Early changes in the brain during Alzheimer’s disease (AD) arise in part due to the pathophysiological effects of oligomeric Aβ (1–42) (Mucke et al., 2000; Fukumoto et al., 2010; Edwards, 2019). A characteristic feature of oligomeric Aβ (1–42) is the impairment of hippocampal long-term potentiation (LTP), whereby acute effects occur (Wang et al., 2004; Klyubin et al., 2005; Kalweit et al., 2015). Furthermore, deficits in both LTP and learning days after intracerebral treatment with Aβ (1–42) have been reported (Zhang et al., 2017; Khodadadi et al., 2018). Examination of the effects of oligomeric Aβ have indicated that topical application of Aβ to the slice chamber causes a suppression of GABA_A receptor function in the hippocampus of young (P25-P40) rats in vitro (Orr et al., 2014). It has been proposed that this can lead to elevated levels of extrasynaptic glutamate that, in turn, enable enhanced activation of GluN2B-containing N-methyl-D-aspartate receptors (NMDAR), which then mediate hyperexcitability (Lei et al., 2016). Others have reported that antagonism of GluN2B-containing NMDAR prevents Aβ (1–42)-mediated deficits in LTP in the hippocampal CA1 region of young adult rats (Hu et al., 2009).

LTP in the CA1 region is predominantly NMDAR-dependent and postsynaptically mediated (Malenka et al., 1988, but see also Falcón-Moya et al., 2020 and Grover and Teyler, 1990 for examples of exceptions). NMDAR are typically composed of two GluN1 subunits and two GluN2 subunits (Dingledine et al., 1999). GluN2A and GluN2B-containing NMDAR play a key role in the enablement of hippocampal LTP (Bartlett et al., 2007; Berberich et al., 2007; Ballesteros et al., 2016) Although GluN2C and GluN2D subunits also occur in NMDAR, these do not appear to play a critical role in LTP (Banerjee et al., 2009). Whereas co-agonist binding of glycine or D-serine occurs at the GluN1 subunit (Hirai et al., 1996; Mothet et al., 2001; Henneberger et al., 2010), glutamate binds to the GluN2 subunit (McBain and Mayer, 1994; Laube et al., 1997). GluN2A-containing NMDAR exhibit faster kinetics compared to GluN2B-containing NMDAR (Punnakkal et al., 2012), lose their Mg²⁺ block at lower membrane potentials compared to GluN2B-containing NMDAR (Clarke and Johnson, 2006; Clarke et al., 2013), but allow half as much charge transfer, deactivate faster, and enable less Ca²⁺-influx per unit of current than GluN2B-containing NMDAR (Vicini et al., 1998; Erreger et al., 2005; Sobczyk et al., 2005; Clarke et al., 2013). Furthermore, GluN2A-containing NMDAR respond to weaker stimuli (Köhr et al., 2003; Berberich et al., 2005, 2007) and enable weaker and less persistent forms of LTP compared to GluN2B-containing NMDAR (Ballesteros et al., 2016).

Excessive activation of NMDAR leads to excitotoxity (Rothman and Olney, 1987) and NMDAR antagonists have proven effective in the treatment of cognitive deficits in early AD (Paoletti et al., 2013; Zhou and Sheng, 2013). It is widely believed that the excitotoxic effects of NMDAR in AD are mediated by excessive extracellular glutamate that leads to overactivation of GluN2B-containing NMDAR (Texidó et al., 2011; Danysz and Parsons, 2012; Paoletti et al., 2013; Talantova et al., 2013; Zhou and Sheng, 2013). In addition, NMDAR have been reported to mediate specific cellular and biochemical actions of Aβ in processes that involve both GluN2A and GluN2B subunits (Roselli et al., 2005; Snyder et al., 2005; Domingues et al., 2007; Abbott et al., 2008; Deshpande et al., 2009; Li et al., 2009) in a process that may involve Aβ–binding to NMDAR (Cowburn et al., 1997; De Felice et al., 2007; Lacor et al., 2007).

