**3. Anti-AQP4 autoantibodies in NMO**

168 Autoimmune Disorders – Current Concepts and Advances from Bedside to Mechanistic Insights

length human MOG (Lalive et al., 2006; Zhou et al., 2006). Although both studies used cell based assays, the frequency of anti-MOG antibodies within the MS disease course varied in both studies. Lalive and colleges reported increased titers of serum anti-MOG antibodies in patients with CIS, RR-MS and to a smaller extend in SP-MS, but not in healthy controls or PP-MS patients (Lalive et al., 2006). This is in contrast to a study of Zhou, observing the highest frequency of pathogenic autoantibodies to MOG in PP-MS patients using a flow cytometry-based assay. Moreover, Zhou and colleges demonstrated a pathogenic role of human anti-MOG antibodies *in vitro* and *in vivo* following injection into susceptible rat models (Zhou et al., 2006). Summarizing, the relevance of anti-MOG antibodies in MS remains controversial. Latest findings using a novel tetramer radioimmunoassay indicated the presence of conformation dependent anti-MOG antibodies is a subset of pediatric patients with acute disseminated encephalomyelitis (ADEM) and pediatric MS, but rarely in

ADEM is a rarely occurring inflammatory demyelinating disease of the CNS, brain and spinal cord, with an unknown relationship to MS. In patients with ADEM, acute or subacute multifocal large bilateral white matter lesions, frequently involving deep grey matter regions are accompanied by the occurrence of encephalopathy (Mikaeloff et al., 2004; Krupp et al., 2007). Although guidelines have been published to support the diagnosis of ADEM, diagnosis can be complicated, as exact diagnostic criteria are missing. Thus, the incidence remains to be investigated. Some publications suggest a prevalence rate of 0.8 per 100,000 affected patients per year (Leake et al., 2004). Whereas some reports describe no gender predisposition (Dale et al., 2000; Leake et al., 2004), most studies indicate a slight male preponderance in ADEM patients (Pavone et al., 2010). Although the majority of patients with ADEM follow a monophasic disease course, recently recurrent or multiphasic forms have been described with a lower incidence rate (Rust, 2000; Hynson et al., 2001). ADEM commonly occurs after a vaccination (post-vaccination encephalomyelitis) or infection (postinfection encephalomyelitis). In a study of Tenembaum, analyzing 84 pediatric ADEM patients, neurological disturbances occurred in 74% of patients following vaccination or infection (Tenembaum et al., 2002). ADEM is more often described in pediatric patients and juveniles (Leake et al., 2004), however, adult cases have also been reported (Schwarz et al., 2001). In contrast to the persistent disease course of MS, 57-89% of ADEM patients show complete recovery (Dale et al., 2000; Tenembaum et al., 2002). Furthermore, acute treatment with corticosteroids, immunoglobulins and plasma exchange often results in amelioration of ADEM patients, for which reason a biomarker is of high relevance in order to stratify MS and ADEM. Primarily at disease onset, ADEM can be misdiagnosed as CIS (Mikaeloff et al., 2007). The International Consensus criteria of 2007 can serve as guidelines for diagnosing CIS or ADEM (Krupp et al., 2007). Recently, a retrospective study was published analyzing the role of MRI in 28 children with MS and 20 ADEM patients (Callen et al., 2009). Herby, Callen et al. demonstrated a lower age of onset for ADEM patients compared to pediatric MS. This study invented new MRI diagnostic criteria to help differentiating RR-MS from monophasic ADEM at disease onset, yielding a high sensitivity (81%) and specificity (95%) (Callen et al., 2009). Contrary to MS which is typically associated with the development of new lesions, ADEM lesions usually resolve or show residual findings (Kesselring et al., 1990). Therefore, a follow-up MRI within a time period not shorter than 6 months is helpful for diagnosis (Kesselring et al., 1990). Analysis of CSF can support the diagnosis, as OCB are

adult onset MS (O'Connor et al., 2007).

**2.2 High titer anti-MOG antibodies in ADEM patients** 

NMO is a rare devastating inflammatory demyelinating disease of the CNS. In former times it was believed to be a severe variant of MS, the most common neurological disease in young adults. In contrast to MS, it has several unique features (Table 1). NMO is characterized by the occurrence of optic neuritis (ON) and longitudinally extensive transverse myelitis (LETM) extending over three or more vertebral segments (Wingerchuk et al., 1999; Cree, 2008), which can lead to blindness and paraplegia within several years of disease onset (Wingerchuk et al., 1999; Wingerchuk & Weinshenker, 2003). Furthermore, NMO commonly follows a more aggressive disease course compared to MS and has a high rate of morbidity and mortality in patients who receive no special treatment (Wingerchuk & Weinshenker, 2003). Especially at disease onset, the diagnosis can be complicated by a long lasting time interval between the occurrence of LETM and ON. Whereas OCB are detected in the CSF of approximately 90% of MS patients (Kabat et al., 1948), they are rarely or transiently present in patients with NMO (0-37%) (Wingerchuk et al., 1999). In addition, the diagnosis of NMO can be supported by the detection of CSF pleocytosis (>50 x106 white blood cell count /L) during acute relapses (Zaffaroni, 2004), which is not indicative for MS. Originally, NMO was described in 1894 by Eugene Devic and Gault as acute monophasic disorder with simultaneous occurrence of ON and LETM (Minagar et al., 2002). Due to a tremendous increase in the scientific interest for this disease, many aspects from the original view of NMO have changed. Nowadays, NMO is characterized as a mainly relapsing disease (80-90%), with a minority of patients suffering from a monophasic course (Wingerchuk et al., 2007; Sellner et al., 2010). Whereas NMO was initially described by a lack of brain MRI lesions, MS-atypical brain lesions are found in some NMO patients primarily at sites of high AQP4 expression (Pittock et al., 2006). However, a negative brain MRI at disease onset is not indicative for MS (Jarius et al., 2008b). The explosive rise in the field of NMO research was mainly due to the discovery of NMO-IgG autoantibodies, mostly IgG1,

