Structural dynamics of bacterial ribosomes

J. Mol. Biol. (1976) 105, 111-130

Structural Dynamics of Bacterial Ribosomes V. Magnesium-dependent Dissociation of Tight Couples into Subunits: Measurements of Dissociation Constants and Exchange Rates MARKUS Nom-f AND HANS NOLL$ Department of Biochemistry and Molecular Biology Northwestern University, Evanston, Ill. 60201, U.S.A. (Received 18 June 1975, and in revised form 28 October 1975) The magnesium ion-dependent, equilibrium of vacant, ribosomc couples with thei subunits k, 50 s + 30 s 70 s s k-% has been studied quantitatively with a novel equilibrium displacement labeling method which is more sensitive and precise than light-scattering. At a concentration of 10e7 M, tight couples (ribosomes most active in protein synthesis) dissociate between 1 and 3 mM-Mg2+ at 37°C with a 50yo point at 1.9 mM. The corresponding association constants K: are 5.1 x lo5 Mm1 (1 mM-Mg2+), 3.5 X lo7 M-l (2 mM), and 1.2 x log M-~ (3 mM), about five orders of magnitude higher bhan the K: value of loose couples studied by Spirin et aE. (1971) and Zitomer & Flaks (1972). In this range of Mg2 + concentrations (37”C, 50 mM-NH,+) t’he rate constants depend exponentially and in opposite ways on t,he Mg2+ concentration: kl = 2.2 x 10-3s-1,k-, = 7.7 x 104~-‘s-l (2mM-Mg2+);k, = 1.5 x 10-4s-1,k_1 = 1.7 x lo7 M-l s-l (5 mM-Mg’+). Under physiological conditions (Mg2+ 1: 4 mM, ribosome concn N 1O-7 M), the equilibrium strongly favors association and the rate of exchange is slow (tt 21 10 min). In the range of dissociation (2 mm-Mg2’), association of subunits proceeds without measurable entropy change and hence dQ’ = AHo. The negative enthalpy change of AH0 = - 10 kcal suggests that association of subunits involves a shape change. Below a critical Mg2 + concentration (2: 2 mM), t’he 50 S subunits are converted irreversibly into the b-form responsible for the transition to loose couples. The results are compatible with two classes of binding sites, one class binding Mg2+ non-co-operatively and contributing to the free energy of association by reduction of electrostatic repulsion, and another class probably consisting of hydrogen bonds between components at opposite interfaces whose critical spatial alignment rapidly denatures in the absence of stabilizing magnesium ions.

1. Introduction Ever since the discovery of the bipartite design of the ribosome and its Mg2+ dependent dissociation into subunits, it has been assumed that the observations can be described by a chemical equilibrium of the type 70s z 50 s + 30 s. k-1 t Present address: Biocenter of the University, Klingelbergstr. $ Author to whom reprint requests should be addressed. 111

(1) 7., CH-4056 Besel, Switzerland.

11. NOLl,

II:!

.-1.XU H. X01,1,

However, evidence for such an equilibrium together w&h a quantitativr evaluation of the parameters has been published only recently (Nell, 1972; Zitomer & Flaks. 1972; H. No11 et al., 1973). In our studies t’hat led to a new model of chain initiation (No11 & Noll, 1972,1974a; H. No11 et al., 1973) it became apparent that a quant,itabivc characterization of the equilibrium governing the association of subunits is crucial both for an understanding of the role of initiation factors as well as for the resolution of the controversial issues related to what has been calbd the ribosome cycle. In this paper we present measurements of two kinds: the position of t,he equilibrium and det’erminations of the rate of dissociation. as a function of the Mg2 + concentration We find that both parameters are strongly dependent on the Mg2+ concentrabion. Other studies from this laboratory showed that the affinity of the subunits for each other is profoundly influenced by the method of ribosome preparation as well. Moreover, the strength of association is direct,ly related to biological activity. Ribosomes that are most active in the translation of the natural messenger R17 RNA (tight couples; H. No11 et al.. 1973) are nearly all in the associated state at’ 5 mikf-Mg2 + and their rate of spont,aneous dissociation in t#he absence of initiation fact’ors is fairly slow, corresponding to a rate constant of 1.5 x 1W4 5-l. The experiments described in this report were carried out with ribosomes that were functionally homogeneous and could be converted nearly quantitatively into initiation complexes with R17 RNA as messenger (Noll, 1972; H. No11 et aZ.. 1973; M. No11 et al.. 1973a,h; No11 & Noll, 1974b).

2. Materials and Methods Most of the experimental 1973a).

details

have been given

in paper

I of this series (M. No11 et wl.,

(a) Reagents Buffer solutions (A to F) have been described (M. No11 et al., 1973a; No11 & Nell, 1974b). Buffer G is 20 mM-Tris*HCl (pH 7.5), 60 mM-NH,Cl, 10.5 mM-Mg acetate, 0.5 mM-EDTA (Staehelin & Maglott, 1971), 10 mM-2-mercaptoethanol. (b) Preparation (i) Pure

tight

vacant

of components

couples

Pure tight vacant couples were prepared from Escherichia coli strain Q13 middle log phase cells purchased as a frozen paste from General Biochemicals. Disruption of the cells and isolat,ion of the riboin buffer A (10 mm-Mg2+) with a Sorvall RMl cell fractionat,or somes by sedimentation of the S-30 extract, through a convex exponential sucrose gradient, (6 mM-Mg2+) in a B29 IEC zonal rotor have been described (M. No11 et al., 1973a). The ribosomes isolated from the 70 S peak (M. No11 et al., 19733, Fig. 1) were still contaminated a, with 3% of 50 S subunits (M. No11 et al., 1973a, Fig. 7(b)). For the final purification, second purification step in the zonal rotor was carried out, using 1000 A,,, units of the first step product in 20 ml of buffer A in which the Mg2+ was raised to 15 ITIM. The proin the gradient cedure was the same as in the first step, except that the Mg2+ concentration was raised to 15 InM and centrifugation was for 18 h at 25,000 revs/min. The resulting ribosomes were free of any detectable contamination wit)h subunits (No11 8: Noll, 19743, Fig. 5). (ii)

‘“C-labeled

subunits

E. coli strain MREGOO cells were grown in 1.26 1 medium containing 1.25 g Tryptone. 6.25 g Casamino acids, 6.0 g NaCl, 1.25 g NH&l, 2.0 g KH2P04, 22 mg uracil, and 16.25 g Tris. The pH was adjusted to 7.5 with cont. HCl, and 0.1 ml Antifoam was added. The

STRrCTURAL

DYNAMICS

OF BACTERIAL

RIBOSOMES.

