Schlick T. Molecular Modeling and Simulation: An Interdisciplinary Guide

Подождите немного. Документ загружается.

220 7. Topics in Nucleic Acids Structure: Noncanonical Helices and RNA Structure

P1.1

U1ARBD

J1/2

J1.1/4

J4/2

P1.1

U1ARBD

J1/2

J1.1/4

J4/2

CGAA

’

P3P8 P7

P9.0

’

AAACC CU

P4 P6

Domain

GG G

CCUUUA

GGGGG

AAA

CUCUUCU

AA ACU

AUA

AAG

UUA

J8/7

J3/4

J6/7

P1.1

U1ARBD

Figure 7.6. Ribozymes with pseudoknots. Top: Hepatitis Delta virus (HDV) ribozyme

(PDB code 1DRZ) shows two pseudoknots, with pseudoknot basepairing pattern drawn

in the circular diagram with basepair number along the perimeter: a main one formed by

regions P1 (red) and P2 (orange-red), and a minor one formed by regions P1.1 (cyan) and

P3 (green). A hairpin formed by the P4 helix (blue) and the U1A ribonucleoprotein binding

domain (purple) is also present. Bottom: P3-P9 domain of the group I intron Tetrahymena

thermophila ribozyme (PDB code 1GRZ) shows one pseudoknot formed by regions P3

(red) and P7 (orange-red). The P4-P6 domain (yellow) folds independently. See Box 7.3

for a discussion of some of those structures. The intertwined or pseudoknotted regions are

clearly seen by the crossings in the circular diagrams.

These applications — together with design of novel RNAs and of RNA se-

quences called aptamers that bind speciﬁc molecular targets (e.g., [166, 199,

724]) — have important potential as regulators of gene expression, therapeutic

agents, molecular switches, and molecular sensors

.See[546, 889, 1207, 1208,

1275,1381], for some examples.

RNA sensors are RNA molecules that are turned on or off when they contact a target, like small

organic molecules, and can trigger a chemical reaction.

7.3. RNA Structure and Function 221

Designing new RNAs based on this idea of aptamers has led to the exciting

experimental ﬁeld of in vitro selection. This experimental technology involves

generating and screening large random-sequence libraries of nucleic acid mole-

cules for a speciﬁc function, such as binding or catalysis. Numerous nucleic acid

molecules binding targets (aptamers) have been developed, as diverse as organic

molecules, antibiotics, proteins, and whole viruses. In addition, new classes of

RNA enzymes (ribozymes) have been discovered by this approach, with applica-

tions in biomolecular engineering, such as in the design of allosteric ribozymes

and aptamer-based biosensors, aptamers capable of inhibiting protein function,

and therapeutic aptamers (e.g., inhibiting the TAR RNA element of HIV-1). Other

emerging applications of engineered RNAs include RNA nanotechnology, where

RNAs are assembled into functional arrays, and RNA synthetic biology, where

designed RNAs are used to control cellular functions (e.g., regulate gene ex-

pression). See Subsection 7.5.2 on analysis of random nucleotide pools to better

understand the relationship between sequence and structure/function and devel-

opment of a computational procedure for mimicking aspects of this selection

approach in silico.

7.3.4 Non-Coding and Micro-RNAs

Such newly found roles for RNAs, especially concerning tiny RNAs that do not

encode proteins (ncRNA for non-coding RNAs) but can inﬂuence gene action won

DNA’s cousin the venerable trophy of “Breakthrough of The Year” by Science

editors in 2002 (see the 20 December 2002 issue of Science, volume 298). Non-

protein coding stretches of mRNAs range in size from only 20 nucleotides to over

10000 nucleotides [1228]; they are required to control the translation from the

mRNA transcript into protein.

The 2002 award recognized a large group of papers that unraveled various fas-

cinating features of small RNAs (affectionately termed nanoRNAs). Micro-RNAs

(miRNAs), generally 21 to 25 nucleotides long, form a regulatory class of

ncRNAs [200]. These RNAs control gene expression by repressing translation

of target genes through, for example, binding to 3



untranslated regions of the

messenger-RNA targets.

