Iodine-Catalyzed Prins Cyclization of Homoallylic Alcohols and Aldehydes

The iodine-catalyzed Prins cyclization of homoallylic alcohols and aldehydes was investigated under metal-free conditions and without additives. Anhydrous conditions and inert atmosphere are not required. The reaction of 2-(3,4-dihydronaphthalen-1-yl)propan-1-ol and 21 aldehydes (aliphatic and aromatic) in CH2Cl2 in the presence of 5 mol % of iodine gave 1,4,5,6-tetrahydro-2H-benzo[f]isochromenes in 54%–86% yield. Under similar conditions, the Prins cyclization of six alcohols containing an endocyclic double bond (primary, secondary, or tertiary) led to dihydropyrans in 52%–91% yield. The acyclic homoallylic alcohols gave 4-iodo-tetrahydropyran in 29%–41% yield in the presence of 50 mol % of iodine. This type of substrate is the main limitation of the methodology. The relative configuration of the products was assigned by NMR and X-ray analysis. The mechanism and the ratio of the products are discussed, based on DFT calculations.


Scope of the Iodine-Catalyzed Prins Cyclization: Aromatic Aldehydes
The Prins cyclization of 1a and p-anisaldehyde (2b) was investigated in detail to optimize the preparation of 7a (Table 1) [31]. We found that the desired product 7a can be obtained in 75% yield using 5 mol % of I 2 (entry 4). This condition was used in the Prins cyclization of 1a and other aldehydes. The reaction of HI and HOI, which are formed in the reaction medium, gives I 2 and H 2 O, thus explaining the catalytic use of I 2 [40]. The regeneration of I 2 is possible because iodide does not act as nucleophile, i.e., is not incorporated in the product, differing from the work of Yadav and co-workers [29,30]. However, we cannot exclude the participation of HI and of HOI to promote the Prins cyclization. Once optimized conditions were found, the reactions of 1a with a broad selection of aromatic aldehydes were performed. Prins cyclization products were obtained in 60%-86% yield ( Table 2). Besides the expected isochromene 7, the isomeric alkenes 12 were also formed, typically as a minor component. The distribution between the pyrans 7 and 12 is discussed in the mechanism section below. The reaction tolerates the presence of electron donating (Me, OMe and NHAc) and electron withdrawing groups (Br and NO 2 ). It can also be performed with aldehydes bearing substituents in the ortho position (entries 4, 10 and 13), including the sterically demanding aldehyde 2g, although in this case the major product is the pyran 12g (entry 5).  In all cases, the two groups in the isochromenes 7 possess a cis relationship with respect to the pyran ring. The relative configurations were assigned by NMR analysis, including NOESY experiments. X-ray analysis of the bromo derivative 7k gave further evidence for the relative configuration of isochromenes 7. Figure 1 is an ORTEP-3 view of 7k, which was solved in the space group P2 1 2 1 2 1 . Since 7k is a chiral molecule crystallized in a chiral space group containing just one molecule in the asymmetric unit its crystal structure contains a pure enantiomer [41]. Moreover, due to the presence of the bromine atom, which has an anomalous scattering large enough to permit the refinement of the Flack parameter [42], the absolute structure of 7k was unambiguously determined in this study. Thus, the chiral atoms present the following configurations: C7(S), C9(R).

Scope of the Iodine-Catalyzed Prins Cyclization: Aliphatic Aldehydes
The iodine-catalyzed Prins cyclization of homoallylic alcohol 1a and several aliphatic aldehydes was next investigated ( Table 3). Depolymerization of paraformaldehyde occurs in situ and the Prins cyclization of the formaldehyde formed in situ with 1a gave 7a (entry 1).  The desired products were also obtained using several aliphatic aldehydes, including sterically demanding ones, such as 2q-s (entries 3-5), albeit for 2s a significant amount of 12s was formed. Treatment of 1a with the α,β-unsaturated aldehydes 2t and 2u gave the desired isochromenes 7t and 7u, respectively, in good yield (entries 6 and 7).

