|Home | About | Journals | Submit | Contact Us | Français|
Via Ca2+-imaging in freely behaving mice that repeatedly explored a familiar environment, we tracked thousands of CA1 pyramidal cells' place fields over weeks. Place coding was dynamic, for each day the ensemble representation of this environment involved a unique subset of cells. Yet, cells within the ~15–25% overlap between any two of these subsets retained the same place fields, which sufficed to preserve an accurate spatial representation across weeks.
CA1 place cells are considered crucial for spatial memory, but data is limited regarding whether their representations of space evolve over time scales of weeks or more1. Some theories suggest place cells should retain stable place fields for long-term retention of familiar environments1. Alternatively, dynamic aspects of place coding may facilitate distinct memory traces of different events occurring in the same environment2.
Due to technical limitations, it has been only partially explored if CA1 representations of familiar environments are stable or evolve over time. Electrical recordings from many tens of cells are feasible3, but it is challenging to record from the same cells longer than a few days. Data on place fields' stability has largely been from small numbers of cells recorded over at most a week4-10. These studies have demonstrated cells with stable place fields, but the data have been too sparse to assess how coding evolves at the ensemble level.
To study long-term coding dynamics, we combined (Fig. 1a): a viral vector (AAV2/5-CamKIIα-GCaMP3) to express the Ca2+-indicator GCaMP311 in pyramidal cells; a chronic mouse preparation for time-lapse imaging of CA1 over weeks12; and a miniaturized (<2 g) microscope for Ca2+-imaging in hundreds of cells in freely behaving mice13. We thereby tracked somatic Ca2+ dynamics of 515–1040 pyramidal cells in individual mice as they repeatedly visited a familiar track over 45 days.
We first verified CA1 neurons' place coding attributes as mice explored various arenas. The data revealed up to 740 cells (range: 73–740 cells; n = 13 mice) undergoing Ca2+ excitation in single fields of view (Fig. 1b–d; Supplementary Fig. 1; and Movie 1). Ca2+ dynamics generally displayed quiescent periods interrupted by prominent transients. This concurs with in vitro studies showing GCaMP3 reports spike bursts well but for solitary spikes has weak signals easily masked by background fluorescence or noise11. We used established computational means14 to extract individual cells and their dynamics from each session's Ca2+-imaging data, without regard to mouse behavior (Online Methods).
As expected of place cells, during active exploration many pyramidal cells exhibited Ca2+-excitation when the mouse occupied a specific portion of its arena (Fig. 1d). When we placed mice in two different arenas at the same location in the room but with distinct shape, color and orientation cues, a subset of cells exhibited re-mapping2, showing spatially distinct patterns of Ca2+-excitation in the two arenas. Consistent with prior work, some cells had place fields in only one arena. Thus, one can optically detect CA1 place cell activity in freely behaving mice, in accord with a Ca2+-imaging study in mice exploring a virtual reality15.
To study place cells over weeks, we trained mice to run back and forth on a linear track; Ca2+-imaging occurred on ten sessions over 45 days (Fig. 2a). As in prior studies in linear environments16, many cells had clear place-coding properties that usually depended strongly on the mouse's running direction (Fig. 2b–d). Overall, ~90% of cells had at least one Ca2+ transient during running.
For detailed analyses we focused on a subset of mice (n = 4) and used a conservative definition of place field by requiring statistically significant mutual information between a cell's Ca2+ excitation events and the mouse's location17. With this definition, ~20% of cells had place fields for left, right, or both running directions (Fig. 2c–e). The set of place fields fully covered the track, with the ends represented more densely than the interior (Fig. 2f,g). The mean place field size was ~27% of the 84-cm track, within the range published for mice15,16,18,19. For each place field we detected Ca2+ activity in 17 ± 14% of running passes (n = 1656 place fields, mean ± s.d.; range: 2-87%). Across days 5–35 the percentages of cells on each day with place fields for right (12 ± 1%; mean ± s.e.m) or left (12 ± 1%) motion did not vary (Kruskal-Wallis ANOVA; P = 0.77 and 0.88 for right and left, respectively; n = 7 sessions; n = 4 mice) (Fig. 2e). Nor were there changes in the distributions of place fields' locations or sizes (Fig. 2g,h) (Kolmogorov-Smirnov test; P=0.06–0.99 for locations and 0.02–0.99 for sizes, both compared to a significance threshold of 2.4·10-3 that includes the Dunn-Sidák correction for the 21 pairwise comparisons). Notably, we saw no discernible changes to cells' morphologies nor substantial changes in mean Ca2+-transient amplitudes or baseline fluorescence within or across sessions (Supplementary Fig. 2). Thus, photobleaching was negligible, and neither GCaMP3 expression nor illumination had perceptibly deleterious effects on cell health.
