Protein Delivery with GlycoCPDs

As shown in the introduction, a strategy for protein delivery with CPDs was developed in the group in 2016.^[56] By interfacing CPDs with biotin-streptavidin technology, efficient delivery of very large cargos, such as quantum dots and nanobodies (around 15 nm diameter) was achieved.^[57] The biotinylated residue necessary for the binding to streptavidin is introduced at the initiator level and

monomer 24 was used to obtain CPDs that could then be conjugated to the protein.^[56-57]

With the success of glycoCPDs, especially glucose-functionalized CPDs, we were intrigued by the possibility of applying the technique of glycosylation through sidechain modification to deliver proteins. In particular, we wanted to evaluate if their excellent solubility and multifunctional uptake compared to unmodified polymers could also enable the delivery of bigger cargos, such as streptavidin with a molecular weight of around 150 kDa versus the 10 kDa of the CPD.

In order to test this, we decided to compare the cellular uptake of two streptavidin adducts containing either an unmodified CPD or a glycoCPD. The first step was, therefore, to synthesize the necessary biotinylated polymers. As shown in Scheme 32, a fluorescent initiator containing a biotin residue was synthesized first. The synthetic route to obtain initiator 224 was slightly modified compared to the one previously reported in the group.^[57] The first step was to form the activated ester of biotin 225 using N-hydroxysuccinimide obtaining derivative 226 which was then reacted with ethylenediamine to obtain 227, as reported in literature.^[230] We then prepared analogously activated ester 228,^[231]

using protected cysteine 229, which was coupled to amine 227 to obtain key intermediate 230 in excellent yield. Removal of both trityl and Boc protecting groups of 230 in acidic conditions afforded derivative 231 which was then coupled to fluorophore 150 following reported procedures,^[57] to give initiator 224.

Scheme 32. Reagents and conditions: a) N-hydroxysuccinimide, EDCI, DMF, rt,

Having obtained biotinylated initiator 224, we proceeded with polymerization using unmodified 24 and azide 160 as monomers and iodoacetamide 23 as terminator, as shown in Scheme 33. CPD 232 was obtained by following reported procedures at 25 °C using monomer 24,^[56-57] while for biotinylated azide CPD 233, monomer 160 was first dissolved at 80 °C in DMF and then initiator 224 was added. Polymerization was conducted at 80 °C in aqueous TEOA buffer and, after 30 min, terminator 23 was added. Purification of the reaction mixture by gel permeation chromatography followed by desalting afforded pure CPDs 232-233.

Scheme 33. Reagents and conditions: a) DMF, TEOA buffer (1.0 M, pH 7.0), 25 °C (232) or 80 °C (233), 1000 rpm, 0.5 h.

CPDs 232-233 were then characterised using GPC to obtain the values of

bAverage values ± errors obtained for 8 independently prepared polymers. ^cMass average molecular weight. ^dNumber average molecular weight.

Scheme 34. Reagents and conditions: a) CuSO₄×H₂O, sodium ascorbate, TBTA, THF/H2O 1:9, 25 °C, 1000 rpm, 1 h, 60-78%.

In order to introduce the glucose unit on the CPD sidechain, CuAAC was performed on azide polymer 233 with glucose alkyne 208. Using the same conditions used on azide 165, biotinylated glycoCPD 234 was obtained in good yield, as shown in Scheme 34.

The prepared biotinylated CPDs 232 and 234 were then conjugated to streptavidin to form the complexes that were then to be evaluated for cellular uptake in HeLa cells. Since streptavidin is a tetrameric protein, four sites were available for the functionalization. We decided to prepare the adducts using the least favourable ratio of polymer to protein of 1:1, meaning functionalization of only one site of the tetramer with the CPD, in order to evaluate the efficiency introduced by the carbohydrate. The other free sites were then occupied by a model cargo, biotinylated fluorophore 235, which was available in the group.

Figure 104 shows the building blocks necessary for the formation of the adducts with streptavidin.

Figure 104. Schematic representation of the building blocks needed for the formation of the protein-CPD adducts: streptavidin 27, biotinylated green fluorophore 235, glycoCPD 234 and reference CPD 232.

The formation of the complexes with streptavidin was conducted in two steps using stoichiometric amounts of each reagent, as shown in Scheme 35.

First, the CPD (232 or 234) was added to streptavidin in a 1:1 ratio and shaken S

free three sites available in the protein. The resulting adducts 236-237 were used without any purification in cellular uptake experiments.

Scheme 35. Reagents and conditions: a) MES buffer (50 mM, 150 mM NaCl, 10% glycerol, pH 5.5), 4 °C, 4 h, 1000 rpm; b)4 °C, 4 h, 1000 rpm.

