Mechanistic and Kinetic Analysis of the DcpS Scavenger Decapping Enzyme*

Decapping is an important process in the control of eukaryotic mRNA degradation. The scavenger decapping enzyme DcpS functions to clear the cell of cap structure following decay of the RNA body by catalyzing the hydrolysis of m7GpppN to m7Gp and ppN. Structural analysis has revealed that DcpS is a dimeric protein with a domain-swapped amino terminus. The protein dimer contains two cap binding/hydrolysis sites and displays a symmetric structure with both binding sites in the open conformation in the ligand-free state and an asymmetric conformation with one site open and one site closed in the ligand-bound state. The structural data are suggestive of a dynamic decapping mechanism where each monomer could alternate between an open and closed state. Using transient state kinetic studies, we show that both the rate-limiting step and rate of decapping are regulated by cap substrate. A regulatory mechanism is established by the intrinsic domain-swapped structure of the DcpS dimer such that the decapping reaction is very efficient at low cap substrate concentrations yet regulated with excess cap substrate. These data provide biochemical evidence to verify experimentally a dynamic and mutually exclusive cap hydrolysis activity of the two cap binding sites of DcpS and provide key insights into its regulation.

The control of mRNA degradation is a critical step in the posttranscriptional regulation of gene expression, and the steady state level of any mRNA species depends on both the rate of mRNA synthesis and its breakdown. In eukaryotic cells, cytoplasmic mRNA degradation proceeds predominantly through an initial removal of the polyadenylated tail (1, 2) followed by 5Ј to 3Ј or 3Ј to 5Ј decay (3,4). In the 5Ј-3Ј decay pathway, the 5Ј cap structure is cleaved by the catalytic activity of the Dcp2 decapping enzyme to release the m 7 Gpp and monophosphorylated RNA (5)(6)(7)(8). The resulting uncapped monophosphorylated RNA is digested by a 5Ј-3Ј exonuclease, XrnI (1,9). In the 3Ј-5Ј pathway, subsequent to deadenylation, the capped RNA body is continuously degraded by an exosome complex (10,11). The resulting capped oligonucleotide m 7 GpppN(pN) n (n Ͻ 9) is hydrolyzed by a second type of decapping enzyme, DcpS, to release the m 7 Gp and ppN(pN) n products (12,13). These pathways need not be mutually exclusive and could occur simultaneously (14,15), and an interplay between the two pathways could also exist. DcpS, which was originally characterized as the decapping enzyme in the 3Ј-5Ј pathway, is also able to hydrolyze the 5Ј-3Ј decapping product m 7 Gpp to release m 7 Gp (16,17). Furthermore, disruption of the yeast DcpS ortholog, Dcs1p, impeded the 5Ј-exonuclease activity (18), indicating that the Dcs1p decapping products might serve as signaling molecules for the 5Ј decay pathway.
DcpS is a member of the histidine triad (HIT) 3 hydrolase family of proteins that contain a stretch of His-X-His-X-His-X residues, where X denotes hydrophobic amino acid residues. HIT proteins are dimeric nucleotide binding proteins that have hydrolase activities (19 -21). The central histidine residue is critical for the hydrolase activity, since it is thought to serve as the nucleophile attacking the phosphate most proximal to the methylated guanosine in m 7 GpppG (22) and is also critical for DcpS hydrolysis (13).
Structural analysis of DcpS has revealed that it is a homodimer with a symmetric structure when in the ligand-free form (16,23) or asymmetric homodimer in the ligand-bound form (16,24). Each DcpS monomer possesses a distinct N-terminal domain and a C-terminal domain containing the HIT motif linked by a hinge region. The N-terminal domain displays a domain-swapped form by exchanging an identical ␣-helix and two antiparallel ␤-strands with the second monomer. The DcpS homodimer contains two cap binding pockets, which serve as the active sites for cap hydrolysis. In the ligand-bound form, DcpS forms a closed conformation on one side and an open conformation on the other, with substrate bound at the C-terminal domain of each side (24). The structure suggests that the closed conformation constitutes the cap hydrolysis productive site, whereas the open site would be nonproductive. The N-terminal domains can both be in the open state, or one side can be in an open state with the second in a productive closed state with the hinge enabling the N terminus to flip back and forth, alternating on each side (16,24).
Here, we examined the enzyme kinetics of DcpS and demonstrate substrate binding and hydrolysis are regulated by negative cooperativity. Furthermore, by comparing the enzymatic behaviors of the wild type homodimer of DcpS and HIT mutant heterodimer, we biochemically confirm and expand upon the previously proposed dynamic mechanism of decapping implied by the structural analysis (16,24).

