Introduction

In PKMζ-knockout mice (PKMζ-KO) the key long-term processes of hippocampal LTP and long-term memory appear normal, and LTP is still inhibited by the aPKC inhibitor ZIP (Lee et al., 2013; Tsokas et al., 2016; Volk et al., 2013). In the same knockout mice, LTP is eliminated in medial prefrontal cortex (mPFC) (Kniffin et al., 2025; Sacktor, 2026). These results seem inconsistent, suggesting that PKMζ is crucial only for LTP maintenance in mPFC, but not in hippocampus. An alternative hypothesis, however, is that another ZIP-sensitive molecule compensates in hippocampus for the absence of Prkcz, the gene for PKMζ. In Tsokas et al., 2016, multiple members of the PKC gene family were found to increase expression in compensation for the loss of PKMζ, and the most promising candidate was the other aPKC, the ZIP-sensitive PKCι/λ, since closely related genes often compensate for one another (Conant and Wagner, 2004; El-Brolosy and Stainier, 2017; Gu et al., 2003; White et al., 2013). Notably, PKCι/λ (referred to as PKCι) is critical to the initial generation of the early phase of LTP and short-term memory in wild-type (WT) mice, and PKMζ compensates for the conditional knockout of the PKCι gene (Prkci) to support both short- and long-term processes (Ren et al., 2013; Sheng et al., 2017; Tsokas et al., 2016; Wang et al., 2016). Perhaps LTP and long-term memory maintenance might require a persistent kinase after all — but not always PKMζ.

The notion that the persistence of PKCι is a biochemical mechanism responsible for late-phase LTP and long-term memory in PKMζ-knockout mice raises two questions: 1) Does PKCι persist in LTP and long-term memory in the absence of PKMζ, and 2) Does eliminating both aPKCs abolish enduring LTP and long-term memory? We address these questions using conditional and double-knockout transgenic mouse strategies.

Results

As neither inducible nor constitutive PKMζ-KO eliminates hippocampal LTP (Volk et al., 2013), it is important to know if compensatory increases in other PKCs are present in the hippocampus of conditional PKMζ-KO mice (ζ-cKO), as observed in PKMζ-null mice (Tsokas et al., 2016). In the hippocampus of PKMζ-null mice, basal levels of PKCι and the conventional PKCβI increased (Tsokas et al., 2016). To examine the hippocampus of ζ-cKO mice, adult CaMKIIα promoter-driven-CreER^T2Prkcz^fl/fl mice were injected with tamoxifen. One week later, expression of PKMζ in hippocampus decreased, and there was a compensatory increase in PKCι, as well as PKCι phosphorylated on its activation-loop (Figure 1A, Figure 1 — figure supplement 1A, Figure 1 — table supplement 1A, 2A). In addition, the expression of all four conventional PKCs (α, βI, βII, γ) increased, as did phosphorylation of the conventional PKC activation-loop (Figure 1B, Figure 1 — figure supplement 1A, 2, table supplement 1B, 2A). However, there was no change in the expression of the four novel PKC isoforms (δ, ε, η, θ) or phosphorylation of the activation-loop of PKCε (Figure 1B, Figure 1 — figure supplement 1A, 2, table supplement 1B, 2A). There also was no change in either the level of CaMKIIα or T286-autophosphorylation of CaMKIIα, which initiates Ca²⁺-independent autonomous kinase activity (Miller and Kennedy, 1986) and LTP induction (Bayer and Giese, 2024; Tullis et al., 2023) (Figure 1 — figure supplement 1B, table supplement 2B). Thus, both conditional and constitutive PKMζ-KO mice express compensatory increases in PKCι, as well as other PKCs, in hippocampus.

Figure 1.

