|Home | About | Journals | Submit | Contact Us | Français|
Estrogen, in addition to its genomic effects, triggers rapid synaptic changes in hippocampus and cortex. Here we summarize evidence that the acute actions of the steroid arise from actin signaling cascades centrally involved in long-term potentiation (LTP). A 10-min infusion of E2 reversibly increased fast EPSPs and promoted theta burst-induced LTP within adult hippocampal slices. The latter effect reflected a lowered threshold and an elevated ceiling for the potentiation effect. E2’s actions on transmission and plasticity were completely blocked by latrunculin, a toxin that prevents actin polymerization. E2 also caused a reversible increase in spine concentrations of filamentous (F-) actin and markedly enhanced polymerization caused by theta burst stimulation (TBS). Estrogen activated the small GTPase RhoA, but not the related GTPase Rac, and phosphorylated (inactivated) synaptic cofilin, an actin severing protein targeted by RhoA. An inhibitor of RhoA kinase (ROCK) thoroughly suppressed the synaptic effects of E2. Collectively, these results indicate that E2 engages a RhoA >ROCK> cofilin> actin pathway also used by brain-derived neurotrophic factor and adenosine, and therefore belongs to a family of ‘synaptic modulators’ that regulate plasticity. Finally, we describe evidence that the acute signaling cascade is critical to the depression of LTP produced by ovariectomy.
Several groups have shown that brief applications of estrogen rapidly increase baseline synaptic transmission, alter synaptic connectivity, and enhance LTP (1–5). These short latency events raise the possibility that estrogen, in addition to its classical genomic actions, is a member of a family of what might be called ‘synaptic modulators’. We define this family as a group of molecules that are released by particular patterns of afferent activity and bind to synaptic receptors but have no direct effects on membrane voltage. They instead act in concert on actin signaling cascades, modify the subsynaptic cytoskeleton, and thereby regulate synaptic plasticity. That estrogen is synthesized in hippocampal synapses, as demonstrated recently (6, 7), and has synaptic receptors (8–10) is consonant with the idea that the steroid is a member of the above-defined family. Accordingly, we tested if E2 acts locally on actin signaling cascades and the spine cytoskeleton in adult hippocampal slices in a manner comparable to that for two established modulators.
Adenosine and brain-derived neurotrophic factor (BDNF) are both released by repetitive afferent activity (11–13), bind to synaptic receptors (A1/A2, and TrkB, respectively), and potently influence the actin polymerization that occurs within spines shortly after induction of LTP (14–16). The two modulators are antagonistic in that adenosine blocks filament assembly, and suppresses LTP consolidation, while BDNF enhances both polymerization and potentiation (15, 17). These actions appear to arise from opposing effects on an actin-signaling cascade consisting of the small GTPase RhoA, its effector ROCK (RhoA kinase), and the actin regulatory protein cofilin (see below). Given these points, the question for our studies was whether estrogen exerts its acute synaptic effects by stimulating LTP-related actin signaling in adult hippocampal slices.
In agreement with previous reports (1–4), we found that fast EPSPs undergo a modest, rapid, and fully reversible increase during 10-minute infusions of 17β-estradiol (E2); comparable effects were obtained with WAY200070 (WAY), an estrogen receptor beta (Erβ) agonist, while a selective estrogen receptor alpha agonist (PPT) had no effect. E2-induced increases in AMPA receptor gated currents were not accompanied by changes in GABA- or NMDA- receptor mediated responses.
Past studies have shown that E2 promotes LTP (1–3), but did not distinguish between an effect on threshold as opposed to an increase in the magnitude of the potentiation effect. We found that subthreshold levels of stimulation (2–3 theta bursts) delivered in the presence of E2 or WAY produced a significant degree of LTP. Suprathreshold stimulation (5 or 10 bursts) generated a percent potentiation that was well above the normal maximum. The positive effects of E2 and WAY appeared to occur at some stage after LTP induction because the compounds did not affect the area, or NMDA receptor dependent component, of theta burst responses.
Infusions of latrunculin A, a toxin that blocks the addition of actin monomers to growing filaments, eliminated the effects of E2 and WAY on field EPSPs and LTP. These results strongly suggest that E2 triggers the same actin assembly process used in the production of LTP (14, 16, 18–20). This was tested by topically applying fluorescence-tagged phalloidin, a toxin that selectively binds to F-actin, to slices immediately after a 20 min treatment of E2 (14, 21). The number of spines containing dense F-actin in area CA1 was significantly elevated by E2, and this effect was completely blocked by pretreatment with latrunculin A. As was the case for EPSPs, the increases in actin polymerization elicited by E2 were eliminated by a sixty-minute washout of the hormone.
We next asked if E2 enhances theta burst-induced actin polymerization. Three theta bursts did not increase the number of F-actin rich spines above control values in vehicle-treated slices collected at 60 minutes post stimulation. However, the same stimulation was effective when applied in the transient presence of E2, indicating that the hormone lowers the threshold for the stable increases in actin polymerization produced by TBS. Work on how E2 promotes actin polymerization in dendritic spines began with cofilin, a constitutively active protein that disassembles actin filaments. We found that a brief infusion of E2, using the same concentration that significantly increased EPSP responses in area CA1, substantially increased phosphorylated (p)-cofilin in hippocampal slices. This effect provides a likely explanation for E2’s influence on actin polymerization because phosphorylation inactivates cofilin, and thereby opens the way to actin filament assembly.