The contribution of different GluN subunits to NMDAR toxicity, or Aβ–mediated pathophysiology, may change along the lifespan of an individual. Developmental changes in the expression of GluN2A and GluN2B have been reported, whereby a systematic increase of GluN2A subunits and a decline of GluN2B subunits occurs in the period encompassing early postnatal stages (12 days postnatally) through early adulthood (35 days postnatally) (Carmignoto and Vicini, 1992). More recent findings suggest that GluN2A and GluN2B levels remain abundant and largely equivalent in later adulthood (2–4 months postnatally), at least in C57BL/6 mice, although relative differences in murine strains occur (Beckmann et al., 2020). Furthermore, differences in GluN2A:GluN2B ratios occur along the dorsoventral axis of the hippocampus (Dubovyk and Manahan-Vaughan, 2018). GluN2A:GluN2B ratios are also modulated by synaptic activity, whereby lower levels lead to an increase in GluN2B and a decrease in GluN2A subunits (Chen and Bear, 2007; Yashiro and Philpot, 2008). The consequence is a prolongation of NMDAR currents and a reduction in LTP thresholds (Chen and Bear, 2007; Yashiro and Philpot, 2008). Thus, reductions in synaptic activity triggered by Aβ (Balleza-Tapia et al., 2010) may lead to a preferential recruitment of GluN2B-containing NMDAR into synaptic plasticity processes.

In the present study, we explored to what extent GluN2A and GluN2B-containing NMDAR contribute to changes in hippocampal excitability and LTP triggered by intracerebral inoculation with oligomeric Aβ (1–42) in aging mice. We treated 8–15 month old GluN2A^+/− and GluN2B^+/− animals, and their wt littermates, with oligomeric Aβ (1–42), or control peptide, 4–5 weeks before assessing neuronal excitability and LTP in the hippocampal slice preparation. Effects of Aβ (1–42) on neuronal excitability were minimal. LTP was reduced in GluN2A^+/− and GluN2B^+/− mice compared to their wt littermates. In wt hippocampi, intracerebral pretreatment with Aβ (1–42) potently reduced the magnitude of LTP. Strikingly, however, pretreatment with Aβ (1–42) had no impact on the profile of LTP expressed in the hippocampi of GluN2A^+/− and GluN2B^+/− mice. These findings suggest that in old age, the detrimental effects of Aβ (1–42) on LTP are not mediated by GluN2A and GluN2B containing NMDAR.

In this study, we report that in 8–15 month old animals, neuronal properties were largely equivalent in the hippocampi of GluN2A^+/− and GluN2B^+/− transgenic mice compared to their wt littermates (Table 4). A limited range of changes in properties of CA1 pyramidal cells were detected 4–5 weeks following intracerebral oligomeric Aβ (1–42)-treatment of wildtype animals, comprising, for example, a higher input resistance and a more negative sag, in wt littermates of GluN2A^+/− mice (Table 4). Aβ-treatment elicited a limited amount of changes in neuronal properties of the transgenic animals, whereby GluN2A^+/− mice were more affected than GluN2B^+/− mice (Table 4). LTP was impaired in both GluN2A^+/− and GluN2B^+/− hippocampi compared to their wt littermates. Furthermore, intracerebral treatment with oligomeric Aβ (1–42) resulted in an impairment of LTP in wt mice. Strikingly, the profile of LTP was unchanged in Aβ-treated GluN2A^+/− or GluN2B^+/− hippocampi compared to control peptide-treated transgenic hippocampi, meaning that the already deficient LTP (compared to wt) was not impaired further by Aβ-treatment. Taken together, these data indicate that in the aging brain, GluN2A-containing NMDAR played an important role in the homeostasis of neuronal excitability. Furthermore, neuronal function was only mildly affected by Aβ-treatment of aging wt or GluN2 deficient mice, and knockdown of GluN2A or GluN2B did not worsen the debilitating effects of oligomeric Aβ (1–42) on hippocampal LTP. This suggests that in the aging hippocampus, NMDAR were not instrumental in propagating the pathophysiological effects of oligomeric Aβ (1–42) on hippocampal function.

