Monozygotic twins discordant for epilepsy differ in the levels of ...

Autoimmunity, March 2005; 38(2): 139–150

Monozygotic twins discordant for epilepsy differ in the levels of potentially pathogenic autoantibodies and cytokines YONATAN GANOR1,†, MICHAEL FREILINGER2,†, OLIVIER DULAC3, & MIA LEVITE1,4 1

Department of Neurobiology, The Weizmann Institute of Science, Rehovot, Israel, 2Department of Pediatrics, University Hospital of Vienna, Vienna, Austria, 3Hopital Necker Enfants Malades, Paris, France, and 4Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel (Received 24 August 2004; accepted 14 February 2005)

Abstract Can autoantibodies (Ab’s) and cytokines play a role in epilepsy? Monozygotic twins discordant for epilepsy (most probably Rasmussen’s encephalitis (RE)), compared to 49 neurologically intact controls, were both found to contain in their serum (at the time of epilepsy diagnosis) significantly elevated levels of specific Ab’s against peptide B (amino acids 372 – 395) of the ionotropic glutamate receptor of AMPA subtype 3 (i.e. GluR3B peptide). Interestingly, both twins also had clinically elevated levels of Ab’s to double-stranded (ds) DNA, glutamic acid decarboxylase, nuclear antigens, b2-glycoprotein I and cardiolipin, as in “classical” autoimmune diseases. Both twins also had significantly elevated levels of IFNg, TNFa, IL-4 and IL-10 in the serum, compared to the controls. Comparing the twins revealed that the epileptic twin had significantly higher levels of five of the above anti-self Ab’s, but significantly lower levels of all four cytokines compared to her healthy sister. Importantly, the epileptic twin, alike three other RE patients tested herein, contained elevated levels of Ab’s to GluR3B and dsDNA also in cerebrospinal fluid (CSF) (unavailable of the healthy twin). Our results suggest that the various autoimmune Ab’s studied herein, all of which are known already to have a potential to be pathogenic in the nervous system and/or peripheral organs, may play a role in some types of epilepsy. The titer of such Ab’s and of key cytokines may be crucial for either facilitating or arresting the development of epilepsy. Our findings also show that antiGluR3B Ab’s in serum are not necessarily detrimental (their presence in the CSF may be more dangerous), and that they are not a mere side effect of already existing epilepsy, as they were found herein in serum of a healthy individual. These findings and suggestions may be of clinical importance and call for further studies.

Keywords: Autoantibodies, cytokines, glutamate receptor GluR3, epilepsy twins, Rasmussen’s encephalitis

Introduction Epilepsy is a major health problem, affecting , 1% of the world population. There are many different types of epilepsy, a substantial number with unknown etiology. Moreover, , 20 –30% of all epilepsy patients are completely refractory to all anti-epileptic drugs (AEDs) and suffer from a low quality of life. Can some types of human epilepsy be in fact associated with autoimmunity and abnormal cytokine secretion? Autoimmunity to a specific type of glutamate receptor is currently suspected to contribute, at least partially, to the etiology and/or pathology of some

human epilepsies. In specific, high levels of autoantibodies (Ab’s) against ionotropic glutamate receptor of the a-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA) subtype 3 (GluR3) were found in Rasmussen’s Encephalitis (RE), as well as in patients with other types of epilepsy [1 – 5]. Furthermore, Ab’s directed specifically against the GluR3B peptide (GluR3 amino acids (aa) 372 – 395) of this receptor significantly associate with frequent seizures (compared to occasional or drug-controlled seizures) [3]. Several additional lines of evidence provide further support to the notion that anti-GluR3 Ab’s may have

Correspondence: M. Levite, Department of Neurobiology, The Weizmann Institute of Science, 76100 Rehovot, Israel. Tel: 972 8 9342315/ 972 54 244814. Fax: 972 8 9344131. E-mail: [email protected] † Equal contribution ISSN 0891-6934 print/ISSN 1607-842X online q 2005 Taylor & Francis Ltd DOI: 10.1080/08916930500100825

140 Y. Ganor et al. an active neuropathological role in epilepsy: (1) Two rabbits immunized with a large portion of the GluR3 extracellular domain (GluR3 aa 245 – 457) developed epileptic seizures and RE-like inflammatory histopathology [4]; (2) Anti-GluR3 Ab’s can kill neurons and astrocytes [1,6 – 11], either via excitotoxicity [6,7]-over activation of glutamate receptors leading to neuronal death (like that caused by excess glutamate in various pathological conditions), and/or via complement fixation [8 – 11]; (3) A transient reduction in serum anti-GluR3 Ab’s (by plasma exchange [4] or long-term selective IgG immunoabsorption [12]) was accompanied in two RE patients by decreased seizure frequency and improved neurological function; (4) Functional hemispherotomy performed to two RE patients, which had high levels of anti-GluR3B Ab’s in serum and CSF, caused a marked decrease in the level of these Ab’s, in association with seizure arrest and neurological improvement [1]; (5) Immunization of mice with the GluR3B peptide leads to production of high titers of the respective anti-GluR3B Ab’s that could bind neurons, induce glutamate-like currents and kill neurons in vitro [7]. The GluR3B-immunized mice also exhibited brain pathology partially resembling that observed in RE patients [13]. Yet, the immunization with either GluR3B or another GluR3-derived peptide (GluR3A aa 245 – 274) did not trigger spontaneous overt epilepsy in mice [7,13] and rats

[14]. Thus, in animal models, the direct experimental evidence (for example by passive transfer) that antiGluR3B Ab’s can by themselves initiate an epileptic process is still missing, and the mechanism by which they can lead to pathology (exitotoxicity, complement fixation or both) is still to be determined. Above all that, it is still unknown, and therefore challenged by the present study, whether the mere presence of antiGluR3B Ab’s in the serum is necessarily detrimental. Besides anti-GluR3 Ab’s, several studies reported that other types of autoimmune Ab’s are found in some epilepsy patients (reviewed in [15]). On the basis of all the above, we tested herein a pair of monozygotic twins, one with epilepsy (diagnosed most probably as RE) and the other healthy, for Ab’s against GluR3B, double-stranded DNA (dsDNA), glutamic acid decarboxylase (GAD), nuclear antigens, b2-glycoprotein I (b2GPI) and cardiolipin. All these Ab’s are known already to have an ability to cause damage to the nervous system and/or peripheral organs (see Table I). We also tested the epileptic and healthy monozygotic twins for the serum levels of four cytokines: IFNg, TNFa, IL-4 and IL-10, all playing a profound role (either beneficial or detrimental, depending on their levels and context) in health and disease. In general, comparing monozygotic twins in relation to a non-genetic disease provides some insight into the course of the disease, the non-affected twin being

