue so far examined. Also on a northern blot from CHO cells (Oelmann et al., 2001), the two mRNA signals can be distinguished, indicating that both forms can even be expressed within one cell line. Thus, there is until now no indication of differential expression of the two splice forms.
Materials and methods
Antibodies
Monoclonal anti-mouse antibody (mAb) M5 directed against the FLAG sequence MDYKDDDDK was from Sigma (St. Louis, MO). A rabbit antiserum against the catalytic domain of α-mannosidase II was a kind gift of Dr. K. Moremen, University of Georgia, Athens, and anticalnexin, directed against the cytoplasmic domain of the calnexin, was a kind gift of Dr. A. Helenius (Swiss Federal Institute of Technology, Zurich). Anti-mouse Ig-Cy3-conjugate was from Sigma and anti-rabbit Ig-ALEXA 488-conjugate from Molecular Probes (Eugene, OR).
Cell lines and plasmids
All transient transfections for imunostaining were done in the CHO mutant Lec8 (Deutscher and Hirschberg, 1986) (ATCC CRL 1737) a cell line with a defect in the UGT gene. Cells were maintained in alpha medium (Biochrome, Berlin, Germany) supplemented with 10% fetal calf serum, 2 mM L-glutamin, 100 units/mL penicillin and 100 µg/mL streptomycin. Cells were grown in a humidified atmosphere at 37°C and 5% CO2. Complementation experiments were carried out in Lec8 cells for UGTs and in CHO Lec2 (ATTC CRL 1736; Stanley and Siminovitch, 1977) for CST constructs.
Generation of N-terminally FLAG-tagged constructs of the hamster UGT and mouse CST in pcDNA3 has been described (Eckhardt et al., 1998; Oelmann et al., 2001) and named hamUGT2 and mCST in this study. cDNA clones for human UGT1 (IMAGp998D17289Q2, IMAGE:163552) and UGT2 (IMAGp998H0411479Q2, IMAGE:5191731) were obtained from RZPD (Berlin), a distributor of the I.M.A.G.E. Consortium [LLNL] cDNA clones (Lennon et al., 1996) and cloned into pcDNA3-FLAG (pCDNA3 [Invitrogen, Carlsbad, CA] with the FLAG sequence tag [GGTACCGCCACCATGGACTACAAGGATGATGATGATAAGGGATCC] cloned into the KpnI-BamHI sites). The fragments were amplified with oligonucleotide HB22 and HB23 for UGT1 and HB22 and HB24 for UGT2, digested with BamHI-XbaI and ligated in the corresponding sites of pcDNA3-FLAG. The resulting constructs have the N-terminal sequence MDYKDDDDKGSN (FLAG tag underlined) followed by the complete open reading frames of human UGT1 and UGT2 and named in this article hUGT1 and hUGT2. To introduce a point mutation to create K396A, hamUGT2, which has an N-terminal FLAG tag followed by an EcoRI site, was amplified using the primers SO23 and RK001 and subcloned in the same vector using EcoRI and XbaI (hamUGT2-K396A). A BamHI site was introduced in hamUGT2 by site directed mutagenesis using the oligonucleotides RK007 and RK008. In the resulting protein the C-terminus is changed from LTKVKGS into GSKVKGS (hamUGT2-BamHI). To create mCST-KVKGS, we the mCST was amplified with the oligonucleotide primers ME71 and ME42, and the resulting product was cloned via EcoRI and BamHI into hamUGT2-BamHI. The sequence KVKVS was replaced in hamUGT2-BamHI and mCST-KVKGS for KKNT by digesting with BamHI-XbaI and ligation of a linker composed of the oligonucleotides RK024 and RK025 (hamUGT2-KKNT and mCST-KKNT). All constructs were confirmed by sequencing. Table I summarizes the obtained constructs and highlights the amino acid sequences of the respective C-termini.
Used oligonucleotides: CTA TGG ATC CAA CAT GGC AGC GGT TGG GGC T(HB22), GTA GTC TAG AAT TGC TGC CAG CCC TCA CT(HB23), GTA GTC TAG AAT CCC AGC GGC TAG GAA C(HB24), CGT CTA GAC TAC GAA CCC GCC ACC TTG GTG(RK001), GCG AAT TCG CAG CGG TTG GGG TTG(SO23), CTG CCA AAG TTG GGA TCC AAG GTG AAG GG(RK007), CCC TTC ACC TTG GAT CCC AAC TTT GGC AG(RK008), GCG AAT TCG CTC AGG CGA GAG AA (ME71), GCG GAT CCC ACA CCA ATG ATT CTC TCT TTT(ME42), GAT CCA AGA AGA CTA ACT AGT(RK024), CTA GAC TAG TTA GTC TTC TTG(RK025). Restriction sites are underlined.
Indirect immunofluorescence and complementation studies
For immunofluorescence analysis, we seeded cells onto glass cover slips. Plasmids were transfected using Metafectene (Biontex, Munich, Germany). Cells were analyzed by immunofluorescence 2 days post transfection. Therefore, cells were fixed in 4% paraformaldehyde and permeabilized for 30 min with 0.1% saponin in phosphate buffered saline (PBS) containing 0.1% bovine serum albumin. After permeabilization, samples were incubated with the respective primary antibodies (anti-FLAG mAb M5 and anti-α-mannosidase II or anticalnexin) followed by incubation with anti-mouse-Cy3 and anti-rabbit-ALEXA 488. Slides were mounted in Mowiol (Calbiochem, Darmstadt, Germany) and analyzed under a Leica DM IRBE.
Complementation of Lec8 and Lec2 cells were essentially done as described (Eckhardt et al., 1996; Bakker et al., 2005) with the exception that the cells have been analyzed by flow cytometry instead of cell surface staining of adherent cells. For this, plasmids were transfected using Metafectene (Biontex, Munich, Germany). Two days post transfection cells were released from the plates by incubation with PBS/2 mM ethylenediamine tetra-acetic acid and incubated with monoclonal antibody L2-412 (Kruse et al., 1984) and anti-Rat-FITC as secondary antibody for Lec8 cells and monoclonal antibody 735 (Frosch et al., 1985) and anti-Mouse-FITC as secondary antibody for Lec2 cells. Per construct 30,000 cells were counted.
Acknowledgements
The authors would like to thank Drs. K. Moreman, University of Georgia, Athens, USA, A. Helenius, Swiss Federal Institute of Technology, Zürich, Switzerland, and B. Sodeik, Medizinische Hochschule, Hannover, Germany for the generous gifts of antibodies, and Dr. Detlef Neumann, MH-Hannover for help with flow cytometry. The research was financially supported by grants obtained from the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie.