|Home | About | Journals | Submit | Contact Us | Français|
Familial Alzheimer’s disease (FAD) is caused by mutations in amyloid precursor protein or presenilins (PS1, PS2). Many FAD-linked PS mutations affect intracellular calcium (Ca2+) homeostasis by mechanisms proximal to and independent of amyloid production, although the molecular details are controversial. Here, we demonstrate that several FAD-causing PS mutants enhance gating of the inositol trisphosphate receptor (InsP3R) Ca2+ release channel by a gain-of-function effect that mirrors the genetics of FAD and is independent of secretase activity. In contrast, wild type PS or PS mutants that cause frontotemporal dementia have no such effect. FAD PS alter InsP3R channel gating by modal switching. Recordings of endogenous InsP3R in lymphoblasts derived from individuals with FAD or cortical neurons of asymptomatic PS1-AD mice revealed they have higher occupancy in a high open probability burst mode compared to that of InsP3R in cells with wild-type PS, resulting in enhanced Ca2+ signaling. These results indicate that exaggerated Ca2+ signaling through InsP3R-PS interaction is a disease-specific and robust proximal mechanism in FAD.
Alzheimer’s disease (AD) is a common form of dementia that involves slowly developing and ultimately fatal neurodegeneration. Most AD is sporadic and idiopathic and develops at ages over 60, but about 5% is inherited in an autosomal dominant manner due to mutations in amyloid precursor protein (APP) or presenilins (PS1, PS2) (1). Although familial Alzheimer’s disease (FAD) develops at ages as early as the late 30s, both familial and sporadic AD share hallmark features that include accumulation of β amyloid (Aβ) in extracellular plaques, intracellular neurofibrillary tangles comprised largely of hyper-phosphorylated tau, and cell atrophy and death in various brain regions (2–4). The consistent phenotypes suggest that both types of AD may share pathogenic origins. Nevertheless, the mechanisms by which these mutant proteins exert such devastating effects, and their roles and relationships in the two forms of AD, are still not clear. Insights into the molecular mechanisms and cellular functions of mutant proteins in FAD are likely to provide important clues into the etiology of AD pathogenesis and the identification of targets for therapeutic interventions.
Presenilins are transmembrane proteins that are synthesized on the endoplasmic reticulum (ER) and localized there (5). Together with nicastrin, APH-1 (anterior pharynx-defective 1), and PEN-2 (presenilin enhancer 2), PS forms a protein complex that is transported to the cell surface and to endosomes, where it functions as a γ-secretase that cleaves several type 1 transmembrane proteins, including APP (6, 7). γ-secretase cleavage of APP releases Aβ peptides, a major component of amyloid plaques in the brains of AD patients. Mutant PS are believed to affect APP processing by either enhancing the total production of Aβ or the relative proportion of the more amyloidogenic Aβ-42 form (8). In the amyloid hypothesis of AD, accumulation of amyloidogenic Aβ aggregates or oligomers is a proximal feature that causes neural toxicity leading to brain pathology (9, 10). However, FAD mutations in PS cause loss of secretase function, in contrast with the dominant gain-of-function indicated by the genetics of the disease (11). In addition to disrupting APP processing, many FAD-linked PS mutations affect intracellular calcium (Ca2+) homeostasis (12, 13). Although extracellular Aβ influences intracellular Ca2+ homeostasis in vitro (14, 15) and in vivo (16, 17), FAD-mutant PS also influences intracellular Ca2+ signaling by proximal, Aβ-independent mechanisms. Such Ca2+ signaling disruptions have manifested as attenuated capacitive Ca2+ entry (18–20), but most commonly as exaggerated Ca2+ liberation from the ER (18, 21–24), the major intracellular Ca2+ storage organelle. The molecular mechanisms underlying exaggerated ER Ca2+ release have been ascribed to enhanced loading of the ER lumen (23) due either to enhanced SERCA (sarco-endoplasmic reticulum Ca2+-ATPase) pump activity (25) or to disruption of a putative Ca2+ channel function of wild-type PS (26, 27). Alternately, exaggerated Ca2+ release has been accounted for by enhanced Ca2+ liberation from normal stores through inositol trisphosphate receptor (InsP3R) (21, 23) or ryanodine receptor (RyR) (22, 28, 29) Ca2+ release channels, both in vivo (22, 24, 28, 29) and in vitro (30–33), either as a consequence of enhanced channel abundance (28, 34–36) or, in the case of the InsP3R, enhanced activity in response to its ligand InsP3 (32, 37). Notably, enhanced agonist-induced InsP3R-mediated Ca2+ signals have been used diagnostically to identify individuals with FAD (31, 32). Biochemical interaction of the InsP3R with both wild-type (WT) and FAD-mutant PS1 and PS2 has been demonstrated (37). Single channel recordings of Sf9 insect cell InsP3R demonstrated that recombinant FAD-mutant PS1 and a FAD mutant-PS2 could enhance InsP3R Ca2+ release channel gating (37). These single channel studies were performed in the absence of Aβ or cellular pathology, suggesting that modulation of InsP3R gating is a fundamental mechanism that contributes to exaggerated Ca2+ signaling in FAD PS-expressing cells.