V

113

following components were added after having been autoclaved separately: 6.25 ml 40::) (W/V) glucose, 1.25 ml 10 mM-FeCl,, and 0.30 ml 40% (w/v) MgSO+ Sterile [14C]uracil (0.3 mCi; New England Nuclear Corp.) was added together with the inoculum of 40 ml of a 6-h culture. The cells were harvested in log phase at A,,, = 2.4 and ground wit,h alumina (twice wet weight of cells) and the addition of 2 pg DNase (Worthington) per gram of cells. The S-30 extract was prepared according to Staehelin & Maglott (1971) using buffer A containing 15 mM-Mg 2+ . The extract, was layered over 32 ml of convex exponential sucrose gradients (c, t = 0.29 M, c, = 0.94 M, r;i = 30 ml) prepared according to No11 (1969). The gradients were made up with buffer A in which the Mg2+ was roduc& to 1 mM in order t.o cause complete dissociation of t*he ribosome couples into subunits. After centrifugation for 14 h at 4°C and 25,000 revs/min in an IEC SB-110 swinging-bucket rotor, the fractions corresponding to the 30 S and 50 S peak were cut out, diluted with an equal volume of buffer A and centrifuged for 21 h at 20,000 revs/min and 4’C in an IEC A170 angle-head rotor. The pellets were taken up in buffer A containing at, -65°C. Analytical gradient,s 15 mM-Mg2’ (A260 = 110) and stored in small portions shoncd no det,ectable contamination of the subunits with csch other (Noll, 1972, Fig. V.1). for the 50 S subunits. 7%~ spcxcific activity was 0.58 Ci/mmol for the 30 S, 1 .25 U/mmol (c)

of D, from

Determinatinn

in 70 S peak

radioa.ctizGty

It’ iV is the tot,al amount of non-radioactive 30 S particles added in the form of vacant coupl(*s and n is the amount# of non-radioactive 30 S subunits present at equilibrium, dissociation is defined by D, = n/N. After displacing the equilibrium toward association by adtlition of an excess n* of uctive [14C]30 S subunits, the radioactivity R entering the 70 R couples is diluted by the factor n*/(n,*-+n) and proportional to t,he fraction n present as 30 S sllhrmits at quilibrium and the specific act*ivity S of the wct,ivc [14C]30 S subunits:

Since t#he recovery of 70 S particles is incomplete because of t,he wall effect, we correct by t,he fact,or a = A2&nput/A2,,recovered. Empirical curves for the wall effect have been published (No11 85 Noll, 1974b). Since TL = A’D,, we obttlin aRn* 03 = jqK-ajj,. H is expressed in total and K fls azsO units.

cts/min

counted

under

t,he 70 S peak, S as cts/min

per A,,,

unit,

II *

3. Results (a)

Validity

Essential

requirements

equilibrium components

the

of measurements for

of equilibria the

has been reached 70 8

validity

by centrifugation of equilibrium

in Sucrose gradients measurements

at a given temperature

are (i) that

and concentration

of the

+ iMg2 + + 70 S.Mg:+

50 S + jMg2 + G-+50 S.Mg; + 30 S + 1Mg2+ = 30 S.Mg:+ and (ii) that,

once established,

the equilibrium

be not perturbed

by the measurement.

Since centrifugation through sucrose gradients was the technique used to determine the equilibrium, we had to choose conditions that prevented shifts in the equilibrium during centrifugation in order to satisfy the second condition. t C!,, sucrose concentration

at top of gradient;

C,, reservoir

concrntmtion;

V,, mixing

volume.

I 14

Al. NOLI,

;iNL,

H.

