Developmental onset of enduring long-term potentiation in mouse hippocampus

Abstract

Analysis of long-term potentiation (LTP) provides a powerful window into cellular mechanisms of learning and memory. Prior work shows late LTP (L-LTP), lasting >3 hr, occurs abruptly at postnatal day 12 (P12) in the stratum radiatum of rat hippocampal area CA1. The goal here was to determine the developmental profile of synaptic plasticity leading to L-LTP in the mouse hippocampus. Two mouse strains and two mutations known to affect synaptic plasticity were chosen: C57BL/6J and Fmr1^−/y on the C57BL/6J background, and 129SVE and Hevin^−/− (Sparcl1^−/−) on the 129SVE background. Like rats, hippocampal slices from all of the mice showed test pulse-induced depression early during development that was gradually resolved with maturation by 5 weeks. All the mouse strains showed a gradual progression between P10-P35 in the expression of short-term potentiation (STP), lasting ≤1 hr. In the 129SVE mice, L-LTP onset (>25% of slices) occurred by 3 weeks, reliable L-LTP (>50% slices) was achieved by 4 weeks, and Hevin^−/− advanced this profile by 1 week. In the C57BL/6J mice, L-LTP onset occurred significantly later, over 3–4 weeks, and reliability was not achieved until 5 weeks. Although some of the Fmr1^−/y mice showed L-LTP before 3 weeks, reliable L-LTP also was not achieved until 5 weeks. L-LTP onset was not advanced in any of the mouse genotypes by multiple bouts of theta-burst stimulation at 90 or 180 min intervals. These findings show important species differences in the onset of STP and L-LTP, which occur at the same age in rats but are sequentially acquired in mice.

1 |. INTRODUCTION

The hippocampus is critical for spatial navigation and processing of new information. It is the main brain region used to study long-term potentiation (LTP), a cellular mechanism of learning and memory. Knowing the maturational profile of synaptic plasticity provides a basis for investigating abnormalities leading to intellectual disabilities and other neurodevelopmental disorders. LTP has been most rigorously studied in stratum radiatum of hippocampal area CA1; hence, the excitatory synapses in this subfield are the focus of this and many prior studies. In Sprague Dawley rats, LTP begins to consolidate in hippocampal area CA1 around postnatal day 21 (P21) (Kramar & Lynch, 2003). Our previous work in Long–Evans rats revealed test pulse depression that lasts until P21 (Cao & Harris, 2012), replicating earlier findings (Abrahamsson, Gustafsson, & Hanse, 2007, 2008). Theta-burst stimulation (TBS) reversed the test pulse depression at P8–P11, but no potentiation was produced above the initial naïve response. At P12, the TBS reliably induced enduring LTP lasting more than 3 hr (late LTP [L-LTP]). When multiple episodes of TBS were delivered, the onset age of L-LTP was advanced to P10 (Cao & Harris, 2012). Here, our goal was to extend these studies to mouse hippocampus.

Mice are a widely used model system to test the effects of genetic manipulations on normal behavior and physiology (Ellenbroek & Youn, 2016; Homberg, Wohr, & Alenina, 2017). Little is known about the developmental profile of synaptic plasticity in mice. Two commonly used wild-type mouse strains, C57BL/6J and 129SVE, were chosen together with Fmr1^−/y and Hevin^−/− (Sparcl1^−/−), which are known for producing aberrations in synaptic plasticity and development. Fmr1^−/y on the C57BL/6J background is a common model of fragile X syndrome (FXS) for mental retardation and autism (He & Portera-Cailliau, 2013; Pfeiffer & Huber, 2009). The cause of FXS is the mutation preventing the synthesis of FMRP, an RNA-binding protein selectively expressed in neurons and responsible for mRNA transport and local protein synthesis in dendrites (Wang et al., 2016). Hevin is a protein released by astrocytes and interneurons that is critical for synapse formation and rearrangement (Mongredien et al., 2019). The synaptogenic activity of Hevin promotes glutamatergic synapse maturation and refines cortical connectivity and plasticity (Risher et al., 2014; Singh et al., 2016).

We tested for the impact of these two strains and two key mutations on the developmental profile of synaptic plasticity in hippocampal area CA1. The outcomes provide a foundation for investigating genetic effects on synaptic plasticity and may help to explain the inter-species variance in synaptogenesis and developmental capacity for learning and memory.

2 |. MATERIALS AND METHODS

2.1 |. Ethical approval

Procedures were approved by the University of Texas at Austin Institutional Animal Care and Use Committee and complied with all NIH requirements for the humane care and use of laboratory mice (protocol # AUP-2012–00127, AUP-2012–00056, and their successor protocols).

2.2 |. Animals

Breeding pairs of C57BL/6J (RRID:IMSR_JAX:000664) and Fmr1^−/y (RRID:MGI:5703659) on this background were kindly donated by Dr D. Brager (Center for Learning and Memory, University of Texas at Austin) who received the founder pair from Dr K. Huber (University of Texas Southwestern). Breeding pairs of 129SVES6 (129SVE) mice were obtained from a supplier (Taconic, Rensselaer, NY; RRID: IMSR_TAC:129sve), and Hevin^−/− (Sparcl1^−/−, allelic composition Sparcl1^tm1Pmc/Sparcl1^tm1Pmc, RRID:MGI:4454665) were on this background. We will be referring to this knock-out as Hevin^−/−. The generation of Fmr1^−/y and Hevin^−/− (Sparcl1^−/−) has been described before (Barker et al., 2005; McKinnon, McLaughlin, Kapsetaki, & Margolskee, 2000; The Dutch-Belgian Fragile X Consortium, 1994).

