Aging is a natural process related to the gradual loss of physiological, behavioral, and social functions. Therefore, it has an essential impact on the nervous system, and it is considered a risk factor for many neurodegenerative and psychiatric illnesses (Hou et al., 2019). Thus, understanding the neurobiology underlying age-related impairment is essential given the growing elderly population. The structural and functional brain changes observed during aging in the central nervous system are the consequence of cellular and molecular alterations, which are in turn regulated by genetic, epigenetic, lifestyle, and environmental factors (Khan et al., 2017). Interestingly, the brain is not homogeneously affected by aging; specific regions are specially altered. Such is the case of the hippocampus, which is our region of interest in the present study, and appears to be among the most affected.

During aging, the hippocampus shows a decrease in volume, which correlates with a decline in learning and memory (Driscoll et al., 2006). Aging is also associated to the presence of neuroinflammation in this region, along with an up-regulation of pro-inflammatory genes, resulting in oxidative stress (Gavilán et al., 2007; Barrientos et al., 2015). Most studies have found no evidence for the death of pyramidal neurons during aging, both in humans and mice (West et al., 1994). In the case of inhibitory neurons there is still controversy, while some studies have found loss of interneuronal markers in the hippocampus of aged animals (Potier et al., 2006; Stanley et al., 2012), others did not find it (Miettinen et al., 1993). At the cellular level, there are structural alterations in hippocampal pyramidal neurons, such as a reduction in dendritic branching (Markham et al., 2005), and reductions in the density of dendritic spines and synapses (Dickstein et al., 2013). However, little is known about the impact of aging on hippocampal inhibitory circuits. Some studies have described decreased GAD65, GAD67 and specifically, somatostatin mRNA levels in the hippocampus during aging (Vela et al., 2003; Gavilán et al., 2007), but nothing is known on the effects of aging on the structure of hippocampal inhibitory neurons.

The study of interneurons in general, and particularly in the hippocampus, represents a challenge due to the existence of several subpopulations, presenting diverse morphological, neurochemical, physiological, and synaptic characteristics (Booker and Vida, 2018). In the hippocampus, interneurons represent 10–15% of the total neuronal cell population (Pelkey et al., 2017). Particularly, the oriens-lacunosum moleculare (O-LM) cells, which are the subpopulation studied in the present work, represent the 4.5% of the total number of CA1 interneurons (Bezaire and Soltesz, 2013). They are characterized by the expression of the neuropeptide somatostatin (SST) (Freund and Buzsáki, 1996; Oliva et al., 2000), and are widely distributed in the different regions of the hippocampus, including CA1 (Köhler and Chan-Palay, 1982; Freund and Buzsáki, 1996). O-LM cells have their soma and main dendritic arbor in the stratum oriens, where they receive inputs from pyramidal neurons of the stratum pyramidale (Lacaille and Schwartzkroin, 1988; Blasco-Ibáñez and Freund, 1995; Katona et al., 1999). The anatomy of these cells is described in detail in Freund and Buzsáki (1996). Most of these synaptic contacts are established on dendritic spines, a peculiar characteristic of SST-expressing interneurons (Freund and Buzsáki, 1996; Guirado et al., 2014). The O-LM cells reciprocally synapse onto pyramidal and non-pyramidal neurons in the stratum lacunosum moleculare, through a dense axonal projection field decorated with abundant en passant boutons (EPB) (Sik et al., 1995; Müller and Remy, 2014). This anatomical arrangement of the O-LM cells allows them to function in a prototypical feedback inhibitory circuit (Leão et al., 2012), in which they mediate theta oscillations (Katona et al., 2016).

Because of their roles as postsynaptic and presynaptic elements, dendritic spines and synaptic boutons respectively are proper proxies for neuronal input and output; therefore, increases in their density have been correlated to increases in neuronal activity (Engert and Bonhoeffer, 1999; Becker et al., 2008). Our laboratory has previously demonstrated the structural remodeling of the dendritic arbor and dendritic spines of O-LM cells by chronic stress (Gilabert-Juan et al., 2017) and by the depletion of the polysialylated form of the neural cell adhesion molecule (PSA-NCAM), a plasticity-related molecule (Guirado et al., 2014). We also showed that the manipulation of NMDA receptors affected both the density of dendritic spines and EPB (Pérez-Rando et al., 2017a,b).

