Unlocking regioselective meta-alkylation with epoxides and oxetanes via dynamic kinetic catalyst control

Regioselective arene C−H bond alkylation is a powerful tool in synthetic chemistry, yet subject to many challenges. Herein, we report the meta-C−H bond alkylation of aromatics bearing N-directing groups using (hetero)aromatic epoxides as alkylating agents. This method results in complete regioselectivity on both the arene as well as the epoxide coupling partners, cleaving exclusively the benzylic C−O bond. Oxetanes, which are normally unreactive, also participate as alkylating reagents under the reaction conditions. Our mechanistic studies reveal an unexpected reversible epoxide ring opening process undergoing catalyst-controlled regioselection, as key for the observed high regioselectivities.

and protonation steps are required, and there is only 30% of carboxylic acid in the catalytic system.Would an additional base (carbonate, carboxylate) bring a positive effect?This has apparently not been investigated in the optimization process (Table 1).Another point is the fate of RuCl2(PPh3)3 in the presence of carboxylic acid at 80 °C.It is known that RuCl2(PPh3)3 is a good catalyst in radical processes such as atom transfer radical addition (Kharasch reaction) or polymerization (Macromolecules 1996, 29, 6979).It is also known that RuCl(O2CR)(PPh3)3 and Ru(O2CR)2(PPh3)2 can be formed and have catalytic properties due to the presence of their internal base (Nat.Commun 2017, 6, 8140; Dalton Trans 2019, 48, 4625).What are the actual catalytic species in this catalytic reaction?
The work presented in this manuscript is very innovative and could be published with the addition of some comments answering some of the above questions.
Reviewer #3 (Remarks to the Author): In this manuscript, Larrosa et al. described a meta-C−H bond alkylation of aromatics bearing Ndirecting groups using (hetero)aromatic epoxides as alkylating agents.It is smart to use NaI as the nucleophilic catalyst, which could be engaged in nucleophilic ring opening of epoxides to yield secondary benzylic iodides.Note that the meta-C−H functionalization with various radical precursors such as alkyl halides by Ru-catalysis is well developed by Ackermann and others.In this method, the in situ generated secondary benzylic iodides are acted as actual radical precursors for further meta-C−H alkylation.Therefore, it suffers from the lack of novelty considering the radical precursors.This manuscript would be a very strong publication in a more specified journal.

Response to reviewers
We are grateful to all reviewers for their time in reviewing our work and their useful suggestions to improve the manuscript.Point-by-point responses follow:

Reviewer #1
This paper describes the meta-selective alkylation of aromatic rings bearing directing groups using a Ru catalyst and epoxides.Similar meta-alkylation reactions using alkyl halides have been reported by Ackerman and Frost.Mechanistic studies have revealed that the reversible ring-opening of epoxides by NaI generates alkyl iodides, which act as alkylating agents.Therefore, from a mechanistic perspective, it can be considered an extension of the known reactions.However, from a synthetic organic chemistry standpoint, it is noteworthy that epoxides are used as alkylating agents, controlling the meta-selectivity in C-H functionalization and the regioselectivity in the ring-opening reaction.This is certainly an important catalytic reaction in organic synthesis and could be suitable for publication in Nature Communications.
Question 1: One point to consider is whether the proposed reaction mechanism is also applicable to the reaction with oxetanes, and it would be beneficial to provide further verification in this regard.
Response: To explore the reaction mechanism when using oxetanes we have carried out the mechanistic studies detailed in the scheme below.
In eq A, the reaction between 2-phenylpyridine 1a and iodohydrin 10 occurs smoothly to yield 5a in 46% yield, indicating that iodohydrin 10 may serve as the key intermediate for meta-alkylation of 1a and 2-phenyloxetane 4.This result parallels those obtained with the epoxide.However, while using the iodohydrin 11 under identical conditions, no detectable amount of 5a is obtained (eq B).
Furthermore, iodohydrin 10 and 11 do not interconvert under the reaction conditions (eq C and eq D).Moreover, the ring opening study of 4a in eq E demonstrates that iodide promoted oxetane opening can only occur at more hindered benzylic C−O bond.These results are in stark contrast to those obtained with epoxides where a dynamic equilibrium was observed.Finally, the reaction does not occur in the absence of NaI (eq F).Taken together, these results imply that iodohydrin 10 is generated as the single regioisomer in the reaction when 2-phenyloxetane reacts with iodide, and then functions as the intermediate in the meta-alkylation reaction.

ACTION:
We have added the above data and a discussion into the SI (pages S70-S74, mechanistic studies section).We have added the following sentence in the mechanistic section of the manuscript: "On the other hand, analogous mechanistic studies on the use of oxetanes as coupling partners suggest in that case a direct and non-reversible iodide-mediated regioselective opening of the oxetane may be responsible for the observed meta-alkylation (see SI for more details)."

Reviewer #2
The authors report on meta-alkylation of phenylpyridine derivatives with epoxides and oxetanes under ruthenium catalysis conditions.The mechanism which is proposed involves radical processes and implies a dual role of the ruthenium catalyst as redox active species and for directed ortho-CH activation/cyclometalation.This type of mechanism has already been reported to explained the regioselective meta-functionalization of phenylpyridine derivatives, which are classical model substrates for this type of catalysis (see ref. 24 for instance).The novelty lies in the use of epoxides as alkylating reagent, and more interestingly the regioselectivity of the created C-C bond, which is reverse to classical epoxide opening.The mechanism has been studied in details leading to the conclusion that a catalyst-controlled dynamic ring opening/ring closing of the epoxide leading to the production of the reactive radical precursor was operating.
Question 1: Concerning the scope of the reaction, it is noticeable that no example of phenylpyridine featuring a meta-substituted phenyl ring is reported (only ortho and para).In the same direction, no bis-meta-alkylated products are formed.Could the authors comment on this?Is is steric?

