Interactions of Antibodies to the Gram-Negative Gastric Bacterium Helicobacter pylori with the Synaptic Calcium Sensor Synaptotagmin 5, Correlate to Impaired Vesicle Recycling in SiMa Human Neuroblastoma Cells

Due to molecular mimicry, maternal antibacterial antibodies are suspected to promote neurodevelopmental changes in the offspring that finally can cause disorders like autism and schizophrenia. Using a human first trimester prenatal brain multiprotein array (MPA), we demonstrate here that antibodies to the digestive tract bacteria Helicobacter pylori (α-HPy) and Campylobacter jejuni (α-CJe) interact with different synaptic proteins, including the calcium sensor synaptotagmin 5 (Syt5). Interactions of both antisera with Syt5 were confirmed by Western blot with a HEK293-cells overexpression lysate of this protein. Immunofluorescence and Western blotting revealed SiMa cells to express Syt5, which also co-migrated with a band/spot labeled by either α-HPy or α-CJe. Functionally, a 12-h pretreatment of SiMa cells with 10 μg/ml of either α-HPy or α-CJe resulted in a significant reduction of acetylcholine(ACh)-dependent calcium signals as compared to controls. Also ACh-dependent vesicle recycling was significantly reduced in cells pretreated with either α-HPy or α-CJe. Similar effects were observed upon pretreatment of SiMa cells with Syt5-specific antibodies. In conclusion, the present study supports the view that prenatal maternal antibacterial immune responses towards HPy and by this to Syt5 are able to cause functional changes, which in the end might contribute also to neurodevelopmental disorders. Electronic supplementary material The online version of this article (10.1007/s12031-020-01670-0) contains supplementary material, which is available to authorized users.


Introduction
With a prevalence of around 0.5% in the general population (Simeone et al. 2015), schizophrenia is one of the most frequent neuropsychiatric and probably also neurodevelopmental disorders worldwide. Patients suffering from this disease are characterized by changes in cognition, emotion, and self-perception, including both negative symptoms like avolition, alogia, apathy, poor or nonexistent social functioning, and positive symptoms like hallucinations and delusions (Tandon et al. 2009).
With regard to its causes, schizophrenia seem not only to depend on psychodynamic effects but also on changes in organic brain structure and functioning (Harrison 1999;Harland et al. 2009;Kahn and Sommer 2015). However, schizophrenia seems to be a multifactorial disease, and therefore, the underlying causes have still not yet been completely identified. Instead a couple of hypotheses have been formulated which still await their experimental confirmation.
One of the most fruitful of these hypotheses is the so called neurodevelopmental hypothesis for schizophrenia pathogenesis (Murray et al. 1992(Murray et al. , 2017, which suggests the disease to start already in utero during prenatal brain development and most probably already during the late first and/or early second trimester of pregnancy (Murray et al. 2017). Based on this hypothesis on one hand, genetic factors have been investigated, leading to the identification of a large number of schizophrenia candidate genes (Hosák et al. 2012;Giusti-Rodríguez and Sullivan 2013). On the other hand, also environmental factors such as infections as well as pre-and perinatal Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12031-020-01670-0) contains supplementary material, which is available to authorized users. complications during pregnancy and birth seem to play a role for schizophrenia pathology. Along this line, pre-and perinatal infections with pathogens like influenza or herpes simplex viruses, or eukaryotic endoparasites like Toxoplasma gondii, have already been investigated at more detail (Khandaker et al. 2013).
In the last decade, also bacterial infections, which have been suspected already during the very early days of psychiatric research to play a role in schizophrenia pathology (Noll 2004(Noll , 2007 have come anew into the focus of scientific interest (Khandaker et al. 2013;Lee et al. 2019). Thus, two population-based studies revealed that prenatal maternal infections during the first trimester of pregnancy with the Gram-negative bacterium Neisseria gonorrhoeae (NGo) correlate to an increased schizophrenia lifetime risk in the affected offspring (Babulas et al. 2006;Sørensen et al. 2009). NGo is widely known as a common cause for clinical and subclinical reproductive tract infections in women and men (Edwards and Butler 2011), and in these cases, antibodies directed to NGo-specific epitopes are a common feature of the blood serum of NGo-infected women (Hoffman et al. 1979). Therefore cross-reactivity of NGo-specific antibodies with specific brain proteins could well be responsible for perturbed brain development following exposure during early pregnancy.
A possible mechanism underlying such effects could be the so called molecular mimicry, which describes the observation that antibodies induced by an infection (Oates et al. 1977) due to molecular similarities are accidentally able to bind also to cellular proteins and by this to change or impair the functions of these proteins (Oldstone 1998). This concept has been already shown to underlie several neurological and neuropsychiatric disorders such as the Guillain-Barré syndrome that can be elicited by infections with Campylobacter jejuni (Speed et al. 1984;Wijdicks and Klein 2017), or Sydenham's chorea, which is sometimes caused by infections with A-type streptococci such as Streptococcus pyogenes and Streptococcus dysgalactiae (Berrios et al. 1985;Kirvan et al. 2006;Cunningham 2014). Based on this concept, we could already previously demonstrate that polyclonal antisera directed to NGo are able to interact with different cellular and synaptic proteins, some of which have already been earlier identified as schizophrenia candidates (Almamy et al. 2017).
Besides the effects of gonorrhea, there are also hints available that other maternal prenatal bacterial infections could contribute to the neurodevelopmental pathology of schizophrenia (Sørensen et al. 2009;Lee et al. 2019). Thus also the microaerophilic helix-shaped Gram-negative bacterium Helicobacter pylori (HPy) has been suggested to play a role as a possible environmental factor for schizophrenia pathogenesis (Yilmaz et al. 2008). HPy, a Gram-positive, facultative anaerobic bacterium, is primarily found in the gastric mucosa (Montecucco and Rappuoli 2001), where it is well known as the primordial pathogenic factor for peptic ulcer disease and even gastric cancer (Caruso and Fucci 1990;Labenz and Börsch 1994). Seropositivity for HPy and/or the closely related bacterium CJe have been found to be associated with several neurological disorders, such as Parkinson's disease (Dobbs et al. 1999), Alzheimer's disease (Roubaud-Baudron et al. 2012;Kountouras et al. 2009), and Guillain-Barré syndrome (Kountouras et al. 2005;Moran and Prendergast 2001), whereas a negative correlation has been found for certain forms of multiple sclerosis (Li et al. 2007). Despite these neurological diseases, HPy has been also suspected to play a role in schizophrenia pathology (Tret'iakov et al. 2006;Yilmaz et al. 2008). This connection has been established based on the fact that HPy is a wellknown causal agent for peptic ulcers of the stomach leading in some cases even to stomach cancer (Marshall and Warren 1983), together with the finding that schizophrenic patients frequently suffer also from ulcer disease (Tret'iakov et al. 2006) and exhibit a comparably high rate of HPy infections (de Hert et al. 1997).
