|Home | About | Journals | Submit | Contact Us | Français|
ALG-2 (apoptosis linked gene 2 product) is a calcium binding protein for which no clear cellular function has been established. In this study we identified Scotin as a novel ALG-2 target protein containing 6 PXY and 4 PYP repeats, earlier identified in the ALG-2 binding regions of AIP1/ALIX and TSG101, respectively. An in vitro synthesized C-terminal fragment of Scotin bound specifically to immobilized recombinant ALG-2 and tagged ALG-2 and Scotin were shown by immunoprecipitation to interact in MCF7 and U2OS cell lines. Furthermore ALG-2 bound to endogenous Scotin in extracts from mouse NIH3T3 cells. Overexpression of ALG-2 led to accumulation of Scotin in MCF7 and H1299 cells. In vitro and in vivo binding of ALG-2 to Scotin was demonstrated to be strictly calcium dependent indicating a role of this interaction in calcium signaling pathways.
ALG-2 was identified in 1996 in a cell death trap assay as a proapoptotic protein since expression of ALG-2 antisense RNA inhibited T-cell receptor induced apoptosis in the 3DO T-cell hybridoma cell line . As ALG-2 is a calcium binding protein of the penta EF-hand protein family (PEF)  with an affinity for calcium in the micromolar range, similar to calmodulin, it has been suggested that ALG-2 could be involved in calcium-dependent apoptosis . However, molecular pathways leading to ALG-2 dependent cell death have not been worked out so far. Moreover, published work (reviewed in  and our unpublished observations) indicate that this protein could also play a role in proliferative processes. An example is its preferential expression in cancer tissues .
Several ALG-2 target proteins have been identified including AIP1/Alix [5,6], annexins 7, 11  and TSG101 , all featuring a proline-rich region. The interaction of ALG-2 with AIP1/Alix has been studied in some detail both on a functional and structural level (reviewed in [9,10]). ALG-2 expression counteracts the anti-apoptotic properties of a truncated form of AIP1/Alix when expressed in HeLa cells  and it has been reported that AIP1/ALIX plays a role in endocytosis (reviewed in [5,10]) and takes part in retrovirus budding [11,12]. ALG-2 binds AIP1/Alix in a calcium-dependent fashion and the ALG-2 interacting region of AIP/Alix has been mapped by two groups independently to a proline-rich sequence at position 801–812  and 794–826 of human AIP1/ALIX . Other identified ALG-2 interacting proteins, annexin 7 and 11  as well as TSG101  contain sequences with similarity to the mentioned AIP1/Alix ALG-2 interacting region. Recently, Sec31A, a novel ALG-2 binding protein containing as well as proline-rich sequence was identified independently by three groups [14–16]. Binding and colocalization with Sec31A, a coat component of the ER to Golgi transport vesicle COPII were found to be Ca2+-dependent [14–16]. It seems therefore likely that ALG-2 exerts its calcium dependent function at least partially at the ER. By searching databases with the ALG-2 binding sequence of AIP1/Alix in combination with an ER localization signature we found Scotin.
Scotin was identified by a differential display approach in search for new genes involved in p53-dependent apoptosis induced by irradiation of normal and p53 nullizygote mice . Scotin has the structure of a transmembrane receptor of type I composed of a signal sequence, a cysteine-rich domain, a transmembrane domain and a proline/tyrosine-rich region and is localized to the endoplasmic reticulum (ER)1 with the proline/tyrosine-rich domain at the cytosolic face of the ER membrane. Transfection of wt Scotin but not of mutant Scotin lacking the N-terminal domain promoted apoptosis in a caspase-dependent manner in tumor cells independently of p53, indicating that Scotin has an intrinsic pro-apoptotic activity localized to the N-terminal cysteine-rich domain. Inhibition of endogenous Scotin expression by siRNA increases resistance to apoptosis induced by UV irradiation despite strong activation of p53, indicating that Scotin plays a significant role in p53-dependent apoptosis. The discovery of Scotin brings to light a role of the endoplasmic reticulum (ER) in p53-dependent apoptosis and prompts investigation on possible connections with ER-linked processes. Here, we provide evidence that ALG-2 binds Scotin in vitro and in vivo and that overexpression of ALG-2 stabilizes Scotin.
