[1] 18w Postischemic fish oil treatment restores dendritic integrity and synaptic proteins levels after transient, global cerebral ischemia in rats

[1] 321w We previously found that fish oil (FO) facilitated memory recovery in the absence of pyramidal neuron rescue after transient, global cerebral ischemia (TGCI). Fish oil preserved the expression of microtubule-associated protein 2 (MAP-2), suggesting a relationship between dendritic plasticity and memory recovery that is mediated by FO after TGCI. The present study examined whether postischemic treatment with FO prevents ischemiainduced loss of dendritic processes in remaining pyramidal neurons. The effects of FO on neuroplasticity-related proteins were also examined after TGCI. Rats were subjected to TGCI (15 min, four-vessel occlusion model) and then received vehicle or FO (300 mg/kg docosahexaenoic acid) once daily for 7 days. The first dose was administered 4 h postischemia. Golgi-Cox staining was used to evaluate dentrict morphology in the pyramidal neurons of hippocampus (CA1 and CA3 subfields) and prefrontal cortex (PFC). Neuronal nuclei protein (NeuN), brain-derived neurotrophic factor (BDNF), growth-associated protein 43 (GAP-43), synaptophysin (SYP), and postsynaptic density protein 95 (PSD-95) levels were measured by Western blot in both structures. Fifteen minutes of TGCI reduced consistently the length of dendrites, number of dendritic branches and dendritic spine density (average of 25%, 43%, 32%, respectively) 7, 14, and 21 days postischemia, indicating that they did not recover spontaneously. This outcome of TGCI was reversed by FO treatment, an effect that was sustained even after treatment cessation. The NeuN and BDNF protein levels were reduced in both the hippocampus and PFC, which were recovered by FO treatment. GAP-43 protein levels decreased after ischemia in the PFC only, and this effect was also mitigated by FO. Neither SYP nor PSD-95 levels were altered by ischemia, but PDS-95 levels almost doubled after FO treatment in the ischemic group. These data support our hypothesis that synaptic plasticity at the level of dendrites may at least partially underlie the memory-protective effect of FO after TGCI and strengthen the possibility that FO has therapeutic potential for treating the sequelae of brain ischemia/ reperfusion injury.

[1] 86w Transient, global cerebral ischemia (TGCI) often results from reversible cardiac arrest. Patients who survive long-term after TGCI generally become functionally disabled because of a wide range of neuropsychological deficits (Anderson and Arciniegas, 2010). Although many compounds have been shown to exert neuroprotective effects in animal models of brain ischemia, these preclinical findings have not yet been translated into clinical practice. Aside from the many factors that contribute to this translational dilemma, the search for therapeutic substances that can effectively treat outcomes of cerebral ischemia is needed.

[2] 332w Using the rat four-vessel occlusion (4-VO) model of TGCI, we found that fish oil (FO) treatment that was initiated before (Fernandes et al., 2008;Bacarin et al., 2013Bacarin et al., , 2015) ) and after (Ferreira et al., 2014;Bacarin et al., 2016;Oliveira et al., 2017) TGCI prevented the loss of spatial retrograde memory that was formed and consolidated prior to ischemia (i.e., retrograde memory). Importantly, this memory-protective effect of FO persisted for up to 5 weeks after treatment cessation. In these studies, however, FO failed to prevent the loss of pyramidal neurons in the hippocampus and in the retrosplenial (RS) and parietal (PtA) cerebral cortices, clearly indicating a dissociation between functional recovery and histological neuroprotection, at least with regard to pyramidal neurons in these brain regions. We then asked how FO could facilitate memory recovery despite pyramidal cell loss in the hippocampus and cerebral cortex, since these brain structures are thought to play a major role in spatial memory, including that measured in the radial maze task (Haijima and Ichitani, 2008;Ramos, 2009). The CA1 and CA3 regions of the hippocampus principally comprise pyramidal cells, and the CA1 and CA3 regions have functional differences with regard to memory processes. For example, in the radial maze task, both CA1 and CA3 pyramidal neurons are required for the recall of remote, spatial memory, whereas granule cells in the dentate gyrus seems to be involved in the recall or relearning of more general aspects of the environment (Schlesiger et al., 2013). In another study, the CA3 region was shown to play a critical role in the temporal storage of long-term spatial working memory (Gilbert and Kesner, 2006). In a contextual fear-conditioning paradigm, both the acquisition and consolidation (or retrieval) of contextual memories are also differentially processed by the CA1 and CA3 subfields. The CA1 subfield appeared more critically involved in memory consolidation, whereas the CA3 subfield supports both the acquisition and consolidation of multisensory informations that build a unified representation of the context (Daumas et al., 2005).

