A packing-density metric for exploring the interior of folded RNA molecules

Article abstract of DOI:10.1002/anie.200353575

The intricate architectures of RNA molecules often contain expansive close-packed interfaces. The folding of the well-defined P4-P6 RNA domain was challenged by using a series of 2′-modified nucleotides with a narrow range of molecular volumes. The resulting folding interferences (see picture) reflect the spatial environment of the 2′-hydroxy groups and thereby define a packing-density metric.

Full text of DOI:10.1002/anie.200353575

Angewandte
Chemie
complex architectures often contain coaxially stacked helices
that adopt specific spatial arrangements through tertiary
interactions between distinctly structured, complementary
surfaces. These interactions result in the formation of
expansive close-packed interfaces.[1] Within these interfaces,
atoms acquire distinct packing environments that together
uniquely specify an RNA fold. To define these packing
environments biochemically, we exploited a series of nucleo-
tide analogues with 2’-substituents that span a range of
molecular volumes (Figure 1). We used nucleotide analogue
Figure 1. A series of nucleotide analogues that span a range of molec-
ular volumes. A) Nucleotide-a-thiotriphosphates for NAIM analysis.
B) Space-filling representations of the uridine analogues. The molecu-
lar volume of the 2’-substituent ranges from 12 to 37 Š3. Molecular vol-
umes were calculated by using the approach described by Connolly[4]
and are consistent with volumes calculated from crystallographic
data.[5]
RNA Folding
A Packing-Density Metric for Exploring the
Interior of Folded RNA Molecules**
Jason P. Schwans, Nan-Sheng Li, and
Joseph A. Piccirilli*
interference mapping (NAIM)[2] to examine the effect of
these volume changes at every residue within the structurally
well-defined, independently folding P4-P6 RNA domain.[3]
Our results show that these analogues together provide a
sensitive metric for the packing density of the atoms
surrounding the 2’-hydroxy groups in an RNA molecule.
We synthesized 2’-chloro, 2’-methyl, and 2’-sulfanylnu-
cleoside analogues of adenosine, cytidine, guanosine, and
uridine and converted these analogues into the corresponding
a-thiotriphosphates. We have reported the synthesis of 2’-
sulfanylnucleotide analogues previously.[6] The 2’-deoxy-2’-
methylnucleoside analogues were synthesized by glycosyla-
tion of methyl 2-a-acetoxymethyl-3,5-di-O-(4-chlorobenzyl)-
2-deoxy-a-d-ribofuranoside, followed by deoxygenation and
deprotection.[7] The procedure described by Verheyden and
Moffatt served as a guide for the synthesis of 2’-chloro-2’-
deoxyuridine.[8] Displacement of the corresponding 2’-b-O-
triflate esters with lithium chloride gave 2’-chloro-2’-deoxy-
cytidine, adenosine, and guanosine.[9] We prepared the a-
thiotriphosphate derivatives of the chloro- and methylnucleo-
sides as described previously for the sulfanylnucleosides.[6]
RNA molecules fold into intricate three-dimensional archi-
tectures that perform important biological functions. These
[*] J. P. Schwans, N.-S. Li, Prof. J. A. Piccirilli
Howard Hughes Medical Institute
The University of Chicago
Department of Biochemistry and Molecular Biology
and Department of Chemistry
5841S. Maryland Avenue, MC1028, Chicago, Il 60637 (USA)
Fax: (+1)773-702-0271
E-mail: jpicciri@midway.uchicago.edu
[**] We thank Rui Sousa for the gift of the Y639F T7 RNA polymerase
plasmid, Kara Juneau and Tom Cech for the DC209 P4-P6 plasmid,
Cecilia Cortez and Shirshendu Deb for assistance with nucleoside
synthesis, Joe Olvera for technical assistance, and Cheryl Small for
assistance with manuscript preparation. We also thank members of
the Piccirilli group and the reviewers for their comments on the
manuscript. N.S.L. is a Research Specialist, and J.A.P. is an
Associate Investigator of the Howard Hughes Medical Institute.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2004, 43, 3033 –3037
DOI: 10.1002/anie.200353575
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3033

Communications
To establish a metric for RNA packing density, we
used interference mapping[2] to determine the influence
of these analogues on the folding of DC209 P4-P6, an
independently folding RNA domain derived from the
Tetrahymena group I intron.[3] This domain serves as an
excellent model system with which to explore the
physical basis of analogue interference. Crystallo-
graphic analyses have revealed the domain structur-
e,[3b,c] which consists of two sets of coaxially stacked
helices that pack against each other to form a compact
tertiary architecture. Biochemical analyses have
allowed the thermodynamics and kinetics of folding
of this domain to be defined.[10] Divalent magnesium
