Increased O-GlcNAcylation rapidly decreases GABAAR currents in hippocampus but depresses neuronal output

O-GlcNAcylation, a post-translational modification involving O-linkage of β-N-acetylglucosamine to Ser/Thr residues on target proteins, is increasingly recognized as a critical regulator of synaptic function. Enzymes that catalyze O-GlcNAcylation are found at both presynaptic and postsynaptic sites, and O-GlcNAcylated proteins localize to synaptosomes. An acute increase in O-GlcNAcylation can affect neuronal communication by inducing long-term depression (LTD) of excitatory transmission at hippocampal CA3-CA1 synapses, as well as suppressing hyperexcitable circuits in vitro and in vivo. Despite these findings, to date, no studies have directly examined how O-GlcNAcylation modulates the efficacy of inhibitory neurotransmission. Here we show an acute increase in O-GlcNAc dampens GABAergic currents onto principal cells in rodent hippocampus likely through a postsynaptic mechanism, and has a variable effect on the excitation/inhibition balance. The overall effect of increased O-GlcNAc is reduced synaptically-driven spike probability via synaptic depression and decreased intrinsic excitability. Our results position O-GlcNAcylation as a novel regulator of the overall excitation/inhibition balance and neuronal output.


Acute increase in protein O-GlcNAcylation depresses GABAergic transmission onto CA1 pyramidal cells and dentate granule cells. To determine if protein O-GlcNAcylation modulates
GABA A R-mediated inhibitory neurotransmission, we used whole-cell voltage clamp to record spontaneous inhibitory post-synaptic currents (sIPSCs) from CA1 pyramidal cells while blocking glutamatergic transmission using DNQX (10 μM) and DL-AP5 (50 μM). Following a 5 min baseline, we bath applied the HBP substrate glucosamine (GlcN, 5 mM) and the OGA inhibitor thiamet-G (TMG, 1μM) to acutely increase protein O-GlcNAc levels, as done previously 18,19 (Fig. 1Ai). We found a significant reduction in sIPSC amplitude (Fig. 1Aii, cumulative probability distribution, p < 0.0001, KS D value = 0.217, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test) and inter-event interval (Fig. 1Aiii, cumulative probability distribution, p < 0.0001, KS D value = 0.084, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test) in CA1 pyramidal cells. To ensure that the change in amplitude and inter-event interval of the sIPSCs was not a consequence of a technical artifact, sIPSCs during baseline and 5 min after GlcN + TMG application were averaged and scaled (Fig. 1Aii,Bii, inset); the traces perfectly overlapped, indicating that the decrease in sIPSC amplitude and frequency observed are not due to an increase in series resistance caused by prolonged recording or by washing on GlcN + TMG. It is also important to note that we observed a shift in holding current following application of GlcN + TMG (baseline: −138.2 ± 13.6 pA vs. GlcN + TMG: −106.4 ± 11.5 pA, n = 9 cells, 5 rats, p = 0.006, paired t-test), suggesting possible modulation of extrasynaptic GABA A Rs, which will be investigated in future experiments.
In order to determine if this decrease in sIPSCs induced by GlcN + TMG is specific to CA1 pyramidal cells or if it occurs at inhibitory synapses on other cells, we recorded sIPSCs from dentate gyrus granule cells (GCs) before and after bath application of GlcN + TMG (Fig. 1Bi). We observed the same effect, namely a decrease in sIPSC amplitude (Fig. 1Bii, cumulative probability distribution, p < 0.0001, KS D value = 0.11, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test) and increase in inter-event interval (Fig. 1Biii, cumulative probability distribution, p < 0.0001, KS D value = 0.055, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test), but no shift in holding current (p = 0.768, paired t-test). These findings suggest that O-GlcNAc-mediated depression of GABAergic transmission is likely to be a general mechanism occurring at many inhibitory synapses in the brain.
