Contemporary Approaches for Site‐Selective Dual Functionalization of Proteins

Abstract Site‐selective protein functionalization serves as an invaluable tool for investigating protein structures and functions in complicated cellular environments and accomplishing semi‐synthetic protein conjugates such as traceable therapeutics with improved features. Dual functionalization of proteins allows the incorporation of two different types of functionalities at distinct location(s), which greatly expands the features of native proteins. The attachment and crosstalk of a fluorescence donor and an acceptor dye provides fundamental insights into the folding and structural changes of proteins upon ligand binding in their native cellular environments. Moreover, the combination of drug molecules with different modes of action, imaging agents or stabilizing polymers provides new avenues to design precision protein therapeutics in a reproducible and well‐characterizable fashion. This review aims to give a timely overview of the recent advancements and a future perspective of this relatively new research area. First, the chemical toolbox for dual functionalization of proteins is discussed and compared. The strengths and limitations of each strategy are summarized in order to enable readers to select the most appropriate method for their envisaged applications. Thereafter, representative applications of these dual‐modified protein bioconjugates benefiting from the synergistic/additive properties of the two synthetic moieties are highlighted.


Introduction
Proteins are ubiquitous in Nature,s erving as the basic building blocks of life.T hey play many essential roles in amyriad of biological processes,such as molecular transport, energy conversion, inter-a nd intramolecular signaling. [1,2] Nature expands protein diversity by post-translational modifications (PTMs) after their biosynthesis in the ribosome,thus vastly enlarging their structural and functional repertoire by up to two orders of magnitude. [3,4] Inspired by Natures elegance,s cientists strived to modify proteins with diverse synthetic moieties,allowing for the creation of bioconjugates with high degree of structural perfection and new functional characteristics. [5] In this context, the past decades revealed significant progress in the development of new methodologies for siteselective protein functionalization to install the desired functionalities at pre-defined sites. [3,[6][7][8][9][10] These well-defined protein conjugates offer great prospects for aw ide range of fields including biomedicine,b ioimaging,b iosensing and materials science. [5,[11][12][13][14][15][16][17] Some promising examples include the conjugation of synthetic polymers to therapeutic proteins to improve their solubility and extend their plasma circulation half-life [13] or the attachment of anticancer drugs to antibodies forming antibody-drug conjugates (ADCs) for cell-targeted cancer therapy. [11,12] Nevertheless,a long with the increasing demand for multifunctional bioconjugates to perform more sophisticated biological studies in vitro as well as in vivo,a ppending only one type of functionality to proteins is often insufficient to customize proteins for the desired applications.F or example,d espite the clinical success of ADCs,c lassical ADCs equipped with as ingle type of anticancer drug could still suffer from low efficacy,d rug resistance,u nfavorable pharmacokinetics,i mmunogenicity,a nd the inherent hydrophobicity of the drug, which greatly hampers their further in vivo applications. [18,19] In addition, there is also an increasing demand for new therapeutic strategies which combine imaging agents and drug molecules within the protein, for example,a ntibody,f or real-time monitoring during the treatment. [20] In another example,ap rotein was functionalized with three copies of cell-targeting somatostatin peptide and an enzyme.R emarkably,t he resultant bioconjugate inhibited tumor growth already at 100-fold lower concentration than ac linically approved antibody acting via as imilar mode of action. Furthermore,c o-administration of this protein bioconjugate with an approved anticancer drug, doxorubicin, boosted its antitumor activity in ac ombination therapy approach. [21] However,t hese multifunctional protein Site-selective protein functionalization serves as an invaluable tool for investigating protein structures and functions in complicated cellular environments and accomplishing semi-synthetic protein conjugates such as traceable therapeutics with improved features.Dual functionalization of proteins allows the incorporation of two different types of functionalities at distinct location(s), whichgreatly expands the features of native proteins.The attachment and crosstalk of afluorescence donor and an acceptor dye provides fundamental insights into the folding and structural changes of proteins upon ligand binding in their native cellular environments.Moreover, the combination of drug molecules with different modes of action, imaging agents or stabilizing polymers provides new avenues to design precision protein therapeutics in areproducible and well-characterizable fashion. This review aims to give atimely overview of the recent advancements and afuture perspective of this relatively new researcha rea. First, the chemical toolboxf or dual functionalization of proteins is discussed and compared. The strengths and limitations of each strategy are summarized in order to enable readers to select the most appropriate method for their envisaged applications.Thereafter,r epresentative applications of these dual-modified protein bioconjugates benefiting from the synergistic/additive properties of the two synthetic moieties are highlighted. conjugates have been mainly achieved by statistical modification on the protein surface, [22,23] which results in heterogeneous mixtures with reduced protein activity as well as batchto-batch variations.I nv iew of biosafety,t hese limitations significantly hamper their further developments.C onsequently,t here is ap ressing need to devise new strategies to generate protein bioconjugates that exhibit higher order of structural and functional complexity but retaining structural perfection. In this regard, site-selective dual functionalization of proteins has emerged as apromising strategy to customize proteins for the respective applications. In this review,d ual functionalization is defined as the incorporation of two different functionalities into proteins in as ite-selective fashion, which is accomplished either at two different amino acids (AA) sites or at asingle AA residue at the protein surface ( Figure 1). In both cases,t he reagents as well as the sequence of the bioconjugation reactions need to be carefully considered. Dual-modified protein bioconjugates can harness the function and properties imparted by the respective payloads,w hich complement the capabilities of biomolecules and significantly expand their functional arsenal. [24] Fore xample,d ual functionalization with two chromophores at distinct sites allows real-time monitoring of changes in protein conformations and dynamics upon ligand binding in native environments or in response to certain stimuli by Fçrster Resonance Energy Tr ansfer (FRET) measurements. [25] This cannot be achieved by mono-functionalization of proteins.M oreover,anew generation of protein therapeutics/diagnostics can be derived by incorporating two different functionalities,a sexemplified by integrating adrug and an imaging agent into one platform for simultaneous therapy and diagnostics applications (theranos-tics) or inserting two different drugs into an antibody for combination therapy. [20,26,27] Owing to the emerging interest for addressing previously "unanswered" fundamental scientific questions as well as offering practical solutions for biomedical applications,s iteselective dual functionalization of proteins has seen rapid developments in the past ten years.I nt his review,w ef irst summarize the synthetic strategies for protein dual functionalization to provide atimely overview of this burgeoning field, which serves as ag uideline for the selection of the most appropriate method for the envisaged applications ( Figure 2). In addition, the rationale and principles behind these synthetic strategies as well as the strength and inherent limitations are discussed. In the last section, some representative applications of dual-functionalized protein conjugates  benefitting from the additive or even synergistic features are highlighted.

