GalT2 (UDP-Gal:GA2/GM2/GD2 β-1,3-galactosyltransferase) is a Golgi-resident type II membrane protein that participates in the synthesis of glycosphingolipids. The molecular determinants for traffic and localization of this and other glycosyltransferases are still poorly characterized. Considering the possibility that interactions with other proteins may influence these processes, in the present study we carried out a yeast two-hybrid screening using elements of the N-terminal domain of GalT2 as bait. In this screening, we identified calsenilin and its close homologue CALP (calsenilin-like protein), both members of the recoverin-NCS (neuronal calcium sensor) family of calcium-binding proteins. In vitro, GalT2 binds to immobilized recombinant CALP, and CALP binds to immobilized peptides with the GalT2 cytoplasmic tail sequence. GalT2 and calsenilin interact physically when co-expressed in CHO (Chinese-hamster ovary)-K1 cells. The expression of CALP or calsenilin affect Golgi localization of GalT2, and of two other glycosyltransferases, SialT2 (CMP-NeuAc:GM3 sialyltransferase) and GalNAcT (UDP-GalNAc:lactosylceramide/GM3/GD3 β1-4 N-acetylgalactosaminyltransferase), by redistributing them from the Golgi to the ER (endoplasmic reticulum), whereas the localization of the VSV-G (G-protein of the vesicular stomatitis virus) or the Golgin GM130 was essentially unaffected. Conversely, the expression of GalT2 affects the localization of calsenilin and CALP by shifting a fraction of the molecules from being mostly diffuse in the cytosol, to clustered structures in the perinuclear region. These combined in vivo and in vitro results suggest that CALP and calsenilin are involved in the trafficking of Golgi glycosyltransferases.
Golgi-complex resident glycosyltransferases are membrane proteins that glycosylate lipids and proteins using sugar nucleotides as donors. With a few exceptions, they are type II membrane proteins with a N-terminal domain that comprises a relatively short CT (cytoplasmic tail) and a TMR (transmembrane, uncleaved signal-anchor region), followed by a stem region and a lumenally oriented C-terminal domain that bears the catalytic site and the sugar-nucleotide-binding site [1–4].
The N-terminal domains are able to concentrate reporter proteins in the Golgi complex [5,6], an indication that they have the minimal molecular determinants necessary for ER (endoplasmic reticulum) exit, transport to the Golgi and retention in this organelle. In some cases, a role for the lumenal domain has also been observed [4,6,7].
Molecular dissection of the N-terminal domain highlighted a role for the CT in these events . The CT of several glycolipid glycosyltransferases participates in the selective concentration of these enzymes as cargo in vesicles that transport them from the ER to the Golgi. Mutational analysis revealed that a dibasic motif [R/K](X)[R/K] in the CTs, proximal to the TMR, is necessary for ER exit and the Golgi concentration of glycolipid glycosyltransferase. The [R/K](X)[R/K] motif binds to the small GTPase Sar1 [9,10] that initiates the formation of COPII (coatamer protein II) coats at ER exit sites [11,12]. The selective concentration of glycosyltransferases along sub-Golgi compartments is also influenced by the nature of their CTs, since they impose the corresponding sub-Golgi location to reporter proteins and also to other glycosyltransferases [13,14]. These observations lend support to the possibility that differential interactions of the CTs with protein partners at the cytoplasmic leaflet of the endomembrane system might influence their traffic and localization. Since glycosyltransferases interact physically forming multienzyme complexes [5,8,15], other members of the complex could passively follow the behaviour and dynamics of the one that interacts directly with these partners.
In the present study, we carried out a yeast two-hybrid screening to identify putative interactors with the CT of GalT2 (UDP-Gal:GA2/GM2/GD2 β-1,3-galactosyltransferase). We identified calsenilin, and its close homologue CALP (calsenilin-like protein). These proteins are structurally related to the recoverin-NCS (neuronal calcium sensor) family of calcium-binding proteins . Calsenilin, also named DREAM (dynorphin response element-antagonist modulator) and KChIP3 (Kv4 channel interacting protein), has been independently discovered as an interacting partner for presenilins , as a protein interacting with DRE  and finally as KChIP3 (see  for a review, ). CALP/KChIP4 was later identified as a presenilin and Kv4 channel interacting partner . Calsenilin and CALP have a diffuse cytoplasmic localization, but in the presence of their interaction partners they associate with membrane compartments within the cell [16,21]. KChIP3/DREAM is detected in Western blot as monomers, dimers and tetramers in nuclear extracts, and as a dimer and tetramer in the cytosolic fraction .