Scope of the Iodine-Catalyzed Prins Cyclization: Homoallylic Alcohols
After studying the reaction of several aliphatic and aromatic aldehydes, the behavior of different homoallylic alcohols was investigated using p-anisaldehyde (2b), as the carbonyl component. The reaction of endocyclic homoallylic alcohols was first studied (Table 4).  The iodine-catalyzed Prins cyclization can also be performed with endocyclic homoallylic alcohols bearing the double bond in five and seven-membered rings (entries 1 and 2). Homoallylic alcohols with different side chains can be used as substrate (entries 3-5), including tertiary (entry 6) and secondary alcohols (entry 7). The bromo derivative 15e (entry 5) was isolated as nice crystals and its structure was confirmed by X-ray analysis ( Figure 2, see Supplementary Information for details), supporting the relative configuration assigned by NMR. Compound 15e is a chiral molecule crystallized in a centrosymmetric space group, P2 1 /c. Therefore, its crystal structure is an 50:50 equimolar mixture of a pair of enantiomers (1R,4S/1S,4R) in a well-defined arrangement. Figure 2 shows the 1R,4S-enantiomer structure C7(S) and C9(R) of 15e.  Analyzing the intramolecular geometry, it is observed that the individual rings in 7k and 15e assume similar geometries: (a) rings A and D are, as expected, very planar; (b) ring B is in half-chair conformation, with atoms O1 and C8 in the flap positions; (c) ring C is in twist-boat conformation. Another similarity is observed comparing the ring B substituents: In both structures the methyl (or ethyl in 15e) and bromophenyl groups are in axial and bisectional positions respectively related to ring B.
In Figure 3, the equivalent 1R,4S-enantiomers of 7k and 15e are superimposed in a capped stick fashion. The overlay of molecular backbones clearly shows the conformational similarity between homologous atoms in rings A and B and their first neighbor atoms. Indeed, 7k and 15e adopt a very similar conformation in terms of the torsion angles about the bonds by which the bromophenyl ring links the polycyclic three ring system: C6-C5-C7-C11 = 119.0(2) and C4-C5-C7-O1 = 62.5(2)° for 7k, and 120.2(2) and 63.3(3)°, respectively, for 15e. On the other hand, the molecular overlay also shows that the homologous atoms in rings C and D do not match. In fact, despite having the same 6-member ring shape, i.e., a twist-boat conformation, the ring C orientation in 7k strongly differs from that one found in 1R,4S-enantiomer of 15e. As consequence the D rings also do not match themselves. The insert of Figure 3 illustrates the rotated conformation of ring C comparing 7k and 15e. Interestingly, the superimposition of the pure 1R,4S-enantiomer of S15 with the opposite one of S97 (1S,4R) shows that rings D and C match ( Figure 4). The molecular geometries were studied through MOGUL [43], a knowledge base that takes a molecule submitted either manually or by another computer program via an instruction-file interface and perform substructure searches of the Cambridge Structural Database (CSD) for, typically, all its bond, angles and torsion angles. In both structures, all bond lengths and angles are in agreement with the expected ones, when compared with the similar structures and considering a good refinement.
The crystal packing in 7k and 15e is dominated by Van der Waals close contacts. In both structures the molecules are self-assembled generating a double chain along the unit cell b axis involving molecules related to 2 1 -fold screw axis ( Figure 5). Surprisingly, despite the chemical, molecular conformational, and space group symmetry differences comparing 7k and 15e, the 1-D structure along its respective unit cell b axis are very similar ( Figure 5). This supramolecular synthon is itself linked by Van der Waals forces along unit cell a and c axis completing the 3-D network of the two structures ( Figure 6).  The final stage to understand the scope of the iodine-catalyzed Prins cyclization was the investigation of the reactivity of a series of acyclic homoallylic alcohols with p-anisaldehyde (2b) ( Table 5). In the reaction of alcohols 1h-l the carbocation intermediate 4 (cf. Scheme 1) reacts with iodide giving 4-iodo-tetrahydropyrans 5h-l, as product, instead of dihydropyran. Iodine is not regenerated in the medium and the catalytic cycle is interrupted. Thus, as expected, the yields of the reaction are clearly related to the amount of iodine. Using 0.5 equiv of iodine the yields were 33%-40% (entries 1-5), whereas using 1 equiv. [29,30], the yield jumped to 81%-84% (entries 6-7). Using 0.2 equiv of iodine, compound 5h was obtained in 5% yield. The 4-iodo-hydropyrans 5h-l were isolated as a single diastereomers, corresponding to the equatorial attack of the iodide [5,44].
In the Prins cyclization with the monoterpene isopulegol (1m), the carbocation intermediate 4m reacts with water (formed in situ) giving the alcohols (+)-16m and (+)-17m, in a 5:1 mixture, respectively. These products were obtained in very good yield (81%), using 5 mol % of iodine. The attack of the water to the carbocation intermediate 4m occurs at the equatorial position preferentially, as expected (Scheme 4) [5,44].

