The severity of behavioural symptoms in FTD is linked to the loss of GABRQ‐expressing VENs and pyramidal neurons

Abstract Aims The loss of von Economo neurons (VENs) and GABA receptor subunit theta (GABRQ) containing neurons is linked to early changes in social–emotional cognition and is seen in frontotemporal dementia (FTD) due to C9orf72 repeat expansion. We investigate the vulnerability of VENs and GABRQ‐expressing neurons in sporadic and genetic forms of FTD with different underlying molecular pathology and their association with the presence and severity of behavioural symptoms. Methods We quantified VENs and GABRQ‐immunopositive neurons in the anterior cingulate cortex (ACC) in FTD with underlying TDP43 (FTLD‐TDP) (n = 34), tau (FTLD‐tau) (n = 24) or FUS (FTLD‐FUS) (n = 8) pathology, neurologically healthy controls (n = 12) and Alzheimer's disease (AD) (n = 7). Second, we quantified VENs and the GABRQ‐expressing population in relation to presence of behavioural symptoms in the first years of disease onset. Results The number of VENs and GABRQ‐expressing neurons and the ratio of VENs and GABRQ‐expressing neurons over total Layer 5 neuronal population decreased in FTLD‐TDP and FTLD‐FUS, but not in FTLD‐tau, compared to control and AD. The severity of early behavioural symptoms in all donors correlated with a lower VEN and GABRQ neuronal count. Conclusion We show that in FTD, a loss of VENs together with GABRQ‐expressing pyramidal neurons is associated with TDP43 and FUS pathology. No significant loss was found in donors with FTLD‐tau pathology; however, this could be due to the specific MAPT mutation studied and small sporadic FTLD‐tau sample size. Overall, we show the GABRQ‐expressing population correlates with behavioural changes and suggest they are key in modulating behaviour in FTD.


INTRODUCTION
Behavioural changes are one of the most prominent clinical symptoms in frontotemporal dementia (FTD). The majority of FTD patients develop one or multiple behavioural symptoms during their disease course, of which the early unifying core feature is a lack of social and emotional cognition. Symptoms include behavioural disinhibition, apathy or inertia, loss of empathy, perseverative or compulsive behaviour and hyperorality or dietary changes [1]. Patients presenting with at least three of these symptoms in the first 3 years of disease onset are diagnosed with the behavioural variant of frontotemporal dementia (bvFTD) [1]. Another clinical variant of FTD is the language variant, named primary progressive aphasia (PPA), and includes semantic variant (svPPA) and a non-fluent variant (nfvPPA). Patients with FTD can also present a spectrum of other symptoms, including motor neuron disease (MND)/ amyotrophic lateral sclerosis (ALS) [2]. Psychotic symptoms, such as delusions and hallucinations, are also commonly seen in bvFTD patients, during or prior to the onset of dementia [3]. The pathology of FTD, termed frontotemporal lobar degeneration (FTLD), is characterised by three subclasses of aggregated misfolded proteins: prevalent, whereas patients with FTD due to FUS aggregation are described as only sporadic [4][5][6][7][8][9].
We recently showed selective loss of the GABA receptor subunit theta (GABRQ) expressing cortical neurons in the anterior cingulate cortex (ACC) in bvFTD patients due to a C9orf72 repeat expansion [10]. The GABRQ-expressing neurons include the von Economo neurons (VENs), of which selective vulnerability has been linked to FTD in the earliest stages of the disease [11][12][13][14][15].
VENs have a unique morphology and are found mainly in Layer 5 of the human ACC and frontoinsular cortex (FI) [16]. They are distinguished from pyramidal neurons by their large bipolar cell body and thick dendrites [17]. VENs have been identified with a similar regional distribution in highly social mammals, such as primates, cetaceans and elephants, but are not found in common laboratory animals, such as mice and rats [18][19][20].
In addition, the GABRQ-expressing population is also absent in mice [21]. Currently, it is not known whether the loss of GABRQ-expressing neurons extends to other genetic and sporadic pathological forms of FTLD, including FTLD-TDP, FTLD-tau and FTLD-FUS and how this relates to the early behavioural symptoms.
Here, we aim to investigate whether there is a loss of GABRQexpressing neurons in the ACC in the different pathological and genetic subtypes of FTD. In addition, we want to establish whether there is a link between the number of GABRQ-expressing neurons and the early manifestation of behavioural symptoms.

Subjects
Post-mortem brain tissue was obtained from the Netherlands Brain • Loss of VENs and GABRQ-expressing neurons in the ACC is observed in donors with FTLD-TDP and FTLD-FUS molecular subclasses.
• No significant loss of VENs and GABRQ-expressing neurons was found in donors with FTLD-tau pathology.
or extensive pathology throughout the cortex that was not related to their main pathological diagnosis) were excluded from the study.

