Integrated omics study delineates the dynamics of lipid droplets in

Integrated omics study delineates the dynamics oflipid droplets in Rhodococcus opacus PD630Yong Chen12 Yunfeng Ding1 Li Yang13 Jinhai Yu13 Guiming Liu4 Xumin Wang4

Shuyan Zhang1 Dan Yu4 Lai Song34 Hangxiao Zhang4 Congyan Zhang13 Linhe Huo4

Chaoxing Huo13 Yang Wang13 Yalan Du5 Huina Zhang1 Peng Zhang13 Huimin Na13

Shimeng Xu13 Yaxin Zhu6 Zhensheng Xie1 Tong He7 Yue Zhang34 Guoliang Wang4

Zhonghua Fan4 Fuquan Yang1 Honglei Liu2 Xiaowo Wang2 Xuegong Zhang2

Michael Q Zhang28 Yanda Li2 Alexander Steinbuchel910 Toyoshi Fujimoto11

Simon Cichello12 Jun Yu4 and Pingsheng Liu1

1National Laboratory of Macromolecules Institute of Biophysics Chinese Academy of Sciences Beijing 100101China 2MOE Key Laboratory of Bioinformatics and Bioinformatics Division Center for Synthetic and SystemsBiology TNLISTDepartment of Automation Tsinghua University Beijing 100084 China 3University of theChinese Academy of Sciences Beijing 100049 China 4Key Laboratory of Genome Sciences and InformationBeijing Institute of Genomics Chinese Academy of Sciences Beijing 100101 China 5Department of Histologyand Embryology University of South China Hengyang Hunan Province 421001 China 6State Key Laboratoryof Microbial Resources Institute of Microbiology Chinese Academy of Sciences Beijing 100101 China7School of Applied Mathematics Central University of Finance and Economics Beijing 102206 China8Department of Molecular and Cell Biology Center for Systems Biology The University of Texas at DallasDallas TX 75083-0688 USA 9Institut fur Molekulare Mikrobiologie und Biotechnologie Westfalische Wilhelms-Universitat Corrensstrasse 3 D-48149 Munster Germany 10Environmental Sciences Department KingAbdulaziz University Jeddah 21589 Saudi Arabia 11Department of Anatomy and Molecular Cell BiologyNagoya University Graduate School of Medicine 65 Tsurumai-cho Showa Nagoya 466-8550 Japan and12School of Life Sciences La Trobe University Melbourne Victoria 3086 Australia

Received August 6 2013 Revised September 22 2013 Accepted September 23 2013

ABSTRACT

Rhodococcus opacus strain PD630 (R opacusPD630) is an oleaginous bacterium and also isone of few prokaryotic organisms that contain lipiddroplets (LDs) LD is an important organelle for lipidstorage but also intercellular communicationregarding energy metabolism and yet is a poorlyunderstood cellular organelle To understand thedynamics of LD using a simple model organismwe conducted a series of comprehensive omicsstudies of R opacus PD630 including completegenome transcriptome and proteome analysisThe genome of R opacus PD630 encodes 8947genes that are significantly enriched in the lipidtransport synthesis and metabolic indicating a

super ability of carbon source biosynthesis andcatabolism The comparative transcriptomeanalysis from three culture conditions revealed thelandscape of gene-altered expressions responsiblefor lipid accumulation The LD proteomes furtheridentified the proteins that mediate lipid synthesisstorage and other biological functions Integratingthese three omics uncovered 177 proteins thatmay be involved in lipid metabolism and LDdynamics A LD structure-like protein LPD06283was further verified to affect the LD morphologyOur omics studies provide not only a first integratedomics study of prokaryotic LD organelle but also asystematic platform for facilitating further prokary-otic LD research and biofuel development

To whom correspondence should be addressed Tel +86 10 64888517 Fax +86 10 64888517 Email pliuibpaccnCorrespondence may also be addressed to Jun Yu Tel +86 10 82995357 Fax +86 10 82995373 Email junyubigaccn

The authors wish it to be known that in their opinion the first four authors should be regarded as Joint First Authors

1052ndash1064 Nucleic Acids Research 2014 Vol 42 No 2 Published online 22 October 2013doi101093nargkt932

The Author(s) 2013 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby30) whichpermits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited

Downloaded from httpsacademicoupcomnararticle-abstract42210521026394by gueston 12 April 2018

INTRODUCTION

Lipid droplets (LDs) are cellular organelles widely foundin fungal plant animal and human cells (1ndash3) They areencapsulated by a phospholipid monolayer and are com-positionally different from other membrane structures (4)They differ in that their primary role is lipid storage butmay also be pivotal in cellular communication with organ-elles such as the mitochondria to regulate energy metab-olism and substrate utilization LD is an importantorganelle related to human metabolic diseases andbiofuel productions For example LD dysfunction isone of the main causes of metabolic disorders such asobesity insulin resistance type 2 diabetes and cardiovas-cular diseases (5ndash9) In biofuel studies triacylglycerol(TAG) in LD of green algae has been investigated anddeveloped for high oil yields by using targeted metabolicengineering (10ndash12) making it a biological candidate forbiofuel production

Delineating the molecular mechanisms of LD dynamicsis essential to understand its formation functions syn-thetic engineering and further biofuel applications Sinceperilipin the first protein of perilipin family (PLIN) wasidentified in 1991 (13) numerous proteins have beenrevealed to be related to LD functions and dynamics(314) LD may also be involved in multiple importantcellular processes such as intermembrane lipid traffic(15) lipid storage (16) lipolysis (17) signaling temporalprotein storage (18) and protein degradation (19) LD isreported functionally interacted with many other organ-elles such as the mitochondria (20) endoplasmic reticulum(2122) endosome (23) and peroxisome (24) Despite thefunctional importance of LDs systematic understandingof the organellersquos biogenesis and dynamics remainselusive In contrast to eukaryotes that have multiple or-ganelles LD is the only membranous organelle found in anumber of bacterial strains that can be used as ideal modelorganisms for LD research Among them Rhodococcusopacus PD630 has the ability to accumulate largeamounts of TAG in the LD (25)

The importance of Rhodococcus opacus strain PD630(R opacus PD630) as a model system is also exemplifiedby its powerful ability to convert carbon sources intolipids Interestingly the TAG storage in R opacusPD630 accounts for up to 87 of the cellular dryweight (26) and thus has higher lipid storage capacitywhen compared with other oleaginous organisms (2627)Early studies reported that R opacus PD630 has 10diacylglycerol acyltransferases (DGAT) that assimilatecellular fatty acids into TAG (1328) Holder et alreported a partial genome and also performed a compara-tive genomic study with a lipid mass analysis (29) whichidentified 16 DGAT and 261 genes that are directlyinvolved in 20 TAG cycle reactions These previousstudies suggest that TAG biosynthesis from carbonsources is a pronounced characteristic of R opacusPD630 Therefore to facilitate the application ofR opacus PD630 LD production for biofuel developmenta complete genome of the organism and integratedanalysis of its transcriptome a proteome of its lipid syn-thesis storage and metabolism are essential

We performed multi-omic studies and present herein thecomplete genome sequence a comparative transcriptomeand a comparative LD proteome of R opacus PD630After integrating the collected data a number of proteinfamilies involved in LD dynamics were identified includinglipid synthesis LD structure-like proteins dynamin-likeand SNARE-like proteins A structure-like proteinLPD06283 was verified by its LD location and its effecton LD size Together these omics are useful tools to in-vestigate the mechanisms of LD dynamics that willenhance our understanding of the lipid storage of LD inbiofuel development

MATERIALS AND METHODS

DNA extraction and genome sequencing and assembly

Cells of R opacus PD630 (30) were obtained from DrSteinbuchelrsquos lab at the University of Munster Cellswere cultured aerobically in 100ml of nutrient broth(NB) at 30C to postlogarithmic phase and then theDNA was extracted The complete nucleotide sequencewas obtained using a combination of paired-endmate-pair Illumina sequencing and 454 sequencing Thesequence gaps were completed by direct sequencing ofpolymerase chain reaction (PCR)-amplified fragmentsFor 454 pyrosequencing genomic DNA was sheared upby nebulization into random fragments of 500ndash800 bp forthe construction of a dispersed library which was thenclonally amplified and sequenced on a 454 GenomeSequencer For Illumina sequencing genomic DNA wasprocessed to construct paired-end libraries with size spansof 300 bp and also mate-pair libraries with size spans of3 kb using an Illumina Genomic DNA Sample Prep kitThe total number of 454 reads obtained was 861 751

giving a 36-fold coverage while the total number ofpaired-end and mate-pair library reads was 40 110 584giving a 445-fold coverage We used two assemblyprograms and combined the primary contigs and paired-end data to build scaffolds in successive assemblies Fourhundred fifty-four sequences were assembled using theRoche GS assembler Newbler (version 25) with defaultparameters The primary contigs were then scaffoldedwith Illumina mate-pair reads using SSPACE-premium(version 21) (31) To close the gaps among scaffoldsread pairs that were uniquely mapped to the contig tailswere extracted for manual assembly Primers weredesigned for the remaining gaps and PCR walking wasused to finish the whole genome Illumina reads (300 bp)were mapped to this assembled whole genome sequence toidentify potential single miss-called nucleotides using theBowtie method (32)

Genome analysis and annotation

Gene models were predicted independently usingGLIMMER (33) and GeneMark (34) The predictedopen reading frames (ORFs) were further evaluated andadjusted using RBSfinder (35) The translated sequencesof the predicted protein-coding genes were searchedagainst UniProt (36) and InterPro (37) The function ofenzymes was assigned using EFICAz2 (38) and searched

Nucleic Acids Research 2014 Vol 42 No 2 1053


against the KEGG database (39) We used the COG clas-sification scheme (40) to further classify gene functionsTwo-tailed Fisher exact test was used to compare the dis-tributions of COG categories between two species Foreach COG category a 2 2 contingency table was con-structed by recording the numbers of genes included ornot included Putative tRNAs were identified usingtRNAscan-SE (41) and TFAM 10 (42) rRNAs weredetected by RNAMMER (43) and confirmed againstknown rRNAs using BLASTN Transposons and repeatelements were identified using ISfinder (44) and searchedagainst Repbase (45) Protein domains were predictedusing the Pfam (46) and NCBI CDD (47) databasesHorizontally transferred genes (HTGs) were predictedusing the WN method (48) For protein sequenceanalysis multiple alignments were generated withCLUSTALX (49) and phylogenetic analysis was per-formed with MEGA40 (50) The operons of PD630were predicted by using a statistic operon predictionmethod (51) All PD630 proteins were compared withthe proteins of R opacus B4 (R opacus B4)Rhodococcus jostii RHA1 (R jostii RHA1) (52) and anearlier reported partial genome of PD630 (29) that weredownloaded from NCBI database (NCBI release data ofMarch 2012) by using BLASTP (53) with a cutoff value of10E-3

RNA extraction sequencing and transcriptome analysis

RNA was extracted from R opacus PD630 under the threeculture conditions NB MSM3 and MSM24 by usingTrizol Reagent (Invitrogen Carlsbad CA USA) follow-ing the standard protocol except that after isopropanoltreatment of the sample was incubated at 20C over-night Further purification and DNase treatment was con-ducted with RNAprep pure cellbacteria and RNAcleanKits (TIANGEN Beijing China) according to the manu-facturerrsquos instructions Rhodococcus opacus PD630 rRNAwas depleted using a RiboMinus Eukaryote Kit(Invitrogen Carlsbad CA USA) After RNA amplifica-tion libraries were constructed for sequencing by using aSOLiD system (Applied Biosystems Inc) according to themanufacturerrsquos specificationsRNA-Seq reads from each mRNA sample were mapped

against our assembled genome by using Bowtie with thelsquobestrsquo strata option (32) Totals of 39 922 375 62 306 706and 51 153 776 reads from the NB and MSM3 andMSM24 samples respectively were mapped with lessthan two mismatches To analyze differential expressionfragments per kilobase of transcript per million mappedreads values (FPKM) were calculated using Cufflinks (54)The fold change between conditions A and B is calculatedas log2

AB

Quantitative real-time PCR

Total RNA from cultured R opacus PD630 was isolatedusing Trizol Reagent (Invitrogen) and purified usingTIANGEN RNAclean Kit (TIANGEN) according tothe manufacturerrsquos instructions For quantitative real-time PCR (qPCR) analysis RNA was reverse transcribedusing the M-MLV Resverse Transcriptase Kit (Promega)

and further used in qPCR reactions containing SYBRgreen fluorescent dye (ABI) Relative expression ofmRNA was determined after normalization with 16Slevels using the DD-Ct method comparing MSM3MSM24 with NB respectively qPCR was performedusing an ABI StepOne PLUS PCR machine

LD purification

LD was isolated according to the method described byDing et al (55) Forty milliliters of R opacus PD630cells were centrifuged in NB and then transferred into400ml of mineral salt medium (MSM) and cultured for24 h for TAG accumulation MSM contains a high carbonsource (10 gl) but low nitrogen source (05 gl) and pri-marily used to induce a stress state in the culture mediumfor TAG accumulation Cells were collected by centrifu-gation at 5000g for 10min and washed twice with 30ml ofphosphate buffered saline (PBS) each time Afterincubating in 30ml of buffer A (25mM tricine 250mMsucrose pH 78) on ice for 20min cells were homogenizedby passing through a French Pressure Cell four times at100MPa and 4C The cell homogenate was centrifugedin a 50-ml tube at 6000 g for 10min to remove cell debrisand unbroken cells The postnuclear supernatant fraction(10ml) overlaid with 2ml of buffer B (20mM HEPES100mM KCl 2mM MgCl2 pH 74) was centrifuged at38 000 rpm for 1 h at 4C (Beckman SW40) The whiteband containing LDs at the top of the gradient was col-lected using a 200-ml pipette tip and transferred to a 15-mlEppendorf tube LDs were washed three times with 200 mlof Buffer B each time One milliliter of chloroformacet-one (11 vv) was added to each sample to dissolve lipidsand precipitate LD proteins The sample was mixed thor-oughly by vortexing and then centrifuged at 20 000 g for10min (Eppendorf centrifuge 5417R) The pellet contain-ing LD proteins was resolved with 50 ml of 2 sodiumdodecyl sulphate (SDS) sample buffer and denatured at95C for 5min The sample was stored at 20C untilrequired

Mass spectrometry (MS) analysis

The bands of interest from the NB and MSM24 sampleswere cut from SDS-polyacrylamide gel electrophoresis(SDS-PAGE) gels Samples were loaded onto a C18 trapcolumn with an auto-sampler and then eluted onto a C18column (100mm 100 mm) packed with Sunchrompacking material (SP-120-3-ODS-A 3 mm) for nano-LC-ESI-LTQ MSMS analysis The linear trapquadrupole (LTQ) mass spectrometer was operated indata-dependent mode with the initial MS scan rangingfrom 400ndash2000 Da All the MSMS data were searchedagainst our assembled and annotated genome sequenceby the SEQUEST program (Thermo USA) Bio-Workssearch parameters were set up as enzyme trypsin precur-sor ion mass tolerance 20 Da and fragment ion masstolerance 10 Da The variable modification was set tooxidation of methionine (Met+1599 Da) and the fixedmodification to carboxyamidomethylation of cysteine(Cys+5702 Da) Results were filtered with Xcorr(charge values) of Xcorr (+ 1)gt 190 Xcorr (+ 2)gt 250

1054 Nucleic Acids Research 2014 Vol 42 No 2


and Xcorr (+ 3)gt 375 where Xcorr is the cross-correl-ation score of a candidate peptide against a searchdatabase The MSMS data were then converted and de-posited at PRIDE database (56)

Construction of LPD06283 deletion mutant

A deletion mutant of structural protein LPD06283 wasconstructed by using homologous recombinationColloidal Blue staining was used to verify the absence ofthe LPD06283 protein bands Different phenotypesbetween the LPD06283 deletion mutant strain and thewild type were observed by EM The upstream and down-stream sequences of the target gene were cloned by PCRusing primers ab and cd respectively and using the wildtype R opacus PD630 genome as a template generatingfragments AB and CD Fragments AB and CD wereligated together sequenced and then cloned into apK18mobsacB plasmid Plasmid pK18mobsacB waskindly provided by Ping Xu from Shanghai Jiao TongUniversity The pK18mobsacB fusion plasmids were trans-formed into R opacus PD630 by electronic transformationPositive mutants were selected with a positive screen using akanamycin cassette and a negative screen using a sacBcassette Primers a and d were used to confirm that thefinal selected cells were positive mutants Primers f and rwere used for further PCR validation

