Abstract
We immunohistochemically and morphometrically examined the expression of gap junction protein connexin (Cx) in normal and crush-injured rat sciatic nerves using confocal laser scanning microscopy. Cx26 was localized in the perineurium and Cx43 was present in the perineurium and the epineurium, whereas Cx32 was confined to the paranodal regions of the nodes of Ranvier. Double labeling for connexins and laminin revealed that Cx43 was localized in multiple layers of the perineurium, whereas Cx26 was confined to the innermost layer. Double labeling for connexins and a tight junction protein, occludin, showed that occludin frequently coexisted with Cx43 but existed separately from Cx26 in the perineurium. After crush injury, the pattern of normal Cx32 expression was initially lost but recovered, whereas Cx43 rapidly appeared in the endoneurium and its expression was subsequently attenuated. Although crush injury produced no apparent alteration in Cx43 and occludin in the perineurium, a rapid increase and a subsequent decrease in the frequency of Cx26-positive spots during nerve regeneration were shown by morphometric analysis. These results indicate that Cx26, Cx32, and Cx43 are expressed differently in various types of cells in peripheral nerves and that their expressions are differentially regulated after injury. The expression of connexins and occludin in the perineurium suggests that perineurial cells are not uniform in type and that the regulation of gap junctions and tight junctions is closely related in the perineurium.
G
The peripheral nervous system consists of three different compartments, i.e., the endoneurium, perineurium, and epineurium (from inside to outside). The endoneurial compartment contains axons and their surrounding Schwann cells, collagen fibers, fibroblasts, capillaries, and a few mast cells. The perineurial compartment that encloses the endoneurial compartment is made of concentric layers of flattened fibroblast-like cells called perineurial cells (Ortiz-Hidalgo and Weller 1997). The presence of basal lamina on both sides of the perineurial cell, and of tight junctions between adjacent perineurial cells, makes it possible for the perineurium to function as a diffusion barrier (Akert et al. 1976; Breathnach and Martin 1980; Ghabriel et al. 1989). The epineurial compartment consists of moderately dense connective tissue binding nerve fascicles together. Cell-cell interaction and communication among these various types of cells enable the peripheral nervous system to maintain its structure and function in physiological and pathological conditions.
Gap junctional intercellular communication constitutes an important part of cell-cell communication in the peripheral nervous system. It has been shown that Cx32 is localized at the paranodal regions of the nodes of Ranvier and Schmidt-Lantermann incisures in the peripheral nervous system (Bergoffen et al. 1993; Miyazaki et al. 1995; Scherer et al. 1995; Chandross et al. 1996) and that mutations in the Cx32 gene result in X-linked Charcot-Marie-Tooth disease, a progressive and slowly developing peripheral neuropathy (Bergoffen et al. 1993; Chance and Pleasure 1993; Fairweather et al. 1994). With regard to connexins other than Cx32, only a few studies have been conducted on localization in the peripheral nervous system. Even among the results from those studies, there are discrepancies. For example, Yoshimura et al. (1996) reported that Cx43 was expressed in the cytoplasm of Schwann cells in human peripheral nerves, whereas Scherer et al. (1995) and Chandross et al. (1996) showed negative data concerning the existence of Cx43 in Schwann cells in vivo. Therefore, it remains to be resolved what kinds of connexins are expressed in which types of cells in the peripheral nervous system.
In this study we performed immunofluorescence studies in intact rat sciatic nerves using antibodies against Cx26-, Cx32-, and Cx43-specific peptides, in combination with confocal laser scanning microscopy, to elucidate which connexins are expressed in the peripheral nervous system. Double labeling was also performed by combining connexin immunolabeling with antibodies to laminin (a component of basal lamina) or occludin, an integral tight junction protein (Furuse et al. 1993). By comparing the anti-connexin labeling with the labeling of these antibodies, the precise distribution of these connexins in the perineurium and the co-localization of gap junctions with tight junctions could be determined. Using a nerve crush injury model, we immunohistochemically and morphometrically analyzed connexin expression after crush to learn whether the expression of connexins is modulated during peripheral nerve regeneration.
