Bioorthogonal Chemistries for Nanomaterial Conjugation and Targeting

DOI 10.1515/ntrev-2012-0083 Nanotechnol Rev 2013; 2(2): 215–227

Review

Maha K. Rahim , Rajesh Kota , Sumi Lee and Jered B. Haun * Bioorthogonal chemistries for nanomaterial conjugation and targeting

Abstract: Bioorthogonal chemistries are covalent reaction 1 Introduction pairs that proceed in the presence of biological compo- nents with complete specificity. A suite of reactions has Nanomaterials continue to generate significant interest been described to date that provides scientists and engi- in biomedical applications owing to their unique proper- neers with diverse operational characteristics for different ties. For example, nanocarriers can provide protection, applications. Nanomaterials in particular have benefitted high cargo capacity, controlled release, and attractive from these new capabilities, resulting in improved cou- pharmacokinetics/biodistribution [1] . Moreover, nanocrys- pling efficiencies and multifunctionality. In this review, tals exhibit unique physical properties such as paramag- we will discuss the application of bioorthogonal chemis- netisim [2, 3], semiconductor fluorescence [4 – 6] , plasmon tries to different nanomaterial systems, highlighting the resonance [7] , and photoluminescence upconversion [8, 9] advantages and limitations for use in bioconjugation. We that can be harnessed as exquisitely sensitive, stable, and/ will also describe how recent improvements in the reac- or multiplexed detection signals. Finally, biomolecules can tion speed of catalyst-free bioorthogonal chemistries have be conjugated to the surface of all types of nanomaterials enabled the successful coupling of nanomaterials directly to impart efficient and multivalent adhesion for molecular to live cells. Using a recently developed reaction pair, targets. Owing to these attributes, targeted nanomaterials tetrazine and trans -cyclooctene, the direct covalent cou- present attractive diagnostic and therapeutic capabilities. pling to cells has been shown to occur on time-scales that However, to maximally exploit their potential, effective are relevant for biological studies and diagnostic applica- bioconjugation strategies are needed that are simple, fast, tions and can even amplify nanomaterial binding greater broadly applicable, have a small footprint, and are suit- than tenfold relative to traditional immunoconjugates. able for complex biological environments. This powerful technique still maintains exquisite specific- Bioorthogonal chemistries are a class of reactions ity, however, yielding robust results in clinical diagnos- whose cognate partners do not interfere with the func- tic applications using human tissue and blood samples. tional groups endogenous to cells. This includes standard Future work will likely focus on further advancement of protein functional groups such as primary amine, thiol, the in situ amplification technique, such as increasing carboxylic acid, and hydroxyl groups, as well as meta- nanomaterial binding, enabling multiplexed detection bolic intermediates like aldehydes and ketones [1] . These through the use of orthogonal reaction systems and exten- recently developed chemistries offer greater control and sion to applications in vivo . flexibility for manipulating biological systems and, thus, are ideal for elucidating the biological processes or engi- Keywords: amplification; bioorthogonal chemistry; neering-specific interactions within a complex cellular or molecular detection; nanomaterials; tetrazine. tissue environment. In this review, we discuss the development of *Corresponding author: Jered B. Haun , Department of Biomedical bioorthogonal coupling reactions and highlight their Engineering and Chao Family Comprehensive Cancer Center, use for probing biomolecule targets with nanomate- University of California Irvine, 3107 Natural Sciences II, Irvine, CA rial sensors in biological systems. We describe how the 92697, USA, e-mail: [email protected] growing suite of bioorthogonal chemistries is suited for Maha K. Rahim, Rajesh Kota and Sumi Lee: Department of Biomedical Engineering and Chao Family Comprehensive Cancer different types of nanomaterials and targeting appli- Center , University of California Irvine, 3107 Natural Sciences II, cations. Specifically, bioorthogonal chemistries have Irvine, CA 92697, USA enabled more efficient and diverse modifications of nanomaterials and subsequent targeting to biomolecules. This includes recent applications in which the 216 M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials reactions, themselves, have been used for coupling these reasons, the CuAAC reaction has become a valuable nanoparticles to live cells in situ (i.e., direct labeling by technique for surface modifications and protein labeling. covalent reaction). Finally, we discuss the clinical trans- The CuAAC has also been referred to as “ click chemistry” lation of bioorthogonal chemistry-based cell diagnostics. or the “ click reaction” . Unfortunately, the CuAAC has Future work in this area will improve the sensitivity of limited applicability for living systems because the Cu(I) nanoparticle detection platforms as well as utilize mul- catalyst is cytotoxic. Another limitation is that the Cu(I) tiple bioorthogonal chemistries simultaneously or in affects the fluorescence yield of fluorescent proteins such concert with classic chemistries for multiplexing capa- as GFP. More directly relevant to this review, the CuAAC is city. Another promising advancement is the translation not compatible with quantum dots because the Cu(I) cat- of bioorthogonal chemistries to in vivo systems for detec- alyst nonreversibly quenches fluorescence [15] . Finally, tion or drug delivery. care must be taken when achieving the reducing conditions required for Cu(I) because they can easily affect acid-labile species such as iron oxide nanocrystals. As an alternative to metal catalysis, the Bertozzi 2 History of development group demonstrated that the introduction of ring strain into the alkyne moiety can effectively drive cycloaddi- Taking inspiration from the classical Staudinger reaction, tion, but with a slower reaction rate [16] . In this cata- Saxon and Bertozzi [10] designed a chemoselective liga- lyst-free approach, the bioorthogonal cycloaddition of tion between an azide and phosphine. Unlike the origi- geometrically strained alkyne, cyclooctyne, reacts with nal reduction reaction whose intermediate hydrolyzes to azide to form a triazole. This reaction is stable under form a primary amine and phosphine oxide, the modified physiological conditions, and cytotoxic effects were not Staudinger reaction introduces an electrophilic trap on observed after labeling azido sugars with a biotin-con- the phosphine reactant. As a result, a stable amide bond jugated cyclooctyne. Though improved biocompatibility forms after hydrolysis. Representing both the origins and is certainly an advantage of the strain-promoted reac- future application goals for this reaction, it was named tion, the slower reaction rate relative to CuAAC is limit- the Staudinger ligation. Utility in a biological setting ing for some applications. The slow reaction has been was first demonstrated using azido-modified sugars that addressed to some degree by modifying cyclooctyne were metabolically incorporated onto cellular surfaces. with substituent groups such as monofluoriniated [17] Azide moieties were then specifically labeled by covalent and difluorinated cyclooctyne [18] . The addition of reaction with a biotinylated phosphine and subsequent electron-withdrawing fluorine groups adjacent to the binding with fluorescein isothiocyanate (FITC)-avidin alkyne encourages reaction with azide. Cyclooctyne vari- [10] . Owing to the small size of the azide reactant, modi- ations, such as 4-dibenzyocylooctynols (DIBO) [19] and fications do not perturb the biomolecule. In addition, biarylazacyclooctynones (BARAC) [20] , introduce aro- azide is entirely absent from biological systems. These matic rings to impose higher ring strain and, thereby, advantages, along with the chemoselectivity of the reac- increase the reactivity of the alkyne. The improved tion, have resulted in extensive use of the Staudinger liga- reaction rates observed with DIBO-azide cycloaddition tion for labeling biomolecules [11] . Despite its effective- prompted Boons et al. to utilize DIBO for a strain-pro- ness, the Staudinger ligation is limited by its relatively moted nitrone cycloaddition [21] . Nitrone cycloadditions, slow reaction kinetics and susceptibility to phosphine though similar, offer faster kinetics than their azide oxidation [11, 12]. counterparts on the order of sixfold. The copper-catalyzed Huisgen cycloaddition was To meet the needs of the most demanding biological developed to address the kinetic limitations of the applications, the groups led by Fox and Weissleder further Staudinger ligation, while still making use of azide as a accelerated catalyst-free capabilities by developing a set reactant. This 1,3 dipolar cycloaddition yields a triazole of reactions between 1,2,4,5-tetrazines (Tzs) and strained linkage by reacting an azide with a terminal alkyne [13] . alkenes that proceeds via inverse-electron-demand Diels- Without a catalyst, this reaction only proceeds at high tem- Alder cycloaddition [22 – 25] . The strained dienophiles that peratures. However, when catalyzed by reduced copper, have been tested include norbornene, cyclooctyne, cyclo- or copper(I), the reaction proceeds at an accelerated rate propene, and trans-cyclooctene (TCO), each displaying at room temperature [14] . The copper-catalyzed azide- varying reaction speeds. Modifying Tz substituents has alkyne cycloaddition (CuAAC) is efficient, selective, and also been shown to affect the reaction speed with TCO, proceeds over a broad pH range in aqueous solutions. For as well as Tz stability in serum [26] . The extremely fast M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials 217 reaction between Tz and TCO has enabled live cell labe- groups in vivo, make them ideal for modifying nanomate- ling both in vitro [28] and in vivo [29, 30] . Interestingly, rials with biomolecules and for targeting disease markers the TCO/Tz chemistry has been shown to be orthogonal for diagnostic purposes. to octyne-azide if a slower-reacting Tz is employed [31] . Despite the potential power of the TCO/Tz reaction, dif- ficulty in synthesizing Tz had hindered widespread application. This problem was recently addressed though with 3 Functionalization of the report of a series of 1,2,4,5-Tzs by metal salt-catalyzed nanomaterials with biomolecules formation from nitriles [32] . The bioorthogonal chemistries that we will discuss using CuAAC in this review are summarized in Table 1 and the general reaction schemes are presented in Figure 1. We will spe- Early application of bioorthogonal chemistries to nano- cifically focus on the application of these bioorthogonal materials focused on the CuAAC at the time in which the chemistries to nanomaterial-based sensing applications. “ click reaction ” was gaining popularity across biology Successful nanomaterial detection strategies require and engineering. The high selectivity in complex biologi- advanced nanomaterial targeting for efficient binding and cal surroundings, rapid reaction rate, mild aqueous con- sensitive biomolecule detection. Nanomaterials should ditions, and reproducible results made the reaction well specifically recognize their biomolecule targets, yet have suited for biocompatible nanoparticle modification. In limited nonspecific binding, and should react rapidly. The some cases, researchers were even able to show that using speed and chemoselectivity of bioorthogonal reactions, as “ click ” chemistry to functionalize nanoparticles could well as their widespread application to labeling functional overcome the limitations that had previously hindered