[3] 107w Fish oil is a rich, natural source of docosahexaenoic acid (DHA), which is considered the most active omega-3 polyunsaturated fatty acid (ω-3 PUFA). The preventive, therapeutic value of ω-3 PUFA has been clinically and systematically investigated in the setting of chronic heart disease, but controversial results have been reported (Goel et al., 2018). In the setting of stroke, one study that investigated the effects of FO in patients 3 months after ischemic stroke did not observe any beneficial effects of FO on emotional end-points (Poppitt et al., 2009). The very late onset of treatment after stroke (i.e., > 3 months) may have contributed to these negative results.

[4] 235w Beyond its antioxidant, antiinflammatory, and antiapoptotic actions, DHA has also been shown to possess neurotrophic and neuroplastic properties (Cao et al., 2009;Su, 2010;Frautschy and Cole, 2010). We recently found that FO facilitated memory recovery and restored microtubule-associated protein 2 (MAP-2) immunoreactivity after TGCI, an effect that was stronger and longer lasting in the CA3 subfield of the hippocampus than in the CA1 subfield and cerebral cortex (Bacarin et al., 2016). Interestingly, FO restored both memory and MAP-2 immunoreactivity in the hippocampus when it was administered 4 h postischemia. None of these effects were observed, however, when FO treatment began 12 h postischemia. This temporal relationship between memory improvement and MAP-2 restoration suggests an association between the effects of FO on memory recovery (or preservation) and dendritic plasticity (Bacarin et al., 2016). MAP-2 is a cytoskeleton protein that is highly compartmentalized in dendrites where it plays a crucial role in maintenance of the structural integrity of dendrites, synapse formation, and neuroplasticity (Conde and Caceres, 2009). Dendritic trees of pyramidal neurons are vastly covered with small protrusions from the surface of dendrites, so-called dendritic spines, which may represent the most active site of neuroplasticity, including synaptogenesis and synaptic remodeling (Spruston, 2008). Neurons that survive an ischemic event can respond with ultrastructural changes, including dendritic restructuring, reactive synaptogenesis, and growth-promoting processes that, in turn, promote functional recovery even in the absence of neuronal rescue (García-Chávez et al., 2008).

[5] 148w In a continuation of our previous study, which reported the protective effects of FO on memory recovery and MAP-2 preservation after ischemia (Bacarin et al., 2016), in the present study we aimed to evaluate the effects of FO on dendrict morphology, including dendrict length, branching, and spine density in pyramidal neurons that survived ischemia in the hippocampus (CA1 and CA3 subfields) and prefrontal cortex (PFC). We also evaluated the effects of ischemia on the levels of some neuroplasticity-related proteins in both brain structures and the effects of FO thereon. Together with other extrahippocampal regions, the PFC is directly interconnected with the hippocampus, and they work together to play a key role in processing spatial memory formation, consolidation, and retrieval (Wirt and Hyman, 2017). Considering our previous findings, our hypothesis is that FO acts on neurons that survive ischemia and stimulates dendritic plasticity both in the hippocampus and PFC.

[1] 97w For each brain, five sections were selected based on overall good quality staining and tissue integrity. In each of these sections, one pyramidal neuron per hemisphere (i.e., 10 neurons/animal) was chosen to estimate the total number of dendritic spines (apical + basal) in the CA1 and CA3 subfields of the hippocampus and PFC. As schematized in Fig. 2, the number of dendritic spines was counted along a 50 μm linear length of an apical and basal (100 μm in total) second-order, oblique dendrite that branched from its parent, first-order dendrite (González-Burgos et al., 2007;García-Chávez et al., 2008).

