Homozygous mutations in the cyclin-dependent kinase-5 regulatory subunit-associated protein 2 gene CDK5RAP2 cause primary autosomal recessive microcephaly (MCPH). MCPH is characterized by a pronounced reduction of brain volume, particularly of the cerebral cortex, and mental retardation. Though it is a rare developmental disorder, MCPH has moved into the spotlight of neuroscience because of its proposed central role in stem-cell biology and brain development. Investigation of the neural basis of genetically defined MCPH has been limited to animal studies and neuroimaging of affected patients as no neuropathological studies have been published. In the present study, we depict the spatiotemporal expression of CDK5RAP2 in the developing brain of mouse and human. We found intriguing concordance between regions of high CDK5RAP2 expression in the mouse and sites of pathology suggested by neuroimaging studies in humans and mouse. Our findings in human tissue confirm those in mouse tissues, underlining the function of CDK5RAP2 in cell proliferation and arguing for a conserved role of this protein in the development of the mammalian cerebral cortex.
Cyclin-dependent kinase-5 regulatory subunit-associated protein 2 (CDK5RAP2) has moved into the spotlight of neuroscience because of its central function in neural stem-cell proliferation, and thus brain development, as well as its proposed role in mammalian brain evolution. Homozygous mutations in the CDK5RAP2 gene cause primary autosomal recessive microcephaly type 3 (MCPH3; Moynihan et al. 2000; Bond et al. 2005), a rare developmental disorder of the brain characterized by a pronounced reduction of brain volume, particularly of the neocortex, and mental retardation (reviewed in Thornton and Woods 2009; Kaindl et al. 2010; Mahmood et al. 2011). Investigation of the neural basis of genetically defined MCPH has been limited in humans to neuroimaging of affected patients as no neuropathological study has been published to date. Neuroimaging studies performed in patients with MCPH provide evidence of microencephaly that becomes apparent by the 24th week of gestation, with a strong reduction of the cerebral cortex volume. The cortical gyration has been described as simplified, and only few patients with further anomalies such as periventricular heterotopias, infra-tentorial anomalies, dysmorphic/enlarged lateral ventricles, corpus callosum agenesis, and micropolygyria and/or dysplasia have been reported (Woods et al. 2005; Passemard et al. 2009). Specifically for MCPH3, neither neuroimaging nor neuropathological studies have been reported to date in humans.
From the human, non-progressive phenotype, it has become clear that CDK5RAP2 controls the brain size during fetal development. One current model for the MCPH phenotype of CDK5RAP2 mutations invokes a premature shift from symmetric to asymmetric neural progenitor-cell divisions (with a subsequent depletion of the progenitor pool and a reduction of the final number of neurons). In addition, a reduction in cell survival has been proposed. Underlying mechanisms include a deregulation of CDK5RAP2′s role in centrosome function, spindle assembly and/or response to DNA damage (reviewed in Kraemer et al. 2011; Megraw et al. 2011). Centrosome dysfunction can disrupt many key processes including cell polarity, cell division/cell-cycle progression, progenitor fate and survival, and DNA damage response. When centrosomes are inactivated in neural progenitors, spindle orientation becomes random and can promote asymmetric divisions producing neurons early at the expense of progenitor pool expansion (Siller and Doe 2009). Similarly, increased cell-cycle exit has been shown to lead to a reduction in neocortical numbers (Tarui et al. 2005). The exact link between DNA damage response and MCPH is unclear.
In support of these mechanisms, the analysis of the MCPH3-model Hertwig's anemia mouse (exon 4 inversion of the Cdk5rap2 gene, deletion of a large part of the y-tubulin ring complex binding domain) with small brains and thin superficial cerebral cortex layers has revealed as possible underlying pathomechanisms: A loss of neural progenitors coinciding with a premature cell-cycle exit and elevated apoptosis rates, a delay in mitotic progression, an abnormal mitotic spindle orientation, indications for a premature switch from symmetric to asymmetric cell division, and decreased neurogenesis (Lizarraga et al. 2010). These mice also display smaller ganglionic eminences, hippocampi, and olfactory bulbs. In contrast to the reported human phenotype, Hertwig's anemia mice have a hematopoietic phenotype (hypoproliferative anemia, leucopenia, predisposition to hematopoietic tumors) and defects in multiple organs including the thymus. In 2 further splice trap mutation mice (Cdk5rap2RRF465, Cdk5rap2RRU031), MCPH was not evident, but in their embryonic fibroblasts centriole amplification, loss of centriole cohesion and engagement, mitotic delay, multipolar spindles, supernumerary cilia, and premature senescence were detected (Barrera et al. 2010). Consistent with this scenario, knockdown of Cdk5rap2 via in utero shRNAi in embryonic brains resulted in reduced proliferation and premature cell-cycle exit of neural progenitors with premature neuronal differentiation and a depletion of the germinal matrix (Buchman et al. 2010). In conjunction with further in vitro studies (often in tumor cells), CDK5RAP2 function specifically contributes to centrosome cohesion and engagement, centrosome asymmetry, centriole replication control, gamma-tubulin recruitment to the centrosome, attachment of centrosomes to spindle poles, kinetochore attachment to spindles, chromosome condensation, binding to the dynactin complex, DNA damage response, checkpoint control, cell-cycle regulation and exit, microtubule plus-end dynamics, mitotic spindle formation and function/orientation, primary cilium formation, and cell survival (reviewed in Megraw et al. 2011). Linking these cellular phenotypes to physiological processes in the developing brain, using animal models and human specimen, will be crucial for an understanding of the mechanisms of neocortex development and the pathomechanism underlying MCPH.
