CDK5 Regulates Cell Adhesion and Migration in Corneal Epithelial Cells

12 Vol. 1, 12–24, November 2002 Molecular Cancer Research

Chun Gao,1 Sewite Negash,1 Hong Tao Guo,1 Dolena Ledee,1 Hwai-Shi Wang,2 and Peggy Zelenka1

1National Eye Institute, NIH, Bethesda, MD; and 2Yang Ming University, Taipei, Taiwan

Abstract CDK5 must form a complex with one of two known regulatory CDK5 and its activator, p35, are expressed in mouse subunits, p35 and p39, both of which are expressed at high corneal epithelium and can be coimmunoprecipited from levels in neurons (5, 18–20). These activators of CDK5 have corneal epithelial cell lysates. Immunostaining shows been shown to bind to cytoskeletal components such as actin CDK5 and p35 in all layers of the corneal epithelium, filaments (16, 21), microtubules (11, 21), and the cytoskeletal especially along the basal side of the basal cells. Stable scaffolding protein, Cables (22). transfection of corneal epithelial cells with CDK5, which A number of observations suggest that CDK5 may also have increases CDK5 kinase activity by approximately 33%, important functions in non-neuronal cells, especially during also increases the number of cells adhering to fibro- development and differentiation. CDK5/p35 activity seems to nectin and the strength of adhesion. CDK5 kinase be an essential factor in monocytic differentiation (23–25). Not activity seems to be required for this effect, because the only do peripheral monocytes express high levels of CDK5 kinase inactive mutation, CDK5-T33, either reduces activity, but overexpression of CDK5/p35 in HL60 cells is adhesion or has no significant effect, depending on the sufficient to force differentiation along the monocytic pathway. level of expression. Using an in vitro scrape wound in CDK5 kinase activity also has been associated with myogenic confluent cultures of stably transfected cells to examine differentiation (26, 27), and seems to be required for expression the effect of CDK5 on cell migration, we show that of the muscle differentiation markers, myogenin and troponin T reoccupation of the wound area is significantly (26). Finally, blocking CDK5 kinase activity by injecting decreased by CDK5 and increased by CDK5-T33. These CDK5-T33 into one of the two dorsal cells of a four-cell findings indicate that CDK5 may be an important Xenopus embryo leads to numerous abnormalities in the deve- regulator of adhesion and migration of corneal epithelial lopment of the eye, including microphthalmia, disruption of eye cells. structure, and malformation of the lens (28). Because this phenotype can be rescued by concomitantly injecting wild-type CDK5, the developmental defects seem to be caused by inhi- Introduction bition of endogenous CDK5 activity. In keeping with the above CDK5 is a member of the cyclin-dependent kinase family of findings, the CDK5 activating protein, p35, has been found in proline-directed protein kinases (1). Although most of the monocytes (24), muscle (29), lens (30), and retina (31). members of this family are involved in cell cycle regulation, Although CDK5 expression and function have been CDK5 is preferentially expressed in terminally differentiated investigated in the lens and retina, the available information neurons of the developing and adult nervous system (2, 3) and provides little insight into its involvement in eye development has an essential role in regulating the complex migration and differentiation. CDK5 expression has been observed in the program of postmitotic neurons during embryogenesis (4–7). retina, where it is correlated with developmental neuroplasticity Several other neuronal functions of CDK5 have also been (32). This finding is consistent with numerous studies identified, including cell-cell adhesion (8), cell-matrix adhesion indicating the importance of CDK5 for neuronal differentiation (9), neurite extension (4), and cytoskeletal regulation (10, 11). (4, 5, 7, 33). In addition, CDK5 has been reported to phosphory- A number of known CDK5 substrates in neurons are either late the regulatory subunit of cGMP phosphodiesterase (Pg)in cytoskeletal proteins or proteins involved in cytoskeletal the rod outer segments, suggesting that it may regulate some regulation. Examples include the neurofilament proteins, NH aspect of the visual transduction cycle (31, 34). Work from our H and NM (10, 12, 13), the microtubule-associated proteins, and laboratory has shown that differentiating fiber cells of the rat Map1b (14, 15), the protein kinase PAK1 (16), which regulates lens contain active CDK5/p35 (30). The present study examines h actin polymerization, and -catenin (8, 17), a multifunctional the expression of CDK5 and p35 in the corneal epithelium and component of cadherens junctions. To be enzymatically active, tests the possibility that CDK5 may have a role in regulating cell-matrix adhesion of corneal epithelial cells.

Received 2/26/02; revised 6/24/02; accepted 7/22/02. Results The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in CDK5 and Its Activator, p35, Are Expressed in Mouse accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Corneal Epithelium Requests for reprints: Peggy S. Zelenka, NIH/NEI, Room 214, Building 6, To examine the expression of CDK5 and its regulatory 6 Center Drive MSC 2730, Bethesda, MD 20892-2730. E-mail: [email protected] subunit, p35, in the corneal epithelium, corneal epithelial Copyright D 2002 American Association for Cancer Research. lysates were immunoblotted using antibodies specific for the

COOH termini of CDK5 and p35. Anti-CDK5 antibody detected a single strong immunoreactive band of the correct molecular weight (Fig. 1A) that comigrated with an immunoreactive band in the brain lysate (Fig. 1A). Antibody against p35 also detected a single immunoreactive band of approximately Mr 35,000 that comigrated with the corresponding immunoreactive band from the brain lysate (Fig. 1A). The specificity of the anti-p35 antibody was confirmed by complete neutralization of the immunoreactivity by the antigenic peptide (not shown; see Fig. 4B). To determine whether CDK5 and p35 form an intracellular protein complex, proteins were immunoprecipitated from lysates of corneal epithelium or brain using anti-CDK5 antibody, then immunoblotted with anti-CDK5 and anti-p35 antibodies. Anti-CDK5 antibody detected CDK5 in the immunoprecipitates from both tissues (Fig. 1B), confirming that the immunoprecipitation was successful. Anti-p35 antibody also detected p35 in the CDK5 immunoprecipitates from both tissues (Fig. 1B), indicating that p35 and CDK5 form an intracellular protein complex in corneal epithelium as well as in the brain. As in the immunoblots of whole cell lysates, the p35 band in CDK5 immunoprecipitates from corneal epithelium comigrated with p35 from the brain, further confirming its identification. As an additional assay for the expression of p35, we performed reverse transcription-PCR (RT-PCR), using oligonucleotides specific for p35 mRNA. A single RT-PCR product of the predicted size (923 bp) was detected (Fig. 1C) in total RNA from corneal epithelium and from A6(1) corneal epithelial cells. Partial sequencing of the isolated PCR product confirmed that it was derived from p35 mRNA. In contrast, RT-PCR did not detect any expression of p39 from total RNA in corneal epithelium or A6(1) cells, although the 160-bp RT-PCR product corresponding to p39 mRNA was observed in an equal amount of total RNA from the brain (Fig. 1D). Thus, corneal epithelial cells appear to express the CDK5 activating protein, p35, as judged by RT-PCR, partial nucleotide sequence, immunoreactivity, comigration with brain p35, and the ability to bind CDK5. However, these cells do not express detectable levels of p39.

CDK5 and p35 Colocalize in Mouse Cornea FIGURE 1. Expression of protein and mRNA for CDK5 and p35 in To examine the distribution and subcellular localization of mouse corneal epithelium. A. Immunoblots of whole cell extracts. Protein extracted from adult mouse corneal epithelium was immunoblotted with CDK5 and p35 in mouse cornea, immunocytochemical staining antibody to CDK5 and p35. The protein bands identified by anti-CDK5 was performed on paraffin sections from 2-month-old normal from corneal epithelium comigrated with CDK5 from the brain and the p35 mice. In general, CDK5 and p35 showed similar localization band in the corneal epithelia also comigrated with p35 from the brain lysate. B. Coimmunoprecipitation experiments. Whole cell lysates from patterns. Positive staining for CDK5 (Fig. 2, A–C) and p35 corneal epithelium and brain were immunoprecipitated using anti-CDK5 (Fig. 2, E–G) was observed in the cytoplasm of all layers of antibody. Immunoblotting was performed on immunoprecipitates using anti-CDK5 and p35 antibodies. Anti-CDK5 detected CDK5 on immuno- corneal epithelial cells, especially along the basal side of the precipitates from both corneal epithelium and brain. A p35 immunoreactive basal cells (Fig. 2, C and G, arrows). Signals were strong in the band was detected in the CDK5 immunoprecipitates from the corneal central corneal epithelium, but immunoreactivity for both CDK5 epithelium, which comigrated with p35 from the brain. C. RT-PCR for p35 mRNA. RT-PCR was performed on total RNA extracted from the corneal (Fig. 2B, arrow) and p35 (Fig. 2F, arrow) was reduced in the epithelium and A6(1) cells. A single product for p35 with correct size (923 limbus region. Interestingly, just beyond the limbus region, in bp) was detected. Controls contained RNA from the corneal epithelium or the conjunctival epithelium, CDK5 and p35 staining became the A6(1) cells, but no reverse transcriptase (ÀRT). D. RT-PCR for p39 mRNA. RT-PCR was performed on total RNA from brain, corneal strongly positive again (Fig. 2, B and F, arrowhead). Corneal epithelium, and the A6(1) cells. The assay did not detect p39 mRNA in endothelium was also positively stained for CDK5 (Fig. 2C). In corneal epithelium or A6(1) cells although it successfully detected p39 mRNA in the brain RNA included as a positive control. This PCR product addition, the inner nuclear layer of the retina (Fig. 2, A and E), was not present in a negative control containing brain RNA without reverse lens epithelial cells, and the cortical fiber cells of the lens stained transcriptase (not shown).

