| To whom correspondence should be addressed: Nobuhiro Nakamura, Department of Molecular Biosciences, Faculty of Life Sciences, Kyoto Sangyo University, Motoyama, Kamigamo, Kita, Kyoto 603-8555, Japan. Tel/Fax: +81–75–705–3018 E-mail: osaru3@cc.kyoto-su.ac.jp Abbreviations: YIPF, Yip domain family; ERGIC, ER-Golgi intermediate compartment; TGN, trans-Golgi network; BN-PAGE, blue native PAGE; GlcNAcTI, N-acetylglucosaminyltransferase I; GFP, green fluorescent protein; VSV-G, vesicular stomatitis virus glycoprotein; sDPPIV, a soluble form of dipeptidyl-peptidase IV. |
Yip1p and Yif1p were first identified in budding yeast as interacting proteins for the Ypt/Rab family small GTPases and were shown to play an essential role in ER to Golgi transport (Matern et al., 2000; Yang et al., 1998). Yip1p and Yif1p are multi-span transmembrane proteins forming a complex and have been proposed to regulate vesicle fusion at the Golgi apparatus (Matern et al., 2000). However, it remains controversial whether they regulate vesicle fusion at the Golgi apparatus (Barrowman et al., 2003) or vesicle budding from the ER (Heidtman et al., 2003, 2005). Budding yeast has two other family members (Yip4p and Yip5p) that interact with Ypt/Rab GTPases. These are non-essential gene products and their functions remain obscure (Calero et al., 2002). Interactome analyses in budding yeast revealed that the Yip1p/Yif1p family proteins interact with a range of gene products that function in the vesicular transport pathway, including SNAREs, Ypt/Rab GTPases, ARFGAP and sorting nexins (Ito et al., 2000; Uetz et al., 2000) (see http://www.yeastgenome.org/ for an update). Yip1p binds to di-geranylgeranylated Rab GTPases including Ypt1p but not to mono-prenylated Rab GTPases (Calero et al., 2003). Therefore, Yip1p has been proposed to function as a specific membrane acceptor for the di-geranylgeranylated Rab proteins. On the other hand, Yip1p and Yif1p interact with Yip3p, a multi-span transmembrane protein similar to Yip1p/Yif1p (Calero and Collins, 2002). PRA1, a mammalian homologue of Yip3p is shown to function as a GDP dissociation inhibitor-displacement factor (GDF) for Rab9 (Sivars et al., 2003). These findings strongly suggest that Yip1p/Yif1p complex function in close relationships with a specific subset of Rab GTPases.
Phylogenic analysis of Yip1p and Yif1p indicates that these belong to a family of multi-span transmembrane proteins, most probably with five transmembrane segments, and are highly conserved in eukarya (Shakoori et al., 2003). The available protein sequence data indicates that there are nine family members in humans (FinGER1~9: HGNC names: YIP domain family (YIPF) member 1~7 [YIPF1~7], YIF1A and YIF1B), and the existence of these nine members is conserved at least in mammals (Shakoori et al., 2003; Yoshida et al., 2008). Among these mammalian Yip1p/Yif1p family members, YIPF5 (Yip1A), the predicted orthologue of budding yeast Yip1p, is shown to interact with COPII and has been proposed to function in the cargo export from the ER (Tang et al., 2001). Consistently, we found that human YIPF5 (Yip1A) and YIF1A, the predicted orthologues of budding yeast Yif1p, form a complex, mainly localize in the cis-Golgi and the ERGIC, and recycle between the ER and the Golgi apparatus (Yoshida et al., 2008). However, the knockdown of YIPF5 or YIF1A does not affect anterograde transport of transmembrane or soluble marker proteins while it causes significant fragmentation of the Golgi apparatus (Yoshida et al., 2008). In addition, it has recently been shown that the knock down of YIPF5 (Yip1A) causes delay in the Golgi to the ER retrograde transport of COPI independent cargo molecules while not affecting the anterograde transport and COPI dependent Golgi to ER retrograde transport pathway (Kano et al., 2009). Therefore, whether YIPF5 (Yip1A) functions in anterograde transport remains uncertain. Interestingly, YIF1B, a closest homologue of YIF1A, has been shown to interact and support the specific targeting of serotonin receptor (5-HT1A) to neuronal dendrites (Carrel et al., 2008). This result implies that at least YIF1B, and possibly other Yip1p/Yif1p family proteins as well, plays a role in the anterograde transport of a limited subset of cargo molecules from the ER to the Golgi and/or from the Golgi to downstream compartments.
We have been analyzing human Yip1p/Yif1p family proteins (Hereafter, called YIPF proteins, collectively) and are particularly interested in YIPF3 and YIPF4 because they are most strictly localized in the cis-Golgi among the family members. Here, we report the characterization of YIPF3 and YIPF4, paralogues of Yif1p and Yip1p, respectively, and show that they are involved in the maintenance of the Golgi structure.
These were performed as described previously (Yoshida et al., 2008).
Site-specific single strand mutagenesis was performed with suitable primers (Kunkel et al., 1991).
Antibodies were produced and affinity purified as described previously (Yoshida et al., 2008). Rabbit anti-rat GRASP65 (His6-GRASP65) antibody (Yoshimura et al., 2005), anti-human GM130 (GST-human GM130) (Sohda et al., 1998), anti-human golgin-84 (Misumi et al., 2001) was previously produced in our laboratory. Mouse anti-ERGIC53 antibody was kindly provided (Schweizer et al., 1988). The following antibodies were purchased: mouse monoclonal anti-GM130 antibody (BD Biosciences Pharmingen, San Diego, CA, USA), anti-FLAG M5 (Sigma-Aldrich, St. Louis, MO, USA), anti-HA (16B12) (Covance Research Products, Inc., Berkeley, CA, USA), anti-HA (3F10) (Roche Diagnostics GmbH, Penzberg, Germany), and anti-calnexin (Stressgen Bioreagents, Victoria, Canada).
