Autoimmune gastrointestinal dysmotility (AGID) is a limited dysautonomia, occurring idiopathically and in paraneoplastic context.1–3 Onset is typically subacute, with early satiety, abdominal pain, nausea, vomiting, weight loss, and slow transit. Autoimmune serology aids the diagnosis. Autoantibodies specific for intracellular antigens are surrogate markers of peptide-specific T cell-mediated effector mechanisms. Antineuronal nuclear-type 1 autoantibody (ANNA-1/anti-Hu) and collapsin response-mediator protein antibody [(CRMP)-5]-IgG both predict underlying lung carcinoma or thymoma.1,4 Autoantibodies specific for extracellular domains of plasma membrane antigens have pathogenic potential.2,3,5 Symptoms improve with antibody-depleting therapies (plasma exchange, B lymphocyte-depleting-IgG, or high-dose intravenous immune globulin).5–7 Specificities include ganglionic-type (α3*) and muscle-type (α1*) nicotinic acetylcholine receptor (nAChR) and neuronal voltage-gated calcium channel (N-type) or potassium channel complex.5,8 Most frequently identified is α3*-nAChR–IgG, the only autoantibody presently proven to cause AGID.5,9 Rabbits immunized with human α3-polypeptide (extracellular residues 1–205) readily produce autoantibody, and AGID signs parallel serum α3*-nAChR-IgG levels.10 Immune serum IgG (rabbit or human) transfers transient gastrointestinal hypomotility to mice.9 We now report AGID induction by actively immunizing mice.
Background Autoimmune gastrointestinal dysmotility (AGID) is a limited form of dysautonomia. The only proven effector to date is IgG specific for ganglionic nicotinic-acetylcholine receptors containing α3 subunits [α3*- nicotinic acetylcholine receptor (nAChR)]. Rabbits immunized with recombinant α3-polypeptide produce α3*-nAChR autoantibodies, and profound AGID ensues. Human and rabbit α3*-nAChR-specific-IgGs induce transient hypomotility when injected into mice. Here, we describe success and problems encountered inducing gastrointestinal hypomotility in mice by active immunization.
Methods We repeatedly injected young adult mice of seven different strains susceptible to autoimmunity (spontaneous diabetes or neural antigen immunization-induced myasthenia gravis or encephalomyelitis) with: (i) α3-polypeptide, intradermally or (ii) live α3*-nAChR-expressing xenogeneic cells, intraperitoneally. We measured serum α3*-nAChR-IgG twice monthly, and terminally assessed blue dye gastrointestinal transit, total small intestinal α3*-nAChR content (radiochemically) and myenteric plexus neuron numbers (immunohistochemically, ileal–jejunal whole-mount preparations).
Key Results Standard cutaneous inoculation with α3-polypeptide was minimally immunogenic, regardless of dose. Intraperitoneally injected live cells were potently immunogenic. Self-reactive α3*-nAChR-IgG was induced only by rodent immunogen; small intestinal transit slowing and enteric α3*-nAChR loss required high serum levels. Ganglionic neurons were not lost.
Conclusions & Inferences Autoimmune gastrointestinal dysmotility is inducible in mice by active immunization. Accompanying enteric α3*-nAChR reduction without neuronal death is consistent with an IgG-mediated rather than T cell-mediated pathogenesis, as is improvement of symptoms in patients receiving antibody-depleting therapies.
Materials and methods
Cell lines and α3-nAChR fusion protein
Human neuroblastoma cells IMR-32 [American Type Culture Collection (ATCC)] and SHSY-5Y (Memorial Sloan Kettering, Cancer Center, New York, NY, USA) were maintained in Gibco minimal essential medium with 10% fetal bovine serum (FBS) and 1% nonessential amino acids and Eagle MEM/Ham F12 medium (1 : 1, supplemented with 15% FBS). Human embryonic kidney cells (HEK-293, ATCC) were maintained in Dulbecco MEM (Gibco, Grand Island, NY, USA) with 10% bovine calf serum. The stably transfected HEK-293 line KXα3ß4R2 was derived and maintained as described.11 Recombinant human α3-polypeptide (residues 1–205 fused to GST) was produced from cDNA derived from IMR-32.10
Native α3*-nAChR preparation
Postnuclear membrane fractions (isolated by differential centrifugation of homogenized IMR-32 cells grown subcutaneously in athymic nude mice and from cultured KXα3ß4R2 cells) were extracted in buffer (10 mmol L−1 Na3PO4, 1 m mol L−1 EDTA, 100 m mol L−1 NaCl, 0.02% NaN3, pH 7.4) containing 2% Triton X-100. Supernates were clarified centrifugally (100 000 g, 30 min).
