Cell, Vol. 11, 641-650,

July

1977, Copyright

0 1977 by MIT

Globin RNA Precursor Molecules: Biosynthesis and Processing in Erythroid Cells Roberto N. Bastos and Haim Aviv Department of Virology Weizmann Institute of Science Rehovot, Israel

Summary Hybridization of labeled RNA with excess amounts of DNA complementary to globin mRNA, in conjunction with a pulse-chase technique, were used to investigate the biosynthetic pathway of globin mRNA in erythroid cells. Three species of molecules sharing common sequences with globin mRNA were detected in the nuclei of these cells, two of which are larger than the cytoplasmic globin mRNA. One species was approximately 7 times larger than globin mRNA (“27S”), and the other (“15s”) was only about twice the size of cytoplasmic globin mRNA. The largest species lacked poly(A) sequences, while the others contained poly(A). After chase, the large RNA species gradually disappeared (tliz = 5 min), while the cytoplasmic 10s species accumulated. From these results, a model is proposed describing the biosynthetic pathway of globin RNA transcription; an early transcription product is the large molecule “279 (-5000 nucleotides long) which is then cleaved into a smaller species “1%” (-1500 nucleotides). This intermediate precursor is then clipped, presumably at the 5’ end, and finally converted to the exported “10s” molecule (-750 nucleotides) which accumulates in the cytoplasm. Introduction In contrast to procaryotes, where transcription of the genes and translation of mRNA are linked, in eucaryotes these processes are physically separated by the nuclear membrane. It has been proposed (Scherrer and Marcaud, 1968; Scherrer, 1973; Darnell, Jelinek and Molloy, 1973) that a multistep process is involved in the information flow from the eucaryotic genome to the site of translation in the cytoplasm. The experimental data supporting this proposal were problematic, however, because of the tendency of nuclear RNA to aggregate (Bramwell, 1972; Macnaughton, Freeman and Bishop, 1974), and because there are difficulties in using pulse-chase techniques in many eucaryotic cells (Singer and Penman, 1973; Murphy and Attardi, 1973). Yet these pulse-chase techniques are extremely valuable in drawing a relationship between precursors and products. Recently, sensitive techniques were developed for quantitation of minute amounts of specific mRNAs (Ross et al., 1972; Kacian et al., 1972; Verma et al., 1972), and the use

of denaturing conditions to separate nuclear RNA classes (Acheson et al., 1971; Imaizumi, Diggelman and Scherrer, 1973; Macnaughton et al., 1974; Spohr, Dettori and Manzari, 1976b) has produced rather convincing data that large molecules of RNA-containing globin sequences exist in nuclei of erythroid cells, but the size of these molecules is controversial. Moreover, the generality of this phenomenon was questioned because nuclear RNA containing ovalbumin mRNA sequences seems to be as large as the cytoplasmic ovalbumin mRNA (McKnight and Schimke, 1974). In all these experiments, however, the level of accumulated RNA sequences was quantitated at steady state conditions by hybridization with labeled complementary DNA (cDNA). Since the nuclear RNA species which are the putative precursors for cytoplasmic RNA may have a short half-life, a more desirable approach would be to follow the fate of labeled RNA sequences using hybridization conditions of excess cDNA. This approach was taken recently by Ross (1976) and by Curtis and Weissmann (1976). In these experiments, labeled globin RNA molecules were observed in nuclei from erythroid cells which are about 2 or 3 times larger than the mature cytoplasmic RNA. We have examined nuclear RNA (nRNA) from DMSO-treated Friend erythroleukemic cells and RNA extracted from spleen cells of anemic mice by sedimenting pulse-labeled RNA from these cells under denaturing conditions and hybridizing the various resulting fractions to excess globin cDNA bound to cellulose (Levy and Aviv, 1976). This technique is sensitive enough to allow us to detect very low amounts of labeled hybridizing RNA speciesas low as 1O-5 of the input. Furthermore, we examined the change in the size distribution of the globin mRNA species after an effective chase of the labeled nucleotide. Our results indicate that in the nuclei of erythroid cells, there exist three species of RNA molecules containing globin sequences, two of which are larger than the mature form of globin mRNA. In very short pulses (5 min), only the two larger species were observed, while following chase, the gradual appearance of a 10s molecule is clearly seen. The largest mRNA molecule hybridizing to globin cDNA is devoid of poly(A), while the two smaller molecules are associated with poly(A). A model is proposed which describes the processing pattern of the precursor globin RNA molecules into the cytoplasmic mature mRNA. Results Three Species of Nuclear Globin RNA and Their Size The major problem encountered in sizing nuclear RNA has been the tendency of this fraction of RNA