All primer sequences were as follows

LPD06283-a CGGAATTCTGAGGAGTTCACTGATGGTGGCG

LPD06283-b CGGGATCCTGCGTGTCGACCTCGTAGGATGGG

LPD06283-c CGGGATCCCGGCTTTCTCCTGTTCAACGGTGG

LPD06283-d CGAAGCTTAAGAAGATCGAGCTGCAGGTGGGG

LPD06283-f CAGGATCCACTGACCAGAAGACCATCGACAGCGT

LPD06283-r CAGGATCCAGCCTTCTTGGCCGGAGCAGCCTT

Thin layer chromatography and western blotting

For thin layer chromatography (TLC) neutral lipids wereextracted twice from purified LD and bacterial sam-ples using chloroformacetone (11 vv) and chloroformmethanolmedium (111 vvv) respectively The organicphases were collected and air dried with nitrogen gas of ahigh purity Total lipids were dissolved in 100 ml of chloro-form for TLC analysis by using Whatman PurasilTM

60FA silica gel plates (Merck Germany) Neutral lipidswere developed using the solvent system hexanediethyletheracetic acid (80201 vvv) and phospholipidsin chloroformmethanolacetic acidH2O (751393 vvvv) TLC plates were visualized using iodine vapor

For western blotting proteins were separated by SDS-PAGE and transferred to a polyvinyl difluoride (PVDF)membrane followed by blotting with the antibodiesindicated and detection using an ECL system Weselected 20 LD proteins based on the proteome analysis

for antibody production Two rabbits were immunizedwith two synthetic peptides per protein

Transmission electron microscopy and confocalmicroscopy

Bacterial cells were examined by transmission electronmicroscopy (TEM) including positive staining andultrathin sectioning methods For positive staining cellswere loaded onto carbon-coated copper grids and subse-quently stained using 2 (wv) phosphotungstic acid for2min The grid was then washed with deionized waterthrice before viewing using a FEI Tecnai 20 (FEI CoNetherlands) electron microscope For ultrathin section-ing cells were prefixed in 25 (wv) glutaraldehyde inPBS (pH 74) overnight at 4C and postfixed in 2(wv) potassium permanganate for 5min at room tem-perature The sample was then dehydrated in ascendingconcentrations of ethanol at room temperature andembedded in Spurrrsquos resin Sections with a thickness of70 nm were cut with a Leica EM UC6 Ultramicrotome(Leica Germany) then stained with 2 (wv) uranylacetate for 15min and lead citrate for 5min at room tem-perature before visualizationFor confocal microscopy PD630 cells were washed

twice with PBS and then mounted onto coverslipspretreated with collagen prepared from rat tail Sampleswere dried for 30min before washing with 1ml of PBSand then incubated for 30min in a 1500 solution ofLipidTOX Red in darkness at room temperatureSamples were mounted onto glass slides with Mowiolmounting media and analyzed by confocal microscopy(Olympus FV1000)

Data access

Genome assemblies together with predicted gene modelsand annotation were deposited at GenBank under theproject accession number PRJNA178618 The accessionnumbers of chromosome and plasmids are CP003949(chromosome) CP003950 (plasmid-1) CP003951(plasmid-2) CP003952 (plasmid-3) CP003953 (plasmid-4) CP003954 (plasmid-5) CP003955 (plasmid-6)CP003956 (plasmid-7) CP003957 (plasmid-8) CP003958(plasmid-9) The expression data sets used in this study areavailable at the NCBI Gene Expression Omnibus (GEO)(httpwwwncbinlmnihgovgeo) under accessionnumber GSE42381

RESULTS

Rhodococcus opacus PD630 genome exhibits a superability of biosynthesis and catabolism

We used a combined dispersed strategy incorporating datagenerated using Roche454 and Illumina sequencingtechnologies and assembled the genome and associatedplasmids of R opacus PD630 The complete genomeconsists of a circular chromosome of 8 376 954 bp inlength and nine plasmids which in combined total are9 169 032 bp (Supplementary Table S1) The genomeencodes 8947 protein-coding genes 51 tRNA and 12



rRNAs genes (Supplementary Tables S2 and S3) Themean size of predicted ORFs is 928 bp with 8028 ofthe ORF between 201 and 1500 bp in length (Figure 1A)Further a few insertion sequences were predicted usingthe ISfinder (44) (Supplementary Table S4) but onlyoccupy 038 of the genome A large number of genes(ie 2563) are duplicated (Figure 1B) Rhodococcus opacusPD630 has a high genomic G+C (guaninendashcytosinecontent) content (6747) with 176 genes having eithera much higher or lower G+C contents than the genomeon average (Figure 1C) Protein functions were manuallyassigned based on Interpro (37) UniProt searches (36)and COG functional classification system (40) Amongthe 22 COG subclasses lipid transport and metabolismwere the largest category occupying 850 genesThe genomes of two other species in the Rhodococcus

genus namely R opacus B4 (R opacus B4) and R jostiiRHA1 (R jostii RHA1) (52) have also been sequencedAlthough they are closely related to R opacus PD630(Figure 1D) comparative analyses reveals that R opacusPD630 may have superior lipid metabolism For exampleR opacus PD630 has 850 COG categories E genes (aminoacid transport and metabolism) 556 of the P genes (inor-ganic ion transport and metabolism) and 580 of the Ggenes (carbohydrate transport and metabolism) which issignificantly more than R jostii RHA1 which has 595 329

and 434 genes respectively and R opacus B4 which has575 328 and 412 genes respectively in the above men-tioned categories (all Plt 1E-04 two-tailed Fisher exacttest Supplementary Table S5) Enzymes were annotatedto metabolic reactions using EFICAz2 (38) and theKEGG database (39) In summation 3200 enzymes arepredicted for 2727 metabolic reactions that are markedlymore than predicted reactions of R jostii RHA1 Ropacus B4 (39) and earlier reported 2017 metabolic reac-tions of PD630 partial genome (29)

As R opacus PD630 is able to grow on a variety ofsubstrates and environments (262957) we searched forHTGs that might confer selective evolutionary advantagesin different environments A total of 532 genes(Supplementary Table S2) were predicted as HTG byusing the WN method (48) Of the 612 genes implicatedin energy production and conversion and the 711 genesinvolved in lipid transport and metabolism 73 and 77respectively were predicted to be HTG providing signifi-cant evolutionary contributions to lipid synthesis and lipidmetabolism in R opacus PD630 (P-values 226E-07 and305E-06 two-tailed Fisher exact test)

Global gene expressions under lipid accumulation

To confirm that R opacus PD630 cells include LDs andalso display their dynamics in different cultures cells were

Figure 1 Basic statistic information of R opacus PD630 genome (A) Length distribution of predicted R opacus PD630 ORFs The average lengthof all genes is 928 bp with most genes ranging from 200 to 1600 bp The number of genes in each category is noted above each bar (B) Distributionof the copy numbers of homologous genes among all 8947 genes Homologous genes were detected using BLASTP with an e-value cutoff of 10E-20and identity of 50 The number of genes in each category is noted above each bar (C) The average C+G content of all genes was 6747 (solidred line) One percent of the genes fell outside the upper and lower green-dotted lines and had markedly higher or lower G+C contents (D) Venndiagram comparing the whole genomes of R opacus PD630 R jostii RHA1 and R opacus B4 All proteins in R opacus PD630 were compared withthose in R jostii RHA1 and R opacus B4 using BLASTP with an e-value cutoff of 10E-20 and an identity of 50



first cultured in NB and subsequently transferred to aMSM for 3 h (MSM3) and 24 h (MSM24) respectivelySince MSM is a low nitrogencarbon ratio medium thatis used as a stressful culture for TAG accumulation(272858) the systematically comparative analysis of dif-ferential gene expressions under NB and MSM cultureswill be helpful to reveal how proteins and biologicalpathways respond to TAG accumulation and LDdynamics Firstly LipidTOX staining showed that theR opacus PD630 cells grown in MSM cultures containmuch larger LDs than those grown in NB culture(Supplementary Figure S1A) This observation is in agree-ment with results from the TLC that TAG is accumulatedto a greater amount in MSM24 than NB but DAG ismuch accumulated in lesser quantities (SupplementaryFigure S1B) Further we imaged cells by TEM to bettervisualize phenotypic differences in LDs Interestingly LDsderived from the MSM24 treatment were much larger indiameter than those cultured in NB (Figure 2A)Moreover R opacus PD630 contains many electron-transparent structures that occupy most of the innerarea of the cell with the TAG content increasing fromthe NB to MSM24 culture (Supplementary Figure S2)These measurements confirm that R opacus PD630

