Abstract
Background:
In Alzheimer’s disease (AD), synaptic dysfunction is thought to occur many years before the onset of cognitive decline.
Objective:
Detecting synaptic dysfunctions at the earliest stage of AD would be desirable in both clinic and research settings.
Methods:
Population voltage imaging allows monitoring of synaptic depolarizations, to which calcium imaging is relatively blind. We developed an AD mouse model (APPswe/PS1dE9 background) expressing a genetically-encoded voltage indicator (GEVI) in the neocortex. GEVI was restricted to the excitatory pyramidal neurons (unlike the voltage-sensitive dyes).
Results:
Expression of GEVI did not disrupt AD model formation of amyloid plaques. GEVI expression was stable in both AD model mice and Control (healthy) littermates (CTRL) over 247 days postnatal. Brain slices were stimulated in layer 2/3. From the evoked voltage waveforms, we extracted several parameters for comparison AD versus CTRL. Some parameters (e.g., temporal summation, refractoriness, and peak latency) were weak predictors, while other parameters (e.g., signal amplitude, attenuation with distance, and duration (half-width) of the evoked transients) were stronger predictors of the AD condition. Around postnatal age 150 days (P150) and especially at P200, synaptically-evoked voltage signals in brain slices were weaker in the AD groups versus the age- and sex-matched CTRL groups, suggesting an AD-mediated synaptic weakening that coincides with the accumulation of plaques. However, at the youngest ages examined, P40 and P80, the AD groups showed differentially stronger signals, suggesting “hyperexcitability” prior to the formation of plaques.
Conclusion:
Our results indicate bidirectional alterations in cortical physiology in AD model mice; occurring both prior (P40-80), and after (P150-200) the amyloid deposition.
INTRODUCTION
The first signs of cognitive impairment may begin 20–30 years before a clinical Alzheimer’s disease (AD) diagnosis is established, and synaptic dysfunction in AD is thought to emerge long before any substantial loss of neurons [1, 2]. In humans, the apparent loss of synapses and nerve cells is manifested by macroscopic brain atrophy even in the prodromal stage of disease [3]. Hence, AD is recognized as a disease of synaptic failure [4, 5]. When cognitive impairments become strongly manifested, large brain areas are already irreversibly damaged and the causal therapies are dubious [1, 6]. It would be useful to identify synaptic dysfunction at an early stage of AD, in both: [a] human patients and [b] animal models of disease; before the onset of irreversible structural changes, so that early interventions for improving synaptic or cognitive functions can be explored.
Mouse models of AD have been valuable for monitoring changes of synaptic function associated with the AD pathology [7–9]. The current popular approach is to monitor long-term potentiation (LTP) or depression (LTD) in the Schaffer collateral pathway [10, 11], or to monitor neuronal spiking via calcium imaging [12]. Modern optical methods allow recordings of neuronal activity occurring in many neurons simultaneously, where all recorded neurons belong to only one cell type of interest (e.g., Layer 2/3 pyramidal neurons) [13]. Optical imaging of membrane potential changes, using Genetically-Encoded Voltage Indicators (GEVIs) [14, 15], is uniquely suited for monitoring subthreshold (synaptic) potentials [16–18]. In brain slice preparations, GEVI measurements can be performed in several cortical layers simultaneously [19], thus giving researchers an experimental tool for quick and thorough assessment of the evoked network responses in Control (CTRL) and Test (AD) animals.
By creating a new mouse strain called APP/PS1-GEVI (AD-GEVI), we could examine functional changes in the neocortical region, unlike the conventional LTP/LTD assays that are mainly restricted to the hippocampal circuits [20]; but see [21]. The neocortical area of the new AD-GEVI mice began accumulating amyloid plaques at age P150 (postnatal day 150). In older animals (P150 and P200), synaptically-evoked responses were weaker in the AD group compared to the age- and sex-matched controls (CTRL). Quite the opposite, at younger ages (P40 and P80) physiological responses were stronger in AD versus CTRL (“hyperexcitation”). In summary, in the new mouse model of AD, we observed bidirectional physiological changes, with a switch in signal polarity occurring at the emergence of plaques (around P150). Before plaque growth, synaptically evoked cortical responses were stronger (hyperexcitation), while after the plaque accumulation the same responses were found to be weaker (hypoexcitation) in the AD groups compared to the Control (healthy) groups.
MATERIALS AND METHODS
Animals
A transgenic animal line (B6.Cg –B6.Cg-Tg(APPswe,PSEN1dE9)85Dbo/Mmjax) was dubb-ed an “APP/PS1 mouse” and used for crossing with the PAN-GEVI mouse line. The transgenic animal line, PAN-GEVI, was donated by Thomas Knopfel (Imperial College London, UK). The PAN-GEVI mice expressed chimeric voltage-sensitive fluorescent protein (chi-VSFP) in all cortical pyramidal neurons (CaMK2A-tTA;tetO-chiVSFP). The offspring of the crossing between APP/PS1 and PAN-GEVI, positive for both the AD and GEVI genes, was dubbed “AD-GEVI” or the “AD group”. All animals were housed in standard conditions with free access to food and water, in a 50% dark/light cycle. Brain slices were harvested from transgenic mice (age 37–247 days, both sexes) according to the animal protocols approved by the UConn Health Institutional Animal Care and Use Committee (IACUC), in accordance with ARRIVE guidelines and institutional regulations.
