Detection of anti-aquaporin-4 antibodies in neuromyelitis optica: current status of the assays.
Abstract: Background: Neuromyelitis optica (NMO) is a severe inflammatory demyelinating disease that predominantly affects the optic nerve and spinal cord. Since the discovery of a specific serum marker for NMO in 2004, and its subsequent identification as an antibody to aquaporin-4 (AQP-4), various methods have been developed to test for the antibodies in patients' sera.

Objective: To assess the principal methods used to measure AQP-4 antibodies in patients' sera, describe their contribution to neuromyelitis spectrum disease and examine their value in the early detection of disease.

Methods: We compared the published data on each assay and used the relapsing NMO cohort as a uniform patient group for direct assay comparison.

Results: The indirect immunofluorescence assay, a cell-based assay (CBA) and a fluorescence-based immuno precipitation assay have broadly similar high sensitivities (86%, 91% and 83%) in the relapsing cohort, but the indirect immunofluorescence has a lower specificity (91%) compared with the other two (both 100% specific). Conclusions: The indirect immunofluorescence assay for NMO-IgG allows the detection of antibodies in routine screening, but the CBA for AQP-4 antibodies is the most sensitive. The fluoroimmunoprecipitation assay is a potentially high-throughput test for identifying positive sera and for serial estimations of antibody levels, but in its present form is slightly less sensitive. Overall, these assays are proving very useful in helping to diagnose those patients with longitudinally extensive transverse myelitis or recurrent optic neuritis who are likely to have relapsing NMO, including patients with myelopathy and Sjogren's syndrome, but it appears to be less often positive in patients with monophasic NMO.


Neuromyelitis Optica; Transverse Myelitis; Multiple Sclerosis; Optic Neuritis; Sjogren's Syndrome, Assay, Immunoprecipitation; Indirect Immunofluorescence
Article Type: Report
Subject: Antibodies (Analysis)
Viral antibodies (Analysis)
Medical research (Analysis)
Medicine, Experimental (Analysis)
Aquaporins (Analysis)
Rheumatoid arthritis (Analysis)
Neuromyelitis optica (Analysis)
Precipitation (Meteorology) (Analysis)
Authors: Waters, P.
Vincent, A.
Pub Date: 11/01/2008
Publication: Name: The International MS Journal Publisher: PAREXEL MMS Europe Ltd. Audience: Academic Format: Magazine/Journal Subject: Health Copyright: COPYRIGHT 2008 PAREXEL MMS Europe Ltd. ISSN: 1352-8963
Issue: Date: Nov, 2008 Source Volume: 15 Source Issue: 3
Product: Product Code: 8000200 Medical Research; 9105220 Health Research Programs; 8000240 Epilepsy & Muscle Disease R&D NAICS Code: 54171 Research and Development in the Physical, Engineering, and Life Sciences; 92312 Administration of Public Health Programs
Accession Number: 221274691
Full Text: Introduction

Aquaporin-4 (AQP-4), the principal water channel in the central nervous system, controls water movement between the brain, cerebrospinal fluid and blood (see Figure 1 for the topology and structure of AQP-4; PDB2D57). (1) AQP-4 was identified as a target antigen in neuromyelitis optica (NMO) in 2005 (2) leading to the recognition of a group of diseases that may have a similar autoimmune process based on the presence of this antibody. (3-5) These include not only patients who fulfil the criteria for NMO but also patients with longitudinally extensive transverse myelitis (LETM), Sjogren's syndrome with myelopathy, and some with recurrent optic neuritis (ON) or systemic lupus erythematoses (SLE). (6) Some of the NMO spectrum diseases can be confused with multiple sclerosis (MS) on initial presentation but disability accrues early in NMO. Consequently, many different groups have tried to develop assays to detect AQP-4 in patients' sera to facilitate early diagnosis and appropriate treatment. Here, we review the five main methods that have been developed, comment on their advantages and disadvantages, and compare them using the relapsing cohort of NMO patients as the most uniform group available from each study.