Chemical toolbox for site-selective functionalization of proteins
Site-selective protein functionalization either relies on the functionalization of canonical AAs on the protein surface or genetically encoded noncanonical AA (ncAA) bearing the appropriate bioorthogonal groups. [28] Different methodologies have been reported to achieve site-selective monofunctionalization of proteins with high efficiencyinaresidue specific manner.Bioconjugations have been accomplished at the N-terminus,t yrosine,c ysteine or serine residues,t ol ist just afew examples.Abrief summary of the commonly used chemical methods for mono-functionalization of natural AA residues is given in Table 1. [3,6,7,[29][30][31][32] Form ore detailed discussions on protein mono-functionalization, we refer the readers to other excellent reviews on this topic. [3,6,7,[29][30][31] Among the 20 AAs,u npaired cysteines have become the primary choice for functionalization owing to their low abundance at the protein surface as well as the unique nucleophilicity and versatile reactivity profile of the thiol groups. [7,[33][34][35] Besides the unpaired cysteine residues,targeting other side chain residues,s uch as disulfide bonds and methionine, have also emerged as valuable alternatives to the more commonly used cysteine-based strategies for protein functionalization ( Table 1). All these available methods potentially offer aversatile bioconjugation toolbox to achieve dual functionalization of proteins at two different AA sites.F or instance, cysteine and methionine residues are exploited for dual modification with maleimide and hypervalent iodine reagents that proceed in as equential manner without cross reactivity. [59] Despite the simplicity and convenience to achieve dual modification at two different AA residues,t here are still only few reports that address two different AA residues at the protein surface.Furthermore,judicious selection and combination of the two mono-functionalization methods is crucial as the bioconjugation reactions need to be orthogonal and also compatible with each other, to ensure high modification efficiency.T herefore,a lthough other methods involving serine,t yrosine or tryptophan potentially offer av ersatile bioconjugation toolbox that is in principle suitable for protein dual functionalization at two different AA residues,t hese reactions have not been reported yet in this context.
Alternatively,t he attachment of am ultifunctional linker containing two reactive orthogonal groups to asingle site on the protein surface,usually an exposed cysteine or adisulfide bond, also provides straightforward access for the incorporation of two different functionalities.T hese bioconjugation reagents include maleimide derivatives, [47] sulfone derivatives [37,38] and other novel reagents, [44,46] which are discussed in Section 3.
Ther ecent advances in bioorthogonal chemistry have greatly promoted the progress of site-selective protein modification. Nowadays,v arious bioorthogonal reactions have been reported with optimized reaction parameters, such as reaction rate,c atalyst type,a nd substrate stability,t o impart the desired functionalities for the envisaged applications. [75,76] Some commonly used bioorthogonal chemistries as well as selected characteristics are summarized in Figure 3. Thec opper-catalyzed azide-alkyne cycloaddition (CuAAC) bears the advantage of fast reaction kinetics but suffers from the usage of toxic copper catalyst. [77] Strain-promoted azidealkyne cycloaddition (SPAAC) utilizes the strained alkyne derivatives as reactive partner to avoid the toxic catalyst. [78] Nonetheless,the major drawbacks of SPAACare the limited water solubility of strained alkyne and slow reaction kinetics. [83] Strained cyclooctyne can also react with other 1,3dipoles,f or example,n itrones,w hich is termed as "strainpromoted alkyne-nitrone cycloaddition (SPANC)". SPANC reactions proceed rapidly with second order rate constants of up to 39 M À1 s À1 ,w hich is about 30 times faster than the SPAACreaction. [84] However,the fast reactivity is associated with the instability of the reactive nitrones,which are prone to hydrolysis in aqueous media. [84] Photoclick reactions offer the advantages of operational simplicity as well as spatial and temporal control due to the usage of light. [85] But recent evidences have shown that photoclick reactions could be limited by potential cross reactivity with, for example,amine residues,a nd its bioorthogonality still remains controversial. [86] Among all the existing bioorthogonal reactions,t he

Angewandte Chemie
Reviews inverse electron demand Diels-Alder reaction (iEDDA) stands out because of the fast reaction kinetics (rate constant of up to 10 6 M À1 s À1 ), high bioorthogonality,c atalyst-free conditions and good biocompatibility. [87] Because of its very fast reaction kinetics,i tr evolutionized bioconjugation of various biomolecules and stimulated labeling in living systems with rate constants comparable to biological reactions. [87] The currently available toolbox of diverse bioorthogonal reactions provides the basis for protein dual functionalization ensuring no cross-reactivity and high yields of the bioconjugates. Figure 4s ummarizes the different combinations of the bioorthogonal reactions that are commonly employed in the literature.D ual modification proceeds either in as equential or simultaneous fashion. Fore xample,i ft wo identical CuAAC( or iEDDA) reactions are combined, protein dual modification has to be executed in as equential manner to prevent cross-reactions,w hich otherwise would result in am ixture of products.I nc ontrast, the combination of CuAACa nd iEDDAi ntroduces two bioorthogonal tags on proteins simultaneously.Since both reactions allow quantitative conversions,d ual-modified bioconjugates can be obtained in ao ne-pot reaction without the need for purification of the single modified product. [82] 3. Synthetic strategies for dual functionalization of proteins Most synthetic strategies for protein dual modification target canonical AAs exposed at the protein surface.T hese residues are immediately accessible without the need for tedious genetic engineering of recombinant protein variants and also mitigate the risk of negative effects on proteins folding and function. In this section, dual functionalization at two different sites as well as at asingle site is discussed.

Dual functionalization at two different sites
Direct modification of two rare AA residues on the protein surface is astraightforward approach to achieve dual modification of proteins.T he combination of two different mono-functionalization methods requires stringent selection to ensure orthogonality,c ompatibility and preferably high modification efficiency.F or example,G aunt and co-workers developed am ethod based on chemoselective labeling of as ingle methionine residue with ah ypervalent iodine reagent. [59] This hypervalent iodine reagent selectively reacted with the moderate nucleophilic methionine residue in the presence of other competitive nucleophilic AA residues,w hich make it compatible and complementary to other bioconjugation strategies targeting other AA residues. This has been demonstrated by first modifying the unpaired cysteine residue of GTP-binding protein fragment Ga with am aleimide derivative in aM ichael reaction to form at hioether bond. Subsequent modification with hypervalent iodine reagent selectively addresses the thioether in amethionine residue yielding the dual modified protein bioconjugate ( Figure 5a). Notably,t he methionine modification showed high site selectivity without any cross reactivity with the thiolmaleimide conjugation. Besides that, Paavola and co-workers have also achieved dual functionalization by combining cysteine modification and an N-terminal transamination  [79] (b) CuAAC + SPAAC. [50] (c) CuAAC + iEDDA. [25] (d) SPAAC + iEDDA. [80] (e) CuAAC + Oxime ligation. [81] (f)CuAAC + iEDDA. [82] (Solid line refers to two moieties reacting with each other;d ashed line refers to their orthogonal reactivities).

