Background
Multiple sclerosis (MS) is a debilitating neurological condition resulting from immune-mediated demyelination [
1]. Various etiologies increase the risk of developing MS, including genetic and environmental origins, but the exact causes are unclear [
2]. Several different murine models exist that partially recapitulate different aspects of MS and demyelination [
3]. Cuprizone is a copper-chelator that impacts cell metabolism and leads to demyelination, and if continued, eventual oligodendrocytic and neuronal death. Cuprizone is administered via food for a defined time period before returning the mice to standard chow to allow remyelination to occur. While the cuprizone model maintains an intact blood-brain barrier with no T cell infiltration, it extensively activates microglia and macrophages [
4‐
6]. The cuprizone model is increasingly used because of its reversibility and reproducibility, reviewed in [
7,
8].
Cuprizone treatment leads to demyelination in the corpus callosum (CC). The CC in turn forms the roof of the subventricular zone (SVZ), a neural stem cell niche that generates neurons and glia throughout life [
9]. The cuprizone model therefore enables investigation into the SVZ response to demyelination and remyelination in the CC. A recent fate-mapping study demonstrated that SVZ cells contributed to remyelination in the CC after cuprizone-induced demyelination [
10]. These cells produced thicker myelin than oligodendrocytes derived from local oligodendrocyte progenitor cells (OPC) [
10]. These findings are consistent with previous research showing SVZ cells can differentiate into oligodendrocytes constitutively as well as after demyelination [
11‐
13] and indicate SVZ cells migrate to the demyelinated CC in MS [
13]. OPCs may have broader reparative effects throughout the CNS than SVZ cells, but SVZ-derived progenitors contribute to repair around the lateral ventricles. This is clinically relevant since it has been known for close to a century that MS lesions frequently emanate from the ventricular surface [
14].
The SVZ thus provides a conveniently positioned nest of stem cells that could benefit MS. Interestingly, data from the past several years suggest inflammation is uniquely regulated in the SVZ, often exhibiting features dramatically different to the CC or other nearby regions. We found that even in the absence of injury, CD45 levels and microglial proliferation were significantly higher in the SVZ than in adjacent regions [
15]. We recently showed the SVZ expresses higher levels of chemokines than the nearby cerebral cortex [
16]. These chemokines rose markedly in the SVZ after Theiler’s murine encephalomyelitis virus (TMEV) injections which induce demyelination. However loss of Gal-3 blocked these increases, suggesting a pivotal role in the inflammatory response in the TMEV model of demyelination [
16]. Perhaps because of these differences, the SVZ can exhibit inflammation in response to brain insults unlike adjacent regions. Traumatic brain injury induced massive inflammation in the CC, but all manifestations of inflammation remained stable in the SVZ [
15]. In contrast, TMEV specifically targeted the SVZ and led to greater inflammation there compared to surrounding regions [
16,
17]. We had reported in a preliminary study that microglial density was maintained in the SVZ while rising in the CC after cuprizone [
17]. We therefore asked if inflammation was differentially regulated in the SVZ compared to the CC after cuprizone treatment. Initial studies suggested that inflammation can be detrimental to neurogenesis and brain repair [
18,
19]. Further studies have shown however that microglia can enhance neurogenesis in vitro and in vivo [
20,
21]. Thus, it is important to determine context-dependent inflammation in the SVZ and how this affects the reparative potential of this largest pool of endogenous brain stem cells.
Another clue that inflammation is uniquely regulated in the SVZ is that the pro-inflammatory protein galectin-3 (Gal-3) is homeostatically expressed in the SVZ and RMS although elsewhere in the central nervous system it is only detectable after injury [
22]. Gal-3 has been implicated in MS; it is expressed at high levels in active lesions around the lateral ventricles, and it modulates animal models of MS [
16,
23]. Gal-3 was initially termed Mac-2 because it was identified in activated macrophages [
24]. However, in the SVZ, it is expressed by astrocytes and ependymal cells and serves to maintain speed and rostral directionality of SVZ neuroblast migration, thereby positively contributing to olfactory bulb neurogenesis rates [
22]. Gal-3 expression increases leukocyte infiltration into the SVZ in the EAE and TMEV demyelination models, leading to a worse phenotype [
16,
25]. Gal-3 has also been reported to increase oligodendrocyte differentiation in vitro and enhance myelination in vivo [
26] and loss of Gal-3 inhibited oligodendrocyte maturation and microglial activation after cuprizone [
27]. Here, we hypothesized that cuprizone-induced demyelination would cause SVZ inflammation and that Gal-3 would regulate SVZ progenitor after cuprizone treatment.
