Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Jan 1.
Published in final edited form as: Methods Mol Biol. 2016;1320:67–76. doi: 10.1007/978-1-4939-2763-0_6

Use of the U1A protein to facilitate crystallization and structure determination of large RNAs

Adrian R Ferré-D’Amaré 1
PMCID: PMC5178131  NIHMSID: NIHMS836314  PMID: 26227038

Summary

The preparation of well-ordered crystals of RNAs with complex three-dimensional architecture can be facilitated by engineering a binding site for the spliceosomal protein U1A into a functionally and structurally dispensable stem-loop of the RNA of interest. Once suitable crystals are obtained, the U1A protein, of known structure, can be employed to facilitate preparation of heavy atom or anomalously scattering atom derivatives, or as a source of partial model phases for the molecular replacement method. Here, we describe the methods for making U1A preparations suitable for cocrystallization with RNA. As an example, the cocrystallization of the tetracycline aptamer with U1A is also described.

Keywords: Crystallization, crystallizability, spliceosomal protein, RNA biding domain (RBD), protein purification

1. Introduction

Determination of the three-dimensional structures of large RNAs with complex architectures by X-ray crystallography requires the preparation of well-ordered crystals. Although the first RNA structure to be successfully determined, that of yeast tRNAPhe, employed crystals of an intact molecule isolated from its biological source, the majority of subsequent RNA structure determinations have employed either synthetic or in vitro transcribed molecules. This makes it possible to engineer for increased “crystallizability” the RNAs whose structures are of interest. Because of the high thermodynamic stability of local structure, which is dominated by the A-form double helix, the periphery of many RNAs can be engineered to alter molecular surface properties without adversely affecting their overall folds or active site structures (1). One approach that has been particularly successful since its invention in 1998 (2) is to replace a solvent-exposed, functionally dispensable stem-loop of an RNA of interest with the cognate binding site for the spliceosomal protein U1A (3). The engineered RNA is complexed with the strongly basic RRM-I domain of the protein (hereafter “U1A”), and the RNA:U1A complex is subjected to crystallization. It has been found empirically that cocrystallization of such a complex often results in better-ordered crystals than crystallization of the parental form of the RNA in the “naked” state. Moreover, if good crystals are obtained, the U1A protein can be used to facilitate structure determination, either by using it to incorporate heavy or anomalously scattering atoms into the crystals for experimental phasing [for instance (3, 4)], or by providing partial model phases for molecular replacement [for instance (5, 6)]. The use of U1A as a “crystallization module” in several structure determination projects has been reviewed elsewhere (612). Here, we provide detailed methods the preparation of recombinant U1A protein suitable for use in facilitating RNA crystallization. Because conditions vary from RNA to RNA, it is not possible to provide a general crystallization protocol. Instead, we describe procedures for the in vitro transcription of an engineered tetracycline aptamer RNA and its cocrystallization with U1A (13) as an example of an RNA structure determination that was facilitated by U1A, both for the preparation of well-ordered crystals and for experimental phase determination.

2. Materials

Except where noted, prepare all solutions using deionized, DEPC-treated water (see Note 1). All solutions used for chromatography need to be filtered through 0.2 μM filters. During purification, protein-containing solutions should be kept on ice or at 4°C. Purified U1A should be stored at −80°C, but can be used at room temperature. Personal protective equipment (lab coat, safety goggles, gloves) should be worn where appropriate, and all waste disposal regulations must be followed.

2.1 U1A Protein Expression and Purification Components

  1. Lysis buffer: 20 mM HEPES-KOH pH 7.5, 10% (v/v) glycerol, 100 mM KCl, 0.05-0.1 trypsin inhibitor unit of aprotinin per liter, 5 μg/ml leupeptin, 0.0005% (v/v) Triton X-100, 0.5 mM EDTA, 0.5 mM phenylmethysulfonyl fluoride (PMSF, see Note 2). It is not necessary to use DEPC-treated water for this solution, as it will encounter abundant bacterial nucleases.

  2. Neutralized Polyethyleneimine (PEI): a 5% (w/v) neutralized PEI solution should be prepared as described by Burgess (14). After sterilization by filtering through 0.2 μm filters, this solution can be stored long-term at 4°C.

