Birt-Hogg-Dubè can be thought of as a ciliopathy. Ciliopathies are a set of rare diseases resulting from defects in cilia; membrane-protruding organelles that provide either propulsion (motile cilia) or a sensory function (primary cilia) in mammalian cells. Like BHD, cystic lesions in the kidneys are a major manifestation of ciliopathies such as polycystic kidney disease and Bardet-Biedl syndrome, which result from mutations in the ciliary proteins that regulate cilia formation and retraction, a process known as ciliogenesis. In 2013 Luijten et al., reported that FLCN localizes to the primary cilium and influences various parameters of ciliogenesis. Furthermore, knock-down of FLCN in polarized epithelial cells of the kidney interferes with canonical Wnt signaling, an important cilia-associated pathway, and distorts the ability of these cells to organize into 3D spheroid cultures.

In 2016, Zhong et al. published an interesting study that further examined the role of FLCN in cilia, this time in the context of flow stress and mTOR signaling. Primary cilia of kidney cells experience flow stress when fluid passes over the cell surface causing the cilia to bend, and this is transduced into a cellular response by the suppression of mTORC1 signaling in an AMPK-dependent manner (Boehlke et al., 2013). Excitingly, the paper by Zhong et al., implicates a role for FLCN in this process, thereby providing another mechanism by which mTOR signaling could be dysregulated in BHD.

Zhong et al. first established that an interaction between FLCN with KIF3A  (a subunit of the cilia motor protein Kinesin-2) occurred in a cilia-dependent manner and was absent in cell populations lacking cilia, and therefore, all subsequent experiments were conduct on ciliated cells. This is consistent with the findings of Dodding et al., 2011, which predicted that FLCN houses a WD motif that enables it to bind Kinesisn-1, a cytoplasmic motor protein that is associated with intracellular transport. Next the study sought to confirm the observation that flow stress on ciliated cells causes a suppression of mTOR signaling by measuring the levels of P-S6, a product of mTOR phosphorylation. In normal cells, flow stress caused a down-regulation of P-S6, which is consistent with a reduction of mTOR activity, however, in FLCN knockout (UOK-257) and knock-down cells (siRNA) this P-S6 down-regulation failed to occur in response to the stress. The investigators also looked at the P-AMPK signaling that is upstream of the pathway and that acts as a repressor of mTOR signaling. In normal cells, the flow stress increased the expression of P-AMPK (which in turn decreases mTOR activity) but this increase was less apparent in the FLCN deficient knockdown cells. This behavior was also mirrored by P-TSC2, a signaling molecule down stream of P-AMPK (see diagram). Taken together, this is strong evidence that FLCN is required to down-regulate mTORC1 in flow stress signals, the very pathway strongly implicated in FLCN tumor suppressor role (Tsun et al., 2013).

Next, using another series of intricate co-immunoprecipitation experiments, the authors demonstrated that FLCN recruits the protein LKB1  to activate AMPK (see diagram below) This is consistent with the findings of Goncharova et al. 2014, who report that FLCN controls AMPK activity via LKB1 for cell survival processes in the lung alveoli. Like previous experiments, this process was cilia dependent and only occurred in response to fluid stress. When FLCN was knocked down the interaction of LKB1 and AMPK was significantly reduced. These experiments were nicely complemented with imaging and FLCN knockdown experiments that demonstrated that FLCN was required for the localisation of both p-AMPK and LKB-1 to the basal bodies, the microtubule structure present at the base of the cilia.

In summary, Zhong et al., have identified a novel role for FLCN in the mechanosensing function of cilia when exposed to fluid stress. A reduction in FLCN levels interferes with the fluid stress–induced suppression of mTORC1 and causes prolonged activation of this pathway. Notably, hyperexcitation of mTOR pathways is well documented in cancer biology and this study is consistent with of a wealth of recent literature that implicates a role for cilia within cancer and tumorigenesis. Unfortunately, the paper did not look to rescue the effects of FLCN knockdown/knockout on fluid stress signaling, which would have given them a good opportunity to test the effects of FLCN mutations on the process. Nevertheless, the study opens some interesting follow-up opportunities, including understanding if the process of mechanosensing involved in BHD cyst formation and tumourigenesis .

Image from Zhong et al., 2016.

Tuberous sclerosis complex (TSC) is an autosomal dominant disease that causes a range of symptoms. Like BHD, it can cause benign growths in several organs including the lungs, kidney and skin. However, TSC patients can also suffer from neurological problems, including intellectual disability, and it is a leading genetic cause of autism and epilepsy. The disease is caused by mutations in the TSC1 and TSC2 genes, which are components of the mTOR pathway. Mutations in TSC1 and TSC2 cause dysregulation of mTOR signaling and this is thought to contribute to the pathophysiology of TSC (Franz et al., 2017).

