The recent Review article by Heilig et al. (Pharmacogenetic approaches to the treatment of alcohol addiction. Nature Rev. Neurosci. 12, 670–684 (2011))1 expertly discussed strategies towards the development of therapeutics for alcoholism. However, we noticed a serious omission in their discussion of the corticotropin-releasing factor (CRF) system.

The authors described evidence supporting a role for CRF receptor 1 (CRF1) in alcohol-related behavioural traits, concluding that phenotypic screening may improve the clinical efficacy of CRF1 antagonists. However, they did not acknowledge that in addition to CRF and CRF1, this system includes three CRF-related ligands (the urocortins (UCNs): UCN1, UCN2 and UCN3), an additional receptor (CRF2) and the CRF binding protein (CRFBP)2. UCN2 and UCN3 bind CRF2 selectively, whereas UCN1 binds both CRF receptors and the CRFBP with greater affinities than does CRF itself2 (Fig. 1).

Figure 1: Relationships among corticotropin-releasing factor (CRF), urocortins and their targets.
figure 1

CRF has high affinity for CRF receptor 1 (CRF1) and CRF binding protein (CRFBP), urocortin 2 (UCN2) and UCN3 are selective for CRF receptor 2 (CRF2), and UCN1 has high affinity for both receptors and the CRFBP. Solid lines indicate high-affinity binding, dashed lines indicate species-dependent affinity.

Full size image

Although such information might be considered extraneous, we would like to correct a mistake in the authors' Review1 that highlights precisely why it is necessary to acknowledge all CRF system components. Heilig et al. claimed that “blockade of stress-induced relapse is in part mediated by CRF1 blockade in the median raphe nucleus (MRN)”, citing a 2002 study3. In fact, those experiments used the non-selective4 antagonist D-Phe CRF12-41, leaving in question whether this effect was mediated by CRF1 or CRF2. On the surface, the error appears to be minor. However, because the authors neglected to mention all CRF system components, they not only falsely implicated CRF1, but also implied that the underlying ligand must be CRF.

This example illustrates a broader problem in the literature. Our analysis of the most recent 120 items retrieved by a PubMed search for CRF (or CRH) system involvement in alcoholism or addiction (excluding articles that focused solely on the hypothalamic–pituitary–adrenal axis) found that only 34.7% of these articles acknowledged UCN peptides. Furthermore, 53.3% of these articles implied a role for CRF without providing evidence against a role for UCNs.

We also identified several cases in which authors applied a ligand exogenously and inferred that the same ligand must mediate the effect endogenously. For example, the 2002 study discussed above found that intra-MRN CRF infusions reinstated alcohol-seeking, and the authors inferred that this mechanism mediated alcohol-seeking endogenously3. We should point out that many brain areas receive co-innervation by multiple ligands of the CRF system2,5,6, complicating this interpretation.

Heilig et al. probably excluded CRF2 and urocortins because orally available CRF2-specific drugs do not exist. However, we might argue that the slower development of CRF2-targeted therapies is a by-product of the issues outlined above. In fact, CRF2-selective agonists are potent inhibitors of alcohol intake7,8, and their therapeutic potential has been discussed9. Aside from the authors' exclusion of CRF2, we remind them that UCN1 also exhibits high affinity for CRF1. Furthermore, genetic deletion of UCN1 dampened alcohol preference and alcohol-induced reward10, justifying the inclusion of UCNs in future conversation.

It is not our intent to diminish the role of CRF in alcohol-related behavioural traits.Rather, we hope that investigators will give careful thought to the endogenous mechanisms by which the CRF system influences behaviour and consider all suspects in order to generate a nuanced dissection of CRF system involvement in stress- and addiction-related disorders.