  3. Saturated ammonium sulfate: a saturated solution suitable for protein purification can be prepared by dissolving 800 g of solid ammonium sulfate in sufficient water to make a 1 liter solution (heating gently, do not exceed 40 °C) and filtering through a 0.2 μM pore filter while warm. Upon cooling to room temperature in a closed bottle, excess ammonium sulfate will crystallize out of the solution. These crystals can be left in the bottle, but care should be taken to only use the supernatant saturated solution.

  4. Cation-exchange chromatography buffers: Buffer A comprises 100 mM KCl, 25 mM HEPES-KOH pH 7.5, 0.5 mM EDTA and 0.5 mM PMSF. Buffer B has the same composition, except that it contains 0.5 M KCl.

  5. Hydroxyapatite chromatography buffers: Buffer C comprises 10 mM potassium phosphate, pH 7.5 and 50 mM KCl. Buffer D has the same composition, with the addition of 0.5 M ammonium sulfate.

  6. Storage buffer: 10 mM Hepes-KOH pH 7.5, 0.1 mM EDTA.

2.2 RNA Transcription and Purification Components

  1. Neutralized ribonucleoside triphosphates (rNTPs): for economy, lyophilized sodium salts of rNTPs can be purchased. These should be dissolved in water and neutralized with NaOH prior to use (it is sufficient to use pH indicator paper with a legibility of one pH unit). In order to make 100 mM stock solutions, one gram of the salt should be brought up to a final volume of 16.5, 17.1, 16.5 and 17 ml for ATP, CTP, GTP and UTP, respectively. After filtering through 0.2 μm syringe filters, these solutions can be aliquoted and stored at −20°C until use.

  2. Transcription buffer: specific conditions need to be optimized for each RNA sequence. As a starting point, transcription reactions contain 30 mM Tris-HCl pH 8.1, 25 mM MgCl2, 0.01% (v/v) Triton X-100, 2 mM spermidine hydrochloride, 2.5 mM each of the four NTPs, 1 mM DTT, 0.05 g/l T7 RNA polymerase, and 1 unit/ml of inorganic pyrophosphatase (see Note 3).

  3. Tris-Borate-EDTA buffer. A 10x stock is prepared by mixing 432 g of Tris base, 220 g of Boric acid, 160 ml of 0.5 M sodium EDTA (pH 8.0) and sufficient water to make 4 liters. Solubilization is hastened by autoclaving for 30–40 minutes.

  4. Denaturing polyacrylamide gels: these are prepared by dissolving urea with water, 10x TBE buffer, and acrylamide:bisacrylamide solution. We find that a commercially available 40% total, 29:1 acrylamide:bisacrylamide solution is adequate for most RNA separations. The acrylamide content of the gel needs to be varied depending on the size of the RNAs being purified. The total volume of gel solution needed will depend on the electrophoresis apparatus used. Table 1 indicates the amounts of urea, acrylamide solution and water needed for different gel compositions. To initiate polymerization, the gel mixture is supplemented with 0.001% (v/v) of tetramethylethylenediamine (TEMED) and 0.002% (v/v) of a 50% (w/v) aqueous solution of ammonium persulfate. 50% aqueous ammonium persulfate can be prepared ahead of time, and stored indefinitely in small aliquots at −20°C until use.

Table 1.

Preparation of gel mixture for polyacylamide electrophoresis1.

Gel percentage Volume of 40% acrylamide stock solution (mL) Purified water (mL)
4 10 43
6 15 37
8 20 31
10 25 27
12 30 16
15 38 8
20 50 0
1

Quantities given are for preparing 100 mL of gel mixture. For this amount, 48 grams of solid urea and 10 ml of 10x TBE need to be included in addition to the acrylamide and water volumes tabulated.

3.3 U1A-RNA Cocrystallization Components

  1. RNA preincubation buffer: 50 mM Tris-HCl pH 7.5, 5 mM MgCl2, 0.5 mM 7-chlorotetracycline, and 0.25 mM spermine hydrochloride (final concentration). A 5x stock solution can be prepared, and kept at −20°C in the dark.

