Development and Application of a Cell-Based Assay for Detecting Anti-Agrin Antibodies Associated With Myasthenia Gravis

Abstract

Background and Purpose

Anti-agrin antibodies (agrin Abs) have recently been identified in patients with myasthenia gravis (MG), sometimes in conjunction with antibodies (Abs) to the acetylcholine receptor (AChR), muscle-specific tyrosine kinase (MuSK), or low-density lipoprotein receptor-related protein 4. This study aimed to develop an in-house cell-based assay (CBA) for detecting agrin Abs, and to test its application to serum samples collected from individuals diagnosed with MG.

Methods

Agrin complementary DNA as cloned into a pCMV6-AC-GFP vector, which was subsequently transfected into human embryonic kidney 293T (HEK293T) cells. Transfected HEK293T cells were incubated with patient serum and antihuman immunoglobulin G Ab conjugated with a red fluorescent dye. Agrin Ab levels were measured using the CBA in 389 serum samples: 340 from patients with MG, 36 from patients with other neuromuscular diseases, and 13 from healthy controls. The presence of agrin Ab was determined based on the fluorescence intensity and colocalization using fluorescence microscopy.

Results

The expression levels of agrin mRNA and protein in transfected HEK293T cells were confirmed using the reverse-transcription polymerase chain reaction and Western blotting, respectively. Agrin expression in cells was further confirmed by immunocytochemistry. Two (0.6%) of the 340 patients with MG tested positive for agrin Ab: 1 of 191 AChR-positive patients and 1 of 54 MuSK-positive patients.

Conclusions

We have developed and validated a novel CBA for detecting agrin Abs. This CBA was successfully applied to detect agrin Abs in serum samples obtained from individuals with MG.

Graphical Abstract

INTRODUCTION

Myasthenia gravis (MG) is an autoantibody-mediated autoimmune disorder that affects the neuromuscular junction (NMJ), and is primarily characterized by fatigability and weakness of the skeletal muscles.¹ The identification of associated autoantibodies is a key step in diagnosing MG that guides subsequent therapeutic strategies. Autoantibodies against the acetylcholine receptor (AChR) or muscle-specific tyrosine kinase (MuSK) are detected in approximately 90% of patients with MG. Some treatment options, including thymectomy and complement inhibitors, are indicated in MG patients with AChR antibodies (Abs) but not in those with MuSK Abs.² Patients with MuSK Abs show poor responses and are commonly intolerant of acetylcholinesterase inhibitors.³,⁴ Autoantibodies against low-density lipoprotein receptor-related protein 4 (LRP4) have reportedly been found in 2%–50% of MG patients without AChR or MuSK Abs.⁵,⁶,⁷,⁸,⁹,¹⁰,¹¹ However, approximately 5% of MG patients show none of these associated autoantibodies.

Agrin is a heparan sulfate proteoglycan that plays a pivotal role in NMJ formation. The neural agrin isoform is expressed in motor neurons and binds to LRP4 to form the agrin/LRP4/MuSK complex, which activates signaling cascades that lead to AChR clustering.¹² Of particular importance is previous studies observing that amino-acid regions in the C-terminus play vital roles in forming the complex that leads to AChR clustering. Genetic defects in this region of the agrin gene can disrupt NMJ function to induce congenital myasthenic syndrome.¹³,¹⁴,¹⁵ Additionally, NMJ formation is significantly impaired in agrin-deficient mice.¹⁶ Anti-agrin Abs (agrin Abs) have been detected in a small proportion of MG patients,⁵,¹⁷,¹⁸,¹⁹,²⁰,²¹,²² suggesting that this is an autoantigen associated with MG. However, detecting agrin Abs is not yet a part of general clinical practice. The resulting inadequate research means that little is known about the characteristics of patients with agrin-Ab-positive MG.

Therefore, the present study aimed to develop a cell-based assay (CBA) for detecting agrin Abs and apply it to serum samples from patients with MG.