contains LDs and accumulate large amounts of TAGunder MSM culture conditionsTo investigate the proteins that are related to dynamics

of R opacus PD630 under different culture conditions wesequenced and compared whole-genome transcriptomes ofthese three cultures NB MSM3 and MSM24 The qualityof our transcriptomes was confirmed by measuring expres-sions of 13 randomly selected genes by using qPCRAmong the 13 genes 9 genes (LPD05955 LPD07778LPD02638 LPD05411 LPD04190 LPD04189LPD05410 LPD02774 LPD06334) presented similartrends in NB to MSM3 and MSM24 and only fourgenes (LPD02936 LPD05356 LPD02250 LPD07707)had little ratio differences (Supplementary Figure S3 andSupplementary Table S6 for detailed expression values)The qPCR result confirmed the transcriptomic data isreliable We then performed a systematic analysis of thegenome-wide expression dynamics under three culturesNB MSM3 and MSM24 Most genes were either ex-pressed under all three conditions (6759 genes) or underat least one condition (7770 genes) but there were 1177genes that were not expressed under any of these condi-tions When cells were changed from NB to MSM condi-tion a drastic response to environmental change was

Figure 2 Whole-genome differential expression analysis (A) EM images (ultra-thin sections) of R opacus PD630-WT cultured in MSM for 3 h(MSM3) or 24 h (MSM24) or grown in NB for 48 h (NB) Bar=05 mm (B) Relative abundance of different categories of differentially expressedgenes in R opacus PD630 are shown under the three culture conditions Pie charts on the left represent upregulated genes while those on the rightrepresent downregulated genes The total number of genes accounted is given below each pie chart Colors correspond to categories in the COGdatabase



observed even after 24 h We observed a marked response3 h after cells were transferred from NB to MSMwith 5699 of the genes upregulated and 3032downregulated and 2115 being upregulated gt2-foldand 936 being downregulated gt2-fold (SupplementaryFigure S4) The functional enrichment analysis was per-formed for differentially expressed genes (2-fold changeFigure 2B Supplementary Table S7) Genes in category J(translation ribosomal structure and biogenesis) were sig-nificantly upregulated (P-value 505E-20 two-tailedFisher exact test) while genes in category K (transcrip-tion) G (carbohydrate transport and metabolism) and C(energy production and conversion) were downregulated(P-values of 60E-07 808E-06 and 456E-05 respectivelytwo-tailed Fisher exact test) The increased expression ofribosomal proteins is consistent with the number of genes(5699) upregulated in the MSM3 treatment The largeupregulated protein proportions and their functionalenriched groups exhibited a dynamical landscape ofproteins and pathways responsible for lipid synthesisand storage in PD630

Enzymes involved in TAG biosynthesis and metabolism

The accumulation of TAG in R opacus PD630 is adynamic balance between lipid synthesis and degradationHere we systematically analyzed the potential proteinsinvolved in the TAG biosynthesis and metabolicpathway The pathway consists of 22 reactions involving457 candidate enzymes (Supplementary Table S8) Thesereactions are further classified into five stages initiallyfatty acid biosynthesis TAG biosynthesis TAG storageTAG degradation and fatty acid degradation (Figure 3Aand C) In these enzyme families the largest one is 1233-oxoacyl-(acyl-carrier-protein) reductase (EC111100)exhibiting a marked preference for acyl-carrier-protein de-rivatives over CoA derivatives as substrates Hundredacyl-CoA dehydrogenase (EC1336EC13993EC139913) performed different specificities for longmedium and short chains Other large gene familiesinclude 16 diacyglycerol O-acyltransferase (DGATsEC23120) 45 lipaseesterase (EC3113) 40 long-chain fatty acid CoA ligase (EC6123) and 53 enoyl-CoA hydratase (EC42117) Among these reactionsmany enzymes were highly expressed or differentiallyexpressed by gt2-fold change (Figure 3B) indicating thata broad spectrum of fatty acids were differentiallysynthesized and catabolized in R opacus PD630 whencultures changedEarlier studies revealed that R opacus PD630 can

synthesize many neutral lipids but the precise enzymesinvolved for different lipids have rarely been reported(1328ndash30) Thus identifying key lipid enzymes and sup-pressing glycogen synthesis appears important to ensuremaximal TAG yields In our study 16 DGATs were pre-dicted as possessing lipid synthases activity and thesecould be further divided into several subclusters usingphylogenetic analysis (Figure 3D) The expression ofgenes in each subcluster was similar under three condi-tions indicating that those genes of subclusters may beinvolved in different lipid synthesis In particular the

gene LPD02996 was significantly upregulated gt2-foldchange in MSM3 compared with NB culture Threegenes LPD01972 LPD05741 and LPD00644 were alsoslightly upregulated with fold change of 139 041 and134 respectively Since TAG accumulates to high levelsin MSM cultures the increased expression of these fourgenes suggests that they may be involved in TAGsynthesis LPD05741 also named artf2 had beenverified to be involved in TAG biosynthesis and accumu-lation in PD630 (59) The elevation of LPD05741 is con-sistent with this study Furthermore the significantupregulation of LPD02996 also suggests it may beanother potential protein related to TAG biosynthesisSix genes LPD03605 LPD04096 LPD04039 LPD04040LPD02252 and LPD04049 were significantlydownregulated gt2-fold change indicating that these sixgenes may be involved in other lipid synthesis or TAGdegradation but blocked in MSM culture to potentiategreater TAG yield A total of 45 genes coding forputative lipaseesterase proteins (17 lipases 22 esterasesand 6 others) were predicted to be involved in neutrallipid degradation Of the 17 lipases the expression ofLPD00504 and LPD03906 were significantly elevated inMSM3 compared with NB (2 fold change) whileLPD03039 LPD00420 and LPD05381 were slightlyincreased with fold change of 02 044 and 184 respect-ively A gene LPD04146 was decreased with a fold changeof 168 in MSM3 compared with NB (Figure 3E) Of the22 esterases LPD00874 and LPD03041 were dramaticallyupregulated (2-fold change) while LPD05391LPD02749 LPD00891 LPD01667 LPD07390 andLPD05053 were slightly upregulated with fold change of158 124 146 186 155 and 031 respectively GeneLPD00479 was significantly downregulated in MSM3compared with NB culture (2-fold change Figure 3F)These proteins can be classified into some subclusters withsimilar expression patterns in each cluster by phylogeneticanalysis For example LPD00874 and LPD02749 areboth increased in MSM3 and then decreased inMSM24 (Figure 3F-F1) A similar tendency is alsoobserved among LPD00891 LPD03041 and LPD01667(Figure 3F-F2) These results presented a precise predic-tion that those differentially expressed lipasesesterasesmay be involved in TAG synthesis and degradation

Identifying proteins associated to prokaryotic LD

To better understand prokaryotic LD proteins weisolated LDs (please see lsquoMaterials and Methodsrsquosection) (Supplementary Figure S5A) and performedproteomic and lipid analyses Initially we determinedthe quality of isolated LDs TEM imaging of isolatedLDs using positive staining displayed few contaminantspresent from other membranes (Supplementary FigureS5B) The size distribution of LDs is bell shaped andbetween 131 and 3168 nm as determined using a DelsaNano C particle analyzer (Supplementary Figure S5C)Total lipids from isolated LDs and from the total cellmembrane were separated by TLC (SupplementaryFigure S5D) Results indicated that the main lipidpresent was TAG (band 2 in Supplementary Figure



S5E) and that the total membrane fraction wasenriched in phosphatidylethanolamine (band 5 inSupplementary Figure S5F) and also an unknown lipid(band 4 in Supplementary Figure S5F) The protein com-position of LDs was distinctly different from that of thetotal membrane cytosol and whole-cell lysates(Figure 4A) further verifying the high quality of theisolated LDs

