PINCH2 belongs, together with PINCH1, to a new family of focal adhesion proteins, the members of which are composed of five LIM domains. PINCH1 and PINCH2 interact, through their first LIM domain, with the integrin-linked kinase and thereby link integrins with several signal transduction pathways. Despite their high similarity, it has been shown that PINCH1 and PINCH2 could exert distinct functions during cell spreading and cell survival. To investigate the function of PINCH2 in vivo, we deleted PINCH2 in mouse using the loxP/Cre system. In contrast to the PINCH1-deficient mice, which die at the peri-implantation stage, PINCH2-null mice are viable, fertile and show no overt phenotype. Histological analysis of tissues that express high levels of PINCH2 such as bladder and kidney revealed no apparent abnormalities, but showed a significant upregulation of PINCH1, suggesting that the two PINCH proteins may have, at least in part, overlapping function in vivo. To further test this possibility, we established PINCH1-null mouse embryonic fibroblasts, which express neither PINCH1 nor PINCH2. We found that in fibroblasts with a PINCH1/2-null background, PINCH2 is able to rescue the spreading and adhesion defects of mutant fibroblasts to the same extent as PINCH1. Furthermore, we show that the LIM1 domain only of either PINCH1 or PINCH2 can prevent ILK degradation despite their failure to localize to focal adhesions. Altogether these results suggest that PINCH1 and PINCH2 share overlapping functions and operate dependently and independently of their subcellular localization.
Integrins are heterodimeric transmembrane glycoproteins that mediate cell-cell and cell-extracellular matrix adhesion (Hynes, 2002). They play pivotal roles in many biological processes including cell shape modulation, proliferation, survival and differentiation (van der Flier and Sonnenberg, 2001). Integrins execute these functions by recruiting a large number of intracellular adaptor and signaling molecules to their short cytoplasmic domain (Zamir and Geiger, 2001). They link integrins with different signaling pathways and connect integrin adhesion sites to the actin cytoskeleton. Integrin-associated molecules can also modulate integrin activation (affinity) and their clustering in focal adhesion sites (FAs; avidity). One FA protein is PINCH (also known as Lims), which is composed of five LIM domains followed by a short C-terminal tail containing putative nuclear localization/export signals. Each LIM domain is composed of two zinc-finger motifs, which are able to engage in protein interactions (Kadrmas and Beckerle, 2004). The second zinc-finger of the LIM1 domain of PINCH binds integrin-linked kinase (ILK) (Li et al., 1999). The LIM4 domain binds the adaptor protein NCK2, which interacts with several tyrosine kinase receptors thereby linking integrins and growth factor signaling (Tu et al., 1998). The LIM5 domain interacts with the Ras suppressor 1 (RSU1), which links PINCH to JNK signaling (Kadrmas et al., 2004; Dougherty et al., 2005).
The first evidence for an essential role of PINCH for integrin function in vivo came from studies on UNC-97, the Caenorhabditis elegans orthologue of PINCH (Hobert et al., 1999). Worms in which UNC-97 has been depleted by RNA interference arrest their development and show paralysis due to disrupted muscle attachment sites to the hypodermis. This phenotype resembles the defects observed in β integrin- and ILK-null worms (Gettner et al., 1995; Mackinnon et al., 2002). Similarly, Drosophila melanogaster lacking the PINCH gene suffers from muscle malfunction and paralysis caused by impaired integrin-dependent cytoskeletal attachment to the plasma membrane (Clark et al., 2003).
Vertebrates have two PINCH isoforms, PINCH1 and PINCH2 (Braun et al., 2003; Zhang et al., 2002c). Both PINCH proteins share the same modular architecture and have high sequence similarity. The most noticeable difference between the two PINCH proteins is the C-terminal tail, which is 11 amino acids longer in PINCH2 than in PINCH1. In addition, the two PINCH genes share the same structure (Braun et al., 2003) suggesting that they have probably evolved by gene duplication of a common ancestor, as have the members of many other vertebrate gene families (Gu et al., 2003). We previously compared the expression pattern of the two PINCH genes during mouse development and in adult mouse tissues (Braun et al., 2003). Interestingly, both in situ and northern blot analysis indicated that only PINCH1 is expressed during early embryonic development, whereas PINCH2 expression is first detectable in the second half of embryogenesis. Although in adult tissues the PINCH1 and PINCH2 transcripts are detected in almost all organs, the expression of the two isoforms can diverge within different cell types of the same organ. For example, in colon PINCH2 expression is restricted to the smooth muscle layer, whereas PINCH1 is also expressed in the epithelial layer (Braun et al., 2003).
In order to analyze PINCH1 function in vivo we and others recently reported the phenotype of PINCH1-deficient mice, which die shortly after implantation (Li et al., 2005; Liang et al., 2005). Although loss of β1 integrin or ILK also leads to peri-implantation lethality (Fässler and Meyer, 1995; Stephens et al., 1995; Sakai et al., 2003), PINCH1-null embryos survive significantly longer (Li et al., 2005). This prolonged development was confirmed in PINCH1-null embryoid bodies (EBs), which we generated from PINCH1-null ES cells. The EB studies revealed that lack of PINCH1 permits differentiation of primitive endoderm and epiblast but their adhesion to the basement membrane (BM), polarity, survival and, surprisingly, cell-cell adhesion are severely compromised (Li et al., 2005).