NOLI,

FixaGon of the equilibrium before analysis with agents such as formaldehyde or glutaraldehyde seemed undesirable, as we had no assurance that t’he iixat’ion react’ion was sufficiently fast bo prevent, shift’s in the equilibrium. Moreover. in our hands glutaraldehyde produced arbifacts such as association products and aggregat’es not8 originally present (h’oll &, Noll, 1974a). A general method t’o circumvent Obese difficulties is to determine t’he actual state of t’he equilibrium by the following equilibrium displacement labeling technique. An excess of 14C-labeled 30 S subunits active in couple formation is added to the equilibrium mixture of non-radioactive 70 S ribosomes in order to shift the equilibrium toward association. The 50 S subunits that had been dissociated are thus reconverted into 70 S couples by combining with radioactive 30 S partners. From the radioactivity transferred into the 70 S peak and stabilized by the high Mg2+ concentration in the gradient,, the fraction of 70 S couples that had been dissociated is t,hen computed (see Mat’erials and Methods). Computations according t’o mass action show that at low Mg2+ concenbrations (1.3 mM and 0% mM) addition of subunits produces less than lOO?A association. However, from the absence of 50 S subunits on the gradients, it is apparent that the actual shift is very close to 100%. This additional shift is caused by the high Mg2+ concentration in the gradient (15 mM). The only error that remains to be considered is the exchange at 0°C taking place during the interval of at most six minutes after the radioactive subunit’s have been added and before the high Mg2 + concentration in the gradient freezes the equilibrium. Equilibration with the Mg2+ concentration in the gradient) is expected to be instantaneous because the thickness of the sample zone (less than 1 mm) is negligible and ensures complete mixing during application. Measurements to estimate this error showed that the exchange at 2.3 mM-Mg2+ and 0°C is linear and reaches 6% after 40 minutes (k, = 2.5 x 10M5 s--l). In our experiments this corresponds to less than 1 o/0 exchange or an error of 14% for a II, value of 0*07 at 2.3 mM-Mg2 + and 0°C. For lower Mg2 + concentrations, this error is estimated to be negligible even under the most unfavorable assumption, namely that k, increases exponentially as we have shown to be the case for 37°C. For at the relatively high 11, values corresponding to these Mg2+ concentrabions, the rate of net incorporation of label into the 70 S part’icles becomes very slow. ln order to e&mate the error of II, at 37°C. we have to consider the additional exchange occurring from the time of addition of the subunits at 37°C until the mixture has been cooled to 0°C. A review of the relevant experimental details suggests a time equivalent of five seconds for exchange at 37°C as a realistic estimate. Thus, after addition of the radioactive subunits to the 70 S particles equilibrated at 37°C and the specified Mg2 + concentration, the reaction mixture (0.10 ml) contained in a thinwalled glass tube was instantly chilled in ice. With the onset of cooling the reaction rates decrease rapidly (at 2.3 mM-Mg 2+ t’he rates at 37°C and 0°C differ by a factor of 65). Taking five seconds as the upper limit, less than 10% of D, is due to exchange at 4 and 5 mM-Mg2 + . At lower Mg2 + concentrations, the exchange causes an error of less than 5% (calculated according to Table 3). The control experiments summarized in Figure 1 provide an evaluation of the critical parameters. Vacant couples free of subunits were dissolved at low Mg2+ concentrations (2.3 mM) and analyzed on gradients after 15 minutes at either 0°C (Fig. l(a)) or 37°C (Fig. l(b)). To p revent pressure-induced dissociation during centrifugation, the Mg2 + concentration in our standard gradients (60,000 revs/min

Sedimentatlan 70

50

value

(S)

30

(a) 2

I---

Effluent

valume

(ml)

FIG. 1. Evaluation of critical parameters in measuring true dissociation state by sedimentation through sucrose gradients. The incubation mixture (0.10 ml) contained (mix): HEPES buffer (Sigma Chemical Co.) pH 7.6 (50), ammonium acetate (50), magnesium acetate (2.3), 2-mercaptoethanol (l), pure vacant couples (0.51 A,,, units) and [14C]30 S subunits that had been activated at 37°C for 60 min (0.55 Azao units as indicated). Radioactive 30 S subunits were added at 0°C to (c) at, and (d) and at 37°C to (e) and (f). (a ) and (c) were kept at O”C, (b) and (d) were equilibrated 37”C, then chilled before addition of subunits to (d); (e) and (f) were incubated at 37°C for 5 ant1 15 min, respectively, and, after addition of subunits, immediately chilled to 0°C and loaded on the gradients. Centrifugation was for 1.2 h at 60,000 revs/min and 4°C in standard sucrose gradient,s. Note that the radioactivity scale is reduced by a factor of 4 to the right of the break. 14C radioactivity; ----, absorbance at, 260 nm. A>-c--,

at 4°C) was raised to 15 mM (EC. No11 et al., 1973; M. No11 et al., 1973a). As evident’ from the results in Figure 1, no dissociation was observed, unless equilibration took place at 37°C. The same results were obtained if an excess of [14C]30 S subunits was added to each sample after incubation and chilling (Fig. 1 (c) and (d)). The absence of subunits in Figure l(a) implies that the rate of thermal dissociation is too slow in the cold to cause measurable concentration-dependent dissociation as the particle zone moves through the gradient. The effect of temperature on equilibration is evident from the fact that dissociation (I), = 0.07) occurred only after incubation at 37°C

Jl.

llti

X01,1,

ANI:,

H. NOLI,

(Fig. l(b)). We found t,hat equilibration was complete after five minut,es at, 37 ‘c’. since there was no further change in dissociation after 15 minmes (Fig. l(e) and (f )). The results in Figure 1 and Table 1 also establish that the state of dissociation at 37°C) is not significantly (5 % ) produced at 2.3 mM-Mg’ + and 0°C (after equilibration in the gradient during centrifugat’ion in t,he reversed by the high Mg2 + concentration cold (4°C). A slightly higher value (7%) was obtained by measuring the radioactivity transferred into the 70 S peak after displacing the equilibrium with an excess of is result is validated by the control (Fig. l(c)) [14C]30 S subunits (Fig. l(d)). Th showing only about 1% exchange on the gradient) when the 30 S particles were added in the cold. TABLE

Equilibrium

Incubation of 70 S couplest Time (mm) 15 15 5 15 15 15

0 37 37 37 0 37

Addition of 30 St

0 0 37 37

1

at 2.3 rnM-Mgzt Radioactivity under 70 S peak after subtraction of background (cts/min) 70 400 1700 1600 -

Fig.

0.01 0.07 0.34 0.32 co.01 0.05

t 70 S couples (0.51 A2a0 units) in 0.0030 ml of buffer G (10 mM-Mg2+) were diluted to 2.3 mix-Mg2 + (see legend to Fig. 1) and incubated as specified. At the end of incubation [‘%I30 S subunits (in 0.0050 ml of buffer G (10 mM-Mg2+) ) were added, the mixtures (0.100 ml) chilled immediately to O”C, and 0.090-ml portions analyzed in sucrose gradients (15 m&r-Mg2 + ).

However, the necessity to “freeze” the equilibrium during analysis by performing the centrifugation at temperatures close to 0°C renders the method inapplicable to the study of equilibria at higher temperatures. For, as solutions equilibrated at, a higher temperature are chilled, the equilibrium is shifted considerably toward association. Thus, in the experiment shown in Figure 1 the dissociation increased from 7% to 33% when the excess of [14C]30 S particles was added at 37”C, immediately before chilling (compare Fig. l(d) with (e) and (f). For this reason all our measurements were carried out by the displacement labeling technique. Determinations of dissociation by the method described here are reproducible within &5% and the accuracy remains high over a wide range because the quantity measured is net transfer of radioactivity rather than a difference in the number of particles, as in the case of measurements by light-scattering. Thus, it is obvious from Figure 1 that dissociations in the range of a few per cent can be determined with the usual analytical precision, provided the equilibrium is established by starting with vacant couples that are free of 50 S subunits. By contrast, we found that determinations by light-scattering gave much less reproducible results and the shape of the dissociation curve at either end was not well-defined (cf. Fig. 2, curve 6).

