Neuronal uptake of anti-Hu antibody, but not anti-Ri antibody, leads to cell death in brain slice cultures

Sera from nine patients with paraneoplastic neurological disorders were studied. Seven of these patients had anti-Hu antibodies and two patients had anti-Ri antibodies. The presence of anti-Hu or anti-Ri antibodies and absence of other known paraneoplastic autoantibodies was confirmed in all patients by: 1) immunohistological staining of neurons typical for anti-Hu or anti-Ri antibody in frozen and fixed sections of human and rat cerebellum; 2) antibody binding restricted to the 35-42 kDa proteins characteristic of Hu antigens or the 55 kDa antigen recognized by anti-Ri antibody in Western blots of neuronal lysates; and/or 3) commercial identification of Hu or Ri antibodies (ARUP, University of Utah, USA) [36]. Serum antibody titers by endpoint dilution ranged from 1:320 to 1:100,000. Control serum samples were from individuals without cancer or neurological disease. All materials were obtained and studied under Institutional Review Board guidelines (University of Utah and Veterans Affairs Salt Lake City Health Care System, USA), either under written informed consent or as de-identified patient samples. Purified IgG was prepared from patient sera by column chromatography as previously described [31].

All aspects of animal handling and care were conducted with local Institutional Animal Care and Use Committee approval in an Association for Assessment and Accreditation of Laboratory Animal Care-approved facility and were conducted according to applicable national and international guidelines. Slice cultures were prepared from the two brain regions most commonly affected in patients with anti-Hu or anti-Ri antibody-associated paraneoplastic syndromes: cerebellum, given the association of the two antibodies with cerebellar degeneration and/or syndromes of opsoclonus-ataxia; and hippocampus, given the association of both antibodies with limbic encephalitis. Limited studies were also done employing slice cultures including the cerebral cortex. Sprague-Dawley rats, ages p10-12 or p23-27 days (Charles River, Germantown, MD, USA), were euthanized with carbon dioxide and cerebellar and hippocampal slice cultures prepared at 250-350 μm thickness using the method of Rothstein and colleagues, as previously described [34],[35],[37]. Horse sera (Life Technologies Grand Island, NY, USA) used in tissue culture media were heated to inactivate complement. Cultures were incubated at 37°C in a 5% CO₂/95% humidified air environment with twice weekly changes of medium. Sera and purified IgGs were used at dilutions of 1:400 to approximate amounts of antineuronal antibodies typically found in patient cerebrospinal fluid.

To detect cell death, we incubated cultures for the final 2 hours prior to harvesting with SYTOX cell viability dyes (SYTOX green or SYTOX orange; Invitrogen, Springfield, OR, USA). SYTOX dyes are fluorescent compounds which stain intracellular nucleic acids; these dyes are excluded from living cells but readily enter cells following cell membrane injury and death [38]. We have previously demonstrated the utility of these dyes in detecting dead or dying neurons in brain slice cultures [34],[35]. Detection of antibody-containing neurons undergoing apoptosis employed the pan-caspase marker FLICA (fluorochrome-labeled inhibitors of caspases; carboxyfluorescein-labeled fluoromethyl ketone peptide inhibitor of caspases; Immunochemistry Technologies, LLC, Bloomingdale, MN, USA) [34]. To detect both cell death and apoptosis, cultures incubated with anti-Hu antibodies, anti-Ri antibodies, or control IgGs were treated for the last 4 hours prior to harvesting with a 1:5 dilution of FLICA and, for the last 2 hours prior to harvesting, with SYTOX dyes [39],[40]. Confirmation of apoptosis in cultures exhibiting FLICA staining was carried out in selected replicate cultures using a terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) in situ cell death detection kit (TMR red, Roche Applied Science, Indianapolis, IN, USA). In these experiments, cultures were incubated with 1:400 dilutions of patient or control sera for 72 hours. FLICA was added to cultures 4 hours prior to fixation and SYTOX dyes were added 2 hours prior to harvesting. Cultures were then fixed in 2% paraformaldehyde, permeabilized, and incubated with TUNEL assay mix at 37°C for 2 hours. Uptake of antibody was confirmed by immunofluorescence staining using Cy5-conjugated donkey anti-human IgG, and cell death was confirmed by SYTOX green exclusion as well as by FLICA or TUNEL staining. Positive controls for apoptosis included permeabilized cultures treated with DNase I (Sigma-Aldrich, St Louis, MO, USA) to induce nicks in DNA to allow TUNEL staining. Negative controls included cultures maintained without IgG, with normal IgG, or with omission of conjugated secondary antibody during postfixation staining. Negative TUNEL controls were cerebellar cultures incubated with the TUNEL assay mix without addition of DNase.