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The pcDNA3-FLAG-DcpS plasmid, which expresses FLAG-tagged DcpS in HEK293T cells, was constructed by inserting the DcpS open reading frame flanked by a BamHI site at the 5Ј-end and an XhoI site at the 3Ј-end into the pcDNA3-FLAG vector (25). A second FLAG-tagged DcpS expression plasmid, pcDNA3-FLAG-DcpS-2, was constructed by inserting the DcpS open reading frame with a FLAG tag fragment at the 5Ј of the open reading frame into the NcoI/XhoI sites of pcDNA3 vector (Invitrogen). Plasmids expressing a series of double-tagged homodimer and heterodimer proteins were generated by inserting two corresponding tagged wildtype or mutated DcpS open reading frames into the two multiple cloning regions of the commercial plasmid pETDuet-1 vector (Novagen, San Diego, CA). Plasmid pETDuet1-His-DcpS/FLAG-DcpS was constructed by inserting the NcoI (Klenow-filled)/XhoI restriction fragment from pcDNA3-FLAG-DcpS-2, as well as the NcoI/XhoI (Klenow-filled) restriction fragment from pET28-DcpS (13), into the EcoRV/ XhoI sites and NcoI/EcoRI (Klenow-filled) sites in pET-Duet-1, respectively. Plasmids pETDuet1-His-DcpS H277N /FLAG-DcpS and pETDuet1-His-DcpS W175A FLAG-DcpS N110A were constructed in a similar way, except with the indicated mutated sites introduced into the DcpS open reading frame. The mutagenesis of DcpS H277N was described in Ref. 13. The mutagenesis of N110A and W175A was carried out using the QuikChange mutagenesis system (Stratagene, La Jolla, CA) with the primer sets 5Ј-CTC CAA TTG CAG TTC TCC GCT  GAT ATA TAC AGC ACC TAT C-3Ј and 5Ј-GAT AGG TGC  TGT ATA TAT CAG CGG AGA ACT GCA ATT GGA G-3Ј  (for N110A) and 5Ј-CAG AGC CTC AGC ATC CAG GCG  GTG TAT AAC ATT CTC GAC-3Ј and 5Ј-GTC GAG AAT  GTT ATA CAC CGC CTG GAT GCT GAG GCT CTG-3Ј (for  W175A).
The single-tagged homodimer proteins His-DcpS BC/BC and FLAG-DcpS BC/BC were used in Fig. 4C and contain the N110A and W175A double mutations. They were expressed from plasmids pET28-DcpS N110A/W175A and pcDNA3-FLAG-DcpS N110A/W175A , respectively. These two plasmids were generated by introducing the N110A and W175A mutations into the plasmids pET28-DcpS and pcDNA3-FLAG-DcpS-2, respectively, using the primer sets described above.
Recombinant Protein Expression and Purification-Recombinant FLAG-DcpS was purified from HEK293T cells transfected with pcDNA3-FLAG-DcpS by using anti-FLAG M2-agarose beads (Sigma). 1 ϫ 10 8 293T cells were transfected with 120 g of pcDNA3-FLAG-DcpS and incubated at 37°C, 5% CO 2 to allow transient protein overexpression. Forty-eight hours posttransfection, cells were harvested and washed with 1ϫ PBS and resuspended in sonication buffer (150 mM KCl, 20 mM Tris-HCl, pH 7.9, 0.2 mM EDTA, 10% glycerol, 0.01% Nonidet P-40), followed by sonication for 30 s. The cell debris was then spun down, and the supernatant was incubated with 200 l of anti-FLAG M2-agarose beads at 4°C for 3 h with mild shaking to allow the FLAG-DcpS protein to bind to the beads.
The expression of double-tagged recombinant homodimer and heterodimer proteins was performed using the pETDuet1 expression system. Plasmids pETDuet1-His-DcpS/FLAG-DcpS, pETDuet1-His-DcpS H277N /FLAG-DcpS, and pET-Duet1-His-DcpS W175A /FLAG-DcpS N110A were used to transform BL21-CodonPlus(DE3)-RIPL competent cells (Stratagene; La Jolla, CA) according to the manufacturer's instructions to generate the DcpS WT/WT , DcpS WT/HIT , and DcpS WT/BC proteins, respectively. A single colony was picked and grown in 2 liters LB medium containing 100 mg/ml ampicillin and 20 mg/ml chloramphenicol to 0.6 A 600 and induced with 0.2 mM isopropyl 1-thio-␤-D-galactopyranoside at 30°C for 2-3 h. The bacterial cells were then washed and resuspended in binding buffer (300 mM urea, 5 mM imidazole, 500 mM NaCl, 20 mM Tris-HCl, pH 7.9) and sonicated three times for 30 s at 1-min intervals in ice. The cell debris was then spun down, and the cell lysate containing the recombinant homo-or heterodimer proteins was subjected to a two-step affinity purification procedure. The recombinant proteins were first purified on a nickel charged column for His-tagged protein binding, according to the manufacturer, except there was 300 mM urea and 0.5% Triton X-100 included in the binding buffer. The bound His-tagged proteins were eluted in 1.5 ml of elution buffer (500 mM imidazole, 500 mM NaCl, 20 mM Tris-HCl, pH 7.9) from the nickel column. The eluted proteins were then subjected to the second affinity column. The eluate was slowly added into 15 ml of FLAG immunoprecipitation binding buffer (150 mM KCl, 20 mM Tris-HCl, pH 7.9, 0.2 mM EDTA, 10% glycerol, 0.01% Nonidet P-40) and incubated with 200 l of prewashed anti-FLAG M2-agarose beads at 4°C for 3 h with mild shaking. The bound proteins were then washed with wash buffer and eluted as for FLAG-DcpS proteins. The eluted proteins are His-and FLAG-double-tagged homo-or heterodimers, since they underwent both His and FLAG column purification. They were subsequently concentrated by Centricon centrifugal filter columns (Amicon, Bedford, MA).
