Acetylcholine (ACh) is one of the main neuromodulators of the central nervous system playing an essential role in attention, regulation of the sleep-wake cycle, learning, and memory (Dannenberg et al., 2017). The brain has two major cholinergic projections: the basal forebrain and the brainstem cholinergic system. The basal forebrain cholinergic system includes the nucleus basalis of Meynert (nucleus basalis magnocellularis in rodents), substantia innominata (NB/SI), the medial septal nucleus, and the horizontal and vertical limbs of the diagonal band of Broca. These regions modulate learning, memory, synaptic plasticity, arousal, and attention (McCormick, 1993; Leanza et al., 1996; Villano et al., 2017). The brainstem cholinergic system comprises the peduncolopontine nucleus and the laterodorsal pontine tegmental nucleus, which has been described as part of the ascending reticular activating system. This system is implicated in the regulation of rapid eye movement (REM) sleep, wakefulness, and vigilance (Shouse and Siegel, 1992; Datta and Siwek, 1997). In rodents, HPC receives cholinergic inputs from the septum-diagonal band complex, whose fibers project to hippocampal subfields and most cell types, including pyramidal cells, granule cells, interneurons, and hilar neurons (Frotscher and Léránth, 1985; Woolf, 1991). The ventral regions of the mPFC (prelimbic and infralimbic), on the other hand, receive strong projections from the horizontal and diagonal band of Broca and only a few projections from the nucleus basalis (Chaves-Coira et al., 2018; Figure 2A).

ACh acts primarily by activating two types of membrane receptors, the G protein-coupled muscarinic acetylcholine receptors (mAChRs) and the ligand-gated nicotinic acetylcholine receptors (nAChRs). Muscarinic acetylcholine receptors (mAChRs) are divided into five subtypes (M₁-M₅). Of these receptors, M₁, M₃, and M₅ are excitatory and are coupled with the G_q/11 family of G proteins, while M₂ and M₄ are inhibitory and are coupled with the G_i/o family (Volpicelli and Levey, 2004). The M₁ and M₂ subtypes are the most abundant mAChRs in the brain (Volpicelli and Levey, 2004). While the M₁ receptor is localized post-synaptically and is expressed abundantly in the cerebral cortex and HPC, the M₂ mAChR is an inhibitory autoreceptor localized mainly on large cholinergic interneurons (Levey et al., 1991; Crook et al., 2001). In contrast, nAChRs belong to the family of the ligand-gated ion channels. There are 16 nAChRs subunits identified (α₁-α₁₀, β₁-β₄, γ, δ, ε), which can combine in multiple forms of receptors according to the brain region, neuronal subtype and animal developmental stage. The homomeric α₇ nAChRs and the heteromeric α₄β₂ are the most abundant in the adult mammalian brain and are highly expressed in the HPC and mPFC (Gotti et al., 2007).

Dopamine (DA) is the most extensively studied modulator of HPC-mPFC synaptic plasticity. Dopaminergic transmission is involved in numerous cognitive, emotional, and motor functions (Tritsch and Sabatini, 2012; Speranza et al., 2021), and impairments of this system are implicated in neuropsychiatric disorders, such as schizophrenia and major depression (Goto et al., 2010; Grace, 2016).

Dopamine is produced in midbrain and hypothalamic neurons. In the midbrain, the ventral tegmental area (VTA) is the primary source of afferents to limbic and cortical structures. There are five subtypes of dopamine receptors (D₁-₅) identified to date. They are grouped into two main classes: D₁-like and D₂-like. All DA receptors are G-protein coupled receptors (GPCR) which generally activate (via Gs/olf: D₁-like) or inhibit (via Gi/o: D₂-like) adenylate cyclase and protein kinase A (PKA) pathway (Beaulieu and Gainetdinov, 2011). The mPFC receives dense projections from the VTA, while the HPC receives more sparse terminals (Santana and Artigas, 2017; Edelmann and Lessmann, 2018). The expression of D₁ receptors in the mPFC is more densely concentrated in the deep layers (V-VI), which also receives limbic excitatory afferents. In contrast, the expression of D₂ receptors is preferentially localized in the superficial cortical layers (I-III), largely comprising intracortical connections (Berger et al., 1991; Gaspar et al., 1995; Santana and Artigas, 2017). In the HPC, there is a gradient of D₁ and D₂ expression across the dorsoventral axis, in which the ventral region aspect shows greater expression than the dorsal part (for a review, see Edelmann and Lessmann, 2018). Remarkably, VTA terminals in the mPFC occur in both dendritic shafts and spines and present proximity with HPC excitatory afferents forming synaptic triads (Carr and Sesack, 1996). This structural characteristic intriguingly suggests a relationship between DA, HPC, and mPFC.

