Polyribosome Metabolism in Escherichia coli II. ^{Formation and Lifetime of Messenger RNA Molecules, Ribosomal Subunit Couples and Polyribosomes}

Abstract

During a continuous label with [³H]uracil of fragile Escherichia coli cells in exponential growth, all or very nearly all the hybridizable mRNA of the cells can be observed in polyribosomes during zonal sedimentation in sucrose gradients. In contrast, all the newly formed rRNA appears to be free of polyribosomes until it is part of complete ribosomal subunits. The separation of the two major classes of newly formed RNA permits the direct observation of the growth with time of mRNA molecules, of ribosomal subunits and of polyribosomes.

A complete chain of ribosomal RNA forms in one to two minutes; completion of either a 30 s or 50 s ribosome requires a minimum of about five minutes more. The specific activities of free ribosomal subunits and those that are present in polyribosomes increase at identical rates, consistent with very rapid exchange of the two classes.

Before any new subunits are finished, the ³H-labeled RNA in polyribosomes is the cellular mRNA, which can therefore be extracted and observed separated from ribosomal RNA. Assuming that their sedimentation rates in sucrose gradients are related to molecular weights as are those of ribosomal rRNA and that no degradation occurs dining extraction, most of the molecules of mRNA are large enough to code for only one or two protein chains of 30,000 molecular weight.

The kinetics of entry of all the new mRNA into polyribosomes is in agreement with the view that a polyribosome forms as the corresponding messenger RNA is synthesized, over a period of at least a minute. 3% of the total cellular RNA in these cultures (doubling time 120 minutes) is mRNA, with an average chemical lifetime of 11 to 12·5 minutes.

1. Introduction

For an understanding of the metabolism of polyribosomes, it is necessary to know the origin and stability of messenger RNA and ribosomal subunits, as well as how they interact. At present there are only fragmentary data, because of the instability of messenger RNA molecules in cells and extracts. Recently, however, several techniques yielded relatively intact preparations of polyribosomes from actively growing cells (Mangiarotti & Schlessinger, 1966). The preparations displayed polyribosomes and ribosomal subunits, but not the “70 s” ribosomes regularly observed in earlier studies. The 70 s ribosomes were shown to have been, most likely, an artifact produced by partial degradation of polyribosomes. With their elimination, and the resultant simplification of the classes of ribosomes into free subunits (30 and 50 s) and polyribosomes, a detailed analysis of polyribosome metabolism became more feasible.

The analysis reported here measures the rate at which components of polyribosomes reach constant specific activity during continuous labeling with [³H]uracil. This technique has been previously used for some studies of the biosynthesis of ribosomal subunits (McCarthy, Britten & Roberts, 1962; Kono & Osawa, 1964). To a culture in which the stable classes of RNA have been uniformly labeled with one precursor, such as [¹⁴C]uracil introduced in a pulse several generations earlier, a large amount of a second labeled precursor, [³H]uracil, is added. As shown by earlier workers (McCarthy & Britten, 1962), the tritiated precursor is incorporated from within seconds after its addition until it is almost all used up. The analysis of the newly labeled RNA is possible because at all times the ratio of new label ³H to old label ¹⁴C in any fraction is a measure of the ratio of new to old RNA. The kinetics of labeling of various fractions of RNA should be rather different. For example, McCarthy et al. (1962) have shown that after a chain of ribosomal RNA is completed, some minutes are required for it to associate with protein to form a complete ribosomal subunit. Therefore, precursors of ribosomal subunits will be labeled first, then completed subunits, and finally mature subunits in polyribosomes.

mRNA will take up label somewhat like a ribosomal precursor. As the molecules of mRNA break down, they are replaced at an equivalent rate. The rate of replacement can therefore provide a direct determination of the lifetime of mRNA. When all the mRNA is labeled, the amount of ³H label in the fraction will increase only slowly, in proportion to net growth.

The various RNA fractions are here separated in sucrose gradients and their capacity to hybridize with DNA is measured. The amount, lifetime and size distribution of mRNA are thereby determined.

2. Materials and Methods

(a). Cultures and extraction of polyribosomes and RNA

Cultures of Escherichia coli K 12 were grown exponentially in fragile form in high salt concentration at 37°C, lysed with buffers containing 0·5% sodium deoxycholate, and the ribosomes displayed in sucrose gradients as previously described (Mangiarotti & Schlessinger, 1966). To ensure quantitative release of cellular RNA and intact polyribosomes with these strains, the lysing medium contained 10 mM-Tris (pH 7·5), 5 mM-MgSO₄, and 0·3 mM-chloramphenicol. In order to avoid precipitation of magnesium deoxycholate, 40 mM-NaCl was added, and the medium was prepared at 0°C, with the deoxycholate added last.

To extract RNA from gradient fractions, 20% sodium dodecyl sulfate (Matheson Chemical Co.) was added to a final concentration of 1%. An equal volume of water-saturated phenol was then mixed with the sample for 3 min. The aqueous and phenol phases were separated by centrifugation at 1500 g for 2 min, and the aqueous phase removed and subjected to two more phenol extractions. Phenol extractions were carried out at 60°C for samples to be analyzed by hybridization assays. All other phenol extractions were carried out at room temperature. The final aqueous layer was dialyzed for 2 hr against the appropriate buffer: at 4°C against 10 mM-Tris (pH 7·3)—1% phenol, for samples to be analyzed by hybridization assays; at 4°C against 10 mM-sodium acetate (pH 5·1) for samples to be analyzed in zonal sedimentation.

(b). Sucrose gradient centrifugation of RNA

RNA preparations were analyzed in 15 to 30% sucrose gradients containing 10 mM-sodium acetate (pH 5·1) and 50 mM-NaCl (Asano, 1965). Although phenol extraction was customarily carried out, identical sedimentation profiles were obtained by an alternative procedure in which samples were instead brought to 0·5% sodium dodecyl sulfate, dialyzed against 10 mM-Tris (pH 7·3), 50 mM-NaCl, 0·5% dodecyl sulfate at 20°C, and then analyzed in gradients containing the same buffer.

(c). Preparation of rRNA

In order to obtain preparations of rRNA^† free of tRNA and fragments of mRNA, ribosomes were prepared by the procedure of Williams (1967, manuscript in preparation), including several precipitations with ammonium sulfate. Such ribosomes are totally dependent on added mRNA for amino acid incorporation, yield preparations of RNA that do not stimulate incorporation of amino acids in pre-incubated extracts, and are devoid of pulse-labeled RNA and amino acyl acceptor RNA (Williams, 1967, and unpublished data). RNA was isolated from the purified ribosomes by adding 20% sodium dodecyl sulfate to a final concentration of 0·5%, then extracting with phenol (method of Hill & Echols, 1966).

(d). Hybridization of RNA and DNA

The reagents and conditions used for hybridization studies of labeled RNA were essentially those of Hill & Echols (1966). E. coli DNA was purchased from General Biochemicals, Chagrin Falls, Ohio; 2 mg/ml. were dissolved in 60 mM-KCl, 5 mM-potassium citrate (pH 6·0), and freed of traces of contaminating protein and ribonuclease by phenol extraction. After dialysis against 50 mM-KCl–5mM-potassium citrate, the solution was diluted to 0·4 mg/ml. DNA, 10 mM-KCl and 1 mM-potassium citrate. To denature the DNA, sample tubes were heated 10 min in boiling water and then rapidly chilled in ice water.

Each hybridization assay combined 75 μg of the denatured DNA and 0·2 to 0·5 μg labeled RNA in 0·3 ml. of 500 mM-KCl, 10 mM-Tris (pH 7·6), 10 mM-potassium citrate (pH 6·0) (hybridization buffer). After the addition of 30 μl. of water-saturated phenol, the sample was incubated at 60°C for 16 hr to permit hybrid formation to take place. The degree of hybrid formation was determined on nitrocellulose membrane filters according to Nygaard & Hall (1963), except that before filtration, the sample, diluted to 5 ml. with the hybridization buffer, was treated for 20 min at room temperature with 10 μg pancreatic ribonuclease/ml. (Worthington Biochemical Corp., Freehold, N.J.).