These experiments were performed as described previously (Yoshida et al., 2008). For detection with anti-FLAG (M5), antibody diluted in blocking solution was pre-incubated with 10% volume of purified rat liver cytosol (Hui et al., 1998) and a protease inhibitor cocktail (Yoshimura et al., 2001) for 30 minutes at room temperature. Densitometry was performed with ImageJ (version 1.38x, National Institute of Health, Bethesda, USA) using digital image data exported from LAS1000 (Fuji Photo Film. Inc, Tokyo, Japan). Area of form I and II or form III was selected with a square and mean density was measured. Same surface area was used for each lane. Relative amount of form III (%) was calculated by the following equation: (mean density of form III)×(surface area of form III)/((mean density of form I+II)×(surface area of form I+II)+(mean density of form III)×(surface area of form III)).
Radio-labeling was performed as described previously with slight modifications (Yoshimura et al., 2001). A total of 5×105 cells were seeded on a 3.5-cm diameter dish and used for labeling with 5.3 MBq Pro-Mix (Amersham Biosciences). The cell lysate was split into two equal volumes and incubated with pre-immune IgG or affinity purified anti-YIPF3 antibody and protein A beads for 16 hours. The beads were then collected and washed with RIPA buffer three times. Bound materials were analyzed by SDS-PAGE followed by autoradiography using a BAS-1800 image analyzer (Fujifilm, Tokyo, Japan). Protein deglycosylation was performed using an enzymatic protein deglycosylation kit (Sigma-Aldrich) following the manufacturer’s protocol.
These experiments were performed as described previously (Shakoori et al., 2003; Yoshimura et al., 2004). Briefly, HeLa cells were transfected with an expression plasmid (pcDNA3 with FLAG and HA tags) for YIPF3/FinGER3 or YIPF4/FinGER4 (Shakoori et al., 2003) and selected by G418.
This was performed as described previously (Yoshida et al., 2008). The siRNA sequence were: YIPF3 (sense: 5'-GGGGAGGGUUCGAAGAAAACAUCCAAG-3', antisense: 5'-UGGAUGUUUUCUUCGAACCCUCCCCAU-3), YIPF4 (5'-GUGGUGAGAGACAAUCCUGTT-3', antisense: 5'-CAGGAUUGGUCUCUCACCACTT-3). The sequence of the control siRNA was described previously (Yoshida et al., 2008).
The cells were fixed in PLP fixative, permeabilized with saponin, and stained by an affinity-purified antibody as described previously (Uchiyama et al., 2002).
We have previously shown that epitope-tagged YIPF3 and YIPF4 are localized at the Golgi apparatus (Shakoori et al., 2003). To determine the precise location of the proteins in the Golgi sub-compartment, anti-YIPF3 and anti-YIPF4 antibodies were produced to analyze the localization of endogenous proteins. Anti-YIPF4 serum recognized one band at the 29 kDa position in HeLa, NRK and rat liver Golgi lysates by western blotting (Fig. 1, lanes 5–7). The estimated size corresponded to the predicted size from the cDNA sequence (27 kDa), suggesting that the 29 kDa band was YIPF4. Occasionally, a minor, slightly lower, mobility band was detected; this was particularly evident in the rat liver Golgi, suggesting that YIPF4 may receive a minor post-translational modification. On the other hand, even after affinity purification, the anti-YIPF3 antibody recognized two bands, at the 36 kDa (III) and 46 kDa (II) positions, in HeLa, NRK and rat liver Golgi lysates (Fig. 1, lanes 1–3). Heavy exposure revealed a third band at the 40 kDa (I) position in the HeLa cell lysate (Fig. 1, lane 4). The predicted size of YIPF3 was 38 kDa, suggesting that YIPF3 received major posttranslational modifications (detailed characterization described below). The specificity of the antibodies was later confirmed by the knockdown of YIPF3 or YIPF4 by siRNA treatment, in which the signals of the above-described bands were strongly reduced (Fig. 8A).
 View Details | Fig. 1. Characterization of anti-YIPF3 and YIPF4 antibodies. Antibodies were produced using purified recombinant human YIPF3 and YIPF4 proteins. Lysates were prepared from HeLa cells (lanes 1, 4 and 5), NRK cells (lanes 2 and 6) and purified rat liver Golgi membranes (lanes 3 and 7) and analyzed by western blotting using the indicated antibodies. An enhanced darker picture shows a third band in the HeLa cell lysate (lane 4). The positions of the molecular weight markers (kDa) are shown on the left. The specific bands detected by the antibodies are indicated by arrows on the right. |
Using our highly specific antibodies for YIPF3 and YIPF4, we then carried out immunofluorescence staining of HeLa cells. Both antibodies showed a clear Golgi staining pattern that almost completely co-localized with GM130, a widely used cis-Golgi marker protein (Fig. 2A, upper panels on upper and lower galleries and Fig. S1, upper galleries) while they were mostly segregated from TGN46 (Fig. S1, lower galleries) (Nakamura et al., 1995). ERGIC-like cytoplasmic structures were not stained by anti-YIPF3 (upper gallery) or YIPF4 (lower gallery) antibody, in contrast with YIPF5 and YIF1A, the majority of which localize in the ERGIC (Yoshida et al., 2008). Localization of YIPF3 and YIPF4 in the medial- and trans-Golgi was evaluated by Brefeldin A treatment (Fig. 2A and Fig. S1). It is known that after Brefeldin-A treatment most of the medial- and trans-Golgi resident proteins are relocalized to the ER while cis-Golgi/ERGIC resident proteins are retained in the cytoplasmic punctate structures, so called “Golgi remnants” (Lippincott-Schwartz et al., 1989; Misumi et al., 1986; Nakamura et al., 1995). Accordingly, YIPF3 and YIPF4 were well colocalized with GM130 in the cytoplasmic punctate structures even after Brefeldin-A treatment (Fig. 2A, lower panels, Fig. S1, upper gallery) while they were clearly segregated from TGN46 (Fig. S1, bottom gallery). These results strongly suggested that YIPF3 and YIPF4 are localized in the cis-Golgi and not in the medial-, trans-Golgi or trans-Golgi network (TGN).
 View Details | Fig. 2. YIPF3 and YIPF4 localize at the cis-Golgi. (A) HeLa cells were fixed and stained with affinity purified anti-YIPF3 (upper gallery) or anti-YIPF4 (lower gallery) antibody together with anti-GM130 antibody before (lower panels) or after (upper panels) Brefeldin A treatment for 30 minutes. Bar=20 μm. (B) HeLa cells were fixed and processed for immuno-gold silver enhancement staining using anti-YIPF3 (upper panel) or anti-YIPF4 (lower panel) antibody. Arrows indicated silver grains. Arrowheads indicated clathrin coated vesicles. Bars=1 μm. |
The cis-Golgi localization of YIPF3 and YIPF4 was further evaluated at the ultra-structural level by immuno-electron microscopy. As shown in Fig. 2B, silver grain deposition were found specifically over the Golgi stacks. Overall, the distribution patterns were very similar between YIPF3 and YIPF4. There was clear accumulation on one side of the Golgi stack, most probably the cis-side of the Golgi apparatus (arrows) because clathrin coated vesicles were found mainly on an opposite side of the stack (arrowheads).
