GAP-43 heterozygous mice show delayed barrel patterning, differentiation of radial glia, and downregulation of GAP-43


  • Vera McIlvain,

    1. Department of Cell and Developmental Biology, SUNY Upstate Medical University, Syracuse, New York
    Search for more papers by this author
  • James S. McCasland

    Corresponding author
    1. Department of Cell and Developmental Biology, SUNY Upstate Medical University, Syracuse, New York
    • Department of Cell and Developmental Biology, SUNY Upstate Medical University, 750 East Adams Street, Syracuse, NY 13210
    Search for more papers by this author
    • Fax: 315-464-8535


GAP-43 heterozygous (HZ) mice exhibit abnormal thalamocortical pathfinding, fasciculation, and terminal arborization at postnatal day 7 (P7). Here we tested whether these defects are correlated with delayed development of HZ cortical patterns. We assessed the rate of barrel segregation and radial glia differentiation in wild-type (WT) and HZ cortices. Since GAP-43 is involved in some forms of neural plasticity, we also compared the duration of the critical period for lesion-induced plasticity in both genotypes. Cytochrome oxidase histochemistry revealed a delay of approximately 1 day in barrel pattern formation in GAP-43 HZ mice. GAP-43 WT barrels showed complete segregation between P2–P3, while HZ barrels did not reach the same level of segregation until P3–P4. We found a similar delay in the transformation of radial glia from monopolar to multipolar phenotypes, from P5 in WT to P7 in HZ cortex. Radial glial cells represent many of the neuronal progenitors in developing cortex and aid in cell migration. Thus, the delay in radial glial differentiation may contribute to the delay in HZ barrel segregation. Interestingly, we found no change in the extent of the critical period for HZ cortical responsiveness to early peripheral damage or in the time course of the cortical response. As expected, GAP-43 expression in HZ cortex is significantly reduced early in development. However, HZ GAP-43 expression remains at maximum levels after P9, when it is normally downregulated. As a result, HZ GAP-43 expression is near-normal by P26, by which time near-normal barrel dimensions have been restored. Our findings indicate that GAP-43 deficiency leads to early delays in barrel development and suggest that these failures are followed by homeostatic responses, including prolonged GAP-43 expression. These compensatory mechanisms may rescue normal cortical reorganization in neonates and near-normal barrel morphology and GAP-43 expression in adulthood. © 2006 Wiley-Liss, Inc.

The presynaptic growth-associated protein GAP-43 is a major constituent of growth cones (1% of total protein). Its synthesis is highly induced in vivo in all neurons that are extending axons (Skene,1989). The timing of GAP-43 expression may regulate several important developmental processes, including axonal pathfinding, axonal regeneration, plasticity, and neurotransmitter release (reviewed in Oestreicher et al.,1997). Reducing or eliminating GAP-43 from navigating axons leads to a number of pathfinding abnormalities. For example, GAP-43 knockout (KO) mice have defects in thalamocortical pathfinding (Maier et al.,1999), forebrain commissural pathways (Shen et al.,2002), retinotectal mapping (Zhu and Julien,1999), and midline crossing of retinal ganglion cell axons at the optic chiasm (Kruger et al.,1998; Maier et al.,1999).

GAP-43 heterozygous (HZ) mice show abnormal pathfinding of thalamocortical afferents (TCAs) and reduced terminal arborization relative to WT (McIlvain et al.,2003). The errors in TCA navigation in GAP-43 HZ mice suggest that cortical patterning is delayed and/or disrupted. We addressed this issue by comparing the rates of barrel segregation and differentiation of radial glia, and the duration of the critical period for lesion-induced plasticity in WT and HZ cortices. We correlated these findings with the time course of GAP-43 expression for both genotypes.

Barrel segregation is integral to cortical morphogenesis. In WT mice, segregation of TCAs into barrel patches occurs in stages between postnatal day 1 (P1) and P3 (Rebsam et al.,2002). A diffuse pattern of cytochrome oxidase (CO) reactivity outlining the barrel field is observed at P1, barrel rows emerge at P2, and individual barrels are clearly outlined at P3. Here we report a delay of 24–36 hr in the timing of barrel segregation in GAP-43 HZ mice.

The delayed barrel segregation of GAP-43 HZ mice could be due to aberrant TCA pathfinding. However, pathfinding errors could be caused by intrinsic defects in HZ corticogenesis. Radial glia perform a variety of essential functions for corticogenesis, from acting as neuronal precursors (Noctor et al.,2001; Fishell and Kriegstein,2003) to aiding in the migration of newborn neurons (Rakic,1975). Radial glia undergo a sequence of morphological transformations late in the period of neuronal migration, from bipolar to monopolar and multipolar forms, then disappear or become astrocytes (Misson et al.,1988,1991; Ma et al.,1989). These transformations occur when radial glia are no longer needed as guides for neuronal migration. Thus, the rate of radial glial differentiation can serve as an indicator of normal and abnormal corticogenesis.

Using RC-2 immunohistochemistry, we analyzed the temporal expression pattern of radial glia. We show that GAP-43 deficiency delays the transformation of radial glia to astrocytes in later phases of neuronal migration, suggesting delayed and/or abnormal corticogenesis.