Materials and Methods
Animals
Adult Wistar rats weighing 250–350 g were sacrificed under deep diethyl ether anesthesia and sciatic nerves were exposed and removed. The specimens were immediately frozen in liquid nitrogen with Tissue-Tek OCT compound (Miles; Elkhart, IN) and stored at −80C.
Nerve Crush Injury
Fifteen adult Wistar rats were anesthetized by IP injection of sodium pentobarbital (4 mg/100 g body weight). Anesthesia was monitored and confirmed by the absence of pain reflexes. One side of the sciatic nerve was exposed and crushed for 10 sec at a sciatic notch with fine forceps. The muscles were then reattached at the suture line and the wound was closed. After various intervals (4, 7, 14, 21, or 28 days after crush injury, three rats/group), animals were sacrificed with diethyl ether and the crushed sciatic nerves were removed for analyses. The sciatic nerve from the contralateral side was collected as a control.
Immunohistochemistry
The applied primary antibodies, species, dilutions, and sources are given in Table 1. Serial 10– 20μm frozen sections were prepared, air-dried, and fixed in absolute ethanol at − 20C for 15 min. After incubation in PBS, pH 7.4, supplemented with 5% skim milk for 15 min at room temperature to block nonspecific labeling, the primary antibodies (see Table 1) were applied overnight at 4C. After three 5-min washes with PBS, incubation with fluorescein isothiocyanate (FITC)-labeled anti-rabbit or anti-mouse antibodies (1:300 dilution; DAKO, Copenhagen, Denmark) was performed for 1 hr at 37C. After final rinses in PBS, the sections were mounted in Vectorshield (Vector Laboratories; Burlingame, CA). All antibody dilutions were made in PBS containing 1% bovine serum albumin (Sigma; St Louis, MO) and 1% sodium azide with 0.1% Triton X-100. As negative controls, specimens were incubated in the absence of the primary antibodies.
Characteristics of antibodies used in this study
aCx, connexin; PGP9.5, protein gene product 9.5.
bZYM, Zymed Laboratories (South San Francisco, CA); CMN, Chemicon International (Temecula, CA); ICN, ICN Biomedicals (Aurora, OH); UCL, UltraClone (Isle of Wright, UK)
A rabbit polyclonal anti-protein gene product 9.5 (PGP9.5) antibody (1:500 dilution; UltraClone, Isle of Wight, UK) was used for serial sections from crush-injured nerves to assess the presence of regenerating axonal fibers (Gulbenkian et al. 1987) at the crushed and 5-10-mm distal sites of the nerves.
Some sections were double labeled for Cx26 and Cx43 by utilizing the rabbit anti-Cx26 (Kuraoka et al. 1993) and the mouse anti-Cx43 antibodies. Some sections were double labeled for either Cx26 or Cx43 and laminin or occludin, in which mouse anti-connexin, rabbit anti-laminin, or anti-occludin antibody was utilized. After blocking nonspecific binding of biotin-avidin system reagents with a Blocking Kit (Vector), a mixture of mouse and rabbit antibodies (for dilution see Table 1) was applied overnight at 4C. After rinses in PBS, both FITC-conjugated anti-rabbit immunoglobulin (used as above) and biotinylated anti-mouse immunoglobulin (1:200 dilution; Vector) were incubated for 1 hr at 37C. Then Texas Red-streptavidin (1:100 dilution; Vector) was further reacted for 1 hr at 37C. After final rinses in PBS, the sections were mounted in Vectorshield. For all immunolabeling, negative controls included incubation with secondary antibodies after omission of the primary antibodies.
Confocal Laser Scanning Microscopy
The sections were viewed with a confocal laser scanning microscope (Olympus FLUOVIEW system; Tokyo, Japan) coupled to an inverted microscope (IX70; Olympus) equipped with an oil immersion objective (PlanApo X60, NA = 1.4; Olympus). An argon-krypton laser generated excitation bands at 488 nm for FITC, 568 nm for Texas Red, and monochromatic light for differential interference contrast (DIC) images. Fluorescence images were collected with emission filters for 510-550 nm for FITC and 585-610 nm for Texas Red. Simultaneous images (800 × 600 pixels, 12 bits each) of FITC and Texas Red labels were obtained and stored. In some cases, DIC images were obtained in parallel with fluorescence images and three-channel images were constructed with FLUOVIEW imaging software. Digital images were transferred to a Macintosh computer (Apple Computer) equipped with Photoshop software (Adobe System) and printed on an ink-jet printer (PM-750C; Epson, Tokyo, Japan).
