Recognition of double-stranded (ds) RNA is an important part of many cellular pathways, including RNA silencing, viral recognition, RNA editing, processing, and transport. dsRNA recognition is often achieved by dsRNA binding domains (dsRBDs). We use atomistic molecular dynamics simulations to examine the binding interface of the transactivation response RNA binding protein (TRBP) dsRBDs to dsRNA substrates. Our results explain the exclusive selectivity of dsRBDs toward dsRNA and against DNA-RNA hybrid and dsDNA duplexes. We also provide corresponding experimental evidence. The dsRNA duplex is recognized by dsRBDs through the A-form of three duplex grooves and by the chemical properties of RNA bases, which have 2'-hydroxyl groups on their sugar rings. Our simulations show that TRBP dsRBD discriminates dsRNA- from DNA-containing duplexes primarily through interactions at two duplex grooves. The simulations also reveal that the conformation of the DNA-RNA duplex can be altered by dsRBD proteins, resulting in a weak binding of dsRBDs to DNA-RNA hybrids. Our study reveals the structural and molecular basis of protein-RNA interaction that gives rise to the observed substrate specificity of dsRNA binding proteins.
Recognition of double-stranded (ds) RNA is an important part of many cellular pathways, including RNA silencing, viral recognition, RNA editing, processing, and transport. dsRNA recognition is often achieved by dsRNA binding domains (dsRBDs). We use atomistic molecular dynamics simulations to examine the binding interface of the transactivation response RNA binding protein (TRBP) dsRBDs to dsRNA substrates. Our results explain the exclusive selectivity of dsRBDs toward dsRNA and against DNA-RNA hybrid and dsDNA duplexes. We also provide corresponding experimental evidence. The dsRNA duplex is recognized by dsRBDs through the A-form of three duplex grooves and by the chemical properties of RNA bases, which have 2'-hydroxyl groups on their sugar rings. Our simulations show that TRBP dsRBD discriminates dsRNA- from DNA-containing duplexes primarily through interactions at two duplex grooves. The simulations also reveal that the conformation of the DNA-RNA duplex can be altered by dsRBD proteins, resulting in a weak binding of dsRBDs to DNA-RNA hybrids. Our study reveals the structural and molecular basis of protein-RNA interaction that gives rise to the observed substrate specificity of dsRNA binding proteins.
RNA
is emerging as an important regulatory element of many cellular processes.[1−4] Double stranded (ds) RNA is a frequent structural element of the
cellular RNA; accordingly, cells synthesize many proteins that recognize
it. dsRNA binding domain (dsRBD) is one of the most abundant RNA binding
domains,[5,6] found in proteins localized in both cell
nuclei and cytoplasm.[7] dsRBD proteins primarily
regulate gene expression and signaling events. For example, the protein
kinase PKR binds viral dsRNA and signals the onset of infection;[8] RNA helicase A (RHA) unwinds dsRNA;[9] the Dicer enzyme binds to and slices dsRNA into
21–22 nucleotide long pieces, which silence mRNA;[10,11] adenosine deaminase acting on RNA (ADAR) edits adenosine bases of
dsRNA and converts them to inosines.[12] Many
dsRBD proteins contain multiple dsRBDs. For example, two dsRBDs are
found in RHA, three dsRBDs are found in ADAR1, in protein activator
of PKR (PACT), and in transactivation response RNA binding protein
(TRBP), and five dsRBDs are found in Staufen.[7]TRBP participates in multiple cellular pathways in cell nuclei
and cytoplasm.[13] In gene silencing by the
RNA interference pathway, TRBP, Dicer, and PACT proteins are essential
components of the RNA-induced silencing complex.[14−17] Biochemical and cryo-electron
microscopy experiments have shown that two of TRBP dsRBDs aid in positioning
of dsRNA in the proper conformation for dsRNA cutting by Dicer.[18] Recent single molecule experiments established
that TRBP and Dicer-TRBP bind to and diffuse on dsRNA duplexes.[19] Interestingly, TRBP does not bind to duplex
types other than dsRNA, namely not to dsDNA and not to DNA–RNA
hybrids.[19,20] Many other dsRBD proteins specifically bind
to dsRNA duplexes, whereas some also weakly bind to DNA–RNA
hybrid duplexes.[21,22]Considering the above experimental
data about dsRBD binding targets, there are two questions of particular
interest: (1) How do dsRBDs discriminate between nucleic acid duplexes
of similar structures, such as dsRNA, DNA–RNA, and dsDNA? (2)
How do dsRBDs discriminate between dsRNA of different sequences, abundant
in the cell nuclei and cytoplasm? In the present study, we address
the first question. Currently, the mechanism of dsRBD discrimination
of dsRNA, DNA–RNA, and dsDNA has been explored only through
static (crystal) structures of several different dsRBDs to short pieces
of dsRNA[5] and short (2 ns) simulations.[23] Here, we examine by experiments and atomistic
molecular dynamics (MD) simulations the binding of TRBP dsRBDs to
dsRNA, DNA–RNA, and dsDNA duplexes. Our simulations identify
the molecular mechanism of dsRNA recognition by dsRBDs; in particular,
they reveal why dsRBDs bind weakly to DNA–RNA duplexes but
do not bind to dsDNA duplexes.
Methods
Electrophoretic Mobility
Shift Assay
To experimentally investigate binding of TRBP-RBD2
to dsRNA, DNA–RNA, and dsDNA, we performed an electrophoretic
mobility shift assay (EMSA). We prepared three types of 25-bp duplexes
with the same sequence (forward, 5′-GCUUGUCGGGAGCGCCACCCUCUGC-3′; reverse, 5′-GCAGAGGGUGGCGCUCCCGACAAGC-3′;
T instead of U for DNA strands), which were labeled with either DY547
or Cy3 at the 3′ end of the reverse strand. We incubated 10
nM of 25-bp duplexes with various concentrations of TRBP-RBD2 ranging
from 1 to 10 μM in binding buffer (20 mM Tris–HCl, pH
7.5, 25 mM NaCl, 1 mM DTT, and 0.1 mg/mL BSA) on the ice for 20 min.
