An altered IGF-I receptor is present in human leukemic cells.

We have characterized and analyzed IGF-I- and insulin-stimulated cell growth, receptor binding, and autophosphorylation in the human leukemic cell line HL-60. IGF-I-stimulated cell growth occurred at low (5 ng/ml) and insulin stimulated only at high (500 ng/ml) concentrations. Binding of 125I-IGF-I to partially purified plasma membrane proteins followed the characteristics of IGF-I receptor binding. 125I-IGF-I binding, as determined by chemical cross-linking, occurred to a 145-kDa protein. IGF-I, as well as insulin, stimulated the autophosphorylation of a 105-kDa band (pp105), but we could not detect a 95-kDa band corresponding to the known molecular mass of the IGF-I and insulin receptor beta-subunits. Phosphorylation of pp105 followed the dose-response characteristics of the IGF-I receptor. The phosphorylation of pp105 occurred at tyrosine and threonine, and the pattern of HPLC tryptic peptide maps showed marked differences when compared with that of a phosphorylated insulin receptor beta-subunit. Enzymatic deglycosylation of pp105 resulted only in a slight reduction of the molecular weight. These data suggest that pp105 is the beta-subunit of an IGF-I receptor variant with a higher molecular weight, similar to that found in fetal tissue. The HL-60 cell may acquire, at least in part, malignant growth characteristics through reexpression of the fetal version of the IGF-I receptor.

followed the dose-response characteristics of the IGF-I receptor.
The phosphorylation of pp105 occurred at tyrosine and threonine, and the pattern of HPLC tryptic peptide maps showed marked differences when compared with that of a phosphorylated insulin receptor @-subunit. Enzymatic deglycosylation of pp105 resulted only in a slight reduction of the molecular weight. These data suggest that pp105 is the @-subunit of an IGF-I receptor variant with a higher molecular weight, similar to that found in fetal tissue. The HL-60 cell may aquire, at least in part, malignant growth characteristics through reexpression of the fetal version of the IGF-I receptor.
Leukemic cells possess a growth advantage over their normal counterparts. Although there are indications that polypeptide growth factors and their membrane receptors may play an important role, it is not known which molecular mechanisms are responsible for this growth advantage. In theory, autonomous production of growth factors (autocrine hypothesis) or increased susceptibility to physiological growth-promoting agents (altered signal transduction hypothesis) could be responsible for their altered growth behavior. HL-60 cells (1) are highly dependent upon the presence of transferrin and insulin under serum-free culture conditions (2). Since supraphysiological concentrations of insulin (500 rig/ml) are required it is likely that this effect occurs through the IGF-I' receptor and that IGF-I is the physiological stim- There is also some evidence that IGF-I is acting through autocrine or paracrine mechanisms in other cell systems (5). The b-subunit of the insulin and IGF-I receptor is a transmembrane protein which contains tyrosine kinase activity in its cytoplasmic region (6)(7)(8)(9). The receptor P-subunits possess 84% homology to each other in their tyrosine kinase domain (10) and are regulated by similar control mechanisms (11). Insulin and IGF-I receptors have been isolated from several human tissues (10, [12][13][14][15][16][17]; the molecular mass of the (Ysubunit varies between 130 and 140 kDa (1, 9, 10, l&20, 21) and that of the p-subunit between 90 and 98 kDa (8,10,12,14,16,17). Here, we report our studies in which we have isolated glycoproteins from membranes of the human leukemic cell line HL-60 and partially purified them using a wheat germ agglutinin column. Specific binding of 1251-IGF-I to WGApurified HL-60 receptor could be demonstrated and covalent cross-linking of lz51-IGF-I to HL-60 IGF-I receptor revealed a specifically labeled band of a 145 kDa which is similar to the known a-subunit of the IGF-I receptor. However, in contrast to other reports (15,(22)(23)(24)(25)(26)(27)(28)(29) we could not identify in HL-60 cells the characteristic receptor @-subunit protein of 95 kDa indicating the presence of only minute or no amounts of normal insulin or IGF-I receptor kinase in these cells. Instead, we have identified an IGF-I-and insulin-dependent phosphorylation of a 105-kDa protein. Our studies suggest that pplO5 represents the P-subunit of an altered IGF-I receptor in HL-60 cells since IGF-I is the primary stimulus for this kinase.

