Engineering of technetium-99m-binding artificial receptors for imaging gene expression

THE JOURNAL OF GENE MEDICINE R E S E A R C H A R T I C L EJ Gene Med 2003; 5: 1056–1066.Published online 11 August 2003 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jgm.448

Engineering of technetium-99m-binding artificialreceptors for imaging gene expression

Maria SimonovaOlena ShtankoNikolai SergeyevRalph WeisslederAlexei Bogdanov Jr*

Center for Molecular ImagingResearch, Department of Radiology,Massachusetts General Hospital, Bldg.149, 13th St., Charlestown, MA02129, USA

*Correspondence to:Alexei Bogdanov Jr, Center forMolecular Imaging Research, Rm.5420, Massachusetts GeneralHospital, Bldg. 149, 13th Street,Charlestown, MA 02129, USA.E-mail:[email protected]

Received: 18 October 2002Revised: 23 April 2003Accepted: 30 May 2003

Abstract

Background Optimization of gene therapy protocols requires accurateand non-invasive quantification of vector delivery and gene expression.To facilitate non-invasive imaging of gene expression, we have genet-ically engineered ‘artificial receptors’, i.e. membrane proteins that bind99mTc-oxotechnetate (99mTcOT) via transchelation from a complex withglucoheptonate. The latter is a component of a widely used clinical imag-ing kit.

Methods The engineered marker proteins were designed as type I andII membrane proteins and consisted of (1) an 99mTcOT-binding domain,metallothionein (MT), and (2) a membrane-anchoring domain. Engineeredconstructs were used for transfection of COS-1 and 293 cells; the expressionof mRNA was verified by RT-PCR.

Results Immunofluorescent analysis, cell fractionation and immunoblottingrevealed expression of marker proteins on plasma membrane. Transfection ofcells resulted in strong positive staining of plasma membrane with anti-His-tagantibodies. Scintigraphic imaging in vitro confirmed the ability of transfectedcells to bind 99mTcOT. The fraction of bound radioactivity reached a peak(3.53%) when 0.93 MBq 99mTcOT was added to transfected COS-1 cells. Theexperiment-to-control signal ratio was equal to 32 at the same added dose.

Conclusions (1) Both types of engineered ‘artificial receptors’ wereexpressed on the surface of eukaryotic cells; (2) marker proteins werefunctional in binding 99mTcOT; and (3) type II membrane proteins weremore efficient in binding 99mTcOT than type I proteins. We anticipatethat the developed approach could be useful for ‘tagging’ transfected cellswith 99mTcOT enabling imaging of tracking in vivo transduced cells or celltherapies. Copyright 2003 John Wiley & Sons, Ltd.

Keywords metallothionein; osteonectin; cell-surface expression; nuclearimaging

Introduction

Non-invasive mapping of gene expression is one of the major goals ofmolecular imaging in vivo. In vivo gene expression imaging has emerged asan essential requirement for advancing the development of novel vectors,defining therapeutic efficacy and for developing new gene replacementmethods [1–7]. Several imaging approaches that can potentially be translatedinto clinical practice, in particular, radionuclide imaging techniques, havebeen suggested for gene expression imaging. Radionuclide-based gene

Copyright 2003 John Wiley & Sons, Ltd.

Artificial Receptors for Imaging Gene Expression 1057

expression imaging relies on detection of two classesof marker proteins: intracellular enzymes [1,2,8–10]or cell-surface-expressed proteins [11–15]. One of theintracellular marker enzymes, herpes simplex virus-1 thymidine kinase (HSV1-TK), has been thoroughlyinvestigated for gene therapy monitoring [9,10,16–20].This enzyme phosphorylates several 18F- or 124I-labeledprodrugs which enable the use of HSV1-TK as a markerfor PET imaging because of trapping of positron-emittingphosphorylated products in cells expressing the enzyme[2,9,17,21].

The alternative approach utilizing cell-surface expres-sion of receptor markers has an advantage that receptorligands, in contrast to intracellular enzyme substrates, donot have to penetrate plasma membranes to reach bind-ing sites. Large Gi-coupled receptors, such as dopaminetype 2 receptor D2R and human type 2 somatostatinreceptor (hSSTr2), are currently being used to detectresults of in vivo viral vector gene transfer. Receptorexpression has been imaged using 18F-labeled spiperoneand 111In-DTPA-d-Phe1-octreotide as the correspondingligands. Both receptors can modulate cAMP concentra-tion in cells upon binding the appropriate ligand. Recentachievements in decoupling of these receptors from sig-nal transduction pathways using mutagenesis will enablefurther improvement of receptor-based marker systems[22,23]. A co-expression of two imaging marker genessuch as HSV-TK and dopamine receptor D2R, as wellas hSSTr2 and HSV1-Tk, has also been proven feasible[24]. The above technique may be useful in detectingconstitutive expression of transduced genes as well astranscriptionally regulated expression.

However, there are several potential drawbacks ofnaturally occurring receptors for imaging gene expression:(1) the presence of receptor in non-transduced tissues(natural background expression), such as the expressionof somatostatin receptor in both normal pancreatictissues as well as in a large number of neuroendocrinegastroenteropancreatic tumors, while dopamine receptorhas a widespread normal expression in the brain; (2) theneed for mutagenesis to decouple receptors from signaltransduction cascades; and (3) the need for elaboratesynthesis and purification of imaging ligands. Here wereport a new model system that uses ‘artificial receptors’,normally not present on the surface of eukaryoticcells. These ‘artificial receptors’ form complexes withthe oxotechnetate cation, readily available throughthe reduction of 99mTc-pertechnetate, a commerciallyavailable radiopharmaceutical. It is well known thatcysteine-rich polypeptides establish stable bonds withtransition metals, including technetium-99m [25]. Inthis study, we compared two cysteine-rich proteinsas potential acceptors of oxotechnetate: osteonectin(secretable protein, acidic rich in cysteine, SPARC) andmetal-binding cytoplasmic metallothionein. Osteonectinis an extracellular matrix-associated glycoprotein thatregulates activation of matrix metalloproteinase-2 on thecell surface [26]. Metallothionein (MT) is a cysteine-rich, metal-binding protein with unknown function,

potentially playing an important role in intracellular redoxsignaling [27–30]. Using this model system we haveshown that MT-based artificial receptors (1) can exhibitoxotechnetate-binding properties upon expression on thecell surface, and (2) can be used as cell-surface-expressedmarkers for nuclear imaging.