We then performed cell fractionation analysis to confirm the localization of YIPF3 and YIPF4. HeLa cells were homogenized and subjected to Nycodenz density gradient centrifugation. As previously reported (Yoshida et al., 2008; Yoshimura et al., 2004), the Golgi markers GM130, GRASP65 and golgin84 were distributed in the lighter fractions, peaking in fraction 6 (Fig. 3). On the other hand, the ER marker calnexin was distributed in the heavier fractions, peaking in fraction 10. ERGIC53 showed two distribution peaks, one in the lighter fractions (fraction 5) and one in the heavier (fraction 10) fractions. YIPF4 and the two major forms of YIPF3 (II and III) were distributed in the lighter fractions with one peak, similar to the Golgi markers. These results indicate that YIPF3 and YIPF4 are concentrated in the Golgi fractions, not the ER or the ERGIC fractions, in contrast with YIPF5 or YIF1A, the majority of which are recovered in the ER/ ERGIC fractions (Yoshida et al., 2008; Fig. 3). Interestingly, a significant amount of form I of YIPF3 was found in the heavier ER/ERGIC fractions (fraction 8–10) suggesting the localization of form I in the ER/ERGIC (discussed below).
 View Details | Fig. 3. Subcellular distribution of YIPF3 and YIPF4. (A) HeLa cells were homogenized and post-nuclear supernatant was prepared. This was then fractionated using a Nycodenz density gradient and analyzed by western blotting using the indicated antibodies. (B) The bands in (A) were densitometrically quantified and the presence in each fraction (%) is presented as the total amount recovered in all the fractions. The average of three independent experiments is shown with the S.E.M. (Bars). |
Taken together, these results indicated that YIPF3 and YIPF4 are mainly localized in the cis-Golgi.
The affinity purified anti-YIPF3 antibody recognized three bands (36, 40 and 46 kDa) by western blotting (Fig. 1). The possibility that these were unrelated proteins with similar antigenic properties was excluded since the signals of all of these bands were specifically reduced by YIPF3 siRNA treatment (Fig. 8A). Other possibility was that these were either products of post-translational modification or splicing variants. To investigate these possibilities, we established a stable cell line mildly expressing YIPF3 tagged with the FLAG-epitope at the N-terminus and the HA-epitope at the C-terminus (FLAG-F3-HA) and analyzed whether this produced similarly sized products (Shakoori et al., 2003). Previously, we have confirmed that the addition of these epitope tags does not affect the targeting, localization and dynamics of YIPF5 and YIF1A (Shakoori et al., 2003; Yoshida et al., 2008). When the cells were immunostained with anti-FLAG (M5) antibody, a clear Golgi pattern was observed (Fig. 4A, top panels), as was the case for the lower level transient expression (Shakoori et al., 2003). FLAG staining nicely co-localized with GM130 distribution, suggesting that epitope-tagging did not affect the localization of the protein. Anti-HA (16B12) antibody gave a similar staining pattern, showing clear Golgi localization (Fig. 4A, middle panels). However, we also saw apparent nuclear envelope and cytoplasmic meshwork staining, indicating localization of the protein in the ER. This appeared to be the result of accumulation of the protein in the ER caused by greater expression of the protein under these conditions compared with endogenous YIPF3 (discussed below). Curiously, anti-HA (3F10) only gave the ER staining pattern (Fig. 4A, bottom panels).
 View Details | Fig. 4. Stable expression of epitope-tagged YIPF3. HeLa cells stably expressing FLAG and HA-epitope tagged YIPF3 were established. (A) Cells were fixed, permeabilized by saponin and stained by anti-FLAG (M5), anti-HA (16B12) or anti-HA (3F10) antibody with anti-GM130 antibody. Bar=20 μm. (B) Cell lysates were prepared from control HeLa cells (C) or HeLa cells stably expressing FLAG-F3-HA (S) and analyzed by western blotting using anti-FLAG (M5), anti-HA (16B12), anti-HA (3F10) or anti-YIPF3 antibody as indicated on the top. Dots on the left of lanes indicate specific bands. The positions of the molecular weight markers (kDa) are shown on the left. |
To understand these immunofluorescence results, western blotting was performed. When the cell lysate of FLAG-F3-HA-expressing HeLa cells was probed with anti-FLAG (M5) antibody, three bands (III', 38 kDa; I', 43 kDa; and II', 49 kDa) were detected (Fig. 4B, lane 4). The estimated molecular weights of the bands corresponded with the endogenous proteins (lane 1, III; 36 kDa, I; 40 kDa, II; 46 kDa) plus the size increase for the epitope tags (FLAG+HA; 2.7 kDa). Anti-YIPF3 antibody also detected three bands at exactly the same positions as those detected by anti-FLAG (M5), indicating that the three bands detected by anti-FLAG (M5) were the proteins derived from the introduced FLAG-F3-HA gene. This strongly suggested that YIPF3 and exogenously expressed FLAG-F3-HA underwent the same post-translational modifications, giving rise to the three different molecular weight forms. Moreover, the endogenous YIPF3 signals had strongly decreased, showing as only faint bands at the corresponding positions (lane 2). These results suggested that exogenously expressed protein replaced the endogenous protein (discussed later).
Band III' was detected only by anti-FLAG and anti-YIPF3 but not by anti-HA, strongly suggesting that the HA-epitope was lost in this form (Fig. 4B, compare lanes 2 or 4 and 6 or 8). Anti-HA gave further intriguing results. In contrast with 16B12, 3F10 detected only band I' but not band II' (Fig. 4B, lane 8), suggesting that form II' received a modification that hampers recognition by 3F10 on the HA-tag. Whatever the nature of the modification, 3F10 gave the ER staining pattern by immunofluorescence (Fig. 4A, bottom panels), suggesting that form I' is localized primarily in the ER. This was confirmed by a cell fractionation experiment (unpublished observation), the results of which corresponded well with the finding that some endogenous YIPF3 form I was found in the ER/ERGIC fractions (Fig. 3). The amount of form I' appeared to be higher than that of the corresponding form I of endogenous YIPF3 (Fig. 4B, compare lanes 1 and 2), suggesting that over expression of FLAG-F3-HA causes the accumulation of the ER form (I'). Forms II' and III' were strongly suggested to be Golgi forms corresponding to form II and III, and this was also confirmed by cell fractionation (unpublished observation).