Gene-specific primers for human Scotin encoding amino acids 68–240 were designed with addition of sequences on both upstream and downstream primers. Upstream primer: 5′-ctttaagaaggagatataccatgAGCGAGGAAAGGTGTGCTGTGCC and downstream primer: 5′-tgatgatgagaaccccccccGAGGGCCGCCTTCGGGGCATC (capital letters indicate gene specific sequences). The 5′-ends of the primers overlap with DNA sequences that contain the coding sequence for the C-terminal His6-tag and all regulatory elements required for expression in a wheat germ cell-free expression system. In a second round of PCR the DNA template was extended with these regulatory DNA sequences supplied in the RTS Wheat Germ Linear Template Generation Set, His6-tag (Roche Applied Science). The Scotin cDNA fragment of 172 bp was amplified from a placenta cDNA library (Stratagene) with the use of the Advantage—GC 2 PCR kit (Clontech). PCR program: 94 °C, 3 min; 35 cycles 94 °C, 30 s and 68 °C, 2 min followed by a 3 min extra elongation step of 68 °C. The second PCR was performed according to the manufacturer′s protocol (Roche Applied Science). From the linear PCR-generated template. Scotin was expressed in a eukaryotic cell-free protein synthesis system based on a wheat germ lysate provided by the RTS 100 Wheat Germ CECF kit according to the manufacturer′s protocol (Roche Applied Science).
The cDNA of ALG-2 and the short isoform, ALG-2.1 , were cloned into the pGEX-2TK expression vector downstream of glutathione S-transferase (GST). Recombinant GST fusion proteins were expressed in BL21 Escherichia coli cells following transformation with either pGEX-2TK, pGEX-2TK/ALG-2, pGEX-2TK/ALG-2.1, pGEX-2T-Scotin-4 or pGEX-2T-Scotin-5. Recombinant GST-Scotin-4 protein was generated by subcloning the cysteine-rich domain of human Scotin (aa 26 to aa 108) fused to a FLAG-tag in frame of GST at its C-terminus in the pGEX-6 plasmid. Recombinant GST-Scotin-5 protein was generated accordingly by subcloning the proline-rich domain of human Scotin (aa 123 to aa 240) fused to a Flag-tag into the same vector. Expression of the proteins was induced at OD600 = 0.4 by 1 mM isopropyl-1-thio-β-d-galactopyranoside (IPTG) for 2 h at 16 or 25 °C. Bacteria were harvested by centrifugation at 5000g for 10 min at 4 °C and lysed in PBS containing 1 mg/ml lysozyme followed by 20 min incubation on ice before freezing at −80 °C. Thawed lysates were cleared by centrifugation at 50,000g for 30 min at 4 °C and the GST fusion proteins were purified on a 1 ml GSTrap FF affinity column according to the manufacturer′s protocol (GE Healthcare).
Immobilized GST fusion proteins were washed three times in binding buffer (0.5% Nonidet P-40, 150 mM NaCl, 1 mM DTT, 0.1% proteinase inhibitor mix P8340 (Sigma–Aldrich), 50 mM Tris, pH 7.4) containing either 1 mM CaCl2 or 1 mM EGTA before use in the pull-down assays. The Sepharose conjugated GST fusion proteins were incubated with His-tagged Scotin (residues 68–240) with gentle rotation at 4 °C overnight. The same procedure was followed when ALG-2 covalently bound to NHS-activated Sepharose beads were used to pull-down His-tagged Scotin. After incubation the samples were centrifuged at 1000g for 1 min at 4 °C. Supernatants were removed and pellets were washed three times in binding buffer containing either 1 mM CaCl2 or 1 mM EGTA. The GST fusion proteins were eluted from the glutathione Sepharose beads by addition of 20 mM l-glutathione (GSH) in Tris pH 8 containing 5 mM DTT. The samples were analyzed on a 12% polyacrylamide gel under reducing and denaturing conditions and transferred onto a PVDF membrane (Hybond-P, GE Healthcare). Immunolabeled proteins were detected by HRP conjugated secondary antibodies followed by the ECL plus chemiluminescence system according to the manufacturer′s manual (GE Healthcare).
