Ligand-induced and small-molecule control ofsubstrate loading in a hexameric helicaseMichael R. Lawsona, Kevin Dyerb, and James M. Bergera,c,1

aDepartment of Molecular and Cell Biology, University of California, Berkeley, CA 94720; bLawrence Berkeley National Laboratory, Berkeley, CA 94720;and cDepartment of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, MD 21205

Contributed by James M. Berger, October 12, 2016 (sent for review August 15, 2016; reviewed by Brandt F. Eichman and Marcelo Nollmann)

Processive, ring-shaped protein and nucleic acid protein translocasescontrol essential biochemical processes throughout biology and areconsidered high-prospect therapeutic targets. The Escherichia coli Rhofactor is an exemplar hexameric RNA translocase that terminates tran-scription in bacteria. As with many ring-shaped motor proteins, Rhoactivity is modulated by a variety of poorly understood mechanisms,including small-molecule therapeutics, protein–protein interactions,and the sequence of its translocation substrate. Here, we establishthe mechanism of action of two Rho effectors, the antibiotic bicyclo-mycin and nucleic acids that bind to Rho’s primary RNA recruitmentsite. Using small-angle X-ray scattering and a fluorescence-basedassay to monitor the ability of Rho to switch between open-ring(RNA-loading) and closed-ring (RNA-translocation) states, we foundbicyclomycin to be a direct antagonist of ring closure. Reciprocally,the binding of nucleic acids to its N-terminal RNA recruitment do-mains is shown to promote the formation of a closed-ring Rho state,with increasing primary-site occupancy providing additive stimula-tory effects. This study establishes bicyclomycin as a conformationalinhibitor of Rho ring dynamics, highlighting the utility of developingassays that read out protein conformation as a prospective screen-ing tool for ring-ATPase inhibitors. Our findings further show thatthe RNA sequence specificity used for guiding Rho-dependent ter-mination derives in part from an intrinsic ability of the motor tocouple the recognition of pyrimidine patterns in nascent transcriptsto RNA loading and activity.

antibiotic | ATPase | helicase | motor protein | transcription

Ring-shaped hexameric helicases and translocases are motorproteins that control myriad essential viral and cellular pro-

cesses. Many hexameric motors undergo substrate-dependentconformational changes that couple activity to the productivebinding of client substrates (1–4). Internal regulatory domains andexogenous proteins or small molecules frequently impact clientsubstrate recruitment and engagement by these enzymes (5–8);however, it is generally unclear how such factors control helicaseor translocase dynamics.Rho is a hexameric helicase responsible for controlling ∼20% of

all transcription termination events in Escherichia coli (9). Rho isinitially recruited to nascent transcripts in an open, lock washer-sha-ped configuration (Fig. 1A) (10, 11), where it binds preferentially topyrimidine-rich sequences (termed “Rho utilization of termination”sequences, or “rut” sites) using a primary RNA-binding site located inthe N-terminal OB folds of the hexamer (12–14). Following rut rec-ognition, Rho converts into a closed-ring form (Fig. 1B), locking theRNA strand into a secondary RNA-binding site formed by twoconserved sequence elements known as the “Q” and “R” loops (15)located within the central pore of the hexamer. This conformationalchange, which we show in an accompanying paper to be both RNA-and ATP-dependent (16), rearranges residues in the Rho ATP-binding pockets into a hydrolysis-competent state (17). Onceengaged, Rhomaintains primary-site contacts with the rut sequence asit translocates 5′ to 3′ along the RNA strand in an ATP-dependentmanner, a process known as “tethered tracking” (18–21). Rho elicitstermination by applying direct or indirect forces to RNA polymerase(22, 23), dislodging it from DNA and the newly made RNA in thetranscription bubble.

The basic steps of Rho-dependent transcription terminationhave been largely elucidated, and it has become increasingly clearthat Rho’s activity, like that of many helicases and translocases,can be controlled by a variety of intrinsic and extrinsic factors. Thesmall molecule bicyclomycin (Fig. 1C), a highly specific chemicalinhibitor of Rho that has been shown to wedge itself in betweensubunits of an open Rho ring (Fig. 1D and Fig. S1) (24), is onesuch example. Previous studies have shown that bicyclomycin is anoncompetitive inhibitor of Rho ATPase activity and a mixedinhibitor of RNA binding to Rho’s secondary site (25, 26). Al-though subsequent structural studies suggested that bicyclomycinantagonizes Rho by sterically preventing the binding of the nu-cleophilic water molecule that initiates ATP hydrolysis (24), howbicyclomycin might interact with catalytically competent Rhostates, such as those thought to accompany ATPase activity andtranslocation (17), has not been defined.It is similarly unclear how other dissociable factors, e.g., regu-

