NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Lodish H, Berk A, Zipursky SL, et al. Molecular Cell Biology. 4th edition. New York: W. H. Freeman; 2000.

  • By agreement with the publisher, this book is accessible by the search feature, but cannot be browsed.
Cover of Molecular Cell Biology

Molecular Cell Biology. 4th edition.

Show details

Section 12.2The DNA Replication Machinery

The cellular mechanisms responsible for DNA replication were uncovered first in bacterial systems; more recently, they have been studied with proteins isolated from yeasts and cultured eukaryotic cells. Because the proteins and reactions involved in E. coli DNA replication are now understood in considerable detail, our discussion focuses mainly on these; however, replication in eukaryotes is carried out by analogous proteins and proceeds via similar reactions.

Before plunging into a description of the various enzymes and numerous other factors that participate in DNA replication, we need to recall certain elementary problems in the copying of DNA by DNA polymerases, mentioned in Chapter 4:

  • DNA polymerases are unable to melt duplex DNA (i.e., break the interchain hydrogen bonds) in order to separate the two strands that are to be copied.
  • All DNA polymerases so far discovered can only elongate a preexisting DNA or RNA strand, the primer; they cannot initiate chains.
  • The two strands in the DNA duplex are opposite (5′ → 3′ and 3′ → 5′) in chemical polarity, but all DNA polymerases catalyze nucleotide addition at the 3′-hydroxyl end of a growing chain, so strands can grow only in the 5′ → 3′ direction.

In this section, we describe the cell’s solutions to the unwinding, priming, and directionality problems resulting from the structure of DNA and the properties of DNA polymerases.

DnaA Protein Initiates Replication in E. coli

Genetic studies first suggested that initiation of replication at oriC most likely depended on the protein encoded by a gene designated dnaA. Initially, mutant strains carrying temperature-sensitive mutations in dnaA were isolated; these cells grew at permissive temperatures (e.g., 30 °C) but not at nonpermissive temperatures (39 – 42 °C). When E. coli cells carrying such conditional lethal mutations had begun DNA replication at the permissive temperature and then were shifted to the higher temperature, they completed the round of DNA synthesis already under way; however, they did not start another round of replication at the nonpermissive temperature. Subsequent genetic studies with recombinant E. coli further pinpointed the DnaA protein as a prime candidate for interaction with oriC. In vitro studies showed that pure DnaA protein binds to the four 9-mers in oriC, forming an initial complex that contains 10 – 20 protein subunits (Figure 12-7). Furthermore, although DnaA can bind to duplex E. coli origin DNA in the relaxed-circle form, it can initiate replication only if the DNA is negatively supercoiled. The reason for this specificity is that DNA molecules with negative supercoils are tightly wound and are easier to melt locally (thus providing a single-stranded template region) than are DNA molecules without supercoils. Supercoiling of DNA and the enzymes that control the degree of DNA supercoiling, called topoisomerases, are discussed in detail later.

Figure 12-7. Model of initiation of replication at E. coli oriC.

Figure 12-7

Model of initiation of replication at E. coli oriC. The 9-mers and 13-mers are the repetitive sequences shown in Figure 12-5. Multiple copies of DnaA protein bind to the 9-mers at the origin and then “melt” (separate the strands of) the (more...)

Binding of DnaA to the oriC 9-mers facilitates the initial strand separation, or “melting,” of E. coli duplex DNA, which occurs at the oriC 13-mers. This process requires ATP and yields a so-called open complex. When a mixture of E. coli DNA and DnaA protein is treated with an endonuclease that specifically recognizes single-stranded DNA, the DNA is cut in the origin region, demonstrating that it is melted.

DnaB Is an E. coli Helicase That Melts Duplex DNA

Further melting of the two strands of the E. coli chromosome to generate unpaired template strands is mediated by the protein product of the dnaB locus, a helicase that is essential for DNA replication. One molecule of DnaB, a hexamer of identical subunits, clamps around each of the two single strands in the open complex formed between DnaA and oriC. This binding requires ATP and the protein product encoded by the dnaC locus, which “escorts” DnaB to the DnaA proteins, yielding the prepriming complex (see Figure 12-7).