[2] 256w A 100-μm ruler that was affixed to the eyepiece of the microscope (BX50, Minato-Ku, Japan) was used to determine the 50 μm linear length of the dendritic segment. The number of dendritic spines (100 x immersion objective) was counted manually in real time. In the hippocampus, apical and basal dendrites were assessed in the stratum radiatum and stratum oriens, respectively. In the PFC, area 1 of the cingulate cortex (Cg1) and prelimbic cortex (PrL) comprised the regions of interest as a whole, and dendritic spines were counted in pyramidal neurons from layer III or V. In a post hoc analysis, the total length of dentrites (apical + basal) and the total number of dendritic branches (apical + basal) were estimated. Neuron images were captured (BX 50 microscope, 20× objective, 3CCD Pro-series camera) and the dendritic arborization was traced by the 3D Simple Neurite Tracer FIJI software (3.1.3 version, http://imagej.net/Simple_Neurite_Tracer). In the brain sections that were used for analysis of dendritic spines, five neurons per animal were randomly selected for further analysis, using the following criteria: (i) neurons relatively isolated, (ii) consistent impregnation throughout the whole cell, (iii) well defined cell body, and (iv) well distinguished dendritic tree (Zhao et al., 2013;Fogaça et al., 2018). Because one or the other of these criteria could not be met during the post hoc analysis, the sample size available for estimating dendritic length and branching was reduced and variable in relation to that used for estimation of dendritic spines. The identity of the groups was not revealed during histological assessment.

[3] 209w Fig. 3 shows the effects of TGCI on dendritic morphology measured at different time points after TGCI. A global, main effect of surgery was revealed for all the three parametrs that were measured in CA1, CA3 and PFC regions, i.e., dendritic length (F 1, 24-26 = 56.64-82.37, p < 0.001), dendritic branches (F 1, 24-26 = 116.00-127.3, p < 0.001) and dendritic spines (F 1, 26 = 41.92-55.63, p < 0.001). Notably, the number of dendritic spines did not differ between the apical and basal dendrites. Considering the three time periods examined (7, 14, and 21 days), ischemia reduced the length of dendrites, the number of branches and the number of spines by 44%, 35% and 34% on average, respectively, compared to sham-operation (p < 0.0001-0.05). In overall, this deteriorating effect of ischemia on dendritic morphology of pyramidal neurons in CA1, CA3 and PFC did not differ as a function of postischemia survival time (F 2, 24-26 = 0.14-1.13, p > 0.05), except for a main effect of time on the parameter branching in the CA3 subfield (F 2, 24 = 7.70, p < 0.01). Based on these data, the intermediate, 14day postischemia period was chosen to test the effect of FO on the ischemia-induced loss of dendritic spines.

[4] 133w Fig. 4 shows the impact of ischemia on the dendritic morphology 14 days after ischemia, and the effects of FO thereon. On average, ischemia decreased the length of dendrites, the number of branches and the number of spines by 36%, 44% and 25%, respectively, considering the CA1, CA3 and PFC regions. In these regions, a main effect of surgery was found for all the three parameters (dendritic length: F 1, 17- 26 = 11,52-57,27, p < 0.001-0.01; dendritic branches: F 1,16- 22 = 9.04-61.97, p < 0.001-0.01; dendritic spines: F 1, 26 = 5.60-29.93, p < 0.001-0.05). This ischemia-induced changes on dendritic morphology was significant in CA1, CA3 and PFC (p < 0.001-0.01), compared to sham-operation. The two-way ANOVA revealed also a main effect of treatment for both dendritic length (F 1, 17-

[5] 296w Fig. 5 shows changes in protein levels following 14 days of ischemia/reperfusion or sham operation and the effects of FO thereon. The two-way ANOVA revealed a main effect of surgery condition on NeuN, BDNF, and PSD-95 protein levels in the hippocampus (F 1,18 = 3.69-11.91, p < 0.01-0.05) and PFC (F 1,18 = 9.61-23.35, p < 0.001-0.01). A main effect of treatment on NeuN, BDNF, and PSD-95 protein levels was found in both the hippocampus and PFC (F 1,18 = 4.35-27.16, p < 0.0001-0.05). A main effect of treatment on GAP-43 protein levels was found in the PFC (F 1,19 = 11.49, p < 0.01), with a surgery condition × treatment interaction for NeuN and PSD-95 protein levels in the hippocampus (F 1,18 = 4.67-5.13, p < 0.05) and BDNF and GAP-43 protein levels in the PFC (F 1,18/19 = 8.63-12.21, p < 0.01). Compared with sham-operated rats, the levels of NeuN and BDNF proteins decreased consistently in both the hippocampus (p < 0.01) and PFC in vehicle-treated animals (p < 0.001-0.05). Ischemia also reduced GAP-43 levels in the PFC (p < 0.001). Treatment with FO mitigated the effects of ischemia on the levels of NeuN, BDNF, and PSD- 95 in both the hippocampus and PFC (p < 0.001-0.05, vs. Veh). Fish oil treatment also restored the levels of GAP-43 protein, but this effect was not statistically significant (p = 0.058, vs. Veh). Compared with sham surgery, the level of PSD-95 in the hippocampus and PFC did not change after ischemia. PSD-95 levels significantly increased in both structures, however, in the ischemic group that was treated with FO (p < 0.05, vs. sham; p < 0.05, vs. Veh). The level of SYP did not significantly change after TGCI in either the vehicle-or FO-treated group (Fig. 6).

