A Thermally Stable SO <sub>2</sub> -Releasing Mechanophore: Facile Activation, Single-Event Spectroscopy, and Molecular Dynamic Simulations

INTRODUCTION

Sulfur dioxide (SO2), known historically as an environmental pollutant and a food preservative, 1,2 has recently gained recognition as an endogenously generated gaseous signaling molecule (GSM) responsible for transmitting chemical signals in organisms, tissues, and cells. 3,4 There is mounting evidence that suggests SO2 may be a fourth gasotransmitter akin to NO, CO, and H2S, but this claim is still being carefully examined and validated. 5 SO2 has also been found to be a potent therapeutic agent with various pathophysiological effects, including its abilities in reactive oxygen species (ROS) upregulation and glutathione (GSH) consumption. [5][6][7][8] Gaseous SO2 and mixed sulfite salts are traditionally used as the SO2 donor in biological assessment, leading to concerns in both dosage control and release kinetics. The development of reliable SO2 donors is therefore highly desirable. 6 Whereas a few controllable SO2 donors have been reported under external stimuli such as light, [9][10][11] heat, 12 chemical reagent, [13][14][15][16] and enzymes, 17 the development of SO2 donors using alternative chemistries and triggering mechanisms will enable new possibilities in future biological investigations. 6,18,19 Polymer mechanochemistry has recently attracted great attention as an enabling tool to release cargo molecules in a controlled manner for potential biomedical applications. [20][21][22][23][24][25][26] Ultrasound is typically employed as an external trigger to generate solvodynamic shear force from the collapsing microbubbles. 27 The resulting mechanical energy efficiently activates stress sensitive motifs embedded in polymer backbones, known as mechanophores, to release desired payloads. This approach, especially when combined with recent advances in using biomedical ultrasound, 28,29 offers promise as a new way to drive chemical reactions remotely, as ultrasound has the unique ability to penetrate deep within biological tissues to achieve mechanochemical transformations noninvasively with precise spatial and temporal control. [30][31][32] While the judicious design of mechanophores has enabled the liberation of various molecules upon activation, the scope of cargos that have biomedically relevant functions is still very limited, [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47] especially when high payload is needed. 20 The development of non-scissile mechanophores that release many drug-like molecules (e.g., GSMs) per polymer chain is still underexplored. 48 Our group recently developed a non-scissile norborn-2-en-7one (NEO) mechanophore that releases carbon monoxide (CO) upon the application of mechanical stimuli. 49 The CO gas was designed to be extruded from an unsaturated [2.2.1] bicyclic ketone motif after the generation of mechano-diradicals from the homolytic activation of C-C bonds (Scheme 1a). We hypothesized that if a sulfone moiety was substituted for the bridging CO in NEO, it would act as an alternative leaving group that would be liberated following a similar pathway, resulting in a new non-scissile SO2 releasing mechanophore. Unfortunately, preliminary trials following this strategy were unsuccessful (Scheme S1), presumably due to the welldocumented poor thermal stability of unsaturated [2.2.1] bicyclic sulfones. The spontaneous SO2 cheletropic extrusion is known to proceed near room temperature when either one or two double bonds are present in the bicyclic ring, generating diene or benzene derivatives respectively (Scheme 1b). 13,50 To overcome the challenge of thermal decomposition in such scaffolds, we herein report the development and comprehensive investigation of a highly thermally stable but mechanically labile SO2 releasing mechanophore based on an 8thiabicyclo[3.2.1]octane 8, ,8-dioxide (TBO) scaffold (Scheme 1c). The mechanically triggered release of SO2 from TBO was demonstrated in both organic solvent (i.e., THF) and aqueous media using pulsed ultrasonication. We quantified the reactivity of TBO under tensile force using single-molecule force spectroscopy (SMFS) and resolved its single-event activation process in force-clamp experiments. The mechanism and stereoselectivity of TBO opening were further investigated using ab initio steered molecular dynamics (AISMD) and results are discussed below.