Mechanistic Insights Based on DFT Calculations
The mechanism of the Prins cyclization of aldehydes and homoallylic alcohols was investigated by DFT theoretical calculations. The following topics were addressed: (i) the preferential formation of the cis diastereomer of compound 7 (Tables 2 and 3); and (ii) the preferential formation of 7 instead of 12 ( Table 2).
The formation of compound 7h (Table 2, Entry 6, 8:1 cis:trans) was selected as the model reaction for the theoretical study. The cis diastereoselectivity has been rationalized assuming the preferential formation of the intermediate (E)-3 as well as repulsive steric interactions that compromise the rotation around the dihedral angle formed between the 1,2-dihydronaphthalene ring and the benzylidene(propyl)oxonium moiety (i.e., Me-CH-(C=C), φ 1 ) [31,45]. To verity this hypothesis, the structure of the intermediate carbocations (E)-and (Z)-3h in two different twist-boat conformations of the 1,2-dihydronaphtalene ring (namely A and B) were optimized in the gas phase and the rotational barrier around the dihedral angle φ 1 was determined.
The barrier for the E/Z isomerization of 3h is around 40 kJ mol -1 (Scheme 5), corroborating the preferential formation of (E)-3h, despite the fact that the non-stereospecific nucleophilic attack of the alcohol 1a on the carbonyl moiety may give, a priori, both E-and Z-oxocarbenium ions 3h. Furthermore, the potential barrier for the rotation of the dihedral angle φ 1 is around 60 kJ mol -1 ( Figure S1), supporting our initial hypothesis that the repulsive steric interaction between the methyl group and the aromatic hydrogen (H16) compromises the rotation around the dihedral angle φ 1 [31,45]. However, the E/Z diastereomers of 3h still can undergo the electrophilic addition step from both faces of the 1,2-dihydronaphtalene ring in two different half-chair conformations of ring C. Scheme 5 provides an -H + outline of the reaction paths leading to oxonium intermediates of type 4h in which the C12-C11 is staggered. Electrophilic addition leading to cyclization is spontaneous in all cases, due to the formation of a more stable carbocation 4h, compared to oxocarbenium ions 3h. As discussed above, (E)-3h is more stable than (Z)-3h in both conformers selected. However, conformer A of (11R)-cis-4h is in average 25 kJ mol -1 more stable than other diastereomers. Consequently, the favorable conversion of (E)-3h into (11R)-cis-4h seems to be the cause for the preferential formation of cis-7h. However, considering the activation barriers for the formation of (11R)-cis-4h and (11R)-trans-4h, (19 kJ mol -1 and 14 kJ mol -1 , respectively) and the more exergonic formation of the trans diastereomer, one cannot rule out the possible preferential formation of the trans diastereomer depending on the substitution pattern of 3 (Table S1). Animations showing the reaction coordinate (imaginary frequency) for electrophilic addition step giving both cis and trans isomers are available in the SI.
The elimination reaction involving cis-or trans-4 was investigated determining the relative stability of products 7 and 12. First, the elimination of 4s was selected as model system since 7s and 12s are produced by elimination in roughly equimolar amounts (7s/12s ratio = 1.5). The Boltzmann ratio 7s/12s is 1.6 (ΔGº = 1.2 kJ mol -1 , Table S1), which (although is in the limit of accuracy of the method) is in very good agreement with the experimental values. Next, the exclusive formation of 7h instead of 12h in the elimination reaction of 4h was investigated comparing the relative stability of both products. Figure 7 shows that both cis-and trans-7h are more stable than the corresponding isomers 12h, independent on the conformation of the ring C.  Briefly, results from theoretical calculations indicate that in the case of 3h, the formation of the E isomer is, as predicted before [31,45], preferable to other isomers because rotation around the φ 1 dihedral angle and E/Z isomerization of oxocarbenion ions 3 are both not thermodynamic favorable (Table S1 and Scheme 5). The relative stability of one of the E isomers favors the formation of the (11R)-cis-4h carbocation compared to (11S)-cis-4h, which originates from the electrophilic attack from the opposite face of the 1,2-dihydronaphtalene ring. The activation barrier for the formation of the cis-and trans-4h indicates an earlier transition state in the case of the conversion of (Z)-3h to trans-4h. Therefore, if the substitution pattern favors the Z isomer of the oxocarbenium ion 3, the preferential formation of the trans product would be possible. The formation of the product 7 is related to its higher stability when compared to 12. Furthermore, NBO analysis indicate that between the two possible cis-4h intermediates, the 11R has the H11 in a better alignment with the p-type LP*(C10), reducing the overall charge at the carbon and favoring the formation of product cis-7h by elimination.

