Caffeic acid recovers ischemia-induced synaptic dysfunction without direct effects on excitatory synaptic transmission and plasticity in mouse hippocampal slices

Caffeic acid is a polyphenolic compound present in a vast array of dietary components. We previously showed that caffeic acid reduces the burden of brain ischemia joining evidence by others that it can attenuate different brain diseases. However


Introduction
Caffeic acid is a polyphenolic compound present in a vast array of dietary components [1]. It is the main hydroxycinnamic acid found in human diet obtained from natural sources like olives, berries, fruits like prunes, potatoes, carrots, propolis, and predominantly coffee [1]. The interest in caffeic acid stems from numerous studies describing its impact on human health with effects ranging from the control of metabolism to cancer [2]. Caffeic acid also affords benefits to brain functions, such as amelioration of memory [3], namely in models of Alzheimer's disease [4], as well as decreasing the extent of brain damage in models of cerebral ischemia [5,6].
Several studies have documented different mechanisms of action of caffeic acid, mainly as a direct antioxidant (e.g. [7]) or through either Nrf2-mediated (e.g. [8]) or prolyl hydroxylase-2-mediated pathways [9], or as an inhibitor of DNA methylation [10] or apoptotic pathways [11]. Caffeic acid is rapidly absorbed, reaching maximum plasma levels within 1 h after consumption and easily permeating the brain [12]. Accordingly, the consumption of caffeic acid can affect brain function [13] and behavior [14]. Strikingly, compared with other polyphenols, caffeic acid has a broader profile of neuroprotection [15] and targets particular molecular pathways to afford neuronal protection [16]. This prompts caffeic acid as a promising candidate for combination therapies [17] and as a novel scaffold for the development of new strategies to Abbreviations: ACSF, artificial cerebrospinal fluid; caffeic acid, 3,4-dihydroxybenzeneacrylic acid; fEPSP, field excitatory post-synaptic potential; HFS, high frequency stimulation; LTP, long-term potentiation; LFS, low frequency stimulation; OGD, oxygen-glucose deprivation. manage brain diseases [18]. Such aims require clarifying if caffeic acid might affect information processing in neuronal networks, which is currently unknown. Additionally, it is also unknown if caffeic acid might protect against synaptic dysfunction, which is present at the onset of different brain diseases [19], mainly affecting excitatory synapses [20,21]. Thus, we now tested if caffeic acid affected excitatory synaptic transmission and plasticity in neuronal networks of hippocampal slices. Since the hippocampus is involved in the processing of spatial memory [22] and is particular susceptible to ischemia [23], we also tested if caffeic acid could prevent the disruption of synaptic transmission caused by an in vitro ischemia model and its consequences on aberrant synaptic plasticity processes, which are considered the neurophysiological basis of memory [24].

Animals
Male C57 black 6 (C57Bl/6) mice with 9 to 12 weeks-old were obtained from Charles River laboratories (Barcelona, Spain). Animals were kept under a 12-h light/dark cycle under controlled temperature (23 ± 2 • C) with free access to food and water.

Electrophysiological recordings
Mice were killed by decapitation after anesthesia under a halothane atmosphere. The brain was removed and placed in ice-cold, oxygenated (95% O 2 , 5% CO 2 ) artificial cerebrospinal fuid (ACSF; in mM: 124.0 NaCl, 3.0 KCl, 1.25 Na 2 HPO 4 , 26.0 NaHCO 3 , 2.0 CaCl 2 , 1.0 MgCl 2 , 10.0 glucose). Slices (400 μm-thick) from the dorsal hippocampus were prepared and allowed to recover for at least 1 h in gassed ACSF at 32 • C, before being transferred to a submerged recording chamber (1 mL capacity) and continuously superfused at a constant rate of 3 mL/min with gassed aCSF kept at 30.5 • C.
Extracellular field recordings were carried out as previously described [25], with stimulation of Schaffer collaterals through a bipolar concentric electrode positioned in the stratum radiatum of the CA1 region to trigger field excitatory postsynaptic potentials (fEPSPs) recorded extracellularly using micropipettes filled with 4 M NaCl (resistance of 1-2 MΩ) placed in the proximal stratum radiatum. Stimulation was delivered with a Grass S44 stimulator (Grass Technologies, RI, USA), every 20 s with rectangular pulses of 0.1 ms. After amplification (ISO-80, World Precision Instruments), the recordings were digitized (Pico ADC-42, Pico Technologies Ltd.), averaged in groups of 3, and analyzed using the WinLTP software. The intensity of stimulation was chosen between 50 and 60% of maximal fEPSP slope, selected based on a previously carried input/output curves. Alterations of synaptic transmission were quantified as the percentage alteration of the fEPSP slope taken from 15 to 20 min after beginning the application of tested drugs in relation to the fEPSP slope during the 5 min that preceded drug application.
The paired-pulse ratio was investigated by applying two pulses with a 25 and 50 ms inter-pulse interval every 20 s. Long-term potentiation (LTP) was induced by high-frequency tetanus (HFS, 1 s train of 100 Hz pulses). We next tested depotentiation, which was induced with a train of 1 Hz during 15 min. We use the term depotentiation rather than longterm depression since the lower frequency train was applied in slices where LTP was previously induced. The amplitude of synaptic plasticity was quantified as the percentage change between the average slope of the fEPSPs between 55 and 60 min after the induction of synaptic plasticity in relation to the average slope of the fEPSPs during the 5 min that preceded the induction of synaptic plasticity. The effect of drugs on synaptic plasticity was assessed by comparing the magnitude of LTP or of depotentiation in the absence and presence of caffeic acid in experiments carried out in different slices from the same animal.

