Attenuation of Aβ-associated hyperactivity reduces Aβ and tau pathology along the entorhinal cortex-hippocampal network

High levels of the amyloid-beta (Aβ) peptide have been shown to disrupt neuronal function and induce hyperexcitability but it is unclear what effects Aβ-associated hyperexcitability may have on tauopathy pathogenesis or propagation in vivo. Using a novel transgenic mouse line to model the impact of hAPP/Aβ accumulation on tauopathy in the entorhinal cortex-hippocampal (EC-HIPP) network, we demonstrate that hAPP aggravates EC tau aggregation and accelerates pathological tau spread into the hippocampus. In vivo recordings revealed a strong role for hAPP/Aβ, but not tau, in the emergence of EC neuronal hyperactivity and impaired theta rhythmicity. Chemogenetic attenuation of Aβ-associated hyperactivity led to reduced hAPP/Aβ accumulation and reduction of pathological tau in downstream hippocampus. These data strongly support the hypothesis that in Alzheimer’s disease (AD), Aβ-associated hyperactivity accelerates the progression of pathological tau along vulnerable neuronal circuits, and demonstrates the utility of chronic, neuromodulatory approaches in ameliorating AD pathology in vivo.


Introduction 27
The accumulation of hyperphosphorylated, misfolded tau proteins into neurofibrillary tangles 28 (NFT), coupled with deposition of amyloid beta (Aβ) into extracellular plaques, are two hallmark 29 pathological features of Alzheimer's disease (AD) in the brain. The severity of cortical NFT 30 accumulation is strongly correlated with Aβ plaque load (1, 2) and is the principal 31 neuropathological variable associated with cognitive impairment in AD (1-4). The entorhinal 32 cortex (EC; Brodmann Areas 28 and 34) is a structure in the parahippocampal gyrus that plays 33 a critical role in spatial representation and navigation (5-7), and it is one of the first structures to 34 exhibit AD-related tauopathy and subsequent neuronal loss (8,9). As AD progresses, 35 considerable accumulation of pathological tau continues downstream into the hippocampus 36 (HIPP), which is extensively connected to the EC. Preclinical investigation into the stereotypical 37 spread of pathological tau along neuronal circuits in AD is an active area of research interest 38 (10-13). However, the biological mechanisms underlying the propagation of tau pathology in the 39 brain are currently unresolved. 40 In vitro studies utilizing rodent primary neurons have provided several mechanistic 41 insights into the pathophysiological relationship between cleaved amyloid precursor protein 42 (APP) fragments and tau. At the cellular level, accumulating evidence implicates Aβ oligomers 43 as a causative agent in the increased phosphorylation of tau at AD relevant epitopes (14) and 44 the missorting of tau and neurofilaments within the cell (15). In mouse models of tauopathy, 45 stereotaxic injection of Aβ oligomers and fibrils into the brain results in significantly elevated 46 phosphorylation of tau (16) and the increased induction of neurofibrillary tangles (NFT) (17). 47 Thus, tauopathy in the brain may be aggravated by the increased production or accumulation of 48 APP fragments in vivo through direct interaction. For reviews of experimental models that 49 examine Aβ-induced tau alterations and pathology, see (18,19). 50 Alternatively, human APP/Aβ accumulation in the brain may trigger the aggregation and 51 acceleration of tau pathology via an intermediate, non-pathogenic mechanism. Indeed, several 52 reports now describe an effect of Aβ accumulation on neuronal network hyperactivity in Aβ 53 generating mouse models (20-22); for review, see (23,24), as well as in humans with mild 54 cognitive impairment (25,26). Spontaneous, non-convulsive epileptiform-like activity has been 55 described in cortical and hippocampal networks of relatively young transgenic mutant hAPP-56 expressing mice (21). In addition, increased proportions of neurons surrounding amyloid 57 plaques exhibit aberrant hyperactivity (20) and are accompanied by the breakdown of slow-58 wave oscillations (27). Interestingly, neuronal hyperactivity has been shown to precede amyloid 59 plaque formation in the hippocampus, suggesting that the abnormal accumulation of soluble Aβ 60 drives aberrant neuronal network activity (28,29). Thus, it is plausible that Aβ-associated 61 hyperactivity can facilitate the progression of pathological tau in the brain, and does so indirectly 62 without the need for direct Aβ-tau interaction. Interestingly, stimulating neuronal activity can 63 facilitate both Aβ and tau release from neurons in vivo (30,31), and exacerbates Aβ deposition 64 and tauopathy in synaptically connected neurons (13,(32)(33)(34). Mature tau pathology may in turn 65 aggravate Aβ-associated neuronal network dysfunction by further altering neuronal firing rates 66 and network oscillations (35), recruiting neuronal populations into a harmful feedback loop 67 involving protein aggregation and aberrant signaling. 68 In these studies, we utilize a newly developed AD mouse line to resolve the individual 69 effects of Aβ and tau pathology on neuronal activity in the EC. Mice that generate Aβ and tau 70 pathology were compared to littermates that generate either Aβ or tau alone, while non-71 transgenic littermates served as controls. We first demonstrate that hAPP/Aβ expression 72 aggravates tau accumulation in the EC and accelerates pathological tau spread into the HIPP, 73 supporting previous findings (36)(37)(38). In vivo electrophysiological recordings in our mice 74 revealed distinct neuronal hyperactivity and network dysfunction associated with EC hAPP/Aβ 75 expression, but not tau expression. We then employed a chemogenetic approach in the 76 transgenic mice to combat EC neuronal hyperactivity, with the goal of reducing the 77 accumulation of pathological Aβ and tau along the entorhinal cortex -hippocampal (EC-HIPP) 78 network. Chronic attenuation of EC neuronal activity dramatically reduced hAPP/Aβ 79 accumulation in downstream hippocampus and reduced abnormally conformed and hyper-80 phosphorylated tau aggregates along the EC-HIPP network. Our data support the emerging 81 view that Aβ-associated hyperactivity plays a role in AD pathogenesis, specifically by acting as 82 an accelerant of tau spread along synaptically connected neuronal circuits in the brain. 83 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https: //doi.org/10.1101/487405 doi: bioRxiv preprint In vivo multi-electrode recordings were performed in the EC of 16-month EC-Tau/hAPP mice 112 and their age-matched, transgenic and non-transgenic littermates, as previously described (39). 113 Briefly, tetrodes from custom-built microdrives (Axona, UK) were positioned to target cell layers 114 II/III in the dorsal EC (see methods). Single-unit activity and local field potentials (LFPs) were 115 collected while mice freely explored a large open field arena. Each experimental mouse 116 underwent four to six recording sessions total over the course of several days, with only one 117 recording session performed per day. Tetrodes for each mouse were moved down 0.1 mm from 118 their previous position 24hr prior to the next recording session, allowing stable electrode 119 positioning and a robust sampling of EC LII/III neuronal activity for each mouse. IHC was 120 performed on post-mortem brain sections at the end of the study to confirm electrode placement 121 ( Figure 2A) and to examine the distribution of Aβ and tau pathology in the EC-HIPP network. 122 A total of 1,260 EC neurons were recorded and analyzed from 20 mice (Control,n=7;123 EC-Tau, n=4; hAPP, n=5; EC-Tau/hAPP, n=4) ( Figure 2) (for detailed methodology, see 124 Methods). Plotting the cumulative frequency distributions of the spontaneous, average firing 125 rates of all recorded neurons showed a clear shift towards higher firing rates in both EC-126 Tau/hAPP mice and hAPP mice versus Control mice ( Figure 2B). Individual two-sample 127 Kolmogorov-Smirnov tests comparing the distributions of EC-Tau/hAPP and hAPP neuronal 128 firing rates to Controls confirmed this shift (p < 0.001). The average firing rates for pooled EC 129 neurons in each genotype were then compared ( Figure 2B, insert). EC-Tau/hAPP and hAPP 130 neurons each exhibited nearly a two-fold increase in their average firing rates versus Control 131 neurons. Interestingly, the average firing rates of EC-Tau neurons was not significantly different 132 from Controls, suggesting that the increased hyperactivity in EC-Tau/hAPP neurons was driven 133 by hAPP expression and/or Aβ deposition, but not tau. To examine task-relevant neuronal firing 134 rates during active exploration, we applied a minimum speed filter (<5cm/sec) to the data to 135 remove EC spiking activity occurring during bouts of immobility (Supplemental Figure 1A). hAPP 136 (2.291 ± 0.141 Hz) and EC-Tau/hAPP (2.658 ± 0.170 Hz) neurons once again exhibited 137 significantly increased average firing rates compared to Control (1.299 ± 0.092 Hz) in speed 138 filtered datasets. We then examined the inter-spike interval (ISI) values for EC single-units as an 139 additional measure of neuronal hyperactivity. Cumulative frequency distributions of the median 140 ISIs showed that EC-Tau/hAPP and hAPP single-units were skewed towards shorter ISIs than 141 Control cells for ~65% of the data ( Figure 2C). Interneuron dysfunction has previously been described in hAPP expressing mice, 149 suggesting that shifts in the excitation-inhibition ratio within cortical and hippocampal networks 150 may be responsible for emergent epileptiform activity (21, 40). We first plotted the average firing 151 rates of all EC single-units as a function of the waveform's averaged spike-width (µs). This 152 revealed a distinct population of hyperactive neurons in Aβ-generating mice that were 153 distributed along narrow and wide spike widths ( Figure 2D, purple). We also noticed what 154 appeared to be a bimodal distribution in the scatterplot. To examine cell-type specific firing 155 patterns in our single-unit recordings, we then plotted cell frequency as a function of average 156 spike width ( Figure 2E) (39,41). This resulted in a clear bimodal distribution of the cell 157 population, wherein narrow-spiking (NS) cells could be separated from wide-spiking (WS) cells 158 ( Figure 2D Figure 1B-C). These data agree with previous 169 reports describing Aβ-associated dysfunction in hippocampal interneurons, and expand this 170 hyperactivity phenotype to both putative interneurons and putative excitatory neurons in EC. 171 Impaired neuronal network activity has been described in human AD (42) and in mouse 172 models of AD pathology (43)(44)(45). To investigate the effects of hAPP/Aβ and tau pathology on 173 EC network activity in vivo, we examined the LFPs and compared oscillatory activity across 174 genotypes ( Figure 3). Initial, visual inspection of the filtered LFP signal in theta (4-12 Hz), low 175 gamma (35-55 Hz), and high gamma (65-120 Hz) frequency ranges, as well as the filtered LFP 176 spectrograms, demonstrate impaired theta rhythmicity in 16-month EC-Tau/hAPP mice 177 compared to non-transgenic Control mice ( Figure 3A-B Tau/hAPP (n=8) mice and hAPP (n=9) mice exhibited a marked decrease in % power 181 distribution within the theta frequency range compared to Control (n=7) mice. Theta power was 182 not significantly affected in EC-Tau (n=5) mice (p > 0.05). No genotype differences were 183 detected in the low gamma or high gamma frequency ranges. As running speed can impact 184 theta power in the hippocampal formation (46, 47), we speed filtered our LFP recordings and re-185 analyzed the data to remove activity during bouts of immobility (Supplemental Figure 1D-F).

