Neuronal Cdc2-like kinase (Nclk) binds and phosphorylates the retinoblastoma protein.

The tumor suppressor retinoblastoma protein (RB) plays a central role in cellular growth regulation, differentiation, and apoptosis. Phosphorylation of RB results in a consequent loss of its ability to inhibit cell cycle progression. However, how RB phosphorylation might be regulated in apoptotic or postmitotic cells, such as neurons, remains unclear. Here we report that neuronal Cdc2-like kinase (Nclk), composed of Cdk5 and a neuronal Cdk5 activator (p25nck5a), can bind and phosphorylate RB. Since RB has been shown recently to associate with D-type G1 cyclins and viral oncoproteins through a common peptide sequence motif of LXCXE, Nclk binding may be mediated by a related sequence motif (LXCXXE) found in p25nck5a. We demonstrate (i) in vitro binding of bacterially expressed p25nck5a to a GST-RB fusion protein, (ii) coprecipitation of GST-RB and reconstituted Cdk5·;p25nck5a, and (iii) phosphorylation of GST-RB by bacterially expressed Cdk5·;p25nck5a kinase and by Cdk5·;p25nck5a kinase purified from bovine brain. Finally, we show that immunoprecipitation of RB from embryonic mouse brain homogenate results in the coprecipitation of Cdk5 and that Cdk5 kinase activity is maximal during late embryonic development, a period when programmed cell death of developing neurons is greatest. Taken together, these results suggest that Nclk can bind to and phosphorylate RB in vitro and in vivo. We infer that Nclk may play an important role in regulating the activity of RB in the brain, including perhaps in apoptosing neurons.

The tumor suppressor retinoblastoma protein (RB) plays a central role in cellular growth regulation, differentiation, and apoptosis. Phosphorylation of RB results in a consequent loss of its ability to inhibit cell cycle progression. However, how RB phosphorylation might be regulated in apoptotic or postmitotic cells, such as neurons, remains unclear. Here we report that neuronal Cdc2-like kinase (Nclk), composed of Cdk5 and a neuronal Cdk5 activator (p25 nck5a ), can bind and phosphorylate RB. Since RB has been shown recently to associate with D-type G 1 cyclins and viral oncoproteins through a common peptide sequence motif of LXCXE, Nclk binding may be mediated by a related sequence motif (LXCXXE) found in p25 nck5a . We demonstrate (i) in vitro binding of bacterially expressed p25 nck5a to a GST-RB fusion protein, (ii) coprecipitation of GST-RB and reconstituted Cdk5⅐p25 nck5a , and (iii) phosphorylation of GST-RB by bacterially expressed Cdk5⅐p25 nck5a kinase and by Cdk5⅐p25 nck5a kinase purified from bovine brain. Finally, we show that immunoprecipitation of RB from embryonic mouse brain homogenate results in the coprecipitation of Cdk5 and that Cdk5 kinase activity is maximal during late embryonic development, a period when programmed cell death of developing neurons is greatest. Taken together, these results suggest that Nclk can bind to and phosphorylate RB in vitro and in vivo. We infer that Nclk may play an important role in regulating the activity of RB in the brain, including perhaps in apoptosing neurons.
The retinoblastoma gene product, RB, 1 initially identified as a tumor suppressor, is now also believed to play a key role in cellular processes such as cellular growth regulation, apoptosis, and differentiation. The growth regulatory function of RB was first demonstrated when loss of RB correlated with the development of retinoblastoma and several other human tumors (for review, see Refs. 1 and 2). Reintroduction of a functional RB gene into RB-deficient tumor cells results in reduced cell growth in culture and in nude mice (3,4). Interestingly, unscheduled inactivation of RB by viral oncoproteins (HPV E7 and SV40 T-antigen) in transgenic mice results in massive apoptosis in differentiating and differentiated neurons and retinal cells (5)(6)(7)(8). In addition, mouse embryos homozygous for mutated RB do not survive past day 16 of gestation as a result of aberrant terminal differentiation of erythrocytes and neurons as well as massive cell death in the central nervous system (9 -11). These results suggest that RB has an essential role in the differentiation and apoptosis of specific-cell types. However, the biochemical mechanisms underlying these various aspects of RB activity are still unclear.