Table 1 Bioorthogonal reaction schemes.

Reaction Reaction scheme Rate constant Comments Reference (m-1 s-1)

Staudinger ligation Triarylphosphine+azide 10-3 - Slow reaction speed [10, 12] - Phosphine oxidation CuAAC Alkyne+azide Up to 104 + Fast reaction speed [13–15] + Simple reactants - Cytotoxic - Reduces protein and quantum dot fluorescence - Potential negative effects of acidic reaction conditions (iron oxide) Cyclooctyne-based, strain- Cyclooctyne+azide 10-3 + No catalyst required [11, 16] promoted cycloaddition Fluorinated cyclooctyne+azide 10-3–10-1 - Slow reaction speeds [17, 18] 4-Dibenzocyclooctynol+azide 2 - Difficult synthesis of cyclooctyne and [19] derivatives BARAC+azide 1 - Poor solubility of dienephilesa [20] Tz-based, inverse-electron- Norbornene+Tz 1 + Extremely fast reaction speed (TCO) [23] demand Diels-Alder Cycloaddition Cyclooctyne+Tz 10 + Can tailor reaction speed or [25] application need with range of dienephiles Cyclopropene+Tz 10 + Can tailor reaction speed with a [24] range of Tzsb Trans-cyclooctene+Tz 104 - Poor solubility of dienephilesa [26, 27] aOwing to the potential solubility concerns for the dienephiles, particularly those with ring structures (TCO, cyclooctyne, cyclopropene), we do not recommend high-density modification of the nanoparticles. bAll the reaction rates shown for the Tz cycloadditions are for the most reactive version: [4-(1,2,4,5-tetrazin-3-yl)phenyl]- methanamine or 1,2,4,5-Tz-benzylamine. Varying substituent groups has been shown to lower the reaction rate as low as 100 m-1 s-1 [26]. 218 M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials

Figure 1 Reaction schemes for the bioorthogonal chemistries discussed in this review and listed in Table 1. in vivo nanoparticle targeting, such as insufficient reac- and estradiol were modified with azide and coupled to tion efficiency, aqueous compatibility, and bio-inertness. alkyne-functionalized magnetic nanoparticles [35] . This Bosseliar et al. demonstrated the utility of the CuAAC procedure was facile, and the resulting triazole linkages reaction for modifying inorganic nanoparticles by func- were stable for several months. tionalizing gold nanocrystals [33] . Previous attempts to The Bhatia group demonstrated the in vivo compat- employ this chemistry on gold nanoparticles were largely ibility of “ clickable ” nanoparticles [36] . A tumor-targeting unsuccessful, leading to low yields [34] . The authors peptide, LyP-1, was decorated with alkyne and reacted via addressed previous problems using a homogenous water- “ click ” chemistry to azide-magnetofluorescent nanopar- THF solution to solubilize all the reactants and inert ticles (MFNP) to form LyP-1 nanoparticles. After intrave- atmospheric conditions to prevent the oxidation of Cu(I). nous injection into mouse tumor xenograft models, LyP-1- As a result, azido-terminated gold nanoparticles were suc- nano particles retained stability while circulating for over cessfully coupled to terminal alkynes to form 1,2,3-tria- 5 h. LyP-1 nanoparticles directly bound to the p-32 target zolyl linkages at near quantitative yield. A variety of termi- expressed by tumor cells, whereas untargeted nanoparti- nal alkynes were investigated, including PEG-containing cles were unable penetrate the tumor. alkynes, which could assist in the biocompatibility of The in vivo compatibility of “ click ” conjugations also gold nanoparticles. All tested alkynes reacted with similar permitted the coupling of a positron emission tomogra- success. phy (PET) tracer to MFNP for trimodality imaging by PET, The CuAAC chemistry has also been employed to con- magnetic resonance imaging, and fluorescence molecular struct small biomolecule libraries on nanoparticles. A tomography [37] . The application of magnetic nanopar- panel including biotin, indocyanine dye, azo dye, taxane, ticles and fluorophores for molecular imaging has been M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials 219 well explored; however, it is currently limited by in vivo however, reports using the catalyst-free alternatives are detection thresholds. “ Click ” chemistry enabled facile beginning to rise. In the case of quantum dots, catalyst-free addition of the PET tracer 18 F to magnetic nanoparticles, chemistries are the only option because Cu(I) irreversibly a labeling that has otherwise been chemically challenging quenches fluorescence. The Bawendi group was the first because of the short half-life of the probe. The addition to demonstrate bioorthogonal modification of quantum of the positron emitter lowered the detection threshold dots using inverse-electron-demand Diels-Alder cycload- two- to fourfold for the trimodality nanoparticles relative dition between Tz and norbornene [15] . Interestingly, as to MRI and FMT and exhibited in vivo stability. the norbornene used to modify the quantum dot does not Thorek et al. utilized CuAAC to investigate the cou- coordinate with the zinc sulfide coating, unlike the thiol pling efficiency of the antibodies to magnetic nanoparti- or amine functional groups, the bioorthogonal modifi- cles in comparison to the alternatives. In fact, they dem- cation was more facile and efficient than the traditional onstrated that CuAAC yields a higher antibody coupling bioconjugation strategies. Norbornene quantum dots density to nanoparticles than carbodiimide chemistry were functionalized with Tz-modified epidermal growth [38] . The superior “ click ” reaction coupled an average of factor (EGF), and the resulting conjugate was used to label almost seven antibodies per nanoparticle, while carbo- EGF receptor (EGFR) on lung cancer cells for fluorescent diimide chemistry only yielded approximately three anti- imaging. Remarkably, the coupling chemistry was rapid bodies. The increased antibody density on CuAAC-derived enough to directly label cells in situ, as demonstrated by magnetic nanoparticles directly resulted in increased first targeting cells with Tz-EGF and then covalently react- binding efficiency for the molecular target due to higher ing norbornene-quantum dots. avidity. Catalyst-free site-specific ligation has also been dem- CuAAC has also been employed in tandem with onstrated using strain-promoted azide-alkyne cycloaddi- expressed protein ligation (EPL) to couple magnetic tion [40] . This was accomplished by first expressing an nanoparticles to recombinant targeting ligands [39] . anti-HER2 single-chain antibody (scFv) that was recom- Expressed protein ligation is a site-specific protein modi- binantly expressed with an N-terminal serine residue. fication technique that can offer controlled orientation Through a series of mild reactions, the hydroxyl group of affinity molecules on the nanoparticle surface. This on this serine was converted to a nitrone functionality for is useful to maximize binding efficiency for target mol- reacting with cyclooctyne. Indeed, the nitrone-scFv could ecules. However, previous attempts to use EPL for nan- be attached to DIBO-modified MFNPs, and the cycloaddi- oparticle modification were plagued by low yields. In tion product selectively bound to its HER2 target leading this study, a two-step approach using CuAAC with EPL to elevated fluorescence and decreased T2 relaxation. significantly improved yield. A HER2-specific affibody Though not yet tested with nanomaterials, genetically with a C-terminus thioester was ligated to an alkynated encoded TCO has been reported using the lipoic acid/ fluorescent peptide via an N-terminus cysteine by EPL. lipoic acid acceptor peptide system [41]. The alkyne-modified affibody was then “ clicked ” via the Owing to the rapid kinetics of TCO/Tz Diels-Alder intermediate peptide to azide-magnetic nanoparticles. cycloaddition, Haun et al. adapted this chemistry for The HER2 affibody-magnetic nanoparticles prepared by functionalizing MFNP with antibodies and compared CuAAC-EPL bound to HER2-positive cancer cells more the binding efficiency to the classic maleimide-thiol efficiently than the ones prepared with CuAAC alone coupling [42] . The monoclonal antibodies were first or carbodiimide chemistry. The anti-HER2 affibody was decorated with one to two TCO residues via standard also conjugated to liposomes and dendrimers in similar amine-reactive chemistry and then directly coupled to fashion to demonstrate the broad utility of this approach. Tz-modified MFNPs. This was performed for the antibodies that were specific for the extracellular cancer biomarkers HER2, EGFR, and epithelial cell adhesion molecule (EpCAM). The nanoparticle immunoconju- 4 Catalyst-free bioorthogonal gates were then targeted to the cancer cells overexpress- functionalization of ing the target proteins, and fluorescence was measured. The immunoconjugates prepared using TCO/Tz chemis- nanomaterials try yielded comparable nanoparticle binding levels relative to the traditional maleimide-thiol coupling chem- The application of bioorthogonal chemistries to nano- istry for two of the cases, but interestingly, the EpCAM materials has been dominated by the CuAAC reaction; signals were much greater. 220 M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials

5 Covalent coupling of this high loading density, the affinity of the antibody was nanomaterials to cells using still unaffected. Charged fluorophores can affect antibody binding even at low loading levels, thus, the nonpolar TCO the Tz cycloaddition is very inert for protein bioconjugations. Most importantly, the high number of TCO reaction sites on the antibod- Haun et al. investigated the capabilities of the exception- ies supported MFNP binding levels that exceeded direct ally fast TCO/Tz reaction for covalent, in situ coupling of immune conjugates prepared by BOND-1 or traditional MFNP to live cells [42] . Using monoclonal antibodies that maleimide-thiol chemistries. Using the BOND-2 scheme, were heavily modified with TCO for pretargeting extracel- MFNP signal was, thus, chemically amplified by attach- lular cancer biomarkers (HER2, EGFR, EpCAM), followed ing multiple nanoparticles per antibody scaffold, and this by the covalent reaction of Tz-MFNP, the theory that the effect extended as high as 15-fold. Some signal amplifica- antibody may be large enough and the covalent linkage tions was also observed when avidin-biotin binding was small enough, to promote the attachment of multiple used for a similar two-step labeling procedure; however, MFNP sensors was tested (Figure 2 ). The antibody would, the large size of avidin limited amplification, and thus, the thus, act as a scaffold to amplify nanoparticle binding. bioorthogonal chemistry-based technique remained two- This strategy was referred to as a two-step bioorthogonal to threefold greater. This superior signal amplification has nanoparticle detection (BOND-2), in comparison to using been attributed to the small size and high valency of the the TCO/Tz chemistry for antibody coupling prior to cell bioorthogonal reactants. labeling (one-step BOND or BOND-1). Initial experiments The bioorthogonal cycloaddition between Tz and TCO demonstrated that modulating TCO loading on monoclonal has been shown to perform well in an intracellular envi- antibodies resulted in a corresponding increase in nano- ronment using fluorescent molecules. Therefore, Haun particle binding. Binding saturation was not observed until et al. adapted the two-step nanoparticle detection scheme approximately 20 reactive sites per antibody, and even at to target intracellular cancer biomarkers [43] . Fixed and

A In situ bioorthogonal amplification

TCO-Antibody

Tz-Nanoparticle

Target cell Labeled cell B HER2 EpCAM 100 100

75 75

50 50

25 25 Fluorescence signal (% max) Fluorescence signal (% max) 0 0 Bond-1Av/biotin Bond-2 Bond-1Av/biotin Bond-2 8 31 54 61318

Figure 2 Overview of the two-step nanoparticle detection strategy (BOND-2). (A) A biomarker of interest is targeted with a monoclonal antibody modified with TCO. Multiple Tz-modified nanoparticles are then covalently coupled to bound antibodies to achieve amplified nanoparticle binding. (B) Fluorescence intensity of labeled cells has the highest achievable signal using the BOND-2 strategy compared to the avidin/biotin two-step labeling or the one-step labeling using bioorthogonal chemistry (BOND-1). (Reprinted with permission from Ref. [42] . Copyright 2010 Nature Publishing Group.) M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials 221 permeabilized cancer cells were labeled with TCO-modi- only the background signal was observed when detecting fied antibodies followed by Tz-MFNPs. As observed with a panel of other bacteria. extracellular biomarker targeting, the two-step bioorthog- Bacterial detection has also been extended to Gram- onal labeling strategy also supported signal amplification positive pathogenic bacteria [45] . Antibiotics vancomycin inside of cells, yielding a > 10-fold higher signal in com- and daptomycin, which are used to treat Gram-positive parison to direct immunoconjugates. Moreover, nanopar- bacterial infections by binding to the bacterial cell wall ticle binding correlated closely with molecular expression and preventing cell wall synthesis, were modified with levels determined by antibody staining and Western blot TCO. The antibiotics maintained the ability to bind to analysis (Figure 3 ). This included numerous cytoplasmic the bacterial cell wall after modification. In this study, and nuclear proteins of interest for cancer, such as acti- synthesized antibiotics containing TCO moieties suc- vated signaling molecules and growth indicators. Finally, cessfully bound to Gram-positive bacteria and were sub- the in situ bioorthogonal coupling technique was extended sequently labeled with Tz-modified MFNPs (Figure 4 ). to quantum dots. So long as antibiotic concentrations were below 20 μ m This two-step nanoparticle detection strategy was suc- to minimize bactericidal effects, Gram-positive bacteria cessfully adapted from cancer to bacterial pathogen detec- were efficiently labeled with almost two orders of mag- tion. Liong et al. modified the antibodies that were highly nitude higher sensitivity than demonstrated with cova- selective for Staphylococcous aureus with TCO, and these lent nanoparticle conjugates. This labeling technique antibodies bound specifically to their bacterial targets in was even suitable for detecting Gram-positive bacteria human sputa [44] . Tz-magnetic nanoparticle probes were within infected macrophages after semipermeabilization then efficiently coupled to TCO-antibodies via cycloaddi- procedures. tion reaction to enable rapid and accurate detection of S. The amplification of the nanoparticle signal beyond aureus. The bacterial signal was specific to S. aureus, as the two-step bioorthogonal scheme has been achieved

