Direct deoxygenative borylation of carboxylic acids

Carboxylic acids are readily available, structurally diverse and shelf-stable; therefore, converting them to the isoelectronic boronic acids, which play pivotal roles in different settings, would be highly enabling. In contrast to the well-recognised decarboxylative borylation, the chemical space of carboxylic-to-boronic acid transformation via deoxygenation remains underexplored due to the thermodynamic and kinetic inertness of carboxylic C-O bonds. Herein, we report a deoxygenative borylation reaction of free carboxylic acids or their sodium salts to synthesise alkylboronates under metal-free conditions. Promoted by a uniquely Lewis acidic and strongly reducing diboron reagent, bis(catecholato)diboron (B2cat2), a library of aromatic carboxylic acids are converted to the benzylboronates. By leveraging the same borylative manifold, a facile triboration process with aliphatic carboxylic acids is also realised, diversifying the pool of available 1,1,2-alkyl(trisboronates) that were otherwise difficult to access. Detailed mechanistic studies reveal a stepwise C-O cleavage profile, which could inspire and encourage future endeavours on more appealing reductive functionalisation of oxygenated feedstocks.

author already reported a reaction of B2cat2, which gives similar products from aldehydes (ref. 25, JACS 2020, 13011). Thus the present reaction is the carboxylic acid version of the previous JACS paper. Besides, this reaction requires a large excess of HMPA, which is a strong carcinogen, as a solvent and relatively high temperature (100 degrees C). The author mentions the advantage of metal-free conditions, but the reaction condition is a toxic and not environmentally benign process. From these points, it is difficult for me to see this paper has enough novelty and usefulness suitable for Nat. Comm.
Other points: 1. The reaction mechanism is still not clear. If the proposed reaction mechanism supposed by the authors can be employed (in Scheme5A), the most important key step is B1 or B2 intermediate to acyl boron compound. This can be investigated by DFT calculations.
2. I don't feel several application studies in Scheme 3 were not variable and appealing. For Scheme 2B, not many researchers want to convert benzoic acid to 1-phenylacetic acid by the present procedure (B2cat2/HMPA; Cu cat/CO2). Similarly, preparations of 2b and 38b'-H2 are little to be gained for the effort. For making these processes more appealing, reactions including more complicated starting materials with functional groups to hard-to-obtain-products should be shown.

Reviewer 1
Li and co-workers report a direct deoxygenative borylation of free carboxylic acids using B 2 cat 2 . It is a novel and elegant paper that develops a simple, practical, and useful protocol for achieving monoand tri-borylated compounds. The authors demonstrate good functional group compatibility (>40 examples) in reasonable yields. The diversifications of products also show the potential of the current methods in synthetic chemistry. The mechanistic studies rationalise the product outcome of the reaction. I believe this work will be well-received by the community.
Response: Thank you for your comment on our work.
Overall, this paper is well written, and the work is thorough, and publication in Nature Communications is strongly recommended after addressing the following issues: 1. In SI, 11 B NMR spectra should be included in addition to 1 H and 13 C NMR.
Response: The 11 B spectra of all boron-containing compounds have been added in the supplementary information for full characterisation.
Response: To study the reactivities of α-branched carboxylic acids, several representative members in this class, including 2-phenylpropionic acid, 2-methylbutanoic acid, ibuprofen, and naproxen, were utilised. We carefully analysed the reaction mixture in the case of 2-phenylpropionic acid using NMR and HRMS. The results were shown below.
Although the desired tris(boronate) product was not observed in this reaction, several boroncontaining products were identified. 1) 1,1-Alkyldiboronate: this side product could be attributed to the selective monoprotodeboronation of desired tris(boronate) product. We believed that the strained scaffold of tris(boronate) product promoted such a hydrolysis step during the reaction since it occurred to a much smaller extent in the case of linear acid examples. In addition, the internal chelation between Bpin groups was proposed to assist the deborylation step, where the carbanion bearing α-phenyl ring was favoured (J. Am. Chem. Soc. 136, 16140-16143 (2014).).