Then, 8 μL of the incubated sample was mixed with 2 μL
of RNA loading dye (NEB) and the mixture was loaded immediately in
the 6% DNA retardation gel (Invitrogen). We took the fluorescence
gel images after running the gel at 100 V for 40 min and analyzed
the band density with ImageJ (http://rsbweb.nih.gov/ij/).
Atomic Resolution Models
To examine mechanisms of dsRNA
recognition by dsRBD domains, we simulated three complexes: (i) TRBP-RBD2:dsRNA,
(ii) TRBP-RBD2:DNA–RNA, and (iii) TRBP-RBD2:dsDNA, where TRBP-RBD2
labels the second dsRBD of TRBP. The TRBP-RBD1:dsRNA complex and dsRNA,
DNA–RNA, and dsDNA duplexes were simulated also separately
for comparison. The initial structures of complexes i–iii were
based on the crystal structure of TRBP-RBD2 bound to coaxially stacked
RNA duplexes (Protein Data Bank entry code 3ADL).[24] The initial
structures of dsRNA, DNA, and DNA–RNA duplexes, 35-base pairs
in length, were generated in their A-forms by the 3DNA software.[25] The simulated dsRNA duplexes contained complementary
strands, where the 5′ → 3′ strand had the sequence
UAACAACCAGAUCAAAGAAAAAACAGACAUUGUCA, as in our previous experiments.[19] DNA strands had sequences analogous to RNA strands,
with the exception that U bases were replaced with T bases. In the
initial structures of complexes, TRBP-RBD2 was positioned approximately
in the middle of the duplexes to avoid its interaction with duplex
edges, as shown in Figure 1.
Figure 1
Binding of TRBP-RBD2
to dsRNA. (a) Snapshot of the TRBP-RBD2:dsRNA complex. TRBP-RBD2 binds
to dsRNA across three grooves, in minor–major–minor
groove pattern. dsRNA minor grooves are lined with 2′ hydroxyl
groups (for clarity, only shown at the binding interface as red and
white spheres), and its major grooves are lined with phosphate groups
(P atoms are shown as tan spheres). Minor and major grooves, which
interact with TRBP-RBD2, are marked with red and tan brackets, respectively,
the two minor grooves being distinguished by labels I and II. The
RNA backbone is shown in gold, the bases are shown as transparent
sticks, and TRBP-RBD2 is colored according to its secondary structure,
where helices are shown in red, β-sheets in blue, and unstructured
loops in cyan. (b) Top view of the binding interface of the TRBP-RBD2:dsRNA
complex. TRBP-RBD2 residues within 3 Å of the dsRNA are shown
both as a transparent surface and in licorice representation. Blue
residues are positively charged, red residues are negatively charged,
green and cyan residues are polar, and white residues are nonpolar.
(c) Sequence alignment of two TRBP dsRBDs. The residues conserved
across different species and proteins are highlighted (Figure S1 (Supporting Information) and ref (5)), where the colors indicate
secondary structure elements, as shown in (a). The arrows point to
TRBP-RBD2 residues that contact dsRNA; red and tan arrows mark minor
and major groove contacts, respectively.
Binding of TRBP-RBD2
to dsRNA. (a) Snapshot of the TRBP-RBD2:dsRNA complex. TRBP-RBD2 binds
to dsRNA across three grooves, in minor–major–minor
groove pattern. dsRNA minor grooves are lined with 2′ hydroxyl
groups (for clarity, only shown at the binding interface as red and
white spheres), and its major grooves are lined with phosphate groups
(P atoms are shown as tan spheres). Minor and major grooves, which
interact with TRBP-RBD2, are marked with red and tan brackets, respectively,
the two minor grooves being distinguished by labels I and II. The
RNA backbone is shown in gold, the bases are shown as transparent
sticks, and TRBP-RBD2 is colored according to its secondary structure,
where helices are shown in red, β-sheets in blue, and unstructured
loops in cyan. (b) Top view of the binding interface of the TRBP-RBD2:dsRNA
complex. TRBP-RBD2 residues within 3 Å of the dsRNA are shown
both as a transparent surface and in licorice representation. Blue
residues are positively charged, red residues are negatively charged,
green and cyan residues are polar, and white residues are nonpolar.
(c) Sequence alignment of two TRBP dsRBDs. The residues conserved
across different species and proteins are highlighted (Figure S1 (Supporting Information) and ref (5)), where the colors indicate
secondary structure elements, as shown in (a). The arrows point to
TRBP-RBD2 residues that contact dsRNA; red and tan arrows mark minor
and major groove contacts, respectively.The cionize code, available as a VMD plugin,[26] was used to place Na+ counterions
neutralizing nucleic acid charges around the prepared duplexes and
complexes. The resulting structures were solvated in TIP3P water and
50 mM NaCl with the VMD plugins solvate and ionize.[26] The final systems contained
approximately 80 000 atoms (in the case of simulations involving
duplexes) and 105 000 atoms (in the case of simulations involving
protein and duplexes).
Molecular Dynamics Simulations
MD
simulations were performed with NAMD2 software,[27] where the systems were described by assuming the AMBER
force field with SB[28] and BSC0[29] corrections, a suitable choice for describing
RNA.[30,31] The particle-mesh Ewald (PME) method[32] was used for evaluation of long-range Coulomb
interactions. The time step was set to 1.0 fs, and long-range interactions
were evaluated every 2 (van der Waals) and 4 timesteps (Coulombic).
After 2000 steps of minimization, ions and water molecules were equilibrated
for 2 ns around duplexes and complexes, which were constrained using
harmonic forces with a spring constant of 1 kcal/(mol Å2). Then, unconstrained duplexes and complexes were simulated for
50 and 100 ns, respectively. The simulations were performed in NpT ensemble, at a constant temperature T = 310 K, a Langevin constant γLang = 1.0 ps–1, and at a constant pressure p =
1 bar.
