Glucagon-Like Peptide-1 Cleavage Product Improves Cognitive Function in a Mouse Model of Down Syndrome

Abstract Currently there is no effective therapy available for cognitive impairments in Down syndrome (DS), one of the most prevalent forms of intellectual disability in humans associated with the chromosomes 21 trisomy. Glucagon-like peptide-1 (GLP-1) is an incretin hormone that maintains glucose homeostasis by stimulating insulin secretion. Its natural cleavage product GLP-1 (9-36) lacks insulinotropic effects and has a low binding affinity for GLP-1 receptors; thus, GLP-1 (9-36) has historically been identified as an inactive metabolite. Conversely, recent work has demonstrated interesting physiological properties of GLP-1 (9-36) such as cardioprotection and neuroprotection. We have previously shown that GLP-1 (9-36) administration enhances neuronal plasticity in young WT mice and ameliorates cognitive deficits in a mouse model of Alzheimer’s disease. Here, we report that systemic administration of GLP-1 (9-36) in Ts65Dn DS model mice of either sex resulted in decreased mitochondrial oxidative stress in hippocampus and improved dendritic spine morphology, increase of mature spines and reduction of immature spines. Importantly, these molecular alterations translated into functional changes in that long-term potentiation failure and cognitive impairments in TsDn65 DS model mice were rescued with GLP-1 (9-36) treatment. We also show that chronic GLP-1 (9-36) treatment did not alter glucose tolerance in either WT or DS model mice. Our findings suggest that GLP-1 (9-36) treatment may have therapeutic potential for DS and other neurodegenerative diseases associated with increased neuronal oxidative stress.


Introduction
Down syndrome (DS) is one of the most common forms of intellectual disabilities with an apparent genetic cause: the trisomic repeat of chromosome 21 (HSA21; Patterson and Costa, 2005;Mégarbané et al., 2009;Lott and Dierssen, 2010 ). Impaired cognition is a hallmark of DS, and there are currently no effective treatments that improve cognitive function in individuals with DS (Patterson, 2009;Lana-Elola et al., 2011;Lott, 2012). Of note, nearly all DS patients develop typical Alzheimer's disease (AD) neuropathology (i.e., senile plaques and neurofibrillary tangles) by age 40 (Nelson et al., 2011;Head et al., 2016), and many go on to develop age-dependent AD type dementia syndromes later in life (Dierssen, 2012;Zigman, 2013;Bayen et al., 2018).
Increased reactive oxygen species (ROS) is a common feature of many neurologic diseases of cognitive impairment, including DS (Lin and Beal, 2006;Pagano and Castello, 2012;Wang et al., 2014;Guo et al., 2017). Excessive neuronal ROS disrupts molecular and cellular mechanisms underlying memory and synaptic plasticity maintenance and can lead to cognitive impairment Guo et al., 2017). Previous studies have demonstrated that decreasing mitochondrial ROS improves memory and hippocampal synaptic plasticity impairments associated with neurodegenerative diseases such as AD (Dumont et al., 2009;Ma et al., 2011;Massaad and Klann, 2011). Therefore, targeting mitochondrial ROS production or clearance may be a feasible strategy to improve cognitive function in individuals with DS.
In the current study, we aim to investigate whether GLP-1 (9-36) treatment can improve DS pathophysiology by administrating GLP-1 (9-36) systemically to a mouse model of DS. A series of experiments were conducted to test effects and relevant cellular/molecular mechanisms of GLP-1 (9-36) treatment on cognitive impairments and synaptic plasticity deficiency in Ts65Dn DS model mice.
Our findings suggest that GLP-1 (9-36) may be a novel therapeutic agent in treating DS-associated cognitive dysfunction and synaptic failure.

Animals
All mice were housed in the Transgenic Mouse Facility at Wake Forest School of Medicine Animal Facility. Mice were kept in compliance with the National Institute of Health (NIH) Guide for Care and Use of Laboratory Animals. The facility kept a 12 h light/dark cycle with regular feeding, cage cleaning, and 24 h access to water. Breeder mice were purchased from The Jackson Laboratory. The mouse colony was maintained by breeding Ts65Dn trisomic females (005252) with B6EiC3Sn.BliAF1 males (003647). Ts65Dn mice used in behavioral experiments did not carry the phosphodiesterase 6b (Pde6b) gene mutation associated with retinal degeneration. Genotyping was determined by PCR. All experiments were conducted on male and female 9-month-old mice; Ts65Dn mice reliably exhibit cognitive impairments at this age based on our observation and previous report (Faizi et al., 2011).

