Abstract
During early development, memory systems gradually mature over time, in parallel with the gradual accumulation of knowledge. Yet, it is unknown whether and to what extent maturation is driven by discrete experience. Sleep is thought to contribute to the formation of long-term memory and knowledge through a systems consolidation process that is driven by specific sleep oscillations (i.e., ripples, spindles, and slow oscillations) in cortical and hippocampal networks. Based on these oscillatory signatures, we show here in rats that discrete spatial experience speeds the functional maturation of spatial memory systems during development. Juvenile male rats were exposed for 5 min periods to changes in the spatial configuration of two identical objects on postnatal day (PD)25, PD27, and PD29 (Spatial experience group), while a Control group was exposed on these occasions to the same two objects without changing their positions. On PD31, both groups were tested on a classical Object Place Recognition (OPR) task with a 3 h retention interval during which the sleep-associated EEG and hippocampal local field potentials were recorded. On PD31, consistent with forgoing studies, Control rats still did not express OPR memory. By contrast, rats with Spatial experience formed significant OPR memory and, in parallel, displayed an increased percentage of hippocampal ripples coupled to parietal slow oscillation-spindle complexes, and a stronger ripple-spindle phase-locking during the retention sleep. Our findings support the idea that experience promotes the maturation of memory systems during development by enhancing cortico-hippocampal information exchange and the formation of integrated knowledge representations during sleep.
SIGNIFICANCE STATEMENT Cognitive and memory capabilities mature early in life. We show here that and how discrete spatial experience contributes to this process. Using a simple recognition paradigm in developing rats, we found that exposure of the rat pups to three short-lasting experiences enhances spatial memory capabilities to adult-like levels. The adult-like capability of building spatial memory was connected to a more precise coupling of ripples in the hippocampus with slow oscillation-spindle complexes in the thalamo-cortical system when the memory was formed during sleep. Our findings support the view that discrete experience accelerates maturation of cognitive and memory capabilities by enhancing the dialogue between hippocampus and cortex when these experiences are reprocessed during sleep.
Introduction
The maturation of cognitive and memory capabilities during development is a slow process over time which is essentially driven by experience and resulting knowledge accumulating over time in parallel. The importance of prior knowledge for the development specifically of mnemonic capabilities has been already highlighted in the classical work of Piaget (1929). In his view, we integrate new incoming information from the environment in relation to our preexisting knowledge about the world, with knowledge accumulating at a particularly rapid rate in this manner during early life. Knowledge accumulation over time is commonly thought of as mediated by a systems consolidation process in which, for example, in the hippocampus-dependent episodic memory system, information is initially encoded in hippocampal networks to be subsequently gradually redistributed toward neocortical knowledge networks, including the mPFC (Sekeres et al., 2018). Such systems consolidation is indeed facilitated when relevant prior knowledge is available (Tse et al., 2007; van Kesteren et al., 2010), and it is also supported by sleep (Durrant et al., 2015; Groch et al., 2017; Klinzing et al., 2019).
Sleep is not only commonly thought to support maturational processes in general. There is also strong evidence that it specifically supports systems consolidation processes (Lewis and Durrant, 2011; Inostroza and Born, 2013), making sleep a likely candidate mediating cognitive maturation by facilitating the accumulation of knowledge. Active system consolidation of memories is established during sleep through the precise temporal coordination of three cardinal oscillations occurring during slow-wave sleep (SWS), that is, the neocortical slow oscillations (SOs), the thalamo-cortical sleep spindles, and hippocampal sharp-wave ripples, which coordinate the dialogue underlying the integration of episodic information into neocortical knowledge networks (Diekelmann and Born, 2010; Klinzing et al., 2019). There is strong evidence showing that, during SWS, spindles occur such that they are nested into the up-state of the SO, and that these SO-spindle complexes phase-lock hippocampal ripple activity to support cortico-hippocampal communication flow (Clemens et al., 2007; Staresina et al., 2015; Helfrich et al., 2019; Skelin et al., 2019; Oyanedel et al., 2020). The precise temporal coordination between these oscillations predicts consolidation of memories that rely on the cortico-hippocampal memory system (Mölle et al., 2009; Inostroza and Born, 2013; Maingret et al., 2016; Latchoumane et al., 2017; Rothschild et al., 2017; Todorova and Zugaro, 2020). Interestingly, these three cardinal sleep oscillations and the cortico-hippocampal memory system undergo remarkable changes across development (Campbell and Feinberg, 2009; Olini et al., 2013; Clawson et al., 2016; Lindemann et al., 2016; Timofeev et al., 2020; Zhang et al., 2021).
Here we demonstrate in juvenile rats that knowledge resulting from discrete spatial experience accelerates the maturation of spatial capabilities using an object-place recognition (OPR) task with a 3 h delay between encoding and test phase. With this long delay, typically only adult rats express robust recognition memory, and only when they sleep in the retention interval between encoding and test (Palchykova et al., 2006; Binder et al., 2012; Ishikawa et al., 2014; Sawangjit et al., 2018). Accordingly, here, we took advantage of a developmental time point (PD31) during which juvenile rats do not yet express adult-like spatial memory (Contreras et al., 2019). Before testing on the OPR task, one group of rats was exposed on 3 different days (for 5 min periods) to changes in the spatial configuration between two identical objects located in an arena. A Control group of rats was subjected to basically the same contextual experiences, except that, instead of changes in the spatial configurations, these rats experienced that the configurations between the two objects remained the same. We found that on PD31, only the rats with Spatial experience, but not the Control rats, were able to form a significant OPR memory, which was paralleled by an increased occurrence of hippocampal ripples and a stronger coupling of SO-spindle complexes over the parietal cortex with hippocampal ripples.
Materials and Methods
Animals
A total of 43 male Long-Evans rats were used for the experiments. Rat pups arrived (from Janvier) in our facilities on postnatal day (PD) 7 or 8 in litters of 5 or 6 pups for the first behavioral experiments, or in litters of 2 pups for the experiments, including electrophysiological recordings. After arrival, the litters were kept undisturbed for at least 7 d to allow acclimatization. The weaning day was on PD21 for animals of the purely behavioral experiments and on PD20 for the electrophysiological experiments. The animal colony was kept at a room temperature of 22 ± 1°C, on a 12 h/12 h light/dark cycle (lights on at 6:00 h). All rats had free access to food and water throughout the experiments. All experimental procedures were performed in accordance with the European animal protection laws (Directive 2010/63/EU, European Community) and were approved by the Baden-Württemberg state authority.
