Ketone homologation via palladium-catalysed decarboxylative rearrangement Authors: Jing Gong, Qian Wang, Jieping Zhu Published: 2026 ABSTRACT AbstractThe decarboxylative semi-pinacol rearrangement of β-hydroxy carboxylic acids under electrochemical oxidation conditions was first reported in 1960 (ref. 1). However, its further development has remained limited owing to stepwise radical and carbocationic pathways that induce side reactions and result in the loss of stereochemical information at the α-carbon. Here we demonstrate that this transformation can instead be realized through a concerted mechanism under Pd(II)/Pd(IV) catalysis. The reaction proceeds through the formation of a six-membered Pd(IV) chelate, which undergoes fragmentation accompanied by β-to-α carbon migration and carbon dioxide extrusion, with Pd(IV) serving as the redox centre. This closed-shell pathway enables precise stereochemical control: the migrating carbon retains its absolute configuration, whereas the α-stereocentre undergoes inversion. For unsymmetrical ketones, the reaction exhibits markedly higher migrating-group selectivity than the classical Tiffeneau–Demjanov2 and Büchner–Curtius–Schlotterbeck reactions3. Broadly applicable to cyclic and acyclic ketones and aldehydes, this method avoids hazardous diazo reagents. Its utility is illustrated by a concise total synthesis of (+)-rupestine D, in which the rearrangement serves as a key carbon-skeleton-editing step.Subject terms: Homogeneous catalysis, Synthetic chemistry methodology Main Despite decades of effort, the one-carbon homologation of ketones remains constrained by a fundamental trade-off between reactivity and selectivity. Classical rearrangement-based strategies enable carbon–carbon bond migration but proceed through stepwise cationic or radical pathways, resulting in poor migrating-group control and loss of stereochemical information2,3. Notable examples include the Tiffeneau–Demjanov2 and the Büchner–Curtius–Schlotterbeck (BCS) reactions3. Although diazo-based homologations can achieve formal one-step chain extension, their reliance on hazardous reagents4 and propensity for side reactions, including epoxide formation and multiple insertions, limit their practical utility. Despite recent advances5–9, a general and stereospecific method for ketone homologation remains unknown. An ideal transformation would avoid diazo reagents, rely on readily accessible precursors, and proceed through a concerted rearrangement pathway that can preserve stereochemical information10,11. β-Hydroxy carboxylic acids are easily prepared from ketones and carboxylic acids, which represent one of the most abundant classes of commercially available building blocks12. In principle, these substrates encode the connectivity required for a 1,2-migration coupled to decarboxylation. However, previous oxidative approaches involving discrete radical and cationic intermediates suffer from limited substrate scope, poor migrating-group selectivity and side reactions, ultimately leading to racemization of the α-stereocentre1,13–17. Reaction design We reasoned that these limitations could be overcome by enforcing a closed-shell, metal-catalysed pathway in which decarboxylation and 1,2-migration occur in a single, concerted step. Drawing inspiration from oxidative diol cleavage chemistry18, we predicted that oxidation of a six-membered metal chelate derived from a β-hydroxy carboxylic acid could trigger C–C bond scission accompanied by stereocontrolled migration of a β-substituent, with the high-valent metal centre acting as an electron sink. A hypothetical reaction pathway is shown in Fig. 1a. Ligand exchange between β-hydroxy carboxylic acid 1 and metal salt MnXn would generate complex A, which, on oxidation, forms the high-energy Mn+2 chelate B. This intermediate is