latory proteins and/or nucleic acids, influence Rho activity. It iswell established that the sequence of the RNA itself has a pro-nounced impact on whether a transcript will be acted upon by Rho(12, 27–29). Binding of pyrimidine-rich sequences to theN-terminal RNA-binding domains of Rho is a particularlywell-known accelerant of Rho’s ATPase activity (30), with pre–steady-state ATPase assays showing that the formation of acatalytically competent Rho ring is governed by a rate-limitingRNA- and ATP-dependent conformational change (8). The ligand-dependence of this isomerization event correlates with the re-quirements for ring closure identified in an accompanying study(16). These findings raise the intriguing possibility that the se-quence specificity of Rho-dependent termination is caused in partby an increased efficiency of ring closure when the N-terminalRNA-binding domains are occupied by pyrimidine-rich sequences.Here we show that bicyclomycin and primary-site occupancy af-

fect the structural state of Rho in an opposing manner. Using

Significance

Many processive, ring-ATPase motor proteins rely on substrate-dependent conformational changes to assist with the loading ofclient substrates into the central pore of the enzyme and sub-sequent translocation. Using the Escherichia coli Rho transcriptionterminator as a model hexameric helicase, we show that twodistinct ligands—the antibiotic bicyclomycin and pyrimidine-richnucleic acids—respectively repress or promote the transition ofRho from an open RNA-loading configuration to a closed-ringactive helicase. Our findings explain several mechanisms bywhich Rho activity is controlled and provide a general illustrationof how intrinsic and extrinsic factors can regulate ring-typeATPase dynamics through diverse mechanisms.

Author contributions: M.R.L. and J.M.B. designed research; M.R.L. and K.D. performedresearch; M.R.L. and J.M.B. analyzed data; and M.R.L. and J.M.B. wrote the paper.

Reviewers: B.F.E., Vanderbilt University; and M.N., Centre Nationale de la RechercheScientifique.

The authors declare no conflict of interest.1To whom correspondence should be addressed. Email: jmberger@jhmi.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1616749113/-/DCSupplemental.

13714–13719 | PNAS | November 29, 2016 | vol. 113 | no. 48 www.pnas.org/cgi/doi/10.1073/pnas.1616749113

Dow

nloa

ded

by g

uest

on

May

22,

202

1

small-angle X-ray scattering (SAXS) and a fluorescence-based assayto track ring status in vitro, we demonstrate that bicyclomycin in-hibits RNA binding to the central pore of Rho by sterically im-peding ring closure. In contrast, the binding of pyrimidine-richnucleic acids to the Rho N-terminal RNA-binding domains pro-motes Rho ring closure, aiding the capture of nonideal, purine-richRNA sequences within Rho’s translocation pore. Collectively, thesefindings help highlight diverse mechanisms by which ligand bindingto discrete sites on ring-type ATPases can activate or repress con-formational changes and govern the function of the motor.

ResultsThe Bicyclomycin-Binding Pocket Collapses upon Rho Ring Closure.During the course of examining the bicyclomycin-binding pocketacross different Rho structural states (17, 16), we noticed that thesite appeared substantially smaller in closed-ring states than in theopen-ring states. Computational analysis confirmed that the pocketvolume is much larger in the open-ring conformation (272 Å3)(Fig. S2A) than in the closed-ring state (56 Å3) (Fig. S2C), a trendthat held true with all open- and closed-ring Rho structures that havebeen observed to date (Fig. S2 and Table S1). Correspondingly,alignment of a bicyclomycin-bound open ring structure [ProteinData Bank (PDB) ID code 1XPO (24)] with a closed-ring Rhomodel [PDB ID code 3ICE (17)] revealed extensive steric clasheswith bicyclomycin and residues P180, K184, and E211 in theclosed-ring state that would appear to preclude binding (compareFig. 1 D and E). From these observations, we postulated thatbicyclomycin might not sterically block the attacking water neededfor ATP hydrolysis, as previously proposed, but instead antagonizethe RNA- and ATP-dependent conformational switching ofRho from an open- to a closed-ring state.

Bicyclomycin Inhibits Rho Ring Closure over a Range of NucleotideConcentrations. To determine the impact of bicyclomycin on theRho conformational state, we first used SAXS to monitor structuraldynamics in solution directly. An accompanying study demonstratesthat open-ring Rho molecules close in the presence of RNA and thenonhydrolyzable ATP analog ADP·BeF3 but not in the presence ofeither ligand individually (16). Because bicyclomycin is a relativelymodest inhibitor of Rho ATPase activity (IC50 = 20 μM) (25), wereasoned that its effects on ring closure might be most evident whenexamined over a range of nucleotide concentrations. As a control,