CDK5RAP2 is widely expressed in mouse and human tissue with high levels detected in the central nervous system (CNS) during early embryonal development (Ching et al. 2000; Bond et al. 2005; Buchman et al. 2010). The aim of the present study was to characterize CDK5RAP2 expression and localization during murine and human brain development and to compare the results between these species. We further aimed to bring our results in perspective with the phenotype described in humans with MCPH and in disease models.
Materials and Methods
C57BL/6 mice were obtained from the animal facility of the Charité – Universitätsmedizin Berlin, Germany. The breeding was performed during the day, the day of insemination was considered as embryonic day (E) 0 (E0), and the day of birth was designated as postnatal day (P) 0 (P0). All experiments were carried out in accordance to the national ethic principles (registration no. T0113/08).
For quantitative real-time PCR (RTQ-PCR) and western blots, the cerebral cortex was quickly dissected from mice at E10, E12, E14, E16, P0, P5, P10, P20, and P56 (n = 6 per group). From the P0 pups, brain, kidney, thymus, lung, heart, bladder, intestines, liver, and placenta were dissected (n = 3–6 per group). The harvested tissues were snap frozen and stored at −80 °C for later RNA and protein extraction.
For histological analysis, brains from mice at E10, E12, E14, E16, P0, P5, P10, and P56 (n= 3–12 per age) were dissected and immersed in 4% paraformaldehyde in 0.12 M TPO4. Further brain fixation was performed in the same solution at 4 °C for 1–2 h (embryonic brains) or overnight (postnatal brains). Brains were cryoprotected through overnight incubation in 10% sucrose 0.12 M TPO4 solution at 4 °C followed by an overnight incubation in 20% sucrose 0.12 M TPO4. The brains were then immersed in a solution of 7.5% gelatin, 20% sucrose in 0.24 M TPO4 for 1 h at 37 °C and subsequently embedded in a block with the same solution for 1 h at 4 °C. The block was frozen in 2-methylbutane at −60 °C and stored at −80 °C. Coronal and sagittal sections of 10 μm thickness were cut on a cryostat and collected on Superfrost plus slides (R. Langenbrinck, Emmendingen, Germany).
Tissues of human embryos and fetuses of 12, 15–16, 23, 30, 32, and 38 gestational weeks (GW) and 10 postnatal weeks of age were used in this study with approval from the local ethic committee (approval no. EA1/212/08). Paraffin-embedded tissue was used and processed for 3,3'-diaminobenzidine tetrahydrochloride (DAB) immunohistology as mentioned below.
To design a peptide antibody directed against mouse Cdk5rap2, we used algorithms for secondary and tertiary protein structure prediction and chose a sequence of the N-terminal protein region. This peptide (NH2-DSGMEEEGALPGTLSGC-COOH; amino acids 2–18 of the Cdk5rap2 mouse sequence, accession no. NP_666102.2), is unique for Cdk5rap2 and was used to immunize rabbits. The antibodies were purified by absorption of the rabbit sera following 2 injections of the peptide to an antigen-coupled column (Cdk5rap2 peptides immobilized onto CNBr-activated sepharose) and eluted at low pH. The specificity of the peptide antibody against Cdk5rap2 was further tested by western blot (data not shown) and on brain sections by peptide precompetition, by applying the secondary antibody only, and for immunofluorescence by using conditional knockout mouse sections (Supplementary Fig. S1). Further primary and secondary antibodies utilized in this study are specified in Supplementary Table S1.
RNA Extraction and Quantitative Real-Time PCR
Total RNA from tissue specimen was extracted using TRI-Reagent® (Sigma-Aldrich, Taufkirchen, Germany) according to the manufacturer's recommendations for RNA isolation from tissue samples, and cDNA was prepared by reverse transcription using the ThermoScript® RT-PCR System (Invitrogen, Karlsruhe, Germany), using a combination of oligo(dT)20 and random hexamer primers. cDNA synthesis was performed on 1 μg of RNA. For quantitative real-time PCR, 1 µL of 1:10 diluted cDNA was used as template. To specifically amplify and detect Cdk5rap2 and Hprt (hypoxanthine-guanine phosphoribosyl-transferase, housekeeping gene) cDNA, we designed the corresponding specific sets of sense and antisense primers and TaqMan probe using the GenScript Real-Time PCR (TaqMan) Primer Design online software (www.genscript.com). Primer and probe sequences were: mCdk5rap2-F 5′-TCAGAGGCGTTGGGTGAGA-3′; mCdk5rap2-R 5′-GGATCAACAAGCCCGTCTTC-3′; mCdk5rap2-probe FAM-5′-CAACAGGCCACTCACCTCTCATTCCC-3′-TAMRA; mHprt-F 5′-ATCATTATGCCGAGGATTTGGAA-3′; mHprt-R 5′-TTGAGCACACAGAGGGCCA-3′; and mHprt-probe FAM-5′-TGGACAGGACTGAAAGACTTGCTCGAGATG-3′-TAMRA. Experiments (n = 3–6 per group) were run in triplicate. PCR was performed in an Applied Biosystems 7500 Fast Real-time PCR System (Applied Biosystems Inc., Norwalk, CT, United States of America) in 96-well microtiter plates using a final volume of 13 µL. The reaction mixture consisted of 1× TaqMan® Universal PCR Master Mix, No AmpErase® UNG (Roche, Branchburg, NJ, United States of America), 385 nM primer F, 385 nM primer R, 230 nM probe, and 1 µL template cDNA. Amplification was performed with the thermal profile of 50 °C for 2 min, initial denaturation at 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 15 s, and a combined primer annealing/extension step at 60°C for 1 min, during which the fluorescence signal was acquired. Ct values were calculated using the 7500 Fast System SDS Software (Applied Biosystems Inc.) and further statistical calculations were performed on GraphPad Prism 5 Software (GraphPad Software Inc., La Jolla, CA, United States of America). The 2−ΔΔCt method was applied for the quantification of the relative expression of the Cdk5rap2 mRNA using the housekeeping gene Hprt as the endogenous control for normalization.
