Computationally-Guided Development of a Chelated NHC-P Iridium(I) Complex for the Directed Hydrogen Isotope Exchange of Aryl Sulfones

Abstract

Herein, we report the rational, computationally-guided design of an iridium(I) catalyst system capable of enabling directed hydrogen isotope exchange (HIE) with the challenging sulfone directing group. Substrate binding energy was used as a parameter to guide rational ligand design via an in silico catalyst screen, resulting in a lead series of chelated iridium(I) NHC-phosphine complexes. Subsequent preparative studies show that the optimal catalyst system displays high levels of activity in HIE, and we demonstrate the labeling of a broad scope of substituted aryl sulfones. We also show that the activity of the catalyst is maintained at low pressures of deuterium gas and apply these conditions to tritium radiolabeling, including the expedient synthesis of a tritium-labeled drug molecule.

The incorporation of heavy isotopes into potential drug molecules has, over time, become an indispensable tool within the pharmaceutical industry.¹ One of the most utilized methods in this area is directed hydrogen isotope exchange (HIE, Scheme 1), wherein hydrogen atoms ortho to a Lewis basic directing group (DG) are replaced with deuterium or tritium.²

Scheme 1. General Reaction Scheme for Directed HIE.

Homogeneous iridium catalysts have been proven to be highly active in ortho-directed HIE.³ These species have enabled the use of a broad range of directing groups in the labeling of small molecules and potential drug candidates.⁴ Related to this, studies within our own laboratory have led to the development of a series of catalytically active iridium(I) NHC/phosphine complexes 1–2 (Figure 1), which deliver heavy isotopes of hydrogen (deuterium, D and tritium, T) to aryl and alkenyl substrates via a directed C–H activation process with a broad range of directing groups.^3d,5 Additionally, iridium(I) NHC chloride complexes of type 3 have shown utility in the labeling of primary aryl sulfonamides and aryl aldehydes.⁶

Figure 1.

Ir(I) precatalysts for directed HIE.

Despite these advances, however, certain high-value functionalities, common throughout pharmaceutical motifs and natural products, still present significant challenges for directed HIE. One such example is the aryl sulfone unit, which is prevalent throughout drug discovery.⁷ For example, sulfones are present in antibiotics, such as dapsone and dextrosulphenidol,^8a,8b and are also components within a range of other medicines, including the non-steroidal anti-inflammatory, rofecoxib, and the retinoid, sumarotene (Figure 2).^8c,8d However, the capacity to exploit the sulfone as a directing group in C–H activation processes,⁹ including HIE,¹⁰ is currently vastly undermet. With specific regard to HIE, Pfaltz and Muri have reported an N,P-chelated Ir(I) catalyst, which mediates deuterium labeling of a series of substrates, including one example of a simple aryl (phenyl methyl) sulfone. This catalyst system does, however, require activation with elevated pressures (2.2 bar) of deuterium.^10a Nonetheless, the abundance of the sulfone moiety in pharmaceutically active agents makes the ability to exploit these functional units as handles for ortho-directed HIE a particularly attractive goal. Additionally, applying a mechanistic approach to such a challenge has the potential to deliver an enhanced, and potentially predictive, understanding of such directed functionalization endeavors, enabling the use of sulfones in a broader range of C–H activation processes.

Figure 2.

Examples of sulfone-containing drug molecules.

Current iridium-catalyzed HIE processes directed by sulfones have a range of limitations. For example, previous studies within our laboratory toward sulfone-directed HIE have employed monodentate iridium(I) catalysts, specifically NHC/phosphine 1a and NHC/Cl analogue 3a. When these catalyst systems were applied to the labeling of phenyl methyl sulfone 4a, low levels of incorporation were observed (Figure 3a).^6a We hypothesized that, in this case, substrate binding was unusually the turnover-limiting step, in contrast to less hindered directing groups where C–H activation is turnover limiting.^5c

Figure 3.

Sulfone labeling: Challenges with monodentate catalysts and a potential chelated catalyst solution.

