Optogenetic activation of parvalbumin and somatostatin interneurons selectively restores theta-nested gamma oscillations and oscillation-induced spike timing-dependent long-term potentiation impaired by amyloid β oligomers
Abnormal accumulation of amyloid β1–42 oligomers (AβO1–42), a hallmark of Alzheimer’s disease, impairs hippocampal theta-nested gamma oscillations and long-term potentiation (LTP) that are believed to underlie learning and memory. Parvalbumin-positive (PV) and somatostatin-positive (SST) interneurons are critically involved in theta-nested gamma oscillogenesis and LTP induction. However, how AβO1–42 affects PV and SST interneuron circuits is unclear. Through optogenetic manipulation of PV and SST interneurons and computational modeling of the hippocampal neural circuits, we dissected the contributions of PV and SST interneuron circuit dysfunctions on AβO1–42-induced impairments of hippocampal theta-nested gamma oscillations and oscillation-induced LTP.
Targeted whole-cell patch-clamp recordings and optogenetic manipulations of PV and SST interneurons during in vivo-like, optogenetically induced theta-nested gamma oscillations in vitro revealed that AβO1–42 causes synapse-specific dysfunction in PV and SST interneurons. AβO1–42 selectively disrupted CA1 pyramidal cells (PC)-to-PV interneuron and PV-to-PC synapses to impair theta-nested gamma oscillogenesis. In contrast, while having no effect on PC-to-SST or SST-to-PC synapses, AβO1–42 selectively disrupted SST interneuron-mediated disinhibition to CA1 PC to impair theta-nested gamma oscillation-induced spike timing-dependent LTP (tLTP). Such AβO1–42-induced impairments of gamma oscillogenesis and oscillation-induced tLTP were fully restored by optogenetic activation of PV and SST interneurons, respectively, further supporting synapse-specific dysfunctions in PV and SST interneurons. Finally, computational modeling of hippocampal neural circuits including CA1 PC, PV, and SST interneurons confirmed the experimental observations and further revealed distinct functional roles of PV and SST interneurons in theta-nested gamma oscillations and tLTP induction.
Our results reveal that AβO1–42 causes synapse-specific dysfunctions in PV and SST interneurons and that optogenetic modulations of these interneurons present potential therapeutic targets for restoring hippocampal network oscillations and synaptic plasticity impairments in Alzheimer’s disease.
KeywordsAlzheimer’s disease Amyloid beta oligomers Hippocampus Optogenetics Parvalbumin interneuron Somatostatin interneuron Theta-nested gamma oscillations Spike timing-dependent long-term potentiation Synapse-specific dysfunction
Alzheimer’s disease is a neurodegenerative disease characterized by a progressive decline in cognitive and mnemonic functions [1, 2]. Abnormal accumulation of amyloid β1–42 oligomers (AβO1–42) is a hallmark of Alzheimer’s disease [1, 2, 3, 4] and AβO1–42-induced impairments of gamma oscillations [5, 6, 7, 8, 9, 10] and long-term synaptic plasticity [3, 4, 11, 12] are believed to contribute to the memory deficits observed in Alzheimer’s disease. In particular, hippocampal theta-nested gamma oscillations observed during spatial memory processing [13, 14, 15] have been shown to support the induction of long-term potentiation (LTP) [16, 17, 18, 19]. Thus, AβO1–42 may impair memory by disrupting GABAergic inhibitory circuits, which underlie oscillogenesis [14, 20, 21, 22, 23, 24, 25]. Indeed, there is now increasing experimental evidence showing that AβO1–42 reduces GABA synaptic transmission [26, 27, 28], causes excitation/inhibition imbalances [9, 12, 27, 28], and even diminishes the number of GABAergic synapses/terminals onto pyramidal cells . Also, parvalbumin-positive (PV) and somatostatin-positive (SST) interneurons, the two major subtypes of hippocampal interneurons  that are critically involved in oscillogenesis [24, 25, 31], are reported to be impaired in mouse models of Alzheimer’s disease [5, 6, 7, 8, 27, 32, 33]. PV interneurons’ spike amplitude, membrane potential, and firing rate are decreased [5, 7] while SST interneurons’ structural plasticity and axonal sprouting are impaired in Alzheimer’s disease mouse models [27, 32]. Surprisingly, the neural circuit mechanism by which dysfunction of PV and SST interneurons contributes to AβO1–42-induced impairment of oscillogenesis and LTP is unclear. If uncovered, it could help researchers find novel therapeutic targets for Alzheimer’s disease. Recently, optogenetic stimulation of channelrhodopsin-2 (ChR2)-expressing hippocampal CA1 pyramidal cells (PCs) at theta-frequency was shown to induce in vivo-like theta-nested gamma oscillations in the CA1 area of acute hippocampal slices in vitro . This provides a novel model in which to perform targeted whole-cell patch-clamp recordings and selective optogenetic modulation of PV or SST interneuron activity during optogenetically induced theta-nested gamma oscillations and LTP induction. We have used this approach to investigate neural circuit dysfunction in hippocampal slices treated with AβO1–42. We found that AβO1–42 caused selective dysfunctions in reciprocal synapses between PC and PV interneurons, which impaired gamma oscillations and desynchronized the spike phases of PC and PV interneurons relative to gamma oscillations. While AβO1–42 had no effect on PC-to-SST or SST-to-PC synapses, it specifically disrupted SST interneuron-mediated disinhibition to PC resulting in the impairment of theta-nested gamma oscillation-induced spike timing-dependent LTP (tLTP). Selective optogenetic activation of PV interneurons restored gamma oscillations while selective optogenetic activation of SST interneurons restored theta-nested gamma oscillation-induced tLTP. These results demonstrate that AβO1–42-induced synapse-specific dysfunctions in PV and SST interneurons may explain the concomitant impairments of hippocampal gamma oscillations and synaptic plasticity in Alzheimer’s disease. Moreover, using a computational network model of PC, PV, and SST interneurons, we further demonstrate that PV and SST interneurons targeting different compartments of the CA1 PC have distinct functional roles in oscillogenesis and tLTP induction.
