An aspartate/insulin receptor chimera mitogenically activates fibroblasts.

A gene encoding the ligand-binding domain of the Escherichia coli aspartate receptor fused to the cytoplasmic domain of the insulin receptor tyrosine kinase to produce the chimeric aspartate insulin receptor (AIR) was expressed in mammalian cells. A murine fibroblast transfectant line designated CA3 was generated that stably expressed the AIR receptor. This 70,000 Mr receptor containing the tyrosine kinase of the insulin receptor was recognized by aspartate receptor-specific antisera. When isolated in cellular membrane preparations, AIR was found to be capable of autophosphorylation and phosphorylation of histone H2B on tyrosine. The receptor was found to be predominately cytoplasmic and to be situated in the endoplasmic reticulum and Golgi membranes by immunofluorescence imaging of CA3 cells. Mitogenic effects of AIR were observed; CA3 cells continued DNA synthesis under serum deprivation conditions that prevented parental cells from cycling. These results demonstrate that a chimeric receptor containing procaryotic transmembrane sequences is expressed by a eucaryotic cell in intracellular membranes and functionally couples to cellular signaling pathways.

A gene encoding the ligand-binding domain of the Escherichia coli aspartate receptor fused to the cytoplasmic domain of the insulin receptor tyrosine kinase to produce the chimeric aspartate insulin receptor (AIR) was expressed in mammalian cells. A murine fibroblast transfectant line designated CA3 was generated that stably expressed the AIR receptor. This 70,000 M r receptor containing the tyrosine kinase of the insulin receptor was recognized by aspartate receptor-specific antisera. When isolated in cellular membrane preparations, AIR was found to be capable of autophosphorylation and phosphorylation of histone H2B on tyrosine. The receptor was found to be predominately cytoplasmic and to be situated in the endoplasmic reticulum and Golgi membranes by immunofluorescence imaging of CA3 cells. Mitogenic effects of AIR were observed; CA3 cells continued DNA synthesis under serum deprivation conditions that prevented parental cells from cycling. These results demonstrate that a chimeric receptor containing procaryotic transmembrane sequences is expressed by a eucaryotic cell in intracellular membranes and functionally couples to cellular signaling pathways.
Chimeric transmembrane receptors of eucaryotic origin have been constructed in which extracellular, transmembrane, and intracellular domains are combined in order to study receptor function. The resulting hybrids have often exhibited transmembrane signaling. For instance, insulin binding to an extracellular insulin receptor domain was shown to activate an intracellular kinase domain derived from the epidermal growth factor receptor (1). A functional hybrid was constructed between the platelet-derived growth factor receptor and the fibroblast growth factor receptor (2). Chimeric procaryotic receptors between domains from the Escherichia coli aspartate receptor (AR) 1 and EnvZ receptor have also been shown to transduce signals (3). In addition, two different hybrids derived from the procaryotic E. coli AR extracellular domain and the human insulin receptor kinase domain have been expressed and studied in E. coli (4,5), one of which was regulated by aspartate.
Despite a lack of sequence similarity or shared biochemical activities, the IR and AR have some congruence. The AR is a homodimer of 60 kDa subunits with noncovalent contacts between the extracellular and possibly the transmembrane domains (6,9). Each subunit has two transmembrane domains that flank the extracellular aspartate binding domain. Two identical aspartate binding sites have been identified in the dimer by equilibrium binding studies and x-ray diffraction imaging that are known to interact leading to negatively cooperative ligand binding (7,8). The IR configuration is analogous in that two ␣␤ transmembrane subunits join as a "dimer" (␣␤) 2 , but they are covalently linked by a disulfide as well as noncovalent bonds (9). Each ␣␤ molecule has an insulin binding site, and the ␣ 2 ␤ 2 receptor exhibits negatively cooperative insulin binding (10,11).
Both receptors transmit signals across the membrane by conformational changes because receptor clustering mechanisms have been ruled out (6,9). Small conformational changes that may be involved in signaling have been observed in AR crystal structures and nuclear magnetic resonance studies (7,12). No similar structural data are available for the IR, but studies of its signaling mechanism have indicated that binding of an insulin molecule by one ␣␤ subunit has the initial effect of inducing autophosphorylation of the other subunit (13). Intersubunit autophosphorylation precedes activation of the exogenous kinase activity of the IR (14).