In addition to our in vivo studies, many in vitro experiments provided insight into the molecular functions of PINCH1. Overexpression of dominant negative PINCH1 protein or siRNA knock downs of PINCH1 in HeLa cells revealed an essential role in controlling cell spreading, adhesion and migration (Fukuda et al., 2003; Zhang et al., 2002a). Furthermore, these studies revealed that the interaction of PINCH1 and ILK is required for the localization of a ternary complex composed of PINCH1, ILK and α-parvin to integrin adhesion sites as well as for the stability of each component of the complex. Depletion of PINCH1 leads to downregulation of ILK protein and, vice versa, depletion of ILK leads to downregulation of PINCH1 through proteasome-mediated protein degradation (Fukuda et al., 2003). These findings are in agreement with our PINCH1-null EBs, which also showed reduced ILK protein levels (Li et al., 2005). The molecular mechanism of how these proteins can mutually prevent their degradation is still unclear.
Evidence from several sources have suggested a role for mammalian PINCH1 in cell survival. The molecular mechanisms underlying PINCH1-mediated cell survival are complex and involve several signaling pathways. Knock down of ILK or PINCH1 in HeLa cells leads to reduced phosphorylation of PKB/Akt at Ser473 (Fukuda et al., 2003), which can be phosphorylated by ILK (Delcommenne et al., 1998). In addition to the diminished Ser473 phosphorylation, PINCH1 knock down cells also displayed a decreased phosphorylation of Thr308, which is phosphorylated by 3-phosphoinositide-dependent kinase 1 (PDK1) (Alessi et al., 1997). How PINCH1 affects PDK1 activity and consequently Thr308 phosphorylation is unclear. Finally, expression of a constitutively active PKB/Akt protein did not rescue these cells from apoptosis, suggesting that PINCH1-dependent survival signaling acts either both upstream and downstream or additionally in parallel to PKB/Akt (Fukuda et al., 2003). The gene knockout studies in mice also pointed to a prominent role of PINCH1 in cell survival (Li et al., 2005; Liang et al., 2005). Interestingly, however, these studies also demonstrated that the survival function is cell specific and does not operate in trophectodermal cells, inner cell mass cells and epiblast but selectively in the primitive endoderm of the implanting embryo (Li et al., 2005).
Whereas in vivo and in vitro studies provided insight into the role of PINCH1, little is known about PINCH2. Overexpression studies in HeLa cells revealed that PINCH2 can compete with PINCH1 for ILK binding and hence alter the balance of PINCH1-ILK to PINCH2-ILK complexes (Zhang et al., 2002c). The appropriate balance seems to be essential since re-expression of PINCH1 fully restored all the defects in PINCH1 knock down cells, whereas expression of PINCH2 rescued ILK protein level, but not cell spreading and Akt phosphorylation (Fukuda et al., 2003; Xu et al., 2005). An additional functional difference between PINCH1 and PINCH2 is the inability of PINCH2 to interact with the PINCH1-binding protein RSU1 (Dougherty et al., 2005). All these findings led to the proposal that PINCH2 may act as a regulator of PINCH1 activity (Wu, 2004).
To investigate PINCH2 function in vivo, we generated PINCH2 (Lims2 – Mouse Genome Informatics) knockout mice. To our surprise, these mice are viable, fertile and age normally. Tissues that normally express high levels of PINCH2 up-regulated PINCH1 expression. Complementation experiments with mouse embryonic fibroblasts (MEFs) that express neither PINCH1 nor PINCH2 revealed that PINCH2 is able to rescue their spreading and adhesion defects to the same extent as PINCH1. Interestingly, the subcellular localization of either PINCH isoforms is essential to restore spreading and adhesion but is not essential to prevent ILK degradation.
Materials and Methods
ES cells culture and generation of PINCH2-null mice
A 500 bp fragment from a PINCH2 EST clone (GeneBank accession number AI325875) was used to screen a PAC library. Several clones were identified and used to generate a conditional targeting PINCH2 construct (PINCH2fl). Briefly, the targeting vector consists of a 4 kb left arm followed by a single loxP site, a 1.6 kb genomic fragment containing exon 3 and 4, a neo-tk cassette flanked by two loxP sites and a 5 kb right arm (for more detailed information please contact email@example.com). The construct was electroporated into R1 ES cells and clones that underwent homologous recombination were isolated, transiently transfected with a Cre expression plasmid and selected in the presence of 1-2′-deoxy-2′-fluoro-b-D-arabinofuranosyl-5-iodouracil (FIAU, Moravek Biochemicals Brea, CA). Clones that lost the neo-tk cassette but not the genomic fragment containing exon 3 and 4 (floxed allele in Fig. 1A) were identified by Southern blot analyses and injected into blastocysts to generate germline chimeras.
PINCH2 mutant mice were genotyped by Southern blot analyses (Fig. 1B) or PCR (not shown). For PCR analyses, a forward primer (5′-CACTCCCAATTCCCCTCCCTGAG-3′) was used in combination with a first reverse primer (5′-AGGGGTCTGAGGTCCTGAGAAGG-3′) to determine the wild-type and floxed allele, or with a second reverse primer (5′-GGACAGAGGGGGCAAAG ACC-3′) to detect the null allele.
