Characterisation of an Endogenous Protein Kinase Activity in Ribonucleoprotein Structures Containing Heterogenous Nuclear RNA in HeLa Cell Nuclei

Eur. J. Biochem. 79, 117-131 (1977)

Characterisation of an Endogenous Protein Kinase Activity in Ribonucleoprotein Structures Containing Heterogenous Nuclear RNA in HeLa Cell Nuclei Jean-Marie BLANCHARD, Claude BRUNEL, and Philippe JEANTEUR Laboratoire de Biologie Moleculaire, Universite des Sciences et Techniques du Languedoc, Montpellier, and Laboratoire de Biochimie, Centre Paul Lamarque, HBpital St-Eloi, Montpellier (Received March 14/June 23, 1977)

Ribonucleoprotein particles containing heterogeneous nuclear RNA (hnRNA) were isolated from sonically disrupted HeLa cells nuclei by sedimentation through a triple sucrose cushion. We found that these particles were able to catalyse the incorporation of radioactive phosphate from [ Y - ~ ~ P I A into T P phosphoserine and phosphothreonine residues of endogenous proteins. Optimal conditions for enzymatic activity were defined after investigation of the influence of several parameters of the reaction. The requirement for a divalent cation was most efficiently met by 10 mM Mg', to a lower extent by Mn2+ and not at all by Ca2'. An apparent K, value for ATP of 0.02 mM was determined and the enzyme was found to be sensitive to p-hydroxymercuribenzoate. Neither cyclic AMP nor cyclic GMP stimulated the reaction over a wide range of concentrations and the same protein substrates were phosphorylated in their presence. After excluding possible contamination by other subcellular fractions, considerable evidence for the association of kinase and phosphate acceptor proteins with ribonucleoprotein structures containing hnRNA was gathered along several lines by showing that the enzymatic activity closely followed tritiated RNA pulse-labeled in the presence of low doses of actinomycin D in the following circumstances: (a) sedimentation in sucrose gradients, (b) isopycnic banding in metrizamide density gradients in the presence or absence of M g ' ions and (c) adsorption to oligo(dT)-cellulose columns under conditions where poly(A)-containing material is adsorbed. Analysis of phosphorylated substrates by electrophoresis on polyacrylamide gels containing sodium dodecylsulphate revealed that, although they spanned a wide range of molecular weights, the most intensely labeled species were localized in bands corresponding to molecular weights of 37000 and 28 000. The above pattern of polypeptides phosphorylated in vitro is compared with that of phosphoproteins labeled in vivo after exposure of cells to [32P]phosphate. The most widely accepted model for messenger RNA formation in eucaryotic cells postulates that it is derived by post-transcriptional modification of the rapidly labeled heterogeneous nuclear RNA (hnRNA, for review see [l]). However, only in few instances could a direct precursor-product relationship between this hnRNA and cytoplasmic messenger RNA be established with good evidence [2,3]. Abbreviations. hnRNA, heterogenous nuclear RNA; mRNA, messenger RNA; hnRNA . proteins, heterogenous nuclear ribonucleoprotein particles; CAMP and cGMP, adenosine and guanosine 3': 5'-monophosphates. Enzymes. Glyceraldehyde-3-phosphatedehydrogenase (EC 1.2.1.12); 3-phosphoglycerate kinase (EC 2.7.1.30); deoxyribonuclease I (EC 3.1.4.5); ribonuclease A (EC 3.1.4.22); ribonuclease TI (EC 3.1.4.8).

Although the exact physiological significance of hnRNA is far from clear, it seems reasonable to speculate that some regulatory processes might act at its level. In fact, most of this hnRNA turns over very rapidly in the nucleus, a half-life of 23 min has been reported in L cells [4], and only a small percentage ever reaches the cytoplasm [4,5]. The association of hnRNA with proteins into hnRNA . proteins is now firmly established on both cytological [6- 111and biochemical grounds [12- 181. We therefore surmised that any process involving hnRNA probably occurs at the level of these particles and is likely to be mediated by some of the associated proteins which have been found to be quite heterogeneous [16-191. The major drawback of this kind of approach is the absence of a functional criterion

118

for hnRNA . proteins. To circumvent this difficulty, we have undertaken a search for enzymatic activities which might be directly or indirectly involved in the regulation of mRNA biosynthesis in the nucleus and its transport to the cytoplasm. The presence of phosphorylated proteins in hnRNA . proteins from rat brain [20] as well as from HeLa cells [21] together with the observation that mRNA release from adenovirus-infected KB cell nuclei in vitro is an ATPdependent process [22] prompted us to look for a protein kinase. We have previously reported on the occurrence in nuclear particles from HeLa cells of an endogenous protein kinase activity [23]. We now present detailed evidence for the association of this kinase and its cognate phosphate acceptor proteins with particles containing hnRNA. Some properties of the enzymatic reaction are described and the acceptor proteins are characterized by sodium dodecylsulphate/acrylamide gel electrophoresis.

Protein Kinase Activity Associated with hnRNA . proteins

deen [24] to remove traces of RNAse. Acrylamide and bis-acrylamide were recrystallized according to Loening [25]. Cell Growth and Labeling Conditions Procedures for growth in suspension culture, collection and washing of HeLa cells (S3 strain) have been previously described [171. Washed cell pellets were always used immediately. Cultures were monthly checked for the absence of mycoplasma contaminations by scanning electron microscopy [26]. Standard labeling conditions were as follows : cells were centrifuged and resuspended to a concentration of 2 x lo6 cells/ml in medium containing dialysed calf serum, incubated for 90 min then exposed to [5-3H]uridine (27-30 Ci/mol; 10 pCi/ml) for 30 min. In most cases, actinomycin D was added to 0.04 pg/ml 20 min before radioactive uridine in order to prevent the synthesis of rRNA [27].

MATERIALS AND METHODS Materials

Preparation of hnRNA . proteins

0-Phospho-L-serine, ribonuclease TI and pronase were purchased from Calbiochem (Los Angeles, Calif.) ; 0-phospho-DL-threonine, p-hydroxymercuribenzoate and actinomycin D from Sigma Chemical Co. (St Louis, Miss.); amido black, sodium dodecylsulphate, chloramine T and formamide from Merck (Darmstadt, Germany) ;Triton X-100 from the British Drug Houses Ltd (London, England) ; acrylamide, and bisacrylamide, N,N,N’,N‘-tetramethylenediamine bovine serum albumin from Fluka (Switzerland) ; metrizamide from Nyegaard and Co. (Oslo, Norway); Coomassie brillant blue R-250 and phenylmethylsulfonyl fluoride from Serva (Heidelberg, Germany) ; oligo(dT)-cellulose type T-2 from Collaborative Research Inc. (Waltham, Mass.) ; pancreatic ribonuclease (code R) and deoxyribonuclease (code DPFF) from Worthington Biochemical Corp. (Freehold, N.J.) ; 2,5-diphenyloxazole (PPO) and 1,4-bis(2,4-dimethyl5-phenyloxazolyl) benzene (dimethyl-POPOP) from Packard Instruments (France). Boehringer (France) supplied the following products : ATP, CAMP,cGMP, poly(U) and Combithek markers as well as glycerate 3-phosphate and a mixture of glyceraldehyde-3-phosphate dehydrogenase and 3-phosphoglycerate kinase for preparation of [Y-~~PIATP. Carrier-free H332P04, carrier-free Nalz5I and [5-3H]uridine (27 - 30 Ci/mmol) were obtained from the CEA (Saclay, France) while [ Y - ~ ~ P ] G Tand P [GX-~~P]ATP were from the Radiochemical Center (Amersham, England). Pancreatic RNAse was boiled for 2 rnin to remove traces of DNAse while DNAse was treated with iodoacetic acid according to Zimmerman and San-

Cells labeled as described above were mixed with a ten-fold excess of unlabeled ones and hnRNA . proteins were extracted using a slight modification of the procedure of Kish and Pederson [28]. All operations were carried out at 0 - 4 “C. The washed cell pellet was resuspended in 3 vol. of buffer 1 (10 mM Tris-HC1 pH 7.4 at 25 “C,10 mM NaC1, 1.5 mM MgC12). Breakage of cells was achieved by a few gentle strokes of a Dounce glass homogenizer and checked by phase-contrast microscopy. Nuclei were pelleted by a 5-min spin at 1160 x g (IEC-PR2 centrifuge), resuspended and homogenized once more in buffer 1 containing 0.5 % Triton X-100 and then centrifuged and washed with plain buffer 1. After centrifugation the clean nuclei were finally resuspended in 1 vol. of buffer 1 and sonically disrupted by one 20-s pulse at 50 W with a microtip-equipped Branson B12 sonifier. Up to 10ml of the above sonicate were immediately layered on top of a 20-ml cushion of 30% (w/v) sucrose in buffer 1 and centrifuged for 20 min at 3000 x g . The material which did not enter the sucrose layer (post-nucleolar supernatant) was withdrawn with a pasteur pipet, layered and centrifuged on a discontinuous three-layer sucrose gradient (20, 2 and 8 ml of respectively 60,45 and 10 % sucrose w/v in buffer 1) for 90 min at 26000 rev./min in the Spinco SW-27 rotor. The wide band of hnRNA . proteins spanning the 45 % sucrose layer was withdrawn with a pasteur pipet or a peristaltic pump, care being taken not to collect too close to the 60% layer so as to minimize possible contamination with chromatin. hnRNA . proteins were in most cases analysed immediately or in some instances stored at -20 “C.

119

J.-M. Blanchard, C. Brunel, and P. Jeanteur

Characterization of Phosphorylated Amino Acids

!iiI ji

i

Fraction number

Fig. 1. isopycnic centrifugation o j ' hnRNA . proteins in cesium chloride. hnRNA . proteins were purified from cells labeled for 30 min in the absence of actinomycin D. Conditions for glutaraldehyde fixation and centrifugation have been described previously [17]. Acid-precipitable 3H radioactivity (O-----O) was determined in each fraction while the density of every fifth fraction was measured by pycnometry 1-(

Integrity of nuclei and absence of cytoplasmic contamination was kindly checked for us on a routine basis by Dr E. Puvion using the procedure of Bernhard [29]. The characteristic low density of hnRNA . proteins was also routinely checked by isopycnic centrifugation in cesium chloride after glutaraldehyde fixation as described previously [17]. A typical profile of hnRNA . proteins pulse-labeled for 30 min with [5-3H]uridine in the absence of actinomycin D is presented in Fig. 1. Protein Kinase Assay Standard reaction mixtures (0.2 ml) were in kinase buffer (10 mM Tris-HC1 pH 8.3 at 25 " C , 10 mM MgC12, 10 mM 2-mercaptoethanol and contained varying amounts of hnRNA . proteins). Reaction was initiated by adding [p3'P]ATP to 0.1 mM (specific activity adjusted to 100- 5000 counts min-l pmol-' with unlabeled ATP). After a 5-min incubation at 30 "C, the reaction was stopped by adding 200 pg of unlabeled ATP and 2 ml of cold 10% trichloroacetic acid. When aliquots of the standard assay were separately precipitated, 50 pg of bovine serum albumin was added as a carrier. Precipitates were collected on Whatman GF/C glass filters, washed successively with cold 5 % trichloroacetic acid and ethanol then counted as described below.

Details of the experiments designed to establish that 32P is covalently incorporated into proteins (Table 1) through an ester phosphate linkage (Fig. 3) will be given in the corresponding legends. Electrophoretic detection of phosphoserine and phosphothreonine after partial acid hydrolysis was carried out as follows. 400 pg of hnRNA . proteins (expressed as protein) were phosphorylated in a scaled-up (0.5 ml) standard reaction mixture, separated from unreacted ATP by filtration on a 0.8 x 30-cm Sephadex G-25 column and recovered in the excluded volume by precipitation with 10 % cold trichloracetic acid. After centrifugation, the pellet was washed with 10 % trichloracetic acid and finally resuspended in 0.2ml 3 M HCl. The tube was then sealed and incubated overnight at 110 "C. After drying under reduced pressure, the residue was dissolved in 0.1 ml water containing 5 pg of both phosphoserine and phosphothreonine markers. 20 1.11 were then spotted on a Whatman 3MM paper sheet and electrophoresed in 1 M formic acid/pyridine (pH 2.1) at 2500 V as described by Martelo et al. [30]. The paper was then dried, sprayed with ninhydrin (0.2 % in acetone) to reveal the markers and autoradiographed on Kodak X-ray film. Sucrose Density Gradient Analysis Fresh hnRNA . proteins from cells labeled for 30 min with [5-3H]uridinein the presence of 0.04 pg/ml actinomycin D were purified through three sucrose cushions as described above. Either the whole hnRNA . protein population (Fig. 6) or a single fraction (Fig. 5) were diluted with buffer 1 so as to reduce the sucrose concentration to about 10 % (w/v) as checked by refractive index; they were divided into aliquots which were then subjected to the various treatments indicated in the legends to Fig. 5 and 6 . Next 0.4-ml samples were layered on top of a 12-ml 15-30% (w/v) sucrose gradient in buffer 1 and centrifuged for the time indicated in the legends at 40000 rev./min in the SW-41 rotor of the Spinco L5-75 ultracentrifuge. 0.4-ml fractions were collected from the bottom of the tubes by means of a peristaltic pump and analyzed for absorbance at 260 nm (not shown in Fig. 5 and 6), acid-precipitable 3H (50-pl aliquots) and kinase activity (20-1-11aliquots). Metrizamide Density Gradient Analysis 0.4-ml samples containing about 500 pg of 3Hlabeled hnRNA . proteins (30-min label in the presence of actinomycin D) or unlabeled ribosomes in either buffer 1 or buffer 1 containing 10 mM EDTA were layered on top of a 4-ml column of 41 % (w/v) metrizamide solution in the corresponding buffer.

120

Centrifugation was at 4 "C for 66 h at 30000 rev./min in the SW-60 Spinco rotor. 3-drop fractions were collected after piercing the bottom of the tube. Refractive index was measured on 20 p1 of every third fraction and the density was calculated from the following formula [31]:

20-p1 aliquots of each fraction were used for kinase assay in all cases and for determination of acidprecipitable 3H radioactivity of hnRNA . proteins. In the case of unlabeled ribosomes, protein content was determined in the remaining volume of each fraction by the method of Kuno and Kihara [32].

Oligo ( d T )-cellulose Column Chromatography 3H-labeled hnRNA . proteins (30-min label in the presence of actinomycin D) were prepared as described above except that the three sucrose cushions were in binding buffer (buffer 1 containing 0.25 M NaCl). 5 ml oligo(dT)-cellulose columns (0.9 x 8 cm) equilibrated with the same buffer were loaded with 2-ml samples (containing about 1 mg of hnRNA . proteins expressed as protein), washed with 15 ml buffer and eluted with 50 % formamide in regular buffer 1. Flow rate was 10 ml/h and 1-ml fractions were collected. 50-pl and 2 0 4 aliquots of each fraction were used respectively for measuring precipitable 3H and kinase activity in standard assay conditions. Preribosomes and ribosomes were chromatographed in similar conditions.