Animals were co-housed and provided with food and water ad libitum on a 12 hr light–dark cycle. The experimental design was originally optimized for the Fmr1^−/y mice, in which the males have the strongest phenotypes; hence, for the appropriate comparison with Fmr1^−/y data, males were used for all the experiments. The exact age of each animal was known. Since the developmental profiles were more gradual in the mice than in rats, and for ease of graphical presentation, data from the mice were grouped by age as P10–13 (<2 weeks), P13–17 (2 weeks), P18–23 (3 weeks), P26–31 (4 weeks), and P32–37 (5 weeks).

2.3 |. Slice preparation

Hippocampal slices were prepared from mouse pups at P8–P38 as previously described (Bourne, Kirov, Sorra, & Harris, 2007). Animals were decapitated under isoflurane anesthesia when appropriate (age older than P33). The brain was removed, and the left hippocampus was dissected out and rinsed with room temperature artificial cerebrospinal fluid (aCSF) containing (in mM) 117 NaCl, 5.3 KCl, 26 NaHCO₃, 1 NaH₂PO₄, 2.5 CaCl₂, 1.3 MgSO₄, and 10 glucose, pH 7.4, and bubbled with 95% O₂/5% CO₂. Four slices (400 μm thick) from the middle third of the hippocampus were cut at 70° transverse to the long axis on a tissue chopper (Stoelting, Wood Dale, IL) and transferred to four individual interface chambers in the Synchroslice system (Lohmann Research Equipment, Castrop-Rauxel, Germany). The slices were placed on a net at the liquid–gas interface between aCSF and humidified 95% O₂/5% CO₂ atmosphere held at 32–33°C. The entire dissection and slice preparation took 5–7 min. The slices recovered in the chambers for 3 hr before the recordings commenced.

2.4 |. Electrophysiology

The stimulation and data acquisition were obtained using the SynchroBrain software (Lohmann Research Equipment). A concentric bipolar stimulating electrode (FHC Inc., Bowdoin, ME) was positioned near the CA3 side, and a metal recording electrode (Thomas Recording, Geissen, Germany) was placed ∼400 μm away from the stimulating electrode, also in the middle of CA1 stratum radiatum. Stimuli consisted of 200 μs biphasic current and each stimulus was applied every 5 min. Response measurements included the amplitude (mV) of the fiber volley (FV) and slope (mV/ms) of the field excitatory postsynaptic potential (fEPSP). The maximum positive amplitude of the FV was measured where it was clearly separated from the stimulus artifact and onset of the negative slope of the fEPSP. Each fEPSP slope was estimated by linear regression in the middle of the initial phase of the fEPSP over an interval ranging from 0.2 to 0.4 ms, depending on response magnitude. The stimulus intensity and response analysis time frames were kept constant for each slice throughout the duration of each experiment. Stimulus intensities were set to obtain about 50% of the maximum response levels typical for each age group of animals. Stimulus intensities were set to obtain about 50% of the maximum response levels typical for each age group of animals. This setting was found to be within 40–60% of the maximum response obtained by delivering increasing stimulus intensities (ranging from 100 to 500 μA). This partial I/O was done at the end of each experiment to avoid response plasticity at higher stimulus intensities before the experiment.

The various TBS paradigms are described in the Results and figure legends. Briefly, the 8T TBS paradigm consisted of eight trains with 30 s intervals with each train containing 10 bursts at 5 Hz and each burst containing four pulses at 100 Hz. The 1T TBS paradigm consisted of one train of the same stimulation pattern. The fEPSP slope is expressed as a percentage of the naïve fEPSP or the averaged baseline response obtained 30 min before delivering the TBS paradigm as indicated in the Results and figure legends. Baseline responses were recorded for 60 min before the delivery of the TBS paradigm. Experiments within an age/genotype were grouped depending on the success of inducing potentiation with a threshold set at 120% of the naïve fEPSP slope. fEPSP slope changes normalized to the first naïve response or the baseline were averaged across the slices for each genotype.

2.5 |. Statistics

Statistical analyses were performed using Prism (GraphPad Software Inc., San Diego, CA). Data are presented throughout as the mean ± SEM. The minimal level of significance was set at p < .05. The outliers were detected and removed using Grubbs’ test (no more than one data point per dataset). The total number of animals and slices of each cohort (genotype and age) used in the 8T LTP experiments presented in Figures 1–8 are presented in Table 1. In other experiments, the number of slices is indicated in parenthesis in Figures 4, 5c, 8g, 9. The specific tests and the outcomes are indicated in the figure legends.

FIGURE 1.

Test pulse-induced depression was resolved by 4–5 weeks for all strains and genotypes. (Top row) Electrode positions in hippocampal area CA1 and slice paradigm to test for late long-term potentiation (L-LTP) and experimental design. Experimental design: Slices were recovered for 3 hr without stimulation (tan frame). Then the stimulus intensity was set to obtain the approximate half-maximal response (40–60%). The stimulus was repeated at this intensity at 5 min intervals for 60 min to obtain the baseline responses. The 8T consisted of eight trains at 30 s intervals with 10 bursts at 5 Hz of four pulses each at 100 Hz. Then, the responses were monitored for 180 min and L-LTP was determined by averaging the response slope over the last 155–180 min. All field excitatory postsynaptic potential (fEPSP) slopes are normalized to the first (naïve) response at time 0. (a–d) Baseline responses to the test pulse stimulation for the first 1 hr of the experiments show an age-dependent decrease in test pulse-induced depression for all four genotypes (two-way analysis of variance [ANOVA], interaction: F (9,208) = 0.427 (p = .92), age: F (3,208) = 24.03 (p < .0001), genotype: F (3,208) = 1.70 (p = .17)). All genotypes revealed significant differences between ages before or at 2 weeks and 4–5 weeks old (Dunnet’s post hoc tests). Data from 4 and 5 week old mice did not differ and are plotted together for Hevin^−/−

FIGURE 8.

