Efficient and irreversible antibody–cysteinebioconjugation using carbonylacrylic reagentsBarbara Bernardim1,2, Maria J. Matos 1, Xhenti Ferhati3, Ismael Compañón3, Ana Guerreiro4,Padma Akkapeddi4, Antonio C. B. Burtoloso5, Gonzalo Jiménez-Osés3, Francisco Corzana 3,Gonçalo J. L. Bernardes 1,4*

There is considerable interest in the development of chemical methods for the precise, site-selective modification ofantibodies for therapeutic applications. In this protocol, we describe a strategy for the irreversible and selectivemodification of cysteine residues on antibodies, using functionalized carbonylacrylic reagents. This protocol is based on athiol–Michael-type addition of native or engineered cysteine residues to carbonylacrylic reagents equipped with functionalcompounds such as cytotoxic drugs. This approach is a robust alternative to the conventional maleimide technique; thereaction is irreversible and uses synthetically accessible reagents. Complete conversion to the conjugates, with improvedquality and homogeneity, is often achieved using a minimal excess (typically between 5 and 10 equiv.) of thecarbonylacrylic reagent. Potential applications of this method cover a broad scope of cysteine-tagged antibodies in variousformats (full-length IgGs, nanobodies) for the site-selective incorporation of cytotoxic drugs without loss of antigen-binding affinity. Both the synthesis of the carbonylacrylic reagent armed with a synthetic molecule of interest and thesubsequent preparation of the chemically defined, homogeneous antibody conjugate can be achieved within 48 h and canbe easily performed by nonspecialists. Importantly, the conjugates formed are stable in human plasma. The use of liquidchromatography–mass spectrometry (LC–MS) analysis is recommended for monitoring the progression of thebioconjugation reactions on protein and antibody substrates with accurate resolution.

Introduction

Protein bioconjugation has been widely used in the design of biologically active conjugates forapplications in biology and medicine1–4. Despite substantial progress in this field5–8, there is stillconsiderable interest in the development of efficient methods for site-selective chemical installation ofmodifications into proteins, especially those that can be applied to the covalent attachment ofcytotoxic drugs to antibodies1,9,10. To date, most approaches for the preparation of antibody–drugconjugates (ADCs) have used methods that typically result in heterogeneous products containing amixture of species with different drug-to-antibody ratios and, potentially, different pharmacokinetics.Current trends in ADC design are related to achieving site-selective antibody conjugation and acontrolled drug-to-antibody ratio while forming products that are stable while in circulation to avoidpremature drug release and thus side toxicity.

We have recently reported the design of a new class of Michael-acceptor reagents, carbonylacrylic(caa) derivatives, that undergo very rapid, chemoselective thiol–Michael addition with cysteineresidues on proteins11. The reaction can proceed to completion even when a single molar equivalentof synthetically accessible reagents is used at low protein concentrations, generating homogeneousconjugates that are fully resistant to degradation in human plasma. The reaction typically results in ahomogenous conjugate; most proteins do not have free cysteine residues, and, for most examples, wehave engineered the proteins to have a unique cysteine. In examples with multiple cysteines, we havefound that all available residues have reacted to form the corresponding thioether. The utility of thisreaction has been demonstrated for the labeling of proteins, including antibodies that are availablewith a free cysteine through large-scale production. The modified proteins (albumin, Annexin-V,C2A domain of synaptotagmin-I (C2Am) and the antibody Thiomab LC-V205C) retained their

1Department of Chemistry, University of Cambridge, Cambridge, UK. 2Departamento de Química, Universidade Federal de São Carlos, São Carlos,Brazil. 3Departamento de Química, Universidad de La Rioja, Centro de Investigación en Síntesis Química, Logroño, Spain. 4Instituto de MedicinaMolecular, Faculdade de Medicina, Universidade de Lisboa, Lisboa, Portugal. 5Instituto de Química de São Carlos, Universidade de São Paulo, SãoCarlos, São Paulo, Brazil. *e-mail: gb453@cam.ac.uk; gbernardes@medicina.ulisboa.pt

86 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

PROTOCOLhttps://doi.org/10.1038/s41596-018-0083-9

1234

5678

90():,;

1234567890():,;

biological activity after chemical conjugation. Using a fluorescently labeled conjugate of C2Am, whichbinds to the apoptotic biomarker phosphatidylserine, we were able to selectively image and detectapoptotic cells.

Overview of the procedureThe protocol presented here describes the preparation and use of simple carbonylacrylic acid deri-vatives to perform cysteine chemoselective and irreversible protein chemical modification. In parti-cular, we exemplify the utility of the method for the construction of antibody conjugates. Antibodiesare an important class of biotherapeutics, and their precise chemical modification to generate ADCsbearing a defined number of drugs at specific sites is critical for their therapeutic activity1,4,9. Takinginto account the importance of the conjugation method in generating ADCs with improved ther-apeutic index12, we decided to demonstrate the general utility of carbonylacrylic reagents for ADCconstruction through selective cysteine modification. The first step involves the facile synthesis offunctionalized carbonylacrylic reagents by coupling a commercially available trans-3-benzoylacrylicacid to payloads bearing a free amine handle. The resulting functionalized Michael-acceptor reagentis then used as a substrate for the modification of cysteine-tagged antibodies. Notably, complexstructures containing a cleavable linker and a cytotoxic drug are readily coupled by using a few (5–10)molar equivalents of the carbonylacrylic reagent, generating stable and fully functional antibodyconjugates.

Applications of the methodDifferent functional groups can be readily attached to the commercially available carbonylacrylicscaffold using classic amidation conditions of the corresponding carboxylic acid reagent. Manyrelevant conjugation reagents, such as fluorescent probes, PEG, carbohydrate, DNA and drug deri-vatives, are commercially available and feature a free reactive amine handle11,13. With the reagent inhand, this simple, irreversible and chemoselective approach can be directly used to modify severalantibodies at cysteines that may be either genetically encoded or generated by disulfide reduction. Inthis protocol, we exemplify the use of the method for the modification of antibodies, including nativefull-length IgG antibodies (modification at cysteines that result from the reduction of native inter-chain disulfides), engineered IgGs with genetically encoded additional cysteine residues in the lightchain, and smaller recombinant antibody fragments, such as a nanobody. For example, this methodwas recently used to create a stable nanobody–drug conjugate bearing a thioether-directingpalladium-cleavable linker for targeted bioorthogonal drug decaging13.