By and large, we detected no marked effects of Aβ-treatment on neuronal properties. We saw for example, a greater positivity of the resting membrane potential in GluN2A^+/− hippocampi compared to control GluN2A^+/− hippocampi, but no significant difference in the membrane potential in GluN2B^+/− compared to control GluN2B^+/− hippocampi. The former difference derived more from differences in control peptide effects in wt and transgenics than from direct effects of Aβ, however. Sag (I_h) ratio was increased in GluN2B^+/− following Aβ-treatment but unaffected in GluN2A^+/− hippocampi following treatment. The I_h stabilises the resting membrane potential, regulates the afterhyperpolarization and influences firing frequency (McCormick and Pape, 1990). Neuronal oscillations are supported by the I_h (McCormick and Pape, 1990; Wahl-Schott and Biel, 2009) and we have reported in the past that hippocampal neuronal oscillations are undermined by oligomeric Aβ (1–42) (Kalweit et al., 2015). Sag is enabled by HCN channels (Robinson and Siegelbaum, 2003). HCN channels modulate glutamate release in the hippocampus and thus, influence NMDAR currents (Neitz et al., 2014). It has been proposed that these channels support hippocampal plasticity processes (Honnuraiah and Narayanan, 2013). NMDAR-dependent spontaneous slow excitatory dendritic potentials are regulated by HCN channels and are mediated by GluN2B-containing NMDARS (Ashhad and Narayanan, 2016). Others have reported a more positive resting membrane potential and enhanced I_h current in the hippocampus after intracerebral Aβ-treatment (Eslamizade et al., 2015). Our finding that sag ratio was more positive after Aβ-treatment of GluN2B^+/− mice, suggesting that GluN2B contributed to these effects.

A role for GluN2B has been described in Aβ-mediated effects in the hippocampus: The reduction in network activity and LTP that occurs following topical application of Aβ to hippocampal slices or neuronal cultures, or intracerebral treatment of rats is prevented by antagonists of GluN2B (Hu et al., 2009; Rönicke et al., 2011). It has also been reported that the enhancement of NMDAR currents and intracellular calcium levels that occur following application of Aβ are mediated by GluN2B-containing NMDAR (Li et al., 2011; Ferreira et al., 2012). Furthermore, Aβ (1–40) triggers an increase in the expression of GluN2B in hippocampal neuronal cultures (Chang et al., 2016) and the GluN2A:GluN2B ratio decreases after Aβ (1–42) application (Huang et al., 2017). Interestingly, inhibition of GluN2B-containing NMDAR prevents Aβ-mediated impairments of LTP (Rönicke et al., 2011; Huang et al., 2017). The frequency-dependency of LTP was not assessed in these studies, but our findings suggest that less potent forms of LTP that do not critically require activation of GluN2B are not affected by Aβ (1–42). This is all the more interesting given the advanced age of the mice in our study: all of the studies mentioned above used young adult animals and treatment regimes of maximally 15 days before hippocampal scrutiny.

We previously reported that the frequency of the afferent input, and impulse number it delivers, determines the recruitment of GluN2A or GluN2B subunit-containing NMDAR to LTP in the CA1 region (Ballesteros et al., 2016). Weak afferent stimulation recruits a GluN2A-dependent form of LTP that is small in magnitude and short (>4 h) in duration. By contrast, strong afferent stimulation recruited LTP that required GluN2B-containing NMDAR that was much larger in magnitude and lasted over 24 h (Ballesteros et al., 2016). The form of LTP examined in the present study had both a GluN2A and a GluN2B-dependent component, as indicated by the reduction in LTP magnitude in GluN2A and GluN2B transgenic hippocampi compared to wt littermates. Effects were more potent in GluN2A transgenics, although the increased variability in responses in the later phase of LTP in GluN2B transgenics may have masked deficits in LTP compared to controls. Although we cannot entirely rule out the possibility that LTP was successfully induced, albeit less potently, in the GluN2A^+/− or the GluN2B^+/− mice by non-ionotropic mechanisms, this seems unlikely. Theta-burst stimulation induces a decremental form of LTP in the mouse hippocampal slice preparation that is distinct from more robust forms of LTP induced by high frequency afferent stimulation (Novkovic et al., 2015). The recruitment of, for example, voltage-dependent calcium channels into hippocampal LTP requires very fast high frequency stimulation (Grover and Teyler, 1990; Manahan-Vaughan et al., 1998) and forms of synaptic potentiation that can be induced by activation of metabotropic glutamate (mGlu), or catecholaminergic, receptors are temporally slow to become manifest (Manahan-Vaughan and Reymann, 1995; Tse et al., 2023) and do not fit the temporal dynamics of the LTP profiles induced in our study. We did not see a complete abolishment of LTP in the GluN2A^+/−, or the GluN2B^+/− hippocampi presumably because the remaining subunits permitted a weaker form of LTP to occur. Evidence for this has been offered by pharmacological studies that showed that LTP, short-term potentiation and forms of synaptic depression can be induced with the same afferent stimulation frequency combined with a graded degree of activation of NMDAR (Cummings et al., 1996).