Preparation of primary and immortalized MEF cells
We previously described the generation of PINCH1+/fl embryonic stem (ES) cells, in which loxP sites flank exon 4 of the PINCH1 gene (Li et al., 2005). The PINCH1+/fl ES cells were injected into blastocysts to generate germline chimeras and then the mice were mated to obtain a PINCH1fl/fl mouse strain. Mouse embryonic fibroblasts (MEFs) were isolated from PINCH1fl/fl mice at E16.5 following standard procedures (Talts et al., 1999). Part of the MEF cells were immortalized by retroviral transduction of the SV40 large T antigen. Immortalized PINCH1fl/fl MEF cells were cloned and subsequently infected with an adenovirus expressing Cre recombinase (4000 units/cell) to obtain PINCH1–/– clones. Primary MEFs were directly infected with a retrovirus expressing Cre recombinase. Deletion of the PINCH1 gene was detected by PCR using specific primers described previously (Li et al., 2005).
RNA isolation, northern blot and RT-PCR
RNA isolation from cells and mouse tissues and northern blot assays were performed as described previously (Braun et al., 2003). Poly(A)+ RNA was purified using a commercial kit (Oligotext mRNA mini kit, Qiagen GmbH, Hilden, Germany) and 1 μg mRNA was gel separated and hybridized with a PINCH1 cDNA probe (Braun et al., 2003), a PINCH2 cDNA probe generated by PCR amplification using specific primers (5′-TCTCTAGTGTCAGTCTCTAGT-3′ and 5′-CTTCCCGTCATGGTGGCCGCG-3′), and a ILK cDNA probe spanning a 340 bp SmaI-NsiI fragment excised from the ILK cDNA. For RT-PCR, 5 μg of total RNA were reverse-transcribed using an oligo(dT)20 primer and Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA), according to the manufacturer's protocol. The single strand cDNA was used as a template for a PCR reaction, using a forward primer hybridizing to exon 2 (5′-GATGCCATGTGCCAGCGCTG-3′) and a reverse primer hybridizing to exon 5 (5′-GAGCAGCTGAAGTGGTCCGG-3′) of the PINCH2 gene.
Preparation of protein lysates for western blotting and coimmunoprecipitation
For PKB/Akt phosphorylation assays, cells were grown to 60-70% confluency, serum-starved for 4 hours and stimulated with 10% FCS for the 10 minutes. Cell lysis and western blotting were performed as described previously (Sakai et al., 2003). For co-immunoprecipitation (IP) experiments, sub-confluent cells in 150-mm dishes were lysed in 1 ml IP buffer (1% Triton X-100, 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 10 mM Na4O7P2, 2 mM Na3VO4 pH 10, 100 mM NaF). For each IP, 700 μg of each protein extract were mixed with 53 μl of a mix of rabbit anti-GFP antibody and protein A-Sepharose slurry (Sigma-Aldrich, St Louis, MO; 3 μl antibody + 50 μl of protein A-Sepharose per sample) in a final volume of 1 ml. After 2 hours at 4°C, the Sepharose was pelletted, washed three times with 1 ml IP buffer and finally resuspended in 2× SDS-PAGE loading buffer for western blot analysis.
All antibodies have been described previously (Li et al., 2005; Sakai et al., 2003) except rabbit anti-GFP ab290 (Abcam Ltd, Cambridge, UK) and mouse monoclonal anti-talin (Sigma-Aldrich). Anti-tenascin-C was a kind gift from Andreas Faissner (Department of Molecular Neurobiology, Ruhr University, Bochum, Germany). Corresponding secondary antibodies were purchased from Jackson Immunoresearch Laboratories Inc. (West Grove, PA), Molecular Probes (Eugene, OR) and BioRad (Hercules, CA).
Histological analysis and Immunocytochemistry
Histological analysis of mouse tissues was performed as described previously (Sakai et al., 2003). For immunocytochemistry, cells were plated for 4 hours on glass slides coated with 10 μg/ml bovine fibronectin (Sigma-Aldrich), then fixed and stained using a standard protocol (Sakai et al., 2003). F-actin was visualized with TRITC-conjugated phalloidin (Sigma-Aldrich). Samples were analyzed using a confocal microscope (Leica, Bensheim, Germany).
Cells (2×105) were washed twice in ice-cold PBS, resuspended in 100 μl of binding buffer (10 mM Hepes pH 7.4, 140 mM NaCl, 2.5 mM CaCl2) and stained with 5 μl of Alexa Fluor 488-conjugated annexin V. After 15 minutes 200 μl of binding buffer and 1 μg/ml propidium iodide were added to each sample (Sigma-Aldrich). Positive control cells were cultured for 24 hours in suspension in serum-free medium in order to induce apoptosis.
EGFP was directionally cloned using the EcoRI and BamHI sites of the pCLMFG retroviral vector (Naviaux et al., 1996), generating a pCLMFG-EGFP vector that was used for the cloning of all further constructs. The following primers were used to amplify PINCH1 and PINCH2 portions: LIM1/PINCH1 (5′-GCGGATCCGGGGCCAATGCCTTGGCCAG-3′ and 5′-CGGGATCCTCAGCAGGGAGCAAAGAG-3′), LIM1/PINCH2 (5′-GCGGATCCGGGTCTGAGTGTTTGGCCGA-3′ and 5′-CGGGATCCTCAGCATGGAGCAAATAG-3′), ΔLIM1/PINCH1 (5′-GCGGATCCGGCCAAATGCTCTTTGCTCC-3′ and 3′-GCGGATCCTATTATTTCCTTCCTAAGGT-5′) and ΔLIM1/PINCH2 (3′-GCGGATCCGGCCAAATGCTATTTGCTCC-5′ and 3′-GCGGATCCTATTAGAGTGAGTTGACGTC-5′). Full-length PINCH1 and 2 cDNAs (Braun et al., 2003) and their deletion constructs were subcloned in the BamHI site in the pCLMFG-EGFP vector. VSV-G-pseudotyped retroviral vectors were produced by transient transfection of 293T (human embryonic kidney) cells. Viral particles were concentrated from cell culture supernatant as previously described (Pfeifer et al., 2000) and used for infection of PINCH1-deficient fibroblasts.
