INTRODUCTION

Transforming growth factor-s (TGF-s)¹ belong to a large family of structurally related cytokines that includes activins, inhibins, and bone morphogenetic proteins (BMPs) (1). The members are involved in a wide variety of biological phenomena such as carcinogenesis, development, and immunological response (2). Molecular cloning of the TGF-₁ receptors has revealed that the type I (TR-I) and the type II (TR-II) receptors are serine/threonine kinases (1, 3). TGF-₁ first binds to TR-II, which is a constitutively active kinase. Upon ligand binding, TR-II recruits and phosphorylates TR-I mostly, if not exclusively, at its juxtamembrane region (4). The juxtamembrane region of the type I receptor for TGF-₁ as well as for other TGF-₁-related molecules contains in its carboxyl-terminal half a highly conserved glycine-serine-rich sequence denoted the GS domain. Mutational analyses have shown that the GS domain is essential for TGF- signaling (4). More recent evidence suggests that the signaling pathway of TGF- may be bifurcated at TR-I. Deletion of the amino-terminal half of the juxtamembrane region, which is relatively variable among the type I receptors, resulted in abrogation of anti-proliferative effect of TGF-₁, whereas transcriptional response to TGF-₁ was retained (5). Taken together, the juxtamembrane region of TR-I seems to function as a major regulatory domain in the TR-I kinase.

A number of candidates that mediate growth-inhibitory effect by TGF-₁ has been identified. Earlier studies showed that TGF-₁ caused rapid down-regulation of the c-Myc expression (6) and overexpression of c-Myc protein blocked growth inhibition by TGF-₁ in mouse keratinocytes (7). The retinoblastoma tumor suppressor gene product (pRb) is one of the key regulators of the G₁ phase progression. Sequestration of pRb by DNA tumor viral proteins abrogated growth arrest by TGF-₁ (6), and TGF-₁ suppressed the pRb hyperphosphorylation that is crucial to the S phase entry (8). Recently a number of cyclin-dependent kinase inhibitors have been identified. TGF-₁ rapidly activated the expression of Ink4b/p15 and consequently suppressed the cyclin-dependent kinase activity in several cell lines sensitive to TGF-₁(9). Until recently only little has been known about the mechanisms of induction of certain genes such as plasminogen activator inhibitor-1 by TGF-₁. TAK-1, which belongs to the mitogen-activated protein kinase kinase kinase family, has been identified as a potential mediator of such transcriptional regulation (10). Genetic studies in Drosophila melanogaster have identified schnurri (shn) and mothers-against-dpp (mad) as components of the signaling pathway of decapentaplegic (dpp), a member of the TGF- superfamily in Drosophila (11, 12, 13). Mammalian homologs of mad are shown to be implicated in BMP signaling (14, 15, 16). It is also intriguing that DPC-4, a putative tumor suppressor gene identified in human pancreatic cancers, belongs to the mad family (17).

Although the number of the possible players in the TGF--specific response is growing as listed above, no direct links between the TGF- receptors and any of those proteins have been established. To elucidate the signaling network emanating from the receptors, several groups including us have employed the yeast two-hybrid system. TRIP-1, a novel protein with Trp-Asp domains, was shown to bind to TR-II both in vitro and in vivo (18). We and others have identified the type II receptor for BMP (19, 20), and farnesyl transferase- (FT) (21, 22) as interactors of TR-I. The most abundant clones in our screen were FKBP12, a binding protein for immunosuppressants such as FK506 and rapamycin, which was the first molecule reported to associate with the cytoplasmic region of the activin and TGF- type I receptors in vitro (23). Here we demonstrate that FKBP12 interacts with TR-I in vivo, and the juxtamembrane region of TR-I is indispensable for the interaction. Finally the role of FKBP12 in TR-I signaling is discussed.