AβO1–42 impairs in vivo-like, optogenetically induced theta-nested gamma oscillations in hippocampal slices
AβO1–42 causes synapse-specific dysfunction of PC-to-PV, but not PC-to-SST synapses
Since the spiking of hippocampal CA1 interneurons is in large part driven by CA1 PC’s excitatory inputs to the interneurons , we investigated whether the treatment of AβO1–42 affected CA1 PC’s excitatory inputs to PV and SST interneurons. We performed voltage-clamp recordings in eYFP-expressing PV or SST interneurons during blue light-induced theta-nested gamma oscillations in DMSO-treated and AβO1–42-treated slices (Fig. 2f). We found that the amplitude of CA1 PC’s excitatory postsynaptic current (EPSC) to PV, but not SST interneuron, was significantly decreased in AβO1–42-treated slices (Fig. 2f, g), while EPSC frequency was unaffected (Fig. 2h). To characterize the AβO1–42-induced synaptic dysfunctions at CA1 PC-to-PV synapse and CA1 PC-to-SST synapse, we first investigated how AβO1–42 affected the stimulus-response (S-R) curve of these synapses by electrically stimulating the axons of CA1 PC in the alveus of CA1 at different intensities (10, 50, 100, 150, 200, and 300 μA) and recording the corresponding PC-evoked EPSCs in eYFP-expressing PV interneuron (Fig. 2i, j) or in eYFP-expressing SST interneuron (Fig. 2m, n). Analysis of the S-R curve revealed that, for each stimulation intensity, AβO1–42 significantly increased the amplitudes of PC-evoked EPSCs in PV (Fig. 2j, right), but not those in SST interneurons (Fig. 2n, right). These results indicate that AβO1–42 increases the initial neurotransmitter release probability of PC-to-PV synapse. To investigate the synaptic locus of EPSC changes, we stimulated the CA1 PC axons using a half-maximal stimulus (based on the S-R curve in Fig. 2j, n, right; 115–210 μA) and an inter-stimulus interval of 20 ms (50 Hz, 10 stimulus) for the analysis of paired-pulse ratio (PPR), total charge, and short-term plasticity of PC-evoked EPSCs in PV (Fig. 2k, l) and SST interneurons (Fig. 2o, p). Paired-pulse facilitation of PC-evoked EPSCs in PV interneurons, as observed in DMSO-treated slices, was converted to paired-pulse depression in AβO1–42-treated slices (Fig. 2k, right). The total charge of PC-evoked EPSCs in PV (Fig. 2l, left), analyzed by the area of the PC-evoked EPSCs in Fig. 2k (left), was significantly decreased by AβO1–42. Furthermore, short-term facilitation of PC-evoked EPSCs in PV interneurons, as observed in DMSO-treated slices, was converted to short-term depression in AβO1–42-treated slices (Fig. 2l, right). These results indicate that AβO1–42 causes presynaptic depression at PC-to-PV synapse, which led to a decrease in CA1 PC-evoked excitatory synaptic inputs onto PV interneurons. Thus, AβO1–42-induced gamma oscillation impairment may be due to dysfunction of presynaptic mechanisms at PC-to-PV synapses. In contrast, AβO1–42 had no effect on PPR, total charge, or short-term plasticity of CA1 PC-evoked EPSCs in SST interneurons (Fig. 2o, p). Therefore, AβO1–42 causes presynaptic dysfunctions at CA1 PC-to-interneuron synapses which is target-specific.
AβO1–42 causes synapse-specific dysfunction of PV-to-PC synapses, but not SST-to-PC synapses
Optogenetic activation of PV interneurons restores AβO1–42-induced impairment of theta-nested gamma oscillations
Optogenetic activation of SST interneurons restores AβO1–42-induced impairment of theta-nested gamma oscillation-induced tLTP
AβO1–42 causes selective dysfunction of SST interneuron-mediated disinhibition to CA1 PC
Distinct functional roles of PV and SST interneurons in gamma oscillogenesis and theta-nested gamma oscillation-induced tLTP
Here we have provided the first experimental evidence on how AβO1–42 causes synapse-specific dysfunction in hippocampal inhibitory circuits to impair theta-nested gamma oscillations and theta-nested gamma oscillation-induced tLTP. AβO1–42 selectively disrupted reciprocal PC-to-PV and PV-to-PC synapses, which decreased the peak power of theta-nested gamma oscillations and desynchronized the phase of spikes and synaptic currents relative to gamma cycles (Fig. 1, 2, 3, 4). In contrast, AβO1–42 had no effect on either PC-to-SST synapse or SST-to-PC synapses, but it did selectively disrupt SST interneuron-mediated disinhibition to block NMDAR-mediated tLTP at CA3-to-CA1 synapses induced by theta-nested gamma oscillation-like stimulation (Figs. 5 and 6). Importantly, optical stimulation of PV and SST interneurons selectively restored theta-nested gamma oscillations and oscillation-induced tLTP, respectively, which strongly supports the conclusion that these phenomena were the result of synapse-specific dysfunctions of PV and SST interneurons induced by AβO1–42.