Wide-field voltage imaging
Following a deep anesthesia with isoflurane, mice of both sexes (ages P37–P247) were decapitated. Brains were extracted with the head immersed in ice-cold saline. The saline contained (in mM) 125 NaCl, 26 NaHCO3, 2.3 KCl, 1.26 KH2PO4, 2 CaCl2, 1 MgSO4, and 10 glucose. Coronal slices (300μm) were cut from the fronto-parietal cortex, incubated at 37°C for 30 min and then at room temperature. Acute brain slices were transferred to an Olympus BX51WI upright microscope and perfused with the aerated (5% CO2/95% O2) saline. All experimental measurements were performed at 34°C. Synaptic stimulation was achieved through a computer-generated TTL pulse and stimulus isolation unit (IsoFlex, A.M.P.I., Israel). The synaptic stimulation electrodes were pulled from 1.5 mm borosilicate glass with filament (resistance ∼2 MOmh) and backfilled with saline. Triplets of synaptic shocks (1 ms duration, 135 nA) at 120 ms inter-stimulus interval, ISI, 8.3 Hz (Train-1) and 12 ms ISI, 83 Hz (Train-2) were delivered in the same optical recording sweep, separated by a 1.1-s interval. Optical trials were typically 3 s of light exposure with at least 10-s intervals between two consecutive sweeps. Optical signals were sampled at 1.020-ms full-frame interval (∼1 kHz frame rate) with NeuroCCD camera (80×80 pixel configuration; RedShirtImaging, Decatur, GA). The GEVI (chi-VSFP) was excited using a 470 nm light emitting diode, LED (pE, CoolLED, Andover, UK) and imaged using a filter set: excitation: 480/40 nm; dichroic 510 nm, and emission: 535/50 nm. The GEVI (chi-VSFP) signals recorded at a green emission filter (535/50 nm) have a negative polarity in our raw data records. In the figures, however, these optical signals have been inverted. We feel that inverted GEVI optical signals (positive with depolarization) are more appropriate for presentations.
Still imaging
At the end of voltage-imaging sessions, brain slices were removed from the recording chamber, fixed in 4% paraformaldehyde, incubated in 1% aqueous Thioflavin-S, mounted on microscope slides and photographed on a Keyence Fluorescence Microscope BZ-X800 using a 2x dry objective. Each brain slice (300μm thick) was examined first on the top side, and then flipped over to examine the bottom side. Both (top and bottom) full-slice surfaces were included in the analysis. Confocal images were obtained on an Olympus BX51 microscope equipped with a Thorlabs confocal laser scanning head and a two-channel detection module. Both green and red images were excited at 488 nm but collected at 510–545 nm (green) and 578–625 nm (red).
At the end of voltage-imaging sessions, brain slices were removed from the recording chamber, fixed in 4% paraformaldehyde, incubated in 1% aqueous Thioflavin-S, mounted on microscope slides and photographed on a Keyence Fluorescence Microscope BZ-X800 using a 2x dry objective. Each brain slice (300μm thick) was examined first on the top side, and then flipped over to examine the bottom side. Both (top and bottom) full-slice surfaces were included in the analysis. Confocal images were obtained on an Olympus BX51 microscope equipped with a Thorlabs confocal laser scanning head and a two-channel detection module. Both green and red images were excited at 488 nm but collected at 510–545 nm (green) and 578–625 nm (red).
Data analysis
Optical traces were conditioned and analyzed in Neuroplex (RedShirtImaging, LLC). Bleach correction was done by subtracting an exponential fit from the optical trace. Each trace was a product of temporal averaging (n = 4 sweeps), spatial averaging (21–37 pixels), low-pass Gaussian filter with 77 Hz cutoff, and high-pass Tau filter (10), unless stated otherwise. Optical signal amplitude was measured in Neuroplex as fractional change in light intensity (ΔF/F in %); typical for calcium and voltage imaging experiments [22]. Resting fluorescence intensity (RFI) was measured at standard illumination –fixed LED intensity, by averaging the resting light intensities of 37 neighboring pixels in cortical layer 2/3. Signal to noise ratio (SNR) was measured as the optical signal amplitude (filtered at 77 Hz) divided by peak-to-peak noise measured at unfiltered baseline preceding a biological signal. Temporal Summation (paired pulse facilitation) was calculated as a ratio of the amplitudes of the third peak divided by the first, in the same optical trace. Distance Attenuation was also calculated as a ratio of corresponding peak amplitudes at two recording locations (ROIs). Quantifications were organized in MS Excel. Results are presented as mean±standard error of the mean (SEM), unless stated otherwise. Statistical significance was determined with unpaired, two-tailed Student’s t-tests. Results were considered statistically significant if the p≤0.05.
RESULTS
Two mouse lines (APP/PS1 and PAN-GEVI) were crossed, resulting in “AD-GEVI” mouse line, which combines an AD pathology (APPswe, PSEN1dE9) with: [1] dense, [2] cell-type-specific (neocortical excitatory pyramidal cells), and [3] cortex-wide expression of genetically-encoded voltage indicator (GEVI) [23]. Sixty (60) male and fifty-two (52) female animals were divided in two experimental groups (AD & CTRL) and used in ex vivo recordings. The youngest animal was 37 days and the oldest animal was 247 days old. Voltage imaging experiments were carried out in 2–4 brain slices belonging to the same animal.
No noticeable effect of GEVI on the amyloid deposition
We first examined whether the introduction of GEVI in APP/PS1 mice would disrupt the course of amyloid deposition. Thioflavin-S-positive plaques were counted in two cortical regions. Region-1 (“superior cortex”, Fig. 1A) contained the dorsal and upper lateral surfaces (e.g., primary motor, primary somatosensory cortex). Region-2 (“inferior cortex”, Fig. 1A) contained the lower lateral and inferior cortical areas (e.g., agranular insular cortex, piriform cortex).