Data Analysis
To analyze the binding of TRBP-RBD2 to
three duplexes, we computed the contact area acon with dependence on time t, defined aswhere adsRBD, adupl, and adsRBD:dupl are the solvent accessible surface
areas (SASA) of TRBP-RBD2, the nucleic acid duplex, and the TRBP-RBD2:duplex
complex, respectively. The evaluation was done by the SASA built-in
VMD plugin,[26] where the van der Waals radius
of 1.4 Å was assigned to atoms to identify the points on a sphere
that are accessible to the solvent. We also calculated the interaction
energy between TRBP-RBD2 and each nucleic acid duplex by the NAMDEnergy
function in VMD.[26]The forms of duplexes
were analyzed from single averaged structures of duplexes and complexes.
These averaged structures were obtained by averaging the coordinates
of aligned duplexes and complexes over specified portions of trajectories.
The helical rise values for nucleic acid duplexes, which characterize
duplex forms, were evaluated with the 3DNA software.[25]
Results
The fact
that TRBP recognizes dsRNA over DNA–RNA and dsDNA has been
established in previous experiments.[19−21,33] Here we examine through MD simulations and experiment how TRBP dsRBDs
actually recognize dsRNA duplexes over nucleic acid duplexes that
contain DNA.
TRBP-RBD2 Binding to Nucleic Acid Duplexes in EMSA Gels
To examine relative binding affinities of TRBP-RBD2 for dsRNA, DNA–RNA,
and dsDNA duplexes, we performed electrophoretic mobility shift assays
(EMSA) for solutions containing these duplexes and 0, 1, 3, or 10
μM TRBP-RBD2. The resulting EMSA gels, shown in Figure 2, contain strong upper bands in lines 2 and 3, indicating
that TRBP-RBD2:dsRNA complexes are formed at 1, 3, and 10 μM
TRBP-RBD2 concentrations. The fact that in Figure 2b the location of the upper band in line 3 is higher than
in line 2 indicates that the number of bound TRBP-RBD2 molecules per
dsRNA duplex increases with TRBP-RBD2 concentration. There is usually
no relative affinity of dsRBDs for DNA–RNA and dsDNA reported
in the literature.[19−21] Here, the EMSA gel in Figure 2b shows that 10 μM TRBP-RBD2 binds to DNA–RNA and does
not bind to dsDNA. Therefore, TRBP-RBD2 has the greatest affinity
for dsRNA, a smaller affinity for DNA–RNA, and no discernible
affinity for dsDNA at the tested TRBP-RBD2 concentrations.
Figure 2
EMSA demonstrates
that the binding affinity of TRBP-RBD2 is strongest for dsRNA, and
weakest for dsDNA. (a) EMSA of 25-bp duplexes incubated with 0, 1,
and 3 μM of TRBP-RBD2. TRBP-RBD2 binds to dsRNA and does not
bind to DNA–RNA or dsDNA. (b) EMSA of 25-bp duplexes incubated
with 0, 1, and 10 μM of TRBP-RBD2. TRBP binds to dsRNA and DNA:RNA
(5–10 μM TRBP-RBD2, as shown in Figure S2, Supporting Information) and does not bind to
dsDNA. The quantified binding fraction of TRBP-RBD2 to each duplex
is displayed below the gel images.
EMSA demonstrates
that the binding affinity of TRBP-RBD2 is strongest for dsRNA, and
weakest for dsDNA. (a) EMSA of 25-bp duplexes incubated with 0, 1,
and 3 μM of TRBP-RBD2. TRBP-RBD2 binds to dsRNA and does not
bind to DNA–RNA or dsDNA. (b) EMSA of 25-bp duplexes incubated
with 0, 1, and 10 μM of TRBP-RBD2. TRBP binds to dsRNA and DNA:RNA
(5–10 μM TRBP-RBD2, as shown in Figure S2, Supporting Information) and does not bind to
dsDNA. The quantified binding fraction of TRBP-RBD2 to each duplex
is displayed below the gel images.
TRBP-RBD2 Binding to dsRNA in MD Simulations
In Figure 1a,b, we show a snapshot of TRBP-RBD2 in complex
with a 35-base pair RNA duplex, based on the crystal structure of
TRBP-RBD2 with coaxially stacked RNA duplexes.[24] TRBP-RBD2, a domain with the αβββα fold, binds to dsRNA along the duplex axis and across three grooves
of the duplex (minor–major–minor grooves), similarly
to other dsRBDs.[34−37] dsRNA minor grooves are lined with 2′-hydroxyl groups on
sugar rings of RNA nucleotides, and the major grooves are lined with
the negatively charged phosphate (−PO4–) groups. dsRNA duplex
assumes the A-form, in which the widths of minor and major grooves
are comparable (Figure 1a).The contacts
between TRBP-RBD2 and dsRNA are either direct (hydrogen bonding) or
water-mediated (hydrogen bonding through water molecules that have
long residence times). Bases of the minor groove I contact polar (His159,
Gln165) and negatively charged (Glu166) residues, and bases of the
minor groove II contact polar (His188) and positively charged (Arg189,
Lys190) residues (Figure 1b). The −PO4– groups
of the dsRNA major groove contact polar (Thr208) and positively charged
(Lys210, Lys211, Lys214, Arg215) residues of helix α2, which lies across the width of the major groove.In Figure 1c, we compare the sequences of two RBDs in TRBP.
Interestingly, not all the residues, which contribute to binding of
TRBP-RBD2 to dsRNA, are conserved: Glu166, His188, Lys210, Lys211,
and Lys214 are conserved, whereas Gln165, Arg189, Lys190, and Arg215
are not conserved.