Such small RNAs in animals, plants, and fungi collectively became associ-

ated with the title RNAi (for RNA interference). The agents that initiate RNAi

in a sequence-speciﬁc manner are double-stranded RNA segments that silence

gene expression (siRNAs, for small interfering RNAs). For example, they may

seek out the messenger RNA and destroy it, or they may bind to chromatin

and/or modify chromatin structure [383, 1004, 1432]. Though initially regarded

as anomalies, work is revealing that such siRNAs regulate gene expression in a

variety of organisms.

Such interference mechanisms by RNA silencing can provide an organism a

natural defense against invading viruses and transposons (DNA segments that mi-

grate within and across organisms and are associated with bacterial pathogenicity)

[11]. Consequently, this natural protection is being exploited by scientists using

222 7. Topics in Nucleic Acids Structure: Noncanonical Helices and RNA Structure

siRNAs to target viral genes that can inhibit the replication of HIV-1, polio, or

other viruses (e.g., [757]). Moreover, RNA interference mechanisms are being in-

vestigated by companies who apply them to discover the functions of genes by

turning them off to determine the effect on the plant or the animal. For exam-

ple, a landmark study on obesity employed RNA interference [626] to inactivate

about 85% of the roundworm’s predicted 19,757 genes that code for proteins in

a single experiment [66]. More recent studies have used RNA interference to un-

cover genes involved in regulating the immune response to pathogenic bacterial

infections [273] and genes regulating programmed cell death (or apoptosis)[229].

These fascinating discoveries regarding RNA’s interference with gene activity

are associated with many epigenetic phenomena — changes in gene expression

(inheritance of features) that do not involve alterations in the genome and persist

across at least one generation [31]. A different kind of epigenetic control was also

discovered by Breaker, Nudler, and coworkers [863,864,889,926,1275,1381]in

some bacterial messenger RNAs containing sequences that sense small molecules

directly to control translation of mRNA into protein. Namely, speciﬁc control

regions of mRNA can bind directly to metabolites associated with vitamin B syn-

thesis and import, and induce a conformational change in RNA’s folding state; this

metabolite-triggered conformational change acts as part of the signal transduction

pathway that senses vitamin level and controls enzyme production. Such RNAs

that control gene expression upon binding metabolites or other ligands have been

termed riboswitches.

A riboswitch can undergo local and global conformational changes upon bind-

ing to its substrate. The secondary and tertiary structures of these natural aptamers

can be very diverse, and several structures have been solved [873]. Examples in-

clude the type-I glmS ribozyme and purine and SAM-II riboswitches in which one

binding region induces local conformational changes upon binding. TPP, SAM-I,

and the M-box magnesium are examples of type II riboswitches because of their

two regions which undergo conformational changes upon binding their metabo-

lite, making possible global as well as local conformational rearrangements (see

Figure 7.7).

Such a switch of RNA conformation between two states in a ligand-dependent

manner (see Figure 7.7) also opens new avenues for thinking about RNA design

in a variety of contexts (e.g., [166,199,242,599,1028,1245]). Like the Paracelsus

challenge for proteins (see Chapter 2), one can formulate a similar challenge for

RNA design — describe minimal changes in the nucleotide sequence to trigger

a conformational rearrangement in the folding of a given RNA molecule — that

can be approached by a combination of computational and experimental wizardry.

Indeed, Science editor Jennifer Couzin [267] writes: “Having exposed RNAs’

hidden talents, scientists now hope to put them to work”.

7.3.5 RNA at Atomic Resolution

Until fairly recently, less was known at atomic resolution on RNA structure

in comparison to DNA, but this has changed as the ‘RNA era’ has begun.

