1. Introduction
Multiple sclerosis (MS) is the most common chronic inflammatory and neurodegenerative disease of the central nervous system. The pathologic hallmark of MS is the formation of focal areas of myelin loss, termed lesions. Besides the commonly described white matter lesions, extensive grey matter lesions are found in MS cerebral cortex and deep grey matter [
1,
2,
3]. Grey matter lesions are reported to occur in up to 90% of patients with chronic MS [
4]. Early-stage MS grey matter lesions were shown to contain myelin-laden macrophages as well as perivascular CD3+ and CD8+ T-cell infiltration [
5,
6]. In contrast, chronic grey matter lesions characteristically lack significant infiltration of immune cells but show diffuse microglial activation [
7,
8]. In both the early and in the chronic stage, grey matter lesions were shown to be associated with meningeal inflammation [
5,
9]. Cortical demyelination also correlates with loss of neurons, axons, synapses, and glial cells [
8,
10,
11,
12,
13]. Grey matter lesions further seemed to have a more efficient myelin repair and less gliosis compared to white matter lesions.
Cortical atrophy and demyelination in MS are associated with lymphoid aggregates in the meninges [
8]. Further, higher levels of interferon gamma (IFNγ) and tumor necrosis factor alpha (TNFα) in the meninges and in the cerebrospinal fluid were identified [
7,
14]. Although grey matter pathology in MS has received rapidly growing interest over the past two decades, many aspects of the chronic pathological processes and the adaptations to a chronically demyelinated environment remain elusive.
Long-standing cortical demyelination may contribute to neuronal damage and degeneration. It is estimated that 40–70% of MS patients suffer from deficits of memory and attention, reduction of cognitive flexibility, and impairment of executive functions [
15,
16]. These symptoms are therapeutically very difficult to address and are mechanistically poorly understood, but it is strongly suspected that these cognitive symptoms are associated with changes in cortical grey matter such as demyelinated lesions and atrophy [
14,
17]. Besides the well-described demyelinated grey matter lesions, diffuse grey matter abnormalities in non-lesional normally myelinated areas, such as diffuse microglial activation, diffuse axonal injury, and astrogliosis, have also been observed [
18,
19,
20,
21].
To study the pathomechanisms and functional consequences of chronic grey matter demyelination and meningeal inflammation in detail, an experimental model replicating the key phenotypes is required. Such a model was established by Merkler et al. in 2006 [
22] by immunizing Lewis rats with recombinant myelin oligodendrocyte glycoprotein (MOG) and incomplete Freund’s adjuvant (IFA) [
6,
22]. This animal model was characterized by highly reproducible focal cortical demyelination and inflammation, followed by rapid resolution and remyelination. This model, however, did not give rise to a chronic cortical pathology.
Based on our results, we report the establishment of an animal model for studying the consequence of chronic cortical grey matter demyelination and meningeal inflammation.
3. Discussion
Our results demonstrate that continuing overexpression of TNFα and IFNγ generated by lentiviral intracortical injection of MOG-immunized rats leads to chronic demyelination and meningeal inflammation.
In a previously described animal model, MOG-immunized rats were injected with recombinant proteins TNFα and IFNγ, giving rise to subpial and intracortical demyelinated lesions with infiltrating immune cells, complement deposition, acute axonal damage, and neuronal cell death, followed by a rapid resolution of the lesions [
22]. A less traumatic version of this model was achieved by injecting the recombinant proteins TNFα and IFNγ into the subarachnoidal space. This approach gave rise to a similar pathology to the intracortical injections [
7]. These models, using recombinant proteins, demonstrated that immunization against MOG and the injection of both TNFα and IFNγ are required for demyelination. In our model, we demonstrate that immunization against MOG is necessary for demyelination. Even a 10-week overexpression of TNFα and IFNγ alone did not lead to demyelination. Demyelination thus cannot be a direct effect of cytokine expression and meningeal inflammation but, in our model, also requires a priming of the immune system against a component of myelin.
