Experimental Allergic Encephalomyelitis in Connexin 43–Heterozygous Mice

Abstract

Alterations in the expression of gap junction proteins (connexins) have previously been observed in experimental allergic encephalomyelitis (EAE). Demyelinating lesions have significantly decreased Cx43, while recovering lesions have greatly increased Cx43 and increased glial fibrillary acidic protein–expressing astrocytes. This suggests an important role for gap-junctional intercellular communication in astrocytes in the recovery from CNS inflammation. To study the effects of decreased Cx43 expression during acute disease (21 days post-immunization) and in recovering spinal cord tissue (55 days post-immunization) we induced EAE in Cx43 heterozygous and wild-type mice. Mice showed signs of disease by day 10, and signs of recovery by day 25. There were no clinical or pathological differences between heterozygous and wild-type mice in the acute disease stage, except that wild-type male mice had fewer clinical signs of disease. Male mice that were heterozygous for Cx43, and therefore had decreased expression of Cx43, had increased EAE disease severity. All demyelinating lesions had reduced numbers of reactive astrocytes and a significant decrease in Cx43 expression. In the 55-day study, all heterozygous and wild-type mice were clinically improved, showed decreased pathological signs, and showed increased laminin expression, indicative of CNS recovery. Furthermore, all heterozygous mice showed a striking increase in Cx43 expression during recovery, suggesting that the regulatory factors affecting Cx43 expression are still present in mice that have only one wild-type Cx43 allele.

Keywords:

• connexin 43
• Cx43
• experimental allergic encephalomyelitis
• EAE
• multiple sclerosis
• MS
• demyelination
• heterozygote
• gender

INTRODUCTION

Astrocytes play an important role in the central nervous system (CNS) in healthy and disease states. Astrocytes take up extracellular potassium and neurotransmitters after action potentials in order to maintain homeostasis for proper neuronal functioning. Spatial buffering through an astrocytic syncitium is possible because of gap-junction channels that allow for direct intercellular communication (Nagy and Rash 2003). Gap junction channels are formed by the oligomerization of six connexin proteins that form a connexon, which then binds to a connexon from an adjacent cell to form a link between the cytoplasm of both cells (Herve et al. 2005; Nicholson and Bruzzone 1997). There are multiple connexin proteins expressed in the CNS; astrocytes primarily express connexin 43 (Cx43) (Altevogt and Paul 2004; Nagy et al. 2004). During CNS inflammation, astrocytes become reactive (gliosis) and therefore begin to proliferate, migrate, and alter normal functions, including down-regulation of gap-junction proteins (Rouach et al. 2002).

Multiple sclerosis (MS) and its animal model, experimental allergic encephalomyelitis (EAE), are neuroinflammatory diseases that involve loss of the protective myelin sheath surrounding neurons in the CNS. Activated immune cells infiltrate the CNS parenchyma, produce large amounts of various cytokines, such as tumor necrosis factor–alpha (TNF-α), interleukin-1 (IL-1), and nitric oxide (NO), which damage the myelin formed by oligodendrocytes, leading to neuronal damage (Kanwar 2005).

Other studies have shown that altered Cx43 expression is linked to various neurological diseases, including Alzheimer disease (Nagy et al. 1996), Huntington disease (Vis et al. 1998), and ischemic brain injury (Nakase et al. 2004); whether the altered expression is due to the disease, or is a cause of it, is not always known. The only human disease that is known, so far, to be caused directly by a mutation in Cx43 is oculodentodigital dysplasia (ODDD) (Roscoe et al. 2005). Some ODDD patients have abnormal cerebral white matter causing paraparesis (Gutmann et al. 1991). Only recently has altered Cx43 expression been linked to EAE in mice (Brand-Schieber et al. 2005), and in chronic progressive EAE in guinea pigs (Roscoe et al. 2007). We have previously shown that Cx43 expression is decreased in demyelinating EAE lesions and it is dramatically increased during remyelination (Roscoe et al. 2007), which implies that Cx43 is important for recovery from neuroinflammation.

