The Role of Tuba1a in Adult Hippocampal Neurogenesis and the Formation of the Dentate Gyrus

The multitubulin hypothesis holds that each tubulin isotype serves a unique role with respect to microtubule function. Here we investigate the role of the α-tubulin subunit Tuba1a in adult hippocampal neurogenesis and the formation of the dentate gyrus. Employing birth date labelling and immunohistological markers, we show that mice harbouring an S140G mutation in Tuba1a present with normal neurogenic potential, but that this neurogenesis is often ectopic. Morphological analysis of the dentate gyrus in adulthood revealed a disorganised subgranular zone and a dispersed granule cell layer. We have shown that these anatomical abnormalities are due to defective migration of prospero-homeobox-1-positive neurons and T-box-brain-2-positive progenitors during development. Such migratory defects may also be responsible for the cytoarchitectural defects observed in the dentate gyrus of patients with mutations in TUBA1A.


Introduction
Microtubules are composed of ␣ -tubulin and ␤ -tubulin heterodimers and are known to play a vital role in numerous cellular processes including intracellular trafficking, migration and mitosis. In mammals there are at least 7 genes that encode ␣ -tubulins and another 7 for ␤tubulins [Villasante et al., 1986;Oakley, 2000;Khodiyar et al., 2007]. While there is a high degree of homology within these two gene families, each tubulin isotype possesses a unique amino acid sequence with varying expression patterns [Sullivan et al., 1985[Sullivan et al., , 1986Khodiyar et al., 2007]. These differences have led to the multitubulin hypothesis, which holds that each tubulin isotype serves a specific role with respect to microtubule function [Fulton and Simpson, 1976;McKean et al., 2001]. Evidence in support of this proposition is limited because the raising of specific antibodies to study the different tubulin isotypes has proved difficult [Linhartová et al., 1992].
An alternative approach is to employ genetic tools, whereby the function of tubulins is investigated by studying the phenotypic consequences of mutating each tubulin isotype. For example, the generation of Tubb1 knockout mice has revealed that this particular tubulin isotype plays an important role in the synthesis, structure and Neurogenesis in Tuba1a Mice Dev Neurosci 2010;32:268-277 269 function of blood platelets [Schwer et al., 2001]. We have recently reported the identification of a dominant S140G mutation in Tuba1a in the Jenna mouse (Jna/+) . The mutation impairs the ability of TUBA1A to bind guanosine triphosphate and results in a dramatic reduction in the efficiency of tubulin heterodimer formation. Consequently, neuronal migration is impaired, which results in a fractured pyramidal cell layer in the hippocampus and subtle layering defects in the cortex. The discovery that this gene is also mutated in humans suffering from lissencephaly further demonstrates the importance of TUBA1A for the migration of neurons . Its role, however, in mitotic division has not been explored.
Mitosis is a cellular process that is heavily dependent on microtubules. First, microtubules attach to kinetochores during prometaphase; then, in metaphase, they form the mitotic spindle facilitating the alignment of the sister chromatids, before mediating their separation in anaphase [Gadde and Heald, 2004]. Studies in the fly Drosophila melanogaster have shown that mutations in an ␣ -tubulin subunit ( ␣ TUB67C) compromise meiotic and mitotic division [Matthews et al., 1993]. Similarly, in the fungus Aspergillus nidulans, mutations in both ␣ -tubulins have been shown to disrupt mitosis by affecting the stability of the mitotic spindle [Morris, 1975;Gambino et al., 1984]. In this paper, we ask whether Tuba1a is required for cell division in the adult nervous system in mice.
We focused on the adult because Tuba1a expression is largely absent from the proliferative ventricular zones during development. In adulthood Tuba1a, in addition to being expressed in a wide array of neuronal structures [Bamji and Miller, 1996], is also expressed in the olfactory bulb, the rostral migratory stream and the subgranular zone (SGZ) of the dentate gyrus [Gloster et al., 1994;Coksaygan et al., 2006]. The generation of new neurons has been shown to persist in these regions postnatally [Altman and Das, 1966;Kaplan and Hinds, 1977], and in the case of the dentate gyrus progenitors located in the SGZ proliferate, migrate and differentiate becoming granule cells. These newly born neurons are thought to play an important role in a number of behaviours including spatial memory, fear conditioning and fear-related behaviours [Snyder et al., 2005;Saxe et al., 2006;Sahay and Hen, 2007]. A molecular defect in tubulin that affects the genesis of neurons could contribute to the abnormal spatial working memory and anxiety-based behaviour we observed in the Jna/+ mutant mice. Here we investigate this hypothesis.