positive for CDK5 (Fig. 2, A and B) and p35 (Fig. 2, E and F). Stable Transfection of A6(1) Cells with CDK5 No specific staining was observed in controls in which the To examine the physiological function of CDK5 in corneal antigenic peptides were included during incubation with the epithelial cells, a mouse corneal epithelial cell line, A6(1), primary antibodies for CDK5 (Fig. 2D) and p35 (Fig. 2H). was stably transfected with GFP-tagged constructs of CDK5 or the kinase inactive form, CDK5-T33, which contains a K Subcellular Distribution of CDK5 and p35 to T substitution at amino acid 33 (4). If expressed at In the brain, CDK5 is found primarily in the cytosol, sufficiently high concentrations, CDK5-T33 can exert a whereas p35 is mostly associated with the membrane fraction dominant negative effect on endogenous CDK5 activity by (35). To investigate whether CDK5 and p35 have similar sequestering the available p35 (4). GFP-CDK5 and GFP- subcellular distributions in mouse corneal epithelium, we CDK5-T33 were expressed at comparable levels in stably prepared cytosolic, cytoskeletal, and membrane fractions of transfected A6(1) cells (Fig. 4A, arrow), and their expression corneal epithelial cells. Immunoblotting showed that although did not affect the levels of endogenous CDK5 (Fig. 4A, CDK5 and p35 were present to some degree in all three arrowhead) or p35 (Fig. 4B, open arrow). Expression of fractions, the cytoplasmic fraction contained the bulk of the CDK5 and CDK5-T33 also had no observable effect on CDK5 (Fig. 3A), whereas the membrane fraction was enriched growth or survival of A6(1) cells (not shown). Although with p35 (Fig. 3B). CDK5 and p35 antibodies recognized expression of GFP-tagged CDK5 and CDK5-T33 was multiple bands in all three fractions (Fig. 3, A and B). These are relatively low as compared to endogenous CDK5 (Fig. 4A), likely due to phosphorylated forms (18, 19, 22, 36, 37), which more than 95% of the stably transfected A6(1) cells expressed are preserved due to the presence of phosphatase inhibitors in the constructs, as judged by microscopic examination of GFP the fractionation buffer. fluorescence (not shown).

FIGURE 2. Immunohistochemical staining for CDK5 (A – D) and p35 (E – H) on paraffin sections from adult mouse cornea. Sections were incubated with antibody to CDK5 (A – D) or p35 (E – H). To determine the level of nonspecific staining, the corresponding blocking peptides were included during incubation with the primary antibody (D,H). Positive staining was seen for CDK5 (A – C) and p35 (E – G) in corneal epithelium, but the staining for both CDK5 (B, arrow) and p35 (F, arrow) was attenuated within the limbus region. CDK5 and p35 were also strongly stained in the conjunctival epithelium (B,F, arrowheads). CDK5 and p35 appear to be present in the cytoplasm of all layers of corneal epithelium, especially along the basal side of basal cells (C,G, arrows). Staining for both CDK5 and p35 was also positive in the epithelium and cortical fiber cells of the lens (A,B and E,F), and inner nuclear layer of the retina (A,E, arrow). Scale bar, 250 AmforA, D, E, and H; 125 Amfor B,F;25AmforC,G. L, lens; R, retina; C, cornea; S, stroma.

was seen only in the cytoplasm, especially in the nuclear/ perinuclear region (Fig. 5A). In contrast, the GFP-tagged CDK5 constructs were localized not only in the cytoplasm, but also in filopodia, and at the edges of lamellipodia, where cells make focal contacts with the extracellular matrix (Fig. 5, B, C, E, and F). Endogenous CDK5 in control cells transfected with GFP only and in untransfected cells was similarly localized in the cytoplasm, in filopodia, and along the edges of lamellipodia (Fig. 5D). Thus, neither overexpression nor the presence of the GFP tag seems to alter the subcellular localization of CDK5.

Cell-to-Matrix Adhesion To test whether CDK5 might be involved in cell adhesion, stably transfected A6(1) cells were allowed to adhere to FIGURE 3. Differential distribution of CDK5 (A) and p35 (B)in subcellular fractions from corneal epithelium. Corneal epithelia were fibronectin for 2 h. The plates were then inverted and the cells separated into three different fractions: cytosol, cytoskeleton, and were centrifuged to determine the force necessary to dislodge membrane by differential detergent extraction. Immunoblotting was then performed on equal amounts of protein from each fraction using antibodies them. When subjected to gentle centrifugation (50 Â g)to against (A) CDK5 and (B) p35. The multiple bands for CDK5 and p35 in these preparations are likely due to post-translational modifications.

CDK5 Kinase Activity To determine whether stable transfection with CDK5 and CDK5T33 altered CDK5 kinase activity, CDK5 was immunoprecipitated from extracts of GFP-transfected control A6(1) cells, GFP-CDK5-A6(1) cells, and GFP-CDK5-T33-A6(1) cells, and in vitro kinase assays were performed using a biotinylated peptide substrate (Fig. 4C). For comparison, we also measured CDK5 kinase activity in immunoprecipitates from extracts of newborn rat brain and mouse corneal epithelium (not shown). The results showed that A6(1) cells contain a low level of endogenous CDK5 activity (approximately 0.24 pmol PO4 transferred/min/mg cellular protein). Under the same assay conditions, newborn rat brain contained about 20 times as much CDK5 activity (4.66 pmol PO4 transferred/min/mg cellular protein), whereas corneal epithelial extracts contained no detectable activity. Stable transfection with CDK5 increased the activity in A6(1) cells to 0.32 pmol PO4 transferred/min/mg cellular protein, an increase of 33%. This increase was statistically significant (P < 0.03). In contrast, stable transfection with CDK5-T33 produced a small, but not statistically significant, reduction in CDK5 activity (to 0.21 pmol PO transferred/min/mg cellular protein). Thus, CDK5 4 FIGURE 4. Stable transfection of A6(1) cells with CDK5 and CDK5- activity is elevated in A6(1) cells stably transfected with CDK5, T33. A. A6(1), a corneal epithelial cell line, was stably transfected with and is not significantly changed in cells expressing an equiv- pGFP only, pGFP-CDK5, and pGFP-CDK5-T33. The expression of pGFP- CDK5 and pGFP-CDK5-T33 was examined by immunoblotting using alent amount of CDK5-T33. antibodies against CDK5. The single arrow points to exogenous CDK5, whereas the arrowhead indicates endogenous CDK5. B. Proteins from A6(1) cells stably transfected with pGFP only, pGFP-CDK5, and pGFP- Localization of GFP-CDK5 and GFP-CDK5-T33 CDK5-T33 were immunoblotted with antibody to p35. The level of ex- Because the biological effects of protein kinases may also pression of endogenous p35 was not affected by the expression of be affected by their subcellular localization, we tested whether exogenous CDK5 (left panel, open arrow). The p35-immunoreactive band was obliterated by the presence of p35 blocking peptide (right panel). C. localization of GFP-CDK5 or GFP-CDK5-T33 corresponded CDK5 kinase activity. CDK5 was immunoprecipitated from GFP-, GFP- to the normal subcellular localization of CDK5. The CDK5-, and GFP-CDK5-T33-transfected A6(1) cell lysates. Kinase activity distribution of GFP, GFP-CDK5, and GFP-CDK5-T33 fluo- of the immunoprecipitated protein was assayed in vitro using a biotinylated peptide substrate (PKTPKKAKKL). The results of six independent rescence in transiently transfected A6(1) cells was compared measurements from three experiments were averaged and expressed with the immunolocalization of CDK5 in control cells as pmol ATP/min/mg of total A6(1) protein. The differences between the GFP-CDK5 and GFP only, and GFP-CDK5- and GFP-CDK5-T33- transfected with GFP alone and in untransfected A6(1) cells. transfected cells are statistically significant (P < 0.03 and P < 0.01, In cells transfected with the GFP tag alone, GFP fluorescence respectively).

