Structural Basis of Transcription Initiation: RNA Polymerase Holoenzyme at 4 Å Resolution Katsuhiko S. Murakami, Shoko Masuda, and Seth A. Darst

Crystallization. Transcriptionally active Taq RNAP holoenzyme was reconstituted in vitro by combining native core RNAP (1) and recombinant Taq ^A expressed and purified from Escherichia coli (2). Holoenzyme was reconstituted by mixing a 1.2-fold molar excess of ^A with core RNAP in 10 mM Tris-HCl, pH 8.0, 0.2 M NaCl, 1 mM DTT, 0.1 mM EDTA, 15% glycerol. The mixture was then incubated at 30°C for 30 min. Initial crystal trials yielded small, needle-like crystals, which could be improved to extremely thin plates using microseeding techniques. Examination of the protein present in the crystals by SDS-PAGE revealed in situ degradation of ^A by unkown contaminating proteases (3). N-terminal sequencing showed that the ^A polypeptide was cleaved at its N-terminus up to residue 92. This shortened ^A fragment (residues 92-438), which lacks the entire region 1.1 (1.1^A; ref. 4), was sub-cloned and used for holoenzyme reconstitution. Holoenzyme with 1.1^A, which was indistinguishable from wild-type holoenzyme in promoter DNA binding and abortive transcription activity assays (see below), was used for subsequent crystallization experiments.

Crystals belonging to the primitive space group P2₁ (a = 155.0 Å, b = 271.2 Å, c = 155.3 Å, = 91.4°), with two 430 kDa holoenzyme molecules per asymmetric unit, were grown by vapor diffusion. Holoenzyme solution (15 mg/ml) was mixed with the same volume of crystallization solution containing 0.1 M Hepes-NaOH, pH 8.0, 3 M sodium formate, and crystal seeds. The mixture was incubated as a hanging drop over crystallization solution but containing 2.4-2.6 M sodium formate. Crystals grew in 2 months to typical dimensions of 0.3 x 0.1 x 0.2 mm at 22° C. For cryocrystallography, crystals were presoaked in crystallization solution, followed by four, 10 min soaks in crystallization solution containing increasing concentrations of sodium formate to a final concentration of 6 M. Crystals were then flash frozen by dunking in liquid ethane held at liquid nitrogen temperature. The crystals grew as thin rectangular plates about 0.2 mm x 0.1 mm x 0.02 mm. Nevertheless, diffraction to 4 Å resolution was obtained at suitably bright synchrotron beam lines (Table 1). Diffraction data were processed using DENZO and SCALEPACK (5).

Structure determination. An excellent electron density map at 4 Å resolution was obtained by a combination of molecular replacement, single isomorphous replacement (SIR), and density modification (Fig. 1A). Using a search model derived from the Taq RNAP holoenzyme complex with fork-junction DNA (6), a molecular replacement solution was obtained. Using diffraction data from 15 to 5 Å resolution, rotation and translation searches were performed using CNS (7). After a solution for one molecule in the asymmetric unit was obtained, its position was fixed and another search was performed. The position for a second molecule was found that did not overlap with the first. The molecular replacement phases were improved by phase combination with SIR phases from a Ta₆Br₁₄ derivative (text Table 1). A spherically averaged model for the structure of the Ta-cluster was used for heavy atom refinement and phasing calculations as described (8). The four-Gaussian fit to the scattering curve yielded the following coefficients (useful to 4.5 Å-resolution): c = 17976.4, a1 = 27.1519, b1 = 673.365, a2 = 6696.77, b2 = 22.1632, a3 = 4994.74, b3 = 22.1958, a4 = -29267.1, b4 = 6.99315. The electron density map was further improved by density modification and non-crystallographic symmetry averaging using CNS (7).

Comparison of the molecular replacement solution with the resulting electron density map revealed significant deviations, indicating that model bias was effectively removed by the inclusion of the Ta-cluster phases. The structure was divided into domains (Table 1) and rigid body refinements were performed. Adjustments to the model were then made by hand to better fit the electron density map. Connecting segments of ^A that were not modeled in the original lower resolution DNA complex (most significantly the extended polypeptide chain and -helix comprising conserved region 3.2), and segments of the RNAP ' subunit that were disordered and not modeled in the core RNAP structure (1, 9), were added (text Fig. 2). Several solvent-exposed loops within the RNAP subunit were missing electron density, presumably because of disorder within the crystals, and were removed. The Zn²⁺-ion was placed according to the ethyl mercury thiosalicylate Fourier difference peak from the holoenzyme-DNA complex (6). Because of the limited resolution of the diffraction data, side chain density was not always clear, but main chain density over all of the modeled portions was unambiguous (Fig. 1A). Because of this, the polypeptide segments added to the model were generally built as poly-alanine. Phases calculated from this new model were combined with the original Ta-cluster SIR phases to generate a new electron density map, but inspection of this map revealed that further changes to the final model (text Table 1) were not warranted. Structural calculations were performed using CNS (7) with very tight non-crystallographic symmetry restraints between the two molecules in the asymmetric unit at all times. Map inspection and model building was done using O (10). The current, unrefined model has an R factor of 0.345 (R_free = 0.397).