Sodium Dodecylsulphate/ Polyacrylamide Gel Electrophoresis 60 pg of hnRNA . proteins (expressed as protein), eventually phosphorylated in vitro or mixed with 2.5 pg lZ5I-labeled bovine serum albumin, were precipitated with cold 10 % trichloracetic acid for 10 min in ice then centrifuged at 6000 rev./min for 15 min. The pellet was washed with an ethyl ether/ethanol mixture (3 : l), dried under vacuum, dissolved in 50 p1 of 0.05 M Tris-HC1(pH 6.8), 2 mM EDTA, 1 % 2-mercaptoethanol, 1 % sodium dodecylsulphorate, 10 % glycerol, 0.02 % bromophenol blue then heated at 100 "C for 5 min. Up to 12 50-p1 samples were loaded into the 5-mm-wide wells of the slab gel apparatus described by Studier [33] using 1.5-mm-thick gels. The discontinuous gel system was that described by Matringe and Jacob 1191 using a 1-cm stacking gel and a 10-cm separating gel containing 10 % acrylamide. Electrophoresis was at 20mA/gel until the dye reached 1 cm above the lower end of the gel. Staining was carried out for 1 h in a 0.25 % solution of Coomassie

Protein Kinase Activity Associated with hnRNA proteins

brilliant blue R-250 in methanol/acetic acid/water (5/1/5), then destained by diffusion in a solution of 7.5% acetic acid and 5% methanol in water. When 32P-labeledproteins were analysed, gels were exposed for 15 min to 10% trichloracetic acid at 90 "C prior to staining. Destained gels were photographed, then transferred to a sheet of Whatman 3MM paper, dried under vacuum as described by Maize1 [34] and autoradiographed on Kodak X-ray film. Molecular weights were estimated according to Weber and Orborne [35] using bovine serum albumin, ovalbumin and RNA polymerase subunits as markers.

Miscellanous Techniques [y3'P]ATP was synthesized and purified according to Glynn and Chappell [36] using carrier-free [32P]orthophosphoric acid. Its radiochemical purity was routinely checked by thin-layer chromatography on cellulose sheets using butan-1-ol/acetone/acetic acid/ 5 % NH40H/H20 (4.5/1.5/1/1/2) as solvent and always found to be higher than 95 %, the remaining small percentage being unreacted free phosphate. Unlabeled ribosomes were prepared as follows. Cells were homogenized in 7 vol. of buffer 1 without Triton X-100 and nuclei were removed as described above. Mitochondria were deposited by centrifugation for 20 min at 30000 x g and the upper 80 % of the supernatant was recentrifuged through a 10-ml 20 % (w/v) sucrose cushion in buffer 1 for 3 h at 50000 rev./ rnin in the Spinco 70 Ti rotor to yield a pellet of ribosomes. 25 I-labeled bovine serum albumin was prepared by the chloramine T method according to Berthoux [37] using a chloramine-to-protein ratio of 1/50. Counting was carried out in a Packard Tricarb Autogamma counter. Protein concentrations were determined according to Lin and Lehman [38] or by the amido black method of Kuno and Kihara [32] both using bovine serum albumin as a standard. Both 3H and 32Pradioactivities were determined on precipitates collected on Whatman GF/C glass filters and counted in an Intertechnique SL-30 scintillation spectrometer using 5 ml of a scintillation fluid containing 5 g of PPO and 0.3 g of dimethyl-POPOP per 1 of toluene. RESULTS

Evidence for Endogenous Protein Kinase Activity in the Particulate Fraction of HeLa Cell Nucleoplasm Fig. 2 shows the kinetics of incorporation of radioactivity from [Y-~'P]ATPinto acid-insoluble material as a function of time or nuclear particle concentration expressed as protein. Incorporation appears to be linear for about 10 min (Fig. 2A) and reaches a

121

J.-M. Blanchard, C. Brunel, and P. Jeanteur

0

20

40

60

Time [min)

0

10

I I 20 30 Protein (bg)

I

40

Fig. 2. Evidence for 3zP incorporation upon incubation ofhnRNA 'proteins with [ Y - ~ ' P / A T P(A) . Kinetics of incorporation as a function of time. 30 pg of hnRNA . proteins were incubated in a standard reaction mixture at 30 "C. At various time intervals, 20-pl aliquots were taken, precipitated and counted. (B) Kinetics of incorporation as a function of hnRNA . protein concentration. Varying amounts of hnRNA . proteins were incubated and processed under standard assay conditions

plateau between 20 and 30 min. This justifies the choice of 5 min as standard incubation time. Prolonged incubation (beyond 1 h) results in a decline of acidprecipitable radioactivity. This can be ascribed either to proteolysis or to the presence of a phosphatase activity, either non-specific (alkaline phosphatase) or specific (phospho-protein phosphatase). The involvement of non-specific serine proteases is excluded by the experiments reported in Fig. 9 below. Some nonspecific alkaline phosphatase activity has indeed been detected and related to some membrane contamination as discussed below. However, current observations in our laboratory show that this activity is unable to dephosphorylate the labeled phosphoproteins and that a phosphoprotein phosphatase can be identified and characterized in the hnRNA . proteins (unpublished results). Incorporation as a function of hnRNA . proteins concentration under standard 5-min incubation conditions (Fig. 2 B) is linear up to at least 100 pg of protein per 0.2-ml assay. It should be noted that these incorporations are not dependent upon the addition of any exogenous protein substrate. In as much as the incorporation has occurred into protein, as will be demonstrated below, the kinase activity studied here is therefore endogenous to the preparation which must contain phosphate acceptor proteins. To determine whether the incorporated radioactivity was associated with proteins and not with RNA or phospholipids, standard incubation mixtures of [ Y - ~ ~ P I Awith T P hnRNA . proteins were precipitated with cold 10% trichloroacetic acid and subjected to treatments known to remove RNA and phospholipids. As can be seen in Table 1 for average values of several experiments, 90% and 100% of the radioactivity precipitated by cold trichloroacetic acid was

Table 1. Evidence for 32P incorporation into the protein moiety of hnRNA .proteins 30 pg of hnRNA . proteins were incubated with [Y-~'P]ATPin a standard kinase assay and three 30-p1 aliquots were separately precipitated and filtered. Each filter was then subjected to a different treatment before being dried and counted in a toluene-based scintillation fluid: (1) control with no further treatment; (2) exposure for 20 rnin to 10 % trichloracetic acid at 90 "C followed by rinsing with cold 10 % trichloracetic acid; (3) exposure for 1 h to a chloroform/methanol mixture (2/1) at 25 "C followed by washing with the same mixture. Two other 30-p1 aliquots were further incubated for 1 h at 37 "C in 0.1 ml kinase buffer containing 0.4 M NaCl and 1 mM ATP in the presence of either: (4) 10 pg pronase or ( 5 ) 10 pg pancreatic ribonuclease, then precipitated with cold 10% trichloracetic acid and counted. These modified ionic conditions were designed to inhibit the hnRNA . protein-associated phosphatase Treatment

Resistant radioactivity

% control 1. Cold trichloracetic acid (control) 2. Hot trichloracetic acid 3. Chloroform/ethanol 4. Pronase 5. RNAse

100 90 98 2 100

resistant respectively to hot trichloracetic acid and a chloroform/ethanol mixture, 98 % of the labeled material was solubilized by pronase while 100% resisted pancreatic RNAse. These data demonstrate that all the 32Pwas indeed incorporated into proteins and allowed us to use only cold trichloracetic acid precipitation for our standard assay. To determine the nature of the chemical linkage of phosphate to amino acid residues, standard incubation mixtures were subjected to treatment with hydroxylamine which cleaves acyl-phosphates [39] and hot alkali which hydrolyses phosphoserine and phosphothreonine but