  2. Crystallization reservoir: 50 mM Hepes-KOH pH 7.0, 20 mM MgCl2, and 12.5–15% (w/v) PEG 8000.

3. Methods

3.1 U1A Protein Purification

  • 1

    Resuspend E. coli from a 9 liter U1A-RBD growth and induction (see Note 4) in 75 ml of ice-cold lysis buffer. Avoid vigorous shaking. Bacterial clumps can be broken down by repeatedly forcing the suspension through a 25 or 50 ml serological pipette. Any residual clumps can be removed using a double thickness of cheesecloth, or a fine metal sieve (such as a repurposed household metal mesh coffee filter).

  • 2

    Lyse bacteria using a French press, EmulsiFlex, or similar high-pressure cell disruption device. Add PMSF immediately prior to lysis. It is important to avoid heating the resuspended bacteria; therefore, sonication is not recommended. If a high-pressure device is not available, several cycles of freeze-thawing complemented with lysozyme can be employed, but protein yield will decrease. Add sufficient PMSF immediately after lysis to a nominal final concentration of 1 mM.

  • 3

    Clarify the lysate by centrifugation at 4°C for 30′ at 75,000 x g (see Note 5).

  • 4

    Transfer the supernatant to fresh centrifuge tubes, and add 1/10 volume of neutralized 5% PEI solution. Mix by gentle inversion and incubate 10′ on ice.

  • 5

    To remove PEI-bound nucleic acids, centrifuge at 4°C for 30′ at 75,000 x g. Discard the pellet.

  • 6

    Place supernatant in a suitable container (e.g. an Erlenmeyer flask) on ice. While stirring gently with a magnetic stirrer, add dropwise sufficient saturated ammonium sulfate solution to a final concentration of 35% saturation (not w/v). Stir for 15′ after addition is complete.

  • 7

    To remove contaminant proteins, centrifuge at 4°C for 30′ at 38,000 x g. Discard the pellet.

  • 8

    To the supernatant, add sufficient solid ammonium sulfate to achieve a final concentration of 80% saturation (not w/v). This requires the addition of 316 g of solid ammonium sulfate per liter of 35% saturated supernatant. Stir magnetically at room temperature until complete dissolution of the salt.

  • 9

    Centrifuge at 4°C for 30′ at 38,000 x g. Discard the supernatant.

  • 10

    Gently dissolve the pellet in the smallest possible volume of a solution of the same composition as Buffer A (add PMSF immediately prior to use), but with 50 mM KCl. Dialyze overnight using 3,000 Da nominal molecular weight cutoff membranes against 20x volume of Buffer A.

  • 12

    After dialysis, eliminate precipitated proteins by centrifuging at 4°C for 30′ at 75,000 x g. Decant supernatant taking care not to dislodge the pellet. Discard pellet.

  • 13

    Equilibrate a 30 ml bed-volume SP-Sepharose Fast Flow (GE Biosciences) column in Buffer A. Load dialyzed protein and run a 10-column volume gradient to 100% Buffer B. U1A elutes at a conductivity approximately equivalent to 300 mM KCl. Concentrate U1A peak by ultrafiltration.

  • 14

    Further purify U1A by size-exclusion chromatography on a Superdex-75 PG (GE Biosciences) column equilibrated and run in Buffer A. U1A elutes at a volume consistent with a molecular mass of ~11 kDa.

  • 15

    At this point in the procedure, it is advisable to clean the chromatography system to eliminate nucleases. In our laboratory, this is done by running a solution of 0.5 M NaOH through the entire system and then flushing it with DEPC-treated water (see Note 6). All further solutions employed in the procedure are made with DEPC-treated water and buffer components are crystallization-grade.

  • 16

    Equilibrate a 10 ml bed-volume CHT-I hydroxyapatite column (BioRad) in Buffer C. To the protein, add enough potassium phosphate (pH 7.5) solution to a final concentration of 10 mM. Load protein on equilibrated column and run a 20-column volume gradient to 100% Buffer D. The protein elutes at a conductivity approximately equivalent to 250 mM ammonium sulfate (see Note 7).

  • 17

    Dialyze protein against 20 volumes of storage buffer. Concentrate by ultrafiltration using 3000 Da molecular weight-cutoff membranes, quantify and store at −80°C (see Note 8).