dislodge nonadherent or weakly adherent cells, twice as many ments indicated that the association of vinculin with the CDK5-transfected cells remained attached as CDK5-T33- cytoskeleton increased about 30% in CDK5-transfected cells transfected or nontransfected A6(1) cells (Fig. 6), indicating (Fig. 7B). This finding suggests that CDK5 overexpression that CDK5 enhances cell-to-matrix adhesion. Increasing the promotes focal adhesion formation and/or linkage to the centrifugation force to 200Âg dislodged many adherent CDK5- cytoskeleton. T33-transfected cells and nontransfected cells, but had no effect on CDK5-transfected A6(1) cells. In fact, even at 450 Â g,the Adhesion in A6(1) Cells Infected with Recombinant CDK5 number of adherent CDK5-transfected cells decreased only Adenovirus Vectors slightly, indicating that CDK5 overexpression increases the As a further test of the physiological role of CDK5 in cell strength of adhesion. The differences between CDK5-trans- adhesion, we infected A6(1) cells with recombinant adenovirus fected cells and CDK5-T33-transfected cells or control cells constructs Adv-CDK5 or Adv-CDK5-T33. The cDNAs carried were statistically significant (P < 0.001) at all centrifugation by the adenovirus vectors were not ‘‘tagged’’ by additional forces tested. residues at the NH2 terminus to minimize possible differences in protein structure. Adenovirus-mediated cDNA transfer Association of Vinculin with the Cytoskeleton seemed to yield higher expression of CDK5 and CDK5-T33 Since cell-to-matrix adhesion is associated with formation relative to endogenous CDK5 than stable transfection (compare of vinculin-containing adhesion plaques linked to the Figs. 4A and 8A). The total level of exogenous protein (CDK5 detergent-insoluble cytoskeletal fraction, we tested whether or CDK5-T33) and endogenous CDK5 in the infected A6(1) CDK5 affects the association of vinculin with the detergent- cells was about twice as high as endogenous CDK5 expression insoluble fraction. Stably transfected A6(1) cells were in noninfected, control cells (Fig. 8A), indicating that the separated into a Triton X-100 soluble fraction (cytosol) exogenous and endogenous proteins are present in approx- and a 9 M urea soluble fraction (cytoskeleton), and both imately equal amounts. fractions were immunoblotted with vinculin antibody. GFP- Fibronectin-coated adhesion strips were used to examine the CDK5-transfected A6(1) cells had more vinculin in the effect of CDK5 and CDK5-T33 on cell adhesion in the cytoskeletal fraction than cells stably transfected with GFP- adenovirus vector-infected cells. Overexpression of CDK5 CDK5-T33 or control cells (Fig. 7, A and B). Quantitative increased the number of adherent cells approximately 2-fold as densitometry of six immunoblots from four separate experi- compared to noninfected controls (P < 0.0005) (Fig. 8B), in

FIGURE 5. Immunolocalization of CDK5 in A6(1) cells transfected with GFP, GFP-CDK5, and CDK5-T33. A6(1) cells were transiently transfected with GFP (A,D), GFP-CDK5 (B,E), or GFP-CDK5-T33 (C,F). GFP fluorescence was viewed to identify transfected cells and to localize the GFP-tagged proteins (A – C). Immunofluores- cence using anti-CDK5 antibody was used to localize endogenous CDK5 (D) as well as the GFP-tagged CDK5 wild-type and mutated fusion proteins (E,F). CDK5 immunostaining was seen in the cytoplasm, in filopodia (arrows in D – F), and in lamellopodia (arrowheads in D – F). Panel D shows a GFP-transfected cell and an untransfected cell in the same field.

of gene expression is different from that of central cornea (42). For example, the corneal differentiation marker, keratin 12, is not expressed in this region (43–45). The limbus region is enriched with a distinct subpopulation of stem cells, serving as a proliferative reserve (40). To generate new epithelial cells, limbal stem cells move centripetally into the basal layer of the corneal epithelium, where they divide, differentiate, detach from the basement membrane, and eventually slough off from the superficial layer. The present study shows that CDK5 and p35 expression is closely correlated with corneal differentiation. CDK5 and p35 are expressed to appreciable levels throughout the corneal epithelium, except for the limbus. In fact, expression of CDK5 and p35 correlates well with the differentiation marker keratin 12, when both are stained in parallel (not shown). Thus, as in other cell types that have been studied (24, 26, 30, 46, 47–50), expression of CDK5 and p35 FIGURE 6. Effect of CDK5 on cell adhesion. Adhesion to fibronectin was examined by a centrifugation assay at 50, 200, and 450 Â g using in the corneal epithelium seems to be associated with cellular CDK5-transfected A6(1) ( 6 ), control-transfected A6(1) cells (o), or differentiation. CDK5-T33-transfected A6(1) cells ( ! ). Centrifugation at 200 Â g The possibility that CDK5 and p35 might play a role in cell- dislodged control cells and CDK5-T33-transfected cells, but had no effect on CDK5-transfected cells. The differences between CDK5-transfected to-matrix adhesion in the corneal epithelium was first cells and control cells, or CDK5-transfected cells and CDK5-T33- transfected cells were significant (P < 0.001) at each centrifugation force tested. good agreement with the results obtained using stably transfected cells and the centrifugation assay. Interestingly, however, with this method of cDNA transfer, CDK5-T33 overexpression significantly reduced the number of adherent cells as compared to controls (P < 0.001).

Migration in A6(1) Cells Stably Transfected with CDK5 Since cell adhesion is an important component of cell migration, the observed effect of CDK5 on corneal epithelial cell adhesion suggested that it might also affect cell migration. We tested this possibility using an in vitro scrape wounding assay (38). GFP-CDK5-, GFP-CDK5-T33-, and GFP-transfected control cells were maintained at 37jC for 5 days in the absence of IFN-g to establish confluence. A uniform scrape wound was made across the culture dish and the ability of the cells to migrate was assessed by monitoring reoccupation of the wound area (Fig. 9A). After 48 h, the area reoccupied by GFP- CDK5-transfected A6(1) was 30% less than that reoccupied by control cells transfected with the vector only (P < 0.01), whereas the area reoccupied by GFP-CDK5-T33-transfected cells was 20% greater than control (P < 0.05) (Fig. 9B). BrdUrd labeling of scrape wounded cultures showed very low levels of proliferation during the 48-h period, with no significant difference among the three cell lines (not shown). Thus, we attribute the differences in reoccupation of the wound area to differences in cell migration. FIGURE 7. Increased association of vinculin with the Triton X-100 insoluble cytoskeletal fraction in A6(1) cell transfected with CDK5. A6(1) cells stably transfected with vector only, pGFP-CDK5, and pGFP-CDK5- Discussion T33 were maintained at 37jC for 5 days in the absence of IFN-g to The corneal epithelium must be constantly renewed to establish confluent, quiescent cultures. Cell lysates were separated into maintain a smooth optical refractive surface and an intact soluble and insoluble fractions. The soluble and insoluble proteins were immunoblotted with anti-vinculin antibody. A. Representative Western blot physical barrier for the eye (39–44). This renewal is showing the immunoreactive band with correct molecular size of vinculin. accomplished by coordinated proliferation, differentiation, and B. Quantitative densitometry of six different immunoblots from four separate experiments. Vinculin in the insoluble fraction from CDK5- migration of stem cells from the limbus, a junctional zone transfected cells was approximately 30% greater than in vector-only between the cornea and conjunctiva. In the limbus, the pattern controls (P < 0.05) or cells transfected with CDK5-T33 (P < 0.03).

