1

Epimerization of Tertiary Carbon Centers via Reversible Radical Cleavage of Unactivated 1 C(sp3)−H Bonds 2

3

Yaxin Wang,1# Xiafei Hu,1# Cristian A. Morales-Rivera,3 Guo-Xing Li,1 Xin Huang,1 Gang He,1,2* Peng 4 Liu,3* and Gong Chen1,2* 5

1State Key Laboratory and Institute of Elemento-Organic Chemistry, College of Chemistry, Nankai Uni-6 versity, Tianjin 300071, China 7

2Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin 300071, 8 China 9

3Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, USA 10

11

12

Abstract 13

Reversible cleavage of C(sp3)-H bonds can enable racemization or epimerization, offering a val-14 uable tool to edit the stereochemistry of organic compounds. While epimerization reactions operating via 15 cleavage of acidic C(sp3)-H bonds, such as the Ca-H of carbonyl compounds, have been widely used in 16 organic synthesis and enzyme-catalyzed biosynthesis, epimerization of tertiary carbons bearing a non-17 acidic C(sp3)-H bond is much more challenging with few practical methods available. Herein, we report 18 the first synthetically useful protocol for the epimerization of tertiary carbons via reversible radical cleav-19 age of unactivated C(sp3)-H bonds with hypervalent iodine reagent benziodoxole azide and H2O under 20 very mild conditions. These reactions exhibit excellent reactivity and selectivity for unactivated 3o C-H 21 bonds of various cycloalkanes and offer a powerful strategy for editing the stereochemical configurations 22 of carbon scaffolds intractable to conventional methods. Mechanistic study suggests that the unique ability 23 of N3• to serve as a catalytic H atom shuttle is critical to reversibly break and reform 3o C-H bond with 24 high efficiency and selectivity. 25

26

27

28

29

30

31

32

33

2

34

Figure 1. Epimerization of tertiary carbon via reversible cleavage of tertiary C-H bonds. 35

Methods for the selective functionalization of alkyl C-H bonds have been greatly advanced over 36 the past few decades, offering streamlined strategies for the synthesis and modification of complex mol-37 ecules.1-4 A wide range of reactions have been developed to transform C(sp3)-H bonds into different func-38 tional groups.5-9 However, reactions featuring reversible cleavage of unactivated C(sp3)-H bonds have 39 received much less attention. Racemization or epimerization via reversible cleavage of C(sp3)-H bonds 40 might offer an invaluable tool for editing the stereochemistry of organic compounds. Additionally, ex-41 changing C(sp3)-H bonds with C-deuterium bonds may provide valuable deuterated compounds for bio-42 medical applications.10,11 Epimerization reactions of tertiary Ca of carbonyl compounds via cleavage of 43 their acidic C(sp3)-H bonds have been routinely used in the synthesis of complex molecules, and in cata-44 lytic process such as dynamic kinetic resolution (Figure 1A). Similar enolization mechanisms have also 45 been widely used by enzyme epimerases for the biosynthesis of natural products.12 Compared with acidic 46 C(sp3)-H bonds, epimerization of tertiary carbons bearing non-acidic C(sp3)-H bonds is much more chal-47 lenging. Interestingly, recent biochemical studies have shown that [4Fe-4S]-cluster radical S-adenosyl-L-48 methionine (SAM) epimerases invert the stereocenters of amino acid or sugar units through radical-me-49 diated pathways.13, 14 For example, an epimerase selectively converts an L-Ile amino acid residue of a 50 peptide to D-allo-Ile via abstraction of Ca-H by a 5’-deoxyadenosyl radical (5’-dA•) and quenching of 51 the carbon radical intermediate by a thiol group of an enzyme cysteine residue (Figure 1B).13 Radical-52 mediated C(sp3)-H cleavage reactions could potentially provide a unique solution to epimerize tradition-53 ally “unepimerizable” tertiary carbon centers in organic synthesis. To achieve such transformations with 54

H

R1 R2O

R3

H

R1R2

O

R3

R1

R2O(H)

R3

A) Epimerization via cleavage of acidic C(sp3)-H bond

B) Epimerization of Ile residue in peptide by radical SAM peptide epimerase

H

R1 R2R3 H

R1R2R3

R1

R2

R3

A A-H DD-H

A : H acceptor D-H : H donor

D) This work: selective epimerization of nonacidic 3o C−H of cyclic alkanes

H

R H

R IO

O

N3

enolization

BIN3 1

1 (2-3 equiv)EtOAc(or PhCl) / H2O

35-50 oC

efficient; selective; simple to operate;compatable with

complex substrate

unactivated3o C-H

NH O

H

NH O

H

L-Ile D-allo-Ile

NH O

OCH2

HOHO

NN

N

N

NH2

5'-dA

SAM

H-S5'-dA

Cys

C) Epimerization strategy via radical cleavage of nonacidic 3o C−H bond

R

[4Fe-4S]

a catalytic H atom shuttleN3

3

useful efficiency, the sequence of H abstraction by a suitable radical H atom acceptor (A•) and quenching 55 of the tertiary C-radical by suitable H atom donor (D-H) must be developed (Figure 1C). Herein, we 56 report an efficient and synthetically useful protocol for epimerizing tertiary carbons via radical cleavage 57 of non-acidic 3o C(sp3)-H bonds with hypervalent iodine reagent benziodoxole azide and H2O under mild 58 conditions (Figure 1D). 59

