Amyloid pathology reduces dynamic range and disrupts neural coding in a mouse model of Alzheimer’s Disease

Amyloid pathology reduces dynamic range and disrupts neural coding in a mouse model of Alzheimer’s Disease | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Amyloid pathology reduces dynamic range and disrupts neural coding in a mouse model of Alzheimer’s Disease Mary Ann Go , Kathrine E. Clarke , Yimei Li , Seigfred V. Prado , Beatriz R. F. Teixeira , Jess J. Yu , Simon R. Schultz doi: https://doi.org/10.1101/2025.06.27.661987 Mary Ann Go a Centre for Neurotechnology and Department of Bioengineering , Imperial College London, South Kensington, London, SW7 2AZ, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Kathrine E. Clarke b Department of Biomedical Engineering, University of Melbourne , Melbourne 3010, Australia Find this author on Google Scholar Find this author on PubMed Search for this author on this site Yimei Li a Centre for Neurotechnology and Department of Bioengineering , Imperial College London, South Kensington, London, SW7 2AZ, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Seigfred V. Prado a Centre for Neurotechnology and Department of Bioengineering , Imperial College London, South Kensington, London, SW7 2AZ, UK c Department of Electronics Engineering, University of Santo Tomas , Manila, Philippines Find this author on Google Scholar Find this author on PubMed Search for this author on this site Beatriz R. F. Teixeira a Centre for Neurotechnology and Department of Bioengineering , Imperial College London, South Kensington, London, SW7 2AZ, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jess J. Yu a Centre for Neurotechnology and Department of Bioengineering , Imperial College London, South Kensington, London, SW7 2AZ, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Simon R. Schultz a Centre for Neurotechnology and Department of Bioengineering , Imperial College London, South Kensington, London, SW7 2AZ, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: s.schultz{at}imperial.ac.uk Abstract Full Text Info/History Metrics Preview PDF Abstract Alzheimer’s Disease (AD) disrupts neural circuits vital for memory and cognition. We used two-photon microscopy to investigate these disruptions in behaving mice, focusing on the link between amyloid plaques - a hallmark of AD - and aberrant neural activity. Using the 5xFAD mouse model, we observed significant changes in hippocampal neurons, including elevated baseline activity and reduced locomotion-driven firing, leading to a diminished neuronal dynamic range. These abnormalities were more pronounced near amyloid plaques. We also found degraded spatial coding, reduced synchrony, and increased variability in neuronal responses. Furthermore, place fields emerged more slowly in both familiar and novel environments, indicative of recall and learning impairments respectively. By showing a specific link between plaque vicinity and neural coding deficits including reduced dynamic range in mice performing spatial tasks, our study offers new insights into the circuit basis of progressive cognitive degradation in AD. 1 Introduction Impairment of memory and cognition is a prominent symptom of Alzheimer’s Disease (AD), with spatial and episodic memory processes particularly affected. While substantial advances have been made in characterizing the molecular neuropathological correlates of AD ( Selkoe, 2001 ), therapeutic progress depends crucially on our understanding how these pathological signatures lead to alterations in the ability of neural circuits to process, store and recall information. A broad range of evidence suggests that the amyloid beta protein (Aβ), a hydrophobic peptide which tends to assemble into long-lived oligomers and polymers, plays an important role in the pathology, both in the earlier stages of the disease in which soluble Aβ levels are elevated, and in later stages of disease progression involving the formation of amyloid plaques ( Busche et al., 2012 ). Historically, the prevailing view was of solely the silencing of neuronal circuits through synaptic dysfunction ( Selkoe, 2002 ). In more recent years, however, it has become clear that the effects on cortical circuits are more complex, with neuronal hyperactivity in the vicinity of plaques ( Busche et al., 2008 ), in addition to a broader hypoactivity, with these together being referred to as aberrant excitability ( Palop et al., 2007 ; Palop and Mucke, 2016 ). While it is clear that synaptic silencing results in disrupted information processing, disruption of the balance of excitation and inhibition in hippocampal and cortical circuits may also directly affect circuit function, providing a primary mechanism for cognitive deficits. In support of this, patients with sporadic AD have been found to have an increased incidence of seizures, and familial early-onset AD is associated with more prominent epilepsy ( Palop and Mucke, 2009 ). It is therefore important that we understand how aberrant excitability affects neural circuit function and leads to cognitive deficits. The hippocampal CA1 circuit provides a good model system to study this problem, as the hippocampus is important for memory function in both humans and rodents ( Squire, 2004 ), it is one of the earliest locations showing amyloid pathology ( Hyman et al., 1984 ), and the CA1 subfield contains many “place cells” cells tuned for a particular spatial location in an environment ( O’Keefe and Dostrovsky, 1971 ), and which remap their spatial tuning in distinct environments ( Muller and Kubie, 1987 ), providing a circuit mechanism for spatial memory. Hippocampal place cells have historically been studied in behaving mice using extracellular electrophysiology ( O’Keefe and Dostrovsky, 1971 ; Wilson and McNaughton, 1993 ), and more recently using head-mounted miniature single-photon fluorescence microscopes ( Ziv et al., 2013 ; Gonzalez et al., 2019 ). However, neither of these techniques provide accurate three-dimensional characterisation of the relative locations of amyloid plaques and recorded neurons, the gold standard for which is provided by in vivo two-photon microscopy. Busche et al. (2012) used two-photon microscopy to image neurons and amyloid plaques in the hippocampus of mice over-expressing both mutant human amyloid precursor protein (APP) and mutant human presenilin 1 (PS1), finding that hyperactive neurons were located exclusively in the vicinity of plaques, with silent and normal neurons distributed throughout the hippocampus, qualitatively similar to what was previously observed in cortex ( Busche et al., 2008 ), and that in young APP/PS1 mice there was an increase in hyperactivity even before the formation of plaques. However, these results were obtained in the anaesthetized mouse, leaving open the question of how they relate to neuronal activity during performance of a spatial memory task. This motivates the use of two-photon microscopy to track neuronal excitability and information processing in the neighbourhood of plaques in the behaving mouse. Head-fixation is currently required for high quality, stable two-photon imaging in behaving animals (although this may change in the near future; see Zong et al., 2022 ). Two-photon calcium imaging has been used to map place fields in head-fixed mice navigating a virtual reality (VR) environment ( Hainmueller and Bartos, 2018 ). The use of VR environments ( Harvey et al., 2009 ) using only distal visual cues to study spatial cognition in rodents has been criticised ( Aghajan et al., 2013 ), however the impaired spatial selectivity in head-fixed animals has been largely ameliorated by the development of “floating cage” environments involving multi-modal, proximal cues ( Go et al., 2021 ). In vivo two-photon microscopy thus now provides us with the ability to study the how the storage and retrieval of cellular contributions to spatial memory is affected by the three-dimensional distribution of amyloid plaques in the vicinity of individual neurons. In this study, we examined the relationship between aberrant excitability, the three-dimensional spatial proximity of amyloid plaques, and spatial memory as read out from the hippocampal CA1 place cell network. We used the 5xFAD mouse model, a well-established model of amyloidosis with five human familial AD mutations, and in which high levels of intracellular Aβ begin to accumulate at 1.5 months of age, with plaque deposition beginning around 2 months, and being widespread in the cortex and hippocampus by 6 months ( Oakley et al., 2006 ). We investigated the progression of spatial information coding deficits during maturation and aging of 5xFAD mice, and the extent to which disruptions were synchronous across the circuit, as opposed to affecting cells independently, and how these deficits affected learning of new spatial memories versus the recalling of old. 2. Results To study how amyloid plaques influence the relationship between neural circuit activity and spatial memory, we used a floating real-world environment behavioural apparatus ( Go et al., 2021 ) together with a multi-photon microscope to image the activity of populations of hippocampal CA1 neurons in head-fixed mice trained to run around a circular track ( Figure 1A ) decorated with patterned visuotactile cues. Hippocampal neurons in 5xFAD and wild-type (WT) mice were labeled with GCaMP6s or jGCaMP7s calcium sensors ( Figure 1B ). We studied two age groups, the first based on the age at which accumulation of Aβ starts, and the second driven by the age at which plaque deposition becomes widespread ( Oakley et al., 2006 ). Mice in the young group were 2.1 to 3.0 months old at the time of the first imaging session (median 2.7), and mice in the older group were 6.4 to 10.1 months of age (median 7.5). Mice were trained to navigate the circular track for 6-7 days and then imaged. The proximity of amyloid plaques to each neuron was mapped approximately one week after the first neuronal activity imaging session (see Methods for full schedule). Neural activity was extracted using non-negative matrix factorization ( Figure 1C ; see STAR Methods) and by deconvolving trains of events (with amplitude) using the OASIS algorithm ( Friedrich et al., 2017a ). To compensate for possible differences in signal to noise ratio, and accordingly OASIS event amplitude, we fit the distribution of event amplitudes for each field of view (FOV) in each mouse to a lognormal distribution ( Figure S1 ), and standardized each distribution to unit mean. Activity rates are calculated by accumulating the amplitudes of events within a certain time window, and dividing by window duration - we thus report them in the account that follows in arbitrary units per minute (arbs/min) - but importantly, the same , cross-calibrated, arbitrary units are used across the entire dataset, facilitating direct comparison between mice and fields of view. Download figure Open in new tab Figure 1. Elevated baseline activity and suppressed firing during locomotion in old 5xFAD mice. (A) Behavioural apparatus: head-fixed mice traverse a floating circular track. (B)( Left ) Typical FOV showing neurons (green, jGCaMP7s) overlaid with the amyloid plaques (magenta, Methoxy-X04) within 100 vertical microns of the calcium imaging plane. This example from a 7.8-month old 5xFAD mouse. ( Right ) 388 segmented ROIs from image on left. (C)( Top ) Linearized position of mouse along outer circumference of track; ( Bottom ) Ca 2+ transients for 15 (of 388) cells from B. (D)Standardized calcium activity rate (see STAR Methods) distributions for 5xFAD and WT mice during resting (speed 2 cm/s) periods. Inverted triangles denote median. Group color codes as in E. Resting, young WT: 4.04± 0.05 arbs/min (mean SEM); young 5xFAD: 4.51± 0.06 arbs/min; old WT: 3.52± 0.07 arbs/min; old 5xFAD: 4.16 ±0.10 arbs/min. Running, young WT: 6.17 ±0.08 arbs/min; young 5xFAD: 6.63± 0.10 arbs/min; old WT: 7.72 ±0.12 arbs/min; old 5xFAD: 5.63 0.05 arbs/min. For D-E, young WT: n = 5,168 cells pooled from 6 animals, 33 sessions; young 5xFAD: n = 4,925 cells, 6 animals, 27 sessions; old WT: n = 3,475 cells, 8 animals, 27 sessions; old 5xFAD: n = 7,332 cells, 6 animals, 31 sessions.3.52 ±0.07 arbs/min; old 5xFAD: 4.16 ±0.10 arbs/min. (E)Cumulative histograms of the distributions in D. (F)Average running speed during the first 50 laps of each session. Each data point indicates one running session. Boxplot edges denote quartiles while black bars denote median. Young WT: 4.81 ±0.31 cm/s (n = 44 sessions pooled from 8 animals); young 5xFAD: 4.84± 0.33 cm/s (n = 32 sessions, 6 animals); old WT: 4.78± 0.61 cm/s (n = 27 sessions, 8 animals); old 5xFAD: 8.14± 0.90 cm/s (n = 31 sessions, 6 animals). (G)Time spent (as a fraction of recording time) at each speed. See also Figures S1 and S2. All statistics were performed with hierarchical bootstrap analysis (see STAR Methods) which provides a direct measure of a hypothesis being true in contrast to other statistical tests which calculate the probability of the null hypothesis being true. The probability that the WT mean is greater than the 5xFAD mean is denoted by P (WT 5xFAD) and is significant if P > (1 −α/2), where α is the significance level. P (WT 5xFAD) 0.975 or P 0.995 or P 0.9975 or P < 0.0025 (α = 0.005). 