Cerebral Cortex Advance Access originally published online on October 25, 2006
Cerebral Cortex 2007 17(8):1918-1933; doi:10.1093/cercor/bhl102
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Comparative Analysis of Layer-Specific Genes in Mammalian Neocortex
1 Division of Brain Biology, National Institute for Basic Biology, 38 Nishigonaka Myodaiji, Okazaki 444-8585, Japan, 2 Department of Basic Biology, The Graduate University for Advanced Studies, 38 Nishigonaka Myodaiji, Okazaki 444-8585, Japan, 3 Laboratories for Cortical Organization and Systematics, 4 Neural Architecture, Brain Science Institute, RIKEN, Wako 351-0198, Japan
Address correspondence to email: yamamori{at}nibb.ac.jp.
 |
Abstract |
|---|
 TopNotes Abstract IntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialReferences |
|---|
We examined the expression patterns of 4 layer-specific genes in monkey and mouse cortices by fluorescence double in situ hybridization. Based on their coexpression profiles, we were able to distinguish several subpopulations of deep layer neurons. One group was characterized by the expression of ER81 and the lack of Nurr1 mRNAs and mainly localized to layer 5. In monkeys, this neuronal group was further subdivided by 5-HT2C receptor mRNA expression. The 5-HT2C+/ER81+ neurons were located in layer 5B in most cortical areas, but they intruded layer 6 in the primary visual area (V1). Another group of neurons, in monkey layer 6, was characterized by Nurr1 mRNA expression and was further subdivided as Nurr1+/connective tissue growth factor (CTGF)– and Nurr1+/CTGF+ neurons in layers 6A and 6B, respectively. The Nurr1+/CTGF+ neurons coexpressed ER81 mRNA in monkeys but not in mice. On the basis of tracer injections in 3 monkeys, we found that the Nurr1+ neurons in layer 6A send some corticocortical, but not corticopulvinar, projections. Although the Nurr1+/CTGF– neurons were restricted to lateral regions in the mouse cortex, they were present throughout the monkey cortex. Thus, an architectonic heterogeneity across areas and species was revealed for the neuronal subpopulations with distinct gene expression profiles.
Key Words: cerebral cortex • corticothalamic • cytoarchitecture • feedback • retrograde tracer • serotonin
|
Introduction |
|---|
|
TopNotesAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialReferences |
|---|
The neocortex is defined by its conspicuous lamination and is generally considered as having a 6-layered bauplan (Brodmann 1909
). In contrast with the ordered stratification of a structure like the retina, however, the fine organization of cortical layers has been more difficult to characterize. Projection neurons to a particular target, for example, are not necessarily contained within a single layer and, conversely, a single layer typically contains neurons projecting to multiple targets. Moreover, cortical connectivity is to some extent area specific. In monkeys, for example, area V1 does not project to the caudate nucleus but the primary somatosensory area does (Jones et al. 1977; Saint-Cyr et al. 1990). For these reasons, laminar-specific connectivity, such as has been elegantly revealed in the retina, remains to be elucidated in the cortex (reviewed in Douglas and Martin 2004; Bannister 2005).
The excitatory neurons that form cortical layers are often viewed as relatively homogeneous, compared with the diverse array of nonpyramidal cell types (DeFelipe 1993; Markram et al. 2004). Recently, however, several genes have been identified in rats and mice that exhibit very high laminar specificity within the neocortex (e.g., see Gong et al. 2003; Hevner et al. 2003; Gray et al. 2004; Arlotta et al. 2005). One such layer-specific gene, fezl, is critically involved in the fate determination of corticospinal motor neurons in mouse layer 5 (Chen, Schaevitz, et al. 2005; Chen, Rasin, et al. 2005; Molyneaux et al. 2005; Molnar and Cheung 2006). Correlation with projection pattern is also reported for ER81 (Hevner et al. 2003, but see Yoneshima et al. 2006), otx1 (Weimann et al. 1999), latexin (Arimatsu et al. 1999, 2003), and Nurr1 (Arimatsu and Ishida 2002) genes. These observations suggest that layer-specific genes may be useful 1) for distinguishing distinct pyramidal cell populations, presumed to share common developmental histories and possibly common functional properties as well and 2) for providing criteria for standardized comparisons across various areas and species.
In this study, we used 4 gene markers to characterize the area- and species-specific features of monkey neocortex. Genes were selected by literature search for reports of layer-specific gene expression either in rodents (rats and mice) or in primates (humans and monkeys). Subsequently, gene expression was visualized by nonradioisotopic in situ hybridization (ISH). This provides single-cell resolution, and for neuroanatomical analyses, is superior to conventional ISH techniques, due to the greater sensitivity and resolution. To distinguish different cell populations, we used a highly sensitive fluorescence double ISH procedure, based on the tyramide signal amplification (TSA) technique (Speel et al. 1999). Finally, in 3 monkeys, we combined ISH with retrograde labeling of thalamically or cortically projecting neurons to investigate the relationship between gene expression and neural connectivity. Our data showed that the deep layer neurons of monkey cortex could be classified into several subpopulations in terms of gene expression profiles. These subpopulations were located in relatively constant laminar positions across areas, but considerable differences were observed in their abundance and relative positioning. We have also observed that despite overall similarity of layer expression, some significant differences exist between monkeys and mice. Thus, we suggest that the genes used in this study can serve as good markers to visualize selective subpopulations of cortical neurons across areas and species. A part of this study has been reported in abstract form (Watakabe et al. 2004).
|
Materials and Methods |
|---|
|
TopNotesAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialReferences |
|---|
Animals and Tissue Fixation
Three young adult Japanese monkeys (Macaca fuscata) were used for the ISH experiments. In addition, 2 adult rhesus monkeys (Macaca mulatta; for V1 and pulvinar injections) and 1 adult Japanese monkey (for area TEO injection) were used for combined experiments with Fast Blue (FB) injections. For tissue processing, monkeys were deeply anesthetized with nembutal (60 mg/kg body weight, intraperitonially [i.p.]) after ketamine pretreatment (16 mg/kg body weight, intramuscularly [i.m.]) and perfused through the heart sequentially, first with 0.9% NaCl containing 2 U/mL heparin and then with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). The brains were postfixed for 4–6 h at room temperature and then cryoprotected with 30% sucrose in 0.1 M phosphate buffer at 4 °C. Adult female mice (C57BL/6) were purchased from Japan SLC, Inc., Hamamatsu, Japan and perfused through the heart with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) under deep anesthesia induced by nembutal (100 mg/kg body weight, i.p.). All experiments followed the animal care guidelines of the National Institute for Basic Biology and National Institute for Physiological Sciences, Japan, and the National Institutes of Health, USA.
