Our experiments were performed on hippocampi prepared from adult male and female mice of young (2–4 months; i.e., young adults) and middle-aged (13–16 months; i.e. old adults) mice, which were either wild-type (wt, n = 34 young and n = 29 middle-aged) with C57Bl/6 J background or littermate transgenic mice expressing a dominant-negative mutant of ActRIB (dnActRIB, n = 37 young and n = 29 middle-aged) (Müller et al., 2006). Mice were housed under standard conditions. All experiments were conducted in accordance with the Animal Protection Law of Germany and the European Communities Council Directive of November (1986 /86/609/EEC) and were approved by local government of Lower Franconia.

Transverse or horizontal slices (350 μm thick) from dorsal hippocampus were prepared from mice anesthetized with isoflurane, as described previously (Müller et al., 2006; Dahlmanns et al., 2023a,b). Briefly, brain slices were cut in ice-cold sucrose-based artificial cerebrospinal fluid (aCSF) containing (in mM) 75 sucrose, 87 NaCl, 3 KCl, 0.5 CaCl₂, 7 MgCl₂, 1.25 NaH₂PO₄, 25 NaHCO₃, and 10 d-glucose. Slices were incubated in the same solution for 10 min at 35°C and then maintained in aCSF containing (in mM) 125 NaCl, 3 KCl, 1 CaCl₂, 3 MgCl₂, 1.25 NaH₂PO₄, 25 NaHCO₃, and 10 d-glucose at room temperature for at least 2 h before being used. Individual slices were transferred to a submerged chamber that was mounted on the stage of an upright microscope, and perfused with normal aCSF with 1.5 mM MgCl₂ and 2.5 mM CaCl₂ at 31°C. All solutions were constantly gassed with 95% O₂ / 5% CO₂. Recorded signals were filtered at 2 kHz and sampled at 20 kHz using a Multiclamp 700B amplifier together with Digidata 1440A interface and pClamp10 software (Molecular Devices, Sunnyvale, CA, United States).

Field potentials in CA1 stratum radiatum were recorded with a glass pipette filled with modified aCSF, in which NaHCO₃ was replaced by HEPES (5 mM) to avoid pH change. A concentric platinum bipolar electrode was inserted into the stratum radium to stimulate Schaffer collaterals (SC), with constant current pulses (0.1 ms width). The input–output relationship of the SC-CA1 synapse was routinely determined by an incremental increase in stimulus strength from 50 to 200 μA. Short-term potentiation (STP) and long-term potentiation (LTP) of the SC-CA1 synapse were examined using stimulus intensities that elicited 30% of the maximum response before high frequency stimulation. With individually adjusted stimuli delivered at 0.1 Hz before and after tetanus, two forms of SC-CA1 LTP were studied. “Classic” NMDA receptor-dependent LTP was induced by applying theta-burst stimuli (TBS: 15 bursts of four pulses at 100 Hz, delivered at an inter-burst interval of 200 ms), twice 10 s apart. Voltage-gated Ca²⁺ channel-dependent LTP was measured in the presence of the NMDA receptor blocker APV (50 μM) and induced by strong high-frequency stimulation at 200 Hz (for 1 s), repeated four times at 5 s intervals (Grover and Teyler, 1990). Data points were obtained by averaging six responses every minute.

Mice were sacrificed under anesthesia with isoflurane and the brain was quickly dissected out. The isolated hippocampus was homogenized in lysis buffer containing 0.32 M sucrose, 5 mM Tris–HCl (pH 8.0), and a protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, United States). Homogenates were centrifuged at 13,000 × g at 4°C for 15 min (twice). Supernatant was collected for assaying levels of activin A and ActRIB according to the manufacturer’s instructions, with ELISA kits from R&D Systems (Activin A Quantikine ELISA Kit; Minneapolis, MN, United States) and ELAab Science Co. Ltd. (Activin Receptor Type IB ELISA Kit; Wuhan, China), respectively.