To exclude the possibility that detection of intraneuronal IgG might be due to leakage of intravascular Igs into brain parenchyma and neurons after death [41],[42], we studied the interaction of anti-Hu and anti-Ri antibodies with neurons in real time as previously described [34]. IgGs containing high titers of anti-Hu or anti-Ri antibody were purified and conjugated to Cy5 using a DyLight 650 Antibody Labeling Kit (Thermo Scientific/Pierce Biotechnology, Rockford, IL, USA) [34]. Cerebellar and hippocampal cultures incubated with anti-Hu or anti-Ri IgGs conjugated to Cy5 were observed at intervals through 24 hours using a microscope stage incubating chamber (SmartSlide microincubation chamber, WaferGen Biosystems, Fremont, CA, USA) and a Nikon Eclipse TE300 inverted microscope (Nikon Biosciences, Melville, NY, USA.

To confirm that neurons in these cultures were viable and that uptake of Cy5-conjugated anti-Hu or anti-Ri antibody seen in fixed cultures corresponded to stages of antibody uptake observed in real time, SYTOX green was added to each culture, and the cultures were fixed with 2% paraformaldehyde after 6 and 24 hours of incubation and subjected to confocal analysis. To minimize artifactual labeling of cells resulting from diffusion of extracellular IgG, fixed cultures incubated with Cy5-conjugated IgG in these real time studies were not subjected to permeabilization with Triton X-100. Distribution of IgG within neurons was compared with that seen in the same cultures during observation in real time.

Cerebellar and hippocampal cultures were incubated with 1:400 dilutions of anti-Hu or anti-Ri sera or corresponding dilutions of normal sera or purified normal IgG. Cultures were harvested at 24-hour intervals from 24 to 144 hours, with limited additional cultures being maintained for 168 to 219 hours. All studies, unless otherwise indicated, were performed in triplicate. Cultures were treated with FLICA and SYTOX and then fixed in 2% paraformaldehyde, permeabilized with 0.2% Triton X-100, treated with ImageiTFx signal enhancer (Life Technologies, Molecular Probes, Eugene, OR, USA), and incubated overnight at 4°C with 1:400 dilutions of Cy5 (Alexa Fluor 647)-conjugated donkey anti-human IgG (Jackson ImmunoResearch, West Grove, PA, USA) as previously described [35].

Sera from seven patients with anti-Hu antibody response and two patients with anti-Ri antibody response were studied. Cultures were incubated for 72 hours with control, anti-Hu, or anti-Ri sera. FLICA was added 4 hours prior to fixation and SYTOX orange was added 2 hours prior to fixation as previously described [34]. Amounts of apoptosis and cell death as determined by immunofluorescence microscopy were quantified by an observer unaware of treatment of the cultures and consisted of counting the number of cells labeled by Cy5-conjugated anti-human IgG that were stained by SYTOX orange or FLICA or which remained unstained by either agent. Live cells were recorded as containing IgG but without staining by either SYTOX orange or FLICA. Dead cells were scored as cells co-labeled for IgG and SYTOX orange, FLICA, or both. Approximately 40 to 90 cells were counted for each field and the average percentage cell death was obtained from a minimum of eight fields captured at 40× magnification for each time point [35]. Neurons in control cultures did not contain amounts of IgG detectable by immunofluorescence. For purposes of quantification, neurons in these cultures were identified by postfixation immunostaining of cultures with anti-Hu IgG followed by Cy5-conjugated donkey anti-human IgG. Statistical significance between groups was determined by non-parametric Kruskal-Wallace test with Dunn’s multiple comparison test using GraphPad Instat statistical software (GraphPad Software, Inc., La Jolla, CA, USA) [34],[35].

