DISCUSSION
In this study, we show that FKBP5 mRNA expression turns on at birth in
brain areas involved in MDD, such as the hippocampal formation,
amygdala, and striatum. According to previous work describing the link
between FKBP5, HPA axis and MDD, these data allow us to postulate that
FKBP5 expression is vulnerable to ELS, and we propose a key role of
FBKP5 in MDD pathogenesis.
The role of FKBP5 for ELS and HPA axis. FKBP5 regulates GR
sensitivity, a fact which is intertwined with the HPA hypothesis of MDD.
For patients with MDD, FKBP5 polymorphisms have been associated with the
extent of GR signaling dysregulation (Menke et al., 2013). DNA
methylation alterations in stress-regulating genes such as FKBP5, are,
in fact, the cause of long-lasting effects of ELS (Wiechmann et al.,
2019). Taking ELS into account illuminates the circular relationship
between FKBP5 and the physiological stress response. Exposure to ELS –
which is essentially a challenge to the glucocorticoid system - creates
lifelong changes within the FKBP5 gene regulatory regions (Wiechmann et
al., 2019). These stress-induced changes to FKBP5 DNA methylation
further impact an individual’s stress regulation system, disposing them
to developing a disorder such as MDD (Wiechmann et al., 2019). This
mechanism has been demonstrated in detail in rodents, implicating the
hippocampus for long-term FKBP5 alterations brought on by ELS, which
increase vulnerability to depressive-like behaviors (Xu et al., 2019).
The role of FKBP5 for immunoregulation. Despite our focus on a
psychiatric disorder, we must not ignore the rest of the body simply
because it is not the brain, especially considering the role of FKBP5 in
immune function. For humans, FKBP5 is broadly expressed throughout
multiple bodily systems (unsurprisingly due to its essential cellular
functions), but it is most strongly expressed in immune system tissues.
According to GeneCards (Safran et al., 2022), FKBP5 is overexpressed in
lymph nodes and peripheral blood mononuclear cells. As mentioned, FKBP5
functions as a co-chaperone for a heat shock protein. These are
molecules that, as a part of the innate immune system, can signal cell
damage to recruit other immune cells to the site of insult (Gong et al.,
2020). Such activation promotes inflammation to aid with regenerative
processes, but excess may result in further disease (Gong et al., 2020),
including psychiatric disorders like MDD. It is also worth mentioning
that ELS causes persistent alterations in peripheral inflammation
(Baumeister et al., 2016), thus tying together the stress and immune
sides of the story.
Putting it all together: the role of FKBP5 for antidepressant
treatment. The HPA axis and immune systems individually affect
antidepressant efficacy. For instance, there are numerous reports that
ELS (which impairs the HPA stress regulation) substantially reduces
antidepressant response (Nanni et al., 2012; Williams et al., 2016;
Menke et al., 2021). Moreover, inflammatory markers decrease after MDD
treatment, and non-responders have higher inflammation at baseline
(Strawbridge et al., 2015). Some studies also indicate inflammation is
associated with poor antidepressant response (Vogelzangs et al., 2014).
These are exciting findings, and could have broad implications for
predicting individual treatment success or even creating new
pharmaceutical options. Yet the crucial insight here is the connection
provided by the gene FKBP5, with a particular influence from ELS, which
offers a more comprehensive explanation of a complex and often variable
conundrum. FKBP5 has been associated with unipolar depression (Zobel et
al., 2010) and even treatment response for depression (Lekman et al.,
2008). Severe life events significantly alter FKBP5 mRNA expression,
which tends to normalize after four weeks of antidepressants (Menke et
al., 2021). ELS-induced alterations of FKBP5 expression even change the
expression of other glucocorticoid-related genes (Yeo et al., 2017). For
its GR-modulating properties and consequent connection to the HPA axis,
FKBP5 polymorphisms have been studied and found to be associated with
faster response to antidepressants (Binder et al., 2004). Lastly and
most strikingly, FKBP5 deletion in the mouse results in antidepressant
behavior, with no change to other motor or cognitive functions (O’Leary
et al., 2011). Such a direct intervention to FKBP5 itself, and with such
precise effects, strongly supports our idea that this gene is a vital
functional link between two large-scale processes which mediate the
success or failure of antidepressants. This evidence aligns with our
findings of FKBP5 expression, anatomically and throughout human and NHP
brain development (Fig. 1), as well as with the finding (Fig. 3) that
FKBP5 expression is upregulated in MDD both clinical (Matosin et al.,
2023) and preclinical settings (Laine et al., 2018).