Animals
Jna/+ mice and wild-type littermate male mice were maintained in a 12: 12-hour light:dark cycle at a temperature of 22 8 1 ° C with a humidity of 60-70%. Where possible, 5 mice were placed in each cage, and animals had access to food ad libitum. The genotype of the animals was determined by PCR analysis, as previously described , and only littermates were selected for experiments. Tuba1a-LacZ mice (K6) were obtained from the Miller group [Gloster et al., 1994], and the presence of the transgene was confirmed by amplifying and s equencing with the following primers: TUBA1A_LACZ_F1 GGGGGAGAGATTACCTCATA, and TUBA1A_LACZ_R1 TGCGCAACTGTTGGGAAG. All experiments were performed in accordance with the UK Animals (Scientific Procedures) Act 1986.

Bromodeoxyuridine Labelling
For our developmental studies, bromodeoxyuridine (BrdU) was injected intraperitoneally at embryonic day (E)14.5, brains from mutants and wild-type littermates were harvested at postnatal day (P)0 (n = 3), drop fixed in 4% paraformaldehyde and then sectioned on a cryostat. For pulse labelling experiments, male mice (n = 9) aged approximately 12 weeks were injected with BrdU (50 g/g of body weight) on a single occasion, and perfused 24 h later with 0.9% NaCl, followed by 4% paraformaldehyde. For cell survival and differentiation experiments, 4 BrdU injections (50 g/g of body weight) were administered over 4 consecutive days at approximately 12 weeks of age (n = 7), and the animals perfused 28 days later. The brains were removed, postfixed for 4-6 h and then dehydrated in 30% sucrose. Then, they were sectioned coronally (40 m) on a sliding microtome (Leica) through the hippocampus and the sections stored at -20 ° C in antifreeze.

Permanent Staining
One in 8 sections was mounted on polylysine-coated slides (VWR Scientific) for immunostaining. Citrate buffer (0.01 M ) antigen retrieval was performed, as previously described [Huang and Herbert, 2005]. For BrdU staining, the sections were then digested in trypsin (0.0125%) for 7 min at 37 ° C, washed 3 times in PBS, followed by a 30-min incubation in 2 N HCl at 37 ° C. Then they were incubated with the primary antibody overnight in 0.5% PBS/Triton with 2% sera at the following concentrations: BrdU (Accurate Chemical and Scientific; 1: 100); doublecortin (DCX, 1: 100; Santa Cruz) or Ki-67 (Vector; 1: 3,000). The next day, following several washes in PBS, biotinylated secondary antibodies were applied for 2 h in 0.5% Triton with 2% serum for 2 h (Vector; 1: 200). Staining was visualised by an avidin-biotin peroxidase system (Vectastain ABC kit; Vector Laboratories) and diaminobenzidine (Sigma).

Quantification
Images of one side of the dentate gyrus were captured on a TE2000 inverted microscope (Nikon) and then analysed using ImageJ (NIH). The total number of cells and the number of ectopic cells were counted in each section spaced 320 m apart. A cell was deemed to be ectopic if it was observed within the granule cell layer (GCL). Total cell counts were determined by multiplying by 8 (for each section) and then doubled to account for both hemispheres. All cell counting was performed by J.C., who was blind to the genotype of the animals. For cell survival and differentiation studies, images were captured on a Zeiss LSM 510 confocal microscope, and a total of 30 BrdU-positive cells were analysed for coexpression with NeuN for each animal. To determine whether differences between groups of animals were significant, one-way ANOVAs were performed. In circumstances where the data failed to meet the assumptions of normality and equality of variance, transformations were performed. For expression studies, at least 100 DCX or PROX1 cells were selected randomly and inspected for colocalisation with LACZ.

