Background
Neuromyelitis optica (NMO) is an autoimmune inflammatory disease of the central nervous system that causes demyelinating lesions in optic nerve and spinal cord, leading to loss of visual and motor function [
1‐
3]. A specific feature of NMO is the presence of serum immunoglobulin G (IgG) autoantibodies (NMO-IgG) against astrocyte water channel aquaporin-4 (AQP4) [
4,
5]. NMO pathogenesis is thought to involve NMO-IgG binding to AQP4 on astrocytes, which causes complement- and cell-mediated astrocyte cytotoxicity, inflammation, and blood–brain barrier (BBB) disruption, with secondary oligodendrocyte and neuron damage [
6‐
8]. Current therapies of NMO include general immunosuppression, B-cell depletion and plasma exchange [
9,
10].
Although optic neuritis (ON) with permanent loss of vision is a major clinical feature of NMO [
11‐
13], adequate models of NMO ON are lacking. The particular sensitivity of the optic nerve in NMO suggests the need to study disease mechanisms and treatment responses in optic nerve-specific NMO models. Disease-relevant animal models of NMO are important for investigating pathogenesis mechanisms, such as the role of inflammatory effector cells [
14‐
16] and for testing of potential therapeutics such as antibodies targeting AQP4 [
17] or complement [
18,
19]. The original models of NMO involved administration of NMO-IgG to rats with pre-existing neuroinflammation produced by experimental autoimmune encephalomyelitis, in which immunization with a myelin oligopeptide produces an anti-myelin T-cell response [
20‐
22]. Subsequently, a passive-transfer mouse model of NMO involving intracranial injection of NMO-IgG and human complement recapitulated key pathological findings in NMO, including loss of AQP4 and glial fibrillary acidic protein (GFAP) immunoreactivity, granulocyte and macrophage infiltration, vasculocentric deposition of activated complement, and demyelination [
23]. However, due in part to the limited diffusion of AQP4-IgG and complement from the injection site, pathology in this model was confined to a small region around the injection site, sparing the optic nerves.
The purpose of this study was to establish an animal model of NMO ON involving passive transfer of NMO-IgG with targeted delivery to the optic nerves. Mice were chosen for these studies because of the availability of relevant knockout strains (AQP4 and CD59). After testing various approaches we established the conditions in which delivery of NMO-IgG to optic nerves produced ON with characteristic NMO pathology.
Methods
Mice
In vivo studies were performed on 8- to 10-week-old, weight-matched AQP4
+/+ and AQP4
-/- mice in CD1 genetic background, which were generated as described previously [
24]. Some experiments were done on CD59
+/+ and CD59
-/- mice on a C57bl/6 background (provided by Dr Xuebin Qin, Harvard University, USA). Littermates were used as wild-type controls for the AQP4 and CD59 knockout mice. Mice were maintained in air-filtered cages and fed normal mouse chow in the University of California, San Francisco (UCSF) Animal Care facility. All procedures were approved by the UCSF Committee on Animal Research.
Neuromyelitis optica (anti-aquaporin-4) antibodies
Recombinant monoclonal NMO antibody rAb-53 (referred to as NMO-IgG) was generated from a clonally expanded plasma blast population from cerebrospinal fluid of an NMO patient, as described and characterized previously [
22,
25]. Purified rAb-53 was used for studies here because of its high affinity for AQP4, and to eliminate the potential variability introduced by using NMO patient serum, which is polyclonal and may contain other antibodies or soluble factors that influence NMO pathogenesis. A NMO ‘superantibody’ with enhanced complement-dependent cytotoxicity (referred as NMO-IgG
CDC+) was generated as described previously [
26] by introducing mutations (G236A/S267E/H268F/S324T/I332E) in the Fc portion of rAb-53 [
27].
Neuromyelitis optica immunoglobulin G antibody delivery to anterior optic nerve and retina
Adult mice were anesthetized with intraperitoneal tribromoethanol (avertin, 250 to 500 mg/kg). Lateral canthotomy was done under a dissecting microscope. Ocular muscles were retracted and anterior optic nerve was exposed to infuse locally 1 μg NMO-IgG and 0.5 μL human complement (Complement Technology, Tyler, TX, USA) in a total volume of 1.5 μL. For intravitreal injection, a 32-gauge needle attached to a 10-μL gas-tight Hamilton syringe was passed through the sclera, next to the limbus, into the vitreous cavity. NMO-IgG (1 or 3 μg) and 0.5 μL human complement in a total volume of 2 μL was injected (0.5 μL per minute) above the optic nerve head.