Comparison with other methods and limitationsCysteine is the residue of choice (over lysine) when modifying proteins because of its low abundance(1.9% for cysteine and 5.9% for lysine) and the high nucleophilicity of its sulfhydryl side chain (pKa =8.0 for cysteine and 10.5 for lysine)14. The most used strategy for cysteine modification on biomo-lecules utilizes maleimide reagents (Fig. 1a)15,16. There are, however, a number of limitations of themaleimide strategy that include the need to use a slightly acidic pH to avoid cross-reactivity withlysine residues and the use of excess equivalents to achieve complete conversion. But the mainlimitation of maleimide conjugation is that the reaction is reversible: the product undergoesretro–Michael addition reactions with biological thiols in plasma, leading to the release of the mal-eimide scaffold (Fig. 1a). Some strategies to slow the retro–Michael reaction have been proposed,including the replacement of endocyclic olefin with an exocyclic one17. However, the kinetics of thereaction is slower and requires a larger excess of reagent to achieve useful conversions on antibodies.Methods to generate stable ring-opened products by hydrolysis of the thio-succinimidyl conjugatehave been described in two steps (Fig. 1b)18. Initially, a basic amino group is installed on themaleimide structure by synthetic chemistry or by genetic engineering of the sequence adjacent tocysteine (Fig. 1b)18. Hydrolysis of the thio-succinimidyl intermediate is promoted by basic mediagenerating a stable ring-opened structure that does not undergo a retro–Michael addition reaction.Although improved in vivo stability is achieved, the requirement of extensive engineering of antibodysubstrates or incomplete hydrolysis can reduce conjugation efficiency (Fig. 1b). By contrast, thepresent method (Fig. 1c) forms conjugates that are stable in the presence of biological thiols and fullystable in plasma. In addition, the reaction shows complete chemoselectivity and proceeds even at aslightly basic pH, with no side reaction observed at competing nucleophilic amino acids such aslysine. The method has a broad substrate scope and allows the installation of a wide range of synthetic

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 87

modifications on different protein scaffolds, including antibodies, without disturbing their nativeantigen-binding properties. The method presented is straightforward and, depending on the reactivityand accessibility of the cysteine residue to be modified, reaction optimization with regard to durationand number of equivalents of the carbonylacrylic reagent may be necessary. A range from 1 to 10equiv. of the functionalized carbonylacrylic reagent, a temperature of 37 °C, and 1–6 h of reactionduration using buffer with pH 7.0–8.0 were found to provide complete conversion to the desiredhomogeneous antibody conjugates.

Experimental designOur protocol involves the use of carbonylacrylic reagents for chemoselective and irreversible cysteinebioconjugation. These reagents undergo rapid thiol-Michael addition under biocompatible conditions(aqueous buffer, near to physiological pH and temperature, and <10% (vol/vol) organic solvent)(Fig. 2a)11. We began our study by reacting the simple carbonylacrylic reagent (E)-N-ethyl-4-oxo-4-phenylbut-2-enamide (1) with three different cysteine-tagged antibodies targeting the Her2 antigen:the 2Rb17c nanobody19 (one engineered, surface-exposed cysteine, Fig. 2b), trastuzumab (onecysteine residue on the light chain and three on the heavy chain, resulting from inter-chain disulfidereduction, Fig. 2c), and Thiomab LC-V205C (one engineered cysteine in the light chain, Fig. 2d)12.

In the case of the 2Rb17c nanobody, complete conversion (>95%, based on LC–MS analysis) in a95% yield (based on antibody concentration as determined by Bradford protein assay) was typicallyobtained using 5–10 equiv. of 1 per cysteine after 2 h at 37 °C in sodium phosphate buffer (50 mMNaPi, pH 8.0) in open air (Fig. 2a). Deconvoluted mass spectra of the site-selectively modifiedantibodies are shown in Fig. 2b–d and Supplementary Figs. 1–5, indicating product homogeneity, i.e.,the starting material was fully consumed and a new product was formed with a precisely installedmodification at the intended cysteine residue. Identical data were obtained for trastuzumab–1 whenthe reaction was left for 24 h (Supplementary Fig. 6). For trastuzumab, a total of eight modificationsare installed per antibody as a result of the modification of the cysteines resulting from the reductionof four disulfides. In the cases of conjugation of a high number of synthetic modifications perantibody, and depending on the hydrophobic nature of the synthetic modifications, it may benecessary to incorporate a hydrophilic linker such as PEG into the reagent in order to reducehydrophobicity and potential aggregation of the final conjugate20.

In the case of Thiomab LC-V205C, two modifications were detected within the light chain,whereas no modifications occurred within the heavy chain (Fig. 2d and Supplementary Figs. 3 and 5).

N OO

Antibody

NH3H3N

pH 6.0–7.0

SH S

N

O

O

R-SH

Rapid retro-Michael

S

N

O

O

R

Conventional and reversible maleimide–cysteine bioconjugationa

b

c

‘Self-hydrolyzing’ maleimide–cysteine bioconjugation

N OO

S

N

O

O

H2OS

HN

O

ONH2

NH2

NH2

This work: carbonylacrylic reagents for selective and irreversible cysteine bioconjugation

NH2H3N

S–

Ph

O

O

HN

One-step, irreversiblepH 7.0–8.0

NH2H3N

S

ONH

PhO

- Cysteine selective- Mild conditions- Rapid kinetics - Easy synthesis of functionalized reagents- Low number of reagent equivalents- Low protein concentration (15 μM)- Stable products

OH

+ +NH3H3N

+ +NH3H3N

SH

+ +

+

NH3H3N

SH

+ +

pH 6.0–7.0

NH3H3N+ +

Base-promotedhydrolysis

NH3H3N+ +

+

+

Antibody Antibody

AntibodyAntibodyAntibody

Antibody

Antibody

Fig. 1 | A general approach for cysteine selective and irreversible bioconjugation of antibodies. a, Conventional and reversible maleimide–cysteinebioconjugation. b, ‘Self-hydrolyzing’ maleimide–cysteine bioconjugation. c, Carbonylacrylic reagents for selective and irreversible cysteinebioconjugation.

PROTOCOL NATURE PROTOCOLS

88 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

This is probably due to inefficient refolding of the disulfide between the light and heavy chains afterantibody production, and because 1 is small, two modifications are introduced into the light chain(Fig. 2d). This is consistent with our data that showed that when Thiomab LC-V205C is reacted withthiol-specific 5,5′-dithiobis(2-nitrobenzoic acid (Ellman’s) reagent, two modifications are detected

SH

SHHS

SH HS

SH

SHHS

HS SS

S S

S

SS

S

a

SH

Ph

O

O

HN

1 (5–10 equiv./cysteine)

2Rb17cnanobody(15 µM)

15,063

2Rb17c–1 calcd. 15,062 Da

Mass (Da)0

100

Per

cent

age

15,000 17,50012,500

Mass (Da)0

23,643100

Per

cent

age

Trastuzumab–1calcd. 23,642 Da

20,000 25,000 30,000

0

23,847100

Per

cent

age

Thiomab–1calcd. 23,846 Da(two modifications oneach light chain)

20,000 25,000

pH 8.0NaPi (50 mM)

2 h, 37 °C

Antibody Antibody S

Ph

O

O

HN

S

b

1 (10 equiv.)

pH 8.0NaPi (50 mM)

2 h, 37 °C

S

c

1 (5 equiv./cysteine)

pH 8.0NaPi (50 mM)

2 h, 37 °C

1 (5 equiv./cysteine)

pH 8.0NaPi (50 mM)

2 h, 37 °C

d

SHHS

Trastuzumab(15 µM)

Thiomab LC-V205C(15 µM)

Mass (Da)

Fre

quen

cy d

istr

ibut

ion

norm

aliz

ed to

mod

e(%

of m

ax.)

