Reactive polymer enables efficient in vivobioorthogonal chemistryNeal K. Devaraj1, Greg M. Thurber1, Edmund J. Keliher, Brett Marinelli, and Ralph Weissleder2

Center for Systems Biology, Massachusetts General Hospital, Harvard Medical School, Simches Research Building, 185 Cambridge Street, Boston,MA 02114

Edited by David A. Tirrell, California Institute of Technology, Pasadena, CA, and approved January 31, 2012 (received for review August 16, 2011)

There has been intense interest in the development of selectivebioorthogonal reactions or “click” chemistry that can proceed inlive animals. Until now however, most reactions still require vastsurpluses of reactants because of steep temporal and spatial con-centration gradients. Using computational modeling and design ofpharmacokinetically optimized reactants, we have developed apredictable method for efficient in vivo click reactions. Specifically,we show that polymer modified tetrazines (PMT) are a key enablerfor in vivo bioorthogonal chemistry based on the very fast andcatalyst-free [4þ 2] tetrazine/trans-cyclooctene cycloaddition.Using fluorescent PMT for cellular resolution and 18F labeled PMTfor whole animal imaging, we show that cancer cell epitopes canbe easily reacted in vivo. This generic strategy should help guidethe design of future chemistries and find widespread use for dif-ferent in vivo bioorthogonal applications, particularly in the biome-dical sciences.

in vivo chemistry ∣ pharmacokinetics ∣ PET imaging ∣ intravital microscopy ∣pretargeting

The ability to perform selective chemistries in living systemssuch as single cells, 3D cultures, invertebrates, or mammals

would have far reaching applications in tracking biomolecules,designing new therapeutic approaches, and in visualizing medi-cally relevant biomarkers. To date, only a few practical bioortho-gonal reactions have been reported, the most popular being theStaudinger ligation and the [3þ 2] cycloaddition “click” reactionbetween azides and alkynes (1, 2). The latter click reactioninvolves copper(I) catalyzed coupling of an azide and terminalalkyne to generate a stable triazole (2). Until recently, the neces-sity of the copper catalyst precluded use of this reaction in bio-logical systems due to toxicity concerns (3, 4). Bertozzi and otherselegantly solved this problem by developing several new ringstrained dienophile derivatives that do not require catalysts (5–9).However, many of these derivatives have low water solubility, re-quire complex multistep synthesis, and possess suboptimal kinetics.Our search for alternative rapid, selective, and chemically acces-sible coupling reactions without need for a catalyst led us andothers to investigate the [4þ 2] inverse Diels-Alder cycloaddition(10–13). We realized that this set of chemistries is more uniquelysuited to biological applications and may indeed represent a uni-versal platform technology (Fig. 1 A and B). Specifically we andothers have shown that the cycloaddition between tetrazine (Tz)and trans-cyclooctene (TCO) can proceed orders of magnitude fas-ter than previously studied azide and alkyne click reactions andimportantly does not require the action of a catalyst. This reactionhas also been adapted for single cell imaging using newer fluoro-genic Tz probes (14).

Although initial work focused on in vitro labeling there hasbeen a surge of recent work applying this reaction for variousin vivo applications (15–17). Despite the above progress, ourresults with initial TCO/Tz reactions in live animals were disap-pointing unless we used excessive amounts of reactants. Thisbrute force approach would ultimately be cost prohibitive andlikely not be biologically viable. We therefore systematicallyeliminated different reasons for the observed discrepancies

(stability, bioconversion of reactants, target delivery, etc.) andidentified simple pharmacokinetic principles as the primary rea-son why the reaction did not proceed at target sites as originallyanticipated. We then performed detailed studies on how the re-action rate and pharmacokinetics (PK) can affect the efficiency ofan in vivo chemical reaction. Using reactants with modified PKprofiles, such as polymer modified tetrazines [(PMT); Fig. 1C]and a newly developed computational model to predict reactionefficiency, we now observe highly efficient in vivo bioorthogonalreaction. Using cancer cell epitopes as a model target (Fig. 1D),we show the specificity and sensitivity of the method in a mousemodel of colon cancer. Specifically, we observe in vivo labelingwith cellular resolution using near infrared fluorophores.Furthermore, we demonstrate use of bioorthogonal reactionswith rapidly decaying positron emission tomography (PET) iso-topes for imaging cancer cell epitopes in solid tumors. We believeour work and findings will guide further development of in vivochemistries for a number of different biomedical applications.

