Lasp-1 has been identified as a signaling molecule that is phosphorylated upon elevation of [cAMP]i in pancreas, intestine and gastric mucosa and is selectively expressed in cells within epithelial tissues. In the gastric parietal cell, cAMP-dependent phosphorylation induces the partial translocation of lasp-1 to the apically directed F-actin-rich canalicular membrane, which is the site of active HCl secretion. Lasp-1 is an unusual modular protein that contains an N-terminal LIM domain, a C-terminal SH3 domain and two internal nebulin repeats. Domain-based analyses have recently categorized this protein as an epithelial representative of the nebulin family, which also includes the actin binding, muscle-specific proteins, nebulin, nebulette and N-RAP.
In this study, we show that lasp-1 binds to non-muscle filamentous (F) actin in vitro in a phosphorylation-dependent manner. In addition, we provide evidence that lasp-1 is concentrated within focal complexes as well as in the leading edges of lamellipodia and the tips of filopodia in non-transformed gastric fibroblasts. In actin pull-down assays, the apparent Kd of bacterially expressed his-tagged lasp-1 binding to F-actin was 2 μM with a saturation stoichiometry of ∼1:7. Phosphorylation of recombinant lasp-1 with recombinant PKA increased the Kd and decreased the Bmax for lasp-1 binding to F-actin. Microsequencing and site-directed mutagenesis localized the major in vivo and in vitro PKA-dependent phosphorylation sites in rabbit lasp-1 to S99 and S146. BLAST searches confirmed that both sites are conserved in human and chicken homologues. Transfection of lasp-1 cDNA encoding for alanine substitutions at S99 and S146, into parietal cells appeared to suppress the cAMP-dependent translocation of lasp-1 to the intracellular canalicular region. In gastric fibroblasts, exposure to the protein kinase C activator, PMA, was correlated with the translocation of lasp-1 into newly formed F-actin-rich lamellipodial extensions and nascent focal complexes. Since lasp-1 does not appear to be phosphorylated by PKC, these data suggest that other mechanisms in addition to cAMP-dependent phosphorylation can mediate the translocation of lasp-1 to regions of dynamic actin turnover. The localization of lasp-1 to these subcellular regions under a range of experimental conditions and the phosphorylation-dependent regulation of this protein in F-actin rich epithelial cells suggests an integral and possibly cell-specific role in modulating cytoskeletal/membrane-based cellular activities.
The actin-based cytoskeleton is essential for cellular activities ranging from maintenance of cell morphology to cellular locomotion and also participates in regulation of secretion, endocytosis and transmembrane signaling ( Mitchison and Cramer, 1996; Schmidt and Hall, 1998). The assembly of this cytoskeleton is regulated by multiple actin-binding proteins which allow for the diversity of actin filament forms and functions such as cytoskeletal remodeling, actin bundling and branching ( Ayscough, 1998). Although much attention has been focused on the role of tyrosine kinases in cytoskeletal organization, there is also substantial evidence that the elevation of cAMP and subsequent activation of cAMP-dependent protein kinase (PKA) alters the morphology of epithelial cells and fibroblasts. Indeed, a cAMP-induced alteration in the shape of epithelial cells was noted as early as 1966 ( Yasumura et al., 1966). Such changes in morphology have been particularly well documented in the terminally differentiated gastric parietal cell in which elevation of [cAMP]i induces dramatic changes in the actin cytoskeleton which are correlated with the activation of HCl secretion (reviewed by Forte and Yao, 1996). In some cultured cells, elevation of [cAMP]i also leads to the loss of stress fibers and focal adhesions ( Schoenwaelder and Burridge, 1999).
Lasp-1 is a recently identified cAMP-dependent signaling protein that may also be involved in tyrosine kinase signaling ( Chew et al., 1998; Schreiber et al., 1998). It is widely expressed, but differentially distributed, in normal epithelial tissues and in brain ( Chew et al., 2000; Chew et al., 1998; Schreiber et al., 1998). There is a particularly prominent expression of this protein in the gastric parietal cell and in other F-actin rich ion-transporting cells including pancreatic and salivary duct cells as well as certain distal tubule and collecting duct cells in the kidney ( Chew et al., 2000). In the parietal cell, elevation of intracellular cAMP ([cAMP]i) induces a partial translocation of lasp-1 to the apically directed F-actin rich intracellular canaliculus, which is the site of active HCl secretion. This stimulus-associated phosphorylation and translocation of lasp-1 to the canalicular region suggests that lasp-1 may play role in the regulation of actin cytoskeleton plasticity and, possibly vesicle trafficking ( Chew et al., 2000).
Lasp-1 was initially identified as pp40, a phosphoprotein that migrated on SDS-PAGE gels with an apparent molecular mass of ∼40 kDa ( Chew and Brown, 1987). Phosphorylation of pp40 was increased in gastric parietal cells following elevation of [cAMP]i and was correlated with histamine H2-receptor-activation of HCl secretion. Subsequently, pp40 was isolated, sequenced and cloned ( Chew et al., 1998) and shown to be identical to lasp-1 (LIM and SH3 domain-containing protein), a product of the human gene, MLN 50, which is amplified in some cancers ( Tomasetto et al., 1995). In addition to an N-terminal LIM domain and a carboxyl terminal SH3 domain, lasp-1 contains two nebulin repeats. Although the specific cellular functions of lasp-1 have not been defined, the presence of several major protein-interacting motifs predicts multiple binding partners. Sequence homology comparisons as well as analyses of physical characteristics further suggest that one or more these interacting proteins is likely to be cytoskeletal ( Chew et al., 1998; Schreiber et al., 1998). In this regard, actin is a strong candidate because lasp-1 has been localized to non-stress fiber, actin-rich subcellular regions ( Chew et al., 2000; Schreiber et al., 1998) and also reportedly associates with actin on blot overlays and in GST pull down assays ( Schreiber et al., 1998).
