Introduction

The processes of C–H activation (functionalization) of non-activated CH-substrates are among the most important processes in modern organic chemistry1. They are the main subjects of study for many scientific groups around the world. Saturated hydrocarbons belong to the most important substrates for directed functionalization due to their availability and low cost2. Transition metal catalysis1,3, radical approaches4, photoredox catalysis5, and carbene insertion processes6,7 are among the most common techniques for C–H functionalization. There are many thousands of examples where these approaches were used.

However, the C–H activation (functionalization) processes that occur via vinyl carbocations received much less attention. They are very rare and poorly studied8,9. This is partially a consequence of difficult control of selectively in such processes for unstabilized vinyl carbocations in the presence of a large number of competing process pathways, often inherent in carbocation reactions in general10. Undoubtedly, further development of these processes is an important task in modern organic chemistry.

A little more than two dozen examples of these processes are probably known at the moment (Fig. 1A, a)10,11,12,13,14,15. The primary mechanism involves the intramolecular concerted insertion of vinyl carbocations into unactivated C–H bonds occurring by a non-classical mechanism and via ambimodal transition states9. Basically, these processes occurring after the stage of vinyl carbocation insertion can be reduced to two types that occur with further proton elimination11,12,14 or with addition of a hydride anion (for example, in the presence of R3SiH)9,13. The mechanisms of these processes are similar to the well-known insertion of carbenes into C–H bonds6. Intermolecular insertion of a vinyl carbocation into aliphatic hydrocarbons can also be carried out9.

Fig. 1: C(sp3)–H functionalization via vinyl carbocations and the general concept of the work.
figure 1

A Known reactivity modes and the double CH2 functionalization approach developed. B The developed example of deep double C–H functionalization of inert CH2-groups driven by the cooperative action of Ga(III) salts & GaHal4 anions. C Study of the process conditions and features.

The utilization of unstabilized vinyl carbocations and other carbocationic transition states often requires the use of bulky and expensive WCAs (weakly coordinating anions), such as [HCB11Cl11] or [B(C6F5)4], for proper control of the reactivity8,9,11. Vinyl carbocations themselves are generated by various methods, in particular, from vinyl triflates9,11, by addition of carbocations to acetylenes12,13, by Au(I)+ catalysis from substrates with a triple C–C bond (carbene like reactions)16, by decomposition of diazo compounds with Lewis acids14, and by some other techniques.

In this work, we suggested the following conceptually new family of processes. They involve the double C–H functionalization of aliphatic inert CH2-groups based on concerted C–H insertion of vinyl carbocations at the 1st step similar to example reported in the literature (Fig. 1A, b). We failed to find any precedents in the literature to date. The main feature is introducing an additional C–H functionalization step at the 2nd C–H bond in the same CH2-group by nucleophile addition, which allows one to create an all-substituted quaternary carbon stereocenter based on CH2. The nucleophile addition was initially implemented as intramolecular addition of a C-nucleophile, but it is not limited to this variant. Our methodology also relies on the cooperative dual action of gallium (Ga(III)/GaHal4). Ga(III) cations are used to activate the reaction cascade. Simultaneously, we used simpler tetrahedral GaHal4- anions instead of classical WCAs for stabilization of carbocationic intermediates. It should be noted that no alternative to the use of Ga(III) has been found to date.

Results

Development of the primary double CH2-functionalization process

To generate vinyl carbocations, we used the authors’ Ga(III)-initiated process of addition of activated cyclopropanes to acetylenes with opening of a three-membered ring17. In this case, incorporation of a large fragment of the initial cyclopropane into the product structure is achieved, which significantly increases the molecular complexity of the process with incorporation of functional groups and creation of additional rings.

