Intellectual disability (ID) is a highly prevalent disorder that affects 1–3% of the population. The Aristaless-related homeobox gene (ARX) is a frequently mutated X-linked ID gene and encodes a transcription factor indispensable for proper forebrain, testis and pancreas development. Polyalanine expansions account for over half of all mutations in ARX and clinically give rise to a spectrum of ID and seizures. To understand how the polyalanine expansions cause the clinical phenotype, we studied mouse models of the two most frequent polyalanine expansion mutations (Arx(GCG)7 and Arx432-455dup24). Neither model showed evidence of protein aggregates; however, a marked reduction of Arx protein abundance within the developing forebrain was striking. Examining the expression of known Arx target genes, we found a more prominent loss of Lmo1 repression in Arx(GCG7)/Y compared with Arx432-455dup24/Y mice at 12.5 and 14.5 dpc, stages of peak neural proliferation and neurogenesis, respectively. Once neurogenesis concludes both mutant mouse models showed similar loss of Lmo1 repression. We propose that this temporal difference in the loss of Lmo1 repression may be one of the causes accounting for the phenotypic differences identified between the Arx(GCG)7and Arx432-455dup24 mouse models. It is yet to be determined what effect these mutations have on ARX protein in affected males in the human setting.
Aristaless-related homeobox gene (ARX) [NM_139058.2] (MIM 300382) is a member of the paired-type homeodomain transcription factor family with critical roles in development. ARX is an important disease-causing gene on the X-chromosome contributing to intellectual disability (ID) and epilepsy in males (1). ID is a complex debilitating condition of considerable medical importance, with as many as 1 in every 50 people affected worldwide (2). When the causative gene is on the X-chromosome, it is referred to as X-linked intellectual disability (XLID). Mutations in ARX, 1 of over 100 XLID genes (1), lead to a broad spectrum of ID and epilepsy phenotypes (3).
ARX is one of eight transcription factors in which expansions of polyalanine tracts cause hereditary diseases, many with neurocognitive phenotypes (4). Over half (63/114, 55%) of all reported ARX mutations leads to expansion of the first and second polyalanine tracts. The predicted mechanism of protein dysfunction for the autosomal dominant disorders due to expanded polyalanine tracts ranges from gain of function, complete or partial loss of function and even a dominant negative effect contributing to associated disease features (5). In contrast, both ARX and SOX3 are located on the X-chromosome and are subject to X-inactivation in females, making it difficult to ascertain if the expanded polyalanine tract mutations may be causing disease due to a dominant gain of function or due to altered or loss of function. In the case of ARX, a complete loss of function is unlikely given the mutations leading to complete loss of ARX function result in severe brain malformation phenotypes, including lissencephaly, hydranencephaly and agensis of the corpus callosum (3,6,7).
There is a common belief based on studies of some proteins with polyalanine expansions that these tracts, above a certain threshold, may induce mis-folding and aberrant protein interactions, degradation, mis-localization and likely aggregation under light microscopy (4,8–11). The polyalanine thresholds at which these events occur vary between proteins, but are a common finding in over-expression studies in routine and explant cell culture (4,8,9,12–14). However, the existence and the contribution of these aggregates to the pathogenesis of disease and ARX-related ID and epilepsy, in particular, remains to be demonstrated. Particularly, there are no data on in vivo aggregation of ARX protein available.
Currently, there is one mouse model for the most common ARX mutation, a polyalanine expansion (from 12 to 20 Ala) in the second tract (6). Moreover, two independent mouse models have been generated for the expansion (from 16 to 23 Ala) in the first polyalanine tract, the second most common ARX mutation (6,15). Patients with these mutations present with a highly variable phenotype, ranging from mild ID as the only feature to ID with early onset of epileptic seizures (Ohtahara Syndrome). Accordingly, these mouse models recapitulate many of the phenotypic presentations of human patients (6,15). However, at the molecular and cellular level, it is still unclear whether protein aggregation or protein mis-localization are the underlying drivers. Kitamura et al. (6) reported only nuclear localization of Arx in 12.5 days post-coitum (dpc) ganglionic eminence (GE)-originated migrating cells and in 18.5 dpc cortical interneurons in the two mouse models, Arx(GCG)7 and Arx432-455dup24, that recapitulate the respective polyalanine tract 1 (c.304ins(GCG)7) and tract 2 (c.429_452dup) mutations. In contrast, Price et al. (15) reported an increased Arx cytoplasmic localization in the cortical neurons from adult brains of their independently generated mouse model of the c.304ins(GCG)7 mutation. Neither study found support for expanded Arx polyalanine aggregate formation.