ions induce a two-state tertiary folding transition that
renders the folded molecules more compact than the
unfolded molecules.[3a] Nondenaturing gel electropho-
resis resolves these states and allows isolation (selec-
tion) of folded species from a population.[6,11]
We incorporated each nucleotide a-thiotriphos-
phate into DC209 P4-P6 at random by in vitro tran-
scription with Y639F T7 RNA polymerase.[12] As
described previously for the 2’-sulfanylnucleotides,[6]
the polymerase efficiently and evenly incorporates
the 2’-chloro- and 2’-methylnucleotides into the RNA
to generate populations with approximately 1–5 ana-
logue substitutions per transcript. Folded species were
selected from the population by nondenaturing gel
electrophoresis. Treatment with I2 induced cleavage of
the RNA backbone at sites containing phosphoro-
[13]
thioate linkages
and thereby revealed the locations
of the nucleotide analogues. Gaps in the I2 cleavage
ladder of the folded molecules compared to those of
the unselected population indicate residues at which an
analogue induces interference.[14] To achieve maximum
sensitivity to folding interference without compromis-
ing the signal, we conducted our experiments at
0.45 mm Mg2+, the concentration corresponding to the
midpoint of the DC209 P4-P6 folding transition.[3c,6]
Under these conditions, the selection assay detects
interferences arising from as little as 0.2 kcalmolÀ1
destabilization.[6]
Figure 2. Interference histograms mapped onto the DC209 P4-P6 secondary
structure. A) Deoxynucleotide interference distinguishes residues whose hy-
droxy groups engage in tertiary interactions from those that bear nonfunctional
(dispensable) hydroxy groups. B) Superimposed interference maps for 2’-Cl, 2’-
CH3, and 2’-SH nucleotides. For clarity, the map shows only interferences at
residues bearing nonfunctional hydroxy groups (no 2’-H interference). We con-
ducted the folding selection at 0.45 mm Mg2+, which corresponds to the mid-
point of the DC209 P4-P6 folding transition,[3c,6] and calculated interference
values as described by Ryder et al.[2e] For residues at which more than one ana-
logue interferes, the histogram shows the maximum interference value
observed. Residues that bear functional hydroxy groups generally exhibit the 2’-
We observed little interference in simple duplex
regions of DC209 P4-P6, which indicates that the minor
groove of A-form helical RNA readily accommodates
our atomic modifications. The interferences occur Cl/2’-CH3/2’-SH interference profile. Sites at which one of the analogues inter-
feres always also exhibit interference when an analogue of larger volume is
present, such that three interference profiles emerge (green, brown, and blue
bars). The data are summarized in Table 1.
primarily in single-stranded regions and span the
entire domain. When mapped onto the tertiary struc-
ture, the interferences cluster tightly in the three most
densely packed regions of the molecule: the hinge
region, the A-rich bulge/trihelical junction, and the
tetraloop/receptor.[6] The residues in these regions frequently
engage in hydrogen-bonding interactions through their 2’-
hydroxy groups, experience significant surface area burial,
and/or exhibit strong deviations from A-form helical geo-
metry.[3b,c]
To explore the physical basis of these interferences, we
obtained a 2’-deoxynucleotide (2’-H) interference map (Fig-
ure 2a) for comparison. Deoxynucleotide interference func-
tionally distinguishes residues bearing dispensable (nonfunc-
tional) 2’-hydroxy groups from those bearing functional 2’-
hydroxy groups that provide a significant energetic contribu-
tion to folding. All deoxynucleotide interferences in DC209
P4-P6 occur at residues whose 2’-hydroxy groups engage in
hydrogen bonds, as inferred from the crystal structure.[3b,c] The
remaining residues in DC209 P4-P6 bear nonfunctional or
dispensable 2’-hydroxy groups whose removal has little or no
energetic effect on folding. Residues bearing functional
hydroxy groups exhibit 2’-SH/2’-CH3/2’-Cl interference, con-
3034
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2004, 43, 3033 –3037

Angewandte
Chemie
sistent with our expectation that the analogues have little or
no capacity to mediate tertiary interactions.[15] The residues
bearing nonfunctional hydroxy groups therefore provide the
basis set of microenvironments with which to define a metric
for packing density. Among these residues, the number of
interferences increases as the analogue volume increases; the
interferences induced by a given analogue comprise a subset
of the interferences induced by an analogue of larger volume
(Figure 2b, Table 1). These results suggest that residues
bearing nonfunctional hydroxy groups in DC209 P4-P6 have
a volume threshold above which RNA folding is destabilized.