Many publications have shown evidence of increased disease in mice heterozygous for a null mutation in the Gja1 gene encoding the Cx43 connexin protein, suggesting that decreased Cx43 expression is detrimental. Cx43 heterozygotes (Gja1^+/−) had larger myocardial infarctions compared to wild-types (Schwanke et al. 2002) and Gja1^+/−mice have impaired radical oxygen species formation and therefore, less protection in cardiomyocytes (Heinzel et al. 2005). Cx43 heterozygotes are more susceptible to lung adenomas (Avanzo et al. 2004), and have defective coronary artery development (Clauss et al. 2006). Nakase et al. (2004) showed that Gja1^+/−mice have increased infarct size after CNS ischemia, and that these mice have significantly more apoptosis within ischemic lesions, leading this group to conclude that Cx43 plays an important role in astrocyte activation. We hypothesized that Cx43 heterozygous mice would also have alterations in astrocyte activation in EAE, leading to increased disease severity compared to wild-type mice.

A further confounding issue in the progression and severity of MS and EAE involves differences between the sexes. More females have MS than males (Dasgupta et al. 2005; Kantarci et al. 2005; Pozzilli et al. 2003; Reske et al. 2006), but gender data are much more variable in EAE studies. For example, female B10.S mice are resistant to proteolipid protein (PLP)–induced EAE (Reddy et al. 2005), parathyroid hormone–induced hypercalcemia reduces EAE in female but not male mice (Meehan et al. 2005), and vitamin D3 has a protective effect only on female mice (Spach and Hayes 2005), an effect not seen in ovariectomized mice. It has also been reported that gender differences in the susceptibility to EAE depend on the strain of mouse used; for example, there were no differences due to gender observed in EAE severity in C57BL/6 mice (Okuda et al. 2002).

We have previously shown that Cx43 expression is decreased in demyelinating spinal cord lesions and it is increased in remyelinating lesions in female guinea pigs 60 days post-immunization (Roscoe et al. 2007). We hypothesized that Cx43 will also be decreased in demyelinating lesions in C57BL/6 mice, 21 days post-immunization, and that Gja1^+/−C57BL/6 mice should have more severe clinical and pathological signs of disease, with no gender differences, as published by Okuda et al. (2002). Furthermore, since myelin oligodendrocyte glycoprotein (MOG) 35-55–induced EAE via a single immunization is an acute monophasic disease in C57BL/6 mice, with recovery beginning by day 25, Gja1^+/−mice should not recover as quickly, if at all, compared to wild-type mice. In fact, our hypothesis was not confirmed: we found that EAE in heterozygous female mice was not different from wild-type female mice, male wild-type mice had a less severe disease than anticipated, and, surprisingly, both genotypes were able to recover.

METHODS

Experimental Model

Six-week-old male and female C57BL/6 mice that were either heterozygous for the Cx43 null mutation (Gja1^+/−), or were wild-type (Gja1^+/+), were used. Because the null mutation is lethal in the homozygous state (Reaume et al. 1995), Gja1^−/−mice could not be studied. Mice were subcutaneously injected in each flank with a 1 mg/mL mixture of 100 mg of MOG (35-55) peptide in complete Freund's adjuvant (CFA), supplemented with 400 μ g of Mycobacterium tuberculosis. Bordetella pertussis toxin (200 ng) (List Biological Laboratories, Cedarlane, Hornby, ON, Canada) was given by IP injection on day 0 and on day 2 post-immunization. Animals were housed in a controlled-light and -temperature environment, and were weighed daily and scored by a blinded observer. Mice were scored on a five-point scale: 0, no abnormality; 1, limp tail; 2, limp tail and poor righting reflex; 3, one-limb paralysis; 4, two-limb paralysis; and 5, moribund. Spinal cords were removed on either day 21 or 55 post-immunization. Male and female mice were used in the 21-day study (sacrificed prior to recovery): male Gja1^{+ /−} (Het), n = 8, male Gja1^+/+ (WT), n = 7, female Gja1^+/−, n = 7, female Gja1^+/+, n = 6, as well as 16 non-immunized controls (four mice for each gender and genotype). All mice in the 55-day study were female: Gja1^+/−, n = 7, Gja1^+/+, n = 5. All animals used in this study were maintained and handled in accordance with the International Guiding Principles for Biomedical Research Involving Animals and using protocols approved by the Animal Use Subcommittee of the University of Western Ontario.