Results and discussion 60

Radical reactions can provide simple and efficient means to selectively cleave 3o and activated 2o 61 C-H bonds of organic compounds due to their relatively weak bond dissociation energy (BDE).15-20 In the 62 absence of activated 2o C-H bonds, 3o C-H bonds could potentially serve as a unique set of targets for 63 selective radical C(sp3)-H functionalization of complex substrates. While conventional radical C-H cleav-64 age reactions often require relatively harsh conditions, recent studies have shown radical reactions can 65 take place under much milder conditions.21-23 Compared with the large number of studies on various rad-66 ical-mediated functionalizations of 3o C-H bonds, epimerization reactions of tertiary carbons via reversi-67 ble cleavage of unactivated 3o C-H bonds have been sporadically studied, mostly more than thirty years 68 ago.24-29 While pioneering works by Mazur,24, 25 Kochi,26 Hill,27 and others have demonstrated the feasi-69 bility of such a transformation, their methods demand relatively harsh operating conditions such as strong 70 UV irradiation, high temperature or use of toxic reagents (e.g. HgBr2), and exhibit narrow substrate scope. 71

In 1996, Zhdankin reported that the reaction of simple alkanes with azidobenziodoxole (BIN3, 1) 72 in the presence of radical initiator benzoyl peroxide (BzOOBz) at elevated temperature led to selective 73 azidation of 3o C−H bonds (see reaction of model substrate cis-decalin 2-cis in entry 1 of Table 1).30 74 Recently, Hartwig discovered that Fe/PyBOX catalysts promote similar C(sp3)−H azidation of more com-75 plex substrates at room temperature (entry 2).31 Subsequently, we developed a visible light (VL)-promoted 76 method to affect selective 3o C−H azidation with photosensitizer Ru(bpy)3Cl2 in hexafluoroisopropanol 77 (HFIP) solvent (entry 3).32, 33 Our visible light-promoted azidation reaction likely starts with the formation 78 of N3• or benziodoxole (Bl•) radicals via single electron transfer (SET) activation of BIN3.34-38 These 79 radicals abstract a H atom from 3o C-H bond to form a tertiary C-radical, which then reacts with BIN3 to 80 give the azidation product. In this VL-promoted system, we noted that C-H halogenation products were 81 obtained in high yield and selectivity when in the presence of halide salts, such as LiCl.32 This successful 82 modulation of the hypervalent-iodine mediated radical reaction pathway prompted us to attempt C-H epi-83 merization using a combination of H donor (D-H)39 and BIN3. As shown in entry 4, we were pleased to 84 see that the desired epimerization product trans-decalin 2-trans was obtained in 35% yield when 1 equiv 85 of H-donor Bu3SnH was included in the reaction mixture. Interestingly, a small amount of 2-trans was 86 formed in the absence of photosensitizer and VL irradiation (entry 5). The yield of 2-trans was improved 87 to 75% when EtOAc solvent was used (entry 6). Use of excess of Bu3SnH (2 equiv) gave lower yield 88 (entry 7). Use of other H-donors such as Ph2P(=O)H, Et3SiH, and Hantzsch ester 7 also gave 2-trans in 89 moderate to excellent yield along with small amount of azidation byproduct 3 (entries 8-10). As seen in 90 entry 11, only trace amount of 2-trans was formed in dry EtOAc solvent in the absence of H-donors. 91 Surprisingly, a clean epimerization with almost complete suppression of azidation was observed when a 92 mixture of EtOAc/H2O (9:1) was used (entry 13). A mixture of PhCl/H2O gave comparable results 93

4

94 entry reagents (equiv) temp

(oC) solvents 2-trans

(2-cis) 3

1 1 (2), BzOOBz (0.1), 24 h 80 DCE <1(36) 62 2 1 (2), Fe(OAc)2 (0.1), PyBOX (0.1), 24 h 23 CH3CN <1 (8) 90 3b 1 (2), Ru(bpy)3Cl2 (0.1%), VL, 24 h 35 HFIP <1 (3) 95 4b 1 (2), Ru(bpy)3Cl2 (0.1%), Bu3SnH (1), VL, 24 h 35 HFIP 35 (13) 52 5 1 (2), Bu3SnH (1), 24 h 35 HFIP 7 (34) 59 6 1 (2), Bu3SnH (1), 24 h 35 EtOAc 75 (7) 17 7 1 (2), Bu3SnH (2), 24 h 35 EtOAc 10 (89) <1 8 1 (2), Ph2P(=O)H (1), 24 h 35 EtOAc 89 (2) 9 9 1 (2), Et3SiH (1), 24 h 35 EtOAc 80 (19) <1 10 1 (2), 7 (1), 24 h 35 EtOAc 91 (2) 7 11 1 (2), 24 h 35 EtOAc (dry) 1 (92) 7 12 1 (2), 24 h 35 EtOAc (+ 1% H2O)c 84 (7) 8 13 1 (2), 24 h 35 EtOAc/H2O (9:1) 97 (<1) <1 14 1 (2), 24 h 35 DCE/H2O (9:1) 87 (2) 10 15 1 (2), 24 h 35 PhCl/H2O (9:1) 95 (2) 3 16 1 (1), 24 h 35 EtOAc/H2O (9:1) 90 (9) <1 17 1 (0.5), 7 d 35 EtOAc/H2O (9:1) 95 (4) <1 18 1 (0.1), 7 d 50 EtOAc/H2O (9:1) 69 (30) <1 19 1 (0.1), HOAc (0.2), 7 d 50 EtOAc/H2O (9:1) 80 (19) <1 20 5 (2), 24 h 35 EtOAc/H2O (9:1) 0 (100) 0 21 6 (2), 24 h 35 EtOAc/H2O (9:1) 0 (100) 0 22 1 (2), 24 h, air 35 EtOAc/H2O (9:1) 0 (95) (4d) 23 1 (2), 24 h, in darkness 35 EtOAc/H2O (9:1) 97 (1) 1

