Active Disruption of an RNA-Protein Interaction by a DExH/D RNA Helicase

See allHide authors and affiliations

Science  05 Jan 2001:
Vol. 291, Issue 5501, pp. 121-125
DOI: 10.1126/science.291.5501.121


All aspects of cellular RNA metabolism and the replication of many viruses require DExH/D proteins that manipulate RNA in a manner that requires nucleoside triphosphates. Although DExH/D proteins have been shown to unwind purified RNA duplexes, most RNA molecules in the cellular environment are complexed with proteins. It has therefore been speculated that DExH/D proteins may also affect RNA-protein interactions. We demonstrate that the DExH protein NPH-II from vaccinia virus can displace the protein U1A from RNA in an active adenosine triphosphate–dependent fashion. NPH-II increases the rate of U1A dissociation by more than three orders of magnitude while retaining helicase processivity. This indicates that DExH/D proteins can effectively catalyze protein displacement from RNA and thereby participate in the structural reorganization of ribonucleoprotein assemblies.

Many DExH/D proteins hydrolyze nucleoside triphosphates (NTPs) in a reaction that is stimulated by nucleic acids, and unwind RNA duplexes in an NTP-dependent fashion in vitro (1). DExH/D proteins are frequently part of large ribonucleoprotein (RNP) assemblies such as the spliceosome or viral replication machineries (2, 3). In some instances, DExH/D proteins have been shown to couple NTP hydrolysis to conformational changes in these complexes (4–6), and it is generally believed that this represents the predominant function of these enzymes in RNP assemblies (2).

Despite the importance of DExH/D proteins, little is known about the mechanisms by which these enzymes effect the numerous conformational changes that occur in RNP machines. It has been demonstrated that DExH/D proteins can function as processive and directional molecular motors for unwinding regular RNA duplexes (7). Although unwinding of regular duplex RNA is clearly important in RNA metabolism, cellular RNA often has a more complex structure and is likely to be bound to proteins. This fact has prompted the attractive hypothesis that DExH/D proteins might not necessarily be “pure” RNA helicases; rather, they may also function to disrupt or rearrange RNA-protein interactions (2). However, such activity by DExH/D proteins has never been demonstrated.

We tested the ability of DExH/D proteins to displace proteins from RNA by investigating whether the DExH protein NPH-II from vaccinia virus can displace the protein U1A from an RNA substrate (8). NPH-II is an RNA helicase that unwinds RNA duplexes processively in the 3′ to 5′ direction with a kinetic step size of roughly one-half helical turn (7). Use of a kinetically well-characterized RNA helicase permits direct comparisons of the RNA unwinding and protein displacement activities. U1A is an ideal target protein because its RNA-binding properties have been characterized (9). U1A binds RNA through an NH2-terminal RNP domain (10), which is the most common motif for mediating specific RNA-protein interactions (11). The active displacement of U1A is of particular interest because it is a constituent of the spliceosomal machinery and a feedback regulator of its own gene expression.

In order to simultaneously monitor U1A displacement and RNA helicase activity, a multifunctional RNA substrate was designed. The substrate contains the U1A binding site from the 3′-untranslated region (UTR) of U1A mRNA (Fig. 1A). U1A binds this motif as a dimer, interacting primarily with two asymmetric loop structures (12) that are imbedded within a set of duplex motifs (11) (Fig. 1D). To transform the U1A binding site into a helicase substrate, the hairpin loop was removed, and the flanking helical regions were lengthened (Fig. 1B). This type of extended two-piece substrate for U1A binding has previously been shown to retain subnanomolar affinity for U1A binding (9). In order to promote high-affinity NPH-II binding, a single-strand 3′ overhang was appended to the duplex region (Fig. 1B) (13). A blunt-ended control substrate was also synthesized, which contained the U1A binding site but lacked the single-strand overhang (Fig. 1C).

Figure 1

Substrate design and structure of the U1A binding site. (A) Sequence and secondary structure of the U1A binding site in the U1A mRNA 3′ UTR (30). Orange and blue letters correspond to the nucleotides retained in the substrates and present in the structure (D). (B) Substrate. Colored letters represent nucleotides retained in the wild-type U1A binding site; black letters correspond to the nucleotides added as described in the text. The duplex regions are identical to sequences included in constructs used to study the structure of the complex (9). The 24 nucleotide single-strand overhang (AN22U-3′) has the sequence 3′-UACAGUAACUACGACAAUCAUGCA. (C) Blunt-end control RNA. (D) Structure of the U1A RNA complex as determined by nuclear magnetic resonance (12) (Protein Data Bank accession no. 1DZ5). The two protein units are drawn as a transparent surface with ribbons representing the backbone. The two RNA strands are drawn as a ladder with the sticks corresponding to the bases and the ribbon corresponding to the backbone. The location of the 3′ end with the single-strand overhang on the RNA substrate is indicated.