STRUCTURAL

DYNAMICS

(b) Dissociation

equilibrium

OF BACTERIAL

as a function

RIHOSO~LES.

of Mg’+

V

117

concentration

Vacant tight couples free of subunits were equilibrated at 37°C and at Mgz+ concentrations ranging from 0.8 mM to 5.3 mM. After addition of an excess of [ 14C]30 S subunits, the mixtures were chilled and analyzed on sucrose gradients. Similar experiments measured the dissociation at 0°C. In this case the radioactive, 30 S subunits were added after chilling the mixtures that had been equilibrated at 37°C. The sedimentation patterns (H. No11 et al., 1973, Fig. 6) showed the expected transfer of radioactivity from the 30 S to the 70 S peak that increased as the Mg2+ concentration was lowered. The absence of a peak corresponding to free 50 S subunit~s implies that displacement is close to 100% association. The dissociation at equilibrium (D,) was computed from a quantitative evaluation of the gradient! data (SW Materials and Methods). If (C,,), is the initial concentration of pure tight couples ant1 (C,,),, (C,,),, etc. are the corresponding equilibrium concentrat,ions, k

= Fkt),/(~,,),

[Mg’+]

= (~30MGJ0~

(mh+)

FIG. 2. Dissociation equilibria of tight and loose couples as a function of the Mg2 + concentration. Equilibria of tight couples at 1.08 x lo-’ M ribosome concentration and 0°C (1) and 37°C (2) determined by displacement labeling and sedimentation as described in the text. (For further experimental details see Nell, 1972.) The dotted curve was calculated for (2) under the assumption that log K: vewua log [Mge+] falls on a straight line (see Fig. 4). (3) Mixture of tight and loose couples at 20°C. (4) Loose couples (data corresponding to (3) and (4) from Spirin et al., 1971). (5) and (6) Loose couples at 24°C (5) and 37°C (6) as determined by light-scattering (Zitomer 8r Flake, 1972).

A plot of D, against the log of the Mg2+ concentration (Fig. 2, curves 1 and 2) shows that below 3 m&r-Mg2 + the equilibrium is strongly dependent on the temperature and very sensitive to small changes in the Mg2+ concentration. The narrow range of dissociation (60% of the particles dissociating between 1.5 and 2.5 mM-Mg2+ at, 37°C) implies that t,he population is fairly homogeneous.

11s

AI. NOLI,

;\NI)

H. NOLI,

Although the calculat’ions assume that t,he act,ivc 30 Is particles added and t host derived from the 70 S couples by dissociation are equally competent to form couples~ we have taken into account that not all 30 S subunits were active. ‘Thus, we determined in control experiments that 507; of t’he radioactive particles were able to form couples when titrated to saturation wit’h an excess of 50 S subunits. It is obvious that the calculated D, values would be too low if the non-radioactive 30 S subunits derived from dissociation were favored, and too high if the labeled 30 S subunits added had a better combining affinity, the potential error being greatest’ for high I), values. Since our data, calculated on the basis of equal chance, show loo)/, dissociation at 37°C and 1 mM-Mg2+, it follows that the values cannot be too low and hence that the unlabeled endogenous 30 S subunits are not favored over their radioactive competitors. Nor is it likely that’ the values are too high, because ib would obviously make no sense to assume that the 30 S particles added as radioactive subunits have a higher combining affinity than those originally present in couples. An overestimation that might result from exchange at 0°C is small as discussed earlier and illustrated by the small amount of radioactivity in the 70 S peak of Figure l(c). For a given Mg2+ concentration, the association (K,) and dissociation (K,) constants and the rate constants for dissociation (k,) and association (Iz- 1) are given by t’he mass action equilibrium K&oh Ka

=

1'Kd

=

k-1 =

(C,,),(C,,),

k,

.

(4)

If we dissolve vacant couples free of subunits, the equilibrium concentration (C,,), is related to the initial concentration (C,,), at a given temperature and Mg2 + concentration: (Go), and Combining

= Go)0

-

V5oh = Go),. equations

(C5o)es

(5)

(‘3)

(3) to (6), we obtain (7)

or

(8)

Equations (7) and (8) show that the extent of dissociation is strongly dependent on the ribosome concentration. For illustration, we have plotted in Figure 3 the dissociation curves calculated from Figure 2 according to equations (7) and (8) for concentrations of (C,,), ranging from 0.01 to 100 times that measured at 37°C (C,). At an average growth rate, the intracellular ribosome concentration corresponds to about 10n5 M (Maalrae & Kjeldgaard, 1966). Since in our experiments the ribosome concentration was 10-l M, the dissociation curve for C, would reflect the intracellular distribution if 1% of the ribosomes was present in the free state. Assuming an upper limit of lo%, we conclude from Figure 3 that more than 90% of the vacant ribosomes would be present in the associated form at a Mg2 + concentration of 2 3 mM. In Figure 4 the log of Ki has been plotted against the log of the Mg2+ concentration for measurements at 0°C and 37°C. A linear relationship is observed at 37°C between

I.0 . I

I

I

I

I

I

I

I IO

.

00

0.0

I

2

3 [Mg*+]

4

5

6

(rnM)

Pm. 3. Dependence of dissociation equilibrium on ribosome concentration. 2. The values for C, = I.08 x 10e7 M are the same as in Fig. 2, curve concentrations ranging from 0.01 C, to 100 C, were calculated using values for G, as basis.

