Genetic deletion of TRPA1 receptor attenuates amyloid beta- 1-42 (Aβ1-42)-induced neurotoxicity in the mouse basal forebrain in vivo

Amyloid β 1-42 peptide (Aβ1-42) accumulates in Alzheimer's disease (AD) that is toxic to the basal forebrain cholinergic (BFC) neurons in substantia innominata-nucleus basalis magnocellularis complex (SI-NBM). Transient Receptor Potential Ankyrin1 (TRPA1) receptor is present in murine brain, however its role in neurotoxic processes is unclear. We investigated the Aβ1-42-induced neurotoxicity in TRPA1 wild-type (TRPA1+/+) and knockout (TRPA1-/-) mice. Expression and neuroanatomical localization of TRPA1 receptor were examined using RT qPCR. Cholinergic fibre loss was determined on acetylcholinesterase (AChE) stained brain slices, and choline acetyltransferase (ChAT) immunohistochemistry was used to assess the cholinergic cell loss. Novel object recognition (NOR), radial arm maze (RAM) and Y-maze tests were used to investigate memory loss. Aβ1-42-injected WT mice showed marked loss of cholinergic fibres and cell bodies, which was significantly attenuated in TRPA1-/- animals. According to the NOR and RAM tests, pronounced memory loss was detected in Aβ1-42-injected TRPA1+/+ mice, but not in TRPA1-/- group. Our findings demonstrate that TRPA1 KO animals show substantially reduced morphological damage and memory loss after Aβ1-42 injection in the SI-NBM. We conclude that TRPA1 receptors may play an important deteriorating role in the Aβ1-42-induced cholinergic neurotoxicity and the consequent memory loss in the murine brain.


Introduction
One of the major pathomorphological hallmarks of Alzheimer's disease (AD) is the significant neuronal death in the cortex, hippocampus and basal forebrain. The cholinergic system of the basal forebrain (BFC) is the pivotal source of cholinergic connections to the cortex and hippocampus, which plays a major role in attention and memory (Zaborszky et al., 1999;Beninger et al., 2001). Accumulation of neurotoxic β-amyloid peptides, such as Aβ 1-42 contributes to neuronal loss in BFC in AD (Maccioni et al., 2001). The Aβ activates astrocytes and microglial cells, also it has direct neurotoxic effect by ligand-like interaction with the N-methyl-D-aspartate (NMDA) glutamate receptors (Harkany et al., 2000). Aβ 1-42 microinjection into the nucleus basalis magnocellularis (NBM) results in severe damage of neurons in BFC and memory deficits in rodents (Harkany et al., 2000(Harkany et al., , 1998Giovannelli et al., 1995;Kwakowsky et al., 2016).
There is an increasing evidence that neuro-immune interactions may play a critical role in the pathomechanism of several neurodegenerative diseases (Candore et al., 2007;Masters et al., 2015, Sághy et al., 2016 and regulatory role of the transient receptor potential (TRP) ion channels is strongly suggested in these processes. TRP receptors form non-selective cation channels, with preferentially high calcium ion permeability. Calcium influx triggers several intracellular pathways. TRP channels are sensitive for various stimuli including mechanical, thermal triggers or chemical ligands. Considering these interactions, they are likely to be sensors for several physiological or https://doi.org/10. 1016/j.mad.2020.111268 Received 23 January 2020; Received in revised form 16 May 2020; Accepted 20 May 2020 pathophysiological stimuli (Khalil et al., 2018). One of the TRP receptors, the transient receptor potential ankyrin 1 (TRPA1) is best known as a sensor for environmental irritants participating in pain, cold and itch sensation. TRPA1 has a wide range of ligands; specifically it is sensitive to low and high temperature, osmotic changes, natural and synthetic irritants. Several TRPA1 agonists are reactive electrophilic ligands (Zimova et al., 2018). Channel gating effect of these compounds is based on covalent modification of cysteine and lysine residues within the N-terminus and transmembrane domains of the receptor (Zygmunt and Högesättt, 2014). Since TRPA1 can be triggered by endogenous compounds generated by tissue injury and inflammation, it can be a potential target for anti-nociceptive and anti-inflammatory drug candidates (Nishida et al., 2015;Horváth et al., 2016). Recent data showed that TRPA1 receptors are also expressed in the brain and play modulatory role in neurodegenerative disorders and neuroinflammation, such as multiple sclerosis and AD (Sághy et al., 2016, Lee et al., 2016Bölcskei et al., 2018;Dalenogare et al., 2020;Ritter et al., 2020). TRPA1 channels have also been found on astrocytes of corpus callosum (Sághy et al., 2016), also in oligodendrocytes of the cerebellum (Hamilton et al., 2016) and in the endothelium of cerebral arteries (Sullivan et al., 2015;Pires and Earley, 2018). Nowadays, Lee and colleagues have provided immunohistochemical evidence that hippocampal astrocytes express TRPA1 in wild-type mice; and immunoreactivity of the receptor protein also occurs in the cortical neurons of amyloid precursor protein/presenilin1 (APP/PS1) transgenic mice (Lee et al., 2016). However, only few in vivo data are available to corroborate the function of TRPA1 in the brain. The principal aims of the present study were to investigate the morphological and the functional aspects of TRPA1-mediated events in BFC using Aβ 1-42 -induced neurotoxicity model in TRPA1 knockout mice.

