TECHNOLOGY FEATURE

Identifying in vivo substrates 226 Screening libraries 228 Laying down the chips 229 Box 1: Phosphospecific antibodies 227 Box 2: Scansite 228 Probing the kinome Researchers are eager to map out all the signaling pathways regulated by protein . Many new

methods protocols and reagents are available to tackle this formidable task. Laura Bonetta reports.

Phosphorylation is the main device a cell for these tests, including off-the-shelf and uses to control the function of its proteins. custom-made kits, phosphospecific .com/nature e The responsible , protein kinases, antibodies, purified kinases, and specific regulate a myriad of processes from cell substrates and inhibitors. .natur

w division in development to neuronal com- The initial kinase assays used radioac- munication in memory. In addition, many tive ATP in the reaction mixture. But the of them have been implicated in disease. trend in recent years has been to move

http://ww The notorious Novartis drug Gleevec, away from using radioactivity. A research-

which has had great success in treating er can now choose from a wide variety of

oup chronic myeloid leukemia and gastroin- fluorescence- and luminescence-based r

G testinal stromal tumors, is a kinase inhibi- assays. Molecular Devices, for example,

tor. It is no surprise then that academic has a fluorescence polarization–based labs and pharmaceutical companies are kinase assay that uses its proprietary The IMAP technology from Molecular Devices lishing hot on the trail of understanding kinase IMAP technology. When fluorescently

b is a homogeneous fluorescent kinase (or function. labeled peptides are incubated with a phosphatase) assay. of Pu Kinases impart their signal by transfer- kinase, the phosphorylated peptides bind a fluorescent substrate is detected by its ring the γ-phosphate group from ATP to to the IMAP reagent—a trivalent metal binding to trivalent metals on nanoparticles and measuring the resulting increase in Nature

a recipient , or with high affinity for phosphates—caus- fluorescence polarization. (Courtesy of 5 residue on a protein. The choice of target ing an increase in the polarization of the Molecular Devices.)

200 is decided by the catalytic domain of the fluorescence. “The assay is highly resistant

© kinase, which usually recognizes a docking to interference and gives a strong signal,” domain on the substrate and a specific, says Richard Sportsman, vice president phosphorylated peptides causing them to short sequence of amino acids (the con- of reagents development. For a new lose the FRET signal. “We have developed sensus motif) that surround the residue to kinase of unknown specificity, Molecular a portfolio of substrates that have been be phosphorylated. Each kinase can have Devices provides the IMAP Substrate validated with over 130 kinases and have several substrates and be itself a substrate Finder products, consisting of arrays of the potential to be used to assay 85% of for other kinases. 55 different peptides in microwell-plate all kinases in Z-LYTE format,” says Chris To start to delineate the complex net- formats, which can be tested against the Armstrong, business area manager for work of pathways involving kinases, kinase. “The substrates are picked from drug discovery solutions. researchers want to match every kinase in all relevant regions of the kinome,” says Invitrogen also markets another assay a cell, representing the so-called kinome, Sportsman. “Typically customers will get based on time resolved (TR)-FRET called to specific phosphorylation sites on target 3–4 hits. They can then order the best sub- LanthaScreen, which can be used to study proteins. Products and assays to accom- strates for further testing.” not only phosphorylation by kinases but a plish this ambitious task are making their Another platform offered by a number whole variety of protein interactions. An way into research labs. of companies is based on fluorescence antibody labeled with a donor terbium resonance energy transfer (FRET)— fluorophore associates specifically with Monitoring kinase activity a detection method that relies on the phosphorylated peptide substrates labeled Not every kinase will work in an in vitro proximity of two fluorescent dye mol- with the acceptor fluorophore, resulting in assay and a peptide that is phosphorylated ecules (fluorophores) to allow the trans- an increased TR-FRET value in the reaction by a kinase in vitro does not necessarily fer of excitation from donor to acceptor mixture. “We have tried to focus on generic reflect that kinase’s physiological target. moiety. Invitrogen’s Z-LYTE kinase assay reagents, so that a researcher can build his However, kinase kits provide an efficient uses synthetic peptides labeled with both or her own assays.” says Armstrong. way to monitor the activity of a kinase donor and acceptor fluorophores. When Promega markets a luminescence- and screen compounds that will affect this a site-specific protease is added to the based system, Kinase-Glo, which mea- function. Many companies offer reagents phosphorylation reaction, it cleaves non- sures changes in ATP levels in a kinase