The first developed example of this class of processes involves the creation of a norbornane ring system based on a cascade reaction of activated cyclopropanes (so-called donor-acceptor cyclopropanes)18,19,20,21,22 with internal acetylenes containing an alkyl chain at least C4 long (Fig. 1B). As a result, a γ-CH2-group is selectively functionalized at both C–H bonds to create a nodal quaternary carbon stereocenter. In this case, activated cyclopropanes act as sources for generation of 1,2-zwitterionic intermediates of types 14 or 15 (Fig. 1C)23,24 containing the required nucleophilic center.

This process requires the conditions to be optimized carefully, since activated cyclopropanes, which are multi-functional substrates, exhibit high and varied activity towards acetylenes. A few different types of reactions under the action of GaHal3 are known17,25, and a few more in the presence of other Lewis acids26,27. The model reaction of cyclopropane 7a with dec-5-yne 8a under partially optimized conditions gives four products 9a–12 (Fig. 1C), but tetralin 13 that is most typical of these reactions25 is not formed. Further directed optimization can increase the yield of the target norbornane 9a from 41–53% to 65–70%. The need to use 1.5 equiv. GaCl3 to generate the required amount of GaCl4 anions (which can also be created by addition of Bu4N+ GaCl4) by a dismutation reaction (14 to 15)28 should be noted among the main features. It is also interesting to note the formation of a minor mono-C–H functionalization product (10a). It is a previously unknown path that will be described below.

To date, gallium(III) is the only and unique metal that allows this process to be carried out. We certainly made a wide screening of various metal halides (for details see Supplementary Information), but they did not show any results. Also, the proper source of gallium(III) is very important. Only GaCl3 gives good results. GaBr3 is already much worse, GaI3 gives only a little product, while GaF3 and Ga(OTf)3 do not give at all any product. The reason, however, lies not only in the exactingness of the stages of C–H functionalization, but also in the complexity of the transformations of the cyclopropane substrate itself, which is also very demanding on the metal catalyst17,23,24,25. Another very important feature of gallium metal is a high tendency to generate tetrahedral GaCl4 anions in high concentration during reaction28, which are very stable and have a suitable balance between vinyl cation stabilization (acting as WCA) and their good reactivity. Other metal halides are either not capable of this, or are capable of it rather poorly, such as AlCl3 and FeCl328. Knowing the reasons, one can try to simulate the action of GaCl4 anions using various additives, starting with simple Bu4N+GaCl4, and ending with more complex and bulky WCA anions. One of the best to date is GaArF4 anion (Ar = 3,5-(CF3)2C6H3), which makes it possible to obtain norbornane 9a up to 83% yield. However, it is irrational to use in practice, since it is very difficult to synthesize and very expensive. And it is a vision for the future. At the same time, its much more affordable boron analogue BArF4 does not work well yet for some unknown reasons.

If we look at the scope for various substituents, we can see that the products of annulation to the aromatic ring (types 10 and 12) are formed rather rarely and in small amounts. Acyclic vinyl chloride 11 is the most “harmful” side product17. However, this process can be significantly suppressed by the addition of NaSbF6 (due to the withdrawal of “halide” anions from the reaction sphere).

A further estimation showed that this process is of rather general scope (Fig. 2). Various activated cyclopropanes 7a–m, both with donor and acceptor aryl substituents (9a–j), with alkyl substituents (for example, 9k), as well as with varying acceptor activating groups (9l,m), successfully react with dec-5-yne 8a (Fig. 2A). The range of acetylenes used (8a–m) is also quite broad. In addition to long n-alkyl chains with a γ-CH2 moiety in the range of C4–C15 (9h, 16b,c,i), branched moieties can be used, including those with additional substituents (16k–m). Moreover, the groups at the other end of the acetylene moiety can be varied (16e–j) by using symmetrical or unsymmetrical acetylenes (Fig. 2B). All these reactions generally give a polysubstituted norbornane frame (9a–m, 16a–m). The product yields are usually decent, although they can be low for various reasons in some cases. For example, norbornanes 16g,n are obtained in low yields from acetylenic substrates with an isopropyl group at the triple bond, or product 16d is obtained from symmetrical oct-4-yne that contains a less reactive methyl group at the γ-position instead of a CH2 moiety.