Cell

642

larger species of about 27s. The proportions between these three RNA species changed with the labeling time. Similarly, three RNA species containing globin sequences could be detected in nuclei from erythroid spleen cells (not shown). In contrast to the nuclear RNA, the cytoplasmic RNA contained only one species of labeled globin RNA, the size of which was about 10s (not shown). The molecular weight of the largest molecule containing globin sequences (27s) was estimated to be about I .5 x IO6 daltons, which corresponds to about 4500-5000 nucleotides, the 15s species was about 0.5 x lo6 daltons (-1600 nucleotides) and the smallest species of 10s (or perhaps as big as 11s) was about 0.25 x IO6 daltons (see Experimental Procedures). Since the purity of cDNA may not be absolute, however, the hybrid formed between the large RNA molecules and cDNA could result from a hybrid between labeled RNA sequences and nonglobin cDNA contaminating the preparation of cDNA. This possibility could be excluded by an experiment where isolated labeled RNA species and cDNA cellulose would be made to compete for unlabeled purified globin RNA. If the hybrid formed between large molecular weight labeled RNA sequences and cDNA were competed by a larger quantity of purified unlabeled globin RNA than needed to compete out a hybrid between labeled purified globin RNA sequences and cDNA, this would indicate that the large labeled RNA molecules are not globin se-

to aggregate (Acheson et al., 1971; Bramwell, 1972). A variety of denaturing conditions were introduced to minimize this problem (Macnaughton et al., 1974; Spohr et al., 1976a). We examined the melting of a hybrid between globin RNA and cDNA in different concentrations of formamide to choose the proper denaturing conditions. Even in 85% formamide at a very low ionic strength, the midpoint of the melting curve was at 19°C (not shown). To minimize aggregation, we therefore decided to preheat the RNA samples to 60°C in 85% formamide before loading onto the gradient and then to run them at 25°C. Under these conditions, aggregation of RNA species is improbable. However, aggregates of RNA due to unusually high GC regions are hard to exclude. We have not experimented with the conditions described by Strauss, Kelly and Sinsheimer (1968) and Spohr et al. (1976a). We then labeled DMSO-treated Friend cells with 3H-uridine. Cells were lysed and nuclei were separated. The labeled RNA was extracted and fractionated on a sucrose gradient under denaturing conditions as described. Every fraction was then analyzed for labeled globin RNA sequences by hybridization to globin cDNA cellulose. Three distinct species of labeled RNA containing globin sequences were detected in the nuclei of these cells (Figures lA, lC, IE and 1G): a small molecule of about 10s (corresponding in size to the newly synthesized cytoplasmic globin RNA), a second species with an S value of about I.5 and a

0

0 5

0

15

XI

25

5

IO

15

20

25

Fraction Figure

1. Nuclear-Labeled

RNA-Containing

Globin

Sequences

5

IO

15

20

25

5

IO

15

20

25

number

and Their

Poly(A)

Content

Cells were grown for 4 days in the presence of DMSO and labeled in 10 ml of medium with 2 mCi 3H-uridine. RNA was split into two parts. One half wasdirectly fractionated on sucrose-formamide gradients as described. The other half was first fractionated on oligo(dT)-cellulose prior to separation on a sucrose gradient. Gradient fractions were hybridized to excess globin cDNA cellulose as described, RNA-eluted and counted. (A) Nuclear RNA after 5 min of labeling; (C) nuclear RNA after 5 min of labeling and IO min of chase; (E) nuclear RNA after 15 min of labeling: (G) nuclear RNA after 30 min of labeling. (B, D, F and H) are the corresponding poly(A)-containing RNA.

Metabolism 643

of Globin

mRNA

quences. On the other hand, if all three RNA species (lOS, 15S, 27s) were removed from the hybrid with cDNA by equivalent amounts of unlabeled globin RNA, this would indicate that the RNA sequences hybridized to globin cDNA are indeed globin RNA. Figure 2 shows that the latter case was correct. Addition of Poly(A) to Nuclear Globin RNA In contrast to cytoplasmic mRNA which is mainly comprised of molecules containing poly(A) at their 3’ end, most heteronuclear RNA is poly(A)-deficient (Greenberg and Perry, 1972; Derman and Darnell, 1974). This can indicate either that poly(A) is never added to most of the nuclear RNA molecules, or that primary precursor molecules exist in the nucleus without poly(A), and poly(A) is added later during processing (see Darnell et al., 1973; Lewin, 1975). We have determined whether or not the precursor forms of globin RNA contain poly(A) first by

IOOB I

I

I I

z 80‘L n 2 cn 60.-c .-c if! z 40E E h

20-

I

I

I I

001

0.05 0.1 0.5 Competitor globin RNA added@g)

Figure 2. Competiton Hybridization cies Containing Globin Sequences

of Labeled Nuclear RNA Spewith Pure Globin mRNA

The three nuclear RNA species were purified on 85% formamide sucrose gradients after hybridization of nRNA samples to globin cDNA and elution (without RNAase treatment) as described, and pooled. The RNA species were divided into five equal aliquots (each totaling approximately 300-400 cpm of globin-specific counts) and added to a hybridization mixture containing 40 ~1 of a 1:9 suspension of cDNA cellulose. Varying amounts of competitor pure mouse reticulocyte 9S globin RNA were added to every sample, and hybridization was carried out overnight at 40°C. Elution and counting of hybridized RNA were performed as described. (O-O) 275 nANA; (A-A) 15s nRNA; (O---Cl) 10s nRNA; (O-O) “51-labeled 10s RNA from reticulocytes (27,000 w-N.