Detection of mitochondrial superoxide in hippocampal slices
After completion of either drug or vehicle treatment, mice were killed and 400 m hippocampal slices were incubated with 5 M MitoSOX Red, a mitochondrial superoxide indicator (prepared as 5 mM stock solution immediately before the experiments; Invitrogen) for 10 min. Slices then were fixed with ice-cold 4% paraformaldehyde in PBS overnight at 4°C. Slices were further cut into 40 m sections and mounted onto pre-subbed slides with VECTASHIELD mounting medium with DAPI (Vector Laboratories). The sections were imaged using a Leica TCS SP5 confocal microscope at 20ϫ magnification. All parameters (pinhole, contrast, gain, offset) were held constant for all sections from the same experiment.

Western blot
Hippocampal slices were flash-frozen on dry ice and sonicated as previously described (Ma et al., 2012). Samples containing equal amounts of protein lysate were loaded on 4 -12% Tris-glycine SDS-PAGE gels for standard gel electrophoresis. Membranes were probed overnight at 4°C using the OXPHOS primary antibody (1:250; Abcam) and GAPDH (1:10,000; Cell Signaling). Densitometric analysis was performed using ImageJ software. Data were normalized to GAPDH.

Golgi-Cox stain
Brains were processed using the FD Rapid GolgiStain Kit in accordance with the manufacturer's instructions (FD Neurotechnologies; catalog #PK401). Transverse sections (125 m) were made using a Leica VT1200S vibratome and mounted unto gelatin-coated slides. Development was performed according to kit instructions. Sections were dehydrated through a graded ethanol series and cleared in xylene. Slides were cover-slipped with Vectamount Permanent Mounting Medium (Vector Laboratories; catalog, #H-5000) and imaged at 100X on a Keyence BZ-X710 microscope. For spine analysis, images were blinded, and spines were manually counted and sorted as previously described (Risher et al., 2014).

Glucose tolerance test
An intraperitoneal glucose tolerance test (IPGTT) was administered to mice before the treatment regimen began and after daily saline or GLP-1 (9-36) treatment (500 ng/g/d). On the day of the experiment, mice were fasted for 4 h and glucose (2.0 g/kg) was administered intraperitoneally. Blood samples were taken from tail veins and blood glucose was measured (0,15,30,45,60,90, 120 min, respectively) using a glucometer (Bound Tree Medical Precision XTRA Glucometer; Fisher).

Hippocampal slice preparation and electrophysiology
Acute 400 m transverse hippocampal slices were prepared using a Leica VT1200S vibratome as described previously. Slices were maintained at room temperature for at least 2 h in an artificial CSF containing the following (in mM): 118 NaCl, 3.5 KCl, 2.5 CaCl 2 , 1.3 MgSO 4 ,1.25NaH 2 PO 4 , and 15 glucose, bubbled with 95% O 2 /5% CO 2 . For electrophys-iology, monophasic constant-current stimuli (100 s) were delivered with a bipolar silver electrode placed in the stratum radiatum of area CA3. Field EPSPs (fEPSPs) were recorded using a glass microelectrode from the stratum radiatum of area CA1. LTP was induced using a high-frequency stimulation (HFS) protocol consisting of two 1 s, 100 Hz trains separated by 60 s. Paired-pulse facilitation (PPF) was conducted using a pair of stimuli with interstimulus intervals (ISIs) of 25, 50, 100, 200, or 300 ms delivered to the CA3, and the respective slopes of the fEPSP were measured. The ratio of the second slope to the first slope, compared with the ISI was plotted. For input/output (I/O) curves, the slopes of fEPSPs and the values of the fiber volley at different stimulation intensities were measured.

Object location memory task
Mice were habituated to an opaque plastic chamber (30 ϫ 21 ϫ 15 cm) with visible spatial cues for 10 min. After 24 h, mice were returned to the chamber with two identical objects in the arena and were allowed to freely explore and interact with the objects for 10 min. Twentyfour hours later, mice were returned to the chamber again, where one of the two objects had been relocated to an adjacent position (Fig. 1A). Objects and changes in object location were randomly determined and counterbalanced. Time spent with each object was measured and calculated as a percentage of the total object interaction time. Novel object preference of Ͻ50% indicates memory impairments. Time with objects was measured both manually and with EthoVision XT tracking software. Mice with a total object interaction time of Ͻ10 s were excluded from analysis. Data collection and analysis were performed blinded.