Experimental design and conditions
The first purely behavioral experiments and the second experiments with electrophysiological recordings each included two groups: a Spatial experience group (first experiment n = 12, second experiment n = 10) and a Control group (n = 11 and n = 10). In these groups, each animal underwent prior experimental “experiences” on PD25, PD27, and PD29, and the final OPR test on PD31 (see Fig. 1a). (Behavioral data for PD25 of the Spatial experience group was part of a previous study published previously (Contreras et al., 2019). After the weaning day, rats were kept in pairs or trios until the end of the experiments. Rats of the second experiment underwent surgical implantation on PD20 or PD21 balanced across groups. To enable the same degree of recovery, in the latter rats, experimental experiences and final OPR testing were postponed by 1 d. For simplicity, and because exploratory analyses revealed no differences between the rats that underwent surgery on PD20 or PD21, we refer here in all cases to the same time schema with PD31 denoting the day of the final OPR task.
For the Spatial experience groups, each of the three spatial experiences comprised the exposure to a change in the spatial configuration of two identical objects, following the principles of an OPR paradigm with a first 5 min exposure period (encoding) during which the rat encountered the arena with two identical objects, and a second 5 min period (retrieval) where the rat returned to the same arena but now one of the objects was displaced to a new location. The two periods of these prior experiences were separated by 3 h. Different spatial configurations and objects were used on each occasion. Rats of the Control groups were exposed to the same arena and objects at the same time points as in the Spatial experience groups with the only difference that, during the two periods of the prior experience on 1 d, the objects remained stationary. The final OPR test on PD31 was identical for the Spatial experience and Control groups and followed the same design as used for the prior experiences in the Spatial experience group, but with different objects and a different spatial configuration of the objects. In the second experiment, electrophysiological recordings were performed during the 3 h retention period between the encoding and retrieval phase of the OPR task.
Experimental procedures
All experiments were run between 7:00 and 15:00 h (i.e., the light phase). The rat's behavior was monitored by a video camera and analyzed offline by experienced experimenters using ANY-maze software (Stoelting Europe). A minimum of 1 s of exploration toward each object during the encoding phase of the OPR test was used as inclusion criterion (only 1 rat of the implanted Control group did not meet this criterion).
Before the experiments proper, all rats received 5 sessions of handling (one per day). During this period, pups later undergoing surgical implantation were familiarized with humid pellets and milk to ensure regular consumption after the weaning/surgery day. Before the first experimental experience, rats underwent three habituation sessions (once per day) consisting of 10 min inside the arena followed by 6 h spent in the home cage with their siblings in the experimental room. Implanted rats spent these 6 h in a recording box in isolation. During the 6 h, rats remained undisturbed.
Apparatus and objects
The arena was a quadratic dark gray open field (43 cm × 43 cm, height of walls 35 cm), and the objects were glass bottles of different shapes (height 10-18 cm), filled with sand of different colors. They had sufficient weight to ensure that the rats could not displace them. For each of the four exposures, different pairs of objects were used and placed at different locations, with the spatial configuration of each pair counterbalanced across rats. The same objects were used for the Spatial experience and Control groups. The identity of the object was also counterbalanced across occasions (i.e., which pair of objects was used at the first, second, etc. occasion). Objects and arena were cleaned thoroughly between trials with 70% ethanol solution.
The arena was placed at the center of the room and surrounded by a circular black curtain, with the south side of the curtain used as the entrance for the experimenter. To promote an allocentric spatial orientation, rats were introduced into the arena each time from a different side and facing the wall. Additionally, three distal cues were placed surrounding the arena (two yellow crosses made of cartoon; size 24 × 24 cm2; attached to the black curtains, and a 3D pinata with a star shape size 20 × 20 × 20 cm3 hanging from the ceiling at the west side of the open field). Two fluorescent strip lights at the side and below the arena provided indirect light. White noise was presented at a constant intensity during all procedures to mask any disturbing sounds.
Surgical implantations and histology
Rats of the electrophysiological experiments were anesthetized with an intraperitoneal injection of fentanyl (0.005 mg/kg of body weight), midazolam (0.15 mg/kg), and medetomidin (2 mg/kg). They were placed into a stereotaxic frame and supplemented with isoflurane (0.5%) if necessary. The scalp was exposed, and four holes were drilled into the skull. Three EEG screw electrodes were implanted: one frontal electrode (AP: 3.0 mm, ML: −1.5 mm, relative to bregma), one parietal electrode (AP: −3.0 mm, ML: 2.5 mm, relative to bregma), and one occipital reference electrode (AP: ∼−10.0 mm, ML: ∼−0.1 mm, relative to bregma). Additionally, one stainless-steel electrode was implanted to record local field potential (LFP) signal into the left dorsal hippocampus (dHC; AP: −3.0 mm, ML: 2.5 mm, DV: −2.5 mm). Electrode positions were confirmed by histologic analysis (see Fig. 2b). Animals whose LFP electrode position could not be verified were discarded from the dHC electrophysiological analysis. For EMG recordings, a stainless-steel wire was implanted in the neck muscle. Electrodes were connected to a 6-channel electrode pedestal (PlasticsOne) and fixed with cold polymerizing dental resin, and the wound was sutured. Rats had 3 d for recovery.
After the last recording session, rats were terminally anesthetized with fentanyl (0.01 mg/kg of body weight), midazolam (0.3 mg/kg), and medetomidin (4 mg/kg). The electrode's positions were marked by electrolytic lesion (10 μA, 30 s). Rats were perfused with physiological saline (50-100 ml) followed by 4% PFA (200-300 ml). After decapitation, the brains were removed and postfixed in 4% PFA for 1 d. Coronal sections (60 µm) were cut using a vibratome, stained with 0.5% toluidine blue, and examined under a light microscope.
Electrophysiological recordings
Rats were habituated to the recording box (dark gray PVC, 30 × 30 cm, height: 40 cm) for 3 d, 6 h per day (right after the habituation to the arena). The third habituation day included recordings. During electrophysiological recordings, the electrodes were connected through a swiveling commutator to an amplifier (model 15A54, Grass Technologies), with the rat's behavior being continuously tracked using a video camera mounted on the recording box. EEG, LFP, and EMG signals were digitalized (sampling rate 1 kHz) using a CED Power 1401 converter and Spike2 software (Cambridge Electronic Design). Signals were amplified and filtered between 0.1 and 300 Hz (EEG), 0.1 and 1000 Hz (LFP), and 30 and 300 Hz (EMG), respectively.