Fluorescence intensity

e f100

80

60

40

20

0

Fre

quen

cy d

istr

ibut

ion

norm

aliz

ed to

mod

e(%

of m

ax.)

100

80

60

40

20

0

105

Fluorescence intensity

104103102101100

Disulfide

1: TOF MS ES+9.59 × 106

1: TOF MS ES+1.21 × 107

1: TOF MS ES+8.79 × 104

105104103102101100

Fig. 2 | Antibody–cysteine bioconjugation with carbonylacrylic reagent 1. a, Schematic of the reaction of cysteine-tagged antibodies withcarbonylacrylic reagent 1. b–d, Scheme and deconvoluted mass spectra for the reaction of nanobody 2Rb17c (calcd. 15,062 Da; obs. 15,063 Da; b),trastuzumab (calcd. 23,642 Da; obs. 23,643 Da; c) and Thiomab LC-V205C (calcd. 23,846 Da (two modifications in each light chain); obs. 23,850 Da;d). See the Supplementary Information for the complete LC–MS analysis. e,f, Flow cytometry analysis of specificity of trastuzumab–1 (e) and Thiomab–1 (f) toward Her2+ SKBR3 cells. Red curves represent the ADC binding; blue curves correspond to secondary antibody control; and gray curvesrepresent unstained cells. ES+, positive electrospray ionization; TOF, time of flight.

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 89

within the light chain, supporting the presence of two reactive cysteine residues (SupplementaryFig. 7)21. Under identical conditions, NaPi (pH 8.0, 50 mM) and 5 equiv. per cysteine, the reaction ofThiomab LC-V205C with the commercially available N-benzylmaleimide affords <5% of the expectedconjugate, which highlights the superior reactivity of carbonylacrylic acid reagents for cysteine bio-conjugation as compared with maleimides (Supplementary Fig. 8). In the case of a native full-lengthIgG such as trastuzumab, and after total disulfide reduction, controlled installation of eight mod-ifications (one in each light chain and three in each heavy chain) was achieved using 5 equiv., of thereagent 1 per cysteine residue (Fig. 2d and Supplementary Figs. 2 and 4). Modification of the cysteineresidues resulting from inter-chain disulfide reduction provides rapid access to sites for modificationbut limits control over the positioning of the modification and leads to a high payload/antibody ratio.Furthermore, although disulfide reduction followed by conjugation of eight modifications has beenshown to generate conjugates that maintained the antibody’s binding capacity22, there may be cases inwhich the destruction of these disulfides may lead to loss of structure and biological function.

We next evaluated the stability of the antibodies conjugated with 1 in human plasma. This is importantin assessing whether the conjugate can survive the conditions while in circulation and thus avoid pre-mature release of the payload. Stability should be checked for any therapeutic conjugate before biologicalevaluation. Importantly, we observed that the thioether conjugate formed after the reaction of the sulf-hydryl side chain of cysteine with the carbonylacrylic reagent exhibited complete resilience to degradation,remaining intact at 37 °C, as demonstrated by LC–MS analysis (Supplementary Figs. 9–11).

It is important to also check that the antibody can still perform its function after conjugation. Forthe recognition of Her2 receptors displayed on the surface of Her2+ cells, SKBR3 cells were stainedwith 100 nM of trastuzumab–1 and Thiomab–1 for 1 h, at 4 °C and then stained with anti-humanIgG Alexa Fluor 647 secondary antibody. The specific binding is illustrated in the histogram FACSplots, in which red curves represent the antibody-conjugate binding, blue curves correspond to thesecondary antibody control, and the gray curves represent unstained cells (Fig. 2e,f). By flow cyto-metry analysis, we were pleased to confirm that the antibody conjugates retain their specificity towardHer2+ cells. This was also true for trastuzumab, which has been modified at the eight cysteineresidues that resulted from prior reduction of the four inter-chain disulfides. These data demonstratewell the mildness of the conjugation process and its potential for the preparation of homogeneous,stable and functional antibody conjugates.

The preparation of two derivatized carbonylacrylic reagents equipped with a drug payload,followed by chemoselective cysteine modification of an antibody of interest, is illustrated in Figs. 3and 4a,b. The first example, a carbonylacrylic-tagged ValCit-PAB-MMAE derivative (2) featuring thecleavable linker-drug ValCit-PAB-MMAE motif, which is routinely used for the preparation ofADCs, was easily synthesized from H-ValCit-PAB-MMAE and trans-3-benzoylacrylic acid (Fig. 3aand Supplementary Figs. 12–17). Drug release from ADCs (e.g., brentuximab vedotin)23 built usingValCit-PAB-MMAE is mediated by cathepsin-B hydrolysis of the protease-sensitive valine–citrulline(ValCit) dipeptide linker, followed by [1,6]-fragmentation of p-aminobenzylcarbamate (PAB)24.Using the same coupling conditions (N-methyl morpholine and isobutyl chloroformate (IBCF)), anon-cleavable carbonylacrylic drug was prepared from trans-3-benzoylacrylic acid and crizotinib(Fig. 3b and Supplementary Fig. 18). Crizotinib is an ALK tyrosine kinase inhibitor indicated forpatients with locally advanced or metastatic non-small-cell lung cancer25. These examples illustratethe synthetic accessibility to carbonylacrylic-tagged reagents for cysteine site-selective and irreversiblemodification from substrates bearing an amine handle.

With the carbonylacrylic-functionalized cleavable linker and drug in hand, we decided todemonstrate the utility of the method by site-selective conjugation of 2 to the engineered cysteine atposition 205 in Thiomab. Unexpectedly, we found that the use of one or excess molar equivalents of 2per cysteine residue did not provide useful conversions to the desired conjugate when the reaction wasperformed in 50 mM NaPi, pH 8.0 or pH 9.0, after 24 h (Supplementary Figs. 21 and 22). However,when the ionic strength of the buffer was changed to 20 mM and the pH was slightly lowered to 7.0,efficient conversion to the desired product was observed at 6 h (Fig. 4a,b and Supplementary Fig. 23).Unlike the reaction with 1, which resulted in two modifications in the light chain, a single modificationwas incorporated, which is in accordance with what has been previously described in the literaturewhen modifying Thiomab LV-V205C using maleimide-ValCit-PAB-MMAE12. Importantly, nomodifications were detected in the heavy chain, as was previously observed for the reaction with 1,which demonstrates the site selectivity of the approach. Finally, we showed that the homogeneousADC, Thiomab LC-V205C–2, retained the antibody’s native binding capacity to the Her2 antigen, asevidenced by flow cytometry analysis in high-Her2-expressing SKBR3 cells (Fig. 4c).