122

Protein Kinase Activity Associated with hnRNA . proteins

neither phospholysine, phosphohistidine or phosphoarginine [40]. Results presented in Fig. 3 show that hydroxylamine had no effect while complete removal of phosphate groups was obtained with alkali. These properties characteristic of an ester linkage are confirmed by the demonstration of labeled phosphoserine and phosphothreonine in mild acid hydrolysates as shown in Fig. 4. Association of Protein Kinase Activity and Phosphate Acceptor Proteins with hnRNA .proteins

All the experiments to be described below have been designed to provide critical evidence that the phosphoprotein kinase and acceptor proteins investigated here are indeed associated with structures containing hnRNA and to exclude the possible contribution to this enzymatic activity of contaminants from other cellular compartments, the more so as protein kinases are known to be quite ubiquitous in eucaryotic cells

0

40

20

60

Time (min)

Fig. 3. Chemical nature of the phosphate-protein bond. 30 pg of hnRNA . proteins were incubated with [y-32P]ATPin a standard kinase assay. At the end of the incubation period, three 30-pl aliquots were taken into 0.1 ml of either kinase buffer containing 0.4 M NaCl and 1 mM ATP (0- -O), 1 M NaOH (-0) or 1 M Incubation hydroxylamine/l M succinic acid pH 5.5 (A-A). was continued at 30 "C and 10-p1aliquots were removed at indicated times, precipitated in the presence of carrier and counted

[41]. Four obvious types of contaminants should a priori be considered : chromatin, soluble proteins, cytoplasmic ribosomes or preribosomes of nucleolar origin. The presence of such kinases has been well documented in nuclear proteins, whether or not associated with chromatin [42- 501 and in ribosomes [51- 541. Although no such activity has to our knowledge been described in nucleolar preribosomes, the proteins which they contain have been shown to be phosphorylated [55,56]. Contamination of hnRNA . proteins by membrane debris has also been considered. Although such a contamination may exist, we will show below that it does in no way interfere with our data. Sucrose Density Gradient Analysis

Nuclear RNA ' protein particles were purified from cells labeled for 30 min with [5-3H]uridine in the presence of low levels of actinomycin D as described above. They were then subjected to various treatments, sedimented through sucrose gradients in hypotonic buffer and analyzed with respect to acidprecipitable 3H radioactivity and protein kinase activity. The control experiment presented in Fig. 5 A shows that rapidly labeled RNA and protein kinase activity comigrate together with a distribution centered around 80 S. hnRNA . proteins from a single fraction of the 45 % sucrose layer were used in this particular experiment which explains the higher homogeneity as compared to the profiles reported by Pederson [IS] and ourselves where the whole hnRNA . protein population was used (Fig. 6). The four types of contaminants mentioned above had now to be ruled out. The presence of preribosomes whose density in cesium chloride is only slightly lower than that of ribosomes [57] is excluded by the presence of a unique peak of labeled RNA shown in Fig. 1 at a density of 1.39 g/cm3. Contamination with soluble proteins which might have adventitiously bound to hnRNA . proteins can be readily ruled out by showing that upon recentrifugation of the 80-S material on a second identical

Fig. 4. Paper electrophoresis of u partial acid hydrolysate oj hnRNA . proteins 32P-lubeledin viiro autoradiography of the electrophoregrum. Conditions for incubation, acid hydrolysis and electrophoresis have been described in Materials and Methods. Migration was from left to right; a, b and c refer to the respective positions of phosphothreonine, phosphoserine and inorganic phosphate

123 J.-M. Blanchard, C. Brunel, and P. Jeanteur

0

bottom

Fraction number

I

top

Fig. 5. Effect of' vurious treutments on the sedimentuiion behavior in sucrose gradient of hnRNA protein-ussociuted protein kinusr. Three identical aliquots of3H-labeled hnRNA . proteins (30-min label with 0.04 pgjml actinomycin) were incubated at 0 "C for 20 min in the presence of the following additions: (A) none; (B) 0.01 M EDTA; (C) 20 pg/ml pancreatic DNAse. Centrifugation was for 2 h in conditions described in Materials and Methods. (+O) 3H radioactivity; (O--O) 32Pradioactivity. The position of the 80-S ribosomes marker was determined on a parallel gradient

sucrose gradient, all the protein kinase activity migrates as a sharp peak with the same sedimentation coefficient (results not shown). Although electron microscopic examination of our purified nuclei at low magnification (not shown) revealed no significant contamination with polysomes, we looked for the presence of 80-S ribosomes by running a gradient in the presence of 1 0 m M EDTA, a treatment previously shown to dissociate them into subunits [27,58]. Fig. 5 B shows that both profiles (3H and 32P) remained unaffected by this treatment. In view of the fact that protein kinases have been found in association with chromatin [42,46,47,49,50], it was important for our experiments to rule out the presence of chromatin. Experiments (not shown here) with particles extracted from cells labeled with tritiated thymidine and sedimented on a sucrose gradient had shown that only traces of radioactive DNA were present under the peak of particles. Assuming a 1/2 mass ratio between DNA and non-histone proteins, it was calculated that these could only contribute less than 2 % of the total protein content of the particles. To further exclude any possible contribution of DNA to the sedimentation pattern observed in Fig. 5A, we treated a sample of particles with 20 pg/ml pancreatic DNAse at 4 "C. As a preliminary to this experiment, it was verified that under these conditions, DNAse is active against exogenous DNA. This control was made necessary by the finding that eucaryotic nuclei contain actin [59] which is known to be a potent inhibitor of pancreatic DNAse [60]. Sedimentation analysis of the DNAse-treated sample yielded a pattern (Fig. 5C) similar in all respects to that of the

control. Having excluded that the protein kinase studied here might be due to contamination by other cellular components, it remains to be established conclusively that the enzyme is actually associated with particles containing hnRNA. The following experiment as well as all those to be described below have been designed to answer this crucial question. When h n R N A . proteins pulse-labeled with [3H]uridine were submitted to a mild RNAse treatment (0.01 pg/ml and 0.5 unit/ml of pancreatic and TI RNAses respectively), the radioactivity was shifted from high to low sedimentation coefficients without loss of acid-precipitable radioactivity. The kinase was also shifted but its distribution remained centered around 60-80 S (Fig. 6A and B). If the RNAse concentrations were increased to 0.1 pg/ml and 5 units/ ml for pancreatic and TI RNAses, the acid-precipitable 3H radioactivity was not only shifted to low sedimentation values but was significantly reduced while kinase activity was observed both in the 60-80-S and soluble regions of the gradients (Fig. 6C). Attempts to completely solubilize 'H radioactivity and ultraviolet-absorbing material with high levels of RNAse (20- 100 pg/ml) have resulted in aggregation artefacts of the kinase which was eventually pelleted under the same centrifugation conditions as above. The direct involvement of RNAse in these aggregates could be clearly established by demonstrating its presence in the pellet by enzymatic assay according to Zimmerman and Sandeen [24]. These experiments show that the hnRNA clipping does affect the sedimentation behavior of the kinase and its substrates although they belong to structures