Figure 5.
Site-selective protein dual modification at different AA residues. (a) Cysteine and methionine sites. [59] (b) Cysteine and disulfide sites. [89] reaction mediated by pyridoxal 5-phosphate. [88] Theperiplasmic glutamine binding protein was site-selectively functionalized by aFRET pair, in which the ligand-induced conformational movements were monitored via changes in FRET efficiency.
An alternative strategy for site-selective protein dual modification is based on the differences in the reactivity of cysteines in their free (thiol) and oxidized (disulfide) forms. [89] Disulfide bonds could be considered as protected thiols, which require activation to form the reduced free thiols for subsequent functionalization. Therefore,s ite-selective dual modification of native proteins proceeds stepwise by modifying an unpaired cysteinew ith am aleimide reagent as the initial step,followed by the disulfide reduction to liberate two additional free thiols,w hich will react with ad isulfide rebridging reagent, for example,anallyl sulfone ( Figure 5b). It is essential that the thiol-maleimide reaction should be applied first to functionalize the unpaired cysteine residue, followed by the disulfide functionalization using allyl sulfone reagents.Otherwise,the allyl sulfone reagents will also react with the unpaired cysteine,r esulting in ah eterogeneous product mixture.

Dual functionalization at asingle site
Despite the simplicity of dual modification at two distinct sites,t he availability of proteins with two different AA residues with orthogonal reactivities is rather limited. Therefore,a lternative methods have been developed to target one specific AA residue with am ultifunctional bioconjugation reagent.
Baker, Caddick, and co-workers have demonstrated protein dual modification using mono-and dibromomaleimide reagents. [47] Thef irst functionality is introduced by reacting these reagents with an accessible thiol group through an addition-elimination reaction. Subsequently,anadditional thiol conjugation introduces the second functionality (Figure 6a). Ac onceptually similar strategy utilizing dibromopyridazinediones is depicted in Figure 6a. [90] Interestingly,the native protein, for example,G rb2 adaptor protein, could be regenerated from the dual-modified conjugate after addition of phosphine or alarge excess of thiols,which opens access to the reversible modulation of proteins function or controlled release of the attached cargos,s uch as drug molecules. [90] In ar ecent example,a nother maleimide analogue,3 -bromo-5methylene pyrrolones (3Br-5MPs), was reported for cysteinespecific dual modification of proteins,w hich has comparable modification efficiency but higher cysteine specificity than the traditional maleimide reagents (Figure 6b). [91] Thed ual modification was achieved by two sequential Michael reactions.F irst, aM ichael reaction of cysteine and 3Br-5MPs generated the bioconjugate that is amenable to as econd Michael addition with another thiol, allowing protein dual functionalization at acysteine site.Due to the slow release of the second functionality,areducing reagent, for example Figure 6. Site-selective protein dual modificationa tasingle cysteine site to introduce multifunctional bioconjugation reagents. (a) Mono and dibromomaleimide, dibromopyridazinediones (from top to down) [47,90] (b) 3-Bbromo-5-methylene pyrrolones (3Br-5MPs) [91] (c) Azabicyclic vinyl sulfone [37] (d) Allyl sulfone [38] (e) Dichloro-1,2,4,5-tetrazine [44] (f)Ethynylbenziodoxolones (EBXs). [46] NaBH 4 ,w as required to retard the elimination reaction to generate as table and bioactive conjugate for subsequent applications. [91] Alternatively,v inyl sulfones are also commonly explored Michael acceptors for protein modification due to its high electrophilic properties that enable their reaction with nucleophiles on the protein surface. [92] Nevertheless,t heir application for dual functionalization was hampered by the cross-reactivity with e.g.,amino or imidazole groups generating heterogeneous products. [93] Recently,B ernardes and coworkers combined the strained [2.2.1]bicyclic systems with the vinyl sulfone systems and developed the azabicyclic vinyl sulfone reagents for dual functionalization (Figure 6c). [37] Such combination results in af ast chemoselective protein modification at the cysteine site,w hile the dienophile in the azabicyclic strained moiety concomitantly offers an opportunity for further bioorthogonal modification via iEDDAt o liberate the energy stored in the strain systems.T he second functionalization could even proceed inside living cells for selective apoptosis imaging.B esides vinyl sulfone,a llyl sulfone reagents with enhanced water solubility and higher reactivity have also been proposed as av iable strategy for dual modification in astepwise fashion. [38] By simply adjusting the pH, allyl sulfone reagents first reacted in aM ichael reaction at pH 6t oa ttach the first functionality,y ielding aconjugated ester system that reacted with the second thiolcontaining moiety at pH 8toachieve dual functionalization of proteins ( Figure 6d). [38] In addition to maleimide analogues and sulfone derivatives,o ther strategies have also been developed for dualmodification. Fore xample,G oncalves and co-workers reported that dichloro-1,2,4,5-tetrazine can undergo two successive nucleophilic aromatic substitutions to introduce the thiol-containing payload at ac ysteine residue with excellent selectivity (Figure 6e). [44] Thet etrazine linkage could serve as the second handle for subsequent bioorthgonal iEDDAr eaction, allowing the preparation of site-selective dual-modified protein conjugates.T he feasibility of this strategy has been shown by the dual labeling of the human serum albumin with am acrocyclic chelator for nuclear imaging and af luorescent probe for fluorescence imaging. [44] Despite the simplicity of this method, it could suffer from lower yield if bulky and hydrophobic functionalities needs to be incorporated. Furthermore,adifferent strategy utilizing the inherent reactivity of the hypervalent bond was also reported. Waser and co-workers showed the dual modification of proteins with ethynylbenziodoxolones (EBXs) in high efficiency and chemoselectivity by introducing two reactive groups,i.e., an azide and ahypervalent iodine (Figure 6f). [46] Dual modification was achieved via as train-release-driven cycloaddition and Suzuki-Miyaura cross-coupling of the vinyl hypervalent iodine bond with using palladium diacetate complex as catalyst.
Due to the emergence of ADCs for targeted cancer therapy,dual functionalization at disulfide site has also gained growing interest because of the presence of accessible disulfide bonds in antibodies and the antigen-binding fragment (Fab). [11,12] Previously,d ibromopyridazinediones have been extensively employed as versatile reagents for dual modification at the single cysteine site. [90] Further reports revealed that it can also serve as adisulfide rebridging reagent to introduce two bioorthogonal tags into disulfide-containing proteins,f or example,a ntibodies or antibody fragments. Chudasama et al. exploited the insertion of dibromopyridazinediones bearing two bioorthogonal tags into the disulfide bonds in full antibody and antibody Fabf ragments (the antigen-binding fragment) without perturbing the internal disulfide bonds that are vital for activity (Figure 7). [50] Such ap lug-and-play platform allowed the introduction of two functionalities via two sequential bioorthogonal reactions in am odular and efficient way,p aving the way for the nextgeneration ADCs.
Besides addressing cysteines and disulfide bonds,o ther modification strategies at less-explored AA residues have also been reported to expand the existing protein functionalization toolkit. Fore xample,t he hypervalent iodine reagents developed by Gaunt and co-workers can be combined with maleimide reagents to achieve dual functionalization at cysteine and methionine sites,w hich is described in Figure 5a.I na ddition, the hypervalent iodine reagents have also been demonstrated to show multifaceted reactivity. [59] Theelectrophilicity of the diazo sulfonium conjugate enables aphotoredox radical cross-coupling reaction with C-4 benzylated Hantzsch ester derivatives to attach the second functionality yielding dual functionalized conjugates with high conversion. [59] Achieving dual modification at one single AA site has less restriction in terms of the choice of reagents compared to the functionalization at two AA sites.O ne possible limitation of the single site strategy is that the two orthogonal groups can be sterically hindered due to close proximity on ar elatively small multifunctional linker.T his could prevent bulkier groups such as polymers to be attached onto the protein. In addition, modification besides cysteine residues is relatively unexplored, thus this field would greatly benefit from further investigations of modification strategies at other low abundant AA residues such as tyrosine,serine or the N-termininus.

