Get - Wiley Online Library
Contents
1. Vizca no J A C t R Reisinger F Foster J M et al A guide to the Proteomics Identifications Database proteo mics data repository Proteomics 2009 9 4276 4283 Aranda B Achuthan P Alam Faruque Y Armean I et al The IntAct molecular interaction database in 2010 Nucleic Acids Res 2010 38 525 531 Jensen L J Kuhn M Stark M Chaffron S et al STRING 8 a global view on proteins and their functional interactions in 630 organisms Nucleic Acids Res 2009 37 D412 D416 www proteomics journal com2. PFL Protein Frequency Library 2011 WILEY VCH Verlag GmbH amp Co KGaA Weinheim matrices e g agarose sepharose and magnetic beads which are essential to the IP protocol are the main contributors to non specific binding with a smaller contribution from protein binding to antibodies and tags Fig 1A SILAC 8 has ushered in a more accurate multiplexed method of condition dependent comparison which has in turn enabled the relative quantitation of putative protein interactors and contaminants in IP experiments 5 6 SILAC labelling utilises artificially increased levels 98 in specific amino acids generally arginine and lysine of stable isotopes i e carbon 13 nitrogen 15 and deuterium Cells of choice can be grown in normal light cell culture media arginine 0 lysine 0 or combinations of arginine 13C6 RG or 13C6 15N4 R10 and lysine 4 4 5 5 D4 K4 13C6 K6 or 13C6 15N2 K8 supplemented media Aside from the convenience of combining the bead control arginine 0 lysine 0 with the IP of interest arginine 6 lysine 4 and if required a third condition e g comparing www proteomics journal com 1154 S ten Have et al A Total Protein E Bait E Strong Interactors m Weak Interactors E Contaminants Complex Sample Whole Organisms or Tissue Complex cell mixtures which require either a label free scenario or y j peptide level labelling when SIL
3. Verlag GmbH amp Co KGaA Weinheim Proteomics 2011 11 1153 1159 of time and resources The benefits of such data conserva tion have been seen with initiatives such as the Cochrane Reviews 9 in the medical trial field which used only randomised controlled medical trials This meant the data going into the analyses was of higher quality randomised controlled trials are a better sampling method for seeing the true effects of medical interventions and therefore signifi cance and the outcomes of a number of trials were collec tively analysed yielding stronger statistics and more accurate conclusions This is a similar strategy to the one employed in the Lamond Laboratory www lamondlab com and the Wellcome Trust Centre for Gene Regulation and Expression http gre lifesci dundee ac uk index html with proteomic approaches We are curating all of the metadata results and machine variables to better understand scruti nise and critically appraise our data with an aim to apply the results rapidly to biology and medicine and to generate publicly available resources such as the PFL When presenting these data in publication form one should also consider the Minimum Information About a Proteomics Experiment MIAPE 10 and Minimum Infor mation about a Molecular Interaction Experiment MIMIx 11 guidelines for what to include Also depositing Proteo mics results in databases such as PRIDE http www ebi a c uk pride 1
4. 2 and interaction data into an IMEx Consortium database such as IntAct 13 allows for cumula tive data analysis and easy access by reviewers for your data Pathway analysis The log ratios or log intensity values alone of proteins which could potentially be interactors i e in the pink region of Fig 2 do not justify their identification as interaction partners Their biological functions and therefore previously known interactions can moreover provide addi tional confidence to justify their inclusion In addition to co IP experiments to verify specific interactors in silico analysis can be done by individually searching the proteins and assessing the literature for their known associations or else several software packages are available with which you can do this It is also helpful to perform follow up experiments using for example Western blot analysis and immuno fluorescence studies to provide additional independent evidence to support the protein interactions identified using MS String analysis software http string db org 14 is freely available and the protein associations are selectable i e you can specify experimental associations etc A more expensive option but more extensive software is the Inge nuity Pathway Analysis package www ingenuity com The authors thank Doulas Lamont and Kenneth Beattie at the University of Dundee s Fingerprints Proteomics facility for technical support Matthias Mann and his L
5. 55 recommended Tricks for the optimisation of IP experi mental design are given in Table 1 to help increase the efficiency of the protein recovery and to reduce and or identify putative contaminants One important step to improve the accuracy of conclu sions drawn from IP data is to characterise the range of non specific binding proteins The non specific proteins identi fied in IP experiments vary considerably and depend on parameters such as cellular fraction utilised cell type bead type etc This was described previously in Identifying specific protein interaction partners using quantitative MS and bead proteomes 6 and has since been developed into a more general approach in the form of the Protein Frequency Library PFL 5 and described below see Data manage ment section To look at and assess the statistics of the entire popula tion of identified proteins is required for labelled and un labelled scenarios alike Figs 2 and 3 The way to go about Log of Protein Intensities in 140 IP Experiments 3000 2500 2000 Protein Count 1500 1000 Protein Intensity Lo PFLFrequencies of Proteins with Intensities lt 6 25 PFLFrequencies of Proteins with PFLFrequency of Highest Intensities Intensities 6 25 7 25 gt 7 25 3500 3000 3500 4 804 2500 320007 gt 704 2000 2500 f 60 so 1500 20007 2 z 1500 07 p 2000 3 1000 z 304 i i E ii H I ol PEPP
