Results

Screening for LPP-interacting proteins by yeast two-hybrid
In a previous study [13], we showed that the LIM domains of LPP are the major units for targeting LPP to focal adhesions. LIM domains are cysteine- and histidine-rich domains that form two zinc fingers capable of mediating protein-protein interactions [20,21]. However, the protein(s) that is/are responsible for the targeting of LPP to focal adhesions, i.e. protein(s) that bind(s) to the LIM domains of LPP, are not yet known. To identify protein binding partners of the LIM domains of LPP, we performed a yeast two-hybrid screening experiment. We made use of a yeast two-hybrid system that is based on transcriptional activation of two reporter genes HIS3 and LacZ whose expression is driven by upstream GAL4 DNA-binding sites. Because all three LIM domains of LPP cooperate to target LPP to focal adhesions [13], we initially focused on a screening using a bait that contained all three LIM domains. Unlike in mammalian cells, where we have shown that the three LIM domains of LPP have transcriptional activation capacity [11], this bait, although well expressed, did not activate the reporter genes in yeast cells (results not shown). This is similar to what has been found for zyxin's LIM domains [22], but in contrast to what has been found for the three LIM domains of TRIP6 that do activate reporter genes in yeast [22]. However, the bait containing all three LIM domains of LPP appeared to be very sticky since thousands of yeast colonies were obtained in which both reporter genes were activated. In an effort to reduce background activity, we deleted the first LIM domain, or the first and the second LIM domain, in the bait, leaving the two most carboxy-terminal, or the most carboxy-terminal LIM domain(s) intact, respectively. These deletions completely abolished all background activity making these baits the baits of choice to perform a library screening. Here, we report about the screening that was performed with the bait containing only the most carboxy-terminal LIM domain of LPP. As described before [13], we showed that the third LIM domain of LPP only has a very weak targeting capacity for focal adhesions. This makes it very unlikely that, by using this bait, we would pick up a protein that targets LPP to these structures, which was the initial goal of our studies. Indeed, our screening did not reveal any focal adhesion binding partners of LPP, however, in stead, we found another very interesting LPP-interacting protein as will be outlined in the following sections.
A mouse embryonal cDNA library was screened using a bait (pGBT9-LPPWT) containing the third LIM domain and carboxy-terminus of human LPP (amino acids 531–612). Among ~1.0 × 106 yeast cotransformants (Leu+ and Trp+), 56 clones were His+ of which 23 were LacZ+ too. PCR analysis of these His+/LacZ+ clones, using prey-specific insert-flanking primers, revealed that 21 of the 23 obtained clones, contained a prey-construct having a 2 kb cDNA insert (results not shown). Subsequent fragmentation of the obtained 2 kb PCR products, representing the cDNA inserts of the prey-constructs, using the HaeIII restriction enzyme (frequent cutter), indicated that all 21 isolated prey-constructs, having a 2 kb insert, were identical. The 2 kb cDNA insert of one representative prey-construct was completely sequenced and the sequence was submitted to the NCBI database (Genbank accession no. AF271735).
A BLAST (Basic Local Alignment Search Tool)-search revealed that this mouse prey-construct encoded an amino- and carboxy-terminally truncated protein comprising four PDZ domains that was almost identical to the human Scrib protein (Fig. 1B), indicating that the prey-construct represented mouse Scrib. The Scrib protein contains a set of 16 leucine-rich repeats (LRRs) near its amino-terminus and four PDZ (PSD-95, Discs large, ZO-1) domains distributed throughout the remainder of the protein (Fig. 1B). The partial mouse Scrib protein, expressed by the prey-construct, corresponded to amino acids 709 – 1242 of human Scrib (Fig. 1B).
Further analysis indicated that the isolated prey-construct, which was named pACT2-mScrib, activated the HIS3 and LacZ reporter genes of the yeast only in the presence of pGBT9-LPPWT, identifying pACT2-mScrib as a true positive (Table 1, upper three rows).
Table 1  Interaction of LPP with Scrib in the yeast two-hybrid system  Yeast cells (CG-1945), cotransformed with a bait and a prey as indicated, were selected on medium containing 5 mM 3-AT, lacking Trp, Leu and His. Yeast colonies were tested for the expression of β-galactosidase.