In addition to regulating axonal navigation, GAP-43 facilitates some forms of cortical plasticity (Oestreicher et al.,1997; Frey et al.,2000; Yamamoto et al.,2000). Overexpression of GAP-43 leads to formation of new synapses and enhanced sprouting after injury (Benowitz and Routtenberg,1997). GAP-43 phosphorylation by PKC increases with the induction of LTP in the rat hippocampus (Gianotti et al.,1992) and is correlated with the magnitude and duration of LTP-induced plasticity (Lovinger et al.,1986). These findings suggest that reduced GAP-43 could alter lesion-induced morphological plasticity (Van der Loos and Woolsey,1973). However, we found no difference in either the duration of critical period or rate of cortical reorganization after follicle ablations in GAP-43 HZ and WT cortex.

We correlated the processes of barrel differentiation, transformation of radial glia, and lesion-induced plasticity with the time course of GAP-43 expression. Since phosphorylated GAP-43 is essential for processes that regulate neuronal growth state (He et al.,1997; Dent and Meiri,1998), we quantified both total and phosphorylated GAP-43 levels from P0–P26 HZ and WT cortices by Western blot analysis. We found that expression of phosphorylated GAP-43 follows a similar time course in WT and HZ cortices. By contrast, expression of total GAP-43 is significantly reduced in HZ mice early in development, but near-normal by P26. This transition correlates with restoration of near-normal barrel dimensions in HZ cortex at P26. Taken together, our findings suggest that compensatory mechanisms in GAP-43 HZ cortex lead to near-normal barrel morphology and GAP-43 expression in adulthood, despite developmental delays and abnormal early GAP-43 expression.



For this study, we analyzed brains of GAP-43 WT and HZ mice (Table 1). The mice were generated by GAP-43 HZ crosses and genotyped as previously described (Maier et al.,1999). GAP-43 KO mice were not used because their barrelless phenotype does not allow the assessment of map development or lesion-induced plasticity. All mice were the progeny of a seventh-generation backcross into the C57BL/6 strain. The treatment of all animals was in strict accordance with both Institutional Animal Care and Use Committee guidelines and those advocated by the National Institutes of Health.

Table 1. Experimental subject
  1. Number of cases processed for all experiments in the present study. Total n = 286.

Assessment of delay in barrel formation:  
 CO (P2)139
 CO (P3)2725
 CO (P4)510
 CO (P27)87
Assessment of plasticity after follicle ablations:  
 Ablated at P138
 Ablated at P21110
 Ablated at P32122
 Ablated at P4513
 Ablated at P533
Western Blot analysis:  
RC-2 immunohistochemistry:  

Cytochrome Oxidase (CO) Histochemistry

Following isofluran anesthesia, animals were transcardially perfused with 4% paraformaldehyde fixative. Brains were removed, postfixed in the same fixative overnight, and cryoprotected with 30% sucrose in 0.1 M phosphate buffer, pH 7.4 (PB). Each hemisphere was flattened tangential to the pial surface overlying the barrel cortex. Serial 40 μm sections were cut on a freezing microtome and processed for CO histochemistry as described previously (McIlvain et al.,2003). After three rinses in PB solution, sections were mounted on subbed glass slides and air-dried. The slides were then dehydrated in ethyl alcohol and xylene and coverslipped.

Follicle Ablation

Neonatal pups of postnatal days 1–5 were anesthetized by hypothermia. An incision with a sharp scalpel was made just ventral to the central row (row C) and follicles C1 through C4 were pulled out through the incision site with extrafine forceps. Whisker pads were carefully examined to ensure that the lesion was complete and all follicles were properly removed. Animals with questionable lesions were excluded from the analysis. The pups were allowed to recover and returned to their home cages.

Classification of Barrel Segregation and Plasticity

For classification of barrel segregation, we analyzed tangential CO reacted sections. Previous reports show that high CO enzymatic activity is found mainly in barrel hollows, where reactive mitochondria reside in many dendrites, some axonal terminals, and a few neuronal perikarya (Wong-Riley and Welt,1980). Thus, CO histochemistry provides a global metabolic perspective on barrel segregation, including postsynaptic and presynaptic components.

To quantify barrel segregation, a trained observer who was blinded to the genotype examined each experimental hemisphere. Each hemisphere was then assigned to one of four categories (Fig. 1A). “No map” was assigned if the area of presumptive barrel field did not show any hints of increased CO activity. “Map outline” was assigned if the area of presumptive barrel field had visible borders. “Partial segregation” reflected the initial segregation of rows or faint barrel outlines. “Complete segregation” indicated hemispheres in which both barrels and septa were clearly distinguishable. The results were analyzed by chi-square test.

Figure 1.

Time course of barrel map segregation in HZ and WT mice. A: Characteristic stages of barrel segregation. Scale bar = 500 μm. B: Distribution of segregation stages in WT and HZ mice at P2, expressed as percent of total animals tested. Barrel development is delayed in HZ mice at P2 compared to WT. The barrels of GAP-43 HZ are significantly less differentiated (P = 0.05, chi-square test). Many of the WT hemispheres are partially segregated, while none of the HZ cortices have reached this level of segregation. C: Distribution of segregation stages at P3, expressed as percent of total animals tested. By this point, some WT cortices achieve complete segregation, while HZ cortices are no more than partially segregated. GAP-43 HZ barrels are significantly less differentiated (P < 0.01, chi-square test).