HuD is a recombinant cDNA cloned from a lambda zap library using anti-Hu antibodies and is widely used as a specific antigenic target in testing for anti-Hu antibody response [15]. Despite its use in diagnostic testing, however, the role of antibodies to the HuD protein in the pathogenesis of neuronal death has not been established [15]. To determine whether antibodies reactive with the HuD protein were required for neuronal cytotoxicity, we compared neuronal accumulation of anti-Hu patient IgG and neuronal death before and after the antiserum had been adsorbed to recombinant HuD protein. In these studies, we used anti-Hu IgG which we had shown to produce extensive cell death in both cerebellar and hippocampal cultures. The coding region (1101 nucleotides, 366 amino acids) for the HuD protein (homo sapiens ELAV, embryonic lethal, abnormal vision, Drosophila-like 4, accession number BC036071.1) [15] was sub-cloned into the plasmid pReceiver-B31 that contains a T7 promoter and a multiple cloning site which allows in-frame fusion of six histidine residues (His tag which facilitates binding to a nickel column) to the C-terminus of the protein. The resulting plasmid (EX-Z-0448-B31) was constructed by GeneCopoeia (Rockville, MD, USA) and transformed into the E. Coli strain BL21 (DE3)/pLysS. Post-induction cell pellets were frozen in buffer (NTB- Clontech) containing protease inhibitors, and the cell lysate containing the 40 kDa recombinant HuD-His tag protein was adsorbed onto a nickel affinity column (Clontech column #635657). As a sham control for nonspecific binding, the same treatment was performed with bacteria containing a control vector which lacked the His-tag HuD antigen. The anti-Hu patient IgG (diluted 1:100 in organotypic culture media) was passed over the His-tagged HuD antigen column or over the sham control nickel column treated with E. coli lysate containing the control vector but lacking the His-tag HuD antigen. Eluates from each column were tested by dot blot assays (spotted with recombinant His-tag HuD antigen) and showed that immunoreactivity to the HuD protein was removed when passed over the His-tag HuD antigen nickel column but was unaffected by passing over the column with control vector lacking the His-tag HuD antigen (data not shown).

Confocal analysis employed a Nikon Eclipse E800 upright microscope (Nikon Biosciences) and the Personal Confocal Microscopy PCM-2000 utilizing Argon-ion and green and red HeNe lasers. Simple Personal Confocal Image software program (Compix, Cranberry Township, PA, USA) was used to acquire digital images and image analysis. A red HeNe laser with a 633 nm excitation filter and 675 LP filter was used to visualize Cy5. An Argon-ion laser with a 514 nm excitation filter was used with a 510 LP filter to image SYTOX green. A green HeNe laser with a 543 nm excitation filter and a 565 LP filter was used to visualize SYTOX orange. The argon laser with a 488 nm excitation filter and a 510 LP filter was used to visualize FLICA. All filters were matched to the peak emission spectra of the fluorochromes employed. General procedures utilized individual fluorochromes with X, Y, and Z scans of 14 to 20 focal planes. Identical focal plane settings for each fluorochrome were used for single visual field analysis to ensure that each corresponding fluorochrome was imaged in the same focal plane. In all studies, stringent uniform experimental parameters and computer software settings were maintained for the respective image analyses. Because the vibratome preparation techniques used to prepare cerebellar slice cultures invariably resulted in death of neurons on the cut surfaces of culture slices, image analyses in all experiments were confined to the interior portions of the cultures.

A potential concern in studies of antibody uptake by neurons is that residual extracellular anti-Hu or anti-Ri antibodies present in cultures following incubation might enter neurons during fixation and permeabilization and result in detectable levels of antibody binding to intracellular Hu or Ri antigens [41]. To exclude this possible artifact, cerebellar and hippocampal cultures maintained in a microscope stage incubation chamber were incubated with 1:400 dilutions of either anti-Hu IgG or anti-Ri IgG conjugated to Cy5. This allowed us to examine by confocal microscopy whether IgG containing anti-Hu or anti-Ri antibody was taken up in real time prior to any fixation. In cultures incubated with Cy5-conjugated IgG from anti-Hu positive patients, bright fluorescence was detected in neurons at 6 hours, with significant increase in uptake by 24 hours (Figure 1). Uptake and accumulation of anti-Hu IgG in cerebellar cultures was observed in Purkinje cells, granule cells, and molecular layers. In hippocampal cultures, anti-Hu IgG was found in neurons in all anatomical regions including the pyramidal layer and the dentate gyrus. Similar widespread neuronal uptake of IgG was seen in cultures incubated with anti-Ri IgG (Figure 2). Distribution of Cy5-conjugated anti-Hu or anti-Ri IgG within neurons excluding SYTOX cell death dyes in fixed cultures studied at each time point was identical to that seen in the same cultures studied in real time (Figures 1 and 2). These observations demonstrate that neuronal uptake of anti-Hu or anti-Ri IgG in both cerebellar and hippocampal cultures had occurred prior to fixation while cells were viable and that observations made using fixed cultures reflected those seen in living cells prior to harvesting (Figures 1 and 2).

To assess the intracellular distribution of anti-Hu or anti-Ri antibodies after incorporation by neurons, serial confocal images through individual neurons were obtained after 6, 24 and 72 hours of incubation. In these cultures, anti-Hu and anti-Ri IgG could be detected in both cytoplasm and nuclei of neurons within as early as 6 hours after incubation, with progressive accumulation of antibody over time (Figure 3; Additional file 1). Neurons incorporating IgG excluded SYTOX dyes, indicating that antibody uptake had occurred in living cells. The distribution of intraneuronal IgG in cultures incubated with anti-Hu or anti-Ri antibody was essentially identical to the reported patterns of labeling produced by these two antibodies in conventional immunohistological studies of fixed tissue [1],[4].