Protein Extraction Procedure and Western Blot
Protein extracts for western blots were isolated from tissues by homogenization in radioimmunoprecipitation assay buffer containing 1 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich) and 1 protease inhibitor cocktail tablet (Complete Mini; Roche Diagnostics, Mannheim, Germany), 20 min incubation on ice, and centrifugation at 4 °C for 10 min at 3000 × g and for 20 min at 16 000 × g. Protein concentrations were determined using a bicinchoninic acid (BCA)-based assay, according to the instructions of the manufacturer (BCA Protein Assay Kit; Pierce Biotechnology, Rockford, IL, United States of America). Protein extracts (30 µg per sample) were denaturated in Laemmli sample loading buffer at 95 °C for 5 min, separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), and electrophoretically transferred in transfer buffer in a semi-dry fashion using Trans-Blot SD Semi-Dry transfer cell (Bio-Rad, Munich, Germany) onto nitrocellulose membrane (Bio-Rad). The membranes were incubated for 1 h at room temperature (RT) in blocking buffer [tris-buffered saline with tween (TBS-T) 1× with 5% bovine serum albumin], rinsed 3 times with TBS-T 1× for 8 min each at RT on a shaker, and then incubated overnight at 4 °C with rabbit anti-Cdk5rap2 (1:500) and mouse anti-beta actin (1:10 000; Sigma-Aldrich) antibodies. After incubation with the corresponding secondary antibodies donkey anti-rabbit (1:2000; Amersham Biosciences, Freiburg, Germany) and goat anti-mouse (1:10 000; Dako, Hamburg, Germany), the immunoreactive proteins were visualized using a technique based on a chemiluminescent reaction. The gel pictures were obtained with a Bio-Rad imager (Bio-Rad) and further statistical calculations were performed on GraphPad Prism 5 Software (GraphPad Software Inc.). Western blot experiments were run in triplicate.
Cryostat sections were air-dried briefly prior to rinsing in phosphate buffered saline (PBS 1×) for 10 min and staining buffer (0.2% gelatin, 0.25% Triton X-100 in PBS 1×) for 20 min. In a 30-min blocking step, sections were incubated in 10% donkey or goat normal serum (DNS or GNS) in staining buffer at RT. Sections were incubated overnight at RT with primary antibodies in staining buffer containing 10% DNS or GNS followed by an incubation with the corresponding secondary antibodies for 2 h at RT. Nuclei were labeled with 4′,6-diamidino-2-phenylindole (DAPI, 1:1000, Sigma-Aldrich). CDK5RAP2 was visualized in human tissue through DAB staining using a commercial rabbit anti-CDK5RAP2 antibody (1:200; Abnova, Heidelberg, Germany). In brief, paraffin sections were deparaffinized, rinsed in staining buffer, incubated 30 min in 10% GNS at RT, and incubated overnight at RT with the primary antibody followed by an incubation with streptavidin goat anti-rabbit biotinylated antibody (Invitrogen) 1:400 for 2 h at RT. Endogenous peroxidase was quenched through subsequent incubation in 0.3% H2O2 for 10 min at RT, and signal amplification was performed using the Vectastain ABC elite system® (Vector laboratories, Orton Southgate, United Kingdom). Color development was achieved by adding 17 µL of H2O2 30% to sections incubated in a solution containing 200 mg/L DAB, 0.05 M Tris, and 0.6% NiNH4SO4. Color development was stopped through rinsing sections in 0.05 M Tris solution, and sections were mounted with Entelan® following dehydration. Human sections were also stained with Hematoxylin and Eosin or anti-MIB1 (Dako) using standard procedures. See Supplementary Table S1 for a comprehensive list of antibodies and concentrations.
Fluorescently labeled sections were analyzed and imaged by a fluorescent Olympus BX51 microscope with the software Magnafire 2.1B (2001) (Olympus, Hamburg, Germany), and confocal microscopy images were taken by an lsm5exciter Zeiss confocal microscope with the software Zen (version 2009, Zeiss, Jena, Germany). All images were processed using Adobe Photoshop.
Cdk5rap2 in Developing Mouse and Human Tissues
Cdk5rap2 is widely expressed and synthesized in murine tissues during development. In the newborn mouse (P0), Cdk5rap2 mRNA levels were highest in brain, thymus, and kidney and detected at lower levels in lung, heart, bladder, and liver (Supplementary Fig. S2a). Similar results were obtained for Cdk5rap2 protein levels through western blot (Supplementary Fig. S2b), and Cdk5rap2 was detected in various P0 mouse pup organs such as brain, kidney, thymus, lung, heart, liver, and intestines through immunofluorescence (Supplementary Fig. S2c). In human fetal tissues at GW16, CDK5RAP2 immunopositivity was likewise present in various organs including brain, thymus, kidney, lung, heart, liver, and intestines, where the protein was localized particularly in regions of high proliferation identified through MIB1 immunopositivity (Supplementary Fig. S3a–f).