7.3. RNA Structure and Function 223

Progress in RNA structure elucidation can be attributed to vast improvements in

crystallization procedures (e.g., RNA structure determination through crystalliza-

tion with a protein that would not interfere with the enzyme’s activity [391]), as

well as alternative approaches for studying RNAs such as high-resolution NMR,

spectroscopy, crosslinking reactions, and phylogenetic data analysis. Our knowl-

edge of RNA structures has also increased dramatically with recent solutions of

ribosomes [85, 201, 206, 210, 910, 922, 1135, 1374, 1379, 1427], since ribosomes

contain numerous tertiary motifs of RNAs and therefore provide a rich resource

of information on RNA structural elements and organization.

The clover-leaf structure of the tRNA molecule has been known for decades,

and for a long time was the only well-characterized major structure of an RNA

molecule [1173]. Its structure whet our appetite for RNA appreciation by reveal-

ing the long-distance tertiary interactions. By now, RNA folds characterized by

X-ray crystallography include the tRNA, hammerhead ribozyme, Tetrahymena

group I intron, hepatitis delta virus ribozyme, Group I intron from Azoarcus and

Twort, Group II intron, various ribozymes (e.g., hairpin, GlmS, Diels-Alder), var-

ious riboswitches (e.g., purine, M-box, and TPP), RNase P types A and B, signal

recognition particle, and various fragments such as kissing hairpin and sarcin/ricin

motifs. See some examples in Figure 7.7 and details of some solved catalytic

RNAs in Box 7.3.

Box 7.3: RNAs at Atomic Resolution

The hammerhead [1005, 1282]andHepatitis Delta Helper virus (HDV) [391]ri-

bozymes, in the family of self-cleaving catalytic RNA, were solved in 1994 and 1998,

respectively. The hammerhead’s Y, or wishbone-shaped, structure has three base-paired

stems resembling the head and handle of a carpenter’s hammer (see Figure 7.4). The RNA

is unpaired in its U-turn core and stabilized by non-WC, non-wobble bps in the stems.

Visualizing its structure led to further analysis of the mechanism of RNA self cleavage

through trapping of intermediates in the ribozyme reaction pathway [331, 886]. While the

hammerhead’s active site is open, that of HDV is hidden (see Figure 7.6), resembling the

catalytic sites of globular proteins, and contains two pseudoknots.

The crystal structure of one self-folding Tetrahymena thermophila Group I intron [488],

was solved in 2004. Its secondary structure consists of nine regions (P1-P9), which fold

into two major domains: P3-P9 and P4-P6. Both domains are stabilized through extensive

inter-domain tertiary interactions. The P3-P9 domain is formed by a coaxial helix between

P3 and P8, with a slight bend with respect to P7, while helix P9 is oriented perpendicular

to P7 (see Figure 7.6). The P4-P6 domain has a helical tetraloop region connected to

another helical segment, the tetraloop receptor,byalargebend(≈150

) (see Figure 7.4).

Thus, its active site is hidden and exempliﬁes the complexity of RNA structure, involving

complex intertwining of secondary structure elements, including pseudoknots. Indeed,

both WC and non-canonical base-paired regions are interspersed with internal loops, and

an adenosine-rich bulge region mediates the long-range tertiary contacts in the RNA,

improving base stacking interactions. The adenosine platform motif emerges from this

structure as an important architectural component of RNA that might have arisen early

224 7. Topics in Nucleic Acids Structure: Noncanonical Helices and RNA Structure

Figure 7.7. Examples of two riboswitches: SAM-I and TPP riboswitch. In the presence

of its substrate, S-adenosynlmethionine (SAM), the SAM riboswitch is composed of a

four-way junction forming a pseudoknot and a pair of coaxial helices P1–P4 and P2a–P3.

The junction is stable due to hydrogen bonding between nucleotides (orange) in helices P1

and P3 which bind to SAM. The bound state of the riboswitch was solved by crystallogra-

phy [872]. In the absence of SAM, the 3



-side of helix P1 forms an anti-terminator element

(AT), allowing the mRNA to be fully transcribed. The TPP riboswitch The thiamine py-

rophosphate (TPP) riboswitch from Escherichia coli thiM mRNA also has two forms

depending on the presence of TPP. The bound-state form has been solved by X-ray crys-

tallography [1166]. Without TPP, the 3



-side of helix P1 binds to an anti-ShineDalgarno

(anti-SD) region, allowing the ribosome to access the SD element so that translation can

start as proposed by the Breaker group [1381].

in evolution. As found in other RNA structures, metal-phosphate coordination is also

important in stabilizing tertiary contacts in this intron RNA domain. In addition to the

Group I intron of the Tetrahymena thermophila, the Group I intron of the Azoarcus [3]and

Twort [466] species are also available.