95

96

Table 1. Epimerization of cis-decalin with BIN3. a) Yields are based on GC-MS analysis of reaction mixture on a 97 0.1 mmol scale at a 0.2 M concentration using ACS grade solvents under Ar atmosphere unless specified otherwise. 98 Source of VL: 18 W compact fluorescent lamp. b) 2 M concentration. c) EtOAc solvent with 1% of H2O (~3 equiv) 99 was used. d) 3o C-H hydroxylation side product 4 was obtained in 4% yield. 100

(entry 15). The use of 0.1 equiv of BIN3 over an extended reaction time (7 days) at 50 oC led to 69% yield 101 of 2-trans, suggesting a catalytic pathway in the BIN3-mediated epimerization reaction (entry 18). In 102 contrast to BIN3, other benziodoxole reagents such as hydroxybenziodoxole BIOH 5 and acetoxybenzio-103 doxole BIOAc 6 exhibit no reactivity (entries 20, 21). A trial conducted under an air atmosphere gave 104 only trace amount of C-H hydroxylation side product 4 (entry 22). Irradiation with visible light had no 105 impact on the epimerization reaction (entry 23). 106

Substrate scope and selectivity. With the optimized conditions in hand, we next explored the 107 scope of this BIN3/H2O-mediated C-H epimerization reaction with representative mono-, bicyclic and 108 acyclic alkanes bearing suitable 3o C(sp3)-H bonds (Figure 2). In general, the efficiency of these epimer-109 ization reactions correlates well with the relative thermodynamic stability of the corresponding epimers. 110

H

H

H

H

N3

H2-cis 2-trans 3

Ara

NH

Me Me

CO2EtEtO2CIO

O

N3I

OO

OH

BlN3 1 BlOH 5

IO

O

OAc

BlOAc 6 7

5

Reactions of epimers with small free energy difference are bidirectional and give similar product distri-111 bution from either epimer starting material, indicating the ability to reach epimerization equilibrium (see 112 8, 11). Reactions of epimers with large free energy difference are unidirectional, predominantly giving 113 the more stable epimer (see 2, 9). The site and chemo-selectivity of this C-H epimerization is strongly 114 influenced by the BDE of C-H bonds and electronic effects. Electron-rich 3o C-H bonds of cyclic alkanes 115 are most reactive, forming only small amount of azidation byproducts (typically less than 5%). Electron-116 withdrawing groups such as alkoxylcarbonyl (e.g. CO2Bn) and carboxylate oxygen (e.g. BzO) not only 117 diminish the epimerization reactivity of neighboring 3o C-H bonds but also lead to more C-H azidation 118 byproduct.40-42 Compared to electronic effects, steric effects appear to have a less significant impact on 119 the reactivity (see 10, 12). 120

As shown in Figure 2A and 2B, the epimerization of 5- and 6-membered mono- and bicyclic 121 alkanes generally works well. While the reactions of 1,4 or 1,2-disubstituted cyclohexanes (10-14, 16) 122 favor the formation of more stable trans isomers, the reaction of 1,3-disubstituted cyclohexane 15 slightly 123 favors the more stable cis-isomer 15-cis. Functional groups such as benzoate (9), benzyl ester (10), 124 phthalimide (13), BocNH (20), and tosyl (12) were tolerated. Epimerization of isomenthol benzoate 19-I 125 takes place at either C5 or C2 position to give a mixture of 19-II (65%) and 19-III (16%). The epimeriza-126 tion of Boc-protected phenylalanine ester 20-cis selectively occurs at the 3o C-H bonds of cyclohexane 127 without reaction at the acidic Ha of the amino acid. Reaction of 14-cis bearing a secondary amide group 128 BzNH under standard conditions gave a complex mixture. However, the reaction was improved with the 129 addition of 2 equiv of Et3SiH. Including Et3SiH in the reaction mixture improved the epimerization per-130 formance in some cases (see Figure 3). The lack of reactivity observed with 23-cis and 24-cis indicates 131 that electron-withdrawing CO2Bn and BzO groups deactivate their adjacent 3o C-H bond (Figure 2C). 132 Accordingly, the epimerization of 10-15 should occur at the more electron-rich 3o C-H bonds highlighted 133 in green. Reactions in EtOAc/H2O usually gave less C-H azidation byproduct than in PhCl/H2O (see 8 134 and 13). The epimerization reactivity of electronically deactivated substrates can be slightly improved 135 with PhCl /H2O solvents (see 9 and 17 bearing two electron-withdrawing OBz groups). Compared with 136 cyclohexane 16 and cyclopropane 17, the lower epimerization efficiency of 18 may be caused by the 137 higher BDE of the cyclobutane 3o C-H bonds. No epimerization reaction was observed for the correspond-138 ing cyclopropane substrate due to its even higher 3o C-H BDE. These OBz-protected substrates were 139 easily prepared in two steps from the corresponding commercially available carboxylic acids. Compared 140 with the clean epimerization reactions of benzyloxycarbonyl, benzoate and tosylate substrates (e.g. 10, 141 11, 12), a trial with benzyl ether 25-cis gave a complex mixture and benzaldehyde byproduct, presumably 142 caused by oxidation of the benzylic C-H bond. As indicated by the slow racemization of 21-S, 3o C-H 143 bonds of acyclic alkanes are less reactive than the 3o C-H bonds of cyclic alkanes. In 22, the target 3o 144 benzylic C-H bond gave mostly azidation byproduct 22-N3 possibly due to the low H abstraction reactivity 145 of the corresponding benzyl radical intermediate. 146

As shown in Figure 3, this epimerization protocol has been successfully applied to steroid sub-147 strates bearing multiple 3o C-H bonds. Reaction of androsterone benzoate 26 selectively gave the C14-148 epimer 27 with a cis C/D ring juncture in good yield under slightly modified conditions with the addition 149 of 0.1 equiv of Et3SiH. Reaction of androstanediol derivative 28 gave C14-epimer 29 in 65% isolated yield. 150

