RESEARCH ARTICLE

complementarity to the true A-form heterodu-

Structure of a Transcribing T7 plex. The inhibitory effect on T7 RNAP of T7
lysozyme, which binds to the side of the poly-

RNA Polymerase Initiation merase opposite the active site (12), is achieved, in
part, by this transcription factor allosterically al-

Complex
tering the structure of the polymerase active site
and its suitability for catalysis.

Graham M. T. Cheetham1 and Thomas A. Steitz1,2,3* The Structure of a Transcribing RNAP

Initiation Complex
The structure of a T7 RNA polymerase ( T7 RNAP) initiation complex captured Because the cocrystal form of the T7 RNAP
transcribing a trinucleotide of RNA from a 17– base pair promoter DNA con- initiation complex with promoter DNA, three
taining a 5-nucleotide single-strand template extension was determined at a nucleotides of transcribed RNA, and a ribonu-
resolution of 2.4 angstroms. Binding of the upstream duplex portion of the cleoside triphosphate analog is nearly isomor-
promoter occurs in the same manner as that in the open promoter complex, phous with the crystal of the binary promoter–
but the single-stranded template is repositioned to place the ⫹4 base at the T7 RNAP complex (T7 RPP), the new structure
catalytic active site. Thus, synthesis of RNA in the initiation phase leads to could be solved by rebuilding models into
accumulation or “scrunching” of the template in the enclosed active site pocket difference maps and subsequent coordinate re-
of T7 RNAP. Only three base pairs of heteroduplex are formed before the RNA finement to a crystallographic R factor for 5% of
peels off the template. data omitted from refinement, Rfree, of 0.26% at
a resolution of 2.4 Å (13). The template exten-

Downloaded from http://science.sciencemag.org/ on January 9, 2018

In comparison to the multisubunit DNA-de- transition or conformational change occurs to sion encodes the RNA transcript 5⬘-GGGAG
pendent RNAPs that synthesize messenger form the more stable elongation complex that is at positions ⫹1 though ⫹5. Thus, with the
RNA in prokaryotes and eukaryotes (1, 2), capable of transcribing the complete template. addition of guanosine 5⬘-triphosphate (GTP)
the DNA-dependent RNAP from a T7 bacte- The intermediate DNA-RNA hybrid duplex is and ␣,␤-methylene adenosine 5⬘-triphosphate
riophage is a relatively simple enzyme (of analogous to the primer-template DNA homo- (ATP), the enzyme synthesizes a ribonucleotide
molecular weight 100 kD) that is capable of duplex product synthesized by DNA poly- trimer, pppGpGpG (p, phosphate group), which
transcribing the complete T7 phage DNA. T7 merases. However, in contrast to the DNA can be seen in the experimental difference elec-
RNAP is homologous to eukaryotic mito- polymerase, the transcript peels away from the tron density maps, and then prematurely stalls.
chondrial, chloroplast, and other phagelike template and exits the transcribing polymerase The incoming ATP analog and two associated
RNAPs (3) and to the well-studied polymer- complex. A final contrast with the DNA poly- magnesium ions are evident in the 2.8 Å reso-
ase I (pol I) family of DNA polymerases (4, merases is that the initiation of transcription by lution structure of a second isomorphous crystal
5). Because it shows most of the functional RNAPs is regulated by many protein transcrip- (14). This ATP substrate can be positioned in
features that are characteristic of RNAPs, its tion factors. the first T7 RNAP complex by superposition of
close relation to the pol I DNA polymerases The structure of a binary complex formed 12 C-␣ atoms located in the active site ␤ strands.
makes it ideal for understanding the structural between T7 RNAP and a 17– base pair open The T7 RNAP in this transcribing quaterna-
bases for the functional differences between promoter DNA (T7 RPP) (10) established that ry complex (Fig. 1) makes the same interactions
DNA and RNA polymerases. domains present in the RNAP but not in the with the duplex part of the promoter (⫺5 to
DNA-dependent RNAPs have several func- DNA polymerase are able to recognize specific ⫺17) as observed previously in T7 RNAP (10),
tional features not seen in DNA polymerases, promoter sequences and denature the duplex to and the protein structures are nearly identical in
although they share a common two-metal-ion form an initiation bubble. The template base at the two complexes. In contrast, the single-
mechanism for phosphodiester bond formation position ⫺1 is bound in a hydrophobic pocket stranded template portion of the open promoter
(6). The RNAPs can initiate transcription, adjacent to the catalytic active site, and this “bubble” (⫺1 to ⫺4) has a dramatically differ-
which involves recognition of specific promoter positions the template strand for initiation by a ent conformation. The primer terminus of the
DNA sequences (7) followed by melting of the single nucleotide. On the basis of the structures RNA transcript (now at ⫹3), the incoming ATP
DNA strands close to the catalytic active site of this T7 RNAP binary complex, the ternary analog, and the template nucleotides to which
and precise positioning of the single-stranded complex of T7 DNA polymerase (11), a model they are paired have nearly the same location in
template at positions ⫹1 and ⫹2 for base pair- of an initiation complex was constructed with a the active site as was modeled for the incoming
ing with incoming ribonucleoside triphos- priming nucleotide at ⫹1 and an incoming and priming NTPs at positions ⫹1 and ⫹2 (10)
phates. In contrast to DNA polymerases, initi- nucleoside triphosphate (NTP) at ⫹2. by homology to the T7 DNA polymerase ter-
ation is primed by a single nucleotide rather The structure of a transcribing initia- nary substrate complex (11). Thus, as synthesis
than an oligonucleotide. The initiation phase is tion complex shows that, during synthesis, proceeds down the template, the catalytic and
characterized by abortive cycling, a process of the template strand has accumulated or promoter recognition sites maintain the same
repeated synthesis and release of short RNA scrunched into a pocket in the enzyme to relative orientation by accumulating or scrunch-
products (8). During initiation, the template place positions ⫹3 and ⫹4 of the template ing the transcribed template in a pocket on the
strand and active site must constantly translate strand at the catalytic site. Only the 3 base enzyme.
in relation to each other while a firm grip of the pairs of heteroduplex observed are likely to Structural studies of DNA polymerases
promoter is retained (9). Synthesis of short form before the transcript peels off and exits from a variety of organisms (4, 5, 11, 15–17)
oligonucleotide products continues until some through its own RNA binding cleft. The spec- have a common overall shape that has been
ificity for synthesis of RNA rather than DNA morphologically compared to a right hand with
1
Department of Molecular Biophysics and Biochemis-
transcripts appears to reside largely in the nu- domains called “thumb,” “fingers,” and “palm.”
try, 2Department of Chemistry, Yale University, merous hydrogen bonds made between the pro- The active site of T7 RNAP is located in a deep
3
Howard Hughes Medical Institute, New Haven, CT tein and 2⬘-OH of the incoming NTP and each pocket bounded by the polymerase fingers,
06520 – 8114, USA. of first three nucleotides of the transcript (at thumb, and palm subdomains and an NH2-
*To whom correspondence should be addressed. least) as well as the enzyme’s structural terminal domain that is unique to the RNAP