Patients with the ARX c.304ins(GCG)7 mutation present with severe ID and epilepsy compared with a milder phenotype in most patients with the c.429_452dup mutation (3). These differences are at least to some extent recapitulated in the mouse models of the two mutations with the c.304ins(GCG)7 model having elevated seizure susceptibility and profound learning impairment (6,15). Loss of interneuron subsets was shown in both Arx(GCG)7 mouse lines (6,15). However, the underlying molecular mechanism for this phenotypic variation remains to be determined.
We have performed comparative analysis of the brain development of Arx(GCG)7 and Arx432-455dup24 mouse models. We found no evidence of apparent Arx protein aggregation or protein mis-localization under light microscopy. However, we found a significant reduction of protein abundance for both Arx mutant proteins. Moreover, we demonstrated a more prominent loss of Lmo1 repression, a well-characterized Arx direct target, in Arx(GCG)7/Y compared with Arx432-455dup24/Y mice. We propose the reduced Arx protein leading to aberrant expression of Lmo1 as a potential molecular cause of the ARX polyalanine expansion mutations pathology and variability of clinical expressivity.
No evidence of protein aggregation in vivo
We performed comparative analysis of Arx(GCG)7 and Arx432-455dup24 mouse models (6) by studying 12.5 dpc Arx protein expression between the wild-type (WT) and the two mutant mouse models at the dorsal telencephalon (presumptive cortex), ventral thalamus (VTh), lateral ganglionic eminence (LGE) and the medial ganglionic eminence (MGE) by immunofluorescence (Fig. 1A and E). Arx protein was detected in cells residing at the mantle zone (MZ) but not the proliferative ventricular zone (VZ) of the LGE and MGE (Fig. 1A and E). Similar strong MZ Arx expression was also observed in the VTh and HTh. In contrast, a lower but uniform level of Arx was present within the VZ of the presumptive cortex and hippocampus (Supplementary Material, Fig. S1A, E, I).
Under light microscopy, we were unable to identify any apparent protein aggregation comparable with that suggested by our and other in vitro studies of Arx mutant proteins in Arx(GCG)7 and Arx432-455dup24embryos (12–14). We analysed over 800 Arx-positive cells, with ∼60% from GEs, ∼30% from VTh and 10% from dorsal telencephalon, from three independent brain samples per Arx(GCG)7 and Arx432-455dup24 genotype (n = 3) (Fig. 2A, A′, C, C′, E, E′). Normal expression pattern was also observed in Arx-positive cells within the cerebral cortex (CCx) and olfactory bulbs of two 18.5 dpc Arx(GCG)7/Y brains (data not shown). Together, these data suggest that the Arx(GCG)7 and Arx432-455dup24 proteins do not form apparent aggregates in vivo.
Reduction of Arx protein abundance in Arx(GCG)7/Y and Arx432-455dup24/Y mutant brains during neurogenesis
We noted a marked reduction of the Arx protein abundance within the developing telencephalon of the 12.5 dpc Arx(GCG)7/Y and Arx432-455dup24/Y embryos by immunofluorescence, including both GEs, VTh (Fig. 1 C and E; and Supplementary Material, Fig. S1 C, C′ G, G′) (n = 3) and the presumptive cortex (data not shown). Semi-quantitative western immunoblot demonstrated a marked reduction of Arx protein in Arx(GCG)7/Y (16–17%) (n = 3) and Arx432-455dup24/Y (8–50%) (n = 3) at 12.5 dpc (Fig. 3A and B showed two out of the three embryos), a stage of peak neural proliferation. This loss of Arx protein was maintained at 14.5 dpc, a stage of peak neurogenesis, 18.5 dpc and postnatal day 10 (P10) (Fig. 3C and D showed one out of two embryos for each time point). Thus, both the Arx(GCG)7/Y and Arx432-455dup24/Y mice modelling the polyalanine expansion mutations lead to similar reduction of Arx protein in the developing brain.
