Depression and Neurogenesis -- Annie Bryant

Both Santarelli et al. and Bessa et al. concur that hippocampal atrophy is central in the behavioral manifestations of depression, and that the hippocampus is a major substrate for antidepressant drug (AD) action. However, it seems (at first) that these two papers are otherwise at odds with each other: Santarelli et al. posit that ADs act primarily via neurogenesis, while Bessa et al. contend that they act primarily via neuronal plasticity and also target the prefrontal cortex (PFC). 

Santarelli et al. demonstrated that ADs increased hippocampal neurogenesis and decreased latency in the novelty-suppressed feeding (NSF) test from baseline in WT mice, as did the 5HT1A receptor agonist 8-OH-DPAT. 5HT1A knockout (KO) mice still responded to TCAs but not to the SSRI fluoxetine or to 8-OH-DPAT, indicating that fluoxetine likely interacts with this receptor to mediate neurogenesis and anti-depressive effects. They also show that ADs didn’t decrease feeding latency in mice who received x-irradiation to the SGZ of the dentate gyrus, from which they conclude ADs can’t act without neurogenesis. At face value these conclusions appear sound, as prior studies have shown that neurogenesis is decreased after stress and restored following chronic AD treatment, and that the 5HT1A receptor has been implicated in modulating mood and anxiety. However, their scope of behavioral and molecular tests was narrow and didn’t really allow for exploration of possible alternatives, and they seem to conflate “temporal coincidence” (as Bessa et al. put it) with direct causality. I’m surprised they didn’t incorporate a depression model, as I would imagine a depression-naïve brain would respond quite differently to ADs than a brain exposed to depression for a protracted interval. There are also potential problems with their method of blocking neurogenesis, namely inflammation and the fact that irradiation requires a delay before ADs can be administered – and Santarelli et al. began ADs concurrently with irradiation.

Six years later, Bessa et al. drew upon reports of reduced hippocampus and PFC volume in human depression to hypothesize that changes in neuronal circuitry are responsible for depression, and accordingly, the amelioration of depressive behavior via ADs. They sought to first settle the issue of whether neurogenesis really is critical for AD-induced improvements in multiple behavior paradigms – namely, anhedonia (sucrose preference), learned helplessness (forced swim test), and anxiety (NSF test). Using the cytostatic agent methylazoxymethanol acetate (MAM) in tandem with a chronic mild stress (CMS) protocol and multiple ADs, they demonstrated that ADs reduced anhedonia and learned helplessness following CMS, even when neurogenesis is blocked via MAM. Bessa et al. went on to explore changes in synaptic plasticity and dendritic arborization in the context of CMS and AD treatment using qPCR and Golgi staining. They chose the synaptic plasticity markers Ncam1 and Syn1 for qPCR analysis, as NCAM1 is involved in neurite outgrowth and SYN1 is involved in axonogenesis and synaptogenesis. While CMS reduced expression of these markers in the hippocampus and PFC, treatment with ADs restored levels to baseline, with or without MAM. Additionally, Golgi staining demonstrated atrophy of apical dendrites (dendrites emerging from pyramidal cells) in the hippocampus as well as layers II-III of the PFC, which receive inputs from the hippocampus. Dendritic atrophy and volumetric reduction were both effectively reversed by treatment with ADs.  Furthermore, they positively correlated the extent of depressive symptoms in behavioral testing with the degree of dendritic atrophy. Based on their findings of dendritic atrophy and reduced synaptic plasticity following CMS, Bessa et al. propose that disturbances to the hippocampus-PFC circuit give rise to behavioral symptoms of depression, and that ADs act upon these mechanisms to restore behavioral homeostasis.

I think the “missing link” connecting these papers lies in Bessa et al.’s finding that MAM increased latency to feed in the NSF test across all groups, including those not exposed to stress and/or treated with ADs. From this they conclude that continuous neurogenesis is necessary to produce the anxiolytic effects of ADs, which lines up with Santarelli et al.’s findings. However, since Santarelli et al. didn’t look at other behavioral paradigms, they extrapolated their findings into an overly generalized conclusion that neurogenesis is necessary for all behavior-alleviating effects of antidepressants. By incorporating multiple tests that encompassed several components of depressive behavior, Bessa et al. were able to specify that neurogenesis is only necessary for this one aspect of AD efficacy, but not for other depressive behaviors like anhedonia and learned helplessness.

Bessa et al. suggest that the new neurons generated following AD treatment may be specifically involved in modulating anxiety. In the same vein, Santarelli et al. cite a study (van Praag et al. 2002), in which researchers showed that young granule neurons generated from the SGZ are functionally integrated into the existing hippocampal circuit. Van Praag et al. propose that these new neurons serve to either replace dying neurons or to enhance plasticity. Both papers from this week assert that neurogenesis and neuronal remodeling both result from AD treatment; I wonder if these are components of a bi-phasic response to the ADs. Perhaps neurons first reorganize in the hippocampus and PFC to their pre-depression configuration, mitigating anhedonia and learned helplessness. Then, within weeks, neurons proliferate in the SGZ and are integrated into this new circuit, mitigating anxiety. A time-course study outlining the respective emergence of synaptic plasticity and neurogenesis would be very informative to better determine their temporal relationship.