of more than 5000 proteins stably expressed in
human embryonic kidney (HEK)–293T cells (20).
This analysis revealed C7orf60, a previously unstudied protein, as a putative interaction partner of
all known components of GATOR1 (Depdc5, Nprl3,
Nprl2) and KICSTOR (Kaptin, ITFG2, C12orf66,
SZT2). For reasons described below, we renamed
C7orf60 as SAMTOR (S-adenosylmethionine sensor upstream of m TORC1).
Using an antibody against SAMTOR to probe
anti-FLAG immunoprecipitates prepared from
cells having endogenously Flag-tagged components of GATOR1 (Depdc5) or GATOR2 (WDR59)
or stably expressing a KICSTOR component (
Flag-Kaptin), we verified that SAMTOR coimmunoprecipitated with GATOR1 and KICSTOR, but not
GATOR2 (Fig. 1A). Moreover, transiently expressed SAMTOR coimmunoprecipitated with
endogenous GATOR1 and KICS TOR, as detected
by the presence of their Nprl3 and SZT2 components, respectively. Loss of a component of GATOR1
or KICSTOR, but not of GATOR2, severely reduced
the interaction of SAMTOR with KICSTOR or
GATOR1, respectively (Fig. 1B). Furthermore, overexpressed GATOR1 coimmunoprecipitated SAMTOR
only when KICSTOR was coexpressed (fig. S1A).
Thus, SAMTOR binds to the supercomplex of
GATOR1 and KICS TOR, and both complexes are
required for the interaction to occur (Fig. 1C).
Orthologs of SAMTOR are encoded in the ge-
nomes of vertebrates and some invertebrates, such
as Drosophila melanogaster. We could not iden-
tify SAMTOR orthologs in Caenorhabditis elegans
or Saccharomyces cerevisiae (Fig. 1D).
To determine whether SAMTOR regulates
m TORC1 signaling, we overexpressed SAMTOR
in HEK-293T cells and monitored the phosphorylation at Thr389 of S6 kinase 1 (S6K1), a canonical
m TORC1 substrate. SAMTOR expression suppressed
m TORC1 signaling in a dose-dependent fashion
(Fig. 2A), establishing SAMTOR as a negative
regulator of the pathway. Amino acids activate
m TORC1 by promoting its localization to the lysosomal surface (4, 8). Consistent with SAMTOR
inhibiting the amino acid sensing pathway upstream of m TORC1, overexpression of green
fluorescent protein (GFP)–tagged SAMTOR displaced m TOR from lysosomes to an extent similar to that seen with GFP-Sestrin2, an inhibitor
of GATOR2 (21, 22) (Fig. 2B).
To position the SAMTOR function within the
m TORC1 pathway, we performed epistasis ex-
periments with established m TORC1 regulators.
Overexpression of SAMTOR inhibited m TORC1
signaling when coexpressed with the wild-type
RagA and RagC heterodimer, but not with the
constitutively active mutant heterodimer (RagA
Q66L and RagC S75N) that bypasses the require-
ment for amino acids for maintaining m TORC1
activity (Fig. 2C) (4, 5). In addition, SAMTOR did
not inhibit mTORC1 signaling in cells lacking
either a GATOR1 or KICSTOR component. Thus,
SAMTOR acts upstream of the Rag GTPases
and requires GATOR1 and KICSTOR to inhibit
mTORC1 signaling (Fig. 2D). In combination
with the interaction data, these results are con-
sistent with SAMTOR promoting the function
of GATOR1 and/or KICSTOR, which are both
negative regulators of mTORC1 signaling.
Sequence analyses predict that SAMTOR contains a class I Rossmann fold methyltransferase
domain (Fig. 3A and fig. S2) (23). These domains
are known to bind S-adenosylmethionine (SAM)
and exist in methyltransferases in bacteria, ar-chaea, and eukaryotes (24). To determine whether
SAMTOR binds SAM, we developed an equilibrium binding assay based on one we used to
detect the binding of leucine to Sestrin2 (16) and
determined that SAMTOR binds SAM with a dissociation constant of approximately 7 mM (Fig. 3B).
A competition binding assay revealed that, as
with other SAM-binding proteins, SAMTOR can
also bind S-adenosylhomocysteine (SAH), the
demethylated form of SAM (Fig. 3B).
Given these findings, we asked whether SAM
and SAH regulate the interaction of SAMTOR
and GATOR1-KICSTOR. Indeed, SAM and SAH,
but not methionine, homocysteine, adenosine,
5-methylthioadenosine, leucine, or isoleucine, disrupted the interaction when added directly to
the immunopurified complex kept at 4°C (Fig. 3,
C and D). Thus, SAM disrupts the interaction
between SAMTOR and GATOR1-KICSTOR analogously to how leucine and arginine induce the
release of Sestrin2 and CASTOR1 from GATOR2,
respectively (16, 19). Given that SAH has the
S Z T 2 Nprl3 D e p d c 5 Nprl2
HEK-293T expressing: endogenousFLAG- Depdc5 FLAG-Kaptin endogenousFLAG- WDR59 FLAG-metap2
Wild-type sgNprl3 sgSZT2 sgWDR24
HEK-293T cell lines:
–––– +++ +
–––– + +++– –
–––––– – –
Fig. 1. SAMTOR interacts with GATOR1 and KICSTOR. (A) GATOR1 and KICSTOR, but not GATOR2,
coimmunoprecipitate SAMTOR. FLAG immunoprecipitates (IP) were prepared from HEK-293T cell lines
that stably expressed FLAG-tagged metap2 or Kaptin, or had endogenously FLAG-tagged Depdc5 or
WDR59. FLAG immunoprecipitates and lysates were analyzed by immunoblotting for the indicated proteins.
FLAG-metap2 served as a negative control. Depdc5 and Nprl3, WDR59 and WDR24, and Kaptin and
SZT2 were used as representative components of the GATOR1, GATOR2, and KICSTOR complexes,
respectively; Raptor was used as a loading control. Short or long exposure indicates relative blot exposure
times. (B) SAMTOR coimmunoprecipitates GATOR1 and KICSTOR, and the interaction requires both GATOR1
and KICSTOR but not GATOR2. FLAG immunoprecipitates were prepared from wild-type, Nprl3-deficient, SZT2-
deficient, or WDR24-deficient HEK-293Tcells transiently expressing the indicated cDNAs. FLAG immunoprecipitates
and lysates were analyzed as in (A). (C) Model showing how SAMTOR interacts with GATOR1 and KICSTOR.
(D) Presence or absence of gene orthologs of SAMTOR in several model organisms.