METHODS

Patients

This study collected 389 serum samples between October 2016 and October 2023. These included 340 samples from 191, 52, 54, and 43 patients who were AChR-Ab positive, AChR-Ab negative, MuSK-Ab positive, and double seronegative (DSN), respectively. Additionally, 36 serum samples were collected from patients with various neurological diseases: 18 with chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), 7 with Lambert–Eaton myasthenic syndrome, 3 with inflammatory myopathy, 2 with amyotrophic lateral sclerosis, 2 with motor neuron disease, 1 with IgM neuropathy, 1 with paroxysmal nocturnal hemoglobinuria syndrome, 1 with congenital myasthenic syndrome, and 1 with multiple sclerosis. Serum samples were also collected from 13 healthy controls (Table 1).

Table 1. Serum samples analyzed using the cell-based assay to detect anti-agrin antibodies.

Variable			Number of participants (n=389)
Generalized MG			239
	AChR-Ab positive		145
		Early onset	52
		Late onset	23
		Thymoma	70
	AChR-Ab negative; MuSK Ab not checked		13
	MuSK-Ab positive		54
	DSN		27
Ocular MG			101
	AChR-Ab positive		46
	AChR-Ab negative; MuSK Ab not checked		39
	DSN		16
Other disease			36
	ALS		2
	CIDP		18
	LEMS		7
	Multiple sclerosis		1
	IgM neuropathy		1
	PNHE syndrome		1
	Motor neuron disease		2
	Inflammatory myopathy		3
	Congenital myasthenic syndrome		1
Healthy controls			13

Ab, antibody; AChR, acetylcholine receptor; ALS, amyotrophic lateral sclerosis; CIDP, chronic inflammatory demyelinating polyradiculoneuropathy; DSN, double seronegative; LEMS, Lambert–Eaton myasthenic syndrome; MG, myasthenia gravis; MuSK, musclespecific tyrosine kinase; PNHE, paroxysmal nocturnal hemoglobinuria.

All participants provided written informed consent, and this study was approved by the Institutional Review Board of Yonsei University Severance Hospital (No. 4-2021-1328).

Agrin plasmid DNA preparation

Agrin complementary DNA (cDNA) was inserted into the SgfI and MluI sites of the pCMV6-AC-GFP vector (Origene), resulting in the expression of agrin as a tagged protein with C-terminal turbo-GFP (tGFP) in mammalian cells. The vector contained ampicillin and neomycin antibiotic resistance genes for Escherichia coli and cell selection, respectively. Transformation was performed using Dyne DH5α Chemically Competent E. coli cells (version 2, Dyne Bio). The bacteria were selected on plates containing ampicillin, and the selected bacteria were cultured in LB medium. The ampicillin-resistant bacteria containing the agrin plasmid each formed a single colony. Agrin plasmids were subsequently isolated from the incubated LB medium using the Plasmid Maxi kit (Invitrogen).

Agrin expression in HEK293T cells by transfection

Human embryonic kidney 293T (HEK293T) cells (Korean Cell Line Bank) were cultured in DMEM (Cytiva) supplemented with 10% FBS (R&D Systems) and 1% penicillin (Gibco) in a 24-well cell culture dish (SPL Life Sciences). One day later, the cell culture medium was replaced with DMEM (Cytiva) without 1% penicillin (Gibco) or 10% FBS (R&D Systems), and the cells were cultured in a 37℃ incubator for 2 h. The pCMV6-Agrin-tGFP vector or pCMV6-AC-GFP vector (the empty vector) was then transfected into HEK293T cells using TransIT-2020 (Mirus Bio Corporation) and Opti-MEM (Gibco). Transfection was performed following the basic protocol provided by the manufacturer for the TransIT-2020 reagent (Mirus Bio Corporation). Transfected cells were cultured in a cell culture incubator at 37℃ and 5% CO₂. After 24 h, the culture medium was replaced with DMEM (Cytiva) containing 1% penicillin (Gibco) and 10% FBS (R&D Systems).