Bacterial extracts (containing GST-Scotin-4 or GST-Scotin-5) were run on a 12% polyacrylamide gel and the proteins were transferred onto a PDVF membrane as described. After blocking for 1 h in AC buffer (containing 10% glycerol, 100 mM NaCl, 20 mM Tris–Hcl, pH 7.4, 0.1% Tween 20) with 2% low fat milk powder, the membrane was incubated with 5 μg/ml recombinant ALG-2 in AC buffer overnight followed by extensive washing in AC buffer. The membrane was then probed with ALG-2 antibodies and labeled antibody detected as described above.
Polyclonal antibodies against mouse Scotin were used as described in . Affinity purified polyclonal anti-mouse ALG-2 antibodies were used as previously described . Mouse monoclonal anti-TSG101 antibodies was from Abcam (Abcam, Cambridge, MA, USA). Mouse monoclonal anti-His6-Peroxidase conjugated antibodies was from Roche Applied Science. Rabbit polyclonal antibodies against AIP1/ALIX and GST were made in our laboratory (unpublished). Secondary HRP-conjugated goat anti-mouse- or goat anti-rabbit-immunoglobulins were from Dako A/S (Glostrup, Denmark). The monoclonal antibody directed against the FLAG tag and the monoclonal antibody directed against the HA tag were purchased from Sigma–Aldrich.
NIH3T3 cells were a kind gift from Professor Berthe M. Willumsen. Cells were grown to confluency and harvested by trypsination. Cell pellets were suspended in lysis buffer containing 10 mM Hepes–NaOH, pH 7.4, 142.5 mM KCl, 0.2% NP-40, 2 mM NaVO4, 20 mM NaF and 0.1% proteinase inhibitor mix P8340 (Sigma–Aldrich), and lysed mechanically by 10 strokes in a Dounce homogenizer. Lysates were spun at 10,000g at 4 °C for 10 min and the protein concentration of the supernatants was adjusted to 2 mg/ml. Either 10 μM CaCl2 or 1 mM EGTA was added to the lysates before incubation with control beads or beads conjugated with either recombinant ALG-2.1 or ALG-2. Recombinant ALG-2.1 and ALG-2 were expressed, purified and conjugated as previously described [4,18]. Beads were incubated with lysates overnight at 4 °C and washed three times in lysis buffer before addition of sample buffer (100 mM Tris, pH 6.8, 200 mM DTT, 4% SDS, 0.2% bromophenol blue and 20% glycerol) followed by boiling for 2 min. Proteins pulled down from the lysates were separated by SDS–PAGE (10%), blotted to a PVDF membrane and immunolabeled with antibodies against Scotin, Tsg101 and AIP1/Alix.
About 1 × 106 H1299, MCF7 or U2OS cells were seeded onto 10-cm tissue culture plates and transfected 24 h later by Fugene (Boehringer). Construction of the EGFP-ALG-2 plasmids is described in  and the HA tag was added of the intact cDNA coding sequence of the ALG-2 variants in the pcDNA3 vector. Typically, 5 μg of expression plasmids for Scotin and for ALG-2 were transfected per 10 cm plate, unless otherwise stated. The DNA amount was normalized to 10 μg with pcDNA3 vector or as otherwise stated.
