Genetically Engineered Bacterial Outer Membrane Vesicles withExpressed Nanoluciferase Reporter for in Vivo BioluminescenceKinetic Modeling through Noninvasive ImagingYikun Huang,†,# Andre O’Reilly Beringhs,‡,# Qi Chen,§ Donghui Song,† Wilfred Chen,§ Xiuling Lu,‡

Tai-Hsi Fan,*,∥ Mu-Ping Nieh,⊥ and Yu Lei*,†,⊥

†Department of Biomedical Engineering, University of Connecticut, Storrs, Connecticut 06269, United States‡Department of Pharmaceutical Sciences, University of Connecticut, Storrs, Connecticut 06269, United States§Department of Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware 19716, United States∥Department of Mechanical Engineering, University of Connecticut, Storrs, Connecticut 06269, United States⊥Department of Chemical and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut 06269, United States

ABSTRACT: Outer membrane vesicles (OMVs) producedby Gram-negative bacteria play significant roles in the bio-medical field as they can be facilely functionalized usinggenetic engineering tools and thus often serve as a versatilemultifunctional nanoparticles for a variety of applications.In this study, we investigated the multifaceted bioluminescencekinetics of a NanoLuc luciferase-expressed outer membranevesicle produced by E. coli. This multifunctional OMV emitsstrong blue luminescence at 460 nm after mixing with the sub-strate furimazine, which potentially can be used for bio-luminescence-based optical imaging. Characterization of thevesicles was performed via dynamic light scattering and nano-particle tracking analysis. A murine animal model was used to observe the in vivo behavior of the bioluminescence producedby outer membrane vesicles through post subcutaneous administration. The bioluminescence signal was tracked by noninvasivein vivo optical imaging, while in vitro cytotoxicity and ex vivo tissue histopathology were studied to demonstrate the biocom-patibility of the engineered OMVs. A theoretical model was also developed to simulate the relevant enzyme−substrate reactionkinetics along with absorption of the in vivo system. The interplay of the reaction and absorption is in good agreement with theexperimental results. The study shows a great potential of the genetically engineered vesicles as an interesting class of functionalnanomaterials for imaging-related biomedical applications.

KEYWORDS: outer membrane vesicles, bioluminescence, NanoLuc luciferase, modeling, in vivo reaction kinetics

■ INTRODUCTION

Functional nanoparticles, including artificial and natural nano-particles, are widely utilized in biomedical research and phar-maceutical development.1−4 Artificial nanoparticles, such asquantum dots or carbon dots, have been successfully used forcell imaging and in vitro protein detection.5,6 However, artificialnanoparticles may have high toxicity due to the chemical agentsor metals applied during the manufacturing process,7 whichgreatly limits their extensive in vivo application. In this scenario,natural nanoparticles such as extracellular vesicles (EVs) can be asignificant alternative material for bioimaging, diagnosis, andmedical treatments due to their intrinsic biocompatibility.8,9

EVs can be secreted from some mammalian cells and Gram-negative bacteria.10,11 EVs isolated from mammalian cells havebeen employed in several applications, such as early cancerdiagnosis or as carriers for therapeutic agents.12,13 However,these vesicles must be isolated from tissues or body fluidsbefore use,14 which typically suffer from low yield. It is also

difficult to endow them multifunctionality. With the developmentof recombinant DNA technology, outer membrane vesicles(OMVs) secreted from E. coli have become excellent alternatives.As a microfactory of OMVs, E. coli can be genetically engineeredto synthesize OMVs with tailored functionality for a variety ofapplications.15 For instance, bioengineered OMVs that containforeign antigens, glycopolymers, or other macromolecules wereused as drug-delivery vehicles for vaccines.15−17 In addition,E. coli culturing for OMV production is generally much easierthan mammalian cell culturing, which often requires strictgrowth conditions.18

On the other hand, the past decades have witnessed adramatic growth of optical imaging (e.g., fluorescence, biolu-minescence, etc.) in biomedical applications such as cell-based

Received: July 30, 2019Accepted: November 4, 2019Published: November 26, 2019

Article

www.acsabm.orgCite This: ACS Appl. Bio Mater. 2019, 2, 5608−5615

© 2019 American Chemical Society 5608 DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

