Macromolecular Crowding - The Journal of Biological Chemistry

THE JOURNAL OF B~JLOGICAL CHEMISTRY 0 1990 by The American Society for Biochemistry

“Macromolecular Protein-Protein

Vol. 265, No. 25, Issue of September 5, pp. 15160-15167, 1990

and Molecular Biology, Inc.

Crowding”: Interactions

Printed in U. S. A.

Thermodynamic Consequences for within the T4 DNA Replication Complex* (Received for publication,

Thale From

C. Jarvis$, the Institute

Dawn of Molecular

M. Ring, Biology

S. Daube, and Peter H. von Hippelg and Department of Chemistry, University of Oregon, Eugene,

December 29, 1989)

Shirley

In vitro biochemical assays are typically performed using very dilute solutions of macromolecular components. On the other hand, total intracellular concentrations of macromolecular solutes are very high, resulting in an in viva environment that is significantly “volume-occupied.” In vitro studies with the DNA replication proteins of bacteriophage T4 have revealed anomalously weak binding of T4 gene 45 protein to the rest of the replication complex. We have used inert macromolecular solutes to mimic typical intracellular solution conditions of high volume occupancy to investigate the effects of “macromolecular crowding” on the binding equilibria involved in the assembly of the T4 polymerase accessory proteins complex. The same approach was also used to study the assembly of this complex with T4 DNA polymerase (gene 43 protein) and T4 single-stranded DNA binding protein (gene 32 protein) to form the five protein “holoenzyme”. We find that the apparent association constant (Km) of gene 45 for gene 44162 proteins in forming both the accessory protein complex and the holoenzyme increases markedly (from -7 x lo6 to -3.5 x 10’ M-‘) as a consequence of adding polymers such as polyethylene glycol and dextran. Although the processivity of the polymerase alone is not directly effected by the addition of such polymers to the solution, macromolecular crowding does significantly stabilize the holoenzyme and thus indirectly increases the observed processivity of the holoenzyme complex. The use of macromolecular crowding to increase the stability of multienzyme complexes in general is discussed, as is the relevance of these results to DNA replication in rho.

The replication of bacteriophage T4 DNA depends on the cooperative interaction of a number of T4-coded gene products to form a multi-protein DNA replication complex. Numerous in vitro studies have resulted in a substantial understanding of the specific roles of the individual components of the replication apparatus (e.g. see Nossal and Alberts, 1983; Cha and Alberts, 1988; Jarvis et al., 199Oc). Here we examine the thermodynamic (and kinetic) consequences of macromo* This work was supported in part by United States Public Health Service (USPHS) Research Grants GM-15792 and GM-29158 (to P. H. vH.) &d by aUSPHS Institutional Research Service Award GM07759 predoctoral traineeship (to T. C. J.). This work was submitted (by T. C. J.) to the Graduate School of the University of Oregon in partial fulfillment of the requirements for the Ph.D. degree in Chemistrv. The costs of publication of this article were defrayed in part by the”payment of page charges. This article must therefore be-hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Present Address: Dept. of Molecular, Cellular, and Developmental Biology, University of Colorado, Boulder, CO 80309. § An American Cancer Society Research Professor of Chemistry.

lecular crowding on protein-protein interactions within the replication

Oregon

97403

and protein-nucleic acid complex.

The proteins that comprise the five-protein “core” of the T4 DNA replication system include the DNA polymerase (gene 43 protein), the T4-coded single-stranded binding protein (gene 32 protein), and three proteins known collectively as the polymerase accessory proteins, encoded by genes 44, 62, and 45. The accessoryproteins exist in solution as discrete homogeneous sub-assemblies: the gene 44/62 proteins as a tightly associated complex containing four gene 44 protein subunits and one gene 62 protein subunit, and the gene 45 protein in the form of trimers (Jarvis et al., 1989a). These proteins interact to form a complex that increases the processivity of the DNA polymerase (i.e. increases the number of nucleotide residues incorporated per association event of the polymerase with the primer-template) via a mechanism that requires the hydrolysis of ATP (Piperno and Alberts, 1978; Nossal and Peterlin, 1979; Newport et al., 1980; Mace and Alberts, 1984; Jarvis et al., 1989b, 1990). In vitro studies involving gene 45 protein have frequently revealed that the binding interaction between gene 45 protein and the other replication proteins is surprisingly weak (i.e., the binding affinity of gene 45 protein for the complex is far less than stoichiometric); in earlier studies from this laboratory it was shown that the apparent association constant for binding of gene 45 protein trimers to gene 44162 protein complexes is on the order of 7 x lo6 M-’ (Jarvis et al., 1989b). This weak binding of gene 45 protein is not limited to the accessory proteins in isolation. Although some evidence exists for a tighter association of the polymerase with gene 45 protein in the absence of gene 44162 protein (Topal and Sinha, 1983; Alberts et al., 1983), stimulation of leading strand DNA synthesis by the full accessory protein complex is generally also characterized by weak gene 45 protein binding. Thus Bedinger and Alberts (1983) found that an 8-11-fold excess of gene 45 protein was required to fully saturate the stimulatory effect of the accessory proteins on the polymerase exonuclease activity, while Newport et al. (1980) reported, in studies of leading strand DNA synthesis by the five protein T4 replication system, that the number of DNA chains initiated per unit time as a function of gene 45 protein concentration increased far beyond the point at which stoichiometric

levels of gene 45 protein had been added. A processive DNA replication complex functioning within the cell must hold together long enough to incorporate thousands of nucleotide residues into the nascent DNA strands during the lifetime of a single binding event. Based on an average

number

of -60 replication

forks per infected

Esche-

cell (Matthews et al., 1988), a reasonable estimate of the total concentration of each T4 replication protein present in the cell during DNA synthesis may be in the low micromolar range. This suggests that an association constant of lo6 to lo7 M-l for gene 45 protein binding to the gene 441 richia