Synthesis of TBO and Main-Chain TBO Containing

Polymers. We set out to synthesize the TBO mechanophore from the cheletropic addition between 1,3-cycloheptadiene and excess liquid SO2 utilizing a pressure tube (Scheme 2). 51 A clean transformation with excellent yield (93%) was observed after 7 days to afford 1 in multigram scale, whose structure was confirmed by X-ray crystallography. Intriguingly, trials on 1,3cyclooctadiene under the same condition led to no conversion to bicyclic sulfones. Next, an osmium-catalyzed dihydroxylation reaction using citric acid as an additive provided 2 in 62% yield. 52 The exo product was observed as the only diastereomer, consistent with the general diastereoselectivity observed in the dihydroxylation of norbornene like structures. 53 A subsequent two-step sequence of acylation and ring-closing metathesis afforded macrocycle 4 as an E/Z mixture in 81% yield. The stereochemistry of 4 was confirmed to be the exo product by X-ray crystallography (Scheme 2). Notably, 19.5 wt% of the monomer 4 comes from SO2. We copolymerized 4 with Z-cyclooctene using ringopening metathesis polymerization (ROMP) to obtain multimechanophore polymers (MMPs) P1 for solution sonication studies. 54 Following the common strategy pioneered by the Craig lab, another comonomer 1,2-epoxy-5-cyclooctene was used to prepare 1:1 statistical copolymer P2 for SMFS studies. 55 The epoxides in the backbone aim to provide strong adhesion to the atomic force microscope (AFM) tip, ensuring the application of nN-regime forces that are typically required to activate covalent bonds. Scheme 2. The synthesis of TBO embedded polymers for solution sonication and SMFS studies. Ultrasonic Activation of TBO Mechanophore. The mechanochemical reactivity of P1 (Mn= 141.1 kDa, Đ = 1.9, 52.4% TBO, 14.9 wt% from SO2) was evaluated in dilute THF solutions using pulsed solution ultrasonication (1 s on/1 s off, -10 °C, 20 kHz, 10.0 W cm -2 ). During 120 min of sonication, several new resonance peaks emerged and intensified in 1 H NMR of the resulting material SP1, with a concomitant decrease of all corresponding TBO peaks. The characteristic peak at  = 7.08 ppm was assigned to the  proton in vinyl ester moieties, suggesting the desired activation of TBO (Figure 1a). The microstructure of SP1 was carefully determined via 2D NMR analysis and was consistent with the formation of a single symmetrical isomer (Figure S3-S8), as evidenced by the one sharp doublet resonance at  = 7.08 ppm. The 3 JHH coupling constant between two new vinyl protons was calculated to be 15.0 Hz (Figure S5), strongly suggesting the trans configuration (i.e., E olefin). The (E,E) isomer was the stereoretentive product from the opening of TBO ring, and the stereoselectivity was similar to our previously reported NEO mechanophores. 49,56 From peak assignments, it was determined that 63% of the TBO mechanophores were activated after 120 min of sonication based on integrations in 1 H NMR. Low molecular weight P1 (Mn = 5.2 kDa) was synthesized as control, as it is known that low molecular weight polymers experience negligible elongational force in solution sonication. 27 We observed no activation under the same condition in this case, indicating the mechanochemical nature of this transformation (Figure S9). Different from the facile mechanochemical activation, no SO2 extrusion nor other decomposition of TBO was observed from 1 H-NMR at 120 °C even after 24 h (Figure S11). We also conducted thermal gravimetric analysis (TGA) on P1 and did not observe significant weight loss until 300 °C (Figure S12), indicating no thermal SO2 extrusion upon moderate heating. The saturated design of the [3.2.1] structure effectively shuts down the cheletropic extrusion pathway, accounting for the excellent thermostability.

To confirm the release of SO2 from P1, we employed a reported fluorogenic probe 5 that selectively reacts with SO2 derived sulfite (SO3 2-) and bisulfite (HSO3 -). 57 Specifically, the nucleophilic attack of SO3 2-and HSO3 -interrupts the πconjugation of a hemicyanine dye, leading to large changes in both absorption and emission spectrums (Figure 1b). After sonication of P1 (Mn = 79.1 kDa, Đ = 1.8, 48.0 mol % TBO) for 10 min in THF, an aliquot of the solution was added to a 10 μM PBS buffer solution of 5. The deep purple color of 5 PBS solution disappeared after 20 min of incubation and turned into faded pink color, indicating the generation of 5-SO3 2-adduct. In contrast, negligible color change was observed after incubation without ultrasonication (Figure 1b). The fluorescent spectra were also monitored before and after sonication. We observed the emergence of a new peak at 478 nm with the decrease of the original peak at 633 nm only after sonication, which is indicative of the generation of 5-SO3 2-adduct (Figure 1b). On the other hand, no significant changes were detected in either the color of the solution or the emission spectrum after the sonication of P1-control (Mn =5.2 kDa). Infrared (IR) spectroscopy analysis was conducted to further elucidate the extrusion of SO2 in SP1. The significantly diminished intensity of S=O stretching peak at ~ 1305 cm -1 after sonication validated the release of SO2 (Figure S14). 58 Additionally, we noticed two new peaks in the IR of SP1 at 1675 cm -1 and 934 cm -1 . These peaks correspond to the stretching and bending of transdisubstituted olefins respectively, in agreement with the proposed vinyl ester groups generated upon activation. 58 We note that the direct ultrasound triggered SO2 release was previously observed in the depolymerization of alternating sulfone copolymers. 59 However, the co-release of many vinyl acetate monomers and the lack of backbone tunability make those polymers impractical as reliable SO2 donors compared to TBO.