Sleep stage determination and event detections
Sleep stages and wakefulness were determined offline based on EEG and EMG recordings, using standard visual scoring procedures for consecutive 10 s epochs as previously described in Oyanedel et al. (2020). In addition to wakefulness, three sleep stages were discriminated: SWS, intermediate stage (IS), and REM sleep. Wakefulness was identified by mixed-frequency EEG and sustained EMG activity, SWS by the presence of high-amplitude low-frequency activity (δ activity: <4.0 Hz) and reduced EMG tone, REM sleep by low-amplitude EEG activity with predominant theta activity (5.0-10.0 Hz), phasic muscle twitches, and a decrease of EMG tone. IS was identified by a decreased δ activity, progressive increase of theta activity, and an increased σ power (10-16 Hz) in the 10 s epoch. Recordings were visually scored by an experienced experimenter.
Slow oscillation (SO) and spindle events were detected in the frontal and parietal EEG, as described by Mölle et al. (2009) and Oyanedel et al. (2020). For SO identification, the EEG signals were filtered between 0.3 and 4.5 Hz (Butterworth filter of third order). An SO event was selected if the following criteria were fulfilled: (1) two consecutive negative-to-positive zero crossings of the signal occurred at an interval between 0.4 and 2.0 s; (2) of these events in an individual rat and channel, the 35% with the highest negative peak amplitude between both zero crossings were selected; and (3) of these events, the 45% with the highest negative-to-positive peak-to-peak amplitude were selected. The criteria resulted in the detection of SOs with downstate peak amplitudes exceeding −80 μV and peak-to-peak amplitudes exceeding 120 μV.
For the detection of spindles, the EEG signal was filtered between 7.0 and 20.0 Hz. Then, the envelope was extracted via the absolute value, that is, the instantaneous amplitude, of the Hilbert transform on the filtered signal, followed by an additional smoothing (moving average with 200 ms window size calculated by the Smooth function of MATLAB). A spindle was identified when the absolute value of the transformed signal exceeded 1.5 SDs of the mean signal in the respective channel, during a rat's SWS epochs, for at least 0.4 s and not >2.0 s. Spindle onset and end were defined by the times when the signal the first time exceeded the 1.5 SD threshold, and again fell below this threshold, respectively. Spindle power was calculated based on the integral of the envelope of the Hilbert-transformed signal between spindle onset and end. For calculating Hilbert transformations, the MATLAB function Hilbert was used. The envelope was extracted using the MATLAB function abs, which returns the absolute value (modulus), that is, the “instantaneous amplitude” of the transformed signal.
Ripples were identified in dHC LFP recordings as described by Oyanedel et al. (2020). The signal was filtered between 150 and 250 Hz (in exploratory analyses, prefiltering for more narrow bands, e.g., 145-245 Hz, was used to assure artifact-free ripple detection). These analyses confirmed the present results and are not reported here). As for spindle detection, the Hilbert transform was calculated, and the signal was smoothed using a moving average (window size 200 ms). A ripple event was identified when the smoothed Hilbert transform value exceeded a threshold of 2.5 SDs from the mean smoothed Hilbert transform of the filtered signal during an animal's SWS epochs, for at least 25 ms (including at least three cycles) and for not >500 ms.
Analysis of event co-occurrence
An SO and a spindle event were considered as co-occurring whenever the maximum amplitude of a spindle was located within the two positive-to-negative zero crossings of the identified SO. Similarly, a ripple-spindle co-occurrence was determined when the beginning and the end of a detected ripple occurred within the beginning and end of a detected spindle. Thus, a triple coupling (i.e., ripple coupled to an SO-spindle complex) was marked when a ripple occurred within a spindle whose maximum amplitude occurred within a detected SO. Event correlation histograms were calculated as supporting analysis for the ripple-spindle co-occurrence. For these histograms, a window size of 2 s (±1 s) around the spindle onset (trigger event) was used. During this time, ripple peaks and troughs were identified and accumulated across spindle events for bins of 100 ms. Counts in every bin were divided by the number of spindles and the bin width to give ripple event rates per millisecond.
In a more fine-grained analysis of the temporal coordination between spindles and ripples, we calculated the phase of the spindle oscillation in which the ripples preferentially occurred. For this, the SWS epochs were first filtered in the original 7-20 Hz range used for spindle detection at 5 different subfrequency bands (7-10, 10-12, 12-14, 14-16, and 17-20 Hz). Then, for each individual spindle event (with a given frequency), the corresponding filtered frequency band was chosen to extract the phases. For this, the “Hilbert” and “angle” functions of MATLAB were used. The CircStat toolbox (Berens, 2009) was used for calculating the average preferred phase.
Time-frequency plots were calculated to identify changes in ripple power after the spindle onset. For this, time-frequency analysis for frequencies between 150 and 250 Hz was performed in a window of ±1.5 s around the spindle onset using the FieldTrip toolbox (http://fieldtriptoolbox.org, function mtmconvol) (Oostenveld et al., 2011). The analysis was done in steps of 1 Hz using a sliding Hanning tapered window with a variable frequency-dependent length that always comprised 9 cycles. Time-locked power values for each frequency of each event were normalized by dividing the value by the average power during a prior baseline interval from −0.5 to 0 s relative to the spindle onset, using the function ft_freqbaseline, baselinetype: “relative” of FieldTrip. Similarly, time-frequency plots of EEG power were calculated with reference to the SO negative peaks (±1 s) comprising frequencies between 0.01 and 20 Hz in steps of 0.25 Hz using a sliding Hanning tapered window with a variable frequency-dependent length that always comprised three cycles. Power values were normalized here by dividing the value by the average power during a prior baseline interval from −2 to −1 s.