Seizures in TSC patients are thought to be partly caused by cortical tubers (brain lesions with distorted cellular architectures) and these are surgically removed in TSC cases where the epilepsy is unresponsive to medication. A recent paper from Mills et al., 2017, examined the gene expression profiles of several cortical tubers from TSC patients to look for clues into the mechanisms that underlie the neuropathology of the disease. The authors measured levels of RNA in cortical tubers from TSC patients (10 surgically removed and 2 autopsied) and cortical samples from age-matched controls who died from unrelated illnesses. Excitingly, the authors simultaneously measured both the protein-coding RNA and non-coding small RNAs (RNA that is not translated but regulates gene expression) using RNA Seq, to produce the first parallel coding and non-coding transcriptome analysis for TSC.

Mills et al. found 438 protein-encoding genes were differentially expressed in cortical tubers (measured by fold change); 269 over-expressed and 169 under-expressed. The authors noted that many of the more highly-transcribed genes were associated with the adaptive and innate immune responses including genes coding for the complement system. This is consistent with other studies that have implicated inflammation genes and the inflammatory response in the pathophysiology of TSC and that have made an association between the activation of the complement system and seizures (Vezanni et al., 2013). Interestingly, no significant differences in gene expression were observed between patients with TSC1 (4 samples) and TSC2 (8 samples) mutations. However, since the analysis was performed on a mixed population of cells, in both the TSC and control samples, it does not reveal which cell types within the tubers are driving these changes. Therefore, the investigators next looked to see which of the differentially expressed genes were exclusive to certain cell types of the cortex. By examining the 269 more highly-transcribed genes in healthy human cortex, it was found that 23 were exclusive to microglia, 3 to oligodendrocytes, 5 to neurons and 8 to astrocytes. The 32 genes that were specific to the astrocytes and microglia were components of the complement system, which is consistent with the role of these cells in neuroinflammation. The study did not report any RNA-seq comparisons between the different cell types of TSC and controls samples, which would have considerably raised the resolution of this analysis.

Next, the authors analysed the small non-coding RNAs and identified differential expression of 991 transcripts, including many different small RNA species. They found that members of the microRNA (miRNA) 34 family were amongst the most over-transcribed in the TSC samples. miRNAs silence gene expression by targeting protein coding mRNA transcripts and labeling them for degradation instead of allowing them to translate into proteins. In situ hybridization targeting of the miR-34 family was performed on TSC and control tissues and, consistent with the RNA-Seq data, showed an increase in expression of miR-34a-5p and miR-34b-5p in the TSC tubers.

To unite the RNA-Seq results of both the coding and small non-coding RNAs, the investigators applied a weighted gene co-expression network approach (WGCNA) to the datasets. This data mining method seeks to examine the implications of gene expression changes in the context of gene modules (a network of genes with related or interconnected functions) instead of at the individual gene level. The WGCNA analysis showed that many of the 169 genes that were transcribed less in the TSC coding RNA-Seq, and whose relevance could not be determined at an individual level, were predominantly involved in neurogenesis or glutamate receptor signaling. Moreover, several miR34 targets were also predicted to be involved in these two processes, suggesting that they could be affected by miRNA species that are differentially expressed in TSC. With this in mind, the authors sought to briefly explore the effect of miR-34 overexpression on neurogenesis in an in vitro hippocampal neuronal culture assay. Here, mimic miR-34b-5p molecules were overexpressed in the neurons and when compared to a non-targeting control, the miR-34 mimic samples had longer and more numerous neurites and larger cell somas. Accordingly, the authors concluded that miR-34 expression could be a contributing factor to the neuropathology of cortical tubers.

Figure from Mills et al., 2017.

Mills et al, 2017 is an ambitious study that uses powerful transcriptome analysis to highlight the complexity of gene regulation processes involved in the pathophysiology of TSC. The analysis of different RNA species, alongside the application of advanced bioinformatic techniques, has much potential as an ‘epigenetic-driven’ strategy for identifying new targets for therapies. The study of rare diseases such as BHD, would greatly benefit from such insight.