PROTOCOL NATURE PROTOCOLS

90 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

Next, we applied the optimal conditions found for the modification of the anti-tenascin C fullyhuman antibody F16, which has displayed promising biodistribution results in mouse models ofcancer and in patients26,27. The F16 antibody was expressed in engineered human IgG1 format, whichcontained a single cysteine residue at the light chain for site-selective conjugation, a C-terminaldisulfide that connected the heavy chains, and a substitution of an asparagine residue in the VLdomain with a glutamine that abolished N-linked glycosylation28. When F16 antibody (5 μM) wastreated with 5 equiv. per cysteine of caa-ValCit-PAB-MMAE (2), >95% conversion to an homo-geneous conjugate with a single modification within the light chain was observed using LC–MS after6h (Fig. 5a,b, Supplementary Fig. 24). No modifications were detected in the heavy chain of theantibody. Using the same conditions, the treatment of F16 antibody with 5 equiv. per cysteine of caa-crizotinib (3) furnished the desired conjugate, with complete conversion (Fig. 5a,c, SupplementaryFig. 25). These data further highlight the robustness of our method for the site-selective conjugationof drugs to tumor-homing antibodies.

Reaction scaleThe procedure can be performed on a variety of scales, with our laboratory having successfullyperformed the conjugation on individual protein and antibody samples ranging from 10 to 200 μM.In all cases, scaling up the reaction resulted in identical efficiency as monitored by LC–MS analysis.Of note, when scaling up the reaction to high protein concentrations, the amount of organic solventused should be adjusted by increasing its total amount up to 10% to avoid precipitation of the reagent.

Reaction monitoringThe use of LC–MS is recommended for monitoring the progression of the bioconjugation reactionson protein and antibody substrates with accurate resolution.

H2N

HN

O

NH

O O N

OHN

O

N

O

OMe

N

O OMe O

OHHN

NH

H2N

OH-ValCit-PAB-MMAE

NMM, IBCF, DMF, –10 °C to RT, 1 h53%

Caa–ValCit–PAB–MMAE (2)

NH

HN

O

NH

O O N

OHN

O

N

O

OMe

N

O OMe O

OHHN

O

Ph O

NH

H2N

OCaa handlefor site-selectiveand irreversible

cysteine modification

Ph

O

O

OH

Amine-bearingpayload

a

b

N N

N

N NH2

O Cl

Cl

FO

PhO

HN N

N

N NH2

O Cl

Cl

F

NMM, IBCF, DMF, –10 °C to RT 1.5 h, 60%

Ph

O

O

OH

Caa–crizotinib (3)Crizotinib

Fig. 3 | Synthesis of carbonylacrylic reagents bearing the desired payload for site-selective antibody–cysteine bioconjugation. a, The payload can beany fluorescent probe, PEG, carbohydrate, DNA or drug derivative bearing a reactive amine group. Caa-ValCit-PAB-MMAE (2) was synthesized fromH-ValCit-PAB-MMAE and trans-3-benzoylacrylic acid (see protocol 1, Supplementary Information and Supplementary Figs. 12–17 for details andcharacterization). b, Caa-crizotinib (3) was prepared in a similar manner from crizotinib and trans-3-benzoylacrylic acid (see SupplementaryInformation and Supplementary Figs. 18–20 for details and characterization). DMF, N,N-dimethylformamide, anhydrous; IBCF, isobutyl chloroformate;NMM, N-methyl morpholine.

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 91

Materials

Biological materials● SKBR3 cells (human breast adenocarcinoma cell line; ATCC, cat. no. ATCC HTB-30)

c CAUTION The cell lines used in your research should be regularly checked to ensure that they areauthentic and that they are not infected with mycoplasma.

Reagents! CAUTION Reagents used in the Procedure are potentially dangerous, and appropriate care should betaken during their manipulation. The reaction of carbonylacrylic acid with a cytotoxic payload should bealways performed with maximum care in a fume hood. The solid and liquid waste products generatedshould be disposed of appropriately, as defined locally. c CRITICAL Use all commercially availablereagents as received unless otherwise noted.● Monomethyl auristatin E (MMAE; Carbosynth, cat. no. FM63498) ! CAUTION MMAE is a highlytoxic compound, and appropriate care should be taken during manipulation, including the wearing ofpersonal protective equipment and conduction of the reaction in a fume hood.

● 9-Fluorenylmethyloxycarbonyl-valyl-citrullyl-(4-aminobenzyl)-(4-nitrophenyl) carbonate (Fmoc-Val-Cit-PAB-PNP; Iris Biotech, cat. no. ADC1000)

● 3-Benzoylacrylic acid (predominantly trans, 97% (vol/vol); Alfa Aesar, cat. no. A11147)● 1-Hydroxybenzotriazole hydrate, for procedures described in the Supplementary Information (HOBt;Sigma-Aldrich, cat. no. 54802)

● N,N-Diisopropylethylamine, for procedures described in the Supplementary Information (DIPEA; TCIChemicals, cat. no. D1599)

● N,N-Dimethylformamide anhydrous (DMF; Alfa Aesar, cat. no. 43997)● Trifluoroacetic acid (TFA; Fluorochem, cat. no. 001271)● Piperidine, for procedures described in the Supplementary Information (Alfa Aesar, cat. no. A12442)● Isobutyl chloroformate (IBCF; Sigma-Aldrich, cat. no. 177989)● N-Methyl morpholine (NMM; Sigma-Aldrich, cat. no. 67870)

2 (5 equiv./cysteine)

pH 7.0NaPi (20 mM)

6 h, 37 °C

a

SHHS

Thiomab LC-V205C(25 µM)

c

Fre

quen

cy d

istr

ibut

ion

norm

aliz

edto

mod

e (%

of m

ax.)

100

80

60

40

20

0

Fluorescence intensity

100 101 102 103 104 105

b

One drug per light chain

Mass (Da)0

24,721100

Per

cent

age

Thiomab–2calcd.

24,721 Da

20,000 25,000 30,000

1,302

1,237

1,178

1,125

1,076 1,903

1,350 1,425

1,546

1,455

0

100

Per

cent

age

1,000 1,200 1,6001,400 1,800m/z

1: Scan ES+3.35 × 106

1: Scan ES+2.80 × 105

Disulfide

Fig. 4 | Synthesis of a functional ADC through site-selective conjugation with caa-ValCit-PAB-MMAE 2.a, Schematic of the reaction of cysteine-tagged antibody Thiomab LC-V205C with 2. b, Combined ion series anddeconvoluted ESI–MS mass spectra of the light chain of Thiomab LC-V205C–2 (calcd. 24,721 Da; obs. 24,721).c, Analysis of specificity of Thiomab–2 toward SKBR3 cells by flow cytometry analysis. Red curve represents theADC binding, blue curve corresponds to secondary antibody control, and the gray curve represents unstained cells.