It is not known whether the effects of FAD PS on InsP3R gating represent a gain or loss of function. Moreover, although many (>100) PS mutations (especially in PS1) that cause FAD have been identified (38), only two FAD-mutant PS have been examined for their effects on InsP3R channel gating (37). In addition, some PS1 mutations result in frontotemporal dementia (FTD), a neurological disorder lacking Aβ accumulation (39, 40). If FAD PS-mediated alteration of InsP3R-mediated Ca2+ signaling is proximal in AD pathogenesis, then other FAD-mutant PS might be expected to have similar enhancing effects on InsP3R channel gating, whereas those associated with FTD might not. Previous studies of the effects of mutant PS on InsP3R investigated endogenous insect (Sf9 ovarian cells) and chicken (DT40 B lymphocytes) InsP3Rs (37), whereas AD, in which the pathological consequences are primarily in brain neurons, affects humans. Consequently, the relevance of these data in appropriate cell types with endogenous amounts of PS and InsP3R are unclear. Here, we studied InsP3R channel kinetics under the influence of several FAD- and FTD-mutant PS in four different systems, including transgenic AD mouse neurons, B-lymphoblasts derived from human FAD patient cells, and fibroblasts from PS 1 and 2 double knock-out cells. All FAD-linked PS mutations enhanced InsP3R single channel gating, leading to exaggerated intracellular Ca2+ signaling, whereas FTD-associated PS1 mutations did not affect InsP3R channel kinetics. Furthermore, the effects of FAD PS mutants were gain-of-function effects, consistent with the genetics of FAD. In contrast, the secretase activity of PS was not required. The results indicate that exaggerated Ca2+ signaling through InsP3R-PS interaction is a disease-specific and robust proximal mechanism in FAD.
To determine whether enhanced InsP3R channel activity is a phenotype conserved in FAD PS-expressing cells, we recorded single InsP3R channel activities in the presence of one of eight different PS mutants (PS1-L113P (leucine at residue 113 substituted with proline), -M146L, -L166P, -G183V, -D257A, -G384A, -D385A and PS2-N141I). We performed single-channel patch-clamp electrophysiology of the outer membrane of isolated Sf9 cell nuclei (41) 48 hr after infecting cells with recombinant baculovirus (Fig. S1). Because enhancement of InsP3R activity is more apparent at sub-saturating InsP3 concentrations (37), we used 100 nM InsP3 and 1 µM Ca2+ to sub-optimally activate channel gating. We consistently detected InsP3R channels with open probability (Po) of 0.27 ± 0.04 in membrane patches from control EVER1-(an irrelevant ER transmembrane protein) infected nuclei (Fig. 1A and B). InsP3R channels recorded in membrane patches from PS1-WT- or PS2-WT-infected cells had Po similar to those from EVER1-infected control cells (Po = 0.32 ± 0.04 and 0.25 ± 0.03, respectively; p >0.05; Fig. 1A and B). In contrast, InsP3R channel Po was significantly enhanced by 250% in nuclei from cells infected with mutant PS1-M146L (Po = 0.81 ± 0.02; Fig. 1A and B) to a degree similar to that achieved with saturating ligand concentrations (37). Increased Po resulted from a marked reduction of channel mean closed-time (τc; Fig. 1C). FAD-mutant PS2 (N141I) also markedly enhanced InsP3R channel activity (Fig. 1A and B), with Po increased by 200% (0.66 ± 0.05; Fig. 1B), also mainly due to a significant reduction of τc (Fig. 1C). Similar results were obtained for two other FAD-causing PS1 mutants: InsP3R channel Po was increased 200% with PS1-L166P (Po = 0.63 ± 0.08) or PS1-G384A (Po = 0.61 ± 0.05; Fig. 1A and B). Thus, all four FAD PS mutants examined had similar effects on InsP3R channel activity. The γ-secretase-dead mutants PS1-D257A and PS1-D385A, which have mutations in intra-membrane sites involved in PS1 catalytic activity, also significantly enhanced InsP3R channel activity, although to a lesser extent than the FAD mutants (Po = 0.50 ± 0.05 and 0.46 ± 0.08, respectively; Fig. 1A and B). Thus, the secretase activity of PS is not required for its effects on InsP3R gating. Po of channels recorded from cells infected with FTD-associated mutant PS1-L113P and PS1-G183V were 0.28 ± 0.04 and 0.29 ± 0.04, respectively, not different from controls (Fig. 1A and B). Thus, several FAD-mutant PS have similar effects on InsP3R gating, and these effects are not recapitulated in PS mutants associated with a different neurological disease.