In Situ Hybridization
For single ISH in monkeys, serial coronal sections of one hemisphere were prepared, and 38 sections each in total were used for Nissl staining, cholineacetyl transferase histochemistry, ISH for neurofilament M, ER81, 5-HT2C receptor, Nurr1, connective tissue growth factor (CTGF), and VGluT1 genes. These sections were equivalent to the entire hemisphere with an approximately 500-µm interval and contained most of the important brain regions. We inspected all the neocortical areas but limited our description only to selected areas. The cDNA fragments of various genes were obtained by polymerase chain reaction (PCR) using the primers listed in the supplemental material and subcloned into the pBlueScriptII vector. The digoxygenin (DIG)- and fluorescein (FITC)-labeled riboprobes were produced using these plasmids as templates for in vitro transcription. ISH was carried out as previously described (Liang et al. 2000; Komatsu et al. 2005). Briefly, free-floating sections were treated with proteinase K (10 µg/mL for monkey tissue and 1 µg/mL for mouse tissue) for 30 min at 37 °C, acetylated, then incubated in a hybridization buffer containing 0.5–1.0 µg/mL DIG-labeled riboprobes at 60 °C. The sections were sequentially treated in 2x standard saline citrate (SSC)/50% formamide/0.1% N-lauroylsarcosine for 20 min at 60 °C, twice; 30 min at 37 °C in RNase buffer (10 mM Tris-HCl, pH 8.0, 1 mM ethylenediaminetetraacetic acid [EDTA], 500 mM NaCl) containing 20 µg/mL RNase A (Sigma-Aldrich, St. Louis, MO); 20 min at 37 °C in 2x SSC/0.1% N-lauroylsarcosine, twice; and 20 min at 37 °C in 0.2x SSC/0.1% N-lauroylsarcosine, twice. The hybridized probe was detected with an alkaline phosphatase–conjugated anti-DIG antibody using a DIG nucleic acid detection kit (Roche Diagnostics, Basel, Switzerland). There were no apparent signals in control sections with the sense probes examined in macaque monkeys and in mice. ISH causes considerable tissue section shrinkage (approximately 80%). The scale bars in this paper are not adjusted for such shrinkage.
Double ISH
Double ISH was carried out using DIG- and FITC-labeled riboprobes. The sections were cut to 15–20 µm. The hybridization and washing were carried out as described above, except that both DIG and FITC probes were used for the hybridization. After blocking in 1% blocking buffer (Roche Diagnostics) for 1 h, DIG and FITC-labeled probes were detected in 2 different ways. For the detection of the FITC probes, the sections were incubated with an anti-FITC antibody conjugated with horseradish peroxidase (Roche Diagnostics, 1:2000 in the blocking buffer) for 2–5 h at room temperature or overnight at 4 °C. After washing in TNT (0.1 M Tris-HCl, pH 7.5, 0.15 M NaCl, 0.1% Tween 20) 3 times for 15 min, the sections were treated with 1:50 diluted TSA-Plus (dinitrophenol [DNP]) reagents for 30 min according to the manufacturer's instruction (Perkin-Elmer, Wellesley, MA), and the FITC signals were converted to DNP signals. After washing in TNT 3 times for 10 min, the sections were incubated for 2–5 h at room temperature or overnight at 4 °C with an anti-DNP antibody conjugated with Alexa488 (1/500, Molecular Probes) in 1% blocking buffer for the fluorescence detection of the DNP signals. At this point, an anti-DIG antibody conjugated with alkaline phosphatase (1/1000, Roche Diagnostics) was included in the incubation, for the detection of the DIG probes. The sections were washed 3 times in TNT, once in TS 8.0 (0.1 M Tris-HCl, pH 8.0, 0.1 M NaCl, 50 mM MgCl2), and the alkaline phosphatase activity was detected using an HNPP fluorescence detection set (Roche Diagnostics) according to the manufacturer's instruction. The incubation for this substrate was carried out for 30 min and stopped by washing in phosphate-buffered saline (PBS) containing 0.5 mM EDTA. The sections were then counterstained with Hoechst 30442 (Molecular Probes) diluted in PBS to 1:1000 for 5 min. After brief washing in PBS containing 0.5 mM EDTA, the sections were mounted onto gelatin-coated slide glass and prepared with PermaFluor (ThermoShandon, Pittsburgh, PA) mounting medium. It is critical to quickly mount the sections because the red fluorescence of HNPP can easily diffuse. In such cases, the HNPP reaction could be repeated on slides (Watakabe A, unpublished data). In the PermaFluor medium, the fluorescence signal was stable for several weeks, when the sections were stored frozen.
The TSA-amplified fluorescence signals tend to be granular and generally less sensitive than the colorimetric single ISH signals. For relatively abundant genes, however, the ISH signals were strong enough to identify each positive neuron. Unless the mRNA is extremely abundant, we observed no cross reaction or bleaching of the coexpressed signals (data not shown).
Triple ISH
Triple ISH was carried out using DIG-, FITC-, and biotin-labeled riboprobes. The sections were cut to 15–20 µm. The hybridization and washing were carried out as described above, except that DIG-, FITC- and biotin-labeled probes were used simultaneously for the hybridization. After blocking in 0.5% TNB (Perkin-Elmer) for 1 h, the hybridized FITC-labeled probes were turned into DNP signals by TSA-Plus (DNP) kit as described above. After the amplification, the horseradish peroxidase conjugated anti-FITC antibody was stripped from the section by immersion in 0.1 M glycine-HCl (pH 2.2), 0.1% Tween 20 for 10 min. After washing in TS 7.5, the section was incubated with streptavidin conjugated with horseradish peroxidase (Perkin-Elmer, 1:2000 in the blocking buffer) for overnight at 4 °C. After washing in TNT, the sections were treated with TSA biotin system according to manufacturer's instruction (Perkin-Elmer). Finally, the sections were incubated with Alexa488-conjugated anti-DNP antibody, Alexa350-conjugated streptavidin, and alkaline phosphatase–conjugated anti-DIG antibody. After the washing, the detection of the alkaline phosphatase activity was carried out using HNPP fluorescent detection set as described above.
