Transcriptome Profile of Nicotinic Receptor-Linked Sensitization of Beta Amyloid Neurotoxicity

Understanding the specific gene changes underlying the prodromic stages of Alzheimer’s disease pathogenesis will aid the development of new, targeted therapeutic strategies for this neurodegenerative disorder. Here, we employed RNA-sequencing to analyze global differential gene expression in a defined model nerve cell line expressing α4β2 nicotinic receptors (nAChRs), high-affinity targets for beta amyloid (Aβ). The nAChR-expressing neuronal cells were treated with nanomolar Aβ1–42 to gain insights into the molecular mechanisms underlying Aβ-induced neurotoxicity in the presence of this sensitizing target receptor. We identified 15 genes (out of 15,336) that were differentially expressed upon receptor-linked Aβ treatment. Genes up-regulated with Aβ treatment were associated with calcium signaling and axonal vesicle transport (including the α4 nAChR subunit, the calcineurin regulator RCAN3, and KIF1C of the kinesin family). Downregulated genes were associated with metabolic, apoptotic or DNA repair pathways (including APBA3, PARP1 and RAB11). Validation of the differential expression was performed via qRT-PCR and immunoblot analysis in the defined model nerve cell line and primary mouse neurons. Further verification was performed using immunocytochemistry. In conclusion, we identified apparent changes in gene expression on Aβ treatment in the presence of the sensitizing nAChRs, linked to early-stage Aβ-induced neurotoxicity, which may represent novel therapeutic targets.

. Top canonical signaling pathways and specific gene expression activated in differentiated nAChR-NG108-15 cells in response to prolonged nanomolar Aβ 1-42 treatment as identified by deep RNA sequencing (A) Volcano plot (log 2 of individual transcript fold-change (FC) as a function of the −log 10 of p-values (P)) showing the differential gene expression of the set of 15,336 genes induced by 100 nM Aβ 1-42 treatment in differentiated NG108-15 cells transfected with α4β2 nAChRs (nAChR-NG108-15). (B) Top canonical signaling pathways activated with Aβ treatment. The connecting lines (orange) indicate the ratios of genes in the identified signaling networks to total number of genes in the canonical pathways. Threshold line (right graph) indicates cut-off point of significance, p < 0.05, using Fisher's exact test for identifying particular pathways. (C) List of differentially regulated genes, identified as significant based on -log 10 P threshold of 2.5. The log 2 fold changes in the expression of identified genes in response to Aβ  treatment is shown at right. (D) An overview of the overlapping interactions between select up-regulated or down-regulated differentially expressed genes following treatment of α4β2 nAChR-transfected cells with Aβ 1-42 for 3 days and Using an in vitro nAChR-reconstituted nerve cell system for which we had previously established a tight timeline for Aβ-triggered toxicity, we discovered that the presence of α4β2 nAChRs, one of the notable high-affinity targets for Aβ, sensitizes the cells to toxic actions of oligomeric Aβ 13,14 , shifting the potency of Aβ for neurotoxicity from micromolar to nanomolar. We further demonstrated that this nAChR-induced Aβ neurotoxicity occurs through the timed alteration of discrete intracellular signaling molecules 14 . This prompted our study to investigate differential changes in downstream pathways underlying Aβ-linked neurotoxicity at a genetic level, possibly revealing new cellular targets for intervention in neurodegenerative processes.
The present study used two model systems. The differentiated rodent hybrid neuroblastoma NG108-15 neuronal cell line transiently expressing exogenous mouse sequences for specific nAChR subunits was employed as the defined in vitro nerve cell model for investigating global differential gene expression via RNA sequencing (RNA-seq) in response to sustained exposure to sensitizing levels (nM) of Aβ for neurotoxicity. Differentially regulated genes were then examined in Aβ-treated mouse hippocampal neurons as a validating primary in vitro neuronal model endogenously expressing nAChRs and in 5xFAD (familial Alzheimer's disease) APP/presenilin 1 (PS1) mutant mouse hippocampus.

