Characterization of Coelenterazine Analogs for Measurements ofRenilla Luciferase Activity in Live Cells and Living Animals

Hui Zhao1 , Timothy C. Doyle1, Ronald J. Wong1, Yuan Cao1, David K. Stevenson1,David Piwnica-Worms2, and Christopher H. Contag1

1Stanford University School of Medicine, and 2Mallinckrodt Institute of Radiology and WashingtonUniversity School of Medicine

AbstractIn vivo imaging of bioluminescent reporters relies on expression

of light-emitting enzymes, luciferases, and delivery of chemical

substrates to expressing cells. Coelenterazine (CLZN) is the

substrate for a group of bioluminescent enzymes obtained from

marine organisms. At present, there are more than 10 commer-

cially available CLZN analogs. To determine which analog is

most suitable for activity measurements in live cells and living

animals, we characterized 10 CLZN analogs using Renilla

luciferase (Rluc) as the reporter enzyme. For each analog, we

monitored enzyme activity, auto-oxidation, and efficiency of

cellular uptake. All CLZN analogs tested showed higher auto-

oxidation signals in serum than was observed in phosphate

buffer or medium, mainly as a result of auto-oxidation by

binding to albumin. CLZN-f, -h, and -e analogs showed 4- to 8-

fold greater Rluc activity, relative to CLZN-native, in cells

expressing the enzyme from a stable integrant. In studies using

living mice expressing Rluc in hepatocytes, administration of

CLZN-e and -native produced the highest signal. Furthermore,

distinct temporal differences in signal for each analog were

revealed following intravenous or intraperitoneal delivery. We

conclude that the CLZN analogs that are presently available vary

with respect to hRluc utilization in culture and in vivo, and that

the effective use of CLZN-utilizing enzymes in living animals

depends on the selection of an appropriate substrate. Mol

Imaging (2004) 3, 43 –54.

Keywords: Renilla luciferase, coelenterazine, in vivo bioluminescence imaging, auto-oxidation,

biophotonic imaging.

IntroductionThe use of in vivo bioluminescence imaging (BLI) forreal-time measurements of biological function in livingcells and animals has been described in many areas ofresearch, including studies of infectious disease (viral,bacterial, and fungal), cancer, gene therapy, drug devel-opment, immunology, and stem cells [1–8]. Three lucif-erase proteins, isolated from bacteria (Lux), firefly(Photinus pyralis, Fluc), and sea pansy (Renilla renifor-mis, Rluc), have been used as reporters in cell-basedassays and, more recently, have been employed for invivo studies. Each of these enzymes requires a different

chemical substrate for light production, and only for Luxhave the enzymes for substrate biosynthesis been coex-pressed in labeled cells. Therefore, for FLuc and RLuc,which have been used to label mammalian cells andfungi, the substrates need to be administered exoge-nously [1,9]. The use of different substrates for theseluciferases allows the activities to be separated biochem-ically, providing the opportunity to look at separateprocesses in the same animal or group of animals[10,11]. FLuc and its substrate, D-Luciferin, constitute awell-established reporter system used for BLI. However,the properties of Rluc and its substrate, coelenterazine(CLZN, Figure 1), are yet to be fully characterized in vivo.

Rluc catalyzes the oxidative decarboxylation of CLZN,producing blue-green light with a spectral peak of 480nm. The reaction requires only O2 [12,13], unlike Fluc,which also requires ATP as an energy source and Mg2+.Rluc has been successfully expressed in mammalian cellswithout any apparent signs of cytotoxicity. A modifiedrluc gene has been optimized for expression in mam-malian cells and is commercially available (hrluc; Prom-ega, Madison, WI). Optimization of the hrluc codingsequence included the use of codons that are preferen-tially used by mammals, and by eliminating any potentialbinding sites for mammalian transcription factors. Thesemodifications have resulted in significantly higher levelsof transcription in mammalian cells in culture. Rluc is auseful reporter protein for cell culture assays and haspotential for in vivo measurements due to its uniquesubstrate that allows for differentiation from Fluc activityas well as its lack of cofactor requirements, which offersversatility in assay development [14].

© 2004 Massachusetts Institute of Technology.

Abbreviations: hRluc, humanized Renilla luciferase protein; hrluc, the gene for humanized

Renilla luciferase; CLZN, coelenterazine; BLI, in vivo bioluminescence imaging; BW, body

weight; ROI, region of interest.

Corresponding author: Christopher H. Contag, PhD, Clark Center, Bio-X Program, 318 Campus

Drive, Room E-150, Stanford, CA 94305-5427, USA; e-mail: ccontag@stanford.edu.

Received 13 November 2003; Accepted 6 February 2004.

RESEARCH ARTICLE Molecular Imaging . Vol. 3, No. 1, January 2004, pp. 43– 54 43

In cell culture, constitutive expression of RLuc hasbeen used as an internal control for expression studieswhere Fluc is used to report regulated expressionpatterns [15,16]. Rluc has also been used as the biolu-minescent donor in bioluminescence resonance energytransfer (BRET, also known as chemiluminescent andluminescent resonance energy transfer, CRET and LRET,respectively) studies of protein–protein interactionsand shift wavelength of emission [17 –19]. In vivomeasurements of Rluc activity were first demonstratedusing tumor cells implanted subcutaneously with CLZNbeing injected intravenously prior to imaging. In thisstudy, Fluc activity could be distinguished from Rluc inthe same host animal by sequential injection of theappropriate substrates followed by imaging with asensitive cooled charge-coupled device (CCD) camera[14]. Subsequently, hRluc has been used in living sub-jects as a split reporter gene for measuring protein–protein interactions [20].