The YIPF proteins were predicted to have five trans-membrane regions with an N-terminal cytoplasmic region and a C-terminal luminal region. The protease protection experiment using cells expressing FLAG-F3-HA confirmed that the N-terminus with a FLAG-epitope was indeed exposed to the cytoplasm and that the C-terminus with a HA-epitope was in the lumen (Fig. S2A). We next sought to understand why 3F10 did not recognize the Golgi form (form II). Tyrosine residues are known to be sulfated in the Golgi apparatus (Friederich et al., 1988). The HA-tag was composed of the residues YPYDVPDYA and it has been found that tyrosine residue(s) are sulfated and that this hampers the recognition of the epitope by some antibodies (Personal communication, Dr. de Graffenried, Max F. Perutz Laboratories). Therefore, the tyrosine residue(s) in the HA-epitope may have been sulfated in the Golgi apparatus and this sulfation then hampers recognition of the HA-epitope by 3F10. To evaluate this possibility, cells expressing FLAG-F3-HA were treated with chlorate, which is a potent inhibitor of sulfation in the Golgi apparatus (Baeuerle and Huttner, 1986). Chlorate treatment induced the Golgi staining pattern of 3F10 (Fig. S2B) in a significant number of cells (35±8%), suggesting that the tyrosine residue of the HA-epitope was indeed sulfated and that the sulfation inhibited recognition of the HA-epitope by 3F10.
It was noticed that 16B12 gave comparatively stronger signal for form II' than form I' while anti-YIPF3 and M5 antibody gave similar levels of signal for form I' and II' (Fig. 4B, lane 6 compare to lane 2 or 4). This may be because 16B12 bound to the sulfated HA-epitope more efficiently than the non-sulfated one.
Next, we carried out tracer experiments to analyze the precursor product relationship between the three forms of YIPF3. When cells were pulse labeled with [35S]-methionine/cysteine for 5 minutes, the 40 kDa ER form (form I) was detected as the major form (Fig. 5A, lane 2). The 46 kDa Golgi form (form II) was also detected at this time point, as minor doublet bands. After a 30-minute chase, the amount of form I had decreased while that of form II had increased concomitantly (lane 4). After a 60-minute chase, the amount of form I had greatly decreased while that of form II had increased further (lane 6). This result strongly suggested that form II is derived from form I. The 36 kDa Golgi form (form III) was under the detectable level up to the 60-minute chase (lane 6). To detect form III, longer labeling (30 minutes) was performed. Form III was then detected after a 120-minute chase (Fig. 5B, lane 4) and the amount increased after a 240-minute chase (lane 6). The amount of form I was under the detectable level at the 120-minute chase, strongly suggesting that form III is generated from the form II during the 120- and 240-minute chases.
 View Details | Fig. 5. Post-translational processing of YIPF3. HeLa cells were pulse-labeled with [35]S-methionine/cysteine and chased for the indicated time. The cells were then analyzed by immunoprecipitation using control (PI) or affinity purified anti-YIPF3 antibody (F3). Precipitated materials were then analyzed by SDS-PAGE followed by autoradiography. In C and D, precipitated materials were treated with the indicated deglycosylation enzyme (N-Glyc ase; N-Glycanase, NeuAc ase; neuraminidase, O-Glyc ase; O-glycanse). The positions of the molecular weight markers (kDa) are shown on the left. The position of form I, II or III is indicated by arrows on the right. Asterisks indicate non-specific precipitate bands. Dots on the left of lanes with alphabets indicate bands of interest described in the text. |
Taken together, these results indicated that YIPF3 is synthesized as the 40 kDa form (form I) in the ER and is then post-translationally processed to the 46 kDa form (form II) and later to the 36 kDa form (form III) in the Golgi apparatus.
The luminal region of YIPF3 has several potential N- and O-glycosylation sites (Fig. 6A). To evaluate the possibility of glycosylation, we examined whether YIPF3 shows mobility shifts upon glycosidase treatment. [35S]-labeled YIPF3 was immunoprecipitated as before and treated with glycosidases (Fig. 5C and D).
 View Details | Fig. 6. Mutagenesis of potential N- and O-glycosylation sites of YIPF3. (A) Amino acid sequence of the predicted C-terminal luminal region of YIPF3. Asterisks indicate glycosylation sites determined. Expected cleavage site(s) are indicated by Δ. (B) Wild type (lanes 1 and 2) or point mutants (lanes 3–18) of epitope-tagged YIPF3 (FL-F3-HA) were constructed and transiently transfected into HeLa cells (lane 2–18). The cells were then lysed and subjected to western blotting analysis using anti-HA (16B12) antibody. Mutations were introduced singly or in combination, as indicated at the top of the panels (+). HeLa cells stably expressing wild type FL-F3-HA were also analyzed as the control (lane 1). (C) Control HeLa cells (Cont; lane 1) and HeLa cells transfected with wild type (WT; lane 2), N-glycosylation defective mutant (N337A; lane 3) or N- and O glycosylation defective mutant (ΔGlyc; lane 4) of YIPF3 were analyzed as in (A) using anti-FLAG M5 antibody. Lower panel shows a darker reproduction of the band III area. (D) The amounts of form I, II and III of FLAG-tagged YIPF3 in (C) were densitometrically determined and relative amount of form III (%) to total expressed FLAG-tagged YIPF3 (I+II+III) was calculated as described in Materials and Methods. The average of three independent experiments is shown with the S. D. (Bars). |
First, the existence of N-linked oligosaccharides was analyzed by N-glycosidase F treatment. When the labeled proteins were analyzed after the 60-minute chase (Fig. 5C), only form II (lane 1; bands a and b) and III (band d) were detected, as was the case after the 120-minute chase (Fig. 5B, lane 4). After the N-glycosidase F treatment, form II (bands a and b) disappeared and higher mobility forms (lane 2, bands e and f) appeared concomitantly. The mobility of form III did not change after N-glycosidase F treatment (band d). Thus, form II (bands a and b) was N-glycosylated and converted into a lower mobility form (bands e and f) by N-glycosidase F treatment while form III was not N-glycosylated. When the labeled proteins were analyzed without a chase, only forms I (Fig. 5D, lane 1; band c) and II (bands a and b) were detected. After N-glycosidase F treatment, both forms disappeared and higher mobility forms (lane 2; bands e, f and g) appeared concomitantly. Since bands e and f were derived from form II, band g was the product of form I (compare Fig. 5C, lanes 1 and 2 and Fig. 5D, lanes 1 and 2). Thus, form I was also N-glycosylated.