Preparation of Aβ 1-42 peptide samples
Aβ 1-42 was expressed recombinantly in E. coli and purified according to Ikenoue and colleagues (Ikenoue et al., 2014). Briefly, the DNA sequence coding the human Aβ 1-42 peptide was artificially constructed using codons preferred by E. coli with an extra Met residue at the N-terminal of the peptide. The DNA construct was inserted into pAED4 vector (Doering and Matsudaira, 1996) and the protein expression was carried out in E. coli BL21 (DE3) pLysS strain (Novagen, Inc., Madison, WI), using 1 mM IPTG for induction. The peptide accumulated in the inclusion bodies of the bacterial cells. The purified inclusion bodies were solubilized in 20 mM NaOH and the peptide was further purified by repeated cycles of amyloid growth at low pH and monomerization in hexafluoroisopropanol combined with centrifugation steps. The monomeric lyophilized peptide was first dissolved in 10 mM NaOH on ice, then it was diluted into appropriate amount of TBS buffer resulting in a final concentration of 300 μM peptide and a pH of 7.4. The solution was aged for 5 days at room temperature. Before administration, the solution was spun down in a tabletop centrifuge at 15,000 g for 2 min to remove larger aggregates.

Measurement of Trpa1 expression
Brains and trigeminal ganglia of TRPA1 +/+ (n = 6) and TRPA1 −/− mice (n = 6) were quickly dissected after decapitation. The samples were immediately frozen in liquid nitrogen and stored at −80°C. Then, brains were sliced by razor blades using a coronal brain matrix (cat.nr. 15050, Ted Pella, Redding, CA) to obtain 1 mm coronal sections (Palkovits et al., 1978). A microdissection needle (cat.nr. 15091, Ted Pella) of 1 mm diameter was used to collect micropunches containing the following brain areas in the marked coronal planes (the distances of coronal sections from the Bregma were indicated in brackets: SI-NBM (0-(−1 mm)), hippocampus and somatosensory cortex (−1 mm-(−2 mm)) (Paxinos and Franklin, 2001). Total RNA content was isolated using Direct-zol RNA MiniPrep kit (Zymo Research, cat.nr: R2060), according to the manufacturer's instruction. Samples were diluted to equal amount of total RNA and cDNA pools were generated by reverse transcription using Maxima First Strand cDNA Synthesis kit (Thermo Fisher Scientific, cat.nr: K1971). The expression level of Trpa1 mRNA was determined with Stratagene Mx3000 qPCR instrument using custom-designed Trpa1 TaqMan assay (forward primer: atgccttcagcaccccattg, reverse primer: gacctcagcaatgtccccaa, probe: 56FAMtggg-cagctZENtattgccttcacaat3IABkFQ) and normalised with the predesigned hypoxanthine phosphoribosyltransferase (Hprt) TaqMan assay (Mm.PT.39a.22214828). The thermal cycling was initiated at 95°C for 5 min followed by 45 cycles of 35 s at 95°C and 30 s at 62°C and 1 min synthesis at 72°C. Primers and probes were obtained from IDT, Leuven, Belgium.