Table 1: DC209 P4-P6 interference profiles at 0.45 mm MgCl2.
Residue
2’-Cl Int
2’-CH3 Int
2’-SH Int
A122
C154
G164
G181
C197
G200
C121
C127
U185
A225
U120
U155
X
X
X
X
X
X
–
–
–
–
–
–
X
X
X
X
X
X
X
X
X
X
–
X
X
X
X
X
X
X
X
X
X
X
X
–
We used the crystal structure of DC209 P4-P6[3c] to
determine the distances between the nonfunctional hydroxy
groups and their nearest neighbors (Figure 3). Qualitatively,
the interference profiles correlate with the number and
closeness of neighboring atoms surrounding the hydroxy
group. Residues for which all three analogues exhibit
interference have the most densely packed hydroxy groups
(Figure 3, left column); multiple close contacts usually occur
with at least one atom residing within 2.65 Š of the hydroxy
group. Residues that exhibit the 2’-SH/2’-CH3 interference
profile have less tightly packed hydroxy groups and usually
only a single close contact within 3.1Š (Figure 3, middle
column). Residues that exhibit only 2’-SH interference have
sparsely packed hydroxy groups whose closest atomic neigh-
bors are between 3.2 and 3.5 Š away (Figure 3, right column).
Residues that exhibit no interference have no close contacts.
These observations suggest that the interference mapping
volume thresholds reflect the spatial environment of the
hydroxy groups in the folded state of DC209 P4-P6.
To define the space surrounding the hydroxy groups
quantitatively, we introduced molecular volume changes at
the hydroxy group sites and calculated the resulting change in
van der Waals energy by using molecular mechanics.[17]
Figure 4 shows volume–energy dependencies for five repre-
sentative positions in DC209 P4-P6. As the molecular volume
increases, the calculated energy rises, which reflects the
increasing van der Waals repulsion caused by steric overlap.
Sites that exhibit 2’-Cl/2’-CH3/2’-SH interference display the
greatest sensitivity to molecular volume. Sites that exhibit
Figure 3. Contacts with neighboring atoms and interference profiles
for nonfunctional 2’-hydroxy groups in DC209 P4-P6. Residues that
exhibit 2’-SH interference have hydroxy groups whose closest atomic
neighbors reside within 3.5 Š. Residues that exhibit interference in the
presence of the intermediate-sized methyl group have hydroxy groups
with closer neighbors (<3.1Š). Residues that exhibit interference
caused by the smallest analogue (chlorine) have the most densely
packed hydroxy groups with one or more close neighbors less than
2.65 Š away. We used Swiss–PDB Viewer software[16] to determine
interatomic distances within the DC209 P4-P6 crystal structure.[3c] The
atomic distances for unit A in the asymmetric unit are shown. Values
for unit B are essentially the same as those for unit A, except that in
unit B the 2’-oxygen atom of A122 is 2.19 Š from the phosphate
oxygen atom of A123.
only 2’-SH interference have repulsive energy only after the
molecular volume exceeds 35 Š3. Sites that exhibit 2’-CH3/2’-
SH interference display an intermediate sensitivity and exert
significant repulsion only when the molecular volume exceeds
Angew. Chem. Int. Ed. 2004, 43, 3033 –3037
www.angewandte.org
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3035

Communications
tion by destabilizing the tertiary structure when located at
residues that populate the 2’-endo conformation in the folded
RNA.[2d,19,20] We observe no link between interference and
nucleoside analogues that preferentially populate the 2’-endo
conformation.[6,19] These sugar pucker preferences evidently
have little energetic influence on the DC209 P4-P6 tertiary
folding transition, which was monitored by using the electro-
phoretic mobility shift assay. We did not observe a link
between interference and the hydrophobic character of the
introduced substituent, though we cannot rule out the
possibility that hydrophobicity has some influence on tertiary
folding stability.[18] Overlapping molecular volumes exert the
most significant energetic force in biology, since the repulsive
energy between two atoms increases exponentially with the
inverse of the distance between the centers of the two
atoms.[21] Our observation that DC209 P4-P6 interferences
correlate with molecular volume is in agreement with this
relationship.