The initial goals of this study were to define the actin binding properties of lasp-1 and to determine if phosphorylation can modulate the interaction. Our results demonstrate that lasp-1 binds to filamentous (F) actin and that cAMP-dependent phosphorylation modifies this interaction in vitro. In the course of these experiments, lasp-1 was found to be highly expressed not only in the gastric parietal cell but also to be present in focal adhesions and focal complexes as well in the extreme tips of lamellipodia and filopodia in gastric mucosal fibroblasts. Since these subcellular regions are rich in F-actin and are associated with a range of activities, including cell migration and membrane trafficking, our results suggest that lasp-1 may play an important signaling-dependent role in the regulation of one or more of these processes.
Materials and Methods
Mixed gastric mucosal cells and parietal cells were isolated from nembutal-anaesthetized, male New Zealand white rabbits and placed in primary culture as previously described ( Chew et al., 1989). Madin Darby kidney (MDCK) cells, passage 20-25, were cultured using standard conditions. Transfection of various plasmids into gastric cells and MDCK cells was accomplished using Effectene (Qiagen) as previously described ( Parente et al., 1999).
DNA constructs and bacterial protein expression
Polyhistidine (his)-tagged lasp-1 protein was generated using the pET15b expression vector (Novagen, Madison, WI) by transformation into BL21(DE3)pLysS bacteria (Promega, Madison, WI) and isopropylthio-(-D-galactoside) induction. His-tagged protein was purified on a Hi Trap chelating column (Amersham-Pharmacia Biotech, Piscataway, NJ) followed by Mono Q purification as previously described ( Chew et al., 1998). Immediately after purification, proteins were dialyzed, aliquoted and lyophilized. Glutathione-S-transferase (GST)-lasp-1 fusion protein was generated by inserting cDNA encoding for the rabbit lasp-1 open reading frame downstream of the GST fragment in the pGEX4T-3 vector (Pharmacia Biotech, Piscataway, NJ) with BamHI and EcoRI restriction sites at the 5′ and 3′ ends, respectively. Bacterial transformations and inductions were performed with lasp-1-pGEX4T-3 and empty pGEX4T-3 plasmid (to generate GST protein). A similar procedure was used to generate mutated lasp-1 GST fusion proteins (below). GST-tagged proteins were purified using glutathione-sepharose 4B beads (Pharmacia Biotech).
For expression in MDCK cells, constructs containing an N-terminal hemagglutinin (HA) tag upstream of the coding region for lasp-1 were subcloned into the pcDNA3 vector (Invitrogen, Carlsbad, CA). Constructs were generated by PCR-amplification using an Advantage®-HF 2 PCR kit (Clontech, Palo Alto, CA). The pET15b plasmid containing lasp-1 cDNA served as the template for primers that generated BamHI and EcoRI restriction sites at the 5′ and 3′ ends, respectively. Primers based on the rabbit lasp-1 cDNA sequence (GenBank accession # AF017438) were designed with OLIGO Primer Analysis software, version 6.6 for MacIntosh (National Biosciences, Plymouth, MN) and synthesized by Gibco BRL Life Sciences, PCR conditions were as follows: Sense primer: 5′ GCC GGA TCC ACC ATG GGC TAC CCA TAC GAT GTT CCA GAT TAC GCT AAC CCC AAC TGC GCC; anti-sense primer: 5′ GGC CGA ATT CTC AGA TGG CTT CCA CGT AGT T; Initial denaturation, 94°C, 30 seconds followed by 35 cycles of 94°C, 30 seconds; 60°C, 30 seconds; 72°C, 45 seconds and a final 10 minute extension at 72°C. PCR products were gel isolated and ligated into EcoRI and BamHI-digested pcDNA3 vector. After transformation into Escherichia coli JM109 bacteria, plasmids containing lasp-1 cDNA were isolated (Qiagen Miniprep kit, Valencia, CA). Positive clones were identified by PCR. The sequences of all clones were confirmed prior to use (Medical College of GA Core DNA Sequencing Facility; ABI Prism 377 automated DNA sequencer; ABI Prism Cycle Sequencing Dye Terminator Ready Reaction kits). For transfections, plasmid DNA was isolated and purified with Qiagen Maxiprep Endo-Free kits as previously described ( Parente et al., 1999).
Site-directed mutagenesis was performed with the Stratagene Quick-Change mutagenesis kit using pcDNA3 vector containing rabbit lasp-1 cDNA as a template. Primers for single serine to alanine substitutions were as follows: S146 (RRDA): sense, 5′ CGA GCG CCG GGA CGC CCA GGA CAG CAG C; antisense, 5′ GCT GCT GTC CTG GGC GTC CCG GCG CTC G; S99 (RGFA): sense, 5′ GGG CAG AGG CTT CGC CGT GGT GGC AGA C; antisense, 5′ GTC TGC CAC CAC GGC GAA GCC TCT GCC C. PCR reactions were performed with Pfu taq using the following conditions: 95°C, 1 minute then 16 cycles; 95°C, 30 seconds, 55°C, 1 minute, 68°C, 12 minutes. After 16 cycles, 1 μl (10 U/μl) DPNI restriction enzyme was added to the reaction and, after a brief centrifugation, samples incubated at 37°C for ∼2 hours to digest supercoiled dsDNA. Vectors were transformed into Epicurean Col:XL1 Blue Super Competent cells as per manufacturer's instructions. Plasmid DNA was isolated and sequences of all constructs confirmed after mutagenesis as described above. For double S99/S146 (RRDA/RGFA) mutants, the same strategy was employed using the RRDA mutant in pcDNA3 vector as the starting material. For in vitro experiments with his-tagged lasp-1 mutants, pcDNA3 plasmids containing the appropriately mutated inserts were transferred to the pET15b plasmid using a PCR-based approach (Advantage®-HF 2 PCR kit) as described above with oligodeoxynucleotide primers containing NdeI (5′) and BamHI (3′) restriction sites respectively as follows: 5′ GGG AAT TCA TAT GAA CCC CAA CTGC GCC CGG TG and 5′ CCG GAT CCT TCA GAT GGC TTC CAC GTA GTT GGC A.