We then conducted comparative proteomic studies onLDs isolated under different culture conditions andfocused our attention to identify proteins that arerelated to LD functions and dynamics LD proteinsfrom bacteria cultured in MSM24 were initially separated

by SDS-PAGE The gel was cut into 42 slices with 430proteins identified using proteomic analysis (Figure 4B)Proteins with a high abundance according to the peptidenumber in the MS data were chosen and respectiveantibodies produced Further Western blotting wasperformed to verify the association of these proteinswith LDs and determine their cellular distribution(Figure 4C) The proteins LPD02850 LPD04067LPD05350 LPD03377 LPD08045 and LPD02496 weremainly present in the LD fraction whereas proteinsLPD01403 and LPD02840 were present in both the LDand cytosol fractions Proteins LPD02062 and LPD02043were ubiquitous distributed within the whole cell

Figure 3 Expression and phylogenetic analysis of gene families involved in TAG biosynthesis and degradation (A) TAG biosynthesis storage anddegradation pathways are divided into five biochemical stages and 22 reactions (B) Heatmap of highly expressed (10 FPKM) or dramaticallydifferentially expressed enzymes (2-fold change) in each reaction (C) EC numbers of enzymes involved in (A) (D) Phylogenetic tree and heatmapof 16 predicted TAG synthases (E) Phylogenetic tree and heatmap of 17 predicted TAG lipases (F) Phylogenetic tree and heatmap of 22 predictedesterases Genes whose expression increased from NB to MSM3 (2-fold change) are marked by red triangles and those which decreased (2-foldchange) are marked by green triangles in D E and F Two examples of gene clusters that have similar sequences and expression patterns are noted asF1 and F2



Figure 4 Functional and expressional analysis of LD proteins (A) Rhodococcus opacus PD630 proteins from whole-cell lysates (W) cytosol (C)membranes (M) and purified LDs (L) were separated by 10 SDS-PAGE followed by silver staining (B) LC-MS band analysis of LD-associatedproteins Arrows indicate the positions at which the gel was sliced Forty-two bands (1ndash42) from the MSM24 sample and eight bands (43ndash50) fromthe NB sample corresponding to the major bands in the MSM24 sample were analyzed (C) Western blotting of LD-associated proteins The sameamount of protein from LDs membranes the cytosol and whole-cell lysates were separated by 10 SDS-PAGE and blotted with the antibodiesindicated Western blotting using anti-LPD02850 anti-LPD04067 anti-LPD05350 anti-LPD03377 anti-LPD08045 anti-LPD01403 anti-LPD02496anti-LPD02840 anti-LPD02062 and anti-LPD02043 (D) 430 LD-associated proteins were identified in MSM24 and categorized into nine groupsbased on searches of our R opacus PD630 genome and the Pfam and NCBI databases (E) Histogram showing the percentage of LD-associatedgenes with dramatic expression changes in the three conditions (F) Heatmap showing the expression of all 430 genes ATP synthase (up) RNApolymerases translation initiation factors and elongation factors (middle) and ribosome proteins (down) are enlarged to the right of the heatmap



The 430 proteins identified in LDs were categorized intonine groups including 231 enzymes 6 transport proteins25 transcription and translation proteins 31 ribosomeproteins 5 chaperone protein 8 stress-induced proteins5 cell divisionndashrelated proteins and 110 other unknownfunctional proteins (Figure 4D and Supplementary TableS9) The 231 enzymes were mainly involved in lipid syn-thesis and degradation including 40 transferases 8 ligases37 dehydrogenases 32 reductases and 29 synthases whichsummed up gt63 of these enzymes To further confirmthese results two independent proteomic analyses ofPD630 LD proteins under MSM24 condition were per-formed Two hundred thirty-eight of these 430 proteinswere also identified at least once (SupplementaryTable S10) The major LD protein bands that showedmarked differences between the NB and MSM24samples (Figure 4B bands 43 to 50 SupplementaryTable S11 for corresponding bands in two samples) werethen subjected to proteomic analysis Most abundantproteins in these bands were similar in these twosamples but increased markedly in the MSM24 samplescompared with NB Interestingly we found that 31 ribo-somal proteins were all increased by gt2-fold change(Supplementary Table S9) Furthermore six ATPenzymes (Figure 4F LPD02785 LPD02782 LPD02805LPD03215 LPD02780 and LPD02777) two RNA poly-merases two translation initiation factors two elongationfactors and 15 transcriptional regulators were also foundto be differentially expressed There were 13 of the 15transcriptional regulators upregulated in MSM3compared with NB Since these expressions of thesefactors were highly correlated they may construct as acomplete transcriptiontranslation system in LD surfaceFor summary the abundant proteins detected in PD630LD revealed that prokaryotic LD are involved in not onlythe lipid synthesis and catabolism but also multiple otherimportant cellular functions

Proteins involved in LD dynamics

Integrating transcriptomic and proteomic data is a usefulmethod to reveal the proteins involved in LD dynamicsTranscriptome analysis specifically for LD proteomeshowed that gt50 of 430 proteins were dramatically dif-ferentially expressed under the varying NB MSM3 andMSM24 conditions Totally 177 of the 430 genes weresignificantly upregulated while 30 genes weredownregulated (2-fold change or 2-fold change) fromNB to MSM3 (Figure 4E) suggesting that these proteinsplay important functions related to LDs when culture con-ditions changed Specifically 99 proteins of the 231enzymes were increased suggesting that the lipid synthesisprocesses are much accelerated We also identified fourdynamin (LPD02043 LPD02044 LPD02062 andLPD02063) and three SNARE-like proteins (LPD02118LPD02119 and LPD03976) that may be involved in LDdynamics (Supplementary Table S12) Each of thedynamin proteins contains a Dynamin_N domain(PF00350) (also named DLP_2 in NCBI CDD database)(Supplementary Figure S6A) Further operon LPD02062and LPD02063 were highly expressed but operon

LPD02043 and LPD02044 showed almost noexpressions The three SNARE-like proteins eachcontain a SNARE_assoc domain (PF09335) (also namedPRK10847 in NCBI CDD database) (SupplementaryFigure S6B) Moreover LPD03976 LPD02118 andLPD02119 were expressed under all three conditions andwere observed to have a 15-fold change from NB toMSM3 culture Since dynamin and SNARE-like proteinsplay important roles in LD budding and fusion processesin eukaryotes (6061) their presence in R opacus PD630suggests that these proteins may also mediate the LDdynamics in prokaryotic organisms thus indicating theirevolutionary origins So far these integrated omics studiessystematically revealed multiple factors that are related toprokaryotic LD dynamics especially the lipid synthesisstorage and LD morphology

A structure-like protein affects the morphology ofprokaryotic LDs

Among all the proposed proteins that may be involved inLD dynamics we further determined the proteinLPD06283 as a structure-like protein We observed thatit was involved in LD dynamics The expression ofLPD06283 was markedly increased from NB (FPKM377) to MSM3 (FPKM 1815) achieving a 227-foldchange It was revealed as the major protein band(Figure 5A) A fragment of LPD06283 (from 26 to127 bp) was predicted to be an apolipoprotein domain(PF01442 Figure 5B) by using Pfam databases (46) Theaverage posterior probability that means the degree ofconfidence in each individual aligned residue is arrangedfrom 65 to 75 showing a high confidence of this predic-tion We deleted it by homologous recombination to in-vestigate its function (Supplementary Figure S7A and B)Compared with the wild type LDs in the LPD06283deletion mutant were dramatically larger however theirnumbers were reduced (Figure 5C Supplementary FigureS8 S9) LPD06283 is sequence similar with the earlierreported proteins Ro02104 in RHA1 (62) and TadA inPD630 (63) These results are consistent with earlystudies demonstrating that LDs are easily fused in theabsence of structural proteins and also consistent withearly studies in eukaryotic cells (6264-66)