www.sciencemag.org SCIENCE VOL 286 17 DECEMBER 1999 2305

 RESEARCH ARTICLE
(10). In contrast, DNA polymerases bind the fixed by a hydrogen bond between the 2⬘-OH Specificity of T7 RNAP for
primer-template DNA homoduplex product in a ribose group of G(⫹2) and the main-chain Synthesizing RNA Rather Than DNA
more open cleft. This difference in the active carbonyl of residue 425. Because of the strict There are several important ways in which this
site has important implications for many of the stereochemical requirements for catalysis in pol I family RNAP differs from the homolo-
fundamental differences between DNA-depen- the active site, the positions of the ribose gous T7 DNA polymerase; these differences
dent RNAPs and DNA polymerases. rings at the 3⬘-terminus of the RNA transcript enable it to synthesize RNA rather than DNA
The newly transcribed RNA chain forms a (⫹3), the 3⬘-end of the primer strand in DNA transcripts. Not only does this enzyme selec-
DNA-RNA hybrid duplex inside the pocket polymerase, and the incoming NTPs (⫹4 in tively bind and incorporate ribonucleotides
(Figs. 1 and 2), has an exact A-form helical T7 RNAP) are very similar. Thus, the active rather than deoxyribonucleotides, but it also
conformation, and occupies a position similar site of T7 RNAP is able to specifically bind binds the 2⬘-OH groups of the transcript RNA
to that of the first few base pairs of primer- ribonucleotides with C3⬘-endo ribose puckers and accommodates the C3⬘-endo ribose confor-
template DNA duplex product seen in DNA and conserve the catalytic geometry in the mation. DNA polymerases enforce specificity
polymerase complexes. The helical axis of active site. for the incorporation of deoxyribose nucleo-
the template-RNA heteroduplex is, however, The helical axis of the upstream B-form tides because a bulky residue (“steric gate”)
rotated by ⬃15° toward the thumb subdo- duplex region of the T7 DNA promoter has a sterically excludes ribose (19). In T7 DNA
main, as compared to the DNA homoduplex similar orientation to that of the heterodu- polymerase, the residue Glu480 functions as a
product bound to T7 DNA polymerase (11). plex, but the two helical axes are offset by steric gate. The corresponding residue in T7
The thumb subdomain of T7 RNAP does not ⬃30 Å (Fig. 1). The promoter duplex is RNAP is Gly542, which is absolutely conserved
interact with the heteroduplex, as does the bound to T7 RNAP at a site remote from the in phagelike RNAPs. The lack of any side chain
primer-template homoduplex in DNA poly- active site pocket. Positions ⫺4 through ⫺1 allows the 2⬘-OH of the incoming ribonucleo-
merases. The position of the heteroduplex is in the template strand connect the DNA-RNA side triphosphate to be accommodated. Residue
His784 can move toward and directly interact