+ indicates strong positive interaction; - indicates no interaction.

LPP binds to the PDZ domains of Scrib via its C-terminal tail
Since the pACT2-mScrib prey-construct contained four PDZ domains, and since PDZ domains are one of the most commonly found protein-protein interaction domains in organisms from bacteria to humans [23], it was most likely that Scrib would bind to LPP via its PDZ domains. The LPP-bait that was used to screen the library was pGBT9-LPPWT containing the third LIM domain and carboxy-terminus of human LPP. Although PDZ domains have been shown to bind LIM domains [24], binding to carboxy-terminal peptides appears to be the typical mode of interaction [25]. The common structure of PDZ domains comprises six β strands (βA-βF) and two α helices (αA and αB), which fold in an overall six-stranded β sandwich [25]. The binding specificity of PDZ domains is critically determined by the interaction of the first residue of helix α B (position αB1) and the side chain of the -2 residue of the C-terminal ligand. This forms the basis for PDZ classification [25]. Since all four PDZ domains of Scrib contain a histidine at position αB1, they are classified as class I PDZ domains. Therefore, based on what has been demonstrated for this subclass of PDZ domains [25,26], the carboxy-terminal sequence of Scrib target proteins is predicted to require a hydrophobic amino acid (h) at the 0 (carboxy-terminus) position, and a serine (S) or threonine (T) at the -2 position.
Theoretically, the carboxy-terminus of the LPP protein, being -STDL, thus completely fulfils the criteria for binding to the PDZ domains of Scrib. To evaluate these predictions experimentally and to demonstrate that the binding of LPP to Scrib is specific, we performed yeast two-hybrid experiments using pGBT9-LPPWT as well as pGBT9-LPPS609A, pGBT9-LPPT610A, pGBT9-LPPD611A and pGBT9-LPPL612A as bait. The last four baits are identical to pGBT9-LPPWT except for a point mutation to alanine, respectively introduced at serine609 (-3 position), threonine610 (-2 position), aspartate611 (-1 position) and leucine612 (position 0). As prey, we used pACT2-mScrib. As summarized in Table 1, this alanine-scan mutant analysis identified threonine610 (-2 position) and leucine612 (0 position) of LPP as being essential for binding to Scrib indicating a PDZ domain-mediated specific interaction between Scrib and the carboxy-terminus of LPP.
Additional yeast two-hybrid analysis showed that LPP did not interact with Erbin, PICK1, PSD-95, Syntenin, CASK, or AF6 PDZ domains, as summarized in Table 2.
Table 2  Interaction of LPP with PDZ domains of proteins different from Scrib  Yeast cells, cotransformed with pGBT9-LPPWT and a pACT2-prey as indicated, were selected on medium lacking Leu and Trp, and either containing His or no His with 5 mM 3-AT.
+ indicates growth of yeast transformants; - indicates no growth of yeast transformants.

