Introduction
Cells that reside within the bone marrow (BM) have long been known to fuse with several distinct types of cells throughout the body, including brain neurons [
2,
41]. While the biological function of BM-derived cell fusion with neurons is not clear, related experiments involving fusion of ‘healthy’ donor BM-derived cells with host hepatocytes, to ameliorate metabolic liver disease, have provided strong clues for a role in targeted tissue regeneration [
40,
43]. Within the brain, fusion of BM-derived cells occurs predominantly, although not exclusively [
10], with Purkinje cells of the cerebellum [
23]. Morphological and mRNA analyses of these fused Purkinje cells reveal the formation of either mononucleate cells or binucleate heterokaryons, with subsequent nuclear reprogramming and expression of donated nuclear genes [
20,
29].
The incidence of BM-derived cells fusing with cerebellar Purkinje cells appears to be very low under normal physiological conditions [
22,
24,
29]. Nevertheless, its biological relevance is suggested by observations that such fusion events in either rodents or humans are substantially increased in number with age [
41,
42]; or after exposure to cytotoxic agents (for example radiation or chemotherapeutics) [
26,
42]; or within an inflammatory microenvironment such as that present in multiple sclerosis [
22] and in animal models of cerebellar disease [
9,
10,
13,
20,
21,
29]. These observations have been taken to suggest that, as Purkinje cells are generated only during early cerebellar development [
28], heterotypic cell fusion (fusion between different cell types) acts as a physiological cell rescue mechanism to counter neuronal injury and maintain Purkinje cell function throughout adulthood. Purkinje cells are some of the largest, most complex and elaborate neurons in the human brain; their axons represent the sole output for conveying nerve impulses from the cerebellar cortex, and as such, they are an essential part of the motor system [
8]. Their loss is characterised clinically by insidious accumulation of disability [
27]. If heterotypic cell fusion attenuates neuronal cell damage and prevents Purkinje cell dysfunction, it may have valuable therapeutic implications for neurodegenerative disease in general, and in particular, for patients with cerebellar injury.
The notion that fusion of BM-derived cells with adult Purkinje cells preserves or restores function in the face of injury currently remains an attractive, even likely, hypothesis. However, there is no direct experimental evidence of cellular repair; functional restoration in Purkinje cells has never been directly tested by assessing the cell morphology or electrical activity of the fused cells, or by comparing the properties of these cells with those of damaged Purkinje cells that have not undergone fusion with BM-derived cells or with Purkinje cells in healthy control animals. At a more fundamental level, it is not even known if an adult brain neuron that has fused with a different type of cell remains capable of firing action potentials; studying and elucidating these aspects of cell fusion in the adult brain is challenging. In an attempt to address these outstanding and crucial questions, we utilize an in vivo central nervous system (CNS) inflammatory disease model in chimeric mice expressing enhanced green fluorescent protein (EGFP)-BM cells. We perform both detailed histological and electrophysiological analysis of single Purkinje cells ex vivo in adult cerebellar slices, to assess the physiological role that cell fusion plays in neuronal protection in the adult brain.
Materials and methods
Animals
All animal experiments were performed in accordance with the UK Animals (Scientific Procedures) Act 1986 and approved by the University of Bristol Animal Welfare and Ethical Review Body. All animal numbers are based on both previous experience and/or power calculations using preliminary data. Blinding could not be performed when comparing Purkinje cells that had or had not fused with BM-derived cells due to reliance on EGFP to identify fused cells.
Transgenic mice ubiquitously expressing EGFP (strain # C57BL/6-Tg(CAG-EGFP)131Osb/LeySopJ, stock # 006567) were purchased from Jackson Laboratory, USA. Wild-type C57BL/6 VAF/Elite mice were provided by Charles River, UK. All mice were housed in a specific pathogen-free facility, with free access to food and water. If subject to a bone marrow transplant (BMT), mice were maintained in filter top cages (pore size 100 μm) during the recovery period.
