The main findings of this work were that BDNF mRNA and protein levels during antidepressant treatments did not correlate, and that induction of mature protein preceded that of mRNA. The measurement of BDNF mRNA expression showed an overall similar temporal profile of induction with the two drugs. RBX acted somewhat faster, inducing BDNF mRNA already at week 1 in both areas, while FLX induced BDNF from week 2. The shape of this induction was also somewhat different; FLX progressively increased the expression up to week 3, while RBX induced BDNF in two distinct waves. After a precocious induction at week 1, the expression of BDNF was not further increased (P/FC) or was reduced (HPC) by RBX between week 1 and 2. A second, more pronounced, wave of BDNF induction by RBX was observed between week 2 and 3. Overall, RBX acted faster on BDNF transcription but FLX outsized RBX with regard to the maximum extent of mRNA induction. Interestingly, the washout week also differently affected the outcome of the two drug treatments in HPC, because it sharply reduced the effect of FLX in both areas and of RBX in P/FC, while in HPC of RBX-treated rats BDNF mRNA level was unchanged at week 3+1. In a companion study , we previously found that 1 washout week entirely abolished the activation of CREB induced by 3 weeks of RBX, but not of FLX, in line with the longer half-life of norFLX, the active FLX metabolite. This discrepancy between the effects of washout on activation of CREB vs. BDNF expression may underscore the complex regulation of the BDNF gene, that contains several different regulatory elements (other than CRE) at the level of promoters .
Surprisingly, the temporal profile of BDNF protein expression in HPC was largely inconsistent with that of mRNA expression (compare Fig. 1A with Fig. 2C). Although with FLX a significant increase of mRNA was not observed until week 2, in the same rats mature BDNF (14 kDa) was already significantly elevated at week 1 and reached maximal level at week 2, when mRNA level was still far below the maximal level attained at week 3. Instead, the level of pro-BDNF (32 kDa) in HPC was significantly elevated only at week 3. A possible explanation for this discrepancy is that, in the first several days of FLX treatment, BDNF mRNA already available is rapidly translated and the proBDNF produced is also rapidly processed to mature BDNF, suggesting that FLX has a more rapid effect on BDNF translation/processing than on the transcription of BDNF gene. Although more in line with the early induction of mRNA by RBX, the profile of induction of mature BDNF by this second drug was also inconsistent with the profile of mRNA induction, which reached maximal level only at week 3 (as for FLX). With RBX the level of mature BDNF was already maximal at week 1 and remained steadily elevated up to week 3, suggesting again a rapid effect of the drug on BDNF translation/processing, perhaps accelerated by the induction of new mRNA at week 1 by RBX.
In P/FC the effects of the two drugs on BDNF translation were different from each other. Maximal level of mature BDNF was induced by RBX already at week 1, as in the HPC, although the extent of increase was somewhat smaller than in HPC. Again, the temporal profile of the protein was inconsistent with that of mRNA, which reached maximal level only at week 3. By contrast, no mature BDNF was detected at any time point during FLX treatment, although moderate levels of proBDNF were attained, suggesting that in P/FC FLX was able to induce transcription of BDNF but little translation and no processing of the protein. Previous studies did not find induction of BDNF protein by FLX or other antidepressants in frontal cortex [15, 17]. This substantial lack of induction of BDNF protein by FLX in P/FC would suggest that downstream effects of this drug on synaptic plasticity are more powerful in HPC than in prefrontal/frontal areas of cerebral cortex.
The present results are in line with the several reports that previously showed up-regulation of BDNF mRNA induced by chronic antidepressant treatments, often increasing with the length of treatment [16, 25]. We also found here, as in two previous studies [16, 17], a lack of correlation between mRNA and protein, to the point that in HPC detectable levels of BDNF protein were raised faster than those of mRNA by the drug treatment. Overall, the differences in levels of BDNF protein we detected are somewhat bigger than in previous studies using the same drugs; this could be due to the fact that we measured here for the first time BDNF levels in a time-course treatment by Western analysis rather than by ELISA, a method that does not allow to distinguish between pro- and mature BDNF.
The present results would suggest that two antidepressants, FLX and RBX, have a particularly fast effect on posttranscriptional regulation of BDNF in HPC (RBX also in P/FC). It has already been suggested that BDNF synthesis could be regulated at posttranscriptional level , as found for other proteins . First, it is known that selected BDNF transcripts are located at dendrites, a mechanism that allows rapid regulation of translational control ; we recently found that in rats treated with FLX or RBX more BDNF mRNA is located at dendrites in CA3 hippocampal area (Tongiorgi E, Popoli M; unpublished). Second, it was recently suggested that microRNAs exert a robust effect on BDNF levels in mature human prefrontal cortex, participating in posttranscriptional fine tuning of BDNF expression . Third, it was recently shown that cleavage of pro- to mature BDNF is regulated by neuronal activity ; changing the processing of pro-BDNF by proteolytic enzymes could be an additional way whereby antidepressants rapidly increase the production of mature BDNF. This could explain why in HPC we found a faster rise of mature BDNF compared to pro-BDNF (compare Fig. 2A with Fig. 2C). Investigation of these mechanisms in subchronic and chronic treatments with antidepressants is warranted.
Finally, the finding that BDNF, a major readout effector, is rapidly elevated in HPC by antidepressant treatment may suggest that later downstream events are required to elicit systemic and behavioral effects of antidepressants. If BDNF production is faster than thus far envisaged, perhaps the correlates of the therapeutic effects should be searched in the actual cellular/molecular changes induced by the action of BDNF, such as the modifications in synaptic plasticity and the increment of neurogenesis. Further work is required to characterize these events.