ResultsWe first compared different reactants in mice to better under-stand achievable in vivo click reaction efficiencies as a functionof PK profiles (Table 1). These studies were performed as serialblood draws, single time point biodistribution studies, or longitu-dinal imaging studies at different resolutions. In a first set of ex-periments, mice were injected with 30 micrograms of anti-CD45monoclonal antibodies labeled with TCO or norbornene dieno-philes. The goal was to provide a reactive target in the plasma toavoid the additional complexities of delivering agents to extravas-cular sites. The dienophile labeling was controlled so on averageeach antibody possessed approximately three reactive groups.After waiting 3 h, an excess of tetrazine chaser was administered.Because the pharmacokinetics of the chaser would likely affectthe efficiency of labeling, we administered either a small mole-cular tetrazine fluorophore (MW 1.3 kDa) or polymer modifiedtetrazines (PMT) using two differently sized dextrans: 10 kDa(PMT10) and 40 kDa (PMT40). Dextrans were chosen as poly-mer scaffold due to their low cost, high stability, and numerousprevious in vivo applications (18–20). These amine modifiedscaffolds were modified on average with approximately one fluor-ochrome and either two (PMT10) or eleven (PMT40) tetrazinesas measured using absorption spectroscopy. Such high couplingefficiencies to amino dextran may be the result of the low mole-cular weight and lipophilic nature of tetrazine affinity probes.Measurement of reaction kinetics demonstrated that conjugationcombined with multivalent presentation did not significantly

Author contributions: N.K.D., G.M.T., and R.W. designed research; N.K.D., G.M.T., E.J.K.,and B.M. performed research; E.J.K. contributed new reagents/analytic tools; N.K.D.,G.M.T., and R.W. analyzed data; and N.K.D., G.M.T., and R.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1N.K.D. and G.M.T. contributed equally to this work.2To whom correspondence should be addressed. E-mail: rweissleder@mgh.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1113466109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1113466109 PNAS Early Edition ∣ 1 of 6

MED

ICALSC

IENCE

SCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0

impede overall probe reactivity compared to small molecule tet-razines (see SI Text). The clearance kinetics of these chaser ima-ging agents were measured separately through serial blood draws(Fig. 2A). The tetrazine fluorophore is a low molecular weightand highly charged probe, and thus it is not surprising that thePK showed rapid tissue redistribution and clearance (21). In con-trast, the PMT’s persisted in the blood and exhibited biexponen-tial clearances with the larger molecular weight dextran PMT40circulating longer.

Fig. 2B shows the labeling of dienophile modified whiteblood cells, measured by flow cytometry, 3 h after administrationof a fluorescent tetrazine chaser. The labeling efficiency isstrongly dependent on the kinetics of the reaction (e.g., trans-cyclooctene ≫ norbornene) and the clearance kinetics of the sec-ondary agent. TCO/Tz reaction efficiencies were very high forboth the fluorescent PMT10 and PMT40 (680 nm). However,tetrazine-VT680 TCO reactions showed a much lower labelingefficiency, presumably due to its much faster clearance rate. Like-wise, targeting PMT40 to a slow reacting norbornene antibodyled to nearly negligible reaction. In order to gain greater quanti-tative insight, computational modeling of the in vivo reaction wasperformed (seeMaterials and Methods). A plot of predicted label-ing efficiency vs. both the kinetics of cycloaddition reaction andthe area under the concentration curve (AUC) is displayed inFig. 2C. The fraction of primary agent reacted is an exponentialfunction of the reaction rate and AUC of the secondary agent.Small molecule tetrazines, while possessing rapid reaction rates,do not possess the requisite AUC for very efficient labeling, ex-plaining our experimental results.

Based on the results from the initial in vivo experiments, wedecided to use PMT10 to demonstrate the utility of our methodfor imaging cancer cell epitopes in a more challenging but clini-cally relevant solid tumor model. PMT10 showed good clearanceafter several hours yet persisted long enough to allow highly effi-cient labeling using the TCO/Tz reaction. As a disease model,we targeted the A33 glycoprotein which is overexpressed in >95%of all human colorectal cancers including the human colorectal

Fig. 1. In Vivo Bioorthogonal Reactions. (A) Tetrazine cycloaddition with trans-cyclooctene forming a dihydropyrazine. (B) Schematic of PMTused in this study.The scaffold consists of dextran that has been aminated to allow attachment of tetrazine reactive groups as well as imaging agents such as near-infraredfluorophores and radioisotopes. (C) In vivo multistep delivery of imaging agent. A slow clearing targeting agent is administered first (green) and is given 24 hfor localization and background clearance. Next, a lower molecular weight secondary agent (red) is delivered that rapidly reacts and is cleared from thebackground tissue much faster than the primary agent. (D) Kinetic parameters of consideration for in vivo clicking. The secondary tetrazine agent reactswith transcyclooctene antibodies at a given rate (kreaction). This rate is in competition with other rates including the clearance of the secondary agent fromthe body (kclearance) and internalization of the antibody (kendocytocis).