186
Similarly, we found an Aβ-associated decrease in % theta power, though this was accompanied 187 by an Aβ-associated increase in % low gamma power. Averaged, speed filtered % high gamma 188 power values were not significantly different across genotypes. 189 Overexpression of hAPP Swedish/Indiana has previously been associated with increased 190 locomotor activity in the hAPP/J20 mouse line (22,48,49). To investigate locomotor activity in 191 16-month EC-Tau/hAPP mice, we analyzed the position data from our recorded mice as they 192 performed a foraging task in the open field arena ( Figure 4). We did not detect significant 193 differences between groups in the total distance traveled in the arena (m), % of arena coverage 194 or average speed (cm/sec) during exploration ( Figure 4A-D). To examine whether group 195 differences in locomotor activity exist within the initial phases of recording, we split the sessions 196 and examined performance measures in the first 5, 10 and 15 min ( Figure 4E-G (1mg -1 /kg -1 /day -1 ) into the peritoneum. EC spiking activity was noticeably decreased during the 229 last three weeks of CNO treatment (n=9 mice, EC-Tau/hAPP and EC-Tau mice) ( Figure 5A), 230 and theta power was significantly reduced at the fifth (T5) and sixth week (T6) of CNO treatment 231 compared to baseline ( Figure 5B), confirming chronic hM4D i DREADDs activation in vivo. After 232 6-weeks of hM4D i EC DREADDs activation, all experimental mice were sacrificed and 233 immunostaining was performed on horizontal brain sections to confirm EC DREADDs 234 expression and electrode placement, and to identify pathological Aβ and/or tau deposition along 235 the EC-HIPP network. An overlay of mCherry signal (hM4D i DREADDs) and eGFP (control 236 virus) expression patterns for EC-Tau/hAPP mice (n=4) and transgenic littermates (hAPP, n=3; 237 EC-Tau, n=5) is shown ( Figure 5C & Supplemental Figure 4A-G). hM4D i DREADDs expression 238 in the right hemisphere was primarily localized to cell bodies and neuropil throughout the EC 239 and pre-and para-subiculum, with occasional mCherry signal present in the subiculum, and at 240 terminal ends of axons in the middle-and outer-molecular layers of the DG ( Figure 5C-D).

241
Importantly, we did not detect DREADDs mCherry crossover into the contralateral left 242 hemisphere. 243 EC-Tau/hAPP mice and hAPP mice that were administered 1mg -1 /kg -1 /day -1 CNO for 6-244 weeks exhibited a dramatic reduction in Aβ accumulation within subregions of the hippocampus 245 downstream of the DREADDs-activated right EC ( Figure 5E-G administered a low dose of 0.5mg -1 /kg -1 /day -1 CNO for 6-weeks ( Figure 5H). No significant 253 differences were detected between right and left hippocampal area (mm 2 ) sampled for 6E10+ 254 immunostaining comparisons (Supplemental Figure 3 A-D). 255 These data provide evidence to support the utility of chronic, neuromodulatory 256 intervention in the EC-HIPP network of hAPP/Aβ-expressing mice, as 6E10+ immunoreactivity 257 was significantly reduced in the downstream HIPP after 6-weeks of hM4D i EC DREADDs 258 activation. 259 260 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint

Chronic attenuation of EC neuronal activity in vivo reduces tau pathology along the EC-261 HIPP network 262
We have previously shown that increased neuronal activity can aggravate tau pathology in EC-263 Tau mice (32). However, it is unclear whether attenuation of neuronal activity, or attenuation of 264 Aβ-associated hyperactivity in the case of EC-Tau/hAPP mice, can ameliorate local tau 265 pathology or arrest its spread in vivo. To address these questions, we performed a series of 266 DAB-IHC staining experiments on horizontal brain sections from chronic hM4D i EC DREADDs 267 activated EC-Tau/hAPP mice (n=4) and EC-Tau (n=5) mice. 268 We first identified tau aggregates in our tissue by staining for the conformation 269 dependent antibody MC1, which recognizes pathological, abnormally conformed tau (52). Initial immunostaining could be seen in ipsilateral, downstream DG granule cells, CA1 pyramidal cells 290 and Subiculum in hM4D i EC DREADDs activated EC-Tau mice ( Figure 6N-O). Selective 291 decreases in MC1+ immunostaining could also be seen along the EC-HIPP network in EC-292 Tau/hAPP mice with Advanced Tau pathology, with notable reductions in somatic MC1+ 293 immunostaining in rPaS and rEC ( Figure 6Q-R) in the DREADDs activated right hemisphere, as 294 well as in rCA1 pyramidal cells ( Figure 6T). 295 We then processed adjacent tissue sections from experimental mice to identify 296 phosphorylated tau using the phosphotau-specific antibody AT8 (Ser 202 /Thr 205 ) (53). Reductions 297 in somatodendritic phosphotau accumulation was also identified along the EC-HIPP network of 298 our hM4D i EC DREADDs activated EC-Tau/hAPP mice and EC-Tau mice (Figure 7). EC-Tau 299 mice exhibiting Early Tau pathology ( Figure 7A) showed reduced AT8+ immunoreactivity in the 300 rPaS ( Figure 7B) and rEC ( Figure 7C) compared to analogous regions in the contralateral 301 hemisphere. EC-Tau/hAPP mice exhibiting Early Tau pathology showed a similar reduction in 302 AT8+ immunostaining ( Figure 7F-H). As significant tau pathology had not progressed into the 303 HIPP of these mice at 16-months, no hemisphere differences in HIPP subfield AT8+ 304 immunoreactivity were detected ( Figure 7D-E, I-J). EC-Tau mice ( Figure 7K) and EC-Tau/hAPP 305 mice ( Figure 7P) exhibiting Advanced Tau pathology showed reductions in AT8+ cell bodies 306 and neurites within the rPaS, rEC and rHIPP subfields. Somatic AT8+ immunostaining was 307 reduced in the rSub, rDG granule cells and rCA1 pyramidal cells of EC-Tau mice ( Figure 7M-O). 308 EC-Tau/hAPP mice showed reduced AT8+ immunostaining in rPaS and rEC of the DREADDs 309 activated right hemisphere, in addition to downstream CA1 pyramidal cells ( Figure 7T). We also 310 stained adjacent tissue sections from experimental mice with the antibody CP27 (human Tau 130-311 150 ) to detect the distribution of total human tau along the EC-HIPP network. Similar to MC1+ 312 and AT8+ immunostaining, CP27+ immunostaining revealed selective decreases in total human 313 tau along the EC-HIPP network in the DREADDs activated right hemisphere of our mice 314 (Supplemental Figure 5). 315 The extent to which chronic hM4D i EC DREADDs activation reduced pathological tau 316 immunoreactivity in the hippocampus varied across mice, with some mice showing strong 317 effects in some regions and others subtle effects. In addition, reduced tau pathology in certain 318 mice was readily detected using some, but not all, tau antibodies. This effect is illustrated in a 319 16-month DREADDs-activated EC-Tau/hAPP mouse, where reduced tau aggregation is 320 apparent in CP27+ immunostaining, but not in MC1+ or AT8+ immunostaining ( Figure 8A). To 321 examine the downstream impact of attenuated EC neuronal activity on hippocampal tau 322 pathology, we performed threshold analysis for each pathological tau marker in the DG, CA1 323 and SUB. We then calculated the % area of MC1+, AT8+ and CP27+ above threshold for each 324 ROI, similar to methods described in Figure 5G-H analysis. Finally, we generated a right 325 hemisphere versus left hemisphere ratio for each immunostained section and pooled them 326 according to hippocampal subfield ( Figure 8B) and pathological tau marker ( Figure 8C). Values 327 less than 1 indicate reduced tau accumulation in the treated right hemisphere, while values 328 greater than 1 indicate increased tau accumulation in the right hemisphere. Plotting the right 329 versus left hemisphere ratios by region revealed an effect of hM4D i EC DREADDs activation on 330 tau pathology in the DG (0.8029 ± 0.067) and CA1 (0.7598 ± 0.104), but not the Sub (1.0000 ± 331 0.117). Plotting the right versus left hemisphere ratios by tau marker revealed the strongest 332 effect of hM4D i EC DREADDs activation on CP27+ immunostaining (0.7714 ± 0.915). AT8+ 333 immunostaining was trending (0.7963 ± 0.097), while MC1+ immunostaining did not show an 334 effect (0.9951 ± 0.108). 335 In conclusion, our experimental data strongly support the emerging hypothesis that 336 hAPP/Aβ accumulation in the brain is associated with aberrant neuronal activity and network 337 impairment. Our data describe a role for Aβ-associated neuronal hyperactivity in accelerating 338 tau pathology along a well-characterized neuronal network that is vulnerable to AD pathology 339 and neurodegeneration. We further show that hyperactivity in this network can be targeted via 340 chronic chemogenetic activation to arrest the accumulation of both hAPP/Aβ and pathological 341 tau along the EC-HIPP network in vivo. 342 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint

Discussion 343
The progressive, stereotypical spread of pathological tau along neuronal circuits in AD is an 344 active area of intense investigation. However, the underlying mechanisms that potentiate 345 pathological tau spread in AD are currently unresolved. We, and others, have proposed that 346 increased neuronal activity can exacerbate tauopathy either by promoting tau release from 347 neurons or by facilitating its uptake in synaptically connected neurons (32, 37). To model the 348 functional interactions between hAPP/Aβ and hTau in a well-characterized neuronal circuit in 349 vivo, we crossed the Aβ-generating hAPP/J20 (Swedish/Indiana) mouse line (21, 54) with the EC- Tau  350 mouse line, wherein mutant hTau (P301L) is predominantly expressed in the EC (10, 11). Network 351 dysfunction and impaired cognitive performance have previously been described in relatively 352 young hAPP mice (21,22,48). Mutant hAPP expression in the EC-Tau/hAPP mouse thus 353 provides a strong in vivo platform to model the impact of Aβ-associated hyperactivity on tau 354 pathology. At 16 months, somatodendritic MC1+ immunostaining was increased in the EC and 355 DG ( Figure 1C-E). By 23-months, the distribution and severity of MC1+ immunostaining 356 resembled that of much older (30+ month) EC-Tau mice ( Figure 1F) (39, 55), suggesting that 357 hAPP/Aβ plays a significant role in promoting the intercellular transfer of pathological tau in vivo.

358
Our data supports previously published findings in similar mouse models, where robust 359 hAPP/Aβ accumulation was associated with accelerated tauopathy along the EC-HIPP network 360 (36-38). 361 We hypothesized that increased Aβ production or accumulation may trigger an 362 intermediate, non-pathogenic cascade of events that impact tau. Indeed, substantial evidence 363 exists to support the hypothesis that increased Aβ leads to neuronal hyperactivity and large-364 scale network dysfunction in the brain (24). This hypothesis may partially explain the 365 accelerated progression of pathological tau in AD and in mouse models of AD pathology. 366 hAPP/Aβ accumulation has been linked to the appearance of epileptiform-like network activity in 367 the brain at a relatively young age (21,22,56). This aberrant network activity is associated with 368 changes in inhibitory neuron profiles and remodeling of the DG. More recently, hyperactive 369 neurons have been shown to disproportionately cluster around Aβ plaques in cortex (20) (EC-Tau/hAPP and hAPP) mice exhibited significant EC single-unit hyperactivity and network 374 dysfunction, characterized by increased average firing rates, decreased median ISIs ( Figure 2B-375 C) and decreased % theta power compared to control mice ( Figure 3A-C). Aβ-associated EC 376 hyperactivity was not attributed to oversampling of neuronal bursting in the hAPP-expressing 377 mice, as no significant group differences were found in the % of EC bursting activity recorded 378 (One-way ANOVA: F (3) = 2.426, p = 0.1034) ( Table 1). Plotting the average firing rates of pooled 379 EC neurons as a function of spike width revealed a distinct population of hyperactive neurons in 380 Aβ-generating mice that was distributed among putative interneurons (NS) and excitatory cells 381 (WS) ( Figure 2D-G). This increased proportion of hyperactive cells was not due to oversampling 382 of NS neurons in Aβ-generating mice (NS, n=212) versus Non Aβ-generating mice (NS, n=196) 383 (Table 1). Additionally, Aβ-associated EC hyperactivity and theta impairment was also present 384 after applying a minimum speed filter to the datasets, which removed neuronal activity during 385 bouts of immobility (Supplemental Figure 1). These data support previous findings describing 386 Aβ-associated interneuron dysfunction and network hyperactivity (21,22,57), and extend them 387 to putative inhibitory and excitatory neurons in EC. hAPP/Aβ-associated disruptions in theta 388 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint oscillatory activity are also in line with previous reports (43,45). Finally, we did not detect 389 hAPP/Aβ-mediated effects on behavioral activity during open field recording sessions ( Figure 4). 390 This is in contrast to several reports describing locomotor hyperactivity in the hAPP/J20 mouse 391 line (22,48,49). We predict that this may have been due to increased motivational drive to 392 actively explore the arena in our mice, as sucrose pellets were administered during recording 393 sessions to encourage foraging behavior and arena coverage. Repeated handling and 394 acclimation to the recording paradigm may have also reduced innate anxiety-like behavior in our 395 mice, leading to similar behavioral activity patterns. Importantly, we can conclude that Aβ-396 associated neuronal hyperactivity and network dysfunction are not due to gross locomotor 397 differences in our mice. 398 Several lines of evidence now support a role for increased neuronal activity in both Aβ 399 accumulation and accelerated tau pathology in vivo. Stimulation of the perforant pathway results 400 in increased Aβ concentrations in hippocampal interstitial fluid (ISF) (30), and increased Aβ 401 deposition in downstream DG (outer molecular layer) (33). Chemogenetic stimulation of cortical 402 neurons is also associated with increased deposition of mature amyloid plaques (34). Likewise, 403 stimulating neuronal activity results in increased ISF hTau concentrations (31), and chronic 404 stimulation of EC neurons enhances local tauopathy and accelerates neurodegeneration (32).