Oxygen/glucose deprivation
A brief oxygen/glucose deprivation (OGD) applied directly to hippocampal slices was used an in vitro ischemia model, as previously described [26]. OGD, achieved by replacing the oxygenated ACSF with an ACSF with 7 mM sucrose/3 mM glucose instead of 10 mM glucose and saturated with 95 % N 2 /5% CO 2 , was applied during 7 min, to allow a partial recovery of synaptic transmission on re-oxygenation, reaching values < 25% of the initial fEPSP magnitude in control slices. Shorter periods of OGD allowed a complete recovery of fEPSP on reoxygenation, whereas longer ODG periods did not allow fEPSP recovery. Caffeic acid was only added on re-oxygenation onwards, i.e. it was absent prior to and during the OGD challenge. The impact of caffeic acid on the recovery of synaptic transmission on re-oxygenation was quantified 20-25 min after re-oxygenation. When testing the impact of OGD on synaptic plasticity, the fEPSP slope was adjusted after re-oxygenation to circa 50% of the initial fEPSP slope before applying the LTP-induction protocol first and subsequently depotentiation, as described above.

Statistical analysis
Values are presented as mean ± S.E.M. with the number of determinations (n, i.e. slices from different mice). Data were screened with the Grubbs test to identify putative outliers. We did not assume or test for normality and only applied non-parametric statistics using GraphPad Prism software (CA, USA). Alterations compared to baseline were estimated with a Wilcoxon Signed Rank test and the comparison of two experimental conditions was performed using a Mann Whitney test. Otherwise, statistical analysis was performed by one-way analysis of variance (ANOVA), followed by a Tukey's post hoc test. Statistical significance was considered at p < 0.05.

Effect of caffeic acid on synaptic transmission and plasticity
We first studied transmission and plasticity in mouse hippocampal slices to test for putative direct effects of caffeic acid on synaptic function. As shown in Fig. 1A, caffeic acid (1 or 10 µM) did not affect basal excitatory synaptic transmission (− 0.62 ± 1.01% and 1.00 ± 2.03% modification of fEPSP slope; p = 0.625 and p = 0.999 versus control, n = 4-5, Wilcoxon Signed Rank test). Caffeic acid (1 or 10 µM) was also devoid of effects on the paired-pulse facilitation ratio, with an interpulse interval of 25 ms or 50 ms, indicating a lack of direct presynaptic effects (Fig. 1B).
Additionally, caffeic acid was devoid of effects on synaptic plasticity under physiological conditions. Thus, caffeic acid (1 or 10 µM) did not affect the magnitude of long-term potentiation, LTP (50.40 ± 7.06% over baseline without drugs, 42.76 ± 4.02% over baseline with 1 µM caffeic acid and 49.58 ± 6.67% over baseline with 10 µM caffeic acid; n = 5, p = 0.631 versus control, one-way ANOVA followed by Tukey's post hoc test) (Fig. 1C,D). Caffeic acid also failed to affect depotentiation, which was tested by applying a low frequency train (1 Hz for 15 min) after the stabilization of LTP. As shown in Fig. 1E,F, caffeic acid (1 or 10 µM) did not affect the magnitude of depotentiation (25.07 ± 2.56% below baseline without drugs, 28.33 ± 7.07% below baseline with 1 µM caffeic acid and 22.20 ± 4.05% below baseline with 10 µM caffeic acid; n = 5, p = 0.688 versus control, one-way ANOVA followed by Tukey's post hoc test). Overall, these findings indicate a lack of direct effects of caffeic acid on synaptic function under physiological conditions.