separating nuclear RNA into poly(A)-containing and poly(A)-deficient molecules using chromatography on oligo(dT)-cellulose (Aviv and Leder, 1972). Both fractions were separated on formamide sucrose gradients and then assayed for labeled globin RNA sequences as described previously. It is clear that only the smaller RNA molecules (155 and 10s) which contain globin sequences contain poly(A) (Figure I), while the larger precursor molecule (27s) is devoid of poly(A). This is consistent with a mechanism in which the largest precursor molecule for globin RNA is first cleaved and poly(A) is added later during the processing period. Do the Large Globin RNA Molecules Contain Tandemly Oriented Globin Sequences? The three species of globin-containing RNA species were purified first by hybridization to globin cDNA cellulose and then fractionated on formamide-sucrose gradients (Figure 3A); the purity of each fraction a (IOS), b (15s) and c (27s) was assayed by its capacity to compete under hybridization conditions with purified ‘251-labeled 10s globin mRNA for excess amounts of cDNA (Figure 3B). It is clear that fraction a, the 10s nuclear globin RNA, is practically pure (>90%), because the competition curve is identical to that of pure 10s globin mRNA which was purified from mouse reticulocytes (Figure 38, open circles and triangles). On the other hand, larger amounts of fractions b and c were required to compete with the hybridization of ‘Z51-labeled 10s globin RNA. 50% competition was reached when 0.055 pg of fraction a were used, while 0.16 pg and 0.35 pg of b and c, respectively, were required for the same result. Thus fraction b contains only 35% globin sequences, and fraction c contains only 16% globin sequences. These values correspond nicely with their relative size (see above), thereby indicating that each of these three fractions is (practically) pure. The gradient fractions containing labeled globin RNA sequences were also assayed by analytical hybridization to excess amounts of cDNA cellulose followed by RNAase treatment. It appears that while 90% of fraction a was protected against RNAase digestion, only 30% and 15% of fractions b and c, respectively, were protected by excess cDNA (Figure 3A, open circles). From the molecular weights of these three species and their degrees of purity, it seems rather improbable to us that fractions b and c contain information for more than one copy of globin RNA in the same molecule (for example, two copies of CY- or two copies of p-globin mRNA). Additional experiments are required, however, to exclude this possibility positively. Processing An effective

of Nuclear procedure

RNA by which

a precursor-prod-

Cell

644

Fraction Figure

3. Quantitation

of Globin

Competitor

number Sequences

in Globin

Nuclear

globin RNA added

(pg)

RNA Species

(A) Total nRNA extracted from nuclei of 5 x IO’-lo9 cells labeled for 15 min with 3H-uridine under the standard conditions (Experimental Procedures) was hybridized to 100 pi of a I:3 suspension of globin cDNA cellulose (Experimental Procedures). After washing the cellulose as described (except that RNAase treatment was omitted), the hybridized RNA was eluted by washing the cellulose with 85% formamide at 60°C. The nRNA resulting from this treatment was fractionated into fractions of 250 pl each, and a 25 ~1 aliquot of each fraction was counted to determine total counts in each of the three peaks of globin mRNA (O---O). The fractions were rehybridized to globin cDNA cellulose as described, and RNA was eluted after the standard washing procedure, including the RNAase treatment (O-O). (6) Isolated globin nRNA species were purified as described in (A) (without RNAase treatment). The three peaks were each pooled and used to compete with a constant amount of ‘*Wabeled pure reticulocyte globin RNA (spec. act. IO6 cpm/+g) for 40 ~1 of a I:9 suspension of globin cDNA cellulose. (O-O) RNA from peak a; (O---U) RNA from peak b; (O-O) RNA from peak c; (A-A) 10s globin RNA purified from reticulocytes.

uct relationship between the larger molecules and the smaller ones can be established is by following the fate of the labeled sequences after chase with unlabeled nucleosides. Fortunately, labeled RNA in Friend cells can be efficiently chased within 1 min by unlabeled uridine and cytidine (not shown). Cells were labeled for 5 or 15 min and then chased for different periods of time. As seen in Figure 4, the proportions of the three nuclear globin RNA species have changed with time of labeling (Figures 4A and 4E). Moreover, during the chase period, a dramatic shift of the relative proportions between the three nuclear species of globin RNA was observed. Both species of large RNA disappeared, and the 10s RNA became the predominant species in the nucleus. The half-life of disappearance of the largest precursor (27s) was approximately 5 min (Figure 5). The pattern describing the change in the different globin RNA species is as follows. While the label in the 27s species disappeared immediately after chase, the 15s species went through a peak IO min after chase. On the other hand, labeled cytoplasmic globin RNA ap-

