Interdomain dynamics and coactivation of the mRNAdecapping enzyme Dcp2 are mediatedby a gatekeeper tryptophanStephen N. Floora,b,1, Mark S. Borjab,c, and John D. Grossb,2

aGraduate Group in Biophysics; bDepartment of Pharmaceutical Chemistry; and cProgram in Chemistry and Chemical Biology, University of California,San Francisco, CA 94158

Edited by Juli Feigon, University of California, Los Angeles, CA, and approved December 22, 2011 (received for review August 18, 2011)

Conformational dynamics in bilobed enzymes can be used toregulate their activity. One such enzyme, the eukaryotic decappingenzyme Dcp2, controls the half-life of mRNA by cleaving the 5′cap structure, which exposes a monophosphate that is efficientlydegraded by exonucleases. Decapping by Dcp2 is thought to becontrolled by an open-to-closed transition involving formationof a composite active site with two domains sandwiching sub-strate, but many details of this process are not understood. Here,using NMR spectroscopy and enzyme kinetics, we show that Trp43of Schizosaccharomyces pombe Dcp2 is a conserved gatekeeperof this open-to-closed transition. We find that Dcp2 samples multi-ple conformations in solution on the millisecond-microsecondtimescale. Mutation of the gatekeeper tryptophan abolishes thedynamic behavior of Dcp2 and attenuates coactivation by a yeastenhancer of decapping (Edc1). Our results determine the dynamicsof the open-to-closed transition in Dcp2, suggest a structural path-way for coactivation, predict that Dcp1 directly contacts the cata-lytic domain of Dcp2, and show that coactivation of decapping byDcp2 is linked to formation of the composite active site.

enzyme dynamics ∣ methyl groups ∣ mRNA decay ∣ protein NMR

Conformational dynamics in enzymes often comprise the rate-limiting step in the catalytic cycle and thus are prime targets

for regulatory cofactors (1–5). Bilobed proteins frequently usean open-to-closed transition to coordinate catalysis on theirsubstrates following cellular cues such as posttranslational mod-ifications or macromolecular interactions (6–8). A recent modelproposes that the eukaryotic mRNA decapping enzyme Dcp2 isregulated by such a transition, where a composite active site isformed using conserved surfaces on each of the two N-terminaldomains (9). According to this model, stimulating or inhibitingthis conformational transition could regulate decapping. How-ever, the structural details of this composite active site and thetimescale of interconversion between closed and open states ofDcp2 are currently unknown. Moreover, whether coactivatorsuse the composite active site to effect decapping is unclear.

Degradation of eukaryotic mRNA is critical to many biologicalprocesses including development (10), stress response (11), clear-ance of the products of pervasive transcription (12), and qualitycontrol of gene expression (13). For example, it has been sug-gested that microRNAs (miRNAs) act primarily by destabilizingmessages and it is known that Dcp2 is a vital component ofmiRNA-induced mRNA decay (14, 15). Further, an entire classof unstable transcripts was recently discovered that is sensitive tothe exonuclease Xrn1, whose members are therefore likely pro-ducts of decapping (12, 16). Each of the variety of pathways thatutilize decapping relies on coactivator proteins that are believedto recruit messages to the decapping machinery and activate it.

A model of decapping coactivation is emerging followingrecent work on the Saccharomyces cerevisiae coactivator Edc1(17). Edc1 is a yeast-specific protein that is required for carbonsource changes and is strongly upregulated during such transi-tions (18–20). It binds directly to Dcp1, which in turn forms a

stable complex with the regulatory domain of Dcp2 (17, 21).Dcp1 has an enabled/VASP homology-1 (EVH1) fold and usesa hydrophobic patch to recognize a proline-rich stretch in theC terminus of Edc1, which is also found in other putative coac-tivators (17, 22). Binding of Edc1 raises the catalytic efficiency ofthe Dcp1∶Dcp2 complex by up to 3,000 times by enhancing boththe KM for mRNA and rate of the catalytic step kmax (17). Inter-estingly, it appears that Edc1 is modular with the N-terminal re-gion responsible for the KM enhancement and the C-terminalregion primarily affecting kmax (17). The mechanism of this en-hancement is unknown though it was proposed that Edc1 maystimulate closure and thereby activity of Dcp2.

Many proteins are dynamic on the millisecond-microsecond(ms-μs) timescale and these motions can be intimately tied toactivity (1). Dcp2 is known to undergo an open-to-closed transi-tion that leads to formation of the composite active site and itwas suggested that both the apo and ligand-bound forms of theenzyme sample multiple conformations (9). Open-to-closed tran-sitions in bilobed proteins like Dcp2 can occur on the ms-μstimescale, which can be monitored with site-specific resolutionby NMR spectroscopy (23, 24). Motions on the ms-μs timescalelead to dephasing of transverse magnetization which manifestsin a rate constant Rex that, together with R2 for molecular tum-bling, determines the resonance linewidth. Carr–Purcell–Mei-boom–Gill (CPMG) NMR spectroscopy allows contributionsto the resonance linewidth from molecular tumbling to be sepa-rated from ms-μs dynamics (25). When coupled with 13C-methylIle, Leu, Val, Met, and Ala (ILVMA) labeling CPMG can beapplied to large molecular weight complexes and highly dynamicproteins (26). Refocusing pulses applied at various frequenciesresult in dispersion curves that are fit to either the generaltwo-site equation (SI Methods) or the fast-exchange limit:

R2;eff ¼ R2;0 þ pApBΔω2

kex

�1 −

4νCPMG

kextanh

kex4νCPMG

�; [1]

where R2;eff is the effective transverse relaxation rate, R2;0 is thetransverse relaxation rate of the major state, pA and pB are thepopulation of the major and minor states,Δω is the chemical shiftdifference between states, kex is the rate of exchange betweenstates, and νCPMG is the refocusing frequency (27–29). In the gen-

Author contributions: S.N.F. and J.D.G. designed research; S.N.F. and M.S.B. performedresearch; S.N.F. contributed new reagents/analytic tools; S.N.F. analyzed data; and S.N.F.and J.D.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1Present address: Department of Molecular and Cell Biology, University of California,Berkeley, CA 94720.

2To whom correspondence may be addressed. E-mail: jdgross@cgl.ucsf.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1113620109/-/DCSupplemental.

2872–2877 ∣ PNAS ∣ February 21, 2012 ∣ vol. 109 ∣ no. 8 www.pnas.org/cgi/doi/10.1073/pnas.1113620109

Dow

nloa

ded

by g

uest

on

June

27,

202

0

eral case kinetics (kex), thermodynamics (pA, pB), and structure(Δω) are accessible by CPMG spectroscopy, but in the fast-exchange limit one can only extract the composite termpApBΔω2 and the exchange rate kex (30). CPMG is sensitive todynamics involving skewed populations with exchange ratesbetween approximately 100 and 5;000 s−1 (28).

Despite its importance in regulation of mRNA decapping,structural, kinetic, and thermodynamic details regarding theopen-to-closed transition of Dcp2 are poorly understood. Addi-tionally, it has been suggested, but not tested, that coactivatorsmay stimulate decapping by promoting closure of Dcp2. Herewe show that Dcp2 exists in a conformational equilibrium insolution between open and closed states in the absence of ligandthat is fast on the NMR timescale. Exchange between states de-pends on a conserved, solvent-exposed tryptophan previouslyshown to bind m7G of cap, and implicated as a critical componentof the composite active site in Dcp2 (9). Enzyme kinetics andNMR experiments reveal that this tryptophan acts as a gate-keeper that promotes formation of the composite active siteand allows Dcp1 and bound coactivators to enhance the catalyticstep.

ResultsDcp2 Samples Multiple Conformations in Solution. To investigate thedynamics of the proposed composite active site in solution, weturned to NMR spectroscopy of the N-terminal 243 residuesof Schizosaccharomyces pombe Dcp2 (spDcp2), which containsthe regulatory and catalytic Nudix hydrolase domains and isconserved from yeast to humans (31). The first 15N heteronuclearsingle quantum correlation (HSQC) spectrum acquired ofspDcp2 was of marginal quality with many missing resonancesand a large variation in peak intensity (Fig. 1), typical of proteinsundergoing ms-μs dynamics in solution. Because there is a pro-posed composite active site between the two domains of Dcp2(9), and Dcp2 was crystallized with ATP in two conformations(21), we reasoned there might be multiple states coexisting in so-lution giving rise to the observed resonance broadening. To testthis possibility we mutated select sites on both domains of spDcp2and found that mutation of Trp43 of the regulatory domain in-duced profound changes in the nitrogen HSQC (Fig. 1). This mu-tation decouples the two domains because the Trp43Alaspectrum is more similar to the spectra of the isolated domainsthan the wild-type spectrum (Fig. S1). Tryptophan 43 is requiredfor decapping in vitro and in vivo, closure by small-angle X-rayscattering (SAXS), and cap binding by the regulatory domain(9, 32). These data suggest that it also mediates an interdomaininteraction in Dcp2 in the absence of substrate.

Quantitative Measurement of Methyl Side-Chain Dynamics in spDcp2.We turned to CPMG spectroscopy to quantify the kinetic andthermodynamic properties of the dynamic behavior observedin the 15N HSQC. Because the cross-peaks for dynamic residuesare broadened in the 15NHSQC (Fig. 1) we used 13C-methyl ILV-MA side-chain labeling of spDcp2 for quantification. We foundthat approximately one third of methyl groups in spDcp2 undergosignificant ms-μs dynamics that are attenuated by CPMG pulses(Fig. 2, Figs. S2–S4). Exchange rates ranged from approximately500 s−1 to 3;000 s−1 and all but three are fit well by the fast-exchange limit formula (SI Methods, Fig. S4, Table S1, andTable S2).

To identify the dynamic residues, we assigned the methylgroups of ILVMA residues of spDcp2. Dynamic residues aredistributed throughout both domains of the protein (Fig. 2B,orange spheres). Crucially, the exchange broadening (Rex) ofnearly all of the methyl groups in the isolated domains of spDcp2is insignificant (Fig. S5), strongly suggesting that the dynamics

15N

(pp

m)

1H (ppm)

130

11 10 9 8 7 6

125

120

115

110

105

G168

I130

A41

V171

N170

F225

A51

G124

G60

Fig. 1. Mutation of Trp43 in the regulatory domain of spDcp2 alleviates re-sonance broadening in the 15N HSQC. Shown are the 15N HSQC spectra ofwild-type S. pombe Dcp2 residues 1–243 (red) and spDcp2 Trp43Ala (black).Selected residues with significant changes uponmutation are indicated; notethat residues following 95 are located on the catalytic domain.