Forms of LTP that are intrinsically linked to learning are enabled by weak afferent activity in the hippocampus (Kemp and Manahan-Vaughan, 2004, 2008; Hagena and Manahan-Vaughan, 2012; Hoang et al., 2021). By contrast, very strong afferent stimulation induces robust LTP that is associated with reduced learning flexibility, reduced reversal learning and an absence of differentiated neuronal encoding in the hippocampus of rats (Barnes, 1979; Barnes et al., 1994; Hoang et al., 2021). This raises the question as to the functional requirement of GluN2B-dependent LTP in adulthood. It has been reported that the expression of GluN2-subunits declines after early postnatal development (Carmignoto and Vicini, 1992), but in adult C57Bl/6 mice, we did not observe an appreciable decline in receptor expression (Beckmann et al., 2020). This would suggest that both GluN2A and GluN2B-containing NMDAR contribute to LTP in adulthood. The kind of LTP (magnitude, persistency) may be determined by the kind of information that is encoded, however.

We previously reported that intracerebral treatment with oligomeric Aβ (1–42) 1 week prior to assessing LTP in 4–10 month old wildtype mice impairs the early phase of LTP (Südkamp et al., 2021). Animals in the present study were 8–15 months old at the time of treatment. Here, wildtypes showed a significant LTP impairment that extended to the later phases of plasticity. One possibility is that the increased age of the wildtypes may have caused a greater vulnerability to the debilitating effects of intracerebral Aβ-treatment. Age-dependent changes in hippocampal function have been reported (Barnes et al., 1994; Wilson et al., 2004; Twarkowski et al., 2016) that could underlie these effects. Another reason for the greater vulnerability of wildtype hippocampi in this study might be the enhanced time-period of exposure to Aβ-treatment. One cannot exclude, however, that although the background strain was identical for the wt mice in this and the abovementioned study (C57BL/6)(Sakimura et al., 1995; von Engelhardt et al., 2008; Dvoriantchikova et al., 2012), substrain-dependent differences influenced the outcome of Aβ-sensitivity. Genetic drift related to separation of breeding pools is likely to have an impact on the precise genomic identity of wildtype littermates derived from the C57BL/6 strain (Manahan-Vaughan, 2018), that could have influenced the sensitivity of the wildtypes to Aβ. For this reason, we included separate wt cohorts for both transgenic lines, whereby only wildtype littermates of either the GluN2A^+/− or the GluN2B^+/− were used.

The relatively mild effects of oligomeric Aβ (1–42) on neuronal responses, as detected by patch clamp in our study, corresponds to reports that the peptide predominantly affects synaptic transmission and thus, dendritic responses (Shankar et al., 2008). It has been reported that application of oligomeric Aβ (1–42) onto hippocampal slices from young (P26-32) rats results in an acute increase in surface expression of the GluA1 subunit of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPAR) (Whitcomb et al., 2015). By contrast, topical application of oligomeric Aβ to hippocampal cultures from P18 rats inhibits AMPAR trafficking (Rui et al., 2010) and reduces surface expression of AMPAR (Guntupalli et al., 2017). Reductions in GluA1 density in the hippocampus have also been reported 30 days after intracerebral inoculation of 18 month old mice with Aβ (1–42) in a dose of 20 μM (Yeung et al., 2020). Scrutiny of the stimulus–response relationship of wildtype 8–15 month old mice that had been treated intracerebrally 4–5 weeks previously with 10 μM oligomeric Aβ (1–42) did not reveal any significant effects. The absence of deficits in the stimulus–response relationship that would indicate that AMPAR density had declined in Aβ (1–42)-treated wildtypes, may reflect differences in the oligomer doses used these two studies, or may indicate that although AMPAR density might have declined in the Aβ (1–42)-treated mice, this did not have a functional impact on synaptic transmission.