Tuba1a Expression in the Dentate Gyrus
Neurons born in the SGZ of the dentate gyrus express specific markers as they mature from radial glial-like progenitors (type 1: GFAP, nestin, PAX6 positive), to intermediate progenitor cells (type 2: TBR2 positive), to neuronal committed intermediate progenitors (type 3: NeuroD, DCX positive) to immature granule cells (type 4: calretinin, NeuN positive) and finally to mature granule cells (type 5: PROX1, calbindin positive) [von Bohlen und Halbach, 2007]. To investigate which of these cell types express Tuba1a, we utilised the Tuba1a-LacZ mouse line, which drives the expression of a LacZ transgene under the control of a rodent Tuba1a promoter. This mouse line has been widely used to define the expression pattern of Tuba1a and in general has been found to mirror the endogenous expression of Tuba1a mRNA [Gloster et al., 1994[Gloster et al., , 1999Bamji and Miller, 1996]. Consistent with previous studies, we observed LACZ staining in the pyramidal cell layers of the hippocampus (CA1, CA2, CA3), and in the GCL of the dentate gyrus ( fig. 1 a). Within the dentate gyrus we did not observe clear colocalisation when staining with GFAP or TBR2 (n = 50), indicating that Tuba1a is not expressed in radial glial-like progenitors or intermediate progenitors ( fig. 1 c, d). We were able to de-tect clear colocalisation when staining with DCX and NeuroD, suggesting that Tuba1a is expressed in type 3 progenitors ( fig. 1 e, f). However, as DCX and NeuroD expression persists in young postmitotic neurons, we additionally stained with an antibody against PH3, a marker for cells in the mitotic phase ( fig. 1 b). We did not observe colocalisation of LacZ with PH3 (n = 40), evidence that LacZ-positive cells, while being DCX and NeuroD positive, are not mitotically active. As expected, LacZ staining was also observed in postmitotic neurons (NeuN positive, fig. 1 g) and mature granule cells (PROX1 positive, fig. 1 h). Cell counting revealed that 23% of DCX-positive cells expressed LacZ (n = 100) (although LacZ staining in these cells was generally less intense than in those LacZ-

Normal Neurogenesis, Differentiation and Cell
Survival in Jna/+ Mutants Given that Tuba1a in the adult hippocampus is limited in expression to postmitotic neurons, the S140G mutation in the Jna/+ mouse should not affect the neurogenic potential. To ascertain whether this is the case, we performed BrdU pulse labelling in male Jna/+ mutants and wild-type littermates (n = 9), and quantified BrdU-positive cells by cell counting blind to the genotype of the animals ( fig. 2 a-c). We observed no significant difference in the number of BrdU-labelled cells between mutants and controls [F(1, 17) ! 1; p 1 0.1]. A drawback of employing BrdU to assess neurogenesis is its reliance on an intraperitoneal injection, and its uncertain uptake by target cells [Wojtowicz and Kee, 2006]. We therefore confirmed this result by staining serial sections with Ki-67, a protein that is present during the active phases of the cell cycle and is indicative of mitotic activity [Yu et al., 1992] ( fig. 2 e-g). Again, we observed no difference between those animals harbouring the S140G mutation and wildtype littermates [F(1, 17) ! 1; p 1 0.5]. As BrdU and Ki-67 fail to distinguish between newly born glia and neurons, we also employed antibodies for DCX [Brown et al., 2003]. Staining with DCX, followed by cell counting, again showed no difference in neurogenesis in Jna/+ mutants, [F(1, 19) ! 1; p 1 0.5] ( fig. 2 i- . 2 d, h, l). We then asked whether there was any difference in the ability of newly born cells to survive in Jna/+ mutants. To investigate this, we injected mutant and littermate controls with BrdU over 4 consecutive days, and sacrificed the animals 28 days later. Staining with a BrdU antibody revealed that there were fewer BrdU-positive cells in mutant animals, which is suggestive of increased cell death; however, this difference was not statistically significant [F(1, 13) = 2.2; p 1 0.1) ( fig. 3 a-c). We also asked whether the S140G mu- tation affects the ability of neurogenic precursors to differentiate into neurons. We employed double labelling, staining with sera for BrdU and NeuN ( fig. 3 d-f), and found that there was no significant difference in the percentage of NeuN-positive. BrdU-stained cells when comparing wild-type littermates (73%) and Jna/+ mutants [75%; F(1, 13) ! 1; p 1 0.5].