Fig. 2: Scope of the reactions.
figure 2

See Supplementary Information for detailed conditions. A Varying activated cyclopropanes. B Varying alkyl acetylenes. C Application of an activated cyclobutane. D Varying methylidene malonates instead of cyclopropanes. E Diastereoselectivity control.

It is important to note that, to expand the scope of the reactions in question, activated cyclobutanes 1729,30 (Fig. 2C), as well as methylidene malonates 1928 can be successfully used instead of activated cyclopropanes for generating vinyl cations by addition of zwitter-ionic intermediates to a triple C-C bond. In the latter case, the scope studied also shows a good variability of substituents (20a–h) (Fig. 2D). Apparently, the examples presented here do not limit all the capabilities of this process, and in principle, it can be expanded to other types of substrates.

The issue of diastereoselectivity and its control is very interesting (Fig. 2E). For example, in the formation of the C(3) and C(7) centers of the norbornane frame, two out of four possible diastereomers with respect to the C(3) atom are formed. The diastereomer ratio can vary from ~1.2/1 to 20/1, depending on the substituents. In this case, poor diastereoselectivity values can be improved significantly by incorporating sterically bulky groups into the acceptor moiety of cyclopropane, for example, using neo-pentyl instead of methyl in ester groups (9a vs. 9m‒o) (Fig. 2E). In some cases, the third diastereomer with respect to C(7) is formed in small amounts, especially in the presence of low-volume substituents at C(7) C(1) and C(2) i.e., norbornanes 16e and 9l. Additional stereocenters can also be incorporated in a controlled manner into the norbornane frame, for example, at C(2), C(6) or C(7) (9l, 16k–q).

The configuration of the main diastereomer at C(3) depends on the substituent type. In fact, the endo-isomer9 is predominantly formed in the case of substituents with an aromatic moiety (CH2Ar, etc.), whereas the exo-isomer20 is formed in the case of alkyl groups (Fig. 2A, D). Moreover, stereodivergent synthesis that selectively gives both diastereomers can be performed with some substituents at controlled process temperature. For example, in the case of the (2,6-Cl2C6H3)CH2 moiety at C(3) and alkyl chains no shorter than C2 at C(1) and C(7) the endo-isomer16 is formed at 0 °C (kinetic control) and the exo-isomer21 is formed at 40 °C (thermodynamic control) with a selectivity no less than 6/1 in each case (Fig. 3A) (see further mechanistic part for explanation).

Fig. 3: Stereodivergent synthesis and further development of the process.
figure 3

A General diastereoselectivity regularities and stereodivergent synthesis. B Initial approach to asymmetric transformation. C Catalytic version of the developed process.

Also, a preliminary approach to asymmetric synthesis was demonstrated. It is a big challenge, because no asymmetric approaches are currently known for such gallium-mediated reactions, and the first asymmetric approach for the reaction of С–H-insertion of vinyl cations was developed only in 2022 year31. Here, we used a simple scheme, consisting in the introduction of two optically active (–)-menthyl substituents into the carboester groups, which are natural and cheap raw materials (Fig. 3B). At the initial stage, this makes it possible to achieve “e.r.” 8/1 for norbornane, taking into account the high steric loading of the reaction centers, without the use of additional chiral ligands, which is a very good result.

And finally, we tried to develop a catalytic version of the current process. To date, there are several examples in the literature where the gallium salts or gallium itself have been used in catalytic amounts for C-C bond formation32,33,34,35. We have focused on the use of super-electrophilic gallium catalyst systems and highly electrophilic gallium salts due to their higher activity and stability to hydrolysis compared to gallium halides. The use of the double catalytic system Ga(SbF6)3 + Na[Ga(C6F5)4] (10 + 15 mol.%) allowed us to successfully introduce cyclopropane 7a and dec-5-yne 8a into the reaction with the formation of norbornane 9a with moderate yield (40%) and good diastereoselectivity (dr = 4/1). (Fig. 3C) The first component of this system catalyzes the main cascade of processes, and the anion of the second component is needed for the C-H insertion stage. It is worth noting that perfluorophenyl metal salts Ga(C6F5)3, Al(C6F5)3, B(C6F5)3 by themselves did not show sufficient catalytic activity.

