ARTICLE pubs.acs.org/jpr
New Structural Proteins of Halobacterium salinarum Gas Vesicle Revealed by Comparative Proteomics Analysis Lichieh Julie Chu,†,‡ Mengchieh Claire Chen,†,‡ Jocelyn Setter,§ Yihsuan Shannon Tsai,§,|| Hanyin Yang,|| Xuefeng Fang,§,^ Ying Sonia Ting,§,|| Scott A. Shaffer,§ Gregory K. Taylor,§ Priska D. von Haller,# David R. Goodlett,§ and Wailap Victor Ng*,‡,||,z, Institute of Biotechnology in Medicine, Institute of Biomedical Informatics, and zDepartment of Biotechnology and Laboratory Science in Medicine, National Yang Ming University, Taipei, Taiwan, Republic of China § Department of Medicinal Chemistry, University of Washington, Seattle, Washington 98195, United States ^ Systems Biology Division, Zhejiang-California Nanosystems Institute, Zhejiang University, Hangzhou 310029, People's Republic of China # UW Proteomics Resource, University of Washington, Seattle, Washington 98109, United States Clinical Biotechnology Research Center, Taipei City Hospital, Taipei, Taiwan, Republic of China )
‡
bS Supporting Information ABSTRACT: The Halobacterium salinarum gas vesicle (GV) is an extremely stable intracellular organelle with air trapped inside a proteinaceous membrane. Reported here is a comparative proteomics analysis of GV and GV depleted lysate (GVD) to reveal the membrane structural proteins. Ten proteins encoded by gvp-1 (gvpMLKJIHGFED-1 and gvpACNO-1) and five proteins encoded by gvp-2 (gvpMLKJIHGFED-2 and gvpACNO-2) gene clusters for the biogenesis of spindle- and cylindrical-, respectively, shaped GV were identified by LC-MS/ MS. The peptides of GvpA1, I1, J1, A2, and J2 were exclusively identified in purified GV, GvpD1, H1, L1, and F2 only in GVD, and GvpC1, N1, O1, F1, H2, and O2 in both samples. The identification of GvpA1, C1, F1, J1, and A2 in GV is in agreement with their previously known structural function. In addition, the detection of GvpI1, N1, O1, H2, J2, and O2 in GV suggested they are new structural proteins. Among these, the structural role of GvpI1 and N1 in GV was further validated by immuno-detection of protein A-tagged GvpI1 and N1 fusion proteins in purified GV. Thus, LC-MS/MS could reveal at least a half dozen gas vesicle structural proteins in the predominant spindle-shaped GV that may be helpful for studying its biogenesis. KEYWORDS: gas vesicle, structural protein, archaea, Halobacterium salinarum, proteomics
’ INTRODUCTION Gas vesicles have been found almost exclusively in aquatic prokaryotes including the halophilic archaea.1 Halobacterium salinarum, an extremely halophilic archaeon living in high salinity environments, synthesizes intracellular gas filled vesicles to modulate cell buoyancy.1-6 The genes required for gas vesicle synthesis have been extensively studied in H. salinarum in the past two decades. The major gas vesicle structural proteins Gvp-A (GvpA1) and its homologue Gvp-B (GvpA2) of the spindle- and cylindrical-shaped GVs, respectively, were initially identified by sequencing of the proteins in purified GV.7 After cloning and sequencing of the gvpA (gvpA1) present on the 191-kb minichromosome pNRC100 by DasSarma et al. (1987), analysis of GV-deficient mutants and further sequencing of the gvpA flanking regions resulted in the identification of the gvp-1 cluster.8-12 H. salinarum has two different gene clusters to synthesize the two types of gas vesicles. Since the discovery of the major gas vesicle structural protein gene gvpA (gvpA1), a total of two gas r 2010 American Chemical Society
vesicle gene clusters, namely p-vac (gvp-1) and c-vac (gvp-2), have been identified and implicated in gas vesicle synthesis.2,7,10-14 Upon the completion of the H. salinarum NRC-1 genome sequence, a total of three copies of gas vesicle gene clusters were found in the minichromosomes.15-17 These included an identical copy of gvp-1 on both pNRC100 (191 kb) and pNRC200 (365 kb) replicons and a unique gvp-2 copy on pNRC200. The gvp-1 cluster (p-vac) consisted of two sets of divergently transcribed genes gvpACNO-1 and gvpDEFGHIJKLM-1 is required for the synthesis of the predominant spindle-shaped gas vesicles. The gvp-2 cluster (c-vac) contained the gvpACNO-2 and gvpDEFGHIJKLM-2 is responsible for the biogenesis of the cylindricalshaped gas vesicles.7,10,12-14 The genes essential for the spindleshaped GV synthesis have been studied by gene deletion and linker scanning mutagenesis analyses. The mapping and deletion Received: September 15, 2010 Published: December 15, 2010 1170
