Central cholinergic function and metabolic changes in streptozotocin‐induced rat brain injury

As glucose hypometabolism in the brain is an early sign of Alzheimer´s dementia (AD), the diabetogenic drug streptozotocin (STZ) has been used to induce Alzheimer‐like pathology in rat brain by intracereboventricular injection (icv‐STZ). However, many details of the pathological mechanism of STZ in this AD model remain unclear. Here, we report metabolic and cholinergic effects of icv‐STZ using microdialysis in freely moving animals. We found that icv‐STZ at a dose of 3 mg/kg (2 × 1.5 mg/kg) causes overt toxicity reflected in body weight loss. Three weeks after STZ administration, histological examination revealed a high number of glial fibrillary acidic protein reactive cells in the hippocampus, accompanied by Fluoro‐Jade C‐positive cells in the CA1 region. Glucose and lactate levels in microdialysates were unchanged, but mitochondrial respiration measured ex vivo was reduced by 9%–15%. High‐affinity choline uptake, choline acetyltransferase, and acetylcholine esterase (AChE) activities in the hippocampus were reduced by 16%, 28%, and 30%, respectively. Importantly, extracellular acetylcholine (ACh) levels in the hippocampus were unchanged and responded to behavioral and pharmacological challenges. In comparison, extracellular ACh levels and cholinergic parameters in the striatum were unchanged or slightly increased. We conclude that the icv‐STZ model poorly reflects central cholinergic dysfunction, an important characteristic of dementia. The icv‐STZ model may be more aptly described as an animal model of hippocampal gliosis.


| INTRODUC TI ON
Animal models are of great importance to understand Alzheimer's disease (AD) pathology and to evaluate potential therapies, but can also be misleading if they do not adequately reflect human disease (Drummond & Wisniewski, 2017). While transgenic approaches imitate early-onset AD only, toxic models may more closely mimic the typical, late-onset dementia in humans (Cavanaugh, Pippin, & Barnard, 2014). One such model in which brain injury is induced by intracerebroventricular injection of streptozotocin in rats (icv-STZ), was developed based on the idea that AD resembles "diabetes of the brain", and is caused by an insulin-resistant brain state (Hoyer, 1998;de la Monte & Tong, 2014). Insulin can cross the blood-brain barrier (Banks, Owen, & Erickson, 2014) and insulin receptors are widely expressed in the brain, e.g., occurring in hippocampus and striatum (Arnold et al., 2018). Recent studies indicate that insulin contributes to cognition in animals and is being tested as a possible therapeutic option in AD (Clarke, Ribeiro, Frozza, De Felice, & Lourenco, 2018;Grillo et al., 2015). The icv-STZ model induces AD-like symptoms such as memory impairment, glucose hypometabolism, mitochondrial dysfunction, increased oxidative stress, neuroinflammation, and neurodegeneration (Knezovic et al., 2015;Salkovic-Petrisic & Hoyer, 2007). More than 50 studies have used this model to investigate potential therapeutic effects of a variety of drugs, including lithium, NMDA receptor antagonists, antioxidants, pioglitazone, and many more (reviewed by Salkovic-Petrisic, Knezovic, Hoyer, & Riederer, 2013).
Two aspects of this model, however, have not been thoroughly investigated. First, the mechanism of STZ action in the brain remains unknown. In the periphery, STZ is taken up into cells (e.g., pancreatic ß-cells) preferentially via the GLUT2 glucose transporter receptor (SLC2A2) which is poorly expressed in the brain. The effects of insulin on blood glucose are mediated by the GLUT4 glucose transporter (SLC2A4). In the brain, however, the insulin-insensitive glucose transporters GLUT1 and 3 are predominant, with GLUT1 preferentially expressed in endothelia and glial cells and GLUT3 in neurons. Second, in humans, central cholinergic dysfunction is closely associated with cognitive function (Ballinger, Ananth, Talmage, & Role, 2016), and current drug treatment of AD is largely dependent on acetylcholinesterase (AChE) inhibitors which increase acetylcholine (ACh) levels in the brain. However, the effects of icv-STZ administration on central cholinergic systems have not been investigated in depth. Some studies determined activities of choline acetyltransferase (ChAT) or acetylcholinesterase (AChE), but these enzymatic activities do not correlate with ACh release. After icv-STZ administration, ChAT activity was reduced in the hippocampus to varying degrees, but was unchanged in the septum (Hellweg, Nitsch, Hock, Jaksch, & Hoyer, 1992;Terwel, Prickaerts, Meng, & Jolles, 1995). ChAT-positive cells were reduced in one study in the basal forebrain (Majkutewicz et al., 2016). AChE activity was found to be increased, and total ACh levels decreased, in some studies (Costa et al., 2016;Sorial & El-Sayed, 2017), but both measurements were done in brain homogenates and, therefore, poorly reflect local cholinergic activity. Another study reported an unchanged number of basal forebrain cholinergic neurons (based on ChAT immunoreactivity) or hippocampal cholinergic terminals (based on vesicular acetylcholine transporter (VAChT) staining) (Shoham, Bejar, Kovalev, Schorer-Apelbaum, & Weinstock, 2007). As cholinergic dysfunction is an important characteristic of AD, a more detailed investigation into cholinergic activity seemed warranted. We here report microdialysis findings of cholinergic activity in icv-STZ-treated rats, and we compare our findings with metabolite and mitochondrial data to monitor changes in energy metabolism.