7.4. Current Challenges in RNA Modeling 225

The complete Group II intron solved in 2008 [1266] is a self-splicing ribozyme that

catalyze their own excision from precursor mRNAs and joins together ﬂanking exons with-

out the help of proteins. The secondary structure is characterized by six domains (I to VI)

composed of step-loop structures connected by a common central core and stabilized by

long-range tertiary interactions. Domain I is the largest element and contains recognition

sequences responsible for the correct assemble of the intron in its active form. Domains

II and III enhance the catalytic efﬁciency of the splicing, while Domain IV contains the

open reading frame (ORF) for expression of a reverse-transcriptase enzyme. Domain V is

the most phylogenetically conserved element and contains a hairpin loop involved in catal-

ysis. Domain VI contains an adenosine that interacts with the splice site during splicing.

The Group II intron and the spliceosome share common structural and functional features,

supporting the hypothesis of a common ancestor [1266].

7.4 Current Challenges in RNA Modeling

7.4.1 RNA Folding

Deducing the functional structure of RNA molecules from the primary sequence

has been called the RNA folding problem [172, 241, 771, 1261]. The challenge

is to understand how the strong electrostatic repulsions between closely packed

phosphates in RNA are alleviated. Indeed, the stability of compact RNA forms

is strongly maintained through interactions with both monovalent and divalent

cations and by pseudoknotting.

As mentioned in connection to the multitude of hydrogen bonding patterns for

polynucleotides in Section 7.2.1, the rich variety of possibilities in RNA, as re-

viewed in [736, 739, 740], has led to proposal for new nomenclature for RNAs

and the annotation of RNA structure through the international RNA Ontology

Consortium,ROC(http://roc.bgsu.edu/)[735].

7.4.2 RNA Motifs

We now recognize at least seven major tertiary interaction motifs for RNA, as

shown in Figure 7.8 — subdivided into three classes. These classes separate in-

teractions between two double stranded helices, between a single strand and a

helix, and between two single stranded regions. A recent annotation study of 54

representative high-resolution solved RNA structures showed the dominance of

A-minor motifs (37%), coaxial helices (32%), and ribose zippers (20%), which

together account for 89% of the total motifs (which number 613) [1403](see

Figure 7.9). Correlations among motifs, such as a pseudoknot or coaxial helix

with A-minor, reveal patterns of higher organization and underscore RNA’s hier-

archical structure (Figure 7.9). Analyses of RNA junctions of orders 4 through 10

further suggested global patterns and subnetworks that organize RNA in complex

226 7. Topics in Nucleic Acids Structure: Noncanonical Helices and RNA Structure

ways as well as a classiﬁcation into distinct families (Figure 7.10)[690, 691].

Such studies can help in the goal of RNA structure prediction by deﬁning major

motifs in RNA and pinpointing speciﬁc sequence/structure relations.

7.4.3 RNA Structure Prediction

Predicting the secondary and tertiary folding of RNA is a difﬁcult and ongoing

enterprise [172, 225, 241, 277, 689, 1022, 1145, 1465]. Secondary structure ele-

ments are easier to identify through free energy minimization combined with

comparative analysis (sequence alignment) using evolutionary and database rela-

tionships [838,839,1460,1461]. The minimized objective function is empirically

derived based on basic physiochemical laws. However, caution is warranted in the

S/S

tRNA D-loop:T-loop

Kissing hairpin

Pseudoknots

Self-complementary nucleotides

Two intertwining

regions forming

Complementary

basepairs

D/T loop

interaction

Ribose zipper

Antiparallel

stem/loop

interaction

5'-CC-3' (Stem)