Morphometric Analysis
For morphometric assessment of connexin-immunolabeled spots, digital images were collected by adjusting the sensitivity of the photomultiplier so that the spectrum of label intensities in each section spanned within the 4096-level scale. Section planes were obtained at 1-μm steps along the z-axis using the objective × 60 and a zoom setting of 2 or 3. A single optical plane was then selected and the image was converted to 8 bits (256 level), and transferred to the NIH image program (Wayne Rasband; NIH, Bethesda, MD) in a Macintosh computer. A binary overlay was created and spots were morphometrically analyzed by counting their number (frequency) and measuring their area (size). The extent of co-localization of occludin (green signals) with connexins (red signals) was evaluated as the percentage of yellow area in the green or red area. In the crush-injury model, each image was obtained 5-10 mm proximal or distal to the site of the crush injury, and gap junctional spots were counted and measured in the same way as above. As a control, data were obtained from untreated sciatic nerves. Data on each group were collected from more than six images and were expressed as the mean ± SD.
Statistical Analysis
Statistical analysis was performed using StatView software (Abacus Concepts). Data on the size and frequency of immunolabeled connexin spots in normal sciatic nerve were examined by the Kruskal-Wallis test or the Mann-Whitney U-test. Data in the crush model were examined by two-factor factorial ANOVA and Scheffé test. Significant differences were defined by p<0.05.
Results
Normal Rat Sciatic Nerves
Indirect immunohistochemistry on transverse sections of normal rat sciatic nerves demonstrated the expression of Cx26, Cx32, and Cx43. The polyclonal rabbit antibody against Cx26-specific peptides (Kuraoka et al. 1993) revealed punctate labeling in the innermost layer of the perineurium, but the distribution of Cx26-positive spots was not even (Figures 1A and 1B). Another rabbit polyclonal antibody against Cx26-specific peptides (Zymed; South San Francisco, CA) (not shown) and the mouse monoclonal antibody against Cx26 (Figures 2A and 4A) also showed a punctate pattern in the perineurium. With all three antibodies against Cx26, we found no clear labeling in the endoneurium or the epineurium.
In transverse sections, Cx32 immunolabeling was detected in some neural fibers of the endoneurium but not in the perineurium or the epineurium (Figures 1C and 1D). Labeling was localized to the paranodal regions of the nodes of Ranvier in the longitudinal sections (Figures 6A and 6B). Identical results were obtained with both mouse and rabbit antibodies against Cx32 (not shown).
With the polyclonal rabbit antibody against Cx43, abundant punctate Cx43 immunolabeling was detectable in multiple layers of the perineurium, and a small number of unevenly distributed punctate spots were also detected in the epineurium (Figures 1E and 1F). However, no Cx43 immunoreactivity was found in the endoneurium except in endoneurial vessels (not shown). Identical results were obtained by use of the monoclonal mouse antibody against Cx43 (Figures 2B, 3, and4B). The immunohistochemical localization of connexins and other proteins in this study is summarized in Table 2.
The anti-Cx26 and anti-Cx32 antibodies demonstrated the expected punctate patterns of gap junctional plaques in rat livers, as did the anti-Cx43 antibodies in rat hearts (not shown). In the absence of the primary antibodies, nonspecific fluorescence was not noted in the sciatic nerve with FITC-labeled anti-rabbit or anti-mouse antibodies (Figures 1G-1J). Table 3 summarizes morphometric data on the size and frequency of immunolabeled connexin spots in the normal rat sciatic nerve. There was no significant difference among the different antibodies for each connexin.
Immunohistochemical localization of connexins in adult rat sciatic nerves in transverse sections. Confocal immunofluorescence images of Cx26 (
Confocal images of double labeling for connexins [Texas Red, Cx26 (
A confocal image of double labeling for Cx26 (FITC) and Cx43 (Texas Red) with DIC imaging in a rat sciatic nerve. Cx26 is localized in the innermost layer of the perineurium, whereas Cx43 is distributed in multiple layers. Arrow indicates Cx43 labeling in the epineurium. epi, epineurium; end, endoneurium. Bar = 5 μm.
