Cdk12 and Cdk13 Regulate Axonal Elongation Through a Common Signaling Pathway That Modulates Cdk5 Expression

Experimental Neurology 261 (2014) 10–21

Contents lists available at ScienceDirect

Experimental Neurology

journal homepage: www.elsevier.com/locate/yexnr

Regular Article Cdk12 and Cdk13 regulate axonal elongation through a common signaling pathway that modulates Cdk5 expression

Hong-Ru Chen a, Guan-Ting Lin b, Chun-Kai Huang a, Ming-Ji Fann a,c,⁎ a Department of Life Sciences and Institute of Genome Sciences, National Yang-Ming University, Taipei 11221, Taiwan, ROC b Institute of Neuroscience, National Yang-Ming University, Taipei 11221, Taiwan, ROC c Brain Research Center, National Yang-Ming University, Taipei 11221, Taiwan, ROC article info abstract

Article history: Cdk12 and Cdk13 are Cdc2-related proteins that share 92% identity in their kinase domains. Using in situ hybrid- Received 25 April 2014 ization and Western blot analysis, we detected the expression of Cdk12 and Cdk13 mRNAs and their proteins in Revised 16 June 2014 developing mouse embryos, especially during development of the nervous system. We explored the roles of Accepted 18 June 2014 Cdk12 and Cdk13 in neuronal differentiation using the P19 neuronal differentiation model. Upon knockdown Available online 3 July 2014 of Cdk12 or Cdk13, no effect on differentiated cell numbers was detected, but a substantial decrease of numbers of neurons with long neurites was identified. Similarly, knockdown of Cdk12 or Cdk13 in primarily cultured cor- Keywords: Cyclin-dependent kinase tical neurons shortens the averaged axonal length. A microarray analysis was used to examine changes in gene Axonal elongation expression after knockdown or overexpression of Cdk12 and we identified Cdk5 as a molecule potentially in- Gene expression volved in mediating the effect of Cdk12 and Cdk13. Depletion of Cdk12 or Cdk13 in P19 cells significantly reduces Cdk5 expression at both the mRNA and protein levels. Expression of Cdk5 protein in the developing mouse brain is also reduced in conditional Cdk12-knockout mice in proportion to the residual amount of Cdk12 protein present. This suggests that the reduced axonal outgrowth after knockdown of Cdk12 or Cdk13 might be due to lower Cdk5 expression. Furthermore, overexpression of Cdk5 protein in P19 cells was able to partially rescue the neurite outgrowth defect observed when Cdk12 or Cdk13 is depleted. Together, these findings suggest that Cdk12 and Cdk13 regulate axonal elongation through a common signaling pathway that modulates Cdk5 expression. © 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Introduction and Hoogenraad, 2011; Stiess and Bradke, 2011). Axons have tau-bound microtubules of uniform orientation with their plus end pointing out- Neurons are highly polarized cells with two functionally distinct do- ward, whereas dendrites have microtubule-associated protein 2 mains that emerge from the cell body; these are the single long axon, (MAP2)-bound microtubules of mixed orientation with many minus which transmits signals, and the multiple shorter dendrites, which spe- ends pointing outward (Baas et al., 1989; Burton, 1988; Conde and cialize in receiving signals (Craig and Banker, 1994). Once axon specifi- Caceres, 2009). Axonal and dendritic microtubule networks further- cation occurs, the axonal process begins to grow at a rate approximately more differ in their patterns of post-translational modifications and 10 times faster than that of the dendritic processes (Dotti et al., 1988); these result in unique interactions with other microtubule-associated this is due to differences in the patterns of microtubule organization be- proteins (Hammond et al., 2010; Witte et al., 2008). Several types of mi- tween the dendrites and the axons (Conde and Caceres, 2009; Kuijpers crotubule post-translational modifications, including tyrosination/ detyrosination and phosphorylation, are also known to modulate the sorting mechanisms of motor proteins in axons and dendrites and these consequently augment differences between axons and dendrites Abbreviations: Cdk5, cyclin-dependent kinase 5; Cdk12, cyclin-dependent kinase 12; (Fukushima et al., 2009; Sirajuddin et al., 2014). For example, Cdk13, cyclin-dependent kinase 13; CNS, central nervous system; DIV, days in vitro; K-S detyrosination modification of microtubules occurs more frequently in test, two sample Kolmogorov–Smirnov test; MAP2, microtubule-associated protein 2; axons, which facilitates the recruitment of kinesin-1 to transport SMI312, antibody against axonal phosphorylated neurofilament. collapsin response mediator protein-2 in axons (Hammond et al., ⁎ Corresponding author at: Department of Life Sciences and Institute of Genome Sciences, National Yang-Ming University, Taipei 112, Taiwan, ROC. Fax: +86 2 2823 4898. 2010; Kimura et al., 2005). The physiological relevance of microtubule E-mail address: [email protected] (M.-J. Fann). phosphorylation in axons and dendrites is another field of active