124

Protein Kinase Activity Associated with hnRNA . proteins 15

18os

C

18os

I

Fraction number

bottom to P Fig. 6 . Effect of mild RNAse and high ionic strength on thc sedimentation behavior in sucrose gradient of'hnRh'A '1,rorein-ussociated protein kinase. Four identical aliquots of Wlabeled hnRNA . proteins (30-min label with 0.04 pg/ml actinomycin) were incubated at 0 ' C for 20min under the following conditions: (A) in buffer 1 (control); (B) in buffer 1 containing 0.01 pg/ml and 0.5 unit/ml of pancreatic and TI RNAse respectively; (C) in buffer 1 containing 0.1 pg/ml and 5 units/ml of pancreatic and TI RNAse respectively; (D) in buffer 1 containing 0.5 M NaC1. Centrifugation was for 1.5 h in conditions described in Materials and Methods

which seem to survive the removal of most hnRNA. These 'core' structures might be due to the presence of a stable RNA component [61] which is now under study in our laboratory. The stability of the association of the kinase with hnRNA . proteins with respect to ionic strength has been tested by exposing the particles to increasing salt concentrations, spinning them down and assaying the remaining endogenous activity in the pellet. The results in Table 2 show that a progressive release of activity occurs above 0.15 M NaCl and is nearly complete at 0.5 M. Confirmation of this observation is obtained by running a sucrose gradient at 0.5 M NaCl which shows that the activity no longer migrates in association with radioactive RNA but remains close to the top of the gradient (Fig. 6D). However, as we pointed out earlier [23], the absence of detectable activity in washed particles could also reflect only the complete removal of the acceptor proteins. The fact that addition of histone HI from calf thymus did not significantly increase the amount of 32Pincorporated rendered this possibility quite unlikely unless the enzyme had such a restricted specificity as to be unable to phosphorylate anything but its own cognate substrates.

Metrizamide Density Gradient Analysis

So far, one of the most widely recognized criterion for non-ribosomal ribonucleoprotein particles has been their low buoyant density in cesium chloride which was attributed to their high protein-to-RNA ratio [62]. However, the requirement for prior fixation with aldehydes to prevent their dissociation makes this analytical method unsuitable for assaying enzymatic activities. Metrizamide has been proposed as an alternative medium for use in isopycnic centrifugation because, being non-ionic, it does not require prior fixation of the material to be analysed. When analysed in metrizamide in the presence of 1.5 mM MgC12, our hnRNA . protein preparations were found to band as a unique peak around 1.281.30 g/cm3 (Fig. 7A) in accordance with the observation reported by Houssais [63] for the corresponding particles from L cells labeled for a brief pulse, while chromatin was found at a density of about 1.2 g/cm3 by us (not shown) and others [31]. The profile of kinase activity closely followed that of labeled RNA (Fig. 7A). These observations further exclude the possibility that the kinase studied here might have originated from contamination by chromatin.

125

J.-M. Blanchard, C. Brunel, and P. Jeanteur Table 2. Progressive release of endogenous kinase activity from hnRNA . proteins with increasing salt concentration Identical ahquots containing 800 pg each of 3H-labeled hnRNA . proteins (30-min label with 0.04 pg/ml actinomycin D) were diluted to 2.5 ml of buffer 1 whose NaCl concentration had been adjusted to the indicated values. After a 2-h incubation at 0 "c,each sample was centrifuged through a 1.5-ml cushion of 12% (w/v) sucrose in the same buffer for 5 h at 50000 rev./min in the SW60 Spinco rotor. Pellets were resuspended in 0.2 ml kinase buffer and 20-4 aliquots were assayed for acid-precipitable 3H radioactivity and 32P incorporation in standard kinase assays. All values of kinase activity were corrected for the slightly variable recovery of hnRNA . proteins in the pellet by normalizing then with reference to the 3H pelleted in the control experiment (0.01 M NaC1) Final NaCl concn

Kinase activity remaining associated with hnRNA. proteins

M

%

0.01 0.15 0.25 0.30 0.50 0.80

100 96 80 43 7 0

Under similar ionic conditions, that is in the presence of 1.5 mg MgClz, ribosomes banded at about the Samedensity as hnRNA . proteins ( ~ i7 c~) . and it therefore appeared to On these grounds a possible contribution of the ribosomeassociated kinase. However, Buckingham and Gros [64] observed that the density in metrizamide of ribosomal subunits from muscle cells was strongly dependent upon the presence of M$+ ions (being only around 1.2 g/cm3 in their absence) while that of mRNA . protein was not affected. Should the same situation apply to hnRNA . proteins and ribosomes from HeLa cells, one would expect no change in the density of either hnRNA . proteins or kinase when analyzed in the presence of EDTA, as long as there are no contaminating ribosomes. This prediction is clearly confirmed in Fig. 7B. The integrity of hnRNA . proteins after metrizamide banding under both conditions was checked by showing that they exhibited the usual low density of 1.4 g/cm3 when fixed and rebanded in cesium chloride. On the other hand, it was verified _ _ _ _ _ _ _ _ _ _ ~

~

p =I . 30 g i c m 3

I

p =1.3 1 gicrn3

C

10

0 0

10

20 Fraction number

10

20

Fig. 7. Distinctive behaviour of hnRNA . proteins a n d r.ih(,soine-a.ssu~i(i/~~(/ / i w r c i i i Itinuse on metrizamide density grudiients. Centrifugation conditions and fraction analysis of 3H radioactivity (0 o), 32Pincorporation in standard kinase assay (.---a) and protein content (@-a) were as described in Materials and Methods. (A) SO0 pg 'H-labeled hnRNA proteins in buffer 1. (B) 500 pg 3H-labeled hnRNA . proteins in buffer 1 containing 10 mM EDTA. (C) 500 pg unlabeled ribosomes in buffer 1. (D) 500 pg unlabeled ribosomes in buffer 1 containing 10 m M EDTA

126

Protein Kinase Activity Associated with hnRNA . proteins

Table 3 . Oligo(dT)-cellulose chromatography Experimental conditions were described in Materials and Methods. All values for experiments involving hnRNA . proteins are averages of at least three different experiments. n.d. = not determined Experimental Conditions

[3H]RNA co1u mn s

Material

unadsorbed

Kinase activity unadsorbed adsorbed

% input hnRNA . proteins hnRNA . proteins hnRNA . proteins hnRNA . proteins Ribosomes Preribosomes

+ poly(U)

unsubstituted cellulose oligo(dT)-cellulose equilibrated with poly(A) oligo(dT)-cellulose oligo(dT)-cellulose oligo(dT)-cellulose oligo(dT)-cellulose

that unlabeled ribosomes were actually shifted to lower densities by such a treatment as was their associated kinase (Fig. 7 D). This result therefore rules out any significant contribution of the ribosomeassociated protein kinase to that described here. The presence on hnRNA . proteins of an alkaline phosphatase activity against p-nitrophenylphosphate has raised the possibility of contamination by membranes. Indeed, this activity hands in metrizamide at a density of about 1.15 g/cm3 as expected for membranes [31] and is therefore effectively and completely separated from hnRNA . protein (results not shown). However, hnRNA . proteins banding at a density of about 1.30 g/cm3 in metrizamide are still able to carry out the phosphorylation of the same polypeptides species as will be shown in Fig. 10D below. This expensive and time-consuming additional purification step appeared therefore useless for our present purpose and was not used on a routine basis.