Downloaded from http://science.sciencemag.org/ on January 9, 2018

heteroduplex and the DNA homoduplex he-
lices. This single-stranded region is bound with the 2⬘-OH ribose position of NTP(I). In
within the active site pocket to form part of DNA polymerase, the position corresponding
the upstream transcription bubble. As ob- to His784 is usually occupied by glutamine (Fig.
served in the T7 RPP structure, the nontem- 3), which is constrained to adopt an alternate
plate strand from positions ⫺3 through ⫺1 is side-chain rotamer because of the steric gate
disordered in electron density maps. residue, Glu480. Residue Tyr639 is also within
hydrogen-bonding distance of the 2⬘-OH of
NTP(I), although it is disordered in the quater-
A K631 nary complex.
F644
The “O helix” in the fingers subdomain of
M635
Y639 D812 pol I family polymerases contains a highly con-
C+5
ATP(I) binding served sequence motif that is important for
pocket
T+4 3' binding the incoming nucleotide. However, its
5' 2'
C+3
conformation in T7 RNAP is different from that
Fingers C+2
G+3
5' in the homologous DNA polymerases. Residue
Tyr639, homologous to Tyr766 in Escherichia
template G+2 coli DNA polymerase I, Tyr530 in T7 DNA
strand C+1 polymerase, and Tyr714 in Bacillus DNA poly-
GTP+1
merase (19), is located in a loop region imme-
diately adjacent to the O helix. It is not part of
B the helix, as observed in DNA polymerase, but
adopts a novel position with respect to NTP(I).
RNA
transcript The mutation Tyr639 3 Phe639 substantially
template 5' reduces the preference of T7 RNAP for ribonu-
cleotide over deoxyribonucleotide substrates
(20). It also increases the misincorporation of
Fig. 1. The structure of a transcribing T7 RNAP bases and the extension of short RNA tran-
complex. (A) The position of the template
(gray) and nontemplate (purple) DNA, RNA 2.8Å scripts. An analogous mutation, Tyr766 3
GTP+1
transcript (green), and ATP (light green) on a 3' Phe766, in E. coli DNA polymerase I affects the
C+1
surface representation of RNAP. The thumb fidelity of deoxyribonucleotide incorporation
(green) is truncated so the heteroduplex and O Fig. 2. The active site heteroduplex. (A) The (21). These data suggest that residue Tyr639
helix (light blue) can be seen in the active site pppGpGpG RNA transcript (green) and DNA may have a multifunctional role in RNAP,
pocket. The fingers are blue, the palm is purple, template (gray) on a surface rendition of the which may be different than its role in DNA
the palm insertion is pink, and the NH2-termi- active site pocket and color coded as in Fig. 1
nal domain is yellow. (B) The structure of the (40). Residue Y639 from the end of the O helix
polymerase. Additional mutations suggest that
template strand (gray), nontemplate strand interacts with T⫹4 and partly overlaps the Tyr639 may distinguish between GTP(I) and
(purple), RNA transcript (green), and next in- ATP(I) binding site in the complex missing the ATP(I) (22).
coming ribonucleoside triphosphate (light ATP. Residues M635 and T636 are conserved in The phenolic ring of Tyr639 plays a role in
green) as seen in the transcribing complex with phagelike RNAPs and provide the hydrophobic positioning the ⫹3 ribonucleotide and, conse-
T7 RNAP. The template-RNA heteroduplex is an surface against which the base of NTP(I) packs. quently, NTP(I) (Fig. 2A) by stacking against
A-form helix that is displaced from the up- (B) Peeling of the transcript away from the
stream promoter duplex. Polymerase elements template after four base pairs. Electron density
the template base (⫹3), which is paired with the
interacting with the promoter duplex are for the mismatched base pair at position ⫹1 is primer terminus. It also helps to redirect the
shown. Figures 1, 2, 3, and 5 were generated shown. Dashed line indicates single hydrogen template strand away from the active site across
with the program RIBBONS (39). bond. the face of the fingers subdomain so that posi-

