Reaction of glia, particularly microglia, is universal to injuries and diseases of the brain and spinal cord. The glial reaction is at least initially a defense mechanism of the central nervous system (CNS). It could, however, subsequently become a mediator or facilitator of the disease process, a mechanism that has been postulated for many neurological disorders. For example, reactive microglia are thought to be the mediator of the death of neurons in the penumbra area of ischemic lesions (Dirnagl et al., 1999). In neurodegenerative and demyelinating diseases, activated microglia are found to be associated with senile plaques in Alzheimer's disease or demyelinated axons in multiple sclerosis (Benveniste, 1997; Frautschy et al., 1998; Stoll and Jander, 1999; Akiyama et al., 2000; Pocock and Liddle, 2001).
Despite the clear association between reactive microglia and disease processes, the precise consequence of microglial activation on neural cell survival and function in vivo remains uncertain. In culture, stimulated microglia produce cytotoxins such as free radicals, excitatory amino acids, and inflammatory cytokines (Frei et al., 1992; Giulian et al., 1993; Merrill et al., 1993), as well as trophic factors (Mallat et al., 1989; Zhang and Fedoroff, 1998; Nakajima et al., 2001), suggesting that activated microglia play dual roles. The outcome is likely the combinatorial effect of both detrimental and beneficial effects, which will depend on the nature of the pathology and the stage of the disease process. Defining the role of reactive microglia and identifying ways to modulate activated microglia in vivo will be instrumental to the design of treatment for neurological disorders (McGeer and McGeer, 1995; Selkoe, 1999).
In vivo experimental neuronal lesion models such as peripheral nerve axotomy have provided significant insights into the nature of microglial reaction in situ (Kreutzberg, 1996). However, how these models relate to neurodegenerative processes is unclear. We have recently identified a myelin mutant rat called Long Evans shaker (les) that carries a mutation in the myelin basic protein gene (O'Connor et al., 1999), which results in severe dysmyelination in the first month and later loss of scattered abnormal myelin (Kwiecien et al., 1998). Accompanying this is a progressive reaction of microglia (Zhang et al., 2001). In the postnatal 4 weeks, reaction of microglia is characterized by increased cell proliferation, enlarged cell volume, and upregulated antigenic expression such as complement 3 receptor. Further activation of microglia is seen around 4 weeks of age by expression of major histocompatibility complex (MHCII), mRNAs of inflammatory cytokines, and nitric oxide synthase with phagocytosis of myelin debris (Zhang et al., 2001). This chronic yet graded microglial reaction is secondary to oligodendroglial pathology and dysmyelination and does not involve a classical inflammatory or immune response. It hence mimics the graded microglial response seen in chronic neural degenerative disorders and provides a unique model to examine the effect of activated microglia on neural cell survival and function. It also offers a system to test the efficacy of compounds that may modulate activated microglia in vivo. To date, there are no known substances that can effectively suppress the function of activated microglia in chronic disorders. Recent reports indicate that a tetracycline derivative, minocycline, can prevent the induction of microglial activation in experimental ischemia, Parkinson's disease, and experimental allergic encephalomyelitis (Yrjanheikki et al., 1998, 1999; Tikka et al., 2001; Popovic et al., 2002; Wu et al., 2002). We have explored the effect of minocycline on chronic preexisting activated microglia in les rats with the survival of and myelination by transplanted glial progenitors as an indicator of functional outcome.
MATERIALS AND METHODS
Culture of Oligodendroglial Progenitors
Oligodendroglial progenitor cultures were established from neonatal Long Evans rat brain according to Zhang et al. (1998) and expanded as free-floating oligospheres, aggregates of purified oligodendroglial progenitors or preprogenitors. The oligosphere cells were transduced with retrovirus PG13 expressing β-galactosidase (Miller et al., 1991) and selected with antibiotics G418 as described (Zhang et al., 1998). Before transplantation, oligospheres (between passages 10 and 15) were triturated into single cell suspension with a Pasteur pipette and the cells were prepared in Hanks balanced salt solution (HBSS) at a concentration of 50,000 live cells/μl. For analyses of acute cell death after transplantation, the disaggregated oligosphere cells were stained with Hoechst 33342 (5 μg/ml for 10 min), washed extensively (six times of wash in PBS), and resuspended in HBSS with propidium iodide (50 μg/ml; to label dead cells after transplantation).