Materials and methods

Expression vectors

SPARC/osteonectin marker protein was expressed asa type II membrane protein containing a hexahisti-dine (His-tag) sequence. The construct designated asPEP-SPARC contained a signal peptide/transmembranedomain derived from neprilysin (neutral endopeptidase-24.1) (neprilysin cDNA was provided by Dr. GuyBoileau, University of Montreal [31]) and a C-terminalosteonectin-containing domain (a gift from Dr. JamesBassuk, University of Washington, Seattle [26]). The pep-tidase fragment was obtained using polymerase chainreaction (PCR) with the sense primer: 5′-gcaagatttggtacc-atgggaagatcag-3′, and antisense primer: 5′-gtcgagcagc-aagctttatgcagtctgatg-3′. The PCR product was digestedwith KpnI and HindIII (here and below: restriction sitesare underlined in primer sequences). The osteonectinPCR fragment was obtained using the direct primer:5′-gggctctggcaaagcttgcccctcagcaagaagccc-3′, and reverseprimer: 5′-ccactctagacgcttccgtgatggtggtgatggtggcttcctcga-gtcacaagatccttgtcgatatcc-3′.

The reverse primer encoded a His-tag motif RGSH-HHHHHGS. The XhoI-XbaI restriction site flanked thatsequence. The obtained PCR product was digested usingHindIII-XbaI and ligated together with the PEP fragmentinto the KpnI-XbaI-digested pcDNA3 vector (Invitrogen,Carlsbad, CA, USA). All PCRs were performed usingPwo DNA polymerase (DNA amplification kit, Roche, CA,USA), and DNA sequences were confirmed by sequencing.

Metallothionein-based markers were engineered astype I and II membrane proteins. The construct designatedas MT-IgAp contained a signal peptide from human alpha-1-antitrypsin [32] (derived from pRcCMV-aHAT vector,a gift from Dr. Frank Szoka Jr., University of Califor-nia, San Francisco), fused with metallothionein (ATCC916071, NF38F08.S1, Rockville, MD, USA) and with theC-terminal membrane-spanning domain of H. influenzaeIgA protease. The pFG 26-IgA vector, a gift from Dr.Andrew G. Plaut (Tufts University-New England Medi-cal Center, Boston), contained a mutated D-epitope withthe following histidine-rich amino acid motif: RGSHHH-HHHD [33]. The DNA sequence encoding the N-terminalsignal peptide-metallothionein fragment of this constructwas obtained in two consecutive PCR reactions in the pres-ence of metallothionein cDNA and the following primers:sense 1: 5′-catcctcctgctggcaggcctagtcgacccgaactgctcctgc-accactgg-3′; sense 2: 5′-atcggtaccatgccctcttctgtctcgtggg-gcatcctcctgctggc-3′; antisense: 5′-ccactcgagcacagcagctg-cagttctccaacgtccc-3′. The PCR products were digested

Copyright 2003 John Wiley & Sons, Ltd. J Gene Med 2003; 5: 1056–1066.

1058 M. Simonova et al.

using KpnI-XhoI. Alternatively, the DNA sequence corre-sponding to the C-terminal membrane-spanning domainfragment was amplified using pFG26-IgA plasmid andthe following primers: sense: 5′-cgtctcgagaggtggaaaag-agaaatcaaactgtcgatacg-3′; antisense: 5′-ggtatctagattagaa-actaaaacttagctttaattctgc-3′. The construct designated asMT-IgAp tr had a truncated membrane-spanning domainat the autolysis site and was obtained using the abovesense and a different antisense primer: 5′-cggctcgagtctc-agagacaactgaaacagtggctg-3′. Each IgAp PCR product wasdigested with XhoI-XbaI restriction endonucleases and lig-ated with the first antitrypsin-MT PCR product togetherwith the pcDNA3 expression vector digested with KpnI-XbaI restriction endonucleases.

The vector designated as PEP-MT contained a sig-nal peptide/transmembrane domain derived from neu-tral endopeptidase and C-terminal metallothionein frag-ments. Peptidase fragment was amplified using PCRas described above for the PEP-SPARC construct. Themetallothionein PCR fragment was obtained usingsense primer: 5′-gctcgagcttgaccccaactgc-3′, and antisenseprimer: 5′-ccactctagacgcttccgtgatggtggtgatggtggcttcctcga-gcacagcagctgcagttctcc-3′. The antisense primer containedthe DNA sequence corresponding to His-tag motif RGSH-HHHHHGS.

The XhoI-XbaI restriction site flanked that sequence.The PCR product was digested with HindIII-XbaI, andligated with neutral endopeptidase PCR product intopcDNA3 digested with KpnI-XbaI. PEP-MT2 vector wasprepared by substituting four Lys to Arg using site-directed mutagenesis. Four mutations were introducedinto the metallothionein cDNA sequence using thefollowing mutagenic primer: 5′-ggctcctgcaggtgcagagag-tgcgaatgcacctcctgcgagaggagctgc-3′ and the QuickChangemutagenesis kit (Stratagene, Cedar Creek, TX, USA).Mutated nucleotides are shown in bold.

DNA sequences were verified using the services of theMGH Sequencing Core (Department of Molecular Biology,Massachusetts General Hospital).

Cell transfection

Cells were propagated in 10% FBS/DMEM (Cellgro,Washington, DC, USA). Transfection was performedat 2 × 106 cells/10-cm dish using Maxfect reagent(Molecular Research Labs, Herndon, VA, USA) asrecommended by the manufacturer in serum/antibiotic-free DMEM for 30–45 min.

RNA isolation and RT-PCR

Total RNA was extracted 72 h post-transfection. The con-trol RNA was isolated from mock-transfected cells usingthe Absolutely RNA RT-PCR miniprep Kit (Stratagene)according to the protocol provided by the manufacturer.Total RNA (200 ng) was used in RT-PCR (TitanOne tubeRT-PCR kit, Roche, Indianapolis, IN, USA). Beta-actin

mRNA levels were determined for comparative purposes.Following amplification, aliquots of the PCR productswere resolved in 1% agarose electrophoreses, and digi-tized using a Kodak digital camera model 40 (EastmanKodak, Rochester, NY, USA).