Oligo ( d T )-cellulose Chromatography hnRNA ' proteins have been shown to bind to oligo(dT)-cellulose [65], presumably through the poly(A) sequence of their hnRNA. Should the kinase be associated in some way with hnRNA . proteins, its behaviour on oligo(dT)-cellulose columns would be expected to closely parallel that of labeled hnRNA. We therefore chromatographed particles labeled in vivo with [5-3H]uridine on such columns and followed both labeled RNA and kinase activity. Results presented in Table 3 show that adsorption of kinase to the columns strictly follows that of hnRNA and that both depend on the possibility of interaction between oligo(dT) and the poly(A) tract of hnRNA. While 85 % of hnRNA and 90 % of kinase are adsorbed on oligo(dT)-cellulose columns under standard conditions, about 100% of both are unadsorbed under either of the following conditions: use of plain cellu-

100 100 35 88 96 91

0

0 85 12 4 3

85 - 95 85-95 10-20 85 - 95 n.d. n.d.

lose, presaturation of oligo(dT)-cellulose with poly(A) or of hnRNA .proteins with poly(U). To exclude interference by protein kinases which are or might be present in ribosomes [51-541 or preribosomes respectively, it was verified that neither of these were adsorbed to the oligo(dT)-cellulose column whether or not they were mixed with excess unlabeled hnRNA . proteins. Critical demonstration that kinase was actually bound and not inactivated would require one to be able to elute it from the column. Although significant levels of activity have been recovered in the formamide buffer used for elution of RNA, this and other attempts (omission of salt and increase of the temperature to 40 "C) to elute the kinase failed to yield quantitative and reproducible recovery. By comparison to the binding of only 25 - 40 % of deproteinized hnRNA to poly(U)-Sepharose [ 17, 661, the 85 "/, level of retention of hnRNA . proteins appeared strikingly high although similar values had already been reported [65]. However, we cannot exclude that binding could occur on account of other factors besides the interaction of poly(A) with oligo(dT) although the abolishment of binding when the column had been presaturated with poly(A) argues against this possibility.

Enzymatic Properties The following experiments have been carried out in order to determine optimal conditions for protein kinase activity. In all cases, purified hnRNA . proteins were used. The following parameters have been investigated. Effect of Cations. Fig. 8 A shows that the kinase reaction has an absolute requirement for either Mg2 or Mn2+ as a divalent cation with a strong preference for Mg2+.A plateau of maximum activity with Mg2+ was obtained at about 10 mM while concentrations +

127

J.-M. Blanchard, C.

A

0

O-

-

A

n 0

h

0-

--

Divalent cation (mM)

I

I

I

I

Monovalent Cation (mM)

Fig. 8. Effect of divalent andmonovalent cations on kinase activity. 30 Fg of unlabeled hnRNA . proteins were incubated and processed under MgClz; (A--A) MnClz; standard assay conditions, except for the concentration of the following cations. (A) Divalent cations: (-0) (M CaCI2; ) (-0) CaC12 10 mM MgC12. (B) Monovalent cations: (-0) NaCl; (w) KC1

+

suggesting a requirement for free - SH groups whether higher than 20 mM were inhibitory (not shown). The at the level of enzyme or acceptor proteins. However, maximum activity obtained with Mn2+ was at least the striking protective effect exerted by unlabeled 4-5 times lower than with Mg2+.Ca2+ was totally ATP when present during preincubation with p-hyineffective in replacing Mg2 and was even inhibitory droxymercuribenzoate favors the idea that the required in its presence (10 mM). As to monovalent cations, - SH groups are carried by the enzyme itself. The use Na+ and K + showed no significant effect between of phenylmethylsulfonyl fluoride as a protease in0 and 150 mM (Fig. SB); above 200 mM the activity hibitor [67]in the experiments reported below required began to decline but this effect must be interpreted preliminary verification that it was not inhibitory. in terms of the release of soluble enzyme and acceptor The last result of expt I shows clearly that such is not proteins from the particles (Table 2 and Fig. 6D) the case. which we had previously shown to result in an apparent Influence of Cyclic Nucleotides. As shown in loss of activity [23]. Following the above data, the Table 4 (expt 11) concentrations of either cAMP or standard assay contained 10 mM Mg2+ and no monocGMP up to 0.01 mM had nearly no detectable effect valent cation except for those included in the enzyme even when assayed in the presence of 25 mM theosample. phyllin (to inhibit cAMP phosphodiesterase) while Influence of A T P Concentration. Maximum velo0.01 mM of either nucleotides was slightly inhibitory. city was attained at 0.1 mM ATP and an apparent However, this latter concentration is too far from K, value of 0.02 mM has been calculated from the cellular levels for this effect to have any physiological corresponding double-reciprocal plot (data not significance. shown). A standard concentration of 0.1 mM has therefore been adopted for all assays. It should be noted that the accuracy of this result remains contingent on the concentration of acceptor proteins not Characterization of Phosphate Acceptor Proteins being rate-limiting, a condition which are presently Before any attempt could be made to identify unable to confirm. the polypeptides phosphorylated in vitro during inOptimal p H and Temperature. Maximum enzyme activity was observed for a temperature of 30 "C and cubation of hnRNA . proteins with [Y-~'P]ATP,the a pH between 8 and 8.5. Therefore, standard assays possible interference of proteases had to be eliminated, were carried out at 30 "C and pH 8.3. the more so as the decline of acid-precipitable 32P Specificity for [y-32P]ATP.A stringent specificity radioactivity observed after prolonged incubation was observed for [Y-~'P]ATPas compared to [Y-~'P]- periods could have been blamed on their action. In GTP for which incorporation is 15 times lower while order to test this possibility control experiments were no incorporation of 32Pis observed with [U-~~P]ATP. performed in which 1251-labeledbovine serum albumin Effect of Various Reagents. Data of Table 4 was mixed and incubated for various times with (expt I) show that 2-mercaptoethanol significantly hnRNA . proteins in standard kinase assay conditions activates the reaction while preincubation with p-hy(except that only unlabeled ATP was present) before droxymercuribenzoate results in a drastic inhibition, electrophoresis on sodium dodecylsulphate acryl+

128

Protein Kinase Activity Associated with hnRNA . proteins

Table 4. Effect of various reagents and cyclic nucleotides 011 I, , i J c / w activity In expt I, five 30-pg samples of unlabeled hnRNA . protein$ were separately preincubated for 1 h at 0 "C in 0.2 ml kinase buffer without 2-mercaptoethanol and with the additions indicated. Incubations were started by raising the temperature to 30 "C and adding [y-32P]ATP to the standard assay concentration. HOHgBzOH = p-hydroxymercuribenzoate; PhMeSOsF = phenylmethylsulfonyl fluoride. In expt 11, 30 pg of hnRNA . proteins were incubated and processed under standard assay conditions except for the presence of varying concentrations of cAMP or cGMP Expt Additions

3 z incorporation ~

pmol

I

To preincubation mixture: none 2-mercaptoethanol(lO mM) HOHgBzOH (1 mM) HOHgBzOH (1 mM) unlabeled ATP (0.1 mM) PhMeS03F 2-mercaptoethanol

+

I1

+

To incubation mixture: none cAMP 0.1 pM 1 PM 10 pM 100 pM 0.1 pM cGMP 1 PM 10 pM 100 pM

86 120 11 70 116 115 124 111 117 97 111 118 121 86

Fig. 9. Assay of protease activity in hnRNA proteins by sodium dodecylsulphatelacrylamide gel electrophoresis. Autoradiography of slab gel electrophoregrams of 60 pg unlabeled hnRNA . proteins and 2.5 pg '251-labeledbovine serum albumin mixed and incubated in a standard 0.2-ml kinase mixture as follows: (a) albumin alone, not incubated; (b) albumin + hnRNA. proteins incubated at hnRNA . proteins incubated at 30 "C for 15 min; (c) albumin 30 "C for 15 h; (d) albumin + hnRNA . proteins incubated at 30 "C for 15 h in the presence of 1 mM phenylmethylsulfonyl hnRNA . proteins not incubated. At the fluoride; (e) albumin end of the incubation period, mixtures were precipitated, washed and electrophoresed as described in Materials and Methods. The arrow indicates the direction of migration 1