405
These data suggest that Aβ-associated hyperactivity can impact pathological tau progression 406 along a defined neural circuit while simultaneously driving increased hAPP/Aβ release. The 407 afflicted circuit could then potentially be recruited into a harmful, positive feedback loop that 408 drives aggressive pathological hAPP/Aβ and tau aggregation in the neuronal network, leading to 409 cognitive impairment and cell death. Aged mice with significant tauopathy along the EC-HIPP 410 network exhibit deficits in spatially modulated grid cell function and impaired spatial learning and 411 memory, as well as excitatory neuron loss, independent of Aβ pathology (39, 55). 412 Surprisingly, the increased aggregation of pathological tau in EC did not appear to affect 413 measures of hyperactivity or network function in 16-month EC-Tau mice. Single-unit average 414 firing rates, ISI medians and % theta power were remarkably similar in EC-Tau/hAPP and hAPP 415 mice, and in EC-Tau and non-transgenic Control mice. These findings support a previous report 416 from our lab describing normal EC single-unit function in 14-month EC-Tau mice, where 417 average firing rates and SM cell firing patterns matched that of non-transgenic controls (39). 418 These data suggest that the accumulation of tau in this model does not strongly impact EC 419 neuronal activity by 16-months in vivo. However, these data should be carefully interpreted, as 420 EC-hTau overexpression has been linked to hypometabolism in ~9-month mice (36) and has 421 recently been shown to blunt Aβ-associated hyperexcitability in vitro and in vivo (58, 59). It is 422 possible then that detection of hTau-mediated effects on neuronal activity is partly dependent on 423 the sensitivity of the assays used to measure them and degree of pathological severity. Our 424 analysis in 16-month animals revealed blunted hyperactivity in NS neurons of EC-Tau/hAPP 425 mice ( Figure 2F), which may represent an early, synergistic effect of Aβ on tau-mediated 426 inhibitory interneuron dysfunction that could precede subsequent impairments in excitatory 427 neurons and gross network function. We have shown that aged (30+ month) EC-Tau mice 428 exhibit significant hypoactivity in excitatory MEC grid cells (39). Thus, we predict that divergent 429 hAPP/Aβ and hTau effects on neuronal activity would be readily observed in the EC-HIPP 430 network of aged EC-Tau/hAPP mice. 431 We implemented a chemogenetic approach in our studies to combat the aggressive 432 progression of Aβ-associated EC hyperactivity on Aβ and tau pathology in vivo. hM4D i 433 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint DREADDs were targeted to EC principal neurons based on the finding that WS (wide-spiking) 434 cells showed hyperactivity in both hAPP mice and EC-Tau/hAPP mice ( Figure 2F-G). 435 Importantly, DREADDs-mediated neuromodulation has been shown to reduce local Aβ 436 deposition in cortex (hM4D i ) (34) and facilitate the transfer of hTau into distal post-synaptic cell 437 populations (hM3D q ) (13). Using a within-subjects design, we probed for hAPP/Aβ and 438 pathological tau immunoreactivity in both the ipsilateral, downstream hippocampus and in the 439 contralateral hippocampus. After 6-weeks of hM4D i EC DREADDs activation, we found a 440 marked reduction in 6E10+ immunoreactivity within ipsilateral hippocampus of EC-Tau/hAPP 441 and hAPP mice, supporting previous reports linking stimulated neuronal activity to Aβ release 442 and Aβ pathology ( Figure 5E-H) (30,33,34,60,61). Chronic hM4D i EC DREADDs activation in 443 hTau-expressing mice also led to selective reductions in pathological tau immunoreactivity 444 within ipsilateral, downstream hippocampal subfields, as well as local regions where hM4D i 445 DREADDs were expressed (Figures 6-7, Supplemental Figure 5). Extensive somatodendritic 446 tau aggregates were readily observed within the hippocampus in a subset of mice 447 (characterized as exhibiting 'Advanced' tau pathology). To visualize trends in hemispheric tau 448 differences according to hippocampal subfield or tau marker, we pooled right versus left 449 hemisphere ratio values and compared group means to a hypothetical mean of 1.0, which would 450 indicate equal distribution of pathological tau in left and right hemisphere ROIs ( Figure 8). We 451 consistently observed reduced tau pathology in ipsilateral rDG granule cells and rCA1 pyramidal 452 neurons, while no group trends were found in rSub ( Figure 8B). These data are the first to 453 demonstrate that attenuated neuronal activity can reduce pathological tau in vivo, and support 454 previous reports showing that stimulated neuronal activity can increase tau release and 455 tauopathy in AD mouse models (12,13,31,32). 456 Recent concerns have been raised in the literature regarding the utility of CNO as an 457 inert DREADDs ligand that easily permeates the blood-brain barrier (BBB) (62-64): for review, 458 see (65). In our studies, we confirmed chronic EC DREADDs activation in vivo using recording 459 metrics derived from single, high-dose CNO injection studies. Acute hM4D i DREADDs activation 460 resulted in decreased EC neuronal activity beginning at ~20 min, with a maximal response at 461 ~30 min that lasted for at least 4 hr (Supplemental Figure 2A -B). This data is supported by 462 previous in vivo research showing strong hM4D i activation in EC 30 min post-CNO, with activity 463 levels recovering towards baseline by 12 hr (66). Acute hM4D i DREADDs activation also led to 464 a robust decrease in % theta power in EC lasting over an hour (Supplemental Figure 2D-E). 465 Therefore, we tracked chronic EC DREADDs activation in vivo by analyzing total spike counts 466 and % theta power ( Figure 5A-B), rather than measuring levels of CNO or converted clozapine 467 in peripheral blood/plasma. Consistent with our single CNO injection findings, total spike counts 468 and % theta power were reduced over chronic CNO treatment, supporting the utility of long-term 469 CNO delivery in indwelling osmotic minipumps to activate DREADDs in vivo (see also (67)). 470 Expression of hM4D i DREADDs was restricted to the right hemisphere of our mice and 471 did not crossover into the contralateral, left hemisphere ( Figure 5C-D and Supplemental Figure  472 4). Thus, we were able to discriminate the effects of chronic EC DREADDs activation on 473 pathology in ipsilateral, downstream HIPP (right hemisphere) and directly compare it to 474 pathology in the contralateral, control left HIPP. We predict that any off-target effects of chronic, 475 converted CNO-to-clozapine on hAPP/Aβ and tau pathology would have impacted both left and 476 right hemispheres in our experimental mice, as continuous 6-week delivery of 1mg -1 /kg -1 /day -1 477 CNO was performed using minipumps. 6E10+ immunostaining revealed strong decreases in 478 hAPP/Aβ in the ipsilateral rHIPP ( Figure 5E-G), downstream of the DREADDs activated rEC. 479 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint We did not detect hemisphere differences in 6E10+ immunoreactivity in age-matched hAPP 480 mice or EC-Tau/hAPP mice sampled from our colony (no DREADDs manipulation), or in hM4D i 481 EC DREADDs expressing EC-Tau/hAPP mice administered a lower CNO dose (0.5mg -1 /kg -482 1 /day -1 ) ( Figure 5H and Supplemental Figure 4H). Furthermore, regional decreases in 483 abnormally conformed tau (MC1+), phosphotau accumulation (AT8+) and total human tau 484 (CP27+) were detected in ipsilateral, rHIPP subfields after chronic hM4D i EC DREADDs 485 activation. The degree to which pathological tau and Aβ accumulation were reduced in 486 downstream HIPP, along with our electrophysiology data showing reduced neuronal activity, is 487 consistent with the hypothesis that ~50-75% maximal DREADDs activation could be achieved 488 with 1mg -1 /kg -1 /day -1 CNO (68). 489 Attenuating neuronal hyperactivity and network dysfunction may prove to be a powerful 490 tool in combating impaired cognition in human AD, especially when paired with therapies aimed 491 at alleviating the aggregation and deposition of Aβ and tau. On its own, Aβ-targeted 492 immunotherapy has proven unsuccessful at relieving AD cognitive symptoms, which may be 493 due to ineffective amelioration or the exacerbation of neuronal hyperactivity (29, 69). Indeed, 494 reducing AD pathology-associated neuronal dysfunction with the antiepileptic drug leviteracetam 495 has shown promise in preclinical mouse models of hAPP overexpression (22, 70), and is being 496 tested for efficacy in AD clinical trials. Our data shows that alleviating Aβ-associated EC 497 hyperactivity reduces downstream accumulation of both Aβ and tau pathology in the 498 hippocampus. An important step forward will be to replicate these findings in non-499 overexpressing hAPP mice (APP NL-F/NL-F ), which show early signs of neuronal hyperexcitability 500 in vitro and impaired gamma oscillations in vivo (71-73). In addition, future studies will be 501 required to determine if relief from pathological Aβ and tau using approaches such as 502 chemogenetics will improve cognitive behavior. Interestingly, reducing murine tau levels has 503 been shown to alleviate locomotor hyperactivity in young hAPP/J20 mice and can blunt 504 pharmacologically induced aberrant over-excitation (49,74). This would indicate that several 505 potential mechanisms exist in the brain to contribute to impaired neuronal activity in AD, and 506 provide ample avenues for investigation into the etiology of AD progression. 507 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.