15160

coli

Macromolecular

Crowding Effects on T4 DNA Replication

62 protein complex will not suffice to support efficient replication. We believe that macromolecular crowding, reflecting the difference between solution conditions in viuo and those normally used in uitro, may be involved in stabilizing important association interactions within the replication complex. Solutions used to conduct typical in vitro biochemical assays generally contain less than 0.1% (w/v) concentrations of macromolecules and thus can be expected, to a first approximation, to behave in a thermodynamically “ideal” fashion. In contrast, it is not uncommon for intracellular soluble proteins to reach 15-25% concentrations by weight (see Fulton, 1982). Estimates of physiological protein concentrations vary with cell type examined (Fulton, 1982), but it is clear that in all cases a significant volume fraction of the intracellular space is occupied by macromolecular components. The non-ideal contribution to the chemical potential in such a “volumeoccupied” solution can have a profound effect on the thermodynamic equilibria we study in these systems. Minton (1981, 1983) has used statistical methods to model the effect of excluded volume on the thermodynamic parameters governing the interactions of globular macromolecules. It must be emphasized that the effects discussed here are not merely due to increasing effective macromolecular reactant concentrations by fractional decreases in solvent volume; for example, raising the effective concentration of the reactive species in a bimolecular association reaction by lo-20% has an almost negligible effect on the fraction bound at equilibrium. In contrast, using inert polymers to decrease the effective volume available to the interacting macromolecular solutes in the solution substantially alters the thermodynamic activity of the macromolecular components and can change the equilibrium constant of a macromolecular reaction by several orders of magnitude. This change in reaction equilibrium results because a protein or nucleic acid in a macromolecularly volume-occupied solution has many fewer configurational degrees of freedom than in a dilute solution of comparable protein concentration; the degree to which the configurational entropy is decreased is a complex function of the relative size and shape of the inert solute and the reacting macromolecules. Therefore, although the exact magnitude of the macromolecular crowding effect may be difficult to predict for a given situation, there are predictable thermodynamic trends. In particular, Minton (1981, 1983) has shown that macromolecular crowding preferentially favors association reactions that result in quasispherical, as opposed to extended, forms of the resulting complex. In order to mimic “crowding” conditions in vitro, it is desirable to use some sort of macromolecule that is available in a variety of sizes and is relatively inert (that is, one that exhibits little enthalpic interaction with the reacting species). Polyethylene glycol (PEG)’ has been utilized in several studies and appears to meet these requirements. The interaction of polyethylene glycol with proteins has been extensively studied (see Murphy et al., 1988). Solutions of polyethylene glycol (average molecular weight of 6,000-35,000) have been shown to promote the association of ribosomal subunits (Zimmerman and Trach, 1988), to tighten the binding of certain DNA polymerases to DNA (Zimmerman and Harrison, 1987), and to increase the effciency of DNA kinasing reactions (Harrison and Zimmerman, 1986). Polyethylene glycol increases the processivity of topoisomerase (Forterre et al., 1985), increases the rate of intermolecular DNA ligation reactions (Pheiffer and Zimmerman, ’ The abbreviation used is: PEG, polyethylene glycol, followed by a number indicating the average molecular weight.

15161

Complex II

1983; Zimmerman and Pheiffer, 1983; Takahashi and Uchida, 1986; Teraoka and Tsukada, 1987) and is essential for the function of an in uitro plasmid replication system (Fuller et al., 1981). In this paper we present experiments in which we have used concentrated solutions of polyethylene glycol and dextran (a polysaccharide) to alter the thermodynamic (and kinetic) interactions of gene 45 protein with the other components of the T4 DNA replication system. MATERIALS

AND

METHODS

Polyethylene Glycols and Dextrans-All sizes of dextrans and PEG 200, 1,500, 12,000 and 35,000 were obtained from Fluka and were used without further purification. PEG 8,000 was from Sigma. No significant differences were observed when reactions were performed with PEG that had been extensively dialyzed against water, suggesting that commercial preparations of PEG do not contain low molecular weight contaminants that might inhibit the accessory protein ATPase activity. PEGS and dextrans were dissolved in distilled water; we report concentrations of these polymers in weight per unit volume percentages. The density of PEG 6,000 is 1.21 z/cm? thus a 10% (w/ ;) soluti& corresponds to an 8.3%‘(v/v) soluti& Protein and Nucleic Acid Preparation and Analysis-The T4 gene 45 protein, gene 44/62 protein complex, gene 32 protein, and gene 43 protein were all purified as described previously (see Jarvis et al., 1989a, 1989b; Jarvis, 1989, and references therein). Protein concentrations were determined spectrophotometricaliy using extinction coefficients derived by the calculation method of Gill and von Hippel (1989), based on published amino acid sequences (Spicer et al., 1982, 1984, 1988; Spicer and Konigsberg, 1983; Rush et al., 1989). Gene 45 protein concentrations are reported in units of nanomolar trimers

and gene 44/62 protein concentrations

as nanomolar

in (4:l) com-

plexes, in accord with the association states of these proteins determined previously (Jarvis et al., 1989a). Gene 32 protein and gene 43 protein concentrations are reported as nanomolar in protein monomers. Oligonucleotides were synthesized and purified as described previously (Jarvis et al.. 1989b). The Ml3 seauencins mimer (5’GTAAACGACGGCCAGT) anneals to M13mp8 at iosition 62866270. Single-stranded M13mp8 DNA was a gift from Dr. Mary Kay Dolejsi (this laboratory). AZ’Pase Assays-ATPase assays were performed spectrophotometrically, essentially as described previously (Jarvis et al., 1989b), using either an HP8450A or a Beckman model 25 UV/visible SnectroDhotometer with temperature-controlled cells set at 37 “C. The presence of PEG does not adversely affect the functioning of the coupling enzymes involved in the spectrophotometric assay, as judged by comparison of the results of this assay with controls carried out using a direct TLC assay (Jarvis et al., 198913). Buffers and Methods-The buffer for the DNA synthesis assays contained 25 mM Tris-OAc (pH 7.5), 5 mM fl-mercaptoethanol, 6 mM Mg(OAc)z, 160 mM KOAc, 250 NM deoxyribonucleotide triphosphates, 1.0 mM ATP, and 50 nM (molecules) of circular single-stranded M13mp8 DNA primed with an equimolar amount of 5’-32P-labeled l7-mer primer. The proteins were pre-mixed and diluted in a buffer containing 25 mM KOAc, 25 mM Tris-OAc (pH 7.5), 0.1 mM EDTA, and 5 mM P-mercaptoethanol. The reactions were started by addition of the protein mixture to the buffer, followed by incubation for 2 min at 37 “C, and quenching with an equal volume of 0.25 M EDTA. The quenched reactions were then phenol/chloroform-extracted and ethanol-precipitated with NH,OAc. The pellets were resuspended in a formamide dye mix (95% deionized formamide, 1 mg/ml bromphenol blue, 1 mg/ml xylene cyanole FF, and 10 mM EDTA), boiled briefly and chilled rapidly prior to electrophoresis on 10% polyacrylamide gels containing 8 M urea (Maxam and Gilbert, 1980). Dried gels were quantitated on an AMBIS radioanalytic scanner. RESULTS