ECM molecule for attachment of these cells, as for other corneal epithelial cells in culture (54, 55), although it is not a major ECM component in the corneal basement membrane in vivo (56). Two different adhesion assays using two different cDNA transfer techniques indicated that CDK5 overexpression strongly promotes adhesion of A6(1) cells to fibronectin. Because these assays allow 2 h for cell attachment, they measure a relatively late stage of cell adhesion that includes assembly of focal adhesion complexes, cytoskeletal engage- ment, and formation of stress fibers (57, 58). Indeed, cell fractionation studies demonstrated that stable transfection with CDK5 significantly enhanced the association of the focal adhesion component, vinculin, with the detergent- insoluble cytoskeletal fraction. We also found that the kinase-inactive construct, CDK5-T33, suppressed adhesion to fibronectin, when expressed at high levels using an adenoviral vector. We attribute this to the ability of CDK5- T33 to exert a dominant negative effect on endogenous CDK5 activity when expressed in high enough concentrations to sequester the available p35. The ability of CDK5 to regulate adhesion to fibronectin is especially interesting in view of the fact that fibronectin is transiently expressed under the flattened, actively migrating cells at the leading edge of a corneal wound (56). Although we did not detect CDK5 kinase activity in the intact corneal epithelium, it is possible that this enzyme may be activated when the epithelium is wounded, to regulate adhesion to fibronectin during wound FIGURE 8. Adenovirus-mediated transfer of CDK5 and CDK5-T33. A. healing. Immunoblots of CDK5 and CDK5-T33 in A6(1) cells following adenovirus- mediated gene transfer. Expression of CDK5 or CDK5-T33 in infected Cell migration studies also support a role for CDK5 in A6(1) cells is increased about 2-fold compared to CDK5 expression in non- regulating cell-to-matrix attachment. Cell migration is a infected cells. B. Effect of CDK5 on cell adhesion to fibronectin. Adhesion coordinated series of integrated events consisting of lamelli- strips were used to examine the effect of CDK5 or CDK5-T33 on cell adhesion to fibronectin in A6(1) cells infected with recombinant CDK5 or podial extension, formation of focal adhesion contacts at the CDK5-T33 adenovirus. Results represent the average of three to five front of the cell, cytoskeletal contraction, and release of cell- independent measurements. The number of adherent cells was significantly increased by Ad-CDK5 and decreased by Ad-CDK5-T33 as to-substrate attachments at the rear (59, 60). The ability of a compared to control, uninfected cells (P < 0.001). cell to migrate is contingent on disengaging trailing focal adhesions and generating new ones at the leading edge (61, 62). As a result, the relationship between adhesion and suggested by their immunolocalization along the basal aspect migration is a bell-shaped curve (59), with the highest rate of of the basal cells, where hemidesmosomes maintain the stable migration occurring at some intermediate value of adhesive attachment of cells to the basement membrane (51). Cell strength. The strong enhancement of cell-to-matrix adhesion fractionation studies confirmed that the bulk of the p35 in the produced by CDK5 overexpression in A6(1) cells could corneal epithelium is associated with the membrane fraction. reduce cell migration by opposing the disassembly of the This localization is also consistent with the presence of an trailing adhesions. Conversely, since disassembly of trailing NH2-terminal myristoylation site in p35, which targets p35 to adhesions is usually considered to be the rate-limiting factor in the membrane (35). migration (59, 61, 63), a slight decrease in CDK5 activity Adhesion studies performed on the mouse corneal epithelial might increase migration rate by promoting this disassembly. cell line, A6(1), provided a second line of evidence supporting At some point, however, any additional decrease in CDK5 a role for CDK5 in cell adhesion. A6(1) cells are conditionally activity would be expected to reduce adhesive strength below transformed by a temperature-sensitive SV40 T-antigen. When the minimum needed to form new attachments effectively, and cultured at the nonpermissive temperature in the absence of the rate of migration would be decreased. This is consistent IFN-g for 5 days, A6(1) cells exhibit some characteristics of with the observation that neuronal migration is inhibited in differentiated corneal epithelial cells and express certain mice with targeted disruptions of CDK5 or p35 (5–7). Thus, differentiation markers such as keratin 12 and transketolase the present results and the available literature on the effect of (52, 53). However, there are also some important differences CDK5 deficiency on cell migration are consistent with a role between these cells and cells of the corneal epithelium. Like for CDK5 in regulating cell-to-matrix adhesion. Nevertheless, most cultured cells, A6(1) cells attach to the matrix through we do not rule out the possibility that CDK5 may also affect integrin-mediated focal adhesion complexes rather than hemi- migration in ways that are independent of its effect on desmosomes. Moreover, fibronectin is the most favorable adhesion.

In addition, we note an interesting reciprocal relationship T33, seems unlikely in view of the overall structural similarity between CDK5 and pinin (DRS/memA), a protein associated between these two proteins (16). Moreover, since we have with desmosomes of corneal epithelial cells (64). Overexpres- relied on the endogenous activating protein, p35, to activate the sion of pinin seems to strengthen cell-cell adhesion and decrease exogenous CDK5, the observed 33% increase in kinase activity cell-matrix adhesion (65), exactly opposite to the effects of is likely to be within physiological limits, thus minimizing the CDK5 observed in this study. Moreover, overexpression of phosphorylation of nonphysiological substrates. Finally, the pinin suppresses CDK5 expression (38). Because conditions simplest explanation for the observation that high levels of that favor cell-cell adhesion often reduce cell-matrix adhesion, CDK5-T33 have the opposite effect on adhesion is that CDK5- and vice versa (65), this reciprocal relationship between pinin T33 inhibits the endogenous CDK5 activity. Certainly, CDK5- and CDK5 lends further support to the view that CDK5 may be T33 has the potential to be a highly specific inhibitor of an important regulator of corneal cell adhesion and migration. endogenous CDK5 activity by sequestering p35. Thus, the The available evidence supports the interpretation that results of this study support a model in which endogenous endogenous CDK5/p35 is active and regulates adhesion and CDK5/p35 regulates corneal epithelial cell adhesion and migration in cultured corneal epithelial cells by phosphorylating migration by phosphorylating specific proteins involved in protein substrates involved in these processes. Accordingly, we these processes. Since our data indicate that p39 is not have found that overexpressing CDK5 increases both CDK5 expressed in the corneal epithelium or in A6(1) cells, CDK5/ activity and adhesive strength. Since the kinase-inactive form, p39 is not an essential component of this regulatory pathway. CDK5-T33, does not affect adhesion, kinase activity seems to One possible mechanism for the observed effects of CDK5 be required for this effect. The alternative explanation, that on adhesion and migration is through phosphorylation of the CDK5 binds to cellular proteins that fail to interact with CDK5- Rac1 effector, PAK1. A previous report has demonstrated that

FIGURE 9. The effect of CDK5 on cell migration. A6(1) cells stably transfected with vector only, pGFP-CDK5, and pGFP-CDK5-T33 were maintained at 37jC for 5 days in the absence of IFN-g to establish confluent, quiescent cultures. A uniform scrape wound was made across the culture dish at 0 h. The ability of A6(1) cells to reoccupy the scraped area was compared between 0 h and after 48 h. A. A representative micrograph showing the effect of CDK5 and CDK5-T33 on cell migration. B. Quantification of the reoccupied area after 48 h by image analysis software. Six measurements were made from three separate experiments for each cell type. Reoccupation of the wound area by CDK5-transfected A6(1) cells was 30% slower than control (P < 0.01), whereas reoccupation by pGFP- CDK5-T33-transfected A6(1) cells was 20% faster (P < 0.05).