PROTOCOL NATURE PROTOCOLS

92 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

● Formic acid (LC–MS grade; Thermo Fisher Scientific, cat. no. 85178)● Acetonitrile for HPLC (ultragradient grade; Carlo Erba, cat. no. 412372000)● Water (deionized water, >15 MΩ/cm resistance, filtered through a 0.2-μm disc filter, Millipore (UK),cat. no. JGWP09025)

● Sodium phosphate dibasic dihydrate (NaPi; Sigma-Aldrich, cat. no. 71638)● Sodium phosphate monobasic dihydrate (NaPi; Sigma-Aldrich, cat. no. 71505)● Antibody of interest containing a cysteine to be modified. In this protocol, we use trastuzumab (20 mg/mL)in PBS (Carbosynth, cat. no. FT65040), Thiomab LC-205C (kindly provided by Genentech) and 2Rb17c-Cnanobody (kindly provided by S. Massa and N. Devoogdt, Vrije Universiteit Brussel (VUM), Brussels).

● D-MEM cell culture medium (Gibco; Life Technologies, cat. no. 21969-035)● Heat-inactivated FBS (Gibco; Life Technologies, cat. no. 10500-064)● GlutaMAX (Gibco; Life Technologies, cat. no. 35050-061)● MEM non-essential amino acids solution NEAA (Gibco; Life Technologies, cat. no. 11140-035)● Pen–Strep (Gibco; Life Technologies, cat. no. 15140-122)● HEPES (Gibco; Life Technologies, cat. no. 15630-080)● TrypLE Express stable trypsin-like enzyme (Life Technologies, cat. no. 12605-028)● Secondary antibody, goat anti-human IgG (H + L) cross-adsorbed secondary antibody, Alexa Fluor647 (Life Technologies, cat. no. A21445)

Equipment● CryoCool cooler (Neslab CryoCool CC 100 II Immersion Cooler; Gemini BV)● Magnetic stir bar● Magnetic stir plate● 10-mL Round-bottom flask● Disposable graduated 1-mL syringes (VWR, cat. no. 612-0161) and injection needles (VWR, cat. no.BRAU9166033V)

● LC–MS system for protein/antibody analysis (Acquity UPLC system (Waters H-Class) equipped with anAcquity UPLC Protein BEH C4 column, 300 Å (1.7 µm, 2.1 × 50 mm) and a photodiode array (PDA) detector)

● Semi-preparative HPLC system for peptide/caa-ValCit-PAB-MMAE (2) purification (Waters2525 system equipped with a Phenomenex Luna C18(2) column (10 μm, 250 × 21.2 mm) and adual absorbance detector)

2 or 3 (5 equiv./cysteine)

pH 7.0NaPi (20 mM)

6 h, 37 °C

a

SHHS

F16 antibody(5 μM)

cb

One drug per light chain

C-terminaldisulfide = 2 or 3

1,333

1,000 1,200 1,6001,400 1,800m/z

F16–2calcd.

23,985 Da

0

100

Per

cent

age

Mass (Da)40,00030,00020,00010,000

23,983 1: Scan ES+2.80 × 105

0

100

Per

cent

age

1,263

1,223

1,9001,845

1,500

1,412

F16–3calcd.

23,312 Da

Mass (Da)40,00030,00020,00010,000

23,307 1: Scan ES+7.75 × 107

1,296

1,000 1,200 1,6001,400 1,800m/z0

100

Per

cent

age

1,228

1,166

1,9431,794

1,457

1,372

1: Scan ES+4.76 × 106

1: Scan ES+4.76 × 106

1,111

0

100

Per

cent

age

Fig. 5 | Construction of homogeneous ADCs through cysteine-selective conjugation with caa-ValCit-PAB-MMAE (2) and caa-crizotinib (3). a,Schematic of the reaction of cysteine-tagged antibody F16 with 2 or 3. b, ESI–MS spectrum of the light chain of F16–2 (calcd. 23,985 Da; obs. 23,983).c, ESI–MS spectrum of the light chain of F16–3 (calcd. 23,312 Da; obs. 23,307).

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 93

● Analytical HPLC system for peptide/caa-ValCit-PAB-MMAE (2) analysis (Waters 1525 systemequipped with a Phenomenex Luna C18(2) column (5 μm, 250 × 4.6 mm) and a dual absorbancedetector)

● MS-ESI system for peptide/caa-ValCit-PAB-MMAE (2) analysis (Waters Acquity UPLC system,equipped with a Bruker MicrOTOF-Q detector)

● MALDI-TOF system for peptide/caa-ValCit-PAB-MMAE (2) analysis (microflex, Bruker)● MS-ESI system for protein/antibody analysis (Water Acquity UPLC system, equipped with a singlequadrupole mass detector 2)

● Lyophilizer (VirTis; Sp Scientific)● Slide-A-Lyzer dialysis cassettes (10 K MWCO, 3 mL; Thermo Scientific, cat. no. 66380)● Zeba Spin desalting column (7 K MWCO, 0.5 mL; Thermo Scientific, cat. no. 89882)● Standard biological pipettes (various sizes)● Eppendorf tubes● Laboratory centrifuge● Vortexer● Eppendorf incubator (37 °C)● EasYFlask (75 cm2 with filter; VWR, cat. no. 734-2066)● Humidified incubator set to 37 °C under 5% CO2 for cell culture● Flow cytometry tubes● Cell analyzer (BD Biosciences, LSR Fortessa)

Reagent setupAntibody reaction bufferThe reaction buffer is 20 or 50 mM sodium phosphate buffer (NaPi) at pH 7.0 or 8.0. For 1, use 20mM, pH 8.0; for 2 or 3, use 50 mM, pH 7.0. Once prepared, this buffer can be stored and used for upto 1 month at room temperature (15–25 °C).

Caa-ValCit-PAB-MMAE (2) stock solutionThe stock solution of this reagent (2.34 mM) can be prepared in N,N-dimethylformamide anhydrous(DMF) on the day of the experiment and stored at −20 °C for up to 1 week. For improved efficiencyof the conjugation, it is advised to use freshly prepared solutions.

HPLC solvent A for peptide/caa-ValCit-PAB-MMAE (2) analysis and purificationHPLC solvent A is deionized H2O and 0.1% (vol/vol) TFA. Store at room temperature for up to1 week.

HPLC solvent B for peptide/caa-ValCit-PAB-MMAE (2) analysis and purificationHPLC solvent B is CH3CN. Store at room temperature for up to 1 month.