Bone marrow transplantation: generation of EGFP-expressing BM chimeric mice
Donor BM cells were harvested, under sterile conditions, from 10 to 12 week old male C57BL/6 EGFP-expressing transgenic mice [Tg(CAG-EGFP)131Osb/LeySopJ; Jackson laboratories, USA]. Briefly, mice were euthanized by cervical dislocation and BM cells harvested by gently flushing both their femurs and tibias (using a 27G needle) with phosphate buffered saline (PBS) (pH 7.4), 2% foetal bovine serum (FBS), 1% Penstrep, 10 units/ml heparin. Cells were subsequently passed through a 40 µm cell strainer, washed twice in PBS (pH 7.4) by centrifugation at 600
g (10 min), and re-suspended in PBS to give a final concentration of ≥ 1 × 10
7 cells/150 µl. Young adult female recipient wild-type C57BL/6 mice (aged 12 weeks) were irradiated, with a single dose of 1000 rad from a 137Cs source, 6 h prior to receiving 1 × 10
7 unfractionated EGFP-expressing BM cells by tail-vein injection (there are well-established differences in the radiosensitivity of different inbred mouse strains [
14]; C57BL/6 mice typically require a single dose of 900–1100 rad to achieve both complete myeloablation and high levels of BM engraftment). Sterile water, antibiotics (Baytril), sterile food and bedding were all provided for 4 weeks post-transplant.
Detection of chimerism
At 12 weeks post BMT (female animals aged 24 weeks), haematopoietic reconstitution was evaluated in peripheral blood by flow cytometry (FACSCalibur, Becton–Dickinson). Briefly, 100 µl of peripheral blood was harvested from the tail vein and suspended in PBS [pH 7.4; ethylenediaminetetraacetic acid (EDTA), 2 mg/ml]. Red cells were removed using red cell lysis buffer, and the remaining nucleated cell population was re-suspended in PBS, 3% FBS and examined for EGFP expression when excited at 488 nm using flow cytometric analysis. Peripheral blood harvested from a non-transplanted C57BL/6 mouse was used as a reference control. Data were evaluated using BD Cellquest™ software.
Induction and evaluation of experimental autoimmune encephalomyelitis (EAE)
At 18 weeks post BMT, mice (female, now aged ~ 30 weeks) were immunised by subcutaneous injection, at the base of the tail, of 100 µl of a sonicated emulsion containing equal volumes of complete Freund’s adjuvant (CFA) (Difco) and PBS containing 200 µg myelin oligodendroglial glycoprotein (MOG) peptide p35–55. CFA was supplemented with 4 mg/ml of heat-killed Mycobacterium tuberculosis (Difco). Pertussis toxin (Sigma Aldrich, P2980) (200 ng) was administered intraperitoneally in 500 µl of PBS directly after immunisation and again 48 h later. Individual mice were assessed twice daily for clinical signs of EAE using the following scoring system: 0, no disease; 1, flaccid tail; 2, hindlimb weakness and/or impaired righting; 3, hindlimb paralysis; 4, hind and forelimb paralysis; 5, moribund.
Cerebellar slices
All female mice were culled aged between 9.5 and 11.5 months (~ 10 to 20 weeks after EAE induction), in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 and the University of Bristol Animal Welfare and Ethical Review Body. Parasagittal slices of cerebellar vermis (225 µm) were cut on a Leica VT1000S vibrating microtome (Leica Microsystems, Nussloch, Germany) in ice-cold solution (in mM: 62 NaCl, 124 sucrose, 1.3 MgSO4, 5 KCl, 1.2 KH2PO4, 26 NaHCO3, 10 D-glucose, 2.4, CaCl2, pH 7.4, bubbled with 95% O2, 5% CO2). They were stored in standard Krebs–Henseleit solution (in mM: 124 NaCl, 1.3 MgSO4, 5 KCl, 2.4 CaCl2, 1.2 KH2PO4, 26 NaHCO3, 10 d-glucose, pH 7.4, bubbled with 95% O2, 5% CO2) at room temperature for 1–8 h prior to extracellular recording from Purkinje cells. After recording, each cerebellar slice was fixed in 4% paraformaldehyde in PBS for 18 h at 4 °C and subsequently stored in PBS containing 0.1% sodium azide at 4 °C prior to immunohistochemistry. Cerebellar slices not used in extracellular recording were also fixed in 4% paraformaldehyde in PBS for 18 h at 4 °C and subsequently stored in PBS, 0.1% sodium azide at 4 °C. Slices were also made from female, age-matched control mice (9.5–11.5 months, C57BL/6 VAF/Elite).
