ABSTRACT
Anti-NMDA receptor (NMDAR) autoantibodies have been postulated to play a role in the pathogenesis of NMDAR hypofunction, which contributes to the etiology of psychotic symptoms. Toxoplasma gondii is a pathogen implicated in psychiatric disorders and associated with elevation of NMDAR autoantibodies. However, it remains unclear whether parasite infection is the cause of NMDAR autoantibodies. By using mouse models, we found that NMDAR autoantibody generation had a strong temporal association with tissue cyst formation, as determined by MAG1 antibody seroreactivity (r = 0.96; P < 0.0001), which is a serologic marker for the cyst burden. The presence of MAG1 antibody response, but not T. gondii IgG response, was required for NMDAR autoantibody production. The pathogenic relevance of NMDAR autoantibodies to behavioral abnormalities (blunted response to amphetamine-triggered activity and decreased locomotor activity and exploration) and reduced expression of synaptic proteins (the GLUN2B subtype of NMDAR and PSD-95) has been demonstrated in infected mice. Our study suggests that NMDAR autoantibodies are specifically induced by persistent T. gondii infection and are most likely triggered by tissue cysts. NMDAR autoantibody seroreactivity may be a novel pathological hallmark of chronic toxoplasmosis, which raises questions about NMDAR hypofunction and neurodegeneration in the infected brain.
INTRODUCTION
Toxoplasma gondii infection is known to cause significant brain and behavioral abnormalities in humans and rodents. The immune response to the parasite is regarded as an important mechanism underlying these changes (1, 2). Upon exposure to the obligate intracellular parasite, rapidly replicating tachyzoites infect a broad spectrum of host cells. Under the pressure of innate and adaptive immune responses, tachyzoites convert into slow-replicating bradyzoites, a semidormant stage that mainly exists as quiescent intracellular cysts in the brain for the lifetime of the host. Persistence of tissue cysts requires a continuous immune response provided by resident central nervous system (CNS) and/or infiltrating peripheral immune cells to prevent cyst reactivation and toxoplasmic encephalitis (3, 4). Consequently, low-grade inflammation persists throughout the brain, as evidenced by microglia and astrocyte activation, an increase in complement C1q, and ventricular dilatation (1, 5–7). The continuing inflammation can cause synaptic and neuronal loss, leading to disruption of brain connectivity and behavioral deficits (1, 8).
N-Methyl-d-aspartate receptors (NMDARs) are ligand-gated ion channels abundantly expressed in the mammalian brain. The receptors are heteromers of GLUN1 subunits that bind glycine and GLUN2 subunits (A, B, C, and D) that bind glutamate. NMDARs are pivotal in synaptic transmission and CNS plasticity. NMDAR overactivity is a proposed underlying mechanism for epilepsy, dementia, and stroke, whereas decreased NMDAR activity is associated with symptoms of schizophrenia (9, 10). The underlying causes of NMDAR dysfunction are unclear, but a particular type of antibody against NMDAR has been postulated as a potential mechanism. These antibodies are reported to decrease the availability of NMDAR by internalizing the receptor (11). Moreover, a previous study reported a link between memory and behavioral deficits and antibody-mediated reduction of NMDAR (12). The origin of NMDAR antibodies was initially associated with the presence of ovarian teratomas (13). However, subsequent studies have found NMDAR autoantibodies in many cases of autoimmune encephalitis and brain disorders, including schizophrenia (14, 15).
Emerging clinical observations suggest that infectious agents, such as herpesvirus, may trigger NMDAR autoantibodies (16). Interestingly, recent studies have reported that NMDAR autoantibodies are present in chronic murine toxoplasmosis (6, 17). This raises the question of whether T. gondii is the origin of the autoantibodies. Since both T. gondii and NMDAR autoantibodies are implicated in the development of brain disorders, understanding their relationship can provide insight into how the parasite affects the brain. By using mouse models of T. gondii infection, we sought to characterize the generation of NMDAR autoantibodies and their pathogenic effects. Our results suggest that NMDAR autoantibodies are most likely triggered by T. gondii tissue cysts. The pathogenic potential of NMDAR autoantibodies has been related to behavioral abnormalities and synaptic loss in infected mice. These findings add an autoimmune mechanism to the host immune response against T. gondii and may represent a new pathological hallmark in chronic toxoplasmosis.