Cell adhesion assay
Cell adhesion assays were carried out as described previously (Sakai et al., 2003).
Cell migration and cell wounding assays
Transwell chambers (Corning Incorporated Life Sciences, Acton, MA) with an 8 μm pore diameter were coated on the lower surfaces with 10 μg/ml fibronectin (FN; Sigma-Aldrich). Cells (3×104) suspended in 0.1 ml DMEM/1% FCS were placed into the chamber and incubated for 5 hours at 37°C. Afterwards cells in the upper surface of the membrane were removed and the cells on the lower surface fixed and stained with 20% methanol/0.1% crystal violet. The cells from five randomly chosen microscopic fields were counted and cell motility was expressed as the number of the cells/mm2 of the microscopic fields±s.d.
To assay cell migration after cell wounding cells were grown to confluence in six-well plates. The monolayers were wounded by scratching with a smoothed glass micropipette. After washing, the cells were incubated in DMEM/10% FCS and observed with a Zeiss Axiovert microscope. To determine cell polarity the Golgi was localized by GM130 staining. Cells in which the Golgi pointed towards the scratch within an angle of 120° were scored positive (Etienne-Manneville and Hall, 2001). For each time point at least 300 cells were examined.
Generation of PINCH2 knockout mice
To disrupt the murine PINCH2 gene we made a conditional PINCH2 targeting vector, in which exons 3 and 4 were flanked by loxP sites and a neo-tk cassette. Upon electroporation and G418 selection 180 ES cell clones were isolated, and 13 clones revealed homologous recombination of the targeting construct within the PINCH2 genomic locus. Out of these, two clones were transiently transfected with a Cre recombinase expression vector in order to remove the neo-tk cassette (Fig. 1A). Southern blot analyses were used to identify these ES cell clones (termed PINCH2+/fl). Four PINCH2+/fl ES cell clones were injected into blastocysts to generate chimeric founder males. After germline transmission, PINCH2+/fl mice were intercrossed to obtain PINCH2fl/fl mice. These mice were normal and expressed normal levels of PINCH2 protein in all tissues analyzed. Finally, these mice were crossed with deletor-Cre mice to generate mice harboring one PINCH2 null allele (PINCH2+/–).
The deletion of exon 3 and 4 causes a reading frame shift within the first LIM domain and should therefore result, if at all, in the expression of a short, non-functional N-terminal PINCH2 polypeptide. Heterozygous PINCH2 mice (PINCH1+/fl; Cre) appeared normal and did not show any overt phenotype. To eliminate the Cre allele, these mice were crossed with C57Bl/6 mice and the progeny was genotyped for Cre. Mice carrying one PINCH2-null allele and lacking the deletor-Cre transgene (PINCH2+/–) were intercrossed to generate PINCH2-deficient (PINCH2–/–) mice (Fig. 1B). Among 43 viable, 3-week-old offspring from heterozygous intercrosses, 9 were wild type, 21 heterozygous and 13 homozygous. PINCH2-deficient mice were indistinguishable from their wild-type or heterozygous littermates, were fertile and did not develop any obvious abnormal phenotype.
Lack of PINCH2 leads to upregulation of PINCH1 protein level
Northern blot experiments and in situ hybridizations revealed PINCH2 expression in a number of different tissues (Braun et al., 2003). In order to test whether the PINCH2 gene is inactivated in our mutant mice we performed northern blots and prepared protein extracts from different organs of wild-type and PINCH2 knockout mice. For northern blot, we designed a specific probe 5′ to the site of deletion and overlapping with exon 1. The PINCH2 mRNA showed up as a band slightly above 18S rRNA in poly(A)+ RNA samples extracted from wild-type bladder (Fig. 1C). A fainter band of slightly shorter mRNA was detected in poly(A)+ RNA samples from PINCH2–/– mice (Fig. 1C). To further characterize the shorter sized band in mutant tissue we performed RT-PCR on RNA from bladder of PINCH2+/+ and PINCH2–/– mice, using a forward primer hybridizing to exon 2 and a reverse primer hybridizing to exon 5. A band consistent with the expected size of 467 bp was amplified from PINCH2+/+ tissue, whereas a band around 280 bp was amplified from PINCH2–/– tissue (Fig. 1D). Cloning and sequencing of the wild-type and mutant bands confirmed that the size difference was due to the deletion of 188 nucleotides corresponding to exons 3 and 4. In the mutant PINCH2 mRNA, exon 2 was spliced to exon 5, causing a reading frame shift and introducing a premature stop codon in exon 5 (see Fig. S1 in supplementary material). The presence of the premature stop codon probably causes degradation of the PINCH2 mutant mRNA through nonsense-mediated mRNA decay (reviewed by Wagner and Lykke-Andersen, 2002) which would account for the weak signal observed in northern blot assay (Fig. 1C).
