Characterization of the non-histone nuclear proteins associated with rapidly labeled heterogeneous nuclear RNA.

Heterogeneous nuclear RNA (HnRNA) is associated with a set of specialized RNA-binding proteins. Mild ribonuclease digestion of intact HnRNA-protein complexes released 15 S ribonucleoprotein (RNP) particles containing poly(A) and its associated protein and 40 S RNP particles containing most of the HnRNA sequences. Highly purified 40 S RNP particles have been obtained from rat liver by centrifugation of nuclear extracts on sucrose density gradients and isopycnic banding on Metrizamide density gradients. These RNP preparations contain 27% of the total HnRNA sequences of rat liver, and appear homogeneous when viewed in negative contrast in the electron microscope and by centrifugation studies using velocity sedimentation in sucrose density gradients or isopyenic banding in density gradients of cesium salts. Analysis of the proteins in the rat liver 40 S RNP particles by two-dimensional gel electrophoresis demonstrated that the 40 S RNP particle is composed of 12 major protein components with molecular weights ranging from 29,000 to 42,000 which accounted for 75% of the total protein mass and 13 minor protein components with molecular weights greater than 42,000. The proteins in the 29,000 to 42,000 group were fractionated by ion exchange chromatography. The amino acid compositions of the purified protein fractions were strikingly similar and shared several unusual features that distinguish these HnRNA-associated proteins, as a group, from the histones and the non-hi&one chromosomal proteins. Each of the RNP proteins have basic charge characteristics (p1 greater than 8.0) high glycine (25 mol %), low cysteine, and little detectable methionine. Like the histones, the HnRNP proteins are subject to extensive postsynthetic modification. We have identified the unusual amino acid NC, NC (CH,),-L-arginine in acid hydrolysates of some of the RNP proteins and shown that this amino acid

on an AA-15 column (0.9 x 20 cm) equilibrated and eluted with 0.05 N HCl. The column was eluted under pressure at a flow rate of 40 ml/h and fractions of 1 ml were collected and monitored for radioactivity.

Electrophoretic Analyses
Proteins associated with RNP particles and other nuclear fractions were characterized first by size separation in linear 8.75 to 14% SDS-polyacrylamide gradient slab gels (1.5 x 150 x 170 mm) using the discontinuous buffer system of Laemmli (45). The protein bands were visualized after staining in 0.1% Coomassie brilliant blue R-250 in 50% ethanol, 7.5% acetic acid, and subsequent destaining in 7.5% acetic acid.
Two-dimensional gel electrophoretic analyses of nuclear proteins were based on mobility differences in acid-urea gels (46)  ASS iociated with HnRNA are recovered in a peak of UV absorbing material with a sedimentation coefficient of approximately 40 S. The particulate nature of this material is evident in electron micrographs which reveal a relatively uniform population of discrete particles approximately 20 nm in diameter (Fig. 2). No contamination of these particles by chromatin, membraneous material, or ribosomal subunits has been observed and no DNA was detectable in the 40 S peak region by the diphenylamine reaction. Direct chemical analysis showed that nearly 25% of the total nuclear RNA can be recovered in the form of 40 S particles containing about 5% of the total nuclear protein (Table I).
In addition to the 40 S RNP particle peak, a smaller peak of 32P activity and UV absorption occurs in the 12 to 20 S region of the gradient (Fig. 1, lower panel). Particles of this size have been shown to contain the 3'-terminal poly(A)-rich sequences of newly synthesized HnRNA chains (10, 14). The relative increase in the labeling of the 12 to 20 S peak with [32P10rthophosphate compared to the labeling of the 12 to 20 S peak with 13H]orotic acid (Fig. 1 to equilibrium in Metrizamide gradients of low ionic strength. Most of the 32P and 3H activity bands as a sharp peak of buoyant density at 1.29 g/cm3. Purification of the HnRNP particles in the way is accompanied by a loss of 10 to 20% of the [3H]leucine-labeled protein present in the 40 S peaks from sucrose gradients. These proteins are recovered as a small peak of [3H]leucine-labeled material banding at 1.38 g/cm3. After Metrizamide gradient centrifugation the RNP fraction contains protein and RNA in a ratio of 3.70 to 1, and more than 10% of the total hepatocyte RNA rapidly labeled in uivo with 13Hlorotic acid (or 27% of the total HnRNA (Table I)).