Table 1. Reported Bioorthogonal Rate Constants andPharmacokinetic Parameters

Reaction type Rate constant (∕M·s) Reference

azide-DIFO 0.076 (CH3CN) (7)phen-tetrazine-

norbornene1.6 (H2O, 20 °C) (11)

pyr-tetrazine-TCO 2,000 (9∶1 MeOH∕H2O, 25 °C) (12)phen-tetrazine-TCO 6,000 (H2O, 37 °C) (10)pyr-tetrazine-TCO 13,090 (H2O, 37 °C) (15)pyr-tetrazine-

strained-TCO22,000 (MeOH, 25 °C) (13)

Molecule AUC at 3 h (2 μM) Reference

Tetrazine-VT680 4.85 × 10−5 M·s measuredDextran, 10 kDa 6.93 × 10−4 M·s measuredDextran, 40 kDa 7.66 × 10−3 M·s measuredMAb 2.07 × 10−2 M·s (28)DOTA 2.05 × 10−4 M·s (21)

2 of 6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1113466109 Devaraj et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0

cancer LS174Tcell line (22). In addition, this antigen is associatedwith tight junction proteins, giving it a surface persistence with ahalf life greater than 2 d (23).

Initially we reacted near-infrared fluorescent PMT imagingagents to obtain cellular resolution of the in vivo bioorthogonalreaction. In addition, fluorescent agents avoid the cost and hand-ling complexities of radionuclides, and are increasingly beingused for in vivo applications (24). To facilitate the in vivo imaging,we used dorsal window chambers for intravital microscopy. Win-dow chamber mice bearing LS174T tumors were injected firstwith anti-A33 monoclonal antibodies bearing TCO (n ¼ 3)dienophiles (25). After waiting 24 h for antibody targeting andclearance, the mice were subsequently injected with 30 μg offluorescent PMT10 (750 nm). After waiting 3 h to allow for thereaction and clearance to occur, the tumors were visualized usingconfocal microscopy. The images revealed beds of LS174T cellswhose surfaces were brightly stained with the fluorescent anti-body as well as the tetrazine dextran probe (Fig. 3). Merging ofthe two channels demonstrated excellent colocalization betweenthe location of the dienophile antibody and the secondary tetra-zine dextran, indicative of a specific bioorthogonal reaction. Con-trol experiments involving mice which received an antibodylacking a dienophile also showed significant antibody stainingof LS174T cell surfaces but lacked subsequent tagging by thetetrazine dextran. These in vivo images and controls looked re-markably similar to corresponding in vitro experiments. These in-itial studies demonstrated that use of a higher molecular weight

PMT10 agent, previously validated in the blood, is also able tolabel specific cell surface biomarkers in the tumor microenvir-onment.

To demonstrate the utility of our method for clinically relevantPET imaging we performed longitudinal studies using a radiola-beled 18F-PMT10 (26). Mice were implanted subcutaneouslywith LS174T xenografts, and once tumors had grown to approxi-mately 1 cm in diameter, these mice were injected with 30 μg ofTCO-antibody against A33 (also labeled with VT680 fluoro-phore). After allowing 24 h for accumulation and clearance ofthe TCO-modified antibody, 18F-PMT10 was injected (30 μg;150 μCi). Cycloaddition of 4 mCi of TCO-18F to 60 μg of dextranproceeded cleanly, leading to an 89.2% decay corrected radioche-mical yield of 18F-PMT10. This construct was used for all PET/Computed Tomography (CT) and subsequent autoradiographyexperiments. PET/CT scans were initiated 3 h after injectionof 18F-PMT10. The images revealed significant renal clearanceof PMT10 as well as widespread tissue distribution. Fig. 4A showsone representative axial slice from the PET/CT reconstruction.Compared to control mice which were administered antibodieslacking TCO, there was significant accumulation of signal inthe vascularized region of the tumor mediated by the in vivoTCO/Tz reaction. To investigate further, mice were euthanized3 h after injection, the tumors were excised, and 1 mm slices wereimaged to determine the extent of cycloaddition. Near infraredfluorescence was used to determine antibody/TCO localizationand autoradiography was used to reveal the 18F-PMT10. Fig. 4Bshows two representative tumor slices. The localization patternsof the chemical probes were distinct and often showed greatestactivity in the periphery of the tumor, likely reflecting the necroticnature of the tumor core. Comparison of the pattern to the lo-calization of the fluorescent antibody demonstrated good coloca-lization, indicating that the Tz probe was primarily localized tothe site of TCO binding. Tumors from mice which were adminis-tered the control antibody lacking reactive TCO showed muchlower uptake of Tz and a lack of correlation to antibody/TCOfluorescence signal.