AD animal model expressing GEVI (AD-GEVI). A) Left (image): Coronal section from an AD-GEVI animal, P202, Female. Thioflavin-S staining of amyloid plaques. L4 –layer 4; c.c. –corpus callosum. Region-1 encompasses superior cortex, while Region-2 encompasses inferior cortex. Right (bar graphs): Plaques were counted inside regions 1 and 2, in four age groups (Age Days). In CTRL animals, we found zero plaques. The number of slice surfaces per experimental group is shown in boxes (n). In AD animals, age 40 and 80 we found zero plaques, with an exception of 1 animal. Older AD animals (e.g., 150) showed high plaque counts. Region-1 is larger than Region-2, hence higher plaque counts in Region-1. B) AD Group, Region-1 data (plaque counts) separated by both age and sex reveal that the emergence of plaques begins at P150, and appears to be stronger in females. #p > 0.05. Asterisk indicates p < 0.01. C) GEVI expression in control animal, P213, Female. Two emission channels (green and red) are displayed, as chi-VSFP is comprised of two fluorophores, mCitrine and mKate2. D) Each confocal image is comprised of two channels (green and red) –age group P200. d1 shows dense GEVI labeling of pyramidal neurons in a plaque-free zone (healthy cortical tissue of an AD-GEVI animal). d2 shows small, rounded plaques surrounded by undisturbed cortical cytoarchitecture. E) Same as in D, except here the thioflavin S-stained amyloid plaques (e1 and e2) are large and with rough edges. Inset: Gray scale blowup of the plaque. White arrow marks a disturbance (swirl) in the cortical layering, found in 33% of the plaques examined (pie chart, n = 150 plaques examined in 8 animals of the P200 AD Group).
The healthy littermates (CTRL), which expressed only GEVI (but not the mutant-APP/PS1), did not develop plaques regardless of the animal age examined (Fig. 1A, Graph, CTRL Group). All CTRL animals (n = 58 animals, n = 132 slice surfaces) across the four age groups (40, 80, 150, and 200 days) were plaque-free (i.e., plaque count = 0). In the AD-GEVI mice (AD Group), the appearance of plaques occurred simultaneously in the superior and inferior cortex, starting with the age group P150, and increasing further in the age group P200 (Fig. 1A, Graph, AD Group, n = 54 animals, n = 134 slice surfaces from the 4 age groups). The robustly developed amyloid plaques were seen in 100% of the older AD animals (P150 and P200), indicating that GEVI expression did not disrupt the formation of amyloid plaques. No plaques were found in the age group AD P40 (n = 16 slice surfaces from 8 animals), and only one AD animal showed some plaques in the age group P80 (n = 16 slice surfaces from 8 animals, Fig. 1A, Graph, AD Group, P80).
Sex is a biological variable that influences the plaque counts
On average, female animals developed more plaques at P150 and P200 (Fig. 1B). At ages P200 and P80, the differences between sexes were not statistically significant (#p > 0.05). At age P150, the female animals developed significantly more plaques than their male counterparts did (*p < 0.01, n = 20 slices from male, and 22 from female animals). These data suggested that in the new AD-GEVI model, the female sex begins to develop plaques before the male sex does.
The plaques have no impact on the GEVI expression
The dual emission of the GEVI, chi-VSFP (Fig. 1C, mCitrine and mKate2), is useful for studying GEVI expression patterns [23]. A darker band in both green and red channels is indicative of fewer GEVI-positive neurons inside the band (Fig. 1C, L4). On the other hand, strong fluorescence in the corpus callosum (Fig. 1A, c.c.) indicates that axons from GEVI-labeled pyramidal neurons are tightly bundled there (“tough body”). Overlapping dendrites and axons of pyramidal neurons form a “sea of fluorescence”, while neuronal cell bodies appear as “dark holes”, in both green and red channels (Fig. 1D, d1). To test if amyloid-β (Aβ) plaques disturb the expression of GEVI, brain slices were counterstained with Thioflavin-S and examined under a confocal microscope. In eight AD-GEVI animals, ages over 190 days and with abundant Aβ plaques, we checked 10–20 individual plaques per slice at 60x (0.9 NA) lens magnification (40μm confocal aperture), and found no evidence that GEVI expression in the immediate vicinity of a plaque (Fig. 1D, d2; Fig. 1E, e1-e2) was any different than the GEVI expression in the healthy, plaque-free cortex (Fig. 1D, d1). However, in 33% (49 out of 150) plaques examined in this way, we observed that cortical layers were disturbed with some “swirling” of cell bodies and neuropil around the plaques (Fig. 1E, white arrows). These disruptions (deviations from the orderly organization of the cerebral cortex) were especially prominent near large-diameter plaques characterized with rough edges (pointy plaque protrusions, Fig. 1E, e1 and e2, Inset). Large diameter plaques appear to push the cortical cells bodies and crowd them inside the halo surrounding the plaque (Fig. 1E, white arrows); where the pyramidal cell bodies appear to deviate from their strict radial orientation—hence the term “swirl”. Both the “halos” and cortical tissue “swirls” were notably less prominent around small plaques characterized with smooth plaque surfaces (Fig. 1D, d2). Specifically, 67% of the plaques examined (101 out of 150) did not show any disturbance in the cortical architecture (Fig. 1E, pie chart, Smooth).