CDK5/p35 specifically binds activated Rac1, thus targeting Immunoblotting and Immunoprecipitation PAK1 for phosphorylation by CDK5 and suppressing its Immunoblotting and immunoprecipitation were performed activity (16). PAK1 is a known regulator of cell adhesion and as previously described (30). Cellular proteins were lysed and migration. Constitutively activated PAK1 reduces the number extracted in PBSTDS (1% Triton X-100, 0.5% sodium of focal adhesions and causes loss of stress fibers (66), increases deoxycholate, 0.1% SDS in PBS) containing one Complete- cell contractility (67, 68), and promotes polarized cell move- Mini2 protease inhibitor cocktail tablet/10 ml. Cell lysate ment (68, 69). Conversely, dominant negative PAK1 has been containing 200 Ag of protein was immunoprecipited using anti- shown to decrease migration of human microvascular endothe- CDK5 rabbit polyclonal IgG (C-8) (sc-173, Santa Cruz lial cells (68). Thus, inactivation of PAK1 by CDK5 would be Biotechnology, Santa Cruz, CA). Control immunoprecipitation expected to promote cell adhesion by increasing focal adhesions was performed with an equivalent amount of cell extract in the and stress fiber formation. At the same time, CDK5-dependent absence of primary antibody. Immunoprecipitated proteins or inactivation of PAK1 could interfere with cell migration by 25–50 Ag of total cell extract were immunoblotted using anti- inhibiting the generation of contractile force needed to pull off CDK5 mouse monoclonal IgG (DC-17; sc-249, Santa Cruz the rear attachment during migration, or alternatively, by Biotechnology) or anti-p35 rabbit polyclonal IgG (C-19; sc- disrupting the coordinated nature of directed movement (68, 820, Santa Cruz Biotechnology). Where indicated, 2–3 Agof 69). Additional studies will be needed to determine whether p35 blocking peptide (sc-820P, Santa Cruz Biotechnology) PAK1 phosphorylation is a component of the regulatory were added during incubation with primary antibody. Immuno- pathway affected by CDK5 in corneal epithelial cells. reactive bands were detected using horseradish peroxidase- linked anti-rabbit IgG (Santa Cruz Biotechnology) by enhanced chemiluminescence (ECL-Plus; Amersham Life Science, Materials and Methods Piscataway, NJ). Isolation of Mouse Corneal Epithelia All animal studies were performed in accordance with the Immunohistochemistry NIH Guidelines for Care and Use of Laboratory Animals. Two- Paraffin sections (10 Am) from 2-month-old mouse eyes on month-old mice were obtained from Charles River Farms silanized slides (Digene, Gaithersburg, MD) were deparaffi- (Charles River Breeding Laboratories, Kingston, NY) and nized by Hemo-De (Fisher, Pittsburgh, PA) twice for 5 min euthanized by asphyxiation in 95% carbon dioxide. The eyes each. After rehydration in a series of decreasing ethanol were enucleated and transected posterior to the corneal limbus concentrations, samples were permeabilized in 0.25% Triton under a dissecting microscope. Corneal tissues were washed in X-100 in PBS for 10 min and postfixed in Bouin’s solution PBS and placed in Dispase II solution at 2.4 units/ml for 1.5– (Sigma Chemical Co., St. Louis, MO) for 15 min. The 2 h at 37jC (Roche Diagnostics, Basel, Switzerland). Sheets of samples were incubated in PBS containing 3% hydrogen corneal epithelium were then carefully removed from the peroxide for 30 min to remove the endogenous peroxidase stroma with forceps. activity. Following several washes in PBS and blocking in 5% normal goat serum in PBS, sections were incubated with RNA Extraction and RT-PCR either anti-CDK5 (C-8, Santa Cruz Biotechnology) or anti-p35 Corneal epithelia were harvested as described above and (C-19, Santa Cruz Biotechnology) rabbit polyclonal antibodies cytoplasmic RNA was isolated using RNAzol (TelTest, Inc., for 1 h. After extensive washing in PBS, samples were Friendswood, TX) (70). The RNA was treated with DNase I incubated for 30 min with secondary biotinylated antibodies, (amplification grade; Invitrogen Life Technologies, Inc., followed by avidin-biotinylated-peroxidase complex (ABC Kit, Carlsbad, CA), 1 unit/Ag RNA for 15 min at room temperature, Vector Laboratories, Inc., Burlingame, CA). Finally, the slides followed by heat inactivation for 10 min at 65jC. RT-PCR of were developed with Vector NovaRED and hydrogen peroxide p35 and p39 was performed according to the manufacturer’s substrate (SG) (Vector Laboratories, Inc.) according to the ma- instructions (Gene Amp RNA PCR core kit; Perkin-Elmer nufacturer’s instructions. Samples were then washed in distilled Corp., Boston, MA). A total of 1 Ag of RNA was used with the water, mounted with Aqua Poly mount (18606, Polysciences, following oligonucleotides: Inc., Warrington, PA), and viewed with a Zeiss Axioplan 2 photomicroscope. Images were captured with a CCD camera For p35: (OPELCO, Dulles, VA). For controls, the antigenic peptides Upstream: 5V-CGGCACGGTGCTGTCCCTGTCT-3V for CDK5 and p35 were included during incubation with Downstream:5V-TCACCGATCCAGGCCTAGGAG-3V primary antibodies. For p39: Upstream: 5V-GGCCGTCCGTGCTCATCTCGGCG- Immunofluorescence CTCA-3V GFP-transfected A6(1) cells were cultured on chamber Downstream: 5V-CGGCCCTTGCGGAGAAGGTTCTC- slides, fixed with 4% paraformaldehyde in PBS for 10 min at GCGGTTGCG-3V room temperature, extracted in 0.25% Triton X-100 for 5 min, and incubated for 1 h at room temperature in a blocking buffer The PCR protocol was 5 min at 95jC, followed by 35 cycles of consisting of PBS with 5% goat serum. Subsequent antibody 1 min at 95jC, 1 min at 55jC, 1 min at 72jC, and a final incubations and washes were also performed in blocking buffer. extension of 10 min at 72jC. Slides were incubated with anti-CDK5 (C-8, Santa Cruz

Biotechnology) for 1 h at room temperature, washed three times, Constructs and Stable Transfection and incubated with rhodamine-conjugated goat anti-rabbit IgG A6(1) cells were transfected with 2 Ag of pGFP, pGFP- (111-295-144, Jackson ImmunoResearch Laboratories, West CDK5, and pGFP-CDK5-T33 cDNA constructs (46) using Grove, PA). Samples were washed and mounted in Aqua Poly FuGENE 6 reagent (Roche Diagnostics) according to the mount (18606, Polysciences, Inc.), and examined with a Zeiss manufacturer’s instructions. Cells carrying the neomycin- Axioplan 2 photomicroscope equipped with epifluorescence. resistance marker were selected by addition of G418 at Images were captured with a CCD camera (OPELCO). 350 Ag/ml, 3 days after transfection. Stably transfected cells were maintained in the presence of G418 at the same Cellular Fractionation concentration. Cellular fractionation was performed as previously described (71–73) with minor modifications. In brief, mouse corneal Protein Kinase Assay epithelia were extracted on ice in 1% Triton X-100 buffer CDK5 kinase activity was determined using SignaTECT containing 10 mM imidazole, 100 mM NaCl, 1 mM MgCl ,5 2 protein kinase assay system (#V6430, Promega, Madison, mM EDTA, 0.5 mM NaF, 1 mM sodium vanadate, at pH 7.4, and 2 WI). Briefly, GFP-, GFP-CDK5-, and GFP-CDK5-T33-trans- one Complete-Mini protease inhibitor cocktail tablet/10 ml fected A6(1) cells were allowed to grow to 80% confluence, (Roche Diagnostics, Indianapolis, IN). Extracts were centri- medium was removed, and cells were treated for 10 min with fuged at 12,000 Â g for 10 min at 4jC. The Triton X-100- Ser/Thr/Tyr phosphatase inhibitor cocktail diluted 1:100 in 1Â soluble supernatant (‘‘cytoplasmic fraction’’) contains cytosolic PBS (#17-317, Upstate Biotechnology, Inc., Lake Placid, NY). proteins and weakly associated membrane proteins. The Triton Cells were then harvested and homogenized in lysis buffer: 50 X-100-insoluble pellet was washed with the 1% Triton X-100 mM Tris-HCl, pH 7.4; 250 mM NaCl, 1 mM EDTA, 0.1% buffer and further fractionated by extraction with RIPA buffer Triton X-100 containing the same phosphatase inhibitor (150 mM NaCl, 1% NP40, 0.1% sodium deoxycholate, 0.1% 2 cocktail (1:100 dilution) and one Complete-Mini protease SDS, 5 mM EDTA, and 50 mM Tris-HCl, pH 7.4) into a inhibitor cocktail tablet/10 ml (Boehringer Mannheim). CDK5 ‘‘cytoskeletal-associated fraction’’ (RIPA soluble) and a was immunoprecipitated from cell lysates as previously ‘‘membrane fraction’’ (RIPA insoluble) as follows. Pellets were 2 described (30) using CDK5 antibody (c-8, Santa Cruz solubilized in RIPA buffer, with one Complete-Mini protease Biotechnology). Following three washes in the lysis buffer, inhibitor cocktail tablet/10 ml, briefly homogenized, incubated the kinase activity of the immunoprecipitates was assayed on ice for 15 min, and centrifuged at 12,000 Â g for 10 min. using a biotinylated peptide substrate (PKTPKKAKKL) The supernatant containing the RIPA-soluble protein was (Promega) according to the SignaTECT protocol provided removed, and the Triton X-100-insoluble/RIPA-insoluble pellet by the manufacturer. was washed in RIPA buffer and solubilized in SDS-containing sample buffer, as above. Protein concentration was measured by the BCA method (Pierce Chemical Co., Rockford, IL) and Preparation of Recombinant Adenovirus and Cell equal amounts of protein from each fraction were analyzed by Infection 12% SDS-PAGE. The shuttle vectors pGEM-CMV-CDK5 and pGEM-CMV- For vinculin fractionation, the Triton X-100-insoluble pellet CDK5-T33 were constructed by inserting CDK5 or CDK5- remaining after extraction of the soluble fraction (see above) T33 cDNA into the pGEM-CMV vector, which contains a was solubilized in 9 M urea, 4% NP40 and 10 mM DTT, human cytomegalovirus immediate early promoter, enhancer, incubated at room temperature for 15 min, and centrifuged at and polyadenylate sequence (kindly provided by Dr. Terete 12,000 Â g for 10 min (74). The supernatant from this step and Borras). Adenovirus DNA was obtained from Ad-CMV-h-gal the Triton X-100-soluble fraction were immunoblotted with (also provided by Dr. Borras) by digestion with XbaI and anti-vinculin antibody (Sigma, V-4505) as described above. ClaI. Recombinant adenoviruses, Adv-CDK5 and Adv- CDK5-T33, were constructed by cotransfection of the shuttle Cell Culture vectors and adenovirus DNA into the human embryonic A6(1) corneal epithelial cells were derived from corneal kidney cell line 293 (ATCC 11573) by calcium phosphate epithelia of the 14-day-old Immorto-Mouse (Charles River precipitation (76). The recombinant adenoviral vectors were Breeding Laboratories). These cells are conditionally immor- screened for the presence of CDK5 or CDK-T33 sequences talized by a temperature-sensitive SV40 T-antigen under control and the absence of adenovirus E1A sequence, and purified of an IFN-g inducible promoter (75). Cells were cultured in three times by plaque assay in 293 cells. For large-scale Medium 500, supplemented with corneal epithelial growth purification of high-titer recombinant adenovirus, the virus supplement (Cascade Biologics, Inc., Portland, OR), IFN-g (5 was purified twice by cesium chloride density gradient units/ml), 20% fetal bovine serum, L-glutamine (60 Ag/ml), centrifugation, dialyzed for 12 h at 4jC against 10 mM Tris- penicillin (20 units/ml), and streptomycin (20 mg/ml) at the HCl, 1 mM MgCl2, 10% glycerol, pH 7.5, and stored at permissive temperature, 33jC, in a humidified atmosphere of À80jC. Adenoviral vectors were titered by plaque assay 10 95% air and 5% CO2. For in vitro migration and adhesion with 293 cells. The titer range was at between 1 Â 10 and assays, subconfluent cultures were moved to the nonpermissive 1 Â 1011 pfu/ml (77). temperature, 37jC, and cultured in the same medium in the For infection of target cells by adenovirus vectors, A6(1) absence of IFN-g. Experiments were initiated after 5 days or cells were plated at 5 Â 106 cells/60-mm Petri dishes, incubated more at 37jC. at 37jC overnight, then infected with Adv-CDK5 or Adv-