The prominent role of the Rho family in regulating cofilin activity in many types of cells led us to test if the small GTPases mediate the effects of E2 (22). Pull down assays indicated that E2 markedly increases the activity of RhoA but not that of two other members of the Rho family (Rac and Cdc42). The RhoA effects were rapid and reversed fully in slices given a 60-minute washout. RhoA typically exerts its downstream effects by activating ROCK (23) and past studies have shown that ROCK inhibitors block cofilin phosphorylation in slices (16). We found that the inhibitors also completely blocked E2’s effects on synaptic transmission without affecting baseline EPSPs.
Collectively, the above results point to the conclusion that the acute effects of estrogen in adult hippocampal slices are due to the activation of a RhoA> ROCK> LIM kinase> cofilin> actin assembly pathway. This cascade has been previously implicated in the production of stable LTP (16), a result suggesting that estrogen’s baseline effects represent a weak form of the potentiation effect. Why then would the hormone’s actions reverse so readily upon washout? The answer may lie in the failure (see above) of E2 to engage a Rac> PAK (p21-activated kinase) sequence suggested by recent studies to stabilize both newly formed filaments and newly induced LTP (16).
It is noteworthy that the proposed mode of action for estrogen overlaps extensively with that for adenosine. The purine depresses RhoA activity while having only minor influence on Cdc42 and Rac; as expected from this, it blocks TBS-induced cofilin phosphorylation, actin polymerization, and LTP consolidation (16). Repetitive afferent activity triggers the release of adenosine and it is possible that also holds for estrogen, particularly given recent evidence that the steroid is locally synthesized within hippocampal synapses (6, 7). If so, then adenosine and estrogen would constitute releasable factors that act in opposing fashions on a signaling cascade that controls the subsynaptic cytoskeleton and the plasticity it generates.
Estrogen’s actions are comparable to those of BDNF with regard to the RhoA pathway, which suggests that the two molecules may be additive or even synergistic with regard to plasticity. Unlike estrogen, BDNF also engages the Rac> PAK pathway and thus is likely to act on additional stages in the production of stable synaptic modifications. In any event, the picture that emerges is one in which synapses possess a family of modulators targeted for the cytoskeleton (and thus enduring plasticity) rather than seconds-long adjustments to transmission (Figure 1). Such an arrangement could add new levels of regulation to the process of encoding information, particularly if the various factors have different release requirements (assuming that estrogen is released in an activity-dependent manner). Different patterns of afferent input could, for example, result in EPSP potentiation of different strength and duration depending on the particular combination of modulators they present to the synapse. More generally, a family of diverse modulators might serve to connect memory encoding to broad biological and psychological domains. Adenosine concentrations vary with the sleep-wakefulness cycle (24) and it has been proposed that fluctuating levels of Gonadotrophin Releasing Hormone regulate estrogen synthesis within hippocampus (25). Along the same lines, BDNF production is readily influenced by behavior (26, 27). The ‘family of synaptic modulators’ hypothesis thus suggests that learning-related plasticity is ‘tuned’ by recent experience, daily cycles, and hormonal rhythms.
It follows from the above arguments that age-related changes in estrogen production and brain adenosine clearance will have important effects on synaptic plasticity, a point that has evident therapeutic implications. Past studies have linked LTP deficits in middle-aged rats to a deterioration in adenosine processing (28) and also showed that the potentiation effect is negatively affected by ovariectomy (29–31). However, in the latter case it is unclear if the impairments are due to the loss of genomic signaling as opposed to a depression of the acute mechanisms described here. Accordingly, we tested if long-term ovariectomy, with and without hormone replacement, affects the actin signaling cascades initiated by acute estrogen. TBS in slices from ovariectomized (OVX) rats without replacement failed to trigger actin polymerization or stable LTP; RhoA levels were also depressed in these animals relative to those found in the replacement group. Surprisingly, brief infusions of E2 produced the same reversible increases in F-actin and baseline transmission in OVX (no replacement) cases as in young adult slices. Consonant with these observations, the infusions also rescued LTP.
Depressed estrogen levels are sometimes accompanied by memory and cognitive problems in humans (32–34). The work described here raises the novel possibility of treating such impairments by manipulating other members of the synaptic modulator family. BDNF is of particular interest in this regard because of work showing that elevated concentrations of the neurotrophin reverse deficits in actin signaling and LTP consolidation in rodent models of aging (35) and early stage Huntington’s Disease (36). We are currently exploring the possibility that similar treatments can offset the plasticity deficits found in OVX rats.
These studies were supported by the National Institute of Neurological Disease and Stroke grant NS045260 to GL and CMG and NS051923 to GL. CSR was supported by NS045540 and LYC by the National Institute of Mental Health fellowship MH083396.
Conflicts of Interest
No potential conflicts of interest to disclose.