The GEVI resting fluorescence intensity does not decline with aging
The GEVI expression levels were tested by measuring the resting fluorescence intensity (RFI) in cortical layer 2/3 on acute brain sections (Fig. 2A). On each sample, we selected three ROIs within the cortical layer 2/3 (L2/3), where each ROI averages light intensity from 37 pixels (Fig. 2B). These three ROI values were then averaged to represent one brain slice. Per each animal, 2–4 brain slices were measured. No significant differences existed between the AD versus the age-matched CTRL animals (Fig. 2C, compare narrow dark-gray versus narrow light-gray bars within each age group). Combined AD + CTRL data (here dubbed “ALL”) indicated that RFI was the lowest in the youngest animals (age P40, Fig. 2C, wide bars with stripes). A polynomial fit through the “ALL” data points, showed a slow rise with stabilization at older ages. Lower RFI values in the age group P40 were most likely due to a rapid growth of dendritic branches and dendritic spines at young age, resulting in new membranes initially void of GEVI—it takes time for GEVI to accrue into an immature membrane. Also, the density of axons and dendrites may be higher in older animals, hence more fluorescent membrane is packed into the unit of volume. In corpus callosum (Fig. 1A, c.c.), the extracellular space is narrow and the density of packed membranes is high, giving this band a significantly higher RFI than the RFI measured in layer 2/3 of the same brain slice.
AD-GEVI animal model: GEVI expression with age and optical signal quality with age. A) Resting fluorescence intensity (RFI) from the surface of acute brain slice, captured by voltage imaging camera. Twelve slices from 12 animals are displayed on the same spatial scale. White labels indicate days of age. B) Three regions of interest (ROIs) are selected in layer 2/3 (L2/3). Each ROI averages RFI from 37 pixels. C) RFI in arbitrary units (a.u.) are averaged by age group. Data is presented for the AD group (narrow dark-gray bar), CTRL group (narrow light-gray bar) and lumped AD+CTRL = ALL (wide bar with stripes). Number of brain slices (n) examined in the AD Group, ages P40, P80, P150, and P200 was 26, 53, 62, and 56, respectively. In the CTRL group, the “n” was 22, 27, 71, and 45, respectively. In the ALL group, the “n” was, 48, 80, 133, and 101, respectively. Trend line is a 2nd order polynomial fit through “ALL” data points. D) A region of interest (ROI) closest to the synaptic stimulation electrode was selected for measuring signal-to-noise ratio (SNR). E) Synaptically-evoked optical traces from 16 animals (4 per age group). Each trace is an average of 12 sweeps. F) Inset: “noise” and “signal” are labeled on the optical trace. Graph: Average SNR values across 4 age groups, separated into AD (narrow dark-gray bar) versus CTRL (narrow light-gray bar), and lumped AD+CTRL = ALL (wide bar with stripes). Asterisk indicates p < 0.05. Trendline is a linear fit through “ALL” data points; slope = –0.32.
The GEVI optical signal quality does not decline with aging
Signal to noise ratio (SNR) was measured at the location at which the synaptic stimulation electrode touches the surface of the brain slice (Fig. 2DE, ROI). No significant differences in SNR existed between the AD versus the age-matched CTRL animals of the same age group (Fig. 2F). No statistically significant differences between three age groups: P40, P80, and P200. However, at P150, the SNR was significantly smaller than the SNR at P40, which is indicated by an asterisk in Fig. 2F. This decrease in the SNR at age P150 caused a small negative slope of the linear trend line (k = –0.32), which we consider negligible and unlikely to compromise the comparisons between age-matched AD versus CTRL groups (AD versus CTRL comparisons inside the same age group).
Evoked cortical depolarizations
All measurements were carried out in 300-μm thick brain slices. The stimulus train delivered in L2/3, Train-1 (8.3 Hz) produced three separate transients (peaks 1, 2, and 3), while Train-2 (83 Hz) in the same layer produced only one signal peak due to temporal summation, here termed the 4th peak (Fig. 3A1, 4th peak). Optical signals were analyzed in three regions of interest (ROIs S, P, and Q). Each ROI was a spatial average of 37 pixels (Fig. 3A1).
Distance dependent attenuation; Temporal summation and Refractory Escape. A1) Image: brain slice, AD group, age 197 days, male. Scale, 300μm. Synaptic stimulation was delivered in L2/3. Three ROIs are selected for data quantification –235μm apart, along the L2/3 lamina. The 1st peak traces (upper set) and the 4th peak traces (lower set) are displayed on the same amplitude and time scale. A2) Same as in A1, except the CTRL group (animal age 202 days). A3) A fraction of the signal amplitude that reaches proximal ROI-P (Propagation) is quantified as a ratio between the signal amplitude in ROI-P divided by the signal amplitude in ROI-S. Smaller fraction indicates stronger distance-dependent attenuation. ‘n’ –indicates number of brain slices. Asterisk, p < 0.05. A4) The P200 data set separated by sex. B1, B2) Sixteen representative synaptically-evoked cortical signals from the age group P200, separated into 4 categories by condition (AD and CTRL) and sex (Male and Female). Each trace belongs to a different animal. Animal age (in Days) is printed on the trace. All 16 traces are on the same vertical (0.1 %) and time (200 ms) scales. Black horizontal lines indicate amplitudes of the 1st peak. Label (%) above the trace indicates: ratio between the 3rd peak and the 1st peak, in %. B3) Average ratio (3rd peak/1st peak) obtained at the stimulation site “ROI-S” was plotted across four age groups - no statistically significant difference between AD versus CTRL. B4) Male and female groups, P200. C1) AD group, female. Image: fluorescence, brain slice with stimulation electrode, three ROIs, and animal age. Scale, 300μm. Optical signal from ROI-P is selected for display. Horizontal gray line marks: the time period between the 3rd and 4th peak, as well as the amplitude-difference between the 3rd and the 4th peak. C2) Same as C1, except CTRL female group. C3) “Refractory Escape” obtained in ROI-S, expressed as ratio of 4th/3rd peak (in %). C4) Male and female groups, P200.