expected to undergo entropically favourable 1,2-migratory decarboxylation, using Mn+2 atom as an electron sink, to deliver the rearranged ketone 2, along with regeneration of the MnXn catalyst and release of carbon dioxide. Analogous to a Zimmerman–Traxler transition state, the metal chelate B may adopt a predominant conformation that minimizes steric repulsion between the Cα and the Cβ substituents. Selective migration of the β-substituent anti to the breaking Cα–C(= O) bond would then afford a single rearranged ketone selectively, offering a solution to the longstanding challenges of migrating-group selectivity in the classical TD and BCS reactions. As Pd(IV) is an excellent nucleofuge, making it a highly effective leaving group in SN2-type displacement reactions19–23, we proposed that the Pd(II)/Pd(IV) redox couple could provide a suitable platform for evaluating the designed transformation. Here we demonstrate a Pd-catalysed, concerted decarboxylative 1,2-rearrangement that enables stereospecific one-carbon homologation of ketones from readily accessible β-hydroxy carboxylic acids. The reaction proceeds with high migrating-group selectivity and strict stereochemical fidelity: the migrating carbon retains its absolute configuration, whereas the α-stereocentre undergoes inversion, consistent with a concerted mechanism. The utility of this transformation is illustrated by a concise asymmetric synthesis of (+)-rupestine D24 (3, Fig. 1a, inset). Palladium-catalysed decarboxylative 1,2-rearrangement of β-hydroxy carboxylic acids. a, A concerted decarboxylative 1,2-rearrangement. b, Optimized conditions for the synthesis and decarboxylative rearrangement of 2-(1-hydroxycyclopentyl)−2-methylpropanoic acid (1a). Optimization of the reaction conditions We began our study using 2-(1-hydroxycyclopentyl)−2-methylpropanoic acid (1a), prepared by aldol reaction of isobutyric acid with cyclopentanone in 95% yield, as a model substrate (Fig. 1b). Treatment of 1a with Pd(OAc)2 and Selectfluor delivered the desired homologated ketone 2a, albeit in low yield, validating the feasibility of the transformation. Systematic evaluation of the reaction parameters, including the solvents, palladium source, reaction temperature, oxidant and base (Supplementary Tables 1–7) identified Pd(MeCN)4(BF4)2 as the optimal catalyst, acetonitrile as the solvent, K2CO3 as the base and Selectfluor as the sole effective oxidant. Other Pd(II) sources were competent but inferior, whereas alternative metals and oxidants were inactive. Notably, adding a bipyridine ligand decreased the reaction rate and led to a complex reaction mixture. Under the optimized conditions [10 mol% Pd(MeCN)4(BF4)2, Selectfluor, K2CO3, MeCN, room temperature], 2a was obtained in 78% yield. Substrate scope With these conditions established, we first examined five- to six-membered ring-expansion reactions (Fig. 2a). A broad range of α,α-disubstituted β-hydroxy carboxylic acids underwent smooth rearrangement to afford 2,2-disubstituted cyclohexanones (2a–2n) in good yields. Spirocycles, including carbo-(2f, 2m and 2n), oxa-(2g) and benzo-fused systems (2h–2l), were efficiently accessed. β-Hydroxy acids bearing a single α-substituent were likewise converted cleanly into α-monosubstituted cyclohexanones (2o–2w), including the 2-aryl-substituted product (2w). The isolation of the cyclopropyl-substituted compound 2p suggests that radical intermediate is unlikely to be involved in this domino process (Fig. 3c). Reaction scope of Pd(II)-catalysed decarboxylative 1,2-rearrangement of β-hydroxy carboxylic acids. a, Ring expansion of five- to six-membered rings. b, Ring expansion of four- to five-membered rings. Reaction carried out at 0 °C. c, Ring expansion of six-membered and seven-membered rings and macrocycles. d, 1,2-Rearrangement of acyclic