TRBP-RBD2 Binding to Noncognate Substrates
Next, we examine the binding of TRBP-RBD2 to nucleic acid duplexes
containing DNA. Snapshots of TRBP-RBD2 in complex with DNA–RNA
and dsDNA, captured at the end of 100 ns long simulations, are shown
in Figure 3a,b, respectively. For comparison,
the TRBP-RBD2:dsRNA complex is overlaid as a shadow. Even though the
simulations were started with DNA–RNA and dsDNA duplexes in
A-forms, these duplexes changed to B-forms within several nanoseconds
of the simulations; in MD simulations with the selected force field,[29] free dsDNA duplexes assume B-forms[38] and free dsRNA duplexes assume A-forms,[30] in agreement with experimental data. Quick (≈1
ns) A → B form changes of DNA strands in dsDNA and DNA:RNA
duplexes were observed and characterized in previous short scale simulations.[39] After 100 ns of equilibration, DNA–RNA
and dsDNA duplexes are significantly longer than dsRNA, despite all
duplexes having the same number of base pairs. During simulations,
DNA–RNA and dsDNA relaxed to forms with significantly wider
major grooves and narrower minor grooves, thus departing from their
initial A-forms. Analyses of selected helical parameters (Table S1, Supporting Information), including twist, slide,
and roll, show that DNA–RNA and dsDNA duplexes in complex with
TRBP-RBD2 are distinctly non-A-form and are similar in form to free
duplexes.[39,40] In addition, Table S1 (Supporting Information) shows that DNA strands of DNA–RNA
and dsDNA duplexes acquire dihedral angles (δ, ϵ) and
phase angles of pseudorotation (P) characteristic
of DNA strands in B-form, comparable to values reported in ref (40).
Figure 3
Binding of TRBP-RBD2
to noncognate duplexes: (a) DNA–RNA hybrid and (b) dsDNA. RNA
strands are shown in gold and DNA strands in orange. For comparison,
the TRBP-RBD2:dsRNA complex is shown in transparent ribbon representation
with protein and RNA backbones aligned to the backbones of the complexes
shown in color. Curved arrows mark displacements of α1 helices of TRBP-RBD2 in complexes with DNA–RNA and dsDNA,
as compared to complexes with dsRNA. Top views of binding interfaces
are shown on the right, where TRBP-RBD2 residues within 3 Å of
duplexes are highlighted in both transparent surface and licorice
representations. The residues are colored according to residue type,
as described in Figure 1b.
Binding of TRBP-RBD2
to noncognate duplexes: (a) DNA–RNA hybrid and (b) dsDNA. RNA
strands are shown in gold and DNA strands in orange. For comparison,
the TRBP-RBD2:dsRNA complex is shown in transparent ribbon representation
with protein and RNA backbones aligned to the backbones of the complexes
shown in color. Curved arrows mark displacements of α1 helices of TRBP-RBD2 in complexes with DNA–RNA and dsDNA,
as compared to complexes with dsRNA. Top views of binding interfaces
are shown on the right, where TRBP-RBD2 residues within 3 Å of
duplexes are highlighted in both transparent surface and licorice
representations. The residues are colored according to residue type,
as described in Figure 1b.Results in Figures 1b and 3 show that DNA–RNA and dsDNA duplexes stay
in contact with a smaller number of TRBP-RBD2 residues than does dsRNA.
Binding of TRBP-RBD2 to minor groove I is significant only for the
dsRNA duplex, where helix α1 lies across the minor
groove I and remains in close contact with it. Minor grooves I of
DNA–RNA and dsDNA are more exposed to solvent, because helix
α1 is lifted from the groove. Binding of TRBP-RBD2
to major grooves is also more prominent for dsRNA. The numbers of
TRBP-RBD2 contacts to the minor groove II are very similar for all
three studied duplexes.Parts a and c of Figure 4 show contact areas and interaction energies of TRBP-RBD2
and three duplexes, evaluated during the last 85 ns of the simulations.
The contact area decreases (Figure 4a) and
binding energy becomes less favorable (Figure 4c) as the DNA content of the duplex increases. The number of hydrogen
bonds between TRBP-RBD2 and three duplexes also decreases with the
increase of DNA content (Figure S3, Supporting
Information), confirming that TRBP-RBD2 binds most favorably
to its cognate substrate, dsRNA, in agreement with our experiments.
Figure 4
Interaction
of TRBP-RBD2 with three duplex types. (a) Contact areas between TRBP-RBD2
and dsRNA, DNA–RNA, and dsDNA. (b) Contact areas between TRBP-RBD2
and individual nucleic acid grooves (minor grooves I and II, major
groove, marked in Figure 1a). Contact areas
were measured between whole duplexes and sets of the TRBP-RBD2 residues
above the selected grooves, H159, E160, V161, G162, A163, Q165, E166,
V168, V169, Q170, R174, L175, Y178 (for minor groove I); F192, T208,
S209, K210, K211, L212, K214, R215 (major groove); E183, P186, A187,
H188, R189, K190, E191 (minor groove II). (c) Interaction (nonbonding)
energy between TRBP-RBD2 and the three duplexes. The plots are shown
for the last 85 ns of trajectories. In (a) and (c) thin lines correspond
to data sampled from trajectories and thick lines show gliding time
averages over 800 ps.
Interaction
of TRBP-RBD2 with three duplex types. (a) Contact areas between TRBP-RBD2
and dsRNA, DNA–RNA, and dsDNA. (b) Contact areas between TRBP-RBD2
and individual nucleic acid grooves (minor grooves I and II, major
groove, marked in Figure 1a). Contact areas
were measured between whole duplexes and sets of the TRBP-RBD2 residues
above the selected grooves, H159, E160, V161, G162, A163, Q165, E166,
V168, V169, Q170, R174, L175, Y178 (for minor groove I); F192, T208,
S209, K210, K211, L212, K214, R215 (major groove); E183, P186, A187,
H188, R189, K190, E191 (minor groove II). (c) Interaction (nonbonding)
energy between TRBP-RBD2 and the three duplexes. The plots are shown
for the last 85 ns of trajectories. In (a) and (c) thin lines correspond
to data sampled from trajectories and thick lines show gliding time
averages over 800 ps.The individual contributions to the net contact area are
shown for each (minor–major–minor) groove in Figure 4b. These contact areas were evaluated between whole
duplexes and selected groups of TRBP-RBD2 amino acid residues directly
above each of the grooves. The results confirm that TRBP-RBD2 has
much less contact with the minor groove I in case of DNA-containing
duplexes than in case of the dsRNA duplex; the contact area between
TRBP-RBD2 and DNA–RNA or dsDNA minor grooves I is only ≈40–50%
of the contact area between TRBP-RBD2 and dsRNA minor groove I. The
contact area between TRBP-RBD2 and the duplex major groove also decreases
with the increased DNA content of the duplex. The results indicate
that TRBP-RBD2 recognizes dsRNA on the basis of interactions in minor
groove I and in the major groove.