RESULTS
Immunoreactivity against NMDAR occurs early in infection and correlates with the dynamics of tissue cysts.We previously established a model of chronic T. gondii infection using strain GT1 in CD-1 mice. The model has the advantage of generating varying degrees of cyst burden and facilitates studies on the role of tissue cysts (18). We investigated the kinetics of NMDAR autoantibody production in relation to the cyst burden using the model. Serum samples obtained at biweekly intervals for 18 weeks following infection were tested. Among this cohort (n = 10), we found consistently that half of the mice had antibodies against NMDAR while the other half lacked detectable anti-NMDAR antibodies (optical densities [ODs], <0.1). The cutoff for seropositivity was defined as an OD value that differed significantly from that of a negative control supplied by the manufacturer or that of uninfected control mice. The kinetics of anti-NMDAR antibody synthesis in the 5 mice capable of producing antibodies are shown in Fig. 1A. Antibodies were detectable in circulation by 2 weeks postinfection, increased gradually over the next 2 months, and remained high until the end of the experiment. In an earlier study (17), the time course of NMDAR antibody generation in mice with T. gondii type II infection showed a similar pattern.
Immunoreactivity against NMDAR correlates with kinetics of tissue cyst generation. (A) Similar time courses of the development of serum levels of MAG1 and NMDAR antibodies. Serum samples were tested at biweekly intervals for 18 weeks in mice (n = 5) following infection with T. gondii strain GT1. Shown are means ± SEM. The data for the anti-MAG1 antibody response are from a previous publication (18). (B) Correlation analysis between levels of NMDAR and MAG1 antibodies over the 18-week time period.
As demonstrated previously (18), all the mice from this cohort generated high levels of anti-T. gondii IgG antibodies. However, only half of the mice from the cohort developed a MAG1 antibody response, which is a serological marker for the cyst burden, and all had a MAG1-high response, as defined in Materials and Methods. Interestingly, we found that the same mice developed NMDAR autoantibodies and MAG1 antibodies. Employing previously published MAG1 data (18), we compared the kinetics of NMDAR and MAG1 antibody generation. As shown in Fig. 1A, the kinetics of the response were similar between serum levels of NMDAR and MAG1 antibodies. We further performed a correlation analysis among the levels of the two antibodies over the 18-week period. Higher levels of NMDAR antibodies were correlated with higher levels of MAG1 antibodies in Spearman's correlation analysis (r = 0.96; P < 0.0001) (Fig. 1B). Our data suggest that the production of anti-NMDAR antibodies is closely associated with tissue cyst formation.
Infection, but not immunization, with T. gondii induces immunoreactivity against NMDAR.Employing UV-inactivated parasites (strain Prugniaud [PRU]) that were able to stimulate a T. gondii-specific immune response in the absence of cyst production in mouse brains, we investigated whether T. gondii cysts residing in the brain were required for NMDAR antibody production. Serum samples from BALB/c mice infected with either live or inactivated parasites were assayed. We first examined humoral responses against the T. gondii whole organism and the MAG1 peptides (Fig. 2), with the intent to determine if infection was established. Both live (n = 13) and inactivated (n = 14) parasites elicited strong serological anti-T. gondii IgG responses in mice (OD, 3.12 ± 0.044 and 2.17 ± 0.12, respectively; the following are all the same OD level), although the response was more robust in the live group than in the inactivated group (P < 0.0001). MAG1 antibody was detected in the sera of all the mice given live parasites (0.91 ± 0.204; n = 13 [9 were MAG1 high; 4 were MAG1 low]), but not in mice given inactivated parasites (0.04 ± 0.0002; n = 14), confirming their lack of cyst formation.
Cyst-forming T. gondii infection of the brain is required for NMDAR autoantibody production. BALB/c mice infected with live or inactivated T. gondii organisms (strain PRU) had elevated T. gondii antibodies at 5 months postinfection. However, only infection with the live organisms resulted in development of both NMDAR and MAG1 antibodies. Live infection, n = 13; inactivated infection, n = 14. ***, P < 0.001.