In the present study, we intended to follow this line and started therefore to identify possible interaction partners for antibodies directed to HPy in the prenatal human brain on the protein level. For this, we used a commercial multiprotein array (MPA, hEXselect, Engine, Berlin, Germany) representing more than 10,000 different proteins, encoded by 23,806 expression clones of a cDNA library extracted of two late first-trimester human fetal brains (Büssow et al. 1998). By this we could identify and confirm among others synaptotagmin 5 (Syt5), a synaptic calcium sensor, and a protein with potential relevance for schizophrenia pathology (Kontkanen et al. 2002;Maycox et al. 2009;Martins-de-Souza et al. 2010) as a cellular interaction partner for α-HPy. Functional consequences of this interaction were then characterized by imaging its influence on both acetylcholinedependent calcium transients and on synaptic vesicle recycling in SiMa neuroblastoma cells, an in vitro cell culture model for human neuronal differentiation (Marini et al. 1999).

Recombinant Proteins and Overexpression Lysates
The following cell lysates were used in this study: Untransfected HEK293 control-lysate (Origene, cat. no. LY500001). Human Syt5-transfected HEK293 overexpress i o n l y s a t e ( O r i g e n e , L Y 4 1 8 8 4 7 ) . H u m a n SLC17A7-transfected HEK293 overexpression lysate (Origene, cat. no

hEXselect Multiprotein Array Analysis
A high-density multiprotein array (MPA) (hEXselect, Engine, Berlin, Germany, Order No. 1003) derived from a cDNA bank of two first trimester human fetal brain samples, containing 23,806 E. coli expression clones representing a total of 10,000 human proteins (Büssow et al. 1998;Horn et al. 2006), was incubated with either α-HPy or α-CJe according to the manufacturer's protocol. In brief, prior to the first use of these arrays, protein spots were fixed on the PVDF membrane for 10 min with 70% ethanol and rinsed twice with distilled water. After removing excess protein material with tissue paper and Tris-buffered saline (TBS) supplemented with 0.05% Tween 20 and 0.5% Triton X-100 (TBS-TT), the MPA was washed three times for 10 min with TBS-TT and rinsed twice for 10 min with TBS alone, followed by a 2-h blocking step with 3% skim milk powder in TBS. The MPA was then incubated overnight with the respective primary antiserum at a dilution of 1:2000 at 4°C on a rocking shaker. On the following day, the MPA was washed three times for 10 min in TBS, containing 0.05% Tween-20 (TBS-T), before incubating it with appropriate secondary antibodies at a dilution of 1:10 000 for 2 h at room temperature. After four washes with TBS-T and two washes with TBS, the location of immunoreactive spots was visualized by exposing the membrane to a SuperRX medical X-ray film (Fuji, Düsseldorf, Germany), during the application of a chemiluminescent peroxidase substrate (0.1-mol/l Tris-HCl (pH 8.6), 0.25-mg/ml luminol, 0.2-mg/ml phydroxycoumaric acid, and 0.1% H 2 O 2 ) in a dark room cabinet. For reprobing, the MPA membrane was stripped with 1mol/l NaOH for 40 min and then processed for a second round of immunodetection as described. All experiments were repeated at least in duplicate. As a control experiment, the hEXselect multiprotein was incubated with secondary antibodies only, to demonstrate the antibody-specificity of the signals obtained with bacteria-specific antisera (see also Supplementary Fig. 1).

Cell Culture
SiMa cells (human neuroblastoma; DSMZ, Braunschweig, Germany; Marini et al. 1999) were maintained in RPMI medium supplemented with 10% FCS, glutamine, and penicillin/ streptomycin, in an incubator at 37°C and under a humidified atmosphere containing 5% CO 2 . Medium was exchanged every other day, and shortly before reaching confluency, cells were mechanically suspended and seeded at a lower density either on 12-mm glass cover slips (Menzel, Braunschweig, Germany) in a 24-well plastic multiwell plate (Sarstedt, Nümbrecht, Germany) for immunocytochemistry or on 6well plastic multiwell plates (Sarstedt, Nümbrecht, Germany) for Western blot analysis. For antibody treatment, upon reaching the requested density, cells were pre-incubated for 4 h in an RPMI medium containing 1% FCS and subsequently for a time span as indicated in the text with 10 μg/ml of the respective antiserum. Prior to use, sodium azide of the applied antisera was removed by microdialysis with Amicon Ultra filter units (Merck, Darmstadt, Germany).

Immunocytochemistry
Immunocytochemistry was performed as described earlier (Dahm et al. 2010). In brief, cells on glass cover slips (Menzel, Braunschweig, Germany) were washed with PBS and fixed for 10′ with 4% paraformaldehyde (PFA) in PBS. After washing, cells were permeabilized for 10′ with a mixture of acetone/methanol (1:1) at − 20°C. Following three washes with phosphate buffered saline (PBS), cells were blocked for 1 h with goat serum (GS) diluted 1:50 in PBS (PBS-GS). Primary antibodies diluted 1:50 in PBS-GS were applied overnight at 4°C, followed by three washes with PBS and a 90′ incubation with Atto488-coupled secondary antibodies (Sigma-Aldrich, Steinheim, Germany), diluted 1:400 in PBS-GS at 37°C. After three washes with PBS, cells were mounted on standard microscope slides using a commercial mounting medium (DAKO, Glostrup, Denmark). Imaging was performed using an Axiocam digital camera system, mounted on an Axiophot microscope (both Zeiss, Jena, Germany).