Data reduction and statistical analyses
Statistical analyses of behavioral parameters were calculated using SPSS software (version 26.0 IBM). Before the evaluation of memory performance, behavioral control parameters, such as exploration to the objects, distance traveled, and time spent in the center of the arena, were examined during the encoding phase of the OPR test and outliers (values > ±2 SDs from the group's mean) were removed to assure homogeneity in the rats' performance (1 case in the Control group of the first behavioral experiment, 1 case each in the Spatial experience and Control group of the second electrophysiological experiment). Analyses of memory performance concentrated on cumulative data across the first 3 min of the retrieval phase as this initial interval is known to most sensitively reflect memory assessed by the response to novelty (Winters et al., 2004; Sawangjit et al., 2018). Spatial memory performance was analyzed using the object discrimination index (DI) defined by the formula: DI = [(exploration time for novel object-location – exploration time for familiar object-location)/(exploration time for novel object-location + exploration time for familiar object-location)]. Two DI outliers were identified, 1 in each the Spatial experience and the Control group of the first experiment. Memory was statistically inferred from a DI value above chance level tested using one-sample t tests. Differences between groups in memory performance were evaluated using independent t test. Statistical differences for the behavioral experiments did not change when the analyses were exempt from any exclusion criteria. Changes in the DI across prior experiences in the Spatial experience group were evaluated using linear regression analysis with Age as independent and DI as dependent variable.
The rat pups undergoing surgery for implantation of electrodes in the second electrophysiological experiment showed more variable behavior, as confirmed also by an exploratory ANOVA across both experiments revealing, in addition to a Spatial experience/Control main effect (F(1,36) = 8.26, p = 0.007), significance for the covariate first/second experiment (F(1,36) = 4.95, p = 0.033). To account for this nonspecific increase in variance, linked to the surgery and related procedural changes, analyses of the electrophysiological experiments focused on animals with DI values deviating less than ±2.5 SD from the respective group mean of the behavioral experiments (leading to removal of 3 cases in the Spatial experience and 2 cases in the Control group). Electrophysiological analyses concentrated on the 3 h retention period during the final OPR task (see Fig. 3a). Event rates, as well as general sleep parameters, were first analyzed by a global ANOVA (with degrees of freedom Greenhouse–Geisser-corrected) including a Group factor (Spatial experience vs Control) and, where appropriate, a repeated-measures factor Area (Frontal, Parietal), followed by a post hoc independent t test and corroborated by a nonparametric test (i.e., Mann–Whitney U) in case of small sample size. Event correlation histograms (for ripple event rates) were compared between groups using cluster-based permutation tests (Monte Carlo Method, 1000 permutations, cluster α < 0.05). Bins of the event correlation histogram found to be part of a significant cluster were further compared using a linear mixed-effects model analysis, including event rates as fixed effect and animals as random effect (using the fitlme function of MATLAB). We used the Rayleigh test to evaluate the preferred phase of an event (e.g., ripples preferentially occurring during a certain phase of the spindle cycle). Preferred phase comparisons between groups were evaluated using the Watson Williams test (i.e., the circular equivalent to a two-sample t test). Time-frequency plots were compared between groups using cluster-based permutation tests calculated with the ft_freqstatistics function of FieldTrip (1000 permutations cluster α = 0.05, statistic = ft_statfun_indepsamplesT).
Results
Prior spatial experience enhances the capability to express adult-like OPR memory in developing rats
The rats that were exposed to the spatial experiences on PD25, PD27, and PD29 showed significant adult-like OPR memory (i.e., a discrimination index during the OPR test phase that was significantly above zero, when tested on PD31; t(10) = 5.240, p < 0.001, for one-sample t test against chance level). In contrast, the Control rats that had not been exposed to the spatial experiences before showed null memory expression (t(8) = −1.423, p = 0.193, Fig. 1b). Moreover, memory performance of the Spatial experience group was higher than that of the Control group (t(18) = 4.637, p < 0.001, for an independent-sample t test). Analyses of control variables at the encoding phase of the OPR task revealed that the groups showed comparable total object exploration time (all p > 0.106). However, the distance traveled and time spent in the center of the arena were longer in the Spatial experience group than in the Control group (both p < 0.01, Fig. 1c), suggesting that prior spatial experiences also shape locomotory activity in an unspecific manner. Analysis of these control variables during the first exposure phase on PD25 did not reveal any differences between the groups (all p > 0.306, Fig. 1d), confirming that baseline exploratory behavior was comparable between groups.
Prior spatial experiences accelerate adult-like expression of spatial memory. a, Study design: Two groups of juvenile rats were exposed to experiences on PD25, PD27, and PD29, followed by testing spatial memory on an OPR task on PD31. For the rats of the Spatial experience group (n = 11), prior experiences comprised a change in the spatial configuration of two identical objects with each experience set up like an OPR task, including a first (encoding) period during which the rat encountered the two identical objects in the arena and, delayed by 3 h, a second (retrieval) period in which 1 of the objects was displaced to a new location. Different spatial configurations and objects were used for each experience. For the rats of the Control group (n = 9), the experiences were set up identically, except that in the second (retrieval) period both objects remained at the same place (i.e., stationary). OPR testing on PD31 was performed with novel objects and also a 3 h delay between encoding and retrieval testing. b, Mean ± SEM memory performance (discrimination index) at OPR testing on PD31 for the Spatial experience and Control groups (dot plots overlaid). Only rats of the Spatial experience group showed above chance OPR memory performance, which was also significantly higher than in the Controls. c, Mean ± SEM total object exploration (in seconds, left), distance traveled (in meters, middle), and time spent in the center of the arena (in seconds, right) during the encoding phase of the OPR task on PD31. d, The same control parameters as in c, but for the first (encoding) period of the prior experience on PD25. e, Line graph represents mean ± SEM memory performance (discrimination index) for the three prior experiences of the Spatial experience group (dot plots overlaid). Memory expression shifts from a significant preference for the object in the familiar location (negative discrimination index) to an adult-like preference toward the object in the novel location (positive discrimination index). +p = 0.026 (linear regression with age). ###p < 0.001; #p < 0.05; one-sample t test against chance level. ***p < 0.005; **p < 0.01; pairwise between-groups comparisons (two-sided t test).
Analysis of memory performance during prior spatial experience
The prior experiences were set up basically in the same way as the OPR task used at PD31; that is, each experience comprised a first 5 min encoding phase with two identical objects placed in the arena and, delayed by 3 h, a test phase with the change in the spatial configuration such that one of the objects was displaced to a novel location (Fig. 1a). This allowed a supplementary analysis of memory performance in terms of discrimination indices across the three exposures on PD25, PD27, and PD29. Interestingly, these analyses revealed a progressive switch in the memory expression shifting from an exploration preference for the object in the familiar location (i.e., a negative discrimination score; t(10) = −2.952, p = 0.014) to a preference toward the object in the novel location (i.e., a positive discrimination score, as indicated by a significant linear regression across age, R2 = 0.15, F(1,32) = 5.451, p = 0.026, Fig. 1e). Significance of the negative discrimination score on PD25 indicates that, already at this age, object location memory is present, although its behavioral expression is opposite to that on PD31 and in adult rats.