Experimental animals 523
Three transgenic mouse lines were used in these studies to model hallmark AD pathologies in 524 the brain. 1 st Line: EC-Tau mice that overexpress human mutant tau (4R0N P301L) 525 predominantly in the EC (10, 11, 75) on a C57BL/6 background. 2 nd Line: hAPP/J20 mice that 526 overexpress hAPP with two familial AD mutations (KM670/671NL, Swedish) (V717F, Indiana) 527 (54, 76) on a FVB/N background. 3 rd Line: EC-Tau/hAPP mice, which generate both hAPP/Aβ 528 and tau pathology in the brain, were created by crossing the EC-Tau and hAPP mouse lines. 529 A total of 58 mice, both males and females, were used as experimental animals in these 530 studies. Total mouse numbers per genotype were as follows: non-transgenic controls (Control, 531 n=8), and age-matched, transgenic littermates (EC-Tau, n=6; hAPP, n=9; EC-Tau/hAPP, n=8). 532 Additionally, two mice from each genotype were sampled from our colony at 10-months, 16-533 months, and 23-months of age for initial pathology studies (n=18) in Figure 1. All mice were 534 housed in a temperature and humidity controlled vivarium at Columbia University Medical 535 Center and maintained on a standard 12 hr light/dark cycle with food and water provided ad 536 libitum. All animal experiments were performed during the light phase in accordance with 537 national guidelines (National Institutes of Health) and approved by the Institutional Animal Care 538 and Use Committee of Columbia University. 539 540

Microdrive construction 541
Microdrives were constructed as described previously (39, 77). Briefly, custom-made reusable 542 16-channel or 32-channel microdrives (Axona, UK) were outfitted with four to eight tetrodes 543 consisting of twisted, 25 mm thick platinum-iridium wires (California Wires, USA) funneled 544 through a 23 ga stainless steel inner cannula. A 19 ga protective, stainless steel outer cannula 545 was slipped over the inner cannula and secured to the microdrive body via modeling clay. 546 Individual electrodes were wrapped tightly around the exposed wires of the microdrive and 547 coated with a layer of Pelco conductive silver paint (Ted Pella, Inc., USA) prior to sealing of the 548 microdrive body with liquid electrical tape (Gardner Bender, USA). Several hours prior to 549 surgery, the tetrodes were cut to an appropriate length and electroplated with a platinum/gold 550 solution until the impedances dropped within a range of 150-200 ohms. 551 552

DREADDs virus microinjection and tetrode implantation 553
A total of 30 mice were implanted and recorded for in vivo electrophysiology studies (Control,554 n=7; EC-Tau, n=6; hAPP, n=9; EC-Tau/hAPP, n=8). On the day of electrode implantation, mice 555 were anesthetized with isoflurane (3-4% for induction; 0.5-3% for maintenance) using a multi-556 channel VetFlo Traditional Anesthesia vaporizer (Kent Scientific) and fixed within a stereotaxic 557 frame (Kopf Instruments). As described previously (39), an incision was made to expose the 558 skull and 3 jeweler's screws were inserted into the skull to support the microdrive implant. A 2 559 mm hole was made on each side of the skull at position 3.0-3.1 mm lateral to lambda and ~ 0.2 560 mm in front of the transverse sinus. Viral delivery of the G i -coupled hM4D i DREADD (AAV5-561 CamKII α -hM4D i -mCherry, 4.1x10 12 ) (Cat No. 50477-AAV5; Addgene, USA) was administered 562 into the right EC via a 33 ga NeuroSyringe (Hamilton, USA) tilted at an angle of 6-7 o in the 563 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint sagittal plane. A control virus (AAV5-CamKII α -eGFP, 4.1x10 12 ) (Cat No. 50469-AAV5; Addgene, 564 USA) was administered into the contralateral left EC using identical parameters. At this point, an 565 additional screw connected with wire was inserted into the skull, serving as a ground/reference 566 for local field potential (LFP) recordings. The prepared 16-channel microdrive was then tilted at 567 6-7 o on a stereotaxic arm and the tetrodes lowered to 1.0 mm from the surface of the brain 568 (below dura) into the DREADDs-delivered right hemisphere. The microdrive ground wire was 569 then soldered to the skull screw wire and the microdrive was secured with dental cement. Mice 570 were then allowed to recover in a cleaned cage atop a warm heating pad until awake (~45 min) 571 before being transported to the housing facility. Mice received Carprofen (5mg/kg) prior to 572 surgery and post-operatively to reduce pain, in additional to a sterile saline injection (s.c.) to aid 573 in hydration. Recording experiments began approximately one week from the time of surgery. 574 An additional cohort of 16-month hAPP (n=3) mice were microinjected with virus and 575 implanted with osmotic minipumps for long-term DREADDs activation, but did not receive 576 microdrive implants for electrophysiological recordings. 577

578
In vivo recording: single-unit and local field potential analyses 579 All mice outfitted with microdrives underwent four to six recording sessions in an arena (70 cm x 580 70 cm), with one recording session performed per day. Tetrodes for each mouse were moved 581 down 100 µm from their previous position 24 hr prior to the next recording session, allowing 582 stable electrode positioning and a robust sampling of EC neuronal activity for each mouse. 583 Additionally, the arena and visual cue were rotated between sessions. 584 Neuronal signals from our mice were recorded using the Axona DacqUSB system, and 585 described previously (39). Briefly, recording signals were amplified 10,000 to 30,000 times and 586 bandpass filtered between 0.8 and 6.7 kHz. The LFP was recorded from four channels of each 587 microdrive, amplified 8,000 to 12,000 times, lowpass filtered at 125 Hz and sampled at 250 Hz. 588 60 Hz noise was eliminated using a Notch filter. Spike sorting was performed offline using TINT 589 cluster-cutting software and Klustakwik automated clustering tool. The resulting clusters were 590 further refined manually and were validated using autocorrelation and cross-correlation 591 functions as additional separation tools. Only cells that produced a minimum of 100 spikes with 592 less than 1% refractory period violations (refractory periods < 1ms) were used for subsequent 593 analysis (41). Single-units with no undershoot in their waveform were discarded (defined as 594 Area Under the Peak < 1.0 µV 2 ). Putative excitatory neurons (WS, wide-spiking) were 595 distinguished from putative interneurons (NS, narrow-spiking) by first examining the frequency 596 distribution histogram of pooled EC single-unit spike widths, and then bisecting the waveform 597 spike widths according to the following calculation: Mice were once again anesthetized with isoflurane as previously described and the fur 618 clipped at the abdomen. A single injection of marcaine (2mg/kg, 0.05mL) was delivered 619 intradermally into the site of incision ~ 5 min prior to surgery, then a midline incision was made 620 in the abdominal wall and a sterile, CNO-filled minipump was maneuvered into the 621 intraperitoneal (i.p.) cavity. The incision was then closed using an absorbable suture (Henry 622 Schein, USA) for the abdominal layer, followed by closure of the skin with a nylon synthetic non-623 absorbable suture (Henry Schein, USA). A topical antibiotic was then administered at the 624 surgical site and the mice were allowed to recover in a cleaned cage atop a warm heating pad 625 until awake (~15 min). Mice implanted with minipumps were administered Carprofen (5mg/kg) 626 prior to surgery and post-operatively to help reduce pain. Non-absorbable sutures were 627 removed ~10 days after surgery. 628