Controlling the local concentration is critical for GSMs delivery, as a number of GSMs including SO2 are known to show critical regulatory functions only at low concentrations, while exhibiting systemic toxicity at high concentrations. 60 To demonstrate the tunable release rate of TBO, various molecular weight P1 samples were synthesized, and the kinetics of activation were investigated. The activation percentage () was monitored by 1 H NMR (Figure 1c). We observed a faster TBO activation with the increase of molecular weight, in agreement with other reported MMPs. 49,61 By tuning the molecular weight and/or the sonication time, TBO activation percentage from 14% to 63% were achieved (Table S1- Triggering the activation in aqueous conditions is required to translate cargo releasing mechanophores to biomedical applications. However, the mechanochemical activation of the cargo releasing mechanophore systems, especially for MMPs, has been mostly investigated in pure organic solvents or their mixtures with water. 26,35,37,43,44,48 In pursuit of both good solubility in water and reasonable incorporations of TBO in the polymer main chain, we synthesized comonomer 6 that contains a ROMP compatible cyclooctene motif with a grafted short polyethylene glycol (PEG) unit (Figure 2a). The resulting high molecular weight copolymers (Mn = 597.8 kDa, Đ = 2.3) with 20.6 mol% of TBO (2.9 wt% from SO2) readily dissolved in water with moderate aggregation (Figure S18). The application of ultrasound to 1.0 mg mL -1 copolymer solution (1 s on/1 s off, ice bath, 20 kHz, 10.0 W/cm 2 ) provided 67% of TBO activation after 60 min, equating the generation of ~ 300 µM of SO2 (Figure S19). We also copolymerized monomer 4 and monomer 6 with our previously developed CO-releasing endoalkyl-NEO to generate statistical terpolymers that concomitantly release CO and SO2 from one polymer chain as an approach for the co-delivery two GSMs (Figure S20). 56 The size of the resulting terpolymer (Mn = 304.6 kDa, Đ =2.4, 15.9 mol% NEO, 13.9 mol% TBO) in water was determined to be around 7 nm on average using dynamic light scattering (DLS) (Figure 2b). From the kinetic study of sonication, we observed slightly greater activation rate of TBO compared to endo-alkyl NEO (Figure 2c). Reasonable  values around 30% were achieved for both mechanophores upon 120 min of sonication, corresponding to around 92 µM of SO2 and 99 µM of CO. We envision these aqueous soluble copolymers as promising materials for biomedical applications in combination with biocompatible ultrasound in future studies. 29 Single-Molecule Force Spectroscopy of TBO. To quantify the mechanochemical reactivity of TBO and compare it with other reported mechanophores, we used SMFS to directly measure the transition force (f*) in a kinetic regime relevant to most potential applications. [62][63][64][65][66][67][68] We prepared P2 with relatively high molecular weights and TBO incorporations (Mn = 73.6 kDa, Đ = 1.8, TBO mol% = 47.2) to sample a large number of mechanophores along one polymer chain.

