4. DISCUSSION
In the current study, we compared in vitro hippocampal network oscillations in three commonly used inbred mouse strains (B6J, B6N, 129) with the aim of elucidating common oscillatory patterns in inbred mouse strains that show aberrant emotional/cognitive behaviour (B6N and 129). We further investigated the density of main GABAergic interneurons in the hippocampal subregions and assessed hippocampal excitability via measuring population spikes in the CA3 and CA1 subregions. We identified three common oscillatory features in B6N and 129 strains that diverged from the B6J strain: (1) Larger amplitude cholinergic gamma oscillations in the CA3; (2) Higher incidence of CA1 SW with a specific increase of SW propagation along the CA3-CA1 axis in the 129 strain; (3) Higher correlation of CA3 and CA1 SW transients. We further detected significant alterations in the number of specific inhibitory interneuron populations in the hippocampal subregions and activation of CA3 associative network that are essential for generation and maintenance of hippocampal network oscillations across three mouse strains. These observations align well with the notion that in vitro network oscillations provide useful reductionist models to investigate neurobiological underpinnings of phenotypic differences and highlight the importance of careful consideration of genetic background when designing experiments.
We used in vitro network oscillations that are dependent on the CA3 associative network activity as our experimental model. Hippocampal CA3 subregion forms an extensive associative network with abundant local recurrent excitatory connections that are under tight control of local inhibitory interneurons (Li et al. , 1994; Le Duigou et al. , 2014). These features make this circuit ideal for generation of local field potential network oscillations such as SW-R and gamma oscillations both in vivo and in vitro (Le Duigou et al. , 2014). Both oscillation types show similar features when recordedin vitro or in vivo (Butler & Paulsen, 2015; Çalışkan & Stork, 2018). Indeed, there is accumulating evidence that ex vivohippocampal network oscillations correlate with specific behaviours (Luet al. , 2011; Çalişkan et al. , 2016) or are modulated by behavioural manipulations (Albrecht et al. , 2013; Mizunumaet al. , 2014; Çalışkan et al. , 2015) suggesting that our findings might be relevant for behavioural phenotypes observed in investigated inbred mouse strains.
In our study, B6J strain was used as control strain as it has been shown to exhibit normal spatial learning, normal anxiety levels, normal contextual fear memory and extinction (Gerlai, 1998a, 1998b, 2002; Stiedl et al. , 1999; Rodgers et al. , 2002; Siegmundet al. , 2005; Camp et al. , 2012; Åhlgren & Voikar, 2019). Of note, both B6N substrain and 129 strains have been consistently shown to exhibit aberrant contextual fear memory / extinction and elevated anxiety levels in comparison to B6J strain (Stiedl et al. , 1999; Rodgers et al. , 2002; Siegmund et al. , 2005; Camp et al. , 2012; Åhlgren & Voikar, 2019). We used hippocampal slices obtained from the ventral-to-mid portion of the hippocampus. Importantly, ventral hippocampus have been repeatedly shown to be involved in stress adaptation, emotional memory formation and anxiety, while the dorsal hippocampus is implicated in cognitive functions such as spatial memory processing (Kjelstrup et al. , 2002; Fanselow & Dong, 2010; Strange et al. , 2014). Therefore, our findings might be more relevant to emotional phenotypes observed in these two mouse strains.
Both B6N and 129 strains show significantly increased number of CA1 SW events and correlation of SW events in the CA3 and CA1 (Fig. 1). Additionally, 129 strain shows a facilitated propagation of SW events from CA3 to CA1 region. These findings aligns well with our previous work demonstrating that both increased propensity for SW-R generation in the ventral CA1 and increased SW propagation along CA3-CA1 axis correlate well with the aberrant fear extinction and persistent fear memories (Çalişkan et al. , 2016). This is also supported byin vivo work showing causal involvement of SW-R in contextual fear memory consolidation (Wang et al. , 2015; Girardeau et al. , 2017; Ognjanovski et al. , 2017). Cholinergic hippocampal gamma oscillations (Fig. 2) are significantly enhanced in mice with anxious phenotype (B6N, 129) in comparison to a less anxious mouse strain (B6J). This observation also supports our previous study demonstrating reduced gamma oscillation in ex vivo slice preparations of mice with lower anxiety (Albrecht et al. , 2013). Indeed, heightened anxiety/emotionality appears to elicit increases ofin vivo gamma oscillations in the ventral hippocampus and hypersynchronous gamma activity has been related to emotional psychopathologies (Headley & Paré, 2013; Dunkley et al. , 2014, 2015; Stujenske et al. , 2014). Future studies testing whether such oscillatory patterns are also observed in vivo are needed to substantiate these in vitro findings. Furthermore, it would be interesting to test whether cholinergic markers that are important for gamma generation (e.g., M1 receptors) are altered in these mouse strains.
