Corticostriatal projection neurons (CStrPN) project from the neocortex to ipsilateral and contralateral striata to control and coordinate motor programs and movement. They are clinically important as the predominant cortical population that degenerates in Huntington's disease and corticobasal ganglionic degeneration, and their injury contributes to multiple forms of cerebral palsy. Together with their well-studied functions in motor control, these clinical connections make them a functionally, behaviorally, and clinically important population of neocortical neurons. Little is known about their development. “Intratelencephalic” CStrPN (CStrPNi), projecting to the contralateral striatum, with their axons fully within the telencephalon (intratelencephalic), are a major population of CStrPN. CStrPNi are of particular interest developmentally because they share hodological and axon guidance characteristics of both callosal projection neurons (CPN) and corticofugal projection neurons (CFuPN); CStrPNi send axons contralaterally before descending into the contralateral striatum. The relationship of CStrPNi development to that of broader CPN and CFuPN populations remains unclear; evidence suggests that CStrPNi might be evolutionary “hybrids” between CFuPN and deep layer CPN—in a sense “chimeric” with both callosal and corticofugal features. Here, we investigated the development of CStrPNi in mice—their birth, maturation, projections, and expression of molecular developmental controls over projection neuron subtype identity.
Corticostriatal projection neurons (CStrPN) are the cortical efferent neurons of corticobasal ganglia circuitry; their degeneration is a predominant feature to Huntington's disease (HD), and they are implicated in the pathophysiology of several neurological conditions including Parkinson's disease and multiple forms of cerebral palsy (Vonsattel et al. 1985; Sotrel et al. 1991; Albright 1996; Martin et al. 1997; Sieradzan and Mann 2001; Rosas et al. 2002; Reading et al. 2004; Stephens et al. 2005). A major population of CStrPN, termed intratelencephalic CStrPN (CStrPNi), projects to targets within the telencephalon (Reiner et al. 2003, 2010). CStrPNi have the unique attribute of being both corticofugal because they project to the striata bilaterally, and callosal because their axons cross the midline (Supplementary Fig. S1A). Adult CStrPNi have been well studied, and are considered to be the major corticostriatal population (Jones et al. 1977; Wise and Jones 1977; Donoghue and Kitai 1981; Royce 1982; Wilson 1986, 1987; Cowan and Wilson 1994; Wright et al. 1999, 2001; Reiner et al. 2003, 2010; Lei et al. 2004). In addition to CStrPNi, there is a small population of subcerebral projection neurons (corticospinal and related corticobrainstem neurons) (Arlotta et al. 2005; Lai et al. 2008) that have axon collaterals to the ipsilateral striatum (pyramidal tract-type CStrPN, CStrPNp) (Wilson 1987; Sheth et al. 1998; Reiner et al. 2003; Lei et al. 2004) (Supplementary Fig. S1B). Despite their significance, very little is known about CStrPNi development.
We applied a range of developmental analyses to investigate the generation, axon outgrowth and projections, target innervation, molecular development, and pruning of CStrPNi in mice. We first investigated the temporal range of birth of CStrPNi; we report that CStrPNi in mice are born predominantly during the initial infragranular phase of neocortical development (embryonic day [E]12.5–E14.5). We applied both anterograde and retrograde axonal projection analyses; we find that the earliest that CStrPNi can be distinguished from pure callosal projection neurons (CPN; neurons having only contralateral cortical targets) is when they first invade the contralateral striatum at around postnatal day (P)3–P4. At this stage, but not before, at least some CStrPNi simultaneously project to the ipsilateral striatum. We investigated the mediolateral and rostrocaudal distribution of CStrPNi through postnatal development until their stabilization; by 2 weeks after birth, there is substantial loss of CStrPNi from caudal neocortex, and the distribution in mice reaches a pattern similar to that of adult rat CStrPNi (McGeorge and Faull 1987, 1989). We applied analysis of recently identified molecular controls over CPN and corticofugal projection neurons (CFuPN; neurons that send their axons away from the neocortex) as indicators of alternative differentiation pathways; our data indicate that CStrPNi have molecular features of both CPN and CFuPN. For example, at P4 we find that, while CStrPNi express Satb2, a critical molecular regulator of CPN development, many also express Sox5, a molecular control over sequential generation of CFuPN subtypes, in striking contrast to pure CPN. These results indicate that CStrPNi possess dual callosal and corticofugal anatomic and molecular characteristics, motivating further work identifying molecular controls over the development of this unique, functionally and clinically important population of projection neurons.