Structural and functional studies reveal that RB is composed of several distinct protein-binding domains (A/B pocket, large A/B pocket, and C pocket) that, when hypophosphorylated, can bind and inhibit the activity of a variety of cellular proteins required for cell cycle progression (for review, see Ref. 2). Viral oncoproteins can displace RB-binding proteins and drive normally resting cells into S phase (Refs. 12-16; for review, see Ref. 17). The ability of RB to bind and inactivate the regulatory proteins is reduced upon its phosphorylation at various sites. During progression through the mammalian cell cycle, RB becomes hyperphosphorylated and inactivated just prior to the G 1 /S phase transition as well as during S and G 2 (for review, see Ref. 18). During later stages of mitosis RB is progressively dephosphorylated and becomes hypophosphorylated in early G 1 phase of the cell cycle. In growth arrested, senescent, or terminally differentiated cells, RB remains largely hypophosphorylated (18,19) and therefore inhibitory to mitotic progression. The kinases responsible for RB phosphorylation in vivo include members of the cyclin-dependent kinase (Cdk) family, including Cdk2 in association with cyclin E and A as well as Cdk4 and Cdk6 in association with D-type cyclins (18).
Neurons express Cdk5, which is a variant member of the Cdk family. Mammalian neuronal Cdc2-like kinase (Nclk) is composed of Cdk5 and a 25-kDa regulatory subunit (20,21). The regulatory subunit is an amino-terminal truncated derivative of a 35-kDa protein expressed specifically in neurons of the central nervous system and these and a related protein are designated as neuronal Cdk5 activators (p25 nck5a , p35 nck5a , and p39 nck5ai ) (22,23). Recent data suggest that Nclk has a role in neurite outgrowth and in regulating neurocytoskeletal dynamics (24 -26). However, the precise function of Nclk and its kinase targets have yet to be clearly defined. RB is present in neurons where Nclk contributes the major Cdc2-like kinase activity (27)(28)(29)(30)(31), and we have therefore explored the possibility that RB may be a potential substrate for Nclk. We find that purified Nclk from bovine brain and reconstituted bacterially expressed Nclk phosphorylate RB in vitro. In addition, the activator protein of Nclk (p25 nck5a ) binds directly to RB, presumably via a LXCXE-related sequence motif (LXCXXE) in p25 nck5a . Finally, Cdk5 coimmunoprecipitates with RB from embryonic mouse brain homogenate, and its kinase activity is maximal during late embryonic development (when RB phosphorylation and neuronal apoptosis are greatest), thus suggesting that the interaction between Nclk and RB may be functionally important in vivo.

EXPERIMENTAL PROCEDURES
Materials-Isopropyl-␤-D-thiogalactopyranoside was purchased from Life Technologies, Inc. [␥-32 P]ATP (4500 Ci/mmol) was obtained from ICN. Nitrocellulose membrane was obtained from Millipore. P81 phosphocellulose was purchased from Whatman. All other chemicals were purchased from Sigma.
In Vitro Histone H1 and RB Phosphorylation Assays-Histone H1 kinase activity of the reconstituted enzymes was measured by incubating the reaction mixture, containing 20 mM MOPS, pH 7.4, 30 mM MgCl 2 , 100 M [␥-32 P]ATP (1000 cpm/pmol), and 100 M histone H1 peptide P 9 KTPKKAKKL 18 , at 30°C for 20 min as described previously (22). [␥-32 P]phosphate incorporation into the substrate peptide was quantitated by liquid scintillation using a Beckman LKB 1215 scintillation counter. RB phosphorylating activity was determined using the expressed RB peptides (0.1 g/l) as substrates under the same conditions. Samples were analyzed by SDS-PAGE followed by autoradiography.