ATPase Activity as a Probe for the Binding of Gene 45 Protein to the Gene 44/62 Protein Complex-As described in Jarvis et al. (1989b), the gene 44/62 protein complex displays a low level of DNA-dependent ATPase activity that is stimulated (>30-fold) by the addition of gene 45 protein. Titration curves, in which the concentration of one of the accessory

15162

Macromolecular


proteins is fixed and that of the other is varied, have been presented previously (see Fig. 4, A and B, in Jarvis et al,, 1989b). Hyperbolic titration curves are obtained, both when the concentration of gene 45 protein is fixed and that of the gene 44162 protein complex is varied and in the reverse situation. The shape of the curves is consistent with a simple bimolecular association reaction with a relatively weak equilibrium constant. (In experiments in which the concentration of gene 45 protein is fixed a correction must be made for the low, but detectable, DNA-dependent ATPase activity of gene 44162 protein alone.) For titrations of this sort, carried out at saturating concentrations of the DNA cofactor and the ATP substrate, we have shown (see Jarvis et al., 1989b) that the rate of ATP hydrolysis is a direct reflection of the amount of gene 45 protein bound to the gene 44/62 protein complex. The plateau rate (V,,,,,) is achieved when all of the available fixed protein is complexed with titrant protein. Thus the ATPase activity of the accessory protein complex can be used as a probe to measure the binding affinity of gene 45 protein for the gene 44162 protein complex. In dilute solution, in the buffer system used here, the association constant for this reaction was previously shown to be -7 x lo6 M-’ (Jarvis et al., 1989b). Here we examine the effect of increasing the mole fraction of solution occupied by “inert” macromolecules on this binding equilibrium. Macromolecular crowding might be expected to perturb not only protein-protein interactions, but also the binding of replication proteins to the cofactor DNA. The complete accessory protein complex has a K, for primertemplate DNA sites of about 10 nM in dilute solution (Jarvis et al., 1989b). The assay concentrations used here (350-450 nM primer-template sites) are well above saturation, and therefore no effect on ATPase rate is expected to result from changes in DNA concentration. In contrast, the gene 44/62 protein complex alone has a much higher K,,, for DNA (1.3 pM sites). Thus in this case tighter binding of gene 44/62 protein to DNA induced by crowding should lead to a slightly higher background ATPase rate. This is, in fact, observed (the details are presented in Fig. 1). The protein-protein binding equilibria reported below have been corrected to eliminate the effect of this phenomenon. A priori, one would not expect macromolecular crowding to alter the catalytic activity of the accessory protein complex beyond the effects normally induced by binding of the macromolecular cofactors DNA and gene 45 protein. The ATP substrate is present at a concentration 5-fold above K,,,, and therefore any increase in ATP binding affinity induced by macromolecular crowding polymers should not affect the measured ATPase rate.* Optimal Polymer Size for the Crowding Effect-A variety of different sizes of polyethylene glycol and dextran were tested as crowding agents (by monitoring these effects on ATPase activity) in order to explore the effect of polymer size on the binding equilibria involved in the assembly of the accessory proteins complex. The results are summarized in Table I. Compared to “dilute” solutions, reactions containing 10% (w/v) polyethylene glycol of average molecular mass 1,500 daltons or smaller show little or no effect on the apparent ’ In addition, ATP has a small molecular volume compared to those of the protein and DNA components. The binding constant for a small molecule substrate is unlikely to be significantly altered by the presence of macromolecular crowding solutes (Minton, 1981). Although polymers such as polyethylene glycol do increase the observed viscosity of the solvent, this effect is primarily on the so-called “macroviscosity” of the solution (Blacklow et al., 1988). Thus the rate at which small molecules are able to diffuse in the medium is not expected to be significantly decreased.

Complex II TABLE

I

Influence

of polymer size on the macromolecular crowding effect on the ATPase of the accessory proteins system Polymers of the indicated size were present at 10% (w/v) concentrations. Reaction mixtures contained saturating amounts of primertemplate DNA (5’-GCG(A)20 annealed to 3’-CGC(T)s$) at a concentration of 350 nM molecules (K,,, for this DNA is approximately 10 nM; Jarvis et al., 198913). The concentration of gene 45 was held constant at 85 nM trimers, and that of gene 44/62 protein complex was varied from 0 to 350 nM. The ATPase rate has been corrected for the low level DNA-dependent activity of gene 44/62 protein in the absence of gene 45 protein and the data (15 to 20 points per curve) have been fitted to a simple 1:l binding equilibrium by a nonlinear curve fitting routine (Duggleby, 1981) to determine the V,,,., and & for the gene 45 protein to gene 44/62 protein interaction. k,., was obtained by dividing V,., by the fixed concentration of gene 45 protein trimers. The errors represent the standard error derived from the nonlinear curve fitting routine. Poivmer No polymer PEG PEG PEG PEG PEG Dextran Dextran Dextran

added

200 1,500 8,000 12,000” 35,000 6,000 15,000-20,000 110.000

K,

k -*.