TRBP-RBD2 Recognizes dsRNA-
over DNA-Containing Substrates
Recognition at the Major
Groove
The binding mode of TRBP-RBD2 to the major groove
of dsRNA is shown in Figure 5a. Five amino
acid residues, Thr208, Lys210, Lys211, Lys214, and Arg215, make direct
contact with the dsRNA backbone at the major groove. Thr208, Lys210,
Lys214, and Arg215 bind to the parts of the dsRNA major groove directly
beneath them. On the other hand, Lys211 lies across the major groove.
Its backbone atoms are on one side of the groove, whereas its positively
charged tip contacts the −PO4– group on the opposite side of the
major groove.
Figure 5
Major groove interactions. (a) Binding of TRBP-RBD2 to
the major groove of dsRNA. The residues that interact through Coulomb
interactions (Lys and Arg) and hydrogen bonding (Thr) are highlighted,
together with their bonding partners, negatively charged phosphate
groups. Oxygen, hydrogen, and phosphorus atoms are shown as red, white,
and tan spheres, respectively, carbon and nitrogen are shown in cyan
and blue. (b) Average distances of TRBP-RBD2 residues from the phosphate
groups nearest to them (phosphorus atoms for Lys and Arg, and oxygen
atom for Thr208), averaged over the last 80 ns of simulations. Red
arrows mark the residues that bind to dsRNA more tightly than to DNA–RNA
or dsDNA. Register fit of the Lys210–Lys214 pair in major grooves
of dsRNA (c), DNA–RNA (d), and dsDNA (e). Lys210–Lys214
distances and the distances between two phosphate groups that coordinate
Lys210 and Lys214 are shown. These distances are calculated between
phosphate atoms of RNA nucleotides 17 and 48, and between nitrogen
atoms of Lys210 and Lys214 side chains. The distance between P17 and P48 characterizes the major groove width
of the duplexes. In (c)–(e) thin lines correspond to data sampled
from trajectories and thick lines show gliding time averages over
800 ps.
Major groove interactions. (a) Binding of TRBP-RBD2 to
the major groove of dsRNA. The residues that interact through Coulomb
interactions (Lys and Arg) and hydrogen bonding (Thr) are highlighted,
together with their bonding partners, negatively charged phosphate
groups. Oxygen, hydrogen, and phosphorus atoms are shown as red, white,
and tan spheres, respectively, carbon and nitrogen are shown in cyan
and blue. (b) Average distances of TRBP-RBD2 residues from the phosphate
groups nearest to them (phosphorus atoms for Lys and Arg, and oxygen
atom for Thr208), averaged over the last 80 ns of simulations. Red
arrows mark the residues that bind to dsRNA more tightly than to DNA–RNA
or dsDNA. Register fit of the Lys210–Lys214 pair in major grooves
of dsRNA (c), DNA–RNA (d), and dsDNA (e). Lys210–Lys214
distances and the distances between two phosphate groups that coordinate
Lys210 and Lys214 are shown. These distances are calculated between
phosphate atoms of RNA nucleotides 17 and 48, and between nitrogen
atoms of Lys210 and Lys214 side chains. The distance between P17 and P48 characterizes the major groove width
of the duplexes. In (c)–(e) thin lines correspond to data sampled
from trajectories and thick lines show gliding time averages over
800 ps.The stability of binding between
the five identified residues to major grooves of dsRNA, DNA–RNA,
and dsDNA is examined by tracking distances of these residues to their
nearest neighbor −PO4– groups. In Figure 5b, we show these distances averaged over the last 85 ns of
trajectories (time evolution of these distances is shown in Figure
S4, Supporting Information). The results
show that Thr208 makes a hydrogen bond to the nearest −PO4– group
in dsRNA and DNA–RNA duplexes but does not engage in hydrogen
bonding in case of the dsDNA duplex. The distances between Lys210
and Arg215 to the nearest −PO4– groups are similar for all
three duplexes, but the distances between Lys211 and Lys214 to the
nearest −PO4– groups are perturbed when the duplex contains DNA,
involving a gradual increase in the average distance of Lys211 to
the nearest −PO4– group with the increase of DNA content
in the duplex. Likewise, the distance between Lys214 to the nearest
−PO4– group becomes longer and less stable for DNA-containing duplexes.Although the results in Figure 5b show that
Lys211 and Lys214 enable TRBP-RBD2 to recognize and bind to dsRNA,
experiments have shown that single mutations of His188, Lys210, and
Lys214 block TRBP-RBD2 binding to dsRNA.[13] A closer look at the binding mode of TRBP-RBD2 and the dsRNA major
groove indicates that Lys210 and Lys214 form a pair that bridges −PO4– groups
on two sides of the dsRNA major groove. In Figure 5c–e, we analyze the distances between Lys210 and Lys214,
and between −PO4– groups that coordinate them (the
distances are measured between the nitrogen atoms of Lys210 and Lys214
side chains, and between the phosphate atoms of RNA nucleotides 17
and 48, respectively). The results show evidence for a “register”
fit of Lys210–Lys214, as the lysine pair fits exactly above
the −PO4– groups that coordinate them. The distance between
positively charged tips of Lys210 and Lys214 residues is always in
the range ≈9–13 Å, whereas the distance between
the bridging −PO4– groups increases from ≈11
Å for dsRNA to ≈17 Å for DNA–RNA and to ≈20
Å for dsDNA. Therefore, the Lys210-Lys214 pair fits poorly in
the major grooves of DNA–RNA and dsDNA duplexes because the
major grooves of these duplexes are significantly wider, as shown
in Figure S5 (Supporting Information).