To investigate the effect of minocycline on the proliferation and differentiation of oligodendroglial progenitors, oligospheres were dissociated into single cell suspension by trituration and then plated onto polyornithine-coated coverslips at a density of 2,500/cm2 in medium consisting of Dulbecco's modified essential medium (DMEM)/F12, N1 supplement, and 30% B104 neuroblastoma-conditioned medium for proliferation and the same medium without B104-conditioned medium for differentiation (Zhang et al., 1998). Minocycline (Sigma) was added to the culture media at various concentrations. Cell morphology and density were monitored daily under a phase contrast microscope and immunostaining with O4 and O1 (GC) antibodies; cell proliferation was assessed by bromodeoxyuridine (BrdU) incorporation (Zhang et al., 1998).
The les rats aged 1, 4, and 8 weeks were recipients of transplanted glial progenitors. The les rats were identified by a high-frequency whole body tremor at 12–14 days of age or older. Neonatal les rats were determined by identifying the mutated DNA sequence using polymerase chain reaction (PCR) as described previously (O'Connor et al., 1999). Similarly, normal rats were also confirmed by PCR analysis to exclude carriers.
Rats were anesthetized with isoflurane mixed with oxygen. A laminectomy was performed at the thoraco-lumbar junction of the spinal column. One microliter of cell suspension or HBSS (as sham-transplant) was slowly injected into the exposed spinal cord with a glass micropipette (30 μm bore size) held in a micromanipulator. The injection site was marked with sterile powdered charcoal before the wound was sutured. The operated animals were maintained in the laboratory animal facility of the School of Veterinary Medicine.
Tissue Preparation and Immunohistochemistry
Rats were deeply anesthetized with an overdose of pentobarbital and then perfused with 4% paraformaldehyde prepared in phosphate-buffered saline (PBS). Spinal cords were dissected and postfixed for 2–4 h. In those rats in which the cells survived and made myelin, a white patch was seen in the dorsal column that was measured and photographed. They were then rinsed with PBS and stained with X-gal as described (Zhang et al., 1998). The blue X-gal reaction product that appeared in the dorsal spinal cord was measured and photographed before the spinal cord was trimmed. For ultrastructural examination, tissue blocks were postfixed in osmium tetroxide, washed in buffer, and processed through graded alcohols before embedding in Epon (Kwiecien et al., 1998). Semithin sections were cut and stained with paraphenylene diamine (PPD) or toluidine blue, and thin sections were further stained with uranyl acetate and lead citrate before being examined with a Philips 410 electron microscope. For immunohistochemistry, tissues were cryoprotected in 30% sucrose overnight. Serial free-floating sections of 12 μm thickness were collected in a 96-well plate and every four consecutive sections were immunostained with the following antibodies according to the protocol detailed previously (Zhang et al., 1998; O'Connor et al., 1999): antiglial fibrillary acidic protein (GFAP; rabbit IgG; Dako) at 1:10,000; OX42 (anti-CD11b; mouse IgG; SeroLab) at 1:1,000; OX6 (anti-MHCII; mouse IgG; SeroLab) at 1:1,000; antimyelin basic protein (MBP; mouse IgG; Boehringer-Manhheim) at 1:500.
Assessment of Cell Death by DNA Fragment End Labeling
Cryosections of the grafted spinal cord that were already labeled with Hoechst (blue) and propidium iodide (red) were examined for apoptotic nuclei using a DNA Fragmentation Detection Kit (Oncogene Research Product, Boston, MA) following the manufacturer's instruction. The staining was revealed by FITC-conjugated streptavidin (green).
RNA Extraction and RT-PCR
The les rats with or without pretreatment with minocycline were anesthetized with an overdose of pentobarbital and perfused transcardially with ice-cold Ringer's solution for 5 min to remove blood. Total RNAs were extracted from the les rat spinal cord, reverse-transcribed, and amplified according to the protocol detailed previously (Zhang et al., 2001). PCR cycles were from 25 to 35.