Does the increase in PKCι persist in long-term memory, such that one aPKC substitutes for the other? Previous research has shown that PKMζ expression in CA1 stratum (st.) radiatum remains elevated for at least a month after training transgenic mice on a spatial memory task, as a component of a PKMζ-engram that traces the tri-synaptic circuit in the dendritic compartments of memory-tagged neurons (Han et al., 2026; Hsieh et al., 2021). We determined if PKMζ-cKO mice would also exhibit a compensatory increase of PKCι in st. radiatum. The PKMζ gene was ablated in adult CaMKIIα-promoter-driven-CreER^T2Prkcz^fl/fl mice (Figure 2A), and 3 weeks later the mice were trained on the active place avoidance task to produce spatial memory. One week after training, compared to vehicle-injected littermates, the PKMζ-cKO mice showed a decrease to ∼25% in PKMζ expression and an increase to ∼400% in PKCι expression (Figure 2B, Figure 2 — table supplement 1), along with memory retention (Figure 2C). Thus, spatial conditioning of ζ-cKO mice creates long-term active place avoidance memory and induces compensatory, persistent increased expression of hippocampal PKCι.

Figure 2.

Does the persistent PKCι expression substitute for the function of PKMζ in LTP and memory of PKMζ-KO mice? As PKCι-null mice are embryonically lethal (Seidl et al., 2013), we determined the functional significance of the compensatory increase of PKCι for LTP by injecting an adeno-associated virus (AAV) expressing Cre-recombinase in one hippocampus of PKCι-floxed/PKMζ-null (Prkci^fl/fl-Prkcz^-/-) mice (Sheng et al., 2017) (Figure 3A). The contralateral hippocampus was injected with a control AAV-eGFP, and ex vivo slices prepared for electrophysiology. PKCι decreased in the ipsilateral hippocampus to ∼20% compared to the contralateral hippocampus (Figure 3A,B).

Figure 3.

If PKCι is important for enduring LTP in PKMζ-KO mice, then this decrease should disrupt LTP. High-frequency afferent stimulation (HFS) of Schaffer collateral/commissural-CA1 synapses in the ipsilateral ι/ζ-dKO slices produced no persistent change in the residual PKCι and a transient LTP only lasting ∼1-2 h (Figures 3, 4A; Figure 4 — figure supplement 1A). In contrast, HFS of the contralateral slices, expressing PKCι but not PKMζ, induced persistent increases of the PKCι to ∼200% and compensatory late-LTP, both lasting at least 3 h, the duration of the recordings (Figures 3 and 4A, Figure 4 — figure supplement 1A). Slices from hippocampus of Prkci^fl/fl-Prkcz^-/-mice injected with AAV expressing Cre by the CaMKIIα-promoter to selectively ablate the PKCι gene in excitatory neurons resulted in similar transient LTP (Figure 4A, inset; Figure 4 — figure supplement 1B, 2A). In addition, we showed the virus injected into PKCι-floxed mice that express PKMζ (Prkci^fl/fl-Prkc^+/+mice) resulted in compensated early-LTP, as previously reported (Sheng et al., 2017) (Figure 4 — figure supplement 1C, 2B). Thus, individual knockout of each aPKC shows compensatory LTP, whereas double-knockout of both aPKCs eliminates late-LTP.

Figure 4.

We tested if late-LTP could be induced in ι/ζ-dKO hippocampus by increasing HFS from two trains, 20 sec apart, which is optimized to produce an early onset of late-LTP (Tsokas et al., 2007), to four trains, spaced 5 min apart, which is optimized to produce maximal late-LTP (Scharf et al., 2002; Serrano et al., 2005) (Figure 4B; Figure 4 — figure supplement 1D). Stronger stimulation induces LTP in ι/ζ-dKO slices that lasts only ∼1-2 h. This LTP was no longer expressed at 3 h post-stimulation, at which time field excitatory postsynaptic potentials (fEPSPs) did not significantly differ from baseline fEPSPs before HFS.

