Introduction

Oxetanes, azetidines, thietanes, and cyclobutanes are all saturated four-membered rings. Their overall carbon frameworks are similar, with the principal difference lying in the atom or group within the ring that connects the three-carbon fragment. In conventional synthesis, however, replacing the endocyclic oxygen atom with nitrogen, sulfur, or carbon generally cannot be accomplished through a routine functional-group transformation. Instead, different starting materials must usually be selected and the four-membered ring constructed anew.

In 2025, Ying-Qi Zhang, Shuo-Han Li, Xinglong Zhang, and Ming Joo Koh reported a photocatalytic oxygen-atom replacement method in Nature. In this method, an oxetane is first selectively ring-opened to form an acyclic 1,3-dibromide intermediate. Different nucleophiles are then used to reconstruct four- or five-membered rings, thereby providing structures such as azetidines, thietanes, cyclobutanes, pyrazolidines, and isoxazolidines.[1]

The central concept of this study is to remove the original endocyclic oxygen linkage and then insert a new atom or structural unit to achieve localized skeletal reconstruction.

1. Why Endocyclic Atom Replacement Is More Complex than a Conventional Substitution Reaction

1.1 The endocyclic oxygen forms two skeletal linkages

A conventional substituent is typically connected to the main molecular framework through only one chemical bond. After that substituent is removed, only one new bond needs to be formed at the same position.

The oxygen atom in an oxetane is different. It is bonded to two ring carbons and forms part of the four-membered-ring framework:

Removing the oxygen atom simultaneously cleaves two carbon–oxygen bonds and opens the four-membered ring. Therefore, although the net transformations may be represented as O→N, O→S, or O→C replacement, they actually involve three chemical processes:

1. Selective activation of the endocyclic oxygen;

2. Opening of the four-membered ring while retaining the three-carbon fragment;

3. Introduction of a new nucleophilic unit followed by ring closure.

Such transformations constitute skeletal editing rather than a conventional single-bond functional-group exchange.[3]

1.2 Conventional routes are difficult to reuse across different four-membered rings

Azetidines are commonly constructed from nitrogen-containing precursors through nucleophilic substitution or cyclization reactions; thietanes require sulfur-containing precursors; and cyclobutanes often rely on carbon–carbon bond-forming reactions. Once the target endocyclic atom is changed, the precursor structure, protecting-group arrangement, and cyclization conditions generally need to be adjusted accordingly.[6]

The problem, therefore, is not simply the low efficiency of a particular reaction step, but rather the lack of a common advanced intermediate that can be used for different target structures.

2. Reaction Design: Converting an Oxetane into a 1,3-Dibromide Intermediate

2.1 An oxetane can be viewed as a cyclized three-carbon fragment

From the perspective of connectivity, an oxetane is equivalent to a three-carbon chain whose two ends are connected by an oxygen atom. If the two carbon–oxygen bonds are replaced by carbon–bromine bonds, the cyclic structure is converted into a 1,3-dibromide:

A 1,3-dibromide contains two carbon–bromine bonds that readily undergo nucleophilic substitution. A nucleophile can first displace one bromide and then replace the second bromide through intramolecular nucleophilic substitution, thereby closing the ring.

The key to the reaction design is therefore not to replace the oxygen directly within the intact four-membered ring, but first to prepare a common intermediate that can react with multiple classes of nucleophiles.

2.2 Representative reaction sequence

In the model O→N reaction, tris(2,2′-bipyridine)ruthenium(II) chloride hexahydrate, Ru(bpy)₃Cl₂·6H₂O, is used as the photocatalyst, and carbon tetrabromide, CBr₄, serves as the bromine source. The reaction mixture is first irradiated with 456 nm blue light in anhydrous N,N-dimethylformamide, DMF, at room temperature for 2 h. The light source is then switched off, and the mixture is heated at 60 °C for 3 h to promote formation of the 1,3-dibromide intermediate. After cooling, 4-methoxyaniline, N,N-diisopropylethylamine, DIPEA, and DMSO are added, and azetidine formation is completed in a mixed DMF/DMSO solvent system. The model product was isolated in 85% yield.[1]

The overall reaction can be summarized as follows:

Oxetane

│ Ru(bpy)₃Cl₂·6H₂O, CBr₄, anhydrous DMF

│ 456 nm blue light, room temperature

↓

Light source switched off; heating at 60 °C

↓

1,3-Dibromide intermediate

│ Nucleophile, base

↓

Reconstructed four- or five-membered ring

The photocatalytic step converts the oxetane into the dibromide intermediate, whereas the final endocyclic atom or structural unit is determined by the nucleophile added in the subsequent step.