To gain deeper insight into the mechanisms of InsP3R channel activation by FAD-mutant PS, we employed modal gating analysis. Previous studies demonstrated that ligand (InsP3, Ca2+) regulation of InsP3R gating is largely mediated by altering the propensity of the channel to gate in particular modes (42). Strongly activated channels gate in a high-Po H mode characterized by long bursting activities; an intermediate-Po I mode is characterized by fast channel openings and closings; and a low-Po L mode is characterized by long closed periods containing brief openings (42). In control nuclei isolated from EVER1-infected cells, the L gating mode was dominant, with the channel spending ~60% of its time in this mode and ~25% in the H mode (Fig. 1D). In nuclei from cells infected with either WT or FTD PS, similar modal gating distributions were observed (Fig. 1D). In contrast, the H mode was the dominant gating mode of InsP3R recorded from all of the FAD-causing mutant PS-expressing cells (Fig. 1D). Thus, FAD-mutant PS enhance InsP3R channel gating by mode switching, causing the channel to spend more time in the H mode at the expense primarily of the L mode (Fig. 1D; Fig. S2).
Enhancement of InsP3R channel activity by heterologous expression of mutant PS has been demonstrated in both Sf9 and DT40 cells [(37) and this study], systems that employ PS over-expressed in non-human cells. To determine the effects of endogenous PS in human cells, we studied InsP3R activity in normal and FAD human B cell lymphoblasts. Currents from endogenous human InsP3R single channels have never been previously recorded. Thus, we initially characterized endogenous InsP3R channels from human B lymphoblasts by nuclear membrane patch-clamp electrophysiology. In the absence of InsP3, no channel activity was apparent (n = 18; Fig. 2B), whereas with InsP3 (10 µM) in the pipette solution, we observed heparin-sensitive single channels with brief openings and long closings (n =15; Fig. 2A and B). These channels showed a linear I/V relationship with slope conductance ~475 pS (Fig. 2C), typical of mammalian InsP3R under these ionic conditions (43). InsP3R currents recorded from human B cells were long-lasting (Fig. 2A), with relatively low Po (0.18 ± 0.02, n=20; Fig. 2D).
We compared InsP3R gating in B lymphoblasts derived from three individuals with FAD, harboring PS1-M146L, PS1-A246E, or PS2-N141I (FAD lymphocytes), with that in B-lymphoblasts from two different age-matched individuals without FAD or FAD-associated PS mutations (control lymphoblasts) (Table 1). InsP3R in control lymphoblasts from the two individuals without FAD had low channel Po (0.18 ± 0.02 and 0.23 ± 0.03, respectively; Fig. 3A,B) with channel activities characterized by brief openings and relatively long closings (Fig. 3A and C). InsP3R Po recorded from lymphoblasts from all three individuals with FAD were increased 200 to 300% when compared with those from control lymphoblasts (PS1-M146L: 0.62 ± 0.05; PS1-A246E: 0.67± 0.06; PS2-N141I: 0.50 ± 0.04; Fig. 3A and B), mainly due to a marked decrease in τc (Fig. 3C), with many channels bursting for extended periods (Fig. 3E). In control lymphoblasts, the L and I gating modes dominated channel kinetics, whereas InsP3R analyzed in FAD lymphoblasts spent 50 to 75% of the time in the high Po H mode (Fig. 3D and E). Analogous results were obtained with low (100 nM) InsP3. InsP3R Po was 0.04 ± 0.01 in control lymphoblasts from an individual without FAD, whereas Po was 0.22 ± 0.05 in PS1-A246E FAD lymphoblasts (Fig. 3F–G). These observations in human B-lymphoblasts with endogenous PS and InsP3R are similar to those made in Sf9 and DT40 cells. FAD-linked PS mutations therefore have a robust, common effect to enhance InsP3R single channel activity in insect, avian, and human cells.