DISCUSSION

LD is an important organelle in both prokaryote andeukaryote organisms but little is known regarding itsfunction dynamics and evolution R opacus PD630 is auseful model for the investigation of LD since the LD isthe only intracellular membranous organelle We havecomprehensively presented the R opacus PD630genome transcriptome and LD proteome and have sys-tematically investigated the key proteins potentiallyinvolved in LD dynamics and functions Integratingthese oimcs data revealed significantly differential expres-sion of 177 LD proteins and further depicted a dynamicallandscape of prokaryotic LD in TAG accumulatingcultures These results not only confirmed several earlystudies of LD proteins but also indicated many novel



candidate proteins for optimizing TAG synthesis andstorageComparative proteomics detected an abundance of

proteins on the PD630 LD surface including enzymestransport protein cell division related proteins transcrip-tiontranslation protein ribosome proteins and manyother unknown proteins These diverse proteins indicatethat prokaryotic LDs as the first intracellular membranesystem are involved in lipid metabolism and other import-ant cellular functions Furthermore this diversity of LDproteins was also observed in earlier reported proteomicsfrom multiple species such as green algae (67) yeast (68)Drosophila (69) Caenorhabditis elegans (70) mice (71)indicating a functional evolutionary path fromprokaryotes to mammals For example many ribosomalproteins were found in the LDs of PD630 yeastC elegans and Drosophila embryos but absent in theLDs of mice The number of enzymes presented in LDsis decreased from PD630 to mice The wide range offunctions carried out by prokaryotic LDs and the special-ization of mammalian LDs suggest that functional special-ization accompanied by the distribution of otherbiochemical processes among other organelles and it isa trend that continues during LD evolution from prokary-otes to eukaryotesIn both prokaryotic and eukaryotic cells LDs have the

same topological structure that lipids are encapsulated byphospholipid monolayer From a topological view LDshould be evolutionarily constructed by cells themselvesand rather different from organelles such as themitochondria and chloroplast that may be original viafusion of prokaryotic cells (7273) An early model ofLD ontogeny in strain PD630 had been proposed thatLDs originate in the internal side of the cytoplasmicmembrane where DGAT enzymes are localized (58)

However this model presents difficulty to explain howTAGs are synthesized and topologically encapsulatedinto LDs in the cytoplasm as TAGs are not soluble inwater Our discovery of dynamin-like proteins in LDs maysuggested that the initialized lipid synthesis may beaccelerated at the cell membrane bilayer and disengagedfrom cell membrane via catalysis of dynamin-like proteinshowever further experiments need to be performed to val-idate this hypothesis We believe the complete R opacusPD630 genome transcriptome and LD proteome pre-sented here provides a starting point not only forunravelling mechanisms of LD dynamics but also forinvestigating the organelle and eukaryotic evolution

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online

ACKNOWLEDGEMENTS

The authors thank Dr Joy Fleming for critical reading ofthis manuscript and useful suggestions The authors alsothank Shufeng Sun for assistance with electron micros-copy and Xudong Zhao and Su Liu for techniquesupport PL JY XZ MQZ YL AS TFdesigned the projects YD LY YC JY DYGL XW YZ CZ performed the experimentsYC YD LY JY GL LS LH CH YWYD HZ PZ HN SX ZX HZ TH GWZF YZ FY HL and XW analyzed the data SZperformed electron microscopy PL JY YC YDLY SC wrote the article All authors read andapproved the final manuscript

Figure 5 Deletion of LPD06283 results in supersized LDs (A) Gel electrophoresis of LD proteins from R opacus PD630-WT and the LPD06283deletion mutant stained by colloidal blue Band 3 the main band disappeared in the LPD06283 deletion mutant (B) Location of the predicteddomain in LPD06283 Apolipoprotein (PF01442) (C) a1-a2 EM images of R opacus PD630-WT cultured in MSM for 24 h after growing in NB for48 h using positive staining methods b1-b2 EM images of the LPD06283 deletion mutant under the same conditions as R opacus PD630-WT Thelower panels give amplified pictures Bar=2 mm



FUNDING

Ministry of Science and Technology of China[2009CB919000 2011CBA00906 2012CB3165002012DFG32160] National Natural Science Foundationof China [30971431 31000365 61273228 3110006891019016] Funding for open access charge Ministry ofScience and Technology of China

Conflict of interest statement None declared

REFERENCES

1 MurphyDJ (2001) The biogenesis and functions of lipid bodiesin animals plants and microorganisms Progr Lipid Res 40325ndash438

2 FareseRV Jr and WaltherTC (2009) Lipid droplets finally geta little R-E-S-P-E-C-T Cell 139 855ndash860

3 YangL DingY ChenY ZhangS HuoC WangY YuJZhangP NaH ZhangH et al (2012) The proteomics of lipiddroplets structure dynamics and functions of the organelleconserved from bacteria to humans J Lipid Res 53 1245ndash1253

4 FujimotoT OhsakiY ChengJ SuzukiM and ShinoharaY(2008) Lipid droplets a classic organelle with new outfitsHistochem Cell Biol 130 263ndash279

5 MeexRC SchrauwenP and HesselinkMK (2009) Modulationof myocellular fat stores lipid droplet dynamics in health anddisease Am J Physiol Regul Integr Comp Physiol 297R913ndashR924

6 Le LayS and DugailI (2009) Connecting lipid droplet biologyand the metabolic syndrome Prog Lipid Res 48 191ndash195

7 MaedaK CaoH KonoK GorgunCZ FuruhashiMUysalKT CaoQ AtsumiG MaloneH KrishnanB et al(2005) Adipocytemacrophage fatty acid binding proteins controlintegrated metabolic responses in obesity and diabetes CellMetab 1 107ndash119

8 EckelRH AlbertiKG GrundySM and ZimmetPZ (2010)The metabolic syndrome Lancet 375 181ndash183

9 SamuelVT and ShulmanGI (2012) Mechanisms for insulinresistance common threads and missing links Cell 148 852ndash871

10 BeerLL BoydES PetersJW and PosewitzMC (2009)Engineering algae for biohydrogen and biofuel production CurrOpin Biotechnol 20 264ndash271

11 MoelleringER and BenningC (2010) RNA interferencesilencing of a major lipid droplet protein affects lipid droplet sizein Chlamydomonas reinhardtii Eukaryotic Cell 9 97ndash106

12 ScottSA DaveyMP DennisJS HorstI HoweCJLea-SmithDJ and SmithAG (2010) Biodiesel from algaechallenges and prospects Curr Opin Biotechnol 21 277ndash286

13 GreenbergAS EganJJ WekSA GartyNB Blanchette-MackieEJ and LondosC (1991) Perilipin a major hormonallyregulated adipocyte-specific phosphoprotein associated with theperiphery of lipid storage droplets J Biol Chem 26611341ndash11346

14 KalscheuerR WaltermannM AlvarezM and SteinbuchelA(2001) Preparative isolation of lipid inclusions from Rhodococcusopacus and Rhodococcus ruber and identification of granule-associated proteins Arch Microbiol 177 20ndash28

15 ZehmerJK HuangY PengG PuJ AndersonRG andLiuP (2009) A role for lipid droplets in inter-membrane lipidtraffic Proteomics 9 914ndash921

16 LondosC BrasaemleDL SchultzCJ SegrestJP andKimmelAR (1999) Perilipins ADRP and other proteins thatassociate with intracellular neutral lipid droplets in animal cellsSemin Cell Dev Biol 10 51ndash58

17 PaarM JungstC SteinerNA MagnesC SinnerF KolbDLassA ZimmermannR ZumbuschA KohlweinSD et al(2012) Remodeling of lipid droplets during lipolysis and growthin adipocytes J Biol Chem 287 11164ndash11173

18 WelteMA (2007) Proteins under new management lipid dropletsdeliver Trends cell Biol 17 363ndash369

19 PloeghHL (2007) A lipid-based model for the creation of anescape hatch from the endoplasmic reticulum Nature 448435ndash438

20 PuJ HaCW ZhangS JungJP HuhWK and LiuP (2011)Interactomic study on interaction between lipid droplets andmitochondria Protein cell 2 487ndash496

21 OzekiS ChengJ Tauchi-SatoK HatanoN TaniguchiH andFujimotoT (2005) Rab18 localizes to lipid droplets and inducestheir close apposition to the endoplasmic reticulum-derivedmembrane J Cell Sci 118 2601ndash2611

22 MartinS DriessenK NixonSJ ZerialM and PartonRG(2005) Regulated localization of Rab18 to lipid droplets effectsof lipolytic stimulation and inhibition of lipid droplet catabolismJ Biol Chem 280 42325ndash42335

23 LiuP BartzR ZehmerJK YingYS ZhuM SerreroG andAndersonRG (2007) Rab-regulated interaction of earlyendosomes with lipid droplets Biochim Biophys Acta 1773784ndash793

