doi:10.1016/S0092-8674(03)00512-9

Cell, Vol. 114, 135–146, July 11, 2003, Copyright 2003 by Cell Press

Structure of the Rho Transcription Terminator:

Mechanism of mRNA Recognition

and Helicase Loading

this process by a number of factors, such as NusG,

which serve as antiterminators to ameliorate the termi-

nation properties of the enzyme (Sullivan and Gottes-

man, 1992).

Tethering of Rho to a rut site is independent of ATP

Emmanuel Skordalakes and James M. Berger*

Department of Molecular and Cell Biology

University of California, Berkeley

239 Hildebrand Hall, #3206

Berkeley, California 94720

binding and/or hydrolysis (Galluppi and Richardson,

1980; McSwiggen et al., 1988) and is mediated by an

N-terminal domain in each protomer of the Rho hexamer

Summary

(Martinez et al., 1996a, 1996b; Allison et al., 1998; Bri-

ercheck et al., 1998). Previous high-resolution NMR and

In bacteria, one of the major transcriptional termina-

X-ray crystal structures of the N-terminal mRNA binding

tion mechanisms requires a RNA/DNA helicase known

domain of Rho have provided a structural view of this

as the Rho factor. We have determined two structures

fold and its interactions with target substrates (Allison

of Rho complexed with nucleic acid recognition site

et al., 1998; Briercheck et al., 1998; Bogden et al., 1999).

mimics in both free and nucleotide bound states to

The domain consists of a three-helix bundle that rests

3.0 A

resolution. Both structures show that Rho forms

on top of a five-stranded ␤ barrel. The barrel belongs

a hexameric ring in which two RNA binding sites—a

to the oligonucleotide/oligossacharide binding (OB) pro-

primary one responsible for target mRNA recognition

tein superfamily and is primarily responsible for Rho’s

and a secondary one required for mRNA translocation

preferential interaction with cytosine-rich nucleic acids

and unwinding—point toward the center of the ring.

(McSwiggen et al., 1988; Geiselmann et al., 1992; Rich-

Rather than forming a closed ring, the Rho hexamer

ardson and Richardson, 1992; Modrak and Richardson,

is split open, resembling a “lock washer” in its global

1994; Platt, 1994). Curiously, the domain does not spe-

architecture. The distance between subunits at the

cifically recognize the 2⬘-hydroxyl of RNA and can asso-

opening is sufficiently wide (12 A

) to accommodate

ciate with ssDNA in vitro, a property that has been used

single-stranded RNA. This open configuration most

to great effect for dissecting Rho function (Oda and

likely resembles a state poised to load onto mRNA

Takanami, 1972; Richardson, 1982; Brennan et al., 1987).

and suggests how related ring-shaped enzymes may

The C-terminal region of Rho contains signature se-

be breached to bind nucleic acids.

quence motifs that are hallmarks of Walker-type ATP

binding proteins. The closest homolog to Rho is the F

Introduction

ATP synthase, an ␣

␤

heterohexameric complex that

assembles into rings and generates ATP from a proton-

Appropriate termination of transcription is essential for

motive force (Boyer, 1997). Consistent with this relation-

proper regulation of gene expression. In bacteria, mRNA

ship, EM studies have shown that Rho also forms hex-

transcription termination is carried out by two distinct

americ rings, although in the absence of RNA, these

mechanisms (Richardson and Greenblatt, 1996). In in-

rings exist in both closed and “notched” open forms

trinsic termination, an mRNA stem loop structure is

(Oda and Takanami, 1972; Gogol et al., 1991; Yu et al.,

thought to induce RNA polymerase to pause, promoting

2000). Notched rings have been interpreted either to be

the release of both the polymerase and the RNA tran-

missing a subunit or to be a natural functional state of

script from the template DNA. This mechanism accounts

the protein for loading RNA into the hole of the hexamer

for approximately half of the termination sites in Esche-

(Yu et al., 2000). Open Rho rings can be converted into

richia coli, and resembles the transcription termination

the closed form by the binding of extended single-

mechanism used by eukaryotic pol III. In contrast, the

stranded nucleic acids, which have been proposed to

second mechanism of transcription termination is pro-

wrap around the perimeter of the hexamer (Galluppi

tein mediated and relies on an enzyme found throughout

and Richardson, 1980; McSwiggen et al., 1988; Gan and

bacteria known as the Rho factor (Brown et al., 1981;

Richardson, 1999; Yu et al., 2000).

Opperman and Richardson, 1994).

As Rho loads onto nascent mRNAs, the single-

Since its discovery in 1969 by Roberts (Roberts, 1969),

stranded substrate is thought to bind in the hole of the

Rho has become a paradigm for understanding how

hexameric ring, where it interacts with a secondary RNA

exogenous proteins can regulate the termination of tran-

binding site in the C-terminal domains (Burgess and

scription by RNA polymerase. In vivo, Rho loads onto

Richardson, 2001a; Burgess and Richardson, 2001b).

mRNAs at a cytosine-rich region of 40 or more bases

Two regions, the Q and R loops, have been implicated

known as a Rho utilization (rut) site (Morgan et al., 1985;

in RNA associations at the secondary site (Miwa et al.,

Alifano et al., 1991). Once bound, Rho acts as a hexa-

1995; Burgess and Richardson, 2001a; Wei and Richard-

meric, 5⬘→3⬘ ATP-dependent helicase to translocate to

son, 2001b), and RNA binding in the hole of the hexa-

the site of transcription and disengage the polymerase

meric ring can stimulate ring closure when the second-

(Oda and Takanami, 1972; Lowery-Goldhammer and

ary sites become affixed to the message (Gogol et al.,

Richardson, 1974; Morgan et al., 1983; Brennan et al.,

1991; Yu et al., 2000). After loading appropriately, the

1987; Richardson, 2002). Rho can be regulated during

C-terminal domains are thought to engage in cycles of

ATP binding and hydrolysis, translocating Rho along the

mRNA until it reaches RNA polymerase and unwinds

*Correspondence: [email protected]

Cell

136

the RNA/DNA heteroduplex at the site of transcription ssDNA and another bound to ssRNA and AMPPNP. The

ssDNA-Rho complex was phased to intermediate reso-

(Lowery-Goldhammer and Richardson, 1974; Brennan

lution (6.0 A

) by single isomorphous replacement from

et al., 1987, 1990; Walstrom et al., 1997; Burgess and

a tantalum bromide derivative, followed by high-resolu-

Richardson, 2001a; Kim and Patel, 2001; Stitt, 2001).

tion phasing using selenomethionine-based MAD. The

Two models have been advanced to describe the physi-

ssRNA/AMPPNP assembly was solved by molecular re-

cal nature of the translocation/unwinding reactions

placement. Both structures were refined to 3.0 A

resolu-

(Richardson, 2002; Delagoutte and von Hippel, 2003). In

tion (Tables 1 and 2).

one, known as the tethered tracking mechanism, Rho

The crystalline asymmetric unit contains six Rho pro-

is thought to remain attached to the rut site while translo-

tomers assembled into a discrete hexameric particle.

cating toward the 3⬘ end of the mRNA (Steinmetz and

Each subunit is peanut-shaped, with dimensions of 65 A

Platt, 1994). In the other model, Rho is thought to detach

in height, 50 A

in depth, and 35 A

in width (Figure 1A),

from the rut site upon the onset of translocation (Geisel-

and is composed of an N- and C-terminal domain joined

mann et al., 1993; von Hippel and Delagoutte, 2001).