In the present study we show that CALP and calsenilin physically interact with GalT2 in CHO (Chinese-hamster ovary)-K1 cells; in addition, overexpression of either protein affects the ER-to-Golgi traffic of GalT2 and also of two other glycosyltransferases, SialT2 (CMP-NeuAc:LacCer/GM3 sialyltransferase) and GalNAcT (UDP-GalNAc:GM3/GD3 N-acetylgalactosaminyltransferase), but not of the Golgin GM130 or the VSV-G (G-protein of the vesicular stomatitis virus). Results indicate that CALP and calsenilin are involved in the trafficking of Golgi glycosyltransferases.
GalT2–HA–YFP (where HA is haemagglutinin and YFP is yellow fluorescent protein), SialT2–YFP and GalNAcT–YFP have been described previously . Briefly, they are pEYFP-N1- (Clontech Laboratories) based vectors containing the N-terminal domains of the transferases (residues 1–52 for GalT2, 1–27 for GalNAcT and 1–57 for SialT2) fused to YFP (Figure 1C). The chimaeric construct containing the thermosensitive VSV-G (ts045) fused to YFP was kindly provided by P. Keller (Max-Planck Institute, Dresden, Germany) . The Escherichia coli expression construct pET-11d His-Sar1wt was kindly provided by W. Balch (Scripps Research Institute, La Jolla, CA, U.S.A.). Human CALP and calsenilin cDNAs were amplified by PCR from the yeast two-hybrid positive clones and introduced into the mammalian expression vector pECFP-N1 (Clontech Laboratories) using the following oligonucleotides: oCALP01, 5′-ATCTCGAGGCCACCATGAATGTGAGGAGGGTGGAAAGC-3′; oCALP02, 5′-GTGGATCCCGAATACAATTTTCAAAGAGCTG-3′; oCals01, 5′-ATCTCGAGGCCACCATGCAGCCGGCTAAGGAAGTGACA-3′; and oCals02, 5′-GTGGATCCCGGATGACATTCTCAAACAGCTG-3′, which introduce a XhoI site at the 5′-ends, a BamHI site at the 3′-ends, and eliminate the stop codon (Figure 1C). For bacterial expression of CALP, the original pACT2-CALP clone was digested with BglII and sub-cloned into the pRSET-C vector (Invitrogen) to generate the His6–HA–CALP (Figure 1C).
Yeast two-hybrid analysis reveals an interaction of GalT2 with calsenilin and with CALP
Yeast two-hybrid screening
The screening was performed following the manufacturer's protocol for the Matchmaker Gal4 Two Hybrid System 2 (Clontech Laboratories). Briefly, the sequence corresponding to amino acids 1–11 of mouse GalT2 was sub-cloned into pAS2-1, to express the peptide fused to the Gal4 binding domain. The resultant plasmid pAS-GalT2 was introduced into the yeast strain CG1945, which was then transformed with a human brain cDNA library cloned into the pACT2 vector. Transformants (1.5×107) were plated on SD (selective drop-out)/−Leu/−Trp/−His agar medium containing 5 mM 3-AT (3-aminotriazole) and the colonies that grew in the selective medium were assayed for β-galactosidase activity. Plasmid DNA was prepared from candidate positive clones. A second round of co-transformation with bait plasmids was then performed to confirm the interaction. Clones that passed the confirmation tests were subjected to DNA sequencing and further analysis.
Cell culture and transfection
CHO-K1 cells were grown in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% fetal calf serum, 100 mg/ml penicillin and 100 mg/ml streptomycin. At approx. 70% confluence, cells were transfected with Lipofectamine™ (Invitrogen) and analysed 18 h after transfection. Proteolysis inhibitors (5 μM MG132, 60 μM chloroquine or 10 μM lactacystin) were added 5 h before harvesting.