2306 17 DECEMBER 1999 VOL 286 SCIENCE www.sciencemag.org

 RESEARCH ARTICLE
tion ⫹5 stacks against residue Phe644, in a binding of dNTP to the polymerase-DNA bina- affinity of the T7 RNAP–lysozyme complex
similar manner to that seen in human immuno- ry complex (26). This large rotation in the for NTPs (30). The palm-insertion module (res-
deficiency virus (HIV)–type 1 reverse tran- fingers subdomain orients catalytically impor- idues 450 through 527) (27), which is not
scriptase (23). On the basis of the observed tant amino acids located in the O helix (which is present in DNA polymerase, closes in behind
position of the incoming nucleoside, ATP(I), conserved in the pol I family) or in an analo- the FOOT, preventing it from retracting out of
and the position of the ⫹4 template base in this gous feature in other polymerase families. In its hydrophobic binding site. The FOOT is lo-
structure, the phenolic ring of residue Tyr639 the transcribing complexes of T7 RNAP, re- cated beneath the active site, more than 8 Å
partially overlaps with the ATP(I) binding ported here, the fingers subdomain is consider- away from the catalytically essential carboxyl-
pocket when the ATP(I) is not present. It ap- ably more closed than in either the unliganded ate residues, and is not in a position to interact
pears that residue Tyr639 can compete with (27) or T7 lysozyme complex (12) crystal with either the incoming ribonucleotide sub-
NTP(I) for similar positions within the T7 structures. However, the arrangement of poly- strate or RNA transcript, as had previously been
RNAP active site. The role of residue Tyr639 in merase domains are unchanged with respect to suggested (31, 32).
T7 RNAP, seen in our structure, is therefore the binary T7 RPP complex. Although this
very similar to that of residue Tyr714 in the closing of the polymerase conformation upon Length of the Transcript-Template
structures of Bacillus DNA polymerase com- promoter binding by T7 RNAP is small (in Heteroduplex
plexed with several DNA substrates (19). relation to the differences observed in the DNA It appears that the intermediate heteroduplex
There are other important structural differ- polymerases), it is sufficient to move residue cannot extend beyond a total of four base pairs,
ences between T7 RNAP and the homologous Lys631, which interacts with the triphosphate including the interaction of the template with
DNA polymerases that enable it to specify the moiety of the incoming ribonucleotide substrate the next incoming substrate, NTP(I). Extension
formation of an RNA-DNA heteroduplex. Dis- (28), and residue Tyr639 6 Å closer to the of the template-RNA heteroduplex in the active
crimination between ribose and deoxyribose is catalytic active site. site pocket by even one additional base pair

Downloaded from http://science.sciencemag.org/ on January 9, 2018

achieved, in part, by hydrogen bonding be- would result in clashes with the polymerase
tween the protein and the 2⬘-OH groups of the Regulation of T7 RNAP Activity by T7 NH2-terminal domain. The non–Watson-Crick
growing RNA transcript. Specificity for the Lysozyme
2⬘-OH at the primer terminus, NTP(P), is T7 lysozyme appears to inhibit transcription
achieved by a direct hydrogen bond between by T7 RNAP in part by allosterically chang- A
I(+4)
Arg425 and the 2⬘-OH ribose group. Residue ing the T7 RNAP active site structure. In the P(+3)
Arg425 also hydrogen bonds to the N3 position complex between T7 RNAP and T7 ly-
of G⫹3 (and could make similar contacts with sozyme (12), the COOH-terminal residues
the N3 position of A or with the O2 position of and the closely associated active site of T7
Y639
U and C), located in the minor groove of the RNAP are in an altered conformation when 3.2Å
3.1Å
A-form template-RNA hybrid. In the T7 DNA compared to the transcribing complex de- .0
Å 2' 2'
D812

Å
3

3.2
polymerase (11) and the Thermus aquaticus scribed here. H784
DNA polymerase (17), primer-template com- The catalytic active site of T7 RNAP is G542
plexes (residues Arg429 and Arg573, respective- located on the surface of three ␤ strands in the R425 H811

ly) make only a single hydrogen bond with the palm subdomain, which lies between the fin-
O4⬘ deoxyribose position of the primer 3⬘-ter- gers and thumb subdomains. The COOH-ter-
minus. The constraints imposed on the position minal residues of T7 RNAP and other phage- B
and orientation of residue Arg425 by the sur- like RNAPs have a sequence that is highly G542

rounding side chains in T7 RNAP therefore conserved and occupy a position that is similar E480 3' D654
define a unique binding site for ribonucleotides to that occupied by a fourth ␤ strand seen in 2'
D812