Immunohistochemistry and imaging
When required, cerebellar slices stored in PBS, 0.1% sodium azide were washed in PBS. For immunofluorescent labelling, non-specific binding was blocked with 10% normal goat/donkey serum diluted in PBS containing 0.1% triton. Sections were incubated at 4 °C overnight with primary antibodies to calbindin-D
28K (Sigma-Aldrich; C2724 1:500), CD11b/c (Abcam; ab1211 1:100), EGFP (Abcam; ab6556 1:500 and Abnova; MAB1765 1:1000), glutamic acid decarboxylase (GAD) (Abcam; ab11070 1:1000); myelin basic protein (MBP) (Serotec; MCA4095 1:100); synaptophysin (Dako; M7315), SMI-34 (Biolegend; 835503 1:500); zebrin II (kindly donated by Prof. Izumi Sugihara, Tokyo Medical and Dental University, Tokyo, Japan [
39]). Sections were washed in PBS and incubated for 45 min in the dark with Alexa Fluor 488/555—goat/donkey anti-mouse (1:500),—goat/donkey anti-rabbit (1:500),—goat/donkey anti-rat (1:500) secondary antibodies (Invitrogen, Paisley, UK), and then mounted in Vectashield medium containing the nuclear dye 4′6′-diamidino-2-phenylindole (DAPI) (H-1200, Vector Laboratories).
A proportion of the cerebellar slices stored in PBS, 0.1% sodium azide were first embedded in paraffin for sectioning (8 µm) on a rotary microtome (Leica LM2135) and mounting on glass slides. Sections were deparaffinised in clearene, dehydrated in 100% ethanol and hydrated in distilled water. Antigen retrieval was performed, through boiling the sections in sodium citrate buffer (0.01 M, pH 6.0, 5 min), prior to immunofluorescent labelling (as described above). Confocal microscopic analysis was performed using either a Leica SP5-AOBS confocal laser-scanning microscope attached to a Leica DMI6000 inverted epifluorescence microscope or a Nikon C1 confocal microscope and EZ viewer software (Nikon). All Z-stack and three-dimensional imaging was created using Leica Application Suite Advanced Fluorescence software (Leica Biosystems) followed by Volocity 3D image software (PerkinElmer, USA).
In situ hybridization
Cerebellar slices were embedded in paraffin for sectioning (8 µm) on a rotary microtome (Leica LM2135) and mounting on glass slides. Sections were deparaffinised in clearene, dehydrated in 100% ethanol, and placed in 0.2 M HCl for 20 min. The sections were washed in water and subsequently in 2× saline-sodium citrate (SSC) buffer before immersion in 10 mM citric acid (pH 6.0) at 80 °C for 2 h. After washing with water and then 2× SSC buffer, proteins were digested for 2 min at 37 °C using pepsin A (Digest-All 3 protease, Invitrogen). The sections were washed in water, dehydrated through an ethanol series and air dried. Fluorescent in situ hybridisation (FISH) probes (mouse control X and Y probes; Empire genomics, New York, USA) were prepared according to the manufacturer’s recommendations and applied directly to the tissue sections; a coverslip was subsequently placed on top and sealed using rubber cement. DNA was denatured at 83 °C for 5 min and then renatured with FISH probes by overnight incubation at 37 °C. The following day, slides were washed by immersion for 2 min in 0.4× SSC, 0.1% Tween-20 buffer at 73 °C, followed by 1 min in 2× SSC, 0.1% Tween-20 buffer, at room temperature. The sections were air dried and mounted in Vectashield medium containing the nuclear dye DAPI (H-1200, Vector Laboratories).
Phenotypic analysis of fused Purkinje cells
Fused Purkinje cells in fixed, fluorescently labelled cerebellar slices were identified according to their location in the sagittal slice, characteristic morphology and co-expression of EGFP and calbindin-D28K. Each slice was scanned along the entire length of the Purkinje cell layer, situated between the granular layer and molecular layer. At least 2000 Purkinje cells from each mouse were examined to determine the frequency of Purkinje cell heterokaryons. The presence of two nuclei in these Purkinje cells was confirmed using confocal microscopy to view serial sections throughout the whole Purkinje cell soma. The frequencies of Purkinje cells expressing SMI-34 in the soma or with hypertrophic (enlarged) axons expressing SMI-34 were counted, scanning the entire Purkinje cell and granular layers.