The decrease in sIPSCs following increased O-GlcNAcylation could arise from a decrease in presynaptic GABA release, or from decreased postsynaptic GABA A R function. To distinguish between these possibilities, we recorded mIPSCs from both CA1 pyramidal cells and GCs in the presence of the voltage-gated sodium channel blocker TTX (1 μM) (Fig. 2Ai,Bi). A change in mIPSC frequency typically reflects a change in presynaptic neurotransmitter release probability, while a change in mIPSC amplitude indicates a change in postsynaptic receptor function. Bath application of GlcN + TMG resulted in a reduction in mIPSC amplitude (Fig. 2Aii, cumulative probability distribution, p < 0.0001, KS D value = 0.24, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test), but not inter-event interval (Fig. 2Aiii, cumulative probability distribution, p = 0.96, KS D value = 0.032, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test) in CA1 pyramidal cells. We observed a similar decrease in mIPSC amplitude in dentate GCs (Fig. 2Bii, cumulative probability distribution, p < 0.0001, KS D value = 0.12, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test), but in contrast to CA1, we also observe a slight but significant increase in mISPC inter-event interval (Fig. 2Biii, cumulative probability distribution, p < 0.0001, KS D value = 0.099, Kolmogorov-Smirnov test; inset: p < 0.0001, Wilcoxon matched-pairs signed rank test). However, the very small difference in mean values between control (373.5 ± 7.3 ms) and GlcN + TMG (374.1 ± 7.4 ms) suggests this significant difference may not be biologically relevant. Collectively, the decrease in mIPSC amplitude in the absence of a decrease in inter-event interval, at least in CA1 pyramical cells, is consistent with the interpretation that the dampening of inhibitory transmission following increased O-GlcNAcylation is due to a reduction in postsynaptic GABA A R function.
www.nature.com/scientificreports www.nature.com/scientificreports/ O-GlcNAcylation induces long-term depression of inhibitory transmission. We next tested if O-GlcNAcylation similarly affects the amplitude of electrically evoked IPSCs (eIPSCs), as evoked transmission may be dependent upon presynaptic vesicle pools (reviewed in 26 ) distinct from those involved in spontaneous release. To address this and remaining questions, we focused our experiments on CA1 pyramidal cells. Following a 5 min stable baseline of eIPSCs (Cs Gluconate pipet solution; E Cl − = −50 mV), bath applied GlcN + TMG elicited a significant reduction in eIPSC amplitude that lasted the duration of the recording (Fig. 3Ai,Aiii, p = 0.001, paired t-test), with no change in decay kinetics (Fig. 3Aii, p = 0.23, paired t-test).
We previously reported that a 10 min exposure to GlcN (5 mM) alone induces a significant increase in O-GlcNAcylation in hippocampus that is readily reversed within a 10 min washout, while O-GlcNAcylation remains elevated for up to 2 hrs following a 10 min exposure to TMG 19 . Importantly, the depression of transmission at CA3-CA1 synapses induced by GlcN outlasted the transient increase in protein O-GlcNAcylation, demonstrating expression of a long-term depression (LTD) at these synapses 19 . To test if O-GlcNAcylation similarly causes long-term depression at GABAergic synapses onto CA1 pyramidal cells, we applied GlcN alone for 10 min and recorded eIPSCs for 10 min during washout and observed a persistent reduction in amplitude (Fig. 3Bi,Biii, p = 0.003, paired t-test), consistent with expression of a form of LTD at GABAergic synapses.

O-GlcNAcylation has a variable effect on the E/I ratio.
Since O-GlcNAcylation induces LTD of both excitatory 19 and inhibitory transmission onto CA1 pyramidal cells (Fig. 3), we wondered what the overall effect of increased O-GlcNAc is on the balance of excitation to inhibition (E/I). Thus, we measured the E/I ratio at CA3-CA1 synapses by recording compound synaptic currents consisting of mono-synaptic EPSCs followed by di-synaptic IPSCs in whole-cell recordings of CA1 pyramidal cells voltage-clamped at −30 mV. We observed a linear trend in the balance of E to I, with the E/I ratio increasing following GlcN + TMG application (Fig. 4Ai, p < 0.0001, repeated measures one-way ANOVA with post hoc test for linear trend). However, at the single cell level, the change in the E/I ratio was variable with just over half of cells (n = 6/10) displaying a statistically www.nature.com/scientificreports www.nature.com/scientificreports/ increased ratio following GlcN + TMG exposure (p ≤ 0.009, paired t-test on each individual cell comparing 5 mins before and 5 min following 5 min exposure to GlcN + TMG), and a statistically decreased E/I ratio in the rest (n = 4/10; p ≤ 0.01, paired t-test on each individual cell comparing 5 mins before and 5 min following 5 min exposure to GlcN + TMG). Thus, comparing the mean E/I ratios before and after GlcN + TMG exposure when all cells were averaged yields no significant difference (Fig. 4Aii, p = 0.08, paired t-test; red circles and blue circles indicate increased and decreased E/I respectively).