Banding RNP particles in Metrizamide gradients of low ionic strength does not substantially alter either the sedimentation coefficient of the 40 S particles (data not shown) or the equilibrium bouyant density of the particles in gradients of cesium salts (see below), indicating that the 40 S RNP particle structure remains intact throughout the isolation procedure. Characterizatiop of Proteins Co-purifjing with HnRNA by Polyacrylamide Gel Electrophoresis -The proteins appearing in nuclear subfractions at successive stages in the preparation of HnRNP particles have been characterized by electrophoresis in SDS-polyacrylamide gradient gels. Complex mixtures of proteins ranging in molecular weight from less than 20,000 to more than 150,000 are released during the washing of the isolated nuclei (data not shown). The subsequent extractions at pH 8.0 remove the bulk of the nuclear HnRNP particles, together with other nuclear proteins which were separated by sucrose density gradient centrifugation (Fig. 4~). About 35 protein species ranging in molecular weight from 25,000 to FIG x-x, density (gmlcm?. over 150,000 co-sediment with RNA in the 40 S RNP particle fraction. Of these proteins more than 70% are represented by a group of proteins of molecular weight from 29,000 to 42,000 (Fig. 4b). The proteins of this group are in constant relative proportions in replicate preparations and the same proportions are observed whether the relative amount of each band is determined by densitometry of gels stained with Fast Green, or Coomassie brilliant blue R-250 or of autoradiographs of W-labeled protein bands. By each of these methods, the major proteins that co-sediment with HnRNA as 40 S particles appear as a unique and constant subset of the total proteins present in the pH 8.0 extract (compare Fig. 4, a and b). Other proteins mainly of molecular weight greater than 42,000 distribute characteristically elsewhere on the sucrose gradients, some of them appearing in the 12 to 20 S particle region (e.g. Fig. lob).
All the proteins in the 29,000 to 42,000 molecular weight group remain tightly associated with RNA during centrifugation of the RNP particles in Metrizamide density gradients, however, many of the proteins of higher molecular weight are separated from the RNP particles (compare Fig. 4, b and c) and observed to band at a density corresponding to that of free protein (Fig. 3). It seems likely that the major proteins co-purifying with RNA through the Metrizamide gradients are structural proteins required for the assembly of the 40 S complex. Treatment of the complex with pancreatic ribonuclease (0.01 pg/ml) for 30 min at 4" or exposure to nonionic detergents such as Triton X-100 or Brij 35 (0.1% (v/v)), and centrifugation on gradients of sucrose or Metrizamide does Nuclear Proteins Associated with HnRNA not alter either the proportions or numbers of the proteins in the complex. In contrast, exposure of the HnRNP particles to high ionic strength (see below) or to protein-denaturing conditions (such as concentrated urea or ionic detergents) permits selective solubilization of some of the proteins (2, 3, 13, 18) but results in disaggregation of the 40 S particles. The association of a characteristic subset of nuclear proteins with HnRNA molecules does not appear to be an artifact of a particular method of nuclear fractionation.
We have prepared 40 S particles by three other methods; involving sonication, hypotonic lysis, and hypertonic lysis of isolated nuclei, followed by cenbrifugation of the released particles in sucrose gradients. Although recoveries of 40 S particles varied somewhat with the method used, the electrophoretic banding patterns of the proteins in the 40 S particles prepared by all four methods are virtually indistinguishable (data not shown). Two-dimensional Gel Electrophoresis of the Protein Components of Purified HnRNP Particles -A better estimate of the complexity of the protein complement of the 40 S HnRNP particles isolated by Metrizamide gradient centrifugation was obtained by two-dimensional. gel electrophoresis. The HnRNP proteins (solubilized in urea) are separated according to charge by electrophoresis in denaturing gels at pH 4.3, and further fractionated by electrophoresis using polyacrylamide gels containing 0.1% SDS, as shown in Fig. 5. Twenty-five proteins are reproducibly resolved; they have been numbered consecutively in order of increasing molecular weight. It is significant that protein bands which appear to contain only one or two components in charge separating gels show higher orders of complexity in the two-dimensional gel system. Thus, the major subset of proteins in the 29,000 to 42,000 molecular weight class (75% of the protein mass of the Metrizamide gradient-purified HnRNP particles) all have very similar mobilities in denaturing gels at pH 4.3. (This accounts for earlier reports on the apparent simplicity of the HnRNAassociated proteins (e.g. 3, 61.) The two-dimensional gel separations show this group of proteins to include five major species (Numbers 4, 5, 6, 8 and 9) and seven minor species (Numbers 1, 2, 3, 7, 10, 11, and 12) (Fig. 5).