U1A bound to both substrate and control RNA with high affinity (Fig. 2, A and C), demonstrating that the base-paired extensions and single-strand overhang did not alter the binding of U1A.

Figure 2

U1A binding to the RNA substrate and its effects on unwinding. (A) U1A binding to substrate RNA (Fig. 1B). Radiolabeled substrate (1 nM) was combined with U1A (10 nM) in a buffer containing 40 mM tris-HCl (pH 8.0) and 4 mM MgCl2 (in a final volume of 10 μl). After incubation at room temperature for 5 min, glycerol was added (8% v/v final) and the mixture was subjected to 8% native PAGE at 4°C, running at 10 V cm−1. Bands were visualized by a PhosphorImager. Species corresponding to free substrate RNA, bound U1A monomer, and bound U1A dimer are indicated at left. The asterisk represents the radiolabel. Left lane, RNA substrate bound to U1A; right lane, free substrate. (B) The effect of U1A binding on duplex unwinding. Reactions were performed at room temperature for 5 min with 1 nM RNA substrate and 20 nM NPH-II in a buffer of 40 mM tris-HCl (pH 8.0); 4 mM MgCl2; and, if applicable, 3.5 mM ATP (10 μl final volume). Where present, the U1A concentration was 10 nM. Substrate and U1A were preincubated for 5 min at room temperature. NPH-II was added and then incubated for 5 min more. The reaction was then started by addition of ATP. Reactions were quenched by adding 10 μl of a solution containing 25 mM EDTA, 0.4% SDS, 0.05% bromphenol blue, 0.05% xylene cyanol, and 10% glycerol. Mixtures were subjected to 15% native PAGE, which was run at room temperature at 20 V cm−1. Lanes from left to right are as follows: unwinding reaction without ATP, NPH-II unwinding reaction, unwinding in the presence of 10 nM U1A, and boiled substrate. Unwound and duplex species are indicated by the cartoons at right. (C) U1A binding to the blunt-end control RNA. Binding reactions were performed as described and shown in (A). (D) Unwinding reactions with the blunt-end control RNA. Lanes correspond to those in (B).

The U1A binding site differs substantially from regular A-form helical geometry (Fig. 1D), and there is evidence that, even without bound U1A, the RNA is extensively bent (9). Despite this distortion in the RNA, NPH-II readily separated the two substrate strands in both the presence and absence of bound U1A (Fig. 2B). These findings establish that NPH-II can displace U1A. They also indicate that NPH-II can traverse loops and tolerate considerable bending in both substrate strands during duplex unwinding (14).

No unwinding was observed for the blunt-ended RNA substrate, regardless of whether U1A was bound (Fig. 2D). This provides two important controls: First, strand separation does not initiate at the internal loops; and second, U1A binding does not provide additional opportunities for NPH-II to initiate unwinding.

Next, it was important to distinguish whether NPH-II displaces U1A actively or in a passive manner. In the latter scenario, NPH-II would wait passively until U1A dissociates and then rearrange the binding site so that U1A can no longer bind. In an active process, NPH-II would affect the kinetics of U1A dissociation from the RNA. We reasoned that it should be possible to distinguish both processes by measuring the effect of NPH-II action on U1A dissociation rates (Fig. 3).

Figure 3

Active displacement of U1A by NPH-II. Dissociation experiments were conducted with 1 nM RNA substrate and 10 nM U1A in a buffer of 40 mM tris-HCl (pH 8.0) and 4 mM MgCl2 at 23°C (final volume, 40 μl). Reactions were performed as described in Fig. 2, except that they were initiated by addition of ATP (3.5 mM final concentration) and U1A trap [200 nM final concentration (15)], as indicated. Aliquots (6 μl) were withdrawn at 0, 1, 4, 8, 12, and 20 min and mixed with 2 μl of 100 mM EDTA and 2.5 μM NPH-II trap [which serves to capture dissociated NPH-II (16)] in 30% glycerol. Each aliquot was then loaded immediately on a 8% native polyacrylamide gel, which was run at 4°C at 10 V cm−1. Bands were visualized by a Phosphor- Imager. U1A-bound and free RNA as well as unwound and duplex RNA species are indicated by the cartoons at left (D, bound U1A dimer; M, bound U1A monomer; F, U1A free duplex substrate; U, unwound substrate). (A) Release of U1A upon addition of ATP (initiated by adding ATP together with U1A trap). (B) Release of U1A upon addition of NPH-II without ATP (initiated by adding only U1A trap). (C) Release of U1A in the presence of ATP and NPH-II (initiated by adding ATP together with U1A trap). (D) Trapping control: U1A trap was added together with RNA substrate to assess trapping efficiency. Aliquots were removed and treated as described above. (E to H) Same reactions as above, but with blunt-end control duplex.