Rluc is one of several luciferases from marine organ-isms that utilize CLZN to produce luminescence. Ae-quorin, the luciferase from the jellyfish Aequorea,requires the presence of calcium ions to catalyze theCLZN oxidation in the photoprotein complex, whichhas allowed the use of aequorin–CLZN as an extremelysensitive method to monitor intracellular free calciumlevels [21,22]. CLZN is able to penetrate the mammaliancell membrane and diffuse throughout the cytosol[23,24], although the mechanism of membrane trans-port is not clear. Toxic effects of CLZN in cells has notbeen demonstrated; it is likely, however, that there are

effects on cell physiology given its properties as anantioxident [25]. Dubuisson et al. [25] have reportedthat native CLZN and synthetic methyl-CLZN (CLZN-m)could protect cells from oxidative stress. Moreover, thecytotoxicity of CLZN, but not CLZN-m, was observed atconcentrations more than 2 £ 10-5 M (equal to 8.5 g/mL) in rat hepatocytes [25].

The efficiency of light emission in the Rluc–CLZNsystem varies depending on which CLZN analog is used(Figure 1). Shimomura et al. [26] have synthesized morethan 25 CLZN analogs and examined the specificity andlight intensity of each analog using recombinant Rlucprotein. Only two analogs (-e and -v) had higher effi-ciencies than that of the native substrate. CLZN-e and -vproduced luminescence at intensities 7.5 and 6.4 timeshigher than that from native CLZN. CLZN-e showed abimodal spectrum with peaks at 418 and 475 nm. CLZN-v showed a red-shifted spectrum with a peak at 512 nm,due to its additional conjugated double bond [27].Longer wavelengths of emission are of interest for BLIbecause red light is not absorbed by mammalian tissuesto the extent of shorter wavelengths (blue or green).There is a dramatic increase in light transmission throughmammalian tissues above 600 nm given that blue-green light is strongly absorbed by hemoglobin [28].

In this study, we characterized 10 different CLZNanalogs for their use in BLI for both cell culture and invivo studies. Different media, routinely used to propa-gate mammalian cells, demonstrated different levels ofauto-oxidization of CLZN analogs. In addition, we foundthat light output was different for different analogs inintact cells, cell lysates, and animals. In living animals,the route of CLZN administration, interperitoneal orintravenous, leads to different kinetics of light output.The reaction conditions used in live cell assays andanimals to assess Rluc–CLZN bioluminescence deter-mine the background levels and the signal intensities;therefore, optimization in a given model is necessary.

Materials and MethodsCLZN Analogs and Luciferase Expression Plasmids

Native CLZN and its semisynthetic analogs desig-nated, -f, -h, -hcp, -fcp, -cp, -n, -i, -ip, and -e (Biotium,Hayward, CA; Figure 1) were dissolved in methanol to aconcentration of 1 mg/mL for use as stock solutions,stored at ¡70 C, and diluted in phosphate buffer (with-out calcium and magnesium) immediately prior to use. Aplasmid containing the humanized RLuc DNA (phRluc-CMV) was purchased from Promega. The Fluc expressionconstruct, pcDNA-fLuc, was constructed by inserting theFluc coding sequence (BglII/BamHI fragment of pGL3;

Figure 1. Structures of the CLZN analogs. The 10 analogs used in the study have

different substitutions at two positions, R1 and R2. The core structure is shown

on the top left, with the R groups listed in the table. CLZN-e differs in structure

from the other analogs by the presence of an extra ring in the core structure

and is shown on the top right.

44 Characterization of CLZN Analogs Zhao et al.

Molecular Imaging . Vol. 3, No. 1, January 2004

Promega) into the BamHI sites of pcDNA 3.1(¡) (Invi-trogen, Carlsbad, CA).

Generation of hRluc Reporter Stable Cell LinesHeLa (human cervical carcinoma) and C6 (rat glio-

ma) cells were obtained from American Type CultureCollection (ATCC, Manassas, VA) and maintained ac-cording to ATCC instructions. To generate a plasmidwhere Rluc was expressed similarly to Fluc, a DNAfragment containing the coding sequence for hRlucwas cut out of the vector phRluc-CMV using therestriction enzymes, NheI and XbaI, and cloned intoNheI/XbaI-digested pCDNA3.1(¡) (Invitrogen). Theresulting plasmid, containing the neomycin resistancegene (neor) and hRluc behind the immediate earlypromoter from human cytomegalovirus (CMV), wastransfected into cells using Lipofectamine 2000 (Invi-trogen). Positive clones were selected in Geneticin(100 g/mL) and identified by light emission as de-tected by the IVIS Model 100 (Xenogen, Alameda, CA).

Low-light imaging was performed using the IVIS,which consists of a cooled integrating CCD cameramounted on a light-tight specimen chamber (darkbox), a camera controller, a camera cooling system,and controlled using a computer. Grayscale referenceimages were obtained with low-light illumination andbioluminescent output was then acquired in completedarkness. The data are represented as pseudocolorimages of light intensity (blue—least intense, red—mostintense) superimposed over the grayscale referenceimages. Light output from specified regions of interest(ROIs) were quantified as the total number of photonsemitted per second using LivingImage (v2.12, Xenogen)and Igor imaging analysis software (v.4.02A, WaveMet-rics, Lake Oswego, OR).

Cultures of positive clones were expanded over aperiod of 7 to 15 days. For cryostorage, the cells wereprepared according to standard protocols of being slowcooled in freezing medium at ¡70 C for overnight, andthen were maintained in liquid nitrogen.

Measurement of Auto-Oxidation of CLZN AnalogsCLZN analogs (final concentration of 2.4 M) were

added into 50 L of different growth media in theabsence of cells onto wells of 96-well black microtiterplates (Greiner Bio-one, Germany) to prevent lightreflection from well to well. Luminescent signals weremeasured immediately after substrate addition and mon-itored over a period of 30 min using the IVIS or a singletube luminometer.

Measurements of Rluc Activity in Live Cell and CellLysate Assays

In live cell assays, HeLa and C6 cell lines, consti-tutively expressing Rluc, were grown in 96-well plateswith growth medium [Dulbecco’s Modified Eagle’sMedium (DMEM) containing 10% fetal bovine serum(FBS)]. Cells not expressing Rluc were used as con-trols. At 80% to 90% confluency (¹3 £ 104 cells/well),growth medium was replaced with phosphate-buff-ered saline (PBS) containing one of the CLZN analogs(to a final concentration of 2.4 M). Bioluminescentsignals were then immediately determined using theIVIS.