Investigation of the mechanism

The occurring processes involve a complex cascade of several carbocationic reactions. At the same time, detailed consideration and study of mechanistic aspects are important tasks due to the nontriviality of the processes. The cascade of reactions begins with the activation and opening of the cyclopropane ring and generation of a 1,2-zwitterionic gallium complex of type 14 (Fig. 4B)36. Further, its dismutation occurs to give an ionic complex of type 15 under the action of excess GaHal3, which results in the required GaHal4 anions28. Ionic complex 15 has a pronounced 1,2-zwitterionic nature with strong polarization of the double bond28. It adds very quickly as a carbocation to the acetylene triple bond.

Fig. 4: Investigation of the Mechanism.
figure 4

A Mechanistic experiments. B The final mechanistic hypothesis for the observed cascade of reactions.

The 1st C–H activation (functionalization) step occurs after the formation of a vinyl carbocation23 by concerted C–H insertion into the γ-CH2 group of the alkyl chain, like in the examples described previously10,11,12,13,14, with γ-selectivity via the most stable ambimodal transition state 24 (Fig. 4B). The intermolecular C–H insertion (like in ref. 9) is also possible in these processes (see below). GaCl4 or GaBr4 anions perfectly stabilize the concerted ambimodal transition state. However, they are not classical WCAs and can be coordinated with carbocations. Moreover, additional solvation with an organochlorine solvent plays an important role in additional stabilization, as demonstrated by the remarkable effect when a solvent is replaced with a similar one.

The concerted C–H insertion mechanism is confirmed by dedicated mechanistic experiments on the distribution of deuterium isotopic labels (16k–D) (Fig. 4A, a) and by the reaction of an acetylenic substrate containing two methyl groups at the β-position of the alkyl chain (16m) (Fig. 4A, b). These experiments unambiguously rule out the possibility of sequential migration of the carbocation along the alkyl chain to the γ-position. Moreover, the participation of substrates with an n-propyl substituent8d in this process and activation of the terminal Me-group in it rules out the possibility of a 1,5-hydride shift37 that is impossible if an unstable primary carbocation is formed37 (Fig. 2B, product 16d).

The 2nd C–H functionalization step raises more questions. Apparently, this process occurs immediately after the 1st concerted C–H insertion via a nonclassical carbocation transition state of type 25 (Fig. 4B). At the same time, the continuation of the process immediately after the 1st step (unlike all the known examples10,11,12,13,14, Fig. 1A) seems to be possible due to enhanced stabilization of the nonclassical carbocation transition state by GaCl4/GaBr4 anions in combination with their zwitter-ionic nature. The ambimodal transition state 25 is a superposition of two classical carbocations on each of two carbon atoms related via a 1,2-hydride shift. However, classical carbocations do not fully explain the observed reactivity pattern, which is too limited and combined with unusually high selectivity. Even for the 8/16k-D substrate with a deuterium label and an additional Me group, almost neither additional redistribution of the isotopic label nor additional carbocationic rearrangements that are so characteristic of such carbocations are observed, which indicates a high consistency of the process (Fig. 4A, a). Moreover, this hypothesis about the 2nd C–H functionalization step is confirmed by the formation of considerable amounts of cyclopentane 22 in the case of ethylhexylacetylene 8i under more drastic conditions (Fig. 4A, c), though as one diastereomer and one regio-isomer with respect to the double bond. The occurrence of free 1,2-hydride shifts in isomeric classical carbocations should eliminate the extremely high diastereoselectivity and result in isomeric cyclopentenes; it fact, it is observed upon double bond migration in cyclopentene 22 in the presence of excess GaCl3 to give diastereomeric mixtures.

