Authors: Leah Baxter (1Department of Neurobiology, University of California San Diego.), Steven Hopkins (2Department of Neurology, University of Texas Southwestern Medical Center.), Kevin C. O’Connor (5Department of Neurology, Yale University School of Medicine.; 6Department of Immunobiology, Yale University School of Medicine), Minh C. Pham (5Department of Neurology, Yale University School of Medicine.), Richard J. Nowak (6Department of Immunobiology, Yale University School of Medicine), Nancy L. Monson (2Department of Neurology, University of Texas Southwestern Medical Center.), Kyle Blackburn (2Department of Neurology, University of Texas Southwestern Medical Center.), Ryan E. Hibbs (1Department of Neurobiology, University of California San Diego.; 3Department of Neuroscience, University of Texas Southwestern Medical Center.; 4Department of Pharmacology, University of California San Diego.), Steven Vernino (2Department of Neurology, University of Texas Southwestern Medical Center.), Colleen M. Noviello (1Department of Neurobiology, University of California San Diego.)
Categories: Article
Source: Journal of neuroimmunology
Authors: Leah Baxter, Steven Hopkins, Kevin C. O’Connor, Minh C. Pham, Richard J. Nowak, Nancy L. Monson, Kyle Blackburn, Ryan E. Hibbs, Steven Vernino, Colleen M. Noviello
Autoimmune autonomic ganglionopathy (AAG) is a rare disease wherein autoantibodies target the ganglionic acetylcholine receptor (gAChR). Current diagnosis in the United States depends upon clinical symptoms and positive autoantibody detection using a radioimmunoprecipitation assay (RIA). Here we offer a proof-of-principle study on an alternative method, fluorescence-detection size-exclusion-chromatography (FSEC). We show FSEC can detect autoantibodies against gAChR from patient sera but not healthy controls or samples from other autoimmune diseases. We compare FSEC to RIA and find good correlation. We discuss potential advantages of using FSEC as an alternative or as a first-step diagnostic prior to pursuing existing methodologies.
Autoimmune autonomic ganglionopathy (AAG) is caused by pathogenic autoantibodies that interfere with neurotransmission in autonomic ganglia. At these synapses, the α3β4 nicotinic acetylcholine receptor (gAChR) is the principal postsynaptic receptor. Autoantibodies against the gAChR are detected in 50–60% of AAG cases (Vernino et al., 2000). Patients with AAG present with severe autonomic failure; characteristic features include orthostatic hypotension, constipation, urinary retention, loss of sweating, and impaired pupillary reflexes. These problems can be life-threatening, but most patients improve with immunotherapies aimed at reducing the levels of pathogenic autoantibodies.
Mechanisms which autoantibodies can utilize against neuronal receptors to induce pathology generally fall into one, or a combination, of three 1) blocking of neurotransmitter binding, 2) immunomodulation via cross-linking induced internalization, and 3) complement fixation. Blocking antibodies in AAG have been detected (Vernino et al., 2000), as have immunomodulation-inducing antibodies (Urriola et al., 2021, Wang et al., 2007). While complement fixation as a mechanism of AAG has not been measured in humans, it is a reasonable assumption that it also contributes to pathology as many of the antibodies from patients are of the IgG1 serotype (Karagiorgou et al., 2022).
Current detection of anti-gAChR antibodies from patients uses radioimmunoprecipitation of solubilized receptors from cell lysates from either transfected cells (Karagiorgou et al., 2022) or the neuroblastoma IMR-32 cell line (Vernino et al., 2000). These receptors are labeled with I^125^ epibatidine, an agonist which occupies the neurotransmitter binding site and may preclude detection of blocking autoantibodies (Vernino et al., 2000). More recently, other techniques have been developed to detect autoantibodies in AAG patients. These include cell-based immunofluorescence microscopy assays, flow cytometry cell-based assays, and luciferase immunoprecipitation assays. Of these methods, only the live cell-based assay (Karagiorgou et al., 2022) could definitively capture all three mechanisms of autoantibody pathology. Unfortunately, many clinical laboratories do not have tissue-culture facilities available to grow live cells or access to sophisticated flow cytometry instrumentation. In this study we developed an alternate method, fluorescence-detection size exclusion chromatography (FSEC), for the detection of autoantibodies from AAG patient sera. In this method, the receptor is fused in an intracellular loop to a fluorescent protein. Binding of antibodies can access any of the extracellular domains on the receptor, and is not blocked by occupation of the neurotransmitter site, nor dependent upon internalization of receptor or fixation of complement. By reducing the fluorescent signal of the receptor, this method can quantitatively detect autoantibodies that act through any of the three mechanisms. We show it has excellent discrimination between healthy and AAG serum samples.