Interestingly, we noted a significantly lower body mass in postnatal Arx(GCG)7/Y and Arx432-455dup24/Y pups from P5 (25 pups for +/Y; 9 pups for Arx(GCG)7/Y and 18 pups for Arx432-455dup24/Y) (Fig. 3E and Supplementary Material, Fig. S2A–C). No apparent body size difference was detected between both mutant and WT embryos at 18.5 dpc (15 embryos for +/Y; 5 embryos for Arx(GCG)7/Y and 12 embryos for Arx432-455dup24/Y) (Fig. 3E and Supplementary Material, Fig. S2A), indicating that the weight difference is a consequence of retarded postnatal growth.
Reduced Arx protein abundance is independent of cell loss and as intrinsic to the expanded polyalanine mutations
To further investigate the mechanism responsible for the loss of polyalanine expanded Arx protein and its developmental consequences, we focused on the two regions with the highest WT Arx protein expression and hence, most striking reduction of Arx polyalanine expanded mutant protein, the GEs and the VTh. Despite some slight tissue disorganization, both Arx(GCG)7/Y and Arx432-455dup2/Y brains showed comparable mass and cellular density by Haematoxylin and Eosin stains in VTh and GEs where Arx protein expression is high (n = 3) (Supplementary Material, Fig. S3A–F and A′-F′). Cell count analysis also suggested that cell densities were invariable in the respective regions of LGE and MGE between +/Y (WT) and Arx(GCG)7/Y (n = 3) and Arxdup432-455dup24/Y embryos (n = 3) (MGE and LGE: Fig. 3F and G; VTh: data not shown). These data indicate that the reduction of Arx protein abundance in both Arx(GCG)7/Y and Arxdup432-455dup24/Y 12.5 dpc mouse brains was primarily due to the presence of polyalanine mutations rather than a secondary consequence of cell or tissue loss.
Loss of Arx protein abundance is not due to the reduction of Arx mRNA expression
We found no significant difference in Arx quantity between WT (+/Y) and the two mutants (Arx(GCG)7/Y and Arx432-455dup24/Y) in 12.5 dpc brains by semi-quantitative RT-PCR and RT-qPCR (n = 4, Fig. 4A and B) analysis, indicating no overall loss of Arx transcript. RNA toxicity from CAG-trinucleotide repeats has been well characterized in polyglutamine diseases (16). Potential cellular toxicity of polyalanine encoding GCG-trinucleotide repeats is unknown. In situ hybridization showed normal Arx spatial expression pattern within the 12.5 dpc forebrain, including the dorsal telencephalon, MGE, LGE and VTh from both Arx(GCG)7/Y and Arx432-455dup24/Y embryos (n = 2, Fig. 4C–F and J–L). Together, these data suggest the reduction in Arx protein levels in both mutant models is unlikely to be due to the loss of Arx mRNA or Arx-positive brain domains, but rather reflects a translational and/or post-translational event.
Differential regulation of Arx transcriptional targets in the Arx(GCG)7/Y and Arx432-455dup24/Y embryonic brains
Arx is well known to be a transcriptional repressor (3,13,14,17,18). We examined the impact on transcriptional regulation of characterized (Lmo1, Shox2, Ebf3, Kdm5c) and candidate (Gria1, Rab39b and Pax6) Arx target genes by Arx(GCG)7/Y and Arx432-455dup24/Y mutations. We undertook RT-qPCR analysis from 12.5 dpc whole brains and telencephalic vesicles (including both the pallium and the subpallium) of the 14.5 and 18.5 dpc WT and mutant embryos.