The description of all the three observed pathways of the process with C(sp3)–H activation perfectly fits the assumed transition state 25 (Fig. 4B). However, a thorough examination of reactions with various substituents did not reveal any additional pathways associated with the C–H activation. As a result, three possibilities are realized1: an intramolecular nucleophilic attack to create a norbornane frame9,16,20 as the main pathway2; H+ abstraction observed only in a couple of cases22; and3 a related concurrent intramolecular electrophilic attack on the aromatic ring (orange arrow in transition state 25) that interrupts the 2nd C–H functionalization of CH227.

Considering the stereochemistry of the process is a challenging task; the process consists of a large number of sequential and also reversible elementary stages, many of which, in an overall complex relationship, affect to the stereochemical outcome of the reaction. The structure of the norbornane product contains three main stereocenters (C(3), C(4), and C(7); C(1) is not independent), which in theory could lead to four different diastereomers. In the experiment, either two diastereomers are formed, or predominantly one diastereomer. The main isomerism is observed at the C(3) center, forming either an endo-isomer (most often) and/or an exo-isomer, with observed a strong dependence on the substituent in the parent cyclopropane / methylidenemalonate (see also discussion above). Figure 5 shows the stereochemical model of the main reaction. The main endo/exo isomerism at C(3) is set at the first stages of the C–H functionalization cascade in the transition states 24a/24b, which are apparently quite close in energy and therefore such a strong dependence on substituents is observed, an additional contribution is also made by weak intramolecular interactions with substituents. In the further intermediates formed, starting from 25a/25b, the relative configuration of the stereocenters C(3)/C(4) is already fixed and cannot simply change, and is maintained until to the final norbornane product. At the same time, the formation of the C–C bond at the stage 24 → 25 is irreversible, and the entire carbocation cascade of C–H functionalization should proceed quickly, and there is no time left to reverse the configuration, since it requires additional steps or reversible transition states, that is why such good selectivity is observed in general. The possibility of carrying out stereodivergent syntheses in a number of cases with the selective production of either endo- or exo-isomer is realized under more stringent conditions and at higher temperatures, that is necessary to speed up the additional steps required to reverse the configuration, since the direct transformation 25a ↔ 25b is impossible. And the change of the configuration between the C(3)/C(4) stereocenters occurs through additional reversible stages of migration of the carbocation center (hydride-anion shift) between C(1)/C(7)/C(4) atoms via intermediate 25c. Thus, in this case it is possible to control the stereoselectivity of the process through kinetic and thermodynamic control of the reaction depending on the temperature and time of the process. For the third stereocenter C(7) with an alkyl substituent, a high selectivity is observed, and here the configuration is set at the last stages of the carbocation cascade (intermediates 25a→25aa→25ab in the Figure) and appears to be regulated mainly by steric factors (Fig. 5).

Fig. 5
figure 5

The origins of the diastereoselectivety of the obtained norbornanes.

The 3rd general pathway of the mechanism is the interrupted double CH2-functionalization process with functionalization of only one C(sp3)–H bond in CH2-group. In this process, the 2nd C–H functionalization of CH2 in interrupted by a concurrent intramolecular electrophilic attack on the aromatic ring, which gives cyclopenta[a]tetralin derivatives 27 with good diastereoselectivity. It can be partially optimized for some CH2-substrates (27a–f) (Fig. 6A). In this case, the use of deficiency of GaCl3 to 1.0–1.2 equiv. (compare with common 1.5 equiv.) is optimal; also, the use of GaBr3 is more preferable. Moreover, in the case of GaBr3 it is a dominant pathway for symmetrical oct-4-yne, (8d) in which the mono C(sp3)–H activation of the terminal methyl group in the alkyl chain occurs successfully (27g). This allows one to assume that this process can be easily implemented for many substrates with a γ-CHR1R2 moiety. However, in this work, we were focused only on the γ-CH2 involvement that is more complicated.

