Effects of cardiac glycoside digoxin on dendritic spines and motor learning performance in mice

Synapse formation following the generation of postsynaptic dendritic spines is essential for motor learning and functional recovery after brain injury. The C -terminal fragment of agrin cleaved by neurotrypsin induces dendritic spine formation in the adult hippocampus. Since the α 3 subunit of sodium-potassium ATPase (Na/K ATPase) is a neuronal receptor for agrin in the central nervous system, cardiac glycosides might facilitate dendritic spine formation and subsequent improvements in learning. This study investigated the effects of the cardiac glycoside digoxin on dendritic spine turnover and learning performance in mice. Golgi-Cox staining revealed that intraperitoneal injection of digoxin less than its IC 50 in brain signi�cantly increased the density of long spines ( ≥ 2 µm) in cerebral cortex and hippocampus in wild-type mice and neurotrypsin-knockout (NT-KO) mice showing impairment of activity-dependent spine formation. Whereas motor learning performance of NT-KO mice showed signi�cantly lower than control wild-type mice under the control condition, low dose of digoxin enhanced performance to a similar degree in both strains. In NT-KO mice, lower doses of digoxin equivalent to clinical doses also signi�cantly improved performance. These data suggest that lower doses of digoxin could modify dendritic spine formation or recycling and facilitate motor learning in compensation for the neurotrypsin-agrin pathway.


Introduction
Neural plasticity is a fundamental property for learning and memory processes and for functional recovery after brain injury.A functional MRI study indicated the plasticity of motor cortex during motor skill learning 1 .One important plastic change is the functional reorganization of synaptic circuits constituted by de novo synapses.Synapse formation in the adult central nervous system (CNS) is requisite for motor learning 2 .Furthermore, such synaptic reorganization was also induced in motor skill enhancement after ischemic cortical damage 3 .The generation of dendritic spines, representing excitatory postsynaptic structures, follows synaptic formation 4 .Structural reorganization of dendritic spines occurred in the basal ganglia after coordinated motor learning using the rotarod test 5 , and motor skill learning in a reaching task promoted the formation and stabilization of dendritic spines in the motor cortex 6 .
Neurotrypsin-dependent cleavage of agrin has been identi ed as a signal pathway for the generation of dendritic spines in the adult hippocampus.Neurotrypsin, a serine protease localized at presynapses 7 , splices out a C-terminal fragment of agrin in response to coactivation of pre-and post-synapses 8 .The agrin C-terminal fragment is essential for the induction of dendritic lopodia, a precursor of dendritic spines 8 .The α 3 subunit of sodium-potassium ATPase (Na/K-ATPase) was identi ed as a neuronal receptor for the C-terminal fragment of agrin in the CNS 9 .Previous studies have suggested that α subunits of Na/K ATPase are involved in spatial learning and dendritic growth 10,11 .These studies raise the possibility that the regulation of Na/K ATPase activity could induce the de novo formation of synapses in the adult brain.
Cardiac glycosides are Na/K ATPase ligands that regulate the activity of this enzyme in an inverted Ushaped manner, inhibiting Na/K ATPase at concentrations near the IC 50 and increasing the pumping by Na/K ATPase at doses below the IC 50 [12][13][14] .One of the cardiac glycosides, digoxin, shows partial penetration of the blood-brain barrier and has been shown to upregulate hippocampal Na/K ATPase activity following intraperitoneal injection into rats at low doses, but inhibits this activity at high doses 14 .Whereas digoxin is a cardiotonic drug that have been used for cardiac failure by inhibiting Na/K ATPase pumping activity 15 , a previous study reported neuroprotective effects of low-dose cardiac glycosides after ischemia 14 .However, it remains unknown whether cardiac glycosides affect synapse turnover or learning behavior, especially at low doses.
Consequently, it is hypothesized that intraperitoneal injection of digoxin could modify the spine turnover and improve motor learning accompanied with spine formation by regulating brain Na/K ATPase activity.
To test this hypothesis, the present study investigated the effect of digoxin on the brain ATPase activity, the turnover of dendritic spines in the cerebral cortex and hippocampus, and motor learning in wild-type mice.As injection doses of digoxin, we adopted doses more and less than its IC 50 in brain, and lower doses equivalent to clinical dose in anticipation of drug repurposing.Since high-dose digoxin reduced spontaneous motor activity 16 which might affect motor performance in the rotarod test, the locomotor activity was also analyzed in this study.In addition, we assumed that neurotrypsin knockout mice (NT-KO) impairing activity-dependent lopodia formation 8 could show lower motor learning performance, and that it could be compensated by digoxin injection.Therefore, the present study also examined the digoxin effect on spine turnover, locomotor activity, and motor learning in NT-KO mice and compared their results with those in wild-type mice.