Results
Prolonged exposure of nAChR-expressing neuronal cells to soluble nanomolar Aβ differentially modulated the expression of 15 genes. As a defined in vitro neuronal model expressing one of the prominent receptor targets for Aβ, namely high affinity α4β2-type nicotinic receptors, which sensitize the cells to Aβ toxicity 14 , neuroblastoma hybrid rodent NG108-15 cells exclusively expressing mouse α4β2-nAChRs (nAChR-NG108-15) were treated daily with 100 nM soluble oligomeric Aβ 1-42 as compared to vehicle-treated, receptor-expressing controls. Analysis of RNA-seq data generated from the treated cell cultures compared the levels of expression of 15,336 genes, as shown by the Volcano plot in Fig. 1A. Figure 1B lists in decreasing order of z-scores the canonical pathways activated in the nAChR-NG108-15 cells by Aβ, as identified by analysis of the RNA-seq data using the Ingenuity Pathway Analysis (IPA) tool and ranked by the highest z-scores. These canonical pathways, as ranked via IPA, included nucleotide and ribonucleotide biosynthesis, calcium signaling and DNA repair pathways including base excision repair (BER) and DNA double strand break repair by non-homologous end joining. Other activated pathways revealed on treatment with Aβ included 'Toll-like Receptor Signaling' , 'TREM1 Signaling' , 'iNOS Signaling' , and 'GranzymeB signaling' .
Of the differentially expressed genes identified via RNA sequencing, 15 were observed to be substantially altered in the neuronal cultures expressing α4β2-nAChRs on Aβ treatment as based on stringent threshold p-values (Fig. 1A,C). Of particular interest were CHRNA4 (the α4 subunit of the nAChR), KIF1C (kinesin family), RCAN3 (also known as DSCR1L2, a calcineurin regulator) and microRNA 344 g, which were up-regulated 1.3 to over 6-(log 2 )fold, and APBA3 (X11 family, APP adapter protein also known as Mint3), PARP1 (DNA repair family, polyADP-ribose polymerase) and microRNA 675, which were down-regulated −1.4 to −3.6-(log 2 )fold. IRAK1 (interleukin receptor kinase), linked to Parp1 through Akt (Fig. 1D) and NFκB regulation, was only modestly changed. Regulation of nAChR, kinesin family and Rab11 family genes on Aβ treatment is consistent with previous findings with the nAChR-NG108-15 cells, where upregulation of nAChR expression and functional responses were linked to enhanced receptor recycling involving Rab11 and altered axonal mitochondrial transport 13 . The other differentially regulated genes, as identified by RNA sequencing, are novel in regard to Aβ regulation.
Aβ-linked alteration of the expression of select genes in nAChR-expressing model neuronal cells confirmed via qRT-PCR, western blot analysis and immunocytochemistry. qRT-PCR was conducted on RNA extracted from control and Aβ 1-42 -treated differentiated nAChR-NG108-15 cells to confirm the changes in Chrna4, Rcan3, Irak1, Kif1c, Apba3 and Parp1 transcripts between untreated and Aβ-treated samples, normalized with respect to GAPDH ( Fig. 2A). The qRT-PCR data showed an upregulation of the levels of Chrna4, Rcan3 and Kif1c and downregulation of the Parp1, Apba1 and Irak1 transcripts, validating the RNA-seq results. A similar trend was observed for the expression levels of these transcripts in qRT-PCR conducted on RNA from primary mouse hippocampal neurons, which express endogenous nAChRs ( Supplementary Fig. S2), treated or not with 1 μM Aβ 1-42 for 7 days (Fig. 2B). (In contrast to nanomolar Aβ being sufficient for neurotoxicity in the sensitized nAChR-NG108-15 model, micromolar levels of Aβ are typically required for inducing neurotoxicity in primary neurons (e.g. see ref. 12 )).
