J. Mol. BioE. (1978)

119, l-20

Precursors of Alpha and Beta Globin Messenger RNAs JEFFREY Ross AND DAVID A. KNECHT McArdle

Laboratory for Cancer Research Universiby of Wisconsi~L Madison, Wk. 53706, 1J.S.A. ( Received 24 Jme

1977)

Several structural properties of the pulse-labeled globin messenger RNA precursor molecules that sediment faster than steady-state (10 S) globin mRNA (Curtis & Weissmann, 1976; Ross, 1976; Kwan et al., 1977; Bastos 8: Aviv, 1977) are investigated. The radioactive, globin-specific RNA resolves into two peaks during electrophoresis in polyacrylamide gels prepared in 98yb formamide. The molecular weights of the larger and the smaller RNAs are 600,000 and 280,000, respectively, as compared with the average molecule weight of cytoplasmic globin mRNA, 200,000. When RNA from cells labeled for 10 minutes is analyzed by formamide gel electrophoresis, a precursor larger than 600,000 .V, is not detected. Most, if not all, of the molecules in the 600,000 and 280,000 .14, RNA fractions are polyadenylated. The 600,000 M, RNA contains /3 globin lnRNA sequences, but little, if any, LXglobin sequences; the 280,000 M, RNA c*ont,ains both c( and ,5?sequences. At steady-state the 14-day-old mouse fetal liver erythroblast contains approxima,tely 50 molecules of the 600,000 1M, RNA, 1000 molecules of the 280,000 M, RNAs, and 60,000 molecules of mature globin mRNA. The half-lives of the 600,000 and 280,000 M, RNAs calculated using steady-state levels are 2 minutes and 17 minutes, respectively. These data demonstrate that /I globin gene sequences are transcribed into a molecule (/l-precursor) that is threefold larger than mature /3 globin mRNA. In contrast, the largest t,ranscript of the u globin genes detected in these experiments is only I.4-fold larger than mature a globin mRNA. The possibility that the 280,000 M,, /I-specific RNA is a processing intermediate generated by cleavage of the 600,000 3, /3-precursor is discussed, along witah alternative hypotheses.

1. Introduction The objective of these and related investigations has been to characterize aspects of struct,ural gene transcription in eukaryotes. The synthesis of globin mRNA innucleated mouse erythroid cells has been exploited as a useful model system, because it permits us to focus on specific mRNAs, rather than total cell mRNA, which consists of thousands of structurally distinct molecules. Certain ambiguities t,hat arise due to the size heterogeneity and sequence complexity of total mRNA are thereby eliminated (see Lewin, 1975). Radioactive globin mRNA nucleotide sequences were detected in pulse-labeled RNA molecules that were larger than steady-state (cytoplasmic) globin mRNA (Curtis & Weissmann, 1976; Ross, 1976; Kwan et al., 1977). The larger, labeled, globin-specific RNA sedimented in a broad peak, with an average s value of 14 to 15 S, and contained a precursor ofglobin mRNA, as determined by pulse-chase experiments (Ross, 1976). A precursor larger t,han 20 S was not observed, either in I I

2

J. ROSS

AND

D. A. KNECHT

cultured mouse erythroleukemia (Curtis & Weissmann, 1976), spleen (Kwan et al., 1977), or fetal liver cells (Ross, 1976). These experiments demonstrated that mouse globin gene nucleotide sequences are transcribed initially into precursors that are two- to threefold larger than cytoplasmic globin mRNA. Since low-resolution sucrose gradients were used in the fetal liver studies, it was not possible to measure precisely the molecular weight of the precursor. It was also impossible to determine if the precursor was processed to 10 S by a single cleavage reaction or by two or more sequential reactions that generated cleavage intermediates, as is the case in ribosomal and transfer RNA processing (see Perry, 1976). The molecular weight of pulse-labeled, globin-specific RNA has been determined by electrophoresis in polyacrylamide gels containing 98% formamide. The data demonstrate two p-globin-specific RNAs (M,, 600,000 and 280,000), both of which are larger than mature /3 globin mRNA (average M,, 215,OOO)t. A single u globin-specific precursor molecule (M,, 280,000) is also detected. Most, if not all, of the CIand /3 precursor molecules contain poly(A) of approximately 145 contiguous adenylate residues.

2. Materials and Methods (a) Cell culture and labeling

and isolation of cellular RNA

Fourteen or 15-day-old mouse embryos were dissected, and liver cells were prepared and cultured as previously described by Ross (1976). A total of approx. *lo’ ~-11s were obtained from each 14-day-old embryonic liver. The mouse fetal liver is an erythroid organ, and more than 85% of its cells are nucleated red blood cell precursors that have reached various stages of maturation. When incubated as primary cultures, these cells synthesize appreciable quantities of adult mouse hemoglobin and globin mRNA (see Ross, 1976). RNA was labeled either with [5, 6-3H]uridine (Amersham/Searle, TRK. 410; 42 to 49 Ci/mmol, 1 mCi/ml) as previously described (Ross, 1976) or with all 4 3H-labsled nucleosides at once. For labeling with all 4 nucleosides, equal volumes (mCi) of [5, Bw3H]uridine, [2, 8-3H]adenosine (New England Nuclear, NET-064; 31 to 33 Ci/mmol, 1 mCi/ml), [5-3H]cytidine (Amerhsam/Searle, TRK.198; 26 to 30 Ci/mmol, 1 mCi/ml), and [ae3H]guanosine (Amersham/Searle, TRK.222; 8 to 19 Ci/mmol, 1 mCi/ml) were combined in a single test tube approx. 18 h prior to us3 and were then lyophilized. Immediately prior to use, the nucleosides were resuspended in a small volume of growth medium, warmed to 37”C, and added to the cultures at a final concn of 200 to 400 &i/ml of each nucleoside. Total cell RNA was extracted with phenol and purified by cesium chloride centrifugacultures were poured into an equal tion as previously described (ROSS, 1976). Briefly, volume of frozen, crushed F12 medium, and the cells were pelleted by brief centrifugation. Whole cells were lysed with a buffer that included 7 M-urea and 2% (w/v) sodium dodecyl sulfate (Holmes & Bonner, 1973). Total cell nucleic acids were extracted with phenol plus chloroform/isoamyl alcohol and were then centrifuged in an isopycnic cesium chloride gradient at 25°C for 16 to 20 h at 32,000 revs/min in the Beckman SW56 or SW60 rotor. The RNA pelleted to the bottom of the tube, and DNA remained in the gradient solution. The RNA pellet was then resuspended and concentrated by precipitation with either trichloroacetic acid or ethanol (see Ross, 1976 for details). In all experiments in which RNA was further analyzed by gel electrophoresis or sucrose gradient centrifugation, RNA pellets from the cesium chloride gradient were precipitated with ethanol. t We refer here to the molecular weight of newly synthesized, mature globin mRNA that is labeled during a 20-min pulse. We also refer in this paper to mature globin mRNA labeled continuously during a 5 h period and to steady-state (unlabeled) globin mRNA, both of which have molecular weights ranging from 185,000 to 200,000. The size difference in these various globin mRNA species is due to time-dependent removal of adenylate residues from the poly(A) (Merkel et al., 1976).