Unlabeled rRNA dilutes the labeled rRNA, and can thereby block it from hybridization. In controls, essentially no labeled rRNA was left hybridized when the reactions were carried out in the presence of 10 μg of unlabeled rRNA. Similarly, when a mixture of labeled rRNA and mRNA was hybridized with DNA in a series of tubes containing increasing amounts of unlabeled rRNA, 10 μg of rRNA was found to block the hybridization of rRNA maximally. In order to distinguish the hybridization of mRNA from that of rRNA in a labeled sample, the reactions were therefore carried out in the absence and in the presence of 10 μg unlabeled rRNA, to give the total hybridization and hybridization of mRNA, respectively.

The RNA content of the cultures was determined in two ways: (1) by the orcinol reaction with AMP as a standard (Dische, Ehrlich, Munoz & Von Sallman, 1953); (2) by using a uracil auxotroph and measuring the uptake of [¹⁴C]uracil of known specific activity. Both estimates agree on an RNA content per cell of 0·068 μμg ± 10%, the equivalent of about 20,000 30 s and 20,000 50 s ribosomes per cell (assuming that 80% of the RNA content of the cell is in ribosomes, each containing a molecular weight of RNA of 1·6 × 10⁶ per pair of 30 and 50 s subunits).

3. Results

(a). Rates of subunit synthesis and exchange of subunit couples

Cell lysates display polyribosomes and free subunits in sucrose gradients (Mangiarotti & Schlessinger, 1966). The question arises: Are the free subunits and those in polyribosomes identical? All subunits might be in free exchange (Fig. 1(a)), so that one 30 s and one 50 s subunit combine, join a polyribosome, participate in the formation of protein, then separate from the polyribosome and dissociate to re-enter the free pool each time a protein chain is finished. However, several alternatives can be visualized. For example, the subunits could be in partial exchange (Fig. 1(b)), so that a single couple composed of one 30 s plus one 50 s subunit (a “70 s monomer”) might traverse several mRNA molecules in turn before dissociating to free subunits. Alternatively, the free 30 and 50 s subunits might not exchange at all with those active in protein synthesis. The free subunits might instead be immature and not yet ready to function on mRNA molecules (Fig. 1(c)); or a sizeable percentage of ribosomes may be permanently blocked from further function after working for some time (Zipser, 1963). In the latter event, part or all of the pool of free 30 and 50 s subunits could represent a dead end (“spent ribosomes”; Fig. 1(d)).

Fig. 1.

Possible models for the relation of free ribosomal subunits and those in polyribosomes (see text).

(a) Free exchange of subunits; (b) partial exchange of subunits; (c) free subunits as immature precursors; (d) free subunits as incompetent end products.

A preliminary experiment suggested that essentially all the subunits—in and out of polyribosomes—exchange. A brief pulse of [³H]uracil was given to a growing culture and followed by a long chase with unlabeled uracil. If the free subunits were merely precursors of those in polyribosomes and, once active, never return to the pool of subunits (Fig. 1(c)), all the radioactive label should move into the fraction of polyribosomes, and the free subunits (“precursors”) should become progressively less radioactive. On the other hand, if they are largely spent or exhausted, all of the radioactive label should finally be trapped in the pool of subunits, and the polyribosomes should become progressively less radioactive. Finally, if there is exchange of free subunits and those in polyribosomes, the specific radioactivity of both should reach the same value (Fig. 1(a) or (b)).

In such experiments, it was found that in less than 60 minutes, the RNA in all regions of an analytical sucrose gradient reached the same specific radioactivity, and exchange of free and polyribosomal subunits must therefore occur. However, the experiments could not distinguish whether exchange takes place very rapidly (Fig. 1(a)) or very slowly (Fig. 1(b)). Similarly, they could not exclude that nearly complete subunits spend an appreciable time (though less than 60 minutes) as immature precursors (Fig. 1(c)). Therefore, in order to determine directly the time required to make functional subunits and their rate of exchange with subunits in polyribosomes, more refined experiments using a continuous label with [³H]uracil (Figs 2 and 3) rather than a pulse were carried out.

Fig. 2.

Rate of appearance of labeled RNA in free and polyribosomal subunits.

(a) Initial gradients. The RNA of a culture of fragile cells was first labeled with a pulse of 0·6 μg [¹⁴C]uracil/ml. (0.33 μc) followed by two generations of growth. At 1·5 × 10⁸ cells/ml. [³H]uracil was added. At intervals, 10-ml. samples were harvested, centrifuged and lysed. Equivalent portions of samples were then run on sucrose gradients at 4°C at 56,000 g for 10 hr, to sediment most of the polyribosomes to the bottom of the gradient and display free 30 and 50 s. The bottom of the gradient is at left.

In order to increase the counts measured at early times without using exorbitant quantities of [³H]uracil to permit long-term labeling, the experiment was run in two flasks. To one (top), 0·15 μg (10 μc)/ml. was added for one set of points, from 30 sec to 6 min; to the other (bottom) 0·75 μg (10 μc)/ml. was added for points from 6 to 30 min. The incorporation of [³H]uracil in the two parts of the experiment was then normalized by multiplying the amount of ³H in each sample from 6 to 30 min by the ratio of specific activities of exogenous uracil in the two series of points (0·75/0·15=5). The two patterns at 6 min thereby become superimposable.

A sample of the large amount of radioactive RNA in polyribosomes at the bottom of the gradient tube was measured for each sample. For 125,000 cts of ¹⁴C the relative amounts of [³H] uracil were: at 0·5 min, 72,000; at 1 min, 142,000; 1·5 min, 215,000; 2 min, 300,000; 4 min, 570,000; 6 min, 860,000; 9 min, 1,290,000; 12 min, 1,720,000; 15 min, 2,150,000; 20 min, 2,870,000; 30 min, 4,380,000. The heavy line indicates the amounts and positions of stable, ¹⁴C-labeled RNA; the other lines show the pattern of ³H-labeled RNA at the times indicated. Gradient fractions 1 to 14 have been plotted with twice their actual number of counts, to increase the resolution of the curves.

(b) Derived subunits. The resuspended polyribosomes of the gradients of Fig. 2(a) were dialyzed 4 hr at 4°C against 0·1 mM-Mg²⁺ to release the derived subunits. Subsequent suorose gradient analyses in 0.1 mM-Mg²⁺ are depicted that display the freed subunits. Heavy line, position of ¹⁴C-labeled RNA; other lines, ³H-labeled RNA at each indicated time.

(c) Free 30 s subunits. 30 s peak fractions of the gradients of Fig. 2(a) were dialyzed 4 hr at 4°C against 0·1 mM-Mg²⁺ to remove sucrose, then re-run in gradients 14 hr at 56,000 g to permit estimation of the ratio ³H/¹⁴C in each subunit. ¹⁴C- and ³H-labeled RNA indicated as in Figs 2(a) and (b).

(d) Free 50 s subunits. 50 s peak fractions of the gradients of Fig.2(a) dialyzed and resedimented in sucrose gradients for 14 hr. Analysis as in Fig. 2(b).

Fig. 3.

Entry of [³H]uraoil into total RNA and polyribosomes.

(a) Entry of [³H]uracil into total RNA (–○– –○–) and polyribosomes (—Δ—Δ—) during a continuous label (from the gradients of Fig. 2(a)). ³H cts/min are expressed as ratio to total ¹⁴C cts/min in RNA. The dashed line is calculated by dividing each experimental point on the solid line by exp k₁t to correct for exponential growth during the experiment (where k₁ is the growth constant, In 2/mass doubling time, here equal to 1/173).

(b) [³H]uracil in total RNA of polyribosomes (—●—●—) and in 30 and 50 s subunits obtained from polyribosomes (—○—○—) (from data of Fig. 2). ³H cts/min are expressed as ratio to ¹⁴C cts/min in polyribosomes (rather than ¹⁴C cts/min in total RNA as in Fig. 3(a)).

Inset : Linear labeling of total polyribosomal RNA at very early times, shown on ctn expanded time scale.

(c) Difference between specific activity (³H/¹⁴C) of polyribosomes and of ribosomal subunits obtained from polyribosomes. The difference between the curves of Fig. 3(b) (—●—●—), identified as mRNA. A small increment attributable to exponential growth during labeling (– – – – –) has been subtracted from the curve for total mRNA.