http://dx.doi.org/10.1016/j.expneurol.2014.06.024 0014-4886/© 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 11 research. It has been suggested that phosphorylation of the microtubule and we previously explored the roles of a number of uncharacterized networks is able to positively or negatively regulate the polymeriza- protein kinases. In this study we have focused on Cdk12 and Cdk13 tion of microtubules (Fourest-Lieuvin et al., 2006; Khawaja et al., due to their abundant expression in the nervous system (Lin and 1988). For example, Cdk1/cyclin B-mediated phosphorylation at Fann, 1998). They share 92% identity in the kinase domain, and are Ser172 of β-tubulin impairs the incorporation of these modiﬁed known to interact with cyclin L and cyclin K (Blazek et al., 2011; subunits into polymers (Fourest-Lieuvin et al., 2006). Similarly, Bosken et al., 2014; Chen et al., 2006, 2007; Dai et al., 2012). At the phosphorylation of β-tubulin at tyrosine residues prevents the in- molecular level, they are able to facilitate transcription of long corporation of these β-tubulin molecules into microtubules (Ley genes with many exons, and are involved in mRNA alternative splic- et al., 1994). ing (Blazek et al., 2011; Chen et al., 2006, 2007; Rodrigues et al., One of the molecules engaging in neuronal microtubule dynamics is 2012). At the cellular level, Cdk12 and Cdk13 are able to maintain Cdk5, a member of cyclin-dependent kinases (Cdk) (Smith, 2003; self-renewal in embryonic stem cells (Dai et al., 2012). In the present Veeranna et al., 2000). Cdk5 phosphorylates a number of neuronal cyto- study, we examined additional possible functions of Cdk12 and skeletal proteins involved in neuronal migration, neurite growth and Cdk13 during neuronal development. Using P19 cells as a neuronal synaptogenesis (Cheung and Ip, 2007; Cheung et al., 2006; Cruz and differentiation model, we asked whether changes in Cdk12 and Tsai, 2004; Xie et al., 2006). Furthermore, substrates of Cdk5, such as Cdk13 expression affect neuronal differentiation and neurite out- tau, MAP2, the dynein-associated protein, Nudel, and SIRT2, are growth. The P19 cells are mouse embryonic carcinoma cells and are known to be involved in cytoskeletal dynamics. This implies that Cdk5 able to differentiate into neurons when there is overexpression of plays a role as a critical signal transmission hub that is positioned be- the proneural gene Ascl1 (Farah et al., 2000). We found that reduced tween a range of cell surface receptors and various cytoskeletal systems expression of Cdk12 or Cdk13 resulted in a shortening of neurite (Pandithage et al., 2008; Smith, 2003; Speranza et al., 2013; Veeranna elongation in P19 cells and an inhibition of axonal extension in et al., 2000). primary cultured cortical neurons, but reduced Cdk12 or Cdk13 Protein kinases are critical regulators that affect the cell cycle, dif- expression had no effects on dendrite extension. We further demon- ferentiation, metabolism, cell death and various other critical cellular strated that Cdk12 and Cdk13 promote axonal elongation through a processes (Ickowicz et al., 2012; Malumbres, 2011; Yuan et al., 2013) common signaling pathway that involves modulation of Cdk5

Fig. 1. Expression of Cdk12 and Cdk13 in the developing nervous system. (A) Cdk12 and Cdk13 mRNA expression in mouse embryos at various developmental stages was analyzed by whole-mount in situ hybridization. The results from the antisense probes are shown in a–d and e–h, while those from the sense probes are shown in a′–d′ and e′–h′. Cdk12 is expressed in all cells, except in the heart (H), but with higher levels in forebrain (F), midbrain (M), primitive streak (PS) and tail bud mesoderm (TBM). A similar expression patterns is observed for Cdk13 mRNA expression (e–h). The scale bar represents 300 μm. (B, C) Expression of the Cdk12 and Cdk13 proteins in developing brains was analyzed by Western blotting. 12 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 expression at the RNA level. This is a novel ﬁnding that couples tubulin antibody (E7, 1:10000) and anti-Myc antibody (9E10, 1:5000) Cdk12 and Cdk13 functioning with Cdk5 signaling during axonal were obtained from the Developmental Studies Hybridoma Bank. elongation. Mouse anti-Flag M2 monoclonal antibody (1:1000) was purchased from Sigma. Mouse anti-Cdk5, rabbit anti-MAP2 and mouse anti-tau1 Materials and methods antibodies were obtained from Millipore. Mouse anti-pan-axonal neu- roﬁlament marker (SMI312) and mouse anti-neuronal class III β- Animals tubulin (TuJ1) were purchased from Covance. TRITC-conjugated goat anti-mouse antiserum was obtained from the Jackson Laboratory. Adult C57BL/6J mice were obtained from the National Laboratory Other secondary antisera were purchased from Chemicon or Bethyl Animal Center (Taipei, Taiwan). All mice were handled according to Laboratories. the university guidelines and all experiments were approved by the National Yang-Ming University Animal Care and Use Committee. For Cell culture and transfection the timed pregnancies, mice were set up in the late afternoon and when plugs were detected the next morning, this was designated P19 cells were cultured in α-minimal essential medium (α-MEM) as E0.5. with 10% fetal bovine serum (FBS), 100 U/ml penicillin, 100 μg/ml

streptomycin, and 2 mM glutamine and incubated in a 5% CO2/95% hu- Whole-mount in situ hybridization midified atmosphere at 37 °C. For neuronal differentiation, P19 cells were transfected with pAscl1 and other plasmids using Lipofectamine Mouse embryos were dissected at different developmental stages, 2000 (Invitrogen) according to the instructions of the manufacturer. fixed in 4% paraformaldehyde, and permeabilized by treatment with The expression plasmid pAscl1 (pUS2-Ascl1) from D. L. Turner was proteinase K. After hybridization and washing, embryos were incubated used for neuronal differentiation, which has previously been shown to with an alkaline phosphatase-conjugated anti-digoxygenin antibody convert P19 cells into a relatively homogenous population of electro- (Roche), washed again and then stained using a NBT/BCIP mixture physiologically differentiated neurons (Farah et al., 2000; Yu et al., (Roche) in the presence of 10% polyvinyl alcohol (31–50 kD; Sigma). 2002). Eight hours after transfection, 8 μg/ml puromycin was applied After visualization of the alkaline phosphatase activity, embryos were to select transfected cell. The cells were then replated onto 6-well plates post-fixed and cleared in 100% glycerol for photography. Sense and an- coated with 2 μg/ml laminin (Invitrogen) at 24 h after transfection, and tisense RNA probes were synthesized using an in vitro transcription kit cultured in 1% FBS OPTI-MEM. The medium was changed at two days from Roche, and were purified using a Sephadex G-50 column after transfection. Cells were harvested for immunocytochemistry or (Amersham Biosciences). RNA inhibitor (RNase) was purchased from protein extraction at three days after transfection. Mouse cortical neu- Promega. rons were prepared from E14.5 mouse embryos. The cortex was me- chanically dissociated by passing through a glass pipette for twenty fi μ fi Vector construction times and then ltered through a 70- m nylon mesh lter (BD Biosci- ences). Cells were plated onto plastic culture plates coated with μ μ The various Cdk12 and Cdk13 plasmids had been constructed pre- 30 g/ml poly-L-lysine and 2 g/ml laminin and cultured in Dulbecco's fi viously (Chen et al., 2006, 2007). Protocols described by Chung et al. modi ed Eagle's medium containing 10% FBS, 100 U/ml penicillin and (2006) were followed to construct the microRNA mimics. Oligonu- streptomycin, and N2 supplement and B27 supplement (Invitrogen). fi cleotides against mouse Cdk5, Cdk12 and Cdk13 sequencesaswell The cells were kept at 37 °C in a humidi ed atmosphere of 5% CO2 as against their equivalent scramble sequences were synthesized and 95% air. To deliver plasmids into cortical neurons, electroporation and cloned into the vector UI4-puro-SIBR to obtain pCdk5-s, was performed before plating using the Amaxa Mouse Neuron pCdk5-i1, pCdk5-i2, pCdk12-s, pCdk12-i1, pCdk12-i2, pCdk13-s, Nucleofector kit and the Amaxa Nucleofecor II (Lonza). Conditions 7 μ pCdk13-i1 and pCdk13-i2, or into the vector UI4-GFP-SIBR to obtain were optimized for 4 × 10 cells and 10 g of plasmid per electropora- fi pCdk12-s-e, pCdk12-i1-e, pCdk13-s-e and pCdk13-i1-e. The se- tion, which gave a survival and transfection ef ciency of 60% and 35% quences of the oligonucleotides are listed in Table S1. Mouse Cdk5 to 40%, respectively. activator, p35, was amplified by PCR and cloned into the vectors pUS2. All constructs were verified by restriction enzyme mapping Immunofluorescence microscopy and neurite counting and sequencing. Cultured neurons were rinsed once with PBS and fixed in 4% parafor- Antibodies maldehyde for 1 h. Samples were stained with the appropriate primary and secondary antibodies and then finally stained with DAPI (1 μg/ml) Rabbit anti-Cdk12 and mouse anti-Cdk13 polyclonal antibodies had for 15 min. Samples were then examined using a fluorescence micro- been produced previously (Chen et al., 2006, 2007). Mouse anti-β- scope (Zeiss). Numbers of TuJ1-positive cells and DAPI-positive cells