To register repeated observations of individual cells, we first examined the precision of image registration (Supplementary Fig. 3). Bootstrap analyses showed errors in aligning cells' locations across sessions were <1 μm. This precision more than sufficed, as even the closest cells had ≥6 μm between centroids. Over the full study each mouse yielded 515–1040 cells (n = 4 mice), more than the maximum (740) seen in one session but consistent with anatomical data.
A majority of cells was active in one or two sessions (57 ± 1%; mean ± s.d.; n = 2960 cells; n = 4 mice). 2.8 ± 0.3% were active in all 10 sessions (Fig. 3a,b). Yet, each session had the same percentage (31 ± 1%) of active cells out of the full tally (Kruskal-Wallis ANOVA; P = 0.46) (Fig. 3b inset). Cells came in and out of this active subset day-by-day, but the overlap in active subsets from any two days was only moderately time-dependent, declining from ~60% for sessions 5 days apart to ~40% for 30 days apart (Fig. 3c).
Comparisons between any two sessions revealed ~15–25% overlap in the subsets of cells with statistically significant place fields, declining from ~25% for sessions 5 days apart to ~15% for 30 days (Fig. 3c). Notably, when individual cells did show place fields in more than one session, the place fields' locations were generally identical (Fig. 3d). This is a compelling, independent validation of our image registration protocol. Though cells came in and out of the place-coding ensemble, place fields' invariant locations plus the slowly declining overlap in place-coding ensembles led to spatial representations that retained a clear resemblance while decaying over time (Fig. 3e–g).
We next sought factors that influenced cells' recurrences in the place-coding ensemble. If cell physiological or coding parameters were key influences, Ca2+-activity or place-coding parameters might correlate with recurrence probabilities. But if network dynamics were more important, the data might reveal no relationships between cells' characteristics and recurrence probabilities. Notably, the numbers of sessions in which cells had Ca2+ activity or statistically significant place fields were uncorrelated with their rates and amplitudes of Ca2+ activation (Supplementary Fig. 4). Cells with high place-coding stability in single sessions had virtually the same recurrence odds as other cells (Supplementary Fig. 5). Neither inclusion of Ca2+ transient amplitudes in the computations of place fields nor variations in how we extracted cells from the raw data altered these findings (Supplementary Figs. 6, 7). Overall, we failed to find parameters predictive of which cells recur in either the active or place-coding ensembles.
Given place fields' invariant locations, did the ~15–25% overlap between different days' coding ensembles suffice to retain a stable spatial representation? To address this, we used Bayesian decoding techniques to study how well we could reconstruct the mouse's location from the Ca2+-imaging data (Supplementary Fig. 8 and Fig. 3h–j). We created a set of decoders of a common mathematical structure, trained each decoder on a portion of one day's data, and tested it on other data. When test and training data were from the same day, estimates of mouse location were excellent (median error nearly always <7 cm) and highly significant compared to shuffled test data (P<10-160; Kolmogorov-Smirnov test). We then asked how well a decoder trained on data from one day would perform on data from other days. In comparisons between decoders using the same number of cells, performance declined only modestly with the interval between training and testing and remained very significant for 30-day intervals (P=10-27; Kolmogorov-Smirnov test). Thus, the ~15% commonality in place-coding subsets across 30 days sufficed to deduce the mouse's trajectory using a decoder trained on data of 30 days prior.