LPP interacts with Scrib PDZ domains in mammalian cells
We verified the Scrib-LPP interaction, which was identified in yeast cells, in mammalian two-hybrid experiments. Doing the assay in mammalian cells rather than in yeast cells, provides a more physiological environment: proteins are more likely to be in their physiological configuration, i.e. appropriately folded and modified posttranslationally, etc. Interaction between bait- and prey-proteins in a mammalian two-hybrid assay takes place in the nucleus. For an accurate performance of this assay, this means that bait- and prey-proteins should be localized in the nucleus. In contrast to the yeast assays, where we used partial bait-proteins, we wanted to use full length bait-proteins in the mammalian assay. However, since LPP contains a nuclear export signal (NES) (amino acids 117–128) in its pre-LIM region [11], we used bait-proteins in which this NES had been deleted. To verify whether deletion of the NES in LPP induced nuclear accumulation of the bait-proteins that were used in the mammalian two-hybrid assay, we introduced wild-type and mutated LPP-bait-proteins in 293T cells. While pM-LPPWT, containing GAL4-fused full length wild-type human LPP with an intact NES, was excluded from nuclei, pM-LPPdNESWT, containing GAL4-fused full length human LPP with a deletion of the NES, was accumulating in the nuclei of the cells (results not shown). These results indicate that deletion of the NES in the LPP bait proteins used in this study indeed induce nuclear accumulation of these proteins. To perform the mammalian two-hybrid experiments, we used as baits: pM-LPPdNESWT, containing full length human LPP with a deletion of its NES, and pM-LPPdNEST610A and pM-LPPdNESL612A, which are identical to pM-LPPdNESWT except for a point mutation to alanine introduced at threonine610 (position -2) and leucine612 (position 0), respectively. As determined in the yeast two-hybrid assay, each of the threonine610 and leucine612 residues is critical for the interaction with Scrib. As prey-protein, we used pSNATCH-hScribPDZ containing a part of the human Scrib protein (amino acids 669–1233) encompassing all four PDZ-domains. As summarized in Fig. 2, the interaction between wild-type full length LPP and Scrib PDZ domains resulted in high levels of luciferase reporter activity. These high levels dropped to background levels when pM-LPPdNEST610A or pM-LPPdNESL612A were used as baits in combination with Scrib as prey. The "background" levels of luciferase that were detected when pM-LPP-baits were used in combination with pSNATCH (empty prey-vector) as prey, are due to the intrinsic transcriptional activation activity of the LPP protein [11].
Figure 2  Scrib interacts with LPP in mammalian cells. pM-bait- and pSNATCH-prey-constructs were cotransfected into 293 cells in the combination indicated, together with a GAL4-regulated luciferase reporter and a CMV-β-galactosidase internal control. Cell lysates were assayed for luciferase activity 18–24 hours after transfection. Relative luciferase activity is reported as the average of three independent duplo experiments (with standard error).   These results indicate that LPP binds to Scrib PDZ domains and that this binding is abolished when amino acids at position 0 or -2 are mutated.

Development and characterization of Scrib antibodies
To analyze expression and intracellular distribution of Scrib in cultured cells, we prepared a Scrib-specific antibody (Scrib-472), as described in the Methods section. The Scrib-472 antibody recognized a protein of an apparent molecular mass of more than 200 kDa in a number of different cell extracts (Fig. 3A). Scrib was easily detected in the epithelial cell lines 293 and MDCKII, in the fibroblast cell line CV-1, and also in the T lymphocyte cell line Jurkat (Fig. 3A). These results indicate that our antibody recognizes Scrib-proteins of different species, being human (Jurkat and 293), monkey (CV-1) and dog (MDCKII). The Scrib-472 antibody also reacted with an Xpress-hScrib fusion protein produced in 293T cells transfected with the corresponding DNA (Fig. 3B). In Fig. 3B, lane 2, which depicts a Western analysis of untransfected 293T cell lysate with Scrib-antibodies, no band of endogenous Scrib is seen. Longer exposure, however, did show a band indicating that Scrib is expressed in these cells, however, 293T cells express much lower levels of endogenous Scrib as compared to 293 cells (our unpublished observations). The Scrib protein was migrating slower in SDS gels than would be expected from its theoretically calculated molecular mass (175 kDa). Possible explanations include anomalous migration per se, and posttranslational modifications. To investigate whether the Scrib-472 antibody not only recognizes denatured Scrib protein on Western blots but also is capable of detecting Scrib in fixed cells, MDCKII cells were grown to confluency on glass coverslips and stained with the Scrib-472 antibodies. From previous studies, it is known that Scrib is localized in cell-cell contacts [17]. As shown in Fig. 3C, the Scrib-472 antibody indeed is capable to detect native Scrib in cell-cell contacts in fixed cells.
