Oxford OX1 3QR, United Kingdom and the YDepartment of Biochemistry and the **Howard Hughes Medical Institute,. University of Texas Southwestern Medical ...
VOl. 266,No. 7.0, Issue of July 15, PP. 13405-13410,1991 Printed in U.S.A.
THEJOURNAL OF BIOLOGICALCHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.
Autophosphorylation of Soluble Insulin Receptor Protein-tyrosine Kinases ’H NMR SPECTRAL CHANGES OBSERVED DURING PHOSPHORYLATION OF MOBILE TYROSINE RESIDUES* (Received for publication, February 21, 1991)
Barry A. Levine#& JeremyM. TavareTII, Erica Alejos**,Beatrice ClackT, Nadira SayedT, and Leland EllisT**$$ From the $Inorganic Chemistry Laboratory and Oxford Center for Molecular Sciences, Uniuersity of Oxford, S o u t h P a r kRoad, Oxford OX1 3QR, United Kingdom and the YDepartment of Biochemistry and the **Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, Texas75235-9050
Autophosphorylation of a soluble -48-kDa deriva- While phosphorylation at sites 1158,1162, and 1163 correlates tive oftheinsulinreceptorprotein-tyrosine kinase with an increase in the specific activity of the enzyme toward occurs at multiple tyrosine residues (analogous to ty- a variety of exogenous substrates, the functional role of sites rosines 1158, 1162, and 1163 in the kinase homology 1328 and 1334 is unclear at present (Tornqvist et al., 1987; region of the native receptor and tyrosines 1328 and Tavar6 and Denton, 1988; Tavar6 et al., 1988; Tornqvist et 1334 in the carboxyl-terminal tail) and is accompanied al., 1988; White et al., 1988). by an increase in the specific activity of the enzyme The cytoplasmic protein-tyrosine kinase domain of the toward exogenous substrates. A comparison of‘H NMR human insulin receptor has been engineered and expressed as spectra of -48- and -38-kDa forms of enzyme (the an active monomeric enzyme both in stably transfectedmamlatter generated by tryptic deletion of -10 kDa from malian cell lines (Ellis et al., 1987) and in insect Sf9 cells via the carboxyl terminusof the -48-kDa protein) allows a recombinant Baculovirus vector (Ellis et al., 1988, Herrera a correlation of observed mobile tyrosine resonances to two of the knownsites of autophosphorylation(res- et al., 1988).This soluble enzyme (401 amino acids, M , 45,297, idues 1328 and 1334). Furthermore, spectra acquired designated the -48-kDa enzyme given its mobility on SDSduring autophosphorylation of the -48-kDa enzyme PAGE)’ exhibits a low basal level of kinase activity following reveal a rapiddownfield shift in the resonances of purification, which increases during autophosphorylation these mobile tail tyrosines consistent with their phos- (Cobb et al., 1989). Furthermore, the sites of autophosphoryphorylation (as confirmed by two-dimensional tryptic lation of this soluble enzyme are entirely typical of those of the intactwild-type receptor ( i e . at sites inthe soluble enzyme phosphopeptidemappingperformedunderidentical conditions). This experimentalstrategy now provides corresponding to tyrosines 1158, 1162, 1163, 1328, and 1334 a means by which to monitor protein-tyrosine kinase of the native receptor (Tavar6 et al., 1991). Thus, even though synthesized free of its membrane anchor and now insulinautophosphorylation in solution in real time. independent, this soluble enzyme exhibits the major functional properties of the kinase of the wild-type transmembrane receptor. In the present study, the analysis of proton Autophosphorylation specifically on tyrosine residues is nuclear magnetic resonance (‘H NMR) spectra of the soluble associated with an increase in the specific activity of a number -48-kDa enzyme, an -38-kDa derivative of the -48-kDa of protein-tyrosine kinases, including those associated with enzyme (with a deletion of -10 kDa from the carboxyl tertransmembrane receptors for polypeptide hormones such as minus of the protein), and the -48-kDa enzyme during auinsulin(Hunterand Cooper, 1985). Insulin binding to its tophosphorylation reveals a correlation of mobile tyrosine receptor on intact cells results in the rapid autophosphoryla- residues observed in the -48-kDa spectra to the two sites of tion of the receptor @subunit on tyrosines 1158, 1162, and receptor autophosphorylation that reside in the carboxyl1163,which reside in the kinase homology region of the terminal tail of the enzyme (i.e. tyrosines 1328 and 1334). receptor primary sequence, and on tyrosines 1328 and 1334 in the carboxyl-terminal tail of the receptor (the numbering MATERIALS ANDMETHODS of insulin receptor residues is according to Ebina et al., 1985). The -48-kDa kinase was purified in 10-20-mg quantities from 800
* This work wassupported by the National Institutes of Health (to L. E.), the Howard Hughes Medical Institute (to L. E.), and the Medical Research Council (to B. A. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 3 Associate member of the Oxford Center for Molecular Sciences. I( Medical Research Council Traveling Fellow. Present address: Dept. of Biochemistry, School of Medical Sciences, University of Bristol, Bristol BS8 lTD, UK. $$ To whom correspondence should be addressed Howard Hughes Medical Inst. and Dept. of Biochemistry, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, T X 752359050.