Although in vitro studies demonstrate the dependence of PFC plasticity on DA, they do not reveal such a clear picture of how this modulation occurs (reviewed in Goto et al., 2010) as it has been observed in the intact brain (Jay et al., 2004). For this reason, we will review only studies that preserved the HPC-mPFC and mesolimbic circuits (Figure 3A). Jay et al. (1995) showed that electrical stimulation of the VTA, known to induce cortical DA release, inhibits the neural response of most (73%) mPFC neurons responsive to vHPC stimulation. This finding provided clear electrophysiological evidence for the convergence between vHPC and VTA afferents onto the mPFC, corroborating previous morphological studies. Additionally, they reported that electrical stimulation of the NAc induced antidromic activation of vHPC-responsive mPFC neurons (Jay et al., 1995). This work also suggested a circuit where vHPC controls DA release from the VTA through the PFC and NAc. Other studies confirmed that vHPC stimulation increases DA levels (Blaha, 1997; Floresco et al., 2001). However, the subsequent studies showed that the vHPC could control the VTA even when the PFC is inhibited by tetrodotoxin (Floresco et al., 2001). The control of the HPC over the VTA release is reviewed in Lisman and Grace (2005).

Noradrenaline (NA) is produced in the brainstem in specific locus coeruleus (LC) neurons. It has widespread terminals onto the forebrain, and it is proposed to be one of the most important brain systems to modulate arousal and attention (Sara, 2009). NA has three main types of receptors, α₁ (Gq GPCR), α₂ (Gi/o GPCR), and β (Gs GPCR), which are subdivided into three subtypes (Marzo et al., 2009). Importantly, NA receptors also have an affinity with epinephrine, a hormone released in the body during stress, suggesting that NA transmission is associated with stress-related modulation of brain function (Ross and Van Bockstaele, 2021). The PFC expresses NA receptors less densely than DA and in a more distributed manner across layers (Santana and Artigas, 2017). The HPC, in turn, expresses more adrenergic receptors than DA, where it is critical for LTP, especially in the Schaffer collateral pathway (CA3-CA1) (Nguyen and Connor, 2019). NA modulation of synaptic plasticity, including at the HPC and PFC, is reviewed in Marzo et al. (2009) (Figure 3A).

Lim et al. (2010) performed a comprehensive investigation demonstrating a positive link between NA and enhancement of vHPC-mPFC LTP. LC electrical stimulation just before bursts of HFS increased vHPC-mPFC LTP. In contrast, inhibition of the LC by local lidocaine injection decreased LTP. Recently, Takeuchi et al. (2016) showed that DA rather than NA released from the LC might play a more important role in plasticity and memory within the HPC. Nevertheless, Lim et al. (2010) showed that a NA reuptake inhibitor and systemic administration of an α2 antagonist, which disinhibits NA release, also increased LTP. In contrast, α2 agonist and partial lesion of NAergic LC neurons by DSP-4 decreased LTP. Noteworthy, no studies manipulated NA transmission directly in the PFC, and we still do not know the roles of specific adrenergic receptors at the mPFC in the facilitatory effect of NA on HPC-PFC plasticity. Nevertheless, NA exerts positive effects on mPFC-related cognitive functions and HPC plasticity via β-mediated signaling, which generally acts through the PKA pathway (Chamberlain and Robbins, 2013; Nguyen and Connor, 2019), while it modulates HPC and intracortical mPFC LTD mainly via α1 (Marzo et al., 2010). In addition, Lim et al. (2017) showed that NA indirectly modulates vHPC-PFC plasticity via the amygdala, a key region for processing fear and stress. NA-mediated activation and inhibition of the amygdala positively and negatively modulated vHPC-PFC LTP, respectively. Besides the role in stress, Bhardwaj et al. (2014) investigated the effects of neonatal lesion of vHPC, which is described as a model of schizophrenia, and reported a disruption of NA modulation of mPFC corticocortical plasticity via alterations of α1 signaling. Notably, NA and DA interact (Xing et al., 2016; Figure 3B). Therefore another possible mechanism of NA modulation could be indirectly by influencing DA transmission. NA manipulations can increase DA levels, and LC stimulation can activate the VTA (Grenhoff et al., 1993). The interaction between these signaling pathways has been investigated in the HPC and the mPFC (Xing et al., 2016) but still not in the HPC-PFC pathway.

Serotonin (5-hydroxytryptamine; 5-HT) is produced in specific neurons of the midbrain that comprise the raphe nuclei (RN). The RN has widespread innervations to all the forebrain, and the HPC and PFC are principal targets of these innervations (Azmitia and Jacobs, 1992). The most relevant nuclei regarding HPC and PFC function are the dorsal raphe nucleus (DRN) and the median raphe nucleus (MRN) (Puig and Gener, 2015). The DRN has widespread terminals in the forebrain and is the primary source of serotonin to the PFC, to whom it has reciprocal connections (Puig and Gulledge, 2011; Pollak Dorocic et al., 2014). In turn, the MRN has more specific connections, and it is the primary source of 5-HT to the septohippocampal system (Vertes and Kocsis, 1997). There are seven families of 5-HT receptors (5-HT₁-₇) divided into fourteen subtypes. All subtypes are GPCR, except for the 5-HT₃, which is ionotropic. The most expressed receptors in both structures are the 5-HT_1A (Gi/o GPCR) and the 5-HT_2A (Gs GPCR). There is a specific profile of expression of the distinct 5-HT receptor throughout neurons. In both PFC and HPC, 5-HT_1A is preferentially expressed in the somata and axons, while the 5-HT_2A is on the apical dendrites of pyramidal cells (Puig and Gener, 2015). PFC pyramidal neurons show co-expression of both receptors in the same neurons (Amargós-Bosch et al., 2004). Therefore, 5-HT has a significant inhibitory effect on the firing rate but promotes excitatory effects on field potentials (Puig and Gulledge, 2011; Figure 3A).

The endocannabinoid (eCB) system is a complex, widespread neuromodulatory pathway in the mammalian brain regulating multiple neural functions involved in cognitive, sensory, and emotional processes (Piomelli, 2003). The eCB system includes cannabinoid receptors, eCB neurotransmitters, and enzymes related to the synthesis and degradation of eCBs (Lu and MacKie, 2016).

The main eCB receptors are the CB₁ and CB₂ receptors, primarily expressed in the central nervous system and the peripheral immune cells, respectively (Chen et al., 2017; Kendall and Yudowski, 2017). The CB₁ and CB₂ receptors are Gi/Go-coupled receptors which inhibits adenylyl cyclases and voltage-dependent calcium channels, while activating MAP kinases and potassium channels, reducing neurotransmitter release (Howlett et al., 2002). The most abundant eCB neurotransmitters are 2-arachidonoylglycerol (2-AG) and anandamide (AEA), which are post-synaptically, synthesized “on-demand” by 2-arachidonoyl-containing phospholipids and N-arachidonoyl phosphatidyl ethanol, respectively (Pacher et al., 2006). In turn, monoacylglycerol lipase degrades 2-AG, and fatty acid amide hydrolase (FAAH) degrades AEA (Di Marzo et al., 1994; Dinh et al., 2002). In the central nervous system, CB₁ receptors are fundamentally pre-synaptically expressed both in excitatory and inhibitory synapses in mesocorticolimbic circuits, mediating complex retrograde signaling (Ohno-Shosaku and Kano, 2014). Not surprisingly, the eCB is a fundamental regulator in synaptic plasticity and functional connectivity in the HPC-mPFC pathway (Heifets and Castillo, 2009; Katona and Freund, 2012; Figure 4A).

Several studies report the role of the HPC-mPFC pathway in cognitive processes, such as spatial working memory and memory consolidation (Laroche et al., 2000; Eichenbaum, 2017). It has been argued that the ventral and intermediate HPC-mPFC projections are more related to working memory tasks, while the dorsal HPC is related to memory consolidation. Optogenetic inhibition of vHPC terminals in the mPFC during the encoding phase of a T-maze working memory task results in memory impairments (Tamura et al., 2017). Further, CNO injection into mPFC of subjects expressing hM4D(Gi) DREADD in HPC inactivated HPC-mPFC communication and disrupted memory consolidation (Ye et al., 2017). Similar approaches have been employed to investigate the mechanisms of plasticity in the HPC-PFC circuit. For example, the firing responses in the mPFC after LTP induction in the iHPC-mPFC projections were attenuated by optogenetically controlling the paraventricular/mediodorsal thalamic area (Bueno-Junior et al., 2018). Functionally, several studies report increases in HPC-mPFC oscillatory synchrony during spatial working memory tasks (Jones and Wilson, 2005; Benchenane et al., 2011; Yu and Frank, 2015) and memory consolidation (Paz et al., 2008; Binder et al., 2019). Despite the few works that directly investigated neuromodulation of HPC-mPFC during these cognitive processes, cholinergic neuromodulation of HPC-mPFC seems to be fundamental.

Microdialysis experiments demonstrated that the cholinergic drive increases in the neocortex and hippocampus during cognitive processes (Pepeu and Giovannini, 2004). This indirect evidence may lead us to conjecture that cholinergic tone can modulate the synchronization dynamics between HPC-mPFC during cognitive tasks. Recent evidence has shown that this interaction is more complex, depending on brain state and cognitive demand. First, the traditional view of the acetylcholine neurotransmission as primarily spatially diffuse and slow-acting, in the scales of minutes, has been challenged (Sarter et al., 2009; Sarter and Lustig, 2020). A recent study using in vivo amperometry and LFP recordings has helped elucidate the dHPC-mPFC acetylcholine release dynamics showing that tonic and phasic ACh release occurs in both regions, occur in different timescales, and conduct different functions (Teles-Grilo Ruivo et al., 2017).

Interestingly, both modes of transmission occur in a coordinated manner in the HPC and PFC. Tonic ACh release mode is strongly associated with a transition to a high-arousal state and the switch from deactivated to activated. On the other hand, phasic ACh release occurred only during a performance in a cognitive task and was strongly related to reward occurrence. In addition, Maharjan et al. (2018) used viral vector-mediated expression of muscarinic M₄ receptor coupled to Gi protein, hM4D(Gi) DREADD, to inactivate excitatory neurons in the dHPC and mPFC in a W-track spatial alternation task that tested both working and spatial memory. Their results demonstrated that contralateral dHPC-PFC inactivation with CNO impaired working but not spatial memory in the W-track test. Taken together, these results demonstrate that while coordinated HPC-PFC tonic ACh release is associated with arousal and deactivated states, creating a purposeful context to dynamical coordination between HPC-PFC during cognitive tasks may involve other neurotransmitters (Benchenane et al., 2010). Synaptic plasticity and functional effects of muscarinic stimulation in iHPC-mPFC are associated with enhancement of dopamine release in mPFC (Lopes Aguiar et al., 2008). In addition, D₁ receptor modulation in HPC-mPFC is critical to performance spatial working memory (Seamans and Yang, 2004). However, as seen before, D₁ modulation in HPC-mPFC synaptic plasticity occurs in an inverted U-shaped dose dependence, and DA facilitates theta oscillation HPC inputs to mPFC engagement (Gurden et al., 2000; Benchenane et al., 2010). Thus, dopaminergic stimulation plays a role in fine-tuning the information flow in HPC-mPFC.

The HPC-PFC synaptic plasticity and functional connectivity have also been implicated in several animal models of schizophrenia. Goto and Grace (2006) observed an aberrant increase of vHPC-mPFC LTP in the neurodevelopmental model of prenatal exposure to methylazoxymethanol acetate (MAM). In a genetic mouse model of schizophrenia, Sigurdsson et al. (2010) observed a decrease of HPC-PFC theta synchrony correlated with poor spatial working memory performance. Dickerson et al. (2010) found a similar reduction of synchrony in the maternal immune activation (MIA) neurodevelopmental model. Ultimately, Spellman et al. (2015) showed an augmented HPC theta-mPFC slow gamma coupling in wild-type mice correlated to spatial working memory task difficulty, which was even higher in a genetic model of schizophrenia that presented an impaired performance on the task. Similarly, pharmacological animal models of schizophrenia using NMDA receptor antagonists also indicate HPC-mPFC synaptic disruption. Ketamine, an NMDA receptor antagonist, reduced vHPC-mPFC LTP induction in rats, while co-administration of the atypical antipsychotic clozapine counteracted these effects (Rame et al., 2017). Alvarez et al. (2020) showed that early ablation of NMDA receptors in corticolimbic parvalbumin-expressing interneurons of mice impacts vHPC-mPFC connectivity before and after adolescence. While juvenile mice exhibited elevated cortical excitability in an uncoordinated manner relative to the HPC, adult mice showed weaker evoked potentials in mPFC and increased LTD, suggesting impaired functional connectivity during adulthood. In iHPC-mPFC, ketamine-induced aberrant oscillatory coupling in mPFC and alterations in the HPC-PFC synaptic efficacy (Lopes-Aguiar et al., 2020). The authors demonstrated that prior induction of LTP in the CA1-mPFC attenuates these effects, highlighting the importance of high-frequency stimulation as a non-pharmacological alternative strategy to investigate HPC-PFC modulation and help in the prevention or attenuation of cognitive impairments in schizophrenia.

Remarkably, dopaminergic and eCB neuromodulation in HPC-mPFC are associated with the impairments in animal models of schizophrenia. D₂ agonism or D₁ antagonism revert aberrant LTP enhancement in the MAM model (Goto and Grace, 2006). However, there is still no evidence for the D₂ attenuating effect of LTP in a non-pathological brain state. Although most studies focused on the positive modulation of LTP by D₁ and showed the independence of this phenomenon on D₂, recent evidence implicates D₂ signaling in the occurrence of LTD. Banks et al. (2015) applied a TBS protocol to induce LTD in the HPC-PFC in vitro. Then, they showed that applying a D₂ antagonist, but not D₁, blocked LTD induction. Strikingly, they showed that LTD induction impaired NMDAr, but not AMPAr, synaptic transmission in this pathway. This finding is intriguing because it provides a mechanism that gathers exaggerated DA, hypofunction of NMDAr, and treatment by D₂ antagonism, which comprises critical neurobiological elements of schizophrenia (Grace, 2016). There is also a complex relationship between eCB neuromodulation and HPC-mPFC (Aguilar et al., 2016; Ruggiero et al., 2017). URB597, a FAAH inhibitor that leads to AEA accumulation, increases mPFC firing rate only in rats submitted to subchronic phencyclidine (PCP) model, while THC does not affect mPFC firing rate following PCP model but decreases firing rate in the control group (Aguilar et al., 2016). The exact mechanism of these effects remains to be elucidated but it is likely related to disruption in other neurotransmission systems (i.e., dopaminergic and GABAergic) induced by PCP, which differently influence eCB effects (Laviolette and Grace, 2006; Volk and Lewis, 2016).

Alterations in synaptic plasticity are among the main effects of a deleterious experience of stress (Pittenger and Duman, 2008). It is proposed that antidepressants, such as SSRIs, counteract the effects of stress on plasticity across the brain (Pittenger and Duman, 2008; Duman, 2014). This pattern also occurs in the HPC-mPFC pathway. Rocher et al. (2004) exposed rats to an unbalanced elevated platform, which induced a lasting increase in stress-related glucocorticoids, and they observed a marked impairment of vHPC-mPFC LTP. Then, they observed that acute systemic injection of fluoxetine, an SSRI, and the atypical antidepressant tianeptine restored LTP. Interestingly, both drugs did not affect LTP in non-stressed animals. Also remarkable, tianeptine presented a more significant effect than fluoxetine (Rocher et al., 2004). Another study applying the same stress procedure compared the effects of tianeptine and imipramine, a general monoamine reuptake blocker, and observed no significant effect of imipramine on restoring stress-induced effects on LTP (Qi et al., 2009). Nevertheless, although these findings have significant implications for the link between HPC-mPFC plasticity and depression etiology and treatment, tianeptine does not have a specific mechanism of action, so it hinders interpretation. Additionally, there are reports that fluoxetine increases the release of both DA and NA in the cortex (Bymaster et al., 2002), which could also be the actual modulators of plasticity. Moreover, recent evidence indicates that antidepressants may bind directly to tyrosine kinase receptors (TrkB) (Casarotto et al., 2021), which are usually activated by the brain-derived neurotrophic factor (BDNF), which is a pivotal regulator of neural plasticity (Castrén and Antila, 2017). In conclusion, 5-HT modulates HPC-PFC LTP negatively, but antidepressants, including SSRIs, facilitate plasticity, possibly through serotonin-independent pathways.

Among the best studied behavioral functions of HPC-PFC transmission is anxiety. A series of studies using the elevated plus-maze showed that mPFC synchronizes preferentially with the vHPC rather than dorsal parts, especially in the theta frequencies at the environment’s safer zones (closed arms) (Adhikari et al., 2010). Then, they also showed that the mPFC neurons that encode aversion-related information, such as the location in the closed or open arms, are strongly phase-locked to vHPC theta (Adhikari et al., 2011). Similar results were found in the open field. Optogenetic inhibition of vHPC-mPFC terminals decreased both theta synchrony in the anxious environment and expression of anxiety-related behaviors (Padilla-Coreano et al., 2016). More recently, it was shown that sinusoidal optogenetic stimulation of vHPC-PFC terminals at the theta frequency, which mimic more precisely endogenous LFP, induced behavioral avoidance to the safer zones more robustly than by applying single pulses (Padilla-Coreano et al., 2019). Anxiety-related behaviors are also linked to serotonin neuromodulation in the vHPC-mPFC circuit. Knockout mice for 5-HT_1A exhibited increases in both anxiety and mPFC theta power, while 5-HT_1B agonist in the mPFC reduced both anxiety and theta power in the elevated plus maze (Adhikari et al., 2010; Kjaerby et al., 2016)

Impairments of HPC-PFC connectivity have also been associated with depression. Zheng and Zhang (2015) investigated the effects of LTP induction on the oscillatory coupling between vCA1 and mPFC in the chronic unpredictable stress (CUS) model of depression in rats. Remarkably, they showed that CUS rats showed both a weaker LTP induction and reduced theta phase synchrony. Furthermore, they reported a positive correlation between the two measures, suggesting a relationship between these two estimates of neural connectivity. Jia et al. (2019) investigated the oscillatory and antidepressant effects of electrical LFS or HFS in the mPFC. They showed that CUS promoted a reduction of HPC-PFC beta and gamma powers, but electrical stimulations promoted both antidepressant responses and increases in beta and gamma HPC-PFC coherence. Carreno et al. (2016) used optogenetics and chemogenetics to show that the selective activation of the vHPC-mPFC ascending pathway also produces an antidepressant-like response comparable to that observed with sub-anesthetic doses of ketamine. Finally, Marques et al. (2019) investigated iHPC-PFC network dynamics during exposure to controllable or uncontrollable stress. They identified a pattern of increased HPC-PFC theta connectivity that accurately predicted resistant animals but was absent in helpless subjects, suggesting a role of HPC-PFC communication in adaptive stress coping.

Recent studies in animal models for Alzheimer’s disease (AD) have provided further evidence on HPC-mPFC disruption. Optogenetic activation of hippocampal cells involved in memory formation resulted in long-term memory retrieval in a genetic model of AD, suggesting that retrieval, rather than storage, is impaired in the disease (Roy et al., 2016). Specifically in the HPC-mPFC, alterations in proteins related to AD, such as the decay of calcium-independent phospholipase A2 (iPLA2) and the increase of amyloid-β peptide (Aβ), decreased prefrontal gamma oscillation engagement during spatial working memory (Liu et al., 2016) and abolished CA1-mPFC LTP induction related to spatial working memory deficits, respectively (Shalini et al., 2014). In addition, Bazzigaluppi et al. (2018) showed that the TgF344 genetic rat model of AD, presents lower HPC-mPFC low-gamma power, decreased theta-gamma PAC, and weaker gamma coherence compared to non-transgenic rats. The same frequency bands were analyzed during a stimuli association task performed by adult rats overexpressing the tau protein in the entorhinal cortex (Tanninen et al., 2017). In this case, tau-expressing rats showed attenuated theta-gamma phase-phase and amplitude-amplitude coupling between dHPC and mPFC. Interestingly, these oscillatory activities are also cholinergically modulated, which is impaired in AD (Ballinger et al., 2016; Ferreira-Vieira et al., 2016; Al-Onaizi et al., 2017). In fact, recent evidence has contributed to the advance of new AD treatment through cholinergic neuromodulation. Esteves et al. (2017) showed that chronic nicotine treatment prevented novel object recognition memory deficits and disruption of iHPC-mPFC LTP in the streptozotocin animal model of AD. However, further research is necessary to evaluate the effects of cholinergic neuromodulation on functional connectivity in HPC-mPFC in AD.

The concept that neuropsychiatric symptoms could emerge from alterations in information processing within neural networks rather than an imbalance of a specific neurotransmitter is now consolidated (Castrén, 2005). In this view, a promising approach is to investigate the effects of neuropsychiatric drugs on specific networks of the brain. Relevant clinical drugs with non-selective actions make it hard to interpret mechanisms, but investigating their effects in cognitive circuits could be exceptionally informative about the therapeutic effect and elucidate the underlying network functioning that can compensate for the cognitive symptoms. For instance, there is evidence that the different systems may interact to promote efficient modulation of synaptic function (Meunier et al., 2017). Indeed, all monoamines are known to interact between themselves (González-Burgos and Feria-Velasco, 2008; Xing et al., 2016) and with both eCB (Cohen et al., 2019) and cholinergic systems (Lester et al., 2010). However, there is still no knowledge on how these interactions may influence HPC-PFC synapses. In this review, we focused on the basic mechanisms of modulation. However, there is significant evidence that some manipulations only show effects when counteracting the impacts of stress or developmental deficits (Dupin et al., 2006; Goto and Grace, 2006), suggesting a more general role in regulating synaptic homeostasis rather than increasing or decreasing connectivity.

We found a profound difference in research interest on the modulations of the intermediate against the ventral hippocampal inputs to the PFC. While monoamines and cannabinoids were only studied in vHPC inputs, cholinergic modulation was mainly investigated in iHPC. Although evidence has accumulated for a functional distinction across the septotemporal axis of the HPC, we still do not know if the ventral and intermediate HPC inputs into the mPFC are differently modulated by the systems addressed here. Interestingly, most modulation transmitters are released into the PFC in a non-synaptic manner, suggesting a convergence of these modulations onto both inputs. However, the coincident phasic activation of brainstem efferent systems could modulate each input differently across time. For instance, vHPC stimulation markedly increases VTA activity and DA release (Lisman and Grace, 2005). However, iHPC tetanic stimulation was shown to decrease monoamine levels (Lopes Aguiar et al., 2008), suggesting distinct influences of the HPC inputs over the neuromodulatory systems.

One important aspect to recognize is that most studies adopted an artificial stimulation protocol to probe synaptic plasticity. Such robust protocols elicit synchronous spike activity from a large number of neurons in a prolonged and high-frequency manner that unlikely replicate what happens in a physiological state (Bocchio et al., 2017). Thus, it is essential to understand the spontaneous network electrophysiological activities related to synaptic plasticity. Understanding the network effects of synaptic plasticity in the LFP measures while controlling for the artificial stimulation could be useful to understand the dynamics of HPC-mPFC interaction during behavior. It is postulated that oscillatory synchrony facilitates neural communication creating a temporal window in which one region could influence or coordinate the activity in another distant brain region (Fries, 2015). Hence, it is theorized that an increase in synaptic influence would also generate a stronger LFP coupling. Zheng and Zhang (2015) showed that LTP induction in the vHPC-mPFC pathway increased theta phase coupling rather than the power of either region. Also, stimulation protocols that resemble more physiological activity, such as TBS, induce theta-gamma cross-frequency coupling and interfere with memory consolidation (Radiske et al., 2020). Future studies need to investigate the relationship between synaptic plasticity and network coupling in freely moving animals and control for the effects of stimulation per se. Furthermore, the effects of other forms of synaptic plasticity in HPC-mPFC network coupling, such as LTD and PPF, remain to be investigated.