Genetic engineering for dual functionalization of proteins
To expand the reactivity beyond what is offered in native proteins,reactive groups for functionalization through genetic incorporation of new AAs (canonical or noncanonical ones) Figure 7. Site-selective dual modification of proteins at the disulfide site of antibody Fabf ragment. [50] has emerged as an indispensable tool for site-selective protein dual functionalization. In this section, genetic methods to integrate new reactive canonical AAs,ncAAs or peptides tags for protein dual functionalization are summarized.

Incorporation of canonical AAs
Theexpression of recombinant proteins containing one or more point mutations is as traightforward and well-established technology. [84] Early attempts to achieve dual functionalization of proteins via genetic methods focused on the preparation of dual cysteine mutants of the target protein.
Tw oc ysteine mutations were incorporated at two distinct locations,and both thiols exhibited different reactivity toward different thiol-reactive reagents. [94] Fori nstance,C addick et al. reported the two-cysteine insertions to genetically engineered antibody mimetic proteins,s o-called designed ankyrin repeat proteins (DARPins). [94] Both thiols are carefully selected and revealed different nucleophilicity,w hich may origin from their different solvent accessibility.T his allowed for the dual functionalization to be executed in as tepwise fashion with high overall labeling efficiency ( Figure 8a). [94] Thea uthors proposed that the less reactive reagent (bromoacetamide) was reacted first with the more nucleophilic cysteine and subsequently,t he more reactive reagent (maleimide) was applied to the second, less nucleophilic cysteine. They successfully demonstrated that the differences in thiol nucleophilicity afforded ah omogeneous product with quantitative conversion at each reaction step, and no purification was needed. Although this concept for dual modification of proteins is very elegant, the delicate balance of thiol reactivity and the selection of the most appropriate cysteine mutation site to prevent heterogeneous product formation (single and dual modified products) could be very challenging.C addick and co-workers also succeeded in protein dual modification by reacting two cysteines with the same thiol-reactive reagent generating two identical sulfoniums. [95] Due to the different accessibilities of the a-protons at the two cysteine mutations,o ne of the sulfonium groups, which had ag ood solvent accessible a-proton on adjacent position, underwent a b-elimination reaction affording dehydroalanine (Figure 8b). In contrast, the other sulfonium group,w hich has as hielded a-proton next to it, remained intact because of the different protein local microenvironment. Thefunctionalities were incorporated via two different chemoselective reactions,o ffering the site-selective dualmodified protein conjugate in high yield.
Thee xamples mentioned above took advantage of the different local microenvironment offered by the native protein. Inspired by Naturese legance,P entelute and coworkers have developed as trategy to create as pecific local chemical environment for the cysteine residue by an ewly developed, fine-tuned four-amino-acid peptide sequence (FCPF), which is termed as "p-clamp" (Figure 8c). [96] This p-clamp enables the conjugation exclusively at this cysteine site with perfluoroaromatic reagents with almost quantitative conversion. Ther eaction proceeds even in the presence of other competing thiols,thus rendering this approach compatible and complementary to other thiol-conjugation strategies. [96] Dual functionalization was demonstrated on amodel protein substrate bearing ac ysteine and a p-clamp mutation, which was functionalized with ap erfluoroaryl probe first based on the p-clamp-mediated conjugation and followed by the thiol-maleimide conjugation reaction. [96]