Passive avoidance task
For the training phase of the passive avoidance task (PAR apparatus, Panlab), mice were placed in a well-lit chamber (25 ϫ 25 ϫ 24 cm). Following a 60 s exploration period a trap door opened to reveal a smaller, dark chamber (19 ϫ 10 ϫ 12 cm). On entry into the dark chamber the trap door closed, and mice were given a mild foot shock (foot shock intensity: 0.1 mA, 2 s duration). Memory was tested 24 h after training when mice were returned to the light chamber, and following a 60 s exploration period, the trap door opened to reveal the dark chamber. On entry into the dark chamber the trap door closed (no foot shock was given on the test day). For training and test day, mice were given a max latency of 600 s. Mice that did not enter the dark chamber on the first day were not included in data analysis.

Data analysis
Data were presented as mean Ϯ SEM. Summary data are presented as group means with SE bars. Normality of the data were tested using GraphPad Prism (GraphPad Software), and the criteria for parametric testing were met. For comparisons between two groups, a two-tailed independent Student's t test was performed using Graph-Pad Prism 7 software. Two-tailed unpaired t tests were performed for within-group analyses. For comparisons between more than two groups, two-way repeated mea- sures ANOVA was used with Tukey's post hoc tests for multiple comparisons. Error probabilities of p Ͻ 0.05 were considered statistically significant.

Experimental design and statistical analysis
All statistical comparisons were done in four mouse groups: WTϩ saline, Ts65Dn ϩ saline, WTϩ GLP-1 (9-36), and Ts65Dn ϩ GLP-1 (9-36). The number of mice (n) for each comparison is given in the corresponding figure legends. For each tested parameter, data distributions are represented by mean Ϯ SEM. For all tests, results were considered statistically significant at p Ͻ 0.05. All data analyses and statistics were performed in GraphPad Prism (GraphPad Software).

Study approval
All methods involving animals were approved by the IACUC of Wake Forest University School of Medicine.

GLP-1 (9-36) treatment decreases DS-associated elevated levels of mitochondrial superoxide
Increased neuronal mitochondria-derived ROS is an important feature of DS and is linked to cognitive impairments Zis et al., 2012). Previous studies have demonstrated that GLP-1 (9-36) treatment decreases ROS production both in cultured cells and animal models (Ban et al., 2010;Ma et al., 2012;Giacco et al., 2015). To determine whether GLP-1 (9-36) treatment is able to decrease DS-associated ROS, we assessed the levels of mitochondrial ROS by staining live slices with MitoSOX Red, a fluorogenic dye that selectively detects mitochondria-specific superoxide, a primary ROS (Ma et al., 2012;Wong et al., 2015;Ishii et al., 2017). Compared with WT mice, Ts65Dn mice demonstrated significantly enhanced MitoSOX fluorescence signal, indicating increased levels of mitochondrial superoxide and ROS (Fig. 1A). Notably, GLP-1 (9-36) treatment blunted the DS-associated increase in mitochondrial superoxide (Fig. 1A). We further examined the effects of GLP-1 (9-36) treatment on mitochondrial electron transport chain (ETC) proteins using Western blot assay. The ETC consists of a series of protein complexes that transfer electrons from electron donors to electron acceptors via redox reactions. The ETC proteins create the proton gradient that drives ATP production; however, ETC complexes I and III produce large amounts of superoxide in this process (Massaad and . We observed no differences in ETC expression according to either genotype or treatment condition (Fig. 1B), indicating that the difference in superoxide production was not because of dysregulations of ETC proteins expression. These findings are also consistent with previous reports demonstrating that GLP-1 (9-36) has antioxidant-like properties (Giacco et al., 2015).