The off rate for U1A was measured by saturating the substrate with U1A and, after complex formation, adding a large excess of RNA that contained another high-affinity U1A binding site (15). This prevented U1A from rebinding the substrate once it detached and enabled us to monitor the rate of U1A release by gel-shift electrophoresis (Fig. 3).

Without NPH-II, roughly 15% of U1A dissociates from the substrate within 20 min, which corresponds to an off rate of koff ∼ 10−2min−1 (Fig. 3A). In the presence of NPH-II, but without adenosine triphosphate (ATP), no unwinding is observed (compare Fig. 1B) and the off rate was not significantly changed (Fig. 3B), which indicates that U1A is not displaced by mere binding of NPH-II to the substrate. However, adding both NPH-II and ATP resulted in a dramatically increased off rate for U1A (Fig. 3C). After only 4 min, U1A was almost completely released from the substrate. This suggests a rate increase of several orders of magnitude and clearly demonstrates that NPH-II dissociates U1A from the substrate in an active energy-dependent fashion.

The rate of U1A dissociation from the blunt-end RNA is similar to the rate of U1A dissociation from the helicase substrate in the absence of NPH-II (Fig. 3E) or in the presence of NPH-II without ATP (Fig. 3F). However, unlike the helicase substrate, NPH-II combined with ATP does not increase the rate of U1A dissociation (Fig. 3G). Thus, displacement of U1A by NPH-II is not caused by the structural peculiarities of the U1A binding site but rather depends on binding of NPH-II to the single-strand overhang of the substrate.

Having established that NPH-II actively displaces U1A in an ATP-dependent fashion, it was of interest to determine how U1A binding impedes the helicase activity of NPH-II and to obtain a kinetic framework for the process of protein displacement by a DExH/D protein. To this end, U1A displacement was monitored under single-cycle conditions with respect to NPH-II; that is, any NPH-II that dissociates from the RNA cannot rebind. This was achieved by adding a large excess of trap RNA together with the ATP that is used to initiate unwinding of the NPH-II/substrate/U1A complex (16). In this manner, it was possible to monitor the relative fractions of U1A-bound RNA substrate, free duplex substrate, and unwound RNA strands (Fig. 4A). Although the decay of substrate bound to U1A was first order (17), the fraction of free duplex substrate passed through a maximum and the fraction of unwound substrate evolved with a small lag phase (Fig. 4, A and C). This indicates a sequential reaction and suggests the presence of a second slow step after U1A has been displaced.

Figure 4

Mechanism of U1A displacement by NPH-II. (A) Time course of U1A displacement and substrate unwinding with and without NPH-II trap. Reactions without NPH-II trap were conducted as described in Fig. 3. The reaction with NPH-II trap was initiated by adding a combination of ATP (3.5 mM final concentration), U1A trap (200 nM final concentration), and NPH-II trap (500 nM final concentration). Aliquots were withdrawn at the times indicated in the plots [(B) and (C)]. U1A- bound, free substrate, and unwound substrate species are indicated by the cartoons at left (bound, U1A dimer and monomer; free, nonbound and nonunwound substrate; unwound, unwound substrate). (B) Plot of reaction without NPH-II trap for bound, free, and unwound substrate [under normalized (26) multiple cycle conditions]. The monomer and dimer forms of bound U1A decayed at roughly the same rate and were therefore combined as the bound fraction. Solid lines are the simulated fits of the data based on the reaction mechanism described below (19), using the emipirically determined rate constants (D). (C) Plot of reaction with NPH-II trap (single-cycle conditions). Solid lines are the best fit to the integrated rate laws derived from the mechanism below (17). (D) Kinetic mechanism of U1A displacement and unwinding by NPH-II. The red circle represents NPH-II, and the blue elipsoids represent U1A. Rate constants were calculated according to integrated rate laws describing single-cycle reaction kinetics (17), using three different time courses (C). Abbreviations are as follows: ES, NPH-II–substrate–U1A complex before reaction initiation; I1, NPH-II–U1A–substrate complex after the first rate-limiting step; I′1, substrate-U1A complex (after NPH-II dissociation); I2, NPH-II–substrate complex (after U1A displacement); I′2, substrate (after U1A displacement and NPH-II dissociation); P, unwound product. At the end of the reaction, NPH-II dissociates rapidly and irreversibly from the substrate. The fraction of substrate bound to U1A consists of the species ES, I1, and I′1. The fraction of free substrate comprises I2 andI′ 2.