To determine if PKCι supports long-term memory in the absence of PKMζ, we compared the effects on spatial memory of ι/ζ-dKO to ζ-KO by injecting PKCι^fl/fl-PKMζ^−/− littermates bilaterally in hippocampus with either AAV-Cre or AAV-eGFP (Figure 5A). Mice received three 30-min training trials separated by 24 hours and a final retention test without shock the next day. We assessed two measures of short-term memory (Figure 5B). First, we examined the time to each entry into the shock zone in the first training trial, as compared to the entries into the shock zone during the pretraining session with the shock off. Avoidance behavior increased within the first 30-min training session in both the ι/ζ-dKO and ζ-KO mice, and the increases were indistinguishable. Second, we measured the maximum avoidance time within each session. Maximum avoidance time reflects the time between shocks, which is controlled by the animal’s behavior and within-trial memory. Note that compared to pretraining inter-shock zone entries, the maximum avoidance time for ζ-KOs and ι/ζ-dKOs increased in the first training trial, indicating that both genotypes acquired short-term memory for the shock zone. The maximum avoidance time for the ζ-KO increased over the 3 daily training sessions, whereas the ι/ζ-dKO did not, suggesting impaired long-term memory in the ι/ζ-dKO compared to the ζ-KO. Our main measure of long-term memory was time to first entry into the shock zone at the beginning of each session (Figure 5C). This increases with long-term memory maintained across days from previous trials. The ζ-KOs’ first entry times increased dramatically from pretraining to both session 3 and retention test, indicating that mice with PKCι maintain long-term memory. In contrast, ι/ζ-dKO mice displayed a minimal increase at session 3 that was not significantly different from pretraining, and no detectable difference between pretraining and the retention test, indicating loss of spatial long-term memory.

Figure 5.

Discussion

Here we found that a second aPKC becomes persistently active to maintain late-LTP and long-term memory in the PKMζ-κΟ. The isoform most closely related to PKMζ, PKCι, which normally plays only a transient role in LTP and short-term memory in WT mice, persistently increases expression in LTP and long-term memory in PKMζ-KO mice (Figures 2, 3). Although LTP is present if PKMζ or PKCι is individually knocked out, when both are genetically eliminated by double-knockout there is no enduring hippocampal LTP or long-term spatial memory (Figures 4, 5, Figure 4 — figure supplement 2B).

The ι/ζ-dKO exhibits an early transient synaptic potentiation that could not be maintained into the late-phase of LTP (Figure 4, Figure 4 — figure supplement 1A, B, and D). In addition, bilateral hippocampal ι/ζ-dKO eliminated long-term spatial memory but did not prevent learning, short-term memory, or expression of the place avoidance behavior (Figure 5). As PKCι is a key contributor to early-LTP and short-term memory in WT mice (Ren et al., 2013; Wang et al., 2016), there must be additional compensation for short-term processes present in the aPKC-dKO.

Our finding that PKCι can substitute for PKMζ raises the question of how the compensation is induced. After gene knockout, compensatory gene expression can be triggered by the fragments of mRNA that are produced by transcription upstream of the site of Cre-recombination (El-Brolosy et al., 2019; El-Brolosy and Stainier, 2017; Ma et al., 2019). This could explain why constitutive and conditional PKMζ-KOs produce compensation and normal-appearing late-LTP/long-term memory, whereas PKMζ-shRNA and PKMζ-antisense oligodeoxynucleotides that retain full-length PKMζ mRNA transcription do not induce compensation and eliminate late-LTP/long-term memory (Tsokas et al., 2016; Wang et al., 2016). In contrast to the hippocampus, both early- and late-LTP are absent in the medial prefrontal cortex of PKMζ-KO mice (Kniffin et al., 2025). This suggests that the PKCι activation mediating early-LTP in the hippocampus of WT mice may not be available to compensate for the loss of PKMζ in the medial prefrontal cortex of ζ-KO mice (Sacktor, 2026).

Once increased, how does PKCι accomplish maintenance? PKMζ maintenance is linked to the kinase’s second messenger-independent, persistent enzymatic activity (Sacktor et al., 1993). PKMζ is autonomously active because the kinase is an independent catalytic domain that lacks the PKCζ autoinhibitory regulatory domain (Hernandez et al., 2003; Sacktor et al., 1993). PKCι, however, is a full-length PKC isoform with a regulatory domain that inhibits its catalytic domain. Therefore, for it to compensate for PKMζ, PKCι likely requires additional posttranslational mechanisms to persistently activate and localize the kinase at active synapses (Figure 2B, 3). PKCι can be activated by postsynaptic proteins such as p62 that bind its regulatory domain (Jiang et al., 2009; Ren et al., 2013). The protein-protein interaction that activates PKCι may sustain its kinase action longer than the rapidly metabolized lipid second messengers that stimulate the conventional/novel PKCs, thus allowing one persistently active aPKC to replace the other.