There was a significant increase in Lmo1 expression within Arx(GCG)7/Y 12.5 dpc whole brain (n = 4) (Fig. 5A) in comparison with that of Arx432-455dup24/Y or WT. In contrast, no significant difference in Lmo1 expression was observed between WT and Arx432-455dup24/Y. At 14.5 and 18.5 dpc, both mutant mouse models demonstrated a significantly higher than WT level of Lmo1 expression (Fig. 5B and C). Interestingly, the loss of Lmo1 repression was more severe in Arx(GCG)7/Y than in Arx432-455dup24/Y in 14.5 dpc embryos during neural development (Fig. 5A and B). In contrast, we found no overall changes in the expression levels of Gria1, Ebf3, Pax6, Kdm5c and Rab39b between WT and either mutant at any time point tested (Fig. 5A–C). We noted a small but significant reduction in Shox2 expression only when comparing Arx432-455dup24/Y mice to Arx(GCG)7/Y mice at 12.5 dpc (n = 4). However, we were not able to amplify Shox2 from WT or mutant telencephalic vesicles of 14.5 and 18.5 dpc embryos (Fig. 5D and data not shown), suggesting the repression on Shox2 expression by Arx was intact in the telencephalic vesicles of 14.5 and 18.5 dpc Arx(GCG)7/Y and Arx432-455dup24/Y embryos.
Loss of Lmo1 expression boundaries in both polyalanine mutant mouse models
We performed in situ hybridization of Lmo1 in 12.5 dpc brain sections from WT, Arx(GCG)7/Y and Arx432-455dup24/Y embryos to test whether there is an ectopic expression of Lmo1 in cells that were normally deprived of its expression. Lmo1 expression was restricted to the VZ of the VTh in WT and both mutants (Fig. 4H and I). A diffuse band of Lmo1 expression was observed in the presumptive cortex of WT 12.5 dpc embryos (Fig. 5M and Supplementary Material, Fig. S4A–C), with a similar expression pattern present in both mutant mouse models (Fig. 5N and O and Supplementary Material, Fig. S4D and E). In contrast, ectopic Lmo1 expression was present within the MZ of MGE, and, at a much less prominent level, within the MZ of LGE of 12.5 dpc Arx(GCG)7/Y embryos (n = 2, Fig. 4N). Despite the lack of significance in RT-qPCR, this expansion of Lmo1 spatial expression pattern was also observed in the Arx432-455dup24/Y embryos (n = 2, Fig. 4O). Hence, our in situ study further supported our RT-qPCR data of a loss of repression of Lmo1 expression in Arx(GCG)7/Y and Arx432-455dup24/Y embryos.
Previous literature (19) and our data indicate that normal Lmo1 expression is exclusive to the VZ, where neural precursors of MGE and LGE reside. Next, we asked whether mutant cells with ectopic Lmo1 expression display some progenitor aspects similar to the VZ precursors (Fig. 4P). Nestin is an intermediate filament found mainly in neural progenitors and phosphohistone-3 is a protein highly associated with mitotic cycles. No apparent differences in the density, length or orientation of nestin filaments were found between either mutant or WT embryos (data not shown). Similarly, the number of phosphohistone-3-positive cells within the MGE were comparable across WT and mutant embryos (data not shown). Together, these data suggest that cells ectopically expressing Lmo1 were not mitotically active nor expressing the progenitor filaments common with the VZ precursors.