Immunohistochemistry
To examine the cholinergic cell loss in the SI-NBM, ChAT immunohistochemistry was accomplished with free-floating technique (Kwakowsky et al., 2016). To determine the rate of cholinergic fibre loss in the somatosensory cortex, AChE histochemistry with silver nitrate intensification was performed (Koszegi et al., 2011). On day 13, Aβ 1-42 -injected and non-injected control animals were deeply anesthetized and perfused as described above. Dissected brains were postfixed for 4 h at room temperature, then incubated overnight in 30 % sucrose diluted in phosphate buffer. Four sets of 30 um thick coronal M. Payrits, et al. Mechanisms of Ageing and Development 189 (2020) 111268 sections were cut on a freezing microtome (Leica, Germany).

Behavioural experiments and analysis
2.6.1. Novel object recognition test (NOR) NOR tests the recognition memory (Li et al., 2011;Bevins and Besheer, 2006;D'Souza et al., 2015). We applied a 3-day-long protocol. On day 1, mice were allowed to move around in the 45 × 45 × 30 cm wooden box for 5 min. On day 2, animals were let freely explore two identical objects for 5 min. On day 3 (24 h later), the animals were presented to one familiar and one novel object, which was similar in size but different in color and shape compared to the familiar one (Fig. 4A). Mice could explore the objects freely for 5 min. Behaviour of animals was video-recorded and analysed with Ethovision XT 11 software (Noldus Information Technology, Netherlands). The obtained data are calculated and represented as discrimination index (DI) = difference in time exploring the novel and familiar object/total exploration time.

Radial arm maze test (RAM)
RAM is a suitable tool for the assessment of both short and longterm memory. Mice on normal diet were trained for three days (habituation and learning period) to find food pellets (Dustless Precision Pellets 45 mg, Sucrose; BioServ, US) in four selected arms of the eightarm radial maze (5 × 35 cm for each arm and 5 cm in diameter for the central platform). The sessions lasted until the animals collected all the four food pellets or 5 min, whichever came first. The learning ability of the animals was assessed on the fourth day of the experiment. Behaviour was video-recorded and analysed with EthoVision XT 11 Software (Noldus Information Technology, Netherlands). The obtained data are calculated and represented as: reference memory error = entries into unbaited arms (Zhang et al., 2012;Li et al., 2011).

Y-maze test
Spatial working memory was measured after placing the animals in one of the arms of a Y-shaped maze (arms 5 cm wide x 35 cm long x 10 cm high) for 5-mins (D 'Souza et al., 2015). The track of the animals was obtained with Ethovision XT 11 software (Noldus Information Technology, Netherlands). The sequence and number of arm entries were used to determine the percentage of spontaneous alternation by the following formula: the sequence of visiting the 3 arms (A, B and C) consecutively in any order was considered as correct alternating behaviour (ABC, ACB, BAC, BCA, CAB, CBA) and was divided by the number of arm entries minus two (representing all possible alternation sequences). Total number of arm entries was also measured for the assessment of locomotor activity.

Statistical analysis
Data in all experiments were expressed as mean ± SEM. Data were analysed by factorial ANOVA followed by Bonferroni's post hoc test with a value of p < 0.05 considered significant. In case of cell/fibre loss, Mann-Whitney test was used. Learning curve was analysed by repeated measures two-way ANOVA followed by Bonferroni's post hoc test. All statistical analysis was performed using GraphPad Prism Software 6.0, except for the factorial ANOVA analysis that was performed using STATISTICA Software (TIBCO Software Inc., CA).