Based on our in vitro experimental observations, we built a computational network model of CA1 PC, PV, and SST interneurons which allowed us to infer possible reasons for why hippocampal oscillations are conducive to LTP in a healthy brain [16, 17, 18, 19]. From our simulation results, we were able to see how perisomatic-targeting PV interneurons entrain both CA1 PC and SST interneurons at gamma-frequency which allowed for the SST interneuron to disinhibit CA3 input-activated feedforward inhibition onto CA1 PCs’ proximal dendrites, creating a time window for tLTP induction (Fig. 7). Thus, PV and SST interneurons have distinct functional roles in the induction of synaptic plasticity in different compartments of the CA1 PC, and the accumulation of AβO1–42 seen in Alzheimer’s disease may cause memory deficits due to impairment of these synaptic plasticity mechanisms.
Although all of our experiments are conducted in vitro, the gamma oscillation impairment observed in our study shares many similarities with the effects of Aβ on kainate-induced gamma oscillations in vitro  as well as gamma oscillations recorded in vivo in mouse models of Alzheimer’s disease [5, 6, 7, 8]. Also, our finding that optical stimulation of PV interneurons can restore gamma oscillations is consistent with previous results showing that manipulations of PV interneurons [5, 8] or PV-like fast-spiking interneurons were able to restore gamma oscillations in Alzheimer’s disease mouse models in vivo . However, unlike previous studies using animal models with the late phase of Alzheimer’s disease [5, 7, 8], the acute effects of AβO1–42 that we uncovered here may only account for the early phase of Alzheimer’s disease. In Alzheimer’s disease mouse models such as APP/PS1 mice  and hAPPJ20 mice , spike firing rates and membrane potentials of PV interneuron are increased while in early phase of Alzheimer’s disease, pathological effects of AβO1–42 are mainly limited to synaptic dysfunctions with the intrinsic neuronal properties are spared , which is consistent with our results (Figs. 2 and 3 and Additional file 4: Figure S4). Thus, optogenetic activation of PV interneurons could have restored theta-nested gamma oscillations by directly depolarizing PV interneurons, which in turn compensate for the AβO1–42-induced reduced PV interneuron-evoked EPSCs to CA1 PC (Fig. 2) to resynchronize CA1 PC spikes during theta-nested gamma oscillations (Fig. 4), consequently leading to the restoration of theta-nested gamma oscillations. In addition to the reduction in gamma oscillation power, epileptic hyper-synchronous activities are widely observed in human patients with Alzheimer’s disease [6, 42] and in genetically modified Alzheimer’s disease mouse models [5, 6, 27, 43, 44]. Since the occurrence of epileptic activities in Alzheimer’s disease mouse models requires the abnormal aggregation of Aβ fibrils  and tau protein , but not AβO1–42 , it may be that hyper-synchrony may develop with Alzheimer’s disease progression [6, 45]. In fact, it is well established that AβO1–42 causes hyperexcitability in excitatory neurons . Also, the increase in EPSC and decrease in IPSC amplitudes in CA1 PC during kainate-induced gamma oscillations under AβO1–42 pathology was observed in vitro . Thus, it may be that the balance between excitation and inhibition is disrupted in Alzheimer’s disease but how the same neural circuit alternates between hypo- and hyper-synchrony requires further investigation.
Although many studies manipulated PV interneurons in Alzheimer’s disease studies [5, 7, 8], our study is the first to directly show how manipulation of SST interneurons could alleviate Alzheimer’s disease-related dysfunctions. In contrast to many studies targeting dysfunctional excitatory synapses [46, 47, 48, 49] or LTP induction-related intracellular cascades in order to restore LTP in Alzheimer’s disease mouse models [49, 50, 51], we show that reinstating SST interneuron-mediated disinhibition  is sufficient for restoring tLTP in AβO1–42-treated slices in vitro (Figs. 5 and 6). In fact, SST interneuron-mediated disinhibition unmasks the back-propagating spike required for the induction of tLTP [52, 53]. Thus, our results suggest that SST interneurons’ neural circuit dysfunction could explain the tLTP impairment caused by acute application of AβO1–42 resembling early stages of Alzheimer’s disease, further supported by our in silico hippocampal network simulation (Fig. 7, Additional file 12: Figure S12). Although we did not get to identify the interneuron subtype that provides disinhibition to CA1 PC through SST interneuron activation, CCK-positive interneurons such as Schaffer collateral-associated cells [54, 55, 56] or bistratified cells  that are located in the stratum radiatum could be potential candidates. Thus, identifying the interneuron subtypes involved in disinhibition could help target the disinhibitory synapse that is impaired by AβO1–42 pathology. A recent study reported that optogenetic activation of OLM interneurons can induce type 2 theta oscillations in vivo , indicating that SST interneurons may also contribute to the generation of theta oscillations in addition to providing disinhibition to CA1 PC in vivo. Since we optically stimulated theta oscillations in order to induce gamma oscillations in vitro, our data cannot resolve the individual contribution of PV or SST interneurons on theta oscillation impairment in Alzheimer’s disease [57, 58]. Moreover, it is possible that theta-nested gamma oscillations could play a role in the induction of synaptic plasticity in interneurons ; thus, the neural circuit mechanism linking theta-nested gamma oscillations and tLTP may be more intricate than suggested in the present study (Fig. 7). Interestingly, a recent study reported re-emergence of LTP in aged Tg2576 Alzheimer’s disease mice which correlates with a decrease in PV interneuron number . Thus, the specific manner in which PV and SST interneurons are affected as the pathologies of Alzheimer’s disease progress with age in vivo to disrupt synaptic plasticity requires further investigation. Nonetheless, our data suggests that targeted manipulation of interneuron populations in the hippocampus may be a promising approach for treatments of early-stage Alzheimer’s disease.