Tissue Collection

Mice were sacrificed using 540 mg/kg euthanyl forte (Bimedia-MTC Animal Health Inc., Cambridge, ON) by intraperitoneal injection. Spinal cords were removed and divided into cervical-thoracic and lumbar-sacral regions, and fixed in 10% neutral buffered formalin. The fixed spinal cord tissues were embedded in paraffin and cut into 6-μ m sections.

Immunohistochemistry and Immunofluorescence

Transverse paraffin sections were stained with hematoxylin and eosin (H&E), to identify areas of inflammation, and Solochrome R cyanin (ScR), to identify areas of demyelination. Antibodies used included: anti–caspase 3, active (R&D Systems, Minneapolis, MN) used with a biotinylated secondary antibody and an Horse Radish Peroxidase-3-amino-9 Ethyl Carbazole (HRP-AEC) detection kit (R&D Systems, Minneapolis, MN); anti-Cx43 antibody (Sigma, St. Louis, MO) used with an anti-rabbit Fluorescein Isothiocyanate (FITC) secondary antibody (Sigma, St. Louis, MO); monoclonal anti–myelin basic protein (anti-MBP) (Abcam, Cambridge, MA), with an anti–mouse rhodamine secondary (Abcam, Cambridge, MA); a polyclonal anti–glial fibrillary acidic protein (anti-GFAP) (Abcam, Cambridge, MA) to label reactive astrocytes, used with an anti–rabbit FITC secondary (Sigma, St. Louis, MO); and anti-laminin (Sigma, St. Louis, MO), with a biotinylated anti-rabbit IgG (Vector Laboratories, Burlingame, CA), developed with a Vectastain ABC reagent (Vector Laboratories) and diaminobenzidine (DAB) substrate (R&D Systems, Minneapolis, MN). The anti-caspase antibody does not detect the precursor form and it can distinguish apoptotic cells from non-apoptotic cells. The laminin antibody is specific for a non-collagenous connective tissue glycoprotein in basement membranes, commonly used for identifying blood vessels.

Microscopy and Quantification

Cx43 expression levels were viewed using a Zeiss 510 META confocal microscope. A total of 112 images were analyzed in the 21 day study (2 lesions and 2 normal-appearing white matter (NAWM) in each EAE animal (n = 28; 15 heterozygous and 13 wild-type)). Spinal cord sections were from the lumbar and cervical regions with lesions forming randomly throughout the white-matter areas. Cx43 expression levels were quantified using ImageJ software to determine the number of FITC-labeled Cx43 pixels per confocal image (magnification 630X) within lesion areas, compared to NAWM. Values were expressed as the total pixel numbers within each lesion area normalized to the same area as the NAWM, 0.4 mm². Areas of demyelination were also quantified from solochrome-stained sections using ImageJ software by determining the percentage of the total white matter per tissue section that was demyelinated. Fifty-six sections, total, were blindly evaluated (two lesions per animal). The laminin and lesion images were evaluated blindly; 78 total laminin-stained lesions were imaged (two to four images per animal) as well as each adjacent solochrome section. Laminin deposition was scored between 0 (none) to 3 (extensive), and demyelination was scored using the solochrome sections from 0 (no abnormality) to 3 (extensive).

Statistical Analysis

To determine differences in the clinical score between the groups in acute EAE we used a Kruskal–Wallis one-way ANOVA on ranks, and an all-pairwise multiple comparison, Dunn's Method. To compare area of demyelination in each gender between genotypes, we used a one-way ANOVA and Tukey Test. Cx43 pixels in lesions and normal-appearing white matter for each genotype and sex were each compared using a two way ANOVA. Cx43 pixels and lesion areas were correlated with clinical score using a Spearman rank–order correlation. To determine the differences in caspase-positive cells between groups, we used a Kruskal–Wallis one-way ANOVA on ranks and Dunn's Method. Clinical scores for the 55-day study were compared using a Mann–Whitney rank sum test. Laminin deposition and extent of lesion abnormality were evaluated using an ANOVA on Ranks and Dunn's Method. We considered p < 0.05 to be significant.