The RNA-seq data were also validated at the protein expression level using immunoblot (western) analysis and immunocytochemistry. For western blot, the cell lysates collected from control and Aβ-treated neuronal cultures were used to determine changes in specific protein expression. We have previously shown an upregulation of other intracellular signaling molecules, having a role in neurological diseases, in general. Color indicates upregulation (green) or down-regulation (red) relative to the control cells. Solid lines represent direct interactions and dashed lines represent indirect interactions. Genes in bold are those identified via RNA sequencing. The interacting genes are: Akt, protein kinase B; Camkk; calmodulin-dependent kinase kinase; CHRN, nicotinic acetylcholine receptor; MAPK, MAP kinase; TNF, tumor necrosis factor; PARP, poly(ADP-ribose) polymerase; GBP, guanylate-binding protein; FUT, fucosyltransferase; HTR, serotonin receptor; IgG, immunoglobulin G; EZH, histone methyltransferase; HRH, histamine receptor; GZMM, granzyme; Dpp, dipeptidyl peptidase; TBKBP, TANK-binding kinase-binding protein; POU, POU homeobox; CCL, CC chemokine; BLVR, biliverdin reductase; LANCL, glutathione s-transferase; NTRK, neurotrophic tyrosine kinase receptor; PDK, pyruvate dehydrogenase kinase; FAIM, Fas apoptotic inhibitor.
surface levels of α4 nAChR in Aβ 1-42 -treated differentiated nAChR-NG108-15 cells 13 , which further confirms the data obtained from RNA-sequencing and also qRT-PCR. Here, we observed an up-regulation of Rcan3 and, conversely, decreased expression of Parp1 in the protein samples collected from treated nAChR-NG108-15 cells as well as treated primary hippocampal neurons (Fig. 3). The expression levels of Irak1 were modestly altered in the nAChR-NG108-15 cells and reduced in the hippocampal neurons ( Fig. 4). Immunostaining for Rcan3, Apba3 and Irak1 (Fig. 4) of control and Aβ-treated differentiated NG108-15 cells further affirmed these results. The difference between means of Aβ-treated and control cells was 25.40 ± 3.962 (p < 0.05) and −1.438 ± 2.021 (p > 0.05) for Rcan3 and Irak1, respectively.
Changes in expression of these representative proteins in the hippocampi from a familiar AD mouse model, 5xFAD (APP/PS1 mutant mice), at 1.5 months of age when Aβ levels begin to rise (see 15 ), as compared to age-matched background control mice, were also observed, particularly upregulation of Rcan3 (Fig. 5), which has a predominant neuronal localization 16 . Parp1 was also upregulated; however, it is expressed across a wide range of cell types but is most predominantly expressed in glia 16 , and thus the change in expression may have been significantly affected by non-neuronal cells. There was no significant difference in the expression of Irak1. These differences were not observed in 8-8.5-month-old 5xFAD hippocampal extracts, an age when the mice display robust AD-like endophenotypes of synaptic dysfunction and spatial memory deficits.

Analysis of pathways modulated by Aβ 1-42 treatment in the presence of α4β2 nAChRs. To
investigate the biological interactions of differentially expressed genes and identify functional networks, the prominent genes differentially expressed in response to Aβ treatment that were identified in the RNA-seq analysis were investigated using IPA, as previously described. Figure 1D shows the interactions between various genes, notably those involved in apoptosis and DNA repair pathways. One of the characteristic hallmarks of AD pathogenesis is an increase in oxidative damage to DNA. We have previously shown in our in vitro nAChR-NG108-15 neurotoxicity model that prolonged exposure to Aβ causes increased oxidative stress in a manner dependent upon the presence of the sensitizing α4β2-nAChRs 14 , leading to apoptosis. The gene network analysis of the differentially regulated genes identified here indicates that the Aβ-triggered increase in oxidative DNA damage and apoptosis ( Fig. 6A) is correlated, in part, with a defective BER pathway (Fig. 1A), in line with ample evidence showing that Aβ can have deleterious effects on DNA repair pathways including downregulation of BER-associated genes 17 .
In addition, Parp1 (poly(ADP-ribosyl) polymerase 1), a DNA repair enzyme that catalyzes the formation of poly ADP-ribose polymers from nicotinamide adenine dinucleotide (NAD + ), is usually activated by single-strand breaks associated with oxidative stress, and in the present study was found to be downregulated in neuronal cultures (Figs. 1-3) but upregulated in APP/PS1 5XFAD mouse hippocampal lysates (Fig. 5), which contain a mixture of neuronal and non-neuronal proteins.