[1] 123w The present study demonstrates for the first time that postischemic FO treatment prevented the ischemia-induced changes in dendritic length, number of branches, and spine density in the hippocampal and cortical pyramidal neurons that survived ischemia. Also, FO mitigated the ischemia-induced decrease of NeuN, BDNF, and GAP-43 protein levels in both the hippocampus and PFC. Importantly, these neuromorphological and biochemical effects of FO occurred despite administration that began 4 h postischemia, indicating a clinically relevant time window of efficacy, which coincide with our previous behavioral studies with young (Bacarin et al., 2015) and middle-aged (Ferreira et al., 2014) rats. Moreover, these protective effects of FO on dendritic morphology and neuroplasticity-related proteins were sustained for at least 7 days after the cessation of FO treatment.

[2] 236w Dendrites, and specially their spines, are a major site of synaptic inputs to pyramidal neurons, which can respond with structural changes that, in turn, may compensate for neuronal death and promote functional recovery after brain ischemia. Importantly, dendritic and axonal remodeling can be stimulated by pharmacological interventions after brain damage (Benowitz and Carmichael, 2010). While the neurotrophic and/or neuroplastic properties of ω-3 PUFA have been well documented in the field of brain development (Cao et al., 2009;Su, 2010) and aging (Frautschy and Cole, 2010;Cutuli et al., 2014), we found no other study in the field of cerebral ischemia with which our present findings can be compared. Our data agree closely, however, with studies that evaluated the effects of FO or DHA after central or peripheral nerve lesions. Treatment with DHA was reported to facilitate the sprouting of uninjured corticospinal fibers and promote neurological recovery (Liu et al., 2017). In another study that administered FO prophylactically in mice with sciatic nerve injury, FO facilitated the recovery of myelinated fibers, increased GAP-43 protein levels, decreased markers of neuroinflammation in lesioned tissue, and improved neurological functions (Silva et al., 2017). These findings are consistent with the present data, which suggest that FO-mediated neuroplasticity may have contributed to memory recovery that was consistently observed in rats subjected to TGCI (Fernandes et al., 2008;Bacarin et al., 2013Bacarin et al., , 2015Bacarin et al., , 2016;;Ferreira et al., 2014;Oliveira et al., 2017).

[3] 175w In the present study, the dendritic morphology results complemented our previous data that showed that FO restored MAP-2 immunoreactivity after TGCI, an effect that was temporally associated with memory improvement (Bacarin et al., 2016). MAP-2 is highly concentrated in dendrite shafts, where it regulates dendritic development, differentiation, remodeling, and synaptic plasticity at dendritic spines (Conde and Caceres, 2009). Therefore, the protective effects of FO on MAP-2 (Bacarin et al., 2016) and dendritic morphology (present) may be closely related. Therefore, our data are consistent with observations that neurons that survive around or distant from the lesion respond with structural remodeling that compensates for those neurons that are lost, a response that can be stimulated if neurons are exposed to a more favorable extracellular environment or are stimulated with appropriate trophic agents (Benowitz and Carmichael, 2010). As mentioned above, the neurotrophic properties of DHA have been well documented (Cao et al., 2009;Su, 2010;Frautschy and Cole, 2010;Cutuli et al., 2014), including its stimulating effect on dendritic spine density (Sakamoto et al., 2007) and branching (Cutuli et al., 2014).

[4] 113w Notably, the postischemia, protective effect of FO on dendritic morphology paralleled its restorative or stimulating effect on the expression of BDNF, GAP-43, and PSD-95 proteins, suggesting a relationship between these proteins, dendritic plasticity, and memory recovery (previous studies) after TGCI. The role of DHA in learning and memory is emphasized by its multiple (i.e., pleiotropic) actions at the molecular, cellular, and integrative levels of function. DHA has been reported to increase the levels of more than a dozen proteins, including BDNF, GAP-43, and PSD-95, improve LTP, increase neurite outgrowth, increase dendritic spine density, increase synaptogenesis, and increase neurogenesis, which may work together for the development and maintenance of learning and memory (Su, 2010).