CDK5-T33 (100 pfu/cell). The cells were continuously cultured quantified by Image Pro Plus morphometric software. In at 37jC for an additional 48–72 h. The level of expression of total, three different cultures were wounded and six different CDK5 and CDK5-T33 proteins was determined by immuno- areas from those three cultures were sampled for each blotting. transfected cell types. Statistical analysis was performed using SigmaStat 2.03. Cell Adhesion Assay Adhesion assays were carried out as previously described Acknowledgments (78). Cultured cells, at 70–80% confluence, were dissociated We thank Dr. L-H. Tsai for the CDK5 and CDK5-T33 cDNA clones, Dr. J. Piatigorsky for A6(1) cells, Dr. Terete Borras for the pGEM-CMV shuttle vector using 2 mM EDTA, and the cell suspension was brought to a and Ad-CMV-hGal recombinant adenovirus, and Drs. J. Piatigorsky and M. A. 5 density of 5 Â 10 cells/ml in PBS. Flat-bottom polyvinyl Stepp for critical reading of the text. chloride micro-titer plates (BD Biosciences, Bedford, MA) were coated with 10 Ag/ml fibronectin (Invitrogen Life Technologies, Inc.), rinsed with PBS to remove unbound References 4 1. Meyerson, M., Enders, G. H., Wu, C. L., Su, L. K., Gorka, C., Nelson, C., substrate, and kept on ice. Cells (5 Â 10 cells/well) were added Harlow, E., and Tsai, L. H. A family of human cdc2-related protein kinases. to each well and the wells were filled with tissue culture EMBO J., 11: 2909 – 2917, 1992. medium without serum. The cells were then forced into contact 2. Tsai, L.-H., Takahashi, T., Caviness, V. S. J., and Harlow, E. Activity and with the substrate by centrifuging the plates for 3 min at 35 Â g expression pattern of cyclin-dependent kinase 5 in the embryonic mouse nervous system. Development, 119: 1029 – 1040, 1993. at 4jC in a low-speed centrifuge with a micro-titer plate carrier. 3. Ino, H., Ishizuka, T., Chiba, T., and Tatibana, M. Expression of Cdk5 Plates were incubated at 37jC for 2 h, inverted, and centrifuged (PSSALRE kinase), a neural cdc2-related protein kinase, in the mature and at 50, 200, or 450 Â g for 5 min to remove weakly bound cells. developing mouse central and peripheral nervous systems. Brain Res., 661: 196 – The remaining cells were stained with 0.2% crystal violet in 206, 1994. 4. Nikolic, M., Dudek, H., Kwon, Y. T., Ramos, Y. F. M., and Tsai, L.-H. The 10% ethanol. The stain was solubilized using a 50:50 mixture cdk5/p35 kinase is essential for neurite outgrowth during neuronal differentiation. of ethanol and 0.1 M NaH2PO4, pH 4.5. Absorbance was Genes Dev., 10: 816 – 825, 1996. measured at 540 nm on a microplate reader. Adhesion of trans- 5. Ohshima, T., Ward, J. M., Huh, C. G., Longenecker, G., Veeranna, Pant, H. C., fected A6(1) cells at each centrifugation force was normalized Brady, R. O., Martin, L. J., and Kulkarni, A. B. Targeted disruption of the cyclin- dependent kinase 5 gene results in abnormal corticogenesis, neuronal pathology to that of pGFP-transfected A6(1) cells at 50 Â g and results and perinatal death. Proc. Natl. Acad. Sci. USA, 93: 11173 – 11178, 1996. were expressed as relative adhesion (percentage of the control). 6. Chae, T., Kwon, Y. T., Bronson, R., Dikkes, P., Li, E., and Tsai, L. H. Mice Statistical analysis was performed using SigmaStat 2.03. lacking p35, a neuronal specific activator of Cdk5, display cortical lamination Alternatively, cell adhesion was measured using Cyto- defects, seizures, and adult lethality. Neuron, 18: 29 – 42, 1997. Matrix2 Cell Adhesion Strips (Chemicon International, Inc., 7. Ohshima, T., Gilmore, E. C., Longenecker, G., Jacobowitz, D. M., Brady, R. O., Herrup, K., and Kulkarni, A. B. Migration defects of cdk5(À/À) neurons in Temecula, CA), according to manufacturer’s recommenda- the developing cerebellum is cell autonomous. J. Neurosci., 19: 6017 – 6026, tions. Cells obtained as above were plated on microstrip 1999. precoated wells at 5 Â 104 cells/well, and incubated for 2 h. 8. Kwon, Y. T., Gupta, A., Zhou, Y., Nicolic, M., and Tsai, L.-H. Regulation of N-cadherin-mediated adhesion by the p35-Cdk5 kinase. Curr. Biol., 10: 363 – At the end of incubation, the strips were washed gently with 372, 2000. PBS to remove unattached cells. Then, the remaining bound 9. Li, B.-S., Zhang, L., Gu, J., Amin, N. D., and Pant, H. C. Integrin cells were stained, solubilized, and quantified as described a1h1-mediated activation of cyclin-dependent kinase activity is involved in above. neurite outgrowth and human neurofilament protein H lys-ser-pro tail domain phosphorylation. J. Neurosci., 20: 6055 – 6062, 2000. As negative control for both assays, cells were tested for 10. Pant, A. C., Veeranna, Pant, H. C., and Amin, N. Phosphorylation of human adhesion to wells coated with BSA (40 mg/ml). All the high-molecular-weight neurofilament protein (HNF-H) by neuronal cyclin- substrate-containing wells were also treated with BSA to block dependent kinase-5. Brain Res., 765: 259 – 266, 1997. nonspecific binding. 11. Veeranna, Shetty, K. T., Takahashi, M., Grant, P., and Pant, H. C. Cdk5 and MAPK are associated with complexes of cytoskeletal proteins in rat brain. Brain Res. Mol. Brain Res., 76: 229 – 236, 2000. In Vitro Scrape Wounding 12. Lew, J., Winkfein, R. J., Paudel, H. K., and Wang, J. H. Brain proline- directed protein kinase is a neurofilament kinase which displays high sequence A6(1) cells stably transfected with pGFP, pGFP-CDK5, homology to p34cdc2. J. Biol. Chem., 267: 25922 – 25926, 1992. and pGFP-CDK5-T33 were cultured to 70–80% confluence 13. Hisanaga, S., Uchiyama, M., Hosoi, T., Yamada, K., Honma, N., Ishiguro, at 33jC, then switched to 37jC and cultured to confluence in K., Uchida, T., Dahl, D., Ohsumi, K., and Kishimoto, T. Porcine brain the same medium in the absence of IFN-g for 5 days. A neurofilament-H tail domain kinase: its identification as cdk5/p26 complex and comparison with cdc2/cyclin B kinase. Cell Motil. Cytoskelet., 31: 283 – 297, central, 2-mm wide, linear scrape wound was made with a 1995. plastic pipette tip and the wound area was marked with three 14. Kobayashi, S., Ishiguro, K., Omori, A., Takamatsu, M., Arioka, M., Imahori, black ink dots for reference. Cultures were rinsed with PBS K., and Uchida, T. A cdc2-related kinase PSSALRE/cdk5 is homologous with the H and incubated in fresh medium for 48 h. Migration of cells 30 kDa subunit of protein kinase II, a proline-directed protein kinase associated with microtubules. FEBS Lett., 335: 171 – 175, 1993. into the wound area was monitored during this period by 15. Ishiguro, K. S., Kobayashi, S., Omore, A., Takamatsu, S., Yonekura, K., phase-contrast microscopy of the area marked by the Anzai, K., Imahori, K., and Uchida, T. Identification of the 23 kDa subunit of H reference dots. After the final phase, contrast image was protein kinase II as a putative activator of Cdk5 in bovine brain. FEBS Lett., 342: taken at 48 h, the culture dishes were fixed and stained with 203 – 208, 1994. 16. Nikolic, M., Chou, M. M., Lu, W., Mayer, B. J., and Tsai, L.-H. The p35/ hematoxylin QS to increase the visibility of cells (Vector Cdk5 kinase is a neuron-specific Rac effector that inhibits PAK1 activity. Nature, Laboratories, Inc.) and the total wound area reoccupied was 395: 194 – 198, 1998.