A CK Ki-67 10,000 3000 PANC-1 PANC-1 8000 SK-BR-3 A549 2000 6000 HT-29 4000 1000 SK-BR-3 MFNP signal (a.u.) MFNP MFNP signal (a.u.) MFNP HeLa 2000 SK-OV-3

SK-OV-3 2 2 U118 R =0.9815 R =0.9289 0 0 0 0.51.0 1.5 2.0 2.5 0246 × -7 Ki-67 Expression (×10-6/cell) B CK Expression ( 10 /cell)

Figure 3 Intracellular biomarker labeling using BOND-2. (A) MFNP fluorescence after labeling using BOND-2 correlates with molecular expression determined by targeting panels of cell lines expressing various amounts of CK or Ki-67. (B) Confocal microscopy images of cells targeted for CK (i –iv) and Ki-67 (v – viii) using antibodies modified with fluorescent dye (i and v), and MFNP-modified antibodies (ii and vi). The signals correlate well, illustrated by merged images (iii and vii). A nonbinding, TCO-modified antibody was used as a control (iv and viii). (Reprinted with permission from Ref. [43] . Copyright 2011 American Chemical Society.) 222 M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials

Figure 4 Labeling Gram-positive bacteria with TCO/Tz chemistry and MFNP. (A) Schematic showing how the antibiotic vancomycin, modified with TCO, binds to the cell wall of Gram-positive bacteria. Subsequent addition of Tz-modified MFNP results in the covalent attachment to the antibiotic and enables bacterial detection. (B) The purity and identity of vancomycin-TCO are confirmed with HPLC (top) and mass spectroscopy (bottom). (C) The relative fluorescence intensity of MNFP-labeled bacteria scales with changes in vanc-TCO concentration for Gram-positive bacteria, but is negligible for Gram-negative bacteria. (D) The two-step bioorthogonal technique more efficiently labeled Gram-positive bacteria than the direct covalent conjugates. [Reprinted (adapted) with permission from Ref. [45] . Copyright 2011 American Chemical Society.]

using a layering approach [46] . This was accomplished dithiol bond between the MFNP and bioorthogonal reac- using separate, alternating rounds of Tz- and TCO-mod- tant allowed the release of MFNP from cells using a reduc- ified nanoparticles. Peterson et al. first targeted extra- ing agent. The ability to separate the nanoparticles after cellular cancer biomarkers by executing the standard targeting obviates the need to purify samples prior to TCO-antibody and Tz-MFNP steps. Subsequent reactions detection, which could be advantageous for clinical appli- with TCO-MFNP and then Tz-MFNP resulted in nanopar- cations. Indeed, nonpurified ascites derived from patients ticle clusters that exhibited a higher transverse relaxa- with pancreatic cancer exhibited 90 % of the signal tion rate, demonstrating an enhanced signal amplifica- observed in purified samples for the expression of EGFR, tion (Figure 5 ). Moreover, the introduction of a cleavable EpCAM, MUC-1, and HER2. M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials 223

A Labeling Amplification

AMP1 AMP2

Cleave: DTT Cleave: DTT Cleave: DTT Amplification ss s s s s

L-C AMP1-C AMP2-C

BC 4 100 AMP1-C ) -1 AMP2-C (s 3 (%) 75 + 2 2 T R Δ Δ 2 50

L-C 1 25 NMR Signal NMR Signal Threshold 0 0 10 100 1000 10,000 L-C Cell number AMP1-CAMP2-C

Figure 5 Layering TCO and Tz nanoparticles results in signal amplification. (A) TCO-modified antibodies are bound to a biomarker target and conjugated with Tz-MFNPs in the initial labeling step. The reaction of separate rounds of complementary orthogonal MFNP with the bound species leads to signal amplification (AMP1, AMP2). Further improvement in the detection capabilities was obtained by introducing a cleavage site on the nanoparticles to remove the cell-associated background. (B and C) The layering and cleavage approach yields a higher signal with each round. [Reprinted (adapted) with permission from Ref. [46] . Copyright 2012 American Chemical Society.]