In order to distinguish pre-existing from newly formed RNA, labeling was carried out with different tracers. Stable RNA was uniformly labeled with [¹⁴C]uracil, and newly formed RNA was then continuously labeled with [³H]uracil. From 0 to 30 minutes after the addition of [³H]uracil, portions of the culture were harvested, lysed, and centrifuged through sucrose gradients for 10 hours to sediment the polyribosomes to the bottom of the tube and display the ribosomal subunits and nascent RNA (Fig. 2(a)). Figure 3(a) illustrates that, from the gradient analyses, about 30% of the total incorporated label enters polyribosomes for at least 12 minutes (±2% in six experiments).

To measure the rate at which new RNA enters ribosomal subunits in and out of polyribosomes, the subunits were separately re-isolated: the polyribosome fraction, as well as samples from the 30 and 50 s ribosomal peaks of the initial gradients, were dialyzed against 0·1 mM-Mg²⁺ (to release mRNA; Schlessinger & Gros, 1963), yielding purified polyribosomal and free subunits, respectively. The free and polyribosomal subunits were then separately re-run in sucrose gradients (Fig. 2(b), (c) and (d)). Repurification of the free subunits was required to separate them adequately from the large amounts of label in ribosomal precursors evident in Fig. 2(a). By choosing the appropriate samples from the initial gradients (about fraction 16 for the 50 s and fraction 24 for the 30 s subunits), the contamination by precursors was sharply reduced. Also, in the subsequent gradients in low magnesium ion concentration (Fig 2(c) and (d)), the remaining precursor particles sediment more slowly relative to subunits than they do in 0·01 M-Mg²⁺, so that sufficient resolution was attained. In some trials (not shown), brief treatment of duplicate samples with 0·05 μg RNase/ml. was used before dialysis of the fractions, in order to destroy precursor particles selectively (Sypherd, 1965); no change in the values for the ratio of ³H to ^l4C in complete subunits was observed.

Nearly identical results have been obtained in six such experiments (including one in which the order of labeling was reversed, i.e. cells were prelabeled with [³H]-uracil, and [¹⁴C]uracil was used as the second label). The ratio of ³H to ¹⁴C at the peak of a repurified subunit fraction was taken as the measure of new to old RNA, and Table 1 and the graphs of Fig. 3 were derived. It appears (Fig. 3) that complete new 30 and 50 s subunits take a minimum of about seven minutes to form under these growth conditions. Until six minutes of labeling had elapsed, the gradient analyses showed no detectable new label moving with complete subunits, and the earlier gradients are therefore not shown in Fig. 2(b), (c) and (d). According to Fig. 2 and the resultant Table 1, complete 30 and 50 s ribosomal subunits are made at about the same rate; as will be analyzed in detail in a further publication, it takes the same time to make a 30 s as a 50 s subunit. Furthermore, and more important, the subunits in and out of polyribosomes are labeled at the same rate (Table 1), which makes it very likely that the exchange of subunits in and out of polyribosomes is very rapid (Fig. 1(a)).

Table 1.

Radioactivity ratio of new to old rRNA (³H/¹⁴C) in free and polyribosomal subunits

Cumulative time of 3H- labeling (min)	Free subunits		Subunits derived from polyribosomes
	30 s	50 s	30 s	50 s
6	0.08	0.02	<0.005	<0.005
7	0.30	0.22	0.27	0.22
9	0.79	0.69	0.73	0.66
12	1.63	1.47	1.61	1.48
15	3.50	3.25	3.36	3.21
20	6.69	6.35	6.68	6.40
30	15.10	14.90	1.495	14.62

A fragile culture prelabeled with [¹⁴C]uracil was continuously labeled with [³H]uracil from time zero (Fig. 2 and text). At the indicated times, samples of cells were lysed and analyzed in sucrose gradients. Polyribosomes from the bottom of the tubes, and peak fractions of free 30 and 50 s subunits were dialyzed in low Mg²⁺ concentration and re-run in gradients. The ³H and ¹⁴C content of each sample was then determined to yield the listed values. The values are those observed in one of six experiments, all of which give very similar results. At times less than 6 min, no ³H-label was detectable in subunits (i.e. the ratio of ³H to ¹⁴C was less than 0·005).

(b). Synthesis of mRNA

Kinetic evidence from the sucrose gradient analyses and hybridization studies both show all the mRNA in polyribosomes.

(i). Evidence from partition of label in sucrose gradients

In the gradients of Fig. 2, summarized in Fig. 3, newly labeled RNA enters the polyribosome fraction linearly and essentially from zero time, although no newly labeled ribosomal subunits (or as shown below, any newly labeled ribosomal RNA) enter the polyribosomes for as much as seven to ten minutes. Thus, a large amount of the newly labeled RNA in the polyribosomes cannot be accounted for as ribosomal subunits. Nor can more than about 3% of the newly labeled RNA be accounted for by soluble RNA (calculated as follows: from the gradient analyses, sRNA comprises 20% of the RNA, or 25 molecules for every 30 to 50 s couple, and is labeled no more rapidly than rRNA (Fig. 2(a)). Since each functioning ribosome probably bears two molecules of sRNA (Warner & Rich, 1964), and half the ribosomes are functioning in polyribosomes, the polyribosomes contain 1/2 × 2/25 × 1/5∼1% of the total label in sRNA. With 30% of the total labeled RNA in polyribosomes, 3% is therefore sRNA.) At each time point during the continuous labeling with [³H]uracil, the contribution of every stable component in polyribosomes to the ratio ³H/¹⁴C is therefore adequately accounted for by the measured specific activity of purified ribosomal subunits (if anything, overestimated because of any appended fragments of mRNA). In Fig. 3(b) are shown the curves for the total radioactivity in polyribosomes and the radioactivity accounted for by stable RNA. The data shown are those for one of the six experiments, all of which closely agree. In Fig. 3(c) the difference between the two curves of Fig. 3(b) is plotted. The difference, which cannot be accounted for by rRNA, sediments in low Mg²⁺ concentration at 8 to 16 s or faster, and is highly sensitive to RNase (Mangiarotti & Schlessinger, 1966), suggesting that it is due to mRNA. About 30% of the total ³H label entering RNA is found in the polyribosomes for times up to about six minutes (Fig. 3(a)). This fraction of the total label corresponds to the estimates for the percentage of mRNA (relative to total RNA) in a pulse-label, obtained in earlier studies of base ratios and hybridization with a variety of organisms (Midgeley & McCarthy, 1962; Bolton & McCarthy, 1962), and is consistent with the notion that the RNA labeled at early times and entering the polyribosomes is the total new mRNA of the cells. Hybridization studies strongly support such an interpretation.

(ii). Evidence from hybridization

If mRNA and rRNA accumulate in polyribosomes in the way that Figs 2 and 3 suggest, then in the experiments using two labels, at early times (up to seven minutes) all the ³H-labeled RNA extracted from polyribosomes should hybridize as mRNA; later, the percentage that hybridizes as rRNA should progressively increase.

In order to distinguish between hybridized mRNA and rRNA, hybridization reactions were run in the presence and in the absence of a large excess of unlabeled, purified rRNA. Conditions were chosen such that essentially all the ¹⁴C-labeled, stable rRNA could be prevented from hybridizing, and any reduction in the hybridization of ³H-labeled RNA observed.

The results of the hybridization studies are reported in Tables 2 and 3, and in Fig. 4. When samples of RNA were hybridized with DNA, the kinetics of mRNA entry into polyribosomes were confirmed. Overlapping data for a number of trials with samples from five experiments are shown (Tables 2 and 3; Fig. 4(a) and (b)). The exact efficiency of hybridization from day to day was somewhat variable, even with portions of samples from a single experiment tested repeatedly on successive days, and duplicates were accurate only to ±5%. It is nevertheless clear that newly labeled RNA entering polyribosomes continues to have equivalent capacity to form hybrids from about 0·5 minute until 12 to 15 minutes after labeling begins: for this time interval, essentially none of the ³H-labeled RNA is prevented from hybridizing by a 20- to 50-fold excess (10 μg) of unlabeled rRNA. That the material excluded from the hybrid at later times by cold rRNA is itself rRNA is confirmed by the data plotted in Fig. 4(c). Authentic 23 s rRNA extracted from free 50 s ribosomes and repurified by zonal sedimentation always has the same ³H/¹⁴C ratio as the fraction of polyribosomal RNA prevented from hybridization by unlabeled rRNA (calculated from the data of Table 2).

