Pax7 is requisite for maintenance of a subpopulation of superior collicular neurons and shows a diverging expression pattern to Pax3 during superior collicular development

Background Pax7 encodes a transcription factor well-established as an important determinant of mesencephalic identity and superior collicular development. Pax7 mutant mice, however, present with no obvious morphological impairments to the superior colliculus. This finding is paradoxical and has been attributed to functional redundancy afforded by its paralogue Pax3. Here we utilise Pax7 mutant mice to investigate the precise role of this important developmental regulator during superior collicular development and neuronal specification/differentiation. We also assess its spatiotemporal relationship with Pax3 during embryonic development. Results Analysis of the superior colliculus of Pax7 mutant and wildtype mice at a variety of developmental timepoints revealed that whilst correct initial specification is maintained, a subpopulation of dorsal mesencephalic neurons is lost at early postnatal stages. Moreover, a comparative analysis of embryonic Pax3 and Pax7 expression profiles indicate that Pax3 expression overlaps extensively with that of Pax7 initially, but their expression domains increasingly diverge as development progresses, coinciding spatiotemporally with neuronal differentiation and maturation of the tissue. Furthermore, Pax3 expression is perturbed within the CNS of embryonic Pax7 mutant mice. Conclusion In summary, these results demonstrate that during superior collicular development, Pax7 is required to maintain a subpopulation of dorsal, mesencephalic neurons and partially regulates, spatiotemporally, Pax3 expression within the CNS. The differential nature of Pax7 and Pax3 with respect to neuronal differentiation may have implications for future stem cell therapies aimed at exploiting their developmental capabilities.


Background
It is evident that Pax7 is a multiplex contributor to correct CNS development. This is exemplified by dynamic spatiotemporal expression patterns, occurring from early development and persisting in restricted regions throughout adulthood. Pax7 expression initially occurs in the neural tube and mesencephalon from very early stages [1,2] and is required for polarisation of the dorsoventral axis of the neural tube [3] and specification of the superior colliculus/tectum from the mesencephalic alar plate [4][5][6][7]. In the developing superior colliculus, graded expression of Pax7 establishes rostrocaudal and dorsoventral polarity. Expression of Pax7 localises within superior collicular neurons as development proceeds. This expression is upregulated during retinal innervation and axonal arborisation but reduced in Pax6 (Sey) mutant mice [7], with reduced (20-30%) retinal innervation [8], confirming that Pax7-expressing cells are responsive to retinal input. Demonstrated colocalisation in superior collicular neurons with the mapping marker ephrin-A2 validates Pax7 participation in retinotopic mapping [7]. Continued, graded expression into adulthood is thought to maintain a small population of dorsal neurons in the mature colliculus [7,9], although the functional requirement for this feature remains obscure.
Given the aforementioned importance of Pax7 in mesencephalic and superior collicular development, the lack of gross defects in this region in Pax7 mutant mice is surprising, and points to rescue by the paralogous Pax3 gene which has overlapping expression domains [9,10]. Here, within the superior colliculus, we seek to determine the developmental role of Pax7 in specification of neurons, and assess its spatiotemporal relationship with Pax3. We have analysed Pax7 mutant mice [10] relative to wildtype at key stages of development and results indicate that a subpopulation of neurons is lost during early postnatal stages. We show that this apparent loss of neurons is not due to aberrant specification or proliferation, or cell-fate switching/transdifferentiation to the astrocytic lineage, but rather appears due to the inability of Pax7 mutant mice to maintain a subpopulation of dorsal superior collicular neurons.
Furthermore, analysis of Pax3 expression in embryonic wildtype and Pax7 mutant mice indicates crossregulation between paralogues, and illustrates a functional divergence during superior collicular development. We propose that within the superior colliculus initial overlapping Pax3 expression ensures correct neuronal specification, and temporospatial separation of expression patterns leads to solitary expression of Pax7 during a critical period of neuronal maturation, which abrogates the ability of Pax3 to compensate, revealing the aberrant phenotype.