Effect of digoxin on ATPase activity
We investigated the effect of intraperitoneal administrations of digoxin at various doses on ATPase activity in the brain.ATPase activity was signi cantly higher with 65 µg/kg of digoxin than with vehicle control (Fig. 1; P < 0.05, Tukey test).On the other hand, the facilitation effect on ATPase activity was abrogated at 650 µg/kg of digoxin.The present dose-dependent change in ATPase activity was consistent with the effect of digoxin on rat hippocampal Na/K ATPase activity in the previous study 14 .

Increment in ≥2 µm dendritic spines by digoxin
In this study, we identi ed dendritic lopodia and spines with length ≥2 µm as an index of spine turnover in mouse cerebral cortex and hippocampus, because the average length of dendritic spines is 0.5-2 µm in the CNS spiny neurons 17 .Young spines such as lopodia and thin-type spines shrink in length as they mature into stubby-and mushroom-type spines less than 1 µm in length 4,17,18 .
In C57BL/6N mice, Golgi-Cox staining revealed the change in densities of ≥2 µm dendritic spines corresponding to the injected dose of digoxin.First, we focused on the rostral half of cerebral cortex including motor cortex where spine formation was accompanied with motor learning 6 .In the rostral half of cerebral cortex, ≥2 µm spines along the basal dendrites of pyramidal neurons in layer 3 signi cantly increased under the condition of 1, 4, or 65 µg/kg in comparison to vehicle control (P < 0.05 between vehicle and 1, 4, or 65 µg/kg of digoxin, Steel-Dwass test; Fig. 2a-f).Meanwhile, 650 µg/kg of digoxin did not affect the density of ≥2 µm dendritic spines.Almost same tendency was also obtained along the basal dendrites of pyramidal neurons in layer 3 of the caudal half of the cerebral cortex (Fig. 2g and Supplementary Fig. S1a-e).In the caudal half of cerebral cortex, 650 µg/kg of digoxin still signi cantly increased the density of ≥2 µm dendritic spines (P < 0.05 vs vehicle, Steel-Dwass test).Along the primary branches of apical dendrites of hippocampal CA1 pyramidal neurons, density of ≥2 µm dendritic spines was signi cantly elevated by 65 µg/kg of digoxin (P < 0.05 vs vehicle, Tukey test; Fig. 2h and Supplementary Fig. S1f-j).Signi cant differences were also detected between

High-dose digoxin reduced spontaneous locomotor activity
A previous report showed that the injections of cardiac glycosides into mice in uenced locomotor activity.For example, digoxin injected intragastrically at ≥1000 µg/kg signi cantly reduced spontaneous locomotor activity 16 .We therefore performed the open eld test to assess the effect of digoxin at the doses used in this study on locomotor activity.To verify the association between the effect of digoxin and activity-dependent formation of dendritic spines, we used NT-KO mice showing impairment of activity-dependent lopodia formation in the hippocampus 8 and its control neurotrypsin wild-type mice (NT-WT mice) for the behavioral testing.Injection of 650 µg/kg of digoxin to both strains signi cantly reduced the total distance travelled for 20 min in the open eld test (P < 0.05 vs vehicle, 1, 4, or 65 µg/kg, Tukey test), whereas 1, 4, or 65 µg/kg of digoxin exerted no signi cant effect in comparison to vehicle control (Fig. 4).The inhibitory effect of 650 µg/kg of digoxin was more profound than that in NT-KO mice (P < 0.05, unpaired t test).
Low-dose digoxin enhanced performance of coordinate motor skills in the rotarod test Both strains of mice showed the gradual increase in performance in the rotarod test by repeating the trials until Day 3, and performance was retained even after a week (Day 10; Fig. 5).However, the extent of improvement in performance differed between NT-WT and NT-KO mice under the vehicle control condition.The performance of NT-WT mice was signi cantly higher than that of NT-KO mice on Days 2, 3, and 10 (P < 0.05, unpaired t test at Day 2 and 10 or Mann-Whitney U test at Day 3).On the other hand, lower-dose digoxin (1 or 4 µg/kg) exerted a signi cant facilitatory effect on the performance in NT-KO mice (Fig. 5c).In addition to the enhancing effects of 65 µg/kg of digoxin on Days 2 and 3 performances (P < 0.05 vs vehicle, Games-Howell test on Day 2 or Tukey test on Days 3), digoxin injections at 1 or 4 µg/kg also signi cantly improved performances on Days 2, 3, and 10 (P < 0.05 vs vehicle, Games-Howell test on Day 2 or Tukey test on Days 3 and 10) to a similar extent as those at 65 µg/kg (P ≥ 0.05).Furthermore, performances of NT-KO mice at Day 3 by injection of 4 µg/kg digoxin was more improved than that in the same condition of NT-WT (P < 0.05, Welch's t test).As opposed to NT-WT mice, 650 µg/kg of digoxin did not affect the performance on all Days (Fig. 5b).At Days 1 and 2 under 650 µg/kg condition, NT-KO mice showed higher performance than NT-WT mice (P < 0.05, Mann-Whitney U test at Day 1 or unpaired t test at Day 2).There were no signi cant differences between NT-KO and NT-WT mice in other digoxin conditions (P ≥ 0.05).