DAB immunohistochemistry and immunofluorescence imaging 630
At 6 weeks, all mice were deeply anesthetized with a cocktail of ketamine/xylazine before being 631 transcardially perfused with ice-cold 100mM phosphate-buffered saline (PBS) pH 7.4., followed 632 by 10% formalin (Fisher Scientific, USA). The last recording position for each microdrive-633 implanted mouse was recorded, and then the microdrive removed. Brains were then harvested 634 and left in 10% formalin overnight, then incubated in 30% sucrose until the brains sank to the 635 bottom of the conical tube (all at 4 o C). Horizontal brain sections were sliced (30µm) using a 636 Leica CM3050 S cryostat and stored in cryoprotectant at -20 o C until immunostaining 637 procedures. For both immunoperoxidase staining and immunofluorescence imaging, EC-638 Tau/hAPP brain sections were processed in parallel with sections from EC-Tau and hAPP mice, 639 and non-transgenic Control mice where appropriate. Finally, all tissue sections included for 640 semiquantitative analysis of hAPP/Aβ ( Figure 4) and tau (Figure 7) were verified to be within the 641 range of DREADDs mCherry expression by first checking for native fluorescence signal in free-642 floating sections on an inverted Olympus epifluorescence microscope. 643 Immunoperoxidase staining was performed using a Mouse-on-Mouse kit (Vector 644 Laboratories, USA) and modified from previous methodology (11,39,55 completely, and then dehydrated in ethanol and cleared with xylenes before being coverslipped. 659 Horizontal tissue sections used to visualize native DREADDs mCherry and eGFP 660 expression in the brain were first rinsed in PBS containing 0.3%Triton X-100 (Sigma Aldrich, 661 USA) (PBST) and then incubated in a working solution of Hoechst 33342 dye (5µg/mL) (Thermo 662 Scientific, USA) to stain cell nuclei for 10 min at room temperature. Subsequent washes in 663 PBST were followed by mounting the tissue onto Superfrost Plus slides and coverslipping using 664 SlowFade gold anti-fade reagent (Life Technologies, USA). All slides were stored in the dark at 665 4 o C until imaging. 666

DAB IHC image analysis 668
Immunoperoxidase-stained tissue sections were analyzed under bright field microscopy using 669 an Olympus BX53 upright microscope. Digital images were acquired using an Olympus DP72 670 12.5 Megapixel cooled digital color camera connected to a Dell computer running the Olympus 671 cellSens software platform. Image files were then coded and analyzed by an investigator 672 blinded to genotype and saved to a Dell Optiplex 7020 (79). 673 Semiquantitative analysis of MC1+ cell counts in EC and DG was performed in 674 horizontal brain sections from 16-month EC-Tau/hAPP mice (n=4) and EC-Tau mice (n=5) 675 ( Figure 1E Fiji using an over/under display mode (red represents pixels above threshold) (Supplemental 684 Figure 3A & 3F). A region of interest (ROI) was then defined within each image and saved via 685 the ROI Manager. The total immunoreactivity for each antibody marker above threshold was 686 saved as a percentage of total ROI area (mm 2 ) and used to compare pathological accumulation 687 of hAPP/Aβ and tau in the right versus left hemispheres. The same minimum threshold value 688 was applied for each pair of images (right vs left ROIs) used in our analysis. For all DAB IHC 689 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint experiments, three immunostained tissue sections were analyzed per mouse and averaged to 690 generate a representative value reflecting hAPP/Aβ and tau pathology. interface. BatchTINTV3 sorts the spike data of each session in a chosen directory using 714 KlustaKwik. The BatchTINTV3 code is freely available and hosted in the following GitHub 715 repository: https://github.com/hussainilab/BatchTINTV3. 716 BatchPowerSpectrum: The percent power values were calculated in MATLAB. A 717 Welch's power spectrum density (PSD) estimate of the LFP was calculated via the 'pwelch' 718 function. Using the PSD, the average powers in each of the desired frequency bands were 719 calculated with the 'bandpower' function. The average power of each band was then divided by 720 the total power of the signal to produce the percentage power in each of the bands. In situations 721 where the data was speed filtered (5-30 cm/sec), the speed of the animal was calculated and 722 then interpolated so there was a speed value for each LFP value. Then the LFP values were 723 chunked to contain consecutive data points where the mouse's movements satisfied the 724 minimum and maximum speed requirements. Chunks containing less than 1 second of data 725 were discarded. The aforementioned power calculations were then performed on each of the 726 LFP chunks, and an average of these chunks would yield the final percentage power values for 727 each of the frequency ranges. The BatchPowerSpectrum code is freely available and hosted in 728 the following GitHub repository: https://github.com/hussainilab/BatchPowerSpectrum. 729 hfoGUI for time-frequency analysis: A time-frequency representation (TFR) of the LFP 730 was visualized using a GUI written in Python called 'hfoGUI.py'. The GUI allows for complete 731 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint control of signal filtering and is equipped with various filter types (butterworth, chebyshev type 1, 732 chebyshev type 2, etc.), along with the ability to specify filter order, cutoffs, etc. We used 3rd 733 order butterworth filters in order to maintain consistency with the filter types and order 734 implemented by the Axona data acquisition software/hardware. Specific time windows of the 735 LFP data were selected and subjected to a Stockwell-Transform (s-transform) to visually 736 represent EC-Tau/hAPP and Control genotypes ( Figure 3A-B). The hfoGUI code is freely 737 available and hosted in the following GitHub repository: https://github.com/hussainilab/hfoGUI. 738 739

Statistical Analysis 740
Statistical analyses were performed in GraphPad Prism 7 and Matlab. All datasets were tested 741 for normality using the Shapiro-Wilk test. Datasets where values were not well modeled by a 742 normal distribution were subjected to non-parametric statistical analyses. Unpaired t-tests with 743 Welsh's correction were used to perform semiquantitative comparisons of tau pathology in 744  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint  Horizontal brain sections from EC-Tau/hAPP mice and age-matched EC-Tau littermates were processed for immunohistochemical detection of hAPP/Aβ (6E10) and abnormal, misfolded tau (MC1). Representative, adjacent brain sections from two mice sampled are shown for both 6E10 and MC1. A-B. 16-month EC-Tau/hAPP mice exhibited robust Aβ accumulation and plaque deposition throughout the EC and HIPP. Diffuse Aβ accumulation comprised the majority of the pathology in these regions, with occasional small, compact plaques and large, dense-core Aβ plaques present (arrows). EC-Tau mice did not exhibit 6E10+ immunoreactivity. Scale bars, 500µm. C-D. MC1+ immunostaining revealed a clear acceleration of tau pathology in the EC of EC-Tau/hAPP mice, characterized by an increased number of cells with abnormally conformed tau localized to the somatodendritic compartment. Scale bars, 250µm. E. A semiquantitative analysis of MC1+ cell counts was performed in the EC and DG. EC-Tau/hAPP brain sections exhibited over three-fold and over ten-fold greater MC1+ cell counts in EC and DG than in EC-Tau sections, respectively. Unpaired t-tests with Welch's correction: EC, p < 0.01; DG, p < 0.05. F. MC1+ immunostaining in EC-Tau/hAPP mice sampled at 10-, 16-, and 23-months of age confirmed the onset of pathological tau spread from the EC into the HIPP at 16-months. Graphs represent mean ± SEM for the averaged ROI values from three independently processed brain sections per mouse. * p < 0.05; ** p < 0.01.

