Abstract
PMD (Pelizaeus-Merzbacher disease) is a rare neurodegenerative disorder that impairs motor and cognitive functions and is associated with a shortened lifespan. The cause of PMD is mutations of the PLP1 [proteolipid protein 1 gene (human)] gene. Transgenic mice with increased Plp1 [proteolipid protein 1 gene (non-human)] copy number model most aspects of PMD patients with duplications. Hypomyelination and demyelination are believed to cause the neurological abnormalities in mammals with PLP1 duplications. We show, for the first time, intense microglial reactivity throughout the grey and white matter of a transgenic mouse line with increased copy number of the native Plp1 gene. Activated microglia in the white and grey matter of transgenic mice are found as early as postnatal day 7, before myelin commences in normal cerebra. This finding indicates that degeneration of myelin does not cause the microglial response. Microglial numbers are doubled due to in situ proliferation. Compared with the jp (jimpy) mouse, which has much more oligodendrocyte death and hardly any myelin, microglia in the ***overexpressors show a more dramatic microglial reactivity than jp, especially in the grey matter. Predictably, many classical markers of an inflammatory response, including TNF-α (tumour necrosis factor-α) and IL-6, are significantly up-regulated manyfold. Because inflammation is believed to contribute to axonal degeneration in multiple sclerosis and other neurodegenerative diseases, inflammation in mammals with increased Plp1 gene dosage may also contribute to axonal degeneration described in patients and rodents with PLP1 increased gene dosage.
INTRODUCTION
PMD (Pelizaeus-Merzbacher disease) is an X-linked neurodegenerative disorder of the CNS (central nervous system) caused by missense mutations, duplications and deletions of the PLP1 [proteolipid protein 1 gene (human)] gene. Duplications/triplications of the native PLP1 gene are most frequent, accounting for nearly 70% of males that have PMD (Garbern et al., 1999). Clinically, nystagmus, psychomotor developmental delay, spasticity, hypotonia and ataxia present within the first 2 months of life in connatal forms of the disease (Hodes et al., 1993). Investigators have concluded that a major biological function of PLP protein is to maintain the structural integrity of the myelin sheath (Duncan et al., 1987; Boison et al., 1995). However, the protein has other functions that include regulation of oligodendrocyte cell death (Knapp, 1996; Yang and Skoff, 1997; Nadon and West, 1998; Cerghet et al., 2001). There are currently no cures or treatment options available for this rare disease, and the cellular and molecular mechanisms that lead to the phenotypes are poorly understood.
Transgenic rodents with Plp1 [proteolipid protein 1 gene (non-human)] increased gene dosage are valid models for PMD because they show the same behavioral deficits that include a shortened lifespan as PMD patients (Mastronardi et al., 1993; Kagawa et al., 1994; Readhead et al., 1994; Bradl et al., 1999). At the cellular level, many different abnormalities have been described in Plp7 transgenics that include not only defects in oligodendrocytes but also neurons. Mice with Plp1 increased gene dosage exhibit major oxidative phosphorylation defects that include 50% reductions in ATP brain levels and decreased mitochondrial membrane potentials in neurons as well as oligodendrocytes (Hüttemann et al., 2009). Axonal degeneration is also a peculiar feature of rodents that overexpress the Plp1 gene (Anderson et al., 1998; Griffiths et al., 1998). However, the sequence of events that begins with increased Plp1 gene transcription and leads to axonal degeneration is unknown. In the present study, we describe microglial activation and up-regulation of pro-inflammatory cytokines in the cerebrum of a transgenic mouse that overexpresses native Plp1 (Plp1tg). The relevance of an inflammatory response cannot be underestimated because inflammation is a known major contributor to axonal damage and subsequent neurological sequelae in many neurodegenerative diseases.
Microglial cells are considered resident macrophages of the CNS that act as immune defence cells. In normal mature brains, microglial cells are responsible for immune surveillance (Block and Hong, 2005) and exhibit a distinctive morphology in normal mammals. Microglia can also become activated and turn into motile, amoeboid cells and play a major role in initiation and progression of neurodegenerative diseases (Dickson et al., 1991; Thomas, 1992; McGeer et al., 1993; Kreutzberg, 1996; Gonzalez-Scarano and Baltuch, 1999). These cells undergo dramatic morphological alterations upon brain injury or immunological stimuli transforming from resting microglia into activated amoeboid microglia (Kreutzberg, 1996). For example, in the demyelinating disorder multiple sclerosis, microglia proliferate and increase lysosomal activity around active sites of demyelination (Matsumoto et al., 1992).