/ 26
2) Vinylboronate: Due in part to the steric hindrance of this branched substrate, the rate of the vicinal diborylation step was decreased, allowing the accumulation of vinylboronate and 1,1vinyldiboronate in the reaction mixture. The formation of these intermediates was consistent with our mechanistic findings, which indicates the important role of vinyl(di)boronates in the deoxygenative borylation of aliphatic substrates.
Similar results were obtained in the cases of 2-methylbutanoic acid, ibuprofen, and naproxen (not shown above).
3. Did the authors try pivalic acid? Logically, the absence of an α-proton could avoid elimination; therefore, a mono-borylated product would be formed.
Response: We have tested the reactivity of some carboxylic acids bearing α-tertiary carbon centres, e.g., pivalic acid and 1-adamantanecarboxylic acid. The results were shown in the scheme below.
As mentioned by the reviewer, the absence of α-protons could avoid elimination; therefore, the formation of 1,1,2-triboronate was impossible with these substrates. However, unlike the aromatic acids (giving benzylmonoboronates), diboronates were obtained as exclusive products in these cases. This result was, indeed, consistent with our previous findings that steric hindrance would decelerate the protodeboronation (J. Am. Chem. Soc. 142, 13011-13020 (2020).). Other than the steric factors, the absence of resonance stabilisation of the anionic intermediate may account for the slow hydrolysis rate, leading to exclusive formation of the gem-diboronate in these cases.
Besides, we noticed the low yields in these examples, which could be attributed to the encumbered structures of these acids, disfavouring the complexation between acid and diboron.

/ 26
Response: We have tested the reactivities of the pyridinecarboxylic acid series, including picolinic acid, nicotinic acid and isonicotinic acid. The results were shown in the scheme below.
All three isomers of the pyridinecarboxylic acids did not give the corresponding pyridylic boronates under our optimal conditions, whether in the free acid or sodium salt forms. This was in accordance with our results in testing the effect of basic additives, in which pyridine showed the inhibitory effect.
Possible rationales for these poor reactivities include: 1) Pyridine shows strong coordination to the diboron compound, which was a typical interaction between Lewis acidic boron compound and the nitrogen base. Characteristic yellow to red colour was observed when mixing the pyridine-based substrates with B 2 cat 2 . For selected references, see J. Am. Chem. Soc. 139, 7440-7443 (2017).; Org. Lett. 19, 4291-4294 (2017).; 2) Some pyridine-diboron complexes could induce dearomatisation of the pyridine ring, which could consume the B 2 cat 2 in our case and hamper the desired reactivity. For selected references, see 5. I notice that this method is compatible with the free hydroxy group. Thus, I recommend the authors try more bioactive substrates cholic acid, and deoxycholic acid, which should add value to the current work.
Response: The reactivities of several bile acids and their derivatives were evaluated under our optimal conditions. The results were shown below.
Among the bile acids tested, although both the cholic acid and deoxycholic acid cannot afford the desired borylated product, 5β-cholanic acid was successfully converted to the corresponding tris(boronate).
To explain such different reactivities, several control experiments were conducted. Under the same conditions, 4-hydroxybenzoic acid was deoxygenatively borylated, while the 4-(hydroxymethyl)benzoic acid did not impart any desired reactivity. These results indicated that our system could tolerate the phenolic proton but not the alcoholic one since the latter, which is more nucleophilic, might transesterify with B 2 cat 2 , generating the B 2 pin 2 -like diboron. This new alcoholbound diboron was shown inactive since B 2 pin 2 was unable to perform the deoxygenative borylation, neither with the free acid nor the sodium carboxylate.
In addition to the bile acid series, we also supplemented several structurally complex samples in the revised manuscript, including some pharmaceutically relevant and isotope-labelled ( 2 H and 13 C) carboxylic acids.