Figure 2. Aβ-associated hyperactivity in EC single-units in vivo
In vivo multi-electrode recordings were performed in the EC of 16-month EC-Tau/hAPP mice (n=4) and their age-matched littermates: EC-Tau (n=4), hAPP (n=5), and non-transgenic control mice (n=7). A. MC1 immunostained horizontal brain section from a 16-month EC-Tau/hAPP mouse. Representative image depicts tetrode tract mark terminating in EC layer II. Scale bar, 500µm. B. Cumulative frequency distributions of the spontaneous, average firing rates of all single-units collected in EC-Tau/hAPP mice and age-matched littermates. Pooled EC-Tau/hAPP neurons and hAPP neurons each exhibit shifts in their distributions towards higher firing rates versus Control. Two-sample Kolmogorov-Smirnov test: (EC-Tau/hAPP vs Control: p < 0.001; hAPP vs Control: p < 0.001). Insert, histograms of mean average firing rates are shown for each genotype. The single-unit average firing rates of EC-Tau/hAPP mice and hAPP mice were nearly two-fold higher than Control. Kruskal-Wallis test: p < 0.001. C. Cumulative frequency distributions of the median inter-spike intervals (ISI) for pooled EC single-units. Two-sample Kolmogorov-Smirnov test: (EC-Tau/hAPP vs Control: p < 0.001; hAPP vs Control: p < 0.001). Insert, box-and-whisker plots depicting median and mean ISI values along with 10-90 percentile limits per genotype. EC single-unit ISI values were significantly decreased in EC-Tau/hAPP mice and hAPP mice compared to Control. Kruskal-Wallis test: p < 0.001. D. EC single-unit average firing rates were plotted as a function of waveform spike-width and color-coded based on whether they belonged to Aβ-generating mice (purple) or Non Aβ-generating mice (gray). A distinct population of hyperactive EC neurons were present in Aβ-generating mice, and were distributed along the x-axis (shorter to longer spike-widths). E. Frequency histogram of neuronal spike widths for all single-units; note the bimodal distribution. A cutoff of 300µm spike-width was used to delineate putative interneurons (NS, narrow-spiking cells; yellow) from putative excitatory neurons (WS, wide-spiking cells; blue) (broken arrow). F. hAPP NS cells exhibit increased average firing rates compared to Controls. Kruskal-Wallis test: p < 0.05. G. EC-Tau/hAPP WS cells and hAPP WS cells each exhibit increased average firing rates versus Control WS cells. Kruskal-Wallis test: p < 0.001. Bar graphs represent mean ± SEM. * p < 0.05; *** p < 0.001.

Figure 3. Aβ-associated impairment of EC network activity in vivo
Local field potentials (LFPs) were collected within the EC of 16-month EC-Tau/hAPP (n=8) mice and age-matched littermates (EC-Tau, n=5; hAPP, n=9; non-transgenic control, n=7). The percentage power values for oscillatory frequency bands were then calculated and compared across genotype. A-B. Representative, filtered LFP waveforms in the theta (4-12 Hz), low gamma (35-55 Hz), and high gamma (65-120 Hz) frequency ranges are shown for one Control mouse and one EC-Tau/hAPP mouse recording session, along with the LFP spectrograms. LFP signals show a clear disturbance in the theta rhythm of 16-month EC-Tau/hAPP mice. Timescale, 2000 ms. C. Quantification of % power values revealed a significant decrease in averaged % theta power in EC-Tau/hAPP (24.85 ± 3.55 %) mice and hAPP (27.88 ± 2.51 %) mice versus Control (43.50 ± 6.86 %) mice. One-way ANOVA test: p < 0.01; Dunn's multiple comparison test, p < 0.05. D-E. No differences in averaged % power spectrum values were detected across genotypes in the low gamma (p > 0.05) or high gamma (p > 0.05) frequency ranges. Bar graphs represent mean ± SEM. Transparent overlays represent individual % power spectrum values per mouse in each genotype. ** p < 0.01.

Figure 4. Aβ and tau pathology does not affect locomotor activity in an open field
Locomotor activity was assessed by analyzing the position data of each mouse during active exploration in an open field (recording session, 30 min). A. Representative trajectories are shown for one recording session per genotype. Scale bar, 20cm. B-D. The following parameters in the open field were analyzed and compared across genotype: the total distance traveled (m), the % arena coverage and average speed (cm/sec). No significant group differences were detected on any measure. One-way ANOVA tests: Total distance (m), F (3,25) = 1.540, p > 0.05; % of arena coverage, F (3,25) = 1.281, p > 0.05; average speed (cm/sec), F (3,25) = 1.632, p > 0.05. E-G. No significant group differences were detected on any dependent measure when analyzing position data in the first 5, 10 or 15 min of the recording sessions (One-way ANOVA tests; p > 0.05). Bar graphs represent mean ± SEM. Individual values (transparent overlays) are representative means from three averaged recording sessions per mouse.

Figure 6. Chronic hM4D i EC DREADDS activation reduces MC1+ staining along the EC-HIPP network
DAB-IHC was performed on horizontal brain sections from 16-month EC-Tau/hAPP (n=4) mice and age-matched, littermate EC-Tau (n=5) mice to evaluate the impact of chronic hM4D i EC DREADDs activation on abnormally conformed tau (MC1). MC1+ immunoreactivity patterns along the EC-HIPP network revealed a distinct separation between mice exhibiting early tau pathology, localized primarily within the EC but not yet in the HIPP, and mice exhibiting significant tau pathology in both the EC and HIPP. We refer to these mice as exhibiting Early Tau Pathology and Advanced Tau Pathology, respectively. A. EC-Tau, Early Tau pathology; F. EC-Tau/hAPP, Early Tau pathology; K. EC-Tau, Advanced Tau pathology; P. EC-Tau/hAPP, Advanced Tau pathology. Scale bars, 500µm. A-E. A subtle reduction in MC1+ neuropil staining could be detected in the right PaS and EC of an EC-Tau mouse. F-J. A clear reduction in MC1+ cell bodies within the right PaS and EC was seen in an EC-Tau/hAPP mouse. K-O. Reduced MC1+ somatic staining was clearly evident in the DG granule cell layer and CA1 pyramidal cell layer of an EC-Tau mouse with Advanced Tau pathology. P-T. Reduced MC1+ somatic staining was detected in PaS, EC and CA1 of an EC-Tau/hAPP mouse with Advanced Tau pathology. Yellow arrows indicate areas of reduced MC1+ staining. All images are representative for human tau-expressing mice that fall into Early Tau pathology and Advanced Tau pathology groupings. Magnified images: scale bars, 100µm. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint Adjacent tissue sections from hM4D i EC DREADDs activated mice were processed for AT8+ immunostaining experiments to examine the distribution of hyperphosphorylated tau expression along the EC-HIPP network. Horizontal brain sections from 16-month EC-Tau/hAPP (n=4) mice and age-matched, littermate EC-Tau (n=5) mice were processed in parallel. A. EC-Tau, Early Tau pathology; F. EC-Tau/hAPP, Early Tau pathology; K. EC-Tau, Advanced Tau pathology; P. EC-Tau/hAPP, Advanced Tau pathology. Scale bars, 500µm. B-E. Chronic hM4D i EC DREADDs activation resulted in reduced AT8+ neuronal staining in the right hemisphere PaS and EC in an EC-Tau mouse exhibiting Early Tau pathology. F-J. Reduced AT8+ immunostaining is shown in the rPaS and rEC of an EC-Tau/hAPP mouse exhibiting Early Tau pathology. K-O. Reduced somatic AT8+ immunostaining was evident in the rSub, rDG granule cell layer and rCA1 pyramidal cell layer of an EC-Tau mouse with Advanced Tau pathology. P-T. Reduced somatic AT8+ immunostaining was evident in the rPaS, rEC and rCA1 pyramidal cell layer of an EC-Tau/hAPP mouse exhibiting Advanced Tau pathology. Yellow arrows indicate areas of reduced AT8+ immunostaining. All images are representative for human tauexpressing mice that fall into Early Tau pathology and Advanced Tau pathology groupings. Magnified images: scale bars, 100µm.

Figure 8. Chronic hM4D i EC DREADDS activation selectively reduces tau pathology in downstream hippocampus
Semi-quantitative analysis of tau pathology expression in HIPP was performed on MC1+, AT8+ and CP27+ immunostained tissue sections of mice exhibiting advanced tau pathology. Withinsubject comparisons of tau immunoreactivity (right versus left hemisphere) were calculated per hippocampal region of interest (CA1, DG and Sub) and then plotted as a ratio to reveal any laterality in pathological tau accumulation. Chronic activation of hM4D i EC DREADDs in the right hemisphere was associated with reduced MC1+, AT8+ and CP27+ immunostaining in DG granule cells. DREADDs activation was also associated with reduced AT8+ and CP27+ immunostaining in CA1 pyramidal cells. Box plots depict the mean ± SEM of right versus left hemisphere % Area ratios for experimental mice exhibiting advanced tau pathology (Total, n=4 mice: EC-Tau, n=2; EC-Tau/hAPP, n=2). Black boxes, MC1; Gray, AT8; White, CP27. Transparent, colored overlays depict individual ratio values for each mouse. Hashed line (1.0 ratio) represents equal right versus left hemisphere tau immunoreactivity.
. CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint Table 1

. Single-unit totals and average firing rates per mouse
In vivo electrophysiological recording parameters are shown for individual mice included in the study. The total number of cells and their average firing rates are listed, as well the average firing rates and total numbers of narrow-spiking (NS) and wide-spiking (WS) cells per mouse. Finally, the averaged bursting activity % per mouse is listed.

