Distinct Functions of Cdk5(Y15) Phosphorylation and Cdk5 Activity in Stress Fiber Formation and Organization

Abstract

Previous studies have shown that Cdk5 promotes lens epithelial cell adhesion. Here we use a cell spreading assay to investigate the mechanism of this effect. As cells spread, forming matrix adhesions and stress fibers, Cdk5(Y15) phosphorylation and Cdk5 kinase activity increased. Cdk5(Y15) phosphorylation was inhibited by PP1, a Src family kinase inhibitor. To identify the PP1-sensitive kinase, we transfected cells with siRNA oligonucleotides for cSrc and related kinases. Only cSrc siRNA oligonucleotides inhibited Cdk5(Y15) phosphorylation. Cdk5(pY15) and its activator, p35, colocalized with actin in stress fibers. To examine Cdk5 function, we inhibited Cdk5 activity under conditions that also prevent phosphorylation at Y15: expression of kinase inactive mutations Cdk5(Y15F) and Cdk5(K33T), and siRNA suppression of Cdk5. Stress fiber formation was severely inhibited. To distinguish between a requirement for Cdk5 kinase activity and a possible adaptor role for Cdk5(pY15), we used two methods that inhibit kinase activity without inhibiting phosphorylation at Y15: pharmacological inhibition with olomoucine and expression of the kinase inactive mutation, Cdk5(D144N). Stress fiber organization was altered, but stress fiber formation was not blocked. These findings indicate that Cdk5(Y15) phosphorylation and Cdk5 activity have distinct functions required for stress fiber formation and organization, respectively.

INTRODUCTION

Cdk5 is an atypical member of the cyclin dependent kinase family, which requires a non-cyclin protein partner (either p35 or p39) for enzymatic activity [1,2]. Cdk5 has been shown to regulate neuronal cytoskeletal dynamics, adhesion, and migration, but has no known role in cell cycle regulation [3]. Although Cdk5 and its activators were originally considered to be neuron-specific, more recent work has shown that they are also expressed in many non-neuronal tissues, where they regulate a variety of processes, including adhesion, migration, differentiation, and senescence [4]. In vivo, Cdk5 is phosphorylated at Y15, significantly increasing its kinase activity [5]. Two tyrosine kinases have been reported to phosphorylate at this site: the Src family kinase, Fyn [6] and the non-receptor protein kinase cAbl [5]. Phosphorylation at Y15 has been shown to be important for semaphorin dependent axon guidance in neurons, thus confirming its physiological relevance [6,7].

Previous findings from this and other laboratories have demonstrated that Cdk5/p35 regulates adhesion and migration in epithelial cells [8–11]. In particular, we have shown that overexpression of Cdk5 increases cell-to-matrix adhesion in lens epithelial cells [10]. The effect of Cdk5 on cell adhesion is most apparent as cells are spreading and forming attachments to the extracellular matrix [10]. Since the Src family kinases and the Rac-dependent kinase, cAbl, are also involved in regulating cell spreading and attachment [12,13], it seemed likely that Y15 phosphorylation of Cdk5 by one or more of these kinases might play a role in the observed effects of Cdk5. Moreover, since many kinases have adaptor functions in addition to their enzymatic activity, we considered the possibility that Y15 phosphorylation of Cdk5 might have a distinct function, apart from its contribution to Cdk5 activity. Thus, the present study was undertaken to examine the relationship between Cdk5(Y15) phosphorylation, Cdk5 kinase activity, and cytoskeletal remodeling during spreading and attachment of lens epithelial cells.