Patient sample collection was approved by the Institutional Review Board at UT Southwestern Medical Center (IRB protocols 092004–041 and 012011–182). With informed consent, serum was collected from patients with autoimmune autonomic ganglionopathy, suspected GABAA encephalitis, suspected NMDA receptor encephalitis, postural orthostatic tachycardia syndrome (POTS), or healthy controls. Serum samples were collected and processed through the Neuroscience Biorepository at University of Texas Southwestern Medical Center using previously established protocols (Estrada et al., 2018, Li et al., 2021). Patient data was stored and organized via REDCap (Research Electronic Data Capture) (Harris et al., 2009). Some of the serum samples used in this study were included in previous serological studies (Vernino et al., 2008, Vernino et al., 2000, Vernino et al., 1998), and some individual patients have been described in published case reports (Gibbons et al., 2012, Schroeder et al., 2005, Stopschinski et al., 2022). For the purposes of this study, AAG was defined by typical clinical symptoms by clinicians and the presence of autoantibodies against the gAChR was determined by RIA prior to selection for screening by FSEC. Suspected autoimmune encephalitis (AE) patients were identified clinically by the authors. Deidentified myasthenia gravis (MG) serum samples, from individuals with laboratory confirmed AChR autoantibody serostatus, were retrieved from a biorepository established at the Yale University School of Medicine under the approval of Yale University’s Institutional Review Board. Patient demographics are described in Table 1.
The RIA for a subset of gAChR patient autoantibodies was performed as previously reported due to concerns about long-term storage affecting antibody titer (Vernino et al., 2008, Vernino et al., 2000). Briefly, solubilized membranes derived from IMR-32 neuroblastoma cells were incubated with ^125^I-radiolabeled epibatidine, a high affinity agonist of nicotinic acetylcholine receptors. After mixing with patient serum, samples were incubated overnight at 4°C. Precipitation with anti-human serum from goat was measured for radioactivity in comparison to negative control samples. Results are measured in pmol I^125^-epibatidine/L serum.
Genes encoding the α3GFP β4 nicotinic acetylcholine receptor in the pEZT plasmid were described previously (Gharpure et al., 2019). The gene for the α7 nicotinic acetylcholine receptor containing a deletion in the intracellular domain has also been described previously (Noviello et al., 2021), however yellow-fluorescent protein is in place of the thermostable apocytochrome B (BRIL) protein. Transient transfections of HEK 293 GnTI^−^ adherent cells (ATCC CRL-3022) (Reeves et al., 2002) were performed using Lipofectamine 2000 as per manufacturer’s protocols. Cells were grown in Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal bovine serum. Cells were incubated for 72 hours at 30°C before harvesting. To produce the α3GFPβ4 nicotinic acetylcholine receptor lysate, GnT1^−^ cells were grown in suspension and transduced with bacmam virus generated from the pEZT plasmids (Morales-Perez et al., 2016). Suspension cells were transduced with titered bacmam virus at an MOI of 1 for each subunit and supplemented with 3mM sodium butyrate to enhance expression. For the α7YFP nicotinic acetylcholine receptor lysate, the pEZT plasmid containing α7YFP (Noviello et al., 2021) was used to generate bacmam virus as for α3GFPβ4. This plasmid was co-tranduced with the pEZT-nAChO construct to aid in expression of this protein (Noviello et al., 2021, Matta et al., 2017). Cells were kept at 37°C instead of moving to 30°C as for α3GFPβ4 (Noviello et al., 2021).