NATURE METHODS | VOL.2 NO.3 | MARCH 2005 | 225 TECHNOLOGY FEATURE

Dual laser detection Identifying in vivo substrates using the Beadlyte The holy grail of kinome analysis is to technology. As determine the physiological substrates of each bead flows all kinases to begin to map out signaling through a cuvette in a single file, a pathways. The KESTREL (kinase sub- red laser detects strate tracking and elucidation) method the identity of the offered by Kinasource identifies specific bead and a green sequences that are phosphorylated in laser quantifies vivo. Cell lysates from many different tis- the fluorescence sues are incubated with a purified active associated with kinase and the resulting phosphoproteins the analyte-bound are separated by gel electrophoresis. Any methods reporter. (Courtesy of Upstate Group LLC.) leads identified in this initial screen are then pursued in a scaled-up protocol to isolate selected proteins from the gel and determine their identity by mass spec- .com/nature e trometry. “Most researchers ask us to purify all the strong signals from the ini- .natur

w tial screen,” says Axel Knebel, managing reaction. ATP from a completed kinase with a broad range of purified kinase-sub- director of Kinasource. The initial screen assay drives a reaction with luciferin and strate combinations” says Michael Curtin, typically requires 2–3 weeks and the puri-

http://ww luciferase to produce oxyluciferin and light; product manager for cellular analysis. “The fication an additional month on average.

the amount of light produced is inversely substrate can be a peptide, protein, lipid, “In theory the experiment is quite easy to

oup correlated with kinase activity. “The nice sugar—just about anything a kinase nor- do, but, like mass spectrometry, you need r

G thing about this assay is that it works mally uses as a substrate”. to accumulate the experience and skill to lishing b Pu

Nature

5 200 ©

226 | VOL.2 NO.3 | MARCH 2005 | NATURE METHODS TECHNOLOGY FEATURE

do it,” says Knebel. Kinasource provides To enrich for proteins that are phos- offers “a simple procedure to selectively KESTREL as a service to pharmaceutical phorylated on tyrosine—the minority enrich for the tyrosine phosphopro- companies and as a mix of collaboration of all phosphorylated proteins in a cell— teome,” says Christopher Bunker, director and service to academic labs. “Academic researchers in Arthur Salomon’s group of new business development at CST. “It researchers get a much better price than the at Brown University added an immuno- avoids the limitations of IMAC methods pharmaceutical companies, but we end up precipitation step using an antibody that such as charge bias, enrichment based on on the published paper,” says Knebel. specifically recognizes proteins phos- abundance and multistep procedures.” According to Knebel the method is suc- phorylated on a tyrosine residue (Box 1) According to Bunker, the company has cessful in finding physiologically relevant before purification by IMAC1. “The goal identified nearly 4,000 phosphotyrosine substrates 80% of the time. “Some kinases is not to find the most phosphorylation sites, of which almost 70% are new. don’t bind ATP that well and cannot be sites. It is to find those associated with To make these procedures amenable to used in the assay,” he says. Another limita- signaling pathways and disease in the large-scale screens, Salomon has automat- methods tion is that KESTREL does not work well shortest amount of time,” says Salomon. ed many of the steps involved in processing for scarce or rare substrates because the “One-third of all proteins in a cell are samples and gathering data. “In one hour technique suffers from a relatively high phosphorylated, but most are not relevant we can get 300 phosphopeptide mass spec- level of background contamination. to signaling. We tried to design a method tra,” says Salomon. The bottleneck, how- .com/nature e In an effort to get around some of these to efficiently extract the most meaningful ever, is having to interpret the information. problems other researchers are finessing data by focusing on proteins phosphory- A number of algorithms are available to .natur

w methods that do not require gel electro- lated on tyrosine.” match mass spectra with protein sequences phoresis. One such technique uses immo- As an alternative, the PhotoScan tech- but they “are not optimized for phospho- bilized metal affinity chromatography nology developed by John Rush et al.2 at peptide spectra,” says Salomon. “Manual

http://ww (IMAC) resins with a high specificity for Technology (CST) uses a validation of the spectra is still necessary.”

phosphates to purify phosphopeptides in phosphotyrosine antibody immobilized Another challenge is determining which

oup bulk from protease-digested cell extracts. on agarose beads to isolate phosphotyro- phosphopeptides represent interesting r

G The isolated peptides are then identified by sine-containing peptides from protease- substrates to pursue in follow-up experi-

reversed-phase liquid chromatography and digested cell extracts, which were then ments. “We have to do experiments like tandem mass spectrometry (LC-MS/MS). analyzed by LC-MS/MS. The strategy time courses to see which phosphorylation lishing b Pu