Incorporation of ncAAs
Thei ntroduction of point mutations has certain limitations as the available functionalities and their respective reactivities could only be selected from the pool of the 20 canonical AAs.However,the cellular biosynthetic machinery can be manipulated to incorporate ncAAs that often represent structurally similar derivatives of the canonical AAs. These ncAAs allow protein labeling with unprecedented molecular precision. [97] To date,adiverse set of ncAAs with various functionalities or bioorthogonal groups has been reported for genetic incorporation into proteins. [3,84] In the auxotrophic strain, organism such as E. coli are applied that are not able to synthesize acertain amino acid required for its growth. ThencAAs need to structurally resemble the natural AA to allow binding to the respective endogenous aminoacyl-tRNAs ynthetase and to replace the natural AA in the polypeptide sequence. [84] This strategy has been mainly employed to introduce azide-or alkyne-containing methionine analogues in am ethionine-auxotrophic E. coli strain. [84] Figure 8. (a) Genetic encoding of two cysteine mutations with different nucleophilicity. [94] (b) Genetic encoding of two cysteine mutations possessingdifferent protein local microenvironment. [95] (c) Genetic encoding of acysteine mutation and a"p-clamp" FCPF peptide sequence creates an ew microenvironment for the cysteine residue. [96] Fore xample,t he Davis group incorporated the methionine analogue,azidohomoalanine (AHA), and acysteinemutation to the target protein based on the combination of site-directed gene mutagenesis and the residue-specific replacement of methionine by its analogues (Figure 9a). [98] Dual modification was accomplished through the initial cysteine conjugation with methanethiosulfonates derivatives,f ollowed by the CuAACr eaction of the azido group in AHA and ethynyl functionalities.T his approach offers the benefit of the established AHA incorporation and its efficient modification by cycloaddition reactions.H owever, the set of available ncAA is limited as they have to bind to their respective tRNA synthetase efficiently,u sing auxotrophic strains and all the AA residues within as equence will be replaced by the respective ncAA analogue.
Theg enetic code expansion technique has been developed as another technique that allows the insertion of abroad variety of ncAAs with spatial precision at virtually any desired site. [97] It is accomplished by using an orthogonal aminoacyl-tRNAs ynthetase (aaRS)/tRNAp air, which is capable of charging ad esigned ncAA in response to an onsense codon, such as the amber stop codon (UAG), due to Figure 9. (a) Site-selective incorporation of one cysteine mutation via site-directed mutagenesis and one methionine analogue via residue-specific replacementexperiment. [98] (b) Genetic encoding of one cysteine mutation via site-directed mutagenesis and one ncAA via genetic code expansion in response to UAG codon. [60] (c) Genetic encoding of two noncanonical AAs via genetic code expansion in response to UAG and UAA codon. [100] (d) Genetic encoding of two noncanonicalAAs via genetic code expansion in response to UAG and AUGA codon. [25] Angewandte Chemie Reviews their minimal occurrence in most organisms. [97,99] This strategy allows the genetic encoding of more than 150 ncAAs containing various reactive handles and functionalities. [84] Therefore,t he incorporation of ncAA via genetic code expansion in combination with the site-directed canonical AA mutation provides av ersatile strategy for protein dual functionalization. Representative work was reported by Deniz and co-workers,i nw hich they utilized an engineered tyrosyl-tRNAs ynthetase (MjTyrRS)/tRNA CUA pair (MjTy rRS/tRNA CUA pair) derived from Methanococcus jannaschi in response to the amber (UAG)stop codon to insert pacetylphenylalanine,aketone bearing ncAA, into T4 lysozyme. [101] In combination with asingle cysteinemutation, dual labeling of the T4 lysozyme mutant was demonstrated by modification with aF RET dye pair through the thiolmaleimide reaction and oxime ligation (Figure 9b). [101] Despite the simplicity of introducing ac ysteinem utation as one of the target sites for protein dual modification, this approach could become problematic if the protein consists of native cysteines that play important structural or functional roles.F urthermore,c ysteiner esidues could form disulfide bonds,w hich could limit expression yields and the reaction reversibility of the thiol-maleimide conjugation could also cause stability problems of the resulting protein bioconjugates. [36] Hence,t here have been much efforts to introduce two different ncAAs with bioothogonal tags that could be functionalized independently.L iu and co-workers applied two mutually orthogonal aaRS/tRNApairs in response to two blank codons. [100] They mutated the pyrrolysyl-tRNAsynthetase (PylRS)/tRNA CUA pair (PylRS/tRNA CUA pair) to produce anew variant, PyIRS/tRNA UUA ,which can suppress the ochre (UAA) stop codon. In combination with the evolved MjTy rRS/tRNA CUA pair,t hese two orthogonal aaRS/tRNA pairs were capable of recognizing and inserting two ncAAs, pazido-l-phenylalanine and N e -propargyloxycarbonyl-l-lysine, into ag lutamine-binding protein in response to the amber UAGc odon and the ochre UAAc odon (Figure 9c). [100] Tw o sequential CuAACr eactions were employed to install two chromophores for the protein dual modification. In as ubsequent report, the authors incorporated the azide-and ketonebearing ncAAs into GFP with the use of an evolved MjTy rRS (AzFRS)/tRNA CUA pair and PyIRS (AcKRS)/tRNA UUA pair, to eliminate protein aggregation and oxidation induced by the copper catalyst. [81] Notably,d ual functionalization was accomplished by SPAACand an oxime ligation in one-pot and catalyst-free fashion. In addition, besides in E. coli,S chultz and co-workers have also successfully incorporated two different ncAAs,w hich contained azido and ketone groups, in mammalian cells by utilizing the two orthogonal Methanosarcina barkeri pyrrolysyl-tRNAs ynthetase (MbPylRS)/ Methanosarcina mazei pyrrolysyl tRNA( MbPylRS/ MmtRNA UUA pair) and tyrosyl-tRNAs ynthetase (EcTyrRS)/tRNA CUA pair (EcTyrRS/tRNA CUA pair). [102] The dual-tagged antibody was subsequently functionalized with atoxic drug payload and afluorophore with high conversion.
Alternatively,i nstead of reassignment of the triplet stop codons,C hin and co-workers have exploited the two orthogonal aaRS/tRNAp airs in response to aq uadruplet blank codon (four-base codon) and as top codon for the incorpo-ration of two ncAAs. [103] However,n atural ribosomes suffer from very low efficiency in decoding the quadruplet codon. In this context, Chin et al. have synthetically evolved an orthogonal lysozyme (ribo-Q1), which was not responsible for synthesizing the proteome as natural lysozyme,f or selectively decoding the quadruplet codon. [79] By the combination of ribo-Q1 with two orthogonal aaRS/tRNAp airs, AzPheRS*/tRNA UCCU (a derivative of MjTy rRS/tRNA) and PylRS/tRNA CUA ,two ncAAs were site-selectively introduced into Calmodulin, forming at riazole intramolecular crosslink through the subsequent CuAACreaction. [79] Nonetheless,the major drawback of this system lies in the low efficiency and specificity of the AzPheRS*/tRNA UCCU pair in directing the corresponding ncAA incorporation. Thee fficiencyo ft he original system was substantially improved based on the evolution of the PylRS/tRNA CUA to obtain an optimized quadruplet decoding variant, PylRS/tRNA UACU pair. [25] As ap roof-of-concept, the evolved PylRS/tRNA UACU pair was combined with the AzPheRS*/tRNA CUA pair to site-selectively introduce two bioorthogonal tags,n orbornene and tetrazine,i nto Calmodulin (Figure 9d). [25] Dual modification was successfully accomplished via two sequential iEDDA reactions. [25] With the established technologies,i nt he latter example,t he authors incorporated the alkyne and cyclopropane handles into Calmodulin by using the orthogonal aaRS-tRNAp air described above,p ermitting the simultaneous dual functionalization in one pot. [82] Genetic code expansion has witnessed incredible achievements in recent years,w hich greatly promotes and revolutionizes the field of site-selective protein dual functionalization. Theadvantages of this strategy lie in the small size of the ncAAs,f lexible incorporation sites but high site-selectivity, and the diverse reactive tags available for incorporation. However,s ome important challenges still remain in this rapidly evolving field, for example,low catalytic efficiencyof engineered aaRS which require tedious evaluation and optimization to improve their performance,r elative low expression yields,r epetitive optimization of the protocols for effective protein engineering,t he scope and compatibilities of the functional groups capable to be inserted. [84] Besides developing the recombinant engineering techniques, the bioorthogonal chemistry toolbox can also be expanded so that there are more choices for compatible bioorthogonal reaction pairs in terms of reaction rate,c atalyst type,a nd substrate solubility for their intended applications.

Incorporation of peptide tags
Besides the incorporation of ncAAs bearing bioorthogonal tags for dual modification of proteins,the insertion of two artificial peptide sequences that can be recognized by two distinct enzymes for the subsequent covalent labeling with user-defined probes also serves as astraightforward approach for dual modification of proteins.T his enzyme-mediated peptide labeling combines the advantage of high specificity of the enzyme towards the peptide tags and excellent labeling efficiency with minimal perturbation to proteins structure and function. [104] Abrief summary of the commonly used peptide Angewandte Chemie Reviews tags and their respective enzymes is given in Table 2a nd relevant examples for dual modification are further discussed in this section. In ar ecent example,t wo distinct enzymes, Sortase Aa nd Butelase 1, which demonstrate orthogonal specificity to LPXTGG and HNV motif,r espectively,h ave been combined for dual modification of IgG antibody (Figure 10 a). [118] Notably,asimple centrifugation process is sufficient to obtain the pure dual-modified protein conjugate, demonstrating the high selectivity and excellent conversion of the enzyme-mediated labeling approach. In addition, peptide tags have also been utilized in combination with ncAAc [105] or "p-clamp" sequence [106] for site-selective dual-modification of proteins.F or example,C hen and co-workers have demonstrated the incorporation of aL AP peptide (LplA acceptor peptide), which can be recognized by al ipoic acid ligase (LpIA) to ligate al ipoic acid derivative,a nd an pyrrolysine analogues which bear an azido group for the site-selective dual labeling of epidermal growth factor receptor (EGFR) on living cells. [105] Despite the simplicity of ao ne-step introduction of the desired functionalities by two independent reactions,t he labeling efficiency may be greatly influenced by the steric effects and hydrophobicity of the introduced probes.T his limitation can be overcome by the incorporation of am ultifunctional scaffold possessing the bioorthogonal tags.D istefano et al. used farnesyl transferase,which can catalyze the transfer of an isoprenoid group from farnesyl diphosphate to the cysteine site of atetrapeptide (CVIA, the letter code is the abbreviation of aspecific AA), to introduce two reactive tags to the model protein, GFP (Figure 10 b). [107] Thebifunctional alkyne-aldehyde modified protein can undergo two independent bioorthogonal reactions simultaneously,offering the dual-modified conjugate in good conversion.
[b] "a" represents small aliphatic amino acids and "X" denotes one of Ala, Ser,Met or Glu residues amino acids in CaaX motif. Figure 10. Dual modification of proteins by using (a) the combinationo fSortase Aand ButelaseA . [118] (b) Farmesyl transferase through the incorporation of ad ual-functional scaffold bearing an alkyne and an aldehyde tag. [107] (c) Microbial transglutaminase through the incorporation of adual-functional scaffold bearing atetrazine and an azido tag. [80] Angewandte Chemie Reviews phobicity-masking PEG side chain in am ix-and-match manner.N evertheless,t his strategy is greatly limited by the protein substrate,w hich requires the recognition motif to be present on the native protein surface at as pecific site.