Mouse serum (5 μL duplicate specimens; serial 10-fold dilutions as necessary) was mixed with solubilized α3*-nAChR (50 fmol complexed with 10-fold excess 125I-epibatidine). Antibodies were detected by radioimmunoprecipitation.10,12,13
The Mayo Institutional Animal Care and Use Committee approved the experiments. Mouse strains susceptible to autoimmunity (spontaneous diabetes mellitus,14 or neural antigen-induced myasthenia gravis15 or encephalomyelitis16) were purchased from Jackson Laboratories (Bar Harbor, ME, USA), acclimated 10 days in a conventional facility, anesthetized with 4% isoflurane/4% O2 and injected with fusion protein emulsified in complete Freund’s adjuvant (i.d., six sites), and with killed B. pertussis organisms (20 billion, s.c., four sites). Live cells were injected i.p., 10 million in 0.25 mL PBS, initially with killed B. pertussis organisms, s.c., as adjuvant. Mice were weighed weekly and bled alternate weekly.
Small intestinal transit
Brilliant Blue ethyl-[4-[[4-[ethyl-[(3-sulfophenyl)methyl]amino]phenyl]-(2-sulfophenyl)methylidene]-1-cyclohexa-2,5-dienylidene] -[(3-sulfophenyl)methyl]azanium (100 μL) was gavaged via 18-gauge needle. Mice were killed 30 min later (CO2 asphyxiation); small intestine was removed (gastroduodenal to ileocecal junction). Transit (distance of dye front from gastroduodenal junction) was expressed as percent total small intestinal length: 100% when dye reached ileocecal junction; 0% when dye remained in stomach.
Neuronal density quantification
Myenteric plexus neurons (ANNA-1-immunoreactive perikarya)17 were enumerated in coded photomicrographs of muscularis propria whole-mount preparations (1 cm of ileo–jejunal junction tissue; Fig. 1B). Images from four corners and center of each (five randomly picked fields of 0.59 mm2) were captured using Olympus AX70 fluorescence microscope (Center Valley, PA, USA), 20× objective. Automated object counting (SvCell; Svision LLC, Bellevue, WA, USA, employing proprietary soft matching algorithms to segment ANNA-1 positive somata) was validated on a randomly chosen set of 143 images. Numbers obtained from manual counts correlated significantly with numbers obtained by the SvCell program (Fig. 1C). Significance between α3-nAChR-immune and control immune groups was compared by t-test.
Small intestinal α3*-nAChR extraction
After luminal flushing with Kreb’s solution (mmol L−1: 137.4 Na+, 5.9 K+, 2.5 Ca2+, 1.2 Mg2+, 134 C1−, 15.5 HCO3−, 1.2 HPO42−, and 11.5 glucose, saturated with 97% O2 and 3% CO2), tissue was weighed and nAChR was extracted as above.
Cell and tissue extracts were incubated with 125I-epibatidine, with and without non-radioactive epibatidine in 10-fold excess, and quantified using glass fiber filtration.18
Despite repeated intradermal injections of recombinant human α3-polypeptide 1–205 and adjuvants, serum autoantibody detection was infrequent (Table 1), and levels lower than 3.00 nmol L−1 (predicts severe AGID in immunized rabbits).10 No gastrointestinal dysmotility signs were evident.
|Mouse strain||Immunogen†||n||Serum α3*-nAChR-IgG values (nmol L−1)|
|Frequency (%)||Median (range)||Frequency (%)||Median (range)|
|C57BL/6J||Human α3, residues 1–205||6||3 (50)||0.07 (0.05–0.12)||2 (33)||0.07 (0.04; 0.10)|
|AKR||Human α3, residues 1–205||7||0||–||0||–|
|SJL||Human α3, residues 1–205||12||1 (8)||0.05||3 (25)||0.07 (0.05–0.10)|
|129S/J||Human α3, residues 1–205||8||0||–||nt||nt|
|NOD||Human α3, residues 1–205||38||5 (13)||0.13 (0.11–0.20)||9 (24)||0.09 (0.03–0.16)|
|NOD-DQ8||Human α3, residues 1–205||7||7 (100)||0.17 (0.07–0.24)||2 (29)||0.04 (0.03; 0.06)|
|ABoDQ8||Human α3, residues 1–205||11||0||–||0||–|
|C57BL/6J||Human neuroblastoma cells (SHSY-5Y)||18||18 (100)||3.40 (1.20–10.0)||0||–|
|C57BL/6J||Rat α3ß4-transfected HEK-293 cells||32||29 (91)||0.41 (0.05–2.40)||32 (100)||14.0 (3.90–22.0)|
|C57BL/6J||Non-transfected HEK-293 cells||24||0||–||0||–|
Live cell immunization
We next investigated the immunogenicity of SHSY-5Y neuroblastoma cells expressing endogenous α3*-nAChR. Three weeks after injection, α3*-nAChR-IgG was detected in 50% of C57BL/6J mice; 100% were seropositive after reinoculating on day 28. Levels peaked after a third inoculation (day 120), far exceeding serum levels of mice injected with α3-polypeptide (Table 1). Mice receiving human kidney cells (HEK-293) intraperitoneally were seronegative. Despite high α3*-nAChR-IgG levels, no mouse had signs of AGID. Small intestinal transit (80 ± 11%) was not significantly different from control values (67 ± 12%; Fig. 2A). The discontinuous dye distribution pattern in the small intestine of both groups indicated normal peristalsis.