Western Blot Analysis
Western blot analysis was performed as described previously (Dahm et al. 2010), with either 5 μg of total cellular protein, 500 ng of a recombinant protein sample, or 1 μg of an overexpression lysate being electrophoretically size separated on an 8.5% SDS polyacrylamide gel (Laemmli 1970) using a Mini Protean Gel System (Biorad, München, Germany). After tank-blot Western transfer onto a polyvinyl difluoride membrane (PVDF; Roth, Karlsruhe, Germany), blocking occurred for 1 h at 4°C in 3% (w/v) nonfat dry milk in TBS with 0.01% Tween-20 (TBST). Blots were then incubated overnight at 4°C with primary antibodies (as listed above), diluted 1:2000 in TBST with 0.1% nonfat dry milk. After washing, rabbit-specific peroxidase-coupled secondary antibodies (Sigma-Aldrich, Steinheim, Germany), diluted at 1:10000 in TBST, with 0.1% nonfat dry milk, were applied for 90′ at room temperature. Subsequent visualization occurred by exposing SuperRX medical X-ray films (Fuji, Düsseldorf, Germany), during the application of a peroxidase chemiluminescence substrate (0.1-mol/l Tris-HCl (pH 8.6), 0.25-mg/ml luminol, 0.2-mg/ml p-hydroxycoumaric acid, and 0.1% H 2 O 2 ) to the blots. For reprobing, blots were stripped with 1-mol/l NaOH for 15′ and then processed for a second round of immunodetection as described. Each blot was repeated at least three times. Also in this case specificity controls were performed incubating similar Western blots with secondary antibodies only. By this it is becoming clear that the signals obtained with the bacteria-specific antisera are not only due to an unspecific interaction of the secondary antibodies (see also Supplementary Figs. 2, 3, and 4).

Two-Dimensional Western Blot Analysis
Isoelectric focusing and 2D-Gel electrophoresis were performed according to a previously described method (Bollag and Edelstein 1994, see also Reuss and Asif 2014) using a Mini-Protean Gel System (Biorad, München, Germany). In brief, after cultivation on a 6-well plastic multiwell plate (Sarstedt, Nümbrecht, Germany) and before reaching confluency, cells were harvested in 2× sample collecting buffer (8-mol/l urea, 2% Triton X-100, 1% 2-mercaptoethanol, 0.1% bromophenol blue), containing also 2.4% of a commercial ampholyte solution (pH 3.5-10; Sigma-Aldrich, Steinheim, Germany). Protein concentration of the homogenate was determined densitometrically (Henkel and Bieger Fig. 1 Interactions of polyclonal antibodies directed to the Gram negative microaerophilic gastric bacterium Helicobacter pylori (α-HPy) with specific protein spots on the hEXselect multiprotein array (MPA). This array contains 23,806 spots of both full-length, and shorter protein fragments, expressed in an E. coli system, representing a total number of around 10 000 different human proteins (Büssow et al. 2000). As revealed by the green stained false color image of an X-ray film exposed to the immune incubation of a hEXselect MPA as visualized by ECL-detection, a number of pairwise immunoreactive spots revealed immunoreactivity with α-HPy, which subsequently could be identified according to their membrane location as 99 different proteins (see Table 1 for a more detailed gene description). For a functional classification see also Table 3.  Phosphoinositide-3-kinase-interacting protein 1 Intracellular signaling 1994), and for isoelectric focusing, 30 μg of total cellular protein were loaded on 5% polyacrylamide gel slices with 2.4% ampholytes (pH 3.5-10) containing also 50% urea for protein denaturation. Gel slices were then run for 30 min at 150 V followed by 2.5 h at 200 V. After equilibrating the slices for 30 min in electrophoresis sample buffer, they were located on top of an 8% SDS polyacrylamide gel followed by electrophoretic separation of the proteins according to their molecular weight. After this secondary run, Western transfer onto PVDF-membrane was performed by tank blot, and the obtained membrane was immunostained with the respective antiserum at a dilution of 1:2000. After photographic documentation, blots were stripped with 1-mol/l NaOH for 15′ and then processed for a second round of immunodetection as described. Each experiment was repeated at least three times. Also for the two-dimensional Western blot analysis, a control with secondary antibodies only reveals that the signals obtained with the bacteria-specific antisera are not detected only due to an unspecific interaction of the secondary antibodies (see also Supplementary Figs. 3 and 4).

Calcium Imaging
For dye loading, cell cultures on 12-mm glass cover slips, with or without a 12-h pretreatment with different antibacterial antisera, were incubated for 30 min at room temperature with 5 mmol/l of the Ca ++ -sensitive fluorescent dye Fluo3-AM (Sigma) diluted in a standard bath solution (NaCl 150 mmol/ l, KCl 5.4 mmol/l, CaCl 2 2 mmol/l, MgCl 2 1 mmol/l, Hepes, 10 mmol/l, glucose 10 mmol/l), the pH of which was adjusted to 7.3 with NaOH. For calcium imaging, dye-loaded cells were transferred to a recording chamber, continuously superfused with the standard bath solution at a rate of 5 ml/min. Acetylcholine (ACh) was applied as indicated in the text by changing the perfusate. Experiments were performed at room temperature (20°C). Fluorescence imaging was performed with a CCD camera system (Princeton Instruments, Trenton, NJ) mounted on an inverted microscope equipped with epifluorescence (Axiovert, Zeiss). In order to detect Ca ++ -transients, light with a wavelength of 485 nm was used for excitation, and fluorescence was measured at an emission wavelength of 540 nm selected with a 30-nm bandpass filter. Images were acquired every 5 s and were subsequently processed with the MetaFluor image analysis software (Universal Imaging, West Chester, PA). To identify and quantify neurotransmitter-reactive cells, fluorescence intensity during stimulation (F) was compared to the fluorescence intensity immediately before stimulation (F 0 ), with the ratio of both values (F/F 0 ) indicating the relative increase. To cover all Ca ++ -transients during a single neurotransmitter application, maximum plots of the complete sequence of pictures taken during every application period were generated using the MetaMorph Image analysis software (Universal Imaging, West Chester, PA).