2.1 Elevated CA1 baseline firing and greater running speed in old 5xFAD mice As hippocampal CA1 neurons have previously been reported to be modulated by animal speed as well as location ( Góis and Tort, 2018 ), and given that hippocampal activity at rest shows different features to activity during locomotion for instance, with place cells in the network displaying spontaneous replay of past trajectories ( Ólafsdóttir et al., 2018 ), it is important to separate out the observed effects on neuronal activity during stationary and locomoting states. During rest (i.e. baseline, which we define as periods in which speed of movement did not exceed 0.5 cm/s), neural activity was higher in old AD mice compared to old WT mice, as shown in Fig. 1D,E (see also Fig. S2 ). This baseline hyperactivity was not evident in the young group. It is notable that in old AD mice, there was a mode at zero, suggesting an increase in the number of cells in near-complete silence. There was also a significant increase in activity rate with age in both WT and AD mice. We wanted to know whether the shift in baseline neural activity was also present during locomotion, during which we expect the activities of CA1 neurons to be elevated due to both place selectivity (when the mouse is passing through a cell’s place field) and speed tuning. To examine this, we analysed neural activity during periods when the mice were running (periods in which speed exceeded 2 cm/s). Activity rates were found to be higher during the running periods compared to the resting periods for all groups ( Fig. 1D,E , right panels; see also Fig. S2 ). Similar to the resting phase, neuronal activity during locomotion in young 5xFAD mice did not show a strong difference to the young WT group. However, there was significantly lower neuronal activity in old AD mice compared to old WT mice. Notably, the activity rate distribution was narrower, with a shift toward higher rates at the low end of the distribution in 5xFAD mice, but fewer very high activity rates observed, resulting in a diminished range of activity. Similar to the resting phase, there was an increase in activity rate with age for both genotypes. Given the strong differences in neuronal activity distributions observed during running, and that 5xFAD mice have been previously found to show some motor deficits by 9 months of age ( Jawhar et al., 2012 ), we thought it important to look at running speed. We found that old 5xFAD mice, on average, ran faster than old WT mice ( Fig. 1F ), while the young groups had comparable speeds. In addition, old 5xFAD mice spent more time moving at higher speeds ( Fig. 1G ). The increase in running speed with age observed in the 5xFAD mice was not apparent in the WT mice. The old 5xFAD mice were thus somewhat more behaviorally hyperactive. 2.2 Reduced dynamic range of activity in 5xFAD mice during locomotion The increased baseline activity and suppressed firing during locomotion we observed in old AD mice is suggestive of a reduction in the dynamic range of neuronal firing, which has important consequences for information coding. To examine this further in the context of locomotion, we characterized the average level of neuronal activity at each speed, producing a speed tuning curve for each cell, as well as an average speed tuning curve for the population firing rate. The dynamic range of the population firing rate is given by the ratio of the maximum and non-zero minimum measurable values. Note that excursions in activity above baseline due to running do not simply reflect speed tuning of the individual neuron, but also that the mouse runs through the place field, for cells which have one. We will refer to this as the speed related dynamic range. The population speed tuning curve for old AD mice began at a higher baseline but was flattened compared to the other groups ( Fig. 2A ), indicating a reduction in dynamic range for speed at population level. To examine this at single cell level, we then measured the dynamic range for each cell from its individual speed tuning curve, finding speed related dynamic range to be significantly lower across the population of cells in old AD mice compared to old WT mice ( Fig. 2B ) but comparable in the young groups. Download figure Open in new tab Figure 2. Old 5xFAD mice show reduced neuronal dynamic range during locomotion. (A)Population average speed tuning curves. Group color codes as in B. (B)Violin plots of dynamic range of neuronal activity during loco- motion, defined as the ratio of the maximum and non-zero minimum points on the speed tuning curve for each cell. Plots show mean (white solid line), median (white circle) and interquartile range (gray vertical line). Young WT: 14.99 ±0.51; young 5xFAD: 13.37 ±0.29; old WT: 19.60± 4.87; old 5xFAD: 11.16± 0.13. * P (WT 5xFAD) = 0.982. (C)Proportion of cells in each group showing significant speed modulation. Each data point indicates a session. Young WT: 0.19 ±0.02; young 5xFAD: 0.15 ±0.01; old WT: 0.18 ±0.02; old 5xFAD: 0.12± 0.01. + P (WT≥ 5xFAD) = 0.968. (D)Temporal shift in instantaneous firing rate that maximizes correlation between instantaneous firing rate and running speed in speed- modulated cells. Young WT: 92.0 ± 9.0 ms; young 5xFAD: 63.0 ± 10 ms; old WT: 102± 11 ms; old 5xFAD: 27.0± 11 ms. 5xFAD) = 0.024. * P (WT > See also Figure S3. All statistics were performed with hierarchical bootstrap analysis. For all panels, young WT: n = 5,310 cells pooled from 6 animals, 33 sessions; young 5xFAD: n = 4,955 cells, 6 animals, 27 sessions; old WT: n = 3,517 cells, 8 animals, 27 sessions; old 5xFAD: n = 7,400 cells, 6 animals, 31 sessions. To examine speed tuning further, we looked at whether the fraction of cells modulated by speed was affected by age or disease phenotype. We found a significantly reduced fraction of cells modulated by speed in old AD mice compared to old WT mice ( Fig. 2C ). We observed three different types of modulation profiles: cells whose firing rate was positively or negatively modulated by speed, and cells whose firing rate was maximum at a “preferred” speed less than 70% of the maximum animal speed ( Fig. S3A,B ). For all groups, the number of positively modulated speed cells was the highest of the three profiles, while the number of negatively modulated speed cells was the lowest ( Fig. S3C ). Notably, the number of negatively modulated speed cells in old AD mice was higher compared to old WT mice ( Fig. S3C ). All other fractions were comparable. It has been reported that hippocampal speed cells code for speed retrospectively, as opposed to median entorhinal cortex (MEC) speed cells which code for speed prospectively (Kropff et al., 2015). To verify this, we calculated the correlation between the instantaneous firing rate and instantaneous animal speed for temporal shifts in the instantaneous firing rate spanning -500 ms to 500 ms. We then identified the temporal shift for which the correlation was maximum. We were able to replicate the Kropff et al. (2015) result, that hippocampal speed cells code for past speed. Moreover, the time interval between the instantaneous firing rate and the past speed was significantly lower in old 5xFAD mice compared to old WT mice ( Fig. 2D ) but not significantly different in the young groups. This indicates that old 5xFAD speed cells code for speed in the more recent past, ie that the hippocampal memory of speed is more limited. 2.3 Neuronal circuit alterations begin in the vicinity of amyloid plaques A particular advantage of our in vivo two-photon microscopy approach is the ability to visualize the locations of plaques in the brains in real time, and relate them to the properties of nearby neurons. We therefore next examined the relationship between the neuronal circuit alterations we observed (such as elevated baseline firing and reduced dynamic range) and the three-dimensional (3D) spatial distribution of amyloid plaques ( Fig. 3A ). We averaged the quantity of interest, e.g. activity rate, for all cells at a certain distance from a plaque, then averaged over all plaques. We found that baseline neuronal activity rate in both young and old 5xFAD mice was elevated relative to the mean resting activity rate of WT cells, regardless of proximity to plaques, but the elevation was especially pronounced in the vicinity of plaques (< 20 µm, Fig. 3B ). In contrast, during running, neuronal activity rate was elevated in young 5xFAD cells except in the vicinity of plaques (< 20 µm, Fig. 3C ). In old 5xFAD mice, neuronal activity during running was lower than the mean activity rate for old WT mice, regardless of distance to plaques. The dynamic range of activity for cells in young AD mice was below the mean (of cells in WT mice) far away from plaques (> 100 µm, Fig. 3D ), whereas for old AD mice the dynamic range was lower regardless of plaque distance, but particularly so close to plaques. Download figure Open in new tab Figure 3. Aberrant excitability and reduced dynamic range of activity near plaques in 5xFAD mice. (A)(i)Three-dimensional volume showing the location of the 490 × 490 µm 2 two-photon jGCaMP7s imaging FOV. (ii) Same 3D field as (i) overlaid with the locations of Methoxy-X04-stained amyloid plaques in a 1100 × 1100 × 300 µm 3 volume relative to jGCaMP7s FOV. (iii) Measurements, e.g. activity rate, are averaged over all cells at a distance d from a plaque, shown schematically, then averaged over all plaques. (B)Average neuronal activity versus distance from plaques for young and old 5xFAD mice during rest (±SEM). Dashed lines in (B-D) denote the average for the corresponding WT group of the same age. Young 5xFAD: n = 1479 cells pooled from 6 mice; old 5xFAD: n = 1365 cells, 4 mice. For WT means, young WT: 5168 cells, 6 mice; old WT: 3475 cells, 8 mice. (C)Average neuronal activity versus distance from plaques during running. (D)Average dynamic range of neuronal activity during locomotion versus distance from plaques. These results on the whole suggest that neuronal circuit alterations begin in the vicinity of amyloid plaques in young AD mice, and are fully manifested in old animals. 