Figure 4. van der Waals repulsion determines the 2’-Cl/2’-CH3/2’-SH
In summary, we have synthesized 2’-chloro-, 2’-methyl-,
and 2’-sulfanylnucleoside-a-thiotriphosphates and have dem-
onstrated that T7 RNA polymerase efficiently incorporates
these compounds into RNA without sequence bias, which
renders interference mapping analysis possible. These nucleo-
side analogues span a narrow range of molecular volume
(DV= 1 0 Š3) and together provide a sensitive measure of the
spatial environment (the packing density) of a 2’-hydroxy
group within a structured RNA molecule. Many residues
within RNA bear nonfunctional hydroxy groups that reside
within the extensive packing interfaces formed by tertiary
folding. Our packing density metric (PDM) defines these
interfaces biochemically. PDM analysis may help to evaluate
the functional validity of crystallographic and biochemical
models of RNA structure and may reveal regions within RNA
that undergo structural changes whilst carrying out their
function.
interference profile. A) The dependence of van der Waals repulsion
energy on molecular volume. The interference profile for a given resi-
due matches its relative volume sensitivity. We used the Sybyl 6.9
molecular mechanics package[17] to calculate the change in van der
Waals energy upon introduction of variously sized atoms at the 2’-posi-
tion (O, Cl, S, Br). The calculation included all force potentials but
DC209 P4-P6 was held rigid so that DE represents solely changes in
van der Waals overlap. The smooth curves serve to guide the eye.
B) Representative close-packed hydroxy groups and their local environ-
ment in space-filling representation. The models contain sulfur atoms
(yellow) in place of the 2’-hydroxy groups.
30 Š3. These trends indicate that the interference profiles
reflect the volume–energy dependencies.
The multifunctional hydroxy group exerts its influence on
RNA structure and function in multiple ways, for example, by
mediating tertiary interactions through hydrogen bonding or
metal-ion coordination, or by serving as a scaffold for the
integral hydration network associated with RNA.[1] The 2’-
hydroxy group may also make indirect contributions to
structure and function as a consequence of its polar character
and capacity to fill space, or its ability to withdraw electrons
inductively or engender a ribonucleotide with a preference
for the 3’-endo sugar conformation.[18] These effects appa-
rently have no energetic significance for folding at residues
bearing dispensable 2’-hydroxy groups (that is, residues that
do not interfere upon 2’-deoxynucleotide substitution). Nev-
ertheless, the nucleotide analogues described herein can
“induce” DC209 P4-P6 folding interference when introduced
at these locations. The analogues may act at several levels,
which include, but are not necessarily limited to, 1) sugar
conformational preferences, 2) hydrophobicity, and 3) molec-
ular volume.
Received: December 18, 2003
Revised: March 15, 2004 [Z53575]
Keywords: hydrogen bonds · interfaces · nucleotides ·
.
RNA structures · RNA
[1] See, for example: a) S. A. Strobel, J. A. Doudna, Trends
Biochem. Sci. 1997, 22, 262 – 266; b) A. R. Ferre-D'Amare,
J. A. Doudna, Annu. Rev. Biophys. Biomol. Struct. 1999, 28, 57 –
73; c) T. Hermann, D. J. Patel, J. Mol. Biol. 1999, 294, 829 – 849;
d) R. T. Batey, R. P. Rambo, J. A. Doudna, Angew. Chem. 1999,
111, 2472 – 2491; Angew. Chem. Int. Ed. 1999, 38, 2326 – 2343;
e) E. A. Doherty, J. A. Batey, B. Masquida, J. A. Doudna, Nat.
Struct. Biol. 2001, 8, 339 – 343; f) D. J. Battle, J. A. Doudna, Proc.
Natl. Acad. Sci. USA 2002, 99, 11676 – 11681.
[2] a) R. B. Waring, Nucleic Acids Res. 1989, 17, 10281 – 10293;
b) E. L. Christian, M. Yarus, J. Mol. Biol. 1992, 228, 743 – 758;
c) S. A. Strobel, K. Shetty, Proc. Natl. Acad. Sci. USA 1997, 94,
2903 – 2908; d) L. Ortoleva-Donnelly, A. A. Szewczak, R. R.