Table I. Characteristic features and pathological functions of the specific autoimmune Ab’s tested in this study. Autoantibody

Pathological function of autoantibodies

Found in

References

Anti-dsDNA

May cause tissue damage by forming immune complexes, or by penetrating into cells Can suppress gamma-amino butyric acid (GABA)-mediated transmission and reduce GABA synthesis in nerve terminals May cause apoptotic death of lymphocytes

Systemic lupus erythematosus (SLE). Recently found also in some epilepsy patients (16% of 80 patients tested) Insulin-dependent diabetes mellitus (IDDM), Stiff-person syndrome (SPS), Chronic cerebellar ataxia, Drug-resistant epilepsy, Myoclonus Mixed-connective tissue disease, SLE, Sjo¨gren’s Syndrome, Systemic sclerosis Sjo¨gren’s syndrome, SLE and neonatal/subacute cutaneous SLE, rheumatoid arthritis

[2,27]

Anti-GAD

Anti-RNP Anti-SS-A

Anti-Cardiolipin

Anti-beta2GPI

Anti-GluR3

Induce conduction abnormalities/block in human fetal heart and in hearts of neonatal and adult rabbits/mice May also cross-react with the serotoninergic 5HT receptor May contribute to tissue damage by forming immune complexes, interfere with the coagulation cascade and increase the risk of vascular complications May contribute to tissue damage by forming immune complexes, increase the risk of thrombocytopenia Induce hyperactive behavior and learning/ memory deficits in mice Induce depolarization of brain synaptoneurosomes Kill neurons and astrocytes, cause epilepsy in rabbits, cause brain histopathology in rabbits/mice/rats

Anti-phospholipid syndrome (APS), SLE, Recurrent fetal loss, Non-thrombotic neurological disorders such as chorea/epilepsy APS and other thrombotic disorders, Fetal loss in the context of SLE

Rasmussen’s Encephalitis (RE), Noninflammatory focal epilepsy, “Catastrophic” epilepsy, Partial and generalized epilepsy

[53]

[54,55] [56,57]

[31]

[33,34,58]

[1–7,10,13,14]

Elevated antibodies and cytokines in epilepsy considered as an image of the condition of the affected twin before the disease onset, and before some precipitating factor occurred. Several previous studies reported on monozygotic twins discordant for epilepsy [16,17]. Interestingly, it appears that siblings and monozygotic twins with similar chromosomal aberration associated with development of epilepsy are often discordant for seizure disorders [18]. Factors found in correlation with the discordance include the presence of major clinical risk factors, epileptogenic lesions (visualized by magnetic resonance imaging (MRI)) and quantitative brain volume abnormalities, as well as the existence of prenatal insults and of genetic abnormalities resulting from the post-fertilization genetic process [16]. However, while all the above studies clearly suggest that the susceptibility to develop epilepsy may be influenced by the combined action of a number of different genes and environmental factors, it is still unknown whether specific immune – autoimmune components (e.g. Ab’s and cytokines) may differ in monozygotic twins and contribute to the epilepsy discordance.

Materials and methods Patients Detailed case report of the epileptic twin. A six-year-old girl of Philippine origin having a healthy HLA-identical twin was admitted for continuous jerking of the left foot. After normal pregnancy and spontaneous delivery (38 gestational weeks, 2850 g, 48 cm), a normal psychomotor development and somatic history except an atopic dermatitis was reported. Neither epilepsy nor febrile convulsions were known in the family. The affected girl had four generalized seizures six weeks before admission, and anticonvulsive medication (valproic acid) was started. Two weeks after these seizures, a tooth was extracted due to an abscess. Two weeks after generalized tonic–clonic seizures (GTCs), epilepsia partialis continua (EPC) of the left foot started and oxcarbazepine was added outwards. Detailed investigations included extensive clotting tests, immunological markers, ammonia, lactate and pyruvate that were all normal, as were the screening for inborn errors of metabolism and serologic tests for Ab’s to cerebral viruses: herpes-simplex virus (HSV), cytomegaly virus (CMV), epstein-barr virus (EBV), human-immunodeficiency virus (HIV), early summer meningoencephalitis (FSME), measles and rubella. Lumbar puncture revealed no white blood cells, normal protein, no Ab’s to the tested viruses, no oligoclonal bands and no lactate acidosis. Electroencephalography (EEG) showed a progressive disappearance of basic activity on the right side, a slow wave focus on the right central area with multifocal epileptiform discharges on the whole right hemisphere. MRI three years after beginning of EPC showed

141

atrophy in the sylvian fissure frontal lobe. Technetium99 mm hexamethylpropylene amin oxim (HMPAO)single-photon emission scan tomography (SPECT) showed discrete higher uptake of the left hemisphere, whereas [18F] fluorodeoxyglucose positron emission tomography (FDG-PET) supported the diagnosis of RE on account of an increased glucose metabolism in the right motor cortex. Anticonvulsive medication ineffective for EPC included valproic acid, oxcarbazepine, vigabatrine, sulthiam, topiramate and benzodiazepines. Intravenous immunoglobulines and corticosteroid bolus therapy showed no effect. High dosages of phenobarbital, topiramate and continuous oral corticosteroids could prevent repeated generalized seizures and propagation of the continuous clonic jerks to the left arm and face. Since the onset of the EPC persisting during sleep, the patient developed left hemiparesis, cognitive deterioration and behavioral difficulties. Due to the side effects of corticosteroids and sedation with phenobarbital, it was decided to switch to tacrolimus (FK506) and reduce the dosage of phenobarbital, and this permitted better speech and vigilance performances without deterioration of EPC. The patient is now in a stable condition with blood levels of tacrolimus within 5–10 ng/ml, able to walk and attending school. The association of EPC with focal EEG and neuroimaging changes allowed the tentative diagnosis of RE [19 – 23]. Other partial lesional epilepsies or systemic diseases with EPC (like mitochondrial diseases) could be excluded. Due to this clinical, neurophysiological and radiological evidence of RE, a biopsy was not performed, in particular knowing the unspecific histopathologic findings [21]. Additional RE patients. Patient HF: a 6-year-old girl, diagnosed with RE on the basis of refractory focal seizures, left frontal atrophy on CT and MRI, left frontal intense signal on T2-weighted MRI and oligoclonal bands in CSF; Patient SA: a 16-year-old girl, diagnosed with RE on the basis of refractory seizures (focal with secondary generalization), hypodense lesion within the right temporal lobe on CT, global atrophy of the right hemisphere on MRI, large hypometabolism of the right hemisphere on FDG-PETand oligoclonal bands in CSF; patient EF: a 16-year-old girl diagnosed with RE on the basis of progressive left hemisphere atrophy on MRI. She was also diagnosed with encephalopathy and acute intermittent porphyria (AIP) based upon her blood and urine results. These RE patients were also investigated in our previous study [1]. Non-epileptic control patients. 49 epilepsy-free and neurologically healthy individuals included 26 males, mean age of all individuals: 7.2 years (95% CI’s 5.5 – 8.9 years).