The experimental curves for the other on the smooth rurvr

II -

IO -

9-

6-

5-

[Mg’+](m~l

PI<:. 4. Non-linearity of plot log h’* ?)WSZI.Slog [i%fg2+ 1. h’i was computed from I>, valws in Fig. 2 according to eqn (7). The solid lines were tletermined from the smooth curves in Fig. 2. .--(I!-.----<:-, Tight couples 0°C; -@---a--, tight c~>uples 37°C; --m-m-, loose cwu~drs (Zitomtr & Flaks, 1972) at 24°C or 37°C (-).

IL’I,

31. NOLI,

.4X1)

H.

NOLI,

I .x and 5.3 mfir-Mg 2 + . To test’ whether t,he deviat’ion fkm lirwarit,y below 1.X mnq-Mg” + is significant, t’he dissociation curve De in Figure 2 has been replot’ted (curve 2. dotted line) under t#he assumption t,hat all log Ki values lie on the straight, line in Figurcb 4. It should be pointed out that in our measurements the error is larger near complete dissociation than near the association end of the curve. Thus, at two points close t,o either end of the dissociation curve (e.g. 7% and 920/6), an error of -+ 6% in the determination of the radioactivity under the 70 S peak would result in uncertainties Accordof dissociation corresponding to 7% -&- 0.5% and 92% A- 70/b, respectivelyj-. ing to this estimat,e, t’he deviation from linearity appears to be real. Evidencr for a non-linear relationship is much stronger at O”C, where the points describing the Mg2 + dependence of Ki at 0°C fall on a curve that exhibit,s a concave curvature at’ low In t’his case the accuracy is sufficient t,o rule out, a linear Mg2 + concentrations. relat#ionship. Since lowering t’he temperature makes dissociation measurable at’ lowtar Mg2 + concent(rations. it, seems likely t’hat the curve for t*hc Mg2 + dependence of K, at 37°C deviat’es in a similar manner from lineariby at, low Mg2 + concentrations. (2). the following If Mg2+ interacbs with the subunibs according to equation relat,ionship can be derived under the assumption of identical and non-interacting binding sit’es (for derivat’ion, see Appendix I) : d In K,/d n-here ?a = average Integration yields

number

In [Mg2 +] = dn,

of Mg 2+ bound

by the respective

(9) ribosome

[70 a log K, = log [50~,,,~~ s,, = An log Wg2+l + 1s K,,

species.

(l(J)

in which the integration constant K, corresponds to the association constant at’ a M. In this treatment Mg2+ is considered a reactant that concentration of 1 Mg2 + cnt#ers into the mass action equilibrium in proportion to the power An of its concenand describes the tration. By contrast, K, does not contain the Mg2+ concentration is supplied as a parameter with no equilibrium only if the Mg2+ concentration assumptions about its mathematical relation to K,. Applying equation (10) to the linear portion of the curve for dissociation at 37°C in Figure 4, we obtain An = 10 (slope) and K, = 1.6 x 1O32 M-l1 (ordinate intercept). For comparison, we have replotted in Figures 2 and 4 values published by Spirin et al. (1971) and by Zitomer & Flaks (1972). The ribosomes examined in our study are much tight’er than those of the other authors, as indicated by the shift in the at 500/, dissociation from about 6 mM to 2 mM-Mg2+ (Fig. 2). It Mg2 + concentration is further evident from Figure 4 that at 4 mM-Mg 2+ the association constant Kk of t,ight couples is nearly five orders of magnitude greater than those of the loose couples studied by Zitomer & Flaks and by Spirin et al. In addition to these data obtained with fairly homogeneous populations of tight and loose couples, the unwashed ribosomes examined by Spirin et al. exhibit a biphasic dissociation curve (curve 3 in Fig. 2) that spans the entire range and obviously reflects the dissociation of a mixture consisting of tight and loose couples. Of particular interest is the finding that above 50% dissociation the log-log plots in Figure 4 exhibit significant deviation from linearity. This implies that below this 7 Errors of D, are amplified in K: according particularly large if D, is close to 1.

to eqn (7), i.e. by at least a factor of 2, and become

STRUCTURAL

DYNAMICS

OF

BACTERIAL

RIBOSOMES.

V

121

the binding of Mg 2+ is co-operative rather than indepencritical Mg2 + concentration dent. However, the significance of the apparent linearity in the region of incipient. dissociation is equally doubtful, and this criticism also applies to the interpretation of bhe slope suggested by the oversimplified analysis in equation (10). Thus, the data given by Zitomer & Flaks show widely different slopes for t’he dissociation curves determined by light-scattering (An = 7.5) of non-fixed ribosomes or by sedimentat,ion analysis (An = 4.3) of formaldehyde-fixed ribosomes. Other evidence from this laboratory established that exposure to Mg2 + concentrat’ions below that corresponding to 50’;/, dissociation leads very rapidly at 37”C, and very slowly at 4”C, to irreversible conformat8ional changes in t’he 50 S subunit that, are responsible for the conversion is lowered, an from tight to loose couples. It follows that as the Mg2+ concentration increasing proport)ion of tight couples are converted to loose couples during equilibration at’ 37”C, wit’h a corresponding change in equilibrium toward dissociation prior tjo analysis on the gradient. As a result’, the measured I), values in t,his region would be too high, a fact which would make t,he actual deviat’ion from linearity in the log-log plot) of Figure 4 even greater. This would explain why the dissociation curve at, 37°C is not parallel to that at 0°C. Nevertheless. the overall pict,ure would be consistent with the conclusion that Mg 2 + binding in tight couples is largely co-operative and becomes largely but not completely independent upon conversion t,o loose couples.