Relevance of Autoantibodies for the Classification and Pathogenesis of Neurological Diseases 171

(Wingerchuk et al., 2007), with a median age of onset at around 40 years (Wingerchuk et al., 2007). Female patients were far more likely to develop the recurrent disease course (Wingerchuk, 2009), yet the gender had no impact on the type or severity of initial attacks, recovery from initial attacks, relapse frequency or disease related mortality (Wingerchuk, 2009). A genetic predisposition for NMO was recently indicated by some case reports, which are mainly based on studies of sibling pairs and a parent-child case (Mirsattari et al., 2001; Rivera & Cabrera, 2001; Braley & Mikol, 2007; Cabrera-Gomez et al., 2009). Furthermore, 12 pedigrees of NMO patients with a total number of 25 patients were recently analyzed by Mattiello and his group, resulting in 3% familial NMO cases in patients with clinical definite NMO (Matiello et al., 2010). This number might be larger when also including patients with high risk NMO, as the disease can have a heterogeneous presentation (Pellkofer et al., 2009).

With the advent of anti-AQP4 antibodies as biomarkers in NMOSD (Lennon et al., 2004; Lennon et al., 2005), various NMO antibody assays have been developed (Lennon et al., 2004; Lennon et al., 2005; Paul et al., 2007; Takahashi et al., 2007; Marignier et al., 2008; Waters & Vincent, 2008; Mader et al., 2010). The choice of assay is crucial for the identification of NMO IgG autoantibodies in serum and CSF samples of patients. The first assay describing the presence of NMO autoantibodies applied an indirect immunofluorescence (IF) assay with a composite substrate of adult mouse cerebellum sections (Lennon et al., 2004). This assay was described by the group of Vanda Lennon, achieving 58-73% sensitivity and 91-100% specificity for NMO. One year later the AQP4 water channel protein was detected as target antigen using human embryonic kidney cells (HEK) transfected with human AQP4 (Lennon et al., 2005). The establishment of cell based assays using transfected HEK cells resulted in an even higher sensitivity than the tissue based IF assays, resembling most likely the native conformation of the AQP4 protein (Takahashi et al., 2006; Takahashi et al., 2007; Mader et al., 2010). For this purpose, the cells were transfected with the AQP4 protein fused to a green fluorescence protein. After addition of the NMO samples, the bound anti-AQP4 antibodies were detected using a secondary antibody. Positive samples were visualized by a co-staining of NMO IgG (red) with the AQP4 expressing cells (green), as demonstrated in Figure 4 (Takahashi et al., 2006;

Fig. 4. Detection of an anti-AQP4 IgG seropositive patient with our live cell staining IF assay (Mader et al., 2010). Human anti-AQP4 IgG (red, B) in patient`s serum bind to the AQP4-

EmGFP transfected cells (green, A), resulting in co-localisation (merge, C).

**3.2 Anti-AQP4 antibodies as biomarkers for NMO-spectrum disorders** 

Mader et al., 2010).

in serum of NMO patients, but not in classical MS or any other controls (Lennon et al., 2004). This marvelous achievement is attributed to Vanda Lennon and her group, discovering one year later the AQP4 water channel protein as target antigen of NMO autoantibodies (Lennon et al., 2004; Lennon et al., 2005). This transmembrane channel protein constitutes an essential part of the blood brain barrier due to its localization in pericapillary endfeet processes and ependymal cells facing the ventricles (Figure 1) (Nielsen et al., 1997; Rash et al., 1998; Nicchia et al., 2004). The discovery and validation of this highly specific biomarker resulted in the incorporation of the anti-AQP4 antibody serostatus in the diagnostic criteria of NMO, achieving high sensitivity (99%) and specificity (90%) (Jarius et al., 2007; Wingerchuk et al., 2007) (Table 2).


Table 2. Revised diagnostic criteria of NMO (Wingerchuk et al., 2007). Definite NMO requires fulfillment of both absolute criteria and of two of the 3 supportive criteria.

Since the discovery of anti-AQP4 IgG, NMO is considered as a separate disease entity with an unknown relationship to MS. Thus, the detection of anti-AQP4 antibodies facilitates an early stratification of NMO and MS, which is highly important due to the different treatment recommendations. Compared to MS, NMO patients have a worse prognosis and require distinct treatment strategies due to the dominant humoral immunopathogenesis. Whereas immunomodulatory therapies are frequently applied for treating MS, immunosuppressive treatment is more promising for NMO (Sellner et al., 2010). Interferon beta (IFN-ß) and glatiramer acetate (GA) were shown to be beneficial in MS, whereas in NMO patients these drugs have an ineffective or even deleterious effect (Papeix et al., 2007; Warabi et al., 2007). Acute attacks are commonly treated by a combination of corticosteroids and immunosuppressive agents. Plasma exchange or treatment with rituximab can prevent NMO attacks in patients not responding to corticosteroids (Cree et al., 2005; Watanabe et al., 2007; Jacob et al., 2008; Bonnan et al., 2009).

#### **3.1 NMO epidemiology and genetic factors**

Limited reports are published concerning the epidemiology of NMO in different ethnic groups (Kira, 2006; Cabrera-Gomez et al., 2009; Collongues et al., 2010), and thus the incidence and prevalence of NMO remains unknown. Some studies indicate a prevalence of one per 100,000 patients (Cabre, 2009; Cabrera-Gomez et al., 2009), however these studies use different antibody assays. Presumably, a proportion of patients remains to be falsely diagnosed as severe variant of MS. Available data suggest a higher incidence in non-Caucasian countries, especially in Latin American, East Asian and African populations compared to Northern European countries (Osuntokun, 1971; Kira et al., 1996; Papais-Alvarenga et al., 2002). NMO occurs up to nine times more often in women than in men (Wingerchuk et al., 2007), with a median age of onset at around 40 years (Wingerchuk et al., 2007). Female patients were far more likely to develop the recurrent disease course (Wingerchuk, 2009), yet the gender had no impact on the type or severity of initial attacks, recovery from initial attacks, relapse frequency or disease related mortality (Wingerchuk, 2009). A genetic predisposition for NMO was recently indicated by some case reports, which are mainly based on studies of sibling pairs and a parent-child case (Mirsattari et al., 2001; Rivera & Cabrera, 2001; Braley & Mikol, 2007; Cabrera-Gomez et al., 2009). Furthermore, 12 pedigrees of NMO patients with a total number of 25 patients were recently analyzed by Mattiello and his group, resulting in 3% familial NMO cases in patients with clinical definite NMO (Matiello et al., 2010). This number might be larger when also including patients with high risk NMO, as the disease can have a heterogeneous presentation (Pellkofer et al., 2009).