MATERIALS AND METHODS

Cell Culture and Transfection--

Monkey kidney epithelial cells (Cos7), rabbit lens epithelial cells (N/N1003), or human lens epithelial cells (FHL124) were cultured at 37 °C in a humidified atmosphere of 95% air and 5% CO₂. Medium for Cos7 cells was Dulbecco’s minimum essential medium (DMEM) (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum, 100units/ml penicillin, 100μg/ml streptomycin; medium for N/N1003 cells was DMEM containing 10% rabbit serum, 100units/ml penicillin, 100μg/ml streptomycin; medium for FHL 124 cells was 1 part KGM (Lonza Biologics Inc, Portsmouth, NH), 4 parts M199 (Invitrogen-GIBCO) 50μg/ml gentamycin, and 10% fetal bovine serum. Cells were transiently transfected according to the manufacturer’s instructions using FuGENE 6 (Roche Diagnostics, Indianapolis, IN) for Cos7 cells or Lipofectamine (Invitrogen) for N/N1003 cells and FHL124 cells, and incubated 24 to 48hrs before use. For suppression of cSrc, Fyn, Yes, cAbl, and Arg expression, FHL124 cells were transfected with 160nM of the appropriate antisense oligonucleotides (cSrc, (sc-29228, Santa Cruz Biotechnology, Santa Cruz, CA); Yes (M-003184, Upstate/Millipore, Billerica, MA,); Fyn (sc-29321, Santa Cruz); cAbl (sc29843, Santa Cruz,); Arg (sc-38945, Santa Cruz); or with a scrambled control, (sc-37007, Santa Cruz)), and were harvested after 48 or 72 hr, as indicated.

Pharmacological Inhibition--

Cells were preincubated overnight in the presence and absence of 10μM PP1 (Biomol International, Plymouth Meeting, PA) to inhibit Src family kinases, or 15μM olomoucine (Calbiochem, San Diego, CA) to inhibit Cdk5.

Cell Spreading Assay--

Cells were detached with trypsin, collected by centrifugation, replated on tissue culture dishes or slide chambers coated with fibronectin (Invitrogen) (10μg/ml), and allowed to spread for the indicated times. If cells were preincubated with a pharmacological inhibitor, the inhibitor was also present during spreading. For immunostaining, cells were allowed to spread on slide chambers for the indicated times, washed in PBS, fixed with 4% paraformaldehyde for 10 min, and rinsed with PBS. Following permeabilization with 0.25% Triton X-100 in PBS, the cells were blocked in 1% BSA for 1 hr, and stained with the indicated antibodies and/or fluorescent probes.

For biochemical analysis, cells plated on culture dishes were collected by scraping, then pelleted and lysed with 50 mM Tris (pH 7.5), 150 mM NaCl, containing 1% Triton X-100, CompleteMini^™ protease inhibitor (Roche Diagnostics), with protein tyrosine and serine/threonine phosphatase inhibitors (Upstate). Cell lysates were used for immunoprecipitation and immunoblotting as previously described [14].

Antibodies and fluorescent probes—

The following antibodies were obtained from Santa Cruz: Cdk5 (C-8); Cdk5 monoclonal (J-3); phosphospecific Cdk5(pY15) (sc-12918-R); c-Src (N-16) (sc-19); Fyn (FYN3-G) sc-16-G; c-Abl monoclonal (sc-23); and Arg (C-20) (sc-6356).Yes antibody (cat. 610376) was from BD Biosciences, San Jose, CA. All monoclonal antibodies were from mouse; all others were rabbit polyclonal. Anti-rabbit and anti-mouse IgG horseradish peroxidase linked secondary antibodies were from Amersham Bioscience (Piscataway, NJ). Alexa568-goat antirabbit IgG, Alexa488-donkey antimouse IgG, and rhodamine-phalloidin were from Molecular Probes-Invitrogen.

Fluorescence microscopy—

Fluorescence-labled cells were viewed using a Zeiss Axioplan II microscope with excitation 488nm to detect transfected GFP-fusion proteins and Alexa488-labeled antibodies, or 568nm to detect rhodamine-phalloidin and Alexa568-labeled antibodies. Confocal fluorescence microscopy and analysis of co-localization were carried out using a Leica laser scanning microscope and its associated software (Leica TCS SP2, Leica Microsystem, Germany).

Cdk5 Kinase Assay--

Cdk5 was immunoprecipitated with anti-Cdk5 antibody. The immunoprecipitate was washed twice with the lysis buffer and once with buffer containing 50 mM Tris pH 7.5, 10 mM MgCl₂, and 2 mM DTT. In vitro kinase activity was measured in a total volume of 30 μl of kinase assay buffer (50mM Tris pH 7.5, 10mM MgCl₂, 2mM DTT, 1mM EGTA, 40mM β-glycerophosphate, 20mM p-nitrophenylphosphate, 0.1mM sodium vanadate, 0.01% Brij35) (Promega, Madison, WI) with 10μCi of [γ-³²P]ATP (MP Biochemicals-ICN), 50μM ATP and 10μg histone H1 (Calbiochem) at 30°C for 30 min. The reaction was stopped by adding 10μl of 4X SDS sample buffer. Proteins were separated by gel electrophoresis, transferred to nitrocellulose membrane, and autoradiographed to detect ³²P incorporation.