Gutell, S. A. Strobel, RNA 1998, 4, 498 – 519; e) S. P. Ryder, L.
Ortoleva-Donnelly, A. B. Kosek, S. A. Strobel, Methods Enzy-
mol. 2000, 317, 92 – 109, and references cited therein.
DC209 P4-P6 can be used as a model system to define the
influence of atomic modifications on folding.[6] In addition to
the analogues described herein, we have defined the influence
of 2’-fluoro-, 2’-amino-, and 2’-methoxynucleotides on folding
in the context of this RNA domain.[6,19] Only the 2’-
fluoronucleotide analogues induce interference in accordance
with their strong preference for the 3’-endo sugar conforma-
[3] a) F. L. Murphy, T. R. Cech, Biochemistry 1993, 32, 5291– 5300;
b) J. H. Cate, A. R. Gooding, E. Podell, K. Zhou, B. L. Golden,
3036
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2004, 43, 3033 –3037

Angewandte
Chemie
C. E. Kundrot, T. R. Cech, J. A. Doudna, Science 1996, 273,
1678 – 1685; c) K. Juneau, E. Podell, D. J. Harrington, T. Cech,
Structure 2001, 9, 221– 231.
[4] M. L. Connolly, J. Am. Chem. Soc. 1985, 107, 1118 – 1124.
[5] H. Ammon, Struct. Chem. 2001, 12, 205 – 212.
[6] J. P. Schwans, C. N. Cortez, J. O. Olvera, J. A. Piccirilli, J. Am.
Chem. Soc. 2003, 125, 10012 – 10018.
[7] The synthesis of the 2’-deoxy-2’-methylnucleosides will be
reported elsewhere.
[8] J. P. H. Verheyden, J. G. Moffatt, J. Org. Chem. 1972, 37, 2289 –
2299.
[9] See the Supporting Information for the synthesis of 2’-chloro-2’-
deoxycytidine, adenosine, and guanosine, and characterization
of an RNA transcript containing a single 2’-chloro-2’-deoxyur-
idine residue.
[10] a) S. K. Silverman, T. R. Cech, Biochemistry 1999, 38, 8691–
8702; b) S. K. Silverman, M. L. Deras, S. A. Woodson, S. A.
Scaringe, T. R. Cech, Biochemistry 2000, 39, 12465 – 12475;
c) S. K. Silverman, T. R. Cech, RNA 2001, 7, 161 – 166.
[11] a) J. H. Cate, R. L. Hanna, J. A. Doudna, Nat. Struct. Biol. 1997,
4, 553 – 558; b) S. Basu, S. A. Strobel, RNA 1999, 5, 1399 – 1407.
[12] R. Sousa, R. Padilla, EMBO J. 1995, 14, 4609 – 4621.
[13] G. Gish, F. Eckstein, Science 1988, 240, 1520 – 1522.
[14] See the Supporting Information for representative DC209 P4-P6
sequencing ladders of transcripts containing phosphorothioates
of guanosine, 2’-deoxyguanosine, and 2’-deoxy-2’-methylguano-
sine.
[15] Residue G150 bears a functional 2’-hydroxy group but exhibits
no 2’-SH interference.[6]
[16] N. Guex, M. C. Peitsch, Electrophoresis 1997, 18, 2714 – 2723.
[17] M. Clark, R. D. Cramer III, N. V. Opdenbosch, J. Comput.
Chem. 1989, 10, 982 – 1012.
[18] P. Gordon, R. Fong, S. K. Deb, N-S. Li, J. P. Schwans, J. A.
Piccirilli, Chem. Biol. 2004, 11, 237 – 246, and references cited
therein.
[19] J. P. Schwans, PhD Thesis, University of Chicago, Department of
Chemistry, Chicago, 2003.
[20] a) S. P. Ryder, S. A. Strobel, J. Mol. Biol. 1999, 291, 295; b) S. P.
Ryder, S. A. Strobel, Nucleic Acids Res. 2002, 30, 1287.
[21] T. E. Creighton, Proteins: Structures and Molecular Properties,
2nd ed. (Ed.: W. H. Freeman), New York, 1993, pp. 140 – 148.
Angew. Chem. Int. Ed. 2004, 43, 3033 –3037
www.angewandte.org
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3037