6 Clinical applications using from patients with suspected intra-abdominal tumors. in situ The limited tissue sample available in FNAs, as well as bioorthogonal nanoparticle the low expression levels of some proteins of interest, has labeling prevented molecular analysis in the clinic. The bioorthogonal amplification/ μNMR detection strategy, however, is Owing to the remarkable simplicity and signal amplifica- highly sensitive and capable of accurately measuring tion capabilities offered by the two-step, in situ bioorthog- the expression levels from sample sizes on the order of onal nanoparticle coupling technique, it was translated 100 – 1000 cells [42, 43] . It is, therefore, well-suited to to clinical applications for molecular profiling of human address the challenges of molecular diagnosis in the cancer samples. This added sensitivity is critical for clinic. Indeed, single FNA specimens contained suffi- detecting cancer biomarkers because detailed molecular cient cellular material to detect nine different surface and descriptions of tumors can assist in the diagnosis and intracellular biomarkers after dividing and labeling in treatment. parallel. Based on the quantitative biomarker expression Haun et al. first used the TCO/Tz bioorthogonal cou- results and evaluation of the consensus diagnosis for 50 pling technique to amplify MNP binding to extracellular patients, a panel of four biomarkers, MUC-1, HER2, EGFR, and intracellular cancer biomarkers for profiling tumor and EpCAM, was identified to provide the most accurate cells using a novel miniaturized diagnostic magnetic prediction of malignant disease. In fact, this four-protein resonance device (μ NMR) [47] . Tumor samples tested malignancy signature had a higher diagnostic accu- were obtained by fine-needle aspiration (FNA) biopsy racy (96 % ) than conventional cytology of FNAs (74 % ) 224 M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials or even histopathology performed on much larger core total microvesicle count and reduced the expression biopsies (84 % ). In addition to being highly accurate, the profiles of the key biomarkers, EGF and EGFRvIII. This bioorthogonal detection scheme had rapid turnaround rapid detection strategy was capable of distinguishing a time of less than an hour. The four-protein signature was dose-dependent decrease in microvesicle counts as well then validated in a test set of 20 new patients, with 100 % as a decline in biomarker expression levels in a cell line accuracy. model (Figure 6 ). The authors applied this platform to Recently, the Weissleder group applied the two- track tumor progression in patients with GBM undergo- step, bioorthogonal coupling technique for protein ing treatment and were successfully able to distinguish biomarker-based detection drug treatment response responder and nonresponder status after TMZ and radi- monitoring of glioblastoma-derived microvesicles ation treatment. [48] . The microvesicles shed from glioblastomas have Bioorthogonal amplification has also been used a molecular expression profile distinct from nontu- clinically to detect rare circulating tumor cells (CTCs) mor host microvesicles. This was exploited to distin- with higher diagnostic accuracy than the current clinical guish tumor from nontumor microvesicles. The protein standard [49] . The two-step TCO/Tz bioorthogonal cou- biomarkers on microvesicles were labeled with MNP pling was used to label CTCs with magnetic nanoparti- using TCO/Tz bioorthogonal coupling and captured on cles in whole blood, after red blood cells were lysed and a microfluidic device, followed by μ NMR detection to removed. Magnetic signatures were then detected using reveal protein expression levels. Notably, the treatment a novel microfluidic chip outfitted with an array of eight of patients with geldanamycin and TMZ decreased the micro-Hall sensors that was capable of detecting the

A BC

D N N N O H N O N

mAb-TCO MNP-TZ Filtration Filtration

Circulating microvesicles EF

CD63 EGFR EGFRvIII ) ξ 0.4 EGFR 1.0 EGFRvIII

0.5 0.2

(normalized) μNMR MV expression ( 0 Cellular expression 0 0 0 0 0.5 1.0 0.5 1.0 0.5 1.0 0 0.5 1.0 EGFR EGFR EGFRvIII EGFRvIII MVs Cells CD63 CD63 0 0.5 1.0 0 0.5 1.0 Dose (μM) Dose (μM)

Figure 6 Microvesicle protein typing for glioblastoma. (A) Scanning electron microscopy image of primary human glioblastoma cell- shedding microvesicles. (B) High-magnification image reveals a microvesicle shape. (C) Transmission electron microscopy shows magnetic nanoparticles targeted to microvesicles shed from glioblastoma. (D) Schematic represents two-step bioorthogonal labeling of microvesicles. (E) Expression profiles of microvesicles determined by flow cytometry and Western blot after geldanamycin treatment. (F) Expression profiles of microvesicles determined by μ NMR assay after geldanamycin treatment. [Reprinted (adapted) with permission from Ref. [48] . Copyright 2012 Nature Publishing Company.] M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials 225 magnetic moment of each passing cell. Flow focusing in has enabled in situ coupling to cells and significant ampli- conjunction with the micro-Hall sensor array produced fication of nanoparticle binding. While these advance- a signal independent of the cell size and only propor- ments have been exciting, we envision a number of areas tional to the number of bound magnetic nanoparticles. where further improvements can be made. For example, The mean Hall voltage due to the magnetic nanoparti- there is continual need to improve detection sensitivity to cles was also proportional to the particle volume; there- enable the interrogation of low-level expression biomark- fore, multiple-sized particles yielded distinguishable ers or even mediators involved in downstream signaling signal peaks. This enabled multiplexed biomarker detec- pathways. To this point, bioorthogonal amplification tion capacity in the micro-Hall detector chip. The blood has relied on the fortuitously large size of antibodies to samples derived from patients with advanced ovarian provide a scaffold. Thus, precisely engineered scaffolds cancer were screened for EGFR, HER2, and MUC-1 with could provide for improved nanoparticle coupling yields this detection approach, leading to the discovery of CTCs and extension to diverse types of affinity molecules. Fur- in 100% of the samples and a diagnostic accuracy of 96% . thermore, multiplexing is the key area for improvements The current conventional approach, CellSearch, was also to be made because specificity is lost if one were to try employed for comparison and was able to detect CTCs in and amplify multiple species with the TCO/Tz chemistry. only 20% of the patients with 25% diagnostic accuracy. Certainly, one avenue is to employ traditional nanopar- Thus, the bioorthogonal nanoparticle amplification/ ticle conjugates with a single bioorthogonally amplified micro-Hall detector represents an exciting CTC detection target. Another approach, however, would be to employ strategy. multiple bioorthogonal chemistries that are orthogonal to each other in concert. One such mutually orthogonal pairing has already been demonstrated for TCO/Tz and cyclooctyne/azide chemistries [30] . Finally, Deveraj et al. 7 Conclusion and future work have established that the TCO/Tz chemistry can be utilized for in vivo detection of cancer using separate injec- In recent years, the targeted nanoparticle delivery vehi- tions of TCO-antibody and a Tz-modified fluorescent cles and diagnostic sensors have shown potential for polymer. Similar work using nanoparticles could establish treating and detecting diseases. As we have discussed at whether bioorthogonal amplification is feasible in vivo for length here, bioorthogonal chemistries have significantly improved detection and drug delivery to tumors or other advanced the capabilities of researchers to conjugate bio- diseases. molecules to nanoparticle surfaces with exceptional efficiency and control. Moreover, the exquisite reaction speed Received December 9, 2012; accepted January 13, 2013; previously and selectivity of the advanced bioorthogonal reactions published online February 23, 2013