A second episode of 8T separated in time augments the initial late long-term potentiation (L-LTP). (a) Experimental design: Baseline, first 8T (black arrowhead), delivery of second 8T (red arrowhead) 180 min after the first. (b–e) Changes in field excitatory postsynaptic potential (fEPSP) slopes were normalized relative to the naïve baseline response and averaged across the slices before and after the onset ages of L-LTP (red dotted line at 120% baseline) and plotted versus time for each genotype. (f) Within each slice, the initial L-LTP was averaged over 155–180 min after the first 8T (orange frame). Next, the L-LTP was considered to be augmented if the fEPSP slope after the second 8T (averaged over 225–260 min, black frame) was greater than the initial L-LTP (orange frame) by at least 10% (red dotted line, C57BL/6J: t = 6.335, df = 5, partial η² = 0.889, **p = .0014; Fmr1^−/y: t = 4.646, df = 5, partial η² = 0.812; *p = .0056; 129SVE: t = 6.221, df = 5, partial η² = 0.886, **p = .0016; Hevin^−/−: t = 5.184, df = 5, partial η² = 0.843, **p = .0035). (For Hevin^−/−, the data from the after onset age group at 3 weeks were separated from 4 weeks because the augmentation of L-LTP at 3 weeks did not reach criterion but did reach criterion at 4 weeks.) (g) Slices from 4 week old animals that had no initial L-LTP also showed no augmentation, in response after the second 8T (C57BL/6: t = 12.11, df = 5, partial η² = 0.967, ***p < .0001; Hevin^−/−: t = 7.369, df = 5, partial η² = 0.916, ***p = .0007)

TABLE 1.

Total number of slices and animals in each strain and age group for Figures 1–8. For Figures 9 and 10, the number of slices in each condition is indicated in parenthesis in the figures themselves

	<2 weeks	2 weeks	3 weeks	4 weeks	5 weeks
Animals
C57BL/6J	8	7	10	8	6
Fmr1−/y	6	4	6	4	7
129SVE	6	6	6	6
Hevin−/−	5	5	6	7	4
Slices
C57BL/6J	14	11	18	11	16
Fmr1−/y	9	10	10	10	17
129SVE	10	11	10	10
Hevin−/−	11	12	13	14	10

FIGURE 4.

Demonstration that 8T is a robust induction paradigm for late long-term potentiation (L-LTP) for all strains and genotypes. Reducing the number of trains in the theta-burst stimulation (TBS) paradigm from 8T to 1–4T resulted in the same magnitude and endurance of L-LTP in (a) C57BL/6, (b) Fmr1^−/y, (c) 129SVE, and (d) Hevin^−/−. Dotted red line is at 120% naïve. The slices were obtained from animals (a–c) at 4 weeks or (d) at 5 weeks, all were after the onset age of L-LTP for 8T

FIGURE 5.

Differences among strains in the probability of late long-term potentiation (L-LTP). The probability of short-term potentiation (STP) (a) and L-LTP (b) by strain and genotype across postnatal age. The probabilities were calculated as the ratio of the number of successful STP or L-LTP experiments relative to the total number of experiments for each condition and age group. No significant differences were detected for STP. Significant differences were detected for the probability of L-LTP (b) at postnatal Week 3 (χ² = 13.16, df = 3; **p = .0043) and postnatal Week 4 (χ² = 16, df = 3; **p = .0012). 129SVE strain had a significantly higher probability of L-LTP than C57BL/6J at 4 weeks (χ² = 12, df = 3; ***p = .0006). Within backgrounds, a significant difference was detected at 3 weeks between C57BL/6J and Fmr1^−/y (χ² = 4.875, df = 1; *p = .0272) and between 129SVE and Hevin^−/− (χ² = 5.490, df = 1; *p = .0191). Both genotype pairs were equal at their corresponding after onset ages. (c) The magnitude of field excitatory postsynaptic potential (fEPSP) potentiation at different time periods post theta-burst stimulation (TBS) did not differ significantly across strains or genotypes in the after onset age groups for each strain or genotype

FIGURE 9.

Second episode of 8T did not produce late long-term potentiation (L-LTP) in slices lacking initial L-LTP in 4-week-old C57BL/6J mice. (a) Experimental design: Each slice was subjected to two identical theta-burst stimulation (TBS) paradigms consisting of either one or eight trains that were spaced by 90 min (1T light blue, 8T pink arrows) or 180 min (1T navy, 8T red arrows). Initial L-LTP was calculated by averaging responses at 60–85 min after the first 8T (gray time frame) or 155–180 min for the 180 min 8T interval (orange time frame). The effect of the second 8T was calculated at 135–160 min for the 90 min interval (green time frame) or at 225–260 min for the 180 min interval (black time frame). (b) Summary of the mean changes in field excitatory postsynaptic potential (fEPSP) slope normalized to the 30 min averaged baseline responses with 8T episodes spaced 90 min (pink) or 180 min (red). (c) Quantification of the experiments at the representative time frames for the experiments in (b). No significant differences were detected at any of the time points for the different separations in 8T. (d) Summary of the mean changes in fEPSP slope normalized to the 30 min of the averaged baseline responses with 1T theta-burst stimulation (TBS) episodes spaced 90 min (blue) or 180 min (navy). (e) Quantification of the experiments at the representative time frames for the experiments in (d). No significant differences were detected between the levels of potentiation at orange and black intervals (pre and post second 1T spaced 180 min after the first 1T, one-way analysis of variance [ANOVA], F(3,9) = 2.736, p = .267). A significant difference was detected by two-way RM ANOVA between 1T-90 m-1T and 1T-180 m-8T by TBS spacing factor (F(1,7) = 6.762; *p = .0354)

3 |. RESULTS

3.1 |. Developmental test pulse depression

In immature rat hippocampus (<P20), test pulse stimulation produces a marked depression of the fEPSPs (Abrahamsson et al., 2007, 2008; Cao & Harris, 2012; Xiao, Wasling, Hanse, & Gustafsson, 2004). Both tetanic stimulation and TBS can reverse this test pulse-induced depression, but they produce no potentiation above the first naïve response at these young ages. To calculate the magnitude of test pulse depression and discern between the reversal of depression and LTP, all responses were normalized relative to the first naïve response. Here, test pulses were given at the lowest frequency of 1 pulse per 5 min that was tested in rats (Figure 1). For mice aged ≤2 weeks, the test pulse depression was significant at 55 min (Table 2). By 4–5 weeks, the depression was gone, and such age dependence was similar between all four mouse strains and genotypes. Thus, like in rats, test pulse depression in mice was age dependent.