Pax gene expression patterns
The Pax7 antibody has previously been demonstrated to be suitable for use in mouse tissue by Western Blot analysis [11]. The Pax3 antibody has been tested in vitro and in vivo for specificity in recognition of mouse Pax3 by Western Blot analysis [12].

Pax7 expression in wildtype mice
Within the anlage of the tectum, at the earliest embryonic stage examined (E12.5), Pax7 expression occurs from the mesencephalic ventricular zone to the most superficial layer. Immunoreactive cells are most dense in the ventricular and subventricular zones, then decrease in the intermediate zone, which is larger rostrally compared to the caudal region at this stage (Fig 1a-b). Pax7 expression can also be detected within the subthalamus, pretectum, pons, and in the ventricular zones of the cerebellar primordium, pons and myelencephalon (4 th ventricle). This profile concurs with in situ hybridisation results reported previously at E13 [9], with the exception of expression detected at the 4 th ventricle. At E15.5, a large number of immunoreactive cells are detected (Fig 2a) in the presumptive superior colliculus whilst expression declines in the caudal tectum, reflecting the emerging distinction of the tectum into the inferior and superior colliculi. Pax7 + cells also recede from the mesencephalic ventricular zone, although remnants of expression can still be detected at the dorsal ventricular surface, with cells also noted at the ventral ventricular surface at certain mediolateral positions. Pax7 + cells can be detected up to the pial surface of the superior colliculus with the exception of the stratum zonale, which is now becoming evident (Fig 1c). Expression remains robust within the pons. Pax7 expression at E18.5 is similar to that noted at E15.5 (Fig 1e), however immunoreactivity is no longer detected within the ventricular zone and expression within the subthalamus is waning. At P5, while a rostral low to caudal high , ventral low to dorsal high gradient is maintained, Pax7 + cell numbers are reduced throughout the superior colliculus (Table 1; Fig  1g; Fig 2a,b,d,e; [7]). Cellular protein levels are similarly graded (Fig 2f-g; [7]). Expression is also reduced at the midbrain/hindbrain boundary, pons and subthalamus. Even fewer Pax7 + cells are detected throughout the young adult superior colliculus (P18.5) (Fig 1i; Fig 2a). Expression persists at the midbrain/hindbrain boundary and rostral to the choroid plexus (4 th ventricle), with weak expression in the subthalamus (n = 3-5 at all stages).

Pax7 expression in Pax7 +/mutant mice
In Pax7 +/embryos, the Pax7 expression profile is grossly indistinguishable from that of wildtype mice (Fig 1d,f); cell counts, wt = 2248.67 ± 94.75; Pax7 +/-= 2373.67 ± 206.88 (p > 0.05; n = 3; Fig 2a). However, at P5, there is a dramatic reduction in the number of Pax7 + cells in the most superficial region of Pax7 +/mice relative to those of wildtype mice (Table 1; Fig 1h; Fig 2b-e; n = 4 each). This manifests as a distinct area near the pial surface that is almost devoid of Pax7 + cells uniformly across the entire rostrocaudal (Fig 2d) and mediolateral (data not shown) axes. Moreover, the number of Pax7 + cells is reduced throughout the dorsal half of the superior colliculus. Although the reduction in the total number of Pax7 + cells in Pax7 +/relative to wildtype mice at P5 does not reach statistical significance, this phenomenon results in the loss of graded cellular distribution across the dorsoventral axis (Table 1; Fig 2e; p > 0.05). Concomitantly there is some variation in cellular protein levels ventrally creating a lack of graded expression medially (Fig 2g). Interestingly, at P18.5 we did not discern any differences in Pax7 expression between heterozygous and wildtype mice; cell numbers, wt = 828.5 ± 108.97; Pax7 +/-= 913.60 ± 69.77 (p > 0.05, Table 1 Pax3 expression in wildtype and Pax7 mutant mice At E12.5, in wildtype and Pax7 mutant mice Pax3 expression can be detected from the pretectum to the tectum in the ventricular and subventricular zones, with rostral expression lower in intensity relative to the mid-caudal region (Fig 3a-b). Immunoreactivity is also evident in the ventricular zones of the cerebellar primordium, pons and myelencephalon, and at the midbrain/hindbrain boundary. By E15.5 Pax3 expression within the superior colliculus of wildtype mice is primarily restricted to the dorsal ventricular zone, and the pons (Figs 3c,e). In Pax7 +/mutant mice at this stage, Pax3 + expression expands ventrally within the myelencephalon and pons (Figs 3d,f). In striking contrast to wildtype (Fig 3g) and Pax7 +/mice, Pax7 -/mice demonstrate large numbers of Pax3 + cells above the rostral ventricular zone (3 rd ventricle) (Fig 3h). Furthermore, Pax3 + cells can be detected at the ventral ventricular surface in Pax7 mutant mice but not in wildtype mice (Figs 3i-j). At E18.5, in wildtype (Fig 3k) and heterozygous mice the pretectum and ventricular zone of the superior colliculus remain Pax3 + , although only a remnant of expression within a small number of weakly stained cells remains within the superior colliculus and in the pons. By contrast, Pax3 immunoreactivity cannot be detected within the ventricular zone of Pax7 -/mice