Helicases constitute a class of enzymes that can move along a DNA duplex utilizing the energy of ATP hydrolysis to separate the strands. In E. coli, the separated strands are inhibited from subsequently reannealing by a single-strand-binding protein (SSB protein), which binds to both separated strands. When temperature-sensitive dnaB mutants are shifted to nonpermissive temperatures, unwinding ceases; as a result, DNA synthesis stops immediately for lack of single-stranded templates. Helicases like DnaB bind to a single-stranded segment of DNA, then move along that strand melting the hydrogen bonds that link it to its complementary strand. Like many proteins that bind to DNA, helicases exhibit a directionality with respect to the unwinding reaction. DnaB moves along the single strand of DNA to which it binds in the direction of its free 3′ end, and in this sense it is said to unwind DNA in the 5′ → 3′ direction (Figure 12-8). DnaB, like many other proteins that act on DNA, is processive. Because it forms a clamp around a single strand of DNA, DnaB does not “fall off” until it reaches the end of that strand or is “unloaded” from DNA by another protein. Other kinds of DNA helicases unwind in the opposite direction, moving along the strand to which they are bound toward the free 5′ end.

Figure 12-8. Helicase activity of E. coli DnaB protein.

Figure 12-8

Helicase activity of E. coli DnaB protein. In the presence of ATP and single-strand-binding (SSB) protein, purified DnaB can unwind a gapped DNA duplex in vitro. Unwinding occurs predominantly in the direction of the 3′ end of the strand to which (more...)

E. coli Primase Catalyzes Formation of RNA Primers for DNA Synthesis

As noted earlier, DNA polymerases can only elongate existing primer strands of DNA or RNA. The primers used during DNA replication in both prokaryotes and eukaryotes are short RNA molecules whose synthesis is catalyzed by the RNA polymerase primase. E. coli strains with temperature-sensitive mutations in dnaG, which encodes primase, cannot replicate their DNA at the nonpermissive temperature, thereby establishing the essential role of primase. Primase is usually recruited to a segment of single-stranded DNA by first binding to a DnaB hexamer already attached at that site. The term primosome is now generally used to denote a complex between primase and helicase, sometimes with other accessory proteins. In initiation of E. coli DNA replication, a primosome is formed by binding of primases to DnaB in the prepriming complex (see Figure 12-7). After the bound primases synthesize short primer RNAs complementary to both strands of duplex DNA, they dissociate from the single-stranded template.

At a Growing Fork One Strand Is Synthesized Discontinuously from Multiple Primers

We have seen how the activities of helicase and primase solve two of the problems inherent to DNA replication — unwinding of the duplex template and the requirement of DNA polymerases for a primer. Remember, though, that both strands of the DNA template are copied as the replication bubble enlarges. Each end of the bubble represents a growing fork where both new strands are synthesized (see Figure 12-2c).

At each growing fork, one strand, called the leading strand, is synthesized continuously from a single primer on the leading-strand template and grows in the 5′ → 3′ direction. Growth of the leading strand proceeds in the same direction as movement of the growing fork (Figure 12-9a). Synthesis of the lagging strand is more complicated, because DNA polymerases can add nucleotides only to the 3′ end of a primer or growing DNA strand. Movement of the growing fork unveils the template strand for lagging-strand synthesis in the 5′ → 3′ direction; thus the overall direction of growth of the lagging strand must be from its 3′ end toward its 5′ end, complementary to the polarity of its template but opposite to the direction of nucleotide addition by DNA polymerases. In both prokaryotes and eukaryotes these apparently incompatible requirements are met by the discontinuous copying of the lagging strand from multiple primers, a process involving several steps.

Figure 12-9. At a growing fork, one strand is synthesized from multiple primers.

Figure 12-9

At a growing fork, one strand is synthesized from multiple primers. (a) The overall structure of a growing fork. Synthesis of the leading strand, catalyzed by DNA polymerase III, occurs by sequential addition of deoxyribonucleotides in the same direction (more...)