We did not notice any drastic (obvious at visual inspection) voltage waveform differences between the AD animals (Fig. 3A1), and the age-matched CTRL animals (Fig. 3A2), except that CTRL animals sometimes appeared to show stronger signal amplitudes. Therefore, we carried out quantifications of the optical data. To investigate the earliest time point for detecting any waveform changes in AD-GEVI mice, we designed four age groups: P40, P80, P150, and P200 days, with a margin±∼10 days. Specifically, into the group “P40”, we included animal ages from 37 to 51 days. Group P80, P150, and P200, included animals in the range 72–92 days, 141–158 days, and 192–247 days, respectively. Each age group contained two conditions (AD versus CTRL), hence our data are comprised of 8 experimental groups (eight bars in Fig. 3A3).
Distance-dependent attenuation
We hypothesized that an impaired physiological function in AD animals will be manifested by a weaker propagation of synaptically-evoked depolarizations through the cortical tissue—a stronger amplitude-decline with distance, which would be manifested by a smaller ROI-P / ROI-S amplitude-ratio. No differences between AD and CTRL were detected in three animal ages, P40, P80, and P150 days (Fig. 3A3). However, in the oldest age group (P200), the ROI-P / ROI-S amplitude ratio in the AD condition was significantly lower compared to the CTRL condition (Fig. 3A3, asterisk). The P200-AD-Male showed a significantly weaker signal propagation strength compared to P200-CTRL-Male (Fig. 3A4, asterisk). The P200-AD-Female group also showed smaller amplitudes than the P200-CTRL-Female, but this difference was not statistically significant (Fig. 3A4, Female). Next, we asked if the physiological impairment was different between sexes (this statistical comparison is not marked in Fig. 3A4). We found no significant differences between P200-AD-Male versus P200-AD-Female (p > 0.05), or P200-CTRL-Male versus P200-CTRL-Female (p > 0.05). Altogether, these data suggest that in younger animals (40, 80, and 150 days), distance dependent attenuation is not affected by a developing AD pathology. However, in older male animals (age P200), the spatial spread of synaptically evoked population signals through the L2/3 parenchyma was weaker in the AD condition.
Temporal summation
“Paired pulse ratio” (PPR) is greater in synapses with the overall smaller release probability or smaller initial EPSP amplitude [24]. Visual inspections of evoked cortical depolarizations in the AD group (Fig. 3B1), and in the age-matched and sex-matched CTRL group (Fig. 3B2), did not reveal any obvious distinctions that would allow us to categorize animals into AD and CTRL categories based solely on the visual inspection of voltage waveforms. Numerical analysis showed that across all ages examined (P40, P80, P150, and P200 days), the efficiency of temporal summation defined as the 3rd Peak/1st Peak ratio, was similar between AD versus CTRL (Fig. 3B3, Summation), regardless of the ROI examined (ROI-S, -P, or –Q; data not shown for ROI-P and ROI-Q). Separation of the P200 age group into Male and Female animals revealed that P200-AD-Male showed stronger temporal summation (Fig. 4B4, asterisk). However, two factors indicate that this physiological parameter “temporal summation” (or PPR), defined as “3rd Peak/1st Peak ratio”, is a very weak marker of an AD-mediated deficit. First, no significant differences (AD versus CTRL) were found at ages 40, 80, 150, or 200 when Male and Female animals were lumped together (Fig. 3B3). Second, no significant differences (AD versus CTRL) were found in the P200 Female cohort (Fig. 3B4).
Refractory escape
If too little time is allowed between any two depolarizing events, the later event may fall into the refractory period of the previous event and show a reduced amplitude. If AD pathology impairs the recovery of synaptically-evoked cortical responses, this would be reflected in the 4th peak/3rd peak amplitude ratio (Fig. 3C1 - 2). A simple visual inspection of voltage waveforms bared no obvious distinctions between the AD and CTRL groups (Fig. 3C2). We quantified the optical signals from 353 brain slices. For each age group, we calculated the ratio 4th/3rd Peak at the stimulation site (ROI-S), but also at the distal site, ROI-P, and found no significant differences between the AD versus CTRL (Fig. 3C3, Amplitude Ratio). Breaking down the P200 age group (ROI-S) into sexes produced mixed responses, where Males supported (asterisk) and Females rejected (asterisk) our working hypothesis (Fig. 3C4). Separation of sexes performed on the ROI-P data revealed no differences between AD and CTRL (not shown). Overall, the “refractory escape” parameter, defined as the 4th/3rd Peak ratio, did not appear to provide solid physiological distinctions between the AD and CTRL brains.
Artifact-to-peak latency (delay)
We measured the period of time for an optical signal to reach its peak. Specifically, we measured time between the synaptic pulse (stimulus artifact) and the signal’s maximum (Fig. 4A1, Delay). We hypothesized that synapses weakened by AD pathology (AD Group) are either slower to respond to stimuli or would charge postsynaptic membranes at lower rates than the healthy synapses (CTRL Group), or would be desynchronized, resulting in longer artifact-to-peak delays in the cortical population signals. Numerical analysis (Fig. 4A2) showed that the AD and CTRL conditions performed at an equal latency (Delay) across all age groups (P40, P80, P150, and P200). Separation of the P200 group into Male and Female animals, failed to resolve any AD versus CTRL differences (Fig. 4A2, Sexes Separated).