GST `pull down' assays were performed at 4°C using parietal cell extracts that were prepared by incubating freshly isolated cells for 15 minutes in lysis buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM EGTA, 1% NP-40, 0.1% SDS, 10 mM NaF containing the following inhibitors: 0.2 mM AEBSF, 5 mM benzamidine, 10 μg/ml each of leupeptin, pepstatin). Cellular debris was removed by centrifugation (10 minutes, 10,000 g) and supernatants (0.5-1 mg protein in 0.5 ml) precleared by incubating with 150 μl of a PBS-washed, 50% GST-sepharose bead slurry for 1 hour on a nutator. Precleared supernatants were collected by centrifugation (500 g, 5 minutes) then incubated (90 minutes, nutator) with lasp-1-GST fusion protein (30 μg) bound to washed glutathione-sepharose beads (7.5 μl bed volume) as per manufacturer's instructions. All samples were run in duplicate with the following controls: GST protein + precleared cell lysate + sepharose beads; precleared cell lysate + sepharose beads; GST lasp-1 + sepharose beads. Beads were harvested by centrifugation (1500 g, 15 seconds), washed three times with cell lysis buffer then solubilized with 40 μl of 2× SDS stop buffer. Supernatants were analyzed on SDS-PAGE gels (8-12%), which were either silver stained (Investigator™ Silver stain kit, ESA, Chemsford, MA) or with a modified Coomassie Blue colloidal staining protocol ( Chew et al., 1998), and by western blot with enhanced chemiluminescent (ECL) detection as previously described ( Chew et al., 2000). Replicate western blots were probed for lasp-1 (anti-lasp-1 monoclonal antibody (mab, clone 3H8, diluted 1:1000) ( Chew et al., 2000) and actin (anti-non-muscle actin mab, clone AC40, Sigma Aldrich, ST Louis, MO, diluted 1:5000). ECL detection was performed with HRP-conjugated sheep anti-mouse Ig (1:5000 dilution, Amersham Pharmacia, Piscataway, NJ) as the secondary antibody.
Actin interaction assays
Actin co-sedimentation assays were performed with bacterially-expressed lasp-1. Monomeric (G)-actin was generated by incubating human platelet actin (≥99% purity, 5:1 β/γ isoforms, Cytoskeleton, Denver, CO) at a concentration of 1 μg/ml in a buffer containing 5 mM Tris-HCl, pH 8.0, 0.2 mM CaCl2, 0.2 mM ATP, 0.5 mM DTT for 1-2 hours, 4°C. Lyophilized his-tagged or GST-tagged lasp-1 was dissolved in the same buffer at a concentration of 1-2 μg/ml then centrifuged (100,000 g, 1 hour, Airfuge (Beckman Instruments) to remove protein aggregates. Lasp-1 (0.5-16 μM) in the resulting supernatants was incubated with G-actin (14-23 μM, 40 minutes, 4°C) then polymerized by addition actin polymerization buffer (2 mM Tris, pH 8, 50 mM KCl, 2 mM MgCl2, 1 mM ATP). After incubation for 30 minutes, room temperature to allow actin polymerization to reach a steady state, samples were centrifuged (100,000 g, 1 hour) to pellet F-actin. Supernatants (which contained G-actin) and pellets were dissolved in SDS-PAGE buffer and resolved on 8% SDS-PAGE gels. Gels were stained with Coomassie® Brilliant Blue R250. Images of destained gels were digitized with a Syngene Gene Genius system and bands quantitated with Gene Tools software (Synoptics, UK) using BSA as a standard. BSA (Sigma) and α-actinin (Cytoskeleton) were used respectively as negative and positive controls for F-actin co-sedimentation. In preliminary experiments in which actin was polymerized prior to lasp-1 addition, similar amounts of lasp-1 were found to co-sediment with F-actin as compared to experiments in which actin was polymerized after lasp-1 addition. This latter approach was used in all subsequent experiments to avoid quantitation problems associated with the transfer of small quantities of F-actin.
To assess the association of endogenous lasp-1 with F-actin, a modification of previously described methods was used ( Weed et al., 2000). Parietal cells were temperature equilibrated then rapidly pelleted, rinsed in cold PBS then lysed by sonicating cells (3×10 seconds, 4°C) in a lysis buffer containing 10 mM imidazole, pH 7.2, 75 mM KCl, 5 mM MgCl2, 1 mM EDTA, 0.5 mM DTT plus proteolytic inhibitors (mini EDTA-free tablet, Roche Diagnostics, Mannheim, Germany). After centrifugation (30 minutes, 100,000 g, 4°C), the resulting supernatants were preincubated with platelet-derived actin (5-8 μM). Samples were sedimented as for the recombinant protein following actin polymerization. Lasp-1 associated with F-actin was detected by western blot using the 3H8 monoclonal antibody and ECL detection as described above. Chemiluminescent signals on western blots were quantitated with a Syngene GeneGnome 16 bit CCD-based chemiluminescent detection system and Gene Tools software (Synoptics).
In vitro and in vivo phosphorylation site analyses
To generate phosphorylated lasp-1 for actin co-sedimentation assays, lyophilized his- or GST-tagged lasp-1 (0.5-1 μg/μl) was dissolved in 1× cAMP-dependent protein kinase buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl2). Following addition of recombinant cAMP-dependent protein kinase (catalytic subunit, New England Biolabs, Beverly, MA; 1-2 U/μg lasp-1), reactions were initiated with 1 mM ATP and continued for 10 minutes, 30°C. For controls, 1 μg synthetic rabbit protein kinase inhibitor (PKI, Sigma Chemicals, St Louis, MO) was included in the reaction mixture. 32P-labeling using [γ-32P]ATP as a substrate (0.2 mM, specific activity 400 cpm/pmol) was used to confirm that this concentration of PKI completely blocked the phosphorylation of lasp-1 by cAMP-dependent protein kinase. At the end of the incubation period, samples were immediately dialyzed (20× volume of 5 mM Tris-HCl, pH 8.0) and concentrated by centrifugation (2500 g, 4°C) in Centricons (Amicon, Beverly, MA). To prevent any further phosphorylation in the ATP-containing actin polymerization buffer, PKI was added to samples in which it was not added initially. Phosphorylated lasp-1 was incubated with actin and polymerization performed as described above. Because recombinant protein kinase migrates close to the phosphorylated form of his-tagged lasp-1 on SDS-PAGE gels, western blot analyses were also performed using the lasp-1 mab in conjunction with ECL detection as above. Quantitation was performed using multiple film exposures to ensure film linearity as previously described ( Chew et al., 2000) or with the Syngene GeneGnome system as above.