CHO-K1 cells from a 10-mm Petri dish were rinsed and gently scraped in PBS plus PI (protease inhibitors; 5 μg/ml aprotinin, 0.5 μg/ml leupeptin and 0.7 μg/ml pepstatin). The cells were collected by centrifugation (1000 g for 5 min at 4 °C), resuspended in 600 μl of homogenization buffer [250 mM sucrose and 3 mM imidazole (pH 7.4)], plus PI and centrifuged at 1200 g for 10 min. The pellet was then re-suspended in 300 μl of homogenization buffer containing 0.5 mM EDTA and passed 20 times through a 25-gauge needle. A post-nuclear supernatant was obtained by sequential centrifugation at 1200 g (for 10 min) and 1500 g (for 5 min). This post-nuclear supernatant was adjusted to 50% (w/v) sucrose. A 300 μl aliquot of this suspension was loaded on to a 1.7 ml 10–60% sucrose gradient in 3 mM imidazole and 0.5 mM EDTA. The gradient was centrifuged at 40000 rev./min for 130 min at 4 °C using the Optima™ TLX Ultracentrifuge with a TLV 100 rotor (Beckman). After centrifugation, 12 fractions were collected from the bottom. Proteins in each fraction were precipitated by the addition of TCA (trichloroacetic acid) to a final concentration of 10%. The protein pellet was collected by centrifugation at 15000 g for 10 min, and 15 μl of 1 M Tris/HCl (pH 7.4) was added to neutralize excess TCA. Then 5 μl of 4× sample buffer was added and the samples were analysed by SDS/PAGE and Western blot.
Immunoprecipitation and in vitro binding assays
CHO-K1 cells were harvested in PBS and collected by low-speed centrifugation (1000 g for 3 min at 4 °C). They were then resuspended in lysis buffer [50 mM Tris/HCl (pH 7.2), 1% (v/v) Triton X-100, 150 mM NaCl, 1 mM PMSF and 1 mM EDTA plus PI] and passed 10 times through a 25-gauge needle. Lysates were cleared by centrifugation at 13000 g for 10 min and the supernatant was used for immunoprecipitation and binding assays. When immunoprecipitation from membrane-enriched fractions was required, Triton X-100 was omitted from the lysis buffer, lysates were centrifuged at 300 g for 5 min, the supernatant was re-centrifuged for 15 min at 13000 g and the pellet was resuspended with lysis buffer plus 1% Triton X-100. GalT2–HA–YFP  was pulled down using an anti-HA rabbit polyclonal antibody, covalently bound to Protein A–Sepharose beads (GE Healthcare) using DMP (dimethyl pimelinediimidate dihydrochloride) as a cross-linker . Calsenilin–CFP (where CFP is cyan fluorescent protein) was pulled down using a rabbit anti-DREAM antibody followed by Protein A–Sepharose. After a 2 h incubation at 4 °C, beads were washed three times with lysis buffer, once with PBS and heated at 50 °C with 4× sample buffer without reducing agents. The eluate was analysed by Western blot.
For CALP–Sepharose bead preparation, recombinant His6–HA–CALP in binding buffer [500 mM NaCl and 20 mM Tris/HCl (pH 7.9)] was incubated with 60 μl of 50% Ni2+-charged Sepharose His-bind at 4 °C for 2 h and then washed three times with binding buffer to remove the unbound material. Lysates of CHO-K1 cells expressing GalT2–HA–YFP were incubated with CALP–Sepharose beads at 4 °C for 2 h, washed three times with binding buffer containing 1% (v/v) Triton X-100, and three times with washing buffer [500 mM NaCl, 20 mM Tris/HCl (pH 7.9) and 1% (v/v) Triton X-100]. The bound proteins were eluted with 25 μl of elution buffer [500 mM NaCl, 100 mM EDTA and 20 mM Tris/HCl (pH 7.9)]. The eluted samples were subjected to SDS/PAGE and Western blot analysis.
Coupling of GalT2 CT peptides to Sepharose 6B was carried out according to the manufacturer's protocol (Sigma–Aldrich). The efficiency of coupling was monitored by measuring 2-thiopyridone release at 343 nm. For binding assays, 15 nmol of Sepharose-bound peptides was used.
The bacterial lysates containing recombinant proteins were prepared as follows: 10 ml of a bacterial culture (D600=0.6) bearing appropriate plasmids was induced with 1 mM IPTG (isopropyl β-D-thiogalactoside; Promega) for 5 h at 37 °C. The cells were collected and resuspended in 1 ml of 50 mM Tris/HCl (pH 8), 100 mM NaCl, 1 mM PMSF and 200 μg/ml lysozyme (Sigma–Aldrich). The resuspension was incubated for 30 min at room temperature (25 °C) with gentle agitation, sonicated to break the cells open, and then centrifuged at 10000 g in a microcentrifuge for 10 min at 4 °C.