Assay for Vesicle Turnover
Effects of α-HPy and α-CJe on acetylcholine(ACh)-elicited exocytosis and subsequent vesicle recycling by endocytosis were visualized by incubation with the fluorescent dye FM1-43 (Chowdhury et al. 2005;Gaffield and Betz 2006). Cells with or without a 12-h pretreatment with 10 μg/ml of a given antibacterial antiserum were transferred to a recording chamber, continuously superfused with a standard bath solution (at a rate of 5 ml/ min). Subsequently the perfusate was changed and the cells perfused for 50 s in standard bath solution containing 10 μmol/l of FM1-43 (Sigma-Aldrich, Steinheim, Germany) first in the absence and then in the presence of 1-nmol/l ACh (Sigma-Aldrich, Steinheim, Germany). After each FM1-43 exposure, excess dye was washed away by another 50 s perfusion with standard bath solution alone, and then FM1-43 fluorescence intensity (as a measure for the amount of exocytosis and subsequent vesicle recycling) was documented photographically with a CCD camera system (Princeton Instruments, Trenton, NJ) mounted on an inverted microscope equipped with epifluorescence (Axiovert, Zeiss). ACh-dependent exocytosis Fig. 2 Interactions of polyclonal antibodies directed to the Gram negative gut bacterium Campylobacter jejuni (α-CJe) with specific protein spots on the hEXselect multiprotein array (MPA). This array contains 23,806 spots of both full-length, and shorter protein fragments, expressed in an E. coli system, representing a total number of around 10 000 different human proteins (Büssow et al. 2000). As revealed by the red stained false color image of an X-ray film exposed to the immune incubation of a hEXselect MPA as visualized by ECL-detection, a number of pairwise immunoreactive spots revealed immunoreactivity with α-CJe, which subsequently could be identified according to their membrane location as 107 different proteins (see Table 2 for a more detailed gene description). For a functional classification see also

MTT-Assay for Cell Viability
To exclude possible effects of α-HPy and α-CJe on cell viability, a so called MTT-assay was performed (Levitz and Diamond 1985). For this, to SiMa cells pretreated with either of both antisera or to untreated controls, MTT (3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide, Merck, Darmstadt, Germany) was added at a final concentration of 50 μg/ml, followed by a 2-h incubation at 37°C under a humidified atmosphere with 5% CO 2 . After this, brightfield photographs of the stained cells were taken with an Axiocam digital camera system, mounted on an Axiophot microscope (Zeiss, Jena, Germany). Photographs were then evaluated densitometrically with the FiJi clone of the open source image analysis program ImageJ (see https://imagej.net), followed by statistical analysis of the obtained results with the free OpenOffice Calc software (see https://www.openoffice.de).

Gene-Specific Knockdown of Syt5 in SiMa Neuroblastoma Cells
In order to confirm interactions of the antibacterial antibodies used here with Syt5, a gene-specific knockdown of Syt5 in SiMa neuroblastoma cells was performed by transfecting them with a commercial Syt5 shRNA expression vector Known schizphrenia candidates are labeled by an asterisk (*)   . Upon transfection this vector expresses a short sequence of the Syt5 mRNA together with a hairpin structure (5′ CCG GCC AGA GTT ACA TAG ACA AGG TCT CGA GAC CTT GTC TAT GTA ACT CTG GTT TTT 3′) leading to a selective degradation of the endogenous Syt5 mRNA. In brief, a single-cell suspension of SiMa cells was washed with serum-free RPMI without antibiotics, and seeded in 0.8-ml of the same medium on a 6-well plate (Sarstedt, Nümbrecht, Germany) at a density of 3 × 10 6 cells per well. Five μg of plasmid DNA were diluted in 100 μl of RPMI medium and then mixed with 100 μl of a suspension of 10% Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, USA) in RPMI. This suspension was then given to the cells which were then incubated overnight at 37°C in a CO 2 -incubator. At the following day, 4 ml of RPMI medium (supplemented with FCS, glutamine, and penicillin/streptomycin, as stated above) was added, and the cells were again incubated for 24 h at 37°C in a CO 2 -incubator. Cells were then harvested for Western blot analysis as described above. Ten (10) μg of total cellular protein was then analyzed by this latter method to clarify whether expression of Syt5 as detected by Syt5-specific antibodies, and of the Syt5-specific bands stained by either α-HPy or α-CJe were diminished in Syt5-shRNA-transfected cells. For control purposes, both cells incubated with Lipofectamine only and cells transfected with 5 μg of a non-reacting control sh-RNA vector (MISSION® pLKO.1-puro:non-mammalian-shRNA; Merck, Darmstadt, Germany) were analyzed. As a control for equal protein loading, blots were also incubated with an antibody directed to the cellular housekeeping protein β-actin.

Results
In a first round of experiments, we tried to identify possible interaction partners for a polyclonal antiserum directed to the Gram-negative bacterium Helicobacter pylori (α-HPy), or the closely related bacterium Campylobacter jejuni (α-CJe) in the first trimester human fetal brain. For this we incubated a commercial hexSelect multiprotein array (MPA) covering the protein expression at this early stage of human brain development, with either α-HPy or α-CJe, and assigned the obtained immunoreactive spots to specific proteins according to the manufacturer's instructions. For α-HPy (Fig. 1), a comparably high number of different immunoreactive protein spots were detected that, according to their specific location, were identified as a set of 99 different proteins (for a detailed description see Table 1) which could be further assigned to different functional groups (see also Table 3). For α-CJe (Fig. 2), a slightly higher number of 107 different proteins (for a detailed description see Table 2) was identified which could also be assigned to several functionally different groups (see Table 3). Of a total number of 153 identified interaction partners, 53 proteins interacted with both antisera (α-HPy and α-CJe), whereas 46 of them reacted only with α-HPy and 54 only with α-CJe. For both α-HPy and α-CJe, several established schizophrenia candidate proteins with cross-reactivity to either of these antibodies could be identified including Syt5, Slc17a7, Stmn4, and Ncan. As a control for the antibodyspecificity of the signals obtained by the bacteria-specific antisera, a hEXselect multiprotein array incubated with secondary antibodies only revealed only a slight background staining (see Supplementary Fig. 1).