Recombinant protein concentrations were determined by spectrophotometry and calculated based on their extinction coefficient (26) obtained from the ExPASY ProtParam Web tool (27). Due to the relative purity of the protein sample shown in Fig. 1A, the detected absorbance was directly used with the DcpS extinction coefficient to determine the concentration. Protein samples shown in Figs. 3B and 4B contained a copurified bacterial protein, GroEL. Therefore, extinction coefficients of both proteins were used in the calculations, and the corresponding concentrations of the DcpS dimers were determined.
Generation of Cap Structures-Unlabeled, uncapped RNA corresponding to the pcDNA3 polylinker spanning from the SP6 promoter to the T7 promoter (pcP) with 16 guanosines at the 3Ј-end was transcribed by SP6 RNA polymerase from a PCR-generated template using the primers 5Ј-CGATTTAG-GTGACACTATAG-3Ј and 5Ј-CCCCCCCCCCCCCCCCCG-TAATACGACTCACTATAGGG-3Ј. Cap-labeled RNA was generated with the vaccinia virus capping enzyme utilizing [␣-32 P]GTP and S-adenosylmethionine to label the first phosphate within the cap relative to the methylated guanosine (m7G*pppG-), and the RNA was gel-purified as described (28). Labeled cap structure without the RNA body was generated by treating the cap-labeled RNA with 1 unit of Nuclease P1 (Sigma) for 1.5 h at 37°C to hydrolyze the RNA body, leaving the intact cap structure as described (12).
In Vitro Decapping Assays and Data Analysis-Decapping assays in Fig. 4C were carried out with the indicated monomer concentrations of proteins and 200 nM cold cap structure (New England Biolabs, Ipswich, MA) added to 32 P-labeled cap structure in decapping buffer (10 mM Tris, pH 7.5, 100 mM KOAc, 2 mM MgOAc, 2 mM dithiothreitol) for 30 s at room temperature. Decapping reactions were terminated by 1.7 N formic acid. An aliquot of each reaction was spotted onto polyethyleneiminecellulose TLC plates (Sigma) that were prerun in H 2 O and airdried, and the products were developed with 0.45 M (NH 4 ) 2 SO 4 at room temperature. The TLC plates were air-dried and exposed to PhosphorImager for quantitation. All quantitations were conducted with an Amersham Biosciences PhosphorImager (Storm860) using ImageQuant-5 software.
The decapping assays were conducted on a rapid quenchedflow instrument (KinTek Corp., Austin, TX) at 25°C. DcpS was loaded in one syringe of the quenched-flow apparatus. The cold cap substrate added to radiolabeled cap substrate with defined concentrations as indicated, were loaded in a second syringe. Decapping reactions were initiated by rapidly mixing equal volumes of solutions from both syringes. After the mixtures of enzyme and substrate were incubated for the indicated times, the reactions were quenched by 2.3 N formic acid added from a third syringe. An aliquot of each reaction was spotted onto TLC plates, developed, dried, and exposed to a PhosphorImager and quantitated as above.
The kinetics in Figs. 1C, 2A, and 4D were fit to Equation 1 using SigmaPlot 10.0 software (Systat Software, Inc., Point Richmond, CA), where Y represents the fraction of product m 7 Gp over total substrate, Y 0 is the interception representing the fraction of background m 7 Gp hydrolyzed from the input substrate in the absence of enzyme, A is the fraction of product generated from actual cleavage of enzyme, k obs is the observed rate constant of cap substrate hydrolysis, and t represents the reaction time.
The kinetics of DcpS WT/HIT under excess enzyme over substrate (single turnover conditions) in Fig. 3D were fit to Equation 2, where Y, Y 0 , A, and k obs represent the same parameters as in Equation 1. B is the amplitude of the second phase in the kinetics, C is the rate constant of the second phase, and t is time.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assays were carried out by incubating proteins with labeled cap structure in RNA binding buffer (75 mM KCl, 10 mM Tris-HCl, pH 7.5, 1.5 mM MgCl 2 , and 0.5 mM dithiothreitol) per reaction on ice for 15 min. The resulting protein-cap complexes were resolved on a 5.6% native polyacrylamide gel. The gel was dried and exposed to a PhosphorImager.