As synthesis of the leading strand progresses, sites uncovered on the single-stranded template of the lagging strand are copied into short RNA primers (<15 nucleotides) by primase (Figure 12-9b). Each of these primers is then elongated by addition of deoxyribonucleotides to its 3′ end. In E. coli, this reaction is catalyzed by DNA polymerase III (Pol III), one of three DNA polymerases produced by E. coli. Thus each lagging strand grows in a direction opposite to that in which the growing fork is moving. The resulting short fragments, containing RNA covalently linked to DNA, are called Okazaki fragments, after their discoverer Reiji Okazaki. In bacteria and bacteriophages, Okazaki fragments contain 1000 – 2000 nucleotides, and a cycle of Okazaki-strand synthesis takes about 2 seconds to complete. In eukaryotic cells, Okazaki fragments are much shorter (100 – 200 nucleotides).

As each newly formed segment of the lagging strand approaches the 5′ end of the adjacent Okazaki fragment (the one just completed), E. coli DNA polymerase I takes over. Unlike polymerase III, polymerase I has 5′ → 3′ exonuclease activity, which removes the RNA primer of the adjacent fragment; the polymerization activity of polymerase I simultaneously fills in the gap between the fragments by addition of deoxyribonucleotides. Finally, another critical enzyme, DNA ligase, joins adjacent completed fragments (Figure 12-9c).

E. coli DNA Polymerase III Catalyzes Nucleotide Addition at the Growing Fork

Three DNA polymerases (I, II, and III) have been purified from E. coli (Table 12-1). In addition to its role in filling the gaps between Okazaki fragments, DNA polymerase I probably is the most important enzyme for gap filling during DNA repair. DNA polymerase II functions in the inducible SOS response discussed later; this polymerase also fills gaps and appears to facilitate DNA synthesis directed by damaged templates. Our discussion here focuses on DNA polymerase III, which catalyzes chain elongation at the growing fork in E. coli.

Table 12-1. Properties of DNA Polymerases.

Table 12-1

Properties of DNA Polymerases.

The DNA polymerase III holoenzyme is a very large (>600 kDa), highly complex protein composed of 10 different polypeptides. The so-called core polymerase is composed of three subunits. The α subunit contains the active site for nucleotide addition, and the ϵ subunit is a 3′ → 5′ exonuclease that removes incorrectly added (mispaired) nucleotides from the end of the growing chain. (This “proofreading” activity of DNA polymerase III is described later.) The function of the Θ subunit is not known.

The central role of the remaining subunits is to convert the core polymerase from a distributive enzyme, which falls off the template strand after forming relatively short stretches of DNA containing 10 – 50 nucleotides, to a processive enzyme, which can form stretches of DNA containing up to 5 × 105 nucleotides without being released from the template. This latter activity is necessary for efficient synthesis of both leading and lagging strands. The key to the processive nature of DNA polymerase III is the ability of the β subunit to form a donut-shaped dimer around duplex DNA and then associate with and hold the catalytic core polymerase near the 3′ terminus of the growing strand (Figure 12-10). Once tightly associated with the DNA, the β-subunit dimer functions like a “clamp,” which can slide freely along the DNA, like a ring on a string, as the associated core polymerase moves. In this way, the active sites of the core polymerase remain near the growing fork and the processivity of the enzyme is maximized. Remarkably, of the six remaining subunits, five (γ, δ, δ′, χ, and ψ) form the so-called γ complex, which mediates two essential tasks: (1) loading of the β-subunit clamp onto the duplex DNA – primer substrate in a reaction that requires hydrolysis of ATP and (2) unloading of the β-subunit clamp after a strand of DNA has been completed. Loading and unloading of the β-subunit clamp requires opening the clamp ring, but exactly how the γ complex accomplishes this feat is not known. The final subunit (τ) acts to dimerize two core polymerases and, as summarized in the next section, is essential for coordinating the synthesis of the leading and lagging strands at each growing fork.

Figure 12-10. A β-subunit dimer tethers the core of E. coli DNA polymerase III to DNA, thereby increasing its processivity.

Figure 12-10

A β-subunit dimer tethers the core of E. coli DNA polymerase III to DNA, thereby increasing its processivity. (a) Space-filling model based on x-ray crystallographic studies of the dimeric β subunit binding to a DNA duplex. Two β (more...)