IZM

min-’

185 + 20

600 f 60

>200 190 f 50 52 + 10 6f2 40 f 8

>200 600 f 650 f 260 + 520 +

150 + 20 65 f 10 75 f 10

590 k 60 470 + 90 610 f 60

90 70 30 60

“The low value for k,., with PEG 12,000 appears to be due to a reversible precipitation of the proteins at PEG concentrations at and above 10%. This issue is examined in greater detail in Table II.

dissociation constant of gene 45 protein to the gene 44162 protein complex. Larger polymers, however, do induce markedly tighter protein-protein association. This effect may be seen in the much decreased Kd values listed in Table I for reactions run in the presence of PEG 8,000, 12,000, and 35,000. PEG 12,000, in particular, significantly increases the binding affinity of gene 45 protein for the gene 44162 proteins complex. This effect clearly exceeds that which would be expected from the decreased volume (and consequent increase in effective concentration of the interacting species) alone.3 Although generally less efficacious than polyethylene glycol, the larger sized dextrans tested also show a distinct macromolecular crowding effect on the binding equilibria. As expected, the maximal ATPase activity observed (Iz,,,) remains essentially unchanged in the presence of any of the polymers with the exception of PEG 12,000. The reason for the rate reductions seen with this crowding polymer will be discussed below. Theoretical calculations by Minton (1981) compare situations in which the molecular volume of the crowding macromolecule is about one-tenth, equal to, or ten times larger than the volume of the reacting species. These calculations predict that the excluded volume effect will be greatest when the crowding molecules are one-tenth the volume of the reacting species. Our results are certainly in qualitative accord with this observation. Gene 45 protein exists in solution as a 74,000-dalton trimer, the gene 44/62 protein complex has a molecular mass of 164,000 daltons; the macromolecular 3 It is easy to calculate that the apparent & for the binding of gene 45 urotein to the comnlex will be decreased by approximately 15% as a consequence of making the solution 7.5%. in PEG 12,000, if the effect were due only to the overall decrease in the solution volume available to the reactants. In contrast, the observed decrease in Kd is approximately 50-fold, clearly demonstrating that the effect is one of macromolecular excluded volume, and is critically dependent on the macromolecular character of the crowding agent.

Macromolecular

Crowding

Effects on T4 DNA Replication

crowding effect is observed to be greater for the PEG 12,000 than for the PEG 35,000 polymer. Minton has not directly explored the lower size limit at which inert macromolecules exhibit a crowding effect. In fact, simple extrapolation of Minton’s equations suggests an everincreasing nonideal crowding effect as the size of the inert polymer is decreased. Clearly this is not in agreement with our experimental results, nor with those of others (Zimmerman and Harrison, 1987). Otto C. Berg (University of Uppsala) has recently developed a more elaborate theoretical treatment of the crowding phenomenon, using a hard-sphere fluid model and explicitly including the primary solvent component (water) as well as the macromolecular solutes.“ One striking consequence of Berg’s treatment is the prediction of an optimal size for the crowding macromolecule. This optimum is shown to be dependent on the size of the crowding macromolecule, relative both to the size of the primary solvent molecule and to the sizes of the reacting solute macromolecules. The optimum size predicted for the crowding macromolecule by the Berg analysis comes at a molecular mass of approximately 200 daltons; this is obviously considerably below the optimum molecular weight actually observed under the conditions of Table I. However, this prediction applies to hard sphere crowding macromolecules. Minton (1983) has suggested that crowding macromolecules that are “nonspherical” in shape may show enhanced crowding effects. Thus it is possible that the Berg calculations, based totally on hard sphere approximations, might be further refined to treat more effectively the non-ideal solution effects of extended (random coil) crowding polymers such as polyethylene glycol. The Crowding Effect Increases with Polymer Concentration--All of the experiments shown in Table I were carried out at 10% (w/v) of the polymer involved and showed that PEG 12,000 is the most effective crowding agent we have tested. We then proceeded to investigate the concentration dependence of this polymer on the macromolecular crowding effect. The results are presented in Table II. Increasing the volume fraction of the solution occupied by PEG 12,000 clearly drives the equilibrium toward increased binding of gene 45 protein to gene 44162 protein, as shown by the decrease in Kd value as the concentration of the polymer is raised toward 7.5%. The value of k,., is not changed over this range of PEG concentrations, showing again that excluded volume changes perturb primarily macromolecular binding equilibria rather than catalytic activity. Concentrations of PEG 12,000 at or above lo%, on the other hand, produce a visible precipitate and a concomitant loss of enzymatic activity. This precipitation is fully reversible and appears to be a natural consequence of the crowding phenomenon (i.e., the association equilibra between macromolecules has become so favorable under these conditions that nonspecific aggregation predominates). The steady decrease in the apparent k,,, at higher PEG concentrations is probably a consequence of precipitation of active gene 45 protein (the fixed protein in the titrations summarized in Table II). Precipitation of gene 44/62 protein would be expected to result in an increased apparent Kd value for the gene 45 to gene 44/62 protein interaction, and this is clearly observed. Thus both gene 45 and gene 44162 protein complexes seem to be affected by this aggregation. We do not know whether the DNA is also subject to structural alteration under these conditions. The results of Lerman (1971) suggest that this is unlikely at the monovalent salt concentrations used in this study. However, it has been shown that the formation of condensed (psi form) DNA is facilitated at 4 0. G. Berg,

(1990)

Biopolymers, submitted

for publication.