BOX 1 PHOSPHOSPECIFIC ANTIBODIES Nature

5 Phosphospecic antibodies recognize the phosphorylated form of specific sequences or consensus binding 200

© motifs, including some that are generic for either tyrosine or serine and threonine phosphorylation. Cell Signaling Technology was the first company to release these products on the market and currently stocks 500–600 phosphospecific antibodies. “We make all of them in house, confirm the antibodies’ performance in multiple applications and provide the exact protocol for using them—and we guarantee they will work,” says Christopher Bunker, director for new business development. Phosphosolutions, with an arsenal of about 60 phosphospecific antibodies, is focusing on the nervous system. “The phosphoantibody field has caught fire in Western blots of phosphorylated and dephosphorylated proteins cancer research but there are not that many phosphospecific to demonstrate the selectivity of a phosphospecific antibody. The antibodies for proteins in the nervous system,” says company pan-specific antibody (Anti-pan) recognized both dephosphorylated president Michael Browning. “Almost all kinases are enriched tyrosine hydroxylase (Dephos-TH) and phosphorylated tyrosine in the brain. It is a target-rich environment and at the hydroxylase (Phos-TH). But the phosphospecific antibody moment we are just scratching the surface.” (Anti-phospho) and dephosphorylation-specific antibody (Anti- The antibodies are already being exploited for many dephospho) recognized only phosphorylated and dephosphorylated different applications, including immunohistochemistry TH, respectively. (Courtesy of Phosphosolutions.) and flow cytometry, to monitor kinase activity in cells. “These antibodies will tell you what a protein is doing and where it is doing it,” says Browning. In 3–5 years researchers hope to have phosphospecific antibodies on a chip to be able to simultaneously measure the phosphorylation state of dozens of distinct targets in a single sample. “The technology is there, but what we are missing is the content,” says Browning. “90 percent of the battle is having good, specific antibodies and lots of them.”

NATURE METHODS | VOL.2 NO.3 | MARCH 2005 | 227 TECHNOLOGY FEATURE

sites are changing and might point to inter- esting pathways,” says Salomon.

Screening libraries An alternative to finding sites that are phosphorylated in vivo is to use peptide libraries to determine the amino acid sequence preference of a given kinase and then draw on bioinformatics tools and databases to uncover candidate sub- strates that contain this sequence motif. In these protocols the kinase of interest is methods incubated in solution with many pools of peptides, all containing a single fixed ser- ine, threonine or tyrosine residue within a sequence consisting of all possible amino Immunofluorescent analysis of cells using an antibody specific for phosphorylation by the JNK .com/nature kinase (green) and Texas Red X–conjugated phalloidin (red) to detect actin. The cells in the right e acids. Only those peptides that contain an panel are untreated. The cells in the left panel were treated with epidermal growth factor, which amino acid sequence favored by the kinase caused activation and nuclear localization of JNK and rearrangement of the actin cytoskeleton. .natur

w are phosphorylated. The phosphorylated (Courtesy of Cell Signaling Technology and Harry Mellor, University of Bristol, UK.) peptides are then separated from the non- phosphorylated ones and sequenced. “You

http://ww are asking the contribution of a given of each amino acid is weighed against that high-throughput screens. In the oriented

residue in a sequence motif based on how of all the others in the sequence (Box 2). array peptide library (OPAL) approach,

oup much it is enriched at each position in The motif can then be exploited to scan hundreds of pools of oriented peptide r

G the peptide mixture,” says Zhou (Sunny) protein databases in search of candidate libraries are synthesized and arranged

Songyang at Baylor College of Medicine. A substrates that can be further tested. as scan arrays, in which each spot cor- consensus motif is then generated in a sta- Recent refinements to the tech- responds to a fixed amino acid at a par- lishing tistical manner, such that the contribution nique have made it better suited to ticular position flanking the tyrosine, b serine or threonine3. The pools of pep- Pu tides are phosphorylated in situ and the BOX 2 SCANSITE amount of radioactivity at each spot on Nature