Though GCaMP3 does not faithfully report single spikes11, our approach can sense isolated spike bursts. To evade analyses of place coding by using only solitary spikes, cells would have to avoid burst spiking across entire sessions while still encoding spatial information. We do not exclude this possibility but consider it unlikely, especially given the key place coding role ascribed to bursts20 and lack of correlation here between cellular Ca2+ activity and involvement in place coding. Improved Ca2+ sensors should reveal a greater portion of spiking activity and could amend our findings with GCaMP3. Our general approach will allow long-term tracking of large neural ensembles in multiple brain areas beyond CA1.
Overall, our data indicate retention of spatial information in CA1 combines stable place field locations with ~15–25% odds an individual cell will recur in the ensemble place code. Prior long-term recordings had stressed place field stability and usually focused on tens or fewer cells, far less than the ~3500 we studied. In vivo imaging allows reliable tracking of CA1 cells over months12, here revealing the fluctuating membership of place coding ensembles. This supports prior reports of individually stable place fields but shows CA1 coding has day-to-day dynamism at the cellular level while preserving spatial information in the ~15–25% overlap between coding ensembles from any two days. Conversely, each episode in a familiar arena has a unique signature via the ~75–85% of cells that do not overlap when comparing coding ensembles from any two sessions (Fig. 3c). Note that our data show the existence of these non-overlapping signatures but do not imply any functional significance. One possibility is that coding turnover is a long-term form of the spike-rate re-mapping seen over shorter intervals2 and might help distinguish traces of distinct events occurring in the same environment.
University of North Carolina Vector Core packaged AAV2/5 vectors (~2·1012 particles/mL) expressing GCaMP3 via the CaMKIIα promoter21. We used immuno-staining of virally infected CA1 tissue to verify GCaMP3 and CaMKII co-expression in the same cells.
Stanford APLAC approved all procedures. Male C57/BL6 mice (aged 8–12 weeks at start, housed 2–3 per cage with a running wheel) underwent two procedures under isoflurane (1.5–2%). We first injected AAV2/5-CaMKIIα-GCaMP311,15 (250 nL) into CA1 (–1.9 mm from Bregma, 1.4 mm mediolateral, –1.65 mm dorsoventral). A week after viral transduction we implanted a glass guide tube just dorsal to CA1, as described12.
We used the integrated microscope as described for imaging CA113, with minor adaptations for time-lapse studies12,22. The first session (~4 weeks after second surgery) began by installing the microendoscope into the guide tube of isoflurane-anesthetized mice, guided by two-photon imaging of CA1 through the microendoscope12,22,23. The microendoscope was a gradient refractive index lens (GRINtech GmbH, 0.44 pitch length, 0.47 NA) and relayed light from CA1 to a focal plane outside the mouse. After verifying GCaMP3 expression, we fixed the microendoscope in the tube using UV-curing adhesive (Norland, NOA 81).
We lowered the integrated microscope towards the microendoscope until we saw GCaMP3 fluorescence using the microscope's LED light source (0.05–0.2 mW). After finding a suitable imaging site, we attached to the cranium the microscope's base plate using dental acrylic and Cerebond. This plate stayed with the mouse even when the microscope was detached. We generally darkened the acrylic with carbon powder (Sigma, 484164).
Mice displayed vigorous activity ~1–2 min after release from anesthesia. We chose isoflurane for its rapid clearance from tissue but nevertheless waited 20–30 min before imaging. Illumination (<0.4 mW) lasted ~3 min per imaging trial. Each session on the track involved 4–7 trials, over which the mouse usually ran >100 roundtrip passes. During ~3 min between trials the mouse rested in a holding chamber. After all trials we waited another 10–15 min, then briefly (~5 min) re-anesthetized the mouse to detach the microscope. A typical session yielded <25 min of video (19.9 Hz; 480 × 480 pixels covering ~0.34 mm2 of CA1).
In following sessions we re-attached the microscope to its base while the mouse was isoflurane-anesthetized (~5 min), then waited 20–30 min before imaging. We verified the field of view matched prior sessions or made slight focal adjustments13. Subsequent steps were as above.
During Ca2+-imaging the mouse explored: a square 46 × 46 × 15 cm3 arena (acrylic); a circular arena (21-cm-radius; red plastic); or an 84 × 4.5 × 4.5 cm3 elevated linear track (aluminum). For three days prior to Ca2+-imaging on this track we trained water-scheduled mice to run back and forth for water rewards at the ends. An overhead camera (Prosilica, EC640) recorded this behavior, using infrared LEDs (Lorex, VQ2120) and dim room lights for illumination.