17. Kesavapany, S., Lau, K., McLoughlin, D., Brownlees, J., Ackerley, S., Leigh, 39. Thoft, R., Wiley, L., and Sundarraj, N. The multipotential cells of the limbus. P., Shaw, C., and Miller, C. p35/cdk5 binds and phosphorylates h-catenin and Eye, 3: 109 – 113, 1989. h regulates -catenin/presenilin-1 interaction. Eur. J. Neurosci., 13: 241 – 247, 40. Dua, H. and Azuara-Blanco, A. Limbal stem cells of the corneal epithelium. 2001. Surv. Ophthalmol., 44: 415 – 425, 2000. 18. Lew, J., Huang, Q.-Q., Qi, Z., Winkfein, R. J., Aebersold, R., Hunt, T., and 41. Lu, L., Reinach, P., and Kao, W. Corneal epithelial wound healing. Exp. Biol. Wang, J. H. A brain-specific activator of cyclin-dependent kinase 5. Nature, 371: Med., 226: 653 – 664, 2001. 423 – 426, 1994. 42. Wolosin, J., Xiong, X., Schutte, M., Stegman, Z., and Tieng, A. Stem cells 19. Tsai, L. H., Delalle, I., Caviness, V. S. J., Chae, T., and Harlow, E. p35 is a and differentiation stages in the limbo-corneal epithelium. Prog. Retinal Eye Res., neural-specific regulatory subunit of cyclin-dependent kinase 5. Nature, 371: 19: 223 – 255, 2000. 419 – 423, 1994. 43. Chaloin-Dufau, C., Sun, T., and Dhouailly, D. Appearance of the keratin pair 20. Tang, D., Yeung, J., Lee, K.-Y., Matsushita, M., Matsui, H., Tomizawa, K., K3/K12 during embryonic and adult corneal epithelial differentiation in the chick Hatase, O., and Wang, J. H. An isoform of the neuronal cyclin-dependent kinase 5 and in the rabbit. Cell Differ. Dev., 32: 97 – 108, 1990. (Cdk5) activator. J. Biol. Chem., 270: 26897 – 26903, 1995. 44. Liu, C., Zhu, G., Westerhausen-Larson, A., Converse, R., Kao, C., Sun, T., 21. Humbert, S., Dhavan, R., and Tsai, L.-H. p39 activates cdk5 in neurons, and and Kao, W. Cornea-specific expression of K12 keratin during mouse develop- is associated with the actin cytoskeleton. J. Cell Sci., 113: 975 – 983, 2000. ment. Curr. Eye Res., 12: 963 – 974, 1993. 22. Zukerberg, L. R., Patrick, G. N., Nikolic, M., Humbert, S., Wu, C.-L., Lanier, 45. Kurpakus, M., Maniaci, M., and Esco, M. Expression of keratins K12, K4 L. M., Gertier, F. B., Vidal, M., Van Etten, R. A., and Tsai, L.-H. Cables links and K14 during development of ocular surface. Curr Eye Res., 13: 805 – 814, Cdk5 and c-Abl and facilitates Cdk5 tyrosine phosphorylation, kinase 1994. upregulation, and neurite outgrowth. Neuron, 26: 633 – 646, 2000. 46. Gao, C., Negash, S., Wang, H. S., Ledee, D., Guo, H., Russell, P., and 23. Chen, F. and Studzinski, G. Cyclin-dependent kinase 5 activity enhances Zelenka, P. Cdk5 mediates changes in morphology and promotes apoptosis of monocytic phenotype and cell cycle traverse in 1,25-dihydroxyvitamin D3-treated astrocytoma cells in response to heat shock. J. Cell Sci., 114: 1145 – 1153, 2001. HL60 cells. Exp. Cell Res., 249: 422 – 428, 1999. 47. Lazaro, J. B., Kitzmann, M., Cavadore, J. C., Muller, Y., Clos, J., Fernandez, 24. Chen, F., Rao, J. and Studzinski, G. P. Specific association of increased A., and Lamb, N. J. C. Cdk5 expression and association with p35nck5a in early cyclin-dependent kinase 5 expression with monocytic lineage of differentiation of stages of rat cerebellum neurogenesis; tyrosine dephosphorylation and activation human leukemia HL60 cells. J. Leukoc. Biol., 67: 559 – 566, 2000. in post-mitotic neurons. Neurosci. Lett., 218: 21 – 24, 1996. 25. Chen, F. and Studzinski, G. Expression of the neuronal cyclin-dependent 48. Lilja, L., Yang, S. N., Webb, D. L., Juntti-Berggren, L., Berggren, P. O., and kinase 5 activator p35Nck5a in human monocytic cells is associated with Bark, C. Cyclin-dependent kinase 5 promotes insulin exocytosis. J. Biol. Chem., differentiation. Blood, 97: 3763 – 3767, 2001. 276: 34199 – 34205, 2001. 26. Lazaro, J.-B., Kitzmann, M., Poul, M.-A., Vandromme, M., Lamb, N. J. C., 49. Musa, F. R., Takenaka, I., Konishi, R., and Tokuda, M. Effects of luteinizing and Fernandez, A. Cyclin dependent kinase 5, cdk5, is a positive regulator of hormone, follicle-stimulating hormone, and epidermal growth factor on myogenesis in mouse C2 cells. J. Cell Sci., 110: 1251 – 1260, 1997. expression and kinase activity of cyclin-dependent kinase 5 in Leydig TM3 and 27. Philpott, A., Porro, E. B., Kirschner, M. W., and Tsai, L.-H. The role of Sertoli TM4 cell lines. J. Androl., 21: 392 – 402, 2000. cyclin-dependent kinase 5 and a novel regulatory subunit in regulating muscle differentiation and patterning. Genes Dev., 11: 1409 – 1421, 1997. 50. Musa, F., Tokuda, M., Kuwata, Y., Ogawa, T., Tomizawa, K., Konishi, R., Takenaka, I., and Hatase, O. Expression of cyclin-dependent kinase 5 and 28. Philpott, A., Tsai, L.-H., and Kirschner, M. W. Neuronal differentiation and associated cyclins in Leydig and Sertoli cells of the testis. J. Androl., 19: 657 – patterning in Xenopus: the role of cdk5 and a novel activator Xp35.2. Dev. Biol., 666, 1998. 207: 119 – 132, 1999. 51. Gipson, I. Adhesive mechanisms of the corneal epithelium. Acta Ophthal- 29. Fu, A., Fu, W., Cheung, J., Tsim, K., Ip, F., Wang, J., and Ip, N. Cdk5 is mol., 202 (Suppl.): 13 – 17, 1992. involved in neuregulin-induced AChR expression at the neuromuscular junction. Nat. Neurosci., 4: 374 – 381, 2001. 52. Creaven, T., Davis, J., Kim, A., and Piatigorsky, J. Conditionality immortalized corneal cell lines that undergo differentiative changes in culture. 30. Gao, C. Y., Zakeri, Z., Zhu, Y., He, H. Y., and Zelenka, P. S. Expression of Invest. Ophthalmol. Vis. Sci., 41 (Suppl.): 1369, 2000. Cdk-5, p35, and Cdk5-associated kinase activity in the developing rat lens. Dev. Genet., 20: 267 – 275, 1997. 53. Kays, W. T., Creaven, T., Kim, A., Carper, D., and Piatigorsky, J. Mouse corneal epithelial cells that differentiate in culture. Invest. Ophthalmol. Vis. Sci., 31. Hayashi, F., Matsuura, I., Kachi, S., Maeda, T., Yamamoto, M., Fujii, Y., 40 (Suppl.): 4150, 1999. Liu, H., Yamazaki, M., Usukura, J., and Yamazaki, A. Phosphorylation by cyclin- dependent protein kinase 5 of the regulatory subunit of retinal cGMP 54. Gipson, I., Watanabe, H., and Zieske, J. Corneal wound healing and phosphodiesterase. II. Its role in the turnoff of phosphodiesterase in vivo. fibronectin. Int. Ophthalmol. Clin., 33: 149 – 163, 1993. J. Biol. Chem., 275: 32958 – 32965, 2000. 55. Ding, M. and Burstein, N. Fibronectin in corneal wound healing. J. Ocul. 32. Hirooka, K., Tomizawa, K., Matsui, H., Tokuda, M., Itano, T., Hasegawa, E., Pharmacol., 4: 75 – 91, 1988. Wang, J. H., and Hatase, O. Developmental alteration of the expression and 56. Frangieh, G., Hayashi, K., Teekhasaenee, C., Wolf, G., Colvin, R., Gipson, I., kinase activity of cyclin-dependent kinase 5 (Cdk5)/p35nck5a in the rat retina. and Kenyon, K. Fibronectin and corneal epithelial wound healing in the vitamin J. Neurochem., 67: 2478 – 2483, 1996. A-deficient rat. Arch. Ophthalmol., 107: 567 – 571, 1989. 33. Gilmore, E. C., Ohshima, T., Goffinet, A. M., Kulkarni, A. B., and Herrup, K. Cyclin-dependent kinase 5-deficient mice demonstrate novel developmental 57. McClay, D., Wessel, G., and Marchase, R. Intercellular recognition: arrest in cerebral cortex. J. Neurosci., 18: 6370 – 6377, 1998. quantitation of initial binding events. Proc. Natl. Acad. Sci. USA, 78: 4975 – 4979, 1981. 34. Matsuura, I., Bondarenko, V., Maeda, T., Kachi, S., Yamazaki, M., Usukura, J., Hayashi, F., and Yamazaki, A. Phosphorylation by cyclin-dependent protein 58. Lotz, M., Burdsal, C., Erickson, H., and McClay, D. Cell adhesion to kinase 5 of the regulatory subunit of retinal cGMP phosphodiesterase. I. fibronectin and tenascin: quantitative measurements of initial binding and Identification of the kinase and its role in the turnoff of phosphodiesterase in subsequent strengthening response. J. Cell Biol., 109: 1795 – 1805, 1989. vitro. J. Biol. Chem., 275: 32950 – 32957, 2000. 59. Holly, S. P., Larson, M. K., and Parise, L. V. Multiple roles of integrins in cell 35. Patrick, G. N., Zukerberg, L., Nikolic, M., de la Monte, S., Dikkes, P., and motility. Exp. Cell Res., 261: 69 – 74, 2000. Tsai, L. H. Conversion of p35 to p25 deregulates Cdk5 activity and promotes 60. Schmitz, A. A., Govek, E. E., Bottner, B., and Van Aelst, L. Rho GTPases: neurodegeneration. Nature, 402: 615 – 622, 1999. signaling, migration, and invasion. Exp. Cell Res., 261: 1 – 12, 2000. 36. Sharma, P., Sharma, M., Amin, N. D., Albers, R. W., and Pant, H. C. 61. Palecek, S. P., Schmidt, C. E., Lauffenburger, D. A., and Horwitz, A. F. Regulation of cyclin-dependent kinase 5 catalytic activity by phosphorylation. Integrin dynamics on the tail region of migrating fibroblasts. J. Cell Sci., 109: Proc. Natl. Acad. Sci. USA, 96: 11156 – 11160, 1999. 941 – 952, 1996. 37. Patrick, G. N., Zhou, P., Kwon, Y. T., Howley, P., and Tsai, L. H. p35, the 62. Laukaitis, C. M., Webb, D. J., Donais, K., and Horwitz, A. F. Differential neuronal-specific activator of cyclin-dependent kinase 5 (Cdk5) is degraded by dynamics of a 5 integrin, paxillin, and a-actinin during formation and the ubiquitin-proteasome pathway. J. Biol. Chem., 273: 24057 – 24064, 1998. disassembly of adhesions in migrating cells. J. Cell Biol., 153: 1427 – 1440, 2001. 38. Shi, Y., Simmons, M. N., Seki, T., Oh, S. P., and Sugrue, S. P. Change in gene 63. Palecek, S. P., Huttenlocher, A., Horwitz, A. F., and Lauffenburger, D. A. expression subsequent to induction of Pnn/DRS/memA: increase in p21(cip1/ Physical and biochemical regulation of integrin release during rear detachment of waf1). Oncogene, 20 (30): 4007 – 4018, 2001. migrating cells. J. Cell Sci., 111: 929 – 940, 1998.