Discussion
This study found that low doses (1, 4, or 65 µg/kg) of digoxin presumably less than the IC 50 in the brain increased the densities of newly formed dendritic spines and improved performance in the motor learning test without affecting locomotor activity.Cardiac glycosides such as digoxin or ouabain at doses below the IC 50 enhanced the NA/K ATPase pumping activity [12][13][14] .Considering the previous study using rats 14 , intraperitoneal injections of digoxin ≤65 µg/kg could also result in concentrations less than the IC 50 in the brain, leading to upregulation of Na/K ATPase activity.Although a change in Na/K ATPase activity was partially masked by other ATPases in Fig. 1, our result supported this speculation that Na/K ATPase is activated by digoxin up to 65 µg/kg.
In spine analysis, we mainly focused the rostral part of cerebral cortex, because this part includes motor cortex where spine formation was accompanied with motor learning 6 .An increase in density of young spines longer than average length (≥2 µm) could represent a change in spine turnover, namely, spine formation or inhibition of recycling of young spines, which results in enhancement of plasticity.In the rostral part of cerebral cortex, low doses (1, 4, or 65 µg/kg) of digoxin signi cantly increased the density of ≥2 µm spines in both NT-WT and NT-KO mice (Figs.2f and 3f).Signi cant increase in ≥2 µm spines were also detected in other areas under condition of 4 or 65 µg/kg (Figs. 2 and 3).Previous reports have revealed that activity-dependent generation of lopodia in the hippocampus requires the C-terminal fragment of agrin 8 , which binds to the α 3 subunit of Na/K ATPase and induces membrane depolarization and an increase in action potential frequency in cortical neurons 9 .As the α 3 and α 2 subunits both show higher a nity for digoxin than the α 1 subunit 19 , the increasing effect of low doses (1, 4, or 65 µg/kg) of digoxin on the densities of newly formed dendritic spines might be due to activation of signals downstream of neurotrypsin-agrin pathway.Although a high concentration of agrin was shown to inhibit Na/K ATPase activity 9 , whether agrin exerts an inverted U-shaped regulation on Na/K ATPase activity similar to cardiac glycosides remains unclear.Na/K ATPase is known to act as signal transducers for processes such as calcium signaling, EGF receptor cascade, and PI3 kinase cascade 20,21 , and the resulting signals are involved in the regulation of lopodia formation; dendritic lopodia extension was facilitated by low levels of local Ca 2+ transients, whereas high levels inhibited lopodia formation 22 .
Downregulation of EGF receptors increased the stability of axonal lopodia 23 , while the inhibition of PI3 kinases reduced lopodia in islet cells 24 .As for dendrite morphology, activation of Na/K ATPase by ouabain triggered CREB activation and dendritic arbor growth in cortical neurons 10 .
The highest dose of digoxin in the present study (650 µg/kg) signi cantly reduced spontaneous locomotor activity in both strains of mice (Figs. 4 and 5), which was inconsistent with the effect on the densities of ≥2 µm dendritic spines in brain (Figs. 2 and 3).The negative effect of high-dose digoxin on locomotor activity might be partially due to its cardiotonic action in addition to its direct effect in brain.
Intragastrically-injected digoxin at high doses downregulated locomotor activity by systemic effect 16 .Conversely, intracerebroventricular injection of high-dose ouabain into brain inducing Na/K ATPase inhibition caused hyperactivity in mice 25 .On the other hand, the effect of 650 µg/kg digoxin on locomotor activity seems to coincide with that on rotarod test; the negative effects on locomotor activity and rotarod performance on Days 1-3 in NT-WT mice was more severe than that in NT-KO mice, which implied that a decrease in locomotor activity could cause a reduction in rotarod performance.
65 µg/kg of digoxin improved motor learning in the rotarod test.Improved performances were retained for at least 7 days after the last injection (Day 10), indicating that these improvements could not be due to the transient action of digoxin on cardiopulmonary endurance.Na/K ATPase is involved in spatial learning as mice heterozygous for α 2 or α 3 displayed spatial memory de cits in the Morris water maze 11 .
The present results suggest that digoxin-induced activation of Na/K ATPase could adversely impact motor learning.Under the vehicle control condition, NT-KO mice exhibited a signi cant lower performance than NT-WT mice, which indicates that the activity-induced lopodia formation dependent on the neurotrypsin-agrin pathway could be important for motor learning.Motor skill learning in mice promoted immediate dendritic spine formation in the motor cortex where neurotrypsin mRNA was expressed 26,27 .65 µg/kg of digoxin improved motor learning performance by NT-KO mice up to the level in NT-WT mice injected with 65 µg/kg of digoxin.The compensation for motor learning impairment in NT-KO mice with digoxin administration suggests that dendritic spine formation could be essential for motor learning.On the other hand, lower doses (1 or 4 µg/kg) of digoxin improved motor learning only in NT-KO mice showing impairment of activity-dependent lopodia formation.In NT-WT mice, lower doses of digoxin signi cantly increased ≥2 µm dendritic spines in cerebral cortex, but intrinsic activity-dependent lopodia formation might mask their effects on motor learning.This raises the possibility that digoxin might be more effective in the reconstitution of synaptic circuit in injured brain.Because the lower doses are equivalent to clinical doses, the present study raises the possibility that digoxin is potentially repurposed for disorders involving brain in addition to cardiac failure.
This research, however, is subject to several limitations.First, only male mice were used.Second, 2D analysis using Golgi-Cox staining loses some important 3D information.Their limitations could be addressed in future research.