Arx expression can be detected in the developing mouse brain as early as 9.5 dpc (20), suggesting a role in early neural development. Its expression then peaks in the forebrain, in particular within the subpallium at 12.5 to 14.5 dpc, which are time points critical for neural proliferation and neurogenesis. This Arx expression pattern persists until the end of embryogenesis and is downregulated during postnatal development (20,21). Hence, we focused our analysis during peak neural proliferation and neurogenesis initiation at 12.5 dpc. Our findings of an absence of apparent in vivo protein aggregation, but an intrinsic and cell-autonomous loss of polyalanine expanded Arx protein abundance are in accordance with the previous report (21). We did not find any difference in the capacity of our Arx antibody to detect WT or polyalanine expanded Arx proteins when expressed in a range of in vitro cell lines, including those with Arx protein aggregations (data not shown). Hence, the difference in protein abundance detected between the WT and mutant mouse models is unlikely to be a result of biased antibody affinity towards the WT Arx. Our in vivo data are in contrast with previous in vitro analyses (13,14) showing formation of polyalanine expanded and length-dependant ARX protein aggregates. This discrepancy may have at least two non-mutually exclusive explanations. First, the formation of aggregates may be dependent on protein abundance such that a physiological level of in vivo protein may not be sufficient to induce protein aggregation. Alternatively, it is plausible that polyalanine expansion mutations result in mis-folded protein and transient protein aggregates that are efficiently removed by in vivo cellular machineries, i.e. the proteasome complex. The latter would explain our observation of reduced Arx protein abundance in both mutant mouse models. In fact, in vivo protein reduction is common across a number of mice modelling polyalanine expansion mutations, i.e. Hoxa13 +10Ala, Hoxd13 +7Ala and Sox3 +12Ala (22–24). Together, these data suggest that the loss of protein, but unlikely mRNA, may be a common in vivo pathogenic hallmark of different by polyalanine expansion mutations in different genes.
To explore if there is a molecular explanation for the phenotypic variation between the Arx(GCG)7/Y and Arx432-455dup24/Y mice (6), we speculated that the phenotypic differences were a result of differential Arx target gene regulation between the two mutations. To test this, we have selected a subset of known and candidate Arx target genes. Lmo1, Shox2 and Ebf3 are well-characterized Arx target genes repressed by Arx in which regulation on these genes is compromised in the absence of Arx (17,18). Recent studies have demonstrated the binding of Arx to the regulatory elements of many other genes, among these genes previously implicated in epilepsy and ID, e.g. KDM5C, GRIA1 and RAB39B (25,26). In particular, activation of KDM5C transcription activity by ARX appears to be reduced by the introduction of polyalanine expansion mutations in vitro (25). While Arx and Pax4 mutually regulate each other in the specification of glucagon-producing and insulin-producing cells during pancreatic development (27), upregulation of Arx expression has been observed in the loss of forebrain Pax6 function, a Pax4 brain paralogue, in both mouse and zebrafish (28,29). Thus, we speculated that the Pax6 may retain similar interaction with Arx during brain development. We interrogated the transcriptional regulation of these selected well-known and candidate Arx target genes (18,26). Out of the seven genes selected, we detected a more severe loss of Lmo1 repression in Arx(GCG)7/Y in comparison with Arx432-455dup24/Y during peak neural proliferation (12.5 dpc) and neurogenesis (14.5 dpc), but were equally affected at 18.5 dpc when neurogenesis subsides. The temporal and quantitative Lmo1 expression differences, as seen in these mice models, are reminiscent of the differences in clinical presentations and ARX patients with the respective mutations. We did not identify any significant changes in the transcriptional regulation of the other six genes in either mouse model. We reason that the lack of in vivo support for these expression differences might be due to the in vitro settings of previous studies (25,30) or due to the temporal or cellular differences. In fact, a loss of Shox2 and Ebf3, but not Lmo1 repression by luciferase reporter in in vitro neurons differentiated from transfected embryonic stem cells with +8Ala in Arx polyalanine tract 1 has been reported by others (30). It is plausible that this discrepancy is a result of different Arx protein abundance in transfected in vitro differentiated neurons (30) versus in vivo neural progenitors in our study. Otherwise, this difference may be a result of the disparity in the addition of eight alanine residues in the transfected Arx mutant in vitro, but only an addition of seven alanines in the Arx(GCG)7 mouse model.
