characterization and use of green fluorescent proteins from renilla mulleri and ptilosarcus guernyi...

507

Journal of Protein Chemistry, Vol. 20, No. 6, August 2001 (© 2001)

0277-8033/01/0800-0507$19.50/0 © 2001 Plenum Publishing Corporation

Characterization and Use of Green Fluorescent Proteinsfrom Renilla mulleri and Ptilosarcus guernyifor theHuman Cell Display of Functional Peptides

Beau Peelle,1,2 Tarikere L. Gururaja, 1,2,3 Donald G. Payan,1 and D. C. Anderson1

Received July 17, 2001

Green fluorescent protein (GFP) is useful as an intracellular scaffold for the display of random pep-tide libraries in yeast. GFPs with a different sequence from Aequorea victoriahave recently beenidentified from Renilla mulleriand Ptilosarcus gurneyi.To examine these proteins as intracellularscaffolds for peptide display in human cells, we have determined the expression level of retrovirallydelivered human codon-optimized versions in Jurkat-E acute lymphoblastic leukemia cells usingfluorescence activated cell sorting and Western blots. Each wild type protein is expressed at 40%higher levels than A. victoriamutants optimized for maximum fluorescence. We have compared thesecondary structure and stability of these GFPs with A. victoriaGFP using circular dichroism (CD).All three GFPs essentially showed a perfect b-strand conformation and their melting temperatures(Tm) are very similar, giving an experimental evidence of a similar overall structure. Folded RenillaGFP allows display of an influenza hemagglutinin epitope tag in several internal insertion sites, in-cluding one which is not permissive for such display in AequoreaGFP, giving greater flexibility inpeptide display options. To test display of a functional peptide, we show that the SV-40 derived nu-clear localization sequence PPKKKRKV, when inserted into two different potential loops, resultsin the complete localization of RenillaGFP to the nucleus of human A549 cells.

KEY WORDS: Green fluorescent proteins; retroviral delivery; peptide libraries; fluorescence; circular dichroism.

ping intracellular signaling pathways. AequoreaGFPhas been used as a peptide presentation scaffold in yeast(Abedi et al.,1998; Caponigro et al.,1998), and has ob-vious advantages as a fluorescent indicator of cells con-taining the library member, as a stable protein scaffoldfor the display of constrained peptides in surface loops,as a protein which is not toxic to mammalian cells, andas a scaffold with a long intracellular half-life. In spiteof the recent availability of a variety of different GFPswith different fluorescence properties, some of en-hanced brightness (Ward and Cormier, 1979; Lorenzet al., 1991; Matz et al., 1999; Szent-Gyorgyi et al.,

1. INTRODUCTION

Fluorescent proteins cloned from marine organisms areof great utility to biological research. Green fluorescentprotein (GFP)4 from the jellyfish Aequorea victoriaandits variants have been used extensively as reporters ofgene expression, for cellular localization of fusion pro-teins, as in vivo sensors, and in biophysical applications(Tsien, 1998). Phenotypic screens of random peptides inyeast have resulted in peptides which can change theyeast phenotype (Colas et al., 1996; Geyer et al., 1999;Norman et al.,1999) and which may be useful for map-

1 Protein Chemistry Department, Rigel Pharmaceuticals, Inc., S. SanFrancisco, California.

2 Both authors have contributed equally to the publication of this work.3 To whom correspondence should be addressed at Protein ChemistryDepartment, Rigel Pharmaceuticals Inc., 240 E. Grand Ave., SouthSan Francisco, California 94080. Tel: (650)624-1126; Fax: (650)624-1101; e-mail: [email protected]

4 Abbreviations: CD, circular dichroism; EGFP, enhanced Aequoreavictoria GFP; FACS, fluorescence-activated cell sorting; GFP, greenfluorescent protein; HEPES, N-2-hydroxyethylpiperazine-N8-2-ethanesulfonic acid, HA, influenza hemagglutinin epitope; NLS, nu-clear localization sequence; PBS, phosphate buffered saline; PCR,polymerase chain reaction; SDS-PAGE, sodium dodecyl sulfate poly-acrylamide gel electrophoresis.

1999), none besides AequoreaGFP has been utilized asa peptide display scaffold and none have been testedin human cells. Some properties, such as intracellularconcentration and tolerance of the scaffold to internalloop replacements or insertions to constrain randompeptides, will be important for a GFP or any other dis-play scaffold.

Here we examine GFPs from the sea pansy Renillamulleri and the sea pen Ptilosarcus gurneyi(Szent-Gyorgyi and Bryan, 1999), which are 25.1% and 22.7%identical to enhanced AequoreaGFP, as potential pep-tide display scaffolds in human cells. We compare theirexpression levels in human acute lymphoblasticleukemia Jurkat-E cells with those of AequoreaGFP(Peelle et al., 2001). Because both are of unknownstructure, we have compared their secondary structureand stability with AequoreaGFP (Ormo et al., 1996;Yang F. et al., 1996) using CD. We also examine thelocation of potential loops and thus loop replacementor insertion sites, by inserting an HA tag into six dif-ferent sites at locations similar to loops in AequoreaGFP. The results of this epitope tag scan suggest thatthis class of Anthozoan GFPs may be suitable peptidedisplay scaffolds for the use of peptides as stable intra-cellular reagents and for screens in mammalian cells,and suggest the existence of a unique peptide displaysite in Renilla GFP. We also demonstrate the activityof an inserted functional peptide in two of the RenillaGFP sites.

2. MATERIALS AND METHODS

2.1. Materials

All the antibodies used in this study were procuredfrom indicated source: anti-A. victoria GFP rabbitserum (Molecular Probes, Eugene, OR), anti-Flag M2agarose affinity gel and goat anti-rabbit IgG-horseradish peroxidase conjugate (Sigma, St. Louis, MO),IkB-a (C-15) agarose conjugate and rabbit polyclonalflag-probe (Santa Cruz Biotechnology, Santa Cruz,CA), and anti-HA affinity matrix (Babco, Berkeley,CA). The purified recombinant EGFP was obtainedfrom Clontech (Palo Alto, CA), and complete EDTA-free protease inhibitor cocktail was from BoehringerMannheim (Chicago, IL). Plasmids coding wild typeR. mulleriand Ptilosarcus gurneyiGFPs were obtainedfrom Dr. C. Szent-Gyorgyi (Prolume Ltd., Pittsburgh,PA). All restriction endonucleases were from New Eng-land Biolabs (Beverly, MA). All cell growth mediawere from JRH Biosciences (Williamsburg, VA), exceptas noted.