[5] 126w Favorable effects of BDNF on neuroplasticity and functional recovery after cerebral ischemia have been reported, reflected by prevention of the ischemia-induced deterioration of LTP, prevention of the disruption of spatial memory (Kiprianova et al., 1999), the stimulation of neuronal remodeling, and improvements in functional motor recovery (Schäbitz et al., 2004). Therefore, the restorative effect of FO on BDNF in the present study may have been important for the recovery (or preservation) of morphological integrity of dendrites and consequently, memory recovery that was reported previously (Ferreira et al., 2014;Bacarin et al., 2015). This is in line with the observation that both the dendritic spine density of cortical pyramidal neurons and cognitive improvement depend on the consistent expression of BDNF in the adult brain (Vigers et al., 2012).

[6] 241w GAP-43 is a major component of axons and presynaptic terminals, where it participates in neuronal differentiation, plasticity, and regeneration, and its expression increases during axon sprouting (Benowitz and Routtenberg, 1997). In the present study, GAP-43 levels decreased ∼50% in the PFC but appeared unchanged in the hippocampus after TGCI. Decrease in GAP-43 mRNA expression was reported also in the cerebral cortex around ischemic tissue (Zendedel et al., 2015). In the hippocampus, the expression of GAP-43 mRNA decreased significantly in the CA1 but not in the CA3 subfields after TGCI, indicating a region-specific effect of ischemia on the expression of GAP-43 in the hippocampus (Schmidt-Kastner et al., 1997). This regionspecific difference may have been masked in the present study because the whole hippocampus was examined, possibly explaining why GAP-43 levels appeared unchanged in the hippocampus. Only two previous studies investigated the effects of FO or DHA on GAP-43 expression. In primary culture of cortical neurons, DHA increased both GAP-43 immunoactivity, the number and length of neurites, and the content of phosphatidylserine and phosphatidylethanolamine, two important phospholipids that are required for neurite growth (Cao et al., 2005). Also, the treatment with a PUFA n3-based emulsion restored the expression of Tau and GAP-43, reduced neuroinflammation, and improved motor and sensorimotor performance after focal cerebral ischemia (Zendedel et al., 2015). Therefore, FO-mediated GAP-43 restoration and dendritic preservation appear to be related events that contribute to the ability of FO to facilitate memory recovery after TGCI.

[7] 333w PSD-95 is a scaffolding protein that is highly concentrated under the postsynaptic membrane, and it may control the interaction between postsynaptic membrane receptors, mainly the NMDA and AMPA glutamate receptors, and their downstream signaling pathways. Such interactions are essential for neurotransmission and provide a basis for neuroplasticity, including changes in dendritic spine density and maturation (Iasevoli et al., 2013). In the present study, PSD-95 levels in the hippocampus and PFC did not change 14 days after TGCI, which diverges partially from other studies. For example, PSD-95 levels decreased as early as 6 h after TGCI in gerbils and persisted for the next 5 days, an effect that occurred in the CA1 region but not CA2 or CA3 region of the hippocampus (Yan et al., 2013). In rats that survived for 24 h after TGCI, PSD-95 levels significantly decreased in the hippocampus but only slightly in the PFC. This effect accompanied an increase in the expression of NMDA receptor protein, suggesting that the excessive activation of NMDA receptors in the ischemic milieu may destabilize the anchoring of PSD-95 molecules in dendritic spines (Zaric et al., 2018). In line with this, the association between PSD-95 and NMDA receptors was disrupted 6 h after TGCI, an effect that was greater in the CA1 hippocampal subfield than in the CA3/dentate gyrus regions (Takagi et al., 2000). Together, these data suggest a region-specific and time-dependent response of PDS-95 to ischemia. In the present study, PSD-95/NMDA receptor destabilization may also have occurred in the hyperacute phase of ischemia, but it recovered spontaneously and gradually as the postischemic survival time increased. This supposed spontaneous recovery of PSD-95 levels may have been stimulated by FO, in which PSD-95 levels nearly doubled in the hippocampus and PFC in ischemic rats. A similar effect on PSD-95 recovery after ischemia/ reperfusion was observed with the neurosteroid dehydroepiandrosterone (Zaric et al., 2018). Together, these data suggest that DHA in FO interacts with surviving neurons in the ischemic milieu and stimulates the expression of PSD-95 protein.