Determination of Layers
The determination of precise laminar identity was very important. Although we could roughly determine the layer positions by Nissl staining of adjacent sections as in Figure 2, the precise determination required double staining of the same section because of the major shrinkage and distortion during ISH. In many areas, including TE and V1, nuclear counterstaining with the Hoechst dye served as a substitute for Nissl staining for determining the more granular layers 4 and 6. In other areas such as the motor cortex, double ISH with a neurofilament gene was helpful. Neurofilament mRNA was strongly expressed by the large pyramidal cells in the deep part of layers 3 and 5, consistent with previous antibody studies (Campbell and Morrison 1989). For counting the positive cells in double ISH (Tables 1 and 2), layers 5 and 6 were determined by the pattern of VGluT1 ISH: neurons in layer 5 were generally larger and more sparsely distributed than those in layers 4 and 6 (see Supplementary Fig. S1). After determining the lamina positions of expression for layer-specific genes, we were able to rely on their patterns for determining the layers. Layer 6B of mice and rats, as designated here, is a very thin sublayer with packed neurons. Some investigators considered this sublayer as a distinct lamina (layer 7), present only in some species (Clancy and Cauller 1999; Reep 2000). Layer 6B of monkeys, in this study, is the lower part of layer 6 that shows colocalization of ER81, Nurr1, and CTGF genes.
|
|
Quantification of the ISH Signals
We carried out 2 kinds of quantification methods to estimate the level of mRNA expression in different areas. In the first method, we analyzed the optical density profiles of the ISH signals to describe the distribution of staining intensity across layers. For this purpose, the images of the 2 adjacent areas (V1/V2 and S1/M1) were skewed by Adobe Photoshop (Adobe systems Inc., San Jose, CA), so that the total thickness in each area would become equal. The optical density of the thickness-adjusted images was then measured by Gel submenu of ImageJ image analysis software (Abramoff et al. 2004).
In the second method, the double-positive cells for the 2 genes of interest were manually counted. Due to the granular nature of the ISH signals, we needed visual inspection to identify the positive cells (Fig. S1). The following procedures were carried out by Adobe Photoshop. The images taken in 3 channels (ISH signals in red/green channels and Hoechst nuclear staining in the blue channel) were layered into a single file. The cells positive for red or green ISH fluorescent signals were separately plotted onto a blank layer manually as dots for later counting. The following criteria were used to count the hybridization signals as one cell. 1) The shape of hybridization signals is recognizable as a single neuron. 2) When compared with the Hoechst nuclear staining, the hybridization signals show the presence of only a single nucleus in the middle. The Hoechst staining was often required to distinguish the close-by cells from a single large cell. In comparing the ISH and Hoechst images, the merged image was created on top of the 2 images and compared directly on screen by changing the visible layers (Supplementary Fig. S1B). After plotting the positive cells for each of the 2 ISH images, the double-positive cells were identified by the overlap of the plotted dots (Supplementary Fig. S1A, Plot). Even when the dots overlapped, we judged them to be distinct cells, if the shapes in the red/green channels did not match. After plotting the single and double-positive cells, the numbers of these dots were counted by Image-Pro Plus image analysis software (Media Cybernetics, Silver Spring, MD).
For the counting, the sections (15-µm width) containing areas 46, M1, S1, TE, V1, and V2 were obtained from 3 monkeys and used for the double ISH. These areas were identified in reference to a standard atlas (Paxinos et al. 1999). For the double staining of VGluT1 and layer-specific genes, a 212-µm-wide window was selected within each area for cell counting. Layers 5 and 6 were determined by the pattern of VGluT1 mRNA expression. The proportions of the ER81+ cells and Nurr1+ cells among the VGluT1+ cells in layers 5 and 6 were calculated for each sample and averaged. For the double staining of the following combinations, 5-HT2C/ER81, CTGF/Nurr1, and ER81/Nurr1 genes, a 424-µm-wide window was selected within each area for cell counting. The proportions of double-positive neurons in different areas for a particular pair of double ISH were examined by one-way analysis of variance (ANOVA) for statistical significance, and the multiple comparison by ryan method was performed to determine the pair of areas that show statistically significant differences.
Retrograde FB Labeling
In 3 monkeys, the retrograde tracer FB (Dr Illing Plastics GmbH, Breuberg, Germany) was injected into V1, area TEO, and the pulvinar nucleus, as previously described (Zhong and Rockland 2004). Surgery was carried out under sterile conditions after the animals were deeply anesthetized with 25-mg/kg nembutal intravenously following a tranquilizing dose of 11 mg/kg ketamine i.m. All procedures for tracer injection were carried out in RIKEN as approved by the Experimental Animal Committee (RIKEN Institute, Wako, Japan). The cortical areas of interest were localized by direct visualization, subsequent to craniotomy and durotomy, in relation to sulcal landmarks (i.e., lunate and superior temporal sulci). For the pulvinar injection, a small craniotomy (5–10 mm) was first opened contralateral to the injection target, in a dorsal position approximately overlying the posterior pulvinar, as estimated by stereotaxic coordinates. By online ultrasound imaging (LOGIQ 9, GE Medical System, Milwaukee, WI; Tokuno et al. 2000), the posterior region of the pulvinar was localized contralaterally, typically in relation to the superior colliculi, which were clearly visible. Then, a second craniotomy was opened over the injection target; and a filled Hamilton syringe (10 µL), coated with Teflon spray (New TFE coat, Finechemical Japan, Tokyo, Japan), was lowered into the pulvinar under ultrasound visualization. Injections were made by pressure (4% FB in 0.1M PBS, 0.5–1.0 µL per injection; Sigma, St Louis, MO). For V1 and TEO injections, 1 µL each of tracers were injected into 3 locations, which were separated by 2 mm. After a survival time of 16 days (for area TEO injection) or 21 days (for V1 and pulvinar injections), animals were perfused transcardially under anesthesia with nembutal (100 mg/kg) as described above. The brains were cut into 20-µm coronal sections using a freezing microtome and used for ISH.