We analyzed videos using MATLAB (Mathworks) and set all pixels to zero or one if their intensities were, respectively, above or below 10% of the median intensity. This demarcated the mouse due to its dark fur. We determined the mouse's position as the centroid of each binary image and calculated its velocity after smoothing the position data (0.5 s sliding average).
Analysis used ImageJ (NIH) and MATLAB routines. Since the microscope's sensor had a Bayer color filter13, we zeroed all pixels in the red and blue channels and de-mosaiced GCaMP3 signals in the green pixels by Bayer interpolation using the MATLAB function demosaic(). Image rows were readout successively; to correct for the slightly variable number of LED pulses illuminating each row we normalized each demosaiced pixel by the mean intensity in its row. The illumination exhibited mild spatial non-uniformity, so we also normalized each pixel by the ratio of the mean intensity along its column to that of a reference column. We coarse-grained images to 240 × 240 pixels each the mean of four pixels at the finer density.
We used rigid image registration to correct lateral displacements of the brain. We created an image stack, F2(t), as the difference between the original stack, F(t), and a smoothed version of F(t) (20-pixel-radius smoothing filter). Within F2(t) we selected a high-contrast sub-region to provide a fiducial marker. To mutually register all frames of F2(t) we used an ImageJ plug-in based on the TurboReg algorithm24. For each registered frame of F2(t) we applied the same coordinate transformation to F(t), yielding the registered stack F′(t).
As typical for Ca2+-imaging, we re-expressed registered images as relative changes in fluorescence, ΔF′(t)/F′0 = (F′(t) – F′0)/F′0, where F′0 is the mean image obtained by averaging the entire movie. We identified spatial filters corresponding to individual cells using an established cell sorting algorithm that applies principal and independent component analyses13,14,25. Cells' spatial filters were based on Ca2+ activity (temporally down-sampled 4×) over the entire session, not just when the mouse was running. For each filter we zeroed all pixels with values <50% of that filter's maximum intensity.
We used each cell's thresholded spatial filter to extract its Ca2+ activity from the ΔF′(t)/F′0 stack. We removed baseline fluctuations (ascribed to Ca2+ activity outside the focal plane or in neuropil) by subtracting the median trace (200 time bins sliding window) and applied a 5-frame (~250 ms) sliding average. We identified Ca2+-transients by searching each trace for local maxima that had: (1) Peak amplitude more than two standard deviations (2σ) from the trace's baseline; (2) ≥10 frames (~0.5 s) when the mean intensity surrounding the peak was >2σ; (3) Separation of >6 frames (~300 ms) from adjacent Ca2+-transients. We set a Ca2+-transient's occurrence to the temporal midpoint in the rise to peak fluorescence from the most recent trough, approximating a time midway within the corresponding spike burst. To correlate Ca2+ activity to mouse behavior, we offset Ca2+-transient occurrences by ~250 ms due to GCaMP3's known delayed response11.
On ~7% of all detected Ca2+-transients, fluorescence increases occupied more pixels than a single spatial filter. To mitigate effects of this spillover we took a conservative approach, allowing only one cell among a group of neighbors to register a Ca2+-transient within a ~250 ms window. We defined neighbors as cells whose spatial filters had non-zero pixels within 30 μm of each other. If multiple Ca2+-transients arose within ~250 ms among neighboring cells, we retained only the transient with the greatest peak ΔF′(t)/F′0 value.
We mapped all cells from each session by assembling their thresholded spatial filters onto a single image. Picking one day's map for reference (usually day 15), we aligned the others to this via a scaled image alignment using TurboReg24 (Supplementary Fig. 3a). This corrected slight translations, rotations, or focus-dependent magnification changes between sessions and yielded each cell's location in the reference coordinate system.
Next, we visually identified candidate cells across sessions that might be the same neuron seen on multiple occasions. We applied two observations: our registration procedures had sub-micron precision (Supplementary Fig. 3b–e); the distance between centroids of neighboring somata was always >6 μm (Supplementary Fig. 3f). We thus considered a candidate set of cells to be the same neuron if all pair-wise separations were ≤6 μm. If any of the pair-wise separations exceeded 6 μm we split the set into two or more.
