Introduction
Neuromyelitis optica spectrum disorders (herein called NMO) is an inflammatory demyelinating disease of the central nervous system with characteristic pathological features in spinal cord and optic nerves, and to a lesser extent in brain. Most NMO patients are seropositive for immunoglobulin G autoantibodies against astrocyte water channel aquaporin-4 (AQP4), called AQP4-IgG (or NMO-IgG) [
13,
14]. The major pathological features in seropositive NMO include astrocyte damage, inflammation with prominent granulocyte and macrophage infiltration, vasculocentric deposition of activated complement, and demyelination, which can produce marked neurological deficits [
10,
19,
26]. There is abundant evidence that pathogenesis in AQP4-IgG seropositive NMO patients involves AQP4-IgG binding to AQP4 on astrocytes and activation of the classical complement system, which causes complement-dependent cytotoxicity (CDC) leading to inflammation, blood–brain barrier disruption and demyelination [
8,
13,
19]. Antibody-dependent cell-mediated cytotoxicity (ADCC) [
24] and sensitized T cells [
22,
35,
36] may also play a role in NMO pathogenesis.
Several lines of evidence implicate a major role for complement activation in NMO, including human pathology showing deposition of activated complement [
16,
18,
26], rodent models showing complement-dependent NMO pathology following passive transfer of AQP4-IgG [
1,
28,
37], and an open-label clinical trial of the C5 convertase inhibitor eculizumab showing efficacy in NMO [
21]. We previously reported that complement inhibitor protein CD59, a phosphoinositol-linked membrane glycoprotein expressed on astrocytes that inhibits formation of the terminal membrane attack complex, may be an important regulator of complement action in NMO [
38]. CD59
−/− mice are highly sensitive to administration of AQP4-IgG and human complement, with longitudinally extensive NMO spinal cord pathology produced by coinjection of AQP4-IgG and complement into the lumbosacral cerebrospinal space. However, a major limitation of mice as models of NMO is the near-zero activity of their classical complement pathway, in part because of complement inhibitory factor(s) present in mouse serum [
25]. The ineffective classical complement pathway in mice precludes the development of clinically relevant NMO models, such as robust passive-transfer models of NMO optic neuritis and transverse myelitis, as well as testing of NMO therapeutics such as complement inhibitors.
To overcome these limitations and to further investigate the role of CD59 in NMO pathogenesis, here we generated CD59
−/− rats and determined their sensitivity to passive transfer of AQP-IgG. We previously showed that passive transfer of AQP4-IgG to rats, without added complement, by a single intracerebral injection produced NMO pathology in brain at the site of injection [
1]. We tested here the prediction that marked NMO pathology might be produced in the central nervous system by passive transfer of AQP4-IgG to CD59
−/− rats, without added complement, under conditions where minimal pathology is produced in CD59
+/+ rats.
Discussion
Our study supports the central involvement of CD59 in modulating complement-mediated injury in AQP4-IgG seropositive NMO. CD59 is expressed in CNS tissues affected in NMO and may play a protective role to contain local, subclinical injury initiated by minor exposures to AQP4-IgG. CD59−/− rats were highly sensitive to passive transfer of AQP4-IgG by intracerebral and intracisternal routes, without the need for added components such as complement or pro-inflammatory factors. Though astrocytes may also express other complement regulator proteins such as CD55, the marked effect of CD59 gene deletion suggests that CD59 is the major complement regulator in rat brain. As an important complement regulator in astrocytes, drugs that enhance astrocytic CD59 expression, perhaps identifiable by compound screens, may be beneficial in NMO, and conversely, reduced astrocytic CD59 expression or subcellular colocalization with AQP4 might trigger NMO exacerbations and worsen disease severity.