| Animals and treatments
Icv-STZ-induced brain injury was induced in male Wistar rats (RRID:RGD_13508588) of 150-250 g body weight (5 weeks of age; Janvier Labs, Le Geneste St. Isle, France). A total of 103 rats were housed in a controlled room (22°C, 50%-65% humidity; day/night cycle of 12/12 hr) with free access to water and standard laboratory diet (Altromin; 1324, 2018) and were numbered Rat 1 to Rat 103.
After 7 days of acclimatization, they were divided into 2 or 3 groups per cage and block randomization according to 3 or 4 rats per cage was used. Rats were divided to receive different treatment in the following order. For two groups per cage: AACC, ACCA, ACAC, CAAC, CACA, and CCAA. For three groups per cage: ABC, ACB, BAC, BCA, CAB, and CBA (Table S1).
For treatments, the high-dose group (group C) received icv-STZ twice at a dose of 1.5 mg/kg (Sigma-Aldrich S0130, 2018) dissolved in water (Fresenius Kabi, 406760000, 2018) on day 0 and day 3. The low-dose group (group B) received icv-STZ once at a dose of 0.6 mg/ kg in saline. The control group (group A) received saline twice on days 0 and 3 (B. Braun, 2350720, 2018). Freshly prepared streptozotocin maintains its effects at neutral pH as reviewed by Deeds et al. (2011). During the experiments, the experimenter was unaware of the animal's group during raw data calculation and data analysis because during those times, the rats were labeled as number 1-103 by another researcher, and de-blinding occurred after data analysis.
Bupivacain (Jenapharm, 30004090000, 2018) was applied for long-lasting local anesthesia. The diffusion of drugs was confirmed by an infusion of fast green dye (Sigma-Aldrich F7258, 2018). Health status was monitored daily and body weights were checked twice a week for 21 days before experiments. The distribution of the 103 rats into experimental groups is illustrated in Table S1. As described above, the rats received saline or STZ (1.5 mg/kg) by i.c.v. injection on days 0 and 3. In the case of the low-dose STZ group, 0.6 mg/kg was administered once by i.c.v. infusion on day 0. This low dose of icv-STZ was only used for microdialysis studies. On day 20 after the initial injection, microdialysis probes were inserted under isoflurane anesthesia as described below. On days 21 and 22, microdialysis experiments were performed (day 21: open field; day 22: scopolamine).
A total of 45 rats were used for this part of the study, but because of blocked probes, 3 rats had to be excluded from analysis so that 42 rats were entered into data analysis. For the other experimental procedures, only two groups of rats were used, saline-treated rats and rats that received high-dose STZ. We used two groups of 9 rats each for histology, two groups of 8 rats each for the measurement of cholinergic parameters, and two groups of 12 rats each for measurement of mitochondrial activities. The flow diagram of the experiments is illustrated in Figure 1.
Both animal treatments and experiments were done in mixed order (saline, STZ; STZ, saline). Experiments were carried out according to the German Law for Animal Protection and were registered with Regierungspräsidium Darmstadt (FR/1005). The experimenter was unaware of the animal's group during raw data calculation and data analysis. The study was not pre-registered and exploratory.

