Introduction
Stress and glucocorticoid dysregulation have been linked extensively to depression1. For example, major stressors precede most depressive illnesses2; pathological excess of glucocorticoids can cause depression3 e.g. Cushing’s syndrome4; and, up to half of depressives have an excess of glucocorticoids3.
The hypothalamic-pituitary-adrenal axis (HPA) is responsible for the initiation of the glucocorticoid stress response5 and is regulated, in part, by the hippocampus6. The hippocampus is densely populated with stress hormone receptors and negatively feedbacks on the paraventricular nucleus (PVN) to regulate the HPA axis5. Neurogenesis occurs in the ventricular-subventricular zone (V-SVZ) and subgranular zone (SGZ) of the hippocampal dentate gyrus throughout life7. Stress and glucocorticoids profoundly inhibit adult hippocampal neurogenesis (AHN)8.
The neurogenesis theory posits that depression ultimately arises from impaired AHN and restoration of AHN leads to recovery9. Evidence supporting this theory includes: rodents models of childhood neglect (a risk factor for both depression) show a depressive-like phenotype and impaired AHN10; AHN is increased by virtually all antidepressant therapies11,12; and, AHN is decreased following acute and chronic stress13. In a seminal paper, Santarelli et al. also showed the behavioural effects of antidepressants in mice is dependent on AHN14.
Snyder et al. have used a transgenic experimental mice model to discover a mechanistic link between neurogenesis, stress and depression.
MAJOR FINDINGS & CRITICAL EVALUATION
Transgenic mice that expressed HSV-TK under the control of a GFAP promotor were first created and treated with valganciclovir (v-TK). TK renders mitotic cells sensitive to valganciclovir15 and GFAP is expressed on neuronal precursors16,17. Antiviral valganciclovir treatment kills mitotic cells whilst sparing post-mitotic cells. Two negative controls were used to validate results – WT mice treated with valganciclovir (v-WT) and TK-mice not treated with the antiviral. Thus demonstrating that any effects seen in v-TK mice compared to controls were not due to antiviral treatment or expression of the transgene. Using confocal microscopy, the group demonstrated that the number of GFAP+ astrocytes was unchanged by treatment with valganciclovir in v-TK mice. This not only mirrors the case in humans, but also demonstrates the specificity of the model in only killing mitotic cells, thus improving the transferability and reliability of the model respectively. Using markers of immaturity and proliferation (DCX and BrdU respectively), the group showed that immature proliferating neurones were reduced in v-TK by >99% compared to both negative controls. The strong agreement between both markers further increases reliability. The group could therefore justly conclude that v-TK mice had specific loss of adult neurogenesis. However, experiments using this model do not indicate the location of the inhibition.
Snyder et al.’s next aim was to evaluate the role of neurogenesis in HPA regulation and glucocorticoid fluctuation. They did this by measuring serum corticosterone by radioimmunoassay at the onset of the light and dark phase AND following restraint. v-TK mice had elevated corticosterone relative to v-WT mice 30 minutes after stress and the levels of corticosterone in dexamethasone-injected v-TK mice were significantly higher than in control. From this, the group justly concluded that impaired adult neurogenesis led to a higher basal HPA tone and HPA dysregulation. To add, v-WT and control v-WT mice had equivalent levels of corticosterone at the onset of the light and dark phase. A 24h profile of corticosterone levels, as opposed to only measuring at the onset of the light and dark phase, would have provided stronger evidence to conclude that neurogenesis doesn’t disrupt glucocorticoid circadian fluctuation.
The group then tested the location of neurogenesis inhibition in their experimental mice model. They exposed v-WT and v-TK mice to prolonged isoflurane anaesthesia, an external stressor shown not to be influenced by the hippocampus, and showed similar elevations in corticosterone between v-TK mice and controls. There was also no reduced cell birth in the hypothalamic PVN of v-TK mice. From this, they concluded that HPA dysregulation in v-TK mice wasn’t global but instead hippocampus-specific. However, it would have been helpful to check corticosterone levels following other physiological stressors that are not regulated by the hippocampus such as inflammation18. To further support this conclusion, they used X-irradiation to inhibit hippocampal neurogenesis (but spare SVZ neurones, vice versa) – the hippocampal-irradiated mice had significantly elevated corticosterone recovery from restraint stress and there was no relation between SVZ neurogenesis inhibition and corticosterone response in irradiated mice. Although it is unlikely that irradiation has a focal effect on hippocampal neurogenesis alone and it is unclear from the experiments the extent of neurones damaged in both hippocampal and SVZ irradiation, the group did provided evidence to suggest that neurogenesis in a small subset of neurones in the hippocampus was driving HPA dysregulation.