Brains from young and old adult wt and dnActRIB mice were extracted as described above, and the hippocampi were isolated and stored at −80°C. RNA was then isolated according to the manufacturer’s protocol (RNeasy Plus Universal Mini Kit; QIAGEN, Hilden, Germany), reverse-transcribed into cDNA (High-Capacity cDNA Reverse Transcription Kit; Applied Biosystems, Waltham, MA, United States), and then stored at −20°C. RNA levels were determined using quantitative real-time PCRs (RT-qPCR) according to the manufacturer’s instructions (ABsolute QPCR Mix SYBR Green no ROX, Thermo Fisher Scientific, Waltham, MA, United States) in a realple × 4 cycler (Eppendorf, Hamburg, Germany). The following primers were used (Eurofins Genomics, Ebersberg, Germany): (1) dnActRIB (NM_007395.3, exon 6–7 and exon 7–8 deleted in the dnActRIB mutant, 130 bp), 5′-tctaccataaccgccagagg-3′ (forward) and 5′-aggtcctcctcggaaatcag-3′ (reverse), and (2) TATA binding protein (TBP; 146 bp), 5′-gccaagagtgaagaacaatcc-3′ (forward) and 5′-ccttccagccttatagggaac-3′ (reverse). Every biological sample was measured as a technical duplicate and then averaged. Then, ΔCq values were calculated by relating dnActRIB to TBP, and 2^−ΔΔCq values were calculated by normalizing values to the mRNA abundance of ActRIB in the hippocampus of dnActRIB mice.

Electrophysiological data analysis was performed off-line with Clampfit 10.2 (Molecular Devices, Sunnyvale, CA, United States). Peak amplitude and initial slope (20–80% of the peak response) of field postsynaptic potentials (fPSPs) were quantified. STP during quadruple-pulse stimulation was determined by normalization peak amplitude of 2nd–4th response to that of first response. For LTP measurements, averaged fPSP slopes (from six responses every minute) were normalized to 100% before tetanus and pooled across mice of the same group.

Data are expressed as means ± SEM. OriginPro 2018G (OriginLab Corporation, Northampton, MA, United States) was used for statistics and graphs. Shapiro–Wilk test was used to assess the normality of data distribution, and the null hypothesis was accepted when p value was larger than 0.05. Statistical comparisons were performed using unpaired or paired Student’s t-test, one-way ANOVA followed by Tukey’s post-hoc test, as appropriate. Significance was assumed for p < 0.05.

We reported previously that in the hippocampus from young mice expressing a dominant-negative mutant of ActRIB (dnActRIB), several features of basal excitatory neurotransmission and synaptic plasticity were altered when compared to wt mice (Müller et al., 2006; Dahlmanns et al., 2023a,b). To examine how age alone and in combination with activin receptor signaling would affect the performance of glutamatergic synapses, we made extracellular recordings in the CA1 region of brain slices from middle-aged wt and dnActRIB mice. Electrodes placed in stratum radiatum served to measure field postsynaptic potentials (fPSPs) in response to electrical stimulation of Schaffer collaterals (SC; Figure 1A, inset). As a basic property, we determined first the input–output (I-O) relationship of the SC-CA1 synapse, which depicts peak fPSP amplitude as function of stimulus intensity (50–200 μA, Figure 1A). Whereas the I-O relationship of the SC-CA1 synapse differed significantly between wt and mutant mice of young age, as reported before (Dahlmanns et al., 2023a,b), the I-O curves in hippocampi from older wt and dnActRIB mice displayed a nearly perfect overlap, with indistinguishable responses across stimulating intensities (e.g., at 200 μA: wt-old, 1.81 ± 0.11 mV, n = 29; dnActRIB-old, 1.96 ± 0.09 mV, n = 26; p = 0.404, unpaired t-test; Figure 1A). We next asked whether the characteristic facilitation of synaptic responses upon repetitive stimulation would be impaired in the CA1 region of older dnActRIB mice. We gauged frequency facilitation by means of a train of four identical stimuli delivered at 20 Hz (Figure 1B). For comparison, fPSP amplitudes were normalized to that of the first response. Again, no difference was observed between hippocampi from the two groups of older mice (ratio for the 4th fPSP: wt-old, 2.04 ± 0.08, n = 29; dnActRIB-old, 2.13 ± 0.07, n = 26; p = 0.345, unpaired t-test; Figure 1B).