2.4 Aberrant neuronal synchrony in 5xFAD mice Network hypersynchrony in the cortex and hippocampus has been reported in other AD mouse models prior to plaque deposition, through cortical electroencephalogram (EEG) recordings, mapping of chronic hippocampal seizure markers, and resting-state functional magnetic resonance imaging (fMRI) ( Verret et al., 2012 ; Bezzina et al., 2015 ; Shah et al., 2016 ). In later stages of the disease, hyposynchrony in resting-state fMRI BOLD signals has also been reported ( Shah et al., 2016 ). We therefore wanted to know if synchrony is altered in the 5xFAD mouse model. To quantify network synchrony, we calculated the Pearson cross-correlation coefficient of neural activity at zero time lag, averaged over all pairs of neurons in each imaging field of view. When the animal was quiescent, we observed greater synchrony between pairs of cells in young 5xFAD mice compared to that in WT mice ( Fig. 4 ). No significant difference was observed in synchrony in old 5xFAD mice in the resting state. During locomotion, however, the hypersynchrony in young 5xFAD mice disappeared, while significant hypo synchrony was observed in old 5xFAD mice. Download figure Open in new tab Figure 4. Aberrant synchrony in 5xFAD mice Synchrony, as measured by the mean pairwise Pearson correlation coefficient between neural activities at zero time lag, for the different groups during resting ( left ) and running ( right ) periods. Each data point indicates average of all pairwise correlations in an imaging session. Resting, young WT: (8.9 ± 0.7) × 10 −4 (global mean ± SEM); young 5xFAD: (1.6 ± 0.3) × 10 −3 ; old WT: (3.1 ± 0.6) × 10 −3 ; old 5xFAD: (1.4 ± 0.8) × 10 −3 . Running, young WT: (9.5 ± 0.2) × 10 4 ; young 5xFAD: (1.5 ± 0.5) × 10 −3 ; old WT: (3.5 ± 0.9) × 10 −3 ; old 5xFAD: (1.1 ± 0.1) × 10 −3 . Young WT: n = 33 sessions pooled from 6 animals; young 5xFAD: n = 27 sessions, 6 animals; old WT: n = 27 sessions, 8 animals; old 5xFAD: n = 31 sessions, 6 animals. * P (WT > 5xFAD) = 0.009, ** P = 0.996, *** P < 0.002, hierarchical bootstrap test. 2.5 Degraded spatial coding in 5xFAD mice Spatial navigation deficits are an important preclinical marker for AD ( Coughlan et al., 2018 ). While navigation performance can be measured behaviourally, a more direct way to quantify spatial memory performance, independent of behavioural task, is to measure the spatial information rates of hippocampal CA1 place cells ( Skaggs et al., 1992 ; Langston et al., 2010 ). In previous work, a decrease in spatial information content in hippocampal place cells was found to correlate with plaque load and reduction in behavioural performance ( Cacucci et al., 2008 ; Jun et al., 2020 ). Previous information analyses of the 5xFAD mouse model, however, have yielded conflicting results ( Zhang et al., 2023 ; Prince et al., 2021 ). This discrepancy might relate to differing measurement modality (electrophysiology versus calcium imaging), or to different normalization (bits per unitary calcium event versus bits per spike). We note that the majority of information analyses in the hippocampal literature use the Skaggs spatial information, which is a first order approximation to the mutual information ( Skaggs et al., 1992 , 1996 ). This approximation, while very easy to estimate, depends only upon the mean instantaneous firing rate map, and is not affected by trialto-trial reliability of spiking or correlations between spikes ( Panzeri et al., 1999 ; Schultz and Panzeri, 2001 ). In view of potential effects of amyloid on either response trial-to-trial variability or autocorrelation, we here calculate the full spatial mutual information rate (see Methods). To facilitate qualitative comparison with bits/spike measures reported in the literature, we normalise by the accumulated standardized neural activity throughout the time window, which is the quantity most comparable to firing rate (as opposed to the number of calcium events, which is less directly related to firing rate). It should be noted that absolute values should not be compared between studies with different measurement technologies. We found that spatial mutual information rates were highest in young WT mice compared to the other groups ( Fig. 5A ). The empirical cumulative density plot showed an apparently substantially leftwards (lower) shift in spatial information values for old 5xFAD mice, however this did not reach statistical significance in the hierarchical bootstrap analysis. Nevertheless, the trend towards poorer spatial information coding in old 5xFAD mice is clear, and consistent with other data (see following). Download figure Open in new tab Figure 5. (previous page) : Degraded spatial coding in 5xFAD mice (A)(i) Neuronal activity maps for two representative cells with dif- ferent spatial information values (shown above map). (ii) Cumu- lative histograms and violin plots (inset) of information comparing WT with 5xFAD for young and old mice. Young WT: 1.56 ±0.05 bits/min; young 5xFAD: 1.32 ±0.01 bits/min; old WT: 1.26± 0.11 bits/min; old 5xFAD: 0.90 0.021 bits/min. (B) (i) Calcium transient location (black dots) overlaid on mouse spatial trajectory (gray line) for two representative cells with differ- ent spatial coherence values. (ii) Cumulative histograms and violin plots (inset) comparing average coherence of WT with ±5xFAD young and old mice. Young WT: 0.769± 0.004; young 5xFAD: 0.807 ±0.005; old WT: 0.805 0.005; old 5xFAD: 0.661 0.003. (C) (i) Place tuning curve for a representative cell, this one a place cell, allowing calculation of spatial response amplitude (in this ex- ample 96.1 arbs/min), defined as the maximum minus the minimum activity across the spatial tuning curve. (ii) Old 5xFAD neurons tend to have lower spatial response amplitude than old WT neurons. Young WT: 20.5 ±0.3 arbs/min; young 5xFAD: 21.9± 0.3 arbs/min; old WT: 25.1 ±0.4 arbs/min; old 5xFAD: 18.2± 0.2 arbs/min. (D)(i) Variance of activity rate in each 2-cm spatial bin plotted against the mean activity rate in that bin, for a representative cell. This plot is double-logarithmic, and the slope of the linear fit (to- tal least-squares regression fit, dash-dot line) indicates the exponent (β) in the relationship between the variability and mean. In this example, β = 1.86. (ii) Cumulative histograms and violin plots (insets) of the slope (β, young WT: 1.89± 2.42 ×10 −3 ; young 5xFAD: 1.87± 2.14 ×10 −3 ; old WT: 1.86 ±2.75 ×10 −3 ; old 5xFAD: 1.88 ±2.22× 10 −3 ) and y-intercept (α, young WT: 3.65 ±5.18 ×10 −3 ; young 5xFAD: 3.63± 4.38 ×10 −3 ; old WT: 3.61± 5.96 ×10 −3 ; old 5xFAD: 3.8 ±5.73 ×10 −3 ). (E)Proportion of cells that meet place cell criteria in each group. Young WT: 29.6 ±2.2%; young 5xFAD: 30.2 ±1.7%; old WT: 31.0 ±3.2%; old 5xFAD: 17.3 ±2.4%. (F)Place field size cumulative histograms (staircase due to 2 cm spatial binning) and violin plots (inset). Young WT: 16.2 ±0.2 cm; young 5xFAD: 16.4 0.2 cm; old WT: 15.9± 0.2 cm; old 5xFAD: 12.9± 0.2 cm. (G)Average mutual information (i), coherence (ii), dynamic range of activity in response to location (iii), variability exponent and y- intercept (iv) and place field size (v) as a function of distance from plaque. Error bars show SEM. All statistics were performed with hierarchical bootstrap analysis. For A-D, young WT: n = 5082 cells pooled from 6 animals, 31 ses- sions; young 5xFAD: 4955 cells, 6 animals, 27 sessions; old WT: 3268 cells, 8 animals, 24 sessions; old 5xFAD: 7400 cells, 6 animals, 31 sessions. For E-F, young WT: n = 1470 place cells, 6 animals, 31 sessions; young 5xFAD: n = 1428 place cells, 6 animals, 27 sessions; old WT: n = 1101 place cells, 8 animals, 24 sessions; old 5xFAD: n = 1162 place cells, 6 animals, 31 sessions. + P (WT > 5xFAD) = 0.967, *0.975. P 0.995 or P . 0.005, *** P = 0.002 . For a comprehensive picture of the quality of spatial coding, we looked at additional, complementary measures of spatial tuning. Spatial coherence is a measure of how contiguous in space the responses of a neuron are. We found spatial coherence to be similar in the young groups, and in old WT mice, but significantly lower in old 5xFAD mice ( Fig. 5B ). To further examine effects on dynamic range as spatial location is varied, we measured the amplitude of place tuning, which we define as the maximum value of the place tuning curve minus the minimum, and term spatial response modulation depth. This was similar in the young groups but significantly decreased in old 5xFAD mice ( Fig. 5C ). The fidelity of information transmission by neurons is limited by the variability of their responses from trial to trial i.e. by how much their activity varies when they revisit the exact circumstances in which the information they signal should be identical. Cortical responses typically show variable responses to each repeat of the same stimulus parameters, with a key property being that stimuli that induce higher firing rate also inducing higher variability ( Tolhurst et al., 1983 ). We have observed a similar result for hippocampal CA1 neurons in a prior study ( Go et al., 2021 ). This variability is typically greater than that predicted by a Poisson process, a commonly used statistical model of spike trains, however becoming closer to that of a Poisson process when as many external variability sources as possible are taken into account ( Gur et al., 1997 ). As response variability is likely driven by a combination of network properties and local amplification ( Carandini, 2004 ), we conjectured that it may be particularly sensitive to disruptions of neural circuitry induced by Alzheimer pathology. We plotted the variance of the activity rates observed when the mouse is in each spatial bin around the track against the mean for that bin over laps (see example in Fig. 5D i), examining y-intercept and slope (variability exponent) parameters (see Methods). Young AD mice showed higher variability exponents overall than young WT mice ( Fig. 5D ), while old AD mice showed greatly increased variability y-intercepts in comparison to their WT controls. Finally, we quantified the number of place cells for the different groups and found a smaller fraction of place cells in the old 5xFAD group in comparison to the old WT group ( Fig. 5E ). We also measured the place field size and found a substantially reduced place field size in the old 5xFAD group compared to the old WT group ( Fig. 5F ). Although this is contrary to what has been reported for some other AD mouse models ( Takamura et al., 2021 ; Jun et al., 2020 ), it is consistent with what has been observed for the 5xFAD mouse model ( Zhang et al., 2023 ). Altogether, these complementary spatial coding measures point to a degradation in spatial tuning in the 5xFAD model, apparent in the old animals according to the specific measure. Finally, we examined the relationship between these degraded spatial tuning parameters and the 3D spatial distribution of plaques. Spatial mutual information in old 5xFAD mice was lower than in old WT mice regardless of distance from plaques ( Fig. 5Gi ). Coherence was lower in old 5xFAD mice compared to their WT controls except near plaques (< 60um, Fig. 5Gii ). Spatial response modulation depth was lower in old 5xFAD mice compared to old WT mice regardless of plaque distance ( Fig. 5Giii ). Variability exponents were comparable between old WT and 5xFAD mice while variability y-intercepts were higher in old 5xFAD mice compared to their WT controls regardless of plaque distance ( Fig. 5Giv ). Finally, place field size was smaller in old 5xFAD mice in comparison to old WT mice regardless of plaque distance. The degraded spatial tuning parameters, with the exception of coherence, were observed everywhere, suggesting a widespread disruption of spatial tuning. 2.6 Delayed place tuning in 5xFAD mice To determine whether AD impacts encoding and recall differently, we imaged mice while navigating a familiar track (FAM) then moved it to a novel track (NOV) in the same recording session. The familiar track was used in behavior training while the novel track was introduced on imaging day. These tracks differed in visual and tactile cues on the walls and floor. Only sessions for which mice ran 40-50 laps in each environment were included in the analysis. To investigate the intra-session stability of place fields in an environment, we divided each session into two equal segments, and calculated the correlation between the place fields obtained separately for the two halves. We found that young WT and AD place fields had comparable intrasession stability in the familiar environment, and that place fields were significantly less stable in both groups in the novel compared to the familiar environment ( Fig. 6A ). For old mice, in the familiar environment, AD place fields were less stable compared to WT place fields. In the novel environment, old WT place fields were less stable compared to the familiar environment. This was also true for old AD place fields, which were even less stable again ( Fig. 6A ). Download figure Open in new tab Figure 6. Delayed tuning of place fields in 5xFAD mice (A) Intra-session stability of place fields, quantified by the correla- tion between the place fields of two segments of the trial. Data shown for familiar (FAM) and novel (NOV) environments. Cumming plots are used to visualize the effect size. Top panels show distribution of correlation coefficients; bottom panels show median difference with left-most group as reference. (B) Cumulative histograms of number of laps before place tuning is reached for the different groups. ( Insets ) Median number of laps to place tuning. Each data point corresponds to a session for one mouse. ( Left ) FAM, young WT: 3.8±0.3 laps; young 5xFAD: 3.7±0.3 laps. NOV, young WT: 5.3 ± 0.2 laps; young 5xFAD: 7.4 ± 0.4 laps. ( Right ) FAM, old WT: 3.1 ± 0.2 laps; old 5xFAD: 6.9 ± 0.3 laps. NOV, old 5xFAD: 7.7 ± 0.2 laps. + P right (WT ≥ 5xFAD) = 0.064, *** P right (WT 5xFAD) < 0.002, one-sided hierarchical bootstrap test. (C) Pairwise median difference between