Site-Directed Mutagenesis of Cdk5--

Site-directed mutation of GFP-Cdk5 [15] was performed using the QuickChange^™ method (Stratagene, La Jolla, CA) to generate GFP-Cdk5(D144N), and GFP-Cdk5(Y15F). All mutations were verified by DNA sequencing to ensure their accuracy. Construction of GFP-Cdk5(K33T) was described previously [15].

RESULTS

Cdk5 Activity and Y15 phosphorylation accompany cytoskeletal rearrangement in spreading cells--

Our previous results indicated that Cdk5 overexpression strengthens adhesion of lens epithelial cells and demonstrated that this effect is most apparent 1-2hr after plating [10]. To determine whether changes in endogenous Cdk5 activity occurred during this period, Cdk5 was immunoprecipitated and kinase activity was assayed in vitro using histone H1 as a substrate. Cdk5 activity increased steadily as cells spread and was significantly increased by 120min (Fig 1A,B). Since phosphorylation of Cdk5 on Y15 has been linked to increased Cdk5 activity [5], we also examined the tyrosine phosphorylation status of Cdk5 during this period in cells transiently transfected with GFP-Cdk5. The Y15 phosphorylation of GFP-Cdk5 closely paralleled the increase in endogenous Cdk5 activity during this period (Fig. 1C,D).

FIG. 1. Cdk5 kinase activity and Cdk5(Y15) phosphorylation increase during cell spreading.

A. Endogenous Cdk5 was immunoprecipitated from N/N1003 cell lysates 0, 15, 30, 60 and 120min after plating and assayed for kinase assay in vitro using histone H1 as the substrate. Cdk5 kinase activity was determined by autoradiography of ³²P incorporation into histone H1 (upper panel); total histone H1 was assessed by Ponceau S staining of the nitrocellulose membrane (lower panel). B. Results of three experiments of the type shown above were quantified by densitometry and normalized to the initial value (t=0) to determine the relative levels of Cdk5 kinase activity at each time. Statistical significance as compared to t=0 (*) was determined by one-way ANOVA (p<0.05). C. N/N1003 cells were transiently transfected with GFP-Cdk5. At 0, 15, 30, 60 and 120min after plating, GFP-Cdk5(Y15) phosphorylation and GFP-Cdk5 expression were determined by immunoblotting with antibodies against Cdk5(pY15) and Cdk5 respectively. D. Results of three experiments of the type shown above were quantified by densitometry to determine the ratio of Cdk5(pY15)/Cdk5 as a function of time (normalized to initial (t=0) values). Statistical significance (*) was determined as in B.

To determine whether these changes in Cdk5 activity and phosphorylation were temporally correlated with changes in cytoskeletal architecture, we examined the actin cytoskeletal structure during this period (Fig. 2). Fluorescence microscopy of cells stained with rhodamine-phalloidin showed a time-dependent reorganization of the actin cytoskeleton, as seen in other cell types [16,17]. Within 15 min after plating on fibronectin, the cells attached and began to spread (Fig. 2A). A uniform band of polymerized actin could be seen around the cell periphery. By 30min, numerous filopodia had formed around the periphery and a few cells had begun to form peripheral stress fibers (Fig. 2B). Stress fiber formation continued over the next 30min, so that by 60min more than 95% of cells contained well-formed stress fibers (Fig. 2C). By 120 min. cells contained prominent stress fibers and distinctly concave cell boundaries consistent with myosin-dependent contraction (Fig. 2D). Human lens epithelial cells (FHL124) underwent the same cytoskeletal changes with a similar time course (not shown).

FIG. 2. The actin cytoskeleton undergoes time-dependent rearrangement during spreading.

A. N/N1003a rabbit lens epithelial cells attached and spread within 15 min after plating on fibronectin. A prominent band of polymerized actin formed around the cell perimeter. B. 30min after plating, the peripheral band of actin began to reorganize into filopodia attached to short radial stress fibers (long arrow) and transverse stress fibers (short arrow), which crossed the radial fibers at right angles. C. By 60min after plating more than 90% of cells contained numerous, well-formed stress fibers. D. After 120min, cell borders became sharply concave as cells appeared to contract. More than 90% of cells retained stress fibers, which often spanned the entire cell. Scale bar = 20μ.