Confirmation of agrin mRNA and protein by RT-PCR and Western blotting

The expression levels of agrin mRNA and protein in transfected HEK293T cells were confirmed using the reverse-transcription polymerase chain reaction (RT-PCR) and Western blotting, respectively, with the following primer pairs: agrin, forward 5′-GTGGCACATAGGGAGCAGAG-3′ and reverse 5′-ACAAAGCCTGTGCCGTAGG-3′; and glyceraldehyde 3-phosphate dehydrogenase (GAPDH), forward 5′-ATGGC ACCGTCAAGGCTGAGA-3′ and reverse 5′-GGTATGG ACTGTGGTCATGAG-3′. HEK293T cells transfected with empty vectors and untransfected HEK293T cells were used as negative controls. Each sequence was selected through sequence comparison and sequence adjustment, and retrieved from the GenBank database of the National Center for Biotechnology Information. Annealing was performed at 62℃, with 30 reaction cycles.

Western blotting was performed using rabbit polyclonal agrin Ab (diluted 1:500; GeneTex) and β-actin Ab (diluted 1:2,000, Thermo Fisher Scientific). Secondary horseradish-peroxidase-conjugated goat antirabbit immunoglobulin G (IgG) Ab (Thermo Fisher Scientific) was added. An enhanced chemiluminescence method was used to detect immunocomplexes.

Immunocytochemistry and CBA

HEK293T cells were grown on 12-mm glass coverslips (Marienfeld Superior) coated with poly-L-lysine (Sigma-Aldrich) in 24-well cell culture plates. The cells were grown in DMEM (Cytiva) containing 10% FBS (R&D Systems) and 1% penicillin (Gibco). When the cells reached 80% confluency after 1 day, pCMV6-Agrin-tGFP was transfected using TransIT-2020, and 48 h later immunofluorescence staining was applied to the agrin-transfected HEK293T cells. The DMEM medium was then removed, and the cells were washed with DMEM containing 2% HEPES (Tech&Innovation). Cells were fixed in 4% paraformaldehyde (Tech&Innovation) dissolved in Dulbecco’s phosphate-buffered saline (DPBS, Gibco) for 15 min at room temperature, and then washed with DPBS. To promote strong Ab binding, the cells were permeabilized for 6 min using 0.5% TritonX-100 (Sigma-Aldrich) dissolved in DPBS. After again washing with DPBS, the cells were blocked with 5% BSA in 1× TBST for 1 h at room temperature. The cells were finally incubated with agrin Abs (diluted 1:20, Thermo Fisher Scientific) in DMEM containing 5% BSA and 2% HEPES for 1 h at room temperature, after which the primary Ab solution was removed. The cells were then incubated with goat antirabbit IgG H&L conjugated to Alexa Fluor 594 (diluted 1:750, Abcam) in DMEM containing 5% BSA and 2% HEPES for 1 h at room temperature in the dark. After washing with DPBS, the coverslips were removed from the 24-well cell culture dish and placed on a microscope slide with VECTASHIELD antifade mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories). The slides were stored in the dark at 4℃ before being observed using fluorescence microscopy (Zeiss).

CBAs were applied to the serum samples according to established immunocytochemistry protocols. Goat antihuman IgG (H+L) conjugated with AF594 (1:750 dilution, Invitrogen) was used to detect agrin Abs in sera. The results were evaluated by fluorescence microscopy, with the fluorescence intensity of tGFP-tagged agrin (green fluorescence) and Alexa Fluor 594 (red fluorescence) evaluated using the following visual scoring system: 0, no labeling of transfected cells; 0.5, very weak labeling of a few cells with no definite colocalization; 1, weak labeling of some cells with colocalization; 2, labeling of 20%–50% of cells with accurate colocalization; 3, labeling of 50%–80% of cells with perfect colocalization; and 4, labeling of all transfected cells showing perfect colocalization.⁸,²³,²⁴ A score of 1 or higher was considered to indicate a positive result. The results were determined by the consensus of independent analyses performed by two investigators (S.H.K. and M.K.) who were blinded to the clinical data.

