In this review we summarize the most important procedures for the preparation of functionalized organzinc and organomagnesium reagents. In addition, new methods for the preparation of polyfunctional aryl- and heteroaryl zinc- and magnesium compounds, as well as new Pd-catalyzed cross-coupling reactions, are reported herein. Experimental details are given for the most important reactions in the Supporting Information File 1 of this article.
Keywords: cross-coupling; heterocycles; insertion; metalation; organomagnesium; organozinc
The functionalization of heterocyclic scaffolds is an important task in current pharmaceutical research. In this review article, we describe the approaches to this problem that use functionalized magnesium and zinc heterocyclic intermediates. Some typical experimental procedures are indicated in each case for the most important methods. New Pd-catalyzed cross-coupling procedures are also presented.
Organozinc compounds [1-3] are important synthetic intermediates as they are compatible with a broad range of functional groups. The reactivity of a carbon–zinc bond is quite low, and therefore, reactions with organic electrophiles often require the use of transition metal catalysts. The preparation of aryl and heteroaryl zinc derivatives is conveniently achieved by three general procedures:
These three methods, developed recently in our laboratories, provide access to numerous heterocyclic zinc reagents (Scheme 1).
Although the direct insertion of zinc dust to alkyl iodides proceeds readily, the insertion to aryl iodides is very slow in THF and requires the use of polar solvents  or highly activated zinc . Recently, we found that the presence of LiCl greatly facilitates the insertion of zinc to aryl iodides . Thus, the insertion of zinc dust (activated with 1,2-dibromoethane and Me3SiCl) to ethyl 4-iodobenzoate (1) at 70 °C provides less than 5% of zinc reagent 2 after a reaction time of 24 h. On the other hand, in the presence of one equivalent of LiCl, the insertion of zinc is completed within 24 h at 25 °C. After the addition of a catalytic amount of CuCN·2LiCl , the arylzinc intermediate is allylated with allyl bromide providing the ester 3 in 94% isolated yield (Scheme 2) .
This method can be extended to a broad variety of functionalized heterocyclic iodides such as the pyridines 4 and 7. The corresponding zinc reagents 5 and 8 are obtained at 25 °C in quantitative yield. The allylation of pyridylzinc derivative 8 with allyl bromide provides pyridine 9 in 85% yield . Interestingly, a diiodide, such as 2,5-diiodothiophene (10) reacts selectively with Zn and LiCl to provide the iodothienyl ketone 11 in 94% yield after benzoylation. Subsequent treatment of 11 with a second amount of Zn and LiCl (1.4 equiv) provides a new intermediate zinc reagent within 10 min, which after allylation provides the 2,5-disubstituted thiophene 12 in 87% yield (Scheme 3) . The insertion reaction proceeds best with aryl and heteroaryl iodides, however, the presence of electron-withdrawing substituents greatly accelerates the zinc insertion rate and electron-poor-heteroaryl bromides, such as the bromofuran 13, react smoothly with Zn and LiCl to furnish the furylzinc reagent 14 within 12 h at 25 °C, which after Pd-catalyzed cross-coupling (Negishi reaction) affords the 5-arylated furan 15 in 89% yield.
Interestingly, a high chemoselectivity is observed with several heterocyclic dihalides [8,9]. Thus, the tribromopyrimidine 16 provides only the 4-zincated pyrimidine 17. After allylation, the expected allylated pyrimidine 18 is obtained in 63% yield. Also, the dibromothiazole 19 allows insertion of zinc only into the most labile C–Br bond (in position 2) leading to the zincated thiazole 20. After Negishi cross-coupling [10-12], the 2-arylated thiazole 21 is obtained in 85% yield. Polar functional groups, such as a tosyloxy-group are able to direct the zincation. Thus, the diiodoquinoline 22 is regioselectively zincated (25 °C, 12 h) to intermediate 23 leading to the polyfunctional quinoline 24 in 78% yield after copper(I)-mediated acylation (Scheme 4 and Supporting Information File 1, Procedure 1) . This regioselectivity is explained by the polar and electron-poor nature of the tosyloxy group, which leads to a strong electron-withdrawing effect and accelerates the insertion of zinc into the neighboring C–I bond. The presence of LiCl amplifies this effect through coordination to the tosyloxy group and to the ortho-iodide, and therefore facilitates the cleavage of this carbon–iodide bond.