6

151

Figure 2. Substrate scope of BIN3/H2O-mediated 3o C-H epimerization. a) Standard reaction conditions: 0.1 mmol 152 of alkane, 0.2 mmol of BIN3, EtOAc/H2O (0.45/0.05 mL), 35 oC, Ar, 24 hours unless specified otherwise. b) EtOAc 153 is replaced with PhCl. c) More azidation byproduct is formed in PhCl/H2O than in EtOAc/H2O. d) BIN3 (3 equiv). 154 e) Additional Et3SiH (2 equiv). Significant oxidation at NH-adjacent 3o C-H occurred in the absence of Et3SiH. f) 155 Reaction temperature: 50 oC. g) t = 48 h. h) t = 3 d. i) t = 5 d, j) Yields are based on GC-MS analysis. k) Yields are 156 based on 1H-NMR analysis. l) Isolated yield. m) about 20% of benzamide byproduct was formed. n) A cis/trans 157 (3/1) mixture was used. o) Yields are analyzed by chiral HPLC. p) azidation of 3o C-H bond. SM: starting material. 158

NH O

OHO

O

HH

HNH O

OHO

O

HH

H

65% (60%l)(+ 23% of 20-cis, 10% of 20-N3

c,p)k

20-cis 20-trans30% (26%l)

(+ 60% of 20-trans, 10% of 20-N3)k

HH

OBz

H

OBz

H

OBzOBz

H

H

H

H

BzO H BzO

HnBu

H H

44%(+ 55% of 11-trans,

<1% of 11-N3)k

11-cis 11-trans57% (50%l)

(+ 42% of 11-cis, <1% of 11-N3)k

BnO2C

H

H

tBu

6%(+ 93% of 10-trans,

<1% of 10-N3)k

10-cis 10-trans66%

(+ 33% of 10-cis, <1% of 10-N3)k

nBu

BzOBzO

HH

51%(+ 47% of 15-trans,

2% of 15-N3)k

15-cis 15-trans26% (21%l)

(+ 72% of 15-cis, 2% of 15-N3)k

TsO H TsO

HtBu

H H

tBu

34%(+ 65% of 12-trans,

<1% of 12-N3)k

12-cis 12-trans55% (50%l)

(+ 45% of 12-cis, <1% of 12-N3)k

13-cis 13-trans73% (68%l)

(+ 16% of 13-cis, 10% of 13-N3

c)k

PhthN H PhthN

H

H H

18-cis 18-trans18%

(+ 78% of 18-cis, 3% of 18-N3)k

H

56%(+ 42% of 16-cis,

2% of 16-N3)k

16-cis 16-trans8%k

(+ 90% of 16-trans, 2% of 16-N3)k

17-cisn 17-trans65%

(+ 29% of 17-cis, 5% of 17-N3)k

OBz

OBz

H

HOBz

H

OBz

HH

OBz

OBzHH

OBzBzOOBz

OBz

H

HOBzH

H

OBz

77%(+ 14% of 8-cis,

9% of 8-N3c) j

8-trans8-cis22%

(+ 76% of 8-trans, 2% of 8-N3) j

9-cis 9-trans75%

(18% of 9-cis, 6% of 9-N3)k

3%(+ 96% of 13-trans,

<1% of 13-N3)k

A) Bicyclic alkanes

B) Monocyclic alkanes

H

H

H

H

2-cis 2-trans97%

(+ <1% of 2-cis, <1% of 2-N3) j

BnO2C H

tBu

H

BnO2C H

CO2Bn

H 23-cis(>95% SM recovered)a

OBz

24-cis(>95% SM recovered)a

OBzH

H

H

OBz

C) Other substrates

*

21-S(enantiopure)

racemization

25-cis

BzO

H

H19-I

BzHN H BzHN

H

H H

5%(+ 69% of 14-trans, <1% of 14-N3)k, m

14-cis 14-trans58 (50%l)

(+ 9% of 14-cis, <1% of 14-N3)k, m

HH

1 (3)d

PhClb

48 hg

1 (3)d

PhClb

48 hg

1 (3)d

PhClb

50 oCf, 48 hg

1 (3)d

PhClb

3 dh

1 (3)d

5 di

1 (3)d

PhClb

50 oCf, 48 hg

1 (3)d

50 oCf

48 hg

BzO H

Ph

H

standard conditionsa

1 (3)d

Et3SiH (2)e

1 (3)d

PhClb

50 oCf, 48 hg

1 (3)d

PhClb

50 oCf, 3 dh

1 (3)d

PhClb

3 dh

1 (2 equiv)EtOAc/H2O (9/1)

35 oC, 24 h

1 (3)d

5 di

OBz*

H

1 (2)a

EtOAca

24 ha

2%(+ 95% of 17-trans,

2 % of 17-N3)k

2%(+ 96% of 18-trans,

<1% of 18-N3)k

BzO

H

H19-II 65%k

19-III 16%k

(+ 18% of 19-I, <1 of 19-N3)k

BzO

H

+

21-R22%

(+ 61% of 21-S, 16% of 21-N3c)o

22(cis/trans: 1:3.6) (+ 25% of 22 (of same cis/trans ratio)) j

1 (3)d

3 dh

BzO N3

Ph

H 22-N3 75% (cis/trans ~ 1:1)

1 (3)d

48 hg

25

O

OBnH

H

O

Hdecomposition(>90% SM consumed) +

~20%byproduct

1 (2)a

PhClbPh

H

7

The epimerization of pregnanediol dibenzoate 30 took place at either C5 or C14 position to give a diastere-159 omeric mixture of 31 (23%) and 32 (57%). The structures of epimerization products have been confirmed 160 by X-ray crystallography. 161