RESULTS
The leukemic promyelocytic cell line HL-60 can be continuously grown in serum-free medium provided it contains insulin (2) or IGF-I (4), respectively, and transferrin. When we compared the growth stimulatory effects of various concentrations of IGF-I and insulin in the presence of a constant amount of transferrin we found IGF-I to be a more potent stimulator than insulin (Fig. 1).
To analyze IGF-I signal transmission by the target cell, glycoproteins from HL-60 membranes were partially purified with a WGA column. High affinity binding sites for IGF-I could then be demonstrated by displacement of 'Y-IGF-I with increasing concentrations of unlabeled IGF-I, IGF-II, and insulin. IGF-I competed in low concentrations, whereas displacement with IGF-II and insulin was less effective at the same concentrations; IGF-II, however, was more potent in displacement than insulin (Fig. 2). To identify the membrane protein responsible for the high  affinity binding, radioactively labeled IGF-I was incubated with aliquots of WGA eluate in the presence of a cross-linking agent (Fig. 3). Ligand specificity of the cross-linking process was evaluated through the additon of an excess (100 nM/liter) of unlabeled IGF-I, IGF-II, or insulin. ""I-IGF-I was crosslinked to a membrane protein yielding a protein with an apparent M, = 145,000, which is slightly above the known molecular weight of the IGF-I receptor, and an additional high molecular weight band. The production of both bands an antibody raised against the insulin receptor, but also crossreacting with the IGF-I receptor, separation of the immunoprecipitate on gel electrophoresis revealed an insulin-stimulated 105-kDa phosphoprotein (Fig. 4). When we compared phosphoproteins from HL-60 cells to the insulin receptor j-l-subunit of human adult skeletal muscle under identical experimental condit.ions, a clear molecular mass difference of 10 kDa between the HL-60 protein and the insulin receptor @-subunit became evident (Fig. 5).
Tryptic of pp105 isolated from HL-60 cells and of insulin receptor /3subunit purified from human skeletal muscle showed different characteristics (Fig. 6). Although phosphopeptide fragments were eluted at the same acetonitrile concentrations, there were marked differences in the phosphorylation pattern. Insulin receptor showed rather hydrophilic phosphopeptide fragments with maximal "P incorporation at 16.5% acetonitrile (fraction 61) and 18.2% acetonitrile (fraction 70). In contrast, the HL-60 derived 105kDa phosphoprotein showed only low "'P incorporation in these fractions and maximal incorporation in a more hydrophobic phosphopeptide fragment which eluted at 20.5% acetonitrile (fraction 83). In order to characterize the phosphorylation sites of the 105-kDa phosphoprotein of HL-60, we analyzed the phosphoamino acid content of the HL-60 HPLC fractions (Fig.7): phosphotyrosine is predominant in pplO5 from HL-60 (peaks 2-5) but, in addition, phosphothreonine is found (peak 6), which corresponds to the fraction with the highest 82P incorporation (20.5% acetonitrile) of all the tryptic peptides. When incubated with different IGF-I or insulin concentrations the phosphorylation of the pp105 occurred in a concentration-dependent manner. Fig. 8 shows the mean values of dose-response curves of five individual receptor phosphorylation experiments after incubation with IGF-I or insulin, respectively.
Half-maximal '*P incorporation after stimulation with IGF-I is found at a loo-fold lower concentration (0.1 nM) than with insulin (10 nM). Maximal '*P incorporation for both ligands is present at concentrations between 100 and 1000 nM. This extreme sensitivity of pp105 phosphorylation to IGF-I strongly suggests that pp105 represents the P-subunit of an IGF-I receptor with an altered molecular weight.
To determine whether the increase of the molecular weight The fractions of peaks 2-7 were lyophilized, subsequently dissolved in 300 ~1 of 6 N HCl, hydrolyzed for 2 h at 110 "C, and further processed as described under "Experimental Procedures." Phosphoamino acid standards were localized with ninhydrin and "2P-labeled amino acids from peaks 2-7 by autoradiography. "P incorporation is determined in percent of maximum '*P incorporation of the experiment. The curves represent the mean value and standard deviation of five individual experiments. of the p-subunit is due to differences in carbohydrate content, we treated the pp105 phosphoprotein with glycopeptidase F or neuraminidase, respectively. Fig. 9 shows that digestion with either enzyme led only to a slight reduction of the apparent molecular mass (2 kDa).