we carried out SAXS studies using a series of ADP·BeF3 concen-trations in the presence of RNA and the absence of bicyclomycin.When the average intensity of the observed scattering (I) as afunction of scattering vector (q) was plotted for a variety ofADP·BeF3 concentrations (Fig. 2A), nucleotide-dependent differ-ences in curve shapes were clearly evident, similar to those reportedin the accompanying study (Fig. S3) (16). Visual inspection of thecurves over a q range from 0.07–0.13 Å−1, the region of the curve inwhich nucleotide-dependent differences are most pronounced (Fig.2A, Inset), revealed that the profiles converged between 1.5 and4 mMADP·BeF3, suggesting that under these conditions Rho existspredominantly as a closed ring. Indeed, quantification of the per-centage of open vs. closed states using a Minimal Ensemble Search(MES) (31) suggested that 1.5 mM ADP·BeF3 was sufficient toclose the majority of Rho rings in solution, with an increase in thepercentage of closed rings observed at 2, 3, and 4 mM ADP·BeF3(Fig. 2C and Table S2). Similar trends also were observed in thereduction of the radius of gyration (Rg) at increasing ADP·BeF3concentrations, consistent with the expected nucleotide-dependentcompaction of the Rho ring upon transition to a predominantlyclosed-ring state (Figs. S4A and S5 and Table S3).After establishing a SAXS regime for looking at changes in ring

state, we next set out to determine whether bicyclomycin wouldantagonize the ability of the Rho ring to close in a nucleotide-dependent fashion. We first preincubated Rho with a high concen-tration of bicyclomycin (400 μM final concentration, an eightfoldmolar excess over Rho monomer) and then added RNA and a va-riety of ADP·BeF3 concentrations. The resulting SAXS curves (Fig.2B) also displayed differences that were clearly dependent uponnucleotide concentration; however, at intermediate concentrationsof ADP·BeF3 (0.5, 1, and 1.5 mM), curves were skewed toward theopen-ring state relative to the drug-free curves (compare Insets inFig. 2 A and B). Quantification of ring state by MES showed asubstantial shift toward the open form when bicyclomycin was pre-sent; this shift was particularly notable at 1.5 and 2 mM ADP·BeF3(Fig. 2C and Table S2). This bicyclomycin-dependent inhibition ofRho ring closure also was consistent with observed increases in Rgupon addition of the drug (Figs. S4A and S6 and Table S3).

The Rho Ring State Can Be Controlled by Varying the BicyclomycinConcentration. After observing that bicyclomycin inhibits ring clo-sure, we wondered whether it would be possible to control the Rhoring state by varying the concentration of bicyclomycin with Rhobefore adding RNA and ADP·BeF3. To address this question, weconducted a SAXS experiment in the presence of 1.5 mMADP·BeF3,a nucleotide concentration that, based on our previous titration, issufficient to close a substantial fraction (but not all) of the Rho ringspresent in solution. Preincubating Rho with varying concentrations ofbicyclomycin yielded significant differences in the observed SAXScurves (Fig. 3A). When we focused on the mid-q region (Fig. 3A,Inset), it was immediately evident that the ring state can be affectedby bicyclomycin. Quantification of the percentage of the open vs.closed ring populations by MES (Fig. 3B and Table S4) shows thatthe ring state changed from predominantly closed in the absence ofbicyclomycin to mostly open at high bicyclomycin concentrations.Observed Rg values ranged from 46.3 Å in the absence of bicyclo-mycin to 48.2 Å in the presence of 3,200 μM bicyclomycin, a findingthat is consistent with bicyclomycin concentration-dependent in-hibition of Rho ring closure (Figs. S4B and S7 and Table S5).

A Fluorescence-Based Assay Tracks Ligand-Dependent Effects on RhoRing Status in Vitro. Because SAXS studies constitute only oneassessment of Rho state, we next set out to develop an in vitroassay to further characterize the ligand-dependent effects on theconformation of the hexamer. We reasoned that a short, fluorescein-derivatized RNA would have a significantly faster tumbling ratewhen free in solution than when stably bound to the secondary(translocation pore) site inside a closed Rho ring (see the diagram inFig. 4A), thus enabling us to track ring closure by monitoring changesin fluorescence anisotropy (FA). To rule out the possibility that thebinding of labeled RNA to Rho’s primary (RNA-recruitment) site

Fig. 1. Bicyclomycin is an inhibitor of the Rho helicase. (A) Crystal structure ofopen-ring and bicyclomycin-bound Rho [PDB ID code 1XPO (24)]. Rho subunitsare alternatingly colored yellow and gold for distinction, bicyclomycin is black,and primary-site RNAs are magenta. (B) Crystal structure of closed-ringand translocation-competent Rho [PDB ID code 3ICE (17)]. Rho subunits arealternatingly colored cyan and dark blue, and RNA bound in the secondary siteis red. Magenta-colored ovals denote the location of the primary sites that arenot occupied in this crystal form. (C) Chemical structure of bicyclomycin. (D) Aclose-up view of the bicyclomycin-binding pocket shows that bicyclomycinnestles into a small pocket between Rho subunit interfaces. (E) Modeling ofbicyclomycin into the closed-ring Rho structure shows clear steric clashes be-tween the drug and Rho.