6. AC is not possible Prepare Extract Immuno precipitation Multiple unlabelled sample replicates to ssn beanalysed or ooponn n peptide label s GTRAQ QconCAT quantification LALI ane 2011 WILEY VCH Verlag GmbH amp Co KGaA Weinheim Weak Wash 0 lt 100mM salt 0 lt 0 05 detergent Medium wash 100 lt 150mM salt 0 05 detergent Strong wash gt 150mM Salt 0 05 0 1 detergent Bead control oe a aw Single Cell Type Sample Cells in Culture SILAC Is possible more predictable pool of proteins Analysis identifying and Quantifying protein protein interactions Giaa ROKA Heavy ROKVUght cols i cols PA SILAC ratio analysis un SOMEONE Proteomics 2011 11 1153 1159 Figure 1 A The above diagram characterises the relative chan ges percentage of protein iden tified in IP results right and total percentage of protein as a fraction of cell extract left in terms of the abundance of the proteins identified in response to different experimental proce dures Whether comparing intensities directly in a label free experiment or utilising a SILAC approach to quantify proteins these changes should be taken into consideration It also indi cates the importance of having a bead control non specific proteins which bind to beads characterised for every experi ment because the non specific proteins identified in bead controls vary
7. EN ee eee h reereee oH 4 ad mot Racca PSRRERSRERE Sas veeoooreoos Frequency of Occurrence Frequency of Occurrence Frequency of Occurrence Figure 2 The graph depicts the normalised distribution of average log protein intensities detected in all protein identifications showing the normalised distribution of the population The three graphs derived from the main graph describe the frequency of occurrence of the proteins in each protein intensity region It is interesting to note that the number of proteins in the highest intensity range is 100 fold less than the numbers seen in the low and mid intensity ranges This indicates that the very high intensity proteins are only a small percentage of the proteins seen Secondly the graphs show a positive correlation between protein intensity and frequency of occurrence which suggests that high intensity proteins have a higher likelihood of being contaminants Therefore using a tool such as the Protein Frequency Library to tease apart significance of these protein identifications is helpful The data shown above consists of 21682 inde pendent protein identifications from 140 IP experiments performed in two different laboratories These IP experiments included GFP tagged protein pull downs endogenous protein pull downs and included the use of agarose sepharose and dynabeads The peak of 600 proteins at 0 is due to the ability of MaxQuant to identify proteins peptides from the MS MS spectra with insuffi
8. Proteomics 2011 11 1153 1159 DOI 10 1002 pmic 201000548 TECHNICAL BRIEF 1153 Mass spectrometry based immuno precipitation proteomics The user s guide Sara ten Have S verine Boulon Yasmeen Ahmad and Angus Lamond Wellcome Trust Centre for Gene Regulation and Expression College of Life Sciences University of Dundee Dundee Scotland UK Immuno precipitation IP experiments using MS provide a sensitive and accurate way of characterising protein complexes and their response to regulatory mechanisms Differences in stoichiometry can be determined as well as the reliable identification of specific binding part ners The quality control of IP and protein interaction studies has its basis in the biology that is being observed Is that unusual protein identification a genuine novelty or an experimental irregularity Antibodies and the solid matrices used in these techniques isolate not only the target protein and its specific interaction partners but also many non specific contaminants requiring a structured analysis strategy These methodological developments and the speed and accuracy of MS machines which has been increasing consistently in the last 5 years have expanded the number of proteins identified and complexity of analysis The European Science Foundation s Frontiers in Functional Genomics programme Quality Control in Proteomics Workshop provided a forum for disseminating knowledge and experience on this subj