To identify in vitro cAMP-dependent protein kinase phosphorylation sites, 20 μg of his-tagged lasp-1 was phosphorylated with 15 U recombinant cAMP-dependent protein kinase catalytic subunit (New England Biolabs) for 5 minutes, 30°C in kinase buffer (25 mM HEPES, pH 7.4, 10 mM Mg (C2H3O2)2, 0.33 mM DTT) containing 0.1 mM [γ-32P]ATP (specific activity, 36,000 cpm/pmol). Phosphorylated protein was resolved on an 8% SDS-PAGE gel and subjected to `in gel' tryptic digestion as previously described ( Chew et al., 1998; Parente et al., 1996). Peptides in digests were resolved on a μRPC C2/C18 column (0-40% linear acetonitrile gradient, 100 μl/minutes) using a Pharamacia SMART system then re-purified on the same column at a flow rate of 50 μl/minutes. Peaks containing radiolabeled peptides were identified by Cerenkov counting of each fraction ( Parente et al., 1996). Sequencing by Edman degradation was performed at the Emory University Microchemical Core Facility, Atlanta, GA. This facility also performed mass spectrum analyses comparing signals in V8 digests of His-tagged lasp-1 before and after phosphorylation with recombinant cAMP-dependent protein kinase catalytic subunit.
Time course experiments were performed by adding [γ-32P]ATP (final concentration 0.2 mM, SA, 400 cpm/pmol) to temperature-equilibrated (5 minutes, 30°C) assay tubes containing 10 μg histagged lasp-1 plus recombinant cAMP-dependent protein kinase catalytic subunit (20 U) in kinase buffer. Aliquots (∼1 μg lasp-1) were withdrawn at different time points, placed in an equal volume of 2× SDS stop solution, boiled for 3 minutes and subjected to SDS-PAGE. Radiolabeled bands were located by autoradiography, excised and quantitated by Cerenkov counting ( Parente et al., 1996).
In vivo phosphorylation site analyses were performed with MDCK cells. Exponentially growing cells were transfected with the pcDNA3 vector containing HA-tagged wild-type lasp-1 cDNA and cDNA from phosphorylation site mutants using Effectene (Qiagen) as previously described ( Parente et al., 1999). Forty-eight hours later, cells were incubated with forskolin (10 μM, 15 minutes) or an equal volume of DMSO vehicle. For one dimensional (1D) Mr band shift analyses, cells were rinsed in cold PBS and immediately lysed in 1× SDS stop buffer. Lysates were fractionated on SDS-PAGE gels and lasp-1 detected by western blot with ECL detection as previously described ( Chew et al., 2000). For two dimensional (2D) analyses, cells were lysed with hot 0.3% SDS-1% βME, 10 mM Tris, pH 7.4. Lysates were precipitated at room temperature with 4× volumes of acetone and redissolved in rehydration buffer (8 M urea, 2% CHAPS, 18 mM DTT, 0.5% IPG buffer, pH 3-10 (Amersham-Pharmacia), 0.001% bromphenol blue). First dimension IEF was performed on IPG strips (pH 3-10 NL or L) with an IPGPhor (Amersham-Pharmacia) as follows: 12 hour rehydration; 500 V, 1 hour; 1000 V, 1 hour; 8000 V→28,000 volt hours. For second dimension SDS-PAGE, strips were incubated for 15 minutes, room temperature in SDS equilibration buffer (50 mM Tris, pH 8.8, 6 M urea, 30% glycerol, 2% SDS, 65 mM DTT, 0.001% bromphenol blue). Resolved proteins were transferred to nitrocellulose for western blot analyses of lasp-1 using the lasp-1 mab and ECL detection. Lasp-1 phosphorylation was defined as an acidic shift resulting from an addition of negatively charged phosphate residue(s) to the protein (the predicted acidic shifts were confirmed in metabolic 32P labeling experiments as previously described ( Chew et al., 1998). Phosphorylation site analyses were performed with the Phosphobase program on the Center for Biological Sequence Analysis (CBS) web site ( Kreegipuu et al., 1999).
Indirect immunofluorescence microscopy
Endogenous lasp-1 was localized by indirect immunofluorescence (primary antibody, anti-lasp-1 mab 3H8); secondary antibody, cyanine (Cy)-5-labeled goat anti-mouse IgG (Jackson Immunoresearch Labs, Westgrove, PA) with primary and secondary antibody controls as previously described ( Chew et al., 2000). In brief, gastric cells grown on glass coverslips were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, blocked with 5% non-fat milk (BioRad) in PBS and sequentially incubated with the lasp-1 antibody (diluted 1:50 in 1% milk/PBS) followed by the Cy-5-labeled secondary antibody (diluted 1:100 in 0.1% milk/PBS). PBS rinses (3-6×5 minutes) were performed after each incubation step. In most experiments, cells were dual labeled for F-actin by adding Oregon Green phalloidin (1:400 dilution, Molecular Probes, Eugene, OR) simultaneously with the secondary antibody. Transfected lasp-1 and mutants were immunolocalized using a similar protocol with monoclonal anti-HA antibody (BabCo/Covance, 1:1,000), Alexa 488 chicken anti-mouse secondary antibody (Molecular Probes, 1:100). Dual labeling for F-actin was accomplished using Alexa 647-labeled phalloidin Molecular Probes, 1:400). Fluorescently labeled cells were optically sectioned using a Molecular Dynamics 2010 confocal microscope equipped with a krypton/argon laser ( Chew et al., 2000).
Where appropriate, values are expressed as means±s.e.m. with n representing the number of independent experiments. For paired samples, data was analyzed for statistical significance using the Student's t-test for paired comparisons. Analysis of variance and Dunnett's tests were used to analyze multiple comparisons ( Chew and Brown, 1987).