To determine whether anti-Hu IgG caused neuronal cell death in cerebellar or hippocampal cultures, we incubated cultures for periods of up to 168 hours with sera from seven patients with anti-Hu antibody response. Uptake of IgG by neurons throughout cerebellar and hippocampal cultures was observed after 6 hours in all cultures incubated with anti-Hu sera, identical to our findings in real-time experiments (data not shown). Neurons in cultures examined after incubation with anti-Hu sera for 24 and 48 hours did not stain with SYTOX cell death dyes above levels seen in control cultures incubated with normal sera. By 72 hours, however, individual neurons in cultures incubated with anti-Hu sera contained SYTOX dyes, indicative of cell membrane disruption and death (Figure 4; Additional file 2). Numbers of dead neurons continued to increase over incubation periods of up to 168 hours.

To determine whether neuronal death associated with anti-Hu antibodies involved apoptosis, cerebellar and hippocampal cultures incubated with anti-Hu antibody for 72 hours were subsequently incubated with FLICA to detect apoptosis and SYTOX orange to detect cell death. Apoptotic cells, as determined by FLICA staining, were detected in both cerebellar and hippocampal cultures incubated with anti-Hu antibody, with some neurons staining with both SYTOX dyes and FLICA (Figure 5). Apoptosis as detected by FLICA staining was confirmed using immunofluorescent TUNEL methods (Figure 6) [34].

Neuronal death was quantified in cultures treated for 72 hours with anti-Hu, anti-Ri, or control sera (see Methods) (Figure 7). Seven anti-Hu sera and two anti-Ri sera were studied. The extent of apoptotic and non-apoptotic cell death in neurons containing IgG, as determined by FLICA staining and by entry of SYTOX dyes, was compared with that seen in normal serum controls as previously described [34]. Hippocampal and cerebellar cultures incubated with control sera exhibited less than 5% background neuronal death (Figure 7). In contrast, although variation in amounts of cell death was seen among anti-Hu sera studied, four of the sera produced death in up to 90% of hippocampal neurons and 80% of cerebellar neurons over 72 hours of incubation. Increased neuronal death was also observed in cerebellar and hippocampal cultures incubated with the other three anti-Hu sera, but statistical significance was not achieved compared with controls. Numbers of dead neurons in hippocampal cultures exceeded those in cerebellar cultures for each serum studied. However, this difference between hippocampal and cerebellar cultures reached statistical significance with only one sample. The apparent differences in cytotoxicity could not be accounted for by differences in antibody uptake or binding as determined by immunohistological methods, and it is possible that the observed differences may have been due to minor variations in Hu protein epitopes expressed in different neuronal populations. Because information concerning clinical syndromes associated with these samples was not available, it could not be determined whether this difference in cytotoxicity for cultured cells corresponded to sites of neurological deficits in living patients.

Because our studies involved incubation of cultures with sera containing anti-Hu antibodies, it was essential to determine the extent to which antibody binding and cell death observed in our cultures was associated with antibodies binding to Hu antigens and was not caused by antibodies to other intracellular neuronal proteins or other components which might be present in the clinical samples tested. To determine the role of anti-Hu antibodies in causing cell death, serum IgG from one patient with high titer of anti-Hu antibody response, demonstrated to produce extensive neuronal death in cultures, was adsorbed using His-tagged HuD expression protein bound to a nickel column. Serial three-fold dilutions of the adsorbed serum were then incubated with cerebellar and hippocampal slice cultures. The same serum passed through a sham nickel column, with bound control vector but without Hu antigen, was used as a control for nonspecific adsorption (see Methods). Cultures were evaluated for cell death using SYTOX and FLICA methods and were compared with amounts of cell death seen in cultures incubated with unadsorbed sera. The unadsorbed and sham-adsorbed samples produced robust neuronal death. In contrast, adsorption with Hu antigen essentially abolished intracellular accumulation of IgG, and neuronal death was reduced to background levels seen in control cultures incubated with normal human IgG (Figure 8, Additional file 3).

The effect of anti-Ri antibodies on cerebellar and hippocampal cultures differed markedly from that induced by anti-Hu antibodies. Cultures incubated with anti-Ri or control antibodies were followed for up to 219 hours. Although extensive IgG uptake was observed in neurons throughout both cerebellum (Figure 9) and hippocampus (Additional file 4), cell death and apoptosis, as determined by SYTOX and FLICA methods, were not observed beyond background levels seen in controls.