9. aboratory for data contribution This work was supported in part by Wellcome Trust Program Grant 073980 Z 03 Z to A I L with additional support from European Union EU FP7 Grant Proteomics Specification in Time and Space PROSPECTS EU Network of Excellence Grant European Alternative Splicing Network EURASNET and an www proteomics journal com Proteomics 2011 11 1153 1159 interdisciplinary Radical Solutions for Researching the Proteome RASOR initiative which is supported by the Biotechnology and Biological Sciences Research Council BBSRC Engineering and Physical Sciences Research Council Scottish Higher Education Funding Council and Medical Research Council MRC The authors have declared no conflict of interest References 1 Gally J A Edelman G M Protein protien interactions among L polypeptide chains of Bence Jones proteins and human gammaz globulins J Exp Med 1964 179 817 836 Heidelberger M Kendall F E A quantitative study of the precipitin reaction between type Ill Pneumococcus poly saccharide and purified homologous antibody J Exp Med 1929 50 809 823 2 3 Kohler G Milstein C Continuous cultures of fused cells secreting antibody of predefined specificity Nature 1975 256 495 497 4 Burnette W Western blotting electrophoretic transfer of proteins from sodium dodecyl sulfate polyacrylamide gels to unmodified nitrocellulose and radiographic detection w
10. al lines indicate the arbitrary borders of significance In general proteins with high SILAC ratios usually correspond to specific interaction partners Ambiguity appears largely in the pink zone where proteins have log ratios close to 0 and can correspond either to contaminants or to specific interaction partners with low affinity and or low abundance To discriminate the PFL can be helpful 2011 WILEY VCH Verlag GmbH amp Co KGaA Weinheim www proteomics journal com Proteomics 2011 11 1153 1159 1157 Table 1 Pitfalls of immuno precipitation methodology Antibody specificity Do you know how specific your antibody is for binding to your target protein Do not rely on specificity of commercial antibodies without checking this It may be the case that it is targeted to a motif of your protein that has high homology in other proteins that have a similar function Check this possibility by blasting your protein http blast ncbi nIm nih gov Blast cgi Do some of the proteins identified in your experiment match these homologous proteins If so the significance of the assumed interaction must be confirmed Antibody affinity There is the possibility that the binding of an antibody to its target is weak or that there is competition within the sample for the binding sites This can be checked by analysing the sample flow through Additionally using a different solid matrix e g agarose sepharose or magnetic beads as an alternative could
11. be considered Antibody specificity and affinity should be checked and the IP protocol optimised prior to MS analysis Pre clearing Many commercial IP methods specify pre clearing of cell extracts with sepharose G beads This does reduce levels of non specific binding proteins but it may also be the case that the genuine target protein has a high affinity for the matrix or is of low abundance and lost during the pre clearing step Avoid this by analysing the eluate of the pre clearing beads you never know what you might find Also keep incubation times short to limit the loss of weak interactions partners Affinity tags Be aware that protein e g GST tags can also bind certain non specific proteins in the extract Additionally they may cause steric hindrance that masks the binding site of an important interactor Counter this problem by the location of the tag i e C and N terminal Bead controls Always characterise non specific binding possibilities Run all control samples exactly the same way as for the analysis of interest with a control antibody or with beads only and compare which proteins are identified In the case of SILAC this is included in the final sample run for analysis in label free scenarios this needs to be run in parallel to the IP This can be treated as your Bead Control By compiling data from separate experiments a global bead proteome can be compiled To verify the legitimacy of either a contaminant o
12. cient information from the MS spectra to determine intensities 2011 WILEY VCH Verlag GmbH amp Co KGaA Weinheim www proteomics journal com 1156 S ten Have et al Proteomics 2011 11 1153 1159 this is described below in two sections firstly for a label are significantly enriched and therefore putative free IPs and secondly for b labelled IP experiments specific interactor s for the bait protein ii Determining significance In Fig 2 the graph was a Unlabelled IP analysis generated from 140 separate IP experiments consist i Population statistics This requires the frequency of ing of 21682 protein identifications using human protein intensities to be measured note raw ion cell lines many different antibodies bead types intensities should not be compared directly but the and GFP tagged proteins from multiple researchers median of the intensities from all peptides identified for in two different laboratories using Thermo Orbitrap a protein with any skewing due to experimental and or XL and Velos mass spectrometers The analysis of machine error inaccuracy factored into these data ion intensities as an example of label free experi Examine the range of these values using the log of the mental design generated a log of peak intensity intensity values as this is more practical to deal with and population centred over 6 75 This value may vary logically divide these evenly into bins Then group for differ