Materials and Methods
All mice used in these experiments were handled according to guidelines of the National Institutes of Health (NIH), and all procedures were conducted with approval of the Institutional Animal Care and Use Committee (IACUC) of Massachusetts General Hospital.
CStrPNi Birthdating Analysis
The timing of CStrPNi birth was assessed through bromodeoxyuridine (BrdU) pulse labeling. Briefly, timed pregnant females (day of vaginal plug is taken to be embryonic day [E] 0.5) at E10.5, E11.5, E12.5, E13.5, E14.5, E15.5, and E16.5 were pulse labeled with BrdU (100 mg/kg body weight) by a single intraperitoneal injection. Retrograde labeling protocols were subsequently performed on the pups from these litters at P12 to label CStrPNi, as described below, and brains processed for analysis at P14. Results are from at least 3 different mice from at least 2 independent litters, and bars representing standard error of the mean are shown (Fig. 2A).
Retrograde labeling was performed as previously described (Fricker-Gates et al. 2002; Arlotta et al. 2005). Postnatal day (P)3 pups or younger were anesthetized with 4 to 5 min of hypothermia, while pups of age P6 and older were deeply anesthetized via an intraperitoneal injection of 0.015 cc/g body weight of Avertin (1.25% of 2-2-2 tribromoethanol in a solvent containing 0.63% isoamyl alcohol by weight in ddH2O). For specific, distinct experimental goals, a number of retrograde labels, 20–40 nL each, were used: 1) red and green fluorescent microspheres (Lumafluor); 2) FluoroGold (FG; Fluorochrome), and 3) Alexa 555- or Alexa 647-conjugated β-subunit of the cholera toxin (CTB555 and CTB647 respectively; Invitrogen). Microspheres/CTB/FluoroGold injections were performed using an ultrasound-guided microinjection system (Vevo 770, Visual Sonics, Toronto, Canada) for pups younger than P3, and stereotaxically (Stoelting, IL) for older mice 24–48 h before perfusion and brain collection. Fezf2- lacZ mice were generated by Hirata et al. (2004).
Mice were deeply anesthetized with a lethal dose of anesthetic (Avertin) or hypothermia, and perfused transcardially with phosphate-buffered saline (PBS), followed by 4% paraformaldehyde. Brains were removed, postfixed overnight in 4% paraformaldehyde, rinsed with PBS, and sectioned coronally at 50 µm on a vibrating microtome (Leica). Sections were blocked in 8% donkey or goat serum, 0.3% BSA (Sigma), and 0.3% Triton x-100 (Sigma) for 1 h at room temperature before incubation with primary antibodies overnight at 4°C. Primary antibodies and dilutions used were 1) rat anti-BrdU (1:500; Accurate Chemical and Scientific, NY); 2) rabbit anti-Cux1 (1:200; Santa Cruz Biotechnology, CA); 3) goat anti-Lmo4 (1:200; Santa Cruz Biotechnology, CA); 4) rat anti-Ctip2 (1:500; Abcam, MA); 5) goat anti-Sox5 (1:200; Santa Cruz Biotechnology, CA [discontinued]) or rabbit anti-Sox5 (1:500; Abcam, MA); 6) goat anti-cholera toxin, β-subunit (1:4000; List Biological Labs, CA); and 7) rabbit anti-β-galactosidase (1:3000; MP Biomedical). Before applying the blocking solution in BrdU immunocytochemistry (ICC) protocols, sections were immersed in 2 N HCl for 2 h at room temperature, and rinsed 3 times in PBS for 5min each. We used anti-β-galactosidase ICC on retrogradely labeled Fezf2-lacZ heterozygous brain tissue to investigate whether CStrPNi express Fezf2. Alexa fluorophore conjugated secondary antibodies from Invitrogen were used at a dilution of 1:500.