Review of the Methods

Indirect Immunofluorescence

NMO-IgG detection by indirect immunofluorescence (IIF) was the original method used to identify an antibody specific for NMO. This technique was adapted from one that has been used in many clinical laboratories to detect auto-antibodies binding in tissue-specific patterns in a variety of autoimmune or paraneoplastic diseases. Dilutions of each serum are incubated with a composite of frozen mouse sections that includes the cerebellum and midbrain. (2,7) After washing, a secondary antibody, fluorescein-conjugated goat anti-human IgG, is applied to the sections. Finally, Hoescht stain is added to allow visualization of the nuclei under the fluorescence microscope. For NMO-IgG, the sera need to be pre-adsorbed with guinea-pig liver powder to reduce binding of IgG to non-neuronal antigens. NMO-IgG positive sera show a typical staining pattern around blood vessels, along the pial surfaces and Virchow-Robin spaces (see Figure 2A for a cartoon representation of this method and Figure 3A for an example). Although screened initially at 1:60, doubling dilutions showed titres between 0 and 1:30,720. Four other groups have published data based on this method (Table 1), and in general this method demonstrates lower sensitivity, 54-73% for NMO, and tends to have lower specificity (91-100%) than the antigen-specific methods (Table 1). Reduced sensitivity might stem from the use of mouse rather than human tissue--there are four amino acid differences out of about 45 extracellular amino acids which could influence binding of AQP-4 antibodies--and the presence of antibodies to other neuronal proteins in sera from patients with neurological diseases may reduce the specificity. On the positive side, the sections allow binding of antibodies to both the intracellular and extracellular regions of AQP-4, which is advantageous as an initial screen to detect antibodies as disease markers, irrespective of whether they have the potential to be pathogenic.

Cell-Based Assay

This assay was first described as proof of the identification of AQP-4 as a target antigen in NMO-IgG positive sera (2) and subsequently adapted by others for routine use. (8-10) Serum is incubated with human embryonic kidney (HEK) cells that had been stably transfected with either human AQP-4 (M1) or vector alone for 1 hour. After washing with phosphate buffered saline, the cells are then incubated with fluorescein-conjugated goat anti-human IgG for 30 minutes. Positive sera are identified by the fluorescence at the cell surface (see Figure 2B for a cartoon of the method). Any positive samples can be titrated in doubling dilutions to ascertain the maximum dilution that gives positive results. This assay seems, on initial examination, to be more sensitive than IIF (91% vs 73% for IIF) with 100% specificity. However, all the sera tested initially (8,9) in this assay were from relapsing NMO patients, taken during active disease, the cohort in which the antibodies are most readily detected by the different methods. Only one other group (10) has reported on this method so far, and in this case they used transient transfection of the cells with AQP-4 into which enhanced green fluorescent protein (EGFP) had been incorporated (Figure 3B). The results showed 80% sensitivity and 100% specificity in an unselected cohort of NMO patients sampled during different stages of the disease. (10)


Radioimmunoprecipitation Assay

To produce a test that could be adapted for routine use, a radioimmunoprecipitation assay (RIPA) was developed. (11) Full length human AQP-4 (M1) was expressed in a cell-free in vitro transcription/translation system using rabbit reticulocyte lysate in the presence of [sup.35]S-methionine to produce [sup.35]S-methionine-AQP-4 in solution. (11) The final product was incubated with the sera and precipitated with protein-A beads. The radioactivity was counted and positivity determined by a ratio of sample counts to control counts (x 10). This method uses human AQP-4 which should increase the sensitivity compared with the IIF. However, the sensitivity was relatively low, perhaps because the concentration of AQP-4 in the incubation was low and the antigen may not be correctly folded (see Table 1 and Figure 2C for a cartoon of the method).


Fluoroimmunoprecipitation Assay

The success of the cell-based approach using AQP-4 tagged with EGFP, and work with other immunoassays from our laboratory (D Beeson and S Maxwell; unpublished results) suggested that the antibodies might be detected routinely using immunoprecipitation of EGFP-AQP-4. Both isoforms of AQP-4 with an N-terminal (intracellular) EGFP tag were transiently transfected into HEK 293 cells. The cells were solubilized in a triton-based buffer and the cell extract, containing tetramers of EGFP-AQP-4, was incubated with serum overnight. The complexes were precipitated with protein-A beads and the fluorescence read on a fluorescence plate reader. The results were positive in 76% of our NMO cohort (correlating highly with the cell-based assay [CBA] results) and in all patients sampled in relapse. (10) This assay is now being used for clinical diagnosis and has the advantage of being highly quantitative and suitable for serial estimations (Figure 3C).

Enzyme-Linked ImmunoSorbent Assay

Finally, an enzyme-linked immunosorbent assay (ELISA) has been developed to detect AQP-4 antibodies in patients' sera. (12) Here a histidine-tagged rat M23-AQP-4 was produced using a viral expression system in S2 cells. The cells were solubilized and the AQP-4 was purified on a nickel-column (nickel binds to the histidine tag). The purified protein was bound to nickel-coated ELISA plates. Patient serum (diluted 1:1000) was incubated in the plates, and binding was detected using peroxidase conjugated goat anti-human IgG and hydrogen peroxide hydrolysis. A sample was considered positive if its value was more than 3 SD's above the mean of 51 normal controls. The results are comparable with the IIF with 71% sensitivity and 98% specificity; however, it seems to be the most labour-intensive of the methods described, and is still not as sensitive or specific as the fluoroimmunoprecipitation assay (FIPA) or the CBA (Figure 4).