together by an extended 30 residue linker. Although

Despite extensive efforts, several questions regarding

there is clear electron density for five out of six

the mechanism of Rho remain unanswered. For exam-

N-terminal domains in the hexamer, the N-terminal do-

ple, it is not known how the primary and secondary RNA

main of protomer B is partially disordered, and electron

binding sites are oriented with respect to each other,

density can only be seen for its C-terminal half.

and whether this arrangement can accommodate the

The Rho N-terminal domain consists of two subdo-

simultaneous RNA binding events predicted by the teth-

mains: a three-helix bundle followed by a five-stranded

ered tracking model. It has also remained unclear how

␤ barrel (Figure 1A). The ␤ barrel is an OB-type fold,

an extended nucleic acid segment gains entry into the

which is found in a wide variety of single-stranded nu-

topologically closed-ring system of the Rho particle, a

cleic acid binding molecules that include proteins such

problem that exists for a variety of hexameric helicases.

as type II aminoacyl tRNA synthases and cold shock

To begin to address these questions, we determined

proteins (Cavarelli et al., 1993; Newkirk et al., 1994;

the structure of the full-length Escherichia coli Rho pro-

Schindelin et al., 1994). An indented face on one side

tein bound to an ssDNA recognition site mimic to 3.0 A

of this subdomain forms the primary RNA binding site

resolution. The same structure bound to ssRNA and the

of Rho (Modrak and Richardson, 1994; Briercheck et al.,

ATP analog AMPPNP (adenosine 5⬘-(␤,␥-imido)triphos-

1998; Bogden et al., 1999). Structural comparisons of

phate) was also determined to 3.0 A

resolution. Both

either the helical-bundle or the OB-fold subdomains be-

structures show that Rho forms hexameric rings in which

tween different protomers and isolated Rho N-terminal

the primary RNA binding sites of the N-terminal domains

domains solved previously by NMR and X-ray crystallog-

face toward the interior of the particle, positioning the

raphy show that the individual structures of these seg-

3⬘ end of nucleic acid exiting from the rut-recognition

ments do not vary greatly (C␣ RMSD 0.73A

) (Allison et

surface directly toward the hole of the hexamer. Nucleo-

al., 1998; Briercheck et al., 1998; Bogden et al., 1999).

tide binds in a crevice between the C-terminal domains

However, when comparing both subdomains in unison,

of neighboring protomers and is liganded by residues

the overall RMSDs for the entire N-terminal domain be-

of the Walker-A and -B motifs. Secondary RNA binding

come higher (C␣ RMSD 1.1A

) due to small variations in

motifs, which comprise the acceptor sites for translocat-

the relative orientations between the helical bundle and

ing along RNA, line the perimeter of the interior hole:

the OB fold. In the Rho hexamer, these differences are

six Q loops together form a narrow constriction in the

most prominent for protomers A and F.

center of the ring, while each R loop lies above the

Each of the six C-terminal domains of Rho consists of

P loop of the Walker-A motif. Strikingly, the Rho ring is

seven parallel ␤ strands (␤6–␤13) sandwiched between

split open in both the AMPPNP bound and unbound

several ␣ helices (Figure 1A). There is clear electron

states, adopting an architecture akin to that of a lock

density for residues 155–417 in all six C-terminal do-

washer. Taken together, these structural features illumi-

mains, although the last 15 residues exhibit higher than

nate the mechanism by which Rho associates with rut

average temperature factors than the rest of the model.

sites and loads the 3⬘ end of the mRNA into the interior

The average C␣ RMSD between each of the six mono-

of the ring prior to the start of translocation.

mers in the hexamer is 0.58 A

. As expected from se-

quence analysis, the fold of the C-terminal domain be-

Results and Discussion

longs to the RecA superfamily of ATP binding proteins.

A structural survey of such folds in the database shows

The Rho Monomer Structure

that this region of Rho is most similar to the mitochon-

The full-length Rho protein used for these studies is

drial F

ATPase (C␣ RMSD 2.1 A

) (Abrahams et al., 1994;

hexameric in solution, as judged by both gel filtration

Bird et al., 1998b).

and dynamic light scattering (data not shown). Although

There are several key signature sequence motifs lo-

preliminary crystals of the purified protein grew readily,

cated in the Rho C-terminal domain. One is the P loop,

extensive optimization of initial crystallization condi-

which is part of the Walker-A motif found in all RecA-

tions was required to obtain crystals that diffracted to

family ATPases and is required for ATP binding and

a useful resolution. The key elements in this process

hydrolysis. Each P loop is formed by the ␤6/␣7 connec-

proved to be cocrystallization of the protein with mild

tor segment and is juxtaposed with an adjacent pro-

detergents and the use of an appropriate length and

tomer. A second important motif in Rho’s C-terminal

sequence of nucleic acid substrate. In all, two different

domain is the R loop, which connects the secondary

structural elements ␤11/␣12 and is located just belowRho structures were obtained: one in complex with

Rho Termination Factor Structure

137

Table 1. Data Collection and Phasing Statistics

Rho-DNA Rho-RNA

Complex Complex Ta

⫹

Se-␭1 Se-␭2

Data set (␭ A

) 1.245 1.0 0.9794

Resolution (A

) 50–3.0 50–3.0 50–4.6 50–3.6 50–3.6

Completeness (%) 97.9 (97.8) 96.1 (95.0) 95.7 (99.1) 94.3 (94.8) 96.2 (96.7)

Redundancy 2.2 (2.2) 2.2 (2.2) 2.5 (2.5) 2.3 (2.3) 2.3 (2.3)

sym

(%)

4.6 (27.2) 5.0 (33.2) 6.9 (30.4) 6.3 (24.5) 6.7 (29.4)

I/␴ 15.5 (3.5) 15.3 (2.5) 10.0 (3.0) 12.9 (5.0) 11 (4.5)

Phasing Analysis Ta

⫹

Resolution (A

) 20–6.0 20–3.6

cullis

(%,acent/cent, iso)

0.78/0.68 0.43/0.40

cullis

(%, anom)

0.80 0.78

Number of sites 6 83

Mean figure of merit (FOM) 0.432 0.308

Values in parentheses are for the highest-resolution bin.

sym

⫽⌺⌺

– 具I典|/⌺具I典, where I

is the recorded intensity of the reflection j and 具I典 is the mean recorded intensity over multiple recordings.