Neuromyelitis optica immunoglobulin G antibody delivery to posterior optic nerve
Adult mice were anesthetized and mounted on a stereotaxic frame. A midline scalp incision was made and a burr hole of diameter 1 mm was drilled in the skull 1-mm right and 1-mm anterior to bregma. For single administration of NMO-IgG, a 30-gauge needle attached to a 50-μL gas-tight syringe was inserted through the brain (6 mm below the dura down to base of the skull) near the optic chiasm to deliver 5 μg NMO-IgG and 5 μL human complement in a total volume of 10 μl. For continuous administration of NMO-IgG, an osmotic minipump (Alzet 1003D, Cupertino, Ca, USA) delivered 3.3 μg NMO-IgG and 16.7 μL human complement per day for 3 days.
Immunofluorescence
Optic nerves were post-fixed for 2 hours in 4% paraformaldehyde. Ten micrometer-thick frozen sections were immunostained at room temperature for 1 hour with antibodies against AQP4 (1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA), GFAP (1:100, Millipore, Temecula, CA, USA), myelin basic protein (MBP; 1:200, Santa Cruz Biotechnology), ionized calcium-binding adaptor molecule-1 (Iba1; 1:1,000; Wako, Richmond, VA, USA), albumin (1:200, Santa Cruz Biotechnology), C5b-9 (1:100, Santa Cruz Biotechnology), neutrophil (Ly-6G, 1:100, Santa Cruz Biotechnology), eosinophil (siglec-F, 1:50, BD Biosciences, Oxford, UK), macrophage (F4/80, 1:100, Santa Cruz Biotechnology) or CD45 (1:10, BD Biosciences) followed by the appropriate fluorescent secondary antibody (1:200, Invitrogen, Grand Island, NY, USA). Immunofluorescence was examined with a Leica (Wetzlar, Germany) DM 4000 B microscope or Nikon (Melville, NY, USA) laser-scanning confocal microscope. Areas were defined by hand and quantified using ImageJ software (National Institutes of Health).
Retinal ganglion cell labeling
Retinal ganglion cells (RGCs) were labeled as described previously [
28‐
30]. Briefly, mice were injected with 1 μL neurotracer dye FluoroGold (4% solution in saline; Fluorochrome, Denver, CO, USA) in the superior colliculus (from the bregma, anterior-posterior, -3 mm; medial-lateral, +0.5 mm; 2 mm below the dura) 7 days before NMO-IgG and human complement delivery to posterior optic nerve. Retinas were flattened 14 days after the FluoroGold injection and whole-mounts were fixed in 4% paraformaldehyde. RGCs were counted manually under a fluorescence microscope with a 40× objective. A total of 16 to 20 images per retina were used for cell counting. As a positive control for RGC loss, optic nerve crush injury was produced by 30-second compression of the optic nerve 2 mm posterior to the globe insertion using cross-action forceps. The optic nerve was accessed by lateral canthotomy and dissection beneath the lateral rectus. RGCs were quantified 7 days after crush.
Statistical analysis
Values are presented as mean ± SEM. Comparisons between two groups were performed using the unpaired Student's t-test. P < 0.05 was considered statistically significant.
Discussion
Current understanding of NMO pathogenesis comes largely from data on brain and spinal cord, as there is little descriptive pathology of optic nerves in human NMO and adequate animal models of NMO ON have not been developed. The bulk of evidence supports a pathogenesis mechanism that involves NMO-IgG access to the central nervous system and binding to AQP4 on astrocytes, which causes complement- and cell-mediated cytotoxicity [
22,
26,
39‐
42]. The primary astrocyte damage initiates an inflammatory reaction with cytokine release, granulocyte and macrophage infiltration, and further BBB disruption, which produces secondary oligodendrocyte injury, demyelination and neuron loss. This mechanism is supported by pathology in human spinal cord and brain [
43,
44], by brain pathology in mice following passive transfer of NMO-IgG by intracerebral injection [
23], and by
ex vivo studies in spinal cord slice cultures exposed to NMO-IgG and various effector molecules and cells [
45]. It has been assumed without direct evidence that a similar pathogenesis mechanism applies to NMO ON. Studies done in
ex vivo optic nerve cultures exposed to NMO-IgG and complement showed an astrocytopathy with demyelination, but this model is limited by the short-term (~1 day) viability of optic nerve cultures [
45].
The main finding of the study here is that passive transfer of NMO-IgG and human complement to mice by continuous intracranial infusion near the optic chiasm produces optic nerve lesions with loss of AQP4 and GFAP immunoreactivity, demyelination, and inflammation with prominent macrophage infiltration. Optic nerve pathology was exacerbated when NMO-IgG was administered to CD59 knockout mice or when a mutated NMO antibody with enhanced complement effector function was administered to wild-type mice, supporting a central role of complement in NMO ON in our model. Control studies, including infusions in AQP4 knockout mice, indicated that the pathological changes require NMO-IgG, complement and AQP4. The passive transfer model of NMO ON established here should be useful in studying mechanisms of NMO pathogenesis specific to the anterior visual pathway, as well as in evaluating potential vision-preserving therapies using clinically relevant in vivo anatomic and functional outcome measures.