Use of the Assays in Patients at Risk of NMO

All of the antibody assays have been shown to be useful in identifying patients with other forms of NMO-spectrum disease. Initially, NMO-IgG was detected in 11/29 patients presenting with LETM, (3) and 6/9 of these patients were found to relapse within 2 years, whereas none of the 14 NMO-IgG negative LETM patients relapsed within 4 years of follow-up. Similarly, although NMO-IgG is less frequent in patients with recurrent ON, 6/12 NMO-Ig positive ON patients developed transverse myelitis within 12 months compared with only 1/15 NMO-Ig negative patients, and antibody presence predicted poor visual outcome. (13)


The antibody also helped to clarify the myelitis that can occur in association with Sjogren's syndrome or SLE. (3,4) In both diseases, myelopathy was previously assumed to be a complication of the primary disease rather than a separate disease entity. It now seems that longitudinally-extensive myelopathy in those diseases, very often associated with AQP-4 antibodies, is likely to represent the co-existence of NMO-spectrum disorder with the primary conditions.

Study Material and Comparison of Different Methods

As in all studies of this kind, many of the early publications are from unselected patients who may have been sampled at different times during their disease course, and after treatment with a wide variety of different immunotherapies. Moreover, it appears from examination of the data, that monophasic NMO and relapsing NMO may not be the same. The assays frequently report >80% positives in the predominantly female cohorts with relapsing NMO (a female:male ratio of nearly 6:1) (7,8,10-12) whereas only 20% of the sera from monophasic patients were positive, there was no sex bias and titres were generally lower. (7,10,11) Therefore, we need to take into account whether each serum was taken during relapse or remission, whether the patients had monophasic or recurrent disease, and whether the patient had already received treatment or not.

To provide a fair comparison, we summarized the results from relapsing NMO patients only (although it is not always clear whether these were undergoing relapse and untreated at the time of sampling). The results are shown graphically in Figure 4. The RIPA and ELISA are the least sensitive assays, and the IIF, CBA and FIPA show similar high sensitivity (between 83% and 91%) and good specificity, although the specificity of the IIF is lower (91%). When comparisons were available, all patients tested positive by IIF were also positive by FIPA or CBA, but the latter two methods detected additional positive patients. (9-11) We believe that the CBA will prove to be the most sensitive and specific assay, but the FIPA may be easier to adapt for use in diagnostic laboratories since the EFGP-AQP-4 protein could be freeze-dried for distribution.

AQP-4 Localization In Vivo and Potential Improvements to the Assays

The main target for the antibodies by IIF is the astrocytic foot processes that abut the abluminal surface of capillaries in the brain; they also bind along the subpial layer and at the glia limitans. (7,14) Freeze fracture images are available of astrocytic foot processes in rat brain and spinal cord, (15-17) which show a variety of regular arrays of intramembrane particles (IMP), believed to be AQP-4 tetramers, arranged into higher order particles called orthogonally arranged particles (OAPs) or square arrays. It has been shown that a variety of arrays from 2-100 IMPs are present in the astrocytic end foot, and that in these arrays the density of the antigen is very high.

An image of a human IgG1 antibody juxtaposed to the antigenic region of AQP-4 (depicted with EGFP attached to the N-termini of three monomers) demonstrates the relative sizes of the antibody and its antigen, and suggests that each tetramer may only be bound by one or two antibody molecules (see Figure 5), although the actual epitopes and their sites on the extracellular domain of the antigen are not yet known. It seems very likely, in analogy with acetylcholine receptor antibodies in myasthenia gravis (see ref 18), that AQP-4 antibodies are capable of divalent binding to adjacent AQP-4 molecules, and that reproducing the high density of the antigen found in vivo could improve sensitivity of the assays even further, as recently shown for myasthenia antibodies. (19) A system that presents the antigen in large high density arrays should allow the detection of even low affinity antibodies if they exist.