Cdk5(pY15), and p35 co-localize with stress fibers--

We next examined the subcellular localization of endogenous Cdk5(pY15) and p35 120min after plating cells, when Cdk5 kinase activity is greatest (Fig. 3). Cdk5(pY15) immunofluorescence showed a close correspondence with actin along stress fibers and at the cell periphery (Fig. 3A–C). Less intense, diffuse staining was also present throughout the cell. The Cdk5 activator, p35, showed a similar distribution on stress fibers and at the cell periphery (Fig. 3D–F), suggesting that the Cdk5(pY15) associated with stress fibers is enzymatically active. An IgG control showed no signal in the red channel (Fig. 3G–I).

FIG. 3. Cdk5(pY15) and p35 co-localize with stress fibers in spreading cells.

A. Alexa588-phalloidin staining of FHL124 lens epithelial cells co-stained with antibody to Cdk5(pY15) 120min after plating on fibronectin. Actin localizes in stress fibers (arrows) and at the cell periphery. B. Immunostaining of the same cells for Cdk5(pY15) shows similar localization along stress fibers (arrows) and at cell periphery. C. Merge of images in A and B showing close correspondence of Cdk5(pY15) and actin staining in stress fibers. White line indicates path scanned for colocalization analysis shown in Fig. 4. D. Alexa588-phalloidin staining of FHL124 lens epithelial cells co-stained with antibody to p35 120min after plating on fibronectin. Actin localizes in stress fibers (arrows) and at the cell periphery. E. Immunostaining of the same cells for p35 shows similar localization in stress fibers (arrows) and in peripheral structures. F. Merge of images in D and E, demonstrating co-localization of p35 and actin on stress fibers (arrows). White line again indicates path scanned for colocalization analysis shown in Fig. 4. G. Alexa588-phalloidin staining of control FHL124 cells mock-immunostained with non-immune rabbit IgG 120min after plating on fibronectin, showing a typical localization of actin in stress fibers (arrows) and at the cell periphery. H. Red channel image of control cells shows no signal above background. I. Merge of images in G and H, showing one of several paths scanned for quantitative assessment of red and green channel signals (white line). Scale bar = 20μm.

For a more exact determination of the correspondence between actin staining on the one hand and Cdk5(pY15) or p35 on the other, we scanned across multiple phalloidin stained stress fibers as illustrated in Figs. 3C,F and compared the fluorescence intensities of the green and red channels along the scanned lines as shown in Fig. 4. In samples that were doubly stained for actin and Cdk5(pY15), 86% of stress fibers analyzed (37/43) showed a corresponding Cdk5(pY15) peak within 200nm, with an average separation between actin and Cdk5(pY15) peaks of 93 +/− 13nm. Similarly, in samples stained for actin and p35, 80% (44/55) of stress fibers had a corresponding peak of p35 immunofluorescence within 200nm, with an average separation of 55 +/− 7 nm. Similar scans of the IgG control samples showed no signal above background in the red channel (not shown). Thus, the subcellular localization of Cdk5(pY15) and p35 is consistent with a functional role for Cdk5 in cytoskeletal reorganization in spreading cells.

FIG. 4. Co-localization analysis of Cdk5(pY15) and p35 shows close association with actin.

A. Green channel (phalloidin) intensity along the 4.0μ path indicated by the white line in Fig. 3C, shows distinct peaks corresponding to two stress fibers (arrows). B. Red channel (Cdk5(pY15)) intensity along the same 4μ path as in A. shows distinct peaks of Cdk5(pY15) immunofluorescence (arrows) that closely correspond to actin staining. C.Green channel (phalloidin) intensity along the 5.2μ path indicated by the white line in Fig. 3F, shows distinct peaks corresponding to several stress fibers, three of which are indicated by arrows. D. Red channel (p35) intensity along the same 5.2μ path shown in C. shows three peaks of p35 immunofluorescence (arrows) that closely correspond to actin staining. (Other corresponding peaks present in this scan have not been marked).