3. The Authors’ Proposed Mechanism for Photocatalytic Ring Opening and Its Basis

The following pathway is a mechanistic model proposed by the authors on the basis of control experiments, DFT calculations, and established photoredox chemistry.

3.1 Generation of the activating reagent through a photoredox cycle

Upon absorbing blue light, the Ru(bpy)₃²⁺ photocatalyst enters an excited state. The authors propose that the excited ruthenium complex undergoes single-electron transfer to CBr₄, generating Ru(bpy)₃³⁺, a tribromomethyl radical, and bromide.

The tribromomethyl radical is subsequently proposed to form a radical adduct with DMF. Oxidation of this adduct by Ru(bpy)₃³⁺ produces cationic intermediate I while regenerating Ru(bpy)₃²⁺, thereby completing the photoredox cycle. This design was inspired by related studies on the visible-light-mediated conversion of alcohols into halides.[1,5]

In this step, light serves to drive single-electron transfer and generate the activating species required for subsequent ring opening, rather than directly cleaving the carbon–oxygen bonds of the oxetane.

3.2 Activation of the endocyclic oxygen by a Vilsmeier–Haack-type reagent

According to the computational pathway proposed by the authors, bromide attacks cationic intermediate I, and subsequent carbon–oxygen bond cleavage generates Vilsmeier–Haack-type reagent II, together with carbonyl dibromide, Br₂CO, and bromide. DFT calculations indicate that this process is thermodynamically favorable.[2]

The oxetane oxygen uses a lone pair to attack the iminium carbon of reagent II, forming oxonium-type intermediate III. At this stage, the original carbon–oxygen bonds are substantially activated because the oxygen atom bears a positive charge. Bromide then attacks a ring carbon through a bimolecular nucleophilic substitution, S_N2, pathway, cleaving the first carbon–oxygen bond and generating ring-opened intermediate IV.

The DFT results show that:

1. The calculated activation free energy for attack of the oxetane on the Vilsmeier–Haack-type reagent is 16.6 kcal·mol⁻¹;

2. The barrier for the subsequent first bromide-induced ring-opening step is only 1.7 kcal·mol⁻¹;

3. Ring-opened intermediate IV is substantially lower in energy than intermediate III, indicating that release of ring strain provides a strong driving force for the reaction.[2]

3.3 Second bromination to form the 1,3-dibromide

One end of ring-opened intermediate IV has already formed a carbon–bromine bond, whereas the other end remains connected through oxygen to a DMF-derived group. A second bromide then undergoes another S_N2 substitution, cleaving the remaining carbon–oxygen bond and releasing DMF to afford 1,3-dibromide intermediate V.

In the calculated reaction pathway, formation of the second carbon–bromine bond is the higher-barrier step. The calculated activation free energy from intermediate IV is 25.1 kcal·mol⁻¹. Increasing the temperature promotes formation of the dibromide and reduces the competing hydrolysis of intermediate IV to a formate byproduct.[2]

3.4 Experimental results support formation of the dibromide intermediate

The authors obtained two key pieces of evidence through control experiments:

These results are consistent with a reaction pathway involving DMF-mediated activation, oxetane ring opening, dibromide formation, and nucleophile-induced ring closure.[1]

4. Origin of Selectivity for the Endocyclic Oxygen

Complex molecules may contain an oxetane together with several other oxygen-containing functional groups, such as acyclic ethers, esters, and amides. If the activating reagent cannot distinguish among these oxygen atoms, bond cleavage may occur at an undesired site.

Model substrate 8 contains both an oxetane and an acyclic ether. Under the standard conditions, the oxetane undergoes O→N replacement, whereas the acyclic ether remains largely unchanged. Computational results attribute this selectivity to two consecutive factors.[1–2]

4.1 Attack by the endocyclic oxygen on the electrophilic reagent is more favorable

For model substrate 8, transition state TS3 for attack of the oxetane oxygen on the Vilsmeier–Haack-type reagent is 10.8 kcal·mol⁻¹ lower in energy than transition state TS3a for the corresponding attack by the acyclic ether oxygen.