R2,

eff (

s-1)

νCPMG (Hz)

30

20

10

00 500 1,000

40

50L113

30

20

10

00 500 1,000

1,0001,000

40

50 L90

30

20

10

00 500

40

50I130

30

20

10

00 500

40

50I179

A

B

Fig. 2. Quantitative analysis of methyl ILVMA side-chain dynamics inspDcp2. (A) Four representative CPMG dispersion curves are shown withexperimental data in circles and fits in lines. Data acquired at 800 and900MHz are in black and red, respectively. Error bars are the pooled standarddeviation (SI Methods). (B) Methyl groups in spDcp2 1–243 are displayed asspheres on PDB ID code 2QKM. Methyls that were fit are orange, with inten-sities that were too weak to quantify are red, insignificantly broadenedby ms-μs dynamics (Rex less than 5 s−1) are gray and unassigned are black.The regulatory domain is shown in purple, catalytic domain in green, andthe catalytic Nudix helix in red with Trp43 in cyan.

Floor et al. PNAS ∣ February 21, 2012 ∣ vol. 109 ∣ no. 8 ∣ 2873

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

27,

202

0

measured in spDcp2 are not due to fluctuations within the indi-vidual domains, but rather a transient interdomain interaction.Moreover, as the dynamic behavior is largely unchanged by two-fold sample dilution, the observed resonance broadening is notdue to intermolecular interactions (Fig. S3). Previous SAXSanalysis showed that spDcp2 undergoes an open-to-closed transi-tion promoted by nucleotide, but closure was not observed with-out bound Dcp1 (21). The results presented here show theenzyme samples multiple states in the absence of cofactors.

A Tryptophan Dynamically Links the Regulatory and CatalyticDomains. If the exchange broadening is due to a concerted processsuch as an open-to-closed transition, then mutations that disruptclosure should alter the exchange broadening in a uniform man-ner. To test this possibility we mutated Trp43 of spDcp2 to alanineand measured ms-μs dynamics by 13C-methyl CPMG. Thismutant was chosen because it blocks closure by SAXS and has aprofound effect on the nitrogen HSQC of spDcp2 (Fig. 1) (9).Remarkably, mutation of Trp43 to alanine strongly damps relaxa-tion in 18 out of 23 methyl groups that are ms-μs dynamic in thewild-type protein (Fig. 3). Of those not affected by mutation ofTrp43, two are dynamic in the isolated catalytic domain (Leu113and Val114) (Fig. S5) and one is in the flexible interdomain linker(Ile102). Mutation of Trp43 quenches relaxation due to dynamicsnot only on the regulatory domain but also the catalytic domain,consistent with its global effect on the nitrogen HSQC (Fig. 1,Fig. 3). These results implicate Trp43 in an interdomain interfaceof an excited state of spDcp2 that is transiently sampled in solu-tion. Notably this interdomain interface is different than theone found in the closed crystal structure because mutation ofthe interaction partner Arg167 does not affect kinetics, closureby SAXS or the nitrogen HSQC (Fig. S6) (9).

The Regulatory and Catalytic Domains Experience CollectiveDynamics. An additional prediction of a concerted dynamic pro-cess is that residues on both domains should be able to be fitcollectively as a group to one global exchange rate. Alternatively,residues may undergo motions on the ms-μs timescale due to lo-cal, independent conformational changes such as loop dynamics.For example, residues Leu113, Val114, Ile193, Leu201, Leu204,and Leu226 are dynamic in the isolated catalytic domain (Fig. S5)as previously observed for the S. cerevisiae ortholog (33). How-ever, many more residues are dynamic in the two-domain spDcp2than in the isolated domains (23 versus 8, Fig. S4, Fig. S5). This

observation, along with the decoupling effect of the Trp43 muta-tion, suggests the dynamics involve an interdomain interaction.

To rigorously determine if there are collective motions withinspDcp2, we isolated a group of fourteen residues to collectivelyfit CPMG exchange curves to (Fig. 4 and Fig. S7; for group mem-bership rationale see SI Methods). A collective exchange rate fitsthe group well, with fits extremely close to the individual fits(Fig. 4A dashed versus solid lines, note that Ile130 has the fourthlargest difference between individual and group fits). Most inter-estingly, six of the group members are on the regulatory domainand three are on the catalytic domain, with no such constraintimposed (Fig. 4B, the remaining five are unassigned). We there-fore conclude that apo spDcp2 experiences collective interdo-main dynamics that depend on the presence of Trp43 with arate of kex ¼ 2;299� 74 s−1, likely the sum of opening and closingrates. Unfortunately, because of protein solubility constraints, wecannot perform similar experiments on substrate-bound proteinat present.