Cullis

⫽⌺⌺||F

⫾ F

| ⫺ F

|/⌺|F

⫾ F

|, where F

and F

are the structure amplitudes of the parent and the heavy-atom derivative and F

the calculated heavy-atom structure factor.

the P loop. A third motif, the Q loop, is formed by an full-length, ring-shaped ATPases observed crystallo-

graphically to date have been imaged with their rings8 residue segment that projects toward the center of

the hexamer. Together, the Q and R loops are thought closed. Open configurations of oligomeric ATPases

have been observed previously for only a few proteins,to form Rho’s secondary RNA binding site (Burgess and

Richardson, 2001a; Wei and Richardson, 2001a, 2001b; including the recombination protein RecA and a trun-

cated form of the bacteriophage T7 gp4 primase/heli-Xu et al., 2002).

case (Story et al., 1992; Sawaya et al., 1999). However,

the particles in these structures do not exist as discreteThe Rho Assembly

The six subunits of Rho pack laterally into a hexameric hexamers and instead assemble into continuous fila-

ments with 6

symmetry. EM analyses have shown thatring (Figures 2A and 2B). Subunit-subunit interactions

occur between both the N- and C-terminal domains of Rho hexamers are able to individually form both open

and closed rings (Oda and Takanami, 1972; Gogol eteach protomer. Contacts between adjacent N-terminal

domains are generated between the ␣2/␣3 connector al., 1991; Yu et al., 2000), although the purpose, function,

and oligomeric organization of the open conformationloop in the helical bundle of one protomer and the ex-

tended linker that joins the N- and C-terminal domains have remained somewhat enigmatic. Our ability to trap

and image an open conformation reminiscent of thatof the adjacent subunit. In the C-terminal domain, ␣11

in one protomer packs against the ␤7/␣8 and ␤8/␣9 seen previously by EM now suggests that this conforma-

tion is a stable and natural form accessed by the protein.junctions of the neighboring subunit. The connector loop

for secondary elements ␤11/␣12 is also located at the The diameter of the Rho ring is 120 A

, with a large

interior hole that varies in width from 20–35 A

. The widestC-terminal subunit-subunit interface and is positioned

adjacent to and above the P loop of the ATP binding point of the hole is defined by the perimeter of the

N-terminal domains, while the most narrow constrictionsite.

Strikingly, the Rho ring is split open, giving rise to a is formed by the Q loop signature sequence motifs. The

lock-washer arrangement of the hexamer arises fromglobal structure reminiscent of a lock washer (Figures 2A

and 2B). This conformation is unusual, because nearly all an upward rotation of one protomer with respect to

Table 2. Model Refinement

Rho-DNA Complex Rho-RNA Complex

Resolution (A

) 20–3.0 20–3.0

free

(%)

29.6 30.3

work

(%)

27.0 27.0

RMSD

bond

) 0.014 0.010

RMSD

angle

(⬚) 1.47 1.25

Favored 85.6 84.6

Additionally allowed 12.4 13.2

Generously allowed 1.5 1.7

Total atoms (protein) 18,881 18,881

Total atoms (substrates) 190 386

Total atoms (Water) 20 15

free

is the R value calculated for a test set of reflections, comprising a randomly selected 5% of the data that is not used during refinement.

work, free

⫽⌺||F

obs

| ⫺ |F

calc

||/|F

obs

Cell

138

Figure 1. Structure and Topology of the Rho Protomer

(A) Structure of Rho protomer-D showing the relative orientations of the N- (cyan) and C- (red) terminal domains. Each N-terminal domain

consists of a helical bundle and an OB-fold subdomain. The C-terminal domain is connected to the N-terminal domain by an extended linker

(yellow) and consists of a RecA-type ATPase binding fold followed by a small helical subdomain. The P loop (blue), the Q loop (magenta),

and the R loop (green) are highlighted. Secondary structure elements are labeled.

(B) Exploded views of the N- and C-terminal domain interface. Both domains are rotated by 90⬚ opposite to each other. Residues involved

in interdomain contacts are shown as blue rods and labeled.

another by approximately 15⬚ about an axis that lies in ating with mRNA in the absence of accessory factors

(Burgess and Richardson, 2001a, 2001b), this structurala plane perpendicular to the pseudo 6-fold rotation axis

of symmetry (Figure 2C). This swivel generates a helical organization appears consistent with a particle state

competent to load onto extended nucleic acid chains.rise of approximately ⵑ8.5 A

along the rotational axis of

the particle. By propagating these displacements about

the ring of the hexamer, the midpoints of protomers A Rho and Primary Site RNA Interactions

Rho has two distinct nucleic acid binding sites. Theand F at either end of the ring are offset from each other

by 45 A

, giving rise to a 12 A

wide gap that breaches primary mRNA binding sites are formed by the

N-terminal domains, which have the ability to bind eitherthe ring and allows access to the interior of the particle

(Figure 2). It is interesting to note that the width of this single-stranded DNA or RNA (Galluppi and Richardson,

1980; Chen et al., 1986; McSwiggen et al., 1988; Modrakgap is sufficient to allow for the entry of a single-

stranded nucleic acid into the hole of the hexamer. and Richardson, 1994). This feature was used to grow

crystals of Rho in complex with target site mimics com-The intersubunit angular rise present in the Rho com-

plex (15⬚) is within a range seen for the related ATPases posed of either nucleic acid. Earlier studies have shown

that each N-terminal domain binds a dinucleotide seg-RecA and T7 gp4. In RecA, however, the subunit-subunit

rocking angle is more extreme and along the 6

helical ment in a network of contacts that explain the preference

of Rho for cytosine (Bogden et al., 1999). The first nucle-axis (18⬚), which leads to filamentous assembly of the

protein (Story et al., 1992). In contrast, for T7 gp4, two otide base packs into a hydrophobic enclosure that is

formed by the side chains of Tyr80, Glu108, and Tyr110,intersubunit rotation angles of 15⬚ in the plane of the

ring are offset by a downward 30⬚ swing every third and is too small to comfortably hold purine bases. For

the second nucleotide, the cytosine base stacks on thesubunit that leads to ring closure (Singleton et al., 2000).

Modulation of subunit-subunit orientations in these pro- aromatic side chain of Phe64, while its O2, N3, and N4

groups interact with the side chains of the neighboringteins is thought to occur in response to the combined

action of ATP turnover and nucleic acid associations, Arg66 and Asp78 (Bogden et al., 1999). No contacts

are seen to the 2⬘ hydroxyl of the bound nucleic acid,which drive either strand exchange (RecA) or DNA trans-

location and unwinding (T7 gp4). In Rho, it appears that explaining why Rho is able to bind both ssDNA and

ssRNA.this common hinging mechanism has been further co-

opted to allow the protein to enter a ring-open conforma- In our full-length Rho complex, the primary RNA bind-

ing sites of the N-terminal domains lie equidistant fromtion without perturbing the oligomerization state of the

protein. Because the interior of Rho is capable of associ- each other about the periphery of the hexamer. There

Rho Termination Factor Structure

139

Figure 2. The Rho Hexamer

(A) Front view of the particle. The six subunits

of Rho pack into an open hexameric ring.

Each protomer is represented by a different

color. The orientation of protomer C (yellow

subunit) is similar to that shown in Figure 1A.

(B) Topdown view of (A). Protomers are la-

beled A–F.

of adjacent Rho subunits as they wind about

the pseudo-6-fold axis of the ring (vertical

line). Ovals are colored according to the color

scheme in (A) and (B). The gap between

monomers A and F is 12 A

, and the helical

pitch is 45 A

is clear electron density for nucleic acid in the OB-fold density for nucleic acid outside of the primary binding

site.binding cleft of five of the six N-terminal domains. The

one N-terminal domain that lacks density for the DNA Surprisingly, the arrangement between the N- and

C-terminal regions of each protomer is such that theor RNA substrate belongs to protomer B; the helical

bundle of this domain is also disordered. As anticipated, primary RNA binding clefts face inward, toward the cen-

tral hole of the ring (Figure 3A). This finding was unex-each binding site accommodates an oligo-(deoxy)

cytadylic acid dinucleotide as seen for the isolated Rho pected, because earlier data had suggested that the

primary mRNA binding cleft might reside on the outerN-terminal domain/RNA complex solved previously

(Bogden et al., 1999). There is no observable electron periphery of the ring (Yu et al., 2000). Several lines of

Cell

140

Figure 3. Rho RNA Binding Sites

(A) Molecular surface (GRASP [Nicholls et al.,

1991]) of the Rho hexamer. Primary RNA

binding sites in the OB-fold of the N-terminal

domain are colored cyan. Secondary (C-ter-

minal) RNA binding sites in the ATPase do-

main are colored magenta. Nucleic acid

bound at the primary RNA binding sites is

shown as yellow rods. View is the same as in

Figure 2B.