STRUCTURE

OF THE

GLOBIN

mRNA

PHECUHSOHS

::

In some experiments involving polyacrylamide gel electrophoresis, 5 to 20 S RNA was isolated by sucrose gradient oentrifugation. Gradients (15% to 30% sucrose, w/v) were prepared approx. 16 h prior to use by layering the following sucrose solutions, one on top of the next, in SW60 nitrocellulose tubes: 1.0 ml of 30%, then 1.1 ml of 25q,,. then 1.1 ml of 20%, and then 1-O ml of 15%. Sucrose solutions were prepared in TE buffc?r (0.002 M-EDTA, 0.01 M-TrisaHCl, pH 7.5) and were filtered through 0.2 pm filt,rrh (Nalgene). Gradients were stored at 4% until used. RNA was pelleted out of ethanol and resuspended in 0.25 ml of sterile TE buffer, heated to 50°C for 5 min, and then rapidly cooled and layered onto the gradient. Centrifugation was for 16 h at 33,000 revs; min, 4”C, in the SW60 rotor (Beckman). Fractions were collected and their absorbanccx at 260 nm was measured. Appropriate fractions were pooled, NaCl was added to a final concn of 0.1 M, and the RNA was precipitated with 2 vol. 1009& ethanol and st,ortd at -- 20°C until used. Purified mouse reticulocyte 10 S globin mRNA was prepared as described by Ross et al. (1974). It is important to note that this RNA was extracted from a total cell lysatjtb. not from polysomes. Therefore, the complementary DNA prepared from this RNA wits from total cell globin mRNA, not from polysomal globin mRNA.

(b) Electrophoresis

of RNA

Cylindrical polyacrylamide gels containing 98yh (v/v) formamidr and 3.59, or 4.5”,, acrylamide were prepared in non-siliconized glass tubes by the method of Pinder et al. (1974), with the following modifications: formamide (Matheson, Coleman and Bell) wss deionized with Rexyn I-300 (Fisher), as described by Forget et al. (1975). Phosphates buffer (pH 6.8, 0.02 M) was substituted for diethylbarbituric acid. Prior t)o polymerization, gels were overlaid with water instead of 70 “/A formamide, and the water was replaced with phosphate-buffered formamide after polymerization. Gels were stored under phosphate-buffered formamide in the dark at room temperature for no more than 4 de,ys. Phosphate buffer was used in the reservoirs and was not recirculated. RNA was pelleted out of ethanol and dried in a gentle stream of air, and marker RNAs (32P-labeled or unlabeled) that were stored in water were then added. The RNAs w(‘r( redried and were then resuspended in deionized formamide (without phosphate buffei~) containing 10yo (w/v) sucrose (Schwarz-Mann; Ultra-Pure) and 0.030,, (w/v) xylem cyanol. RNA to be analyzed in 0.6-cm diameter gels was resuspended in 0.04 ml; RNA for 1.0 or 1.46-cm diameter gels was resuspended in 0.075 ml. The maximum quantit,ies of RNA applied to 0.6, 1.0, or 1.46.cm diameter gels were 20, 70, and 100 pg, respectively. The RNA solution was heated to 50°C for 5 min and was then layered onto the gel surface* under phosphate-buffered formamide. Gels were run at 50 V for 30 min and for 80 V for the remainder of the run. Unless otherwise noted, gels were removed from the tube am1 stained with methylene blue or were sliced and crushed wit,h an aut,omat,ed gel slic*r,r (Gilson). To elute RNA from the gel, the slices were first crushed into small fragment)s, either by the gel slicer or by pushing a gel piece through a 23-gauge needle on a 3 or 5ml plast’ic syringe. Plastic tubes were used to collect the crushed gel fragments, which stick to glass. Two volumes of urea buffer were added per volume of gel. Urea buffer contains 7 ~:urea (Schwarz-Mann; Ultra-Pure), 2% (w/v) sodium dodecyl sulfate (Matheson, Coleman and Bell; not recrystallized), 0.35 an-N&l, 0.001 M-EDTA, and 0.01 M-Tris.HCl, pH 8.0 (Holmes & Bonner, 1973). The gel fragments were shaken vigorously for 2 h at 45°C’. < Glycerol (60%, v/v) was carefully layered under the suspended gel fragment,s, and t,he material was centrifuged at 2400 revs/min for 15 min at room temperature in the PR-.J centrifuge (IEC) to pellet out the gel fragments. The liquid above the glycerol cushion was removed to a separate tube and was extracted with phenol and chloroform/isoampl alcohol, as previously described (Ross, 1976). RNA to be used in hybridization assays was then concentrated by trichloroacetic acid precipitation (Ross, 1976). Although recoveries of RNA have been somewhat variable from experiment to experiment (46 to 90”,,), the urea elution technique yields greater recoveries, on the average, than methods whirh utilize sodium acetate buffers (e.g., see Forget. et al., 1975).

4

J. ROSS AND

D. A. KNECHT

For those experiments in which the total radioactivity in gel slices was counted directly (Fig. 6), crushed gel fragments were placed into counting vials along with 0.3 ml of water. NCS (2.5 ml) (Amersham/Searle) was added, the vials were shaken vigorously for 2 to 4 min, and 10 ml of toluene/PPO was added. (c) Hybridization assays Assays utilizing [3H]cDNA to detect steady-state (unlabeled) globin mRNA were performed as previously described (Ross, 1976). Annealing reactions utilizing unlabeled cDNA to detect radioactive globin mRNA sequences were also performed as described (Ross, 1976), but the RNAase assay was modified, to account for the [3H]adenylate residues in poly(A). At the end of the annealing reaction, 0.4 ml of RNAase buffer (0.1 M-sodium acetate/acetic acid buffer (pH 4.5), 0.002 M-EDTA, 0.15 M-NaCl) was added to the reaction tubes with or without 4 units of RNAase T, (Calbiochem). All subsequent steps were as described by Ross (1976). Some of the RNA labeled with all four [3H]nucleosides contained radioactive poly(A). Since globin cDNA contains poly(dT) (Ross et al., 1973), it is important to note that the poly(A) portion of this RNA does not form detectable hybrids with poly(dT), when assayed by RNAase T, digestion, as described above (J. Ross, unpublished observations). In addition, neither [3H]poly(A) nor [3H]AMP-labeled RNA from cells that do not synthesize globin mRNA form detectable hybrids with unlabeled globin cDNA (J. Ross, unpublished observations). On the basis of previous data (Ross et al., 1973), we assume that [3H]poly(A) does anneal with the poly(dT), but the poly(A) separates from the poly(dT) at pH 4.5 and forms hydrogen-bonded structures with itself (Rich et al., 1961; Rottman et al., 1974). As a result, most, if not all, of the poly(A) that had been annealed with poly(dT) becomes sensitive to RNAase Ts, under these salt and temperature conditions (see Results). In summary, hybrids between labeled poly(A) and unlabeled globin cDNA are not detected under the RNAase Tz conditions described here. For hybridizations with the phage X Hb DNAs, the A DNA in TE buffer was heated to 95°C for 15 min and was then rapidly chilled. All subsequent steps in these reactions were performed as described above, except that the RNAase Tz digestion was carried out at 3O”C, instead of 45°C. This modification was introduced, because the globin-specific portions of one or both h Hb clones are probably significantly smaller than globin cDNA synthesized in vitro (see page 12). The RNAase reaction is so stringent that the RNA portion of a short DNA-RNA duplex is partially digested at 45%; at 30°C the hybridization specificity remains the same, but the extent of hybridization is greater (unpublished observations). (d) Partial pur&ation of unlabeled ccand B globin mRNAs Initial attempts to separate a from /3 globin mRNA by electrophoresis in formamidecontaining polyacrylamide gels yielded RNAs that were 90 to 95% pure at best. Elution of the partially separated c( and p RNAs followed by re-electrophoresis in separate gels yielded only a moderate increase in purity. This degree of purity was unacceptable for the experiments described here. We assumed that the cross-contamination of the partially separated RNAs was due, at least in part, to the heterogeneous poly(A) lengths in reticulocyte globin mRNA. The poly(A) of mouse globin mRNA ranges from 35 to 150 nucleotides in length (Gorski et al., 1974; Merkel et al., 1976). It seemed likely that the faster-migrating Q globin mRNA was contaminated with /3 globin mRNA containing few adenylate residues; conversely the slower-migrating j3 globin mRNA was contaminated with c( globin mRNA containing many adenylate residues. To decrease this cross-contamination, the crude globin mRNA was first fractionated by the Millipore binding procedure into 2 classes based on poly(A) size (Gorski et al., 1974). RNA that was bound to Millipore filters contained 50 or more adenylate residues in its poly(A); RNA that failed to bind contained 50 or less adenylate residues (Gorski et al., 1974). The percentage of RNA bound to the filter varied from 45 to 65% in various experiments. The Millipore-bound RNA was eluted and concentrated by precipitation with ethanol. This RNA was then electrophoresed in a cylindrical 4.5% polyacrylamide