This method has been extended to the preparation of benzylic zinc reagents . A remarkable chemoselectivity is observed and functional groups, such as an acetyl group, are perfectly compatible with such synthesis. Thus, the reaction of the benzylic chloride 25 with zinc dust (1.5 equiv) and LiCl (1.5 equiv) in THF at 25 °C for 3.5 h provides the corresponding zinc reagent 26 in 68% yield. Its half-life at 25 °C is approximately two days. The copper(I)-mediated acylation of 26 provides the expected diketone 27 in 74% yield (Scheme 5) [13,14].
A broad range of functional groups are tolerated, and homocoupling products account for less than 5% of the total. These benzylic zinc reagents give access to biologically important phenylacetic acids. Thus, the treatment of the chloro-substituted benzylic zinc compound 28 with ClCO2Et in the presence of Pd(PPh3)4 (5 mol %) furnishes the phenylacetic derivative 29 in 81% yield. Alternatively, a copper (I)-mediated reaction with NC-CO2Et provides the same product in 77% yield when a dummy ligand is added (Scheme 5) . Electron-rich benzylic chlorides, such as 30 are also readily converted to the desired zinc reagents 31. The Pd-catalyzed cross-coupling of 31 with the iodoquinoline 32 and with S-Phos as ligand [15-17] provides the alkaloid papaverine (33) in 68% yield (Scheme 5) . Ni-catalyzed cross-couplings can also be realized . Thus, the reaction of the benzylic zinc reagent 34, obtained by direct zinc insertion in the presence of LiCl, with the chloropyridine 35 in the presence of Ni(acac)2 (0.5 mol %) and PPh3 (2 mol %) affords the polyfunctional pyridine 36 in 90% yield (Scheme 6 and Supporting Information File 1, Procedure 2) [14,18,19].
Although the LiCl-mediated zinc insertion represents a major preparative advance for the synthesis of polyfunctional zinc reagents, this method has an intrinsic limitation due to the use of zinc as a reducing agent. Zinc has only moderate reducing properties, therefore its insertion into organic halides only proceeds smoothly in the case of aryl iodides and electron-poor substituted aryl bromides. The use of highly reactive zinc (Rieke-zinc) [20,21] solves the problem only partially. It is expensive and most aryl or heteroaryl chlorides do not react. Therefore, we used a stronger reducing reagent, magnesium. Magnesium turnings readily insert into a variety of aryl chlorides or bromides in the presence of LiCl. However, arylmagnesium reagents are compatible with fewer functional groups. Thus, methyl esters react rapidly with arylmagnesium reagents at 0 °C. In order to solve this problem, we have performed a Barbier-type preparation of aryl and heteroaryl zinc reagents by treating the aryl bromide or chloride with magnesium turnings in the presence of zinc chloride and LiCl. Under these conditions, the relatively unreactive aryl bromides and chlorides readily react. Furthermore, sensitive functionalities are tolerated since the reactive arylmagnesium species generated is immediately trapped with zinc chloride (Scheme 7) . Thus, methyl 3-bromobenzoate (37) reacts with magnesium powder in the presence of LiCl (1.5 equiv) and ZnCl2 (1.1 equiv) to provide the intermediate magnesium species 38, which is immediately trapped with ZnCl2 leading to the zinc reagent 39 in high yields. Subsequent Pd-catalyzed cross-coupling of 39 with an aryl iodide provides the cross-coupling product 40 in 79% yield (Scheme 7) .
The preparation of zinc reagents by a directed deprotonation was of limited use as no soluble zinc base was available [23,24]. We found that the treatment of commercially available TMPMgCl·LiCl (41) [25-27] with ZnCl2 (0.5 equiv) at 25 °C provides the new base TMP2Zn·2MgCl2·2LiCl (42) . All three metals Zn, Mg and Li are important in this mixed base . The role of LiCl is to increase the solubility of the base, the role of MgCl2 is to increase its reactivity and the role of zinc is essential since it confers to this base an exceptional chemoselectivity (Scheme 8). Thus, the 1,3,4-oxadiazole 43 is readily converted to the zinc reagent 44 by the reaction with TMP2Zn·2MgCl2·2LiCl (42, 0.55 equiv; 25 °C, 20 min). It should be noted that both TMP-moieties are used and that no fragmentation of this sensitive heterocycle is observed, as is the case for the corresponding Mg- and Li-derivatives.