HnRNP Particles Are Macromolecular Complexes Stabilized by Electrostatic
Interactions Between RNA and Proteins -The close association between 32P-labeled HnRNA and a characteristic subset of [3H]leucine-labeled proteins persists throughout centrifugation of the unfixed 40 S particles in neutral sucrose or Metrizamide gradients provided the ionic strength of the medium is kept low (Fig. 3A, Fig. 4c), but raising the salt concentration to 600 mM NaCl leads to a disruption of the particles and release of the RNA-associated proteins. Analysis of the doubly labeled 40 S particles in Metrizamide gradients containing 600 mM NaCl shows a separation of the 32P-labeled nucleic acids, which band at 1.25 g/cm3, a buoyant density which corresponds to that of free RNA in Metrizamide (371, from the 13H11eucine-labeled proteins, which band at a density of 1.33 g/cm3 (Fig. 3I3).
The stability of the HnRNP particles at low ionic strength and their dissociation at high salt concentrations was confirmed by isopycnic banding of the peaks from Metrizamide gradients in density gradients of Cs,SO, or CsCl (Fig. 6). HnRNP particles, labeled in uiuo with nucleotide precursors were purified in sucrose gradients and fixed with glutaraldehyde before centrifugation in gradients of cesium salts. The 40 S particles, unable to dissociate because of fixation, banded as sharp peaks with a density of 1.43 g/cm3 in Cs,SO, and 1.385 g/cm" in CsCl (Fig. 6, A and D). After further puritication of the 40 S particles by centrifugation in Metrizamide gradients of low ionic strength, the glutaraldehyde-fixed RNP particles banded with a density of 1.44 g/cm3 in C&SO, and 1.395 g/cm" in CsCl (Fig. 6, B and E). However, if 40 S particles from sucrose gradients were exposed to 600 mM NaCl during Metrizamide gradient centrifugation, the particles dissociated and most of the resulting RNA banded in C&JO, at a buoyant density of 1.66 g/cm3, very close to the density of free RNA (Fig. 6C). Similarly, when the 40 S particles were exposed to high ionic strength during Metrizamide gradient centrifugation, most of the radioactivity applied to the CsCl gradients was recovered in the pellet, while less than 10% of the applied radioactivitv was observed in a small peak at 1. 44 5. Two-dimensional gel electrophoresis of the proteins assosis. Strip at the top of the two-dimensional gel shows the proteinciated with 40 S RNP purified by centrifugation in Metrizamide banding pattern after electrophoresis in the first demension of pH (Fig. 3). Reduced and alkvlated proteins were subiected to electrophoresis at pH 4.3 in 7.5% acryiamide gels containing 6 M urea.
4.3, while the strip at the right of the two-dimensional gel shows Strips of the first dimension gel were applied to SDS-polvacrvlamide protein banding pattern after one-dimensional SDS-gradient gel gradient gel slabs and elec<rophoresed for 18 h. The fractionated electrophoresis. The schematic diagram at left plots the distribution of protein spots resolved by two-dimensional electrophoresis and proteins were visualized by staining with Coomassie Brilliant Blue R-250. The composite photograph at right shows the distribution of assigns numbers to the spots in order of increasing molecular weight. A molecular weight scale plotting the mobilities of standard the 25 major RNP protein spots after two-dimensional electrophore-proteins of known molecular weight (Fig. 4) is shown at the far left. stable macromolecular complex which migrates as a single sharp band in 1% agarose gels (Fig. @I). Cleavage of this complex, by reduction with 2-mercaptoethanol, releases each of the proteins of the original 40 S particle (Fig. 8) The proteins which were not retained on the DEAE-Sephadex column (73% of the total 'C-labeled protein) were examined by SDS-polyacrylamide gel electrophoresis (Fig. 4d). Each of the proteins in the 29,000 to 42,000 molecular weight group (with the exception of protein 9) were separated from the more acidic proteins of higher molecular weight. Comparable chromatographic purification of these proteins may be obtained using either the total pH 8.0 nuclear extracts or 40 S RNP particles isolated from sucrose or Metrizamide gradients as starting material. When total rat liver non-histone nuclear proteins ars chromatographed on DEAE-Sephadex columns in Buffer F, more than 60% of the proteins in the excluded fraction are the RNA-binding proteins in the 29,000 to 42,000 molecular weight group (data not shown), suggesting that the RNAbinding proteins are chemically distinct from the majority of the chromosomal non-hi&one nuclear proteins.