STRUCTURE

OF THE

GLOBIN

mRNA

.5

PRECURSORS

gel containing 98% formamide; the gel size was either 1.46 cm x 11.5 cm or 1.0 cm 11.5 cm, depending on the quantity of RNA (sse section (b), above), and gels were run at 50 V for 30 min

and then

70 V for an additional

16.5 h. The gels were

stained

and

destained, and 2 bands, separated from each other by 0.6 t,o 0.7 cm, were observed. Thr faster-migrating band was designated cc globin mRNA, the slower-migrating band /? globin mRNA, in accordance with the results of Kazazian et al. (1974). The band-widths were relatively broad (0.2 to 0.4 cm), as compared with RNA standards used in ot.hcr gels, such as the Escherichiu coli 23 S and 16 S rRNAs and the brome mosaic virus 4 12 S RNA. Each band was sliced out, and the RNA was eluted and precipitated with ethanol. A flocculent precipitate formed; this precipitate was due to urea or sodium dodecyl sulf’itte or to linear polyacrylate that had not been extracted by the phenol (Hamlyn 6 Gould, 1975). The RNAs were chromatographed on oligo(dT)-cellulose to remove tht> material that formed the flocculent precipitate, and the adsorbed RNA was precipitated with ethanol. The RNAs were then electrophoresed in 0.6 cm 2: 11.5 cm formamidtlcontaining gels, a,s described above. A single band was observed in racb gel, after staining and destaining. The faster-migrating two-thirds of the c( band and the slower-migrating with t,wo-t,hirds of the fi band were sliced out. The RNA was elut’ed and precipitated c%hanol, and was subsequently pzlletod, resuspended in water, and stored at - 70’(‘. Clobin mRNAs fractionated in this manner wel’e 98 to 99O,A pure (Fig. 3). The recovrritss ~1s1 elol)irl for bot,h RN=Zs were 80,, i.e. 496 of the total globin mHN.4 WHS rrcovcred mRNA, and 4”;, was rccovc%rcd as ,!l glohin mRNA. (0) Base a&y.sis Base analy& was psrform-d by incubating t’he labaletl RNA for 20 min at 45°C with 0.25 pg of RNAase Pl (Yamasa Shoyu Co., Tokyo) in a total vol. of 0.01 ml containing 0.01 M-sodium acetate/acetic acid buffer (pH 5.3), 10e4 M-zinc chloride. The react’ion mix was applied directly to a plastic sheet coated with polyethylene imine cellulose (Baker). Five pg of each of the 4 unlabeled 5’-mononucleotides (PL Riochemicals) wert added at the site of application of the labeled RNA. In order to remove excess salt, the, sheet was immersed in anhydrous methanol for 5 min, dried, and re-immersed in fresh anhydrous methanol for another 5 min with intermittent gentle shaking. Affer drying. the chromatogram was run to 5 cm above thp origin in 1 x;-a&ic acid and was then transferred without drying to 1 M-lithium chloride, in which it was developed tjo 15 cm ;rhove the origin. The spots were identified by ultraviolet, light and were cut out imtl plact,d in counting vials. One ml of 1 N-HCl was added, and, after 20 min at room temptxrat#ure, t,he radioactivit,y was rount,ed in 10 ml of KIA counting solution (Researrh Products Incorporated).

3. Results (a) What is the molecular weight of the globin messenger RNA precursor; do one or more cleavage intermediates exist? Previous studies indicated that pulse-labeled globin mRNA nucleotide sequences sedimented fast,er than steady-state (cytoplasmic) globin mRNA (Curtis & Weissmann. 1976: Ross, 1976; see also Kwan et al., 1977). The sedimentation rates of the globinspecific RNA sequences were compared directly in the same sucrose gradient. This approach ensured that the faster sedimenting, labeled sequences were not aggregates of mature globin mRNA molecules. The larger, globin-specific RNA had the properties expected of a precursor, because it was “chased” to approximately the same size as globin mRNA with actinomycin D and with unlabeled nucleosides (Ross, 1976). In order to obtain more precise estimates of the molecular weight, of the precursor. pulse-labeled fetal mouse liver RNA was fractionated in formamide-containing polyacrylamide gels. This gel system affords high resolution based on molecular weight and is independent’ of base composition (Pinder et al., 1974). It also reduces and may

6

J. ROSS

AND

D. A. KNECHT

the aggregation of nuclear RNA t)hat can occur in some non-denaturing gel systems (Kung, 1974). Total cell RNA from cultured fetal mouse liver cells labeled for ten minutes was prefractionated on a sucrose gradient. The 4 to 20 S RNA from that gradient was electrophoresed in a cylindrical 4.5% gel, and the quantity of unlabeled and labeled globin mRNA nucleotide sequences eluted from gel slices was 32P-labeled RNAs (23 5, M, 1.07 x 106; 18 S, M, determined by hybridization. 7 x 105; and 12 S, M, 2.8 x 105) were co-electrophoresed as internal molecular weight markers. The unlabeled globin mRNA migrated in a single band with a molecular weight of 185,000 (Fig. 1, open circles). In contrast, the pulse-labeled globin mRNA sequences migrated in two bands with molecular weights of 600,000 and 280,000 (Fig. 1, closed circles). The quantity of hybridized RNA in t’he lower molecular weight, peak was CT’WI elimirlate

FIG. 1. Electrophoresis of pulse-labeled and steady-state globin-specific RNA in a formamidecontaining, 4.5% polyacrylamide gel. A total of 3.5 x IO7 fourteen-day-old mouse fetal liver cells were cultured at a concn of 1.4 x 107/ml. After 110 min in culture, [3H]nucleosides, each at a final concn of 400 &i/ml, were added. After 10 min of labeling, cells were harvested, total cell RNA was extracted, and 5 to 20 S RNA was prepared by sucrose gradient centrifugation. This 5 to 20 8 RNA was electrophoresed for 10.5 h in a cylindrical 4.5% polyacrylamide gel (0.6 cm x as 11.5 cm) containing 98% formamide. Marker RNAs laheled with azP were co-electrophoresed internal molecular weight standards. The gel was cut into 2-mm slices, and RNA was extracted from each slice and concentated to 85 ~1 by trichloroacetic acid precipitation. The positions of the 32P markers were determined by Cerenkov counting. To detect the steady-state (unlabeled) glohin mRNA, 2 pl of the concentrated RNA were incubated with [3H]cDNA for 2 h; 2:/, of the cDNA was S, nuclease-resistant in the absence of RNA, and this background was subtracted to determine the percentage of hybridization. The quantity (disints/min) of radioactive globin mRNA nucleotide sequences was determined by incubating samples of the concentrated RNA with excess unlabeled glohin cDNA for 18 h (Ross, 1976). RNAase-resistant radioactivity in the presence and absence of cDNA was determined, and the data are plotted as the total disints/min per fraction in glohin-specific sequences. Recovery of steady-state globin mRNA from this gel was 77%, as determined by hydridization (C,t) analysis of RNA applied to versus that recovered from the gel. Recovery of radioactivity (total disintslmin) was 84%. The arrows indicate the positions of the [32P]RNA markers; 1070k, E. co& 23 S rRNA (Mp, I.07 x 106); 700k, mouse 18 S rRNA (M,, 7 x 105); 280k, RNA component 4 from hrome mosair virus (M,, 2.8 x 105). - - 0 - - 0 - - , Hybridization to 3H-labeled globin rDNA; -@-•-, hybridization to unlabeled glohin cDNA.