Immunofluorescent analysis oftransiently transfected cells

Cell-surface expression of marker proteins was revealedusing anti-His Tag antibody (clone 4D11; UpstateBiotechnology, Lake Placid, NY, USA) diluted at 1 : 1000or with mouse anti-metallothionein antibody (clone E9;Zymed Inc., San Francisco, CA, USA) diluted at 1 : 50.Antibodies were diluted in incubation/wash buffer (10%horse serum, 1% BSA in 1× PBS) and incubated withcells for 1 h. Following incubation, cells were rinsed withwash buffer and incubated with rhodamine-labeled goatanti-mouse antibody (Pierce, Rockford, IL, USA). Washedcells were fixed using 2% formaldehyde for 30 min,and analyzed using a Zeiss Axiovert 100TV (Wetzlar,Germany) fluorescent microscope equipped with a CCDcamera (Photometrics, Tuscon, AZ, USA).

Cell fractionation and Western blotting

HEK 293 and COS-1 cells (plated in 10-cm dishes)were transfected in triplicate. Seventy-two hours post-transfection, cells were washed with HBSS. Cells weredetached using cold 1× PBS, pelleted by centrifugationat 1000 rpm for 10 min (RT 6000B, Sorvall) andresuspended in 10 ml of cold 2 M glycerol in 50 mMTris-HCl, pH 8. The suspension was incubated at 4 ◦C for20 min. Cell pellets were then homogenized in a buffercontaining 0.25 M sucrose, 25 mM KCl, 5 mM MgCl2,20 mM Tris HCl, pH 7.8, and centrifuged at 1000 rpm for10 min (RT 6000B, Sorvall). The pellets were resuspendedin 1.5 ml of 35% OptiPrep solution (Nycomed PharmaAS, Oslo, Norway) and separated in 30–10% OptiPrepgradient at 28 500 rpm at 4 ◦C for 19 h (SW-41, BeckmanInstruments). Gradients were divided into 1-ml fractionsand alkaline phosphatase activity (a marker for plasmamembrane) was determined in each fraction as describedpreviously [34] using p-nitrophenyl phosphate (pNPP)as a substrate. Peak fractions containing enzyme activitywere collected, lysed by the addition of 50 mM Tris (pH6.8), 0.1% SDS, 0.1% Igepal, supplemented with 1 mMPMSF and Complete inhibitors (Roche). Lysates wereanalyzed using 10% SDS-PAGE and transferred onto PVDFmembranes (Bio-Rad, Hercules CA, USA), followed byblocking in 1% defatted milk (Bio-Rad), 0.1% Tween-20in PBS (pH 7.4). Washed membranes were then incubatedwith Penta-His-HRP conjugate (Qiagen Inc., CA, USA)diluted 1 : 1000 to reveal His-tagged proteins.



Glucoheptonate preparation

Glucoheptonate kits were prepared using 20 mg/ml ofsodium glucoheptonate (Sigma Chemical Co.) solution inMilliQ-grade autoclaved water saturated with nitrogen.Then, 20 µM stannous(I) chloride hexahydrate (20 µl)was added, the pH was verified (7.0), kits were packagedunder argon and stored frozen. Kits were test-labeledusing 37–185 MBq 99mTc-pertechnetate (20 min atroom temperature) and analyzed using Gelman ITLC-SG fiber plates (VWR, West Chester, PA, USA) andacetone as mobile phase. Prior to cell labeling a testexperiment was performed using 50% FBS/PBS solutionto determine whether oxotechnetate labels serum proteinsvia transchelation. 50% FBS/DMEM solution (50 µl) wasincubated with 50 µl of labeled kit for various times. Themixture was loaded on BioSpin P6 minicolumns (Bio-Rad)equilibrated with PBS and the high-molecular-weightprotein fraction was separated from the radioactivity inthe total volume. The void volume and the remainingradioactivity were separately counted using a dosecalibrator (CRC-712; Capintec, Ramsey NJ, USA).

99mTc-oxotechnetate binding totransfected cells

HEK 293 or COS-1 cells were plated in 10-cm dishes (n =3/time point), transfected and analyzed for oxotechnetatebinding 24, 48 and 72 h post-transfection. Cells were firstwashed with HBSS and fresh 10% FBS/DMEM mediumwas added 1–2 h before the experiment. Glucoheptonatekit (250 µl) was mixed with 111 MBq of 99mTc-pertechnetate (Syncor, Woburn, MA, USA), incubatedat 37 ◦C for 30 min and used within 30 min. An aliquotof 99mTc-glucoheptonate (7.4 MBq) was added to thecells and incubated for 30 min at 37 ◦C. Cells werewashed three times, lysed using 1% Igepal (Sigma)and radioactivity was determined by gamma-counting.Oxotechnetate binding was normalized by protein usingthe BCA assay (Pierce).

Scintigraphic imaging

COS-1 cells were plated in 10-cm Petri dishes at a densityof 50–60% 24 h before transfection. The cells reached adensity of 3 × 106 cells/plate (corresponding to 0.3 mgprotein/million cells) before the start of scintigraphicimaging experiments. COS-1 cells transfected withPEP-MT and PEP-MT2 constructs as well as mock-transfected cells were incubated with 7.4 MBq ofoxotechnetate per plate and washed as described above.Control and experimental plates were then subjected toscintigraphic imaging (total imaging time 2 min) using anM.CAM gamma camera equipped with a medium-energycollimator (Siemens Medical, Hoffman Estates, IL, USA).

Results

SPARC and metallothionein expressionvectors

Osteonectin/SPARC-based marker protein was designedas a type II membrane protein. Its native signal peptide(first 17 N-terminal amino acids) was replaced by a sig-nal peptide/membrane-anchoring domain (52 N-terminalamino acids) from rabbit neutral endopeptidase [31].Downstream of the PEP-SPARC fusion protein part aHis-tag sequence was inserted to enable easier immunode-tection of the recombinant protein (Figures 1A and 1B-I).Metallothionein-based type I membrane proteins wereanchored to the cell membrane using a large IgA pro-tease membrane-spanning fragment. Both MT-IgAp andMT-IgAp tr constructs had similar N-termini (15 aminoacids from anti-trypsin signal peptide [32] and MT frag-ment), and differed only in C-terminal domains. TheMT-IgAp construct had a full-size membrane-spanningdomain beginning at Ser1014 and a 25 amino-acidspacer upstream of the helper peptide. The MT-IgAptr was lacking a spacer upstream of the helper pep-tide, and began at Ser1040, the IgA protease autolysissite (Figures 1A and 1B-III). The PEP-MT construct wasexpressed as a type II membrane protein and had the sig-nal peptide/membrane-anchoring domain from neutralendopeptidase [31]. The PEP-MT2 construct containedfour lysines (20, 22, 30 and 31) in MT mutated to Argin an attempt to achieve better transfer of nascent MTpolypeptide through the endoplasmic reticulum mem-brane (Figures 1A and 1B-II). All constructs contained aHis-tag sequence to enable easier immunodetection ofrecombinant protein.