Dow

nloa

ded

via

UN

IV O

F D

EL

AW

AR

E o

n D

ecem

ber

19, 2

019

at 1

4:22

:15

(UT

C).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

assay and animal imaging.3,19−21 Among different opticalimaging methods, bioluminescence-based imaging has its ownadvantages. Since excitation is not required for biolumines-cence to occur, autofluorescence, photoxicity, and photo-bleaching issues that are commonly encountered in fluorescenceoptical imaging therefore can be avoided.22 Traditionally,Firefly and Renilla luciferases are the most commonly usedluciferase enzymes to produce bioluminescence.23 However,these luciferases are only useful in a small scope of applicationsdue to their intrinsic low stability and strict reaction condi-tions, requiring the presence of ATP.24 The novel NanoLucluciferase targets a much wider range of applications because itis an ATP-independent luciferase and displays a relatively longhalf time and high stability under a wider pH range.25 In thissense, the in vivo subcutaneous use of NanoLuc-expressing bio-luminescent OMVs can be an extraordinary tool for a variety ofbiomedical applications. For example, we recently reported abioluminescence-based immunoassay relying on multifunc-tional outer membrane vesicles for immunoglobulin G detec-tion, which exhibits a comparable detection limit as that of thecommercial IgG ELISA kit.26

To further exploit the application of OMVs, we conductedthe in vivo and in silico study of bioluminescence kinetics ofgenetically engineered multifunctional OMVs in this work. Theouter membrane vesicles containing NanoLuc luciferase andthe Z-domain (Figure 1) are secreted and isolated from thegenetically engineered E. coli. As shown in Figure 1b, the outermembrane lipoprotein SlyB is utilized as an anchor to packagethe NanoLuc inside the vesicle. Another anchor, the ice nucle-ation protein (INP), was employed to decorate the surface ofthe vesicle with Z-domain through a trifunctional scaffold(consisting of three cohesion domains from Clostridiumcellulolyticum (CC), Clostridium thermocellum (CT), andRuminococcus flavefaciens (RF)).27 The outer membrane vesi-cles used in this study potentially have many other applicationsas they can be further decorated with dockerin-tagged heter-ologous proteins based on the high affinity of cohesion−dockerin interaction,28 while the Z-domain has high bindingaffinity with the IgG antibody; thus, outer membrane vesiclescan be further functionalized with an antibody of interest torecognize biomarkers on cancer cells for bioluminescence-based targeted tumor imaging. These vesicles exhibited smalland uniform size and high concentration according to dynamiclight scattering and nanoparticle tracking analysis. Cytotoxicityand tissue histology analysis indicate that the OMVs alsopossess high biocompatibility. The in vivo bioluminescencekinetics was experimentally analyzed through local absorptionreaction and circulation studies based on noninvasive luminescence

optical imaging. To better understand the in vivo behaviorof these OMVs, an enzyme−substrate reaction kinetics andabsorption model is proposed to predict the in vivo kinetics ofbioluminescence which holds a promise for further in vivoquantitative analysis.

■ EXPERIMENTSPreparation of OMVs. OMVs were manufactured and collected

from the cell culture medium following the procedure reportedelsewhere.27 Briefly, the engineered bacteria were cultured andinduced to secrete vesicles. In order to separate cells and culturalmedia, the cell culture was centrifuged for 20 min at 4000 rpm and4 °C. The supernatant containing OMVs was filtered with a 0.2 μmsyringe filter to remove the remaining cells and large aggregates.Then, the residual aggregation was removed by centrifugation for 1 hat 40 000g at 4 °C. Ultracentrifugation was applied to concentrate theOMVs for 3 h at 1 440 000g and 4 °C. The final pellet was resuspendedin pH 7.4 PBS buffer and freshly used for subsequent experiments.

Characterization of OMVs. The concentration of vesicles insuspension was determined by nanoparticle tracking analysis (NTA)using the NanoSight NS500 system (Malvern Instruments, UK).Briefly, 2 μL of purified OMV samples were diluted in 1998 μL ofultrapure water prior to measurement. The final concentration wascorrected by the dilution factor and expressed as an average of fivemeasurements. Zetasizer Nano ZS (Malvern Instruments, UK) wasemployed to measure the hydrodynamic diameter of the OMV samplevia dynamic light scattering. The refractive index and viscosity ofwater were used as parameter inputs.

In Vitro Cytotoxicity. In vitro cytotoxicity of OMVs was testedagainst human lung carcinoma A549 cells (ATCC CCL-185TM).Cells were seeded on a 96-well polystyrene microplate at a seedingdensity of 10 000 cells/well and allowed to attach and grow for 48 h(37 °C, 5% CO2), embedded in Dulbecco’s Modified Eagle Medium(DMEM) supplemented with 10% (v/v) fetal bovine serum. OMVsuspension was neutralized to pH 7.4 and diluted in supplementedDMEM. Dilutions (1 × 1011, 1 × 1010, 1 × 108, 1 × 106, 1 × 104, and1 × 102 vesicles/mL) were used to replace the cell culture medium ofthe polystyrene microplate. Cells were incubated with OMVs for4 and 24 h in two distinct cell culture plates (37 °C, 5% CO2; thenumber of plates n = 3). At the end of each incubation period, OMV-containing medium was removed from the corresponding plate andreplaced with fresh supplemented DMEM, followed by the addition of10 μL of 2 mM tetrazolium salt (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)) solution. After 4 h of incu-bation, the medium was removed and the content in each celldissolved in dimethyl sulfoxide. Cell viability was calculated bymeasuring the absorbance of each well at 540 nm using the following

equation:ÄÇÅÅÅÅÅÅÅÅ

ÉÖÑÑÑÑÑÑÑÑ= ×

−

−Viability (%) 100A A

A A

( )

( )sample blank

0 blank, where Asample, Ablank,

and A0 refer to the values of absorbance from the sample, the blank,and the control, respectively.

