|Home | About | Journals | Submit | Contact Us | Français|
This column highlights recently published articles that are of interest to the readership of this publication. We encourage ABRF members to forward information on articles they feel are important and useful to Clive slaughter, hartwell center, St. Jude children’s Research hospital, 332 north Lauderdale St., Memphis TN 38105–2794. Tel: (901) 495–4844; Fax: (901) 495–2945; email: email@example.com or to any member of the editorial board. Article summaries reflect the reviewer’s opinions and not necessarily those of the Association.
Korbel JO, Urban AE, Grubert F, Du J, Royce TE, Starr P, Zhong G, Emanuel BS, Weissman SM, Snyder M, Gerstein MB. Systematic prediction and validation of breakpoints associated with copy-number variants in the human genome. Proceedings of the National Academy of Sciences, U.S.A. 2007;104:10,110–10,115.
Although copy number variants (CNVs) cause a level of variation comparable to that from single nucleotide polymorphisms, knowledge of CNVs is less well developed in so far as the precise breakpoint sequences producing them are rarely known. The methods usually employed for CNV detection provide effective resolution of only tens to hundreds of kilobases in defining breakpoints, leaving considerable uncertainty about which genes are included between them. Comparative genome hybridization (CGH) now achieves high resolution through the use of high-density tiling arrays, but noise levels are high, owing to cross-hybridization caused by the shortness of the probes and the complexity of genomic DNA. The present paper provides computational tools for analysis of the data from high-resolution CGH arrays for precise definition of CNV breakpoints. Resolution of approximately 300 bp is achieved.
Ponchon L, Dardel F. Recombinant RNA technology: The tRNA scaffold. Nature Methods. 2007;4:571–576. [PubMed]
Structural, biochemical and pharmacological studies of RNA require large quantities of homogeneous RNA species. Chemical synthesis can be costly, and the heterogeneity of RNAs made by in-vitro transcription using phage RNA polymerase necessitates control by procedures that are time consuming to optimize. The present paper provides a strategy for expressing and purifying structured RNA in E. coli. The problem of susceptibility to cellular RNases is overcome by incorporating target RNA species into the anticodon stem-loop of plasmid-encoded human tRNALys or E. coli tRNAMet. The tRNA acts as a protective scaffold. The RNA chimeras are expressed constitutively from a strong bacterial promoter, and are processed and modified by the same enzymes as endogenous transcripts. The methodology is expected to be widely applicable for large-scale structural and molecular studies of RNA function.
Bowman MJ, Zaia J. Tags for the stable isotopic labeling of carbohydrates and quantitative analysis by mass spectrometry. Analytical Chemistry. 2007;79:5777–5784. [PubMed]
A series of compounds is described that permit isotopic labeling of carbohydrates for relative quantification by procedures analogous to those employed for relative quantification of peptides in proteomics. The tag compounds incorporate an aromatic amine that allows for UV detection of carbohydrates that generally lack UV absorbance. They couple to the reducing end of glycans by reductive amination in high yield and with minimal side reactions, and confer an increase in hydrophobicity to promote ionization. Isotopic labeling is via hydrogen atoms bound directly to carbons. The compounds are made in four isotopically distinguishable forms to allow comparison of four samples in a single experiment.
Savitski MM, Kjeldsen F, Nielsen ML, Zubarev RA. Relative specificities of water and ammonia losses from backbone fragments in collision-activated dissociation. Journal of Proteome Research. 2007;6:2669–2673. [PubMed]
It is generally assumed that in collision-induced dissociation (CID) the loss of ammonia (17 Da) from b ions is indicative of the presence of amino acids with nitrogen in the side chain, whereas the loss of water (18 Da) from y ions indicates the presence of side-chain oxygen atoms. Here, these propositions are tested using a database of high-resolution CID spectra for 20,000 different tryptic peptide sequences. The spectra are acquired with a Fourier transform mass spectrometer capable of distinguishing the signal for ammonia loss from the first isotope peak of water loss when both are present together, enabling these signals to be independently and accurately quantified. Ammonia loss from b ions is indeed shown to be highly correlated (>95%) with the presence of Asn, Gln, His, Lys, and Arg. However, water loss from y ions is not specific enough to rely upon for detecting amino acids with side chains containing oxygen. Ammonia loss from y ions of Glu-C peptides is also diagnostic of Asn, Gln, His, Lys, and Arg; y ions with an N-terminal Gln residue exhibit strong water and ammonia losses. Ser and Thr in the next-to-last position in the b ions derived from tryptic peptides exhibit strong water losses. These rules will be useful for the development of algorithms for de novo sequence interpretation of CID spectra.