Cell lysates were prepared fresh for each Rluc activitymeasurements. Cells (1 £ 106) expressing Rluc werelysed in 1 mL of Passive Lysis Buffer (PLB, Promega) for30 min at 37 C as recommended by the manufacturer.Twenty microliters of diluted lysate (1:100) were mixedwith 100 L of CLZN solution (final concentration of1 M) and the light intensity was measured every 10 secfor 2 min in a single tube luminometer or every 5 minfor 30 min using the IVIS.

Luminometry was performed in a Sirius single tubeluminometer (Berthold Detection Systems, Germany).Reaction conditions were determined by testing serialdilutions of lysate and various concentrations of CLZN.Twenty microliters of lysate were mixed with 100 L of1 M CLZN and resulting light output was measuredevery 10 sec for 2 min and expressed as relative lightunits (RLU) per cell.

In Vivo Liver TransfectionFVB mice (6- to 8-week-old males) were obtained

from Charles River Laboratories (Wilmington, MA) andmaintained under strict adherence to Stanford Universi-ty institutional guidelines. In vivo transfection of cells inmouse livers were performed using the hydrodynamicmethod as described previously [29,30]. Briefly, theplasmid DNA was diluted in phosphate buffer (withoutMg2+, Ca2+) in a volume of 1 mL/10 g body weight (BW)and was rapidly (within 10 sec) injected into the tailveins of mice using a 3-mL syringe fitted with a 27-gaugeneedle. The amount of DNA injected ranged from 0.1 to5 g per mouse.

For measurements of light output from both report-ers, a mixture of the two plasmids containing 1 g eachof Rluc and Fluc vectors was injected into mice via thetail vein. Twenty-four hours posttransfection, CLZN-native (1 mg/kg BW) was administrated intravenouslyand Rluc signals were measured 1 min after injection byBLI. Two to four hours later, when the Rluc signal

Characterization of CLZN Analogs Zhao et al. 45

Molecular Imaging . Vol. 3, No. 1, January 2004

diminished to background levels, D-Luciferin (150 mg/kgBW) was injected intraperitoneally into the mice andFluc signal was determined after 5 min.

BLI of Rluc Activity in MiceTwenty-two to twenty-four hours post liver transfec-

tion, mice were anesthetized by intraperitoneal injec-tions of 100 to 150 L 2,2,2-tribromoethanol/10 g BW(Avertin, 25 mg/mL, Aldrich Chemical, Milwaukee, WI).CLZN analogs were then administrated into anesthe-tized mice, intraperitoneally or intravenously, at a doseof 1 mg/kg BW just prior to BLI [2]. Light output wasacquired using the IVIS for 1 to 60 sec depending onsignal intensity. Following imaging, animals were keptwarm and allowed to recover. Light intensity was deter-mined from an ROI corresponding to the liver region ina ventral view of the animals and expressed as photons/sec/ROI.

Measurements of Renilla Luciferase Activity in LiverHomogenates

Liver tissue from mice that had been transfected usingthe phRluc-CMV plasmid, and from normal controls werecollected and homogenized in PLB (0.2 g liver/mL) usingTissue Tearor (Model 985-370, Biospec Product, Bartles-ville, OK). The tissue lysates were kept on ice for no morethan 2 hr prior to measurements. Rluc activity wasdetermined from a mixture of 20 L diluted lysate(1:1000) and 100 L of 1 M CLZN using either a singletube luminometer or using the IVIS and was expressedas RLU/g liver tissue or photons/sec/g liver.

Statistical AnalysisAnalysis of variance (ANOVA) was performed to de-

termine the statistical significance. A p value of less than.05 was considered to be significant. Slopes of lightintensity kinetics were calculated by linear regression.

Figure 2. Comparison of autoluminescence intensity from CLZN analogs in different media. Light emission (photons/sec/50 L of medium) was measured after the

addition of each CLZN analog (-native, -h, -f, -cp, -n, -hcp, -e) to each medium. Integration times were 1 min. (A) The autoluminescence from FBS, MS, DMEM, PBS, and

PLB are shown. (B) Measurements of autoluminescence from CLZN analogs in PBS or in media containing 10% FBS or 1% albumin were determined. All values are

shown as mean ± SD (n = 4).

46 Characterization of CLZN Analogs Zhao et al.

Molecular Imaging . Vol. 3, No. 1, January 2004

ResultsAuto-Oxidation of CLZN Analogs in Different Media

CLZNs contain an imidazopyrazine structure and aretherefore unstable and prone to oxidation, especiallyin solution [26]. If significant amounts of CLZN wereauto-oxidized (enzyme-independent oxidation) in cellculture medium or in the circulation and tissues ofmice, then less CLZN substrate would be available tothe luciferase-expressing target cells. More importantly,the emitted auto-luminescence would interfere withRluc luminescent signal. Therefore, we evaluated thelevels of auto-luminescence for each substrate underconditions that would be used in live cell assays and inliving animals.

When auto-luminescence was measured from thevarious CLZN analogs, we found that the presence ofserum from different species in the assay medium re-sulted in a significant increase in luminescent signal, withthe highest signals obtained in the presence of FBS(Figure 2A and Table 1). The signals were 30- to 100-foldhigher than those obtained in wells containing PBS orDMEM in the absence of serum ( p = .001). The recom-mended luciferase assay buffer, PLB (1£, Promega),produced background autoluminescence that was equiv-alent to that observed in the presence of mouse serum(MS). Therefore, for all subsequent measurements ofRluc activity in live cells, growth medium was replacedwith either PBS or DMEM in the absence of any sera.

In addition, among the analogs tested, CLZN-e hadthe greatest signals due to auto-oxidation, producingup to 1 £ 107 photons/sec/well in FBS and 2.6 £ 105

photons/sec/well in PBS without serum. The intensities

were approximately 24- and 14-fold higher than that ofCLZN-native in serum and PBS, respectively (Figure 2A).