Methods
Animals
129Sv wild-type mice (Gal-3
+/+
) were obtained through the University of Oxford Biomedical Services Specific Pathogens Free Breeding Unit. Gal-3 knockout mice on a 129Sv background (Gal-3
−/−
) were obtained from Françoise Poirier’s laboratory (Institut Jacques Monod, Paris; Colnot et al., 1998). Gal-3
+/+
and Gal-3
−/−
mice were bred to produce heterozygote mice (Gal-3
+/−
), which were subsequently mated with one another to produce Gal-3
+/+
and Gal-3
−/−
littermate controls. For experiments using adult animals, these animals were used. For experiments using postnatal animals, the littermate controls were bred with other mice of the same genotype to produce entire litters of Gal-3
+/+
or Gal-3
−/−
pups. Animals were maintained in individually ventilated cages on 12-h light/dark cycles with free access to food and water. Procedures were performed with University of Oxford Research Ethics Committee approval in accordance with the Animals (Scientific Procedures) Act of 1986 (UK). All efforts were made to minimize animal suffering and distress.
Bromodeoxyuridine injections
To create label retaining SVZ cells, bromodeoxyuridine (BrdU) was administered via intraperitoneal (IP) injection at 50 mg/kg. It was given in six 12-hourly doses during the first 3 days of cuprizone administration.
Cuprizone administration
Cuprizone was administered at 0.2 % to 8-week-old male and female mice ad libitum in chow (International Product Supplies). Treated mice received cuprizone chow for 3 or 6 weeks to cause demyelination, or 6 weeks followed by 10 days of control chow to induce remyelination. Control mice received the same chow without cuprizone. The 6-week cuprizone period was chosen as oligodendrocytes decrease greatly by that time, while the speed of oligodendrogenesis should have peaked after ~10 days of remyelination [
28].
Fluorescent immunohistochemistry
Brains for immunohistochemistry were removed following IP pentobarbitone overdose and transcardiac perfusion with cold normal saline and 4 % paraformaldehyde (PFA). They were postfixed in 4 % PFA overnight and then cryoprotected in 30 % sucrose in 0.1 M phosphate buffer. They were frozen in dry ice, stored at −80 °C and then sectioned into 30-μm coronal sections using a freezing microtome. Fluorescent immunohistochemistry was performed as described previously [
29]. Primary antibodies are outlined in Table
1. BrdU antigen retrieval was achieved with 1 M HCl at 38.5 °C for 1 h.
Table 1
Antibodies used in study
BrdU | 1:500 | Sheep | Abcam | ab1893 |
CD45 | 1:500 | Rat | Chemicon | CBL1326 |
Dcx | 1:100 | Goat | Santa Cruz | sc-8066 |
Gal-3 | 1:100 | Rabbit | Santa Cruz | sc-20157 |
Gal-3 | 1:100 | Rat | Santa Cruz | sc-23938 |
GFAP | 1:400 | Rat | Life Technologies | 13-0300 |
Iba1 | 1:400 | Goat | Abcam | Ab5076 |
Il1-β | 1:100 | Rabbit | AbD Serotec | AAM13G |
Mash1 | 1:200 | Mouse | BD Pharmingen | 556604 |
MBP | 1:100 | Goat | Santa Cruz | sc-13914 |
Olig2 | 1:1000 | Rabbit | Chemicon | AB9610 |
PHi3 | 1:400 | Rabbit | Upstate | 06-570 |
Microscopy
Epifluorescence images were obtained using a Leica DMIRB microscope with a Hamamatsu C4742-95 digital camera or a Leica DMR microscope with a Leica DFC-500 digital camera. Images were acquired in Openlab software (Improvision) and processed in Volocity 4 software (PerkinElmer), or acquired in Leica Firecam 3.4.1 software. Confocal images were obtained on a Zeiss LSM 710 laser scanning confocal microscope using the Z-stack and tile functions as appropriate. They were analyzed in LSM Image Browser 4.2 (Zeiss). The majority of imaging, including all images for threshold analysis, used constant camera settings within a given experiment.