Higher-level functionalization of proteins
Driven by exciting developments in the discovery of new synthetic procedures as well as scientific curiosity,a chieving higher level of protein modification in asite-selective manner is considered crucial for advancing different fields in chemistry,b iology and material science.C ompared to mono-and dual functionalization of proteins,t riple functionalization strategies allow the incorporation of three different functionalities into proteins,t hus offering well-defined protein bioconjugates with further expanded structural and functional diversities that is beyond what current synthetic strategies can accomplish. These endeavors will greatly boost our capability to investigate,m odulate and re-design the chemical and physical properties of proteins.
However,triple functionalization of proteins is even more demanding than mono-and dual functionalization as the combination of three orthogonal chemistries have to be applied in aqueous solution and in the presence of many reactive proteinogenic groups.Furthermore,byincreasing the number of functionalities attached to the protein surface,i t may lead to potential detrimental influence to their functional and structural integrity.I na ddition, the selection of the reactive groups requires more deliberate chemical design as the increase in the hydrophobicity of the bioconjugation reagents may cause protein aggregation. Thel abeling efficiency may also be compromised due to the steric hindrance from the insertion of three different functionalities.
Recent examples have demonstrated the successful triple functionalization of proteins.Chatterjee and co-workers have reported the introduction of three ncAAs into proteins via genetic code expansion. In this work, they assigned the EcTrp, MjTy r, and Pyl pairs to suppress UGA, UAGa nd UAA codons for the incorporation of three ncAAs,5-hydroxytryptophan, p-azidophenylalanine,a nd cyclopropene-lysine,i n the engineered E. coli strain ATMW1 (Figure 11 a). [119] The triple functionalization of the protein was successfully achieved by three mutually compatible reactions,S PA AC, iEDDA, and chemoselective rapid azo-coupling reaction (CRACR), in which the electron-rich 5-hydroxyindole ring in 5-hydroxytryptopan reacts with electron-deficient aromatic diazonium ions with fast reaction kinetics and high conversion. [120] This is the first example of triple modification of proteins where three ncAAs were incorporated into the target protein in living cells,which further expands the simultaneous ncAAs coding capacity via genetic code expansion. However, this strategy utilized three different aaRS/tRNAp airs and three stop codons,w hich leaves no codon for termination of endogenous genes and therefore greatly interferes with the translation termination inside cells.S hortly after, Chin and co-workers have demonstrated the genetic encoding of three distinct ncAAs into proteins with three aaRS/tRNAp airs in ad ifferent strategy. [121] In their work, they identified new DNPyIRS/ DNPyI tRNAp airs,w hich lack the N-terminal domains,a nd revealed that they can be assigned into two classes,c lass Aa nd B, based on their sequences.Specifically, class A DNPyIRS preferentially function with class A DNPyI tRNAa nd vice versa. Next, they discovered a MmPyIRS/Spe PyI tRNAp air, in which Spe PyI tRNAi s orthogonal to both class Aa nd B DNPyIRS.I nt his context, the triply orthogonal aaRS/tRNApairs were identified, which contained the MmPyIRS/Spe PyI tRNAp air, an evolved class A DNPyIRS/ DNPyI tRNAp air and an evolved class B DNPyIRS/ DNPyI tRNAp air. Three ncAAs,N e-((tert-butoxy)carbonyl)-l-lysine,3-methyl-l-histidine and Ne-(carbobenzyloxy)-l-lysine,w ere genetically encoded into proteins in response to the UAG, AGGA and AGUA codons.However, subsequent functionalizations have not demonstrated yet. Compared to the reassignment of three stop codons,this work utilizes the combination of as top codon and quadruplet codons,w hich do not interfere with the translation termination and serve as am ore general and applicable strategy to achieve higher-level protein functionalization. Even though there is no application shown and there are still obvious limitations,f or example,l ow encoding efficiency, these examples represent the first proof-of-concept for protein triple modifications.T hey still represent significant technological advancements for the development of multifunctional protein conjugates,w hich holds immense potential for drug delivery,bioimaging,and material science.
Besides increasing the number of functionalities that can be incorporated into proteins,more complex protein bioconjugates have also been reported. Fore xample,C hudasama et al. described an efficient and modular strategy for the generation of ab ispecific antibodies as well as the dual functionalization of the resulting bioconjugate (Figure 11 b). [122] By fine-tuning the bioorthogonal chemistry employed, two dual-modified antibodies were first conjugated together via the iEDDAreaction of the two bioorthogonal handles from each antibody (Figure 11 b). Subsequently, two other bioorthogonal handles from each antibody allow for the dual functionalization of the chemically constructed bispecific antibodies.D espite these seminal studies,t he preparation of protein conjugates with higher level of modification or structural complexity remains relatively unexplored. We envision that with technical breakthrough in methodology,e xciting yet unexplored applications will be discovered, which will undoubtedly revolutionize fundamental studies and applications of proteins.