Forms I and II did not produce the same mobility product after removal of the N-linked oligosaccharide, suggesting that form II is produced by O-glycosylation of form I in the Golgi apparatus. Therefore, we next analyzed the existence of O-linked oligosaccharides using O-glycosidase. O-glycosidase is only able to remove free Gal-β(1-3)-GalNAc core-saccharides attached to serine or threonine and not the common mono-, di- and tri-sialylated saccharides. Therefore, α-2(3,6,8,9) neuraminidase was added to remove NeuAc from the Gal-β(1-3)-GalNAc core to completely remove the O-linked oligosaccharides. After α-2(3,6,8,9) neuraminidase treatment, form II showed a mobility shift similar to that seen after N-glycosidase F treatment (Fig. 5C, lane 5; band j). After double treatment with α-2(3,6,8,9) neuraminidase and O-glycosidase, form II was converted into a higher mobility form (lane 6; band k) similar to form I (lane 7, band c). Slightly lower mobility of band k compared with band c was probably the result of the modification of N-liked sugar chain in the Golgi apparatus. Double treatment of the no-chase sample produced no additional form with higher mobility than form I (Fig. 5D, lane 5; band c/k, compare with Fig. 5C, lane 6; band k) confirming that form I is the ER form and not O-glycosylated.
Form II showed a further mobility shift by simultaneous removal of the N-linked and O-linked oligosaccharides (Fig. 5C, compare lane 4; band i with lane 2; band e or f, or lane 6; band k). Complete removal of the N-linked and O-linked oligosaccharides from form II yielded the same mobility product as form I (Fig. 5D, lane 4; band g/i, compare with Fig. 5C, lane 4; band i). Again, form III showed no mobility shift under any conditions (Fig. 5C; band d), suggesting that form III is neither N- nor O-glycosylated. The size of form III (Fig. 5C; band d) was apparently smaller than de-glycosylated form I or II (Fig. 5C and D; band g or i), strongly suggesting that the luminal glycosylated region is lost in form III.
Taken together, these findings indicated that YIPF3 is N-glycosylated in the ER (form I) and then O-glycosylated in the Golgi apparatus (form II) on its luminal region. The luminal region of YIPF3 is later removed in the Golgi apparatus to produce form III.
When a C-terminal deletion mutant (ΔC332; C-terminal 18 residues were deleted) was expressed in HeLa cells and analyzed by western blotting, a major band of 39 kDa was detected (Fig. S2C, lanes 3 and 6) which corresponded well with the predicted size (38 kDa). In contrast, the size of the major band estimated for the wild type (Fig. 4 and Fig. S2C, form I': 43 kDa) was larger than the predicted size (40 kDa). As described above, the difference (3 kDa) was the contribution of N-linked sugar chain. This result indicated that the N-glycosylation site was on the C-terminal 18 residues of YIPF3. Furthermore, lower mobility species corresponding to the form II' was not detected for ΔC332 (Fig. S2C, compare lane 5 and 6). Therefore, it was suggested that the major O-glycosylation sites were also on the C-terminal 18 residues. Based on these results, we attempted to determine the glycosylation sites of YIPF3 by point mutagenesis. An epitope-tagged YIPF3 construct (FLAG-F3-HA) was used as the template and potential glycosylation sites (serine, threonine and asparagine; see Fig. 6A) were mutated to alanine in various combinations.
When wild type FLAG-F3-HA was expressed transiently and analyzed by western blotting using anti-HA (16B12), forms I' and II' were detected, as was the case in cells stably expressing FLAG-F3-HA (Fig. 6B, lanes 1 and 2). Mutation of the asparagine residue at 337 (N337) to alanine caused disappearance of form I', as well as an increase in the mobility of form II' (compare lane 2 with lane 3). The extent of the mobility shift of form II' was consistent with the result of N-glycosidase F treatment (Fig. 5D, lanes 1 and 2). These results indicated that N337 is the N-glycosylation site. Unexpectedly, a higher mobility band (band 0) was detected under the band of form I', even for wild type FLAG-F3-HA (lane 2). This was not observed in cells stably expressing wild type FLAG-F3-HA (lane 1). Since an identical mobility band was detected when all the glycosylation sites were mutated (lane 18), the band 0 was thought to be a non-glycosylated form of FLAG-F3-HA produced by over expression of the protein.
To identify the O-glycosylation site(s), point mutants were further produced on the N337A mutant background to simplify interpretation of the results. Firstly, we selected three potential O-glycosylation sites (T333, T339 or T346) that were predicted to have higher possibility of glycosylation by the NetOGlyc 3.1 program (Julenius et al., 2005). When the potential O-glycosylation sites (T333, T339 or T346) were singly mutated to alanine, a significant mobility shift of form II' was observed (Fig. 7B, lanes 4–6). A further mobility shift was observed by the combination of two sites (lanes 7–9). The mobility of the band became highest when all the three sites were mutated (lane 10) suggesting that all these sites are O-glycosylated. However, a slightly lower mobility band was still detected for this triple O-glycosylation site mutant suggesting that there are yet other O-glycosylation site(s). Therefore we introduced further mutations (T306, T334 or S349) on the triple O-glycosylation site mutant. Only the T334 mutation induced the disappearance of the lower mobility band (lanes 13, 15, 17 or 18) while the T306 or S349 mutation preserved the lower mobility band (lanes 12, 14 or 16). These findings indicated that YIPF3 is O-glycosylated on T333, T334, T339 and T346.