Spatiotemporal assessment of neuronal proliferation and differentiation within the superior colliculus
To assess whether the loss of Pax7 + cells in the dorsal superior collicular region of Pax7 mutant mice was due to altered proliferation we explored Ki67 expression at E12.5. We could not detect any variation in expression patterns between wildtype and mutant mice (data not shown), indicating that cellular proliferation proceeds normally, thus excluding neuronal precocity at this stage (n = 3-4).
We next sought to relate observed differences in the Pax7 and Pax3 expression profiles to temporal differences in neuronal differentiation. We have shown previously that Pax7 co-localises with βIII-tubulin and ephrin-A2 in the mouse superior colliculus at P5 [7]. We therefore used these and other markers of early neuronal differentiation, α-internexin and doublecortin as well as the mature neuronal marker, NeuN to analyse neuronal differentiation in wildtype and mutant mice.
The efficacy of βIII-tubulin to discriminate differentiating neurons and tract formation in the mouse embryo has been demonstrated previously [13]. At E12.5 we observed βIII-tubulin + cells predominantly located within the intermediate zone at the pial surface of the tectum, identifying the emerging stratum profundum (SP) (Fig 4a) containing the first differentiating neurons of the superior colliculus [14]. Immunolabelled cells were also observed further ventrally at the rostral tectum, but were reduced in number throughout the mid-caudal region, indicating a rostral to caudal cellular maturation, with the occasional immunoreactive cell seen in more ventral regions.
At E15.5 the SP is clearly visible in the ventral tectum, and immunolabelled, horizontally-oriented processes are now visible extending throughout the dorsorostral superior colliculus (see Additional file 1). These processes extend more caudally at E18.5 and most likely represent the incoming axons of retinal ganglion cells, the major Pax7 expression profiles in the developing superior colliculus of Pax7 mutant and wildtype mice   afferent projection to the superficial superior colliculus [15].
When we compared βIII-tubulin expression in the embryonic superior colliculus of wildtype mice to that of Pax7 mutant mice we could not discern any differences in expression profiles (Fig 4a-b) (n = 2-3 at all embryonic stages).
Similar results were obtained at E12.5 for α-internexin ating [17,18] neurons, respectively. Results confirm that in mutant mice (n = 3 each) at E12.5, neuronal differentiation occurs normally within the intermediate zone close to the pial surface in a rostral high to mid-caudal low manner.
During all embryonic stages investigated, we could not detect any difference in the cellular distribution of NeuN + within the superior colliculus of Pax7 mutant mice relative to that of wildtype mice (n = 2-3). The tectum of wildtype mice at E12.5 contains a small number of postmitotic, NeuN + cells within the intermediate zone, distributed in a rostral high to caudal low manner (Fig 5a-b). At all other embryonic stages examined, NeuN + cells are distributed throughout the superior colliculus from ventral to pial surfaces (Fig 5c-d).
By contrast, at P5, the most superficial region of the superior colliculus of Pax7 mutant mice is almost completely devoid of NeuN + cells (Fig 5f) [15], would be expected to impact on superior collicular polarity and retinotopic mapping. We therefore assessed ephrin-A2 expression in Pax7 mutant mice relative to expression in wildtype littermates. Consistent with the altered Pax7 and NeuN profile, at P5 ephrin-A2 + cells are missing in the most superficial region (Fig 5h,j), in contrast to wildtype expression displaying immunoreactivity close to the pial surface (Fig 5g,i), providing further evidence of neuronal loss and changes to polarity.