Applications
With the emergence of protein dual functionalization techniques,homogeneous protein bioconjugates with remarkable functional complexity have been achieved and studied for diverse applications.Inthe following,three major fields of applications are highlighted that have greatly benefited from these novel synthesis opportunities. Angewandte Chemie Reviews 6.1. Probing protein dynamics through Fçrster resonance energy transfer Thee xcitation energy transfer from ad onor to an acceptor chromophore,a lso termed Fçrster resonance energy transfer (FRET), allows assessing important parameters such as chromophore distances and their orientations in real time and in complex biological environments.A lthough cryo-electron microscopy is becoming apopular technique to determine structural snapshots of biomacromolecules at atomic resolution, samples need to be frozen and studies cannot be carried out in native environment. In contrast, FRET bears the advantage of easy accessibility and enables the real-time monitoring of the dynamics and conformational changes of proteins in their native environment, thus serving as an important complementary methodology for elucidating proteins structural dynamics. [123] Ac ritical requirement for FRET studies is the site-selective incorporation of the respective donor and acceptor chromophores at pre-defined locations on the protein surface without interfering with the proteins structure and function. [123] Fluorescent proteins (FPs), for example,cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP), can be genetically encoded into the target proteins and are widely utilized for the investigation of proteins inside living systems. [15] Forexample,agonist-induced activation of G-protein coupled receptor (GPCRs) is thought to cause ac onformational rearrangement of their seven transmembrane a-helices. [124] Thei nsertion of CFP and YFP to the G-protein coupled receptors (GPCR) offered the dual tagged protein bioconjugate,w hich allows real-time monitoring of the activation switch of GPCRs in living cells. [125] However,d ue to the large size of FPs (molecular weight close to 30 kDa), the CFP-YFP FRET pair was found to perturb the proteins structure and function and impair the downstream signal transduction. [125] In order to overcome these limitations,t he self-labeling peptide tags have been developed, which can bind or react with small-molecule reagents to attach the fluorophore to target proteins.F or example,t etracysteine- Figure 11. (a) Genetic incorporation of three ncAAc into protein via genetic code expansioni nr esponse to the UAA, UAG and UGA codons for trifunctionalization of proteins. [119] (b) Preparation of antibody bispecifics by the conjugation of two dual-modified antibody Fabfragments. [122] containing motif CCXXCC (C is the abbreviation of cysteine and XX can be virtually any AA sequence), can selectively react with fluorescein arsenical hairpin binder (FlAsH) to form fluorescent product (Figure 12 a). [126] Lohse and coworkers used the FlAsH-tetracysteine system to label the human adenosine A 2A receptor in as ite-specific fashion, which, in conjunction with aC FP,g enerated aF RET construct (A 2A -FlAsH-CFP) that exhibited a5-fold enhanced agonist-triggered FRET signal compared to the dual-tagged CFP and YFP conjugate (Figure 12 b). [127] In comparison to FPs,small molecule chromophores bear the advantage of small size,excellent photostability and high quantum yield. [128] So the incorporation of two chromophores into target proteins as FRET pair have represented as amore advantageous strategy to probe protein dynamics without interfering their structure and functions.O ne representative example is the introduction of two fluorophore into acalciumbinding protein, Calmodulin (CaM), via genetic code expansion to study its conformation changes as af unction of the calcium ion (Ca 2+ )c oncentration. CaM consists of two domains,the C-terminal and N-terminal domain that contain two EF-hand motifs to bind Ca 2+ . [129] Upon Ca 2+ binding,t he conformation of CaM changes,w hich affects its binding to different target enzymes. [129] Thes tructures of CaM with four Ca 2+ and without Ca 2+ have been resolved in detail via NMR or X-ray crystallography.H owever,t he transition states that dynamically occur during the Ca 2+ binding processes and that play an important role in modulating the interaction of CaM with its binding partners could not be resolved with conventional structure analysis by NMR and X-Ray. [130] Therefore,dual-tagged CaM was obtained by genetic code expansion and conjugated with the FRET pairs,BODIPY-TMR-X (Bodipy-tetramethylrhodamine,a cceptor) and BODIPY-FL (4,4-difluoro-5,7dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid, donor), respectively,t oinvestigate the local conformation changes occurring in the N-terminal domain of CaM in response to the Ca 2+ concentration (Figure 12 c). [25] After excitation at 485 nm, the dual chromophore tagged CaM reveals two distinct signal corresponding to the emission of the BODIPY-FL donor at 515 nm and the BODIPY-TMR-X acceptor at 570 nm. With increasing Ca 2+ concentrations,the emission intensity of the acceptor dye decreased while the emission of donor dye increased indicating that the FRET pair at the N-terminal domain moved apart as ar esponse of Ca 2+ binding.T he current applications utilizing FRET to follow protein dynamics are mainly applying ac ertain stimulus in test tubes rather than measurements inside cells. In the future,c ontinuous advancements in dual modification of proteins could provide valuable tools to shed light on the dynamics of proteins in living cells or even in organisms. There is much interest and progress in probing protein labeling reactions proceeding in living environments.H ow-ever,n od ual labeling reactions have been achieved inside cells that would allow FRET studies until now.

Combination therapy
ADCs have emerged as one of the most powerful and promising strategies for targeted cancer therapy. [12] This new class of biopharmaceuticals combines the exquisite targetspecificity of antibodies with ah ighly potent cytotoxic payload via various conjugation technologies and enables the selective delivery of the payload to cancer cells with minimal off-target effects. [11,12] Owing to the elegant concept, there are already four ADCs approved by the US Food and Drug Administration on the market and over 60 ADCs are in clinical trials,d emonstrating its great prospect for targeted therapeutic applications. [12] Despite the clinical success,most ADCs are still restricted to the conjugation with asingle type of payload, for example ac ertain anticancer drug, to target as pecific type of cancer cells.I nf act, this kind of single-functionalized ADCs often suffer from the unfavorable pharmacokinetics and the inherent hydrophobicity of the chemotherapeutic drugs, which could significantly limit the in vivo therapeutic efficiency. [11] Dual modification of antibodies can serve as an elegant strategy to alleviate these drawbacks,allowing for the incorporation of two complementary modalities to achieve synergistic effects.T his can be exemplified by the studies on extending the plasma circulation half-life and attenuating immunogenicity of proteins through the attachment of hydrophilic polyethylene glycol (PEG) chain to proteins.T he hydrodynamic radius of the biomolecule also increases after attaching along PEG chain, thereby mitigating the renal filtration and circumventing its degradation by proteases and recognition by the immune system, which would otherwise lead to faster plasma clearance. [131] In this context, dual modification of antibodies allows the simultaneous introduction of ac ytotoxic drug molecule and aP EG chain in as ite-selective manner to afford well-defined dualfunctional ADCs with improved pharmacokinetics.S enter and co-workers have reported that the accelerated plasma clearance of ADCs originated from their hydrophobicity, which could be reduced by introducing aP EG chain with different configuration. [132] As such, three different drug linkers were designed, in which the first drug linker 4 had no PEG chain while both the drug linker 5 and 6 consisted of au nfunctionalized PEG chain (PEG 24 ,containing 24 repeating units) but with different configuration (Figure 13 a). [132] Specially,drug linker 6 incorporated abranched scaffold. The cAC10 antibody (CD30-directed antibody) was conjugated with the three different drug linkers via thiol-maleimide chemistry offering three different ADCs,c AC10-4 (without PEG24), cAC10-5 (with PEG24) and cAC10-6 (with branched PEG24) as depicted in Figure 13 b. [132] Both the in vitro and in vivo experiments revealed an inverse correlation between hydrophobicity and tumor volume (Hodgkins lymphoma), namely that cAC10-6 exhibited slower plasma clearance and much slower tumor growth compared to cAC10-4 and cAC10-5 (Figure 13 c,d). [132] In particular, the improved pharmacokinetics and therapeutic efficiencyofthe branched conjugate cAC10-6 compared to the linear conjugate cAC10-5 highlights the importance of the hydrophilic linkers for optimizing the pharmacokinetic parameters.
ity. [12] Although the exact mechanism of drug resistance is still under much investigation, current available clinical data indicate that malignant cells,which are resistant to aparticular drug, could still respond to other drugs. [19] Therefore,t he attachment of two drugs with different modes of action to an antibody can afford more complex ADCs.T his could be codelivered to cancer cells to overcome drug resistance,which represents an emerging field in targeted cancer therapy. [18] For example,L evengood and co-workers have reported ad ualconjugation strategy for the preparation of novel ADCs including two complementary drugs. [27] In their work, two different auristatin molecules,m onomethyl auristatin E-(MMAE) and monomethyl auristatin F( MMAF), were selected to conjugate to the cAC10 antibody due to their complementary physicochemical properties and anticancer activities (Figure 14 a,b). [27] Specifically,M MAE is cell-permeable and exhibits bystander activity capable of killing neighboring antigen-negative cells. [133] However,M MAE is as ubstrate for multidrug efflux pump exporters,h ence showing diminished activity on cells with high pump expression. [134] In contrast, MMAF is susceptible to drug export, but not cell-permeable and exhibits negligible bystander effect. [133] In vitro and in vivo experiments demonstrated that the dual-auristatin ADCs were active on cells and tumors that were refractory to treatment with either of the individual component drugs (Figure 14 c,d). This work highlights the potential of dual modification of antibody for delivering two synergistic and complementary drug payloads for improved antitumor activities,which represents anotable advancement of the ADCs technology.
Furthermore,d ual modification of antibodies also provides great opportunities for simultaneous therapy and diagnostic applications,s o-called theranostics,i nw hich the antibody is conjugated to an anticancer drug and ac ontrast agent to allow tumor diagnosis and therapy at the same time.
In one representative example,Z eglis and co-workers reported the dual labeling of aH ER2-targeting trastuzumab with atoxin, MMAE, and apositron-emitting radiometal 89 Zr for theranostic applications (Figure 15 a). [26] Thei nv ivo experiment indicated that the resulting 89 Zr-trastuzumab-MMAE bioconjugate demonstrated excellent tumor targeting and therapeutic efficacy (Figure 15 b). Importantly,t he dual labeled ADCs represents atargeted drug delivery system that could be tracked in vivo using PET providing information of the in vivo biodistribution and real-time drug doses during the treatment (Figure 15 c).