 View Details | Fig. 7. Oligomer formation of YIPF3 and YIPF4. (A–C) Rat liver Golgi membranes were treated with digitonin containing buffer and insoluble materials (P) were precipitated by centrifuge. The soluble extract (S) was then analyzed by SDS-PAGE (A), blue native PAGE (B) or blue native PAGE for the first dimension and SDS-PAGE for the second dimension (C) followed by western blotting using anti-YIPF3 or anti-YIPF4 antibody. The positions of the molecular weight markers (kDa) are shown on the left in B and on the top and left in C. The estimated molecular weight of each band is indicated on the right in A and B. The arrows in B and C indicate the major lower mobility complexes (520 and 260 kDa). The arrowhead indicates the minor higher mobility complexes (450 and 220 kDa). (D–F) HeLa cells stably expressing epitope-tagged YIPF3 (D and E) or YIPF4 (F) were lysed and immunoprecipitated by affinity purified anti-YIPF3 antibody (F3, IM), anti-YIPF4 serum (F4, IM) or their corresponding preimmune IgG or serum (F3, PI or F4, PI). (D) The precipitated materials were analyzed by western blotting using anti-FLAG M5 (lanes 1–5), anti-HA 16B12 (lanes 6–10) or anti-HA 3F10 (lanes 11–15) antibody. Asterisks on the left of M5 panel indicated non-specific reaction signals of the M5 antibody for the cell lysate. An asterisk on the right indicated a cross-reaction signal for IgG heavy chain. (E) The samples in (D) were analyzed using anti-YIF1A antibody. (F) The samples were analyzed by anti-HA 16B12 antibody (lanes 1–5). A 30% amount of the lysate used for precipitation was also analyzed as a control (L). |
We noticed that the amount of form III' decreases in glycosylation mutants probing with anti-FLAG (M5) antibody. Significant amount of form III' was detected in the lysate of HeLa cells transiently transfected with the wild type YIPF3 (Fig. 6C, lane 2). The result of densitometry showed that about 8% of exogenously expressed YIPF3 was in form III' (Fig. 6D). In contrast, apparently lower amount of form III' (~4%) was detected with the N-glycosylation defective mutant (Fig. 6C and D, N337A). Much lower amount of form III' (~2%) was detected with the N- and O-glycosylation double defective mutant (ΔGlyc; N337A, T333A, T334A, T339A, T346A). The reduction of form III' by the removal of N- and O- glycosylation suggested that the glycosylation is important for the accumulation of form III'.
Recently we have demonstrated that YIPF5 and YIF1A, the human orthologues of yeast Yip1p and Yif1p, respectively, form a complex (Yoshida et al., 2008). YIPF3 and YIPF4 are paralogues of YIF1A and YIPF5, respectively. Therefore, it is probable that YIPF3 and YIPF4 also form a complex. To evaluate this possibility, we first tried to detect an oligomer of YIPF3 and YIPF4 by blue native PAGE (BN-PAGE). For this purpose, we used rat liver Golgi lysate, which contains only the Golgi forms to simplify interpretation of the results (Fig. 7A). As shown in Fig. 7B, major bands were detected at about 520 kDa and 260 kDa. Minor bands were also detected at about 450 kDa and 220 kDa. YIPF3 and YIPF4 were detected at almost exactly the same positions, strongly suggesting that these are in the same complexes. Since YIPF3 has two Golgi forms (Fig. 7A, upper panel; forms II and III), we next tried to determine whether both of these were in the complexes. For this purpose, 2 dimensional PAGE combining BN-PAGE and SDS-PAGE was performed (Fig. 7C). Spots corresponding to both forms II and III were detected in the higher molecular weight regions of dimension I (upper panel, arrows and arrowheads). YIPF4 was also detected in the same regions (lower panel, arrows and arrowheads). This result strongly suggested that both forms II and III form complexes with YIPF4. Majority of form II was detected at the lower mobility positions of the first dimension (520 kDa and 260 kDa, arrows). On the other hand, the majority of form III was detected at the higher mobility positions (450 kDa and 220 kDa, arrowheads). This result suggested that the major lower mobility complexes are mainly composed of YIPF3 form II and YIPF4 while the minor higher mobility complexes are mainly composed of YIPF3 form III and YIPF4. Furthermore, clear spots of form II were detected at the 220 kDa position (an arrowhead) suggesting that form II is in the minor 220 kDa complex together with form III.
We then performed co-immunoprecipitation experiments to confirm the complex formation of YIPF3 and YIPF4 using cells stably expressing YIPF3 (FL-F3-HA). The cell lysate was prepared and immunoprecipitated with anti-YIPF3 or anti-YIPF4 antibody and then probed with anti-FLAG (M5), anti-HA (16B12) or anti-HA (3F10). As shown in Fig. 7D, all three forms of YIPF3 were precipitated by the anti-YIPF3 antibody and detected by anti-FLAG (M5) (lane 3; forms I and III were detected), anti-HA (16B12) (lane 8; forms I and II were detected) and anti-HA (3F10) (lane 13; form I was detected) antibodies. Interestingly, anti-YIPF4 antibody also effectively precipitated forms II and III (lanes 5 and 10), but not form I (lanes 5, 10 and 15). This result strongly suggested that apparently lower amount of the ER form (form I) of YIPF3 are in a complex with YIPF4 compared with the Golgi forms (form II and III). YIF1A was not significantly precipitated with anti-YIPF3 or anti-YIPF4 antibody (Fig. 7E, lanes 3 and 5) confirming the specificity of the precipitation.
To unequivocally demonstrate complex formation, the converse experiment was performed using cells stably expressing YIPF4 (FL-F4-HA). As shown in Fig. 7F, YIPF4 was effectively precipitated by anti-YIPF4 (lane 5) and also by anti-YIPF3 antibody (lane 3) confirming complex formation by YIPF3 and YIPF4.
We then tried to knockdown YIPF3 and YIPF4 to explore the function of these proteins. As shown in Fig. 8A, YIPF3 or YIPF4 was successfully knocked down 3 days after siRNA transfection. YIPF3 or YIPF4 was efficiently reduced to about 10% or 20% by YIPF3- or YIPF4-targeted siRNA, respectively, compared to the control double strand RNA (Fig. 8B or C). Interestingly, YIPF3 was also effectively reduced by YIPF4 targeted siRNA (Fig. 8B; F4 KD). On the other hand, YIPF4 was only slightly reduced by siRNA targeted for YIPF3 (Fig. 8C; F3 KD, ~65%). These results suggested that the expression of YIPF4 is necessary for the expression of YIPF3 but not vice versa. YIF1A and actin were not affected, confirming that the effects of the siRNAs are specific (Fig. 8A).
