The first neural processes that appear in these paradigms, i.e., recognition of threats and creating fear responses, mainly involve the hippocampus, amygdala, and medial prefrontal cortex–areas that seem to be enveloped by orexinergic neurons. The lateral hypothalamus/perifornical area directly innervates hippocampal formation, with CA1, CA2, and the dentate gyrus (DG) expressing mainly OX1R, CA3 mainly expressing OX2R (Trivedi et al., 1998; Lu et al., 2000; Hervieu et al., 2001; Marcus et al., 2001; Akbari et al., 2011), and the basolateral amygdala with orexinergic fibers acting through both OX1 and OX2 receptors which, in turn, modulate central nucleus output (Bisetti et al., 2006).

In addition to its direct innervation, the LH/PFA region sends orexinergic neurons to the locus coeruleus, a nucleus involved in stress and fear responses. Activation of orexin neurons in the LC has been observed in response to aversive stimuli, such as shock, and this activation, in turn, affects the amygdala via LC-noradrenaline fibers (Chen and Sara, 2007; Sara, 2009; Soya et al., 2017).

The hippocampus plays a role in integrating contextual cues and receives input from the collaterals of sensory pathways through the ventral tegmental area (VTA). Meanwhile, the amygdala processes early aversive memory traces, and the medial prefrontal cortex facilitates memory encoding (Izquierdo et al., 2016). The wide distribution of orexins in the brain suggests that they may contribute to associative learning and, in collaboration with other neurotransmitters, participate in the selection of new memory traces for further consolidation and the formation of stable engrams.

To further support the involvement of orexin receptors in fear conditioning, experiments utilizing intracerebroventricular infusion of an OX1R antagonist called SB-334867 have shown that blocking these receptors immediately before training impairs fear response. This finding was reported by Sears et al. (2013) in the context of Pavlovian fear conditioning. The same authors showed that blocking OX2R with specific antagonist TCS-OX2-29 before training did not affect learning. In another study, an improvement in encoding was observed after OX-B ICV infusion before training, which was completely abolished by ICV infusion of the orexin receptor blocker EMPA (Palotai et al., 2014). Therefore, it might be noted that the elevated concentration of OX-B can facilitate memory encoding mainly through OX2 receptors, but it is not sufficient to sustain even a base level of learning in the case of an OX1 receptor block. On a similar basis, OX1 receptors located in the hippocampus, amygdala, and locus coeruleus are pivotal for the creation of memory traces strong enough to undergo later consolidation and long-term memory formation.

Jaeger et al. (2002) investigated the role of orexins in fear memory for the first time, using a T-maze active avoidance paradigm. They administered orexin-A intracerebroventricularly, which improved the results of 12-month-old SAMP8 mice in this paradigm in a dose-dependent manner. In 4-month-old mice, only the highest dose resulted in a better score; however, the SAMP8 strain develops memory impairments at approximately 6 months of age, which would mean that the second case showed improvement in clinically healthy individuals. Additionally, they tested the impact of OX-A ICV infusion on CD-1 mice in a passive avoidance paradigm. Both experiments showed that the peptide improves memory retention when administered immediately after training but not 24 h later. Considering that ICV administration mostly targets the hippocampus, the results are consistent with the notion that fear memory consolidation is a hippocampus-dependent process susceptible to loss over time. In another study, researchers managed to restore consolidation to base levels in mice with a genetic loss of orexins by injecting OX-A into the CA1 region, using a two-step active avoidance paradigm, further corroborating hippocampus involvement in this process (Mavanji et al., 2017). Subsequently, blocking the OX1 receptor with selective antagonist SB-334867 immediately after training impaired consolidation in contextual fear conditioning (Flores et al., 2014) and passive avoidance (Ardeshiri et al., 2017). In the first experiment, the drug was injected intraperitoneally, and in the second, directly into the basolateral amygdala. On the other hand, Sears et al. (2013) noted no effect of SB-334867 on the consolidation phase after post-training ICV infusion in cued fear conditioning. As the used doses were similar, these somewhat contradictory findings may result from different paradigms, as it was theorized that the amygdala might not be crucial for contextual conditioning as opposed to paradigms using auditory cues (Vazdarjanova and McGaugh, 1998). The involvement of orexinergic neurons in amygdala-dependent learning could benefit from further studies using c-Fos expression techniques to elucidate their engagement and amygdala response in different paradigms. The OX2 receptor seems to be significantly less involved in consolidation, as was proven by several different studies. The blockade of this receptor by ICV or intra BLA infusion of antagonist TCS-OX2-29 did not alter the performance of animal models in cued conditioning or passive avoidance (Sears et al., 2013; Ardeshiri et al., 2017), except for one study by Flores et al. (2014) in which intraperitoneal injection of the drug effectively impaired cued and contextual fear conditioning in rats. However, the dosage of TCS in this experiment was significantly higher than in the previous ones, which might have potentially blocked OX1 receptors as well, despite the high selectivity of the drug. Furthermore, the orexinergic neurons activate noradrenergic input of the locus coeruleus to the amygdala during the consolidation phase in cued and contextual fear conditioning (Soya et al., 2013). It was proven that mice lacking OX1R on noradrenergic LC fibers show impaired memory formation during cued conditioning, which can be abolished by restoring the receptors with an AAV vector. Subsequently, contextual conditioning was impaired in both OX1R and OX2R knockouts, which was not abolished by restoring the receptors. These results were further corroborated with DREADD techniques, showing that orexinergic innervation of the locus coeruleus is crucial for the activation of the amygdala in auditory-based paradigms and much less important in contextual learning (Soya et al., 2017; Dustrude et al., 2018). However, the exact circuits between the lateral hypothalamus, locus coeruleus, and amygdala in different disorder models should be investigated.