Tryptophan 43 Links Dcp1 and the Catalytic Domain of Dcp2. BothDcp1 and Trp43 are thought to promote closure of Dcp2, so wereasoned there might be a coupling between the two (9, 21). Totest this possibility we formed complexes between 13C-methylILVMA labeled spDcp2 and unlabeled S. pombe Dcp1 andmonitored chemical shift changes. Because this heterodimer isdynamic and approaching 50 kDa, which obliterated the nitrogenHSQC, we turned to the 13C-heteronuclear multiple quantumcorrelation (HMQC), also known as methyl-transverse relaxationoptimized spectroscopy (methyl-TROSY) (34). Stable complexes

30

20

10

00 500

40

50

30

20

10

00 500 1,000

40

50

30

20

10

00 500 1,000

40

50

L90

I196I130

30

20

10

00 500 1,000 1,000

40

50A41

R2,

eff (

s-1)

νCPMG (Hz)

Fig. 3. Mutation of Trp43 in Dcp2 strongly attenuates the majority of ms-μsdynamic methyl groups. CPMG data for four representative residues at800 MHz for wild-type or Trp43Ala Dcp2 are in black and pink, respectively.All but the following residues had Rex values less than experimental errorafter Trp43Ala mutation: Val112, Leu113, Val114, Ile179, and Ile102. Errorbars are the pooled SD.

30

20

10

0

40

50

0 500 1,000

L90

30

20

10

0

40

50

0 500 1,000

I130

R2,

eff (

s-1)

νCPMG (Hz)

A

B

Fig. 4. The regulatory and catalytic domains of spDcp2 are involved in a col-lective exchange process. (A) Representative CPMG group fits are shown forresidues Leu90 and Ile130. Four different fits are shown in these plots: 800and 900MHz individual fits (black and red, solid) and 800 and 900MHz groupfits (gray and orange, dashed). The ratio of group to individual F-statistics forLeu90 is 1.02 and for Ile130 is 1.14. (B) The nine assigned members of thefourteen-member group are displayed on PDB ID code 2QKM as orangespheres with other colors as in Fig. 2.

2874 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1113620109 Floor et al.

Dow

nloa

ded

by g

uest

on

June

27,

202

0

of Dcp1 and spDcp2 for both wild-type and Trp43Ala constructsyield high quality ILVMA spectra with widespread chemical shiftchanges (Fig. S8). Because we were unable to obtain full reso-nance assignments of the Dcp1∶Dcp2 complex due to exceedinglyhigh Rex, we used the minimum chemical shift metric to measurechemical shift changes, which places a conservative lower boundon the perturbation of assigned Dcp2 resonances (35). As ex-pected, residues proximal to the Dcp1∶Dcp2 regulatory domaininterface observed by crystallography are shifted considerablyupon addition of Dcp1, which depended little on mutation ofTrp43 (Fig. 5, Fig. S8).

A second subset of residues outside the Dcp1 binding site isshifted by the addition of Dcp1 only when Trp43 is not mutated(Fig. 5, Fig. S8). The shift changes are small but larger than aresonance linewidth, consistent with an unfavorable equilibriumin fast exchange, similar to other systems in a preexisting equili-brium that is shifted by posttranslational modifications (36),suggesting either Dcp1 is directly contacting regions of Dcp2in the closed, Trp43-dependent state, or Dcp1 is enhancing thepopulation of the closed state but can only do so when Trp43is present. For example, residues Val112, Ile162, Leu201, andIle233 on the catalytic domain of Dcp2 may be directly influenced

by Dcp1 because mutation of Trp43, which should also block for-mation of the closed state, has little effect on their chemical shift(Fig. 5). Some of the residues outside the Dcp1 binding site onDcp2 experience Trp43-dependent exchange broadening in apoDcp2 (Leu68, Leu74, Ile162, Ile196, Leu201, and Ile233; Fig. S2and Fig. S4). We conclude that there is a coupling between Trp43and Dcp1 and that Dcp1 might directly interact with regions ofthe Dcp2 catalytic domain (Fig. 5C).

Tryptophan 43 Couples Coactivators to Catalysis by Dcp2.Addition ofDcp1 to Dcp2 stimulates catalysis 10-fold in vitro and is believedto enhance formation of the closed, active conformation of Dcp2(9, 21). Therefore, if Dcp1 interacts with the catalytic domain ofDcp2 in a Trp43-dependent fashion (Fig. 5), then mutation atTrp43 should block kinetic stimulation by Dcp1. We tested thisprediction directly by comparing the stimulation of Dcp2 activityafforded by Dcp1 to the stimulation in the Trp43Ala enzyme.Whereas Dcp1 simulates catalysis by wild-type spDcp2 by a factorof ten, the Trp43Ala mutant is refractory to Dcp1 stimulation(Fig. 6A). Therefore Trp43 is critical for both contact betweenDcp1 and the catalytic domain in the absence of substrate (Fig. 5),and kinetic stimulation by Dcp1 in the presence of substrate(Fig. 6A).

Dcp1 is believed to be a protein-protein interaction platformbased on its EVH1 fold and essentiality for decapping in yeast(22, 37, 38). The S. cerevisiae protein Edc1 binds to the prolinerecognition site of Dcp1 and is a model coactivator of decapping(17). The C-terminal 30 residues of Edc1 (Edc1CTR) bind to Dcp1and stimulate decapping in vitro, largely via enhancement of therate of the catalytic step, kmax (17). Given the aforementionedresults suggesting a coupling of Dcp1 to Dcp2 via Trp43, wehypothesized that mutation of this residue would abolish coacti-vation by Edc1. To test this hypothesis we made the correspond-ing mutation in the S. cerevisiae Dcp1∶Dcp2 complex (Trp50Ala)and tested its activity (Fig. 6B). Whereas the wild-type Dcp1∶Dcp2 complex was stimulated by Edc1CTR by a factor of 13, itonly enhanced catalysis in the Trp50Ala mutant by a factor ofthree. These results are consistent with a model where Edc1CTRis contacting either the catalytic domain of Dcp2 or substrateitself, and that an interaction between Trp50 and cap mediatesthis coactivation.