Trpa1 mRNA expression in the brain
To elucidate whether Trpa1 is expressed in brain regions relevant to this work, we performed RT-qPCR in micropunched areas obtained from TRPA1 +/+ or TRPA1 −/− animals. Trpa1 expression became significant at 35-36 cycles. Trpa1 mRNA was detectable in SI-NBM, hippocampus, somatosensory cortex and trigeminal ganglion isolated from TRPA1 +/+ mice. The Trpa1 expression level of the TRPA1 +/+ samples was compared to their TRPA1 -/counterparts (Fig. 2).
In the Y-maze test there was no difference in locomotion between groups reflected by the total number of arm entries.

Discussion
Our study provides the first evidence for the putative role of TRPA1 receptor in neuronal loss and cognitive impairment in Aβ 1-42 -induced murine neurotoxicity model. With the help of neuroanatomical evaluation and in vivo functional studies we have demonstrated that genetic deletion of TRPA1 receptor attenuates Aβ 1-42 toxicity in cholinergic neurons in the SI-NBM.
There are several dementia models available: spontaneous, chemically induced (e.g. Aβ, scopolamine, okadaic acid) and transgenic (e.g. APP/PS1 knockout mice, tau-related) models (Tayebati, 2006;Neha Sodhi et al., 2014). None of them is able to show all pathophysiological features of the human AD. Transgenic models are widely used, because non-invasive procedures are needed and intracellular Aβ-toxicity can  Data were presented as the mean ± SEM (n = 6-10) and were analysed with Mann-Whitney test (*** p < 0.001).
M. Payrits, et al. Mechanisms of Ageing and Development 189 (2020) 111268 also be triggered (LaFerla and Green, 2012). However, in gene-manipulated animals, the massive loss of neurons and fibres, as one of the most typical changes of human disease is not characteristic. In contrast to the genetic models of AD (e.g. APP/PS1 knockout mice) (Lee et al., 2016), our experimental approach with intracerebral injection of Aβ acutely generate AD-like neurotoxicity and cognitive deficits. Early recognition of the AD is almost impossible because of the atypical symptoms of the slow progression. The current treatment focuses on the maintenance of the remaining cholinergic functions applying cholinesterase inhibitors and NMDA receptor antagonists (Talesa, 2001;Eleti, 2016). Despite of some moderately effective drugs, the pharmacotherapy of the AD is still an unsolved problem. Therefore, useful animal models that properly mimic the pathophysiological characteristics of AD are still essential for the discovery of novel therapeutic targets. Aβ microinjection elicits significant loss of neuronal elements with remarkable memory impairment that can be detected with classical behavioural tests (Morris water maze, Y/T-maze, NOR). This model represents the later phase of the disease when patients' relatives realize the memory problems.
TRPA1 is a widely studied non-selective cation channel, expressed by the primary afferent neurons regulating pain sensation (Nishida et al., 2015;Jardín et al., 2017). In addition, increasing evidence has been published about its function in the CNS: TRPA1 can be found in neurons of the rodent hippocampus, magnocellular neurosecretory cells, the visceral afferent pathway and cortex (Nilius et al., 2012;Kheradpezhouh et al., 2017). Also its expression has been investigated in glial cells and its putative role in the pathophysiology of AD (Lee et al., 2016, Shigetomi et al., 2013. Since it is well-known, that TRPA1 is a key regulating receptor of calcium homeostasis, it may influence cell death or survival (Bosson et al., 2017, Stueber et al., 2017. Aβ 1-42 stereotaxic microinjection caused robust damage of cholinergic neurons and fibres in the WT animals, which indicates that this method is suitable model of the typical pathomorphological changes of cholinergic neurotoxicity. Interestingly, in TRPA1 −/− animals the Aβ 1-42 -induced cell and fibre loss of cholinergic neurons were nearly absent. These results suggest pivotal role of TRPA1 in the neurodegenerative processes of the brain, especially in the basal forebrain cholinergic neurons. Morphological changes evoked by Aβ injection Fig. 4. Effects of vehicle or Aβ 1-42 injection on the performance in the novel object recognition (NOR) task in TRPA1 +/+ and TRPA1 −/− mice. On day 1 of the 3-day-long NOR test (A) distance moved (B), on day 2 location preference (C) was measured. On day 3 memory function was determined as discrimination index (D). Data are presented