Animals
Male mice (8-11 weeks old) from the following lines were used: C57BL/6N mice; NT-KO mice lacking the region encoding the proteolytic domain 28 ; and its control NT-WT mice.All animals were housed in groups on standard a 12 h light: dark cycle with ad libitum access to food and water.The C57BL/6N strain was used for ATPase assay and Golgi-Cox staining.NT-KO and NT-WT mice were used for behavioral testing.
All behavioral experiments were started at Zeitgeber time 4.All behavioral tests and intraperitoneal administration of digoxin were performed without anesthesia.Cervical dislocation was performed by an expert as a method of euthanasia before ATPase activity assay and spine analysis, and after behavioral tests.All experimental protocols were approved by animal care committee at Osaka Metropolitan University.All methods were carried out in accordance with relevant guidelines and regulations, and are reported below in accordance with ARRIVE guidelines.

Administration of digoxin
Digoxin (nacalai tesque, Kyoto, Japan) was dissolved in pyridine at a concentration of 0.02 mg/kg, 0.08 mg/kg, 1.3 mg/kg or 13 mg/kg, then diluted with saline up to a nal concentration of 1 µg/kg, 4 µg/kg, 65 µg/kg, and 650 µg/kg of digoxin per 100 µl.Digoxin at the above doses was administered intraperitoneally without anesthesia.Na/K ATPase activity in rat hippocampus was reportedly increased by intraperitoneal administrations of digoxin at low concentration (65 µg/kg), but decreased at high concentration (650 µg/kg) 14 .Intraperitoneal administration of 65 or 650 µg/kg of digoxin corresponded to a dose below or above its IC 50 for brain Na/K ATPase inhibition (25 nM for the α 2 and α 3 subunits; 130 µg/kg is equivalent to IC 50 ), respectively, which was calculated from a previous study 29 .Concentrations of 1 µg/kg and 4 µg/kg were additionally adopted as doses approximately corresponding to clinical doses for humans.The effects of 1 and 4 µg/kg digoxin were analyzed in anticipation of drug repurposing as treatment for brain disorders in addition to cardiac failure.