Because many components in the serum could acti-vate the auto-oxidation of CLZN, we considered albuminas a prime candidate and assayed light production inmedia containing albumin. Autoluminescence in mediacontaining 1% albumin was measured and comparedwith that from CLZN in PBS containing 10% FBS(which contains an equivalent amount of albumin).Similar intensities were measured from both solutions(Figure 2B), suggesting that albumin alone can auto-oxidize CLZN and may be the primary catalyst.

Comparison of CLZN Analogs for Measurements of RlucActivity in Mammalian Cells

To determine the photon yield of each analog andtheir availability to intracellular Rluc, HeLa and C6 celllines were used in both live cell and cell lysate assays(Figure 3, HeLa cell data shown only). Of the 10 analogstested, three (-ip, -i, -m) produced lower biolumine-scent signals than the native substrate (data not shown).CLZN-e showed a significant higher emission rate(7.5 £ 104 ± 6.18 £ 103 photons/sec/cell, p < .001),which was 8-fold higher than that of CLZN-native(9.4 £ 103 ± 2.8 £ 103 photons/sec/cell). Two otheranalogs (-f, -h), also demonstrated significantly greatersignals (4-fold, p < .001). For all analogs, peak signalswere observed 1 min after CLZN addition, with theexception of -n (peak at 10 min) and -hcp (peak at5 min) (Figure 3A and Table 1). Control cells producednegligible signals (2.90 ± 0.55 photons/sec/cell).

Table 1. Properties of CLZN Analogs in Cells and Mice

Auto-Oxidation Rluc Activity

Medium Mouse Cell Culture Mouse

Serum PBS Intraperitoneal Intravenous Whole Cells Cell Lysate Whole Body Liver Lysate

-native 1a 1b 1c 1d 1e 1f 1g 1h

-e 23.7 14.4 150 37.0 8.0 7.5 3.1 4.8

-h 1.2 0.7 – – 4.8 1.0 0.19 1.5

-f 0.9 0.6 – – 4.1 0.8 0.25 1.0

-cp 4.6 0.7 – – 0.5 0.4 0.19 0.5

-n 3.5 0.7 – – 0.7 0.4 0.02 0.6

-hcp 3.5 0.6 – – 0.9 0.1 – –

-ip 2.8 – – – 0.2 0.03 – –

Data are expressed as fold activity relative to that of native CLZN.a 4.38 £ 105 photons/sec/50 L.b 1.79 £ 104 photons/sec/50 L.c 5.06 £ 105 photons/sec/ROI (ROI = whole body).d 4.08 £ 105 photons/sec/ROI (ROI = whole body).e 9.4 £ 103 photons/sec/cell.f 3.26 £ 103 RLU/cell (luminometer).g 1.95 £ 109 photons/sec/ROI (ROI = liver region).h 6.5 £ 109 RLU/g liver ( luminometer).

Characterization of CLZN Analogs Zhao et al. 47

Molecular Imaging . Vol. 3, No. 1, January 2004

Figure 3. Comparison of Rluc activity with CLZN analogs in mammalian cells. (A) Light intensity from live HeLa cells after the addition of CLZN analogs (-e, -native, -h,

-f, -cp, -n, -hcp, -ip) was measured in 96-well plates using the IVIS, and are expressed as total photons emitted per sec per 3 £ 104 cells/well (mean, n = 4). (B) Light

intensity of cell lysates was measured by single tube luminometer following addition of various CLZN analogs (-e, -native, -h, -f, -cp, -n, -hcp, -ip) and expressed as

relative light units (RLU) per cell (mean, n = 3). (C) The ratio of relative photons/cell ( Pcell) to photons/lysate ( Plysate) was determined and plotted. For various CLZN

analogs (-native, -e, -h, -f, -cp, -n), the ratio of light intensity measured from live cell assays to that from cell lysate assays after 1 min was normalized and compared

with the CLZN-native value.

48 Characterization of CLZN Analogs Zhao et al.

Molecular Imaging . Vol. 3, No. 1, January 2004

To determine if the live cell measurements reflectedthe levels of Rluc, activity measurements were also madein cell lysates using the various CLZN analogs. Measure-ments could not be made immediately after addition ofsubstrate using the IVIS due to the time required forconfiguration of the detector (approximately 1 min).Therefore, the IVIS was used for measurements over atime interval of 1 to 30 min. A single tube luminometerwas used for measurements of time intervals from 1 secto 2 min after substrate addition. The relative magnitudeof signal from each CLZN analog in each instrumentwas similar, confirming data acquisition by either instru-ment was comparable. Single tube luminometer mea-surements of CLZN showed that the light intensitypeaked within 10 sec and decreased gradually over time.Of all the analogs tested, CLZN-e was observed toproduce the greatest signal ( p < .001). Surprisingly,-h and -f showed similar intensities to that of CLZN-native and was different from that observed in the livecell assays (Figure 3B and Table 1).

In order to determine the efficiency of substrateavailability within cells, we calculated the ratio of theinitial light intensity from live cell assay to cell lysateassay ( Pcell/Plysate) 1 min post-CLZN addition. These ratioswere then normalized to the ratio found with the CLZN-native analog and varied from one analog to the other. Inaddition, when both HeLa and C6 cells were tested,results were similar: Pcell/Plysate from CLZN-h, -f were 4-to 5-fold higher than -e and -native (Figure 3C).

Background Signals from CLZN-Native and -e In VivoThe observation that autoluminescence of CLZN in

the presence of serum produced significant signalssuggested that background signals from CLZNs in ani-

mals may present problems when reporter gene expres-sion levels are low. We tested the background signals forthe -native and -e analogs by injecting them intountreated wild-type mice via intraperitoneal or intrave-nous routes. For both routes of administration, back-ground signals were only observed ventrally (images notshown). Signals obtained at 1 min post-injection variedamong the analogs. CLZN-e displayed much more in-tense signals (7.6 £ 107 ± 5.9 £ 106 photons/sec ip and1.5 £ 107 ± 7.7 £ 106 photons/sec for iv) than CLZN-native (5.0 £ 105 ± 7.4 £ 104 photons/sec ip and 4.1 £105 ± 3.5 £ 104 photons/sec iv). Following intraperito-neal injections, signals emanated primarily from theabdominal region and persisted above the background(2.2 £ 105 ± 2.1 £ 104 photons/sec/mouse) for morethan 40 min. In contrast to intravenous injections ofCLZN-e, signals were observed initially over the wholebody, then localized to the neck and abdominal region,and diminished to background within 20 min (Figure 4).