LC–MS solvent A for protein/antibody analysisLC–MS solvent A is deionized H2O and 0.1% (vol/vol) formic acid. Store at room temperature for upto 1 week.

LC–MS solvent B for protein/antibody analysisLC–MS solvent B is 71% (vol/vol) MeCN, 29% (vol/vol) H2O, and 0.075% (vol/vol) formic acid. Storeat room temperature for up to 1 week.

Complete cell culture mediumComplete cell culture medium is D-MEM with 10% (vol/vol) FBS, 1× GlutaMAX, 1× MEM NEAA,1× Pen–Strep, and 10 mM HEPES. Store at 4 °C for up to 1 month.

Flow cytometry bufferFlow cytometry buffer is 10% (vol/vol) FBS in 1× PBS. Store at 4 °C for up to 1 month.

PROTOCOL NATURE PROTOCOLS

94 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

Equipment setupSemi-preparative HPLC for peptide/caa-ValCit-PAB-MMAE purificationDeionized water containing 0.1% (vol/vol) TFA (solvent A) and CH3CN (solvent B) are used as themobile phase for gradient elution. The gradient program for the purification of caa-ValCit-PAB-MMAE (2) is reported in the following table (flow 20 mL/min).

Time interval (min) % (vol/vol) Solvent B

0 32.5

20 42.5

30 100

Mass spectrometry for peptide/caa-ValCit-PAB-MMAE analysisMS analysis is performed using a Waters Acquity system, with parameters as displayed in the tablebelow.

Method parameter Value

Capillary voltage 4,500 V

End plate offset −500 V

Nitrogen is used as the desolvation gas and nebulizer at a total flow of 7.0 L/min. Exact massspectra are recorded using sodium formate as calibrator.

UPLC for protein/antibody analysisDeionized water containing 0.1% (vol/vol) of formic acid (solvent A) and 71% (vol/vol) MeCN, 29%(vol/vol) H2O, 0.075% (vol/vol) formic acid (solvent B) are used as mobile phase for gradient elution.The gradient program for the purification of protein/antibody is reported in the following table.

Time interval (min) % (vol/vol) Solvent B

0 28

12 71.2

13 100

16 100

16.50 28

20* 28

*Flow 0.2 mL/min from 0 to 20 min and flow 0.05 mL/min from 20 to 30 min).

Mass spectrometry for protein/antibody analysisMS analysis is performed using a Waters Acquity system, with parameters as displayed in the tablebelow.

Method parameter Value

Capillary voltage 3,000 V

Cone 20 V

Desolvation temperature 400 °C

Nitrogen is used as the desolvation gas and nebulizer at a total flow of 8.0 L/h. Exact mass spectraare recorded using sodium formate as calibrator.

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 95

Cell analyzer for binding affinity detectionFlow cytometry analysis is performed using an LSR Fortessa cell analyzer from BD Biosciences. Todetect antibody binding to SKBR3 cells, the equipment is set up with a 640-nm laser and a 670/14-nmband-pass filter (combination used for APC detection).

Procedure

Synthesis of functionalized carbonylacrylic reagent (2) ● Timing 4 h

c CRITICAL In these steps, we describe how to synthesize caa-ValCit-PAB-MMAE (2). For thesynthesis of caa-crizotinib (3), see the Supplementary Methods.1 Prepare the carboxylic acid solution. In a 10-mL round-bottom flask, dissolve 30 mg of trans-3-

benzoylacrylic acid (5 equiv.) in 1 mL of dry DMF and cool the resulting solution at −10 °C in anacetone bath cooled with a CryoCool cooler.

c CRITICAL STEP Cooling can also be achieved using an NaCl ice bath; the CryoCool system is,however, better suited to maintaining the low temperature.

2 To this solution, add 26 μL of IBCF (6 equiv.) and 22 μL of NMM (6 equiv.). Keep stirring for10 min at −10 °C to activate the carboxylic acid (solution 1).

c CRITICAL STEP Carefully control the temperature in this step.? TROUBLESHOOTING

3 In another flask, dissolve H-ValCit-PAB-MMAE in 500 μL of dry DMF and add 4 μL (1 equiv.) ofNMM (solution 2).

4 Add 500 μL of solution 1 to solution 2 and stir the reaction at −10 °C for 15 min; then let thereaction warm up to room temperature.

5 Check the reaction by MS, and when no starting material is detected (usually it takes ~1 h forcomplete conversion at room temperature), dilute the reaction mixture with 2 mL of CH3CN.For MS analysis, a typical volume of 10 μL of the reaction mixture at a concentration of10–30 μM is used.

6 Purify the crude reaction mixture by semi-preparative HPLC (32.5–42.5% of solvent B (vol/vol)over 20 min; then 42.5–100% over 10 min, with a flow of 20 mL/min). Check the peaks by MS, andcollect the fractions containing the desired product by mass (retention time: 26.8 min).

7 Lyophilize the sample to obtain the desired compound as a white solid.

c CRITICAL STEP The product must be immediately lyophilized after purification to avoiddegradation.

j PAUSE POINT The product can be stored as a solid at −20 °C for 4 months.

Procedure for antibody modification ● Timing 20–22 h8 Prepare a stock solution of caa-ValCit-PAB-MMAE (2) in an Eppendorf tube. Dissolve the desired

functionalized reagent (3 mg) in 1,000 μL of DMF to a concentration of 2.3 mM. Vortex for 10 s, ifnecessary, to obtain a homogeneous solution.

j PAUSE POINT The caa stock solution in DMF was found to be stable for ~1 week at −20 °C,depending on the structure. Caa functionalized reagents with fluorescent probes tend to be lessstable than caa reagents bearing PEG units or drugs.

9 Prepare a 20 mM solution of NaPi at pH 7.0.

c CRITICAL STEP 20 and 50 mM solutions of NaPi at pH 8.0 work for most antibodies and caareagents tested. In the case of modification using caa-ValCit-PAB-MMAE (2), the use of NaPi (20mM, pH 7.0) was critical to achieving complete conversion. When labeling high-molecular weightproteins, we advise trying a lower pH and lower-ionic-strength buffer.

10 In an Eppendorf tube, prepare the reaction medium by dissolving 12.5 μL of the antibody stocksolution (Thiomab LC-V205C at an initial concentration of 80 μM) in 23.5 μL of NaPi (20 mM, pH7.0). Vortex the mixture for 10 s until all materials are dissolved.

11 Initiate the reaction by pipetting 4.3 μL (10 equiv.) of the caa-ValCit-PAB-MMAE (2) stocksolution directly into the Eppendorf tube containing the antibody. Vortex the tube for 10 s. Totalvolume = 40 μL; final antibody concentration = 25 μM.