number of laps to place tun- ing in NOV and FAM environments. Young WT: 1.5 ± 0.2; young 5xFAD: 3.7±0.5; old WT: 1.2±0.4; old 5xFAD: 0.8±0.3. + P right (WT≥ 5xFAD) = 0.064, * P right (WT ≥5xFAD) = 0.024, *** P right (WT≥ 5xFAD) = 0.003, one-sided hierarchical bootstrap test. See also Figure S5. For all panels, young WT: n = 449 place cells pooled from 6 animals, 11 sessions; young 5xFAD: n = 390 place cells, 5 animals, 8 sessions; old WT: n = 280 place cells, 6 animals, 6 sessions; old 5xFAD: n = 291 place cells, 5 animals, 8 sessions. We investigated this effect further, by finding the first lap in each session on which place cells exhibited placetuned firing. By pooling all cells in each group, we observed that in all groups, some place cells have tuning from the beginning of the session, whereas place tuning in others was gradually recalled or encoded with experience of the environment ( Fig. 6B ). Our first hypothesis was that Alzheimer’s disease slows the rate of recall leading to a decrease in intra-session stability in old AD place fields in the familiar environment. To test this, we compared the median number of laps to acquire tuning in the familiar environment, averaged over all familiar recording sessions. We found that old AD place cells took a significantly larger number of laps to gain tuning compared with old WT ( Fig. 6B right inset), while young AD place cells did not take a significantly larger number of laps to gain tuning compared with young WT ( Fig. 6B left inset), providing evidence in support of our hypothesis. Our second hypothesis was that Alzheimer’s disease slows the rate of encoding of place fields, leading to the decrease in intra-session place field stability in the novel environment in old AD mice. We analysed the novel environment data, finding that old 5xFAD place cells took a significantly larger number of laps to gain tuning compared with old wildtype cells. On the other hand, young 5xFAD place cells took a marginally larger number of laps to gain tuning compared with young wildtype, suggesting a weaker effect in young mice. The evidence is thus in support of this second hypothsis also. To understand the mechanisms behind these differences further, we divided the data into place cells that were tuned instantly, and those that developed tuning with experience. We compared AD place cell populations to their WT counterparts in both novel and familiar environments, to determine if they (A) had a smaller proportion of initially tuned place cells, or (B) had on average a slower tuning acquisition rate for cells that developed with experience. We found that in all environments, old AD place cell populations had both a smaller proportion of cells appearing initially, and slower tuning development compared with the old WT population (Fig. S5). Young AD place cell populations had no significant differences in the initial proportion of tuned cells, but were slower to encode novel environments after the initial laps compared with young WT (Fig. S5). In healthy place cells, we expect recall to be quicker than encoding. Our third hypothesis was that Alzheimer pathology interrupts the process of recall, resulting in a novel encoding happening in both familiar and novel environments. We tested the hypothesis that the time to gain tuning in novel environments is equal to or greater than that for familiar environments in all groups, using paired sessions recorded on the same day. For old AD mice, place cells in the novel environment did not take a significantly larger number of laps to gain tuning compared with the familiar environment ( Fig. 6C right), while in all other groups, place fields took a significantly larger number of laps to gain tuning in novel than familiar environments ( Fig. 6C ). We found a similar effect when looking at the proportion of cells initially tuned, where old AD mice did not have a significantly smaller proportion of place cells tuned initially in the novel environment compared with the familiar environment, but all other groups did (see Fig. S3 ). Interestingly the rate of tuning acquisition of place cells that were not initially tuned was not significantly slower in novel compared with familiar environments for old AD, old WT, and young WT cell populations, however in young AD place cell populations, cells not tuned initially did take significantly longer to reach tuning in novel environments compared with familiar (see Fig. S3 ). 3. Discussion The most prominent amongst our findings was that amyloid pathology in old 5xFAD mice induces an elevated baseline firing, together with a reduction in the increase in firing rate that CA1 neurons typically undergo during locomotion. Together, these result in a reduction in the dynamic range of neural activity. This is to our knowledge the first experimental report of reduced dynamic range of activity in an AD mouse model. We believe that this is an important consequence of the effects of amyloid pathology (together with homeostatic tissue responses to it) on cellular and circuit function, which is fundamental to the impaired spatial, memory and cognitive function observed in the disease. This reduction in dynamic range was accompanied by impairments in neural coding for speed and place, as well as by a slowing in both the rate of encoding of new place fields, and the rate of recall of spatial tuning in familiar environments. We found hippocampal CA1 cells in old 5xFAD mice to have elevated baseline firing compared to those in old WT mice. In contrast, Zhang et al. (2023) observed lower calcium event rates in tethered 5xFAD mice as early as 4-5 months of age during immobile periods in circular arenas. When we compared calcium event rates in our own dataset (i.e. if we did not account for the amplitudes of individual events), we found comparable rates in 5xFAD and WT groups (not shown). However, elevated baseline firing is consistent with some earlier reports. Busche et al. (2012) described a complex excitability scenario in the APP/PS1 mouse, with a marked increase in the fraction of both “silent” (or hypoactive) neurons and hyperactive neurons in the plaque-bearing CA1 region of older transgenic mice. We see a similar trend in our data. In older AD mice, there is a marked increase in the number of cells with near-zero activity rates ( Fig. 1D,E ). We observed higher average running speeds in our headfixed old 5xFAD mice as they traversed a circular track. This hyperactivity in behaviour was not observed in APP knock-in mice ( Takamura et al., 2021 ) or in tethered 5xFAD mice exploring an open field ( Zhang et al., 2023 ). It has however been observed in APP-PS1 mice ( Lagadeca et al., 2012 ) and is consistent with restlessness which has been noted in AD patients ( Teri et al., 1988 ). We also observed lower neuronal activity rates in old 5xFAD during locomotion. As noted above, this, combined with the increased baseline firing, leads to a reduced dynamic range of activity in old 5xFAD mice. The dendritic tree of a neuron is believed to be important for its dynamic range, with larger dendritic trees tending to have larger dynamic ranges ( Gollo et al., 2009 ). Dendritic degeneration, characterized by loss of dendritic spines and reduction in dendritic arborization, is well known in AD ( Moolman et al., 2004 ; Tsai et al., 2004 ; Grutzendler et al., 2007 ; Šišková et al., 2014 ), and we thus speculate that such a reduction in the dendritic tree may underpin the reduction in dynamic range we observe. We identified speed-modulated cells in the CA1, consistent with earlier studies ( B. L. McNaughton and O’Keefe, 1983 ; Czurkó et al., 1999 ; Kropff et al., 2015) and found there were fewer speed-modulated CA1 cells in old 5xFAD mice compared to old WT mice. Moreover, these speed cells encode for speed in the more recent past. This suggests memory impairment and likely has implications for the updating of spatial maps during navigation, based on the path integration theory of spatial navigation, which posits that animals map their environment by integrating information on their locomotion direction and displacement, the latter requiring speed information coded in the entorhinal-hippocampal network ( McNaughton et al., 2006 ; Kropff et al., 2015; Góis and Tort, 2018 ). One of the advantages of two-photon imaging is the ability to relate the locations of amyloid plaques to effects observed in nearby neurons. We found aberrant excitability and reduced dynamic range were especially pronounced near plaques in old 5xFAD mice. The former is consistent with previous reports ( Busche et al., 2008 ). Amyloid plaques have been shown to induce abnormalities in dendritic architecture ( Le et al., 2001 ). Moreover, dendritic structural degeneration is linked to increased excitability ( Šišková et al., 2014 ). These may explain why aberrant excitability and reduction in dynamic range are most prominent close to plaques. Old 5xFAD neurons also had reduced spatial information content, lower spatial coherence, reduced spatial response modulation depth and greater trial-to-trial variability, suggesting degraded spatial tuning. While impairment of spatial representation is well known in AD ( Cheng and Ji, 2013 ; Mably et al., 2017 ; Jun et al., 2020 ), our study offers new insights into the dynamics of spatial memory encoding and recall. We observed that place fields emerged more slowly in a familiar environment in old 5xFAD compared to old WT mice ( Fig. 6A ). This is consistent with studies reporting impaired memory recall in AD animal models ( Zhao et al., 2014 ; Poll et al., 2020 ) and in human AD patients ( Haj and Robin, 2021 ). We also observed a slower rate of memory encoding in both young and old 5xFAD as evidenced by the slow emergence of place fields in old 5xFAD mice when the animals were in a novel environment. This agrees with earlier reports of lack of improvement in spatial information content over 3 days of animals learning to navigate a novel environment ( Zhao et al., 2014 ; Broussard et al., 2022 ). Sharp wave ripples (SWRs), which are important for memory, are fewer and shorter in both young ( Iaccarino et al., 2016 ) and old ( Prince et al., 2021 ) 5xFAD mice and may underlie the slow emergence of place fields. Neuronal hyperactivity and hypersynchrony are potentially interconnected in that increased neuronal activity can lead to enhanced synchronization of neuronal firing ( Kazim et al., 2021 ). In our study, in resting mice we observed neuronal hypersynchrony without hyperactivity in young 5xFAD mice, and neuronal hyperactivity without hypersynchrony in old 5xFAD mice. We did however observed neuronal hypoactivity alongside hyposynchrony in old 5xFAD mice during locomotion. Chen et al. (2025) report synchronous ensembles and correlated subthreshold membrane potentials outside of SWRs in hippocampal CA1 pyramidal neurons when mice explored a novel environment. These synchronous ensembles are believed to play a crucial role in memory consolidation. The hyposynchrony we observe in old 5xFAD during locomotion may contribute to the slower rate of memory encoding in old 5xFAD mice. We compared the number of laps to attain tuning between novel and familiar environments to try and determine whether AD impacts either encoding or recall differentially ( Fig. 5 B). That there is no significant difference in tuning acquisition time between novel and familiar environments in old AD mice seems to suggests that old AD mice are doing encoding in both environments. Since we only find the lap where the place field is initiated, this is not to say that other aspects of tuning quality may be different between novel and familiar environment. This is apparent in Fig. 5A , where we see greater instability in novel environments compared to familiar in Old AD mice. Additionally it should be noted that systematic differences in tuning consistency between groups may introduce some bias into the estimation of acquisition lap. We have, however taken steps to avoid this, by only using cells that are verified as place cells for the whole trial. In wild type mice we found that the earlier tuning in familiar environments compared with novel was accounted for by a larger proportion of cells having tuning on the very first lap, as opposed to a difference in rate of emergence with experience. That young AD mice had a faster rate of place field emergence after the first lap in the novel environment may indicate some compensatory mechanism in early stages of disease