The maintenance properties of PKMζ depend not only on continuous activity but also on continuous binding to the postsynaptic scaffolding protein KIBRA/WWC1 (kidney and brain protein/WW and C2 domain protein 1) (Tsokas et al., 2024). This sustained interaction perpetually targets PKMζ to active synapses in a process of persistent synaptic tagging (Hsieh et al., 2026; Shouval et al., 2025; Tsokas et al., 2024). KIBRA also binds PKCι, albeit more weakly than PKMζ (Tsokas et al., 2024). In WT mice, the strong binding of PKMζ to KIBRA might allow it to displace PKCι at active synapses in the transition from early- to late-LTP. In contrast, in the hippocampus of PKMζ-KO mice, the PKCι at active synapses would not be replaced by PKMζ. PKMζ and PKCι also compete for binding to PAR3 (partitioning defective protein 3), another postsynaptic protein that localizes aPKCs within neurons (Parker et al., 2013; Zhang and Wei, 2022). Thus, in the PKMζ-KO synaptic tags such as KIBRA and PAR3 that are components of the PKMζ maintenance mechanism may now shift to PKCι to sustain hippocampus-dependent LTP and long-term memory.

Our finding that persistent biochemical action by PKMζ or PKCι is crucial for maintaining synaptic potentiation and memory appears at odds with the widely held notion that synapses sustain memory through stable changes in structure without the need for ongoing enzymatic activity dedicated to storing information. The structural view, as introduced, hypothesized, and established by Ramón y Cajal, Hebb, and Kandel, has led to identifying structural plasticity and non-enzymatic molecules that contribute to establishing LTP and memory (Bailey and Kandel, 1993; Hebb, 1949; Ramón y Cajal, 1894), including cytoskeletal actin and perineuronal nets (Matus, 2000; Tsien, 2013). Memory maintenance by ongoing, persistent biochemical processes was an alternative hypothesis proposed by Crick, Lisman, and Schwartz (Crick, 1984; Lisman, 1985; Schwartz, 1993). The search for biochemical processes that maintain LTP for hours and long-term memory for days focused on two persistently active protein kinases, CaMKII and PKMζ (Lisman, 2017; Sacktor and Fenton, 2018). The kinase action of CaMKII, however, is only crucial for initiating but not perpetuating LTP and memory (Bayer and Giese, 2024; Tullis et al., 2023). By contrast, PKMζ in the physiological conditions of WT mice, and PKCι in the compensatory response of PKMζ-KO mice, play the crucial role of maintaining the molecularly distinct, enduring late phase of LTP and long-term memory (Pastalkova et al., 2006; Shema et al., 2011; Shema et al., 2007; Tsokas et al., 2024; Wang et al., 2016). Future work will be required to determine if persistent changes in synaptic structure are sustained by the ongoing aPKC activity that maintains long-term memory (Chen et al., 2014).

Materials and Methods

Reagents

Reagents were from MilliporeSigma unless otherwise stated. The tamoxifen vehicle for i.p. injections was sunflower seed oil.

Animals

Male mice were on C57BL/6 background and at least 4-months-old for all experiments. The PKMζ-null mouse line was previously described (Lee et al., 2013) and provided by Robert O. Messing (Univ. Texas at Austin, TX, USA). Conditional PKMζ and PKCι mice were generated by Sourav Ghosh as previously described (El Allam et al., 2024; Mercau et al., 2024; Scott et al., 2019). Camk2a-CreER^T2mice were from Jackson Labs.