The lesions were largest in animals, which were immunized with MOG and chronically overexpressing TNFα and IFNγ. However, the control virus leading to chronic β-galactosidase expression also gave rise to subpial demyelination comparable to the injection of the recombinant proteins TNFα and IFNγ. This could be a consequence of the direct traumatic tissue damage caused by the intraparenchymal injection, leading to an opening of the blood–brain barrier. Alternatively, it could be a consequence of lentiviral particles being injected into the parenchyma or β-galactosidase provoking an immune response. Merkler et al. found only limited demyelination when injecting phosphate-buffered saline into MOG-immunized animals instead of the recombinant proteins TNFα and IFNγ [
22], suggesting that both proposed mechanisms might play a role in the animal model shown here.
It has further been demonstrated that one cytokine alone is not sufficient to generate widespread acute cortical demyelination [
7,
22]. We have demonstrated that the simultaneous injection of both LV-
tnfa and LV-
infg leads to demyelination in MOG-immunized animals, but we did not test the lentiviruses alone. It is not evident that both cytokines together are necessary and one alone may be sufficient to initiate and uphold the demyelination and inflammation.
The previous acute models recovered within two to four weeks. In comparison, our model causes stable subpial demyelination for at least 10 weeks and possibly much longer. In another approach by [
24], a catheter was installed at the border of the white and cortical grey matter of MOG-immunized animals, allowing multiple injections of recombinant protein TNFα and IFNγ that generate demyelinating lesions for at least four weeks. This approach, however, causes direct trauma to the brain parenchyma, potentially greater than the injection of the lentivirus. It depends, in addition, on the permanent installation of foreign material, penetrating the skin, skull, and blood–brain barrier.
Further, we have shown that the meningeal inflammation depends only on the chronic overexpression of TNFα and IFNγ, as described before in the acute models [
7]. The parenchymal infiltration detected was primarily limited to the injection site and occurred irrespective of the type of injection. It decreased in all groups over time.
After an initial phase with partial remyelination between five days and four weeks post injection, the extent of demyelinated lesions, the meningeal inflammation, and the number of oligodendrocyte lineage cells were stable until at least ten weeks post injection. The previous model with recombinant cytokine injection led to a rapid remyelination, even after repeated acute demyelination [
25]. In our model, complete remyelination might fail directly due to the chronic release of TNFα and IFNγ or indirectly by effects inflicted by the meningeal inflammation. The density of oligodendrocyte lineage cells did not decrease in our model. We speculate that in the initial phase a certain loss of oligodendrocytes and a reactive proliferation of oligodendrocyte precursor cells occur. To support this, both TNFα and IFNγ have previously been shown to lead to oligodendrocyte death and demyelination [
26,
27]. However, we might have overlooked this due to the temporal resolution of our analysis. We could not demonstrate any significant changes in the overall number of oligodendrocytes or oligodendrocyte precursor cells.
In a parallel study, a very similar model has been described [
28]. The authors immunized Dark Agouti rats against MOG and injected lentiviruses to overexpress TNFα and IFNγ in the subarachnoidal space, efficiently transducing meningocytes, in contrast to the intracerebral injections in the animal model described here. Their approach gave rise to bihemispheric, symmetric meningeal inflammation and subpial demyelination, closely resembling aspects of human cortical MS pathology. While the changes described here are very similar, we thus see predominantly unilateral pathology. This is an advantage, enabling us to use the weakly affected contralateral hemisphere as an internal control. We see our approach as an alternate version of the model suggested by James et al. [
28].
In MS research, lesions spanning the whole cortical thickness from the subpial area to the white matter, termed type III cortical lesions, have been shown to be the most common type of cortical MS lesions and have been demonstrated to be unique to MS pathology [
29,
30]. Interestingly, the acute models described previously and the model established here lead to a subpial demyelination pattern as seen in MS. While Merkler et al. discussed the drainage of the recombinant proteins along predetermined anatomical routes from the white matter to the meninges as a possible cause of this pattern [
22], Gardner et al. suggested that cytokines produced from the meningeal infiltrates may diffuse into the brain parenchyma and cause a subpial pattern of demyelination [
7]. In our model, both mechanisms are possible: a direct effect by the lentivirus, leading to chronic overexpression of TNFα and INFγ, or an indirect effect caused by cytokine production from the meningeal infiltrates. In MS, an association between subpial demyelination and meningeal inflammation has been discussed [
8,
9,
18], but the nature of any causal connection between the two phenomena remains elusive. Analysis of our data revealed a significant correlation between the demyelinated fraction of the cortex and the meningeal inflammation, whereas the parenchymal infiltration did not show such a correlation. However, we cannot conclude that meningeal inflammation is necessary for the development of subpial demyelination from our model.