| Mitochondrial oxygen consumption
Twenty-one days after the first icv-STZ injection, hippocampi of 24 rats (12 rats of the saline group and 12 rats of the highdose STZ group) were collected quickly after decapitation and mitochondria were isolated according to Hagl et al. (2013).
Rats were allowed to recover over night and microdialysis experiments were conducted for 2 days after probe implantation between 09:00 and 17:00 hr. The probes were perfused with artificial cerebrospinal fluid (147 mM NaCl, 4 mM KCl, 1.2 mM CaCl 2 , and 1.2 mM MgCl 2 ) at a rate of 2 µl/min. After 30 min of equilibrium between perfusion liquid and tissue, dialysates of 15 min intervals were collected for 90 min (6 × 30 µL) and used for analysis.
For extracellular acetylcholine level measurement, the artificial cerebrospinal fluid was supplemented with 0.1 µM neostigmine (Sigma-Aldrich, N2001, 2015) to stabilize extracellular levels of ACh. After collection of the first 90 min dialysates (basal data), rats were moved to an open field box (35 × 32 × 20 cm) as exploration and increased locomotor activity are known to physiologically stimulate hippocampal ACh release (König, Berlin, et al., 2018).
After collection of the second 90 min dialysates, the rats were transferred back to the microdialysis cage and dialysates were collected for another 90 min.
On day 2, after collection of dialysates under basal conditions, 1 µM scopolamine (Sigma-Aldrich, S0929, 2015) was added to the perfusion solution to pharmacologically stimulate ACh release, and dialysates were collected for 90 min followed by another 90 min of dialysis without scopolamine. Afterwards, rats were killed in isoflurane anesthesia and the brains were removed for further analysis and probe location confirmation.

| Determination of acetylcholine and choline
ACh and choline levels were determined by directly injecting 10 µl dialysate into a high-performance liquid chromatograph coupled with an electrochemical detector system (HPLC-ECD). The system consisted of an autosampler, a pump with degasser, a pre-column, a separation column, an enzyme reactor (AChE and choline oxidase), and an electrochemical detector (a platinum electrode working at + 500 mV by Eicom, Kyoto, Japan). The mobile phase consisted of KHCO 3 50 mM, EDTA-Na 2 134.3 mM, and sodium decane-1-sulfonate 1.64 mM in RotisolV® HPLC gradient grade water with a pH of 8.4 and flow rate at 150 µl/min.

| Cholinergic parameters (HACU, ChAT, AChE)
A total of 16 rats from the saline group and high-dose STZ group (8 rats each) were used for the study of cholinergic parameters. The hippocampus and striatum were collected from each rat and homogenates of each brain region were made. Each homogenate was divided into three subsamples for HACU, ChAT, and AChE studies.

| Single-cell sequencing analyses
Raw single-cell expression data of the adult mouse nervous system (Zeisel et al., 2018) were downloaded from http://mouse brain. org/downl oads.html in two resolution levels: at the single-cell level and at cell-type level. Data were imported into the R environment using R/loomR (https://github.com/mojav eazur e/loomR) and R/ LoomExperiment (Morgan & Van Twisk, 2019) and analyzed regarding cholinergic transcripts and glucose transporters. Results were visualized using R/ggplot2 (Wickham, 2016).
Correlation of expression of the selected genes was determined via Pearson's product-moment correlation coefficient (PPMCC, using the cor() function of R/stats). P-values for each PPMCC were calculated for the entire gene-sample matrix using the cor_pmat() function, and plotted using the ggcorrplot() function of R/ggcorrplot (https://github.com/kassa mbara/ ggcorrplot). Correlation was assumed significant at p < .05. All analyses were performed using R software version 3.5.1 ("Feather Spray").

