Reagent-controlled regiodivergent ring expansions of steroids

Ring expansion provides a powerful way of introducing a heteroatom substituent into a carbocyclic framework. However, such reactions are often limited by the tendency of a given substrate to afford only one of the two rearrangement products or fail to achieve high selectivity at all. These limitations are particularly acute when seeking to carry out late-stage functionalization of natural products as starting points in drug discovery. In this work, we present a stereoelectronically controlled ring expansion sequence towards selective and flexible access to complementary ring systems derived from common steroidal substrates. Chemical diversification of the reaction intermediate affords over 100 isomerically pure analogs with spatial and functional diversity. This regiodivergent rearrangement, and the concept of using chiral reagents to affect regiocontrol in chiral natural products, should be broadly applicable to late-stage natural product diversification programs.


Supplementary
solution of sodium 4-methyl thiophenoxide was prepared immediately prior to use by adding sodium (1.0 equiv) to a solution of 4-methylbenzenethiol (1.1 equiv) in anhydrous DMF (15.0 mL) at 0 ˚C and stirring at room temperature overnight. c NaBH4, MeOH. d 51% isolated yield of C9 was also obtained from LAH reduction of C2. e Hydrogenation using 10% Pd/C, EtOH. f Inversion of stereochemistry. g 58% isolated yield of D9 was also obtained from LAH of D2.

Supplementary Discussion
General mechanism of Schmidt reactions enabled by hydroxyalkyl azides 2,3 This reaction has been extensively studied by experimental and computational methods. For general information, see a relevant review 1 . The specific features of this reaction relevant to the present project are noted here for the reader's convenience. Following this description, specific details for selected examples are provided.
The reaction begins with attack of the alcohol group onto the carbonyl reactant, followed by elimination to afford an oxonium ion intermediate. The attached azide attacks the oxonium ion to afford a spirocyclic intermediate, which undergoes migration to provide an iminium ether. This intermediate can be isolated but is usually reacted in situ with a nucleophile to provide an N-substituted lactam product.
Supplementary Figure 1. General mechanism of ring expansions using hydroxyalkyl azides.

Factors that determine reaction regio-and stereochemistry
Most of the work pertaining to the regiochemical outcome of this reaction was carried out in the context of asymmetric ring expansions of achiral cyclohexanones to diastereoselectively make substituted caprolactams. In brief, the outcomes of these reactions depend on three considerations. They are the direction of azide attack relative to pre-existing substitution on the ketone reactant, selective formation and reaction of the most stable new heterocyclic ring (1,3-oxazinane from 3-azidopropanol or oxazoline from 2-azidoethanol), and antiperiplanar CN migration to afford the iminium ether product. Importantly, all steps leading to the actual migration are reversible on the reaction time scale and therefore thermodynamically controlled. This is supported by DFT calculationsthe barrier for CN migration is only ca. 2 kcal/mol higher than reversion to azide and oxonium ion 4

.
A representative case is shown and annotated below; this discussion will focus on the better-understood reactions of substituted cyclohexanones with 3-azidopropanols, but similar considerations apply to other versions as well. Overall, one can consider this complex process to arise from a combination of three stereochemical and conformational factors. Figure 2. The three determinants of stereo-or regioselectivity in ring expansion reactions mediated by chiral hydroxyalkyl azides. a, In intermediates derived from six-membered rings, equatorial attack onto the more stable cyclohexanone derivative has been proposed and supported by computation 4 ; the stereoselectivity is related to the relative populations of the possible conformations of the starting ketones (typically in chair conformations). Attack onto ketones of other ring sizes depend on steric accessibility (see specific examples below). Practically, since it is not possible to directly observe the spirocyclic intermediate, the direction of attack in non-obvious cases may need to be inferred from analyzing the outcomes in the context of considerations 2 and 3 below (see, for example, discussion of C-17 reactivity below in the SI). b, For intermediates derived from monosubstituted 3azidopropanols, the major products arise from spirocyclic 1,3-oxazinanes containing the carbon substituent in an equatorial orientation. It has been reported 4 that 2-aryl-or 2-alkoxy 3-azidopropanols lead to greater quantities of product arising from spirocyclic 1,3-oxazinanes having the aryl group in an axial position (due to a -cation interaction between the aryl group and the N2 + group, which is now in a potentially 1,3-diaxial relationship), but that is not relevant in this paper since we did not explore any hydroxyalkyl azides of these types in this study. In all cases, the 1,3-oxazinane ring is presumed to adopt the most stable chair-like conformation, recognizing that the two chair forms can interconvert either by reversion to oxonium ion and azide or by ring flip between the two forms. c, All known CN migrations in azido-Schmidt reactions entail concerted antiperiplanar migration to the N2 + leaving group 4 . For the spirocyclic intermediates arising from hydroxylalkyl azides, this means that the reaction coordinate necessarily goes through an axial N2 + group, since when this group is equatorial, the only antiperiplanar options are a C-O bond (very unlikely to migrate, as it would form a new, weak N-O bond) or a C-C bond that, if broken, would lead to a cation unstabilized by a neighboring oxygen atom. In general, the N2 + group is axial in the ground-state conformation as revealed by DFT calculations 4 .