Transcription assays. E. coli core RNAP (40 nM), ⁷⁰ or ⁷⁰C (160 nM), and -10 con promoter DNA fragment from -150 to +71 (40 nM) were mixed in 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10 mM MgCl₂, 0.1 mM EDTA, 1 mM DTT, 25 g/ml BSA, and incubated for 10 min at 37°C. RNA synthesis was initiated by the addition of NTPs (GTP, 1 mM; ATP and CTP, 100 M; UTP, 10 M; [-³²P]UTP, 1 Ci) and the samples were incubated for 10 min at 37°C. Radiolabeled RNA products were analyzed by electrophoresis on a 20% polyacrylamide/7 M urea gel. The -10 con promoter (11) is an optimized extended -10 promoter that does not require regions 3.2 or 4 for transcription activity (3, 12).

References and Notes

1. G. Zhang et al., Cell 98, 811-824 (1999).
2. L. Minakhin, S. Nechaev, E. A. Campbell, K. Severinov, Journal of Bacteriology 183, 71-76 (2001).
3. E. A. Campbell et al., Molecular Cell, in press (2002).
4. M. Lonetto, M. Gribskov, C. A. Gross, Journal of Bacteriology 174, 3843-3849 (1992).
5. Z. Otwinowski, W. Minor, Methods in Enzymology 276, 307-326 (1997).
6. Accompanying manuscript.
7. P. D. Adams, N. S. Pannu, R. J. Read, A. T. Brunger, Proceedings of the National Academy of Sciences USA 94, 5018-5023 (1997).
8. J. Fu et al., Cell 98, 799-810 (1999).
9. E. A. Campbell et al., Cell 104, 901-912 (2001).
10. T. A. Jones, J.-Y. Zou, S. Cowan, M. Kjeldgaard, Acta crystallographica A47, 110-119 (1991).
11. T. Gaal et al., Molecular Microbiology 42, 939-954 (2001).
12. A. Kumar et al., Journal of Molecular Biology 232, 406-418 (1993).
13. T. M. Gruber, D. A. Bryant, Journal of Bacteriology 179, 1734-47 (1997).

Supplemental Table 1. Core RNAP mobile modules.
Module	Subunit	Residues (total)	Maximum C atom displacement (Å) (residue number)
Core	I	6-229	-
	II	6-225
		2-21, 131-141, 325-335, 393-704, 829-1057
	'	620-1435, 1460-
		2-93
		(1975)
1		22-130, 336-392	12.9
		(166)	(365)
2		142-324	6.3
		(183)	(245)
flap		705-828	13.0
		(124)	(773)
Clamp		1058-1116	14.3
	'	3-619, 1436-1459	('561)
		(700)

Supplemental Table 2. -Core RNAP interaction areas.
		Contact Area (Å2)			Contact Area/Total Area (%)			Target Contact Area/Total Contact Area (%)
		Target			Target			Target
Segment	Total Area (Å2)	'		Core RNAP	'		Core RNAP	'
2	11464	2670	76	2745	23.3	0.7	23.9	97.2	2.8
3	5222	638	1691	2329	12.2	32.4	44.6	27.4	72.6
3.2	1924	391	1182	1573	20.3	61.4	81.8	24.9	75.1
4	5890	351	1233	1584	6.0	20.9	26.9	22.1	77.9
Total	24500	4049	4182	8231	16.5	17.1	33.6	49.1	50.8

Figure Legends

Supplemental Figure 1. Taq RNAP holoenzyme - electron density and structure. A) Stereo views of regions 2.2-2.4 interaction with the 'cc (top), and the ₄ interaction with the flap-tip helix (bottom). The -carbon backbone of is colored orange, ' pink, cyan. Electron density, calculated using F_o coefficients, is shown (blue net, contoured at 1.2), and was computed using molecular replacement and SIR phases combined, followed by density modification. B) (top) The thick bar represents the 438 amino acid Taq ^A primary sequence, with amino acid numbering below. Evolutionarily conserved regions are labeled and color-coded (3, 4, 13). The domain architecture of the segment in the holoenzyme crystals is indicated above the bar, with rectangles indicating the three structured domains, and lines representing the flexible linkers. Disordered segments within the holoenzyme structure are indicated by dashes. (middle) Stereo view of the Taq ^A structure as it occurs in the RNAP holoenzyme, with the conserved regions color-coded as above. -helices are shown as cylinders. The RNAP active site Mg²⁺ is shown as a magenta sphere. Disordered segments of the polypeptide are indicated by dashed lines. (bottom) Stereo view of the Taq RNAP holoenzyme structure. Core RNAP components are shown as molecular surfaces, color-coded as follows: I, II, , grey; , cyan; ', pink.

Medium version | Full size version

Supplemental Figure 2. Amino acid sequence alignment between Taq ^A and E. coli ⁷⁰. Shown is a sequence alignment within conserved regions 1.2-4.2 of primary or group 1 's from Taq and E. coli factors (3, 4, 13). The sequences are presented in one-letter amino acid code and are identified by the species in the rightmost column. Numbers at the beginning of each line indicate amino acid positions relative to the start of each protein sequence. Numbers at the top indicate the amino acid positions in Taq ^A/E. coli K-12 ⁷⁰ (Taq/E. coli). Amino acid identity in >50% of a full alignment of 53 group 1 's (3, 13) is indicated by a black background, amino acid similarity by a blue background. Groups of residues considered similar are: ST, RK, DE, NQ, FYW, and ILVM. Gaps are indicated by dashes The domain architecture of the segment in the holoenzyme crystals is indicated above the bar, with rectangles indicating the three structured domains, and lines representing the flexible linkers. Disordered segments within the holoenzyme structure are indicated by dashes. The histogram at the top represents the level of sequence identity at each position in the full alignment. Sequence identity of 100% is represented by a tall red bar, less than 20% is represented by a small blue bar, intermediate levels are represented by orange, light green, and light blue bars.

Medium version | Full size version