ATPase activity assay
To assess the Na/K ATPase activity in brain, we calculated the amount of free phosphate ion released from ATP by ATPase using an ATPase Activity Assay Kit (BioVision, Waltham, MA, USA), because approximately 50% of ATP is consumed by Na/K ATPase in the brain 30,31 .Brie y, Following cervical dislocation and decapitation 1 h after intraperitoneal administration of vehicle only or digoxin at 1, 4, 65, or 650 µg/kg, we isolated and homogenized 40 mg of cerebral cortex from the parietal lobe of 10-weekold male C57BL/6N mice.After removing endogenous phosphate using the ammonium sulfate method, the amount of free phosphate ion released by ATPase was measured using Malachite Green Reagent according to the procedure recommended by the manufacturer.The absorbance of samples was measured at 650 nm using a microplate reader (MPR-A100; AS ONE Corp., Osaka, Japan).

Dendritic spine analysis
To visualize the dendritic spine, we performed Golgi-Cox staining using the FD Rapid Golgistain™ Kit (FD NeuroTechnologies, Columbia, MD, USA).Although the two-dimensional analysis using Golgi-Cox staining has limitations in acquiring 3D information, this method is still useful for comparing the spine morphology under various conditions.We used 10-week-old male C57BL/6N mice and 8-week-old male NT-KO mice for dendritic spine analysis.Following cervical dislocation and decapitation 1 h after intraperitoneal administration of vehicle only or digoxin at 1, 4, 65, or 650 µg/kg, we rapidly removed the brain.Removed brains were divided into equal thirds along the rostrocaudal axis, then the rostral and middle parts of the brain including cerebral cortex and hippocampus were stained by the Golgi-Cox method according to the instructions from the manufacturer and a previous study 32 .The border between rostral and middle parts was approximately located at the bregma level.Brie y, fresh brain blocks were immersed in a mixture of equal volumes of kit Solution A and B, then stored for 3 weeks under dark conditions.This solution was replaced with fresh solution after the rst 24 h of immersion.Samples were subsequently immersed in Solution C for 1 week.Solution C was also replaced with fresh solution after the rst 24 h of immersion.After immersion in Solution C, brain blocks were rapidly frozen in powdered dry ice, then tissue sections of 200 µm thickness along the rostrocaudal axis were sliced at -22 to -24°C using a cryostat microtome (HM525NX; PHC Corp., Tokyo, Japan).Sections were mounted with Solution C on glass slides precoated with 0.5% gelatin, then dried naturally at room temperature.Subsequently, the sections were stained using a solution containing 1 part Solution D, 1 part Solution E, and 2 parts distilled water (DW) for 5 min.After staining, slides were rinsed in DW twice for 4 min each, then dehydrated in a serial dilution of ethanol and cleared in xylene.Finally, slides were coverslipped and sealed using Entellan (Merck, Darmstadt, Germany).We performed two times of independent staining using 2 mice under each condition.
Bright eld images of stained sections were acquired using a microscope (BX50; Olympus Corp., Tokyo, Japan) with the NY-X9 digital photographic device system (Microscope Network Corp, Saitama, Japan).The morphological characterization of newly formed dendritic spines ( lopodia and thin-type spines) is that the spine is longer than a matured spine (stubby-and mushroom-type spines; <1 µm in length) 4,17,18 .
Because the average length of dendritic spines is 0.5-2 µm in the CNS spiny neurons 17 , we counted protrusions with a length ≥2 µm from dendritic shafts as an index of spine turnover, that is, spine formation or inhibition of recycling of young spines, along the basal dendrites of pyramidal neurons in layer 3 of the cerebral cortex, and along primary branches of the apical dendrites of CA1 pyramidal neurons.All researchers who counted dendritic spines were blinded to the experimental condition of each image.

Open eld test
To evaluate the locomotor activity of mice, the open eld test was performed using the video tracking software (Smart 3.0; Panlab Harvard Apparatus, Barcelona, Spain).Male NT-KO and NT-WT mice at 11 weeks old were intraperitoneally injected with vehicle only or digoxin (1, 4, 65, or 650 µg/kg) 1 h before the test.We measured the total travel distance by freely moving mice in a gray acryl box (30 × 30 × 30 cm) for 20 min.