We also tested for biomarker specificity with mice simulta-neously implanted with LS174T tumors and A431 tumors whichlack expression of the A33 glycoprotein epitope (27). Mice wereadministered anti-A33 TCO antibody and 18F-PMT10 as pre-viously described and imaged using PET/CT. Fig. 4C shows onerepresentative coronal slice from the PET/CT reconstruction.LS174T tumors can be clearly visualized whereas the controlA431 tumor shows much lower Tz uptake. Autoradiography ofexcised tumor slices further demonstrate the greater uptake ofTz in biomarker positive LS174T tumors vs. the control A431tumors (Fig. 4D).

DiscussionThis work demonstrates that bioorthogonal TCO/Tz reactionscan efficiently occur in live animals when reaction partner phar-macokinetics are optimized. Based on modeling of reaction ki-netics and PK clearance rates, we designed and tested specificpolymer-tetrazine adducts bearing “imaging” reporters (near in-frared fluorochromes and 18F) to visualize reactions in live miceboth at the cellular and whole animal level. The current genera-tion of polymer-tetrazine conjugates included dextran derivatives(PMT10 and PMT40) because dextrans have a long history of invivo use and the PK are well understood (18–20). Based on thekinetics of the TCO component, we would envision that the PMTcould be based on a number of different polymers for furtheroptimization (e.g., PEGs, PL, PLGA, nanoparticles, etc.). Weexpect that the methodology used for developing these probesand modulating the reaction rate and clearance kinetics shouldbe applicable to other bioorthogonal reactions, imaging agents,and therapeutics.

Fig. 2. Optimization of In Vivo Chemistry with Polymer Modified Tetrazines(PMT). (A) Clearance kinetics of tetrazine-VT680 (red), PMT10 (blue) andPMT40 (green). (B) Efficiency of in vivo labelling of trans-cyclooctene mod-ified leukocytes using various tetrazine cycloadditions. (C) Heat map showingpredicted efficiency of reaction given a reaction rate and secondary clearancerate. The 1.3 kDa point represents the small molecule tetrazine-VT680. Thehighest efficiencies were observed for the fast tetrazine/trans-cyclooctenereaction using slow clearing PMT10 or PMT40. Use of either a faster clearingtetrazine or a much slower reaction greatly diminishes the efficiency of re-action as seen in plot in (B).

Devaraj et al. PNAS Early Edition ∣ 3 of 6

MED

ICALSC

IENCE

SCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0

The TCO/Tz reaction is an ideal bioorthogonal chemicalsystem for in vivo use given its fast reactions kinetics(>6;000 M−1 sec−1)(10). Despite these impressive kinetics forbenzylamine Tz/TCO, PK modifications are required for in vivouse. These improved PK properties can theoretically be achievedby small molecule tetrazine modifications, such as those thatwould alter pharmacokinetics without affecting reaction rates

such as side groups with protein binding abilities. In this respectwe have recently looked at reaction rates of a number of differentTZ analogs. However, although binding to proteins would de-crease blood clearance, it was unclear if the protein bound Tzwould be viable for chemical reaction with TCO targets. Instead,to improve PK of the reactants we decided to conjugate Tz todextran creating PMTof different chain lengths but with identical

Fig. 3. In Vitro and In Vivo Microscopy of Fluorescent PMT10 Targeting a Cancer Cell Epitope. Confocal imaging demonstrating targeting of PMT10 (750 nm)to trans-cyclooctene (TCO) antibodies (680 nm) on the surface of LS174T cells both in vitro and in vivo. (A) LS174T cells in culture labeled with TCO monoclonalantibodies followed by PMT10 (750 nm) labeling. (B) Same as previous except monoclonal antibodies lack TCO. (C) Cells in a LS174T xenograft in vivo labeledwith TCO monoclonal antibodies followed by PMT10 (750 nm) labeling. (D) Same as previous except monoclonal antibodies lack TCO. See Materials andMethods for experimental details.