162

163

Figure 3. 3o C-H epimerization of steroids. Reactions are conducted on a 0.1 mmol scale. a) Yields are based on 164 1H-NMR analysis. b) Structure of its oxime derivative was confirmed by X-ray crystallography. c) Isolated yield. 165 See SI for detailed X-ray structures. 166

Mechanistic study. Control experiments and density functional theory (DFT) calculations43 have 167 been carried out to understand the mechanism of this BIN3/H2O-mediated C(sp3)-H epimerization reaction. 168 As outlined in Figure 1C, a radical 3o C-H epimerization reaction would require efficient H-abstraction 169 by a H-acceptor (A•) as well as selective quenching of the resulting carbon radical intermediate by a H-170 donor (D-H). Ideally, A• should not react with D-H, avoiding non-productive consumption of donor and 171 acceptor. 40-45 Furthermore, competing reaction pathways of the carbon radical intermediates need to be 172 suppressed to achieve clean epimerization. 1) Previous studies have shown that BIN3 is uniquely effective 173 at initiating the C-H activation step through the formation of H-acceptor Bl• 33 or N3• via dissociative 174 SET or homolytic cleavage of I-N bond (Figure 4A).32 In this reaction system, homolytic I-N cleavage of 175 BIN3 is expected to occur at ambient temperature in the absence of light irradiation. This is supported by 176 the DFT calculations that indicated a very small BDE of 27.8 kcal/mol for the Bl-N3 bond (eq 1), com-177 pared to the calculated I-O BDEs for Bl-OH 5 and Bl-OAc 6 (42.3 and 42.5 kcal/mol, respectively). More 178 importantly, H-N3 has a BDE of 93.3 kcal/mol, which is very close to the BDE of unactivated 3o C-H 179

CH3CH3

H

H

H

CH3CH3

HH

H

HOBz

O

CH3CH3

HH

H

HAcO

OBz

CH3CH3

H

HH

HOBz

O

3 5

149

8

H

OBz

OBz

EtOAc/H2O (9/1) 50 oC, 4 d

CH3CH3

HH

H

HOBz

OBz

3 5A B

C D

EtOAc/H2O (9/1) 35 oC, 6 d

14

Bl-N3 (3 equiv)

EtOAc/H2O (9/1) 50 oC, 4 d

CH3CH3

H

HH

HAcO 14

OBz

5 +

514

Bl-N3 (3 equiv)Et3SiH (0.1 equiv)

27b

40%a (+ 50% of 26) without Et3SiH72%a (+ 25% of 26) with Et3SiH

29 76%a (65%c)(+ 23% of 28)

31, 23%c

30 pregnanediol

26 androsterone

28 3β-androstanediol

BlN3 (3 equiv)(Et3SiH 0.1 equiv)

CH3CH3

HH

HH

OBz

OBz 32, 57%c

(+ 10% of 30)14

x-rayx-ray

8

180

Figure 4. Mechanistic studies. DFT calculations were performed at the M06-2X/6-311++G(d,p)-181 SDD/SMD(EtOAc)//M06-2X/6-31+G(d)-SDD/SMD(EtOAc) level of theory, See Supporting Information for 182 more details. a) 64% yield of 5 was formed for forward reaction of 1 (1 equiv) in PhCl/H2O (9/1, H2O ~ 9 equiv). 183 5% yield of 1 was formed for reverse reaction of 5 (1 equiv) and aq. HN3 (~ 3 equiv) in PhCl/H2O. b) aq. solution 184 of HN3 [~5 M] is prepared from reacting NaN3 with aq. H2SO4 at rt (see Supporting Information). 185

bonds (93.3 and 96.1 kcal/mol for 3o C-H of 2-cis and 2-trans, respectively, Figure 4C). In comparison, 186 the 2o C-H bonds of cyclohexane have a larger BDE of 97.5 kcal/mol. The close match in BDE of H-N3 187 and unactivated 3o C-H not only renders N3• a competent H-acceptor for 3o C-H bonds but also makes H-188 N3 a suitable H-donor to 3o carbon radical intermediates, making N3• an effective hydrogen atom shuttle 189 for 3o C-H bonds. 2) Although H2O is critical to the success of this epimerization reaction, H2O is unlikely 190 the immediate H-donor due to the high BDE of H-OH (~ 117.2 kcal/mol).46 Control experiments indicate 191 that BIN3 readily reacts with H2O to form BlOH 5 and HN3 in EtOAc or PhCl (>60% conversion in 30 192 min, eq 3). Furthermore, reacting BlOH with aq. HN3, prepared by mixing NaN3 with aqueous H2SO4, 193 can also form small amount of BIN3, suggesting a reversible reaction of BIN3 and H2O. Reaction of 2-cis 194 using D2O as co-solvent gave the deuterated product 34 in excellent yield (eq 4), suggesting DN3 as the 195 deuterium donor. 3) As shown in eq 5, the use of aq. HN3 alone did not give any epimerization product 196 under the reaction conditions. Similarly, BlOH alone cannot promote the epimerization reaction (entry 20, 197 Table 1). However, a combination of BlOH and aq. HN3 is effective (eq 6). These results suggest that the 198 residual BIN3, rather than HN3 or BlOH, is responsible for the initiation of C-H epimerization.32 Since 199 BIN3 is an excellent azidation reagent for carbon radicals, the low concentration of BIN3 and the abun-200 dance of HN3 in the reaction system might contribute to the suppression of the competing C-H azidation 201 pathway. 4) As outlined in the proposed reaction pathway in Figure 4B, this epimerization reaction likely 202