Edc1binding

m7Gbinding

m7Gbinding

Dcp1shifts

Dcp1shifts

B C

A

∆δ (p

pm)

0.0

0.1

0.2

0.3

L15

L15

L21

L21

L68

L74

V112

V114

I162

I196

L201 I2

33

Dcp1 binding site Catalytic domainReg.domain

D1D2WT

D2W43AD1D2W43A

Fig. 5. Dcp1 causes chemical shift changes on Dcp2 outside its binding sitethat depend on Trp43. (A) The minimum chemical shift change for select re-sidues is plotted for Dcp1∶Dcp2 (orange), Dcp2 Trp43Ala (black), andDcp1∶Dcp2 Trp43Ala (gray). Each chemical shift change is calculated with re-spect to the wild-type Dcp2 HMQC. The horizontal line indicates a chemicalshift change of 0.05 ppm (approximately one linewidth). (B) Shifted residuesfrom A are shown in orange sticks on a model of the S. pombe Dcp1•Dcp2open structure using PDB ID codes 2QKM and 2A6T (SI Methods). Colors areas in Fig. 2 with Dcp1 in gold. (C) An illustration of the Dcp1•Dcp2 complexwith Edc1 and cap (m7G) binding sites in gray, regions shifted by Dcp1 addi-tion in orange, and other colors as in B.

scDcp

1:Dcp

2 (1-2

45)

scD1:

D2 +

Edc1 CTR

scD1:

D2 W50

A

scD1:

D2 W50

A +

Edc

1 CTR

k obs (

min

-1)

k obs (

min

-1)

0.001

0.01

0.1

1

spDcp

2 (1-2

43)

spDcp

1:Dcp

2

spDcp

2 W43

A

spDcp

1:Dcp

2 W43

A

0.001

0.01

0.1

10x

2x

13x

3x

1A B

Fig. 6. Mutation of Trp43Ala blocks Dcp1 activation in S. pombe and coac-tivation by Edc1 in S. cerevisiae. (A) Observed rates for 5 μM S. pombe dec-apping proteins (kmax conditions) at 0.1 °C are 0.02 min−1 for spDcp2,0.2 min−1 for spDcp1∶Dcp2, 0.002 min−1 for spDcp2 Trp43Ala, and0.006 min−1 for spDcp1∶Dcp2 Trp43Ala. (B) Observed rates for the S. cerevi-siae Dcp1∶Dcp2ð1–245Þ complex with and without Edc1CTR at 20 nM at 4 °C are0.014 for scDcp1∶Dcp2, 0.18 for scDcp1∶Dcp2þ Edc1CTR, 0.0019 forscDcp1∶Dcp2 Trp50Ala, and 0.0059 for scDcp1∶Dcp2 Trp50Ala þ Edc1CTR.Rates were measured under kmax∕KM conditions for budding yeast Dcp2 be-cause reactions were too fast to follow by manual pipetting and previouswork showed the Edc1CTR affects KM only threefold (17). Error is SEM fromthree independent experiments.

Floor et al. PNAS ∣ February 21, 2012 ∣ vol. 109 ∣ no. 8 ∣ 2875

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

27,

202

0

DiscussionThe activity of many enzymes is regulated by macroscopicopen-to-closed transitions, for example adenylate kinase (23, 39),imidazole glycerol phosphate synthase (24), DEAD-(Asp-Glu-Ala-Asp) box helicases (40, 41), and, it seems, some ubiquitinE2 enzymes by closure of an attached ubiquitin (42, 43). Nudixenzymes can also be regulated by domains outside the Nudixmotif (44, 45). In the case of Dcp2, a composite active site isformed upon cap recognition involving absolutely conserved re-gions on both domains (9). Tryptophan 43 is intimately tied toformation of this composite active site, as its mutation blockscap binding to the regulatory domain, enzyme closure, and dec-apping in vitro and in vivo (9, 32). Ligand binding promotesclosure of Dcp2 in a process that is mediated by contactsbetween the m7G cap structure and both the regulatory and cat-alytic domains of Dcp2 (9).

Previously we proposed, but did not test, the idea that Dcp1and associated coactivators like Edc1 use the Trp43-mediatedconformational change to enhance catalysis by Dcp2 (9, 17).Furthermore, it was puzzling to us how in the absence of Dcp1and coactivators the regulatory domain could still contribute 100-fold to kcat, because closure is not observed by SAXS under theseconditions (9, 17, 21). Here, we found using NMR that Dcp2undergoes collective fast-exchange between open and closedstates in the absence of any cofactors (Fig. 2, Fig. 4). An impor-tant observation is that mutation of Trp43 blocks both exchangeobserved by NMR and the ability of Dcp1 and Edc1CTR toenhance the catalytic step of Dcp2 (Fig. 3, Fig. 6). Therefore,it follows that cofactors such as Dcp1 and Edc1 stabilize a dyna-mically labile composite active site. This work provides insightsinto how decapping can be controlled by multiple layers of pro-tein-protein interactions: Dcp1 binds the regulatory domain andpromotes closure via Trp43, possibly by interacting with the cat-alytic domain of Dcp2; Edc1 likely consolidates through interac-tions with substrate or the catalytic domain of Dcp2. Altogether,the regulatory domain of Dcp2, its essential activator Dcp1, andthe coactivator Edc1 contribute four log-units to the catalytic step(9, 17, 33), which is mediated by Trp43 (Fig. 6). Tryptophan 43 istherefore a gatekeeper of closure with (9) and without bound sub-strate and is central to conformational changes in Dcp2 (Fig. 1,Fig. 3). It is thus not surprising that, in yeast, mutation of Trp43phenocopies deletion of the entire regulatory domain (32).