For the binding experiments, the bacterial lysates containing recombinant CALP or His6–Sar1 were incubated with the immobilized GalT2 CT peptides for 1 h at 4 °C [9,24]. The beads were collected, washed five times and eluted with Laemmli sample buffer for Western blot analysis with an anti-HA or an anti-Sar1 antibody.
Fluorescence microscopy and Western blots
Cells grown on coverslips were fixed for 7 min in methanol at −20 °C, incubated with the specific antibody and fluorescently labelled secondary antibodies (see below). The coverslips were mounted with FluorSave (Calbiochem) and observed in an Olympus FV 1000 confocal microscope with a 100× planapochromat oil-immersion objective and appropriate filters for CFP, YFP, rhodamine and FITC. The quantification of calsenilin–CFP present in perinuclear structures upon co-transfection with glycosyltransferases was made using Metamorph 4.5 Imaging System (Universal Imaging Corporation) software. The structures were considered positive when their intensity was at least 2-fold higher than the average fluorescence intensity of the cell. Structures smaller than 50 nm were excluded from the analysis (n=20 for each case).
For Western blot analysis, membranes were blocked with 5% (w/v) non-fat dried skimmed milk in PBS for 1 h at room temperature, followed by three 5 min washes with PBS, and reacted with appropriate specific antibodies. The following antibodies were used: mouse monoclonal antibody anti-GFP (1:1500; Roche Applied Science), mouse monoclonal antibody anti-GM130 (1:200; BD Biosciences), rabbit polyclonal anti-DREAM antibody (1:500; Santa Cruz Biotechnology), rabbit polyclonal antibody anti-calreticulin (1:1000; Affinity BioReagents) and rabbit anti-calnexin (1:300; Santa Cruz Biotechnology). Fluorescently labelled secondary antibodies were goat anti-rabbit or goat anti-mouse coupled to Alexa Fluor® 546 or Alexa Fluor® 660 (each 1:1000; Molecular Probes). Blots were probed using secondary antibodies coupled to IRDye 700 (anti-rabbit) or IRDye 800 (anti-mouse) (LI-COR Biosciences) at 1:20000 and then scanned using an Odyssey IR Imager (LI-COR Biosciences).
Yeast two-hybrid screening identifies calsenilin and CALP as GalT2-interacting proteins
To identify GalT2-interacting proteins, a human brain cDNA library was screened using a Gal4-based yeast two-hybrid system (Matchmaker System 2, Clontech Laboratories). As bait, we used a fusion of the Gal4-binding domain to the MPLSLFRRVLL peptide, which comprises the GalT2 CT plus three amino acids (VLL), which are predicted to be part of the TMR .
Four positive clones were isolated by their ability to grow on minimal medium containing 5 mM 3-AT and lacking histidine, and by their β-galactosidase activity (Figure 1A). One clone contained the full-length sequence of calsenilin, two clones contained the full-length sequence of CALP and the fourth contained a truncated version of CALP lacking the first 23 amino acids.
CALP and GalT2 interact in vitro
The lysates of CHO-K1 cells expressing a fusion of the GalT2 N-terminal domain to an HA epitope and to YFP (Figure 1C)  were incubated with recombinant His6–HA–CALP (Figure 1C) immobilized in Ni2+-charged Sepharose. GalT2–HA–YFP was retained by CALP–Sepharose beads, but not by glucogenin–Sepharose beads, an unrelated protein used as a negative control (Figure 2A). A positive control for this methodology was the binding of GalT2 to Sar1–Sepharose beads, as previously described  (Figure 2A). In an alternative protocol, a synthetic peptide with the sequence of the GalT2 CT was covalently coupled to thiopropyl–Sepharose 6B and incubated with a lysate from bacteria expressing His6–HA–CALP. We used fresh bacterial lysates rather than purified CALP because we found it to aggregate rapidly after purification. Figure 2(B) shows that recombinant CALP is specifically bound to the GalT2 CT peptide (top panel, WT). The binding of CALP was not dependent on the presence of the Arg–Arg motif present in these peptides, since replacement of these residues by alanines (top panel, RR-AA) did not affect it, whereas the binding of Sar1 was affected (Figure 2B, lower panel, RR-AA), as reported previously .