A haploid human genome comprises of 23 chromosomes, roughly 20,000 genes and is encoded by approximately 3 million base pairs (the building blocks of DNA – A, T, C and G). The full code for life’s blueprint for making a human, was revealed back in 2003 through an international collaboration known as The Human Genome Project (HGP). Considered as one of the greatest scientific achievements of the century, the HGP gave scientists never-before-seen insight into human genetics and sparked a wider interest in the study of human genomics.

Taken alone, a single human genome provides limited diagnostic information for a disease of unknown cause: the power of human genomic studies lies in being able to make comparisons. In 2012, another international effort known as the 1000 Genomes Project was completed, this time with many more genomes to study, and the project gave scientists a fresh look at the extent of human genetic variation (Nature, 2012). The 1000 Genomes project greatly benefited from faster and cheaper DNA sequencing technology (Mardis, 2011), a field that is still advancing today (Scott, 2016). Whereas the HGP took 13 years and nearly $2.7 billion to obtain the sequence of one mosaiced human genome (samples were provided by multiple donors), today a person’s genome can be sequenced for around $1,000 and in about a day (Scott, 2016). The same year the 1000 Genomes Project announced it had completed 1092 genome sequences, another exciting and ambitious genomics initiative was launched, this time entirely in the UK; the 100,000 Genomes Project.

The 100,000 Genome Project is funded by the Department of Health and it aims to annex a genomic medicine program to the NHS while also providing a platform for UK genomic scientific investigation. At this point, you may be wondering what the 100,000 genomes project has to do with BHD? Well, the 100,000 genomes project largely focuses on two areas, cancer and rare diseases, two fields that are relevant to the BHD community. In fact, one of the criteria for rare disease patients wanting to volunteer is familial (family-history associated) primary spontaneous pneumothorax. Recurrent pneumothorax can be suggestive of BHD especially if it runs in the family, and BHD patients are believed to be 50 times more likely to experience a pneumothorax than the general population (Zbar et al. 2002).

Unfortunately, anyone who is known to have folliculin (FLCN) mutations will not be able to sign up to the 100,000 Genome Project because it is only recruiting patients with primary pneumothorax, (whereas BHD patients suffer from secondary pneumothorax associated with an underlying pulmonary cause). There is a requirement that prospective volunteers with suspected familial pulmonary primary pneumothorax be screened for FLCN mutations prior to entering the study, to rule out BHD as the potential cause. The BHD Foundation thinks that there are some positive messages to be taken away from this. Firstly, BHD awareness has clearly infiltrated the study of rare diseases well, which is important given that it is so rare, even by rare disease standards. Currently there are roughly 600 families worldwide with a BHD diagnosis (published BHD families) but BHD is certainly underdiagnosed, perhaps significantly. Secondly, the requirement of FLCN screening before entering the study will hopefully increase the diagnosis rate of people with a FLCN mutation: quicker and earlier diagnosis of BHD is important because it allows for better management and surveillance for all possible BHD symptoms, including renal cell carcinoma. 

It is well known that mutations in the folliculin gene (FLCN) cause BHD syndrome; a disorder that can result in fibrofolliculomas, lung and kidney cysts, pneumothorax and renal cancers. The FLCN protein has long been suspected of acting as a tumour suppressor and has been identified as a modulator of the AMPK-mTOR pathway (Baba et al. 2006), a signaling cascade involved in cell metabolic processes and cancer. Unfortunately, the precise mechanism by which FLCN exerts its tumour suppressor role remains unknown. This is addressed by a recent research paper by Laviolette et al. 2017, which finds that FLCN acts as a GTPase activating protein (GAP) for RAB7A and proposes that normal FLCN-RAB7A interactions may suppress tumour growth by modulating the receptor tyrosine kinase (RTK) EGFR (epidermal growth factor receptor).

Elevated EGFR signaling has already been heavily associated with cancer mechanisms (reviewed in Jones et al. 2014) and consistent with this, the paper demonstrates increased levels of EGFR and EFGR signaling in FLCN defective cells (human follicular thyroid carcinoma cell line) relative to FLCN wildtype cells. Upon stimulation with EGF ligand, higher levels of phosphorylated EGFR (and of phosphorylated downstream effector molecules pERK and pS6) were detected in both FLCN deficient cells and those expressing tumour-associated FLCN mutations. In addition to this, high levels of EGFR-activated signaling molecules were detected (via immunohistochemistry) in the renal cysts and tumours from a conditional FLCN knockout mouse and in various types of renal cell carcinomas biopsied from BHD patients. Taken together, this is evidence for the involvement of heightened EGFR signaling in the tumorigenesis associated with FLCN dysfunction. The implications of this finding are that a loss of FLCN function may result in increased mTOR activity (which is downstream of the EGFR pathway) and the paper states that this is contrary to studies claiming FLCN activates mTOR (Petit et al. 2013). However, the authors acknowledge that many of those experiment were performed under different stimulating conditions (amino acid vs EGF excitation) and raises the hypothesis that the interactions and dynamics of FLCN could vary according to the stimulus.