[8] 283w Synaptophysin is another important marker of synaptic plasticity and synaptogenesis. It is highly expressed and accumulates during synapse formation in culture (Tarsa and Goda, 2002). In the present study, ischemia did not change the level of SYP significantly, although it was above and below the control levels in the hippocampus and PFC, respectively. In gerbils, SYP levels in the CA1 subfield of the hippocampus significantly decreased 2 and 4 days after TGCI, returned to the control levels 14 days postischemia, and then decreased again to below control levels 30 days postischemia (Ishimaru et al., 2001). Consistent with these data, SYP immunoreactivity increased in the CA1 subfield 7 days after TGCI in rats (Miyazawa et al., 1993). In other studies, however, the immunoreactivity of SYP and other synaptic proteins was preserved in the CA1 subfield of the hippocampus at different timepoints postischemia (Miyazawa et al., 1993). Despite differences in the magnitude of SYP levels, our findings agree with the aforementioned studies, and the slight increase and reduction of SYP levels in the hippocampus and PFC, respectively, suggest that this synaptic protein, similar to GAP-43 and PSD-95, responds to ischemia in a region-specific manner. To our knowledge, no previous study has investigated the effects of FO or DHA on the expression of SYP under normal or ischemia conditions, thus precluding comparisons with the present findings. However, DHA has been shown to stimulate the expression of presynaptic proteins (e.g., synapsin I and syntaxin-3), PSD-95, and cytoskeleton protein F-actin in the hippocampus in normal, adult gerbils (Sakamoto et al., 2007;Cansev and Wurtman, 2017). Further investigations are needed to understand the role of SYP in mediating the beneficial effects of FO or DHA under conditions of brain ischemia.

[9] 227w Finally, the levels of NeuN were decreased by TGCI and recovered by FO in both the hippocampus and PFC. NeuN protein is a marker of global neuronal viability (i.e., a universal, adult neuron-specific marker; Gusel'nikova and Korzhevskiy, 2015). Therefore, the restorative effect of FO on NeuN protein levels, measured by Western blot, appears to be paradoxical when compared with our previous studies where FO failed to prevent pyramidal cell death, evaluated by Nissl staining (Fernandes et al., 2008;Bacarin et al., 2013Bacarin et al., , 2015Bacarin et al., , 2016;;Ferreira et al., 2014) and in situ NeuN immunohistochemistry (Oliveira et al., 2017). Although we have no satisfactory explanation for these discrepant results, we can speculate, however, that the failure of FO to prevent pyramidal cell loss (assessed by either Nissl staining or in situ NeuN immunohistochemistry), in contrast to its effectiveness in restoring NeuN protein levels (assessed by Western blot), may reflect methodological limitations. Indeed, Nissl staining and in situ immunohistochemistry reveal intact-appearing pyramidal neurons in situ, whereas Western blot provides a panoramic view of regional levels of NeuN protein as it was measured in the hippocampus as a whole. Alternatively, the increase in NeuN protein levels in the Isch/FO group may reflect stimulation of the expression of NeuN protein by neurons that survived ischemia. If so, then the meaning of such an effect of FO is unknown.

[10] 73w In conclusion, the present findings supported the hypothesis that FO-mediated synaptic plasticity at the morphological (dendrite) and biochemical (protein) levels may be one mechanism that underlies the memory-protective effect of FO that has been consistently observed in our previous studies, despite the presence of extensive pyramidal cell death. Together with other behavioral, neurohistological, immunohistochemical, and biochemical findings, the present results suggest that FO may have therapeutic utility for the treatment of cerebral ischemia.

[1] 82w Throughout the experiments, male Wistar rats (270-300 g body weight) were housed under controlled temperature (22 °C ± 1 °C) on a 12 h/12 h light/dark cycle (lights on at 7:00 AM). Tap water and a commercial chow diet (Nutrilab-CR1, Nuvital Nutrients, Curitiba, PR, Brazil) were provided ad libitum. Before any experimental manipulations, the rats were acclimated to the laboratory vivarium for 1 week. The local Ethics Committee on the Use of Animals (CEUA) approved the experimental procedures (CEUA authorization no. 2879100816).