|
Results |
|---|
|
TopNotesAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialReferences |
|---|
We have selected 4 genes with particularly high layer specificity for detailed expression analyses in monkey and mouse neocortices. These genes were ER81, serotonin (5-HT) 2C receptor, Nurr1, and CTGF genes. The expression patterns of ER81 (Weimann et al. 1999
; Xu et al. 2000; Sugitani et al. 2002), Nurr1 (Xing et al. 1997; Arimatsu et al. 2003), and CTGF (Heuer et al. 2003) genes were originally reported for rodents but not for primates. The expression of 5-HT2C receptor mRNA had been reported for monkey and human neocortices (Pasqualetti et al. 1999; Lopez-Gimenez et al. 2001). Although 5-HT2C receptor gene is not well expressed in the neocortical regions of rodents (Wright et al. 1995), and therefore cannot be used for cross-species comparison, we selected this gene because of characteristic expression pattern that was useful to define distinct neuronal subpopulation in monkey neocortices. First, we describe our ISH analysis of ER81, Nurr1, and CTGF genes in the mouse cortex, which confirmed and extended previous findings. Second, based on this analysis, we compared the expression pattern of these genes in monkeys and mice. Expression of Layer-Specific Genes in Mouse Cortex
The ISH patterns of ER81 mRNA expression at 3 different coronal levels of the mouse brain are shown in Figure 1A. In the mouse cortex, ER81 mRNA expression was mostly restricted to layer 5, although we observed sparse signals in layers 2/3 and layer 6. The abundance of expression in layer 5 showed subtle but clear area-specific differences. In general, ER81 mRNA signals were low in the primary sensory areas, such as the somatosensory (S1), auditory (Au1), and visual (V1) areas. In most cortical areas, ER81+ cells were present evenly in both the upper and lower part of layer 5. However, in motor (M1) and adjacent sensorimotor areas, ER81 mRNA was only sparsely expressed in the lower part of layer 5 compared with the upper part (Fig. 1A, Fig. S2).
|
Consistent with the previous reports of radioisotopic ISH (Xing et al. 1997
) and immunohistochemistry (Arimatsu et al. 2003), Nurr1 mRNA showed striking area differences in its distribution patterns. Nurr1+ cells were abundant in the deep layers of the lateral cortical areas (e.g., S2 and secondary auditory cortex) but were sparse or absent in the medial areas except in the thin layer 6B (e.g., M1, S1, and V1). The distribution pattern of Nurr1 mRNA appeared to change at the borders of different areas. For example, the distribution pattern of Nurr1 mRNA changed at the S1/S2 border (Fig. 1A,C): in S2, Nurr1 mRNA was expressed throughout layer 6, but in S1, there was a gap between the upper part of layer 6 and 6B. In addition, the primary auditory area (Au1) showed slightly less abundant expression of Nurr1 mRNA than did V2L, despite its lateral position (Fig. 1A, Nurr1, occipital). Area-specific differences were also observed in the expression in layers 2/3 in the occipital cortex, which may be faintly seen in Figure 1. This upper-layer expression was not observed in the more frontal sections.
As reported previously (Heuer et al. 2003), CTGF mRNA was expressed in the thin layer 6B across areas (Fig. 1A). The specificity was very high and we observed virtually no expression of CTGF mRNA in layers 1–6A. In addition to the neuronal expression, we observed CTGF mRNA in the pia matter as well.
Although these 3 genes exhibit highly layer-specific expression patterns, some of their expressions overlapped. To further characterize the specificity of gene expression, we developed a sensitive procedure for fluorescence double ISH and examined the coexpression profiles of the 3 genes within the deep layers (see Materials and Methods).
First, we examined the coexpression of ER81 and Nurr1 genes. In the laterocaudal areas, Nurr1+ cells scattered into layer 5 (Fig. 1A, filled arrow b) and intermingled with ER81+ cells (Fig. 1B). Despite this intermingling, the 2 mRNAs were not expressed in the same cells (Fig. 1B). We also observed a few ER81+ cells in layer 6, but they also did not coexpress Nurr1 mRNA (data not shown). Next, we investigated the coexpression of Nurr1 and CTGF mRNAs in layer 6B. As expected from the single ISH, the Nurr1+ cells in layer 6B generally coexpressed CTGF mRNA (Fig. 1C). There were also CTGF+/Nurr1– cells in layer 6B.
Expression of Layer-Specific Genes in Macaque Monkey Cortex
Figure 2 shows the distribution of the 4 genes in 4 cortical areas. Among these genes, ER81 gene has been recently reported to be expressed in layer 5 of neonatal monkeys (Yoneshima et al. 2006). With our ISH procedure, we were able to detect the expression of ER81 mRNA even in the adult monkey cortex. 5-HT2C receptor gene is reported to be expressed in layer 5 of the primate neocortex, but no mRNA expression had been observed specifically in V1 (Pasqualetti et al. 1999; Lopez-Gimenez et al. 2001). We confirmed the layer 5 expression of 5-HT2C receptor mRNA throughout monkey areas including V1, although at a lower level. Nurr1 and CTGF mRNAs were both expressed in layer 6 in monkeys, which was consistent with the expression pattern in mice.
|
To investigate the monkey-specific features of gene expression patterns in more detail, we examined the precise lamina localization of each gene with reference to Nissl staining and each other (Fig. 2). In area 46, for example, ER81 mRNA was enriched throughout layer 5. In addition, lower expression of ER81 mRNA was also observed in layer 6, unlike in mice. Compared with ER81 mRNA expression, 5-HT2C receptor mRNA was observed in a much more restricted subpopulation of neurons located in the lower half of layer 5. Although Nurr1 and CTGF mRNAs were both observed in layer 6, the upper border of CTGF mRNA appeared to be slightly lower than that of Nurr1 mRNA. Nevertheless, unlike the mouse pattern, the CTGF mRNA expression in the monkey cortex was more widespread throughout layer 6.