Signal Latency and Duration. A1) Synaptically evoked cortical depolarizations in L2/3. “Delay” is measured as a time (in ms) between synaptic stimulation pulse and peak of the optical signal. A2) Average “Delay” values, per age (P40 to P200) and per condition (AD versus CTRL). B) “Half-width” is measured as a duration (in ms) at signal’s half amplitude. C1) Upper: Optical traces inside experimental group P40 AD were scaled to the same height, averaged, superimposed, and compared (T-Test) to the corresponding group (P40 CTRL). Error bars, s.e.m. Number of brain slices (n) obtained from N animals. Lower: p values from unpaired T-Test (AD versus CTRL) are displayed for each time point on the voltage waveform. C2 - 4) Same as C1 except different age groups were analyzed. Dashed rectangle marks the section of the voltage waveform with statistically significant amplitude difference between AD versus CTRL. D1) Comparison between P200 AD Females versus P200 CTRL Females obtained at ROI-S. D2) Same as D1, except ROI-P was used. E1) A table used to calculate the cumulative T-Test score. A significant AD versus CTRL difference (p < 0.05) is assigned “+1” if it supports the working hypothesis (e.g., depolarizations decay faster in the CTRL group). Significant AD versus CTRL differences, which did not support the working hypothesis (e.g., depolarizations decay more slowly in the AD group) were assigned value “–1”. No statistically significant difference between AD and CTLR voltage waveforms carries a score of “0”. E2) The cumulative score (calculated in the table by adding values along the horizontal rows) is presented in the form of a bar diagram. Physiological markers suggesting dysfunctional synapses in the AD Model (positive scores) coincide with the accumulation of Thioflavin-S-positive plaques in cerebral cortex (age P150, see Fig. 1).
Duration
We hypothesize that sluggish repolarization (slower kinetics of the decay phase) is a sign of AD pathology. Slower decay will be manifested in the duration of the voltage waveform measured at half amplitude (Fig. 4B, Half-Width). These measurements showed statistically significant differences between AD versus CTRL for the older ages P200 and P150, but not for the younger age groups P80 and P40 (not shown). In addition to these half-width quantifications, we analyzed the shape of the voltage waveform by averaging optical traces. Optical traces were scaled to the same height and averaged within an experimental group (e.g., AD). Superposition of the resultant traces revealed minimal waveform discrepancies between AD and CTRL subjects at younger ages: P40 (Fig. 4C1) and P80 (Fig. 4C2). An unpaired T-Test was used to compare the corresponding time points between AD versus CTRL. Since the optical traces consisted of 273 samples (273 ms), this produced 273 p values (Fig. 4C1, bottom graph). On some instances during the course of the optical trace, the p value dropped below 0.05 (Fig. 4C1 - 2, bottom graph), but this was extremely brief (couple of ms), and was contributed to noise in optical recordings.
At older ages (P150 and P200), superposition of the averaged traces revealed larger voltage waveform discrepancies between AD and CTRL groups (Fig. 4C3 - 4). Discrepancies were predominantly found on the decaying phase of the voltage signal, with the AD group showing slower repolarization—slower return to the baseline. The unpaired T-Test identified a long string of samples (more than 100 ms long) during which the AD group was significantly less repolarized than the CTRL group. The sections of the optical traces with statistically significant AD-to-CTRL differences are marked by dashed rectangles (Fig. 4C3 - 4).
In P200 female animals, at location ROI-S, the AD-versus-CTRL discrepancies were obvious in both voltage trace display (Fig. 4D1, upper) and T-Test analysis (Fig. 4D1, lower). Identical analyses performed on the AD and CTRL waveforms obtained from a more distal region of interest, ROI-P (Fig. 4D2), were consistent with the ROI-S data (Fig. 4D1). These statistical tests are summarized using a table (Fig. 4E1). Lack of significant difference between AD and CTRL was assigned value “0”. Significant AD-to-CTRL differences were given value “+1” or “–1” depending on whether the outcome of the statistical test supported (+1) or rejected (–1) the working hypothesis. For example, “+1” supports our working hypothesis that an AD pathology impaired the decay of evoked depolarizations, resulting in a sluggish return to baseline (longer duration at half amplitude). As animals aged in our study from P40 to P200, the sluggish waveform score AD-versus-CTRL increased (Fig. 3E2). The significant slowing of the cortical depolarization decay phase coincided with the accumulation of plaques in the cerebral cortex (Fig. 3E2, Plaque Onset).
Amplitude of the evoked cortical response
To analyze signal amplitude (ΔF/F in %), the evoked optical signals (L2/3) were averaged within the experimental group (e.g., P40-AD, n = 29 brain slices from N = 9 animals). The corresponding group was P40-CTRL with n = 44 brain slices from N = 13 animals). Superposition of the results of averaging P40-AD and P40-CTRL revealed obvious AD-to-CTRL amplitude differences (Fig. 5A1, top), which were statistically very significant (Fig, 5A1, bottom). All 4 age groups were processed in this way, and at each age examined (P40, P80, P150, and P200), the amplitude differences between AD and CTRL were significant (Fig, 5A1 - 4). However, at younger age groups (P40 and P80), the evoked cortical responses were stronger in the AD group; against our working hypothesis. At the same time, in older age groups (P150 and P200), the evoked cortical responses were smaller in the AD; supporting our working hypothesis.