Table 2.

Hybridization of labeled polyribosoml RNA

Cumulative time of 3H labeling (min)	Input		Hybridized		Hybridized in presence of 10 μg unlabeled RNA
	14C	3H	14C	3H	14C		3H
	(cts/min)	(cts/min)	(cts/min)	(cts/min)	(cts/min)	%	(cts/min)	%
Exp. 1
0·5	4040	1350	1030	670	85	2·1	609	45
1	4220	2810	1075	1090	102	2·4	1155	41
2·5	5420	5490	1730	2520	101	1·8	2410	44
1	4850	6570	1465	2760	72	1·5	2560	39
4	5168	14040	1730	5770	84	1·6	5970	42
6	4530	18350	1270	7020	97	2·1	6920	38
1 Total extract	3715	2510	972	705	33	0·9	345	13·7
Exp. 2
1	2743	1947	610	1010	54	1·9	1008	52
6	4180	5252	940	2810	53	1·2	2877	54
9	3556	7426	840	3688	61	1·7	3575	50·5
12	2902	8563	742	5082	32	1·1	4847	56
15	3305	11920	756	6435	29	0·9	5980	50·1
20	3416	16423	779	7096	37	1·1	6153	37
30	2765	19450	715	7970	25	0·9	5890	30
1 Total extract	7156	6363	1417	1286	82	1·1	971	15·2

Cells were pre-labeled with [¹⁴C]uracil and then continuously labeled with [³H]uracil from time zero (see text and Fig. 2). Data from two representative experiments are shown, one using 7·5 μc [³H]uracil/ml. to label cells from 0·5 to 6 min (Exp. 1), the other (Exp. 2) using 15 μc [³H]uracil/ml. to label cells from 6 to 30 min and a control flask labeled for 1 min with 7·5 μc/ml. At each time, a portion of cells was lysed and a part of the lysate centrifuged in a sucrose gradient for 10 hr at 56,000 g to yield a pellet of polyribosomes. 0·2 to 0·5 μg of RNA extracted from the polyribosomes was then hybridized with 75 μg DNA in duplicate samples (see Materials and Methods). In another sample, hybridization was carried out in presence of 10 μg unlabeled ribosomal RNA, to prevent hybridization of labeled rRNA by competition. To compare efficiencies from one experiment to another, a sample of total labeled RNA was similarly hybridized.

Table 3.

Hybridization of labeled RNA extracted from whole cells

Cumulative time of 3H labeling (min)	Input		Hybridized		Hybridized in presence of 10 μg unlabeled RNA
	14C	3H	14C	3H	14C		3H
	(cts/min)	(cts/min)	(cts/min)	(cts/min)	(cts/min)	%	(cts/min)	%
0·5	6024	1884	1860	660	171	2·8	302	16·0
1	6922	4245	2010	1490	129	1·8	650	15·4
2	6720	7471	1972	2492	20	0·3	1012	13·4
4	6778	15375	1764	5900	152	2·2	2704	17·6
6	5115	17896	1442	6940	21	0·4	2962	16·6
9	6333	9695	1728	3374	102	1·0	1550	16·0
12	7219	15665	1804	5124	95	1·3	1972	12·6
15	6521	19037	1842	6126	22	0·2	2326	12·2
20	7261	27694	2240	8309	105	1·4	3104	11·2
30	5709	30820	1704	8074	88	1·5	2462	8·0

Fragile cultures pre-labeled with [¹⁴C]uracil were continuously labeled with [³H]uracil from time zero. At intervals, portions of the culture were lysed, RNA extracted, and hybridization carried out (see legend for Table 2). In order to reduce the amount of [³H]uracil used, the labeling was done in two flasks, with 4 μc (0·68 μg) [³H]uracil/ml. for the series 0·5 to 6 min, and 4 μc (2·6 μg) [³H]uracil/ml. for the series 6 to 30 min.

Fig. 4.

Kinetics of entry of hybridizable, labeled RNA into polyribosomes.

As in the experiment of Fig. 2, [¹⁴C]uracil was used to label stable RNA; [³H]uracil was then added to follow the newly synthesized RNA. At intervals, portions of culture were lysed and centrifuged in a sucrose gradient to obtain a pellet of polyribosomes. RNA extracted from the polyribosomes was hybridized in presence or in absence of an excess of cold rRNA (as described in Table 2).

(a), (b) Curves for two different experiments are shown for sampling times of 0.5 to 6 min after the addition of [³H]uracil (a); curves for five experiments show the results from 6 to 30 min (b). Due to variations in the amount of tracers used, the absolute numbers of counts in the polyribosomes varied from experiment to experiment (Table 2). However, in all the experiments, for a constant amount of ¹⁴C-label in stable RNA, the amount of aH-label in polyribosomes increased linearly with time. To compare one experiment with another, the number of ³H counts found in polyribosomes after 6 min or 30 min of labeling were set equal: the curves representing the appearanoe of newly formed ³H-labeled RNA in polyribosomes with time (—▲—▲—) thereby became superimposable. The fractions of ³H-labeled RNA hybridized at each time, as indicated, in presence of an excess of cold rRNA were plotted for each experiment after a corresponding normalization (—○—○— and —●—●—). The efficiencies of hybridization vary from one experiment to another, but the shape of the curves (4(b)) is always the same.

(c) At various times during labeling, free 50 s ribosomes were isolated by gradient centrifugation and the 23 s rRNA extracted. The measured specific activity (ratio ³H/¹⁴C) in the purified rRNA (—○—○—) is compared with the ratio ³H/¹⁴C for the polyribosomal RNA blocked from hybridization by an excess of cold rRNA (—●—●–) (calculated from the data of Table 2).

Since little if any of the rapidly labeled RNA that enters polyribosomes early in the experiment is blocked from hybridization by cold rRNA, all of it is identified as mRNA. The observed efficiency of hybridization (35 to 60%) is consistent with the efficiency of mRNA hybridization observed in other studies (for example, McConkey & Dubin, 1965; Friesen, 1966).

Hybridization of the mRNA in total pulse-labeled RNA provides another argument that the new RNA entering polyribosomes at early times is mRNA. The percentage of pulse-labeled RNA that hybridizes as mRNA should be constant for some time (Bolton & McCarthy, 1962), and should follow the percentage of total RNA detected as mRNA in sucrose gradients; this it does (Table 3; also compare Figs 3(c) and 4(b)).

(c). Co-ordination of polyribosome formation with mRNA synthesis

The evidence shows that a polyribosome forms along with its corresponding mRNA.

All, or very nearly all the mRNA detectable by hybridization of pulse-labeled RNA is found in the polyribosomes (Fig. 4 and Tables 2 and 3), and therefore unfinished chains must already be associated with the polyribosomes. One might anticipate that incipient polyribosomes would appear in cell extracts still bound to DNA, but in our studies treatment with deoxycholate was used to release the nascent polyribosomes from DNA (Mangiarotti & Schlessinger, 1966), presumably by denaturing some part of the relevant structure such as RNA polymerase (Bremer & Konrad, 1964).

Table 4 shows that the newly formed RNA from the regions of a gradient in which ribosomal subunits and smaller entities are found hybridizes as rRNA, as expected; its association with DNA is almost total blocked by cold rRNA. Very likely, slowly sedimenting “mRNA” observed free of ribosomes in some earlier studies arose during the runoff of subunits and fragmentation of polyribosomes incurred during extraction, both of which are minimized by the present use of fragile cells (Mangiarotti & Schlessinger, 1966).

Table 4.