This is the first study that shows microglial activation in a transgenic mouse that overexpresses native Plp1. Microglial activation in the spinal cord of the jp (jimpy) mouse, a mouse with a Plp1 missense mutation, characterized by its severe hypomyelination has been described (Vela et al., 1995). The optic nerve and spinal cord of Plp1tg mice are characterized by demyelination, but the striatum and cortex are surprisingly normal (Readhead et al., 1994). Therefore it would not be expected that microglia exhibit intense reactivity in areas of the cerebrum if microglial activation is based solely on hypomyelination/demyelination. To our surprise, we found that microglial activation begins before myelination commences in the normal cerebrum. Accordingly, interactions between oligodendrocytes and axons rather than frank demyelination and/or hypomyelination must lead to axonal degeneration (Garbern et al., 2002). Thus strategies to modulate oligodendrocyte function in vivo are necessary to alter the course of PMD if myelin abnormalities are not the direct cause of certain debilitating diseases (Garbern et al., 2002).
To support the hypothesis that an immune response is occurring in Plp1tg mice, we found extensive microglial activation accompanied by an increase in many cytokines that are also elevated in CNS disorders. For example, TNF-α (tumour necrosis factor-α) and IL-6 (interleukin-6) are elevated in several CNS disorders (Navikas and Link, 1996; Vladic et al., 2002; Baraczka et al., 2004). Recent studies demonstrate that TNF-α exacerbates focal ischaemic brain injury and that blocking endogenous TNF-α is neuroprotective (Barone et al., 1997). What role these cytokines play in exacerbating oligodendrocytes and neuronal dystrophies in animals with Plp1 increased gene dosage is yet to be determined.
MATERIALS AND METHODS
Animals
Male and female 7, 19-27, 50 and 70 DPN (day postnatal) B6CBA mice (Jackson Laboratory, Bar Harbor, ME, U.S.A.) and male and female Plp1tg mice (Readhead et al., 1994) were used for genotyping, immunocytochemistry, cytokine assays and electron microscopy. Plp1tg mice aged 30 days and older were first identified by tremors and seizures followed by extraction of DNA from tails with the Qiagen DNA easy kit (Qiagen, Valencia, CA, U.S.A.). DNA from both tremoring Plp1tg mice and presumptive Plp1tg mice younger than 30 days were used to measure copy number. Copy number was calculated with the δδCt (Ct is threshold cycle value) method using a wild-type male with one Plp1 gene as the calibrator (Regis et al., 2005). Plp1tg mice that had at least two extra copies of the native Plp1 gene and exhibited tremors and seizures were studied. QPCR (quantitative PCR) was performed with an Eppendorf Realplex PCR system using SYBR Green (Stratagene, La Jolla, CA, U.S.A.). Seventeen-19 DPN jp mice (jp+/+Ta; Ta is tabby) were obtained from our breeding colony housed in the Division of Laboratory Animal Resources, a federally approved animal facility. Plp1/jp/Ta female carriers (jp+/+Ta) and males (+Ta) were purchased from Jackson Laboratory. jp mice at 17-19 DPN were identified by tremors and seizures and then genotyped by PCR following the procedure in Hüttemann et al. (2009). All mice were housed in DLAR animal facilities that are under the direct supervision of IACUC, a federally approved institutional committee; all procedures were approved by the Wayne State University IACUC.
Immunocytochemistry
Mice were anaesthetized with an intraperitoneal lethal dose of chloral hydrate. All animals were perfused intracardially with 4% (w/v) paraformaldehyde and their brains were removed and placed in buffer. Then, 50 μm sections were cut with a Vibratome, washed with PBS, treated with 10% Triton for 1 h, treated with 3% H2O2 for 25 min, blocked with 10% (v/v) goat serum for 30 min, and incubated in polyclonal anti-Iba1 (ionized calcium-binding adaptor molecule 1) antibody (Wako, Richmond, VA, U.S.A.) diluted 1:500 overnight in 0.1 M PBS at room temperature (20°C). The next day, sections were rinsed in PBS and incubated in goat anti-rabbit IgG HRP (horseradish peroxidase; Jackson Immuno-Research)-conjugated antibody diluted 1:100 and detected using 0.5 mg/ml DAB (diaminobenzidine) without or with 0.2 mg/ml cobalt chloride (Sigma-Aldrich, St. Louis, MO, U.S.A.).
For combined Iba1 and BrdU (bromodeoxyuridine) immunostaining of 8 DPN mice, mice were injected with 100 μg of BrdU/g of body weight intraperitoneally and perfused with 4% paraformaldehyde 40 h later. The interval from injection to sacrifice permits microglia/microglial precursors to divide and differentiate. Then, 50 μm sections were cut with a Vibratome, washed in PBS, treated with 2 M HCl for 30 min at 37°C, incubated in 0.1 M sodium borate twice for 30 min at room temperature, treated with 3% H2O2, incubated in PBS-blocking buffer (PBS containing 0.2% BSA, 0.2% milk and 1.0% Triton) for 30 min and incubated with anti-BrdU (BD Biosciences, San Jose, CA, U.S.A.) diluted 1:100 at 4°C overnight. Sections were rinsed in PBS, incubated in anti-mouse HRP (Jackson ImmunoResearch) conjugated and developed with DAB. Sections were then rinsed in PBS and processed for Iba1 immunocytochemistry as previously described and developed using DAB without cobalt.