In the murine brain, Cdk5rap2 immunopositivity was marked within the ventricular zone (VZ) and the rhombic lip at E10 (Fig. 1c–j) and E12 (Supplementary Fig. S4 and data not shown), within the rhombic lip, the VZ, the cortical plate, the cerebellum, and the hippocampus at E14 and E16, and within the VZ and subventricular zone (SVZ) at E16 (Figs 2, 8, 11, Supplementary Fig. S4 and data not shown). At P0, Cdk5rap2 could be visualized particularly in the VZ, SVZ, and the upper cerebral cortex layers; marked levels were also distinguishable within the hippocampus, the cerebellum, and the medial septum with a sparing of the lateral septum (Figs 1k–p and 2; Supplementary Figs S4 and S5). At this age, areas of predominantly white matter such as the corpus callosum and the anterior comissure were largely spared with only few Cdk5rap2-positive cells (Fig. 1q–t). In the adult brain, Cdk5rap2 immunopositivity was low and seen mainly within the anterior SVZ and the cerebellum (Figs 2, 11, Supplementary Figs S4 and S5). Similar results were obtained for the developing human fetal brains at GW12, 14–15, 18, and 38 (Fig. 4a–d and data not shown). During the early developmental period from GW12 to GW18, CDK5RAP2 immunopositivity was present throughout the hemispheres, with highest density in the VZ and the superficial cerebral cortex layers. The density of CDK5RAP2-positive cells decreased with maturation similar to the finding in the murine brain. In the full-term infant at GW38, CDK5RAP2 was distributed in low density throughout the hemispheres, with highest levels in the VZ/SVZ. Here, CDK5RAP2 was not localized predominantly in the superficial layers of the cerebral cortex anymore, similar to our findings in the maturing mouse brain.
Cdk5rap2 in the Developing Murine and Human Cortex
In the murine cerebral cortex, Cdk5rap2 protein and mRNA were detected at high levels between E10 and E16, levels were still high at P0, subsequently decreased with development and were hardly detectable at adult age of P56 (Fig. 1a,b). To determine the regional expression of Cdk5rap2 in the cortex during development, we performed immunohistological stainings at various ages from E10 to P56 (Fig. 2, Supplementary Figs S4–S6). At E10, a dense punctate Cdk5rap2 staining, colocalizing with the centrosome marker γ-tubulin, was present along the wall of the neural tube (Figs 2 and 5, Supplementary Figs S4 and S6). At this age, the Cdk5rap2 signal also appeared in a mesh-like arrangement within the VZ indicating an additional non-centrosomal localization. To analyze whether this corresponds to the previously reported localization of Cdk5rap2 to the Golgi network (Wang et al. 2010), we performed a double immunostaining with the cis Golgi-marker GM130 and detected only a partial colocalization (Supplementary Fig. S7). At E12, the centrosomal Cdk5rap2 signal was preserved in the VZ and became denser within the cortical preplate (Supplementary Fig. S4). At E14, corticogenesis has progressed and has lead to the formation of the SVZ, the subplate, the cortical plate (CP), and the marginal zone (MZ) in mice (O'Leary and Nakagawa 2002). At this time, the centrosomal localization of Cdk5rap2 was preserved in the VZ, SVZ, and CP and to a lesser extent in the MZ (Supplementary Fig. S4). By E16, a stage of corticogenesis when division activity in the VZ is decreased but still high in the SVZ due to the generation of neurons for the superficial layers of the cortex (Tarabykin et al. 2001; Haubensak et al. 2004; Noctor et al. 2004), the Cdk5rap2 immunopositivity within the VZ decreased further and was present especially in the SVZ (shift from VZ to SVZ at this stage; Fig. 2, Supplementary Fig. S4). Denser centrosomal Cdk5rap2 staining also appeared within the CP, and the dispersed centrosomal MZ staining was retained. The non-centrosomal staining pattern seen at E10 decreased strongly with maturation and was hardly discernable by E16. At P0, the Cdk5rap2 staining was present throughout the cortex with a dense centrosomal Cdk5rap2 signal within the SVZ and the upper cortical layers (Fig. 2). Much lower Cdk5rap2 signal density and staining differences between cortical layers were observed at P5 and P10 (data not shown). At the adult age of P56, the Cdk5rap2 staining was hardly discernable, and individual centrosomal Cdk5rap2 signals were evenly distributed between the cortical layers (Fig. 2, Supplementary Figs S4 and S6).
To analyze more in detail the localization of Cdk5rap2 in the developing neocortex, we performed costainings with layer-specific markers. At E16 and even more clearly at P0, centrosomal Cdk5rap2 signal was localized below that of the Reelin staining, which is specific for layer I (Alcantara et al. 1998), colocalized with the upper layer stained with the layers II–IV marker Cux1 (Nieto et al. 2004), and to a lesser extent partially colocalized with the deep layers V–VI marker Ctip2 at E16 (even less colocalization with the deep layers at P0; Leid et al. 2004; Figure 3). While Cux1 in the adult brain is confined almost exclusively to the upper layers, this protein has been detected in proliferative regions of the developing cerebral cortex and in the emerging population of postmitotic upper-layer neurons in the cortical mantle (Nieto et al. 2004). Similar to previous reports, we detected Cux1 also in the SVZ/VZ and in deeper layers of the cortex (possibly in cells migrating to the upper layers) at E16 (data not shown), that is, in regions of the neocortex where Cdk5rap2 is also localized.