Prior spatial experience increases the temporal coordination between SOs, spindles, and ripples
The effects of prior spatial experience on oscillatory signatures of memory processing were investigated in two new groups of rats with implanted electrodes for surface EEG (frontal, parietal) and dHC LFP recordings. Recordings were obtained during sleep in the 3 h retention interval at OPR testing on PD31. OPR performance on PD31 in the Spatial experience group of these experiments was above chance level (t(5) = 2.654, p = 0.045), and higher than the Control group (t(10) = 2.438, p = 0.035 for independent-sample t test, Fig. 2c), Analyses of the behavioral control variables did not reveal any significant differences between groups during the encoding phase of the OPR test (all p > 0.159) or during the first period of the prior experience on PD25 (all p > 0.087, Fig. 2d), with the latter finding confirming that baseline exploratory behavior was comparable between the groups.
Prior spatial experience does not affect SO-spindle co-occurrence or the phase-coupling of spindles to the SO cycle. a, Left, Schema of recordings of EEG from skull electrodes over left frontal and right parietal cortex, and of LFPs from left dHC. EMG was recorded from the neck muscle. Reference: occipital skull electrode. Right, General procedure for the implanted rats, undergoing surgery on PD20, Spatial experience versus Control experience between PD25-PD29, and OPR testing on PD31. (For half of the rats, surgery and behavioral tests took place 1 d later; see Materials and Methods). EEG/LFP signals were recorded during the 3 h retention period of OPR testing on PD31. b, Coronal histologic sections showing LFP electrode sites (dots) in the dHC. c, Mean ± SEM memory performance (discrimination index) at OPR testing on PD31 for the Spatial experience and Control groups (dot plots overlaid, each group n = 6 rats). #p < 0.05 for one-sample t test against chance level. *p < 0.05 for pairwise between-groups comparisons (two-sided t tests). d, Mean ± SEM total object exploration (in seconds, left), distance traveled (in meters, middle), and time spent in the center of the arena (in seconds, right) during the encoding phase of the OPR task on PD31 (top) and for the first (encoding) period of the prior experience on PD25 (bottom). There were no significant differences between the Spatial experience and Control groups. e, Mean ± SEM percentage of spindles co-occurring with an SO in recordings from (left) frontal and (right) parietal areas for the Spatial experience and Control group (dot plots overlaid). f, Circular histograms of preferred phase for the occurrence of spindles during an individual SO in frontal (left) and parietal (right) recordings for the Spatial experience (up, frontal: n = 113 co-occurring events, parietal n = 97) and Control group (bottom, frontal, n = 126, parietal, n = 89; 20°/bin, SO negative peak at 180°). Orientation of red line within the histogram indicates average phase which was, in the Spatial experience group −4.02° and 7.09° for frontal and parietal recordings, respectively. For the Control group, the respective phase angles were 17.5° and −16.95°. Phase-locking of spindles to the SO cycle was significant in both groups and at both recording sites (**p < 0.01; ***p < 0.001), with no significant differences between groups (all p > 0.08).
Total sleep time during the OPR retention interval on PD31 as well as the percentage of time spent in the different sleep stages (i.e., SWS, IS, and REM sleep) were comparable between the groups (Table 1). Analyses of EEG SOs and spindles, and ripples in dorsal hippocampal LFP recordings during SWS epochs revealed that, on average, 18.76 ± 1.27% and 22.13 ± 2.04% of detected spindles co-occurred with an SO in frontal and parietal EEG recordings, respectively, with no differences between groups (all p > 0.222, Fig. 2e). Also, in both groups, spindles co-occurring with SOs were tightly phase-locked to the up-state of the SO, over frontal and parietal cortex areas (all p < 0.002 for Rayleigh tests) with no significant differences between groups (Fig. 2f).
Sleep architecture and features of SOs, spindles and ripplesa
To assess the integration of hippocampal with thalamo-cortical memory-related activity, we determined the percentage of ripples identified in dHC recordings that were coupled to an SO-spindle complex (i.e., ripples that occurred during an ongoing spindle of an SO-spindle complex). Interestingly, the percentage of ripples in this way coupled to an SO-spindle complex was higher in the Spatial experience than Control group, with this difference restricted to the parietal cortex (F(1,6) = 6.238, p = 0.047 for an Area × Group interaction, t(6) = 2.545 p = 0.044 for post hoc t test on parietal recordings; Fig. 3a). Consistent with this finding, the Spatial experience group also showed a greater percentage of ripples that generally co-occurred with parietal spindles (regardless of whether or not it was part of an SO-spindle complex) compared with the Controls (t(6) = 0.222, p = 0.016, Fig. 3b).
Prior spatial experience increases co-occurrence of parietal SOs, spindles, and hippocampal ripples. a, b, Mean ± SEM percentage of ripples co-occurring with frontal and parietal (a) SO-spindle complexes and (b) spindles, for the Spatial experience and Control group (dot plots overlaid). Note the enhanced percentage of hippocampal ripples co-occurring with parietal SO-spindle complexes and spindles in the Spatial experience group. *p < 0.05, pairwise between-groups comparisons (two-sided t tests). c, Example of ripple events co-occurring with a parietal SO-spindle complex. From top to bottom: Raw parietal EEG signal with dashed lines framing interval enlarged below. Parietal EEG filtered between 0.3 and 4.5 Hz, for detection of SOs, and filtered between 7 and 20 Hz, for detection of spindles, and LFP signal from dHC filtered between 100 and 250 Hz, for detection of ripples. Red represents detected SO and spindle ripples (red cross on top for coupled ripples).
Prior spatial experience strengthens the phase-coupling between spindles and ripples
In a more fine-grained analysis, we assessed the phase-locking of hippocampal ripples to the spindle oscillation, for ripples that co-occurred with parietal spindles. To this end, we calculated the mean circular vector to quantify the phase of the individual EEG spindle oscillation in which the highest peak of a hippocampal ripple occurred. We found for the rats of the Spatial experience group a significant phase-locking of ripple events preferentially occurring in the up-to-down phase of the spindle cycle (42.59 ± 0.686° for circular mean, p = 0.008), whereas no such phase-locking was found in the Control group (p = 0.234, for Rayleigh tests, F(1,61) = 5.16, p = 0.026 for difference between groups for Watson-William test, p = 1.476, p = 0.22, for nonparametric multisample test of equal median directions, Fig. 4a). Thus, prior spatial experience appears to boost the phase-locked occurrence of ripples within the individual spindle oscillation. Corresponding analyses for frontal cortical spindles did not reveal consistent phase-locking of ripples (all p > 0.529, for Rayleigh test).