Amplitude of the cortical response. A1) Upper: Optical traces (synaptically evoked cortical depolarizations in L2/3) were averaged and compared (T-Test) between the AD and CTRL group. Error bars, s.e.m. Lower: p values from unpaired T-Test (AD versus CTRL) are displayed for each time point on the voltage waveforms shown above. Dashed rectangle marks the section of the voltage waveform with statistically significant amplitude difference between AD versus CTRL. A2 - 4) Same as A1 except different age groups were analyzed. B1) Comparison between P200 AD Females versus P200 CTRL Females obtained at ROI-S. Comparatively weaker synaptic responses were found in the AD group. B2) Same as B1, except ROI-P was used. C1) A table used to calculate the cumulative T-test score. A significant AD versus CTRL difference (p < 0.05) is assigned “+1” if it supports the working hypothesis (e.g., synapses are weaker in the AD group). Significant AD versus CTRL differences, which did not support the working hypothesis (e.g., synapses are stronger in the AD group) were assigned value “-1”. No statistically significant difference between AD and CTLR carries a score of “0”. C2) The cumulative score (calculated in the table by adding values along the horizontal rows) is presented in the form of a bar diagram. Physiological markers suggesting dysfunctional synapses in the AD Model (positive scores) coincide with the accumulation of Thioflavin-S-positive plaques in cerebral cortex (age P150). However, before the onset of plaques (ages P40 and P80), significantly stronger synaptic responses were detected in the AD group.
Segregation of the P200 data into the age- and sex-matched experimental groups (Fig. 5B1 - 2) did not change the conclusions previously reached with the experimental groups made of both sexes (Fig. 5A4). The outcomes of these statistical testing (AD versus CTRL) were summarized in the form of a table (Fig. 5C1). Lack of difference between the matching AD and CTRL groups has been assigned a value “0”. Significant AD-to-CTRL differences were assigned either “+1” or “–1”, depending on the support (+) or rejection (–) of the working hypothesis: “cortical AD pathology weakens cortical synapses; hence synaptic responses are smaller in the AD condition”.
As animals aged in our study from P40 to P200, the amplitude “score” increased (Fig. 5C2, Amplitude –Cumulative Score Graph). At two older ages (P150 and P200), cortical responses in the AD group were notably weaker than in the CTRL group, but at young ages these scores were decisively negative (Fig. 5C2). At young ages too, the AD process changes cortical physiology –the cortical responses become stronger than normal (i.e., hyperexcitation). Overall, our data revealed that the AD condition causes bidirectional changes in cortical physiological responses, stronger than normal and weaker than normal, with the switch in direction occurring at the age when Aβ plaques begin to accumulate in cerebral cortex (Fig. 5C2, Plaque Onset).
DISCUSSION
Subthreshold depolarizations
Voltage imaging signals (Figs. 2E 3A) mostly reflect subthreshold synaptic potentials [25, 26], just like the local field potential, LFP [27, 28]. This is in stark contrast to the calcium imaging methods, which predominantly detect neuronal spiking (action potentials).
Population imaging
Population imaging studies (e.g., wide-field imaging) lack single-cell resolution. The optical signal represents a “mean” response of many neurons, many dendrites and axons [29–33]. In contrast to LFP, which senses extracellular voltages and flips signal’s polarity if a depolarization wave travels under the electrode, the GEVI method senses genuine trans-membrane voltages, in which membrane depolarization is always with positive polarity, and membrane hyperpolarization is always negative [19, 26].
Transgenic mouse line
Neuronal labeling with GEVIs is not trivial [34]. Typically, labeling is achieved by injecting brains with viral vectors (AAV). Alternatively, one can use transgenic animal lines. The transgenic strategy, is more robust (repeatable), more uniform (wide cortical areas are uniformly labeled), more humane (it is not necessary to drill holes in the animal’s head), and less labor intensive than the AAV strategy, but we are seeing new developments with the noninvasive AAV approaches [35]. Transgenic mice provide a valuable resource for AD research, both for understanding disease mechanisms and for testing potential therapies [36–38]. In the APP/PS1 mouse model (used in the current study to generate the AD-GEVI mouse line), human Aβ42 is preferentially generated over Aβ40, but levels of both increase with age. Strong and cortex-wide expression of GEVI in cortical pyramidal neurons (Fig. 1CD) did not jeopardize the fundamental aspects of the AD mouse model: age-dependent accumulation of amyloid plaques in the neocortex, onset between 80–150 days postnatal, with a slightly earlier onset in females compared to males [39–42].
The new AD-GEVI mouse line allows population voltage imaging ex vivo—in brain slice preparations (current study). An identical GEVI expression pattern, as reported in Fig. 1, has already been successfully used for studying sensory-evoked cortical voltage transients in vivo [43], as well as a large-scale spontaneous activity in vivo [44].
GEVI expression around amyloid plaques
The small-diameter smooth-surface Aβ plaques appeared as “foreign bodies” embedded into the GEVI-labeled cortical parenchyma, with no obvious interaction with the pyramidal neuropil (dendrites and axons, Fig. 1D). However, large-diameter plaques that projected their amyloid spikes into the surrounding cortical parenchyma did cause disturbances in the cortical microarchitecture, here dubbed “swirls” (Fig. 1E, white arrows).