The aspartate/insulin receptor chimera (AIR) when expressed in procaryotes responds to aspartate binding with stimulated kinase activity (5). It was of interest to examine how a eucaryotic system would handle this hybrid containing procaryotic ligand-binding and transmembrane domains and whether the resulting receptor could function. We have therefore expressed AIR in mammalian cells and studied its enzymatic activity, subcellular location, and interactions with cellular mitogenic pathways.

MATERIALS AND METHODS
Cells and Antisera-Murine 3T3 crip cells (15) were obtained from the laboratory of Richard Mulligan and were cultured under standard conditions in Dulbecco's modified Eagle's medium containing 10% calf serum and antibiotics. Rabbit antiserum 9207 was generated by immunization of a female New Zealand White rabbit with purified AR expressed in E. coli. AR-specific IgG was isolated by affinity purification on a column of recombinant AR cross-linked to Affi-gel 10 (Bio-Rad) essentially as described (16).
Expression of AIR-Plasmid pLJAIR was constructed by subcloning the BamHI-SalI DNA fragment from pLAIR (5) containing the AIR gene into retroviral expression vector pLJ (17) that had been cut with the same two enzymes. CsCl-purified DNA was used in calcium phosphate transfections of the fibroblast line crip . Colonies were isolated after selection of transfected cells in 500 g/ml G418, and clone CA3 was chosen for further study on the basis of immunoblot screening of cell lysates with anti-AR serum 9207. AIR expression was found to be maintained for more than 20 passages under G418 selection, and experiments were carried out on AIR transformants that had been passaged 3-15 times.
Cell Lysis and Labeling-Particulate and soluble fractions were pre-* This work was supported by Grant 09765 from the NIDDK, National Institutes of Health (to D. E. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Metabolic labeling of cells with [ 32 P]orthophosphate was performed with nearly confluent 100-mm dishes of cells that had been cultured in 0.5% serum for 22 h, washed in phosphate-and serum-free medium, and labeled with 0.1 mCi/ml [ 32 P]orthophosphate in the same medium for 2 h. Phosphate-buffered saline (PBS; 150 mM NaCl. 20 mM sodium phosphate, pH 7)-washed monolayers were lysed on ice in 1.5 ml of RIPA buffer (150 mM NaCl, 25 mM Tris-Cl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS and 40 g/ml phenylmethylsulfonyl fluoride, pH 8) plus 0.1 mM sodium vanadate, and solublized proteins were isolated by centrifugation at 200,000 ϫ g for 12 min.
Thymidine incorporation was measured in six-well plates with subconfluent cells that had been starved in Dulbecco's modified Eagle's medium (0.5% calf serum) for 30 h and then stimulated with 10% calf serum for 18 h or left unstimulated. 1 Ci/ml [ 3 H]thymidine was added for 1 h before harvesting, cells were washed in PBS, and measurement of acid-insoluble counts was as described (18).
Receptor Analyses-Immunoprecipitations were performed in 1 ml for 1.5 h at 4°C in lysis buffer plus Triton X-100 (for kinase assays) or RIPA buffer plus 0.1 mM sodium vanadate (for labeled cells) with affinity-purified anti-AR antibodies (approximately 3-10 g that had been prebound to protein A beads) and followed by four washes in the same buffer. Immunoprecipitates corresponding to approximately 10 6 cells were electrophoresed on 10% denaturing polyacrylamide gels, and autoradiography was performed.
Fifty-microliter kinase reactions were carried out for 15 min at room temperature in 20 mM HEPES, 10 mM MgCl 2 , 10 mM MnCl 2 , 40 g/ml phenylmethylsulfonyl fluoride, 20 M [␥-32 P]ATP (5 Ci), pH 7.3, with or without 0.5 mg/ml histone H2B (Boehringer Mannheim). Densitometry of SDS-polyacrylamide gel autoradiograms was performed with a digital imaging system (Alpha Innotech IS-1000). Phosphoamino acid analysis of proteins in gel slices was performed by electrophoresis in one dimension only at pH 3.5 essentially as described (5). All autoradiography was with Kodak X-AR film at Ϫ70°C. Anti-AR immunoblotting (approximately 50 g of protein/lane) was performed essentially as described (5). Immunoblotting with a rabbit anti-phosphotyrosine antibody (approximately 100 g of lysate protein/lane) was essentially as described (19).