After a reaction of the heterocyclic zinc reagent 44 with PhSO2SPh the corresponding thio-derivative 45 is obtained in 75% yield. Coumarine (46) can be directed zincated leading to the zinc reagent 47. After a Negishi cross-coupling with an aromatic iodide, the substituted coumarine 48 is obtained in 83% yield (Scheme 8 and Supporting Information File 1, Procedure 3) . This procedure tolerates most of the important functional groups in organic chemistry. Thus, the zincation of the formyl-substituted indole 49 is complete within 45 min at 25 °C leading to the zinc reagent 50. After allylation, the 2,3-disubstituted indole 51 is obtained in 71% yield (Scheme 9). Similarly, 2-nitrobenzofuran (52) is zincated without reacting with the nitro group, leading to the nitro-substituted zinc reagent 53. After allylation, the benzofuran 54 is obtained in 80% yield. The polyfunctional pyridine 55 is zincated with TMP2Zn·2MgCl2·2LiCl (42) leading to the zinc reagent 56. Subsequent allylation furnishes the trisubstituted pyridine 57 in 80% yield (Scheme 9) .
In some cases, the zincation using TMP2Zn·2MgCl2·2LiCl (42) is slow and requires long reaction times. This is the case for benzofuran (58), which requires 9 days at 25 °C for a complete zincation in position 2 leading to 59. The reaction time can be dramatically decreased by means of microwave irradiation. Under these conditions, the zincation is complete within 1 h at 120 °C (Scheme 10). Similarly, the functionalized pyridine 61 is zincated within 1 h at 80 °C under microwave irradiation leading to 62. The success of this procedure is a result of the high thermal stability of organozinc reagents. A Pd-catalyzed cross-coupling of 59 or a copper(I)-mediated acylation of 62 affords the products 60 and 63 in 80–95% yield (Scheme 10 and Supporting Information File 1, Procedure 4) .
Unexpectedly, recent research work from our laboratories showed that the preparation of heteroarylmagnesium reagents is compatible with numerous functional groups [31-33]. There are three important synthetic methods for the preparation of polyfunctional heteroarylmagnesium reagents:
Due to the higher polarity of the carbon–magnesium bond, these heterocyclic organometallics are significantly more reactive than the corresponding zinc reagents. This makes their preparation especially important.
Compared to the halogen/lithium exchange, discovered in 1939 by Wittig and Gilman, the halogen/magnesium exchange is much slower. Whereas aryl and electron-poor unsaturated iodides readily react with iPrMgCl and undergo a metal-metathesis to provide the more stable heteroarylmagnesium reagent (Scheme 11) , the reaction of aryl and heteroaryl bromides is slow when iPrMgCl is used as an exchange reagent.
However, with the aid of the mixed Li/Mg-reagent iPrMgCl·LiCl (64), an efficient exchange reaction is also effective with a wide range of aryl and heteroaryl bromides [31-33,35]. This reagent (64) is commercially available as an approx. 1 M THF solution from Chemetall GmbH . Recently, we have applied this exchange reaction for the regioselective functionalization of quinolines. Thus, the 2,3-dibromoquinoline (65) is regioselectively converted to the 3-magnesiated quinoline derivative 66. Using the same exchange reagent, iPrMgCl·LiCl (64) and 2,4-dibromoquinoline (68), it is now possible to obtain the 4-magnesiated quinoline 69. All these magnesiations proceed at low temperature (−50 °C to −78 °C) and are complete within 2 h reaction time. After reaction with TsCN, the corresponding nitriles 67 and 70 were obtained in 84–85% yield (Scheme 12 and Supporting Information File 1, Procedure 5) .