4. Materials and Methods
4.1. Subcloning of Lentiviral Plasmids
The lentiviral plasmid pULTRA was used as s backbone and was a gift from Malcolm Moore (Addgene plasmid #24129, Addgene, Watertown, MA, USA) [
31]. The sequence for
tnfa (reference sequence: NM_012675.3, nucleotides 5′-145-852-3′) was acquired from rat cDNA. The sequence for
ifng (reference sequence: NM_138880.2, nucleotides 5′-10-480-3′) was amplified by polymerase chain reaction (PCR) from a cDNA ORF clone plasmid (RG80234-UT, Sino Biological, Wayne, PA, USA). Primers for the PCR of
tnfa and
ifng were designed with specific restriction enzyme sequences at the 5′ and 3′ sites (
Figure 1A). The PCR products were separated by gel electrophoresis and the specific band purified by a gel DNA recovery kit according to the manufacturer’s protocol (D4007, Zymo Irvine, CA, USA). The sequence for
lacZ was obtained by restriction digest with BamHI (R3136S, New England Biolabs, Ipswich, MA, USA) of the plasmid LV-
lacZ received as a gift from Inder Verma (Addgene plasmid #12108) [
32]. Restriction digests were performed in CutSmart buffer (B7204S, New England Biolabs, Ipswich, MA, USA), using 6 units of restriction enzyme per microgram of cDNA at 37 °C for 3 h. cDNAs were subcloned into the pULTRA backbone with the Rapid DNA Dephos & Ligation Kit (4898117001, Hoffmann-La Roche, Basel, CH, Switzerland) according to the manufacturer’s protocol. Amplification of the plasmids was carried out in Stbl3 bacteria (C737303, Thermo Fisher Scientific, Waltham, MA, USA). The plasmids were thereafter purified with the miniprep classic kit (D4015, Zymo Irvine, CA, USA) and sent to Microsynth (Balgach, Switzerland) for sequencing.
4.2. HEK293T Cells
HEK293T cells were used for lentivirus production and lentiviral quality control.
4.3. Lentivirus Production
Lentiviruses containing the sequence for enhanced green fluorescent protein (eGFP) with either
lacZ (LV-
lacz),
tnfa (LV-
tnfa), or
ifng (LV-
ifng) were generated in HEK293T cells. The lentiviral helper plasmids pMD2.VSV-G, (Addgene plasmid #12259), pMDLg/pRRE (Addgene plasmid #12251), and pRSV-Rev (Addgene plasmid #12253) were a gift from Didier Trono (Addgene, Watertown, MA, USA) [
33]. For a T75 flask with 70% confluent HEK293T cells, a 1.5 mL tube of Opti-MEM medium (31985062, Thermo Fisher Scientific) was mixed with 5 μg of each helper plasmid and 6.5 μg of the lentiviral plasmid. Further, one 15 mL tube with 0.75 mL Opti-MEM medium was mixed with 60 μL Lipofectamine 2000 Transfection Reagent (No. 11668027, Thermo Fisher Scientific). The diluted plasmids were mixed with the Lipofectamine solution, vortexed, and incubated for 5 min at room temperature. The DNA–lipid complex was then added to the cells and incubated for 18 h, removed, and replaced with 7 mL fresh HEK293T culture medium. After 48 h, the cell culture medium was harvested, transferred to a 15 mL tube, and centrifuged for 15 min at 3000×
g to pellet cells and debris. The supernatant was then filtered through a 0.45 μm filter on a syringe. To concentrate the viral particles, 2.3 mL Lenti-X-concentrator was added before processing according to the manufacturer’s protocol (631232, Takara, Kyoto, Japan). The lentiviral particles were resuspended in PBS and aliquoted and stored at −80 °C.
4.4. Lentivirus Quantification
Concentration of lentiviral particles was determined by estimating the concentration of lentivirus-associated p24 protein with the QuickTiter Lentivirus Titer Kit according to the manufacturer’s protocol (VPK-107, Cell Biolabs, Inc., San Diego, CA, USA). For calculating the approximate number of transfected units, the conservative assumption of 1000 P24 proteins per lentiviral particle was applied (range suggested by the manufacturer: 100–1000).