Immunohistochemical procedures and GABRQ quantification
Samples from the ACC were collected and processed using immunohistochemistry as described previously [10]. Briefly, the right hemisphere was fixed for 4 weeks in PFA. The ACC was dissected perpendicular to the corpus callosum, caudal to the genu and embedded in paraffin. Ten micrometre-thick sequential sections were cut, and immunohistochemical analysis was performed with  (Table S2).

Clinical information
For all donors, extensive clinical information was available. The neurological symptoms were evaluated by a neurologist (M.S.). Donors presented varying phenotypes, including mild behavioural symptoms and prominent language symptoms, to more prominent behavioural symptoms (see Table S1 for clinical diagnosis). The behavioural symptoms were scored using the clinical criteria of Rascovsky et al. [1]. The framework consists of six main categories each with two or three subcategories. If a behavioural symptom was present in the first 3 years of disease onset, it was scored as present (1) or absent (0). If symptoms were not explicitly mentioned and could not be extrapolated from the records, they were considered absent. The sum of the behavioural scores resulted in a behavioural score ranging from 0-14.

Statistics
Statistics were performed using SPSS 22. Pearson correlation was used to assess relation between clinical profile and scores for neurodegeneration, and differences between groups were assessed using one-way ANOVA with Tukey's post hoc analysis.

FTLD-TDP and FTLD-FUS but not FTLD-tau donors show loss of GABRQ-expressing neurons
We quantified the number of GABRQ-expressing pyramidal neurons in the ACC (Figure 1). See Table S2

Relation between VENs and GABRQ-expressing neurons and behavioural symptoms
We investigated the presence of behavioural symptoms in the first RNA translation [26][27][28][29]. In FTLD-TDP and FTLD-FUS, TDP43 and FUS, respectively, are depleted from the nucleus and aggregate in the cytoplasm. Sequestration of these proteins from the nucleus has profound effects on the cell that could ultimately result in neuronal death [30,31]. Based on our data, it is likely that the GABRQ-expressing neuron and VEN loss are a consequence of upstream modulators of hnRNPs.
Early behavioural symptoms correlate with the number of GABRQ-expressing neurons donor's disease course [32], we can assume that the loss of the GABRQ-expressing neuronal population in the ACC occurs proportionally to the disease duration and therefore focussing on the initial stages can provide insight into the anatomical basis of symptoms. In addition, previous work has shown that the FI also shows selective loss of VENs and fork cells in the early stages FTD [14], and for future research, it would be interesting to explore the GABRQneuronal population in this region.
One important consideration is that this study is limited to the expression of one marker, GABRQ. In our data, we see that the percentage of GABRQ-negative VENs over total VENs increases in FTLD. This could indicate that either the GABRQ-negative VENs are less likely to degenerate or that some GABRQ-expression is lost in FTLD without neurodegeneration. If the latter is the case, this would also reflect on the pyramidal population, leading to an overestimation of neuronal loss in FTLD. However, the possibility remains that the neurons that lost expression of GABRQ will also have altered functioning. Future functional studies should investigate this in more detail.
To conclude, TDP43 and FUS pathology show a reduction of the GABRQ-expressing neuronal population in the ACC. Donors from our FTLD-tau group (sporadic and MAPT P301L mutation) showed no selective reduction of the GABRQ-expressing neuronal population, but this has to be explored across more pathological and clinical variants of FTLD-tau. The severity of behavioural symptoms is correlated with the GABRQ-expressing neuronal population, suggesting that this neuronal population is a key modulator of behaviour in FTD.

ACKNOWLEDGEMENT
This study was supported by a grant from the Memorabel ZonMw (733050507).

CONFLICT OF INTEREST
The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS
AAD and PGP designed the study and wrote the manuscript. PGP, MS and AAD performed the experiments and quantification of the data.
PGP and AAD analysed the data. JCvS, ABS, YALP and JJHM provided intellectual contribution and participated in discussion. FHB, BDCB, AJMR and NBB were responsible for the autopsy, storage of the post-mortem material and clinical and neuropathological evaluation. NBB also assisted in the selection of suitable tissue from its bank. All authors read and approved the final manuscript.

ETHICS STATEMENT
All procedures performed in the study were in accordance with the ethical standards of Amsterdam University Medical Centre location VUmc. Ethical approval for the NBB procedures and forms was given by the Medical Ethics Committee of the VU University Medical Centre (Amsterdam, Netherlands). Informed consent and tissue collection were also carried out in accordance with the Code of conduct for Brain Banking and Declaration of Helsinki [33].

INFORMED CONSENT
All donors gave informed consent for autopsy, storage and use of their tissue and anonymised clinical and neuropathological data for research purposes.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/nan.12798.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.