Immunofluorescence-Cells were grown on coverslips and fixed, permeablized, and stained under conditions essentially as described (20). Coverslips were washed twice with PBS and fixed with 2% paraformaldehyde in PBS, quenched, and then permeabilized in 0.3% Triton X-100/PBS. Nonspecific binding sites were blocked by sequential incubations in PBS solutions containing 2% normal goat serum, 2% egg albumin, and 5% nonfat dry milk. Affinity-purified anti-AR antibodies (approximately IgG concentration, 10 g/ml in 5% nonfat dry milk/PBS) or preimmune IgG were incubated with cells in the presence and the absence of a concanavalin A-fluorescein isothiocyanate conjugate (ICN 153245). Washed coverslips were treated with a biotin-conjugated antirabbit F(ab)Ј2 (Boehringer Mannheim 1214659) followed by streptavidin-Texas Red (Boehringer Mannheim 1131575). Fluorescence photomicrography was performed with a Nikon AFX-2A microscope on Kodak Ektachrome film exposed at E.I. 400-1200.

RESULTS
The AIR gene, described previously (5), contains sequences encoding the N-terminal 212 amino acids of the AR including both transmembrane domains and the intervening periplasmic domain that binds aspartate. The cytoplasmic domain of the human insulin receptor is fused to the cytoplasmic end of the second AR transmembrane domain, resulting in a receptor with an extracellular aspartate binding domain and an intracellular kinase domain.
Stable Expression of the AIR Kinase in 3T3 Cells-A eucaryotic expression vector pLJAIR was constructed to encode the AIR gene downstream of the retroviral promoter in plasmid pLJ. pLJAIR was transfected into the murine 3T3 line crip , and a neomycin-resistant transformant termed CA3 was iso-lated that expressed AIR. The rabbit antibody 9207 that specifically reacts with the AR recognized the AIR tyrosine kinase in CA3 cells. Fig. 1A shows an immunoblot with this antibody of particulate proteins from the 3T3 parental line (lane 1) and the CA3 line (lane 2), revealing an immunoreactive band of approximately 68 kDa specifically in CA3 cells. Expression of immunoreactive kinase activity is demonstrated in Fig. 1B; anti-AR immune complexes from CA3 cells contained histone H2B kinase activity that was not recognized by preimmune serum and absent in untransfected 3T3 crip cells. Phosphoamino acid analysis showed that this immunoreactive histone phosphorylation is entirely on tyrosine, as would be expected from the specificity the IR kinase domain (Fig. 1B,  inset).
Phosphorylation of AIR-Efficient autophosphorylation of the IR occurs when the two cytoplasmic domains of the receptor phosphorylate each other (9). AIR lacks covalent links between its subunits that would necessitate a dimeric configuration, however noncovalent interactions between the aspartate-binding N-terminal domains of the subunits and the transmembrane domains provide strong intersubunit forces. This is shown by the fact that the full-length AR is dimeric (6) and that when the cytoplasmic domain of one AR is deleted the two subunits remain associated (21). To determine whether the B, immune complex histone kinase activity from control and CA3 cells using anti-AR antibody. Radioactivity was quantified by densitometry of autoradiogram that had been exposed 4 h. Inset, autoradiogram from phosphoamino acid analysis of phosphorylated histones. Positions of phosphoamino acids were determined with ninhydrin-stained standards. ability of the IR to autophosphorylate in vitro is conserved in AIR, isolated CA3 cellular membranes were incubated with radiolabeled ATP and then immunoprecipitated with anti-AR antibody. Fig. 2 shows that whereas no phosphorylated bands were recognized in 3T3 parental cell membranes (lane 1), a single radiolabeled band of 70 kDa is found in CA3 membranes (lane 2). No band was observed in analyses of cytoplasmic proteins (not shown). The slight increase in apparent molecular mass of the protein in the phosphorylated band relative to the band in Fig. 1A is likely to be attributable to a different phosphorylation state. Phosphatase experiments demonstrated that AIR autophosphorylation levels regulate the exogenous kinase activity of the receptor in immunoprecipitates. Treatment of immunoprecipitates with 10 units acid phosphatase for 5 min followed by extensive washing reduced histone kinase activity by 50% (data not shown).