142 Y. Ganor et al. Informed consents for sampling of blood and CSF of the epileptic twin, her healthy sister and all the other patients included in this study, were obtained from all patients or their parents/guardians upon approval by the Helsinki Committee at their institution. Detection of anti-GluR3B and anti-dsDNA Ab’s The levels of anti-GluR3B Ab’s were tested by ELISA, as described previously [1,7], measuring in parallel the specific binding to the GluR3B peptide (NEYERFVPFSDQQISNDSSSSENR, synthesized at the Weizmann Institute of Science) and the non-specific binding to bovine serum albumin (BSA). The levels of anti-dsDNA Ab’s were tested by ELISA, using plates coated with Lambda phage dsDNA (Worthington Biochemical Corporation, Lakewood, NJ, USA) as previously described [24]. Evaluation of the relative levels of anti-GluR3B and anti-dsDNA Ab’s was based on an estimated threshold value, calculated separately for each type of Ab, as the mean Ab level of the neurologically intact nonepileptic control group þ 2SD. Detection of Ab’s directed against GAD, RNP-70, SS-A, cardiolipin and b2GPI The Ab’s against GAD, RNP-70, SS-A, cardiolipin and b2-GPI were detected using the commercial ELISA diagnostic kits used routinely in hospitals worldwide. Their clinical evaluation was performed as instructed by the respective manufacturer, which also supplied the negative and positive controls for each Ab type, and the exact cutoff values above which a patient is considered clinically positive. Specifically, the diagnostic kits and the clinical cutoffs were as follows: anti-GAD IgG diagnostic kit (Isletest-GAD; Biomerica, Newport Beach, CA, USA), anti-GAD Ab’s level above 1.0 “GAD value” is considered clinically positive and a level of 1.0 –1.05 as borderline; antiRNP-70 and anti-SS-A IgG diagnostic kits

(Medizym anti-RNP and Medizym anti-SS-A; Medipan Diagnostica, Selchow, Germany), anti-RNP Ab’s level above 15.0 U/ml and anti-SS-A Ab’s level above 10.0 U/ml are considered clinically elevated; anticardiolipin and anti-b2GPI diagnostic kits (ORG515 and ORG521; Orgentech Diagnostika GmbH, Mainz, Germany), anti-cardiolipin IgG level above 10.0 U/ml and anti-b2GPI level (both IgM and IgG) above 8.0 U/ml are considered clinically elevated, and levels of 5.0 –8.0 U/ml are borderline. Measurements of cytokine levels Measurements of IFN-g, TNF-a, IL-4, IL-10 in the sera of the monozygotic twins and control nonepileptic individuals were performed by quantitative sandwich ELISA kits (R & D systems, Minneapolis, MN, USA) according to the manufacturer instructions. Each sample was tested in triplicate microtiter wells. Serial dilutions of IFN-g, TNF-a, IL-4 and IL10 standards (supplied by the manufacturer) were seeded in each experiment, in order to obtain standard curves and to allow the translation of the OD measured into pg/ml of the respective cytokine. Statistical analysis Statistical significance was analyzed by Student’s t-test. Results The majority of the results are summarized in Table II. Both epileptic and healthy twins harbor elevated serum levels of anti-GluR3B Ab’s We tested for Ab’s against GluR3B peptide believed to be the primary antigenic epitope for excitotoxic antiGluR3 Ab’s, in the serum of the epileptic twin, three additional RE patients and 49 neurologically intact non-epileptic controls. Each individual serum was

Table II. The main findings of the study in regards to the levels of various autoimmune Ab’s and of IFNg, TNFa, IL-4 and IL-10, in the serum of monozygotic twins discordant for epilepsy (most probably RE). Epileptic twin Antibodies Anti-GluR3B Anti-cardiolipin Anti-dsDNA Anti-GAD Anti-RNP Anti-SS-A Anti-beta2GPI Cytokines IFN-g TNF-a IL-10 IL-4

Elevated Clinically elevated Elevated Clinically elevated Clinically elevated Clinically elevated Above clinical borderline Elevated, but less Elevated, but less Elevated, but less Elevated, but less

Comparison between the twins

¼ (no significant difference) ¼ (no significant difference) . P ¼ 0:003 . P ¼ 0:017 . P ¼ 0:045 . P ¼ 0:034 . ,P ,P ,P ,P

¼ 0:0005 ¼ 0:0002 ¼ 0:0097 ¼ 0:0004

Healthy twin

Elevated Clinically elevated Elevated, but less Clinically elevated, but less Clinically elevated, but less Clinically elevated, but less Below clinical borderline Elevated Elevated Elevated Elevated

Elevated antibodies and cytokines in epilepsy tested in parallel for specific binding to GluR3B (in GluR3B-coated plates) and for non-specific binding (in BSA-coated plates). The results, expressed for each sample as (specific binding to GluR3B peptide— non-specific binding to BSA), showed that the epileptic twin and the other RE patients contained elevated anti-GluR3B Ab’s in serum (Figure 1A; cutoff value calculated as the mean þ 2SD Ab level of the control group). We found that the serum of both monozygotic twins, one of which is epileptic and the other healthy (taken three months after the first seizure in the epileptic twin) contained significantly higher levels of anti-GluR3B Ab’s, compared to the controls (Figure 1B, *p , 0:05; Student’s t-test) and that the magnitude of elevation of anti-GluR3B Ab’s