For immunoprecipitation (IP), H1299 MCF7, or UO2 S cells were transfected and treated as indicated. Upon treatment, cells were fixed with 1% formaldehyde for 10 min at 37 °C, and then blocked for 5 min in 125 mM glycine at room temperature. Proteins were extracted using a lysis buffer containing 50 mM Tris, pH 7.5, 1% Triton X-100, 0.27 M sucrose, NaCl 100 mM, complete protease inhibitor cocktail (Roche Applied Science) and passed several times through a narrow gauge needle. The immunoprecipitating antibody was covalently coupled to protein G-Sepharose using the dimethyl pimelimidate method. For each IP 2 μg of antibody was coupled to 20 μl of packed volume protein G-Sepharose and incubated with 1 mg of cell extract for 1 h at 4 °C. IPs were washed four times in 1 ml of IP buffer. The immunoprecipitated proteins were eluted end-over-end for 30 min at ambient temperature in 50 μl of Laemmli sample buffer and boiled with 5% (v/v) β-mercaptoethanol, prior to SDS–PAGE (NuPAGE pre-cast 12% gels using MES running buffer from Invitrogen). For Western-blots proteins were extracted and analyzed as previously described .
Elucidation of the binding of ALG-2 to TSG101, AIP1/Alix, Annexin 7 and 11 (reviewed in ) encouraged database searches for protein sequences containing similar proline/tyrosine-rich regions with the goal to find novel ALG-2 binding partners. One of the candidate proteins found by this in silico approach was Scotin (Fig. 1a). Comparisons of the proline/tyrosine-rich region of Scotin with known ALG-2 interaction partners revealed that the proline-rich region of Scotin does contain six PXY repeats (residues 166–168, 171–173, 196–198, 206–208, 225–227, 229–231) as described for the ALG-2 binding region in AIP1/ALIX  and 4 PYP repeats (residues 157–159, 189–191, 197–199, 219–221) as well as 1 YPP sequence (residues 194–196), as described for the ALG-2 binding region in TSG101  (Fig. 1b).
Comparisons of the proline-rich region of Scotin (residues 157–240) with AIP1/Alix and TSG101 revealed that the N-terminal part of this region (residues 157–177) has highest similarity with the ALG-2 interacting sequence of AIP1/ALIX (residues 794–812), which has been shown to be sufficient for the interaction with ALG-2 . The C-terminal adjacent region in Scotin (residues 178–207) has higher similarity to TSG101 (residues 163–192) .
In order to examine binding between Scotin and ALG-2, we synthesized the predicted ALG-2 interacting protein fragment of Scotin in vitro. The fragment encoding amino acid residues 68–240 was amplified from a placenta cDNA library (Fig. 1a). At the 3′-end of the coding region a sequence encoding a His-tag was added and the PCR product was directly used as template in an in vitro transcription/translation system. The produced His-tagged Scotin fragment was then used in pull-down assays with either GST, GST-ALG-2, GST-ALG-2.1 or recombinant ALG-2 directly conjugated to NHS-activated beads or non-conjugated control beads.
ALG-2 was able to pull-down the Scotin fragment in a calcium dependent manner (Fig. 2a, top panel, lanes 3 and 4), whereas the two amino acids shorter ALG-2.1 isoform was unable to bind the Scotin fragment (Fig. 2a, top panel, lanes 1 and 2). Scotin could be pulled down under calcium conditions when ALG-2 was directly conjugated to beads (Fig. 2a, top panel, lane 7), but not with control beads (Fig. 2a, top panel, lane 8). Equal amounts of bait protein used for pull-down of Scotin were confirmed using antibodies against ALG-2 (Fig. 2a, middle panel) and GST (Fig. 2a, bottom panel). The strong signal in Fig. 2a, middle panel, lane 7 is probably due to noncovalent recombinant ALG-2 dimerization formed during conjugation. The presence of the bands of nonfused ALG-2 (Fig. 2a, middle panel lanes 1–4) and GST (Fig. 2a, bottom panel lanes 1–6) indicates that the GST-ALG-2 protein is unstable following sample buffer treatment.