AIR was found to be phosphorylated in cells metabolically labeled with [ 32 P]orthophosphate. Fig. 3A shows a phosphorylated doublet in immunoprecipitates from solublized CA3 cells (lane 4) that is not recognized by preimmune serum (lane 2) and is absent in untransfected parental cells (lanes 1 and 3). Evidence that phosphotyrosine was present on AIR within the cell is presented in the phosphotyrosine Western blot in Fig.  3B. Proteins from the particulate fraction of CA3 cells had a predominant band of the appropriate size (lane 1) that was not present in parental cells (lane 2). Subcellular Location of AIR-The presence of an active fulllength AIR kinase in the particulate fraction of CA3 cells suggested that the receptor was folded to the active conformation and was inserted into cellular membranes. Double-label immunofluorescence revealed that significant amounts of AIR could be detected only in the perinuclear region corresponding to the endoplasmic reticulum and Golgi membranes and not in the plasma membrane by the aspartate receptor antibody (Fig. 4A). This specific subcellular location is suggested because doublelabel immunofluorescence of CA3 cells with AR Ab and concanavalin A-fluorescein isothiocyanate conjugate showed very similar staining patterns (Fig. 4, A and B). Confirmation that the AR antibody was indeed detecting AIR is provided by the absence of staining of CA3 cells when the antibody had been preincubated with excess AR protein (Fig. 4C). Likewise, no staining was observed with AR antibody on parental cells (Fig.  4D) or with preimmune antibody on CA3 cells (not shown). Because the N-terminal region containing the transmembrane domains has procaryotic sequences, the failure of transport to the exterior membrane is understandable.
Induction of DNA Synthesis by AIR-To see whether the chimeric receptor exhibited some of the insulin receptor's regulatory properties, we measured thymidine uptake in serumdeprived cells. AIR was found to induce thymidine uptake as does the native insulin receptor kinase. Subconfluent monolayers of CA3 and parental cells that had been serum-deprived for 30 h and then stimulated with calf serum or left untreated were labeled with [ 3 H]thymidine, and the incorporated radiolabel was measured. Fig. 5 shows that serum-deprived 3T3 parental cells lacking AIR incorporated minimal amounts of thymidine but upon serum stimulation took up large amounts of thymidine. CA3 cells, on the other hand, took up thymidine with or without stimulation, indicating that they were mitogenically activated independently of external growth factors. DISCUSSION We report here that the hybrid AIR can be expressed in mouse 3T3 cells and enters the endoplasmic reticulum where it is apparently retained. This intracellular protein is an active tyrosine kinase that is capable of stimulating DNA synthesis under serum deprivation conditions. Evidently the foreign extracellular sequences present in AIR and the absence of ␣ subunit glycosylation sites do not prevent the expression of a kinetically functional receptor but do affect its intracellular transport. These examples of endoplasmic reticulum retention are consistent with prevailing views of trafficking of receptors in general (22,23) and the IR in particular (24). Insulin receptors lacking most of the ␣ subunit sequences and that are therefore unable to bind insulin have been shown to be constitutively activated (25)(26)(27). These observations have been interpreted as evidence that the IR ligand binding domain is inhibitory; binding or deletion of this domain apparently relieves an inhibition of the cytoplasmic kinase via a conformational change. Additionally, such truncated insulin receptors were not transported to the plasma membrane but were retained in the endoplasmic reticulum when expressed in Chinese hamster ovary cells (24). Our studies add to the existing data on compatibility of foreign sequences because AIR contains a procaryotic four-helix bundle domain flanked by two procaryotic transmembrane domains. These N-terminal domains are tolerated by the mammalian expression machinery, giving rise to a kinetically functional membrane-associated receptor. This receptor is a good candidate to be expressed in membranes because we have been able to reconstitute the AR in vitro in the absence of secretory machinery (29).
Whereas the constitutively active truncated receptors in the studies referenced are believed to be monomeric, AIR is likely to form homodimers analogous to the native IR heterotetramer. Stable monomeric ␣␤ IR half-receptors have been shown to result after treatment of the native IR with dithiothreitol indicating that the ␣-␣ disulfide is essential for the ␣ 2 ␤ 2 configuration (10). We propose that AIR is dimeric because its ligand binding and transmembrane domains have been shown to dimerize via noncovalent intersubunit interactions (28). Our data support a model in which the AIR subunits transphosphorylate on tyrosine in vivo activating their activity toward exog-enous substrates. Our in vitro phosphatase experiments confirm that the high AIR exogenous kinase activity we see is regulated by autophosphorylation as in the case of the IR.
This report shows that AIR is permanently "turned on" in vivo, and we propose that as with the insulin-bound IR, transphosphorylation takes place and cellular substrates are phosphorylated that transduce signals to the metabolic machinery. The mechanism of activation may be independent of aspartate ligand, analogous to the active truncated insulin receptors cited above. It remains possible, however, that intracellular aspartate plays a stimulatory role in the initial activation and autophosphorylation of AIR. Indeed, we have occasionally observed 2-fold aspartate stimulation of AIR autophosphorylation in membrane preparations (not shown). However, these observations have not been reproducible so further evaluation of reaction conditions will be necessary to verify such an effect.