Hippocampal slice recording and stimulation

Acute mouse hippocampal slices (450 µm) were prepared as previously described (Tsokas et al., 2016; Tsokas et al., 2019). Hippocampi were dissected, bathed in ice-cold dissection buffer, and sliced with a McIlwain tissue slicer in a cold room at 4°C. The dissection buffer contained (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 11 glucose, 10 MgCl₂, and 0.5 CaCl₂, and was bubbled with 95% O₂/5% CO₂ to maintain pH at 7.4. After dissection the slices were transferred to an Oslo-type interface recording chamber (31.5 ± 1°C) (Tsokas et al., 2019). The recording superfusate consisted of (in mM): 118 NaCl, 3.5 KCl, 2.5 CaCl₂, 1.3 MgSO₄, 1.25 NaH₂PO₄, 24 NaHCO₃, and 15 glucose, bubbled with 95% O₂/5% CO₂, with a flow rate of 0.5 ml/min.

Field EPSPs were recorded with a glass extracellular recording electrode (2–5 MΩ) placed in the CA1 st. radiatum, and concentric bipolar stimulating electrodes (CBBRE75 and 30200; FHC, Bowdoin, ME) were placed on either side within CA3 or CA1. Test stimulation rate was once every 30 sec, alternating every 15 sec between stimulating electrodes. Based upon a pre-established exclusion criterion, a slice was not used if fEPSP spike threshold was < 2 mV on initial input-output analysis. Pathway independence was confirmed by the absence of paired-pulse facilitation between the two pathways. A single stimulating electrode was used for immunocytochemistry with a test stimulation rate of once every 30 sec. HFS optimized to produce a relatively rapid onset of protein synthesis-dependent late-LTP, consisted of two 100 Hz-1 s tetanic trains, at 25% of spike threshold, spaced 20 sec apart (Tsokas et al., 2005). HFS optimized to produce maximal late-LTP consisted of four 100 Hz-1 s tetanic trains, at 25% of spike threshold, spaced 5 min apart (Scharf et al., 2002; Serrano et al., 2005). The maximum slope of the rise of the fEPSP was analyzed on a PC using the WinLTP data acquisition program (Anderson and Collingridge, 2007).

Immunoblots and Immunohistochemistry

Immunoblots of total hippocampus were performed as previously described, using antibodies in the Key Resources Table (Tsokas et al., 2016). Immunoblots were stained with multiple antisera to visualize multiple PKCs on the same immunoblot. Isoforms with similar molecular weights (e.g., the cPKCs) were either stained with antisera of different species or examined on separate blots.

Quantitative immunohistochemistry for Figure 3 and Figure 4 — figure supplement 2 was as described (Hsieh et al., 2021; Tsokas et al., 2024), using mouse anti-PKCι primary antibody (1:1000, E-7, Santa Cruz SC-376344).

Immunohistochemistry for Figure 2B was performed as follows. Free-floating sections were permeabilized with phosphate-buffered saline (PBS) containing 0.1% Tween20 (PBS-T) for 1 h at room temperature and blocked with 10% normal goat serum in PBS-T (blocking buffer) for 2.5 h at room temperature. One batch of sections was incubated overnight at 4 °C with rabbit anti-PKMζ C-2 antisera primary antibody (1:1,000) (Hernandez et al., 2003) and a second batch of sections with rabbit anti-PKCι (1:1000, Cell Signaling #2998S) in blocking buffer. After washing 3 times for 10 min each in PBS-T, both batches of sections were incubated with the secondary antibody goat anti-rabbit-Alexa 647 (1:500 in blocking buffer; Jackson ImmunoResearch) for 2 h at room temperature. After washing 3 times for 10 min each in PBS-T and extensive washing with PBS, the sections were mounted with Vectashield with 4′,6-diamidino-2- phenylindole (DAPI, Vector Laboratories). Three sections from each mouse were examined using an upright Leica SP8 confocal microscope and analyzed using ImageJ (version 1.53a). For each section, 8.5 µm-thick Z-stacks of the dorsal CA1 were created using the maximum intensity projection function in ImageJ. For each st. pyramidale, radiatum, and lacunosum-moleculare, two square regions of interest were centered in each stack. Measurements were made from each mouse in each region of interest. The raw integrated density (defined as the sum of the values for all pixels) of the Z-stack region of interest expressing the fluorescent label was measured for the volume of target pixels, and the average of each measurement was taken as representative for the region and each mouse.