Hydrolysis of Cap Structure by DcpS Is Rate-limiting at the Binding
Step under Single Turnover Conditions-To gain a better understanding of the mechanism by which the DcpS scavenger decapping enzyme functions to hydrolyze cap structure, we undertook an enzymatic kinetic analysis of DcpS decapping. To measure the rate of the decapping reaction, in vitro decapping assays were performed under conditions where enzyme concentration was in excess of the substrate concentration (single turnover conditions), and time courses were measured using a rapid chemical quenched-flow apparatus. The DcpS monomer concentration was varied between 20, 50, 100, and 200 nM, and the cap substrate was kept constant at 10 nM and was spiked with ␣-32 P-labeled cap to follow the decapping products and reactants. The label is exclusively at the first phosphate following the methylated guanosine to enable detection of both the cap structure substrate (m 7 Gp*ppG) and the decapping product (m 7 Gp*). A Sypro Ruby (Invitrogen)-stained gel of FLAG-tagged DcpS purified from HEK293T cells is shown in Fig. 1A. The kinetics of cap hydrolysis was monitored from 0.1 s to 2 min at 25°C. The substrate and product were resolved by polyethyleneimine-cellulose TLC (Fig. 1B), and the fraction of generated product was plotted against time (Fig. 1C). The kinetics were fit to a single exponential equation (Equation 1) that provided the decapping rate constants at each enzyme concentration ( Table 1).
As shown in Fig. 1D and summarized in Table 1, the observed decapping rate constants depended on the DcpS concentrations and increased linearly with increasing DcpS. This dependence indicates that the observed rate of decapping is limited by the formation of the initial enzyme-substrate complex (i.e. the binding step is rate-limiting under the conditions used where the enzyme is in excess of substrate). If the rate-limiting step occurred after the enzyme-substrate binding step, increasing the amount of enzyme would have no effect on the observed reaction rate. Thus, the data are consistent with the rate-limiting step being the initial substrate binding step under the conditions of the experiments. We infer from these data that the decapping reaction rate constant is Ͼ1.3 s Ϫ1 ( Table 1). The bimolecular rate constant of substrate binding was obtained from the slope of the line in Fig. 1D and is equal to 6.8 Ϯ 0.6 ϫ 10 6 M Ϫ1 s Ϫ1 . To obtain the apparent dissociation constant (K d ), the amount of m 7 Gp product produced after the reaction reached equilibrium was plotted against the respective DcpS concentrations (20 -200 nM). The saturated amplitudes indicate that the DcpS-cap complex K d is Ͻ20 nM (Fig. 1E), consistent with the K d of 75 nM we previously reported (29).

The Open to Closed Conformational Step Is the Rate-limiting
Step under High Substrate Conditions-The above kinetic analysis demonstrates that the binding between DcpS and cap substrate is the rate-limiting step at the concentrations used in the above experiments and where the enzyme is in excess over the substrate. We next tested parameters with excess substrate and limiting enzyme concentrations. The decapping reaction was performed in the rapid chemical quenched-flow apparatus as above, except the DcpS monomer concentration of 100 nM and excess cap substrate concentrations of 200, 400, 800, and 1600 nM were used. The reaction products were resolved by thin layer chromatography as in Fig. 1B and plotted in Fig. 2A. The kinetics of the decapping reaction under these multiple turnover reaction conditions was initially linear, and then as substrate was depleted, the kinetics became nonlinear. The kinetic curves in Fig. 2B were fit to obtain the initial rate of decapping ( Table 1). The initial rate (nM/s) was divided by [DcpS] to obtain the psuedo-first order rate constants of decapping, which were plotted against the substrate concentrations (200 -1600 nM). As shown in Fig. 2B  To identify the rate-limiting step under conditions of excess cap over DcpS, we examined the presteady state portion of the kinetics of decapping in Fig. 2A. When product release of a reaction is rate-limiting, the reaction proceeds through an initial burst (with an amplitude close to the enzyme concentration) for the first round of substrate turnover. An initial steep slope with a subsequent shallow slope would indicate that the reaction proceeds through an initial rapid burst, but the subsequent rounds of hydrolysis are slower due to a slow release of the product. The amplitude of the initial burst would be close to FIGURE 1. Decapping kinetics under conditions of excess DcpS relative to cap substrate. A, recombinant FLAG-DcpS was overexpressed and purified from HEK293T cells and resolved by SDS-PAGE, followed by Sypro-Ruby staining. B, decapping reactions were carried out using monomer concentrations of 20, 50, 100, and 200 nM DcpS and 10 nM unlabeled cap structure substrate spiked with ␣-32 P-labeled cap by a rapid quenched-flow instrument at 25°C. The hydrolysis product and unhydrolyzed substrate were resolved by TLC developed in 0.45 M (NH 4 ) 2 SO 4 and exposed to a phosphor screen and developed on a PhosphorImager (see "Experimental Procedures"). C, the fraction of generated products was quantitated and plotted against the logarithmic value of the reaction time. The data points of each experiment were fit to a single exponential equation (Equation 1) to obtain the psuedo-first order decapping rate constants. The data represented here are an average of two sets of independent experiments carried out on two different days. The error bars indicate the data range for each time point of the experiment. D, the rate constants derived from C were plotted against DcpS concentrations. The slope of this linear plot gave the biomolecular rate constant for the cap hydrolysis. This value ranged between 2.3 Ϯ 0.1 ϫ 10 6 and 6.8 Ϯ 0.6 ϫ 10 6 M Ϫ1 s Ϫ1 based on four different data sets obtained through independent experimentation. E, the amount of m 7 Gp product generated was calculated from C and plotted against DcpS concentrations. Even under the lowest enzyme concentration tested here (20 nM), the reaction went to completion, indicating that the enzyme exhibits a very tight binding for the substrate and has an equilibrium dissociation constant (K d ) of less than 20 nM.  (24). By virtue of the hinge that separates the two domains, the N terminus has the capacity to flip from one side to the other, creating a closed complex bound to a cap substrate on one side to initiate hydrolysis, which in turn would force an open conformation on the other side and vice versa (24). Therefore, the structural data are consistent with a mutually exclusive activity of the two sites. If this proposed mechanism is true, both sides of the DcpS protein are not able to close at the same time. A prediction would be that when one side is locked into the closed conformation, the other side would remain open and would not be able to close freely to cleave its bound cap. To test this hypothesis biochemically, a heterodimer with one wild type active site and the other containing an asparagine substitution at His 277 in the HIT motif (DcpS WT/HIT ), which renders the protein inactive (13), was generated (Fig. 3, A and B).   (24). B, recombinant DcpS WT/HIT was overexpressed and purified from BL21-CodonPlus(DE3)-RIPL bacterial cells and resolved by SDS-PAGE, followed by Sypro-Ruby staining. Migration of the DcpS WT/HIT heterodimer is indicated as is the copurifying bacterial GroEL, whose identity was determined by mass spectrometry. C, an electrophoretic mobility shift assay was used to test the ability of DcpS WT/HIT heterodimer to bind 32 P-labeled cap analog. The DcpS WT/HIT heterodimer containing two different tags that were used to selectively purify the protein possesses a FLAG-tagged wild type protomer and a His-tagged HIT mutant protomer (lane 3). The His-tagged wild type homodimer was used in lane 2, and His-tagged HIT mutant homodimer was used in lane 4. The electrophoretic mobility shift assay reactions were carried out with 1 g of the indicated proteins and resolved on a 5.6% polyacrylamide gel. D, the rapid quenched-flow decapping assays of DcpS WT/HIT were carried out under single and multiple turnover conditions and resolved by TLC. Single turnover conditions of 100 nM monomer concentration of enzyme and 10 nM substrate are denoted by the circles. ϫ, multiple turnover conditions of 100 nM monomer concentration of enzyme and 800 nM substrate. The fractions of generated decapping product in each set of experiments were quantitated and plotted against the reaction times. The data points of the single turnover condition were fit to a biphasic, double exponential curve (Equation 2). The data represented here are an average of two sets of independent experiments carried out on different days. The error bars indicate the data range for each time point of the experiment. The maximal decapping rate of the multiple turnover experiment is lower than 2%.
To confirm the HIT mutant protomer within the DcpS WT/HIT heterodimer has the capacity to bind cap structure, we tested its binding by an electrophoretic mobility shift assay. As shown in Fig. 3C, lane 3, the DcpS WT/HIT heterodimer was capable of binding cap structure. This binding is probably mediated through the HIT mutant protomer, since a wild type protomer is expected to hydrolyze the cap and not stably bind it in this assay system. Consistent with this hypothesis, the wild type DcpS homodimer does not stably bind cap structure (lane 2), whereas as expected, the DcpS H277N (DcpS HIT/HIT ) homodimer, which can stably bind but not hydrolyze two cap structure molecules per homodimer protein (24), formed a detectable complex (lane 4). Therefore, the HIT mutant cap binding site within the DcpS WT/HIT should contain the capacity to bind and trap the cap substrate without hydrolyzing it. We expect the HIT protomer to bind and possibly interfere with hydrolysis on the wild type side of the heterodimer.
Rapid quenched-flow decapping assays were carried out with the DcpS WT/HIT heterodimer under limiting substrate single turnover (100 nM protein monomer, 10 nM substrate) and excess substrate multiple turnover (100 nM protein monomer, 800 nM substrate) conditions, and the decapping products were resolved by TLC. The data points from the single turnover experiments were fit to a biphasic, double exponential curve (Fig. 3D), and the kinetic values were calculated from Equation 2 and listed in Table 2. Under single turnover conditions, the DcpS WT/HIT protein was catalytically active. The decapping amplitude reached 20% in a single exponential phase, compared with ϳ100% for the DcpS WT/WT (compare Figs. 3D and 1C). These data suggest that 80% of the cap substrate was captured at the inactive HIT side, whereas only 20% was bound by the WT side and hydrolyzed. Moreover, the data suggest the HIT mutant side had a stronger affinity for the cap substrate and was able to bind and trap a majority of the cap substrate. The second exponential phase might represent the fact that substrates bound at the HIT side were slowly released and eventually hydrolyzed by the wild type side.