Neuronal loss is not due to astrocytic cell-fate switching or transdifferentiation
To assess the likelihood of astrocytic cell-fate switching or transdifferentiation, we analysed GFAP expression in postnatal stages as the majority of gliogenesis occurs within the first postnatal week [19]. We could not detect co-expression of Pax7 and GFAP at P5 or P18.5 in either wildtype or Pax7 mutant mice, consistent with previous results determined in the chick tectum [20]. GFAP + processes can be seen extending from cells located at the pial surface, however the dorsal regions of the superior colliculus in Pax7 mutant mice are similar to wildtype mice and are not populated by GFAP + cells within the region of per-turbation at P5 or P18.5 (Fig 6a-d), indicating that cell fate switching or transdifferentiation to the astrocytic lineage is not responsible for the reduced dorsal expression of Pax7, NeuN and ephrin-A2 at this stage (n = 3-4).

Neuronal loss may be attributable to cellular regression
To evaluate whether neuronal loss was due to apoptosis, we investigated activated caspase 3 expression from E15.5 to P5, however we did not detect either co-expression with Pax7 at any stage, or increased numbers of cells in mutant mice relative to wildtype (data not shown). This marker, however, proved difficult to analyse as elevated and widespread immunoreactivity detected at the earlier embryonic stages tested may be due to the emerging role of caspase 3 in neuronal differentiation [21]. We therefore utilised Hoechst as an indicator of pyknotic nuclei, and assessed co-expression with activated caspase 3 as confirmation of the apoptotic status of the cells, but only a few apoptotic cells were detected at any given time point. The identification of apoptotic cells would, however, require knowledge of the exact timeframe in which the neurons were lost, due to rapid neuronal degeneration noted previously [22].

Mesencephalic boundary formation
To explore the effects of perturbed Pax7 expression on formation of superior collicular boundaries, ie between the diencephalon and mesencephalon and between the superior colliculus (dorsal) and tegmentum (ventral mesencephalon), we assessed Pax6 expression at the rostral and ventral boundaries of the superior colliculus [4,9,23]. At E12.5, there are a few weakly-stained Pax6 + cells superior to the interpeduncular fossa. At E15.5 in wildtype mice, Pax6 + cells can be detected dorsally in the pretectum (Fig  6e,f) sometimes spreading to the rostral margin of the ventricular zone. Pax6 immunoreactivity could not be detected within the pretectum at E18.5 in either wildtype or mutant mice. We could not detect any abnormality in the expression domain of Pax6 in Pax7 mutant mice at these midgestational stages (n = 3).
To investigate alterations to mesencephalic/metencephalic boundary formation, we assessed the expression of the marker En-1 [24] from E15.5 to P18.5. En-1 + cells can be detected at the midbrain/hindbrain boundary, from the interpeduncular fossa to the rhombencephalic isthmus (Fig 6g,h) and expression wanes as development proceeds until P18.5, when En-1 expression in the midbrain is usually restricted to a cluster of cells superior to the interpeduncular fossa. Throughout all stages examined, we could not detect any difference between wildtype and mutant mice (n = 2-3).
Comparative NeuN and ephrin-A2 expression profiles in the superior colliculus of Pax7 mutant and wildtype mice Comparable astrocytic profile in the dorsal superior colliculus of wildtype and Pax7 mutant mice Figure 6 Comparable astrocytic profile in the dorsal superior colliculus of wildtype and Pax7 mutant mice. Immunofluorescent detection of GFAP indicates a normal astrocytic profile at P5 (a-b) and P18.5 (c-d) for wildtype (a,c) and Pax7 mutant mice (b,d). (a') Positive control indicating GFAP expression in astrocytes of the myelencephalon. GFAP+ processes extend from cells located at the pial surface, however the dorsal half of the superior colliculus in Pax7 -/mice, like that of wildtype, is not populated by GFAP+ cells. Therefore, cell fate switching and/or transdifferentiation towards the astrocytic lineage does not account for the reduction in Pax7 + cells dorsally.