Across areas, the relative layer positions of these genes were generally conserved, although 5-HT2C receptor mRNA appeared to be expressed in layer 6 in V1 (we analyze this point in detail later). In this regard, the expression patterns of Nurr1 mRNA in monkeys appear to be more constant across areas compared with the highly area-specific distribution in the mouse cortex. The constancy of laminar expression was particularly striking at sharp borders such as those between V1 and V2 (Fig. 3A–D) or between M1 and S1 (Fig. 3E–H). Even across these sharp borders, the neurons positive for each gene formed a continuous layer of expression.
|
Layer-Specific Genes Exhibit Area-Specific Lamina Profiles in Monkey Cortex
Although the lamina positions of the 4 genes were relatively constant across areas, we have observed considerable area differences in other respects. In Figure 3, we showed the changes of the ISH patterns of each gene across V1 and V2 and across S1 and M1 by optical density profiles of the ISH signals. The normalized profiles show 2 kinds of area differences. First, the abundance of mRNA expression, represented by the size of the peak zone in the profile, was variable. For example, the overall expression level of ER81 mRNA in V1 was much lower than that in V2 (Fig. 3A). The expression of 5-HT2C receptor mRNA exhibited an even greater difference across the V1/V2 border (Fig. 3B). On the other hand, the expression of Nurr1 mRNA was similar between V1 and V2 (Fig. 3C). Second, the relative width of the peak zone for each gene varied greatly across areas. This reflects the large area difference in the relative thickness of layers 5 and 6 (see Supplementary Fig. S3). For example, the relative thickness of both layers 5 and 6 in M1 were disproportionately large compared with other areas. Reflecting such cytoarchitecture, the optical density profiles of the 4 layer-specific genes were generally broad and without sharp peaks in M1 (Fig. 3E–H).
In addition to the quantitative differences described above, we also observed differences in the spatial configuration of positive neurons at cellular resolution. For example, the Nurr1+ neurons were segregated into 2 sublayers in V1, and to a lesser extent in V2 (shown by the filled and open arrowheads in Fig. 3C), but not in other areas (Fig. 2). Nurr1 mRNA expression also showed variability in the expression in layer 5. In M1, we observed strongly positive Nurr1+ neurons scattered into layer 5 (Fig. 2). Such incursion of Nurr1+ neurons into layer 5 was also conspicuous in area S2 or in the supratemporal gyrus (data not shown), moderate in TE, but was rare in V1, S1, and area 46.
Layer-Specific Genes are a Subpopulation of Excitatory Neurons
The layer-specific expression strongly suggests that the 4 genes are expressed in excitatory neurons. To confirm this, and to estimate their proportion among the excitatory neurons, we carried out double ISH of the layer-specific genes with the excitatory marker, vesicular glutamate transporter 1 (VGluT1) gene (Fujiyama et al. 2001). Double ISH revealed that more than 95% of the ER81+ and Nurr1+ cells coexpress VGluT1 mRNA (data not shown, see Supplementary Fig. S1). Using double ISH, we counted the numbers of ER81+ and Nurr1+ cells among the VGluT1+ cells in 6 distinct areas of the monkey neocortex; that is, areas 46, M1, S1, TE, V1, and V2 (Table 1, Fig. S4). This analysis showed that approximately 57–71% of the VGluT1+ cells in layer 5 and 21–29% of the VGluT1+ cells in layer 6 expressed ER81 mRNA. We did not detect statistically significant differences among the 6 areas examined (one-way ANOVA, P > 0.05). We also found that 20–31% of the VGluT1+ cells express Nurr1 mRNA in layer 6 (Table 1, Fig. S4). The area difference was found to be statistically significant (one-way ANOVA, P < 0.05), but a multiple comparison by the ryan's method failed to find statistically significant pairs among the 6 areas tested.
5-HT2C+ Neurons are a Subpopulation of ER81+ Neurons in Layer 5
Considering that ER81 and Nurr1 mRNAs were expressed in a subpopulation of excitatory neurons, it was critical to determine the colocalization of the 4 layer-specific genes with each other. First, we investigated the coexpression of ER81 and 5-HT2C receptor mRNAs, which are both expressed in layer 5. Figure 4A shows double ISH of the 5-HT2C receptor and ER81 genes in area TE. In this experiment, 5-HT2C receptor mRNA was detected by the red fluorescence of HNPP/Fast Red deposition and ER81 mRNA was detected by the green fluorescence of the Alexa488-conjugated anti-DNP antibody, which recognizes the TSA-amplified signal. The merged image unambiguously demonstrated that 5-HT2C+ neurons occupy only the lower half of the zone of dense ER81 mRNA expression, as expected from the single ISH experiments. Importantly, we found that most of the 5-HT2C+ neurons coexpress ER81 mRNA (insets for Fig. 4A). In our counts, more than 95% of the 5-HT2C+ cells expressed ER81 mRNA in areas other than V1 (data not shown). In V1, 78 ± 16% (means ± standard deviation [SD], n = 3) of the 5-HT2C+ cells expressed ER81 mRNA. We suspect that the ER81 mRNA expression was less efficiently detected in V1 due to the lower level of ER81 mRNA per cell.
|
In Table 1, we show the ratio of the 5-HT2C+ cells in the ER81+ cells in layer 5 of 6 areas, areas 46, M1, S1, TE, V1, and V2 (for the ratio in V1, see the footnote for the table). We found that in the 6 areas examined, 5-HT2C receptor mRNA was present in 14–22% of ER81+ cells in layer 5 (Table 1, Fig. S4). We did not find any significant differences among the 6 areas including V1 and V2 (one-way ANOVA, P > 0.05). In V1 and V2, the absolute numbers of 5-HT2C+ cells in 424-µm-wide windows were different by more than 2-fold (14 ± 4.6 in V1 and 35 ± 9.6 in V2). This difference, nevertheless, was proportional to the numbers of ER81+ cells in layer 5 of V1 and V2 and was not significant.
Distinct Subgroups of ER81+ and Nurr1+ Neurons can be Defined by Coexpression Pattern
Next, we investigated the coexpression of ER81 and Nurr1 mRNAs in area TE. Here, we found 2 distinct modes of coexpression patterns. Within layer 5 and in the upper part of layer 6, ER81 and Nurr1 mRNAs were generally not coexpressed within the same cell (Fig. 4B), even when there was an extensive intermingling of Nurr1+ and ER81+ cells. This is consistent with the mouse result. In contrast, in the lower part of layer 6, many ER81+ cells did coexpress Nurr1 mRNA. It is worth noting that despite the sparse distribution of ER81+ and Nurr1+ cells in the lower part of layer 6, they coincide almost perfectly in this sublayer (compare the magnified view in the bottom row). On the basis of these orderly coexpression patterns, we distinguished 3 different subpopulations of pyramidal cells in layers 5 and 6: ER81+/Nurr1– cells mainly present in layer 5, ER81–/Nurr1+ cells in the upper part of layer 6, and ER81+/Nurr1+ cells in the lower part of layer 6.