peared only during the chase period, and the pattern of accumulation was linear with the time of chase. The 10s RNA species in the nucleus seems to be confined to the nucleus and not a contaminant of cytoplasmic RNA, since the pattern of its behavior is different from that of the cytoplasmic RNA. The label in the 10s nuclear RNA also went through a peak very similar to the 15s RNA species. We also show below (Figure 7) that in very short chase periods, the 10s nuclear RNA species was labeled before any label appeared in the cytoplasm. It is therefore improbable that nuclear RNA was contaminated with cytoplasmic 10s RNA. For a finer analysis of the half-life of the 27s RNA species, actinomycin D was used to stop abruptly further incorporation of uridine. Cells were harvested rapidly 1, 3, 5 and 10 min after exposure to the drug, and the labeled globin RNA species was analyzed as described previously. Chase with unlabeled nucleosides is also shown in comparison (Figure 6). The picture that emerges from these experiments is very similar to that which was described earlier. The half-life of the 27s RNA species

Metabolism

of Globin

mRNA

645

S value 0

200

20

30

IO

20

30

I

0

Time Figure Chase

5

IO

15

20

25

Fraction Figure 4. Pulse-Labeling Min of Labeling)

and

Chase

5

IO

15

20

25

number of Nuclear

RNA

(5 and

15

Cells grown for 4 days in the presence of DMSO were labeled under the described conditions in 10 ml medium with 2 mCi of $Huridine. Labeling of RNA was performed for either 5 or 15 min, after which the cells were quickly collected by centrifugation in a clinical centrifuge and resuspended in the same volume of medium containing 20 mM uridine and 15 mM cytidine. Samples were taken at the indicated times, nuclei were isolated and nRNA was prepared. The RNA pellet was resuspended in 200 ~1 of 65% formamide containing 1 mM EDTA and 10 mM Tris-HCI (pH 7.6), and fractionated on sucrose-formamide gradients. Each of the gradient fractions was hybridized to excess globin cDNA cellulose. The hybridized RNA waseluted and counted asdescribed. Left column: (A-D) 5 min labeling time followed by chase; A = zero time chase; B = IO min after chase; C = 20 min after chase; D = 40 min after chase. Right column: (E-H) 15 min labeling time followed by chase; E = zero time chase; F = 10 min after chase: G = 20 min after chase: H = 40 min after chase.

was about 4-5 min (Figure 5). The proportion between the different species is shown in Figure 7. Here again, the disappearance of the 27s RNA species was followed by the appearance of the smaller nuclear RNA species. Surprisingly, however, in the cells treated with this concentration of actinomycin (10 pg/ml), the labeled globin RNA did not appear in the cytoplasm even after 10 min of chase (Figure 7). We do not know whether this concentration of

5. Decay

of

Labeled

after Nuclear

chase ( min) RNA

Species

Following

RNA amounts used to calculate the decay of globin RNA sequences were taken from Figures 4 and 6. Decay of labeled 27s RNA counts after chase by cold nucleoside (O+) or by actinomycin D (O-O). Decay of total labeled globin nuclear RNA counts following cold chase after labeling for 5 min (A-A) or after labeling for 15 min (A-A). Decay of total labeled nuclear RNA counts following cold chase (O--U).

actinomycin blocks completely the transport of globin nuclear RNA or only slows down this process. It is also clear from this experiment that the 10s globin nuclear RNA is a bona fide nuclear species and not merely a cytoplasmic contamination, since it is present not only before label appears in cytoplasm, but even in the complete absence of cytoplasmic 10s RNA. The half-life values of the 15s -+ 10s and of the 10s (nuclear) -+ 10s (cytoplasmic) transitions can be calculated mathematically to fit the experimental data (equations not shown). The best fit obtained was when all three processes occurred with a t,,2 around 5 min. It is interesting to compare the conservation of labeled globin and nonglobin sequences. While most pulse-labeled total RNA does not appear in the cytoplasm after chase (not shown), practically all pulse-labeled globin RNA sequences seem to be conserved in the cytoplasm (Figure 7, insert). In experiments where much longer chase was used, however, a small fraction (-30%) of pulse-labeled globin RNA was not recovered during the chase period (not shown).

Cell 646

Discussion From the experiments presented above, we conclude that the biosynthesis of globin RNA in erythroid cells is a multistep process, in which an early event is the synthesis of a large precursor molecule (“27s”) lacking poly(A), which is subsequently cleaved to an intermediary species (“15s”) containing poly(A) and is then processed to the “1OS” species which is exported to the cytoplasm of these cells. A schematic presentation of this process is shown in Figure 8.