Similar results were obtained for human fetal brains at GW12, 14–15, 18, and 38 (Fig. 4). The earliest ages of human fetal brain specimen analyzed in this study (GW12) correspond to rather late stages of corticogenesis in the mouse (about E16; Molnar and Clowry 2012). Neurogenesis and neuronal migration toward the expanding cortical plate are both ongoing at GW12. At this time, a strong CDK5RAP2 signal was observed in the VZ (also referred to as neuroepithilium), most prominently in cells lining the ventricular wall where neural stem cells reside (Altman and Bayer 2002). This staining decreased toward the SVZ and the intermediate zone (IZ; also referred to as stratified transitional field; Altman and Bayer 2002). A further staining density gradient could be discerned within the IZ, containing migrating neurons with less CDK5RAP2-positive cells in the outer than in the inner IZ. In the cortical plate, most of the tightly packed cells are also CDK5RAP2 positive with a slight accentuation of the superficial layers. The MZ (layer I) contained almost no CDK5RAP2-positive cells. At GW14–15, the high proliferation zone of the ganglionic eminence contained mainly CDK5RAP2-positive cells, delineating it from the adjacent basal ganglia (caudate and putamen) with significantly less CDK5RAP2-positive cells. The internal capsule and the IZ contained only few cells, but these displayed strong CDK5RAP2 punctae. In the frontal lobe, the VZ, SVZ, and inner IZ were prominently CDK5RAP2 labeled with a staining gradient in the VZ/SVZ from lateral (denser) to medial (less dense staining). These regions were demarcated from the IZ, where less cells were CDK5RAP2 positive. The cortical plate was packed with CDK5RAP2-positive cells. At this age, there was still an accentuation of the superficial layers in the lateral neocortex that could not be discerned anymore in the medial cortex. Again, layer I contained only very few cells and almost no CDK5RAP2 signal. In contrast, the subpial granular layer, where cells including Cajal Retzius cells settle, contained CDK5RAP2 signals. At GW18, the CDK5RAP2 labeling was globally reduced in comparison with previous ages. Cells of the ganglionic eminence retained a CDK5RAP2 staining; however, it was most prominent close to the ventricle, where MIB1-positive proliferating cells reside. The basal ganglia, rich in more mature neurons, contained only very few CDK5RAP2-positive cells. At this age, a cytoplasmatic-appearing CDK5RAP2 staining pattern appeared in cells within the internal capsule. The latter was, thus, similar to the staining observed in murine GFAP-positive glial cells in this region. In the frontoparietal cortical mantle, the CDK5RAP2 signal was also less dense with strongest labeling in the VZ/SVZ and also in the cortical plate with exception of layer I (no accentuation of the ventricular lining or the superficial cortical layers at this age); there was sparse signal in the IZ. Again, the subpial granular layer contained CDK5RAP2-positive cells. In comparison with GW18, the number of CDK5RAP2-positive cells was further decreased within the cortical mantle at GW22 and almost not discernable at GW38.
Cdk5rap2 in Neural Cells
Cdk5rap2 signal colocalizes with cells positive for the progenitor marker Nestin at the VZ and SVZ during murine brain development (Fig. 5a). All pH3-positive M-phase proliferating cells within the VZ at E10 and within the VZ/SVZ at E16 were also Cdk5rap2 positive. At E10, all dividing cells at the VZ were Cdk5rap2 positive, independent of their division plane orientation (77 cells, n= 3). In these cells, Cdk5rap2 colocalizes with the centrosomal marker γ-tubulin throughout the cell cycle, from interphase to anaphase (Fig. 5b,c).
Cdk5rap2 further colocalizes strongly with the markers of glial cells in the developing and the mature brain (Fig. 6 and data not shown). In the adult murine brain, about 99% of analyzed GFAP-positive astrocytes were also Cdk5rap2 positive (90 of 91 GFAP+ cells), and about 90% of analyzed Iba1-positive microglia were Cdk5rap2 positive (122 of 135 Iba1+ cells) when assessed in the SVZ (n = 3 per group). Cdk5rap2 immunopositivity was also detected in the white matter exhibiting high myelin basic protein (MBP) immunopositivity as a marker for oligodendrocyte sheaths (Fig. 6).
To further address the prevalent gray matter reduction in MCPH patients, we analyzed the occurrence of Cdk5rap2 in cells belonging to the neuronal lineage. Here, Cdk5rap2 is present in early neurons (Fig. 7). Within the cortex, Cdk5rap2 signal colocalizes with regions exhibiting high levels of early neuron marker Tuj1 (neuronal class III β-tubulin) immunopositivity at E10, E16, and P0 (Fig. 7a–c). This is not the case for cells expressing the more mature neuron marker NeuN at P0 and P56 (Fig. 7d,e). In adult mice, most NeuN-positive mature neurons in the upper layers of the parietal cerebral cortex were negative for Cdk5rap2 (13 Cdk5rap2+/NeuN+ cells of 388 NeuN+ cells analyzed [0.03%], n = 3). The few NeuN-positive cells that were also Cdk5rap2 positive displayed only a very faint Cdk5rap2 signal (Fig. 7e). Cdk5rap2 immunopositivity could also be detected within distinct tangential migratory pathways, including the rostral migratory stream (RMS), where progenitors and early differentiating cells migrate toward the olfactory bulb (Supplementary Fig. S8).