Animals. In vivo experiments were conducted using femaleathymic nude mice. All experiments were conducted in accordance

Figure 1. Schematic illustration (not to scale) of the secretion of outer membrane vesicles from genetically engineered E. coli (a) and the structurescheme of outer membrane vesicles (b).

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5609

with the University of Connecticut Institutional Animal Care and UseCommittee approved protocol. Humane use and care of vertebrateanimals were ensured in the research.In Vivo Bioluminescence Kinetics. To prepare the solution for

animal experiments, outer membrane vesicle suspension and Nano-Glo (Promega, USA) luciferase assay substrate were diluted by PBS(pH = 7.4) under the injection concentration. The final concentrationof OMVs used for in vivo study was 6.25 × 1010 vesicles/mL, while thefinal concentration of the substrate was equivalent to 7.3 × 10−4 M.All materials and reagents for injection were sterilized by eitherheating or filtration prior to the experiment.Local Absorption and Reaction Kinetics Study. To study the

local absorption of the substrate and outer membrane vesicles as wellas the bioluminescence kinetics, 200 μL of OMV dilution was injectedsubcutaneously to the right flank of female athymic nude mice (thenumber of mice n = 3). Three minutes after the OMVs injection,200 μL of substrate solution was injected subcutaneously into thesame location of the OMV injection. To generate a kinetic curve for abioluminescence signal, the imaging acquisition was continuallyperformed after the substrate injection by using the IVIS SpectrumCT(PerkinElmer, USA) under the luminescence mode. Bioluminescenceoptical imaging was collected at different time intervals under autoex-posure, medium binning, and open emission filter. Dynamic changesin bioluminescence were monitored subsequently by injecting 200 μLof substrate solution at 0 and 5 h post OMV injection. After eachinjection, the bioluminescence signal was tracked until the biolumi-nescence signal is back to the background level. Data were expressedas a function of total radiant efficiency over time.Circulation and Reaction Kinetics Study. To study the delivery

of the substrate through a circulation system, 200 μL of OMVsuspension was injected into the right flank of mice (the number ofmice n = 3) subcutaneously prior to the administration of the sub-strate. Then, the as-prepared substrate solution was intravenouslyinjected in the mice through the tail vein using a 30G BD syringe.To track the substrate delivery and local consumption, imagingsessions were performed every 3 min after the substrate injection untilthe signal reduced to the same level as the background. The dynamicchanges in bioluminescence were monitored by injecting 200 μL ofthe substrate solution at 0 and 40 min post OMV injection. After eachinjection, the bioluminescence signal was tracked until the signalintensity decayed to the level similar to the background. Data wereexpressed as a function of total radiant efficiency over time. In allexperiments, animals were sedated in an anesthesia induction boxwith 4% isoflurane and then kept under isoflurane gas anesthesia (2%isoflurane, O2 0.5 L/min) during injections and imaging sessions.Ex Vivo Biocompatibility Study. Subcutaneous tissue samples

and major organs (liver, kidneys, and spleen) from OMV-treated micewere collected at the end of the in vivo bioluminescence kineticsexperiment and preserved in 10% neutral buffered formalin. Collectedtissues were blocked in paraffin and sectioned into 5 μm slices andstained with hemotoxylin and eosin (H&E). Histopathological analy-sis was performed at the Connecticut Veterinary Medical DiagnosticLaboratory.Modeling of Reaction and Absorption Kinetics. The

luciferase reaction kinetics and OMV/substrate absorption processobserved from the in vivo bioluminescence can be quantified theo-retically for a better understanding of the mechanisms. A lumped(zero-dimensional) approximation of the coupled reaction andabsorption model is proposed next. Considering the enzyme−substrate reaction that generates the luminescence

+ + ← →⎯⎯ → + + +−S E C P E hvO light COk k k

2,

21 1 2

(1)

where S is the substrate furimazine; E is the enzyme concentrationwhich is represented using enzyme-expressed OMV concentration asan analog; C is the assumed substrate−enzyme complex; P refers tothe reaction products along with the released photon hv; k1 and k−1are the association and dissociation reaction rate constants for theformation of the substrate−enzyme complex, respectively; and k2 isthe catalytic rate constant. Due to the very low enzyme concentration

and the sufficiency of oxygen diffusion, oxygen concentration duringin vivo experiments is assumed to be a constant. Therefore, O2 con-centration is accommodated to the k1 value. The transient reactionkinetics and the continuous absorption effect can be described by thefollowing differential equations