STBUCTCRE

OF THE

GLOBIN

mRNA

PRECURSORS

7

approximately fivefold greater than that in the larger molecular weight, peak. This ratio was not, due to unequal recoveries of large versus small RNA, since RNA was recovered with equal efficiencies across these gels (unpublished observations). The shoulder on the heavy side of the 280,000 M,. RNA might represent a separate entit? that. does not, resolve from the major peak; a similar shoulder has been observed in ot,hcr 4.5% gels (Fig. 5 and unpublished data). The data indicate that there are two pulse-labeled, globin-specific RNA fractions containing molecules t,hat are approximat,+ threefold larger and l+fold larger than steady-st,at’c globin mRNA. (b) (km a precursor

larger than 600,000 molecular

weight DP detected?

pulse-labeled globin mRNA nucleotide sequences have been det)ected consistent)lJ in the 10 t,o 16 S regions of gels and sucrose gradients (200,000 t,o 600,000 M,.) (Curt& & Weissmann? 1976; Ross, 1976: Kwan et al., 1977: Fig. 1). Small quantit,ies of radioactive globin mRNA sequences have been observed in single. isolated fractions in t’he higher molecular weight regions of some gradients (Ross, 1976). However, a total of five similar experiments has been performed in ohis laborat,ory, in an effort to detect radioact,ive globin-specific RNA larger than 600,000 M,. In bwo experiments a small quantity of hybridized RNA sedimenting fa&er than 16 S was observed. However. these results were not reproducible, since t’he hybridizable RN,4s were in singk fract,ions and were located in different regions of the gradients in the t,wo experiments. In t’he three other experiment,s, globin-specific RNS larger than 16 8 was not observed (unpublished data). In short, we have never consistently det’ectetl a globit~ analysis ()t’ mRNA precursor larger t’han 600,000 M,. In addit,ion, stoichiometlric p&-chase experiments indicat’ed t(hat most, if not all. of the globin-specific RSX that lvould eventually “chase!’ into globin mRNA was 600.000 M,. or less (Ross, 1976). This result provided additional evidence against a larger precursor. In spite of these negative results, we at,tempted to det’ect a larger precursor by gel analysis. This experiment, was performed for two reasons. (1) To confirm or to rcftitc, by an independent method the conclusions drawn from sucrose gradient analyses. (2) To provide bett,er resolution between 18 S and 28 8 RliA, in order to det,ect potential precursors, 18 S or larger, t,hat were somehow overlooked in that region of the sucrose’ gradients. Fet)al mouse liver cells were incubated for 10 min wit,h [3HJnucleosides. and total cell RNA (riot fractionated on a sucrose gradient) was elect.rophoreseti in it formamide containing 3.5% polyacrylamide gel. after staining and destaining: to determine the positions of the internal molecular weight markers. the gel was slicctl into O&cm pieces. and RNA was extracted and annealed with [3HlcDNA and \vit II unlabeled cDXi4. The radioactive. globin-specific sequences (Fig. 2(b). closed circles) ware in t\vo peaks. both of which were larger than the steady-state. 10 S globin mRNA (Fig. 2(b). open circles). This result confirmed that of Figure 1. Relat,ivel,v large gel slices (0.5 cm) were analpzcd in this experiment, and the marker notations have been placccl within the gel slice in which t,hey appeared, regardless of t,heir position within tlu* slice. These technical details probably account for t)he fact that t,he larger and smaller molecules do not, coincide precisely with the expected.molecular weight,5 of 600,000 and 280,000. This assumption is support’ed by the fact that, these RNAs consistentl> migrated at 600.000 and 280,000 M, in gels fractionat,ed into W%cm slices (Figs 1 ad 5: .I. Ross, accompanying paper). The major result was the absencc~ of pulse-labeh~tl globin mRN-1 sequences in molecules larger than 600,000 X,. Although most, of tlaa

J. ROSS

8

AND

D. A. KNECHT

total radioactivity was in RNA larger than 1 x lo6 M, (Fig. 2(a)), no globin-specific sequences were detected in the larger RNA. This analysis thus confirms the results of most of our previous experiments, and we conclude that we are unable to detect a globin mRNA precursor larger than 600,000 M,. However, due to limitations imposed by the pulse-labeling techniques and by the sensitivity of the assays, these experiments do not disprove beyond doubt the existence of such a precursor (see Discussion).

l-

bl

1700k 34

107Ok

55Ok J

3

2aok 6

,

60

4

0

12 16 Sltce no

20

24

3-J 210

FIG. 2. Eleotrophorssis of pulse-labeled and steady-state globin-specific RNA in a formamidecontaining, 3.5% polyacrylamide gel. A total of 4 x 10’ liver cells from 14-day-old mouse embryos were isolated and cultured at a concn of 2.3 x 10’ per ml. After 110 min in culture, all 4 [3H]nucleosides were added, each at a final concn of 300 &i/ml. After 10 min of labeling, the cells were harvested and total RNA was isolated and ethanol precipitated. The RNA was pelleted out of ethanol and air dried, and unlabeled marker RNA8 (2 pg of each of E. coli 23 S and 10 S and BMV-4 12 X) were added. The RNA was then electrophoresed in a 3.5% polyacrylamide gel (1.46 cm x 11 cm) containing 98% formamide. Electrophoresis wa8 for 20 min at 50 V, followed by 11 h at 80 V. The gel was removed from the tube, stained for 3U min, and destained for 1.5 h. The positions of the markers were noted, and the gel was then cut into 0.5.om slices and RNA was extracted as described in Materials and Methods. 70% of the labeled RNA that wa8 applied to the gel was recovered. Quantitation of steady-state and newly synthesized globin mRNA nucleotide sequences was performed as described in the legend of Fig. 1. 1700k, 1700,000 M, etc. (a) Total aH]RNA recovered from each slice (disintsjmin). Hybridization to globin [3H]cDNA; (b) -O-O-----, --+---a---, total [3H]RNA (d isints/min) that hybridized to unlabeled globin oDNA.

STRUCTIJRE

OF

THE

GLOBIN

mRNA

0

PRECURSORS

(c) Do the 600,000 and 280,000 M, RNAs each con.tain, ccand /3 globin mRNA sequences? The hybridization assays to detect radioactive globin mRNA nucleotide sequences were performed with excess unlabeled DNA complementary to both M and ,8 globin mRNAs. Therefore, the 600,000 and 280,000 .iW, RNAs observed in Figures 1 and 2 could conceivably contain only ct or only /3 globin mRNA sequences or both. It was unlikely that u and /3 sequences co-existled in t,he same RNA precursor, since the u and ,8 genes are on different, chromosomes in the mouse (Russell & McFarland, 1974). ‘1’0 determine if the radioactive, globin-specific RNAs contained sequences corresponding to one or both globin mRNAs, a hybridization competition experiment was performed with partially purified, unlabeled CIor /3 globin mRNA. To determint, the purit,y of these RNAs, a hybridization C,tt analysis was performed, in which either the separated tl or /3 RNA or a 1 : 1 mixture of t,hese RNAs was annealed with [3H.]cDNA prepared from total globin mRNA (Fig. 3). The C,t curve of thr 1 : I mixture of the separated RNAs (Fig. 3(c)) resembled that of t)he unfractionat(etl globin mRNA (Fig. 3(d)), in that both curves spanned two decades and contained no Tn contrast, the curves for t’he individual c(or p mRNAs spannetl apparent, plateau. /

I a

-o\ 10-3

I 10-l

10-2

oI00

10-z

10-l RNA (ng)

RNA (,ul)

I00

IO’

3. Hybridization of partially purified cc and fi globin mRNAs to globin [3H]cDNA. Unlabeled mouse reticulocyte globin mRNA was fractionated by Millipore filtration and gel electrophoresis (Materials and Methods). Increasing quantities of the separated RNAs were incubated for 24 h with globin [3H]cDNA prepared from total unfractionated (a plus /3) globin mRNA. 2% of the input cts/min represented S1 nuclease-resistant material, and this background was subtracted to determine the O/o hybridization. See text for details roncerning the purities of the RNAs, as calculated from these data. (a) Purified OLmRNA; (b) purified fi mRNA; (c) 1 : 1 mixture of purified a and b mRNAs; (d) unfractionated 10 S reticulocyte globin mRNA (a plus p). FIG.

t Abbreviation

used:

C,t,

product

of concentration

and

time.