Fig. 2. Knockdown of Cdk12 and Cdk13 decreases neurite elongation in P19 cells. (A) A schematic depiction of the experimental procedure. (B) Representative images of P19 cells transfected with the indicated Cdk12 constructs. (C) Differentiation rates of P19 cells transfected with the indicated constructs. Differentiated neuronal cells were detected by TuJ1 staining. The number of TuJ1-positive cells was divided by the number of DAPI positive cells in 30 randomly selected fields. (D) Average neurite length of P19 cells transfected with the indicated constructs. The lengths of neurites marked by TuJ1 staining were manually quantified for each selected neuron using Neuron J software. In each experiment, 300 neurons from randomly selected fields for each transfected condition were evaluated. Data from three independent experiments were analyzed by Student's t-test and are presented as mean ± SEM. **, p b 0.01; ***, p b 0.001. (E) The distribution of neurite lengths using P19 cells transfected with the indicated constructs. Data from three independent experiments are presented as mean ± SEM. Statistical analysis using two sample Kolmogorov–Smirnov test (K–S test) shows p N 0.05 for pCdk12 when compared to pEF, and p N 0.05, p b 0.01, and p b 0.001 for pCdk12-s, pCdk12-i1, and pCdk12-i2, respectively, when compared to pUI4. (F) Representative images of P19 cells transfected with the indicated Cdk13 constructs. (G) Differentiation rates of P19 cells transfected with the indicated constructs. (H) Average neurite length of P19 cells transfected with the indicated constructs. Data from three independent experiments are presented as mean ± SEM. ***, p b 0.001 by Student's t-test. (I) Distribution of neurite lengths of P19 cells transfected with the indicated constructs. Data from three independent experiments are presented as mean ± SEM. The two sample K–S test shows p N 0.05 for pCdk13 when compared to pEF, and p N 0.05, p b 0.001, and p b 0.001 for pCdk13-s, pCdk13-i1, and pCdk13-i2, respectively, when compared to pUI4. Scale bars 100 μm. H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 13 were counted using the Image J plugin, NeurphologyJ (Ho et al., 2011). neurites and cortical neurons neurites was performed manually, and The differentiation rate was calculated by dividing TuJ1-positive cell the length of each neurite was quantified using the Image J plugin, numbers by DAPI-positive cell numbers. The tracking of P19 cell ImageJ. In each experiment, 300 neurons from randomly selected fields 14 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 15 for each condition were evaluated. All experiments were repeated at expression of Cdk12 and Cdk13 at the protein level in the develop- least three times. ing brain at later developmental stages and found that Cdk12 and Cdk13 proteins were consistently expressed in the brain at all em- Targeted disruption of the mouse Cdk12 gene bryonic stages (Figs. 1B and C), which suggests that Cdk12 and Cdk13 may play a general role in the development of the nervous To generate Cdk12 conditional knockout mice, a targeting vector system. containing Cdk12 sequences between exon 3 and exon 4 flanked by fl loxP, and FRT- anked Neo cassette was prepared from a BAC clone Knockdown of Cdk12 or Cdk13 leads to an axonal extension defect through using the recombineering approach (Copeland et al., 2001; Liu a common signaling pathway et al., 2003). The targeting vector was linearized and electroporated into embryonic stem (ES) cells, which was performed by Transgenic Initially, we examined whether Cdk12 and Cdk13 participate in Mouse Models Core Facility, Taiwan. After selection with antibiotics, neuronal differentiation using the P19 neuronal differentiation fi cell clones with correct homologous recombination were identi ed model. P19 is a mouse embryonal carcinoma cell line that is able by Southern blot hybridization. After Neo pop-out, a correctly to be induced to differentiate into neuronal cells by ectopic expres- targeted ES clone was injected into blastocysts and the resultant chi- sion of proneural gene Ascl1 and gives rise to unipolar or bipolar meras were bred with C57BL/6J mice. Genotyping was performed by neurites. Overexpression constructs that increase Cdk12 or Cdk13 Southern blotting and/or PCR using the probe and primers listed in protein expression more than nine folds in P19 cells, and two RNA Table S1. To conditionally delete Cdk12 in the nervous system, interference constructs for each gene that decrease Cdk12 or T2 NestinCreER mice were used for Cre-mediated recombination Cdk13 protein expression to less than 40% of the control level after injecting Tamoxifen. Most mice used for this study were on a were constructed (Figs. S1A and C). The Cdk12 interference C57BL/6-129Sv mixed background. constructs do not inhibit Cdk13 protein expression, and vice versa (Figs. S1B and D). Differentiated P19 cells exhibit unipolar or bipolar Protein extraction and Western blotting neurites whose lengths can be revealed by staining with TuJ1 antibody, which recognizes a neuronal specific beta-III-tubulin Previously described procedures were followed during these exper- expressed in early postmitotic neurons. Overexpression of either iments (Wong et al., 2010). Cdk12 or Cdk13 in P19 cells was found not to affect differentiation of P19 cells cultured for 3 days after transfection, either in terms of Quantitative real-time PCR assay the numbers of differentiated TuJ1-positive cells or in terms of the average length of outgrown neurites (Figs. 2Bb,C,D,Fb,G,andH). Total RNA was extracted using a RNeasy Mini kit (Qiagen) follow- Surprisingly, however, knockdown of Cdk12 or Cdk13 decreased ing the manufacturer's suggestions. Reverse transcription was per- the average lengths of neurites in differentiated P19 neurons formed with 1.5 μg of total RNA using SuperScript III reverse (312.66 ± 0.34 μm, mean ± SEM, for the pUI4 control, 227.17 ± transcriptase (Invitrogen). The real-time quantitative PCR was per- 7.27 μm for pCdk12-i1, Fig. 2D; and 306.95 ± 4.3 μm for pUI4 control, formed on a LightCycler 480 using the Universal Probe Library 202.52 ± 5.75 μm for pCdk13-i1, Fig. 2H), although there was no (UPL) (Roche). A web service (http://qpcr.probefinder.com/roche3. change in the numbers of differentiated cells (Figs. 2CandG).De- html) was used to select targeted input sequences, UPL probes, and tailed analysis of the neurite length distribution revealed that cells primer pairs. The sequences of the primers are listed in Table S1. with a deficiency in either Cdk12 or Cdk13 were able to send out pro- Negative controls containing no cDNA template were included in cesses but had reduced capability to extend these processes to long each experiment. The results were analyzed by LightCycler software distances (N300 μm) (Figs. 2E and I). Thus, distribution of the neurite version 3.5 (Roche) to determine the threshold cycle (Cp) for each length was skewed to the right in Cdk12-depleted and Cdk13- reaction. depleted cells, suggesting that Cdk12 and Cdk13 are critical for neurite elongation in P19 cells. Results Next, we analyzed whether Cdk12 and Cdk13 affect axonal elongation through a common pathway. Interestingly, overexpression of Expression of Cdk12 and Cdk13 in the developing nervous system Cdk13 in Cdk12-depleted cells reduced the percentage of short neurite (100–200 μm) from 38% to 27%, and increased the percent- We used whole-mount in situ hybridization and Western blotting to age of longer neurite (300–400 μm) from 16% to 23% (Figs. 3Band detect the spatial and temporal expression patterns of Cdk12 and F). Overexpression of Cdk12 in Cdk13-depleted cells was also able Cdk13 at both the mRNA and protein levels. Cdk12 mRNA is present to decrease the percentage of short neurite (100–200 μm) from in E6.5 embryo proper and trophoblasts (Fig. 1Aa). During E7–E7.5, 44% to 33%, and to raise percentage of the longer neurite (300– Cdk12 is expressed in all cells, but with stronger expression in the 400 μm) from 8% to 13% (Figs. 3D and H). Furthermore, the average primitive streak (PS) (Fig. 1Ab). In E8.5 and E9.5 embryos, stronger neurite lengths after overexpression approached those of the control signals are detected in the forebrain, midbrain and tail bud meso- cells, when compared to those of the Cdk12-depleted or Cdk13- derm (TBM), but not in the heart (Figs. 1Ac and Ad). Cdk13 is also depleted P19 cells (Figs. 3E and G). These results suggest that the ubiquitously expressed but not in the heart at the various devel- same underlying signal pathway, at least in part, may mediate action opmental stages examined (Figs. 1Ae–Ah). We further analyzed of both Cdk12 and Cdk13 (Fig. 3).