We then measured the levels of NMDAR autoantibodies. In mice inoculated with live parasites, we detected elevated levels of NMDAR autoantibody (0.18 ± 0.0125; n = 13). In contrast, mice inoculated with inactivated parasites were seronegative for NMDAR antibodies (0.062 ± 0.026; n = 14). Neither MAG1 nor NMDAR antibodies were present in the control group (n = 10) (data not shown). Compared to controls, seroreactivity in the live group was increased 3-fold. Our data indicated that tissue cysts residing in the brain are required for the generation of NMDAR autoantibodies, although in this model, no correlation was noted between levels of NMDAR antibody and MAG1 antibody (P = 0.249). In addition, there was no significant difference in the levels of NMDAR antibody between the MAG1-high and MAG1-low groups (mean antibody levels, 0.18 ± 0.016 versus 0.19 ± 0.021, respectively; P = 0.69).
NMDAR and MAG1 antibodies are not cross-reactive.Because the antibody responses to MAG1 and NMDAR were observed in the same mouse and with similar kinetics, we investigated the possibility of antibody cross-reactivity. We pooled sera from three mice that had produced high levels of NMDAR autoantibody and preabsorbed these antibodies with MAG1 peptides. As a positive control, we also preabsorbed the NMDAR antibodies with protein A agarose. As shown in Fig. 3, preincubation of sera with protein A agarose completely removed the NMDAR antibody signal detected in the enzyme-linked immunosorbent assay (ELISA), and they could be recovered following elution (pH 2.8), readjusting the pH to 8.0, and reassaying the recovery solution. When sera were preincubated with different MAG1 peptides (MAG1_4 and MAG1_5), NMDAR antibody activity was not quenched. In contrast, we found a dilution-dependent reduction of NMDAR antibody activity regardless of the presence or absence of MAG1 peptides. Our data indicate that NMDAR antibodies were distinct from MAG1 antibodies.
Absence of cross-reactivity between NMDAR and MAG1 antibodies. Serial dilutions of pooled sera (n = 3) were incubated with protein A agarose and MAG1_4 and MAG1_5 peptides. The sera were then tested for anti-NMDAR antibody using an ELISA kit. Protein A was employed as a positive control at a serum dilution of 1:100. Antibody was recovered by incubation with eluting buffer. The experiment was repeated three times using different pooled sera with similar results. Shown are means plus SEM.
Levels of NMDAR antibody are correlated with behavioral changes in a mouse model of chronic toxoplasmosis.NMDAR autoantibodies in humans are associated with autoimmune encephalitis, which variably consists of psychosis, epilepsy, cognitive decline, and extrapyramidal symptoms (14). T. gondii infection is known to induce behavioral changes in rodents. We therefore investigated whether the infection-induced behavioral abnormalities were related to the infection-induced NMDAR autoantibodies. In our previous study (18), we observed several behavioral changes in the mouse model of chronic T. gondii type I infection. Most of the changes were directly correlated with levels of MAG1 antibody. Mice with high MAG1 antibody levels (OD ≥ 0.5) displayed reduced locomotor and exploratory activity, impaired object recognition memory, and lack of response to amphetamine-induced activity. The changes were not found in mice with low MAG1 antibody levels (OD < 0.5) or mice that lacked detectable MAG1 antibody. Employing sera collected upon the completion of behavioral testing of this mouse cohort, we measured NMDAR autoantibodies to explore their behavioral relevance. We found that sera from uninfected control mice and infected, MAG1-seronegative mice did not contain NMDAR autoantibodies (0.091 ± 0.011; n = 24). However, NMDAR autoantibodies were detected in MAG1-seropositive mice, although the levels varied greatly (0.426 ± 0.103; n = 21). When the mice were stratified based on MAG1 antibody levels, as previously described (18), we found that NMDAR autoantibody levels differed significantly between mice categorized as MAG1 high and MAG1 low (mean antibody levels, 0.69 ± 0.19 versus 0.19 ± 0.022, respectively; P = 0.012). Compared to controls, seroreactivity in the MAG1-high group was increased 7-fold and that in the MAG1-low group 2-fold. We then examined the relationship between levels of NMDAR and MAG1 antibodies. Spearman's correlation analysis revealed a positive correlation between them (r = 0.70; P = 0.0005; n = 21).