Representative force vs separation curves obtained from constant-velocity (CV) SMFS experiments are shown in Figure 3a. In CV SMFS, the tip of the atomic force microscope (AFM) and the substrate on which the polymer is deposited are separated at a constant retraction velocity of 300 nm s -1 . As a polymer chain that is attached to both surfaces is stretched, the corresponding force along its backbone gradually increases. Eventually, the force becomes high enough to accelerate the TBO reaction to observable rates. These reactions result in the lengthening of the polymer. Further stretching continues to accelerate the reaction, and as the rate at which the polymer lengthens approaches the rate at which the polymer is being stretched (300 nm s -1 ), a plateau appears in the force-separation curve. The plateau is characterized by a transition force, f*, taken from the inflection point of the transition. The absolute rate of chain extension at a given force is proportional to the total number of mechanophores in the trapped region of the polymer, and so f* in CV SMFS varies with both the length of the trapped chain region and with the loading of mechanophore along the backbone.

In the case of the TBO polymers examined, f* was determined to be 1520 ± 46 pN (Figure 3a). The relatively low f* value of TBO indicates its higher reactivity compared to most cyclobutane and ladderane mechanophores, 63,68,69 and it is comparable to the exo-alkyl-NEO mechanophore reported by us previously (f* = 1526 ± 33 pN). 56 We also performed retracing experiments and observed full hysteresis (i.e., no plateau in the 2 nd stretching) (Figure 3b), implying the irreversible activation of TBO in SMFS.

To further characterize TBO mechanochemistry, we obtained the distance vs time curves of P2 using the complementary technique of constant-force SMFS, also known as force-clamp (FC) spectroscopy. In FC experiments, feedback control of the AFM allows the polymer chains to be held at a constant tension that is maintained by automated adjustment of the tip position to restore force that is lost when a TBO opening releases stored length that relaxes the chain. Here, polymer chains were successfully clamped for up to 18 s at average forces of 1440 pN, just below the plateau force observed in the CV SMFS experiments (Figure 4a and more details in the SI). As the high force is applied to P2 over this time, the TBO mechanophore will open, extend, and release the stored length in the seven membered ring, leading to an increase in the polymer contour length. These events lead to jumps in the distance vs time curves, and the time between events reflects how much time the polymer spends at a given degree of unravelling before another unravelling event occurs.

Using this approach, we were able to resolve the single activation event of TBO mechanophore in the FC experiments at the applied force of 1440 pN, as evidenced by the staircase pattern shown in Figure 4a. 70 The distribution step sizes (L) were characterized across all the pulls and used to determine the average change in length between two steps. Fitting the collected data with a Gaussian distribution yields the average value of 0.68 ± 0.10 nm for P2 (Figure 4b). To validate the nature of the observed transitions, we calculated the contour length of the TBO repeating unit before and after activation under the experimental force (i.e., 1440 pN), using the external force explicitly included (EFEI) method. 71,72 The optimized equilibrium geometries of the nascent TBO and its fully unraveled state were calculated under 1440 pN as shown in Figure 4c. The simulated stored length of a single TBO unit equals 0.71 nm, in good agreement with the experimental results from SMFS. A second small peak is observed in the experimental distribution of step sizes (Figure 4b). The corresponding L value for that second peak is 1.32 ± 0.13 nm (Figure S35), and that value is consistent with the activation of two TBO mechanophores that cannot be resolved individually. Looking forward, the overall gain in length enabled by TBO makes it a potential candidate for reactive strand extension in the service of tougher bulk polymeric networks. 73,74 The bifunctional character of the TBO mechanophore (i.e., 2.13 nm 1.42 nm triggered release of both SO2 and stored length) is appealing, as tougher networks enable greater stress to be loaded into a material and, as a result, higher levels of cargo release to be obtained. We therefore speculate that the dual functions of TBO might be synergistic. The force-coupled rate constants of TBO opening were extracted from both CV and single-event FC experiments (Figure S30). Rate constants were determined in two ways. In the FC experiments, distance vs. time data were fit to a single exponential decay (first order rate law) to obtain a rate constant at the applied clamping force. In the CV SMFS experiments, force-coupled rate constants were obtained by fitting the plateau region to a kinetic model that accounts for and extracts the change in rate as the force varies across the pull. Combining these data gives rate constant k(f) as a function of force. For the data shown in Figure 3a, at the 1520 pN f* typical of the CV SMFS experiments, the rate constant was determined to be 9.0 s -1 , whereas the rate constant at 1440 pN in the FC experiments was found to be 1.8 s -1 . The full range of extracted rate vs. force data across multiple CV and FC SMFS experiments are provided in Figure S30. According to the simplest model of mechanochemical coupling (eq 1), 75 ln(𝑘(𝑓)) = 𝐶 + 𝐹∆𝑥 ‡ 𝑅𝑇 (1) the empirical mechanochemical coupling constant x ‡ was calculated to be 0.82 ± 0.2 Å. The number is comparable to the reported values on NEO mechanophores. Because x ‡ is strongly associated with the structure of the transition state relative to reactant, the agreement between TBO and NEO suggests a similar mechanism for the two mechanophores, 65 previously attributed to a rate-limiting step of the initial C-C bond cleavage under ~ 1.5 nN force. 56 We note that whereas many covalent reactions of mechanophores have been characterized by SMFS, very few reports have demonstrated individually resolved activation events. 70,76,77 Those rare examples have been limited to macrocyclic cyclobutane mechanophores that have a sufficiently large stored length (> 3 nm per event). The resolved single activation event of relatively small TBO domains (~ 0.68 nm) in FC implies the possibility of further generalizing the scope of covalent mechanophores in single-event spectroscopy. The FC spectroscopy has been widely used to study protein unfolding [78][79][80] and the current results suggest that it may hold unique advantages compared to CV pulls for the study of covalent mechanophores.