This study identifies a novel interaction between FLCN and GTPase RAB7A by co-immunoprecipitation experiments and mass spectrometry. The FLCN-RAB7A interaction was investigated in more detail using RAB7A proteins locked in several conformations, including a constitutively active GTP-bound form (induced by a Q67L mutation) and a dominant negative form (T22N mutation). Here, preferential precipitation of both the constitutively active and wildtype forms of RAB7A suggested that FLCN could be acting as a GTPase activating protein (GAP) that promotes hydrolysing activity in RAB7A. This hypothesis was confirmed in a functional assay that revealed FLCN enhances the GTPase activity of RAB7A and that GAP activity was lost in FLCN when it housed the tumour-associated mutation K508R.

FLCN is known to regulate the spatial distribution of lysosomes via regulating Rab-RILP interactions in a process that is dependent on nutrient status (Starling et al. 2016). Furthermore, Rab7A is known to regulate the endocytic recycling and lysosomal degradation of EGFRs (Ceresa et al. 2006, Rush et al. 2013). Therefore, the authors of this paper looked to examine how FLCN, as a Rab7A GAP, effects EGFR trafficking and signaling. Intricate imaging experiments showed that FLCN-deficient cells (renal carcinoma cell line) had slower endocytic trafficking of EGFRs compared to normal cells and this manifested as an accumulation of EGFR in early endosomes, alongside a delayed transition into late endosomes. The authors noted that delayed endocytic trafficking could result in more recycled and less degraded EGFR, causing an increase and prolonging of EGFR signaling. Accordingly, this provides a potential explanation for the observed heightened EGFR signaling in FLCN deficient cells and BHD tumours (see above).

Having implicated the involvement of Rab7A and EGFR signaling in FLCN’s tumour suppressor role, the authors examined the mechanism in vivo. Afatinib, an inhibitor of EGFR signaling, was used to treat FLCN-negative induced tumour xenografts in mice. Consistent with the general findings of this paper, that there is heighten EGFR signaling in FLCN defective cells, the study found that Afatinib significantly hindered the growth of these tumours compared to the control-treated mice but, notably, there was no tumour regression. Although the focus of this paper was on EGFR signaling, the authors highlight the importance of investigating the role of other RTK pathways in FLCN-associated renal cancers. Furthermore, it would be interesting to see with further research and study, whether or not Afatinib holds any promise as a future treatment of BHD.

Pulmonary cysts are a frequent clinical manifestation of Birt-Hogg-Dubé syndrome and are assumed to be the cause of pneumothorax in BHD patients (Johannesma et al. 2014f). Pneumothorax is when lungs collapse due to air leaking into the pleural cavity and in BHD patients the number of cysts is strongly correlated to the incidence of recurrent pneumothorax (Johannesma et al., 2014f). Since many lung diseases carry an increased risk of pneumothorax, it is hard to identify potential BHD cases when pulmonary symptoms present alone.

Takegahara et al. 2017, reports on a 40-year-old man who had experienced recurrent bilateral pneumothorax from the age of 22. At the time of this report, X-ray showed that the patient had a pneumothorax in his left lung, although it came to light that he had previously experienced three pneumothoraces on his right side. His case of pneumothorax did not subside following thoracostomy tube insertion and the decision was made to perform bilateral thoracoscopic surgery after a chest CT revealed the presence of cysts in both lungs. This enabled surgeons to remove the larger cyst suspected to be responsible for the problem, seal any lesions and performed a pleural covering with a polyglycolic sheet with the intention of reducing the recurrence of pneumothorax.

Despite the patient only exhibiting pneumothorax, clinicians suspected there to be an underlying genetic cause such as BHD, because of several observations including the patient’s family history of pneumothorax (mother and maternal grandmother). Subsequent genetic testing of the affected family members revealed a duplication mutation (c.1285dupC) in exon 11 of the FLCN gene, confirming them to be cases of BHD syndrome. Further examinations revealed no fibrofolliculomas or renal symptoms but long-term monitoring was recommended including for further pneumothoraces. Such a diagnosis is important because it can lead to early recognition and the better treatment of life-threatening BHD symptoms such as renal cell carcinoma, should they develop at a later age.