Complex II TABLE

II

ATPase parameters as a function of concentration for PEG 12,000 Conditions as in Table I, except that the concentration protein trimers was 95 nM and that of primer-template Cd% was 450 nM. PEG 12,000 (w/v) % 0.0 2.5 5.0 7.5 10.ob 12.5 15.0 20.0

Kd IlM

165 57 15 3.0

k 2 + f

30 6 3 1.5

10 f 4 (55)

(280) (z-500)

kc., min-’ 580 k 650 + 600 + 600 f

of gene 45 DNA mole-

40 40 20 20

450 + 20 350 280 >160

a Note that the rates observed (even in the absence of polyethylene glycol) are somewhat lower than those reported previously (Jarvis et al., 198913). The kcat values obtained previously were derived from assays performed at very low protein concentrations, requiring a double extrapolation to saturating concentrations of gene 45 protein and DNA. Thus the results of our earlier assays involve a larger standard error than are those presented here. b A visible precipitate forms at a threshold concentration of about 10% PEG 12,000, increases in magnitude with increasing protein concentration, and is accompanied by an apparent loss of ATPase activity. This precipitation process seems to be fully reversible, since dilution of a 15% PEG solution to 7.5% PEG results in >95% recovery of the normal activity observed in a solution initially containing 7.5% PEG. This precipitation makes curve fitting to a simple binding equilibrium difficult; thus the Kd values shown for PEG concentrations of 12.5% and above are placed in parentheses to indicate their approximate nature. The difference in rates between Table I and II at 10% PEG 12,000 probably reflects experimental error in solution preparation that is magnified near the threshold of precipitation (i.e. if the experiment in Table I actually contained 10.5% PEG and Table II contained 9.5% PEG, the values for k,,, would be significantly different).

relatively low concentrations of magnesium ions (although higher than those used here (see Auer, 1978)). While the background DNA-dependent ATPase activity of the gene 44162 protein complex (in the absence of gene 45 protein) normally increases linearly with added gene 44/62 protein, this activity exhibits a less monotonic behavior at PEG concentrations greater than 10%. We find that the observed ATPase activity is normal at low protein concentrations and then actually decreases at higher protein concentrations. Fig. 1 shows the DNA-dependent ATPase activity of gene 44162 protein alone as a function of polymer concentrations for three different protein concentrations. At low concentrations of polyethylene glycol the activity observed for a given amount of protein increases, consistent with the notion that the DNA binding affinity increases (recall that under these conditions the DNA concentration is below the K, value for the interaction of DNA with the gene 44/62 protein complex in the absence of gene 45 protein). Aggregation causes the measured ATPase rate to fall at polymer concentrations above 7.5%; this effect is most notable at high accessory protein concentrations. Note that the curves cross between 10 and 15% PEG. This phenomenon could be explained by a model in which higher protein concentrations nucleate the aggregation effect induced by high concentrations of crowding polymer. The Apparent Stoichiometry of the Accessory Protein Complex is 1:1 in Gene 4416.2 Protein Complexes and Gene 45 Protein Trimers-The binding constant (-3.5 X 10’ M-l, expressed as KJ for the binding interactions of the polymerase accessory proteins summarized in Table II (for 7.5% PEG 12,000) corresponds to about 50-fold tighter binding of gene

15164

Macromolecular

I 0

5

Effects on T4 DNA Replication

Crowding

10 % PEG 12.OI33(w/v)

I 20

15

FIG. 1. Effect of PEG 12,000 on the DNA-dependent ATPase rate of gene 44/62 protein alone. The ATPase rate of gene 44/62 protein in the presence of 450 nM of primer-template DNA (as in Table II) is plotted as a function of the concentration of PEG 12.000. The concentration of gene 44/62 protein complex = 95 nM (Oj, 185 nM (m), and 370 n; (O), resp&tively. The highest rate attained (at 7.5% PEG) equals, but does not exceed, the maximum rate of ATP hydrolysis previously measured for the gene 44/62 protein complex in the presence of saturating DNA primer-template cofactor (Jarvis et al., 198913). This is consistent with models that suggest that macromolecular crowding increases the affinity of gene 44162 protein for DNA (a lo-fold increase in binding affinity would be sufficient to explain the result), but does not change the catalytic activity of the protein.

100

200

IgW62pl

300

‘loo

(nM) complexes

FIG. 2. Macromolecular crowding increases the association constant of gene 45 protein to the proteins complex. Corrected ATPase rate data fit binding equilibrium by nonlinear curve fitting are shown containing 0% (O), 2.5% (O), 5.0% (m), and 7.5% (0) Reaction conditions as in Table II.

45 protein

trimers

to the

gene

44/62

protein

equilibrium gene 44162 to a simple for reactions PEG 12,000.

complexes

than

observed in dilute solution. It should be noted that under these conditions it becomes difficult to measure binding constants accurately because, as shown in Fig. 2, nearly stoichiometric titration of gene 45 protein has been attained. Thus the dissociation constant obtained by curve fitting here should be considered an upper limit for the true Kd value. Ideally one would like to repeat the experiment at lo- or loo-fold lower fixed protein concentrations in order to determine the true binding constant. However, due to the low ATPase rate obtained under such conditions, this measurement is not experimentally attainable. We can, however, exploit the results shown in Fig. 2 in another

way.

Thus,

as a result

of the

significant

increase

in

binding due to the addition of PEG, we can estimate the stoichiometry of gene 45 protein trimers binding to the gene 44/62 protein complex. For reasons discussed previously (Jarvis et al., 1989b), it seems unlikely that the gene 45 protein trimers

disproportionate

The titration

experiments

upon

binding

summarized

to gene

44162

protein.