RESULTS

Agrin gene transfection into HEK293T cells

HEK293T cells were transfected with agrin cDNA to overexpress the agrin protein. Green fluorescence was emitted by cells expressing agrin tagged with tGFP but not by untransfected HEK293T cells, for which only DAPI²⁵ was detected in the nucleus. Green fluorescence was detected when tGFP-tagged agrin were transfected with HEK293T cells (Fig. 1A).

Fig. 1. Expression of agrin in agrin-transfected HEK293T cells. A: Untransfected HEK293T cells showed no green fluorescence, with nuclei stained by DAPI. Green fluorescence was observed in HEK293T cells transfected with a plasmid containing GFP without agrin, confirming GFP expression. Green fluorescence was also observed when tGFP-tagged agrin plasmid was transfected, indicating the expression of the tGFP-tagged agrin protein. B: HEK293T cells transfected with agrin were incubated with a commercially available polyclonal agrin Ab and an anti-immunoglobulin G antibody conjugated with Alexa Fluor 594. The colocalization of red and green fluorescence in the transfected HEK293T cells indicates binding of the agrin Abs to agrin. Ab, antibody; agrin Ab, anti-agrin Ab; DAPI, 4′,6-diamidino-2-phenylindole; GFP, green fluorescent protein; HEK293T, human embryonic kidney 293T; tGFP, turbo-GFP.

The binding between agrin and agrin Abs in agrin-transfected HEK293T cells was evaluated by staining cells with agrin Abs and secondary anti-IgG conjugated with Alexa Fluor 594. Green fluorescence was observed in transfected cells not treated with agrin Abs, in which red fluorescence was not detected. In contrast, colocalization of green and red fluorescence was observed in transfected cells treated with agrin Abs, indicating binding of agrin Abs to the agrin protein (Fig. 1B).

Confirmation of agrin mRNA and protein expression in agrin-transfected HEK293T cells

To confirm successful agrin transfection, the expression levels of agrin mRNA and protein were evaluated using RT-PCR and Western blotting, respectively, in untransfected, empty-vector-transfected, and agrin-transfected cells (Fig. 2). Agrin mRNA expression was confirmed in RT-PCR by the presence of a band at 176 bp, while expression of the GAPDH housekeeping gene was measured in all three groups as a control (Fig. 2A). Agrin protein expression was confirmed by Western blotting. The agrin protein tagged with tGFP was 268 kDa, the agrin protein was 214 kDa, and the tGFP protein was 54 kDa. Western blotting detected 268-kDa bands in the lanes of agrin-transfected cells. Additionally, β-actin expression was detected in all three groups (Fig. 2B).

Fig. 2. Confirmation of agrin mRNA and protein in agrin-transfected HEK293T cells. A: The reverse-transcription polymerase chain reaction confirmed the expression of agrin mRNA (176-bp band) in agrin-transfected HEK293T cells, whereas untransfected HEK293T cells and HEK293T cells transfected with only an empty vector showed no detectable bands. B: Agrin protein expression (268-kDa band) in agrin-transfected HEK293T cells was confirmed by Western blotting. No bands were observed in the lanes of untransfected HEK293T cells or HEK293T cells transfected with an empty vector. Empty vector, HEK293T cells transfected with plasmids that did not contain the agrin gene; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; HEK293T, human embryonic kidney 293T; UT, untransfected cells.

Application of the CBA for detecting agrin Abs associated with MG

Immunocytochemistry was performed to evaluate the presence of autoantibodies against agrin in the patient serausing the developed CBA. Nuclei were stained with DAPI, while agrin protein overexpression was identified by tGFP expression in the cells. Serum samples were stained with an antihuman IgG Ab conjugated to Alexa Fluor 594. The CBA findings were scored based on the proportion of cells with colocalization of green and red fluorescence (Fig. 3). Assessments of the agrin Ab levels using the CBA in 389 serum samples resulted in 2 samples being scored as 2 and 3 (Fig. 3B and C). Additionally, 6 serum samples received a score of 0.5, while the remaining 381 samples were scored as 0. Three of the six serum samples with a score of 0.5 were from patients with AChR-Ab-positive MG: one from a patient with AChR-Ab-negative MG, one from a patient with DSN-MG, and one from a patient with CIDP.