Lawson et al. PNAS | November 29, 2016 | vol. 113 | no. 48 | 13715

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

May

22,

202

1

might convolute the signal, we first preincubated the pro-tein with a short pyrimidine-rich DNA (dC5) at a concentrationequimolar to that of Rho monomer [polypyrimidine DNAs bindboth tightly and selectively to the primary sites of Rho (12, 30)].No anisotropy changes were observed with a fluorescein-labeledpoly-U RNA (rU12*) in the absence of nucleotide and in thepresence of a primary-site competitor (Fig. 4B, black), indicat-ing that rU12* alone does not have high affinity for the secondarysite in open-ring Rho. By contrast, robust rU12* binding was ob-served as nucleotide was added (Fig. 4B), showing that the sec-ondary site becomes capable of stably binding RNA as the Rhoring closes. These findings are consistent with the SAXS studiescarried out in the accompanying study (16), showing that bothRNA and nucleotide are needed to promote ring closure coop-eratively in Rho.To characterize further the mechanism by which bicyclomycin

inhibits Rho, we next conducted an order-of-addition experimentexamining the impact of bicyclomycin on rU12* binding either beforeor after Rho ring closure. As expected, preincubation of bicyclomycinwith Rho before the addition of RNA and nucleotide inhibited rU12*binding in a bicyclomycin concentration-dependent manner (Fig.4C). Conversely, the impact of bicyclomycin on rU12* binding was farless evident when closed rings were formed before the addition of thedrug (Fig. 4D). These data indicate that, although bicyclomycin isable to bind to an open Rho ring and prevent ring closure, themolecule is unable to force open a preclosed Rho ring.

Fig. 2. Bicyclomycin inhibits Rho ring closure over a range of nucleotideconcentrations. (A) Scaled SAXS intensities of Rho in the presence of onerU12 RNA per hexamer and a variety of concentrations of the ATP mimeticADP·BeF3. (Inset) Zoom-in of the mid-q region, in which differences betweenthe various SAXS curves are most pronounced. (B) Scaled SAXS intensities ofRho preincubated with 400 μM bicyclomycin and then combined with onerU12 RNA per hexamer and a variety of concentrations of ADP·BeF3. (Inset)Zoom-in of the mid-q region reveals that the SAXS curves are skewed towardthe open state in the presence of bicyclomycin. (C) Quantification of thecurves in A and B by MES shows a clear skewing toward the open-ring statein the presence of bicyclomycin.

Fig. 3. The Rho ring state can be controlled by varying bicyclomycin con-centration. (A) Scaled SAXS intensities of a variety of bicyclomycin concen-trations before the addition of one rU12 RNA per Rho hexamer and 1.5 mMADP·BeF3. (Inset) Zoom-in of the mid-q region shows that ring state can betitrated from closed to open by increasing the bicyclomycin concentration.(B) Quantification of the curves in A shows a clear titration of Rho ring statethat is dependent upon bicyclomycin concentration.

13716 | www.pnas.org/cgi/doi/10.1073/pnas.1616749113 Lawson et al.

Dow

nloa

ded

by g

uest

on

May

22,

202

1

Rho Ring Closure Is Promoted by the Binding of Pyrimidine-ContainingNucleic Acids to the N-Terminal Primary-Site Domains.Although RNAbinding to Rho’s secondary site (the helicase translocation pore) isa major effector of ring closure, RNA binding to Rho’s primarysite (its OB folds) also has been implicated in controlling Rhofunction (27). To probe the role of primary-site ligands in mod-ulating ring closure, we tested whether the presence or absence ofdC5 could impact the binding of rU12* to the secondary site. Wealso conducted the same experiments with fluorescein-labeledpoly-A RNA (rA12*), which is a weaker secondary-site ligand thanpoly-U and is incompatible with primary-site binding (14, 32). Aclear increase for rA12* affinity was evident in the presence of dC5(Fig. S8), especially at low ADP·BeF3 concentrations. Under allnucleotide concentrations tested, rU12* bound more tightly toRho than did rA12* (Fig. S8), with dC5 subtly promoting ringclosure at moderate (1 mM ADP·BeF3) nucleotide concentra-tions. Primary-site occupancy appeared to have either no effect(1.5 mM ADP·BeF3) or a slight inhibitory effect (2–3 mMADP·BeF3) on rU12* binding at higher nucleotide concentrations.Collectively, these results signify that nucleic acids bound to theprimary site promote ring closure under conditions that are sub-optimal for RNA binding to the secondary site.Based on the spacing of the OB folds in the Rho structure and