Expression and Purification of GST-fusion Proteins from Bacteria-GST-fusion proteins were purified as described (32). pGEX-2T recombinant plasmids expressing p70 N-RB (N-terminal portion of human RB, residues 1-379), p45 C-RB (C-terminal portion of human RB, residues 773-928), and p82 L-A/B-RB (large A/B pocket of human RB, residues 379 -928) were kindly provided by Drs. Harlow and Kaelin and were transformed into Escherichia coli strain DH5␣. The host cells were cultured in 4 liters of LB medium containing 100 g/ml ampicillin to A 600 nm ϭ ϳ0.7 at 37°C. The expression of GST-fusion proteins was induced with 0.4 mM isopropyl-␤-D-thiogalactopyranoside at room temperature overnight. The cells were harvested by centrifugation at 5000 ϫ g for 15 min, washed with 1 ϫ phosphate-buffered saline, suspended in 30 ml of lysis buffer (50 mM Tris-HCl, pH 7.4, 2 mM EDTA, 1 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml aprotinin, and 1 g/ml antipain) and lysed three times using a French press (Aminco) with the cell pressurized to 16,000 p.s.i. The resulting lysate was centrifuged at 18,000 ϫ g for 20 min. The supernatant was incubated for 1 h at 4°C with 4 ml of glutathione-agarose (Sigma) previously equilibrated with the same buffer. After loading the slurry onto a 10-ml column, the column was washed with 10 bed volumes of 1 ϫ phosphate-buffered saline supplemented with 0.25 M KCl, 0.1% Tween 20, 1 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml aprotinin and 1 g/ml antipain, followed by 5 bed volumes of 1 ϫ phosphate-buffered saline supplemented with 1 mM DTT. The expressed GST-fusion proteins were eluted with 5 mM reduced glutathione in 50 mM Tris-HCl, pH 8.0, 1 mM DTT.
In Vitro Binding Assays-0.3 g of GST-p25 nck5a and GST-Cdk5 were incubated with 1 unit of thrombin at room temperature for 40 min, and the digested GST portions were removed by precipitating with glutathione-agarose beads. GST-removed supernatant fractions containing either p25 nck5a alone (0.1 g) or Cdk5⅐p25 nck5a (0.1 g/each) were mixed with 1 g of GST-p82 L-A/B-RB and incubated at room temperature for 40 min. Complex formation between GST-p82 L-A/B-RB and p25 nck5a (or Cdk5⅐p25 nck5a ) was assessed by measuring the ability of p25 nck5a or Cdk5⅐p25 nck5a to bind the glutathione-agarose beads. The presence of p25 nck5a and Cdk5⅐p25 nck5a was determined by Western blot analysis with p25 nck5a -and Cdk5-specific antibodies as described below.
SDS-PAGE, Immunoblots, and Antibodies-SDS-PAGE was performed by the method of Laemmli (33) in 10 or 12.5% vertical slab gels. For Western blot analyses, samples were analyzed by SDS-PAGE followed by transfer to nitrocellulose membranes (Millipore). The blots were probed with primary antibodies (1/500 dilution of a rabbit polyclonal p25 nck5a -specific antibody (22) or 1/2000 dilution of a rabbit monoclonal Cdk5-specific (DC-17) antibody (Santa Cruz Biotechnology)) followed by incubation with a 1/2500 dilution of secondary antibodies (either goat anti-rabbit IgG or goat anti-mouse IgG conjugated with horseradish peroxidase (Pierce)) and developed by ECL reagents as specified by the manufacturer (Amersham Life Science, Inc.).