= − + + −−dS dt k SE k C S k S/ 1 1 in dS (2a)

= − + + −−dE dt k SE k C k C k E/ 1 1 2 dE (2b)

= − − −−dC dt k SE k C k C k C/ 1 1 2 dC (2c)

=dP dt k C/ 2 (2d)

where the decaying coefficients kdS, kdE, and kdC account for overalldegradation and absorption of the substrate, enzyme, and complex,respectively, assuming kdE ≈ kdC. The source term S in represents asummation of multiple injections of furimazine at two intermittenttimes (including blood circulating effect kconv), expressed as

∑ = { − + Δ }=

S k m H t H t t( ) ( )i

n

i i iin0

conv(3)

where mi is the molar flow rate for the i-th injection; H is the Heavisidestep function; and Δt is the estimated intermittent injection time. Notethat the clearance effect due to blood flow through the reaction site isdifficult to evaluate. For the first test case without blood circulationeffect on the substrate, kconv is 1, whereas for the second case wherethat substrate is injected intravenously, we assume a significant reductionof the concentration to a level that only about 0.1% of the substratesarrives at the targeted site according to an estimation based on thebioluminescence intensity. The OMVs are implanted in place firstbefore the substrate injection. At t = 0, the substrate injection processstarts, but it takes ∼1 min to complete the substrate injection in bothmodels. Therefore, the initial conditions are thus defined as S(0) = 0,E(0) = E0, C(0) = 0, and P(0) = 0. From the conservation relations,eq 2a−2d reduce to substrate and complex equations, expressed as

= − + + + −−−dS dt k E e S k S k C S k S/ ( )k t

1 0 1 1 in dSdC (4a)

= − + + +−−dC dt k E e S k S k k k C/ ( )k t

1 0 1 1 2 dCdC (4b)

Presuming the short time behavior is not of interest, one can apply theMichaelis−Menten kinetics by assuming that the variation of complexconcentration remains quasi-steady at longer time, that is, dC/dt ≈ 0,29

and thus

≈+ +

−C t

E e S tS t K k k

( )( )

( ) /

k t0

M dC 1

dC

(5)

where KM = (k−1 + k2)/k1 is the Michaelis constant. Since the smallvariation of the complex concentration in the long-time regime isneglected in the Michaelis−Menten kinetics, we further consider theconstant decaying and clearance coefficients kdE, kdC, and kconv for thewhole process, whereas the coefficient kdS can be better approximatedby a linear dependency on the substrate concentration and is formu-lated as kdS(S) = kdS0S(t)/S0, in which S0 represents the injectedsubstrate concentration. Finally, by substituting eq 5 into eq 4a, thesubstrate equation for S(t) as a simplified reaction model can bewritten as

≈− +

+ ++ −

−dSdt

E e k k SS t K k k

Sk S

S( )

( ) /

k t0 2 dC

M dC 1in

dS02

0

dC

(6)

The resulting production velocity of the bioluminescent light, andthus the corresponding radiant intensity, is assumed to be linearlyproportional to the catalyzed reaction, expressed as

∝ ≥dhvdt

k C t t( ) for 02 (7)

which can be further scaled by the maximum observable intensity thatappeared at the initial stage for the comparison with experimental

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5610

data. Assuming k−1 is about the same order of magnitude as k2, andthus k1 can be calculated from KM, the simplified model requires theinputs of two kinetic coefficients, the Michaelis constant KM and thecatalytic rate constant k2, and four absorbing or decaying constants,

kdS0, kdE, kdC, and kconv, along with the experimental setting for theinitial concentrations of furimazine and enzyme and the amounts ofintermittent injections of furimazine. The proposed model isnumerically solved by the Runge−Kutta fourth-order time integrationscheme.

■ RESULTS AND DISCUSSION

Vesicle Size Distribution and Concentration. Thehydrodynamic diameter of OMVs was determined to be55 ± 1 nm according to the result from Zetasizer Nano ZS.The OMVs were washed and concentrated via ultra-centrifugation, and the final OMV concentration of the stockwas (6.25 ± 0.53) × 1011 vesicles/mL. It was diluted by 10-foldbefore injection. There were no noticeable large size particles invesicle size distribution data after ultracentrifugation, indicatingthat there was no aggregation of the vesicles under high cen-trifugal force. The concentrated samples were employed insubsequent experiments.