Discussion
In order to understand the role of Pax7 in superior collicular development we investigated the superior colliculus of Pax7 mutant mice relative to wildtype at a variety of developmental stages. Whilst both Pax3 and Pax7 are known to define the early tectum subsequent to mesencephalic determination, and the ability to form ectopic tectum subsequent to misexpression within the diencephalon and ventral mesencephalon testifies to their critical nature in specification of tectal identity [5], individual roles during further development have remained elusive.
Here we show that loss of Pax7 dramatically alters Pax3 expression domains and disturbs the neuronal profile in the dorsal layers of the developing superior colliculus.

Pax7 and Pax3 have separate roles during neuronal differentiation in the superior colliculus
Since Pax7 and Pax3 have similar expression domains in the developing superior colliculus at mid-gestation [9], we undertook a thorough, comparative analysis of their expression patterns in wildtype and mutant mice. It was anticipated that this analysis would highlight any changes caused by loss of Pax7 and shed light on their individual roles during superior collicular development.
Results indicate that as the superior colliculus emerges and the dorsal laminae are formed by cellular migration and differentiation, Pax7 expression becomes progressively more refined towards the upper, developing strata whereas Pax3 expression becomes increasingly restricted towards the proliferative region (Fig 7). At E12.5, Pax7and Pax3-expressing cells overlap in the ventricular and subventricular zones, but not in the emerging, differentiating intermediate zone.
At E15.5 tectal Pax7 expression within the ventricular zone decreases, and is no longer evident at E18.5, with expression persisting within the upper developing primary layers (strata profundum, intermedium and superficiale). However, Pax3 expression becomes restricted to the ventricular zone, with a few scattered cells within the intermediate region at E15.5 decreasing at E18.5. Postnatally, expression of Pax7 is detected throughout the superior colliculus, numerically declining with maturation yet persisting at all developmental stages examined, whilst Pax3 expression is extinguished.
The divergent Pax3/Pax7 expression profiles, as demonstrated in this paper, suggest a dichotomy of function with respect to neuronal differentiation. The Pax7 expression profile, without co-expression of Pax3, appears spatially associated with neuronal differentiation and maturation of the superior colliculus, whilst Pax3 expression appears to be negatively associated with these characteristics. A role for Pax3 in maintaining the undifferentiated pheno-type has been demonstrated previously, both in vitro and in vivo for neuronal cells [25,26], for Schwann cells [27] and for melanoblasts [28,29]. Conversely, transfection of Pax7 into P19 mouse embryonal cells in vitro directs these cells along a neuronal pathway [30]. This dichotomy may, therefore, represent an important functional divergence between paralogues which may have implications for future stem cell therapies designed to treat midbrain disorders.