To determine whether these effects observed at the single-channel level are associated with altered [Ca2+]i signaling, we measured InsP3R-mediated Ca2+ signals in B lymphoblasts from the same individuals with FAD that were used for single-channel studies. InsP3R-mediated Ca2+ signals were elicited by cross-linking the B cell receptor (BCR) with IgM antibody. At high [IgM] (5µg/ml), 20% of cells responded with similar Ca2+ oscillations and spiking in both control and PS1-A246E FAD lymphoblasts (Fig. 4B and D), whereas a further 27% of the FAD lymphoblasts responded with exaggerated high-amplitude transient responses (Fig. 4A, B and C). With low-dose anti-IgM stimulation (50 ng/ml), Ca2+ oscillations/spiking were triggered in 19% ± 2% of control cells (Fig. 4E and G). Perfusion with xestospongin B, a membrane-permeable specific InsP3R inhibitor (44), reversibly inhibited them indicating that they were due to periodic Ca2+ release through the InsP3R (Fig. S3). In FAD lymphoblasts, both the percentage of responding cells and the oscillation and spiking frequency were increased (Fig. 4E, G and H). Perfusion with culture medium containing 10% FBS, which generates ongoing low InsP3 production (45), induced spontaneous Ca2+ oscillations/spiking in 25 ± 5% of control lymphoblasts (Fig. 4F and G). In contrast, the percentage of PS1 FAD lymphocytes displaying spontaneous Ca2+ oscillations was increased by 100% and the oscillation and spiking frequency doubled (Fig. 4F, G and H). The percentage of spontaneously oscillating PS2-N141I FAD cells was similar to that in control lymphoblasts, however, the oscillation frequency was increased (Fig. 4F, G and H). These responses are consistent with an enhanced sensitivity and activity of InsP3-mediated Ca2+ release in human FAD lymphoblasts, consistent with the enhanced InsP3R channel activity recorded in these cells.
Ca2+ signaling disruption has been observed in fibroblast or lymphoblast lines derived from human FAD cells [here and (30, 32, 46)]. Our results above implicate mutant PS-enhanced InsP3R channel gating as the underlying mechanism. To determine if this molecular mechanism also operates in brain neurons, we isolated cortical neurons from embryonic day 14 to 16 (E14 to E16) WT C57BL/6 and 3×Tg-AD mice and recorded single InsP3R channel activities in nuclear envelopes from isolated nuclei. 3xTg-AD mice contain PS1-M146V knocked into the PS1 locus, and exhibit age-dependent amyloid plaques, neurofibrillary tangles, and cognitive decline starting at 3 to 6 months of age (3, 47). In nuclei isolated from control C57BL/6 mice, channel currents were not observed in the absence of InsP3 (Fig. 5B). With 10 µM InsP3, and 1 µM Ca2+, heparin-sensitive (Fig. 5B), channels with a linear slope conductance of ~375 pS (Fig. 5C) were recorded (Fig. 5A and B) with gating characterized by short openings (τo = 2.25 ± 0.11 ms) and relatively long closures (τc = 52.7 ± 12.7 ms) with Po = 0.06 ± 0.01 (Fig. 5D). Po was enhanced by 700% (0.43 ± 0.05; Fig. 5B and D) in nuclei isolated from 3xTg-AD mice. Increased Po was caused by markedly prolonged τo (10.22 ± 1.57 ms) together with shortened τc (14.61 ± 3.04 ms). The I and L modes dominated channel gating in control C57BL/6 neurons, whereas the H mode was the major gating mode in 3xTg-AD neurons (Fig. 5B,E, and F).