RESULTS

Immunization of C57BL/6 mice with MOG (35-55), CFA, and pertussis, induced EAE and produced clinical signs of disease beginning on day 10. In the females, the predicted difference between heterozygotes and wild-type mice was not observed. In the males, the heterozygous mice showed similar signs of disease as both female genotypes and the wild-type males had a statistically significant reduction in clinical disease severity compared to all other mice. This contradicts the results of Okuda et al. (2002), who did not find gender-related differences in severity of EAE also in C57BL/6 mice, possibly due to differences in mouse lineage (see Discussion). Heterozygous male mice had the same clinical disease severity as both the heterozygous and wild-type female mice (Figure 1), indicating that loss of one Gja1 allele abolishes the difference in clinical disease severity seen between genders. Mean clinical scores are shown in Figure 1 for day 1 to day 21 post-immunization, with clinical recovery beginning between day 18 and 21.

Figure 1 Wild-type male mice have less severe EAE than female mice. Mean clinical scores for male and female Cx43 heterozygous and wild-type mice are shown for days 1 to 21 post-immunization. Animals begin to show clinical signs of disease by day 10. MOG-induced EAE mice begin to recover between days 18 and 21. There is a significant decrease in disease severity in the wild-type males* compared to all other groups. Data are shown as the mean ± S.E.M. (Kruskal-Wallis one way ANOVA on ranks, p < 0.001, and Dunn's Method, p < 0.05).

The pathological signs of disease were compared in at least two spinal cord tissue sections from each mouse. Infiltration and demyelination were seen in 96% of mice induced with EAE (Figure 2); one wild-type (WT) male mouse did not show any clinical or pathological signs of disease. Solochrome-stained sections show that all groups had demyelinated regions. Likewise, serially sectioned spinal cord images showed a corresponding cellular infiltration in the same areas where demyelination occurred (Figure 2).

Figure 2 Infiltration and demyelination occurs in all EAE-induced mice. Solochrome-stained sections show areas of demyelination in male Gja1^+/−, female Gja1^+/, male WT, and female WT, compared to healthy controls. Hematoxylin- and eosin-stained sections show cellular infiltration in all demyelinated areas in all groups compared to healthy controls. Magnification 250X.

The areas of demyelination were calculated in order to compare any differences between the groups of mice. Two to four sections of spinal cord from each animal were imaged with the demyelinated areas being calculated as a percentage ± SEM of the total white matter for each image using ImageJ software (Figure 3). Male WT mice had less demyelination than female WT and female Gja1^+/− mice (p < 0.050), there was no difference between the genotypes in the females, and male WT mice showed a trend towards a decrease in demyelinated area compared to male Gja1^+/− mice, although this difference was not statistically significant (p = 0.252).

Figure 3 Female WT and Female Het mice had the same amount of demyelination as Male Het mice (p < 0.05), denoted with the “a” above the bars. The Male WT mice had less demyelination than both female groups but had the same as the Male Het mice (p < 0.05), denoted with the “b” above the bars. At least two images from each animal were evaluated blindly and the mean precent of demyelination (± SEM) for each group was determined using ImageJ software.

All tissue sections were immunolabeled with an anti-Cx43 antibody and an anti–myelin basic protein (anti-MBP) antibody (Figure 4) to show the localization of Cx43 (green) with respect to the areas lacking myelin (red). There is an obvious decrease in Cx43 expression in all lesion areas compared to the healthy WT controls. Although male WT mice had fewer and smaller lesion areas, the lesions that were present had substantially decreased Cx43 expression.

Figure 4 Cx43 expression is decreased in all lesions regardless of genotype or sex. All lesions, determined by the decreased myelin (red) are devoid of Cx43 (green) regardless of sex or genotype compared to healthy WT controls. Images are at 630X magnification.

The amount of Cx43 was quantified to determine if any differences occurred between the lesion areas or the normal-appearing white matter (NAWM) areas between the groups (Figure 5). FITC-Cx43 pixels were counted from two lesion areas and two NAWM areas from each animal; a total of 112 images were analyzed. The amount of Cx43 within the lesions was the same in both sexes (p = 0.451), and in both genotypes (p = 0.598). In the NAWM, the expression of Cx43 was not different between gender (p = 0.853) and, as expected, the heterozygous mice expressed significantly less Cx43 than the wild-types (p = 0.001).

Figure 5 The amount of Cx43 within lesion areas was the same in all groups. FITC-Cx43 pixels were counted from two lesion areas and two normal appearing white matter (NAWM) areas from every animal. Mice from both genotypes and sexes had less than 50 pixels/0.4 mm² of Cx43 in lesion areas, whereas, wildtype mice (b in bar graph) had more FITC-Cx43 than the heterozygotes (a in bar graph) in NAWM (two-way ANOVA, p < 0.001).