In a next step, we tried to confirm several of the protein interactions revealed by the MPA with an independent method. For this we tested interactions of α-HPy with a set of commercial overexpression lysates of these proteins in HEK293 cells, by Western blotting. As shown in Fig. 3a, α-HPy revealed again strong cross-reactivity with Syt5 and a somewhat weaker labeling of SLC17A7, demonstrating distinct immunoreactive bands at the appropriate molecular weight of~43 kDa for Syt5 and~62 kDa for Slc17a7. In contrast to this, interactions with other proteins showing cross-reactivity on the hexSelect MPA like Stmn4 or Ncan could not be confirmed by Western blot analysis with commercial overexpression lysates of these proteins. Also the totally unrelated protein Srf revealed no interaction with this antiserum. In a similar manner cross-reactivity of the closely related antiserum α-CJe with the same HEK293 overexpression lysates was also analyzed by Western blotting. As shown in Fig. 3b, α-CJe revealed an even stronger cross-reactivity with Syt5 and with SLC17A7, demonstrating distinct immunoreactive bands at the appropriate molecular weight of4 3 kDa for Syt5 and~62 kDa for Slc17a7. Again, interactions with Stmn4 and Ncan could not be confirmed by Western blot analysis with a commercial overexpression lysate of this protein, and also the totally unrelated protein Srf revealed no interaction with this antiserum. Again for control purposes, a similar Western blot was incubated with secondary antibodies only, revealing only a slight background staining (see Supplementary Fig. 2).
Due to its prominent role in exocytosis and synaptic transmission, and its exclusive expression in the human brain with high abundance in frontal cortex and the amygdala, we selected the interactions of α-HPy and α-CJe with Syt5 for a more detailed functional analysis. For this, we used here SiMa cells, a human neuroblastoma cell line with a neuronal phenotype as revealed by the formation of neurites and of cholinergic synapses. As shown in Figs. 4, 5, and 6, we clarified first whether these cells expressed Syt5, using immunocytochemistry Fig. 3 Western blot analysis of the cross-reactivity of antibodies directed to Helicobacter pylori (α-HPy) and Campylobacter jejuni (α-CJe) with different protein samples as provided by commercial HEK-293 overexpression lysates. a Cross-reactivity of α-HPy as revealed by a distinct immunopositive band can be observed for Syt5 and Vglut1 (Slc17a7), whereas Stmn4 and Ncan reveal no such band. Also a control lysate of non-transfected HEK293 cells, as well as an overexpression lysate of Srf is negative. b For α-CJe cross-reactivity as revealed by a distinct immunopositive band can be observed for Syt5 and Vglut1, whereas also in this case Stmn4 and Ncan reveal no such band. Again a control lysate of non-transfected HEK293 cells, as well as an overexpression lysate of Srf is negative. A corresponding negative control incubated with secondary antibodies only is shown in Supplementary Fig. 2  (Fig. 4) and both standard and 2D-Western blot analysis (Figs. 5 and 6). Thus, immunofluorescent labeling revealed SiMa cells to express Syt5 in synaptic boutons at their neurite terminals (Fig. 4). Further along this line, in a standard Western blot (Fig. 5a), a whole cell protein extract of SiMa cells, incubated with a polyclonal antibody to Syt5 (α-Syt5) revealed a distinctly labeled protein band at~43 kDa. After stripping and reincubation of the same blotting membrane with α-HPy, a similar band was labeled, and finally, the false color overlay image revealed both bands to run at the same size (Fig.  5a). Similarly the false color overlay image of a 2D-Western blot analysis of a SiMa cell whole cell extract (Fig. 5b) revealed α-Syt5 to label a protein spot of4 3 kDa with a characteristic isoelectric point of pH~9.3. Again, after stripping and reincubation with α-HPy, the same protein spot was labeled (Fig. 5b). Similarly, as shown in Fig. 6, another standard Western blot with a whole cell protein extract of SiMa cells (Fig. 6a), a polyclonal antibody to Syt5 (α-Syt5), labeled again a specific protein band of~43 kDa, and in this case after stripping and reincubation of the same blotting membrane with α-CJe, a protein band of the same size was labeled which in the false color overlay image turned out to be identical to the one labeled by α-Syt5 (Fig. 6a). Also in this case the false color overlay image of a 2D-Western blot analysis of a SiMa cell whole cell extract (Fig. 6b) revealed α-Syt5 to label a protein spot of~43 kDa with a characteristic isoelectric point of pH~9.3, and after stripping and reincubation with α-CJe, the same protein spot was labeled (Fig. 6b). Also in this case, control experiments were performed by incubating similar Western blots with secondary antibodies only, revealing a g a i n o n l y a s l i g h t b a c k g r o u n d s t a i n i n g ( s e e Supplementary Figs. 3 and 4).
In order to further confirm both, expression of Syt5 in SiMa neuroblastoma cells and interactions of α-HPy and α-CJe with Syt5 in these cells, a-gene specific knockdown of the Syt5 mRNA was performed by transfecting these cells with a commercial Syt5 shRNA expression vector ( Supplementary  Fig. 5). As revealed by the Western blot analysis for Syt5 ( Supplementary Fig. 5a), immunoreactivity for this protein was distinctly decreased in the whole cell protein extract from SiMa neuroblastoma cells transfected with the Syt5 shRNA expression vector, as compared with untreated cells, and also to cells transfected with a non-mammalian shRNA expression vector for control purposes. Likewise a similar Western blot, however in this case incubated with α-HPy ( Supplementary  Fig. 5b) revealed a distinctly decreased staining of the protein band representing interactions of this antiserum with Syt5. Incubation of the same Western blot with an antibody directed to β-actin confirmed the amount of protein loaded on each lane to be identical. The same was true for α-Syt5 and α-CJe, since also in this case a distinct reduction in immunoreactivity for both α-Syt5 (Supplementary Fig. 5d) and for the protein band representing the interactions of α-CJe with Syt5  Fig. 5e) could be observed. Also in this case incubation of the same Western blot with an antibody directed to β-actin confirmed the amount of protein loaded on each lane to be identical.