the observation that the isolated N-terminal domain engages only apyrimidine dinucleotide (Fig. 1 A and B) (10, 14), each dC5 sub-strate would be expected to bind one primary site at a time. Havingobserved that this short oligonucleotide nonetheless can promoteRho ring closure, we wondered whether longer DNAs capable ofbridging multiple primary sites in a single Rho hexamer would havea more pronounced impact on ring state (see the diagram in Fig.5A). We repeated the rU12*- and rA12*-binding assays with either a15-mer DNA capable of bridging two primary sites (dC15) or a 79-mer DNA potentially sufficiently long to bridge all six primarysites (dC75), using an ADP·BeF3 concentration (1 mM) at whichprimary-site binding of DNA has the most significant effect onRNA binding to the secondary site. The number of putative sitesthat could be occupied by the DNAs was held constant relative tothe dC5 experiments by adjusting the relative molar ratios between

Rho and the DNA (Methods). For both rU12* and rA12*, inclusionof dC15 or dC75 promoted RNA binding more robustly than theshorter dC5 oligo (Fig. 5 B and C). Using fluorescein-labeledderivatives of the various primary-site ligands (denoted as dC5*,dC15*, and dC75*), titration experiments revealed that DNAslong enough to bridge multiple primary sites have a much higheraffinity for Rho than does dC5 (Fig. 5D); by fixing Rho at aconcentration insufficient to drive ring closure with rU12* orrA12* and varying the concentration of primary-site ligands, wefound that ring closure can be driven by sufficiently high con-centrations of dC5, dC15, or dC75 (Fig. 5 E and F). These dataindicate that primary-site occupancy favors the binding of bothoptimal (polypyrimidine) and nonoptimal (polypurine) RNAs toRho’s secondary site.

DiscussionRing-type helicases and translocases control numerous essentialbiological processes (33, 34). Although these enzymes generally arecapable of stand-alone motor activity, multiple lines of evidenceincreasingly highlight complex mechanisms by which these enzymesare regulated. Protein–protein and protein–ligand interactions,small molecule effectors, and posttranslational modifications haveall been implicated in controlling disparate types of ring-ATPases;how a majority of these factors exert their various stimulatory orantagonistic effects is generally not well understood.In the present study, we discovered that bicyclomycin, one of

the few small-molecule agents known to inhibit a ring-type motor,acts on its target, the Rho transcription termination factor, bysterically blocking a conformational change from an open- to aclosed-ring state. Rho ring closure, which an accompanying studyshows is dependent upon both ATP and secondary-site–boundRNA (16), triggers RNA strand engagement and rearranges res-idues in the ATP-binding pocket into a hydrolysis-competentstate. By comparing a bicyclomycin-bound, open-ring conforma-tion of Rho with other structures of Rho substates, we found thatthe drug-binding pocket collapses upon ring closure (Fig. 1).Follow-up SAXS studies show that bicyclomycin does not simply

Fig. 4. FA-based RNA-binding assay to track Rho ring closure in vitro. (A) Schematic illustrating how the tumbling rate of a short fluorescein-tagged RNA isslowed by capture within the Rho ring. The Rho primary sites were blocked with short DNAs, represented in magenta. (B) Binding of rU12* to Rho is de-pendent upon the ADP·BeF3 concentration. (C) Preincubation of Rho with bicyclomycin before the addition of RNA and ADP·BeF3 demonstrates bicyclomycinconcentration-dependent inhibition of rU12* binding. (D) Incubation of bicyclomycin with a preclosed Rho ring has little effect on rU12* binding.

Lawson et al. PNAS | November 29, 2016 | vol. 113 | no. 48 | 13717

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

May

22,

202

1

occlude the binding of a catalytic water, as previously proposed(24), but instead directly impedes nucleotide-dependent closure ofthe Rho ring (Figs. 2 and 3). Using an FA-based assay to track ringclosure in vitro, we further observed that bicyclomycin acts onpreopened Rho rings and has little effect on ring topology afterring closure (Fig. 4). Because inhibition of secondary-site RNAbinding is indicative of bicyclomycin binding to Rho (26), theseresults strongly indicate that bicyclomycin is unable to bind topreclosed Rho rings. Thus, bicyclomycin appears to inhibit Rho bystabilizing a conformation that is incapable of ATP hydrolysis andis incompatible with stable RNA binding to its motor regions.