with C3⬘-endo ribose conformation. The 2⬘-OH homologous DNA polymerase structures. In the Y639 I(+4) 3'
of the second nucleotide from the primer termi- absence of a bound promoter, residues Phe882- Y530
nus, G(⫹2), is hydrogen bonded to a backbone Ala883-COOH can be cleaved in solution by H784 R425 P(+3)
carbonyl oxygen. carboxypeptidase A; however, this “FOOT” be-
o4'
The binding site for the RNA-DNA hetero- comes resistant to proteolysis during transcrip-
duplex is complementary to the true A-form tion initiation (29). In the T7 RNAP–T7 ly- H811
Q615 H653
duplex rather than to the conformation seen sozyme complex (12), the partially ordered R429
with DNA polymerases (11, 17, 19, 24). The FOOT adopts an extended conformation,
ribose backbone of the heteroduplex bound to whereas in the transcribing complex, Phe882 Fig. 3. Comparison of the nucleotide binding sites
T7 RNAP superimposes on the crystal structure and Ala883 are bound in a hydrophobic pocket in T7 RNAP (red and blue) and T7 DNA polymer-
of an RNA-DNA heteroduplex (25) with a root beneath the three ␤ strands. This protects them ase (gray). The structures were aligned by super-
mean square deviation (rmsd) of 1.0 Å and on from proteolysis and induces a structural rear- imposing 15 C-␣ positions located in three cor-
the corresponding portion of the primer tem- rangement in the active site conformation that responding active site ␤ strands. The secondary
plate bound to T7 DNAP with a rmsd of 1.9 Å. orients the catalytically essential residues, structure of only T7 RNAP is depicted as ribbons,
whereas T7 DNA polymerase residues are gray
Unlike duplex DNA bound to DNA poly- Asp537 and Asp812 (28) (Fig. 4), and thus cre- (40). Side chains in palm and fingers subdomains
merases, the heteroduplex shows all the char- ates the binding sites for magnesium ions and are red and blue, respectively. (A) The structures
acteristics of A form, including base pairs being ribonucleotide substrates. of ATP(I⫹4) (light green) and primer terminus
tilted and displaced in relation to the helix axis. Thus, it appears that the T7 lysozyme inter- (P⫹3) (green) at the active site with selected
action with T7 RNAP inhibits the binding of important interactions with T7 RNAP interac-
Substrate-Induced Conformational the COOH-terminal FOOT to the palm subdo- tions. (B) Comparison with T7 DNAP viewed from
Changes in T7 RNAP the top of (A). Residues Y639 and Y530 are in
main, thereby inhibiting the formation of an dramatically different orientations, whereas
The three DNA polymerases and HIV reverse active site that can stably bind NTP and RNA H784 and R425 interact with 2⬘-OH groups. The
transcriptase structural studies have shown a transcript in the initial stages of transcription. ribose groups for the ⫹3(P) and ⫹4(I) substrates
substantial rotation of the fingers domain upon This is consistent with an observed reduced in T7 RNAP are shown in green.

www.sciencemag.org SCIENCE VOL 286 17 DECEMBER 1999 2307

 RESEARCH ARTICLE
base pair seen at position ⫹1, 3 base pairs from tween the template base C(⫹1) and pppG base Comparison of the crystal structure of the
the primer terminus, is likely to end the hetero- at the 5⬘-end of the RNA product may be the transcribing RNAP with that of the T7 RPP
duplex (Fig. 2). The single hydrogen bond be- point at which the transcript starts to peel away complex (10) provides the first direct structural
from the template (Fig. 2B). In contrast, the evidence for accumulation or scrunching of nu-
A open cleft in DNA polymerases allows the prim- cleic acid chains in the catalytic active site. In
C+3 3'
er-template DNA homoduplex product to ex- order to position the template base at ⫹4 in the
C+2 GTP+2
tend out of the enzyme as a continuous duplex. active site opposite the incoming NTP(I), the
5' C+1
GTP+1 single-stranded template substantially changes
DNA Scrunching During Transcription its conformation within the active site pocket
Initiation and Abortive Cycling (Fig. 5). The template base at position ⫺1 ori-
T–1 5'
Ikeda and Richardson (9) have previously ents the ⫹1 and ⫹2 template positions to bind
W422 shown that, upon synthesis of pppGpGpG, the two guanosine ribonucleotide triphosphate sub-
pocket
footprint of T7 RNAP on the template strand is strates during the first step of initiation. After the
A–2 extended in the downstream direction from po- synthesis and translocation of a trinucleotide,
sition ⫹3 to position ⫹8. In contrast, the up- the base at ⫺1 has flipped out of the hydropho-
T–3
stream footprint remains constant, consistent bic pocket formed by Trp422, His300, and Tyr739
with the enzyme-retaining interactions with the and has assumed a new position (Fig. 4). In
B C+5
ATP+4 3' upstream promoter duplex during initiation and contrast, the structure of the protein does not
T+4
abortive cycling. Only after the synthesis of an change substantially during this process. The
5'
C+3 G+3
8- to 10-nucleotide (nt) RNA transcript (32) and polymerase and NH2-terminal domains of these
transition to processive RNA synthesis does T7 two complexes superimpose with a rmsd of 0.84

Downloaded from http://science.sciencemag.org/ on January 9, 2018

C+2 G+2 5' RNAP release the upstream promoter contacts. Å for all 862 C-␣ positions. Thus, it appears that
C+1 Thus, during abortive cycling, the downstream the template is accumulating in the active site as
GTP+1
contacts between the polymerase and the tem- RNA synthesis proceeds.
plate expand and contract while the upstream The volume of the active site pocket ap-
contacts are constant. These footprinting data pears to be sufficient to accommodate the
A–2 T–1 are consistent with three popular models for template required to synthesize a transcript
T–3 transcription initiation: “polymerase inchworm- that is 6 to 9 nt long. Its volume is defined by
ing,” in which the enzyme progressively extends the positions of the base pair at ⫹1 and the
C on the downstream template and nontemplate NH2-terminal domain and is ⬃3850 Å3 (Fig.
5' GTP(I) strands; “DNA scrunching,” in which the con- 5). This volume is sufficient to accommodate
3'
formation of the polymerase remains unchanged four further template bases (eight template
while the transcribed template strand accumu- bases in total) in the active site pocket, as-
template
lates within the active site pocket; and a third suming an approximate solvent content of
model in which the DNA slides forward and 30% and a Matthews coefficient, Vm (33),
back. equal to 1.9 to 2.0. Thus, the observed aver-
primer
age transcript length of abortive products (8
thumb
to 10 bases) (32) and the number of template
bases that can be accommodated appear to be
N-terminal correlated.