ml of insect Sf9 cells grown in suspension as described (Ellis et al., 1988; Cobb et al., 1989). Truncation of the -48-kDa enzyme to generate an -38-kDa species was achieved by limited proteolysis with trypsin, which removes -10 kDa from the carboxyl terminus of the -48-kDa enzyme (TavarC et al., 1991). The -48-kDa enzyme is monomeric, as judged both by ultracentrifugation (Cobb et al., 1989) The abbreviations used are: SDS-PAGE, polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate; ‘H NMR, proton nuclear magnetic resonance spectroscopy; -48-kDa enzyme, a recombinant soluble derivative of the cytoplasmic protein-tyrosine kinase domain of the human insulin receptor; -38-kDa enzyme, derived by limited digestion of the -48-kDa enzyme with trypsin; DTT, dithiothreitol.
13405
Autophosphorylution of Soluble Insulin Receptor Kinases
13406
2,6H?3,5H
0‘
h
>. . . . I , . . . , . . . ,
8.00
I 1 . 8 , . . . .
. I . . , . . . .
250
7.00
!
.
.
.
l
,
r
n
6.50
PPM
FIG. 1. Aromatic ‘H NMR spectra of the soluble -48- and -38-kDa insulin receptor protein-tyrosine kinases. Spectra were recorded in 10 mM NaHC03-buffered DzO (pH 7.8) plus 2 mM DTT and 5 mM MgCl, at a temperature of 300 “K (27 “C) with an enzyme concentration of 200 PM for the -38-kDa enzyme and 820 PM for the -48-kDa enzyme. No discernible changes were observed in the spectra upon the addition of MgCl, to the buffered enzymes. Instrument settings were as described under “Materials and Methods.” Note that the gain of this and subsequent (Fig. 5) aromatic spectra is 8 X that of the aliphatic spectra of Figs. 2 and 6. The top two traces are two-pulse (Carr-Purcell) spin echo spectra (see “Materials and Methods”) of the -48- and -38-kDa enzymes. The positions of resonances due to theC-2 and C-4 ring protons of histidines (His 2H and 4H, respectively), the C-3,5 and C-2,6 ring protons of respectively), and theC-2,6 and phenylalanines (Phe 3,5H and 2,6H, C-3,5 ring protons of tyrosines (Tyr 2,6H and 3,5H, respectively) are indicated. and by gel filtration chromatography, while the -38-kDa protein has some propensity to dimerize even in the presence of 2 mM DTT? Autophosphorylation of the enzymes was carried out as described in the figure legends. Phosphoamino acid analysis and two-dimensional tryptic phosphopeptide mapping were carried out as described in the figure legends (see also Tavar6 and Denton (1988) and Tavar6 et al. (1988). NMR spectra were recorded on a Brucker 500-MHz spectrometer using a sweep width of 6024 Hz. A gated presaturation pulse (0.4-s duration) was used to suppress the residual HzO signal, and spectra of 256 scans were collected in 8192-byte memory blocks with a total accumulation time of 5 min. A line broadening of 1 Hz wasused prior to Fourier transformation. Two-pulse (Carr-Purcell) spinecho experiments (90-7-180-7)were carried out using an interpulse delay ( 7 ) of 60 ms to obtain inverion of doublets with resolvable coupling constants of-7 Hz (Jardetzky and Roberts, 1981). Kinase autophosphorylation was monitored directly during incubations in 0.5-ml reactions in a 5-mm diameter quartz NMR tube with conditions as described in the figure legends. RESULTS AND DISCUSSION
‘HNMR Spectra Reveal Distinct Changes in the Aromatic Region following Deletion of the Carboxyl-terminal Tail of the -48-kDa Enzyme-NMR spectroscopy is sensitive to both main chain and side chain motions in the range of nanoseconds to a few seconds. Thus, changes in conformation can be directly monitored as the resonance energy and relaxation properties of specific spectral signals reflect the spatial disposition and structuralflexibility of the corresponding groups (Jardetzky and Roberts, 1981; Wuthrich, 1986; Ernst et al., 1987). In Figs. 1 and 2, the aromatic (8.25-6.00 ppm) and aliphatic (3.50 to -0.15 ppm) regions, respectively, of a oneS. Hubbard, W. A. Hendrickson, and L. Ellis, unpublished observations.