Hybridization of non-polyribosomal labeled RNA

Cumulative time of 3H labeling (min)	Input		Hybridized		Hybridized in presence of 10 μg unlabeled RNA
	14C	3H	14C	3H	14C		3H
	(cts/min)	(cts/min)	(cts/min)	(cts/min)	(cts/min)	%	(cts/min)	%
0·5	5224	1230	2610	590	34	0·6	46	3·5
1	2960	1610	1620	805	28	0·9	65	4·6
1·5	1778	1600	1044	735	26	1·4	82	5·1
2	1746	2760	1074	1430	24	1·3	142	5·1
4	2252	7060	1210	3870	32	1·5	285	4·0
6	3580	18810	1840	8470	28	0·8	736	4·0

From master gradients in an experiment like those analyzed in Fig. 2 and Table 2, the samples extending from 4 s to 50 s were pooled and RNA extracted and concentrated by precipitation with 2 vol. 95% ethanol at −10°C. The resuspended RNA, from cells labeled with 0·33 μc [¹⁴C]-uracil/ml. and 5 μc [³H]uracil/ml. was dialyzed and assayed for hybridization as in Materials and Methods.

The small, variable percentage of radioactive material remaining hybridized in the presence of cold rRNA (Table 4) requires additional comment (from 0·3 to 5°1%, varying from point to point and from experiment to experiment). It is possible that this represents only a variable amount of mRNA released by runoff from some polyribosomes. Alternatively, it might correspond to small, newly initiated mRNA chains still bound to DNA in the lysates; a small percentage of the total pulse-labeled RNA (up to 1%) usually moves at 30 to 50 s, but barely sediments after DNase treatment (Mangiarotti & Schlessinger, 1966, and unpublished results).

The kinetic data demonstrate that the newly formed [³H]RNA free of polyribosomes and still “hybridized” in presence of cold rRNA cannot be an obligate precursor pool of completed mRNA chains that have not yet entered polyribosomes. If each newly formed mRNA passed through such a pool, a kinetic precursor of polyribosomal mRNA would become increasingly hot, while a lag would be observed before the mRNA in polyribosomes began to be labeled at a maximal rate. Instead, the rate of entry of newly labeled mRNA into polyribosomes is a maximum within 10 seconds (Fig. 3(b), inset). Therefore, if there is a free pool of finished mRNA, then chains must be made and cycle through it in a few seconds. Since the synthesis of a typical chain of rRNA or mRNA requires of the order of one to two minutes or more (see below and Discussion), we conclude that there is no free pool of completed mRNA that has not yet entered polyribosomes.

The sedimentation properties of RNA extracted from polyribomes after a pulse label are also consistent with the conclusion that the mRNA of the cells is localized in polyribosomes. The pulse-labeled RNA extracted from polyribosomes at early times displays a broad peak not unlike the distribution of mRNA molecules in phage-infected cells (Asano, 1965), with no easily discernible relation to the 16 and 23 s rRNA (Figs 5 and 7(c)). Only at 9 to 12 minutes, when an appreciable fraction of the new label ([³H]uracil) in polyribosomes is contributed by newly formed ribosomal subunits, does the pattern begin to include well-defined peaks of 16 and 23 s rRNA (Fig. 5). In contrast, as one might expect for a fraction that contains new rRNA, extraction of the non-polyribosomal part of the gradient (50 s and less), yielded only 16 and 23 s and smaller bits of rRNA, along with some 4 s material that is presumably tRNA (unpublished results).

Fig. 5.

Zonal sedimentation profile of RNA extracted from polyribosomes.

A culture of fragile cells received a pulse of 0·1 μc [¹⁴C]uracil/ml. followed by two generations of growth to label stable RNA. At 1·5 × 10⁸ cells/ml., 5 μc [³H]uracil/ml. were added. At intervals 10-ml. samples were harvested, lysed and the lysates centrifuged in sucrose gradients at 56,000 g for 12 hr to sediment most of the polyribosomes to the bottom of the gradient. From the polyribosomal pellet of each master gradient, RNA was extracted with 0.5% sodium dodecyl sulfate, then resedimented in a sucrose gradient at 56,000 g at 16°C for 18 hr. For clarity, the pattern of stable, ¹⁴C-labeled RNA is omitted.

Top: ³H-labeled RNA extracted after a labeling time of 0·5 min(—▲—▲—); 1 min (—Δ—Δ—); 1·5 min (—○—○—); 2 min (—●—●—).

Bottom: ³H-labeled RNA at 6 min (—○—○—); 12 min (—●—●—).

Fig. 7.

Zonal sedimentation profile of mRNA extracted from different size classes of polyribosomes.

a) A 3-min pulse of [³H]uracil was given to fragile cells prelabeled with [¹⁴C]uracil (as in the experiment of Fig. 2). Analysis of a sucrose gradient after centrifugation for 2 hr at 56,000 g at 4°C to spread out the polyribosomes. Regions arbitrarily demarcated 1 to 6 from largest to smallest polyribosomes. (– –●– –●– –) ³H-labeled RNA; (—●—●—) ¹⁴C-labeled RNA.

(b) Zonal sedimentation analyses of RNA from some of the regions defined in the gradient of (a). Fractions in each numbered region of (a) were pooled and the RNA extracted and recentrifuged in sucrose gradients containing 0·5% sodium dodecyl sulfate for 20 hr at 16°C to show 16 s and 23 s markers (– – –) and ³H-labeled mRNA (———).

(c) Sedimentation profile of total mRNA after a 3-min pulse of [³H]uracil compared with the summation of analyses of subfractions. (—○—○—) Observed distribution of total mRNA extracted from the polyribosomal pellet of a master gradient centrifuged 10 hr; (—●—●—) calculated distribution of total mRNA reconstituted by superimposition and summation of the analyses of the 6 subfractions indicated in (a) (4 of the 6 analyses of subfractions are shown in 7(b)). (– – –) ¹⁴C-labeled marker RNA.

(d). Precursors of complete rRNA chains

Since little if any rRNA appeared in polyribosomes except in finished subunits, the newly formed rRNA chains could be expected to sediment in precursors. When a gradient of a cell extract was centrifuged for 20 hours rather than 10 hours, in order to spread out the region in which nascent rRNA should appear, a precursor region containing 6 to 25 s material was revealed (Fig. 6(a)). To estimate the size of the RNA chains sedimenting in this region, fractions demarcated by the arrows on the gradient shown in Fig. 6(a) were pooled, extracted with 1% sodium dodecyl sulfate and phenol, and recentrifuged. As expected, in addition to complete 16 and 23 S chains, there were many that sedimented much more slowly. That essentially all this material, of very high (³H/¹⁴C) content, was rRNA was once again confirmed by hybridization; the considerable hybridization observed was essentially all competed for by cold rRNA (Table 4). The conclusion that unfinished rRNA chains are not included in polyribosomes is thereby strongly supported.

Fig. 6.

Precursors of complete rRNA chains free of polyribosomes.

Gradient analysis of a lysate from cells prelabeled with [¹⁴C]uracil to label stable RNA, then pulse-labeled 2 min with [³H]uracil. The cell lysate was centrifuged for 20 hr at 56,000 g to display the region between 4 and 30 S. (a) (– –●– –●– –) ³H-labeled RNA; (—●—●—) ¹⁴C-labeled RNA. A broad distribution of small ³H-labeled chains is observed. The inset in (b) shows how the (³H/¹⁴C) ratio in the regions demarcated 2 and 3 in (a) by arrows is saturated with time.

Zonal sedimentation analyses of purified unfinished rRNA chains are also shown (b). Gradient fractions in each of the three regions demarcated by arrows in (a) were pooled, RNA extracted by the phenol method, and gradient centrifugation done in 0·5% sodium dodecyl sulfate, 20 mM-Tris (pH 7·5), 50 mM-NaCl, at 15°C and 56,000 g for 24 hr. To avoid confusion, only the peak positions of the ¹⁴C-labeled marker 16 s and 23 s RNA are indicated. Curve 1, RNA from region 1 of Fig. 6(a); curve 2 from region 2; curve 3, region 3.

When analysis of samples during a pulse label were compared, the intermediate region between 4 s and about 20 s rose rapidly in ³H content, then leveled off at about two minutes (data plotted in the inset of Fig. 6(b)). The fraction therefore behaves like a precursor pool containing rRNA chains destined to become the 16 and 23 s RNA of finished ribosomal particles. Presumably, one to two minutes is the time required to make a complete chain of rRNA; the pool of unfinished precursors is then saturated. (Additional supporting evidence for this interpretation is presented in a forthcoming paper, in which the biosynthesis of ribosomes is examined in detail.)

