CR is dynamically expressed in L5a pyramidal neurons of the barrel cortex
CR, a calcium-binding protein, is reported to be an important modulator of neuronal excitability [13], with involvement in synaptogenesis, axonal elongation and dendritic remodeling [13],[14]. Here, we detected the postnatal expression and possible role of CR in the mouse somatosensory cortex. At birth, CR was scarce in the somatosensory cortex (Figure1A). At P3, a row-like band of CR expression could be detected in the deep layer of the barrel cortex (Figure1B), and the expression level of CR gradually increased until it peaked at P8 (Figure1D), as previously reported [14]. Scattered CR-positive neurons throughout the cortex were also observed (Figure1A-F), and these neurons are likely interneurons according to previous reports [15],[16]. We further found that neurons that strongly expressed CR in the row-like zone within the deeper layer of the barrel cortex are horizontally aligned in a serrated pattern (Figure1C-D). After P8, the expression level of CR then gradually decreased and barely detectable at P30, except in scattered interneurons (Figure1E-F). This dynamic expression of CR in the deeper layer of the barrel cortex is consistent with the time window during which cortical barrels develop, which suggests a possible role for CR in barrel cortex development. To identify the features and functions of the CR-positive cells, we first performed a BrdU birth-dating analysis. BrdU was injected at E12.5, E13.5, E14.5, E15.5, respectively, corresponding to the time points when layer 6 (L6), L5 and upper layer neurons are born. The brains were then harvested at P8, when the expression of CR is strongest in the deeper layer in the barrel cortex and almost all of the projection neurons had migrated to their final destinations in the cortical plate. Double immunostaining for CR and BrdU revealed that approximately 43.40??2.70% of neurons with strong CR expression in the row-like zone were also positive for BrdU administered at E13.5, which is the time point when the majority of L5 neurons are generated (Figure1G,H,I, arrow and insert). But few of the strongly CR-positive neurons were positive for BrdU labeling applied at other time points (Figure1J,K,L), suggesting that CR-positive neurons are mainly born at E13.5.
In the barrel cortex, L5 consists of two populations of L5a and L5b neurons that receive paralemniscal and lemniscal projections, respectively. We further determined the exact localization of CR-positive neurons within L5 at P8. When slices were co-labeled with Cux1, an upper layer (L2/3 and L4) excitatory neuronal marker, and Ctip2, a marker for L5b excitatory neurons [17],[18], CR-positive neurons were found between the layers of the Cux1-positive and L5b Ctip2-positive neurons without co-localization with either marker (Figure2A-F). Meanwhile, two transcription factors that are widely expressed in cortical excitatory neurons, neurogenic differentiation 2 (NeuroD2) [19] and forkhead box proteins 2 (Foxp2) [20], were found to be expressed in most of the CR-positive neurons (Figure2G-L). Statistical analysis revealed that in the row-like zone, 96.22??2.94% of CR-positive neurons were NeuroD2-positive and 97.04??2.55% were Foxp2-positve. Taken together, our data indicate that CR-positive neurons are L5a excitatory pyramidal neurons, the main target of POm projections [4].
To exclude the possibility that CR-positive neurons in L5a are interneurons, double immunostaining for anti-CR and anti-gamma-aminobutyric acid (anti-GABA) was performed. Within the row-like zone, no CR-positive neurons were found to also be GABA-positive, although CR- and GABA-positive interneurons were found scattered throughout the cortex (Figure3C,C). This result was also confirmed by co-labeling CR with anti-GFP using the GAD67-GFP knock-in mouse line in which GFP is expressed in almost all GABAergic interneurons under the control of the endogenous GAD67 promoter [21]. Abundant GFP-positive interneurons were detected throughout the cortex, but few GFP-positive interneurons within L5a were found to also be CR-positive (Figure3F,F). These data all indicate that CR-positive neurons within L5a are excitatory pyramidal neurons and not GABAergic interneurons.
L5a pyramidal neurons form a unique serrated pattern that requires sensory input during early postnatal days for maintenance
In the barrel cortex, L5a pyramidal neurons specifically receive inputs from the POm [1],[2],[4],[7], whose axon terminals display a row-like pattern with regularly spaced triangular structures in L5a. It has been reported that the POm projects more axons to L5a neurons vertically aligned under septa than to the adjacent neurons below the barrels [1],[2],[7]. This distinct projection pattern of the POm afferents may recruit their postsynaptic partner L5a pyramidal neurons to form a projection-related distribution pattern. To test this hypothesis, a detailed analysis of the distribution of CR-positive L5a neurons was performed by immunostaining at various time points during development of the barrel cortex. At P3, the radial processes of L5a neurons started to extend into the overlying L4 (Figure4A,B). At P4, CR was observed to be expressed in more L5a neurons and formed a clear row-like structure (Figure4C). Furthermore, we found that the CR-positive L5a neurons displayed a non-homogeneous distribution with some cells aggregated into serrated structures, and the pattern was quite similar to that of their presynaptic POm projections (Figure4C,D). Regularly spaced CR-positive processes were observed to extend from the serrated structures into L4 (Figure4D). The expression of CR peaked at P8 (Figure4I), and the serrated distribution pattern became more defined, with the neuropil spanning all of L4 and forming a prominent septa-like pattern (Figure4J). The septa-like pattern could be observed until P15, but the expression level of CR in L5a decreased from P8 on (Figure4K). The structures between the septa-like CR-positive processes might represent prospective barrels (Figure4I-K). By P30, the CR-positive row-like structure with septa-like processes had almost disappeared, and the barrel structures became more apparent (Figure4L). These observations suggested that L5a neurons could be recruited by inputs from the POm to form serrated structures with their processes extending to L4 as septa-like structures that closely resembled the projection pattern of the POm into L5a.
