Interaction of RAFT1 with Gephyrin Required for Rapamycin-Sensitive Signaling

See allHide authors and affiliations

Science  14 May 1999:
Vol. 284, Issue 5417, pp. 1161-1164
DOI: 10.1126/science.284.5417.1161


RAFT1 (rapamycin and FKBP12 target 1; also called FRAP or mTOR) is a member of the ATM (ataxia telangiectasia mutated)–related family of proteins and functions as the in vivo mediator of the effects of the immunosuppressant rapamycin and as an important regulator of messenger RNA translation. In mammalian cells RAFT1 interacted with gephyrin, a widely expressed protein necessary for the clustering of glycine receptors at the cell membrane of neurons. RAFT1 mutants that could not associate with gephyrin failed to signal to downstream molecules, including the p70 ribosomal S6 kinase and the eIF-4E binding protein, 4E-BP1. The interaction with gephyrin ascribes a function to the large amino-terminal region of an ATM-related protein and reveals a role in signal transduction for the clustering protein gephyrin.

Proteins of the ATM family participate in cell cycle progression by linking signals from growth factor receptors and internal checkpoints to the cell cycle machinery. These cell cycle regulators are members of the kinase superfamily and include the gene product of the ataxia telangiectasia locus (ATM), the catalytic subunit of the DNA-activated protein kinase (DNA-PKcs), RAFT1 or FRAP, and the products of the yeast genes TOR1, TOR2, and TEL1(1).

RAFT1 and its yeast homologs, the TOR proteins, are the in vivo targets for the complex of rapamycin with its intracellular receptor, FKBP12. Rapamycin is a potent immunosuppressant that prevents progression through the G1 phase of the cell cycle in various cell types, including T lymphocytes and budding yeast (2). The effects of rapamycin point to a role for RAFT1 and the TORs in cell cycle regulation, and increasing evidence indicates that they participate in mitogen-stimulated signaling pathways that control mRNA translation. In mammalian cells RAFT1 controls the rapamycin-sensitive phosphorylation of at least two translational regulators: the p70 ribosomal S6 kinase (p70S6k) and 4E-BP1, an inhibitor of the cap binding protein eIF-4E (3,4). In yeast, TOR function is necessary for initiation of mRNA translation and may contribute to a checkpoint that prevents cell cycle progression in the absence of nutrients (5).

All ATM-related proteins are large (greater than 200 kD) and have a COOH-terminal region with similarity to the catalytic domains of phosphoinositide 3- and 4-kinases (6). Despite this similarity, RAFT1 and related proteins appear to be protein kinases, and RAFT1 immunoprecipitates directly phosphorylate p70S6k(7) and 4E-BP1 in vitro (4,7).

To identify proteins that may interact with the first 2000 residues of RAFT1, we arbitrarily subdivided this region and used each part individually as bait in a yeast two-hybrid screen of a rat hippocampal expression library (8). Amino acids 944 to 1338 of RAFT1 interacted with gephyrin, a tubulin-binding protein necessary for the postsynaptic clustering of glycine receptors in spinal cord neurons (9). Residues 1010 to 1128 are the smallest part of RAFT1 that interacted with gephyrin in the yeast two-hybrid assay and represent the minimal gephyrin-binding domain (GBD) of RAFT1 (Fig. 1A). Alignment of the 118–amino acid GBD and the intracellular loop of the β subunit of the glycine receptor revealed that residues 1017 to 1046 of RAFT1 share 45% sequence similarity to residues 405 to 435 of the glycine receptor (Fig. 1B). This is the same region of the receptor that interacts with gephyrin (10). By mutating several of the residues conserved between RAFT1 and the glycine receptor we created RAFT1 mutants that do not interact with gephyrin (RAFT1 I1034K and IV1034-5KR) or that show decreased binding to gephyrin relative to that of the wild-type protein (RAFT1 I1034A) (Fig. 1B).

Figure 1

Specific interaction between gephyrin and amino acids (aa) 944 to 1338 of RAFT1. (A) Interaction of indicated RAFT1 fragments with gephyrin in yeast two-hybrid system (8). The results were scored as follows: for β-Gal: +++, blue color in <15 min; ++, blue color in 15 to 45 min; +, blue color in 45 to 120 min;, no blue color in <24 hours. For HIS3:+, growth or , no growth on histidine-free media. (B) Residues 1017 to 1046 of RAFT1 are 45% similar to residues 405 to 435 of the β-chain of the glycine receptor (20). Isoleucine 1034 is critical for the RAFT1-gephyrin interaction; replacement with lysine or alanine abolishes or reduces, respectively, the association (8).