The rate of the Br/Mg-exchange depends on the electronic density of the heterocyclic rings. The electron-poor ring systems undergo considerably faster Br/Mg-exchange reactions than do heterocyclic ring systems bearing electron-rich substituents [31-35]. Therefore, in order to achieve a regioselective exchange with the very electron-poor tribromoquinoline 73, it was necessary to reduce the reactivity of the exchange reagent and thus, to switch from iPrMgCl·LiCl (64) to the less reactive mesitylmagnesium reagent MesMgCl·LiCl (71). This reagent is readily prepared by the reaction of mesityl bromide with magnesium turnings in the presence of LiCl (25 °C, 12 h; Scheme 13) . The lower reactivity of 71 allows a perfectly regioselective exchange reaction of 73, to afford the 3-magnesiated quinoline 74 only. A differentiation between the reactivity of a 3-bromo- and a 4-bromo-substituted quinoline is more difficult and even the use of the less reactive exchange reagent MesMgBr·LiCl is not satisfactory. This reactivity can be further tuned: First by preparing the dimesitylmagnesium reagent Mes2Mg·2LiBr (which has a higher reactivity than 71) and then by adding a complexation reagent, such as TMEDA (1 equiv), which considerably lowers the reactivity [37,38]. The new resulting reagent Mes2Mg·2LiBr·TMEDA (72) now reacts smoothly with 3,4-dibromoquinoline (76) providing selectively the 3-magnesiated 4-bromoquinoline 77. The quenching of 74 and 77 with TsCN and PhSO2SMe, respectively, leads to the regioselectively functionalized quinolines 75 and 78 in 79–88% yield (Scheme 13) .
This fine tuning is usually not necessary and numerous Br/Mg-exchange reactions making use of the commercially available reagent iPrMgCl·LiCl (64) have been reported in the literature [31-34,39].
The use of iPrMgCl·LiCl also proves to be very practical for the generation of polyfunctional alkenylmagnesium reagents, which react only slowly with iPrMgCl [40,41], as well as for the preparation of arylmagnesium reagents bearing sensitive functionalities, such as triazene. Thus, aryl iodide 79 is treated with iPrMgCl·LiCl (64) at −40 °C for 1 h leading to an intermediate magnesium reagent, which after transmetalation to the corresponding zinc reagent using ZnBr2 provides, after Negishi cross-coupling reaction with the bromoquinoline 80, the polyfunctinal triazene 81 in 75% yield. The conversion of the triazene functionality to an azide group is readily achieved by treating 81 with NaN3/BF3·OEt2-CF3CO2H in CH2Cl2 leading to the aryl azide 82 in 78% yield. Heating of 82 in mesitylene at reflux for 6 h provides ellipticine 83, a potent antitumor agent in 57% yield (Scheme 14) .
The structural variations of pyrimidines and purines are very important for the design of antiviral agents. The amination of this class of heterocycles is of particular importance. Recently, we developed an oxidative amination procedure for lithium derivatives using chloranil as oxidation agent . We applied this procedure in the preparation of a CDK inhibitor, purvalanol A (84). Thus, an I/Mg-exchange on the purine 85 with iPrMgCl·LiCl (64), followed by the transmetalation to the corresponding copper derivative with CuCl·2LiCl, and the addition of the lithiated aniline derivative 86, furnishes the amidocuprate 87. In the presence of chloranil amidocuprate 87 undergoes an oxidative coupling providing the adenine derivative 88 in 71% yield. A treatment with D-valinol (89) affords the desired CDK inhibitor, purvalanol A (84) in 65% yield (Scheme 15) .
The presence of LiCl facilitates greatly the insertion of many metals into carbon-halogen bonds and avoids the use of expensive activated forms of Mg, such as “Rieke-magnesium”. This property of LiCl for accelerating the insertion of Mg to organic halides has found numerous applications in the preparation of new polyfunctional arylmagnesium reagents. Thus, the rapid reaction of Mg/LiCl with aryl bromides 90, 93 and 96 allows an expeditive synthesis of the new arylmagnesium derivatives 91, 94 and 97. Quenching with typical electrophiles provides the expected products 92, 95 and 98 in 76–95% yield (Scheme 16) [22,45].
Remarkably, this insertion proceeds also with readily available and inexpensive aryl and heteroaryl chlorides, such as 99, 102 and 105, providing the functionalized magnesium reagents 100, 103 and 106 under mild conditions. The cross-coupling reaction of these Grignard reagents and transmetalation to zinc organometallics with ZnCl2 affords the expected products 101, 104 and 107 in 69–82% (Scheme 17 and Supporting Information File 1, Procedure 6) [9,22].