progression. One aspect of our analysis deserves comment. After extracting fluorescence time series from ROIs corresponding to individual neurons, we deconvolved the activity using the OASIS algorithm ( Friedrich et al., 2017b ) as implemented in CaImAn Giovannucci et al. (2019) . This yields trains of events with varying amplitude, as calcium transients may reflect differing numbers of action potentials. Binarizing these event-amplitude trains should be avoided for two reasons. The first is that it throws away information. The second is that it can be confounding: OASIS may split deconvolved events into two, for instance considering one event of amplitude 1.0 and two events of amplitude 0.5, nearby in time, to be very similar occurrences. Its tendency to do so depends upon the OASIS parameter settings, and also upon the signal to noise ratio (SNR) of the data and the amplitude of the 6F/F events. Both the SNR and calcium transient event amplitude can vary with experimental parameters such as imaging depth, degree of viral expression, laser power, and how scattering the tissue is, which may depend upon animal age and quality of surgical preparation. All of these properties may vary between animals and even between different fields of view in one animal (in particular, different fields of view are often at slightly different depths) - laser power can and should be adjusted to compensate as much as possible, but this is typically performed based on visual inspection and is very rough. In particular in studies where animal age is a factor, this can be a confounding effect. To control for this source of variability in our data, we took advantage of the fact that the distribution of deconvolved calcium transient event amplitudes in any field of view was very nicely fit by a lognormal distribution, and by simply multiplying by a scaling factor to make the mean 1.0, the amplitude distributions from different fields of view and different animals very nicely overlie each other ( Fig S1 ). We refer to this as standardisation of the amplitude distribution. This results in arbitrary activity units, although notably it is the same arbitrary units across our entire dataset. We find that this reduces variability in the analysis, although we also note ( Fig S2 ) that key results did not depend upon this procedure. This standardization procedure may be more widely applicable to other studies in which calcium transient deconvolution (as opposed to inference of individual spike times) is employed. In summary, in this study of the 5xFAD mouse model of Alzheimer’s Disease, we found a number of circuit alterations that began to be apparent in young (2.7 month old median age) mice and were observed more strongly in older (7.5 mo) mice. These effects included an increase in baseline neural activity together with reduced locomotiondriven firing, leading to a reduction in the dynamic range of individual neurons; aberrant synchrony; degraded spatial coding; increased variability; and delayed acquisition of spatial tuning in both familar and novel environments, indicative of deficits in both recall and learning. Our work offers new insights into the progression of neural circuit abnormalities in mouse models of amyloid disorders, and how these effects lead to memory and cognitive impairment, with potential implications for the development of new therapeutic approaches for AD. 5 Rights retention This research was funded in whole, or in part, by the Wellcome Trust [Grant number 221522/Z/20/Z]. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. Contributions MAG and SRS conceived the study and designed the experiments; MAG and YL performed the imaging and behavioural experiments; MAG, KEC, SVP, BRFT, JJY and SRS analyzed the data; MAG and SRS wrote the article. Declaration of interests The authors declare no competing interests. STAR*METHODS KEY RESOURCES TABLE Available as a separate file. RESOURCE AVAILABIILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Simon Schultz]( s.schultz{at}imperial.ac.uk ) Materials availability This study did not generate new unique reagents. Data and code availability All data reported in this paper will be shared by the Lead contact upon reasonable request. Code to generate the figures from this paper is available at the following github link: Any additonal information required to reanalyze the data reported in this paper is available from the Lead contact upon reasonable request EXPERIMENTAL MODEL AND SUBJECT DETAILS Male hemizygous 5xFAD mice (B6SJL-Tg(APPSwFlLon,PSEN1*M146L*L286V)6799Vas/Mmjax, MMRRC stock # 034840-JAX) were crossed with female C57BL6/J mice (JAX stock # 000664) to maintain 5xFAD and non-transgenic wild-type colonies under standard animal breeding at Imperial College London. Non-transgenic littermates were used as controls. The animal room was kept on a 12:12 light-dark cycle daily. Animals were divided into two age groups: young group aged 1.5 −2.4 months at the time of viral injection, 2.1-3.0 months at the time of imaging; and old group aged 5.6 −9.3 months during viral injection, 6.4-10.1 months during imaging. Data were analyzed from 6 young wild type, 6 young 5xFAD, 8 old wild type and 6 old 5xFAD mice. Both male and female mice were used for this study. All experimental procedures were carried out under the Animals (Scientific Procedures) Act 1986 and according to Home Office and institutional guidelines. METHOD DETAILS Virus injection and hippocampal window surgery Surgical procedures followed Go et al. (2021) closely. Mice were anaesthetised with 1.5-3% isofluorane (Iso-Vet, Chanelle Pharma) and body temperature was maintained at 37°C. Analgesia was administered pre-operatively with Carprofen (5 mg/kg, Rimadyl ® , Zoetis) and buprenorphine (0.07 mg/kg, Vetergesic ® , Ceva Animal Health Ltd). Anaesthetic depth was assessed via pedal flexion every 10 min for the duration of surgery. A small (∼0.5 mm) craniotomy was made and a cocktail of two viruses (pGP-AAV-syn-jGCaMP7s-WPRE, Addgene catalog 104487-AAV9, titer 1.9 1013 vg/ml, 50 nL and pAAV-CAG-tdTomato, Addgene catalog 59462-AAV1, titer 1.9 1013 vg/ml, 50 nL) was injected into the hippocampal CA1 region (coordinates from bregma, in mm: -1.3 and -1.5 DV, -1.8 ML, 2.0 AP). The first virus delivers a green genetically encoded calcium indicator (jGCaMP7s), and the second, an activity-independent red fluorescent protein (tdTomato) which is helpful for tracking cell body movements over time. pAAV1-hSyn1-mRuby2-GSG-P2A- GCaMP6s-WPRE-pA, which expresses the red fluorescent (mRuby2) and the green calcium indicator (GCaMP6s), was injected in a few early cohorts of animals. Two weeks post-injection, a hippocampal window was implanted as described by Dombeck et al. (2010). A circular craniotomy was made and the cortex (including parietal cortex and parts of visual and hindlimb sensory cortex) above the injection site was aspirated using a 27 gauge needle connected to a water pump until the fibers of the corpus callosum became visible. A stainless steel cannula (diameter: 3 mm, height: 1.5 mm) with a glass bottom was pressed down into the tissue and fixed in place using histoacryl glue (B.Braun Surgical). The surrounding skull was roughened using a scalpel blade and a stainless steel headplate (aperture: 8.5 mm) was glued to the skull, centred on the craniotomy. Exposed skull outside the headplate aperture was covered with dental cement mixed with black powder paint. Mice were given carprofen (5mg/kg/24hrs) in oral water for 3 days after surgeries. Mice were allowed 5-7 days to recover before behavioural training began. Behavioral training Behavioral training procedures are described in detail in Go et al. (2021) . Approximately one week after the hippocampal window was implanted, the animals were habituated to the apparatus and experimenter, and placed under water restriction. Animals were trained to move in the dark along a circular track (outer diameter: 32.5 cm, width: 5 cm) floating on an air table (Mobile HomeCage Large, Neurotar). Infrared (IR) light illuminated the training area and an IR camera was used to monitor the animals. The floating tracks were made of carbon fibre (weight: 100±2.8 g) and had 4-cm high walls lined with visual (phosphorescent tape, Gebildet E055 and E068) and tactile cues (sandpaper, cardboard, foam, bubble wrap). The phosphorescent tapes emitted light at 500 (blue) and 520 (green) nm throughout the imaging session ( Go et al., 2021 ). Mouse position on the track was measured using a magnetic tracking system (Neurotar) which enabled closed-loop position-based reward delivery via a lickspout. Liquid rewards were accompanied by a beep. Animals were trained twice daily in 45-min sessions. Mice were trained with one circular track in the morning and a second circular track with different visual and tactile cues in the afternoon. Water rewards of 4 µL per loop traversed were delivered at random locations. Daily water intake was limited to 1-3 mL individually adjusted for each mouse to maintain 85% of pre-restriction weight. At the start of each training session, two rewards were given to motivate the mice to lick for water. There was no limit to the number of rewards animals could receive during sessions. If the animal did not reach the target volume for the day during training, the remaining volume was given at the end of the last training session for the day. Animals were trained for 11-14 sessions before imaging sessions began. In vivo two-photon imaging Imaging was performed with a two-photon resonant scanning microscope (Scientifica VivoScope) equipped with a tiltable objective mount and a 16× water-immersion objective (LWD 0.8 NA, Nikon). We used 50% concentration ultrasound gel as the immersion fluid. For imaging jGCaMP7s and tdTomato, laser (Mai Tai HP, Newport) wavelength was set to 940 nm. Laser power underneath the objective was 60 - 166 mW. Mice were imaged while navigating a familiar (floating) circular track in sessions which lasted up to an hour. Images (512 ×512 pixels, 490× 490 µm field of view) were acquired at 30 Hz. In learning experiments, mice were imaged for up to an hour in a familiar track then moved to a novel track where they were imaged for up to another hour. Sessions for which mice ran 40-50 laps were included for analysis. SciScan software (Scientifica) was used for microscope control, and image acquisition was TTL-synchronized to position tracking and reward timing signals. Mouse position and speed data were acquired at 100 Hz with the Neurotar software. Light from the phosphorescent tapes on the floating track walls used as visual cues was not detectable on either red or green imaging channels. Approximately one week after the first imaging session, amyloid plaques were mapped in vivo at least 24 hours following i.p. injection of Methoxy-x04 (dose: 10 mg/kg), using 740 nm laser excitation. Calcium imaging data processing Processing of two-photon calcium imaging data was performed as described in Go et al. (2021) . A customised image processing pipeline based on the MATLAB version of CaImAn ( Giovannucci et al., 2019 ) was used. Motion artefacts were removed using rigid and then non-rigid image registration. Regions of interest (ROIs) were automatically identified and overlapping ROIs excluded. To track cells across multiple images, we temporally concatenated the images and segmented the concatenated video. Neural activity was deconvolved from the fluorescence traces using OASIS ( Friedrich et al., 2017a ). This produced an event train which preserved both time of onset and amplitude of inferred calcium transient events. In the absence of per-cell calibration to true spike counts, we consider the units of the amplitude of this signal train to be arbitrary (arbs), inherited from the original time series extracted from each ROI. Standardized activity rate To account for differences in signal-to-noise ratio, and consequently, OASIS event amplitude, across mice and fields of view (FOV), we fitted the distribution of event amplitudes for each FOV in each mouse to a lognormal distribution ( Figure S1 ), and standardized each distribution to unit mean. Activity rates were calculated by integrating the amplitudes of events within a certain time window and dividing by window duration and are reported in arbitrary units per minute (arbs/min). DATA ANALYSIS All analyses were performed with custom MATLAB scripts. Resting and running periods Periods in which animal speed was less than 0.5 