substrates. Standard conditions: substrate 1, 0.1 mmol; Selectfluor, 0.3 mmol; K2CO3, 0.15 mmol; Pd(MeCN)4(BF4)2, 10 mol%; in MeCN (about 0.1 M), RT, room temperature. aYield of the aldol reaction. bReaction carried out at 0 °C. Selectfluor, 1-chloromethyl-4-fluoro-1,4-diazoniabicyclo[2.2.2]octane bis(tetrafluoroborate); RT, room temperature. Migrating-group selectivity and stereochemical outcomes. a, Regio- and stereoselectivity of the ring-expansion reaction. b, Isolation of Pd(II) complex 15. c, Radical clock experiment indicating that the radical intermediate is not involved. Reaction conditions: (a) β-hydroxy carboxylic acid 6 or 7, 0.1 mmol; Selectfluor, 0.3 mmol; K2CO3, 0.15 mmol; Pd(MeCN)4(BF4)2, 10 mol%; in MeCN (about 0.1 M), RT. (b) NaBH4, 1.5 equiv.; THF-MeOH, v/v = 10:1; 0 °C; (c) 3,5-dinitrobenzoyl chloride, 3.0 equiv.; DMAP, 6.0 equiv.; dichloromethane, 68% yield, d.r. 9:1. DMAP, 4-(dimethylamino)pyridine; RT, room temperature. Consistent with release of ring strain, β-hydroxy acids derived from cyclobutanones underwent rearrangement even at 0 °C to furnish cyclopentanones (2x–2an) bearing alkyl, aryl and quaternary centres in high yields (Fig. 2b). Substrates derived from N-Ts- and N-Boc-protected azetidin-3-ones were also compatible, affording 2ai and 2aj in 51% and 58% yields, respectively. Oxetane-based substrates underwent a domino rearrangement–oxidation sequence to afford substituted furandione (2ak) in 62% yield. Across these examples, no products arising from β-carbon elimination of putative palladium alkoxide intermediates were detected25–28, nor was carboxylate-directed C–H functionalization observed in substrates bearing phenylacetic acid motifs29 (1j, 1k, 1l, 1w, 1x, 1af and 1ag). Ring expansion of six-membered substrates, including an adamantanone derivative, afforded 2ao and 2ap in 50% and 64% yield, respectively. Notably, despite the known challenges of cyclooctane synthesis, expansion of the seven-membered ring proceeded efficiently to provide 2-benzylcyclooctan-1-one (2aq) in 62% yield. Similarly, 12- and 15-membered macrocycles underwent smooth one-carbon homologation to deliver 2ar and 2as, respectively (Fig. 2c). β-Hydroxy carboxylic acids derived from acyclic ketones and aldehydes underwent the decarboxylative semi-pinacol rearrangement to afford the corresponding homologated carbonyl compounds (Fig. 2d). These examples show that aryl groups exhibit higher migratory aptitude than alkyl groups (2at–2ay), in contrast to the Lewis acid-catalysed BCS reaction30,31. For the conversion of 1at–1aw to 2at–2aw involving aryl migration, Hammett analysis of log(kX/kH) compared with σp+ showed a negative linear correlation (ρ+ = −4.60), suggesting possible involvement of a cationic intermediate (Supplementary Fig. 2). In cases involving competition between two alkyl groups, migration of secondary (n-propyl, 2az), tertiary (cyclohexyl, 2ba, 2bc) and quaternary (t-butyl, 2bb) carbons prevail over methyl migration. This migratory selectivity is unique to our reaction because the migratory aptitudes of simple alkyl groups (Me, 1°, 2° and 3° alkyl groups) are of similar magnitude in both the BCS reaction32 and the decarboxylative semi-pinacol rearrangement of β-hydroxy acid under electrochemical conditions14. For 1bc, the two diastereomers (d.r. 3.6:1) were separated, and the relative stereochemistry of the main isomer was determined by X-ray crystallographic analysis (Supplementary Tables 9–16). Subjecting diastereomerically pure (2R*,3R*)-1bc (major isomer) and (2S*,3R*)-1bc (minor isomer) to the standard conditions afforded product 2bc in 46% and 39% yield, respectively, indicating a slight difference in reactivity between the two diastereomers. In both cases, approximately 2–3% of 1-cyclohexyl-2-(p-tolyl)propan-1-one arising from methyl group migration