3.2 Transcription and Purification of Tetracycline Aptamer

  1. Prepare double-stranded DNA template (see Note 9) by the polymerase chain reaction (PCR), as described (13).

  2. Prepare transcription reactions by adding 1/10 volume of PCR reaction to the transcription reaction. Perform transcription at 37°C for 2–4 hours.

  3. Add two volumes of absolute ethanol (chilled to −20°C) to the reaction. Mix well by inversion and incubate at −20°C for 30 minutes.

  4. Centrifuge at 7000 x g for 10′.

  5. Decant supernatant. Rinse pellet (without dislodging) with −20°C 70% (v/v) aqueous ethanol. Decant ethanol.

  6. Remove most of the ethanol from the pellet by placing it under vacuum (a vacuum aspirator or a centrifugal drier-used without the rotor- are sufficient) for 5–10 minutes.

  7. Dissolve pellet in the smallest possible volume of 250 mM sodium EDTA (pH 8.0). Add 0.9 volumes of formamide and 0.1 volume of 10 x TBE buffer. Heat to 50°C for 5 minutes.

  8. Load RNA onto pre-run 8M urea polyacrylamide gel, run until separation is achieved (see Note 10). Tracking dyes (for instance Bromophenol Blue and Xylene Cyanol) dissolved in neat formamide can be loaded into a small parallel lane.

  9. Visualize band by briefly shadowing the gel with a hand-held ultraviolet light while the slab gel is kept over a fluorescent indicator-containing TLC plate (see Note 11).

  10. Excise the portion of the gel containing the RNA with a sterile scalpel or razor blade, transfer to a sterile polypropylene centrifuge tube, and crush using a sterile spatula. Add approximately 25 ml DEPC-treated water per ml of excised gel.

  11. Gently mix or tumble overnight at 4°C

  12. Filter the gel-water suspension through 0.2 μm filter, discard gel fragments.

  13. Using either a centrifugal or a gas-pressure driven ultrafiltration device equipped with a 3000 Da nominal molecular weight cut-off membrane, concentrate eluted RNA to a small volume. Dilute at least 25-fold with 1 M KCl, and concentrate again. Then, sequentially, dilute 25-fold or more with DEPC-treated water and concentrate three times, so that the final concentration of KCl is less than 1 mM. Concentrated RNA can be stored at 4°C until use.

3.3 Cocrystallization of U1A and the Tetracycline Aptamer

  1. Incubate RNA and a stoichiometric amount of U1A in preincubation buffer at a final macromolecular complex concentration of 0.25 mM for 20′ at 37°C.

  2. Set up hanging drop vapor-diffusion experiments by placing 0.5 ml of reservoir solution in the reservoir. Then, mix 1 μl of this solution with 1 μl of preincubated RNA: protein: chlorotetracycline mixture to make the hanging drop (see Note 12). Seal the system and incubate in the dark at 21°C for 2–3 weeks. Cube-shaped crystals appear within 4 weeks and grow to maximum dimensions of 0.15 μm3 over two additional weeks.

  3. For diffraction experiments, crystals can be cryoprotected by (see Note 13) by exchanging the mother liquor over the course of 10–15 minutes with a solution comprising 50 mM Hepes-KOH (pH 7.0), 20 mM MgCl2, 15% (w/v) PEG 8000, 0.5 mM sperimine, 0.5 mM 7-chlorotetracyline, and 30% (v/v) glycerol. Crystals are then mounted in nylon loops and flash-frozen by plunging into liquid nitrogen.

Acknowledgments

We thank K. Nagai (Medical Research Council Laboratory of Molecular Biology, Cambridge, UK) for generously sharing U1A expression plasmids. This work was supported by the Intramural Research Program of the NIH, National Heart, Lung and Blood Institute.

Footnotes

1

. Although most laboratory-grade purification systems will deliver high resistivity (18.2 MΩ cm−1 at 25°C) deionized water, this may still contain ribonucleases (RNases). Residual RNases can be destroyed by treatment with the alkylating agent diethylpyrocarbonate (DEPC). Unreacted DEPC can in turn be destroyed by autoclaving. Mix 1 ml of DEPC with 4 liters of purified water in a bottle with a tightly sealing cap. Shake vigorously for ten seconds. Loosen the cap, cover loosely with aluminum foil and autoclave for at least 30 minutes. DEPC is highly toxic. It should be handled only in a chemical fume hood and with appropriate personal protective equipment. Unreacted DEPC is converted into ethanol and carbon dioxide through autoclaving.