Apart from these molecular and structural changes, functional impairments are also observed during aging, such as alterations in long-term potentiation (LTP) and depression (LTD) (Lister and Barnes, 2009). Some of the most extensively studied molecules involved in these physiological mechanisms, which underlie basic cognitive processes, are the N-methyl-d-aspartate receptors (NMDARs). These are ionotropic glutamate receptors, which are highly expressed in the neocortex and the hippocampus, both in pyramidal and inhibitory neurons, including the O-LM cells (Alvarez et al., 2007; Oren et al., 2009; Pérez-Rando et al., 2017b). The NMDARs are assembled as tetramers composed of 4 subunits: two obligatory GluN1 subunits, along with 2 GluN2 or GluN3 subunits. There are 6 subtypes of non-obligatory subunits: 4 GluN2 (GluN2A–GluN2D) and 2 GluN3 (GluN3A and GluN3B) (Hansen et al., 2017). Alterations in NMDAR complex expression and physiology have been described during aging, especially in the hippocampus (Clayton et al., 2002; Boric et al., 2008; Foster et al., 2017). The GluN2B subunit is more affected by aging than the other GluN2 subunits, showing more significant decreases in mRNA and protein expression in aged animals (Magnusson et al., 2002). The effects of aging on hippocampal GluN1 expression have rendered conflicting results, with significant declines described in some studies, but not in others (Magnusson et al., 2005; Das and Magnusson, 2011). However, the majority of these studies have been performed using western blot or RT-PCR, which does not allow the discrimination of the expression in different hippocampal regions, layers or neuronal populations. SST-expressing interneurons are a very interesting subject to study during aging in this regard, because we know that they express NMDA receptors (Pérez-Rando et al., 2017b) in young adult animals and we have also evidence that these receptors regulate structure and connectivity of the SST-expressing interneurons (Pérez-Rando et al., 2017a,b).

Sex is a very important factor to take into account when studying the nervous system and its pathologies; the study of the female brain in preclinical research is of paramount importance, but studies using both sexes have started to appear published only recently (Shansky and Murphy, 2021). Specifically in the hippocampus, there are sex differences in the dendritic spine density of pyramidal neurons (Shors et al., 2001). These differences are not restricted to excitatory neurons; a recent study from our laboratory has shown changes in the density of dendritic spines and EPB in the O-LM cells during the estrous cycle (Pérez-Rando et al., 2021). The expression of hippocampal NMDARs also appears to be affected by sex and levels of estrogens: Male rats express higher levels of NMDARs in the hippocampus than females, but only when females are in the estrus phase (Brandt et al., 2020). All these studies have been performed in adult-young animals. However, to our knowledge there are still no studies exploring the differential effects of sex on interneuronal morphology or NMDARs expression during aging.

The present study aimed to understand the impact of age on the structure of O-LM cells in the CA1 hippocampal region of male and female mice. We have also studied the expression of different subunits of NMDARs (GluN1 and GluN2B) in this interneuronal subpopulation.

