Jorick J.
Bruins
a,
Daniel
Blanco-Ania
b,
Vincent
van der Doef
a,
Floris L.
van Delft
*a and
Bauke
Albada
*a
aLaboratory of Organic Chemistry, Wageningen University & Research, Stippeneng 4, 6708 WE, Wageningen, the Netherlands. E-mail: floris.vandelft@wur.nl; bauke.albada@wur.nl
bInstitute for Molecules and Materials, Radboud University, Nijmegen, the Netherlands
First published on 7th June 2018
Reaction of cyclopropanated trans-cyclooctene (cpTCO) with in situ generated ortho-quinone is an efficient tool for bioorthogonal protein conjugation. The (4+2)-cycloaddition of cpTCO with ortho-quinone is significantly faster than its cyclooctyne counterpart (BCN). Orthogonal, tandem cpTCO–quinone and BCN–azide cycloadditions afforded a homogeneous, dual labelled antibody–drug conjugate.
Before we embarked on the potential application of trans-cyclooctenes for SPOCQ modification of proteins, we tested if the strain-promoted click reaction between trans-cyclooctenes and ortho-quinone would take place. To this end, 4-tert-butyl-1,2-quinone 1 was mixed with cpTCO 2 (Scheme 1), and the disappearance of the absorbance that corresponds to the presence of ortho-quinone 1 at 395 nm was monitored.20 Much to our delight, the reaction of quinone 1 with cpTCO 2 not only occurs (Fig. 1 and Fig. S3–S5, ESI†), but was actually found to take place faster than that with bicyclo[6.1.0]non-4-yne (BCN)213. Specifically, the SPOCQ reaction of quinone 1 with cpTCO 2 proceeded with a rate-constant of 2900 ± 115 M−1 s−1, a near 3-fold increase versus the (corrected) rate constant of BCN 3 with quinone 1 (1112 ± 8 M−1 s−1, N = 3).12 In the same setup we found a rate constant of 10.4 ± 1.8 M−1 s−1 for the reaction of regular TCO with quinone 1. Apparently, the fused cyclopropane ring increases the rate approximately 300-fold, in analogy with enhancement of reaction rate with tetrazine.18 For comparison, cpTCO–SPOCQ is more than 103 times faster than reaction of quinone 1 with a cyclopropene (<1.7–1.9 M−1 s−1)22 and strain-promoted cycloadditions involving azides (0.01–1 M−1 s−1),7 however slightly slower than SPIEDAC reaction of regular TCO with tetrazines (2000 to >5000 M−1 s−1),13 and much slower than the fastest known cpTCO–tetrazine ligations (reaching above 106 M−1 s−1).23 Nevertheless, in contrast to azide and tetrazine, an ortho-quinone is readily generated from a canonical amino acid (tyrosine), and therefore has the inherent advantage of (inducible) bioconjugation chemistry to native proteins.
![]() | ||
Scheme 1 Reaction between quinone 1 and either exo-cpTCO 2 or endo-BCN 3 to give alkene-SPOCQ product 4 (this work) or alkyne-SPOCQ product 5.14 |
![]() | ||
Fig. 1 Determination of the reaction rate constant of quinone 1 (100 μM) with cpTCO 2 (100 μM) in H2O![]() ![]() ![]() ![]() |
After this, we explored the suitability of cpTCO-quinone cycloaddition for protein conjugation (Fig. 2 and Fig. S6–S9, ESI†). To this end, a cpTCO-functionalized lissamine–PEG conjugate was synthesized and compared head-to-head to its BCN counterpart in labelling of a model antibody (trastuzumab), which contained an exposed G4Y-tag on each heavy chain (abbreviated as Tras[HC]G4Y). Previously, we already showed that without this G4Y-tag, labelling of a smaller protein or an antibody did not proceed.12 In analogy to our previous work, we studied the alkene–SPOCQ reaction between cpTCO–PEG–lissamine and Tras[HC]G4Y upon incubation with mushroom tyrosinase (mTyr) to oxidize the exposed tyrosine phenol ring of the G4Y-tag to ortho-quinone.19 Indeed, when Tras[HC]G4Y was mixed with mTyr in the presence of cpTCO–lissamine, rapid labelling of the antibody was observed that apparently levelled off after ∼90 minutes (Fig. 2, lanes 4–11); no labelling was observed in the absence of mTyr (Fig. 2, lane 3). In the absence of a G4Y-tag on the light chain, labelling of this part of the antibody was not observed (i.e. no fluorescent bands around 25 kDa were observed), thus demonstrating (1) the bioorthogonality of alkene–SPOCQ, and (2) the need for a G4Y-tag on the protein. Based on HPLC analysis, a yield of ∼95% may be calculated (Fig. S10–S12, ESI†), with reduced side-product formation in comparison to alkyne-SPOCQ (Fig. S9, ESI†), i.e. antibody oligomers formed by cross-linking of the ortho-quinone with endogenous protein nucleophiles (Fig. 2, lane 2). The reduced formation of protein oligomers for cpTCO versus BCN further corroborates the higher reaction rate with ortho-quinones. Since oxidation of tyrosine residues by mTyr is the rate-limiting step in either SPOCQ reaction,20 the total reaction time does not decrease despite the higher reaction rate for cpTCO. However, the short lifetime of the generated quinone favours labelling with the fast reacting cpTCO by reducing side product formation.