J. ROSS

10

AND

D. A. KNECHT

at least three decades and contained a plateau or transition at 35 t.o 50% hybridization (Fig. 3(a) and (b)). With sufficient RNA, 80 to 90% of the 13H]cDNA was hybridized. This result indicated that the separated RNAs each contained a mixture of u and /3 globin mRNA, but that a significant degree of purification had been achieved. The purity of each fraction was estimated by determining t,he half-saturation values of the first and second portions of the curves. For the u globin mRNA the half-saturation value of the first portion of the curve (from 0 to 50% hybridization) was 4.5 x 10e3, and that of the second portion (50 to 90% hybridization) was 2 x 10-l. Therefore, the M. preparation contained a 44-fold excess of tc globin mRNA (2 x lo-l)/(4*5 x 10m3), and was thus 98% pure. The half-saturation values for the p globin curve were 3.5~10~~ and 3 to 6 x10-l, corresponding t,o an 86 to l’il-fold excess of /3 globin mRNA of approximately 99% purity. TABLE 1 The presence globin-specijic

of u or p globin mRNA sequences in purified, RNA molecules as determined by competition purified, unlabeled a or /3 glohin mRNA Hybridized

Experiment

RNA comTb)titor

no.

I

II

600,000 280,000

466 143

600,000 280,000

209 354

pulse-labeled, with partially

RNA (disintsjmin) in the presence of 0.3 pg of 0.15 pg each of “/ Inhibition by ~~~~~f fi globin a and /l globin a /l m+P mRNA mRN.4 mRNA ~~ ___-___ 497 d 63 45 9 86 90 x3 55 24 42 62 83 202 207

0 80

15

7

37

42

100 77

93 90

Highly purified, unlabeled D! and j3 globin mRNAs were prepared as described in Materials and Methods and were 98 and 99% pure, respectively (Fig. 3 and text). Unlabeled competitor mRNA was mixed with radioactive 600,000 and 280,000 M, RNA that had been purified from fetal mouse liver cells pulse-labeled for 20 min with [3H]uridine (J. Ross, accompanying paper). The RNAs were added to the hybridization reactions prior to the addition of unlabeled cDNA. The quantities of unlabeled 0: and ,6 globin mRNAs were determined from the hybridization analysis with r3H]cDNA (Fig. 3(a), (b), and (d)). The quantity of unlabeled cDNA was not measured, but was approximately 30 ng per reaction, assuming that the reverse transcriptase reaction used to prepare the cDNA was 15 to 20% efficient (J. Ross, unpublished data). Experiments I and II were performed with radioactive RNAs prepared in 2 separate experiments. The hybridized RNA (disints/min) represents the quantity of RNAase-resistant RNA in the presence of cDNA minus the RNAase-resistant RNA in the absenrc of rDNA.

The radioactive 600,000 and 280,000 M, RNAs were purified (J. Ross, accompanying paper) and incubated with unlabeled globin cDNA in the presence and absence of excess, unlabeled GC, /3 or GC plus jI globin mRNA. If the radioactive RNAs contained sequences corresponding to only one of the globin mRNAs, hybridization should be blocked by the corresponding unlabeled mRNA but not by the other mRNA. This result was in fact observed for the 600,000 M, RNA (Table 1). The a globin mRNA blocked hybridization by less than lo%, while a comparable quantity of /I globin mRNA blocked hybridization by 86 to lOOO/o.This result indicated that little, if any, of the 600,000 M, RNA contained dcglobin mRNA sequences. This RNA will be referred t’o as the j&precursor. In contrast, both tc and /I globin mRNAs blocked

STRIrCTIiRE

OF THE

GLOBCN

mRNA

PRECURSORS

11

hybridization of the 280,000 M, RNA. Therefore, this RNA, which will be referred to as the Q, fl28Ok RNA, contains both u and /3 globin mRNA sequences. Two additional experiments confirmed the conclusion derived from the competit,ion experiment. Globin [3H]cDNA was incubated with increasing quantities of ,&precursor or a, /3 280k RNA, each of which had been highly purified (J. Ross, accompanying paper). The prediction was that an RNA containing only p-specific sequences would not form S, nuclease-resistant hybrids with more than 50% or so of t’hfl When /I-precursor \vas 13H]cDNA prepared from total (a plus /?) globin mRNA. incubated with [3H]cDNA, the maximum extent of hybridization was 30:/b (Fig. 4(a). open circles). In contrast, sufficient quantities of the cc,fi 280k RNA drove 90% of thti r3H]cDNA into nuclease-resistant hybrids (Fig. 4(a), closed circles). This result confirmed the observations that there were little, if an.v. a-specific sequences in the P-precursor preparation, and that the a, /3 280k RNA contained b&h a and p RX--\ sequencrs.

40

l\

\

l \

IO5

l

\O

\ 1

IO’

106

i-

1 108

RNA (from the mdlcoted no. of cells)

(0)

-60

IO’ IO’ IO’ Rekulocyte globm mRNA (P9) (b)

PM:. 4. Hybridization of the 600,000 and 280,000 M, globin-specific RNAs to globin [“H]cUS;\, and 280,000 M, (a, ,G 280k) RNAs were purified as described The 600,000 M, (fl-precur3or) samples of these RP?‘As were incubated with (J. Ross, accompanying paper). Increasing 380 rts/min of globin [3H]cDNA for 70 h. Following S, nuclrase digestion, the reactions WWY~ incubated with alkali to digest the L3H]RNA that was not hvdrolyzed by the nurleesr. Thr, reaction mixtures were then precipitated and counted for 20 min (Materials and Methods). The, quantities of /3-precursor and CC,/3 280k RNA are given as the nunlber of ~(41s uwtl to ~~wlw~~ the RNA included in each reaction. (a) Hybridization of /l-precursor or z, fi 28CJk RNA to globin [3H]~~I>XA. i, --- cj--, p-precursor; -----e---e--, cx:,,Ll280k RNA. (b) H,vbrirlization of purified Ii) S reticuloayte globin mKK.4 to globin [ “H]PI>NA.

The incidental observation that the maximum extent of hybridizat#ion with tlrc* p-precursor was 3076, instead of 50%. is interesting, in view of the fact that thtb ( 3H]cDNA was made from globin mRNA purified from a t’ot’al cell lysate and not, from polysomes (Materials and Methods). Rabbit’ reticulocyt,e polysomal RNA contains approximately 1.5fold more a than /3 globin mRNA (Lodish, 1971), and the postribosomal (supernatant) ribonucleoprotein fraction contains a, but little, if any, /? globin mRNA (Jacobs-Lorena & Baglioni, 1972). Therefore, 13HJcDNA prepared from a tot’al cell (polysomal plus supernatant) lysat,e should contain excess CCcDNA if the two mRNAs are reverse transcribed with equal efficiency and if the a//3 ratio is great,er t’han one in t,he mouse. The 30% hybridization level observed in Figure 4(a) is