| Statistical analysis
If not indicated otherwise, data are presented as means ± SEM of N animals. All data were tested for normal distribution by the Kolmogorov-Smirnov test (GraphPad Prism 5.03). Potential outliers (>2 SD) were identified by the Grubbs test (https://www.graph pad.com/quick calcs/ grubbs). Sample size was calculated using N = 2 SD 2 × power index/ delta 2 . From many years of experience with microdialysis, we expected an SD of 20% and a treatment effect of 25%. The value for the power index (α = 0.05, two-sided; for multiple pair-wise comparisons. One-way ANOVA followed by Tukey's multiple comparison test was used to compare means between three groups on basal values of metabolites (Table 1) and on metabolites (glucose, lactate; Figure S2). Means between two groups were compared by Student's t-test for mitochondrial respiration (Figure 4), cholinergic parameters (Tables 2 and 3), and quantification of fluorescence intensity ( Figure S1). p-values <.05 were considered to be statistically significant. All data in Figures 2, 4-6, Tables 1-3, and Figure S1 and S2 were normally distributed, and no outliers were detected. Table S1 shows the number of excluded animals as a result of failure in dialysates collection because of damaged probe during implantation or incorrect dialysates volume.

| Icv-STZ induces toxicity, gliosis, and neurodegeneration in hippocampus
Icv-STZ was clearly toxic to the animals as seen in the development of body weight (Figure 2). While the rats developed normally after administration of the low dose (0.6 mg/kg STZ), there was initially no increase in body weight after administration of the high dose of icv-STZ, followed by a period of weight gain. At the end of the experiment (on day 21), high-dose STZ-treated rats had 15% less body weight compared to controls (saline 331 ± 35 g versus high dose 283 ± 24 g, N = 15; p < .01).
On day 21, rats were killed and histology was performed to monitor cellular toxicity. The immunohistochemical study with the GFAP primary antibody revealed glial activation after high-dose icv-STZ that was visible in both dorsal and ventral hippocampus, and especially at the CA1 region (Figure 3a and b). The quantification of GFAP staining showed a reproducible and highly significant increase in GFAP expression ( Figure S1a). Fluoro-Jade C staining which indicates apoptosis was also positive in the hippocampal CA1 region, however, only 3 of 9 rats showed a strong response (Figure 3c and Figure S1b).

| Metabolic screening after icv-STZ
We used microdialysis to monitor various indicators of energy metabolism and cellular toxicity. We found that icv-STZ did not affect extracellular glucose levels in the brain ( Figure S2a).
There was also no significant change in glucose levels during behavioral or pharmacological stimulation ( Figure S2a). Lactate levels in brain microdialysates were reduced by 24% after both low-as well as high-dose STZ ( Figure S2b) while pyruvate levels were increased (Table 1), however, the changes were not statistically significant ( Table 1). The extracellular concentration of ß-hydroxybutyrate (a ketone body) was unchanged (Table 1).
Interestingly, glycerol and especially choline, both breakdown products of membrane phospholipids, were increased by STZ (Table 1; p < .05 for choline). The concentration of isoprostanes, which indicate lipid peroxidation and oxidative stress, was unchanged following STZ (Table 1).
In addition, we measured mitochondrial oxygen respiration in mitochondria that was isolated from rat hippocampi 21 days after high-dose icv-STZ. After normalization of the data for citrate synthase activity, all complexes showed a reduction in oxygen consumption by 9%-15% that was statistically significant for FIGURE 2 Body weight of rats after treatment with saline (twice) or streptozotocin (STZ) at low (once 0.6 mg/kg) or high (twice 1.5 mg/ kg) doses. Data are means ± SD of N = 17-20 animals. Statistics: two-way ANOVA with Bonferroni post-test: **p < .01 versus saline. Note: Data are means ± SEM (N = 6 animals) and were analyzed by one-way ANOVA followed by Tukey's multiple comparison test. *p < .05 versus saline.

TA B L E 1 Extracellular concentrations of metabolites as determined by microdialysis under basal conditions (not corrected for recovery)
complexes II (succinate dehydrogenase) and IV (cytochrome oxidase) (Figure 4).