How are the dynamics observed in Dcp2 related to catalysis?For some enzymes like dihydrofolate reductase (5, 46), adenylatekinase (23), triosephosphate isomerase (47, 48), RNase A (49),and cyclophilin A (50) it seems that the rate-limiting step in thecatalytic cycle is a conformational change. Some enzymes areeven preorganized for catalysis by experiencing dynamics alongthe catalytic reaction coordinate without substrate (23, 50). Forunliganded Dcp2, the exchange rate between conformations ismuch faster than kcat (∼2;300 s−1, Table S1 versus ∼0.2 min−1,Fig. 6). Because previous studies on Dcp2 showed that productrelease is fast (33), the catalytic step is the slow step of the cat-alytic cycle (33), and that closure occurs after substrate binding asa substep of kcat (9, 33), we suggest the population of closed state,not the rates of interconversion per se, imposes a limit on thecatalytic rate. Preexisting equilibria play a key role in activationof signaling proteins where phosphorylation shifts a dynamic

equilibrium from inactive to active states (36, 51–53). Likewise,protein cofactors of Dcp2 may shift a highly skewed dynamicequilibrium between open, inactive, and closed active states, cul-minating in formation of the composite active site and catalyticrate enhancement. Confirmation of this prediction requiresdynamics studies in the presence of substrate, which remains achallenge for the future because we have not been able to gen-erate stable, concentrated samples of Dcp2 in a buffer compatiblewith ligand binding.

There are at least two possibilities for the structural natureof the excited state of Dcp2 in the absence of substrate. Onepossibility is that the excited state is in fact the closed, active con-formation. This model, that the apo dynamics are on-pathway, isconsistent with the abrogation of cap binding by the regulatorydomain, dynamics, stimulation by Dcp1, and coactivation byEdc1 upon mutation of Trp43 to alanine (Fig. 6) (9). However,we cannot exclude the possibility that the dynamics are due to anoff-pathway state. For example, the Dcp1∶Dcp2 complex wascrystallized in the presence of ATP in a closed conformation, soother states are not unprecedented (21). This ATP-bound closedstate is neither the active state nor the excited state though, be-cause mutation of the interdomain interface opposite Trp43 inthe closed crystal state does not affect kinetics, closure by SAXS,or the nitrogen HSQC without substrate (Fig. S6) (9). Due to thecentral nature of Trp43 in apo dynamics, closure, catalysis, andcoactivation, we favor the model that the apo dynamics are on-pathway. Independent of these distinctions is the key findingthat the conserved residue Trp43 functions as a gatekeeper ofthe composite active site, allowing decapping activity to be stimu-lated by multiple layers of protein-protein interactions, whichis an important regulatory event for control of 5′-3′mRNA decayin eukaryotic cells.

MethodsProteins were expressed in Escherichia coli and purified as described (9, 17,33). Methyl labeling was achieved by addition of 13C 2H labeled precursorsfor Ile (50 mgL−1), Leu/Val (100 mgL−1), Met (250 mgL−1), and Ala(100 mgL−1) 40 min prior to induction. All Dcp2 1–243 samples were perdeut-erated using 2H glucose as the sole carbon source and D2O, as described inref. 54. NMR experiments are explained in detail in SI Methods. Briefly, denovo assignment of spDcp2 ILVMA methyls was achieved using the (H)C(CO)NH-total correlation spectroscopy (TOCSY) (55) on independent domainsof spDcp2 and a methyl NOE compared to the crystal structure [Protein DataBank (PDB) ID code 2QKM]. CPMG experiments are described (56); intensitiesfor peaks were fit using Function and Data Analysis (FuDA, http://www.biochem.ucl.ac.uk/hansen/fuda) and CPMG dispersion curves were fit usingcpmg_fitd8 (a gift from D. Korzhnev and L. Kay, University of Toronto, Tor-onto, ON). Enzyme kinetics were measured under single-turnover condi-tions as described (9, 17, 57). Structure figures were made using PyMOL(http://pymol.org).

ACKNOWLEDGMENTS. We thank Mark Kelly and Jeff Pelton for NMR support,along with Dmitry Korzhnev for the cpmg_fitd8 software and advice andD. Flemming Hansen for Function and Data Analysis and fitting advice. Thiswork was supported in part or in full by fellowships from the Sandler FamilyFoundation for Basic Sciences and the Achievement Awards for CollegeScientists (ARCS) Foundation (S.N.F.), a US National Institutes of GeneralMedical Sciences predoctoral fellowship 1R25GM56847 (to M.S.B.), a USNational Institutes of Health Grant R01GM078360 (to J.D.G.), and a USNational Institutes of Health Grant GM68933 for the Central California900-MHz facility.