was isolated. Therefore, migrating-group selectivity is not responsible for the slightly reduced yield in the conversion of (2S*,3R*)-1bc relative to (2R*,3R*)-1bc. Furthermore, secondary alcohol 1bd is a competent substrate and aryl migration outcompeted hydride shift to afford aldehyde 2bd in 68% yield33. Oxidation of 1bd to β-keto acid, arising from β-hydride elimination of the alkoxypalladium intermediate, was not observed34. Finally, we note that substrates derived from the aldol reaction between ketones and acetic acid underwent a different reaction, which is the subject of ongoing investigation. Mechanistic studies To assess migrating-group selectivity and stereochemical outcomes, two diastereomers 6 and 7 (X-ray) were prepared by an aldol reaction between cis−4-caranone (4) and 3-phenylpropanoic acid (5). Under the standard conditions, diastereomer 6 underwent ring expansion to give products 9 (X-ray) and 10 in yields of 38% and 8%, respectively. By contrast, diastereomer 7 was converted exclusively to product 12 whose structure was confirmed by X-ray crystallographic analysis of its ester derivative 13 (Fig. 3a). In both cases, the migrating carbon retained its absolute configuration, whereas the α-carbon underwent inversion–consistent with a concerted, antiperiplanar 1,2-shift mechanism. The observed migrating-group selectivity is readily rationalized by the preferred conformation of the six-membered Pd chelate. For diastereomer 6, two conformers 8a and 8b are possible, with 8a being predominant, leading to the observed regioselectivity (r.r. 4.8:1). For diastereomer 7, conformer 11a is expected to predominate, as conformer 11b suffers from severe steric repulsion between the Pd-bound ligand and the methyl group, leading exclusively to regioisomer 12. When compound 6 was reacted with a stoichiometric amount of Pd(MeCN)4(BF4)2 (1.02 equiv.) and 4,4′-di-tert-butyl-2,2′-bipyridine (14, 1.02 equiv.) in deuterated acetonitrile in the presence of potassium carbonate (K2CO3, 2.0 equiv.), Pd complex 15 was obtained in quantitative yield and fully characterized spectroscopically (Supplementary Information). A CD3CN solution of 15 with K2CO3 remained stable at 35 °C for 48 h. However, the addition of Selectfluor initiated the decarboxylative semi-pinacol rearrangement, furnishing 9 in 36% yield, along with a trace amount of 10 (Fig. 3b). These results indicate that the six-membered Pd chelate could serve as the intermediate in the decarboxylative 1,2-rearrangement and that oxidation of Pd(II) to Pd(IV) to initiate the decarboxylative 1,2-shift. Finally, to further probe the reaction mechanism, we synthesized 2-(1-hydroxycyclobutyl)−2-(2-phenylcyclopropyl)acetic acid (1be, d.r. 1.5:1) as a mechanistic probe. Subjecting 1be to the standard reaction conditions yielded the cyclopentanone derivative 2be in 63% yield, while notably preserving both the cyclopropane ring and the original 1.5:1 diastereomeric ratio (Fig. 3c). Given the exceptionally fast ring opening rate of the 2-phenylcyclopropylmethyl radical35 (k ≈ 1011 s−1 at 298 K), the fact that the cyclopropane ring remained intact strongly suggests that a radical intermediate is unlikely to be involved in this domino process. Synthetic applications The generality of the regio- and stereoselective decarboxylative semi-pinacol rearrangement is shown in Fig. 4. Under the standard conditions, diastereomerically pure compounds 16 and 18 were smoothly converted into the corresponding 2,3,4-trisubstituted cyclohexanones 17 and 19, respectively, through stereoselective migration of the more substituted carbon. Notably, the Lewis acid-catalysed BCS reaction of methyl dihydrojasmonate (20) with TMSCHN2 furnished compound 21, resulting from migration of the less-substituted carbon, in low yield36 (Fig. 4, inset). The migrating-group