Anterograde Labeling with DiI Crystals
DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate [DiIC18(3)]) crystals (Invitrogen) were used to anterogradely label CStrPNi projections from the cortex. Crystals were placed into postfixed brains of P0, P2, P3, and P4 mice. Brains were incubated at 37°C for 1–2 weeks in PBS to allow for anterograde transport of dye along axons innervating the contralateral striatum.
Visualization and Analysis
Image acquisition was performed using either: 1) a Nikon Eclipse E1000 epifluorescence microscope with a QImaging Retiga EX cooled CCD digital camera (QImaging Corporation, Surrey, Canada); 2) a Nikon 90i epifluorescence microscope with a Clara DR-328G cooled CCD digital camera (Andor Technology, Belfast, Northern Ireland); or 3) a BioRad Radiance 2100 Rainbow laser-scanning confocal microscope based on a Nikon E800 microscope. Images were assembled in Adobe Photoshop and Illustrator (CS3), with adjustments for contrast, brightness, and color balance to obtain optimal visual reproduction of data.
Retrograde Labeling of Intratelencephalic Corticostriatal Projection Neurons
We developed a reproducible approach to retrogradely label (intratelencephalic CStrPN, CStrPNi) by injecting tracers in the dorsolateral sector of the striatum (Fig. 1A). Using DAPI staining to better visualize the laminar architecture (Fig. 1B), we identified the location of CStrPNi (Fig. 1C–D). As observed in rats (McGeorge and Faull 1987, 1989) and primates (Jones et al. 1977), CStrPNi are mostly located in layer V and lower layer II/III, though we also occasionally found CStrPNi in the deeper infragranular layers (Fig. 1C).
CStrPNi are Predominantly Born Between E12.5 and E14.5
In the mouse, deep layer neurons are predominantly born between E10.5 and E13.5, while superficial layer neurons are born between E14.5 and E16.5. To assess the time-course of CStrPNi birth, timed pregnant females were pulse labeled with BrdU at each embryonic day from E10.5 to E16.5; the resulting mice were retrogradely labeled for CStrPNi at P12, and brains were collected at P14. The percentage of BrdU-positive CStrPNi was quantified in the cortex contralateral to the injection site (Fig. 2A–C). A small number of CStrPNi are born at E10.5 and E11.5 (5 ± 2.5% and 6 ± 2.2%, respectively). The greatest percentage of CStrPNi birth occurs at E12.5 and E13.5 (32 ± 0.9% and 42 ± 4.0%, respectively), while this percentage decreases progressively at E14.5, E15.5, and E16.5 (26 ± 3.8%, 13 ± 2.0%, and 5 ± 0.7%, respectively) (Fig. 2D; statistics reported ± standard error of the mean). These data indicate that most CStrPNi are born between E12.5 and E14.5, comparable to other populations of early born, evolutionarily older, corticofugal, and deep layer callosal projection neurons (Molyneaux et al. 2007).