dx.doi.org/10.1021/pr1009383 | J. Proteome Res. 2011, 10, 1170–1178
Journal of Proteome Research analysis of the minimal gvp-1 gene cluster for gas vesicle synthesis suggested 8 of the 14 genes are necessary and sufficient for gas vesicle biogenesis.11,18,19 Another study that analyzed the genes in the gvp-1 cluster by linker scanning mutagenesis indicated at least 10 genes in the cluster are required for gas vesicle formation.20 Over the past two decades, the structural and nonstructural proteins encoded by the Halobacterium gvp-1 gene cluster have been studied by biochemical, immunological, and genetics approaches, but the overall composition of the proteinaceous envelope of this gas-filled organelle remains unclear. In addition to the major gas vesicle structural protein GvpA1 that was discovered by protein sequencing, six other structural proteins (GvpC1, F1, G1, J1, L1, and M1 also known as GvpC, F, G, J, L, and M) in the spindle-shaped GV had been identified by immunological assays using polyclonal antibodies raised against individual recombinant proteins or synthetic peptides derived from the pNRC100 gvp sequences.21,22 On the other hand, functional analysis of GvpDs and GvpEs encoded by p-vac (gvp-1) and c-vac (gvp-2) indicated that they are regulators responsible for repression and transcription activation, respectively, of gas vesicle formation.4,5,23-25 Furthermore, in a gene deletion analysis a positive transcription regulatory role had been suggested for p-GvpO (GvpO1), which is required for the transcription of p-gvpACN (gvpACN-1) mRNA.26 Taken together, at least one role, structural or regulatory, has been assigned to 10 of the 14 proteins encoded by the gvp-1 gene cluster in previous studies. The functions of the remaining proteins, namely GvpH1, I1, K1, and N1, have yet to be elucidated. Due to the unusually low solubility in commonly available solubilizing agents and resistance to proteases,27 the investigation of the structural components of the major spindle-shaped gas vesicles in H. salinarum has been technically challenging. Another difficulty is posed by the wide dynamic range of the structural proteins. Except for the major structural protein GvpA1, the other Gvps are present in a relatively low copy number. With the advances in mass spectrometry technology, we have initiated the comparative proteomic analysis on purified gas vesicle and gas vesicle-depleted cell lysate to reveal the structural components of this extraordinarily stable protein complex.
’ MATERIALS AND METHODS GV and GV-Depleted Cell Lysate
Halobacterium salinarum NRC-1 (ATCC700922) starter culture was prepared by inoculating a single colony in 5 mL of Halobacterium medium containing trace metals (CMþ)28 and grown at 37 C with shaking at 225 rpm for one week. An aliquot of 300 μL of the culture was spread on 15-cm diameter CMþ plates containing 2% of agarose. After 7-10 days of incubation at 37 C, the GVs from cells on 10 plates were harvested using the centrifugation facilitated flotation method as previously described except the GVs were washed for a total of 5 times instead of the suggested 3 times to remove nonspecific bound proteins.29 After collecting the gas vesicles in the primary cell lysate, the cell membrane and residual GV were removed by ultracentrifugation at 53 000 g, 4 C for 16 h twice to obtain the GV-depleted lysate (GVD). The protein concentration was then estimated by a colorimetric method using the DC Protein Assay kit (Biorad, Hercules, CA). Trypsin Digestion of GV and GVD
Approximately 100 μg of GV or GV-depleted lysate were digested with sequencing grade modified trypsin (Promega, Madison, WI) in a total volume of 500 μL of 50 mM of ammonium bicarbonate
ARTICLE
(pH 8.3) at 37 C for 16 h at a substrate to enzyme ratio of 50:1 (wt:wt). The hydrolysates were vacuum-dried in a SpeedVac (Thermo Scientific, Waltham, MA) and purified using a C18 mini-spin column (Pierce, Rockford, IL) according to the manufacturer’s procedure. The peptides were lyophilized and stored at -20 C until use for MS analysis. Mass Spectrometry