143

found in the twins was similar (Figure 1B, p ¼ 0:191; Student’s t-test, Table II). Thus, anti-GluR3B Ab’s can be found also in the serum of a healthy person, suggesting that serum anti-GluR3B Ab’s are not necessarily detrimental, and that they are also not produced as a side effect of already existing epilepsy. Anti-GluR3B Ab’s are found in the CSF of the epileptic twin Importantly, we found that the epileptic twin contains anti-GluR3B Ab’s in her CSF, i.e. across the blood brain barrier (BBB) and so do the three other RE patients tested (Figure 1C). Due to the unavailability of CSF of the healthy twin, we unfortunately could not

Figure 1. Elevated levels of anti-GluR3B Ab’s are found in the serum of both twins and in the CSF of the epileptic twin. (A, B) Anti-GluR3B IgG in the serum of the epileptic twin, three additional RE patients and 49 neurologically intact non-epileptic controls (A) and in the healthy monozygotic twin (B). (C) Anti-GluR3B IgG in the CSF of the epileptic twin, three additional RE patients and neurologically intact non-epileptic controls. Values represent the mean OD of three independent experiments and considered elevated if exceeding a threshold (black horizontal line in (A) and dashed line in (C)) calculated as the mean anti-GluR3B IgG level of the control group þ2SD. In all experiments performed, each individual serum or CSF was tested in parallel for specific binding to GluR3B and for non-specific binding to BSA. The final level of specific anti-GluR3B Ab’s was calculated as (specific binding to GluR3B minus non-specific binding to BSA). The results in the figure represent anti-GluR3B IgG levels in 1:100 serum dilution and 1:10 CSF dilution. In (B), *p , 0:05 vs. control, Student’s t-test. In addition, the p value of the difference between the epileptic and healthy twin is shown.

144 Y. Ganor et al. test our prediction that her CSF (unlike her epileptic twin) may be devoid of such Ab’s. In principle, Ab’s found in the CSF can either origin from intrathecal Ab production or ooze from the periphery, via a leaky BBB. To evaluate whether the Ab’s found in the CSF of the epileptic twin derive from intrathecal production, we adopted the commonly accepted calculation of the IgG index ¼ (CSF IgG/serum IgG)/(CSF albumin/serum albumin), the cutoff being 0.7 (i.e. . 0.7 index being a sign for intrathecal Ab production) [25,26]. The epileptic twin was found to have an IgG index of 0.52, arguing against an intrathecal Ab production. Both twins harbor elevated serum levels of anti-dsDNA Ab’s, but the levels in the epileptic twin are significantly higher Do the epileptic twin and the other RE patients have in their sera elevated levels of anti-dsDNA Ab’s (see Table I), known for their ability to cause tissue damage [27]? While anti-dsDNA Ab’s are believed to be present almost exclusively in systemic lupus erythematosus (SLE), we recently found that 16% of epilepsy patients (13 patients out of 80 studied) harbor elevated levels of such anti-dsDNA Ab’s [2]. A possible association between Ab’s to GluR3B and dsDNA was also suggested by our previous observations made with GluR3B-immunized mice, which developed not only the corresponding anti-GluR3B Ab’s, but also high titers of anti-DNA Ab’s [13]. Furthermore, we recently found specific anti-glutamate/N-methyl-D -aspartate (NMDA) Ab’s in 31% of SLE patients (34 patients out of 109 studied) [28]. Finally, a recently reported case of RE followed by SLE suggested a possible connection between the two diseases [29]. Herein, we found that the epileptic twin, as well as one of the three additional RE patients, harbor significantly elevated anti-dsDNA Ab’s in their serum, even at dilution as high as 1:250 (Figure 2A; cutoff value calculated as the mean þ 2SD Ab level of the control group). Interestingly, elevated levels of antidsDNA Ab’s were also found in the serum of the healthy twin, but at significantly lower levels than her epileptic sister (Figure 2B, p ¼ 0:0031; Student’s t-test; Table II). Importantly, the epileptic twin and the other RE patient found herein to be positive for serum anti-dsDNA Ab’s (Figure 2A) also harbor these Ab’s in their CSF (Figure 2C). Again, the unavailability of CSF of the healthy twin prevented the testing for anti-dsDNA Ab’s in her CSF. Both twins harbor clinically elevated serum levels of Ab’s against GAD, RNP-70 and SS-A, but their levels in the epileptic twin are significantly higher Table I summarizes the additional autoimmune Ab’s tested in this study, the autoimmune diseases in which

they are classically found, and some of their known pathogenic activities in the nervous system and peripheral organs of humans and/or animal models. Of note, unlike the tests described above for Ab’s to GluR3B and dsDNA, the screen for the additional autoimmune Ab’s was performed using diagnostic kits commonly utilized in hospitals worldwide, which contain build-in negative controls, and which provide precise “clinical thresholds” above which a patient is considered positive. We found clinically elevated levels of anti-GAD Ab’s in the serum of the epileptic twin, in an additional RE patient, and also in the serum of the healthy twin (Figure 3A). Interestingly, the anti-GAD Ab’s levels were significantly higher in the epileptic twin than in her healthy sister ðp ¼ 0:017; Student’s t-test). The CSF of the epileptic twin did not contain anti-GAD Ab’s (data not shown). As to anti-nuclear Ab’s (ANA) (Table I), we detected clinically elevated levels of anti-ribonucleoprotein 70 (RNP-70) (Figure 3B) and anti-SS-A Ab’s (Figure 3C) in the serum of the epileptic twin and in two additional RE patients. Ab’s to RNP-70 and SS-A were also found in the serum of the healthy twin, but in significantly lower levels compared to her epileptic twin (Figure 3B, p ¼ 0:045 and Figure 3C, p ¼ 0:034; Student’s t-test). The CSF of the epileptic twin did not contain Ab’s to neither RNP-70 nor SS-A (data not shown). Both twins have clinically elevated serum levels of anticardiolipin Ab’s, but only the epileptic twin has elevated levels of anti-b2GPI Ab’s We next tested for anti-cardiolipin Ab’s, found not only in autoimmune diseases ([30,31] and Table I), but also in non-thrombotic neurological disorders of the central nervous system (CNS), such as various epileptic diseases [32] and chorea. We found that the epileptic twin, as well as her healthy twin, both contain in their sera clinically elevated levels of anti-cardiolipin Ab’s of the IgG subclass. Of note, anti-cardiolipin IgG’s are found characteristically in progressive stages of manifested autoimmune disorders, as opposed to anti-cardiolipin IgM’s which are usually an indicator of the beginning of autoimmune diseases (Figure 3D, Table I). The levels of anticardiolipin IgG’s in the epileptic twin and healthy twin were of similar magnitude (i.e. not statistically different). The CSF of the epileptic twin was devoid of anti-cardiolipin IgG (data not shown). As for anti-b2GPI Ab’s ([33 –35] and Table I), we found that their level in the serum of the epileptic twin was above the clinical borderline level. In contrast, no signs for elevated anti-b2GPI IgG were found in the serum of the healthy sister (Figure 3E). The CSF of the epileptic twin was devoid of anti-b2GPI IgG (data not shown).