CStrPNi Collaterals Reach Their Targets Between P3 and P4
To establish distinct and potentially critical developmental stages during postmitotic CStrPNi maturation (e.g. midline crossing, axonal entry into the striatum, axonal branching, completion of innervation, axonal pruning), it is important to understand the timing and course of their axonal projections into the contralateral striatum. To visualize the fine axonal projections of CStrPNi during development, crystals of the carbocyanine dye DiI were placed in the cortex of fixed brains at ages P0, P2, P3, and P4 to anterogradely label fibers in the contralateral hemisphere (Fig. 3A). Sufficient time was allowed for full DiI diffusion into the most distant axonal terminals and growth cones. At P0 and P2, no corticostriatal axons contralateral to the DiI deposit were observed entering the striatum yet. The corpus callosum overlying the striatum contained pioneering axons that had already crossed the midline but did not exhibit entry or collateralization into the striatum (Fig. 3B–C). By P3–P4, collateralization of callosal axons was observed to occur into the dorsolateral striatum. These axons display multiple secondary collateral branches (Fig. 3D–F). These data indicate that CStrPNi initially reach the contralateral striatum around P3–P4, defining ∼P3–P4 as the earliest developmental age at which the CStrPNi population can be identified by connectivity. In the absence of subtype-specific markers or other genetic reagents, axonal connectivity enables future investigation of subtype-specific molecular controls, using established approaches originally applied to corticospinal motor neurons and callosal projection neurons (e.g. Arlotta et al. 2005; Molyneaux et al. 2009).
We next asked when CStrPNi collaterals invade the ipsilateral striatum. We specifically investigated whether ipsilateral collaterals enter the striatum before, at the same time as, or after collaterals reach the contralateral striatum. At P0, P1, and P2, we retrogradely labeled from the entire contralateral callosum with CTB555 and simultaneously injected the volume of the ipsilateral striatum with CTB647, and collected brains 24 h later (at P1, P2, and P3, respectively). We did not observe any double-labeled neurons at any of these stages, indicating that ipsilateral striatal collaterals do not enter in advance of contralateral striatal innervation. However, bilateral striatal labeling at P3 (Fig. 3G) revealed double-labeled neurons (Fig. 3H–J). These data indicate that CStrPNi simultaneously project to both striata beginning at P3–4 in mouse. Previous work (Mizuno et al. 2007; Wang et al. 2007) indicates that contralateral cortex is innervated by CPN at around P5, coincident with or just after we find that CStrPNi axons are entering the contralateral striatum. Taken together, these data indicate that CStrPNi send their ipsilateral and contralateral axonal projections simultaneously to their bilateral targets at ∼P3–P4 in the mouse.
Distribution of CStrPNi During Postnatal Development
To directly examine the rostrocaudal and mediolateral distribution of CStrPNi, we injected fluorescent latex microspheres into the dorsolateral striatum at P12–13, and the contralateral cortex was examined for retrogradely labeled neurons at P15. These experiments identified that CStrPNi are predominantly located in the M1/M2 regions of the cortex (Fig. 4Ca–d) consistent with the published distribution in adult rats (McGeorge and Faull 1987, 1989) and primates (Jones et al. 1977). Thus, at the frontal pole, CStrPNi populate the entire dorsoventral and mediolateral expanse of the cortex. At the level of the anterior commissure, the distribution is substantially confined to deep layer II/III and layer V (Fig. 4Cc). A smaller population of CStrPNi is located in layer VI. Further caudal, at the level of the hippocampus (Fig. 4Cd), CStrPNi become sparse. A population of CStrPNi located lateroventrally (Fig. 4Cb–d) might represent a subpopulation of CStrPNi crossing the midline through the anterior commissure, rather than through the corpus callosum (Lent and Guimaraes 1991). We found that the rostrocaudal location of CStrPNi in >3-month-old adult mice is the same as that at P14 (data not shown). Taken together, these results confirm that the distribution of mouse CStrPNi is similar to the distributions in rats and primates.
We next investigated the distribution of CStrPNi through early postnatal development, to investigate whether pruning, neuronal elimination, and/or refinement of other types occurs. We retrogradely labeled the dorsolateral sector of contralateral striatum at P3, when CStrPNi first project axons into the contralateral striatum (Fig. 3). At this early developmental stage, there is a substantially more diffuse distribution of CStrPNi in the neocortex along the rostrocaudal axis, mainly in the deeper layers (Fig. 4Ba–d). Comparison of the distribution of CStrPNi at P15 with the P4 distribution (Fig. 4Ba–d vs. 4Ca–d) shows that, while the distribution remains the same rostrally, the caudal projections to the contralateral striatum are largely absent at P15. This could be due to either retraction of initial collaterals to the contralateral striatum, or death of an early developmental population of caudal CStrPNi (Spreafico et al. 1995; Innocenti and Price 2005).