To confirm direct binding of ALG-2 to Scotin variants we carried out Far Western blots using immobilized recombinant GST-Scotin (Fig. 2b and c). GST-Scotin-4 and GST-Scotin-5 fragments (Scotin-4 containing the N-terminal part and Scotin-5, containing the C-terminal fragment of Scotin with the proline-rich sequence, see Fig. 1a) fused to GST were expressed in bacteria and total extracts as well as soluble fractions were analyzed on a Far Western blot using recombinant ALG-2 as a probe. A strong single signal was detected in total lysates from GST-Scotin-5 expressing bacteria (Fig. 2b, lane 3), but no signal was seen in the soluble fraction (Fig. 2b, lane 4) and in total lysates as well as in the soluble fraction from of GST-Scotin-4 expressing bacteria (Fig. 2a and b, lanes 1 and 2) even though a similar amount of GST-Scotin-4 and GST-Scotin-5 was loaded on the gel (Fig. 2c) as demonstrated by probing for FLAG. These results confirm that the C-terminal part of Scotin binds to ALG-2.
To determine whether Scotin interacts with ALG-2 in a cellular environment, MCF7 cells were cotransfected with Scotin-FLAG tagged expression vectors in the presence or absence of HA-tagged ALG-2. Cells were harvested 48 h after transfection and protein extracts were incubated with HA monoclonal antibody for co-immunoprecipitation experiments (Fig. 3a). Scotin (theoretical size of 28 kDa) gives three bands (28, 25 and 17 kDa) after transfection in MCF-7 cells. All forms of Scotin co-immunoprecipitated with HA-ALG-2.
To determine whether the interaction between ALG-2 and Scotin was specific, FLAG-tagged Scotin-was cotransfected in MCF7 cell lines with different ALG-2 variants fused to EGFP or EGFP alone. Proteins were extracted 48 h after transfection and incubated with GFP antibody for immunoprecipitation (Fig. 3b). Only the EGFP-ALG-2 variant co-immunoprecipitated with Scotin. Interestingly, the ALG-2.1 variant and the ALG-2-EF mutant that does not bind calcium, do not co-immunoprecipitate with any form of Scotin although the ALG-2.1 and mutant ALG-2-EF are expressed at the same level as ALG-2.
To determine whether the interaction between ALG-2 and Scotin was calcium-dependent in vivo, Scotin-FLAG was cotransfected in U2OS cell lines with the three ALG-2 variants fused to EGFP or EGFP alone. Forty-eight hours after transfection, cells were treated or not treated with ionomycin (2 μM) to increase the intracellular calcium concentration. Proteins were extracted and incubated with GFP antibody beads for immunoprecipitation (Fig. 3c). In the absence of ionomycin, consistent with the results obtained in vitro only EGFP-ALG-2 co-immunoprecipitated with Scotin but not the variant ALG-2.1 and the mutant ALG-2-EF. Only the 28/25 kDa Scotin forms interact with ALG-2 upon ionomycin treatment. This indicates that the interaction between Scotin and ALG-2 is calcium-dependent in vivo and that in U2OS cells the 17 kDa Scotin species does not interact with ALG-2.
The ability of ALG-2 to bind endogenous mouse Scotin was tested in protein extracts from NIH3T3 cells. Scotin was expressed at a detectable level in NIH3T3 cells without induction of p53 (Fig. 4, lane 7). Scotin could be pulled down by ALG-2-beads from the lysates of NIH3T3 cells, however, with lower efficiency as compared to TSG101 (Fig. 4, lanes 3 and 7). We also confirmed that the Scotin interaction with ALG-2 was specific for the long form of ALG-2 because, similar to the in vitro situation, ALG-2.1-beads did not pull-down mouse Scotin from NIH3T3 lysates (Fig. 4, lanes 1 and 2). In addition, the binding of ALG-2 to Scotin from cell lysates required the presence of Ca2+ (Fig. 4, lane 3 and 4). As a positive control we show that antibodies directed against the previously described ALG-2 binding partners, AIP1/Alix  and TSG101  reacted with the corresponding antigens in ALG-2 pull-downs from NIH3T3 cell lysates. It should be noted that the presence of a strong ALG-2 signal in the ALG-2.1 and ALG-2 pull-down lanes in Fig. 4 is due to recombinant ALG-2 eluted from the ALG-2 beads following boiling of the samples (as is also seen in Fig. 2) which served as a loading control. The NIH3T3 cells express small amounts of ALG-2. However, this is only visible by longer exposures of the film (data not shown).