Similarly, Lys211 cannot effectively bridge the 17–20 Å
distance across the major groove of DNA–RNA and dsDNA.
Recognition
at Minor Grooves
dsRBDs bind to two minor grooves of dsRNA,[24,34−37] in the binding mode shown in Figure 1a. Minor
grooves of dsRNA are lined with 2′-hydroxyl (OH) groups, located
on the sugar rings of RNA bases. However, DNA bases lack 2′-hydroxyl
groups. To explore the effect of the 2′-OH groups on dsRBD
binding to three duplexes, we analyze the dsRBD binding to minor grooves
I and II.Snapshots of TRBP-RBD2 bound to minor grooves I of
dsRNA and dsDNA are shown in Figure 6a,b. When
TRBP-RBD2 binds to dsRNA, its α1-helix extends across
the groove. His159 and Gln165 residues of the α1-helix
make three direct hydrogen bonds to dsRNA, and Glu166 makes one water-mediated
bond. The three direct hydrogen bonds remain stable for most of the
time during the last 85 ns of the trajectory, whereas the water-mediated
bond exhibits significant fluctuation (Figure 6c). On the other hand, TRBP-RBD2 does not bind across the minor groove
I of dsDNA. Figure 6b,d shows that for this
complex only Gln165 makes a hydrogen bond to the phosphate group of
the nearest DNA nucleotide, albeit only a fluctuating one. Binding
of TRBP-RBD2 to DNA–RNA is very similar to its binding to dsDNA
(not shown).
Figure 6
Interactions of TRBP-RBD2 residues with minor grooves
I of (a) dsRNA and (b) dsDNA. (c) Time dependent distances of stable
contacts between TRBP-RBD2 and dsRNA minor groove I. (d) Same as in
(c) for TRBP-RBD2 and dsDNA minor groove I. In (d) thin lines correspond
to data sampled from trajectories and thick lines show gliding time
averages over 800 ps.
Interactions of TRBP-RBD2 residues with minor grooves
I of (a) dsRNA and (b) dsDNA. (c) Time dependent distances of stable
contacts between TRBP-RBD2 and dsRNA minor groove I. (d) Same as in
(c) for TRBP-RBD2 and dsDNA minor groove I. In (d) thin lines correspond
to data sampled from trajectories and thick lines show gliding time
averages over 800 ps.Figure 6a,b demonstrates that minor
grooves I of dsRNA and dsDNA have different forms: the minor groove
I of dsRNA is wide and shallow (A-form like), with its bases being
in contact with TRBP-RBD2 helix α1, whereas minor
groove I of dsDNA is narrower and much deeper (B-form like), with
its bases being buried below the α1 helix. The results
in Figure 6 show clearly that the A-form of
the minor groove I facilitates TRBP-RBD2 binding.TRBP-RBD2
exhibits a binding (contact) area to minor groove II that is similar
for all three duplexes (Figure 4b). However,
the binding modes in this groove differ between the three duplexes.
For TRBP-RBD2 bound to dsRNA minor groove II, the His188hydrogen
binds to the 2′-OH group, whereas Arg189 and Lys190 occasionally
interact with the 2′-OH groups or the backbone phosphate groups.
For TRBP-RBD2 bound to DNA–RNA minor groove II, His188 is unbound,
whereas Arg189 and Lys190 form stable hydrogen bonds to two bases.
Lastly, for TRBP-RBD2 bound to dsDNA, His188 makes a hydrogen bond
to one of the bases, whereas Arg189 and Lys190 bind Coulombically
to the negatively charged backbone phosphate groups. Because Arg189
and Lys190 residues are not conserved, dsRBDs without these residues
would exhibit weaker binding to the minor groove II of DNA-containing
duplexes.
Duplex Form at the TRBP-RBD2 Binding Site
Results in Figures 5 and 7 already show that the duplex form with its minor and major
grooves plays a significant role in optimal TRBP-RBD2 binding. Here,
we examine the complete duplex forms at the sites of binding to TRBP-RBD2.
To compare in this regard the dsRNA, DNA–RNA, and dsDNA duplex
forms at TRBP-RBD2 binding sites, single average structures of the
respective complexes were characterized by considering structures
averaged over the last 40 ns of trajectories and then aligning the
TRBP-RBD2 domains of these structures.
Figure 7
Duplex forms at TRBP-RBD2 binding sites. The
structures of complexes analyzed in (a) and (b) are obtained by averaging
the coordinates of each system over the last 40 ns of simulation.
(a) Duplex backbones at TRBP-RBD2 binding sites. dsRNA strands are
shown in gold, dsDNA strands are shown in orange, and strands of the
DNA–RNA hybrid are shown as black (DNA) and white (RNA) tubes.
TRBP-RBD2 domains, bound to the shown duplexes, are aligned. The shaded
area outlines the TRBP-RBD2 residues in contact with the dsRNA duplex
(within 3.0 Å of it). The minor–major–minor groove
pattern is labeled and highlighted by the black dashed lines. Arrows
point to major differences in duplex forms in binding regions. (b)
Helical rise for duplex base pairs of A-form dsRNA, B-form dsDNA,
and averaged structures of dsRNA, DNA–RNA, and dsDNA bound
to TRBP-RBD2. The shaded region of the graph marks the base pairs
that form the binding site for TRBP-RBD2. A- and B-form structures
are generated with 3-DNA software.[25]
Duplex forms at TRBP-RBD2 binding sites. The
structures of complexes analyzed in (a) and (b) are obtained by averaging
the coordinates of each system over the last 40 ns of simulation.
(a) Duplex backbones at TRBP-RBD2 binding sites. dsRNA strands are
shown in gold, dsDNA strands are shown in orange, and strands of the
DNA–RNA hybrid are shown as black (DNA) and white (RNA) tubes.