Image analysis
All image acquisition, analysis and quantification, was performed blinded to the genotype and treatment, except for the initial acquisition of CC and cortical myelin basic protein (MBP) images. The sections analyzed were between Bregma 0.6 and 0.2 mm. The most caudal sections where just anterior to crossing of the anterior commissure. Live counting was used for BrdU+ and phosphohistone H3 (PHi3+) cells in the SVZ and CC, and Olig2+ cells in the SVZ. Sample images were acquired of other cell markers and then counted using ImageJ. Macros were developed that displayed only the DAPI channel, so that the appropriate region could be selected and cropped without the bias of seeing other channels. The macros then switched the view to other channels for cell counting. For MBP threshold analysis, regions were initially selected using only the DAPI channel. Histograms for fluorescence intensity levels between 0 and 255 were then obtained. Microsoft Excel was used to plot the intensity distribution and select an appropriate threshold above which MBP labeling was considered positive. The percentage of pixels with intensity greater than the threshold value was then calculated.
Real-time polymerase chain reaction
SVZ tissue from individual 6-week Gal-3
+/+
and Gal-3
−/−
mice was microdissected in Hanks’ Buffered Saline Solution (HBSS). The sample was centrifuged to pellet tissue and aspirate excess HBSS. It was snap frozen in liquid nitrogen and stored at −80 °C. RNA extraction was performed using an RNeasy Mini Kit. Before extraction, individual samples were pooled to provide three groups of four animals for each genotype. All samples provided a 260/280 ratio between 2.08 and 2.10, which was considered to provide satisfactory RNA quality. Reverse transcription was performed using a High Capacity RNA to cDNA Kit (Applied Biosystems). Quantitative PCR (qPCR) used TaqMan Gene Expression Assays (Life Technologies) and TaqMan Gene Expression Master Mix (Life Technologies). It was performed on a 7900HT Fast Real-Time PCR System (Applied Biosystems) using technical triplicates and standard curves and following the TaqMan Gene Expression Master Mix Protocol for 384-well plates. Data was initially analyzed using SDS Software v2.4, and aberrant replicates were removed. EGFr and B2m had efficiency between 90 and 110 % and R
2 ≥ 0.998 while CCL2 had R
2 = 0.965, Ccr2 R
2 = 0.948, and Mmp9 R
2 = 0.987 (Egfr Mm00433023 m1; Ccl2 Mm00441242 m1; Ccr2 Mm00438270 m1; Mmp9 Mm00442991 m1; B2m Mm00437762 m1). Galectin-1 (Lgals1) Mm00839408 g1; Galectin-3 (Lgals3) Mm00802901 m1; Galectin-8 (Lgals8) Mm01332239 m1; Galectin-9 (Lgals9) Mm00495295 m1. Data were exported to Microsoft Excel, where individual gene quantities were normalized to B2m and divided by Gal-3
+/+
quantities to create fold-change ratios. Statistics were performed on normalized values using a Student’s t test as described later.
Neurosphere proliferation assay
Primary SVZ cultures were obtained from P4-5
Gal-3
+/+
and
Gal-3
−/−
mice. Animals were sacrificed using anesthesia (hypothermia) followed by decapitation. Whole brains were dissected then cut into 500-μm coronal sections using a McIlwain tissue chopper. SVZ was microdissected in HBSS and stored on ice until dissection was complete. Cells were dissociated mechanically with fine iris scissors and enzymatically through Accutase incubation (Sigma, A6964) for 10 min at 37 °C. Cells were then washed three times in Neurobasal A+ (1×, B27 1×, glutamine 2 mM and penicillin/streptomycin 50k U/L) including centrifugation for 5 min at 1200 rpm for each wash, suspended in NB-A+ with growth factors (EGF and FGF-2, 20 ng/ml, R&D Systems), counted, and seeded in 96-well plates at 5 × 10
4/mL of NB-GF. Cultures were incubated at 37 °C with 5 % CO
2 for 7 days before whole wells were imaged using an AMF4300 EVOS® FL Imaging System (Life Technologies). Neurospheres were aspirated and then dissociated enzymatically, washed, suspended, counted, and seeded at 5 × 10
3/mL of NB-GF. Cultures were then incubated for a further 7 days before imaging of secondary neurospheres. Images were analyzed in ImageJ using the Stack Sorter plugin to merge images (
http://www.optinav.com/Stack-Sorter.htm).