Immunoprecipitation-Embryonic and adult mice brains (E10, E12, E14, E16, E18, E20, and adult) were isolated as described previously by Drago et al. (34) and homogenized in buffer containing 20 mM HEPES, pH 7.4, 1 mM EDTA, 1 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml aprotinin, and 1 g/ml antipain. 150 g of the homogenates were precleared with 20 l of protein G Plus-agarose beads (Santa Cruz Biotechnology) and incubated with 20 l (2 g) of RB-specific antibody (IF8; Santa Cruz Biotechnology) for 2 h on a rotating platform at 4°C. The immunocomplexes were absorbed to protein G Plus-agarose beads and washed extensively with 1 ϫ Trisbuffered saline with 0.1% Tween 20. The presence of RB and Cdk5 in the immunoprecipitated samples was revealed by immunoblotting with a RB-specific (C-15) antibody and a Cdk5-specific (DC-17) antibody from Santa Cruz Biotechnology.
Protein Concentration Determination-Protein concentration was determined by Bradford microassay (35).

RESULTS AND DISCUSSION
Nclk Phosphorylates RB in Vitro-Protein kinases capable of phosphorylating retinoblastoma protein have proven to be members of the cyclin-dependent kinase (Cdk) family. Cdc2, Cdk2, Cdk4, and Cdk6 phosphorylate RB in vitro at a subset of sites that are known to be phosphorylated in vivo, and both Cdc2 and Cdk2 have been found in association with RB in vivo (36 -41). The cyclin component(s) of these RB kinases have yet to be definitively established, although transfection of cyclins A, D, and E can stimulate RB phosphorylation in vivo (42,43). In postmitotic neuronal cells, the major Cdk is Cdk5. It has been shown that cyclin D1 can bind to Cdk5, but the complex displays no demonstrable kinase activity, possibly due to the presence in these complexes of the Cdk inhibitor, p21 CIP1 (44). Recently, a highly active neuronal Cdc2-like kinase (Nclk) that is composed of Cdk5 and a 25-kDa regulatory subunit (p25 nck5a ) has been purified from postmitotic neurons of the central nervous system (20,21). Since Nclk can phosphorylate cytoskeletal and membrane-associated proteins (neurofilament proteins, Tau, and synapsin-I; 25, 26, 45) and its expression pattern and activity correlate with terminal differentiation of neurons (46), it has been suggested that Nclk has a role in regulating neurocytoskeletal dynamics and/or neuronal differentiation (21,24,47). Another prominent feature of late embryonic brain development is the apoptosis of many neurons that apparently have failed to form proper connections. Because Cdk5 kinase activity increases during brain development while Cdc2, Cdk2, and Cdk4 activities decrease (47,48), and because phosphorylation of RB is involved with both cell cycle progression (12)(13)(14)(15)(16)(17)(18)(19) and apoptosis (7,8,49,50) and because loss of RB activity can lead to enhanced neuronal death (5-11), we asked whether RB might be a substrate for Nclk.
We first established by several means that RB could serve as a substrate for Nclk in vitro. Nclk purified from bovine brain and known to exhibit high histone H1 kinase activity (21) was examined for its RB-phosphorylating activity using various truncated forms of human GST-RB proteins: GST-p70 N-RB , GST-p45 C-RB and GST-p82 L-A/B-RB . As shown in Fig. 1A, each RB peptide was capable of being phosphorylated by Nclk. To demonstrate that the RB-phosphorylating activity of bovine brain Nclk results from Cdk5⅐p25 nck5a but not from Cdk5⅐cyclin D1 complexes, we performed similar assays using bacterially expressed and reconstituted GST-Cdk5⅐GST-p25 nck5a as well as GST-Cdk5⅐GST-cyclin D1. We previously had shown that reconstitution of GST-Cdk5 and GST-p25 nck5a resulted in an active H1 kinase (Ref. 51 and Fig. 1C), and when tested with RB we found that reconstituted Nclk was also capable of efficiently phosphorylating all three RB substrates (Fig. 1B). In contrast, GST-Cdk5 complexed with GST-cyclin D1 showed negligible histone H1 kinase activity (Fig. 1C) nor was it capable of phosphorylating RB (data not shown). The addition of purified Cdk-activating kinase from bovine brain (which in separate tests activated Cdk2⅐cyclin A kinase activity 20-fold; Ref. 52) did not significantly increase Cdk5⅐cyclin D1 histone H1 kinase activity. However, when increasing concentrations of cyclin D1 were added to the mixture of Cdk5 and p25 nck5a , a substantial decrease in histone H1 kinase activity was observed. These results confirm that excess cyclin D1 can interact with Cdk5, disrupting its interaction with p25 nck5a and resulting in a kinase-inactive complex (even in the absence of added p21 CIP1 ; Ref. 44). Our results therefore suggest that in vivo purified or in vitro reconstituted Cdk5⅐p25 nck5a can phosphorylate RB and that Cdk5⅐cyclin D1 complexes are either inactive or they may target presently unidentified substrates.