AAV injections

Mice were anesthetized in a closed chamber filled with the inhalation anesthetic isoflurane (RWD Life Science, R510-22-10) and then fixed in a stereotaxic apparatus (Stoelting Co.). Anesthesia was maintained with isoflurane inhalation (1%−2.5% via trachea). The eyes of the mice were safeguarded using erythromycin ophthalmic ointment (0.5%). The skull was exposed and cleaned using 3% hydrogen peroxide. Small holes in the skull were then drilled with the following stereotaxic coordinates: left hippocampus (triple injection: AP: −1/ ML: −0.7/ DV: − 1.65; AP: −1.8/ ML: −1.5/ DV: −2; AP: −2.7/ ML: −2/ DV: −2, below the skull surface) and right hippocampus (triple injection: AP: −1/ ML: +0.7/ DV: −1.65; AP: −1.8/ ML: +1.5/ DV −2; AP: − 2.7/ ML: +2/ DV: −2, below the skull surface). The virus was injected using a 34-gauge needle with a Hamilton syringe at 0.1 µl/min rate into target regions. At all injected points, the tip of the needle was positioned 0.05 mm below the target coordinate and returned to the target site after 2 min. After injection, the needle stayed in place for an additional 7 min and was slowly withdrawn. AAVs expressing Cre-recombinase and eGFP were from Addgene. For physiology, 0.5 ul of virus pENN.AAV.CMVs.PI.Cre.rBG (AAV2.9) (1 x 10¹³ viral genomes [vg]/ml) was injected into one CA1, and 0.5 ul of virus pAAV.CMV.PI.EGFP.WPRE.bGH (AAV2.9) (1 x 10¹³ vg/ml) was injected in the contralateral side. Virus pENN.AAV.CamKIIa 0.4 Cre SV40 (AAV2.9) (1 x 10¹³ vg/ml) was used to express Cre-recombinase by the CaMKIIα promoter.

Conditioning

Active place avoidance was conducted with a commercial computer-controlled system (Bio-Signal Group, Acton, MA). The mouse was placed on a 40-cm diameter circular arena rotating at 1 rpm. The specialized software, Tracker (Bio-Signal Group, Acton, MA), was used to detect the animal’s position 30 times per second by video tracking from an overhead camera. A clear wall made from polyethylene terephthalate glycol-modified (PET-G) was placed on the arena to prevent the animal from jumping off the elevated arena surface. A 5-pole shock grid was placed on the rotating arena, and the shock was scrambled across the 5-poles when the mouse entered the shock zone. All experiments used the “Room+Arena-“ task variant that challenges the mouse on the rotating arena to avoid a shock zone that was a stationary 60° sector (Pastalkova et al., 2006). Every 33 ms, the software determined the mouse’s position, whether it was in the shock zone, and whether to deliver shock. After the animal enters the shock zone for 500 ms, a constant current foot-shock (60 Hz, 500 ms) was delivered and repeated with an interval of 1500 ms until the mouse left the shock zone. The shock intensity was 0.2 or 0.3 mA, which was the minimum amplitude to elicit flinch or escape responses. The animal was forced to actively avoid the designated shock zone because the arena rotation periodically transported it into the shock area. A pretraining period on the apparatus equivalent in time to a training session, but without shock, was provided.

The tracked animal positions with timestamps were analyzed offline (TrackAnalysis, Bio-Signal Group, Acton, MA) to extract several end-point measures. The time to first enter the shock zone estimates ability to avoid shock and was taken as an index of between-session long-term place avoidance memory. Short-term memory was assessed by two measures. First, the times to each entry into the shock zone in the first training trial were compared to the times for each entry into the shock zone with the shock off during the pretraining session. Avoidance behavior is observed as an increase in the amplitude of the times for entering the shock zone. Second, the maximum time without receiving a shock was determined for each session. Short-term memory for avoidance behavior is measured as an increase in the maximum time between entrances into the shock zone without shock during pretraining and the maximum time between shocks in the first training trial.