Fig. 3. Cdk12 and Cdk13 regulate neurite elongation through a common signaling pathway in differentiated P19 cells. (A, C) The expression levels of Cdk12, Cdk13, and Cdk5 proteins in P19 cells transfected with the indicated constructed. Ectopically expressed Cdk12 and Cdk13 were detected by anti-myc antibody. (B, D) Representative images of P19 cells transfected with the indicated constructs are shown. (E, F) The average neurite lengths and the neurite length distribution of P19 cells transfected with the indicated Cdk13 and Cdk12 constructs are shown. *, p b 0.05. There is a signiﬁcant difference between the pCdk12-i1 and pCdk12-i1 + pCdk13 groups by the two sample K–Stest(p b 0.01). (G, H) The average neurite lengths and neurite length distribution in P19 cells transfected with the indicated Cdk12 and Cdk13 constructs are shown. *, p b 0.05. There is a signiﬁcant difference between pCdk13-i1 and pCdk13-i1 + pCdk12 groups by the two sample K–Stest(p b 0.001). Scale bars 100 μm. 16 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 17

As Cdk12 and Cdk13 proteins are detected in mouse E14.5 cortical Tamoxifen at E10.5 into pregnant mice. Whole brain was then dis- neurons (Fig. S2), we examined whether Cdk12 and Cdk13 are sected from the E14.5 embryos and protein extracts from different able to modulate neurite extension in cultured cortical neurons. embryos were subjected to Western analysis using anti-Cdk12 anti- Dissociated cortical neurons derived from mouse E14.5 embryos bodies and anti-Cdk5 antibodies (Fig. 5D). A linear correlation be- were electroporated with the various constructs. We used anti- tween Cdk12 expression and Cdk5 expression with a value for r of MAP2 antibody to demarkate dendrites, and anti-phosphorylated 0.98 was found (Fig. 5E). neurofilament antibody (SMI312) and anti-tau1 antibody to iden- tify axons. After 3 days of culture in vitro (DIV), depletion of either Overexpression of Cdk5 partially rescues effects of Cdk12 or Cdk13 Cdk12 or Cdk13 did not affect the neuronal polarity in cultured cor- knockdown tical neurons (Fig. S3). These neurons exhibited several MAP2- positive processes, and one long SMI312-positive process. In To examine whether Cdk5 is the downstream gene that is con- Cdk12-depleted and Cdk13-depleted cortical neurons, the average trolled by Cdk12 and Cdk13 in P19 cells, we performed RNA inter- μ length of dendrites was not affected (12.67 ± 1.19 m in control ference experiments to knockdown Cdk5 mRNA and analyzed the μ μ cells vs. 12.33 ± 1.44 m and 12.36 ± 0.72 minCdk12-depleted effect on neurite extension in P19 cell model. Transfection of two and Cdk13-depleted neurons, respectively, Figs. 4A, B); however, RNA interference constructs (pCdk5-i1 and pCdk5-i2) was found μ the average length of axons was decreased (101 ± 5.25 m in control to reduce Cdk5 protein expression to 50% of the control (Fig. S1E). μ μ cells vs. 42 ± 3.3 mand41±2.05 minCdk12-depletedand Upon knockdown of Cdk5, the number of differentiated P19 cells Cdk13-depleted neurons, respectively, Figs. 4C, D). At 5 DIV, den- does not change (Fig. 5G), but it was found that the neurite length drites and axons of the control neurons had grown further distribution was skewed to the right in a pattern that is similar to – (Figs. 4E H). Dendrite growth had taken place in the Cdk12- what we observed for Cdk12-depleted or Cdk13-depleted cells depleted and Cdk13-depleted neurons (compare Figs. 4FwithB); (Figs. 5H, I). It should be noted that the decrease in average neurite however, the growth of the axons of Cdk12-depleted and Cdk13- length and change in skewness after knockdown of Cdk5 were depleted neurons was arrested during these two days (compare found to be less than the changes induced by Cdk12 and Cdk13 Figs. 4H with D). Staining with tau1 gave similar results at 3 DIV depletion. fi (Fig.S4).Together,these ndings support the idea that Cdk12 and We next examine whether overexpression of Cdk5 was able to res- Cdk13 are essential for axonal extension. cue the defects found in Cdk12-depleted and Cdk13-depleted P19 cells. When Cdk5 was ectopically expressed in Cdk12-depleted and Expression of Cdk5 is regulated by Cdk12 and Cdk13 Cdk13-depleted P19 cells with Cdk5 activator, p35, the neurites grew to a longer length; furthermore, the average neurite length and the dis- To dissect the underlying mechanisms by which Cdk12 and tribution of the neurite length approached that of the control cells, Cdk13 regulate axonal elongation, we performed microarray analy- when compared to the Cdk12-depleted or Cdk13-depleted P19 cells sis and compared the global gene expression profiles of the control, (Fig. 6). Thus, the defect of neurite extension due to knockdown of Cdk12-overexpressing and Cdk12-knockdown P19 cells. Using Cdk12 or Cdk13 was found to be partially rescued by Cdk5 overexpres- these findings, we then selected candidate genes whose expression sion. However, overexpression of Cdk12 or Cdk13 in Cdk5-depleted P19 was decreased in the Cdk12-knockdown P19 cells but did not change cells had no effect on neurite outgrowth in comparison to knockdown of fi in the Cdk12-overexpressing P19 cells, both in comparison to cells Cdk5 alone (Fig. S6). These ndings suggest that the Cdk5 is one of the transfected with the control vector. Among these candidates identi- downstream mediators for Cdk12 and Cdk13 when axonal elongation fied, Cdk5 mRNA was found to be down-regulated in Cdk12- occurs. knockdown P19 cells to 27% of the control using the microarray analysis. As Cdk5 is an important molecule in terms of neurite outgrowth Discussion during neural development, we then investigated whether Cdk12 and Cdk13 were indeed able to regulate Cdk5 mRNA expression Cdk12 and Cdk13 are essential for axonal elongation using quantitative RT-PCR analysis. Knockdown of either Cdk12 or Cdk13 was found to reduce Cdk5 mRNA level to ~40% of the control Our investigation provides evidence that Cdk12 and Cdk13 contrib- level (Fig. 5A), which is a level compatible with the decreased levels ute to axonal growth. Upon depletion of Cdk12 or Cdk13, cultured P19 of Cdk12 or Cdk13 protein when their corresponding RNA interfer- cells and primary cortical neurons lose the ability to extend their ence constructs are transfected into P19 cells. Cdk5 protein expres- axons to greater distances, although they are still able to initiate sion is also reduced to ~50% of the control (Fig. 5B, C), which neurites (Figs. 2 and 4). Overexpression of Cdk13 is able to partially res- suggests that regulation of Cdk5 by Cdk12 and Cdk13 is primarily cue the neurite outgrowth defect in Cdk12-depleted P19 cells, and vice at the RNA level. versa, which suggests that they act through common downstream Previously we have generated a Cdk12 conditional knockout effector(s) to promote axonal elongation (Fig. 3). One protein affected mouse by inserting loxP sites into intron 2 and intron 4 (Fig. S5A). by both Cdk12 and Cdk13 is Cdk5, and both are able to modulate Cdk5 We took advantage of variable efficiency of the Cre recombinase in protein expression in vitro; furthermore, overexpression of Cdk5 in embryos after injecting Tamoxifen into pregnant mice, and asked Cdk12-depleted and Cdk13-depleted P19 cells is able to partially rescue whether there is a dose–response relationship between the in vivo the neurite outgrowth defect (Fig. 6). Interestingly, overexpression of expression of Cdk12 and the expression of Cdk5. We crossed Cdk13 increases the endogenous Cdk5 protein expression level and is Cdk12f/f mice with Nestin-CreERT2; Cdk12del/+mice, and injected able to rescue partially the defects in Cdk12-depleted P19 cells, also