Lasp-1 binds to non-muscle F-actin in vitro
Detailed domain-based analyses coupled with the inability to detect lasp-1 expression in skeletal, cardiac and non-vascular smooth muscle have led us to propone this protein as an epithelial representative of the nebulin repeat family of proteins ( Chew et al., 2002). The primary member, nebulin, is a large (∼600-800 kDa) skeletal muscle protein that is thought to play a role in defining the length of thin filaments. Nebulin repeat fragments bind to actin ( Wang, 1996) and also promote actin polymerization and bundling ( Chen et al., 1993; Gonsior et al., 1998). Although lasp-1 contains only two nebulin repeats and does not appear to modulate either actin polymerization or bundling ( Chew et al., 2002), it has been reported to bind to actin on blot overlays and in GST pull down assays ( Schreiber et al., 1998).
To characterize further the interactions between lasp-1 and actin, actin co-sedimentation assays were performed using highly purified, platelet-derived non-muscle actin in order to mimic epithelial cell physiology as closely as possible. In initial experiments we confirmed that both endogenous and bacterially expressed lasp-1 cosediment with F-actin ( Fig. 1A,B). The well-characterized actin binding protein, α-actinin, also co-sedimented with F-actin in these assays but BSA, which does not bind actin, did not ( Fig. 1A). Cumulative data from several independent experiments demonstrated reproducible and saturable lasp-1-F-actin co-sedimentation that was significantly greater than controls ( Fig. 1C).
To determine if lasp-1 also binds to monomeric (G)-actin, a range of assay conditions were tested including GST pull down assays with GST-lasp vs GST alone, blot overlay assays with 32P-labeled lasp-1, and far western blots using his-tagged lasp-1 in conjunction with the anti-lasp-1 monoclonal antibody. No actin binding was detected in blot overlays or far westerns (not shown). Although actin was present in GST pull downs, there was no significant difference between samples containing GST-lasp vs GST alone ( Fig. 1D). Thus, a variety of approaches indicated that lasp-1 does not bind to monomeric actin; however, additional studies are required to establish this point unequivocally.
Identification of major cAMP-dependent phosphorylation sites in vitro and in vivo by microsequencing, mass spectrometry and site-directed mutagenesis
Before testing the effects of phosphorylation on the interaction between recombinant lasp-1 and F-actin, it was essential to define the pattern of in vitro phosphorylation and to determine if the phosphorylation sites targeted by PKA in vitro were the same as those regulated by cAMP in vivo. In intact cells, lasp-1 is phosphorylated on serine residues following elevation of [cAMP]i ( Chew et al., 1998). This could be the result or either a direct or an indirect involvement of cAMP-dependent protein kinase (PKA). Previous work predicted that lasp-1 is a direct substrate for PKA because the recombinant his-tagged protein was strongly phosphorylated by recombinant cAMP-dependent protein kinase catalytic subunit, but not by several other serine/threonine kinases including protein kinase C ( Chew et al., 1998). As shown in Fig. 2A, the kinetics of lasp-1 phosphorylation with PKA (recombinant catalytic subunit) is also rapid and monophasic, reaching a maximum within ∼30 minutes.
Two cAMP-dependent protein kinase phosphorylation consensus sites, both of which contain serine residues, have been identified in the rabbit lasp-1 protein ( Chew et al., 1998; Chew et al., 2002). Blast searches of the GenBank have confirmed that these sites are conserved in the human (NM_006148) and chicken (#BI394039) homologues but only the K/RGFS99 site is conserved in rat (NM_032613) and mouse (NM_010688) (C.S.C., unpublished). As shown in the diagram in Fig. 2A, the first consensus phosphorylation site falls between the two nebulin repeats (RGFS99) and the second is immediately downstream of this region (RRDS146). To confirm that these sites are, indeed, targeted by PKA, his-tagged lasp-1 was incubated with recombinant cAMP-dependent protein kinase catalytic subunit (using [γ-32P]ATP as a substrate) then subjected to tryptic digestion followed by micro-HPLC purification and microsequencing (as described in Materials and Methods). Both PKA consensus sites were tentatively identified within the three major peaks containing phosphorylated peptides ( Fig. 2B). Mass spectrum analyses were also performed on the same preparation of his-tagged lasp-1 before and after phosphorylation with recombinant PKA. Nanocolumn eluates were analyzed using a nanospray device and positive, negative and phosphate ion scans. Positive and negative scans revealed that proteins were digested and fragments produced. Analysis of his-tagged lasp-1 revealed only background signal in the phosphate ion scans. Significant phosphate ion peaks were identified in the 50% methanol eluate of phosphorylated lasp-1. Based on the predicted proteolytic map, two proteolytic fragments were identified with a high degree of certainty (GFS99VADTPELQR) and (MGPSGGEGAEPERRDS146QDSSNYR).
Because other serine residues were present in the sequenced peptides and in one of the two mass spectrum-based phosphate analysis products, neither sequencing nor mass spectrum analyses unequivocally identified Ser99 and Ser146 as specific phosphorylation sites. Therefore, additional analyses were performed with mutated his-tagged lasp-1 proteins in which one or both of these putative cAMP-dependent protein kinase phosphorylation sites were mutated to alanines. As shown in Fig. 3A, cAMP-dependent protein kinase catalyzed 32P incorporation into both RRDA146 and RGFA99 mutants (65% and 26% of total 32P incorporation into wild-type lasp-1, respectively). There was also a lesser degree of 32P incorporation into the double (R/R) mutant. The combined data supported the conclusion that the most significant in vitro phosphorylation catalyzed by cAMP-dependent protein kinase occurs at the Ser99 and Ser146 residues with Ser99 being the most prominent site.
Mutation of Ser146, but not Ser99, to alanine blocked the phosphorylation-induced Mr band shift ( Fig. 3A). Thus, this band shift, which has been observed both in vivo and in vitro ( Chew et al., 2000; Chew et al., 1998), appeared to result from the phosphorylation of Ser146. To test this hypothesis more directly, MDCK cells were transiently transfected with pcDNA3 plasmids containing HA-tagged wild-type lasp-1 cDNA or lasp-1 cDNA containing S→A substitutions for Ser99 and Ser146 respectively. Transfected cells were incubated with forskolin (10 μM, 15 minutes) to elevate endogenous cAMP and to activate endogenous cAMP-dependent protein kinase. Cell lysates were analyzed for lasp-1 phosphorylation by western blot. [The lasp-1 mab (clone 3H8) was used for direct analysis of the expressed rabbit lasp-1 protein because this antibody does not recognize the endogenous canine protein (see mock transfection lanes in Fig. 3B).] In these experiments, S→A mutation of the RRDS146, but not the RGFS99 site, did indeed block the Mr band shift of lasp-1 ( Fig. 3B). This experimental approach confirmed that Ser146 is an in vivo target for cAMP-dependent protein kinase and that phosphorylation of this site induces a Mr band shift.