3.00
2.50
2.00
q.50
4.00
.50
0.0
PPM
FIG. 2. Aliphatic ‘H NMR spectra of the soluble -48- and -38-kDa insulin receptor protein-tyrosine kinases. The top two traces are Carr-Purcell spectraof the -48- and -38-kDa enzymes. The positions of resonances due to the &methylene protons of arginines (Arg 6CH2),the c-methylene protons of lysines (Lys tCH2),the c-methyl protons of methionines (Met SCH,), the P-methyl protons of alanines (Ala PCH,), the y-methyl protons of threonines (Thr yCH3), and the broad envelope derived from the methyl protons of are indicated. Several sharp isoleucines, leucines, and valines (CH,) signals in these spectrathat derive from small molecules in the sample (e.g. DTT at-2.70 ppm (included in the kinase buffers) and acetate at -2.20 ppm) are indicated (*).
dimensional ‘H N M R spectrum of the soluble -48-kDa enzyme are illustrated. Given the large molecular mass of the enzyme (401 residues, M , 45,297) andits relatively slow overall rotational tumbling time in solution, the spectrum is comprised of a broad envelope with relatively few sharp and well resolved peaks. However, closer inspection reveals that discrete signals are present, and they provide information concerning both residue type and chemical environment as expected on the basis of a statistical analysis of resonance positions in proteins (cf. Wuthrich, 1986).Upfield of the major envelope of methyl protons (i.e. cO.9 ppm) occur methyl group signals (e.g. of isoleucine, leucine, and valine residues) from side chains closely packed to aromatic rings (hence, the ring current shift) indicative of hydrophobic cluster(s). Similarly, the upfield shifted aromatic signals (6.80-6.20 ppm) also provide a“fingerprint” for the hydrophobic “core” of the enzyme. Consistent with such a core, downfield of 8 ppm occur backbone amide protons from residues in regions that are least accessible to solvent (after the enzyme has been in D 2 0 for 24 h). A further diagnostic for secondary folds is signals at -5.3-5.4 and -5.1 ~ p m which , ~ derive from backbone C, protons of residues in regions of p-sheet (Dalgarno et al., 1983). Other signals of the aliphatic spectrum directly assignable on the basis of chemical shift position include the &methylene protons of arginines (-3.20 ppm)and the tmethylene protons of lysines (-3.00 ppm). The correlation of other signals to amino acid type is based on the multiplicity and position of resonances. One can distinguish signals of doublet character (e.g. the C-2,6 ring protons of phenylalanines (-7.25 ppm), the C-2,6 and C-3,5 ring protons of tyrosines (-7.07 and -6.80 ppm, respectively), the @-methylprotons of alanines (-1.38 ppm), and the ymethyl protons of threonines (-1.22 pprn)) from those of B. A. Levine and L. Ellis, unpublished observations.