(B) Schematic of the primary (N-terminal) RNA

binding site configuration. The N- and C-ter-

minal domains are colored green and red,

respectively. Solid black lines represent the

positions for the single-stranded nucleic acid,

which binds across the primary RNA binding

site and orients the 3⬘ end toward the hole of

the ring. The broken black line shows the path

needed to be traversed by nucleic acid be-

tween adjacent binding sites.

diagram of the Rho hexamer showing the lo-

cation of the P loops (blue), the Q loops (ma-

genta) and the R loops (green). View is from

the “bottom,” rotated 180⬚ from the perspec-

tive of Figure 2B.

evidence, however, indicate that the observed configu- the length of an ssRNA sufficient to span the entire

N-terminal periphery of the ring would be approximatelyration represents the natural architecture of the Rho

particle. First, each of the six protomers in the asymmet- 70–80 bases, a value in excellent agreement with ribo-

nuclease A digestion experiments of RNAs bound to theric unit independently adopts the same conformation

and orientation. Second, there are extensive hydropho- Rho hexamer (Bear et al., 1988; Zhu and von Hippel,

1998b, 1998a).bic contacts between the N- and C-terminal domains,

burying a total surface area of ⵑ850 A

per protomer

(Figure 1B). Finally, this arrangement places the ex- Rho and Secondary Site RNA Interactions

mRNA translocation and unwinding are thought to betended linker that connects the N- and C-terminal do-

mains on the exterior of the ring, consistent with prote- catalyzed at Rho’s secondary RNA binding site. This

function depends on two sequence motifs known as thease mapping experiments that have shown this region

to be accessible and labile (Bear et al., 1985; Dolan et Q and R loops (Burgess and Richardson, 2001a; Wei

and Richardson, 2001b, 2001a; Xu et al., 2002). Bothal., 1990).

The relative orientation of the N- and C-terminal do- loops, as predicted from a homology model based on

the F

ATPase (Miwa et al., 1995; Burgess and Richard-mains has important consequences for RNA recogni-

tion. The primary RNA binding sites are arranged such son, 2001a), line the interior hole of the hexamer. Each

Q loop lies on the upper segment of the C-terminalthat they bind substrate at an angle of 75⬚ to a plane

perpendicular to the pseudo 6-fold rotation axis of the domain and extends into the center of the ring. The

constellation of the six Q loops in the hexamer togetherring. The substrate is oriented such that the 3⬘ end points

down, toward the interior hole of the ring (Figure 3B). form the narrowest constriction (diameter 20 A

) of the

interior hole (Figures 3A and 3C). Part of the Q loopThis configuration constrains the minimal distance be-

tween the 3⬘ and 5⬘ ends of successive RNA binding of each protomer is disordered, reflecting the intrinsic

mobility of this region, and may result from an absencesites to be 35 A

. As a result, a 12–13 base RNA oligonu-

cleotide would be minimally required to span two adja- of observable contacts with the nucleic acid substrate.

In contrast to the Q loops, the R loops are implicatedcent N-terminal domains and, in contrast to models sug-

gesting that RNA might directly feed from one domain in both ATP and RNA binding (Xu et al., 2002). Each R

loop resides on a segment located at the subunit-sub-to another (Bogden et al., 1999), any RNA long enough

to bridge two or more protomers would be forced to unit interface between the C-terminal domains, and lies

both adjacent to and above the P loop of the ATP bindingzigzag between the primary sites (Figure 3B). In total,

Rho Termination Factor Structure

141

Figure 4. ATP Associations

(A) Stereo view of specific Rho and AMP-PNP interactions. F

-F

electron density contoured at 2␴ is shown for the bound nucleotide. Residues

important for nucleotide binding and hydrolysis are shown in ball and stick. Residues from two adjacent C-terminal domains form the active

site and are colored blue and green.

(B) Rho nucleotide binding pocket conformation. The two adjacent C-terminal domains are colored red and gray (protomers D and E). The

active site is partly closed and can bind nucleotide, but does not appear fully organized to carry out hydrolysis. The P loop (blue) is shown

next to the R loop motif (green).

ATPase nucleotide binding pocket conformation (1bmf). Two ATPase domains are colored yellow and cyan, and bound nucleotide is

shown as black rods. The active site cleft is fully closed and competent to carry out nucleotide hydrolysis.

pocket (Figures 3C and 4B). Part of each R loop also explaining why RNA appears to be readily captured for

translocation following primary site occupancy (Kim andlines the interior hole of the Rho hexamer. The position

of the R loop and its contact with the P loop indicates Patel, 2001).

that this motif could function as part of an allosteric

effector switch that directly couples RNA binding in the Nucleotide Associations and Rho Function

The ATP binding pocket of Rho is located at the interfacehole of the hexamer to the ATP binding and hydrolysis

site (Richardson and Conaway, 1980; Shigesada and between the C-terminal domains of adjacent protomers

(Figure 4B). Part of this region is formed by signatureWu, 1980; Richardson, 1982; Engel and Richardson,

1984; Kim and Patel, 2001). Consistent with this hypoth- sequence motifs such as the adjacent Walker-A

(GXGXXGK(S/T) and Walker-B (D(D/E)XX) segments. Foresis, RNA binding to the secondary state coincides with

closure of the hexameric ring and stimulation of the native crystals grown in the presence of AMPPNP, clear

electron density for bound nucleotide is seen in all sixATPase activity (Gogol et al., 1991; Gan and Richard-

son, 1999), presumably by introducing conformational ATP binding pockets of the hexameric ring (Figure 4A).

In contrast, weak electron density is observed for nucle-changes between subunits and residues around the ATP

binding site (Lowery-Goldhammer and Richardson, otide in the structure determined with the selenomethio-

nine protein, indicating that the occupancy at this site1974). It is interesting to note that the positioning of the

primary RNA binding sites places the 3⬘ RNA end near is low. This difference appears due to the insertion of

the hydrophobic side chain SeMet186 of the seleno-the secondary sites from the outset of mRNA binding,

Cell

142

Figure 5. Schematic Model for Rho Function

Numbers correspond to stages outlined in the

text. Asterisks represent catalytic sites

thought to be competent for ATP hydrolysis.

methionine protein into the adenine binding pocket of ing both an entry point into what would otherwise be

the topologically confined interior of the ring and sug-

each of the six subunits. Nonetheless, superposition

gesting a means by which the protein may enter a nu-

of the ATP bound and -free structures shows that the

cleic acid “loading” configuration. This general organi-

organization and conformation of the ATP binding sites

zation provides an elegant means to ensure that target

are the same.

mRNAs are efficiently captured by the protein, thereby

Cocrystallization (as opposed to soaking) of Rho with

activating the particle and initiating the process of tran-

AMPPNP shows that nucleotide can associate with an

scription termination.

open-ring state of the protein. The interactions between

The significance of the Rho open-ring state may be

Rho and the adenosine base of bound AMPPNP are

applicable to other toroidal helicases and translocases.