In the course of investigating bicyclomycin’s effects on Rho, wealso found that binding of nucleic acids to Rho’s primary RNArecruitment site promotes the switching of Rho from an open- to aclosed-ring state. This finding is in accord with prior studiesshowing that Rho’s ATPase activity can be stimulated by primary-site occupancy (30) and that Rho’s ATPase rate is regulated inpart by an RNA- and ATP-dependent conformational change (8),presumably into a closed-ring state (16, 17). The inclusion ofpolypyrimidine DNAs, which bind selectively to the primary siteand also promote ATPase activity (30), increased Rho’s affinityfor both rU12 and rA12 substrates at nucleotide concentrationsthat otherwise are too low to drive ring closure stably (Fig. S8).This result suggests that the binding of a rut sequence to Rho’sprimary site, which is formed by an N-terminal OB-fold in theprotein (12, 13), allosterically promotes ring closure.We further found that DNAs capable of bridging two (dC15) or

six (dC75) primary sites within the Rho hexamer promoted ringclosure more robustly than short (dC5) DNAs capable of occu-pying only one primary site (Fig. 5 A–C). The more robust impactof these longer DNAs is likely caused by their higher affinity forRho (Fig. 5D) rather than by a particular signal that might bepropagated by a nucleic acid segment bridging multiple primarysites, because at sufficiently high concentrations all three ligandsdrive complete ring closure (Fig. 5 E and F). We also found thatring closure was more sensitive to primary-site occupancy by rA12than with rU12; this finding is of interest because polypurines bindto the secondary site more weakly than polypyrimidines (18).Collectively, these results suggest that the RNA recruitment siteson Rho work together to count the number and spacing of py-rimidine recruitment motifs in an RNA transcript and promotering closure around nonoptimal secondary-site binding sequencessuch as purine-rich regions (Fig. 6). Future studies looking atdifferent types of di-pyrimidine patterns on both natural andsynthetic substrates will be needed to probe how this interrogationoccurs at a molecular level.In using an auxiliary ligand-binding element to aid activity, Rho

joins a growing list of processive ring helicases and translocasesthat are subject to both intrinsic and extrinsic regulation. Thesefactors include the bacterial DnaB helicase (6), the MCM2-7helicase (7, 35), and the proteasomal Rpt1-6 ATPases (5). In eachof these motors, a built-in accessory domain or a dissociable set of

Fig. 5. Primary-site occupancy promotes Rho ring closure. (A) Cartoon representation of how dC5, dC15, and dC75 could occupy one, two, or six primary sites,respectively. (B) Binding of rU12* to Rho is promoted by polycytidylic DNAs of various lengths. (C) Binding of rA12* to Rho is promoted by polycytidylic DNAs ofvarious lengths. (D) Longer DNAs (dC15* and dC75*) capable of bridging multiple primary sites progressively promote secondary-site binding more effectivelythan does a short DNA (dC5*) that can occupy only one primary site. (E) Binding of rU12* can be fully driven at saturating concentrations of dC5, dC15, or dC75.(F) Binding of rA12* can be fully driven at saturating concentrations of dC5, dC15, or dC75. Polypurine DNAs (dA5 and dA15, light and dark green, respectively)that do not bind Rho’s primary site do not facilitate Rho ring closure.

Fig. 6. Model for how bicyclomycin and primary-site ligands regulate Rho ringclosure and thereby the capacity of the helicase to promote transcription ter-mination. After a rut-containing RNA first binds to the Rho primary sites andsubsequently the secondary site, the Rho ring snaps shut to trap the RNAwithinthe central pore. This conformational change to a translocation-competentclosed ring state is less efficient in the presence of a weakly binding rut se-quence with nonoptimal di-pyrimidine number/spacing (Top) and is completelyblocked by bicyclomycin (Bottom, black wedges). Open-ring Rho must bind tothe rut sequence via primary-site interactions and close before its ATPase ac-tivity can promote 5′-to-3′ RNA translocation and transcription termination.

13718 | www.pnas.org/cgi/doi/10.1073/pnas.1616749113 Lawson et al.

Dow

nloa

ded

by g

uest

on

May

22,

202

1

adaptor proteins is used to choreograph proper substrate re-cruitment and engagement with motor activity. How controllinkages between the accessory elements and the motor regionsare established through long-range binding and allosteric tran-sitions remains a frontier question.In closing, it is worth noting that there is substantial interest in

identifying small-molecule approaches to inhibiting ring-ATPases(36). Unfortunately, because ATP-binding pockets are often wellconserved across different motor families, developing such inhibitorshas been challenging. Interestingly, Rho is one of the few ring-typemotor proteins for which a small molecule with therapeutic proper-ties has been identified. Here we have shown that this agent, bicy-clomycin, does not antagonize the chemistry of the ATPase reactionbut instead acts as a conformational inhibitor of ring dynamics, en-tering a binding pocket that is accessible only during a portion of theRho RNA loading and translocation cycle. These findings suggestthat by using appropriate screens for ring dynamics, coupled withcounter screens against general ATPase activity, it may be possible toidentify new generations of selective inhibitors with clinical potential.