### **3.2 Anti-AQP4 antibodies as biomarkers for NMO-spectrum disorders**

170 Autoimmune Disorders – Current Concepts and Advances from Bedside to Mechanistic Insights

in serum of NMO patients, but not in classical MS or any other controls (Lennon et al., 2004). This marvelous achievement is attributed to Vanda Lennon and her group, discovering one year later the AQP4 water channel protein as target antigen of NMO autoantibodies (Lennon et al., 2004; Lennon et al., 2005). This transmembrane channel protein constitutes an essential part of the blood brain barrier due to its localization in pericapillary endfeet processes and ependymal cells facing the ventricles (Figure 1) (Nielsen et al., 1997; Rash et al., 1998; Nicchia et al., 2004). The discovery and validation of this highly specific biomarker resulted in the incorporation of the anti-AQP4 antibody serostatus in the diagnostic criteria of NMO, achieving high sensitivity (99%) and specificity (90%) (Jarius et al., 2007;

Wingerchuk et al., 2007) (Table 2).

2007; Jacob et al., 2008; Bonnan et al., 2009).

**3.1 NMO epidemiology and genetic factors** 

2. Acute myelitis

**Absolute criteria Supportive criteria** 

1. Optic neuritis 1. Brain MRI atypical for MS

Table 2. Revised diagnostic criteria of NMO (Wingerchuk et al., 2007). Definite NMO requires fulfillment of both absolute criteria and of two of the 3 supportive criteria.

Since the discovery of anti-AQP4 IgG, NMO is considered as a separate disease entity with an unknown relationship to MS. Thus, the detection of anti-AQP4 antibodies facilitates an early stratification of NMO and MS, which is highly important due to the different treatment recommendations. Compared to MS, NMO patients have a worse prognosis and require distinct treatment strategies due to the dominant humoral immunopathogenesis. Whereas immunomodulatory therapies are frequently applied for treating MS, immunosuppressive treatment is more promising for NMO (Sellner et al., 2010). Interferon beta (IFN-ß) and glatiramer acetate (GA) were shown to be beneficial in MS, whereas in NMO patients these drugs have an ineffective or even deleterious effect (Papeix et al., 2007; Warabi et al., 2007). Acute attacks are commonly treated by a combination of corticosteroids and immunosuppressive agents. Plasma exchange or treatment with rituximab can prevent NMO attacks in patients not responding to corticosteroids (Cree et al., 2005; Watanabe et al.,

Limited reports are published concerning the epidemiology of NMO in different ethnic groups (Kira, 2006; Cabrera-Gomez et al., 2009; Collongues et al., 2010), and thus the incidence and prevalence of NMO remains unknown. Some studies indicate a prevalence of one per 100,000 patients (Cabre, 2009; Cabrera-Gomez et al., 2009), however these studies use different antibody assays. Presumably, a proportion of patients remains to be falsely diagnosed as severe variant of MS. Available data suggest a higher incidence in non-Caucasian countries, especially in Latin American, East Asian and African populations compared to Northern European countries (Osuntokun, 1971; Kira et al., 1996; Papais-Alvarenga et al., 2002). NMO occurs up to nine times more often in women than in men

2. Spinal cord MRI with contiguous T2 weighted signal abnormality extending over 3 or more vertebral segments 3. Anti-AQP4 IgG seropositive status

With the advent of anti-AQP4 antibodies as biomarkers in NMOSD (Lennon et al., 2004; Lennon et al., 2005), various NMO antibody assays have been developed (Lennon et al., 2004; Lennon et al., 2005; Paul et al., 2007; Takahashi et al., 2007; Marignier et al., 2008; Waters & Vincent, 2008; Mader et al., 2010). The choice of assay is crucial for the identification of NMO IgG autoantibodies in serum and CSF samples of patients. The first assay describing the presence of NMO autoantibodies applied an indirect immunofluorescence (IF) assay with a composite substrate of adult mouse cerebellum sections (Lennon et al., 2004). This assay was described by the group of Vanda Lennon, achieving 58-73% sensitivity and 91-100% specificity for NMO. One year later the AQP4 water channel protein was detected as target antigen using human embryonic kidney cells (HEK) transfected with human AQP4 (Lennon et al., 2005). The establishment of cell based assays using transfected HEK cells resulted in an even higher sensitivity than the tissue based IF assays, resembling most likely the native conformation of the AQP4 protein (Takahashi et al., 2006; Takahashi et al., 2007; Mader et al., 2010). For this purpose, the cells were transfected with the AQP4 protein fused to a green fluorescence protein. After addition of the NMO samples, the bound anti-AQP4 antibodies were detected using a secondary antibody. Positive samples were visualized by a co-staining of NMO IgG (red) with the AQP4 expressing cells (green), as demonstrated in Figure 4 (Takahashi et al., 2006; Mader et al., 2010).

Fig. 4. Detection of an anti-AQP4 IgG seropositive patient with our live cell staining IF assay (Mader et al., 2010). Human anti-AQP4 IgG (red, B) in patient`s serum bind to the AQP4- EmGFP transfected cells (green, A), resulting in co-localisation (merge, C).