Effect of caffeic acid on ischemia-induced depression of synaptic transmission
We then tested if caffeic acid could prevent the deterioration of synaptic function. We resorted to an in vitro model of ischemia-induced depression of synaptic transmission, modelled by exposing hippocampal slices to oxygen and glucose deprivation (OGD) [26]. Caffeic acid was applied at the onset of re-oxygenation since we have previously shown that caffeic acid prevents ischemia-induced brain damage when applied after the onset of ischemia [5]. As shown in Fig. 2A, an OGD challenge during 7 min abolished synaptic transmission, which recovered to 18.46 ± 3.46% of pre-OGD values (n = 7) within 20-25 min of re-oxygenation in control slices. The presence of caffeic acid (10 µM) at the onset of reoxygenation increased the recovery of synaptic transmission upon reoxygenation to 64.30 ± 13.84% of pre-ischemic values (n = 7, p = 0.026 versus control, Mann Whitney test) ( Fig. 2A, B).
The protective effect of caffeic acid on OGD-induced depression of synaptic function was re-enforced by the ability of hippocampal synapses to implement patterns of synaptic plasticity in slices treated with caffeic acid after OGD. Thus, as shown in Fig. 2C, D, a high-frequency stimulation (100 pulses at 100 Hz) failed to trigger a consistent LTP in control slices subject to OGD (1.33 ± 2.04% of baseline value, n = 5, p = 0.813 versus baseline, Wilcoxon Signed Rank test). In contrast, slices treated with 10 µM caffeic acid from re-oxygenation onwards after OGD, displayed a robust LTP (42.12 ± 6.64% of baseline value, n = 5, p = 0.008 versus control, Mann Whitney test) (Fig. 2C, D). Furthermore, control slices subject to a depotentiation protocol (following LTP) after OGD also failed to display consistent changes of synaptic efficiency (-4.87 ± 5.73% of baseline value, n = 5, p = 0.625 versus baseline, Wilcoxon Signed Rank test), whereas slices treated with 10 µM caffeic acid displayed a robust alteration of synaptic efficiency upon applying the prolonged low frequency train (23.45 ± 2.91% of baseline value, n = 5, p = 0.008 versus control, Mann Whitney test) (Fig. 2C, D).
These findings indicate that caffeic acid displays a selective ability to recover synaptic deterioration in this in vitro model of ischemia.