Elevated antibodies and cytokines in epilepsy

145

Figure 2. Elevated levels of anti-dsDNA Ab’s are found in the serum of both twins and in the CSF of the epileptic twin. (A, B) Anti-dsDNA IgG in the serum of the epileptic twin, three additional RE patients and 45 neurologically intact non-epileptic controls (A) and in the healthy monozygotic twin (B). (C) Anti-dsDNA IgG in the CSF of the epileptic twin, three additional RE patients and the controls. Values represent the mean OD of two (serum) or one (CSF) experiments and considered elevated if exceeding a threshold (black horizontal line in (A) and dashed line in (C)) calculated as the mean anti-dsDNA IgG level of the control group þ 2SD. The results in the figure represent anti-dsDNA IgG levels in 1:250 serum dilution and 1:10 CSF dilution. In (B), *p , 0:05 vs. control, Student’s t-test. In addition, the p value of the difference between the epileptic and healthy twin is shown.

Both twins have elevated serum levels of IFNg, TNFa, IL-4 and IL-10 compared to controls, but the levels in the epileptic twin are significantly lower than those of her sister We found that in comparison to control sera ðn ¼ 17Þ; the serum of the epileptic twin, as well as of her healthy twin (the sera collected three months after the first seizure in the epileptic twin), contained significantly elevated levels of four key Th1 and Th2 cytokines: IFNg, TNFa, IL-4 and IL-10 (Figure 4A – D, Table II, *p , 0:05; Student’s t-test). Surprisingly, the levels of all these cytokines were significantly lower in the epileptic twin than in the healthy twin (Figure 4A – D). The p values for the differences between the epileptic and healthy twins in regards to the four cytokines are shown in Table II.

Discussion We found that monozygotic twins discordant for human epilepsy (most probably RE) contain in their serum elevated levels of several anti-self Ab’s, directed against: the B peptide of glutamate/AMPA receptor subtype 3 (GluR3B), dsDNA, GAD, the nuclear antigens RNP and SS-A and cardiolipin. The epileptic twin had also somewhat elevated levels of anti-b2GPI Ab’s. All these Ab’s are known for their pathogenic potential, which if unveiled in-vivo may lead to damage to peripheral organs and/or to the nervous system (Table I). Furthermore, the “classical” autoimmune Ab’s, found at clinically elevated levels in the serum of the twins, are of IgG isotype, a characteristic feature of progressive autoimmune diseases (e.g. [36]).

146 Y. Ganor et al.

Figure 3. Both twins contain clinically elevated levels of Ab’s to GAD, nuclear antigens and cardiolipin, but the levels of most Ab’s are significantly higher in the epileptic twin. (A–C) The presence of anti-GAD IgG (A) and anti-nuclear antigens RNP-70 (B) and SS-A (C) IgG’s was analyzed by specific commercial ELISA kits used for clinical diagnostic purposes. Results represent the levels of the respective Ab’s in the serum of the epileptic twin, three additional RE patients, the healthy twin and non-epileptic controls. (A) Mean anti-GAD IgG concentration (termed GAD value) in serum (1:100 dilution). Of note, GAD value above 1.05 is considered clinically positive according to the manufacturer’s guidelines. Statistical analysis comparing values in the epileptic twin vs. healthy twin: p ¼ 0:017; Student’s t-test. (B, C) Mean anti-RNP (B) and anti-SS-A (C) IgG concentration (U/ml) in the serum (diluted 1:100). The results of one experiment out of two performed are presented. Anti-RNP IgG level above 15.0 and an anti-SS-A IgG level above 10.0 are considered clinically positive according to the manufacturer’s guidelines. Statistical analysis comparing anti-RNP and anti-SS-A IgG in the epileptic twin vs. healthy twin: p ¼ 0:045 (B) and p ¼ 0:034 (C), Student’s t-test. (D, E) Mean IgG’s against cardiolipin (D) and b2GPI (E) in the serum. Numbers represent the mean IgG concentration (U/ml) detected in the serum (1:100 dilution) of one experiment out of three performed. According to the manufacturer’s guidelines, anti-cardiolipin IgG level above 10.0 is considered clinically abnormal; As to anti-b2GPI IgG, a level above 8.0 is considered clinically positive and between 5.0 and 8.0 is considered borderline. NS ¼ not significant ðp . 0:05Þ:


147

Figure 4. The serum of both twins contains elevated levels of key cytokines, but the levels in the healthy twin are significantly higher. The levels of IFNg (A), TNFa (B), IL-4 (C) and IL-10 (D) in the serum of the epileptic twin, her healthy monozygotic twin and neurologically intact non-epileptic controls (white bars, mean OD ^ SD of n controls). The results of one representative experiment out of two performed are presented. Values (pg/ml) represent mean ^ SD concentration of triplicate serum samples. *p , 0:05 vs. control, Student’s t-test. Statistical analysis comparing the cytokine levels between the epileptic and healthy twins was performed by Student’s t-test and the values are shown in the respective figures.