J. ROSS AND

12

D. A. KNECHT

thus consistent with the notion that mouse reticulocytes contain excess a globin mRNA. The third experiment to confirm the p-specific sequences in the &precursor was to attempt to hybridize the purified [3H]RNA with cloned bacteriophage h DNA containing mouse a or /3 nucleotide sequences. The clones were obtained in X vectors, as previously described (Blattner et al., 1977). Clones Hbll and Hb4 contained a and @pecific sequences, respectively (0. Smithies, F. Blatt’ner, L. Furlong, H. Faber & J. Ross, unpublished data). Each clone drove 30% of the input [3H]globin cDNA into S, nuclease-resistant hybrids. Therefore, if the C3H]cDNA represented a full-length transcript of a and /3 globin mRNAs in a 1 : 1 ratio, Hbll and Hb4 would contain approximately 60% of the nucleotides in globin mRNA (0.6 x 600 nucleotides = 360 nucleotides). If the [3H]cDNA contained 70% M.and 30% /3-specific sequences, the Hbll and Hb4 clones would contain 257 and 600 nucleotides, respectively. In either case the h Hb clones are of sufficient length to form sbable hybrids with globinspecific RNA. Hbll or Hb4 DNAs were heat-denatured and then incubated in excess with the purified [3H]RNAs. The hybridization dat,a indicate that little, if any, of the a clone DNA (Hbll) anneals to the /?-precursor, but the ,4 clone DNA (Hb4) does anneal (Table 2). In contrast, both a and /3 clone DNAs anneal to the a, /3 280k RNA. The lower extent of hybridization with the a clone might be due to either or both of the following: (1) The globin-specific sequences in the a clone are significantly smaller in size than those in the ,!l clone; (2) the quantity of /l-specific sequences in the a, /3 280k RNA is greater than that of the a-specific sequences. In either case, this result confirms that the 600,000 M,, globin-specific RNA observed in Figure 1 contains /3, but little or no a, globin mRNA sequences. Similar results have been obtained for the 15 S precursor from Friend erythroleukemia cells (Curtis et ab., 1977). (cl) Are the /3-precursor and a, /3 280k molecules polyadenylated?

Although a portion of the pulse-labeled nuclear RNA in different cell types contains poly(A) (Edmonds et al., 1971; Groner & Phillips, 1975), there is no conclusive evidence that all such nuclear molecules are destined to become cytoplasmic mRNA TABLE 2 The presence of a or /I globin mRNA sequences in purified, pulselabeled, globin-speci$c RNA molecules as determined by hybridization to bacteriophage h DNA containing ‘A or /3 globin-specific sequences

RNA

RNAase-resistant No DNA

600,000 280,000

41 56

RNA (disints/min) Hbll DNA (E) 56 289

in the presence of Hb4 DNA (8) 399 1661

The 600,000 and 280,000 Af, RNAs were purified from fetal liver cells labeled with [eH]uridine for 20 min (J. Ross, accompanying paper); 2930 and 6680 (disintslmin) of the 600,000 and 280,000 M, RNAs were used, respectively. Cloned X DNAs containing c( or /3 globin sequences were heat-denatured and incubated with the [3H]RNAs for 16 h. RNAase reactions were performed at 3O”C, instead of 45°C (see Materials and Methods). No hybridization was observed when the rH]RNAs were incubated with X DNA lacking the globin-specific nucleotide sequences (data not shown).

STRUCTURE

OF THE

GLOBIN

mRNA

PRECURSORS

I3

(Levy & McCarthy, 1976; RyBel, 1976; Herman et al., 1976). Kinetic analysis of potential precursor-product relationships is difficult., in part because of the large nucleotide sequence complexity of the RNA and also because some of the poly(A) is degraded wit’hin the nucleus (LaTorre & Perry, 1973; Perry et aE., 1974; Levy B McCarthy: 1976). To determine if the /3-precursor and u, p 280k RNAs contain an adenylate-rich region, total RNA from fetal liver cells labeled for 20 minutes with [3H]nucleosides was chromatographed on oligo(dT)-cellulose. Two RNA fractions, dT-non-adsorbed and dT-adsorbed, were obtained. The total RNA from the dT-adsorbed fraction was analyzed by electrophoresis, as described above. The steady-state globin mRNA migrated at a molecular weight of approximately 185,000 (Fig. 5, open circles). The radioactive globin mRNA sequences were in three peaks, the largest of which had a tnolecular weight of 600,000 (Fig. 5, closed circles). The two smaller RNAs had molecular weights of 280,000 and 215,000. The 215,000 M, RNA is presumed to be fully cleaved globin mRNA. These data demonstrate that at least’ a portion of thfl molecules in the P-precursor and the tl, @280k RNA fractions adsorb to oligo(dT)cellulose.

FIG. 5. Electrophoresis of oligo(dT)-celiulose-adsorbed RNA from pulse-labeled fetal mouse liver cells in a formamide-containing polyacrylamide gel. Hybridization analyses for steady-state and pulse-labeled globin mRNA nucleotide sequences. A total of 1.5 x lo7 fourteen-day-old fetal mouse liver cells were cultured at a concn of 3.5 x 107/ml. After 110 min in culture, [3H]nucleosides, each at a final concn of 400 &i/ml, were added. After 20 min of labeling, cells were harvested, and total cell RNA was isolated and chromatographed on an oligo(dT)-cellulose column. The RNA that adsorbed to the column was eluted, and electrophoresed for 17.5 h in a cylindrical 4.5% concentrated by ethanol precipitation, polyacrylamide gel (0.6 cm x 11.5 cm) containing 98% formamide. [3ZP]RNA standards were co-electrophoreJed, to serve as internal molecular weight markers (550k = 550,000 etc). The gel was sliced into g-mm pieces, and the RNA was eluted, concentrated by trichloroacetic acid precipitation, and hybridized to 3H-labeled and to unlabeled globin cDNA, as described in the Fig. 1 legend. Recoveries were not determined with this gel. The arrows indicate the positions of the internal [32P]RNA markers. - -- o- :j - -, Hybridization to 3H-labeled globin cDNA; ~ a--•--, hybridization to unlabeled globin cD?\TA.

14

J. ROSS

AND

D. A. KNECHT

However, this result does not prove that these RNAs are polyadenylated. Some nuclear RNA molecules contain oligo(A), as well as poly(A) (Nakazato et aZ., 1973; Jacobson et al., 1974), and the oligo(A) binds to dT-cellulose (Lingrel et al., 1974). In order to distinguish between oligo versus poly(A) sequences the molecular weight of the adenylate-rich region of the globin-specific RNAs was determined directly. For these experiments, radioactive /3-precursor, a, fl 280k RNA, and globin mRNA were labeled with all four [3H]nucleosides and purified by a series of chromatographic and electrophoretic steps (J. Ross, accompnaying paper). The /3-precursor and cc,/I 280k RNAs were prepared from cells labeled for 20 minutes; globin mRNA was from cells labeled for five hours. The purity of these fractions was assessed in several ways and was at least 80%. The RNAs were treated with RNAases A and T,, under conditions in which the heteropolymeric portion, but not the poly(A), was hydrolyzed (Ross et al., 1972; see also the legend to Fig. 6). The RNA&se-resistant material was then electrophoresed in formamide-containing polyacrylamide gels. Each RNA was present in a single peak that migrated more slowly than the 4 S RNA marker (Fig. 6). This result demonstrated that each of the RNAs contained poly(A). Since the migration rate of RNA is independent of its base composition in formamide gels (Pinder et al., 1974): the approximate molecular weights of the poly(A)s can be calculated and are 36,000 for the mature globin mRNA (c) and 48,000 for the

E 500 > E z 0 250

8

16

24

8

16

24

8

16

2

Slice no. (a)

(b)

(c)

6. The molecular weights of the adenylate-rich RNA of highly purified /I-precursor, t(, /I 280k RNA, and globin mRNA, as determined by electrophoresis in formamide-containing polyacrylamide gels. The RNAs labeled with all 4 [3H]nucleosides were purified (J. Ross, accompanying paper). The j3-precursor and a, ,6 280k RNAs were prepared from cells labeled for 20 min; the mature globin mRNA was from cells labeled for 6 h. The purities of the RNAs were 78, 100, and Sly:, respectively. Each RNA was incubated with RNAases A and T,, as previously described (Ross et al., 1972). The percentages of RNAase-resistant, trichloroacetic acid-precipitable radioactivity in /?-precursor, c(, p 280k RNA, and globin mRNA were 11, 24, and 19, respectively. After RNAase incubation, the RNAase-resistant material was phenol extracted and ethanol precipitated. One portion of the RNA was hydrolyzed with Pl RNAase, and the nucleotides were chromatographed on PEI-cellulose thin-layer plates (Materials and Methods). More than 96% of the radioactivity from each RNA preparation was in AMP (data not shown). Another portion was electrophoresed for 4.5 h in a 4.5% polyacrylamide gel (0.6 cm x 11.5 cm) containing 98% formamide. Component 4 RNA from brome mosaic virus (M, 23 x lOa, 280k), and E. coli 5 S (M, 3.6 x IO*, 36k) and 4 S (M, 2.5 x 104, 25k) RNAs were co-electrophoresed as unlabeled markers. The gels were stained and destained, to identify the markers, and were then cut into O.&cm slices. The radioactivity was determined by crushing each slice into a counting vial, immersing the gel pieces in NCS solution, and adding 10 ml of toluene-based scintillation fluid (Rosa, 1976). Recoveries of counts from the fi-precursor, a, 6 280k RNA, and globin mRNA gels were 91%, 94%, and 100°/& respectively. (a) ,%precursor; (b) a, f 280k RNA; (c) globin mRNA. FIG.