Fig. 4. Experimental evidence for scrunching of Escape from the Initiation Phase to
the template in the active site pocket. The struc- Processive Elongation
tures of the single-stranded template strand
(gray) (A) at initiation and (B) after synthesis of
We hypothesize that the escape from abortive
three nucleotides. (A) Before formation of the cycling to the processive elongation phase oc-
first 3⬘-5⬘ phosphodiester bond by T7 RNAP, the curs when the additional template can no longer
template strand at position–1 binds in a hydro- be accommodated in the active site pocket and
phobic pocket adjacent to the catalytic active site the transcript is long enough to bind tightly to
(1; white) (40). This orients the ⫹1 and ⫹2 the NH2-terminal domain. Once the electrostat-
template positions (dark gray) to create the bind-
ing sites for two GTP substrates (light green) in
ically positive pocket is filled with the template,
the incoming and priming positions. The template Fig. 5. The active site pocket and transcript exit either the message dissociates and synthesis
strand overhang (positions ⫹1 through ⫹3) and channel. A view of the active site with the protein begins again (abortive synthesis) or the duplex
the GTP substrates were modeled on the basis of represented as a surface. The polymerase thumb promoter releases from the protein to allow
the T7 DNA polymerase structure and the tran- subdomain (green), NH2-terminal domain ( yel- removal of the template. A longer transcript is
scribing T7 RNAP complex. (B) After synthesis of low), specificity loop (blue), and palm (purple) likely to favor the latter possibility, because a
pppGpGpG (dark green), the ⫹3 and ⫹4 posi- from the active site pocket. The deep pocket into
tions of the template strand are now located in which the single-stranded template (gray) accu-
transcript-induced change in the thumb confor-
priming and incoming positions. The template mulates is highly positively charged and adjacent mation might allow it to cover the cleft, thereby
base T–1 flips out of the small hydrophobic pock- to the exit channel for the RNA transcript. The forming a tunnel through which the elongating
et into the larger active site pocket. The structure remaining volume of the active site pocket is RNA would pass.
of template residues T–3 to T–1 has completely sufficient to accommodate a further four-tem-
changed. (C) Extension of a primer template be- plate-strand base (eight in total), enabling the
yond three base pairs taken from the T7 DNA RNA transcript (8-nt oligomer) to bridge the gap References and Notes
polymerase structure results in clashes with the to the exit channel. A possible binding site for the 1. R. Tjian, Philos. Trans. R. Soc. London Ser. B 351, 491
protein. Figure 4 was generated with the program transcript exiting the active site is shown by a (1996).
GRASP (41). Thumb subdomain and NH2-termi- green arrow. The fingers subdomain has been 2. R. L. Tracy and D. B. Stern, Curr. Genet. 28, 205
nal domain are green and yellow, respectively. removed to reveal the pocket. (1995).