Expression of Rluc in Murine Liver and TemporalAnalyses of Rluc Activity Following Intraperitonealor Intravenous Injections of CLZN Analogs

After hydrodynamic transfection using the plasmidhRluc-CMV (at a concentration of 5 g/mouse), Rlucactivity was highly expressed in mouse liver at 22 to24 hr after administration of various CLZN analogs.Temporal analyses were performed to determine thetime of peak light emission in vivo. In animals that re-ceived CLZN-native (1 mg/kg BW; administrated eitherintraperitoneally or intravenously), signals were local-ized to a region over the liver (Figure 5A). However, adifferent temporal pattern and rate of light productionwas observed between the two routes of delivery

Figure 4. Background signals from CLZN-native and -e in vivo. Representative images are shown of wild-type FVB mice injected with CLZN-native or -e via

intraperitoneal or intravenous routes. Pseudocolor bioluminescent images were superimposed over grayscale reference images of mice. All images were collected at

1 min after CLZN administration unless otherwise indicated.

Characterization of CLZN Analogs Zhao et al. 49

Molecular Imaging . Vol. 3, No. 1, January 2004

[slope for intravenous injections was 0.083 log(photon/sec)/min, and 0.015 log(photon/sec)/min for intraperi-toneal injections; Figure 5B]. Light intensity from theintravenously injected group peaked immediately afterCLZN injection and then rapidly declined over 30 min.In contrast, peak light production by intraperitonealinjection was reached after 2 to 3 min and was relativelystable for about 10 min. In addition, light producedwithin the first min of intravenous injection was morethan 10-fold greater than that by intraperitoneal injection.

A dose response of the native CLZN was thenevaluated for each route of administration at four con-centrations (0.1, 0.5, 1.0, and 4.0 mg/kg BW; Figure 5C).Signal intensity was observed to be dose-dependentwith highest overall signals produced by intravenousadministration. Decreases of signal intensity were sim-ilar when the same route of administration was used(data not shown).

To determine whether quantitative measurementsof Rluc activity can be made when the reporter is ex-pressed in deep tissues (e.g., liver) using CLZN concen-trations that are below the levels that would saturate theenzyme, in vivo liver transfections were performed usingdifferent amounts of hRluc plasmid DNA (0.2, 1.0, and5.0 g/mouse) to vary the hRluc levels. Although higherconcentrations of transfected DNA produced higher sig-nals, the difference was not significant ( p > .05). How-ever, the slopes of signal loss were slightly different [slopefor 0.2 g DNA is 0.055 log(photon/sec)/min; slope for 1 gDNA is 0.078 log(photon/sec)/min; and slope for 5 gDNA is 0.083 log(photon/sec)/min; Figure 5D].

Dual Expression of Rluc and Fluc Activities in the LiverWe found that both Rluc and Fluc activities can be

measured within the same liver and the expression ofRluc or Fluc could be distinguished by injecting CLZN

Figure 5. Temporal expression of Rluc in mice after hydrodynamic transfection and administration of CLZN-h by intraperitoneal or intravenous routes of injection.

(A) Rluc activity was measured in mice transfected with a plasmid that encodes hRluc (5.0 g/mouse). Images were acquired 24 hr after intraperitoneal or intravenous

injections of CLZN-h. A representative ROI (red box) is shown. (B) Comparison of light output kinetics (mean ± SD photons/sec/ROI, n = 3 measurements for each time

point) from an ROI corresponding to the region over the liver following intraperitoneal or intravenous injections of CLZN-h. Measurements were made over 30 and

90 min. (C) Dose response for CLZN-h. Various concentrations of CLZN-h were injected intraperitoneally or intravenously into transfected mice and total flux from

livers was compared at 1 min after CLZN-h administration (n = 4). (D) The effect of hRluc plasmid dose on in vivo light intensity. Three concentrations (0.2, 1.0, and

5.0 g/mouse) of hRluc-CMV plasmid were used to transfect cells in the liver (n = 4 mice for each concentration) and light intensity was plotted over 30 min.

50 Characterization of CLZN Analogs Zhao et al.

Molecular Imaging . Vol. 3, No. 1, January 2004

Figure 6. Comparison of CLZN analogs in living mice and liver tissue expressing Rluc. (A) The light output was determined for each analog in mice. Mice were first

transfected with the phRluc-CMV plasmid (5.0 g/mouse) and then injected intravenously with CLZN analogs (-e, -native, -h, -f, -cp, -n). Mean light intensities (photons/

sec/ROI, ROI = region over the liver, n = 4 for each time point) were collected using the IVIS over a period of 15 to 20 min. (B) Comparisons of CLZN analogs in

homogenates of Rluc-transfected liver were performed using a luminometer. Mean light intensity (RLU per g liver) was determined (n = 3 for each time point) over a

120-min period following the addition of CLZN analogs (-e, -native, -h, -f, -cp, -n). (C) Comparison of auto-oxidation and Rluc activity was assessed in liver

homogenates. Light intensity (mean ± SD RLU/g liver) from liver homogenates of mice without transfection (n = 3) or after hRluc transfection (n = 3) was analyzed

using a luminometer 10 sec after addition of CLZN analogs (-e, -native, -h, -f, -cp, -n).