Figure 3  Characterization of anti-Scrib antibodies. (A) Total cell extracts were prepared from the following cell lines: human embryonic kidney epithelial cells (293) (lane 1), dog normal kidney epithelial cells (MDCK) (lane 2), human T lymphocytes (Jurkat) (lane 3), and African green monkey kidney fibroblast cells (CV-1) (lane 4). Approximately 30 μg of protein from each extract was analysed by SDS-PAGE and Western blotting with the Scrib-472 antibodies. The position of molecular markers are as shown. (B) Total cell extracts of 293T cells, either not transfected (lane 2), or transiently transfected with Xpress-hScrib that is composed of the full length human Scrib protein fused to an Xpress-epitope-tag at its amino-terminus (lanes 1 and 3) were analyzed by SDS-PAGE and Western blotting with an anti-Xpress antibody (lane 1) or with the Scrib-472 antibody (lanes 2 and 3). The position of molecular markers are as shown. (C) MDCKII cells, grown on glass coverslips, were fixed and stained with Scrib-472-antibodies. Immunofluorescence was visualized by epifluorescence microscopy.

Scrib is not localized in focal adhesions in CV-1 and MDCKII cells, and is dispensable for targeting LPP to these structures
We have shown before that LPP is localized in cell-cell contacts [11] and also for human Scrib, it was shown that it is localized in these structures [17] (also shown in Fig. 3C and 5, upper right panel). Since LPP is not only localized in cell-cell contacts but also in focal adhesions [11,13], we investigated whether also Scrib had the ability to localize at these structures. For this purpose, we used two different cell lines: the epithelial cell line MDCKII and the fibroblast cell line CV-1. However, in contrast to LPP, Scrib could not be detected in focal adhesions as shown by staining CV-1 cells with Scrib-472 antibodies (Fig. 4, upper left panel). Identical results were obtained in MDCKII cells (results not shown). Focal adhesions were indeed present, as these structures could be stained using vinculin antibodies used as a marker for focal adhesions (CV-1 cells: Fig. 4, upper right panel; MDCKII cells: results not shown). If Scrib had been present in focal adhesions, we would have detected it there, because, as shown in Fig. 3A, Scrib is highly expressed in CV-1 as well as in MDCKII cells, and as shown in Fig. 3C, Scrib-472 antibodies are able to detect Scrib in its native conformation in fixed cells. Moreover, a hScrib-GFP protein expressed in CV-1 or MDCKII cells was never detected in focal adhesions (results not shown) but was localized in cell-cell contacts (MDCKII cells: Fig. 5, lower left panel). The nature of the nuclear staining observed in CV1-cells stained with the Scrib-472 antibody (Fig. 4, upper left panel) is aspecific, as it is also obtained with the corresponding pre-immuneserum. In addition, nuclear staining was never obtained when an hScrib-GFP protein was transiently overexpressed in these cells (results not shown). Nuclear staining was also not detected in MDCKII cells as shown in Fig. 3C and 5, upper right panel. These results indicate that, in contrast to LPP, which is localized both in focal adhesions and in cell-cell contacts in CV-1 and MDCKII cells, Scrib is only localized in cell-cell contacts but not in focal adhesions in these cells.
Figure 4  Scrib is not localized in focal adhesions in CV-1 cells, and is dispensable for targeting LPP to these structures. Upper panels: CV-1 cells, grown on glass coverslips, were double labelled with Scrib-472 antibodies (left panel) and anti-vinculin antibodies (right panel) used as a marker for focal adhesions. Lower panels: CV-1 cells were transiently transfected with wild-type human LPP (left panel), or LPP with a mutated carboxy-terminus (T610A) (right panel), as GFP-fusions. GFP-fluorescence was visualized by epifluorescence microscopy.
Figure 5  Scrib and LPP are localized in cell-cell contacts but are dispensable for targeting each other to these structures. Upper panels: MDCKII cells, grown on glass coverslips, were double labelled with anti-LPP antibodies (left panel) and anti-Scrib antibodies (right panel). Lower panels: MDCKII stable cell lines, expressing GFP-fusion proteins containing wild-type human LPP (upper left panel), LPP with a mutated carboxy-terminus (T610A) (upper right panel), human wild-type Scrib (lower left panel), or Scrib with a deletion of all its PDZ domains (lower right panel), were grown on glass coverslips (Scrib) or on Transwell-Clear polyester membranes (LPP). GFP-fluorescence was visualized by epifluorescence microscopy (Scrib) or by confocal microscopy (LPP).   As deduced from these results, we hypothesized that Scrib was not involved in targeting LPP to focal adhesions. Indeed, evidence for this hypothesis was obtained by transfecting CV-1 cells with a construct expressing GFP-LPPWT containing full length wild-type LPP, or GFP-LPPT610A, which is identical to GFP-LPPWT except for a point mutation to alanine introduced at threonine610, which abolishes binding to Scrib. No difference in focal adhesion localization could be detected between wild-type and mutated GFP-LPP fusion proteins (Fig. 4, lower panels).