| Central cholinergic activity after icv-STZ
To follow cholinergic activity in situ in real time, we used microdialysis and an HPLC-based, highly sensitive detection method for acetylcholine (ACh). The results of these studies are shown in Figure 5 (hippocampus) and 6 (striatum). Notably, basal levels of extracellular ACh in hippocampus were unchanged after icv-STZ (saline 8.71 ± 1.33 nM, low dose 7.29 ± 0.84 nM, high dose 7.48 ± 1.58 nM, N = 6; p > .3) (Table 1) On day 2, an infusion of scopolamine was used to increase ACh release. All groups responded to the treatment with a threefold increase in ACh levels ( Figure 5b) and showed similar maximum increases at 75 min (saline 36.10 ± 5.94 nM, low-dose STZ 29.19 ± 8.17 nM, high-dose STZ 34.2 ± 6.31 nM; N = 6, p > .3).
Again, two-way ANOVA revealed no significant difference between groups.
As the lack of cholinergic response seemed surprising, we repeated these experiments in the Corpus striatum, a brain area with rich cholinergic innervation that was exposed to STZ after i.c.v. administration. As shown in Figure 6, basal ACh levels in the striatum were higher than in the hippocampus but still similar between experimental groups (p > .05). Surprisingly, the highest ACh values were observed after high-dose STZ (Figure 6a). While the response to the open field was less prominent in the striatum than in the hippocampus, twoway ANOVA revealed a significant difference between high-dose STZ and saline which was strongest at time 90 min of the open field test (p < .05; Figure 6a). Similar to the hippocampal data, scopolamine caused a more than threefold increase in ACh release in the striatum, but STZ treatment did not modify the response of cholinergic neurons in this paradigm (Figure 6b).

| Single-cell sequencing analyses
The apparent lack of specific toxicity toward cholinergic neurons brought up questions about the mechanism of STZ toxicity in the

TA B L E 3
Cholinergic parameters as determined in rat striatum 3 weeks after icv-STZ administration brain. Since, in the periphery, STZ-mediated toxicity depends on specific glucose transport (reportedly via GLUT2), we elected to analyze recent single-cell expression data with regard to cholinergic markers and the relevant glucose transport proteins. We used raw single-cell expression data of the adult mouse nervous system (Zeisel et al., 2018) to determine expression and correlation of cholinergic and glucose transporter genes. Expression levels of cholinergic markers (AChE; ChAT; and SLC5A7, also known as CHT-1, the high-affinity choline transporter) and glucose transmembrane transporters (SLC2A1, SLC2A2, SLC2A3, and SLC2A4; also known as GLUT1, GLUT2, GLUT3, and GLUT4, respectively) in selected brain regions (hippocampus and striatum) were determined at single-cell level, and displayed via violin plot of log 2 (expression + 1) ( Figure S3). The data show a widespread expression of AChE but a low expression of ChAT and CHT-1, as is consistent with the literature. Among the glucose transporters, GLUT 3 is widely expressed, GLUT1 and GLUT4 are found in hippocampus and striatum in small amounts, while GLUT2 mRNA is basically absent apart from the ventral striatum.
Correlation coefficients of expression of the selected genes were calculated for all 265 tissues of the cell-type-level data of the murine nervous system ( Figure S4a), and at the single-cell-level for all neurons ( Figure S4b) and for cholinergic neurons alone ( Figure   S4c; defined as expressing the vesicular ACh-transporter SLC18A3).
We observed high correlations between ChAT and SLC5A7 (CHT- Substrates, uncoupler and inhibitors were used to measure each complex as described in Methods. Raw data as obtained from respirometry were normalized to mitochondrial citrate synthase activity. Statistical analysis: Data are means ± SEM (N = 12 animals) and were analyzed by unpaired t-test. *p < .05 compared to saline

| D ISCUSS I ON
The present data can be interpreted in two ways, namely, (a.) referring to STZ´s mechanism of action in the brain and (b.) to STZ´s effects on the central cholinergic system.