Experiments around the retrieval phase are based on the recognition of familiar stimuli, which, in turn, activates engram cells in a specific pattern. The timing of drug administration is pivotal for determining their influence on this process. Jaeger et al. (2002) administered OX-A intracerebroventricularly 24 h after training, immediately before a test session in passive avoidance and T-maze footshock avoidance paradigms, and concluded that the neuropeptide had no effect on test scores in animal models. However, OX-A infused 30 min before the test session improved the passive avoidance results in rats (Telegdy and Adamik, 2002), as was the case with OX-B in the same paradigm (Palotai et al., 2014), which may suggest that a longer period allows higher receptor saturation, which facilitates engram activation. However, more research is needed to test this hypothesis or propose a different one. On the other hand, the total blockade of either OX1 or OX2 receptor by selective antagonists administered intra BLA immediately before the test session impaired the retrieval phase in the passive avoidance paradigm (Ardeshiri et al., 2017), which suggests that both receptors take part in engram activation during fear memory recall. The study from Salehabadi et al. (2020) showed that, in Single Prolonged Stress (SPS) rats, which are the animal models of PTSD, three consecutive injections of SB-334867 into the amygdala (each after one test session, 24 h apart) could significantly decrease freezing behavior in a contextual fear conditioning paradigm. In a different study, a single injection of OX1R receptor antagonist after the first of two extinction sessions in contextual fear-conditioned mice enhanced fear extinction compared with the control group (Flores et al., 2017). Perhaps controlling orexinergic activation might prove useful in shifting memory processes toward extinction in severe anxiety cases or states such as PTSD. However, the downregulation of orexinergic innervation and its effect on memory in animal models of different disorders needs further elucidation.

The Morris water maze, which is predominantly used to assess spatial memory, tests the ability of animals to find and memorize the location of a transparent platform hidden in the pool in which they swim, allowing them to rest upon it. Aou et al. (2003) tested the effects of different doses of OX-A on performance in this paradigm, subsequently assessing long-term potentiation and long-term depression in Shaffer collaterals in rats. The results showed that intracerebroventricular infusion of the neuropeptide impaired spatial memory formation in all used doses. However, the drug was administered 2 h before training sessions, which deviates from other studies in this matter. The electrophysiology of hippocampal slices showed significant suppression of LTP by a higher dose, while LTD was not affected. A possible explanation for this phenomenon is that elevated OX-A increased levels of noradrenaline as well as activated GABAergic inhibitory interneurons, which, in turn, suppressed the induction of LTP. The lack of synaptic plasticity events within the medial temporal lobe resulted in the inhibited formation of memory traces, thus slowing the learning process (Sil’kis, 2013). In a different study using the same paradigm, the blockade of the OX1 receptor through the injection of SB-334867 directly into CA1 15 min before the test session significantly impaired memory encoding (Akbari et al., 2006). This may suggest that orexin A is pivotal for base-level spatial memory formation, which is on a par with the well-described OX-A-mechanism-activating ERK1/2 pathway (Ramanjaneya et al., 2009; Liu et al., 2015; Greene et al., 2020), which, in turn, triggers or strengthens LTP in the hippocampus (Peng et al., 2010). It might be useful to further elucidate the effects of orexins on spatial memory encoding at different times of drug administration to establish more concise frames for encoding studies.