We recently generated isoform-specific PINCH1 and PINCH2 antibodies (Li et al., 2005). Western blots, using the PINCH2 antibody, revealed a 36 kDa band from wild-type bladder, kidney, liver (Fig. 1E) and other tissue extracts (data not shown). Among all tissues tested, bladder showed the highest PINCH2 expression. Long exposure of the blot also revealed weaker signals in kidney, followed by liver, heart and diaphragm (data not shown). No signal could be detected in knockout tissues even after long exposure, confirming the inactivation of the PINCH2 gene.
Since both PINCH proteins stabilize cellular ILK protein levels (Fukuda et al., 2003) we wondered whether deletion of PINCH2 results in lower ILK levels in vivo. As shown in Fig. 1E, ILK levels were unaltered in tissue extracts from bladder, kidney and liver, as well as in other organs (data not shown). However, PINCH1 protein levels were increased in bladder and also moderately in kidney and liver from PINCH2 knockout mice (Fig. 1E). This increase was also observed in other organs expressing PINCH2 (heart and diaphragm, not shown). In order to test whether the upregulation of PINCH1 occurs at the transcriptional or post-transcriptional level, we performed northern blot analyses. As shown in Fig. 1C, PINCH1 mRNA level was unaltered in mRNA extracts prepared from PINCH2 knockout bladder, suggesting that upregulation of PINCH1 protein is the result of a post-transcriptional mechanism and not by increased mRNA transcription.
Deletion of PINCH2 does not affect the morphology of organs
Since PINCH2-deficient mice revealed no obvious phenotype, we performed histological analyses of several tissues. Internal organs from PINCH2–/– mice appeared, from gross inspection to be of normal size and morphology. Tissues that normally express PINCH2 such as bladder, kidney, liver (Fig. 2A-C) and colon (not shown), showed no histological abnormalities. In addition, staining sections for tenascin-C, a protein that is overexpressed during tissue regeneration and many pathological conditions such as inflammation and neoplasia (for a recent review see Chiquet-Ehrismann and Tucker, 2004) was normal in knockout tissues (data not shown).
Next, we tested whether cells normally expressing PINCH2 upregulate PINCH1 protein. In bladder, both PINCH1 and 2 are expressed in the smooth muscle layer. As shown in Fig. 2D PINCH1 protein expression was elevated in the smooth muscle layer of the bladder of PINCH2 knockout mice.
PINCH2 rescues spreading and adhesion defects of PINCH1-null fibroblasts
The absence of an abnormal phenotype of PINCH2-null mice and the upregulation of PINCH1 in PINCH2-deficient tissues suggest that PINCH1 may compensate for the lack of PINCH2, which would imply that the two PINCH isoforms have, at least in part, overlapping functions. To further explore such a possibility we decided to test to what extent PINCH1 and 2 can complement each other in vitro.
PINCH1 is ubiquitously expressed throughout embryogenesis. For this reason, we could not develop a cell model to study the consequences of PINCH2 deletion in the absence of PINCH1 in vitro. However, we recently described the development of ES cells carrying a conditional PINCH1-null (floxed) allele (Li et al., 2005). We used them to generate PINCH1fl/fl mice and then used the mice to derive PINCH1 fl/fl mouse embryonic fibroblast (MEF) cell lines. The PINCH1fl/fl cells were immortalized, cloned and infected with a Cre transducing adenovirus to delete PINCH1. Successful deletion was confirmed by western blot analysis and immunofluorescence staining (Fig. 3A; Fig. 4B). The PINCH1–/– cell clones and the parental PINCH1fl/fl cells also lacked PINCH2 protein (Fig. 3A), consistent with the finding that the expression of PINCH2 is limited to a few embryonic organs (Braun et al., 2003). Hence the mutant cells represent an excellent in vitro system that allows discrimination between PINCH1 and 2 functions.
Consistent with previous reports describing PINCH1-depleted HeLa cells (Fukuda et al., 2003), the PINCH1-null MEFs showed a clear spreading defect, both in the presence and absence of serum and on both plastic and fibronectin-coated glass slides (compare Fig. 3B and C). Infection of these cells with a retroviral vector expressing EGFP-tagged PINCH1 rescued the spreading defect (Fig. 3D). Similarly, expression of EGFP-PINCH2 also led to normal spreading (Fig. 3E). Mutant cells infected with a control EGFP retrovirus, however, did not rescue the reduced cell spreading (data not shown).
In addition to the spreading defect, PINCH1-null MEFs showed fewer FAs, both short and elongated as determined by paxillin staining (Fig. 4A,B). Furthermore, phalloidin staining revealed the presence of a few, thin stress fibers, which tended to accumulate cortically. To analyze cell spreading at different time points, we use time-lapse phase contrast microscopy and performed immunostaining of cells fixed at different times after seeding on fibronectin. These analyses revealed that the floxed and PINCH1-null cells had already adhered 5 minutes after plating and had begun to form focal complexes and to spread 10 minutes after plating (Fig. 5A,B). The extent of spreading, however, was severely reduced. Furthermore, the PINCH1-null cells failed to organize their FA and actin stress fibers to the same degree as the parental PINCH1fl/fl cells (Fig. 5C,D).