DISCUSSION
HL-60 cells were derived from a patient with acute promyelocytic leukemia (1). In culture they can be continuously grown in defined serum-free medium only when IGF-I or insulin, respectively, and transferrin are present. Our studies demonstrate that the growth-stimulatory effect of IGF-I is mediated by high affinity binding sites for IGF-I present in the membranes of HL-60 cells.
Cross-linking with '251-IGF-I revealed a specifically labeled 145-kDa band which is similar to the a-subunit of IGF-I receptors analyzed in other cells (10, [18][19][20][21]. However, this band migrates electrophoretically somewhat slower, which may be due to a different oligosaccharide content of the (Ysubunit (33) or to tissue-specific differences (34). An additional specifically labeled higher molecular weight band may be a subunit polymer of the IGF-I receptor since its production can be equally well inhibited by excess cold IGF-I. (The 68-kDa band appearing in Fig. 3  When membrane glycoprotein extracts from HL-60 cells were immunoprecipitated with an anti-insulin receptor antibody which also cross-reacts with the IGF-I receptor protein we identified a protein pp105 which is larger than the fisubunits of insulin receptor (12, 13, 15-17, 22-25,35,36) and IGF-I receptor (8-10, 14, 37) in different tissues.
In addition, this protein is phosphorylated not only in tyrosine but also in threonine residues. Since the tryptic peptide map of insulin-stimulated pp105 is different from the peptide map of insulin-stimulated insulin receptor P-subunit, it is unlikely that pp105 is related to the insulin receptor. This conclusion is also supported by the marked insensitivity of pp105 phosphorylation in response to insulin, whereas a very high sensitivity in response to IGF-I is found. This high sensitivity to IGF-I strongly suggest that the protein which is recognized by our antibody represents an altered IGF-I receptor P-subunit. This is also in accordance with our cell culture studies in which we have identified low dosages of IGF-I as a potent stimulus of HL-60 cell growth.
Molecular weight heterogeneities of the a-subunit of insulin and IGF-I receptors of different tissues have been identified and found to be due to differences in oligosaccharide content (14,36,38). However, in these studies there was no evidence for a molecular weight difference in the P-subunit as observed in our experiments. We tried to deglycosylate the pp105 with neuraminidase and glycopeptidase F in order to determine whether the molecular weight increase of the pp105 is due to an increased oligosaccharide content. Only a slight (2 kDa) decrease in the molecular mass was observed, which indicates that the molecular mass increase is only partly due to different glycosylation.
A similar molecular weight increase of the P-subunit, in this case the insulin receptor (40), has been described for another malignant cell line (U-937 monocytes). However, there is no evidence that the P-subunits of erythrocytes and monocytes from non-leukemic donors possess an altered flsubunit (17,41). These results and our studies raise the question whether the molecular weight increase is specifically associated with the malignant state of these cells, and whether there are consequences for cell proliferation and differentia-kDa P-subunit has recently been reported to be expressed in fetal, but not in adult, muscle (42). This could imply that during a developmental process alterations in the signaling pathways in the same cell type qualitatively alter the response to the same growth factor at different stages of development.
Prehn (43) has hypothesized that tumor-specific transplantation antigens found in many induced tumors may be altered growth factor receptors which make the cell to react in a different manner to growth regulators provided by adjacent normal cells.
Therefore, it could be that the altered IGF-I receptor identified in our experiments is at least partly responsible for the malignant transformation of the HL-60 cell. Since a similar IGF-I receptor has been described in a fetal tissue, the malignant transformation may be partially conferred to the leukemic cell through genetic reactivation of a signal transduction program present in early development. To further address this question studies are in progress to analyze the structure of the IGF-I receptor in normal human promyelocytes and myeloblasts as well as to study the genetic mechanism underlying the alteration of the IGF-I receptor present in HL-60 cells.