235 232 234

27 a)

238 239

236 237

b)

We then evaluated the cellular uptake of adducts 236-237 in HeLa cells.

After 6 h incubation at 37 °C, overall a relatively weak red and green fluorescence was observed for adduct 236 bearing a non-glycosylated CPD, as shown in Figure 105. The distribution of the complexes was uneven, with some cells exhibiting punctuate emission from the cytosol and others showing mainly membrane staining.

Figure 105. CLSM images of HeLa cells after 6 h incubation with 625 nM concentration of adduct 236 with Hoechst 33342 added the last 10 min of incubation at 37 °C in Leibovitz’s medium. a) From left to right: Hoechst 33342 (blue, LP = 6.5%); green fluorophore in 235 (green, LP = 0.4%); red fluorophore in CPD 232 (red, LP = 0.4%). b) Merged images from left to right: Hoechst 33342 with 235; Hoechst 33342 with 232; 235 with 232. Scale bar = 10 µm.

In sharp contrast, the cellular uptake of adduct 237 containing the glycosylated CPD was far superior to 236 which lacked the carbohydrate moiety.

The protein complex localized mainly in the cytosol and nucleoli, with faint emission coming from the nucleus as well, as shown in Figure 106. A strong colocalization could be observed for both green and red fluorophores, with some

a)

b)

membrane. This could be explained by the fact that a disulfide was used to connect the green fluorophore and the biotin of the model substrate. Therefore, disulfide reduction, operated by glutathione, would liberate the green emitting fluorophore inside the cytosol before reaching the nucleoli. In the case of CPD 234, reductive depolymerization to liberate the red fluorophore enabled it to efficiently reach the nucleoli where it was most likely attracted by polyanionic DNA. On the other hand, more green compared to red fluorescence could also be observed on the cell membrane, suggesting that disulfide exchange with exofacial thiols liberated the fluorophore that could not enter into the cell if not attached to the CPD.

Figure 106. CLSM images of HeLa cells after 6 h incubation with 625 nM concentration of adduct 237 with Hoechst 33342 added the last 10 min of incubation at 37 °C in Leibovitz’s medium. a) From left to right: Hoechst 33342 (blue, LP = 6.5%); green fluorophore in 235 (green, LP = 0.4%); red fluorophore in CPD 234 (red, LP = 0.4%). b) Merged images from left to right: Hoechst 33342 with 235; Hoechst 33342 with 234; 235 with 234. Scale bar = 10 µm.

a)

b)

Finally, the last step was to investigate which mechanisms of uptake were involved in the efficient delivery of the protein complex 237 into HeLa cells. In particular, we were interested in evaluating whether, even with such a considerable large cargo, thiol-mediated and glucose-mediated uptake were involved. For this purpose, we decided to use confocal laser scanning microscopy and a pre-incubation of 30 minutes with either Ellman’s reagent DTNB to oxidize exofacial thiols, or D-glucose, to saturate the binding sites of glucose receptors.

As shown in Figure 107, cellular uptake of adduct 237 was inhibited in both cases. The contrast between images taken in the absence or in the presence of inhibitors proved that the efficiency of uptake when using glycosylated CPDs as delivery systems stemmed from the combination of counterion-, thiol- and glucose-mediated mechanism of entry. Even if counterion-mediated uptake was not directly proven, previous insensitivity towards endocytosis inhibitors of streptavidin adducts with CPDs suggested that also with this similar construct direct translocation would be taking place.

Figure 107. CLSM images of HeLa cells after 6 h incubation with 625 nM concentration of adduct 237 and after 30 min pre-incubation with a) no inhibitors, b) 1.2 mM DTNB and c) 100 mM D-glucose. LP = 4%; scale bar = 10 µm.

In conclusion, sidechain engineering was applied to solve the solubility issue occurring with standard CPDs at high concentration. The superior uptake efficiency was evaluated qualitatively by confocal microscopy showing no

a) b) c)

units on their sidechain, and quantitatively by flow cytometry which revealed the highest fluorescence counts for CPDs containing glucose units.

Moreover, mechanistic investigations were carried out using inhibitors of thiol-mediated uptake and carbohydrate-mediated uptake, revealing an efficient combination of interactions of the glycosylated CPDs with exofacial thiols, anions and specific carbohydrate receptors. This multifunctional cellular uptake also enabled the delivery to cytosol and nuclei of proteins carrying model substrates interfacing CPDs with biotin-streptavidin technology.

3.3. Terminator Engineering in Cell-Penetrating