Because Nclk evidently can phosphorylate nonoverlapping NH 2 -terminal and COOH-terminal RB peptide fragments in vitro (Fig. 1), we considered the distribution within the intact protein of residues that might serve as targets for the enzyme. Of the 16 Ser/Thr-Pro motifs that are potential Cdk phosphorylation sites in RB, at least eight (Ser-249, Thr-252, Thr-373, Ser-780, Ser-807, Thr-821, and Thr-826) have been shown to be phosphorylated in vivo (36,(53)(54)(55)(56), three of which are in GST-p70 N-RB , four in GST-p82 L-A/B-RB and three in GST-p45 C-RB . RB is therefore a potentially valid substrate of both in vitro reconstituted and purified in vivo Nclk, although it remains to be determined which sites are phosphorylated in vivo by this enzyme and whether their phosphorylation alters RB function. Because recent studies suggest that Ser-780 is preferentially phosphorylated by Cdk4⅐cyclin D1, but not by Cdk2⅐cyclin E/A (56), it is likely that Nclk may also display specific RB phosphorylation patterns in vivo.

Nclk Forms a Stable Complex with RB both in Vitro and in Vivo-The in vitro phosphorylation of RB by Nclk suggests that
Nclk may interact with RB in vivo. Because LXCXE and the related LXCXXE sequence motifs in D-type cyclins and several viral oncoproteins (such as adenovirus E1A protein, SV-40 T-antigen, and human papilloma virus-16 E7 protein) mediate the formation of complexes with RB (Refs. 42 and 57; for review, see Ref. 58), we searched for LXCXE and LXCXXE motifs in Nck5a. We found the LXCXXE sequence motif (L 152 RCLGE 157 ) in Nck5a, as well as related motifs (L 187 RCLGD 192 ) in p39 nck5ai and (L 372 AMGTD 377 ) in p67, another Cdk5 activator found in rat spinal cord (59). Thus, the presence of these related motifs in all the Cdk5-activating proteins raises the intriguing possibility of functional convergence for binding to RB.
Because p25 nck5a contains a potential LXCXXE binding site for RB, we next determined whether it can bind directly to RB in vitro. GST-p82 L-A/B-RB was incubated with GST-free p25 nck5a and the ability of the p25 nck5a to coprecipitate following the addition of glutathione-agarose beads was assessed. As demonstrated in Fig. 2 (lane 4), Western blot analysis with Nck5aspecific antibody showed that a significant portion of p25 nck5a precipitated with glutathione-agarose beads due to its interaction with GST-p82 L-A/B-RB (but not with GST alone; lane 6). As a further control, GST-free p25 nck5a by itself did not precipitate with the beads under the same conditions (lane 2) but remained in the supernatant fraction (lane 3). Therefore, our results suggest that p25 nck5a can bind directly to the expressed RB fragment containing the large A/B pocket. In contrast, similar in vitro binding assays with GST-free Cdk5 revealed no significant coprecipitation with GST-p82 L-A/B-RB (data not shown). Similar binding experiments were conducted with p25 nck5a and GST-p70 N-RB and GST-p45 C-RB , revealing that p25 nck5a also associates with the other fragments of RB (consistent with its ability to phosphorylate them), although the efficiency of its binding appears somewhat lower (data not shown).