We next examined the compound EPSCs and IPSCs individually as both can be depressed by inceased O-GlcNAc. GlcN + TMG application caused a significant reduction in the IPSC component of the compound current ( Fig. 4Bi-Biii, open circles shown as %baseline current, paired t-test, p = 0.03), with no significant change in the EPSC component ( Fig. 4Bi-Biii, closed circles shown as % baseline current, p = 0.43, Wilcoxon matched-pairs signed rank test). These data indicate that on average, increasing O-GlcNAc elicits a greater depression of the disynaptic GABA A R-mediated IPSC than the mono-synaptic glutamatergic EPSC. Normally, GABA A R-mediated transmission limits the amplitude of EPCSs through shunting the excitatory potential, and when the inhibitory shunt is decreased or blocked, the EPSC amplitude increases. Thus, decreasing the inhibitory shunt via the O-GlcNAc-mediated decrease in IPSC amplitude predicts that the EPSC amplitude should increase 27 , which we observed in a fraction of the cells (n = 4/10, Fig. 4Aiii). On the other hand, the GluA2-AMPAR mediated O-GlcNAc LTD at CA3-CA1 synapses 19 , occurring in parallel with the decrease in IPSC amplitude could occlude the expected increase in EPSC amplitude cause by the decreased inhibitory shunt, which we observed in the majority of our cells (n = 6/10, Fig. 4Aiii). Thus, the overall effect of increasing O-GlcNAc is extremely complicated given that exposure to GlcN + TMG causes synaptic depression at both glutamatergic and GABAergic synapses within minutes, and the impact on the E/I ratio will be driven by the magnitude of the change in the EPSC vs IPSC, with increases in the E/I ratio being driven by a larger increase in the EPSC or smaller decrease in the EPSC amplitude compared to the IPSC (compare blue and red circles in Fig. 4Aiii). We also confirmed this variable effect on the E/I ratio in current clamp recordings from CA1 pyramidal cells, and found, once again, a linear trend for an increase in the E/I ratio (

O-GlcNAcylation reduces action potential probability in CA1 pyramidal cells. How does this
variable effect of increased O-GlcNAc on the E/I ratio affect neuronal output? In a previous study, we reported that increasing O-GlcNAc decreased basal spontaneous activity of CA3 pyramidal cells in an intact circuit, and depressed epileptic activity in area CA1 in hyperexcitable conditions generated by blocking GABA A Rs 18 . Therefore, to determine the net effect of increasing protein O-GlcNAcylation on CA1 pyramidal cell output when the circuit is intact, we carried out current-clamp recordings from CA1 pyramidal cells and measured synaptically driven AP probability before and after application of GlcN + TMG. Stimulus intensity was set to achieve a ~25-50% success rate of generating a synaptically driven AP during the 5 min baseline (Fig. 5Ai, 0-5 min) after which GlcN + TMG were bath applied to increase O-GlcNAc levels (Fig. 5Ai, 5-20 min). Increasing O-GlcNAcylation caused a significant reduction in AP probability in all recorded cells, despite the increase in E/I ratio (Fig. 5Aii,Aiii, p < 0.0001, paired t-test). Importantly, control experiments were interleaved to rule out technical artifacts since AP probability is particularly sensitive to any change in the quality of the recording (Fig.  5Bi-Biii, p = 0.56, paired t-test).