To compare the degree of plastic changes following ablation of row C at P2–4, a trained observer who was blinded to the genotype assigned each hemisphere to one of two categories: “no fusion” and “fusion.” The no-fusion category was assigned to all hemispheres with clearly distinguishable barrels and septa of row C (for example, see Fig. 4G). The remaining cortices that showed various degrees of cytoarchitectonic reorganization of row C (for examples, see Fig. 4A–F) were assigned to the fusion category. The results were analyzed by chi-square test.

RC-2 Immunohistochemistry

Animals were transcardially perfused with 2% PLP fixative (2% paraformaldehyde, 0.01 M sodium metaperiodate, 0.075 M lysine in 0.1 M PB) under deep isoflurane anesthesia. Following perfusion, brains were postfixed in the same fixative for 4 hr and cryoprotected with 30% sucrose in 0.1 M PB. Each hemisphere was cut coronally at 40 μm on a freezing microtome and processed for RC-2 immunohistochemistry (1:5; Developmental Studies Hybridoma Bank, University of Iowa). Briefly, free-floating sections were incubated in blocking solution [0.2% Triton-X, 4% normal goat serum (JRHBiosciences, Lenexa, KS), 0.5% albumin, bovine (Sigma, St. Louis, MO) in PB] for 1 hr. Then sections were incubated with the primary antibody in blocking solution overnight, followed by secondary biotinylated goat antimouse IgM antibody (1:200; 1 hr; Vector Laboratories, Burlingame, CA), and finally in the avidin-biotin-peroxidase complex (1:80; 1 hr; Vectastain ABC Elite kit; Vector Laboratories). Peroxidase activity was developed with 0.03% diaminobenzine in 0.01% hydrogen peroxide.

Quantitative Western Blot Analysis

Cortical hemispheres (which included cortex, underlying white matter, and a small area of striatum) were dissected from P4, P9, P11, P15, P18, and P26 GAP-43 HZ and WT mice (Table 1 for number of animals used). Sample preparation and Western blot analysis was performed as described previously (McIlvain et al.,2003). Briefly, 15 μg of total protein (determined by the Bio-Rad Protein Assay) was loaded per lane and transferred to a nitrocellulose membrane after electrophoresis. The blots were blocked in 5% nonfat dry milk/Tris-buffered saline and incubated overnight at 4°C with either monoclonal antibody against total GAP-43 (7B10: recognizes phosphorylated and unphosphorylated forms of GAP-43) or monoclonal antibody against phosphorylated form of GAP-43 (2G12; both antibodies were gifts from K. Meiri, Tufts University). Subsequently, the membranes were washed and incubated with [125I] antimouse IgG (15 μCi/μg; Amersham, Arlington Heights, IL) for 2 hr at room temperature. The membranes were then washed again, and radioactive bands were quantified using a Storm 840 PhosphorImager and Image Quant software (Molecular Dynamics, Sunnyvale, CA). Data were presented as mean ± standard error of mean (SEM) and analyzed with two-tailed Student's t-test assuming equal variance and Tukey posthoc (or multiple comparison) comparisons of GAP-43 expression levels across ages within genotype.

Barrel Reconstruction and Quantitative Morphometric Analysis

Serial CO-stained tangential sections at P26 were processed for morphometric analysis as described (McIlvain et al.,2003). Briefly, the nine representative barrels for each genotype (B1-B3, C1-C3, D1-D3; core barrel field) near the center of the posteromedial barrel subfield were assessed for size. To quantify the area of individual barrels by planimetric analysis, a representative section with the clearest outline of the barrel field was chosen from each hemisphere. Each barrel was traced manually, and the resulting outline was analyzed using NIH Image software. In addition to barrel area, individual barrel volumes were calculated. All CO-stained serial sections that contained visible barrels of interest (B1-B3, C1-C3, D1-D3) were photographed and the outlines of individual barrels (excluding septa) in each consecutive section were manually traced to determine the sum of all areas. The volume of each barrel was calculated by multiplying the sum of individual barrel areas by section thickness (40 μm) (McIlvain et al.,2003). We assumed that tissue shrinkage during processing was the same on average in GAP-43 WT and HZ mice, since their brain weight and size were similar. To avoid observer-based bias, all morphometric analysis was performed without prior knowledge of the genotype. Data were analyzed with a two-tailed Student's t-test, assuming equal variance and presented as mean ± SEM.

Photomicroscopy and Figure Preparation

All slides were photographed with an RT Slider SPOT camera (Diagnostic Instruments, Sterling Heights, MI). Photomicrographs were exported to Adobe Photoshop (Adobe Systems, Mountain View, CA) and adjusted as needed to enhance contrast.


Barrels Emerge Later in HZ Than in WT Littermates

To test whether TCA pathfinding errors in GAP-43 HZ mice (McIlvain et al.,2003) correspond to delays in barrel segregation, we analyzed CO-reacted somatosensory cortices of GAP-43 WT and HZ mice at P0–P5. We classified barrel refinement by four basic stages (Fig. 1A). In the first stage, CO reactivity in presumptive barrel cortex is low, and no pattern or clustering is apparent (no map, Fig. 1A). Next, CO reactivity increases and becomes diffusely distributed within barrel cortex. No barrel pattern is observed, but the outer border of the barrel field is sharply defined (map outline, Fig. 1A). The third stage is indicated by crude barrel segregation, with individual domains but only rudimentary separation between nascent barrels (partial segregation, Fig. 1A). The final stage is complete segregation, with clear barrels and septa (complete segregation, Fig. 1A). Each of the GAP-43 WT and HZ CO-reacted hemispheres were assigned to one of these four stages by an observer who was blinded to the genotype. We then compared the distribution of segregation stages.