the membrane is determined. “OPAL cut 5 Say you have found a phosphorylated protein by mass spectrometry and you want to the step of sequencing peptides so that 200 know where the phosphorylation sites are and which kinase might be responsible for you can do the procedure much faster,” © phosphorylating them. Or maybe you have found a potential kinase consensus motif says Songyang. by screening a peptide library and want to know which proteins contain the motif. Applying the same principle, Ben Turk Scansite (http://scansite.mit.edu) is a web-based tool designed to help with these at Yale University School of Medicine in kinds of analyses. “Our goal was never to make a public tool,” says Michael Yaffe who collaboration with Lewis Cantley’s group codirects the site with Lewis Cantley. “We had started developing Scansite to make at Harvard Medical School synthesized something to use for our own lab.” 198 separate degenerate peptide libraries, The current release of the site contains 63 motifs derived from oriented peptide each containing a fixed serine or threonine library screens, each combined with a scoring matrix that indicates the preference residue and another fixed amino acid in for each amino acid at each position within the motif. A researcher can enter a one of nine flanking positions4. However, protein sequence or ID and Scansite will identify all the motifs within the protein unlike with the OPAL approach, the phos- ranked in order of preference, indicating possible substrate-kinase interactions. To phorylation reactions were carried out in accommodate large proteomics experiments, “users can input up to 10,000 protein solution and libraries were later captured sequences at once,” says John C. Obenauer at St. Jude Children’s Research Hospital. onto membranes. “We successfully pro- Scansite can also be used to scan through different protein databases to find filed 40 kinases but the method did not proteins that contain a particular consensus motif. “It weighs and searches through work with 4. Perhaps it will not work for computation rather than by text searching,” says Songyang. 5–10% of all kinases,” says Turk. He is “You always find lots of hits but the sites may not be accessible depending on planning to work with anaspec to resyn- the conformation, for example, or the candidate substrate and kinase may be thesize the library and make portions of it differentially localized in the cell,” says Turk. “The challenge is to find matches commercially available. “The library cost between what has been done experimentally and the list obtained with Scansite.” To us about $30,000 to make. We are try- help with this process, Scansite also links to Cell Signaling Technology’s Phosphosite ing to commercialize it so that it will be database (http://www.phosphosite.org), a catalogue of in vivo phosphorylation sites. cheaper,” he says. “If a protein is actually known to be phosphorylated in cells at a site identified in The identification of a phosphorylation Scansite, it is a good candidate substrate,” says Turk. motif does not, however, directly lead to the identification of a protein substrate. “A

228 | VOL.2 NO.3 | MARCH 2005 | NATURE METHODS TECHNOLOGY FEATURE

kinase has to have a consensus sequence to recognize many different phosphopro- The Phosphosite Detector array consists phosphorylate a protein but the consensus teins, as long as they are phosphorylated of overlapping peptides from a substrate. does not in itself tell a protein ‘I am a sub- within a limited sequence context (Box “You look for signals and then based on strate.’ You need to have the right confor- 1). “We hope that in a matter of a few the alignment of the sequences determine mation and other conditions,” says Axel years we will have identified all the kinase which amino acid is phosphorylated,” says Knebel of Kinasource. “Many consensus motifs and generated phosphospecific JPT’s Mike Schutkowski. motifs are known, but it is impossible to antibodies to them,” says Turk. Researchers in the kinase field are eager- know in silico which will be substrates.” ly awaiting the development of affordable The key is to follow up the leads pro- Laying down the chips protein arrays that represent the entire vided by these in vitro screens with the Much like the ever-popular DNA micro- component of proteins in a human cell right experiments. “Once you have a list arrays, peptide arrays provide a quick and (proteome). Such arrays will provide an of candidate proteins you need to express easy test for profiling kinase function, invaluable tool for testing a kinase or mix- methods them to show that the motif is phosphory- requiring small quantities of reagents. ture of kinases against all possible protein lated in vivo,” says Michael Yaffe at the Pepscan Technologies’ PepChip Kinase targets. At the moment, several companies Massachusetts Institute for Technology. contains 1,200 peptide substrates in dupli- sell protein arrays that cover a fraction of “Two big advances in the field would be cate on a single chip in standard micro- the human proteome. .com/nature e to have access to full length cDNAs that array format. JPT Peptide Technologies are tagged so that they can be expressed synthesized arrays containing up to Finding kinases for substrates .natur

w in cells and to protein arrays with 32,000 11,000 phosphorylation-site peptides Peptide screening methods start with a human proteins.” in triplicate that they screen as a service kinase and end up, if all goes well, with The consensus motif can also be used to to pharmaceutical company clients—a a candidate substrate to study. A recent

http://ww generate antibodies to facilitate the search smaller 720-peptide version of the array publication by Kevan Shokat’s group

for substrates in vivo. Unlike phosphoan- is sold commercially to individual labs. used the reverse approach5. Because many

oup tibodies raised against a single amino acid JPT has also adapted the microarray tech- kinase substrate interactions are short r