Lehmann WD, Krüger R, Salek M, Hung C-W. Neutral loss-based phosphopeptide recognition: A collection of caveats. Journal of Proteome Research. 2007;6:2866–2873. [PubMed]
A popular method for detecting Ser and Thr phosphorylation of peptides is the observation of neutral loss of H3PO4 (98.0 Da/z) during CID. This paper documents processes that can result in false positives in the method. Among these processes are neutral losses of Pro (97.0/z) and Val (99.0/z), which produce mass losses close to that of phosphate, particularly from peptides with N-terminal sequences PP and VP. Other neutral losses may be confused with the loss of H3PO4 when the charge state of the precursor ion is mis-assigned. For example, loss of sulfenic acid from a doubly charged precursor (64/2 = 32.0) produces a loss in mass close to that of the loss of H3PO4 from a triply charged precursor (98/3 = 32.7). The coincidental occurrence of a fragment ion of m/z in the spectral region of interest can also lead to false positives when accompanied by charge-state mis-assignment.
König s, Kollas O, Dreisewerd K. Generation of highly charged peptide and protein ions by atmospheric pressure matrix-assisted infrared laser desorption/ionization ion trap mass spectrometry. Analytical Chemistry. 2007;79:5484–5488. [PubMed]
MALDI, unlike electrospray ionization, typically produces ions of very low charge state. In this preliminary report, the production of highly charged ions by atmospheric pressure MALDI is described. The method employs a pulsed infrared laser and a glycerol matrix. Small peptides are ionized in doubly charged form; the most abundant charge state of insulin is +4, and for cyto-chrome c is +10. These high charge states promise to bring ions of high mass within the m/z range of mass analyzers such as Paul traps. The potential advantages of using MALDI to produce multiply charged ions instead of electrospray include control over the ionization process in which ions are formed only when the laser is switched on, introduction of multiple samples into the ion source on a single target plate, and high tolerance to salts and detergents. It is speculated that the mechanism by which multiple ions are generated may rely on an electrospray-like process in which the laser generates an expanding plume of matrix droplets that undergo collisions with the atmospheric pressure gas in the source, promoting droplet break-up.
Ibáñez AJ, Muck A, Svatoš A. metal-chelating plastic MALDI (pMALDI) chips for the enhancement of phosphorylated-peptide/protein signals. Journal of Proteome Research. 2007;6:1183–1189. [PubMed]
Blacken GR, Volneý M, Valsar T, Sadílek M, Tureček F. In situ enrichment of phosphopeptides on MALDI plates functionalized by reactive landing of zirconium(IV)-n-propoxide ions. Analytical Chemistry. 2007;7:5449–5456. [PubMed]
Wang Y, Chen W, Wu J, Guo Y, Xia X. highly efficient and selective enrichment of phosphopeptides using porous anodic alumina membrane for MALDI-TOF MS analysis. Journal of the American Society for Mass Spectrometry. 2007;18:1387–1395. [PubMed]
These three papers present alternative approaches to selective capture of phosphopeptides on MALDI targets. Ibáñez et al. describe a disposable polymeric nitrilotri-acetate (NTA) target that is activated with the metal cations, Ga(III), and Ni(II). Blacken et al. produce a durable surface functionalized with a zirconium oxide coating that is produced by a reactive landing process in which zirconium(IV)-n-propoxide solutions in 1-propanol are electrosprayed onto plasma-oxidized stainless steel. Wang et al. use a porous anodic alumina membrane, a self-ordered nanochannel aluminum oxide material formed by anodization of aluminum in suitable multiprotic acid solutions. Each paper reports selectivity of binding and improved phosphopeptide detectability.
Mazor Y, Van Blarcom T, Mabry R, Iverson BL, Georgiou G. Isolation of engineered, full-length antibodies from libraries expressed in Escherishia coli. Nature Biotechnology. 2007;25:563–565.