Autophosphorylation of Soluble Insulin Receptor Kinases triplet or singlet character (e.g. the C-2 and C-4 ring protons of histidines (-7.73 and -7.00 ppm, respectively), the C-3,5 ring protons of phenylalanines (-7.32 ppm), and the t-methyl protons of methionines (-2.08 pprn)) by the use of a twoT) sequence where signal inverpulse ( ~ O - T - ~ ~ O - Carr-Purcell sion (doublets appear negative (down in the top two traces of Figs. 1 and 2) and singlets and triplets appearpositive (up in the toptwo traces ofFigs. 1 and 2)) is afunction of the coupling constant and thedelay time (7).This modulation of signal appearance is possible provided that the spin-spin relaxation time (T2)is sufficiently long for J-coupling to be resolvable, i.e. the observation of modulation is also an indicator of mobility of the corresponding groups (i.e. long Tz sharper signal more mobile). With the delay time (T = 60 ms)in the experiments of Fig. 1, an inversion of two Jcoupled doublets is observed in positions typical of the C-2,6 (-7.07 ppm) andthe C-3,5(-6.80 ppm) ring protons of tyrosine residues. Note that theC-3,5 protons aremore mobile than the C-2,6 protons (i.e the former signal is more intense in Fig. l),perhaps a consequence of their increased exposure away from the polypeptide backbone or signal overlap of the C-2,6 protons with other resonances that contribute topositive intensity (e.g. C-4 protons of histidine residues other than those resolved in this spectrum). In contrast,singlets consistent with the C-2 and C-4 protons of histidines (-7.73 and -7.00 ppm, respectively) are notinverted, and they appear up in the Carr-Purcell spectrum of Fig. 1. (Further supporting the assignment of these signals to histidine ring protons is the pH sensitivity of their chemical shift positions.3) All signals in these Carr-Purcell spectra (Figs. 1 and 2) derive from groups with segmental mobility greater than the restof the molecular framework of the enzyme (which could derive from their exposure or location in more flexible regions of the enzyme) and thus provide prescriptive signals for different aspects of tertiary structure to monitor during subsequent phosphorylation experiments. Protein-tyrosine kinases often have a carboxyl-terminaltail that extends beyond the endof the tyrosine kinase homology region of their primary sequence (Hunter and Cooper, 1985; Hanks et al., 1988). In the human insulin receptor, a pair of hydrophobic residues (leucines 1262 and 1263) demarcate this boundary. As so defined, the tailof this receptor is comprised of 92 amino acids (Mr10,465),a numberof which are charged (17.3% glutamate, 9.2% arginine, 6.1% lysine, 5.5% aspartate) or polar (9.2% serine, 5.5% asparagine). There are only two tyrosine residues (1328 and 1334) in this stretch of sequence that are known to be sites of autophosphorylation of the wildtype insulin receptor both in vivoand in vitroand of the -48kDa soluble kinase in vitro(see Introduction). Limitedtryptic digestion of the -48-kDa enzyme removes -10 kDa from the carboxyl terminus of the protein (as assessed by the relative mobility of the two enzymes on SDS-PAGE; while tryptic cleavage following either lysine 1264 or 1283 would yield a deletion of the size observed, the precise location of the site of cleavage is not known a t present). Thus, two-dimensional tryptic phosphopeptide maps of the -38-kDa enzyme followingautophosphorylation lack thetryptic phosphopeptide (SYEEHIPYTHMNGGK) thatincludes both of the tail tyrosines (residues1328 and 1334) (Tavari. et al.,1991). Further, this truncated -38-kDa enzyme is still catalytically active and exhibits both mono- and diphosphorylation of synthetic dodecapeptide substrates that containmultiple tyrosine residues.4 As -20% of the enzyme mass has been deleted, one might expect that a correspondingly large difference would be
-
-
Levine, B. A., and Ellis, L. (1991) J. Biol. Chem. 266, 1236912371.