Relevance of Autoantibodies for the Classification and Pathogenesis of Neurological Diseases 173

samples (unpublished results). For this reasons, one has to be careful when dealing with the term "seronegative NMO". Moreover, seronegative NMO might resemble another disease course with overlapping clinical features. Particularly, pediatric NMO can present itself with diverse clinical features, and therefore stratification from MS can be difficult especially at disease onset (Lotze et al., 2008). The diagnosis might further be supported by analyzing CSF of patients. Recently, glial fibrillar acidic protein (GFAP), a marker of astrocytic damage, was shown to be significantly elevated in the CSF of NMO patients compared to classical MS (Misu et al., 2009), however this increase was primarily detectable during relapse (Misu et al., 2009). In conclusion, the anti-AQP4 antibody serostatus should be repeatedly analyzed in NMOSD using highly sensitive and specific cell based assays. The absence of anti-AQP4 antibodies over a long time interval indicates a different disease pathomechanism compared to patients with "AQP4 autoimmune channelopathy". Consequently ongoing research should focus on the discovery of new biomarkers for anti-

Apart from clinical definite NMO, anti-AQP4 IgG antibodies are frequently detected in limited forms of NMO (Wingerchuk et al., 2007). These patients do not fulfill the complete diagnostic criteria of NMO, but harbor a high risk of developing clinically definite NMO (Pittock et al., 2008; Mader et al., 2010). Therefore, NMO and high risk NMO patients represent the group of NMOSD, suffering either from monophasic bilateral or recurrent ON or LETM (idiopathic, isolated or recurrent) (Wingerchuk et al., 2007). Currently NMO IgG positive patients with recurrent ON were shown to have a poor visual outcome and were more likely to develop NMO in a longitudinal study (Matiello et al., 2008). Anti-AQP4 IgG seropositivity predicted a relapse in patients with a first episode of LETM event extending over three or more vertebral segments (Weinshenker et al., 2006). In 50% of these anti-AQP4 IgG seropositive LETM patients either ON occurred or LETM relapsed within half a year (Weinshenker et al., 2006). Furthermore, anti-AQP4 antibodies have been frequently detected in systemic autoimmune disorders presenting themselves with ON or LETM, such as neuropsychiatric systemic lupus erythematosus (SLE), Sjogren's syndrome, myasthenia gravis or thyroiditis (Wingerchuk et al., 2007). However, anti-AQP4 antibodies were detected exclusively in systemic autoimmune disorders in combination with NMO or High Risk NMO symptoms. The presence of anti-AQP4 antibodies could indicate a coexistence of systemic autoimmune disorders with NMO (Pittock et al., 2008; Wandinger et al., 2010), rather than an epiphenomenon (Pittock et al., 2008; Wandinger et al., 2010), yet their

The role of serum anti-AQP4 antibody titers remains controversially described. Takahashi and his group showed an involvement of AQP4-IgG antibody titers in disease pathogenesis, detecting a correlation with spinal cord lesion length (Takahashi et al., 2006). This study analyzed 148 serum samples of Japanese patients including 35 patients with NMO-spectrum disorders and demonstrated elevated AQP4-IgG titer levels in patients with permanent complete blindness, LETM and extensive or large cerebral lesions (Takahashi et al., 2006). In addition, a longitudinal study of eight NMO-IgG

AQP4 seronegative patients with NMO.

**3.4 NMO-spectrum disorders** 

relationship remains unidentified.

**3.5 Serum titer levels of anti-AQP4 IgG antibodies** 

This assay has the advantage of determining titer values of NMO antibody positive patients by serial dilutions of serum samples until loss of signal. However, the relevance of these titer levels remains controversial. As the AQP4 transmembrane protein is either expressed as full length M1 or as 23 amino acid shorter M23 AQP4 (Figure 5) (Neely et al., 1999; Furman et al., 2003), many studies lack the information regarding the usage of the AQP4 isoform. Recently, our group demonstrated that anti-AQP4 antibodies primarily target the shorter M23 AQP4 isoform, whereas antibodies to full length AQP4 were developed with increasing disease duration and number of relapses (Mader et al., 2010). For this purpose we used a live cell staining IF assay with transiently transfected HEK cells, resulting in 97% sensitivity for NMO and 65% for high risk NMO, with a specificity of 100% compared to controls (Mader et al., 2010). Our assay showed different staining patterns for M1 and M23 AQP4 transfected cells (Mader et al., 2010). In contrast to M1 AQP4, M23 AQP4 forms orthogonal arrays of particles (Figure 5 B), which are currently believed to be potential targets of antibody binding (Nicchia et al., 2009). Consequently the NMO IF assay yields highest sensitivity when using cell-based assay with M23 AQP4 transfected cells.

Fig. 5. Structure (A) and expression pattern (B) of the M1 and M23 AQP4 isoforms.

#### **3.3 "Anti-AQP4 seronegative NMO"**

The terminus "anti-AQP4 seronegative NMO" should be handled with care, as several factors contribute to the antibody serostatus. A broad range of antibody assays is available resulting in diverse sensitivity and specificity. Consequently, the percentage of seronegative NMO patients is fluctuating depending on the methodology approach. Cell-based assays using the M1 or M23 AQP4 isoform have an impact on the number of seronegative NMO patients. A negative antibody status might be credited to an administered therapy prior to testing. A depletion of antibodies below a detectable threshold could explain a negative serostatus. Although, anti-AQP4 antibodies have been detected up to ten years before disease onset (Nishiyama et al., 2009), we have analyzed a small number of patients who were initially negative for NMO IgG and turned out to be low titer positive in longitudinal samples (unpublished results). For this reasons, one has to be careful when dealing with the term "seronegative NMO". Moreover, seronegative NMO might resemble another disease course with overlapping clinical features. Particularly, pediatric NMO can present itself with diverse clinical features, and therefore stratification from MS can be difficult especially at disease onset (Lotze et al., 2008). The diagnosis might further be supported by analyzing CSF of patients. Recently, glial fibrillar acidic protein (GFAP), a marker of astrocytic damage, was shown to be significantly elevated in the CSF of NMO patients compared to classical MS (Misu et al., 2009), however this increase was primarily detectable during relapse (Misu et al., 2009). In conclusion, the anti-AQP4 antibody serostatus should be repeatedly analyzed in NMOSD using highly sensitive and specific cell based assays. The absence of anti-AQP4 antibodies over a long time interval indicates a different disease pathomechanism compared to patients with "AQP4 autoimmune channelopathy". Consequently ongoing research should focus on the discovery of new biomarkers for anti-AQP4 seronegative patients with NMO.