One of the other prominent pathways that emerged in this study is calcium signaling (Figs. 1B and 6B). We have previously shown that Aβ induces changes in Ca 2+ levels via exogenous α7-nAChRs or α4β2-nAChRs expressed in the somata and axonal varicosities of differentiated NG108-15 cells 14,18,19 , as an early event in the Aβ toxicity timeline. The data from RNA sequencing highlighted a critical molecule involved in calcium signaling, notably Rcan3, a calcineurin regulator. The specific role of Rcan in Aβ neurotoxicity remains to be determined. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
A direct agonist-like action of soluble picomolar-nanomolar Aβ via nAChRs in the regulation of presynaptic calcium was previously observed 6,18,19 . The presence of α4β2-nAChRs, in particular, was found to significantly sensitize the cells to chronic Aβ-induced oxidative stress and, ultimately, apoptosis 13 . In order to elucidate the pathways involved in chronic Aβ-induced toxicity linked to a defined Aβ target, we attempted to determine the differential gene expression on prolonged treatment with nanomolar Aβ in the presence of sensitizing nAChRs through a transcriptome profile. Our study is the first of its kind to use a defined in vitro neuronal toxicity model reconstituted with high-affinity targets for Aβ conferring sensitization for the initiation of Aβ neurotoxicity. In this study, we discovered that prolonged exposure of a model nerve cell line exclusively expressing α4β2 nAChRs (nAChR-NG108-15 cells) to nanomolar Aβ resulted in substantial alteration in the expression of a unique set of 15 genes, confirmed first by quantitative transcript analysis (qPCR) and then, for prominent examples, protein analysis (immunoblot and immunostaining). Most of the 15 differentially expressed genes are distinct from genes www.nature.com/scientificreports www.nature.com/scientificreports/ differentially expressed in a different neuroblastoma model subjected to short-term treatment with micromolar levels of an Aβ toxic fragment (Aβ 25-35 ) using microarray analysis 20 . However, many of the canonical pathways noted in the present study, such as those involved in preapoptotic or apoptotic processes, were also prominent in the latter study 20 . The genes and pathways identified here in our acute Aβ neurotoxicity models are also in contrast to genes strongly correlated with late-onset AD, including apolipoprotein E (APOE), triggering receptor expressed on myeloid cells 2 (TREM2) and cluster of differentiation 33 (CD33) 21-24 among others, suggesting differences in gene regulation with early rises in Aβ in brain (prodromic period) as compared to AD.
Out of several genes displaying differential expression in the model nerve cells expressing α4β2-nAChRs on treatment with nanomolar Aβ, upregulation of CHRNA4 and RCAN3 is of particular interest. While upregulation of high-affinity Aβ target nAChRs with prolonged Aβ was previously described, regulation of RCAN by Aβ is a novel observation. Rcans (regulators of calcineurin 25 ), also previously known as calcipressins as well as Down Syndrome Critical Region-1 (DSCR1)-like proteins, constitute a conserved family of proteins from yeast to humans and bind calcineurin to modulate its activity 26,27 . Studies have shown that the transcripts for all mammalian Rcans (Rcan1, Rcan2, Rcan3) are expressed in the brain 28,29 . The critical role that Rcan proteins play in the physiology of brain is highlighted in reports showing increased locomotor activity and impaired working memory in Rcan1/Rcan2 double-knockout mice 30 . Interestingly, some studies show that there is a link between oxidative stress-induced Rcan levels and aging (and AD-related pathology). Specifically, Cook and colleagues 31 demonstrated that Rcan protein expression was upregulated in the pyramidal neurons of the temporal lobe with aging. It was further shown that there was a positive correlation between the total number of calcipressin (Rcan)-positive pyramidal neurons and the number of neurofibrillary tangles in the temporal cortex. www.nature.com/scientificreports www.nature.com/scientificreports/ Among the other genes altered in response to treatment with Aβ in the presence of nAChRs was Parp1, where, interestingly, down-regulation in our in vitro neuronal culture model as well as primary mouse hippocampal neurons was observed, while upregulation was evident in APP/PS1 (5XFAD) mouse hippocampus at 1.5 month of age when Aβ levels first rise significantly 15 , consistent with increased Parp1 activity found in hippocampus and entorhinal cortex of TgCRND8 (double mutant APP at KM670/671NL + V717F) mice at an equivalent stage (3 months) 32 . A prominent role for Parp1 in carrying out polyADP-ribosylation (up to 93% in the brain) and hence, maintaining DNA integrity has been described 33 . However, there is now ample evidence indicating that Parp1 www.nature.com/scientificreports www.nature.com/scientificreports/ also plays a significant role in cell death processes as well as regulation of mitochondrial function 34 . We have also previously shown a disruption in mitochondrial function and transport along axons in our in vitro model nerve cell culture system (nAChR-NG10815) in response to Aβ treatment 13 . There are also reports suggesting that Parp1 may play a crucial role in the possible interactions between molecules involved in AD-related pathology and regulation of mitochondrial function 35 . As Parp1 has been found to have predominantly glial expression (see 16), a putative differential regulation of Parp1 in neurons vs. glia by Aβ, inferred from our findings, may indicate, in turn, a differential regulation of mitochondrial function in different cell types in brain. Thus, the impact of Aβ in brain on Parp1 activity should be revisited for neuron vs. glia in susceptible brain regions.