Although the data presented here strongly support this model, they are based mainly on kinetic studies, which can also be interpreted in a different way. For example, the 27s RNA can be processed simultaneously to 1% RNA as well as to 10s nuclear RNA. It should also be stressed that in the experiments described here, no distinction was made between the behavior of (Y- and p- globin mRNA. We do not know whether they are identical or not. In view of the experiments of Orkin, Swan and Leder (1975) which indicate that the two globin mRNAs accumulate in Friend cells at a different rate after DMSO treatment, it is conceivable that the kinetics of precursor processing of these two messengers could also be different, In addition, one should keep in mind that the oc,p mRNA representation in the cDNA used is also not known. The common experience that p mRNA is transcribed much more effectively than O( mRNA may affect these measurements. Several additional features of this process await clarification. For example, is the “27s” species the primary transcription product of the globin genes or already a processed molecule? What is the topological relationship between the products and the presumed precursors-that is, are the mRNA sequences located near the 3’ end of the precursor molecules (Herman, Williams and Penman, 1976) or not? Is poly(A) added before or after the cleavage of the “275” species to the “15s” species? Large RNA molecules sharing common sequences with globin mRNA were also detected by other investigators (see Introduction). The size of these molecules ranged from about twice the size of mature cytoplasmic 10s RNA to very large molecules. In most of these studies, however, RNA was detected by hybridization of labeled cDNA to excess amounts of RNA, which measures the steady state concentrations of globin RNA sequences, Figure 6. Pulse Actinomycin)

Fraction

number

Labeling

and Chase

(Unlabeled

Nucleosides

and

Friend cells (strain 745, clone 39) were grown for 4 days in the presence of 1.7% DMSO and labeled for 5 min with 2 mCi of 3H-uridine in 10 ml of medium at approximately 10’ cells per ml. In one experiment, the nucleoside was chased by transfer to medium containing unlabeled nucleosides, while in another, this was done by addition of actinomycin D to a final concentration of 10 pg/ml (Experimental Procedures). Samples were taken at the indicated times, and nuclear RNA was prepared and analyzed on sucroseformamide gradients as described. Globin RNA sequences on the various resulting fractions were determined by hybridization to excess globin cDNA cellulose. Left column: chase with cold nucleoside. (A) zero time of chase: (B) 1 min chase time (according to incorporation data, equivalent to 6 min labeling); (C) 3 min chase time (actual chase time = 2 min); (D) 5 min chase time (actual chase time = 4 min); (E) 10 min chase time (actual chase time = 9 min). Right column: actinomycin D chase. (F) zero time of chase; (G) 1 min chase time; (H) 3 min chase time; (I) 5 min chase time; (J) 10 min chase time.

Metabolism 647

of Globin

mRNA

1

I ’

B

E

nut

z200m” n

\

Time

*

s

o--

IOOm

\ ‘-2 6 ‘O 15s

I

o

I f

x

I ’ I

I

2

(min)

6 Time

IO (mid

Ioscyto

\

E I M

w f

I

I

I

2

4

6

8

P 0

I

--I

-1-1

IO

2

I

4

6

8

IO

Time after chase (mid Figure

7. Proportions

Values content RNA. Insert:

for the plot of amounts of nuclear RNAs was determined. (A-A) 275 nuclear labeled

globin

of Globin

Nuclear

RNA following

and Cytoplasmic

chase

RNA Species

in this figure RNA: (O-O)

(O-O)

Following

Chase

were taken from Figure 6. Cytoplasmic RNA was prepared, 1% nuclear RNA: (0-D) 10s nuclear RNA; (O-O)

in nuclei

and

rather than newly synthesized molecules. Recent results by Ross (1976) in fetal liver, by Curtis and Weissmann (1976) in Friend cells, and by Kwan, Wood and Lingrel (1977) have demonstrated the presence of labeled globin RNA containing molecules which are 2-3 times larger than the mature cytoplasmic molecules under hybridization conditions using excess cDNA. In contrast to our results, these investigators did not see molecules as large as the “275” RNA that we have detected. The reasons for this discrepancy are not clear. A possibility that cannot be overlooked in this context is that the relative proportions and halflives of the three globin nuclear RNA species may be tightly coupled to the metabolic state of the cell. Thus cells which are growing exponentially may process the primary transcript at a rate too fast to be observed, while cells under quasi-stationary conditions will not be under so stringent a pressure for fast processing and export of specific mRNA sequences.

(04)

in cytoplasm,

and (A-A)

total

globin

and its globin RNA cytoplasmic globin RNA counts.

A striking feature of the interrelationship of the three globin nuclear RNA species observed here is the discreteness of processing. In contrast to the expectation where the transition between precursor and product would be a gradual smooth one (presumably via exonucleolytic RNAase activity on the primary transcript), these results indicate that a rather sharp transition occurs between the three RNA species. The enzymes possibly involved and the signals directing such a process are unknown. If large precursor RNA molecules are indeed the general pattern for the biosynthetic pathway of many eucaryotic cellular messenger RNAs, one may wonder what is the biological significance for this difference in the metabolism of mRNA between procaryotes and eucaryotes. Several years ago the “cascade regulation hypothesis” was proposed by Scherrer (Sch.errer and Marcaud, 1968; Scherrer, 1973). In essence, Scherrer suggested that the “selection effort” required to choose one cistron out of a large number of cistrons is much less in a