4.5. Lentivirus Quality Control
To confirm the production of functional lentiviruses, production of TNFα and IFNγ was verified by enzyme-linked immunosorbent assay (ELISA). HEK293T cells were seeded into 6-well plates at a density of 200,000 cells per well and cultivated for 18 h before transduction with estimated viral particles per cell of either LV-tnfa or LV-ifng. The cells were transferred to fresh flasks and the medium was exchanged three times per week. After one week, the supernatant and the cells were assumed free of active lentiviral particles. Cell culture medium was then replaced with 1 mL fresh medium without serum and 24 h later ELISA was performed. For TNFα and IFNγ ELISA, an unlabeled TNFα capture antibody (dilution 1:167, 506102, BioLegend, San Diego, CA, USA) or an unlabeled IFNγ capture antibody (diluted 1:500, 507802, BioLegend), respectively, was diluted in coating buffer (421701, BioLegend) and 100 μL per well was distributed in a 96-well plate (423501, Biolegend). After 18 h of incubation at 4 °C, wells were washed with 0.05% Tween 20 in PBS three times and 200 μL blocking solution (421203, BioLegend) was added per well for one hour. Wells were then washed and incubated for three hours with 100 μL pure or 1:100 diluted supernatants from the HEK293 cell culture transfected with LV-tnfa or LV-ifng. Wells were washed and the biotin-labeled detection antibody against TNFα (diluted 1:1000, 516,003, BioLegend, San Diego, CA, USA) or IFNγ (diluted 1:500, 518,803, BioLegend, San Diego, CA, USA), respectively, was added for one hour. Wells were washed and 100 μL per well of avidin-horseradish peroxidase conjugate was added (diluted 1:1000 in blocking buffer, 405103, BioLegend). Thereafter, wells were washed, and 200 μL substrate solution (421101, BioLegend) was added per well. Upon sufficient color development, 100 μL stop solution (423001, BioLegend) was added. Absorption was read at 450 nm.
4.6. Animals
Adult female inbred Lewis rats (185–250 g) were purchased from Janvier Labs (Saint-Berthevin, France) and used as described before [
34]. Animal housing and all animal experiments were performed at the experimental animal facility of the University Medical Center Göttingen, Germany. The rats were housed in cages with up to five animals each under constant temperature and humidity, a 12/12 h light/dark cycle, and with free access to food and water. All animal experiments were performed in accordance with the European Communities Council Directive of 22 September 2010 (2010/63/EU) and were approved by the Government of Lower Saxony, Germany.
4.7. Immunization and Intracerebral Stereotactic Injection
The immunization and intracerebral stereotactic injection were performed as described before [
22]. Eighteen days after immunization, intracerebral stereotactic injection was performed as described [
22]. For this, rats were anesthetized by intraperitoneal injection of ketamine/xylazine and fixed in a stereotactic frame. A one-centimeter incision of the skin on the midline of the skull was made to expose the bregma and a hole was drilled 2 mm lateral and 1 mm caudal of the bregma. Using a finely calibrated glass capillary, 2.5 μL solution was carefully injected into the cortex, containing either LV-
tnfa and LV-
ifng mixed 1:1, LV-
lacZ, or the recombinant cytokines TNFα (250 ng; P16599, R&D Systems, Abingdon, UK) and IFNγ (150 U; 400–20, PeproTech, London, UK). Monastral blue was added to mark the injection site. After the injection, the glass capillary was retracted and the skin was closed with a suture [
34]. Rats were perfused after 5 days, 4 weeks, 7 weeks, or 10 weeks to investigate lesion evolution. For this, the animals were anesthetized with isoflurane, receiving a lethal dose of ketamine/xylazine and perfused transcardially with PBS followed by 4% paraformaldehyde. Subsequently, the heads were postfixed in 4% paraformaldehyde at 4 °C for two days before removing the brain and embedding it in paraffin.