Transplant-Derived Myelination at Initiation and Progression of Microglial Activation
The effect of progressive microglial activation on grafted neural cells was examined by transplanting oligodendroglial progenitors into the spinal cord of les rats at postnatal day 7, the time when microglial activation begins (Zhang et al., 2001). Two weeks after transplantation, grafted cells produce myelin as seen by the presence of a white patch along the otherwise semitransparent dorsal column of the amyelinated spinal cord. The length of the white patch along the dorsal spinal cord was about 2 mm (2.1 ± 1.2; n = 6), similar to that observed after transplanting oligodendroglial progenitors into the myelin-deficient mutant rat (Zhang et al., 1998). By 5 weeks posttransplantation, the white patch was substantially longer (4.4 ± 1.4 mm; n = 5). The white streak corresponded directly to the localization of the transplanted cells at both time points as determined by X-gal staining (Fig. 1A and B). Immunohistochemical staining through the grafted area indicated that numerous MBP+ myelin sheaths surrounding the blue cells occupied the whole dorsal funiculus, whereas the lateral and ventral columns of the spinal cord were completely devoid of MBP staining (Fig. 1C). The dorsal column beyond (rostral or caudal to) the blue area was also devoid of myelin sheaths. The presence of myelin sheaths in the area with transplanted cells was confirmed in PPD-stained 1 μm coronal sections (Fig. 1D). Rats that received a sham-transplant (HBSS) had no gross evidence of myelin and were negative on X-gal staining.
Loss of Grafted Cells and Lack of Myelination During Peak Microglial Activation
Microglial reaction in the les rats peaks at 1 month of age with the highest density of activated microglia, expression of MHCII and mRNAs of inflammatory cytokines and iNOS (Zhang et al., 2001). When the oligodendroglial progenitors were transplanted into the spinal cord of 1-month-old les rats, evidence for the presence of grafted cells and myelination was found in only 1 of the 10 grafted rats at 2 and 4 weeks after transplantation (Table 1). Similarly, no evidence of myelination was observed after the cells were transplanted into the spinal cord of 2-month-old les rats when the activated microglia persist (Table 1). Immunosuppression with cyclosporin A (10 mg/kg body weight; i.p. daily, beginning 24 h prior to transplantation) did not improve the transplant outcome, suggesting that the absence of grafted cells was not due to immune rejection. In contrast, grafted cells survived and spread for 2 mm in length 2 weeks following transplantation into the spinal cord of their 1-month-old normal littermates (not shown).
Table 1. Myelination after transplant into 1- and 2-month-old rats
Myelin after transplantation
1/5 indicates that one out of five transplanted rats shows myelination as indicated by white streak and Xgal staining.
To identify the time window during which the grafted cells were lost, 1- to 2-month-old les rats that received glial transplants were examined 1, 2, 3, and 7 days following transplantation. One day following transplant, grafted X-gal+ blue cells were found along the needle track in the spinal cord. However, by 3 days, there were no X-gal+ cells in the transplanted area. We therefore labeled the cells with Hoechst 33342 and examined the transplant site 2 days posttransplantation. The majority of the Hoechst-labeled oligodendroglial progenitors were stained with propidium iodide (Fig. 2A–C). We then examined the cell death by TUNEL staining. As shown in Figure 2A–C, very few propidium iodide-stained cells were labeled by TUNEL staining, suggesting that the grafted cells had died shortly after transplantation but not by apoptosis.
Pretreatment With Minocycline During Peak Microglial Reaction Resulted in Graft-Derived Myelination
To overcome the destructive aspect of activated microglia, recipient rats were treated with minocycline (45 mg/kg, i.p.) daily, modified from the protocol of Yrjanheikki et al. (1998) by reducing the dosage from twice a day to once a day. In all the rats that received grafts and minocycline treatment at the same time as transplantation, no myelin formation was observed 2 and 4 weeks later (Fig. 3A, Table 2). In contrast, rats that received minocycline treatment 3 days before transplantation showed cell survival at 2 days (Fig. 2D–F) and myelination 2 and 4 weeks following transplantation (Fig. 3B, Table 2). The length of myelin patch was 1–2 mm (1.75 ± 0.9 mm; n = 4) along the dorsal funiculus 2 weeks posttransplant, which was greater at 4 weeks posttransplant (3.2 ± 0.8 mm; n = 5). Coronal sections of the grafted cord indicated that the myelin was located primarily in the dorsal funiculus and the majority of axons in the dorsal column were myelinated (Fig. 3C and D). Ungrafted cord was completely devoid of any myelin sheaths at this age (Fig. 3E). Electron microscopy analyses indicated that most axons were wrapped with thin compact myelin sheaths, although some axons were completely unsheathed or surrounded by glial processes (Fig. 3F).