Although the optogenetic manipulation technique we adopted in this study targeted CA1 PV and SST interneurons, in CA1 alone, there are more than 20 interneuron subtypes [61, 62] and PV and SST interneurons do not relate to specific interneuron types, nor indeed are these two markers entirely non-overlapping in CA1 [63, 64, 65, 66, 67, 68]. PV can be expressed in both axo-axonic and fast-spiking interneurons, and SST can be found not only in oriens lacunosum-moleculare interneurons, but in various long-range projecting interneurons, too. Indeed, bistratified cells (found in stratum oriens) express both PV and SST [54, 69, 70, 71]. Therefore, care is warranted in interpreting our results.
In summary, by optogenetically manipulating PV and SST interneurons, here we showed for the first time that AβO1–42 causes synapse-specific dysfunctions in PV and SST interneurons’ synapses, which allows us to uncover how AβO1–42 causes concomitant impairments of hippocampal theta-nested gamma oscillations and oscillation-induced tLTP at CA3-to-CA1 synapses. Thus, our findings provide crucial insight that will help guide future studies aimed at identifying the molecular target that gives rise to AβO1–42-induced synapse-specific dysfunctions, potentially leading to novel therapeutic targets for Alzheimer’s disease.
Three different lines of mice, C57BL/6 mice, PV-Cre knock-in mice (C57BL/6 background, Jackson Laboratory, stock #017320), and SST-IRES-Cre (C57BL/6 background, Jackson Laboratory, stock #013044) knock-in mice (4–11 weeks old) were used . All animals were kept in 12:12-h light-dark cycles with food and water available ad libitum. All animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee of Korea University (KUIACUC-2017-112).
AAV particles were purchased from the UNC Vector Core. To express ChR2  selectively in CA1 PC, AAV5-CaMKII-hChR2(E123T/T159C)-p2A-mCherry-WPRE (3.8 × 1012 virus molecules/ml, 1 μl) was injected in all three different lines of mice bilaterally into the hippocampus. For the selective expression of eYFP, Arch, ChR2, or C1V1 on PV or SST interneurons, AAV2-EF1a-DIO-EYFP (4.6 × 1012 virus molecules/ml, 1 μl), AAV5-EF1a-DIO-eArch3.0-EYFP (5 × 1012 virus molecules/ml, 1 μl), AAV5-EF1a-DIO-hChR2(E123T/T159C)-p2A-mCherry-WPRE (3.8 × 1012 virus molecules/ml, 1 μl), or AAV2-EF1a-DIO-C1V1(E162T)-TS-p2A-EYFP-WPRE (3 × 1012 virus molecules/ml, 1 μl) were injected bilaterally into the hippocampus of in PV-Cre or SST-Cre mice.
Stereotaxic virus injections
Mice were deeply anesthetized under 2% isoflurane (2 ml/min flow rate) and head-fixed into a stereotaxic frame (Stoelting Co.). Craniotomies were made bilaterally to target CA1 area of the hippocampus for viral injections (from bregma: anteroposterior − 2.70 mm, lateral ± 2.50 mm, and dorsoventral − 1.75 mm or anteroposterior − 2.56 mm, lateral ± 2.6 mm, and dorsoventral − 1.85 mm). One microliter of each virus suspension was injected into the CA1 area of the hippocampus at a rate of 0.15 μl/min through a Hamilton syringe using a motorized stereotaxic injector (Stoetling Co.). The syringe was left in the brain for more than 5 min to allow for virus diffusion. The scalp was sutured and disinfected with antibiotic, after which the mice were returned to their home cage for recovery for at least 14 days.
Preparation and treatment of AβO1–42 to hippocampal slices
Soluble AβO1–42 was prepared following methods in Lambert et al.  with a slight modification . Aβ1–42 or Aβ42–1 powder (Bachem) was dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, Sigma Aldrich) for monomerization at a final concentration of 1 mM and incubated for 90 min. HFIP was evaporated under vacuum condition (SpeedVac). The remaining thin and clear film of Aβ1–42 or Aβ42–1 was dissolved in dimethyl sulfoxide (DMSO, Sigma Aldrich) to make 5 mM Aβ1–42 or Aβ42–1 stock, which was aliquoted and frozen at − 20 °C. The Aβ1–42 or Aβ42–1 stock was thawed and diluted to 100 μM in artificial cerebrospinal fluid (aCSF, containing (in mM): 126 NaCl, 3 KCl, 1.25 NaH2PO4, 2 MgSO4, 2 CaCl2, 25 NaHCO3, and 10 glucose at pH 7.2–7.4 bubbled with 95% O2/5% CO2). After dilution, Aβ1–42 or Aβ42–1 solution was incubated for 18 h at 4 °C for Aβ oligomerization. Before the recording, 2% DMSO (vehicle) and 100 μM AβO1–42 or AβO42–1 were treated into hippocampal slices in 31.2 ml of aCSF for 20 min by diluting it to a final concentration of 200 nM AβO1–42 or AβO42–1 in 0.004% DMSO for each condition.