The blockade of orexin receptors immediately after the training session by selective antagonist infusion into CA1 resulted in reduced time spent in the target quadrant in a Morris water maze (Akbari et al., 2006). Interestingly, it seems that the effect of this blockade can be mitigated by extended training. After antagonizing OX1 or OX2 receptors in the basolateral amygdala, the animals showed impaired spatial memory formation in the short Morris water maze paradigm, which was ameliorated with longer training (Ardeshiri et al., 2019). Furthermore, in a six-trial Morris water maze, the animal models were able to memorize the location of the hidden platform, similar to the base level learning, as opposed to the two-trial paradigm, during which the learning was significantly impaired by the OX1R blockade (García-Brito et al., 2018). The researchers subsequently assessed c-Fos protein expression in the dentate gyrus, CA2, CA2, dorsal retrosplenial cortex, prelimbic cortex, and thalamic nuclei. The immunohistochemistry data revealed that c-Fos expression in these areas was significantly higher in control animals and even higher in animals subjected to antagonist infusion and a prolonged training protocol, a finding that overlaps with the mitigation of the antagonist effect. The mechanism behind this phenomenon remains to be elucidated. Another way to alleviate the antagonist injection is intracranial self-stimulation (ICSS), as shown by the same authors (García-Brito et al., 2020). The procedure was able to compensate for receptor blockade in the Morris water maze and visual discrimination task. However, as the stimulation procedure was reported to be highly pleasurable, its effects may be explained by acting on the reward system, aside from memory formation, because the ventral tegmental area, a part of the medial forebrain bundle sending dopaminergic fibers, which is stimulated during ICSS, is innervated by orexinergic neurons and participates in memory formation processes. Likewise, the blockade of the OX1 receptor in the dorsal raphe nucleus, the most prominent serotonin source in the brain, impairs the consolidation of spatial memory (Khodabande et al., 2021). On the other hand, antagonizing the OX2 receptor with TCS-OX2-29 did not affect the results in the Morris water maze, despite the expression of OX2R. In one knockout mice experiment, animals without OX receptors in the dorsal raphe nucleus exhibited impaired spatial working memory, with locomotor activity largely intact in an exploration task and a delayed non-matching-to-place T-maze task. DRN serotonergic neurons promote wakefulness and regulate the sleep cycle (Sörman et al., 2011), as well as modulate theta rhythms in the hippocampus (Adidharma et al., 2019), which means that these results may not present a clear view of the involvement of orexins in DRN-related memory processes. What is more, the dorsal raphe nucleus is also densely innervated by noradrenergic fibers, which may excite it through a similar mechanism to orexins (Aghajanian, 1985). Therefore, to elucidate the precise effects of orexins on DRN in memory processes, further studies are required utilizing rigorous conditions to assess the overall state of the animal models.

The effect of orexins on spatial memory was also investigated in SPS rats. During one study, the control group animals showed impaired spatial memory in the Morris water maze test, decreased food intake and downregulation of OX-A. Subsequently, an upregulation of OX1 and OX2 receptors in the hippocampus was discovered, possibly as a compensatory mechanism. ICV injection of OX-A alleviated these symptoms, improving memory, enhancing appetite, and partially reversing the changes in receptor density (Han et al., 2020). Such a decrease in orexin levels was also suggested to contribute to age-related (Stanley and Fadel, 2012) and Parkinson’s disease deficits (Fronczek et al., 2012). Animal Parkinson’s models showed a correlation between declarative and spatial memory impairments and depletion of orexinergic neurons in the lateral hypothalamus/perifornical area (Oliveira et al., 2020). Destroying dopaminergic and orexinergic neurons by injecting 6-OHDA intrastriatally and Saporin intra-LH, respectively, and subsequently evaluating the orexin system with immunofluorescence techniques revealed that maximal memory deficiency was induced by the joint destruction of 88% of dopaminergic and 29% of orexinergic fibers. The role of orexins alone in this model needs to be further examined.

Spatial memory retrieval is the least researched phase in the recent literature. It has been shown to be affected by a pre-test session OX1R antagonist SB-334867 injection directly into CA1 in the Morris water maze paradigm (Akbari et al., 2006). Each used dose impaired memory retrieval in animal models which manifested in less time spent in the target quadrant, while visual and motor functions showed no alterations compared with the control group. Direct intervention in the trisynaptic loop diminished the ability to utilize an allocentric strategy in navigating the environment. However, injecting either SB-334867 or TCS-OX2-29 directly into the basolateral amygdala did not have any effect on the retrieval in rats (Ardeshiri et al., 2019). The results corroborate the notion that the role of the amygdala in the retrieval phase is time-dependent (Liang et al., 1982) and continually diminishes to the point that activating this structure via the orexinergic system does not produce any effect. Furthermore, no significant difference between the control group and experimental groups was found in the Morris water maze test using either OX1R or OX2R blockade in the dorsal raphe nucleus directly before the test session (Khodabande et al., 2021). Considering the differences between the effects of orexins on various structures and the time-dependent manner of these effects, more studies are needed to draw consistent conclusions. The effects of orexins or their antagonists were briefly summarized (Table 2).