Prior models of general neuroinflammation include optic nerve pathology, though the pathogenesis mechanisms are very different from that of NMO, which involves a humorally mediated astrocytopathy. As mentioned in the Introduction, ON is a well-described feature of experimental autoimmune encephalomyelitis. ON has also been seen in a subpopulation of transgenic mice expressing myelin oligodendrocyte glycoprotein (MOG)-specific T cell receptors [
46]. Crossing this mouse with a MOG-specific Ig heavy-chain knock-in mouse produced mice with both T and B cell MOG reactivity [
29]. This double transgenic mouse manifests selective optic nerve and spinal cord pathology (with 60% penetrance) and a Th17 differentiation bias, reminiscent of NMO [
47]. While ON is seen in these various models, they are probably not useful to study NMO pathogenesis mechanisms or test NMO therapeutics.
Several approaches to deliver NMO-IgG and complement were tested to produce robust ON in mice. Delivery of NMO-IgG and complement to the anterior optic nerve following lateral canthotomy did not produce optic nerve pathology, probably because of limited access of infused macromolecules to the ensheathed anterior optic nerve. Though intravitreal injection of NMO-IgG and complement resulted in NMO-IgG binding to AQP4 in retinal Müller cells, no retinal pathology was seen, perhaps because of limited access of some complement components to inner retinal layers. Also, intravitreally delivered solutes and macromolecules are unable to diffuse into the optic nerve. From these initial studies, we postulated that delivery of NMO-IgG and complement near the unsheathed perichiasmal posterior optic nerve might produce NMO ON. Though a single intracerebral injection of NMO-IgG and complement with the needle tip near the optic chiasm resulted in some NMO-IgG binding to optic nerve, as well as NMO pathology in brain near the optic nerve, little NMO pathology was found. Robust NMO pathology required continuous intracerebral injection of NMO-IgG and complement near the optic nerve. We conclude that the sustained exposure of target tissues to pathogenic macromolecules afforded by continuous infusion better recapitulates human NMO in which progressive pathology involves an amplifying cycle of astrocyte cytotoxicity, inflammation and BBB disruption. We cannot exclude, however, the possibility that the prechiasmal optic nerve and optic chiasm may be more susceptible to NMO-IgG-mediated pathology than the anterior optic nerve.
Though the NMO ON model developed here produced robust lesions with characteristic NMO pathology, there are a number of limitations of the model and potential directions for future advances. Continuous intracerebral infusion with precise needle placement is invasive and technically challenging. The direct administration of human complement, which was necessary because of the weak activity of mouse complement and the presence of complement inhibitory factor(s) in mouse serum [
48], does not accurately recapitulate the human disease in which endogenous complement proteins derive primarily from the serum. We recently established a rat model of NMO involving intracerebral administration of NMO-IgG without added complement, which produced robust NMO pathology in brain around the needle track [
16]. Pathology required active rat complement, as complement inactivation by cobra venom factor prevented the astrocyte cytotoxicity and demyelination. In a recent variation of the rat model, NMO pathology was produced in brain by peripheral NMO-IgG administration and focal mechanical BBB disruption (Asavapanumas and colleagues, unpublished results). Models of NMO ON in rats involving peripheral or optic nerve-targeted NMO-IgG delivery, perhaps in combination with maneuvers to disrupt the blood-optic nerve barrier, may produce ON without the need to administer complement.
It remains unclear why NMO pathology is primarily restricted to optic nerve and spinal cord, and to a lesser extent in brain, with little or no pathology in peripheral AQP4-expressing tissues. Optic nerve susceptibility in NMO is unlikely due solely to NMO-IgG access, and may involve impaired diffusion of NMO-IgG, soluble pro-inflammatory factors and complement proteins from focal areas of central nervous system entry. Optic nerve susceptibility in NMO might also arise from the high AQP4 expression in the optic nerve compared to brain [
47] and the abundance of large orthogonal arrays of particles in perivascular astrocytic end-feet of the optic nerve [
49‐
51] that promote tight binding of AQP4-IgG and efficient CDC [
52]. Plasmablasts in the cerebrospinal fluid secreting NMO-IgG locally [
22] and/or regional variations in the expression of complement regulator or key BBB proteins may also play a role.
Acknowledgments
Supported by grants EY13574, EB00415, DK35124, HL73856, DK86125 and DK72517 from the National Institutes of Health, and a grant from the Guthy-Jackson Charitable Foundation. We thank Accelerated Cure (Waltham, MA, USA) for providing human NMO sera.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
NA and JR carried out experimental work and wrote the manuscript draft. MCP, JLB, MHL and ASV designed experiments and edited the manuscript. All authors read and approved the final manuscript.