Disorganisation of the SGZ and GCL in Jna/+ Mutants
Our neurogenesis experiments showed that while the neurogenic potential in Jna/+ mutants is normal, ectopic neurogenesis is abundant. To investigate whether that is due to a disorganised SGZ, we stained adult Jna/+ mu-tants and littermate controls (n = 3) with GFAP to label radial glial progenitors ( fig. 4 a-d). In wild-type controls, GFAP-positive cell bodies were located in the SGZ, with fibres that extended perpendicularly into the GCL. In Jna/+ mutants the glial framework was present; however, some GFAP-positive cell bodies were located within the GCL, and the processes appeared less orthodox. To investigate this phenotype further, we employed sera to label NeuroD-positive cells that are normally located at the boundary of the GCL [von Bohlen und Halbach, 2007]. In Jna/+ mutants, NeuroD-positive cells were observed within the GCL and hilus of the dentate gyrus, confirming disorganisation of the SGZ ( fig. 4 i-l). Next we employed PROX1 staining to investigate the integrity of the Coronal sections of the dentate gyrus in Jna/+ mutants ( b , d , f , h , j , l ) and littermate controls ( a , c , e , g , i , k ) when stained with PROX1 ( a-d , i , j ) and TBR2 ( e-h , k , l ) at P10 ( a-h ) and P4 ( i-l ). The panels on the right ( c , d , g , h ) show high-magnification images of the GCL in panels a , b , e and f . PROX1-positive granule cells are dispersed in Jna/+ mutants at P10 ( b ), and more severely at P4 ( j ). There is a notable presence of PROX1-positive cells in the hilus of mutant animals at P4 ( j , arrow). TBR2 staining at P10 ( e-h ) shows positively stained cells scattered throughout the GCL, and in the molecular cell layer in Jna/+ mutants ( f , h ). At P4, a trail of TBR2-positive cells ( l , arrow) is observed along the subpial stream, suggesting a defect in migration in Jna/+ mutants. BrdU labelling at E14.5, followed by sacrificing at P0 ( m , n ), confirms a migration phenotype. In Jna/+ mutants ( n ), BrdU-positive cells form a concave cluster in comparison to the clearly discernable suprapyramidal blade in wild-type controls ( m ). Scale bars = 100 m ( l , n ) and 20 m ( h ).

The Disorganised SGZ and GCL Are due to a Defect in Migration during Development
To ascertain whether the observed granule cell dispersion and disorganisation of the SGZ was the result of abnormal developmental processes or defective neuronal migration during adulthood, we examined wild-type and Jna/+ mutants at P4 (n = 3) when the radial organisation of the dentate cell layer is accomplished, and then again at P10 (n = 3) when it is further condensed and the neurogenic SGZ is in place [Li and Pleasure, 2007]. At P10, PROX1 staining revealed granule cell dispersion in Jna/+ mutants which was more severe than in the adult phenotype, which was notably worse at P4 ( fig. 5 a-d). At P4, both the suprapyramidal and infrapyramidal blades of the dentate gyrus lack their characteristic structure with a large number of PROX1-positive cells present in the hilus ( fig. 5 i, j). TBR2 staining of intermediate progenitors revealed the origins of the disorganised SGZ in Jna/+ mutants. In mutant animals, TBR2-positive cells were scattered throughout the GCL and were visible in the molecular layer, whereas in control animals most TBR2-positive cells were predominantly found at the helm of the GCL or in the hilus ( fig. 5 e-h). At P4, we observed a cluster of TBR2-positive cells located at the tip of the infrapyramidal blade in both Jna/+ mutants and wild-type controls in addition to a loose scattering throughout the hilus, GCL and molecular layer. While little difference was observed between mutants and wild-type controls in this respect, there was a striking trail of TBR2-positive cells along the subpial route in Jna/+ mutants, indicative of a defect in migration ( fig. 5 k, l). To investigate this further, we undertook a birth date labelling study, injecting BrdU at E14.5 (a time when both TBR2-positive progenitor cells and dentate granular cells are proliferating in the dentate notch), and then sectioned and stained Jna/+ mutants and wild-type littermates at P0 with sera for BrdU. In wild-type animals a clearly defined suprapyramidal blade consisting of BrdU-positive cells was apparent, whereas in Jna/+ mutants we observed an underdeveloped concave cluster of BrdU-positive cells ( fig. 5 m, n). Taken together, these data point towards a defect in the migration of both TBR2-and PROX1-positive cells. The result is a disorganised GCL and SGZ of the dentate gyrus that progressively improves but is still apparent in adulthood.