The time consuming and expensive nature of hybridoma technology had spurred interest in methods involving the screening of combinatorial expression libraries, but these have so far been restricted to single-chain variable region fragments or Fab antigen–binding fragments, which lack important functional characteristics and exhibit short half-lives in vivo. The present paper reports an E. coli combinatorial library system expressing full-length heavy and light chains. These chains are secreted into the periplasm, where they assemble into IgGs. The IgGs are captured by an Fc-binding protein on the inner membrane. After permeabilizing the outer membrane, spheroplasts expressing such antibodies may be incubated with fluorescently labeled antigen and then selected by flow cytometry. The method is validated by isolating high-affinity antibodies from a library constructed from a mouse immunized with a purified antigen from Bacillus anthracis.
Wittig I, Karas M, Schägger H. High resolution clear native electrophoresis for in-gel functional assays and fluorescence studies of membrane protein complexes. Molecular and Cellular Proteomics. 2007;6:1215–1225. [PubMed]
Blue native electrophoresis, a method for separation of high-molecular-weight complexes, is useful also for separating membrane complexes because the negatively charged Coomassie Blue dye maintains membrane proteins in solution in the absence of detergents by binding to hydrophobic surfaces and imparting to them negative charge, which causes electrostatic repulsion between protein molecules. However, the technique does disrupt some physiological protein interactions, and its colored nature also interferes with the use of fluorescence for detection and activity measurement. The present paper employs mixtures of anionic and non-ionic detergents in place of Coomassie Blue dye to solubilize membrane complexes to better preserve physiological interactions but still to confer excess negative charge on proteins to provide for anodal electrophoretic mobility. High-resolution and in-gel catalytic assays for mitochondrial complexes I–V are demonstrated, and detection of fluorescently tagged proteins is described.
James TL, Milos NC. Elimination of vertical streaking in polyacrylamide gel electrophoresis. Analytical Biochemistry. 2007;367:134–136. [PubMed]
This paper addresses the problem of vertical streaking of proteins in non-denaturing polyacrylamide gel electrophoresis. Sample volume is pinpointed as the most important variable in generating streaks. Minimizing sample volume to eliminate the problem is found to be more effective than reducing the amount of protein loaded.
Aramini JM, Rossi P, Anklin C, Xiao R, Montelione GT. Microgram-scale protein structure determination by NMR. Nature Methods. 2007;4:491–493. [PubMed]
Sakellariou D, Le Goff G, Jacquinot J-F. High-resolution, high-sensitivity NMR of nanolitre anisotropic samples by coil spinning. Nature. 2007;447:694–698. [PubMed]
These papers describe new methods for improving the sensitivity of NMR spectroscopy. Aramini et al. extend the use of microcoil-probe technology to achieve for the first time near complete and automated resonance assignments and 3D structure determination of a small protein (68 residues) using only 72 μg of protein sample. A 1-mm triple-resonance microcoil NMR probe with an active NMR volume of 2.5 μL and a practical minimum sample volume of 5 μL is employed. Protein structure determination is feasible with protein concentrations >1.2 mM and molecular weights <15 kDa. Sakellariou et al. improve sensitivity with non-liquid samples, which must be spun rapidly to obtain well-resolved spectra. They employ co-rotation of a micro-sized detector coil and the sample, and wireless inductive coupling. Signals from nL-sized samples of organic powders and biological tissue are found to be 10X improved relative to standard NMR techniques.
Pekar A, Sukumar M. Quantitation of aggregates in therapeutic proteins using sedimentation velocity analytical ultracentrifugation: Practical considerations that affect precision and accuracy. Analytical Biochemistry. 2007;367:225–237. [PubMed]
The precision and accuracy with which analytical ultracentrifuge sedimentation velocity analysis can measure the extent of aggregation of pharmaceutical mono-clonal antibodies are assessed. Using samples spiked with known quantities of aggregated antibody, and employing the Sedfit program for data analysis, a precision of approximately ±0.5% over a range of aggregate content of 0.6–67%, and an accuracy of approximately 1.5%, are observed when nine measurements are averaged. The limit of detection is 0.2% and the limit of quantitation is 0.6%. Precision and accuracy are sensitive to the quality of the centerpieces. The integrity of the surfaces and septum should be examined microscopically to avoid not only poor precision but also spurious peaks.