All mice progressed to various clinical stages of disease (clinical scores were between 0 and 4) throughout the 21 days post-immunization period, with a mixture of clinical scores within each group. For this reason, the Cx43 expression levels were also evaluated with respect to the clinical score in order to observe any temporal expression differences due to the stage of the disease (Figure 6). Lesions formed randomly throughout the spinal cord white matter and these areas of demyelination were quantified for each clinical stage of disease and correlated with the clinical score (Figure 6(a)), showing that lesion size increased with the clinical stage of disease (Spearman correlation coefficient = 0.643, p < 0.001). Cx43 was also quantified within lesion areas at each stage of disease, normalized to 0.4 mm², and correlated with the clinical score (Figure 6(b)). Cx43 is depleted in all lesions regardless of the size of lesion or stage of disease (correlation coefficient = 0.007, p = 0.959).

Figure 6 All lesions lack Cx43 regardless of lesion size or stage of disease. Lesions form randomly throughout the spinal cord white matter areas. (a) Areas of demyelination were quantified within lesion areas for each clinical stage of disease, showing that as the disease progresses, lesion area increases (correlation coefficient = 0.643, p < 0.001). (b) Cx43 was also quantified for lesion areas at each stage of disease and normalized to 0.4 mm², showing that there is no correlation between the amount of Cx43 in lesions compared to the clinical score when the lesion areas are normalized to an area of 0.4 mm² (correlation coefficient = 0.007, p = 0.959). Cx43 is depleted in all lesions regardless of the size of lesion or stage of disease.

Previous research has shown that the presence of gap-junction channels can enhance the rate of apoptosis due to the transfer of apoptotic signals between cells, a phenomenon called the ‘bystander effect’ (Colombo et al. 1995). To determine if the lack of the expected increase in disease severity in the heterozygotes was due to a decrease in programmed cell death, caspase 3 levels were detected and quantified (Figure 7). Active caspase was detected in all lesion areas and was the lowest in the WT male mice (A) (Dunn's Method, p < 0.05), which were the mice that also had the smallest amount of lesion area as well as the lowest clinical scores. The most variability in expression was found in both the WT and heterozygous females (Figure 7(b)). The expected increase in apoptosis in WT mice, expressing more Cx43 due to the presence of both healthy alleles, was not observed.

Figure 7 (a) Male wild-type mice had the fewest caspase-positive cells. (b) Active caspase immunolabeling was detected in all lesion areas, with increased expression in larger lesions, and the male WT mice having the least expression (a) and the female mice of both genotypes having the most variability (ANOVA on ranks, p < 0.001, Dunn's Method, p < 0.050). Caspase was correlated with lesion area (correlation coefficient = 0.702, p < 0.001), data not shown.

We have previously shown that Cx43 expression is considerably increased in remyelinating lesions of EAE guinea pigs (Roscoe et al. 2007), indicating that Cx43 may play an important role in recovery from EAE. In this study, we predicted that heterozygous mice, having less Cx43 before the onset of disease, would have a reduced ability to recover. At 55 days post-immunization, C57BL/6 heterozygous and wild-type mice both recovered clinically from EAE (p = 0.931) (Figure 8). All animals showed clinical signs of disease by day 10, which included hind limb paralysis, and they all began to recover between days 18 and 25, while previously paralyzed mice recovered ambulatory ability by day 35 post-immunization.

Figure 8 Heterozygous and wild-type mice can recover from EAE. Mean clinical scores for female Cx43 heterozygous and wild-type mice are shown for days 1 to 55 post-immunization. Animals began to show clinical signs of disease by day 10. MOG-induced EAE mice began to recover between days 18 and 25 and reached a clinical steady-state by day 35. There was no statistical difference between heterozygous and wild-type mice (Mann–Whitney Rank Sum Test, p = 0.931).

Spinal cord sections were double-labeled with anti-Cx43 and anti-MBP antibodies to determine any possible differences between the genotypes. Both heterozygous and wild-type female mice had increased Cx43 in recovering lesions (Figure 9). Furthermore, there is no observable difference in the expression levels of Cx43 within the recovering lesions of the wild-type and heterozygous mice, indicating that factors regulating Cx43 expression are still intact in the heterozygous mice.