Syt5 has already been shown to play a role in Ca ++dependent synaptic transmission. As many other neuroblastoma cell lines, SiMa cells are known to express also receptors for acetylcholine (ACh), ligand binding of which is able to elicit intracellular Ca ++ -transients. In order to investigate possible functional consequences of α-HPy binding to Syt5, we analyzed now, whether α-HPy was able to interfere with synaptic activity as revealed by ACh-dependent Ca ++ -transients in this cell line. As revealed in Fig. 7, control cells stained with the Ca ++ -dependent fluorescent dye Fluo3-AM (Fig.  7a, b) revealed a strong transient intracellular calcium signal (F/F 0 ) upon stimulation with 10-nmol/l ACh. In contrast to that in cells pretreated for 12 h with 10-μg/ml α-HPy and then stained with the Ca ++ -dependent fluorescent dye Fluo3-AM (Fig. 7c), the intensity of ACh-dependent intracellular Ca ++ -transients (F/F 0 ) was distinctly reduced as compared with the control cultures. Likewise a pretreatment of SiMa cells with 10-μg/ml α-CJe (Fig. 7d) resulted also in a reduction of the intensity of ACh-dependent Ca ++ -transients as compared with control cultures. The statistical evaluation of a series of such experiments confirmed the effect of α-HPy on AChdependent Ca ++ -transients to be statistically significant (Fig. 7e). Statistical evaluation of a similar series of experiments, however in this case besides an incubation with α-HPy pretreated also with a polyclonal antiserum to Syt5, revealed a similar statistically significant effect Fig. 5 Identification of synaptotagmin 5 (Syt5) as an interaction partner of α-HPy in SiMa human neuroblastoma cells as revealed by one-and two-dimensional Western blot analysis. a Standard Western blot analysis of a whole cell protein extract of SiMa cells incubated with either α-HPy (blue) or α-Syt5 (red) revealing in both cases a distinct immunoreactive band with a molecular weight of~43 kDa the pink staining in the overlay image reveals them to be identical. b Overlay image of a two-dimensional Western blot analysis of a whole cell protein extract of SiMa cells incubated first with α-HPy (blue) and after stripping with α-Syt5 (red) revealing an immunoreactive spot with a molecular weight of~43 kDa and an isoelectric pH of~9.3, with a distinct co-labeling as revealed by its pink color. A corresponding negative control incubated with secondary antibodies only is shown in Supplementary Fig. 3 Fig. 6 Identification of synaptotagmin 5 (Syt5) as an interaction partner of α-CJe in SiMa human neuroblastoma cells as revealed by one-and twodimensional Western blot analysis. a Standard Western blot analysis of a whole cell protein extract of SiMa cells incubated with either α-CJe (blue) or after stripping with α-Syt5 (red) revealing in both cases a distinct immunoreactive band with a molecular weight of~43 kDa, the pink staining in the overlay image reveals them to be identical. b Overlay image of a two-dimensional Western blot analysis of a whole cell protein extract of SiMa cells incubated first with α-CJe (blue) and after stripping with α-Syt5 (red) revealing an immunoreactive spot with a molecular weight of~43 kDa and an isoelectric pH of~9.3 with a distinct colabeling as revealed by its pink color. A corresponding negative control incubated with secondary antibodies only is shown in Supplementary Fig. 4  Fig. 7 Effects of the antibacterial antisera α-HPy, and α-CJe on acetylcholine (ACh) dependent calcium transients in cultures of SiMa human neuroblastoma cells, as revealed by imaging with the calcium sensitive fluorescent dye Fluo3-AM. a Image of the maximum plot of the fluorescence intensity (F/F 0 ) of a non-treated control culture of SiMa human neuroblastoma cells revealing weak changes in the intracellular calcium concentration of different cells, representing the normal background activity. b Image of the maximum plot of the fluorescence intensity (F/F 0 ) of a non-treated control culture of SiMa human neuroblastoma cells, however in this case perfused for 50 s with 10 nmol/l of ACh in standard bath solution, revealing a distinct increase in intracellular calcium concentration. c Image of the maximum plot of the fluorescence intensity (F/F 0 ) of a Fluo3-stained culture of SiMa human neuroblastoma cells, pretreated for 12 h with 10 μg/ml of α-HPy, and subsequently perfused for 50 s with 10 nmol/l of ACh in standard bath solution. In this case the ACh-dependent increase in intracellular calcium concentration as revealed by Fluo3-AM is distinctly weaker as compared with the control experiment shown in (b). d Image of the maximum plot of the fluorescence intensity (F/F 0 ) of a Fluo3-stained culture of SiMa human neuroblastoma cells pretreated for 12 h with 10 μg/ml of α-CJe, and subsequently perfused for 50 s with 10 nmol/l of ACh in standard bath solution. Also in this case the AChdependent increase in intracellular calcium concentration as revealed by Fluo3-AM is distinctly weaker as compared with the control experiment shown in (b). e Diagram of the statistical evaluation of a series of experiments as shown in (a-d), revealing the reduction in the intensity of ACh-dependent calcium transients of human SiMa neuroblastoma cells pretreated for 12 h with either 10-μg/ml α-HPy or α-CJe to be highly significant. f Diagram of the statistical evaluation of a similar series of experiments as shown in (a-d), however in this case pretreated with α-HPy or α-Syt5, revealing the equally observable reduction in the intensity of ACh-dependent calcium transients of human SiMa neuroblastoma cells pretreated for 12 h with either 10 μg/ml α-HPy or α-Syt5 in comparison with control cultures, to be highly significant. **p < 0.01; Bar in (d) = 50 μm of this latter antiserum on ACh-dependent Ca ++ -transients (Fig. 7e).
Chemical neurotransmission is characterized by the exocytotic release of neurotransmitters, followed by the reuptake and recycling of the vesicular membrane by clathrinmediated endocytosis. Endocytotic vesicle recycling is therefore also an indicator of synaptic activity that can be visualized by the vesicular uptake and incorporation of the fluorescent dye FM1-43. We used this effect here to further characterize and quantify the effects of α-HPy, α-CJe, and α-Syt5 on ACh-dependent synaptic activity in SiMa neuroblastoma cells. As shown in Fig. 8, control cells incubated with FM1-43 in the absence of ACh revealed a distinct increase in FM1-43 incorporation upon stimulation with 10-nmol/l ACh (Fig.  8a, b). In cells pretreated for 12 h with 10-μg/ml α-HPy and then incubated with FM1-43 (Fig. 8c), the level of ACh-dependent uptake of FM1-43 was distinctly reduced as compared with the control cultures. Likewise a pretreatment of SiMa cells with 10-μg/ml α-CJe (Fig. 8d) resulted also in a reduction of ACh-dependent incorporation of FM1-43 as compared with control cultures. The statistical evaluation of a series of such experiments confirmed the effect of α-HPy on ACh-dependent incorporation of FM1-43 in SiMa cells to be statistically significant (Fig. 8e). Statistical evaluation of a similar series of experiments, however in this case besides an incubation with α-HPy pretreated also with a polyclonal antiserum to Syt5 (α-Syt5), revealed a similar statistically significant effect of both antisera on ACh-dependent uptake of FM1-43 in SiMa cells (Fig. 8e).