Fig. 4. PETand Autoradiography using 18F Tetrazine Agents. (A) PET/CT fusion of LS174T tumor xenograft labeled using either trans-cyclooctene (TCO) mono-clonal antibodies (MAb TCO) or control unlabeled antibodies (MAb) followed by 18F-PMT10. Arrows indicate location of the tumor xenograft. Bladder has beenomitted for clarity. (B) Imaging using autoradiography (left side) and fluorescence reflectance (right side) of 1 mm LS174T tumor slices after targeting withfluorescent TCO monoclonal antibody and 18F-PMT10. (C) PET/CT fusion of mouse bearing A431 and LS174T tumors after targeting with anti-A33 TCO mono-clonal antibodies followed by 18F-PMT10. Arrows indicate location of tumors and the liver has been omitted for clarity. (D) Autoradiography of representative1 mm LS174T or A431 tumor slices after multistep targeting.

4 of 6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1113466109 Devaraj et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0

reactivities to decouple and independently tune the clearance andreaction rates.

We developed a simple mathematical model that takes TCO/Tz reaction rates and pharmacokinetics of reaction partners intoaccount (Fig. 2) to better predict the ideal design and dosages ofreaction partners for in vivo use. The model predicts that the sec-ondary agent must remain at a high enough concentration for asufficient time to undergo the reaction in vivo. These experimen-tal and computational results allow approximation of labelingefficiency in vivo given a chemical reaction rate and secondaryagent clearance rate when the secondary agent is administeredin excess. These conditions are commonly encountered when la-beling surface exposed molecules and is directly comparable tothe experiments targeting cell surface antigens in the blood. Forother targets, such as solid tumors, the concentration profile ofthe primary and secondary agents are more complex given theblood flow, extravasation, diffusion, binding, and clearance ratesfound in tumors (28), but the basic principles still apply. We be-lieve these insights are not only important for our study but mayoffer helpful guidance to future studies where one is interested inperforming in vivo reactions.

We envision a number of practical applications of the devel-oped in vivo click approach based on TCO/PMTand other smallmolecule probes. For instance, the short positron decay half-life (1.8 h) of 18F requires affinity ligands that show significanttarget accumulation and rapid background clearance in hours.These results are simply not possible by directly labeling highAUC affinity agents such as monoclonal antibodies. Multisteptechniques with bioorthognal chemistries could overcome thisproblem. Additionally, in vivo assembly could be used to improvetherapeutic drug delivery by delivering components as small mo-lecules to facilitate transport (e.g., across the blood brain barrier,blood vessel endothelium, intestinal wall) followed by specificcovalent assembly. The technique may also provide informationon the concentration of drugs or metabolic markers labeled withchemical tags in different tissue compartments. Finally, as wehave demonstrated in this work, in vivo chemistry shows greatpromise for imaging biomarkers and targeting agents with rapidlydecaying isotopes.

Materials and MethodsMeasurement of Clearance Kinetics of Tetrazine Agents. Fluorescent tetrazineagents (tetrazine-VT680, PMT10, PMT40) were injected via tail vein into mice.Serial retroorbital 30 μL blood draws were taken starting at 1 min after in-jection and periodically up to 2 h. The blood was diluted in equal volume5 mM EDTA in phosphate buffered saline, and the fluorescence intensitywas measured using a plate reader. The concentration of the fluorescentagent in the blood was determined by comparison with a standard curveusing blood mixtures with known concentrations of fluorescent agent.

Modeling of Reaction Efficiency. The efficiency of tetrazine imaging agent re-action with trans-cyclooctene modified antibodies on the surface of whiteblood cells was modeled using simple mass action kinetics. Assuming amouseplasma volume of approximately 2 mL, the initial antibody bolus dose shouldresult in an antibody plasma concentration of 100 nM. After 24 h, the plasmaconcentration will drop due to clearance (28) and potentially from uptake bytarget white blood cells (29). After clearance, the tetrazine probe is deliveredat an initial concentration of 2 μM, well in excess of the antibody plasma con-centration. Given the large excess of tetrazine probe, the loss in tetrazine dueto reaction is considered negligible. The beta phase of antibody clearance hasa 7 d half life, so, after 24 h, antibody clearance is assumed to be negligibleover 3 h. Because the reaction occurs in the plasma, no tissue transport para-meters were required.

The mass balance for antibody concentration is therefore:

d½Ab�dt

¼ −krxn½Ab�½tet�;

where ½Ab� is the trans-cyclooctene modified antibody concentration, krxn isthe second order reaction rate constant, and ½tet� is the concentration of tet-razine probe.