Our results reveal that FAD-mutant PS consistently enhances InsP3R channel gating. To explore the mechanisms involved, we recorded endogenous InsP3R channels in nuclei from embryonic fibroblasts (MEF) derived from PS double-knockout mice (48, 49). In the absence of PS, the endogenous MEF InsP3R Po was 0.30 ± 0.03 (Fig. 6). Stable expression of human PS1 by retroviral transduction was without effect on InsP3R Po (0.27 ± 0.05), whereas FAD mutant PS1-M146L approximately doubled channel gating activity (0.54 ± 0.05), by enhancing H-mode gating (Fig. 6). Similar results were obtained in independently-derived MEF clones (Fig. S4). These results indicate that the effects of FAD-mutant PS on InsP3R channel involve a gain of function. As shown above in Sf9 cells, this function is independent of PS secretase activity, because the secretase-dead PS1-D257A also enhanced channel activity (Fig. 6).
In summary, the above results demonstrate a consistent and robust phenotype associated with the presence of mutant PS linked to FAD. In five different cell systems (four here and DT40 cells previously) from four species, FAD-causing mutant PS resulted in exaggerated responses of InsP3R Ca2+ release channels and exaggerated Ca2+ signals in response to agonist stimulation, as well as a small degree of constitutive Ca2+ signaling. The FAD-mutant PS phenotype involves gain-of-function effects, consistent with disease genetics, and is independent of the secretase function of PS. Moreover, the FAD-mutant PS phenotype is not observed in cells harboring either wild-type PS or PS mutants associated with a different disease, FTD. The FAD-mutant PS phenotype is manifested independently of any pathology associated with AD, and, in the mouse model, precedes such pathology. Moreover, it is apparent in physiologically-relevant cell types (cells derived from humans with FAD and AD mouse neurons) with all proteins present in endogenous amounts. We propose that exaggerated Ca2+ signaling through an InsP3R-PS interaction is a robust proximal gain-of-function molecular mechanism in FAD.
Our single channel analyses demonstrate that FAD-mutant PS enhances single channel activity of the InsP3R by affecting modal gating kinetics, the major mechanism by which InsP3 and Ca2+ regulate the channel (42). That FAD-mutant PS drives the channel into the H mode may have important physiological implications. The channel open time when it in the L gating mode (~10 ms) is short enough that it may not increase local [Ca2+] sufficiently to recruit additional InsP3R- or RyR-mediated Ca2+ release by Ca2+-induced Ca2+ release (CICR). In contrast, the much longer activity bursts of the channel in the H mode (>200 ms) will provide a sufficiently large flux of Ca2+ to enable a normally local Ca2+ signal to be amplified and propagated by CICR (50). Because InsP3R and RyR are clustered and spatially localized to different regions of cells to provide local [Ca2+]i signals as a critical element of physiological specificity, mode-shifting by mutant PS-induced FAD may result not only in exaggerated local Ca2+ signaling, but also a disruption of spatial specificity by enabling CICR to transmit the signals more globally (42, 50). Exaggerated and spatially disrupted Ca2+ signaling may in turn impinge on APP processing (16, 51–54), calpain activation (16, 54), and tau phosphorylation (55, 56), linking our findings here to the amyloid hypothesis of AD (Fig. 7).
Spodoptera frugiperda cells (Sf9, BD Biosciences) were maintained as described (37, 41). Human PS baculovirus constructs (PS1-WT, PS1-L113P, PS1-M146L, PS1-L166P, PS1-G183V, PS1-D257A, PS1-G384A, PS1-D384A, PS2-WT and PS2-N141I) were subcloned into pFastBac1 and baculoviruses were generated using the Bac-to-Bac system (Invitrogen). Expression was confirmed by Western blotting with antibodies directed against PS1 or PS2 (anti-PS1 and anti-PS2, respectively) as described (37). B-lymphoblast lines derived from human FAD patients and normal individuals (Table I; Coriell Institute, Camden, NJ) were maintained at 37°C (95/5% air/CO2) in RPMI 1640 (Invitrogen) supplemented with 15% fetal bovine serum (Hyclone), 2 mM L-glutamine, 100 units/ml penicillin, and 100 µg/ml streptomycin. PS−/− (genetically deficient in PS1 and PS2), stable human PS1-WT, mutant PS1-M146L and PS1-D257A MEF cells were grown in DMEM supplemented with 10% fetal bovine serum (57, 58). To generate stable lines expressing comparable amounts of PS1 proteins, human PS1 cDNAs were introduced into pMX-IRES-EGFP retroviral vector, and PS retroviruses generated using Retro-X system (Clontech) were added to the parental PS−/− MEF cells, and GFP positive cells were sorted by FACS. PS expression was confirmed by Western blot.