Characterization of CLZN Analogs Zhao et al. 51

Molecular Imaging . Vol. 3, No. 1, January 2004

and D-Luciferin sequentially. Similar to other reports[14], no cross-reactivity was observed between the twoluciferases and their substrates. Using identical con-structs, differing only in the coding sequences of thereporter genes, and the conditions described in theMethods section, firefly luciferase produced a muchgreater (>15-fold) and more stable signal than Rluc(data not shown). Optimization of substrate delivery,wavelength of emission and other parameters mayincrease the detectable signal and stability of lightproduction for Rluc.

Comparison of CLZN Analogs in Living Mice andTissues

To compare the light production from CLZN analogs invivo, six analogs (-e, -native, -h, -f, cp, and -n) wereselected, based on the live cell assay results, for injectioninto Rluc-transfected animals. Each analog was adminis-trated intravenously into mice and Rluc activity mea-sured by BLI. For all analogs, light production from theregion over the liver was maximal within the first minuteof administration and decreased at similar rates [roughly0.080 log(photons/sec)/min], except for CLZN-n, whichdecreased at a much slower rate [0.049 log(photons/sec)/min]. Of the analogs tested, CLZN-e and -nativeproduced the greatest signal intensities, averaging 6.1 £109and 1.95 £ 109photons/sec, respectively. Rluc activitywas weaker after administration of CLZN-h, -f, and -cp,being approximately 1/5 of that observed with the -nativeanalog (Figure 6A and Table 1).

Using the luminometer to make measurements of lucactivity in tissue homogenates, very low backgroundsignals were detected from the Rluc control liver homo-genates (without transfected hRluc plasmid) (¹3.0 £ 107

RLU/g liver) (Figure 6B). Consistent with the in vivodata, CLZN-e gave the strongest signal in the liver homo-genates (¹3.1 £ 1010 RLU/g liver, p < .001). However,Rluc activity was similar between -native, -f, and -h analogs(Figure 6C and Table 1). Using the IVIS, measurementsfrom liver homogenates and living animals with CLZN-native, as the substrate, were compared. Rluc activityfrom liver homogenates (2.43 £ 1011photons/sec/g liver)using 1 M CLZN was more than 100-fold greater than thatmeasured from whole body (1.95 £ 109 photons/seccollected from liver region, each liver was roughly 1 g).

DiscussionIn this study, we investigated various CLZN analogs foruse as substrates by the Rluc reporter system both in livecell cultures and in animals. Measurements made in livecells and animals were validated using measurements

in cell lysates and tissue homogenates. We found signif-icant differences in signals among the CLZN analogs aswell as different patterns of activation following deliveryvia intraperitoneal and intravenous routes to living mice.These differences indicate that for quantitative and reli-able measurements of Rluc activity, it is essential thatrigorous standards be used as shown by studies with Fluc.Detection of a signal in the absence of standardizationand without considering the properties of light transmis-sion through tissues can lead to inaccurate measure-ments for any optical reporter or tracer.

All CLZN analogs tested produced background lumi-nescence in the absence of enzyme in cell culture media,especially those containing serum, due to the auto-oxi-dation of CLZN analogs. CLZNs are generally unstableeven in their solid state and in solutions, where they aresignificantly oxidized within 2 to 3 hr at 0 to 25 C [26].Because signals due to auto-oxidation could interferewith measurements of Rluc activity, especially when Rlucexpression levels are low, these assays should be per-formed only in PBS or DMEM. Because 1% albuminproduced similar auto-luminescence signals as 10% FBS,albumin is a likely contributor to CLZN oxidation. This isconsistent with the report that the solvent cage of albu-min can bind to CLZN and produce bioluminescence [21].Similar effects could be present in other buffers as welland they should be rigorously tested before use in theRluc–CLZN system.

Among the 10 analogs tested in cell culture and 6analogs in living animals, CLZN-e, -native, -h, and -fproduced the greatest signals. Two analogs, -h and -f,showed the most promising signals in live cell assays inPBS, and CLZN-native may be the most appropriateanalog for use in mouse experiments—at least for hRlucexpression in the liver. CLZN-e is the most recent com-mercially available analog and according to the manufac-turer’s report, its initial light intensity is more thanseven times higher than that of CLZN-native [27]. Thisis corroborated by our results where we found that the -eanalog produced as much as 6- to 7-fold higher signalintensity than did the -native analog in cell studies, and a3- to 4-fold greater signal production in the liver studies.However, -e appeared to be more readily oxidized andproduced the greatest auto-luminescence signal in allmedia we tested as well as mice. Furthermore, thesignal-to-noise ratio (SNR) will depend on the Rluc ex-pression levels and assay conditions. In our cell cultureand animal models, the Rluc-mediated signals were 100-to 1000-fold greater than those due to auto-oxidation,which did not contribute significantly to the overall signal.However, in cases where reporter gene expression may

52 Characterization of CLZN Analogs Zhao et al.

Molecular Imaging . Vol. 3, No. 1, January 2004

be low, the noise due to auto-oxidation could interferewith the assay.

Because light output from the bioluminescence reac-tion will depend on substrate availability, the permeabilityof the cell membrane to CLZN can directly affect themeasurement of Rluc activity in living cells. Although it iswidely accepted that CLZN-native and its analogs cansufficiently permeate through various cell membranes[31,32], thorough studies have yet to be performed.Permeability may vary considerably for different celltypes and substitution groups on the CLZN analogs. CLZNavailability has been shown to be the rate-limiting step inintracellular aequorin reactions due to its slow diffusioninto Drosophila oocytes [22]. The membrane permeabil-ity of CLZN analogs has been measured in killifish eggsand CLZN-e is found to be poorly permeable. In contrast,the -f analog is reported to have the highest permeability[33]. In our study using HeLa and C6 cells, we observeda difference in Pcell/Plysate values from the analogs tested.If the Rluc catalytic reaction is assumed to be the same inlive cells and cell lysates, the difference of Pcell/Plysate

values should reflect an analog’s intracellular availability.Assays with CLZN-h and -f showed higher Pcell/Plysate

values, possibly an indication of higher influx or lowerefflux. Because bioluminescence assays have typicallybeen performed in cell lysates, very little is known aboutthe mechanisms of influx or efflux of CLZN and itsderivatives in live cells. However, with the developmentof imaging strategies that employ in vivo reporters, it hasbecome imperative that these processes be evaluated. Ina recent study, it was reported that CLZN shares structuralproperties with compounds that are substrates for themultidrug resistance (MDR1) P-glycoprotein (Pgp) [34].This study supported the hypothesis that Pgp mediatesthe efflux of CLZN. They further demonstrated that CLZN-native, -f, -h, and -hcp were transported by Pgp, whereasCLZN-n and -cp were not [34]. In addtion to Pgp, manyaspects, including other membrane transporters, seques-tration of substrate within cells and potentially auto-oxidation within cells, might also contribute to the vari-ation of CLZN intracellular availability. All of these need tobe addressed specifically in future studies.