Cdk5rap2 in Developing Murine and Human Hippocampus
We further analyzed Cdk5rap2 in the developing hippocampus as a further source of neurons in the developing brain and in the dentate gyrus (DG), a region of neurogenesis retained in the adult brain (Gage 2000). Hippocampal cells proliferate prenatally, followed by dynamic growth and differentiation around the postnatal period and synaptogenesis in the first postnatal week in mice (Forster et al. 2006). At E16, the DG and the CA3–CA1 region were rich in centrosomal Cdk5rap2 (Fig. 8a). The general distribution pattern of Cdk5rap2 was similar at P0 (Fig. 8b). Here, Cdk5rap2 immunopositivity was detected in areas of early neuron marker Tuj1 labeling, surrounding the pyramidal neuron layer in the CA3–CA1 region and the DG; it also showed an overlay with progenitor marker nestin (data not shown) and proliferation marker pH3 (Fig. 8d,e). At P5 and P10, Cdk5rap2 immunopositivity was further decreased along the CA3–CA1 region with only scattered signals, while low-density centrosomal Cdk5rap2 could still be visualized in the DG (data not shown). In the adult hippocampus, a centrosomal Cdk5rap2 signal was still evident at the DG and a low-density signal pattern was scattered within the CA3–CA1 region. In the mature brain hippocampus at P56, Cdk5rap2-positive cells were predominantly GFAP-positive glial cells (Fig. 8c,h), and the few pH3-positive proliferating cells detected were Cdk5rap2 positive (Fig. 8g). Again, similar to the findings in the cerebral cortex, most NeuN-positive mature neurons did not display Cdk5rap2 (Fig. 8f).
Similar results were obtained for the hippocampus in human fetal brains at GW15–16 and GW30 (Fig. 9). The ages of human hippocampus specimen correspond approximately to the developmental stages analyzed in the mouse, that is, late prenatal and postnatal hippocampal development. The earliest production of granule cells in the mouse corresponds to the timing in the human (Angevine 1965; Rakic and Nowakowski 1981), that is, the first granule cells in the DG are formed at about the same time as the first pyramidal cells in the hippocampus. At GW15–16, the DG and the cornu ammonis have started to infold giving the hippocampus its characteristic flexure, and the pyramidal layer is clearly discernable. At that age, CDK5RAP2 labeling was of similar density in cells of the cornu ammonis and the DG. Later in development, at GW30, the DG had adopted a narrow U-shape and was still a high proliferative zone with a prominent CDK5RAP2 signal. In CA4, individual cells contained punctuate CDK5RAP2 signals, while the other pyramidal layers were bare of such labeling.
Cdk5rap2 in Developing Murine and Human Cerebellum
Cerebellar development begins late during vertebrate development with the formation of the rhombic lip, located at the surface of the neural tube where it is connected to the roof plate of the fourth ventricle (Wingate 2001). Granule neuron precursors generated at the rhombic lip give rise to the external granule layer (EGL) of the cerebellum, further proliferation within the EGL is followed by an inward migration of postmitotic cells and thus generation of the deeper cerebellar cortex lamina. In murine brains at E10, the wall of the fourth ventricle was stained with punctate, centrosomal Cdk5rap2 signal pattern and—similar to the finding in the VZ of the E10 cerebral cortex—with a non-centrosomal mesh-like staining pattern (Fig. 10). At E16, the EGL became visible as a dense multi-layer formation of nuclei located as the most outer part of the cerebellum primordium, and here, Cdk5rap2 appeared as dense centrosomal signal pattern. Below the EGL, in the molecular layer (ML), only very low nuclei density and also Cdk5rap2 immunoreactivity were detected. Within the internal granular layer (IGL), Cdk5rap2 immunoreactivity showed higher densities at the lateral part of the cerebellum primordia near the end of the vermis than within the medial region. At P0 and P5, the EGL, ML, and IGL were visible as distinct layers: Centrosomal Cdk5rap2 signal was highly evident in the EGL and IGL, and low in the ML, correlating also with the lower nuclei density within this latter layer, and it was evenly distributed throughout the matrix. In adult cerebellum, Cdk5rap2 immunoreactivity was detected within the Purkinje cell layer at the IGL. This Cdk5rap2 immunopositivity partially colocalized with GFAP-positive Bergman glia cells but not with Calbindin-positive Purkinje cells within the Purkinje layer (Fig. 11a–a′′′,c–c′′′). In contrast, almost no Cdk5rap2 immunopositivity was detected within the molecular layer, that is, the outer cerebellar layer consisting of vertically oriented glial fibers that project from Bergmann glia cells (Das 1976), as well as within the Calretinin-positive IGL (Fig. 11b–b′′′).
Similar results were obtained for the cerebellum in human fetal brains at GW15–16, GW32, and 10 postnatal weeks (Fig. 12). The timing of cerebellar development with respect to birth differs strikingly in human and rodent; in the latter, the cerebellum is relatively immature at birth and proliferation in the EGL, formation of the IGL, and foliation occur to a large part postnatally (Rakic and Sidman 1970; Sidman and Rakic 1973; Altman and Bayer 1997; Volpe 2009). In human fetuses in the second trimester of pregnancy, the entire surface of the cerebellar cortex is covered with a prominent EGL that is actively producing neural cells and reaches its peak thickness by about GW25 (Rakic and Sidman 1970; Bayer and Altman 2005). Underneath, a thin molecular layer is present in the nearly unlobulated hemispheres, and at this age, nearly all Purkinje cells are migrating (Zecevic and Rakic 1976; Bayer and Altman 2005). In the third trimester, proliferation of granule cells in the EGL followed by inward migration through the ML guided by Bergmann radial glial fibers and subsequent formation of the IGL is by far the dominant cellular determinant of cerebellar growth (Rakic and Sidman 1970). Postnatally, cerebellar growth slowly decelerates in humans, the EGL dissipates, and the IGL increases in size in the first years of life (Rakic and Sidman 1970).