(e). Size of mRNA in subclasses of polyribosomes

The distribution of ³H-label in zonal sedimentation analysis of the RNA extracted from the polyribosomes at intermediate time samples (Fig. 7(c)) should approximate the weight sedimentation rate distribution of mRNA. (At later times, a very similar distribution is obtained if the measured contributions of ³H-labeled 16 and 23 s RNA in complete subunits is subtracted.) The analysis can be extended to subfractions of mRNA extracted from classes of polyribosomes spread out in a gradient centrifuged for two hours (Fig. 7(a)). RNA samples from the six polyribosomal regions demarcated with arrows (Fig. 7(a)) were then re-run to determine the distribution of mRNA sizes in each region. Four of these subfractions were used for Fig. 7(b). While larger mRNA molecules tend to appear in faster-moving polyribosomes, the spread of mRNA sizes in each subclass of polyribosomes is very large.

(f). Precursors of complete mRNA chains

Before newly labeled rRNA appears in complete ribosomes during a continuous labeling, the only labeled RNA in the polyribosomal fraction is mRNA. A zonal sedimentation analysis of mRNA extracted from polyribosomes is therefore possible. Since the lifetime of mRNA is much longer than a few minutes, the synthesis and growth of new chains can be examined during the first minutes of labeling, assuming that no degradation by endonuclease occurs in the cells or during extraction of the RNA. At first, just as with rRNA (Fig. 6), small chains, including precursors of larger ones, show a relative excess of radioactivity (Fig. 5). Then the larger chains of mRNA become increasingly radioactive as they begin to include molecules fully labeled with [³H]uracil. From the data of Fig. 5, one can estimate that, just as with rRNA, it takes of the order of one to two minutes to form a complete mRNA molecule of 3000 nucleotides (23 s).

4. Discussion

Since all, or very nearly all, of the mRNA is in polyribosomes during exponential growth (Figs 3 and 4) and ribosomal precursors are free of polyribosomes (Figs 2, 3 and 4), the maturation of each can be followed. The following life cycles are suggested.

To make a complete chain of RNA (rRNA or mRNA), requires of the order of one to two minutes. Each chain of 16 or 23 s rRNA then requires at least an additional five minutes to gain its complete complement of ribosomal proteins. When completed, but not before, the 30 and 50 s ribosomal subunits can enter polyribosomes. From their free pool, 30 and 50 s subunits periodically couple and traverse a molecule of mRNA in a polyribosome. Each time a polypeptide chain is completed, the couple separates to return to the free pool.

In contrast to rRNA, as each molecule of mRNA is formed, it combines with ribosomes in a polyribosome. The finished mRNA presumably continues to function on more ribosomes for up to 9 to 11 minutes before it is broken down to nucleotides.

(a). Rate and amount of subunit synthesis

The average time required to make a subunit can be estimated by extrapolating the final rate of formation back to the abscissa (Fig. 3(b)), and is approximately 15 to 18 minutes. From the growth rate of the cells (doubling time 120 minutes), it can then be calculated that about 10% of the rRNA is in the form of precursors at any time.

A number of previous reports have indicated that 30 S particles are made more quickly than 50 s ribosomal subunits (for example, Gros et al., 1961; Levinthal, Keynan & Higa, 1962; McCarthy et al., 1962). Our results suggest that the complete subunits are formed at the same rate. Possibly the fraction of 50 s precursors moving at about 30 s in sucrose gradients, or fragments of bound mRNA, were earlier taken to be authentic 30 s particles. The 50 s precursors (Figs 2 and 6 and experiments to be published) include two components that sediment at about 32 and 43 s, as suggested earlier by McCarthy et al. (1962); the principal 30 S precursor sediments at about 26 s.

(b). Rate of couple exchange

Earlier results that runoff of ribosomes to the pool of subunits can continue when cells are rapidly cooled (Mangiarotti & Schlessinger, 1966) suggested that the subunits come apart each time a protein chain is finished. Nevertheless, while no “monomers” (“70 s particles”) were observed in extracts, it was still possible that all functioning ribosomes were monomers in polyribosomes, with each 30 to 50 s couple moving from one mRNA molecule to another. No exchange—or very little—of free with polyribosomal subunits would then be observed. However, the data reported here strongly support the suggestion that the exchange of 30 and 50 s subunits in polyribosomes with the pool of free subunits must be very rapid (Fig. 1(a)), since subunits in and out of polyribosomes always have the same ³H^/14C ratio (Table 1).

While equilibration is rapid, one cannot conclude that 30 and 50 s subunits change partners as often as they uncouple. Geographical proximity may favor the repeated coupling of some 30 and 50 s subunits each time they separate in the cell, especially if they continue directly to the next cistron of a polycistronic message. The true rate of exchange of coupled partners with others in a free pool can probably be definitively measured only by a transfer experiment in CsCl (see Meselson, Nomura, Brenner, Davern & Schlessinger, 1964).

(c). Amount and chemical lifetime of mRNA

While no direct information about the functional lifetime of mRNA is available from these experiments, the location of all mRNA in polyribosomes suggests that it continues to function for most of its chemical lifetime. A number of estimates agree that, in the cultures studied, the average chemical lifetime is about 11 to 12·5 minutes, and that about 2·8% of the total RNA is mRNA.

(i). Amount and lifetime of mRNA from its fraction in new RNA and in total RNA

An estimate of the amount of mRNA can be obtained from the percentage of total incorporated radioactive material in mRNA after a time of labeling long enough to ensure that mRNA has reached a constant specific radioactivity. In the graph of Fig. 3(c), it is apparent that 30 minutes is an adequate time. 18% of the total incorporated label (Fig. 3(a) and (c)) is then in mRNA. The appropriate equation (Levinthal et al., 1962) is:

$$\frac{\text{Cts}∕\text{min stable}}{\text{Cts}∕\text{min total}}=1-\frac{M}{M+S}\cdot \frac{\text{exp}{k}_{1}t}{\text{exp}{k}_{1}t-1},$$ (1)

where k₁ is the exponential growth constant, In 2/mass-doubling time of the culture, here = 1/173; M is the amount of mRNA; S is the amount of stable RNA; and we solve for M/(M + S), the fraction of mRNA in total RNA. Since, at t = 30 minutes, the ratio cts/min stable/cts/min total is 0·82, it is calculated that 2·8% of the total RNA is mRNA. Unlike the calculations in sections (iii) and (iv) below, this estimate is independent of the rate at which pools and mRNA become saturated with label (Levinthal et al., 1962). Equation (1) includes a correction for the unlabeled nucleotides present in mRNA at the start of the experiment. Correction for the unlabeled nucleotides in the initial precursor pool is small enough to be ignored (Levinthal et al., 1962), and in our case is insignificant, since the pool is very small. This can be seen from Fig. 8, for example, where the existing precursor pool produces no detectable lag in the uptake of label into cellular RNA.

Fig. 8.

Uptake of [³H]uracil into fragile cells; comparison of control cells and cells pre-incubated with unlabeled uracil.

Cultures were first grown to 10⁸ cells/ml. Then, to one flask 3 μg [³H]uracil/ml. (0·055 μc/μg) was added; to a second, 3 μg unlabeled urecil/ml. was added 10 min before the addition of sufficient [³H]uracil to yield the same final specific activity (0·055 μc/μg). To ensure that each sample contained the same number of cells and at least 1000 cts/min of [³H]uracil, sample sizes varied from 10 ml. (0.25 min) to 0.01 ml. (115 min), and a complementary volume of unlabeled cells was then added to the later samples before precipitation with acid and filtration onto nitrocellulose filters. Two sets of points indicate the uptake of [³H]uracil into the control culture (●) and the pre-incubated culture (○). The curve is a theoretical one for exponential incorporation of label from time-zero R* = C_l (exp k₁t−1), with k₁ = 1/173 (see text). C₁, is chosen to give coincidence with the experimental points at the latest times. The lower curve and points plot the data up to 5 min on a time scale expanded 10-fold.