The Leading and Lagging Strands Are Synthesized Concurrently

Once the prepriming complex and an RNA primer are formed at the E. coli replication origin, chain elongation to yield the leading strand proceeds with little difficulty. As we’ve seen, however, lagging-strand synthesis proceeds discontinuously from multiple primers. Two molecules of core DNA polymerase III are bound at each growing fork; one adds nucleotides to the leading strand, and the other adds nucleotides to the lagging strand. Coordination between elongation of the leading and lagging strand is essential; otherwise one template strand would be incorporated into a duplex with a newly synthesized complementary strand while large parts of the other template strand would remain single-stranded.

Figure 12-11 shows how this coordination is achieved. The two core-polymerase molecules at the fork are linked together by a τ-subunit dimer. The core polymerase synthesizing the leading strand moves, together with its β-subunit clamp, along its template in the direction of the movement of the fork, elongating the leading strand. It follows closely the movement of the DnaB helicase bound to the lagging-strand template as the helicase melts the duplex DNA at the fork. Since this core-polymerase molecule remains attached to the DNA template, leading-strand synthesis occurs continuously.

Figure 12-11. Schematic model of the relationship between E. coli replication proteins at a growing fork.

Figure 12-11

Schematic model of the relationship between E. coli replication proteins at a growing fork. (1) A single DnaB helicase moves along the lagging-strand template toward its 3′ end, thereby melting the duplex DNA at the fork. (2) One core polymerase (more...)

The other core-polymerase molecule, which elongates the lagging strand, moves with its β-subunit clamp in the direction opposite to that of the fork movement. As elongation of the lagging strand proceeds, the size of the DNA “loop” between this core polymerase and the fork increases. One way to see this is to imagine the core 2 polymerase fixed in space, linked to core 1; double-stranded DNA newly synthesized by core 2 would be “pushed” into the loop. Eventually the core polymerase synthesizing the lagging strand will complete an Okazaki fragment; it then dissociates from the DNA template, but the τ-subunit dimer continues to tether it to the forkprotein complex. Simultaneously, a primase binds to a site adjacent to the DnaB helicase on the single-stranded segment of the lagging-strand template and initiates synthesis of another RNA primer. The resulting DNA-primer complex attracts another β clamp to this segment of the lagging-strand template, followed by re-binding of the core polymerase, which is still attached to the fork complex. This polymerase molecule then proceeds to elongate the RNA primer to form another Okazaki fragment. As mentioned earlier, as each Okazaki fragment nears completion, the RNA primer of the previous fragment is removed by the 5′ → 3′ exonuclease activity of DNA polymerase I. This enzyme also fills in the gaps between the lagging-strand fragments, which then are ligated together by DNA ligase (see Figure 12-9b).

Although the two core polymerase molecules are linked by the τ-subunit dimer, they are oriented in opposite directions (see Figure 12-11). Thus, the 3′ growing ends of both the leading and lagging strands are close together but offset from each other. For this reason, the point in the template at which the lagging strand is being copied is displaced from the point in the template at which leading-strand copying is occurring. Nonetheless, the two core polymerases can add deoxyribonucleotides to the growing strands at the same time and rate, so that leading- and lagging-strand synthesis occurs concurrently.

One τ subunit also contacts the DnaB helicase at the fork. Experiments with purified replication proteins have shown that this interaction increases the normally slow unwinding rate of the helicase (≈35 bp/s) over tenfold, thereby enabling the fork to move at rates up to 1000 bp/s. Thus, there is a physical and functional link between the two major replication machines at the fork — the two core polymerases and the primosome complex of primase and DnaB. By closely coordinating all the events depicted in Figures 12-9 and 12-11, the growing fork moves 500 – 1000 bp/s while both strands are being replicated.

Eukaryotic Replication Machinery Is Generally Similar to That of E. coli

As in E. coli, researchers investigating DNA replication in eukaryotes initially concentrated on characterizing the different DNA polymerases present in eukaryotic cells (see Table 12-1). This work was followed by development of in vitro systems for copying small chromosomes from animal viruses (e.g., SV40) whose replication is dependent almost entirely on host-cell proteins. As a result of these studies, the SV40 chromosome now can be replicated in vitro using only eight purified components from mammalian cells. The specific functions of these proteins are highly reminiscent of the E. coli proteins required for replication of plasmids carrying oriC. Thus, the mechanistic problems involved in DNA replication, which are similar in all organisms, have been solved in most cases by use of similar types of proteins. Like DNA replication in E. coli, eukaryotic DNA replication occurs bidirectionally from RNA primers made by a primase; synthesis of the leading strand is continuous, while synthesis of the lagging strand is discontinuous. In contrast to the situation in E. coli, however, two distinct DNA polymerases, α and either δ or ϵ, function at the eukaryotic growing fork.