Next we determined whether expression of EGFP-PINCH2 rescued the actin and FA defects of PINCH1-null MEFs to the same extent as re-expression of EGFP-PINCH1. We found that expression of either EGFP-PINCH1 or EGFP-PINCH2 resulted in the formation of abundant stress fibers and well-developed FAs, which were indistinguishable from those present in the parental PINCH1fl/fl cells (Fig. 4C,D).
Cell adhesion assays revealed that control cells adhered strongly to FN and vitronectin (VN) and weakly to laminin (LN) but PINCH1-deficient cells showed a clear reduction in adhesion to all ECM substrates analyzed (Fig. 6). Like the abnormal spreading and cell morphology, the adhesion defects could also be rescued with either EGFP-PINCH1 or EGFP-PINCH2 (Fig. 6).
Finally, cell migration was tested in a scratch assay. However, PINCH1fl/fl and PINCH1-null cells migrated with the same speed and polarized normally (Fig. S2A-C in supplementary material). In addition, we performed cell motility assay using Transwell chambers in which the undersurfaces of the membranes were coated with FN. Also here no significant difference between PINCH1-null and PINCH1fl/fl cells could be detected (Fig. S2D in supplementary material).
Deletion of PINCH1 in primary MEFs neither impairs survival nor phosphorylation of PKB/Akt
It has been reported that the two PINCH isoforms differently affect cell survival and PKB/Akt phosphorylation of HeLa cells (Fukuda et al., 2003). To test these parameters we prepared primary embryonic fibroblasts from 14.5-day-old PINCH1fl/fl embryos and transduced them with a retrovirus expressing Cre recombinase. Five days after Cre transduction most cells began to retract and 9 days after Cre transduction all of them were smaller, whereas non-infected MEFs were flat and well spread (Fig. 7A,B). PCR genotyping of Cre-transduced cells revealed the loss of the floxed PINCH1 gene (Fig. 7C) and western blot analysis and immunofluorescence staining showed the absence of the PINCH1 protein (Fig. 7D-F). Consistent with previously published data (Fukuda et al., 2003; Li et al., 2005), the level of ILK decreased in the PINCH1-null primary MEFs (Fig. 7D), whereas the level of vinculin was unchanged. Staining of focal adhesions and F-actin revealed thin focal adhesions and actin stress fibers accumulating at the cell cortex of PINCH1–/– primary MEFs (Fig. 7E,F).
Apoptosis was examined by staining cells with Alexa Fluor 488-conjugated annexin V and the non-vital dye propidium iodide (PI) and determining the extent of apoptotic and necrotic cells by FACS (Vermes et al., 1995). As a positive control for apoptosis primary PINCH1fl/fl MEFs were cultured in suspension and in serum-free medium. Cells undergoing apoptosis are positive for annexin V but negative for PI (Fig. 7G). Both adherent PINCH1fl/fl primary MEFs and PINCH1–/– primary MEFs contained around 8% apoptotic cells, whereas the number of apoptotic MEFs grown in suspension rose to 19.05%. The number of dead cells that were positive for both annexin V and PI was similar in the PINCH1fl/fl and PINCH1–/– primary MEFs (7.43 and 6.49%, respectively), whereas in the MEFs grown in suspension the number increased to 61.08%. This indicates that loss of PINCH1 in primary fibroblasts does not alter their survival.
Next we analyzed the phosphorylation level of PKB/Akt in the PINCH1 primary MEFs. Western blot cell lysates derived from PINCH1fl/fl and PINCH1–/– cells growing in serum-supplemented medium revealed similar levels of phosphorylated Ser473 (Fig. 7D). We therefore serum induced PINCH1fl/fl and PINCH1–/– primary MEFs after starvation and tested PKB/Akt phosphorylation. As shown in Fig. 7H, phosphorylation of Ser473 of PKB/Akt was not significantly changed in PINCH1-null cells, whereas phosphorylation of Thr308 was sometimes normal and sometimes slightly reduced.
ILK is stabilized by PINCH1 and PINCH2 independent of their subcellular localization
A recent report showed that PINCH1 knockdown cells have lower ILK protein levels because of a proteasome-mediated degradation of the ILK protein (Fukuda et al., 2003). The degradation of ILK was prevented by the expression of PINCH1 or PINCH2 in PINCH1-depleted cells (Fukuda et al., 2003). Furthermore, it has been shown that PINCH1 is also required for localizing ILK to FAs (Zhang et al., 2002a). It is unclear, however, whether ILK degradation is a result of its intracellular mislocalization or if sole binding of PINCH1 or 2 to ILK is already sufficient for stabilization. Furthermore, it is also unclear whether the two PINCH isoforms prevent ILK degradation through the same mechanism. To answer these questions, we built deletion mutants of both PINCH proteins. Since PINCH1 and 2 bind ILK through their first LIM domains (Li et al., 1999; Zhang et al., 2002c), we fused EGFP to LIM1 derived from PINCH1 and PINCH2 (LIM1/P1, LIM1/P2). Furthermore, we created two other EGFP-tagged mutants, in which the LIM1 domain was removed from PINCH1 and PINCH2 (ΔLIM1/P1, ΔLIM1/P2).