EXPERIMENTAL PROCEDURES

pEG202 and pJG4-5 are yeast expression vectors: the former for a bait and the latter for a prey (24). Construction of FT and the wild type and mutant TR-I in these vectors have been described (21). The entire coding region of the human FKBP12 was amplified by polymerase chain reaction (PCR) from one of the positive clones and inserted between the EcoRI and XhoI sites of pEG202 and pJG4-5. Glutathione S-transferase (GST) fusion protein expression plasmids of TR-I and FT were described (21). GST-FKBP12 was constructed by subcloning the insert of FKBP12 in pJG4-5 into pGEX-4T-1 (Pharmacia Biotech Inc.) at the same cloning sites. Viral hemagglutinin (HA)-tagged TR-I and hexahistidine (His)-tagged TR-II in pCMV5(4) were gifts from J. Massagué. More HA-tagged TR-I expression plasmids were made to ensure a similar expression level of the wild type and mutants. First, pcDNA3-HA, a mammalian expression vector with an HA tag within the multicloning site, was made by inserting an annealed oligonucleotide (5-TCGAGGGGTATCCGTACGATGTGCCCGACTATGCTTAAGTCGACTCTAG-3) between the XhoI and XbaI sites of pcDNA3 (Invitrogen). The whole coding region of TR-I was amplified by PCR with an EcoRI site and an XhoI site added before the starting codon and in place of the stop codon, respectively, followed by insertion between the EcoRI and XhoI sites of pcDNA3-HA. Internal EcoRI and XhoI sites of TR-I were destroyed beforehand to facilitate subcloning. The JD1, JD2, JM1, JM2, JM3, JM123, and K232R mutants of TR-I in pMEP4 have been described (5). JD3 has a deletion of amino acids 182-206, and T204D has an aspartic acid in place of threonine 204. The cDNA between the Bsp1407I and PflMI sites of the wild type TR-I in pcDNA3-HA was replaced with the corresponding region of each mutant TR-I in pMEP4, yielding a series of HA-tagged TR-I. pcDNA3-FLAG was made by inserting an annealed oligonucleotide (5-TCGAGGGGGACTATAAGGACGATGATGACAAATAAGTCGACTCTAG-3) between the XhoI and XbaI sites of pcDNA3. The coding region of FKBP12 without the stop codon was subcloned between the EcoRI and XhoI sites of pcDNA-FLAG, yielding FLAG-tagged FKBP12. The details of the plasmid construction including the oligonucleotide sequences are available upon request. All of the PCR products were sequenced.

Interaction in yeast was tested using the interaction trap as described (19). Briefly, a combination of the reporter, the bait, and the prey plasmids was introduced into EGY48 and transformants were selected on appropriate media. The interaction was evaluated in the -galactosidase assay.

293 cells derived from human embryonic kidney were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 units/ml penicillin, and 4.5 g/liter glucose. Cells were transfected with 10 µg of each plasmid using the calcium phosphate precipitation method (25). After 2 days, cells were labeled with 22.8 mCi/ml [³⁵S]methionine and cysteine mixture (Amersham) overnight, and lysed in 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, and 0.5% Triton X-100. In Fig. 2, indicated concentrations of FK506 were added 40 h before the lysis. Cleared lysates were divided into two tubes and incubated with anti-FLAG M2 (Eastman Kodak Co.) or anti-HA (Boehringer Mannheim) antibodies in the presence of protein A-Sepharose (Kabi-Pharmacia). Rabbit anti-mouse immunoglobulin antibodies (DAKO A/S) were added in the precipitation with anti-FLAG antibodies. The precipitates were washed and subjected to SDS-polyacrylamide gel elecrophoresis (SDS-PAGE) (7-20% gradient gel), fluorography, and analyses with Fuji BAS 2000 Bio-Imaging Analyzer (Fuji Photo Film).

Effect of FK506 on the interaction of FKBP12 with TR-I.

Affinity cross-linking experiments were done essentially as described (25). The labeled lysates were immunoprecipitated with anti-HA or anti-FLAG antibodies and analyzed as described above.

In vitro phosphorylation was done essentially as described (21). 560 ng of GST-FT or GST-FKBP12 was mixed with 50 ng of GST-TR-I in 30 ml of phosphorylation buffer. Phosphorylated samples were then subjected to SDS-PAGE, Coomassie Blue staining, and autoradiography.

RESULTS

To identify proteins that interact with the cytoplasmic region of TR-I, we conducted an interaction trap screen of the HeLa cDNA library (24). The final clones from the screen were classified into three groups. One clone encoded a novel type II serine/threonine kinase receptor (19), which was later shown to bind BMPs (20, 26). Ten clones were FT, which is also shared by geranyl-geranyl transferase-I (27). Seventeen of the the rest of the positives encoded FKBP12 (data not shown). We examined the interaction of FKBP12 with TR-I in vivo. FKBP12 was epitope-tagged with FLAG at the carboxyl terminus (FKBP12-FLAG). FKBP12-FLAG interacted with TR-I in the yeast system (data not shown). HA-tagged TR-I and/or FKBP12-FLAG were transiently expressed in 293 cells. Labeled lysates were immunoprecipitated with anti-HA or anti-FLAG antibodies and then subjected to SDS-PAGE and fluorography (Fig. 1A). Each antibody specifically recognized TR-I-HA and FKBP12-FLAG. Anti-FLAG antibodies coprecipitated TR-I only when FKBP12 was coexpressed, demonstrating that FKBP12 interacts with TR-I in vivo. Reciprocal precipitation with anti-HA antibodies in the coexpression of TR-I and FKBP12 failed to precipitate FKBP12 (see ``Discussion''). We thus used anti-FLAG antibodies in the following coprecipitation experiments.

interaction of FKBP12 with TR-I and the effect of ligand-binding.