Prior spatial experience strengthens parietal spindle ripple phase-coupling. a, Circular histogram of preferred phase for occurrence of ripples during individual oscillations of parietal spindles for the Spatial experience (left, n = 48 co-occurring events) and Control (right, n = 14) groups (20°/bin, spindle trough at 180°). Orientation of red line within the histogram indicates average phase (42.5° and −25.7° for Spatial experience and Control groups, respectively). Phase-locking of ripples (to the peak-to-trough transition) of the spindle cycle is significant only in the Spatial experience group (**p < 0.01). b, Event correlation histograms of the occurrence of ripple events time-locked to the onset of parietal spindles (0 s) across animals. Event rate refers to ripple peaks and troughs. Gray shaded area represents a significant increase in ripple events in the Spatial experience group (n = 151 events) compared with the Control group (n = 122) as revealed by cluster-based permutation analysis (**p < 0.007). c, Time-frequency plots of power in the ripple frequency band (150-250 Hz) during the first 0.5 s after parietal spindle onset (0 s) for the Spatial experience (top, n = 298 spindles) and Control groups (middle, n = 300). Power is color-coded and normalized (i.e., divided by the average power during a 0.5 s baseline interval before spindle onset). Bottom, Map of t values for the comparison between groups. Significant cluster is outlined (p = 0.004, cluster-based permutation test).
We further explored the modulation of hippocampal ripples by parietal spindles by creating event correlation histograms for ripple events (defined by their peaks and troughs) time-locked to the spindle onset. A cluster-based permutation analysis revealed a significant increase in ripple events extending over the first 200 ms after spindle onset in the Spatial experience group compared with the Controls (p = 0.007, cluster α = 0.05, 1000 permutations, Fig. 4b). This finding was confirmed in a linear mixed-effects model analysis (including groups as a fixed factor and animals as a random factor) excluding the possibility that the increase in the ripple event rate following spindle onset was driven by values of single animals (p < 0.006, for the respective Group main effect). The increase in hippocampal ripple activity following onset of parietal spindles was likewise obtained in a time-frequency analysis of power in the ripple frequency band (150-200 Hz) performed time-locked to the parietal spindle onset. This analysis showed a significant increase in ripple band power within 100 ms following spindle onset in the Spatial experience group, compared with the Control group (p = 0.004, α = 0.05, 1000 permutations, Fig. 4c).
Prior spatial experience does not change rates of SOs, spindles, and ripples themselves
The effect of prior spatial experiences on the temporal coordination of sleep oscillations was not accompanied by general changes in the density or other features of these events (Table 1). Overall density of SO, spindle, and ripple events was well comparable between the Spatial experience and Control groups (all p > 0.105, Fig. 5a). Also, amplitude and power of SOs and spindles, respectively, as well as mean spindle frequencies and duration did not significantly differ between the groups (all p > 0.066). Independent of the experimental group, SO density was generally higher and spindle density was generally lower in parietal than frontal cortical recordings (F(1,10) = 17.554 and 63.560, respectively, p ≤ 0.002, for ANOVA main effects of Area; Fig. 5). In parallel, SO amplitude was higher and spindle power was lower over parietal than frontal cortex (F(1,10) = 11.260 and 6.671, p = 0.027, for main effects of Area).
Prior experience does not affect SOs, spindles, or ripples themselves. a, Mean ± SEM density (events/min) of SOs (left), spindles (middle) detected in frontal and parietal EEG, and of ripples (right) detected in the dHC LFP for the Spatial experience and Control groups (dot plots overlaid). Densities differed depending on the area, but not between the groups (***p < 0.001, **p < 0.01; for Area effects). b, Characterization of sleep oscillations in the Spatial experience (top panels) and Control group (bottom panels). Left, Time-frequency plots of power in the 0.01-20 Hz frequency band with the grand mean (± SEM) SO overlaid time-locked to SO negative peak (0 s) over frontal (left) and parietal (right) cortex. Power is color-coded and normalized (i.e., divided by average power during a 1 s baseline interval). Middle, Spindle grand average in the filtered signal time-locked to the maximum trough (0 s) with the group's mean frequency indicated above. Right, Grand average of detected ripples in the filtered (150-250 Hz; left) and raw signal (right) with the group's mean frequency indicated above.
Spatial experience enhances the distance traveled and the time spent in the center of the arena
Repeated experience of changes in spatial configurations might advance expression of OPR memory by fostering spatial and exploratory behaviors that regulate how the animal samples the relevant environmental information. This is suggested by the finding of our behavioral experiments indicating an increased distance traveled and time spent in the center of the arena during the encoding phase of the OPR task in the Spatial experience group compared with the Control group (Fig. 1d). To follow this question, we analyzed the dynamics of the behavioral control variables (total object exploration, distance traveled, time spent in the center of the arena) across the (first) encoding periods of the experiences on PD25, PD27, PD29, and of the OPR task on PD31 (Fig. 6). To increase statistical power, the analyses were run on the pooled data from both behavioral and electrophysiological experiments (after significant differences between the experiments had been excluded by prior comparisons, all p > 0.183). Total object exploration decreased across experiences but did not differ between the Spatial experience and Control groups (F(3,58) = 13.216, p < 0.001, for main effect of Experience in a 2 (Groups) × 4 (Experiences) ANOVA), likely reflecting unspecific contextual habituation. Interestingly, with repeated experiences, the distance traveled tended to increase in the Spatial experience group, but not in the Control group. Indeed, pairwise testing revealed significance for the respective group differences at the two last encoding experiences (on PD29 and PD31) but not at the two first experiences (Fig. 6b), although the respective Groups × Experience interaction remained without significance (p = 0.3; F(1,30) = 8.129, p = 0.008, for Group main effect). This divergent dynamics between the groups was even more pronounced for the time the animals spent in the center of the arena (F(3,90) = 5.395, p = 0.002; Fig. 6c). These findings indeed suggest that the repeated exposure to spatial configurational changes specifically shapes the way the animals move around the arena and how they gather and encode the relevant environmental information.