24 GoodmanJM (2008) The gregarious lipid droplet J BiolChem 283 28005ndash28009

25 WaltermannM and SteinbuchelA (2005) Neutral lipid bodies inprokaryotes recent insights into structure formation andrelationship to eukaryotic lipid depots J Bacteriol 1873607ndash3619

26 AlvarezHM and SteinbuchelA (2002) Triacylglycerols inprokaryotic microorganisms Appl Microbiol Biotechnol 60367ndash376

27 WaltermannM LuftmannH BaumeisterD KalscheuerR andSteinbuchelA (2000) Rhodococcus opacus strain PD630 as a newsource of high-value single-cell oil Isolation and characterizationof triacylglycerols and other storage lipids Microbiology146(Pt 5) 1143ndash1149

28 AlvarezAF AlvarezHM KalscheuerR WaltermannM andSteinbuchelA (2008) Cloning and characterization of a geneinvolved in triacylglycerol biosynthesis and identification ofadditional homologous genes in the oleaginous bacteriumRhodococcus opacus PD630 Microbiology 154 2327ndash2335

29 HolderJW UlrichJC DeBonoAC GodfreyPADesjardinsCA ZuckerJ ZengQ LeachAL GhivirigaIDancelC et al (2011) Comparative and functional genomics ofRhodococcus opacus PD630 for biofuels development PLoSGenetics 7 e1002219

30 AlvarezHM MayerF FabritiusD and SteinbuchelA (1996)Formation of intracytoplasmic lipid inclusions by Rhodococcusopacus strain PD630 Arch Microbiol 165 377ndash386

31 BoetzerM HenkelCV JansenHJ ButlerD and PirovanoW(2011) Scaffolding pre-assembled contigs using SSPACEBioinformatics 27 578ndash579

32 LangmeadB TrapnellC PopM and SalzbergSL (2009)Ultrafast and memory-efficient alignment of short DNAsequences to the human genome Genome Biol 10 R25

33 DelcherAL BratkeKA PowersEC and SalzbergSL (2007)Identifying bacterial genes and endosymbiont DNA withGlimmer Bioinformatics 23 673ndash679

34 BesemerJ and BorodovskyM (2005) GeneMark web softwarefor gene finding in prokaryotes eukaryotes and viruses NucleicAcids Res 33 W451ndashW454

35 SuzekBE ErmolaevaMD SchreiberM and SalzbergSL(2001) A probabilistic method for identifying start codons inbacterial genomes Bioinformatics 17 1123ndash1130

36 MagraneM and ConsortiumU (2011) UniProt Knowledgebasea hub of integrated protein data Database 2011 bar009

37 HunterS JonesP MitchellA ApweilerR AttwoodTKBatemanA BernardT BinnsD BorkP BurgeS et al (2012)InterPro in 2011 new developments in the family and domainprediction database Nucleic Acids Res 40 D306ndash312

38 ArakakiAK HuangY and SkolnickJ (2009) EFICAz2enzyme function inference by a combined approach enhanced bymachine learning BMC Bioinformatics 10 107

39 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114

40 TatusovRL FedorovaND JacksonJD JacobsARKiryutinB KooninEV KrylovDM MazumderR



MekhedovSL NikolskayaAN et al (2003) The COGdatabase an updated version includes eukaryotes BMCBioinformatics 4 41

41 SchattnerP BrooksAN and LoweTM (2005) The tRNAscan-SE snoscan and snoGPS web servers for the detection of tRNAsand snoRNAs Nucleic Acids Res 33 W686ndashW689

42 TaquistH CuiY and ArdellDH (2007) TFAM 10 an onlinetRNA function classifier Nucleic Acids Res 35 W350ndashW353

43 LagesenK HallinP RodlandEA StaerfeldtHH RognesTand UsseryDW (2007) RNAmmer consistent and rapidannotation of ribosomal RNA genes Nucleic Acids Res 353100ndash3108

44 SiguierP PerochonJ LestradeL MahillonJ and ChandlerM(2006) ISfinder the reference centre for bacterial insertionsequences Nucleic Acids Res 34 D32ndashD36

45 JurkaJ KapitonovVV PavlicekA KlonowskiP KohanyOand WalichiewiczJ (2005) Repbase Update a database ofeukaryotic repetitive elements Cytogenet Genome Res 110462ndash467

46 PuntaM CoggillPC EberhardtRY MistryJ TateJBoursnellC PangN ForslundK CericG ClementsJ et al(2012) The Pfam protein families database Nucleic Acids Res40 D290ndashD301

47 Marchler-BauerA LuS AndersonJB ChitsazFDerbyshireMK DeWeese-ScottC FongJH GeerLYGeerRC GonzalesNR et al (2011) CDD a ConservedDomain Database for the functional annotation of proteinsNucleic Acids Res 39 D225ndashD229

48 TsirigosA and RigoutsosI (2005) A new computational methodfor the detection of horizontal gene transfer events Nucleic AcidsRes 33 922ndash933

49 ThompsonJD GibsonTJ and HigginsDG (2002) Multiplesequence alignment using ClustalW and ClustalX Chapter 2Unit 23

50 TamuraK DudleyJ NeiM and KumarS (2007) MEGA4Molecular Evolutionary Genetics Analysis (MEGA) softwareversion 40 Mol Biol Evol 24 1596ndash1599

51 DamP OlmanV HarrisK SuZ and XuY (2007) Operonprediction using both genome-specific and general genomicinformation Nucleic Acids Res 35 288ndash298

52 McLeodMP WarrenRL HsiaoWW ArakiN MyhreMFernandesC MiyazawaD WongW LillquistAL WangDet al (2006) The complete genome of Rhodococcus sp RHA1provides insights into a catabolic powerhouse Proc Natl AcadSci USA 103 15582ndash15587

53 YeJ McGinnisS and MaddenTL (2006) BLASTimprovements for better sequence analysis Nucleic Acids Res 34W6ndashW9

54 TrapnellC WilliamsBA PerteaG MortazaviA KwanGvan BarenMJ SalzbergSL WoldBJ and PachterL (2010)Transcript assembly and quantification by RNA-Seq revealsunannotated transcripts and isoform switching during celldifferentiation Nat Biotechnol 28 511ndash515

55 DingY ZhangS YangL NaH ZhangP ZhangHWangY ChenY YuJ HuoC et al (2013) Isolating lipiddroplets from multiple species Nat Protoc 8 43ndash51

56 VizcainoJA CoteR ReisingerF BarsnesH FosterJMRamesederJ HermjakobH and MartensL (2010) TheProteomics Identifications database 2010 update Nucleic AcidsRes 38 D736ndashD742

57 AlvarezHM SilvaRA CesariAC ZamitALPeressuttiSR ReicheltR KellerU MalkusU RaschCMaskowT et al (2004) Physiological and morphologicalresponses of the soil bacterium Rhodococcus opacus strain PD630to water stress FEMS Microbiol Ecol 50 75ndash86

58 WaltermannM HinzA RobenekH TroyerD ReicheltRMalkusU GallaHJ KalscheuerR StovekenT vonLandenbergP et al (2005) Mechanism of lipid-body formation inprokaryotes how bacteria fatten up Mol Microbiol 55750ndash763

59 HernandezMA ArabolazaA RodriguezE GramajoH andAlvarezHM (2013) The atf2 gene is involved in triacylglycerolbiosynthesis and accumulation in the oleaginous Rhodococcusopacus PD630 Appl Microbiol Biotechnol 97 2119ndash2130

60 BostromP AnderssonL RutbergM PermanJ LidbergUJohanssonBR Fernandez-RodriguezJ EricsonJ NilssonTBorenJ et al (2007) SNARE proteins mediate fusion betweencytosolic lipid droplets and are implicated in insulin sensitivityNat Cell Biol 9 1286ndash1293

61 LowHH SachseC AmosLA and LoweJ (2009) Structure ofa bacterial dynamin-like protein lipid tube provides a mechanismfor assembly and membrane curving Cell 139 1342ndash1352

62 DingY YangL ZhangS WangY DuY PuJ PengGChenY ZhangH YuJ et al (2012) Identification of the majorfunctional proteins of prokaryotic lipid droplets J Lipid Res 53399ndash411

63 MacEachranDP PropheteME and SinskeyAJ (2010) TheRhodococcus opacus PD630 heparin-binding hemagglutininhomolog TadA mediates lipid body formation Appl EnvironMicrobiol 76 7217ndash7225