1. Henzler-WildmanK, KernD (2007)Dynamic personalities of proteins.Nature 450:964–972.

2. McElroy C, Manfredo A, Wendt A, Gollnick P, Foster M (2002) TROSY-NMR studies of

the 91 kDa TRAP protein reveal allosteric control of a gene regulatory protein by

ligand-altered flexibility. J Mol Biol 323:463–473.

3. Tzeng S-R, Kalodimos CG (2009) Dynamic activation of an allosteric regulatory protein.

Nature 462:368–372.

4. Boehr DD, Nussinov R, Wright PE (2009) The role of dynamic conformational ensem-

bles in biomolecular recognition. Nat Chem Biol 5:789–796.

5. Bhabha G, et al. (2011) A dynamic knockout reveals that conformational fluctuations

influence the chemical step of enzyme catalysis. Science 332:234–238.

6. Hyeon C, Jennings PA, Adams JA, Onuchic JN (2009) Ligand-induced global transitions

in the catalytic domain of protein kinase A. Proc Natl Acad Sci USA 106:3023–3028.

7. Zou P, Bortolus M, McHaourab HS (2009) Conformational cycle of the ABC transporter

MsbA in liposomes: Detailed analysis using double electron-electron resonance spec-

troscopy. J Mol Biol 393:586–597.

8. Mallam AL, et al. (2011) Solution structures of DEAD-box RNA chaperones reveal

conformational changes and nucleic acid tethering by a basic tail. Proc Natl Acad

Sci USA 108:12254–12259.

9. Floor SN, Jones BN, Hernandez GA, Gross JD (2010) A split active site couples cap

recognition by Dcp2 to activation. Nat Struct Mol Biol 17:1096–1101.

2876 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1113620109 Floor et al.

Dow

nloa

ded

by g

uest

on

June

27,

202

0

10. Schier AF (2007) The maternal-zygotic transition: Death and birth of RNAs. Science316:406–407.

11. Hilgers V, Teixeira D, Parker R (2006) Translation-independent inhibition of mRNAdeadenylation during stress in Saccharomyces cerevisiae. RNA 12:1835–1845.

12. Thompson DM, Parker R (2007) Cytoplasmic decay of intergenic transcripts in Sacchar-omyces cerevisiae. Mol Cell Biol 27:92–101.

13. Isken O, Maquat LE (2007) Quality control of eukaryotic mRNA: Safeguarding cellsfrom abnormal mRNA function. Genes Dev 21:1833–1856.

14. Behm-Ansmant I, et al. (2006) mRNA degradation by miRNAs and GW182 requiresboth CCR4∶NOT deadenylase and Dcp1∶Dcp2 decapping complexes. Genes Dev20:1885–1898.

15. Guo H, Ingolia NT, Weissman JS, Bartel DP (2010) Mammalian microRNAs predomi-nantly act to decrease target mRNA levels. Nature 466:835–840.

16. van Dijk EL, et al. (2011) XUTs are a class of Xrn1-sensitive antisense regulatorynon-coding RNA in yeast. Nature 475:114–117.

17. Borja MS, Piotukh K, Freund C, Gross JD (2011) Dcp1 links coactivators of mRNAdecapping to Dcp2 by proline recognition. RNA 17:278–290.

18. Dunckley T, Tucker M, Parker R (2001) Two related proteins, Edc1p and Edc2p,stimulate mRNA decapping in Saccharomyces cerevisiae. Genetics 157:27–37.

19. Schwartz D, Decker CJ, Parker R (2003) The enhancer of decapping proteins, Edc1p andEdc2p, bind RNA and stimulate the activity of the decapping enzyme. RNA 9:239–251.

20. DeRisi JL, Iyer VR, Brown PO (1997) Exploring the metabolic and genetic control ofgene expression on a genomic scale. Science 278:680–686.

21. She M, et al. (2008) Structural basis of dcp2 recognition and activation by dcp1. MolCell 29:337–349.

22. She M, et al. (2004) Crystal structure of Dcp1p and its functional implications in mRNAdecapping. Nat Struct Mol Biol 11:249–256.

23. Henzler-Wildman KA, et al. (2007) Intrinsic motions along an enzymatic reactiontrajectory. Nature 450:838–844.

24. Lipchock J, Loria J (2009) Millisecond dynamics in the allosteric enzyme imidazoleglycerol phosphate synthase (IGPS) from Thermotoga maritima. J Biomol NMR45:73–84.

25. Mittermaier A, Kay LE (2006) New tools provide new insights in NMR studies ofprotein dynamics. Science 312:224–228.

26. Ruschak A, Kay L (2009) Methyl groups as probes of supra-molecular structure,dynamics and function. J Biomol NMR 46:75–87.

27. Carver J, Richards R (1972) General 2-site solution for chemical exchange produceddependence of T2 upon Carr–Purcell pulse separation. J Magn Reson 6:89–105.

28. Palmer AG, Gray MJ, Wang C (2005) Solution NMR spin relaxation methods forcharacterizing chemical exchange in high-molecular-weight systems. Methods Enzy-mol 394:430–465.

29. Davis DG, PerlmanME, London RE (1994) Direct measurements of the dissociation-rateconstant for inhibitor-enzyme complexes via the T1 rho and T2 (CPMG) methods.J Magn Reson B 104:266–275.

30. Allerhand A, Thiele E (1966) Analysis of Carr–Purcell spin-echo NMR experiments onmultiple-spin systems. II. The effect of chemical exchange. J Chem Phys 45:902–916.