in Fig. 2 each involved

Complex II

adding gene 44/62 protein to solutions containing 95 nM of gene 45 protein trimers. The ATPase rate in these experiments (carried out in 7.5% PEG) reaches a plateau at a concentration of -100 nM gene 44162 protein complexes. This strongly suggests a 1:l stoichiometry of binding of gene 45 protein trimers to gene 44/62 protein complexes in forming the functional accessory proteins complex, assuming that the relative specific activities of the two protein preparations are approximately the same. Similar results obtained from several different batches of each protein support the validity of this assumption. Macromolecular Crowding Stabilizes the T4 Holoenzyme Replication Complex-We have shown, in the results presented above, that macromolecular crowding can significantly strengthen the interactions between the components of the polymerase accessory protein complex. It is obviously of interest also to ask whether such volume-occupied solution conditions have a comparable effect on the interaction of the accessory protein complex with the T4 DNA polymerase. The following experiments address this question by examining the amount of added gene 45 protein required for maximal stimulations of DNA synthesis by polymerase in the presence of gene 44162 and gene 32 proteins. Thus we use gene 45 protein as a probe to examine the stability of the functional five protein “holoenzyme” complex. The T4 polymerase accessory proteins are known to stimulate the processivity of the polymerase, especially at physiological salt concentrations (Newport et al., 1980; Mace and Alberts, 1984; Jarvis et al., 199Oc). This stimulation requires ATP hydrolysis, as well as gene 44/62 and gene 45 proteins. These proteins are thought to form a holoenzyme complex with the polymerase.” Here we are concerned specifically with the influence of macromolecular crowding on the stability and processivity of the entire five-protein leading strand DNA replication complex. Under these conditions the binding of gene 32 protein to single-stranded DNA is quite tight and the protein would be expected to bind at close to stoichiometric levels (Newport et al., 1981). Thus, although crowding probably influences the DNA binding constant of gene 32 protein, such effects should not perturb these experiments. We have examined the processivity of polymerase on primed natural sequence single-stranded DNA. In order to measure processivity straightforwardly, it is necessary to work under conditions where any particular DNA primer will be acted upon by polymerase no more than once during the course of the assay (i.e. we have used “single-hit” kinetic conditions for these studies; see Newport et al., 1980). Under these conditions the distribution of sizes of the extended primer products (displayed on a gel) accurately reflects the DNA synthesized in a single polymerase binding event. One strategy for attaining single-hit conditions is to provide a large excess of template over polymerase. The reaction is quenched after sufficient time has elapsed to allow each polymerase to perform multiple rounds of synthesis, but under conditions where less than 10% of the primers have been extended. Thus the probability of a polymerase binding to and extending a primer that has already been extended will 5 Intact T4 holoenzyme complexes have not been isolated, in contrast to certain other DNA replication systems for which holoenzymes have been at least partially purified (e.g. see McHenry and Crow, 1979; Vishwanatha et al., 1986). Dr. Nancy Nossal and coworkers have recently shown, using gel filtration techniques, that the T4 DNA nolvmerase and accessory proteins can be isolated together in the breience of ATP and DNA (personal communication). Therefore it seems likelv that under some conditions the T4 DNA replication proteins can form a stable complex with a half-life of at least several minutes, even in dilute solution.

Macromolecular PEG


Complex II

15165

+ PEG

123456789

[gene 45 protein] (nM)

FIG. 3. Macromolecular crowding facilitates DNA polymerase holoenzyme complex formation at low protein concentrations. Primer extension assays were performed as described under “Materials and Methods.” The reactions shown in lanes 5-9 contained 8% PEG 12,000, while those in lanes I-4 were performed under normal dilute solution conditions. All the reactions contained 0.1 nM gene 43 protein, 20 nM gene 44/62 protein complex, 1 pM gene 32 protein, and the indicated molarities of gene 45 protein trimers. Lanes I and 5 are controls showing 5’-“‘P-end-labeled DNA alone.

be less than one in ten. This strategy has the added advantage in this case of utilizing extremely low polymerase concentrations. This should exacerbate the consequences of weak protein-protein binding affinities within the holoenzyme complex, leading to dissociation and thus providing a good test for the efficiency with which macromolecular crowding agents stabilize this complex. The results of such an assay are shown in Fig. 3. The primers have been 5’-end-labeled with “P, allowing us readily to ascertain that single-hit conditions do prevail. This also means that the amount of radioactivity at any position in the gel is a direct measure of the number of primers extended to that size. The processivity of T4 polymerase alone has been shown to decrease with increasing salt concentration (Newport et al., 1980);’ therefore we have chosen to carry out these experiments at relatively high salt concentrations. Polymerase alone rarely extends any primer more than about 80 nucleotide residues under these conditions (see lanes 2 and 6, Fig. 3). On the other hand, much longer products are seen in the presence of added accessory proteins (plus gene 32 protein and ATP; see lanes 3-4 and 7-9, Fig. 3). Most importantly, it is clear that much less gene 45 protein is required in the presence of PEG 12,000 to achieve the distribution of longer extended primers that characterize DNA synthesis by the five-protein holoenzyme complex. A summary of the results of this and similar experiments is presented in Fig. 4. Since virtually all primers that are extended by more than 100 nucleotide residues must represent holoenzyme products (polymerase alone being much less processive at these high salt conditions), we can quantitate the minimal percentage of elongated primers that must have been produced by the holoenzyme. Although this form of analysis is too crude to yield a binding constant for gene 45 protein to the rest of the holoenzyme complex, it does show that approximately 50 times as much gene 45 protein is required in dilute solution as in the macromolecularly crowded solution to produce comparable amounts of extended holoenzyme product (compare the ef‘F. data.

D. Fairfield,

J. D.

Linn,

and

P. H. van

Hippel,

unpublished

FIG. 4. Holoenzyme products as a function of gene 45 protein concentration. The results of several experiments (similar to that presented in Fig. 3) are compiled to show the percentage of extended primers that are longer than 100 nucleotide residues, plotted as a function of gene 45 protein concentration in the presence and absence of PEG 12,000.