An F4/80 antibody (Jackson ImmunoResearch) was used to detect macrophages in 3 months Plp1tg cervical spinal cords. Then 50 μm sections (prepared as described above) were rinsed in PBS, treated with 3% H2O2 for 20 min, blocked with 10% goat serum for 1 h and incubated in polyclonal anti-F4/ 80 antibody diluted 1:100 overnight in 0.1 M PBS at 4°C. The next day, sections were rinsed in PBS and incubated in goat anti-rat IgG HRP-conjugated antibody (Jackson ImmunoResearch) diluted 1 100 and detected using 0.5 mg/ml DAB (Sigma-Aldrich). As a positive control, spinal cord sections from EAE (experimental allergic encephalomyelitis) mice injected with MOG (myelin oligodendrocyte glycoprotein) were stained alongside Plp1tg sections.
Quantification of microglia
Microglia in the cortex and striatum of Plp1tg 19-27 DPN and B6CBA 23-27 DPN brain sections were counted using a Leitz (Wetzlar, Germany) Laborlux microscope. Only cobalt enhanced sections were used for counting. Cells were counted with a × 50 oil objective and a × 10 eyepiece containing a 1 cm2 grid divided into 1 mm squares. The grid was placed over the dorsal cortex, immediately superior to the corpus callosum and adjacent to the interhemispheric sulcus, the ventral-lateral cortex, adjacent to the most ventral tip of the corpus callosum, and the striatum, immediately inferior to the corpus callosum and adjacent to the lateral ventricle (Figure 6). For each animal, three matched transverse sections at the level of the anterior commissure were used. The number of cells for each area was summed and the averages determined. Microglia were only counted if the nuclear membrane was completely intact as visualized by focusing completely through the section.
(A)A50 μm Vibratomed hemi-section from a 30 DPN mouse cerebrum at low magnification shows boxed areas used for quantification of Iba1 immunostained cells. Cells were counted at ×500 directly under the microscope at the level of the anterior commissure (arrow). Microglial cells with distinct nuclei were counted in the three areas selected: (A) dorsal cortex, (B) striatum and (C) ventral-lateral cortex. (B) Quantification of Iba1-immunostained cells in the cerebrum of 19-27 DPN Plp1tg and 23-30 DPN B6CBA mice. Values are means ± S.E.M. for Iba1 cells in the three areas selected, three sections per animal were averaged, and the number of animals from both groups is shown in parentheses. The two-tailed Student's t test was used to compare Plp11tg mice with B6CBA mice; P<0.0001.
Plastic embedded tissue
Animals were perfused intracardially with 4% paraformaldehyde and 0.5% glutaraldehyde. Cerebra were dissected and 200 μm thick sections were embedded in Araldite plastic following routine laboratory procedures (Bessert and Skoff, 1999). Semi-thin sections (1 micron) were stained with Toluidine Blue. Ultrathin sections were cut, stained with uranyl acetate and lead citrate and examined with a Jeol 1010 transmission electron microscope. Negatives were scanned at 1600 dpi (dots per inch) using Adobe Photoshop.
Mouse inflammatory cytokines and receptors PCR array
The Mouse Inflammatory Cytokines and Receptors RT2 Profiler™ PCR Array (SABiosciences, Frederick, MD, U.S.A.) that profiles the expression of 84 key genes involved in a focused inflammatory pathway was run to screen for inflammation in a Plp1tg mouse. Total RNA was isolated from the brain of a Plp1tg mouse and a C57BL/6 control at 27 DPN using TRIzol® reagent (Invitrogen). RNA was purified using RNeasy® Min Elute™ cleanup columns (Qiagen). RNA (0.5-1 μg) from each sample was used to synthesize the reverse transcriptase (cDNA) first strand following the instructions of RetroScript (Ambion, CA, U.S.A.). The cDNA is used as a template in the RT profiler™ PCR Super Array, and QPCR was performed following a two-step cycling programme on Eppendorf Mastercycler® Eppendorf Realplex. The fold change in gene expression was calculated using the δδCt Data Analysis Method; the Plp1tg mouse used had four copies of the Plp1 gene and showed tremors and seizures.