SPARC and metallothionein mRNAexpression in 293 HEK and COS-1 cells

Expression of marker protein mRNA was studied intransiently transfected cells. In a typical experiment,75% cells were transfected, as determined using greenfluorescent protein (GFP) as a transfection marker. Toanalyze steady-state levels of marker protein mRNAexpression, total RNA isolated from transfected cellswas subjected to RT-PCR analysis. Both SPARC and MTmarkers expressed as type II membrane proteins hadstable and similar mRNA steady-state levels in both celllines (Figures 2A–2D). However, as expected, in COS cellsmRNA levels were higher than in 293 HEK cells due toexpression vector replication in COS-1 cells. In 293 HEKcells, the mRNA level of type I membrane proteins wasthe lowest one (Figure 2D). Metallothionein expressedas type I membrane proteins in COS-1 cells had lowersteady-state mRNA levels than type II fusion proteins. Theresults obtained in the experiments were normalized bythose of beta-actin mRNA levels (Figures 2A–2D).



RGSHHHHHHGSV

His-tag

SPARCPEP

50 AA 286 AA

PEP

A

RGSHHHHHHGSV

PEP

His-tag60 AA

MT

50 AA

B

MTAT

23 AA 60 AA 642 AA D-epitope

IgA protease

RGSHHHHHHD

Type II protein

Type II protein

Type I protein

C

N

C

N

Type II proteinType I protein

I

II

III

Figure 1. (A) Design of metallothionein- and osteonectin/SPARC-based protein markers expressed as type I and type II membranefusion proteins. (B) Scheme of engineered vectors. I: SPARC/osteonectin type II membrane fusion protein; II: metallothionein astype II membrane fusion protein; III: metallothionein as type I membrane fusion protein

Immunofluorescent analysis ofcell-surface expression in transfectedCOS-1 cells

To test whether SPARC and MT fusion proteins wereexpressed on the cell surface, intact transfected or controlcells were incubated with antibodies recognizing His-tagor metallothionein, respectively, and then fixed with 2%formaldehyde (Figures 3A and 3B). The fusion PEP-MTwas detected on the surface using anti-MT monoclonalantibodies. Fusion proteins were organized into well-discernable sub-micron clusters (Figure 3A). Cell-surfaceexpression of PEP-SPARC showed a different patternas this fusion protein clearly did not show extensiveclustering on the cell surface (Figure 3B).

Fractionation and Western blotanalysis of MT-IgAp tr and PEP-MTfusion proteins

Expression of the MT-based constructs was studiedby transient transfections in COS-1 cells. Membranehomogenates of two constructs, MT-IgAp tr and PEP-MT, were first subjected to fractionation on a low-viscosity density gradient, followed by immunoblottingof each fraction using anti-His-tag antibodies. Accordingto alkaline phosphatase activity (a plasma membranemarker), fractions 6, 7 in MT-IgAp tr transfected cells

(Figure 4A) and fractions 6, 7, 8 in PEP-MT transfectedcells (Figure 4B) were enriched with plasma membranefragments. Electrophoresis and immunoblotting of thesefractions showed that both types of MT membraneproteins formed complexes or multimers, as previouslydescribed [35,36]. Two major bands in the 100–175kD region were identified in cells expressing MT-IgAptr protein (Figure 4A). In PEP-MT transfected cells(Figure 4B), two major protein bands of 25 and 75 kDresembled fractions present in commercially availablepurified MT (Figure 4D) [35,36].

We also compared protein expression levels in 293and COS-1 cells by Western blot analysis using PEP-MT fusion protein (Figure 4C). Densitometry of thesebands showed that relative signal intensities in COS-1cells were 2.2–2.8-times higher than those of 293 cells(see Figure 4C, legend).

99mTc-oxotechnetate binding to SPARC-and MT-expressing 293 HEK and COS-1cells

Initially, we prepared and tested glucoheptonate kits forreproducible reduction of pertechnetate in the presenceof Sn(I) and glucoheptonic acid. This was necessary toeliminate kit-to-kit variability in commercially availablepreparations. The kits gave >98% of pertechnetatereduction as determined using ITLC-G. This result was



Figure 2. RT-PCR analysis performed at the 72 h time point. (A) PEP-SPARC fusion protein. Lane M: 100-bp DNA ladder; lanes: 1, 3:mock-transfected 293 and COS-1 cells; lanes 2, 4: PEP-SPARC mRNA expression level in 293 and COS-1 cells; lanes 5, 6: beta-actinmRNA expression level in 293 and COS-1 cells. (B) RT-PCR analysis of PEP-MT and PEP-MT2 fusion proteins expressed in 293 cells.M: 100-bp DNA ladder; lane 1: mock-transfected 293 cells; lane 2: PEP-MT mRNA expression level (0.39 kb); lane 3: PEP-MT2mRNA expression level (0.39 kb); lanes 4–6: beta-actin mRNA expression levels in mock-, PEP-MT-, PEP-MT2-transfected 293 cells(0.6 kb). (C) RT-PCR analysis of PEP-MT and PEP-MT2 fusion proteins expressed in COS-1 cells. Lane M: 100-bp DNA ladder; lane 1:mock-transfected COS-1 cells; lane 2: PEP-MT mRNA expression level (0.39 kb); lane 3: PEP-MT2 mRNA expression level (0.39 kb);lanes 4–6: beta-actin mRNA expression levels in mock-, PEP-MT-, PEP-MT2-transfected COS-1 cells (0.6 kb). (D) RT-PCR analysis ofMT-IgAp and MT-IgAp tr fusion proteins expressed in 293 cells. Lane M: DNA standards (λ HindIII digest 0.5 kb; 2 kb; 2.3 kb); lane1: mock-transfected 293 cells; lane 2: MT-IgAp mRNA expression level (2.2 kb); lane 3: MT-IgAp tr mRNA expression level (2.0 kb);lanes 4–6: beta-actin mRNA expression level in mock-, MT-IgAp-, MT-IgAp tr-transfected 293 cells (0.6 kb). (E) RT-PCR analysis ofMT-IgAp and MT-IgAp tr fusion proteins expressed in COS-1 cells. Lane M: DNA standards (λ HindIII digest 0.5 kb; 2 kb; 2.3 kb);lane 1: mock-transfected COS-1 cells; lane 2: MT-IgAp mRNA expression level (2.2 kb); lane 3: MT-IgAp tr mRNA expression level(2.0 kb); lanes 4–6: beta-actin mRNA expression level in mock-, MT-IgAp-, MT-IgAp tr-transfected COS-1 cells (0.6 kb)