fects of 500 nM gene 45 protein without PEG to those of 10 nM gene 45 protein in the presence of PEG). Therefore it appears that the binding increment that is obtained by the addition of PEG 12,000 to the synthesis buffer is approximately the same for the polymerase holoenzyme complex as was observed previously for the assembly of the polymerase accessoryproteins complex in isolation. We assume, therefore, that the primary effect of crowding the holoenzyme complex is also to increase the affinity of this system for gene 45 protein. The Processivity of Polymerase Is Not Affected by Macromolecular Crowding-We have also asked whether macromolecular crowding has an effect on the processivity of DNA synthesis, either by polymerase alone or by the holoenzyme complex. It is difficult to separate protein-protein interaction effects of macromolecular crowding on the holoenzyme complex from possible effects on the intrinsic processivity of the holoenzyme; i.e. the holoenzyme complex may simply be unstable due to weak protein-protein affinities and thus fall apart, reducing the apparent processivity of the holoenzyme complex as a consequence. Therefore we first examine the effects of added PEG on the processivity of DNA synthesis by polymerase alone. At first glance the results presented in Fig. 3 appear to suggest that the addition of PEG 12,000 to polymerase promotes the formation of more longer extended primers (although still none that are extended beyond -80 nucleotide residues) than are seen in the polymerase reaction in dilute solution (compare lanes 2 and 6). Quantitation of these lanes, however, shows that the product size distribution in these reactions is not significantly altered by macromolecule crowding. However, the number of primers extended per polymerase per unit time does increase slightly for the “polymerase alone” reactions in the presence of PEG 12,000. This finding is consistent with the results of Zimmerman and Harrison (1987), who reported that the overall rate of DNA synthesis by E. coli polymerase I (Klenow fragment) increases as a result of crowding, but that no change in the overall processivity is observed. Although our results and those of Zimmerman and Harrison (1987) do not comprise an exhaustive study, they clearly do demonstrate that the effects of crowding agents on the apparent processivity of the polymerase holoenzyme complex are not likely to be due to increases in the intrinsic processivity of the polymerase itself. According to the theory of Minton (1983), excluded volume

15166

Macromolecular


effects drive macromolecular association reactions by raising the configurational entropy of the reactants (several independent macromolecules) more than that of the product (a single macromolecular complex). Thus, Minton (1983) concludes that the primary effect of crowding agents will be on the rate of association of the reactants, rather than on their rate of dissociation. If true, this conclusion is consistent with our observation that the processivity of polymerase is not substantially altered by macromolecular crowding. Since processivity is strictly a property of the already formed proteinDNA complex, it should be affected by changes in the dissociation rate of the protein from the DNA, but would not reflect changes in the association rate. The increase in the percentage of primers elongated per unit time by polymerase alone in the presence of PEG could, however, reflect an increase in the rate of association of the primer-template DNA with the polymerase. DISCUSSION

In this study, we have shown that macromolecular crowding promotes association reactions between the proteins of the T4 DNA replication system. The magnitude of this effect in 7.5% PEG 12,000 is roughly in agreement with the predictions of Minton (1981) for the contribution of that degree of volume occupancy to the non-ideal free energy of such a system. A factor of 50 increase in the association constant corresponds to a value of -3.9 for AGOm/RT (the nonideal contribution to the standard state free energy of reaction). This value is slightly larger than that predicted by Minton for a simple bimolecular association without change of shape (see Minton, 1983, Fig. 5). However, Minton has also shown that these interactions are highly dependent on the shapes of the macromolecular reactants and products. Since we know that both the gene 45 protein trimer and the gene 44/62 protein complex are quite asymmetric (Jarvis et al., 1989a), an additional increment of free energy may be gained from macromolecular crowding if the complete accessory protein complex assumes a more globular shape (see Minton, 1983, Fig. 8). While it is clear that, at least under some conditions, macromolecular crowding can shift the thermodynamic equilibrium of a macromolecular association reaction, the exact effect on the association and dissociation rates remains undetermined. Minton (1983) has proposed that the primary contribution is due to an increase in the association rate because the total entropy loss of the reactants will be greater than that of the product; the increase in the configurational entropy of the transition state due to crowding is assumed to be approximately the same as that of the product, since both exist as single particles (as opposed to the multiple reactants with their additional degrees of freedom). This suggests that no large effect of crowding will be seen when the association rate is diffusion limited. We note that it could be argued that the activation barrier to dissociation might also increase as a result of crowding if the transition state were particularly asymmetric compared to the final product, since crowding tends to stabilize complexes of decreased asymmetry. The theoretical studies of Minton (1981, 1983) provide a powerful basis for modeling the interactions of macromolecules in volume-occupied solutions. These studies have brought attention to an important and relatively unexplored area of physical biochemistry. Obviously there is still much to learn about the thermodynamic properties of non-ideal solutions with respect to the behavior of proteins, DNA, and other macromolecular species. For example, the theoretical curves of Minton (1981) depict a non-ideal component of

Complex II

macromolecular equilibra that diverges steeply at volume occupancies of 30% and above. Under the conditions prevalent in uiuo, this suggests that many macromolecular reactions would be essentially irreversible, rendering critical regulatory interactions impossible. The results of Berg therefore constitute an important step forward in understanding the scope and significance of the macromolecular crowding effect.3 By explicitly treating the primary solvent as a component of the solution, Berg has demonstrated that the magnitude of the crowding effect is considerably smaller, particularly at high volume occupancy, than was originally predicted by Minton. Clearly it is of interest to learn more about the underlying thermodynamics of volume occupied solutions, not only from the standpoint of understanding a particular process such as replication, but also from the general standpoint of in vitro biochemistry. Thus, in order to understand the fundamental principles governing macromolecular processes of biological interest, it is important to mimic in uiuo solution conditions as closely as possible. At the same time, one must strive to maintain the minimum number of reacting components, the hallmark of a controlled in vitro system. Addition of inert macromolecular components may thus prove to be a valuable tool in understanding the biological relevance of experimental results obtained in uitro. Our present study on protein-protein interactions within the T4 DNA replication complex clearly shows that the thermodynamic activity of associating macromolecules is subject to profound changes when studied under volume-occupied, rather than dilute solution conditions. We find that as little as 7.5% of an inert macromolecule such as polyethylene glycol increases the equilibrium association constant for the gene 45 protein to the gene 44/62 protein complex by a factor of 50. A similar effect is seen for the association of gene 45 protein with the polymerase and gene 44/62 protein to form a holoenzyme complex. Although it appears that macromolecular crowding does not increase the intrinsic processivity of the polymerase, it does stabilize the multi-protein replication complex, which itself is much more processive than polymerase alone. These results suggest that the volume-occupied solution conditions prevalent in uiuo can significantly stabilize holoenzyme DNA replication complexes, and thus support high rates and high apparent processivities of DNA synthesis. Acknowledgments-We are grateful to Dr. Otto G. Berg of the University of Uppsala for communicating his results on macromolecular crowding theory to us prior to publication, and to both Otto Berg and Dr. Michael Reddy of this laboratory for helpful comments on the manuscript. REFERENCES Alberts, B. M., Barry, J., Bedinger, P., Formosa, T., Jongeneel, C. V., and Kreuzer, K. N. (1983) Cold Spring Harbor Symp. Quant. Biol. 47,655-668 Auer, C. (1978) The Influence of the Ionic Environment on the Structure of DNA, Ph.D. Thesis, Vanderbilt University Bedinger, P., and Alberts, B. M. (1983) J. Biol. Chem. 258, 964% 9656 Blacklow, S. C., Raines, R. T., Lim, W. A., Zamore, P. D., and Knowles. J. R. (1988) Biochemistry 27,1158-1167 Cha, T.-A.: and Aiberts, B. M. (1988) in Cancer Cells (Kelly, T., and Stillman, B., eds) Vol. 6, pp. l-10, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Duggleby, R. G. (1981) Anal. Biochem. 110, 9-18 Forterre, P., Mirambeau, G., Jaxel, C., Nadal, M., and Duguet, M. (1985) EMBO J. 4,2123 Fuller, R. S., Kaguni, J. M., and Kornberg, A. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 7370-7374 Fulton, A. B. (1982) CeU 30,345-347