Cells were transiently transfected or transduced with pEZT- α3GFP and pEZT-β4 plasmids to express the gAChR. They were harvested by centrifugation incubations indicated above. Resuspended cells were solubilized with 40 mM dodecyl-maltoside (DDM, Anatrace) in TBS (20 mM Tris pH 7.4, 150 mM NaCl) supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF). This method preserves the pentameric integrity of the receptor (Gharpure et al., 2019, Kim et al., 2020, Noviello et al., 2021, Zhu et al., 2018). Lysate was rocked for 40 minutes at 4 °C. Insoluble material was pelleted by ultracentrifugation for 40 minutes at 71,000g at 4 °C. Supernatant was quickly transferred to a fresh tube and kept on ice for immediate analysis via FSEC before aliquoting and storing at −20 °C. In the case of the GABAA receptor, lysate did not perform well after freezing and thawing and therefore was used immediately after preparation for each experiment. For the gAChR, the frozen FSEC lysate is stable for at least 2 years, but is not stable upon repeated freeze/thaw cycles.
To perform the FSEC assay, 5 μl of patient serum were mixed with 40 μl of lysate and incubated on ice for 1 h. Three separate tubes of serum plus lysate were generated for each patient. After precipitation of large aggregates by ultracentrifugation at 21,130 g for 15 minutes, supernatant was transferred into fresh tubes. Twenty μl of lysate and serum were injected over an SRT SEC-500 column (Sepax Technologies) with 1 mM DDM/TBS as the mobile phase. FSEC trace fluorescence at the predetermined pentameric elution time of ~8.3 minutes was measured at excitation/emission wavelengths of 475 nm and 510 nm for GFP-tagged proteins. We determined the amount of receptor fluorescence able to detect patient autoantibodies to be around 5,000 μV of fluorescence. In each FSEC run we include untransfected cell lysate, which we subtract from the pentameric peak fluorescence at an elution of time of ~8.6 minutes on an SRT SEC-500 column run at 0.35 ml/minute. We also optimized incubation time, injection volume, and excitation/emission wavelengths to reduce the time the assay requires, and increase the signal-to-noise ratio. Our laboratory has an anti-gAChR monoclonal antibody developed for structural biology (Gharpure et al., 2019). We use this antibody as an assay positive control at a final concentration of 0.003 µg/ml diluted in commercially available pooled human serum (Sigma). This concentration was determined to be equivalent to a lower antibody titer in myasthenia gravis patients (Leeman et al., 2018, Mane-Damas et al., 2022). At this concentration, our laboratory-generated antibody reproducibly decreases the fluoresecent signal of the pentamer at the standardized elution time by 74%, well beyond a decision limit threshold.
With each experimental run we include receptor plus pooled commercial serum as our negative control, and receptor plus our specific anti-gAChR antibody as our assay positive control. As an additional control we run untransfected cell lysate and subtract its signal at the pentameric elution time. That is then used as the denominator in determining fraction of reduction caused by patient autoantibodies. The fraction is subtracted from 1, then multiplied by 100 to obtain the percent reduction (See equation below) PercentReduction=100∗1−Testsamplepentamerfluorescence−emptylysateControlhumanserapentamerfluorescence−emptylysate
Decision limits on significant receptor reduction were determined by a combination of two factors. First, by taking the average of the healthy control samples (−8.5% reduction) and adding four times the standard deviation to the mean, (4*8.04, or +32.16%) a minimum reduction of 23.5% was reached. Next, we examined the related disease controls. Of the twenty myasthenia gravis patients and four GABAA autoimmune encephalitis patients, their percent reduction values ranged from 2.5%-18.5% reduction. gAChR FSEC assay results between AAG patients and healthy controls were compared using a one-way ANOVA with Sidak’s multiple comparisons test. The myasthenia gravis group reduced gAChR pentamer in a statistically significant fashion when reduction was compared using one-way ANOVA analysis. Indeed, one patient reduced the pentamer peak by 18.5%. Nonetheless, lower levels of gAChR autoantibodies have been detected in other autoimmune diseases, including MG (Vernino et al., 2008, Vernino et al., 2001, Peltier et al., 2010, Balestra et al., 2000). Thus, after confirming the MG patient with the 18.5% reduction exhibited no autonomic dysfunction by examining their medical record, we feel confident that the decision limit of 23.5% reduction is appropriate. Sequence identity between the extracellular domain of the muscle AChR and gAChR subunits was performed with the human reference sequences obtained from GenBank and analyzed with the Sequence Manipulation Site Ident and Sim function(Stothard, 2000).