In Vitro Cytotoxicity Study. The biocompatibility ofOMVs is primordial for their application as a bioluminescentimaging tool. In this sense, understanding whether the OMVsare able to coexist with cells without promoting cell death is ofthe upmost importance prior to any preclinical assessment.The cytotoxicity of the OMVs was determined by treatingA549 cells with serial concentrations of the vesicles. A549 cellswere used as a model cell line for this study as they are fastgrowing, allowing easier detection of reduction in viability dueto the interference in cell replication or due to cell death.As shown in Figure 2, all concentrations of OMVs tested didnot lead to cytotoxicity after 4 h of incubation (p > 0.05 versuscontrol group), indicating that the OMVs do not promote anacute toxic effect to this cell line. It was also noted that therewas no reduction in cell viability after 24 h of incubation (p >0.05 versus control group), indicating no direct interferencewith the replication of this cell line.

In Vivo Local Absorption and Reaction Biolumines-cence Kinetics. For local absorption and reaction kineticsstudy, the bioluminescence generated by reaction of substrateand OMVs is observed quantitatively and qualitatively inFigures 3 and Figure 4, respectively. The maximum signal isachieved right after OMV injection, up to an impressive totalflux of 1.07 × 1010 p/s with 200 μL of OMV suspensioninjected. The substrate is consumed by the luciferase enzymesexpressed in the OMVs, and the second injection of thesubstrate is performed 5 h after the first injection. In this case,a significant reduction in maximum total flux is observed,which indicates a reduction of the luciferase available to consumethe substrate. Such a reduction can be associated with absorption

Figure 3. In vivo luminescence tracking of OMVs over time (localabsorption and reaction bioluminescence kinetics study). Substratewas subcutaneously injected (in vivo data) at 0 (1st) and 5 (2nd)hours upon the start of the experiment.

Figure 4. Bioluminescence tracking of animals treated subcutaneously with OMVs injected to the right flanks. Substrate injection was performed at0 (1st) and 5 (2nd) hours post injection of OMVs. Only the images taken within the first 120 min post each substrate injection are shown here.

Figure 2. Cell viability for different concentrations of OMVs uponincubation for 4 and 24 h, respectively.

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5611

or degradation of the vesicles since the second administrationof the substrate has the same dosage as the first one. Theincreased signal-to-noise ratio between the bioluminescencesignal and the luminescence background in mice highlights oneof the main advantages of bioluminescence optical imaging, asthere is low interference from the animals’ tissues.Reaction Kinetics and Absorption Model. Figure 5a

shows the comparison of modeling and experimental results oftransient luminescence radiant intensity for the in vivo kineticstest without a circulation clearance effect. The simulationparameters k2 are derived from the literature;30 mi and Δt areestimated from experimental settings; and initial concen-trations of substrate and enzyme are based on in vivo experi-mental data. Other parameters including the Michaelis−Menten constant KM and the decaying coefficients kdS0, kdE,and kdC are parameters adjusted to mimic the reaction andabsorption behaviors observed from the experimental data.The process is primarily dominated by the catalytic reaction,which is controlled by KM and the absorption of substrateswithin the tissue, phenomenologically represented by kdS0.Overall the physics-based lumped approximation has generatedreasonable and consistent results comparing the experimentalobservation. The proposed model predicts the longer timereaction and absorption process triggered by the additions ofsubstrates with reasonable trends.In Vivo Circulation and Reaction Bioluminescence

Kinetics. After the local absorption and reaction bio-luminescence kinetics study, we designed another experimentto mimic the circumstances in real applications. In this experi-ment, the OMVs were injected into the animal subcutaneously,and the substrate was injected through a tail vein. Comparedwith the local absorption reaction (Figure 3) and circulation

reaction studies (Figure 6), it is noticeable that the reductionin bioluminescence is much faster with circulation. Figure 6shows that the maximum signal of circulation reaction is∼1000-fold lower than that from the local absorption reactionstudy shown in Figure 5. The differences between the twoin vivo experiments demonstrated that only a very small fraction(∼0.1%) of the substrate has been transported to the OMVinjection site as estimated. As shown in Figures 6 and 7, thestrongest bioluminescence signal is detected quickly after thesubstrate injection which can be attributed to the rapid

Figure 5. (a) Comparison of modeling and experimental results on the bioluminescent intensity. The corresponding modeling results show the transientconcentrations of (b) substrate furimazine, (c) enzyme, and (d) complex, respectively. Parameters used: KM = 5 × 10−3 M, k2 = k−1 = 6.6 s−1, andthus k1 = 2.64 × 103 M−1 s−1, also kdS0 = 2 × 10−2 s−1, kdE = kdC = 4 × 10−5 s−1, m1 = m2 = 1.2167 × 10−5 M s−1, Δt = 60 s, S0 = 0 M, and E0 =1.04 × 10−10 M.