Western blot analysis
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)
AβO1–42 were prepared as described above and resolved on a nonreducing 4–15% tris-glycine–SDS-PAGE gels with LDS sample buffers . The gel was transferred on to a 0.2-μm PVDF membrane (Bio-Rad) according to the manufacturer’s recommendation. Membranes were blocked in 5% bovine serum albumin (BSA) in tris-buffered saline containing 0.01% Tween 20 for 1 h at room temperature. Blots were incubated in the primary antibody mOC64 (rabbit monoclonal against amino acid residues 3–6 of Aβ; Cat# ab201060, Lot# GR3235744-4, RRID: AB_2818982, Abcam)  at 1:200 dilution overnight at 4 °C. Immunoreactivity was detected with enhanced chemiluminescence (Bio-Rad) and imaged using Fluorchem E system (ProteinSimple). Molecular weight values were estimated using Precision Plus Protein™ Dual Color Standards (Bio-rad).
AβO sample was diluted with native PAGE sample buffer (Bio-rad) and then subjected to native PAGE using a 4–15% tris-glycine gel with the tris-glycine running buffer (Bio-rad). Following transfer to PVDF membrane, membranes were blocked in 5% BSA in Tris-buffered saline containing 0.01% Tween 20 for 1 h at room temperature. Blots were probed using rabbit monoclonal Aβ antibody (mOC64, 1:200, Cat# ab201060, Lot# GR3235744-4, RRID: AB_2818982, Abcam) overnight at 4 °C. Immunoreactivity and imaging were performed as described above.
In vitro hippocampal slice preparation
Mice were deeply anesthetized using 1.25% Avertin solution (8 g of 2, 2, 2-Tribromoethanol and 5.1 ml of 2-methyl-2-butanol in 402.9 ml saline, Sigma Aldrich) at a delivery rate of 0.2 ml/10 g body weight and perfused with ice-cold cutting solution (containing (in mM): 180 sucrose, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 11 glucose, 2 MgSO4, and 1 CaCl2 at pH 7.2–7.4 oxygenated with 95% O2/5% CO2). Either coronal or horizontal hippocampal slices (300–400 μm) were cut using a vibratome (VT 1000 S, Leica Microsystems). Slices were allowed to recover for 20 min in a mixture of cutting solution and aCSF solution at 1:1 ratio, after which the slices were further incubated in aCSF for at least 1 h at 30–32 °C before performing electrophysiological recordings. To compare between DMSO and AβO1–42 conditions in the same slice (Fig. 1, Fig. 4c–i), hippocampal slice was first treated with 2% DMSO in aCSF for 20 min and then the same hippocampal slice was treated with 100 μM AβO1–42 or AβO42–1 in aCSF by diluting to a final concentration of 200 nM for 20 min. In all other experiments (Figs. 2, 3, 5, and 6 and Additional file 3: Figure S3, Additional file 4: Figure S4, and Additional file 11: Figure S11), hippocampal slices were treated with either 2% DMSO or 100 μM AβO1–42 or AβO42–1 in aCSF by diluting to a final concentration of 200 nM for 20 min before performing electrophysiological recordings.
In vitro field and patch-clamp recordings
Light-induced theta-nested gamma oscillations and gamma phase analysis
For the induction of theta-nested gamma oscillations, ChR2-expressing PCs were activated by sinusoidal (5 Hz) blue light (470 nm)  (Fig. 1, 2, 3, and 4 and Additional file 2: Figure S2, Additional file 3: Figure S3, Additional file 5: Figure S5, Additional file 6: Figure S6, and Additional file 7: Figure S7). Blue light was delivered using a digital micromirror device (DMD, Polygon400, Mightex) through the objective (× 40) of the microscope (BX51W, Olympus), which covered the 550-μm diameter circle of the CA1 area with the center of the illumination positioned at the field electrode. The intensity of the blue light varied between 0 to a maximum intensity of 15 mW, which was controlled using a custom-made Arduino-based controller. Igor Pro was used to control DMD and synchronize optical stimulation with the electrophysiological recordings. LFP data were first down-sampled to 1 kHz and band-pass filtered between 20 and 120 Hz for gamma oscillations. Welch’s power spectral densities (PSD) of gamma oscillations (3 repetitions of 1-s theta-nested gamma oscillations) were analyzed to quantify the peak power and peak frequency (Figs. 1h–j and 4e–g and Additional file 2: Figure S2, Additional file 3: Figure S3, Additional file 5: Figure S5, Additional file 6: Figure S6, and Additional file 7: Figure S7). Spectrogram of gamma oscillations was generated using short-time Fourier transform with window size = 100 ms and step size = 1 ms. Phase histogram (Fig. 4k) of spike or PSC was generated by calculating the instantaneous phase of spikes or PSCs using the Hilbert transform of simultaneously recorded gamma oscillations. The zero phase of gamma oscillations was defined as the peak of the gamma cycle. Probability of spike or PSCs as a function of the phase of reference gamma oscillations was obtained using 20 bins. Resultant vectors were calculated from the phase histogram and plotted in the polar plot (Fig. 4l) from which vector length (Fig. 4m) and vector phase (Fig. 4n) were calculated. Mean value and statistical significance of vector phase were calculated using the Circular Statistics Toolbox in MATLAB (R2018a) . To generate phase-amplitude comodulograms of theta-nested gamma oscillations (Figs. 1k and 4h and Additional file 3: Figure S3, Additional file 5: Figure S5, and Additional file 6: Figure S6), theta phase was calculated using Hilbert transformation and binned into 20 phase bins with 18° intervals. At each theta bin, the power spectrogram of gamma oscillations was calculated using short-time Fourier transform. The zero phase of theta oscillations was defined as the peak of the theta cycle. To analyze the phase-amplitude coupling strength of theta-nested gamma oscillations (Figs. 1l, 4i, Additional file 3: Figure S3, Additional file 5: Figure S5 and Additional file 6: Figure S6), we calculated the modulation index which is defined as the normalized Kullback-Leibler distance between probability distribution of gamma amplitude per each theta phase bin (18 bins with 20° intervals) and uniform distribution . To obtain the probability distribution of gamma amplitude, mean amplitude of gamma oscillations for each bin was normalized by the sum of gamma amplitude of total bins. Modulation index value of 0 indicates the absence of phase-amplitude coupling, and the higher modulation index value indicates the stronger phase-amplitude coupling.