TABLE 2.

The magnitude of test pulse depression across ages and genotypes. The responses were normalized to the first naïve slope and the difference between the normalized average slopes at Time 0 and 55 min is shown with means and 95% confidence interval (upper and lower limits)

	C57BL/6J			Fmr1 −/y			129SVE			Hevin −/−
	Mean	Upper limit	Lower limit	Mean	Upper limit	Lower limit	Mean	Upper limit	Lower limit	Mean	Upper limit	Lower limit
<2 weeks	−0.19	−0.15	−0.23	−0.17	−0.12	−0.23	−0.16	−0.09	−0.24	−0.17	−0.10	−0.24
2 weeks	−0.16	−0.08	−0.23	−0.18	−0.12	−0.24	−0.17	−0.12	−0.21	−0.14	−0.05	−0.23
3 weeks	−0.08	−0.01	−0.15	−0.07	0.02	−0.15	−0.09	−0.06	−0.11	−0.02	0.06	−0.10
4 weeks	−0.06	0.00	−0.13	0.02	0.12	−0.08	−0.03	0.08	−0.14	0.09	0.19	−0.01
5 weeks	−0.04	0.02	−0.11	0.00	0.05	−0.06				−0.02	0.09	−0.13
4–5 weeks	−0.05	−0.01	−0.10	0.00	0.05	−0.04				0.05	0.12	−0.03

3.2 |. Strain- and genotype-specific differences in developmental onset of STP and L-LTP

Eight trains of TBS (8T) reliably produced L-LTP at P12 in rats (Cao & Harris, 2012) and young adult C57BL/6J mice (Cao & Harris, 2014). Hence, we used this 8T protocol to determine the onset age of L-LTP in acute slices from mouse hippocampal area CA1 (Figure 1, top right). The LTP threshold was set at 120% of the naïve response. For each age, the number of experiments was tabulated where 8T failed (none) or succeeded in producing short-term potentiation lasting 1 hr (STP) or L-LTP lasting at least 3 hr (LTP). Age groupings are shown on weekly basis (Figures 1–5), and also as the relative frequency of slices showing L-LTP (Figures 6–9, Supplementary Figures S1 and S2). The before onset groups comprised ages where L-LTP was induced in 25% or less of tested slices. The onset groups had 25–50% success rate among tested slices. The after onset groups had L-LTP in more than 50% of tested slices. The percentage was calculated by dividing the number of slices that showed only STP (gray sectors) or L-LTP (red sectors) by the total number of slices tested in the corresponding age group as shown in pie charts in Figures 2 and 3 (and Supplementary Figures S1 and S2).

FIGURE 6.

Magnitude of starting naïve responses did not predict success of late long-term potentiation (L-LTP) across ages, strains, or genotypes. Slices in each age group lacking L-LTP (black dots, gray bars) or producing 3 hr L-LTP (red dots and bars). Age groups are indicated relative to when L-LTP could first be produced as before onset, onset, and after onset for each genotype. No significant differences were detected across the genotypes or developmental stages of L-LTP onset for the before onset age group, or for the onset versus after onset age groups (two-way analysis of variance [ANOVA], interaction: F(7,127) = 0.691 (p = .680), age: F (1,127) = 0.0722 (p = .789), genotype: F(1,127) = 1.57 (p = .151)). Individual slice values are plotted as dots. Only three slices showed L-LTP in the before onset group, so these were not included in the statistical analyses

FIGURE 2.

Week-by-week analysis of short-term potentiation (STP) and late long-term potentiation (L-LTP) in the C57BL/6J (a1–a5) and Fmr1^−/y (b1–b5) mice. In time course plots and pie charts, the experiments with no potentiation (none) are colored black, those with STP lasting less than 1 hr (STP 1 hr) are colored gray, and those with LTP lasting 3 hr (LTP 3 hr) are colored red. Representative waveforms for pre-theta-burst stimulation (TBS) baseline responses are colored gray, for 3 hr post-TBS are colored black for no potentiation and red for L-LTP at 3 hr. The pie charts show the relative fractions with the actual number of slices in each fraction for each age. The numbers of all animals and slices are also listed in Table 1

FIGURE 3.

Week-by-week analysis of 1 and 3 hr LTP in the 129SVE (a1–4) and Hevin^−/− (b1–4) mice. The same color and labeling schemes as in Figure S2

In mice from the C57BL/6J strain and Fmr1^−/y on C57BL/6J background, 8T reversed test pulse depression in all slices before 2 weeks of age (Figure 2 a1,b1, Supplementary Figure S1a1,b1). By 2 weeks, about 16% of slices from the C57BL/6J wild-type mice showed STP, but none had L-LTP (Figure 2a1,2, Supplementary Figure S1a1). By 3 weeks, 31% of slices from the Fmr1^−/y mutant mice showed only STP and 7% showed L-LTP (Figure 2b1–3, Supplementary Figure S1b1). Between 3 and 4 weeks, 41% of slices produced STP and 26% produced L-LTP in the C57BL/6J wild-type mice (Figure 2a3,4, Supplementary Figure S1a2); whereas, in the Fmr1^−/y mutants, at 4 weeks 30% of slices produced STP and 40% produced L-LTP (Figure 2b4). At 5 weeks, L-LTP was reliably produced in more than 50% of slices from both the C57BL/6J wild-type mice and Fmr1^−/y mutants (Figure 2a5,b5). These findings suggest a gradual onset for the production of STP and L-LTP in the C57BL/6J strain, an effect that was apparently delayed in Fmr1^−/y.