+

+

amide gels. The pictures presented in Fig.9 show that no degradation at all of labeled albumin had occurred even after 15 h of incubation. Similarly, no effect could be detected upon addition of 1 mM phenylmethylsulfonyl fluoride which is known to inhibit serine proteases [67] and which we have shown not to be inhibitory to the present kinase activity (see Table 4). These results unambiguously ruled out any interfering non-specific protease activity. The autoradiogram presented in Fig. 10b shows that, although some radioactivity is consistantly observed in higher-molecular-weight proteins, most of the 32P is found associated with two regions of molecular weight respectively 37 000 1000 and 28000 1000 (average of seven determinations). The same pattern of phosphorylated proteins was observed when incubation was performed in the presence of either 1 mM phenylmethylsulfonyl fluoride, 1 mM cAMP or cGMP (not shown). Comparison with the stained gel (Fig. 10a) shows that these phosphorylated species comigrate with some of the most prominent proteins found in the 23 000 -42000M , range called zone 3 according to the nomenclature of Matringe and Jacob [19]. Of course, this does not prove that phosphorylated species are identical to the stained ones. On some occasions, the labeled

proteins of the 31 000-M, regions appeared as a double band of two very close species. This occasional observation was not further investigated. As we had observed that histone H1 from calf thymus migrated only slightly faster on our gels than the 28000-Mr band (not shown), we felt it necessary to exclude the remote possibility that this band might be a phosphorylated histone. Should this be the case, it should be extractable from hnRNA proteins by treatment with 0.2 M sulfuric acid [68]. The autoradiogram of 32P-labeledhnRNA . proteins subjected to this treatment (Fig. lOc) before electrophoresis shows clearly that it had no effect on the presence and amount of the 28000-M, band which therefore cannot be an histone. Phosphorylation of hnRNA . proteins previously banded in metrizamide under conditions where membrane contaminants are effectively removed yields the same labeled species (Fig. 10d). The possibility that membranes could have contributed either the kinase itself or the acceptor proteins can therefore be eliminated.

J.-M. Blanchard, C. Brunel, and P. Jeanteur

129

Fig. 10. Identification of phosphate acceptor proteins by sodium dodecylsulphute~polyacrylumidegel electrophoresis. 60 pg hnRNA . proteins were phosphorylated in vitro under standard conditions except for a 15-min incubation time, precipitated and electrophoresed as described in Materials and Methods. (a) Stained gel previously treated with hot trichloracetic acid. (b) Autoradiogram of the same gel. (c) Autoradiogram of hnRNA . proteins phosphorylated as above but treated with 0.2 M sulfuric acid [67] before precipitation and electrophoresis. (d) Autoradiogram of hnRNA . proteins phosphorylated after purification in metrizamide gradient

DISCUSSION As an approach to the study of regulatory processes involved in eucaryotic mRNA biosynthesis which might operate at the level of ribonucleoprotein complexes containing hnRNA, a search was undertaken for enzymes which might be associated with these particles. The present investigation demonstrates that hnRNA . proteins prepared from sonically disrupted HeLa cell nuclei according to Kish and Pederson [28] contain a phosphoprotein kinase and several phosphate acceptor proteins which can be phosphorylated on serine and threonine residues. We had previously reported a preliminary characterization of such an activity in HeLa hnRNA . proteins [23] prepared by diffusion out of nuclei [I71 in which case no rearrangement due to sonication could be invoked. A similar activity has also been observed in nuclear particles from rat liver [69]. Prolonged incubation periods led to a decline in acid-precipitable 32P radioactivity which could reflect the presence of a proteolytic activity in our hnRNA . proteins. Controls show, however, that this putative activity is not due to a serine protease nor is it able to

hydrolyse bovine serum albumin. Therefore, only a highly specific protease cannot yet be ruled out. Alternatively, involvement of a phosphoprotein phosphatase remains likely (unpublished results). Should this be the case, it would suggest that an equilibrium between phosphorylated and dephosphorylated states of some hnRNA . proteins might be relevant to their physiological function. We have devoted considerable effort to exclude the possible contribution to the enzymatic activity observed of any conceivable cellular contamination and to provide unambiguous evidence for the association of kinase and acceptor proteins with structures containing hnRNA. These were defined by their content of RNA pulse-labeled in the presence of 0.04 pg/ml actinomycin D and their ability to bind to oligo(dT)-cellulose. The following possible sources of contamination have been considered and ruled out : soluble proteins of any origin, chromatin, cytoplasmic polysomes or nuclear preribosomes. Although possible at the stage of purified unfractionated hnRNA . proteins, contamination by soluble proteins would not be expected to withstand treatments like sedimentation through sucrose or metrizamide gradients. The presence of

130

chromatin was excluded by several arguments: no significant levels of labeled DNA were found within the sucrose gradient, lack of effect of DNAse treatment on sedimentation behaviour of both [3H]RNA and enzyme activity, absence of activity at the density of chromatin (about 1.2 g/cm3) in metrizamide, and absence of histones in gels. Contamination by preribosomes was made unlikely by the absence of any peak of radioactive RNA (labeled without actinomycin D) at densities higher 1.4 g/cm3 in cesium chloride gradients. The contribution of ribosomes was excluded by experiments carried out in the presence of EDTA which showed no dissociation effect in sucrose gradient and no density shift in metrizamide as well as by electron-microscopic observations of nuclei. As to membrane contaminants, although there seems to be some, their contribution to the present data was conclusively ruled out. Direct evidence for the association of kinase with hnRNA . proteins comes from the demonstration that its behaviour paralleled that of hnRNA in three types of experiments carried out under a variety of conditions : (a) the sedimentation behaviour of both hnRNA ' proteins and kinase in sucrose gradients; (b) isopycnic centrifugation in metrizamide showed that hnRNA proteins and kinase banded at the same density (about 1.3 g/cm3) regardless of the presence of EDTA; (c) retention of hnRNA . proteins and kinase by oligo(dT)-cellulose was found to be dependent on the availability of both poly(A) from the particles and oligo(dT) from the cellulose. Analysis of phosphorylated substrates by electrophoresis on sodium dodecylsulphate/polyacrylamide gels revealed that, although they spanned a wide range of molecular weights, the most intensely labeled species were localized in regions around M , 37000 and 28000. We have also observed that two polypeptides of similar molecular weight are also labeled in vivo after exposure of HeLa cells to [32P]phosphate [21]. In this latter case, at least two additional species of M , 30000 and 52000 appear to be also consistently labeled. The data in this paper are part of the doctoral thesis of J. M. Blanchard. We are grateful to Dr E. Puvion for electron microscopy of nuclei and to Dr M. Jacob for stimulating discussion and criticism throughout this work. Thanks are due to Mrs A. Vi6 for expert technical assistance and to Miss P. Laur for her help in typing the manuscript. This work was supported by grants from Centre National de la Recherche Scientifque, Institut National de la Santt et de la Recherche Medicale, Delegation General a la Recherche Scientifique et Technique, the Ligue Nationale Contre le Cancer and the Fondation pour la Recherche Medicale.

REFERENCES 1. Darnell, J. E. (1975) Harvey Lect. 69, 1-48. 2. Ruiz-Carillo, A,, Beato, M., Schurtz, G., Feigelson, P. & Allfrey, V. G. (1973) Proc. Nut1 Acad. Sci. U.S.A. 70, 36413645.