The highly dispersive nature of the mouse vascular system is likely torapidly result in homogenous distribution of reactive agents in the plasma,so a biexponential decay model is used for plasma concentration:

½tet� ¼ ½tet�0ðAe−kα ·t þ Bekβ·tÞ;where ½tet�0 is the initial plasma concentration of tetrazine, A and B are thefractions of alpha vs. beta phase clearance, t is the time after injection of thetetrazine agent, and kα and kβ are the alpha and beta phase clearance rateconstants, respectively. It is important to capture both phases in measure-ments, particularly for imaging agents that are given as a single bolus dose.While small molecular weight agents may have beta (i.e., clearance) phases ofseveral minutes or longer, this often is preceded by a rapid drop in concen-tration, which can be greater than an order of magnitude, with a half life oftens of seconds (21).

Because the tetrazine concentration is assumed independent from thereaction:

d½Ab�dt

¼ −krxn½Ab�½tet�0ðAe−kα ·t þ Be−kβ ·tÞ:

Integrating:

Z ½Ab�final

½Ab�0

d½Ab�½Ab� ¼ −krxn½tet�0

Zt

0

ðAe−kα ·t þ Be−kβ ·tÞdt

½Ab�final ¼ ½Ab�0 exp�−krxn½tet�0

�Akα

ð1 − e−kα·tÞ

þ Bkβ

ð1 − e−kβ·tÞ��

fraction reacted ¼ 1 − exp�−krxn½tet�0

�Akα

ð1 − e−kα ·tÞ

þ Bkβ

ð1 − e−kβ ·tÞ��

:

The integral of the tetrazine concentration-time curve is commonly definedas the area under the curve or AUC:

AUCtet ≡Z

t

0

½tet�0ðAe−kα ·t þ Be−kβ ·tÞdt

AUCtet ¼ ½tet�0�Akα

ð1 − e−kα·tÞ þ Bkβ

ð1 − e−kβ ·tÞ�:

Substituting into the equation for fraction of TCO reacted:

fraction reacted ¼ 1 − expf−krxnAUCtetg:This equation is graphed as a heat map in Fig. 2C.

Measurement of Reaction Efficiency on Circulating White Blood Cells. Dieno-phile modified anti-CD45 antibody was injected into mice (n ¼ 3) via tail vein.After 1 h, a 30 μL retroorbital blood sample was taken from each mouse andexposed to an excess of tetrazine agent to determine the saturated labelingdensity. The tetrazine agent was then administered via tail vein into themice, and after 3 h, a second retroorbital blood sample was taken. The bloodsamples were exposed to 1 mL BD pharmlyse solution for 15 min at roomtemperature, centrifuged, and the pellet resuspended in order to removelysed red blood cells. The white blood cells were run on a flow cytometerto measure the average fluorescent intensity. The in vivo intensity measure-ments were then divided by the in vitro saturated intensity in order to de-termine the labeling efficiency.

Fluorescence Imaging of Tetrazine Cycloadditions In Vitro and In Vivo. LS174Tcells were trypsinized, plated on cover glass dishes, and allowed to attachovernight. Cells were then exposed to 100 nM of TCO modified anti-A33modified with a VT680 probe. After washing, the cells were incubated with

Devaraj et al. PNAS Early Edition ∣ 5 of 6

MED

ICALSC

IENCE

SCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0

100 nM of fluorescent PMT10 (750 nm). The reaction was allowed to proceedfor 45 min after which the cells were washed and imaged using scanningconfocal microscopy. Control cells were treated identically except that theyreceived an antibody lacking TCO modification.

Window chambered nude mice (n ¼ 3) were prepared as previouslydescribed. To create the LS174T xenograft, approximately 1 million LS174Tcells were subcutaneously injected and allowed to develop into a tumorof approximately 2mm in diameter. 30 μg of TCOmodified anti-A33modifiedwith a VT680 probe was injected by tail vein. After 24 h, a secondary injectionof 30 μg of fluorescent PMT10 (750 nm) was injected via tail vein. After 3 h,the tumors were imaged using scanning confocal microscopy. Control mice(n ¼ 3) were treated identically except that they received an antibody lackingTCO modification.