Dual-modality imaging
Molecular imaging is ap owerful and invaluable tool for noninvasive visualization of physiological processes that occur in living organisms at cellular levels. [5] Until now, various modern imaging technologies,i ncluding optical imaging (OI), magnetic resonance imaging (MRI), positron emission tomography (PET) and computed tomography (CT), have been developed and widely used to monitor the structural, functional and dynamic changes in cancer tissues. [135] Each imaging modality has its own unique strength and intrinsic limitations.Consequently,combining two modal imaging methods has received increasing attention, as it allows the collection of complementary imaging data, thereby improving the reliability and accuracy of the diagnosis.F or example,PET can provide real-time images of tumor lesions as well as monitor their whole-body distribution and migration, while optical imaging can provide high-resolution imaging to support surgeons in identifying tumor margins during surgical resection. [136,137] By combining both modalities into as ingle imaging agent, doctors are able to assess the extent of the disease before and after surgery by PET imaging,m eanwhile fluorescence imaging can be utilized for image-guided surgery. [135] Site-selective dual modification of monoclonal antibodies provides an elegant chemical platform to implement the combination of ar adionuclide and af luorescent dye for dual modality PET and fluorescence imaging. One representative example focused on the conjugation of 18 F and far-red dye sulfonate cyanine 5( sCy5) to anti-prostate stem cell antigen (PSCA) cysteine mutated diabody A2 via ad ual-modality linker,o ffering the dual-modified imaging probe ( 18 F-sCy5-A2cDb) ( Figure 16). [138]1 8 F-immuno-PET showed fast and specific tumor uptake of prostate cancer xenografts,s uggesting high-contrast whole-body images with organ-level biodistribution as early as 1h after injection ( Figure 16). [138] Postmortem optical imaging confirmed highcontrast fluorescence in the PSCA-expressing tumors and excellent delineation of cancerous cellsfrom surrounding tissue ( Figure 16). Thed ual-modality imaging provides complemental data, significantly contributing to the reliable and accurate diagnostic applications.

Conclusion and Outlook
Thet echnical breakthroughs in protein bioconjugation chemistry has served as am ajor driving force for the elucidation of protein trafficking and interactions in living cells as well as the emergence of protein therapeutics.W ith the increasing need for personalized treatments to address great challenges in biomedical research such as efficient cell or organ targeting,o vercoming drug resistance and reducing systemic toxicity,t here is ah igh demand for protein therapeutics with improved features.M ultifunctional protein conjugates provide more than one functionality,a nd they are expected to enhance in vivo performance of therapeutic proteins by combining different functional groups possessing therapy,diagnostics and imaging properties in asingle protein system that is tailored to the requirements of the patient. Nevertheless,dual functionalization of proteins remains much more challenging relative to single functionalization owing to the plethora of reactive functional groups on the protein surface and the requirement for the optimal combination of different orthogonal reactions with high efficiencies.I nt his review,wehave summarized the remarkable progress in both synthetic and genetic engineering strategies that have overcome these significant hurdles and thus allow covalent functionalization of proteins with two different functionalities at distinct sites.
Undoubtedly,d ual functionalization of proteins has witnessed significant advancements over the past ten years which offers an ew arsenal of functional protein conjugates with advantages over singly-functionalized proteins in probing protein dynamics,c ombination therapy and bioimaging. We envision that tremendous efforts will continue to be devoted to enriching the current methodology toolkit by exploring new chemistries to improve the specificity as well as efficiency of dual functionalization or to achieve ah igher level of functionalization, for example,t riple functionalization. Moreover,t he proper combination of different sophisticated strategies,f or example,c hemical methods,g enetic methods,t oc apitalize the advantage of different approaches Figure 15. (a) Incorporation of atoxic drug (MMAE) and imaging reagent ( 89 Zr) for targetedt heranostic applications; (b) Biodistribution data of athymic nude mice bearing HER2-expressing BT474 breast cancer xenografts after 120 hofadministration of the corresponding bioconjugate;(c) PET images of athymic nude mice bearing HER2-expressingB T474 breast cancer xenografts after the injection of 89 Zrtrastuzumab-MMAE bioconjugate. (b, c) Adapted with permission from ref. [26].C opyright (2018) American Chemical Society. Figure 16. Incorporation of radiometal and 18 Fand sCy5 into A2cDb diabodies for PET imaging and optical imaging. Adapted with permission from ref. [138].C opyright (2019) Society of Nuclear Medicine and Molecular Imaging.

Angewandte Chemie
Reviews from the currently expanding toolbox will provide new insights for the preparation of am yriad of functional nanomaterials.U ltimately,w eb elieve that this will pave the way towards "smart" and "intelligent" protein conjugates that can self-adapt to the microenvironment of diseases and provide as elf-feedback loop to achieve an output or decision, for example,t ermination/activation. In this way,c urrent bottlenecks and challenges in therapeutic applications can be addressed with acompletely new perspective through rationale chemical design.