FISH was also used to visualise nuclear chromosomal content of Purkinje cells. For X/Y chromosomal numeration and identification of Polyploid cells, FISH-labelled sections were viewed using confocal microscopy. Fluorescently labelled X and Y probes yielded green and red signals, respectively. Using confocal microscopy, cells were again scanned throughout the entire cell soma to observe the X/Y chromosomal frequency.
The cell soma area was measured in fixed sections stained with antibodies to EGFP and/or calbindin-D28K by manually outlining the perimeter in Z-stacked confocal microscopic images using Image J software (National Institutes of Health). Measurements were made of EGFP-positive/calbindin-D28K positive and adjacent or nearby EGFP-negative/calbindin-D28K positive Purkinje cells (n = ≥ 5 animals, ≥ 10 cells measured/per animal). ≥ 16 individual cells/group were used to compare the soma size of EGFP-positive Purkinje cells containing two dispersed nuclei and EGFP-negative Purkinje cells with both a dispersed and a compact nucleus.
Quantifying cerebellar inflammation and demyelination
CD11b/c was used to identify macrophages/microglia within tissue sections. All EGFP or CD11b/c positive cells were counted, using at least six randomly assigned fields per slice within both the white matter and grey matter regions of the cerebellum. The number of EGFP or CD11b/c positive cells per mm2 was subsequently calculated.
The degree of demyelination in both the cerebellar white and grey matter (granular layer) was assessed by quantifying anti-MBP labelling by optical density. Using at least three randomly assigned sagittal cerebellar slices per animal, confocal images of MBP immuno-labelled cerebellar sections from each animal were processed using Image J and MBP density was calculated as a percentage of the total area.
Individual slices were viewed on a Zeiss FS Axioskop microscope (Carl Zeiss Ltd., Welwyn Garden City, UK) 1–8 h after cutting and superfused with standard Krebs–Henseleit solution at near physiological temperature (~ 34 ± 1 °C, maintained with a home-made Peltier-controlled recording chamber). Purkinje cells were readily identified by their position in the slice and characteristic size and shape. In slices from female mice with EAE, Purkinje cells fused with BM-derived cells were identified by EGFP fluorescence. Since < 1% of Purkinje cells were EGFP-positive, it was rare for such a cell to be at the surface of a slice and readily accessible for recording. Therefore, once an EGFP-positive cell was identified, it was often necessary to remove overlying tissue prior to recording. This was achieved with a fine stream of extracellular solution from a macropipette (tip diameter, ~ 5 µm). Extracellular recordings were made from the soma for at least 10 min in loose cell-attached mode with low resistance pipettes (1.2–1.9 MΩ, thin-walled borosilicate glass; GC150Tf-10, Harvard apparatus; filled with extracellular solution). To ensure that spontaneous firing was not influenced by the presence of the pipette on the soma, we prevented formation of a tight ‘seal’ between the pipette and soma by pre-exposing the pipette tip to the tissue in the slice [
1]. The ‘seal’ resistance was < 9 MΩ. Spontaneous electrical activity was recorded in voltage-clamp with an Axopatch 200A or 200B amplifier (Axon Instruments, Union City, CA). Current recordings were low-pass filtered at 5 kHz (4 pole Bessel filter in the Axopatch 200 A/B amplifier) and digitised on-line at 40 kHz with a Cambridge Electronic Design (CED) power 1401 A/D interface using Spike2 software (v. 6.16) (CED, Cambridge, UK). In slices from mice with EAE, recordings were made from EGFP-positive cells and adjacent or nearby EGFP-negative cells. Recordings were also made from Purkinje cells in slices of sex- and age-matched control mice (female, no BMT, no EAE induction, 9.5–11.5 months).