In contrast to the extent of decapping observed with limiting substrate, under conditions of excess substrate over enzyme, less then 2% of decapping was observed (Fig. 3D). These latter data suggest that the HIT mutant sides of almost all of the DcpS dimers were bound and locked by the cap substrates; therefore, the WT side was forced to remain in the open confirmation and unable to hydrolyze other cap substrates. These data confirm the proposed mechanism in which the two N termini of DcpS dimer cannot simultaneously hydrolyze the caps at both binding sites with both sites in the closed conformation, indicating that when one side is closed, the other side is forced open.

Decapping by a DcpS Heterodimer Containing One Bindingcompromised Mutant Subunit Displayed Decreased Negative
Cooperativity-To further confirm the negative cooperativity between the two active sites within the DcpS dimer, a DcpS heterodimer containing one wild-type active site and a mutated second site with minimal cap binding capacity and unable to hydrolyze cap substrate was generated. With a heterodimer containing only one active site and a second that was rendered binding-compromised, we would expect there to be minimal cooperativity within this heterodimer, since only one side of the protein can effectively interact with the cap substrate. To generate the heterodimer with one binding-compromised site, mutants were constructed such that two residues within one active site of the dimer critical for cap structure binding were substituted with alanine to abolish the cap binding function. Two residues that each make critical contacts to the methyl guanosine moiety of the cap structure (24) and essential for cap hydrolysis, asparagines 110 and tryptophan 175 (24), were mutated to alanine. The rationale was to maximize the disruption of cap binding by using a double mutation. Two different mutant proteins were generated for this analysis, each with a distinct tag (FLAG tag and His tag). The first contained an asparagine 110 to alanine (N110A), whereas the second contained tryptophan 175 to alanine (W175A). Due to the domainswapped nature of DcpS, a heterodimer of these two mutant proteins is assembled in a manner where both mutated residues are located in the same cap binding site. The resulting heterodimer protein would contain one wild type active site and one inactive site housing both mutants (Fig. 4A, DcpS WT/BC ). Co-expression of FLAG-DcpS N110A and His-DcpS W175A in bacterial cells followed by sequential purification by a FLAG column and a nickel column ensured heterodimer isolation ( Fig. 4B; see "Experimental Procedures").
We previously showed that a homodimer containing either an N110A or W175A mutation at the cap binding sites possessed less than 5% activity of the wild type DcpS (24). To ascertain whether a homodimer containing a double mutation of both N110A and W175A retained any catalytic activity, a DcpS double binding-compromised mutant was generated (DcpS BC/BC ) and tested for decapping activity. A DcpS BC/BC homodimer contains both N110A and W175A mutations at each active site (total of 4 mutations/dimer). As shown in Fig.  4C, the DcpS BC/BC homodimer did not contain detectable decapping activity (lanes 6 -13) as opposed to 100% decapping activity observed with the same concentration of wild type protein (DcpS WT/WT ; lanes [2][3][4][5]. The decapping results demonstrate that the presence of both the N110A and W175A substitutions within the same active site on both monomers has the capacity to disrupt decapping to undetectable levels. Further- more, these data suggest that a heterodimer with one wild type site and the second site with the N110A and W175A double mutation should contain only one active site capable of binding and decapping cap structure, whereas the site with the double mutation should be inactive and retain minimal binding capacity to the cap structure. Decapping of the double-tagged DcpS WT/WT as well as the heterodimer with one wild type monomer and a second binding-compromised monomer (DcpS WT/BC ) was next tested. Decapping assays were carried out with DcpS WT/WT or DcpS WT/BC under single turnover conditions (100 nM monomer concentration of DcpS enzyme, 10 nM cap substrate) and multiple turnover conditions (100 nM monomer concentration of DcpS enzyme, 800 nM cap substrate). The decapping products were resolved by TLC, and the fraction of product generated was determined and plotted in Fig. 4D for the DcpS WT/BC heterodimer. The curves were fit to a single exponential equation (Equation 1). The rate constants and maximal fraction of generated decapping product were determined from the equation coefficients and listed in Table 2. We show that the rate constant for DcpS WT/WT , under multiple turnover conditions, was reduced by 8-fold compared with the single turnover conditions (0.735 Ϯ 0.037 s Ϫ1 versus 0.0955 Ϯ 0.0009 s Ϫ1 ), suggesting the existence of negative cooperativity in the wild-type homodimer. Interestingly, under the same conditions, the rate constant of the heterodimer DcpS WT/BC was reduced by only 2-fold compared with the single turnover conditions (0.08 Ϯ 0.002 s Ϫ1 versus 0.042 Ϯ 0.0009 s Ϫ1 ), indicating that higher concentrations of substrate could still negatively impact hydrolysis at the active site, but the impact is significantly lower compared with that observed with the wild type homodimer. The failure of the DcpS WT/BC heterodimer