Cell 648

Transcription

and

processing

of

globin

mRNA

globln gene I

r

1

/

Nuclear Globin RNA Precursors

1 primary tronscr~pt (2 I

5bPPY3,

5”3,

275 RNA (-5000~~)

Poly (A) addiiion

15s RNA (-1600nuc)

5’

1

cleavage of 5’ end

IOSRNA I-750nuC)

-(AIn

Cyioplasmic

cells. However, the presence of some large globin RNA molecules in noninduced Friend cells was observed by other investigators (P. Harrison, A. Sckoulchi and D. Shafritiz, personal communication). It is interesting that some globin RNA sequences were detected in nuclei of early chicken embryos before their appearance in the cytoplasm (Chan, 1976). An attractive way to circumvent the “waste” involved in the existence of RNA precursor molecules larger than the mature mRNA would be if the nonglobin sequences in the molecule contained sequences for other mRNA species or information for other purposes such as regulation of transcription in the nucleus (see Darnell et al., 1973; Davidson and Britten, 1973). Whatever the biological significance of these large molecules, it would be interesting to purify and characterize the sequences adjacent to the globin mRNA sequences. These studies are in progress in our laboratory.

globin RNA

Experimental IOS RNA (750nucl

-(AIn Figure essing

8. A Proposed of Globin RNA

Model

for

the

Post-Transcriptional

Proc-

In this model, we postulate that the transcription of the globin gene starts in a region to the left of the portion which is conserved as cytoplasmic mRNA and extends into a region to the right of it. The conserved region (dashed) is not more than 15% of the transcription product. We would speculate that the 15s RNA is generated by the removal of the 3’ end of the 27s RNA, thus exposing a sequence which serves as a signal for the addition of poly(A). Subsequently, the 5’ end of the molecule is removed, perhaps exposing a sequence recognized by the capping enzymes. The final product is a 10s RNA molecule which is capped by poly(A) at the 3’ end and by 7-methyl-G at the 5’ end. This “double capping” perhaps renders stability to eucaryotic mRNAs. We would speculate that processing of precursor molecules is a sophisticated trimming process, evolved to correct the relative imprecision of the eukaryotic transcription process, rather than having a positive regulatory role. Several essential features predicted from this model can be tested experimentally, and these experiments are in progress.

multistep selection process (“cascade”) as compared to a unistep selection process. Although this theory is indeed appealing, several questions should be raised. First, the theory would predict that in tissues where one particular gene product does not appear in the cytoplasm (for example, hemoglobin in nonerythroid tissues), the proper selection process of that specific cistron did not take place. Thus nonprocessed molecules should be present in the nuclei of at least some of these tissues. Experimental data on this point are rather controversial. Groudine et al. (1974) could not detect globin RNA sequences in nonerythroid tissues of ducks, but Humphries, Windass and Williamson (1976) found globin RNA sequences in many mouse

Procedures

Cell Culture Friend cells (strain 745, clone 39) were cultured as described previously for 4.5 days in the presence of 1.7% DMSO (Fluka) (Aviv et al., 1976). Labeling and Chase of 3H-Uridine 17-24 hr before addition of labeled uridine, the medium was exchanged for fresh medium containing 1.7% DMSO. Just before labeling, cells were collected by centrifugation and resuspended in fresh medium at a concentration of approximately IO’cells per ml. For each 10 ml of culture, l-2 mCi of 5.6 3H-uridine (45 Ci/ mmole; Amersham) were added, and labeling proceeded for the indicated periods of time. When the nucleotide labeling was to be chased, two alternative methods were used. With the first, the cells were rapidly chilled, collected by centrifugation (1 min, 2000 x g) and immediately resuspended in warm medium containing 20 mM uridine and 15 mM cytidine. Alternatively, actinomycin D was added to the culture to a final concentration of 10 pg/ml, a procedure that immediately halted incorporation of label into RNA (not shown). Extraction of Nuclear RNA (nRNA) Aliquots of the cell culture were removed into a 5 fold excess of medium at 0°C. The cells were harvested by centrifugation as described above, washed once with PBS containing 2% BSA and lysed by addition of 2 ml of a buffer containing 5 mM NaCI, 3 mM MgCI,, 10 mM Hepes (pH 7.8), 8.55% sucrose and 0.5% Triton X100, and gentle pipetting with a Pasteur pipette. Nuclei were collected by centrifugation and washed once with the same buffer without Triton. The nuclear pellet was lysed by suspension in 1 ml of a solution containing 0.5 M NaCI, 20 mM MgClz and 80 pg of DNAase A (RNAase-free; Worthington). The suspension was incubated at room temperature for 5 min with frequent shaking, after which 30 mM EDTA were added along with 0.5% SDS (Penman, 1966). The lysate was repeatedly extracted with 1 vol each of phenol and 24:l chloroform:isoamyl alcohol until no interphase remained, and the resulting RNA was precipitated with 2 vol of ethanol and 0.1 M sodium acetate (pH 5.0) overnight at -20°C and collected by centrifugation at 20,000 g for 30 min. Oligo(dT)-Cellulose Chromatography Separation of nRNA samples into poly(A)-containing

and poly(A)-

Metabolism 649

deficient tography

of Globin

mRNA

material was performed as previously described

by oligo(dT)-cellulose (Aviv et al., 1976).