There are at least two forms of AQP-4 expressed in the astrocytes. (20) The M23 isoform tends to form large arrays in cells in vitro and is thought to form OAPs, whereas the M1 form exists as tetramers, but not large arrays. Although both isoforms can be expressed in HEK cells, there is no clear evidence of arrays forming in these cells, unlike in AQP-4 transfected Chinese hamster ovary (CHO) cells where square arrays have been demonstrated by freeze fracture. (16) It appears, therefore, that both the current CBA and the FIPA detect binding to simple tetramers. Both assays may be improved by providing OAPs as a target rather than the tetrameric IMPs. It would be worthwhile comparing an assay with M23 in the HEK and CHO cells. Moreover, although it is convenient and helpful to have AQP-4 tagged with EGFP for the CBA, its presence at the N-terminus may interfere with array formation and limit antigen density. For the FIPA, potentially larger soluble rafts may be produced by CHO cells (16) which could increase the sensitivity of this assay, depending on their stability in detergent. And finally, there are additional changes that could increase the signal to noise ratio, e.g. use of an enzymatic-like tag (luciferase), where the signal is amplified over time.


Over a period of 3 years, results of several assays for AQP-4 antibodies have been reported, demonstrating the clinical relevance of measuring this antibody. Although the assays are different in nature, the highly specific results from most and the improving specificity of the IIF (Table 1), clearly indicate that the AQP-4 antibody is specific for NMO spectrum diseases, that AQP-4 is the main, or only, target for the antibodies detected, and suggests that these antibodies are likely to be pathogenic (see ref 18). The assays perform particularly well in detecting antibody in relapsing NMO and Sjogren's syndrome with myelopathy, but monophasic NMO patients are less often positive. It may be that the assays need to be more sensitive to detect lower values in patients in remission or in the monophasic disease. The use of human AQP-4 M23 expressed in CHO cells rather than in HEK cells may be one way forward.


Conflicts of Interest

The authors' department receives payments and royalties for antibody tests.

Key Points

* AQP-4 is the main water channel in the CNS where it is expressed around cerebral microvessels, pia mater and Virchow-Robin spaces

* Antibodies specific for AQP-4 have been identified in NMO and related disorders that need to be distinguished from MS

* Five different antibody assays for their detection have now been reported

* All assays show good specificity for relapsing NMO but sensitivities differ

* A cell-based technique performs best, but immunoprecipitation methods may prove more suitable for routine diagnosis and serial estimations

Received: 8 August 2008

Accepted: 8 August 2008


(1.) Hiroaki Y, Tani K, Kamegawa A, et al. Implications of the aquaporin-4 structure on array formation and cell adhesion. J Mol Biol 2006; 355: 628-639.

(2.) Lennon VA, Kryzer TJ, Pittock SJ, et al. IgG marker of optic-spinal multiple sclerosis binds to the aquaporin-4 water channel. J Exp Med 2005; 202: 473-477.

(3.) Weinshenker BG, Wingerchuk DM, Vukusic S, et al. Neuromyelitis optica IgG predicts relapse after longitudinally extensive transverse myelitis. Ann Neurol 2006; 59: 566-569.

(4.) Wingerchuk DM, Lennon VA, Lucchinetti CF, et al. The spectrum of neuromyelitis optica. Lancet Neurol 2007; 6: 805-815.

(5.) Wingerchuk DM, Pittock SJ, Lucchinetti CF, et al. A secondary progressive clinical course is uncommon in neuromyelitis optica. Neurology 2007; 68: 603-605.

(6.) Pittock SJ, Lennon VA, de Seze J, et al. Neuromyelitis optica and non-organ specific autoimmunity. Arch Neurol 2008 65: 78-83.

(7.) Lennon VA, Wingerchuk DM, Kryzer TJ, et al. A serum autoantibody marker of neuromyelitis optica: distinction from multiple sclerosis. Lancet 2004; 364: 2106-2112.

(8.) Takahashi T, Fujihara K, Nakashima I, et al. Establishment of a new sensitive assay for anti-human aquaporin-4 antibody in neuromyelitis optica. Tohoku J Exp Med 2006; 210: 307-313.

(9.) Takahashi T, Fujihara K, Nakashima I, et al. Anti-aquaporin-4 antibody is involved in the pathogenesis of NMO: a study on antibody titre. Brain 2007; 130: 1235-1243.

(10.) Waters P, Jarius S, Littleton E, et al. Aquaporin-4 antibodies in neuromyelitis optica and longitudinally-extensive transverse myelitis. Arch Neurol 2007; 65: 913-919.

(11.) Paul F, Jarius S, Aktas O, et al. Antibody to aquaporin 4 in the diagnosis of neuromyelitis optica. PLoSMed 2007; 4: e133.

(12.) Hayakawa S, Mori M, Okuta A, et al. Neuromyelitis optica and anti-aquaporin-4 antibodies measured by an enzyme-linked immunosorbent assay. J Neuroimmunol 2008; 196: 181-187.

(13.) Matiello M, Lennon VA, Jacob A, et al. NMO-Ig predicts the outcome of recurrent optic neuritis. Neurology 2008: [in press].