As depicted in Figure 12-12, replication of SV40 DNA is initiated at a unique site, the replication origin, by binding of a virus-encoded protein called T antigen, or Tag (step 1). This multifunctional, site-specific DNA-binding protein locally melts duplex DNA through its helicase activity. Opening of the duplex at the SV40 origin also requires ATP and replication protein A (RPA), a host-cell single-strand-binding protein with a function similar to that of SSB protein in E. coli cells (step 2). One molecule of polymerase α (Pol α), tightly associated with a primase, then binds to each unwound template strand. The primases form RNA primers, which are elongated for a short stretch by Pol α, forming the first part of the leading strands, which grow from the origins on the two template strands in opposite directions (step 3). The activity of Pol α is stimulated by replication factor C (RFC).

Figure 12-12. Model of in vitro replication of SV40 DNA by eukaryotic enzymes.

Figure 12-12

Model of in vitro replication of SV40 DNA by eukaryotic enzymes. This figure depicts both growing forks originating from the SV40 origin: the top strand is the leading-strand template for the leftward-moving fork, and the bottom strand is the leading-strand template (more...)

PCNA (proliferating cell nuclear antigen) then binds at the primer-template 3′ termini, displacing Pol α from both leading-strand templates and thus interrupting leading-strand synthesis (Figure 12-12, step 4). Next, Pol δ binds to PCNA at the 3′ ends of the growing strands. The association of Pol δ with PCNA increases the processivity of the polymerase, so that it can continue synthesis of the leading strands without further interruption (step 5). The function of PCNA thus is highly analogous to that of the β-subunit clamp of E. coli polymerase III, as both proteins form similar “rings” through which the DNA slides. However, their amino acid sequences are dissimilar, and the β clamp is a dimer, whereas PCNA is a trimer.

As melting of the duplex DNA, catalyzed by a hexameric form of Tag, progresses farther away from the origin, the primase – Pol α complex associates with the melted template strands downstream from the leading-strand primers. Synthesis of the lagging strand then is carried out by combined action of primase and Pol α, along with RFC, while leading-strand synthesis on the other side of the origin also proceeds (see Figure 12-12, step 5). Finally, in eukaryotes, as in E. coli, topoisomerases play an important role in relieving torsional stress induced by growing-fork movement and in separating the two daughter chromosomes.

Much has been learned about the eukaryotic proteins that can carry out replication of SV40 viral DNA in vitro. As noted above, initiation of SV40 replication in vitro requires the viral protein T antigen. Studies on replication of eukaryotic cellular DNA in vitro have been hampered by the lack of eukaryotic experimental systems that can sustain in vitro replication initiated at cellular origins and by the lack of in vitro replication systems prepared from extracts of genetically tractable organisms such as yeast. As discussed in Chapter 13, the recent identification of a protein complex that binds to yeast chromosomal origins may be an important first step toward detailed research on cellular DNA replication using a combined genetic and biochemical strategy, which has proved so profitable in E. coli.

Telomerase Prevents Progressive Shortening of Lagging Strands during Eukaryotic DNA Replication

Unlike bacterial chromosomes, which are circular, eukaryotic chromosomes are linear and carry specialized ends called telomeres. As discussed in Chapter 9, telomeres consist of repetitive oligomeric sequences; for example, the yeast telomeric repeat sequence is 5′-G1 – 3 T-3′. The need for a specialized region at the ends of eukaryotic chromosomes is apparent when we consider that all known DNA polymerases elongate DNA chains from the 3′ end, and all require an RNA or DNA primer. As the growing fork approaches the end of a linear chromosome, synthesis of the leading strand continues to the end of the DNA template strand; the resulting completely replicated daughter DNA double helix then is released. However, because the lagging-strand template is copied in a discontinuous fashion, it cannot be replicated in its entirety (see Figure 12-9a). When the final RNA primer is removed, there is no upstream strand onto which DNA polymerase can build to fill the resulting gap. Without some special mechanism, the daughter DNA strand resulting from lagging-strand synthesis would be shortened at each cell division.