2

. PMSF is a powerful neurotoxin. It is important to avoid any direct contact with PMSF-containing solutions. Because PMSF is slowly destroyed by water, it needs to be dissolved freshly prior to use, or dissolved in an anhydrous solvent and kept free of water. A convenient method to achieve this is to dissolve the PMSF in dry isopropyl alcohol and to store it over 3 Å molecular sieves. Molecular sieves need to be activated by heating to 200 °C for ~15 hours (or more) in a vacuum oven, and then allowing them to cool to room temperature in a desiccator prior to use. Isopropyl alcohol itself (which forms an azeotrope with water) can be dried by stirring over activated molecular sieves.

3

. Because magnesium pyrophosphate has very low solubility, accumulation of pyrophosphate during transcription leads to sequestration of magnesium ion in a precipitate. Since RNA polymerase requires free magnesium ion for activity, this decreases transcription yield. To overcome this, inorganic pyrophosphatase can be added to the reaction (9).

4

. Expression and induction of U1A are described elsewhere (8, 15).

5

. Longer centrifugation times will be needed if a preparative centrifuge/rotor combination capable of these centrifugal forces is not available.

6

. It will be necessary to first establish if the chromatography system being used tolerates exposure to 0.5 M NaOH. Otherwise, other cleaning methods will be needed. Personal protective equipment must be worn at all times when handling sodium hydroxide solutions.

7

. We find that for optimal separation, no more than 15 mg of U1A should be loaded on a 10 ml bed-volume CHT-I column. All solutions that come into contact with the CHT-I column need to contain 10 mM phosphate.

8

. U1A protein has a molecular mass of 11.3 kDa. A 1 g/l solution of the wild-type and double mutant (15) proteins produce an absorbance at 280 nM of 0.46 and 0.34, respectively. Although nucleases can prove serendipitous (16), in general it is important that the U1A preparation be scrupulously nuclease free. This can be assayed as described (10).

9

. For this example, an RNA with a limited amount of 3′ heterogeneity proved sufficient. For other crystallization projects, homogeneous RNA prepared, for instance, through ribozyme cleavage (17, 18), may be necessary.

10

. We employ 3 mm-thick polyacrylamide slab gels. For a 10 ml volume transcription, a gel that is 32 cm wide and 23 cm high is sufficient.

11

. UV irradiation can damage RNA (19). It is imperative to limit exposure of the RNA to a few seconds. It is also essential to wear UV-protective clothing and face and hand protection.

12

. Different geometries for crystallization experiments are discussed by McPherson (20).

13

. A tabulation of the range of cryoprotectants that have been employed successfully for RNA and RNA:protein complex crystals appears in ref. (12).