We further found that both the epileptic and her healthy twin had significantly elevated levels of four pro-inflammatory and anti-inflammatory key cytokines: IFNg, TNFa, IL-4 and IL-10. Interestingly, while there was no significant difference between the twins in regards to anti-GluR3B Ab’s (i.e. they had comparably elevated levels of these Ab’s), significant differences between the twins were evident in regards to the titers of the other autoimmune Ab’s and of the four cytokines: the epileptic twin, compared to her healthy twin, had significantly higher levels of Ab’s to dsDNA, GAD, RNP-70, SS-A and b2GPI, but significantly lower levels of all four key cytokines tested (Table II). Importantly, the epileptic twin contains both antiGluR3B Ab’s and anti-dsDNA Ab’s not only in the serum but also in her CSF (currently not testable in the healthy twin). In principle, the presence of anti-GluR3B Ab’s in the CSF may be detrimental, in view of their documented ability to kill neurons (regardless of the exact mechanism: either

excitotoxicity, or complement fixation or both) shown by now by several different studies [1,6 – 8,10,11], and in view of their potential to lead to seizures (thus far reported by a single study) [4]. In principle, Ab’s found in the CSF are not necessarily produced intrathecally, as seems to be the case of the epileptic twin studied herein. Rather, the Ab’s found across her BBB were most probably produced in the periphery, perhaps in response to a T-cell derived GluR3, since we recently identified high levels of GluR3 on the cell-surface of human T-cells [37]. Later, such peripheral anti-GluR3 Ab’s might have oozed into the CSF across a leaky or damaged BBB. In line with these results, a recent study finds no evidence for intrathecal synthesis of anti-GluR3A (GluR3 aa 274 –302) IgG in four RE patients [3] despite their presence in the CSF and serum. What caused the production of a kaleidoscope of autoimmune Ab’s detected herein in the serum of both twins is currently unknown. This is especially hard to answer in retrospect and in view of the absence of

148 Y. Ganor et al. a detailed documentation of all the potentially relevant “immune events” that may have affected the twins in the weeks/months/years prior to the outburst of epilepsy. The picture is even further complicated by the unavailability of samples of serum and more importantly CSF of both twins at various time points before and after the outburst of epilepsy in one of them. On these grounds it is also impossible to challenge retrospectively our speculation that the higher titers of Ab’s and the lower levels of cytokines produced at a certain time point (perhaps in response to some “immune” irritation alike viral/bacterial infection affecting both twins) might have made one sister more prone to develop seizures. Yet, despite all these open questions and speculations, which clearly require additional investigation in further studies, some lessons may already be learned from several potentially relevant studies reporting that: (1) monozygotic twins can be discordant for “classical” autoimmune diseases, such as insulin-dependent diabetes mellitus (IDDM) [38,39], SLE [40] and myasthenia gravis [41,42]; (2) both ill and healthy twins discordant for a particular autoimmune disease may contain Ab’s to disease-specific antigens [43]; (3) monozygotic twins discordant for an autoimmune disease may differ in their cytokine milieu [44]; (4) high levels of the proinflammatory cytokine IFNg produced at early stages of an infection (e.g. by coxsackievirus) is not necessarily detrimental and can in fact be crucially protective, e.g. it can arrest the potential deterioration of a regulated immune/autoimmune response against cardiac myosin into a full blown detrimental autoimmune disease, namely experimental autoimmune myocarditis [45 – 47]; (5) high levels of the proinflammatory cytokine TNFa, considered for many years to be necessarily detrimental, can also be beneficial in some contexts, for example, by protecting neuronal tissue from injury, by promoting proliferation of oligodendrocyte precursor cells and differentiation into myelinating oligodendrocytes, and by stimulating neurite outgrowth [48]; (6) cytokines may play a role in certain types of epilepsy (e.g. temporal lobe epilepsy, febrile convulsions and tonic –clonic seizures; reviewed in [49]), and an imbalance in serum cytokine levels may have a role in the pathogenesis of seizures (e.g. west syndrome [50] and febrile convulsions [51]). Taken together, we believe that all the findings and ideas raised herein call for further studies on various types of autoimmune Ab’s and cytokines (those tested in this study and others) in a large number of epilepsy patients, and if possible, in additional pairs of monozygotic twins discordant for epilepsy. Exploring the possible immune/autoimmune face of some epilepsies may pave the way for novel strategies to diagnose and treat some immune/autoimmune associated epilepsies, entitled “autoimmune epilepsy” [52],

preferably at an early stage of the disease, and before drastic brain surgery disease, and especially in severe cases of intractable seizures.

Acknowledgements The authors thank Professor Z. Vogel (WIS) and Professor D. Lichtenberg (TLV university) for their kind and important support, and Professor V.I. Teichberg (WIS) for the encouraging scientific environment and the stimulating discussions. We further thank Dr Hadassa Goldberg-Stern, Dr Tally Lerman-Sagie (Israel), Dr Dina Amrom, Dr Denis Verheulpen and Professor Patrick Van Bogaert (Belgium) for the serum/CSF and clinical information of the RE patients. The study was supported by grants to M.L. from the Rochlin foundation, USA.