Some CStrPNi Project to Both Striata and the Contralateral Cortex
To establish whether subsets of CStrPNi might project to multiple contralateral and ipsilateral targets, red and green fluorescent latex microspheres were simultaneously injected into both striata of 2–week-old mice (Fig. 5A). These experiments identified a subset of double-labeled neurons (Fig. 5B–D; ∼15% of CStrPNi), indicating that at least a subset of CStrPNi project bilaterally to both striata, consistent with what is known in rats and primates.
We next investigated whether mouse CStrPNi have collaterals in the contralateral cortex, as is the case for rats and nonhuman primates (Jones et al. 1977; Wise and Jones 1977; Takada et al. 1998; Yeterian and Pandya 1998; Tokuno et al. 1999; Parent and Parent 2006). We employed a broad grid of diffusible FG injections spanning rostrocaudally and mediolaterally to broadly infiltrate unilateral cortex with FG as a retrograde label (Mitchell and Macklis 2005), and we simultaneously stereotaxically targeted the contralateral striatum with fluorescent latex microspheres as a second, distinct retrograde label (Fig. 5F–H). Consistent with data regarding rat and primate CStrPNi, these experiments demonstrate the presence of mouse CStrPNi with projections to both the contralateral striatum and the contralateral cortex.
In a further experiment investigating whether at least some CStrPNi might project to all 3 targets at once, P11–P12 mice were bilaterally retrogradely labeled from both striata with distinct red versus green fluorescent latex microspheres, and with broad injection of FG in the contralateral cortex, as described above (Fig. 5I; ∼8% of CStrPNi). These experiments reveal that a subset of CStrPNi are triple labeled, indicating that a subset of CStrPNi send collaterals to both striata as well as the contralateral cortex (Fig. 5J–M), consistent with what has been reported in primates and rats (Jones et al. 1977; Wise and Jones 1977; Takada et al. 1998; Yeterian and Pandya 1998; Tokuno et al. 1999; Parent and Parent 2006). The actual number of CStrPNi that have axons extending into both striata and/or contralateral cortex is likely to be higher than the estimates we provide here, as retrograde labeling studies typically do not capture all neurons projecting to a region, especially when axon terminals are distributed over a wide area like the dorsolateral sector of the striatum. As a broad population, CStrPNi appear to broadly distribute information both contralaterally to striatum and homotopic cortex, and ipsilaterally within the striatum.
CStrPNi Express a Hybrid “Signature” of Known Molecular Controls of Neocortical Projection Neuron Subtype Identity
Projection neurons in the neocortex can be broadly classified based on their projection patterns into neurons that send their axons either to targets within the opposite hemisphere (callosal projection neurons, CPN) or to targets outside the neocortex (corticofugal projection neurons, CFuPN) (Molyneaux et al. 2007). CStrPNi are a unique population of anatomically “hybrid” neurons that send their axons both across the midline via the corpus callosum, and also away from the cortex to the subcortical striatum. We investigated whether this hybrid connectivity/hodology might be reflected in simultaneous, hybrid molecular character.
To establish whether critical molecular controls for CPN and CFuPN might be expressed distinctly or in combination by developing CStrPNi, we performed ICC for select, recently identified transcriptional regulator controls over development of these distinct populations of cortical projection neurons. Satb2 is a chromatin binding protein necessary for CPN differentiation, and is not expressed by mature CFuPN (Britanova et al. 2005, 2008; Alcamo et al. 2008). Fezf2 (Arlotta et al. 2005; Chen et al. 2005a, 2005b; Molyneaux et al. 2005) and Ctip2 (Arlotta et al. 2005;) are 2 recently identified, functionally important transcription factors controlling specification (Fezf2) and connectivity (Ctip2), expressed specifically by CFuPN (and not by corticocortical CPN) (Molyneaux et al. 2007; Chen et al. 2008; Han et al. 2011; Ip et al. 2011; McKenna et al. 2011; Shim et al. 2012). We found that CStrPNi express Satb2 at P4 (Fig. 6A–D), but do not express either Fezf2 (Fig. 6E–H) or Ctip2 (Fig. 6I–L). Whether CStrPNi might express CFuPN controls earlier in their development cannot be addressed until specific controls over, and, therefore, markers of early CStrPNi development are identified, enabling analysis prior to their innervation of the contralateral striatum (now required for their identification).