Limitations and future directions. This research effort is
limited by its observational nature. Nevertheless, it is important to
point out that our results, initially obtained from supposedly control
subjects (Fig. 1B) and extrapolated to their potential relevance for
MDD, were then validated in clinical and preclinical datasets (Fig. 3).
Further, the low number of subjects analyzed at each time point in the
developmental study (Fig. 1B) might have reduced the power of our data.
This is an intrinsic limitation of the Allen Brain Atlases, which are
built on a relatively low number of experimental subjects. However, the
finding that FKBP5 mRNA expression is upregulated in both human MDD
(Matosin et al., 2023) and mouse models of the disorder (Laine et al.,
2018) (see also Fig. 3) strengthens our hypotheses that FKBP5
dysfunction is linked to two major causes of MDD and to the response to
antidepressant treatment. In order to verify whether FKBP5 expression is
normalized by antidepressant treatment, future work might be extended to
human and mouse gene expression datasets containing control subjects and
subjects treated with antidepressant drugs.
We also acknowledge that studying the expression of one single gene
related to MDD does not allow us to unravel the transcriptional
complexity of this disorder, which clearly has a multifactorial origin.
However, our long (> 5 years) experience with this type of
interactive teaching activity showed us that this simplified view (“one
gene – one disease”) allows students to generate an adequate amount of
data for their final exam and, most importantly, to mechanistically
address the function of a specific gene in the pathogenesis of a complex
brain disease.
Considerations on practical coursework. In this study, we
described an example of how a teaching approach engaging students in
practical coursework may contribute to generate scientifically
meaningful results and novel data-driven hypotheses. Most importantly,
we think this approach positively contributes to the students’
educational experience. First, students learn to collect and analyze
data from publicly accessible sources, becoming familiar with open
science methods; this provides students with new skills that might
become useful for real research projects in the laboratory. In addition,
being asked to write a research article for their final exam, students
develop or improve scientific writing skills in preparation of their
Master’s thesis. Lastly, our Master Degree in Cognitive Sciences does
not include bioinformatic classes, though students do acquire advanced
skills in programming and data analysis. Thus, our approach gives
students the opportunity to become familiar with bioinformatic
techniques.
We acknowledge that a randomization strategy with course delivery would
help to quantitatively address the advantages of this educational
experience. However, such a strategy is far beyond the scope of this
approach: our aim is to develop an interactive teaching approach that
helps students develop new skills. We thus think it might be interesting
to share our experience with the neuroscience community. Our experience
shows that classes aiming at involving students in projects requiring
data collection and analysis from publicly available gene expression
databases (such as the Allen Brain Atlases) may effectively result in
new ideas for neuroscience research.
Acknowledgements. The authors thank the staff of the Master in
Cognitive Sciences (CIMeC, University of Trento) for excellent
administrative assistance, Prof. Enrico Domenici (CIBIO, University of
Trento) for insightful comments, and Prof. Elizabeth Binder (Department
of Translational Research in Psychiatry, Max-Planck Institute of
Psychiatry, Munich, Germany) for allowing data re-analysis from Matosin
et al., 2023 (Fig. 1a).
Competing Interests. The authors declare no competing financial
interests.
Author Contributions: HS wrote the paper and contributed to
making the figures; AD analyzed the data and contributed to making the
figures; LB supervised students’ activities, analyzed the data, and
contributed to making the figures; YB designed the project, analyzed
data, contributed to making the figures, and edited the manuscript. This
research was not supported by any specific grant from any funding
agency.
Data accessibility. This study made use of freely available
data from Allen Brain Atlas and Gene Expression Omnibus. The code used
to analyze data is freely available online on Zenodo and GitHub (see
details in the Materials and Methods section).
Abbreviations. AMG, amygdaloid complex; CB, cerebellum; Ctrl,
control; CRH, corticotropin-releasing hormone; CSDS, chronic social
defeat stress; CTXsp, cortical subplate; dHPC, dorsal hippocampus; ELS,
early life stress; GEO, Gene Expression Omnibus database; GR,
glucocorticoid receptor; HPA, hypothalamus-pituitary-adrenal axis; HPC,
hippocampus; HPF hippocampal formation; HY, hypothalamus; MB, midbrain;
MDD, major depressive disorder; mPFC, medial prefrontal cortex; MY,
medulla oblongata; NHP, non-human primate; OLF, olfactory areas; OMIM,
Online Mendelian Inheritance in Man database; P, pons; PAL, globus
pallidus; PFC, prefrontal cortex; RNAseq, RNA sequencing; ROI, region of
interest; STR, striatum; TH, thalamus; vHPC, ventral hippocampus; V1,
primary visual cortex.