2.2. Cells and Retroviral Constructs

Phoenix retroviral packaging cells (Swift et al.,1999) were carried in 10% fetal bovine serum with 1%penicillin-streptomycin and Dulbecco’s modified Eaglemedia (Mediatech Cellgro, Herndon, VA). Human acutelymphoblastic leukemia Jurkat-E cells stably expressingthe ecotropic receptor were cultured in RPMI1640medium, supplemented with 10% fetal calf serum plus1% penicillin-streptomycin in a 37°C incubator with 5%CO2. Calcium phosphate transfection of Phoenix cellsand infection of Jurkat-E cells was carried out as de-scribed in Swift et al. (1999). Retroviral constructs werebased on p96.7, a retroviral vector with a compositeCMV promoter (Lorens et al., 2000). The pCGFP wascreated from 96.7 and contains human codon-optimizedenhanced GFP (Cormack et al., 1996). The EGFP genewas obtained from Clontech (Palo Alto, CA), with aKozak consensus start sequence. p96.7EGFP-cFlag wasmade by ligation of cFlagUP plus cFlagDN intoXhoI/NotI digested p96.7EGFP, which is identical topCGFP but has additional restriction sites in the openreading frame of EGFP resulting in eight nonoptimizedcodons.

p96.7rmg and p96.7pgg are retroviral expressionvectors containing human codon-optimized P. gurneyiand R. mulleri GFPs, with 11 and 9 nonoptimizedcodons to create restriction sites, respectively. Each hasa Kozak consensus start and backbone vector sequenceidentical to that of p96.7EGFP. These vectors weremade by annealing and ligating 20 synthetic oligonu-cleotides (R1-R20, P1-P20), followed by amplificationof the fragments by PCR with rmgUP plus rmDN, andpggUP plus pgDN primers, respectively, digestion withEcoRI/NotI, and ligation into the EcoRI/NotI digested96.7EGFP vector. C-terminal flag tags were added tothese GFPs by ligation of annealed oligos rm/pgFlag UPplus rm/pgFlag DN into BamHI/NotI digested parentvectors to create p96.7rmgGFP-Flag and p96.7pggGFP-Flag. The p96.7RM cDNA was made by removing thewild type gene from the pET-34 plasmid containing R.mulleri GFP by PCR with primers RMcDNA UP plusRMcDNA DN, digestion with EcoRI/NotI, and ligationinto EcoRI/NotI digested 96.7 EGFP. p96.7rmgA.HA(and loops B, C, D, E, and F) were constructed by PCRligation of two fragments made by PCR of 96.7 rmgGFPwith rmgGFP UP plus rmgA DN, and with rmgA UPplus rmGFP DN, followed by digestion with EcoRI/NotI and ligation into EcoRI/NotI digested 96.7 EGFP.The resulting constructs contain codon optimized RenillaGFP with a linker-HAtag-linker sequence inserted intoeach position A-F. C-terminal flags were added to these

508 Peelle, Gururaja, Payan, and Anderson

vectors in the same manner as stated above, creating96.7rmgA.HA-Flag (and equivalents for loops B, C, D,E and F).

The bacterial expression vector for purification ofthe PtilosarcusGFP, pGEX6P-pgGFP was created byPCR of p96.7pggGFP with G6Ppgg UP plus pgGFP DN,digestion with BglII/Not I and ligation into BamHI/NotIdigested pGEX6P-1 (Pharmacia Biotech, Piscataway,NJ). pGEX6P-rmg8GE8GFP was created by PCR liga-tion as follows. The fragment was created by annealingand extending rmgGE UP plus rmgGE DN was ligatedwith a second fragment made by PCR of 96.7rmgGFP

with rmg58 UP plus RMG DN, using primers G6PrmgUP plus RMG DN. The resulting fragment was digestedwith BglII/NotI and ligated into BamHI/NotI digestedpGEX6P-1. This vector expresses R. mulleri GFP withC10G and C35E mutations to aid in folding of the pro-tein in bacteria.

2.3. Synthetic Oligonucleotides Used in VectorConstruction

All oligonucleotides were synthesized by Biosource,Inc. (Foster City, CA).

Renilla GFP as a Peptide Library Scaffold 509

EGFP-UP 58GATCGTCGACGAATTCCACCATGGTGAGCAAGGGCGAGGHdDN 58TAGATCAAGCTTGTGCCCCAGAATATTGCCGTCCTCCTTGAAATCGATGCCCTTCAGCTCAbUP 58GATCCTCGAGAAGCTTGAGTACAACTACAACAGCCACAACGTGTATATCATGGCCGA

CAAGCAGAAGAACGGCATCAAGGTAACTTCAAGATAbDN 58GATCGAATTCAATTGGGGTGTTCTGCTGGTAGTGGTCGGCGAGCTGCACGGATCCGTC

CTCGATGTTGTGGCGGATCTTGAAGTTAACCTTGATTxUP 58GATCCAATTGGCGACGGGCCCGTGCTGCTGCCCGACAACCACTACCTGAGCACCCA

GAGCGCTCTTTCGAAAGACCCCAACGAGAAGCGCGATCATATGGTxDN 58GATCGCGGCCGCTTACTTGTACAGCTCGTCCATGCCGAGAGTGATCCCGGCGGCGGT

CACGAACTCGAGCAGGACCATATGATCGCGCTTCTCGTTGGGGL3LUP 58AACTTCAAGATCCGCCACAACATCGAGGGCCAGGGCGGTGGCCAAGCAGGTGGAG

GTGGCGL3LDN 58GATCCGCCACCTCCACCTGCTTGGCCACCGCCCTGGCCCTCGATGTTGTGGCG

GATCTTGAAGTTZUP 58GATCTACTCCAGGGCGGTGGGACATGTTGCAGATCCATGZDN 58TCTAGATCCCACCTGCTTGGCCTGATTCATTCCCCAGCG3.HA UP 58GTGGCTACCCCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAG3.HA DN 58TTGGCCCAGGCTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCC3.SV40 UP 58GTGGCCCTCCAAAAAAGAAGAGAAAGGTAGCTGGCCAAGCAG3.SV40 DN 58TTGGCCAGCTACCTTTCTCTTCTTTTTTGGAGGGCCACCGCCcFlagUP 58P TCGAGTTCGTGACCGCCGCCGGGATCACTCTCGGCATGGACGAGCTGTACAAGGAG