Gutmann DAP, Mizohata E, Newstead S, Ferrandon S, Henderson PJF, Van Veen HW, Byrne B. A high-throughput method for membrane protein solubility screening: The ultracentrifugation dispersity sedimentation assay. Protein Science. 2007;16:1422–1428. [PubMed]
Maintenance of membrane proteins in a mono-dispersed state is important for successful crystallization. These proteins are, however, prone to aggregation. This paper reports the use of sedimentation velocity analysis as a method for rapid screening of detergents for their effectiveness in inhibiting the aggregation of membrane proteins destined for crystallization. The method employs as little as 5 μL of purified protein. Used in combination with SDS-PAGE, it provides information about aggregation more rapidly and with smaller sample consumption than size-exclusion chromatography.
Speers AE, Blacker AR, Wu CC. Shotgun analysis of integral membrane proteins facilitated by elevated temperature. Analytical Chemistry. 2007;79:4613–4620. [PubMed]
Methods for LC-MS analysis of peptides derived from the transmembrane domains of integral membrane proteins are presented. Transmembrane segments are prepared from a plasma membrane fraction by first digesting away non-embedded polypeptide segments with a non-specific protease, proteinase K. During a subsequent high-speed spin to sediment the membranes, the enzyme-accessible segments remain in the supernatant. The pellet is then solubilized in 90% formic acid, and digested with cyanogen bromide to release peptides from the transmembrane segments. LC-MS analysis of the resulting hydrophobic peptides is improved by performing on-line reverse-phase chromatography at elevated temperature. The number of peptides identified is five times greater when a column temperature of 60° C is employed compared with room temperature.
Sugiyama N, Masuda T, Shinoda K, Nakamura A, Tomita M, Ischihara Y. Phosphopeptide enrichment by aliphatic hydroxy acid-modified metal oxide chromatography for nano-LC-MS/MS in proteomics applications. Molecular and Cellular Proteomics. 2007;6:1103–1109. [PubMed]
In order to suppress binding of non-phosphorylated but acidic peptides during purification of phosphopeptides on titanium dioxide and zirconium dioxide tips, peptide mixtures are bound in the presence of organic hydroxyacids. These bind less strongly than phosphate but more strongly than carboxylic acids. For subsequent LC-MS analysis, lactic acid works best with titanium and β-hydroxypropanoic acid with zirconium. Recoveries of phosphopeptides from the stationary phases differ widely between phosphopeptides, ranging from less than 10% to nearly 100%.
Huttlin EL, Hegeman AD, Harms AC, Sussman MR. Comparison of full versus partial metabolic labeling for quantitative proteomics analysis in Arabidopsis thaliana. Molecular and Cellular Proteomics. 2007;6:860–880. [PubMed]
Although relative quantification through mixing protein samples obtained from cells grown in media of natural isotopic abundance and cells grown in media in which 14N has been completely replaced by 15N has become widely accepted, the procedure is problematic for large multicellular eukaryotes, where it may be impractically expensive or detrimental to the organism’s health. To circumvent these problems, it’s desirable to perform partial rather than complete replacement of the isotope. Quantification then relies on measuring changes in the relative intensity of peaks within the single isotopic envelope for each peptide rather than the production of two or more entirely separate envelopes corresponding to the peptide derived from the different sources being compared. The present paper provides algorithms for automated quantification from the resulting data. Once a peptide sequence has been assigned, the elemental composition is calculated. Single-ion mass chromatograms for all the ions in an isotopic cluster are then constructed, and regression analysis is used to plot their intensities relative to the monoisotopic signal. The patterns are compared with a library of isotopic distributions based on the elemental composition and known expression ratios, again by linear regression. Partial labeling is shown to provide results comparable to full labeling in terms of dynamic range, accuracy, and reproducibility.
Hervey WJ, Strader MB, Hurst GB. Comparison of digestion protocols for microgram quantities of enriched protein samples. Journal of Proteome Research. 2007;6:3054–3061. [PubMed]
Protocols for trypsin digestion of very small quantities (2 μg) of protein in solution are compared for their effectiveness in conjunction with “shotgun” LC-MS/MS analysis. Specifically, two protocols performed in 80% acetonitrile are compared with digestion of proteins denatured with 6 M guanidine hydrochloride, sequential digestion of urea-denatured protein with Lys-C and trypsin, and digestion in the presence of 0.1% RapiGest cleavable detergent. Eighty percent acetonitrile is found to perform as well as, or in some instances better than, the other solvents, possibly because fewer sample manipulation steps are demanded in the procedures employed for its use in this paper.