GLP-1 (9-36) treatment does not affect glucose clearance in Ts65Dn or WT mice
GLP-1 (7-36), the precursor peptide of GLP-1(9-36), is known as an insulinotropic hormone; it enhances insulin secretion and promotes glucose clearance in humans. In contrast, GLP-1 (9-36) has low binding affinity for the GLP-1R and usually is not considered to have any insulinotropic effects (Vahl et al., 2003). Nevertheless, given that we are treating mice with GLP-1 (9-36) at supraphysiological doses, we sought to determine whether GLP-1 (9-36) affects glucose tolerance at the dose administered. To measure this, we performed an IPGTT (Macklin et al., continued experiments (20ϫ, scale bar, 50 m; n ϭ 3 per group). B, Western blot experiment revealed no significant difference in the expression of oxidative phosphorylation complex proteins among all four experimental groups (n ϭ 8 per group). Figure 3, saline or GLP-1 (9-36) treatment did not alter glucose tolerance in both WT and Ts65Dn mice, as indicated by measurement of the glucose x time clearance curve and the area under the curve (AUC; Fig. 3A-H).

GLP-1 (9-36) treatment alleviates cognitive deficits in Ts65Dn mice
To determine whether GLP-1 (9-36) treatment can rescue DS-associated cognitive deficits, we treated Ts65Dn mice and their WT littermates with GLP-1 (9-36) (500 ng/g/d; i.p.) or a saline control continuously for 14 d (Day et al., 2017). We then subjected mice to a series of behavioral tasks to evaluate their cognitive performance. We assessed the spatial learning and memory by testing the mice on object location memory (OLM) task (Kleschevnikov et al., 2012). In the OLM test, WT mice treated with either saline or GLP-1 (9-36) exhibited normal cognition, spending a significantly greater percentage of time interacting with the relocated object than with the object in the old location (Fig. 5A,B). In contrast, Ts65Dn mice treated with saline failed to recognize the relocated object and spent similar amounts of time with either object, indicating impaired spatial learning and memory. Notably, Ts65Dn mice treated with GLP-1 (9-36) displayed normal cognition, as indicated by a preference for objects in the novel location (Fig. 5B).