The calculated electronic-structure parameters show the same trend:

For this model substrate, the local Fukui functions and calculated charges are consistent with the oxetane oxygen having greater nucleophilic reactivity. It is therefore more likely to form a bond first with the electrophilic iminium carbon.[2]

4.2 Four-membered-ring strain lowers the barrier to subsequent ring opening

After formation of the oxonium intermediate, the calculated barrier for bromide attack at an oxetane ring carbon is 1.7 kcal·mol⁻¹. The corresponding barrier for substitution at the carbon–oxygen bond of the acyclic ether is 7.4 kcal·mol⁻¹.

Opening of the four-membered ring releases ring strain, whereas the acyclic ether lacks this driving force. Ring strain therefore mainly promotes carbon–oxygen bond cleavage after oxygen activation, rather than direct recognition of the substrate by the photocatalyst.[2]

The selectivity can be summarized in two stages:

Greater nucleophilicity of the endocyclic oxygen → Preferential formation of the oxetane-derived oxonium intermediate → Release of four-membered-ring strain → Preferential bromide-induced ring opening

5. Nucleophile-Driven Ring Reconstruction Determines the Product Type

After formation of the 1,3-dibromide, different nucleophiles construct different ring systems through successive nucleophilic substitutions. The principal transformations investigated in the paper are summarized below.[1]

5.1 O→N: Formation of azetidines

The amine first displaces bromide at one end of the dibromide, forming an open-chain amine that retains the remaining carbon–bromine bond. The nitrogen atom then undergoes an intramolecular S_N2 substitution, forming the second carbon–nitrogen bond and closing the azetidine ring.

The reaction is applicable to a variety of arylamines, alkylamines, and sulfonamides. It can also accommodate substrates containing esters, amides, aryl halides, boronate esters, heteroarenes, and other functional groups.[1]

5.2 O→S: Formation of thietanes

When sodium sulfide nonahydrate is used as the sulfur source, the sulfur nucleophile can sequentially displace both bromides and close the ring to form a thietane. Compared with nitrogen, sulfur can also be further oxidized to a sulfoxide or sulfone, providing sulfur-containing structures in different oxidation states on the same carbon framework.

The O→S transformations reported in the paper encompass a variety of substituted oxetanes and were also applied to the late-stage modification of complex molecules.[1]

5.3 O→C: Replacement with carbon-based units and formation of cyclobutanes

Electron-deficient nitriles and malonate derivatives can be deprotonated with a base to form relatively stabilized carbon nucleophiles. The carbon nucleophile first undergoes intermolecular substitution and then closes the ring through intramolecular carbon–carbon bond formation, affording a substituted cyclobutane.

These reactions demonstrate that the inserted unit is not limited to a heteroatom, but can also be a carbon center bearing a cyano or ester group. The resulting functional groups can subsequently undergo hydrolysis, reduction, deoxyfluorination, or other transformations.[1]

5.4 Two-atom replacement and ring expansion

Hydrazine and hydroxylamine derivatives contain two adjacent nucleophilic atoms. When they react with the 1,3-dibromide intermediate, the original four-membered ring is not restored. Instead, an N–N or N–O unit is inserted between the two ends of the three-carbon fragment, producing a five-membered pyrazolidine or isoxazolidine.

This demonstrates that the reaction can not only change the endocyclic atom but also alter the ring size through insertion of a two-atom unit.[1]

6. Applications in the Synthesis of Drug-Related Molecules

6.1 Late-stage modification of the endocyclic atom

Skeletal editing is suitable for comparing compounds with closely related structures. Researchers can preserve the aromatic rings, side chains, and stereogenic centers of a molecule while changing only the oxygen, nitrogen, sulfur, or carbon center within the four-membered ring, thereby investigating the effects of the endocyclic atom on activity, polarity, metabolic stability, and conformation.