To analyze place fields we identified ‘movement periods’ when the mouse ran continuously >0.5 cm/s. Additionally, in open field arenas the speed had to exceed 1 cm/s at some point during the movement; on the track it had to transiently exceed 9.2 cm/s. These criteria rejected small movements such as grooming, rearing, or head-turning.
On the linear track we considered 3.5-cm spatial bins and excluded the last 7 cm at each end where water rewards were given. In open field arenas, bins were 4 cm2. We divided the number of Ca2+ transients in each bin by the mouse's total occupancy time there, applied a Gaussian smoothing filter (linear track: σ = 8.75 cm; open field: σ = 3.5 cm), and normalized each place field by its maximum value. On the track we separately considered place fields for left and right running directions. The number of bins in which a place field had a value ≥50% of its maximum determined the place field's width. We tabulated each place field's position as its centroid.
For each place field (calculated for one running direction) we computed the mutual information26 between Ca2+-transients and the mouse's location (7-cm bins). We also performed 10,000 distinct shuffles of the Ca2+-transient times and calculated the mutual information for each shuffle. This yielded the P-value of the true mutual information relative to the shuffles. P ≤ 0.05 indicated a significant place field for that running direction.
To generate the null hypothesis for place fields' displacements between a pair of days, we used the place fields' measured locations but shuffled cells' identities on each of the days. We calculated the distribution of all displacements, averaged over 1000 distinct pairs of shuffles. Fig. 3e shows the mean null hypothesis curve found by averaging over all pairs of days.
Supplementary Figure 1 Ca2+ dynamics of CA1 pyramidal cells visualized during active mouse behavior.
Supplementary Figure 2 Fluorescence imaging with the integrated microscope caused no photobleaching or evident deterioration in neuronal health.
Supplementary Figure 3 Image registration had 1-µm accuracy and enabled identification of the same CA1 neurons for over a month.
Supplementary Figure 4 Activity-related parameters predict neither the re-appearance of neurons across imaging sessions nor the odds that CA1 neurons retain place fields.
Supplementary Figure 5 The coding properties of cells with especially stable place fields in individual imaging sessions evolve similarly over the long-term as other place cells.
Supplementary Figure 6 The long-term evolution of place-coding properties is robust to changes in the threshold used for cell identification.
Supplementary Figure 7 Inclusion of the amplitudes of Ca2+-activation in the data analyses did not alter conclusions regarding the timeevolution of neural coding.
Supplementary Figure 8 Bayesian decoding of place cells' Ca2+ dynamics permits accurate reconstruction of the mouse's trajectory.
Supplementary Video 1 (or Audio or Spreadsheet) Movie 1 | Ca2+-imaging in hundreds of CA1 pyramidal cells in a freely behaving mouse.
A video showing a mouse exploring a circular arena (left panel) and the simultaneously acquired brain-imaging data of CA1 pyramidal cell Ca2+ activity, displayed as relative changes in fluorescence (ΔF/F) (right panel). 705 pyramidal cells were identified in the total data set and correspond to the neurons of Fig. 1b. The Ca2+-imaging frame rate was 20 Hz, but these data are shown down-sampled to 5 Hz to aid visualization of the Ca2+ transients. The video playback rate is sped up four-fold from how the events actually occurred.
We thank L. Looger for GCaMP3 plasmid and A. Attardo, T. Davidson, J. Fitzgerald, J. Li, J.Z. Li, A. Lui, C. Ramachandran, O. Yizhar, and T. Zhang for helpful conversations and assistance. We appreciate fellowships from Stanford (LDB, EOH, LJK), NSF (LDB, LJK), Simons (LJK), Machiah (YZ), and Rothschild Foundations (YZ), and research funding to MJS from the Paul G. Allen Family Foundation, NIMH, NIA, and an NIH Director's Pioneer award.
Author contributions: YZ, LDB, and MJS designed experiments. YZ, LDB, and EOH acquired data. LDB and LJK analyzed data. LDB, EDC, and KKG built equipment. YZ, LDB, LJK and MJS wrote the paper. AEG and MJS supervised the research