GEVI expression does not diminish with aging
In the AD translational research, almost all aspects of AD (histological, behavioral, and physiological) are manifested only in older animals [40–42]. Therefore, one important prerequisite for using the new mouse line in AD research is that the expression of GEVI is stable with age. We found no evidence that cortical GEVI expression diminished in older animals. Contrary to our expectation, we found that the RFI was slightly lower in the youngest animals, age group P40 (Fig. 2C). This is most likely caused by two factors: 1) At age P40, neurons undergo experience mediated changes in wiring. Rapid growth of new membranes constituting dendritic spines and axon branches is not immediately accompanied by trafficking and insertion of the GEVI molecules. 2) In older animals (P80, P150, and P200), the packing of the cortical neuropil (dendrites & axons) may be tighter, resulting in a greater amount of the GEVI-expressing membrane per unit of volume.
Amplitude
Animals carrying the APP/PS1 mutation and accumulating plaques (AD Group) had statistically significant reduction in the optical signal size compared to the age-matched CTRL Group (healthy littermates). The physiological evidence of weaker cortical responses in the AD group emerge at age 150 days (Fig. 5A3); the same age group in which a significant accumulation of amyloid plaques begins (Fig. 1AB). The amplitude reduction of evoked depolarizations in the cortical tissue infiltrated with plaques (P150 and P200) is not surprising [21]. The surprising finding is that at younger ages (P40 and P80) the AD amplitudes are significantly higher than the matching controls (healthy littermates), indicating hyperexcitability caused by the developing AD pathology. In the same cortical circuit (somatosensory cortex, layer 2/3), which at older ages exhibits plaques and significant decline in cortical responses, at the younger ages we found abnormally strong cortical responses (Fig. 5C).
Does early-age hyperexcitability lead to the old-age plaque-mediated synaptic dysfunction? Overt neuronal activity has been linked to the enhanced plaque deposition [45]. Areas of the human brain that develop the most Aβ plaques also have the highest basal rates of metabolic and neural activity, as measured by PET and fMRI, when an individual is not performing a specific mental task, the so-called “default state”—discussed in [46].
Duration
At older age groups (P150 and P200), we found significantly longer voltage transient half-widths in AD versus CTRL (Fig. 4). This may be related to the slower deactivation of extrasynaptic NMDAR-mediated currents in pyramidal cells [9]. Impairment of voltage-gated K+ currents (by an AD pathology) can also produce slower return to base line.
Limitations of the current study
No obvious changes in voltage waveforms
One would like to detect AD-induced synaptic dysfunction by visual examination of synaptically-evoked voltage waveforms. Ideally, after just a few experimental trials, the experimenter would be able to tell if the brain slices belonged to an AD animal infiltrated with plaques, or to a Control animal free of plaques. Clearly, this was not the case in the current study; the reported physiological differences between AD and CTRL are subtle [8, 47] and emerge from numerical analysis performed on a large number of experiments (Figs. 4 5).
Poor SNR
A relatively weak signal-to-noise ratio of the voltage optical signal (compared to the SNR in calcium optical signal, for example) is of concern [19]. Sharper distinctions between experimental groups (AD versus Control) could potentially be achieved by using new-generation voltage indicators with improved SNR [14, 15], or additional stimulation paradigms that are capable of: (a) exploring a wider range of stimulation frequencies, (b) delivering stimuli in different cortical layers, and (c) applying LTP induction protocols, typically used to evaluate physiological deficits in the AD model animals [10, 11].
Summary
In the past, measurements of the AD-mediated physiological changes were largely restricted to the hippocampus area, and they relied on the LTP paradigm [40, 48–50]. We developed a new animal line (AD-GEVI) which allows for consistent ex vivo (acute brain slice) measurements of evoked physiological responses in the cerebral cortex. The AD-GEVI animals carry mutant APP/PS1 genes and accumulate Aβ plaques in the neocortex. The expression of GEVI is stable with animal age (247 days examined). The formation of amyloid plaques does not diminish the GEVI expression in the membranes surrounding plaques. Apparent weakening of the evoked cortical depolarizations at older ages (P150 and P200) may potentially be related to the phenomenon of Synaptic Failure [1–5, 52]. Intriguingly, prior to formation of amyloid plaques, at ages P40 and P80, the same synaptically evoked cortical responses appear to be somewhat stronger in the AD group compared to their healthy littermates (CTRL group); this may potentially be related to the phenomenon of Hyperexcitability or Hyperactivity [7, 53–55]. Future studies may also investigate if pharmacological manipulations of neuronal excitability at younger ages alleviate the amyloid pathology and weakness in synaptic strength at older ages. Furthermore, the new AD-GEVI model would allow one to investigate the impact of chronically administered candidate therapies (e.g., memantine) on the evoked cortical depolarizations.
Footnotes
ACKNOWLEDGMENTS
We are thankful to Hannah Jacob (South Windsor, CT) for help with data analysis; to Thomas Knopfel’s laboratory (Imperial College London) for donating PAN-GEVI transgenic animals; to Center for Mouse Genome Modification (Siu-Pok Yee, UConn Health), Center for Comparative Medicine (Ramaswamy M. Chidambaram, UConn Health) and Molecular Core (John Glynn & Kevin Claffey, UConn Health) for assistance with animal re-derivation, breeding, and genotyping; and to RedShirtImaging (Chun Bleau and Charlie Bleau) for assistance with optical recording equipment and software.
This research was funded by the Cure Alzheimer’s Fund award to SDA and RY; the National Institute on Aging grant AG064554 to SDA and RY; and the UConn Alcohol Research Center (ARC) / Kasowitz Medical Research Fund grant P50AA027055 to SDA.