Quantitative real-time PCR analysis of cytokines and chemokines
CCL3 (CC chemokine ligand 3), CCL8 (CC chemokine ligand 8) and CRP (C-reactive protein) (Table 1) were determined by QPCR. Reverse transcriptase was obtained as described above and the reverse transcript was subjected to real-time PCR on the Eppendorf Mastercycler® Eppendorf Realplex in the presence of SYBR Green I dye (Stratagene). A two-step QPCR was performed for 40 cycles (10 min at 95°C, 15 s at 95°C and 1 min at 60°C). Product identity was confirmed by melting point analysis. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) served as an internal reference gene in all the experiments. Real-time PCR was performed in triplicates for both the target gene and GAPDH. The level of gene expression was calculated after normalizing against GAPDH in each sample and is presented as relative mRNA expression. The total cDNA subjected to amplification was calculated using the δδCt method.
List of primers used in QPCR analysis
RESULTS
Microglial morphology of Plp1tg compared with B6CBA mice
Iba1 is a polyclonal antibody that recognizes an ionized calcium-binding adaptor molecule that is only expressed in microglia in the brain (Imai et al., 1996; Graeber et al., 1998; Ito et al., 1998). This antibody has been used to identify microglia in many mammalian species (Tonchev et al., 2003; Yamada et al., 2006). Iba1 immunostained microglial cells are present throughout the grey and white matter of both the normal B6CBA and the Plp1tg mouse cerebrum (Figure 1). Typical microglia in normal cerebra, often referred to as ramified, have characteristic long and thin processes of even calibre (Figure 1A). Their primary processes are usually twice as long as the long axis of the perikaryon. Secondary processes branch acutely from primary processes, and sometimes tertiary processes are distinguishable. Perikarya are characteristically oval-, pear- or triangular-shaped and are easily distinguished from the smoothly contoured round oligodendroglia perikarya and the larger, irregularly shaped astroglial perikarya. The ramified morphology is similar across all age groups that we studied in normal B6CBA mice.
High magnification of a typical ramified microglial cell (A) shows several long, thin processes emanating from an ellipsoid cell body. These primary processes branch into secondary processes (arrows). In areas of the cortex (B) and striatum (C), resting microglia are evenly dispersed throughout grey matter with typical ramified morphology. In the corpus callosum (D), microglia are less abundant than in grey matter, but retain their ramified morphology. The cingulum is at the top of the Figure. Scale bars: (A)10 μm and
Low-magnification pictures show that the distribution of microglia in normal cerebra is evenly spaced, with minimal contact between processes of neighbour microglia in both the cortex (Figure 1B), striatum (Figure 1C) and corpus callosum (Figure 1D).
In the Plp1tg mice, the morphology and dispersion of these cells in the cortical, striatal and callosal regions of cerebra drastically differ from the B6CBA controls (Figures 2A and 2B) in all age groups analysed (Figures 2C-2F). Activated microglia in all Plp1tg age groups are distinguished by coarse and short processes, a notable increase in the amount of these processes, and an enlargement of the cell body within the cortical (Figures 2C and 2E) and striatal (Figures 2D and 2F) regions. In the cortices, microglia are often clustered together with an entanglement of brush-like processes. This arrangement has been previously described and is due to actively proliferating microglia cells as an early response to neurodegeneration (Dissing-Olesen et al., 2007). This pattern is most easily observed between the cortex and the white matter in the ventral cortex. In the striatum, microglial processes run parallel with each other with long straight processes, as if they are following axonal trajectories.
Microglia from B6CBA mice at 50 DPN have similar morphology to microglia at younger ages. Activated microglia are often clustered together with their cell bodies having a fuzzy appearance due to numerous, short, thick processes. In the striatum, more elongated and straighter processes are visible, possibly following axonal trajectories. Microglia exhibit more processes at 30 DPN
Corpora callosa of normal B6CBA mice appear to have fewer microglia than grey matter (Figure 1D). When compared with controls (Figures 1, 2A and 2B), white matter in Plp1tg have an abundance of activated microglia (Figures 3A-3C), with coarse, straight processes that parallel the trajectory of callosal axons.
At 30 and 50 DPN, microglia are more activated in the white matter compared with the 19 DPN. Scale bar, 50 μm.
Microglial morphology of Plp1tg mice changes as the mice age
With age, the activation of microglia increases in the Plp1tg cerebra. In the cortices and striata at 30 DPN (Figures 2E and 2F), microglia have stubby processes and are found in clusters with other brush-like microglia. At 19 DPN (Figures 2C and 2D), the transition from a ramified morphology to a reactive morphology is noticeable as processes shorten in length and diameter. In the striatum at 19 DPN (Figure 2D), cell processes do not follow a particular path as do the 30 DPN (Figure 2F) processes.