The most plausible explanation for the differential regulation of Lmo1 activity between the two Arx mutant mouse models, at least in our view, could be the difference in the amount of the Arx protein loss (16–17% in Arx(GCG)7/Y versus 8–50% in Arx432-455dup24/Y 12.5 dpc brains), which suggests a different, partial quantitative (and thus variable) loss of function. Moreover, similar ectopic Lmo1 expression pattern has been reported in the subpallium of 14.5 and 18.5 dpc Arx deficient mice, further supporting this proposition (17,18). Similarly, for Hoxa13 and Sox3, the partial loss of function secondary to protein reduction is proposed to be responsible for most of the phenotypic outcomes of the polyalanine expansion mutations due to an overlapping phenotype spectrum between the null and polyalanine expansion mutant mouse models (22,23). Nevertheless, since Lmo1 is directly repressed by Arx (18), the in vivo primary impact of polyalanine expansion mutations on transcription regulation by residual Arx should not be overlooked. In particular, both Arx(GCG)7/Y and Arx432-455dup24/Y mice displayed intact Shox2 and Ebf3 repression, indicating normal residual functions of these mutant proteins in the context of those genes (17,18).
In vitro studies have readily demonstrated that peptides with >15 alanine residues have higher propensity to form stable macromolecular β-sheets (31,32) and alter molecular folding. The functional disparity between Arx(GCG)7 and Arx432-455dup24 mutation could be a consequence of the positional difference between the two polyalanine expansion mutations, resulting in distinct conformation changes within the Arx protein, affecting the folding of different protein domains and ultimately their function. In fact, functional alteration by polyalanine expansion mutations in transcription factor Zic2 and Hoxd13 has been proposed as the main contributor to the pathogenic mechanisms (24,33). Thus, it is likely that the impact of Arx target gene de-regulation in Arx(GCG)7/Y and Arx432-455dup24/Y mice is a combinatorial outcome of a partial loss of function, possibly secondary to protein reduction, and an altered function of the residual mutant Arx protein.
The ectopic Lmo1 expression strongly suggests that those cells within the MZ of the MGE and LGE (mostly maturing interneurons) already have a change of ‘identity’ by 12.5 dpc in both mutant mouse models. However, our evidence of nestin and phosphohistone-3 expression suggested they do not share these two progenitor properties with the VZ precursor cells. Further studies will need to establish their lineages and the identity of these ‘supposing’ maturing interneurons and the impact of ectopic Lmo1 on their cellular fates.
Despite LMO1 being a well-studied oncogene that promotes proliferation in neuroblastoma cells and maintenance of self-renewal in T-cell lymphoblastic leukaemia (34,35), very little is known regarding a role in neural development. Given the function of LMO1 in cancer cells, and the restricted forebrain expression within the VZ, where proliferative neural progenitors reside, it would be reasonable to speculate that Lmo1 may have a pro-progenitor property in neural development. Lmo3 is a highly conserved and functionally redundant paralogue of Lmo1 and is directly bound by Arx within its regulatory region in 15.5 dpc mouse brain and neuroblastoma cells (26,36). Moreover, Lmo1, Lmo3 and Lmo4, a highly conserved Lmo family member important for neurulation (37), were all ectopically expressed within the subpallium of 14.5 Arx-deficient mice (17). It will be of interest to investigate if Lmo3 and Lmo4 are also ectopically expressed within the MGE and LGE of the Arx(GCG)7 and Arx432-455dup24 mouse models.
Arx(GCG)7 mutant mice displayed morphologically normal CCx but mild interneuron deficiency from the subpallium (6). Unlike the MGE and LGE, we found Lmo1 expression pattern was intact within the dorsal telencephalon of both mutant models. This observation is in agreement with a study demonstrating that the residual function of Arx(GCG)7 mutant protein was sufficient for a near complete rescue of cortical proliferation and migration in an Arx null mouse model (30). Similarly, there was no ectopic Lmo1 expression in the MZ of the VTh. Together, our results support the observation of a dynamic role for Arx in regulating target genes during dorsal (developing cortex) and ventral (GEs and VTh) forebrain development (30,38). Interestingly, a recent study has shown that Arx is required for the patterning and development of dopaminergic neurons in zebrafish VTh (39). Future work will need to address whether this role of Arx in the VTh is conserved in mammals and how polyalanine expansion mutations affect the development of this domain.