To investigate this further, we performed a comparative assessment of βIII-tubulin, α-internexin, doublecortin and  [31]. However, a more thorough investigation at closer temporal increments from E13 would be required to thoroughly preclude this possibility, as precocious or more rapid neuronogenesis may occur between the time frames investigated in this study. Furthermore, the perturbation to Pax3 expression within the ventricular zone at the later embryonic stages investigated, suggests that the early processes of specification and/or migration may be accelerated somewhat with insufficient Pax7 levels. The unaltered proliferation/specification of tectal cells does not preclude a role for Pax7 in these processes per se, but may indicate compensation via Pax3.
The observed perturbations in Pax3 expression in Pax7 mutant mice may indicate that Pax7 acts to limit the expression domain of Pax3. This is supported by the rostral, ventral and caudal expansion of Pax3 + cells in the presence of reduced or absent Pax7 expression. These regions of expanded expression occur where Pax7 expression would normally be encountered, and suggests that Pax7 acts to dorsalise (or at least positionally constrain) Pax3 expression. A cell autonomous relationship between Pax7 and Pax3 has been determined whereby misexpression of one paralogue represses expression of the other, and a balancing mechanism may exist to produce correct total expression levels [5]. The exact mechanism behind this relationship is ambiguous, however it may be significant that the expression of Pax7 precedes that of Pax3 within the headfold and primitive fold ectoderm in the developing chick [2]. Moreover, it is clear that this relationship is also dosage-sensitive, as perturbations occur in the haploinsufficient state.  To understand the significance of this anomaly we reviewed previous studies, which indicate that the majority of superior collicular neurons are born throughout the period E11 to E13 [14,19] and distinct temporal and spatial migration patterns culminate in dictation of the correct cytoarchitecture of the developing superior colliculus (Fig 8). Cells generated at E11 initially migrate to populate the superior colliculus, and are subsequently divided dorsoventrally by cells generated at E13 [14]. The cells that eventually reside within the most superficial region of the superior colliculus undergo their final postmitotic division at E13 while in the ventricular/subventricular zones, and migrate to subdivide their predecessors in the intermediate zone at E15. Thereafter, a subpopulation assume their final position superficially at E17 [14] which coincides with retinal ganglion cell innervation of both superficial and intermediate zones [15,32]. The stratum superficiale thus contains cells generated at both E11 and E13, with the latter cells being more superficially placed. In contrast, the ventral regions of the superior colliculus, associated with auditory and motor responses, undergo earlier development than the superficial regions and are thus populated by earlier-born neurons [14,33,34]. This temporal specification of cells is similarly observed in the chick tectum, where two populations of cells are generated, each with distinct laminar fates, with later born neurons expressing Pax7 fated to reside in the more dorsal laminae [35]. It is therefore evident that the timely migration of cells from the ventricular zone contributes to the correct placement of cells within the superficial laminae, and this temporal-based specification of neurons ensures that placement of cells within laminae coincides with their maturation and initiation of circuitry.