At GW15–16 and GW32, CDK5RAP2 immunoreactivity was detected within the EGL, ML, and IGL (Fig. 12a′′–b′′), and CDK5RAP2 was still present at 10 postnatal weeks (Fig. 12c′′). While the broad and multi-layered EGL at the early ages in humans (GW15–16, GW32) contained CDK5RAP2-positive cells, particularly in the apical layers, its size was strongly reduced postnatally, and the CDK5RAP2 signal was less prominent in comparison with the prenatal ages. Postnatally, some labeled cells within the white matter could be first discerned. CDK5RAP2 immunopositivity overlapped with regions of high MIB1 positivity, detected especially in the outer proliferative EGL (Fig. 12a′-b′,a′′–b′′). Similar to the situation in mice, CDK5RAP2 immunopositivity was detected within the Purkinje cell layer at GW32 and at 10 postnatal weeks (Fig. 12c′′). Here, higher magnification revealed that the Purkinje cell perikaryon was homogenously CDK5RAP2 positive, but negative for a punctate (centrosomal) labeling. In contrast, adjacent cells, most likely Bergmann glia cells, displayed punctate CDK5RAP2 immunopositivity (Fig. 12 insert in c′′).
This study provides the first systematic description of the temporal and spatial expression pattern of CDK5RAP2 in the pre- and postnatal developing murine and human brain. Our findings are generally in line with current concepts regarding CDK5RAP2 function. The present results concerning CDK5RAP2 distribution during murine and human brain development indicate that the protein is neither uniformly expressed throughout the brain nor limited to just one brain area such as the neocortex, but instead shows restricted expression in a number of related brain structures. Brain regions in which we detected only very low levels of CDK5RAP2 include white-matter tracts such as the corpus callosum and the anterior commissure as well as the lateral septum. In the immature brain, CDK5RAP2 levels are especially high in regions of high proliferation, and brain structures expressing CDK5RAP2 during adulthood also express the protein during development. As the brain matures, CDK5RAP2 is refined to specific substructures within positive regions, which correspond to preserved proliferation zones/proliferating cells. For example, early expression throughout the hippocampus and DG becomes confined to the DG, a region of neurogenesis retained in the adult brain (Gage 2000). Similarly, cerebellar centrosomal Cdk5rap2 expression is restricted to the Bergmann glia within the Purkinje cell layer by adulthood. Thus, CDK5RAP2 appears to be tightly regulated both spatially and temporally during CNS development.
CDK5RAP2 in the Developing Brain
The timing of CDK5RAP2 expression coincides with the timing and localization of high proliferation rates during ontogenesis in both humans and mice. In our study, the highest Cdk5rap2 levels were detected at the initiation of murine cortical development around E10.5–E12.5 (Figs 1 and 2) that corresponds to a period of intense neural proliferation and neurogenesis. Also, Cdk5rap2 colocalized with progenitor cells and other proliferating cells (Fig. 5). Cdk5rap2 was detected in glial cells (Fig. 6) and in early neurons, but only rarely in mature neurons (Fig. 7). Similar findings were observed in the hippocampus and the cerebellum (Figs 8, 10, and 11). In human brain specimen, CDK5RAP2 was similarly localized to MIB1-positive proliferative zones, with highest labeling in earliest specimen analyzed at GW12, and a strong reduction of CDK5RAP2-positive cells with ongoing brain maturation (Figs 4, 9, and 12).
CDK5RAP2 Expression in Sites of Pathology Identified in Mouse Models and Humans
Previous studies of the neuroanatomical basis of MCPH have been limited in humans to brain imaging analyses (Woods et al. 2005; Passemard et al. 2009). These have illustrated that MCPH is a disorder of fetal brain growth with normal in utero head measurements until mid-gestation and grossly reduced CNS volume by birth (Woods et al. 2005). Such studies provided insight into the main structural abnormality in humans of predominant neocortex involvement that results from mutations in MCPH genes such as CDK5RAP2, but were unable to shed light on the exact developmental course that leads to these abnormalities (no studies exist specifically for MCPH3 to date). Based on recent in vivo and in vitro studies, the human MCPH phenotype is considered to be the result of a premature shift from symmetric to asymmetric neural progenitor-cell divisions (with a subsequent depletion of the progenitor pool) as well as of a reduction in cell survival. Mechanisms underlying response have recently been thoroughly reviewed (Kraemer et al. 2011; Megraw et al. 2011). There is a high level of concordance between structures implicated by our study and those suggested by complementary investigations of affected individuals with MCPH. Most notable, our observation of high CDK5RAP2 expression in the germinal matrix and the neocortex of murine embryos (Figs 1, 2, and 5) as well as human fetuses (Fig. 4) is in concordance with results of neuroimaging studies in MCPH patients, demonstrating a reduced brain volume that affects especially the neocortex (Woods et al. 2005; Passemard et al. 2009); it also agrees with the results of experimental studies that highlight the key role of CDK5RAP2 in neural progenitor proliferation (reviewed in Megraw et al. 2011).
We identified a preferential localization of CDK5RAP2 immunopositivity within the superficial layers of developing murine and human cerebral cortices (Figs 2–4). This is intriguing, as a thinning especially of the superficial cerebral cortex layers has been described in a MCPH3 mouse model (Lizarraga et al. 2010). Moreover, in neuropathological postmortem studies performed on non-genotyped Microcephaly vera patients, a reduction of the cerebral cortex has been described, and here, the superficial layers were reported to be especially thinned (Bamatter and Rabinowicz 1969; Robain and Lyon 1972). It remains to be elucidated if this preferential CDK5RAP2 localization in the superficial cortical layers is due to a specific role in the emerging upper layer neurons or rather a question of developmental timing. Cdk5rap2 immunopositivity was prominent in early, but not in mature neurons, and this finding is again in line with the current belief that Cdk5rap2 entails a key role in proliferation (Fig. 7). Patients with MCPH have been described to also have a slight reduced white-matter volume, and in line with this, we detected marked Cdk5rap2 immunopositivity in glial cells (Fig. 6). Why the white matter is not more severely affected in MCPH remains unclear. Future studies will need to address the question to what extent white-matter disease also contributes to brain size reduction in MCPH patients.