GAGGCCGCCAAGGCCGACTACAAGGACGACGACGACAAGTAGGCCCGTGAGGCCCTAAGC

cFlagDN 58P GGCCGCTTAGGGCCTCACGGGCCTACTTGTCGTCGTCGTCCTTGTAGTCGGCCTTGGCGGCCTCCTCCTTGTACAGCTCGTCCATGCCGAGAGTGATCCCGGCGGCGGTCACGAAC

RMG UP 58GATCATAGAATTCGCCACCATGGGCAGCAAGCAGATCCTGAAGAACACCTGCCTGRMG DN 58ATGATCGCGGCCGCTACACCCACTCGTGCAGGGATCCCAGGGGCTTGCCGATGR1 58GCAGATCCTGAAGAACACCTGCCTGCAGGAGGTGATGAGCTACAAGGTGAACCTG

GAGGGCATCGTTAACAAR2 58CCACGTGTTCACCATGGAGGGCTGCGGCAAGGGCAACATCCTGTTCGGCAACCAATTG

GTGCAGATCCGCGTR3 58GACCAAGGGCGCCCCCCTGCCCTTCGCCTTCGACATCGTGAGCCCCGCCTTCCAGTACG

GCAACCGTACGTTR4 58CACCAAGTACCCCAACGACATCAGCGACTACTTCATCCAGAGCTTCCCCGCCG

GCTTCATGTACGAGCGCACR5 58CCTGCGCTACGAGGACGGCGGCCTGGTGGAGATCCGCAGCGACATCAACCTGATCGAG

GACAAGTTCGTGTAR6 58CCGCGTGGAGTACAAGGGCAGCAACTTCCCCGACGACGGGCCCGTGATGCGGAAGAC

CATCCTGGGCATCGA


R7 58GCCCAGCTTCGAGGCCATGTACATGAACAACGGCGTGCTGGTGGGCGAGGTGATCCTGGTGTACAAGCTTAA

R8 58CAGCGGCAAGTACTACAGCTGCCACATGAAGACCCTGATGAAGAGCAAGGGCGTGGTGAAGGAGTTCCCCAG

R9 58CTACCACTTCATCCAGCACCGCCTCGAGAAGACCTACGTGGAGGACGGCGGCTTCGTGGAGCAGCACGAGAC

R10 58CGCCATCGCCCAGATGACCAGCATCGGCAAGCCCCTGGGATCCCTGCAR11 58TGCAGGGATCCCAGGGGCTTGCCGATGCTGGTCATCTGGGCGATGGCGGTCTCGTGCT

GCTCCACGAAGCCGCCGTCCTCCACGR12 58TAGGTCTTCTCGAGGCGGTGCTGGATGAAGTGGTAGCTGGGGAACTCCTTCACCACGC

CCTTGCTCTTCATCR13 58AGGGTCTTCATGTGGCAGCTGTAGTACTTGCCGCTGFTAAGCTTGTACACCAGGATCAC

CTCGCCCACCAGCR14 58ACGCCGTTGTTCATGTACATGGCCTCGAAGCTGGGCTCGATGCCCAGGATGGTCTTCTG

CATCACGGGCCCGR15 58TCGTCGGGGAAGTTGCTGCCCTTGTACTCCACGCGGTACACGAACTTGTCCTC

GATCAGGTTGATGTCGCTGR16 58CGGATCTCCACCAGGCCGCCGTCCTCGTAGCGCAGGGTGCGCTCGTACATGAAGCCG

GCGGGGAAGCTCTGGR17 58ATGAAGTAGTCGCTGATGTCGTTGGGGTACTTGGTGAACGTACGGTTGCCGTACTG

GAAGGCGGGGCTCACGR18 58ATGTCGAAGGCGAAGGGCAGGGGGGCGCCCTTGGTCACGCGGATCTGCACCAATTG

GTTGCCGAACAGGATGR19 58TTGCCCTTGCCGCAGCCCTCCATGGTGAACACGTGGTTGTTAACGATGCC

CTCCAGGTTCACCTTGTAGCTCR20 58ATCACCTCCTGCAGGCAGGTGTTCTTCAGGATCTGCPGG UP 58GATCATAGAATTCGCCACCATGGGCAACCGCAACGTGCTGAAGAACACCGGCCTGPGG DN 58ATGATCGCGGCCGCTACACCCACTCGTGCAGGGATCCCAGGGGCTTGCCGATGP1 58CAACGTGCTGAAGAACACCGGCCTGAAGGAGATCATGAGCGCCAAGGCCAGCGTG

GAGGGCATCGTTAACAAP2 58CCACGTGTTCAGCATGGAGGGCTTCGGCAAGGGCAACGTGCTGTTCGGCAAC

CAATTGATGCAGATCCGCGTP3 58GACCAAGGGCGGCCCCCTGCCCTTCGCCTTCGACATCGTGAGCATCGCCTTCCAGTACG

GCAACCGTACGTTP4 58CACCAAGTACCCCGACGACATCGCCGACTACTTCGTGCAGAGCTTCCCCGCCG

GCTTCTTCTACGAGCGCAAP5 58CCTGCGCTTCGAGGACGGCGCCATCGTGGACATCCGCAGCGACATCAGCCTGGAGGAC

GACAAGTTCCACTAP6 58CAAGGTGGAGTACCGCGGCAACGGCTTCCCCAGCAACGGGCCCGTGATGCAGAAGGC

CATCCTGGGCATGGAP7 58GCCCAGCTTCGAGGTGGTGTACATGAACAGCGGCGTGCTGGTGGGCGAGGTGGACCTG

GTGTACAAGCTTGAP8 58GAGCGGCAACTACTACAGCTGCCACATGAAGACCTTCTACCGTTCGAAGGGCGGCGT

GAAGGAGTTCCCCGAP9 58GTACCACTTCATCCACCACCGCCTCGAGAAGACCTACGTGGAGGAGGGCAGCTTCGTG

GAGCAGCACGAGACP10 58CGCCATCGCCCAGCTGACCACCATCGGCAAGCCCCTGGGATCCCTGCAP11 58TGCAGGGATCCCAGGGGCTTGCCGATGGTGGTCAGCTGGGCGATGGCGGTCTCGTGCT