13407
observed in the spectra of the two enzyme species. Surprisingly, the most conspicuous alteration of the NMR spectrum of the -38-kDa enzyme (compared with that of the -48-kDa enzyme) occurs in the aromatic region, with reduction in the intensity of signals due to theC-2,6 and C-3,5 ring protons of tyrosine (the signals at -7.07 and -6.80 ppm, respectively, of Fig. 1).Thus, thisproteolytic deletion is accompanied by the loss of signals due to some of the most mobile tyrosine residues of the -48-kDa enzyme as is also evident in a comparison of the Carr-Purcell spectra of the two enzymes (cf. the top two traces of Fig. 1).Furthermore, there is also a loss of histidine signal intensity (5of the 10 histidines of the -48-kDa enzyme reside in the tail), but there is little change in the intensity attributable tophenylalanines (there are five phenylalanines (based on cleavage following residue 1264) and no tryptophans in the tail). These observations suggest that some but not all of the most mobile residues visible in the aromatic region of the NMR spectra of the -48-kDa enzyme derive from tyrosine and histidine residues that reside inthetail end of the carboxyl terminus of the protein. Retention of the Fold of the Enzyme Core following Deletion of the Carboxyl Terminus of the -48-kDa Enzyme; Structural Organization of the Carboxyl-terminal Tail withRespect to the Enzyme Core-In contrast to these conspicuous changes in the tyrosine and histidine signals as would be expected if those residues were at a molecular tail region, residual aromatic region resonances and the features of the aliphatic spectra of the -48- and -38-kDa enzymes are quite comparable (Figs. 1and 2). Since strong spectral homology of signal dispersion is observed for the resonances that report on the tertiary fold (see above), the spectra for the two enzyme species yield direct evidence for the retention of the native 2'
5'
0- -Pi
- P-s - P-T
- ori FIG. 3. Phosphoamino acid analysis of the -48-kDa enzyme. -48-kDa enzyme (350 p?d) was incubated in 50 mM Tris (pH 7.8), 2 mM DTT, and 8 mM MgCl, for 10 min at 27 "C.Phosphorylation was initiated with the addition of ["PIATP (4 mM; 400 cpm/ pmol). After incubation at 27 "C for 0, 2, and 5 min, an aliquot was removed forthe assessment of exogenous kinase activity with peptide RRDIFENDYFRK (see text), and the reaction was terminated by the addition ofLaemmli sample buffer. Proteins wereresolved by SDS-PAGE (10%acrylamide). "YP-Labeledkinase bands at each time point were electroeluted from SDS-polyacrylamide gels, digested with trypsin, and hydrolyzed with HCl. The resulting phosphoamino acids were separated by chromatographyat pH 3.5 and visualized (at 2 ( 2 ' ) and 5 (5') min) by autoradiography (see "Materials and Methods"). The migration of inorganic phosphate (Pi)and carrier phosphoserine ( P - S ) ,phosphothreonine ( P - T ) and , phosphotyrosine ( P - Y ) ,and the origin of sample application (ori) are indicated.
Autophosphorylation of Soluble Insulin Receptor
13408 2’
Kinases TpP
0
I
All-0
5‘
0
l
8.0
’
l
7.8
’
l
7.6
’
l
7.4
’
l
7.2 PPM
’
l
’
l
7.0 6.4 6.6 6.8
’
l
’
l
FIG. 4. Two-dimensional tryptic phosphopeptide maps of the -48-kDa enzyme. ‘”P-labeled kinase bands a t each time point
FIG. 5. Aromatic spectra of the -48-kDa enzyme recorded before and after autophosphorylation. The basal spectrum (0
(as described in the legend to Fig. 3) were electroeluted from SDSpolyacrylamide gels, digested with trypsin, and phosphopeptides were analyzed followingseparation intwo dimensions and autoradiography (see “Materials and Methods”). Phosphopeptide maps following autophosphorylation for 2 (2’) and 5 (5’)min are illustrated. It is apparent, with reference to the cartoon in the lower panel (a key to the identification of tryptic phosphopeptides derived from the wildtype insulin receptor P-subunit), that phosphorylation has occurred a t the major sites of insulin receptor autophosphorylation, i.e. tyrosines 1158, 1162, and 1163 in the core and tyrosines 1328 and 1334 in the tail. In brief, the three major tyrosine autophosphorylation sites (residues 1158, 1162, and 1163) are recovered as five phosphobis- (B2 peptides (sequence DIYETDYYRK), which are mono- (CI), and B3), or tris- (AI and A2) phosphorylated, and are cleaved by trypsin a t arginine 1155 and either arginine 1164 (CI,B3, andA2) or lysine 1165 (B2 and A I ) . Tyrosines 1328 and 1334 are recovered as a single bisphosphopeptide (sequence SYEEHIPYTHMNGGK) that migrates as peptide B1. A peptide that includes an additional tyrosine autophosphorylation site(s) (probably one or more of residues 965, 972, and 984) migrates as peptide D. The arrows in all three panels indicate the origin of sample application. The dotted ovals indicate the position of migration of a markercompound, dinitrophenyl lysine.
min) was recorded in a solution of 50 mM deuterated Tris (pH 7.8), 2 mM DTT, 4 mM ATP, and 350 p~ -48-kDa enzyme. Phosphorylation was initiated by the addition of MgC12 to a final concentration of 8 mM with resulting dilution of the enzyme concentration by -5%. Since the addition of either MgCl, or ATP alone could cause conformational effects, the binding of either reagent alone was assessed by its separate addition to thebasal enzyme. A relative paucity of spectral changes due to either addition was observed. Sequential addition in either order of reagent led to thedetection of the same time-dependent spectral effects (see text) upon autophosphorylation. Tyr-P, phosphorylated tyrosine residues. Other annotations of spectral peaks are as defined in the legend to Fig. 1. The lower trace is that of the basal enzyme in the presence of ATP, and themiddle trace is that recorded 11 min after the addition ofMgC12. The top trace is a difference spectrum (the basal spectrum subtracted from the spectrum a t 11 min), which illustrates the major spectral changes that occur over the course of the incubation. The pH of the solution was checked after each addition and at theend of autophosphorylation.