Confocal images of double labeling for connexins [Texas Red, Cx26 (
Immunohistochemical localization of Cx26 in normal (
Double Labeling
To elucidate the localization of Cx26 and Cx43 in the perineurium in more detail, we performed double labeling for Cx26 or Cx43 and laminin, a component of the basal lamina of perineurial cells. The rabbit polyclonal anti-laminin antibody revealed linear labeling in multiple layers of the perineurium and around neural fibers (Figure 2), in accordance with previously reported findings (Jaakkola et al. 1993; Patton et al. 1997). Cx26-positive spots were confined to the innermost layer of the perineurium (Figure 2A, whereas Cx43 spots were localized in all the layers of the perineurium labeled with laminin (Figure 2B.
To assess whether both connexins are co-localized at the same gap junction plaque, double labeling for Cx26 and Cx43 was performed using a rabbit anti-Cx26 antibody (Kuraoka et al. 1993) and a mouse antibody against Cx43. Double labeling for Cx26 and Cx43 showed that Cx26 was localized in the innermost layer of the perineurium, whereas Cx43 was distributed in multiple layers (Figure 3). Although some Cx26-positive spots were found to be in contact with Cx43-positive ones even in a single confocal slice (1-2 μm in thickness), little co-localization of these connexins in the same gap junctional plaque was detected (Figure 3).
Immunolabeling for occludin, an integral tight junction protein, was utilized to investigate the localization relationship between connexins and tight junctions in the perineurium. Occludin-positive punctate spots formed short lines in multiple layers of the perineurium (Figures 4A and 4B) and in endoneurial vessels (not shown). Double labeling for Cx26 and occludin showed different localization of the proteins. Cx26 was present in the innermost layer of the perineurium, whereas the expression of occludin was distributed in multiple layers but not across layers (Figure 4A. In contrast, double labeling for Cx43 and occludin revealed that punctate spots of both proteins combined and formed a continuous linear structure, and that both proteins frequently coexisted in the same linear structure (Figure 4B. At the resolution of confocal microscopy, Cx43-positive spots appeared to be located between occludin-positive spots, and some yellow spots indicating co-localization of Cx43 (red) and occludin (green) were observed. Morphometric analysis of uninjured sciatic nerves showed that the ratio of co-localization (yellow) area to the occludin (green) and to the Cx43 (red) areas was 13.1 ± 2.2% and 36.4 ± 6.1% (mean ± SD), respectively. In contrast, little yellow area was found in double labeling for Cx26 (red) and occludin (green), and the ratio of co-localization (yellow) area to the occludin (green) and Cx26 (red) areas was 0.1 ± 0.2% and 1.2 ± 2.2% (mean ± SD), respectively.
Immunohistochemical localization of Cx32 in normal (
Crush-injured Sciatic Nerves
To examine alterations in the expression of connexins after peripheral nerve crush injury, immunolabeling followed by morphometric analysis for Cx26, Cx32, and Cx43 was performed at various time points after an experimentally induced nerve crush. In the longitudinal sections of uninjured rat sciatic nerves, intact axons were revealed as continuous fibers with anti-PGP9.5 antibody. Four days after injury, the continuous fibers disappeared at the crushed and distal sites. Fourteen days after injury, a few axonal fibers were observed at the crushed site but not at the distal site. At 21 days after injury, many regenerating fibers appeared at the distal site (not shown).
Immunohistochemical localization of proteins in rat sciatic nerve
a -, no expression; ±, weak labeling; +, strong labeling;
∗, labeled in vessels.
Size and frequency of immunolabeled connexin spots in rat sciatic nerve a
aValues are the mean ± SD; ND, not determined.