The major HnRNA-associated proteins were separated by a series of chromatographic steps using the cation exchange resins phosphocellulose and carboxymethylcellulose (Fig. 9). Typical purified protein fractions are shown in Fig. 10 which depicts a SDS-polyacrylamide gel analysis of the homogeneity of the isolated proteins. Using this fractionation scheme it was possible to obtain samples of proteins 3, 4, 5, and 8 with greater than 80% electrophoretic homogeneity. Proteins 6 and 7 were only partially resolved by the chromatographic procedures, and unresolved by SDS-gel electrophoresis.
Proteins 10 Table II. The amino acid compositions of the purified proteins are strikingly similar, and share several unusual features that distinguish the RNP proteins, as a group, from the rest of the non-hi&one nuclear proteins and from the histones (Table II). Each of the proteins in this group have extremely high glycine contents (typically 25 mol %), little detectable methionine, and low cysteine contents. The proteins have between 1.55 and 2.26 mol 8 histidine, and between 4.93 and 6.16 mol % lysine. Protein fractions 2, 6 + 7, 8, and 10 + !2 contain the unusual amino acid NCflG(CH,)2-targinine.
The identification and in vivo synthesis of this amino acid is described below. The arginine contents of the proteins varies from 4.98 to 7.62 mol % (including the contribution of NG,NG(CH,),+arginine), while the dimethylarginine content of the protein fractions varies from undetectable (protein 4) to 32% of the arginine residues (proteins 6 + 7). In contrast, amino acid analyses of total rat liver histone fractions show the presence of considerably more methionine, lysine, and arginine (1.80, 12.00, and 11.10 mol %, respectively) in the histones than in the HnRNA-associated proteins. The histones have significantly less glycine than do the RNP proteins (8.7 mol % compared to 25.0 mol I) and no detectable NG,NG(CH,),-r,-arginine (Table II). Differences in amino acid composition between the chromosomal non-histone proteins and the HnRNA-associated proteins are also evident (Table II).
Extensive Methylation of the Proteins of 40 S RNP Particles Leads to Formation of the Unusual Amino Acid NG,NG(CHdr  (50). We have identified this amino acid asN",iV'YCH,),-L-arginine by chromatography with synthetic standards using an amino acid analyzer modified to permit complete separation of all the methylated derivatives of the basic amino acids (Fig. 11,Refs. 43 and 44). The evidence may be summarized as follows: (a) the amino acid elutes from the amino acid analyzer in the same position as N"JVG(CH,),-L-arginine but   (Fig. 9). a, nuclear extracts (pH 8.0) containing 40 S particles; 6, proteins in 12 to 20 S region of sucrose gradients (Fig. 1); c, 40 S nuclear RNP particles prepared by sucrose gradient centrifugation (Fig. 1); d, proteins excluded from DEAE-Sephadex A-25 (Fig. 4d). e, protein Fraction 3; f, protein Fraction 6 + 7; g, protein Fraction 4; h, protein Fraction 5; i, protein Fraction 8. Scale at left indicates mobility of protein molecular weight standards (Fig. 4).
3Hlmethionine, suggesting that this unusual amino acid arises from postsynthetic modification of arginine residues in vivo (43,51,52). After labeling in vivo with [methyZJH]methionine (Fig. 11) more than 70% of the radioactivity found in the 40 S particle proteins could be recovered as W,W(CH,),+arginine. Almost 10% of the radioactivity was found in the region where W(CH,)-L-arginine elutes (although no ninhydrin-positive material could be detected). This amino acid may represent an intermediate in the formation of the dimethylarginine. A small unidentified peak of radioactivity was also observed to elute between ammonium and dimethylarginine.