Polyalanine expansion mutations account for over half of mutations found in patients with ARX mutations. Unlike mutations that cause brain malformations, patients with polyalanine expansion mutations showed no gross morphological defects and suffer from varying degrees of ID with and without epilepsy (3). However, the molecular mechanism that underlies the pathogenesis of these polyalanine expansion mutations is not well understood. Here, we undertook comparative expression analysis of the mouse lines modelling the two most common ARX polyalanine expansion mutations. First, we did not find any support for in vivo Arx mutant protein aggregation in two mouse models studied. Secondly, we demonstrated a reduction of Arx protein abundance without loss of cells in both mutant mouse models, including postnatal growth retardation. We have identified a specific impact on Lmo1 gene expression in mutant mouse models. Overall our data lead us to propose a region- and mutation-specific loss of Lmo1 transcription regulation as one of the contributors to the more severe phenotypic symptoms observed in Arx(GCG)7/Y mice compared with Arx432-455dup24 mice, and possibly human patients with the equivalent mutations.
MATERIALS AND METHODS
Animals and tissue collection
All animal procedures were approved by the Animal Ethics committee of the University of Adelaide, the SA Pathology Animal Ethics committee and the Animal Ethics committee of the Women's and Children's Hospital, Adelaide. Five ArxGCG7/+ (BRC number: 03654) and four Arx432-455dup/+ (BRC number: 03653) heterozygote females were imported from RIKEN Bioresource Centre, Japan (6). Both mouse strains were maintained in the C57BL/6 background. Pregnant dams were euthanized by cervical dislocation followed by decapitation of pups or embryos. For RNA and protein extraction, tissues were snap frozen at −80°C. For frozen tissue sections, whole mount tissue was prepared as described previously (40) and sectioned at 10 μm thick using Microm HM505E.
Genotyping gDNA was extracted as per Maxwell® 16 Tissue DNA purification Kit manual (Promega). Genotyping PCR was performed using FailSafe™ PCR 2X PreMix J (Epicentre) as follows: 35 cycles of 30 s of 94°C for denaturation, 30 s of 60°C for annealing and 40 s of 72°C for elongation. Primers to amplify the Arx knock-in region were described in ref. (6). We also included Sry sexing PCR as part of our genotyping pipeline (sequences in Supplementary Material, Table S1).
Prior to Arx immunofluorescence, frozen tissue sections were air-dried for 1 h at room temperature. All procedures were performed in a humidified chamber to prevent drying of tissue sections. Tissue sections were permeabilized in 1×PBS + 0.5% Triton for 5 min, blocked with blocking solution (10% horse serum and 10% BSA in 1×PBS + 0.1% Triton) for 1 h at room temperature. Sections were then incubated in primary Arx antibody [rabbit anti-ARX at 1/500, (20,21)] in blocking solution at 4°C overnight. Sections were washed in 1×PBS + 0.01%Tween 20 for 10 min three times. They were then incubated at in 1/400 donkey anti-rabbit IgG Alexa488 secondary antibody (Life Technologies) at room temperature for 4 h. Sections were washed again in 1×PBS + 0.01%Tween 20 for 10 min three times prior to be immediately mounted with ProLong Gold Antifade Reagent with DAPI (Life Technologies). Mounted sections were allowed to cure in the dark at room temperature overnight prior to analysis and further storage at 4°C in the dark.
All images were analysed using Olympus IX81 inverted microscope equipped with CellSens 1.3 Software. Immunofluorescence images were acquired by Olympus XM10 black and white camera, while bright field images were captured using Olympus DP70 digital colour camera. All captured images were processed by Adobe Photoshop CS5.