Behavioral control parameters during the first (encoding) phase of the experiences on PD25, PD27, PD29, and of the OPR task on PD31. a, Mean ± SEM total object exploration (in seconds). b, Distance traveled (in meters). c, Time spent in the center of the arena (in seconds) during the encoding phase on PD25, PD27, PD29, and PD31. *p < 0.05; **p < 0.01; post hoc independent t test. Analyses were performed on pooled data from the behavioral and electrophysiological experiments. n = 17 for Spatial experience group; n = 15 for Control group.
Discussion
Our study was based on the idea that the maturation of cognitive capabilities during early development is in part a consequence of an accumulation of knowledge over time (Piaget, 1929). Spatial capabilities mature because persisting knowledge forms that integrates multiple spatial experiences over time. Sleep is of primary importance for this maturation process because it is thought to support the formation of integrative knowledge across experienced episodes by transforming the newly acquired episodic memories through an active system consolidation process that eventually enables the assimilation of the episodic information residing in hippocampal networks, into neocortical knowledge networks (Inostroza and Born, 2013; Klinzing et al., 2019). Against this conceptual backdrop, the present experiments aimed to clarify how several discrete spatial experiences feed into the maturation of spatial capabilities during early development, focusing in particular on brain oscillatory signatures of active systems consolidation during sleep.
Our findings in juvenile rats show that three discrete spatial experiences (i.e., 5 min exposures to a change in the spatial configuration of two identical objects), occurring between PD25 and PD29, are sufficient to promote the adult-like expression of an OPR memory when these rats were tested on PD31. By contrast, the rats of the Control group, which did not experience the specifically spatial changes in the configuration of two identical objects but otherwise were exposed to identical contextual experiences (i.e., the same arena including the same objects), did not express any significant OPR memory at PD31. This confirms previous findings indicating that, at this age, rats are typically still not able to behaviorally express an adult-like OPR memory (Contreras et al., 2019). These behavioral findings are thus in line with the notion that prior spatial experiences can accelerate the maturation of spatial capabilities.
The present findings regarding brain oscillatory signatures of memory processing during sleep, and specifically the enhanced coupling of ripples with spindles and SO-spindle complexes in the rats of the Spatial experience group, are of even greater relevance here, inasmuch these findings point to an accelerated functional maturation of the machinery mediating knowledge formation during sleep. Ripples occurring during SWS accompany and may even drive the replay of neuronal firing patterns in hippocampal place cell ensembles that had been activated for the encoding of spatial information during prior wakefulness (Girardeau and Zugaro, 2011; Drieu et al., 2018; Todorova and Zugaro, 2020), and can thus be generally considered as signs of reactivation of episodic memory information in hippocampal networks. During early development, the replay of hippocampal place cell activity linked to sharp wave ripples emerges between PD17 and PD32, gradually comprising longer trajectories that cover more of the environment, thereby eventually enhancing the efficacy of mapping for linking the objects to the spatial positions in which they were experienced (Muessig et al., 2019).
Hippocampal ripples tend to nest into the excitable troughs of the spindle oscillation, and spindles tend to nest into the excitable upstate of the SO in thalamo-cortical circuitry (Sirota et al., 2003; Clemens et al., 2011; Staresina et al., 2015; Oyanedel et al., 2020). There is growing evidence that this triple-nesting of ripples, spindles, and SOs mediates systems consolidation during sleep (Maingret et al., 2016; Latchoumane et al., 2017; Niethard et al., 2018; Varela and Wilson, 2020) and, specifically, the integration of hippocampal episodic memory information into neocortical knowledge networks (Diekelmann and Born, 2010; Klinzing et al., 2019). Notably, the present experiments do not reveal any alterations in the expression of ripples, spindles, or SOs themselves after spatial experience, but rather a distinctly enhanced coupling of hippocampal ripples to thalamic-cortical spindles and SO-spindle complexes. That means, prior spatial experience leads to a better, and presumably more effective, coupling of episodic memory reactivations in hippocampal networks with memory processing in cortical knowledge networks, with this more effective coupling presumably facilitating the integration of the reactivated episodic information into cortical knowledge representations.
Interestingly, prior spatial experiences strengthened the coupling of hippocampal ripples with thalamo-cortical spindles and SO-spindle complexes specifically in recordings over the parietal cortex but not in frontal cortical recordings. One possible explanation for this preferential connecting of hippocampal ripples to parietal spindles and SO-spindle complexes refers to the pattern of cortical synaptic maturation which follows a posterior-to-anterior progression (Kurth et al., 2010; Buchmann et al., 2011; Gerván et al., 2017). Thus, cortico-hippocampal communication during SWS may preferentially involve parietal oscillations because the parietal cortex matures earlier than the PFC (Casey et al., 2005). However, more distinct coupling of hippocampal ripples to parietal than frontal spindles is a consistent finding also in adult rats (Clemens et al., 2011; Oyanedel et al., 2020). Hence, another and, perhaps more plausible, explanation is that ripples were coupled to parietal, and not to frontal sleep oscillations because the parietal cortex plays an essential role in the processing of spatial information, as it has been observed in adult studies in humans and rodents (Save et al., 1992; Whitlock et al., 2008; Chersi and Burgess, 2015; Brodt et al., 2016). The parietal cortex together with the hippocampus forms part of a wider spatial memory system involved in the conversion of egocentric to allocentric spatial mapping and respective organization of locomotory activity (Whitlock et al., 2008; Bicanski and Burgess, 2018). The offline processing of prior spatial experiences might, thus, specifically serve to potentiate the development of this entire network.