Correlation between FSEC and RIA was measured using the Spearman test for nonparametric data. Receiver operator characteristic curves were determined by comparing AAG patient serum samples (determined by the RIA cutoff of 50 pmol/L) vs all other samples, including non-AAG autoimmune disease controls, and the area under the curve (AUC) was determined. All statistical analyses were performed in GraphPad Prism (Prism v9.5.1).
Data will be made available to any qualified investigator within reason.
In our previous work, we used FSEC to detect antibody binding to neurotransmitter receptors. Evidence of binding in FSEC can be concluded from one of two a shift to an earlier elution volume, indicating an increase in size of the receptor bound to antibodies versus the receptor alone, or if the signal from the receptor decreased, even without an earlier peak. The latter indicates extensive cross-linking and subsequent precipitation. (Althoff et al., 2014, Gharpure et al., 2019, Hibbs and Gouaux, 2011, Kim et al., 2020, Morales-Perez et al., 2016, Noviello et al., 2022). We extended this technique to patient samples after our recent structural study examined autoimmune encephalitis autoantibody fragments in complex with the GABAA receptor. Our laboratory has also previously determined near-atomic resolution structures of the gAChR (Gharpure et al., 2019), which is targeted in AAG, and the related α7 nicotinic acetylcholine receptor (Noviello et al., 2021, Burke et al., 2024). We utilized our receptor preparations from these works as a starting point for development of the FSEC assay in detection of autoantibodies from human serum.
Our overall methodology is outlined in Figure 1A. Each receptor protein is expressed recombinantly in mammalian cells (Morales-Perez et al., 2016). As these are pentameric receptors, each target protein contained at least one subunit fused to a fluorescent reporter protein. The fluorescent protein facilitates detection via FSEC without purification, and is located in an intracellular domain inaccessible to autoantibodies. Locating the 27-kilodalton fluorescent protein to the intracellular domain may also reduce detection of autoantibodies against this domain, which are of questionable physiological relevance (Koneczny and Herbst, 2019). In size-exclusion chromatography, proteins elute based on their molecular size, and antibody binding shifts this elution time earlier. Cross-linking of receptors by bivalent immunoglobulin autoantibodies can also cause target proteins to precipitate out of solution. In either case, antibody binding to its target results in a decrease in peak amplitude at the normal elution time (Figure 1B, peak underlined 2). To standardize our detection of autoantibodies across experiments, we first performed dose-response experiments using increasing amounts of laboratory generated monoclonal antibodies (Figure 1B) (Gharpure et al., 2019) We set our initial range of antibody concentrations based on the highest antibody titers for the related myasthenia gravis disease (Mane-Damas et al., 2022). We settled upon 0.003 µg/ml of laboratory antibody as a positive assay control (for details, please see Materials and methods).
We optimized the reproducibility of the FSEC assay by examining such variables as excitation and emission wavelengths, ratio of lysate to serum, and incubation times (data not shown). We then generated a decision limit cutoff using healthy controls. We obtained 12 serum samples from healthy controls, and measured the variability across triplicate runs on the pentameric peak fluorescence. We noted that human serum does cause some nonspecific precipitation at the pentameric peak, and used the mean plus four times the standard deviation of these samples to settle on a decision limit of 23.5% reduction (for further discussion please see Materials and methods).
Next we screened AAG patient sera with clinically significant positive RIA (>200 pmol/L) and AAG patient sera with borderline RIA (90–110 pmol/L) for gAChR autoantibodies via FSEC. Of the 15 patients with an RIA >200 pmol/L, all tested positive via FSEC with a range of 25.3%-76.9% reduction. For the five samples with borderline RIA values, all tested negative by FSEC (Figure 2A, “AAG RIA^−^” group). We then tested eighteen patients with a related autonomic disease, postural orthostatic tachycardia syndrome (Figure 2A, “POTS” group). All tested FSEC-which correlates with their RIA values (<50 pmol/L). Furthermore, we screened an additional six patients with suspected gAChR autoantibody involvement who tested negative for both RIA and FSEC.