Our data show that barrel segregation is delayed by approximately 1 day in GAP-43 HZ mice. At P2, 33% of HZ mice showed no visible increase in CO reactivity in the presumptive PMBSF relative to surrounding cortex (no map, Fig. 1B), while 67% had a defined barrel field (map outline). GAP-43 WT mice of the same age were more segregated than HZ littermates, with 54% of the cases having advanced to partial barrel segregation (Fig. 1B). Chi-square analysis confirmed that WT mice have more developed barrels at P2, despite considerable variation within genotypes (chi-square = 7.62; P = 0.05).

At P3, WT mice showed all four categories of barrel segregation, and most showed either partial (55%) or complete segregation (19%; Fig. 1C). A small percentage of WT animals had no map (7%) or map outline (19%). Barrel segregation was much less advanced at P3 in HZ mice—a large percentage showed no map (44%)—while a map outline was observed in 20% of the cases and partial segregation in 36%. None of the HZ cortices showed complete segregation. The delay in barrel maturation at P3 between WT and HZ mice was significant (chi-square = 12.67; P < 0.01).

By P4, the majority of both WT and HZ mice showed clearly segregated barrels (Fig. 2), though variation in patterning was still evident within and between genotypes (data not shown). Taken together, our findings indicate that barrel development in GAP-43 HZ mice is delayed by 24–36 hr.

Figure 2.

Representative CO-reacted sections of WT and HZ mice with fully differentiated barrels at P4. The majority of barrels are visibly segregated in both GAP-43 WT and HZ mice.

Presence of Radial Glia Is Prolonged in GAP-43 HZ Mice

Delayed barrel segregation in GAP-43 HZ mice may represent a combination of defects in thalamocortical pathfinding and abnormal corticogenesis. We assayed WT and HZ corticogenesis by comparing the rate of transformation of radial glia from bipolar to astrocytes.

Radial glia are marked by expression of RC-2 antigen (Misson et al.,1988). In normal mice, RC-2 immunoreactivity disappears when radial glia are transformed into astrocytes at the end of the first postnatal week (Misson et al.,1991). We used RC-2 labeling to compare radial glial scaffolds in GAP-43 WT and HZ mice at P3, P5, and P7: middle to late phases of the radial glia-astrocyte transition.

In both GAP-43 WT and HZ mice at P3, RC-2 immunohistochemistry revealed a dense scaffold of radial glia-like processes oriented perpendicular to the surface and spanning the thickness of the cortex, as well as shorter astrocytic processes that are especially dense in the cortical plate (Fig. 3A and B). This coexistence of radial (monopolar glia, Fig. 3D, insert) and astrocytic (multipolar, Fig. 3C, insert) phenotypes is typical at this age (Cameron and Rakic,1991). The morphology and intensity of RC-2–positive fibers were similar in WT and HZ cortices at this stage.

Figure 3.

Transformation of radial glia in early postnatal cortex in developing WT and HZ mice. Pattern of RC-2–positive cell distribution in coronal sections of WT (top) and HZ cortices (bottom) at P3, P5, and P7. A and B: At P3, monopolar and multipolar forms of radial glia are present in a dense fiber scaffold in both WT (A) and HZ (B) mice. Monopolar radial glia are radially aligned and typically do not extend beyond the cortical plate. Multipolar glia stain densely in the cortical plate, demarcating the emerging layer IV. C: In GAP-43 WT mice at P5, most monopolar radial glia have been transformed to the multipolar phenotype (see enlargement). Multipolar radial glia are especially dense in layer IV, clearly delineating the barrels. D: By contrast, monopolar radial glia are still apparent in GAP-43 HZ mice at P5 (see enlargement). Layer IV has differentiated, and multipolar glia clearly outline the barrels. E: By P7, WT mice show very little immunoreactivity with RC-2 antigens. The multipolar glia disappear so that barrels are no longer visible. F: The disappearance of RC-2 immunoreactivity in HZ mice at P7 is delayed. Although staining intensity is decreased, barrel outlines and some multipolar glial forms remain visible at P7. CP, cortical plate; IV, cortical layer IV. Scale bars = 100 μm.

At P5, the transition from monopolar radial glia to multipolar (stellate) glia was evident in WT but not HZ cortex. GAP-43 WT cortices exhibited predominantly multipolar glia with few or no radial processes (Fig. 3C). The WT multipolar processes were not radially aligned and had variable morphologies (Fig. 3C, inset). RC-2 labeling was especially extensive in layer IV in both genotypes, where a dense network of multipolar glial cells outlined the barrels. However, GAP-43 HZ radial glia displayed more immature forms in lower layers at P5 (Fig. 3F).

At P7, GAP-43 WT cortices showed very weak immunoreactivity for RC-2 (Fig. 3E), and barrel outlines were no longer visible. This loss of RC-2 immunoreactivity at P7 is consistent with previous reports (Misson et al.,1991). By contrast, the multipolar astroglial phenotype was still evident in HZ mice (Fig. 3F). The HZ sections at P7 appeared very similar to those of WT at P5 (compare Fig. 3C and F). The barrels were clearly distinguishable in layer IV, and there was abundant staining of multipolar astroglia through the remaining layers.

