*2.2. Stepwise Approach 1 (Knoevenagel/Allylic Oxidation/Wittig)*

The disappointing results of the direct Knoevenagel strategy prompted us to apply a stepwise approach (Scheme 3). Using an acetophenone as a "partner" of Ti(O*<sup>i</sup>* Pr)4/pyridinemediated Knoevenagel condensation, 3-methyleneoxindole **5** was easily obtained from oxindole **1** in 93% yield with good *Z*-stereoselectivity (*Z*:*E* = 5:1). The preference for the *Z*-isomer could be explained by a chelation-controlled transition state [14]. The geometry of each isomer was confirmed by comparing 1H NMR data for **(***E***)-5a** [19] and the chemical shift of H4 [6.84 ppm (*Z*-isomer), 6.14 ppm (*E*-isomer)]. On 1H NMR analysis of known 3-arylmethylenoxindoles, the chemical shift of H4 is upfield (generally 6.50–6.00 ppm) compared to the usual aromatic area when the aryl group attached to the 3-methylene position of oxindole is located close to H4 [6,9]. Next, the methyl group should be transformed into a proper functional group to introduce the second olefin. Under radical bromination conditions [20], allylic bromide **6** was obtained as a single geometric isomer, regardless of the geometry of **5a**. The structure of **6** was elucidated by intensive NMR studies, including HSQC, HMBC, COSY, and ROESY. In addition, the chemical shift of H4 was 6.11 ppm, which supported the (*Z*)-geometry of **6**. The Krische group reported similar scrambling of olefin geometry during radical allylic bromination [21]. Unfortunately, the Wittig reaction of the corresponding ylide derived from **6** did not afford the desired **3aa**. Given these disappointing results, we exchanged the positions of the functional groups in the Wittig reaction. Therefore, the second functionalization of the methyl group was allylic oxidation. After analyzing several oxidation conditions, we found that SeO2 oxidation [22] afforded aldehyde **7a** in 84% yield, including as a single geometric isomer from both (*Z*)- and (*E*)-**5a**. Interestingly, the olefin geometry of **7a** had the *Z*-configuration, which was confirmed by comparison with previously reported 1H NMR data for **7a** [23]. In addition, the chemical shift of H4 for **7a** also appeared at 6.26 ppm. The *Z*-stereoselectivity of allylic oxidation may have been due to coordination of the oxindole carbonyl group to selenium [24]. However, considering the high reaction temperature, the possibility of isomerization of **(***E***)-7a** to the more stable **(***Z***)-7a** during the reaction could not be excluded. Then, the Wittig reaction of aldehyde **7a** with the ylide proceeded smoothly and provided the desired **3aa** in 84% yield. The 1H and 13C NMR data of **3aa** exactly matched the results obtained in our previous study [6].

**Scheme 3.** Stepwise approach to 3-(1,3-diphenylallylidene)oxindole **3aa**.

By applying the successful stepwise approach to 3-(1,3-diphenylallylidene)oxindole, we investigated the substrate scope of aldehyde **7** (Table 1). The Knoevenagel condensation

of oxindole **1** and acetophenones with chloro, nitro, and methoxy substituents proceeded well, affording **5b**–**d** in good yield (77–95%) with *Z*-stereoselectivity (*Z*:*E* = 4–10:1) (Entries 1–3). SeO2-mediated allylic oxidation of **5b** and **5c** also proceeded smoothly to afford the corresponding **7** in 85% and 87% yield, respectively. However, the oxidation of **5d**, bearing a methoxy substituent at the aryl group, was unsuccessful and resulted in complete decomposition. Unfortunately, neither a lower reaction temperature nor other oxidation conditions overcame this decomposition problem.

**Table 1.** Substrate scope for preparation of aldehyde **7**.


<sup>1</sup> Sum of isolated yields of (*Z*)- and (*E*)-isomers, <sup>2</sup> Ratio of isolated yield of (*Z*)- and (*E*)-isomers, <sup>3</sup> Isolated yield. <sup>4</sup> Ratio in 1H NMR of mixture of two isomers, which could not be isolated.

Setting aside the problematic **5d**, we assessed the substrate scope of the final Wittig reaction of **5a**–**c** with ylides bearing various substituents on the aryl group (Table 2). Fortunately, all reactions afforded 3-(1,3-diarylallylidene)oxindoles **3,** regardless of the substituent combination, in moderate to good yield (53–95%) (Entries 1–12).

**Table 2.** Substrate scope of the Wittig reaction.


<sup>1</sup> Isolated yield.

#### *2.3. Stepwise Approach 2 (Knoevenagel/Aldol/Dehydration)*

As shown above, the allylic oxidation/Wittig reaction strategy allowed synthesis of various 3-(1,3-diarylallylidene)oxindoles **3** from Knoevenagel adducts **5**. However, decomposition of **5d** in SeO2 oxidation limited the application scope of this strategy. Therefore, we investigated another stepwise approach, which could be applied to **5d** and overcome the limitation of the first strategy. Based on the fact that the functional handle of the methyl group was located at the γ-position of the α,β-unsaturated carbonyl moiety, we assumed that aldol reaction may be feasible. In addition, several examples of similar aldol reactions were found in the literature [24–26]. According to the results of base screening, only *n*-BuLi could provide the desired aldol product, **8aa**, in good yield (Scheme 4). The olefin geometry of **8aa** was assigned as *E*, as the chemical shift of H4 appeared at 6.14 ppm. Notably, unlike bromination and allylic oxidation, the olefin geometry of **5a** significantly affected the aldol reaction rate. Under optimized conditions, (*Z*)-**5a** was rapidly converted to **8aa** in 91% yield, while the reaction of (*E*)-**5a** afforded **8aa** in 55% yield. A longer reaction time and/or elevated reaction temperature did not increase product yield. The difference in reaction rate may have been caused by the lithiated intermediate from (*E*)-**5a** assuming a stable chelated form via coordination of the oxindole carbonyl group. Dehydration of **8aa** proceeded smoothly under acidic conditions to provide **3aa** in 95% yield [27,28].

**Scheme 4.** The second stepwise approach utilizing aldol reaction/dehydration.

Next, we examined whether the second stepwise approach (aldol/dehydration) was applicable to **5d** (Table 3). The first aldol reaction of **5d** proceeded well with various benzaldehydes, giving aldol adduct **8** in moderate to good yields (Entries 1–3 and 5). With the exception of **8dc**, TFA-mediated dehydration of **8** also proceeded well to afford **3** in excellent yields (Entries 1, 2, and 5). The strong electron-withdrawing action of the nitro group in **8dc** may hamper dehydration under acidic conditions. Even under reflux conditions, the desired **3dc** was produced in only 45% yield (Entry 3). After several tests, we found that, under basic conditions (TsCl, DMAP, NEt3, CH2Cl2, room temperature, 3 h), **3dc** formed in 79% yield (Entry 4). The aldol/dehydration approach could serve as an additional option for synthesis of 3-(1,3-diarylallylidene)oxindoles **3** with the previous allylic oxidation/Wittig strategy.


**Table 3.** Substrate scope of the aldol/dehydration strategy from **(***Z***)-5d**.

<sup>1</sup> Isolated yield, <sup>2</sup> Reflux, 20 h, <sup>3</sup> TsCl, DMAP, NEt3, CH2Cl2, rt, 3 h.