We further investigated whether the Cdk5⅐p25 nck5a complex can bind stably to RB and whether this interaction facilitates Procedures." C, reconstituted GST-Cdk5⅐GST-cyclin D1 enzyme shows no histone H1 kinase activity. 0.1 g of GST-Cdk5 was reconstituted with varying amounts of GST-cyclin D1 and analyzed for histone H1 kinase activity. 5 l of partially purified Cdk-activating kinase (which can activate Cdk2⅐cyclin A substantially; Ref. 52) was added for further activation. For the p25 nck5a -cyclin D1 competition assay, 0.1 g of GST-Cdk5 was mixed with 0.1 g of GST-p25 nck5a and the indicated amount of GST-cyclin D1 and reconstituted at room temperature for 40 min .   FIG. 2. p25 nck5a forms a complex with RB in vitro in the absence of Cdk5. The GST-portion was removed from GST-p25 nck5a by thrombin digestion followed by precipitation with glutathione-agarose beads, and GST-free p25 nck5a was then incubated with 1 g of GST-p82 L-A/B-RB . Complexes recovered on glutathione-agarose beads were analyzed by 12.5% SDS-PAGE and Western blot analysis with Nck5aspecific antibody. S, supernatant; P, precipitate. RB phosphorylation. To do so, we assessed the coprecipitation of Cdk5⅐p25 nck5a with GST-p82 L-A/B-RB . We first removed the GST portion from both GST-Cdk5 and GST-p25 nck5a by thrombin digestion, followed by clearance of residual GST-containing fragments by precipitation with glutathione-agarose beads. The remaining Cdk5 and p25 nck5a polypeptides were combined, incubated with GST-p82 L-A/B-RB , and reprecipitated with glutathione-agarose beads. As shown in Fig. 3A, Western blot analyses with Cdk5-and Nck5a-specific antibodies indicated that the addition of the glutathione-agarose beads to a mixture of GST-p82 L-A/B-RB and reconstituted Cdk5⅐p25 nck5a results in precipitation of both Cdk5 and p25 nck5a , whereas precipitation was not observed in the absence of GST-p82 L-A/B-RB . In addition, when the precipitated sample was incubated with [␥-32 P]ATP and Mg 2ϩ , phosphorylation of GST-p82 L-A/B-RB was observed (Fig. 3B), consistent with the formation of a kinaseactive complex between Cdk5⅐p25 nck5a and RB.
Previous work has demonstrated the expression of RB in a variety of tissues, including brain, during development (27)(28)(29).
To determine whether the binding between Nclk and RB that we detect in vitro might also exist in vivo, we immunoprecipitated RB-containing complexes from embryonic mouse brain and tested for the coprecipitation of Nclk. As shown in Fig. 4, RB-specific antibody precipitated RB from crude homogenates of embryonic mouse brain (E18) and coprecipitated Cdk5, whereas nonspecific mouse IgG precipitated neither (the comigration of p25 nck5a with antibody light chains precluded its specific detection in this assay). Our results therefore strongly suggest the existence of Nclk⅐RB complexes in vivo.