Supplemental Figure 2. Acute EC DREADDs activation in AD mouse models
Single injections of CNO (i.p.) were used to determine salient recording measures of EC DREADDs activation in vivo. A-B. Long-term recordings were first performed in naïve 12-13 month EC-Tau mice to determine the onset and duration of altered single-unit activity due to EC DREADDs activation (hM3D q EC, n=1 mouse; hM4D i EC, n=1 mouse). Left, Total spike counts (normalized to baseline measures) are shown for each mouse following CNO and Saline administration (60 min post-treatment). EC DREADDs activation occurs ~20 min post-CNO compared to Saline and lasts at least 4 hr (right). Square (green), hM3D q EC DREADDs; Diamond (purple), hM4D i EC DREADDs; Circle (white), Saline. C. Sagittal sections from an EC-Tau mouse expressing hM3D q EC DREADDs is shown. DREADDs-mCherry expression is present along the dorsoventral axis of the EC. Tetrode tract is shown. Scale bars, 500µm. D. The effects of acute hM4D i EC DREADDs activation on LFP measures in EC are shown in a 12month hAPP/J20 mouse. LFP traces are shown for both CNO and Saline conditions. Notable . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint differences in EC theta modulation were present after 10mg/kg CNO treatment. E. To determine the long-term effects of CNO treatment on LFP measures, the magnitude difference in % theta, % low gamma and % high gamma power versus Saline were calculated. % theta power and % high gamma power in EC were robustly affected by CNO treatment for over an hour. TD Ratio is the normalized Treatment Difference ratio showing magnitude difference of CNO treatment vs Saline.

Supplemental Figure 3. Right vs left hemisphere ROI measures for DAB IHC analysis
The ipsilateral HIPP regions of interest (ROI) area (mm 2 ) downstream from the hM4D i DREADDs-expressing EC were compared to contralateral HIPP ROIs. Positive immunoreactivity for each marker was also analyzed within each defined ROI and compared across hemispheres as described in Methods. A. Top, 8-bit gray scale images of 6E10+ immunoreactivity in a horizontal brain section from a 16-month EC-Tau/hAPP mouse. A minimum threshold value was first applied to each image (186), and then the ROIs (lHIPP and rHIPP) were defined (magenta). Finally, the % area of 6E10+ immunoreactivity above threshold was used to quantify hAPP/Aβ accumulation in the right and left HIPP ROIs. Scale bars, 500µm. Bottom, Higher magnification of 6E10+ immunoreactivity within ROIs. Black pixels depict 6E10+ immunoreactivity above threshold on a white background. Scale bars, 250µm. B. Right vs left hippocampal ROI values (area, mm 2 ) for each tissue section analyzed in Figure 4G-H. The coefficient of determination (R 2 , 0.6027) is shown. C. The averaged ROI area values did not differ between the right hemispheres (1.731 ± 0.039 mm 2 ) and left hemispheres (1.737 ± 0.039 mm 2 ). Paired t test: t (6) = 0.1557, p > 0.05. Three sections averaged per mouse. EC-Tau/hAPP, n=4 mice; hAPP, n=3 mice. D. The averaged ROI area values did not differ between the right hemispheres (1.910 ± 0.040 mm 2 ) and the left hemispheres (1.899 ± 0.037 mm 2 ) in control conditions. Paired t test: t (6) = 0.5382, p > 0.05. E. 8-bit gray scale images of MC1+ immunoreactivity in the left and right dentate gyrus (DG) of a 16-month EC-Tau/hAPP mouse. A minimum threshold value (180) was applied and the granule cell layers were delineated in black. Within the ROIs, the % area of MC1+ immunoreactivity above threshold was quantified. Accompanying black and white over/under images are shown to emphasis MC1+ signal in DG ROIs. Scale bars, 100µm. F-H. The right versus left ROI values (mm 2 ) for hippocampal subregions analyzed were plotted. CA1: R 2 =0.8507; DG granule cells: R 2 =0.6529; Subiculum: R 2 =0.5711. Bar graphs represent mean ± SEM. Colored bar overlays depict the mean right versus left HIPP ROI area (mm 2 ) values for each mouse.

Supplemental Figure 4. hM4D i EC DREADDs expression in experimental mice
Regional mCherry (hM4D i DREADDS) and eGFP (Control) expression patterns were verified in all mice. For each group, illustrated mCherry and eGFP expression patterns were overlaid onto a mouse stereotaxic brain atlas. A. EC-Tau mice (n=5) exhibit robust hM4D i DREADDs expression throughout the rEC, rPaS and rPrS. mCherry signal was present in the rSub of two mice, and in terminal ends of EC projection neurons (outer molecular layers of rDG and rCA2) from one mouse. No mCherry crossover was detected in the left hemisphere. Widespread eGFP expression was detected in lEC and lPaS, as well as lSub, lCA1, and outer molecular layers of lDG. No crossover of eGFP expression into the right hemisphere was detected. B. mCherry and eGFP expression patterns are shown for three individual EC-Tau mice. C. hAPP . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/487405 doi: bioRxiv preprint mice (n=3) exhibit hM4D i DREADDs expression throughout the rMEC, rPaS and rPrS. DREADDs were expressed in the rSub of two mice, while mCherry signal could be seen in the terminal ends of EC projection neurons from one mouse. No crossover of mCherry expression was detected in the left hemisphere. eGFP signal was confined to the left hemisphere, with robust expression in the lMEC, lPaS, lPrS and lSub. eGFP signal was also detected in the outer molecular layers of lDG and in some lDG granule cells. D. mCherry and eGFP expression patterns are shown for all three hAPP mice. E. EC-Tau/hAPP mice (n=4) exhibit hM4D i DREADDs expression throughout the rEC, rPaS and rPrS. No crossover of mCherry expression was detected in the left hemisphere. eGFP expression was less regionally constrained in this group, and appeared in portions of the lMEC , lPaS, and lSub, as well as in the outer molecular layers of lDG, and in lCA1 (stratum radiatum and pyramidale). eGFP expression crossed over into the rPaS of one mouse. F. mCherry and eGFP expression patterns are shown for three individual EC-Tau/hAPP mice. G. mCherry and eGFP expression patterns were combined for all animals (Total, n=12 mice). H. Merged hM4D i DREADDs mCherry expression patterns are shown for 16-month EC-Tau/hAPP mice (n=4) chronically treated with low dose CNO (0.5mg -1 /kg -1 /day -1 ). Scale bars, 250µm.

Supplemental Figure 5. Chronic hM4D i EC DREADDS activation reduces total human tau along the EC-HIPP network
Adjacent tissue sections from hM4D i EC DREADDs activated mice were processed for CP27+ immunostaining experiments to examine the distribution of total human tau expression along the EC-HIPP network. Horizontal brain sections from 16-month EC-Tau/hAPP (n=4) mice and agematched, littermate EC-Tau (n=5) mice were processed in parallel. A. EC-Tau, Early Tau pathology; F. EC-Tau/hAPP, Early Tau pathology; K. EC-Tau, Advanced Tau pathology; P. EC-Tau/hAPP, Advanced Tau pathology. Scale bars, 500µm. B-E. An equal distribution pattern of CP27+ immunostaining was detected in PaS cell somas and EC neuropil in an EC-Tau mouse exhibiting Early Tau pathology. F-J. Reduced CP27+ immunostaining was detected in the rPaS and rEC of an EC-Tau/hAPP mouse with Early Tau pathology. K-O. Reduced CP27+ staining was evident in the rDG granule cell layer and rCA1 pyramidal cell layer in an EC-Tau mouse exhibiting Advanced Tau pathology. P-T. Reduced CP27+ immunostaining was evident in the rPaS, rEC, rDG granule cell layer and rCA1 pyramidal cell layer in an EC-Tau/hAPP mouse with Advanced Tau pathology. Yellow arrows indicate areas of reduced AT8+ immunostaining. All images are representative for human tau-expressing mice that fall into Early Tau pathology and Advanced Tau pathology groupings. Magnified images: scale bars, 100µm.