GCTCCACGAAGCTGCCCTCCTCCACGP12 58TAGGTCTTCTCGAGGCGGTGGTGGATGAAGTGGTACTCGGGGAACTCCTTCACGCCGC

CCTTCGAACGGTAG


P13 58AAGGTCTTCATGTGGCAGCTGTAGTAGTTGCCGCTCTCAAGCTTGTACACCAGGTCCACCTCGCCCACCAGC

P14 58ACGCCGCTGTTCATGTACACCACCTCGAAGCTGGGCTCCATGCCCAGGATGGCCTTCTGCATCACGGGCCCG

P15 58TTGCTGGGGAAGCCGTTGCCGCGGTACTCCACCTTGTAGTGGAACTTGTCGTCCTCCAGGCTGATGTCGCTG

P16 58CGGATGTCCACGATGGCGCCGTCCTCGAAGCGCAGGTTGCGCTCGTAGAAGAAGCCGGCGGGGAAGCTCTGC

P17 58ACGAAGTAGTCGGCGATGTCGTCGGGGTACTTGGTGAACGTACGGTTGCCGTACTGGAAGGCGATGCTCACG

P18 58ATGTCGAAGGCGAAGGGCAGGGGGCCGCCCTTGGTCACGCGGATCTGCATCAATTGGTTGCCGAACAGCACG

P19 58TTGCCCTTGCCGAAGCCCTCCATGCTGAACACGTGGTTGTTAACGATGCCCTCCACGCTGGCCTTGGCGCTC

P20 58ATGATCTCCTTCAGGCCGGTGTTCTTCAGCACGTTGrm/pgFlagU 58GATCCCTGCACGAGTGGGTGGAGGAGGCCGCCAAGGCCGACTACAAGGACGACGAC

GACAAGTAGGCCCGTGAGGCCCTAAGCrm/pgFlagDN 58GGCCGCTTAGGGCCTCACGGGCCTACTTGTCGTCGTCGTCCTTGTAGTCGGCCTTGGCG

GCCTCCTCCACCCACTCGTGCAGGRMcDNA UP 58GATCATGAATTCGCCACCATGAGTAAACAAATATTGAAGAACACTRMcDNA DN 58TAGATCGCGGCCGCTTAAACCCATTCGTGTAAGGATCCTAGTGGrmgA UP 58CCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAGGTGGAGGCGACGGCGGC

CTGGTGGAGATCCGCArmgA DN 58CTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCCCTGGCCCTCGTAGCGCAGGGT

GCGCTCGTACrmgB UP 58CCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAGGTGGAGGCGACAAGTTCGT

GTACCGCGTGGAGTrmgB DN 58CTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCCCTGGCCCTCGATCAGGTTGAT

GTCGCTGCGGrmgC UP 58CCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAGGTGGAGGCAACGGCGT

GCTGGTGGGCGAGGTGArmgC DN 58CTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCCCTGGCCGTTCATGTACATG

GCCTCGAAGCTGrmgD UP 58CCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAGGTGGAGGCAGCGGCAAG

TACTACAGCTGCCACArmgD DN 58CTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCCCTGGCCGTTAAGCTTGTA

CACCAGGATCACCrmgE UP 58CCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAGGTGGAGGCGTGGTGAA

GAGTTCCCCAGCTACCrmgE DN 58CTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCCCTGGCCGCCCTTGCTCTTCAT

CAGGGTCTTCrmgF UP 58CCTACGACGTGCCCGACTACGCCAGCCTGGGCCAAGCAGGTGGAGGCTTCGTG

GAGCAGCACGAGACCGCCArmgF DN 58CTGGCGTAGTCGGGCACGTCGTAGGGGTAGCCACCGCCCTGGCCGCCGCCGTCCTC

CACGTAGGTCTTCG6PrmgUp 58AGATCATAGATCTGAATTCATGGGrmgGE UP 58AGATCATAGATCTGAATTCATGGGCAGCAAGCAGATCCTGAAGAACACCGGCTG

GCAGGAGGTGATGAGCTACAAGGTGAACCTGGAGGrmgGE DN 58GCCGAACAGGATGTTGCCCTTGCCCTCGCCCTCCATGGTGAACACGTGGTGGTTAAC

GATGCCCTCCAGGTTCACCTTGTAGCTCATCACG6PpggUP 58AGATCATAGATCTATGGGCAACCGCAACGTGCTGAAGAACACCGGCCTG

2.4. FACS and Microscopy

Flow-cytometry analysis and cell sorting of GFPexpressing cells were performed on a FACScan (Beck-ton-Dickson, San Jose, CA) or a MoFlo (Cytomation,Fort Collins, CO) instrument, and data were analyzedusing FloJo software (Treestar Software, San Carlos,CA). Live cells were gated by scater and propidiumiodide (PI) staining during data analysis. GFP fluores-cence intensity measurements were of GFP positivecells only. For Fig. 4 (Column A), FL1 channel GFP in-tensity output was shifted ca. 2.6 log units left relativeto Fig. 4 (Column B) by adjustment of FL1 voltagecompensation on the MoFlo cytometer so that highGFP fluorescence values were within dynamic range.Cells expressing GFP were visualized using a NikonElipse TE300 fluorescence microscope.