As a further test of the effects of increasing O-GlcNAcylation on the net activity in the intact circuit, we performed a complementary experiment measuring the output of the CA1 population by recording extracellular population spikes (pSpike) in the CA1 pyramidal cell layer during a 10 min application of GlcN + TMG (Fig. 5Ci, 10-15 min). Consistent with the single-cell recordings, we observed a significant and sustained reduction in pSpike amplitude following an increase in O-GlcNAcylation (Fig. 5Cii, p = 0.004, paired t-test), which supports our previous report of decreased CA3 output in the context of increased protein O-GlcNAcylation 18 .   www.nature.com/scientificreports www.nature.com/scientificreports/ Therefore, we next tested the possibility that O-GlcNAcylation decreases intrinsic excitability by modulating the active and passive properties of CA1 pyramidal cells. We used whole-cell current clamp recordings of CA1 pyramidal cells during hyperpolarizing and depolarizing current injections in the presence of pharmacological blockers of GABA A Rs, AMPARs, and NMDARs, to isolate the cells from synaptic inputs (Fig. 6A). We found that O-GlcNAcylation caused a significant reduction in the input resistance (Fig. 6C, p = 0.031, paired t-test) when cells were isolated from the synaptic network, and a significant increase in rheobase, or the minimum current required to elicit a single action potential (Fig. 6D, p = 0.036, paired t-test). We also observed an increase in AP threshold (Fig. 6E, p = 0.030, paired t-test), but no difference in AP shape (Fig. 6F-H, AP amplitude: p = 0.441; AP rise: p = 0.143; AP half width: p = 0.653, paired t-test). Additionally, we found a small but significant decrease in the number of action potentials generated with higher current injection steps following GlcN + TMG application (F O-GlcNAc (1,8) = 5.81, p < 0.05, two-way repeated measures ANOVA). These changes are consistent with a decrease in intrinsic excitability.

Reduced intrinsic excitability of CA1 pyramidal cells contributes to the O-GlcNAc
Collectively, our current and previously published data 18 suggest that both synaptic and intrinsic mechanisms contribute to a net decrease in CA1 pyramidal cell excitability, despite a larger O-GlcNAc-mediated depression of inhibitory versus excitatory transmission. Therefore, to distinguish the decreased intrinsic excitability from the synaptic depression at excitatory synapses, we repeated the pSpike recordings in GluA2 KO mice in which O-GlcNAc LTD at CA3-CA1 synapses will be absent (Fig. 7Ai,Aii). Interestingly, increasing O-GlcNAcylation depressed pSpike amplitude in both the wild-type and GluA2 KO mice (Fig. 7Aiii, control: p < 0.01 KO: p < 0.01, Sidak's multiple comparisons test), while the persistent depression (O-GlcNAc LTD) following GlcN + TMG wash-out was only observed in wild-type mice, as expected (Fig. 7Aiii, control: p < 0.01 KO: p > 0.05, Sidak's multiple comparison test). Thus, the early depression of the pSpike amplitude during GlcN + TMG application www.nature.com/scientificreports www.nature.com/scientificreports/ is likely a consequence of decreasing CA1 pyramidal cell intrinsic excitability, while the persistent depression of excitability is a consequence of synaptic depression mediated by O-GlcNAc LTD.

Discussion
We provide, for the first time, empirical evidence of a role for protein O-GlcNAcylation in regulating synaptic inhibition in principal cells of the hippocampus. First, we show that acutely increasing O-GlcNAcylation reduces the frequency and magnitude of sIPSCs in CA1 pyramidal cells and dentate granule cells, and selectively decreases the amplitude of mIPSCs, suggesting a postsynaptic mechanism. The amplitude of evoked IPSCs onto CA1 pyramidal cells is similarly reduced, and the depression is long-lasting, suggesting a form of LTD induced by O-GlcNAc at GABAergic synapses. Second, in a hippocampal circuit with intact excitation and inhibition, an acute increase in O-GlcNAcylation has a variable effect on the E/I ratio, despite a greater effect on inhibition vs excitation. Finally, we show that O-GlcNAcylation modulates the final output of CA1 pyramidal cells by reducing the action potential probability through intrinsic and synaptic mechanisms. These results position protein O-GlcNAcylation as a potent regulator of both synaptic inhibition as well as excitation, as we reported previously 18,19 .