typical of those seen for other RecA-type proteins, in

A long-standing puzzle has been to understand how

which aliphatic or aromatic side chains sandwich the

extended nucleic acid segments gain entry to the interior

purine moiety in a “hydrophobic pincer” (Figure 4A)

of such protein rings, where unwinding and transloca-

(Boyer, 1997; Bird et al., 1998a; Singleton et al., 2000;

tion are thought to occur. Many helicases, such as DnaB,

Xu et al., 2000). Other nucleotide associations are not

papilloma virus E1, and the MCM complex, use special-

canonical, however. For example, residues in the

ized loading proteins to accomplish this task (Lusky et

Walker-A and -B regions do not make standard contacts

al., 1993; Seo et al., 1993; Donovan et al., 1997; Tanaka

to the ␤- and ␥-phosphates of AMPPNP, nor is magne-

et al., 1997; Weinreich et al., 1999; Barcena et al., 2001).

sium evident in our structure (Figure 4A). These altered

Others, such as Rho, appear to load onto target sub-

features arise because the subunit-subunit orientation

strates without accessory factors. The ability of Rho to

of the open ring splays apart the ATPase domains as

spontaneously switch between open- and closed-ring

compared to other RecA-type proteins that have formed

structures provides an immediate solution to the loading

competent active sites (Figures 4B and 4C). Given that

problem (Gogol et al., 1991; Yu et al., 2000; Richardson,

the open-ring form of Rho imaged here appears consis-

2002). By utilizing intersubunit hinging motions evident

tent with an mRNA loading state, as opposed to an

in a number of RecA-type ATPases, our structure shows

actively translocating helicase, such atypical nucleotide

how Rho can configure the relative orientation of neigh-

associations are perhaps not surprising, although these

boring RecA-type folds to allow ring opening without

data indicate that ring opening can still occur when Rho

fully exposing subunit interfaces that might otherwise

is bound to ATP, as would be expected when the protein

foster filamentation. As a result, the Rho model may be

operates in vivo.

applicable for understanding how other ring systems

are breached to bind nucleic acids, and may potentially

Rho Architecture and Mechanistic Implications

represent a structural state accessed for some helicases

As a collective, the structures observed here explain

through loader protein activity.

several important features of Rho function. First, Rho’s

Another interesting finding is that the Rho hexamer

primary RNA binding sites line the perimeter of the hex-

can remain open and, thus, competent to bind mRNA

amer and face inward toward the center of the particle.

at its secondary binding sites even when Rho’s ATP and

Second, the primary sites orient the 3⬘ end of nucleic

primary mRNA binding sites are occupied. This observa-

acid target sequences into the interior hole, which is

tion allows us to expand upon models explaining how

formed by motifs known to be critical for coupling mRNA

Rho recognizes rut sites and loads onto RNA in vivo

translocation and unwinding to ATP turnover. Third, the

(Richardson, 2002; Delagoutte and von Hippel, 2003).

Extensive previous work, together with our structuralring of the hexamer is split by a gap ⵑ12 A

wide, provid-

Rho Termination Factor Structure

143

the native protein overexpressed in 2XYT media by induction at an

data, suggest that Rho can oscillate between closed-

600

of 0.3–0.5 with 1 mM IPTG at 37⬚C for 3 hr. Cells were lysed

and open-ring states, either of which may associate with

by sonication in 50 mM Tris-HCl (pH 7.5), 10% glycerol, 50 mM KCl,

ATP and target mRNAs (Figure 5, stages 1 and 2). In the

1 mM TCEP (Tris(2-carboxyethyl)phosphine hydrochloride) on ice.

open state, ATP would be loosely associated with Rho

The protein was twice purified over a POROS-HS column (Perseptive

and only partially liganded as observed here. In contrast,

Biosystems) then further purified with two successive Sephacryl

S-300 (Amersham) sizing columns in 10 mM Tris-HCl (pH 7.5), 50

entry into the closed state would properly position and

mM NaCl, and 1 mM TCEP. The protein at this stage was more than

coordinate nucleotide for hydrolysis. Natural conversion

99% pure as judged by SDS-PAGE and Coomassie staining and

between the two structural states would allow ATP hy-

was monodisperse as judged by dynamic light scattering (DynaPro,

drolysis to be observed in the absence of RNA (Stitt,

model 99-CP, Protein Solutions). Protein was concentrated by ultra-

2001), but turnover would not be fully activated, both

filtration (Centriprep-10, Amicon) to 15 mg/ml. Selenomethionine

because of the lack of RNA effector signals through

protein was overexpressed in minimal media using the protocol of

Van Duyne and coworkers (Van Duyne et al., 1993) and was purified

the Q and R loops and because a population of Rho

similar to native Rho.

molecules might be open at any given moment.

Loading and priming of Rho would begin through the

Crystallization and Data Collection

association of a rut site with the primary RNA binding

Initial crystals of full-length Rho diffracted poorly. Cocrystals with

sites in the N-terminal domains. The configuration of

DNA or RNA substrates resulted in crystals of different morphology

the N-terminal domains automatically orients the 3⬘ end

to those obtained in the absence of nucleic acid and proved instru-

of message toward the interior hole of the protein. Were

mental in solving the Rho structure. A number of single-stranded

the ring to be open during rut site binding, mRNA could

DNA and RNA substrates were tested in cocrystallization trials.

DNAs and RNAs varied in length (6–19 bases) and composition (one

simply thread its way through the gap between mono-

or two binding sites), and were designed based on Rho’s specificity

mers and into the interior of the hexamer to associate

for the unstructured, cytosine-rich mRNA regions present in rut

with the secondary RNA binding sites (Figure 5, stage

sites. The best quality crystals in terms of size, stability, and diffrac-

4). In contrast, if the ring were closed upon rut site

tion were those cocrystallized with a 15-mer DNA substrate (AACC

recognition (Figure 5, stage 3), mRNA entry would need

CAAGAACCCAA) or with an 8-mer RNA substrate (CU)

to wait until the ring spontaneously opens. It remains a

Cocrystals of Rho in complex with the single-stranded DNA sub-

strate were grown by vapor diffusion at room temperature. Stock

possibility that mRNA binding to the primary, N-terminal

protein solutions were dialyzed in 10 mM Tris-HCl (pH 7.5) and 50

domain binding sites might directly favor ring opening

mM NaCl at 4⬚C prior to crystallization. One volume of protein solu-

(Yu et al., 2000; Kim and Patel, 2001).

tion was mixed with one volume of crystallization solution containing

Once bound in the interior, RNA would be sensed by

100 mM Na•Cacodylate (pH 6.5), 100 mM NaCl, 5% PEG 8K, 40%

the Q and R loop regions, allosterically activating Rho

glycerol, 0.6 mM n-Nonyl-␤-D-thiomaltoside, and 2 mM TCEP; the

and converting the protein into a stable, closed configu-

well solution was diluted with an equivalent amount of H

O prior to

sealing the drop over the reservoir. Small crystals appeared over-

ration (Figure 5, stage 5). Based on comparisons with

night and grew to average dimensions of 100 ⫻ 100 ⫻ 200 ␮Min

closed-ring RecA-type proteins, ring closure would pre-

1 week. Crystals were introduced into cryoprotectant, containing a

sumably occur through conformational changes be-

mix of 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 3% PEG 8K, 25%

tween a subset of the ATPase domains, both forming

glycerol, and 5% PEG 400, in a fast, single step, then flash frozen

competent set of active sites and compensating for the

in liquid nitrogen. The crystals belong to the monoclinic space group

15⬚ subunit-subunit angular rise observed for the open-

C2 with unit cell dimensions of a ⫽ 119.2 A

,b⫽ 205.8 A

,c⫽ 148.0 A

and ␤⫽95.3⬚. There is one hexamer per asymmetric unit.