chroma-

Fractionation of RNA in Sucrose Gradients The nRNA samples were suspended in a small volume (usually 100 ~1) of 85% formamide containing IO mM Tris (pH 7.6) and 1 mM EDTA, and heated at 60°C for 5 min before loading on the top of a 2-10% sucrose gradient in 85% formamide + 1 mM EDTA buffered with IO mM Tris (pH 7.6) (Macnaughton et al., 1974). The gradient was spun at either 49K rpm in a Beckman SW50.1 rotor for 9 hr or at 49K rpm in a SW65 rotor for 6 hr, both at 25°C. The gradient was fractionated into 250 ~1 fractions by pushing from the bottom. Hybridization of RNA Samples to Globin cDNA Hybridization to globin cDNA cellulose (prepared according to the method of Venetianer and Leder: 1974) was performed as previously described (Levy and Aviv, 1976). cDNA cellulose was used as a 13 suspension (volume-wise relative to volume of packed cellulose) in water. Each reaction contained 100 ~1 of thissuspension, which is equivalent to 0.7-I 5 fig of cDNA. Hybridized RNA was determined by washing the cellulose twice with IO ml of 2 x SSC, followed by treatment with 50 Pg/ml RNAase A (Worthington) for 30 min at room temperature. The RNAase was removed, and the cellulose was washed twice with 10 ml of a buffer containing 1 mM EDTA, 0.1% SDS and 10 mM Tris (pH 7.6). The hybridized RNA was eluted by three successive washes with 1 ml of 0.1 N NaOH and neutralized with HCI, and radioactivity was counted in Insta-Gel (Packard). Each sample was counted for at least 10 min. Estimation of Molecular Weights The molecular weights of the three species of RNA were estimated from the S values using rRNA from the same cells as internal standards (l&S = 0.7 x IO6 daltons and 28s = 1.7 x lo6 daltons) (Loening, 1968; Spohr et al., 1976a). A plot of log MW against log S was used for calibration (Spirin, 1963). Two sets of S values were used for rRNA-one of 18s and 28S, and the other of 145 and 21s (Macnaughton et al., 1974).

It is our privilege to acknowledge Drs. M. Edelman, M. Revel, C. Prives and 2. Volloch for many helpful suggestions and interesting discussions. We thank J. Ross and P. Curtis for making available to us manuscripts of their work prior to publication. We thank Drs. J. Beard and J. Gruber for purified AMV transcriptase. R. B. is a recipient of a Lady Tata Memorial Trust fellowship. This work was partially supported by a contract from the US National Cancer Institute. December

17, 1976;

revised

March

29, 1977

References Acheson, N. H., Buetti, E., Scherrer, K. and Weil, R. (1971). Transcription of the polyoma virus genome: synthesis and clearage of giant late polyoma-specific RNA. Proc. Nat. Acad. Sci. USA 68, 2231-2235. Aviv, H. and Leder, P. (1972). Purification of biologically active globin messenger RNA by chromatography on oligothymidylic acid-cellulose. Proc. Nat. Acad. Sci. USA 69, 1408-1412.

Friend cells and detection J. Mol. Biol. 106, 1061-

Darnell, J. E., Jelinek, W. R. and Molloy, G. (1973). Biogenesis of mRNA: genetic regulation in mammalian cells. Science 787, 12151221. Davidson, tion, and 565-613.

E. H. and Britten, R. J. (1973). Organization, transcripregulation in the animal genome. Quart. Rev. Biol. 48,

Derman, E. and Darnell, J. E. (1974). Relationship of chain transcription to poly(A) addition and processing of hnRNA in HeLa cells. Cell 3, 255-264. Greenberg, J. R. and Perry, R. P. (1972). Relative occurrence polyadenylic acid sequences in messenger and heterogeneous nuclear RNA of L cells as determined by poly(U) hydroxylapatite chromatography. J. Mol. Biol. 72, 91-98.

of

Groudine, M., Holtzer, H., Scherrer, K. and Therwath, A. (1974). Lineage-dependent transcription of globin genes. Cell 3, 243-247. Herman, R. C., Williams, J. G. and Penman, S. (1976). Message and non-message sequences adjacent to poly(A) in steady state heterogeneous nuclear RNA of HeLa cells. Cell 7, 429-437. Humphries, S., Windass, globin gene expression Cell 7, 267-277.

J. and Williamson, R. (1976). in erythroid and non-erythroid

Mouse tissues.

