Witkowska_DVH_TAP

Bacterial interactomes: interacting protein partners share similar function and are validated in independent assays more frequently than previously reported

M. Shatsky, S. Allen, B.L. Gold, N.L. Liu, T.R. Juba, S.A. Reveco, D.A. Elias, R. Prathapam, J. He, W. Yang, E.D. Szakal, H. Liu, M.E. Singer, J.T. Geller, B.R. Lam, A. Saini, V.V. Trotter, S.C. Hall, S.J. Fisher, S.E. Brenner, S.R. Chhabra, T.C. Hazen, J.D. Wall, H.E. Witkowska, M.D. Biggin, J.-M. Chandonia, G. Butland

Spectral library for Desulfovibrio vulgaris Hildenborough (DvH) proteins identified in the course of the large-scale mapping of protein-protein interactions using tandem affinity purification (TAP) methods: in addition to DvH peptides, common contaminants and standards that were spiked into washes separating sample injections are also included. DvH wild-type ATCC29579 was genetically engineered to encode locus specific, affinity purification (AP)-tagged fusion proteins using high-throughput construction and electroporation of non-replicating “suicide constructs”, primarily using the Sequence and Ligation Independent Cloning (SLIC) method for plasmid construction. Construct generation and affinity purification methods were previously described (Chhabra et al., 2011. Appl. Environ. Micriobiol. 77:7595-7604). Out of 1401 genes that were successfully fused with AP tag, 957 produced fusion proteins that were detected by MS as baits.  In total, 1323 bait and prey proteins were identified by at least two peptides. AP-pulldowns were processed in parallel using gel-free and gel-based workflows. In gel-free workflow protein mixtures were digested with trypsin on PVDF membrane using Millipore 96-well PVDF plates, as described in Chhabra et al., 2011). In gel-based workflow, proteins were fractionated by SDS-PAGE prior to in-gel tryptic digestion. With few exceptions, the majority of gel-free derived samples were analyzed by LC MS using Thermo LTQ Velos Orbitrap. The majority of gel-based samples were analyzed by LC MS using Thermo LTQ XL linear ion trap. Peak lists were extracted from .raw files using the Mascot Distiller 2.3.2.0 software (Matrix Science). Data were searched with an in-house Mascot version 2.2.04 search engine (Matrix Science) against a custom protein database containing all potential protein products generated via 6-frame translation of the D. vulgaris genome supplemented by frequently observed contaminants and concatenated with the decoy database generated by reversing all D. vulgaris protein sequences (102,572 sequences; 9,848,210 residues) (Elias and Gygi, 2007. Nat Methods 4:207-214).

Clustergrammer Heatmap
Flag FileDownloadCreatedProteinsPeptidesPrecursorsTransitionsReplicates
DvH TAP 02032016 Panorama.sky.zip (234 MB)2016-02-082,19738,01249,81649,8160
Bacterial interactomes: interacting protein partners share similar function and are validated in independent assays more frequently than previously reported

  • Organism: Desulfovibrio vulgaris Hildenborough wild-type ATCC29579
  • Instrument: Thermo LTQ Velos Orbitrap, Thermo LTQ XL and Sciex QSTAR XL
  • SpikeIn: No
Abstract
Of the 3525 predicted D. vulgaris protein coding genes, 2086 were originally selected for the AP-protein bait generation using a combination of criteria i.e., detection of protein complexes using tagless strategy of multidimensional protein fractionation (Mark Biggin et al, unpublished observations), functional interest (e.g., energy generation), and expected complexes based on E .coli interologs. AP-tagged alleles were constructed for 1401 of genes, leading to MS detection of 957 fusion AP-bait proteins and more than 1600 associated prey proteins. Data analysis employed MS data-based filtering (match quality, potential carryover, spectral counts), calculating dice and completeness scores and feature-based filtering. Gold standard positive and negative interaction sets were constructed as a benchmark for FDR calculation. A network consisting of 459 high confidence protein-protein interactions was defined. The library contains all the D.vulgaris-related peptide MS/MS spectra that were generated in the course of the project.
Experiment Description
D. vulgaris Hildenborough wild-type ATCC29579 was genetically engineered to encode locus specific, affinity purification (AP)-tagged fusion proteins using electroporation of non-replicating “suicide constructs” (Chhabra et al.,2011. Appl. Environ. Microbiol. 77:7595-604). We constructed plasmids for generating chromosomal AP-tagged alleles for 1963 of the priority genes, 1681 of which were successfully integrated into the D. vulgaris chromosome. From this set, 1498 strains expressing an AP-tagged fusion protein were verified by Western blot, of which 1,415 were constructed using Sequence and Ligation Independent Cloning (SLIC), 77 using Gateway and 6 using recombineering procedures. The primary AP-tag utilized was Strep-TEV-FLAG (1,231 strains); however Strep-TEV-FLAG-6xHis (237 strains) and Sequential Peptide Affinity tag (30 strains), were also used. A non-redundant total of 1,401 unique genes are represented as AP-tagged alleles in the 1,498 strains constructed. All affinity purifications were performed as previously described (Chhabra et al.,2011). In all cases, Strep-TEV-FLAG-6xHis strains were treated exactly as Strep-TEV-FLAG strains for the purposes of affinity purification of protein complexes. All protein pull-downs were analyzed using gel-free workflow. For the majority of constructs, a parallel gel-based workflow was also employed. Shotgun LC MS analyses were performed using either LTQ Velos Orbitrap or LTQ XL linear trap Extensive column washing procedures and standard protein runs were employed in-between injections of samples derived from different pull-downs to minimize and control sample carryover.
Sample Description
Samples that originated from the gel-based workflow were in-gel digested with trypsin using Progest robot (Digilab). In a gel-free approach, AP-isolated proteins were digested with trypsin utilizing a 96-well PVDF membrane-based protocol. The majority of gel-based samples were analyzed by LC MS using LTQ XL linear trap while the majority of the gel-free samples were analyzed using LTQ Velos Orbitrap, essentially as described by Chhabra et al. 2011, and Roan et al., 2014. J Infect. Dis. 210:1062-6, respectively. Peak lists were extracted from MS raw files using the Mascot Distiller 2.3.2.0 software (Matrix Science). Data were searched with an in-house Mascot version 2.2.04 search engine (Matrix Science) against a custom protein database containing all potential protein products generated via 6-frame translation of the D. vulgaris genome supplemented by frequently observed contaminants and concatenated with the decoy database generated by reversing all D. vulgaris protein sequences (102,572 sequences; 9,848,210 residues). Search was limited to doubly- and triply-charged precursors; peptides matched at a confidence of at least 95% were included for analysis. Skyline library contains data on DvH proteins identified in the course of the study. Peptides that were not automatically matched to parent proteins are included in a "Library Peptides" category and their protein assignment is provided in the enclosed supplemental table entitled "Skyline Library Peptides 02032016".
Created on 2/8/16, 3:48 PM