Table 2. Myelination by transplanted glia in minocycline-treated les rats*
Myelination after transplantation
The les rats were treated with minocycline at the same day (day 0) or 3 days before (day −3) transplant. Myelination was assessed at 2 and 4 weeks posttransplant by gross myelin appearance and histology in the number of animals out of the total grafted rats in each group.
Effect of Minocycline on Oligodendroglial Progenitors and Microglia In Vivo and In Vitro
To understand how minocycline pretreatment leads to successful transplantation, we examined the effect of minocycline on the survival, proliferation, and differentiation of oligodendroglial progenitors. Two days after transplantation, very few Hoechst-labeled oligodendroglial progenitors were stained with propidium iodide in contrast to extensive cell death of the grafted cells in untreated group (Fig. 2D–F). Few propidium iodide-stained cells were labeled with TUNEL (Fig. 2F). Hence, pretreatment of the recipient rats with minocycline protected the grafted cells from damage, which led to myelination.
The direct effect of minocycline on oligodendroglial progenitors was examined in culture. At dosages from 0.1, 1, 5, 10, to 20 μg/ml, minocycline did not affect the cell density, percentage of BrdU-incorporated cells when grown in proliferation conditions, or proportion of O4+ oligodendrocytes when grown in differentiation medium. However, at dosages higher than 40 μg/ml, minocycline decreased the density of oligodendroglial progenitors. Morphologically, cells did not exhibit branched processes as in normal differentiation cultures and died mostly after 3 days of minocycline treatment. The above observations suggest that the protective effect of minocycline on grafted cell survival and myelination is achieved not via its direct effect on oligodendroglial progenitors.
The alternative mechanism is that minocycline modulates the spinal cord environment to promote the grafted cell survival. Immunohistochemical examination of the spinal cord tissues of minocycline-pretreated les rats showed no obvious change in microglial morphology and the intensity of CD11b or MHCII staining. Similarly, there was no change in astrocyte morphology or staining intensity for GFAP (not shown). RT-PCR analyses showed that the expression of IL-1β, TNF-α, and iNOS mRNAs was not altered by the treatment with minocycline (Fig. 4).
The present study, employing the paradigm of neural transplantation into an environment with progressive microglial activation, demonstrates that the consequence of activated microglia on the transplanted cells depends on their functional states. Moderately activated microglia, as indicated by morphological transformation and increased CD11b expression but without upregulation of MHCII and iNOS (Zhang et al., 2001), do not appear to interfere with the survival and function of the grafted oligodendroglial progenitors. A similar degree of myelination is achieved by grafted oligodendroglial progenitors in other myelin mutants such as myelin-deficient rats where there is mild microglial reaction (Zhang et al., 1998). In fact, a substantially larger area of the dorsal column was myelinated in the following 2 weeks even though there was a progressive glial reaction in the les CNS. Hence, reactive glia may facilitate myelination by grafted cells during that period. However, fully activated microglia, as indicated by expression of MHCII, iNOS, and inflammatory cytokines and phagocytosis (Zhang et al., 2001), are apparently detrimental to the survival of implanted glial progenitors. Such a phenomenon has been previously demonstrated in culture where moderately activated microglia (due to cell culture procedure) support neuronal survival and neurite outgrowth yet stimulated microglia (by lipopolysaccharide) actively kill neurons (Zhang and Fedoroff, 1996). Although the transplant outcome is the combinatorial effect of both activated microglia and astrocytes, reactive astrocytes are generally regarded as supportive for cell survival unless they result in scar formation (Giulian, 1993). In fact, in other animal models such as aged myelin-deficient rats and canine shaking pups in which the astroglial reaction is much more prominent than a microglial reaction, transplanted glial progenitors survive and produce myelin (Archer et al., 1997). Therefore, the inhibitory effect on grafted cell survival in les rats is likely attributed to the activated microglia.
The failure of myelination by grafted oligodendroglial progenitors at the peak of gliosis is due to the death of grafted cells within 3 days. We have previously shown that a substantial proportion of grafted cells will die within several hours after transplantation whether the cells are grafted into normal or dysmyelinated environment and that the cells die by apoptosis (Zhang et al., 1999). In the les rats, however, grafted cells are eliminated 2–3 days after transplantation, and TUNEL analysis indicates that few dead cells are labeled. This suggests that the grafted cells are actively damaged during peak gliosis. While the molecular mechanism that leads to the death of grafted cells needs investigation, inflammatory cytokines and free radicals produced by the activated glial cells during peak gliosis (Zhang et al., 2001) may play a role.