Statistics
Data from the appropriate quantification was collated into Microsoft Excel. It was then analyzed using scripts written for SPSS 21. For experiments comparing one factor between two groups, a Student’s t test was performed. For experiments comparing one factor between three or more groups, a one-way ANOVA was performed. For experiments comparing two factors, a two-way ANOVA was performed. For experiments comparing two factors for a variable quantified in multiple SVZ or CC anatomical regions, a repeated-measures ANOVA was performed. Bonferroni post hoc tests were used to determine significant pairwise comparisons in ANOVAs. Graphs were produced in Prism 5 (GraphPad) and show mean ± standard error of the mean (SEM).
Discussion
This study showed that the number of Phi3+, Ki67+, CD45+, and Gal-3+ cells decreased in the SVZ after cuprizone treatment which is in contrast to TMEV [
16] and most other disease models. The decrease was remarkable since immediately above the SVZ in the CC Phi3+, Ki67+, CD45+, and Gal-3+, cell numbers increased. Because inflammation was decreased in the SVZ, we hypothesized that proliferation would increase in the niche [
18]; however, it decreased after cuprizone treatment, suggesting the relationship between inflammation and proliferation in the niche is complex. We also showed that SVZ proliferation was not affected in
Gal-3 knockouts in vivo or in the neurosphere assay. This is in line with our previous work showing that Gal-3 loss does not directly affect SVZ proliferation [
16,
22]. However, we labeled SVZ cells with multiple doses of BrdU and found that 3 weeks after cuprizone administration,
Gal-3
−/−
mice had significantly fewer cells in the SVZ compared to WT controls, suggesting Gal-3 limits emigration of SVZ cells.
The decrease in Gal-3+ SVZ cells was probably not due to the decrease in CD45+ cells since the large majority of SVZ Gal-3 expression is not in CD45+ microglial cells but in SVZ astrocytes and ependymal cells [
22]. Interestingly, we showed that cuprizone induced activation of GFAP expression and morphological changes in the SVZ consistent with reactive astrocytosis. Gal-3 expression can increase in striatal astrocytes in response to TMEV [
16]. In these experiments, despite the SVZ astrocyte reactivity, Gal-3 expression actually decreased, suggesting Gal-3 is not required for SVZ astrocytic reactivity. Similarly, Gal-3 knockout mice exhibited robust increases in GFAP expression both in the SVZ and the CC. A recent study showed that stab wound injury-induced reactive astrocytosis in the cerebral cortex depends on Gal-3 expression [
41]. Thus, it is likely that Gal-3 has different effects on astrocyte reactivity depending on the brain region or type of pathology.
This work adds to the growing list of studies revealing the SVZ as a unique inflammatory niche. We were surprised a number of years ago to find that SVZ microglia are semi-activated in healthy animals [
15]. We then made the remarkable observation that traumatic brain injury (TBI) [
15] and stroke [
32,
42] induce massive inflammation in forebrain parenchyma, but not in the adjacent SVZ. This was in contrast to TMEV infection which was associated with consistent inflammation in the SVZ [
17]. In this manuscript, we provide data suggesting that SVZ microglial numbers and inflammation are actually decreased after cuprizone treatment. We have found that Gal-3 is expressed by the same cells in the healthy SVZ as in disease models. It was expressed by astrocytes and ependymal cells homeostatically, after TBI, TMEV, stroke, and cuprizone [
16,
22,
32] (and Szele lab unpublished studies). It is particularly surprising that few microglial cells express Gal-3 in the SVZ before or after injury since Gal-3 has been used as a marker of activated macrophages. Taken together, this study and previous work shows that inflammation in the SVZ is tightly regulated and is sensitive to pathological context.