While the effect of the analogous post translational modification, phosphorylation, on synaptic transmission has been extensively characterized, little is known about the role of O-GlcNAcylation on excitatory and inhibitory transmission. We have previously shown that acute increases in O-GlcNAcylation causes a GluA2 AMPAR subunit dependent long-term depression of excitatory transmission, O-GlcNAc LTD, in CA1 pyramidal cell dendritic field potentials and CA1 population spikes 18,19 . Here, we report that an acute increase in protein www.nature.com/scientificreports www.nature.com/scientificreports/ O-GlcNAcylation similarly causes a depression of inhibition onto CA1 pyramidal cells and dentate granule cells. This includes a reduction in the amplitude but not frequency of mIPSCs onto hippocampal principal cells, suggesting a post-synaptic site of action. This reduction in inhibition could involve a number of mechanisms including internalization of GABA A Rs, a change in the conductance of individual GABA A Rs, or destabilization of scaffolding proteins that anchor GABA A Rs to the membrane. Indeed, recent work 22 suggests the reduction in mEPSC amplitude following increased O-GlcNAcylation involves the endocytosis of GluA2 subunit containing AMPARs, which likely explains expression of the GluA2-dependent LTD we previously reported 19 . Additionally, phosphorylation of specific serine/threonine residues on GABA A Rs modulates inhibitory transmission, an effect that includes a reduction in GABA A R currents 9 , which might also occur following increased O-GlcNAc.
Notably, phosphorylation alters inhibition in the brain in a kinase-and region-specific manner (for review, see 9 ). Activation of PKA depresses synaptic GABA A R function in CA1 pyramidal cells but not dentate gyrus granule cells (GCs), whereas PKC activation alters mini-IPSC amplitude in GCs alone 28 . O-GlcNAcylation, however, is uniquely positioned to turn down global inhibition in a non region-specific manner as only one enzyme, OGT, is known to catalyze its addition to proteins. While phosphorylation and O-GlcNAcylation on some proteins can occlude one another [29][30][31][32][33] , other times they can act independently 19 or synergistically (for review, see 34 ). Whether there is a negative or positive interplay between the effects of phosphorylation and O-GlcNAcylation on inhibition remains to be determined.
We show that acute increases in O-GlcNAcylation rapidly depress the di-synaptic IPSC while having a variable effect on the amplitude of the EPSC. This variability is likely explained by competing mechanisms including an O-GlcNAc mediated decrease in the inhibitory shunt and O-GlcNAc LTD at excitatory synapses. These changes are complex as indicated by the variable effect on the E/I ratio with some cells experiencing a significant increase and others a decrease such that a significant change in the averaged dataset is not observed. Despite this, there is a consistent reduction in the synaptically-evoked action potential probability in pyramidal cells. While subthreshold EPSPs can spatially and temporally summate to generate APs, the efficacy of EPSP to spike coupling is thought to depend on several factors including IPSP magnitude and timing 35 , as well as passive and active properties of the post-synaptic neuron 36 . Interestingly, (i) individual pyramidal cells vary widely in their threshold afferent stimulation required for spiking, but this threshold is controlled by the magnitude of the EPSC and not the amplitude or timing of inhibition 37 ; and (ii) increasing O-GlcNAcylation leads to a depression of field EPSPs in area CA1, likely owing to an internalization of GluA2 containing AMPARs 18,22 . Thus, it stands to reason that the decrease in AP probability could stem from a concomitant depression of the CA3-CA1 EPSC. We found that O-GlcNAcylation not only reduces the AMPAR mediated dendritic depolarization 19 , but also changes the intrinsic properties of CA1 neurons, making it harder for the cell to spike. This change in intrinsic properties following an increase in O-GlcNAc we report here in rats is consistent with recently published work by Hwang & Rhim 22 showing a reduction in depolarizing voltage-gated sodium channel currents and an increase in hyperpolarizing voltage-gated potassium currents in recordings from mice. The slightly larger effect observed by Hwang & Rhim compared to what we report here is likely due to a species difference, as we observed a larger decrease in CA1 pyrmaidal cell excitability following GlcN + TMG application in our previous report in mice compared to rats 18 . We have also previously reported a reduction in basal spontaneous spiking activity in area CA3 of hippocampus following increased O-GlcNAc 18 , further supporting a suppressive effect of O-GlcNAcylation on neuronal output. Notably, this finding is contrary to studies using genetic strategies to increase O-GlcNAcylation 20 , which reported no changes in intrinsic excitability, or mini-EPSCs and -IPSCs following heterozygous loss of function mutation of OGA, whose protein product catalyzes the removal of O-GlcNAc from proteins. The conflicting results Pop spike amplitude before (black), after (blue) GlcN + TMG and following wash-out (grey) in control mice (baseline vs. GlcN + TMG: p < 0.05; baseline vs. wash-out: p < 0.05, repeated measures one-way ANOVA with Sidak's multiple comparison's test, n = 5 slices, 2 rats) and pSpike amplitude before (red), after (blue) GlcN + TMG, and following wash-out (pink) in GluA2 KO mice (baseline vs. GlcN + TMG: p < 0.01; baseline vs. wash-out: p > 0.05, repeated measures one-way ANOVA with Sidak's multiple comparison's test, n = 5 slices, 2 rats). *p < 0.05, **p < 0.01.