TRBP-RBD2 domains, bound to the shown duplexes, are aligned. The shaded
area outlines the TRBP-RBD2 residues in contact with the dsRNA duplex
(within 3.0 Å of it). The minor–major–minor groove
pattern is labeled and highlighted by the black dashed lines. Arrows
point to major differences in duplex forms in binding regions. (b)
Helical rise for duplex base pairs of A-form dsRNA, B-form dsDNA,
and averaged structures of dsRNA, DNA–RNA, and dsDNA bound
to TRBP-RBD2. The shaded region of the graph marks the base pairs
that form the binding site for TRBP-RBD2. A- and B-form structures
are generated with 3-DNA software.[25]The resulting duplexes at the
binding site are shown in Figure 7a. Several
structural features can be clearly discerned. First, the width of
the major groove is narrowest for dsRNA, wider for DNA–RNA,
and widest for dsDNA. The increase in the major groove width with
the increased DNA content is the reason for the gradual decrease of
TRBP-RBD2 binding to the major groove of the three duplexes (Figure 4b). Second, minor grooves I of DNA–RNA and
dsDNA can be almost perfectly superimposed on each other, but not
to the more compressed minor groove I of dsRNA; only dsRNA achieves
stable binding to TRBP-RBD2. Third, the minor groove II of DNA–RNA
assumes a form almost identical to that of dsRNA, whereas dsDNA exhibits
a much narrower minor groove II.When DNA–RNA is simulated
without bound TRBP-RBD2, both of its minor grooves resemble closely
the dsDNA minor grooves (Figure S6b, Supporting
Information). Therefore, through the presence of TRBP-RBD2,
the minor grooves of DNA–RNA adopt either dsRNA or dsDNA forms,
depending on the interactions with the protein (Figure S5b, Supporting Information). Similar duplex flexibility
and duplex adaptation to bound proteins has been observed for dsDNA
duplexes.[41] Interestingly, it is the DNA
strand of the DNA–RNA minor groove II that changes its conformation
and is almost perfectly overlaid with the RNA strand of the dsRNA,
as shown in Figure 7a.In Figure 7b, we compare the helical rise of the duplexes simulated
with bound TRBP-RBD2. The helical rise is the distance between neighboring
bases in the duplex helix. For perfect A-form dsRNA it is ≈2.5
Å, whereas for B-form dsDNA it is 3.4 Å. For duplexes bound
to TRBP-RBD2, the helical rise is evaluated for the averaged structures.
The results show for dsRNA a perfect A-form, except in one region
of widened major groove (experimentally confirmed for dsRBDs[34]). For DNA–RNA, the helical rise is between
A- and B-forms, as observed in other studies of free DNA–RNA
duplexes;[39,40] for dsDNA the helical rise is again between
A- and B-form, but very close to the B-form value. dsRNA has the smallest
helical rise and remains closest to the A-form.In Table S1
(Supporting Information), we further quantify
forms of duplexes bound to TRBP-RBD2 and list their selected average
helical parameters (twist, slide, roll), as well as dihedral angles
(δ, ϵ) and phase angles of pseudorotation (P). Comparison of Table S1 (Supporting Information) to parameters of free duplexes[40] shows
that free and dsRBD-bound duplexes have similar forms. Overall, DNA
strands in dsDNA and DNA–RNA duplexes have parameters characteristic
of the B-form, whereas RNA strands in dsRNA and DNA–RNA have
parameters characteristic of the A-form. However, dihedral angles
δ and phase angles P of DNA nucleotides in
TRBP-RBD2:DNA–RNA complex, shown in Figure S8 (Supporting Information), show that DNA nucleotides
can switch to the A-form when bound to TRBP-RBD2.
Discussion
In the present article, we have clarified
by means of MD simulations the experimental observation that dsRBDs
recognize and bind to dsRNA and avoid binding to similar DNA–RNA
and dsDNA duplexes. Previous experiments showed that isolated first
and second dsRBDs of TRBP bind stably to dsRNA, with binding affinities
of 220 and 110 nM, respectively.[42] Our
experiments, shown in Figure 2, confirmed that
TRBP-RBD2 (1–10 μM) binds to dsRNA. In agreement with
experiments, MD simulations showed that TRBP dsRBDs bind stably to
dsRNA, with the second dsRBD having a more favorable interaction energy
with dsRNA than the first dsRBD (Figure S9, Supporting
Information).Although there are examples of domains
and proteins that can bind to both RNA and DNA,[43−46] TRBP and other dsRBD-containing
proteins do not bind well to DNA–RNA and dsDNA duplexes.[19−21] For example, dsRBDs of PKR bind to dsRNA and DNA–RNA with Kd = 0.17 μM and ≥500 μM,
respectively, whereas no binding to dsDNA was observed in the tested
experimental conditions.[21] The whole TRBP
protein (10 nM to 2 μM concentration) binds to dsRNA, but not
to DNA–RNA and dsDNA.[19,20] However, some dsRBD
proteins, such as 4F protein that contains two dsRBDs, can bind strongly
to both dsRNA and DNA–RNA duplexes; DNA–RNA duplexes
at 50 pM concentration were able to compete with dsRNA duplexes for
binding to 4F.[22] Our experimental results
in Figure 2 show that at the concentration
of 5–10 μM, TRBP-RBD2 binds to 20–90% of the present
DNA–RNA and does not bind to dsDNA. Therefore, TRBP-RBD2 can
discriminate between dsRNA and DNA–RNA more effectively than
4F and less effectively than PKR. In agreement with the experimental
results, complexes of TRBP-RBD2 with DNA–RNA and dsDNA were
less stable in the performed MD simulations than complexes with dsRNA,
as shown in Figures 1, 3, and 4.MD simulations reveal that
the duplex form plays an important role in the recognition of dsRNA-
over DNA-containing duplexes by dsRBDs, as has already been hypothesized
on the basis of dsRBD:dsRNA complex structures.[5] A proper binding of a dsRBD to a nucleic acid duplex is
determined by a proper “register fit” of the dsRBD into
three successive duplex grooves, which is in turn largely determined
by the duplex form. Though dsRNA duplex has the A-form, DNA-containing
duplexes quickly adopt an intermediate form (DNA–RNA) and the
B-form (dsDNA), as seen in Figures 7 and S6
(Supporting Information). By adopting B-like
forms, DNA-containing duplexes have wider major grooves and narrower
minor grooves than dsRNA. For example, the major grooves of DNA-containing
duplexes become 6–9 Å wider than the major groove of dsRNA
(Figure 5c–e).As a result of
changed forms, binding of TRBP-RBD2 to DNA–RNA and dsDNA becomes
significantly impaired in major grooves and in minor grooves I, but
not in the minor groove II, as quantified in Figure 4b. Delayed changes of the dsDNA major groove width at the
binding site, seen in Figure 5e, show the importance
of the duplex form for dsRBD binding. Preserved contacts between the
dsDNA major groove and three basic residues, Lys 210, Lys 211, and
Lys 214, keep dsDNA from adopting its preferred major groove width
during the first 50 ns of simulations. In comparison, dsDNA without
the bound dsRBD adopts very quickly wide major grooves of the B-form.