[2] 242w The 4-VO surgical procedure (Pulsinelli et al., 1982), with modifications, was used to induce TGCI. Briefly, under isoflurane/oxygen anesthesia (Isoforine, Cristália, São Paulo, Brazil), the vertebral arteries were permanently occluded by electrocoagulation (3-4 mA) at the level of the first cervical vertebrae. Afterward, a longitudinal incision was made in the ventral neck to expose the common carotid arteries, which were carefully separated from adjacent tissues and loosely tied with silk thread, the ends of which were externalized and attached to the back of the animal by an aneurism-like clip. The incision was closed, and the animal was allowed to recover from anesthesia for 4-5 h, after which the silk thread was detached and carefully tightened while the animal was conscious and spontaneously ventilating. As soon as the animal ceased any movement, it was positioned in the lateral decubitus position, and the tightened silk thread was again attached to the back (time zero of ischemia) for 15 min. During occlusion, loss of the righting reflex, unresponsiveness to gentle touch, mydriasis, and/or tonic extension of the paws were considered indicative of effective ischemia. If the animal recovered its righting reflex during the occlusion period, then it was eliminated from the study. One hour before and after occlusion, the animals were maintained in a warming box (37 °C ± 1 °C) to avoid eventual cerebral hypothermia (Seif el Nasr et al., 1992). Sham animals underwent the same surgical procedure, but the arteries were not occluded.

[3] 312w Fig. 1 illustrates the timelines of the morphometric and biochemical measurements relative to the period of TGCI and sham surgery and vehicle and FO treatment. In Experiment 1, the experimental groups consisted of rats that were subjected to sham surgery or TGCI (time 0). Both groups were treated with vehicle to achieve balance with Experiment 2. Seven, 14 and 21 days after ischemia, the dendritic length, number of dendritic branches, and number of dendritic spines in pyramidal neurons of the hippocampus (CA1 and CA3 subfields) and PFC were estimated. In Experiment 1, we sought to identify a postischemia time-point at which the magnitude of the ischemia-induced changes in dendrict morphology would be more intense or consistent. In Experiment 2, the following groups were tested: sham + vehicle (Sham/Veh), sham + FO (Sham/FO), ischemia + vehicle (Isch/Veh), and ischemia + FO (Isch/FO). In both Experiments 1 and 2, vehicle (1.0 ml/ kg extra virgin olive oil) or FO (equivalent to 300 mg/kg DHA) was administered orally by gavage. The first dose was administered 4 h postischemia and then repeated once daily for 7 days. Based on the results of Experiment 1, in Experiment 2 the time-point of 14 days postischemia was used to determine whether FO treatment prevents the ischemia-induced dendritic changes. As part of Experiment 2, separate groups of rats were formed to evaluate the effects of TGCI on the expression of neuroplasticity-related proteins and the effects of FO thereon. The FO formulation that was used herein was a standardized, high-grade formulation that contained 250 mg DHA, 50 mg EPA, and 1 mg vitamin E as an antioxidant (Omega-3 DHA 250, Biotik do Brasil Indústria e Comércio Ltda, São Paulo, Brazil). The total fatty acid content was determined elsewhere (Bacarin et al., 2015). Preparation of the FO solution from a 0.4 ml capsule was performed as described previously (Fernandes et al., 2008).

[4] 216w The Golgi-Cox staining procedure used here was based on previous studies (Gibb and Kolb, 1998;Zaqout and Kaindl, 2016), with modifications. Under deep anesthesia (Thiopentax, 75 mg/kg, i.p., Cristália, São Paulo, Brazil), the brains were fixed in situ by intracardiac perfusion with phosphate-buffered saline (PBS) at a flow rate of 22 ml/min for 15 min. The brains (with the exception of the olfactory bulbs and cerebellum) were then maintained in Golgi solution for 24 h at 37 °C (Gull et al., 2015) and then transferred to a new Golgi solution where they remained for 19 days in the dark. The brains were then immersed in a 30% sucrose/saline solution for at least 6 days until sectioning in a cryostat at -21 °C (Criocut 1800, Reichert-Jung, Heidelberg, Germany). Before the brains were transferred to the cryostat, they were rapidly immersed in liquid nitrogen. Starting from the frontal pole of the cerebral hemispheres, the first 13 coronal sections (100 μm thick each) were discarded. From this point onward (approximately 4.50 mm anterior to bregma), coronal sections (100 μm thick) that comprised the PFC (4.70 mm to 2.20 mm anterior to bregma) and hippocampus (3.60 mm to 4.30 mm posterior to bregma; Paxinos and Watson, 1998) were taken and kept in a humidified chamber for 24 h in the dark.