as the mean ± SEM (n = 6-10) and were analysed by factorial ANOVA followed by Bonferroni's post hoc test and location preference with one sample t-test in comparison to 50. Fig. 5. Effects of vehicle and Aβ 1-42 injection on the performance in the radial arm maze (RAM) and Y-maze task in TRPA1 +/+ and TRPA1 −/− mice. Learning curve on the basis of reference memory error (A) and the reference memory error on day 4 (B) was determined in RAM. Spontaneous alternation (C) and average entries (D) was measured in Ymaze test. Data are presented as the mean ± SEM (n = 6-10). Learning curve was analysed by repeated measures two-way ANOVA followed by Bonferroni's post hoc test, and reference memory error and average entries were analysed by factorial ANOVA followed by Bonferroni's post hoc test (*p < 0.05).
have also been supported by functional investigations. Bilateral administration of Aβ 1-42 induced significantly impaired memory function in TRPA1 +/+ mice. These changes could be detected in both NOR and RAM, but Y-maze was not suitable for the demonstration this type of cognitive impairment. NOR test is one of the most appropriate methods assessing memory loss in rodents. It has been used in numerous different (transgenic and non-transgenic) AD models. It has several advantages, such as short testing period, no food restriction is needed, and it is based on the natural exploratory behaviour of the mice (Zhang et al., 2012;Grayson et al., 2015). RAM is also frequently used in AD mouse models (Webster et al., 2014), and it is a suitable for evaluation of both short-and long-term memory (Puzzo et al., 2014). Despite of the Aβ treatment, the memory performance of the TRPA1 −/− animals was comparable with the naïve mice in these tests. This means, that the measured morphological changes are manifested at the behavioural level. These results verify the observations of the former transgenic model, that genetic lack of TRPA1 can significantly attenuate the memory loss (Lee et al., 2016), nevertheless not all of the tests showed the same results. Our Y-maze methodology was slightly different from the data of the above mentioned research group. Lee et al. divided Ymaze into start, familiar and novel arms, and they measured the visiting activity of mice in the novel arm, which is very similar to the aspect of NOR. On the contrary, we considered the arms as equal, and the alternation activity was assessed during the experiments. In this case we gain information about the spatial memory.
Our findings showed that genetic deletion of TRPA1 receptors substantially inhibited the Aβ 1-42 -induced cholinergic cell body loss and fibre loss. We demonstrated that Aβ 1-42 injection into the SI-NBM caused significantly reduced neuronal damage and memory loss in the TRPA1 KO animals. The exact molecular mechanism, how TRPA1 receptors contribute to the Aβ 1-42 -induced cholinergic cell loss is still not fully elucidated. Previously, Harkany and colleagues published that Aβ 1-42 induces cholinergic cell death via NMDA receptor mediated intracellular Ca 2+ increase in SI-NBM (Harkány et al., 2000). We speculate that TRPA1 receptor-mediated calcium influx may enhance the neurotoxic effect of the Aβ 1-42 due to the cytotoxic Ca 2+ overload in cholinergic neurons in SI-NBM. This theory warrants further investigation.

Conclusion
Our study has presented the first evidence that the genetic lack of TRPA1 receptor significantly attenuates the Aβ 1-42 -induced cholinergic neurotoxicity in the SI-NBM. We also demonstrated that TRPA1 KO animals show substantially reduced memory loss after Aβ 1-42 lesion compared to the wild-type counterparts. Based on our findings, inhibition of the TRPA1 receptors might be a novel promising prospect in the pharmacotherapy of AD. The limitation of our study is, that its main conclusions are based exclusively on morphological and functional data provided by experiments on TRPA1 wildtype and KO mice.

Availability of data and material
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Author's contribution
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors read, revised and approved the final version of the submitted manuscript. EP had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design: MP, ÉB, EP. Acquisition of data: MP, ÉB, SG, DE, ÁK, EK, ÉSZ, JK. Analysis and interpretation of data: MP, ÉB, SG, ÁK, EK, JK, ÉSZ, EP.

Declaration of Competing Interest
The authors declare no conflict of interest.