64. Ouyang, P. and Sugrue, S. Characterization of pinin, a novel protein 72. Walker, J. L. and Menko, A. S. a6 Integrin is regulated with lens cell associated with the desmosome-intermediate filament complex. J. Cell Biol., 135: differentiation by linkage to the cytoskeleton and isoform switching. Dev. Biol., 1027 – 1042, 1996. 210: 497 – 511, 1999. 65. Chen, H., Paradies, N., Fedor-Chaiken, M., and Brackenbury, R. E-cadherin 73. Wu, Y., Ozaki, Y., Inoue, K., Satoh, K., Ohmori, T., Yatomi, Y., and Owadab, mediates adhesion and suppresses cell motility via distinct mechanisms. J. Cell K. Differential activation and redistribution of c-Src and Fyn in platelets, assessed Sci., 110: 345 – 356, 1997. by MoAb specific for C-terminal tyrosine-dephosphorylated c-Src and Fyn. 66. Manser, E., Huang, H., Loo, T., Chen, X., Dong, J., Leung, T., and Lim, L. Biochim. Biophys. Acta., 1497: 27 – 36, 2000. Expression of constitutively active a-PAK reveals effects of the kinase on actin 74. Ito, H., Okamoto, K., Nakayama, H., Isobe, T., and Kato, K. Phosphorylation and focal complexes. Mol. Cell. Biol., 17: 1129 – 1143, 1997. of aB-crystallin in response to various types of stress. J. Biol. Chem., 272: 67. Sanders, L., Matsumura, F., Bokoch, G., and de Lanerolle, P. Inhibition of 29934 – 29941, 1997. myosin light chain kinase by p21-activated kinase. Science, 283: 2083 – 2085, 75. Jat, P., Noble, M., Ataliotis, P., Tanaka, Y., Yannoutsos, N., Larsen, L., 1999. and Kioussis, D. Direct derivation of conditionally immortal cell lines from an 68. Kiosses, W., Shattil, S., Pampori, N., and Schwartz, M. Rac recruits high- H-2Kb-tsA58 transgenic mouse. Proc. Natl. Acad. Sci. USA, 88: 5096 – 5100, affinity integrin avh3 to lamellipodia in endothelial cell migration. Nat. Cell 1991. Biol., 3: 316 – 320, 2001. 76. McGrory, W., Bautista, D., and Graham, F. A simple technique for the rescue 69. Sells, M., Boyd, J., and Chernoff, J. p21-Activated kinase 1 (PAK1) regulates of early region I mutations into infectious human adenovirus type 5. Virology, cell motility in mammalian fibroblasts. J. Cell Biol., 145: 837 – 849, 1999. 163: 614 – 617, 1988. 70. Gough, N. M. Rapid and quantitative preparation of cytoplasmic RNA from 77. Graham, F. and Prevec, L. Methods for construction of adenovirus vectors. small numbers of cells. Anal. Biochem., 173: 93 – 95, 1988. Mol. Biotechnol., 3: 207 – 220, 1995. 71. Brady-Kalnay, S., Rimm, D., and Tonks, N. Receptor protein tyrosine 78. McClay, D. R., Wessel, G. M., and Marchase, R. B. Intercellular recognition: phosphatase PTPmu associates with cadherins and catenins in vivo. J. Cell Biol., quantitation of initial binding events. Proc. Natl. Acad. Sci. USA, 78: 4975 – 130: 977 – 986, 1995. 4979, 1981.

Chun Gao, Sewite Negash, Hong Tao Guo, et al.

Mol Cancer Res 2002;1:12-24.

Updated version Access the most recent version of this article at: http://mcr.aacrjournals.org/content/1/1/12

Cited articles This article cites 75 articles, 31 of which you can access for free at: http://mcr.aacrjournals.org/content/1/1/12.full#ref-list-1

Citing articles This article has been cited by 13 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/1/1/12.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/1/1/12. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.