Bearing in mind that catalysts 1–3 possess very large NHC and (for 1 and 2) phosphine ligands, coordinated in a trans relationship (Figure 3b), the tetrahedral nature of the sulfone directing group results in significant steric repulsion between the substrate and ligand when compared to more planar directing groups, such as acetyl. This may inhibit substrate binding, which in turn would severely limit the isotope incorporation in sulfone-derived substrates, as observed with 1a and 3a.^6a We hypothesized that the use of a tethered N-heterocyclic carbene-phosphine ligand (NHC-P, Figure 3b) would result in a less hindered catalyst environment, more able to accommodate the sulfone unit.

The paradigm of rational ligand design is emerging as an appreciably powerful tool through which the knowledge and understanding gained from mechanistic insights allows an effective catalyst to be accessed rapidly.^11,12 We selected this approach to address the challenge of developing a broadly effective chelated catalyst system with the ability to label sulfone-bearing substrates to high levels of isotope incorporation under mild reaction conditions. Specifically, a system derived from rational, computational design would not only provide a solution to the challenge of sulfone labeling but would also potentially facilitate the application of the designed system to a more diverse array of substrate classes.

To initiate our studies, and guided by our postulate that substrate binding was limiting in the sulfone case (Figure 3), computational modeling was used to calculate the binding energy (E_Bind) of a model substrate methyl phenyl sulfone 4a to catalytically relevant^5c iridium(III) hydride complexes of varying designs (Figure 4). The binding energy was calculated using the counterpoise method, as described by Boys and Bernardi.¹³ We observed that while methyl phenyl sulfone 4a could coordinate to the monodentate NHC/phosphine iridium(III) hydride derived from 1a (Figure 4a), a modest binding energy of only −15.3 kcal mol^–1 was calculated, with a similarly poor binding energy of −15.9 kcal mol^–1 for precatalyst 3a (Figure 4b). To place these binding energies in context, when the same method was applied to the binding of acetophenone 5 (which has been shown to label to high levels using precatalyst 1a(5c)), a significantly more negative and therefore more favorable binding energy of −23.1 kcal mol^–1 was found (Figure 4c).

Figure 4.

Calculation of binding energy with monodentate ligand systems.

In terms of selecting a chelating ligand system to overcome these binding issues, a number of tethered NHC-P ligand motifs have been reported.¹⁴ Due to the range of tethers and combinations of NHC and phosphine substituents already established, an in silico screening was carried out to determine which characteristics of the chelating ligand would lead to an increased sulfone binding energy, and thus a potentially effective catalyst system. Accordingly, we proposed a virtual library of eighteen structurally diverse NHC-P ligands, covering a number of chelate sizes (from four to seven membered rings), and a range of steric and electronic parameters, as dictated by the substituents on the NHC and phosphine moieties. In each case, the binding energy of methyl phenyl sulfone to the relevant chelated iridium(III) hydride was calculated (Scheme 2).

Scheme 2. In Silico Screen of Bidentate Ligands and the Calculated Binding Energy for Methyl Phenyl Sulfone.

Firstly, it was noted that only those ligands with N-aryl substituents gave significantly higher sulfone binding energies than the monodentate systems (cf. 6 vs 7). In general, ethylene tethered systems (6, 9–13) showed a favorable range of binding energies, from −23.7 kcal mol^–1 for 13 to −29.5 kcal mol^–1 for 10. Also, aryl and alkyl phosphines both gave enhanced binding energies when compared to the aforementioned monodentate catalysts. Furthermore, substitution on the backbone of the NHC appeared to deliver a moderately more favorable binding energy in some cases (cf. 12 vs 11). The larger, benzyl-tethered ligands, 14–18, also showed favorable sulfone binding energies with a similar pattern to the corresponding ethylene-bridged systems being observed in terms of the NHC substituent. Dicyclohexyl phosphinyl variant 16, however, displayed a reduced binding energy with respect to the diphenylphosphinyl analogue 15 and, as such, no further dialkylphosphines were considered. The effect of the chelate ring size and rigidity was next examined, with the ortho-phenylene, 19, methylene bridged, 20, and directly N-P bound ligand, 21, all displaying favorable binding energies. Interestingly, phosphite, 22, returned the highest binding energy of all ligand variations examined. Finally, the benzimidazole-derived NHC, 23, was also found to be favorable, resulting in a binding energy of −30.8 kcal mol^–1. The calculated binding energies for each of the 18 bidentate complexes, as well as monodentate complexes 1a and 3a, are shown graphically in Figure 5.