 View Details | Fig. 8. Knockdown of YIPF3 and YIPF4 in HeLa cells. HeLa cells were transfected with siRNA designed for YIPF3 (F3 KD), YIPF4 (F4 KD) or control (Cont). (A) The cells were analyzed by western blotting using anti-YIPF3 (upper panel) or anti-YIPF4 (upper middle panel) antibody. They were also analyzed using anti-YIF1A (lower middle panel) and anti-actin (lower panel) antibodies as controls. (B and C) The amount of YIPF3 (B) or YIPF4 (C) in (A) was densitometrically quantified. The result is the mean of three independent experiments. Bars indicated S.E.M. (D) The cells were subjected to immunofluorescence staining using anti-YIPF3 and anti-YIPF4 antibodies with anti-GM130 antibody. Bar=20 μm. (E) The number of cells showing fragmentation of the Golgi apparatus was counted and presented as a percentage of the total number of cells. The result is the mean of three independent experiments. Bars indicated S.E.M. |
We then observed the knockdown cells by immunofluorescent staining. As shown in Fig. 8D, the fluorescent intensity was greatly reduced for YIPF3 or YIPF4 by siRNA targeted for YIPF3 (middle gallery) or YIPF4 (right gallery), respectively, compared with the control (left gallery). The intensity of YIPF3 staining was also reduced by siRNA targeted for YIPF4, corresponding with the western blotting result. Interestingly, the intensity of YIPF4 staining was also strongly reduced by siRNA targeted for YIPF3 (middle gallery, lower panels). This result suggested that YIPF4 is not able to localize to the Golgi apparatus, most probably retained in the ER, by the knockdown of YIPF3. Actually, large proportion of YIPF4 was recovered in heavier ER fractions by subcellular fractionation of YIPF3 knockdown cells strongly supporting this possibility (Fig. S2D and E).
We noticed that the Golgi apparatus was fragmented in YIPF3 or YIPF4 knockdown cells when observed by GM130 (Fig. 8D, compare left panels of left gallery with that of middle and right gallery). The effect was similar for YIPF3 knockdown (middle gallery) and YIPF4 knockdown (right gallery). To substantiate the effect, we quantified the number of cells showing fragmentation of the Golgi apparatus. As shown in Fig. 8E, fragmentation was observed only in about 5% of the cells transfected with the control double strand RNA. Medial-Golgi (N-acetylglucosaminyltransferase I green fluorescent protein fusion protein; GlcNAcTI-GFP), trans-Golgi (golgin97) and TGN (TGN46) marker proteins also showed similar fragmentation patterns (Fig. S3A) suggesting that all the compartments of the Golgi apparatus were affected by the knockdown of YIPF3 or YIPF4. On the other hand, HSP47 positive structure was not significantly altered by the knockdown (Fig. S3B). The appearance of ERGIC53 positive structures was not changed essentially while the clustering around the Golgi area was lost corresponding with the fragmentation of the Golgi apparatus (Fig. S3B). These results strongly suggested that the ER and the ERGIC were not affected by the knockdown.
Similar results were obtained using two other siRNAs for YIPF3 or YIPF4 (Fig. S4A and B) excluding the possibility that these are off-target side-effects.
We then analyzed whether the knockdown of YIPF3 or YIPF4 affected the anterograde transport by monitoring the transport of VSV-G(tsO45)-GFP as a marker transmembrane protein or a soluble form of dipeptidyl-peptidase IV (sDPPIV) as a marker secretory protein (Yoshida et al., 2008). As a result, VSV-G(tsO45)-GFP was transported to the Golgi apparatus and then to the plasma membrane in a rate that was indistinguishable from the control cells after the knockdown of YIPF3 or YIPF4 (Fig. S4C). Also, the rate of secretion of sDPPIV was not significantly altered in YIPF3 or YIPF4 knockdown cells (Fig. S4D). Furthermore, the N-linked sugar chain on the secreted sDPPIV was normally processed into a complex form and sialylated in YIPF3 or YIPF4 knockdown cells (Fig. S4E) suggesting that the Golgi apparatus was functioning normally in the knockdown cells.
Here we showed that YIPF3 and YIPF4 localize mainly in the cis-Golgi (Fig. 2, Fig. 3 and Fig. S1). This is in marked contrast with their paralogues, YIF1A and YIPF5, which localize mainly in the ERGIC and recycle between the ER and Golgi apparatus. It is possible that YIPF3 and YIPF4 recycle within the Golgi apparatus and to some extent between the ER and the ERGIC. However, the apparent concentration of YIPF3 and YIPF4 in the cis-Golgi (Fig. 3) suggests that they are not actively recycling between the Golgi apparatus and the ER or the ERGIC, compared with YIPF5 and YIF1A. Instead, if they do recycle, it is probable that YIPF3 and YIPF4 recycle within the Golgi apparatus. In this respect, it is intriguing that HA-tagged YIPF3 (FL-F3-HA) is sulfated on the luminal HA-tag region (Fig. S2B and C). Since the tyrosine residue is reported to be sulfated in trans-Golgi (Baeuerle and Huttner, 1987), FL-F3-HA is probably transported to the trans-Golgi for sulfation. This implies that YIPF3 recycles between the cis- and trans-Golgi. Whether YIPF3 and YIPF4 recycle in and out of the Golgi apparatus, and to what extent this occurs, require further analysis.
YIPF3 and YIPF4 form a complex (Fig. 7) similar to their paralogues, YIF1A and YIPF5 (Yoshida et al., 2008), suggesting that complex formation by a Yip1p homologue (YIPF4, YIPF5) and a Yif1p homologue (YIPF3, YIF1A) is a general rule (Matern et al., 2000; Yoshida et al., 2008). The two dimensional PAGE analysis revealed that two different size forms of YIPF3 are in the complex (Fig. 7C). Taking this and the size of the complexes detected by BN-PAGE (Fig. 7B; 220~260 kD and 450~520 kDa) into account, we predicted that YIPF3 and YIPF4 form 4~16 mer complexes. However, the stoichiometry of the complex remains to be carefully determined because other molecules can be incorporated in the complex. Preliminary experiments have revealed that YIPF3 and YIPF4 in HeLa cells also form complexes of similar sizes as that in rat liver Golgi membrane. Similarly, we have shown previously that YIF1A and YIPF5 can form similar 4~12 mer complexes in HeLa cells (Yoshida et al., 2008). However, the estimated molecular weights of the complexes are apparently different from those of YIPF3 and YIPF4 suggesting that YIPF3-YIPF4 and YIF1A-YIPF5 are distinct complexes. This was confirmed by the specific immunoprecipitation of YIPF3-YIPF4 complex (Fig. 7). Therefore, the recognition of a partner protein for complex formation appears to be specific for each YIPF protein. Whether this is a general rule for all members of YIPF needs to be determined.
