Basal forebrain cholinergic systems as circuits through which traumatic stress disrupts emotional memory regulation

Abstract

Contextual and spatial systems facilitate changes in emotional memory regulation brought on by traumatic stress. Cholinergic basal forebrain (chBF) neurons provide input to contextual/spatial systems and although chBF neurons are important for emotional memory, it is unknown how they contribute to the traumatic stress effects on emotional memory. Clusters of chBF neurons that project to the prefrontal cortex (PFC) modulate fear conditioned suppression and passive avoidance, while clusters of chBF neurons that project to the hippocampus (Hipp) and PFC (i.e. cholinergic medial septum and diagonal bands of Broca (chMS/DBB neurons) are critical for fear extinction. Interestingly, neither Hipp nor PFC projecting chMS/DBB neurons are critical for fear extinction. The retrosplenial cortex (RSC) is a contextual/spatial memory system that receives input from chMS/DBB neurons, but whether this chMS/DBB-RSC circuit facilitates traumatic stress effects on emotional memory remain unexplored. Traumatic stress leads to neuroinflammation and the buildup of reactive oxygen species. These two molecular processes may converge to disrupt chBF circuits enhancing the impact of traumatic stress on emotional memory.

Introduction

Mammalian organisms use contextual and spatial features to navigate their environment, learn associative rules, and regulate emotional expression ^1–6. Cholinergic basal forebrain (chBF) neurons project to neural substrates that represent contexts and space (i.e. contextual/spatial systems) ^7–10 and modulate physiological and behavioral processes that are dependent on these systems. Examples include cortical sensory plasticity ^11–13, place field re-mapping in the hippocampus (Hipp) ¹⁴, and spatial learning and memory ^15,16. Contextual/spatial systems are sensitive to traumatic stress and have been implicated in traumatic stress-induced psychiatric disorders such as post traumatic stress disorder (PTSD), especially with regard to traumatic stress effects on emotional memory regulation ^2,17. This would suggest that chBF neurons could have a role in facilitating the impact traumatic stress has on emotional memory regulation. This is especially likely, because chBF neurons are sensitive to neuroinflammation and the buildup of reactive species; two molecular processes that emerge with traumatic stress. Also, chBF neurons are critical for emotional regulation as well (see below). Unfortunately, the potential role of chBF neurons in traumatic stress-induced disruptions in emotional memory regulation remains largely unexplored.

In this article, we review studies that point to a role for chBF neurons in traumatic stress effects on emotional memory regulation. First, we describe the impact traumatic stress has on contextual/spatial memory systems and how these traumatic stress effects can lead to changes in hormonal systems that regulate emotional memory and behavioral and physiological measures of emotional memory regulation. Next, we describe what is known about the role of chBF neurons in emotional memory regulation and how chBF neurons may form circuits with contextual/spatial memory systems to regulate emotional memory. A significant gap in the literature that emerges from a review of the literature is the lack of studies examining the role of the retrosplenial cortex (RSC, a contextual/spatial memory system) in traumatic stress effects on emotional memory regulation and chBF neuronal input to the RSC in regulating emotional memory.

In the last sections of the article, we review studies that demonstrate chBF neurons are sensitive to neuroinflammation and the buildup of reactive species, and that disruptions in the integrity of chBF neurons can lead to neuroinflammation. We discuss how traumatic stress could compromise the integrity of chBF neurons, which may lead to increased inflammation in efferent targets of these neurons. Via these mechanisms (i.e. neuroinflammation, oxidative stress) traumatic stress may act via chBF neurons to produce changes in contextual/spatial systems that lead to disruptions in emotional regulation.

2. Contextual/spatial memory systems in PTSD

Two contextual/spatial memory systems that have been consistently implicated in PTSD are the Hipp and prefrontal cortex (PFC) ^{1–3,5,18–20}. The Hipp of PTSD patients is smaller than matched controls and this may emerge from smaller hippocampi as a result of trauma exposure ^21–25. Alternatively, smaller hippocampi prior to trauma exposure may represent increased susceptibility to developing PTSD ^26,27. Studies have consistently reported smaller PFC volumes in PTSD patients ^17,28,29 and a recent consortium study has shown that several PFC regions (e.g. lateral orbitofrontal gyri, left superior temporal gyrus) are smaller in PTSD patients ³⁰.

The inability to regulate fear and anxiety that stem from traumatic events is a core feature of PTSD ³¹. This deficit in emotional regulation may not be restricted to fear generated from traumatic memory (i.e. re-experiencing), but could also represent a broader deficit in the ability to regulate fear memory ^32,33. To examine fear memory formation, expression, and regulation, Pavlovian fear conditioning experiments are often utilized. Pavlovian fear conditioning is a relatively simple behavioral paradigm where a neutral stimulus such as a tone or a light is paired with an aversive event, such as electrical shock ^19,34–37. The neutral stimulus, referred to as the cued conditioned stimulus (CS), is paired once or several times with the aversive event, referred to as the unconditioned aversive stimulus (UCS). When a specific cued CS is paired with a UCS (specifically co-terminates with the UCS), the behavioral protocol is referred to as cued (or delayed) fear conditioning ^35,36,38. The CS can also be a distinct space or arrangement of objects (e.g. type of background noise, odor, or texture places into a distinct space, virtual reality experience) that form a context ^1,2,39. When the CS is a context, fear conditioning is referred to as contextual fear conditioning ^1,2,19,36,39. Repeated exposure of a feared CS in the absence of UCS presentation is referred to as fear extinction, which results in the formation of a separate memory that inhibits fear memory expression ^1,33,40–45. After the formation of extinction memory, fear and extinction memory compete for expression (i.e. emotional memory regulation).

Below, we review clinical and preclinical studies that have examined the roles of the Hipp and PFC in traumatic stress effects on emotional memory regulation. Adrenal hormones are critical for memory consolidation ^46–49, PTSD is characterized by changes in the hypothalamic-pituitary-adrenal (HPA) axis, and contextual/spatial systems regulate HPA axis activity (see below). For these reasons, we review the potential role of traumatic stress-induced changes in HPA axis function that may lead to changes in emotional memory regulation.

2.1. Traumatic stress, HPA axis, and emotional memory

Glucocorticoid receptors (GRs) are densely expressed in the Hipp and PFC. Increased glucocorticoid (e.g. corticosterone, cortisol) release from the adrenal glands into the circulation and binding of glucocorticoids to GRs in the Hipp and PFC influence the HPA axis to subsequently lower the overall levels of circulating glucocorticoids (i.e. negative feedback of the HPA axis) ^50–55. Enhanced negative feedback of the HPA axis is often observed in PTSD patients ^56–62 (though for contradictory findings see ^63,64). Findings of enhanced negative feedback of the HPA axis in PTSD suggest enhanced GR sensitivity in the HPA axis and/or substrates that regulate the HPA axis ^26,65. While tangential, studies examining GR dynamics on lymphocytes support this assertion. GR binding on lymphocytes can be enhanced in PTSD patients ^66,67, GR levels on lymphocytes can be elevated in PTSD patients ^68,69, and enhanced lymphocyte GR levels and binding represent risk factors for developing PTSD after trauma exposure ^66,70. Together these findings suggest that enhanced negative feedback of the HPA axis is a feature of PTSD and is driven by enhanced GR expression.

Preclinical experiments are useful in examining neurobiological mechanisms of psychiatric disorders and can be used to further explore how increased GR sensitivity within the brain leads to specific physiological and behavioral effects ^71–73. The single prolonged stress (SPS) rodent model of traumatic stress refers to the serial application of multiple stressors, followed by a quiescent period (stress incubation period). The stressors most commonly used for SPS are restraint, forced swim, and ether exposure with stress incubation periods that are at least seven days, but can be as long as 30 days ^71–75. Studies have consistently reported that SPS enhances negative feedback of the HPA axis ^76–81 and GR sensitivity (most often by enhancing GR levels) in the dorsal Hipp (dHipp, which is believed to correspond to the region of the Hipp in humans that is critical for representation of contexts and space ^2,82–85) ^{78,81,86–95}. Furthermore, antagonizing GRs during SPS blocks both upregulation of dHipp GRs and enhancements in negative feedback of the HPA axis ^94,96. These findings directly implicate SPS-enhanced dHipp GR expression with SPS-enhanced negative feedback of the HPA axis. GR enhancements in the PFC ⁹⁰ and ventral Hipp (vHipp) ⁹⁷ have been reported, but these effects are not as consistently observed ^88,90. It is still not clear if SPS differentially enhances dHipp GR levels in the cytoplasm or nucleus of cells as reports are inconsistent ^88,94. It should be noted that SPS also affects GR function in cells within brain regions outside of the Hipp and PFC ^{88 97}.

Enhanced negative feedback of the HPA axis may be linked to enhancements in dHipp GR levels within PTSD and within the SPS model. Furthermore, enhanced dHipp GR levels are also associated with broader deficits in hippocampal function within the SPS model, with some studies implicating enhanced dHipp GRs with emotional memory dysregulation in the SPS model (see below).

2.2. Emotional memory regulation

When PTSD patients are subjected to fear conditioning and extinction procedures, these patients consistently exhibit enhanced fear memory ^98–100 and/or an inability to acquire or maintain extinction memory ^32,101–105. After acquisition of fear extinction, a CS (which at this time is typically referred to as an extinguished CS) can signal either threat or safety. To disambiguate the meaning of an extinguished CS, organisms can use contextual information to determine when a CS potentially signals threat or safety ^1,3. In this way, contextual features can act as an occasion-setter and prime the retrieval of extinction memory (i.e. contextual occasion-setting ^1,3). This suggests that disruptions in contextual occasion-setting could lead to deficits in emotional regulation (e.g. increased re-experiencing symptoms) in PTSD patients. While less studied, two reports suggest that PTSD may be associated with a disruption in contextual occasion-setting ^103,105. In Steiger et al., (2015) valence ratings in response to a feared CS presented in a context that was also paired with electrical shock was resistant to extinction in PTSD patients when compared to non-trauma controls. In Garfinkel et al., (2014) PTSD patients exhibited deficits in the maintenance of extinction in a safe context. In a threat context, PTSD patients showed lower levels of conditioned responding to a feared CS. These results could be interpreted to mean that PTSD patients were unable to use contextual information to regulate fear responses ¹⁰³.

Both the PFC ^18,106–108 and Hipp ^35,36,84 are critical for representing contextual information and the rodent dHipp is also critical for contextual occasion-setting ^{1,2,109–111}. Regions of the PFC also have direct roles in regulating expression of fear by inhibiting neural activity within amygdala nuclei ^17,40,43,112. From these observations it follows that the PFC and Hipp would have roles in facilitating changes in emotional memory regulation that characterize PTSD. Results from clinical studies that utilize fMRI techniques during fear conditioning and extinction experiments support this hypothesis. PTSD is associated with decreased BOLD signal in regions of the PFC (e.g. ventromedial PFC (vmPFC)) and Hipp during acquisition and/or recall of extinction ^103,113,114. During extinction testing, exaggerated amygdala BOLD signal has been observed ^115–117. A similar pattern of neural activity (i.e. decreased BOLD signal in the PFC and/or increased signal in amygdala nuclei) are also observed when protocols that require inhibition of behavioral or emotional output ^118–120. Because PFC inhibition of amygdala nuclei are critical for regulation of fear ^{42,43,121–123}, a deficit in PFC-mediated inhibition of amygdala nuclei is believed to represent a critical circuit through which dysregulation of fear memory manifests in PTSD ^{113,114,124,125}

In clinical experiments decreased PFC and Hipp BOLD signals have been observed when an extinguished CS is presented in a threat context (i.e. fear renewal). During fear renewal, contextual information can be used to determine if a CS signals threat and fear responses to an extinguished CS typically re-emerge. Diminished BOLD activity in the PFC and Hipp during fear renewal could suggest deficits in contextual occasion-setting ¹⁰³. However, this effect has also been observed when an extinguished CS is presented in a safe context ^113,114 and it should be noted that at least one study has reported enhanced Hipp BOLD activity during contextual fear conditioning in PTSD patients ¹⁰⁵. While it is possible that deficits in emotional memory regulation within PTSD emerges because of deficits in contextual processing brought on by changes in neural function within the PFC and Hipp of PTSD patients, further research is needed to better define how (or if) deficits in PFC and Hipp function within PTSD patients lead to deficits in emotional memory regulation via deficits in contextual processing.

The preclinical literature also supports a role of the medial PFC (mPFC) and dHipp in deficits in emotional memory regulation brought on by traumatic stress. This has been explored extensively using the SPS model of traumatic stress. SPS leads to decreased excitatory tone within the mPFC ^126,127 and LTP in the CA1 region of the dHipp ⁹¹. Enhancements in dHipp GR expression are associated with enhanced contextual fear memory and deficits in maintenance of extinguished fear ^{87,90,94,128,129}. During acquisition of extinction memory neural activity is enhanced in the mPFC ^122,130 and SPS disrupts this enhancement specifically within the infralimbic cortex (IL) of the mPFC ^130,131; a PFC region that is critical for extinction memory ^40,43. Both optogenetic ¹³² and chemogenetic ¹³¹ stimulation of the IL after SPS attenuates SPS effects on extinction memory. These findings suggest that the IL is a locus through which traumatic stress-induced deficits in the maintenance of extinction manifests. In most experiments where the effects of SPS on extinction memory are examined, fear conditioning occurs in one context (e.g. threat context) and extinction training and testing occurs in another context (e.g. safe context) ^{74,87,90,92,129,133}. Because organisms use contextual features to determine if a CS represents threat or safety (see above), the experimental design adopted in many SPS conditioning experiments raise the possibility that deficits in the maintenance of extinction are driven by deficits in the ability to use contextual features to assign meaning to a cued CS (i.e. contextual occasion-setting). One study has observed the IL is important for hierarchal occasion-setting ¹³⁴, but more research is needed to determine the role of the IL (if any) in contextual occasion-setting and if traumatic stress effects on extinction maintenance are linked to traumatic stress effects on IL modulation of contextual occasion-setting. This is particularly so, because alternative explanations of the role of the IL in facilitating SPS deficits in the maintenance of extinction exists. The effects of SPS on IL neural activity could be construed as a failure to inhibit neural activity in subcortical substrates such as the lateral amygdala and/or central nucleus of the amygdala. This would represent a deficit in extinction memory formation and/or retrieval. In support of this interpretation, SPS disrupts functional connectivity between the mPFC and amygdala ¹²⁷ and specifically does this during acquisition of extinction memory ¹³⁵. Still other possibilities exist. The results of other studies point to an enhancement in the resistance of fear memory to extinction as the cause of extinction memory deficits within the SPS model ^129,136. It should be noted that these mechanisms are not mutually exclusive. It is possible that traumatic stress, acting via multiple behavioral mechanisms facilitated by overlapping, but distinct, circuits lead to fear memory that is difficult to regulate.

While prior research has examined the contribution of the PFC and Hipp to traumatic stress effects on emotional memory regulation, a contextual/spatial system that has received very little attention is the RSC. The RSC is critical for contextual fear conditioning ^137–139, binding elements within an environment into a single context (i.e. configural processing) ¹⁴⁰, and association of elements in the absence of reinforcement (i.e. sensory preconditioning) ¹⁴¹. The RSC is connected to brain regions critical for spatial navigation ^137,142,143 and neural activity within the RSC may serve as a, ‘homing signal’, during spatial navigation ¹⁴⁴. As such, it is a contextual/spatial system, though the impact of traumatic stress on the RSC remains under explored.

To address this, we conducted an experiment where SPS and unstressed (i.e. controls) rats were subjected to fear conditioning, extinction training, and extinction testing (CS-fear). Fear conditioning comprised of five auditory CSs (10s, 2kHz) and footshocks (1s, 1mA) presentations where CSs and footshocks co-terminated together. Extinction training was 30 CS presentations and extinction testing was 10 CS presentations, both in the absence of footshocks. We also subjected another set of SPS and control rats to similar procedures, except during fear conditioning footshock presentation was omitted (CS-only controls). All stimuli presentations were separated by an interval of 60s and behavioral sessions (e.g. fear conditioning, extinction training) were separated by a 1-day interval. We then measured c-Fos levels in the RSC (granular and agranular regions) in subsets of rats that were euthanized after fear conditioning (CS-fear: SPS = 5, Control = 8; CS-only: SPS = 5, Control = 8), extinction training (CS-fear: SPS = 13, Control = 12; CS-only: SPS = 12, Control = 12), extinction testing (CS-fear: SPS = 8, Control = 10; CS-only: SPS = 12, Control = 10) and immediate removal from the housing colony to establish baseline levels of c-Fos (SPS = 17, Control = 8). These data are novel, but are from a larger dataset whose protocols, regulatory approval, and behavioral results have been previously published ^130,135. SPS may have decreased c-Fos levels in the RSC (granular and agranular regions) at baseline as this effect approached statistical significance [SPS vs. control: t₍₂₃₎ = 1.951, p = .063; Figure 1]. C-fos levels during all behavioral sessions were subjected to separate stress (SPS vs. control) x session (fear conditioning, extinction training, extinction testing) factor designs for the CS-fear and CS-only treatments. For the CS-fear treatment, SPS may have decreased c-Fos levels during fear conditioning, as a stress x session interaction approached significance [F_(2,56) = 2.949, p = .063]. For the CS-only treatment there was a main effect of session [F_(2,59) = 8.219, p < .001], which reflected increased c-Fos RSC levels during fear conditioning in comparison to extinction training and testing. Importantly, there were no stress effects (ps > .05).

Figure 1.

Effects of SPS on c-Fos levels in the RSC (granular and agranular regions) at baseline, fear conditioning, extinction training, and extinction testing. A) representative near-infrared immunohistochemical image staining for c-Fos in the RSC. B) Effects of SPS on c-Fos levels during baseline and behavioral tasks. Baseline c-Fos levels in the RSC was lower in SPS rats when compared to controls and this effect approached significance. For the CS-Fear treatment c-Fos levels were lower during fear conditioning alone and this potential interaction (stress x treatment) approached significance. For the CS-Only treatment there were no stress effects.

Together, these results raise the possibility that SPS decreases activation of the RSC during fear memory formation and this could contribute to extinction retention deficits (i.e. inability to maintain extinction) within the SPS model, though further research is needed to identify mechanisms via which SPS changes in RSC function lead to SPS extinction retention deficits.

3. Role of chBF neuronal input to spatial/contextual memory systems in emotional memory regulation

3.1. chBF neurons

ChBF projection neurons are located in clusters in the base of the forebrain and can be found from the olfactory tubercle to the rostral end of the lateral geniculate bodies. There are excellent reviews and empirical studies about chBF neuron clusters and properties ^{7–10,145–149}. This section provides a brief summary of chBF neurons and their properties, and is not meant to be a comprehensive description of chBF neurons or cholinergic receptor pharmacology.

The activation of chBF and consequent membrane depolarization and Ca²⁺ influx of presynaptic terminals results into the release of acetylcholine (ACh) into the target terminal fields ^150,151. Termination of ACh actions on receptors is accomplished by metabolism of ACh by acetylcholinesterase (AChE) ¹⁵² and axonal choline reuptake ^153–155. This choline is then used to synthesize ACh at the axon terminal ^153,154; a reaction catalyzed by choline acetyltransferase (ChAT) ¹⁵⁶. ACh is then packaged into synaptic vesicles via the vesicular acetylcholine transporter (VAChT) for synaptic release ¹⁵⁷.

Cholinergic neurons signal through metabotropic muscarinic receptors (mAChRs) and ionotropic nicotinic receptors (nAChRs) ¹⁵⁵. There are five broad types of mAChRs with m1AChR and m3–5AChRs being post synaptic and m2AChR being presynaptic ^155,158,159. Activation of mAChRs lead to stimulation of phosphoinositol synthesis and or inhibition of cAMP synthesis ¹⁶⁰ and changes in the permeability of the neuronal membrane to K⁺, Ca²⁺, and Cl⁻ channels ¹⁶¹. NAChRs are pentameric ion channels comprised of α and β subunits in different configurations ¹⁵⁵. There are at least 11 variants of the α subunit and eight variants of the β subunit ¹⁶². The two most common types of nAChRs in the brain comprise of α4β2 subunits and α7 subunits exclusively ^162,163. Furthermore, even though nAChRs form a non-specific cation pore, receptors with different subunits can have selectivity for different types of cation species ¹⁶⁴. ChBF neuronal clusters in the anterior region of the BF project to midline cortical regions, Hipp, and medial habenula, while more posterior BF regions project to the dorsolateral neocortex, PFC, and amygdala. Input to chBF neurons originate from multiple systems in the brain with chBFs receiving neocortical, amygdala, and subcortical input ^{10,145,165–167}.

3.2. Role of chNBM neurons in emotional memory regulation

Given the role of the PFC in facilitating traumatic stress effects on emotional memory regulation, chBF input to the PFC could form circuits through which traumatic stress alters these emotional processes. This is especially so, because clusters of cholinergic neurons that project to the PFC have been implicated in emotional memory (see below). ChBF input to the PFC originates from two heterogeneous clusters of cholinergic neurons. Nucleus basalis of Meynert (chNBM), which also has cholinergic neurons that project to the amygdala, and medial septum and diagonal band of Broca (chMS/DBB), which also has cholinergic neurons that project to the Hipp and medial habenula ^8–10,168. Other chBF corticopetal neurons that project to the mPFC can be found in the medial and lateral preoptic areas ^8,9.

Many studies investigating the role of chNBM corticopetal neurons in mediating emotional memory have utilized the selective cholinergic immunotoxin 192 IgG saporin to induce permanent chNBM corticopetal lesions. ChBF neurons express the p75 receptor ¹⁶⁹ and are they are the only known neuronal group in the BF that express this receptor ^169,170. The 192 IgG saporin toxin does not affect chNBM amygdalopetal neurons and has minimal effects on non-cholinergic neurons ^11,169,170. Thus, when this toxin is infused into the NBM (or the axonal fields of chNBM neurons), selective cholinergic corticopetal neuronal loss can be induced. However, with this treatment residual cholinergic input to the mPFC remains ^171–173 and most studies that have used this toxin did not specifically implicate chNBM neuronal input to the PFC as being critical for emotional memory regulation.

chNBM corticopetal lesions attenuate cued and contextual fear conditioned suppression (i.e. cessation of bar pressing for food induced by presentation of a feared CS) ^171,172,174, without impacting operant performance per se, and animals with chNBM cholinergic lesions can still display operant suppression to footshocks ^171,172,175. Surprisingly, complete chBF or chNBM corticopetal lesions have no effect on cued or contextual fear conditioned freezing ^{11,168,170,174,176} and non-specific excitatory lesions in the NBM have no effects on fear potentiated startle ¹⁷⁷. Two previous studies have reported that contextual fear learning within the inhibitory avoidance paradigm is sensitive to chNBM corticopetal lesions. Within the inhibitory avoidance paradigm, an animal is placed in a light chamber of a passive avoidance box that is connected to a dark chamber by a doorway. Upon entering the dark chamber, the door shuts and animals receive a footshock. One day later, animals are returned to the light chamber of the passive avoidance box and can enter the dark chamber. With conditioning, passive/inhibitory avoidance is observed. In the inhibitory avoidance paradigm, chNBM corticopetal lesions enhance contextual avoidance ^173,178. It should be noted that another study has reported that chNBM corticopetal lesions have no effects on inhibitory avoidance, but instead is critical for the memory enhancing effects of norepinephrine release within the basolateral amygdala ¹⁷⁹.

Results obtained with permanent chBF neural lesions (selective or nonspecific) can be different to results obtained with temporary manipulation of chBF neurons ¹⁸⁰. One study has observed that enhancing cholinergic input to the mPFC enhances contextual fear memory ¹⁸¹ and another study has reported that chNBM input to the auditory cortex is critical for auditory Pavlovian fear conditioning ¹⁸². These studies used either pharmacological or optogenetic methods to manipulate chNBM neurons. With regard to extinction memory, one study has observed that chNBM corticopetal lesions have no impact on extinction memory ¹⁶⁸, but another study has observed that muscarinic antagonism within the IL disrupts extinction memory ¹⁸³. However, it is not clear if this finding specifically implicates chNBM input to the IL, because chBF input to the IL may originate from chMS/DBB neurons ¹⁶⁸.

Taken together studies examining the role of chNBM corticopetal cholinergic neurons on emotional memory are mixed and difficult to interpret in a straightforward manner. Results to date have observed that chNBM corticopetal lesions can disrupt contextual fear memory in one behavioral paradigm (i.e. conditioned suppression), have no effect in other behavioral paradigms (i.e. conditioned freezing and startle), and inhibit avoidance in yet another paradigm (i.e. inhibitory avoidance). Furthermore, inferences made about the role of chNBM corticopetal neurons in emotional memory can vary depending on the methods used to manipulate these neurons (e.g. optogenetic, pharmacological, permanent lesions). ChNBM neurons project to the PFC, which has been implicated in mediating traumatic stress effects on emotional memory and regulation ^{17,33,114,119,124}. ChNBM corticopetal neurons are also intermingled with chNBM amygdalopetal neurons, which are critical for different aspects of fear memory ^184,185. Neurons in the central nucleus of the amygdala project to the NBM ^186–189 and central amygdala neurons are critical for conditioned fear memory ^190,191. Given these findings, one would expect chNBM corticopetal neurons to be critical for fear memory, but unresolved issues remain. These issues may be resolved with further research that utilizes techniques that can specifically target selective groups of chNBM neurons. The existence of rodents that express cre recombinase in cholinergic neurons should facilitate this endeavor, and a number of studies have used this technology already to explore the role of chNBM neurons in emotional memory ^184,185,192. Lastly, selective manipulation of chMS/DBB and preoptic input to the PFC on emotional memory regulation are largely unexplored, but a direction for future research effort.

3.3. Role of chMS/DBB hippocampal projecting neurons in emotional memory

The MS/DBB contains cholinergic neurons that project to the Hipp, but there are neurons in the MS/DBB that also project to the mPFC, midline cortical regions and the medial habenula as well ^147,168. Many studies have used pharmacological approaches to specifically target chMS/DBB neurons that project to the dHipp and other studies have selectively targeted the axons of chMS/DBB neurons in their terminal fields, while others have manipulated chMS/DBB neurons without specifically manipulating a select population of chMS/DBB neurons.

3.3.1. Fear memory

ChMS/DBB input to the Hipp is virtually the exclusive source of ACh in the Hipp ^7–9,147. As a result, pharmacological manipulation of cholinergic receptors in the Hipp directly manipulates chMS/DBB neuronal input to the Hipp. Research measuring ACh levels in the dHipp (e.g. using microdialysis), pharmacologically manipulating cholinergic receptors in the dHipp, or optogenetic activation of chMS/DBB neurons suggest that chMS/DBB hippocampal neurons are critical for mediating contextual fear conditioned freezing. Hipp ACh levels are enhanced during Pavlovian contextual fear conditioning ^193,194 and blockade of mAChRs in the dHipp disrupts acquisition ^195,196 and consolidation ^196,197 of Pavlovian contextual fear conditioning. Enhancing ACh levels ¹⁹⁸ and nAChR activation in the dHipp and vHipp enhances contextual fear memory retrieval ^195,199,200, although nAChR antagonism in the dHipp has no effect on Pavlovian contextual fear conditioning ²⁰¹. Finally, optogenetic stimulation of chMS/DBB neurons enhances contextual fear conditioning ²⁰². Similar results are obtained when contextual fear memory is measured in the inhibitory avoidance paradigm ^{193,203–211}. Together, these findings suggest that chMS/DBB input to the Hipp is critical for contextual fear memory.

The results of studies employing selective chMS/DBB lesions are not consistent with results from pharmacological or optogenetic studies. ChMS/DBB lesions have no effects on contextual Pavlovian fear conditioning ^{170,176,178,212} or inhibitory avoidance ^173,178. However, chMS/DBB lesions disrupt contextual fear memory discrimination ¹⁶⁸. If animals are fear conditioned in one context, then tested for fear memory in a novel context, contextual fear expression is observed in this novel context in rats with chMS/DBB lesions, even though this context was never paired with a UCS ¹⁶⁸. The pool of chMS/DBB neurons that are critical for this phenomena are unclear as neither loss of chMS/DBB input to the mPFC, dHipp, and mPFC and dHipp replicates this effect ²¹³. ChMS/DBB neurons also project to the RSC ^8,9, which is critical for contextual learning and memory ^138,140. While chMS/DBB input to the RSC could be critical for contextual fear memory discrimination, this hypothesis remains untested. Disrupting chMS/DBB input to the vHipp disrupts auditory fear conditioning ²¹⁴, which suggests chMS/DBB input to the vHipp is critical for fear memory.

The reasons for the discrepancies observed when using temporary manipulation in the Hipp vs. chMS/DBB lesions are unclear, but pharmacological, optogenetic, and selective permanent lesion point to a role of chMS/DBB neurons being critical for aspects of contextual fear memory (e.g. acquisition, consolidation, discrimination). Furthermore, chMS/DBB input to the vHipp is critical for cued fear memory

3.3.2. Extinction memory

ChMS/DBB neuronal lesions disrupt acquisition of contextual extinction memory in the fear conditioned freezing paradigm ¹⁷⁶ and induce deficits in acquisition and maintenance of cued extinction memory in the fear conditioned freezing paradigm ¹⁶⁸. The results from these two studies suggests that chMS/DBB neurons have significant roles to play in facilitating acquisition and maintenance of extinction memory. Through what circuits might chMS/DBB neurons facilitate extinction memory? ChMS/DBB neurons project to the IL, which has a significant role in extinction memory ^{43,122,215,216}. Muscarinic antagonism in the IL disrupts extinction memory ¹⁸³, which raises the possibility that chMS/DBB input to the IL is critical for extinction memory. This assertion is supported by the observation that chMS/DBB neurons may be the predominant source of ACh input to the IL ¹⁶⁸. However, removal of cholinergic input to the mPFC has no effect on extinction memory ²¹⁴, which raises questions about the role of chMS/DBB input to the IL in associative extinction memory (i.e. CS-no UCS memory).

Contextual processing has a significant role to play in acquisition and maintenance of fear extinction, where contexts can help determine the meaning of a CS that is initially paired with a UCS (i.e. fear conditioning) and then paired with safety (i.e. fear extinction) ^1–3. In many fear conditioning paradigms, animals are first conditioned in a one context (e.g. fear context), then subjected to extinction training in a distinct context (e.g. extinction context). A number of studies have shown that animals use contextual information to determine when the CS signals fear vs. extinction (i.e. contextual occasion-setting) ^109,110,217 and that the Hipp has a significant role to play in contextual occasion-setting ^109,110,217. ChMS/DBB input to the Hipp could facilitate extinction memory by facilitating contextual occasion-setting, but selective removal of chMS/DBB input to the dHipp or vHipp has no effect on extinction memory ²¹⁴. The RSC is a contextual/spatial system that receives chMS/DBB input, but the role of the RSC in contextual occasion-setting has not been sufficiently explored ¹³⁷. To determine how chMS/DBB neurons facilitate extinction memory, future studies examining the role of chMS/DBB input to the RSC in fear extinction (e.g. acquisition, maintenance, contextual occasion-setting) is needed.

4. Involvement of cholinergic systems in neuroinflammation and PTSD

4.1. Neuroinflammation and PTSD

Inflammation is a normal physiological response against disrupted homeostasis due to infection, injury, or trauma. This response consists of a complex series of immune reactions and is initiated to neutralize invading pathogens, repair injured tissues, and promote wound healing. The onset of inflammation is characterized by activation of immune cells that release pro-/anti-inflammatory mediators, and other key signaling molecules involved in the inflammatory process including tumor necrosis factor (TNF), interleukin (IL)-1, IL-4, IL-6, IL-10, interferon γ (IFN-γ), transforming growth factor β (TGF-β), adhesion molecules, vasoactive mediators, and reactive oxygen species ^218–220. Neuroinflammation refers to inflammatory changes that occur within the central nervous system (CNS) that primarily involves activation of resident brain immune cells such as microglia and astrocytes ^221–223. However, peripheral pro-inflammatory cytokines released through macrophages or other peripheral immune cells can also trigger an inflammatory response in brain tissues by crossing the blood-brain barrier ^224,225.

Prior research suggests a possible link between neuroinflammation and PTSD ^226–228. Clinical studies have shown that PTSD is associated with a dysregulated immune response resulting in neuroinflammation and oxidative stress. In a prospective study involving war zone-deployed marines, pre-existing C-Reactive Protein (CRP, a biomarker of acute inflammatory response) concentrations in the plasma were directly correlated with the occurrence and severity of PTSD three months after deployment ²²⁹. Another study that recruited patients with orthopedic injuries found elevated serum levels of IL-6, IL-8, and TGF-β as compared to health control subjects ²³⁰. Post-traumatic stress symptoms (PTSS) in injured subjects assessed one month later were predicted by higher serum levels of IL-6 and IL-8 and lower levels of TGF-β. Likewise, multiple clinical studies have reported higher blood levels of pro-inflammatory biomarkers and reduced anti-inflammatory cytokines in patients with PTSD as compared to healthy controls ^226,231,235.

Recent neuroimaging studies in PTSD patients demonstrated structural and functional alterations in brain circuits implicated in stress and emotion regulation. A positron emission tomography (PET) study utilizing [¹¹C]PBR28, a PET ligand that binds to 18-kDa translocator protein (TSPO, a microglia biomarker) reported a negative association between PTSD symptoms and reduced TSPO availability in the corticolimbic circuits in patients with PTSD ²³⁶. Moreover, reduced TSPO availability and PTSD severity also correlated with higher CRP levels in the plasma of these patients suggesting that peripheral immune activation in PTSD is associated with microglia dysfunction. Another investigation employed 18F-fluorodeoxyglucose (FDG)-PET to determine the relationship between peripheral inflammatory markers and amygdala activity ²³⁷. Although the magnitude of the FDG signal in the amygdala was not associated with blood levels of pro-inflammatory cytokines, it did correlate with signal in the spleen and bone marrow. Together, these studies support a role for inflammation in driving PTSD.

An increasing body of evidence points towards a modulatory relationship between the buildup of reactive oxidative species (i.e. ROS, oxidative stress) and neuroinflammation in PTSD. Markers of oxidative stress are shown to be upregulated in PTSD patients ^{226,228,238–241}. Additionally, genetic investigations identified single nucleotide polymorphisms of several oxidative stress-linked genes that either moderate the association between PTSD and cortical thickness (ALOX 12 and ALOX 15) or regulate neuroinflammatory and psychpathological manifestations of PTSD (BDNFVal66Met and RORA) ^242,243. Stress-induced neuroinflammatory changes are known to trigger microglia activation which may further facilitate oxidative stress in the brain ^244–246. Thus, neuroinflammation and oxidative stress may act in concert to contribute to PTSD symptomatology ^247,248.

Consistent with clinical studies, preclinical investigations utilizing animal models of traumatic stress show that inflammatory responses may have a causative role to play in mediating traumatic stress effects in neural circuits. Rats exposed to a predator (an animal model of traumatic stress ^71–73) exhibited higher expression of pro-inflammatory cytokines and elevated ROS in the Hipp and the PFC ²⁴⁹. Predator odor stress in mice also increased serum IL-1β levels and anxiety-like behavior via NF-κB signaling, while blockade of this pathway normalized these changes ²⁵⁰. Mice subjected to chronic social defeat (an animal model of traumatic stress ^27,36,251), have transcripts associated with innate immunity and inflammatory signaling upregulated in both the blood and brain and defeated mice exhibited impaired hippocampal neurogenesis ²⁵². Exposure to chronic social defeat stress leads to increased reactive species and activation of microglia, and both processes are implicated in behavioral effects induced by chronic social defeat stress ^253,254. Intracerebroventricular infusions of an IL-1β antagonist prevented stress-enhanced fear learning (an animal model of traumatic stress ^71–73) in rats ²⁵⁵. Using the SPS model of traumatic stress a previous study observed that SPS leads to activation of microglia in the vHipp and this may contribute to extinction retention deficits within the SPS model ²⁵⁶.

It is worthwhile noting two additional mechanisms via which neuroinflammation and the build-up of ROS may alter circuits after traumatic stress exposure. As discussed above, a dysregulated HPA axis and reduced GR sensitivity has been observed with traumatic stress exposure. A reduction in GR activity (or enhanced GR resistance) is suggested to increase the expression of proinflammatory proteins which may cross the blood brain barrier and contribute to neuroinflammation ²⁵⁷. The buildup of reactive species can lead to the induction of apoptosis and neuronal loss ^258–260. SPS is associated with increased activity within apoptotic pathways ^94,261,262 and misfolding of proteins within the endoplasmic reticulum ²⁶², which leads to apoptosis ^251,262. Furthermore, increases in apoptosis has been implicated in extinction retention deficits within the SPS model ⁹⁴.

Although the existing evidence support a close relationship between neuroinflammation, reactive species buildup, and PTSD, more research is needed to identify specific circuits and mechanisms via which traumatic stress-induced changes in neuroinflammation and the buildup of reactive species lead to emotional memory dysregulation.

4.2. Cholinergic regulation of immune response and neuroinflammation: implications for PTSD

As discussed above, chBF systems are critical for emotional memory regulation. Central and peripheral cholinergic systems are also implicated in the modulation of inflammation and neuroimune communication. Expression of nAChRs and mAChRs on immune cells, including T cells, B cells, macrophages, microglia, and astrocytes, and cholinergic stimulation of these receptors has been shown to evoke various downstream functional effects on the immune system ^263–267. In particular, the role of α7 nAChRs in orchestrating the cholinergic anti-inflammatory pathway is well studied. For example, vagus nerve stimulation that increases parasympathetic cholinergic tone attenuated the release of proinflammatory cytokines in a model of septic shock and this effect was mediated via activation of α7 nAChRs on macrophages ^263,268. Likewise, the release of TNFα and IL-1β in response to toll-like receptor-mediated activation of human white blood cells was inhibited by selective stimulation of α7 nAChRs ²⁶⁹. Chronic administration of GAT-107, an allosteric agonist and positive allosteric modulator of α7 nAChR, suppressed T-cell proliferation and the production of pro-inflammatory cytokines while increased the secretion of IL-10 (an anti-inflammatory cytokine) in a mouse model of encephalopathy ²⁷⁰. Furthermore, transgenic α7 nAChR knockout mice displayed higher production and secretion of pro-inflammatory cytokines from antigen-stimulated spleen cells ²⁷¹, and enhanced proliferative response in anti-CD40 antibody-stimulated B-Cells ²⁶⁷. Although α7 nAChRs are traditionally characterized as ionotropic receptors, their anti-inflammatory responses in non-neuronal cells occur in metabotropic ways with downstream signaling involving JAK/STAT and PI3K/Akt pathways that exert negative modulatory effects via nuclear factor NFκB on the expression of pro-inflammatory cytokines ^272,273.

Similar to peripheral immune cells, α7 nAChRs regulate neuroinflammatory changes in the central nervous system. For instance, the activation of α7 nAChRs in cultured rat microglia cells, either via exposure to ACh or nicotine, inhibited the LPS-induced release of the proinflammatory cytokine TNF-α and this effect was driven by the suppression of JNK and p38 MAP Kinase pathway ^274,275. Moreover, in organotypic hippocampal cultures subjected to oxygen and glucose deprivation, exposure to PNU282987 (an α7 nAChR agonist) reduced TNF-α, ROS production, and cell death, and this effect was abolished in microglia depleted cultures ²⁷⁶. The same study also reported that the treatment with PNU282987 reduced infarct size and improved motor skills in a mouse model of stroke demonstrating neuroprotective effects of cholinergic stimulation. Likewise, stimulation of astrocytic α7 nAChRs has been shown to suppress proinflammatory cytokines following immunogenic stimuli such as LPS and IL-1β ^277,278. Furthermore, several studies demonstrated that Aβ-induced and MPTP-induced neuroinflammation and cognitive impairments is ameliorated by stimulation of glial α7 nAChRs implicating central cholinergic system as a key component involved in the regulation of neuroinflammatory changes in age-related neurodegenerative pathologies ^279–284. In addition to the α7 nAChRs, muscarinic receptors and other subtypes of nicotinic receptors have also been suggested to suppress neuroinflammation and oxidative stress by regulating the activity of glial cells ^285,287.

ChBF neurons are sensitive to increased inflammation and oxidative stress. Chronic LPS infusion into the BF of young rats reduces ChAT activity and increases the number of activated microglia and this effect was prevented by systemic administration of anti-inflammatory drugs ²⁸⁸. Given that activated microglia can generate ROS in the local environment ²⁸⁹, the detrimental effects of neuroinflammation on cholinergic neurons could be driven by oxidative stress. Indeed, a recent study that utilized an olfactory bulbectomy model of cholinergic degeneration, found the loss of chMS/DBB neurons and reduced hippocampal ChAT activity was preceded by accumulation of protein-bound carbonyls suggesting an involvement of oxidative stress in cholinergic disruption ²⁹⁰. Additionally, oxidative stress and elevated cytosolic ROS has been shown to disrupt muscarinic and nicotinic cholinergic signaling and downstream pathways ^291,292.

Nerve growth factor is a primary neutrotrophic factor that supports the growth, survival, and maturation of chBF neurons by binding to the high-affinity tropomysin-related kinase A (trkA) receptor ²⁹³. Extensive evidence from in vitro and in vivo studies indicate that oxidative stress or disruption of anti-oxidant defense systems reduces trkA expression resulting in cholinergic dysfunction in Alzheimer’s Disease ^292,294,295. Additionally, our previous research suggested that disruption of NGF-trkA signaling increases the vulnerability of chBF neurons and cognitive capacities in aging model systems ^296,297.

Collectively, the current evidence supports a scenario where neuroinflammation and buildup of reactive species associated with traumatic stress can disrupt chBF circuits leading to emotional memory dysregulation. Further research is warranted to understand the complex relationship between chBF systems, neuroinflammation, reactive species buildup, and traumatic stress.

5. Conclusion

A relatively large body of research has shown that chBF neurons may represent aspects of contextual/spatial systems, these neurons are critical for emotional regulation (e.g. maintenance of fear extinction), and may be sensitive to traumatic stress. Indeed, traumatic stress may disrupt emotional memory regulation by having direct effects on chBF neurons (e.g. neuroinflammation, reactive species buildup, apoptosis). Understanding the cellular and molecular underpinnings of how traumatic stress may compromise chBF neurons may provide insights into the underlying mechanism of PTSD and potentially inform the development of therapeutic approaches. Equally important is the identification of circuits through which chBF neurons facilitate emotional memory regulation. Thus far, a critically unexplored circuit involves chMS/DBB input to the RSC, because the RSC is a contextual/spatial system that may be sensitive to traumatic stress. Future studies examining the potential importance of chMS/DBB input to the RSC in regulating fear and extinction memory and the sensitivity of this circuit to traumatic stress is needed.

Highlights.

• chBF systems project to contextual/spatial systems

• Contextual/spatial systems are sensitive to traumatic stress

• chBF systems are critical for emotional memory regulation

• chBF systems may be sensitive to traumatic stress

• Traumatic stress may disrupt emotional memory regulation via chBF systems