Discussion
In the current study, we have shown that GLP-1 (9-36), a natural peptide and the primary cleavage product of the incretin hormone GLP-1 (7-36), ameliorates cognitive deficits and synaptic plasticity impairments in a mouse model of DS. With regards to potential molecular/cellular mechanisms, we found that GLP-1 (9-36) treatment decreased the DS-associated elevated mitochondrial super-oxide, a major ROS that is linked to memory impairments and synaptic failure Guo et al., 2017). We also observed that GLP-1 (9-36) treatment led to increased mature dendritc spines in hippocampus. Together, these findings demonstrate that GLP-1 (9-36) may exert beneficial effects on the molecular mechanisms underlying impairments of memory and synaptic plasticity in DS. This is the first study we are aware of that has characterized dendritic spine morphology in Ts65Dn mouse model of DS. Spines are dynamic structures whose morphologic structure is directly related to function (Berry and Nedivi, 2017), representing an essential mechanism underlying memory and synaptic plasticity. Immature spines are highly dynamic structures that have little function in synaptic transmission but are essential in initiating contact with nearby axons to form synaptic connections (Lohmann et al., 2005). Furthermore, synaptic weakening induced by long-term depression destabilizes and eliminates existing spines (i.e., increases immature spine density; Nagerl et al., 2004;Zhou et al., 2004). Our findings that DS model mice have increased immature spine density are in line with previous work demonstrating that DS model mice have increased GABAergic innervation to the hippocampus (Mojabi et al., 2016). Conversely, mature Figure 5. GLP-1 (9-36) treatment alleviates cognitive deficits in Ts65Dn mice. A, Schematic of OLM task and object preference for familiar and new locations during the test session. B, Spatial memory was impaired in Ts65Dn mice compared with WT controls, and the impairments were rescued with GLP-1 (9-36) treatment. Preference for the new location Ͻ50% of the total interaction time indicates cognitive impairment (WTϩ Saline: n ϭ 10; Ts65Dn ϩ Saline: n ϭ 11; WTϩ GLP-1 (9-36): n ϭ 12; Ts65Dn ϩ GLP-1 (9-36): n ϭ 13; ‫‪p‬ءء‬ Ͻ 0.01, ‫‪p‬ءءء‬ Ͻ 0.001, independent t test). C, Schematic of passive avoidance paradigm and latency to dark compartment on Day 1 and Day 2 of the test. D, No differences were observed in latency to dark compartment on acquisition day (Day 1; one-way ANOVA; p Ͼ 0.05). On test day (Day 2), Ts65Dn mice had a significantly shorter latency to enter into the dark compartment than WT control groups, indicating cognitive impairments. In contrast, GLP-1 (9-36) treatment significantly increased latency to the dark compartment in Ts65Dn mice, indicating improved cognitive function. None of the WT mice treated with GLP-1 (9-36) entered the dark compartment on testing day. (WTϩ Saline: n ϭ 7; Ts65Dn ϩ Saline: n ϭ 3; WTϩ GLP-1 (9-36): n ϭ 9; and Ts65Dn ϩ GLP-1 (9-36): n ϭ 7). One-way repeated-measures ANOVA with Tukey's post hoc tests; ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01, ‫‪p‬ءءء‬ Ͻ 0.001, ‫‪p‬ءءءء‬ Ͻ 0.0001. spines form stable connections with presynaptic terminals, and are highly sensitive to glutamatergic transmission (Zito et al., 2009;Berry and Nedivi, 2017). Previous studies have shown that excitatory, glutamatergic transmission increases spine volume and stabilizes newlyformed spines (i.e., increases mature spine density; Lang et al., 2004;Kopec et al., 2006;Zito et al., 2009). The spine's ability to alter their morphology in response to increased or decreased synaptic activity has led some to categorize immature spines as "learning" spines, and mature spines as "memory" spines (Bourne and Harris, 2007). The increased immature spine density in salinetreated Ts65Dn mice may explain their memory impairments. Thus, the shift in spine composition (mature Ͼ Ͼ immature) observed in GLP-1 (9-36)-treated Ts65Dn mice may provide a mechanistic explanation for their improved memory and synaptic plasticity.
That the alterations in dendritic spine morphology were associated with decreased mitochondrial oxidative stress with GLP-1 (9-36) treatment is especially interesting. Increased neuronal oxidative stress is a pathologic hallmark of DS and is associated with aging and age-related neurodegenerative diseases (Zana et al., 2007;Dumont et al., 2009;Ma et al., 2011;Massaad and Klann, 2011;Zis et al., 2012;Wang et al., 2014;Panel et al., 2018). AD model mice have elevated mitochondrial superoxide production and impaired memory and synaptic plasiticity (Ma et al., 2012). The association beetween oxidative stress and dendritic spine morphology remains elusive. One study found that a single dose of cranial irradiation, which causes a persistent increase in ROS, caused significantly decreased immature spines and decreased mature spines (Chakraborti et al., 2012). Another study showed that rats with excessive neuronal ROS had a low mature spine density, which was restored with antioxidant treatment (Dos Reis et al., 2013). In the context of our own work, these studies demonstrate a possible causative relationship between excessive ROS and dendritic spine morphology in DS. Meanwhile, future studies are warranted to determine the detailed molecular mechanisms through which oxidative stress interfers with spine maturation in DS.
Currently, there are no effective pharmacological therapies capable of improving cognition in DS patients. We are excited by the findings that GLP-1 (9-36) treatment is able to correct aberrant cognition and synatpic plasticity in the DS model mice. Moreover, the observation that DS-associated increased levels mitochondrial superoxide is blunted by GLP-1 (9-36) treatment agrees with a body of evidence linking mitochondrial ROS to aging-related cognitive and synatpic pathology Wang et al., 2014;Guo et al., 2017). Previous studies have attempted to improve cognitive decline in DS patients with antioxidant treatment. Although preclinical studies have found success, they have yet to prove effective in clinical trials (Ani et al., 2000;Lott et al., 2011;Sano et al., 2016). These clinical trials have attempted to improve cognition through either nutritional supplementation or with traditional ROS scavengers such as ␣-tocopherol (vitamin E; Ani et al., 2000;Lott et al., 2011;Sano et al., 2016). A potential problem with this approach is that it only prevents ROS from oxidizing cellular components, but does nothing to prevent the excessive ROS genesis. Conversely, as we and others have shown, GLP-1 (9-36) protects against oxidative stress and decreased ROS production. It has also been shown to decrease oxidative stress in animal models of cognitive disease, known to have increased ROS production (Ban et al., 2010;Ma et al., 2012;Giacco et al., 2015). Therefore our findings here, in conjuction with previous studies (Ma et al., 2012), suggest that the natural peptide and GLP-1 cleavage product GLP-1 (9-36) has therapeutic potential to improve cognition in DS and other neuronal diseases associated with excessive mitochondrial ROS.