2.5. SDS-PAGE, Immunoprecipitation and WesternBlots

For preparation of whole-cell lysates, cells werecounted, collected, washed in PBS, and lysed by freeze-thaw/vortexing in lysis buffer (50 mM HEPES pH 7.4,150 mM NaCl, 5 mM EDTA, 5 mM EGTA, 1% TritonX-100) with added complete protease inhibitor cocktail(Boehringer Mannheim, Chicago, IL). Lysate removedfrom centrifuged samples was separated on 4% to 12%NuPage SDS polyacrylamide gels (Novex, San Diego,CA), per the manufacturer’s recommendations. For im-munoprecipitations, antibody conjugated agarose beadswere added to the cell lysate, incubated for 4 hours, thebeads were washed in lysis buffer, and samples were sep-arated by SDS-PAGE as above. Samples transferred toPVDF membranes were blocked using PBS buffer con-taining 10% Milk, 0.1% Tween20 overnight at 4°C. Pri-mary antibodies were used at a 1:2000 dilution, second-ary antibodies were used at 1:5000 dilution. Membraneswere developed using ECL plus enhanced chemilumines-cence kit (Amersham Pharmacia, Piscataway, NJ) anddetected using Hyperfilm ECL (Amersham Life Science,Buckinghamshire, UK). For comparative Western blotanalysis, GFPs containing a C-terminal flag tag wereused. Exposed film was scanned with a Hewlett Packard(Palo Alto, CA) ScanJet 4C scanner and band intensitieswere integrated using the program NIH Image (http://rsb.info.nih.gov/nihimage/about.html).

2.6. GFP Purification and Characterization

All components used for purification of the GFPgene products were from Pharmacia Biotech (Piscat-away, NJ). The codon optimized gene for each protein

was expressed in BL21 TIL codon plus (DE3) E. coli(Stratagene, San Diego, CA) as a fusion protein with glu-tathione S-transferase from pGEX6P-1 derived vectors.Each protein was purified using Glutathione Sepharose4B beads, per the manufacturer’s recommendations, andthe mature GFP was removed from the fusion proteinwith Precision Protease. The purified proteins ran as sin-gle bands by SDS-PAGE and appeared as single peaksof the expected molecular mass when examined with aBruker Reflex III time of flight mass spectrometer(Bruker Daltonics, Billerica, MA). Owing to the cloningstrategy employed, purified R. mulleri GFP has theamino acids GPLGSEF-and P. gurneyiGFP the residuesGPLGS-fused to their N-termini.

2.7. Circular Dichroism (CD) Studies

CD spectra were recorded on an AVIV 62A DS CDspectropolarimeter (Lakewood, NJ) equipped with aPeltier temperature control unit. The temperature of theinstrument was maintained constantly below 20°C usingNeslab CFT-33 refrigerated recirculator water bath. Thedevice was periodically calibrated with the ammoniumsalt of (1)-10-camphorsulfonic acid, according to man-ufacturer’s recommendations. Spectra were recorded be-tween 250 and 200 nm at 0.2 nm intervals with a timeconstant of 1 s at 25°C in 10 mM phosphate buffer con-taining 100 mM KF (pH 7.5), as described previously(Gururaja et al., 2000). A cylindrical quartz cell of pathlength 0.1 cm was used for the spectral range with thesample concentration of 5 to 10 mM, as determined byLowry et al., (1951) method. Mean residual ellipticity(MRE) is expressed in deg.cm2/dmol. The thermaldenaturation was measured at 218 nm over a range of4°C to 98°C with a temperature step of 2°C and a2-minute equilibration time and a 60-second signal aver-aging time. The Tm data were fitted to a logistic sigmoidequation using the Levenberg-Marquardt algorithm inUltrafit (Biosoft, Cambridge, UK). CD spectra were de-convoluted with the program CDNN (Bohm et al.,1992)downloaded from http://bioinformatic.biochemtech.uni-halle.de/cdnn/index.html.

3. RESULTS

3.1. Expression in Mammalian Cells

The R. mulleriand P. gurneyiGFP genes were con-structed with a glycine following the initial methionineto optimize translation (Kozak, 1986), and were codon-optimized for efficient expression in human cells (Yang,


et al., 1996). Both GFPs, as well as EGFP (Cormacket al.,1996), were retrovirally expressed in Jurkat-E cellsusing the protocol of Swift et al. (1999). Based on FACSanalysis of scatter and PI staining of cell populationsfrom 13 hours to 8 days post-infection, there was noobserved toxicity of either Ptilosarcusor Renilla GFP(data not shown). By 2 days post-infection, the accumu-lation of intracellular GFP had appeared to slow to asteady-state level. Based on FL1 channel fluorescence,this occurred more rapidly for the Ptilosarcusand RenillaGFPs than for EGFP (data not shown). The excitationand emission maxima were 501 and 511nm, 498 and509 nm, and 489 and 510 nm for PtilosarcusGFP, RenillaGFP, and AequoreaGFP, respectively, which agrees withresults from Szent-Gyorgyi et al. (1999) and Cormacket al. (1996).

The relative levels of wild type and codon opti-mized Renilla GFP and EGFP were analyzed by FACSat 4 days post-infection (Fig. 1A). Based on geometricmean fluorescence values in the FL1 channel, humancodon-optimized Renilla GFP was expressed greaterthan 28-fold higher than the wild type cDNA sequence,and was 1.4-fold brighter than EGFP. PtilosarcusGFP,Renilla GFP, and EGFP were also fused at their C-ter-mini to a linker-flat tag sequence, -EEAAKA-DYKDDDDK, expressed in Jurkat-E cells, and their fluores-cence levels compared by FACS (Fig. 1A). ThesePtilosarcusand RenillaGFPs were, on average, 1.4-foldand 1.2-fold more fluorescent than EGFP, respectively.Lysates from 2.8 3 104 Jurkat-E cells, sorted 8 days afterinfection for GFP fluorescence, were compared by West-ern blot using anti-flag antibody (data not shown). AllGFPs gave only a single band. EGFP-flag ran at aslightly higher molecular mass than the other two flag-tagged GFPs. The integrated intensity values derivedusing NIH Image were 3200, 3206, and 2314 for eachband, and had ratios of 1.4: 1.4: 1.0 for PtilosarcusGFP,RenillaGFP, and EGFP, respectively. Thus, both Renillaand Ptilosarcusare expressed at slightly higher levelsthan EGFP in these cells and are likely to be efficientreporters of gene expression.

3.2. Bioinformatics of Renilla and PtilosarcusGFPs

A clustal W 1.8 (Thompson et al.,1994) alignmentof the recently released Renillaand PtilosarcusGFP se-quences (Szent-Gyorgyi and Bryan, 1999) with otherAnthozoan GFP sequences, excluding red emitting pro-teins, and EGFP is shown in Fig. 2. The Renilla andPtilosarcusGFPs are 25.1% and 22.7% identical to thatfrom Aequorea.Loops or turns in the AequoreaGFPcrystal structure (Yang F et al.,1996) are underlined. Be-

cause the Renilla and PtilosarcusGFPs and the otherAnthozoan GFPs as a group appear to be rather differentfrom the Aequoreasequence, we examined the similari-ties at the level of secondary structure using CD.