BlN3 1 H2OI

OO

OH

BlOH 5

HN3+

A) Selected calculated BDE values and control experiments

IO

O

N3

BlN3 1

IO

O

Bl

N3+

HN3 H N3+

ΔH = 93.3(kcal/mol)

ΔH = 27.8(kcal/mol)

2-cisEtOAc (0.4 mL)

35 oC, 48 h

2-cis

D

D

BlN3 (2 equiv)

PhCl/D2O (9/1)35 oC, 3 d

34

2-trans

Bl

H

R N3

R

N3

HN3 H

R

BlN3H2O

(or I )I II III

B) Proposed reaction mechanism

H abstraction

2-cisEtOAc/H2 (0.4 mL)

35 oC, 48 h

2-trans

aq. HN3 (~20 equiv, 0.4 mL)b

BlOH (5 equiv)

radical chain

initiation

(eq 1)

(eq 2)

(eq 4)

(eq 6)

(eq 5)

(eq 3)

(>95%)

33

(>95%)

+PhCla

35 oC, 30 min

aq. HN3 (~20 equiv, 0.4 mL)b

C) DFT calculation of epimerization of 2-cis

(kcal/mol)

ΔG(ΔH)

H

H

H

H

0.0(0.0)

11.2(1.5)

8.7(-0.7)TS1

N3 TS3N3

-3.0(-2.8)

HN3

0.1(2.0) -2.0

(0.0)

3.7(5.0)

TS2

HN3

TS1 TS3

2-cis2'

2'' 2-trans

TS2

BIOH

9

starts with a homolytic cleavage of the residual BIN3, generating Bl• 33 and N3• radical. N3• then selec-203 tively cleaves a 3o C-H bond of alkane I forming carbon radical II and HN3.38 Nucleophilic carbon radical 204 II then reacts with the electrophilic H-donor HN347 to give the epimerization product III or I and regen-205 erate N3•, thus propagating a radical chain reaction. As shown in Figure 4C, DFT calculations of the 206 epimerization of cis-decalin 2-cis show that both the initial H abstraction by N3• (TS1) and the subsequent 207 quenching of tertiary carbon radical with HN3 (TS3) proceed with low energy barriers.48,49 The late tran-208 sition state (r(C-H) = 1.32 Å in TS1) suggests the rate and selectivity of the C-H abstraction are sensitive 209 to the BDE of the C-H bonds.50 DFT calculations also indicate the more sterically hindered Bl• 33 is less 210 effective than N3• at cleaving the 3o C-H bond of 2-cis (see Supporting Information for details). Overall, 211 the facile activation of BIN3 at ambient temperature, the proper equilibrium between BIN3 and HN3 in the 212 presence of H2O, and the unique ability of N3• as a catalytic hydrogen atom shuttle enable an efficient, 213 selective and clean radical 3o C-H epimerization under very mild conditions. 214

Conclusion: In summary, we have developed the first synthetically useful protocol for radical-mediated 215 C-H epimerization reactions of tertiary carbon centers, using a hypervalent iodine reagent and H2O under 216 mild conditions. These reactions show excellent reactivity and selectivity toward unactivated 3o C-H 217 bonds of various cycloalkanes and offer a powerful tool to edit stereochemical configurations that are 218 intractable by conventional methods. Using this method, we have demonstrated easy access to novel ster-219 oid scaffolds. We hope that use of N3• as a catalytic hydrogen atom shuttle for unactivated C(sp3)-H bonds 220 might facilitate other challenging radical C-H functionalization transformations, such as C-H deuteration 221 and dynamic kinetic resolution, in future investigations. 222

223

Methods: 13-cis (24.3 mg, 0.1 mmol, 1.0 equiv) and BIN3 1 (86.7 mg, 0.3 mmol, 3.0 equiv) was dispersed 224 in mixture solvents of EtOAc (0.45 mL) / H2O (0.05 mL) (9:1) in a 4 mL glass vial at rt. The reaction vial 225 was purged with Ar for 1 min and sealed with a PTEF cap. The reaction mixture was vigorously stirred 226 at 35 oC for 5 days. The mixture was concentrated in vacuo, and the resulting residue was dissolved in 0.5 227 mL of CDCl3 along with Cl2CHCHCl2 (20 µL) as an internal standard for 1H-NMR analysis. The compo-228 sition of reaction mixture was analyzed based on the peaks at δ 2.56-2.31 (m, 2H) for compound 13-cis, 229 δ 2.26 (qd, J = 12.7, 3.4 Hz, 2H) for compound 13-trans and δ 2.86-2.71 (m, 2H) for compound 13-N3. 230 Yields of 13-cis (16%), 13-trans (73%) and 13-N3 (10%) are based on 1H-NMR analysis of the mixture. 231 The reaction mixture was then purified by silica gel flash chromatography (eluted with a gradient of hex-232 ane/acetone (v/v 100:0 to 10:1)) to give product 13-trans in 68% yield. 233

References: 234

1. Godula, K. & Sames, D. C-H bond functionalization in complex organic synthesis. Science 312, 67-235 72 (2006). 236

2. Davies, H. M. & Manning, J. R. Catalytic C-H functionalization by metal carbenoid and nitrenoid 237 insertion. Nature 451, 417-424 (2008). 238

3. Gutekunst, W. R. & Baran, P. S. C-H functionalization logic in total synthesis. Chem. Soc. Rev. 40, 239 1976-1991 (2011). 240

10

4. White, M. C. Adding aliphatic C-H bond oxidations to synthesis. Science 335, 807-809 (2012). 241 5. Chen, M. S. & White, M. C. A predictably selective aliphatic C-H oxidation reaction for complex 242

molecule synthesis. Science 318, 783-787 (2007). 243 6. Hartwig, J. F. Evolution of C-H bonds functionalization from methane to methodology. J. Am. Chem. 244