achieved by using an optimized kit/pertechnetate volumeratio established by using ITLC-G chromatography. Thestability of the complex was tested by incubating with50% mouse plasma. Size-exclusion chromatography onBio Spin P6 columns showed that less than 10% of99mTc-OT was associated with plasma proteins after a1-h incubation. In addition, the stability of the 99mTc-OT complex with membrane proteins was estimatedby challenging transfected cells in the presence of L-cysteine. Transfected cells were incubated for 30 min with1 mM cysteine containing isotonic saline. Radioactivitydetermined in cell supernatants after the incubationcorresponded to 15 ± 6% of the total bound 99mTc-OT.

In SPARC-expressing cells, 99mTcOT binding reachedthe highest levels at 48 h post-transfection (data notshown). SPARC-expressing cells bound only 1.5-timesmore radioactivity than mock-transfected cells, and theabsolute numbers were similar to numbers obtained forthe type I fusion proteins expressed in COS-1 cells (0.67KBq/mg protein). In MT-based fusion proteins, 99mTcOTbinding increased over time reaching the highest levelat 72 h post-transfection. The absolute counts of 99mTcradioactivity for both type I and II fusion proteins weresimilar in 293 transfected cells (1.3–1.5 KBq/mg protein).The transfected to mock-transfected cell radioactivityratio was approximately 5 : 1 (Figure 5). A maximum



A

B

C

10 µm

Figure 3. Immunofluorescent analysis. (A) PEP-MT constructexpressed in COS-1 cells; (B) PEP-SPARC construct expressedin COS-1 cells; and (C) COS-1 mock-transfected cells

radioactivity ratio of more than 7-times was observed inPEP-MT expressed in COS-1 cells. Interestingly, PEP-MTand PEP-MT2 fusion proteins expressed in COS-1 cellsshowed 99mTcOT-binding capacity four times higher thanMT-IgAp and MT-IgAp tr fusion proteins expressed in thesame COS-1 cells (2.8 KBq/mg protein compared with0.67 KBq/mg protein; Figure 5, Table 1).

Table 1. Ratios of bound 99mTcOT radioactivity for transfectedversus mock-transfected cells (normalized per mg of cell protein)

Ratios of bound 99mTcOT radioactivity/mg protein(transfected versus mock-transfected cells)

293 cells timeafter transfection

COS-1 cells timeafter transfection

Constructs 24 h 48 h 72 h 24 h 48 h 72 h

IgAp 2.8 3.1 5.9 1.4 1.8 2.3IgAp tr 2.3 2.4 5.2 1.3 1.8 2.3PEP-MT 2.3 3.1 3.1 2.9 5.2 6.2PEP-MT2 2.2 2.6 3.0 2.7 4.7 6.0PEP-SPARC ND 1.1 1.4 ND 1.4 1.4

Scintigraphic imaging

Scintigraphic imaging was performed using COS-1 cellstransfected with PEP-MT and PEP-MT2 constructs asdescribed in ‘Materials and methods’. Cells expressingmembrane-anchored MT showed a 7 : 1 experimental-to-control ratio of bound radioactivity (Figure 6A). Thecorresponding ROI measured over the area of the cell-plated dish is shown in Figure 6B. The percentage ofradioactivity bound per plate was 0.056 ± 0.007 forcontrol cells, 0.395 ± 0.151 for PEP-MT-, and 0.369 ±0.192 for PEP-MT2-transfected cells.

Optimization of the 99mTc-OT-bindingefficiency

The specific binding levels in the experiment (PEP-MTtype II proteins transfected in COS cells) significantlyexceeded the background (by 6-fold, Figure 5). Thepercentage of cell-bound radioactivity was initiallylow (0.03–0.05%). The 99mTcOT-binding yield wasincreased approximately 10-fold to 0.4% by increasingthe relative volume of the glucoheptonate kit atthe stage of pertechnetate reduction (Figure 6). Wefurther performed incubations at various doses of added99mTcOT radioactivity using mock-transfected and PEP-MT-transfected COS-1 cells. 99mTc-OT was added tomock-transfected and PEP-MT-transfected COS-1 cells ata dose ranging between 0.11 and 7.4 MBq (Table 2). Theobtained maximum binding efficiency was 3.53% at thedose of 0.93 MBq added to cells (Table 2). At this addeddose we observed an overall 100-fold increase in binding(compare with Figure 5). At this added dose PEP-MT-expressing cells bound 32-times more radioactivity thancontrol, mock-transfected cells.

Discussion

In this study we set out to investigate whether geneticallyengineered, cysteine-rich cytoplasmic (metallothionein)or secretable (osteonectin) proteins can be fused tomembrane-attachment anchors and expressed on thesurface of transfected cells. We also hypothesized that,

Table 2. Dose dependence of 99mTcOT binding to COS-1 cellsnormalized per mg of protein. Data presented as mean ± SD

Addedradioactivity, MBq

% boundradioactivity/mg

protein in PEP-MT-transfected cells

% boundradioactivity/mgprotein in mock-transfected cells

0.11 0.853 ± 0.001 0.145 ± 0.0010.22 0.651 ± 0.003 0.091 ± 0.0010.44 0.675 ± 0.003 0.135 ± 0.0010.93 3.532 ± 0.014 0.111 ± 0.0011.85 2.031 ± 0.012 0.095 ± 0.0013.70 1.015 ± 0.007 0.092 ± 0.0017.40 0.472 ± 0.001 0.085 ± 0.001



30MT-IgA

ALP

Act

ivity

(A

U)

01 121110

# Fraction98765432

10

20

30.00

ALP

Act

ivity

(A

U)