Optical modulation of opsin-expressing PV and SST interneurons during patch-clamp recordings
We expressed Arch or C1V1 in PV and SST interneurons and ChR2 in PC in the same hippocampal slice to optically inactivate (Fig. 3b–e, Additional file 5: Figure S5, and Additional file 6: Figure S6) or activate (Fig. 4a–d) interneurons during theta-nested gamma oscillations, respectively. The optimal wavelength for stimulating Arch is a green-colored 565-nm light. However, since 565-nm green light also induced excitatory synaptic currents by activating ChR2-expressing PCs (Additional file 7: Figure S7b, d) as well as inducing gamma oscillations in the LFP (Additional file 7: Figure S7b, e) while 590-nm yellow light had no direct effect on ChR2-expressing PC (Additional file 7: Figure S7c, d), we used 590-nm yellow light in activating both Arch- and C1V1-expressing interneurons during blue light-induced theta-nested gamma oscillations. The effectiveness of 590-nm yellow light on Arch-expressing PV and SST interneurons was tested by performing whole-cell voltage-clamp recordings in PV-Cre or SST-Cre mice, respectively (Additional file 8: Figure S8). For the inactivation of Arch-expressing interneurons during theta-nested gamma oscillations (Fig. 3d, e, Additional file 5: Figure S6, and Additional file 6: Figure S6), a tonic yellow light of a fixed light intensity (1 s, 3 mW) was delivered using the DMD. For the activation of C1V1-expressing PV interneuron during theta-nested gamma oscillations (Fig. 4c, d), a sinusoidal (5 Hz) yellow light (590 nm) was delivered through DMD with the intensity of light sinusoidally varied between 0 and 3 mW using a custom-made Arduino-based controller. To record IPSC evoked by PV and SST interneurons in CA1 PC, ChR2-expressing PV and SST interneurons were optically stimulated with blue light (470 nm) in PV-Cre and SST-Cre mice, respectively, during whole-cell voltage-clamp recordings with the membrane held at + 10 mV (Fig. 3i, n). To analyze the S-R curve of PV/SST interneuron-evoked IPSCs in CA1 PC, a single light pulse (470 nm, 5 ms) was delivered to ChR2-expressing PV or SST interneurons at different light powers (5, 10, 25, 50, 75, 100% of maximal light power (15 mW), Fig. 3j, o). The light power which gave 50% of the maximal IPSC response (half-maximal stimulus, 3.75–9 mW) was used for the subsequent PPR and short-term plasticity analysis, for which a train of ten blue light pulses at 50 Hz were delivered (470-nm light, 5-ms duration, Fig. 3k, p; 3.75–9 mW). The total charge of PV/SST-evoked IPSCs was calculated by integrating the area under the IPSC train (Fig. 3l, q).
Theta-nested gamma oscillation-induced tLTP induction protocol
In order to induce theta-nested gamma oscillation-induced tLTP at CA3-CA1 synapse during theta-nested gamma oscillation-like activity, we paired the presynaptic EPSP evoked by SC stimulation with postsynaptic bursts (4 spikes at 100 Hz, each spike elicited with 3 ms current steps, 800 pA) with a 10-ms time window repeated at 5 Hz  for 200 times. EPSPs were evoked every 6 s using two stimulating electrodes placed in the stratum radiatum of the CA1 area to activate SC, one for monitoring EPSPs in the control pathway and one for test pathway (Fig. 5a, b). Test and control pathways were stimulated 2 s apart. EPSP amplitudes were in the range of 3–5 mV (150–400 μA, 20–80 μs, Digitimer Ltd.) and were recorded at membrane voltage held at − 75 mV. Following 10 min of baseline EPSP recordings of both pathways, tLTP induction protocol was delivered to the test pathway, after which EPSPs were evoked every 6 s in both pathways in either DMSO-treated or AβO1–42-treated hippocampal slices prepared from C57BL/6 mice (Fig. 5c–e). To investigate the effect of activation of PV and SST interneurons on tLTP in AβO1–42-treated hippocampal slices, we expressed ChR2 in either PV or SST interneurons and optically stimulated ChR2-expressing PV or SST interneurons using tonic blue light (470 nm, X-cite 110LED, Excelitas Tech., 100% light intensity) during the tLTP induction in AβO1–42-treated hippocampal slices prepared from PV-Cre or SST-Cre mice, respectively (Fig. 5g–j). tLTP induction was repeated in the presence of 50 μM D-AP5 to see if the tLTP is NMDA receptor-dependent (Fig. 5d, i). The slope of EPSP was calculated as an index of synaptic efficacy, measured by performing a linear fit on the rising slope of the EPSP between time points corresponding to 20 and 80% of the EPSP peak amplitude. Changes in synaptic efficacy were estimated as percentage change relative to the mean EPSP slope during the first 10 min of baseline recordings. To compare synaptic efficacy between neurons and experimental conditions, the mean of the normalized EPSP slope in the time period between 25 and 30 min after the tLTP induction was calculated (Fig. 5f, k).