#### **3.4 NMO-spectrum disorders**

172 Autoimmune Disorders – Current Concepts and Advances from Bedside to Mechanistic Insights

This assay has the advantage of determining titer values of NMO antibody positive patients by serial dilutions of serum samples until loss of signal. However, the relevance of these titer levels remains controversial. As the AQP4 transmembrane protein is either expressed as full length M1 or as 23 amino acid shorter M23 AQP4 (Figure 5) (Neely et al., 1999; Furman et al., 2003), many studies lack the information regarding the usage of the AQP4 isoform. Recently, our group demonstrated that anti-AQP4 antibodies primarily target the shorter M23 AQP4 isoform, whereas antibodies to full length AQP4 were developed with increasing disease duration and number of relapses (Mader et al., 2010). For this purpose we used a live cell staining IF assay with transiently transfected HEK cells, resulting in 97% sensitivity for NMO and 65% for high risk NMO, with a specificity of 100% compared to controls (Mader et al., 2010). Our assay showed different staining patterns for M1 and M23 AQP4 transfected cells (Mader et al., 2010). In contrast to M1 AQP4, M23 AQP4 forms orthogonal arrays of particles (Figure 5 B), which are currently believed to be potential targets of antibody binding (Nicchia et al., 2009). Consequently the NMO IF assay yields

highest sensitivity when using cell-based assay with M23 AQP4 transfected cells.

Fig. 5. Structure (A) and expression pattern (B) of the M1 and M23 AQP4 isoforms.

The terminus "anti-AQP4 seronegative NMO" should be handled with care, as several factors contribute to the antibody serostatus. A broad range of antibody assays is available resulting in diverse sensitivity and specificity. Consequently, the percentage of seronegative NMO patients is fluctuating depending on the methodology approach. Cell-based assays using the M1 or M23 AQP4 isoform have an impact on the number of seronegative NMO patients. A negative antibody status might be credited to an administered therapy prior to testing. A depletion of antibodies below a detectable threshold could explain a negative serostatus. Although, anti-AQP4 antibodies have been detected up to ten years before disease onset (Nishiyama et al., 2009), we have analyzed a small number of patients who were initially negative for NMO IgG and turned out to be low titer positive in longitudinal

**3.3 "Anti-AQP4 seronegative NMO"** 

Apart from clinical definite NMO, anti-AQP4 IgG antibodies are frequently detected in limited forms of NMO (Wingerchuk et al., 2007). These patients do not fulfill the complete diagnostic criteria of NMO, but harbor a high risk of developing clinically definite NMO (Pittock et al., 2008; Mader et al., 2010). Therefore, NMO and high risk NMO patients represent the group of NMOSD, suffering either from monophasic bilateral or recurrent ON or LETM (idiopathic, isolated or recurrent) (Wingerchuk et al., 2007). Currently NMO IgG positive patients with recurrent ON were shown to have a poor visual outcome and were more likely to develop NMO in a longitudinal study (Matiello et al., 2008). Anti-AQP4 IgG seropositivity predicted a relapse in patients with a first episode of LETM event extending over three or more vertebral segments (Weinshenker et al., 2006). In 50% of these anti-AQP4 IgG seropositive LETM patients either ON occurred or LETM relapsed within half a year (Weinshenker et al., 2006). Furthermore, anti-AQP4 antibodies have been frequently detected in systemic autoimmune disorders presenting themselves with ON or LETM, such as neuropsychiatric systemic lupus erythematosus (SLE), Sjogren's syndrome, myasthenia gravis or thyroiditis (Wingerchuk et al., 2007). However, anti-AQP4 antibodies were detected exclusively in systemic autoimmune disorders in combination with NMO or High Risk NMO symptoms. The presence of anti-AQP4 antibodies could indicate a coexistence of systemic autoimmune disorders with NMO (Pittock et al., 2008; Wandinger et al., 2010), rather than an epiphenomenon (Pittock et al., 2008; Wandinger et al., 2010), yet their relationship remains unidentified.

#### **3.5 Serum titer levels of anti-AQP4 IgG antibodies**

The role of serum anti-AQP4 antibody titers remains controversially described. Takahashi and his group showed an involvement of AQP4-IgG antibody titers in disease pathogenesis, detecting a correlation with spinal cord lesion length (Takahashi et al., 2006). This study analyzed 148 serum samples of Japanese patients including 35 patients with NMO-spectrum disorders and demonstrated elevated AQP4-IgG titer levels in patients with permanent complete blindness, LETM and extensive or large cerebral lesions (Takahashi et al., 2006). In addition, a longitudinal study of eight NMO-IgG

Relevance of Autoantibodies for the Classification and Pathogenesis of Neurological Diseases 175