We extended our screen to patients with autoimmune diseases that are caused by autoantibodies against proteins closely related to the gAChR. The GABAA receptor is part of the Cys-loop receptor superfamily, which also includes the gAChR. In rare cases, autoimmune encephalitis can be caused by autoantibodies against GABAA receptors. We tested two GABAA autoimmune encephalitis patients for autoantibodies against the gAChR. We also tested two patients with autoantibodies against NMDA receptors, which is not in the receptor superfamily (Figure 2A, “AE” group). All four AE patients are negative in the FSEC assay for autoantibodies against gAChR.
A more closely related protein is the nicotinic acetylcholine receptor (AChR) found at the neuromuscular junction, which is targeted in the autoimmune disease myasthenia gravis (MG). We obtained twenty AChR autoantibody positive MG patient samples. Using the decision limit cutoff of 23.5% reduction, none of the MG patients screened via FSEC tested positive for autoantibodies against the gAChR (Figure 2A, “MG” group). There was, however, some lower-level reduction present in the MG cohort. While uncommon, previous studies have noted the coexistence of autoantibodies against gAChR and the muscle-type AChR in MG patients (Peltier et al., 2010, Balestra et al., 2000, Vernino et al., 2008, Vernino et al., 2001). The muscle-type AChR is composed of four subunits in a fixed 2α1, β, γ, and δ. In the extracellular domain, the regions that are targeted by autoantibodies, the sequence similarity between these subunits and the α3 and β4 gAChR subunits is 50.2%-60.3%, with the highest similarity between the α1 and α3 subunits.
The α1 and α3 subunits are both thought to be the main target in each disease. Indeed, the antibody used to detect gAChR in a recent flow-cytometry based assay also binds to the muscle-type AChR, underscoring the overlap in potential cross-reactivity of autoantibodies that affect the two receptors (Urriola et al., 2021, Noridomi et al., 2017). The epitope for that antibody is the main immunogenic region, or MIR. Given the proposed dominance of the MIR in MG (Tzartos et al., 1991), it is surprising that cross-reactive autoantibodies are not more common between the two diseases. However, upon review of the medical history for the patient with the highest (18.5%) reduction of gAChR pentamer, there were no autonomic symptoms noted. Thus, we feel our decision limit cutoff of 23.5% excludes lower levels of gAChR cross-reactive autoantibodies that lack documented clinical significance.
We further compared cross-reactivity by examining whether AAG patients contained autoantibodies cross-reactive for the related α7 nicotinic acetylcholine receptor. Despite containing a similar overall architecture of the MIR, no AAG patient sera reduced the fluorescent signal of the α7 acetylcholine receptor pentamer peak (Figure 2B).
We next performed serial dilutions on the patients who tested positive via FSEC. These are shown in Figure 3A. Correlating the titers of the patient samples, we find that our assay can detect up to 40 pmol/L of I^125^-labeled epibatidine. However, the clinical relevance of this RIA level is questionable. To illustrate this point, we examined how clinical interventions for AAG are reflected in the FSEC assay (Figure 3B). We tested serum from a patient prior to treatment with CellCept and prednisone, and after resolution of symptoms as recorded by the clinician. Prior to treatment, the patient’s serum was tested via RIA and gave a value of 400 pmol/L; this sample reduced the gAChR peak in the FSEC assay by 25.4 %. After treatment, with symptom resolution, the serum RIA level was 90 pmol/L, which is in a range of potentially positive results by some evaluations (Karagiorgou et al., 2022, Vernino et al., 1998). However, the FSEC result was a 6.55% reduction, well below our decision limit cutoff (Figure 3B). Thus, FSEC is capable of discriminating between symptomatic vs. asymptomatic levels of gAChR autoantibodies and we hope to expand this to more patients pre- and post-treatment. RIA values of >200 pmol/L (Li et al., 2015, McKeon et al., 2009) 100 pmol/L (Urriola et al., 2021) or >50 pmol/L (Karagiorgou et al., 2022, Vernino et al., 1998) have all been proposed for thresholds of clinical significance of AAG autoantibodies detected by RIA. Given our data, we agree with the cutoff of 200 pmol/L as clinically significant and confirm the ability of FSEC to discriminate between pathological levels of gAChR autoantibodies and background levels.