Role for Pax7 in timing of neuronal specification
Overall, our results tend to suggest that Pax7 may partially contribute to the correct timing of neuronogenesis by regulating migration of Pax3 + cells from the ventricular zone into the mid-collicular region in a timely and orderly fashion, a characteristic that is lost in Pax7 -/mice between E15.5 and E18.5, at the time when wildtype cells destined to reside in the more superficial strata are migrating and assuming their final destination.

A putative anti-apoptotic role for Pax7 in postnatal superior collicular development
As earlier quantification did not detect any numerical changes in the Pax7 cellular profile in Pax7 +/mice, it became apparent that these cells had either altered their differentiation profile or had regressed within the first postnatal week. We assessed GFAP expression postnatally and as there was no change in the cellular distribution of GFAP + cells in Pax7 mutant mice we excluded cell fate switching or transdifferentiation to the astrocytic lineage as a causative factor for the reduced number of Pax7 + cells dorsally. We were, however, unable to detect increased apoptosis within the superior colliculus using activated caspase 3, cresyl violet or Hoechst staining. A more detailed examination commencing from birth would be required to determine the exact time interval when perturbation occurs in order to conclusively demonstrate apoptotic mechanisms. Apoptotic labelling methods such as TUNEL or caspase 3, concomitant with Pax7 immunolabelling, would be required to address this issue, however rapid neuronal degeneration coupled with a progressive rostrocaudal maturation to the tissue which is likely to result in a punctuated apoptotic phenomenon, will likely impede the quantitative nature of the analysis required to differentiate between normal (wildtype) and increased (mutant) apoptosis. However, the loss of neurons evi-A diagrammatic representation of the primary layers of the developing superior colliculus regions. At E15, cells generated at E11 populate the SS and SP, whereas cells generated at E13 intersect these regions to populate the SI, with a subsequent migration of a subpopulation upwards to the SS by E17. Thus, the SS consists of two different populations of cells, with the most recently generated cells residing in the more superficial regions. Adapted from [14]. Abbrev. SS, strata superficiale; SI, intermedium; SP, profundum. "Cells Migrated" column relates to the developmental timepoint when the cells have migrated to their final destination.
denced in Pax7 mutant mice indicates a loss of cells, rather than simply extinction of Pax7 expression.
Relating phenotype to the specification of cells during superior collicular development [14] suggests that the second population of neurons, generated at E13, may be less capable of long-term survival than their earlier-produced counterparts, and as such are sensitive to gene dosage ( Fig  9). Whilst the differential placement of late-born neurons identified in previous studies [14,35] coincides spatially with the perturbed neuronal profile of Pax7 mutant mice further experimentation is required to distinguish a link between the generation of subpopulations of cells and their relationship with Pax7 and postnatal cytoarchitecture. Furthermore, differential sensitivity to neuron death has been suggested for different functional populations of neurons within the superior colliculus [36]. Anti-apoptotic roles for Pax genes are well-documented to-date [37][38][39][40][41] and, importantly, the ciliary neurotrophic factor receptor has recently been demonstrated as a downstream target of Pax7 within the mouse embryo [42].
Apparent neuronal recovery at P18. 5 We have demonstrated altered Pax7 + and neuronal profiles in Pax7 mutant mice at P5, however at P18.5 we can no longer detect these variations between Pax7 mutant and wildtype mice. Whilst it is feasible that the similar number of Pax7 + cells in Pax7 +/mice relative to wildtype at P18.5 occurs because Pax7 + cells in the adult do not normally include the subpopulation of cells which are missing at P5 in Pax7 mutant mice, the apparent recovery of the neuronal profile by P18.5 has proven enigmatic. We have drawn from established studies of CNS and superior collicular development to address this issue.
It is clear that early postnatal stages of superior collicular development are tumultuous. The colliculus becomes invaded by retinal afferents just prior to birth and this continues within the first postnatal week, with formation of collaterals giving rise to axonal arborisation and commencement of synaptogenesis [15,43,44]. During this process, a dramatic transformation occurs in which almost half of all fibre bundles are removed from the upper half of the stratum superficiale (the region including the perturbation where Pax7 + cells are essentially absent) [15,44]. This process coincides with naturally occurring neuron death in the superior colliculus [36,45], indicating neuronal formation in excess of adult requirements. Excessive neuron formation is a recurrent theme in CNS development [22,46,47]. Therefore it is likely that excessive neuron formation masks the loss of the more susceptible, Pax7-deficient neurons, with sufficient cells remaining to satisfy functional requirements. Subsequently, displacement of cells due to fibre invasion and A diagram detailing differential placement of separately generated populations of Pax7+ cells at P5 Figure 9 A diagram detailing differential placement of separately generated populations of Pax7+ cells at P5. A diagram of Pax7 + cells in the superior colliculus of wildtype (a) and Pax7 mutant (b) mice at P5. Cells generated at E11 (red) and E13 (purple) show overlapping (medial) and unique (ventral [E11]/dorsal [E13]) expression patterns (a). We propose that cells generated at E13 show reduced capacity for long term maintenance in Pax7 mutant mice (b), generating a phenotype whereby a superficial region is absent of Pax7 + cells, and the region immediately ventral to this exhibits a reduction in the number of Pax7 + cells (refer Fig 2e). Abbrev. CC, cerebral cortex; IC, inferior colliculus.
the ensuring maturation of the tissue could possibly account for the apparent phenotypic equivalence of NeuN + cells noted at P18.5. It should be noted that functional studies have not been performed on Pax7 mutant mice to-date to assess visual acuity.