www.nature.com/scientificreports www.nature.com/scientificreports/ are mostly likely due to differential effects of acute versus chronic upregulation in protein O-GlcNAcylation, as chronic changes can produce compensatory mechanisms to maintain synaptic function.
Studying the effect of acute, moment-to-moment changes in global O-GlcNAc levels on synaptic transmission likely reflects the impact of metabolically driven changes in O-GlcNAcylation occurring in vivo under both physiological and pathophysiological conditions. The OGT substrate UDP-GlcNAc is produced by the flux of glucose through the hexosamine biosynthetic pathway, which incorporates nucleotides, fatty acids, and amino acids. As such, O-GlcNAcylation is positioned to operate as a nutrient sensor (for review, see 38 ). For instance, fasting promotes energy conservation via the upregulation of OGT and O-GlcNAc and its effect on the activity of hypothalamic neurons 17 . Conversely, chronic elevation of glucose can also increase O-GlcNAcylation of certain targets, contributing to cardiac and neuronal pathophysiology 39 . Additionally, O-GlcNAcylation plays a crucial role in pathological states such as diabetes 39,40 where its elevation can contribute to disease pathogenesis, in epilepsy where increasing O-GlcNAc is protective 18,41 , and in neurodegeneration [42][43][44][45][46][47][48][49] , where increasing O-GlcNAc can decrease pathological accumulation of phosphorylated tau or α-synuclein. Collectively, studies published by us and others suggest that maintaining proper balance in O-GlcNAcylation is critical to maintaining normal hippocampal function, as too much or too little O-GlcNAc has been linked to deficits in learning and memory 19,20,42 . Revealing the complex changes in neuronal and synaptic function induced by alterations in O-GlcNAcylation in non-pathological and pathological states will require further investigation.

Materials and methods
All experimental procedures were approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee and follow the National Institutes of Health experimental guidelines.
Modulation of O-GlcNAc levels. As previously described 18,19 O-GlcNAcylation was acutely increased via bath application of glucosamine (GlcN, 5 mM; Sigma) alone or in combination with the selective OGA inhibitor TMG (1 μM; Chem Molecules), the most potent OGA inhibitor (Ki = 21 nM; human OGA) currently available 46,50 . GlcN and TMG were used in combination to ensure robust and lasting increases in protein O-GlcNAc levels.
In some experiments, we tested whether applying the HBP substrate GlcN alone was sufficient to modulate inhibitory transmission, as this approach more closely recapitulates the endogenous regulation of cellular O-GlcNAcylation.
Experimental design and statistical analysis. Recordings were analyzed using Clampfit 10.6. Outliers were determined as per outlier test in GraphPad Prism 8.1.0, La Jolla, CA, and excluded. For two-group comparisons, statistical significance was determined by two-tailed paired or unpaired Student's t-tests (parametric), or Wilcoxon matched-pairs signed rank test (non-parametric, paired). Multi-groups were analyzed using repeated measure one-way ANOVA with Sidak correction (parametric) or repeated measure two-way ANOVA with Sidak's test (non-parametric) (GraphPad Prism 8.1.0, La Jolla, CA). Data are displayed as mean ± SEM and p values less than 0.05 were considered statistically significant.