| Mechanism of action of streptozotocin in the brain
Reduced glucose consumption is an early sign of Alzheimer´s disease (AD), preceding clinical symptoms of memory loss and dementia by several years. As diabetes is known to increase the risk for AD by 1.5-to 2-fold, AD has been hypothesized to be an insulin-resistant state of the brain and has even been termed "diabetes type 3" (de la Monte, 2014). However, the connection among diabetes, insulin, and AD is far from proven (Salas & de Strooper, 2019;Schilling, 2016).
While animal studies suggest a role for insulin in cognition, the situation in the human is more complex ( Reagan, 2015). Streptozotocin (STZ) is a drug that, after systemic administration, induces experimental type 1 diabetes in rats and mice by preferential uptake via GLUT2 glucose transporters into pancreatic ß-cells and subsequent cellular toxicity (alkylation and DNA damage) (Grieb, 2016). The icv-STZ model was developed to imitate this mechanism in the brain, although there is little evidence that the brain contains insulin-producing cells (Blazquez, Velazquez, Hurtado-Carneiro, & Ruiz-Albusac, 2014). The mechanism of action of STZ in the brain, however, is far from clear (Grieb, 2016).
According to present knowledge, the glucose transporters GLUT2 and the insulin-dependent GLUT4 are present in the brain in very low amounts, whereas the glial GLUT1 and the neuronal GLUT3 are highly expressed. Our analysis of murine single-cell expression data confirms that GLUT2 is basically absent from the brain. While GLUT4 was suggested to play a major role in insulin action in the brain (McNay and Pearson-Leary, 2020), our analysis shows low expression and no correlation of GLUT4 expression with cholinergic markers. More likely, then, STZ may enter brain cells through the ubiquitous GLUT1 and GLUT3 transporters. Hence, a specific effect of STZ on a subset of brain cells is unlikely; more likely, many neuronal cell types as well as glial cells will be exposed to STZ toxicity after i.c.v. injection. According to our analysis ( Figure S4), GLUT3 expression has a significant correlation with cholinergic genes, hence, STZ may enter cholinergic cells via GLUT3 (SLC2A3).
Following icv-STZ, we found GFAP-positive staining in all treated rats which indicates astrogliosis. This finding corroborates previous reports of reactive gliosis in this model (Knezovic et al., 2017;Mishra, Singh, Shukla, & Shukla, 2018;Rai, Kamat, Nath, & Shukla, 2014;Santos, Mazucanti, Xavier, & Torrão, 2012). We also found cells that stained positive for Fluoro-Jade C (FJ) in some but not all of the animals. Fluoro-Jade C is an acidic dye postulated to label basic molecules of degenerating neurons. It is known that FJ staining is strongly time dependent and may disappear after several days (Santos et al., 2012); this may explain the lack of staining in several animals.
Additional signs of neurodegeneration were observed in long-lasting studies (Knezovic et al., 2015;Salkovic-Petrisic et al., 2011). These signs of toxicity may not have shown up in our study because we used the most popular regimen, i.e., testing of the animals 3 weeks after STZ administration. Importantly, icv-STZ rats gained very little weight for 2 weeks after administration of STZ, a finding that clearly reflects the toxicity of the compound. It is remarkable that throughout our experiments, the low dose of 0.6 mg/kg (once) did not cause any significant changes in the parameters measured; the high dose (twice 1.5 mg/kg) was required to induce toxicity. We also found signs of toxicity in the microdialysis study: Increases in choline and glycerol are known to signify membrane damage because they are breakdown products of cellular membranes (Klein, 2000;de Lima Oliveira et al., 2014). Taken together, our findings demonstrate that the usual dose of icv-STZ induces severe toxicity for 10 days followed by a recovery period, but even after 3 weeks, signs of neurodegeneration and astrogliosis are clearly visible.
Although abnormalities in glucose metabolism are implicitly assumed to occur in the STZ model, we here show that extracellular glucose levels in the brain are not affected by icv-STZ. We found a decrease in lactate levels after icv-STZ, but the effect did not reach statistical significance. In the absence of ischemia, lactate in the brain is mainly formed by astrocytes to supply neurons with metabolic energy (Dienel, 2012), and a reduction in lactate may be caused by glial dysfunction induced by astrogliosis (see above). Similarly, decreased lactate levels were observed in other AD models (Lu et al., 2015;Zhang et al., 2018).
We also measured glucose and lactate levels in dialysates during behavioral and pharmacological stimulation. Similar to earlier findings (Hartmann, Kiewert, & Klein, 2010), glucose levels did not change significantly, whereas lactate levels were increased during the open field test. The latter finding is probably as a result of neuronal activation during exploration (Hartmann et al., 2010). Again, all treatment groups responded similarly during stimulation.
To further characterize metabolic dysfunction in the icv-STZ model, we isolated hippocampal mitochondria and determined the activities of the complexes of the electron transport chain. For this purpose, we used high-resolution respirometry combined with substrates-uncoupler-inhibitors titration (SUIT) protocol developed previously (Hagl et al., 2013). Similar to previous work (Correia et al., 2013;Paidi et al., 2015), we found reduced oxygen consumption in hippocampal mitochondria. In our hands, all complexes showed reduced activity with the most significant changes occurring in complexes II and IV. Reductions in complex I were not statistically significant, and in agreement with this finding, we did not observe a change in isoprostane levels in dialysate samples which means that oxidative stress was not visibly increased. Of note, a major reduction in mitochondrial activity would be expected to reduce pyruvate consumption in the TCA cycle and to increase the lactate/pyruvate ratio, a response that was not observed here.  (Chen et al., 2012). The factors influencing AChE expression are well known (Soreq & Seidman, 2001), however, our measurements in tissue homogenates do not allow a distinction between changes in expression in specific cell types.