In order to address the pathogenic relevance of anti-AQP4 antibodies, several *in vitro* and *in vivo* studies have been performed so far*.* Tissue sections showed the distribution of anti-AQP4 antibodies and products of complement activation surrounding hyalinized blood vessels in a rosette-like pattern (Lucchinetti et al., 2002). This deposition of NMO antibodies and complement on astrocytes at the glia limitans was accompanied by a loss of the AQP4 water channel protein (Lennon et al., 2004; Roemer et al., 2007). The cytotoxic effect of anti-AQP4 antibodies has been demonstrated in several studies so far (Jarius et al., 2008b; Hinson et al., 2009; Sabater et al., 2009; Kinoshita et al., 2010). The binding of the anti-AQP4 antibody led to an activation of the classical complement cascade, resulting in lysis of NMO antibody opsonized and AQP4-transfected cells and astrocytes (Jarius et al., 2008b). The pathogenic role of anti-AQP4 antibodies has been further supported by several *in vivo* studies using rat and mouse models (Bennett et al., 2009; Bradl et al., 2009; Kinoshita et al., 2009; Saadoun et al., 2010). Therefore, NMO-IgG antibodies were purified from the plasma exchange material of AQP4-IgG positive and negative NMO patients, MS patients and control subjects and then injected into animal models (Bradl et al., 2009). Three studies demonstrated the formation of NMO-like lesions in Lewis rats following injection of isolated human NMO-IgG in the presence of acute T cell mediated CNS inflammation (Bennett et al., 2009; Bradl et al., 2009; Kinoshita et al., 2009). In contrast, Saadoun proved the formation of NMO like lesions after injection of NMO-IgG into mouse brain only in the presence of complement, thus by-passing the damage of the blood brain barrier (Saadoun et al., 2010). In order to confirm that the NMO like pathology in the animals was due to the anti-AQP4 IgG and not based on other antibodies in the plasma exchange material, pre-absorption experiments using cells expressing AQP4 were performed (Bradl et al., 2009). This preabsorption experiment resulted in a massive decrease of lesion size and was associated with less astrocytic damage, confirming the pathogenicity of anti-AQP4 autoantibodies (Bradl et al., 2009). As anti-AQP4 antibodies are not sufficient to induce NMO like lesions without support of T cells (Bradl et al., 2009) and/or complement (Saadoun et al., 2010), the role of T cell mediated immune responses against AQP4 is currently an issue of interest (Nelson et al., 2010; Kalluri et al., 2011; Pohl et al., 2011). Supporting evidence for a possible pathogenic role of T cells comes from observations showing no formation of NMO like lesions in immature rats after injection of anti-AQP4 autoantibodies, although these animals posses a leaky blood brain barrier (Bradl et al., 2009). Latest findings by Pohl et al. showed that AQP4 specific T cells are capable of inducing brain inflammation mainly in astrocytic glia limitans and therefore enable an entry of anti-AQP4 autoantibodies (Pohl et al., 2011). As anti-AQP4 antibodies are detectable more than ten years before disease onset (Nishiyama et al., 2009), the time point when these antibodies lead to NMO symptoms remains unresolved. It is tempting to speculate that anti-AQP4 antibodies are not harmful if they circulate peripherally and as long as they are excluded from the blood brain barrier. Whether a large amount of anti-AQP4 autoantibodies is necessary for the patients to develop symptoms remains to be investigated. Having access to the blood brain barrier, anti-AQP4 autoantibodies bind to their target antigen and result in complement activation. This leads to vascular hyalinization, necrosis, demyelination and axonal injury (Wingerchuk et al., 2007). The titer levels seem to have an impact on the disease pathogenesis as antibodies in the CSF are only detectable in high titer seropositive patients (Jarius et al., 2010b; Dujmovic et al., 2011). Latest findings indicated an influx of systemically produced anti-AQP4

**3.8 Pathogenic role of anti-AQP4 antibodies and T cells in NMO** 

positive patients reported a correlation of serum anti-AQP4 Ig with clinical disease activity (Jarius et al., 2008a), demonstrating a threefold intra-individual increase of AQP4 IgG titers during relapse, which was not accompanied by other serum antibodies (Takahashi et al., 2006). Some papers suggest an effect of treatment on antibody titers, showing a reduction of NMO antibody titer levels after immunosuppressive treatment (Takahashi et al., 2006; Jarius et al., 2008a). Recently, an increase of anti-AQP4 antibody titers was described in one NMO patient following immunomodulatory treatment with IFN-ß (Palace et al., 2010). Applying conventional immunosuppressive therapy, the antibody titers decreased again in this patient (Palace et al., 2010), high lightening the importance of an early stratification of MS and NMO.

### **3.6 AQP4-IgM antibodies**

Although, IgM antibodies binding to AQP4 were described at NMO lesion sites (Lucchinetti et al., 2002; Roemer et al., 2007), their role in the disease course remains unresolved. Most studies investigating autoantibodies against AQP4 refer to IgG antibodies. We addressed this issue in a recent study analyzing IgG and IgM antibodies directed to M23 AQP4 in serum of patients with NMO-spectrum disorders and in other disease groups using a live cell staining IF assay (Mader et al., 2010). In contrast to NMO IgG, which was exclusively detected in 97% of NMO patients and 65% of suspected NMO, M23 IgM antibodies were elevated in NMO (27%) and high risk NMO (12%). However, IgM antibodies to M23 AQP4 were additionally present in NMO IgG seronegative patients with isolated myelitis, MS (4%) and OND (4%). Furthermore, titer levels were much lower for IgM than for IgG AQP4. Antibodies of subtype IgM that bind to full length AQP4 were present in 10% of NMO and 8% of High Risk NMO, but not in any controls (Mader et al., 2010). This is in accordance with a study of Jarius et al. detecting IgM antibodies in almost 10% of NMO patients (4 /42) but not in any controls (Jarius et al., 2010a). Larger studies are warranted to further analyze anti-AQP4 IgG and IgM double positive patients. As anti-AQP4 IgM antibodies are more potent to activate the complement cascade, it would be tempting to further investigate the clinical parameters of anti-AQP4 IgM positive patients with NMOSD.

#### **3.7 Cerebrospinal fluid anti-AQP4 antibodies**

Several studies are available concerning anti-AQP4 antibodies in serum samples of patients with NMO-spectrum disorders. In contrast, few studies focused on the presence and relevance of NMO antibodies in the CSF (Takahashi et al., 2007; Klawiter et al., 2009; Jarius et al., 2010b; Dujmovic et al., 2011). Klawiter and colleges reported the presence of CSF anti-AQP4 antibodies in three seronegative NMO patients (Klawiter et al., 2009), which could not be reproduced in other publications. Recently, Jarius et al., detected CSF anti-AQP4 antibodies in NMO-IgG seropositive patients exceeding anti-AQP4 antibody serum titers ≥ 1:250, but not in anti-AQP4 antibody negative patients (Jarius et al., 2010b). In cooperation with Dujmovic we analyzed the temporal dynamics of CSF anti-AQP4 antibodies in 12 patients with NMOSD (Dujmovic et al., 2011). Thereby, we could show that longitudinal CSF anti-AQP4 IgG correlated with clinical parameters. CSF AQP4-IgG were present in patients with high serum titers and correlated with spinal MRI lesion length and CSF parameters. Moreover, clinical improvement was associated with a decrease in CSF, but not serum, anti-AQP4 IgG titers. Summarizing, CSF AQP4-IgG were associated with clinical activity and neuroinflammation (Dujmovic et al., 2011).