For Figure 5 the training schedule was as follows: 1 day after a 30-min pretraining session, the animals received three 30-min training trials, with an intertrial interval of 1 day. Long-term memory retention was tested the following day without shock. Pre-established exclusion criterion was if cannulae were found to be incorrectly targeted. No mice were excluded.

Statistics

All experiments were performed blindly, except LTP experiments comparing slices transfected with AAV-eGFP vs. AAV-Cre because the overexpressed eGFP is discernible by visual examination during recordings. Sample sizes vary for the different experimental approaches (biochemistry, extracellular field potential physiology, and behavior). The hypothesis that PKMζ is compensated predicts all-or-none effects in the experiments, and this provided a basis for sample size estimates. Power analyses were performed using G*Power Version 3.1.9.7 with α = 0.05 and β = 0.8 and large effect sizes of 1.5–2.0. The effect size estimates were based on prior studies that demonstrated essentially all-or-none effects of PKMζ inhibition on the immunoblot, immunohistochemical, physiological, and behavioral assays used here (Hsieh et al., 2021; Tsokas et al., 2024; Tsokas et al., 2016). Two-population Student t tests with Bonferroni corrections were performed to compare protein levels by immunoblot and immunocytochemistry in the PKMζ-cKO and control mice. For LTP experiments the responses to test stimuli were averaged across 5 min for statistical comparisons. Repeated measures ANOVA was used to compare the change in the potentiated response at the time points described. Multi-factor comparisons were performed using mixed-design ANOVA with repeated measures, as appropriate. The degrees of freedom for the critical t values of the t tests and the F values of the ANOVAs are reported as subscripts. Post-hoc multiple comparisons were performed by Newman-Keuls tests as appropriate. Statistical significance was accepted at P < 0.05. Effect sizes for binary comparisons and one-way ANOVAs are reported as Cohen’s d and as η²_p for multi-factor ANOVA effects.

Supplementary Materials

Figure 1 — figure supplement 1.

Figure 1 — figure supplement 2.

Figure 4 — figure supplement 1.

Figure 4 — figure supplement 2.

Figure 1 — table supplement 1.

Figure 1 — table supplement 2.

Figure 2 — table supplement 1.

Data Availability

All data are available in the main text or the supplementary materials.

Acknowledgements

P.T. is an Alexander S. Onassis Public Benefit Foundation Scholar.

Additional information

Funding

National Institutes of Health grant R37 MH057068 (TCS)

National Institutes of Health grant R01 MH115304 (TCS and AAF)

National Institutes of Health grant R01 NS105472 (AAF)

National Institutes of Health grant R01 MH132204 (AAF)

National Institutes of Health grant R01 NS108190 (PJB and TCS)

The Garry & Sarah S. Sklar Fund (PT)

Author contributions

Conceptualization: TCS, AAF

Methodology: PT, CH, AG-P, LK, SG

Investigation: PT, CH, AG-P, LK, LMR-V, DAC, KDA, HJHS, SK, BJW, SS, REF-O

Visualization: CH, AG-P

Funding acquisition: TCS, AAF, JEC

Project administration: TCS, AAF Supervision: TCS, AAF

Writing – original draft: TCS, AAF, PJB, JR

Writing – review & editing: TCS, AAF, PJB, JR, JEC

Funding

HHS | NIH | National Institute of Mental Health (NIMH) (R37 MH057068)

• Todd Charlton Sacktor

HHS | NIH | National Institute of Mental Health (NIMH) (R01 MH115304)

• Todd Charlton Sacktor

HHS | NIH | National Institute of Neurological Disorders and Stroke (NINDS) (R01 NS105472)

• Andre Fenton

HHS | NIH | National Institute of Mental Health (NIMH) (R01 MH132204)

• Andre Fenton

HHS | NIH | National Institute of Neurological Disorders and Stroke (NINDS) (R01 NS108190)

• Peter John Bergold

The Garry and Sarah S. Sklar Fund

• Panayiotis Tsokas