01 121110

# Fraction98765432

10.00

20.00

32

1

A

B

C D

PEP-MT

Figure 4. Fractionation and immunoblot analysis of type I and II MT-based membrane proteins (anti-His tag antibody wasused). (A) MT-IgAp tr construct expressed in COS-1 transfected cells. (B) PEP-MT construct expressed in COS-1 transfected cells.(C) Fractionation and immunoblot analysis of COS-1 and 293 cells transfected with PEP-MT construct. Relative density of bands(a) COS-1 cells: 1: 27.5; 2: 61.5; 3: 93.5 (total -182.5 AU); (b) 293 cells: 1: 11.8; 2: 27.9; 3: 33.5 (total -73.2 AU). (D) Commerciallyavailable metallothionein. Alkaline phosphatase was used as plasma membrane marker, and its activity was determined in eachfraction. Arrows show major bands of MT complexes

0

0.5

1

1.5

2

2.5

3

3.5

kB/m

g pr

otei

n

293 COS-1

Mock

PEP-MT2

PEP-MT

IgAp tr

IgAp

Figure 5. Oxotechnetate binding to MT-based fusion proteinsexpressed as type I and type II membrane proteins in293 and COS-1 cells. Cells expressing MT-IgAp; MT-IgAp trand PEP-MT; PEP-MT2 markers as well as mock-transfectedcells were analyzed for the ability of 99mTc-OT binding72 h post-transfection. One million 293 cells corresponded to0.175 mg protein; one million COS-1 cells corresponded to0.300 mg protein

when exposed to a non-oxidizing extracellular milieu,cysteine-rich proteins expressed on the cell surface partic-ipate in cysteine-metal binding clustering with radioactiveoxotechnetate, and therefore may be useful for imaging

gene expression. We specifically investigated cysteine-rich proteins as potential macromolecular acceptors of99mTcOT because of their ability to bind transitionalmetal cations, including the oxotechnetate(V) cation[37,38]. The approach suggested has several theoreticaladvantages: (1) cell-surface expression of marker proteinseliminates the need for the imaging drug to penetrate intocells, thus circumventing an additional delivery barrier;(2) the corresponding cDNA insert is small (0.35–2 kb)and can therefore be used in almost any expression system(viral or non-viral); (3) cell-surface expression of proteinsmay be less toxic to the cell than intracellular protein over-expression; and (4) nuclear imaging systems are widelyavailable and 99mTc pertechnetate is a commonly used,inexpensive radioisotope.

The modification and targeting of cysteine-rich proteinexpression to the cell surface are challenging tasks becauseof potential protein aggregation and resultant retentionin the endoplasmic reticulum [39,40]. However, priorevidence suggested that MT, fused with the C-terminalanchoring domain of bacterial IgA protease, can betransported to the cell surface of eukaryotic cells asa functional type I membrane protein. This membraneMT expressed in plants significantly decreased the toxiceffects of heavy metals scavenged from soil [41]. We alsoused an alternative design (type II membrane protein)



50

0

10

Mock PEP-MT2PEP-MT

20

30

kBq

/ p

late

40

Figure 6. Scintigraphic imaging of cell culture plates (A), and ROI analysis (B) performed with PEP-MT and PEP-MT-2 constructsand control mock-transfected COS-1 cells using a M.CAM gamma camera (Siemens Medical Systems). Each plate containedapproximately 3 × 106 cells (1 mg of protein)

by fusing the N-terminal signal peptide/transmembranedomain derived from the neutral endopeptidase-24.11(neprilysin [31]) with cysteine-rich proteins. Previously,we had successfully used the latter design to target theexpression of green fluorescent protein to the cell surface[42].

In this study we tested five engineered fusion proteinswith 99mTc-oxotechnetate-binding properties (Figures 1Aand 1B). First, osteonectin (SPARC), a cysteine-richprotein, was converted into a type II membrane protein(with cytoplasmic N-terminus) and fused with His-tagpositioned at the extracellular C-terminus of SPARC.After transfecting cells with this fusion protein, highlevels of mRNA were expressed (Figure 2A) and theprotein was clearly targeted to the surface of eukaryoticplasma membranes, as demonstrated by staining cell-associated protein using antibodies specific for His-tag(Figure 3B). However, the OT-binding study revealedthat the PEP-SPARC fusion protein did not exhibit anysignificant OT binding (data not shown). This resultsuggested that the presence of cysteine-rich motifs alonewas not sufficient for 99mTcOT binding, presumablybecause osteonectin cysteines were involved in stableS–S bond formation. We further investigated expressionof metallothionein, an alternative cysteine-rich metal-binding protein, a natural metal scavenger [27,41]. Intotal, four MT-based type I and II membrane proteinswere engineered and tested in vitro. Metallothionein isan intracellular protein, and the presence of positivelycharged lysines in its structure was initially perceived byus as a potential stop-transfer signal [43,44] that mightcomplicate cell-surface expression. Thus, we prepared thePEP-MT2 mutant containing arginine residues replacingfour lysines in an attempt to facilitate a bettertransfer of nascent polypeptide through the endoplasmicreticulum membrane. The MT-IgAp and MT-IgAp trvariants were engineered to verify whether a mutantcontaining deletion of the first two autolysis sites in theIgA protease membrane-spanning domain can influence

protein expression. We found that there was no significantdifference in protein expression and OT-binding amongthe pairs of type I and II membrane proteins. However, MTmarkers expressed as MT type I membrane proteins hadlower mRNA steady-state levels in both cell lines than MTtype II membrane proteins (Figures 2B–2D), especially ifexpressed in 293 cells (Figure 2D). Western blot analysis(Figures 4A and 4B) showed that both type I and typeII membrane-anchored variants of MT were localizedin the membrane fractions, as expected (Figure 4; oneof each type of fusion proteins is shown on Westernblots). According to data described in relevant literature[35,36] and our Western blot analysis, both types ofMT membrane proteins formed complexes, similarly tothe commercially available purified MT (Figure 4D). TheOT-binding study suggested that the presence of cysteine-rich motifs was not sufficient for OT binding, and onlymetallothionein-based fusion proteins demonstrated OT-binding properties (Figure 5, Table 1). Technetium-99mbinding to type I and II proteins was similar in 293 cells.The highest levels of 99mTc-OT binding (more than 6-times compared with control cells) were observed in PEP-MT and PEP-MT2 expressed in COS cells. Interestingly, inCOS-1 cells, PEP-MT and PEP-MT2 fusion proteins showedbinding capacity that was four times higher than MT-IgApand MT-IgAp tr variants expressed in the same cell line(Figure 5). This result could be explained by differentprocessing of these proteins. The ability to bind 99mTc-OTby PEP-MT or PEP-MT2 was 1.5-times higher in COS-1transfected cells (P < 0.01) than in transfected 293 cells,presumably as a consequence of vector DNA replicationin COS-1 cells and not in 293 cells. This hypothesiswas verified by Western blot analysis (Figure 4C). Basedon densitometry of immunoblots, recombinant membraneprotein expression in COS-1 cells was 2.5-fold higher thanin 293 cells. The differences in the COS-1/293 ratios ofcell-surface-expressed metallothionein levels (2.5-times)and the ratio reflecting differences in 99mTc-OT binding(1.5-times) may be explained by the observed formation



of protein complexes (Figures 4A and 4B). Complexformation and clustering of membrane proteins couldpotentially lead to the screening of MT-binding sites andto a certain decrease in ligand-binding efficacy.