The enzyme that prevents this progressive shortening of the lagging strand is a modified reverse transcriptase called telomerase, which can elongate the lagging-strand template from its 3′-hydroxyl end. This unusual enzyme contains a catalytic site that polymerizes deoxyribonucleotides directed by an RNA template, and the RNA template itself, which is brought to the site of catalysis as part of the enzyme (Figure 12-13). The repetitive sequence added by telomerase is determined by the RNA associated with the enzyme, which varies among telomerases from different sources. Once the 3′ end of the lagging-strand template is sufficiently elongated, synthesis of the lagging strand can take place, presumably from additional primers.

Figure 12-13. Mechanism of action of telomerase.

Figure 12-13

Mechanism of action of telomerase. This ribonucleoprotein complex elongates the 3′ telomeric end of the lagging-strand DNA template by a reiterative reverse transcription mechanism. The action of the telomerase from Oxytricha, which adds a T4 (more...)

For cells in many organisms, telomere length is increased many times over early in development. Most human somatic cells replicate in the absence of telomerase activity and thus gradually consume the telomeric repeats added earlier in development. The progressive shortening of the chromosome ends and eventual loss of genetic information that results has been linked to cell death, and it has even been suggested that life span is determined by the number of telomeres with which an individual starts. Indeed, an inverse relationship between age and telomeric length has been observed.


  •  The enzymes and other protein factors that carry out DNA replication in E. coli and in eukaryotic cells are analogous, suggesting that the biochemical mechanism of DNA replication is similar in all cells.
  •  The enzymatic events at the growing fork are a consequence of two properties of the DNA double helix and two of DNA polymerases. The DNA helix contains antiparallel strands (i.e., the 5′ → 3′ direction of one strand is opposite to the 5′ → 3′ direction of the other), and the two strands are interwound, so they cannot simply be melted along their entire lengths all at once. DNA polymerases require a nucleic acid primer — either a DNA or an RNA molecule — to begin synthesis, and all DNA chain growth occurs by nucleotide addition at the 3′ end.
  •  In all cells, one new DNA strand, the leading strand, is synthesized continuously in the direction of movement of the growing fork by elongation from the 3′ end of an RNA primer base-paired to a template strand. Synthesis of the other strand, the lagging strand, occurs in the direction opposite to the overall direction of growing fork movement from a series of short RNA primers formed by primase at multiple sites on the second template strand. The resulting segments of RNA plus DNA are called Okazaki fragments. After the primers are removed and the gaps between fragments are filled, they are joined.
  •  Initiation of DNA replication in E. coli occurs by binding of DnaA to oriC, followed by attachment of DnaB, a helicase that melts DNA at the fork (see Figure 12-7). Association of primase with this complex forms a primosome. After primer synthesis, primase dissociates.
  • E. coli DNA polymerase III catalyzes nucleotide addition to both the leading and the lagging strands. DNA polymerase I removes the RNA primers from Okazaki fragments and fills in the gaps on the lagging strand. Finally, DNA ligase joins adjacent completed Okazaki fragments (see Figure 12-9).
  •  Eukaryotic proteins that replicate SV40 DNA in vitro exhibit similarities with E. coli replication proteins. A viral protein called T antigen functions similarly to the E. coli DnaB helicase, and host-cell PCNA (proliferating cell nuclear antigen) is similar to the β-subunit clamp associated with E. coli DNA polymerase III. However, two distinct mammalian polymerases, α and δ or ϵ, function at the eukaryotic growing fork (see Figure 12-12).
  •  The processivity of DNA polymerases is essential for efficient polymerization and is facilitated by their association with the β-subunit clamp in E. coli and PCNA in eukaryotes.
  •  Telomerase, a reverse transcriptase that contains an RNA template, adds nucleotides to the 3′ end of the lagging-strand template and thus prevents shortening of lagging strands during replication of linear DNA molecules such as those of eukaryotic chromosomes.
Image ch12f5
Image ch12f2

By agreement with the publisher, this book is accessible by the search feature, but cannot be browsed.

Copyright © 2000, W. H. Freeman and Company.
Bookshelf ID: NBK21751