CALP and GalT2 interact in vitro
Calsenilin interacts with GalT2 in CHO-K1 cells
In attempts to co-immunoprecipitate GalT2 and calsenilin, we noticed that co-transfection of calsenilin–CFP and GalT2–HA–YFP resulted in a substantial decrease in the levels of expressed calsenilin–CFP whereas the levels of GalT2 remained essentially unchanged (Figure 3A, compare lanes 1 and 2 with lane 3). This was observed by Western blot analysis and also under the fluorescence microscope, where only a few cells showed detectable levels of calsenilin–CFP in GalT2–HA–YFP co-transfected cells (results not shown). To check whether the influence of GalT2–HA–YFP on calsenilin–CFP expression was exerted at the level of its stability, co-expressing cells were treated with inhibitors of proteasomal and lysosomal activities. The proteasome inhibitors MG132 and lactacystin restored calsenilin–CFP to approx. 50% of its level in non-cotransfected cells. The lysosome inhibitor chloroquine was less effective (Figure 3A). The co-expression of calsenilin–CFP with another membrane protein, VSV-G–YFP, did not affect calsenilin–CFP stability, since its level was comparable in single and double transfectant cells (Figure 3B). These results suggest a functional interaction between calsenilin and GalT2. We next assessed whether a physical interaction between calsenilin and GalT2 could be shown by co-immunoprecipitation. Experiments were carried out using membrane-enriched fractions from cells co-expressing GalT2–HA–YFP and calsenilin–CFP in the presence of MG132 to increase the calsenilin content of the cells (see Figure 3A). A membrane-enriched fraction was used because, although calsenilin and CALP are mostly soluble proteins, it has been reported that they associate to the ER and Golgi membranes upon overexpression of membrane-bound partners such as the Kv4 channel and presenilin [17,20,25]. Under these conditions, a fraction of calsenilin–CFP disappeared from the lysate after immunoprecipitation with anti-HA beads and appeared co-immunoprecipitated with GalT2–HA–YFP (Figure 3C, right-hand side). No calsenilin–CFP was detected when the identical immunoprecipitation assay was carried out with lysates from cells transfected with calsenilin–CFP alone (Figure 3C, left-hand side). Consistent results were obtained by immunoprecipitating calsenilin with an anti-DREAM antibody and probing for the presence of GalT2–HA–YFP in the immunoprecipitate with an anti-HA antibody (Figure 3D).
Calsenilin expressed in CHO-K1 cells interacts with co-expressed GalT2
Co-expressed calsenilin and glycosyltransferases mutually affect their subcellular distribution
Cells expressing calsenilin–CFP alone showed fluorescence distributed in the cytoplasm and within the nucleus (Figure 4A, panel a), whereas cells expressing GalT2–HA–YFP alone showed fluorescence in the Golgi complex (Figure 4A, panel b). However, when both proteins were co-expressed, they mutually affected their subcellular distribution: calsenilin–CFP changed its broad, rather uniform distribution to a more patchy one, with some clear dots in the perinuclear region (Figure 4A, panel c). GalT2–HA–YFP is now present both in Golgi-like perinuclear structures and in the ER, as shown by its abundant co-localization with the ER marker calnexin (Figure 4A, panel e and the merged image is shown in panel f). The relocation trend observed for glycosyltransferases was not a generalized effect on Golgi proteins, since the subcellular localization of the Golgin GM130 was essentially unaffected in cells overexpressing calsenilin–CFP (Figure 4A, panel i). The same mutual influences in subcellular distribution were observed when calsenilin–CFP was co-expressed with SialT2–YFP (Figure 4A, panels k and l) or GalNAcT–YFP (Figure 4A, panels m and n). These results are in line with previous reports that calsenilin changes its localization to membranous compartments upon overexpression of their partners [17,20,24]. The calsenilin–CFP redistribution shown in Figure 4(A) was quantified by determining the amount of its fluorescence associated to perinuclear structures; as shown in Figure 4(B), in calsenilin–CFP expressing cells that co-express GalT2–HA–YFP, GalNAcT–YFP or SialT2–YFP, the percentage of calsenilin–CFP in these structures was between 2.5- and 4-fold higher than in cells that express calsenilin–CFP alone or co-express it with VSV-G–YFP (see below and Figure 4C). CALP–CFP produced essentially the same effect as calsenilin–CFP; also, although images of Figure 4(A) were taken from the few cells that had detectable levels of calsenilin–CFP in the absence of MG132, results were similar in cells treated with the proteasomal inhibitor MG132 (results not shown).