By P9, there were no RC-2–positive cells in WT mice. HZ cortex showed very weak RC-2 immunoreactivity at P9, similar to WT cortex at P7 (data not shown). We conclude that the differentiation of radial glia into astrocytes is delayed in GAP-43 HZ cortex by 1–2 days. We have additional evidence that cell migration is abnormal in GAP-43 KO and HZ mice (data not shown). Thus, the observed delays of barrel segregation may be due in part to delayed/abnormal cortical development, not merely to pathfinding errors of TCAs.

Normal Critical Period for Barrel Plasticity Following Peripheral Damage in GAP-43 HZ Mice

The delays in HZ barrel segregation and glial differentiation suggest that the critical period for barrel plasticity may be extended. Furthermore, the link between GAP-43 levels and plasticity events suggests impaired plasticity in HZ cortex. We tested these hypotheses by assessing the critical period for lesion-induced cortical plasticity and the relative rates of cortical reorganization in GAP-43 HZ and WT mice.

When a row of whiskers in the periphery is ablated in WT mice during the critical period, animals develop a dramatically altered barrel pattern (Van der Loos and Woolsey,1973). Barrels corresponding to ablated whiskers shrink, fuse, and even disappear while adjacent barrels enlarge. Follicle ablation after the critical period does not alter the barrel pattern. Early ablations have more dramatic effects (Woolsey and Wann,1976). Follicle ablation can therefore be used to determine the length of the critical period and degree of cortical plasticity in response to peripheral damage.

We performed follicle ablations on row C whisker follicles of GAP-43 WT and HZ littermates at ages P1–P5 and sacrificed at P8–P30. To ensure sufficient time for plasticity events, the initial group survived for 20 days after ablation. Once the critical periods for plasticity responses were assessed in WT and HZ mice, the survival time after follicle ablation was gradually decreased to compare the time course of cortical responses.

The extent of plasticity in WT and HZ mice was assessed by visual analysis of flattened CO-reacted sections. When row C follicles were ablated early in the critical period (P1), both WT and HZ cortices formed a narrow band of fused row C barrels while the patches of surrounding B and D rows were enlarged (compare Fig. 4A and B). At P2, the fused bands appeared wider for both genotypes relative to comparable bands at P1 (Fig. 4C and D). The last day for lesion-induced plasticity was P3, with either partial fusion of row C (Fig. 4E and F) or no change in barrel appearance (no fusion; data not shown). We found some variability in row width within WT and HZ genotypes in each time point of ablation, but there were no significant differences between groups. The cytoarchitectonic barrel patterns of either genotype were no longer altered by follicle ablations at P4 and thereafter (Fig. 4G–J).

Figure 4.

Duration of critical period for lesion-induced plasticity in WT and HZ cortex. Row C follicles were ablated on postnatal days 1–5. Day of ablation (DOA) is indicated at top. Paired arrows mark the row C barrels, corresponding to ablated follicles. In general, later lesions produced less dramatic effects in both genotypes. Lesion at P1 produces a narrow fused band in both HZ and WT mice (A and B, respectively). The band becomes wider when the follicles are ablated at P2 (C and D) and shows only partial fusion when ablated at P3 (E and F). However, when vibrissal damage is performed at P4 or later (GJ), the barrel pattern can no longer be altered in either genotype. Note that for panels A–D and G–J, mice were sacrificed in adulthood (P21–P30), while for panels E and F, mice were sacrificed at P6.

We compared the duration of critical periods in WT and HZ mice by identifying the day when follicle ablations no longer induce changes in the barrel pattern. WT and HZ follicles were ablated on P2, P3, or P4, and the subsequent plastic changes were analyzed qualitatively. One of two categories (fusion or no fusion) was assigned to each hemisphere by an observer who was blinded to the genotype. Follicle ablations at P2 led to fusion of row C barrels for both genotypes (Fig. 5A). Follicle ablation at P3 marked the last day of the critical period for both genotypes. In some cases, barrel maps were no longer plastic, while in others, lesion-induced barrel fusion still occurred (Fig. 5B). The ratios of fused to unfused barrels between WT and HZ genotypes were remarkably close (P > 0.853, chi-square test). By P4, follicle ablation did not produce cytoarchitectonic changes in barrel appearance in either genotype (Fig. 5C). We conclude that the critical period of lesion-induced plasticity, as assessed with CO patterning, is not extended in GAP-43 HZ mice.

Figure 5.

No change in duration of critical period for barrel map plasticity in GAP-43 HZ mice. A: CO stain reveals barrel pattern reorganization after C row whisker lesion at P2 in WT and HZ mice. B: When row C is ablated at P3, similar percentages of each genotype show plasticity (fusion) or do not (no fusion; P > 0.853, chi-square test). This similarity suggests a simultaneous closure of the critical period for lesion-induced plasticity in WT and HZ mice. C: The critical period ends by P4 in both WT and HZ mice. DOA, day of ablation.

The time course of reorganization could be altered in HZ cortex, even though the critical period was normal. We assessed this possibility by ablating row C follicles of WT and HZ mice at P2, the last day for which both genotypes showed fusion of barrels after 20-day survival. We shortened the survival time and found that the rearrangement of cortical connections was visible after only 1 day following ablations in both genotypes (compare Fig. 6A and B). However, in some cases, it was difficult to determine whether the fusion of row C in HZ cortices was due to peripheral damage or delay in barrel segregation. Lesion-induced plastic changes were especially clear if animals were allowed to survive 2 days after ablation to P4, when barrel segregation was normally complete in both genotypes (Fig. 6C and D). Thus, our results suggest that central reorganization occurred with similar efficiency in HZ and WT mice.