Fig. 6: Application and further development based on the discovered double CH2-functionalization process.
figure 6

A The interrupted process with mono-CH2-activation and annulation on aromatic ring. B Norbornane and cyclopenta[a]tetralin frames in nature. C Some further synthetic transformations of norbornane products. D Representation of 3D 1H,13C NMR spectra, widely used for the structural and stereochemical analysis of poly-alkyl norbornane products. EG Preliminary experiments for expanding the process scope.

Application and further development based on the discovered process

The entire family of processes described above provides access to a number of interesting frames, of which norbornane and cyclopenta[a]tetralin have been mainly studied. These two types of frames commonly occur in nature, including pharmaceutically important compounds (Fig. 6B). For example, the norbornane bicycle is contained in many terpenoids38, while cyclopenta[a]tetralin39 is also found in analogs of steroid molecules40. All this, apart from organic chemistry, emphasizes the significance of the discovered reactions. For example, simple further modifications can significantly expand the diversity of functional groups in the products of interest (Fig. 6C). Possibility of steroid mimetic compound synthesis, along with preliminary data obtained by molecular modeling, synthetic schemes leading to norbornane derivatives of the 9/16/20 type are promising in producing new androgen receptor inhibitors. Nuclear steroid hormone receptors, such as androgen and estrogen receptors, are expressed in different tissues of human body, and directly affect proliferatory activity of cells in these tissues41. They are, therefore, present important targets for steroidal and mimetic class drugs with wide range of possible therapeutic effects, particularly anti-cancer properties. In particular, androgen receptor antagonists and degrading agents are currently actively being used for treatment of prostate cancer42. Selective androgen receptor modulators (SARMs) are another therapeutically important class of compounds43, useful in treatment of certain kinds of muscle dystrophy. Biological membranes are of crucial importance for signal transduction in live cells, and they affect virtually all cellular signaling pathways, either via incorporated receptor proteins or by modulation of lipid bilayer physico-chemical properties. Synthesis of norbornanes 20 with multiple long alkyl tails opens new ways in design of new potential membrane probes. For example, compound 30 on Fig. 6C contains nitroxide fragment as a spin label, making it suitable for detection by EPR and NMR spectroscopy. Spin labels are of broad use for investigation of cell membrane mediated processes in both natural and artificial (model) lipid bilayers44.

An interesting and important point that deserves mentioning is the protocol for determining the structure of polyalkyl norbornane derivatives 9/16/20 and their stereochemistry (see Supplementary Information), because this is a nontrivial problem for solving by traditional methods due to the very strong overlapping of numerous complex multiplets in the alkyl region of the 1H NMR spectrum. The presence of a branched alkane frame prevents their normal crystallization and makes it impossible to obtain any XRD data of decent quality suitable for publication, save for preliminary data. This challenge was solved elegantly by applying a three-dimensional (3D) 1H–1H–13C NMR technique45 primarily embodied as 3D TOCSY-HSQC and NOESY-HSQC spectra (Fig. 6D), which allows one to clear up the overlapping in the spectra in the 3rd dimension, in combination with 2D 13C–13C INADEQUATE spectra.

Discussion

We have designed a new family of processes that allow one to perform a double C(sp3)–H functionalization (activation) of inert aliphatic CH2 groups in an alkyl chain to form a quaternary carbon stereocenter. Our methodology is based on concerted C–H insertion of vinyl carbocations into C(sp3)–H bonds, with control by cooperative action of Ga(III) salts & GaHal4 anions. The main synthetic implementation was made as a cascade of reactions of alkyl acetylenes with activated cyclopropanes/alkenes/cyclobutanes to give the norbornane frame with extensive scopes, and a related mono C(sp3)–H functionalization process to give cyclopenta[a]tetralin derivatives. A feature of the processes that we developed is a high resulting molecular complexity. So, they present not just another C–H activation of aliphatic C(sp3)–H bonds but also combine the simultaneous deep polyfunctionalization of simple C–H substrates in one stage with formation of several new cycles and stereocenters and incorporation of functional groups.