Fig. 3. Cell-based assay for detecting anti-agrin antibodies. A: Agrin-transfected HEK293T cells were incubated with a serum sample obtained from a healthy control followed by an anti-IgG Ab conjugated with Alexa Fluor 594. No red fluorescence was observed, and agrin was not detected. B and C: When serum samples from two patients with myasthenia gravis were incubated with antihuman IgG Ab conjugated with Alexa Fluor 594 in agrin-transfected HEK293T cells, colocalization of red and green fluorescence was observed on the cell membrane. Ab, antibody; DAPI, 4′,6-diamidino-2-phenylindole; GFP, green fluorescent protein; HEK293T, human embryonic kidney 293T; IgG, immunoglobulin G.

Clinical features of the patients with agrin Abs

Patient 1

A 45-year-old female developed diplopia, followed 2 months later by blepharoptosis, difficulty swallowing, and dysarthria. A neurological examination revealed weakness of the neck and masticatory muscles. Repetitive nerve stimulation revealed decreasing responses in the right orbicularis oculi, nasalis, and trapezius muscles. The patient was negative for AChR Abs but positive for MuSK Abs, with significant elevation at 9.80 nmol/L. Chest computed tomography (CT) showed normal results with no thymic abnormalities. The patient was treated with prednisolone and mycophenolate mofetil. Serum samples were collected at 65 years of age, at which time her MG activities of daily living score was 7 despite treatment with 15 mg of prednisolone daily and mycophenolate mofetil. Her main symptoms were dysphagia and respiratory weakness, along with limb weakness.

Patient 2

Patient 2 was a male who developed right eyelid ptosis at the age of 40 years, with limb muscle weakness developing 1 week later. Difficulty swallowing and dysarthria were also observed in a neurological examination, with a quantitative MG score of 15. Repetitive nerve stimulation revealed decreasing responses in the right abductor digiti minimi, flexor carpi ulnaris, trapezius, orbicularis oculi, and nasalis muscles. His initial AChR Ab titer was 10.127 n/M. Chest CT revealed an anterior mediastinal mass, and the patient underwent extended thymectomy via midsternotomy. A histopathological examination revealed type B2 thymoma according to the World Health Organization histological classification,²⁶ and follicular hyperplasia was observed in the adjacent nonneoplastic thymic tissue. The Masaoka-Koga classification of the thymoma was stage I.²⁷ At 2 days after thymectomy the patient experienced a myasthenic crisis and was treated with intravenous immunoglobulin at 400 mg/kg/day for five consecutive days and prednisolone at 60 mg/day. The patient gradually recovered from myasthenic weakness, and the prednisolone dose was tapered. The serum sample analyzed in the present study was collected at the age of 48 years when the patient was in pharmacological remission, at which time his AChR Ab titer was 5.39 n/M.

DISCUSSION

This study has developed a CBA for detecting agrin Abs. Applying this CBA to 389 serum samples identified 2 patients with MG with agrin Abs. Agrin Abs were not detected in any of the included patients with other neurological diseases or in healthy controls.

To the best of our knowledge there have been only seven previous studies on agrin Ab assays in patients with MG,⁵,¹⁷,¹⁸,¹⁹,²⁰,²¹,²² and they varied in terms of the characteristics and numbers of subjects, Ab assay methods, and positivity rates of agrin Ab. The present study detected agrin Abs in the sera of 2 (0.6%) of 340 MG patients, which is comparable to the agrin Ab positivity rate of 0.9% found in a previous study that applied a CBA using full-length agrin, which is similar to our method.²² Notably, the subjects in that previous study were East Asians, as in our study.²² In another recent study using both an enzyme-linked immunosorbent assay (ELISA) and CBA to detect agrin Abs, none of 135 MG patients tested positive for agrin Abs.²⁰ That latter study used a Chinese hamster ovary (CHO)-cell-line-derived human agrin protein (Ala1260 to Pro2045) with an N-terminal 6-His tag in the ELISA, while human mini-agrin cDNA was used in the CBA.²⁰ Five previous studies have previously applied ELISA to measure agrin Abs. One of those studies used commercially available CHO-derived agrin protein, while another study used mini-agrin purified from the supernatants of transiently transfected HEK293T cells. The remaining three studies applied the same ELISA method using Flag/His-tagged rat agrin (His1137 to Pro1940). With the exclusion of the study that used both ELISA and CBA, which found an agrin Ab positivity rate of 0%, the agrin Ab positivity rate ranged from 4% to 14%. This variability in the agrin Ab positivity rate may be attributed to differences in assay methods, such as using CBA versus ELISA.