Interestingly, YIPF4 appear to form a complex only with the Golgi form of YIPF3, suggesting that YIPF3 and YIPF4 do not form a complex in the ER (Fig. 7D). On the other hand, the knockdown of either protein resulted in a reduction in the counterpart protein from the Golgi apparatus (Fig. 8). Therefore, complex formation is suggested to be important for the localization of YIPF3 and YIPF4 in the Golgi apparatus. Regulation of complex formation and localization of YIPF3 and YIPF4 is not yet clear. An intriguing possibility is that YIPF3 and YIPF4 are retained in or retrieved to the ER while they are not in a complex. The fact that higher level expression of YIPF3 causes accumulation of form I in the ER (Fig. 4) supports this possibility.
The knockdown of YIPF4 caused a reduction in YIPF3 but not vice versa (Fig. 8). The reason for this non-symmetrical result remains obscure. Nevertheless, the data clearly indicated that the amount of YIPF3 is restricted by YIPF4. Correspondingly, exogenous expression of YIPF3 resulted in the reduction of endogenous YIPF3 (Fig. 4). Therefore, it is probable that YIPF3 is only stable in a complex with YIPF4 and that the free YIPF3 is degraded. Alternatively, free YIPF3 may negatively regulate endogenous YIPF3 transcription and/or translation.
YIPF3 was shown to receive N- and O-glycosylation on its C-terminal luminal region (Fig. 5 and Fig. 6). The luminal region was then cleaved off in the Golgi apparatus. Preliminary experiments using deletion mutants showed that the luminal region is cleaved near to the 320th residue (Fig. 6A, Δ). The glycosylation of the luminal region was shown to be important for the accumulation of form III, the cleaved form (Fig. 6C and D). It is possible that the cleavage of the luminal region was reduced by the defect in glycosylation. For example, the glycosylation of the luminal region may promote the ER to Golgi transport of the protein and the following cleavage of the luminal region. However, all the glycosylation deficient mutants of YIPF3 were localized to the Golgi apparatus in similar level with the wild type YIPF3 (unpublished observation). Therefore, it is less likely that the deficiency in glycosylation affects the ER to Golgi transport of YIPF3 although we cannot formally exclude the possibility that the ER to Golgi transport of YIPF3 was reduced in an undetectable level in our experimental system. Alternative possibility is that glycosylated YIPF3 becomes enzymatically susceptible for the cleavage. It is also possible that the un-glycosylated form is actively turned over before producing form III. To understand how the glycosylation affects the cleavage of the luminal region of YIPF3, it has to be understood where the cleavage occurs and which enzyme is responsible. These are future problems to be elucidated.
Importantly, it remains unclear whether these post-translational processing play any roles in the functioning of YIPF3. The fact that similar processing was observed in zebrafish (unpublished observation) implies a conserved function for the post-translational modification of YIPF3. Since only forms II and III, the Golgi forms, but not form I, the ER form, were in a complex with YIPF4, it is possible that the modifications regulate complex formation with YIPF4.
The knockdown of either YIPF3 or YIPF4 caused apparent fragmentation of the Golgi apparatus. Correspondingly, knockdown of either YIPF3 or YIPF4 resulted in a reduction of the counterpart protein from the Golgi apparatus (Fig. 8). Therefore, in both cases, the amount of YIPF3-YIPF4 complex in the Golgi apparatus was greatly reduced and this may have induced the fragmentation of the Golgi apparatus. However, there was no detect in anterograde vesicular transport through the Golgi apparatus and the Golgi apparatus was functioning normally after the knockdown of either YIPF3 or YIPF4 in HeLa cells (Fig. S4 C–D). This is similar to the results of knockdown of the paralogues YIF1A and YIPF5 (Yoshida et al., 2008). Furthermore, we tried to knockdown all four members (YIPF3, YIPF4, YIPF5 and YIF1A) simultaneously, but again, no significant effect was observed (unpublished observation). As we have discussed previously (Yoshida et al., 2008), it is possible that loss of YIPF3 or YIPF4 is somehow compensated for by other proteins with overlapping function, for example, other YIPF members. It is also possible that there is an effect on the anterograde or retrograde transport of some proteins, which was not detectable in our experiments. It has been shown that the knockdown of YIPF5/YIP1A causes a reduction in COPI independent Golgi to ER retrograde transport (Kano et al., 2009). Therefore, it is possible that YIPF3 and YIPF4 also function in Golgi to ER retrograde transport. Alternatively, YIPF3 and YIPF4 may function in transport of some special molecule(s) necessary for the differentiation or development of vertebrates. In support of this possibility, YIF1B, another member of the YIPF proteins, has been recently shown to function in targeting serotonin receptor (5-HT1A) to the dendrites of neuronal cells (Carrel et al., 2008). These possibilities are future problem to be elucidated.
We would like to thank Drs. Y. Yoshida (Kyushu Univ.), S. Yoshimura (Osaka Univ.) and Mr. R. Nakazato (Kanazawa Univ.) for experimental help and discussions, Mr. D. Yamada (Kanazawa Univ.) for preparation of the antibodies and Dr. H.-P. Hauri for providing the antibody, all the previous members of Nakamura Lab for their assistance. Special thanks goes to Drs. Minori Shinya and Noriyoshi Sakai (National Institute of Genetics of Japan) for their helpful comments and discussions. This work was supported by Grants-in-Aid for Scientific Research (#18570173), and for Scientific Research on Priority Areas (#16044218, #17028016, #21025012) from MEXT of Japan, research grants from The Naito Foundation (2006), Kanazawa University (2008, #2023202) and Kyoto Sangyo University (for N.N.).
|
Index