The most important observation, however, was that a sizable fraction of the substrate was unwound by NPH-II under single-cycle conditions; i.e., NPH-II was able to displace U1A and continue unwinding the substrate without necessarily falling off during the course of reaction. Thus, processivity was not eliminated by the binding of U1A. Nevertheless, U1A caused substantial defects in the processivity of NPH-II, as indicated by a plateau in the decay of bound U1A, the fact that the amplitude of free substrate did not return to zero, and the fact that unwinding did not go to completion but only to roughly 40% (Fig. 4, A and C). Taking all these observations together, it was possible to derive explicit equations describing the time courses and to model a basic kinetic mechanism for the reaction (Fig. 4D) (17).

In this mechanism, NPH initiates the displacement/unwinding reaction with a rate constant of k1 = 3.5 min−1. This rate constant is identical to that of the rate-limiting step for unwinding a regular duplex during the NPH-II helicase reaction, which involves a slow step at the junction between the single-strand overhang and the duplex region (7). After this initiation step, NPH-II proceeds to displace U1A. This step is fast compared to the initiation step. The actual rate for U1A displacement is therefore kinetically invisible. However, a lower limit for U1A displacement of k2 > 50 min−1 can be estimated (18), which is more then three orders of magnitude faster than the rate of U1A dissociation in the absence of NPH-II and ATP (∼10−2 min−1). Even before U1A is displaced, NPH-II dissociates with a rate of 0.7 · k2, which explains why only ∼60% of U1A molecules are released. After U1A is displaced, another slow step occurs (k3 = 1 min−1), in which a fraction of NPH-II dissociates from the substrate (k3d = 0.4 min−1). This second slow step (k3) is strictly dependent on the presence of U1A and was not observed during unwinding of the substrate without U1A (14). The kinetic steps above are likely to describe composite processes; i.e., the rate constants do not necessarily reflect microscopic reaction steps. Analysis of unwinding/displacement under multiple cycle conditions [in which dissociated NPH-II can rebind the substrate (19) (Fig. 4A, left panel)] indicated that no additional rate-altering steps other than rebinding events affect the reaction (Fig. 4B).

Four major mechanistic insights follow from the kinetic analysis: First, physical displacement of U1A is not the slowest step in the reaction, despite the high affinity of U1A to the substrate. Second, NPH-II increases the dissociation rate of U1A by more then three orders of magnitude. Third, NPH-II retains a significant level of processivity while displacing U1A. Fourth, after U1A is displaced, NPH-II needs to be reoriented or repositioned in order to complete substrate unwinding, as suggested by the second slow step (k3). There are at least two models by which NPH-II accelerates the dissociation of U1A protein: NPH-II may alter the conformation of RNA around the U1A binding site, or it may directly “plow” U1A off the RNA. Although the methods used here cannot distinguish these scenarios, the presence of intermediate species I2 (Fig. 4D) indicates that U1A displacement does not require the complete unwinding of the RNA duplex, thereby suggesting that a form of “snowplow” model is possible.

By showing that NPH-II actively displaces U1A, this study establishes that DExH/D proteins are capable of efficiently dislodging other proteins from RNA molecules. This RNP displacement, or “RNPase” function, is a form of enzymatic activity that is driven by ATP hydrolysis and which, like RNA helicase activity, is likely to have many different manifestations in cellular RNA metabolism. The observation that helicase processivity is not eliminated during U1A displacement suggests that DExH/D proteins may be able to switch back and forth between helicase and protein displacement functions, indicating that both activities can reside in the same protein and can function in the same macromolecular context (20). By obviating the need for numerous additional cofactors, this function may considerably simplify the requirements for RNP disassembly or rearrangement during processes such as pre-mRNA splicing or ribosome assembly.

  • * To whom correspondence should be addressed. E-mail: amp11{at}


View Abstract

Navigate This Article