12 Incubate the reaction mixture at 37 °C for 6 h (shaking at 600 r.p.m.).13 Analyze the reaction outcome by LC–MS. 10-μL aliquots of antibody reaction mixture can be

directly used for LC–MS analysis. The calculated mass for the light chain is 24,721 Da, and theobserved mass was 24,721 Da (Fig. 4). No modifications are expected on the heavy chain.? TROUBLESHOOTING

PROTOCOL NATURE PROTOCOLS

96 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

14 (Optional) For further applications, reagent excess can be removed by loading the sample onto aZeba Spin desalting column pre-equilibrated with the desired storage buffer (20 mM NaPi, pH 7.0or 50 mM NaPi, pH 8.0). Perform the purification, eluting the sample by following themanufacturer’s instructions. Check the sample concentration after this procedure, using, forexample, the Bradford protein assay. In this protocol, when the reaction was scaled up for in vitroand in vivo studies, the conjugate was further dialyzed (Slide-A-Lyzer Dialysis Cassettes, 10-KMWCO, 3 mL) against three changes of 20 mM NaPi (pH 7.0) at room temperature for 12 h.

j PAUSE POINT The antibody conjugate can be stored at −20 °C for up to 6 months.

Procedure for antibody affinity detection ● Timing 49 h15 Grow the SKBR3 cells in T75 flasks with complete cell culture medium inside the humidified

incubator until 4 h before performing the flow cytometry analysis.16 Remove the cells from the culture flask by adding TrypLE Express and incubating for 5 min at 37 °C.17 Count the cells and add 1 million cells per sample.18 Centrifuge the cells to remove the medium (1,500g, 10 °C, 4 min).19 Block the cells with 100 μL of flow cytometry buffer for 1 h at room temperature.20 Centrifuge the cells as in Step 18 to remove flow cytometry buffer.21 Incubate the cells with 50 μL of a 100 nM antibody dilution in complete medium for 30 min to 1 h

at 4 °C.22 Centrifuge the cells (1,500g, 10 °C, 4 min) to remove the antibodies, and wash with 100 μL of flow

cytometry buffer.23 Repeat Step 22.24 Incubate the cells with 50 μL of secondary antibody for 30 min to 1 h at 4 °C.25 Repeat Step 22.26 Resuspend cells in 300 μL of flow cytometry buffer, and analyze the cells in the LSR Fortessa

instrument.

Troubleshooting

Troubleshooting advice can be found in Table 1.

Timing

Steps 1–4, preparation of caa-ValCit-PAB-MMAE (2): 2 hSteps 5–7, analysis and purification of reaction mixture: 2 hSteps 8 and 9, preparation of stock solution of caa and NaPi: 30 minSteps 10 and 11, preparation of antibody substrate solution: 30 min

Table 1 | Troubleshooting table

Step Problem Possible reason Solution

2 Partial by-product formation Inadequate temperaturecontrol

Lower temperatures lead to incomplete acidactivation and to the presence of a high excess ofIBCF in solution. This produces a by-product thatcannot be separated from caa-ValCit-PAB-MMAE(2) with the purification method proposed in thisprotocol. Carefully control the temperature; it shouldbe −10 °C

13 Products with a lower MWthan the starting material ora MW not observed byLC–MS

High proportion of DMFin solution

Adjust the proportion of DMF. The ratio of DMF/NaPi should be no more than 1:9. In the protocoldescribed, 1% (vol/vol) DMF was used, but identicalresults can be obtained in a range of 1–10% DMF

No or partial conversion tothe desired product

Degradation of caa stocksolution

Prepare a fresh caa solution

Incorrect ionic strengthand/or pH of the buffer

Change the ionic strength of buffer from 50 mM to20 mM and/or the pH from 8.0 to 7.0

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 97

Step 12, incubation of reaction: variable, 6–8 h depending on scaleSteps 13 and 14, analysis and purification of reaction mixture: 13 hStep 15, growing of cells: 44 hSteps 16–18, preparation of the cells: 30 minSteps 19 and 20, blocking of the cells: 1 hSteps 21–23, incubation with antibodies: 1 h 30 minSteps 24 and 25, incubation with secondary antibody: 1 hStep 26, flow cytometry analyzer: 1 h

Anticipated results

Full procedures and characterization data can be found in the Supplementary Material of this pro-tocol and also in the Supporting Information in Bernardim and co-workers8.

Following the procedure described for the activation and amidation of trans-3-benzoylacrylic acid,the carbonylacrylic reagent equipped with the synthetic modification of interest should be achieved inmoderate to good yields (~40–80%). If impurities are detected, the activation step may have partiallyfailed due to lack of temperature control. By using the obtained functionalized carbonylacrylic reagentand following the described bioconjugation procedure, site-selective, homogeneous antibodies shouldbe formed by LC–MS as single species with a >95% conversion rate and 80–95% yield of the desiredproduct, as judged by calculation from peak intensities in the deconvoluted spectrum and calculationof non-modified versus modified antibody concentration using the Bradford protein assay. In the caseof Thiomab, it should be noted that two modifications were detected in the light chain when usingcarbonylacrylic reagents with low-molecular weight, whereas a single modification was detected whenusing the larger caa-ValCit-PAB-MMAE (2) reagent. No peaks below the expected starting materialmass should be observed; if such peaks are observed, follow Table 1 to adjust the concentration ofDFM in solution (up to 10% (vol/vol)) or the buffer ionic strength (20 or 50 mM) and pH (7.0 or 8.0).

References

1. Chudasama, V., Maruani, A. & Caddick, S. Recent advances in the construction of antibody–drug conjugates.Nat. Chem. 8, 114–119 (2016).

2. Xue, L., Karpenko, I. A., Hiblot, J. & Johnsson, K. Imaging and manipulating proteins in live cells throughcovalent labeling. Nat. Chem. Biol. 11, 917–923 (2015).

3. Rodrigues, T. & Bernardes, G. J. L. The missing link. Nat. Chem. 8, 1088–1090 (2016).4. Krall, N., da Cruz, F. P., Boutureira, O. & Bernardes, G. J. L. Site-selective protein-modification chemistry for

basic biology and drug development. Nat. Chem. 8, 103–113 (2015).5. MacDonald, J. I., Munch, H. K., Moore, T. & Francis, M. B. One-step site-specific modification of native

proteins with 2-pyridinecarboxyaldehydes. Nat. Chem. Biol. 11, 326–331 (2015).6. Vinogradova, E. V., Zhang, C., Spokoyny, A. M., Pentelute, B. L. & Buchwald, S. L. Organometallic palladium

reagents for cysteine bioconjugation. Nature 526, 687–691 (2015).7. Boutureira, O. & Bernardes, G. J. L. Advances in chemical protein modification. Chem. Rev. 115,

2174–2195 (2015).8. Matos, M. J. et al. Chemo- and regioselective lysine modification on native proteins. J. Am. Chem. Soc. 140,

4004–4017 (2018).9. Beck, A., Goetsch, L., Dumontet, C. & Corvaïa, N. Strategies and challenges for the next generation of

antibody–drug conjugates. Nat. Rev. Drug Discov. 16, 315–337 (2017).10. Sievers, E. L. & Senter, P. D. Antibody-drug conjugates in cancer therapy. Ann. Rev. Med. 64, 15–29 (2013).11. Bernardim, B. et al. Stoichiometric and irreversible cysteine-selective protein modification using carbony-

lacrylic reagents. Nat. Commun. 7, 13128 (2016).12. Junutula, J. R. et al. Site-specific conjugation of a cytotoxic drug to an antibody improves the therapeutic

index. Nat. Biotechnol. 26, 925–932 (2008).13. Stenton, B. J., Oliveira, B. L., Matos, M. J., Sinatra, L. & Bernardes, G. J. L. A thioether-directed palladium-

cleavable linker for targeted bioorthogonal drug decaging. Chem. Sci. 9, 4185–4189 (2018).14. Rosen, C. B. & Francis, M. B. Targeting the N terminus for site-selective protein modification. Nat. Chem.