In the present study, we demonstrate age-dependent alterations in the structure of SST-expressing interneurons in the CA1 stratum oriens. There was a significant decrease in the density of the dendritic spines between 3− and 9-month-old females. However, we did not observe such changes in males. We also studied the density of the en passant boutons (EPB) of these interneurons in the stratum lacunosum moleculare, where no changes were found related to aging. We have focused our study in the stratum oriens of the CA1 region. It should be noted that although there is a small proportion of the GFP + cells in this stratum that project to the striatum, most of them are O-LM cells, which project to stratum lacunosum moleculare (Oliva et al., 2000; Pérez-Rando et al., 2017b). The O-LM cells are an interneuronal subpopulation characterized by the presence of abundant spines in their dendritic tree (Freund and Buzsáki, 1996) and numerous EPB in their axonal projection field in the stratum lacunosum moleculare (Müller and Remy, 2014). The spines and EPB of O-LM cells have been studied during aging previously, using in vivo two-photon imaging in GIN mice, but only from 4 to 11 months of age. The density of EPB in the stratum lacunosum moleculare remained constant, similar to what we have observed in fixed tissue. However, in contrast with our results, the spine density of O-LM interneurons increased by 12% from 4 to 11 months (Schmid et al., 2016). However, the big differences in methodology between the 2 experiments have to be taken into account. Ablation of neocortex covering the hippocampus, type of imaging and the age of the animals compared. It is important to note that both sexes were analyzed together in this experiment, without control over the estrous cycle, which can influence spine density in O-LM interneurons (Pérez-Rando et al., 2021). Moreover, this study examined dendritic spines in behaving animals and, consequently, spine density can be influenced by this behavior, possibly through the activation NMDA receptors. To our knowledge the present study is the first to analyze the structure of SST-expressing neurons in intact animals and to include aged animals from both sexes. Age-related structural changes have been also studied in other subpopulations of interneurons, particularly analyzing their dendritic arborization. In the visual cortex, the dendritic arbors of interneurons (in general) were simplified in aged mice (Eavri et al., 2018). However, in other specific interneuronal types, such as the vasoactive intestinal peptide/calretinin-expressing cells in the CA1 region of the hippocampus, no differences were found in dendritic length or number of intersections between young and old mice (Francavilla et al., 2020). However, it has to be noted that we have observed changes in the dendritic spine density only in 9-month females and when grouping both sexes. These results suggest that the structural changes are not the result of aging but rather of the influence of female sex hormones. The estrous cycle of female mice usually starts between the 6th – 8th week (White, 2007), but can be conditioned for many factors (Drickamer, 1974, 1984; Kruczek and Gruca, 1990), and prolongs itself to 14 months-old, when the animals are no longer reproductive (Flurkey et al., 2007). Furthermore, the variations we find in females, but not in males, when studying O-LM dendritic spine density are in accordance with previous findings from our laboratory studying these interneurons in intact and ovariectomized female GIN mice. We demonstrated that this parameter changes depending on the estrous phase in which the mouse is in, and the alterations seen in ovariectomized mice were also restored by 17β-Estradiol administration (Pérez-Rando et al., 2021). These structural changes had only been previously studied in pyramidal neurons (Woolley and McEwen, 1994; Luine and Frankfurt, 2013).

Interestingly, some of the effects of estrogens on hippocampal structure/connectivity appear to be mediated by NMDAR (El-Bakri et al., 2004). However, a more recent study revealed no differences in the mRNA levels of GluN1 and GluN2B in the dorsal hippocampus when comparing control and ovariectomized mice (McCarthny et al., 2018). Particularly, the study of NMDARs and hormonal levels during senescence demonstrated that NMDAR subunit mRNA levels in the hippocampus were much more prominently affected by the chronological age than by the reproductive status of the animals (Adams et al., 2001). In the present work we did not control the estrous cycle of the females, which could probably affect some parameters and constitute one of the limitations of the study.

Although we have not studied the density of SST-positive neurons, different studies have previously demonstrated a loss during aging in the hippocampus of rats, particularly in the stratum oriens (Potier et al., 2006; Stanley et al., 2012), although some reports did not find decreases in this cell type (Miettinen et al., 1993). Studies in mice are scarcer: one study described detrimental effects of aging on the number of these cells in the dentate gyrus (Koh et al., 2014), while another, using GIN mice, found an age-dependent decrease of O-LM interneurons from 5 to 9 months, but not from 9 to 12 months (Schmid et al., 2016). It is possible that these reductions of cell density may have an impact on the influence of SST-expressing interneurons during aging, in addition to the structural alterations that we have observed.

Interestingly, different studies have shown that aging is associated with a decrease in the excitability of hippocampal pyramidal neurons and with alterations in synaptic plasticity (LTP and LTD), which may underlie the deficits in hippocampal-dependent learning tasks (Oh and Disterhoft, 2020). These changes may be compensatory for a decrease in inhibition, because in the CA1 aging is associated with decreased GABAergic transmission (Billard et al., 1995; Potier et al., 2006), decreased number of GAD-67 + interneurons (Shetty and Turner, 1998; Stanley and Shetty, 2004), and decreased expression of GAD-67 mRNA in the CA1 (Vela et al., 2003). Specifically, the selective loss of O-LM cells in an age-dependent manner has been related to a decrease in the amplitude of evoked inhibitory postsynaptic currents (IPSCs) and a decrease in the frequency of spontaneous IPSCs in CA1 pyramidal neurons (Potier et al., 2006).

Previous studies in our laboratory have demonstrated that the structure of hippocampal SST-expressing interneurons in the stratum oriens of CA1 is regulated by NMDAR. Interestingly, we found that these glutamatergic receptors were expressed in the somata and dendritic spines of SST-expressing interneurons (Pérez-Rando et al., 2017a). In our previous studies using organotypic cultures, we observed that the relative density of spines and their appearance rate increased 24 h after the NMDA infusion (Pérez-Rando et al., 2017a). By contrast, the application of an NMDAR antagonist produced opposite results (Pérez-Rando et al., 2017b).