An average lifetime can be obtained by solving the following equation:

$$\frac{M}{M+S}=\frac{\frac{1}{k_2}(a)}{\frac{1}{k_2}(a)+\frac{1}{k_1}(b)}$$ (2)

Here $\frac{M}{M+S}$ is again the fraction of mRNA in total RNA (0·028); 1/k₂ is the average lifetime of mRNA (with random or ordered decay; see section (iv) below); (a) is the fraction of newly formed RNA that is mRNA; (b) is the fraction of newly formed RNA that is stable; and 1/k₁ is the generation time of the culture (173 minutes). Thus (1/k₂) (a) is the RNA synthesis required to make an equivalent of the initial amount of mRNA; 173 (b) is the synthesis necessary to make an equivalent of the initial stable RNA. (a) is 0·3 and (b) is 0·7 (from the initial slopes in Fig 3(a)) assuming, as is usually done, that one pool of RNA precursors serves both for stable and unstable RNA (see e.g. Midgeley & McCarthy, 1962).

The lifetime calculated from equation (2) is 11·7 minutes. This value is independent of the specific radioactivity of the pool of RNA precursors, since it depends only on relative rates of synthesis.

(ii). Relative radioactivity of precursor pools during the experiment

In order to calculate the percentage and lifetime of mRNA in other ways (sections (3) and (4) below), the variation in specific radioactivity of the RNA precursor pool with time must be known. While we have made no radioactivity measurements on extracted nucleotides, the degree of saturation of the pool with label at any time can be derived from our data as follows.

At early times, when mRNA takes up 30% of the incorporated label, the total rate of incorporation of label from a fully labeled pool should be 30% greater than at 30 minutes, when unstable mRNA has been completely replaced and nearly all incorporation (97%) is into stable RNA. Instead, label appears in total RNA at a constant rate, after correction for exponential growth, from the time that [³H]uracil is added to the medium (Figs 3 and 8). Therefore, the relative specific activity of the pool of nucleotide precursors must increase from about 0·7 to 1·0 over the course of the experiment.

This argument can be stated in another way. The uptake of label into stable RNA is directly proportional to the radioactivity of the precursor pool, and is given in these experiments (Fig. 3(a) and (c), by the difference at any time between total uptake and that into mRNA. Since the total rate of uptake remains roughly constant, while the mRNA declines from 30 to 3%, the relative rate of labeling of stable RNA—and therefore, the relative specific activity of the pool—increases from 0·7 to 1·0. Furthermore, since the rates of synthesis of total and stable RNA do not vary for at least 10 to 12 minutes (Figs 3 and 4), the specific radioactivity of RNA precursors must be nearly constant during that interval.

That the pool is 70% saturated with label from zero time strongly suggests that a small pool of precursors expands greatly when uracil is added to the medium. A number of supplementary experiments has shown that the supposed expansion, similar to that observed in Bacillus subtilis (Levinthal et al., 1962 and personal communication), does in fact occur. For example, Fig. 8 shows the depression in the rate of uracil uptake by pre-incubation of a culture in cold uracil, compared to the culture which has had no earlier addition of cold uracil. The observed delay of uptake of radioactivity indicates that cold uracil in the medium probably creates a larger pool in the cells; the larger pool takes more time to equilibrate. In this important respect, the fragile cultures used differ from non-fragile E. coli, which show no such pool expansion and therefore have very different initial kinetics of labeling (McCarthy & Britten, 1962).

(iii). Lifetime and amount of mRNA from time of linear labeling and fraction in new RNA

A second, independent estimate of the minimum lifetime of mRNA can be obtained directly from Figs 3 and 4 and the knowledge that the rate of uptake of label into the cells is very nearly constant for at least 12 minutes. The argument followed is that of Bolton & McCarthy (1962). For 10 to 12 minutes a constant fraction of 30% of the label incorporated per unit time enters mRNA. Thus, for 10 to 12 minutes, since no decline in the rate of uptake is observed, no measurable fraction of the newly synthesized mRNA is destroyed. The decay of some mRNA might be missed if it contained very little label—if, for example, the RNA precursor pool were poorly labeled for some time and then increased several fold in specific radioactivity. However, no such change in the pool occurs (see section (ii) above). Therefore, no sizable fraction of mRNA can have a lifetime shorter than about 12 minutes; this minimum estimate is in good agreement with the average estimate of section (i) above.

Obviously the independent estimate of lifetime can be substituted in equation (2), and the equation solved to give a second estimate of 2·9% mRNA in total RNA, in agreement with that of section (i).

(iv). Lifetime of mRNA from the shape of the mRNA labeling curve

A third estimate of the lifetime can be derived from the “absolute” rate at which label enters mRNA (Fig. 3(c)). However, such estimates depend greatly on the way in which nucleotide precursor pools become saturated with label. We first compute the maximum lifetimes for extreme models of ordered and random breakdown, then discuss the extent to which the variation in the specific activity of the pool during the course of the experiment lowers the estimates.

If decay is ordered—i.e. if every mRNA molecule synthesized lasts chemically for a fixed time—then labeling of mRNA follows the relation:

$$M^{\ast }=C_1(k_{2}t)\text{exp}{k}_{1}t$$ (3)

where M* is the amount of label in mRNA, C₁ is the radioactivity mRNA would contain at zero time if it were fully labeled, k₂ is the decay constant of mRNA in minutes (the reciprocal of its chemical lifetime), and k₁ the growth constant (1/173), until all the mRNA existing at the time of the addition of label to the medium has decayed and been replaced by labeled mRNA (t = 1/k₂). Thereafter, the further increase in M* will follow exponential growth,

$$M^{\ast }=C_1\text{exp}{k}_{1}t.$$ (4)

After dividing each point by exp k₁t to correct for the exponential growth of the culture (Fig. 3(c)), the average lifetime of mRNA (1/k₂) is therefore the time required to reach the saturation value C₁. The lifetime from Fig. 3(c) would then be 16 to 18 minutes.

If instead, decay of the mRNA is random (exponential), then mRNA would be labeled like a precursor pool (Bolton & McCarthy, 1962); i.e.,

$$M^{\ast }=C_1(1\text{-exp}-k_{2}t)\text{exp}{k}_{1}t.$$ (5)

The lifetime of such a fraction can be defined as the time required to synthesize an amount of mRNA equivalent to that existing at zero time. The same value is obtained, for exponential processes, either by observing the time at which 0·63 of the final labeling is achieved (decay to 1/e), or by extrapolating the initial slope of the curve to the time at which it intersects the final plateau (C₁). The second procedure is more accurate here, because the initial slope is relatively easy to determine. Again, the average lifetime is found to be 16 to 18 minutes.

The relationships used assume that the pool is of constant specific activity throughout the experiment. Since the pool is increasing in specific activity, label will continue to enter mRNA after it has all been degraded and replaced once; the lifetime obtained is therefore a maximum value. However, the estimate can be corrected, since the degree of saturation of the pool at the start of the experiment is known to be 0·7. The lifetimes estimated by extrapolation of initial slopes must therefore be multiplied by 0·7. A corrected lifetime of 11·2 to 12·6 minutes is obtained, once again in good agreement with the other estimates.

The data do not give enough resolution to distinguish between ordered and random decay of mRNA. Nor is it clear whether there are classes of unstable mRNA that have a lifetime very different from the average. However, since the lifetime of the bulk of the mRNA is at least 10 minutes, any fractions of relatively short lifetime must be small.

(d). Size distribution of mRNA

An approximation to the distribution of sizes of messenger RNA molecules is shown in Fig. 7(c), in which the mRNA extracted from polyribosomes is displayed. The over-all sizes of mRNA show peaks at about 14 to 16 s, similar to distributions already reported for phage-infected E. coli mRNA (Asano, 1965) and for normal E. coli mRNA (Monier, Naono, Hayes, Hayes & Gros, 1962), if the larger contribution of contaminating rRNA is estimated and subtracted from their patterns.