Next, we tested mice from the 129SVE strain and Hevin^−/− on the 129SVE background. Before 2 weeks, all slices showed reversal of test pulse depression, 10% of slices from 129SVE mice showed minimal L-LTP (Figure 3a1), and 9% of slices from Hevin^−/− mice showed subtle STP (Figure 3b1), but most slices showed no potentiation at all. Between 2 and 3 weeks, 38% of slices showed either STP or L-LTP in the 129SVE mice (14% STP and 24% L-LTP, Figure 3a2,3, Supplementary Figure S2a2), contrasting with 33% of slices from Hevin^−/− showing L-LTP a week earlier by 2 weeks (Figure 3b2). At 4 weeks, 80% of slices showed L-LTP and 10% had STP in the 129SVE mice (Figure 3a4) versus 74% and 15% in Hevin^−/− between 3 and 4 weeks (Figure 3b3,4, Supplementary Figure S2b3). Thus, reliable L-LTP occurred at 4 weeks in the 129SVE strain (Figure 3a4) and even earlier, at 3 weeks, for Hevin^−/− (Figure 3b3).

In young adult rats and C57BL/6J mice (7–9 weeks old), L-LTP is saturated by eight trains of TBS (8T) delivered in stratum radiatum of hippocampal area CA1 (Cao & Harris, 2014). Saturating in this context means that another episode of TBS given 5 min after the first episode produces no additional LTP. We performed additional experiments demonstrating that 1T, 2T, and 8T produced L-LTP of the same magnitude and endurance at 4–5 weeks (Figure 4). Hence, the 8T paradigm used here produced saturating L-LTP, consistent with prior experiments (Abraham & Huggett, 1997; Cao & Harris, 2014; Kramar et al., 2012).

Both Fmr1^−/y and Hevin^−/− accelerated L-LTP onset relative to their wild-type backgrounds.

The probability of producing at least STP (including STP and L-LTP slices) or L-LTP was compared across ages, strains, and genotypes (Figure 5). More slices showed STP in Hevin^−/− mice by Week 3 than other strains; however, this effect was not statistically significant, and all strains had ≥50% of slices showing STP at this age (Figure 5a). The onset age for L-LTP was significantly earlier at 3 weeks for the Hevin^−/− (corresponds to the LTP probability between 50 and 75%) and at 4 weeks for Fmr1^−/y versus C57BL/6J (Figure 5b). The after onset age at 4 weeks in 129SVE and the Hevin^−/− (LTP probability ≥75%) was earlier when compared with 5 weeks in C57BL/6J and the Fmr1^−/y (Figure 5b). The differences between C57BL/6J and Fmr1^−/y and between 129SVE and Hevin^−/− were also significant at 3 weeks (Figure 5b). Once established, there were no significant differences in the magnitude of L-LTP between mouse strains or genotypes across time post-induction (Figure 5c).

Another measure of developmental onset would be the coincidence of no potentiation and potentiation occurring in different slices from the same animal. This coincidence was calculated as the percentage of animals that had slices showing both no potentiation and potentiation lasting for at least 1 hr out of the total number of animals at the onset age for each genotype (Table 3). More than 60% of animals at the L-LTP onset age showed this coincidence in support of the hypothesis that these were indeed the relative onset ages for each genotype.

TABLE 3.

The percent of animals at the onset age of L-LTP from each genotype that had some slices showing no potentiation and other slices from the same animal showing at least 1 hr of potentiation (coincidence)

Genotype	Coincidence (%)
C57BL/6J	66
Fmr1 −/y	75
129SVE	75
Hevin −/−	60

Abbreviation: L-LTP, late long-term potentiation.

3.3 |. The magnitude of the naïve fEPSP did not predict the developmental onset of L-LTP

The naïve fEPSP slopes were compared across experiments to test their potential effect in determining when L-LTP was first produced. In each of the three key age groups (before onset, onset, and after onset ages of L-LTP), the slices were divided as having no L-LTP at 3 hr or having L-LTP at 3 hr. No significant differences in the naïve slopes were detected across strains, genotypes, or age groups (Figure 6). Thus, the magnitude of the naïve responses did not determine the occurrence of L-LTP in these experiments.

3.4 |. Age dependence of basal synaptic transmission

To test for age- and genotype-dependent effects on synaptic transmission we measured the relationship between FV amplitudes and fEPSP slopes during baseline stimulation (Figure 7). In both the C57BL/6J and Fmr1^−/y mice, the ratio between FV and fEPSP was significantly less before than after the onset age of LTP, with no significant differences between these genotypes (Figure 7a–c). Similarly, this ratio was less prior to LTP onset in the Hevin^−/− and trended less in the 129SVE mice with no differences between the genotypes (Figure 7d–f). Thus, baseline synaptic transmission increases in parallel with the onset of L-LTP.

FIGURE 7.