While ELISA is a practical and high-throughput method that is widely applied to assay human autoantibodies, its low specificity may lead to nonspecific binding and result in false-positive results.²⁸,²⁹ It is also important to consider the possibility of some of the higher agrin Ab positivity rates in previous ELISA studies being caused by nonspecific binding. Nevertheless, these discrepancies in positivity rates cannot be solely attributed to methodological differences between CBA and ELISA. Agrin is a large protein that is expressed in several isoforms. The A/y and B/z sites contain 0 or 4, and 0, 8, 11, or 19 amino-acid inserts, respectively,³⁰ resulting in multiple agrin variants, some of which are secreted while others function as transmembrane proteins. Various forms of agrin have been used as the antigen in different studies,⁵,¹⁷,¹⁸,²⁰,²¹,²² which may also affect positivity rates. This makes it necessary to identify the optimal Ab assay method and agrin protein form for accurately detecting agrin Ab.

The Ab status of a patient plays an important role in diagnosing MG. The absence of AChR and MuSK Abs in approximately 10% of patients with MG, a condition referred to as DSN-MG, presents particular diagnosis difficulties.¹⁹ Therefore, the detection of agrin Abs in DSN-MG has diagnostic significance. The literature contains reports on 84 patients with MG-associated agrin Abs.⁵,¹⁷,¹⁸,¹⁹,²⁰,²¹,²² Of 862 patients with DSN-MG, 49 (5.7%) tested positive for agrin Ab.⁵,¹⁷,¹⁸,¹⁹,²⁰,²¹,²² The remaining 35 patients who were positive for agrin Ab also tested positive for AChR Ab or MuSK Ab.⁵,¹⁷,¹⁸,¹⁹,²² Two of the studies with relatively large numbers of agrin-Ab-positive patients investigated the clinical features of agrin-Ab-positive MG. One study of 181 patients with DSN-MG found that 26 (14.3%) tested positive for agrin Ab, most of whom developed generalized MG and responded to standard MG treatment.²¹ The other study, which involved 1,948 Chinese MG patients, found positivity for agrin Ab in 18 patients, who tended to present with more severe disease activity and association with thymoma, and show favorable responses to a combination of pyridostigmine and prednisone.²² Both of the MG patients in our study who tested positive for agrin Ab were also positive for AChR Ab or MuSK Ab. The small number of agrin-Ab-positive patients with MG meant that it was not possible to further investigate the clinical significance of agrin Ab. In summary, the clinical value of agrin Ab is not yet understood owing to the insufficient numbers of studies and patients associated with the low prevalence of agrin-Ab-positive MG, and so international collaborations may be necessary to accurately identify its clinical features.

This study had several limitations. First, there is no gold-standard assay for detecting agrin Abs that could have been applied to evaluate the performance of the CBA established in this study. Consequently, the possibility of false positives or false negatives cannot be ruled out. It will therefore be important to compare and interpret the results of other assays such as ELISA or radioimmunoprecipitation assays with those of a CBA. Second, this study was conducted at a single center and involved the application of the CBA to a relatively small patient population. The reliability of the results could be improved by including a larger number of patient samples, which would enable more-comprehensive investigations of the clinical features of patients with agrin-Ab-positive MG.

In conclusion, we have developed and validated a novel CBA for identifying agrin Abs associated with MG, and have demonstrated that this CBA can identify agrin Abs in the serum of patients with MG.

Availability of Data and Material

The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.