In both white and grey matter at 7 DPN, the earliest age studied, activation of microglia has already begun in the Plp1tg cerebra (Figures 4C-4E). While many microglia have a normal morphology, others exhibit shortened and numerous processes. Individual microglia appear to have long, thin processes typical of ramified microglia on one side of the cell and shortened, numerous processes typical of reactive microglia at the opposite side of the cell. Microglia in normal brains at this age (Figures 4A and 4B) are similar to microglia at older ages, having long processes that often end in an expanded ball. At 7 DPN in the normal brain, myelinated fibres are not yet detectable in 1 μm plastic sections (Figure 4F). Myelinated axons have dark rings surrounding an opaque centre and easily visible in semi-thin plastic sections. Therefore the activation of microglia at this age is not due to abnormal myelin structure or demyelination.
Microglia in the cortices (A) and striatum (B) of normal mice already display the characteristic morphology of microglia at older ages. Their long processes sometimes terminate in boutons (arrows) that are less prominent in microglia from older mice. In the cortex (C), striatum (D) and corpus callosum (E), some microglial cells in Plp1tg mice are already activated at this young age, indicated by short, numerous processes emanating from microglia with large perikarya (arrows). Note the absence of long processes on some of these activated microglia. A mix of ramified and activated microglial cells is visible in all three regions. (F) A 1 μm plastic section of a normal 7 DPN B6CBA striatum has no myelin around axons at this age. Scale bars:
Generation of new microglia in 8 DPN Plp1tg mice
In addition to microglia activation beginning at 7DPN, microglia are stimulated to divide at this young age. Combined Iba1 and BrdU immunostaining was performed on Plp1tg cerebra with two or more copies of Plp1. Mice were injected intraperitoneally with BrdU on 6 DPN and killed 40 h later on 8 DPN. Generation of new microglia is abundant because Iba1+/BrdU+ microglia are present in the cerebrum (Figures 5A-5D). Microglia in the metaphase stage of mitosis (Figure 5B) have been detected, indicating that microglia divided approx. 40 h earlier at the time of the BrdU injection and shortly before killing. Figure 5 also shows many BrdU+ cells that are not Iba1+; these cells serve as a control to show that the BrdU-labelled cells are not non-specifically labelled with the Iba1 antibody. Some of these BrdU+ cells are Olig1+, indicating that oligodendrocyte progenitors are stimulated to divide (results not shown).
Mice were injected with BrdU at 6 DPN and killed 40 h later on 8 DPN. Brown chromogen indicates Iba1 cytoplasmic staining and black chromagen indicates BrdU nuclear staining. (A) A double-labelled microglia cell with numerous processes indicates that this cell underwent cell division approx. 40 h before the mouse was killed. Two BrdU+/Iba1−- cells (arrows) are shown in the upper right. (B) An Iba1+ cell that is in mitosis (arrow). (C) Four BrdU+/ Iba1−- cells (arrows), a BrdU−-/Iba1+ cell (arrowhead) and a pair of lightly labelled Brdu+/Iba1+ cells (crossed arrows). (D) A double-labelled microglial cell with one process (arrow). Scale bar, 50 μm.
The number of microglia is increased in Plp1tg mice
Plp1tg mice with increased gene dosage lead to an increase in microglia cell number in both white and grey matter of the cerebrum when compared with B6CBA mice. A low-magnification micrograph (Figure 6A) shows a hemi-section at the level of the anterior commissure at 30 DPN of the mice. For quantification, only cobalt-enhanced sections were used for both control and Plp1tgs. The number of Iba1+ microglial cells was counted directly through the microscope with a × 50 oil objective and a × 10 eyepiece (see the Materials and methods section). It can be seen that 19-27-day-old Plp1tg have significant increases (P<0.0001; Figure 6B) in all three areas counted: Dorsal cortex immediately superior to the corpus callosum and adjacent to the interhemispheric sulcus, the ventral-lateral cortex adjacent to the most ventral tip of the corpus callosum, and the striatum immediately inferior to the corpus callosum and adjacent to the lateral ventricle. The results indicate that the density of microglia cells is higher in Plp1tgs (Figure 6B), with a 76% increase in the dorsal cortex, a 95% increase in the ventral-lateral cortex and 85% increase in the striatum. It is likely that the increase in microglia in Plp1tg mice is even greater than the cell counts because microglia are clustered together and sometimes it is difficult to determine the exact number.