Fig. 4. Cdk12 and Cdk13 enhance axonal elongation in primary cultured cortical neurons. Cortical neurons from the E14.5 mouse cortex were transfected with the indicated constructs. The cells were cultured, fixed, and immunostained with anti-MAP2 (A, E) or anti-SMI312 (C, G) antibodies at 3 DIV (A–D) and 5 DIV (E–H). The arrows indicate MAP2-positive and SMI312- positive neurites. (B, D, F, H) Quantification of the length of MAP2-positive dendrites and the SMI312-positive axons as shown in A, C, E, and G was performed. For each condition, the data were collected from 300 randomly selected cells. Data from three independent experiments were analyzed by Student's t-test and are presented as mean ± SEM. **, p b 0.01; ***, p b 0.001. Scale bars 50 μm. 18 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 19 showing causality relationship. Overexpression of Cdk12 in Cdk13- Cdk13 are able to regulate the axonal specific microtubule organiza- depleted P19 cells gives the same results (Figs. 3A, C). The findings tion but not dendritic cytoskeletal organization, although expression from Cdk12 knockout mice further confirm that expression of Cdk5 is of Cdk12 and Cdk13 is distributed evenly across both dendrites and indeed regulated by the level of Cdk12 proteins in vivo. However, the axons (data not shown). Currently, the underlying mechanisms by observation that Cdk5 overexpression in Cdk12-depleted and Cdk13- which Cdk12 and Cdk13 are able to generate this differential effect depleted P19 cells brings about only a partial rescue of the neurite remain unclear. It seems likely that Cdk12 and Cdk13 regulate tubu- elongation defect, which implies that there are other effectors in- lin polymerization and microtubule stability, which is highly critical volved in Cdk12-dependent and Cdk13-dependent axonal elonga- in axons. Silencing of Cdk12 and/or Cdk13 may also bring abnormal tion (Fig. 7). microtubule remodeling and a change in microtubule dynamics in axons. Details of the downstream Cdk12 and Cdk13 substrate proteins and how they contribute to this differential effect on axons Cdk12 and Cdk13 have overlapping, but not identical, functions and dendrites remain obscure. Future investigation in this area will further strengthen our understanding of the development of axons Cdk12 and Cdk13 have similar protein sequences and structural and dendrites. motifs. They are known to be engaged in a number of similar biolog- Supplementary data to this article can be found online at http://dx. ical processes (Chen et al., 2006, 2007; Dai et al., 2012; Kohoutek and doi.org/10.1016/j.expneurol.2014.06.024. Blazek, 2012). For example, both Cdk12 and Cdk13 play an indis- pensable role in maintaining the pluripotency of mouse embryonic stem cells (Dai et al., 2012). Furthermore, both Cdk12 and Cdk13 Author contribution take part in RNA splicing regulation in a similar manner (Chen et al., 2006, 2007). In this study, we have shown that their expression H.-R. Chen, G.-T. Ling, and M.-J. Fann conceived and designed the ex- patterns seem also to be very similar (Fig. 1). Expression of Cdk12 periments. H.-R. Chen, G.-T. Ling, and C.-K. Huang conducted experi- and Cdk13 is initiated quite early during development, beginning ments. H.-R. Chen and M.-J. Fann wrote the manuscript. as early as the E6.5 mouse embryo. Both are expressed ubiquitously, except that they are both absent from the developing heart. This sim- Conflicts of interest ilarity in expression pattern and function between Cdk12 and Cdk13 fi suggests that de ciency in either protein is unlikely to generate an Authors declare no conflict of interest. obvious phenotype in mouse embryos. It is thus surprising that Cdk12 knockout mice die at post-implantation stage at E5.5 (data Acknowledgments not shown) and this suggests that Cdk13 is not a redundant version of Cdk12 and is not able to completely complement Cdk12 deficiency. We thank L.-S. Kao, J.-Y. Yu, Y.-C. Chern, E. Hwang, and M.-m. Poo for Furthermore, it has been shown that Cdk12, rather than Cdk13, acti- their discussion and critical reading of the manuscript, E. Hwang is vates transcription via the phosphorylation of Ser2 at the carboxyl- thanked for providing NeurphologyJ software. We would also like to terminal domain of RNAPII and this then facilitates transcription of thank the Developmental Studies Hybridoma Bank for providing mono- long exons during elongation (Bartkowiak et al., 2010; Blazek et al., clonal antibodies. This research was supported by the National Science 2011; Cheng et al., 2012; Ghamari et al., 2013). It is likely that Council (NSC101-2320-B-010-064) and the Ministry of Education these two genes are derived from an ancestor gene but have then (Aim for the Top University Plan) to M.-J. Fann. evolved divergently and are now an evolutionary distinct far part in vertebrates (Malumbres et al., 2009). This may have resulted in subtle differences in terms of cellular localization as well as changes References in their preference in terms of substrates and cyclins. A more thor- Baas, P.W., Black, M.M., Banker, G.A., 1989. Changes in microtubule polarity orientation ough study is needed in the future in order to explore these fascinat- during the development of hippocampal neurons in culture. J. Cell Biol. 109, ing possibilities. 3085–3094. Bartkowiak, B., Liu, P., Phatnani, H.P., Fuda, N.J., Cooper, J.J., Price, D.H., Adelman, K., Lis, J.T., Greenleaf, A.L., 2010. CDK12 is a transcription elongation-associated CTD kinase, the Axons, but not dendrites, are affected by the activity of Cdk12 and Cdk13 metazoan ortholog of yeast Ctk1. Genes Dev. 24, 2303–2316. Blazek, D., Kohoutek, J., Bartholomeeusen, K., Johansen, E., Hulinkova, P., Luo, Z., Cimermancic, P., Ule, J., Peterlin, B.M., 2011. The cyclin K/Cdk12 complex maintains Axonal outgrowth involves many steps. The roles of Cdk12 and genomic stability via regulation of expression of DNA damage response genes. Cdk13 seem to be limited to axonal elongation. They are not required Genes Dev. 25, 2158–2172. for specification and maintenance of neuronal polarity, since knock- Bosken, C.A., Farnung, L., Hintermair, C., Merzel Schachter, M., Vogel-Bachmayr, K., Blazek, D., Anand, K., Fisher, R.P., Eick, D., Geyer, M., 2014. The structure and substrate spec- down of Cdk12 and Cdk13 does not affect the numbers of SMI312- ificity of human Cdk12/Cyclin K. Nat. Commun. 5, 3505. positive cortical neurons or MAP2-positive cortical neurons at 3 Burton, P.R., 1988. Dendrites of mitral cell neurons contain microtubules of opposite po- DIV and 5 DIV (Figs. 4 and S4). However, knockdown of Cdk12 and larity. Brain Res. 473, 107–115. Chen, H.H., Wang, Y.C., Fann, M.J., 2006. Identification and characterization of the CDK12/ Cdk13 dose affect axonal elongation and this effect occurs without cyclin L1 complex involved in alternative splicing regulation. Mol. Cell. Biol. 26, affecting dendrite extension. This suggests that both Cdk12 and 2736–2745.