Primary cortical neurons were prepared from embryonic day 14 to 16 (E14 to E16) 3xTg-AD mice as described (37). Neurons from C57BL/6 mice (Charles River) served as controls. In brief, dams were killed with CO2, and embryos were removed by cesarean section. Brains from littermates were removed and placed into PBS. After the meninges were removed, cerebral cortices were dissected, minced, and digested with 0.25% trypsin in PBS at 37°C for 20 min. Dissociated cells were washed twice with DMEM supplemented with 10% FBS, triturated with a fire-polished Pasteur pipette and re-suspended in Neurobasal medium supplemented with 1x B27 (Invitrogen). All animal procedures were approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC).
Human B-lymphoblasts (Coriell Institute, Camden, NJ) were plated onto a CellTek-(BD Biosciences) coated glass-bottom perfusion chamber mounted on the stage of an inverted microscope (Eclipse TE2000; Nikon, Melville, NY) and incubated with fura-2 AM (2 µM; Invitrogen) for 30 min at room temperature in Hanks’ balanced salt solution (HBSS, Sigma, St. Louis, MO) containing 1% BSA. Cells were then continuously perfused with HBSS containing 1.8 mM CaCl2 and 0.8 mM MgCl2 (pH 7.4). Ca2+ signals were elicited by cross-linking the B cell receptor (BCR) with 50 ng/ml anti-human IgM antibody (SouthernBiotech, Birmingham, AL). In some experiments, cells were perfused with complete culture medium containing 10% FBS. Fura-2 was alternately excited at 340 and 380 nm, and the emitted fluorescence filtered at 510 nm was collected and recorded (37, 45) using a CCD-based imaging system running Ultraview software (PerkinElmer, Waltham, MA). Dye calibration was achieved by applying experimentally determined constants to the standard equation [Ca2+] = Kd·β·(R − R min)/(R max − R).
Preparation of isolated nuclei from cells was performed as described (37, 41, 45). In brief, cells were washed twice with PBS and suspended in nuclear isolation solution containing (in mM): 150 KCl, 250 sucrose, 1.5 β-mercapoethanol, 10 Tris-HCl, 0.05 phenylmethylsulphonyl fluoride and protease inhibitor cocktail (Complete, Roche Diagnosis, Indianapolis, IN), pH 7.3. Nuclei were isolated using a Dounce glass homogenizer and plated onto a 1-ml glass-bottomed dish containing standard bath solution (in mM): 140 KCl, 10 HEPES, 0.5 BAPTA, and 0.192 CaCl2 (free [Ca2+] = 90 nM). The pipette solution contained (in mM): 140 KCl, 10 HEPES, 0.5 dibromo-BAPTA, and 0.001 free Ca2+, pH 7.3. Free [Ca2+] in solutions was adjusted by Ca2+ chelators with appropriate affinities and confirmed by fluorometry as described (41). Data were recorded at room temperature and acquired using an Axopatch 200A amplifier (Axon Instruments), filtered at 1 kHz, and digitized at 5 kHz with an ITC-16 interface (Instrutech) and Pulse software (HEKA Electronik).
Segment of current records exhibiting current levels for a single InsP3R channel were idealized using QuB software (University of Buffalo) with SKM algorithm (59, 60). Channel gating kinetics and modal gating behaviors were characterized as described (42). In brief, very short closing events (< 10 ms), presumably caused by ligand-independent transitions, were removed by burst analysis (61) after idealization with QuB. Modal gating assignment was then achieved by plotting and examining durations of channel burst (tb) and burst-terminating gaps (tg) as described (42). In Sf9 cells, we set Tb =100 ms and Tg = 200 ms for the detection of modal transitions. In both human B-lymphocytes and mouse cortical neurons, we set Tb = 50 ms and Tg = 100 ms for the detection of modal transitions. Data were summarized as the mean ± SEM, and the statistical significance of differences between means was assessed by using unpaired t tests or one-way ANOVA with Dunnett’s post hoc comparison test. Differences between means were accepted as statistically significant at the 95% level (p <0.05).