The paper converted an oxetane-containing phosphodiesterase 4, PDE4, inhibitor into the corresponding thietane analogue:

1. The original oxetane compound had a half-maximal inhibitory concentration, IC₅₀, of 52 nM;

2. The thietane analogue had an IC₅₀ of 17 nM;

3. The O→S transformation proceeded in 41% yield.

In this example, the sulfur-containing analogue showed approximately threefold higher inhibitory activity in vitro. This demonstrates that endocyclic atom replacement can rapidly provide matched analogues for direct biological-activity comparisons. However, a single example does not imply that every O→S transformation will improve biological activity.[1]

The study also converted an oxetane-containing hepatitis B virus inhibitor scaffold into the corresponding thietane derivative, demonstrating that late-stage editing remained possible in the presence of multiple nucleophilic and electrophilic functional groups.[1]

6.2 Preparation of 1,1-difluorocyclobutane bioisosteres

Using precursor 6 of an anti-tuberculosis candidate compound as the substrate, the authors first carried out photocatalytic ring opening and then completed O→C cyclization with p-toluenesulfonylmethyl isocyanide, TosMIC, and NaH. Subsequent acidic hydrolysis afforded cyclobutanone 74 in 41% overall yield. Deoxyfluorination with diethylaminosulfur trifluoride, DAST, then provided 1,1-difluorocyclobutane 7 in 68% yield.

This route preserves the spirocyclic carbon framework and arylamine side chain of the original molecule, while converting the oxetane into a 1,1-difluorocyclobutane at a late stage. It therefore reduces the need to reconstruct the entire synthetic route specifically for the fluorinated cyclobutane.[1]

6.3 Shorter synthetic routes through later-stage structural divergence

The paper also completed streamlined syntheses of two advanced pharmaceutical intermediates:

For precursor 79 of the respiratory syncytial virus, RSV, inhibitor, the four-step route had an overall yield of approximately 31%, whereas the eight-step route cited in the paper had an overall yield below 18%. Precursor 2 of the cyclin-dependent kinase 2, CDK2, inhibitor was prepared from an oxetane precursor through reductive amination, O→S replacement, and subsequent sulfur oxidation and other operations.[1]

Two different ways of counting the steps should be distinguished:

1. A single skeletal-editing sequence: the paper describes the overall transformation from oxetane ring opening to ring reconstruction as a single operation. The model O→N reaction can be completed sequentially in the same reaction system, whereas some O→S, O→C, and two-atom insertion procedures include filtration, solvent removal, or transfer operations. Therefore, not all of the reactions should be understood as strictly one-pot processes;

2. A complete synthetic route: this also includes preparation of the oxetane precursor, protecting-group manipulations, oxidation, and subsequent derivatization.

The study did not compress every complete eight- to twelve-step synthesis into a single step. Rather, it replaced multiple conventional functional-group transformations and de novo ring-construction steps with a key skeletal-editing sequence, shortening representative complete routes to four steps.

7. Chemical Basis for Shortening the Synthetic Route

7.1 A common precursor reduces repetitive synthesis

Conventional methods often require separate preparation of nitrogen-, sulfur-, and carbon-containing four-membered-ring precursors:

Nitrogen-containing precursor ──multiple steps──→ Azetidine

Sulfur-containing precursor ──multiple steps──→ Thietane

Carbon-containing precursor ──multiple steps──→ Cyclobutane

The oxygen-atom replacement strategy instead allows an oxetane to be prepared first, followed by the selection of different nucleophiles at a later stage:

In this way, the synthetic steps preceding the oxetane can be shared among multiple target structures. Route shortening mainly arises from the following changes:

1. A separate ring-forming precursor no longer needs to be designed for each type of endocyclic atom;

2. Fewer oxidation-state adjustments and protecting-group operations are required;

3. Selection of the endocyclic atom is postponed until after construction of the main molecular framework;

4. A series of matched analogues can be prepared from the same advanced intermediate.

7.2 Structural changes are better suited to establishing matched-analogue relationships

If two candidate molecules are otherwise identical and differ only in the endocyclic atom, they form a more clearly defined matched-analogue pair. Changes in biological activity or physicochemical properties can therefore be associated more directly with O→N, O→S, or O→C replacement.

8. Applicable Conditions and Outstanding Challenges

8.1 Functional-group compatibility is not unlimited

The method tolerates various ethers, esters, amides, heteroarenes, and aryl halides. However, the authors reported that an oxetane bearing a free aliphatic alcohol did not provide the expected azetidine, owing to bromination of the hydroxy group under the reaction conditions.[1]

The presence of groups that are readily brominated or that react with strongly electrophilic species must therefore be evaluated on a substrate-specific basis.