significant mass of the tail were freely moving, one would expect to see more discernible differences (-48 uersus -38 kDa) in regions of the spectrum derived from residues other than tyrosine and histidine). One can therefore infer in the structure of the protein core following proteolytic digestion. A more “floppy” conformational average, however, results as latter case that removal of the tail (by peeling it away) from the surface of the enzyme might well change the chemical seen from the change in relaxation time of signals within the character of discrete regions of the protein surface, which envelope comprised of resonances of the methyl groups of might be manifest during the interaction of the enzyme with isoleucines, leucines, and valines ( i e . centered -0.9 ppm). itself or with peptide substrates. While it is difficult to exclude Cleavage of the tail results in the loss of only 2 alanine the possibility of millisecond fluctuations of the carboxyl residues as opposed to 9 residues that could contribute signals terminus without monitoring the effect of elevated temperaunder this resonance. Yet the relaxation-filtered Carr-Purcell ture on the spectra observed for the enzymes (which was spectra indicate that the latterresidue type possesses signifi- avoided in this study in order to maintain kinase activity), cantly increased segmental freedom upon removal of the two independent observations argue for the association of the carboxyl-terminal region that is likely to reflect agreater tail with the core enzyme structure. Given the known enzyme flexibility between elements of secondary structure of the concentration used for NMR study, comparatively poor sig-38-kDa enzyme that include certain of its methyl groups nal-to-noise was observed for the -38-kDa species suggesting (hence, a more floppy conformational average). Considering that, unlike the -48-kDa protein, the smaller enzyme was the intrinsic mobility of tail tyrosine and histidine residues partially aggregated. This conclusion is supported by gel fil(see above) and that other regions of the spectrum are not tration studies of the -38-kDa enzyme, which show that the dramatically different when -48- and -38-kDa spectra are molecule has atendency to dimerize under experimental compared, these observations can be rationalized either in conditions where its -48-kDa progenitor is monomeric.*Agterms of “wagging” of the tail per se (with consequent inter- gregation is presumably facilitated by the loss of the carboxyl mediate exchange conditions pertaining that give rise to in- terminus (which is of course present inthe -48-kDa enzyme). trinsically broad signals) or asa consequence of the fact that In summary, the datasuggest that intermolecular association a proportion of the 92 residues of the tail interactswith the of the kinase domain is hampered by the carboxyl terminus surface of the enzyme, with the mobile tyrosine and histidine and that interactions of the tailwith the core enzyme thereby residues (see above) being more exposed to solvent (i.e. if mask a potential “dimer” interface. The possibility that au-
Autophosphorylation of Soluble Insulin Receptor Kinases
13409
resonances at -7.07 and -6.82 ppm (attributable in part to tyrosines 1328 and 1334 in the carboxyl-terminal tail (see above)), changes that are entirely consistent with the alteration in chemical shift position observed upon phosphorylation of tyrosine residues in synthetic peptides (Levine et al., 1991). The onset of autophosphorylation also results in the displacement to low field of several histidine resonances (Fig. 5) whose u* n corresponding residues experience more deshielded microenvironments in the phosphorylated form of the enzyme (e.g. the two histidines (residues 1331 and 1336) in the vicinity of the tyrosine phosphorylation sites of the carboxyl-terminal tail). Relatively minor spectral changes occur for the remaining aromatic side chains, while prominent downfield shifts are observed in three areas of the aliphatic spectrum (Fig. 6) at chemical shift positions expected for (i) the t-methyl protons of methionine residues (the resonance centered at -2.08 ppm), (ii) they-methyl protons of threonine residues (-1.12 ppm), and (iii) the y-methyl protons of isoleucine and valine 3.0 2.5 2.0 1.5 1.0 .5 0.0 residues and the &methyl protons of isoleucine and leucine PPM FIG. 6. Aliphatic spectra of the -48-kDa enzyme recorded residues (-0.90 ppm) (cf. the difference spectrum, top trace of changes observed at these positions, before and after autophosphorylation. Conditions are as de- Fig. 6). In contrast to the scribed in the legend toFig. 