Fig. 5. Depletion of Cdk12 or Cdk13 leads to down-regulation of Cdk5 and depletion of Cdk5 causes a neurite elongation defect in P19 cells. (A) Quantitative PCR was performed to measure Cdk5 expression in P19 cells transfected with various Cdk12 or Cdk13 constructs. Data from three independent experiments were analyzed by Student's t-test and are presented as mean ± SEM. **, p b 0.01. (B) Knockdown of Cdk12 or Cdk13 in P19 cells reduces Cdk5 protein expression as revealed by Western blotting. (C) Quantiﬁcation of (B). The data from three independent experiments were analyzed by Student's t-test and are presented as mean ± SEM. **, p b 0.01. (D) Western blot analysis of Cdk12, Cdk13, and Cdk5 protein expression in embryonic brains with different Cdk12 knockout backgrounds. (E) The amount of Cdk5 protein present was plotted against the amount of Cdk12 proteins present using mouse embryos that had different genotypes and different levels of Cre activities. Expression levels of Cdk12 and Cdk5 in a randomly selected Cdk12f/+mouse (marked with *) were deﬁned as having an arbitrary unit (A.U.) value of one. (F) Representative images of P19 cells transfected with the indicated Cdk5 constructs. (G) Differentiation rates of P19 cells transfected with the indicated constructs. (H) The average neurite lengths of P19 cells transfected with the indicated constructs. Data from three independent experiments are presented as mean ± SEM. **, p b 0.01; ***, p b 0.001 by Student's t-test. (I) The distribution of neurite lengths of P19 cells transfected with the indicated constructs. Data from three independent experiments are presented as mean ± SEM. The two sample K–Stestshowsp N 0.05, p b 0.05, and p b 0.05 for pCdk5-s, pCdk5-i1, and pCdk5-i2, respectively, when compared to pUI4. Scale bars 100 μm. 20 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 H.-R. Chen et al. / Experimental Neurology 261 (2014) 10–21 21