Calsenilin and glycosyltransferases mutually affect their ER-Golgi distribution
To examine whether calsenilin affects the intracellular localization of other membrane-bound proteins of the secretory pathway, we determined the fate of the thermosensitive mutant VSV-G–YFP (ts045) in cells expressing calsenilin–CFP. In this case, the mutual influences between calsenilin and glycosyltransferases in their subcellular distribution (Figure 4A) were not observed, the subcellular distribution of ts045 essentially being the same whether or not calsenilin–CFP was co-expressed (Figure 4C).
The use of ts045 allows us to investigate some aspects of its intracellular trafficking. At 39 °C this protein has a folding defect that affects its ER exit; lowering the temperature to 32 °C allows it to exit the ER and to progress along the secretory pathway, concentrating in the Golgi and the plasma membrane . We found that at 39 °C ts045 is mainly associated to intracellular structures, consistent with its reported ER and Golgi localization at that temperature (Figure 4C, top left-hand side); this behaviour was essentially the same in cells that also express calsenilin–CFP (Figure 4B, top centre and right-hand side). After transferring the cells to 32 °C for 2 h (bottom row), the expected appearance of VSV-G–YFP at the plasma membrane was observed, and again this transport step was not affected in cells that also express calsenilin–CFP (Figure 4C, bottom row, compare left-hand side with centre and right-hand side). This indicates that trafficking of ts045 VSV-G–YFP along the endomembrane system is not obviously affected by co-expressed calsenilin–CFP. Similar results were obtained when CALP–CFP was used instead of calsenilin–CFP. The effect of calsenilin–CFP expression on redistribution of GalT2–HA–YFP observed under the fluorescence microscope and shown in Figure 4(A) (2nd and 3rd rows) was confirmed by sucrose-gradient separation of the membranes from these cells (Figure 5). In single transfectant cells, GalT2–HA–YFP-containing membranes showed a fractionation trend comparable with Golgi membranes marked by the Golgi marker GM130 (fractions 5–10). On the other hand, when co-expressed with calsenilin–CFP, GalT2–HA–YFP also fractionated with membranes floating along denser fractions of the gradient, co-distributing with membranes marked by the ER marker calreticulin. Calsenilin–CFP itself was also found in dense fractions, but it should be noted that this may correspond in part to the soluble, non-membrane-associated fraction.
Calsenilin affects subcellular distribution of GalT2
In the present study we have identified calsenilin and CALP as proteins interacting with the peptide MPLSLFRRVLL, which comprises the GalT2 N-terminal CT plus three amino acids (VLL), which are predicted to be part of the cytoplasmic half of the TMR . CALP also binds to a short 10-amino-acid peptide in the C-terminus of PS2 (presenilin 2) . Sequence alignment of GalT2 CT and the region of the PS2 C-terminus used in the original two-hybrid screening that identified CALP shows that the amino acidic sequence in that region of PS2 (L398AIFKKAL405) is homologous with the sequence of the GalT2 CT peptide used as bait (L3SLFRRVL10). However, deletion analysis did not point to this region being as responsible for the binding in PS2 . The region bearing the EF domains mediates the interaction between CALP and PS2, independently of the CALP N-terminus . Our two-hybrid results are consistent with this, since we found a CALP clone lacking 23 amino acids from the N-terminus as a positive interactor of GalT2.