2308 17 DECEMBER 1999 VOL 286 SCIENCE www.sciencemag.org

 RESEARCH ARTICLE
3. B. Hedtke, T. Börner, A. Weihe, Science 277, 809 (40 to 2.4 Å), Rmerge, of 0.078. Subsequent analysis 23. H. Huang, R. Chopra, G. L. Verdine, S. C. Harrison,
(1997). was performed with the CCP4 software suite (35). Science 282, 1669 (1998).
4. D. L. Ollis, P. Brick, R. Hamlin, N. G. Xuong, T. A. Steitz, Despite an R factor of 24% between these data and 24. S. H. Eom, J. Wang, T. A. Steitz, Nature 382, 278
Nature 313, 762 (1985). data from the T7 RPP complex, the template strand (1996).
5. C. A. Brautigam and T. A. Steitz, Curr. Opin. Struct. 5⬘-overhang and 5⬘-pppGpGpG-3⬘ RNA transcript 25. A. H. Wang et al., Nature 299, 601 (1982).
Biol. 8, 54 (1998). could clearly be seen in difference Fourier maps 26. T. A. Steitz, J. Biol. Chem. 274, 17395 (1999).
6. T. A. Steitz, S. J. Smerdon, J. Jäger, C. M. Joyce, Science calculated with the observed amplitudes from the 27. R. Sousa, Y. J. Chung, J. P. Rose, B.-C. Wang, Nature
266, 2022 (1994). two crystals and phases derived from the T7 RPP 364, 595 (1993).
7. P. H. von-Hippel, D. G. Bear, W. D. Morgan, J. A. coordinates. The initial crystallographic R factor for 28. P. A. Osumi-Davis, M. C. Aguilera, R. W. Woody,
McSwiggen, Annu. Rev. Biochem. 53, 389 (1984). the completed model was 0.40 (Rfree ⫽ 0.42). A. Y. M. Woody, J. Mol. Biol. 226, 37 (1992).
8. A. J. Carpousis and J. D. Gralla, Biochemistry 19, 3245 Model refinement (36) and rebuilding (37) reduced 29. L. P. Gardner, K. A. Mookhtiar, J. E. Coleman, Biochem-
(1980). the R factor to 0.22 (Rfree ⫽ 0.26) for all data at a istry 36, 2908 (1997).
9. R. A. Ikeda and C. C. Richardson, Proc. Natl. Acad. Sci. 40 to 2.4 Å resolution; F ⬎ 2␴(F). 30. J. Villemain and R. Sousa, J. Mol. Biol. 281, 793
U.S.A. 83, 3614 (1986). 14. The absence of ␣,␤-methylene ATP in these crystals (1998).
10. G. M. T. Cheetham, D. Jeruzalmi, T. A. Steitz, Nature may have resulted from its loss by diffusion when the 31. J. Lykke-Andersen and J. Christiansen, Nucleic Acids
399, 80 (1999). crystals were soaked in cryoprotection solution: The Res. 26, 5630 (1998).
11. S. Doublié, S. Tabor, A. M. Long, C. C. Richardson, T. Michaelis constant Km for the interaction between T7 32. C. T. Martin, D. K. Muller, J. E. Coleman, Biochemistry
Ellenberger, Nature 391, 251 (1998). RNAP and the NTP in the incoming binding site is ⬃0.1 27, 3966 (1988).
12. D. Jeruzalmi and T. A. Steitz, EMBO J. 17, 4101 mM, determined for optimized in vitro transcription 33. B. M. Matthews, J. Mol. Biol. 33, 491 (1968).
(1998). conditions (38), and ␣,␤-methylene ATP has poor sol- 34. A. G. W. Leslie, Joint CCP4 and ESF-EACMB Newsletter
13. Crystals of the quaternary complex reported here were ubility at high concentrations of polyethyleneglycol/ Protein Crystallography No. 26 (Daresbury Laborato-
grown similarly to those of the binary complex de- propyleneglycol solution. Alternatively, however, the ry, Warrington, UK, 1992).
scribed previously (10); T7 RNAP was mixed (0.3 mM) conformation of the enzyme in this crystal form may 35. Collaborative Computational Project No. 4, Acta
with a T7 promoter (0.35 mM), in the presence of play an important role in determining the captured Crystallogr. D50, 760 (1994).
ribonucleotides GTP, uridine 5⬘-triphosphate, cytidine transcription intermediate. Several other crystals that 36. A. T. Brünger et al., Acta Crystallogr. D54, 905 (1998).
5⬘-triphosphate, and ␣,␤-methylene ATP (4 mM each) diffracted to a resolution of 2.8 Å clearly have bound 37. T. A Jones, J. Y. Zou, S. W. Cowan, M. Kjeldgaard, Acta
and 15 mM MgCl2. The promoter was constructed by magnesium ions and partially occupied ␣,␤-methylene Crystallogr. A47, 110 (1991).
annealing together two synthetic oligonucleotides that ATP at the ⫹4 position in the active site. In this in- 38. J. L. Oakley, R. E. Strothkamp, A. H. Sarris, J. E.