STRUCTURE

OF

THE

GLOBIN

mRNA

PRECURSORS

I .i

,&precursor (a) and the a, j3 280k RNA (b). These values correspond to 145 adenylat’e residues in the poly(A) of the larger RNAs, 110 in the globin mRNA. The size difference is compatible with a time-dependent shortening of poly(A) in the mature globin mRNA, which was labeled continuously during a five-hour period, as compared with the p-precursor and u, /3 280k RNAs, which were labeled for 20 minutes (Merkel et al.. 1976). These gels were run for a sufficiently short time to retain molecules with as feu as ten adenylate residues. However, there is no evidence for oligo(A) in the &precursor. If t’he fi-precursor contained one poly(A) sequence (145 residues) and one oligo(A) sequence (20 residues), approximately 10 to 15% of the total radioactivity woultl have been detected as oligo(A). However, additional experiments are required to prove conclusively that the /I-precursor lacks oligo(A) (see Discussion). (e) Quantities

of @precursor

ami C-C, p 28Ok RNAs

in fetal mouse liver cells

The hybridization data shown in Figure 4 were used to calculate the quantity of each globin-specific RNA at, steady-state in fetal liver cells (Table 3). There arc approximately 50, 1000, and 60,000 molecules per cell of/3-precursor, CC,/3 280k RNA. and mature globin mRNA, respectively. The quantities of ,&precursor and a, p 280k RNA as determined from the reannealing curves agree quite well with the relative peak areas of these RNAs in Figures 1 and 2. These numbers represent averages. sincc~ TABLE

Quantities

3

of /l-precursor, M, ,!I 280k RNA, and wwhre in Il-clay-old mouse fetul liver cells

Fetal liver RNA

Amount

per cell (PC)

,&Precursor (600,000) a, /9 280k (280,000) Mature globin mRNA

Number cwlrs

17-85 x 10-S 160-800x10-” 2x10-2

globin mRNA

of moleper cell

17-85 340-1700 6 x 104

(200.000) Calculations for the /&precursor and a, /3 280k RNAs am from the data of Fig. 4. Calculations for mature fetal liver globin mRNA are from Ross (1976). The following formula WRS used to determine pg/cell : pg globin mRNA x _Mr of RNA >ell no. value M, of globin mRNA. To calculabe pg per cell of j$precursor, the 15% hybridization levels for the globin RNA standard (25 pg; Fig. 4(b)) and /3-precursor (4.5 x lo6 cells; Fig. 4(a), open circles) were rompared. The answer varied less than 2-fold if other hybridization levels between 8 and 2294 were used. For the c(, ,8 280k RNA the 60% hybridization levels were compared. The molecular weight c*orrection factor is based on the observation that the /3-precursor and OL, /3 280k RNAs each contain a single globin mRNA moiety per molecule (J. Ross, accompanying paper). We aasurnc> that the efficiencies of hybridization of the globin-specific portion of these RNAs to L3H]cDNA are equivalent. The number of molecules per cell wa8 determined with the following formula: g/cell

of the RNA

x 6 x 10z3 -__.~-

M, of the RNA

’

The data for the /t-precursor and a, /l 280k RNAs are presented as ranges. The first number represents the observed value, caloulated as described above. The larger number is related to thr fact that the recovery of globin mRNA from fetal liver cells is 20% with the procedure used to purify the ,%precursor and c(, fl 280k RNAs (J. Ross, accompanying paper). We have not determined the actual recoveries of these RNAs, but, if they are also 20%, then the observed valur~ of RN.4 per cell will be b-fold lower than the amount actually present in the cells. Thus, a &fold range is presented.

16

J. ROSS AND D. A. KNECHT

the fetal liver cell population is heterogeneous. Thus, individual cells will contain different ratios of fl-precursor to globin mRNA, depending on their stage of differentiation. Since approximately 80% of the liver cells are a mixture of basophilic, polychromatophilic, and orthochromic erythroblasts (Marks 6 Rifkind, 1972), u-e assume that the numbers in Table 3 essentially reflect the situation in these cell types.

4. Discussion These data indicate that there are least two size classes of RNA that contain globin mRNA nucleotide sequences and are larger than steady-state globin mRNA (Fig. 1). The largest RNA, designated the &precursor, has a molecular weight of 600,000 and contains the nucleotide sequences of /3 globin mRNA. The smaller RNA, molecular weight 280,000, contains both ccand /3globin mRNA sequences. With longer labeling times, newly synthesized globin mRNA with an average molecular weight of 215,000 is observed (Fig. 5). All three globin-specific RNAs contain poly(A) (Fig. 6). A number of unsuccessful attempts have been made in several laboratories to detect pulse-labeled globin mRNA precursor molecules larger than 600,000 1M, (Fig. 2 and unpublished data; see also Curtis & Weissmann, 1976; Ross, 1976; Kwan et al., 1977). There are two interpretations of this negative result. (1) The 600,000 N, molecule is the primary transcript of the &globin genes; (2) there is a larger precursor with such a rapid turnover rate that it is not detected in these experiments. Putative globin mRNA precursors with molecular weights greater than 1.7 x lo6 were reported in experiments in which only unlabeled (steady-state) globin mRNA nucleotide sequences were assayed (Imaizumi et al., 1973; Ruiz-Carrillo et CL, 1973; Williamson et al., 1973; Spohr et al., 1974). The reason for the variable results is not clear. The hybridization assay for unlabeled globin mRNA is more sensitive than that for radioactive globin mRNA sequences (Ross, 1976). However, it seems unlikely that this differential sensitivity accounts for the failure to detect larger, globin-specific RNA with the pulse-labeling techniques, for the following reason. In several of the nonkinetic experiments with avian reticulocytes, the quantity of unlabeled globin mRNA sequences that sedimented at 16 to 18 S was approximately equivalent to that at 28 S (Ruiz-Carrillo et al., 1973; Spohr et aZ., 1974). If a similar situation existed in the fetal mouse liver cells, i.e. if there were a 28 S (1.7 x lo6 X,) precursor containing the same quantity of globin-specific radioactivity as the 600,000 M, molecule, the 28 S precursor would have been detected. Moreover, stoichiometric data from pulse-chase experiments demonstrated that moat, if not all, of the pulse-labeled RNA that would eventually become mature globin mRNA was 18 S or less (Ross, 1976). In summary, if a larger precursor exists, its turnover rate is so rapid that it does not accumulate in sufficient quantities to be detected. In the experiment shown in Figure 2, for example, the quantity (disints/min) of hybridizable (globin-specific) radioactivity in the peak fraction at 600,000 M, was threefold larger than the background. A still larger precursor that turned over threefold more rapidly than the 600,000 M, RNA would have escaped detection. One approach to this question is to determine if the 5’ terminus of the pur$ed 600,000 M, RNA is triphosphorylated (Schmincke et al., 1976; McGuire et al., 1976; Schibler & Perry, 1976). However, the presence of a 5’-triphosphate group would not unequivocally prove that this RNA is the primary gene transcript, because RNA with a 5’-triphosphate could be a cleavage product derived from the 5’ portion of a still larger RNA.