The directed magnesiation of aromatic substrates using TMPMgCl·LiCl (41) constitutes an economical preparation of a range of functionalized arylmagnesium compounds [25,26]. Sensitive heterocycles such as pyrimidines can be readily magnesiated with commercially available TMPMgCl·LiCl (41) . Thus, electron-poor 2-bromopyrimidine (108) is converted within 1.5 h at −55 °C in the presence of TMPMgCl·LiCl (41) to the corresponding magnesium reagent 109. A low reaction temperature is required in this case, since the sensitive heterocycle 108 undergoes ring addition reactions at temperatures above −30 °C. Quenching of the 4-magnesiated pyrimidine 109 with MeSO2SMe provides the thiomethyl derivative 110 in 81% yield (Scheme 18 and Supporting Information File 1, Procedure 7) .
The presence of a thiomethyl substituent considerably increases the electron density of this pyrimidine and the addition of a Grignard reagent to this heterocycle can no longer occur. Therefore, a subsequent magnesiation of 110 with TMPMgCl·LiCl (1.0 equiv) can be performed at 25 °C. After 5 min reaction time at this temperature, the resulting 6-magnesiated pyrimidine 111 is obtained quantitatively. Quenching of 111 with Cl2FCCClF2 (112) provides the trisubstituted pyrimidine 113 in 76% yield. A final functionalization in position 5 is readily achieved by treating 113 with a further equivalent of TMPMgCl·LiCl (41, 25 °C, 20 min) providing the 5-magnesiated pyrimidine 114. Quenching with benzoyl chloride furnishes the expected unsaturated ketone, which by treatment with hydrazine (NH2-NH2·H2O, THF, 25 °C, 10 min) leads to the pyrazolopyrimidine 115 in 56% overall yield (Scheme 18) . A similar approach has been used to prepare the p38 kinase inhibitor 119 in 72% overall yield, as well as the sPLA2 inhibitor 123, in a short reaction sequence (Scheme 19) .
Using TMPMgCl·LiCl (41), it is possible to prepare fully substituted indoles, such as 128 (Scheme 20) . Thus, starting from the aniline 124, an ortho-directed chlorination with N-chlorosuccinimide at 90 °C followed by an iodination with iodine and Ag2SO4 furnishes the tetrasubstituted aniline 125. Protection of the free amino-group followed by a Negishi-reaction provides the scaffold 126 in 80% yield (Scheme 20).
Successive magnesiations at the positions 5 and 3 of the tetrasubstituted anilines 126 with TMPMgCl·LiCl (41) can be performed. The strongly electron-withdrawing properties of the chloro-substituent favor a metalation at position 5. After the addition of pivaldehyde, the subsequent addition of a second equivalent of TMPMgCl·LiCl (41; −30 °C, 1.5 h) allows now a magnesiation in position 3. Quenching with TsCN and deprotection of the silylated aniline with KF and HCl furnishes the hexa-substituted aniline 127 in 76% overall yield. Potassium hydride mediated ring closure in NMP  affords the desired indole 128 in 96% yield (Scheme 20) .
In some cases, TMPMgCl·2LiCl (41) is not reactive enough to achieve a magnesiation in a reasonable time frame. We therefore prepared a more reactive bis-TMP base, TMP2Mg·2LiCl (129), by mixing TMPLi with the commercially available base 41 . The metalation temperature using such a base is low enough that functional groups such as a Boc-group or an aryl ketone are readily tolerated. Thus, the Boc-substituted benzophenone 130 reacts with TMP2Mg·2LiCl (1.1 equiv, −20 °C, 4 h) providing the expected aryl magnesium amide 131, which after a copper-mediated benzoylation leads to the 1,2,3-trisubstituted diketone 132 in 72% yield. This reagent allows a smooth functionalization of heterocycles such as the dicarbethoxypyridine 133, which is readily magnesiated with the base 129 at −40 °C within 3 h, leading to 134. After a Negishi cross-coupling reaction with an aromatic iodide, the 2-functionalized pyridine 135 is obtained in 73% yield (Scheme 21, Procedure 8) .