In the last years, neuroimaging studies have revealed that the cerebral cortex is not the only region affected in MCPH patients as previously believed. Rather, structural abnormalities have been reported in other brain regions such as the cerebellum (cerebellar hypoplasia) and the corpus callosum (agenesis), and periventricular heterotopias in individual MCPH patients indicate possible underlying migration defects. No obvious migration defect with abnormal cortical layering was detected in an MCPH3 mouse model (“Hertwig's anemia” mouse; Lizarraga et al. 2010). In these mice, brain regions other than the neocortex were not described in detail, but smaller ganglionic eminences, hippocampi, and olfactory bulbs were reported. In line with these descriptions, we found that the cerebral cortex is not the sole site of CDK5RAP2 expression in the murine and human brain. Striking CDK5RAP2 immunopositivity was, for example, also detected in the developing hippocampus and cerebellum (Figs 9–12). Moreover, Cdk5rap2 immunopositivity could also be detected within distinct migratory pathways, including the RMS, where progenitors and early differentiating cells migrate toward the olfactory bulb, and the stream from the ganglionic eminence toward the cerebral cortex (Fig. 8).
In addition to the brain, high levels of CDK5RAP2 were detected in other murine and human organs including the thymus and the kidney (Supplementary Fig. S2 and S3). In relation to this, it has been reported that Hertwig's anemia mice display “defects in multiple organs” including the thymus (Lizarraga et al. 2010). So far, such findings have not been reported for the 3 published families with MCPH3. In other MCPH subtypes, individual patients have been reported with short stature (MCPH1, MCPH5), early puberty, renal agenesia, and multicystic kidneys [reviewed in (Kaindl et al. 2010)]. Therefore, our data, which demonstrate the CDK5RAP2 also in various human tissues during development, draw attention to a possible link between the findings in humans and mice, highlighting a point that warrants further investigation in patients.
Evolutionary Neocortical Expansion and Highly Concordant CDK5RAP2 Expression Patterns in Mouse and Human Brain
A key finding of our study is the high degree of similarity between mouse and human CDK5RAP2 expression patterns in the developing brain, despite the given difference in timing of specific developmental processes, the limited human specimen availability and the fact that the cerebral cortex is smooth in mice in comparison with the prominently gyrated neocortex in humans. We did not find any evidence for regions of CDK5RAP2 expression that are observed only in humans or only in mice during brain development. Still, a main evolutionary step from murine to human neocortex is its obvious expansion and also an elaboration of neuronal connections (Rakic 2009). According to the radial glia hypothesis, this exponential increase is achieved through an increase in the number of neural stem cells (founder cells) by symmetrical divisions before the onset of neurogenesis, and this is underlined by the significant larger germinal zone in the human than in the mouse brain (Rakic 2009). Further explanations include an increase of radial columns and an increase of the number of asymmetric divisions that neural progenitors undergo. Thus, the positive selection of human-specific changes in the CDK5RAP2 protein sequence indicated in evolutionary studies rather than the timing and localization of the protein may be an explanation for neocortical expansion with the emergence of “higher intelligence”; see review by (Kraemer et al. 2011). Most relevant in relation to CDK5RAP2, the close regulation of symmetric and asymmetric cell divisions has been suggested to affect the cortical size in various species during evolutionary expansions (Rakic 2009). Our data suggest that CDK5RAP2 might be generally implicated in aspects of cell proliferation in mammalian species. Thus, positive selection of CDK5RAP2 protein changes leads probably to modifications in preexisting brain systems, rather than to acquisition of novel ones. Further studies will need to address the formation of the convolutions, since gyration has been described as “simplified” in many MCPH patients.
Brain size at birth is largely determined by the relative rates of proliferation and cell death. By highlighting regions of physiological CDK5RAP2 expression in human fetuses and infants, we offer a further glimpse into how a disruption of the CDK5RAP2 gene may impact on the development of particular brain systems in humans. CDK5RAP2 localizes to the germinal zones of the cortex in mice and humans, and its colocalization with markers of proliferating/progenitor cells underlines its proposed role in symmetric and asymmetric progenitor cell divisions and subsequent neocortical expansion during brain development. MCPH is considered as a predominant “neuronal disorder”. However, our results indicate a further function of Cdk5rap2 in glia cells, where Cdk5rap2 is also expressed. Future studies will need to address the molecular function of Cdk5rap2 in the white matter. Moreover, studying the development of other brain regions and other organs in animal models and patients (by neuroimaging and on the basis of postmortem samples) is warranted.
Our research was supported by the German Research Foundation (SFB 665), the Sonnenfeld Stiftung, and the Berliner Krebsgesellschaft e.V. We thank the SFB 665 brain bank for providing human samples.
The authors thank Victor Tarabykin, Robert Nitsch, Angelika Zwirner, Julia König, Stefanie Endesfelder, Marianne Peters, Pierre Gressens, Shyamala Mani, Nanette Sarioglu, Marta Rosario, Gregory Wulczyn, Catherine Verney, Annebelle Henrion, and Susanne Kosanke for discussions and technical assistance. Conflict of Interest: None declared.