STRUCTURE

OF THE GLOBIN

mRNA PRECURSOR8

17

The data presented here demonstrate that there are a globin mRNA sequences in the 280,000 but none, or very little, in the 600,000 M, RNA. These results are compatible with the following interpretations. (1) The /3 globin genes are transcribed into a 600,000 2M, precursor that is cleaved once to generate a 280,900 &f, cleavage intermediate, which undergoes a second cleavage to generate a globin mRNA-sized molecule. The a globin genes are transcribed into a 280,009 M, precursor that undergoes a single cleavage reaction. (2) The a globin genes are transcribed into a 600,006 M, precursor that is cleaved far more rapidly than the P-precursor. The cleavage reaction might be so rapid that the quantity of a-specific precursor is below t,he level of detection of the hybridization assays. (3) One of the two /I globin genes is t,ranscribed into a 600,000 M, precursor, the other into a 280,006 M, precursor. The CD1 mice used in these experiments have two distinct /3 globin genetic loci located on chromosome 7 and designated @d major and ,Ed minor. The /3 globin proteins coded for by these genes differ in only nine amino acids (Gilman, 1976). These genes might bc transcribed into precursors of different sizes. If so, the 600,000 M, p-precursor t,ranscribed from one of the /l loci might be cleaved directly t’o a globin mRNA-sized molecule in a single step. The /?-specific sequences in the 280,000 RNA fraction would thus represent a /%precursor transcribed from the other /3 locus. There are two important limitations concerning these experiments that should be noted. (1) As discussed above, a larger precursor with a very rapid turnover rate has not been excluded. (2) We have not provided kinetic proof that the 600,000 and 280,000 M, RNAs are precursors of /? and a globin mRNA, respectively. Since each RNA must be t,reated as an individual entity, it has not been possible to perform kinetic (pulsechase) experiments or to measure precursor--product stoichiometry with the mixed (a plus /3) cDNA probe. These experiments should be feasible with the mouse globin h clones that are now available. With these limitations in mind, we favor the interpretation that the /3 glohin mRNA precursor is approximately threefold larger than the mature /3 mRNA; the a globin precursor is approximately 1*4-fold larger than mature M.globin mRNA. These results are consistent with some experiments that have dealt with total cell mRNA synthesis (Price et al., 1974; Derman et al., 1976). They are also consistent with experiments concerning specific RNA precursors that are only slightly larger to fourfold larger t#han the mature mRNAs themselves (Firtel & Lodish, 1973; Macnaughtou cbab., 1974; McGuire et al., 1976; Craig & Raskas, 1976; Haseltine & Baltimore, 1976). The cytoplasmic mRNAs for silk fibroin and for the protein coded by Balbiani ring :! are approximately the same size as their nuclear count,erpart,s (Lizardi, 1976; Dancbolt,, 1975), but these insect mRNAs are quite large, unlike most mammalian mRNA. An ovalbumin mRNA precursor larger than 28 S was not detected (McKnight & Schimke, 1974) but the existence of a precursor that was only t)wo- to threefold larger hhan ovalbumin mRNA (less than 28 S but greater than 1X S) was not excluded. If it, is valid to generalize from these few specific examples that are now available. XX’ predict that most mRNA precursors,if they exist, are only two- to fourfold larger t,han the mRNAs themselves. The concept of giant precursors that are manyfold larger than cytoplasmic mRNA might not be relevant for most mRNAs. Since polyadenylation occurs after gene transcription, the polyadenylated 600,000 M, RNA (Figs 5 and 6) is presumably derived from a non-polyadenylated molecule. In an effort to detect such a molecule, tota)l cell pulse-labeled RNA that failed to adsorb to oligo(dT)-cellulose was analyzed by gel electrophoresis and hybridization.

18

J. ROSS

AND

D. A. KNECHT

The quantity of non-absorbed, 600,000 M,, globin-specific RNA represented, at most, 10 to 20% of the comparable 600,000 M, RNA recovered from the adsorbed fraction (unpublished observations from three separate experiments). Since 5 to 10% of pulselabeled globin mRNA (M,. 215,000) also fails to bind to oligo(dT)-cellulose (unpublished observations), it is likely that the binding reaction is not completely efficient. If so, then the precursor molecules in the non-adsorbed fraction might be polyadenylated as well. The results suggest that the precursor is polyadenylated very soon after it is transcribed, as is the case with silk fibroin mRNA (Lizardi, 1976). On the other hand, Spohr et al. (1976) observed that most of the unlabeled globin mRNA nucleotide sequences in duck erythroblast nuclei failed to bind to poly(U)-Sepharose, and they suggested that polyadenylstion occurred in the later stages of globin mRNA formation. Perry et al. (1974) have also suggested that adenylation is a late step in mRNA biosynthesis. The experiment,s reported here indicate, but do not prove, that the globin mRNA precursors lack oligo(A) (Fig. 6). In the first place, the RNAase-resistant RNA was precipitated with ethanol after digestion, and, although oligo(A) is reported to be ethanol-precipitable (Kinniburgh & Martin, 1976; A. Kinniburgh, personal communication), appropriate controls to this point were not performed in this laboratory, In the second place, the gels were stained and destained after electrophoresis, and oligo(A) might leak out during these procedures. Although 90% or more of the radioactivity was recovered from each gel (see Fig. 6 legend), the recovery from the precursor gel was only 91%. Perhaps the other 9% represents oligo(A) that was lost during extraction. For reasons cited above, the half-lives of the 600,000 and 280,000 M, molecules were not measued directly. Nevertheless, their turnover rates can be estimated from their steady-state levels (Table 3). We shall assume that the half-life of globin mRNA in fetal liver cells is 17 hours, as it is in adult mouse spleen cells (Bastos et al., 1977), and that the a, p 280k RNA contains approximately equivalent amounts of a and /3 molecules that are cleaved stoichiometrically (without wastage) into mature globin mRNA. If these assumptions are valid, then the half-life of the a, fl 280k RNA is derived from the following equation: Molecules per cell of a, /3 280k RNA x half-life of globin mRNA ’ Molecules per cell of globin mRNA 1000 = x 17 hours = 17 minutes. 60,000 If the /?-precursor is cleaved without wastage into a 280,000 M, intermediate, p-precursor half-life is derived from the following: Molecules per cell of /%precursor Molecules per cell of 280k p-intermediate

the

x half-life of 280k p-intermediate

50 x 17 minutes = 102 seconds. = 1000 x 0.5 If the transcription rate of the /3 globin genes is comparable to that of the ribosomal genes in HeLa cells (approx. 120 nucleotides/s; Greenberg & Penman, 1966), then the heteropolymeric portion of the p-precursor is transcribed in approximately 14 seconds (1720 nucleotides/l20). Thus, the half-life of the /l-precursor might be only sevenfold longer than its synthesis time.

STRUCTURE

OF

THE

GLOBIN

mRNA

PRECURSORS

19

We are grateful to Mrs Karen Everlith for expert technical assistance, to Mr David Wright for his help in establishing conditions for gel electrophoresis, and to Mr Larry Beach for setting up the PEI-cellulose chromatography. We thank Drs R. Dasgupt,a and P. Kaesberg for their generous supply of both unlabeled and 32P-labeled bromo mosaic virus RNAs. Drs F. Blattner, H. Faber, L. Furlong, and 0. Smithies kind]) provided us with h Hb DNAs. Purified reverse transcriptase was supplied by the Offiw of l’rogram Resources and Logistics, Viral Cancer Program, Viral Oncology, Division of Cancer Cause and Prevention, National Cancer Institute, Bethesda, Md 20014. \Vc t,hank Drs *J. Beard and M. A. Chirigos for their help in obtaining t,he reverse transcript,as