cm/s were considered resting periods. Periods in which animal speed exceeded 2.0 cm/s were considered running periods. Analyses on speed scores, speed modulated cells, and place tuned cells were limited to running periods to distinguish speed and place-correlated effects from those related to changes in animal behavior (grooming, pausing for rewards, foraging). Temporal binning The neural activity temporal resolution is defined by the image sampling rate (30 Hz). For the analysis in this paper, except for Fig. 5 , we reduced the temporal resolution of neural activity by summing event amplitudes in 5 consecutive time bins (effective rate 6 Hz). In place tuning analysis ( Fig. 5 ), we summed the event amplitudes in 10 consecutive time bins (effective rate 3 Hz). For speed and position data, we first downsampled the data from 100 Hz to 30 Hz. We then took the average value over the required number of time bins to match the temporal resolution of the neural activity. Dynamic range and speed modulated cells To construct speed tuning curves, neural activity and speed time series were sampled at 5 ms bins. We used 1 cm/s speed bins, except for 0.5 cm/s bins below 1 cm/s, and calculated speed tuning curves for each cell by dividing the total activity for a speed bin by the total time spent in it, then smoothing using a boxcar average over 3 bins. The dynamic range of activity due to locomotion was defined as ratio of the the maximum of the speed tuning curve minus the non-zero minimum. The speed response modulation depth of a cell was defined as the maximum of the speed tuning curve minus the minimum. Following Kropff et al. (2015), neurons were classified as speed modulated cells if their speed response modulation depth exceeded chance level. Chance-level was determined by a shuffling procedure. For each shuffle, the neural activity time series was time-shifted by a random number between 10 seconds and the time series duration minus 10 seconds. A speed tuning curve and its corresponding modulation depth was then calculated for the shuffled neural activity. The cell was considered speed modulated if its modulation depth exceeded the 95th percentile of the distribution of values collected from 500 shuffles. Following Muzzu et al. (2018) , we classified speed modulated cells according to their responses as positively modulated, negatively modulated or having a preferred speed. For each speed modulated cell, we took the running segment of the speed tuning curve (> 2 cm/s), included speed bins with total duration of at least 4 seconds, then used the best-fit curve (linear, quadratic, double exponential, Gaussian, or double Gaussian) to determine the response type of the cell. Specifically, speed modulated cells were classified as 1) positively modulated if their maximum firing rate exceeded the firing rate during rest (<0.5 cm/s) and was recorded at a speed greater than 70% of the maximum animal speed; (2) negatively modulated if their minimum firing rate was less than the firing rate during rest and was recorded at a speed greater than 70% of the maximum animal speed; (3) having a preferred speed if the maximum firing rate exceeded the firing rate during rest and was recorded at a speed less than 70% of the maximum animal speed. Synchrony We calculated the Pearson correlation coefficient for all cell pairs that were imaged in a field of view, normalized such that the auto-correlations at zero lag were identically 1.0. Synchrony was defined as the mean pairwise correlation at zero time lag. Averages were calculated by first taking the Fisher’s transform of the Pearson’s r values: z = 0.5 ln ((1 + r)/(1− r)), calculating the mean of the z values, and then transforming the average z value back to an r value. Place cell analysis Mouse position was linearized by converting angular distance to Euclidean distance using the known circumference of the circular track. Neural activity and position time series were sampled at 33 ms bins. We divided the track into 2-cm spatial bins and constructed neural activity rate maps for each cell by dividing the total activity during the occupancy of a spatial bin by the total time it was occupied, smoothing using a boxcar average over 4 bins and normalizing each map by its maximum value. The dynamic range of activity in response to location was defined as the maximum of the activity rate map minus the minimum. Spatial information rate (in bits/arb) for each cell was computed using the Kraskov method (see Spatial information). Neurons were classified as place cells if they met the following criteria: (1) calcium transient events were present for at least 30% of the laps through the circular track, (2) mean in-field activity rate was at least 1.5 times the mean out-of-field activity rate and (3) the cell contained spatial information greater than chance. Chance-level spatial information for a cell was determined by a shuffling procedure. For each shuffle, the neural activity time series is time-shifted by a random number between 10 seconds and the time series duration minus 10 seconds. The spatial information rate for the shuffled neural activity is then calculated and chance-level values are pooled from 500 repetitions. The cell was considered a place cell if its information rate exceeded the 95% percentile of the shuffled data. The location of the place field was defined by the bin location of maximum activity rate while the place field size was determined by the number of bins for which the neural activity rate was at least 50% of the maximum. To quantify intra-trial place field stability, each trial (session) was divided into 2 segments. For each cell, the Pearson correlation coefficient between activity rate maps generated from each segment was then calculated. Place field tuning acquisition For each place cell, the lap of the track at which place cell tuning begins was determined using the binary place field locations identified from the first 50 laps. We looked at firing rates in the place field in a rolling 5 lap window, to find the first tuned 5 lap window. We defined being tuned as a window where the firing rate in the place field locations exceeded the average firing rate from all locations. We assessed this on the whole 5 lap window, as well as for the individual laps. The first window in which the place field firing rate was greater than the mean on more than one lap as well as the whole window was taken to be the tuning acquisition lap. Cells that met the above criteria at no point in the first 50 laps were removed from analysis. To compare between groups we used calculated the median tuning acquisition lap for each recording sessions. Additionally, we measured the proportion of place cells where tuning was present on the first lap in each recording session. Finally, we compared the median tuning acquisition lap of cells where tuning was not present on the first lap. These statistics were determined using session averages pooled over all sessions in each group. We directly compared place field appearance novel and familiar environments using paired novel familiar experiments recorded on the same day. For each of the three measurements discussed above, we measured the pairwise difference between session populations in paired novel and familiar environments to determine if the novel environment had delayed tuning in each group. Spatial information The mutual information between neural activity and spatial location is given by where R is the random variable describing the neural response and S is the random variable describing spatial location. r is the neural activity rate in a 300 ms window (10 consecutive fluorescence time bins), obtained by accumulating the standardised activities across the window, and dividing by the window duration. s was the average spatial position within the time window. To estimate mutual information we used the Kraskov estimator ( Kraskov et al., 2004 ) as implemented in the JAVA Information Dynamics Toolkit (JIDT) ( Lizier, 2014 ). This is a binless estimator which uses the ditances between nearest neighbours to estimate information, reducing bias and adapting to the local data density, making it particularly suitable for real-valued variables. The Kraskov estimator has a parameter k representing the number of nearest neighbours considered, which we set to 4, chosen to balance the trade-off between bias and variance. Neuronal variability analysis Neuronal variability analyses was performed by adapting methods from classical visual neuroscience literature ( Tolhurst et al., 1983 ) to hippocampal place cells and calcium transient amplitude-event trains, as described in Go et al. (2021) . This analysis shows how variable responses are to laps around the same track. Briefly, for each cell, we calculated for each lap the activity rate for each 2-cm bin in the track. We then calculated the mean and the variance of this quantity across laps for all spatial bins for each cell. We used total least squares linear regression to fit a power law model y = αx /3 to the relationship between activity variance and mean for each individual cell. The vertical intercept parameter α describes the overall variability of the cell, with a Poisson process having α = 1. Also of interest is the power law exponent β, which can be read off from the slope of the fitted line, describing how the variability scales with firing rate. We suggest that these quantities provide potentially highly useful information about the reliability of neuronal signaling with disturbances of excitability altering the scaling behaviour. An exponent of 1 indicates reliability equivalent to that of a Poisson process; above 1 implies additional sources of variability. Coherence Spatial coherence is a measure of the spatial contiguity of the activity of a neuron. Similar to Zhang et al. (2014) , it was obtained by calculating the correlation coefficient between a neuron’s activity rate map and the same rate map smoothed by a boxcar average over 4 bins. Coherence was defined as the Fisher’s transform of the Pearson’s r value: z = 0.5 ln((1 + r)/(1 − r)). Amyloid plaque detection Two-photon images of amyloid plaques were averaged for each z plane and denoised with a gaussian filter. Images from adjacent FOVs were stitched to create a montage 1.1 um x 1.1 um centred on the imaging FOV. We developed a convolution neural network (CNN) for automatic detection of the centroids of amyloid plaques in images. The training data set consisted of 70 images with manually selected plaque ROIs, augmented to 3360 images by one or more of the following processes: vertical and horizontal flipping, rotations of 90°, 180° and 270°, and geometrical operations of rotation, zoom, skew, random distortion and shear. The samples were randomly split into 75% for training and 25% as testing data for validation. Analysis of distance to plaques To examine the spatial relationship between amyloid plaques and cells, the distance from each cell to every plaque was measured. For each plaque, annular rings of width 20 µm were drawn centred on the plaque, with the outer radius (i.e. distance from a plaque) ranging from 20 to 200 µm. Cells were included in the annular ring where they were located and a cell could be included in multiple annular rings of different outer radii centred on different plaques. The quantity of interest (e.g. activity rate, dynamic range, spatial information, etc.) was averaged over all cells within all annular rings of the same outer radius. In cases when cell count (e.g. speed modulated cells, negatively modulated cells) was quantified, the distance from a cell to the nearest plaque was used and cells were only counted once. QUANTIFICATION AND STATISTICAL ANALYSIS Hierarchical bootstrapping In this paper, we consider many neurons which are hierarchically grouped, and as such, not all neurons in a group may be independent samples. Summary statistics (e.g. mean) reported here are obtained using the pooled distribution of all neurons, but a bootstrapping approach ( Saravanan et al., 2020 ) is used to account for the dependence inherent in the hierarchical structure of the recorded data when assessing the significance of the results. Data was collected in the following nested hierarchies: Neurons, within recording days, within microscope fields of view, within mice. Bootstrapping was performed on each level of this hierarchy 1000 times. The same number of samples as originally recorded are resampled with replacement from the