MethodsCalculation of Bicyclomycin-Binding Pocket Volumes. Pocket volumes werecalculated using POVME (see SI Methods for program parameters) (37).

SAXS Sample Preparation. Rho was purified as described previously (17). SAXSsamples were assembled as described in SI Methods, flash-frozen in liquidnitrogen, and stored at −80 °C.

SAXS Data Collection and Analysis. SAXS data were collected using an auto-mated, high-throughput system at the Advanced Light Source Beamline 12.3.1(ALS BL12.3.1), Lawrence Berkeley National Laboratory (38). SAXS curves weregenerated, analyzed, and plotted as described in SI Methods.

FA-Based RNA-Binding Data Collection and Analysis. DNA oligonucleotides [dC5,dC15, and dC75 (dC15TC15TC15TC15TC15)] were purchased from IDT. FA sampleswere prepared, measured, and analyzed as described in SI Methods.

ACKNOWLEDGMENTS. We thank members of the J.M.B. laboratory, AndreasMartin’s laboratory (University of California, Berkeley and Howard Hughes Med-ical Institute), Nathan Thomsen, and Michael Bellecourt (University of Wisconsin-Madison) for helpful discussions and careful reading of the manuscript and GregHura and Michal Hammel at the ALS BL 12.3.1, Lawrence Berkeley NationalLaboratory, for guidance on SAXS sample preparation and data processing.This research was supported by National Institute of General Medical ScienceGrants R01-GM071747 (to J.M.B.) and T32-GM066698 (to the Department ofMolecular and Cell Biology, University of California, Berkeley), and by theG. Harold and Leila Y. Mathers Foundation (J.M.B.). M.R.L. is the recipient ofa National Science Foundation Graduate Research Fellowship.

1. O’Shea VL, Berger JM (2014) Loading strategies of ring-shaped nucleic acid translocasesand helicases. Curr Opin Struct Biol 25:16–24.

2. Singleton MR, DillinghamMS, Wigley DB (2007) Structure and mechanism of helicasesand nucleic acid translocases. Annu Rev Biochem 76:23–50.

3. Lyubimov AY, Strycharska M, Berger JM (2011) The nuts and bolts of ring-translocasestructure and mechanism. Curr Opin Struct Biol 21(2):240–248.

4. Matyskiela ME, Martin A (2013) Design principles of a universal protein degradationmachine. J Mol Biol 425(2):199–213.

5. Bashore C, et al. (2015) Ubp6 deubiquitinase controls conformational dynamics andsubstrate degradation of the 26S proteasome. Nat Struct Mol Biol 22(9):712–719.

6. Strycharska MS, et al. (2013) Nucleotide and partner-protein control of bacterialreplicative helicase structure and function. Mol Cell 52(6):844–854.

7. Froelich CA, Kang S, Epling LB, Bell SP, Enemark EJ (2014) A conserved MCM single-stranded DNA binding element is essential for replication initiation. eLife 3:e01993.

8. Jeong YJ, Kim DE, Patel SS (2004) Nucleotide binding induces conformational changes inEscherichia coli transcription termination factor Rho. J Biol Chem 279(18):18370–18376.

9. Peters JM, et al. (2009) Rho directs widespread termination of intragenic and stableRNA transcription. Proc Natl Acad Sci USA 106(36):15406–15411.

10. Skordalakes E, Berger JM (2003) Structure of the Rho transcription terminator:Mechanism of mRNA recognition and helicase loading. Cell 114(1):135–146.

11. Yu X, Horiguchi T, Shigesada K, Egelman EH (2000) Three-dimensional reconstructionof transcription termination factor rho: Orientation of the N-terminal domain andvisualization of an RNA-binding site. J Mol Biol 299(5):1279–1287.

12. Morgan WD, Bear DG, Litchman BL, von Hippel PH (1985) RNA sequence and sec-ondary structure requirements for rho-dependent transcription termination. NucleicAcids Res 13(10):3739–3754.

13. Dolan JW, Marshall NF, Richardson JP (1990) Transcription termination factor rho hasthree distinct structural domains. J Biol Chem 265(10):5747–5754.

14. Bogden CE, Fass D, Bergman N, Nichols MD, Berger JM (1999) The structural basis forterminator recognition by the Rho transcription termination factor.Mol Cell 3(4):487–493.

15. Miwa Y, Horiguchi T, Shigesada K (1995) Structural and functional dissections of tran-scription termination factor rho by random mutagenesis. J Mol Biol 254(5):815–837.