We used an in vitro binding assay to determine that full-length RAFT1 also interacts with gephyrin (11). Endogenous RAFT1 from detergent lysates of human embryonic kidney (HEK293) cells bound to a glutathione S-transferase (GST)–gephyrin affinity resin but not to one made with GST alone (Fig. 2A). The gephyrin-RAFT1 interaction was unaffected by the addition of 10 nM rapamycin to the cell lysates. We also expressed gephyrin in HEK293 cells with wild-type RAFT1 or variants incorporating mutations that reduced or eliminated the GBD-gephyrin interaction in yeast (12). Antibodies to gephyrin coimmunoprecipitated wild-type RAFT1 but not the I1034K or IV1034-5KR RAFT1 mutants (Fig. 2B). Similarly, RAFT1 I1034A, which retains partial binding to gephyrin in yeast, coimmunoprecipitated with gephyrin but in smaller amounts than the wild-type protein (Fig. 2B). Thus, RAFT1 interacts with gephyrin in HEK293 cells, and point mutations in the GBD reduce or eliminate the interaction in yeast and mammalian cells.

Figure 2

Interaction of full-length RAFT1 with gephyrin in vitro and in vivo, and effects of mutations in the GBD of RAFT1 on the association. (A) Endogenous RAFT1 from HEK293 cells interacted with GST-gephyrin but not with GST alone. The addition of 10 nM rapamycin (Rapa) to the cell lysates did not disrupt the interaction. Results are representative of three separate experiments (11). (B) The indicated mycRAFT1 variants were expressed with epitope-tagged gephyrin in HEK293 cells. RAFT1 coimmunoprecipitating with anti-gephyrin immunoprecipitates was detected with immunoblotting (middle panel) as were expression of RAFT1 variants (top) and gephyrin (bottom). Results are representative of three separate experiments (12).

All tissues examined contained RAFT1 and gephyrin, including HEK293 cells, which express full-length RAFT1 and the 93-kD form of gephyrin (13). In rat embryos RAFT1 and gephyrin mRNAs are widely expressed, with the largest amounts found in the developing central nervous system, kidney, thymus, and small intestine (14). In subcellular fractions (Fig. 3A) of rat brain both RAFT1 and gephyrin were abundant in the LP2 fraction, which is highly enriched in synaptic vesicles and presynaptic membranes (15). RAFT1 was not detected in the soluble fraction of synaptosomes (LS2) whereas gephyrin was present in small amounts (Fig. 3A).

Figure 3

Subcellular localization of RAFT1 and gephyrin and effects of overexpression of the RAFT1 GBD. (A) Subcellular localization of RAFT1, gephyrin, and control proteins in rat brain fractions (15). H, homogenate; P1, crude nuclei; P2, mitochondria and synaptosomes; P3, crude microsomes; S3, cytosol; LP1, postsynaptic densities; LP2, presynaptic membranes and synaptic vesicles; LS2, soluble synaptosomal components, and synapt., synaptophysin. (B) Enrichment of RAFT1 in non-nuclear cellular membranes. Subcellular fractions of HeLa cells (21) were analyzed by immunoblotting with affinity-purified 782 antibody to RAFT1 (19). (C) Altered subcellular localization of endogenous RAFT1 in cells expressing the RAFT1 GBD (residues 1010 to 1128). HeLa cells transfected with a mycGBD construct were stained with mouse antibody 9E-10 to myc (red cells, right panel) and rabbit 782 antibody to endogenous RAFT1 (green cells, left panel) (22). Cells that are green in the left panel and not red in the right panel serve as controls for the localization of endogenous RAFT1 in HeLa cells in the absence of mycRAFT1 GDB expression. Quantitation of two separate experiments indicated that 64 out of 70 cells expressing mycGBD showed RAFT1 localization similar to that shown. Expression of control myc-tagged proteins did not disrupt RAFT1 localization (13). (D) Expression in HeLa cells of mycRAFT1 (green) with gephyrin causes mycRAFT1 to appear in large clumps in the cytoplasm. mycRAFT1 was detected with antibody 9E-10 to myc (22). Similar results were obtained with COS cells (13).

We examined whether gephyrin might contribute to the intracellular localization of RAFT1 as it does for the glycine receptor (9). In HeLa cell subcellular fractions, as in the brain (Fig. 3A), RAFT1 was enriched in cellular membranes (Fig. 3B). Immunofluorescence staining of HeLa cells with an antibody to RAFT1 revealed a fine punctate pattern in the cytoplasm that is not characteristic of any particular subcellular compartment (Fig. 3C) but is similar in appearance to the cytoplasmic localization of the related protein ATM (16). Overexpression in HeLa cells of the minimal GBD of RAFT1 (amino acids 1010 to 1128) disrupted the normal subcellular localization of RAFT1, causing the protein to appear in clumps dispersed throughout the cytoplasm (Fig. 3C). When mycRAFT1 was expressed in HeLa cells, it was evenly distributed throughout the cytoplasm (Fig. 3D), with an appearance similar to that of endogenous RAFT1 (Fig. 3C). However, when mycRAFT1 was expressed with gephyrin, it appeared in large aggregates in the cell cytoplasm (Fig. 3D). The glycine receptor, a known gephyrin-binding protein, displays a similar localization when expressed with gephyrin in HEK293 cells (10).