Figure 9 Heterozygous and wild-type mice have increased Cx43 in recovering lesions. Spinal cord tissue was labeled for Cx43 (green) and MBP (red) (630X). Both heterozygous and wild-type mice have increased Cx43 expression in recovering lesions compared with healthy controls. The heterozygous mice have increased Cx43 expression in remyelinating lesion areas that is qualitatively similar to the wild-type mice, even though the heterozygous mice only have one functioning Gja1^+/− allele.

Spinal cord sections were double-labeled with an anti–glial fibrillary acidic protein (anti-GFAP) antibody and anti-MBP antibody to show the relative appearance of reactive astrocytes (gliosis) in demyelinating and remyelinating lesions (Figure 10). There were fewer GFAP-expressing astrocytes present in the acute phase of EAE, as compared with the substantial increase in reactive astrocytes in the recovering mice of both genotypes; this is consistent with the relative expression of Cx43 in both types of lesions.

Figure 10 Fewer GFAP-expressing astrocytes are present in demyelinating lesions and more are present in recovering lesions. Spinal cord tissue was labeled for GFAP (green) and MBP (red) to show the presence of reactive astrocytes within demyelinating versus recovering lesions. Demyelinating lesions for both genotypes have fewer reactive astrocytes, and recovering lesions of both genotypes have increased GFAP-expressing astrocytes. Images of WT animals shown. Magnification 630X.

We found that laminin expression, indicative of CNS recovery, is increased in recovering lesions (Figure 11), compared to demyelinating lesions regardless of genotype, which implies that heterozygous mice were recovering pathologically as well as clinically. The laminin immunoreactivity was limited to meningeal and vascular structures in healthy and 21 day heterozygous and wild-type mice, whereas 55 day mice of both genotypes showed substantial laminin deposition in the recovering lesions (Table 1).

Figure 11 Laminin expression is increased in recovering mice compared to acute and healthy lesions in both genotypes. Spinal cord tissue was stained with an antibody against the extracellular matrix protein laminin, which is expressed in higher amounts in the recovering lesions of both genotypes, compared to healthy spinal cord and demyelinating lesions. Magnification 250X.

TABLE 1 Laminin expression is significantly increased in all recovering mice.

	Laminin	Lesion Severity	Spearman
Healthy	0 ± 0	0 ± 0
Acute	0.78 ± 0.15	1.36 ± 0.21	−0.17
Recovering	1.82 ± 0.25	0.86 ± 0.14	0.79**

*p < 0.05.

**p < 0.001.

DISCUSSION

Cx43 is the most abundant connexin in astrocytes and research has shown that the presence of astrocytes in a demyelinated lesion plays a significant role in the ability of the CNS to remyelinate (Blakemore et al. 2003). It has been shown that EAE mice have decreased Cx43 expression in areas where there is a high level of monocyte infiltration (Brand-Schieber et al. 2005). Our previous studies in the EAE guinea pig have shown that Cx43 is decreased in chronic demyelinating lesions which are 60 days post-immunization, whereas drug-induced remyelination leads to significantly increased Cx43 expression in recovering lesions (Roscoe et al. 2007). Our current data indicate that Cx43 is decreased in acute demyelinating lesions, 21 days post-immunization, where there are fewer GFAP-expressing astrocytes, and recovering lesions have more reactive astrocytes, increased Cx43 expression, and increased laminin expression, in both wild-type mice and mice that have reduced Cx43 expression due to the loss of one Gja1 allele.

Immunization of C57BL/6 mice induced EAE in 96% of the mice studied, where only one male wild-type mouse did not show any signs of disease. Clinical signs, including a limp tail and poor righting reflex, began in these animals by day 10 but the anticipated difference between heterozygotes and wild-type mice was not observed. During EAE disease progression, pro-inflammatory cytokines such as IL-1, TNF-alpha, and IFN-γ are secreted by resident microglia and infiltrating macrophages and T cells. Pro-inflammatory cytokines cause astrocytes to decrease Cx43 expression (Hinkerohe et al. 2005; John et al. 1999; Koulakoff et al. 2003; Meme et al. 2006), which likely leaves the astrocyte free of attachments to other cells, and which therefore enables them to proliferate and migrate to the site of inflammation. It is possible that during this initial phase of disease, astrocytes had not yet begun to up-regulate GFAP expression, or that some of the apoptotic cells observed in these early lesions might have been astrocytes. Either way, both heterozygous and wild-type mice showed minimal Cx43 expression in acute lesions and they both showed similar clinical and pathological signs of disease, with the exception of the male wild-types.