Figure 6.

Barrels of both WT and HZ mice fuse within 1 day of follicle ablation. Row C follicles were ablated at P2 and mice were sacrificed at P3 or P4. One day was sufficient for fusion of the ablated barrel row in both WT and HZ mice (A and B, respectively). More prominent fusion was apparent with survival times of 2 days after ablation in both genotypes (C and D).

Developmental Changes of GAP-43 Expression in WT and HZ Mice

To understand how the delays observed in GAP-43 HZ mice correlate with the temporal changes of GAP-43 expression, we quantified GAP-43 in WT and HZ cortices by Western blot analysis. Precise quantification of total GAP-43 was possible due to the specificity of the 7B10 antibody, which labeled single bands (Fig. 7A). GAP-43 expression was measured from birth (P0) to young adulthood (P26).

Figure 7.

Comparison of total GAP-43 protein levels in developing WT and HZ mice. A: Representative immunoblot for total GAP-43, isolated from whole cortices of GAP-43 WT and HZ mice at indicated ages. Fifteen micrograms of total protein were loaded per lane, and each sample was loaded in duplicate to control for loading errors. B: Quantification of GAP-43 protein levels in GAP-43 WT and HZ mice during development. Bars represent the mean cpm/band ± SEM (three or four animals analyzed in duplicate for each bar). GAP-43 expression steadily increased in both genotypes up to P9. Then, in WT mice, GAP-43 levels fluctuated, eventually decreasing to base level. By contrast, the total level of GAP-43 in HZ mice remained relatively constant throughout. C: Plot of total GAP-43 levels in HZ mice as a percent of WT. As little as 25% of total GAP-43 is expressed in HZ mice at P0. However, this percentage increases steadily during postnatal development and approaches 100% at P26. An exponential regression trendline shows that GAP-43 levels in HZ mice approach those in WT by P26.

In both WT and HZ mice (Fig. 7B), total GAP-43 expression increased from P0 to P9 (P < 0.01 WT and P < 0.01 HZ, ANOVA). From P9 to P11, total GAP-43 dropped more than 30% in WT (P < 0.005, Student's t-test), with no significant decrease in HZ (P > 0.306, Student's t-test). From P11 to P15, there was a second wave of upregulation, with total GAP-43 levels increased by 23% (P < 0.024, Student's t-test) in WT but not HZ mice (P > 0.807, Student's t-test). From P15 to P26, WT mice again showed significant downregulation of GAP-43 expression (P < 0.002, ANOVA), while GAP-43 levels in HZ mice again remained constant at maximum levels (P > 0.288, ANOVA).

We also compared the protein levels between age groups within each genotype with Tukey posthoc multiple comparison analysis (Table 2). The Tukey test is considered to be more stringent than a series of individual t-tests, since it eliminates the chance of false significance. The results support our conclusion that GAP-43 HZ mice do not downregulate GAP-43 levels into adulthood as in WT (Table 2, panels A and B).

Table 2. Summaries of tukey post hoc tests
A Total GAP-43 levels in GAP-43 WT mice
  1. increase = statisitically significant increase in protein levels between two test ages.

  2. decrease = statistically significant reduction of protein levels when compared between two test ages.

  3. ns = no significant differences in protein levels were found between two test ages.

B Total GAP-43 levels in GAP-43 HZ mice
C Phosphorylated GAP-43 levels in GAP-43 WT mice
D Phosphorylated GAP-43 levels in GAP-43 HZ mice

Interestingly, the ratio of GAP-43 in HZ to WT mice increases with age (Fig. 7C). The expression of GAP-43 in HZ relative to WT cortex was approximately 25% at P0, more than 50% at P18, and 100% by P26. The increase fits an exponential function with a time constant of approximately 24 days (R2 = 0.74). Thus, the deficiency of GAP-43 expression in HZ cortex early in development was continuously reduced and eliminated by young adulthood. The % WT expression in HZ cortex rises because WT expression falls during this period, while HZ expression remains at near-maximal levels.

To quantify the levels of phosphorylated GAP-43 during different stages of development, we used monoclonal antibodies against the phosphorylated form of GAP-43 (2G12). As with 7B10, 2G12 antibodies also have a very high specificity and show a single sharp band on immunoblots (Fig. 8A).

Figure 8.

Comparison of phosphorylated GAP-43 levels in developing GAP-43 WT and HZ mice. A: Representative immunoblot for phosphorylated GAP-43 isolated from the cerebral cortices of GAP-43 WT and HZ mice at indicated ages. Fifteen micrograms of total protein were loaded per lane, and each sample was loaded in duplicate to control for loading errors. B: Quantification of phosphorylated GAP-43 protein levels in GAP-43 WT and HZ mice during development. Bars represent the mean cpm/band ± SEM (three or four animals analyzed in duplicate for each bar). The levels of phosphorylated GAP-43 in HZ mice were significantly lower than in WT. However, the temporal pattern of phosphorylated GAP-43 expression was similar in both genotypes.