Ghamari, A., van de Corput, M.P., Thongjuea, S., van Cappellen, W.A., van Ijcken, W., van Haren, J., Soler, E., Eick, D., Lenhard, B., Grosveld, F.G., 2013. In vivo live imaging of RNA polymerase II transcription factories in primary cells. Genes Dev. 27, 767–777. Hammond, J.W., Huang, C.F., Kaech, S., Jacobson, C., Banker, G., Verhey, K.J., 2010. Post- translational modifications of tubulin and the polarized transport of kinesin-1 in neurons. Mol. Biol. Cell 21, 572–583. Ho, S.Y., Chao, C.Y., Huang, H.L., Chiu, T.W., Charoenkwan, P., Hwang, E., 2011. NeurphologyJ: an automatic neuronal morphology quantification method and its application in pharmacological discovery. BMC Bioinforma. 12, 230. Ickowicz, D., Finkelstein, M., Breitbart, H., 2012. Mechanism of sperm capacitation and the acrosome reaction: role of protein kinases. Asian J. Androl. 14, 816–821. Khawaja, S., Gundersen, G.G., Bulinski, J.C., 1988. Enhanced stability of microtubules enriched in detyrosinated tubulin is not a direct function of detyrosination level. J. Cell Biol. 106, 141–149. Kimura, T., Watanabe, H., Iwamatsu, A., Kaibuchi, K., 2005. Tubulin and CRMP-2 complex is transported via Kinesin-1. J. Neurochem. 93, 1371–1382. Kohoutek, J., Blazek, D., 2012. Cyclin K goes with Cdk12 and Cdk13. Cell Div. 7, 12. Kuijpers, M., Hoogenraad, C.C., 2011. Centrosomes, microtubules and neuronal development. Mol. Cell. Neurosci. 48, 349–358. Ley, S.C., Verbi, W., Pappin, D.J., Druker, B., Davies, A.A., Crumpton, M.J., 1994. Tyrosine phosphorylation of alpha tubulin in human T lymphocytes. Eur. J. Immunol. 24, 99–106. Lin, S.D., Fann, M.J., 1998. Differential expression of protein kinases in cultured primary Fig. 7. A proposed model for the action of Cdk12 and Cdk13 in differentiated neurons. neurons derived from the cerebral cortex, hippocampus, and sympathetic ganglia. – Cdk12 and Cdk13 independently regulate axonal elongation in differentiated neurons, in J. Biomed. Sci. 5, 111 119. fi part, through modulation of Cdk5 expression. Liu, P., Jenkins, N.A., Copeland, N.G., 2003. Ahighlyef cient recombineering-based method for generating conditional knockout mutations. Genome Res. 13, 476–484. Malumbres, M., 2011. Physiological relevance of cell cycle kinases. Physiol. Rev. 91, 973–1007. Chen, H.H., Wong, Y.H., Geneviere, A.M., Fann, M.J., 2007. CDK13/CDC2L5 interacts Malumbres, M., Harlow, E., Hunt, T., Hunter, T., Lahti, J.M., Manning, G., Morgan, D.O., Tsai, with L-type cyclins and regulates alternative splicing. Biochem. Biophys. Res. L.H., Wolgemuth, D.J., 2009. Cyclin-dependent kinases: a family portrait. Nat. Cell Biol. – Commun. 354, 735 740. 11, 1275–1276. Cheng, S.W., Kuzyk, M.A., Moradian, A., Ichu, T.A., Chang, V.C., Tien, J.F., Vollett, S.E., Pandithage,R.,Lilischkis,R.,Harting,K.,Wolf,A.,Jedamzik, B., Luscher-Firzlaff, J., Vervoorts, J., fi Grif th, M., Marra, M.A., Morin, G.B., 2012. Interaction of cyclin-dependent kinase Lasonder, E., Kremmer, E., Knoll, B., Luscher, B., 2008. The regulation of SIRT2 function by 12/CrkRS with cyclin K1 is required for the phosphorylation of the C-terminal domain cyclin-dependent kinases affects cell motility. J. Cell Biol. 180, 915–929. – of RNA polymerase II. Mol. Cell. Biol. 32, 4691 4704. Rodrigues, F., Thuma, L., Klambt, C., 2012. The regulation of glial-specificsplicingof Cheung, Z.H., Ip, N.Y., 2007. The roles of cyclin-dependent kinase 5 in dendrite and synap- Neurexin IV requires HOW and Cdk12 activity. Development 139, 1765–1776. – se development. Biotechnol. J. 2, 949 957. Sirajuddin, M., Rice, L.M., Vale, R.D., 2014. Regulation of microtubule motors by tubulin Cheung, Z.H., Fu, A.K., Ip, N.Y., 2006. Synaptic roles of Cdk5: implications in higher cogni- isotypes and post-translational modifications. Nat. Cell Biol. 16, 335–344. – tive functions and neurodegenerative diseases. Neuron 50, 13 18. Smith, D., 2003. Cdk5 in neuroskeletal dynamics. Neurosignals 12, 239–251. Chung, K.H., Hart, C.C., Al-Bassam, S., Avery, A., Taylor, J., Patel, P.D., Vojtek, A.B., Turner, D. Speranza, L., Chambery, A., Di Domenico, M., Crispino, M., Severino, V., Volpicelli, F., L., 2006. Polycistronic RNA polymerase II expression vectors for RNA interference Leopoldo, M., Bellenchi, G.C., di Porzio, U., Perrone-Capano, C., 2013. The serotonin re- based on BIC/miR-155. Nucleic Acids Res. 34, e53. ceptor 7 promotes neurite outgrowth via ERK and Cdk5 signaling pathways. Neuro- Conde, C., Caceres, A., 2009. Microtubule assembly, organization and dynamics in axons pharmacology 67, 155–167. – and dendrites. Nat. Rev. Neurosci. 