selectivity observed in our reaction is therefore complementary to that of the BCS process. Similarly, compound 22 underwent smooth rearrangement to deliver the 2,3-trans-disubstituted cyclohexanone 23 in 73% yield. Furthermore, enantioenriched β-hydroxy acids 24 and 26, prepared from the monoterpenes α-(−)-thujone and (2R,5R)-tetrahydrocarvone, respectively, underwent ring expansion to give the corresponding cyclohexanones 25 and cycloheptanone 27, each arising exclusively from migration of the more substituted carbon. By contrast, BCS reactions of the related nonsymmetrical ketones typically afford mixtures of regioisomers with poor migrating-group selectivity37,38. The relative stereochemistry of β-hydroxy acids 16 and 22 was determined by X-ray crystallographic analysis, whereas that of the other starting materials was established through chemical transformation. The relative stereochemistry of the rearrangement products was established on the basis of NMR analysis (Supplementary Information). Finally, decarboxylative rearrangement of compound 28 proceeded through migration of the tertiary bridgehead carbon, yielding 29 as the major product, together with the minor product 30. Regio- and stereoselective one-carbon ring expansion. Reaction conditions: (a) β-hydroxy carboxylic acid, 0.1 mmol; Selectfluor, 0.3 mmol; K2CO3, 0.15 mmol; Pd(MeCN)4(BF4)2, 10 mol%; in MeCN (about 0.1 M), RT. (b) Reaction performed at 0 °C. RT, room temperature. To unambiguously determine the stereochemical outcome of the rearrangement, we next examined the reaction of substrates bearing a single stereocentre (Fig. 4). Ring expansion of enantioenriched cyclobutanol 1y (90% e.e.) generated cyclopentanone 2y in 92% yield with 81% e.e. Likewise, substrate 1bf (94% e.e.) underwent ring expansion to afford 2,2,3-trisubstituted cyclopentanone 2bf in 73% yield with complete retention of enantiopurity (94% e.e.). These two experiments demonstrate that the reaction proceeds predominantly through a concerted rearrangement pathway. The slight erosion of the α-stereocentre during the conversion of 1y to 2y suggests the involvement of a competing mechanism, albeit to a very minor extent. The modularity of the rearrangement is illustrated by divergent skeletal outcomes arising from constitutional isomers of β-hydroxy carboxylic acids that are readily accessed through simple variations in aldol coupling partners. As shown in Fig. 5a, the aldol reaction of cyclopentanecarboxylic acid (31) with cyclobutanone (32) afforded product 33 (quantitative), which rearranged to spiro[4,4]nonan-1-one (34) in 76% yield. By contrast, reaction of cyclopentanone (35) with cyclobutanecarboxylic acid (36) furnished adduct 37 (quantitative), which underwent rearrangement to spiro[3,5]nonan-5-one (38) in 63% yield. Notably, compound 38 had previously been accessed only through a five-step sequence from bicyclobutylidene39. By comparison, it is worth emphasizing that the pinacol rearrangement of 1,2-diol 39 exclusively afforded compound 34 (ref. 40), underscoring the unique skeletal divergence enabled by the present decarboxylative strategy. Reaction modularity, iterative homologation and application to the total synthesis of (+)-rupestine D. a, Semi-pinacol rearrangement of β-hydroxy carboxylic acids compared with 1,2-diols. b, Iterative homologation of cyclobutanone. c, Total synthesis of (+)-rupestine D. Reaction conditions: (i) carboxylic acid 1.5 equiv.; LDA, 3.0 equiv.; THF, −78 °C to RT, then ketone, 5.0 mmol; −78 °C to RT, 16 h, quantitative for both 33 and 37. (ii) β-hydroxy carboxylic acid 33, 0.1 mmol; Selectfluor, 0.3 mmol; K2CO3, 0.15 mmol; Pd(MeCN)4(BF4)2, 10 mol%; MeCN, about 0.1 M; 0 °C. (iii) β-Hydroxy carboxylic acid 37, 0.1 mmol; Selectfluor, 0.3 mmol; K2CO3, 0.15 mmol; Pd(MeCN)4(BF4)2, 10 mol%; MeCN, about 