Materials and Methods
All commercially available reagents were used without further purification unless otherwise noted. Commercially available isopulegol was purified by flash column chromatography (15% AcOEt in hexanes) and other homoallylic alcohols 1a-l were synthesized as previously reported [36,46]. 1m is commercially available. THF and Benzene were freshly distilled from sodium/benzophenone. CH 2 Cl 2 was freshly distilled over CaH 2 . TLC analyses were performed in silica gel plates, using UV and/or p-anisaldehyde solution for visualization. Flash column chromatography was performed using silica gel 200-400 mesh (Aldrich, St. Louis, USA). Melting points are uncorrected. All NMR analyses were recorded using CDCl 3 as solvent and TMS as internal pattern in Bruker (AC200) or Varian (INOVA300) spectrometers. IR spectra were measured on a Perkin-Elmer 1750-FT. HRMS analysis were performed on a Bruker Daltonics Microtof Eletrospray. The experimental procedures for the preparation of compounds 6a, 7a-b were previously reported [31].

Computational Details
Gaussian09 revision A.01 was used for all calculations [47]. All structures were optimized at the B3LYP/6-31+G(d,p) theoretical levels [48]. Stationary points were characterized as minima by vibrational analysis. All reported energies include zero-point energy (ZPE) as well as thermal corrections (T = 298.15 K) from frequency calculations. Relaxed potential energy surface scan was carried out at the B3LYP/6-31G(d) level. Natural-bonding orbitals (NBO) analysis [49] was carried out by NBO 3.1 as implemented in the Gaussian09 suite of programs.

X-ray Crystallography
Well-shaped single crystals of 7k and 15e were chosen for the X-ray experiments. The single crystal X-ray diffraction measurements were performed at 150K on a Gemini A-Ultra diffractometer equipped with an Atlas CCD detector using mirror monochromatized CuKα radiation (λ = 1.5418 Å). The programs CrysAlis CCD and CrysAlis RED [50] were used for data collection, cell refinement and data reduction. The structures were solved by direct methods using the software Sir92 [51] and refined by full-matrix least squares on F² using the software SHELXL-2013 [52]. All non-hydrogen atoms were clearly identified and refined with least square of complete matrix in F² with anisotropic parameters considered. H atoms on C atoms were positioned stereochemically and were refined with fixed individual displacement parameters [Uiso(H) = 1.5Ueq(C) for methyl groups or 1.2Ueq(C) for aromatic, methine and methylene groups], using the SHELXL riding model with C-H bond lengths of 0.95, 0.96, 1.00 and 0.99 Å for aromatic, methyl, methine and methylene groups, respectively. WINGX software [53] was used to analyze and prepare the data for publication. Molecular graphics were prepared using ORTEP-3 for Windows [54] and Mercury [55]. Crystal data, data collection procedures, structure determination methods and refinement results are summarized in Table S1. Crystallographic data for the structural analysis of the compounds discussed here have been deposited at the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK, and are available on request quoting the deposition numbers CCDC 948196 and 948057, for 7k and 15e, respectively.

Conclusions
The Prins cyclization of homoallylic alcohols and aldehydes can be performed using catalytic amounts of iodine as Lewis acid. Anhydrous conditions and inert atmosphere are not required in this metal-free protocol. The desired O-heterocycles were obtained in 52%-91% yield for 30 examples, including different homoallylic alcohols and several aliphatic and aromatic aldehydes. The main limitation is the use of some acyclic homoallylic alcohols, where the use of stoichiometric iodine is required to obtain good yield of the product. The mechanism and the ratio of the products could be explained by DFT calculations.