Macromolecular


Gill, S. C., and von Hippel, P. H. (1989) Anal. Biochem. 182, 319328 Harrison, B., and Zimmerman, S. B. (1986) Anal. Biochem. 158, 307-315 Jarvis, T. C. (1989) The Role of the Polymeruse Accessory Proteins in Bacteriophage T4 DNA Replication, Ph.D. Thesis, University of Oregon Jarvis, T. C., Paul, L. S., and von Hippel, P. H. (1989a) J. Viol. &em. 264,12709-12716 Jarvis, T. C., Paul, L. S., Hockensmith, J. W., and von Hippel, P. H. (1989b) J. Biol. Chen. 264, 12717-12729 Jarvis, T. C., Dolejsi, M. K., Kubasek, W. L., Paul, L. S., Reddy, M. K., and von Hippel, P. H. (1990) UCLA Symp. Mol. Cell. Biol. 127, 261-275 Lerman, L. S. (1971) Proc. N&l. Acad. Sci. 7% S. A. 68, 1886-1890 Mace, D. C., and Alberts, B. M. (1984) J. Mol. Biol. 177, 313-327 Mathews, C. K., Moen, L. K., Wang, Y., and Sargent, R. G. (1988) Trends Biochem. Sci. 13,394-397 Maxam, A. M., and Gilbert, W. (1980) Methods Enzymol. 65, 499560 McHenry, C. S., and Crow, W. (1979) J. Biol. Chem. 254.1748-1753 Minton,A. P. (1981) Biopolymers 20,2093-2120 Minton. A. P. (1983) Mol. Cell. Biochem. 55.119-140 Murthy; N. S.,‘Bras’well, E. H., and Knox, j. R. (1988) Biopolymers 27,865881 Newport, J. W., Kowalczykowski, S. C., Lonberg, N., Paul, L. S., and von Hippel, P. H. (1980) ICN-UCLA Symp. Mol. Cell. Biol. 19, 485-505 Newport, J. W., Lonberg, N., Kowalczykowski, S. C., and von Hippel, P. H. (1981) J. Mol. Biol. 145, 105-121 Nossal, N. G., and Alberts, B. M. (1983) in Bacteriophage T4 (Mathews, C. K., Kutter, E. M., Mosig, G., and Berget, P. B., eds) pp. 71-81, American Society for Microbiology, Washington, D. C.

Complex II

15167

Nossal, N. G., and Peterlin, B. M. (1979) J. Biol. Chem. 254, 60326037 Pheiffer, B. H., and Zimmerman, S. B. (1983) Nucleic Acids Res. 11, 7853-7871 Piperno, J. R., and Alberts, B. M. (1978) J. Biol. Chem. 253, 51745179 Rush, J., Lin, T.-C., Quinones, M., Spicer, E. K., Douglas, I., Williams, K. R., and Konigsberg, W. H. (1989) J. Biol. Chem. 264, 1094310953 Spicer, E. K., and Konigsberg, W. H. (1983) in Bacteriophage T4 (Mathews, C. K., Kutter, E. M., Mosig, G., and Berget, P. B., eds) pp. 291-301, American Society for Microbiology, Washington, D. C. Spicer, E. K., Noble, J. A., Nossal, N. G., Konigsberg, W. H., and Williams, K. R. (1982) J. Eiol. Chem. 257,8972-8979 Spicer, E. K., Nossal, N. G., and Williams, K. R. (1984) J. Biol. Chem. 259,15425-15432 Spicer, E. K., Rush, J., Fung, C., Reha-Krantz, L. J., Karam, J. D., and Konigsberg, W. H. (1988) J. Biol. Chem. 263, 7478-7486 Takahashi, M., and Uchida, T. (1986) J. Biochem. (Tokyo) 100,123131 Teraoka, H., and Tsukada, K. (1987) J. Biochem. (Tokyo) 101,225-

231 Topal, M. D., and Sinha, N. K. (1983) J. Biol. Chem. 258, 1227412279 Vishwanatha, J. K., Coughlin, S. A., Wesolowski-Owen, M., and Baril, E. F. (1986) J. Biol. Chem. 261. 6619-6628 Zimmerman, S. B., and Harrison,‘B. (1987) Proc. Nutl. Acad. 5%. U. S. A. 84, 1871-1875 Zimmerman, S. B., and Pheiffer, B. H. (1983) Proc. N&l. Acad. Sci. U. S. A. 80,5852-5856 Zimmerman, S. B., and Trach, S. 0. (1988) Nucleic Acids Res. 16,

6309-6326