16. Thomsen ND, Lawson MR, Witkowsky LB, Qu S, Berger JM (2016) Molecular mecha-nisms of substrate-controlled ring dynamics and sub-stepping in a nucleic-acid de-pendent hexameric motor. Proc Natl Acad Sci USA 113:E7691–E7700.

17. Thomsen ND, Berger JM (2009) Running in reverse: The structural basis for translo-cation polarity in hexameric helicases. Cell 139(3):523–534.

18. Galluppi GR, Richardson JP (1980) ATP-induced changes in the binding of RNA syn-thesis termination protein Rho to RNA. J Mol Biol 138(3):513–539.

19. Koslover DJ, Fazal FM, Mooney RA, Landick R, Block SM (2012) Binding and translo-cation of termination factor rho studied at the single-molecule level. J Mol Biol423(5):664–676.

20. Steinmetz EJ, Platt T (1994) Evidence supporting a tethered tracking model for helicaseactivity of Escherichia coli Rho factor. Proc Natl Acad Sci USA 91(4):1401–1405.

21. Gocheva V, Le Gall A, Boudvillain M, Margeat E, Nollmann M (2015) Direct observa-tion of the translocation mechanism of transcription termination factor Rho. NucleicAcids Res 43(4):2367–2377.

22. Park JS, Roberts JW (2006) Role of DNA bubble rewinding in enzymatic transcriptiontermination. Proc Natl Acad Sci USA 103(13):4870–4875.

23. Schwartz A,Margeat E, Rahmouni AR, BoudvillainM (2007) Transcription termination factorrho can displace streptavidin from biotinylated RNA. J Biol Chem 282(43):31469–31476.

24. Skordalakes E, Brogan AP, Park BS, Kohn H, Berger JM (2005) Structural mechanism ofinhibition of the Rho transcription termination factor by the antibiotic bicyclomycin.Structure 13(1):99–109.

25. Park HG, et al. (1995) Bicyclomycin and dihydrobicyclomycin inhibition kinetics ofEscherichia coli rho-dependent transcription termination factor ATPase activity. ArchBiochem Biophys 323(2):447–454.

26. Magyar A, Zhang X, Kohn H, Widger WR (1996) The antibiotic bicyclomycin affectsthe secondary RNA binding site of Escherichia coli transcription termination factorRho. J Biol Chem 271(41):25369–25374.

27. McSwiggen JA, Bear DG, von Hippel PH (1988) Interactions of Escherichia coli tran-scription termination factor rho with RNA. I. Binding stoichiometries and free ener-gies. J Mol Biol 199(4):609–622.

28. Hart CM, Roberts JW (1991) Rho-dependent transcription termination. Characterization ofthe requirement for cytidine in the nascent transcript. J Biol Chem 266(35):24140–24148.

29. Schwartz A, Walmacq C, Rahmouni AR, Boudvillain M (2007) Noncanonical interac-tions in the management of RNA structural blocks by the transcription terminationrho helicase. Biochemistry 46(33):9366–9379.

30. Richardson JP (1982) Activation of rho protein ATPase requires simultaneous inter-action at two kinds of nucleic acid-binding sites. J Biol Chem 257(10):5760–5766.

31. Schneidman-Duhovny D, Hammel M, Sali A (2010) FoXS: A web server for rapid com-putation and fitting of SAXS profiles. Nucleic Acids Res 38(Web Server issue):W540–544.

32. Faus I, Richardson JP (1990) Structural and functional properties of the segments of lambdacro mRNA that interact with transcription termination factor Rho. J Mol Biol 212(1):53–66.

33. Roberts JW (1969) Termination factor for RNA synthesis. Nature 224(5225):1168–1174.34. Etlinger JD, Goldberg AL (1977) A soluble ATP-dependent proteolytic system responsible for

the degradation of abnormal proteins in reticulocytes. Proc Natl Acad Sci USA 74(1):54–58.35. Barry ER, McGeoch AT, Kelman Z, Bell SD (2007) Archaeal MCM has separable proc-

essivity, substrate choice and helicase domains. Nucleic Acids Res 35(3):988–998.36. Deshaies RJ (2014) Proteotoxic crisis, the ubiquitin-proteasome system, and cancer

therapy. BMC Biol 12(94):94.37. Durrant JD, de Oliveira CA, McCammon JA (2011) POVME: An algorithm for mea-

suring binding-pocket volumes. J Mol Graph Model 29(5):773–776.38. Dyer KN, et al. (2014) High-throughput SAXS for the characterization of biomolecules

in solution: A practical approach. Methods Mol Biol 1091:245–258.39. Rambo RP (2015) Resolving Individual Components in Protein-RNA Complexes Using

Small-Angle X-ray Scattering Experiments. Methods Enzymol 558:363–390.

Lawson et al. PNAS | November 29, 2016 | vol. 113 | no. 48 | 13719

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

May

22,

202

1