Mechanistic Study of TBO Mechanophore. We first computationally investigated the activation of TBO using constrained geometries simulate external force (CoGEF) as a widely applied screening tool. 81 We observed SO2 extrusion with a relatively low rupture force (Fmax) of 3.3 nN (Figure S37). CoGEF suggests a one-step transformation that involves simultaneous C-C bond scission and SO2 release using a step increment of 0.1 Å, in contrast to the previous stepwise process observed in the NEO mechanophores. 56 Intriguingly, a stepwise mechanism was observed instead when using a smaller step increment of 0.05 Å near bond cleavages (Figure S38). CoGEF is well-known to omit thermal and dynamic effects, 81 we thus reasoned the mechanochemical activation of TBO merits more systematic computational studies with a higher-level method.

We employed ab initio steered molecular dynamics (AISMD) to investigate the mechanochemical opening of 7 under different forces from 1.5 to 2.5 nN during 500 fs durations at 300 K using uB3LYP-D3BJ/6-31G* (Figure 5a). 71 Only about 10% of the population proceeded to 9 under 1.5 nN within the 500 fs timeframe of dynamic simulation. With the application of 2.0 nN force, we saw successful opening of 7 to 9 in all 30 trajectories. By following the bond elongation over time, we observed the first C-C bond scission event after ~ 225 fs to generate 8 (Figure 5b The diradical lifetime was estimated to be ~ 630 fs under 2.0 nN. At higher forces of 2.25 and 2.50 nN, the stepwise nature of 7 opening remains unchanged despite the accelerated kinetics (Figure S40). Decreased diradical lifetime was observed with the increase of applied force (Table S8).

Figure 6. Simulated stereochemical outcomes of 7 opening under different forces in AISMD simulations. We next analyzed the stereochemical distribution of the formed olefin products 9 from the AISMD results. 82 Under relatively low force of 2.0 nN, all three possible isomers were observed with the distribution of 77% EE, 20% EZ, and 3% ZZ (Figure 6). With the increase of force, the opening of 7 tends to be more stereoselective towards the EE configuration. A sole EE isomer formation was observed computationally under a high force value of 2.5 nN. Therefore, large forces not only effectively accelerate the TBO activation but also drive the reaction into a single product, accounting for the observed facile activation and near quantitative product selectivity in the μs time scale of sonochemical conditions. 83 This increased stereoselectivity towards the stereoretentive product at high forces was also reported in the flyby trajectories of cyclobutane mechanophore ring-opening. 84 The isomeric configuration of mechanophore is known to dramatically impact its mechanochemical reactivity. 56,62,85 We therefore sought to investigate the reactivity of the other two configurations of TBO (i.e., endo and trans) (Table S9). Different from the facile opening of all trajectories of 7 under 2.0 nN in AISMD within 500 fs, only about 10% of the population of endo-TBO were activated under 2.0 nN, indicating the diminished reactivity in the endo configuration. The trans-TBO exhibited an even more limited mechanochemical reactivity that no dissociations were observed at 3.0 nN of force within 500 fs. Experimentally accessing the endo-and trans-TBO will be helpful to further validate the computed reactivity trend, however, the synthesis of these two stereoisomers has been challenging despite the extensive synthetic efforts (Figure S24).