8.2 O→C transformations depend on stabilized carbon nucleophiles

The O→C reactions reported in the paper mainly use α-cyano compounds, malonate derivatives, and related nucleophiles capable of forming stabilized carbanions. Whether ordinary alkyl carbon nucleophiles can achieve comparable results was not broadly investigated in this study.

Compared with the O→N and O→S transformations, the range of nucleophile classes and product structures demonstrated for O→C replacement is relatively concentrated.

8.3 Scale-up studies must consider photochemistry and bromine-containing reagents

The reaction uses a ruthenium photocatalyst, CBr₄, and blue-light irradiation. From a process-development perspective, the following issues require further evaluation:

1. Light penetration and mass transfer during scale-up;

2. Safe handling of CBr₄ and bromine-containing byproducts;

3. Control of residual ruthenium catalyst;

4. Treatment of brominated waste;

5. Whether continuous-flow photoreactors can improve scale-up efficiency.

A reduction in the number of synthetic steps may lower some material and operational requirements, but it does not directly mean that the overall environmental burden has been reduced. A complete assessment must also consider reagent consumption, solvents, reaction concentration, separation processes, and waste composition.

In addition, the representative reactions described in the Supplementary Information were mainly conducted on a 0.05–0.2 mmol scale and generally used anhydrous solvents under a nitrogen atmosphere. The applicability of the method to gram-scale synthesis, its batch-to-batch consistency, and its performance at higher concentrations therefore require further validation.

9. Representative Chemicals Related to Photocatalytic Atom Replacement in Oxetanes

Table 1. Photocatalytic Ring Opening and Reaction System

Table 2. Nitrogen Nucleophiles for Oxygen-to-Nitrogen Skeletal Reconstruction

Table 3. Reagents for Oxygen-to-Sulfur and Oxygen-to-Carbon Transformations, Two-Atom Insertion, and Subsequent Derivatization

Table 4. Examples of Four-Membered-Ring Building Blocks Related to Oxetane Skeletal Editing

Note: The products listed above are representative Aladdin products related to scientific research and formulation studies. The table includes representative building blocks relevant to this research area, and not all of them were directly used as reagents in the experiments reported in the original paper. Additional information on product specifications, grades, and certificates of analysis, COAs, can be retrieved from the Aladdin website using the product name, CAS number, or product number.

References

[1] Zhang Y.-Q., Li S.-H., Zhang X., Koh M. J. Photocatalytic oxygen-atom transmutation of oxetanes. Nature, 2025, 647: 906–912. DOI: 10.1038/s41586-025-09723-3.

[2] Supplementary Information for: Zhang Y.-Q., Li S.-H., Zhang X., Koh M. J. Photocatalytic oxygen-atom transmutation of oxetanes. Nature, 2025, 647: 906–912. DOI: 10.1038/s41586-025-09723-3.

[3] Jurczyk J., Woo J., Kim S. F., Dherange B. D., Sarpong R., Levin M. D. Single-atom logic for heterocycle editing. Nature Synthesis, 2022, 1: 352–364. DOI: 10.1038/s44160-022-00052-1.

[4] Rojas J. J., Bull J. A. Oxetanes in drug discovery campaigns. Journal of Medicinal Chemistry, 2023, 66: 12697–12709. DOI: 10.1021/acs.jmedchem.3c01101.

[5] Dai C., Narayanam J. M. R., Stephenson C. R. J. Visible-light-mediated conversion of alcohols to halides. Nature Chemistry, 2011, 3: 140–145. DOI: 10.1038/nchem.949.

[6] Malarney K. P., KC S., Schmidt V. A. Recent strategies used in the synthesis of saturated four-membered heterocycles. Organic & Biomolecular Chemistry, 2021, 19: 8425–8441. DOI: 10.1039/D1OB00988E.

For more related articles, see below:

Oxetane: Property-Window Optimization and a Building-Block Selection Guide (Tables 1–4)

Cyclic isomers--Azabicyclic molecular building blocks to aid drug design

Haloheterocycles and Cross-Coupling: A Research-Oriented Selection Framework from Substrate Identification to Bond-Forming Routes (Including Product Navigation and Tables 1–5)

Why Do We So Often Add a “Morpholine Ring” to Molecules? — Definitions, Structural Features, and a Selection Guide to Morpholine/Thiomorpholine (with Tables 1–4)