In the longitudinal sections of uninjured sciatic nerves, punctate Cx26 immunoreactivity was localized to the innermost layer of the perineurium (Figures 5A and 5B), consistent with the results in transverse sections (Figure 1B. The Cx26 localization in the perineurium did not change after crush injury (Figures 5C and 5D). However, morphometric analysis at the distal site of the perineurium after injury revealed that Cx26 frequency increased at 4 days after injury and then decreased, the number of the Cx26-positive signals at 7-28 days after injury being significantly smaller than at 4 days after injury (Figure 8A. The sizes of Cx26-positive spots at 21 and 28 days after injury showed a tendency to become smaller than those of uninjured nerves, although there was no significant difference (p=0.57 for control vs 21 days distal and p=0.16 for control vs 28 days distal) (Figure 8B. No obvious change in the frequency and size of Cx26-immunoreactive spots was found in the perineurium at the proximal site of the injury (Figures 8A and 8B). No clear immunolabeling for Cx26 was detected in the endoneurium and the epineurium of either uninjured or injured nerves.
Cx32 immunoreactivity was restricted to the paranodal regions of the nodes of Ranvier in the endoneurium in the longitudinal sections of uninjured nerves and showed a symmetrical punctate pattern (Figures 6A and 6B). Four days after injury, the normal symmetrical pattern of Cx32 was lost and Cx32-positive spots appeared fragmented at the distal site (Figure 6C. Fourteen days after injury, the crushed and distal regions of injury showed a complete reduction in Cx32 immunoreactivity (Figure 6D. Twenty-one days after injury, small spotty and irregular labeling appeared at the distal site of the endoneurium (Figure 6E. Twenty-eight days after injury, the symmetrical configuration of Cx32 immunoreactivity recovered in the distal segments (Figure 6F. No obvious change in Cx32 was detected at the proximal site of the injury (not shown).
In the longitudinal sections of uninjured sciatic nerves, immunolabeling for Cx43 demonstrated many spots in multiple layers of the perineurium and a small number of spots in the epineurium, but no immunoreactivity in the endoneurium (Figures 7A and 7B), similar to that in the transverse sections (Figure 1F. No alterations in Cx43 immunolabeling (Figures 7C-7G) or in the frequency and size of Cx43-positive signals (Figures 8C and 8D) were detected in the perineurium during the entire period after injury. However, Cx43-positive spots appeared in the endoneurium after crush. Four days after injury, some Cx43-positive spots were detected in the endoneurium not only at the proximal site adjacent to the injury but also at distal sites (Figure 7C. Fourteen days after injury, the number of Cx43-positive spots in the endoneurium gradually attenuated (Figure 7D. DIC images of Cx43 immunolabeling and adjacent sections stained with hematoxylin-eosin indicated that Cx43-positive spots in the endoneurium were localized in/around axonal and myelin debris (not shown). Twenty-one days after injury, a few Cx43-positive spots were detected in the endoneurium, some of which were localized in/around the debris and others of which were located in fibrous structures in the DIC image (Figures 7E and 7F). At this stage, regenerating axonal fibers were first found in these fibrous structures at the distal site of the injury by immunohistochemical labeling for PGP9.5 (not shown). Twenty-eight days after injury, the number of Cx43-positive spots in the endoneurium further decreased, but the spots existed outside neural fibers (Figure 7G.
Morphometric analysis demonstrated that the frequency of Cx43-positive spots at the distal site of the endoneurium markedly increased at 4 days after injury and then significantly decreased at 21 or 28 days after injury compared with that at 4 days after injury (Figure 8E. The endoneurial Cx43 that appeared at the proximal and distal sites after crush injury showed no significant change in size after injury (Figure 8F.
Double labeling for connexins and occludin was also performed to determine whether the relationship among these proteins is changed after crush injury. No obvious changes in the immunolabeling for occludin (Figures 5E and 7H) and laminin (not shown) were observed in the perineurium at either the proximal or the distal sites after crush injury. Double labeling for Cx26 and occludin showed that the relationship between them did not change after injury, i.e., they were rarely co-localized in the perineurium in crushed nerves (Figures 5E and 5F). In contrast, double labeling for Cx43 and occludin revealed that Cx43 frequently coexisted with occludin and that they were sometimes co-localized in multiple layers of the perineurium even after injury (Figures 7H and 7I). Morphometric data confirmed no significant change in the extent of co-localization of Cx43 with occludin at either the proximal or the distal sites after injury (Figure 9).