G sequence, phosphomotif antibodies (first nology to search for phosphorylation lived, Shokat designed a new crosslinker

developed by Cell Signaling Technology) sites within kinase substrate proteins. that connects a substrate of interest to lishing b Pu

Nature

5 200 ©

NATURE METHODS | VOL.2 NO.3 | MARCH 2005 | 229 TECHNOLOGY FEATURE

its upstream kinase, thereby facilitating the purification of the entire complex with antibodies. Whereas the preliminary results are promising, many challenges remain. “We don’t know the limitations of the proce- dure yet. We need to look at it more sys- tematically,” says Shokat. The crosslinker may soon find its way to the market place as several companies have expressed an interest in developing the product. One of the exciting applications of the methods crosslinker method is that it can be used to validate results obtained with bioin- formatics tools. “If you screen for candi- date substrates of a kinase, the challenge

.com/nature The competitive inhibition of the mDECOY e is to confirm the hits. One way to do it approach. (Courtesy of Genomatix is to crosslink each candidate protein and Corporation.) .natur

w then look for the corresponding kinase in cell extracts,” says Shokat. “What you are doing is mapping a pathway in one direc- Cellular Genetics Inc. (CGI) has devel-

http://ww tion and then going back in the other.” oped a stock of genetically engineered

kinases that are fully functional but can

oup Knocking out (or down) kinase function be inhibited in a specific, inducible man- r

G One of the more traditional methods for ner by small molecule inhibitors. “The

probing the function of a protein is to knock approach can be used to systematically out its function and look at the effects in cells generate conditional alleles of protein lishing or whole animals. Methods to knock out kinases allowing for rapid, functional b kinase function include engineering genetic characterization of kinase activity,” says Pu knockout animals and using siRNA or chem- Shokat who pioneered the technology. ical inhibitors. Many companies provide the CGI scientists use their reagents, includ- Nature

necessary reagents for these studies. ing genetically modified mice carrying 5 Genomatix Corporation has devel- the mutant kinases, to provide a screen-

200 oped a different strategy. Their approach ing service to pharmaceutical companies

© combines a subcellular localization sig- to test inhibitors and find new substrates. nal with a mutant substrate ‘decoy’ that “You look at all the phosphoproteins in inhibits a kinase by competing with the a cell, then knock out a particular kinase wild-type substrate. The end result is the and do a subtraction experiment. Any knockdown of kinase function in only proteins that are no longer phosphorylat- one subcellular domain. “Our tool allows ed can be identified by mass spectrometry,” researchers to microdissect out the roles of says Shokat. a kinase in different cellular localizations,” says chief science officer Thomas Reed. Phosphorylation goes live The company’s mDECOY transgenes can Whereas kits for monitoring kinase acti- be quickly assembled from a selection of vity in vitro are widely available, there genetic modules to the customer’s speci- are fewer choices for researchers look- fications. “The vectors can then be used ing for cell-based assays. Invitrogen has to study kinase activity in cells or to make developed a panel of CellSensor live cell transgenic animals,” says Reed. assays using its GeneBLAzer technology, The system may have some advantages which combines a β-lactamase reporter over methods that knock out (or down) gene (bla) with FRET-based detection. A kinase function throughout the cell. substrate labeled with two fluorophores “Kinases are not only involved in signal is loaded into cells. In the absence of bla transduction but also in regulating diverse expression, the excitation of one fluoro- effector proteins throughout the cell,” phore results in FRET to the other, causing says Reed. “By localizing the reagent to a it to emit green light. But, if a particular specific cellular compartment scientists pathway is activated, thereby causing bla can begin to differentiate between these expression, the substrate is cleaved sepa- different roles.” rating the two fluorophores. The result is