The constructs were expressed in PINCH1-null MEFs by retroviral transduction. In contrast to the full-length PINCH1 and 2 proteins (see Fig. 3D,E), none of the PINCH1 and 2 mutants rescued the cell spreading defect caused by PINCH1 deletion. Cell lysates from PINCH1–/– cells expressing the full-length EGFP-tagged PINCH1 and PINCH2 or EGFP-tagged LIM1/P1, LIM1/P2, ΔLIM1/P1 and ΔLIM1/P2 were analyzed by western blot assay. Probing with an anti-GFP antibody showed robust expression of proteins of the expected size corresponding to each construct (Fig. 8A). Similar to the PINCH1–/– primary MEFs and in agreement with previously published data (Fukuda et al., 2003; Li et al., 2005) the level of ILK protein was decreased in PINCH1-null cells (Fig. 8A,B) but not the level of ILK mRNA. The protein levels of other FA components including talin, vinculin and paxillin were unchanged (Fig. 8A). These data confirm a previous report showing that downregulation of ILK depends on a post-transcriptional mechanism that affects ILK but not other FA proteins (Fukuda et al., 2003).
The expression of EGFP-PINCH1 or EGFP-PINCH2 in PINCH1-null cells restored normal ILK levels, whereas expression of EGFP alone had no effect (Fig. 8A). In addition, ILK levels were also restored by expressing either LIM1/P1 or LIM1/P2, but not by expressing ΔLIM1/P1 or ΔLIM1/P2. In order to verify the binding of LIM1/P1 as well as LIM1/P2 with ILK, we performed immunoprecipitation of the EGFP-tagged constructs and subsequently detected ILK by western blotting. As shown in Fig. 8C, both LIM1/P1 and LIM1/P2 co-precipitated ILK. The amount of ILK co-precipitated by LIM1/P1 and LIM1/P2 was similar to the amount of ILK co-precipitated by the full-length EGFP-PINCH1 and EGFP-PINCH 2 proteins. No ILK was co-immunoprecipitated with the ΔLIM1/P1 and ΔLIM1/P2 constructs or with EGFP alone.
Immunofluorescent staining revealed that none of the constructs localized efficiently to FAs (compare Fig. 8D-G with Fig. 4C,D). Only very faint signals were observed in a few ECM adhesion sites, especially with the ΔLIM1/P1 and ΔLIM1/P2 fusion proteins (arrowheads in Fig. 8). In addition, all cells expressing ΔLIM1/P1 or ΔLIM1/P2 had a strong nuclear EGFP signal, which was rarely observed with the LIM1/P1 and LIM1/P2 constructs and never with the full-length EGFP-PINCH1 or EGFP-PINCH2. Consistent with the ability of LIM1/P1 and LIM1/P2 to stabilize ILK, immunofluorescence analysis showed also that PINCH1-null cells transfected with these constructs had increased ILK staining in the cytoplasm. This was not the case in PINCH1-null cells expressing EGFP only or the ΔLIM1/P1 and ΔLIM1/P2 constructs (data not shown).
Here we describe the consequences of PINCH2 deletion in vivo. PINCH2-null mice are viable, fertile and show no overt abnormal phenotype. Histological analyses of several organs did not reveal any morphological abnormality or evidence of impaired functionality. Although it is possible that subtle alterations escaped our detection, these findings suggest that if PINCH2 acts as a regulator of PINCH1 activity, the regulation seems not to be essential for viability. It is, however, still possible, though not mandatory, that PINCH2 function could be important under stress situation such as wound healing, inflammation, etc. However, we show here that the lack of PINCH2 leads to a significant increase of PINCH1 protein level, resulting from a post-transcriptional mechanism and not from changes in gene transcription. It is noteworthy that upregulation of PINCH1 occurs in the same cell types that would normally express high levels of PINCH2. These observations, together with the high protein sequence similarity suggest that PINCH1 may replace PINCH2 function in vivo.
It is unclear whether PINCH2 can substitute for PINCH1 in vivo. PINCH2 is not expressed during early embryonic development (Braun et al., 2003) and hence absent at the peri-implantation period when the PINCH1-null mice die (Liang et al., 2005; Li et al., 2005). However, a tissue-specific deletion of the PINCH1 gene in cardiac muscle cells does not lead to an abnormal phenotype (Liang et al., 2005) suggesting that the remaining PINCH2 expressed in cardiac muscle cells substitutes for PINCH1. Since in vitro evidence indicates functional differences between the two PINCH isoforms (Fukuda et al., 2003) it will be important to test whether this compensatory mechanism holds true for other cells and tissues.
Owing to the current lack of an in vivo model that allowed us to determine the identical and the different functional properties of the two PINCH isoforms we developed an in vitro cell system. Our western blot and immunohistochemistry analysis of mouse tissues revealed no organ, tissue or cell subpopulation in which PINCH2 is the solely expressed PINCH isoform. Even in organs showing highest PINCH2 expression (e.g. bladder), PINCH1 is present and even upregulated after PINCH2 ablation. For these reasons, we could not use PINCH2-deficient cells to develop an in vitro system that allowed investigation of PINCH2 deletion or the potential complementation of PINCH2 by PINCH1. We were able, however, to establish several PINCH1-null MEF cell lines that because of their embryonic origin do not express PINCH2. We could therefore use these cell lines as a suitable in vitro system to compare the roles of the two PINCH isoforms.