Response:
We have tested four meta-substituted phenylpyridines with two of them giving product, albeit in lower yield than orthoand para-substituted substrates: meta-Substituted phenylpyridines tend to be poor substrates in these reactions due to cyclometalation occurring para to the substitution, due to steric hindrance.This selectivity then hinders the alkylation, as alkylation is thought to occur para to the ruthenium.Thus limiting the reactivity and yield for these substrates.This same reasoning also explains the lack of bis-alkylation observed in these reactions.Both of the substrates that gave low yield contained a fluorine in the meta-position, which does not have a large steric impact and can help direct the cyclometalation next to it, leading to the product.

ACTION:
We have added the results of 3k and 3l to Fig 2A in the manuscript, and pages S26-S27 in the SI for compounds data, and pages S104-S107 in the SI for copies of NMR spectrum.We have included the following comment in the manuscript: "In accordance with previous Ru-catalyzed meta-alkylation, 40-53 meta-subsitution on the arene is not well tolerated as it forces the cyclometalation to occur on the distal ortho-position and then blocks reactivity.This reasoning is also why bis-alkylation does not occur in these reactions." The extension to oxetanes with the same regioselectivities is a good achievement, which represent another innovation of this research.

Question 2: Concerning the mechanism, it is noted that the role of the generated base (carboxylate) is required for the C-H activation. It should be useful to mention the nature of this base in the text (concomitant in situ formation of a base that is required …). By the way in the catalytic cycle, both
deprotonation and protonation steps are required, and there is only 30% of carboxylic acid in the catalytic system.Would an additional base (carbonate, carboxylate) bring a positive effect?This has apparently not been investigated in the optimization process (Table 1).
Response: This was an intriguing suggestion.We have tested the effect of different additional carboxylate and carbonate in the reaction (Scheme R1).Specifically, we added 15% of carboxylate and simultaneously increased the loading of 2-ethylbutanoic acid (A4) from 30% to 60% in the reaction (entries 1-5, Table R1), in order to in situ generate the carboxylate of A4 (30%) and maintain the same concentration of A4 (30%) present in the standard conditions.While the addition of K2CO3, Cs2CO3, Li2CO3 and MgCO3 have a minimal effect on the reaction outcome, Na2CO3 slightly inhibits the reaction.Moreover, the addition of 30% of carbonates (A4 was added in 30%, entries 6-10, Table R1) also show a negligible effect on the yield of 3a.

Table R1. The effect of additional base
These results demonstrate that 30% of carboxylic acid additive itself is well-suited for the reaction, likely functioning as a proton shuttle in the reaction: it protonates the alkoxide that is generated from the ring opening of epoxides and oxetanes, and the resulting carboxylate then can act as a base to facilitate ortho-C−H ruthenation.In this paradigm, only a catalytic amount of carboxylic acid is needed in the reaction.The results observed are consistent with the small KIE and the H/D scrambling observed in the reaction, both of which suggest that the C-H activation is not a rate determining step.

ACTION:
We have added this table of results into the SI (page S17, optimization section).We have added following comment in the revised manuscript to explain the possible roles of carboxylic acid additive: "The carboxylic acid additive may function as a proton shuttle in the reaction by protonating the alkoxide generated after the epoxide ring opening, with the resulting carboxylate acting as a base to facilitate ortho-C−H ruthenation."Question 3: Another point is the fate of RuCl2(PPh3)3 in the presence of carboxylic acid at 80 °C.It is known that RuCl2(PPh3)3 is a good catalyst in radical processes such as atom transfer radical addition (Kharasch reaction) or polymerization (Macromolecules 1996, 29, 6979).It is also known that RuCl(O2CR)(PPh3)3 and Ru(O2CR)2(PPh3)2 can be formed and have catalytic properties due to the presence of their internal base (Nat.Commun 2017, 6, 8140; Dalton Trans 2019, 48,   4625).What are the actual catalytic species in this catalytic reaction?
Response: This is an interesting point as Ru(O2CR)2(PPh3)2 is definitely a possible intermediate in the reaction.After heating RuCl2(PPh3)3 with the carboxylic acid in dioxane, only PPh3 and OPPh3 were observed by 31 P NMR analysis.However, upon addition of K2CO3, Ru(O2CR)2(PPh3)2 was also observed by 31 P NMR.
The standard reaction was stopped after 22 h, diluted in dioxane, filtered and submitted for 31 P NMR and mass spectral analysis.Ru(O2CR)2(PPh3)2 was not observed by either analysis.
Instead, the 31 P NMR spectrum shows two peaks (23.0 and 32.4 ppm) which are consistent with literature values for mono-cyclometalated phenylpyridine-ruthenium species, with one and two triphenylphosphines bound to the ruthenium respectively. 1The slight difference in the chemical shift can be explained by change of solvent from d3-acetonitrile to 1,4-dioxane, which can both coordinate to the ruthenium.There were an extra 2 peaks that could not be assigned at around 42 ppm, but these do not correspond to either Ru(O2CR)2(PPh3)2 or RuCl(O2CR)(PPh3)3. 2,3 SI-MS of this reaction mixture shows the masses of monocyclometaled-phenylpyridines (both substrate and product) plus one and two triphenylphosphines.This spectrum is consistent with the tentative assignment of the above 31 P NMR.
Formation of Ru(O2CR)2(PPh3)2 was attempted, but unfortunately purification of this complex proved difficult and was only able to be obtained as a mixture with PPh3.This mixture was shown to be a competent precatalysis for the reaction, with and to a lesser extent without the acid additive.