The transcription factor Sox5 is of particular interest in this regard, since it is specifically expressed by the major classes of CFuPN (subplate, corticothalamic, and all subcerebral subtypes), but not by corticocortical CPN, and it controls sequential generation of these subtypes by its own progressive downregulation (Lai et al. 2008). Quite interestingly, our experiments reveal that approximately 50% of CStrPNi express Sox5 at P4, and that by P14, it is expressed by all CStrPNi (Fig. 6M–P; U–W, Y). These results suggest that CStrPNi originate as CFuPN rather than as pure corticocortical CPN, and acquire their hybrid ability to cross contralaterally via the corpus callosum via unique mechanisms, with increasing Sox5 repression of subcerebral characteristics.
Additional results reinforce this interpretation. Lmo4, a LIM domain-containing protein, is expressed by corticocortical CPN in layers II/III and V, but is excluded from corticospinal motor neurons postnatally (Bulchand et al. 2003; Arlotta et al. 2005; Azim et al. 2009). During earlier development, Lmo4 is expressed by both CPN and CFuPN, but its expression is refined and is subsequently restricted to CPN (Azim et al. 2009). The current experiments reveal that Lmo4 expression by CStrPNi increases with time; at P4, only some express Lmo4 (Fig. 6Q–T), but by P14 most CStrPNi express Lmo4 (Fig. 6U–V, X–Y). Reinforcing the earlier BrdU birthdating results and these transcriptional regulator expression results together reveal that CStrPNi partially resemble deep layer, CFuPN early, then acquire hybrid deep layer CPN/CFuPN character, the current experiments reveal that CStrPNi do not express Cux1 protein (Fig. 6Z, a–c). Cux1 is expressed by neurons in layers II/III and IV, and data indicate that it might act as a determinant of superficial layer neuron fate (Nieto et al. 2004). The absence of Cux1 from CStrPNi reinforces that CStrPNi neurons are an infragranular, evolutionarily older population, distinct from superficial layer CPN.
Taken together, these results reveal that CStrPNi exhibit hybrid molecular characteristics of both early-born, deep layer CPN with some shared features of typical, noncallosally projecting CFuPN. This might indicate hybrid and/or novel developmental controls enabling this combination of otherwise typically distinct characteristics.
By integrating information from regions of the neocortex and the striatum, corticostriatal projection neurons (CStrPN) participate in the modulation of neuronal activity associated with cognitive, affective, and performance aspects of motor function (Dure et al. 1992). Here, we investigated in mice, the birth, axon extension, and maturation of intratelencephalic CStrPN (CStrPNi), a predominant population of corticostriatal projection neurons in mice (Reiner et al. 2003, 2010).
CStrPNi Distribute Information via Bilateral Connectivity in Mice, Rats, and Primates
We find that at least a subset CStrPNi in mice project to both striata, and a subset also projects contralaterally to the homotopic cortex (Fig. 5), similar to what is known in rats and primates (Jones et al. 1977; Wise and Jones 1977; Donoghue and Kitai 1981; McGeorge and Faull 1987; Cowan and Wilson 1994; Reiner et al. 2003). There is evidence from rat and monkey investigations that there are likely multiple subsets of CStrPNi, some that project only bilaterally to both contralateral and ipsilateral striata, and others that project to both striata as well as contralateral cortex (Wilson 1987; Cowan and Wilson 1994; Kincaid and Wilson 1996; Wright et al. 2001; Parent and Parent 2006). It will be of interest in the future to investigate whether molecularly identifiable subsets exist, and whether such information enables anatomical identification of homologous subsets of CStrPNi in mice, toward further investigation into molecular controls over the specificity of their axonal, synaptic, and functional connections.