Discussion
The present study shows that caffeic acid can prevent ischemiainduced dysfunction of synaptic function without directly affecting synaptic transmission and plasticity in hippocampal slices. This selective ability of caffeic acid to prevent synaptic dysfunction rather than synaptic function under physiological conditions prompts the conclusion that caffeic acid may be considered as a selective neuroprotectant without global side effects on synaptic function. In view of the robust neuroprotection afforded by caffeic acid [2], broader than that afforded by other polyphenols [15], this bolsters the interest in exploiting caffeic acid as a promising template to develop novel selective neuroprotective drugs. Furthermore, the recognition that caffeic acid may target multiple pathways to control cellular damage (e.g. [9,15,16]) paves the way to identify critical cellular pathways involved in the selective control of synaptic dysfunction, one of the earliest features associated with brain diseases [19], to develop novel neuroprotective strategies.
Our investigation of the effects of caffeic acid in a concentration range achieved after the consumption of nutrients rich in caffeic acid such as coffee [12,27], showed an absence of effects on basal excitatory synaptic transmission, as previously described for other polyphenols [26]. Consistently, caffeic acid did not modify short term plasticity under paired-pulse facilitation protocols, which are indicative of putative presynaptic site of action of drugs affecting excitatory synaptic transmission [28]. Furthermore, there were also no significant alterations of long-term plasticity processes such as LTP or depotentiation in these excitatory synapses, which involve different molecular pathways argued to be potentially altered by other polyphenols (e.g. [29]). Overall, these findings allow concluding that micromolar concentrations of caffeic acid do not modify synaptic function under physiological conditions.
In sharp contrast, we observed that the addition of caffeic acid after an ischemic-like insult to hippocampal slices was able to attenuate synaptic dysfunction. This is in agreement with the previously observed ability of caffeic acid to prevent the extent of ischemic damage, when administered to rats after the ischemic insult [5]. Thus, caffeic acid can target molecular pathways able to prevent ischemic-induced synaptic dysfunction that are not involved in the control of synaptic function in physiological conditions. Since synaptic dysfunction is an early consequence of ischemic insults (reviewed in [30]), namely in the hippocampus (e.g. [31]), the identification of these molecular pathways may unveil critical pathways selectively involved in the control of early synaptic dysfunction following brain ischemia. One group of candidates could be related to the control of intracellular metabolism: it critically sustains synaptic function [32] and undergoes remodeling upon brain ischemia [33], while it is expected to be controlled by caffeic acid, based on its antioxidant and mitochondria-preventive effects [2]. This ability of caffeic acid to attenuate synaptic dysfunction may also result from indirect effects of caffeic acid on glial cells supporting synaptic transmission rather than from a direct effect on synapses. Glial cells, both astrocytes and microglia, are affected by OGD insults in hippocampal slices (e.g. [34,35]) and tinkering with glial pathways affects OGDinduced synaptic dysfunction (e.g. [36,37]). This links to the ability of caffeic acid to directly control glia cells [38,39]. Additionally, gliarelated pathways of neuroinflammation should also be considered as another possible mechanism since they are bolstered upon ischemia contributing for synaptic damage [40] and caffeic acid can attenuate neuroinflammation [3,39,41]. These hypotheses that caffeic acid might affect either metabolic or antioxidant processes in astrocytes [38] or neuro-inflammatory-like reactivity in microglia [39,41] to indirectly control synaptic dysfunction upon hypoxia are in accordance with the ability of caffeic acid to bolster antioxidant and inflammatory processes related to glial cells in in vivo models of ischemia [5,6]. Interestingly, the synaptic effects of caffeic acid are different from the effects of caffeine acting through adenosine receptors [18], suggesting that the health benefits of coffee intake might result from a synergic action of caffeine and caffeic acid, both abundant in coffee [42].
The future identification of the mechanisms operated by caffeic acid to afford synaptoprotection may provide clues on the mechanisms operated by caffeic acid to attenuate the burden and evolution of neurodegenerative diseases that also display synaptotoxicity features at their onset [20] and may pave the way for the development of novel neuroprotective strategies devoid of side effects in non-afflicted neuronal circuits, as now documented by the absence of effects of caffeic acid on synaptic function under physiological conditions.

Conclusion
It is concluded that, in physiological situations, caffeic acid does not alter synaptic transmission and plasticity whereas after an ischemic event caffeic acid was able to protect against synaptic dysfunction. The mechanism underlying the protective effects of caffeic acid may be related to its antioxidant/metabolic effects on neurons and/or glia cells, which should be elucidated in future studies.

Ethics statement
Animal experiments were performed following the ARRIVE and European Union guidelines (EU Directive 2010/63) and approved by the Ethical Committee of the Center for Neuroscience and Cell Biology of the University of Coimbra (ORBEA 138-2016/15072016). All efforts were made to reduce the number of animals used and to minimize their stress and discomfort.  Caffeic acid (10 µM) accelerated and increased the extent of recovery of both hippocampal synaptic transmission and synaptic plasticity after exposure to oxygen-glucose deprivation (OGD), modelling ischemia. (A) A period of 7 min of OGD (red bar) caused a profound depression of hippocampal synaptic transmission, measured as the slope of field excitatory postsynaptic potentials (fEPSP) recorded extracellularly in the stratum radiatum of the CA1 area upon stimulation of the afferent Schaffer collaterals. (A) Upon glucose and oxygen readmission (re-oxygenation), there was a discrete recovery of synaptic transmission in control conditions (no added drugs, grey symbols), which was more robust in the presence of caffeic acid (10 µM; filled blue circles), added when indicated by the arrow. (B) Quantification of fEPSP slopes between 20 and 25 min of re-oxygenation after OGD. (C) The average time course experiments of LTP induction with a high-frequency stimulation train (HFS: 100 pulses delivered at 100 Hz) in hippocampal slices subjected to OGD followed by re-oxygenation recovery, revealed that LTP magnitude was discrete in slices not exposed to any drug (control, open grey circles) and was more evident in slices treated with caffeic acid (10 µM) during re-oxygenation (D). (E) The subsequent application of a low-frequency stimulation train (LFS: 15 min at 1 Hz) was also devoid of effects in slices not exposed to any drug (grey symbols) and triggered a potentiation in slices treated with caffeic acid (10 µM) during re-oxygenation (F). Data are mean ± S.E.M. of 5-7 experiments; *p < 0.05 and **p < 0.01, Mann Whitney test.