3.3. Conformational Analysis

Figure 3A shows overlaid CD spectra from eachprotein, present at 5 to 10 mM, at 25°C. All three GFPsshowed a broad positive π-π* band around 195 nm andnegative n-π* band centered around 218 nm, which es-sentially suggests a b-strand conformation with minorpopulation of disordered polypeptide chain. The putativeb-structure peptide has been found to posses a positiveπ-π* band around 195 nm and a characteristic negativen-π* band around 217 nm (Greenfield and Fasman,1969). The three spectra appear very similar, suggestingthat the overall secondary structure of these three pro-teins is also quite similar. The spectra were deconvolutedinto contributions of individual secondary structuresusing CDNN (Bohm et al.,1992) and the results for eachGFP were identical (data not shown). PtilosarcusGFP(filled squares) has a minor second minimum at 223 to224 nm, which is not present in the other two proteins.This observation could be due (Woody and Dunker,1996) to an increased percentage of aromatic aminoacids, which comprise 13.9% of PtilosarcusGFP, com-pared with 12.6% and 10.0% for RenillaGFP and EGFP,respectively. To compare the relative stabilities of allthree GFPs we examined their thermal melting curves at218 nm (Fig. 3B). AequoreaGFP has a measured Tm of83.7°C, similar to the reported value of 81.9°C (Topellet al., 1999). The calculated Tm values for Ptilosarcusand Renilla GFPs are 80.5°C and 86.1°C, respectively.Thus, all three GFPs have structures of similar stabilityas indicated by high Tm values.

3.4. Epitope Tag Insertion Scan for Loops Suitablefor Presentation of Peptides

To test for the location of potential surface loops inRenilla GFP, we inserted the sequence GQGGGYPYD-VPDYASLGQAGGG, which contains the influenzahemagglutinin epitope tag (underlined) flanked by twoflexible linker sequences, in candidate sites based on theAequoreaGFP structure. We then examined the fluores-cence of the resulting modified retrovirally expressedGFPs in human cells. It is expected that insertion of this22mer in the middle of the b-strands may cause mis-folding, while insertion into surface loops or turns mayallow GFP to fold and thus fluoresce (Abedi et al.,1998).


Modified EGFPs, with residues in three different loop sites(indicated by red type in Fig. 2) replaced with this standardHA insert, have been shown to be fluorescent whenexpressed in human cells (Peelle et al., 2001). Six differ-ent insertion sites, A-F, were tested in codon-optimizedRenilla GFP, as indicated in Fig. 2. The residues oneither side of the insertion are marked in bold white fontshaving black background. Figure 4 shows the fluores-

cence of the different modified RenillaGFPs retrovirallyexpressed in Jurkat-E cells, analyzed by FACS 4 dayspost-infection. The geometric mean fluorescence valuesfor the populations indicated by the gates are shown inthe upper right corner of each FACS plot. Comparisonsof these values are for samples that have populationspresent within the same dynamic range. All modified Re-nilla GFPs, except that with insertion into position A,


Fig. 1. Retroviral expression of human codon-optimized Renilla mulleri, Ptilosarcus gurneyi,and Aequorea victoriaGFPs in Jurkat-E cells. FACS plots of wild type (WT) and codon-optimized (R) Renilla GFP, AequoreaGFP (E), and flag-tagged versions of PtilosarcusGFP(Pf), RenillaGFP (Rf) and AequoreaGFP (Ef) were obtained 4 days after infection of Jurkat-Ecells. Codon optimization increases RenillaGFP expression more than 21-fold. Both Ptilosarcusand RenillaGFPs have higher fluorescence intensities than AequoreaGFP. Uninfected cells areshown off-scale owing to shift of the dynamic range ca 2.6 log units to the left by FL1compensation on the cytometer. Geometric mean fluorescence values are listed in the upper rightcorner for each population within the gated region shown in blue.

Fig. 2. Alignment of Anthozoan GFPs withAequoreaGFP using Clustal W. TheRenilla mulleri(RENM) andPtilosarcus gurneyi(PTIL) sequencesare shown below theAequoreaGFP sequence at the bottom. The sequences of the bottom 4 Anthozoan GFPs that emit light between 483–506 nm(ANEM, Anemonia majanoGFP; DSFP,Discosoma striataGFP; FP48,Clavularia GFP; ZFP5,ZoanthusGFP) are from Matzet al. (1999). The first35 residues are removed from the N-terminus of FP48. A consensus residue was listed if at least 4 of the 7 residues were identical. Residuescomprising turns and loops between beta strands in theAequoreaGFP based on visual analysis of theAequoreacrystal structure (Yanget al.,1996)are underlined. The two residues on either side of the site of addition of the 22mer peptide in theRenilla mullerisequence are listed in bold whitefonts having black background and designated as loops A-F in bold type. The corresponding replacement sites inAequoreaGFP that allow formationof a fluorescent protein (Peelleet al.,2001) are in bold black fonts having gray background. The fluorophore sequences are shown in bold italics type.

D and F were ca. 49% and 47%, and B, C, and E lessthan 1%. Thus, Renilla GFP with HA tags inserted intopositions D and F best tolerate insertion of the 22merpeptide. Renilla GFP with the position D insertion was,on average, 2.3-fold more fluorescent than AequoreaEGFP with the identical 22mer present in its most fluo-rescent loop (Peelle et al.,2001). In EGFP, position D isa surface loop about 5Å across the top of the loop, whileposition F is a larger surface loop, about 11Å across thetop of the loop.