Yet another class of RNA that showed novel changes in differential expression on exposure to Aβ in the presence of sensitizing nAChRs were microRNAs (or miRNAs). MicroRNAs, which are short (~21-23 nucleotides) conserved non-protein-coding RNAs transcribed from the genome 36 , constitute a very important class of regulators of gene expression 37,38 . Our data showed that the microRNA miR344g, whose role in the nervous system is yet to be determined, was the most strongly upregulated transcript of all those detected with RNA-seq on treatment of nAChR-NG108-15 cells with nanomolar Aβ. In addition, several other microRNAs including miR147, linked to Toll-like receptors, and miR24-1, linked to enhancer RNA expression, were among the strongly upregulated transcripts found on exposure of the nAChR-NG108-15 cultures to Aβ. In contrast, the microRNA miR675, a regulator of cell proliferation, was significantly downregulated on treatment with Aβ. At present, the possible roles for these various microRNAs in Aβ neurotoxicity remain to be discovered, as are the ramifications of these findings for the prodromic period of AD.
In sum, our transcriptome profile of model neuronal systems expressing high-affinity, sensitizing target nAChRs exposed to prolonged treatment with Aβ revealed novel gene regulation, including two notable genes, RCAN and PARP1, involved in the regulation of two key early events in Aβ neurotoxicity, namely calcium signaling and mitochondrial function, respectively. The differential expression of these two genes could thus serve as novel biomarkers for Aβ toxicity, perhaps in the prodromic stage prior to AD, to be confirmed in AD models and patients. Furthermore, characterizing the genes associated with neuronal dysfunction and death after Aβ treatment will have significant impact on developing neuroprotective agents to reduce or prevent pathogenesis leading to AD.

Materials and Methods
Nerve cell culture and transfection. Differentiated hybrid neuroblastoma NG108-15 cells were used as a defined model nerve cell system, as they are normally devoid of functional nAChRs. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 15% FBS and hypoxanthine/ aminopterin/ thymidine (HAT selection). Cells were differentiated on poly-L-lysine (plates) or Cell-Tak (coverslips) with 1 mM dibutyryl cyclic AMP in DMEM in the presence of reduced serum (1% FBS) and penicillin-streptomycin-glutamine for 72 h, as described previously 13 . pcDNA3.1 expression vectors harboring mouse sequences for α4and β2-nAChR subunits were transfected at 1:4 ratio, respectively, into the differentiated cells using FuGENE HD (ThermoFisher), a lipid-based transfection reagent, and the cultures were incubated for 48 h. The transfected cells were then treated or not (control), with 100 nM (sensitizing concentration) of Aβ 1-42 for 3 days, the time period based on the established timeframe for Aβ-induced toxicity in this nerve cell model system 13,14 . The Aβ treatment was therefore initiated following nerve cell differentiation and was listed as 0-3 days.
