Bioorthogonal Ligation

Conventionally, protein-polymer conjugation is based on the use of the amines of lysine residues.^[159] However, most proteins present multiple lysines often leading to a loss of their function since critical sites are blocked, as shown in Figure 42a.^[162-163] Moreover, heterogeneous conjugate mixtures can be obtained which are often tedious to separate. For this reason, bioorthogonal ligation chemistries were developed as, by definition, they do not interfere with biological processes.^[164-167] These reactions are characterized by fast rates under physiological conditions and high specificity for just one functional group of the protein sequence.^[168] They rely on the introduction of a “handle” on a specific amino acid residue of the protein which can then be coupled to its partner installed on the polymer (Figure 42b).^[159]

Figure 42. Strategies for protein-polymer conjugation: conjugation to a) random amines of the protein and b) a specific site using bioorthogonal handles and their corresponding partners. Figure adapted from reference.^[159]

Among the strategies used to introduce a bioorthogonal handle on a specific site of the protein, we can encounter N-terminal modification. Indeed, the α-amino group of the N-terminus of a protein is more reactive compared to the sidechain ε-amino groups of lysine residues due to the lower pK_a (7.7 versus 10.5). However, the type of reaction needs to be chosen carefully in order to avoid a mixture of conjugates following unwanted side-reactions with other amines (i.e. lysines).^[159] A strategy reported by Francis and coworkers^[169]

involves the specific reaction of the terminal amine of the protein, such as BSA 99, with 2-pyridinecarboxaldehyde derivative 100 to form cyclic imidazolidinone 101 (Figure 43). Several native proteins (aldolase, lysozyme, BSA, cytochrome C, etc), with a wide range of molecular weight, were modified with reagent 100 substituted with either fluorophores, biotin or handles, such as

NH₂ NH₂

NH₂ H₂N H₂N

critical site of the protein

: NH₂-reactive functional group H₂N

alkynes and azides, to enable further transformations. In addition, these reactions were also carried out at physiological conditions and the chemical transformation did not affect the function of the protein.^[169]

Figure 43. Introduction of a bioorthogonal handle on BSA 99 (N-terminus in red) using reagent 100 to obtain modified protein 101. Figure adapted from reference.^[169]

Other strategies involve the use of enzymes which recognize a sequence in the protein and then chemically modify it,^[170-172] the incorporation of a non-natural amino acid carrying the handle into the sequence,^[173-174] or the transformation of thiols of cysteine residues (section 1.4.2.).^[168]

As far as the chemistries are concerned, among the ones that are most frequently adopted we can find CuAAC, strain-promoted azide-alkyne cycloaddition (SPAAC), Staundinger ligation, inverse-electron demand Diels-Alder reactions and carbonyl ligation.^[159,168] Reactions involving thiols and disulfides are also commonly used and will be discussed in more detail in the following chapter.

H₂N N H O

99

N O

H N

N R O

:

N N R

O

N HN

N O 100

101

The CuAAC reaction described in section 1.2.2. is also used to site-specifically conjugate polymers bearing an azide or an alkyne group to proteins containing the corresponding partner.^[174-175] There are several commercially available non-natural amino acids containing azide and alkyne handles, however, a possible limitation of CuAAC can be the difficulty to completely remove the copper catalyst necessary for the transformation.^[176] Traces of the copper catalyst can indeed cause cytotoxicity, especially for in vivo applications. This drawback has led to the development, by Bertozzi and coworkers in 2004,^[177] of a copper-free version of the cycloaddition, the so-called SPAAC where the absence of the catalyst is compensated by a highly reactive ring-strained alkyne. Usually cyclooctynes, such as 102 (Figure 44), are used as alkynes in these transformations, that, although biocompatible, exhibit slower reaction rates compared to CuAAC.[176-181]

For example Sun et al. applied SPAAC in 2016^[181] to conjugate recombinant anti-coagulant protein 103 with glycopolymer 102 to increase its serum half-life time (Figure 44). Protein 103 was engineered to contain an azide linker at the C-terminus, while a cyclooctyne group was introduced at the terminus of the lactose-containing polymer 102. The obtained glycoconjugate 104 was confirmed by electrophoresis to have been obtained in a good yield of 75% and its antithrombic activity assessed both in vitro and in vivo.^[181]

Figure 44. SPAAC between azide-containing protein 103 and glycopolymer 102, bearing a ring-strained alkyne group, to obtain conjugate 104. Figure adapted from reference.^[181]

Azide groups can also be used in another type of bioorthogonal reaction: the Staudinger ligation. Indeed, in 2000, Bertozzi and coworkers modified the known Staudinger reduction of azides with triphenylphosphine,^[182] by placing an ester group on one of the aryl substituents of the phosphine, like in derivative 105 (Figure 45).^[183] This modification allows, after reaction with azide 106 and nitrogen release, for the formation of aza-ylide intermediate 107 which then undergoes an intramolecular amide bond formation to 108. This step is crucial since the lack of the ester group would lead to hydrolysis of the aza-ylide intermediate to give the corresponding amine and phosphine oxide. On the other hand, hydrolysis of 108 leads to the stable ligation product 109, as shown in Figure 45.[168, 183]

103

N₃

N O R

102 N NN

N O R R :

HN O

O O

O HN

O HOO OHO OH

O OH HO

HO OH

4 n

104

Figure 45. Staudinger ligation: reaction between triarylphosphine 105 and azide-containing protein 106 gives intermediate 107 followed by amide 108 and final hydrolysis to give conjugate 109. Figure adapted from reference.^[168]

This type of ligation was used, for example, for the PEGylation of a protein equipped with an azido-phenylalanine residue.^[184] However, a major drawback of the Staudinger ligation is the potential cross-reactivity with thiols and disulfides since thiols can, in certain cases, reduce azides and phosphines can reduce disulfide bonds.^[168]

Another type of ligation is the inverse electron-demand Diels-Alder reaction between tetrazines and strained alkenes. It is commonly used because of its compatibility with CuAAC, allowing for the use of both orthogonal conjugation strategies.[159, 186] The reaction was developed in 2008 by Fox et al.^[187] and tetrazines, such as 110, are used as dienes, while strained alkenes, e.g. 111, represent the dienophile (Figure 46). This type of ligation has been applied to CPDs by Yao and coworkers.^[58] Indeed, tetrazine-containing thiol 110 was used as initiator in the disulfide-exchange polymerization to obtain CPD 111 which was then conjugated to modified antibody 112, containing the more reactive

trans-cyclooctene residue, affording product 113 that could be efficiently internalized due to the attachment of the CPD (Figure 46).^[58]

Figure 46. CPD-mediated antibody delivery achieved through tetrazine-alkyne ligation. Formation of tetrazine-containing CPD 111 using initiator 110, and conjugation with trans-cyclooctene antibody 112 to give adduct 113. Figure adapted from reference.^[58]

The above described strategies represent just a few selected examples of a wide range of reactions that can be applied for the formation of protein-polymer conjugates. The next chapter will focus on strategies that involve the use of thiols for biomolecule-functionalization.