The decrease in markers of inflammation in the SVZ was in sharp contrast with the increased number of Gal-3+ and CD45+ cells in the adjacent corpus callosum. Similar sharp contrasts between indices of inflammation in the SVZ and the adjacent CC, striatum, or cerebral cortex have been noted in TBI and TMEV [
15‐
17]. CD45 labels all cells in the hematopoietic lineage so is insufficient to definitively phenotype them. Here, we found that the number of CD45+ SVZ cells that expressed Iba1 increased slightly after cuprizone. Since Iba1 is associated with microglia, these data suggest that though cuprizone decreased the number of CD45+ SVZ cells, the relative percent of microglia increased. Our previous work found a small number of dendritic cells [
17] and T cells [
16] in the SVZ; thus, it is possible that some of the CD45+/Iba1-negative cells we found in this study in the SVZ belonged to one of these cell types. Alternatively, a minority SVZ CD45+ microglial cells expressed Iba1 below the level of immunohistochemical detection. It is likely the unique extracellular molecular environment in the SVZ selectively autoregulates inflammation in the niche [
16,
43,
44]. We recently demonstrated the SVZ expresses higher levels of chemokine ligands than adjacent regions [
16]. A better understanding of how inflammation is regulated in the SVZ may help develop approaches to augment neurogenesis since immune cells cross talk with neural cells and can be either beneficial or detrimental to their homeostatic and reparative functions [
45].
Many types of brain injury increase SVZ proliferation [
38,
46,
47]; however, activation of neurogenesis depends on the model, species, and level of inflammation. Inflammation was originally shown to dampen adult neurogenesis [
18,
19], but recent work has defined multiple modes of microglial activation, some of which increase neurogenesis [
20,
21,
45]. Therefore, we were uncertain whether the SVZ phenotype induced by cuprizone described above would be associated with altered proliferation. The demyelination groups had significantly diminished numbers of proliferative SVZ cells and rebounded to control levels in the remyelination group, suggesting a rapid return to homeostasis in the SVZ upon cessation of cuprizone. In contrast to the SVZ, CC proliferation increased at the 6-week and remyelination time points. This paralleled the discrepancy between the SVZ and CC described above and further suggests unique regulation of inflammation in the SVZ.
The proliferative response to cuprizone in the SVZ and the CC was not affected by loss of Gal-3. This was supported by our in vitro data: the number and size of SVZ neurospheres were unaffected by loss of Gal-3. Neurospheres were prepared from P4-5 pups, and it could be that loss of Gal-3 would affect neurosphere growth when prepared from adults. However, loss of Gal-3 did not alter constitutive proliferation in the adult SVZ in three separate murine strains [
16,
22]. Similar results were obtained in the MCAO model of stroke—there were no differences between WT and Gal-3 nulls in SVZ or parenchymal proliferation [
32]. This is in contrast to a recent study in which loss of Gal-3 inhibited astrocyte proliferation after stab wound injury [
41]. Gal-3 binds to a large variety of proteins, has multiple functional effects depending on context, and is frequently expressed de novo after brain injury, together positioning it to have selective effects in different contexts.
One of the key features of SVZ progenitors is their capacity for long-distance migration to the OB and emigration towards injured regions of the brain. There is good evidence that SVZ progenitors are beneficial by a combination of neural replacement and neuroprotection [
48,
49]. To track potential SVZ progenitor emigration, we labeled cells with BrdU at the beginning of cuprizone administration. The large majority of cells that are proliferative and get labeled by BrdU in the forebrain before damage are SVZ cells. Therefore, BrdU+ cells outside of the SVZ at later time points could be inferred to have migrated from the SVZ. Interestingly, we showed decreased BrdU+ cell numbers in the SVZ but increased numbers in the CC after cuprizone. At 3 weeks, the number of BrdU+ cells that were gone from the SVZ was significantly greater in
Gal-3
−/−
than in WT mice, suggesting that Gal-3 normally inhibits emigration. This is supported by our previous finding that lack of Gal-3 causes SVZ progenitors to switch migration from their rostral direction to a local exploratory motility as a prelude to emigration [
22]. It is also consistent with our TMEV demyelination model wherein loss of Gal-3 increased SVZ progenitor emigration [
16]. These data are consistent with cuprizone inducing BrdU+ cells to migrate from the SVZ to the CC.