Downloaded from http://science.sciencemag.org/ on January 9, 2018

consisted of a 22-nt template strand of sequence 5⬘- stance, the template strand and residue Tyr639 appear Coleman, Biochemistry 18, 528 (1979).
CTCCCTATAGTGAGTCGTAT TA-3⬘ and a 17-nt non- to be poorly ordered in electron density maps. 39. M. Carson, J. Appl. Crystallogr. 24, 958 (1991).
template strand of sequence 5⬘-TAATACGACTCAC- 15. L. A. Kohlstaedt, J. Wang, J. M. Friedman, P. A. Rice, T. 40. Single-letter abbreviations for the amino acid resi-
TATA-3⬘. This produced a 17– base pair duplex and a A Steitz, Science 256, 1783 (1992). dues are as follows: D, Asp; E, Glu; F, Phe; G, Gly; H,
5-nt template overhang. In this way, the T7 RNAP 16. J. Wang et al., Cell 89, 1087 (1997). His; K, Lys; M, Met; Q, Gln; R, Arg; W, Trp; and Y, Tyr.
synthesized a 3-nt RNA and stopped at position ⫹4 of 17. Y. Li, S. Korolev, G. Waksman, EMBO J. 17, 7514 41. A. Nicholls, R. Bharadwaj, B. Honig, Biophys. J. 64,
the template strand before crystallization. Data were (1998). A166 (1993).
collected at Brookhaven National Laboratory (station 18. G. Gao, M. Orlova, M. M. Georgiadis, W. A. Hendrick- 42. We thank J. Ippolito, D. Kennedy, and J. Wang for
X25), using x-rays of wavelength 1.10 Å, which were son, S. P. Goff, Proc. Natl. Acad. Sci. U.S.A. 94, 407 discussions and P. Eatherton for assistance in the
recorded with a Brandeis B4 charge-coupled device (1997). preparation of this manuscript. This research was
detector and reduced with the program MOSFLM (34). 19. J. R. Kiefer, C. Mao, J. C. Braman, L. S. Beese, Nature supported by NIH grant number GM-22778 ( T.A.S.).
The crystals are orthorhombic space group P21212 391, 304 (1998). The complete refined coordinates of the T7 RNAP
with cell dimensions of a ⫽ 221.2 Å, b ⫽ 73.6 Å, 20. R. Sousa and R. Padilla, EMBO J. 14, 4609 (1995). transcribing complex and the observed structure fac-
and c ⫽ 80.9 Å. The number of unique reflections 21. M. Astatke, N. D. F. Grindley, C. M. Joyce, J. Biol. tor amplitudes have been deposited in the Protein
recorded was 51,537 (97.8% complete, with 40 to Chem. 270, 1945 (1995). Data Bank with numbers 1QLN and R1QLNSF.
2.4 Å resolution), with an overall multiplicity of 3.8 22. V. O. Rechinsky, D. A Kostyuk, V. L. Tunitskaya, S. N.
and a crystallographic R factor for merging all data Kochetkov, FEBS Lett. 306, 129 (1992). 5 October 1999; accepted 15 November 1999

REPORTS

Phase-Coherent Amplification amplify a matter wave by using an appropriate

gain medium. Bose-Einstein condensation (the

of Matter Waves macroscopic occupation of a single quantum

state) in dilute alkali gases (2, 3) has the poten-
tial of being such a gain medium because it is
Mikio Kozuma,1 Yoichi Suzuki,1 Yoshio Torii,2 Toshiaki Sugiura,1 an ideal source of highly coherent matter
Takahiro Kuga,1 E. W. Hagley,3 L. Deng3 waves. What has long since happened for the
optical laser is now becoming a reality for
Phase-coherent matter-wave amplification was demonstrated using Bose- matter waves. In fact, many passive matter-
Einstein–condensed rubidium-87 atoms. A small seed matter wave was created wave elements have already been demonstrat-
with coherent optical Bragg diffraction. Amplification of this seed matter wave ed. For example, Bragg diffraction (4) of a
was achieved by using the initial condensate as a gain medium through the Bose-Einstein condensate (BEC) by a moving,
superradiance effect. The coherence properties of the amplified matter wave, optical standing wave (5) can be used to coher-
studied with a matter-wave interferometer, were shown to be locked to those ently diffract any fraction of a BEC into a
of the initial seed wave. The active matter-wave device demonstrated here has selectable momentum state, making it an ideal
great potential in the fields of atom optics, atom lithography, and precision mirror or beam splitter. These passive elements
measurements. were key in the development of a highly effi-
cient (nearly 100% contrast) Mach-Zehnder
Advances in laser science and technology active ones. By passing through such an ac- BEC interferometer (6, 7). In addition, most
have been supported by the development of a tive device, a weak laser beam can be ampli-
variety of optical elements that can be cate- fied while maintaining its phase coherence: 1
Institute of Physics, University of Tokyo, 3-8-1
gorized as either passive or active. Mirrors, the mechanism of lasing itself. Komaba, Meguro-ku, Tokyo 153-8902, Japan. 2De-
partment of Physics, Gakushuin University, Mejiro
beam splitters, and polarizers are typical ex- In the field of atom optics it has long been 1-5-1, Toshima-ku, Tokyo 171-8588, Japan. 3Physics
amples of passive elements, whereas devices speculated, in direct analogy with optical am- Lab, National Institute of Standards and Technology,
such as optical pulsed-dye amplifiers (1) are plifiers, that it should be possible to coherently Gaithersburg, MD 20899, USA.

www.sciencemag.org SCIENCE VOL 286 17 DECEMBER 1999 2309

Structure of a Transcribing T7 RNA Polymerase Initiation Complex
Graham M. T. Cheetham and and Thomas A. Steitz

Science 286 (5448), 2305-2309.

DOI: 10.1126/science.286.5448.2305

Downloaded from http://science.sciencemag.org/ on January 9, 2018

ARTICLE TOOLS http://science.sciencemag.org/content/286/5448/2305

REFERENCES This article cites 35 articles, 11 of which you can access for free
http://science.sciencemag.org/content/286/5448/2305#BIBL

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive licensee
American Association for the Advancement of Science. No claim to original U.S. Government Works. The title Science is a
registered trademark of AAAS.