Rotarod test
We performed the rotarod test to assess motor learning in mice.The rotarod test is a widely used primary assay for the study of motor learning 33,34 , because coordinate sensorimotor responses must be learned in order to remain on the accelerating rotarod without falling.We adopted the accelerating rotation condition (from 4 to 40 rpm in 5 min) because a steep learning curve till 15 trials was obtained 35 and morphological change of spines was detected in striatum under this condition 5 .We used a rotarod machine with automatic timers, falling sensors and a Effect Effect of digoxin on rotarod test performance Fig. 5b).No signi cant difference was seen between performances under the vehicle control condition and that with 1 or 4 µg/kg of digoxin.Contrary to low doses of digoxin, 650 µg/kg of digoxin signi cantly reduced the performance on Days 1, 2, and 3 in comparison to vehicle (P < 0.05, Steel-Dwass test on Day 1, Games-Howell test on Day 2, or Tukey test on Day 3; Fig. 5b).
of intraperitoneally injected digoxin on ATPase activity in the cerebral cortex Intraperitoneally injected digoxin shows an inverted U-shaped dose-response curve for ATPase activities in the cerebral cortex.Digoxin at 65 µg/kg signi cantly increased ATPase activity in comparison to vehicle.Sample numbers are 9 from 3 mice in each condition.Error bars indicate SEM.*: P< 0.05 vs vehicle by Tukey test.

Figure 2 Effect
Figure 2

Figure 3 Effect
Figure 3

Figure 4 Effect
Figure 4 The time to fall from the accelerating rotarod was recorded.Seven days after Day 3 (Day 10), retention of motor memory was assessed by two additional trials on the accelerating rotarod at the same rate without injecting digoxin to exclude a transient in uence of changes in cardiopulmonary endurance by digoxin on test performance.The trial interval on each day was set to approximately 5 min.The timeline of rotarod test was shown in Fig.5a.software (version R-2.8.1 for Windows) was used for statistical data analysis.To evaluate differences between wild-type and NT-KO, unpaired t testing, Welch's t test, or Mann-Whitney U test was performed according to the results of the Shapiro-Wilk normality test and Levene homoscedasticity test.Differences among four or ve conditions of digoxin concentrations (vehicle only, 1, 4, 65, or 650 µg/kg) were examined by one-way ANOVA, Welch's ANOVA, or the Kruskal-Wallis test and their corresponding post-hoc test (Tukey test, Games-Howell test, or Steel-Dwass test), respectively, according to the results of the Shapiro-Wilk normality test and Levene homoscedasticity test.Statistical signi cance was set at the level of P < 0.05. R See Supplemental Figure S1 online for typical images of dendrites in caudal part of cerebral cortex and hippocampus.f) Bar graph showing the average density of dendritic spines with length ≥2 µm in the rostral part of the cerebral cortex.The Y axis indicates the number of ≥2 µm spines per 10-µm dendrite.Intraperitoneally injected digoxin at 1, 4, or 65 µg/kg signi cantly increased ³2 µmspine density in comparison to vehicle control.Sample numbers: 37 dendrites for vehicle; 44 dendrites for 1 µg/kg of digoxin; 41 dendrites for 4 µg/kg of digoxin; 33 dendrites for 65 µg/kg of digoxin; 33 dendrites for 650 µg/kg of digoxin.Error bars indicate SEM.‡: P < 0.05 vs vehicle, Steel-Dwass test.g) Bar graph showing the average density of spines with length ≥2 µm in the caudal part of the cerebral cortex.The Y axis indicates the number of dendritic spines ≥2 µm per 10-µm dendrite.Intraperitoneally injected digoxin at 1, 4, 65, or 650 µg/kg signi cantly increased the density in comparison to vehicle control.Sample numbers: 50 dendrites for vehicle; 47 dendrites for 1 µg/kg of digoxin; 48 dendrites for 4 µg/kg of digoxin; 17 dendrites for 65 µg/kg of digoxin; 14 dendrites for 650 µg/kg of digoxin.Error bars indicate SEM.‡: P < 0.05 vs vehicle, Steel-Dwass test.h) Bar graph showing the average density of spines with length ≥2 µm in hippocampal CA1 area.The Y axis indicates the number of dendritic spines ≥2 µm per 10-µm dendrite.Intraperitoneally injected digoxin at 65 µg/kg signi cantly increased the density in comparison to vehicle control.Sample numbers: 42 dendrites for vehicle; 37 dendrites for 1 µg/kg of digoxin; 52 dendrites for 4 µg/kg of digoxin; 27 dendrites for 65 µg/kg of digoxin; 29 dendrites for 650 µg/kg of digoxin.Error bars indicate SEM.*: P < 0.05 vs vehicle, Tukey test.