Ultrastructure of microglia
Microglia in Plp1tg mice (Figures 7A and 7B) exhibit the same nuclear and cytoplasmic features as ramified microglia in normal mice. The chromatin forms a dense layer immediately beneath the nuclear membrane and the density of the cytoplasm is intermediate to that of oligodendrocytes and astrocytes (Skoff, 1975). However, the activated microglia in Plp1tg mice often contain degenerating material of unknown origin. They tend to have larger perikarya and thicker processes than ramified microglia. These features probably correspond to the larger cell body and thick coarse processes visualized with Iba1 immunostaining. In addition to the activated microglia present in Plp1tg, small, dark cells are also occasionally present (Figures 7C and 7D). These cells have too little cytoplasm to permit definitive classification. The cell in Figure7(C) has the nuclear features of an oligodendrocyte progenitor; the cell in Figure 7(D) has the nuclear features of a microglial cell or oligodendrocyte as these cell types often directly abut the perikarya of neuronal somata. Conversely, they may be immune cells that have invaded the CNS.
Microglial response in jp mice
jp mice have a missense mutation in the Plp1 gene, and they exhibit an intense astrocytic hypertrophy and hyperplasia that accompanies the virtual lack of myelin (Skoff, 1976). In contrast with the Plp1tg mice, microglial reactivity in jp is restricted mainly to the white matter (Figures 8A and 8B). Furthermore, the Iba1 immunostaining in jp does not appear to be as intense as in the Plp1tgs.
Up-regulation of cytokines
Based on the intense reactivity of microglia in Plp1tg mice, we predicted the up-regulation of classical inflammatory markers. qRT-PCR (quantitative reverse transcription-PCR) expression of different inflammatory markers showed significant increases in two classical inflammatory markers, TNF-α and IL-6; they were significantly increased by 24- and 7-fold respectively at 7-8 weeks after birth compared with controls (Table 2). However, there was no increase in IL-1β or iNOS (inducible nitric oxide synthase) at 7-8 weeks of age. Based on a Mouse Inflammatory Cytokine and Receptor RT2 Profiler™ (SABiosciences) that profiles the expression of 84 key genes, we selected the chemokines CCL3, CCL8 and inflammatory marker CRP for qRT-PCR. Expression of CCL8 was significantly increased by 3.5-fold at 3 weeks and then significantly increased by 15-fold by 8 weeks postnatal. Expression of CCL3 was increased by 4-fold at 3 weeks and only slightly increased by 4.8-fold at 8 weeks postnatal. Similarly, CRP increased modestly from 3 to 8 weeks postnatal compared with normal mice. These results indicate that activation of inflammatory markers is well advanced by 3 weeks of age, and even increases for some markers as the disease progresses.
qRT–PCR analysis of mRNA expression of cytokines and chemokines in brain homogenates of Plp1tg mice
NS, not significant.
The qRT-PCR data shown above confirm the results of a cytokine and receptor Super Array profile performed on a Plp1tg and a wild-type mice at 27 DPN that showed similar increases in message levels for these chemokines (Table 3).
Overview of cytokine and chemokine mRNA expression in brain of Plp1tg mice (with >3-fold change)
The cell types expressing these inflammatory markers and their receptors are unknown but are likely to involve a combination of macroglia, microglia and, possibly, immune cells because markers of T and B immune cells have been detected in the Plp1tg mice (Ip et al., 2006). We investigated whether macrophages invaded the CNS in the Plp1tg mice. The F4/80 surface antigen was easily detected in spinal cord sections of C57BL/6 mice with EAE injected with MOG-(35-55) (Figure 9A). However, spinal cord sections from Plp1tg mice processed alongside the EAE sections showed non-specific staining, mainly in axons (Figure 9B). The picture shown here represents the most intense staining we found in the overexpressors.
(A) The EAE spinal cord shows macrophage infiltration (arrows) indicated by cytoplasmic and process staining of membranes. (B) The picture from the Plp11tg ventral funiculus illustrates the most intense staining observed in these mutants. The axoplasm is non-specifically stained in these mutants. Scale bar, 5 μm.
DISCUSSION
The coding region of the PLP1 gene in humans is identical with the Plp1 gene at the amino acid level in rodents (e.g. Stoffel et al., 1985; Macklin et al., 1987), suggesting that the same mutations in rodents as in humans might exhibit the same cellular abnormalities and behavioural deficits. Indeed, increased gene dosage of the PLP1 gene in humans and Plp1 gene in transgenic rodents exhibit many similar behavioural abnormalities that include ataxia, tremors and often a shortened lifespan (Garbern et al., 1999). At the tissue level, both rodents and humans with increased gene dosage exhibit demyelination and/or hypomyelination. At the cellular level, death of oligodendrocytes and axonal degeneration have been described in humans and in rodents with higher copy number (Gow et al., 1998; Griffiths et al., 1998; Anderson et al., 1999; Cerghet et al., 2001; Skoff et al., 2004a, 2004b). The molecular and cellular sequence of events that cause the tissue and behavioural abnormalities is poorly understood. Since PLP is expressed at high levels only in oligodendrocytes, it is generally assumed that oligodendrocytes and/or abnormal structure of myelin must directly cause all the abnormalities, including axonal degeneration. However, in many neurodegenerative diseases, secondary microglial activation and an inflammatory response have been described in many human diseases including Alzheimer's, Parkinson's and amyotrophic lateral sclerosis (Wojtera et al., 2005). In these neurodegenerative diseases, the contribution of microglia to inflammation and neuronal destruction is a subject of intense investigation. Hints that an inflammatory response was present in a human case of PMD have been described because a young male patient positively responded to corticosteroids (Gorman et al., 2007). In rodents with higher copy number, markers of immune and macrophage cells have been described (Ip et al., 2006; Edgar et al., 2010), but whether, and to what extent, inflammatory molecules are activated has not been examined.