In our first experiment, the transfected-to-control signalratio was more than a factor of 6 (Figure 5); however, thepercentage of bound-to-cells radioactivity was not suffi-cient (0.03–0.05%). To increase the binding efficiency wefirst attempted to shift the pertechnetate-oxotechnetateequilibrium in the direction of oxotechnetate formation.As a result of this optimization, scintigraphic imaging per-formed with PEP-MT and PEP-MT2 constructs expressedin COS-1 cells showed an increase in 99mTc-OT bindingby an order of magnitude (0.4%, Figures 6A and 6B).The transfected-to-control signal ratio was similarly highat 6 : 1. While the obtained binding efficiency was suffi-cient in our in vitro experiments, for in vivo experimentsa higher bound radioactivity fraction is required. Thus,we further maximized the bound radioactivity fraction byoptimizing the added 99mTc-OT dose. At a dose of 0.93MBq, the binding fraction peaked at 3.53% (Table 2),which constitutes a 100-fold increase in binding capacityif compared with our initial results. The experiment-to-control signal ratio reached 32 at the same added dose.We expect that the binding efficiency and experiment-to-background ratio demonstrated here will enable us todifferentiate between normal and transfected cells in vivo.

In conclusion, our results suggest that (1) both typeI and II proteins of engineered ‘artificial receptors’ wereexpressed on the cell surface; (2) the presence of cysteine-rich motifs in protein structures is necessary but notsufficient for OT binding; (3) of the proteins studied, onlyMT-based (metal-scavenger protein) OT-binding fusionproteins were functional in eukaryotic cells; and (4) typeII membrane proteins were more efficient in binding99mTcOT than type I proteins. We expect the proposedapproach to be useful for the further development of‘tagging’ of transfected cells with 99mTc-OT for monitoringcell-based therapies, i.e. early events of cell biodistributionafter local or systemic administration as well as forscintigraphic assessment of gene expression in vivo.

Acknowledgements

The authors thank Dr. Guy Boileau (University of Montreal)for a neutral endopeptidase-24.1/neprilysin plasmid, Dr. JamesBassuk (University of Washington, Seattle) for a gift ofosteonectin/SPARC cDNA, Dr. Frank Szoka, Jr. (Universityof California, San Francisco) for an antitrypsin plasmid,Dr. Andrew G. Plaut (Tufts University-New England MedicalCenter, Boston) for sharing H. influenzae IgA protease plasmidand helpful suggestions. The authors are grateful to Dr. XandraBreakefield for stimulating discussions. This study was supportedby NIH grants PO1 CA 69246 and R21 CA85657.

References1. Tjuvajev JG, Stockhammer G, Desai R, et al. Imaging the

expression of transfected genes in vivo. Cancer Res 1995; 55:6126–6132.

2. Wiebe L, Morin K, Knaus E. Radiopharmaceuticals to monitorgene transfer. Q J Nucl Med 1997; 41: 79–89.

3. Larson SM, Tjuvajev J, Blasberg R. Triumph over mischance: arole for nuclear medicine in gene therapy. J Nucl Med 1997; 38:1230–1233.

4. Service R. Molecular imaging: new probes open windows ongene expression, and more. Science 1998; 280: 1010–1011.

5. Bogdanov AJ, Weissleder R. The development of in vivo imagingsystems to study gene expression. Trends Biotechnol 1998; 16:5–10.

6. Fehse B, Li Z, Schade U, Uhde A, Zander A. Impact of a newgeneration of gene transfer markers on gene therapy. Gene Ther1988; 5: 429–430.

7. Thomas D, Lythgoe M, Gadian D, et al. Rapid simultaneousmapping of T2 and T2* by multiple acquisition of spinand gradient echoes using interleaved echo planar imaging(MASAGE-IEPI). Neuroimage 2002; 6: 992–1002.

8. Tjuvajev JG, Finn R, Watanabe K, et al. Noninvasive imaging ofherpes virus thymidine kinase gene transfer and expression: apotential method for monitoring clinical gene therapy. CancerRes 1996; 56: 4087–4095.

9. Gambhir S, Barrio J, Phelps M, et al. Imaging adenoviral-directed reporter gene expression in living animals with positronemission tomography. Proc Natl Acad Sci U S A 1999; 96: 5.

10. Tjuvajev JG, Avril N, Oku T, et al. Imaging herpes virusthymidine kinase gene transfer and expression by positronemission tomography. Cancer Res 1998; 58: 4333–4341.

11. Bogdanov AJ, Simonova M, Weissleder R. Towards in vivoimaging of gene expression: the engineering of metal-bindingsites. In Advances in Gene Technology: Biomolecular Design,Form and Function, Ahmad F (ed). Oxford University Press:Ft. Lauderdale, FL: IRL 1997; 22.

12. Bogdanov A Jr, Wright SC, Marecos EM, et al. A long-circulatingco-polymer in ‘‘passive targeting’’ to solid tumors. J DrugTargeting 1997; 4: 321–330.

13. Rogers BE, Rosenfeld ME, Khazaeli MB, et al. Localization ofiodine-125-mIP-Des-Met14-bombesin (7–13) NH2 in ovariancarcinoma induced to express the gastrin releasing peptidereceptor by adenoviral vector-mediated gene transfer. J NuclMed 1997; 38: 1221–1229.

14. Zinn K, Buchsbaum D, Mountz J, et al. Application of type 2human somatostatin receptor (hSSTr2) as a reporter for imaginggene transfer. Sixth Int Conf Anticancer Research. Kallithea,Halkidiki, Greece, 1998; 18: 4995 (#382).