Discussion
In this paper, we have investigated the expression pattern of Tuba1a in the dentate gyrus as well as the neurogenic potential and morphology of the dentate gyrus in mice with an S140G mutation in Tuba1a. Employing a Tuba1a-LacZ mouse, we have observed colocalisation of LacZ with NeuroD, DCX, NeuN and PROX1, but not with GFAP, TBR2 or PH3. This result indicates that Tuba1a is expressed in early born postmitotic neurons (type 4) and mature granule cells (type 5), but not in radial glial-like (type 1) or intermediate progenitors (types 2 and 3) undergoing cellular division. It is not clear why only a percentage of DCX-, NeuN-and PROX1-positive cells stained positive for LacZ, given that Tuba1a is generally considered to be panneuronal. One explanation could be that partial penetrance plays a role, which may be a result of the genomic insertion site of the transgene [Bamji and Miller, 1996].
Consistent with the absence of Tuba1a in mitotically active progenitors, we have shown that Jna/+ mutant mice exhibit similar levels of adult neurogenesis to wildtype mice when assessed by three different means (BrdU, DCX, Ki-67), with no significant differences in the ability of new neurons to differentiate. Newly born neurons are, however, more likely to be ectopic, which is a consequence of a disorganised SGZ that is accompanied by granule cell dispersion. These findings are similar to those described in the Lis1(+/-) mutant mouse, which models another form of dominant lissencephaly. Like the Jna/+ mouse, the Lis1(+/-) mouse exhibits no defects in the genesis of neurons in the adult hippocampus, and presents with granule cell dispersion in the dentate gyrus with evidence of ectopic neurogenesis [Wang and Baraban, 2007]. Granule cell dispersion has also been reported in the reeler mutant mouse [Drakew et al., 2002;Zhao et al., 2004], and it has been shown that Reelin is required for the correct formation of the radial glial scaffold, for the migration of PROX1-positive granule cells, and for the transition of neurogenic precursors from the subpial zone to SGZ [Frotscher et al., 2003;Li et al., 2009].
Like the reeler mouse, the disorganisation of the dentate gyrus in the Jna/+ mouse has its origins in development. In the Jna/+ mouse, we observed granule cell dispersion at P10, and mislocalisation of PROX1-positive cells in the hilus at P4. In addition, TBR2-positive progenitors still line the subpial route at P4, and TBR2-positive cells are scattered throughout the GCL at P10. These observations suggest a defect in the migration of both these cellular populations. Neurogenic precursors (both nestin and TBR2 positive) as well as the first granule cell neurons begin their migration from a region of the dorsomedial part of the telencephalic vesicles, known as the dentate notch, at around E13.5-14.5 [Li and Pleasure, 2007]. These cells migrate along the subpial stream before nestin-positive progenitors fan out and occupy the hilus (E17.5), the granule cells start forming the suprapyramidal blade (E18.5) and the TBR2-positive progenitors form the neurogenic subpial zone (E18.5). In due course, PROX1-positive granule cells are generated within the hilus, forming the infrapyramidal blade, and the TBR2positive cells migrate through the molecular cell layer, settling in the SGZ alongside the nestin-positive progenitors [Li et al., 2009]. In the case of the Jna/+ mice our results indicate that, in addition to defective migration of postmitotic PROX1-positive granule cells, both the ventricular-to-subpial and subpial-to-subgranular migration of TBR2-positive cells is impaired. Such migratory defects may also be responsible for the unusual cytoarchitecture of the dentate gyrus reported in humans with mutations in TUBA1A [Fallet-Bianco et al., 2008]. For in-stance, a 24-week-old fetus carrying an I238V mutation in TUBA1A has an abnormally shaped dentate gyrus, lacking a compact defined GCL. Similarly, a 22-week-old fetus carrying a P263T mutation in TUBA1A was found to have an undeveloped hippocampus, a near absence of granule cells and little evidence that the dentate gyrus had started to form [Fallet-Bianco et al., 2008].
In summary, our results indicate that Tuba1a is vital for the proper formation of the dentate gyrus, but that it is not essential for adult hippocampal neurogenesis. This result suggests that another of the six ␣ -tubulins is responsible for generating the tubulin heterodimers necessary for spindle formation and mitotic division in neurogenic progenitors. The identification and characterisation of this tubulin isoform will be of interest.