Although numerous cross-coupling methods have been recently described in the literature [50-52], there is still the need for new convenient procedures. We would like to focus on the chemoselectivity problem in cross-couplings in this short section and report two protocols recently developed in our laboratories:
The ability to perform cross-couplings is certainly one of the most versatile functions of heterocyclic zinc intermediates. Recently, we have shown that NiCl2 (0.05 mol %) constitutes an economical method for performing Negishi cross-couplings [18,19], however, it does not solve the problem of the moderate chemoselectivity of organozinc reagents towards substrates bearing acidic hydrogen atoms, such as N–H and O–H bonds. This is an important limitation of the Negishi cross-coupling, especially compared to the Suzuki cross-coupling based on boronic acid derivatives, which are much more tolerant toward acidic NH- and OH-groups. In the course of our studies, we found that by using an active catalyst system, such as S-Phos, developed by S. L. Buchwald [15-17], a smooth cross-coupling can be achieved between benzylic, aromatic and alkyl zinc reagents with substrates bearing an NH- or an OH-group (Scheme 22) [53,54].
Remarkably, this reaction protocol was extended to substrates bearing an α-aminoester moiety, such as 152 providing the cross-coupling product 153 in 85% yield (Scheme 23 and Supporting Information File 1, Procedure 9) .
As aryl- and heteroarylmagnesium reagents are readily available, it would be highly desirable if cross-couplings could be directly realized using these organometallics without the need of further transmetalation to zinc, boron or other metals. However, the disadvantage of this cross-coupling, known in the literature as Kumada cross-coupling [55,56], is that it only proceeds with relatively nonfunctional molecules as the C–Mg bond can competitively attack the functional group present in the aromatic or heterocyclic electrophile instead of undergoing the desired cross-coupling. We have found that the presence of iPrI (or another alkyl iodide) catalyzes the Kumada cross-coupling reaction, such that highly reactive functional groups, such as ketones, esters or nitriles, are perfectly tolerated (Scheme 24 and Supporting Information File 1, Procedure 10) [57,58].
The mechanism of the reaction has been shown to be of radical nature, and it affords the cross-coupling products in very short reaction times, often less than 5 min.
The addition reactions of organometallic reagents to ketones, aldehydes and carbon dioxide are essential transformations in organic synthesis as they provide a convenient access to various types of alcohols or carboxylic acids. Usually, organozinc reagents only react with these types of electrophiles in the presence of catalytic amounts of transition metal salts and in a very limited scope. Recently, we showed that the cheap and non-toxic main group Lewis acid MgCl2 allows smooth addition reactions of different aromatic, heteroaromatic, alkyl and benzylic zinc reagents to various carbonyl derivatives and carbon dioxide without the use of polar cosolvents (Scheme 25 and Supporting Information File 1, Procedure 11). The Lewis acid MgCl2 is usually generated during the formation of the organozinc reagent by a magnesium insertion in the presence of ZnCl2 (compare section 1.2) [59,60].
Thus, 2-fluorophenylzinc bromide 163 and the pyrazolylzinc chloride 165 react at room temperature with the aromatic aldehydes to provide the secondary alcohols 164 and 166 in 87–91% yield. The alkyl zinc reagent 167 adds to α,α,α-trifluoromethylacetophenone in 2 h and the corresponding alcohol 168 was isolated in 76% yield. Furthermore, the method was applied to the synthesis of the blockbuster drug ibuprofen (170). To achieve this, the secondary benzylic zinc reagent 169 was reacted with CO2 gas to provide the phenylacetic acid 170 in 89% yield.
We have summarized the most important procedures for the preparation of functionalized organzinc and organomagnesium reagents in this short review. Although, these reagents were introduced to synthetic organic chemistry at the turn of the 20th century, they are now more than ever essential organometallic intermediates. The progress in the 5 last years in our laboratories shows that much is still unknown in this field, and that the important synthetic preparation methods developed recently will lead to a revolution in the field and considerably expand the use of these organometallics in synthesis.
Experimental details for the most important reactions of this review are given in the Supporting Information File 1.
|Supporting Information File 1: Experimental section.|
|Format: PDF||Size: 544.0 KB||Download|
This research was funded by the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013) ERC grant agreement n° 227763. Furthermore, we thank the DFG (SFB 749) for financial support. We also thank Chemetall GmbH (Frankfurt), Umicore AG (Angleur, Belgium), Heraeus Holding GmbH (Hanau) and BASF SE (Ludwigshafen) for their generous donation of chemicals.