References

  • 1.Ferré-D’Amaré AR, Zhou K, Doudna JA. A general module for RNA crystallization. J Mol Biol. 1998;279:621–631. doi: 10.1006/jmbi.1998.1789. [DOI] [PubMed] [Google Scholar]
  • 2.Ferré-D’Amaré AR, Zhou K, Doudna JA. Crystal structure of a hepatitis delta virus ribozyme. Nature. 1998;395:567–574. doi: 10.1038/26912. [DOI] [PubMed] [Google Scholar]
  • 3.Nagai K, Oubridge C, Jessen TH, Li J, Evans PR. Crystal structure of the RNA-binding domain of the U1 small nuclear ribonucleoprotein A. Nature. 1990;348:515–520. doi: 10.1038/348515a0. [DOI] [PubMed] [Google Scholar]
  • 4.Rupert PB, Ferré-D’Amaré AR. Crystal structure of a hairpin ribozyme-inhibitor complex with implications for catalysis. Nature. 2001;410:780–786. doi: 10.1038/35071009. [DOI] [PubMed] [Google Scholar]
  • 5.Kulshina N, Baird NJ, Ferré-D’Amaré AR. Recognition of the bacterial second messenger cyclic diguanylate by its cognate riboswitch. Nat Struct Mol Biol. 2009;16:1212–1217. doi: 10.1038/nsmb.1701. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Xiao H, Murakami H, Suga H, Ferré-D’Amaré AR. Structural basis of specific tRNA aminoacylation by a small in vitro selected ribozyme. Nature. 2008;454:358–361. doi: 10.1038/nature07033. [DOI] [PubMed] [Google Scholar]
  • 7.Ferré-D’Amaré AR, Doudna JA. Methods to crystallize RNA. In: Beaucage SL, Bergstrom DE, Glick GD, Jones RA, editors. Current Protocols in Nucleic Acid Chemistry. New York: John Wiley & Sons; 2000. pp. 7.6.1–7.6.10. [DOI] [PubMed] [Google Scholar]
  • 8.Ferré-D’Amaré AR, Doudna JA. Crystallization and structure determination of a hepatitis delta virus ribozyme: use of the RNA-binding protein U1A as a crystallization module. J Mol Biol. 2000;295:541–556. doi: 10.1006/jmbi.1999.3398. [DOI] [PubMed] [Google Scholar]
  • 9.Rupert PB, Ferré-D’Amaré AR. Crystallization of the hairpin ribozyme: illustrative protocols. Methods Mol Biol. 2004;252:303–311. doi: 10.1385/1-59259-746-7:303. [DOI] [PubMed] [Google Scholar]
  • 10.Ferré-D’Amaré AR. Use of the spliceosomal protein U1A to facilitate crystallization and structure determination of complex RNAs. Methods. 2010;52:159–167. doi: 10.1016/j.ymeth.2010.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kulshina N, Edwards TE, Ferré-D’Amaré AR. Thermodynamic analysis of ligand binding and ligand binding-induced tertiary structure formation by the thiamine pyrophosphate riboswitch. RNA. 2010;16:186–196. doi: 10.1261/rna.1847310. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Ferré-D’Amaré AR. Crystallization of RNA for structure determination by X-ray crystallography. In: Klostermeier D, Hammann C, editors. Structure and Folding of RNA. Berlin: deGruyter; 2013. in press. [Google Scholar]
  • 13.Xiao H, Edwards TE, Ferré-D’Amaré AR. Structural basis for specific, high-affinity tetracycline binding by an in vitro evolved aptamer and artificial riboswitch. Chem Biol. 2008;15:1125–1137. doi: 10.1016/j.chembiol.2008.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Burgess RR. Use of polyethyleneimine in purification of DNA-binding proteins. Meth Enzymol. 1991;208:3–10. doi: 10.1016/0076-6879(91)08003-z. [DOI] [PubMed] [Google Scholar]
  • 15.Oubridge C, Ito N, Teo CH, Fearnley I, Nagai K. Crystallization of RNA-protein complexes II. The application of protein engineering for crystallization of the U1A protein-RNA complex. J Mol Biol. 1995;249:409–423. doi: 10.1006/jmbi.1995.0306. [DOI] [PubMed] [Google Scholar]
  • 16.Klein DJ, Ferré-D’Amaré AR. Crystallization of the glmS ribozyme-riboswitch. Methods Mol Biol. 2009;540:129–139. doi: 10.1007/978-1-59745-558-9_10. [DOI] [PubMed] [Google Scholar]
  • 17.Ferré-D’Amaré AR, Doudna JA. Use of cis- and trans-ribozymes to remove 5′ and 3′ heterogeneities from milligrams of in vitro transcribed RNA. Nucleic Acids Res. 1996;24:977–978. doi: 10.1093/nar/24.5.977. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Price SR, Ito N, Oubridge C, Avis JM, Nagai K. Crystallization of RNA-protein complexes I. Methods for the large-scale preparation of RNA suitable for crystallographic studies. J Mol Biol. 1995;249:398–408. doi: 10.1006/jmbi.1995.0305. [DOI] [PubMed] [Google Scholar]
  • 19.Kladwang W, Hum J, Das R. Ultraviolet Shadowing of RNA Can Cause Significant Chemical Damage in Seconds. Sci Rep. 2012;2:1–6. doi: 10.1038/srep00517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.McPherson A. Crystallization of biological macromolecules. Cold Spring Harbor: Cold Spring Harbor Laboratory Press; 1999. [Google Scholar]

RESOURCES