Chen EI, Cociorva D, Norris JL, Yates JR. Optimization of mass spectrometry–compatible surfactants for shotgun proteomics. Journal of Proteome Research. 2007;6:2529–2538. [PubMed]
Three commercially available MS-compatible detergents are compared for their effectiveness in promoting trypsin digestion in conjunction with LC-MS/MS analysis: RapiGest (Waters Corp.), Invitrosol (Invitrogen), and PPS Silent Surfactant (Protein Discovery, Inc.). All three act as mild denaturants to increase substrate availability for digestion in solution and increase numbers of proteins identified, but may influence the types of peptides observed. Organic solvent can be used in conjunction with MS-compatible detergents to further improve numbers of hydrophobic peptides identified.
Krogh M, Fernandez C, Teilum M, Bengtsson S, James P. A probabilistic treatment of the missing spot problem in 2D gel electrophoresis experiments. Journal of Proteome Research. 2007;6:3335–3343. [PubMed]
In 2D electrophoresis experiments, there are two standard approaches to dealing with spots that are detected on some gels but not on others. In the first, all missing values for spot staining intensity are replaced with zero-intensity values. This approach is problematic because some spots may seem to be missing, but their absence is due to detection errors, and they are actually not missing at all. The second approach is to ignore all missing values and perform statistical analysis only on spots for which staining intensity is measured on all the gels. This is problematic because proteins with potentially the highest-fold changes are thereby excluded from the analysis. An alternative approach is offered in this paper. Missing spots are included in the overall analysis by supposing that a missing value among a set of replicate measurements is likely to be due to error if the replicates show relatively large spot volumes at the position concerned, and if the incidence of missing spots among those of comparable volume is low. A model formalizing this concept is presented. When used in conjunction with bootstrap sampling, the model permits confidence intervals and statistical significance to be assigned for expression changes in comparisons between groups of samples. A program implementing the analysis is available at http://bioinfo.thep.lu.se/MissingValues2D-gels.html.
Emanuelsson O, Nagalakshmi U, Zheng D, Rozowsky JS, Urban AE, DU J, Lian Z, Stolc V, Weissman S, Snyder M, Gerstein MB. Assessing the performance of different high-density tiling microarray strategies for mapping transcribed regions of the human genome. Genome Research. 2007;17:886–897. [PubMed]
Two oligonucleotide tiling array platforms manufactured by in situ probe synthesis are compared for their performance in mapping transcribed regions of the human genome. The first platform is the Affymetrix ENCODE array, which covers one strand of the entire ENCODE region tiled with 25-mer oligonucleotides at an average distance between oligonucleotide start sites of 21 bases. The second platform, produced by maskless array synthesis (MAS) in a NimbleGen array synthesizer, contains 36-mer oligonucleotides tiling both strands of the non-repetitive sequence of the ENCODE regions end-to-end. Identical biological samples are used for both platforms, and unified scoring and procedures for assigning transcription-ally active regions (TARs) and untranscribed regions are employed. The Affymetrix platform is observed to yield TARs that better agree with known annotation, and exons in multiple-exon transcripts that show better coherence. The difference is ascribed to the higher density of probes and to the presence of mismatch probes in the Affymetrix platform. However, the two platforms are almost equal in their ability to detect novel transcription, as revealed in validation of TARs by RT-PCR. Nevertheless, advantages afforded by the MAS technology are rapid manufacturing of customized arrays and small array series. Their performance might be enhanced by incorporating true mismatches in the array design.
Euskirchen GM, Rozowsky JS, Wei C-L, Lee WH, Zhang ZD, Hartman S, Emanuelsson O, Stolc V, Weissman S, Gerstein MB, Ruan Y, Snyder M. Mapping of transcription factor binding regions in mammalian cells by Chip: Comparison of array- and sequencing-based technologies. Genome Research. 2007;17:898–909. [PubMed]
This paper first explores parameters for experimental design of studies in which transcription factor binding sites are identified by chromatin immunoprecipitation (ChIP) with the factor of interest followed by identification of precipitated DNA targets by probing a DNA tiling microarray (ChIP-chip). The best performance is observed with high probe density, oligonucleotides of ≥50 bases, and the presence of Cot-1 DNA as a competitor. The ChIP-chip methodology is then compared to ChIP with DNA sequencing for segment identification. Sequencing is performed on concatenations of Paired-End diTags cloned from the 5′ and 3′ ends of each ChIPDNA fragment (ChIP-PET). These two approaches show good agreement for the highest ranked targets, but less overlap for the lower ranked targets. Each method identifies targets that are missed by the other. The methods are therefore complementary and the results may be merged to build the most comprehensive list of targets.