We then investigated whether NMDAR autoantibodies are related to behavioral changes. Since there were 4 behavioral outcomes, we first performed principal-component analysis (PCA) to reduce the number of variables. We found that the first component contained 52% of the variance, and it was considered the principal component (PC) (Table 1). The Spearman correlation analysis between the first component and anti-NMDAR antibodies revealed a significant association (r = −0.6361; P = 0.0026; n = 21). Having established that NMDAR antibodies correlated with the behaviors, we further explored the relevance of the antibodies to each behavior index. As depicted in Fig. 4, higher levels of NMDAR autoantibody were correlated with lower locomotor activity, calculated using the number of beams broken (r = −0.51; P = 0.018); lower exploration, assessed by the number of rears (r = −0.50; P = 0.021); and lower amphetamine-triggered locomotor activity, recorded as the total number of beams broken within a 60-min period (r = −0.77; P < 0.0001). There was no association of NMDAR autoantibodies with recognition memory, assessed after a 24-h delay using a preference score (r = −0.182; P = 0.443) (Fig. 4).
Eigenvectors, eigenvalues, proportion, and accumulated variation of the four components from the correlation matrix based on rankings of the four behavioral outcomes
Correlation of NMDAR autoantibodies and behavioral abnormalities in mice (n = 21) with chronic toxoplasmosis. NMDAR antibody levels (OD) were inversely correlated with unstimulated locomotor activity (A), rears in novelty-induced activity (B), and amphetamine-triggered locomotor activity (D), but not the recognition index (C), in cognitive performance after a 24-h delay. Behavioral data were from a previous publication (18).
Antibody-associated reduction of NMDA receptors.Since NMDAR autoantibodies decrease the availability of receptors through cross-linking and internalization (11), we examined the expression of NMDAR in T. gondii-infected mice. Brains of mock-infected (NMDAR-seronegative) and GT1-infected (NMDAR-seropositive) mice were stained for the GLUN2B subunit of the NMDAR. We also stained for postsynaptic density protein 95 (PSD-95), a major anchoring and scaffolding protein associated with the receptor. As shown in Fig. 5, there was loss of GLUN2B and PSD-95 immunoreactivity in the prefrontal cortical area in infected mice compared to mock-infected controls. The prefrontal cortex is a brain region that contains high densities of NMDAR.
Decreased intensity of immunostaining for GLUN2B and PSD-95 in the prefrontal cortical area in T. gondii GT1-infected mice that were NMDAR seropositive. Scale bars, 200 μm. The staining was repeated in two other mice with a similar staining pattern.
As another measure of protein expression, Western blotting was used to quantify GLUN2B and PSD-95 levels in the prefrontal cortex. As shown in Fig. 6A and B, infected mice (n = 10) had reduced expression of GLUN2B (−32%) and PSD-95 (−42%) compared to mock-infected controls (n = 5). We examined the relationship between levels of protein expression and NMDAR autoantibodies. As depicted in Fig. 6C and D, higher levels of NMDAR autoantibody were correlated with lower expression of GLUN2B (r = −0.70; P = 0.031) and PSD-95 (r = −0.70; P = 0.031) in Spearman's correlation analysis.
Reduced expression of synaptically related proteins in brains of T. gondii GT1-infected mice that were NMDAR seropositive. (A) Western blot detection of NMDAR subunit GLUN2B and PSD-95 in cortical extracts from control (n = 5) and infected (n = 10) mice at 5 months postinfection, alongside β-actin loading controls. (B) Quantification of normalized levels of GLUN2B and PSD-95 between control and infected mice. (C and D) Protein expression of GLUN2B and PSD-95 had negative correlations with NMDAR autoantibodies. The error bars represent SEM. *, P < 0.05.
DISCUSSION
T. gondii persists in the brain in the form of intracellular tissue cysts that are controlled but not eliminated by the immune system. The recognition that T. gondii persists in neurons (19, 20) raises concerns about the potential effect on neuronal structure and function. NMDAR is a major neuronal glutamate receptor subtype that is critical for learning and memory. The presence of NMDAR autoantibodies decreases the availability of NMDA receptors via antibody-mediated receptor capping and internalization (11). The occurrence of NMDAR autoantibodies in response to infection prompted us to investigate whether T. gondii is the origin of these antibodies. In the present study, our results showed that tissue cysts, the hallmark of chronic T. gondii infection, are the most likely trigger for NMDAR autoantibodies. The pathogenic relevance of NMDAR autoantibodies has been related to behavioral abnormalities and reduced expression of synaptic proteins in infected mice. NMDAR autoantibody seroreactivity may be a novel pathological hallmark of chronic toxoplasmosis.