Here, we employed a nonviral hydrodynamic livertransfection technique where up to 50% of hepatocytescan be transfected by plasmid DNA after injection intothe tail vein of mice [29,30]. In this model, signals fromRLuc expressed in the liver could be detected whenCLZNs were administrated via either intraperitoneal orintravenous injection, albeit the dynamics of light pro-duction appeared to differ substantially between tworoutes. Intravenous administration produced signals of

greater intensity, which also decreased more rapidly thanthose following intraperitoneal injections; however, in-traperitoneal injections provided detectable signals for alonger duration than in intravenous administration (10min) with relative stability, but the intensities were sig-nificantly less than those following intravenous injections.The longer duration of detectable signal may be beneficialin some applications, however, the lower intensity may beproblematic. The kinetics of signal production that weobserved after intraperitoneal injections have never beenreported before while those from intravenous injectionconfirmed previously reported data [14]. Based on ourdata, it was possible to detect the activities of Rluc andFluc in the same organ with no apparent cross-reactivity.Therefore, even in mouse tissues, it is possible to useboth reporters.

Various concentrations of CLZN were tested and adose-dependent response was observed. The CLZNdose of 1 mg/kg BW did not saturate the reaction. Infact, substrate saturation may not have achieved evenusing highest substrate concentration (4 mg/kg BW;Figure 5C). However, considering cost and potentialtoxicity, 1 mg/kg BW might be an optimal concentration.Because the substrate is not in excess, the level of lightproduction may be due to substrate availability and notenzyme concentration. Quantitative measurements ofRluc activity are, therefore, relative to the site of expres-sion, route of substrate delivery, time post-delivery thatthe measurement is made, Pgp status of the expressingcell, and possibly cell type. Effective use of Rluc andother CLZN-utilizing enzymes will depend on improvingour understanding of CLZN biodistribution, cellularpermeability, and effects of Pgp and other cellularproteins on CLZN availability.

Development of CLZN-utilizing enzymes as effectivein vivo reporters should include the design and charac-terization of substrates with improved biological prop-erties. This is evidenced by the observations that signalsfrom Rluc can be increased through the injection ofmore substrate [35]. The fact that saturation of theenzyme with substrate is difficult to achieve with CLZN-utilizing enzymes, due to cost and biodistribution,makes quantitation difficult as signals may representsubstrate levels and not enzyme expression levels.

Careful selection of CLZN analogs for in vivo mea-surements of Rluc activity appears to be warranted fromthese studies, and these data suggest a need for thedevelopment and testing of additional analogs for im-proved imaging of Rluc in vitro and in vivo. Based on thecriteria of high signal intensity and low background fromauto-oxidation, the -h and -f analogs appear to be the best

Characterization of CLZN Analogs Zhao et al. 53

Molecular Imaging . Vol. 3, No. 1, January 2004

available substrates for cell culture studies and -nativeis optimal for studies in animals.

AcknowledgmentsWe thank Dr. Sanjiv (Sam) Gambhir for his critical review of thismanuscript and Drs. Stacy M. Burns and Hamid R. Kazerouni Zadeh for

their help with this study. This work is supported, in part, by a

Technology Innovation grant from the NCI (CA 88303 [CHC]), other

grants from National Institutes of Health (HL 58013 [DKS, CHC]; CA94056 [DP]) and unrestricted gifts from the Mary L. Johnson and Hess

Research Funds.

References[1] Contag CH, Bachmann MH (2002). Advances in in vivo bio-

luminescence imaging of gene expression. Annu Rev BiomedEng. 4:235 –260.

[2] Contag CH, Contag PR, Mullins JI, Spilman SD, Stevenson DK,Benaron DA (1995). Photonic detection of bacterial pathogensin living hosts. Mol Microbiol. 18:593 –603.

[3] Contag PR, Olomu IN, Stevenson DK, Contag CH (1998).Bioluminescent indicators in living mammals. Nat Med. 4:245–247.

[4] Costa GL, Sandora MR, Nakajima A, Nguyen EV, Taylor-EdwardsC, Slavin AJ, Contag CH, Fathman CG, Benson JM (2001).Adoptive immunotherapy of experimental autoimmune ence-phalomyelitis via T cell delivery of the IL-12 p40 subunit. JImmunol. 167:2379 –2387.

[5] Hardy J, Edinger M, Bachmann MH, Negrin RS, Fathman CG,Contag CH (2001). Bioluminescence imaging of lymphocytetrafficking in vivo. Exp Hematol. 29:1353 –1360.

[6] Lipshutz GS, Titre D, Brindle M, Bisconte AR, Contag CH,Gaensler KM (2003). Comparison of gene expression afterintraperitoneal delivery of AAV2 or AAV5 in utero. Mol Ther.8:90–98.

[7] Rehemtulla A, Stegman LD, Cardozo SJ, Gupta S, Hall DE,Contag CH, Ross BD (2000). Rapid and quantitative assessmentof cancer treatment response using in vivo bioluminescenceimaging. Neoplasia. 2:491–495.

[8] Sweeney TJ, Mailander V, Tucker AA, Olomu AB, Zhang W, Cao Y,Negrin RS, Contag CH (1999). Visualizing the kinetics of tumor-cell clearance in living animals. Proc Natl Acad Sci USA.96:12044 –12049.