SST interneuron-mediated disinhibition
To measure SST interneuron-mediated disinhibition during tLTP induction, we performed whole-cell voltage-clamp recordings in PC to record SC stimulation-evoked IPSC before and during tLTP induction. tLTP induction was performed by pairing of presynaptic EPSP and postsynaptic PC spikes by stimulating the SC and evoking postsynaptic spikes by stimulating the CA1 axons in the alveus at 100 Hz (4 pulses) with 10-ms time window, repeated at 5 Hz for 20 times (Fig. 6b, Additional file 10: Figure S10). All recordings were performed in the presence of D-AP5 (50 μM) to prevent synaptic plasticity during tLTP induction. To test if alveus stimulation can elicit spikes in PV and SST interneurons similar to that during blue light-induced theta-nested gamma oscillations as in Fig. 2c, we performed current-clamp recordings in PV and SST interneurons and stimulated alveus at 100 Hz (4 stimuli) repeated at 5 Hz (Additional file 9: Figure S9b, d, top). To ensure that alveus stimulation activated PC axons and is not a result of direct stimulation of other pathways, we repeated the experiments in the presence of D-AP5 (50 μM) and CNQX (20 μM) to block NMDA and AMPA receptors (Additional file 9: Figure S9b, d, bottom). Since alveus stimulation can activate both PV and SST interneurons to provide direct inhibition to PC, we isolated the SC stimulated IPSC during tLTP induction (Additional file 10: Figure S10b, (4), gray) by subtracting the IPSC evoked by alveus stimulation alone (Additional file 10: Figure S10b, (2) Alveus stim, light brown) from the IPSC evoked by pairing SC stimulation with alveus stimulation (Additional file 10: Figure S10b, (3) SC + alveus stim, brown). In calculating the SST interneuron-mediated disinhibition, we took the difference between the IPSC amplitude evoked by SC stimulation alone (Additional file 10: Figure S10b, (1) SC stim, black) and IPSC amplitude calculated in (4) (Additional file 10: Figure S10b, gray). In order to directly test the effect of the activation of SST interneurons on SC stimulation-evoked IPSC, we optically activated ChR2-expressing SST interneurons simultaneously with SC stimulation in the DMSO-treated and AβO1–42-treated hippocampal slices prepared from SST-Cre mice (Additional file 11: Figure S11).
CNQX, SR95531 (GABAzine), and D-AP5 were purchased from Tocris. PBS, Urea, and Aβ1–42/Aβ42–1 powder were purchased from Gibco, Affymetix, and Bachem, respectively. DMSO and the other regents were all purchased from Sigma. For western blot analysis, rabbit monoclonal antibody mOC64was purchased from Abcam (Cat# ab201060, Lot# GR3235744-4, RRID: AB_2818982). Horseradish peroxidase (HRP)-conjugated anti-rabbit antibodies (Cat# 170-6515, Control# 64170140, RRID: AB_2617112), Mini-PROTEAN TGX 4–15% tris-glycine gels, 4x Laemmli sample buffer, Native sample buffer, and running buffer were all purchased from Bio-Rad.
To confirm the expression of opsins in PC, PV, and SST interneurons, hippocampal slices were post-fixed overnight in 4% paraformaldehyde at 4 °C and subsequently washed in PBS. Washed slices were mounted with CUBIC mount solution , a tissue clearing technique that removes lipids from the sample to enhance transparency in imaging. Images were acquired using a confocal microscope (LSM-700, ZEISS) under a × 10 and × 20 objective.
CA3-CA1 hippocampal network model
All data analysis was conducted using Igor Pro or MATLAB with custom-written scripts. Excel (Microsoft) and SPSS (IBM) software were used for statistical analyses.
Data are represented as mean with individual data values or mean ± SEM. Statistical significance was measured using Student’s t test or one-way, one-way repeated-measures, and two-way ANOVA followed by post hoc Tukey’s test. p value less than 0.05 was considered statistically significant. Statistical significance of spike phases was tested using Watson-Williams multi-sample circular test .
We gratefully thank Danny Kanghoon Seo for advices on the optogenetic experimental technique.
KP, JL, and JK designed, conducted, and analyzed all in vitro experiments; BAR and MMK helped setup optogenetic experimental methods; HJJ and JK designed the computational model simulation; HJJ conducted and analyzed computational model simulation data; JK wrote the original draft; and KP, JL, HJJ, MMK, BAR, and JK reviewed and edited the manuscript. All authors read and approved the final manuscript.