Peptides resuspended in a solution containing 5% acetonitrile [0.1% formic acid (v/v)] were analyzed using a nano-HPLC (nanoAcquity systems; Waters, Milford, MA) coupled to a hybrid linear ion trap Orbitrap (LTQ-Orbitrap) mass spectrometer (Thermo Scientific, San Jose, CA) similar to previously described.30-32 The MS survey scans were acquired over a full m/z range of 400-2000 and within four smaller windows of 400-520, 516-690, 685-968, and 963-2000 m/z. The five most intense precursor ions were sequentially isolated using a data-dependent mode and subjected to collision induced dissociation (CID) in the linear ion trap. Computational Analysis of Mass Spectra
The raw data in Thermo XCalibur (version 2.0) binary format were converted to the mzXML open data format using a modified version of the ReAdW program.33 The SEQUEST (version 27) algorithm34 was applied to search the MS data against a targetdecoy database35 containing the forward and reversed (decoy) sequences of 2427 H. salinarum NRC-1 unique proteins from the EMBL-EBI Integr8 proteome database (www.ebi.ac.uk/integr8/ EBI-Integr8-HomePage.do).16 The search parameters included a peptide mass tolerance of (2.1 Da and possible oxidation of methionine (þ16.0 Da) residues. The protein cleaving agent was not specified for nonconstraint database searches. The search results were further analyzed using the Trans-Proteomic Pipeline.36 A probability of 0.9 (equivalent to an estimated 10% error rate) was used as both the PeptideProphet (p) and ProteinProphet (P) cutoffs. Only the proteins with two or more unique peptides identified were considered as positive identifications. Additional identification criteria included a maximum of two missed tryptic cleavage and precursor ion carried up to five protons. Spectra matched to identical peptide sequences shared by two or more proteins were indicated in the search results. Gvp-Protein A Fusion Protein Constructs
Recombinant DNAs were constructed using the general cloning procedures as previously described.37 The complete coding sequences of H. salinarum NRC-1 gvpA1, D1, E1, I1, and N1 were amplified by PCR with 50 -phosphorylated primers (gvpA1F: 50 -ATGGCGCAACCAGATTCTTCA-30 and gvpA1R: 50 -GGCCTCGGGTGCCGCCTCGGC-30 ; gvpD1F: 50 -ATGAGTTCACCCAATCTAGCT-30 and gvpD1R: 50 -GACCATCTCCGTGAGCGTGAT-30 ; gvpE1F: 50 -ATGGACGACTTGCTGGAGGAA-30 and gvpE1R: 50 -GTCATTGGTCTCTCTTCCTTG-30 ; gvpI1F: 50 -ATGAGCGACAAACAACAGCAA-30 and gvpI1R: 50 -CTCATCGTTCACCTCGTCCTC-30 ; and gvpN1F: 50 -ATGACGAACGAGTCCCGTAAA-30 and gvpN1_R: 50 -AGAAAGGGCGACTTCCATGTC-30 ) and KOD Hot start DNA polymerase (Novagen, Gibbstown, NJ). Agarose gel purified PCR amplicons were inserted upstream of the protein A coding sequence of NdeI digested, CIAP dephosphorylated, and T4 DNA polymerase blunt-ended Halobacterium-E. coli shuttle expression vector pNBPA.38 The gvp gene and the vector-insert junctions containing the upstream ferrodoxin promoter and downstream protein A tag sequence in the recombinant DNAs purified from E. coli DH5R transformants were 1171
dx.doi.org/10.1021/pr1009383 |J. Proteome Res. 2011, 10, 1170–1178
Journal of Proteome Research
ARTICLE
Table 1. Summary of Peptides and Proteins in GV and GVD Identified by MS/MS Shown with the Numbers of Peptides Detected with Different PeptideProphet Probabilitiesa GV PeptideProphet probabilities
GVD
1
0.9
0.8
0.7
0.6
0.5
1
0.9
0.8
0.7
0.6
0.5
Total number of matched peptides
3390
5569
477
304
259
194
6418
9418
988
607
448
415
NRC-1 matches
3389
5531
451
283
233
176
6417
9375
950
572
411
368
Decoy matches False detection rate (FDR), %
1 0.06
38 1.36
26 10.90
21 13.82
26 20.08
18 18.56
1 0.03
43 0.91
38 7.69
35 11.53
37 16.52
47 22.65
Number of GV peptides
255
356
21
12
10
3
49
105
11
6
2
3
Number of non-GV peptides
3134
5175
431
271
223
173
6368
9270
939
566
409
365
Total number of peptides (p g 0.9)
8920
15792
% GV peptides
6.85
0.98
Total number of proteins
479
696
9 470
7 689
1
0
(P g 0.9 and g 2 unique peptides) Number of GV proteins Number of non-GV proteins Total number of decoy proteins (P g 0.9 and g 2 unique peptides) a
One spectrum matched to both NRC-1 and Decoy sequences was counted as Decoy hit.