In order to exclude that the effects of the antibacterial antibodies on SiMa neuroblastoma cells revealed in the present study were not only the result of impaired cell metabolism, an MTT assay for cell viability was performed. As presented in Fig. 9, control cells incubated with MTT revealed the typical mitochondrial staining pattern in almost all cells (Fig. 9a). A similar MTT staining, both with regard to intensity and distribution, was obtained also in SiMa cells pretreated for 12 h with either 10-μg/ml α-HPy (Fig. 9b) or 10-μg/ml α-CJe (Fig. 9c). As revealed by the statistical evaluation of a series of such experiments, a 12-h pretreatment of SiMa cells with either α-HPy or α-CJe was not able to cause significant differences in MTT staining intensity (Fig. 9d).
Further along this line, expression of tyrosine hydroxylase, an established marker for neuroblastoma cells, was investigated by immunofluorescent staining. As revealed in Fig. 10, SiMa cells incubated with only the secondary antibody (SAO) exhibited almost no fluorescent staining (Fig. 10a). In contrast, cells incubated in addition with an antibody to tyrosine hydroxylase revealed a typical cytoplasmic staining (Fig. 10b). A similar staining, both with regard to intensity and distribution, was also obtained for SiMa cells pretreated for 12 h with either 10-μg/ml α-HPy (Fig. 10c) or 10-μg/ml α-CJe (Fig. 10d). As revealed by the statistical evaluation of a series of such experiments, a 12-h pretreatment of SiMa cells with either α-HPy or α-CJe was not able to cause significant differences in tyrosine hydroxylase immunoreactivity (Fig. 10e).

Discussion
Prenatal maternal bacterial infections have been reported to play a role in the outbreak of psychosis in the offspring later in life (Babulas et al. 2006;Sørensen et al. 2009). Thus, gonococcal infections seem to be associated with an increased schizophrenia risk; however, upper respiratory tract infections as well as ear-nose-throat infections also seem to have a similar effect (Sørensen et al. 2009). In a previous study (Almamy et al. 2017), we have already demonstrated that polyclonal Fig. 8 Effects of different antibacterial antisera (α-HPy, α-CJe) on AChstimulated exocytosis/vesicle recycling in cultures of SiMa human neuroblastoma cells, as visualized by life staining with the fluorescent vesicle marker FM1-43. a Fluorescence image of a non-treated control culture of SiMa human neuroblastoma cells, perfused for 50 s with 10 μmol/l of FM1-43 in standard bath solution, followed by 50 s with standard bath solution alone, revealing a distinct fluorescence signal in intracellular storage vesicles all over the cells. b Fluorescence image of the same cells as shown in (a), perfused again for 50 s with 10 μmol/l of FM1-43 in standard bath solution however in this case in the presence of 10 nmol/l ACh, (followed by 50 s with standard bath solution alone, revealing a distinct ACh-dependent increase in the fluorescence signal in the intracellular storage vesicles of these cells. c Fluorescence image of a culture of SiMa human neuroblastoma cells pretreated for 12 h with 10μg/ml α-HPy, and then perfused for 50 s with 10 μmol/l of FM1-43 in standard bath solution, followed by 50 s with standard bath solution alone, revealing a distinct fluorescence signal in intracellular storage vesicles all over the cells. d Fluorescence image of the same cells as shown in (c), also perfused for 50 s with 10 μmol/l of FM1-43 in standard bath solution, however in this case in the presence of 10-nmol/ l ACh, followed by 50 s with standard bath solution alone, revealing only a moderate ACh-dependent increase in the fluorescence signal in the intracellular storage vesicles of these cells. e Fluorescence image of a culture of SiMa human neuroblastoma cells pretreated for 12 h with 10μg/ml α-CJe, and then perfused for 50 s with 10 μmol/l of FM1-43 in standard bath solution, followed by 50 s with standard bath solution alone, revealing a distinct fluorescence signal in intracellular storage vesicles all over the cells. f Fluorescence image of the same cells as shown in (e), also perfused for 50 s with 10 μmol/l of FM1-43 in standard bath solution, however in this case in the presence of 10 nmol/ l ACh, followed by 50 s with standard bath solution alone, revealing again only a moderate ACh-dependent increase in the fluorescence signal in the intracellular storage vesicles of these cells. g Diagram of the statistical evaluation of a series of experiments as shown in (a-f), revealing the reduction in the ACh-dependent increase in FM1-43 fluorescence of human SiMa neuroblastoma cells pretreated for 12 h with either 10 μg/ml α-HPy or α-CJe to be highly significant. h Diagram of the statistical evaluation of a similar series of experiments as shown in (a-f), however in this case pretreated with α-HPy or α-Syt5, revealing the reduction in the ACh-dependent increase in FM1-43 fluorescence of human SiMa neuroblastoma cells pretreated for 12 h with either 10-μg/ml α-HPy or α-Syt5 as compared WITH control cultures to be highly significant. **p < 0,01; Bar in (f) = 50 μm antisera directed to Neisseria gonorrhoeae are indeed able to interact with different neuronal proteins, among which several schizophrenia candidate proteins can be found (Almamy et al. 2017). In the present study, we have now extended these findings on interaction partners and functional effects of antisera to the microaerophilic gastric bacterium Helicobacter pylori (HPy) and the closely related intestinal tract bacterium Campylobacter jejuni (CJe), which have also been implicated as a risk factor for schizophrenia pathology (de Hert et al. 1997;Yilmaz et al. 2008). By this we were able to document for the first time an interaction of polyclonal antisera directed to both HPy and CJe with the human synaptic calcium sensor synaptotagmin 5 (Syt5), and could correlate this interaction in SiMa human neuroblastoma cells to functional changes such as a diminished capability to respond to acetylcholine (ACh) with intracellular Ca ++ -transients, and a diminished rate of ACh-dependent vesicle recycling.