Imaizumi, T., Diggelmann, H. and Scherrer, K. (1973). Demonstration of globin messenger sequences in giant nuclear precursors of messenger RNA of avian etythroblasts. Proc. Nat. Acad. Sci. USA 70, 1122-1126. Kacian, D. L., Spiegelman, S., Bank, A., Terada, M., Metafora, J., Dow, L. and Marks, P. A. (1972). In vitro synthesis of DNA components of human genes for globins. Nature New Biol. 235, 167-169. Kwan, S., Wood, T. G. and Lingrel, J. B. (1977). Purification of a putative precursor of globin messenger RNA from mouse nucleated erythroid cells. Proc. Nat. Acad. Sci. USA 74, 178-182. Levy, S. and Aviv, H. (1976). senger RNA by hybridization covalently bound to cellulose.

Acknowledgments

Received

mRNA from dimethylsulfoxide-induced of a putative globin mRNA precursor. 1075.

Quantitation of labeled globin with excess complementary Biochemistry 75, 1844-1847.

mesDNA

Lewin, B. (1975). Units of transcription and translation: the relationship between heterogeneous nuclear RNA and messenger RNA. Cell 4, 1 l-20. Loening, V. E. (1968). Molecular weights of ribosomal relation to evolution. J. Mol. Biol. 38, 355-365. McKnight, G. S. and Schimke, RNA: evidence that the initial size as polysomal ovalbumin USA 71, 4327-4331.

RNA

in

R. T. (1974). Ovalbumin messenger product of transcription is the same messenger. Proc. Nat. Acad. Sci.

Macnaughton, M., Freeman, K. B. and Bishop, J. 0. (1974). A precursor to hemoglobin mRNA in nuclei of immature duck red blood cells. Cell 7, 117-125. Murphy, W. and Attradi, G. (1973). Stability of cytoplasmic messenger RNA in HeLa cells. Proc. Nat. Acad. Sci. USA70, 115-119. Orkin, S. H., Swan, D. and Leder, P. (1975). Differential expression of (Y and p globin genes during differentiation of cultured erythroleukemic cells. J. Biol. Chem. 250, 8753-8760. Penman, S. (1966). RNA metabolism Mol. Biol. 77, 117-130.

in the HeLa

cell nucleus.

J.

Aviv, H., Voloch, Z., Bastos, R. and Levy, S. (1976). Biosynthesis and stability of globin mRNA in cultured erythroleukemic Friend cells. Cell 8, 495-503.

Ross, J. (1976). A precursor Biol. 106, 403-420.

Bramwell, M. E. (1972). A comparison of cell electrophoresis density gradient centrifugation of heterogeneous nuclear Biochim. Biophys. Acta 281, 329-337.

Ross, J., Aviv, H., Scolnik, E. and Leder, P. (1972). Globin messenger-RNA induction during erythroid differentiation of cultured leukemia cells. Proc. Nat. Acad. Sci. USA 69, 3620-3623.

Chan, L. N. L. (1976). Transport of globin cytoplasm in differentiating embryonic 261, 158-159. Curtis,

P. J. and

Weissmann,

C. (1976).

and RNA.

mRNA from nucleus into red blood cells. Nature Purification

of globin

of globin

messenger

RNA. J. Mol.

Scherrer, K. (1973). Control of gene expression in animal ceils. In Control of Gene Expression, A. Kohn and A. Shatkay, eds. (New York: Plenum Press), p. 169. Scherrer,

K. and

Marcaud,

L. (1968).

Messenger

RNA

in avian

Cell 650

erythroblasts at the transcriptional problem of regulation in animal 212. Singer, R.H. cells: kinetics

and translational levels and the cells. J. Cell. Physiol. 72, 161-

and Penman, S. (1973). Messenger RNA in HeLa of formation and decay. J. Mol. Biol. 78, 321-334.

Spirin, A. S. (1963). Some problems ular structure of ribonucleic acids. 345.

concerning Prog. Nucl.

the macromolecAcid Res. 7, 301-

Spohr, G., Mirault, M. E., Imaizumi, T. and Scherrer, K. (1976a). Molecular-weight determination of animal-cell RNA by electrophoresis in formamide under fully denaturing conditions on exponential polyacrylamide cells. Eur. J. Biochem. 62, 313-332. Spohr, G., Dettori, G. and Manzari, V. (1976b). Globin mRNA sequences in polyadenylated and nonpolyadenylated nuclear precursor-messenger RNA from avian erythroblasts. Cell 8, 505-512. Strauss, F. M., Kelly, R. B. and Sinsheimer, R. L. (1966). Denaturation of RNA with dimethyl sulfoxide. Biopolymers 6, 793-807. Venetianer, phase-bound lian genes.

P. and Leder, P. (1974). Enzymatic synthesis DNA sequences corresponding to specific Proc. Nat. Acad. Sci. USA 71, 3892-3895.

of solid mamma-

Verma, I. M., Temple, F. G., Fan, H. and Baltimore, D. (1972). In vitro synthesis of DNA complementary to rabbit reticulocyte 10s RNA. Nature New Biol. 235, 163-167.