Next, we investigated the immunogenicity of KXα3ß4R2 cells (stably expressing rat α3 and ß4 nAChR subunits, and 5-fold more α3*-nAChR protein than neuroblastoma cells; Fig. 2B). After three intraperitoneal inoculations, 29 of 32 mice (91%) were α3*-nAChR-IgG-seropositive; levels were lower than in mice receiving neuroblastoma cells (Table 1). However, terminal tissue harvest revealed pooling of gavaged dye in the proximal intestine of 14 mice (45%), suggesting loss of coordinated peristalsis (Fig. 1A). Furthermore, despite lower frequency and levels of α3*-nAChR-IgG, small intestinal transit was significantly slowed (49 ± 16%; Fig. 2A).
Severe hypomotility despite relatively modest α3*-nAChR-IgG levels raised the possibility that antibody induced by rodent α3*-nAChR immunization was more reactive with rodent antigen than with human. Specific testing revealed serum IgG reactive with rodent α3*-nAChR in 100% of mice immunized with rat α3*-nAChR-expressing cells, but in no mouse immunized with human α3*-nAChR-expressing cells (Table 1). Rat and mouse α3-nAChR subunit amino acid sequences are identical, but human α3-nAChR subunit differs in 10 of 205 extracellular domain residues accessible to circulating IgG. These data implicate autoantibody as the cause of slowed small intestinal transit and loss of coordinated peristalsis.
To evaluate histometric evidence of neuronal cytotoxicity, we counted neurons in whole-mount preparations of jejuno–ileal junction tissue (Fig. 1B). Neuron numbers did not differ significantly in enteric ganglia of mice immunized with human α3*-nAChR, rodent α3*-nAChR, or non-transfected HEK-293 cells (Fig. 2C; respectively, 100 ± 19 neurons mm−2, 102 ± 14 neurons mm−2, and 113 ± 16 neurons mm−2). Thus, neuronal cytotoxicity did not explain the hypomotility.
Total small intestinal α3*-nAChR content (γ-emission from Triton-X100-solubilized membrane macromolecules ligated with 125I-epibatidine and captured on glass fiber filters) was 85% less in tissues of mice immunized with rodent α3*-nAChR-expressing cells than in tissues of mice immunized with human α3*-nAChR-expressing cells or with non-transfected cells (respectively, 61 ± 41 fmol g−1, 409 ± 173 fmol g−1 and 384 ± 91 fmol g−1; Fig. 2D).
‘Self’-α3*-nAChR-reactive autoantibody production is required to develop AGID in actively immunized mice. Profound intestinal α3*-nAChR loss without loss of enteric neurons indicates that hypomotility was not attributable to neuronal death. The most plausible mechanism is α3*-nAChR internalization and degradation following its cross-linking in neuronal plasma membranes by circulating IgG. This mechanism underlies α1*-nAChR loss from muscle cells exposed to serum from rodents with experimental myasthenia gravis19 and partial aquaporin-4 water channel loss from astrocytes exposed to serum from patients with neuromyelitis optica.20α3*-nAChR internalization is compatible with: (i) reduction of agonist-induced depolarization of α3*-nAChR-expressing cells following application of human autoantibody,21 (ii) transient gastrointestinal hypomotility in mice injected with α3*-nAChR-IgG9, and (iii) rapid patient recovery from AGID symptoms following antibody-depleting therapies.7,22 Reproducible induction of a mouse model of AGID by active immunization will enable detailed analysis of genetic susceptibility determinants, pathogenic and immunoregulatory mechanisms, and implementation of novel preclinical therapeutic trials.
We thank Tara Adams, Stacy Hall, Christine Hachfeld, and James Thoreson for excellent technical assistance.
This work was supported by the National Institutes of Health (DK71209; DK68055).
The authors disclose no conflicts. Serological testing for neural autoantibodies is offered on a service basis by Mayo Collaborative Service, Inc., an agency of Mayo Foundation. Neither Dr. Lennon nor her laboratory benefit financially from this testing.
JWM was involved in study concept and design, acquisition of data, analysis and interpretation of data, drafting of manuscript; KEH did acquisition of data, analysis and interpretation of data; JPF designed the study and provided technical support; TJK was involved in study concept and design, analysis and interpretation of data; SJG was involved in study design and technical support; YX provided material support and critically revised the manuscript; VAL was involved in study concept and design, analysis and interpretation of data, drafting of manuscript, critical revision of the manuscript for important intellectual content, obtained funding and study supervision.