230 | VOL.2 NO.3 | MARCH 2005 | NATURE METHODS TECHNOLOGY FEATURE

a blue fluorescent cell. “You can challenge fluorescent reporter then binds the cap- far. But the future looks bright. As more these cells with a kinase inhibitor and tured molecule. As the bead passes through cell-based assays become available and look at the effects on a particular biologi- the Luminex reader, one laser excites the number of phosphospecific anti- cal pathway,” says Armstrong. “Right now the bead to identify it, and another laser bodies and other reagents continues to we have a focused collection of CellSensor excites the fluorescent tag on the second grow, researchers will have access to a lines which cover several key pathways in antibody to quantitate the amount of cap- stockpile of powerful tools to decode the various cell backgrounds, but we plan tured analyte. “We can monitor multiple pathways regulated to significantly increase the breadth and signal transduction pathways at once with by kinases. depth of our portfolio.” our repertoire of state-specific monoclonal 1. Brill, L.M. et al. Anal. Chem. 76, 2763–2772 Another cell-based assay, Upstate’s antibodies,” says Till. “For example, we can (2004). Beadlyte technology, works on the apply this technique to follow the effects of 2. Rush J. et al. Nat. Biotechnol. 23, 94–101 Luminex xMAP platform—an assay kinase inhibitors in a physiologically rel- (2005). methods 3. Rodriguez, M. et al. J. Biol. Chem. 279, 8802– system based on fluorescently labeled evant environment to identify off-target 8807 (2004). beads and reporter molecules that can interactions.” Right now Upstate provides 4. Hutti, J.E. et al. Nat. Methods 1, 27–29 detect hundreds of soluble agents in the this technology as a service to pharmaceu- (2004). same assay well. Each Beadlyte bead is tical clients, but is developing a ‘desktop’ 5. Maly, D.J. et al. J. Am. Chem. Soc. 126, 9160– .com/nature 9161 (2004). e labeled with a protein or phosphospe- version for researchers. cific antibody that can specifically recog- The family of human kinases is thought Laura Bonetta is a freelance writer based .natur

w nize kinases and their substrates in a cell to have more than 500 members, of which in the Washington, DC area. extract. Another antibody labeled with a only a fraction have been described thus [email protected] http://ww

oup r G lishing b Pu

Nature

5 200 ©

NATURE METHODS | VOL.2 NO.3 | MARCH 2005 | 231 TECHNOLOGY FEATURE

SUPPLIERS GUIDE: COMPANIES OFFERING KINOME ANALYSIS PRODUCTS Company Web Address Amersham Biosciences http://www4.amershambiosciences.com/ AnaSpec http://www.anaspec.com/ Applied Biosystems http://www.appliedbiosystems.com/ BD Biosciences (Pharmingen) http://www.bdbiosciences.com/pharmingen/ Biaffin GmbH http://www.biaffin.com/ Biocan http://www.biocan.com/ Biorad http://www.bio-rad.com/ BioSource International http://www.biosource.com/ methods Biotrend Chemikalien GmbH http://www.biotrend.com/ BioVisioN AG http://www.biovision-discovery.de/ Calbiochem (EMD Biosciences) http://www.emdbiosciences.com/html/CBC/home.html Cambrex Corp. http://www.cambrex.com/default.asp .com/nature e CHEMICON http://www.chemicon.com/ Cell Signaling Technology http://www.cellsignal.com/ .natur

w Cellular Genomics Inc. http://www.cellulargenomics.com/ Clontech (EMD Biosciences) http://www.bdbiosciences.com/clontech/ DiscoveRx http://www.discoverx.com/ http://ww

EMD Biosciences http://www.emdbiosciences.com/

Exalpha Biologicals Inc. http://www.exalpha.com/ oup r GE Healthcare http://www.gehealthcare.com/usen/index.html G Genomatix Corp. http://www.genomatix.com/kinasemdecoys.cfm Globozymes http://www.globozymes.com/

lishing Imgenex http://www.imgenex.com/ b Invitrogen http://www.invitrogen.com/ Pu Jerini AG http://www.jerini.de/ Kinasource http://www.kinasource.co.uk/ Nature

MBL International Corporation http://www.mblintl.com/mbli/ 5 Molecular Devices http://www.moleculardevices.com/

200 Molecular Probes (Invitrogen Detection Technologies) http://www.probes.com/ © New England Biolabs http://www.neb.com/ Pepscan Systems http://www.pepscan.nl/ PerkinElmer http://www.perkinelmer.com/ Pierce Biotechnology Inc. http://www.piercenet.com/ Phosphosolutions http://www.phosphosolutions.com/ Promega Corp. http://www.promega.com/ Rockland Immunochemicals for Research http://www.rockland-inc.com/ Stratagene http://www.stratagene.com Stressgen Bioreagents http://antibody.stressgen.com/ Upstate http://www.upstate.com/ USBiological http://www.usbio.net/ Qiagen http://www1.qiagen.com/

232 | VOL.2 NO.3 | MARCH 2005 | NATURE METHODS