It is possible that the pattern is distorted by some degradation or aggregation. Aggregation seems unlikely, as conditions adequate to prevent intermolecular aggregation of T4 mRNA (Asano, 1965) have been used, and addition to the preparations of tenfold more unlabeled total RNA or isolated ribosomal RNA did not change the gradient patterns (unpublished results). Degradation is much more difficult to exclude, here and in the studies mentioned above. However, the following arguments suggest that the preparations are relatively intact. (1) The same sedimentation distributions of labeled mRNA were observed when preparations were simply put into 0·5% sodium dodecyl sulfate before analysis or also extracted with phenol to eliminate traces of protein. (2) The size distributions of mRNA derived from fractions of polyribosomes (Fig. 7(b)) demonstrably sum up to the distribution observed in the total unfractionated mRNA (Fig. 7(c)). (3) The RNA molecules should be relatively insensitive to shear compared to polyribosomes, and probably not much more sensitive to traces of ribonuclease; since the preparative methods are adequate to avoid most degradation of polyribosomes, they should be adequate to preserve mRNA. (4) Fractions of rapidly sedimenting mRNA can be re-run in another gradient, and sediment again at the same rate (unpublished results). (5) Essentially the same extraction procedures have yielded sharp peaks of very large mRNA molecules specific for the lactose and histidine operons (see below).

Since mRNA is metabolically unstable, some of the observed heterogeneity may arise from molecules partially degraded or in process of formation in the cell. However, the contribution of degraded chains can be estimated as small. The average mRNA is formed in about a minute, judging from Figs 5 and 7, and has a life expectancy of about 12 minutes; therefore, assuming that the rate at which a molecule of mRNA is degraded is not too different from its rate of formation, only about 20% of the mRNA chains in the cell are incomplete at any instant.

If one thus assumes that the sedimentation distribution of cellular mRNA is represented by Fig. 7(c), and that the relation between sedimentation rate of mRNA and molecular weight is the same as for rRNA, the size distribution of mRNA can be estimated from the relation:

$$S=K{M}^{0.52},$$ (6)

which is calculated on the basis that 16 s and 23 s rRNA have molecular weights of 0·55 and 1·1 × 10⁶ (Kurland, 1960). The results are indicated in Fig. 7(c) by the molecular weight scale above the gradient pattern. Boedtker (1966) has pointed out the limitations of this simple approach, especially the possibility that two RNA molecules of the same molecular weight may sediment very differently if they differ in secondary structure. However, in line with her suggestion (Boedtker, 1967), we have repeated the gradient analyses with RNA heated and sedimented in 3% formaldehyde to abolish most secondary structure, and the sedimentation profile of mRNA relative to marker rRNA did not change appreciably. Therefore, the molecular weight scale of Fig. 7(c) is probably a fair first approximation.

Molecular weights of 2·25, 4·5 and 6·75 × 10⁵ (Fig. 7(c), top scale) indicate the approximate location of molecules containing about 750, 1500 and 2250 nucleotides, or enough RNA to code for one, two, and more than two protein chains of 30,000 molecular weight. Very likely, the 60% of the mRNA molecules than can code for one chain of 30,000 or less are monocistronic. If all mRNA molecules coded for one, two or three proteins of molecular weight 30,000, then according to Fig. 7(c), with 3% of the cellular RNA in mRNA and 80% in 20,000 30 s and 20,000 50 s subunits, each cell would contain about 1350 molecules of mRNA coding for one protein, 600 coding for two proteins and 600 coding for three proteins. Very large polycistronic messages, like those reported for the lactose (Guttman & Novick, 1963; Kiho & Rich, 1964; Attardi, Naono, Rouvière, Jacob & Gros, 1963; Hayashi, Spiegelman, Franklin & Luria, 1963), tryptophan (Imamoto, Morikawa & Sato, 1965), and histidine operons (Martin, 1963), seem to be relatively rare.

(e). Sedimentation rate of polyribosomes and size of mRNA

Since the ratio of mRNA per ribosome is roughly constant across the distribution of polyribosomes (Fig. 7(a)) the simplest interpretation is that the loading of different mRNA molecules is roughly comparable. One would then have expected that larger RNA molecules would appear in faster-moving polyribosomes (Kiho & Rich, 1964). In general, this is true (Fig. 7(b)). However, some large mRNA molecules are found in slow-moving polyribosomes, and some small mRNA molecules are found in polyribosomes that sediment very rapidly (Fig. 7(b)). Furthermore, it remains unexplained how a peaked distribution of sizes of mRNA molecules, comparably loaded with ribosomes, can generate a flat distribution of polyribosomes. Among the speculative possibilities are:

(1) that the sedimentation rate of polyribosome is not dependent only on its content of ribosomes: for example, polyribosomes could exist in different states of coiling in solution, so that some small, tightly coiled ones could move relatively fast;

(2) that the true distribution of mRNA sizes could be more nearly flat than the one observed, with breakage of molecules at weak points during extraction yielding the observed peaked distribution; in that case, the number of large molecules of mRNA will have been underestimated above or;

(3) a number of secondary processes, such as some aggregation of small polyribosomes or runoff of ribosomes from some large mRNA molecules, may alter the distribution somewhat from that in the cell.

(f). Polyribosomes form with mRNA but not with rRNA

Regardless of the exact distribution of ribosomes in polyribosomes, the kinetic analysis and the direct determination of hybridizable mRNA suggest that polyribosomes form as the mRNA is being synthesized, a notion that has been advanced earlier by various authors, including Noll, Staehelin & Wettstein (1963); Bremer & Konrad (1964); Byrne, Levin, Bladen & Nirenberg (1964); Fox, Gumport & Weiss (1965); Naono, Rouviere & Gros (1966); Stent (1965); and Shin & Moldave (1966). Apparently, bacteria contain no “transport” form like that which carries mammalian mRNA from nucleus to cytoplasm (Henshaw, Revel & Hiatt, 1965; McConkey & Hopkins, 1965; Joklik & Becker, 1965; Perry & Kelly, 1966; Spirin, 1964).

While polyribosomes form with nascent mRNA, complete mRNA molecules can also bind to ribosomes and function, as has been shown in vivo (Lodish, Cooper & Zinder, 1964; Conconi, Banks & Marks, 1966) and in vitro (Barondes & Nirenberg, 1962; Spyrides & Lipmann, 1962; Gilbert, 1963; Haselkom & Pried, 1964). The most reasonable conclusion from all these studies is that a polyribosome can form with a molecule of mRNA whether or not it is complete.

In sharp contrast to the results with mRNA, ribosomal RNA does not seem to appear in polyribosomes in an incomplete form. One can thus infer that rRNA never functions as mRNA in the cell. Apparently, rRNA is released from DNA without the necessity for attendant attachment to ribosomes (suggested, for example, by Stent (1965). Thus, a selective mechanism must exist to permit ribosomes to join on to and begin to move along those RNA molecules destined to function as messenger RNA, but not to rRNA (or presumably tRNA). Perhaps a particular 5′-terminal sequence is required for this recognition; or the attachment of ribosomal proteins to rRNA (as suggested by Kurland & Maaløe, 1962; Stent, 1964) might prevent them from joining to ribosomes.

We have repeatedly benefited from advice and discussions with David Apirion and David Kennell and, in many of the later experiments, from the technical help of Lorenzo Silengo. This work was supported in part from U.S. Public Health Service grants GM-10447, HD-01956, and training grant T1 AI 257. One of us (G. M.) is on leave from the Institute di Chimica Biologica, Genoa.

Notes added in proof:

1. Although there is a large amount of new ³H-labeled RNA in the ribosomal precursors in Fig. 2, it should be noted that these precursors contain only about 10% of the total cellular rRNA (McCarthy et al., 1962; Mangiarotti, Apirion, Schlessinger & Silengo, results to be published). Hence, no precursor peaks appear in the steady-state pattern of stable ¹⁴C-labeled RNA.

2. It has been assumed, in our discussion of the rate of exchange of subunits in and out of polyribosomes, that an appreciable fraction of the observed free subunits exist free of polyribosomes in the cell. If most of the free particles are produced, instead, by runoff of ribosomes from polyribosomes during harvest and lysis of cells, then the appearance of new label in any authentic free subunits would be masked by the backflow of label from polyribosomes, and conclusions regarding the rate of exchange would then be impossible. The evidence for the existence of a sizable fraction of free subunits in our cells is detailed in Mangiarotti & Schlessinger (1966). Coupled with other recent evidence (Kaempfer, R., Raskas, H. & Meselson, M., manuscript submitted; Nomura & Lowry, 1967; Schlessinger, Mangiarotti & Apirion, 1967), the conclusion of very rapid exchange—possibly every time a polypeptide chain is completed—seems increasingly justified.