Immunohistochemical localization of Cx43 in normal (
Discussion
In this study we demonstrated immunohistochemically that Cx26, Cx32, and Cx43 exhibited different expression patterns in intact peripheral nerves and that their expression was differentially modulated after crush injury. In intact nerves, Cx26-positive spots were unevenly distributed in the innermost layer of the perineurium, whereas abundant punctate Cx43 immunolabeling was detectable in multiple layers of the perineurium and a small number of punctate spots were also detected in the epineurium. Cx32 was confined to the paranodal regions of the nodes of Ranvier. Double labeling for Cx43 and occludin revealed that both proteins frequently existed in the same linear structures and were sometimes co-localized. During peripheral nerve regeneration after crush injury, the expression of Cx32 was altered and Cx43-positive spots appeared in the endoneurium. Morphometric data on Cx26 in the perineurium after injury showed a rapid increase and a subsequent decrease in frequency. There were no obvious changes in the expression of occludin and Cx43 in the perineurium.
The frequency and size of gap junctional spots in rat sciatic nerves after crush injury. The frequency of Cx26 (
We have good grounds for believing that our immunohistochemical data shown in this study exhibit credibility. For each connexin, Cx26, Cx32, and Cx43, identical results were obtained with different antibodies, including mouse monoclonal and rabbit polyclonal antibodies. Furthermore, confocal laser scanning microscopy enabled us to acquire simultaneous high-resolution images of two different connexins, or one connexin with occludin or laminin, as well as DIC images.
As far as we are aware, this is the first report on Cx26 localization in the perineurium of intact peripheral nerves. Before this study, although Yoshimura et al. (1996) reported that a low amount of Cx26 mRNA was detected in rat sciatic nerve using Northern blot analysis, it was unknown where Cx26 protein was localized in the peripheral nerve. We have clarified here that Cx26-positive spots were confined to the innermost layer of the perineurium but rarely co-localized with Cx43 and occludin. Zhao and Spray (1998) recently reported that low expression of Cx26 was seen in myelin sheaths, Schmidt-Lantermann incisures, and on membranes throughout Schwann cells in a study using teased mouse sciatic nerve fibers. Although we tried to detect Cx26 in the endoneurium with three different anti-Cx26 antibodies, we could obtain no clear labeling for Cx26 except for very weak labeling in the myelin sheath by use of highly concentrated polyclonal antibodies (not shown). Therefore, it appears reasonable that, if Cx26 is expressed in intact endoneurium, the amount is small.
For Cx43, our present study showed abundant punctate Cx43 immunolabeling in multiple layers of the perineurium and a few punctate spots in the epineurium, but no Cx43 immunoreactivity in the endoneurium except in endoneurial vessels. These results are consistent with previous reports showing that Cx43 immunoreactivity is localized to perineurial cell layers, the epineurium, and blood vessels of the endoneurium (Chandross et al. 1996), and that sciatic fibroblasts obtained by culturing the perineurium, not myelinating Schwann cells, express Cx43 mRNA (Scherer et al. 1995). In contrast, our results are inconsistent with reports by Yoshimura et al. (1996) and by Zhao and Spray (1998) showing that Cx43 was present in the cytoplasm of Schwann cells in human peripheral nerves and that Cx43 was found at the nodes of Ranvier in teased sciatic nerve fibers of adult mice. One explanation for this discrepancy may be the difference in animal species used in these studies.
Morphometric data on the co-localization of occludin and Cx43 in rat sciatic nerve after crush injury. The extent of co-localization of occludin (green signals) with Cx43 (red signals) was evaluated as the percentage of yellow area in the green area. Values represent the mean ± SD. No significant difference among each group is found (Scheffé test). cont, control.
Our results of double labeling for Cx43 and occludin demonstrated that punctate spots of both proteins combined and formed continuous linear structures, with some co-localization in multiple layers of the perineurium, suggesting a close relationship between gap junctions and tight junctions among perineurial cells in each perineurial layer. The early freeze-fracture study by Akert et al. (1976) described well-developed tight junctions in the perineurial cells of various adult vertebrates. Tight junctions have been considered an important component of the perineurial barrier (Ghabriel et al. 1989; Beamish et al. 1991). Other freeze-fracture studies showed that gap junctions were localized in maculae occludentes or along zonulae occludentes in the perineurium (Reale et al. 1975; Breathnach and Martin 1980; Beamish et al. 1991; Schiavinato et al. 1991). The present data indicate that gap junctions associated with tight junctions in the perineurium were made of Cx43, not Cx26. Furthermore, these results suggest that gap junctional intercellular communication among perineurial cells in each layer is important for the regulation of perineurial cell function, including modulation of structure and function of tight junctions.