Reviewer 2
The manuscript from Li and co-workers describes a deoxygenative borylation reaction, which takes advantage of carboxylic acids and B 2 cat 2 as substrates and is assisted by HMPA. Although the same group has disclosed the deoxygenative borylation reaction of aldehyde (J. Am. Chem. Soc. 2020, 142, 30, 13011) recently, the novelty of this work is also characterised by its carboxylic starting materials. Generally, this manuscript is well written, and the provided data can well support the conclusion. This work should be not only a supplement to the existing methods that convert acids to boronic derivatives but also a creative one-carbon homologation strategy. Also, this work is featured by the diverse transformation that takes the best use of this deoxygenative borylation reaction. The provided abundant mechanism information gives a detailed reaction pathway for this reaction. In view of this, the reviewer recommends the publication of this manuscript in Nature Communications.
Response: Thank you for your comment on our work.
Before acceptance of this manuscript, several issues that are mentioned below need further discussion: 1. In the substrate scope of deoxygenative borylation of aromatic carboxylic acids, it seems only the benzylic position of the phenyl ring is reactive? Has any other heterocyclic substrate been tried?
Response: Several representative heteroaromatic carboxylic acids/carboxylates with the carboxyl/carboxylate groups on heteroaromatic rings were subjected to our optimal conditions. The results were shown in the scheme below.
In general, the carboxylic/carboxylate groups on heteroaromatic rings showed lower reactivity compared to benzoic acid and its derivatives. This is possibly caused by the intrinsic instability of heteroaromatic rings, which could undergo hydrolytic ring opening or reductive dearomatisation under our conditions (B 2 cat 2 could behave as strong Lewis acid and reductant). For those with fivemembered heterocyclic moieties, their electron-rich carboxyl/carboxylate groups were more reluctant toward reduction.
To be noticed, all three isomers of the pyridinecarboxylic acids did not give the corresponding pyridylic boronates under our optimal conditions, whether in the free acid or sodium salt forms. This

/ 26
was in accordance with our results in testing the effect of basic additives, in which pyridine showed an inhibitory effect even in a catalytic amount.
Possible rationales for these poor reactivities include: 1) Pyridine shows strong coordination to the diboron compound, which was a typical interaction between Lewis acidic boron compound and the nitrogen base. Characteristic yellow to red colour was observed when mixing the pyridine-based substrates with B 2 cat 2 . For selected references, see J. Am. Chem. Soc. 139, 7440-7443 (2017).; Org. Lett. 19, 4291-4294 (2017).; 2) Some pyridine-diboron complexes could induce dearomatisation of the pyridine ring, which could consume the B 2 cat 2 in our case and hamper the desired reactivity. For selected references, see Chem. Sci. 9, 2711Sci. 9, -2722Sci. 9, (2018 We have included these samples and the corresponding description in the supplementary information.
2. Has any amino acid been used as starting material?
Response: Amino acids are valuable starting materials in organic synthesis. Considering the broad availability of glycine and its derivative, some unprotected and N-protected glycines were submitted to our deoxygenative borylation conditions. The results were summarised below.
Although the amino group should, in principle, be compatible with our borylation conditions (see the example of 4-N,N-dimethylaminobenzoic acid), the desired boron products from glycines with or with various protecting groups were not observed in NMR and GC-MS analysis. We suspected that the basic and coordinating amino group adjacent to the carboxylic group would interfere with the complexation between the B 2 cat 2 and the latter, inhibiting the desirable reactivities and remaining as one of the limitations of our methods.

/ 26
3. In Scheme 4d, acyl BMIDA intermediate is used to probe the reaction mechanism. After treating with pinacol, the Bpin product 2b is finally obtained. Which type of boronic species could be obtained before this treatment? BMIDA or Bcat species?
Response: The acylboron compound 2a-BMIDA was chosen in this mechanistic study because it is one of the few acylboronates that were fairly stable and readily accessible (Angew. Chem., Int. Ed. 59, 16847-16858 (2020).).
To identify the active boron species before transesterification, we carefully analysed the reaction mixture before transesterification workup, and we did not observe the Bn-BMIDA 2j in both NMR and GC-MS.
To determine the stability of 2j under transesterification conditions, we performed the following experiments.
In both cases, the 2j remained intact during the transesterification with pinacol. Most of the MIDA ester was recovered, and the pinacol ester was absent or present in trace amount.
Taken together, we concluded that the active boron species in the reaction crude of deoxygenation of acylboron is R-Bcat (due to its instability, isolation were unsuccessful). 4. In supporting information, the authors noted "all the boronates were stored at -20 ℃ to prevent significant decomposition." Since the products are isolated after repeated extraction using water, they should not be very sensitive towards hydrolysis. Could some information about such decomposition given? Should hydrolysis or oxidation be prevented while storing such compounds?
Response: We added this note to the supplementary information by referring to the storage conditions recommended by Combi Blocks. For all the benzylboronic pinacol esters sold by this 10 / 26 vendor, it mentioned, "Store under -20 ℃ or -40 ℃ if to be stored for more than 3 months. Keep the container tightly closed in a dry and well-ventilated place. Containers which are opened must be carefully resealed and kept upright to prevent leakage" (4-Methylbenzylboronic acid pinacol ester as a representative example, please see https://www.combi-blocks.com/cgi-bin/find.cgi?PN-8008).
To provide more information about handling and storing this type of compounds, we examined their sensitivity towards light, air and moisture using benzylboronic acid pinacol ester. The results were shown below. In all these stability test experiments, a high recovery of 2b was obtained, and an insignificant amount of decomposition products, including oxidation and hydrolysis, was observed; therefore, at least in the timeframe from half a day to a month, we could conclude that 2b was not susceptible towards ambient light and atmosphere (O 2 , H 2 O, etc.).
To be noticed, we observed a relatively lower recovery rate when the samples were exposed to the air, which might attribute to the evaporation over time. Besides, due to the limited time, other functionalised benzylboronic acid pinacol esters were not tested, and the stability of 2b should only serve as a standard for a small part of the members in the boronate family. Therefore, following the procedure of Combi Blocks to properly store these kinds of compounds are highly advised for long term storage. Table S5.1.1, should "borylating reagent" be "additive"?