Nclk Phosphorylation of RB Is Maximal during Late Fetal Development of the Brain-To assess the possible significance of the binding and phosphorylation of RB by Nclk, we measured Nclk kinase activity during development using RB as a substrate (Fig. 5). Brain tissues were isolated from adult mice and embryos at 10 -20 days post coitum. Nclk immunoprecipitated from these tissue homogenates was then immunoblotted with Cdk5-specific antibody (Fig. 5A), revealing a gradual increase in Cdk5 abundance as described previously (46). The same preparations were then incubated with GST-p82 L-A/B-RB , [␥-32 P]ATP, and Mg 2ϩ and phosphorylation of RB peptide assessed by autoradiography (Fig. 5B). After correction for changes in the abundance of Cdk5, we found that Nclk kinase FIG. 4. Nclk⅐RB complex exists in vivo. RB-containing complexes were immunoprecipitated from 150 g of homogenates from embryonic day 18 (E18) mouse brain with 2 g of RB-specific (IF8) antibody. The presence of RB in the immunoprecipitated samples was examined by immunoblotting with RB-specific (C-15) antibody (left panel). Coprecipitation of Cdk5 was examined by immunoblotting with Cdk5-specific (DC-17) antibody (right panel). For controls, 2 g of nonspecific mouse IgG instead of RB-specific (IF8) antibody was used for the immunoprecipitation assay. In A, Cdk5 abundance in the immunoprecipitated samples was examined by immunoblotting with Cdk5-specific (DC-17) antibody. In C, relative RB-phosphorylating activities of the samples were corrected for changes in the abundance of Cdk5 with the kinase activity in adult brain arbitrarily set at a value of 1.

FIG. 3. Nclk forms a stable kinase-active complex with RB in vitro.
A, Cdk5⅐p25 nck5a coprecipitates with GST-p82 L-A/B-RB . GST-free Cdk5 and p25 nck5a were incubated with 1 g of GST-p82 L-A/B-RB . Complexes recovered on the glutathione-agarose beads were analyzed by 12.5% SDS-PAGE and the presence of Cdk5 and p25 nck5a examined by Western blot analysis with Nck5a-specific antibody (upper panel) and Cdk5-specific (DC-17) antibody (lower panel). Digestion Cdk5 by thrombin generated two major and several minor anti-Cdk5 immunoreactive bands. While the upper bands appear to be incomplete digestion products that bind poorly to p25 nck5a and RB, the lower band migrates as expected and binds efficiently to p25 nck5a and RB. S, supernatant; P, precipitate. specific activity increased rapidly at 16 days of development, was maximal by 18 days, and declined to lower levels in adulthood (Fig. 5C). Similar results were obtained when histone H1 peptide was used as a substrate for Nclk (data not shown), and we estimate that maximal Nclk kinase specific activity at E18 was 5-fold greater than the level observed in adult brain. The down-regulation of Nclk activities for E20 and adult that we observe is clearly not due to a diminished abundance of Cdk5, which is essentially constant over this time period. Instead, it appears most likely that the decreased kinase specific activity arises either from a decrease in the abundance of an activator protein or the enhancement of an inhibitory pathway. These possibilities are currently under investigation.
It is striking that peak Nclk activity is observed at a time when neurogenesis in the mouse brain is already largely complete and apoptosis of nerve cells is most prominent. Approximately half of all neurons generated during development of the mammalian nervous system die (60) and in the cerebral cortex of the fetal mouse brain neuronal apoptosis is maximal between E14 and E18 (61), just as we see for Nclk activation. Previous work has already documented differential RB phosphorylation during embryonic mouse brain development (29), although the precise significance of this observation has been unclear. Because the inactivation of RB by viral oncoproteins can induce apoptosis in neurons (5-8) and mice homozygous for defective alleles of RB show depleted neuronal populations (9 -11), it is intriguing to speculate that highly active Nclk in late fetal development may be responsible for the hyperphosphorylation and inactivation of RB as a prelude to apoptosis in those neurons that are destined to die. It is also possible that inappropriate Nclk phosphorylation of target proteins is an important feature of changes arising during neurodegenerative diseases, where neuronal death and Nclk activation are both observed (21,46,47). 2 In summary, our results suggest that Nclk can bind to and phosphorylate RB in vitro, that it also associates with RB in vivo, and that these interactions are mediated by a direct association between p25 nck5a and RB. Furthermore, Nclk kinase activity is maximal during late fetal development of the brain, a period when neuronal apoptosis is prominent. These relationships, and the mechanisms by which Nclk kinase activity are developmentally regulated, are currently under investigation.