As activated microglia are negative to cell survival and function, finding a way of reversing or reducing their cytotoxicity will have therapeutic value. During our analyses on the effect of activated microglia on grafted cell behavior, Yrjanheikki et al. (1998) reported that minocycline suppresses the activation of microglia in an ischemic mouse model. While in that study minocycline appears to block the induction of microglial activation in the breach of blood-brain barrier, our present study demonstrates that minocycline also reduces the cytotoxicity of the already activated glia, most likely microglia in the presence of an intact blood-brain barrier. It should be noted that degree of myelination and the thickness of myelin sheaths are less than those obtained with transplantation at the neonatal stage. This is likely due to the different CNS environment. Grafted oligodendroglial progenitors proliferate, migrate, and/or produce myelin more in the neonatal stage than they do at the stage when myelination is completed. Studies in other animal models such as ischemia, Huntington's disease, amyotrophic lateral sclerosis, and Parkinson's disease suggest that activated microglia, rather than reactive astrocytes, are the target of minocycline (Yrjanheikki et al., 1998, 1999; Chen et al., 2000; Tikka et al., 2001; Wu et al., 2002). Our recent observations also indicate that minocycline not only blocks the induction of experimental allergic encephalomyelitis in Lews rats but also reduces progression or relapses of the disease when it is applied after the disease onset, at least partly through the regulation of microglia/macrophage activities (Popovic et al., 2002). The ineffectiveness of minocycline treatment at the time of transplantation in les rats contrasts the effective treatment of minocycline at the onset of ischemia in mice (Yrjanheikki et al., 1998), which can be explained by the preexisting chronic microglial activation in the les rats.
The mechanism by which minocycline facilitates grafted cell survival/myelination is likely an indirect effect through the downregulation of activated microglia as minocycline itself, in a wide range of dosages, does not promote the survival and proliferation of cultured oligodendroglial progenitors. Minocycline, or other members of tetracycline family, has been shown to inhibit the production of inflammatory cytokines IL-1, caspase 1/3, and expression of iNOS by cultured microglia (Chen et al., 2000; Tikka et al., 2001; Wu et al., 2002). However, we have failed to demonstrate that minocycline affects the mRNA expression of IL-1β, TNF-α, and iNOS. We did not observe obvious change in reactive microglia at the cellular level either. Our inability to detect clear changes in inflammatory elements may be due to the fact that the change in microglial cells is diluted when we use the whole spinal cord tissue as opposed to the purified cells that are stimulated by LPS artificially (Tikka and Koistinaho, 2001; Tikka et al., 2001). Minocycline has been shown to modulate other cellular functions in addition to inflammatory cytokines (Chen et al., 2000; Brundula et al., 2002; Popovic et al., 2002; Zhu et al., 2002). Furthermore, minocycline not only affects transcription and translation but also influences cellular function at the posttranslational level, such as scavenge of NO (Amin et al., 1996). Hence, a small degree of change in the expression of individual molecules would be difficult to discern in the whole tissue by semiquantitative RT-PCR or immunohistochemistry. Yet, summation of such changes may reduce the cytotoxicity to the threshold that renders the environment to be supportive to the grafted cells. In any case, the present study suggests that minocycline modulates the glial environment to promote neural cell survival and function.
The primary defect of les rats is the mutation in the mbp gene (O'Connor et al., 1999). This genetic defect causes abnormality in oligodendrocytes and dysmyelination (Kwiecien et al., 1998), which in turn leads to secondary microglial reaction (Zhang et al., 2001). The graded microglial reaction in the les CNS does not involve the breach of blood-brain barrier and classical inflammation (Zhang et al., 2001). The phenotypes of microglial activation in the les rats resemble those in many neurological diseases such as Alzheimer's disease and Parkinson's disease (McGeer and McGeer, 1995; Selkoe, 1999; Akiyama et al., 2000). Given the difficulty in assessing the causative effect of the secondarily activated microglia in neurodegenerative models, myelination in the les rats is a good indicator of functional consequence. Hence, the les rat offers a useful model to study the function of activated microglia in neurodegeneration and to evaluate compounds that may regulate microglial activity.
The authors appreciate J. Beck and J. Dean for technical assistance.