In
Response: Thank you for your suggestion. We have corrected this mistake by replacing "borylating reagent" with "radical tapper".

/ 26
Reviewer 3 This study report an interesting reaction that can transform carboxylic acid into benzylic boron compounds or alkytrisboron compounds with B 2 cat 2 without transition metal catalysts. This provides a new transformation pattern for the production of organoboron compounds. But the author already reported a reaction of B 2 cat 2 , which gives similar products from aldehydes (ref. 25, JACS 2020, 13011). Thus the present reaction is the carboxylic acid version of the previous JACS paper. Besides, this reaction requires a large excess of HMPA, which is a strong carcinogen, as a solvent and relatively high temperature (100 degrees C). The author mentions the advantage of metal-free conditions, but the reaction condition is a toxic and not environmentally benign process. From these points, it is difficult for me to see this paper has enough novelty and usefulness suitable for Nat. Comm..
Response: Thank you for your comment on our work, which mainly focused on 1) similarity to our previous work; 2) high reaction temperature; 3) toxicity of HMPA solvent. Below, we would like to reply to these concerns point by point.

1) Similarity to our previous work
Our idea of using carboxylic acids as unconventional electrophiles in reductive borylation reaction was inspired by our prior efforts on deoxygenative borylation of aldehydes/ketones, where we suffered some intrinsic disadvantages due to the limited availability and poor stability of the latter carbonyl compounds. Therefore, we thought that using carboxylic acids, which benefit from being readily available, cost-effective, shelf-stable and structurally diverse, could advance our previous protocol for boronate synthesis.
During the exploration of this work, we noticed some different and interesting features between our current and previous work. Therefore, we made efforts and supplemented some comparison experiments. a) Alkoxylated benzoic acids and heteroaromatic acids were generally ineffective in the previous protocol; however, under the new conditions, the yields of corresponding benzylboronates were significantly boosted, ranging from 2 to 10 times. We believed that this would be an important addition to the scope of deoxygenative borylation. b) We were unable to convert the sterically hindered aldehydes/ketones such as those bearing -quaternary carbon centres to the desired product. However, we found that the carboxylic acids and their sodium salts, which might undergo different elementary steps in the boronate-forming mechanism, could exhibit different reactivities. Taking 43b' and 44b' as examples, the corresponding aldehydes showed little to no yields, while both acids and sodium salts could give the 1,1-diboronate products.
In particular, the tris[(pinacolato)boryl]ethane (35b), which has not been documented before, represented an important synthetic challenge and could potentially serve as a functionalised ethylating reagent. Although it could be accessed from acetaldehyde (35c) using our previous protocol, such a transformation was inefficient due in part to the high volatility of low-weight aldehyde and some off-target reactivities, such as oligomerisation and condensation of acetaldehyde.
For the synthesis of bisborylmethane (41b'), our attempts to exploit formaldehyde, either in aqueous solution or paraformaldehyde form, were unsuccessful as our aldehyde conditions could not tolerate excessive water and polymeric substrates. Fortunately, simple formic acid was proved viable starting material under the new conditions. d) Taking advantage of our acid method allowed us to synthesise several valuable isotopelabelled boronate building blocks using readily accessible and low-cost acid starting materials. These deuterated and 13 C-labelled boronates could be diversified into various isotopic products that were useful in different contexts. However, simple and convenient routes to prepare the same boronate products still remained obscure in the literature. Arguably, these building blocks could not be easily obtained based on the known borylation protocols.
In principle, the same products could be obtained from the corresponding aldehyde, albeit at much higher costs (prices provided by Sigma Aldrich, and the prices for octadecanal-1-13 C, hydrocinnamic acid-2-13 C and octadecanal-d 35 were not found, therefore, not shown). It is worth mentioning that these aldehydes were mostly derived through multi-step synthesis involving the redox adjustment of the corresponding carboxylic acids.
From the above points, we wished to convey that the upgrade from aldehydes/ketones to carboxylic acids for boronate preparation is not a simple extension but leads to a more practical and enabling approach.
In addition, reductive functionalisation of carboxylic acids or their derivative in a controllable manner was difficult to achieve. Such difficulty could partially be attributed to the formation of unstable aldehyde intermediates, which would be soon reduced into alcohol or methylene as side products (J. Am. Chem. Soc. 142, 8109-8115 (2020).). To relive these unreactive end product (e.g., alcohols and methylarenes for boronate synthesis, extra steps and reagents were required.