(14.) Jarius S, Franciotta D, Bergamaschi R, et al. NMO-IgG in the diagnosis of neuromyelitis optica. Neurology 2007; 68: 1076-1077.

(15.) Furman CS, Gorelick-Feldman DA, Davidson KG, et al. Aquaporin-4 square array assembly: opposing actions of M1 and M23 isoforms. Proc Natl Acad Sci USA 2003; 100: 13609-13614.

(16.) Rash JE, Davidson KG, Yasumura T, Furman CS. Freeze-fracture and immunogold analysis of aquaporin-4 (AQP4) square arrays, with models of AQP4 lattice assembly. Neuroscience 2004; 129: 915-934.

(17.) Rash JE, Yasumura T, Hudson CS, et al. Direct immunogold labeling of aquaporin-4 in square arrays of astrocyte and ependymocyte plasma membranes in rat brain and spinal cord. Proc Natl Acad Sci USA 1998; 95: 11981-11986.

(18.) Jarius S, Paul F, Franciotta D, et al. Aquaporin-4 antibodies in neuromyelitis optica. Nat Clin Pract Neurol. 2008 4(4): 202-214.

(19.) Leite MI, et al. IgG1 antibodies to acetylcholine receptors in 'seronegative' myasthenia gravis. Brain 2008; 131: 1940-1952.

(20.) Humphreys W, Dalke A, Schulten K. VMD: visual molecular dynamics. J Mol Graph 1996; 14: 33-38.

(21.) Moe SE, Sorbo JG, Sogaard R, et al. New isoforms of rat Aquaporin-4. Genomics 2008; 91: 367-377.

(22.) Guex N, Peitsch MC. Swiss model and the SwissPdbViewer: An environment for comparative protein modeling. Electrophoresis 1997; 18: 2714-2723.

(23.) Clark, M. Antibody Engineering IgG Effector Mechanisms. Chemical Immunol 1997; 65: 88-110.

(24.) Marignier R, De Seze J, Vukusic S, et al. NMO-IgG and Devic's neuromyelitis optica: a French experience. Mult Scler. 2008; 14: 440-445.

(25.) Saiz A, Zuliani L, Blanco Y, et al; Spanish-Italian NMO Study Group. Revised diagnostic criteria for neuromyelitis optica (NMO). Application in a series of suspected patients. J Neurol. 2007; 254(9): 1233-127.

P Waters, A Vincent

Neurosciences Group, Weatherall Institute of Molecular Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK

Address for correspondence

Angela Vincent, Neurosciences Group, Weatherall Institute of Molecular Medicine, University of Oxford, John Radcliffe Hospital, Headington, Oxford OX3 9DS, UK Phone: +44 1865 222321; Fax: +44 1865 222402; E-mail:
Table 1: Sensitivity and specificity of methods used to detect
NMO-Ig/AQP-4 antibody in unselected cohorts. Results comparing
the relapsing NMO cohort from the original publications of each
method are shown in Figure 4

Method           Author              Year            NMO

IIF              Lennon (7)          2004            33/45
IIF              Lennon (7)          2004            6/11
IIF              Jarius (14)         2007            22/36
IIF              Marignier (24)      2008            14/26
IHC              Saiz (25)           2007            10/16
IIF              Waters (10)         2008            14/24
CBA              Takahashi (8,9)     2006, 2007      20/22
Modified CBA     Waters (10)         2008            20/25
RIPA             Paul (11)           2007            21/37
ELISA            Hayakawa (12)       2008            15/21
FIPA             Waters (10)         2008            19/25

Method           MS                  Sensitivity     Specificity

IIF              2/22                73%             91%
IIF              0/5                 58%             100%
IIF              1/80                61%             99%
IIF              5/52                54%             94%
IHC              0/127               63%             100%
IIF              0/38                58%             99%
CBA              0/53                91%             100%
Modified CBA     0/26                80%             100%
RIPA             4/144               57%             98%
ELISA            2/46                71%             98%
FIPA             0/38                76%             100%

NMO: Neuromyelitis optica; MS: multiple sclerosis.

Figure 4. Comparison of the sensitivity and specificity of
the different methods used to detect AQP-4 antibody or
NMO-Ig using the relapsing NMO cohort from the
original publications.

Replasing NMO only


          Sensitivity     Specificity

RIPA *        57               98
ELISA         71               98
IIF           86               91
FIPA          83              100
CBA           91              100

* No information was available on the patient
breakdown in this assay, which may lead to an
underestimation of the performance.

Note: Table made from bar graph.
Gale Copyright: Copyright 2008 Gale, Cengage Learning. All rights reserved.