#### **3.8 Pathogenic role of anti-AQP4 antibodies and T cells in NMO**

174 Autoimmune Disorders – Current Concepts and Advances from Bedside to Mechanistic Insights

positive patients reported a correlation of serum anti-AQP4 Ig with clinical disease activity (Jarius et al., 2008a), demonstrating a threefold intra-individual increase of AQP4 IgG titers during relapse, which was not accompanied by other serum antibodies (Takahashi et al., 2006). Some papers suggest an effect of treatment on antibody titers, showing a reduction of NMO antibody titer levels after immunosuppressive treatment (Takahashi et al., 2006; Jarius et al., 2008a). Recently, an increase of anti-AQP4 antibody titers was described in one NMO patient following immunomodulatory treatment with IFN-ß (Palace et al., 2010). Applying conventional immunosuppressive therapy, the antibody titers decreased again in this patient (Palace et al., 2010), high lightening the

Although, IgM antibodies binding to AQP4 were described at NMO lesion sites (Lucchinetti et al., 2002; Roemer et al., 2007), their role in the disease course remains unresolved. Most studies investigating autoantibodies against AQP4 refer to IgG antibodies. We addressed this issue in a recent study analyzing IgG and IgM antibodies directed to M23 AQP4 in serum of patients with NMO-spectrum disorders and in other disease groups using a live cell staining IF assay (Mader et al., 2010). In contrast to NMO IgG, which was exclusively detected in 97% of NMO patients and 65% of suspected NMO, M23 IgM antibodies were elevated in NMO (27%) and high risk NMO (12%). However, IgM antibodies to M23 AQP4 were additionally present in NMO IgG seronegative patients with isolated myelitis, MS (4%) and OND (4%). Furthermore, titer levels were much lower for IgM than for IgG AQP4. Antibodies of subtype IgM that bind to full length AQP4 were present in 10% of NMO and 8% of High Risk NMO, but not in any controls (Mader et al., 2010). This is in accordance with a study of Jarius et al. detecting IgM antibodies in almost 10% of NMO patients (4 /42) but not in any controls (Jarius et al., 2010a). Larger studies are warranted to further analyze anti-AQP4 IgG and IgM double positive patients. As anti-AQP4 IgM antibodies are more potent to activate the complement cascade, it would be tempting to further investigate the

Several studies are available concerning anti-AQP4 antibodies in serum samples of patients with NMO-spectrum disorders. In contrast, few studies focused on the presence and relevance of NMO antibodies in the CSF (Takahashi et al., 2007; Klawiter et al., 2009; Jarius et al., 2010b; Dujmovic et al., 2011). Klawiter and colleges reported the presence of CSF anti-AQP4 antibodies in three seronegative NMO patients (Klawiter et al., 2009), which could not be reproduced in other publications. Recently, Jarius et al., detected CSF anti-AQP4 antibodies in NMO-IgG seropositive patients exceeding anti-AQP4 antibody serum titers ≥ 1:250, but not in anti-AQP4 antibody negative patients (Jarius et al., 2010b). In cooperation with Dujmovic we analyzed the temporal dynamics of CSF anti-AQP4 antibodies in 12 patients with NMOSD (Dujmovic et al., 2011). Thereby, we could show that longitudinal CSF anti-AQP4 IgG correlated with clinical parameters. CSF AQP4-IgG were present in patients with high serum titers and correlated with spinal MRI lesion length and CSF parameters. Moreover, clinical improvement was associated with a decrease in CSF, but not serum, anti-AQP4 IgG titers. Summarizing, CSF AQP4-IgG were associated with clinical

importance of an early stratification of MS and NMO.

clinical parameters of anti-AQP4 IgM positive patients with NMOSD.

**3.7 Cerebrospinal fluid anti-AQP4 antibodies** 

activity and neuroinflammation (Dujmovic et al., 2011).