| Effects of streptozotocin on central cholinergic transmission
In the present study, we used microdialysis to measure extracellular levels of ACh and its changes during behavioral or pharmacological stimulation. Surprisingly, basal levels of extracellular ACh (approx. 7 nM) in the hippocampal microdialysates were unchanged 3 weeks after icv-STZ. Exploration of the open field caused an increase in ACh release that was more pronounced in hippocampus than in striatum. However, ACh release was not impaired after icv-STZ; in fact, there were slightly higher ACh levels in the striatal dialysates after icv-STZ. In both brain regions, the infusion of scopolamine caused a several-fold increase in ACh levels as a result of the blockade of negative feedback through M2/M4-type receptors (Mohr et al., 2013). Again, the response of the cholinergic pathways was identical in all groups. Evidently, icv-STZ did not cause a major dysfunction of central cholinergic pathways.
Our measurements of cholinergic parameters and the bioinformatic analyses of murine single-cell data allow us to draw a differential picture of the cholinergic response in two brain areas with prominent cholinergic innervation, namely, hippocampus and striatum. Two limitations must be noted for this analysis. First, as discussed by Grieb (2016), the pattern of damage is influenced by the limited diffusion of STZ from the ventricular space. In preliminary experiments (data not shown), we found that a dye-injected i.c.v.
reached hippocampal and striatal tissue but not the basal forebrain (e.g., the septal area where hippocampal cholinergic fibers originate).
This corroborates an earlier study that also failed to see cholinergic damage in the septal area (Terwel et al., 1995). Second, the analyzed single-cell data were of murine origin, and for the present purpose we have to hypothesize that the expression of cholinergic transcripts and glucose transporters is similar in rats and mice.
These limitations notwithstanding, we can conclude from our data that icv-STZ causes a mild reduction in cholinergic parameters in the hippocampus which is, however, not reflected in a change in extracellular ACh levels. It seems likely that the reduction in ChAT and HACU activities is caused by STZ entering cholinergic axons or nerve terminals in the hippocampus through GLUT3 transporters.
Subsequent toxicity may damage some cholinergic neurons which may cause a reduction in ChAT activity which is, however, not rate limiting for ACh synthesis and release (Brandon et al., 2004;König, Berlin, et al., 2018). Concomitantly, STZ may also enter neighboring neurons or astrocytes through GLUT1 or GLUT3 and reduce the expression of AChE. AChE is more closely coupled to extracellular ACh levels (Erb et al., 2001), but a reduction in AChE would evidently increase extracellular ACh. We hypothesize that the small changes in ChAT and AChE activities do not influence ACh synthesis and degradation. Alternatively, the small reduction of ChAT activity may lower ACh release, but this effect may be balanced by the small reduction in AChE activity which would prolong its lifetime. Clearly, a cholinergic dysfunction does not result from these effects.

| CON CLUS ION
Taken together, these data show preserved cholinergic function 3 weeks after icv-STZ administration. In the hippocampus, cholinergic nerve terminals may sustain some minor damage that does not, however, impair their ability to release ACh during behavioral or pharmacological stimulation. In the striatum, cholinergic interneurons seem to be protected from STZ toxicity. In contrast, we saw clear signs of astrogliosis, neurodegeneration, and some impairment of energy metabolism. Hence, the icv-STZ model lacks an important component of AD pathology, namely, cholinergic dysfunction, and, therefore, is a poor model of human AD. It may be more correctly qualified as an animal model of hippocampal gliosis and toxicity.