Soc. 138, 2-24 (2016) 245 7. He, J., Wasa, M., Chan, K. S. L., Shao, Q. & Yu, J.-Q. Palladium-catalyzed transformations of alkyl 246

C-H bonds. Chem. Rev. 117, 8754-8786 (2017). 247 8. Liu, W. et al. Oxidative aliphatic C-H fluorination with fluoride ion catalyzed by a manganese por-248

phyrin. Science 337, 1322-1325 (2012). 249 9. Le, C., Liang, Y., Evans, R. W. Li, X. & MacMillan, D. W. C. Selective sp3 C-H alkylation via 250

polarity-match-based cross-coupling. Nature 547, 79-83 (2017). 251 10. Loh, Y. Y. et al. Photoredox-catalyzed deuteration and tritiation of pharmaceutical compounds. 252

Science 358, 1182-1187 (2017). 253 11. Soulard, V., Villa, G., Vollmar, D. P. & Renaud, P. Radical deuteration with D2O: catalysis and 254

mechanistic insights. J. Am. Chem. Soc. 140, 155-158 (2018) 255 12. Tanner, M. E. Understanding nature’s strategies for enzyme-catalyzed racemization and epimeriza-256

tion. Acc. Chem. Res. 35, 237-246 (2002). 257 13. Benjdia, A., Guillot, A, Ruffie, P., Leprince, J. & Berteau, O. Post-translational modification of ri-258

bosomally synthesized peptides by a radical SAM epimerase in bacillus subtilis. Nat. Chem. 9, 698-259 707 (2017). 260

14. Kudo, F., Hoshi, S., Kawashima, T., Kamachi, T. & Eguchi, T. Characterization of a radical S-ade-261 nosyl-L-methionine epimerase, NeoN, in the last step of neomycin B biosynthesis. J. Am. Chem. Soc. 262 136, 13909-13915 (2014). 263

15. Newhouse, T. & Baran, P. S. If C-H bonds could talk: selective C-H bond oxidation. Angew. Chem. 264 Int. Ed. 50, 3362-3374 (2011). 265

16. Yi, H. et al. Recent advances in radical C-H activation/radical cross-coupling. Chem. Rev. 117, 9016-266 9085 (2017). 267

17. Mukherjee, S., Maji, B., Tlahuext-Aca, A. & Glorius, F. Visible-light-promoted activation of unac-268 tivated C(sp3)-H bonds and their selective trifluoromethylthiolation. J. Am. Chem. Soc. 138, 16200-269 16203 (2016). 270

18. Halperin, S. D., Fan, H., Chang, S., Martin, R. E. & Britton, R. A convenient photocatalytic fluori-271 nation of unactivated C-H bonds. Angew. Chem. Int. Ed. 53, 4690-4693 (2014). 272

19. Amaoka, Y. et al. Photochemically induced radical alkenylation of C(sp3)-H bonds. Chem. Sci. 5, 273 4339-4345 (2014). 274

20. Brodsky, B. H. & Du Bois, J. Oxaziridine-mediated catalytic hydroxylation of unactivated 3o C-H 275 bonds using hydrogen peroxide. J. Am. Chem. Soc. 127, 15391-15393 (2005). 276

21. Yan, M., Lo, J. C., Edwards, J. T. & Baran, P. S. Radicals: reactive intermediates with translational 277 potential. J. Am. Chem. Soc. 138, 12692-12714 (2016). 278

22. Studer, A. & Curran, D. P. Catalysis of radical reactions: a radical chemistry perspective. Angew. 279 Chem. Int. Ed. 54, 4436-4454 (2015). 280

11

23. Prier, C. K., Rankic, D. A., MacMillan, D. W. C. Visible light photoredox catalysis with transition 281 metal complexes: application in organic synthesis. Chem. Rev. 113, 5322-5363 (2013). 282

24. Gorodetsky, M. & Mazur, Y. Epimerization about unactivated tertiary carbon atoms in substituted 283 cyclopentane derivatives. J. Am. Chem. Soc. 90, 6540-6541 (1968). 284

25. Mazur, Y. Functionalization of non-activated carbon centers. Pure Appl. Chem. 21, 145-166 (1975). 285 26. Salomon, R. G. & Kochi, J. K. Acetone-sensitized photoepimerization about unactivated tertiary car-286

bon atoms. Tetrahedron Lett. 4387-4390 (1973). 287 27. Combs-Walker, L. A. & Hill, C. L. Use of excited-state and ground-state redox properties of poly-288

oxometalates for selective transformation of unactivated carbon-hydrogen centers remote from the 289 functional group in ketones. J. Am. Chem. Soc. 114, 938-946 (1992). 290

28. Augustine, R. L. & Caputo, J. A. The thermal equilibration of cis- and trans-b-decalones. An example 291 of an unusually large entropy of isomerization. J. Am. Chem. Soc. 86, 2751-2752 (1964). 292

29. Dang, H.-S. & Robert, B. P. Strategy for contra-thermodynamic radical-chain epimerisation of 1,2-293 diols using polarity-reversal catalysis. Tetrahedron Lett. 41, 8595-8599 (2000). 294

30. Zhdankin, V. V. et al. Preparation, X-ray crystal structure, and chemistry of stable azidoiodinanes - 295 derivatives of benziodoxole. J. Am. Chem. Soc. 118, 5192-5197 (1996) 296

31. Sharma, A. & Hartwig, J. F. Metal-catalysed azidation of tertiary C–H bonds suitable for late-stage 297 functionalization. Nature 517, 600-604 (2015). 298