Semi-quantitative RT-PCR and RT-qPCR
Collected tissues were homogenized with 21G needles and total RNA was extracted using Trizol (Invitrogen) and RNeasy Mini Kit (Qiagen) and treated with DNase I (Qiagen) according to the manufacturer's instruction. cDNA was prepared as described in SuperScript III reverse transcriptase (Invitrogen) manual with 1 μg of RNA primed by random hexanucleotides. Cycling conditions for RT-PCR are similar to the above genotyping protocols. RT-qPCR was performed as described previously (40). No template and no reverse transcriptase controls were included for product specificity. For Lmo1, Shox2 and Ebf3, reactions were prepared as described in Taqman® PreAmp Master Mix Kit user guide (Applied Biosystem). Expression values were normalized to reference gene Gapdh. Taqman® assays ID are: Mm00475438_m1 FAM (Lmo1), Mm00443183_m1 FAM (Shox2), Mm00438637 FAM (Ebf3) and Mm99999915_g1 VIC (Gapdh). For Arx, Gria1, Kdm5c, Pax6 as well as Lmo1, reactions were setup as described in SYBR® Green PCR Master Mix and RT-PCR Reagents Kit user guide (Applied Biosystem). Melt curve analysis was performed to ensure amplification efficiency. Expression values were normalized to reference gene Sdha. Analyses of Lmo1 expression at 12.5 and 14.5 dpc time point were performed using both SYBR® and Taqman® assay. Comparable results were obtained between the two assays while only values from SYBR® assay were presented in the Results section. Primers are listed in Supplementary Material, Table S1.
Protein extraction, SDS–PAGE and western blot
Dissected tissue was homogenized by 21G needles prior cell lysates extraction for western immunoblot as described previously (14,41). Primary antibodies used were: rabbit anti-Arx (1/700, (21)); and mouse anti-Actb (1/2000, Sigma Aldrich). Secondary antibodies used are: goat anti-rabbit IgG HRP (Dako); and goat anti-mouse IgG HRP (Dako). To quantify Arx protein abundance, the intensity histogram for each band/sample was obtained by ImageJ. Signal intensity was calculated by the area under the signal peak for each histogram. All values were normalized to WT samples within each blot. Percentages of Arx protein abundance were derived from normalized values.
Haematoxylin and Eosin stains
Haematoxylin and Eosin stains were performed as previously (40).
Protocols are as performed previously (40). Arx and Lmo1 sequences were amplified from cDNA cloned into pGEM®T Vector (Promega) prior to riboprobe synthesis (See Supplementary Material, Table S1 for primers). Riboprobes were synthesized by T7 after linearization with SpeI.
Cell counts were performed on DAPI stained nuclei. Cell density analysis was only performed on sections of anatomically comparable planes. For each anatomical domain, there is a high-density zone (made up of regions mostly with proliferating cells) and a low-density zone (made up with regions with cells that are low or have no proliferation). Each zone was arbitrarily divided into four bins of equal area. Cell count was performed twice in each bin as number of cells/mm2 and an average was taken. Then, the cell count of a zone was calculated by averaging the cell counts of the four bins. A schematic diagram of with zoning and binning strategy was presented as Figure 3F and G in Results section.
To perform statistical analyses for postnatal weight, cell counts and gene expression, the median and number of observations were given to the three groups, i.e. +/Y, Arx(GCG)7/Y and Arx432-455dup24/Y. A Kruskal–Wallis test was performed to establish whether there were any significant differences between +/Y, Arx(GCG)7/Y and Arx432-455dup24/Y. If a significant difference was found among all groups, pair-wise comparisons by Kruskal–Wallis tests were then carried out to determine the significance between each paired group.
This work was supported by grants from the National Health and Medical Research Council of Australia (Project Grant 1002732 to C.S.; Senior Principal Research Fellowship 1041920 to J.G.) and Australian Research Council (Future Fellowship FT 120100086 to C.S.).
We would like to thank Laboratory Animal Services Facility at the Women's and Children's Hospital (Adelaide) for their kind assistance. We appreciate the statistical help from Mr Thomas Sullivan, Ms Nancy Briggs and Ms Michelle Lorimer from Data Management and Analysis Centre at the University of Adelaide. We would like to thank Dr K. Kitamura, Ms M. Yanazawa and Riken BRC and Mitsubishi Chemical Corporation for kindly providing Arx(GCG)7/Y and Arx432-455dup24/Y KI mice.
Conflict of Interest statement. None declared.