No modification of lysine or histidine residues was detectable, either by ninhydrin reaction or uptake of label in vivo, although methylation reactions of this kind are known to occur in histones (44).
Quantitative analysis of the NC~G(CH3)2-L-arginine content of various nuclear fractions demonstrates that more than 67% of the total nuclear NG,NG(CH&-Garginine may be recovered in 40 S particles isolated on sucrose gradients (data not shown). Since the rihonucleoprotein particles are not quantitatively recovered during the nuclear fractionation procedures (Table I), these data suggest strongly that the methylation of arginine residues found in nuclear protein fractions almost exclusively involves the proteins associated with HnRNP complexes.

Phosphorylation of RNP Proteins
-Evidence that the 40 S proteins contain phosphorylated amino acids was obtained from analyses of the 3ZP-l+beled amino acid content of 40 S RNP proteins subjected to partial acid hydrolysis. In a typical experiment, 53% of the total protein 3zP activity was recover- Rat liver 40 S HnRNP particles were purified by sucrose gradient NC, NG(CH&bArginine elutes between ammonium and arginine centrifugation as described under "Experimental Procedures." The using standard amino acid analysis procedures (19). Total rat liver 40 S nuclear RNP particle proteins were solubilized in 6 M urea and histone was prepared by extraction of purified rat liver nuclei with fractionated by ion exchange chromatography on DEAE-Sephadex 0.25 M HCl (36). The non-histone chromosomal proteins were pre-A-25, phosphocellulose, and carboxymethylcelluloae. Aliquota of the pared by phenol extraction of chromatin depleted of RNP particles purified proteins (numbered as indicated in Figs. 5 and 9) were by extraction with Buffer B and histones by extraction with 0.25 M hydrolyzed in 6 M HCl, 18 h, llo", and analyzed with a Beckman HCl (36). model 120 B amino acid analyzer or a Durrum amino acid analyzer. able as phosphoserine, 3% as phosphothreonine, and the remainder as inorganic phosphate generated by hydrolysis of the phosphorylated amino acids. The phosphoproteins appearing in the HnRNP particle fraction have been characterized by SDS-polyacrylamide gel electrophoresis of RNP proteins labeled in uivo for 45 min with [32P]orthophosphate (Fig. 12). Most of the 32P activity in the 40 S particle fraction is incorporated by proteins 6, 7, 8, and 9 (Fig. 12c). In contrast, protein 4, which is the protein present in greatest molar quantities in the 40 S particle fraction was not detectably phosphorylated.
Since the serine and threonine contents of the RNP proteins do not differ significantly (Table II), these differences in the content of phosphorylated amino acids probably result from sequencespecific recognition of phosphorylation sites by nuclear protein kinases. Other nuclear phosphoproteins are recovered in pH 8.0 nuclear extracts (Fig. l&z) and in the 12 to 20 S region of sucrose gradients (Fig. 1%).
The proteins in the 40 S RNP complex may also be phosphorylated in vitro by protein kinases (Fig. 13). We have confirmed a previous report (53) that 40 S RNP particles contain an associated protein kinase activity, simply by incubating 40 S particles in the presence of [y-32PlATP and an appropriate buffer. Consistent with the results of the in vivo phosphorylation experiments, proteins 6, 7, 8, and 9 rapidly incorporate 3zP radioactivity while protein 4 does not appear to incorporate 32P radioactivity.
It is of interest that proteins 6, 7, and 8 are also subject to modification by methylation of arginine residues while protein 4 is not methylated (Table II). The protein kinase also labeled to high specific activities a number of proteins with molecular weights less than 27,000 and greater than 67,000 that were not phosphorylated in vivo. Addition of CAMP did not modify the rate or the specificity of the 40 S RNP-associated protein kinase. Addition of partially purified protein kinase from calf thymus (gift of Dr. E. M. Johnson, The Rockefeller University) increased the rate of phosphorylation of the 40 S RNP proteins (Fig. 13), but did not signilicantly alter the substrate specificity of the reaction. The rate of phosphorylation of protein 9 was particularly increased by the exogenous kinase. Even in the presence of protein kinase from another tissue, protein 4 remained unphosphorylated. the position of protein 4, and brackets indicate the positions of the two major rat liver histone H-l proteins. Scale at let% plots the mobility of protein molecular weight standards (Fig. 4).