Although we sampled disparate cortical areas, such as S1, S2, M1, auditory cortex, parietal cortices (LIP, PGM), temporal cortices (TE, TEO), frontal cortex (area 9, 46), and visual cortices (V1, V2, V4), in all these areas, ER81+/Nurr1–, ER81–/Nurr1+, and ER81+/Nurr1+ cells were generally positioned in layer 5, the upper part of layer 6, and the lower part of layer 6, respectively. The extent of intermingling, however, was quite variable across areas. In area V1, ER81+/Nurr1– cells and ER81–/Nurr1+ cells were intermingled, at the border between layers 5 and 6, whereas ER81+/Nurr1+ cells were separated in layer 6B (see below for more detail). In area S2, ER81–/Nurr1+ cells and ER81+/Nurr1+ cells were intermingled within layer 6. We counted the numbers of these cell types in layer 6 of 6 selected areas and found statistically significant differences among areas (Table 2, see below for more details).
V1-Specific Distribution Pattern of ER81+/5-HT2C+ Cells
In the course of the double ISH experiments, several laminar specializations were observed specifically in V1 and not in A1 or in S1. First, the zone of overlap between the mRNA expressions of ER81 and Nurr1 genes was larger at the layer 5/6 border (Fig. 5A). Hoechst nuclear counterstaining revealed that Nurr1+ cells were still mainly within layer 6 (the border is shown by a green line in Fig. 5C) but that there was more intrusion of ER81+ cells into this layer (Fig. 5A–C). In the upper part of layer 6, as mentioned above, the ER81+ cells did not coexpress Nurr1 mRNA, but those in the lower part of layer 6 did.
|
Second, 5-HT2C+ cells occupied a strikingly different position in V1 compared with other cortical areas, including even the adjacent extrastriate cortex, area V2 (Fig. 5D). That is, whereas 5-HT2C+ cells were located mainly above the Nurr1+ cells in the extrastriate areas, these cells were mostly intermingled with the Nurr1+ cells in V1. Importantly, however, even if these cells were intermingled, 5-HT2C receptor and Nurr1 mRNAs were not expressed within the same cells (Fig. 5E'–G'). As expected, these 5-HT2C+ cells coexpressed ER81 mRNA (cells with arrows in Fig. 5H'–J'). From these results, we concluded that the ER81+/5-HT2C+ cells located in the lower half of layer 5 in most cortical areas shifted their position deep into layer 6 in V1. It is also worth noting that 5-HT2C receptor and Nurr1 mRNA expression segregated in M1, too, where some Nurr1+ cells were dispersed into layer 5 (Fig. S5).
Distinct Subpopulations of Layer 6 Cells can be Identified by the Expression of Nurr1 and CTGF mRNAs
Next, we investigated the coexpressions of Nurr1 and CTGF mRNAs, which occurred together mainly in the lower part of layer 6 (Figs 2 and 3). First, we investigated the coexpression of these genes in V1 and V2, where we observed the segregation of 2 sublayers. As shown in Figure 6A,B, a strong CTGF mRNA signal was present in only a subset of cells in the lower part of layer 6 of V1 and V2. The double ISH showed that these cells correspond exactly to those expressing the Nurr1 mRNA. Next, we examined the coexpression patterns of these genes in other cortical areas, where Nurr1 mRNA formed a single band. The typical pattern was as shown for area TE. In TE, the Nurr1+ cells in the upper part did not coexpress CTGF mRNA at a high level, whereas those in the lower part coexpressed CTGF mRNA (Fig. 6C). In most cortical areas, Nurr1+/CTGF– cells (indicated by the green bar) and Nurr1+/CTGF+ cells (indicated by the orange bar) were separable into 2 compartments within layer 6. In area S2, however, these 2 populations of cells were intermingled at the border (Fig. 6D: note the overlap of green and orange bars). These results demonstrate that, unlike in the mouse cortex, Nurr1+/CTGF– cells are present in the upper part of layer 6 throughout the monkey areas.
|
Coexpression of ER81, Nurr1, and CTGF mRNAs in Layer 6
In the double ISH experiments of ER81/Nurr1 and Nurr1/CTGF genes, we have shown that Nurr1 single-positive cells mainly resided in the upper part of layer 6, whereas the Nurr1-positive cells in the lower part coexpressed ER81 or CTGF mRNAs (Figs 4 and 6). This observation strongly suggests that the 3 genes are coexpressed in the lower part of layer 6. To test this directly, we performed triple ISH of these genes in M1, where the cell density is low and the mRNA level per cell for each gene is relatively high and clearly found triple-positive cells (Fig. 7B, white dots, see Fig. 7D for the ISH images). In Figure 7C, we schematically show the relative abundance of the 7 types of cells plotted in Figure 7B. This diagram showed that ER81, Nurr1, and CTGF mRNAs are coexpressed in high proportion.
|
Next, we investigated whether there is any area difference in the proportions of overlaps in the expression for ER81, Nurr1, and CTGF genes, by counting the numbers of single- and double-positive cells for double ISH of ER81/Nurr1 and Nurr1/CTGF genes in various areas. In Figure 8A,D (see Supplementary Fig. S6 for the color version), we present typical results for areas 46 and V1. Similar to the pattern shown in Figure 4B for area TE, Nurr1 single-positive cells (gray squares) and Nurr1/ER81 double-positive cells (filled dots) were generally segregated into upper and lower parts in areas 46 and V1, although the extent of intermingling was quite different in the 2 areas (Fig. 8A). A similar pattern was observed for the double ISH of Nurr1 and CTGF mRNAs: Nurr1 single-positive cells (gray squares) and Nurr1/CTGF double-positive cells (filled dots) were segregated into the upper and lower compartments (Fig. 8D). Comparing the patterns between Nurr1/ER81 (Fig. 8A) and Nurr1/CTGF (Fig. 8D) double ISH, the distribution for the ‘Nurr1 single-positive’ cells and the ‘double-positive’ cells was quite similar in both double ISH in each area, although more double-positive cells appeared to exist in the upper sublayer for Nurr1/CTGF double ISH. This can be explained if the Nurr1+ cells in layer 6B mostly consist of Nurr1/ER81/CTGF triple-positive cells, as we have observed in M1.