To determine if the RGFS99 site ( Fig. 3A) is also targeted by cAMP-dependent protein kinase in vivo, pcDNA3 plasmids containing either wild-type lasp-1, the RRDA146 construct, or the double mutant (R/R) construct were transfected into MDCK cells. Cells were stimulated with forskolin (as above), lysed and subjected to 2D gel electrophoresis followed by western blot analysis with the anti-lasp-1 mab. Fig. 3C demonstrates that forskolin stimulation of cells transfected with wild-type lasp-1 led to the expected shift in apparent Mr which was accompanied by an acidic shift (presumably reflecting the increased negative charge of the phosphate groups). With the RRDA146 mutant, there was an acidic shift, but no shift in apparent Mr. Both the acidic and Mr shifts were abolished in the double mutant. The combined results from microsequencing, mass spectrum and mutational analysis studies, therefore, demonstrate that Ser99 and Ser146 are the major in vitro targets of cAMP-dependent protein kinase and are consistent with direct PKA phosphorylation of these sites in vivo.
Phosphorylation of lasp-1 inhibits cosedimentation with F-actin in vitro
To determine if phosphorylation of Ser99 and Ser146 by cAMP-dependent protein kinase modifies the interaction between lasp-1 and F-actin, his-tagged lasp-1 was phosphorylated with recombinant PKA prior to actin cosedimentation assay. As shown in Fig. 4, phosphorylation suppressed the interaction (as detected by both Coomassie blue staining ( Fig. 4A) and western blot (not shown)). In more detailed kinetic analyses, his-tagged lasp-1 was found to bind to F-actin in a saturable, concentration-dependent manner with a Kd of 2.2±0.3 μM (n=4) and stoichiometry of 1 mol lasp-1:7 mol actin ( Fig. 4B). PKA-dependent phosphorylation of his-tagged lasp-1 increased the Kd by more than three times ( Fig. 4B).
Lasp-1 is concentrated within focal contacts, the edges of lamellipodial extensions, the tips of microfilaments and in orthogonal filament junctions
Lasp-1 has been detected in cell membrane extensions in some breast cancer cell lines and in the cortical region in several normal epithelial cell types ( Schreiber et al., 1998; Chew et al., 1998; Chew et al., 2000). In contrast, lasp-1 does not appear to be present along stress fibers ( Chew et al., 2000; Schreiber et al., 1998) or in focal contacts in the cancer cell lines in which it has been characterized ( Schreiber et al., 1998). As shown in Fig. 5 (a-c, arrowheads), however, lasp-1 is prominently present within focal contacts in non-transformed gastric mucosal fibroblasts, where it co-localizes with F-actin. Lasp-1 is also present within the extreme tips of F-actin enriched filopodia ( Fig. 5d-f) and lamellipodial membrane extensions ( Fig. 5a-c, arrows). Finally, lasp-1 can be detected within the orthogonal actin filament junctions that abut lamellipodial extensions ( Fig. 5a'-c', arrowheads).
Elevation of cAMP leads to disruption of focal adhesions and stress fibers and translocation of lasp-1 from the cortex to the cell interior
Elevation of [cAMP]i in some cultured cells is known to cause a reduction in stress fibers and to disrupt focal adhesions ( Schoenwaelder and Burridge, 1999). To determine if such cytoskeletal alterations could (1) be induced in cultured gastric fibroblasts and (2) alter lasp-1 distribution, gastric mucosal cells were incubated with forskolin then fixed and processed for immunofluorescent localization of lasp-1 and F-actin as described in Materials and Methods. Fibroblasts were identified based on the presence of stress fibers and focal contacts which were not found to be present in differentiated mucosal cells. As shown if Fig. 6, the addition of forskolin (10 μM), which elevates [cAMP]i through activation of adenylyl cyclase, promotes this response in gastric fibroblasts whereas unstimulated cells that were cultured from the same cellular isolates retained well-developed stress fibers and focal contacts ( Fig. 6a). After a 30 minutes exposure to forskolin, stress fibers were disrupted and focal contacts had disappeared in most fibroblasts ( Fig. 6b,c). In agreement with data shown in Fig. 5, lasp-1 was clearly localized within focal contacts in controls but, after forskolin stimulation, was translocated to intracellular regions that frequently contained complexes of F-actin.
Elevation of [cAMP]i in cultured parietal cells induces the recruitment of lasp-1 to the canalicular membrane region and to the leading edge of lamellopodia
In gastric parietal cells in the same primary cultures, lasp-1 was predominately localized to the cortical membrane region along with F-actin ( Fig. 7a,b). Following forskolin stimulation, the lasp-1 signal in the expanded F-actin rich canalicular membrane region is increased ( Fig. 7c-f, arrows). A similar change in lasp-1 distribution was previously observed in freshly isolated parietal cells in gastric glands ( Chew et al., 2000). In addition to canalicular expansion, parietal cells in culture generate cell extensions following stimulation with forskolin ( Ammar et al., 2001). As shown in Fig. 7g,h, these newly formed membranes extensions contain lasp-1 and F-actin at their leading edges. Thus, elevation of [cAMP]i induces lasp-1 recruitment to F-actin rich cellular compartments not only in gastric fibroblasts but also in two different intracellular compartments in cultured gastric parietal cells.
To determine if PKA-dependent phosphorylation is necessary for lasp-1 translocation following elevation of [cAMP]i, experiments were designed to compare the distribution profile of wild-type HA-tagged lasp-1 and the HA-tagged phosphorylation site mutants following transfection into cultured parietal cells. The general outcome of these experiments supported the conclusion that cAMP-dependent phosphorylation is required for lasp-1 translocation and suggested that lasp-1 may be be involved in the initiation of the acid secretory process. Thus, in parietal cells transfected with the wild-type lasp-1 construct, the expressed HA-tagged protein was targeted mainly to the cortical cell membrane. In transfected cells that responded to forskolin stimulation (based on gross morphological changes ( Chew et al., 1989), HA-tagged lasp-1 was recruited to the F-actin rich canalicular membrane region ( Fig. 8a-d). In cells expressing the mutant protein, however, no similar redistribution of the HA-tagged RRDA146/RGFA99 protein could be detected. In addition, no gross morphological changes in cells expressing the HA-tagged RRDA146/RGFA99 protein were identified ( Fig. 8e-h).