**3.6 AQP4-IgM antibodies** 

In order to address the pathogenic relevance of anti-AQP4 antibodies, several *in vitro* and *in vivo* studies have been performed so far*.* Tissue sections showed the distribution of anti-AQP4 antibodies and products of complement activation surrounding hyalinized blood vessels in a rosette-like pattern (Lucchinetti et al., 2002). This deposition of NMO antibodies and complement on astrocytes at the glia limitans was accompanied by a loss of the AQP4 water channel protein (Lennon et al., 2004; Roemer et al., 2007). The cytotoxic effect of anti-AQP4 antibodies has been demonstrated in several studies so far (Jarius et al., 2008b; Hinson et al., 2009; Sabater et al., 2009; Kinoshita et al., 2010). The binding of the anti-AQP4 antibody led to an activation of the classical complement cascade, resulting in lysis of NMO antibody opsonized and AQP4-transfected cells and astrocytes (Jarius et al., 2008b). The pathogenic role of anti-AQP4 antibodies has been further supported by several *in vivo* studies using rat and mouse models (Bennett et al., 2009; Bradl et al., 2009; Kinoshita et al., 2009; Saadoun et al., 2010). Therefore, NMO-IgG antibodies were purified from the plasma exchange material of AQP4-IgG positive and negative NMO patients, MS patients and control subjects and then injected into animal models (Bradl et al., 2009). Three studies demonstrated the formation of NMO-like lesions in Lewis rats following injection of isolated human NMO-IgG in the presence of acute T cell mediated CNS inflammation (Bennett et al., 2009; Bradl et al., 2009; Kinoshita et al., 2009). In contrast, Saadoun proved the formation of NMO like lesions after injection of NMO-IgG into mouse brain only in the presence of complement, thus by-passing the damage of the blood brain barrier (Saadoun et al., 2010). In order to confirm that the NMO like pathology in the animals was due to the anti-AQP4 IgG and not based on other antibodies in the plasma exchange material, pre-absorption experiments using cells expressing AQP4 were performed (Bradl et al., 2009). This preabsorption experiment resulted in a massive decrease of lesion size and was associated with less astrocytic damage, confirming the pathogenicity of anti-AQP4 autoantibodies (Bradl et al., 2009). As anti-AQP4 antibodies are not sufficient to induce NMO like lesions without support of T cells (Bradl et al., 2009) and/or complement (Saadoun et al., 2010), the role of T cell mediated immune responses against AQP4 is currently an issue of interest (Nelson et al., 2010; Kalluri et al., 2011; Pohl et al., 2011). Supporting evidence for a possible pathogenic role of T cells comes from observations showing no formation of NMO like lesions in immature rats after injection of anti-AQP4 autoantibodies, although these animals posses a leaky blood brain barrier (Bradl et al., 2009). Latest findings by Pohl et al. showed that AQP4 specific T cells are capable of inducing brain inflammation mainly in astrocytic glia limitans and therefore enable an entry of anti-AQP4 autoantibodies (Pohl et al., 2011). As anti-AQP4 antibodies are detectable more than ten years before disease onset (Nishiyama et al., 2009), the time point when these antibodies lead to NMO symptoms remains unresolved. It is tempting to speculate that anti-AQP4 antibodies are not harmful if they circulate peripherally and as long as they are excluded from the blood brain barrier. Whether a large amount of anti-AQP4 autoantibodies is necessary for the patients to develop symptoms remains to be investigated. Having access to the blood brain barrier, anti-AQP4 autoantibodies bind to their target antigen and result in complement activation. This leads to vascular hyalinization, necrosis, demyelination and axonal injury (Wingerchuk et al., 2007). The titer levels seem to have an impact on the disease pathogenesis as antibodies in the CSF are only detectable in high titer seropositive patients (Jarius et al., 2010b; Dujmovic et al., 2011). Latest findings indicated an influx of systemically produced anti-AQP4

Relevance of Autoantibodies for the Classification and Pathogenesis of Neurological Diseases 177

The characteristic symptoms of anti-NMDA-R positive patients are of prominent psychiatric and behavioral nature, including rapid memory loss, seizures, abnormal movements (dyskinesias), hypoventilation and autonomic instability. This disease usually progresses from initial neuropsychiatric symptoms into a state of unresponsiveness with catatonic features, commonly associated with abnormal movements, and autonomic- and breathing instability. Additionally, most patients show prodomal symptoms such as headache, fever, nausea, vomiting, diarrhea or upper respiratory-tract symptoms. Brain MRI data show no or only minor changes which usually occur transiently despite severity of symptoms. Concerning CSF parameters, 60% of patients show OCB and mild lymphocyte pleocytosis (Dalmau et al., 2011). Interestingly, intrathecal NMDA-R antibody synthesis was observed in a majority of patients and CSF titer levels were more likely to correlate with clinical severity, compared to serum titers (Dalmau et al., 2008; Dale et al., 2009; Florance et al., 2009;

Although the disease can be lethal in some rare cases and despite the severity of the symptoms, more than 70% of patients recover after treatment and less than 30% of patients show incomplete recovery with memory, cognitive and motor deficits. Treatment options include immunotherapy (corticosteroids, intravenous immunoglobulin or plasma exchange) and/or tumor removal with the aim to reduce anti-NMDA-R autoantibody levels. Recent studies showed that antibodies to the NMDA-R were predominantly of the IgG1 subclass and are able to activate complement on NMDA-R expressing human embryonic kidney cells (Irani et al., 2010). However, the role of complement activation remains controversial as other findings indicated a complement-independent mode of action. Several studies have addressed the issue regarding the binding site of the autoantibodies and possible functional consequences on the targeted NMDA-R. Dalmau et al. described the NR1 isoform or NR1/NR2 heterodimers of the NMDA-R as recognition site of anti-NMDA-R antibodies (Dalmau et al., 2011). Providing further insight into the mode of action, *in vitro* and *in vivo* studies nicely demonstrated that antibodies from patients with anti-NMDA-R encephalitis caused a rapid and reversible loss of surface NMDA-R by antibody-mediated capping and internalization, resulting in abrogation of NMDA-R-mediated synaptic function (Dalmau et

Thus, similar to the role of anti-AQP4 IgG antibodies in NMO, anti-NMDA-R antibodies helped to define a new clinical syndrome, anti-NMDA receptor encephalitis (Dalmau et al.,

Several reports confirmed the presence of autoantibodies to NMDA-R, particularly the NR2 isoform, in the majority of patients with neuropsychiatric SLE (DeGiorgio et al., 2001; Emmer et al., 2006; Hanly et al., 2006; Kowal et al., 2006; Lapteva et al., 2006; Arinuma et al., 2008; Fragoso-Loyo et al., 2008). These autoantibodies were not only detected in serum, but also in the CSF and brain parenchyma of some SLE patients. Furthermore, CSF titers correlate with neuropsychiatric symptoms. In SLE, anti-NMDA-R antibodies were demonstrated to bind to a small peptide (DWEYS) present in the extracellular, aminoterminal domain of NR2A and NR2B subunits (DeGiorgio et al., 2001; Gielen et al., 2009). Injection of murine or human monoclonal antibodies against this peptide into the hippocampus and cerebral cortex of mice resulted in local loss of neurons and induced activation of caspase-3 in cultured human and murine neurons (DeGiorgio et al., 2001; Kowal et al., 2006; Gielen et al., 2009). Furthermore, several experimental studies in mice

**4.2 Anti-NMDA receptor antibodies in neuropsychiatric SLE** 

Irani et al., 2010).

al., 2008; Hughes et al., 2010).

2011).

antibodies through the area postrema (Popescu et al., 2011). This was supported by findings of patients suffering from intractable vomiting and nausea as initial symptoms of NMO (Popescu et al., 2011).