Here, we describe for the first time intense microglial reactivity in Plp1 transgenic mice and the up-regulation of classical inflammatory markers. Although it remains to be determined whether and how the inflammation contributes to the neural degeneration in the Plp1tg mice, the well-known role of cytokine and chemokine activation in other neurodegenerative diseases makes it likely that microglial activation contributes to disease progression.
(A) Microglial staining is strongest in corpus callosum (CC), but weak in cerebral cortex (CeCo) and striatum (St). Microglial staining in grey matter is limited to white matter bundles including cingulum in cortex and Combs fibres in striatum. (B) The microglia in jp corpus callosum do not exhibit the intense reactivity found in Plp1tg mice manifested by short, thick processes. Scale bar, 100 μm.
We find that increased Plp1 gene dosage in line 66 of the Plp1tg mice (Readhead et al., 1994) leads to dramatic morphological changes of microglia. The morphological changes in microglia are as dramatic in the grey matter, if not more so, than in the white matter. From 3 weeks postnatal, shortening and thickening of microglial processes became progressively more pronounced in grey matter, and the number of microglial processes was increased. In the striatum and corpus callosum, microglial processes were increased, but they were longer and followed axonal trajectories. Whether they are interposed between oligodendrocyte processes and axons or directly abut axons is unclear. In active grey matter lesions in multiple sclerosis, activated microglia appose and ensheath apical dendrites, neurites and neuronal perikarya (Peterson et al., 2001). In the cerebral cortex in multiple sclerosis patients, more recent studies have demonstrated microglial activation associated with extensive demyelination (Bo et al., 2003; Kutzelnigg et al., 2005; Vercellino et al., 2005; Gray et al., 2008) but, more interestingly, microglial activation was noted in regions that are not demyelinated (Kutzelnigg et al., 2005). Following an experimentally induced axonal lesion in C57BL/6 mice, activated microglia were proliferating within 3 days and most of the proliferating microglia were aggregating into multicellular clusters (Dissing-Olesen et al., 2007), similar to that observed in our study. Previous studies have examined the distribution and morphology of microglia cells in diseased or injured spinal cord and noted greater microglia activation in white matter as opposed to grey matter (Vela et al., 1995; Vela et al., 1996; Zhang et al., 2008). In jp, the numbers of white matter microglia were 3-5-fold greater than in normal animals but only 60% greater in grey matter than normal animals. We found nearly a doubling of microglia in the grey matter of the Plp1tg mice and, as noted in the Results section, these numbers are likely to be underestimates because microglia are densely clustered together. These and our present study (see below) suggest that microglial activation is caused by other factors/molecules besides degeneration of myelin.
Most importantly, by the first postnatal week, transformation of resting microglia into activated microglia has begun in grey and white matter. Some microglia retain a normal morphology, whereas others are retracting and thickening their processes. The presence of myelin was examined in 1 μm plastic sections of complete transverse cerebral sections, in which individual myelinated axons can be resolved. At this age, axons are still unmyelinated in normal cerebrum. This observation demonstrates that microglial activation is neither due to demyelination nor due to hypomyelination. Of course, the primary cause of the microglial activation must be overexpression of the Plp1 gene, but not necessarily defects in myelination. Both Plp1 mRNA and the protein are detectable in mouse cerebra at this age (Macklin et al., 1991). While the principal function of PLP is often considered to be that of an adhesive molecule in myelin (Boison et al., 1995; Gudz et al., 2002), numerous studies show that PLP and DM20 have other functions (Nadon and West, 1998; Skoff and Knapp et al., 1990; Knapp et al., 1993; Campagnoni and Skoff, 2001) including regulation of cell death (Skoff et al., 2004a, 2004b). Recently, we described major oxidative phosphorylation defects in the Plp1tg mice that lead to 50% reduction of tissue ATP and loss of mitochondrial membrane potential. The cause of these metabolic defects is likely due to the insertion of PLP into mitochondria when the gene is overexpressed (Hüttemann et al., 2009). Activation of microglia by increases in extracellular or exogenous ATP has been described in numerous studies (Lambert et al., 2010; Gyoneva et al., 2009; Monif et al., 2009; Puthussery and Fletcher, 2009). Exogenous ATP, in turn, activates microglial purinergic receptors. However, as presented above, total ATP levels in tissue homogenates are reduced in these mice. The possibility that ATP or its degradation products are secreted into the extracellular space at higher-than-normal amounts by neuroglia is under investigation in our laboratory.