What kind of knowledge accumulates over time to advance spatial capabilities in the developing rats? Although in our task paradigm at each of the three experimental experiences preceding the OPR test at PD31, different objects and a different spatial configuration of these objects were introduced, the arena context, including the experimenter, remained the same across these experiences, and were also the same at OPR testing on PD31. Hence, the young rats may have simply learned an association between the arena context and the occurrence of a change in the spatial configuration of the two objects in this arena (i.e., an associative representation which first generalizes over the different types of objects and configurations of the three prior experiences) and finally also to the objects and configuration used in the OPR test at PD31. Such simple learning of a generalized association between context and the occurrence of any spatial configurational changes might well explain the enhanced and adult-like memory performance at PD31 OPR testing in the rats of our Spatial experience group. But can it also explain the improvement in the coupling of hippocampal ripples to spindles and SO-spindle complexes which occurred in these rats at PD31 during sleep after encoding of the OPR task, and before the test phase, given the fact that the stimulus configuration during the encoding phase at OPR testing on PD31 was identical for the Spatial experience and Control groups, and also that the spatial configurational changes (during the prior experiences and OPR testing) always occurred with a rather long delay of 3 h? Particularly, in light of the long 3 h delay, a simple form of associative learning seems not to be a plausible mechanism explaining the enhanced hippocampal-neocortical coupling during postencoding SWS in the Spatial experience group. Alternatively, we propose that, as a consequence of the three prior spatial experiences, generalized representations are formed which reside in parietal cortical knowledge networks and associate the arena context with possible spatial configurational changes. Assuming that hippocampal ripples reflect the reactivation of newly encoded episodic memory information (Diba and Buzsáki, 2007; Girardeau and Zugaro, 2011; Drieu et al., 2018; Todorova and Zugaro, 2020), for example, during OPR encoding before sleep, the enhanced coupling of ripples to thalamo-cortical SO-spindle complexes may then reflect that the parietal cortical networks containing the relevant spatial information from prior experiences, connected to the arena-related episodic information that was newly encoded in hippocampal networks. In this view, the relatively diminished coupling of memory processing during sleep between hippocampal and thalamo-cortical networks in the Control rats would reflect the lack of spatial change in the prior experiences of these animals.
Our findings complement studies demonstrating a close positive correlation between the maturation of temporal SO-spindle coupling during SWS and the improvement of hippocampal-dependent memory performance across development (Muehlroth et al., 2019; Hahn et al., 2020). Enhanced SO-spindle coupling was also found to predict overnight memory consolidation in adults as well as children (Helfrich et al., 2018; Kurz et al., 2021). However, these studies were performed in healthy humans and did not include recordings of hippocampal ripples. Therefore, it could only be speculated about how ripples during sleep feed into the formation of long-term memory and knowledge, presumed to underlie the maturation of cognitive capabilities. Interestingly, deviating from those human studies, our manipulation of prior experience did not affect the coupling between SO and spindles. This difference might be partly related to our experimental design where, unlike in those human studies, the encoding phase before sleep did not serve the presentation of rather high amounts of novel information (to study their consolidation into long-term memory) but, instead, aimed to activate knowledge already existing in cortical long-term stores. On the other side, our finding of effects of prior knowledge selectively pertaining to the coupling of hippocampal ripples to signs of memory processing in thalamocortical networks, underline the key role hippocampal ripples play for processing and integrating spatial information during sleep (Girardeau et al., 2009; Girardeau and Zugaro, 2011; Fernández-Ruiz et al., 2019). To our knowledge, the present study is the first to show the existence and the experience-dependent emergence of a precisely tuned synchronization of hippocampal ripples to the spindle oscillations and to SO-spindle complexes during development, pointing to an involvement of this coupling in the formation of long-term memory during development.
A surprising secondary finding of our study was revealed by analyses of the rats' behavior during the spatial experiences before OPR testing on PD31, indicating that the rats were basically able to form an OPR memory already on PD25. Notably, unlike OPR memory on PD31 in these rats, this memory on PD25 expressed itself in a negative discrimination index (i.e., a preferential exploration of the stationary rather than displaced object, which is not unusual in very young rats) (Contreras et al., 2019), and it occurred in the absence of any prior spatial experience with similar configurational changes. In combination, these differences tempt us to speculate that expression of these object-place memories in the animals on PD25 relies on separate representations (i.e., a kind of harbinger memory not involving neocortical knowledge representations but solely hippocampal representations), which eventually may also be more crude and mainly used to identify (and avoid) any kind of novelty in the environment (Berlyne, 1950; Dix and Aggleton, 1999).
In addition, the analyses of the rats' behavior during the spatial experiences on the days before OPR testing revealed that the experienced changes in spatial configuration induced a systematic change in spatial behaviors (i.e., an increase in distance traveled and the time the pups spent in the center of the arena with repeated experiences) compared with the Control group. How such alterations in spatial behavior, perhaps by changing the animal's way to encode the relevant information, contributing to the development of the hippocampal memory system, are in need of further investigations. Indeed, since our Control group did not experience any change in the object configuration during the second period of each experimental experience, the experiences of spatial novelty may have potentiated the development of mnemonic capabilities in a nonspecific manner (e.g., by increasing the levels acetylcholine or noradrenaline boosting synaptic connectivity in the relevant areas) (Hasselmo, 2006; Barry et al., 2012; Kjaerby et al., 2022). This hypothesis could be tested by examining whether the effects of prior spatial experience generalize to other forms of hippocampal learning.
In conclusion, our findings support the idea that prior experiences through driving the formation of integrated knowledge representations in cortical networks promote the maturation of cognitive capabilities during development. The idea of repetitive experiences in the hippocampus-dependent episodic memory system, promoting during early development the maturation of capabilities in the same cognitive domain, has been proposed in foregoing studies showing that prior experience produces signs of a lasting upregulation of plasticity (immediate early-gene activation, upregulation of excitatory synaptic markers, etc.) in hippocampal networks (Bessières et al., 2020). Our results extend those findings to the systems level by showing that prior hippocampal-dependent experiences also strengthen the cortico-hippocampal communication during postencoding sleep. Indeed, providing evidence that the accelerated behavioral expression of adult-like spatial memory by prior experiences is paralleled by an enhanced coupling of hippocampal ripples with corticothalamic signatures of memory processing during sleep, our experiments identify sleep (and associated active systems consolidation processes) as a candidate mechanism mediating knowledge accumulation over time and, thus, substantially contributing to the functional maturation of cognitive capabilities in the developing brain.
Footnotes
This work was supported by Deutsche Forschungsgemeinschaft Grant DFG In 279/1-1 to M.I.; and European Research Council ERC AdG 883098 SleepBalance to J.B. M.I. was supported by the Hertie Foundation Hertie Network of Excellence in Clinical Neuroscience. We thank Ilona Sauter for technical assistance; and Carlos Oyanedel, Ernesto Duran, Shan Xia, Eva-Maria Kurz, and Ioannis Zouridis for help with conducting the experiments and data analysis.
The authors declare no competing financial interests.
- Correspondence should be addressed to Marion Inostroza at marion.inostroza{at}uni-tuebingen.de or Jan Born at jan.born{at}uni-tuebingen.de