Figure 5.

Calculated binding energies with sulfone 4a.

From this in silico screen, a total of six ligands (11, 13, 15, 17, 18, and 23) were selected for progression based on their binding energy of methyl phenyl sulfone 4a, coupled with their synthetic tractability in each case. The structures of the targeted iridium(I) precatalysts 24–29 are shown in Figure 6, with the selected ligands predicted to give complexes with a range of substrate binding energies from −23.7 to −30.8 kcal mol^–1. The ligands for the iridium(I) precatalysts shown were then synthesized via modification of existing literature procedures.^14d−14g,15

Figure 6.

Summary of the targeted Iridium(I) precatalysts.

With the iridium(I) precatalysts 24–29 in hand, their labeling of methyl phenyl sulfone 4a was investigated. This initial catalyst screen was performed at 25 °C in dichloromethane with a standard catalyst loading of 5 mol % and a reaction time of 16 h. In the majority of cases (i.e., with the exception of 25 and 27), a marked increase in the isotope incorporation was observed compared to the monodentate systems 1a and 3a (Table 1).

Table 1. Catalyst Screen^a.

entry	catalyst	% D incorporationa
1	1a	96a
2	3a	176a
3	24	35
4	25	12
5	26	23
6	27	13
7	28	36
8	29	43

^aReactions performed in triplicate.

Based on the labeling results shown in Table 1, the most promising catalyst, 29, was selected for further optimization, beginning with a solvent screen and exploiting the versatility of the catalyst’s BAr_F counterion in this regard.^5d As shown in Figure 7, the observed incorporation reached significantly higher levels in a broad range of solvents compared to the initial labeling result in DCM. In particular, chlorobenzene and fluorobenzene delivered the highest levels of labeling, affording excellent 93 and 91% incorporations, respectively. Additionally, the use of non-halogenated solvents such as di-iso-propyl ether, MTBE, diethyl ether, and toluene also provided high levels (>70%) of deuterium incorporation.

Figure 7.

Solvent applicability in sulfone labeling.¹⁵

Accordingly, chlorobenzene was chosen as the optimal solvent for this sulfone labeling process with catalyst 29. Not only did this medium allow for the best incorporation from the eleven solvents examined but also additional technical advantages were delivered, in that the higher boiling nature of chlorobenzene meant that cooling to −78 °C could be avoided when exchanging the atmosphere within the reaction manifold from nitrogen to deuterium, thus simplifying the practical procedure. Notably, maintaining catalyst loadings at 5 mol %, the reaction time could be reduced to 1 h, with no drop in the observed isotope incorporation.¹⁵ These conditions were then applied to a broad range of aryl sulfones to establish the scope of our developed process (Scheme 3). A total of eighteen additional substrates were examined, encompassing a diverse range of sulfones with generally excellent deuterium incorporation effectiveness observed throughout the series. Both electron-donating and electron-withdrawing substituents in the ortho (4b–4d) and meta (4e–4 h) positions of the aryl ring were well tolerated, leading to high levels of incorporation. Notably, meta-trifluoromethyl substrate 4 g exhibits almost no incorporation at the considerably hindered position between both aryl substituents, but displays excellent levels of deuterium labeling at the less hindered position ortho to the sulfone. With less sterically encumbered meta substituents (4f and 4 h) both positions ortho to the sulfone are labeled to a high degree. A range of electronically distinct para-substituents are also well tolerated (4i–4l, 4n). In the case of para-nitro substrate 4 m, this alternative directing group is shown to mildly outcompete the sulfone, but does not prevent an acceptable level of isotope incorporation through sulfone-directed HIE. Furthermore, restricting the orientation of the sulfone, as in cyclic substrate 4o, did not result in a decrease in the excellent levels of incorporation generally observed. We next turned our attention to the effects of increasing the steric bulk around the sulfone group, with substrates 4p–4r. While a slight decrease in the levels of incorporated deuterium were observed in diphenyl sulfone 4p and iso-propyl phenyl sulfone 4q (57 and 66%, respectively), an excellent incorporation of 80% was observed with the bulky tert-butyl phenyl sulfone 4r. We also investigated labeling of benzyl methyl sulfone 4s, where the sulfone would direct labeling via a 6-membered metallacyclic intermediate (6-mmi), which is considerably less favored than the more common 5-mmi.^5c Nonetheless, moderate levels of incorporation were still observed in this more challenging substrate under these mild standard conditions with a low catalyst loading of 5 mol %.