Deletion of PINCH1 in MEF cells led to impaired spreading and adhesion on several substrates, with dramatic changes in the morphology of FAs and the actin cytoskeleton. Nevertheless, despite their abnormal F-actin distribution PINCH1-null MEFs established a normal polarity and showed normal cell wound closure or migration through Transwell chambers. The normal migration of PINCH1-null fibroblasts is somehow surprising, since cells defective in organizing their actin cytoskeleton and in adhesion to ECM are expected to show altered motility. In this respect it is worth noting that loss of Rac1 expression in macrophages also results in defective spreading and reduced adhesion to FN but still normal migration, indicating that an abnormal actin cytoskeleton is not always associated with impaired migration (Wells et al., 2004).
The defective spreading in PINCH1-null MEFs is in line with previous studies with HeLa cells (Fukuda et al., 2003). However, unlike the HeLa cell studies we could fully rescue the spreading defect in PINCH1-null MEFs by expressing PINCH2. Furthermore, complementation of PINCH1-null MEFs with PINCH2 restored normal organization of the actin cytoskeleton and FAs and rescued the adhesion defect. These data suggest that the two PINCH isoforms can play similar functions both in cell spreading and adhesion in our fibroblast cell lines. A possible explanation for the discrepancy between our data and the previously published findings could be that the effects of PINCH2 on spreading may vary among different cell types.
Loss of PINCH1 in primary MEFs grown in serum or in serum-free medium did not affect cell survival and PKB/Akt phosphorylation. This result also differs from PINCH1 depleted HeLa cells (Fukuda et al., 2003) suggesting that the PINCH/ILK complex also regulates cell survival and PKB/Akt signaling in a cell type-specific manner. Our previous observation that apoptosis is detectable only in the endodermal layer of PINCH1–/– EBs (Li et al., 2005) supports the idea of a cell type-specific role of PINCH1 for sustaining cell survival. It has been shown that expression of PINCH2 in the PINCH1 knockdown HeLa cells failed to restore normal PKB/Akt phosphorylation and to prevent apoptosis, which represents a potentially important functional difference between the two PINCH isoforms (Fukuda et al., 2003; Xu et al., 2005). Since the PINCH1-null MEFs showed normal cell survival and PKB/Akt phosphorylation we could not use them to test whether a similar functional divergence also operates in fibroblasts.
Our in vitro experiments with PINCH1-null MEFs also provided some insight into the mechanism of how these proteins prevent ILK degradation. As reported for PINCH1-depleted HeLa cells (Fukuda et al., 2003), loss of PINCH1 in MEFs led to a post-transcriptional downregulation of the ILK protein, which was fully recovered by the expression of PINCH1 or PINCH2 (Fig. 7). However, we found that expression of the LIM1 domain of PINCH1 or 2 was equally sufficient to rescue ILK levels, despite their inability to localize to FAs. This demonstrated that the stabilization of at least part of the ILK cellular pool depends on the mere binding to the LIM1 of PINCH1 or PINCH2. Findings described in a recent report by Xu et al. (Xu et al., 2005) provide further support to this notion, and our preliminary experiments in ILK-null fibroblasts indicate that ILK can prevent PINCH1 degradation in a similar manner. In fact, downregulation of PINCH1 in ILK–/– MEFs can be blocked by expressing ILK mutants that are unable to localize in FA (our unpublished observations). Altogether these observations are consistent with the finding that the interaction of PINCH, ILK and α-parvin occurs before their localization into FAs (Zhang et al., 2002b). Furthermore, they suggest that this complex system of post translation regulation may account not only for the downregulation of ILK and PINCH1 observed in PINCH1 and ILK mutant cells, respectively (Fukuda et al., 2003; Sakai et al., 2003; Li et al., 2005), but also for the upregulation of PINCH1 in PINCH2 mice. It is very possible that the deletion of PINCH2 leaves a number of ILK molecules unbound, which then bind to and stabilize more PINCH1.
In summary, our study shows that mouse development and postnatal aging can proceed without overt abnormalities when the PINCH2 gene is disrupted. Furthermore, PINCH1 and PINCH2 may have more overlapping functions than previous studies indicated (Fukuda et al., 2003) and hence may compensate each other under certain circumstances. However, whereas the PINCH2 function seems to be dispensable in vivo, we cannot rule out that the PINCH2/ILK/parvin complex still serves specific tasks under stress situations that are unique to PINCH2 and cannot be rescued by the PINCH1/ILK/parvin complex. Likewise, PINCH1 may have certain functions (possibly for cell survival) that are not compensated by PINCH2. It is reasonable to assume that a member of a structurally related protein family such as PINCH2 that is not expressed in the first half of embryogenesis lost functional properties that are executed by PINCH1 during this important developmental period. This and many other assumptions can be now tested by deleting either PINCH1 or PINCH2 or both in a spatially or temporally restricted manner in vivo.
We thank Heidi Sebald and Michal Grejszczyk for excellent technical assistance, Claudia Nicolae, Tatjana Dorn, Ling-Wei Chang and Hao-Ven Wang for helpful suggestions, Andreas Faissner for the tenascin-C antibody, Werner Müller for the Cre-expression plasmid and Ernst Pöschl for the Alexa Fluor 488-conjugated annexin V.R.B.is supported by a PhD fellowship from the University of Copenhagen and O.K. is supported by an Erwin Schrödinger fellowship from the Austrian Science Foundation (FWF). The work was supported by the DFG (SFB413), Fonds der Chemischen Industrie and the Max Planck Society.