10, 319 332. Stiess, M., Bradke, F., 2011. Neuronal polarization: the cytoskeleton leads the way. Dev. Copeland, N.G., Jenkins, N.A., Court, D.L., 2001. Recombineering: a powerful new tool for Neurobiol. 71, 430–444. – mouse functional genomics. Nat. Rev. Genet. 2, 769 779. Veeranna, Shetty, K.T., Takahashi, M., Grant, P., Pant, H.C., 2000. Cdk5 and MAPK are asso- – Craig, A.M., Banker, G., 1994. Neuronal polarity. Annu. Rev. Neurosci. 17, 267 310. ciated with complexes of cytoskeletal proteins in rat brain. Brain Res. Mol. Brain Res. Cruz, J.C., Tsai, L.H., 2004. A Jekyll and Hyde kinase: roles for Cdk5 in brain development 76, 229–236. – and disease. Curr. Opin. Neurobiol. 14, 390 394. Witte, H., Neukirchen, D., Bradke, F., 2008. Microtubule stabilization specifies initial neu- Dai, Q., Lei, T., Zhao, C., Zhong, J., Tang, Y.Z., Chen, B., Yang, J., Li, C., Wang, S., Song, X., Li, L., ronal polarization. J. Cell Biol. 180, 619–632. Li, Q., 2012. Cyclin K-containing kinase complexes maintain self-renewal in murine Wong, Y.H., Lu, A.C., Wang, Y.C., Cheng, H.C., Chang, C., Chen, P.H., Yu, J.Y., Fann, M.J., 2010. – embryonic stem cells. J. Biol. Chem. 287, 25344 25352. Protogenin defines a transition stage during embryonic neurogenesis and prevents Dotti, C.G., Sullivan, C.A., Banker, G.A., 1988. The establishment of polarity by hippocampal precocious neuronal differentiation. J. Neurosci. 30, 4428–4439. – neurons in culture. J. Neurosci. 8, 1454 1468. Xie, Z., Samuels, B.A., Tsai, L.H., 2006. Cyclin-dependent kinase 5 permits efficient cy- Farah, M.H., Olson, J.M., Sucic, H.B., Hume, R.I., Tapscott, S.J., Turner, D.L., 2000. Generation toskeletal remodeling–a hypothesis on neuronal migration. Cereb. Cortex 16, of neurons by transient expression of neural bHLH proteins in mammalian cells. De- i64–i68. – velopment 127, 693 702. Yu, J.Y., DeRuiter, S.L., Turner, D.L., 2002. RNA interference by expression of short- Fourest-Lieuvin, A., Peris, L., Gache, V., Garcia-Saez, I., Juillan-Binard, C., Lantez, V., Job, D., interfering RNAs and hairpin RNAs in mammalian cells. Proc. Natl. Acad. Sci. 99, 2006. Microtubule regulation in mitosis: tubulin phosphorylation by the cyclin- 6047–6052. – dependent kinase Cdk1. Mol. Biol. Cell 17, 1041 1050. Yuan, H.X., Xiong, Y., Guan, K.L., 2013. Nutrient sensing, metabolism, and cell growth con- Fukushima, N., Furuta, D., Hidaka, Y., Moriyama, R., Tsujiuchi, T., 2009. Post-translational trol. Mol. Cell 49, 379–387. modifications of tubulin in the nervous system. J. Neurochem. 109, 683–693.

Fig. 6. Overexpression of Cdk5 partially rescues neurite defects in Cdk12-depleted and Cdk13-depleted P19 cells. (A, C) The expression levels of Cdk12, Cdk13, and Cdk5 proteins in P19 cells transfected with the indicated constructed. (B, D) Representative images of P19 cells transfected with the indicated constructs and pp35 are shown. (E, F) The average neurite lengths and the neurite length distribution of P19 cells transfected with the indicated Cdk5 and Cdk12 constructs are shown. **, p b 0.01. There is a sig- niﬁcant difference between the pCdk12-i1 + pp35 and pCdk12-i1 + pCdk5 + pp35 groups by the two sample K–Stest(p b 0.01). (G, H) The average neurite lengths and neurite length distribution in P19 cells transfected with the indicated Cdk5 and Cdk13 constructs are shown. **, p b 0.01. There is a signiﬁcant difference between pCdk13- i1 + pp35 and pCdk13-i1 + pCdk5 + pp35 groups by the two sample K–Stest(p b 0.001). Scale bars 100 μm.