Lasp-1 is recruited into some regions of dynamic actin assembly independent of [cAMP]i elevation
Since other signaling pathways may also be involved in regulating lasp-1 interactions with the actin cytoskeleton, we sought to determine if cAMP-independent manipulations that induce the formation of lamellipodial extensions would alter the distribution of lasp-1. A well recognized strategy for inducing the formation of cellular extensions is to expose fibroblasts deprived of growth factors to the phorbol ester, PMA. Growth factor deprivation reduces the formation of lamellipodial extensions and, under these conditions, actin incorporation occurs only at the sites of protrusive activity ( Chan et al., 1998). The addition of PMA induces lamellipodial formation and actin assembly ( Downey et al., 1992; Schliwa et al., 1984). Thus, sites of protrusive activity can be identified under growth factor-free conditions and newly formed membrane extensions can be identified following the addition of PMA ( Kellie et al., 1985). Data depicted in Fig. 9 support this prediction. Twenty-four hours after the removal of growth factors, lasp-1 was prominently localized within focal contacts and microspikes present at the tips of plasma membrane extensions in gastric fibroblasts ( Fig. 9a-c). Within two minutes after addition of PMA (100 nM), lamellipodial extensions began to form and weak lasp-1 and F-actin signals could be detected within these newly generated structures ( Fig. 9d-f). After 5 minutes, an uninterrupted band of co-localized lasp-1 and F-actin became obvious at the membrane-actin interface of growing extensions ( Fig. 9g-h). At 10 minutes there was some broadening of this band and focal contacts in most cells appeared as punctate regions in which lasp-1 and F-actin signals persisted ( Fig. 9j-l). Taken together, these results demonstrate the recruitment of lasp-1 to F-actin rich regions that are associated with active membrane protrusion in gastric fibroblasts. Since lasp-1 does not appear to be a direct target of PKC ( Chew et al., 1998), the recruitment mechanism is probably not dependent on direct phosphorylation by PKC but may be initiated by the phosphorylation of associated protein(s).
Present results are the first to identify lasp-1 as an F-actin binding protein and to confirm and extend earlier observations ( Chew et al., 2000; Schreiber et al., 1998) that the subcellular distribution of lasp-1 is at least partially correlated with that of F-actin under a range of conditions, including those which drastically modify the actin cytoskeletal dynamics and structure. The F-actin binding studies presented here suggest that this association involves a direct interaction between lasp-1 and F-actin. The cellular observations further suggest that this interaction may be regulated in vivo by both phosphorylation-dependent and -independent mechanisms. The Kd for lasp-1 binding to F-actin in vitro falls within the range previously reported for several other actin binding proteins ( Pollard, 1999) as does the saturation stoichometry ( Ishikawa et al., 1994). This low saturation stoichiometry is consistent with binding along filaments rather than a monomeric association ( Pollard, 1999).
Two major serine phosphorylation sites have been identified in lasp-1 and several lines of evidence indicate that this protein is a direct downstream substrate for PKA in vivo. Based on the parietal cell transfection studies it appears that cAMP-dependent phosphorylation of lasp-1 is required for the recruitment of this protein to the intracellular canalicular region of parietal cells. Although these findings were consistent, it should be noted that there are several potential problems in their interpretation. First, the mutated proteins displayed reduced affinity for lasp-1 in vitro. Thus, it is possible that vivo functions were similarly compromised. Also, the transfection efficiency with parietal cells is low and not all parietal cells in primary culture respond to acid secretory agonists with obvious changes in cytoskeletal morphology ( Parente et al., 1999). Clearly, new experimental approaches will be required to confirm or discount a direct role for cAMP-dependent phosphorylation in lasp-1 translocation and to define the specific cellular activities that are modulated during this translocation process.
In gastric fibroblasts, it appears that lasp-1 can also recruited to regions of dynamic actin/membrane turnover by cAMP-independent mechanisms. In addition, lasp-1 appears to retain its association with F-actin in forskolin-stimulated fibroblasts in which stress fibers are disrupted. Thus, this novel signaling protein may also be regulated either directly or indirectly by several mechanisms, with the predominant mechanism being dependent on the specific cell type. Tyrosine phosphorylation might also play a regulatory role in some cell types. Although we have not been able to detect tyrosine phosphorylation of lasp-1 in the terminally differentiated parietal cell (C.S.C., X.C. and H.-Y.Q., unpublished), there are two conserved tyrosine phosphorylation consensus sites in lasp-1 and this protein can be phosphorylated on tyrosine residues in cell lines overexpressing Src kinase ( Schreiber et al., 1998). The finding that a cytoskeletal-associated protein is differentially phosphorylated in normal epithelial cells has precedence in that cytoskeletal-membrane linking protein, ezrin, is phosphorylated on tyrosine residues in A431 cells ( Bretscher, 1989) but on serine/threonine residues in the parietal cell ( Urushidani et al., 1989). These differences in phosphorylation patterns may reflect differences in cellular protein kinase expression profiles and/or alterations in the coupling of signaling pathways to the cytoskeleton.