References

[1] Algar WR, Prasuhn DE, Stewart MH, Jennings TL, Blanco- live cells, in vivo imaging, and diagnostics. Science 2005, 307, Canosa JB, Dawson PE, Medintz IL. The controlled display 538 – 544. of biomolecules on nanoparticles: a challenge suited to [6] Medintz IL, Uyeda HT, Goldman ER, Mattoussi H. Quantum dot bioorthogonal chemistry. Bioconjugate Chem. 2011, 22, bioconjugates for imaging, labeling and sensing. Nat. Mater. 825 – 858. 2005, 4, 435 – 446. [2] Haun JB, Yoon TJ, Lee H, Weissleder R. Magnetic nanoparticle [7] Anker JN, Hall WP, Lyandres O, Shah NC, Zhao J, Van Duyne RP. biosensors. WIREs. Nanomed. Nanobiotechnol. 2010, 2, 291– 304. Biosensing with plasmonic nanosensors. Nat. Mater. 2008, 7, [3] Lee JH, Huh YM, Jun YW, Seo JW, Jang JT, Song HT, Kim S, Cho 442 – 453. EJ, Yoon HG, Suh JS, Cheon J. Artificially engineered magnetic [8] Hilderbrand SA, Shao F, Salthouse C, Mahmood U, nanoparticles for ultrasensitive molecular imaging. Nat. Med. Weissleder R. Upconverting luminescent nanomaterials: 2007, 13, 95 – 99. application to in vivo bioimaging. Chem. Commun. 2009, [4] Alivisatos P. The use of nanocrystals in biological detection. Nat. 4188 – 4190. Biotechnol. 2004, 22, 47 – 52. [9] Wang F, Liu X. Recent advances in the chemistry of lanthanide- [5] Michalet X, Pinaud FF, Bentolila LA, Tsay JM, Doose S, Li JJ, doped upconversion nanocrystals. Chem. Soc. Rev. 2009, 38, Sundaresan G, Wu AM, Gambhir SS, Weiss S. Quantum dots for 976 – 989. 226 M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials

[10] Saxon E, Bertozzi CR. Cell surface engineering by a modified nanoparticles and small molecules for label-free monitoring Staudinger reaction. Science 2000, 287, 2000 – 2007. of protein binding and reaction kinetics. Lab Chip 2012, 12, [11] Sletten EM, Bertozzi CR. Bioorthogonal chemistry: fishing 3103– 3110. for selectivity in a sea of functionality. Angew. Chem. Int. Ed. [28] Devaraj NK, Upadhyay R, Haun JB, Hilderbrand SA, Weissleder 2009, 48, 6974 – 6998. R. Fast and sensitive pretargeted labeling of cancer cells via [12] Lin FL, Hoyt HM, Van Halbeek H, Bergman RG, Bertozzi CR. tetrazine/trans-cyclooctene cycloaddition. Angew. Chem. Int. Mechanistic investigation of Staudinger ligation. J. Am. Chem. Ed. 2009, 48, 7013 – 7016. Soc. 2005, 127, 2686 – 2695. [29] Rossin R, Verkerk PR, van den Bosch SM, Vulder RCM, [13] Rostovstev VV, Green LG, Foken VV, Sharpless KB. A stepwise Verel I, Lub J, Robillard MS. In vivo chemistry for pretargeted Huisgen cycloaddition process copper(I) catalyzed regiose- tumor imaging in live mice. Angew. Chem. Int. Ed. 2010, 49, lective ligation of azides and terminal alkynes. Angew. Chem. 3375 – 3378. Int. Ed. 2002, 114, 2708 – 2711. [30] Devaraj NK, Thurber GM, Keliher EJ, Marinelli B, Weissleder [14] Persolski SI, Hong V, Cho S, Finn MG. Tailored ligand R. Reactive polymer enables efficient in vivo bioorthogonal acceleration of the Cu-catalyzed azide-alkyne cycloaddition chemistry. PNAS 2012, 109, 4762 – 4767. reaction: practical and mechanistic implications. J. Am. Chem. [31] Karver MR, Weissleder R, Hilderbrand SA. Bioorthogonal Soc. 2010, 132, 14570 – 14576. reaction pairs enable simultaneous, selective, multi-target [15] Han H, Devaraj NK, Lee J, Hilderbrand SA, Weissleder R, imaging. Angew. Chem. Int. Ed. 2012, 50, 920 – 922. Bawendi M. Development of a bioorthogonal and highly [32] Yang J, Karver MR, Li W, Sahu S, Devaraj NK. Metal- efficient conjugation method for quantum dots using tetrazine- catalyzed one-pot synthesis of tetrazines directly from norbornene cycloaddition. J. Am. Chem. Soc. 2010, 132, aliphatic nitriles and hydrazine. Angew. Chem. Int. 2012, 51, 7838 – 7839. 5222 – 5225. [16] Agard NJ, Prescher JA, Bertozzi CR. A strain-promoted [3 + 2] [33] Boisselier E, Salmon L, Ruiz J, Astruc D. How to very efficiently azide-alkyne cycloaddition for covalent modification of functionalize gold nanoparticles by “ click ” chemistry. Chem. biomolecules in living systems. J. Am. Chem. Soc. 2004, 126, Commun. 2008, 5788 – 5790. 15046 – 15047. [34] Fleming DA, Thode CJ, Williams ME. Triazole cycloaddition as a [17] Agard NJ, Baskin JM, Prescher JA, Lo A, Bertozzi CR. A general route for functionalization of Au nanoparticles. Chem. comparative study of bioorthogonal reactions with azides. ACS Matter. 2006, 18, 2327 – 2334. Chem. Biol. 2006, 1, 644 – 648. [35] Sun EY, Josephson L, Weissleder R. Clickable nanoparticles for [18] Baskin JM, Prescher JA, Laughlin ST, Agard NJ, Chang PV, Miller targeted imaging. Mol. Imaging 2006, 5, 122 – 128. IA, Lo A, Codelli JA, Bertozzi CR. Copper-free click chemistry for [36] von Maltzahn G, Ren Y, Park J, Min D, Kotamraju VR, Jayakumar dynamic in vivo imaging. Proc. Natl. Acad. Sci. USA 2007, 104, J, Fogal V, Sailor MJ, Ruoslahti E, Bhatia SN. In vivo tumor cell 16793 – 16797. targeting with click nanoparticles. Bioconjugate Chem. 2008, [19] Ning X, Guo J, Wolfert MA, Boons G. Angew. Chem. Int. Ed. 19, 1570 – 1578. 2008, 47, 2253 – 2255. [37] Devaraj NK, Keliher EJ, Thurber GM, Nahrendorf M, Weissleder [20] Jewett JC, Sletten EM, Bertozzi CR. Rapid Cu-free click R. 18F labeled nanoparticles for in-vivo PET-CT imaging. chemistry with readily synthesized biarylazacyclooctynones. Bioconjugate Chem. 2009, 20, 397 – 401. J. Am. Chem. Soc. 2010, 132, 3688 – 3690. [38] Thorek DL, Elias DR, Tsourkas A. Comparative analysis of [21] Ning X, Temming RP, Dommerholt J, Guo J, Ania DB, Debets nanoparticle-antibody conjugations: carbodiimide versus click MF, Wolfert MA, Boons G, van Delft FL. Protein modification by chemistry. Mol. Imaging 2006, 8, 221 – 229. strain-promoted alkyne-nitrone cycloaddtion. Angew. Chem. [39] Elias DR, Cheng Z, Tsourkas A. An intein-mediated Int. Ed. 2010, 49, 3065 – 3068. site-specific click conjugation strategy for improved tumor [22] Blackman ML, Royzen M, Fox JM. The tetrazine ligation: targeting of nanoparticle systems. Bioconjugation 2010, 6, fast bioconjugation-based on inverse-electron-demand 2460 – 2468. Diels-Alder reactivity. J. Am. Chem. Soc. 2008, 130, [40] Colombo M, Sommaruga S, Mazzucchelli S, Polito L, Verderio 13518 – 13519. P, Galeffi P, Corsi F, Tortora P, Prosperi D. Site-specific [23] Devaraj NK, Weissleder R, Hilderbrand SA. Tetrazine-based conjugation of ScFvs antibodies to nanoparticles by cycloadditions: applications to pretargeted live cell imaging. bioorthogonal strain-promoted alkyne-nitrone cycloaddition. Bioconjugate Chem. 2008, 19, 2297 – 2299. Angew. Chem. Int. Ed. 2012, 51, 496 – 499. [24] Yang J, Seckute J, Cole CM, Devaraj NK. Live-cell imaging of [41] Liu DS, Tangpeerachaikul A, Selvaraj R, Taylor MT, Fox JM, cyclopropene tags with fluorogenic tetrazine cycloadditions. Ting AY. Diels-Alder cycloaddition for fluorophore targeting to Angew. Chem. Int. Ed. 2012, 51, 7476 – 7479. specific proteins inside living cells. J. Am. Chem. Soc. 2012, [25] Chen W, Wang D, Dai C, Hamelbeg D, Wang B. Clicking 1,2,4,5- 134, 792 – 795. tetrazines and cyclooctynes with tunable reaction rates. Chem. [42] Haun JB, Devaraj NK, Hilderbrand SA, Lee H, Weissleder R. Commun. 2012, 48, 1736 – 1738. Bioorthogonal chemistry amplifies nanoparticle binding [26] Karver MR, Weissleder R, Hilderbrand SA. Synthesis and and enhances signal detection. Nat. Nanotech. 2010, 5, evaluation of a series of 1,2,4,5-tetrazines for bioorthogonal 660 – 665. conjugation. Bioconjugate Chem. 2011, 22, 2263 – 2270. [43] Haun JB, Devaraj NK, Marinelli BS, Lee H, Weissleder R. [27] Tassa C, Liong M, Hilderbrand S, Sandler JE, Riener T, Probing intracellular biomarkers and mediators of cell Keilher EJ, Weissleder R, Shaw SY. On-chip bioorthogonal activation using nanosensors and bioorthogonal chemistry. chemistry enables immobilization of in situ modified ACS Nano 2011, 5, 3204 – 3213. M.K. Rahim et al.: Bioorthogonal chemistries for nanomaterials 227

[44] Liong M, Fernandez-Suarez M, Issadore D, Min C, Tassa C, [47] Haun JB, Castro CM, Wang R, Peterson VM, Marinelli BS, Lee Reiner T, Fortune SM, Toner M, Lee H, Weissleder R. Specific H, Weissleder R. Micro-NMR for rapid molecular analysis of pathogen detection using bioorthogonal chemistry and human tumor samples. Sci. Trans. Med. 2011, 3, 1 – 13. diagnostic magnetic resonance. Bioconjugate Chem. 2011, 22, [48] Shao H, Chung J, Balaj L, Charest A, Bigner DD, Carter 2390 – 2394. BS, Hochberg FH, Breakefield XO, Weissleder R, Lee H. [45] Chung HJ, Reiner T, Budin G, Min C, Liong M, Issadore D, Lee H, Protein typing of circulating microvesicles allows real-time Weissleder R. Ubiquitous detection of Gram-positive bacteria monitoring of glioblastoma therapy. Nat. Med. 2012, 18, with bioorthogonal magnetofluorescent nanoparticles. ACS 1835 – 1840. Nano 2011, 5, 8834 – 8841. [49] Issadore D, Chung J, Shao H, Liong M, Ghazani A, Castro CM, [46] Peterson VM, Castro CM, Lee H, Weissleder R. Orthogonal Weissleder R, Lee H. Ultrasensitive clinical enumeration of amplification of nanoparticles for improved diagnostic rare cells ex vivo using a micro-hall detector. Sci. Transl. Med. sensing. ACS Nano 2012, 6, 3507 – 3513. 2012, 4, 1 – 10.