Developmental change in relationship of fiber volley (FV) to field excitatory postsynaptic potential (fEPSP) slope. The fEPSP slope and FV values were averaged during baseline recordings from each slice. (a–c) The FV-slope relationship is stronger with age for C57BL/6J and Fmr1^−/y (two-way analysis of variance [ANOVA], interaction: F(1,42) = 0.002 (p > .05), age: F (1,42) = 21.3 (***p < .0001), Tukey’s post hoc age: (**p = .009 for C57BL/6J and *p = .01 for Fmr1^−/y), for both genotypes (F (1,42) = 3.52 (p > .05)). (d–f) The FV-slope ratio increased significantly across the two age groups for Hevin^−/− (two-way analysis of variance [ANOVA], interaction: F (1,35) = 0.96 (p > .05), age: F(1,35) = 11.35 (**p = .002), Tukey’s post hoc test p = .014) and did not differ significantly from its 129SVE background genotype (F (1,35) = 0.37 (p > .05)). The averages from each slice included 2–3 baseline measurements before onset and up to 12 baseline measurements after late long-term potentiation (L-LTP) onset to avoid test pulse-induced depression

3.5 |. Age dependence of the augmentation of L-LTP

Prior work in the rat hippocampus revealed that multiple time-separated episodes of TBS result in additional potentiation or augmentation of L-LTP. The timing of this effect is strain and age-related in the rat hippocampus (Bowden, Abraham, & Harris, 2012; Cao & Harris, 2012, 2014; Kramar et al., 2012; Manahan-Vaughan, 2000; Manahan-Vaughan & Schwegler, 2011). We tested whether a pair of 8T episodes spaced 180 min apart would augment L-LTP in mouse hippocampus (Figure 8a), as it did in rat. Two criteria were set for augmentation. First, the initial potentiation had to be ≥120% of the naïve response at 3 hr following the first 8T. Then, the second 8T had to elevate the fEPSP slope ≥10% above the initial L-LTP, and last for at least 70 min. Longer monitoring of the responses in slices from developing animals could become unreliable after 11 hr in vitro so experiments were terminated by 9 hr (i.e., 3 hr recovery plus 6 hr of recording).

Slices from the different mouse genotypes were compared in their respective age groups relative to the onset age when L-LTP first occurred. The time course of L-LTP and augmentation of L-LTP is illustrated for all four genotypes (Figure 8b–e). At the onset age of L-LTP, some slices from the Fmr1^−/y mutant mice that had threshold levels of initial L-LTP, produced augmentation of L-LTP; however, none of the other genotypes met the 10% augmentation criterion at their onset ages of L-LTP (Figure 8f). After the onset age of L-LTP, slices from all four genotypes produced augmentation of L-LTP (Figure 8b–f). To illustrate the age-dependent differences for the Hevin^−/− mutants, slices tested for augmentation of L-LTP at 3 weeks are shown with the onset group and 4 weeks with the after onset group (Figure 8f).

Even after the onset age of L-LTP, some slices showed no initial L-LTP. In those 4 week old slices, the second 8T also produced no potentiation and hence no augmentation (Figure 8g). In fact, for slices without initial L-LTP in the C57BL/6J and Hevin^−/− mice, the second 8T episode resulted in a significant reduction in fEPSP slope 45–70 min later (Figure 8g). This depression could reflect the failure to reverse the ongoing decline in the fEPSP slope normally observed over time in slices from developing animals that fail to produce initial L-LTP.

3.6 |. Two episodes of TBS do not enable L-LTP in slices initially lacking production of L-LTP

Since two episodes of 8T produced no potentiation in slices lacking initial L-LTP, further testing was done varying the timing (90 vs. 180 min intervals) and strength (8T vs. 1T) of the episodes (Figure 9a). Four-week-old C57BL/6J mice were chosen as this age was at the onset age when a single episode of 8T did not reliably produce L-LTP. When two episodes of 8T were spaced by 90 or 180 min, and STP was present initially for 1 hr, no additional STP was produced after the second 8T and the response dropped back to baseline or below by 3 hr (Figure 9b,c). When the gentler 1T episodes were spaced by 90 min, substantial potentiation was induced after both the first and the second 1T episodes; however, it did not last (Figure 9d,e). If the 1T episodes were spaced by 180 min, the second episode produced much less potentiation than the first, and it also did not last (Figure 9d,e). Thus, neither change in timing or strength produced reliable initial L-LTP nor augmentation of STP to produce L-LTP at the earlier developmental stage.

4 |. DISCUSSION

In the past, two major induction protocols have been used to discern the developmental onset of synaptic plasticity in the rat hippocampus. Repeated tetanic stimulation (three times at 100 Hz for 1 s each) first produced L-LTP at P15 (Harris & Teyler, 1984; Jackson, Suppes, & Harris, 1993), while the 8T paradigm produced L-LTP earlier, at P12 (Cao & Harris, 2012). Hence, the more efficient 8T paradigm was adopted here to investigate the developmental onset of plasticity in the mouse hippocampus. In rats, STP and L-LTP developmental onset occurred at the same time. In contrast, three stages of plasticity emerged in mouse hippocampus. First, STP gradually emerged between P10 and P28 in all four mouse genotypes. Second, L-LTP onset (25–50% of slices) occurred at different ages depending on mouse genotype. Third, the after onset age of L-LTP (>50% of slices) also depended on mouse genotype. In the C57BL/6J mice, L-LTP onset occurred between 3 and 4 weeks, and reliable L-LTP was not achieved until 5 weeks. Although some Fmr1^−/y mice showed L-LTP before 3 weeks, reliable L-LTP was also not achieved until 5 weeks in this genotype. In the 129SVE mice, L-LTP onset occurred by 3 weeks, reliable L-LTP was achieved by 4 weeks, and Hevin^−/− advanced this profile by 1 week.

4.1 |. Differential effects of development on basal synaptic transmission

The relationship of presynaptic FV amplitude to the slope of fEPSP provides a measure of the strength of synaptic transmission needed to evoke a postsynaptic response. The absolute magnitude of the naïve fEPSP slope used during TBS did not differ across strains or genotypes. In addition, for all strains and genotypes tested here, the ratio between fEPSP slope and the presynaptic FV was lower prior to, than after the onset of L-LTP. Earlier observations showed that the Fmr1/Fxr2 double knock-out disrupted this relationship, and the single Fmr1^−/y had no effect (Zhang, Hou, Klann, & Nelson, 2009), consistent with our findings. Thus, the differences in L-LTP onset ages were not explained by differences in the initial strength of activation or basal synaptic transmission in mice reported here, or previously in rats (Cao & Harris, 2012).