The observation that PKA-dependent phosphorylation reduces the affinity of lasp-1 for F-actin in vitro raises interesting questions regarding in vivo mechanisms. If PKA-dependent phosphorylation also reduces the affinity of lasp-1 for F-actin in vivo, why does lasp-1 translocate to the F-actin rich canalicular membrane region in the parietal cell? There are a several possible explanations. In the gastric parietal cell, there are at least two distinct pools of actin. At the cell cortex, γ-actin predominates whereasβ -actin predominates in the canalicular region ( Yao et al., 1995) The cortical actin pool is exquisitely sensitive to latrunculin B, which binds monomeric actin, whereas the formation of microvillar filaments appear to be highly resistant to this inhibitor as well as to cytochalasin D, which severs actin filaments ( Ammar et al., 2001). Thus, at the cell cortex, a reduced affinity for actin could assist in the rapid remodeling of the actin cytoskeleton ( Pollard, 1999). The source of lasp-1 that translocates to the canalicular region is not yet established. However, both detergent-soluble and -insoluble pools of lasp-1 have been identified ( Chew et al., 2002; Chew et al., 1998) so it is likely that lasp-1 is distributed within several subcellular compartments, possibly with stronger and weaker linkages to the actin cytoskeleton. Since lasp-1 contains both LIM and SH3 protein association domains, it is also likely that other interacting proteins play a direct and/or indirect role in regulating the affinity of lasp-1 for F-actin in vivo. The phosphorylation of lasp-1 by other protein kinase(s) may also regulate this interaction. Finally, the behavior of actin itself is extensively modified by cell membranes as well as by cytosolic extracts ( Jahraus et al., 2001). Thus, the local intracellular environment may also differentially affect the association between lasp-1 and F-actin depending on the specific pool of actin that is present.
The presence of lasp-1 in both classical and nascent focal contacts or adhesions in gastric fibroblasts is intriguing and it will be important to determine if this association is regulated by vesicular trafficking and/or phosphorylation-dependent mechanisms and to determine whether F-actin and/or other proteins within these complexes interact directly with lasp-1. Although classical focal contacts are viewed as relatively stable structures, nascent focal contacts and the cortical actin cytoskeleton appear to contain critical sites for the initiation of actin polymerization ( Beningo et al., 2001; Small et al., 1998; Wang, 1985). To date, more than 50 proteins have been detected in focal contacts and it is becoming increasing evident that there is molecular diversity as well as complex signaling activity and vesicular trafficking of proteins into and out of these regions ( Sastry and Burridge, 2000). Since no systematic or comprehensive search for focal contact proteins has been conducted, the extent of cell-type specific distribution has not yet been established ( Zamir and Geiger, 2001). However, because lasp-1 does not appear to be present in focal contacts of the human BT-474 breast cancer cell line ( Schreiber et al., 1998), the distribution of this signaling protein may indeed be cell-type specific. Defining the distribution pattern and functions of lasp-1 in a range transformed and non-transformed cell lines as well as normal epithelial cells may provide useful insights into the functions of this recently identified signaling molecule.
The specific localization of lasp-1 to proximal junctional branch points within the cortical actin filament network as well as in lamellipodial tips, where the Arp 2/3 complex proteins have also been localized ( Svitkina and Borisy, 1999), suggests the possibility that lasp-1 might interact either directly or indirectly with this important actin-nucleating complex. The actin-binding protein, cortactin ( Schuuring et al., 1993), which is also known as amplaxin and oncogene EMS-1 gene, has recently been found to bind to the Arp 2/3 complex thereby regulating actin polymerization ( Uruno et al., 2001; Weaver et al., 2001). Lasp-1 shares a relatively high degree of homology with cortactin in the SH3 domain and proline-rich regions as well as in their actin binding regions ( Sparks et al., 1996) (C.S.C., unpublished). As with lasp-1, in vitro phosphorylation of cortactin suppresses F-actin binding ( Huang et al., 1997) and cortactin undergoes phosphorylation-dependent translocation in intact cells ( Ozawa et al., 1995). Cortactin also has no direct homologue in yeast and is primarily localized within in peripheral cell regions associated with dynamic actin assembly in cultured cells ( Wu and Parsons, 1993). Most earlier studies of cortactin focused on pp60 src kinase-dependent tyrosine phosphorylation. However, epidermal growth factor induces the phosphorylation of cortactin on serine and threonine residues, and it is the phosphorylation of these residues which induces a mobility shift on western blots ( Campbell et al., 1999). In addition, it has recently been proposed that cortactin translocation is mediated by serine/threonine phosphorylation rather than by tyrosine phosphorylation ( Lopez et al., 2001). Thus, cortactin, like ezrin and possibly lasp-1 may be differentially regulated by serine/threonine kinases and tyrosine kinase dependent on the developmental stage and/or cell type involved.
Because lasp-1 lacks the conserved Arp 2/3 interaction site (the acidic `A' domain (DD/EW), which is present in cortactin as well as WASP-family members and several other Arp 2/3 binding proteins ( Olazabal and Machesky, 2001), it appears unlikely that there is a direct interaction between lasp-1 and the Arp 2/3 complex. However, lasp-1 could modulate this complex indirectly. Several SH3 domain-containing proteins have been found to bind to proline-rich sequences in WASP and N-WASP ( Higgs and Pollard, 2001). In addition, the large GTPase, dynamin, which has been implicated in endocytotic trafficking ( Kessels et al., 2000), has been shown to complex with WASP indirectly through an interaction with the WASP-binding protein, syndapin I ( Qualmann et al., 1999). Recent work has shown that lasp-1 associates with dynamin II in vitro (Okamoto, 2001). Thus, it will be of interest to determine whether or not lasp-1 has the ability to modulate the Arp 2/3 complex via direct connections with WASP family members and/or indirect associations via dynamin, for example. The answer to this question is particularly critical given the current lack of knowledge of the molecular mechanisms involved in directing membrane trafficking to sites of actin polymerization at the leading edge of migrating cells and in the formation of microvillar extensions in ion-transporting cells such as the gastric parietal cell. Another important goal for future research will be to identify additional protein binding partners for lasp-1 and to determine whether it is an indirect modulator of actin assembly and/or membrane trafficking and cell adhesion and motility.
We are grateful to Carlos Isales and Wendy Bolag for sharing equipment and expertise and to James Goldenring and Jennifer Griner for supplying MDCK cells. We also thank Carolyn Leitner in the MCG Molecular Biology Core Facility for DNA sequencing, Jon Pohl at the Emory University Microsequencing Facility for phosphopeptide sequencing, and Boris Babakov and Steve Vogel in the MCG Imaging Core for valuable assistance with sequencing and image production, respectively. This work was supported by National Institutes of Health grants R01 DK31900 and F32 DK09447.
- Accepted September 15, 2002.
- © The Company of Biologists Limited 2002