A possible mechanism underlying such an effect could be molecular mimicry, consisting of the induction of antibacterial or antiviral antibodies as a result of an infection, and, due to molecular similarities, an erroneous interaction of these antibodies with cellular proteins, functions of which can be impaired by antibody binding (Oates et al. 1977;Oldstone 1998). The present study tested this hypothesis by investigating interaction partners of antisera to HPy (α-HPy) and CJe (α-CJe) on a commercial multiprotein array and was able to identify comparably high numbers of different proteins with the ability to bind these antibodies. Thus, for both α-HPy and α-CJe proteins, interacting with these antisera belonged to a wide variety of functional groups including nuclear proteins and transcription factors such as Hdac10 and Atf4, transcriptional and translational regulators like NUDT16 and RPS27, membrane receptors and intracellular signaling molecules like Gpr173 and AKT1, proteins involved in intracellular transport and protein degradation like Rab11b and several ubiquitinylation factors, proteins involved in cell metabolism like Fvt1 and Sult1a1, mitochondrial proteins like Mros11 and Uqcrfs1, cytoskeletal regulators like Actr1a and Kif5b, extracellular signaling molecules like Fst and Ncan, as well as proteins involved in endocytosis like Cd130. Finally, a large group of proteins interacting with α-HPy and α-CJe comprised factors involved in neural development and synaptic transmission such as Npdc1 and Syt5.
Not all proteins with antibacterial cross-reactivity revealed affinity to both of the investigated antisera, such that of a total number of 153 identified interaction partners, only 53 bound to both antisera (α-HPy and α-CJe), whereas 46 and 54 proteins reacted with either α-HPy or α-CJe, respectively. Proteins with affinity to only α-HPy and an importance for Fig. 9 Effects of different antibacterial antisera (α-HPy, α-CJe) on staining of SiMa neuroblastoma cells with the metabolic marker dye 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT). a Untreated control culture of SiMa human neuroblastoma cells, revealing the typical staining pattern for this dye. b Culture of SiMa human neuroblastoma cells, pretreated for 12 h with 10 μg/ml of α-HPy, which shows an almost similar MTT staining intensity as in control treated cells. c Culture of SiMa human neuroblastoma cells, pretreated for 12 h with 10 μg/ml of α-CJe, revealing an almost similar MTT staining intensity as control treated cells. d Diagram of the statistical evaluation of a series of experiments as shown in (a-c), revealing no significant changes in MTT-staining and thus in cell viability of human SiMa neuroblastoma cells pretreated for 12 h with either 10 μg/ml of α-HPy or α-CJe, as compared to untreated controls. Bar in (c) = 40 μm neurodevelopment include the neuron migration factor astrotactin-2 and the growth factor Bmp1, whereas for α-CJe the proteins Dll3 and Fus would be interesting candidates for further investigations. However, due to the multitude of interaction partners for α-HPy and α-CJe on the heXselect protein array, the selection of proteins with putative importance for neurodevelopment and schizophrenia pathology remains a difficult task. We tried to tackle this by selecting proteins with known relevance as schizophrenia candidate genes, or proteins with a role in neurodevelopment and high expression in schizophrenia-relevant brain regions like the frontal and/or cingulate cortex, as well as the amygdala. This led us to proteins like synaptotagmin 5 (Syt5; Maycox et al. 2009;Martins-de-Souza et al. 2010), vesicular glutamate transporter 1 (Slc17a7; Eastwood and Harrison 2005;Bitanihirwe et al. 2009), Stathmin-4 (English et al. 2009Wang et al. 2010), and neurocan (Ncan; Schultz et al. 2014;Wang et al. 2016), which we have chosen for further analysis. By different Western blotting techniques, we could confirm at least some of the interactions demonstrated on the multiprotein array such as the interaction of α-HPy and α-CJe with Vglut-1 and Syt5, whereas the interactions of these Fluorescence image of an untreated control culture of SiMa human neuroblastoma cells, in this case incubated after fixation with an antibody specific for tyrosine hydroxylase (TH), revealing an intense cytoplasmic staining in the cells. c Fluorescence image of a culture of SiMa human neuroblastoma cells, pretreated for 12 h with 10 μg/ml of α-HPy, which after fixation were stained for TH, revealing an intense cytoplasmic labeling in most of the cells. d Fluorescence image of a culture of SiMa human neuroblastoma cells, pretreated for 12 h with 10 μg/ml of α-CJe, and after fixation stained for TH, revealing an intense cytoplasmic labeling in most of the cells. e Diagram of the statistical evaluation of a series of experiments as shown in (a-d), revealing no significant differences in TH-fluorescence in human SiMa neuroblastoma cells pretreated for 12 h with either 10-μg/ml α-HPy or α-CJe. Bar in (d) = 40 μm antibacterial antiserum on the MPA, a corresponding interaction partner by an alternative method like Western blotanalysis can be detected, and also the inverse case cannot be ruled out. This may be due to the fact that the proteins on the MPA were derived from cDNAs spotted on a PVDF membrane, which has then been translated in situ, using an in vitro translation system. Therefore these proteins lack most posttranslational modifications such as phosphorylation or glycosylation and therefore may be immunologically different from their natural occurring counterparts. Nevertheless it seems that the hexSelect MPA is a valuable tool for a first screening for possible immunological interaction partners of a given antiserum; however, every positive signal needs further confirmation and functional characterization in order to allow any statement on a putative pathological relevance!
In conclusion, results of the present study confirm and extend previous findings on interactions of antibacterial antibodies with human neuronal proteins, demonstrating for the first time cross-reactivity of antisera directed to the gastric mucosal bacterium Helicobacter pylori and the intestinal bacterium Campylobacter jejuni with the human synaptic protein Syt5, which seems to correlate also to functional changes in affected cells such as impaired ACh-dependent synaptic activity. These findings could be of importance for a better understanding of cellular and molecular mechanisms underlying the role of maternal bacterial infections for the increased schizophrenia risk in affected children later in life.