In order to exploit the lack of reactivity of cpTCO for azides, and high reaction rate with ortho-quinones, we envisioned that tandem, dual labelling of an antibody should be accessible by orthogonal reactions of BCN and cpTCO with azide and (exposed) tyrosine, respectively. To test this, we explore the possibility of tandem installation of two different functional moieties onto a glycan-remodelled monoclonal antibody24 by cpTCO–SPOCQ and BCN–SPAAC, executed in either order. We started by the cpTCO–SPOCQ reaction on the Tras[HC]G4Y C-terminus followed by SPAAC modification of azido-glycan with BCN–MMAE (method A), but also checked method B (Fig. 3A).‡
In order to obtain the same conjugate via both routes, we used cpTCO–PEG–lissamine and BCN–sulfamide–MMAE (see ESI† for structures) with a cathepsin-cleavable linker (MMAE = monomethyl auristatin E, a potent antimitotic agent that inhibits cell division). In addition, the sulfamide spacer in the BCN unit enhanced solubility and conjugation efficiency.25
Thus, clean and efficient enzymatic glycan remodelling was achieved by endoglycosidase and glycosyltransferase, to install 6-azido-GalNAc onto the core-GlcNAc of transiently expressed Tras[HC]G4Y (Fig. 3A and Fig. S13, ESI†).14 Next, a SPOCQ reaction between cpTCO–lissamine and the G4Y-tag on the antibody heavy chain was performed in the presence of mTyr (Fig. S14, ESI†). The observed mass of 25768 Da (Fig. 3B, yellow peak) corresponds to the SPOCQ product of the azido-glycan antibody and cpTCO–lissamine (expected: 25
766 Da), and corroborates the complete lack of reactivity of the cpTCO moiety with the azido-group on the glycan chain. Subsequently, SPAAC modification of the azido-functionalized glycan chain with BCN–MMAE resulted in the clean formation of the dual labelled fluorescent antibody–drug conjugate. HPLC analysis showed almost complete conversion (Fig. S15, ESI†), and MS analysis identified the presence of the target compound with a FW of 26
516 Da (expected: 27
275 Da) (Fig. 3B, green peak in top spectrum). The difference between the observed an expected mass corresponds to loss of the C-terminal part of MMAE due to fragmentation in the mass chamber (Δ 762 Da).
Lastly, the tandem labelling steps were performed in reverse order, i.e. first SPAAC reaction on the remodelled azido-glycan chain using BCN–MMAE SPAAC (Fig. S16, ESI†), then cpTCO–lissamine SPOCQ reaction on the G4Y-tag upon tyrosinase oxidation (Fig. S17, ESI†). As expected, the same final product is obtained, albeit via a different intermediate product. We were happy to find that in this case, besides the mass of the linker-fragmented final conjugate (at 26514 Da), the mass of the desired dual labelled ADC is also observed at 27
279 Da (expected: 27
275 Da) (Fig. 3B, lower set of MS spectra). This difference is presumably due to some experimental variability during the MS analysis.
In conclusion, we have uncovered a useful addition to the toolbox of copper-free click chemistry, i.e. the rapid reaction of strained alkenes and ortho-quinones, and used this reaction to establish a convenient and high-yielding new strategy for the labelling of antibodies. The rate of the SPOCQ reaction between cpTCO and ortho-quinone is ∼2.5 times higher than that for BCN, ∼300 times higher than that for TCO (Fig. S30 and S31, ESI†), and >103 times higher than that for cyclopropene; the last two strained alkenes would be too slow for efficient SPOCQ labelling of proteins, due to the high reactivity of the quinone, e.g. with competing nucleophiles. This high reactivity of cpTCO enables faster and cleaner conjugation, as superior labelling of a G4Y-derivatized antibody with a fluorescent moiety was demonstrated when compared to its BCN counterpart. Ultimately, we developed a convenient protocol for the dual labelling of an antibody with a toxic payload and a dye using the orthogonal character of cpTCO and BCN, which may find application in the assessment of internalization potential of glycan-conjugated antibody–drug conjugates for targeted therapy.25 Our setup allows a convenient preparation of a glycan-remodelled azido-containing antibody that also contains a peptidic G4Y tag, and to use this in a two-step dual-labelling strategy to obtain homogeneous ADCs. We note that other methods have been reported for metal-free click conjugation to unnatural amino acids genetically encoded into monoclonal antibodies, for example p-acetylphenylalanine26 or p-azidomethylphenylalanine,27 followed by oxime ligation or SPAAC, respectively. However, low antibody titers (1 g L−1 maximum), which are significantly lower than expression titers on monoclonal antibodies based on canonical amino acids alone (3–7 g L−1), limit such applications. Similar arguments apply to genetic encoding of L-DOPA into proteins, which is also suitable for SPOCQ modification.28 We expect that our current method will find widespread application, not only in the preparation of ADCs but extending to other applications in the field of bioconjugation, such as PEGylation, immobilization, protein–protein dimer formation, and beyond. Efforts along those lines are currently ongoing in our laboratory.
Footnotes |
† Electronic supplementary information (ESI) available: Detailed experimental procedures, synthetic protocols, HR-MS analysis of probes, MS-analysis of conjugates, results of control experiments. See DOI: 10.1039/c8cc02638f |
‡ This work is funded by the NWO Gravity Program Institute for Chemical Immunology (ICI). We thank Synaffix for providing the reagents and expertise to perform the glycan remodelling and metal-free conjugation chemistry, in particular Dr R. van Geel. |
This journal is © The Royal Society of Chemistry 2018 |