5. Glu yCH2, y-methylene protons of the overall spectrum of the enzyme (and in particular those glutamic acids; Tyr @CHz,@-methylene protonsof phosphorylated signals derived from the hydrophobic core) remains largely tyrosines. Other annotationsare as described in the legend to Fig. 2. unperturbed.Thus, autophosphorylation is not associated The upfield shoulders of the broad methyl envelope centered at -0.9 with a large reorganization of the enzyme core structure in ppm are shown at 4 X higher gain in the inserts at the right of the that there is no evidence suggestive of a partial unfolding of aliphatic spectra recordedat 0 and 11 min. Thetop trace is a difference spectrum (the basal spectrum subtracted from the spectrum at 11 a significant proportion of the tertiary structureof the enzyme min), which illustrates themajor spectral changes that occur over the as a consequence of the increase in its negative charge with course of the incubation. autophosphorylation of multiple tyrosines. Rather, the data are consistent with small relative movements of secondary tophosphorylation alters either the core enzyme fold and/or structural elements toaccommodate this modification. the conformational properties of the carboxyl terminus was Examination of the spectrum of the -38-kDa enzyme durtherefore next investigated by 'H NMR. ing autophosphorylation reveals a comparable sequence of Distinct Changes in the Aromatic Spectrum of the -48-kDa spectral alterations (apartfrom the absence of the prominent Enzyme Transpire during Autophosphorylation-The addi- tyrosine resonances at -7.07 and -6.82 ~ p m )The . ~ spectral tion of MgC12 (8 mM) to a buffered (50 mM Tris (pH7.8) plus perturbations observed for both kinases, changes in the chem2 mM DTT) solution of the -48-kDa enzyme (350 p M ) and ical shift positions of a variety of residue types (including ATP (4 mM) results in the rapid autophosphorylation of the those of tyrosines), are adirect result of phosphorylation. The enzyme exclusively on tyrosine residues, as illustrated by the fact that the overall spectral characteristics of the two prophosphoamino acid analysis presented in Fig. 3. It is of teins that reflect their tertiary fold are apparently unaltered interest to note that the addition of protamine to such reac- (e.g.the ring-shifted methyl protons at CO.9 ppm) (cf. Fig. 6) tions also results in the phosphorylation of serine residues on is consistent with the absence of an internal reorganization the enzyme (Tavar6 et al., 1991) presumably by stimulating a of the protein conformation. Thus the spectral changes obminor contaminating serine-specific protein kinase. Auto- served during autophosphorylation are likely to be, at least in phosphorylation reaches a maximum within 5 min under these part, due to localized rather than global changes (see above). Autophosphorylation of the -48-kDa enzyme occurs at core conditions with a concomitant activationof the enzyme -90fold (as measured by exogenous kinase activity with the tyrosines 1158, 1162, and 1163 and tail tyrosines 1328 and peptide RRDIFENDYFRK). Furthermore, two-dimensional 1334 (cf. Fig. 4) (Tavar6 et al., 1991) within the first 5 min of tryptic phosphopeptide mapsdemonstrate thatthe phos- incubation. The fact that the rapid changes in the aromatic phorylation of multiple tyrosines occurs within 2 min (and 'H NMR spectra are observed during the same time course, reaches a maximum level by 5 min) and that all five of the and that the most conspicuous aromatic signals derive from known sites typical of the native insulin receptor are utilized, mobile tyrosines that are absent in spectra of the -38-kDa i.e. at tyrosines analogous to native receptor residues 1158, enzyme, suggests that theNMR changes observed correspond 1162, and 1163 in thekinase homology region and at tyrosines to the phosphorylation of the