Great overexpression of any protein substantially inhibits p70S6k activity and 4E-BP1 phosphorylation (17), precluding us from testing the effects of GBD expression on the activity of these downstream signaling molecules. Instead, to determine whether the association with gephyrin is necessary for RAFT1 function, we tested whether RAFT1 mutants incapable of interacting with gephyrin were able to signal to p70S6kand 4E-BP1. We incorporated the GBD mutations into RAFT1 S2035T, a version of the protein that can activate downstream transducers even in the presence of rapamycin because it does not bind the rapamycin-FKBP12 complex (3). In quiescent cells treated with rapamycin, serum induces a fivefold increase in p70S6k activity in cells expressing S2035T but not in cells expressing wild-type RAFT1 (Fig. 4A). The rapamycin-resistant function of RAFT1 S2035T depends on its wild-type GBD because the addition of mutations that disrupt gephyrin binding (I1034K or IV1034-5KR) also eliminates its ability to activate p70S6kin the presence of rapamycin (Fig. 4A). RAFT1 I1034A retains partial binding to gephyrin, and the I1034A/S2035T double mutant allowed an intermediate amount of rapamycin-resistant, serum-inducible activation of p70S6k (Fig. 4A).

Figure 4

GBD mutations that disrupt the RAFT1-gephyrin interaction prevent RAFT1-mediated signaling. (A) Serum-stimulated p70S6k activation. Quiescent HEK293 cells expressing the indicated mycRAFT1 variants and HA-p70S6k were treated with 10 nM rapamycin for 30 min. Cells were lysed after rapamycin treatment (shaded bars) or after serum stimulation for 30 min (black bars), and S6 kinase activity was determined (7). Results are means ± SE for three separate experiments and normalized to the S6 kinase activity in rapamycin-treated cells expressing only p70S6k. Expression of p70S6k (top panel) and RAFT1 variants (bottom panel) was monitored with immunoblotting (23). (B) Serum-stimulated 4E-BP1 phosphorylation. Cells expressing the indicated mycRAFT1 variants and HA–4E-BP1 were treated as in (A). The amount of 4E-BP1 bound to eIF-4E was determined by immunoblotting of proteins interacting with a 7-methyl–guanosine triphosphate (GTP) affinity resin (middle) (7). Expression of 4E-BP1 (top) and RAFT1 variants (bottom) was analyzed with immunoblotting (23). Results are representative of three separate experiments. (C) Phosphorylation of 4E-BP1 in actively growing cells. Twenty-four hours after transfection, HEK293 cells expressing the indicated mycRAFT1 variants and HA–4E-BP1 were treated with 10 nM rapamycin or ethanol vehicle for 30 min. Cells were then lysed and the amount of 4E-BP1 bound to eIF-4E (middle) was determined as in (B). Expression of 4E-BP1 (top) and RAFT1 variants (bottom) was detected with immunoblotting (23). Results are representative of four separate experiments. (D) The GDB mutations do not impair the capacity of mycRAFT1 to autophosphorylate (middle) or phosphorylate 4E-BP1 (bottom). Kinase assays were done as described (24) after immunoprecipitation with 9E-10 antibody to myc. Expression of mycRAFT1 was monitored with immunoblotting (top).

The RAFT1 mutants deficient in gephyrin binding could not mediate signaling to 4E-BP1, the small translational repressor whose interaction with eIF-4E is blocked by rapamycin-sensitive phosphorylation (3). Expression of RAFT1 S2035T having a wild-type or I1034A GBD allowed serum-stimulated, rapamycin-resistant phosphorylation of 4E-BP1. We detected 4E-BP1 phosphorylation in two ways: decreased migration of 4E-BP1 on SDS-PAGE and a decrease in the amount of 4E-BP1 that interacted with eIF-4E (Fig. 4B). Incorporation of the I1034K or IV1034-5KR mutations eliminated the ability of RAFT1 S2035T to signal to 4E-BP1 in the presence of rapamycin (Fig. 4B).

In addition to preventing serum-stimulated 4E-BP1 phosphorylation in quiescent cells, rapamycin treatment causes dephosphorylation of this protein in actively growing cells (3). 4E-BP1 was protected from rapamycin-induced dephosphorylation in cells expressing 4E-BP1 with RAFT1 S2035T or I1034A/S2035T (Fig. 4C). In cells expressing only 4E-BP1 or 4E-BP1 with wild-type or S2035T/I1034K RAFT1, rapamycin produced the characteristic signs of 4E-BP1 dephosphorylation (Fig. 4C). The perturbed function of the RAFT1 GBD mutants appears not to be due to an effect of the mutations on RAFT1 kinase activity because the GBD mutants autophosphorylated and phosphorylated 4E-BP1 as efficiently as did the wild-type enzyme (Fig. 4D).

There is increasing awareness of the role translational controls play in many biological phenomena. Signaling pathways that regulate protein synthesis are likely to be important for the translation of mRNAs localized to distinct cellular compartments, such as neuronal dendrites (18). In neurons gephyrin is localized to synaptic sites in dendrites (9). Our finding that RAFT1 interacts with gephyrin makes RAFT1 an attractive candidate for a regulator of dendritic mRNA translation.

  • * Present address: Whitehead Institute for Biomedical Research, 9 Cambridge Center, Cambridge, MA 02142, USA.

  • To whom correspondence should be addressed. E-mail: ssnyder{at}


View Abstract

Stay Connected to Science

Navigate This Article