Scrib and LPP are dispensable to target each other to cell-cell contacts
Since Scrib and LPP both localize in cell-cell contacts [11,17] (Fig. 5, upper panels), we investigated whether Scrib was responsible for targeting LPP to cell-cell contacts. For this, we made stable MDCKII cell lines expressing wild-type and mutated GFP-coupled forms of the LPP protein, of which the mutant form is not able to bind anymore to Scrib. However, as shown in Fig. 5, lower panels, LPP proteins that could not bind to Scrib anymore were still able to localize in cell-cell contacts in a similar way as their wild-type counterparts. These results indicate that Scrib is not responsible for targeting LPP to cell-cell contacts.
We next investigated whether LPP was responsible for targeting Scrib to cell-cell contacts. To look into this aspect, we made stable MDCKII cell lines expressing either wild-type full length Scrib-GFP or a mutated Scrib-GFP protein lacking all four PDZ domains (deletion of amino acids 725–1227). However, both the full length Scrib-GFP protein as well as the mutated form lacking all four PDZ domains localized equally well in cell-cell contacts (Fig. 5, lower panels). These results indicate that the PDZ domains of Scrib are dispensable for targeting the protein to cell-cell contacts, and as a consequence LPP is not necessary to locate Scrib in cell-cell contacts.
In summary, these results indicate that LPP and Scrib are dispensable to target each other to cell-cell contacts.

There is a direct interaction between the carboxy-terminus of LPP and the PDZ domains of Scrib
To further assess the binding between LPP and Scrib, we investigated whether there is a direct interaction between these two proteins. For this, we performed GST pull-down experiments. In vitro translated full length Scrib was tested for binding with glutathione beads, which were coupled with GST-LPP-LTWT, GST-LPP-LTL612A, or GST alone. GST-LPP-LTWT contains 40 amino acids of the pre-LIM region, the three LIM domains, and the wild-type carboxy-terminal tail of human LPP. GST-LPP-LTL612A is identical to GST-LPP-LTWT except for a point mutation to alanine introduced at leucine612 (position 0). All GST-fusion proteins as well as GST alone were expressed well in E. coli (Fig. 6A). As shown in Fig. 6B, Scrib interacted specifically with the wild-type LPP protein but not with its mutated form, GST-LPP-LTL612A or with GST alone. These results indicate that there is a specific and direct interaction between LPP and Scrib.
Figure 6  Direct interaction between the carboxy-terminus of LPP and the PDZ domains of Scrib. (A) GST fused to either wild-type LPP (40 amino acids of the pre-LIM region, the three LIM domains and the tail), or a similar LPP molecule with a mutated carboxy-terminus (L612A) and GST alone were expressed in E. coli, purified and analyzed by SDS-PAGE and Coomassie Blue staining. All proteins were expressed well. Protein markers are as indicated. (B) In vitro synthesized [35S]-methionine-labelled full length Scrib was incubated with immobilized GST or with either one of the above-described GST fusion proteins and allowed to interact over night at 4°C. After extensive washing, bound proteins were eluted in sample buffer, separated by SDS-PAGE and visualized by autoradiography. The amount of synthesized protein loaded as a reference on the gel corresponds to 10% of the input used in each binding experiment. (C) All four PDZ domains of Scrib (amino acids 616 – 1490), either wild-type or mutated as indicated, were synthesized in vitro and [35S]-methionine-labelled. these labelled proteins were incubated with immobilized GST or with GST-LPP-LTWT and allowed to interact over night at 4°C. Bound proteins were eluted in sample buffer, separated by SDS-PAGE and visualized by autoradiography. The amount of synthesized protein loaded as a reference on the gel corresponds to 10% of the input used in each binding experiment.   To further investigate the requirements in the Scrib protein for binding to LPP, we performed additional GST pull-down experiments. From our previously described experiments (yeast and mammalian two-hybrid), it was clear that the PDZ domains of Scrib bind to LPP. These findings were confirmed by using GST pull-down: as shown in Fig. 6C, upper panel, a portion of the Scrib protein encompassing all four PDZ domains was efficiently pulled down by GST-LPP-LTWT. To find out which of the four PDZ domains of Scrib was responsible for the observed interaction with LPP, we mutated the PDZ domains of Scrib, one at the time, by destroying their carboxylate binding loop (LG → AE), and tested how efficiently these mutated proteins were pulled down by GST-LPP-LTWT. From the results, which are presented in Fig. 6C, we can conclude that all four PDZ domains of Scrib more or less contribute to the binding to LPP, but that PDZ 3 is most important, since binding to GST-LPP-LTWT was almost completely abolished when the carboxylate binding loop of this PDZ domain was destroyed.

