We report here that downregulation of folliculin leads to lower levels of phospho-ribosomal protein S6 both in cell culture and in mice with heterozygous inactivation of Bhd
. This is both an expected and a surprising result. We had previously found that in S. pombe
, the Bhd and Tsc homologs function in opposing pathways. Since downregulation of the TSC proteins leads to Tor activation in S. pombe
as well as in mammalian cells, our finding that downregulation of BHD leads to mTOR inhibition was consistent with the S. pombe
results. However, the fact that BHD downregulation leads to TORC1 inhibition in vitro
and in mice carrying targeted inactivation of Bhd
is surprising for multiple reasons. First, since TSC and BHD patients develop hamartomatous skin tumors, lung cysts, and renal tumors, one might expect the TSC proteins and folliculin to have similar effects on mTOR activity. Second, Baba et al.
found that BHD patient-derived UOK257 cells lacking folliculin had higher levels of mTOR activation than UOK257 cells re-expressing folliculin (Baba et al., 2006
). Third, two groups have reported that mice with targeted inactivation of Bhd
in the kidney (via a cadherin 16 [KSP]-Cre transgene) develop massively cystic kidneys with increased phosphorylation of ribosomal protein S6 (Baba et al., 2008
; Chen et al., 2008
), and that renal carcinomas from BHD patients have weak to moderate staining of phospho-mTOR (Baba et al., 2008
However, alongside these data suggesting that loss of BHD results in mTOR activation, there are other indications that the relationship between folliculin and mTOR is complex. For example, Baba et al.
found that cells lacking BHD had lower levels of mTOR activation than cells re-expressing BHD in a specific cellular condition (deprivation of amino acids), and Takagi et al.
recently reported that in HeLa cells, siRNA downregulation of BHD causes a reduction in phospho-S6 Kinase (Thr389), although the serum conditions for these experiments were not specified (Takagi et al., 2008
The reasons for these different effects of folliculin on mTOR are not yet clear. One possibility is that the earliest impact of BHD loss results in mTOR inhibition, consistent with the S. pombe
data, and that subsequent events associated with tumor progression activate mTOR. Consistent with this notion, Baba et al.
found that multiple kinase cascades, including Raf/MEK/Erk, p90Rsk, and Akt, were hyperactive in the cystic kidneys of mice with KSP-Cre driven inactivation of Bhd
(Baba et al., 2008
). ERK, Rsk1, and Akt all directly phosphorylate and inactivate tuberin, thereby leading to mTOR activation (Ballif et al., 2005
; Cai et al., 2006
; Ma et al., 2005
; Manning et al., 2002
; Rolfe et al., 2005
; Roux et al., 2004
; Tee et al., 2003a
). Therefore, it is possible that in vivo
, mTOR activation occurs not as a primary consequence of BHD mutations, but due to activation of kinases upstream of the TSC/Rheb/mTOR pathway. This could explain the low levels of phospho-S6 in the cysts and tumors in our mouse model, all of which were relatively small and likely represent early lesions.
It is also possible that the observed differences in mTOR activation reflect differences between loss of BHD expression in siRNA-treated cells and mutational inactivation of BHD in UOK257 cells, which have an insertion at nucleotide 1733, leading to a frameshift mutation (Yang et al., 2008
). Interestingly, nearly all identified BHD
mutations are predicted to truncate prematurely the folliculin protein (Khoo et al., 2002
; Leter et al., 2008
; Nickerson et al., 2002
; Schmidt et al., 2005
; Vocke et al., 2005
). Most identified germline mutations are either frameshift or nonsense mutations, and roughly 44% occur in a “hot spot” for BHD
frameshift mutations in a poly-C tract within exon 11 (Schmidt et al., 2005
). It is possible that truncation of folliculin removes a carboxy-terminal region involved in mTOR inhibition or leads to a dominant negative effect, while loss of the entire protein removes an amino-terminal region required for mTOR activation. Finally, it is possible that folliculin regulates mTOR in a cell type or context specific manner that is not fully recapitulated by either the cell culture experiments or the targeted inactivation in vivo
. However, we note that we found that BHD knockdown resulted in a reduction of phospho-S6 (Ser235/236) in three cell types: U251, HEK293, and HK-2.
The mice reported here are the first Bhd
mouse model that develops solid renal tumors. Previous animal models of BHD include the German shepherd dog, which develops hereditary multifocal renal cystadenocarcinoma and nodular dermatofibrosis (Lingaas et al., 2003
), the Nihon rat, which develops renal cell adenomas and carcinomas, both cystic and solid in nature (Hino et al., 2001
; Kouchi et al., 2006
; Okimoto et al., 2004a
; Okimoto et al., 2004b
), and the previously discussed mice with Cre-driven inactivation of Bhd
in the kidney, that die of renal failure by 3 weeks of age, some of which develop cystic renal cell carcinomas (Baba et al., 2008
; Chen et al., 2008
). The solid tumors in our Bhd+/-
mice were composed of oncocytic cells with variable nuclear pleomorphism, reminiscent of the spectrum of tumors found in BHD patients, which include chromophobe renal tumors and oncocytic hybrid tumors (Murakami et al., 2007
; Pavlovich et al., 2005
; Pavlovich et al., 2002
). In addition to providing a model of BHD renal carcinogenesis, this is also, to our knowledge, the first mouse model of oncocytic and chromophobe renal tumors. While BHD
mutations are not frequent in sporadic chromophobe renal carcinomas (Nagy et al., 2004
), Warren et al.
observed weak to absent BHD
mRNA levels in renal oncocytomas, chromophobe renal cancer, and oncocytic hybrid tumors which is in contrast to the high BHD
mRNA levels observed in other cancers including breast, ovarian and prostate (Warren et al., 2004
). Intriguingly, we have observed weak or absent phospho-S6 levels in sporadic oncocytomas and chromophobe renal cell carcinomas, in contrast to high levels of phospho-S6 in the majority of sporadic clear cell renal carcinomas (Robb et al., 2007
In conclusion, we report here that loss of folliculin leads to mTOR pathway inhibition both in vitro and in vivo, consistent with data from S. pombe in which the Bhd and Tsc pathways have opposing roles in amino acid homeostasis. We speculate that the most “ancient” evolutionarily conserved function of folliculin is related to mTOR activation. In more complex organisms folliculin may have additional context specific regulatory functions, highlighting the potential complexity of mTOR regulation in cells carrying BHD mutations. Understanding the role of folliculin in mTOR regulation may elucidate not only the pathogenesis of BHD syndrome, but also the pathogenesis of sporadic chromophobe and oncocytic-hybrid renal carcinomas and the normal homeostatic regulation of mTOR activity. Finally, these studies may contribute to the development of targeted therapeutic strategies for BHD patients, which are urgently needed.