In initial experiments to study the interplay between Scotin and ALG-2, H1299 and MCF7 cells were transfected with Scotin-FLAG tagged expression vectors in the presence or absence of HA-tagged ALG-2. Proteins from H1299 and MCF7 were extracted 24 and 48 h after transfection and analyzed by Western blot (Fig. 5). After 24 h of transfection in H1299, the human Scotin-FLAG expression vector gives rise to two bands specifically revealed by the FLAG antibody (28 and 25 kDa), while three Scotin bands are revealed in MCF7 protein extracts (28, 25 and 17 kDa) as previously observed (see Fig. 3). After 48 h of transfection, the FLAG antibody revealed three bands in H1299 (28, 25 and 17 kDa) with the 25 kDa band being the most abundant. However, 48 h after transfection in MCF7 cells, Scotin is mostly a 17 kDa protein with the 28 and 25 kDa Scotin forms are barely detectable. We observed the same patterns of bands after transfection of untagged human Scotin and mouse Scotin protein in MCF7, H1299, and U2OS cells (data not shown). The addition of MG132, an inhibitor of proteasomal degradation, calpain, cathepsin and cysteine proteases, onto Scotin transfected cells stabilized the 28/25 kDa Scotin form indicating that Scotin is cleaved in vivo by one or more of these proteases (data not shown). As the FLAG-tag is located at the C-terminus end of Scotin, we suggest that the 17 kDa bands is the result of the cleavage in the N-terminal domain of the 28/25 kDa Scotin protein giving rise to a truncated Scotin protein deleted of the cysteine-rich pro-apoptotic domain.
To determine whether stabilization of the Scotin 28/25 kDa forms was dependent on the amount of ALG-2, we cotransfected in parallel H1299 and MCF7 cells with FLAG-tagged Scotin and an increasing amount of ALG-2 or ALG-2.1 expression vectors. Cells were harvested in parallel 24 and 48 h after transfection. Twenty-four hours after transfection of H1299 cells, the 25 kDa Scotin form is accumulated in ALG-2 and in ALG-2.1 transfected cells (Fig. 5a). Twenty-four hours after transfection in MCF-7 cells, the 28, 25 and 17 kDa Scotin bands showed a higher abundance in ALG-2 than in ALG-2.1 transfected cells although ALG-2 and ALG-2.1 expression levels were similar.
Forty-eight hours after transfection in H1299 cells, the 25 kDa Scotin isoform but not the 28 and 17 kDa forms are accumulated in an ALG-2 dose-dependent manner (Fig. 5b). The variant ALG-2.1 promotes accumulation of the 28/25 kDa Scotin forms to a lesser degree. Forty-eight hours after transfection in MCF7 cells, the major Scotin form is the 17 kDa protein. However, co-transfection of ALG-2 with Scotin promotes strong accumulation of the 28, 25 and 17 kDa Scotin forms in an ALG-2 dose-dependent manner. The ALG-2.1 variant seems again to be less active than ALG-2 to promote accumulation of the 28 and 25 kDa Scotin species although ALG-2 and ALG-2.1 are expressed at similar levels. This indicates that ALG-2 and ALG-2.1 can stabilize and inhibit cleavage of Scotin. However, depending on the cell lines, ALG-2.1 may be less efficient compared to ALG-2 to stabilize Scotin.
By utilising the so far available information on ALG-2 target recognition sites, we were able to identify Scotin as a novel ALG-2 interacting protein. Comparison of the Scotin sequence with the mapped ALG-2 binding sequence in AIP1/Alix and putative ALG-2 recognition sequences in other ALG-2 targets, Annexin 7, 11, and TSG101, led us to predict that in human Scotin, it is the sequence region between position 158 and 222, which is responsible for ALG-2 binding. However, only mapping experiments will show whether this is correct. A further indication that binding takes place within this sequence is given by the finding that the interaction between ALG-2 and Scotin shows several features in common with ALG-2 binding to other target proteins. The interaction is calcium dependent and it does not take place with the ALG-2 splice variant ALG-2.1, as is also the case for the AIP1/ALIX and TSG101 interaction with ALG-2 [8,9].
