In the current study we examined the effect of mutant tau (P301L) on synaptic plasticity and behavior in aged JNPL3 mice in a C57BL/6 background, a mouse model for tauopathy. We find that aged JNPL3 (BL6) mice show altered hippocampal long-lasting LTP, behavioral abnormalities such as deficits in both short and long term contextual fear memory and reduced sensorimotor gating. Electrophysiological effects in JNPL3 (BL6) mice were corrected by enhancing GABAergic function in the hippocampus. Consistent with a loss of GABAergic tone in Tg mice, we found a reduced number of GABAergic interneurons in area CA1 and dentate gyrus of the hippocampus. Collectively, our data show for the first time that mutant tau has toxic effects on GABAergic interneurons that progresses with age, leading to a loss of GABAergic function, altered hippocampal synaptic plasticity, and impaired memory and sensorigating in aged JNPL3 (BL6) mice.
Aged JNPL3 (BL6) mice show tau pathology consistent with an age-progressive onset of tau pathological markers (Figure 1). Pathology was found in CA1 hippocampal region, especially in the molecular layer containing the dendrites. The tau pathology we detect is, however, less robust compared to JNPL3 mice in a mixed background . This is consistent with previous reports showing that JNPL3 (BL6) mice display a milder pathology than in a mixed background . These findings, combined with ours, establish again that the strain background is an important factor to consider when studying mouse models of human disease. Although the pathology we observe is milder, we think that our study using this C57BL/6 background is very important because it is more comparable to other AD mouse models which use the same genetic background, for example TauRD (pro and anti-aggregant) [19, 45] and hAPPJ20 .
We found that expression of P301L tau in aged Tg mice did not impair either basal synaptic transmission or transient forms of synaptic plasticity in the hippocampal Schaffer collateral-area CA1 synaptic circuit. Although other reports have shown alterations in short-term synaptic changes in the hippocampus of AD mouse models [17, 47], little was known about long-lasting synaptic changes in aged JNPL3 (BL6) mice. Surprisingly, we found that L-LTP was enhanced in aged JNPL3 (BL6) mice (Figure 2a, Additional file 3: Figure S3). This enhanced L-LTP was rescued with treatment with zolpidem, a GABAA-receptor agonist (Figure 2b). These results suggest that GABAergic function is impaired in JNPL3 (BL6) mice. Indeed, supporting this idea, we found that hippocampal GABAergic interneurons in area CA1 were reduced in Tg mice (Figure 3). The L-LTP rescue we observed with zolpidem treatment was not due to over activation of GABAARs in the CA3-CA1 circuit, as we titrated zolpidem to preclude detectable effects on the L-LTP responses in WT slices. These results, combined with our observations that I/O and PPF were essentially normal in aged JNPL3 (BL6) mice, suggest that although GABAergic interneuronal function is compromised, homeostatic compensation is able, at least partially, to maintain essential circuit integrity in the aged JNPL3 (BL6) hippocampus. Functional deficits may only appear after the synaptic circuit is challenged by activity above ‘nominal’ levels, i.e. in response to behavioral experience or strong stimulation. Future studies, using a more detailed examination of single cell properties may reveal more subtle GABAergic deficits not detectable using field recording approaches. Additionally, it would be interesting to examine if these mice have altered synaptic plasticity in other brain areas, such as the dentate gyrus, where we also found a loss of GABAergic interneurons and a co-localization of pathological tau markers with GABAergic interneurons (Additional file 6: Figure S6 and Additional file 4: Figure S4). Additionally, other AD mouse models have shown GABAergic interneuron and synaptic plasticity deficits in the dentate gyrus [15, 17, 47, 48]. Strikingly, aged JNPL3 (BL6) mice had severe memory deficits and impaired PPI that may be explained by the loss of inhibitory control involved in memory formation and sensorimotor gating. Finding enhanced L-LTP and impaired behavioral performance may seem at first glance to be counterintuitive, but other mouse models have shown enhanced LTP and impaired memory [49–51]. Regardless, these results clearly show that tau plays an important role, separate from Aβ, in synaptic plasticity mechanisms.
Enhanced hippocampal LTP was also reported in very young P301L mice (5–7 weeks of age), but only in the dentate gyrus . This group also showed that young P301L mice had improved performance in the novel object recognition assay. Since no tau pathology was present at this age, the authors suggested that the over-expression of tau resulted in improved trafficking of glutamate receptors, enhancing synaptic transmission and improving memory. Alternative tau-dependent mechanisms may regulate hippocampal synaptic plasticity in an age-dependent fashion. Indeed, this may be likely as we found that aged JNPL3 (BL6) mice with mild but detectable tau pathology showed enhanced L-LTP but also severe memory deficits. Our data show the age-progressive loss of GABAergic interneurons in area CA1, and these data coupled with our zolpidem rescue data support the idea that tau-mediated regulation of GABAergic function is responsible for the enhanced L-LTP we detected. This notion is further supported by the impaired PPI we observe, a behavior that depends on GABAergic function, including in the hippocampus [43, 52]. Interestingly, co-staining of pathological-tau antibodies with GABAergic interneuron markers showed extensive co-localization in the hippocampus (Figure 4). This finding indicates that pathological tau is present in GABAergic interneurons. One explanation could be that the prion promoter, which is used to drive the transgene in JNPL3 (BL6), has greater activity in GABAergic interneurons than in pyramidal cells, resulting in greater expression in interneurons. However, this is unlikely because the prion promoter has been shown to be mainly active in the excitatory neurons within the pyramidal layer in the hippocampus . Therefore, it might be that tau pathology in GABAergic interneurons is developed from pathogenic isoforms contributed extracellularly or that GABAergic neurons are particularly susceptible to pathogenic tau. More research is necessary to study the role of GABAergic interneurons in the development of pathology and memory loss in AD.