Figure 6. In vivo bioluminescence tracking (circulation and reactionbioluminescence kinetics study). Substrate injection was performedintravenously through the tail vein at 0 and 0.67 h post administrationof OMVs.

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5612

clearance in the body. For each injection, the bioluminescenceis tracked until it decayed to the similar level as the back-ground. The signal decay phase is around 30 min which can beexplained by the high furimazine elimination rate fromcirculation in mice.31

Furthermore, the reaction kinetics and absorption modelwith circulation clearance were also used to simulate the in vivobioluminescence kinetics study with blood circulation effect.Here, the simulation parameters k2, km, kdS0, and kdE are thesame as the previous model. Figure 8a shows the comparisonbetween experimental data and the modeling results using thesame reaction and sorption parameters. The overall signaldecaying behavior and the time scale shown in the modelingresults agree with the experimental data, but the amplitude ofthe bioluminescence flux overpredicts the observed results.This is possibly due to the fact that the actual concentrationdelivered to the targeted site may be lower than the value used

in the model. Additionally, compared with the modeling resultof local absorption and reaction (Figure 5), no obvious enzymechange is exhibited at the injection points in simulation ofbioluminescence kinetics with blood circulation effect, whichmight be attributed to the small amount of substrate reachingthe administration site of OMVs. In this animal model, pre-sumably the consumption of enzyme is mainly caused by tissueabsorption, since substrate concentration in the reaction site issignificantly lower than in the case where the substrate isdirectly injected to the administration site of OMVs.

In Vivo Biocompatibility Study of OMVs. In order tostudy the biocompatibility of OMVs, the tissue samples frommajor organs and injection location (subcutaneous tissue) ofthe experimental animals were collected for histology analysis.As shown in Figure 9, mild hyperkeratosis in some hair folliclesof the skin was observed for the mice treated with OMVs, butthis was within the limits of normal for nude mice and in

Figure 7. Bioluminescence tracking of animals treated subcutaneously with OMVs injected to the right flanks. Substrate injection was performedintravenously through the tail vein at 0 (1st) and 0.67 (2nd) hours post injection of OMVs.

Figure 8. (a) Comparison of modeling and experimental results on the normalized bioluminescent intensity. The corresponding modeling resultsshow the transient concentrations of (b) substrate furimazine, (c) enzyme, and (d) complex, respectively.

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5613

accordance with the control group. No lesions were found inliver, spleen, and kidney tissue samples. The histology analysisresult demonstrated the good biocompatibility of OMVs dur-ing the full time course of our experiments, which typicallyspanned 48 h (from the injection of OMVs to the sacrifice ofmice for histology analysis).

■ CONCLUSIONSIn summary, we demonstrate that engineering multifunctionalouter membrane vesicles can be used for bioimaging based ontheir unique properties. The low cytotoxicity and good bio-compatibility are confirmed by cytotoxicity tests and ex vivohistology analysis. Two in vivo experiments were designed tostudy the bioluminescence kinetics. The circulation andreaction kinetics study mimics the real application of OMVs.The strong signal detected from the experimental animalsindicates that these vesicles have great potential for bioimagingapplications. Moreover, the intensities of bioluminescence pro-duced by outer membrane vesicles in both local absorptionreaction and circulation reaction studies are tracked and com-pared. The mathematical models are developed to connect thebioluminescence signal with the local absorption reaction andcirculation reaction kinetics and have successfully describedthe in vivo experimental results.

■ AUTHOR INFORMATIONCorresponding Authors*Tel.: +1 860 486 4554. Fax: +1 860 486 2959. E-mail:yu.lei@uconn.edu*E-mail: tai-hsi.fan@uconn.edu.ORCIDDonghui Song: 0000-0001-9081-4275Wilfred Chen: 0000-0002-6386-6958Yu Lei: 0000-0002-0184-0373Author Contributions#Y.H. and A.O.B. contributed equally.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe greatly appreciate the funding from NSF (CBET1604925and CBET1604826). We also thank Professor Yangchao Luo’sgroup for help using the Zetasizer Nano ZS. This work waspartially supported by a fellowship grant from GE’s Industrial

Solutions Business Unit under a GE-UConn partnership agree-ment. The views and conclusions contained in this documentare those of the authors and should not be interpreted asnecessarily representing the official policies, either expressed orimplied, of Industrial Solutions or UConn.