(c) Thermaiynamics of the equilibrium Because of the profound effects of Mg2+ on ribosome structure, it seems more instead of a realistic to consider Mg2+ a parameter in the mass action equilibrium participant, as proposed by Zitomer & Flaks (1972). Hence we computed for each the standard free energy change (AGO) and the corresponding Mg2 + concentration changes in standard enthalpy (AH”) and the entropy t’erm (T AS”) from K: according t,o the equations

AC0 = -~-RT 111Kg = AI1 --- T AS-’ and

AHo = -R ln KL(Td -1/T,

(11)

- ln K&CT21 l/T,

The results are summarized in Table 2 and compared with t’he data given by Zitomer & Flaks (1972). The Table shows that tight couples exhibit less negative AH” and AS” t,han the loose couples. These findings would be consistent with the interpretat’ion that’ association of tight couples involves less of a shape change and more participation of close range forces (hydrogen or hydrophobic bonds) between complementary components of the interfaces than in the association of loose couples. Thus, while an increase in Mg 2+ binding can largelv compensate for the loss of this primary interaction, the struct’ural alterations responsible for t’he formation of loose couples have changed the quality of the subunit interaction, as evident from the much greater susceptibility to t,hermal dissociation. (d) Kinetics of equilibrium For a more complete description of the equilibrium, we need to know one of the rate constants in addition to the dissociation constant. The rate of spontaneous dissociation, k,. was determined at 37°C and several Mg2 + concentrations from the

T4HTE .I J 2 xg z + -deplderze

of cha,uges in free ewrgy, Pjatropy wd rw~,thalpy tlurirrg dissociatiorL of tight and loose rihosom,e couples T AS”

AH”

(kcal/mol) Tight couples I .o 1.4 1.8 2.2 2.6

Loose couples 4.0 6.0 8.0 10 15 20

0°C -8.4 -9.2 -10.0

37°C -8.1 -9.2 - 10.2

-11.1

-11.1

0

~ 12.9

-12.0

-6.7

- 7.6

37°C -8.1

3 1°C’ -37

37°C ~ 37

.-9.6

--“X

--2x

- 10.5 -11.2 ~ 12.8 - 14.1

-32 -49 -58 -- 76

-33 -- 50 -59 -77

31°C -8.X -.. 10.2 -11.1 - 12.2 - 13.9 -15.6

O’C -2..5 -0.6 -1.1.5

(kcal/mol) 37°C -2.8 -0.6 -I- 1.7

-11 --9.8 ~ 8.5

0

-~11 -20

-~45 -38 --43 --62 -71 g,

initial rate at which pure labeled 50 S particles appeared in the 70 S ribosomes after they were mixed with purenon-radioactive vacant couples. Samples removed from the reaction mixture were analyzed on isokinetic sucrose gradients (H. No11 et al., 1973, Fig. 8). In all of these exchange experiments the purity of the particles is essential, for contamination with the complementary subunit would lead to the immediate formation of radioactive couples and hence to a large zero time background and corresponding error. In our experiments no radioactivity was present in the 70 S peak at zero time. The time-course of exchange at 37°C and 4*1,5*1 and 56 m&r-Mg2 + is illustrated in Figure 5. The experimental values follow closely the theoretical curve (solid line) representing

the solution

of the differential

equation

describing

the exchange

process

(see Appendix 2) :

Go)0+ (*c,o)o(*GcJt= 1 _ exp --~Vmh + (*G3)qkt (*~,,)o (*G3& (Go), l .>. (

(13)

The rate constant of dissociation

t:o was determined from the slope corresponding to the initial rate of incorporation of [14C]50 S subunits into ribosome couples, (dC,,/dt),,,, and the initial concentrations of vacant couples, (C,,),. An additional value for k, was obtained at 2.3 mm-Mg2 + . At this concentration the measurement of the exchange rate is complicated by the fact that 25% of the ribosomes are dissociated and addition of the radioactive 30 S subunits causes a shift toward association from 75% to 95.5%.

STRUCTURAL

DYNAMICS

OF BACTERIAL

RlI~OS0MES.

1

123

lncubatlontame(mln) FIG. 5. Time-course of exchange of tight vacant couples with radioactive 50 S subunits at 37°C. Incubation mixtures in HEPES buffer (as indicated in legend to Fig. 1) containing pura vacant couples (0.60 Azao units), pure [‘%I50 S subunits (0.46 Azao units), GTP (0.5 mw), and Mg2 + (as specified) were incubated at 37°C for the times shown on the abscissa. Incubation was followed by sucrose gradient analysis under standard conditions ( 15mM-Mg2+ ). The computed effective Mg2+ concentration (4.1 mM --@-a--, 5.1 mM-A----A--, and 5.6 mM --m---m-) takes into account that the vacant couples and [14C]50 S subunits together contribute 0.6 mMMga + and that GTP binds equimolar amounts. The [‘%I50 S subunits were 50% active in couple format,ion; however, they consisted of only 33% 50 S a-particles, while 17% were 50 S b-subunits (H. No11 et al., 1973). In a first approximation, only 50 S a-subunits have to be considered to take part in the exchange reaction as their affinity is much higher for the 30 S subunits (Figs 2 and 4). The radioactivity under the 70 S peak is plotted against incubation time. The values on the ordinate on the left are not corrected for the wall effect, whereas those on the right are corrected and given in the units defined in Appendix 2. The kl values were determined from the slope of the initial rates or computed from the theoretical curves (eqn 13) selected according to the best fit (solid lines). (A) [WI50 S (cts/min x 1.08 x 10d3); (u, 0) [WI50 S (cts/min x 10-3).

Our estimate of k, was obtained from an extrapolation of the exchange curve measured after the shift. The theoretical curve (Appendix 2) x0 x = 1 - exp ( - ~~~~~ k, (t -t t,)) -__+ Yo-~ ?/o x0 was fitted through the experimental points at five and ten minutes. The initial rat,e k, (1.6 x 10e3 s-l) was then determined from the slope of this curve at t = - t,. Evidently this curve has the positive ordinate intercept of 1 - exp ( - ~ xc, + ‘a k,t,) YO and the negative intercept on the abscissa of t = -to corresponding to the virtual start of the exchange. This estimate hinges on the validity of the assumption that the new equilibrium after addition of the radioactive subunits had been reached at the times the exchange was measured. An estimate of this equilibrium time was obtained by numerical solution of the coupled differential equations describing the exchange. By computer it was found to be less than two minutes (A. Wishnia, personal communication) . Figure 5 shows that at 5 mM-Mg 2+ tight couples dissociate rather slowly with a 50% exchange time of 15 minutes. As expected, the rate of exchange increased

I”4

111. NO1,L

ANI)

H.