PET Imaging of Tetrazine Cycloadditions. Approximately 106 LS174T or1.5 × 106 A431 cells were injected subcutaneously in either the hind limbor upper dorsal region in nude mice (n ¼ 3). The difference in cell numberwas used to ensure similar sized tumors at the time of imaging due to dif-ferent growth rates. Tumors were allowed to grow for 2 w after which micewere injected via tail vein with 30 μg of TCOmodified anti-A33 modified witha VT680 label. After allowing 24 h for targeting and clearance of the anti-body, the mice received 30 μg of 18F-PMT (30 μg, 150 μCi) via a tail veininjection. All PET-CT images were acquired on a Siemens Inveon PET-CT. EachPETacquisition was approximately 60 min in duration. PETwas reconstructedfrom 600 million coincidental, 511 keV photon counts on a series of LSO(lutetium oxyorthosilicate) scintillating crystal rings. Counts were rebinnedin 3D by registering photons spanning no more than three consecutive rings,then reconstructed into sinograms by utilizing a high resolution FourierRebin algorithm. A reconstruction of sinograms yielded a 3D mapping of po-sitron signal using a 2D filtered back-projection algorithm, with a Ramp filterat a Nyquist cut-off of 0.5. Image pixel size was anisotropic, with dimensionsof 0.796 mm in the z direction and 0.861 mm in the x and y directions, for atotal of 128 × 128 × 159 pixels.

CT was reconstructed from 360 projections of X-rays with a cone beamangle of 9.3 ° over 360 ° perpendicular to the animal bed. 80 keV X-rays weretransmitted from a 500 μA anode source, 347 mm from the center of rotationand recorded on a CCD detector, containing 2,048 transaxial and 3,072 axialpixels. Projections were calibrated using 70 dark and 70 light images, inter-polated bilinearly, processed through a Shepp-Logan filter, then recon-structed using a filtered back projection algorithm. Isotropic CT pixel sizewas 110.6 μm , with a total of 512 × 512 × 768 pixels. Scaling to HounsfieldUnits, calibration was done using a 8.0 cm cylindrical phantom containingwater prior to CT acquisition. During CT acquisition, iodine contrast wasinfused into the tail vein at a rate of 35 μL∕min to enhance intravascularcontrast. Projections were acquired at end expiration using a BioVet gatingsystem (M2M Imaging) and CT acquisition time was approximately 10 min.Reconstruction of datasets, PET-CT fusion, and image analysis were doneusing IRW software (Siemens). 3D visualizations were produced with theDICOM viewer OsiriX (The OsiriX foundation).

Imaging Antibody and Dextran Uptake in Tumor Slices. Xenograft bearing micewere euthanized following PET-CT, tumors were excised and cut into 1 mmthick slices using a heart slicer (Zivic Laboratories). Tumor slices were thenexposed to a phosphor screen (GE Healthcare) for 24 h. The next day, thescreen was scanned using a phosphor imager (Typhoon 9410; GE) yieldingautoradiographic imaging of 18F TCO localization in tumors. In the exactpositioning used for autoradiography exposure, samples then underwentfluorescence reflective imaging (OV-110; Olympus) with a 0.56X objectiveand 280 ms exposure for localization of VT680 conjugated A33 antibody.

ACKNOWLEDGMENTS. The authors gratefully acknowledge S.A. Hilderbrand,J.B. Haun, and R. Mazitschek for many helpful discussions and suggestions.This work was funded in part by National Institutes of Health (NIH) grantsP50CA86355, R01EB010011, K01EB010078, and T32CA079443.

1. Prescher JA, Dube DH, Bertozzi CR (2004) Chemical remodelling of cell surfaces inliving animals. Nature 430:873–877.

2. Rostovtsev VV, Green LG, Fokin VV, Sharpless KB (2002) A stepwise huisgen cycloaddi-tion process: copper(I)-catalyzed regioselective "ligation" of azides and terminalalkynes. Angew Chem Int Ed Engl 41:2596–2599.

3. Hong V, Steinmetz NF, Manchester M, Finn MG (2010) Labeling live cells by copper-catalyzed alkyne-azide click chemistry. Bioconjug Chem 21:1912–1916.

4. Jiang H, et al. (2011) Imaging glycans in zebrafish embryos by metabolic labeling andbioorthogonal click chemistry. Journal of visualized experiments JoVE (52).

5. Agard NJ, Prescher JA, Bertozzi CR (2004) A strain-promoted [3þ 2] azide-alkynecycloaddition for covalent modification of biomolecules in living systems. J Am ChemSoc 126:15046–15047.

6. Agard NJ, et al. (2006) A comparative study of bioorthogonal reactions with azides.ACS Chemical Biology 1:644–648.

7. Baskin JM, et al. (2007) Copper-free click chemistry for dynamic in vivo imaging. ProcNatl Acad Sci USA 104:16793–16797.

8. Ning XH, Guo J, Wolfert MA, Boons GJ (2008) Visualizing metabolically labeledglycoconjugates of living cells by copper-free and fast huisgen cycloadditions. AngewChem Int Ed 47:2253–2255.