■ REFERENCES(1) Fais, S.; O’Driscoll, L.; Borras, F. E.; Buzas, E.; Camussi, G.;Cappello, F.; Carvalho, J.; Cordeiro da Silva, A.; Del Portillo, H.; ElAndaloussi, S.; Ficko Trcek, T.; Furlan, R.; Hendrix, A.; Gursel, I.;Kralj-Iglic, V.; Kaeffer, B.; Kosanovic, M.; Lekka, M. E.; Lipps, G.;Logozzi, M.; Marcilla, A.; Sammar, M.; Llorente, A.; Nazarenko, I.;Oliveira, C.; Pocsfalvi, G.; Rajendran, L.; Raposo, G.; Rohde, E.;Siljander, P.; van Niel, G.; Vasconcelos, M. H.; Yanez-Mo, M.;Yliperttula, M. L.; Zarovni, N.; Zavec, A. B.; Giebel, B. Evidence-Based Clinical Use of Nanoscale Extracellular Vesicles in Nano-medicine. ACS Nano 2016, 10 (4), 3886−99.(2) Ma, X.; Hargrove, D.; Dong, Q.; Song, D.; Chen, J.; Wang, S.;Lu, X.; Cho, Y. K.; Fan, T.-H.; Lei, Y. Novel green and redautofluorescent protein nanoparticles for cell imaging and in vivobiodegradation imaging and modeling. RSC Adv. 2016, 6 (55),50091−50099.(3) Ma, X.; Sun, X.; Chen, J.; Lei, Y. Natural or natural-synthetichybrid polymer-based fluorescent polymeric materials for bio-imaging-related applications. Appl. Biochem. Biotechnol. 2017, 183(2), 461−487.(4) Selvan, S. T.; Tan, T. T. Y.; Yi, D. K.; Jana, N. R. Functional andmultifunctional nanoparticles for bioimaging and biosensing.Langmuir 2010, 26 (14), 11631−11641.(5) Xu, B.; Zhao, C.; Wei, W.; Ren, J.; Miyoshi, D.; Sugimoto, N.;Qu, X. Aptamer carbon nanodot sandwich used for fluorescentdetection of protein. Analyst 2012, 137 (23), 5483−5486.(6) Xu, J.; Tehrani, K. F.; Kner, P. Multicolor 3D super-resolutionimaging by quantum dot stochastic optical reconstruction microscopy.ACS Nano 2015, 9 (3), 2917−2925.(7) Song, Y.; Feng, D.; Shi, W.; Li, X.; Ma, H. Parallel comparativestudies on the toxic effects of unmodified CdTe quantum dots, goldnanoparticles, and carbon nanodots on live cells as well as green gramsprouts. Talanta 2013, 116, 237−244.(8) Stremersch, S.; De Smedt, S. C.; Raemdonck, K. Therapeuticand diagnostic applications of extracellular vesicles. J. ControlledRelease 2016, 244, 167−183.(9) Wyss, R.; Grasso, L.; Wolf, C.; Grosse, W.; Demurtas, D.; Vogel,H. Molecular and dimensional profiling of highly purified extracellularvesicles by fluorescence fluctuation spectroscopy. Anal. Chem. 2014,86 (15), 7229−7233.(10) Furi, I.; Momen-Heravi, F.; Szabo, G. Extracellular vesicleisolation: present and future. Annals of translational medicine 2017, 5(12), 263.(11) Klimentova, J.; Stulík, J. Methods of isolation and purificationof outer membrane vesicles from gram-negative bacteria. Microbiol.Res. 2015, 170, 1−9.(12) Aline, F.; Bout, D.; Amigorena, S.; Roingeard, P.; Dimier-Poisson, I. Toxoplasma gondii antigen-pulsed-dendritic cell-derivedexosomes induce a protective immune response against T. gondiiinfection. Infection and immunity 2004, 72 (7), 4127−4137.(13) Zocco, D.; Ferruzzi, P.; Cappello, F.; Kuo, W. P.; Fais, S.Extracellular vesicles as shuttles of tumor biomarkers and anti-tumordrugs. Front. Oncol. 2014, 4, 267.(14) Armstrong, J. P.; Holme, M. N.; Stevens, M. M. Re-engineeringextracellular vesicles as smart nanoscale therapeutics. ACS Nano 2017,11 (1), 69−83.(15) Gujrati, V.; Kim, S.; Kim, S. H.; Min, J. J.; Choy, H. E.; Kim, S.C.; Jon, S. Bioengineered bacterial outer membrane vesicles as cell-specific drug-delivery vehicles for cancer therapy. ACS Nano 2014, 8(2), 1525−37.(16) Chen, D. J.; Osterrieder, N.; Metzger, S. M.; Buckles, E.;Doody, A. M.; DeLisa, M. P.; Putnam, D. Delivery of foreign antigens

Figure 9. Histological staining of skin tissue and major organs fromOMV-treated mice. Representative H&E stained slices for (a) skintissue, (b) liver, (c) kidney, and (d) spleen, respectively.