Because of the decreased rostral migration previously observed in
Gal-3
−/−
mice [
22], the decreased number of BrdU+ cells in the SVZ of
Gal-3
−/−
mice in this study was probably not caused by greater rostral migration towards the OB. We demonstrated decreased MBP immunofluorescence in the OB of WT mice after cuprizone, suggesting demyelination. However, we do not know if OB demyelination was increased in
Gal-3
−/−
mice relative to WTs, which could have increased rostral migration and explained the decreased numbers of BrdU+ cells in the SVZ at 3 weeks post-cuprizone. Using immunohistochemistry for Gal-3 and CD45, we previously noticed that cuprizone and TMEV caused inflammation in multiple forebrain regions [
17], such as the OB. We have other data suggesting OB demyelination after TMEV based on reduced MBP immunofluorescence (Szele lab, unpublished observations). The extent of OB and olfactory tract demyelination in MS and neuromyelitis optica patients is correlated with disease severity [
50]. We propose the cuprizone and TMEV-IDD animal models will help elucidate the pathophysiological mechanisms and functional consequences of OB inflammation and demyelination in MS.
The majority of cells that emigrated from the SVZ were likely in the oligodendrocyte lineage as many BrdU+ labeled cells in the CC were Olig2+. There is ample evidence that human MS and models of demyelination induce SVZ-derived oligodendrocytes to emigrate [
10‐
13,
51]. In this study, the number of Olig2+ cells in the CC decreased at 3 weeks post-cuprizone and then increased suggesting cuprizone initially killed oligodendrocytes which were then replaced by SVZ-derived oligodendrocyte precursors. Recent work has suggested Gal-3 alters the rate of oligodendrocyte maturation and extent of demyelination [
26]. We did not find any evidence for differences in Olig2+ cells in the SVZ or CC between WT and
Gal-3
−/−
mice. We also examined the expression of the mature oligodendrocyte marker CC1 [
35]. We did not find that the percent of Olig2+ cells that expressed CC1 differed between Wt and Gal-3
−/−
mice. Both genotypes also exhibited an increased percent of Olig2+ that were CC1+ in the remyelination group. These results suggest Gal-3 loss does not affect oligodendrocyte differentiation. Finally, although SVZ neuroblast migration towards demyelination has been documented [
52], we found very few BrdU+ cells in the CC that expressed the immature neuronal marker Dcx.
A study by Hoyos and colleagues showed that loss of Gal-3 inhibited remyelination in the CC after cuprizone as measured by MBP, suggesting that Gal-3 normally has a positive effect on remyelination [
27]. Similar to that study, we showed loss of MBP immunofluorescence in the CC in the 6-week post-cuprizone group suggesting demyelination. However, we did not find that loss of Gal-3 affected MBP loss or return to normal values in the CC. We are not certain what accounts for this discrepancy; however, it is important to note that Hoyos et al. performed experiments in a different Gal-3 knockout mouse, which lacks exon V [
27], whereas the mouse used here lacks exons II, III, and IV [
53].
Abbreviations
ANOVA, analysis of variance; BrdU, 5-bromo-2′-deoxyuridine; CC, corpus callosum; CD45, complement of differentiation 45; Dcx, doublecortin; Gal-3, galectin-3; GFAP, glial fibrillary acidic protein; Iba1, ionized calcium-binding adapter molecule 1; LV, lateral ventricle; MBP, myelin basic protein; MS, multiple sclerosis; OB, olfactory bulb; Olig2, oligodendrocyte lineage transcription factor 2; PHi3, phosphohistone 3; qPCR, quantitative polymerase chain reaction; RMS, rostral migratory stream; SEM, standard error of the mean; SVZ, subventricular zone; TAP, transit amplifying progenitor