|
In Table 2, we summarize the results of the cell counting experiments. As this table shows, the area differences in the proportions of overlaps between ER81, Nurr1, and CTGF genes were statistically significant. To illustrate this point better, we graphically demonstrated the area differences of the proportion of the ‘single-positive cells’ among each positive neuron (Fig. 8). As this graph shows, the proportion of the Nurr1+/ER81– cells (Fig. 8B) and Nurr1+/CTGF– cells (Fig. 8E) among the Nurr1+ cells showed a similar variability across areas, suggesting a considerable overlap in these cell populations; we found that the proportion of the Nurr1 single-positive cells is highest in areas V1 among the areas tested. This area difference was statistically significant at least between V1 and TE (for Nurr1/ER81 double ISH). We also found area differences in the proportions of ER81 single-positive (Nurr1–) cells (Fig. 8C) and CTGF single-positive (Nurr1–) cells (Fig. 8F). Although we did not mention the presence of the CTGF single-positive cells in Figure 6C, the cell counts revealed that the CTGF single-positive cells were found in high proportion in the association areas 46 and TE (Fig. 8F). We note that because the expression level of the CTGF mRNA was generally lower in the CTGF single-positive cells than that in the CTGF/Nurr1 double-positive cells, its presence was not obvious in Figure 6C.
Differential Connectivity of Nurr1+ Cells in Layer 6
It has been shown in the rat that the Nurr1+ neurons send corticocortical but not corticothalamic projections (Arimatsu et al. 2003). To examine the connectivity properties of the Nurr1+ cells in the monkey cortex, we combined retrograde tracing and ISH experiments. Although the harsh treatment of the tissue sections during ISH reduced the FB fluorescence considerably, we were still able to clearly observe the residual signals (e.g., Fig. 9A). To label the corticocortical cells, we injected FB into V1 and TEO. After injections into these areas, retrogradely labeled cells were abundantly observed in layers 2/3 and layer 6 (both layers 6A and 6B) of V2 and TE, respectively (Fig. 9A and data not shown). Fluorescence ISH for the Nurr1 gene showed that some FB-positive cells in layer 6 expressed Nurr1 mRNA in both areas (Fig. 9C,D,G). On the other hand, after injection of FB into the pulvinar nucleus, many corticothalamic neurons were observed in layer 6 of area PO but few of them expressed Nurr1 mRNA (Fig. 9E,F,G). Thus, the Nurr1+ cells in layer 6 are considered to be corticocortical cells, as in rats. It is, however, noteworthy that only 20–25% of the FB-positive cells expressed Nurr1 mRNA after a cortical injection (Fig. 9G).
|
As described above, the Nurr1+/ER81–/CTGF– cells and the triple-positive cells were segregated in V2 (see Fig. 6B). By counting the Nurr1+/FB+ cells in the upper and lower parts of layer 6, separately, we found that almost all of the double-positive cells were located in the upper part (Fig. 9D,H). This result suggests that the Nurr1+ cells in upper layer 6 and the Nurr1+ cells in the lower part have different connectivity.
|
Discussion |
|---|
|
TopNotesAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialReferences |
|---|
By using single, double, and triple ISH with cellular resolution, we report 5 main results in this paper. First, 3 of the genes investigated in this study exhibit conserved layer preferences in the adult neocortices of monkeys and mice: ER81 for layer 5, Nurr1 for layer 6, and CTGF for layer 6B. Second, despite general conservation between the 2 species, several significant species differences were also found. ER81 mRNA, for example, was expressed in layer 6 throughout the neocortex in monkeys but not in mice. Furthermore, the Nurr1+/CTGF– neurons were present throughout monkey cortical areas but were restricted to the lateral regions in the mouse cortex. Third, the combination of the 4 gene markers revealed distinct neuronal subpopulations with preferred sublamina localization across areas. In monkeys, the major subpopulations that we observed were as follows: ER81+/5-HT2C– cells in layer 5, ER81+/5-HT2C+ cells in layer 5B (but in layer 6 in V1), Nurr1+/CTGF– (and/or ER81–) cells in layer 6A, Nurr1+/CTGF+ (and ER81+) cells in layer 6B, and Nurr1–/CTGF+ cells in layer 6B. These subpopulations were well separated into sublayers in some areas (e.g., V1) but were intermingled in others (e.g., M1). Fourth, the absolute numbers of these subpopulations differ across areas. This area difference is partly attributable to the general difference of the cytoarchitecture, but some of the subpopulations showed statistically significant area differences in their relative abundance. Finally, projection neurons that were retrogradely labeled by tracer injections in cortical or thalamic structures showed an orderly concordance with molecular markers; that is, Nurr1 mRNA expression correlated with corticocortical connectivity, although the correlation was only partial.
Technical Considerations: Advantage and Limitation of the ISH Technique for Use in Neuroanatomy
The ISH technique has been used successfully to determine the expression patterns of various genes in the nervous system. However, due to the low sensitivity and/or resolution of the previous double ISH methods, its use in neuroanatomy was limited compared with the immunohistochemical techniques. As we have shown in this and previous studies (Komatsu et al. 2005; Miyashita et al. 2005; Takahata et al. 2006), the optimization of the procedure enabled us to determine the colocalization of various genes at cellular resolution. Furthermore, the fluorescent ISH technique was compatible even with the retrograde tracer, FB. The advantage of ISH method is obvious, considering that the list of genes that are expressed in specified neuronal populations is rapidly increasing (Gong et al. 2003; Gray et al. 2004; Toledo-Rodriguez et al. 2004; Sugino et al. 2006). With an increasing number of these genes, ISH technique is quite suitable for cross-species comparison: making a species-specific gene probe by PCR with 100% sequence match for hybridization is much easier than producing an antibody that shows high specificity for each species.