Further studies
A possible cause for the loss of superficial neurons in early postnatal stages could be altered circuitry. Neurons of the superficial superior colliculus communicate with the deeper collicular regions through superficial neuronal axons [48,49], or via dendrites from neurons of the deep layers [50] which in turn contain a variety of efferent projections from regions such as the cortex, retina, zona incerta (region surrounding subthalamus) [32,[51][52][53] and indirectly from the subthalamus via the substantia nigra pars reticulata or enteropeduncular nucleus [54][55][56]. Taking this into consideration, an interesting finding from this research is the persistent Pax7 + /Pax3expression profile of the subthalamus throughout all timepoints examined. Providing Pax3 is not expressed in this region at an earlier developmental stage than examined in this study, the subthalamus may represent a novel opportunity to further dissect the functional repertoire of Pax7 from that of Pax3. Additional work would therefore be required to characterize the subthalamus of Pax7 mutant mice to address this issue, as this important brain region is a target of deep brain stimulation to treat disorders such as Parkinson's Disease and epilepsy [56,57].
Further studies on the role of Pax7 in superior collicular development would benefit from investigation into the axonal projection and circuity of Pax7 + cells, which may be elaborated by the generation of Pax7/tau-labelled X Pax7 mutant mice. Cell lineage tracing experiments in Pax7 mutant mice would be required to trace the progression of cells generated at E11 and E13 to conclusively resolve the questions related to the early postnatal perturbation and the apparent neuronal recovery in the adult. Moreover, functional studies assessing the ability of Pax7 mutant mice to evoke a co-ordinated response to stimuli will be required to assess the integrity of the mature superior colliculus.

Conclusion
In summary Pax7, while not required for neuronogenesis or neuronal differentiation at early stages of superior collicular development, is absolutely requisite in a dosagedependent manner for long term maintenance of a subpopulation of dorsal mesencephalic neurons. This characteristic may well impart prophylactic properties to stem cells utilised in future replacement therapies to enhance treatment of neurodegenerative diseases of the midbrain. Furthermore, comparative expression analyses indicate a functionally divergent role for Pax7 and Pax3 during neuronal differentiation within the superior colliculus.

Immunohistochemistry
Whole embryo tissue from E12.5 mice was subjected to antigen retrieval by microwave heating in sodium citrate buffer pH6.0 prior to processing. All tissue sections were treated with 0.2% Triton-X100/PBS (10 min

Immunohistochemical quantification and analysis
Images were captured with a Leica DC300 camera attached to an Olympus BX41 microscope and analysed using Optimas 6.5 Digital Image Analysis software (Media Cybernetics, Bethesda, MD, USA). For cellular quantification at E15.5 and P18.5 images were captured within Optimas (100× magnification) and Pax7 + cells were marked and counted. For optical density and cell distribution measurements at P5, the microscope was equilibrated prior to analysis, and microscope and image software settings were standardised. Measurements of Pax7 immunostaining were obtained on a frame-by-frame basis (400× magnification), and encompassed the entire Pax7 + portion of the superior colliculus, producing a serial reconstruction of the region (Coggeshall and Lekan 1996). Each individual cell measurement was topographically replotted within Excel and values were normalised against immunonegative adjacent tissue. Results were subsequently analysed producing maps of mean cellular distribution, or graphed as either mean cellular protein level or mean cell number at each axis point. Standard errors are shown.

Statistical analysis (Table 1)
To assess variation between wildtype and mutant mice at E15.5, P5 and P18.5, we compared the number of Pax7 + cells in wildtype mice with those in Pax7 +/mice using paired t test (two-tailed). For determination of graded expression at P5, mean values at three points across each axis (rostral, mid and caudal, or ventral, mid and dorsal) were calculated by averaging cell numbers in two frames at each axis position. Graded expression across each axis was assessed by analysis of significant variation between axis points using the paired t test