SOLl,

FIG. 6. 14C-labeled 30 S subunits (0.55 &a0 units in 0.0050 ml) that were found to be 50% active in couple formation were added to unlabeled 70 S ribosomes (0.51 Azao units in 0.095 ml of standard buffer containing 1.8 mM-Mg2*) and equilibrated at 37°C resulting in a final Mg2+ concentration of 2.3 mM. The exchange was terminated and measured by quick chilling of the samples to 0°C after 0, 5 and 10 min, and immediate loading on sucrose gradients containing 15 mivr-MgZ+. Determination of k, is explained in the text. The values at 4.1, 5.1 5.6 ITIM Mgz+ were obtained from the experiments described in Fig. 5.

greatly as the Mg2 + concentration was lowered. Washing of ribosomes with high concentrations of monovalent ions (1 M-NH&I) also increased the rate of exchange (Hapke & Noll, unpublished results). It is evident from the semi-logarithmic plot of the rate constants versus Mg2+ concentration that the rate of dissociation increases exponentially as the Mg2+ is reduced from 6 to 2 mM. For this range of Mg2+ concentrations we are now able to compute k-, according to equation (4), using the values for K6 (Fig. 4) and kr (Fig. 6). The values for Ki, Ic, and k _ 1are listed in Table 3. It is evident that the Mg2+ concentration affects the rate of dissociation and associa-

TABLE 3 Dependence

De

of rate constants of dissociation (k,) and association tight couples on magnesium ion concentration at 37°C Mg2 + concn (md

0.95

1.6

0.40 0.085 0.020

2.0

0.008 0.004

3.0 4.0 5.0 6.0

(M-IX

k-1 (PI-1 s-1 x lo-@)

K:, 1o-g) 0~00051 0.035 1.2 23

(kFI) of

54 2.2

0.0028 0.077

0.91

1.1

0.37

X.5

110

0.15

700

0.062

17 43

All values are taken from the smoothed curves (solid line) in Figs 2, 4 and 6. The values for k, shown above are lower by about a factor of 2 than those published previously (Noll, 1972; No11 & Noll, 1972; H. No11 el a.Z., 1973) because we used a wrong value for (*C,,),, in equation (13), even though the correct value had been measured (see legend to Fig. V.2 of Noll, 1972).

STRUCTURAL

DYNAMICS

tion in opposit$e ways and that exponential in the range studied.

OP BACTERIAL

the concentrabion

RIBOSOMES.

dependence

V

1%

is approximately

4. Discussion If the Mg2 + concentration is kept above 10 mM, over 85% of the ribosomes isolated from fresh extracts of bacteria harvested by slow cooling are in the form of what has been called tight vacant couples (H. No11 et al., 1973). Other work from this laboratory (summarized by H. No11 et al., 1973) has shown that these ribosomes are the most active biologically and come closest to what must be regarded as the native state. At8 37°C and 10 mm-Mg2+ such ribosomes are all in the associated state and no measurable exchange with subunits is observed (No11 & Noll, 1972). At 5 mM-Mg2+, a slow exchange takes place although most of the ribosomes are in the associated form. Further reduct’ion in the Mg2+ concentration produces a progressive shift in the equilibrium toward dissociation. The negative enthalpy change suggests that dissociation is accompanied by conformational changes. Other evidence implies that these changes are reversible above and irreversible below a critical Mg2 + concentration (z 1.5 mM) corresponding to about 50% dissociation. Thus, it, has been found that above the critical concentration, incubation at 37°C very slowly converts tight into loose couples (H. No11 et al., 1973; Hapke & Noll, 1976) whereas below 1 to 2 mM-Mg2+ this conversion rate increases very rapidly (Stahli, 1975; M. Noll, unpublished observation). It follows from these studies that our equilibrium measurements are unaffected by this denaturing process only above about 1.5 mM-Mg2 + . Below this Mg2 + concentration. t,he dissociation values tend to be too high because of partial denaturation. This t’ransition from tight to loose couples is caused by an irreversible structural change of the 50 S subunit (H. No11 et al., 1973) that appears to be conformational (Stahli, 1975). Previous reports that in experiments with unwashed ribosomes reassociation required higher Mg2 + concentrations than dissociation (Spirin et al., 1971: Zitomer & Flaks, 1972) are explained by the denaturat,ion of tight couples in preparations containing various proportions of both forms. It’ is reasonable to postulate that the subunits are held together by hydrogen and/or hydrophobic forces that are strongly co-operative and require the precise alignment, of the complementary surfaces, and that the role of Mg2 + consists in the electrostaGc shielding of the repelling negative charges (Walters & Van OS, 1970) of the phosphate groups on the RNA backbone. It follows that if the contribution of the co-operative interface bonds is large (tight couples), less Mg2 + is required to maintain a given degree of association, and dissociation as a function of the Mg2+ concentration should reflect, mostly co-operative behavior. On the other hand, the model would explain why reduction of interface bonds by irreversible conformational changes (loose couples) coultl be compensated by a more efficient electrostatic shielding at higher Mg2+ conccbntrations. Consequently, the dissociation curves of such particles should exhibit less co-operative behavior. Indeed, above the Mg2+ concentration corresponding to 50(?;, dissociation, the Mg 2+ dependence of association (K,) is proport’ional to 1/p&2 + p. consistent with a model in ,which the Mg 2+ binding sites are identical and non-interacting. It should be emphasized, however, that despite this limited formal agreement with the simplified Wyman equation (Appendix l), the underlying concc>pt fails to take into account the important secondary effects of Mg2 + binding on ribosomc~ st,ru&ure. For this reason we have t,reat,ed Mg2 + as a parameter and we

120

111. NOLL

BND

H.