Our study suggests that NMDAR autoantibodies are specifically precipitated by persistent infection with T. gondii tissue cysts. We found that NMDAR autoantibody generation has a temporal association with cyst formation, as determined by MAG1 antibody seroreactivity. The presence of a MAG1 antibody response, but not an IgG response to whole T. gondii organisms, was required for NMDAR antibody production. In spite of the high degree of association, we found no cross-reactivity between NMDAR and MAG1 antibodies. Therefore, our results showed that tissue cysts residing in the brain most likely trigger the generation of NMDAR autoantibodies. The increase of NMDAR antibodies is unlikely to be a generalized immune response to infection, since we found no difference in the concentrations of total IgG in sera from mice displaying different antibody profiles (e.g., presence or absence of IgG antibodies to T. gondii, MAG1, and NMDAR) (data not shown).
One mechanism through which autoreactive immune components are generated is molecular mimicry. There is evidence to suggest that pathogens might be involved in human CNS autoimmune diseases via molecular mimicry (21, 22). A recent finding suggested that nodding syndrome, a severe and puzzling form of epilepsy, is an autoimmune reaction to the parasitic worm Onchocerca volvulus (23). Given its dependency on tissue cysts, NMDAR autoantibodies may be triggered through molecular mimicry between a non-MAG1 tissue cyst-derived antigen and an antigenic region of the NMDAR, as suggested recently (24). The hypothesis is also supported by the fact that mice designated MAG1 low have negligible levels of NMDAR antibodies, presumably due to an insufficient level of cyst-derived antigens needed to stimulate the autoimmune response.
In mice with chronic T. gondii infection, NMDAR autoantibody levels showed relevance to behavioral abnormalities, such as decreased locomotor activity, reduced exploration, and blunted response to amphetamine-triggered activity. Previous studies have found that administration of NMDA receptor antagonists effectively prevented or attenuated a variety of amphetamine-induced responses, such as behavioral sensitization, neuronal degeneration, and gene expression (25–28). These results suggest that activation of the NMDA receptor is required for amphetamine-induced neuropathological responses. Our findings of a blunted response to amphetamine, the presence of NMDAR autoantibodies, and decreased expression of NMDA receptors are consistent with this assumption. Nevertheless, the altered response to amphetamine may also be due to other mechanisms. For example, the interaction of amphetamine with dopamine transporter (DAT), which induces dopamine overflow into the synaptic cleft, is thought to mediate the acute behavioral effects of the psychostimulant (29). In the current study, there was no correlation between impairment of object recognition memory and levels of NMDAR autoantibodies. We proposed previously that this cognitive deficit is not related to persistent T. gondii infection (18). Consistent with our hypothesis, the cyst-dependent NMDAR autoantibodies were not correlated with cognitive deficit. Since the pathological role of NMDAR autoantibodies has been established (11, 12), our findings of a correlation between the NMDAR antibody level and behavioral abnormalities provide a potential mechanistic explanation for the behavioral abnormalities (18). The sample size (n = 21) of this study is not adequate to distinguish independent effects of NMDAR and MAG1 antibodies on behaviors. Future studies, perhaps with antibody or cellular transfer experiments, are needed to elucidate their respective roles.
Consistent with the pathogenic role of NMDAR autoantibodies, we detected reduced expression of the LUN2B subtype of NMDA receptors in infected mice, which was positively correlated with levels of NMDAR autoantibodies. Converging evidence supports a significant disruption in glutamate signaling following infection, as shown by (i) a significant reduction in astrocytic glutamate transporter (GLT-1) (30); (ii) an increase in extracellular glutamate (30); (iii) an elevated level of kynurenic acid (KYNA), an antagonist of the NMDAR (31); (iv) and changes in the distribution of glutamic acid decarboxylase 67 (GAD67), the enzyme responsible for converting glutamate to GABA (32). Our findings add to the growing evidence that NMDAR hypofunction is involved in the pathogenesis of murine toxoplasmosis and potentially leads to changes in brain activity and connectivity. Moreover, NMDA receptors are linked to PSD-95 at synaptic sites, whose expression was found to be decreased in mice with T. gondii infection (8). Similarly, we found a reduced level of PSD-95 in the prefrontal cortex that was positively correlated with levels of NMDAR autoantibodies.