By sequence alignment it was found that ALG-2 belongs to the group of penta EF-hand calcium binding proteins . Later, we showed that the affinity for calcium is in the μM range indicating that ALG-2 could sense physiological calcium transients similar to calmodulin . It has been known since many years that calcium is an important signaling molecule in apoptosis (reviewed in ). Several calcium sensing proteins such as calmodulin, S-100, calcineurin and calpain have been shown to modulate apoptotic processes [20,21]. However, only few pathways with well-defined molecular mechanisms have been established. One of these is the activation of the pro-apoptotic protein Bcl-2 family protein Bad by calcineurin, which dephosphorylates the former protein, enabling it to relocate to the mitochondria and exert its pro-apoptotic function .
The endoplasmic reticulum is the main intracellular calcium store and mechanisms that regulate intake and release of calcium from the ER are highly regulated. Disturbance of such mechanisms, e.g. failure of IP3 receptor function (reviewed in ) may lead to cell death. Several ER membrane linked proteins have been shown to be potential players in calcium-dependent apoptosis (reviewed in ). Rao et al.  have shown that ALG-2 mediates the ER stress induced apoptotic pathway. Thapsigargin but not Brefeldin A induced cell death was to some degree reduced when the ALG-2 levels were reduced by transfection of cells with ALG-2 siRNA, indicating that ALG-2 has a specific function in ER-stress induced cell death. Here we report that the ER localized Scotin, which is expressed as a 28/25 and 17 kDa protein in MCF7, H1299 and U2OS cell lines, is stabilized after transfection of ALG-2 leading to accumulation of the 28/25 kDa forms. This effect is ALG-2 specific as the variant ALG-2.1 as well as the mutant ALG-2-EF are deficient or less effective in accumulating the 28/25 kDa form of Scotin. As treatment of Scotin transfected cells with MG132, an inhibitor of cellular proteolytic activity, also stabilizes in vivo the 28/25 kDa Scotin forms, ALG-2 might protect Scotin from proteolytic degradation. Since the N-terminus of Scotin, needed for its apoptotic function , is partially lost if ALG-2 is only present at low levels, ALG-2 might modulate the apoptotic function of Scotin. Further studies will be required to determine whether the direct interaction between ALG-2 and Scotin is involved in the observed stabilization and what are the biological consequences of Scotin stabilization.
Based on our findings, we speculate that ALG-2 and Scotin are functionally linked in the regulation of signal transduction leading to calcium and ER-dependent apoptosis. One possible regulatory mechanism could be that ALG-2 is brought to the ER when calcium is transiently increased in the cell followed by interaction with and modulation of Scotin or other ER localized proteins. How the interaction of ALG-2 with Scotin would affect the function of the latter protein and/or calcium handling processes in the ER and what could be the significance of this protein interaction in apoptosis remain open questions. One possibility is the modulation of the regulation of the ER to Golgi protein transport since ALG-2 binds Sec31A, which is a component of the COPII particle . This paper reports on the interaction between Scotin and ALG-2 and the stabilization of Scotin upon ALG-2 expression. It is a first step towards the understanding of a novel and complex endoplasmic reticulum-dependent apoptotic pathway.
We thank the Danish Cancer Society, the Danish Research Council, the Lundbeck Foundation and Cancer Research UK for financial support. We thank Svetlana Tarabykina for critical reading of the manuscript and for providing the human placenta cDNA library. Ulla Mortensen is acknowledged for excellent technical help. Y.L.W. is a recipient of a Foulkes Foundation Fellowship 2005. J-C.B. is funded by Cancer Research-UK (CR-UK).
1Abbreviations used: ER, endoplasmic reticulum; GST, glutathione S-transferase; IPTG, isopropyl-1-thio-β-d-galactopyranoside; IP, immunoprecipitation.