Next-generation disulfide stapling: reduction and functional re-bridging all in one

A next-generation disulfide stapling reagent, incorporating both reducing and re-bridging functions, is shown to be successful across various proteins.

In view of the above, we set about designing a reagent that could incorporate both reducing and re-bridging functions (Fig. 1). During the course of our previous studies, we observed dithiophenolpyridazinediones to be unreactive towards commonly used disulde reducing agent tris(2-carboxyethyl) phosphine (TCEP). In light of this, TCEP moieties were a logical choice for incorporation into a dithiophenolpyridazinedione. More specically, the TCEP functional moieties were to be tethered onto the thiophenol groups of the dithiophenolpyridazinedione since these groups would be extruded post-bioconjugation with a reduced disulde.
With dithioaryl(TCEP)pyridazinedione 5 in hand, we appraised its suitability as a bioconjugation reagent for both disulde reduction and functional re-bridging. To do this, Fig. 1 Illustration highlighting previous strategies towards disulfide stapling and the novel strategy described in this manuscript. pyridazinedione 5 was incubated with a selection of biologically relevant disulde containing peptides and proteins, i.e. somatostatin, octreotide and a Fab (fragment antigen-binding) arm of Herceptin™. To our delight, in each and every case, pyridazinedione 5 was shown to reduce and functionally rebridge the singly accessible disulde (see Scheme 2 and ESI for further details †). Moreover, only a small excess of "2-in-1" reagent 5, 1.25 equivalents, was required to effect complete conversion. Another favourable property of pyridazinedione 5 is that it is a solid which can be stored with complete stability over a protracted period at À18 C under argon.
At this stage we rationalised that a molecule with both reducing and re-bridging functions would minimise the residency time of the cysteines liberated from disulde reduction. This is especially in view of no reduced protein being observed upon incubation of the above peptides and proteins with reagent 5 (see ESI for details †). To appraise this further, we incubated the Fab fragment of Herceptin™ with 1, 2 and 5 equivalents of dipropyl-pyridazinedione 6 prior to incubation with 2 equivalents of diethylpyridazinedione 5 (Scheme 3). Validating our hypothesis, re-bridging was only observed with the TCEP-bearing diethylpyridazinedione. The control reaction of reducing the Fab prior to adding a mixture of pyridazinedione 5 and 6 afforded a mixture of bioconjugates (see ESI for details †).
With the knowledge of the use of dithioaryl(TCEP)pyridazinedione 5 resulting in a high local concentration of the specic pyridazinedione incorporated into the "2-in-1" scaffold, we appraised the use of the reagent in the context of a multi-disulde system. To do this, we chose to use Herceptin™an antibody comprising four disulde bondswhose disulde bonds can be scrambled on attempted functional disulde rebridging. 9h Although this scrambling can be minimised, the leading strategy is reagent specic. 9j We rationalised that the use of pyridazinedione 5 would ensure minimisation of  disulde scrambling in a general sense. To this end, pyridazinediones 3 (with no internal reducing agent function) and 5 (with reducing agent function) were reacted with Herceptin™ under the appropriate reaction conditions, i.e. reaction with pyridazinedione 3 required reduction of Herceptin™ with TCEP. Gratifyingly, no disulde scrambling was observed by SDS-PAGE for reagent 5, with complete re-bridging of all disuldes conrmed by UV-Vis (Fig. 2, lane 4). Analogous reagent 3, with no inherent reducing capability, afforded a disulde scrambled product (Fig. 2, lane 2). Even when dithiopyridazinedione 3 was used in excess and TCEP was added in small portions multiple times (6 Â 0.33 eq per disulde every 30 min), i.e. to minimise the number of open disuldes at any given time, disulde scrambling was still observed (Fig. 2, lane  3). Although scrambling was far less pronounced, the reaction protocol is highly cumbersome and inefficient. Reaction of dithiopyridazinedione 3 at 37 C also afforded a mixture of products (Fig. 2, lane 5). Furthermore, reaction of pyridazinedione 2 at 4 C and 37 C also afforded a mixture of correctly and incorrectly re-bridged modied Herceptin™ conjugates (see ESI for details †).
Providing a reagent that can functionally re-bridge the disulde bonds of Herceptin™ without disulde scrambling is a major contribution in view of the desire to create homogenous conjugates in the eld of antibody-drug conjugates. 14 Use of Herceptin™, in view of its clinical validation alone and as the antibody component of FDA-approved ADC Kadcyla™, 15 provides direct applicability of the chemistry to this exciting area of targeted therapy.
To make this approach modular and expand its scope, we synthesised an analogue of pyridazinedione 5, which contained an alkyne handle for "click" functionalisation, in alkyne-pyridazinedione 7 (see Scheme 4). An analogous route to that described in Scheme 1 was followed (see ESI for further details †). The appraisal of this molecule was carried on the Fab fragment of Herceptin™ as this would allow extensive analysis by UV-Vis, MS, SDS-PAGE and ELISA (for binding). The optimised conditions for the insertion of dithioaryl(TCEP)pyridazinedione 5, was applied to the alkyne bearing analogue for the functional re-bridging of the Fab fragment of Herceptin™ to form conjugate 8. The efficiency of reduction and re-bridging was translated cleanly from reagent 5 to alkyne analogue 7 by MS, SDS-PAGE and UV-Vis (see Scheme 4 and ESI for further details †).
Finally, conjugate 8 was functionalised by "click" modication using doxorubicin, AlexaFluor488™ and sulfo-cyanine5 azides (see Scheme 4). In all cases, complete conversion was observed to afford functionalised conjugates 9a-c, thus demonstrating how the platform may be used for efficient introduction of functional modalities as well as reduction and re-bridging. Moreover, the binding of the Fab protein was not compromised, by ELISA analysis (see ESI for details †), through the chemistry applied. Pyridazinedione 7 was also shown to successfully re-bridge the full antibody system of Herceptin™ and be amenable to "click" functionalisation with doxorubicin azide (see ESI for details †).

Conclusions
In conclusion, we have provided an important step towards delivering on next-generation disulde stapling. This rst-in class technology allows for reduction and functional re-bridging by the use of a single reagent. Moreover, this strategy has been shown to result in a high local concentration of bridging agent, which has been exploited for the functional re-bridging of a multi-disulde system (i.e. Herceptin™) without disulde scrambling. Finally, facile "click" functionalisation and retention of binding affinity, using our strategy, has been demonstrated on a Fab of Herceptin™.