3'-AA-5' (Loop)

A-minor motif

Clustering

of adenines

G-C preferred

Tetraloop receptor

Tetraloop/internal loop

5'- GAAA -3'

5'-CC-UAAG-3'

S/H

H/H

Coaxial helices

Junction,

“pseudo-stem”

3' 5'

3'5'

3' 5'

3'5'

3' 5'

3'5'

3' 5'

Two hairpins

3' 5'

RNA Tertiary Motifs

Figure 7.8. Seven major motifs of tertiary interactions in RNA, involving two double

stranded helices (H/H), single strand and a helix (S/H), and two single stranded regions

(S/S). Coaxial helices form when nucleotide bases from two separate helices stack and

align axes to form a pseudo-continuous coaxial helix. A-minor motifs originate from clus-

tering of adenosine interactions, often present within other motifs such as coaxial helices.

Tetraloop receptor is an interaction between a tetraloop (GNRA, where N = A,G,C,U and

R = A,G) and an internal loop plus two GC bps. Ribose zippers form when two consecutive

residues from one chain segment interact in an antiparallel fashion with two consecu-

tive residues from another chain segment distant in sequence but close in space. Kissing

hairpins form by base pairing between single-stranded residues of two hairpins with com-

plementary sequences, and tRNA D-loop:T-loop are interactions between two conserved

hairpins in tRNA. See glossary of [1403] for more details and references to original motif

deﬁnitions.

7.4. Current Challenges in RNA Modeling 227

Coaxial helix

32%

A-minor motif

37%

Ribose zipper

20%

Other

11%

TPP riboswitch

(2GDI)

5’

3’

A-minor motif

Ribose Zipper

Loop-loop receptor

Coaxial helix

Figure 7.9. The distribution of RNA tertiary motifs in a non-redundant set of 54 high-

-resolution crystal structures and annotated diagram of the TPP riboswitch (PDB 2GDI)

showing correlated motifs [1403].

interpretations of such 2D structures since predictions are imperfect, especially

for more than 200 nucleotides. The new structures, however, provide opportuni-

ties for learning what works, as well as what fails, in structure prediction. Thus,

discriminating among the possible tertiary interactions to obtain the ﬁnal folded

state remains a challenge [277, 1115,1261]. Still, ﬁndings concerning the folding

kinetics of Tetrahymena ribozyme [1149] have suggested that, as thermodynamic

data on tertiary structure interactions become available [1261], the RNA folding

problem might be easier to solve than protein folding [101,1261]. Therefore, with

advances in RNA synthesis and structure determination [558] and the availability

of thermodynamic data on tertiary interactions [1261], it is likely that our under-

standing of RNA structure, RNA folding, and RNA’s role in enzyme evolution

will dramatically increase in the coming decade. These developments are pro-

pelling studies in RNA informatics (or ribonomics)[332] and RNA design (e.g.,

[166,199,241,242,359,599,600,689,1115,1142,1245, 1245, 1251, 1281]).

Emerging themes in RNA structure include the importance of metal ions and

loops for structural stability, hidden active sites in some ribozymes like cat-

alytic sites of proteins, various groove binding motifs (e.g., adenines interacting

with minor-groove helices to stabilize tertiary contacts, as deduced from crys-

tal structures [327, 911] and statistical analyses [1383]), architectural motifs

tailored for intermolecular interactions [546], hierarchical folding, fast estab-

lishment of 2D elements, and extreme ﬂexibility of the molecule as a whole

[172,277,390,493,1261]. See recent review [1465].

Some key challenges concerning RNA include ﬁnding novel RNA genes,

identifying the biological roles of these RNA genes, determining the structural

repertoire of RNA, determining RNA tertiary folds from sequence, and designing

novel RNAs [1112]. Bioinformatics tools hold great promise in addressing these

challenges.