It has been reported that the first postnatal week is critical for experience-dependent pattern formation in the barrel cortex. The afferent signal is the principal element in experience-dependent neuronal patterning [3],[22],[23]. It is possible that the distribution pattern of CR-positive L5a neurons may also be input-dependent. To further detect whether the distribution pattern of CR in L5a depends on sensory input, an early transection of the infraorbital nerve (ION) experiment was carried out. ION transection at P2 led to markedly reduced expression of CR in the neuropil and cell bodies of the L5a neurons by P3 (Figure4E,F). By P4, the CR-positive processes that extend to L4 had become shorter, and the septa-like pattern disappeared (Figure4G,H). At P8 and P15, few septa-like CR-positive processes were observed to extend to L4. More interestingly, the serrated distribution pattern of CR-positive L5a neurons also disappeared, and only the uniform row-like CR-positive zone remained (Figure4M,N,O). By P30, expression of CR in L5a could no longer be detected, similar to observations of the control. Taken together, our data indicate that the unique serrated alignment of CR-positive L5a pyramidal neurons is related to the paralemniscal pathway and depends on peripheral sensory input.
The alignment of L5a neurons develops synchronously with formation of the cortical barrels
Timing of the formation of the distinct serrated alignment of L5a neurons corresponds to the time window of cortical barrel formation and is input dependent, suggesting that the lemniscal and paralemniscal pathways might develop synchronously. We next compared the temporospatial development of the neuronal pattern of L5a with that of the barrels in L4 within the barrel cortex.
Previously, we reported a transgenic mouse line Fzd10-TauLacZ in which the thalamocortical afferents (TCAs) from the VPM are specifically labeled [24]. We used the same Fzd10 promoter to generate an EGFP reporter mouse line (Figure5A). Using this mouse line, we found that the GFP-positive TCAs had already reached the deep layer in the barrel cortex at E17.5 (Figure5B) and had extended their arbors into L4 by P0 (Figure5C). However, at P0, no CR expression was detected in L5a (Figure5C). By P3 and P4, the GFP-positive TCA terminals had invaded L4 (Figure5D,E,F) and segregated into clusters, but the clusters were not completely separated from each other (Figure5E,F). Meanwhile, CR expression was observed in L5a, and some of the CR-positive neuropil had begun to extend into the overlying L4 (Figure5D,G,H). At P8, the GFP-positive TCA terminals in L4 were completely segregated into distinct clusters to form barrels (Figure5M,N). At the boundary of L4 and L5, the CR-positive L5a pyramidal neurons intruded into regions between the TCA terminal clusters and formed regularly spaced triangular alignment structures from which the CR-positive neuropil extended into L4 to form a septa-like pattern (Figure5M,N). The alignment of the CR-positive L5a neurons complemented this pattern of GFP-positive TCA terminals. From P8 to P15, when the barrels were nearly mature, the GFP-positive TCA clusters became denser, and the expression intensity of CR decreased (Figure5O,P). We then performed ION lesions at P2 and found that the unique alignment of CR-positive neurons was disrupted and that the CR-positive neuropil became shorter. Although the GFP-positive TCAs were able to reach L4, no barrels formed. The serrated pattern of the L5a neurons disappeared; instead, a uniform CR-positive band was detected, and no septa-like pattern was observed (Figure5K-L, Q-T). These results suggest that segregation of the VPM projections and formation of the CR-positive L5a pattern may occur in parallel during development of the cortical barrels.
The serrated alignment of L5a neurons is disrupted in CR mutants
The distinct expression pattern of CR in L5a neurons raised the possibility that CR may be important for development of the paralemniscal pathway. As a member of the calmodulin superfamily, CR plays an important role in the precise regulation of intracellular calcium signals and neuronal excitability [13],[25]-[27]. Therefore, we next examined the alignment of L5a neurons in CR knockouts [28]. In situ hybridization was performed to detect the distribution of the ets variant 1 (Etv1) gene, which is a transcription factor reported to be specifically expressed in L5a neurons [29],[30]. At P8 in controls, Etv1-positive L5a neurons displayed a distinctive serrated pattern with regularly spaced triangular structures protruding towards the overlaying layer that corresponded to the pattern of CR (Figure6C and E, arrows). However, in CR mutants, the serrated pattern was disrupted with no triangular structures and a uniform distribution of Etv1-positive L5a neurons (Figure6D-F), indicating that CR is important for the formation of the distinct serrated arrangement of L5a pyramidal neurons.
CR is also specifically expressed in the relay nucleus of the paralemniscal pathway
The paralemniscal pathway originates from large neurons in the interpolar subnucleus of the spinal trigeminal nucleus (Sp5i), which project to the POm in the thalamus [31], and these POm afferents target L1, L5a and the septum-related L4 neurons in the S1 cortex. We found that, in addition to being expressed in L5a pyramidal neurons, CR was also dynamically expressed in the POm and Sp5i, which are the relay nuclei of the paralemniscal pathway. At E14.5, the expression of CR could be observed in the POm, which is located lateral to the parafascicular thalamic nucleus (PF) (Figure7A) and the Sp5i (Figure7G). This expression became stronger at E16.5 (Figure7B,H) and continued to persist (Figure7C-F,I-L). To determine whether CR-positive neurons are excitatory neurons or interneurons, double labeling of anti-CR and GFP was performed in the GAD67-GFP knock-in mouse line at P8, when the pattern of the paralemniscal pathway had formed in L5a. No co-localization was detected in the POm or Sp5i (Figure8A-F, insert), which indicates that these CR-positive neurons are excitatory projection neurons. The expression of CR in the developing paralemniscal pathway suggests that CR may be involved in development and maturation of the paralemniscal pathway as a modulator of neuronal excitability and of calcium signaling pathways.