64 MiuraS GanJW BrzostowskiJ ParisiMJ SchultzCJLondosC OliverB and KimmelAR (2002) Functionalconservation for lipid storage droplet association among PerilipinADRP and TIP47 (PAT)-related proteins in mammalsDrosophila and Dictyostelium J Biol Chem 277 32253ndash32257

65 BickelPE TanseyJT and WelteMA (2009) PAT proteins anancient family of lipid droplet proteins that regulate cellular lipidstores Biochimica Biophys Acta 1791 419ndash440

66 StraubBK HerpelE SingerS ZimbelmannR BreuhahnKMacher-GoeppingerS WarthA Lehmann-KochJLongerichT HeidH et al (2010) Lipid droplet-associated PAT-proteins show frequent and differential expression in neoplasticsteatogenesis Modern Pathol 23 480ndash492

67 NguyenHM BaudetM CuineS AdrianoJM BartheDBillonE BruleyC BeissonF PeltierG FerroM et al (2011)Proteomic profiling of oil bodies isolated from the unicellulargreen microalga Chlamydomonas reinhardtii with focus onproteins involved in lipid metabolism Proteomics 11 4266ndash4273

68 AthenstaedtK ZweytickD JandrositzA KohlweinSD andDaumG (1999) Identification and characterization of major lipidparticle proteins of the yeast Saccharomyces cerevisiae JBacteriol 181 6441ndash6448

69 BellerM RiedelD JanschL DieterichG WehlandJJackleH and KuhnleinRP (2006) Characterization of theDrosophila lipid droplet subproteome Mol Cell Proteomics 51082ndash1094

70 ZhangP NaH LiuZ ZhangS XueP ChenY PuJPengG HuangX YangF et al (2012) Proteomic study andmarker protein identification of Caenorhabditis elegans lipiddroplets Mol Cell Proteomics 11 317ndash328

71 WangH GilhamD and LehnerR (2007) Proteomic and lipidcharacterization of apolipoprotein B-free luminal lipid dropletsfrom mouse liver microsomes implications for very low densitylipoprotein assembly J Biol Chem 282 33218ndash33226

72 RiveraMC and LakeJA (2004) The ring of life providesevidence for a genome fusion origin of eukaryotes Nature 431152ndash155

73 EmbleyTM and MartinW (2006) Eukaryotic evolution changesand challenges Nature 440 623ndash630



INTRODUCTION

Lipid droplets (LDs) are cellular organelles widely foundin fungal plant animal and human cells (1ndash3) They areencapsulated by a phospholipid monolayer and are com-positionally different from other membrane structures (4)They differ in that their primary role is lipid storage butmay also be pivotal in cellular communication with organ-elles such as the mitochondria to regulate energy metab-olism and substrate utilization LD is an importantorganelle related to human metabolic diseases andbiofuel productions For example LD dysfunction isone of the main causes of metabolic disorders such asobesity insulin resistance type 2 diabetes and cardiovas-cular diseases (5ndash9) In biofuel studies triacylglycerol(TAG) in LD of green algae has been investigated anddeveloped for high oil yields by using targeted metabolicengineering (10ndash12) making it a biological candidate forbiofuel production

Delineating the molecular mechanisms of LD dynamicsis essential to understand its formation functions syn-thetic engineering and further biofuel applications Sinceperilipin the first protein of perilipin family (PLIN) wasidentified in 1991 (13) numerous proteins have beenrevealed to be related to LD functions and dynamics(314) LD may also be involved in multiple importantcellular processes such as intermembrane lipid traffic(15) lipid storage (16) lipolysis (17) signaling temporalprotein storage (18) and protein degradation (19) LD isreported functionally interacted with many other organ-elles such as the mitochondria (20) endoplasmic reticulum(2122) endosome (23) and peroxisome (24) Despite thefunctional importance of LDs systematic understandingof the organellersquos biogenesis and dynamics remainselusive In contrast to eukaryotes that have multiple or-ganelles LD is the only membranous organelle found in anumber of bacterial strains that can be used as ideal modelorganisms for LD research Among them Rhodococcusopacus PD630 has the ability to accumulate largeamounts of TAG in the LD (25)

The importance of Rhodococcus opacus strain PD630(R opacus PD630) as a model system is also exemplifiedby its powerful ability to convert carbon sources intolipids Interestingly the TAG storage in R opacusPD630 accounts for up to 87 of the cellular dryweight (26) and thus has higher lipid storage capacitywhen compared with other oleaginous organisms (2627)Early studies reported that R opacus PD630 has 10diacylglycerol acyltransferases (DGAT) that assimilatecellular fatty acids into TAG (1328) Holder et alreported a partial genome and also performed a compara-tive genomic study with a lipid mass analysis (29) whichidentified 16 DGAT and 261 genes that are directlyinvolved in 20 TAG cycle reactions These previousstudies suggest that TAG biosynthesis from carbonsources is a pronounced characteristic of R opacusPD630 Therefore to facilitate the application ofR opacus PD630 LD production for biofuel developmenta complete genome of the organism and integratedanalysis of its transcriptome a proteome of its lipid syn-thesis storage and metabolism are essential

We performed multi-omic studies and present herein thecomplete genome sequence a comparative transcriptomeand a comparative LD proteome of R opacus PD630After integrating the collected data a number of proteinfamilies involved in LD dynamics were identified includinglipid synthesis LD structure-like proteins dynamin-likeand SNARE-like proteins A structure-like proteinLPD06283 was verified by its LD location and its effecton LD size Together these omics are useful tools to in-vestigate the mechanisms of LD dynamics that willenhance our understanding of the lipid storage of LD inbiofuel development

MATERIALS AND METHODS

DNA extraction and genome sequencing and assembly

Cells of R opacus PD630 (30) were obtained from DrSteinbuchelrsquos lab at the University of Munster Cellswere cultured aerobically in 100ml of nutrient broth(NB) at 30C to postlogarithmic phase and then theDNA was extracted The complete nucleotide sequencewas obtained using a combination of paired-endmate-pair Illumina sequencing and 454 sequencing Thesequence gaps were completed by direct sequencing ofpolymerase chain reaction (PCR)-amplified fragmentsFor 454 pyrosequencing genomic DNA was sheared upby nebulization into random fragments of 500ndash800 bp forthe construction of a dispersed library which was thenclonally amplified and sequenced on a 454 GenomeSequencer For Illumina sequencing genomic DNA wasprocessed to construct paired-end libraries with size spansof 300 bp and also mate-pair libraries with size spans of3 kb using an Illumina Genomic DNA Sample Prep kitThe total number of 454 reads obtained was 861 751

giving a 36-fold coverage while the total number ofpaired-end and mate-pair library reads was 40 110 584giving a 445-fold coverage We used two assemblyprograms and combined the primary contigs and paired-end data to build scaffolds in successive assemblies Fourhundred fifty-four sequences were assembled using theRoche GS assembler Newbler (version 25) with defaultparameters The primary contigs were then scaffoldedwith Illumina mate-pair reads using SSPACE-premium(version 21) (31) To close the gaps among scaffoldsread pairs that were uniquely mapped to the contig tailswere extracted for manual assembly Primers weredesigned for the remaining gaps and PCR walking wasused to finish the whole genome Illumina reads (300 bp)were mapped to this assembled whole genome sequence toidentify potential single miss-called nucleotides using theBowtie method (32)

Genome analysis and annotation

Gene models were predicted independently usingGLIMMER (33) and GeneMark (34) The predictedopen reading frames (ORFs) were further evaluated andadjusted using RBSfinder (35) The translated sequencesof the predicted protein-coding genes were searchedagainst UniProt (36) and InterPro (37) The function ofenzymes was assigned using EFICAz2 (38) and searched







AB




LD purification
























Data access


RESULTS












































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES



















































































AB




LD purification
























Data access


RESULTS












































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES

































































































Data access


RESULTS












































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES






















































































































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES













































































































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES







































































































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES






























































































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES
























































































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES





















































































DISCUSSION








SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES



















































































SUPPLEMENTARY DATA


ACKNOWLEDGEMENTS





FUNDING



REFERENCES















































































FUNDING



REFERENCES



















































































































Documents

Integrated omics study delineates the dynamics of lipid droplets in