Similarly, TRBP-RBD2 easily contacts the bases in the wide and shallow
minor groove I of dsRNA but cannot contact the bases in the deeper
and narrower minor groove I of dsDNA (Figure 6). In principle, dsRBDs could bind dsDNA if dsDNA was flexible enough
to adopt the A-form. However, although B-form DNA was shown to be
highly flexible, its distortions do not necessarily change its global
helical conformation.[47] Otherwise, dsRBDs
could bind DNA-containing duplexes if dsRBDs were flexible and able
to structurally conform to these duplexes. However, the simulations
show that dsRBDs resist structural changes in the presence of dsDNA
and DNA–RNA and validate previous hypotheses and observations
on dsRBD rigidity.[5]The contribution
of 2′-OH groups to binding of dsRBDs to dsRNA was previously
examined for PKR, a protein with two dsRBDs, binding to chimeric dsRNA
duplexes with several 2′-OH groups substituted to 2′-H
or 2′-OCH3.[21] Several
results indicated that 2′-OH groups play a more important role
than the duplex form in PKR binding to dsRNA. For example, only one
ionic contact was found between PKR and dsRNA, and the mutagenesis
studies showed that one Lys residue is required for PKR:dsRNA binding
(Lys 60 in PKR-RBD1, analogous to Lys 210 in TRBP-RBD2).[48] However, more recently, crystal structures of
dsRBD:dsRNA complexes identified multiple ionic interactions between
dsRBDs and dsRNA phosphate groups.[24,34] Mutagenesis
studies of TRBP showed that three residues of TRBP-RBD2, His 188,
Lys 210, and Lys 214, are required for TRBP-RBD2 binding to dsRNA.[13] Smaller involvement of ionic interactions between
PKR and dsRNA could be due to PKR lacking several basic residues that
contribute to strong binding of TRBP-RBD2 to dsRNA and its major groove.
PKR-RBD1 lacks the residues analogous to Arg 190 and Arg 215 of TRBP-RBD2,
whereas PKR-RBD2 lacks the residues analogous to Arg 190, Lys 191,
Lys 211, and Arg 215 of TRBP-RBD2.[5] The
observation that the Lys 210 and Lys 214 pair of TRBP-RBD2 has to
fit well into the major groove of the duplex for the dsRBD to bind
stably to the duplex, as shown in Figure 5 (c-e),
highlights the importance of shape on binding of dsRBDs to duplexes.Although dsRBDs are known to not bind to DNA–RNA and dsDNA,[19−21] experimental results in Figures 2 and S2
(Supporting Information) show that TRBP-RBD2
at concentration >5 μM has a higher binding affinity for
DNA–RNA than for dsDNA. The higher affinity of TRBP-RBD2 for
DNA–RNA is likely due to the narrower major groove of DNA–RNA
(Figure 5d) and the fact that TRBP-RBD2 can
alter the width of DNA–RNA grooves and strengthen the binding.
The latter was observed in the minor groove II of the DNA–RNA
duplex, which acquired the width of an A-form duplex, due to DNA nucleotides
changing from B- to A-form conformations (Figure S8, Supporting Information) upon binding to TRBP-RBD2 (Figure
S7b, Supporting Information). The observed
B- → A-form change of DNA nucleotides and the resulting minor
groove width modulation potentially explain how other dsRBD proteins,
such as 4F,[22] can also bind to DNA–RNA
hybrids.In the present article we have shown how dsRNA binding
proteins distinguish between dsRNA- and DNA-containing duplexes. Additional
studies are needed to explain the observed sequence specific binding
of other dsRNA binding proteins,[5,36,49] and to understand dsRBD binding to dsRNA with mismatches, small
bulges and loops, common in microRNA (miRNA) in the silencing by RNA
interference pathways.[15,16]
Authors: Atul Rangadurai; Huiqing Zhou; Dawn K Merriman; Nathalie Meiser; Bei Liu; Honglue Shi; Eric S Szymanski; Hashim M Al-Hashimi Journal: Nucleic Acids Res Date: 2018-11-16 Impact factor: 16.971
Authors: Fabricio Barbosa Monteiro Arraes; Diogo Martins-de-Sa; Daniel D Noriega Vasquez; Bruno Paes Melo; Muhammad Faheem; Leonardo Lima Pepino de Macedo; Carolina Vianna Morgante; Joao Alexandre R G Barbosa; Roberto Coiti Togawa; Valdeir Junio Vaz Moreira; Etienne G J Danchin; Maria Fatima Grossi-de-Sa Journal: RNA Biol Date: 2020-12-31 Impact factor: 4.652
Authors: Caitlin A O'Brien; Jody Hobson-Peters; Alice Wei Yee Yam; Agathe M G Colmant; Breeanna J McLean; Natalie A Prow; Daniel Watterson; Sonja Hall-Mendelin; David Warrilow; Mah-Lee Ng; Alexander A Khromykh; Roy A Hall Journal: PLoS Negl Trop Dis Date: 2015-03-23