[9] Contag CH, Spilman SD, Contag PR, Oshiro M, Eames B,Dennery P, Stevenson DK, Benaron DA (1997). Visualizing geneexpression in living mammals using a bioluminescent reporter.Photochem Photobiol. 66:523–531.

[10] Hastings JW, Johnson CH (2003). Bioluminescence and chem-iluminescence. Methods Enzymol. 360:75 –104.

[11] Wilson T, Hastings JW (1998). Bioluminescence. Annu Rev CellDev Biol. 14:197–230.

[12] Matthews JC, Hori K, Cormier MJ (1977). Purification and pro-perties of Renilla reniformis luciferase. Biochemistry. 16:85 –91.

[13] DeLuca M, Dempsey ME, Hori K, Wampler JE, Cormier MJ (1971).Mechanism of oxidative carbon dioxide production during Re-nilla reniformis bioluminescence. Proc Natl Acad Sci USA.68:1658 –1660.

[14] Bhaumik S, Gambhir SS (2002). Optical imaging of Renillaluciferase reporter gene expression in living mice. Proc NatlAcad Sci USA. 99:377 –382.

[15] Grentzmann G, Ingram JA, Kelly PJ, Gesteland RF, Atkins JF(1998). A dual-luciferase reporter system for studying recodingsignals. RNA. 4:479–486.

[16] Tabuchi A, Sano K, Nakaoka R, Nakatani C, Tsuda M (1998).Inducibility of BDNF gene promoter I detected by calcium-phosphate-mediated DNA transfection is confined to neuronal butnot to glial cells. Biochem Biophys Res Commun. 253:818 –823.

[17] Wang Y, Wang G, O’Kane DJ, Szalay AA (2001). A study ofprotein –protein interactions in living cells using luminescenceresonance energy transfer (LRET) from Renilla luciferase toAequorea GFP. Mol Gen Genet. 264:578–587.

[18] Arai R, Nakagawa H, Tsumoto K, Mahoney W, Kumagai I, Ueda H,Nagamune T (2001). Demonstration of a homogeneous non-competitive immunoassay based on bioluminescence resonanceenergy transfer. Anal Biochem. 289:77–81.

[19] Angers S, Salahpour A, Joly E, Hilairet S, Chelsky D, Dennis M,Bouvier M (2000). Detection of beta 2-adrenergic receptordimerization in living cells using bioluminescence resonanceenergy transfer (BRET). Proc Natl Acad Sci USA. 97:3684 –3689.

[20] Paulmurugan R, Gambhir SS (2003). Monitoring protein –proteininteractions using split synthetic renilla luciferase protein-frag-ment-assisted complementation. Anal Chem. 75:1584 –1589.

[21] Campbell AK, Patel A, Houston WA, Scolding NJ, Frith S, MorganBP, Compston DA (1989). Photoproteins as indicators of intra-cellular free Ca2+. J Biolumin Chemilumin. 4:463–474.

[22] Creton R, Steele ME, Jaffe LF (1997). Expression of apo-aequorinduring embryonic development; how much is needed forcalcium imaging? Cell Calcium. 22:439 –446.

[23] Kendall JM, Dormer RL, Campbell AK (1992). Targetingaequorin to the endoplasmic reticulum of living cells. BiochemBiophys Res Commun. 189:1008 –1016.

[24] Kendall JM, Badminton MN (1998). Aequorea victoria bio-luminescence moves into an exciting new era. Trends Biotech-nol. 16:216 –224.

[25] Dubuisson ML, de Wergifosse B, Trouet A, Baguet F, Marchand-Brynaert J, Rees JF (2000). Antioxidative properties of naturalcoelenterazine and synthetic methyl coelenterazine in rat hep-atocytes subjected to tert-butyl hydroperoxide-induced oxidativestress. Biochem Pharmacol. 60:471–478.

[26] Shimomura O, Kishi Y, Inouye S (1993). The relative rate ofaequorin regeneration from apoaequorin and coelenterazineanalogues. Biochem J. 296:549–551.

[27] Inouye S, Shimomura O (1997). The use of Renilla luciferase,Oplophorus luciferase, and apoaequorin as bioluminescentreporter protein in the presence of coelenterazine analogues assubstrate. Biochem Biophys Res Commun. 233:349 –353.

[28] Rice BW, Cable MD, Nelson MB (2001). In vivo imaging of light-emitting probes. J Biomed Opt. 6:432 –440.

[29] Liu F, Song Y, Liu D (1999). Hydrodynamics-based transfectionin animals by systemic administration of plasmid DNA. GeneTher. 6:1258 –1266.

[30] Zhang G, Budker V, Wolff JA (1999). High levels of foreign geneexpression in hepatocytes after tail vein injections of nakedplasmid DNA. Hum Gene Ther. 10:1735 –1737.

[31] Knight MR, Campbell AK, Smith SM, Trewavas AJ (1991). Trans-genic plant aequorin reports the effects of touch and cold-shockand elicitors on cytoplasmic calcium. Nature. 352:524–526.

[32] Knight MR, Read ND, Campbell AK, Trewavas AJ (1993). Imagingcalcium dynamics in living plants using semi-synthetic recombi-nant aequorins. J Cell Biol. 121:83 –90.

[33] Shimomura O (1997). Membrane permeability of coelenterazineanalogues measured with fish eggs. Biochem J. 326:297–298.

[34] Pichler A, Prior JL, Piwnica-Worms D (2004). Imaging reversal ofmultidrug resistance in living mice with bioluminescence: MDR1P-glycoprotein transports coelenterazine analogues. Proc NatlAcad Sci USA. 101:1702 –1707.

[35] Bhaumik S, Lewis XZ, Gambhir SS (in press). Optical imaging ofRenilla luciferase, synthetic Renilla luciferase, and firefly lucifer-ase reporter gene expression in living mice. J Biomed Opt.

54 Characterization of CLZN Analogs Zhao et al.

Molecular Imaging . Vol. 3, No. 1, January 2004