Biol. 13, 697–705 (2017).15. Chalker, J. M., Bernardes, G. J. L., Lin, Y. A. & Davis, B. G. Chemical modification of proteins at cysteine:

opportunities in chemistry and biology. Chem. Asian J. 4, 630–640 (2009).16. Pimenta Góis, P. M., Ravasco, J., Faustino, H. & Trindade, A. Bioconjugation with maleimides: a useful tool

for chemical biology. Chem. Eur. J. doi:10.1002/chem.201803174 (2018).17. Kalia, D., Malekar Pushpa, V. & Parthasarathy, M. Exocyclic olefinic maleimides: synthesis and application

for stable and thiol‐selective bioconjugation. Angew. Chem. Int. Ed. Engl. 55, 1432–1435 (2015).18. Lyon, R. P. et al. Self-hydrolyzing maleimides improve the stability and pharmacological properties of

antibody-drug conjugates. Nat. Biotechnol. 32, 1059–1062 (2014).

PROTOCOL NATURE PROTOCOLS

98 NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot

19. Vaneycken, I. et al. Preclinical screening of anti-HER2 nanobodies for molecular imaging of breast cancer.FASEB J. 25, 2433–2446 (2011).

20. Lyon, R. P. et al. Reducing hydrophobicity of homogeneous antibody-drug conjugates improves pharma-cokinetics and therapeutic index. Nat. Biotechnol. 33, 733–735 (2015).

21. Freedy, A. M. et al. Chemoselective installation of amine bonds on proteins through aza-Michael ligation.J. Am. Chem. Soc. 139, 18365–18375 (2017).

22. Levengood, M. R. et al. Orthogonal cysteine protection enables homogeneous multi-drug antibody–drugconjugates. Angew. Chem. Int. Ed. Engl. 56, 733–737 (2017).

23. Senter, P. D. & Sievers, E. L. The discovery and development of brentuximab vedotin for use in relapsedHodgkin lymphoma and systemic anaplastic large cell lymphoma. Nat. Biotechnol. 30, 631–637 (2012).

24. Carl, P. L., Chakravarty, P. K. & Katzenellenbogen, J. A. A novel connector linkage applicable in prodrugdesign. J. Med. Chem. 24, 479–480 (1981).

25. Cui, J. J. et al. Structure based drug design of crizotinib (PF-02341066), a potent and selective dual inhibitorof mesenchymal–epithelial transition factor (c-MET) kinase and anaplastic lymphoma kinase (ALK). J. Med.Chem. 54, 6342–6363 (2011).

26. Brack, S. S., Silacci, M., Birchler, M. & Neri, D. Tumor-targeting properties of novel antibodies specific to thelarge isoform of tenascin-c. Clin. Cancer Res. 12, 3200–3208 (2006).

27. Heuveling, D. A. et al. Phase 0 microdosing PET study using the human mini antibody F16SIP in head andneck cancer patients. J. Nucl. Med. 54, 397–401 (2013).

28. Gébleux, R., Stringhini, M., Casanova, R., Soltermann, A. & Neri, D. Non‐internalizing antibody–drugconjugates display potent anti‐cancer activity upon proteolytic release of monomethyl auristatin E in thesubendothelial extracellular matrix. Int. J. Cancer 140, 1670–1679 (2016).

AcknowledgementsWe thank FAPESP (BEPE 2015/07509-1 and 2017/13168-8 to B.B., and 2013/25504-1 to A.C.B.B.), Xunta de Galicia (M.J.M.), FCTPortugal (FCT Investigator to G.J.L.B., IF/00624/2015), the EU (Marie Sklodowska-Curie ITN Protein Conjugates, GA 675007, includinga PhD Studentship to X.F.), the Ministerio de Economía y Competitividad (projects CTQ2015-67727-R and UNLR13-4E-1931 to F.C. andCTQ2015-70524-R and RYC-2013-14706 to G.J.O.) and the Universidad de La Rioja (FPI Studentship to I.C.). We also thank S. Massaand N. Devoogdt (Vrije Universiteit Brussel (VUM), Brussels) for the generous gift of the Her2-targeting nanobody 2Rb17c, Genentechfor providing Thiomab LC-V205C and trastuzumab antibodies, and D. Neri’s laboratory (Swiss Federal Institute of Technology (ETHZürich), Zurich) for the generous gift of the F16 antibody. G.J.L.B. is a Royal Society University Research Fellow (UF110046 and URF/R/180019) and the recipient of a European Research Council Starting Grant (TagIt, GA 676832).

Author contributionsG.J.L.B. designed and supervised the research. B.B. and M.J.M. performed reactions on antibodies and MS analysis. F.C., X.F. and I.C.designed and performed the synthesis of caa-ValCit-PAB-MMAE and caa-crizotinib. A.G. and P.A. studied antibody binding andperformed flow cytometry analysis. A.C.B.B. and G.J.-O. contributed to the design of the carbonylacrylic reagent. B.B. and G.J.L.B. wrotethe paper with contributions from all authors.

Competing interestsThe authors declare no competing interests.

Additional informationSupplementary information is available for this paper at https://doi.org/10.1038/s41596-018-0083-9.

Reprints and permissions information is available at www.nature.com/reprints.

Correspondence and requests for materials should be addressed to G .J..L..B.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Published online: 23 November 2018

Related linksKey references using this protocolStenton, B. J., Oliveira, B. L., Matos, M. J., Sinatra, L. & Bernardes, G. J. L. Chem. Sci. 9, 4185–4189 (2018): http://pubs.rsc.org/en/Content/ArticleLanding/2018/SC/C8SC00256H#!divAbstractBernardim, B. et al. Nat. Commun. 7, 13128 (2016): https://www.nature.com/articles/ncomms13128

NATURE PROTOCOLS PROTOCOL

NATURE PROTOCOLS | VOL 14 | JANUARY 2019 | 86–99 |www.nature.com/nprot 99