Identification of the kinase responsible for Cdk5(Y15) phosphorylation-

Both Src family kinases [6] and the related, Rac-dependent kinase, cAbl, [5] have been reported to phosphorylate Cdk5(Y15) in neurons. To determine whether these kinases play a role in the phosphorylation of Cdk5(Y15) in lens epithelial cells, we incubated cells overnight in the presence of PP1, a Src family kinase inhibitor. Cells were then lifted with trypsin and replated as before. PP1-treated cells attached, but did not spread appreciably (not shown). Quantitative densitometry of immunoblots of whole cell lysates showed that PP1 inhibited Cdk5(Y15) phosphorylation by 88+/−7 % (Figure 5A). The residual phosphorylation of Cdk5(Y15) may represent incomplete inhibition of Src-related kinases by PP1 or involvement of an additional kinase that is not inhibited by PP1.

FIG. 5. Cdk5(Y15) phosphorylation is Src-dependent.

A. N/N1003 cells were transfected with GFP-Cdk5 and incubated overnight to allow expression of the transgene. The following day 10μM PP1 was added to experimental samples (+), cells were again incubated overnight, replated in the presence (+) or absence (−) of 10μM PP1, and harvested after 90min. Samples were immunoblotted with antibodies for phospho-Cdk5(Y15) (arrow) and Cdk5. B. Human FHL124 lens epithelial cells were transiently transfected with GFP-Cdk5 cDNA and siRNA oligonucleotides against cSrc. After 72hrs, cells were harvested and replated on tissue culture flasks for 120 min. Levels of cSrc, GFP-Cdk5(pY15) (arrow), GFP-Cdk5, and tubulin were determined by immunoblotting with the corresponding antibodies. C. Experimental conditions as in B., using siRNA oligonucleotides against Fyn. D. Experimental conditions as in B., using siRNA oligonucleotides against Yes. E. Experimental conditions as in B., using siRNA oligonucleotides against Abl. F. Experimental conditions as in B., using siRNA oligonucleotides against Arg.

To identify which PP1-sensitive kinase has the greatest effect on Cdk5(Y15) phosphorylation in spreading cells, human lens epithelial cells (FHL124) were transfected with specific siRNA oligonucleotides to suppress endogenous expression of the individual Src family kinases expressed in FHL124 cells. Since PP1 has been reported to inhibit cAbl and Arg, in addition to the Src family kinases [18], we also tested the effect of siRNA suppression of these kinases. Transfection with cSrc siRNA oligonucleotides suppressed cSrc expression by 50.1+/−0.5% and reduced Cdk5(Y15) phosphorylation to 73 +/− 11 % of control values (untransfected cells or cells transfected with a scrambled, control oligonucleotide) (Fig 5B, Table 1). In contrast, equivalent or greater suppression of Fyn, Yes, cAbl, and Arg, did not reduce phosphorylation of Y15 (Fig. 5C–F, Table 1). Indeed, suppression of cAbl not only failed to reduce Cdk5(Y15) phosphorylation, but consistently increased it (Fig. 5E, Table 1). These findings identify cSrc as the kinase primarily responsible for phosphorylating Cdk5(Y15) in spreading lens epithelial cells, although other kinases may also contribute, particularly when cSrc is depleted.

Table 1.

siRNA suppression of cSrc decreases Cdk5(pY15) phosphorylation

siRNA	Relative target expression	Relative Cdk5(Y15) phosphorylation
Src	0.50 +/− 0.05	0.73 +/− 0.11
Fyn	0.27 +/− 0.06	1.01 +/− 0.12
Yes	0.07 +/− 0.04	1.21 +/− 0.04
Abl	0.50 +/− 0.13	2.67 +/− 0.60
Arg	0.35 +/− 0.09	1.07 +/− 0.26

The effectiveness of siRNA suppression of the target protein (relative target expression) and the relative Cdk5(Y15) phosphorylation are represented as the ratio of expression in siRNA transfected cells v/s non-transfected cells determined by densitometric scanning of immunoblots. Values represent the average of 3 independent experiments +/− s.e.m.

Mutations that inhibit Cdk5 activity have differing effects on Y15 phosphorylation –