In AD, it is known that cholinergic and glutamatergic neurotransmission are disrupted, while inhibitory GABAergic neurotransmission, mediated by interneurons, is thought to be well-conserved (reviewed in ). Recently however, more evidence is emerging that also GABAergic function is compromised. Limon et al. showed that functional GABAA receptors are lost from the brains of AD patients . Furthermore, during the course of normal aging, hippocampal GABAergic interneurons lose contact boutons , while this process is accelerated in hAPP mice (J20 line). An AD mouse model expressing human amyloid precursor protein (hAPP) with Swedish and Indiana mutations . This suggests that the excitation/inhibition balance during aging in hAPP (J20) mice is more severely disrupted. Interestingly, aged hAPP (J20) mice show no loss of GABAergic interneurons. Another AD mouse model, a triple transgenic mouse (TauPS2APP) does show a loss of GABAergic interneurons in the hippocampus of aged mutant mice compared to WT mice . Although this study did not examine persistent forms of LTP in the CA3-CA1 circuit, they found early phase-LTP enhancements in the dentate gyrus of tauPS2APP mice. However, because this mouse model carries three transgenes, it was not clear whether all or one mutation contributed to the loss of GABAergic interneurons. Our results suggest that expression of mutant tau protein alone likely promotes the loss of hippocampal interneurons. Furthermore, several reports provide evidence that the development of AD leads to hyperexcitability, shown by the increased incidence of epileptic activity in sporadic AD, which is particularly high in early-onset autosomal-dominant AD [20, 21, 56, 57]. One possible explanation for this may be that GABAergic interneuronal function, which is critical for maintaining excitatory/inhibitory balance in the brain, is also impacted in tauopathies like AD. AD mouse models provide experimental support for this model of AD-related hyperexcitability. For example, the hAPP (J20) mice, was shown to have spontaneous epileptic activity, indicating network hypersynchrony [46, 55]. Interestingly, this network hypersynchrony in the hAPPJ20 model resulted from PV cell dysfunction. Furthermore, in an earlier study the same research group showed that reduction of tau in the hAPP (J20) mouse model prevented behavioral deficits and excitotoxicity . Finally, the apoE4 knock-in (KI) mouse model has impaired neurogenesis in the dentate gyrus that was due to impaired presynaptic GABAergic input and resulted in loss of GABAergic interneurons [15, 48]. These mice also showed spatial learning and memory deficits, which could be rescued with treatment with the GABAAR potentiator pentobarbital . Interestingly, the GABAergic impairment was dependent on tau because apoE4 KI mice in a tau knockout background did not exhibit this phenotype . These results support the idea that tau plays an important role in the survival and function of GABAergic interneurons. To note, these findings highlight the importance of tau in AD-related phenotypes but do not address whether tau lesions on their own manifest effects in AD-related synaptic plasticity and hyperexcitability. Our present findings provide evidence supporting a model directly implicating tau function in the maintenance of balanced neuronal signaling networks.
Two other tau mouse models, which express either an anti- or a pro-aggregant tau protein have been studied to examine the effects of tau aggregation on synaptic plasticity and behavior. While these models are not represented by naturally occurring human mutations, they do provide insight into the role of tau in synaptic plasticity. Pro-aggregant tau resulted in impaired L-LTP in the CA3-CA1 hippocampal pathway, while an anti-aggregant tau model showed enhanced CA1-LTP [19, 58]. However, only the pro-aggregant mice showed memory deficits, while the anti-aggregant displayed normal memory, as assessed by Morris water maze and the passive avoidance task. The Tg mice in our study exhibit severe memory deficits and enhanced L-LTP and express a mutant tau isoform with properties that more closely align with pro-aggregant tau. While differences in the effects on LTP between this study and our own may be explained by the different models used, another intriguing possibility is that greater levels of tau aggregation may affect other neuronal signaling pathways involved in the expression of L-LTP that are not impacted by the less severe pathology present in our model. This finding also supports the idea that tau conformation can have different effects on synaptic function and memory, and tau mutations that promote pathological aggregation may impair hippocampal function. This possibility may be useful to consider for developing therapeutic strategies that target tau aggregation.
In summary, this study provides valuable new evidence for a role of tau independent of Aβ in AD-related synaptic deficits. Our data support a model in which tau helps to maintain proper network excitability through the regulation of GABAergic function. ‘Normal’ cognition requires intact neuromolecular pathways for the regulation of synaptic plasticity. Altered synaptic function and memory deficits are major hallmarks of dementia. Results from this study suggest a promising therapeutic avenue for the treatment of AD and FTD may be developing new drug regimens based on removing existing or inhibiting the development of tau pathology. However, effective application of this strategy requires detailed knowledge about the effects of tau pathology on cognition in preclinical model systems. By identifying GABAergic function as a crucial pathway influenced by the expression of pathological tau, we may in the short-term be able to take advantage of several FDA approved drugs regulating GABAergic function for use in AD treatment. In the long-term, these studies may provide new insight into the neuronal signaling pathways and molecular targets that are regulated by tau to develop more effective approaches for tauopathy treatment.