The levels of phosphorylated GAP-43 were always significantly lower in HZ than WT mice (Fig. 8B). However, the temporal pattern of expression was similar in both genotypes: steadily increasing from birth to P9 (ANOVA, P < 0.004 for WT; P < 0.001 for HZ), then gradually decreasing to minimal levels at P26 (ANOVA, P < 0.008 and P < 0.004 for WT and HZ, respectively). Tukey posthoc multiple comparison tests showed similar temporal oscillations in phosphorylated GAP-43 expression (Table 2, panels C and D). The data suggest that GAP-43 phosphorylation by PKC is a dynamic process and is not disrupted by reduced GAP-43 expression in HZ mice.

GAP-43 HZ Mice Have Normal Barrels at P26

How do delays in barrel segregation, glial differentiation, and downregulation of GAP-43 levels in HZ mice affect the barrels in adulthood? To determine whether the aberrant development of HZ mice affects their adult barrel phenotype, we compared WT and HZ barrel sizes in adulthood (P26). CO-reacted tangential sections from GAP-43 WT and HZ cortices at P26 showed no obvious differences in overall appearance of barrel patterns (Fig. 9A). Individual barrel volumes (Fig. 9B) and areas (Fig. 9C) were quantified for both genotypes. All nine P26 barrels analyzed were similar in size in WT and HZ mice (P > 0.05, Student's t-test). We conclude that adult barrel cytoarchitecture is at least partially recovered from earlier deficits in GAP-43 HZ mice.

Figure 9.

Restoration of normal barrel volume in GAP-43 HZ mice at P26. A: Photomicrographs of representative WT (left) and HZ (right) barrel fields at P26. B: The mean volume ± SEM of individual barrels in GAP-43 WT (light bars) or HZ (dark bars) cortices. The mean area of individual barrels ± SEM in GAP-43 WT (light bars) and GAP-43 HZ (dark bars) cortices. Barrel volumes and individual barrel areas were similar in WT and HZ mice (P > 0.05, Student's t-test; n = 9 for each bar). Scale bar = 500 μm.


We have shown that reduced GAP-43 expression delays a number of developmental processes in cortex, including segregation of barrels and transformation of radial glia to astrocytes. However, we found a similar critical period for cortical plasticity, as well as similar cortical responsiveness to peripheral damage, in both GAP-43 WT and HZ cortex. Finally, the abnormally large barrel phenotype in GAP-43 HZ mice at P7 is rescued by young adulthood (P26), despite earlier delays. Below we discuss possible mechanisms of delays and rescue, with respect to GAP-43 expression patterns.

Mechanisms of Delayed Barrel Segregation

There are many possible mechanisms for delayed segregation of barrels in GAP-43 HZ mice. TCA pathfinding errors (McIlvain et al.,2003) are likely to contribute. TCAs of HZ mice show increased fasciculation in the internal capsule and inappropriate turns in subcortical white matter. This inefficient recognition of guidance molecules throughout the nascent TCA projection could prolong pathfinding.

Inefficient arborization of HZ TCAs could also delay barrel segregation. HZ TCAs have sparse terminal arbors (McIlvain et al.,2003), indicating fewer synapses with layer IV neurons. This deficiency could reduce both the displacement of layer IV neurons from barrel hollows to walls and the subsequent reorientation of dendrites into cell-sparse hollows. Both effects would delay barrel segregation.

The delayed transformation of radial glia (present results) may also delay barrel segregation. Radial glia perform a variety of functions during corticogenesis, from aiding in migration of newborn neurons (Rakic,1975) to acting as neuronal precursors (Noctor et al.,2001; Fishell and Kriegstein,2003). GAP-43 KO mice show failures in both timely cell migration and neuronal proliferation (data not shown). It is likely that more subtle but otherwise similar abnormalities occur in HZ mice, and these could delay barrel formation. Finally, reduced serotonergic (5-HT) innervation in GAP-43 HZ mice (Donovan et al.,2002) could also delay barrel segregation, since pharmacological depletion of 5-HT delays barrel development by 1 day (Blue et al.,1991). Our findings may reflect a combination of these and other factors.

Why Is the Critical Period Not Extended in GAP-43 HZ Mice?

GAP-43 is implicated in some forms of plasticity (Oestreicher et al.,1997; Frey et al.,2000; Yamamoto et al.,2000), suggesting that plasticity should be impaired in GAP-43–deficient barrel cortex. We expected an extended critical period for HZ lesion-induced plasticity, consistent with the delay of barrel segregation. Surprisingly, the critical period and time course of cortical responses to peripheral lesions were similar in WT and HZ mice. The delay in barrel segregation, but not in duration of critical period, is especially difficult to reconcile. We will consider several possible explanations.

It is likely that plasticity mechanisms are highly redundant, with efficient backup mechanisms to maintain the normal critical period when any one factor is removed. Neurotrophins, NMDA receptors, and GABAergic inhibition have all been suggested as determinants of the critical period for lesion-induced plasticity. However, several studies suggest that removal of a single factor may not significantly affect the critical period. For example, knockout mice for the NR2A receptor subunit show immature current kinetics and NMDA receptor composition, but normal critical periods in somatosensory cortex (Lu et al.,2001). Similarly, lesion-induced plasticity of TCA, and the critical period are not altered in CxNR1KO mice, which lack NMDA receptor function in cortex (Datwani et al.,2002).

To date, neurotrophin levels have shown the clearest effect on critical period and plasticity events. For example, antibody blockade of an endogenous growth factor prolonged the critical period in visual cortex (Domenici et al.,1994). Conversely, overexpression of brain-derived neurotrophic factor (BDNF) accelerates forebrain maturation of GABAergic circuitry and closure of the critical period (Huang et al.,1999).