Protein Kinase Activity Associated with hnRNA . proteins 3. Imaizumi, T., Diggelmann, H. & Scherrer, K. (1973) Proc. Natl Acad. Sci. U.S.A. 70, 1122-1126. 4. Brandhorst, B. P. & McConkey, E. H. (1974) J . Mol. Biol. 85,451-464. 5. Soeiro, R., Vaughan, M., Warner, J. R. & Darnell, J. E. (1968) J . Cell. Biol. 39, 112-118. 6. Gall, J. G. (1956) 1. Biochem. Biophys. Cytol. 2 (suppl.) 393396. 7. Gallan, H. G. & Lloyd, I. (1960) Phil. Trans. R. Soc. Lond. Ser. B , Biol. Sci. 243, 135-219. 8. Stevens, B. J. & Swift, H. J. (1966) J . Cell. Biol. 31, 55-77. 9. Monneron, A. & Bernhard, W. (1969) J. Ultrastruct. Res. 27, 266-288. 10. Miller, 0 . L. & Bakken, A. H. (1973) Karolinska Symp. Res. Methods Reprod. Endocrinol. 5, 155- 177. 11. Puvion, E. & Bernhard, W. (1975) J . Cell. Biol. 67,200-214. 12. Samarina, 0. P., Lukanidin, E. M., Molnar, J. & Georgiev, G. P. (1968) J . Mol. Biol. 33, 251-2263, 13. Moule, Y. & Chauveau, J. (1968) J . Mol. Biol. 33, 465-481. 14. Kdhler, K. & Arends, S. (1968) Eur. J . Biochem. 5, 500-506. 15. Stevenin, J., Mandel, P. & Jacob, M. (1970) Bull. Soc. Chim. Biol. 52, 703 - 720. 16. Niessing, J. & Sekeris, C. E. (1971) Biochim. Biophys. Acta, 247, 391 -403. 17. Ducamp, C. & Jeanteur, Ph. (1973) Biochimie (Paris) 55, 1235- 1243. 18. Pederson, T. (1974) J . Mol. Biol. 83, 163-183. 19. Matringe, H. &Jacob, M. (1972) Biochimie (Paris) 54, 11691178. 20. Gallinaro-Matringe, H. & Jacob, M. (1973) FEBS Lett. 36, 105-108. 21. Blanchard, J. M., Brunel, C. & Jeanteur, Ph. (1977) Biochrm. Soc. Trans. 5, 670- 671. 22. Raskas, H. J. (1971) Nut. New Bid. 233, 134-136. 23. Blanchard, J. M., Ducamp, C. & Jeanteur, Ph. (1975) Nature (Lond.) 253,467 - 468. 24. Zimmerman, S. & Sanden, G. (1966) Anal. Biochem. 14, 269 - 277. 25. Loening, U. E. (1967) Biochem. J. 202, 251 -257. 26. Brown, S., Teplitz, M. & Revel, J. P. (1974) Proc. Natl Acad. Sci. U.S.A. 71,464-468. 27. Perry, R. P. & Kelley, D. E. (1968) J . Mol. Biol. 35, 37-59. 28. Kish, V. M. & Pederson, T. (1975) J . Mol. Biol. 95, 227-238. 29. Bernhard, W. (1969) J . Ultrastruct. Res. 27, 250-265. 30. Martelo, 0. J., Woo, S. L. C. & Davie, E. W. (1974) J . Mol. Biol. 87,685 - 696. 31. Rickwood, D. & Birnie, G. D. (1975) FEBS Lett. 50, 102- 110. 32. Kuno, N. & Kihara, H. K. (1967) Nature (Lond.) 215, 974975. 33. Studier, F. W. (1973) J. Mol. Biol. 79, 237-248. 34. Maizel, J. V., Jr (1971) Method. Virol. 5, 179-246. 35. Weber, K. & Osborne, M. (1969) J . Biol. Chem. 244, 44064412. 36. Glynn, I. M. & Chappel, J. B. (1964) Biochem. J . 90, 147- 149. 37. Berthoux, F. C. (1973) Pathol. Biol. 21, 83-97. 38. Linn, S. & Lehman, I. R. (1965) J . Biol. Chem. 240,1287- 1293. 39. Martonosi, A. (1969) J . Bid. Chem. 244, 613-620. 40. Chi-Ching Chen, Smith, D. L., Bruegger, B. B., Halper, R. M. & Smith, R. A. (1974) Biochemistry, 13, 3785-3789. 41. Rubin, C. S. & Rosen, 0. M. (1975) Annu. Rev. Biochem. 44, 831 - 887. 42. Takeda, M., Yamamura, H. & Ohga, V. (1971) Biochem. Biophys. Res. Commun. 62, 642-650. 43. Roddon, R. W. & Anderson, S. L. (1972) Biochem. Biophys. Res. Commun. 44, 1345-1350. 44. Desjardins, P. R., Lue, P. F., Liew, C. C. & Gormdll, A. G. (1972) Com. J . Biochem. 50, 1249- 1259. 45. Kish, V. M. & Kleinsmith, L. J. (1974) J . Biol. Chem. 249, 750- 760.

131

J.-M. Blanchard, C. Brunel, and P. Jeanteur 46. Lake, R. S. (1973) J. Cell. Biol. 58, 317-331. 47. Gibson, K., Tichonicky, L. & Kruh, J. (1974) Biochimie (Paris) 56,1417-1423. 48. Farago, A,, Antonini, F. &Fabian, F. (1974) Biochim. Biophys. Acta, 370, 459-467. 49. Kruh, J., Courtois, Y. & Tichonicky, L. (1975) Biochimie (Paris) 57, 1323- 1329. 50. Schlepper, J. & Knippers, R. (1975) Eur. J . Biochem. 60, 209 - 220. 51. Eil, C. & Wool, I. G. (1971) Biochem. Biophys. Res. Commun. 43,1001 - 1009. 52. Kabat, D. (1971) Biochemistry, 10, 197-203. 53. Walton, G. M. & Gill, G. N. (1973) Biochemistry, 12, 2604261 1. 54. Quirin-Stricker, C., Schmitt, M., Egly, J. M. & Kempf, J. (1976) Eur. J. Biochem. 62, 199 - 209. 55. Olson, M. 0. J., Prestayko, A. W., Jones, C. E. & Busch, H. (1974) J . Mol. Biol. 90, 161- 168. 56. Grummt, I. & Grummt, F. (1974) FEBS Lett. 99, 129-132. 57. Mirault, M. E. & Scherrer, K. (1971) Eur. J . Biochem. 23, 372- 386.

58. Girard, M., Latham, H., Penman, S. & Darnell, J. E. (1965) J . Mol. Biol. 11, 187-201. 59. Douvas, A. S., Harrington, C. A. & Bonner, J. (1975) Proc. Natl Acad. Sci. U.S.A. 72, 3902-3906. 60. Lazarides, E. & Lindberg, V. (1974) Proc. Nut1 Acad. Sci. U . S . A . 71,4742-4746. 61. Sekeris, C. E. & Niessing, J. (1975) Biochem. Biophys. Res. Commun. 62,642-650. 62. Spirin, A. S. (1969) Eur. J . Biochem. 10, 20-35. 63. Houssais, J. F. (1975) FEBS Lett. 56, 341 - 347. 64. Buckingham, M. E. & Gros, F. (1975)FEBSLett. 53,355-359. 65. Kumar, A. & Pederson, T. (1975) J . Mol. Biol. 96, 353-365. 66. Jelinek, W., Adesnik, M., Salditt, M., Sheiness, D., Wall, R., Molloy, G., Philipson, L. & Darnell, J. E. (1973) J . Mol. Biol. 75,515-532. 67. Gold, A. M. & Fahrney, D. (1964) Biochemistry, 3, 783-791. 68. Bhorjee, J. S. & Pederson, T. (1972) Proc. Natl Acad. Sci. U.S.A. 69, 3345 - 3349 69. Schweiger, A. & Schmidt, D. (1974) FEBS Lett. 41, 17-19.

J.-M. Blanchard, The Rockefeller University, 66th Street and York Avenue, New York City, New York, U.S.A. 10021 C. Brunel and P. Jeanteur *, Laboratoire de Biologie Moleculaire, Universite des Sciences et Techniques de Languedoc, Place Eugtne-Bataillon, F-34060 Montpellier-Cedex, France

* To whom correspondence should be addressed