to be completely homodimer has no detectable decapping activity with the decapping conditions employed. An in vitro decapping assay was carried out with the indicated monomer concentrations of DcpS BC/BC homodimer and 200 nM unlabeled cap structure spiked with 32 P-labeled cap structure. The reactions were incubated for 30 s at room temperature and terminated by 1.7 N formic acid. The hydrolyzed product m 7 Gp and unhydrolyzed cap structure substrate were resolved by TLC and exposed to the phosphor screen and developed on a PhosphorImager. D, rapid quenched-flow decapping assays of DcpS WT/BC carried out under single turnover and multiple turnover conditions. A 100 nM monomer concentration of DcpS WT/BC was used with 10 nM cap substrate spiked with 32 P-labeled cap substrate under single turnover conditions. 100 nM monomer concentration of DcpS WT/BC was used with 800 nM cap substrate spiked with 32 P-labeled cap substrate under multiple turnover conditions. The fraction of decapping product generated in each experiment was carried out as duplicates on two different days. The averages of both of these data sets have been quantitated and plotted against the reaction times, with the error bars indicating the data range obtained with each set. The data points for each set of experiments were fit to a single exponential equation (Equation 1) to obtain the rate constant. resistant to negative cooperativity could be due to the ability of the double mutant site still retaining residual cap binding capacity (see "Discussion").

DISCUSSION
In this report, we present kinetic analysis of DcpS and elucidate several novel insights into its decapping mechanism at the subunit level. Our kinetic analysis shows that the rate-limiting step in the DcpS-catalyzed decapping reaction depends on the substrate concentration relative to the enzyme. When DcpS concentration is in excess of cap substrate, the decapping reaction rate is fast, and the observed rate is limited at the initial substrate binding step. Under high substrate conditions, the rate-limiting step is shifted to the conformational change/hydrolysis step (Table 1). We demonstrate that a dynamic conformational change of the N-terminal domain relative to the C-terminal domain is essential for hydrolysis (Fig. 3), confirm-ing a mutually exclusive hydrolysis function between the two catalytic active sites.
An important mechanistic observation in these studies was that the rate-limiting step of DcpS decapping changed from the initial substrate binding step to a subsequent conformational change/hydrolysis step when the amount of substrate was higher than DcpS ( Figs. 1 and 2). Therefore, at excess substrate concentrations, when both active sites of the DcpS dimer are occupied by substrate, we propose that the open to closed conformational change necessary for cap hydrolysis is slow (Fig. 5). Such a negative allosteric regulation of DcpS is supported by several observations. First, the observed rate constant was reduced by 8-fold with wild type DcpS homodimer under high substrate conditions, whereas the DcpS WT/BC heterodimer with one wild type active site and a binding-compromised active site only displayed a 2-fold reduction of the rate constant ( Table 2). Although we expected the DcpS WT/BC to be com- FIGURE 5. Models of decapping mechanism under single (low substrate/enzyme ratio) and multiple (high substrate/enzyme ratio) turnover conditions. A, decapping model under conditions when substrate concentration is less than enzyme active sites displays no cooperativity between the two protomers. The ligand-free DcpS dimer exhibits a symmetric conformation or open conformation. Upon binding to the cap substrate at one binding site, the N terminus at this site closes at a fast rate for substrate cleavage (closed conformation), followed by releasing of the decapping product m 7 Gp and ppN to complete a catalytic cycle. Only one protomer is used per cycle; therefore, there is no allosteric communication between the first and second protomers. B, DcpS hydrolysis of cap substrate displays negative cooperativity under conditions where substrate concentration is in excess of enzyme active sites. Under high substrate conditions, two substrates are bound to the dimer, and the conformational change involved in the closing of the N terminus on one of the substrates is either an unfavorable step or a slow limiting step. The rigidity of the domain-swapped region restrains the N terminus of the second protomer to stay in an open position. After the substrate bound at the first site is hydrolyzed, the decapping products are released, and the N terminus of the second site closes at the same slow rate for hydrolysis, followed by product release to complete the catalytic cycle.
pletely resistant to negative cooperativity rather than partially resistant, our experimental results indicate that the bindingcompromised monomer retains residual cap binding. This is not surprising, considering the extensive network of amino acid contacts between DcpS and the cap structure (24). Additional multiple mutations to completely disrupt cap binding were not attempted, since, as indicated in Table 2, the decapping rate constant under single turnover conditions for the DcpS WT/BC was reduced relative to the wild type DcpS WT/WT protein.
These data suggest that mutations in the cap binding are not well tolerated by the protein and further mutations were not tested.