And though we demonstrate the capabilities of the double CH2 functionalization for a couple of rather narrow types of processes, they certainly do not limit the possibilities of using this concept. What’s next? How much can its application be expanded? To illustrate this, we made some conceptual expansions in a preliminary version, namely, we involved an external nucleophile and an external non-activated CH2-substrate in the process (Fig. 6E, F), as well as made a decyne-5 cyclization without activated cyclopropane (Fig. 6G). However, although the conditions are still quite specific and the yields are not high, this is a key next step for the further use of this approach. Moreover, it allows one to bypass the main limitation of intramolecular implementation based on the 1st generation of the processes.

Methods

For general experimental, instrumental methods, detailed synthetic procedures and full compound characterization, see Supplementary information.

Materials

All the reagents used in this work were purchased from commercial sources and were used as received. The synthesis of commercially unavailable starting alkynes and donor-acceptor cyclopropanes was performed according to literature protocols with slight modifications (see Supplementary information).

General procedure for GaCl3-mediated C–H activation of alkynes with donor–acceptor cyclopropanes to form substituted norbornanes

All operations were performed under dry argon atmosphere. A solution of ACDC 7a‒m (1 eq, 0.6 mmol) in dry CH2Cl2 (6 mL) was cooled or heated to a target temperature; solid GaCl3 (1.5 eq., 0.9 mmol) was added in one portion and the reaction mixture was stirred at the same temperature during 10–75 min until completion the generation of 1,2-zwitterion. (Detailed conditions are presented in Supplementary Table 1). Then the reaction mixture was cooled or heated to a target temperature of the “Stage 2” (indicated in Supplementary Table 1) and a solution of alkyne 8a‒m (3 eq., 1.8 mmol) in dry CH2Cl2 (2 mL) was added. The reaction mixture was stirred at the conditions indicated in Supplementary Table 1 until reaction completion. After that, a 10 ml of aqueous solution of HCl (10%) was added and the reaction mixture was extracted with CH2Cl2 (3 × 15 mL). The organic layer was dried over MgSO4 and the solvent was removed in vacuo. The residue was purified by column chromatography on silica gel (hexane to benzene–EtOAc, 50:1) to afford title compounds 9a‒m and 16a‒o as thick colorless oils. The resulting compounds can be additionally purified on a Silufol chromatographic plate (20 × 20 cm) eluting with hexane–acetone, 10:1 to afford the pure products, if it is necessary.

General procedure and spectroscopic data for reaction of methylidenemalonates with alkynes

Anhydrous GaCl3 (1.5 equiv.) was added to a solution of substituted methylidenemalonate (1 equiv.) and corresponding acetylene (3 equiv.) in 5–10 mL CH2Cl2 under dry argon atmosphere, and reaction mixture was stirred 4 h at 40 °C. Then the mixture was treated with an aqueous solution of HCl (10 mL, 10%), the organic layer was separated and the aqueous layer was extracted with CH2Cl2 (3 × 10 mL). The combined organic extracts were dried with anhydrous Na2SO4 and evaporated in vacuo. The residue was purified by column chromatography over silica gel (eluent: petroleum ether – EtOAc, 30:1) and the corresponding products, substituted norbornanes 20, were obtained.

General procedure and spectroscopic data GaCl3-mediated C-H activation of alkynes with donor-acceptor cyclopropanes to form substituted hexahydrocyclopenta[a]naphthalenes 27

All operations were performed under dry argon atmosphere. A solution of ACDC 7a,b,i,j,o (1 eq, 0.6 mmol) in dry CH2Cl2 (6 mL) was cooled or heated to a target temperature; solid GaCl3 or GaBr3 (1.5 eq., 0.9 mmol) was added in one portion and the reaction mixture was stirred at the same temperature during 10–15 min until completion the generation of 1,2-zwitterion I. (Detailed conditions are presented in Supplementary Table 3). Then the reaction mixture was cooled or heated to a target temperature of the “Stage 2” (indicated in Supplementary Table 3) and a solution of alkyne 8a,b,d (3–5 eq., 1.8–3 mmol) in dry CH2Cl2 (2 mL) was added. The reaction mixture was stirred at the conditions indicated in Supplementary Table 3 until reaction completion.