Despite the advantages of the ISH technique as described above, there are also limitations. First of all, we should caution that the sensitivity of the fluorescent double ISH is still limited. Even with our procedure, cellular identification becomes very difficult for low-abundance genes, due to the granular hybridization signals. The relatively large SD of cell counting in some samples (Tables 1 and 2) may be partly attributable to the inconsistency of cell identification with low signals. Second, ISH does not work well if the gene is too GC rich or AT rich. Third, ISH technique cannot reveal the subcellular localization of a protein product. Finally, combined use of FB and ISH is not so sensitive because the FB signals faded considerably after the ISH procedure. This last problem can be addressed by improvements, such as the use of virus vector (Tomioka and Rockland 2006).
Possible Function of the Layer-Specific Genes
The contribution of the layer-specific genes to processing in the adult cortical circuits remains to be studied. Among the genes investigated here, ER81 belongs to the PEA3 subfamily of ETS transcription factors (Sharrocks 2001); and in the spinal cord, it is expressed in a pair of sensory and motor neuron pools that form functional circuits with each other (Lin et al. 1998). The lack of ER81 expression leads to the failure of the proprioceptive sensory neurons to arborize in the vicinity of motor neuron dendrites and to form appropriate connections (Arber et al. 2000). Thus, the ER81 gene may similarly be involved in layer-specific circuit formation in the cortex during development (Yoneshima et al. 2006).
Although one recent study did not observe ER81 mRNA in the adult rat and monkey cortex (Yoneshima et al. 2006), ER81 protein is detected by immunohistochemistry in adult mouse (Hevner et al. 2003) and rat (Yoneshima et al. 2006) cortices in a layer-specific fashion. Furthermore, in the present investigation, we were able to clearly show the presence of ER81 mRNA in the adult mouse, rat, and monkey cortex (Figs 1 and 8 and data not shown). The difference between the previous report and our data is likely to be due to the difference in the sensitivity of the ISH methods utilized.
Nurr1, a member of the nuclear steroid/thyroid hormone receptor superfamily, is intensely expressed in the substantia nigra and plays a critical role in the differentiation of dopaminergic neurons (Zetterstrom et al. 1997). Nurr1 can induce the transcription of several dopaminergic phenotypic markers such as tyrosine hydroxylase in embryonic stem cells, independent of neurogenesis (Sonntag et al. 2004). However, because tyrosine hydroxylase is not expressed in the cortex, there may be an additional control of gene transcription by Nurr1.
Whereas ER81 and Nurr1 genes are transcription factors and may play regulatory roles in fate determination, 5-HT2C receptor and CTGF genes could directly affect the neuron's functional property. For example, the 5-HT2C receptor is expected to modulate neuronal activity by serotonin. Although expression of 5-HT2C receptor gene is not strongly detected in mouse or rat neocortical areas, it showed layer 5 specificity in the cat cortex (data not shown), suggesting that its lack of expression may be a specific trait for rodents. CTGF is a member of the CCN (for CTGF, cyr61/cef10, nov) family of secreted proteins and may be important for cell adhesion, migration, and organogenesis (Perbal 2004). These properties suggest a role in intercellular communication.
Area and Species Differences of the Neocortex Revealed by Layer-Specific Genes
Among the monkey cortical areas, area V1 was characterized by several distinctive features. Quantitative differences were most conspicuous in V1, where the band of ER81+ and 5-HT2C+ neurons was unusually thin (Fig. 3). In addition, in area V1 but not in other areas, the majority of 5-HT2C+ neurons were displaced into layer 6 and intermingled with Nurr1+ neurons (Fig. 5). Furthermore, a distinct gap separated the Nurr1+/CTGF– neurons and Nurr1+/CTGF+ neurons in V1 but not in other areas. These observations are consistent with the extreme specialization of the primate V1 (reviewed in Lund 1988; Peters and Rockland 1994), where the overtly stratified structure can be considered as correlated with the segregated processing of various visual attributes (reviewed in Sincich and Horton 2005). It would be interesting to compare these laminar patterns with those in other primates with different habitats and specializations, such as New World monkeys and nocturnal monkeys, to reveal how the laminar architecture of V1 differentiated through evolution (Northcutt and Kaas 1995).
Regarding species comparisons, the conservation of laminar patterns for ER81, Nurr1, and CTGF mRNAs in mouse and monkey cortices suggests conserved functional roles for these genes in each species. Of particular interest is the coexpression profiles of Nurr1 and CTGF mRNAs. In both species, Nurr1+ neurons in the upper part of layer 6 are devoid of CTGF mRNA expression, whereas those in the lower part of layer 6 mostly express CTGF mRNA. This observation suggests that the Nurr1+/CTGF– cells in layer 6A and Nurr1+/CTGF+ cells in layer 6B are distinct cell populations. This is consistent with the previous reports that Nurr1+ cells in layers 6A and 6B have different ‘birthdates’ and show differential colocalization with latexin (Arimatsu et al. 2003). In terms of connectivity, Nurr1+/CTGF– neurons in layer 6 exhibited corticocortical but not corticothalamic projections in the tracer injections of 3 monkeys in this study as well as in rats (Arimatsu et al. 2003). We also point out that approximately 70–80% of the excitatory neurons in layer 6 did not express Nurr1 mRNA in monkey cortex (Table 1). The corticothalamic neurons are likely to be included in this Nurr1-negative subpopulation in both monkeys and rodents.
Several significant species differences were noted. For example, the Nurr1+/CTGF– neurons exhibited differential areal expression in the monkey and mouse cortices (compare Figs 1 and 6). Furthermore, the Nurr1+/CTGF+ neurons in monkeys expressed ER81 mRNA, which we have shown by triple ISH (Fig. 7), whereas this population in layer 6B of mice does not express ER81 mRNA (Fig. 1). We also did not detect latexin mRNA in monkey neocortices, which is expressed in the Nurr1+/CTGF– neurons of rats and mice (data not shown). Finally, in rats, but not monkeys, the neurons in layer 6B exhibit corticocortical connectivity (Clancy and Cauller 1999; Arimatsu et al. 2003; Mitchell and Macklis 2005).
Ontogenic and Evolutionary Implications of the Layer-Specific Gene Expression
The use of double ISH indicated a distinct sublaminar stratification for neurons expressing a particular combination of the gene markers. Although such pattern is consistent with the conserved lamina organization across areas, as envisaged by Brodmann (1909), it is not clear what neuronal properties correlate with the gene expression phenotypes. One obvious possibility is that the sublaminar stratification reflects the commonality of birthdates: the neuronal subpo