Scrib can target LPP to an ectopic location in vivo through its PDZ domains
Evidence for an in vivo interaction between Scrib and LPP was obtained by performing mitochondrial targeting experiments. We tested if Scrib was sufficient to recruit LPP to an ectopic location in living cells. The membrane anchor of the ActA sequence has been shown previously to be sufficient to target proteins expressed in mammalian cells to the surface of mitochondria [27,28]. This ectopic localization allows testing ligand recruitment in vivo. For this purpose, we generated a chimera named Xpress-hScrib-mito made up by an Xpress-epitope tag fused to the amino-terminus of human full length Scrib and linked in frame to the membrane anchor of the Listeria monocytogenes protein ActA (mito). Expression of this construct was confirmed by Western blotting with the use of an anti-Xpress antibody (results not shown). CV-1 cells were transiently transfected with Xpress-hScrib-mito and full length wild-type or carboxy-terminally mutated LPP green fluorescent protein fusions. Cells were stained with an anti-Xpress antibody and examined by fluorescence microscopy. In all transfected cells, the Xpress-hScrib-mito chimera localized to mitochondria, as shown in Fig. 7, left upper and middle panels. As shown in Fig. 7, upper right panels, wild-type LPP can be recruited to Xpress-hScrib-mito on mitochondria. This recruitment of LPP to Xpress-Scrib-mito-coated mitochondria was completely abolished when the carboxy-terminus of LPP was mutated (Fig. 7, middle panels).
Figure 7  Scrib can recruit LPP to an ectopic location in vivo through its PDZ domains. CV-1 cells were transiently co-transfected with Xpress-hScrib-mito or Xpress-hScribdPDZ-mito, and GFP-fusions of wild-type full length human LPP, or LPP with a mutated carboxy-terminus (T610A). Xpress-hScrib-mito and Xpress-hScribdPDZ-mito are composed of the human full length Scrib protein with or without its PDZ domains, respectively, which is fused to an Xpress-epitope-tag at its amino-terminus, and to an ActA-derived mitochondrial membrane anchor at its carboxy-terminus, Cells were stained with an anti-Xpress antibody to detect Xpress-Scrib(dPDZ)-mito. Immunofluorescence and GFP were visualized by epifluorescence microscopy. The focal adhesion localization of the GFP-LPP proteins is not visible in these pictures because a focal plane corresponding to mitochondrial staining is shown.   To investigate the importance of the PDZ domains of Scrib in this recruitment of LPP, we deleted all four PDZ domains (amino acids 724–1192) from Xpress-hScrib-mito (=Xpress-hScribdPDZ-mito) and tested whether this PDZ-less protein still was able to recruit LPP to mitochondria. As shown in Fig. 7, lower panels, Xpress-hScribdPDZ-mito lost its ability to recruit LPP to mitochondria. These results indicate that Scrib can recruit LPP to an ectopic location in vivo, and that the PDZ domains of Scrib are an absolute requirement for this activity.
As mentioned above, the Xpress-hScrib-mito chimera localized to mitochondria in all cells that expressed this protein. However, LPP, which was co-expressed, was only recruited to mitochondria in a small fraction of these cells. This issue will be further addressed in the Discussion section.