After that, a 10 ml of aqueous solution of HCl (10%) was added and the reaction mixture was extracted with CH2Cl2 (3 × 15 mL). The organic layer was dried over MgSO4 and the solvent was removed in vacuo. The residue was purified by column chromatography on silica gel (hexane to benzene–EtOAc, 50:1) to afford title compounds 27 as thick colorless oils. The resulting compounds can be additionally purified on a Silufol chromatographic plate (20 × 20 cm) eluting with hexane–acetone, 10:1 to afford the pure products, if it is necessary.

General procedure and spectroscopic data for synthesis of 3-benzyl-7-butyl-1-methyl-2′H-spiro[bicyclo[2.2.1]heptane-2,5′-pyrimidine]−2′,4′,6′(1′H,3′H)-trione 28

All operations were performed under dry argon atmosphere. An urea (6 eq, 1.3 mmol) and potassium tert-butoxide (2.2 eq, 0.48 mmol) were added to a solution of substituted norbornane 9a (1 eq, 0.22 mmol) in dry DMSO (2 mL) at room temperature. The reaction mixture was stirred at this temperature 1 h and diluted CH2Cl2 (15 ml). The resulting mixture was washed with 10 ml of aqueous solution of HCl (10%), organic layer was separated and an aqueous layer was extracted with CH2Cl2 (3 × 10 mL). The organic layer was dried over MgSO4 and the solvent was removed in vacuo. The residue was purified by column chromatography on silica gel (hexane to benzene–EtOAc, 50:1) to afford title compound 28 as thick light yellow oil in 58 % yield (47 mg) as a single diastereomer. The resulting compound can be additionally purified on a Silufol chromatographic plate (20 × 20 cm) eluting with hexane–acetone, 10:1 to afford the pure products, if it is necessary. IR (CHCl3) 3389 br (NH), 3086, 3054, 3041, 3033, 3009, 2961, 2932, 2874, 2861, 1754, 1730 and 1700 br (C = O), 1602, 1496, 1454, 1425, 1385, 1346, 1312, 1245, 1188, 1054, 1030, 909, 816, 740 cm–1. HRMS (ESI) calcd for C22H28N2O3: M + H, 369.2173; M + Na, 391.1992. Found: m/z 369.2174, 391.1994. 1H NMR (CDCl3, 300.1 MHz): δ 0.92 (t, 3H, Me(4′), 3J = 5.9 Hz), 1.04 (s, 3H, Me), 1.11–1.79 (m, 9H, H(5), H(6), H(1′), H(2′) and H(3′)), 1.96–2.10 (m, 1H, H(5)), 2.14 (br. t, 1H, H(4), 3J = 3.6 Hz), 2.35 (br. t, 1H, H(7), 3J = 6.1 Hz), 2.93–3.09 (m, 2H, H(5′)), 3.31–3.45 (m, 1H, H(3)), 6.91–7.47 (m, 5H, o-H, m-H and p-H), 8.16 and 8.21 (both s, 2H, 2 NH) ppm. 13C NMR (CDCl3, 75.5 MHz): δ 14.0 (Me(4′)), 15.5 (Me), 20.9 (C(5)), 23.1 (C(3′)), 24.7 (C(2′)), 31.2 (C(1′) and C(6)), 33.1 (C(5′)), 42.9 (C(4)), 48.4 (C(3)), 51.5 (C(7)), 61.0 (C(1)), 62.5 (C(2)), 126.1 (p-CH), 128.3 (2 m-CH), 129.2 (2 o-CH), 148.8, 168.4 and 172.5 (3 CO) ppm. 15N NMR (CDCl3, 30.4 MHz): δ 148.1 (s, 1 N), 151.3 (s, 1 N) ppm.