9. Codelli JA, Baskin JM, Agard NJ, Bertozzi CR (2008) Second-generation difluorinatedcyclooctynes for copper-free click chemistry. J Am Chem Soc 130:11486–11493.

10. Devaraj NK, Upadhyay R, Haun JB, Hilderbrand SA, Weissleder R (2009) Fast andsensitive pretargeted labeling of cancer cells through a tetrazine/trans-cyclooctenecycloaddition. Angew Chem Int Ed Engl 48:7013–7016.

11. Devaraj NK, Weissleder R, Hilderbrand SA (2008) Tetrazine-based cycloadditions:application to pretargeted live cell imaging. Bioconjug Chem 19:2297–2299.

12. Blackman ML, Royzen M, Fox JM (2008) Tetrazine ligation: fast bioconjugation basedon inverse-electron-demand Diels-Alder reactivity. J Am Chem Soc 130:13518–13519.

13. Taylor MT, Blackman ML, Dmitrenko O, Fox JM (2011) Design and synthesis of highlyreactive dienophiles for the tetrazine-trans-cyclooctene ligation. J Am Chem Soc133:9646–9649.

14. Devaraj NK, Hilderbrand S, Upadhyay R, Mazitschek R, Weissleder R (2010) Bioortho-gonal turn-on probes for imaging small molecules inside living cells. Angew Chem IntEd Engl 49:2869–2872.

15. Rossin R, et al. (2010) In vivo chemistry for pretargeted tumor imaging in live mice.Angew Chem Int Ed Engl 49:3375–3378.

16. Selvaraj R, et al. (2011) Tetrazine-trans-cyclooctene ligation for the rapid constructionof integrin alpha(v)beta(3) targeted PET tracer based on a cyclic RGD peptide. BioorgMed Chem Lett 21:5011–5014.

17. Reiner T, Keliher EJ, Earley S, Marinelli B, Weissleder R (2011) Synthesis and in vivoimaging of a 18F-labeled PARP1 inhibitor using a chemically orthogonal scavenger-assisted high-performance method. Angew Chem Int Ed Engl 50:1922–1925.

18. Wingardh K, Strand SE (1989) Evaluation in vitro and in vivo of two labelling techni-ques of different 99 mTc-dextrans for lymphoscintigraphy. Eur J Nucl Med 15:146–151.

19. Bisht S, Maitra A (2009) Dextran-doxorubicin/chitosan nanoparticles for solid tumortherapy. Wiley Interdisciplinary Reviews-Nanomedicine and Nanobiotechnology1:415–425.

20. Niemi TT, Miyashita R, YamakageM (2010) Colloid solutions: a clinical update. J Anesth24:913–925.

21. Orcutt KD, Nasr KA, Whitehead DG, Frangioni JV, Wittrup KD (2011) Biodistributionand clearance of small molecule hapten chelates for pretargeted radioimmunother-apy. Mol Imaging Biol 13:215–221.

22. Heath JK, et al. (1997) The human A33 antigen is a transmembrane glycoproteinand a novel member of the immunoglobulin superfamily. Proc Natl Acad Sci USA94:469–474.

23. Ackerman ME, et al. (2008) A33 antigen displays persistent surface expression. CancerImmunol Immunother 57:1017–1027.

24. Weissleder R, Pittet MJ (2008) Imaging in the era of molecular oncology. Nature452:580–589.

25. Lehr HA, Leunig M, Menger MD, Nolte D, Messmer K (1993) Dorsal skinfold chambertechnique for intravital microscopy in nude mice. Am J Pathol 143:1055–1062.

26. Keliher EJ, Reiner T, Turetsky A, Hilderbrand SA, Weissleder R (2011) High-yielding,two-step 18F labeling strategy for 18F-PARP1 inhibitors. ChemMedChem 6:424–427.

27. Lee FT, et al. (2005) Enhanced efficacy of radioimmunotherapy with 90Y-CHX-A”'-DTPA-hu3S193 by inhibition of epidermal growth factor receptor (EGFR) signalingwith EGFR tyrosine kinase inhibitor AG1478. Clin Cancer Res 11:7080s–7086s.

28. Thurber GM, Zajic SC, Wittrup KD (2007) Theoretic criteria for antibody penetrationinto solid tumors and micrometastases. J Nucl Med 48:995–999.

29. Mager DE (2006) Target-mediated drug disposition and dynamics. Biochem Pharmacol72:1–10.

6 of 6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1113466109 Devaraj et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0