Experiments in vitro showing that GalT2 from cell lysates is able to bind to immobilized CALP, and that CALP binds to an immobilized GalT2 CT peptide, reinforces the evidence for the interaction. CALP binding to immobilized peptide with the GalT2 CT sequence was independent of the presence of the Arg–Arg residues that specify binding to Sar1, indicating that Sar1 and CALP and possibly calsenilin, have somewhat different requirements for binding to GalT2 CT. This is also shown by the fact that calsenilin and CALP do not physically interact, at least directly, with the GalNAcT CT (results not shown), whereas Sar1 does . Co-expression of CALP or calsenilin with the N-terminal domain of GalT2 in CHO-K1 cells results in an interaction that is shown by the accelerated proteolysis of calsenilin via the proteasomal pathway and is demonstrated by co-immunoprecipitation with GalT2. Calsenilin and CALP are calcium-binding proteins, but we have not observed significant effects of calcium in either in vitro or in vivo binding experiments. Binding between calsenilin and PS2 has also been shown to be calcium independent .
CALP or calsenilin expression affects ER/Golgi distribution of GalT2, GalNAcT and SialT2, but not of the Golgin GM130 or VSV-G, a type I membrane protein, indicating some specificity of the effect on glycosyltransferases rather than a generalized effect on membrane protein dynamics. Conversely, GalT2, GalNAcT and SialT2 affect the subcellular distribution of calsenilin. Although results from the in vitro binding studies do not evidence a direct interaction of GalNAcT and SialT2 with calsenilin and CALP, the mutual effects on their distribution in vivo can be explained by the fact that GalNAcT and GalT2 are able to interact with each other, and they do so also in the ER [5,8]. SialT2 has also been found as an interactor of GalNAcT in F-11 cells . Thus the possibility that re-localization of GalT2 influences the localization of its partners is worthy of consideration. In any case, our results are in line, and reinforce, the suggestion that interactions of calsenilin and other KChIPs with membrane protein complexes regulate their metabolism, traffic and/or function . The localization of calsenilin itself is influenced by the presence of GalT2. A fraction of the molecules are redistributed to the perinuclear region co-localizing with the fraction of GalT2 that is not redistributed to the ER.
The exit of proteins from the ER is a selective process that leads to signal-mediated concentration as cargo in COPII transport vesicles. For the case of Golgi glycosyltransferases, a [R/K](X)[R/K] motif in the CT, close to the transmembrane domain, interacts with Sar1 , which in its GTP form binds to ER membranes and recruits other proteins of the COPII coat (Sec 23/24, Sec 13/31) promoting formation of ER exit sites. Transport vesicles generated from these sites convey the proteins to the Golgi . At steady state the concentration of glycosyltransferases in Golgi membranes relative to ER membranes was calculated to be 9:1 . It has been proposed that Golgi glycosyltransferases constitutively cycle through the ER, the final Golgi localization being the result of the balance between the rates of forward transport from the ER and retrograde transport from the Golgi . However, this claim has been recently challenged, by proposing that under physiological conditions, proteins constantly remain in the Golgi . In any case, the redistribution of GalT2 to the ER that we observe as a consequence of calsenilin expression could arise from delayed ER exit of newly synthesized or of recycled GalT2. It is possible that overexpression of calsenilin may interfere with the interaction of the GalT2 CT with Sar1, thus decreasing the ER exiting of GalT2 and driving a new steady-state balance more favourable to the ER and less favourable to the Golgi complex.
Calsenilin, CALP and Sar1 are the only proteins identified so far as interacting with the CT of some glycolipid glycosyltransferases. Although further studies are necessary to interpret in molecular terms the interaction of calsenilin and CALP with GalT2, a competence between Sar1 and calsenilin for GalT2 binding at ER exit sites may modulate the exit of GalT2 from the ER exit and hence its traffic to the Golgi.
This work was supported in part by Grants from Howard Hughes Medical Institute of the U.S.A., Secretaría de Ciencia y Tecnología, Universidad Nacional de Córdoba, Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina, Agencia Nacional de Promoción Científica y Tecnológica, Argentina (to H. J. F. M.). The technical assistance of G. Schachner and S. Deza with cell cultures and of C. Mas and S. Dehner with confocal microscopy is also acknowledged. C. A. Q. and M. L. F. are recipients of CONICET Fellowships. H. J. F. M. and J. V.-T. are Career Investigators of CONICET (Argentina).
cyan fluorescent protein
coatamer protein II
dynorphin response element-antagonist modulator
UDP-GalNAc:lactosylceramide/GM3/GD3 β1-4 N-acetylgalactosaminyltransferase
UDP-Gal:GA2/GM2/GD2 β-1,3 galactosyltransferase
Kv4 channel interacting protein
G-protein of the vesicular stomatitis virus
yellow fluorescent protein