The parasite T. gondii has a clonal population structure, and the strain types display differences in virulence and clinical presentation (33). Using two different strains of T. gondii, we demonstrated the generality of NMDAR autoantibody response across strains (types I and II), irrespective of host strain genetics (outbred and inbred) and gender. In the mouse model of type I infection, the cyst burden seems to be an important factor influencing the level of NMDAR autoantibodies, given that they have positive correlations. Interestingly, the GT1 strain does not readily convert to bradyzoites, and the overall cyst burden was relatively low (∼200 cysts per brain). However, in the mouse model of type II infection, we found no correlation between the cyst burden and levels of NMDAR autoantibody. Type II infection leads to a low level of NMDAR autoantibodies (3-fold higher than that for uninfected controls), despite the fact that type II is the more cyst-competent strain. These results suggest that the parasite genotype plays a role in determining the level of NMDAR autoantibodies.
In conclusion, our study shows that NMDAR autoantibodies are produced in chronic T. gondii infection, suggesting a secondary autoimmune mechanism is involved in the pathogenesis of chronic murine toxoplasmosis. Our finding of NMDAR autoantibodies provides an additional mechanism to explain changes in glutamate signaling following T. gondii infection. The pathogenic potential of NMDAR antibodies raises concerns about NMDAR hypofunction and neurodegeneration in brains of T. gondii-infected hosts.
MATERIALS AND METHODS
Chronic models of T. gondii infection.Samples (sera and tissues) for the present study were collected as part of previously published or ongoing studies in which mice were infected with T. gondii (18, 34), as described below.
Chronic T. gondii type I infection.As a model of chronic T. gondii type I infection, female outbred CD-1 mice (7- to 9-week-old adults) were infected with 500 tachyzoites of the GT1 strain, as previously described (18). Control mice received vehicle only (phosphate-buffered saline [PBS]). To control tachyzoite proliferation during the acute stage, both infected and control mice were treated with sulfadiazine sodium in drinking water (400 mg/liter) from days 5 to 30. This model of infection generates varying degrees of cyst burden. In one cohort of mice (n = 10), serum samples were collected weekly from the tail vein between 2 weeks and 18 weeks following infection. A second cohort of mice (n = 46) was evaluated for behavioral phenotypes between 12 and 21 weeks postinfection. Behavioral abnormalities were observed in novelty-induced activity in the open field, novel-object recognition, and amphetamine-induced activity in the open field. Upon the completion of behavior testing, mice were sacrificed (5 months postinfection) and sera were collected. Brains were harvested and divided sagittally. Half of the brain was used for immunofluorescence staining, and the other half was used for dissection of the prefrontal cortex.
Chronic T. gondii type II infection.As a model of T. gondii type II infection, mice were infected with 400 live tachyzoites of the PRU strain, as previously described (34). As a model of immunization with inactivated T. gondii, mice were administered 300,000 UV-inactivated tachyzoites of the PRU strain with an adjuvant, also as previously described (35). These experiments employed inbred male BALB/c mice, which were 4.5 weeks old when infected. Serum samples for both groups were collected at approximately 5 months postinfection. As demonstrated previously (35), the inactivation procedure completely inhibited parasite replication and cyst formation in vitro and in vivo. However, the mice developed a robust anti-T. gondii immune response.
Serum antibody measurements.Serum antibodies to T. gondii, NMDAR, and MAG1 were measured (dilutions at 1:100, 1:50, and 1:100, respectively) via ELISA. Total anti-T. gondii IgG was measured using a modified commercial ELISA kit, as previously described (36), with a negative control and calibrators included in each run. Antibody directed against the cyst antigen MAG1 was measured using our previously developed MAG1 peptide ELISA (MAG1_4 and MAG1_5) (18, 37). The MAG1 antibody level was reported for either MAG1_4 or MAG1_5, depending on which one was greater. We used the median value of MAG1 antibody absorbance (0.5) as the cutoff designation to stratify mice into high and low MAG1 (MAG1-high and MAG1-low) groups, as previously described (5). Antibodies to NMDAR were measured with a Gold Dot GLUN2 antibody test kit (CIS Biotech, Inc.). The assay is designed for the quantitative determination of antibodies to the GLUN2 subunit of NMDAR. Positive and negative controls and manufacturer-supplied calibrators were included in each run.