References [1] Ganor Y, Goldberg-Stern H, Amrom D, Lerman-Sagie T, Teichberg VI, Pelled D, Futerman AH, Ben-Zeev B, Freilinger M, Verheulpen D, Van Bogaert P, Levite M. Autoimmune epilepsy: Some epilepsy patients harbor autoantibodies to glutamate receptors and dsDNA on both sides of the bloodbrain barrier, which may kill neurons and decrease in brain fluids after Hemispherotomy. Clin Dev Immunol 2004;11:241–252. [2] Ganor Y, Goldberg-Stern H, Lerman-Sagie T, Teichberg VI, Levite M. Autoimmune epilepsy: Distinct subpopulations of epilepsy patients harbor serum autoantibodies to either glutamate/AMPA receptor GluR3, glutamate/NMDA receptor subunit NR2A or double-stranded DNA (Revised) 2005. [3] Mantegazza R, Bernasconi P, Baggi F, Spreafico R, Ragona F, Antozzi C, Bernardi G, Granata T. Antibodies against GluR3 peptides are not specific for Rasmussen’s encephalitis but are also present in epilepsy patients with severe, early onset disease and intractable seizures. J Neuroimmunol 2002;131:179 –185. [4] Rogers SW, Andrews PI, Gahring LC, Whisenand T, Cauley K, Crain B, Hughes TE, Heinemann SF, McNamara JO. Autoantibodies to glutamate receptor GluR3 in Rasmussen’s encephalitis. Science 1994;265:648–651. [5] Wiendl H, Bien CG, Bernasconi P, Fleckenstein B, Elger CE, Dichgans J, Mantegazza R, Melms A. GluR3 antibodies: Prevalence in focal epilepsy but no specificity for Rasmussen’s encephalitis. Neurology 2001;57:1511 –1514. [6] Koustova E, Sei Y, Fossom L, Wei ML, Usherwood PN, Keele NB, Rogawski MA, Basile AS. LP-BM5 virus-infected mice produce activating autoantibodies to the AMPA receptor. J Clin Investig 2001;107:737 –744. [7] Levite M, Fleidervish IA, Schwarz A, Pelled D, Futerman AH. Autoantibodies to the glutamate receptor kill neurons via activation of the receptor ion channel. J Autoimmun 1999;13:61 –72. [8] He XP, Patel M, Whitney KD, Janumpalli S, Tenner A, McNamara JO. Glutamate receptor GluR3 antibodies and death of cortical cells. Neuron 1998;20:153–163. [9] Whitney KD, Andrews PI, McNamara JO. Immunoglobulin G and complement immunoreactivity in the cerebral cortex of patients with Rasmussen’s encephalitis. Neurology 1999;53:699–708. [10] Whitney KD, McNamara JO. GluR3 autoantibodies destroy neural cells in a complement-dependent manner modulated


[11]

[12]

[13]

[14]

[15] [16]

[17] [18] [19] [20]

[21]

[22] [23]

[24]

[25]

[26]

[27] [28]

[29]

[30]

by complement regulatory proteins. J Neurosci 2000; 20:7307–7316. Xiong ZQ, McNamara JO. Fleeting activation of ionotropic glutamate receptors sensitizes cortical neurons to complement attack. Neuron 2002;36:363–374. Antozzi C, Granata T, Aurisano N, Zardini G, Confalonieri P, Airaghi G, Mantegazza R, Spreafico R. Long-term selective IgG immuno-adsorption improves Rasmussen’s encephalitis. Neurology 1998;51:302– 305. Levite M, Hermelin A. Autoimmunity to the glutamate receptor in mice–a model for Rasmussen’s encephalitis? J Autoimmun 1999;13:73 –82. Ganor Y, Gottlieb M, Eilam R, Otmy H, Teichberg VI, Levite M. Immunization with the glutamate receptor derived peptide GluR3B induces neuronal death and reactive gliosis, but confers partial protection from pentylenetetrazole-induced seizures (Revised) 2005. Levite M, Hart IK. Immunotherapy for epilepsy. Expert Rev Neurotherapeutics 2002;2:804 –819. Briellmann RS, Jackson GD, Torn-Broers Y, Berkovic SF. Causes of epilepsies: Insights from discordant monozygous twins. Ann Neurol 2001;49:45 –52. Brodtkorb E. Monozygous twins discordant for epilepsy. Ann Neurol 2002;51:137. Schinzel A, Niedrist D. Chromosome imbalances associated with epilepsy. Am J Med Genet 2001;106:119– 124. Dulac O. Rasmussen’s syndrome. Curr Opin Neurol 1996;9:75–77. Granata T, Gobbi G, Spreafico R, Vigevano F, Capovilla G, Ragona F, Freri E, Chiapparini L, Bernasconi P, Giordano L, Bertani G, Casazza M, Dalla Bernardina B, Fusco L. Rasmussen’s encephalitis: Early characteristics allow diagnosis. Neurology 2003;60:422–425. Prayson RA, Frater JL. Rasmussen encephalitis: A clinicopathologic and immunohistochemical study of seven patients. Am J Clin Pathol 2002;117:776 –782. Rasmussen T, Olszewski J, Lloyd-Smith D. Focal seizures due to chronic localized encephalitis. Neurology 1958;8:435 –445. Rasmussen T. Further observations on the syndrome of chronic encephalitis and epilepsy. Appl Neurophysiol 1978; 41:1–12. Eilat E, Dayan M, Zinger H, Mozes E. The mechanism by which a peptide based on complementarity-determining region-1 of a pathogenic anti-DNA auto-Ab ameliorates experimental systemic lupus erythematosus. Proc Natl Acad Sci USA 2001;98:1148 –1153. Lefvert AK, Link H. IgG production within the central nervous system: A critical review of proposed formulae. Ann Neurol 1985;17:13 –20. Link H, Tibbling G. Principles of albumin and IgG analyses in neurological disorders. III. Evaluation of IgG synthesis within the central nervous system in multiple sclerosis. Scand J Clin Lab Investig 1977;37:397–401. Hahn BH. Antibodies to DNA. N Engl J Med 1998; 338:1359–1368. Husebye ES, Sthoeger ZM, Dayan M, Zinger H, Elbirt D, Levite M, Mozes E. Autoantibodies to a NR2A peptide of the glutamate/NMDA receptor in sera of patients with systemic lupus erythematosus. Ann Rheum Dis 2005;11:11. Lascelles K, Dean AF, Robinson RO. Rasmussen’s encephalitis followed by lupus erythematosus. Dev Med Child Neurol 2002;44:572–574. Cervera R, Piette JC, Font J, Khamashta MA, Shoenfeld Y, Camps MT, Jacobsen S, Lakos G, Tincani A, KontopoulouGriva I, Galeazzi M, Meroni PL, Derksen RH, de Groot PG, Gromnica-Ihle E, Baleva M, Mosca M, Bombardieri S, Houssiau F, Gris JC, Quere I, Hachulla E, Vasconcelos C, Roch B, Fernandez-Nebro A, Boffa MC, Hughes GR, Ingelmo M. Antiphospholipid syndrome: Clinical and immunologic

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42] [43]

[44]

[45]

[46]

[47]