Using weak to moderate theta-burst stimulation (TBS) protocols, we found previously that activin signaling augments NMDA receptor-dependent LTP at the SC-CA1 synapse in young adult mice (Müller et al., 2006; Dahlmanns et al., 2023a,b). Here, we repeated this experiment in slices from older wt mice. Employing a comparably moderate TBS paradigm (2 × TBS, separated by 10 s), we observed an increase in fPSP slope to 157.73 ± 7.22% of control value when analyzed 36–40 min after TBS (n = 12; Figures 1C,D). Notably, the amount of potentiation was not significantly different from that previously determined in young wt hippocampus (Müller et al., 2006). However, in contrast to the plastic deficit in the young mutant, time course and extent of TBS-induced LTP were virtually identical in hippocampi from both genotypes when measured in the older cohorts (dnActRIB-old, 166.78 ± 6.02%, n = 14; p = 0.342 vs. wt-old, unpaired t-test; Figure 1D).

In addition to “classic” NMDA receptor-mediated LTP, the SC-CA1 synapse is known to exhibit an NMDA receptor-independent form of LTP that requires much stronger stimulation (200 Hz trains for 1 s, repeated four times every 5 s), and relies on activation of voltage-gated Ca²⁺ channels (VDCC; Grover and Teyler, 1990). Notably, the latter form of LTP (VDCC-LTP) was found to increase in aged rats (Shankar et al., 1998; Boric et al., 2008), as does the L-type Ca²⁺ current in CA1 pyramidal cells of old rats (Thibault and Landfield, 1996). Whether the increase in this type of LTP can be regarded as a compensatory mechanism for the aging-associated decline in NMDA receptor-dependent LTP to preserve cognitive function in old rodents remains unclear (Burke and Barnes, 2010; Foster, 2012). To examine whether activin signaling is also involved in VDCC-LTP, we subjected hippocampi from younger and older mice of both genotypes to the strong induction protocol described above (Figure 2). This set of experiments, which we performed in the presence of the NMDA receptor antagonist APV (50 μM), yielded the following results: (i) The magnitude of VDCC-LTP in hippocampi from young dnActRIB mice was significantly smaller compared to that from young wt mice (36–40 min post-tetanus: wt-young: 158.98 ± 7.00%, n = 6; dnActRIB-young: 129.31 ± 7.71%, n = 6; p = 0.017, unpaired t-test; Figures 2B,D); (ii) Reminiscent of our finding for NMDA receptor-dependent LTP in older mice (Figure 1D), the remarkable discrepancy in the amount of VDCC-LTP between wt and mutant preparations at young age vanished when re-examined in the older cohorts, which exhibited virtually identical amounts of VDCC-LTP (36–40 min post-tetanus: wt-old: 150.66 ± 6.44%, n = 8; dnActRIB-old: 147.39 ± 7.58%, n = 7; p = 0.741, ANOVA with Tukey’s post-hoc test) (Figures 2A,C,D); (iii) There was no difference in VDCC-LTP between young and old control groups (wt-young vs. wt-old, p = 0.404, unpaired t-test). To verify the Ca²⁺ channel-dependence of this form of LTP, we superfused slices from old adult wt mice with the L-type blocker nifedipine (10 μM). The drug produced a uniform increase in synaptic transmission, which reached steady-state within 16–20 min (from 0.49 ± 0.05 to 0.69 ± 0.06 mV, n = 6, p = 0.009, paired t-test). Reassuringly, nifedipine suppressed the subsequent induction of VDCC-LTP (110.94 ± 2.73% of control value, n = 5; p = 0.029 vs. wt-old, ANOVA with Tukey’s post-hoc test; Figure 2D), confirming the essential involvement of L-type Ca²⁺ channels.