3.5. Intracellular Presentation of an Active Peptidein Loops D and F

To further examine Renilla GFP as a peptidedisplay scaffold, we inserted the SV-40 derived NLS-PPKKKRKV- flanked by the glycine linkers used in theepitope tag scan in sites D and F. This peptide interactswith karyopherins in the nuclear pore complex for trans-port into the nucleus Moroianu et al.,1995; (Radu et al.,1995; Rexach and Blobel, 1995). Thus, successful dis-play of this peptide in a folded GFP structure shouldresult in GFP fluorescence mainly in the nucleus of acell. Figure 5 shows the results of this experiment. About106 A549 cells with retrovirally expressed Renillasite Dor F-inserted peptide were grown for 14 days and thenobserved by fluorescence microscopy. The HA epitopetag flanked by 4 glycines, ,G4YPYDVPDYASLG4,,was inserted along with the linker residues as a controlfor each experiment. GFP with this tag inserted in bothsites D and F fluoresced throughout the cell, while theNLS-containing insert showed only nuclear fluores-cence, with some preferential localization to intra-nuclear structures for the loop D NLS insert. The posi-tion of GFP in the nucleus was confirmed by observationof the cell by light microscopy (data not shown).

4. DISCUSSION

In this paper we have examined GFP from R. mul-leri and P. gurneyias scaffolds for the presentation offunctional peptides within a mammalian cell. These pro-teins have a significantly different sequence from that ofenhanced AequoreaGFP, with sequence identities of25.1% and 22.7% using FASTA 2.0 comparisons of thefull-length proteins (Pearson and Lipman, 1988; data notshown). Such low sequence identities make homologymodeling to a known structure difficult, and can causeproblems in global sequence alignments (Sauder et al.,2000). They have a higher sequence homology to otherrecently reported Anthozoans (Matz et al., 1999), rang-


Fig. 3. CD studies of Aequorea victoria, Renilla mulleri,andPtilosarcus gurneyiGFPs. (A) CD spectra from 200–250 nm of EGFP(open circles), Renilla (open gray squares) and Ptilosarcus (filledsquares) GFPs, taken at pH 7.5 in 10 mM potassium phosphate bufferwith 0.1 M potassium fluoride. Deconvolution of these spectra indicatesthe secondary structure content of all three GFPs to be identical.(B) Thermal denaturation curves for the three GFPs as detected by CD.The most stable protein was Renilla GFP (open circles), with a Tm of86.1°C, followed by EGFP (filled squares) with a Tm of 83.7°C andPtilosarcusGFP (open triangles) with a Tm of 80.5°C.

A

B

were expressed and fluorescent. The rank order of fluo-rescence intensities was D . F .. B . E 5 C. Rela-tive to unaltered Renilla GFP, the average expressionlevels of Renilla GFP with the HA peptide in positions

Fig. 4. FACS analysis of Jurkat-E cell expression of Renilla GFP with a 22mer HA epitope tag inserted into positions A-F. Plots in Column A areshown with a standard fluorescence scale. For plots in Column B, FL1 channel compensation was used to shift the fluorescence detection range ca.2.6 log units to the left to observe the high levels of GFP fluorescence. RenillaGFP is shown without an insert (R), and with inserts in positions A,B, C, D, E, and F as labeled. AequoreaGFP is shown without an insert (EGFP) and with the same insert in its equivalent of position D (EGFP3).The sites of insertion, A-F, are shown in blue type in Fig. 2. The constructs were retrovirally expressed in Jurkat-E cells, and analyzed by FACS 4days post-infection. The GFP geometric mean fluorescence values from the gated regions are listed in the upper right of each plot. D, F, and EGFP3retain 30% to 49% of their respective parent GFP fluorescence levels. B, C, and E had observable but much lower levels of fluorescence than theparent RenillaGFP. The position A insert had almost no measurable fluorescence above background.

ing from 39% to 48% homology for Renillaand 39% to47% for Ptilosarcus,and thus may be representative ofthis class of GFP scaffolds. A high level of expressionmay be important for the use of these scaffolds for intra-cellular peptide display, because the concentration of arandom peptide library member at the time of a func-tional screen will control the saturation of binding to anypotential binding partner that is coupled to the desiredphenotype of the screen. Lower levels of peptide maythus saturate fewer binding partners and the screens maythus find fewer intracellular targets coupled to desiredphenotypes. We found that each scaffold is expressed atca. 40% higher levels than EGFP, suggesting that eachmay be at least as useful for peptide presentation. Thebright fluorescence of wild type Renilla and PtilosarcusGFPs in mammalian cells, with no optimization ofexpression by mutagenesis as has occurred for EGFP(Cubitt et al.,1995) suggests that mutagenesis could cre-ate both more highly expressed and brighter (more sen-sitive) intracellular reporters.


Fig. 5. Display of an NLS peptide inserted into sites D and F of Renilla retrovirally expressed in A549 cells. Expression of an HA epitope tagin sites D and F is shown in panels 1 and 2, respectively, and results in fluorescence throughout the cell. Expression of the NLS-peptide, derivedfrom SV-40, in sites D and F in panels 3 and 4 results in fluorescence only in the nucleus. This shows that the inserted NLS sequence is functionalwhen presented in this fashion by the RenillaGFP scaffold, which is folded, when containing the peptide inserts, due to its fluorescence.

A second desirable property for a scaffold is its abil-ity to present constrained peptides, which in this case arerepresented as peptides inserted into surface loops of thescaffold. Such loops are readily discerned in the structureof AequoreaGFP (Abedi et al.,1998; Peelle et al.,2001)but useful insertion sites are not completely obvious forthe Renillaand PtilosarcusGFPs owing to their relativelylow sequence identity compared with AequoreaGFP anda rather different sequence in the vicinity of Aequorealoops C, D, E, and F in the alignments shown in Fig. 2.The CD spectra and Tms were very similar to those ofEGFP, suggesting that the secondary structure contentmight be very similar in all three proteins. Indeed, whenthe CD spectra were examined using CDNN, the derivedsecondary structure contents were identical. The resultsof the epitope tag insertion scan suggest that the 22merpeptide insertion site in positions B, C, D, E, and F maybe in the solvent-exposed part of the loops, and that theseloops may be similar to those in the analogous locationsin AequoreaGFP. Combined with the observed sequence

homologies and CD results, this suggests that the overallfold of Renilla and PtilosarcusGFPs is very similar tothat of AequoreaGFP. The main difference between in-serts in Aequorealoops and potential loop sites in Renillawas in position F. In EGFP, this site is a loop betweentwo twisted b strands with a distance across the top of theloop of ca. 11 Å. An 8mer peptide inserted into AequoreaGFP at the equivalent of position F is only 0.6% as fluo-rescent as the parent GFP when expressed in yeast (Abediet al., 1998), whereas insertion of a 22mer HA tag intoposition F of RenillaGFP (this paper) retains 32% of theparent protein fluorescence. Although the exact details ofthe experiments differ, the RenillaGFP structure appearsto be significantly more tolerant than AequoreaGFP toinsertion of peptides into this particular site. Although theposition F site is likely to be surface-exposed in bothGFPs, its structure or significance in the folding pathwayof RenillaGFP may differ from that in AequoreaGFP.