Animal models of NMO have been useful in characterizing NMO pathogenesis mechanisms and for testing potential NMO therapeutics. Mouse models have been developed involving passive transfer of AQP4-IgG together with human complement by direct injections into the brain [
28] or spinal fluid [
3,
35] to produce NMO-like pathology in brain, spinal cord and optic nerve. As mentioned in the Introduction, a fundamental limitation of mice to study NMO is their lack of an effective classical complement activation pathway [
5,
25]. Early rat models involved administration of AQP4-IgG following induction of experimental autoimmune encephalomyelitis (EAE) [
4]; however, the pathogenic mechanism in EAE – myelin targeting by T cells – is very different from the humoral immune response in NMO, making it difficult to reach conclusions about NMO pathogenesis mechanisms. We found that intracerebral injection of AQP4-IgG produced robust NMO-like pathology in rat brain [
1], and that while systemic administration of AQP4-IgG alone did not produce disease, NMO-like brain pathology was seen following a small needle stab in seropositive rats [
2], which presumably allowed circulating AQP4-IgG leakage into brain parenchyma to access astrocytes, and perhaps produce a local inflammatory response. Creation of NMO spinal cord or optic nerve pathology in rats has been challenging. One study involving continuous AQP4-IgG infusion using intrathecal catheters showed reversible AQP4 loss in spinal cord but without inflammation or demyelination [
9], and a similar more recent study reported AQP4 loss in spinal cord and optic nerves, as well as mildly reduction in myelin in spinal cord [
17]. The marked amplification of NMO pathology by knockout of CD59 in rats produced astrocytopathy as well as inflammation and deposition of activated complement.
CD59
−/− rats did not manifest overt phenotypes, except for mild reticulocytosis and reduced hemoglobin, which is likely due to low-grade hemolysis as seen in humans lacking CD59 [
31] rather than a possible off-target effect in genome editing that can occur using CRISPR methods. The active classical complement system in rats, which has similar activity to that in human [
5,
33], is presumably the reason for the low basal hemolytic activity. As such, CD59
−/− rats may be useful to model complement-initiated diseases in various neurodegenerative, hematological, renal and skeletal muscle diseases [
6,
11,
31]. Although the mechanism of high morbidity in CD59
−/− rats receiving cobra venom factor was not established here, there appeared to be hemolysis and organ injury, which is likely due to complement activation and consumption by cobra venom factor, which is the mechanism of its complement depletion action [
32,
33]. With regard to NMO, the amplified response of CD59
−/− rats to AQP4-IgG may be useful in testing drugs that target distinct steps in the AQP4-IgG/complement injury pathway, as well as in investigating outstanding questions in NMO pathogenesis mechanisms such as the role of sensitized T cells and the explanation for the absence of significant pathology in peripheral AQP4-expressing tissues despite their sustained direct exposure to serum AQP4-IgG.
The marked NMO pathology seen in CD59
−/− rats following AQP4-IgG administration contrasts with the conclusions of Saadoun and Papadopoulos [
27], who concluded that complement inhibitors, including CD59, are not protective against complement injury in CNS tissues. Their findings were based on immunofluorescence in mouse brain in which CD59 expression was seen on astrocytes, but not at AQP4-rich foot-processes abutting microvessels. Detection sensitivity rather than species differences may account for the disparate conclusions, as we previously showed marked NMO pathology in CD59
−/− mice following intracerebral or lumbosacral administration of AQP4-IgG with human complement [
38]. Our recent development of super-resolution microscopy methods to image AQP4 on astrocytes in fixed CNS tissues [
29] may overcome the limited resolution and sensitivity of conventional fluorescence microscopy to detect CD59 in subcellular regions of astrocytes. Saadoun and Papadopoulos [
27] also speculated that the absence of significant NMO disease in peripheral AQP4-expressing tissues such as skeletal muscle and kidney was a consequence of CD59 and AQP4 coexpression, which should be amenable to testing using CD59
−/− rats.
Acknowledgments
This work was supported by grants EY13574, EB00415, DK35124, and DK72517 from the National Institutes of Health, and a grant from the Guthy-Jackson Charitable Foundation. We thank Dr. Jeffrey Bennett (Univ. Colorado Denver, Aurora, CO) for providing recombinant monoclonal NMO antibodies and Tao Su (UCSF) for help in astrocyte and slice culture studies.