ring form. Ensuing cycles of ATP turnover would serve to

modulate the relative orientation of neighboring ATPase

domains, providing directed structural changes to pro-

Structure Determination and Refinement

All data were collected at Beamline 8.3.1 at the Advanced Light

pel the protein 5⬘→3⬘ along the message (Brennan et

Source (ALS) using an ADSC Quantum-Q210 CCD detector. Diffrac-

al., 1987; Singleton et al., 2000). As translocation begins,

tion data were processed using DENZO and SCALEPACK (Otwinow-

Rho could either dissociate from the rut site or could

ski, 1997) (Table 1). Initial attempts to calculate phases using seleno-

remain bound to the target motif as predicted by the

methionine MAD or SAD experiments were unsuccessful, so a

tethered tracking model (Figure 5, stages 6 and 7). Al-

tantalum derivative was prepared by soaking a crystal overnight

with saturated solution of Ta

⫹

. Heavy atom positions were

though our structure cannot rule out either mechanism,

identified by SOLVE (Terwilliger and Berendzen, 1999), and

it is interesting to note that the N-terminal domains open

MLPHARE (CCP4, 1994) was used to calculate initial phases to 6.0 A

out from the center of the ring, creating a wide depres-

based on isomorphous and anomalous differences for this derivative

sion across the top of the particle. The size of this de-

(Table 1). Selenium sites (83 out of 96) were located by calculating

pression appears ample enough to remain associated

anomalous difference maps to 3.6 A

resolution using the tantalum

with a RNA segment while single-stranded RNA is

phases (Table 1). The selenium sites were refined initially with

MLPHARE and then with SHARP (de la Fortelle and Bricogne, 1997),

spooled away from the protein during translocation. Fu-

followed by DM solvent flattening (Cowtan, 1994) to yield starting

ture efforts focused toward assembling RNA bound,

electron density maps.

closed ring structures will be needed to help distinguish

Five out of the six N-terminal domains were located in the solvent-

between these models and to better illuminate the physi-

flattened density by the program FFFEAR (Cowtan K., 1998) using

cal coupling between ATP usage, translocation, and un-

the X-ray model 2a8v (Bogden et al., 1999). The sixth domain was

winding.

placed into the map manually and the fit optimized using RSR-RIGID

in O (Jones T.A., 1991). The N-terminal domains were used to derive

the six NCS (noncrystallographic symmetry) operators within theExperimental Procedures

asymmetric unit and the experimental electron density map was

subsequently improved using solvent flattening with NCS multido-Protein Expression and Purification

The open reading frame encoding the full-length Escherichia coli main averaging and phase extension in DM. The C-terminal domain

of the mitochondrial F

ATPase model (1bmf) was manually dockedRho protein was PCR amplified and cloned into pET24b. The plasmid

was transformed into the Escherichia coli strain BL21 (pLysS), and and rebuilt into this DM-averaged density using O.

Cell

144

The model was refined with REFMAC5 (Murshudov G.N., 1997), Brennan, C.A., Steinmetz, E.J., Spear, P., and Platt, T. (1990). Speci-

ficity and efficiency of rho-factor helicase activity depends on mag-using ARP (Lamzin and Wilson, 1993) for water building. In the final

cycles of refinement, TLS (translation libration and screw-rotation) nesium concentration and energy coupling to NTP hydrolysis. J.

Biol. Chem. 265, 5440–5447.refinement of rigid groups (Winn et al., 2001) was carried out as

implemented in REFMAC5 (Table 2). The refined selenomethionine

Briercheck, D.M., Wood, T.C., Allison, T.J., Richardson, J.P., and

Rho/DNA structure was used to solve the native Rho/RNA•AMPPNP

Rule, G.S. (1998). The NMR structure of the RNA binding domain of

complex by molecular replacement using the program AMORE (Na-

E. coli rho factor suggests possible RNA-protein interactions. Nat.

vaza, 2001). The resultant solution was DM solvent flattened and

Struct. Biol. 5, 393–399.

multidomain averaged, revealing clear electron density in F

-F

maps

Brown, S., Brickman, E.R., and Beckwith, J. (1981). Blue ghosts: a

for AMPPNP in all six ATP binding sites. The structure was then

new method for isolating amber mutants defective in essential genes

refined using REFMAC5. The Rho/DNA and the Rho/RNA structures

of Escherichia coli. J. Bacteriol. 146, 422–425.

were refined to a final R

work

27.0% and 27.0% and an R

free

29.6%

Burgess, B.R., and Richardson, J.P. (2001a). RNA passes through

and 30.3%, respectively. A total of 2472 out of 2502 residues are

the hole of the protein hexamer in the complex with the Escherichia

accounted for in either of the two structures (Table 2). Geometric

coli Rho factor. J. Biol. Chem. 276, 4182–4189.

analysis were carried out with Procheck (Laskowski et al., 1993)

and structural superpositions with LSQKAB (CCP4, 1994).

Burgess, B.R., and Richardson, J.P. (2001b). Transcription factor

Rho does not require a free end to act as an RNA-DNA helicase on

Acknowledgments

an RNA. J. Biol. Chem. 276, 17106–17110.

Cavarelli, J., Rees, B., Ruff, M., Thierry, J.C., and Moras, D. (1993).

The authors are grateful to James Holton at Beamline 8.3.1 of the

Yeast tRNA(Asp) recognition by its cognate class II aminoacyl-tRNA

Advanced Light source for assistance with data acquisition and to

synthetase. Nature 362, 181–184.

David King for mass spectroscopy analyses. We would also like to

CCP4 (Collaborative Computational Project 4) (1994). The CCP4

thank James Keck, Deborah Fass, David Akey, Jan Erzberger, and

suite: programs for protein crystallography. Acta Crystallogr. D 50,

Scott Gradia for critical reading the manuscript, as well as members

760–763.

of the Berger Lab for helpful discussions and insights. This work

Chen, C.Y., Galluppi, G.R., and Richardson, J.P. (1986). Transcrip-

was supported by generous assistance from the G. Harold and Leila

tion termination at lambda tR1 is mediated by interaction of rho with

Y. Mathers Charitable Foundation.

specific single-stranded domains near the 3⬘ end of cro mRNA. Cell

46, 1023–1028.

Received: March 19, 2003

Revised: May 20, 2003

Cowtan, K. (1994). A CCP4 density modification package. Joint

Accepted: June 25, 2003

CCP4 ESF-EACBM. Newslett. Prot. Crystallogr. 31, 34–38.

Published: July 10, 2003

Cowtan, K. (1998). Modified phased translation functions and their

application to molecular fragment location. Acta Crystallogr. D 54,

References

750–756.

de la Fortelle, E., and Bricogne, G. (1997). Maximum-likelihood heavy

Abrahams, J.P., Leslie, A.G., Lutter, R., and Walker, J.E. (1994).

atom parameter refinement for multiple isomorphous replacement

Structure at 2.8 A

resolution of F1-ATPase from bovine heart mito-

and multiwavelength anomalous diffraction methods. Methods En-

chondria. Nature 370, 621–628.

zymol. 276, 472–494.

Alifano, P., Rivellini, F., Limauro, D., Bruni, C.B., and Carlomagno,

Delagoutte, E., and von Hippel, P.H. (2003). Helicase mechanisms

M.S. (1991). A consensus motif common to all Rho-dependent pro-

and the coupling of helicases within macromolecular machines. Part

karyotic transcription terminators. Cell 64, 553–563.