Neurotrophins may also regulate plasticity. For example, subcutaneous administration of BDNF and neurotrophin 3 (NT-3) partially rescues barrel development after peripheral deafferentation at birth (Calia et al.,1998). GAP-43 HZ mice appear to express normal levels of mRNAs for most neurotrophins (microarray data not shown), and this could account for their normal critical period and plasticity.

Regulation of GAP-43 Expression With Respect to Developmental Processes

A steady increase in GAP-43 levels in the first postnatal week in both genotypes is consistent with the idea that GAP-43 plays an important role in axon elongation, terminal arborization, and early synaptogenesis (Moya et al.,1988,1989). The subsequent decline in total WT GAP-43 levels after the first postnatal week (∼ P9) may be initiated by target contact (Karimi-Abdolrezaee et al.,2002), probably via retrograde signals, or contact with CNS myelin (Kapfhammer and Schwab,1994a,1994b; Zagrebelsky et al.,1998).

We found that total GAP-43 in HZ cortex remained at maximum levels from P9, when it is normally downregulated, to P26. The failure to downregulate is likely due to weak or delayed retrograde feedback signals from HZ cortex, a consequence of delayed TCA pathfinding. Reduced neurotransmission (Dekker et al.,1989) may also contribute. The failure of downregulation may also reflect compensation for reduced GAP-43 expression in HZ mice throughout development. This compensatory response may assist in normalizing barrel morphology in GAP-43 HZ mice by P26 (Fig. 9).

The brief increase in GAP-43 levels of WT mice at P15 (Fig. 7A) is most likely associated with the formation of cortical synapses in layer II/III (Lendvai et al.,2000) and development of intracortical projections (Miller et al.,2001), both of which occur at that time. After P15, the decrease in GAP-43 levels of WT mice may coincide with the molecular maturation of cortical synapses (Patterson and Skene,1999).

Model of Barrel Development in Normal and GAP-43–Deficient Cortex

Figure 10 shows a conceptual model for involvement of GAP-43 in barrel formation. Reduction of GAP-43 levels leads to aberrant TCA pathfinding (McIlvain et al.,2003), including extensive fasciculation in the internal capsule and inappropriate turns in subcortical white matter. In our model, these errors delay the arrival of TCAs in the cortical plate. Subsequent barrel development can be divided into three steps.

Figure 10.

Hypothetical steps of barrel segregation in GAP-43 WT and HZ mice. Step 1: target acquisition. GAP-43 HZ mice exhibit delayed target acquisition due to low GAP-43 expression and aberrant pathfinding. Step 2: synaptogenesis. GAP-43 WT mice form stable synapses and initiate activity-dependent refinement of barrel boundaries. In HZ mice, fewer stable synapses form and refinement of barrel boundaries does not occur. Step 3: arborization. In GAP-43 WT mice, the density of thalamocortical terminal arbors continues to increase while GAP-43 is downregulated. The HZ barrel phenotype of mice remains in the synaptogenesis stage. By P26, normal barrel volume is restored in HZ cortex.

Target acquisition (step 1).

This involves recognition of a target layer IV WT neuron by an ingrowing TCA. WT TCAs arrive at layer IV by P0–P2 (Rebsam et al.,2002). At this time, TCAs are uniformly distributed with no apparent clusters. In contrast, HZ TCAs are still navigating to their targets due to extensive fasciculation in the internal capsule and inappropriate turns in subcortical white matter (McIlvain et al.,2003).

Synaptogenesis (step 2).

This involves formation and stabilization of WT synapses and initial refinement of imprecise connections by synaptic activity. Hebbian mechanisms strengthen synaptic terminals in the presumptive barrel by correlated pre- and postsynaptic activity. The strengthened synapses facilitate localized sprouting of new branches. Terminals outside the presumptive barrel are persistently weakened by uncorrelated activity and will eventually withdraw from the postsynaptic cell. HZ cortex forms fewer stable synapses due to reduced TCA neurotransmission and cytoskeletal stability (He et al.,1997). The reduction of stable synapses hinders activity-dependent pruning of initially imprecise connections and terminal branching of HZ TCAs.

Arborization (step 3).

This involves elaboration and stabilization of the nascent patterns from step 2 by means of further synaptogenesis and TCA branching. In GAP-43 WT mice, GAP-43 declines in the second postnatal week, presumably in response to a retrograde signal from postsynaptic cells, and TCA arbors become highly branched. By contrast, HZ TCAs continue to express GAP-43 at maximum levels due to failures of synaptogenesis and subsequent retrograde signaling. HZ TCA arbors are still poorly developed at this time.

The enlarged barrel phenotype of P7 GAP-43 HZ mice is rescued by young adulthood (P26), when both genotypes exhibit similar barrel dimensions and total GAP-43 levels (Figs. 7 and 9). This partial recovery of normal phenotype is consistent with compensatory mechanisms in GAP-43 HZ cortex during postnatal development.


The authors thank Drs. Mary Ann Wilson, Dennis Stelzner, Emily Kelly, and Jacob Dubroff for expert advice, ideas, and critical reading of the manuscript. Dr. Kelly also provided excellent assistance with blinded analysis of barrel patterns. They also thank Ginny Grieb for excellent technical assistance. Supported by grants to J.S.M. from the National Institutes of Health, U.S. Public Health Service (NS31829 and NS40779), and the National Science Foundation (IBN-9724102).