Test pulse depression, indicated by a gradual decline in fEPSP response to test pulses, has been ascribed to reversible silencing of AMPARs (Abrahamsson et al., 2007, 2008). In rats and all four mouse genotypes, test pulse depression no longer occurred at 5 weeks, coincident with the latest onset age of L-LTP in mice, but well after L-LTP onset in rats. Thus, differences between species, strains, and genotypes were not explained by a capacity to avoid test pulse depression during development.

4.2 |. Developmental metaplasticity

Some patterns of stimulation have no direct effect on synaptic strength but instead modulate the subsequent expression of plasticity, a phenomenon known as metaplasticity (Abraham & Bear, 1996; Abraham & Tate, 1997; Young & Nguyen, 2005). Spaced learning produces longer memories than massed learning, and the efficacy of memory is dependent on the interval between episodes of learning (Ebbinghaus, 1885; Fields, 2005). Similarly, spacing episodes of plasticity induction is considered to be a good model for understanding the cellular mechanisms of spaced learning (Kramar et al., 2012; Lynch & Gall, 2013; Lynch, Kramar, Babayan, Rumbaugh, & Gall, 2013). Regarding LTP, sufficient time must pass between the TBS episodes to augment LTP after a second episode of TBS. In adult rat, augmentation of previously saturated LTP was first observed at a 90 min spacing between bouts of 8T, and prolonging the time between TBS episodes increased the probability of augmentation (Cao & Harris, 2014). The delay between episodes of TBS reflects the time needed to enlarge the postsynaptic area after the initial induction of LTP in adult rats (Bell et al., 2014). In adults, this synaptic enlargement is also homeostatically balanced by stalled spine outgrowth that reflects temporal dynamics of resource reallocation to clusters of potentiated synapses (Bell et al., 2014; Bourne & Harris, 2007; Chirillo, Waters, Lindsey, Bourne, & Harris, 2019).

The patterns of augmentation of L-LTP are also developmentally regulated. In rats, a second episode of 8T delivered 90 min after the first episode produced L-LTP at P10-P11 but not at P8-P9. In the C57BL/6J mice, applying a second 8T episode 90 or 180 min after the first did not produce L-LTP even at 4 weeks of age, when STP could be produced in most slices. Instead, in mouse hippocampus, augmentation could be achieved only after L-LTP was reliably established for C57BL/6J, 129SVE, and Hevin^−/− genotypes at after onset age. Curiously, the augmentation of L-LTP was observed earlier in Fmr1^−/y mice than other genotypes, in slices that had initial L-LTP. These observations suggest that the development of L-LTP and metaplasticity involve processes that depend on species, strain, and genotype, but are independent from those processes that result in STP.

4.3 |. Genetic differences between mice and rats

Such striking differences between mice and rats in their developmental profiles of synaptic plasticity are consistent with genetic analysis. Almost half of the ∼1 K genes tested so far also show differential expression between mouse and rat hippocampal dendrites, with much less divergence in the other tissues (Francis et al., 2014). There are also large differences between rat and mouse adult hippocampal neurogenesis, a process that is especially important for learning and memory (Lazarov & Hollands, 2016; Snyder et al., 2009). Rats have more adult-born, death-resistant neurons, and these neurons mature faster in rats than in mice. The young neurons show a much higher contribution to fear learning tasks in rats than mice (Miller & Hen, 2015). These genetic and functional differences are consistent with rats having an earlier and more discrete onset age of L-LTP than mice.

4.4 |. Contrasting onset ages of L-LTP and spinogenesis in rat and mouse hippocampus

In this work, we chose gene manipulations that had been reported to alter dendritic spines and synaptic plasticity. Fmr1^−/y neurons have been characterized by an overproduction of underdeveloped spines that might not support the plasticity events (He & Portera-Cailliau,−2013). Treatment of neonatal Fmr1^−/y mice with the antibiotic minocycline resulted in better learning outcomes along with enhanced spine maturation (Bilousova et al., 2009). Furthermore, in adult Fmr1^{−/ y} mice (3–5 months old), spaced trials rescued learning deficits (Seese, Wang, Yao, Lynch, & Gall, 2014). However, in all the mouse genotypes tested here, including the developing Fmr1^−/y mice, STP was not augmented to produce L-LTP upon spaced bouts of 8T in slices that had no initial L-LTP. Thus, future work will be needed to know whether those spaced learning effects resulted from the augmentation of L-LTP or other processes.

Hevin is required for the development of thalamocortical connectivity between P14 and P25 mouse cortex (Risher et al., 2014). Moreover, when Hevin is absent, cortical dendritic spines show significant immaturity demonstrated by fewer but longer spines and a distinct refinement problem. In the second week of development, cortical spines often receive innervations from one cortical and one thalamic axon. By P25, these multiply innervated spines are refined to receive either a thalamocortical or intracortical synapse in wild-type (129SVE) mice. In the Hevin^−/− mice this pruning effect does not occur uniformly and the ratio between thalamocortical and intracortical inputs is altered, retaining more of the intracortical synapses at the expense of thalamocortical connections. The role of Hevin in refining hippocampal dendritic spines is unknown; however, its absence in Hevin^−/− mice appears to advance the developmental onset age of L-LTP. This finding is consistent with the hypothesis that lack of refinement of CA3-CA1 synapses by Hevin promotes the earlier maturation of plasticity. Application of Hevin protein to autaptic cortical neurons results in a robust induction of NR2B containing NMDAR activity (Singh et al., 2016). Perhaps, the lack of regulation of the NR2B subunit in the Hevin^−/− mice stimulates the maturation of synapses at an earlier developmental stage.