Estimates of Protein Stoichiometry in Rat Liver 40 S HnRNP Particle - Table III summarizes the available chemical data concerning the major proteins found in rat liver 40 S ribonucleoprotein particles. The proteins in the 29,000 to 42,000 molecular weight group appear to be the structural proteins of the HnRNP particles by the following criteria. These proteins represent more than 75% of the protein mass of RNP particle fractions and are closely related by charge (Table III) and amino acid compositions (Table II). Each of Nuclear Proteins Associated with HnRNA 7319 the proteins in this fraction are present in apparently constant Assuming a protein to RNA ratio of 3.7:1 (Table I, Refs. 3, 11,  proportions throughout a rigorous isolation procedure (Fig. 13,and 52), the total mass of the 40 S particle may then be 41. The proteins of higher molecular weight which are someestimated to be 1.3 to 1.5 x lo6 daltons. This molecular mass times present in HnRNP particle preparations (4,9,12,13, estimate is in close agreement with the estimates for the 18) are not likely to be structural proteins. These proteins are molecular mass of the 40 S particle (1.4 to 1.6 x lo6 daltons) present in limited quantities of 40 S particles isolated by based on comparison to the molecular masses of ribosomal sucrose gradient centrifugation, and many are lost when 40 subunits (57,58). S particles are further purified by centrifugation on Metriza-Although these estimates should not be regarded as accumide gradients (Fig. 4). Proteins in this fraction contain rate until confirmed by alternative procedures, they provide a enzymatic activities, including a protein kinase activity (541, useful illustration that a unique protein structure composed endoribonuclease (55), and a homopolymer synthetase activity of major basic RNP proteins can be reasonably proposed. (56).
Whether other proteins are required to maintain this core Making the assumptions that only the proteins in the structure, and whether the high molecular weight proteins 29,000 to 42,000 molecular weight group are required for the have organizational roles in the RNP structure awaits further maintenance of the 40 S particle, and that the RNP fractions investigation. . Side by side comparisons with proteins of rat liver 40 S number per particle of the more abundant proteins may be particles (Fig. 14c) show that the proteins obtained from each estimated. By multiplying the estimated copy number per particle by the molecular mass of the proteins estimated by abcdef g SDS-polyacrylamide gel electrophoresis, and adding the calculated protein masses, the total protein mass of the 40 S particle may be estimated to be 1.1 to 1.2 x lo* daltons. in uivo into proteins fractionated by SDS-polyfrom the column were applied to a 150-ml LKB isoelectric focusing acrylamide gel electrophoresis (Fig. 12). The relative abundance of column containing 1% (v/v) Ampholines, pH 3 to 10, 4 M urea, and a the proteins was estimated by comparing the radioactivity incorpogradient of 10 to 40%' sucrose. The apparatus was operated at 4" for rated in vitro into proteins separated by chromatography on phos-12 h at 200 V and for 24 h at 500 V using 0.1 N HCl as the anodic phocellulose ( Fig. 9Al and confirmed by analysis of densitometry buffer, and 0.1 N NaOH as the buffer at the cathode. At the end of tracings of the stained proteins fractionated by SDS-polyacrylamide the electrophoresis period, when the current had stabilized, 1.5-ml gradient gel electrophoresis (Fig. 4). In this paper, we have proposed a possible model for the 40 S RNP particle of rat liver, based on the assumption that each of the proteins in the 29,000 to 41,000 group is present in at least one copy per 40 S particle (Table  III) 11). and HeLa S-3 cells (62) and iV",Am(CH,,),-arginine has been observed as a component of HeLa S-3 cell and P. polycephalum RNAbinding proteins (52). Similarly mammalian, avian, and amphibian ribonucleoprotein particle-associated proteins have been reported to share a number of antigenic determinants (59).* However, side by side comparisons of HnRNA-associated proteins from a variety of species (Fig.  14) show differences in the electrophoretic mobility of the major proteins.

Content of modified amino acids
The extent of homologies between HnRNP-associated proteins in different species will not be properly understood until sequence studies of the purified proteins have been completed.