Correlations between optical density and serum dilution.For antibodies against NMDAR and MAG1, we determined the correlation between the optical density and the serum dilution. As shown in Fig. 7 for representative sera, there was a linear relationship between the log10 OD value and the serum dilution for both antibodies, demonstrating that the serum dilution we used falls within the linear range of the assay.
Effects of serum dilution on absorbance values of MAG1 (A) and NMDAR (B) antibodies (Ab) in representative sera.
Peptide-blocking assay.Serial dilutions of sera were incubated with the respective MAG1 peptide (MAG1_4 or MAG1_5; 8 μg/ml) for 1 h at 37°C. The sera were then analyzed for anti-NMDAR antibody using ELISA. As a positive control, sera (1:100 dilution) were incubated with protein A agarose (2.25 mg; 10001D; Thermo Fisher Scientific) according to the manufacturer's instructions. Subsequent antibody recovery was done by incubation of the agarose gel with elution buffer (pH 2.8). The experiment was repeated three times using different pooled sera.
Immunoblot analyses.Total protein was extracted from the prefrontal cortices of mice, as described previously (38). In brief, prefrontal cortex was homogenized in radioimmunoprecipitation assay (RIPA) buffer (Sigma, St. Louis, MO, USA) containing protease inhibitors, sonicated for 5 min at 4°C, and centrifuged for 5 min at 10,000 × g. Proteins were probed with primary antibodies for GLUN2B (1:1,000; catalog no. MA1-2014; monoclonal antibodies; Thermo Fisher) and PSD-95 (1:1,000; catalog no. ab18258; polyclonal antibodies; Abcam). Bands were visualized using enhanced chemiluminescence (ECL Prime Western blotting detection reagent; GE Healthcare Life Sciences). Protein values were normalized for corresponding values of β-actin. Relative optical density was assessed using Scanalytics image analysis software (Bio-Rad, Hercules, CA, USA).
Immunofluorescence with brain tissue.Immunohistochemistry for all antibodies was performed using 5-μm-thick slides and standard protocols. Paraffin-embedded sections (3 mice per group; 5 sections per mouse) were treated with xylene and a graduated alcohol series and rinsed in distilled water. Antigen retrieval was performed by boiling in antigen unmasking solutions (Vector Laboratories; H-3300) for 30 min. The slides were blocked with 10% normal donkey serum for 30 min, followed by 0.1 mg/ml Fab fragment donkey anti-mouse IgG for 1 h (Jackson ImmunoResearch; 715-007-003). The slides were stained overnight using mouse monoclonal antibodies against GLUN2B (1: 250; catalog no. ab93610) and rabbit polyclonal antibodies against PSD-95 (1:200; catalog no. ab18258) from Abcam. Secondary antibodies were purchased from Life Technologies, Inc. (Frederick, MD, USA). Images were visualized using an Olympus BX41 microscope and a reflected-fluorescence system.
Statistical analyses.Data are presented as means and standard errors of the mean (SEM). Serum samples from the behavioral cohort were tested twice for MAG1 and NMDAR antibodies on 2 separate days to calculate the coefficients of variation (CV). For MAG1 antibodies, the minimum, maximum, and median CV were 1.0%, 60%, and 12%, respectively. For NMDAR antibodies, the range of the CV was 3.4% to 28%, with a median CV of 18%.
Differences between two groups were analyzed by Student's t test. Correlation analysis between NMDAR antibodies and measured parameters (behaviors and protein expression levels) was performed using a two-tailed Spearman's correlation coefficient (r). We recorded four behavioral outcomes, as previously described (18). To avoid spurious associations due to multiple testing, we first conducted a PCA for behaviors to reduce the number of variables. Before performing PCA, we generated rankings for each behavioral measure, since the scaling differed among behaviors. Correlation analysis was performed between the principal component and anti-NMDAR antibodies. Having established that NMDAR antibodies correlated with behavior, we further explored associations with each of the four behavioral indices. Due to the exploratory nature of the study with limited numbers of samples, Spearman corrections were not made for the possible effects of multiple comparisons. Statistical analyses were conducted in Graph-Pad Prism V7.03. Significance was considered a P value of <0.05.
ACKNOWLEDGMENT
This work was supported by the Stanley Medical Research Institute.
FOOTNOTES
- Received 21 May 2018.
- Accepted 13 July 2018.
- Accepted manuscript posted online 23 July 2018.
REFERENCES
- Copyright © 2018 American Society for Microbiology.