The maintenance of later phases of LTP beginning at around 90 min post-induction, are supported by phospholipase -C coupled (group 1) receptors (Hagena and Manahan-Vaughan, 2022; Mukherjee and Manahan-Vaughan, 2023). Group 1 mGlu receptors can functionally interact with NMDAR and alter NMDAR currents (Rosenbrock et al., 2010). Correspondingly pharmacological antagonists of group 1 mGlu receptors can alter the induction profile of hippocampal LTP (Neyman and Manahan-Vaughan, 2008). Aberrant mGlu5 receptor signaling is triggered by interactions of the receptor with Aβ (1–42) (Haas and Strittmatter, 2016) and mGlu5 receptors contribute to impairments of hippocampal LTP in 8–11 week old rats that are caused by acute cerebral treatment with Aβ (1–42) (Wang et al., 2004; Hu et al., 2014). Others have reported that Aβ (1–42) can form a complex with GluN2B and mGlu1 receptors (Taniguchi et al., 2022) and that mGlu5 receptors can bind with prion protein that serves as a target for Aβ (1–42) (Hu et al., 2014). We did not see any exacerbation of LTP deficits in GluN2A^+/− or the GluN2B^+/− mice that were treated with Aβ (1–42). This raises the interesting question as to whether the knockdown of NMDAR subunits left fewer interaction partners for the putative creation of an Aβ-prion protein-mGlu5 complex (Hu et al., 2014), or an Aβ-GluN2B-mGlu1 complex (Taniguchi et al., 2022), that would otherwise serve to further disrupt LTP in the transgenics.

In conclusion, this study shows that knockdown of GluN2A or GluN2B did not elicit substantial changes in neuronal properties within the hippocampus of aging (8–15 month old) mice. Nonetheless, loss of GluN2A appeared to result in a greater degree of change in neuronal properties suggesting that this subunit is more relevant than GluN2B for neuronal homeostasis in the aging hippocampus. Intracerebral treatment with oligomeric Aβ (1–42) 4–5 weeks before testing resulted in some changes in neuronal properties, but these were mostly apparent in GluN2A transgenics and a comparison of Aβ-mediated effects in the GluN2 transgenics and their wt littermates showed that responses were equivalent, suggesting that the knockdown of the subunits only slightly increased the vulnerability of the hippocampus to oligomeric Aβ (1–42). A similar profile was apparent with regard to hippocampal LTP: knockdown of GluN2A or GluN2B significantly impaired LTP in wt littermates. Treatment with oligomeric Aβ (1–42) resulted in an impaired LTP in wt littermates that was equivalent in magnitude to LTP in GluN2 transgenics. This impaired LTP was not debilitated further by Aβ (1–42)-treatment. The findings of this study show that although the aging hippocampus was affected by the intracerebral presence of oligomeric Aβ (1–42), and although knockdown of GluN2A or GluN2B impaired LTP, changes in the composition of the NMDAR did not contribute appreciably to the effects on neuronal properties or LTP caused by oligomeric Aβ (1–42). In sum, a loss of GluN2 subunit content in the hippocampus did not increase the vulnerability of this structure to the debilitating effects of oligomeric Aβ (1–42).

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

The animal study was approved by Landesamt für Arbeitsschutz, Naturschutz, Umweltschutz und Verbraucherschutz, NRW, Germany. The study was conducted in accordance with the local legislation and institutional requirements.

NS: Formal analysis, Investigation, Methodology, Writing – review & editing, Data curation. OS: Formal analysis, Investigation, Methodology, Writing – review & editing, Data curation. DM-V: Conceptualization, Formal analysis, Funding acquisition, Methodology, Resources, Supervision, Writing – original draft, Writing – review & editing.