Next, we compared our assay to the existing RIA via calculation of the Spearman rho correlation coefficient (Figure 3C). We find a rho of 0.77, with perfect qualitative correlation between the two assays when using the cutoff for RIA+ of 200 pmol/L. The slight quantitative discrepancy in correlation may be related to the presence of autoantibodies in patient serum that bind to the neurotransmitter site. In the FSEC assay, we are able to detect these autoantibodies, but in the RIA the high-affinity agonist I^125^-epibatidine is used to label the gAChR and may reduce binding of autoantibodies that occupy the same site or are conformationally dependent. For example, one patient with an RIA value of 270nM (on the lower end of positive) reduced the FSEC signal by 62.53%, equal to patients with RIA values hundreds of times higher. To further examine the ability of FSEC to discriminate between patients with AAG and without, we generated a receiver-operator characteristic (ROC) curve. In this case, given the ambiguity on how to define AAG+ with relation to the different RIA cutoffs, we included 3 seronegative AAG patients as AAG+. With this strict criterion, our area under the curve was 0.88 (Figure 4), indicating excellent sensitivity and specificity for AAG+ samples.
We adapted the standard biochemistry FSEC assay to detect autoantibodies directly from patient serum. We demonstrate feasibility for the α3β4 ganglionic acetylcholine receptor, the primary target in AAG.
The detection of autoantibodies via RIA enabled clinical diagnosis of numerous autoimmune diseases including AAG. Since then, the difficulty in maintaining radioactive licenses, combined with their comparative danger, inspired other methods to detect autoantibodies. For autoantibodies against the gAChR, these include immunoprecipitation of luciferase-conjugated solubilized protein, cell-based flow cytometry, and immunofluorescence staining of live cells (Urriola et al., 2021, Karagiorgou et al., 2022). There are limitations to each method that inspired us to explore FSEC as an alternative. Radioligands are not readily available for many neuronal receptors, and may also have high background signal or be overly sensitive, detecting autoantibodies of questionable relevance to disease pathology (Karagiorgou et al., 2022). Furthermore, RIA for AAG and related diseases such as myasthenia gravis uses radioactive ligands that bind to the neurotransmitter site, which can preclude binding by autoantibodies via direct competition or by conformational changes induced by the ligand binding (Drachman et al., 1982, Noviello et al., 2022, Crisp et al., 2019). Live cell-based assays require a tissue culture facility on site, as well as microscopy for qualitative assessments or a flow cytometer for quantitative determination. However, they do ensure an appropriate lipidic environment for these transmembrane proteins. In our assay we use a fluorescently labeled, frozen cell lysate which can be stored at −20°C, the temperature of many clinical laboratory freezers. This lysate can be prepared in a central laboratory with tissue culture facilities and then shared with other laboratories that do not have live cell culture. Furthermore, the signal of receptor can be standardized across batches of lysate, ensuring that the assay is performed with the same ratio of receptor across patients and between sites. We acknowledge the initial adoption of an HPLC coupled to a fluorescent detector is certainly a hurdle. We hope this negative will be offset by the advantage of FSEC not needing specialized licenses or rooms (as radiation or tissue culture do).
Furthermore, by measuring only receptor signals at the predetermined pentameric elution time, we ensure we are detecting antibody effects on appropriately assembled receptors. While we have removed the receptor from its native lipid environment, we have previously demonstrated that gAChR purified via our method maintains ion channel function (Gharpure et al., 2019). We also acknowledge the limit of sensitivity of our fluorescence-based assay versus the radioisotope-based method. Nonetheless, this may be a strength as we have demonstrated the ability to detect AAG in 100% of symptomatic patients, and not in asymptomatic patients. Further confirmation of this by expanding the assay to a broader pool of patients is desirable but feasibility of this work is limited by the rarity of the disease.
While our findings broaden the utility of FSEC from beyond biochemistry to the medical field, we acknowledge there are limitations to this technique. FSEC is less sensitive than RIA, and possibly less sensitive than the recently described CBA. To improve sensitivity, we are exploring newer variants of fluorescent protein fusions that will enable a lower serum ratio. Most clinical laboratories do not have a high-pressure liquid chromatography system connected to a fluorescence detector, and each patient sample, run in triplicate with controls, takes up to four hours total. Scaling up is limited to 30 patients per day (most of the time is in preparation of the samples, not the actual FSEC runs). However, it is our hope that this proof-of-principle study demonstrates the feasibility of this approach to other diseases, encouraging adoption of FSEC as a tool for detecting autoantibodies.