NOLL

believe that the model proposed by Zitomer dz Plaks (1972), which considers Mg”+ as a reactant, is inadequate. The fact that higher Mg2+ concentrations restore the association equilibrium of loose couples has long obscured the profound difference between the two forms. Thus, tight couples seem to have a more rigid internal structure that is reflected in their greater resistance to thermal dissociation (Pig. 2 and Table l), in their higher affinity for tRNA binding and in their strict dependence on elongation factor G and GTP for translocation (H. No11 et al., 1973). Removal of Mg2 + reduces the association constant by weakening the internal structure, at first reversibly, t,hen irreversibly by allowing rearrangements that would tend to stabilize the 50 S subunit in the b-conformation. The resulting loose couples are characterized by a greater plasticity that manifests itself in the ability to undergo spontaneous translocation (H. No11 et al., 1973) and in a reduced resistance to thermal dissociation. If we consider that the ribosome consists of more than 50 protein subunits and three large RNA molecules, it is not surprising that such a complex system can exist in a number of different states. Indeed, many of the specific functions, such as translocation and initiation, seem to require reversible transitions between various states (Schreier & Noll, 1971). An important aspect of ribosome function is the efficient control of proteins (initiation and elongation these transitions by a family of extraribosomal factors) as well as tRNA that act as allosteric effecters. It is obvious that the properties of tight couples make more sense in such a concept than those of loose couples. (N 4 mM) loose couples are mostly Thus, while at physiological Mg2+ concentrations dissociated, tight couples are nearly all associated and exchanging their subunits only slowly. Since we have already shown that each initiation is preceded by a dissociation event (No11 & Noll, 1972), the present results explain the need for factors that stimulate dissociation (initiation factors 1 and 3) and shift the equilibrium toward dissociation that prevents denaturation. It is better (initiation factor 3) at a Mg2+ concentration to open a safe with keys than with a crowbar, especially if you want to put something in. After this paper was submitted, two papers on the same subject appeared: Wishnia et al. (1975) and Debey et al. (1975). Although the results of these authors are generally similar to ours, it should be pointed out that their ribosomes are at least partially denatured, as evident from their lower K; and higher k, values, as well as from the absence of hysteresis on reassociation. The variability and reduced tightness of their preparations (50% dissociation ranging from 2.5 to 3.4 mM-Mg2 + at 25°C) is attributable to the denaturing effect of washing with 1.5 M-NH:. Denaturation always results in a shift of the 50% dissociation point to higher Mg2+ concentrations, and hence the hysteresis seen on reassociation is a manifestation of tight rather than loose couples, contrary to the statement made by Wishnia et al. (1975). The results reported in this paper have been published previously as part of a doctora, thesis: M. Noll, Northwestern University, 1972. The authors are grateful to Drs I. M. Klotz and C. Stahli for critical comments and stimulating discussions. This work was supported by research grant no. P381F from the American Cancer Society and grant no. 5ROlCA-11797 from the National Institutes of Health, United States Public Health Service. Professor of the American Cancer Society.

One of the

authors

(H.

N.)

is a Career

APPENDIX

127

1

REFERENCES Debey, P., Hui Bon Hoa, G., Douzou, P., Godefroy-Colburn, T., Graffc, M. S: GrunbcrgManago, M. (197.5). &ochemietry, 14, 1553-1559. Hapke, B. & Noll, H. (1976). J. Mol. Biol. 105, 97-109. Maaloe, 0. & Kjeldgaard, N. 0. (1966). In Control oj’ illacro?noEecct~r Synthesis, 1’. 70, W. A. Benjamin, Inc., New York. Noll, H. (1969). In Techniques in Protein BiO8ynthe8i8 (Sargent., J. & Campbell, P. N., eds), vol. 2, p. 101, Academic Press, London. Noll, H., Noll, M., Hapke, B. & van Dieijen, G. (1973). In Regulation of Transcription and Tranakxtion, MoabachColloquium No. 24 (Bautz, E.. ed.), pp. 257-311, Springer Verlag, Heidelberg. Noll, M. (1972). Thesis, Northwestern University. Noll, M. & Noll, H. (1972). Nature New BioZ. 238, 225-228. Noll, M. & Noll, H. (1974a). J. Mol. BioZ. 89, 477-494. Noll, M. & Noll, H. (19746). J. Mol. BioZ. 90, 237-251. Noll, M., Hapke, B., Schreier, M. H. & Noll, H. (197%). J. Mol. B&Z. 75, 281-294. Noll, M., Hapke, B. & Noll, H. (1973b). J. MOE. BioZ. 80, 519-529. Schreier, M. H. & Noll, H. (1971). Proc. Nat. Acad. Sci., IT.S.A. 68, 805-809. Spirin, A. S., Sabo, B. & Kovalenko, V. A. (1971). FEBS Letters, 15, 197-200. Sitaehelin, T. & Maglott, D. R. (1971). In Methods in Enzymology, Nucleic Acids and Protein Synthesis (Moldave, K. & Grossman, L., eds), vol. 20, part C, pp. 449-456, Academic Press, New York. Stahli, C. (1975). Thesis, Northwestern University. Walters, J. A. L. I. & Van OS, G. A. J. (1970). Biochim. Biophys. Acta, 199, 453-463. Wishnia, A., Boussert, A., Graffe, M., Dessen, P. & Grunberg-Manago, M., (1975). J. Afol. Biol., 93, 499-515. Zitomrr, R. S. & Flaks, J. 0. (1972). J. Mol. Biol., 71, 263-279.

APPENDIX

Derivation

1

of equation (9) from main paper I.M. KLOTZ

Since both subunits and the 70 S association product are binding Mg2 + reversibly, the following equilibria and association constants apply:

50s + iMg2+ + 5OSMgi KsOSM~ 30 S +jMg2+ = 3OS%, K,,,,, 70s +ZMg2+ = 7OSMg, Kmm .

(Al)

The multiple equilibria between ribosomal species can be represented concisely by the equation “70 %a % LX BOSMg, + YZ SOSMg, + Z 70SMg, K, , W4 0

0

0

where ‘n = total sites available on each respective species of ribosome, and K: is the