Mounting evidence suggests that an age-associated hypofunction of the NMDAR can contribute to the impairment in spatial learning and memory, particularly affecting the Schaffer collateral pathway (Foster and Norris, 1997; Kumar and Foster, 2013). A decrease in the expression of NMDARs has been observed in the hippocampus during normal aging (Magnusson et al., 2006; Billard and Rouaud, 2007; Zhao et al., 2009), particularly in the CA1 region (Magnusson and Cotman, 1993; Wenk and Barnes, 2000). However, the response to aging was different depending on the NMDAR subunit studied (Avila et al., 2017). There are no significant changes in GluN1 protein expression in the whole hippocampus of aged mice (Zhao et al., 2009) and particularly in the expression of GluN1 mRNA in the CA1 pyramidal layer (Magnusson et al., 2005). In the present report we have studied the density of GluN1 + puncta in the O-LM cells, but we did not observe changes during aging, which is in accordance with the previous data. However, decreases in the expression of GluN1 protein (Eckles-Smith et al., 2000; Mesches et al., 2004; Liu et al., 2008) and mRNA (Adams et al., 2001) in the whole hippocampus have been reported in aged rats. By contrast with GluN1, there is a widespread agreement that the GluN2B subunit is especially affected by aging. There is a decrease of its protein (Clayton and Browning, 2001; Mesches et al., 2004; Zhao et al., 2009) and mRNA expression (Adams et al., 2001; Clayton and Browning, 2001) in the whole hippocampus, and particularly in the CA1 pyramidal layer (Magnusson, 2001), both in rats and mice. It has to be noted, however, that these studies analyzed the whole hippocampus and not specific cell types as our experiment. Notwithstanding, in concordance with these previous studies, we observed a decrease of the GluN2B + puncta in the O-LM cells in our oldest group, which is significant in females and when analyzing both sexes together. However, the area covered by GluN2B + puncta in O-LM cells is considerably larger in the oldest animals, suggesting the presence of clusters of these receptors. In fact, it has been suggested that an increased association of GluN2B receptors with scaffolding proteins in aged animals may contribute to the age-related memory impairment (Zamzow et al., 2013). Additionally, there is data supporting that the age-associated hypofunction of NMDAR may be due to decrease in the activity-dependent changes in the surface distribution of NMDAR (Clayton et al., 2002) increased association with membrane scaffolding proteins (Zamzow et al., 2013) and to decreased trafficking of GluN2B to the synapse (Kumar et al., 2019). The clustering of GluN2B subunits may be the physical manifestation of poor trafficking of NMDAR, which possibly begins in middle-age.

It is interesting to note that the genetic deletion of GluN2B in interneurons prevents the formation of glutamatergic synapses on hippocampal interneurons (Kelsch et al., 2014). Consequently, it is possible that the age-related decrease in GluN2B expression in O-LM cells also interferes with the maintenance of their glutamatergic input, affecting thus the density of dendritic spines on these cells. Our results show that alterations in NMDAR do not only affect pyramidal neurons, but also SST-expressing cells and, consequently, the inhibition that these interneurons exert on excitatory cells. In fact, SST has a potent effect on LTP in CA1 (Rostampour et al., 2002; Fan and Fu, 2014).

An important limitation of our study is the age of the oldest animals (16 months). These mice are not considered aged in most studies, although cognitive impairment starts to appear around this age in rodents (Kumar and Foster, 2013). However, in our hands, most animals of this transgenic strain died shortly after this age and many of them presented tumors and considerable hair loss. Unfortunately, we were not able to perform learning tests in these animals and there are no reports analyzing cognitive tasks in aged animals of this strain or from this genetic background (FVB).

Altogether, our results help to shed light on how aging and sex modulate the structural plasticity and the NMDARs expression of hippocampal interneurons, particularly of O-LM cells. The study of these receptors is important since changes on their expression can lead to neuronal potentiation or depression, and alterations in their physiology may underlie behavioral and cognitive dysfunctions.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Committee on Bioethics of the Universitat de València.

JN designed the experiment. YG, PK, SC, and MP-R performed the experiments and quantifications. YG, PK, and JN wrote the manuscript and all the authors revised and edited the final version. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.