Our results showed that female mice had more severe EAE than males, but when the males had less Cx43 (the heterozygous mice), they had the same disease severity as the female mice. Our data contradict one study that showed no difference between male and female EAE in C57BL/6 mice (Okuda et al. 2002), which is possibly due to differences in the C57BL/6 lineages. The Cx43 heterozygous mice used in our study were bred onto a C57BL/6 background from the original mixed background (Reaume et al. 1995) by back-crossing with Harlan C57BL/6 mice for a minimum of nine generations, but can not be considered a “pure” strain.

There have been other studies showing that males have decreased EAE disease severity compared to females, depending on the strain of mouse studied. For example, male SJL mice have less severe EAE disease and also secrete less of the pro-inflammatory cytokine, IFN-γ, and secrete more of the anti-inflammatory cytokine, IL-10 (Bebo, Jr. et al. 1999). Female MBP-induced EAE mice are more susceptible than males that secrete more IL-10 (Dalal et al. 1997). It has also been shown that female mice produce more inducible nitric oxide synthase (iNOS), IL-6, and TNF-α than males (Dasgupta et al. 2005), and female Lewis rats with increased progesterone have increased EAE severity (Hoffman et al. 2001). Sex differences have been reported in several neurological diseases, including multiple sclerosis (MS), stroke, and neurodegenerative diseases such as Parkinson disease (PD), Alzheimer disease (AD) and amyotrophic lateral sclerosis (ALS) (Czlonkowska et al. 2005).

Our results show that the temporal expression of Cx43 was the same in all groups throughout all stages of disease. Although the amount of demyelination increased as EAE progressed, all lesions were virtually devoid of Cx43 expression, regardless of genotype or gender. The male wild-type mice had smaller lesions but those lesions were also devoid of Cx43.

In our experiments on recovery, we have shown that both heterozygous and wild-type mice clinically recover from EAE and both genotypes have greatly increased Cx43 expression, indicating that astrocytes were possibly aiding in the re-establishment of extracellular homeostasis. Astrocytic Cx43 expression allows the astrocytes to form a syncytium that enhances their ability to buffer extracellular ion and neurotransmitter concentrations (Nagy and Rash 2003). Blocking astrocytic gap-junctional communication increases neuronal injury, and augmentation of gap-junctional communication may contribute to neuroprotection (Naus et al. 2001). Our data show a substantial increase in Cx43 in recovering animals, both heterozygous and wild-type, which suggests that Cx43 is important for the CNS to restore equilibrium. Whatever factors cause Cx43 to increase in astrocytes are still functioning in mice that have only one wild-type Gja1 allele. These regulatory factors have yet to be elucidated; however, this up-regulation is probably due to the presence of anti-inflammatory cytokines, such as interleukin-4 (IL-4), interleukin-10 (IL-10), or interferon-beta (IFN-β), which are present during recovery from EAE. Future studies are required to clarify the factors that up-regulate Cx43 expression.

Previous research has indicated that the extracellular matrix protein laminin may play a protective role in CNS injury and may be particularly important for repair of the blood–brain barrier (Milner and Campbell 2002; Tate et al. 2007). We looked at laminin expression in our studies for two reasons. Firstly, we wanted to show further evidence of pathological recovery from EAE, and, secondly, we wanted to compare laminin expression in the Cx43 heterozygotes and wild-types. Our data show that laminin is increased in recovering lesions of both genotypes.

It is interesting that male mice have the mildest clinical and pathological EAE disease unless they are deficient in Cx43. It is also interesting that heterozygotes up-regulate Cx43 expression and recover at the same rate and same clinical level as wild-type mice. The increased presence of astrocytes that express GFAP and Cx43 in the recovering lesions of both genotypes leads us to believe that astrocytic cellular communication is important for recovery from neuroinflammation.

This work was supported by an operating grant from the Multiple Sclerosis Society of Canada (to S. J. Karlik) and from the Canadian Institutes of Health Research (to G.M. Kidder). W. A. Roscoe is supported by a Postgraduate Scholarship from the Natural Sciences and Engineering Research Council of Canada.