two tail tyrosines, i.e. residues 1328 and 1334 inthe carboxyl-terminal tail (Fig. 4) (for 1328 and 1334. The exposed location of tyrosine phosphates further discussion, see Tavar6 et al. (1991)). Thus, within 5 in the activated enzyme is indeed shown by the appearance min of the onset of autophosphorylation, tyrosine residues in the -48-kDa aliphatic spectrum of the PCH, protons of that reside in two distinct regions of the receptor primary phosphorylated tyrosines (Fig. 6), whichwould not be examino acid sequence (i.e.at three sites in the core and attwo pected to be resolved unless derived from residues that are sites in the tail)undergo autophosphorylation. located on a more flexible and/or exposed portion of the With the onset of autophosphorylation,distinctive 'H molecular backbone. The fact that phosphorylation of the NMR spectral changes are observed to evolve in slow ex- core tyrosines (1158,1162, and 1163) of the -38-kDa enzyme change. These changes are apparent within 2 min of the does not result in a significant change in the tendency of the addition of divalent cations to theenzyme and areillustrated enzyme to aggregate suggests that theputative dimer interface a t 11min in Fig. 5. The most prominent perturbations in the revealed by deletion of the carboxyl terminus is not influenced aromatic region are the downfield shifts that occur for the by the state of phosphorylation of these sites. '
1
1
1
~
"
"
~
"
"
~
"
"
~
"
"
~
"
"
~
"
"
,
"
13410
Autophosphorylation of Soluble Insulin Receptor Kinases
Ellis, L., Morgan, D. O., Clauser, E., Roth, R. A., and Rutter, W. J . (1987) Mol. Endocrinol. 1, 15-24 J . Virol. Ellis. L.. Levitan,. A... Cobb. M. H.. and Ramos.. P. (1988) . 62,1634-1639 Ernst. R. R.. Bodenhausen. G.. and Wokaun. A. (1987) Princioles of Nuclear Magnetic Reso&nckin One and TLo Dimensions, Clarendon Press, Oxford Hanks, S. K., Quinn, A. M., and Hunter, T. (1988) Science 241,4252 Herrera, R., Lebwohl, D., de Herreros, A. G., Kallen, R.G., and Rosen, 0. M. (1988). J . Biol. Chem. 263,5560-5568 Hunter, T., and Cooper, J. A. (1985) Annu. Rev. Biochen. 54, 897930 Jardetzky, O., and Roberts, G. C. K. (1981) N M R in Molecular Acknowledgments-We thank our colleagues in Oxford and Dallas, Biology, Academic Press, New York especially Drs. Dan Raleigh and Erik Schaefer and Prof. R. J. P. Levine, B.A., Clack, B., and Ellis, L. (1991) J. Biol. Chem. 266, Williams, for helpful discussions during the course of this work and 3565-3570 for their comments on the manuscript, and we thank the R. J. P. Tavarb, J. M., and Denton, R. M. (1988) Biochem. J. 2 5 2 , 607-615 Williams group for their generous hospitality during Dr. Leland Ellis’ Tavark, J . M., O’Brien, R. M., Siddle, K., and Denton, R. M. (1988) visits in Oxford. Biochem. J. 253, 783-788 Tavarb, J. M., Clack, B., and Ellis, L. (1991) J. Biol. Chem. 266, 1390-1395 REFERENCES Tornqvist, H. E., Pierce, M. W., Frackelton, A. R., Nemenoff, R. A., Cobb, M. H., Sang, B.-C., Gonzalez, R., Goldsmith, E., and Ellis, L. and Avruch, J. (1987) J. Biol. Chem. 262, 10212-10219 (1989) J. Biol. Chem. 264,18701-18706 Tornqvist, H. E., Gunsalus, J. R., Nemenoff, R. A., Frackelton, A. R., Dalgarno, D. C . , Levine, B. A., and Williams, R. J. P. (1983) Biosci. Pierce, M. W., and Avruch, J. (1988) J. Biol. Chem. 263, 350-359 Rep. 3,443-452 White, M. F., Shoelson, S. E., Keutmann, H., and Kahn, C. R. (1988) Ebina, Y., Ellis, L., Jarnagin, K., Edery, M., Graf, L., Clauser, E., Ou, J. Biol. Chem. 263,2969-2980 J., Masiarz, F., Kan, Y. W., Goldfine, I. D., Roth, R. A., and Rutter, Wuthrich, K. (1986) N M R of Proteins and Nucleic Acids,John Wiley and Sons, New York W. J. (1985) Cell 40, 747-758
In summary, the use of soluble derivatives of the cytoplasmic protein-tyrosine kinase domain of the human insulin receptor renders feasible the use of biophysical methods such as NMR to begin to explore the structural transitions that transpire during autophosphorylation of these enzymes. This experimental strategy, together with the judicious use of sitedirected point mutations and truncationsof the enzyme coding sequence, provides a new avenue by which to study the autophosphorylation of this member of the protein-tyrosine kinase family in further detail.