To examine the functional consequences of Cdk5(Y15) phosphorylation, we first tested the ability of several known dominant negative mutations of Cdk5 to be phosphorylated on Y15: Cdk5(K33T) [19–21], Cdk5(D144N) [19,21,22], and Cdk5(Y15F) [5,6,20]. These mutations all showed similar expression levels (Fig. 6A) when over-expressed as GFP fusion proteins in lens epithelial cells. However, immunoblotting with a phosphospecific antibody revealed that they are not equivalent with respect to Y15 phosphorylation (Fig. 6). As expected, the Cdk5(Y15F) mutation was not phosphorylated at Y15; however, we were surprised to find that Cdk5(K33T) also failed to be phosphorylated, while phosphorylation of Cdk5(D144N) was consistently 3 to 4 fold greater than that of wild type Cdk5. Inhibiting Cdk5 activity with the pharmacological inhibitor, olomoucine, also produced a 2-3 fold increase in Y15 phosphorylation of wild type GFP-Cdk5 (Fig. 6B). Similar results were obtained when these constructs were expressed in N/N1003 cells (not shown). The increased Y15 phosphorylation produced by olomoucine and Cdk5(D144N) is consistent with the finding of cSrc-dependent phosphorylation of Y15 (Fig. 5, Table 1), since previous studies have shown that inhibition of Cdk5 activity increases the level of active cSrc [9,23].

FIG. 6. Dominant negative mutations of Cdk5 have differing effects on Y15 phosphorylation.

A. Cos7 cells were transfected with GFP or with GFP-tagged constructs of wild type Cdk5, Cdk5(Y15F), Cdk5(K33T), or Cdk5(D144N). After overnight incubation to allow expression of the transfected product, proteins were extracted and immunoblotted with phosphospecific antibody to detect GFP-Cdk5(pY15) and Cdk5 antibody to detect GFP-Cdk5. Results shown are typical of three independent experiments. B. N/N1003 cells were transfected with wild type GFP-Cdk5 or GFP-Cdk5(Y15F) as a negative control. Proteins were extracted after overnight incubation in the presence (+) or absence (−) of 15μM olomoucine and immunoblotted with phosphospecific antibody to detect GFP-Cdk5(pY15) or Cdk5 antibody to detect GFP-Cdk5. Results shown are typical of three independent experiments.

Cdk5(Y15) phosphorylation is required for stress fiber formation -

We next examined the cytoskeletal architecture of cells transfected with each of these mutant proteins. Cells were transiently transfected with the GFP-tagged fusion proteins, incubated overnight to allow expression, then replated on a fibronectin-coated surface and examined after 60min. Transfected cells were identified by GFP fluorescence (Suppl. Fig. 1) and cytoskeletal organization was viewed by rhodamine-phalloidin staining of actin (Fig. 7). In cultures transfected with GFP alone (Fig. 7A) or with wild type GFP-Cdk5 (Fig. 7B), less than 5% of cells failed to form stress fibers within 60min after plating (Table 2). In contrast, there were no visible stress fibers in 75% of cells transfected with Cdk5(Y15F) and 77% of cells transfected with Cdk5(K33T) at this time (Fig. 7C,D, Table 2). Moreover, cells transfected with the Cdk5(D144N) mutation, which lacks Cdk5 activity, but is well phosphorylated on Y15, retained the ability to form stress fibers (Fig. 7E, Table 2), suggesting that Y15 phosphorylation may be required, even in the absence of Cdk5 activity. We noted, however, that cell morphology and the subcellular organization of stress fibers were altered in cells expressing Cdk5(D144N), suggesting a functional role for Cdk5 activity as well (Fig. 7E). In addition, the effects of Cdk5(D144N) were mimicked by the pharmacological inhibitor, olomoucine, which also blocks Cdk5 activity without inhibiting Y15 phosphorylation (Fig. 7F, Table 2).

FIG. 7. Stress fiber formation requires Cdk5(Y15) phosphorylation independent of Cdk5 activity.

N/N1003 cells were transiently transfected with the indicated GFP-tagged construct or were incubated with 15μM olomoucine. The following day, cells were replated on fibronectin-coated chamber slides for 60min, fixed, and stained with rhodamine-phalloidin. Transfected cells were identified by GFP fluorescence (Suppl. Fig. 1). A. Cells transfected with the GFP parental vector (arrow) were indistinguishable from untransfected cells in the same field with respect to cytoskeletal architecture. B. Cells transfected with GFP-Cdk5 contained multiple, well-formed stress fibers at 60min. Both radial and transverse stress fibers were apparent. C. Cells transfected with GFP-Cdk5(Y15F) typically contained little polymerized actin and failed to form stress fibers. D. Cells transfected with GFP-Cdk5(K33T) also frequently showed weak overall F-actin staining and failed to form stress fibers. A portion of an untransfected cell containing stress fibers is visible in the lower left corner. E. Cells transfected with GFP-Cdk5(D144N) contained numerous stress fibers at 60min, but cell morphology and stress fiber arrangement were perturbed. F. Cells incubated in 15μM olomoucine overnight, then replated in the presence of the same concentration of olomoucine, also contained abundant stress fibers. Cells resembled GFP-Cdk5(D144N) transfected cells in morphology and stress fiber organization. G. Cells transfected with siRNA oligonucleotides for Cdk5 also often lacked stress fibers. (arrow indicates transfected cell) H.Cells transfected with a control, scrambled oligonucleotide showed no defect in stress fiber formation (arrow indicates transfected cell). Scale bar = 20μ.