Our results of the localization of Cx26, Cx43, and occludin in the perineurium also indicate that perineurial cells are not uniform in type through the multiple layers and may be made up of different populations. This is because some perineurial cells in the innermost layer possess Cx26, whereas outer perineurial cells in multiple layers express Cx43 and occludin but not Cx26. The difference in the properties of perineurial cells may be reflected in our present data on the expression of connexins after crush injury, showing that Cx26 in the innermost layer of the perineurium was changed during axonal regeneration, whereas Cx43 in the multiple layers did not alter. In addition, perineurial cells in the innermost layer may have different functions in addition to the diffusion barrier via tight junctions. For example, these innermost cells might receive signals from the endoneurial environment and regulate functions in the outer perineurial cells.
Cx32 is the first gap junction protein discovered in peripheral nerves (Bergoffen et al. 1993). Our present results on Cx32 localization are consistent with those previously reported, showing that Cx32 is expressed in myelinating Schwann cells and is localized to the paranodal regions of the nodes of Ranvier and Schmidt-Lantermann incisures (Bergoffen et al. 1993; Miyazaki et al. 1995; Scherer et al. 1995; Chandross et al. 1996). The significance of Cx32 in the maintenance of structure and function of peripheral nerves was proved by the findings that mutations in the Cx32 gene result in X-linked Charcot-Marie-Tooth disease (Bergoffen et al. 1993; Chance and Pleasure 1993; Fairweather et al. 1994) and that Cx32 null-knockout mice exhibit a late-onset progressive peripheral neuropathy (Anzini et al. 1997).
In our crush injury study, changes in the expression of Cx32 and appearance of Cx43 spots were observed in the endoneurium at the distal site of the injury. After the loss of PGP9.5 expression in degenerated nerve fibers, the expression of Cx32 was reduced. Cx32 expression was recovered in the distal segments after the neural fibers regenerated, a process that appears to correspond to remyelination after nerve crush, as previous reports have indicated (Scherer et al. 1995; Chandross et al. 1996; Satake et al. 1997). Our results show that crush injury caused a rapid increase in Cx43 immunoreactivity in and around degenerating debris in the endoneurium, consistent with previous results obtained by Northern blotting (Yoshimura et al. 1996), in situ hybridization, and immunohistochemistry (Chandross et al. 1996). One can safely state that the rapid appearance of Cx43 in the endoneurium is closely related to an acute inflammatory reaction, although the origin of Cx43-expressing cells remains to be elucidated.
In contrast, in the perineurium, crush injuries caused no apparent alterations in the expression of occludin and Cx43 at the distal site, whereas the frequency and size of Cx26-positive spots in the innermost layer were modulated during axonal regeneration. These findings indicate that the endoneurial environment may influence the innermost layer of the perineurium but not the outer layers. The maintenance of a close relationship between occludin and Cx43 in the outer perineurial cells may be relevant to barrier functions against stimuli from the endoneurium.
In conclusion, our present study demonstrates that Cx26, Cx32, and Cx43 are expressed differently in various types of cells in intact peripheral nerves, and that these expressions are differentially modulated during axonal regeneration after crush injury. Because upregulation of Cx46 has been reported after nerve injury (Chandross et al. 1996) and the presence of other connexins in myelinating Schwann cells was indicated from the study of Cx32 null-knockout mice (Balice-Gordon et al. 1998), more connexins are surely involved in the peripheral nervous system under physiological and pathological conditions.
Footnotes
Acknowledgements
Supported in part by a Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Science and Culture.
We gratefully acknowledge the generous gift of the polyclonal anti-connexin26 antibody from Dr Y. Shibata (Kyushu University School of Medicine; Fukuoka, Japan) and that of the monoclonal anti-connexin32 (6-3G11) antibody from Dr A. Takeda (Department of Medical Biology, Ehime University; Ehime, Japan). We thank Mr D. Mrozek for advice on English usage.