The experiments reported here establish that mouse CStrPNi are largely confined to motor (M1) and premotor (M2) cortices; CStrPNi most densely populate broad rostral regions of the cortex, and are more confined to medial areas more caudally. CStrPNi reside mostly in layer V, but smaller subsets are also located in deep layer II/III and in layer VI (Fig. 1), consistent with results previously described in rats (McGeorge and Faull 1987). Together, M1 and M2 cortex coordinate planning and execution of movement (Reiner et al. 2010). CStrPNi in M1 and M2 project to the dorsolateral striatum, and thus into the motor circuitry of the basal ganglia, consistent with their function in motor cognition and planning across species.
CStrPNi are Born During the Early Phase of Cortical Neurogenesis
Consistent with their predominantly layer V location in infragranular cortex, the current experiments find that most CStrPNi are born between E12.5 and E14.5 (Fig. 2), during the early phase of cortical development when corticofugal and deep layer callosal projection neurons are born (Angevine and Sidman 1961; Molyneaux et al. 2007; Fame et al. 2011). Neurons have been identified in turtle dorsal cortex that resembles deep layer output neurons of the mammalian neocortex (Reiner, 1991, 1993). These and other data suggest that deep layer output neurons existed prior to mammalian neocortical evolution. The predominant location of CStrPNi in layer V, along with their early generation and projection to subcortical targets in the striata, suggest that CStrPNi are developmentally related to evolutionarily older corticofugal neurons. CStrPNi appear to be largely distinct from superficial layer CPN (∼80% of CPN) that constitute the most extensive evolutionary addition underlying the striking and relatively rapid expansion of cortical thickness from sauropsids to rodents to primates (Fame et al. 2011).
CStrPNi Have Hybrid Callosal and Corticofugal Features, and Undergo Postnatal Refinement of Projections
One important distinguishing feature in neocortical projection neuron subtype specificity and diversity is whether or not axons cross the midline within the cerebrum (Koester and O'Leary 1993, 1994; O'Leary and Koester 1993; Richards et al. 1997). We find that CStrPNi extend their axons across the midline along with other deep layer CPN, then send collaterals subcortically to the striata by P3–P4 (Fig. 3), developing hybrid callosal and corticofugal connectivity. Interestingly, axons of pure corticocortical CPN enter the contralateral cortex at the same time (Mizuno et al. 2007) that CStrPNi axons enter the striata. Also, we found that at least some CStrPNi project to both the contralateral striatum and the contralateral cortex. Identification of molecular controls regulating midline crossing by CStrPNi axons without disabling later subcortical axonal growth and collateralization might likely elucidate multistep axonal guidance mechanisms relevant to other forebrain neuronal populations.
These experiments also establish that CStrPNi express a hybrid set of both callosal and corticofugal molecular developmental controls. At P4, all CStrPNi express Satb2 (Fig. 6A–D), a chromatin binding protein that regulates callosal identity of projection neurons (Alcamo et al. 2008; Britanova et al. 2008) while they do not express Fezf2-lacZ or Ctip2 (Fig. 6E–L), transcription factors required for specification and axon outgrowth and fasciculation of subcerebral projection neurons (Arlotta et al. 2005). Consistent with the interpretation that CStrPNi are related to deep layer CPN, but not to superficial layer, evolutionarily newer CPN, these experiments reveal that CStrPNi do not express Cux1.