Table 2.

Cdk5(Y15) Phosphorylation is Required for Stress Fiber Formation

Treatment	% w/o Stress Fibers	Chi-Square v/s control	p-Value
None	5% (19/377)	N/A	N/A
GFP	1% (3/25)	N/A	N/A
GFP-Cdk5	3% (5/144)	3.435	0.064
GFP-Cdk5(Y15F)	75% (21/28)	21.157	0.000*
GFP-Cdk5(K33T)	77% (54/97)	36.480	0.000*
GFP-Cdk5(D144N)	8% (4/50)	0.207	0.649
Olomoucine (15μM)	9% (9/100)	2.243	0.134
siRNA Cdk5	22% (66/299)	41.220	0.000*
Scrambled oligo	4% (12/290)	N/A	N/A

The percentage of cells without (w/o) visible stress fibers was determined by immunofluorescence of rhodamine phalloidin-stained samples under identical conditions at 40X magnification. All cells in one or more randomly selected fields were counted. Numbers in parentheses indicate number of cells without visible stress fibers divided by total number of cells counted. Chi-square analysis was performed in reference to the appropriate control: i.e. GFP for GFP-tagged Cdk5 constructs; no treatment (“None”) for olomoucine, and the scrambled oligonucleotide control for siRNA of Cdk5. N/A = not applicable;

^*indicates statistical significance. All treatments that blocked Y15 phosphorylation significantly suppressed stress fiber formation.

As an additional test of the need for Cdk5(Y15) phosphorylation, we used specific siRNA oligonucleotides to suppress expression of endogenous Cdk5 in the human lens epithelial cell line, FHL124. Since both Cdk5 activity and Cdk5(Y15) phosphorylation are reduced by suppressing Cdk5 expression, we expected cells transfected with siRNA oligonucleotides to resemble cells expressing Cdk5 mutations that lacked both activity and Y15 phosphorylation. Reducing Cdk5 expression to 48% of control by transient transfection with siRNA oligonucleotides, increased the percentage of cells that failed to form stress fibers more than 4-fold (22%, as compared to only 5% of cells transfected with a scrambled, control oligonucleotide) (Fig. 7G,H, Table 2). Thus, all treatments that blocked phosphorylation at Y15 significantly reduced stress fiber formation, whereas no significant effect was seen with treatments that did not affect Y15 phosphorylation. Together these findings indicate that cells have a stringent requirement for phosphorylated Cdk5(Y15) during stress fiber formation, which is independent of Cdk5 activity.

DISCUSSION

The present study demonstrates that phosphorylation of Cdk5 at Y15 is essential for stress fiber formation, even when Cdk5 kinase activity is blocked by site-specific mutation or by a pharmacological inhibitor. This finding suggests that Cdk5(pY15) may have adaptor functions, independent of Cdk5 activity, which are required to recruit SH2-domain proteins needed for assembly or stabilization of stress fibers. Such a role for Cdk5(pY15) is consistent with its subcellular localization on stress fibers and at the periphery of spreading cells, where focal adhesions and cytoskeletal regulatory proteins govern the formation of stress fibers and ensure their attachment to the extracellular matrix via integrins. Moreover, focal adhesions contain a number of SH2 domain proteins, which offer potential binding sites for Cdk5(pY15). These include the adaptor proteins Nck [24]and Crk [25], the Src regulatory protein, Csk [26], and the protein tyrosine kinases FAK [27]and Src [12]. Reports that both FAK and Src can be phosphorylated by Cdk5 in cells of neuronal origin [28] further suggest a role for Cdk5 at focal adhesions. Alternatively, Cdk5(pY15) may be involved in stabilizing stress fibers, by recruiting factors necessary for myosin-dependent contraction. The identification of proteins that interact preferentially with Cdk5(pY15) may provide additional insight into these possibilities.