31. Wang Z, Jiao X, Carr-Schmid A, Kiledjian M (2002) The hDcp2 protein is a mammalianmRNA decapping enzyme. Proc Natl Acad Sci USA 99:12663–12668.

32. She M, et al. (2006) Crystal structure and functional analysis of Dcp2p from Schizosac-charomyces pombe. Nat Struct Mol Biol 13:63–70.

33. DeshmukhMV, et al. (2008) mRNA decapping is promoted by an RNA-binding channelin Dcp2. Mol Cell 29:324–336.

34. Tugarinov V, Hwang PM, Ollerenshaw JE, Kay LE (2003) Cross-correlated relaxationenhanced 1H[bond]13C NMR spectroscopy of methyl groups in very high molecularweight proteins and protein complexes. J Am Chem Soc 125:10420–10428.

35. Farmer BT, et al. (1996) Localizing the NADP+ binding site on the MurB enzyme byNMR. Nat Struct Biol 3:995–997.

36. Yu B, et al. (2010) Structural and energetic mechanisms of cooperative autoinhibitionand activation of Vav1. Cell 140:246–256.

37. Sakuno T, et al. (2004) Decapping reaction of mRNA requires Dcp1 in fission yeast: Itscharacterization in different species from yeast to human. J Biochem 136:805–812.

38. Beelman CA, et al. (1996) An essential component of the decapping enzyme requiredfor normal rates of mRNA turnover. Nature 382:642–646.

39. Wolf-Watz M, et al. (2004) Linkage between dynamics and catalysis in a thermophilic-mesophilic enzyme pair. Nat Struct Mol Biol 11:945–949.

40. Montpetit B, et al. (2011) A conserved mechanism of DEAD-box ATPase activation bynucleoporins and InsP6 in mRNA export. Nature 472:238–242.

41. Caruthers JM, McKay DB (2002) Helicase structure and mechanism. Curr Opin StructBiol 12:123–133.

42. Wickliffe KE, Lorenz S, Wemmer DE, Kuriyan J, Rape M (2011) The mechanism oflinkage-specific ubiquitin chain elongation by a single-subunit E2. Cell 144:769–781.

43. Pruneda JN, Stoll KE, Bolton LJ, Brzovic PS, Klevit RE (2011) Ubiquitin in motion: Struc-tural studies of the ubiquitin-conjugating enzyme < ubiquitin conjugate. Biochemistry50:1624–1633.

44. Gabelli SB, Bianchet MA, Bessman MJ, Amzel LM (2001) The structure of ADP-ribosepyrophosphatase reveals the structural basis for the versatility of the Nudix family.NatStruct Biol 8:467–472.

45. Mildvan AS, et al. (2005) Structures and mechanisms of Nudix hydrolases. ArchBiochem Biophys 433:129–143.

46. Boehr DD, McElheny D, Dyson HJ, Wright PE (2006) The dynamic energy landscapeof dihydrofolate reductase catalysis. Science 313:1638–1642.

47. Rozovsky S, Jogl G, Tong L, McDermott AE (2001) Solution-state NMR investigationsof triosephosphate isomerase active site loop motion: Ligand release in relation toactive site loop dynamics. J Mol Biol 310:271–280.

48. Rozovsky S, McDermott AE (2001) The timescale of the catalytic loop motion intriosephosphate isomerase. J Mol Biol 310:259–270.

49. Beach H, Cole R, Gill ML, Loria JP (2005) Conservation of mus-ms enzyme motions inthe apo- and substrate-mimicked state. J Am Chem Soc 127:9167–9176.

50. Eisenmesser EZ, et al. (2005) Intrinsic dynamics of an enzyme underlies catalysis.Nature 438:117–121.

51. Gardino AK, et al. (2009) Transient non-native hydrogen bonds promote activationof a signaling protein. Cell 139:1109–1118.

52. Volkman BF, Lipson D, Wemmer DE, Kern D (2001) Two-state allosteric behavior in asingle-domain signaling protein. Science 291:2429–2433.

53. Li P, Martins IRS, Amarasinghe GK, Rosen MK (2008) Internal dynamics control activa-tion and activity of the autoinhibited Vav DH domain. Nat Struct Mol Biol 15:613–618.

54. Tugarinov V, Kanelis V, Kay LE (2006) Isotope labeling strategies for the study ofhigh-molecular-weight proteins by solution NMR spectroscopy. Nat Protoc 1:749–754.

55. Gardner KH, Konrat R, Rosen MK, Kay LE (1996) An (H)C(CO)NH-TOCSY pulse schemefor sequential assignment of protonated methyl groups in otherwise deuterated (15)N, (13)C-labeled proteins. J Biomol NMR 8:351–356.

56. Skrynnikov NR, Mulder FA, Hon B, Dahlquist FW, Kay LE (2001) Probing slow timescaledynamics at methyl-containing side chains in proteins by relaxation dispersion NMRmeasurements: Application to methionine residues in a cavity mutant of T4 lysozyme.J Am Chem Soc 123:4556–4566.

57. Jones BN, Quang-Dang D-U, Oku Y, Gross JD (2008) A kinetic assay to monitor RNAdecapping under single-turnover conditions. Meth Enzymol 448:23–40.

Floor et al. PNAS ∣ February 21, 2012 ∣ vol. 109 ∣ no. 8 ∣ 2877

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

27,

202

0