Primary hippocampal culture. Primary hippocampal neuron cultures were prepared from neonatal (1-2d old) C57/B6J mouse pups 39 under an approved University of Hawaii IACUC protocol (16-2282-3) in accordance to all guidelines and regulations. Following rapid decapitation, brains were removed from the mice into ice-cold Neurobasal A medium (NB) containing B-27 supplement, 5% fetal bovine serum and Gentamicin (Serum NB). Hippocampi were then dissected out under a stereomicroscope. The hippocampi were digested with activated papain (Worthington) in Hanks buffer with 10 mM cysteine at 37 °C for 15 mins. The preparations were washed by centrifugation in Serum NB. The cells were dissociated using sequential trituration with fire-polished Pasteur pipettes of decreasing diameter and collected by low-speed centrifugation. The dissociated cells were pre-plated in standard tissue culture dishes to remove adherent non-neuronal cells (glia; fibroblasts) for 10-15 min. The neuron-enriched supernatant was diluted to 1 × 10 5 cells/mL and plated into poly-D-lysine-coated 6-well plates in Serum NB. The cultures were maintained in Neurobasal A medium containing B-27 and Gentamicin for 7 days to select for neurons 14 , followed by treatment with 1 µM Aβ 1-42 for another 7 days, the treatment time period based on the observed timeframe for toxicity 39 . The Aβ treatment was therefore initiated following establishment of neuronal cultures and was listed as 0-7 days.
Animals. Animal husbandry and euthanasia were performed under an approved Institutional Animal Care and Use (IACUC) protocol (11-1219-6/16-2282-2) in conjunction with NIH guidelines for use of vertebrate animals in research. The transgenic mutant APP/PS1 mouse line, 5×FAD (familial AD) on the B6.SJL background (B6SJL-Tg (APPSwFlLOn, PS1 (PSEN1)*M146L*L286V) 6799Vas/Mmjax originally obtained from JAX stock #006554, MMRRC034840 hemizygous), was used as a well characterized model for Aβ-based pathology and neurodegeneration 15 , along with age-matched control (B6.SJL background) mice (MMRRC034840 Non-carrier). Age-matched mice (either sex) from the 5XFAD and B6.SJL colonies were housed in ventilated enrichment cages in the John A. Burns School of Medicine AAALAC-accredited Vivarium with ad libitum access to food and water. Mice were used at 1.5 months of age and 8-8.5 months of age.
Validation by immunocytochemistry. Cell cultures were fixed with freshly prepared 4% paraformaldehyde in HBS at room temperature for 40 min and rinsed with phosphate-buffered saline (PBS) for 30 min. The cultures were then permeabilized using 0.1% Triton-X in Tris-buffered saline (TBS) followed by washing with TBS for 20 min. Thereafter, a blocking buffer containing 5% bovine serum albumin and 10% normal goat serum in TBS was added to the cells for 30 min to block nonspecific binding. Affinity-purified primary antibodies (anti-Rcan3 (calcipressin3), anti-Apba3 (RRID:AB_2057069) and anti-Irak1; at 1:100; Fisher Scientific) were then added to the cultures and incubated overnight at 4 °C. The cultures were washed with 10% goat serum in TBS for 30 min, and incubated with the FITC-conjugated secondary IgG antibodies (typically at 1:500) and DAPI to label cell nuclei for 30 min at room temperature. The coverslips were finally washed with 10% normal goat serum and TBS and plated onto glass microscope slides, and sealed in Vectashield anti-fade mounting media (Vector Laboratories). The immunostained preparations were subsequently visualized using an Olympus IX71 fluorescence microscope with appropriate fluorescence filters via a 20X objective and images captured via CCD camera. Digitized images were analyzed via ImageJ.
Statistical analysis. Samples, as biological replicates, were from independent cultures (for each condition: RNA-seq, n = 4; qRT-PCR, n = 3; immunoblot, n = 3). Differentially expressed genes (treated vs. untreated) were assessed as significant based on an FDR (false discovery rate)-adjusted p-value (q-value) of <0.05 under the Cuffdiff test statistics (Cufflinks) 39 . For qRT-PCR and immunoblot analysis, Student t-tests (treated vs. untreated) were performed using Graphpad Prism (v.8) following testing for data normality, with p < 0.05 the minimum value for significance (as rejection of the null hypothesis).

Data availability
Datasets generated on analysis of the RNA-sequencing are available from the corresponding author on reasonable request. The model neuronal cell line used, NG108-15, is readily available from the American Type Culture Collection.