analyzed by automated DNA sequencing (ABI 3730 DNA analyzer; Applied Biosystems, Foster City, CA).
Using the same criteria, only one decoy protein was found in the GV data and none were found in the GVD data.
Validation of GV Structural Proteins
GV Proteins in GV and GVD
H. salinarum NRC-1 was transformed with the Gvp-protein A fusion protein constructs using a PEG-mediated transformation procedure.39 The transformants grown on CMþ agarose plate containing 20 μg/mL mevinolin (Mev) were examined by colony PCR with Pro Tag Plus DNA polymerase (Protech, Taipei, Taiwan) and vector specific primers (50 -CGTGGCAGTACGCTGGCCCGA-3 0 and 5 0 -GCGGCGTGGAGCCAGTATTGG-3). 38 Positive transformants were inoculated in 5 mL of medium (CMþMev) and plated on 10 agarose plates (CMþMev) for GV, GVD, and total cell lysate preparations as described above except the GV were washed for a total of 3 times as previously suggested.29 GV, GVD, and cell lysate proteins were immobilized on PVDV membrane (Millipore, Bedford, MA) using a slot blot apparatus (Biorad, Hercules, CA). The membranes were blocked and incubated with human IgG-HRP (Amersham Biosciences, Piscataway, NJ) antibody and then chemiluminescent HRP substrate (Millipore) to detect the Gvp-protein A fusion proteins with a Fujifilm LAS-4000 imaging system (Fujifilm, Toyko, Japan) according to the manufacturer’s suggested procedures. (H. salinarum is a nonpathogenic microorganism. Recombinant DNA construction and gene expression were conducted in accordance with the safety requirements of the Institutional Biosafety Committee in National Yang Ming University).
A higher number of peptides from gas vesicle proteins were identified in GV than GVD samples. Among all of the identified peptides, 611 (6.85%) and 154 (0.98%) peptides from GV and GVD, respectively, were derived from the gas vesicle proteins. This included a total of 15 Gvps, among which 10 are encoded by gvp-1 and five by gvp-2 gene clusters. As shown in Table 2, the major gas vesicle structural proteins GvpA1 (spindle-shaped GV) and GvpA2 (cylindrical-shaped GV), and several likely minor structural proteins, GvpI1, J1 and J2, were only detected in purified GV. Large number of GvpA1 peptides were only found in GV suggested it is a stable bound structural protein. GvpH1, D1, L1, and F2 were only found in GVD. However, the peptides of GvpC1 (a gas vesicle surface bound protein), F1, N1, O1, H2, and O2 were identified in both GV and GVD samples. A comparison of the number of peptides from GvpA1 and GvpA2 indicated the spindle-shaped GV is more predominant than the cylindrical-shaped GV. GvpA1 (76 aa) and A2 (79 aa) are highly homologous proteins differing by only five amino acids in three locations. Approximately 30 times more unique peptides of GvpA1 than GvpA2 were identified by MS/MS (Table 2). The locations and frequencies of peptides identified in the GvpAs are shown in Supplementary Figure 1, Supporting Information.
’ RESULTS Protein Identifications
Upon SEQUEST search against the target-decoy database of the H. salinarum NRC-1 proteome, 8920 and 15792 tandem mass spectra of GV and GVD, respectively, matched to NRC-1 sequences with a PeptideProphet (p) peptide identification probability above 0.9 and a false detection rate