Recordings were analysed with CED Spike2 software. Three different patterns of firing of control Purkinje cells were defined by the presence or absence of distinct, long pauses in firing (> 1.6 s), the distribution of interspike intervals (ISI) during firing periods and the shape of plots of instantaneous frequency (IF) against time. For tonically firing cells, the ISI distribution showed a single peak with a mean between 8 and 21 ms and a coefficient of variation (CV) between 0.08 and 0.80. This tonic category includes cells firing persistently in a regular tonic pattern (mean, 9–21 ms: CV, 0.08–0.21; e.g. see Online Resource 1; Fig. S1ai) or a regular tonic pattern interrupted by bursts of firing (mean, 8–13 ms; CV, 0.28–0.85; e.g. see Online Resource 1; Fig. S1aii). For cells firing in a trimodal manner (see Online Resource 1; Fig. S1b), there were distinct long pauses in firing (1.6–15 s) and the distribution of ISI during firing periods showed multiple components. The shortest ISI (< 10 ms) occurred within bursts and the longer ISI (25–50 ms) occurred between bursts, while intermediate ISI (~ 20 ms) separated spikes during regular tonic firing. The differing ISI resulted in a high CV (0.6–2.4). Cells were placed into this category if they showed all three modes of firing irrespective of the order; for example, it includes cells following a tonic, burst, pause, tonic, burst, pause pattern as well as cells following a tonic, burst, tonic, burst, pause pattern. For cells classified as firing in an irregular pattern (see Online Resource 1 Fig. S1c), spikes occurred as single events or in clusters of closely spaced two, three or four events. The ISI distribution was broad (2–500 ms) or it had two components, in which the first component represented intervals between spikes within clusters and the longer component represented intervals between clusters. The mean and CV ISI values were 15–43 and 0.8–1.3 ms. In some of these cells, the firing was separated by distinct pauses in firing (7–10 s).
Statistical analysis
Statistical tests were performed using GraphPad Prism (v. 6 or 7, GraphPad Software Inc, USA). All tests were two-sided; values of p < 0.05 were considered statistically significant. For histological analysis, at least five independent animals from each group were included. Where data were known or predicted to violate assumptions for parametric statistical testing, an equivalent non-parametric test was performed (normality was analysed using the Shapiro–Wilk test; unequal variances between groups were subsequently analysed using either the Bartlett’s test or F test). Data between two groups were analysed using paired/unpaired Student’s t tests, Wilcoxon matched-pairs signed rank tests or Mann–Whitney U tests. Statistical comparisons for more than two groups were analysed using either one-way analysis of variance (ANOVA) followed by Holm–Sidak’s multiple comparisons test or Kruskal–Wallis followed by Dunn’s multiple comparison test between groups. Box and whisker plots represent the median, upper quartile and lower quartile values (box) and the minimum and maximum values (whiskers). For electrophysiological data, the Chi-squared test was used to compare the frequency of different modes of firing. One-way or two-way ANOVA followed by the Holm–Sidak’s multiple comparisons test was used to compare different firing parameters. Number of cells is denoted as n. Origin (v. 6 or 7, Microcal, Northampton, MA, USA) was used to plot voltage traces.
Discussion
This study supports the hypothesis that by fusing with neurons in the adult brain, BM-derived cells can prevent or repair pathological changes triggered in mature brain neurons by neuroinflammation. We show that the cell fusion in vivo maintains not only the structure and subcellular composition of cerebellar Purkinje cells; it also maintains the electrical firing properties of these neurons. Our study finally establishes a likely functional role for heterotypic cell fusion, a process identified more than a decade ago [
2,
41] and proposed as a mechanism for the existence of Purkinje cells with two nuclei, first recognised almost 80 years ago [
3]. The conservation of normal spontaneous firing typical of the original native neuron demonstrates that the heterokaryon retains neuronal function, which is to fire action potentials and hence transfer information through the neuronal circuitry.