Johnson DS, Mortazavi A, Myers RM, Wold B. Genomewide mapping of in vivo protein-DNA interactions. Science. 2007;316:1497–1502. [PubMed]
This paper illustrates the use of ultra-high-throughput DNA sequencing technology in association with ChIP for mapping transcription-factor binding sites. Unlike ChIP-PET, it involves no plasmid library construction. Sequencing is performed with the Solexa/Illumina platform, which produces very large numbers of short sequences in each instrument run—40 million sequences of 25 nucleotides each. Quantities of a given sequence are measured by the frequency with which the sequence is identified in the sample. This technology is used to construct a high-resolution map of sites binding the neuron-specific silencer factor, NRSF. This protein is shown to bind to 1946 sites, and the sites are defined with a resolution of ± 50 bp.
The ENCODE Project Consortium. Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature. 2007;447:799–816. [PubMed]
This paper reports results from the pilot phase of the Encyclopedia of DNA Elements (ENCODE) Project, which aims to provide a biologically informative representation of the human genome through the use of high-throughput methods to identify and catalog functional encoded elements. Thirty-five groups contributed data sets that examine 30 Mb of DNA, representing about 1% of the human genome. Many findamental observations have emerged. For example, the human genome is found to be pervasively transcribed, with the majority of bases associated with at least one primary transcript. Many novel transcripts that do not encode protein are identified, some overlapping protein-coding genes and others in regions hitherto believed to be silent. Numerous novel transcription start sites are identified. Many of these exhibit chromatin structure and contain sequences with characteristic protein-binding properties similar to well-understood promoters. Chromatin accessibility and histone modifications are highly predictive of the presence of transcription start sites and start-site activity.
Dennis JH, Fan H-Y, Reynolds SM, Yuan G, Meldrim JC, Richter DJ, Peterson DG, Rando OJ, Noble WS, Kingston RE. Independent and complementary methods for large-scale structural analysis of mammalian chromatin. Genome Research. 2007;17:928–939. [PubMed]
Methods for high-throughput analysis of nucleosome protection over broad areas of the human genome are presented. Two independent and complementary methods are described. One uses a tiling microarray technique to measure nucleosome occupancy, and the other uses a primer-extension technique to measure nuclease accessibility. These measures are linked insofar as the operational definition of regions occupied by nuceleosomes is related to their resistance to cleavage by micrococcal nuclease (MNase). For both methods, formaldehyde-crosslinked chromatin is employed. The chromatin is subjected to limited MNase digestion to study the protected DNA. For the microarray method, mononucleosomal DNA is purified by gel electrophoresis and used to probe a tiling microarray. For the primer-extension method, DNA fragments of 150–3000 bases are subjected to ligation-mediated polymerase chain reaction (LM-PCR) and the products are analyzed on a capillary electrophoresis sequencer. This allows changes in chromatin cleavage sensitivity to be mapped with single-nucleotide resolution. The combined methods provide reliable data on mammalian chromatin structure and are suitable for routine, high-throughput operation.
The Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447:661–678. [PubMed]
Some 50 different groups in the U.K. contribute to the largest genome-wide association study hitherto undertaken. Genetic risk factors for bipolar disorder, coronary heart disease, Crohn’s disease, rheumatoid arthritis, type 1 diabetes, and type 2 diabetes are identified. The study utilizes the Affymetrix GeneChip 500K Mapping Array Set to study approximately 2000 individuals with each of seven major diseases. A key feature of the study’s design is the utilization of a shared set of 3000 controls. Large sample sizes are necessitated by the modest size of the effects exerted by most of the loci identified. The study detects previously known associations, and reveals 58 further associations with single-point P values between 1 × 10−5 and 5 × 10−7 that are likely to yield new susceptibility loci. The work provides a template for future studies of even larger scale.