To further examine the potential of positions D andF for the presentation of a functional peptide, we re-placed the HA epitope tag insert with an SV-40 derivedNLS peptide. The resulting GFPs clearly localize to thenucleus in with the insert in both positions. The insertedpeptide is thus solvent exposed and can functionally in-teract with its target in the cell. Both of these sites maythus be viable peptide library insertion sites in RenillaGFP. Thus, the use of Renilla GFP as a scaffold allowsthe use of an additional GFP peptide display site, perhapswith a different structural bias, in a surface loop (at leastin EGFP) with a distance across the tip about twice thatfor the loop-containing site D, for phenotypic screeningof the libraries.

REFERENCES

Abedi, M. R., Caponigro, G., and Kamb, A. (1998). Nucleic Acids Res.26, 623–630.

Bohm, G., Muhr, R., and Jaenicke, R. (1992). Protein Eng.5, 191–195.Colas, P., Cohen, B., Jessen, T., Grishina, I., McCoy, J., and Brent, R.

(1996). Nature380, 548–550.Cormack, B. P., Valdivia, R. H., and Falkow, S. (1996). Gene173,

33–38.Caponigro, G., Abedi, M. R., Hurlburt, A. P., Maxfield, A., Judd, W.,

and Kamb. (1998). Proc. Natl. Acad. Sci.95, 7508–7513.Cubitt, A., Heim, R., Adams, S., Boyd, A., Gross, L., and Tsien, R.

(1995). Trends Biochem. Sci.20, 448–455.

Geyer, C., Colman-Lerner, A., and Brent, R. (1999). Proc. Natl. Acad.Sci. USA96, 8567–8572.

Greenfield, N. and Fasman, G. D. (1969). Biochemistry8, 4108–4116.Gururaja, T. L., Narasimhamurthy, S., Payan, D., and Anderson, D. C.

(2000). Chem. Biol.7, 515–527.Kozak, M. (1986). Cell 44, 283–292.Lorens, J., Bennett, M., Pearsall, D., Throndset, W., Rossi, A., Arm-

strong, R., Fox, B., Chan, E., Luo, Y., Masuda, E., Ferrick, D.,Anderson, D., Payan D., and Nolan, G. (2000). Mol Ther. 1,438– 447.

Lorenz, W., McCann, R., Longiaru, M., and Cormier, M. (1991). Proc.Natl. Acad. Sci. USA88, 4438–4442.

Lowry, O., Rosenbrough, N., Farr, A., and Randall, R. (1951). J. Biol.Chem.193, 265–275.

Matz, M., Fradkov, A., Labas, Y., Savitsky, A., Zaraisky, A.,Markelov, M., and Lukyanov, S. (1999). Nature Biotechnol.17,969–973.

Moroianu, J., Blobel, G., and Radu, A. (1995). Proc. Natl. Acad. Sci.USA92, 2008–2011.

Norman, T., Smith, D., Sorger, P., Drees, B., O’Rourke, S., Hughes, T.,Roberts, C., Friend, S., Fields, S., and Murray, A. (1999). Science285, 591–595.

Ormo, M., Cubitt, A., Kallio, K., Gross, L., Tsien, R., and Remington,S. (1996). Science273, 1392–1395.

Pearson, W. and Lipman, D. (1988). Proc. Natl. Acad. Sci. USA85,2444 –2448.

Peelle, B., Lorens, J., Li, W., Bogenberger, J., Payan, D., and Ander-son, D. C. (2001). Chem. Biol.8, 521–534.

Radu, A., Blobel, G., and Moore, M. S. (1995). Proc. Natl. Acad. Sci.USA92, 1769–1773.

Rexach, M. and Blobel, G. (1995). Cell 83, 683–692.Sauder, J., Arthur, J., and Dunbrack, R. (2000). Proteins40, 6–22.Swift, S., Lorens, J., Achacoso, P., and Nolan, G. (1999). In Current

protocols in immunology(Coligan, J., Kruisbeek, A., Margulies,D., Shevach, E. and Strober, W., Eds.), Vol. 10.17C, Wiley andSons, New York, pp. 1–17.

Szent-Gyorgyi, C. and Bryan, B. (1999). Luciferases, fluorescent pro-teins, nucleic acids encoding the luciferases and fluorescent pro-teins and the use thereof in diagnostics, high throughput screen-ing and novelty items. World patent application WO9949019.

Szent-Gyorgyi, C., Ballou, B., Dagnal, E., and Bryan, B. (1999). InBiomedical imaging: Reporters, dyes and instrumentation(Born-hop, D, Contag, C. and Sevick-Muraca, E., Eds.), SPIE, Belling-ham, Washington, pp. 4 –11.

Thompson, J., Higgins, D., and Gibson, T. (1994). Nucleic Acids Res.22, 4673–4680.

Topell, S., Hennecke, J., and Glockshuber, R. (1999). FEBS Lett.457,283–289.

Tsien, R. (1998). Ann. Rev. Biochem.67, 509–544.Ward, W. and Cormier, M. (1979). J. Biol. Chem.254, 781–788.Ward, W., Cody, D., Hart, R., and Cormier, M. (1980). Photochem.

Photobiol.31, 611–615.Woody, R. W. and Dunker, A. K. (1996) In Circular dichroism and the

conformational analysis of biomolecules(Fasman, G. D., Ed.),Plenum, New York, pp. 109–157.

Yang, F., Moss, L., and Phillips, G. (1996). Nature Biotechnol.14,1246–1251.

Yang, T., Cheng, L., and Kain, S. (1996). Nucleic Acids Res.24,4592–4593.


characterization and use of green fluorescent proteins from renilla mulleri and ptilosarcus guernyi...

Documents