15. Zinn K, Buchsbaum D, Chaudhuri T, et al. Noninvasivemonitoring of gene transfer using a reporter imaged with ahigh-affinity peptide radiolabeled with 99mTc or 188Re. J NuclMed 2000; 41: 887–895.

16. Haberkorn U, Altmann A, Morr I, et al. Monitoring gene therapywith herpes simplex virus thymidine kinase in hepatoma cells:uptake of specific substrates. J Nucl Med 1997; 38: 287–294.

17. Gambhir SS, Barrio JR, Wu L, et al. Imaging of adenoviral-directed herpes simplex virus type 1 thymidine kinase reportergene expression in mice with radiolabeled ganciclovir. J NuclMed 1998; 39: 2003–2011.

18. Luker G, Sharma V, Pica C, et al. Noninvasive imaging ofprotein-protein interactions in living animals. Proc Natl AcadSci U S A 2002; 99: 6961–6966.

19. Dubey PSH, Adonai NDS, Rosato A, et al. Quantitative imagingof the T cell antitumor response by positron-emissiontomography. Proc Natl Acad Sci U S A 2003; 100: 1232–1237.

20. Koehne G, Doubrovin M, Doubrovina E, et al. Serial in vivoimaging of the targeted migration of human HSV-TK-transduced antigen-specific lymphocytes. Nat Biotechnol 2003;21: 405–413.

21. Gambhir SS, Barrio JR, Herschman HR, Phelps ME. Assays fornoninvasive imaging of reporter gene expression. Nucl Med Biol1999; 26: 481–490.

22. MacLaren DC, Gambhir SS, Satyamurthy N, et al. Repetitive,non-invasive imaging of the dopamine D2 receptor as a reportergene in living animals. Gene Ther 1999; 6: 785–791.

23. Rogers BE, McLean SF, Kirkman RL, et al. In vivo localizationof [111In]-DTPA-D-Phe-octreotide to human ovarian tumorxenografts induced to express the somatostatin receptor subtype2 using an adenoviral vector. Clin Cancer Res 1999; 5: 383–393.

24. Zinn KR, Chaudhuri TR, Krasnykhin VN, et al. Gamma cameradual imaging with somatostatin receptor and thymidine kinaseafter gene transfer with a bicistronic adenovirus. Radiology 2002;223: 417–425.



25. Nordberg M. Metallothioneins: historic review and state ofknowledge. Talanta 1998; 46: 243–254.

26. Gille C, Bassuk J, Pulyaeva H, et al. SPARC/osteonectin inducesmatrix metalloproteinase 2 activation in human breast cancercell lines. Cancer Res 1998; 58: 5529–5536.

27. Klaassen C, Liu J, Choudhuri SC. Metallothionein: anintracellular protein to protect against cadmium toxicity. AnnuRev Pharmacol Toxicol 1999; 39: 267–294.

28. Maret W, Vallee B. Thiolate ligands in metallothionein conferredox activity on zinc clusters. Proc Natl Acad Sci U S A 1998;95: 3478–3482.

29. Jiang L, Maret W, Vallee B. The ATP-metallothionein complex.Proc Natl Acad Sci U S A 1998; 95: 146–149.

30. Jacob C, Maret W, Vallee B. Selenium redox biochemistry ofzinc-sulfur coordination sites in proteins and enzymes. Proc NatlAcad Sci U S A 1999; 96: 1910–1914.

31. Roy P, Chatellard C, Lemay G, et al. Transformation of the signalpeptide membrane anchor domain of a type II transmembraneprotein into a cleavable signal peptide. J Biol Chem 1993; 268:2699–2704.

32. Meyer K, Thompson M, Levy M, et al. Intracheal gene delivery tothe mouse airway: characterization of plasmid DNA expressionand pharmacokinetics. Gene Ther 1995; 2: 450–460.

33. Plaut A, Qiu J, Geme JI. Human lactoferrin proteolytic activity:analysis of the cleaved region in the IgA protease of Haemophilusinfluenzae. Vaccine 2001; 19: S148–S152.

34. Walter K, Schutt C. Methods of Enzymatic Analysis. EnzymaticAssay of Alkaline Phosphatase (EC 3.1.3.1) (2nd edn). vol. II.Academic Press: New York, 1974.

35. Aoki Y, Suzuki KT. Methods Enzymol 1991; 205: 108–114.

36. Jasani B, Elmes ME. Methods Enzymol 1991; 205:95–107.

37. Xue LY, Noujaim AA, Sykes TR, Woo TK, Wang XB. Role oftranschelation in the uptake of 99mTc-MAb in liver and kidney.Q J Nucl Med 1997; 41: 10–17.

38. Pietersz GA, Patrick MR, Chester KA. Preclinical characteriza-tion and in vivo imaging studies of an engineered recombi-nant technetium-99m-labeled metallothionein-containing anti-carcinoembryonic antigen single-chain. J Nucl Med 1998; 39:47–56.

39. Reddy PS, Corley RB. Assembly, sorting, and exit of oligomericproteins from the endoplasmic reticulum. Bioessays 1998; 20:546–554.

40. Maggi D, Cordera R. Cys 786 and Cys 776 in theposttranslational processing of the insulin and IGF-I receptors.Biochem Biophys Res Commun 2001; 280: 836–841.

41. Valls M, Atrian S, de Lorenzo V, Fernandez L. Engineering amouse metallothionein on the cell surface of Ralstonia eutrophaCH34 for immobilization of heavy metals in soil. Nat Biotechnol2000; 18: 661–665.

42. Simonova M, Weissleder R, Sergeyev N, et al. Targeting of greenfluorescence protein expression to the cell surface. BiochemBiophys Res Commun 1999; 262: 638–642.

43. Fu J, Kreibich G. Retention of subunits of the oligosaccharyl-transferase complex in the endoplasmic reticulum. J Biol Chem2000; 275: 3984–3990.

44. Suokas M, Myllyla R, Kellokumpu S. A single C-terminal peptidesegment mediates both membrane association and localizationof lysyl hydroxylase in the endoplasmic reticulum. J Biol Chem2000; 275: 17 863–17 868.


Engineering of technetium-99m-binding artificial receptors for imaging gene expression

Documents

Transcript of Engineering of technetium-99m-binding artificial receptors for imaging gene expression