In Plp1tg mice, the number of Iba1 + microglia is doubled by the third postnatal week, due to their endogenous proliferation. Injection of BrdU on 6 DPN and killing 40 h later on 8 DPN followed by Iba1 immunostaining show BrdU+/ Iba1+ microglia with long, branching processes characteristic of resting microglia and microglia with short, slightly thickened processes characteristic of activated microglia. The cell proliferation studies, carried out between 6 and 8 DPN, lend further support to the hypothesis that microglial activation is not due to demyelination/dysmyelination. Our findings agree with the data from jp mouse studies that show that endogenous microglia proliferate. In jp, activated microglial cells in the grey matter of the spinal cord were not associated with neurons, but were intermingled with nerve fibres crossing the grey matter laminae (Vela et al., 1995). The origin of additional microglia cells is unknown, but it is suggested that the main source of macrophage/reactive microglia cells in jp is the proliferation of endogenous microglial cells (Vela et al., 1995).
The intense microglial reactivity in Plp1tg mice suggests activation of multiple cytokines and chemokines. Not surprisingly, mRNA expression of classical inflammatory markers TNF-α and IL-6 were elevated 24- and 7-fold respectively in 2-month-old animals. Since TNF-α and IL-6 are activated by other regulatory molecules, we performed a multianalyte PCR analysis of inflammatory chemokines and their receptors from a 27 DPN Plp1tg mouse. mRNA levels of nine chemokines were elevated by more than 3-fold and several were down-regulated. For some of these chemokines such as CCL3/MIP-1 α (macrophage inflammatory protein), its receptor (CCR1) is also up-regulated.
Many types of immune and glial cells secrete the chemokines and cytokines listed above. CCL3 and its receptor CCR1 have been identified on astrocytes, oligodendrocytes and radial glia after spinal cord injury (Knerlich-Lukoschus et al., 2010). CCL3 has also been identified in microglial cell lines of Sandhoff disease model mice (Kawashita et al., 2009). CRP, usually associated with susceptibility to stroke (Di Napoli et al., 2005), suppresses EAE, suggesting that it has a protective effect in this experimental model of multiple sclerosis (Szalai et al., 2002). CXCL13 (CXC chemokine ligand 13; BCA-1), a chemo-attractant for B lymphocytes, is down-regulated in these mice, suggesting that B-lymphocytes may not be attracted to the CNS. However, lymphocyte and macrophage markers (CD8+ and CD11b+) have been detected in aged heterozygous Plp1tg mice (Ip et al., 2006). We failed to find evidence of macrophages in the Plp1tg mice using the classical F4/80 antibody that we do not think is due to technical issues because in the same immunostaining experiment we identified F4/80+ cells in EAE mice. Our failure to find vascular derived macrophage cells may be due to the differences in age of the transgenic mice used in the different studies. Increases in CD11b+ cells were first detected at approx. 4 months in Plp1tg heterozygous mice and their numbers increased with age (Ip et al., 2006); in contrast, we used 1-2-month-old homozygous mice as they die at approx. 2 months of age. Because the CD11b+ marker is also expressed by endogenous microglial cells (Clausen et al., 2008; Santos et al., 2010; Bokhari et al., 2009), many CD11b+ cells may be microglia. Other studies have shown that microglial precursors derived from embryonic stem cells co-label with Iba1 and CD11b, CD45, F4/80 (Napoli et al., 2009) and other immune cell markers, making it difficult to distinguish between resident and invasive immune cells. Since presentation of antigens by microglia to T-cells requires a period of time (Brown and Neher, 2010), a combination of both microglia and immune cells is likely present in Plp1tg mice, the ratios of innate to humoral cells changing as the rodents age. More important than identifying all the cell types that contribute to inflammation in the Plp1tgs, the molecular events that cause the inflammation and how the different cytokines modulate neuronal function need to be investigated. Importantly, the extent and type of inflammation in PMD patients needs to be examined and compared with rodents that overexpress Plp1.
Footnotes
Acknowledgements
We thank Dr Klaus Armin-Nave (Max-Planck Institute for Experimental Medicine, Göttingen, Germany) and Dr Carol Readhead (California Institute of Techology, Pasadena, CA, U.S.A.) for kindly providing Plp1tg mice.
Funding
This research was supported by the