II: integration of helicases into cellular processes. Q. Rev. Biophys.

Allison, T.J., Wood, T.C., Briercheck, D.M., Rastinejad, F., Richard-

36, 1–69.

son, J.P., and Rule, G.S. (1998). Crystal structure of the RNA-binding

Dolan, J.W., Marshall, N.F., and Richardson, J.P. (1990). Transcrip-

domain from transcription termination factor rho. Nat. Struct. Biol.

tion termination factor rho has three distinct structural domains. J.

5, 352–356.

Biol. Chem. 265, 5747–5754.

Barcena, M., Ruiz, T., Donate, L.E., Brown, S.E., Dixon, N.E., Rader-

Donovan, S., Harwood, J., Drury, L.S., and Diffley, J.F. (1997).

macher, M., and Carazo, J.M. (2001). The DnaB.DnaC complex: a

Cdc6p-dependent loading of Mcm proteins onto pre-replicative

structure based on dimers assembled around an occluded channel.

chromatin in budding yeast. Proc. Natl. Acad. Sci. USA 94, 5611–

EMBO J. 20, 1462–1468.

5616.

Bear, D.G., Andrews, C.L., Singer, J.D., Morgan, W.D., Grant, R.A.,

Engel, D., and Richardson, J.P. (1984). Conformational alterations

von Hippel, P.H., and Platt, T. (1985). Escherichia coli transcription

of transcription termination protein rho induced by ATP and by RNA.

termination factor rho has a two-domain structure in its activated

Nucleic Acids Res. 12, 7389–7400.

form. Proc. Natl. Acad. Sci. USA 82, 1911–1915.

Galluppi, G.R., and Richardson, J.P. (1980). ATP-induced changes

Bear, D.G., Hicks, P.S., Escudero, K.W., Andrews, C.L., McSwiggen,

in the binding of RNA synthesis termination protein Rho to RNA. J.

J.A., and von Hippel, P.H. (1988). Interactions of Escherichia coli

Mol. Biol. 138, 513–539.

transcription termination factor rho with RNA. II. Electron micros-

copy and nuclease protection experiments. J. Mol. Biol. 199,

Gan, E., and Richardson, J.P. (1999). ATP and other nucleotides

623–635.

stabilize the Rho-mRNA complex. Biochemistry 38, 16882–16888.

Bird, L.E., Brannigan, J.A., Subramanya, H.S., and Wigley, D.B.

Geiselmann, J., Yager, T.D., and von Hippel, P.H. (1992). Functional

(1998a). Characterisation of Bacillus stearothermophilus PcrA heli-

interactions of ligand cofactors with Escherichia coli transcription

case: evidence against an active rolling mechanism. Nucleic Acids

termination factor rho. II. Binding of RNA. Protein Sci. 1, 861–873.

Res. 26, 2686–2693.

Geiselmann, J., Wang, Y., Seifried, S.E., and von Hippel, P.H. (1993).

Bird, L.E., Subramanya, H.S., and Wigley, D.B. (1998b). Helicases:

A physical model for the translocation and helicase activities of

a unifying structural theme? Curr. Opin. Struct. Biol. 8, 14–18.

Escherichia coli transcription termination protein Rho. Proc. Natl.

Acad. Sci. USA 90, 7754–7758.

Bogden, C.E., Fass, D., Bergman, N., Nichols, M.D., and Berger,

J.M. (1999). The structural basis for terminator recognition by the

Gogol, E.P., Seifried, S.E., and von Hippel, P.H. (1991). Structure

Rho transcription termination factor. Mol. Cell 3, 487–493.

and assembly of the Escherichia coli transcription termination factor

rho and its interaction with RNA. I. Cryoelectron microscopic stud-

Boyer, P.D. (1997). The ATP synthase–a splendid molecular ma-

ies. J. Mol. Biol. 221, 1127–1138.

chine. Annu. Rev. Biochem. 66, 717–749.

Brennan, C.A., Dombroski, A.J., and Platt, T. (1987). Transcription Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991).

Improved methods for building protein models in electron densitytermination factor rho is an RNA-DNA helicase. Cell 48, 945–952.

Rho Termination Factor Structure

145

maps and the location of errors in these models. Acta Crystallogr. simultaneous interaction at two kinds of nucleic acid-binding sites.

J. Biol. Chem. 257, 5760–5766.A 47, 110–119.

Richardson, J.P., and Conaway, R. (1980). Ribonucleic acid release

Kim, D.E., and Patel, S.S. (2001). The kinetic pathway of RNA binding

activity of transcription termination protein rho is dependent on the

to the Escherichia coli transcription termination factor Rho. J. Biol.

hydrolysis of nucleoside triphosphates. Biochemistry 19, 4293–

Chem. 276, 13902–13910.

4299.

Lamzin, V.S., and Wilson, K.S. (1993). Automated refinement of pro-

Richardson, J.P., and Greenblatt, J. (1996). Escherichia coli and

tein models. Acta Crystallogr. D 49, 129–147.

salmonella. In Cellular and Molecular Biology, Second Edition, F.C.

Laskowski, A.R., MacArthur, W.M., Moss, S.D., and Thornton, M.J.

Neidhardt, R. Curtiss III, J.L. Ingraham, E.C.C. Lin, K.B. Low, B.

(1993). PROCHECK: a program to check the stereochemical quality

Magasanik, W.S. Reznikoff, M. Riley, M. Schaechter, and H.E. Um-

of protein structures. J. Appl. Cryst. 26, 283–291.

barger, eds. (Washington, D.C.: American Society for Microbiology),

Lowery-Goldhammer, C., and Richardson, J.P. (1974). An RNA-

pp. 822–848.

dependent nucleoside triphosphate phosphohydrolase (ATPase)

Richardson, L.V., and Richardson, J.P. (1992). Cytosine nucleoside

associated with rho termination factor. Proc. Natl. Acad. Sci. USA

inhibition of the ATPase of Escherichia coli termination factor rho:

71, 2003–2007.

evidence for a base specific interaction between rho and RNA. Nu-

Lusky, M., Hurwitz, J., and Seo, Y.S. (1993). Cooperative assembly

cleic Acids Res. 20, 5383–5387.

of the bovine papilloma virus E1 and E2 proteins on the replication

Roberts, J.W. (1969). Termination factor for RNA synthesis. Nature

origin requires an intact E2 binding site. J. Biol. Chem. 268, 15795–

224, 1168–1174.

15803.

Sawaya, M.R., Guo, S., Tabor, S., Richardson, C.C., and Ellenberger,

Martinez, A., Burns, C.M., and Richardson, J.P. (1996a). Residues

T. (1999). Crystal structure of the helicase domain from the replica-

in the RNP1-like sequence motif of Rho protein are involved in RNA-

tive helicase-primase of bacteriophage T7. Cell 99, 167–177.

binding affinity and discrimination. J. Mol. Biol. 257, 909–918.

Schindelin, H., Jiang, W., Inouye, M., and Heinemann, U. (1994).