Augmented network oscillations might be due to an altered CA3 associative collateral network activation in B6N substrain and 129 strain (Çalışkan et al. , 2015). To assess the activation of CA3 associative collateral network (Fig. 3), we analysed antidromic populations spikes and synaptically-mediated orthodromic population spikes triggered by antidromic stimulation of CA3 subregion (Behrenset al. , 2005; Fano et al. , 2012). We found no major effect on the antidromic population spikes in the CA3 subregion except for a minor decrease in the 129 strain at higher stimulation strengths. However, the synaptically-mediated orthodromic population spikes were significantly reduced primarily in 129 strain. Local CA3 GABAergic interneurons control the excitation mediated by excitatory CA3-CA3 synapses (Li et al. , 1994; Le Duigou et al. , 2014). Thus, a reduction in the excitatory interactions within the CA3 recurrent collaterals might be due to an increased local inhibitory tonus in the 129 strain. Therefore, we analysed the expression of distinct markers for inhibitory interneurons that are crucial for generation and modulation hippocampal network oscillations (Fig. 4) (Mann et al. , 2005; Oren et al. , 2006; Ellender et al. , 2010). We used glutamate decarboxylase 67 (GAD67) as the pan-GABAergic marker. Furthermore, somatostatin (SST) and parvalbumin (PV) were used to quantify SST+ and PV+ interneuron populations. We observed a prominent overall increase in the number of GAD67+ interneurons in the hippocampal CA3-CA1 of 129 strain. This finding supports the notion that an increased inhibition might be present in the 129 strain which, in turn, can cause a reduction in the CA3 orthodromic population spikes for a given antidromic stimulation.
Changes in the number of interneuron populations can either cause the observed oscillatory features or reflect a compensatory mechanism. During both CA3 SW events and carbachol-induced gamma oscillation cycles, CA3 pyramidal cells receive a relatively stronger inhibitory input in comparison to an excitatory input (Oren et al. , 2006; Hajos et al. , 2013; Zemankovics et al. , 2013; Bazelotet al. , 2016). These compound intracellular excitatory/inhibitory events with a relatively stronger inhibitory component can be recorded as extracellular field potentials from the pyramidal cell layer as positive going voltage deflections (Bazelot et al. , 2010, 2016; Beyeler et al. , 2013). On the other hand, both perisomatic-region targeting PV+ interneurons and distal dendrite-targeting SST+ interneurons appear to receive relatively stronger excitation in comparison to inhibition during each gamma or SW event (Oren et al. , 2006; Hajos et al. , 2013; Pangalos et al. , 2013). Thus, a relatively increased inhibition during these oscillatory cycles can cause such an increase in gamma oscillation power as observed in the CA3 of both B6N and 129 strains.
Some evidence suggests that perisomatic inhibition onto pyramidal cells provided by PV+ basket cells is the source of gamma-range field potentials induced by carbachol (Gulyás et al. , 2010). Inconsistently, recent evidence shows a paradoxical increase in broadband gamma power in mice deficient for PV+ interneuron function, however, with reduced cellular synchronization to gamma oscillations (Guyon et al. , 2021). We observed a reduced PV+ interneuron number in both CA3 and CA1 of B6N strain. It is tempting to speculate that the reduction in PV+ interneuron number can cause the increased gamma power observed in the CA3 and CA1 of B6N strain. However, we detected an increased local gamma correlations in the CA3 and CA1 suggesting an increased gamma synchronization potentially mediated via increased PV+ interneuron functionality. Thus, as an alternative explanation, more efficient recruitment of PV+ interneurons by excitatory pyramidal cells could compensate for reduced PV+ interneuron abundance and increase perisomatic inhibition augmenting gamma field potential oscillations. Such increase in perisomatic inhibition can also lead to heightened SW amplitude as observed in the CA3 and CA1 of B6N strain. Furthermore, the reduction in the number of PV+ interneuron could alter the reciprocal interactions between PV+ interneurons during SW events that mediate the generation of fast ripple oscillations (Schlingloff et al. , 2014). In line, we observed a reduced CA3 ripple frequency in the B6N strain. Future studies assessing excitation-inhibition balance using intracellular recordings during field potential oscillations need to confirm whether such alterations are present in the B6N strain.
On the other hand, the increased number of SST+ interneurons observed in the hippocampus of 129 strain might underlie the increased gamma power. Indeed, a recent study suggests that SST+ interneurons are necessary for the maintenance of carbachol-induced gamma oscillations in hippocampal slices prepared similarly to our study (Antonoudiou et al. , 2020). Furthermore, the same study shows that optogenetic activation of SST+ interneurons leads to an increase in the peak frequency of cholinergic gamma oscillations in vitro . Indeed, in the 129 strain, the increased SST+ interneuron number was accompanied by faster gamma oscillation cycles evident by an increased peak frequency of gamma oscillations. However, it still remains to be elucidated how increased number of SST+ interneurons could directly affect the entrainment of pyramidal cells leading to augmented gamma oscillations as they mostly target the distal dendrites of the pyramidal neurons in the CA3 (Müller & Remy, 2014). Thus, it is likely that, during cholinergic gamma oscillations, they modulate perisomatic inhibition via altering the excitability of principal neurons and spiking precision of PV+ interneurons as previously described for the dentate gyrus circuit (Savanthrapadian et al. , 2014). Alternatively, changes in the GABA-A receptor subunit composition might also underlie the alterations in the peak frequency of cholinergic gamma oscillations as previously described (Heistek et al. , 2010).
Taken together, we identified several common oscillatory and cellular features in the hippocampus of inbred mouse strains with aberrant emotional/cognitive phenotype (B6N and 129). Noteworthy, while there is substantial evidence for behavioural differences among C57/BL6 substrains (B6N vs. B6J), to the best of our knowledge, our study is the first to show potential interneuronal and oscillatory correlates of these differences in C57/BL6 substrains. Collectively, our study provides further evidence for altered hippocampal physiology among commonly used inbred mouse strains and cautions scientific community to consider genetic background as an important variable when designing experiments and interpreting their results.