In rats, the developmental onset of L-LTP at P12 is coincident with the emergence of dendritic spines, suggesting dendritic spines are necessary for sustained synaptic plasticity (Cao & Harris, 2012; Fiala, Feinberg, Popov, & Harris, 1998; Kirov, Goddard, & Harris, 2004). Initial 3D reconstructions in perfusion-fixed rat hippocampus show evidence for mature dendritic spines at P12, but not at P8 or P10 (Smith, 2019). Preliminary data from rat hippocampal slices showed that 90 min after the initial 8T, dendritic spines were not produced at P8 (Harris, Watson, Kuwajima, & Cao, 2012). In rat hippocampal slices at P10–11, preliminary data suggest that spines were produced 90 min after the initial 8T (Smith, 2019). Thus, a shift from shaft synapses and filopodia to spines might account for this developmental shift in L-LTP onset for rat hippocampus.

The onset of L-LTP in Fmr1^−/y at 4 weeks was delayed relative to the background C57BL/6J strain at 3 weeks. This pattern contrasted with the earlier onset age in Hevin^−/− by 2 weeks, relative to its background 129SVE strain between 2 and 3 weeks. All genotypes showed reliable L-LTP by 5 weeks. These findings contrast with the developmental onset ages of dendritic spines. Mature dendritic spines have been reported by P15 in C57BL/6J mouse hippocampus (Bilousova et al., 2009), well before the onset age of reliable L-LTP at 5 weeks. Confocal microscopy studies reveal a few mushroom spines by 9–12 days in organotypic slices from mouse hippocampus (Parnass, Tashiro, & Yuste, 2000). Similarly, at 14 days in organotypic slices from C57BL/6J and Fmr1^−/y more than 40% of the protrusions were classified as mushroom spines, although less than 10% had mature heads with a diameter greater than 0.5 μm (Bilousova et al., 2009). Reconstructions from serial section EM show mature spines by P24 in the C57BL/6J hippocampus (Nikonenko et al., 2013). Thus, the onset of L-LTP appears to be later than the onset of dendritic spines in mouse hippocampus, suggesting that spines might be necessary but not sufficient.

4.5 |. Other factors that may influence variation in the developmental onset of L-LTP

The wide variance in L-LTP onset ages among individual mice may reflect divergence in many factors that lead to the maturation of neurons. In rats, the discrete onset age of L-LTP may reflect less variation in these factors between animals. Prior work also showed a high rate of failure of LTP induction in P16-P30 C57BL/6J mice when five tetanic stimuli were used in normal calcium concentration (Adesnik & Nicoll, 2007). The success rate of LTP induction was improved by increasing the calcium concentration. One possible explanation is the involvement of different mechanisms of LTP over developmental stages. There is evidence that in 2-week-old mice of mixed 129SVE-C57BL/6J background the initial LTP is mediated by postsynaptic insertion of GluR2-lacking subunits which are later exchanged for GluR2-containing AMPA receptors that are less permeable to calcium (Jia et al., 1996; Plant et al., 2006; Purkey et al., 2018; Sanderson, Gorski, & Dell’Acqua, 2016). LTP in young mice is less dependent on the phosphorylation of GluR1 under mild stimulation conditions (Adesnik & Nicoll, 2007; Lee et al., 2003; Lu et al., 2007; Wikstrom, Matthews, Roberts, Collingridge, & Bortolotto, 2003). Thus, in addition to age, the exact induction protocol may influence the success of LTP.

Another factor that could influence the exact onset age of LTP is the maturation of the inhibitory system and the resulting excitation/inhibition (E/I) balance (Ben-Ari, Khalilov, Kahle, & Cherubini, 2012; Eichler & Meier, 2008). The GABAergic system is depolarizing early during development but switches with maturation to hyperpolarizing. Varying the inter-stimulus interval shows that paired-pulse inhibition first occurs in the developing rat hippocampus at P6 (Harris & Teyler, 1983). Patch-clamp experiments, in hippocampus from both mice and rats, reveal that prior to the maturation of the GABAergic system, the degree of postsynaptic depolarization needed to induce LTP is less (Meredith, Floyer-Lea, & Paulsen, 2003). Furthermore, the Fmr1^−/y mice show a dramatic E/I imbalance mediated by reduced GABAergic inhibition in both hippocampus and subiculum (Cea-Del Rio & Huntsman, 2014; Curia, Papouin, Seguela, & Avoli, 2009; Eichler & Meier, 2008; Paluszkiewicz, Martin, & Huntsman, 2011; Sabanov et al., 2017). The saturating TBS protocol used here would overcome inhibitory effects at all ages; hence, the potential E/I imbalance would not explain the gradual onset of enduring LTP in mice (Pike, Meredith, Olding, & Paulsen, 1999; Thomas, Watabe, Moody, Makhinson, & O’Dell, 1998).

The gradual onset of L-LTP in mice could also stem from variation in the rate of maturation of neurons along the septal temporal axis (Altman, 1966; Altman & Das, 1966; Angevine, 1965; Bayer, 1980a, 1980b; Bayer & Altman, 1975). Taking four slices from the middle of the mouse hippocampus might overlap this developmental axis, whereas the larger rat hippocampus does not. This hypothesis is further supported by our finding that at onset, slices from the same mouse hippocampus could express no potentiation or potentiation lasting ≥1 hr. This coincidence occurred in a majority of mice from all genotypes at the onset ages of L-LTP. Such a finding might also reflect coincidental natural variation between neighboring hippocampal slices in the capacity for L-LTP and spine maturity. Thus, future experiments will need to measure both the capacity for L-LTP and spine structure in individual mouse slices to address the necessity and sufficiency of dendritic spines for L-LTP and the impact of these mutations on that process.