13. control with identical protein loading MS and HPLC conditions unchanged i e having a ratio of 1 whereas with label free experiments the proteins with log intensities gt 7 25 b Labelled IP analysis SILAC iTRAQ etc will comprise both specific enriched proteins and i Population statistics It should be noted that although a abundant contaminant proteins The remaining proteins in the lower intensity ranges lt 7 25 may contain both contaminants and lower abundance specific interaction proteins In the case of label free experiments it is therefore important to have a well characterised bead control for your experiment to help identify likely contaminant proteins Quantification generally requires at least three technical and biological replicates of the 2011 WILEY VCH Verlag GmbH amp Co KGaA Weinheim level of significance can be determined proteins with label ratios values below this significance level may still be specific and of interest Fig 3 The normalised curve should in a labelled context be centred over a log ratio value of zero assuming mixing of labelled samples was 1 1 because the majority of proteins which are non specific binding proteins or contaminants in the samples should be unchanged and therefore have www proteomics journal com 1158 S ten Have et al equivalent ratios In cases where the centre of the curve is located over log ratio of 0 08 for example this visually indica
14. ect Our aim in this technical brief is to outline clearly for the scientists wanting to carry out this kind of experiment and recommend what in our experience are the best potential ways to design an IP experiment to help identify possible pitfalls discuss important controls and outline how to manage and analyse the large amount of data generated Detailed experimental methodologies have been referenced but not described in the form of protocols Received August 31 2010 Revised December 7 2010 Accepted December 10 2010 Keywords Cell biology Cumulative analysis Immuno precipitation Protein frequency Quality control SILAC The ability to purify and specifically produce antibodies in the late 1960s and 1970s 1 3 facilitated the development of targeted protein analysis Antibodies facilitated protein Western blotting 4 Protein interaction studies began analysing one protein at a time Today the use of MS 5 6 in combination with immuno precipitation IP 7 allows hundreds of proteins to be identified in a single experiment However usually the majority of proteins iden tified in IP experiments are non specific binders 6 The solid Correspondence Dr Sara ten Have Wellcome Trust Centre for Gene Regulation and Expression College of Life Sciences University of Dundee Dow Street Dundee DD1 5EH Scotland UK E mail s m tenhave dundee ac uk Fax 44 1382 348072 Abbreviations IP immuno precipitation
15. ent mass spectrometers and or experimental proteins by their corresponding bins This gives the set ups Be aware it is dependent on the accuracy frequency of the average intensity values of the identified and level of detection possible in the mass spectro proteins Fig 2 This is useful for two main reasons It meter but the graph should still have a bell shaped provides a measure of the quality of the data i e it distribution should show a normal or bell shaped curve if not the With label free analysis the margins of siginificance data are biased or skewed and highlights which proteins are less clear than with the SILAC or other labelling Mostly Experimental R Potential Environmental Contaminants Contaminants EE ANAC Eee 350 y Proteins which may have a ratio akin to contaminants but should be checked for significance 200 Protein Number 150 100 50 Protein Ratio log Figure 3 An example of protein ratio frequency graph showing the normalised distribution and the median value of the data The normalised bell shaped curve is centred over a log ratio of 0 this means the mixing of SILAC samples was done accurately i e exactly equal protein levels from each extract mixed If the ratios deviate significantly from this then likely an error was made when mixing and ratio values will need to be adjusted accordingly see Determining significance section The green and red vertic
16. for different cell lines antibodies beads etc B The immuno precipitation workflow Protein protein inter actions analysis utilising IP techniques can be approached in many different ways using complex samples such as tissue biopsies or single cell type samples and with labelled or label free scenarios illustrated by the flow chart www proteomics journal com Proteomics 2011 11 1153 1159 interaction partners of wild type and mutant proteins arginine 10 lysine 8 this protocol can reduce or eliminate both machine variation and human error The IP preparations from each sample are mixed in equal ratios 1 1 1 therefore proteins that do not change between conditions experimental contaminants will have an expected log 2 ratio of 0 in practice 0 32 lt log 2 ratio lt 0 26 which is not symmetrical but characterised experimentally Proteins that have been enriched relative to the control putative specific interaction partners will have increased log 2 ratios i e usually gt 0 26 and environmental contaminants gener ally have a low log2 ratio typically lt 0 32 see Fig 3 Protocols and information regarding these SILAC meth odologies are available at www LamondLab com Experimental design is dependent on the question being asked Fig 1B and therefore dictates control s required to accurately distinguish changes due to biologically relevant effects An initial exploratory IP experiment is usually 11