When TR-I and FKBP12 were coexpressed with TR-II in 293 cells, TR-I still coprecipitated with FKBP12. Addition of TGF-₁ in this condition did not change the amount of the coprecipitated TR-I (data not shown), suggesting that TGF-₁ does not affect the interaction of FKBP12 with TR-I. To study this more precisely, we did affinity cross-linking experiments. Cells transfected with appropriate combinations of plasmids were affinity-labeled with ¹²⁵I-TGF-₁, and the lysates were precipitated with anti-FLAG antibodies (Fig. 1B). In the absence of TR-I or FKBP12, no cross-linked receptor complexes were detected, while both TR-I and TR-II precipitated in the presence of FKBP12. Therefore FKBP12 associates with TR-I in the presence of ligand.

FK506 binds to FKBP12 and was shown to interfere with the interaction of FKBP12 with TR-I in yeast (23). We examined the effect of various concentrations of FK506 on the interaction in 293 cells (Fig. 2). A significant decrease in the intensity of the coprecipitated TR-I was observed at 100 nM (30% of the control) and 500 nM (12%), suggesting that the conformational change of FKBP12 induced by FK506 binding resulted in the dissociation of FKBP12 from TR-I or the drug and TR-I share the same binding site on FKBP12 (23). Similar results were observed in the rapamycin treatment (data not shown).

FT, another interactor of TR-I, showed different affinities to the wild type and mutant TR-I (21). We compared FKBP12 with FT in its interaction with different forms of TR-I (Fig. 3A). The T200V mutant, in which threonine 200 was converted to valine, was shown to be inactive, whereas the T204D mutant with aspartic acid in place of threonine 204 was constitutively active in vivo (28). K232R is a kinase-negative mutant where the ATP-binding site conserved in kinases was disrupted. JM lacks the entire juxtamembrane region (amino acids 148-204). FT bound to T204D most efficiently but did not associate with T200V, as has been shown quantitatively before (21). FT interacted with JM. In remarkable contrast, FKBP12 did not interact with JM but bound to all the other forms of TR-I, suggesting that the juxtamembrane region is indispensable for FKBP12 to bind to TR-I. However, the juxtamembrane region alone does not bind FKBP12 (23).

Interaction of FKBP12 with the wild type and mutant TR-I.

The juxtamembrane region of TR-I contains the GS domain, a major phosphorylation site in TR-I by TR-II, and seems to have a regulatory role in TR-I activation (4). As this region was implicated in the interaction of FKBP12 with TR-I in yeast, a deletion mutant of the whole juxtamembrane region (JD2; 150-206) was constructed and tested for interaction in mammalian cells (Fig. 3B). JD2 was comparable to the wild type in the expression level but did not coprecipitate with FKBP12. We recently showed that deletion of the amino-terminal half of the juxtamembrane region (JD1; 150-181) or point mutations, JM2 (S172A) or JM3 (T176V), in this region abrogated growth inhibition by TGF-₁ but not transcriptional response (5). It is thus of interest whether FKBP12 interacts with any of these mutants. JD1 failed to bind FKBP12, indicating that the interaction of FKBP12 with TR-I is not necessary at least for transcriptional regulation by TGF-₁. Similarly another deletion mutant, JD3 (182-206), which lacks the carboxyl-terminal half of the juxtamembrane region including the GS domain, did not associate with FKBP12. However, all of the point mutants tested including JM1 (S165A) and JM123 (S165A/S172A/T176V) still bound FKBP12. These results argue that the overall structure of the juxtamembrane region is necessary for TR-I to bind FKBP12. Furthermore, both constitutively active T204D and kinase-negative K232R bound FKBP12 in conformity with the results in yeast (Fig. 3B).