Martinez, A., Opperman, T., and Richardson, J.P. (1996b). Mutational

Crystal structure of CspA, the major cold shock protein of Esche-

analysis and secondary structure model of the RNP1-like sequence

richia coli. Proc. Natl. Acad. Sci. USA 91, 5119–5123.

motif of transcription termination factor Rho. J. Mol. Biol. 257,

Seo, Y.S., Muller, F., Lusky, M., Gibbs, E., Kim, H.Y., Phillips, B.,

895–908.

and Hurwitz, J. (1993). Bovine papilloma virus (BPV)-encoded E2

McSwiggen, J.A., Bear, D.G., and von Hippel, P.H. (1988). Interac-

protein enhances binding of E1 protein to the BPV replication origin.

tions of Escherichia coli transcription termination factor rho with

Proc. Natl. Acad. Sci. USA 90, 2865–2869.

RNA. I. Binding stoichiometries and free energies. J. Mol. Biol. 199,

Shigesada, K., and Wu, C.W. (1980). Studies of RNA release reaction

609–622.

catalyzed by E. coli transcription termination factor rho using iso-

Miwa, Y., Horiguchi, T., and Shigesada, K. (1995). Structural and

lated ternary transcription complexes. Nucleic Acids Res. 8, 3355–

functional dissections of transcription termination factor rho by ran-

3369.

dom mutagenesis. J. Mol. Biol. 254, 815–837.

Singleton, M.R., Sawaya, M.R., Ellenberger, T., and Wigley, D.B.

Modrak, D., and Richardson, J.P. (1994). The RNA-binding domain

(2000). Crystal structure of T7 gene 4 ring helicase indicates a mech-

of transcription termination factor rho: isolation, characterization,

anism for sequential hydrolysis of nucleotides. Cell 101, 589–600.

and determination of sequence limits. Biochemistry 33, 8292–8299.

Steinmetz, E.J., and Platt, T. (1994). Evidence supporting a tethered

Morgan, W.D., Bear, D.G., and von Hippel, P.H. (1983). Rho-depen-

tracking model for helicase activity of Escherichia coli Rho factor.

dent termination of transcription. I. Identification and characteriza-

Proc. Natl. Acad. Sci. USA 91, 1401–1405.

tion of termination sites for transcription from the bacteriophage

Stitt, B.L. (2001). Escherichia coli transcription termination factor

lambda PR promoter. J. Biol. Chem. 258, 9553–9564.

Rho binds and hydrolyzes ATP using a single class of three sites.

Morgan, W.D., Bear, D.G., Litchman, B.L., and von Hippel, P.H.

Biochemistry 40, 2276–2281.

(1985). RNA sequence and secondary structure requirements for

Story, R.M., Weber, I.T., and Steitz, T.A. (1992). The structure of the

rho-dependent transcription termination. Nucleic Acids Res. 13,

E. coli recA protein monomer and polymer. Nature 355, 318–325.

3739–3754.

Sullivan, S.L., and Gottesman, M.E. (1992). Requirement for E. coli

Murshudov G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement

NusG protein in factor-dependent transcription termination. Cell 68,

of macromolecular structures by the maximum-likelihood method.

989–994.

Acta Crystallogr. D 53, 240–255.

Tanaka, T., Knapp, D., and Nasmyth, K. (1997). Loading of an Mcm

Navaza, J. (2001). Implementation of molecular replacement in

protein onto DNA replication origins is regulated by Cdc6p and

AMoRe. Acta Crystallogr. D Biol. Crystallogr. 57, 1367–1372.

CDKs. Cell 90, 649–660.

Newkirk, K., Feng, W., Jiang, W., Tejero, R., Emerson, S.D., Inouye,

Terwilliger, T.C., and Berendzen, J. (1999). Automated MAD and

M., and Montelione, G.T. (1994). Solution NMR structure of the major

MIR structure solution. Acta Crystallogr. D 55, 849–861.

cold shock protein (CspA) from Escherichia coli: identification of a

Van Duyne, G.D., Standaert, R.F., Karplus, P.A., Schreiber, S.L.,

binding epitope for DNA. Proc. Natl. Acad. Sci. USA 91, 5114–5118.

and Clardy, J. (1993). Atomic structures of the human immunophilin

Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and

FKBP-12 complexes with FK506 and rapamycin. J. Mol. Biol. 229,

association: insights from the interfacial and thermodynamic prop-

105–124.

erties of hydrocarbons. Proteins 11, 281–296.

von Hippel, P.H., and Delagoutte, E. (2001). A general model for

Oda, T., and Takanami, M. (1972). Observations on the structure of

nucleic acid helicases and their “coupling” within macromolecular

the termination factor rho and its attachment to DNA. J. Mol. Biol.

machines. Cell 104, 177–190.

71, 799–802.

Walstrom, K.M., Dozono, J.M., Robic, S., and von Hippel, P.H. (1997).

Opperman, T., and Richardson, J.P. (1994). Phylogenetic analysis

Kinetics of the RNA-DNA helicase activity of Escherichia coli tran-

of sequences from diverse bacteria with homology to the Esche-

scription termination factor rho. 1. Characterization and analysis of

richia coli rho gene. J. Bacteriol. 176, 5033–5043.

the reaction. Biochemistry 36, 7980–7992.

Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction

Wei, R.R., and Richardson, J.P. (2001a). Identification of an RNA-

data collected in oscillation mode. Methods Enzymol. 276, 307–326.

binding Site in the ATP binding domain of Escherichia coli Rho by

Platt, T. (1994). Rho and RNA: models for recognition and response.

H2O2/Fe-EDTA cleavage protection studies. J. Biol. Chem. 276,

Mol. Microbiol. 11, 983–990.

28380–28387.

Richardson, J. (2002). Rho-dependent termination and ATPases in

Wei, R.R., and Richardson, J.P. (2001b). Mutational changes of con-

transcript termination. Biochim. Biophys. Acta 1577, 251–260.

served residues in the Q-loop region of transcription factor Rho greatly

reduce secondary site RNA-binding. J. Mol. Biol. 314, 1007–1015.Richardson, J.P. (1982). Activation of rho protein ATPase requires

Cell

146

Weinreich, M., Liang, C., and Stillman, B. (1999). The Cdc6p nucleo-

tide-binding motif is required for loading mcm proteins onto chroma-

tin. Proc. Natl. Acad. Sci. USA 96, 441–446.

Winn, M.D., Isupov, M.N., and Murshudov, G.N. (2001). Use of TLS

parameters to model anisotropic displacements in macromolecular

refinement. Acta Crystallogr. D 57, 122–133.

Xu, H., Frank, J., Niedenzu, T., and Saenger, W. (2000). DNA helicase

RepA: cooperative ATPase activity and binding of nucleotides. Bio-

chemistry 39, 12225–12233.

Xu, Y., Kohn, H., and Widger, W.R. (2002). Mutations in the rho

transcription termination factor that affect RNA tracking. J. Biol.

Chem. 277, 30023–30030.

Yu, X., Horiguchi, T., Shigesada, K., and Egelman, E.H. (2000). Three-

dimensional reconstruction of transcription termination factor rho:

orientation of the N-terminal domain and visualization of an RNA-

binding site. J. Mol. Biol. 299, 1279–1287.

Zhu, A.Q., and von Hippel, P.H. (1998a). Rho-dependent termination

within the trp t’ terminator. I. Effects of rho loading and template

sequence. Biochemistry 37, 11202–11214.

Zhu, A.Q., and von Hippel, P.H. (1998b). Rho-dependent termination

within the trp t’ terminator. II. Effects of kinetic competition and rho

processivity. Biochemistry 37, 11215–11222.

Accession Numbers

Coordinates for both Rho complexes have been deposited in the

RCSB PDB database and are available under the accession codes

1PV4 and 1PVO.