Additional molecular expression analyses reinforce that CStrPNi initially resemble deep layer CFuPN, then transition to suppress subcerebral transcriptional control and adopt more CPN signature. At P4, when CStrPNi axons and collaterals are just beginning to enter the striata, about 50% of CStrPNi express Sox5, a transcription factor expressed by all CFuPN with lower expression levels as sequential populations are generated, whose decreasing transcriptional repression of coordinately regulated sets of genes regulates the sequential generation of CFuPN subtypes (Lai et al. 2008). At this same stage, only a subset of CStrPNi express Lmo4, a LIM domain-containing transcriptional coregulator excluded from corticospinal motor neurons and all subcerebral projection neurons (Arlotta et al. 2005). Interestingly, at P14, all CStrPNi coexpress Sox5+ and Lmo4+ (Fig. 6Q–Y), indicating progressively increased repression of subcerebral molecular character, and progressively increased expression of CPN character. This emerging hybrid expression of these otherwise exclusionary transcriptional regulators in the CStrPNi population suggests the progressive maturation of CStrPNi to possess hybrid characteristics of callosal contralateral projection with ultimate subcortical CFuPN connectivity.
Our data establish substantial refinement of the CStrPNi population and/or their projections in caudal regions of the neocortex. There is substantial loss of projections to the contralateral striatum from the caudal neocortex during early postnatal development (Fig. 5). There are at least 2 possible explanations for this refinement. First, there might be retraction of collaterals to the striatum by caudally located CPN, with these neurons maintaining their projections to the contralateral cortex itself. Such exuberant projections to inappropriate targets, with subsequent retraction during the early postnatal period by both callosal projection neurons (of which CStrPNi are a subset) and subcerebral projection neurons have been reported in cats and rodents (Innocenti, 1981; O'Leary et al. 1981; Stanfield et al. 1982; Stanfield and O'Leary, 1985; Weimann et al. 1999; Polleux et al. 2001; Arlotta et al. 2005; Innocenti and Price, 2005; Low et al. 2008). A second possibility is that there is well-characterized developmental apoptosis ongoing in the neocortex during this same time in rodents (Spreafico et al. 1995; Verney et al. 2000). Therefore, it is possible that immature CStrPNi in caudal cortex might undergo programmed cell death. The progressive stabilization of more rostral CStrPNi axon collaterals might support the final distribution of CStrPNi in adult cortex.
Taken together, these results indicate that CStrPNi are quite a unique and hybrid population of cortical projection neurons, with a mixed set of cardinal features and molecular expression of both CFuPN and CPN. Data from the current experiments indicate a developmental origin common to CFuPN, with molecular genetic modifications to enable telencephalic midline crossing, then later progressive acquisition of increased mature CPN character. These results suggest possible evolutionary origin from CFuPN, with acquisition of molecular mechanisms enabling bilateral collateral distribution of motor planning and control information via distributed bilateral axonal connectivity. Future identification and functional analysis of specific molecular controls over the development of this clinically important neuronal population might potentially both elucidate the circuitry underlying complex mammalian motor control, and provide insight into the pathophysiology of important neurological disease, including Huntington's disease.
This work was partially supported by CHDI, the National Institutes of Health (grants NS49553 and NS45523), the Harvard Stem Cell Institute, the Jane and Lee Seidman Fund for CNS Research, and the Emily and Robert Pearlstein Fund for Nervous System Repair (to J.D.M.), with additional infrastructure support by National Institutes of Health GrantNS41590 (to J.D.M.). U.S.S. was partially supported by CHDI (to J.D.M.) and a seed grant from the Harvard Stem Cell Institute (to U.S.S.). H.K.P. was partially supported by CHDI (to J.D.M.), an International Brain Research Organization Fellowship, and a McKnight Brain Research Institute/Regeneration Project Fellowship (to H.K.P. and J.D.M.). J.R.L.M. was partially supported by a CNPq-Brazil fellowship and an NIH-NINDS Fogarty International Fellowship.
We thank Elizabeth Baratta, Patrick Davis, Peter Garas, Ted Yamamoto, and Kathryn Quinn for technical assistance. We thank Ryann Fame, Jessica MacDonald, Alex Poulopoulos, and Maria Galazo, and other members of the Macklis laboratory for critical reading of the manuscript. Conflict of Interest: None declared.