FT was phosphorylated by TR-I in vitro, suggesting that FT may be a direct substrate of the TR-I kinase (21). We studied whether FKBP12 was phosphorylated under the same condition. GST fusion proteins were prepared and in vitro kinase assay was performed as described (21). The same amount of GST-FT and GST-FKBP12 was applied to the reaction mixture as shown in Coomassie Blue staining (Fig. 4). TR-I phosphorylated FT efficiently, but only very little phosphorylation was detected in FKBP12. FKBP12 stayed bound to TR-I regardless of the ligand occupancy of the receptor (Fig. 1B) and constitutively active T204D or kinase-negative K232R still bound FKBP12 both in vitro and in vivo (Fig. 3). Taken together, FKBP12 does not seem to be a substrate of TR-I.

phosphorylation of the interactors by TR-I.

DISCUSSION

In the present report have we shown that: 1) FKBP12 interacts with TR-I both in vitro and in vivo, 2) the interaction can be seen in the presence of ligand, 3) FK506 interferes with the interaction, 4) the juxtamembrane region is indispensable for the interaction, and 5) FKBP12 is not phosphorylated by TR-I in vitro. Furthermore, our present and previous observations (5) suggest that FKBP12 is not necessary at least for transcriptional response to TGF-₁.

Our initial attempt to coprecipitate FKBP12 with TR-I by anti-HA antibodies failed. Although the three-dimensional structure of TR-I is unknown, it is possible that anti-HA antibodies interfere with the interaction if the carboxyl terminus of TR-I is located near the juxtamembrane region. TR-I has 33 methionines and cysteines in its mature form, whereas FKBP12 has only 5 amino acids to be labeled with ³⁵S. This may also explain the difficulty in detecting FKBP12 bands in the coprecipitation with anti-HA antibodies.

FKBP12 was identified as an intracellular receptor for FK506, and was later shown to bind rapamycin as well (29). Both drugs strongly suppress the immune activity, but the mechanisms whereby these drugs exert their effects on lymphocytes are quite different. FK506 suppresses the expression of interleukin-2 in T-cells thereby inhibits the cells to exit from the G₀ to G₁ phase. In contrast, rapamycin inhibits the cyclin-dependent kinase activity and arrest cells at the G₁/S boundary. This is intriguing since TGF-₁ also causes the G₁/S arrest of proliferating cells (6). FKBP12 thus may be a common component of the growth-suppressive pathway of rapamycin and TGF-₁. Here we showed that FKBP12 is not necessary for transcriptional regulation by TGF-₁ but do not have any direct evidence whether it is one of components of the anti-proliferative pathway of TGF-₁. However, the interaction of FKBP12 with TR-I was observed in the presence and absence of ligand, and FKBP12 was not phosphorylated by TR-I whereas the case is the opposite with FT. We thus speculate that FKBP12 is not a direct target of the TR-I kinase. What, then, is the role of FKBP12 in TGF-₁ signaling?

Another function of FKBP12 is the peptidyl-propyl cis/trans isomerase activity. This function, however, is not thought to play a role in the immunosuppression since the effective concentrations of the drugs are different for each maximal activity (29, 30). Perhaps this function may help the TR-I proteins to fold correctly at the cell membrane, namely FKBP12 could function as a chaperon for TR-I (30). Alternatively, FKBP12 may modulate TGF-₁ signaling through its interaction with the juxtamembrane region. FKBP12 binds to ryanodine receptor (RyR), a calcium (Ca²⁺) release channel of the sarcoplasmic and endoplasmic reticula (31). When bound to RyR, FKBP12 modulates channel gating by making the channel harder to open but more stable once it is open. FK506 and rapamycin disrupts the complex and reverse the stabilizing effects of FKBP12. FKBP12 is also physiologically associated with inositol 1,4,5-triphosphate receptor (IP₃-R), another Ca²⁺ channel (32). Addition of FK506 disrupts the complex formation and alters the channel conductance as in RyR (31). Furthermore calcineurin, which is a Ca²⁺-sensitive serine/threonine phosphatase, was shown to be anchored to IP₃-R via FKBP12, regulating the phosphorylation status of the receptor (32).

FKBP12 may modulate the regulatory function of the juxtamembrane region in analogy to RyR or IP₃-R. More specifically, FKBP12 may facilitate the dephosphorylation of the juxtamembrane region by recruiting phosphatases. FKBP12 may interfere with the transphosphorylation of TR-I by TR-II by hiding the phosphorylation sites. Or FKBP12 may affect the heterodimerization of TR-I and TR-II as a wedge or an adaptor. Further studies will be needed to address these hypotheses.