Although the requirement for Cdk5(pY15) in stress fiber formation is independent of Cdk5 kinase activity, several observations suggest that Cdk5 activity also may have a role in stress fiber regulation. First, Cdk5 kinase activity increases in parallel with Cdk5(Y15) phosphorylation as cells spread, form stress fibers, and contract. In addition, both Cdk5(pY15) and p35 are closely associated with actin in stress fibers and at the cell periphery, suggesting that enzymatically active Cdk5/p35 kinase is likely to be present at these sites. This localization is consistent with the known interaction of p35 (and p39) with α-actinin [29]. Finally, the present results show that inhibiting Cdk5 kinase activity under conditions that do not block Y15 phosphorylation (such as expression of Cdk5(D144N) or treatment with olomoucine) does not reduce the percentage of cells that form stress fibers, but consistently alters cell morphology and stress fiber organization. Thus, our data suggest a model in which Cdk5(pY15) and Cdk5 kinase activity have distinct functions, with Cdk5(pY15) first serving as an adaptor to organize a multiprotein complex necessary for stress fiber formation and Cdk5 kinase activity subsequently regulating stress fiber organization, contraction, and cell shape.

The present results also link Y15 phosphorylation of Cdk5 during cell spreading to the Src family kinases and demonstrate that endogenous cSrc plays a major role in phosphorylating this site in lens epithelial cells. Previous studies have identified only two tyrosine kinases that phosphorylate Cdk5(Y15) in vivo: the non-receptor protein kinase cAbl [5], and the Src family kinase, Fyn [6]. Although cSrc has been shown to phosphorylate Cdk5(Y15) when overexpressed in transfected cells [6], the present study provides the first evidence that cSrc phosphorylates this site under physiological conditions. This is consistent with the known activation of cSrc in spreading cells, and its many functions in cytoskeletal reorganization [12]. Src activity is required for the formation of focal contacts [12], and subsequently limits cell spreading by targeting the Nck-PAK-PIX-PKL complex to focal adhesions [24]. In addition, Src family kinase function is required for cytoskeletal reorganization leading to stress fiber formation in spreading SYF −/− fibroblasts [17]. The present findings thus place phosphorylation of Cdk5 on Y15 among the many Src-dependent phosphorylations necessary for formation of focal contacts, their maturation into focal adhesions, and the subsequent formation of stress fibers.

In lens epithelial cells, stress fibers and the signals generated through cytoskeletal tension are important determinants of differentiation and cell fate [30]. The epithelial cells that normally cover the anterior surface of the lens contain a basal array of stress fibers [30,31]. As the epithelial cells differentiate at the lens equator to form terminally differentiated lens fiber cells [32], these stress fibers are disassembled and reorganized into a cortical ring [30]. This cytoskeletal reorganization appears to be a critical step, since the dissociation of stress fibers with cytochalasin or with pharmacological inhibitors of Rho signaling is sufficient to promote expression of the fiber cell differentiation markers, filensin and CP49 in vitro [30]. Interestingly, these same markers of differentiation are induced when Src family kinases are inhibited in lens epithelial cells [33]. Since Src-dependent phosphorylation of Cdk5(Y15) is necessary for stress fiber formation in lens epithelial cells, the present results suggest that Cdk5(Y15) phosphorylation may be an important step linking Src activity with suppression of fiber cell differentiation and maintenance of the epithelial cell state.

Supplementary Material

01

SUPPL. FIG. 1. GFP-fluorescence identifies transfected cells. Green-channel fluorescence of the images shown in Fig. 6 was used to identify the transfected cells in each field. A. Cell transfected with GFP only. B. Cell transfected with GFP-Cdk5. C. Cell transfected with GFP-Cdk5(Y15F), showing accumulation of the GFP-tagged protein along the cell periphery (arrows). D. Cell transfected with GFP-Cdk5(K33T), also showing slight accumulation at cell periphery (arrows). E. Cell transfected with GFP-Cdk5(D144N). F. Untransfected cells treated with olomoucine. Lack of fluorescence in this image confirms that no phalloidin immunofluorescence was observed in the green channel. G. Cell transfected with siRNA oligonucleotides to Cdk5. H. Cell transfected with control, scrambled oligonucleotide.

Abbreviations:

Cdk

cyclin-dependent kinase

GFP

green fluorescent protein

PBS

phosphate-buffered saline

BisTris

2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol