CRF release from a unique subpopulation of accumbal neurons constrains action-outcome acquisition in reward learning, 2023, Eckenwiler et al

[ScoutB]

CRF release from a unique subpopulation of accumbal neurons constrains action-outcome acquisition in reward learning

Spoiler: Authors

Eckenwiler, Elizabeth A.; Ingebretson, Anna E.; Stolley, Jeffrey J.; Fusaro, Maxine A.; Romportl, Alyssa M.; Ross, Jack M.; Petersen, Christopher L.; Kale, Eera M.; Clark, Michael S.; Schattauer, Selena S.; Zweifel, Larry S.; Lemos, Julia C.

Abstract
Background
The nucleus accumbens (NAc) mediates reward learning and motivation. Despite an abundance of neuropeptides, peptidergic neurotransmission from the NAc has not been integrated into current models of reward learning. The existence of a sparse population of neurons containing corticotropin releasing factor (CRF) has been previously documented. Here we provide a comprehensive analysis of their identity and functional role in shaping reward learning.

Methods
To do this, we took a multidisciplinary approach that included florescent in situ hybridization (N mice ≥ 3), tract tracing (N mice = 5), ex vivo electrophysiology (N cells ≥ 30), in vivo calcium imaging with fiber photometry (N mice ≥ 4) and use of viral strategies in transgenic lines to selectively delete CRF peptide from NAc neurons (N mice ≥ 4). Behaviors used were instrumental learning, sucrose preference and spontaneous exploration in an open field.

Results
Here we show that the vast majority of NAc CRF-containing (NAc CRF ) neurons are spiny projection neurons (SPNs) comprised of dopamine D1-, D2- or D1/D2-containing SPNs that primarily project and connect to the ventral pallidum and to a lesser extent the ventral midbrain. As a population, they display mature and immature SPN firing properties. We demonstrate that NAc CRF neurons track reward outcomes during operant reward learning and that CRF release from these neurons acts to constrain initial acquisition of action-outcome learning, and at the same time facilitates flexibility in the face of changing contingencies.

Conclusion
We conclude that CRF release from this sparse population of SPNs is critical for reward learning under normal conditions.

Web | DOI | PMC | PDF | bioRxiv
Previously mentioned: Eccentric medium spiny neuron (eMSN)

 

[ScoutB]

I've had this paper open in a tab for a while to read for CRH reasons, but then I saw it mentions eccentric medium spiny neurons and that ME/CFS Science Blog mentioned it in their first post:

More precisely [this paper] found that some spiny projection neurons in the nucleus accumbens produce CRF and that most of these have more markers typical of eMSN.

Note:
This paper uses the name "spiny projection neurons (SPNs)" as an alternate name for "medium spiny neurons (MSNs)."
And CRF is the same thing as CRH.

The researchers were looking at how these NAc-CRF neurons (i.e. neurons in the NAc = nucleus accumbens that produce CRF) are involved in reward processing and learning.

NAc-CRF neurons are a subpopulation of SPNs that mostly contain D1 and to a lesser extent D2 receptors, are enriched in “eccentric” SPN markers, primarily project to the [ventral pallidum], and are more excitable than SPNs as a whole.

According to wiki: The ventral pallidum is a core component of the reward system which forms part of the limbic loop of the basal ganglia.

My rough understanding:
The researchers knocked-out CRF in these NAc-CRF neurons and checked how this changed the behaviour of the mice. They found, when learning a new task (ie. poke nose in hole, get treat), the CRF-knock-out mice were quicker to learn over the first 3 days, but no different from controls by day 10 of training. On the other hand, the CRF-knock-out mice explored less when exposed to an open field than control mice. Also, when trained on a "reversal learning task" (training the animal to stop doing the thing it had previously been positively reinforced into doing), the CRF-knock-out mice faired much more poorly than controls.

Together, these data indicate a deficit in updating reward contingencies. [...] This suggests that CRF may play a role in preventing hyperfocus (or encouraging flexibility) during certain behaviors that allows for exploration of the environment.

---

Some more quotes I found interesting.

A sparse population of neurons containing the neuropeptide neurotransmitter corticotropin releasing factor (CRF) was initially identified in the rat NAc and later confirmed in mice. These accumbal CRF neurons have been presumed to be a type of interneuron, primarily due to their scarcity. Yet, optogenetic stimulation of this population increases cFos activity [a marker researchers use to observe neuronal activation] in other brain regions such as the ventral tegmental area, ventral pallidum and bednucleus of the stria terminalis (BNST), in addition to the NAc, leaving open the possibility for long-range projections.

CRF has historically been characterized as a peptide that encodes the aversive properties of stress. However, within the corticomesolimbic pathway it may play a more complex role, such as the invigoration of exploratory and appetitive behaviors. In 2021, Baumgartner and colleagues demonstrated that optogenetic stimulation of NAc-CRF neurons in the rat promoted reward consumption and supported intracranial self-stimulation by increasing incentive motivation, not by producing aversion. To our knowledge, this was the first behavior linked to NAc-CRF neurons, yet it was unclear if these results were due to CRF release itself.

A combination of in situ hybridization, viral tract tracing and electrophysiology led us to conclude that NAc-CRF neurons are a unique subpopulation of SPNs that primarily project to the ventral pallidum and are, as a population, more excitable than SPNs. We find that NAc-CRF neurons increase their activity during reward seeking and that this activity scales depending on reward delivery. Importantly, CRF deletion from accumbal neurons facilitates faster acquisition of action outcome reward learning and impairs reversal learning. Together, these results provide the first in-depth transcriptional, anatomical, and electrophysiological analysis of NAc-CRF neurons and identifies a critical role for accumbal CRF release in reward learning.

Do we know if eMSN are more excitable than other medium spiny neurons?

So I guess the claimed overlap is specifically that about 70% of these NAc-CRF have at least one of the 4 markers used to identify eccentric medium spiny neuron. (Haven't looked into these markers though.)

Saunders et al., (2018) identified a third cluster of “eccentric” SPNs in the dorsal striatum that could be distinguished by the presence of one of four markers (Otof, Tac2, Pcdh8 and Cacng5)(23). Using HiPlex techniques, we found that 69 ± 2% of Crh+ cells contained at least one of these markers (Figure 1g,h), compared to 60 ± 1% of all Drd1+ neurons and 46 ± 2% of all Drd2+ neurons (one-way ANOVA with Sidak; main effect F2,21 = 37.91, p < 0.0001). These data indicate that, despite being a relatively small population, CRF cells overrepresent “eccentric” markers compared to SPN populations overall.

 

[ME/CFS Science Blog]

Thanks!

Perhaps this thread would be more suitable for discussing the CRH connection, because it's more of a side issue in the eMSN finding.

I was thinking about a hypothesis taking the following steps:

- Genetic data suggests that eMSN are the main cell type in ME/CFS pathology
- The current study suggests these cells have CRH as co-factor.
- eMSN are constantly overfiring in ME/CFS, creating too much CRH as a side-effect
- They are located in multiple brain regions, so the CRH might spill over and have an effect on other cells and brain regions.
- Too much CRH causes epigenetic silencing of other CRH-producing cells, the most prominent being the PVN cells in the hypothalamus.
- That might explain the findings of the Dutch autopsy studies where they found CRH-producing cells to be depleted in the hypothalamus.

@jnmaciuch pointed out some flaws in the theory, namely about the last two assumptions.

- CRH doesn't travel far, so unclear how it would go from the eMSN to the PVN cells in the hypothalamus
- Unclear if CRH has this inhibitory effect on CRH production

 

[ME/CFS Science Blog]

Unclear if CRH has this inhibitory effect on CRH production

My guess is that a little bit of CRH stimulates production, a bit like seeing other footballers on the pitch as a sign that the game has started. But that too much CRH gives negative feedback as so often happens in biology, a bit like seeing too many football players on the pitch already, so you could sit on the bench.

Most research focuses on the negative feedback from end results of CRH such as cortisol and the effect of stress. But there are also signs of a shorter direct feedback loop from CRH itself.

Found this 2011 review that states:

4.5 Autoregulation of the CRH neuron by CRH and other neuropeptides​

There is evidence that CRH can regulate its own expression in CRH neurons of the PVN. Intracerebroventricular injection of low doses of CRH, devoid of ACTH releasing activity on their own, markedly potentiates ether-stimulated ACTH secretion suggesting a positive feedback effect by CRH [29, 98]. While some of these effects could be indirect via activation of CRH-R in nuclei with interconnections to the PVN, such as other hypothalamic nuclei, the amygdala, locus coeruleus and nucleus of the solitary tract, reports of inhibition of feeding behavior and increases in sympathetic outflow after local application of CRH in the PVN suggest a direct effect [15, 58].

The existence of a direct autoregulatory role for CRH in the CRH neuron is strongly supported by the demonstration that stress induces CRHR1 mRNA and CRH binding in the PVN. CRH receptors are very low in basal conditions but increase markedly following stress [74, 79, 106]. The topographic pattern of CRHR1 expression is stress specific (Fig 3), with increases in the parvocellular division of the PVN after physical-psychological stress (repeated immobilization, ip hypertonic saline injection, acute immune challenge by lipopolysaccharide injection), and in the magnocellular PVN and in the supraoptic nuclei during osmotic stimulation [74]. The stress-specific induction of CRHR1 in CRH neurons suggests that CRH plays an autoregulatory role on its own expression. Supporting this possibility, icv injection of a CRH antagonist attenuates the increases in CRH mRNA in the PVN induced by acute immobilization [49]. CRHR1 is coupled to adenylate cyclase, thus, its activation by locally secreted CRH would provide a source of cyclic AMP, which is necessary for activation of CRH transcription.

THE MOLECULAR PHYSIOLOGY OF CRH NEURONS - PMC

When googling I also found references to this old 1988 study by Calogero et al. which stated:

Exogenous CRH was able to inhibit both unstimulated and neurotransmitter-stimulated iCRH secretion in vitro, a result compatible [...] Neither the source nor the site of action of endogenous CRH involved in an ultrashort-loop negative feedback are known with certainty. The median eminence that contains CRH in nerve terminals may be both as ource of endogenous CRH and a target for the inhibitory effects of CRH.

Multiple feedback regulatory loops upon rat hypothalamic corticotropin-releasing hormone secretion. Potential clinical implications - PubMed

 

[ME/CFS Science Blog]

CRH doesn't travel far, so unclear how it would go from the eMSN to the PVN cells in the hypothalamus

Some speculation about this:

- Perhaps some eMSN are in close enough proximity and well enough connected (e.g. those in the nucleus accumbens) to influence the hypothalamus
- Perhaps eMSN CRH has influences through the ventricles
- Perhaps there is a middle-man that interprets the excess of CRH and then silences the hypothalamus.

All speculation though, and I realize the mechanisms don't fit yet. But feedback loops are so common in biology, and for some we might not fully understand the mechanism yet, so I think it's still something to consider. Especially if the CRH-depletion autopsy findings are confirmed by others.

 

[Jonathan Edwards]

Some speculation about this:

Feedback in neurons is likely to make use of axonal signalling so you just need cells to be connected. All these areas are likely to be connected. CSF might do it but I think less likely.

 

[jnmaciuch]

Perhaps some eMSN are in close enough proximity and well enough connected (e.g. those in the nucleus accumbens) to influence the hypothalamus
- Perhaps eMSN CRH has influences through the ventricles
- Perhaps there is a middle-man that interprets the excess of CRH and then silences the hypothalamus.

These are more plausible. Though you also have the additional issue of the intracellular negative feedback loop that shuts down CRH production—I posted at least one paper on this in another thread, I think my CRH/sickness behavior one.

TLDR the same cascades that trigger CRH transcription also induce CREM, which is an inhibitory TF for CRH transcription. This is the mechanism that prevents CRH neurons from overproducing CRH even with high levels of stimulation during e.g. the cytokines that signal directly to the PVN during infection.

So you’d have to explain how that negative feedback loop avoids getting triggered in eMSNs. It’s possible there’s some epigenetic weirdness happening that affects the inhibitory TF locus but not the CRH locus. In which case, we have to ask whether something happening in the eMSNs is causal for the autopsy study findings, or whether the two lines of evidence are more likely to be downstream of a third weird thing. That could very well be the case if, for example, the genetic hits all seem to point to eMSNs simply because the hits are related to some general process like glutaminergic signaling and eMSNs just happen to be a highly glutaminergic population (using glutaminergic as a placeholder concept here). And since the number of signaling cascades that effect CRH signaling are plenty, the possibility of a “third connection” seems more likely just numerically.

(And, to clarify, I don’t bring these points up to discourage the theorizing here. it’s definitely worthwhile and useful. I’m just taking on the role of the one poking holes)