original samples, mimicking a sampling distribution. This resampling is performed at each layer of the hierarchy. For each replicate, summary statistics are calculated on the pooled resampled statistic. Where population level statistics for a session are reported, cells within each session are averaged, and then higher level resampling is performed, as normal. SEM values reported are from the resampled population. As in Prince et al. (2021) , reported P values were determined as the proportion of pairs of replicates where the statistic for group A exceeded that of group B. This is a direct probability P of the population of A being greater than or equal to the population of B: P(µ B > µ A ) and equates to the proportion of the joint probability distribution that falls above the x = y line. This is in contrast with p values used in statistical tests such as the t-test or ANOVA, which indicate the probability of obtaining results as extreme as those observed, given that the null hypothesis is true. The probability is significant if P > (1 α/2) or if P (1 α/2) denotes that the probability that the WT mean is greater than the 5xFAD mean, P(WT 5xFAD), is significant while P 0.95 or P 0.975 or P 0.995 or P 0.9975 or P < 0.0025 (α = 0.005). Where a one sided comparison is made using bootstrapping, this probability is expressed as either P right or P left for right and left sided tests respectively, where P right = p and P left = 1− p. Lower significance thresholds are doubled in single sided tests, as follows: + P right < 0.1 (α = 0.10), *P right < 0.05 (α = 0.05), **P right < 0.01 (α = 0.01), ***P right < 0.005 (α = 0.005), and the same for P left . Estimation statistics and Cumming plots Estimation statistics reports the magnitude of the effect size and its confidence interval in contrast to significance testing. We used the Cumming plot from the DABEST framework ( Ho et al., 2020 ) to visualize the effect size and to show comparison between multiple groups. The upper part of the plot shows the distribution of raw data with the mean ±SEM. The lower part shows the median difference, with 95% confidence interval, between a group and the left-most group which is taken as the reference. The median difference is calculated from 5000 bootstrap resamples with replacement. Supplemental Information Download figure Open in new tab Figure S1. Standardising neural activity units across the dataset (A) Representative GCaMP image (top) and the identified cells (bottom). (B) df/f traces for 6 of the 183 cells identified. (C) Deconvolved activity for df/f traces shown in B. (D) Log-normal distributions of the amplitudes of the deconvolved activity shown for all experiments for all mice. (E) Log-normal distributions of the standardised amplitudes of the deconvolved activity shown for all experiments for all mice. (F) The scaling factors that give the scaled distributions in (E) each a mean of 1. Download figure Open in new tab Figure S2. Aberrant excitability results are not due to our procedure for standardising deconvolved calcium event amplitude distributions across fields of view (A) Distributions of raw (unstandardised) activity rates across all cells for 5xFAD and WT mice during resting (speed 2 cm/s) periods. Inverted triangles denote median. Group color codes as in B. (B) Cumulative histograms of raw activity rates show more neurons in old 5xFAD mice with higher and lower firing rates during resting and running periods, respectively, consistent with trends obtained with standardised activity rates. * p boot = 0.006, *** p boot = 0.002, all statistics were performed with hierarchical bootstrap analysis (see STAR Methods). Download figure Open in new tab Figure S3. Increased proportion of negatively modulated speed cells in 5xFAD mice (A) Instantaneous speed (top) and firing rate (bottom) for three different types of speed-modulated cells. (B) Speed tuning curves for the different types of speed-modulated cells in A. Solid lines denote best fit lines. (C) Numbers of the different types of speed-modulated cells as a fraction of the number of speed- modulated cells. There are more negatively-modulated cells in old 5xFAD mice. * p boot = 0.018. Download figure Open in new tab Figure S4. Young AD mice show slower rate of tuning acquisition in novel environments. Old AD mice show impairments in both initially tuned proportion, and tuning acquisition rate in all environments (A) Proportion of place cells appearing initially. The young AD initial tuning proportion was not smaller than that of young WT in both the novel and familiar environment . The old AD initial tuning proportion was smaller than that of old WT in both the novel and familiar environment . (B) Pairwise difference between novel and familiar proportions of place cells tuned initially. Place cells in the novel environment had a significantly smaller proportion of cells initially tuned than in the familiar environment for young WT , young AD and old WT mice , but not in old AD mice (C) Number of laps until tuning for place cells that were not initially tuned. The young AD place cells that did not appear initially took more laps to appear in the novel environment than the young WT but did not take significantly more in the familiar environment . The old AD place cells that were not initially tuned took more laps to appear than the old WT in both the novel and familiar environments . (D) Pairwise difference between novel and familiar number of laps until tuning for place cells that were not initially tuned. Place cells in the novel environment that were not initially tuned took more laps than the in the familiar environment for young AD , but not for young WT , old WT mice or old AD mice . 4 Acknowledgements This work was supported by UKRI/Wellcome grant EP/W024020/1 to SRS; Alzheimer’s Research UK (ARUK-NC2019-IMP) grant to SRS and MAG; EPSRC grant EP/J021199/1 to SRS; EPSRC CDT in Neurotechnology for Life and Health (EP/L016737/1) studentship to SVP; BBSRC grant BB/R022437/1 to SRS; Wellcome Trust grant 221522/Z/20/Z to SRS; Chan-Zuckerberg Initiative award NC/W000903/1 to SRS, and a philanthropic donation from Mrs Anne Uren and the Michael Uren Foundation to SRS. We thank M. Sastre for providing the 5xFAD mice, and D. Dupret, N. Zabouri, R. MitchellHeggs, M. Sastre and S. Barnes for useful discussions. Footnotes ↵ * m.go{at}imperial.ac.uk ↵ ** s.schultz{at}imperial.ac.uk References ↵ Aghajan , Z.M. , Acharya , L. , Cushman , J. , Vuong , C. , Moore , J. , Mehta , M.R. , 2013 . Hippocampal motifs . bioRxiv , 001636 . ↵ B. L. McNaughton , C.A.B. , O’Keefe , J. , 1983 . The contributions of position, direction, and velocity to single unit activity in the hippocampus of freely-moving rats . Exp Brain Res 52 , 41 – 49 . OpenUrl CrossRef PubMed Web of Science ↵ Bezzina , C. , Verret , L. , Juan , C. , Remaud , J. , Halley , H. , Rampon , C. , Dahan , L. , 2015 . Early onset of hypersynchronous network activity and expression of a marker of chronic seizures in the tg2576 mouse model of alzheimer’s disease . PLoS ONE 10 , e0119910 . OpenUrl CrossRef PubMed ↵ Broussard , J.I. , Redell , J.B. , Maynard , M.E. , Zhao , J. , Moore , A. , Mills , R.W. , Hood , K.N. , Underwood , E. , Roysam , B. , Dash , P.K. , 2022 . Impaired experience-dependent refinement of place cells in a rat model of alzheimer’s disease . J Alzheimers Dis 86 , 1907 – 1916 . OpenUrl CrossRef PubMed ↵ Busche , M.A. , Chen , X. , Henning , H.A. , Reichwald , J. , Staufenbiel , M. , Sakmann , B. , Konnerth , A. , 2012 . Critical role of soluble amyloid-/3 for early hippocampal hyperactivity in a mouse model of alzheimer’s disease . Proceedings of the National Academy of Sciences 109 , 8740 – 8745 . OpenUrl Abstract / FREE Full Text ↵ Busche , M.A. , Eichhff , G. , Adelsberger , H. , Abramowski , D. , Wiederhold , K.H. , Haass , C. , Staufenbiel , M. , Konnerth , A. , Garaschuk , O. , 2008 . Clusters of hyperactive neurons near amyloid plaques in a mouse model of alzheimer’s disease . Science 321 , 1686 – 1689 . OpenUrl Abstract / FREE Full Text ↵ Cacucci , F. , Yi , M. , Wills , T.J. , Chapman , P. , O’Keefe , J. , 2008 . Place cell firing correlates with memory deficits and amyloid plaque burden in tg2576 alzheimer mouse model . Proceedings of the National Academy of Sciences 105 , 7863 – 7868 . OpenUrl Abstract / FREE Full Text ↵ Carandini , M. , 2004 . Amplification of trial-to-trial response variability by neurons in visual cortex . PLoS biology 2 , e264 . OpenUrl CrossRef PubMed ↵ Chen , E.L. , Chen , T.W. , Schreiter , E.R. , Lin , B.J. , 2025 . Synchronous ensembles of hippocampal ca1 pyramidal neurons associated with theta but not ripple oscillations during novel exploration . eLife 13 , e96718 . OpenUrl ↵ Cheng , J. , Ji , D. , 2013 . Rigid firing sequences undermine spatial memory codes in a neurodegenerative mouse model . eLife 2 , e00647 . OpenUrl CrossRef PubMed ↵ Coughlan , G. , Laczó , J. , Hort , J. , Minihane , A.M. , Hornberger , M. , 2018 . Spatial navigation deficits—overlooked cognitive marker for preclinical alzheimer disease? Nature Reviews Neurology 14 , 496 – 506 . OpenUrl CrossRef PubMed ↵ Czurkó , A. , Hirase , H. , Csicsvari , J. , Buzsáki , G. , 1999 . Sustained activation of hippocampal pyramidal cells by ‘space clamping’ in a running wheel . Eur J Neurosci 1 , 344 – 352 . OpenUrl ↵ Friedrich , J. , Yang , W. , Soudry , D. , Mu , Y. , Ahrens , M.B. , Yuste , R. , Peterka , D.S. , Paninski , L. , 2017a . Multi-scale approaches for high-speed imaging and analysis of large neural populations . PLOS Comput. Biol . 13 , e1005685 . doi: 10.1371/journal.pcbi.1005685 . OpenUrl CrossRef PubMed ↵ Friedrich , J. , Zhou , P. , Paninski , L. , 2017b . Fast online deconvolution of calcium imaging data . PLoS computational biology 13 , e1005423 . OpenUrl CrossRef PubMed ↵ Giovannucci , A. , Friedrich , J. , Gunn , P. , Kalfon , J. , Brown , B.L. , Koay , S.A. , Taxidis , J. , Najafi , F. , Gauthier , J.L. , Zhou , P. , Khakh , B.S. , Tank , D.W. , Chklovskii , D.B. , Pnevmatikakis , E.A. , 2019 . CaImAn an open source tool for scalable calcium imaging data analysis . eLife 8 , 38173 . doi: 10.7554/eLife.38173 . OpenUrl CrossRef ↵ Go , M.A. , Rogers , J. , Gava , G.P. , Davey , C.E. , Prado , S. , Liu , Y. , Schultz , S.R. , 2021 . Place cells in head-fixed mice navigating a floating real-world environment . Frontiers in Cellular Neuroscience 15 , 19 . OpenUrl ↵ Góis , Z.H.T.D. , Tort , A.B.L. , 2018 . Characterizing speed cells in the rat hippocampus . Cell Reports 25 , 1872 – 1884 . OpenUrl CrossRef PubMed ↵ Gollo , L.L. , Kinouchi , O. , Copelli , M. , 2009 . Active dendrites enhance neuronal dynamic range . PLoS Computat Biol 5 , e1000402 . OpenUrl CrossRef ↵ Gonzalez , W.G. , Zhang , H. , Harutyunyan , A. , Lois , C. , 2019 . Persistence of neuronal representations through time and damage in the hippocampus . Science 365 , 821 – 825 . OpenUrl Abstract / FREE Full Text ↵ Grutzendler , J. , Helmin , K. , Tsai , J. , Gan , W. , 2007 . Various dendritic abnormalities are associated with fibrillar amyloid deposits in alzheimer’s disease . Ann N Y Acad Sci 1097 , 30 – 39 . OpenUrl CrossRef PubMed Web of Science ↵ Gur , M. , Beylin , A. , Snodderly , D.M. , 1997 . Response variability of neurons in primary visual cortex (v1) of alert monkeys . J Neurosci 17 , 2914 – 2920 . OpenUrl Abstract / FREE Full Text ↵ Hainmueller , T. , Bartos , M. , 2018 . Parallel emergence of stable and dynamic memory engrams in the hippocampus . Nature 558 , 292 – 296 . OpenUrl CrossRef PubMed ↵ Haj , M.E. , Robin , F. , 2021 . Repeated recall on source misattribution in alzheimer’s disease . Memory 29 , 1354 – 1361 . OpenUrl CrossRef PubMed ↵ Harvey , C.D. , Collman , F. , Dombeck , D.A. , Tank , D.W. , 2009 . Intracellular dynamics of hippocampal place cells during virtual navigation . Nature 461 , 941 – 946 . OpenUrl CrossRef PubMed Web of Science ↵ Ho , J. , Tumkaya , T. , Aryal , S. , Choi , H. , Claridge-Chang , A. , 2020 . Moving beyond p values: data analysis with estimation graphics . Nature Methods 16 , 565 – 566 . OpenUrl ↵ Hyman , B.T. , Van Hoesen , G.W. , Damasio , A.R. , Barnes , C.L. , 1984 . Alzheimer’s disease: cell-specific pathology isolates the hippocampal formation . Science 225 , 1168 – 1170 . OpenUrl Abstract / FREE Full Text ↵ Iaccarino , H.F. , Singer , A.C. , Martorell , A.J. , Rudenko , A. , Gao , F. , Gillingham , T.Z. , Mathys , H. , Seo , J. , Kritskiy , O. , Abdurrob , F. , Adaikkan , C. , Canter , R.G. , Rueda , R. , Brown , E.N. , Boyden , E.S. , Tsai , L.H. , 2016 . Gamma frequency entrainment attenuates amyloid load and modifies microglia . Nature 540 , 230 – 235 . OpenUrl CrossRef PubMed ↵ Jawhar , S. , Trawicka , A. , Jenneckens , C. , Bayer , T.A. , Wirths , O. , 2012 . Motor deficits, neuron loss, and reduced anxiety coinciding with axonal degeneration and intraneuronal a/3 aggregation in the 5xfad mouse model of alzheimer’s disease . Neurobiology of aging 33 , 196 – e29 . OpenUrl ↵ Jun , H. , Bramian , A. , Soma , S. , Saito , T. , Saido , T.C. , Igarashi , K.M. , 2020 . Disrupted place cell remapping and impaired grid cells in a knockin model of alzheimer’s disease . Neuron 107 , 1095 – 1112 . OpenUrl CrossRef PubMed ↵ Kazim , S.F. , Seo , J.H. , Bianchi , R. , Larson , C.S. , Sharma , A. , Wong , R.K.S. , Gorbachev , K.Y. , Pereira , A.C. , 2021 . Neuronal network excitability in alzheimer’s disease: the puzzle of similar versus divergent roles of amyloid /3 and tau . eNeuro 8 , ENEURO.0418– 20.2020. ↵ Kraskov , A. , Stögbauer , H. , Grassberger , P. , 2004 . Estimating mutual information . Physical Review E—Statistical, Nonlinear, and Soft Matter Physics 69 , 066138 . OpenUrl Krop? , E. , Carmichael , J.E. , Moser , M.B. , Moser , E.I. , 2015 . Speed cells in the medial entorhinal cortex . Nature 523 , 419 – 424 . doi: 10.1038/nature14622 . OpenUrl CrossRef PubMed ↵ Lagadeca , S. , Rotureaua , L. , Hémarb , A. , Macreza , N. , Delcassoa , S. , Jeanteta , Y. , Cho , Y.H. , 2012 . Early temporal short-term memory deficits in double transgenic app/ps1 mice . Neurobiol Aging 33 , 203.e1 – 203.e11 . OpenUrl CrossRef PubMed ↵ Langston , R.F. , Ainge , J.A. , Couey , J.J. , Canto , C.B. , Bjerknes , T.L. , Witter , M.P. , Moser , E.I. , Moser , M.B. , 2010 . Development of the spatial representation system in the rat . Science 328 , 1576 – 1580 . OpenUrl Abstract / FREE Full Text ↵ Le , R. , Cruz , L. , Urbanc , B. , Knowles , R.B. , Hsiao-Ashe , K. , Duff , K. , Irizzary , M.C. , Stanley , H.E. , Hyman , B.T. , 2001 . Plaque-induced abnormalities in neurite geometry in transgenic models of alzheimer disease: Implications for neural system disruption . J Neuropathol Exp Neurol 60 , 735 – 758 . OpenUrl ↵ Lizier , J.T. , 2014 . Jidt: An information-theoretic toolkit for studying the dynamics of complex systems . Frontiers in Robotics and AI 1 , 11 . OpenUrl ↵ Mably , A.J. , Gereke , B.J. , Jones , D.T. , Colgin , L.L. , 2017 . Impairments in spatial representations and rhythmic coordination of place cells in the 3xtg mouse model of alzheimer’s disease . Hippocampus 27 , 378 – 392 . OpenUrl CrossRef PubMed ↵ McNaughton , B.L. , Battaglia , F.P. , Jensen , O. , Moser , E.I. , Moser , M.B. , 2006 . Path integration and the neural basis of the ‘cognitive map’ . Nat Rev Neurosci 7 , 663 – 678 . OpenUrl CrossRef PubMed Web of Science ↵ Moolman , D.L. , Vitolo , O.V. , Vonsattel , J.P.G. , Shelanski , M.L. , 2004 . Dendrite and dendritic spine alterations in alzheimer models . J Neurocytol 33 , 377 – 387 . OpenUrl CrossRef PubMed Web of Science ↵ Muller , R.U. , Kubie , J.L. , 1987 . The e?ects of changes in the environment on the spatial firing of hippocampal complex-spike cells . Journal of Neuroscience 7 , 1951 – 1968 . OpenUrl Abstract / FREE Full Text ↵ Muzzu , T. , Mitolo , S. , Gava , G.P. , Schultz , S.R. , 2018 . Encoding of locomotion kinematics in the mouse cerebellum . PLOS ONE 13 , e0203900 . OpenUrl CrossRef PubMed ↵ Oakley , H. , Cole , S.L. , Logan , S. , Maus , E. , Shao , P. , Craft , J. , Guillozet-Bongaarts , A. , Ohno , M. , Disterhoft , J. , Van Eldik , L. , et al. , 2006 . Intraneuronal /3-amyloid aggregates, neurode-generation, and neuron loss in transgenic mice with five familial alzheimer’s disease mutations: potential factors in amyloid plaque formation . Journal of Neuroscience 26 , 10129 – 10140 . OpenUrl Abstract / FREE Full Text ↵ O’Keefe , J. , Dostrovsky , J. , 1971 . The hippocampus as a spatial map: Preliminary evidence from unit activity in the freely-moving rat . Brain Research 34 , 171 – 175 . OpenUrl CrossRef PubMed Web of Science ↵ Ólafsdóttir , H.F. , Bush , D. , Barry , C. , 2018 . The role of hippocampal replay in memory and planning . Current Biology 28 , R37 – R50 . OpenUrl CrossRef PubMed ↵ Palop , J.J. , Chin , J. , Roberson , E.D. , Wang , J. , Thwin , M.T. , Bien-Ly , N. , Yoo , J. , Ho , K.O. , Yu , G.Q. , Kreitzer , A. , et al. , 2007 . Aberrant excitatory neuronal activity and compensatory remodeling of inhibitory hippocampal circuits in mouse models of alzheimer’s disease . Neuron 55 , 697 – 711 . OpenUrl CrossRef PubMed Web of Science ↵ Palop , J.J. , Mucke , L. , 2009 . Epilepsy and cognitive impairments in alzheimer disease . Archives of neurology 66 , 435 – 440 . OpenUrl CrossRef PubMed Web of Science ↵ Palop , J.J. , Mucke , L. , 2016 . Network abnormalities and interneuron dysfunction in alzheimer disease . Nature Reviews Neuroscience 17 , 777 – 792 . OpenUrl CrossRef PubMed ↵ Panzeri , S. , Schultz , S.R. , Treves , A. , Rolls , E.T. , 1999 . Correlations and the encoding of information in the nervous system . Proceedings of the Royal Society of London. Series B: Biological Sciences 266 , 1001 – 1012 . OpenUrl CrossRef PubMed Web of Science ↵ Poll , S. , Mittag , M. , Musacchio , F. , Justus , L.C. , Giovannetti , E.A. , Steffen , J. , Wagner , J. , Zohren , L. , Schoch , S. , Schmidt , B. , Jackson , W.S. , Ehninger , D. , Fuhrmann , M. , 2020 . Memory trace interference impairs recall in a mouse model of alzheimer’s disease . Nat Neurosci 23 , 952 – 958 . OpenUrl CrossRef PubMed ↵ Prince , S.M. , Paulson , A.L. , Jeong , N. , Zhang , L. , Amigues , S. , Singer , A.C. , 2021 . Alzheimer’s pathology causes impaired inhibitory connections and reactivation of spatial codes during spatial navigation . Cell Reports 35 , 109008 . OpenUrl CrossRef PubMed ↵ Saravanan , V. , Berman , G.J. , Sober , S.J. , 2020 . Application of the hierarchical bootstrap to multi-level data in neuroscience . Neuron Behav Data Anal Theory 3 . ↵ Schultz , S.R. , Panzeri , S. , 2001 . Temporal correlations and neural spike train entropy . Physical Review Letters 86 , 5823 . OpenUrl CrossRef PubMed Web of Science ↵ Selkoe , D.J. , 2001 . Alzheimer’s disease: genes, proteins, and therapy . Physiological Reviews 81 , 741 – 766 . OpenUrl CrossRef PubMed Web of Science ↵ Selkoe , D.J. , 2002 . Alzheimer’s disease is a synaptic failure . Science 298 , 789 – 791 . OpenUrl Abstract / FREE Full Text ↵ Shah , D. , Praet , J. , Hernandez , A.L. , Hoöfling , C. , Anckaerts , C. , Bard , F. , Morawski , M. , Detrez , J.R. , Prinsen , E. , Villa , A. , Vos , W.H.D. , Maggi , A. , D’Hooge , R. , Balschun , D. , Rossner , S. , Verhoye , M. , der Linden , A.V. , 2016 . Early pathologic amyloid induces hypersynchrony of bold resting-state networks in transgenic mice and provides an early therapeutic window before amyloid plaque deposition . Alzheimer’s & Dementia 12 , 964 – 976 . OpenUrl CrossRef ↵ Skaggs , W. , Mcnaughton , B. , Gothard , K. , 1992 . An information-theoretic approach to deciphering the hippocampal code . Advances in neural information processing systems 5 . ↵ Skaggs , W.E. , McNaughton , B.L. , Wilson , M.A. , Barnes , C.A. , 1996 . Theta phase precession in hippocampal neuronal populations and the compression of temporal sequences . Hippocampus 6 , 149 – 172 . OpenUrl CrossRef PubMed Web of Science ↵ Squire , L.R. , 2004 . Memory systems of the brain: a brief history and current perspective . Neurobiology of Learning and Memory 82 , 171 – 177 . OpenUrl CrossRef PubMed Web of Science ↵ Takamura , R. , Mizuta , K. , Sekine , Y. , Islam , T. , Saito , T. , Sato , M. , Ohkura , M. , Nakai , J. , Ohshima , T. , Saido , T. , Hayashi , Y. , 2021 . Modality-specific impairment of hippocampal ca1 neurons of alzheimer’s disease model mice . Journal of Neuroscience 41 , 5315 – 4329 . OpenUrl Abstract / FREE Full Text ↵ Teri , L. , Larson , E.B. , Reifler , B.V. , 1988 . Behavioral disturbance in dementia of the alzheimer’s type . Journal of the American Geriatrics Society 36 , 1 – 6 . OpenUrl CrossRef PubMed Web of Science ↵ Tolhurst , D.J. , Movshon , J.A. , Dean , A.F. , 1983 . The statistical reliability of signals in single neurons in cat and monkey visual cortex . Vision research 23 , 775 – 785 . OpenUrl CrossRef PubMed Web of Science ↵ Tsai , J. , Grutzendler , J. , Duff , K. , Gan , W. , 2004 . Fibrillar amyloid deposition leads to local synaptic abnormalities and breakage of neuronal branches . Nat Neurosci 7 , 1181 – 1183 . OpenUrl CrossRef PubMed Web of Science ↵ Šišková , Z. , Justus , D. , Kaneko , H. , Friedrichs , D. , Henneberg , N. , Beutel , T. , Pitsch , J. , Schoch , S. , Becker , A. , von der Kammer , H. , Remy , S. , 2014 . Dendritic structural degeneration is functionally linked to cellular hyperexcitability in a mouse model of alzheimer’s disease . Neuron 84 , 1023 – 1033 . OpenUrl CrossRef PubMed ↵ Verret , L. , Mann , E.O. , Hang , G.B. , Barth , A.M. , Cobos , I. , Ho , K. , Devidze , N. , Masliah , E. , Kreitzer , A.C. , Mody , I. , Mucke , L. , Palop , J.J. , 2012 . Inhibitory interneuron deficit links altered network activity and cognitive dysfunction in alzheimer model . Cell 149 , 708 – 721 . OpenUrl CrossRef PubMed Web of Science ↵ Wilson , M.A. , McNaughton , B.L. , 1993 . Dynamics of the hippocampal ensemble code for space . Science 261 , 1055 – 1058 . OpenUrl Abstract / FREE Full Text ↵ Zhang , H. , Chen , L. , Johnston , K.G. , Crapser , J. , Green , K.N. , Ha , N.M.L. , Tenner , A.J. , Holmes , T.C. , Nitz , D.A. , Xu , X. , 2023 . Degenerate mapping of environmental location presages deficits in object-location encoding and memory in the 5xfad mouse model for alzheimer’s disease . Neurobiology of disease 176 , 105939 . OpenUrl CrossRef PubMed ↵ Zhang , S. , Schönfeld , F. , Wiskott , L. , Manahan-Vaughan , D. , 2014 . Spatial representations of place cells in darkness are supported by path integration and border information . Front Behav Neurosci 8 , 222 . OpenUrl CrossRef PubMed ↵ Zhao , R. , Fowler , S.W. , Chiang , A.C. , Ji , D. ,, Jankowsky , J.L. , 2014 . Impairments in experience-dependent scaling and stability of hippocampal place fields limit spatial learning in a mouse model of alzheimer’s disease . Hippocampus 24 , 963 – 978 . OpenUrl CrossRef PubMed ↵ Ziv , Y. , Burns , L.D. , Cocker , E.D. , Hamel , E.O. , Ghosh , K.K. , Kitch , L.J. , Gamal , A.E. , Schnitzer , M.J. , 2013 . Long-term dynamics of ca1 hippocampal place codes . Nature Neuroscience 16 , 264 – 266 . OpenUrl CrossRef PubMed ↵ Zong , W. , Obenhaus , H.A. , Skytøen , E.R. , Eneqvist , H. , de Jong , N.L. , Vale , R. , Jorge , M.R. , Moser , M.B. , Moser , E.I. , 2022 . Large-scale two-photon calcium imaging in freely moving mice . Cell 185 , 1240 – 1256 . OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted June 28, 2025. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Amyloid pathology reduces dynamic range and disrupts neural coding in a mouse model of Alzheimer’s Disease Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Amyloid pathology reduces dynamic range and disrupts neural coding in a mouse model of Alzheimer’s Disease Mary Ann Go , Kathrine E. Clarke , Yimei Li , Seigfred V. Prado , Beatriz R. F. Teixeira , Jess J. Yu , Simon R. Schultz bioRxiv 2025.06.27.661987; doi: https://doi.org/10.1101/2025.06.27.661987 Share This Article: Copy Citation Tools Amyloid pathology reduces dynamic range and disrupts neural coding in a mouse model of Alzheimer’s Disease Mary Ann Go , Kathrine E. Clarke , Yimei Li , Seigfred V. Prado , Beatriz R. F. Teixeira , Jess J. Yu , Simon R. Schultz bioRxiv 2025.06.27.661987; doi: https://doi.org/10.1101/2025.06.27.661987 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Neuroscience Subject Areas All Articles Animal Behavior and Cognition (7635) Biochemistry (17697) Bioengineering (13895) Bioinformatics (41951) Biophysics (21456) Cancer Biology (18594) Cell Biology (25520) Clinical Trials (138) Developmental Biology (13381) Ecology (19903) Epidemiology (2067) Evolutionary Biology (24323) Genetics (15612) Genomics (22510) Immunology (17738) Microbiology (40401) Molecular Biology (17184) Neuroscience (88622) Paleontology (667) Pathology (2833) Pharmacology and Toxicology (4825) Physiology (7644) Plant Biology (15158) Scientific Communication and Education (2046) Synthetic Biology (4296) Systems Biology (9825) Zoology (2271)