32. Wang, Y., Li, G.-X., Yang, G., He, G. & Chen, G. A visible-light-promoted radical reaction system 299 for azidation and halogenation of tertiary aliphatic C−H bonds. Chem. Sci. 7, 2679-2683 (2016). 300

33. Li, G.-X. et al, A unified photoredox-catalysis strategy for C(sp3)–H hydroxylation and amidation 301 using hypervalent iodine. Chem. Sci., 8, 7180-7185 (2017). 302

34. Wirth, T. et al. Hypervalent iodine chemistry-modern developments in organic synthesis. Top. Curr. 303 Chem. Springer, Berlin, 2003. 304

35. Zhdankin, V. V. & Stang, P. J. Chemistry of polyvalent iodine. Chem. Rev. 108, 5299-5358 (2008). 305 36. Dohi, T., Kita, Y. New site-selective organoradical based on hypervalent iodine reagent for controlled 306

alkane sp3 C-H oxidations. ChemCatChem, 6, 76-78 (2014). 307 37. Y. Li, D. P. Hari, M. V. Vita, J. Waser, Cyclic hypervalent iodine reagents for atom-transfer reactions: 308

beyond trifluoromethylation. Angew. Chem. Int. Ed. 55, 4436-4454 (2016). 309 38. Antonchick, A. P. & Burgmann, L. Direct selective oxidative cross-coupling of simple alkanes with 310

heteroarenes. Angew. Chem. Int. Ed. 52, 3267-3271 (2013). 311 39. Gansäuer, A. et al. Hydrogen atom donors: recent developments. Top. Curr. Chem. 320, 93-120 312

(2012). 313 40. Heberger, K. & Lopata, A. Assessment of nucleophilicity and electrophilicity of radicals, and of polar 314

and enthalpy effects on radical addition reactions. J. Org. Chem. 63, 8646-8653 (1998). 315 41. De Vleeschouwer, F., Van Speybroeck, V., Waroquier, M., Geerlings, P. & De Proft, F. Electro-316

philicity and nucleophilicity index for radicals. Org. Lett. 9, 2721-2724 (2007). 317 42. Fisher, H. & Radom, L. Factors controlling the addition of carbon-centered radicals to alkenes - an 318

experimental and theoretical perspective. Angew. Chem. Int. Ed. 40, 1340-1371 (2001). 319

12

43. Mayer, J. M. Understanding hydrogen atom transfer: from bond strengths to Marcus theory. Acc. 320 Chem. Res. 44, 36-46 (2011). 321

44. Capaldo, L. & Ravelli, D. Hydrogen atom transfer (HAT): a versatile strategy for substrate activation 322 in photocatalyzed organic synthesis. Eur. J. Org. Chem. 2056-2071 (2017). 323

45. Crossley, S. W. M., Barabe, F. & Shenvi, R. A. Simple, chemoselective, catalytic olefin isomeriza-324 tion. J. Am. Chem. Soc. 136, 16788-16791 (2014). 325

46. Hioe, J. & Zipse, H. Radical stability and its role in synthesis and catalysis. Org. Biomol. Chem. 8, 326 3609-3617 (2010). 327

47. Workentin, M. S., Wagner, B. D., Lusztyk, J. & Wayner, D. D. M. Azidyl radical reactivity. N6•- as 328 a kinetic probe for the addition reactions of azidyl radicals with olefins. J. Am. Chem. Soc. 117, 119-329 126 (1995). 330

48. Barluenga, J., Campos-Gomez, E., Rodriguez, D., Gonzalez-Bobes, F. & Gonzalez, J. M. New io-331 dination reactions of saturated hydrocarbons. Angew. Chem. Int. Ed. 44, 5851-5854 (2005). 332

49. Ochiai, M. et al. Hypervalent (tert-butylperoxy)iodanes generate iodine-centered radicals at room 333 temperature in solution: oxidation and deprotection of benzyl and allyl ethers, and evidence for gen-334 eration of a-oxy carbon radicals. J. Am. Chem. Soc. 118, 7716-7730 (1996). 335

50. Liu, F., Yang, Z., Yu, Y., Mei, Y. & Houk, K. N. Bimodal Evans−Polanyi relationships in dioxirane 336 oxidations of sp3 C−H: non-perfect synchronization in generation of delocalized radical intermediates. 337 J. Am. Chem. Soc. 139, 16650-16656 (2017). 338

339 Supplementary Information: Detailed synthetic procedures, compound characterization, NMR spectra, 340 X-ray crystallographic data, and computational details are provided. 341

Acknowledgements: G.C. and G. H. thanks NSFC-21421062, -21672105, -21725204, and -91753124 342 for financial support of the experimental part of this work. P.L. thanks the NSF (CHE-1654122) for fi-343 nancial support for the computational part of the work. C.A.M.R. acknowledges an NSF research fel-344 lowship (DGE-1747452). Calculations were performed at the Center for Research Computing at the 345 University of Pittsburgh and the Extreme Science and Engineering Discovery Environment (XSEDE) 346 supported by the National Science Foundation. 347

Author Contributions: equal contribution# from Y.W. and X.H. Y.W. and X.H. carried out most of re-348 action discovery, study of substrate cope, structural determination of products, and prepared the support-349 ing information. C.A.M. conducted the computational studies. X.H. expanded the substrate scope. G.H. 350 supervised experimental studies. P.L. supervised the computational studies. P.L. prepared the computa-351 tional sections of the manuscript. G.C. formulated the initial ideas of this work, supervised the project, 352 coordinated with P.L. on computational studies, and prepared most of the manuscript. 353

Author Information: 354

There are no competing financial interests in this work. 355

Correspondence and requests for materials should be addressed to hegang@nankai.edu.cn, 356 pengliu@pitt.edu, gongchen@nankai.edu.cn 357

358