Utilising several behavioural tests used in the screening of antidepressants, Snyder et al. examined the role of adult neurogenesis in behavioural responses to stress. Using the NST (novelty-suppressed feeding test), they showed similar feeding latencies between v-TK mice and v-WT mice. However, 30m of restraint stress prior to the test significantly increased feeding latency in v-TK mice. However, there was no significant difference between control and v-TK mice at baseline or following restraint stress in the EPM. To assess the role of neurogenesis in hedonic behaviour, v-TK and control mice underwent a sucrose preference test. v-TK mice had decreased preference at baseline but restraint stress didn’t significantly alter the v-TK mice preference.. Finally, using the FST (forced swim test), they showed neurogenesis-deficient mice became immobile more rapidly at baseline, also indicative of a depressive phenotype as with the sucrose phenotype test. The group also changed the water in the FST, thus eliminating the confounding of thermoregulation. Restraint stress reduced latency to become immobile in WT mice but didn’t significantly affect v-TK mice. This is in disagreement with the final conclusion that adult hippocampal neurogenesis buffers the effect of stress, but suggests that neurogenesis-deficiency can cause a depressive phenotype.
Overall, the group provided evidence to justly conclude that inhibition of AHN led to depressive-like phenotype and HPA dysregulation which can be potentiated by inescapable stress.
LIMITATIONS AND CLINICAL IMPLICATIONS
Neurogenesis-deficient mice had a depressive phenotype at baseline in only two out of four behavioural tests, which limits the strength of the conclusion that neurogenesis deficiency leads to depression. Also, although the behavioural tests are useful antidepressant screening tools, the clinical efficacy of antidepressants is controversial19 and approximately 30% of patients never respond to therapy20. Perhaps the results are not transferrable in humans at all as the behavioural tests do not fully recapitulate the psychiatric syndrome. To add, it is unclear from the experiments whether ablating the same number of mature dentate neurones could also potentiate the stress response. Additionally, the experiments need to be repeated in non-human primates to ascertain how transferable this is.
AN ALTERNATIVE INTERPRETATION
The behavioural tests used focus on quantitatively assaying potentially adaptive features following stress e.g. from an evolutionary perspective, it makes sense to conserve energy and remain immobile when placed in a situation you cannot escape from. The findings could therefore show a protective or adaptive measure against environmental stressors. Further experiments need to be done in order categorise this as the depressive behaviour and impaired stress response seen in humans.
FIELD IMPACT
The group broadened the knowledge in the field by providing evidence to suggest that a small subset of hippocampal dentate gyrus neurones regulates HPA axis and reconfirmation of the role of adult neurogenesis in depressive illness. These findings support the neurogenesis theory of depression, allowing stress to act as a link, thus addressing some criticisms of the hypothesis.
BEYOND 2011
Since the publication of the paper, there have been exciting developments in the field of stress, neurogenesis and depression. Firstly, there are now methodologically more rigorous ways of researching AHN – namely optogenetics and pharmacogenetics. For example, Seib et al., use mice deficient in Dkk1 (Wnt signalling agonist) to inhibit neurogenesis and demonstrate reduced sucrose preference and immobility in tail suspension at baseline21.
Due to the unethical nature of assessing AHN in humans in vivo, it remains to be known whether increasing AHN alone cures depression or ablating AHN in humans causes depression. Post-mortem studies do show a link between antidepressants and hippocampal volume, as well as depression and granule cell number23,24. Additionally, cancer patients receiving therapy that ablates AHN also have increased levels of depression and anxiety25, however several confounding factors are likely to influence results. Recently, P7C3, a neuroprotective drug that enhances hippocampal neurogenesis, has shown promising antidepressant efficacy in mice22. Despite these findings, more direct evidence is needed to determine the mechanism linking stress, neurogenesis and depression in humans.
A study published earlier this month, demonstrated that AHN falls to almost undetectable levels in the adult human26. This finding suggests that the evidence provided in decades of mammalian studies may not be transferable to humans. However, future studies will need to replicate these results before such a claim can be made.