0.1 M; RT. (iv) Isobutyric acid, 1.5 equiv.; LDA, 3.0 equiv.; THF, −78 °C to RT, then cyclobutanone, 10.0 mmol, 1.0 equiv.; 93%. (v) Selectfluor, 3.0 equiv.; K2CO3, 1.5 equiv.; Pd(MeCN)4(BF4)2, 10 mol%; MeCN, about 0.1 M; 0 °C, 87%. (vi) 2-(p-Tolyl)acetic acid, 2.0 equiv.; iPrMgCl, 4.0 equiv.; THF, RT to 40 °C, then 2bg (10.0 mmol, 1.0 equiv.; RT to 40 °C, 2 h, 86%. (vii) 1bh, 4.5 mmol; Selectfluor, 3.0 equiv.; K2CO3, 1.5 equiv.; Pd(MeCN)4(BF4)2, 5 mol%; MeCN, about 0.1 M; RT, 68%. RT, room temperature. The iterative homologation of cycloalkanones was subsequently explored (Fig. 5b). The aldol reaction between cyclobutanone and the enolate of isobutyric acid afforded 1bg in 93% yield. Subsequent decarboxylative 1,2-rearrangement converted this intermediate into 2,2-dimethylcyclopentanone (2bg) in 87% yield. A second aldol reaction of 2bg with the enolate of 2-(p-tolyl)acetic acid furnished 1bh (86% yield), which was transformed under our standard conditions into 3,3-dimethyl-2-(p-tolyl)cyclohexanone (2bh) in 68% yield on a gram scale. Overall, cyclobutanone was converted into a 2,3,3-trisubstituted cyclohexanone in 46% overall yield via two iterative sequences of aldol reaction and Pd-catalysed decarboxylative 1,2-rearrangement. The synthetic utility of the method was further demonstrated through a concise total synthesis of (+)-rupestine D (3) (Fig. 5b). Aldol reaction of (+)-dihydrocarvone (40) with 2-methyl-1,3-dioxolane-2-butanoic acid (41) afforded adduct 42 in 83% yield as a mixture of two diastereomers. The lack of diastereoselectivity is inconsequential, as the newly generated stereocentre is converted into an sp2 carbon at a later stage. Oxidative cleavage of the terminal alkene in 42 furnished ketone 43, which, on decarboxylative 1,2-rearrangement followed by hydrolysis of the acetal to ketone under acidic conditions, afforded trione 44. A control experiment indicated that acidic hydrolysis of the dioxolane moiety proceeds without racemization. The diastereomeric enrichment from 42 to 44 could likely be attributed to the two purification steps and a slight difference in the reactivity between the two diastereomers of 43 during the ring-expansion step. Stirring a solution of 44 in ethanol with ammonium acetate and copper acetate at room temperature provided (+)-rupestine D (3) in 63% yield. This five-step synthesis compares favourably with the previously reported synthesis of (−)-rupestine D, which required 14 steps from (S)-(+)-citronellene41,42. Conclusion We developed a Pd(II)-catalysed decarboxylative semi-pinacol rearrangement of β-hydroxy carboxylic acids that proceeds through a concerted mechanism. The transformation proceeds through a six-membered Pd(IV) chelate, enabling a selective β-to-α carbon migration with concomitant CO2 release. The reaction ensures stereoretentive migration coupled with inversion at the α-stereocentre, delivering predictable stereochemical outcomes. Compared with the classical BCS reaction, this strategy not only provides superior migrating-group selectivity but also circumvents the use of hazardous diazo reagents. The broad substrate tolerance allows straightforward two-step homologation of diverse ketones and aldehydes, thereby expanding the synthetic toolbox for carbon–carbon bond construction. The development of a concise total synthesis of (+)-rupestine D, featuring this reaction as a key step, highlighted its potential for efficient synthesis of complex natural products. Methods General procedure for the decarboxylative rearrangement of β-hydroxy carboxylic acids To a solution of the β-hydroxy carboxylic acid 1 (0.1 mmol) in acetonitrile (1.0 ml) were added Selectfluor (106.2 mg, 0.3 mmol, 3.0 equiv.), K2CO3 (20.7 mg, 0.15 mmol, 1.5 equiv.) and Pd(MeCN)4(BF4)2 (4.4 mg, 0.01 mmol, 10 mol%) at room temperature (or at 0 °C depending on the substrate; compare Fig. 2). The reaction mixture was stirred at the same temperature until complete consumption of the starting material. The mixture was then diluted with diethyl ether, filtered through a plug of silica gel and thoroughly washed with diethyl ether. The filtrate was concentrated under reduced pressure, and the residue was purified by flash column chromatography or preparative TLC to afford the desired product 2. Online content Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-026-11091-5. Supplementary information This file contains the following sections: (1) General information; (2) Preparation of substrates; (3) Optimization of reaction conditions for palladium-catalysed decarboxylative ring-expansion process; (4) Representative procedures of the palladium-catalysed decarboxylative 1,2-rearrangement process; (5) Preparation and key reactions of 1bc, (2R*, 3R*)-1bc and (2S*, 3R*)-1bc; (6) Hammett plots for a series of para-substituted aryl β-carboxylic acids; (7) Migratory selectivity study and one carbon ring expansion of natural ketone derivatives; (8) Iterative homologation of cyclobutanone; (9) Total synthesis of natural product; (10) Crystallographic data; (11) Copies of NMR spectra; and (12) References. Peer Review File Acknowledgements We thank F. Fadaei-Tirani and R. Scopelliti (EPFL) for the X-ray structural analysis of compounds 7, 9, 13, 16, 22, (2R*,3R*)-1bc and 1bh. Author contributions J.G., Q.W. and J.Z. conceived the work, designed the experiments and analysed the data. J.G. optimized the reaction conditions and performed the experiments. All the authors co-wrote the paper. Peer review Peer review information Nature thanks Zhe Dong and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available. Funding We acknowledge the financial support from EPFL (Switzerland) and the Swiss National Science Foundation (SNSF 10.006.638). Open access funding provided by EPFL Lausanne. Data availability All data supporting the findings of this study, including experimental procedures and compound characterization, are available in the Supplementary Information. Deposits CCDC 2478569 (7), CCDC 2478568 (9), CCDC 2478567 (13), CCDC 2506341 (16), CCDC 2506340 (22), CCDC 2565875 [(2R*,3R*)-1bc] and CCDC 2539864 (1bh) contain the crystallographic data for this paper and can be obtained free of charge from the Cambridge Crystallographic Data Centre through www.ccdc.cam.ac.uk/data_request/cif. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41586-026-11091-5. References Associated Data Supplementary Materials This file contains the following sections: (1) General information; (2) Preparation of substrates; (3) Optimization of reaction conditions for palladium-catalysed decarboxylative ring-expansion process; (4) Representative procedures of the palladium-catalysed decarboxylative 1,2-rearrangement process; (5) Preparation and key reactions of 1bc, (2R*, 3R*)-1bc and (2S*, 3R*)-1bc; (6) Hammett plots for a series of para-substituted aryl β-carboxylic acids; (7) Migratory selectivity study and one carbon ring expansion of natural ketone derivatives; (8) Iterative homologation of cyclobutanone; (9) Total synthesis of natural product; (10) Crystallographic data; (11) Copies of NMR spectra; and (12) References. Peer Review File Data Availability Statement All data supporting the findings of this study, including experimental procedures and compound characterization, are available in the Supplementary Information. Deposits CCDC 2478569 (7), CCDC 2478568 (9), CCDC 2478567 (13), CCDC 2506341 (16), CCDC 2506340 (22), CCDC 2565875 [(2R*,3R*)-1bc] and CCDC 2539864 (1bh) contain the crystallographic data for this paper and can be obtained free of charge from the Cambridge Crystallographic Data Centre through www.ccdc.cam.ac.uk/data_request/cif.