Our data are consistent with an inherent negative cooperativity of the two active sites within DcpS, as indicated by the cap-bound co-crystal structure (16,24). A model was proposed whereby the N termini of both monomers within the DcpS dimer acts as a single inflexible domain that can alternate back and forth to form hydrolysis-competent closed and nonproductive open conformations (24). This was expanded upon to include an intermediate state of both sites in the open confirmation (16). The high rigidity of the two intertwined N-terminal domains prevents them from closing at the same time. Therefore, when an N terminus on a monomer closes, the one on the other is forced into an open position, and only the closed form is active in decapping. The site in the open conformation is capable of cap substrate binding (24) but not its hydrolysis. Therefore, the DcpS dimer displays inherent negative cooperativity in decapping. Our results with the DcpS WT/HIT heterodimer provide experimental confirmation for the negative allosteric model, where less then 2% decapping activity was detected under conditions of high substrate relative to enzyme (Fig. 3D). Therefore, the biochemical analysis indicates that the WT site was not able to conform to a closed productive confirmation for hydrolysis when the HIT mutant site was bound and "locked" by the substrate. Our biochemical demonstration of cooperativity between the two active sites is consistent with a recent molecular dynamics simulation of the apo-form of DcpS, which revealed that the protein dimer moves in a cooperative manner, where one side closes and the other side opens (30).
Collectively, our analysis of the heterodimers DcpS WT/HIT and DcpS WT/BC combined with the decapping kinetics of the wild type DcpS homodimer under both low and high substrates as well as the known structural properties of DcpS suggest the following model of hydrolysis; under single turnover conditions, upon binding to the cap substrate at one binding site, the N terminus at this site closes at a fast rate for substrate cleavage, followed by release of the decapping products m 7 Gp and ppN to complete a catalytic cycle (Fig. 5A). Since the amount of substrate is lower than the enzyme, only one monomer is used per cycle. Therefore, there is no allosteric communication between the first and second monomer. The observed rate up to 200 nM DcpS concentration is limited by substrate binding. Under conditions where the substrate concentration is in excess of DcpS, both subunits are bound with the cap substrate. We propose that the conformational change from open to closed is required for decapping on either side, and this conformational change is slow or unfavorable under excess substrate concentrations, and consequently the observed rate of decap-ping is slower. After the substrate bound at the first site is hydrolyzed, the hydrolyzed products must be released from the active site, and the vacant active site is reoccupied by another substrate molecule or by the product. Therefore, the same open to closed conformational change limits the hydrolysis of cap substrate at the second site (Fig. 5B). Although less likely, the formal possibility exists that the hydrolysis and release steps are slow or unfavorable.
An interesting mechanism that regulates DcpS protein activity has been revealed in Saccharomyces cerevisiae. This organism contains two DcpS homologs, Dcs1p and Dcs2p, with only Dcs1p containing decapping activity (13). Dcs1p and Dcs2p are involved in a stress coping mechanism (31). Under glucosedeprived conditions, the catalytically inactive paralog, Dcs2p, was shown to heterodimerize with catalytically active Dcs1p and compromised its substrate specificity and k cat (31,32). The DcpS WT/HIT human protein heterodimer is analogous to this situation, with one active and one inactive subunit, and similarly led to a dramatic decrease in decapping (Fig. 3). There does not appear to be a Dcs2p-like protein in human cells, suggesting that a heterodimer-mediated inhibition is unlikely; however, our data indicate that nonhydrolyzable substrates or compounds could bind in one active site and inhibit decapping by trapping the protein in an inactive state to inhibit DcpS decapping.
DcpS has been characterized as an mRNA decapping enzyme that is involved in mRNA decay. Whether DcpS is involved in other cellular processes is still an open question. The obvious metabolic pathways that DcpS may impact include nucleotide biogenesis and catabolic pathways due to its ability to hydrolyze cap dinucleotides. In addition, DcpS activity may also impact earlier steps of mRNA decay, since disruption of the yeast Dcs1p decapping activity results in inhibition of 5Ј to 3Ј exonuclease activity (18). As shown in this report, local increase of the substrate would inhibit DcpS decapping, which is analogous to the disruption of Dcs1. Thus, substrate and product inhibition of DcpS, which is part of the last step in mRNA decay, could be a means to feedback and regulate earlier steps in mRNA decay. Therefore, the negative cooperative nature of DcpS activity may provide a regulatory point for cross-talk between both pathways. Studies are under way to test this hypothesis.
In conclusion, our data provide evidence that DcpS exhibits different enzymatic kinetics under low and high substrate conditions. The presence of negative cooperativity between two wild type subunits as well as the dramatically reduced decapping activity displayed by the DcpS WT/HIT heterodimer under high substrate conditions validate our previous dynamic decapping model (24). Future studies to determine the local concentrations of DcpS protein and cap dinucleotide will begin to test the negative allosteric regulatory model in cells.