17. ich of the proteins in this region will be contaminants or putative interaction partners Data management The following two sections apply to label free and labelled scenarios alike As previously mentioned typically the numbers of proteins identified using MS in IP experiments range from 70 to 600 depending on washing conditions antibody affinity etc Fig 1A Generating a dynamic record of which proteins are detected under which conditions e g bead type cell type antibody etc is a beneficial accurate and in the long term time saving exercise This has been done using data management systems derived from Busi ness Intelligence methodologies providing a dynamic continually updated list of proteins with statistics of occurrence and significance in relation to experimental metadata The PFL 5 helps to evaluate objectively whether a protein identified is a genuine interactor or is likely to be a non specific binder see http www proteinfrequency library com The magnitude of data now being produced in MS analyses is not in our opinion a reason for employing purification techniques with greater stringency which risks losing important specific interaction partners The technologies have increased in speed and accuracy with the rationale of allowing more peptides to be identified and quantified in each experiment run Therefore utilising all of these data is a more economical and sensible application 2011 WILEY VCH
18. ith antibody and radioiodinated protein A Anal Biochem 1981 112 195 203 Boulon S Ahmad Y Trinkle Mulcahy L Verheggen C et al Establishment of a Protein Frequency Library and its application in the reliable identification of specific protein interaction partners Mol Cell Proteomics 2010 9 861 879 5 2011 WILEY VCH Verlag GmbH amp Co KGaA Weinheim 6 7 8 9 10 11 12 13 14 1159 Trinkle Mulcahy L Boulon S Lam Y W Urcia R et al Identifying specific protein interaction partners using quantitative mass spectrometry and bead proteomes J Cell Biol 2008 183 223 239 Bonifacino J S Dell Angelica E C Springer T A Current Protocols in Immunology Wiley New York 2001 pp 8 3 1 8 3 28 Ong S E Blagoev B Kratchmarova l Kristensen D B et al Stable isotope labeling by amino acids in cell culture SILAC as a simple and accurate approach to expression proteomics Mol Cell Proteomics 2002 1 376 386 Levin A The Cochrane Collaboration Ann Intern Med 2001 135 309 312 Taylor C F Paton N W Lilley K S Binz P A et al The minimum information about a proteomics experiment MIAPE Nat Biotech 2007 25 887 893 Orchard S Salwinski L Kerrien S Montecchi Palazzi L et al The minimum information required for reporting a molecular interaction experiment MIMIx Nat Biotech 2007 25 887 893
19. r a putative interactor check proteins against the Protein Frequency Library www proteinfrequencylibrary com Statistical analysis It is crucial to remember that an IP enriches a specific group of proteins To normalise the data the contaminating proteins which are inherent with IPs can be used see Figs 2 and 3 Washing stringency Washing steps which are common in all IP protocols are a major determining factor of the final protein identifications Fig 1 Having a high number of washing steps gt 3 with high salt concentration gt 150mM salt component will increase the risk of losing weak interacting proteins and also increase the chance of disassembling protein complexes The best way to perform IPs to increase detection of weakly interacting proteins is to use short incubation times 30 min to 1h preferably at 4 C and with minimal low salt washing Sample complexity Due to many of the above described pitfalls IPs are despite being an enriched sample still inherently complex To eliminate co elution of peptides and the statistical and quantitative issues that may arise from this performing pre fractionation of your samples is practical This can be done for example with size exclusion and or reverse phase chromatography as well as by in gel digestion or IEF fractionation techniques strategy ratios This is due to the SILAC ratios of high abundance non specific binding proteins being control IP specific IP and bead
20. tes there has been a mixing error where more heavy labelled proteins were mixed in with the light label and all ratios should be adjusted accordingly i e all ratios should be recalculated with the increase of log ratio 0 08 compensated for The MaxQuant output is in txt file format and generates ratio information in H L H M and M L which are also reversible to necessitate label swapping experiments and also intensity informa tion for label free analysis allowing convenient manip ulation via either custom software or Microsoft Excel and comprises detailed SILAC information peptide identifi cation and statistical significance values on the peptide and protein levels ii Determining significance This is done initially by generating the graph described in Fig 3 The cut off designated on the graph shown is arbitrary and should be decided by the scientist It is important to note that there are inevitably some limitations in this experimen tal method due to non stoichiometric binding of low abundance and or weakly binding genuine interaction partners This means the proteins identified in the region coloured pink in Fig 3 may nonetheless contain some specific proteins of interest Within the current scope of one single experiment this significance cannot be determined unambiguously Therefore the use of the PFL with its cumulative statistical strength based on large numbers hundreds of IP experiments can help to predict wh
Download Pdf Manuals
Related Search