Scheme 3. Scope of Aryl Sulfone Labeling.

Incorporation averaged across both positions.

To further expand the utility of our newly developed catalyst system, we investigated the effects of employing a reduced pressure of deuterium gas using a TRITEC manifold system, with these conditions more closely emulating those deployed for radiolabeling with tritium gas within a pharmaceutical industry setting. Following only minimal optimization of our standard conditions,¹⁵ we obtained high levels of labeling under these low-pressure conditions (Scheme 4). Employing phenyl methyl sulfone 4a, it was found that with a deuterium pressure of only ∼400 mbar, a mildly elevated catalyst loading of 7.5 mol % allowed for similar levels of deuterium incorporation to the standard (non-TRITEC) laboratory set-up. Notably, when employing the most active catalyst reported by Pfaltz and Muri,^10a very low levels of labeling of this same substrate, 4a, were observed when employing 1 atm of D₂ for catalyst activation,¹⁵ as is routinely employed with our suite of Ir(I) catalysts, including 29. Indeed, the requirement for supra-atmospheric catalyst activation with this previously reported system may present practical challenges when applying the methodology to low-pressure tritiations (vide infra).

Scheme 4. Reduced Pressure Deuterium and Tritium Labeling of Methyl Phenyl Sulfone 4a.

With a successful reduced atmosphere protocol in hand using catalyst 29, the conditions were then applied to the tritium labeling of the same sulfone 4a. Exposure of 4a to 7.5 mol % of precatalyst 29 under 405 mbar of tritium gas afforded t-4a with a high activity of 51.3 Ci/mmol, corresponding to a tritium incorporation of 88% across both ortho positions. Additionally, the major mass ion of [M + 4] confirmed that the sample had indeed been labeled with two units of tritium (Scheme 4).

Finally, to further demonstrate the power of our developed catalyst as applied to radiolabeling, as shown in Scheme 5 we targeted a tritium-labeled sample of the highly potent GPR119 agonist 30.¹⁶ Accordingly, benzylic bromide-containing sulfone 4t could be readily tritiated and alkylated in one pot to afford t-30 with excellent levels of radiolabel incorporation.

Scheme 5. Tritium Labeling of GRP119 Agonist 30.

In conclusion, we have employed a computationally-guided, rational ligand design approach to target a series of iridium(I) NHC-P complexes for the directed HIE of aryl sulfones. The resulting optimal complex proved to be highly active in a range of solvents, and an extensive substrate scope has been established, with high levels of deuterium incorporation being exhibited across the series. Furthermore, the catalyst system has been shown to retain its activity when applied to the low-pressure labeling systems currently used extensively in the pharmaceutical industry. Finally, the emerging iridium(I) NHC-P catalyst has been applied to tritium labeling, furnishing a selectively tritiated sample of GPR119 agonist t-30 with high levels of specific activity. Our current studies are focused on extending our understanding of these catalytic systems in order to further refine our design process to encompass even more challenging substrates.

Glossary

Abbreviations

BAr_F

tetrakis(3,5-bis(trifluoromethyl)phenyl)borate

Bn

benzyl

CPME

cyclopentyl methyl ether

DCE

1,2-dichloroethane

DCM

dichloromethane

DiPP

2,6-di-iso-propylphenyl

Mes

2,4,6-trimethylphenyl

MTBE

methyl tert-butyl ether

NHC

N-heterocyclic carbene

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.0c03031.

• Details of experimental procedures and computational methods (PDF)

The authors declare no competing financial interest.