2) High reaction temperature
Direct deoxygenative of carboxylic acids to functionalised alkane (R-CO 2 H to R-CH 2 FG) often demanded high thermal energy input because the carboxylic group was generally 15 / 26 thermodynamically stable and kinetically inert. Depending on the catalysed or non-catalysed systems, the reaction temperature generally ranged from 80 ℃ to 220 ℃, and the examples below this range remained rare. For representative samples, please see below.
We attempted to decrease the reaction temperature, which unfortunately led to lower conversion and lower yields in both cases of free acids and their sodium salts.

3) Toxicity of HMPA solvent
In light of the detrimental effect of HMPA, we made several attempts to avoid its usage, including reducing its loading and testing other solvents.

/ 26
More than 25 common organic solvents in different concentrations and their combination in various ratios were examined in the deoxygenative borylation of 4-biphenylcarboxylic acid and its sodium salt. Some representative results were shown in the tables above. From these data, we could see some trends: 1) This diboron-mediated deoxygenation reaction exhibited a very strong and unique dependence on the nature of the solvent. Among all the solvents and their mixtures tested, the desired product was observed in reasonable yield only if the amide-typed solvent was present. This solvent effect could be rationalised by that a) exposing B 2 cat 2 to a coordinating environment was indispensable for its activation and subsequent reactivity, and also proper solubility; b) some solvents were intrinsically incompatible with B 2 cat 2 , which possesses moderate Lewis acidity and reducing capability. For instance, EtOAc and CH 3 CN were prone to hydrolysis in the presence of B 2 cat 2 , and the oxidising DMSO were incompatible with our optimal conditions since DMS was observed.
2) Decreasing the amide solvent loading resulted in a lower yield of the benzylboronate product, presumably due to the less efficient formation of the amide-diboron adduct, which was proposed to be the key intermediate in this transformation.
A quick spot check of some representative substrates showed that reactions in DMA could lead to comparable productivities to those in HMPA. Other examples involving complex carboxylic acid and some isotope-labelled substrates also showed promising results when DMA was employed as solvent.
During the substate scope exploration with DMA, 1,1,2-trisborylated ethane 35b was occasionally observed in small quantity, whose identity was confirmed by both NMR and GC-MS. The presence of such a side product indicated the background reaction between B 2 cat 2 and DMA solvent, which might account for the generally lower yields of reactions conducted in DMA. Armed with this information, a more robust amide solvent, N,N'-diethylacetamide (DEA), was evaluated; however, it yielded less 18 / 26 desired product, possibly because it is too hindered to complex with B 2 cat 2 (see last entry of the left Table C above).
Taken together, if the harmful effect of HMPA was a major concern when using our acid deoxygenative borylation protocol, especially for large-scale synthesis, we showed that DMA served as a more user-friendly yet analogously efficient surrogate for the HMPA in our deoxygenative borylation chemistry. Besides, exploring a more robust coordinating solvent, which was more resistant toward hydrolysis and reduction, would be a promising way to further improve the current protocol.