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5614

by engineered outer membrane vesicle vaccines. Proc. Natl. Acad. Sci.U. S. A. 2010, 107 (7), 3099−104.(17) Stevenson, T. C.; Cywes-Bentley, C.; Moeller, T. D.; Weyant,K. B.; Putnam, D.; Chang, Y. F.; Jones, B. D.; Pier, G. B.; DeLisa, M.P. Immunization with outer membrane vesicles displaying conservedsurface polysaccharide antigen elicits broadly antimicrobial antibodies.Proc. Natl. Acad. Sci. U. S. A. 2018, 115 (14), E3106−E3115.(18) Gospodarowicz, D.; Moran, J. S. Growth factors in mammaliancell culture. Annu. Rev. Biochem. 1976, 45, 531−58.(19) Ma, X.; Sun, X.; Hargrove, D.; Chen, J.; Song, D.; Dong, Q.;Lu, X.; Fan, T. H.; Fu, Y.; Lei, Y. A Biocompatible and BiodegradableProtein Hydrogel with Green and Red Autofluorescence: Preparation,Characterization and In Vivo Biodegradation Tracking and Modeling.Sci. Rep. 2016, 6, 19370.(20) Ma, X.; Wang, T.; Song, D.; Hargrove, D.; Dong, Q.; Luo, Z.;Chen, J.; Lu, X.; Luo, Y.; Fan, T.-H. Protein microspheres with uniquegreen and red autofluorescence for noninvasively tracking andmodeling their in vivo biodegradation. ACS Biomater. Sci. Eng.2016, 2 (6), 954−962.(21) Stacer, A. C.; Nyati, S.; Moudgil, P.; Iyengar, R.; Luker, K. E.;Rehemtulla, A.; Luker, G. D. NanoLuc reporter for dual luciferaseimaging in living animals. Mol. Imaging 2013, 12 (7), 1−13.(22) Schaub, F. X.; Reza, M. S.; Flaveny, C. A.; Li, W.; Musicant, A.M.; Hoxha, S.; Guo, M.; Cleveland, J. L.; Amelio, A. L. Fluorophore-NanoLuc BRET Reporters Enable Sensitive In Vivo Optical Imagingand Flow Cytometry for Monitoring Tumorigenesis. Cancer Res.2015, 75 (23), 5023−33.(23) Rathbun, C. M.; Prescher, J. A. Bioluminescent Probes forImaging Biology beyond the Culture Dish. Biochemistry 2017, 56(39), 5178−5184.(24) England, C. G.; Ehlerding, E. B.; Cai, W. NanoLuc: A SmallLuciferase Is Brightening Up the Field of Bioluminescence.Bioconjugate Chem. 2016, 27 (5), 1175−1187.(25) Hall, M. P.; Unch, J.; Binkowski, B. F.; Valley, M. P.; Butler, B.L.; Wood, M. G.; Otto, P.; Zimmerman, K.; Vidugiris, G.; Machleidt,T.; Robers, M. B.; Benink, H. A.; Eggers, C. T.; Slater, M. R.;Meisenheimer, P. L.; Klaubert, D. H.; Fan, F.; Encell, L. P.; Wood, K.V. Engineered luciferase reporter from a deep sea shrimp utilizing anovel imidazopyrazinone substrate. ACS Chem. Biol. 2012, 7 (11),1848−57.(26) Huang, Y.; Haomin, L.; Chen, W.; Mu-Ping, N.; Lei, Y.Genetically Engineered Bio-nanoparticles with Co-expressed EnzymeReporter and Recognition Element for IgG Immunoassay. Sensors andActuators Reports 2019.(27) Chen, Q.; Rozovsky, S.; Chen, W. Engineering multi-functionalbacterial outer membrane vesicles as modular nanodevices forbiosensing and bioimaging. Chem. Commun. (Cambridge, U. K.)2017, 53 (54), 7569−7572.(28) Chen, Q.; Rozovsky, S.; Chen, W. Engineering multi-functionalbacterial outer membrane vesicles as modular nanodevices forbiosensing and bioimaging. Chem. Commun. 2017, 53 (54), 7569−7572.(29) Segel, L. A.; Slemrod, M. The quasi-steady-state assumption: acase study in perturbation. SIAM Rev. 1989, 31 (3), 446−477.(30) Suzuki, K.; Kimura, T.; Shinoda, H.; Bai, G.; Daniels, M. J.;Arai, Y.; Nakano, M.; Nagai, T. Five colour variants of brightluminescent protein for real-time multicolour bioimaging. Nat.Commun. 2016, 7, 13718.(31) Shipunova, V.; Shilova, O.; Shramova, E.; Deyev, S.; Proshkina,G. A highly specific substrate for NanoLUC luciferase furimazine istoxic in vitro and in vivo. Russ. J. Bioorg. Chem. 2018, 44 (2), 225−228.

ACS Applied Bio Materials Article

DOI: 10.1021/acsabm.9b00690ACS Appl. Bio Mater. 2019, 2, 5608−5615

5615