149

manifestations and patterns of disease expression in a cohort of 1,000 patients. Arthritis Rheum 2002;46:1019 –1027. Luzzana C, Gerosa M, Riboldi P, Meroni PL. Update on the antiphospholipid syndrome. J Nephrol 2002;15:342–348. Cimaz R, Romeo A, Scarano A, Avcin T, Viri M, Veggiotti P, Gatti A, Lodi M, Catelli L, Panzeri P, Cecchini G, Meroni PL. Prevalence of anti-cardiolipin, anti-beta2 glycoprotein I and anti-prothrombin antibodies in young patients with epilepsy. Epilepsia 2002;43:52–59. George J, Gilburd B, Langevitz P, Levy Y, Nezlin R, Harats D, Shoenfeld Y. Beta2 glycoprotein I containing immune-complexes in lupus patients: Association with thrombocytopenia and lipoprotein (a) levels. Lupus 1999;8:116–120. Shoenfeld Y, Gharavi A, Koike T. Beta2GP-I in the anti phospholipid (Hughes’) syndrome–from a cofactor to an autoantigen–from induction to prevention of antiphospholipid syndrome. Lupus 1998;7:503 –506. Blank M, Krause I, Fridkin M, Keller N, Kopolovic J, Goldberg I, Tobar A, Shoenfeld Y. Bacterial induction of autoantibodies to beta2-glycoprotein-I accounts for the infectious etiology of antiphospholipid syndrome. J Clin Investig 2002;109:797 –804. Carreras LO, Forastiero RR, Martinuzzo ME. Which are the best biological markers of the antiphospholipid syndrome? J Autoimmun 2000;15:163–172. Ganor Y, Besser M, Ben-Zakay N, Unger T, Levite M. Human T cells express a functional ionotropic glutamate receptor GluR3, and glutamate by itself triggers integrin-mediated adhesion to laminin and fibronectin and chemotactic migration. J Immunol 2003;170:4362–4372. Petersen JS, Kyvik KO, Bingley PJ, Gale EA, Green A, Dyrberg T, Beck-Nielsen H. Population based study of prevalence of islet cell autoantibodies in monozygotic and dizygotic Danish twin pairs with insulin dependent diabetes mellitus. BMJ 1997;314:1575–1579. Redondo MJ, Rewers M, Yu L, Garg S, Pilcher CC, Elliott RB, Eisenbarth GS. Genetic determination of islet cell autoimmunity in monozygotic twin, dizygotic twin, and non-twin siblings of patients with type 1 diabetes: prospective twin study. BMJ 1999;318:698 –702. Bustabad S, Gonzalez T, Trujillo E. Systemic lupus erythematosus in childhood. Concordance in a pair of monozygotic twins for anti-dsDNA antibodies and discordance for disease expression. J Rheumatol 1997;24:1450–1451. Lefvert AK, Pirskanen R, Eng H, Sundewall AC, Svanborg E. B cell and autoantibody repertoire in a pair of monozygotic twins discordant for myasthenia gravis. Clin Immunol Immunopathol 1989;53:161–170. Murphy J, Murphy SF. Myasthenia gravis in identical twins. Neurology 1986;36:78–80. Bergkvist M, Sandberg-Wollheim M. Serological differences in monozygotic twin pairs discordant for multiple sclerosis. Acta Neurol Scand 2001;104:262 –265. Kallmann BA, Lampeter EF, Hanifi-Moghaddam P, Hawa M, Leslie RD, Kolb H. Cytokine secretion patterns in twins discordant for Type I diabetes. Diabetologia 1999; 42:1080–1085. Afanasyeva M, Wang Y, Kaya Z, Park S, Zilliox MJ, Schofield BH, Hill SL, Rose NR. Experimental autoimmune myocarditis in A/J mice is an interleukin-4dependent disease with a Th2 phenotype. Am J Pathol 2001;159:193–203. Rose NR, Afanasyeva M, Wang Y, Kaya Z, Fairweather D. Infection to autoimmunity: The delicate balance. Autoimmun Rev 2002;1:14, (abstract). Horwitz MS, La Cava A, Fine C, Rodriguez E, Ilic A, Sarvetnick N. Pancreatic expression of interferon-gamma

150 Y. Ganor et al.

[48] [49] [50]

[51]

[52] [53]

[54]

protects mice from lethal coxsackievirus B3 infection and subsequent myocarditis. Nat Med 2000;6:693–697. Finsen B, Antel J, Owens T. TNFalpha: Kill or cure for demyelinating disease? Mol Psychiatr 2002;7:820 –821. Aarli JA. Role of cytokines in neurological disorders. Curr Med Chem 2003;10:1931–1937. Liu ZS, Wang QW, Wang FL, Yang LZ. Serum cytokine levels are altered in patients with west syndrome. Brain Dev 2001;23:548– 551. Virta M, Hurme M, Helminen M. Increased plasma levels of pro- and anti-inflammatory cytokines in patients with febrile seizures. Epilepsia 2002;43:920–923. Levite M. Autoimmune epilepsy. Nat Immunol 2002;3:500. Vianello M, Tavolato B, Giometto B. Glutamic acid decarboxylase autoantibodies and neurological disorders. Neurol Sci 2002;23:145–151. Alarcon-Segovia D, Llorente L, Ruiz-Arguelles A. The penetration of autoantibodies into cells may induce tolerance

[55]

[56]

[57]

[58]

to self by apoptosis of autoreactive lymphocytes and cause autoimmune disease by dysregulation and/or cell damage. J Autoimmun 1996;9:295– 300. Hiepe F, Dorner T, Burmester G. Antinuclear antibody- and extractable nuclear antigen-related diseases. Int Arch Allergy Immunol 2000;123:5 –9. Eftekhari P, Salle L, Lezoualc’h F, Mialet J, Gastineau M, Briand JP, Isenberg DA, Fournie GJ, Argibay J, Fischmeister R, Muller S, Hoebeke J. Anti-SSA/Ro52 autoantibodies blocking the cardiac 5-HT4 serotoninergic receptor could explain neonatal lupus congenital heart block. Eur J Immunol 2000;30:2782–2790. Mavragani CP, Tzioufas AG, Moutsopoulos HM. Sjogren’s syndrome: Autoantibodies to cellular antigens. Clinical and molecular aspects. Int Arch Allergy Immunol 2000; 123:46–57. Katzav A, Chapman J, Shoenfeld Y. CNS dysfunction in the antiphospholipid syndrome. Lupus 2003;12:903 – 907.