4.8. Composition of Study Groups
In accordance with the 3R principle to reduce, refine, and replace animal experimentation, not all groups were analyzed at all time points, but the experiments were performed with four different models at four time points to address specific questions (
Figure 1B). Animals immunized with MOG and transduced with LV-
tnfa and LV-
ifng represent the experimental model and were studied at 5, 28, 49, and 70 days to assess the development over time. This was compared with the previously published cytokine-induced experimental grey matter lesion model using MOG-immunized and recombinant TNFα- and IFNγ-injected animals at 5 days (peak of inflammation and demyelination in cytokine model) and 28 days (restitutio ad integrum in cytokine model) [
21]. Also, we compared this experimental model to MOG-immunized animals transduced with LV-
lacZ as a negative control at the time points 5 days and 28 days to control for early and late effects of the viral transduction. In addition, we compared the experimental group with non-immunized animals injected with LV-
tnfa and LV-
ifng to control for the MOG immunization at the time points 5, 28, and 70 days.
4.9. Enzyme-Linked Immunosorbent Assay for MOG Titers
For the measurement of the serum anti-MOG antibody titers, an ELISA was performed as described before [
22]. Values two-fold above the background levels were considered positive.
4.10. Immunohistochemistry
Immunohistology was performed on 3 μm thin, formalin-fixed, paraffin-embedded (FFPE) coronary sections using antibodies for macrophages/activated microglia (CD68, clone ED1, Bio-Rad AbD Serotec, Oxford, UK), CD3+ T-cells (clone CD3-12Bio-Rad AbD Serotec, Oxford, UK), CD45R+ B-cells (clone HIS24, BD Biosciences, Franklin Lakes, NJ, USA), myelin basic protein (MBP, Abcam, Cambridge, UK), anti-neuronal nuclei (NeuN, Abcam, Cambridge, UK), oligodendrocyte lineage factor 2 (OLIG2, clone 211F1.1, Merck Millipore, Darmstadt, Germany), mature myelin-maintaining oligodendrocytes (P25/TPPP, clone EPR3316, Abcam, Cambridge, UK), pre-resp. actively myelinating oligodendrocytes (BCAS1, Santa Cruz, Heidelberg, Germany), FITC (HRP conjugated, Dako Deutschland GmbH, Hamburg, Germany), and green fluorescent protein (GFP, clone 6AT316, Abcam, Cambridge, UK).
4.11. Image Analysis
All stained tissue sections were scanned with an Olympus VS120-L100-W scanner fitted with a VC50 camera (Olympus, Shinjuku, Tokyo, Japan) with a 200× magnification at a resolution of 0.345 micrometers per pixel. For CD3 and CD45R staining, the cells were counted manually using NIS-Elements software (Version 5.11.00, Nikon, Tokyo, Japan). For demyelination, areas were delineated manually using CellSense software (Version 1.6, Olympus, Tokyo, Japan). For OLIG2, BCAS1, and P25 analysis, areas of 1 mm by 0.75 mm (0.75 mm
2) were manually delineated in Adobe Photoshop (Version 22.1.0, Adobe Inc., San José, CA, USA) subpially above the injection site, where the pathology was most pronounced. OLIG2 and P25 were then automatically processed with Fiji (image processing package including ImageJ, Version 1.53a, Wayne Rasband, National Institutes of Health, Bethesda, MD, USA [
35]), using the count objects function with the parameters: hue 106–189 (OLIG2) or 45–220 (P25), saturation 37–255 (OLIG2) or 80–255 (P25), brightness 0–255 (OLIG2) or 0–180 (P25), and size 11.9–238 µm
2. Counted objects were marked and saved and all images were visually controlled for accurate counting. BCAS1-positive cells were counted manually using Adobe Photoshop (Version 24.7.0, Adobe, San José, CA, USA).
4.12. Statistical Analysis
Statistical analysis was performed with R. To compare means of two groups, two-sided Welch t-tests were performed and the degrees of freedom (df) and the p-values are reported. For comparisons of multiple time points, an analysis of variance (ANOVA) with subsequent Tukey post hoc correction was performed and adjusted p-values, the between-group and in-group degrees of freedom, and the F-value are reported as follows: F (between-group, in-group) = F-value. If the ANOVA test did not reach significance, the Tukey post hoc results are not reported. For correlations, Pearson correlations were calculated, and the correlation coefficient and the p-value are reported. For the analysis of the oligodendrocyte lineage cells (OLIG2, P25, BCAS1), the data were normalized by dividing by the number of cells in the same area in the contralateral hemisphere. Data were then log-transformed with base 10. Paired two-sided Welch t-tests were used to test for statistically significant changes between the ipsi- and contralateral cortex. For all tests, a p-value or adjusted p-value below 0.05 was considered significant.