Human Frontiers Science Program (HFSP) Young Investigator Award (RGY0073/2015) to B.A.R., M.M.K., and J.K., a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health & Welfare, Republic of Korea (HI15C3086, HI17C0212, HI19C0646) to J.K., the Brain Convergence Research Program (NRF-2019M3E5D2A01058328) through the National Research Foundation (NRF) funded by the Korean Government (MIST) to J.K., and Global Ph.D Fellowship Program through the National Research Foundation of Korea (NRF) funded by Ministry of Education (NRF-2016H1A2A1907615) to K.P.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
- 4.Lambert MP, Barlow AK, Chromy BA, Edwards C, Freed R, Liosatos M, Morgan TE, Rozovsky I, Trommer B, Viola KL, et al. Diffusible, nonfibrillar ligands derived from Abeta1-42 are potent central nervous system neurotoxins. Proc Natl Acad Sci U S A. 1998;95(11):6448–53.PubMedPubMedCentralCrossRefGoogle Scholar
- 7.Martinez-Losa M, Tracy TE, Ma K, Verret L, Clemente-Perez A, Khan AS, Cobos I, Ho K, Gan L, Mucke L, et al. Nav1.1-overexpressing interneuron transplants restore brain rhythms and cognition in a mouse model of Alzheimer’s disease. Neuron. 2018;98(1):75–89 e75.PubMedPubMedCentralCrossRefGoogle Scholar
- 9.Kurudenkandy FR, Zilberter M, Biverstal H, Presto J, Honcharenko D, Stromberg R, Johansson J, Winblad B, Fisahn A. Amyloid-beta-induced action potential desynchronization and degradation of hippocampal gamma oscillations is prevented by interference with peptide conformation change and aggregation. J Neurosci. 2014;34(34):11416–25.PubMedPubMedCentralCrossRefGoogle Scholar
- 24.Gulyas AI, Szabo GG, Ulbert I, Holderith N, Monyer H, Erdelyi F, Szabo G, Freund TF, Hajos N. Parvalbumin-containing fast-spiking basket cells generate the field potential oscillations induced by cholinergic receptor activation in the hippocampus. J Neurosci. 2010;30(45):15134–45.PubMedPubMedCentralCrossRefGoogle Scholar
- 33.Chen L, Saito T, Saido TC, Mody I. Novel quantitative analyses of spontaneous synaptic events in cortical pyramidal cells reveal subtle parvalbumin-expressing interneuron dysfunction in a knock-in mouse model of Alzheimer’s disease. eNeuro. 2018;5(4):0059-18.2018.PubMedPubMedCentralCrossRefGoogle Scholar
- 40.Hijazi S, Heistek TS, Scheltens P, Neumann U, Shimshek DR, Mansvelder HD, Smit AB, van Kesteren RE. Early restoration of parvalbumin interneuron activity prevents memory loss and network hyperexcitability in a mouse model of Alzheimer’s disease. Mol Psychiatry. 2019. https://doi.org/10.1038/s41380-019-0483-4.
- 44.Roberson ED, Halabisky B, Yoo JW, Yao J, Chin J, Yan F, Wu T, Hamto P, Devidze N, Yu GQ, et al. Amyloid-beta/Fyn-induced synaptic, network, and cognitive impairments depend on tau levels in multiple mouse models of Alzheimer’s disease. J Neurosci. 2011;31(2):700–11.PubMedPubMedCentralCrossRefGoogle Scholar
- 47.Shankar GM, Bloodgood BL, Townsend M, Walsh DM, Selkoe DJ, Sabatini BL. Natural oligomers of the Alzheimer amyloid-beta protein induce reversible synapse loss by modulating an NMDA-type glutamate receptor-dependent signaling pathway. J Neurosci. 2007;27(11):2866–75.PubMedPubMedCentralCrossRefGoogle Scholar
- 48.Lacor PN, Buniel MC, Furlow PW, Clemente AS, Velasco PT, Wood M, Viola KL, Klein WL. Abeta oligomer-induced aberrations in synapse composition, shape, and density provide a molecular basis for loss of connectivity in Alzheimer’s disease. J Neurosci. 2007;27(4):796–807.PubMedPubMedCentralCrossRefGoogle Scholar
- 51.Wang Q, Walsh DM, Rowan MJ, Selkoe DJ, Anwyl R. Block of long-term potentiation by naturally secreted and synthetic amyloid beta-peptide in hippocampal slices is mediated via activation of the kinases c-Jun N-terminal kinase, cyclin-dependent kinase 5, and p38 mitogen-activated protein kinase as well as metabotropic glutamate receptor type 5. J Neurosci. 2004;24(13):3370–8.PubMedPubMedCentralCrossRefGoogle Scholar
- 57.Villette V, Poindessous-Jazat F, Simon A, Lena C, Roullot E, Bellessort B, Epelbaum J, Dutar P, Stephan A. Decreased rhythmic GABAergic septal activity and memory-associated theta oscillations after hippocampal amyloid-beta pathology in the rat. J Neurosci. 2010;30(33):10991–1003.PubMedPubMedCentralCrossRefGoogle Scholar
- 71.Baude A, Bleasdale C, Dalezios Y, Somogyi P, Klausberger T. Immunoreactivity for the GABAA receptor alpha1 subunit, somatostatin and Connexin36 distinguishes axoaxonic, basket, and bistratified interneurons of the rat hippocampus. Cereb Cortex. 2007;17(9):2094–107.PubMedCrossRefPubMedCentralGoogle Scholar
- 77.Berens PJJSS: CircStat: a MATLAB toolbox for circular statistics. 2009, 31(10):1–21.Google Scholar
- 82.Koch C. Biophysics of computation: information processing in single neurons. New York: Oxford university press; 2004.Google Scholar
- 86.Zar JH. Statistical procedures for biological-research - a citation classic commentary on biostatistical analysis by Zar,J.H. Cc/Agr Biol Environ. 1989;6:20.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.