228 7. Topics in Nucleic Acids Structure: Noncanonical Helices and RNA Structure

P64

P66

P65

P67

P61

P63

P62P64

Three-Way Junctions

P57

P59

P58

Family A

23S rRNA (2J01)

P75

P79

P76

P28

P43

P29

P75

P79

P76

P28

P43

P29

3'5'

Family B

16S rRNA (2J00)

P57

P59

P58

3' 5'

3'5'

Family C

23S rRNA (1S72)

Four-Way Junctions

Family H

(RNase P type B 1NBS)

P3b

P3aP3c

P10

P3P2

P51

P53

P52

P54

P11

P13

P12

P14

5'3'

P3b

P3a

P3c

5'3'

P3P2

Family cH

(HCV IRES domain 1KH6)

Family cL

(tRNA-Phe 1EHZ)

P29

P41

P30

P42

P27

P29

P28

P31

P75

P79

P76

P79

Family cK

(23S rRNA 1S72)

Family π

(Ribonuclease P 1U9S)

5' 3'

P11

P13

P12

P14

5' 3'

5'3'

P61

P63

P62

P64

5'3'

Family cW

(23S rRNA 2J01)

Family

(23S rRNA 2J01)

P64

P66

P65

P67

3' 5'

Family cX

(16S rRNA 2J00)

5'3'

P29

P41

P30

P42

5' 3'

P27

P29

P28

P31

Family X

(16S rRNA 2J00)

3' 5'

P10

5' 3'

Base pair interactions

Watson-Crick edge

Hoogsteen edge

Sugar edge

Base involved in

tertiary interaction

Cis Trans

Hydrogen bond interaction

AU cis Watson-Crick base pair

GC cis Watson-Crick base pair

Helix-packing interaction

RII

Bifurcated base pair

Ribo-base interaction type I

Ribo-base interaction type II

Figure 7.10. Classiﬁcation of RNA junctions into families according to coaxial stacking

properties, perpendicular helical conﬁgurations, and ﬂexible helical arms. Lines inside the

helices represent the canonical WC bps GC, AU, and the GU wobble bp. The network

symbology follows the Leontis Westhof notation [738].

7.5. Application of Graph Theory to Studies of RNA Structureand Function 229

P2 6

P4 7

P2 5

P7 2

P7 3

P9 4

P9 8

P9 9

P3 6

P3 7

P3 9

P4 0

P3 8

P4 1

P4 5

P4 A

P11

P12

P13

P14

P8 0

P7 4

P7 5

P8 1

P8 2

P8 8

P15

P15.1

P15.2

P7.1

P10

P14

P16

P15

P21

P22

Five-Way Junctions

Group I Azoarcus

intron (1U6B)

23S rRNA

(

1S72

)

tRNA

(2BTE)

5'3'

P13

P14

P10

P27

P29

P28

P30

P31

5'3'

5' 3'

P7 . 1

P10

5' 3'

5'3'

P14

P15

P16

P2 1

P2 2

16S rRNA

(2AVY)

P6P7

P14

P13

Group II

Intron (3EOH)

5' 3'

Pi i

23S rRNA

(1S72)

P3 0

P2 7

P2 8

P3 1

P2 9

3' 5'

Ribonuclease P

(2A64)

23S rRNA

(2AW4)

23S rRNA

(2J01)

P4 A

P11

P12

P13

P14

P15

P15.1

P15.2

P8 0

P7 4

P7 5

P8 1

P8 2

P8 8

5'3'

23S rRNA

(2AW4)

RII

P3 6

P3 7

P3 9

P4 0

P3 8

P4 1

P4 5

23S rRNA

(2AW4)

P2 6

P4 7

P2 5

P7 2

P7 3

P9 4

P9 8

P9 9

Six-Way Junctions

Seven-Way

Junctions

Ten-Way Junctions

Classiﬁcation of RNA junctions (continued).

7.5 Application of Graph Theory to Studies of RNA

Structure and Function

7.5.1 Graph Theory

The application of graph theory to describe secondary structure motifs of RNA,

as pioneered by Waterman in 1978 and extended by others [116, 438, 708, 1350],