To substantiate the physiological significance of our findings, we had to rule out that it is simply a decline in transgene expression with age that explains the apparent failure of the dominant-negative receptor approach to replicate the LTP-phenotype of the young mice in the old ones. As an antibody specifically recognizing dominant-negative ActRIB (dn-ActRIB) is not available, we performed RT-qPCR to demonstrate the sustained mRNA abundance of mutant receptors in aged hippocampi (2^-ΔΔCt for dnActRIB transgene/β-Actin: dnActRIB-old, 3.19 ± 0.12, n = 6), which was not different from that in mutant young adults (dnActRIB-young, 3.58 ± 0.13, n = 4; p = 0.063, unpaired t-test; Figure 3A). As expected, dnActRIB transcripts were absent in wt hippocampi (wt-old, n = 6; wt-young, n = 5; Figure 3A).

In the human cerebrospinal fluid, the level of activin A increases with age (Ebert et al., 2006; Baird et al., 2012). To determine whether such an age-dependent elevation also holds for mouse hippocampus, we measured the level of activin A in young and aged hippocampi by means of ELISA. When the hippocampi from both hemispheres were prepared for ELISA measurements, we noted a small, but, significant difference in weight between young and old hippocampi (Figure 3B). In quantitative terms, hippocampi from aged wt mice had less weight (wt-old, 28.59 ± 0.66 mg, n = 18) compared to those from young wt mice (wt-young, 31.14 ± 0.53 mg, n = 24; p = 0.004, unpaired t-test) (Figure 3B). Such decrease in weight was also true for dnActRIB mice: older hippocampi were lighter (28.26 ± 0.53 mg, n = 16) than young ones (31.27 ± 0.44 mg, n = 26, p = 0.0001, unpaired t-test; Figure 3B). ELISA measurements revealed significantly elevated levels of activin A in aged hippocampi of mice with either genotype (wt-old, 1.39 ± 0.09 pg/mg, n = 10; wt-young, 1.02 ± 0.03 pg/mg, n = 15, p = 0.003, unpaired t-test; dnActRIB-old, 1.35 ± 0.08 pg/mg, n = 10; dnActRIB-young, 1.08 ± 0.05 pg/mg, n = 12, p = 0.011, unpaired t-test; Figure 3C). Importantly, the rise in activin A level with age was accompanied by a parallel increase in the amounts of ActRIB in aged hippocampi from mice of both genotypes (Figure 3D). The comparable levels of functional (endogenous) ActRIB between genotypes in each age group, young hippocampi (wt-young, 78.94 ± 2.34 pg/mg, n = 8; dnActRIB-young, 79.34 ± 2.03 pg/mg, n = 5, p = 0.168, unpaired t-test) and old hippocampi (wt-old, 115.20 ± 6.06 pg/mg, n = 7; dnActRIB-old, 110.87 ± 8.50 pg/mg, n = 6, p = 0.750, unpaired t-test), argue against the antibody recognizing the dominant negative activin receptors.

It is tempting to speculate that the age-associated increase in endogenous activin A together with the rise in its main signal-transducing receptor is sufficient to override the inhibitory effect of the dominant-negative receptors on LTP that we observed in young hippocampus where the level of activin A is lower. Thus, we predicted that application of recombinant activin A should restore the extent of LTP in hippocampi from young dnActRIB mice, by enhancing the probability for formation of heterotetramers consisting of functional full-length (endogenous) receptors. As shown in Figure 4A, pre-incubation of slices from young dnActRIB mice with recombinant activin A was indeed capable of rescuing the LTP phenotype. Consistent with findings from our earlier multi-electrode array-based study (Müller et al., 2006), moderate TBS (2 × TBS) of the SC-CA1 synapse in young hippocampi induced significantly weaker LTP in dnActRIB slices compared to wt slices (wt-young, 153.54 ± 5.43%, n = 10; dnActRIB-young, 123.16 ± 3.58%, n = 9; p = 0.018, ANOVA with Tukey’s post-hoc test; Figure 4). However, when we employed the same TBS protocol in dnActRIB slices after 3–8 h of incubation with recombinant activin A (50 ng/mL; R&D Systems), the LTP deficit in the mutant hippocampus was fully rescued (Figure 4). The slope of fPSPs was 161.87 ± 7.14% of control value (36–40 min after TBS, dnActRIB-young + activin A, n = 8), which was significantly different from that in mutant controls (p = 0.002 vs. dnActRIB-young), but not that in wt young controls (p = 0.451 vs. wt-young; one-way ANOVA followed by Tukey’s post-hoc test).