[5] 92w After washing in distilled water, the sections were collected on gelatinized slides (four to six sections/slide) and allowed to dry during 2 days at room temperature. The mounted tissue was dehydrated in successive alcohol baths of 50% (2x/5 min), 70% (2x/5 min), 90% (3x/ 5 min), and 100% (3x/5 min), followed by xylene baths (2x/10 min). The slides were then covered by coverslips fixed with permount and maintained in a dark enviorment for 15 days without any manipulation. where the number of dendritic spines was estimated along a 50 μm linear length.

[6] 365w Under deep anesthesia, the brain was freshly removed, and the whole hippocampus and PFC were dissected according to methods described elsewhere (Spijker, 2011), with modification. The tissue samples were lysed with lysis buffer (10% glycerol, 137 mM NaC, and 20 mM Tris HCl, pH 7.5) that contained protease and phosphatase inhibitors (Sigma-Aldrich, St. Louis, MO, USA) and centrifuged at 16,700 × g for 15 min at 4 °C. The supernatant was separated. Sodium dodecyl sulfate (SDS) buffer (125 mM Tris [pH 6.8], 2% SDS, 20% glycerol, 0.0001% bromophenol blue, and 10% β-mercaptoethanol) was then added to the supernatant, followed by the determination of total protein content using the Bio-Rad Lowry Protein Assay (Bio-Rad, Hercules, CA, USA). Homogenate samples were diluted to reach a 30 μg protein concentration and separated on 15% SDS-polyacrylamide gel electrophoresis (PAGE) gel (BDNF) or 12% SDS-PAGE gel (neuronal nuclei protein [NeuN], growth-associated protein 43 , synaptophysin [SYP], and postsynaptic density protein 95 [PSD-95]) using a total of four different blots to measure the different proteins. After protein transfer to a nitrocellulose membrane (Bio-Rad, Hercules, CA, USA), the membranes were incubated for 1.5 h in Tris buffer at room temperature, blocked with 5% skim milk powder in Tris-buffered saline (TBS), and washed three times (10 min each) in TBS-T. Afterward, the membranes were incubated overnight for 16 h at 4 °C with the following primary antibodies: rabbit anti-NeuN (1:1000, Abcam), rabbit anti-BDNF (1:400, Santa Cruz Biotechnology), mouse anti-SYP (1:1500, Abcam), mouse anti-PSD-95 (1:1500, QED Bioscience), and mouse anti-GAP-43 (1:1500, Sigma-Aldrich). Brain tissue homogenates in sample buffer were separated on 15% SDS-PAGE gel (BDNF) or 12% SDS-PAGE gel (NeuN, GAP-43, SYP, and PSD-95) using a total of four different blots to measure the different proteins. The blots were stripped to protein control with glyceraldehyde-3-phosphate dehydrogenase (GAPDH). After a washing step with TBS, the membranes were incubated for 1 h with secondary antibodies (goat anti-rabbit [1:1000] or goat antimouse [1:2500]; Abcam, Cambridge, MA, USA) and developed using ECLplus (Invitrogen, Carlsbad, CA, USA). The bands were visualized using the ChemiDoc Imaging System (Bio-Rad, Hercules, CA, USA). The integrated optical densities (IOD) of the specific bands were quantified using ImageJ software and normalized to GAPDH levels.

[7] 78w The data were first analyzed for assumptions of a normal Gaussian distribution (Omnibus D'Agostino and Pearson test) and homoscedasticity (Bartlett's test and Brown-Forsythe test). Two-way analysis of variance (ANOVA), followed by Sidak´s or Tukey's multiple-comparison post hoc test, was used for between-group comparisons (GraphPad Prism 6.0 software), with surgery condition (sham surgery and ischemia) and treatment (vehicle and FO) as factors. The data are expressed as mean ± SEM. Values of p < 0.05 were considered statistically significant.

[1] 50w The data, their analyses and interpretation, and the conduct of the research were under the full responsibility of the principal (corresponding) author. The sponsors have no role in the study design; the collection, analysis, and interpretation of data; writing of the report; and decision to submit the article for publication.