Neuroinflammation induced by EAE caused both structural and functional abnormalities in cerebellar Purkinje cells. There was also an increased incidence (20-fold) of fusion between BM-derived cells and Purkinje cells; these elevations in fusion are similar to those reported in EAE mice by Johansson et al. [
20]. Pathologically, we found demyelination of Purkinje cell axons in the cerebellar cortex and evidence of injury to both the Purkinje cell axon and soma; these changes bear significant resemblance to features of multiple sclerosis, where abnormal neurofilament phosphorylation in Purkinje cells and robust Purkinje cell loss within lesional areas are reported [
5,
11,
25,
34]. The shift from a tonic mode to a trimodal mode of firing of Purkinje cells in ex vivo cerebellar slices represents an increased tendency to burst rather than tonic firing. This is in agreement with the altered firing patterns recorded previously from in vivo Purkinje cells in EAE animals [
35]. Aberrant expression of the sensory neuron specific sodium channel, Na
v1.8, may be partly responsible [
35] although changes in expression levels of the various ion channels and transporters that shape Purkinje cell firing are likely to be implicated [
5,
7,
11,
16,
17]. It is possible, therefore, that fusion with BM-derived cells maintains Purkinje cell firing and/or interaction with neighbouring cells through normalisation of ion channel, transporter or receptor expression. This is predicted to occur by nuclear transfer from the BM-derived cells and subsequent exposure to transcription factors within the Purkinje cell fusion partner, resulting in the expression of donated genes to restore ‘normal’ gene expression. A clue that this occurs is the observed translation of the
EGFP gene, donated by the BM-derived cell following fusion. Further insight into the genes expressed by fused Purkinje cells could be obtained through exploiting species-mismatched BMT (transplantation of rat BM into mice [
20]). Subsequent single-cell mRNA analysis of fused cells, using species-specific primers to genes involved in neurophysiological function known to be differentially expressed in EAE (e.g.
Cacna1a,
Scn10a;
S100a10 [
5,
7,
11]), may provide key answers. Characterisation of the corrective changes in gene expression and identification of the channel or transporter genes involved requires further study.
In response to cerebellar injury, why BM-derived cells fuse preferentially with Purkinje cells is unknown. Purkinje cells are likely to express membrane-specific molecules rendering them fusogenic. As there is no evidence for Purkinje cell generation in adult life [
28], the potential evolutionary-driven capacity to express these molecules, allowing the targeted fusion of BM cells with Purkinje cells to prevent their loss, could relate to their unique physiological significance in the cerebellum. The notion that cell fusion in the brain may have a role in neuronal repair has significant implications for regenerative medicine. Degeneration of the cerebellum, and particularly Purkinje cells, occurs in numerous acquired and inherited neurological disorders [
4,
6,
19,
24,
34,
37]. Moreover, many of these disorders are thought to share common pathogenic pathways that result in altered intrinsic activity of Purkinje cells [
32].
BM-derived cells fusing with Purkinje cells in response to inflammation is of particular importance to cerebellar injury, as cell fusion may be an immune-mediated phenomenon aimed at protecting Purkinje cells against inflammatory or toxic insults [
21]. As we report here in mice, inflammatory disease-associated increases in BM-derived cells fusing with Purkinje cells have also been observed in patients who had multiple sclerosis [
22]. There is also evidence that underlying levels of cerebellar inflammation influence Purkinje cell fusion and heterokaryon formation in patients with genetic ataxias [
24]. Exploiting heterotypic cell fusion as a mode of cell rescue, to introduce ‘therapeutic’ genetic material to boost neuronal cell survival may, therefore, hold hugely valuable implications for a wide range of patients with otherwise untreatable neurodegenerative or inflammatory disease. Indeed, there are equally important implications for aging in the CNS and the preservation of highly complex neuronal sub-populations.
Whole BM and selected BM-derived haematopoietic and mesenchymal stem cell fractions have been studied experimentally with the ultimate purpose of treating cerebellar disease [
12]. Yet, there are still significant deficiencies in our understanding of which precise BM cell therapy protocol is most efficacious. Central to this, understanding whether selected BM cell populations or the whole BM are required to reduce disease activity and effect repair is of paramount importance. Understanding whether there are fundamental differences in the ‘fusogenic’ properties of different BM cell types will help aid the design of future therapeutic trials utilizing transplantation of BM cells in these disorders. Exploring these precise interactions between BM-derived cells and Purkinje cells is likely to aid the search for much needed novel reparative and regenerative therapeutic approaches into ways in which nerve cells can be protected and their survival prolonged.