Supplementary
Specific cases are discussed on the following pages to illustrate the above principles.

Regioselectivity for 5-substituted steroids
In this section, we provide one example of a reaction of each of three substituted 3-azidopropanol reagents, in both enantiomeric forms, with 3-oxosteroid 2.
Supplementary Figure 3. Proposed mechanism for the regioselectivity of 3-oxosteroid 2 with (R)-3azido-1-phenylpropanol (R)-7. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinanne ring is anchored by the 1-phenyl substituent. This leads to migration of the C2 carbon and affords isomer C2.

Supplementary Figure 4.
Proposed mechanism for the regioselectivity of 3-oxosteroid 2 with (R)-3azido-2-methylpropanol (R)-8. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 2-methyl substituent. This leads to migration of the C2 carbon and affords isomer C4.

Supplementary Figure 5.
Proposed mechanism for the regioselectivity of 3-oxosteroid 2 with (S)-3azido-3-phenylpropanol (S)-9. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 3-phenyl substituent. This leads to migration of the C2 carbon and affords isomer C5. Supplementary Figure 6. Proposed mechanism for the regioselectivity of 3-oxosteroid 2 with (S)-3azido-1-phenylpropanol (S)-7. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 1-phenyl substituent. This leads to migration of the C4 carbon and affords isomer D2.

Supplementary Figure 7.
Proposed mechanism for the regioselectivity of 3-oxosteroid 2 with (S)-3azido-2-methylpropanol (S)-8. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 2-methyl substituent. This leads to migration of the C4 carbon and affords isomer D4.
Supplementary Figure 8. Proposed mechanism for the regioselectivity of 3-oxosteroid 2 with (R)-3azido-3-phenylpropanol (R)-9. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 3-phenyl substituent. This leads to migration of the C4 carbon and affords isomer D5.

Regioselectivity for 5-substituted steroids
In this section, we provide two examples to show how regiochemistry differs in response to changing the steroid configuration at C5; compare these examples with those in Supplementary Figure 3 and 6.
Supplementary Figure 9. Proposed mechanism for the regioselectivity of 3-oxosteroid 3 with (S)-3azido-1-phenylpropanol (S)-7. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 1-phenyl substituent. This leads to migration of the C2 carbon and affords isomer E1.
Supplementary Figure 10. Proposed mechanism for the regioselectivity of 3-oxosteroid 3 with (R)-3azido-1-phenylpropanol (R)-7. In this example, /equatorial attack by the azide is preferred and the conformation of the newly-formed 1,3-oxazinane ring is anchored by the 1-phenyl substituent. This leads to migration of the C4 carbon and affords isomer F1.

Regioselectivity in 17-oxosteroid ring expansions
In this section, we include descriptions of all of the reported approaches to control ring expansions of 17-oxosteroids, namely the Beckmann rearrangement, the intramolecular Schmidt reaction, and the reactions of 17-oxosteroids with 3-azidopropanols (both unsubstituted and substituted). For the last, a detailed description of how we assigned  vs.  azide addition to the ketone has been included. Figure 11. Beckmann rearrangement at C-17 in trans-androsterone 4. The mechanism of the Beckmann rearrangement has long been established 5 and shown to occur through selective reaction of the sterically least hindered oxime stereoisomer (here, the E isomer) and migration of the trans-antiperiplanar bond as depicted. Figure 12. Intramolecular Schmidt reaction. The intramolecular Schmidt reaction of a tethered azido ketone entails formation of the most stable ring fusion upon azide addition to the activated (protonated) ketone and subsequent migration of a C-C bond antiperiplanar to the N2 + leaving group. In these reactions, the leaving group generally occupies an equatorial position and migration of the ring fusion carbon ensues. Under special conditions, axial N2 + is possible, leading to bridged adducts, but these cases are rare 6 . In the case shown, -addition to C-17 is proposed because cis [4.3.0] bicyclic ring systems are generally preferred over the trans version, but the same outcome would be predicted from either direction of attack. Figure 13. Regiochemistry of D-ring expansion reaction with 3-azidopropanol 6. This reaction may occur from a combination of two stereochemical/conformational considerations, neither of which is obvious based on precedent. These are  vs.  attack onto the C-17 oxonium ion and the formation of a particular 1,3-oxazinane ring chair conformation. All possible transformations are assumed to involve antiperiplanar migration to an axial N2 + leaving group as no exceptions to this arrangement are currently known (see general mechanistic comments above).

Supplementary
The newly-formed 1,3-oxazinane ring can exist in two conformations as shown. This spirocyclic center is attached to C16 and C13 of the steroid; the former is ethyl-like and the latter fully substituted, and therefore tert-butyl-like. Accordingly, C13 should occupy an equatorial position in preference to the smaller C16 group as shown in the scheme above.
The matter of  vs.  attack is less obvious because in each case described above there is a stable chair conformation that would lead to the observed stereochemistry, as shown: The literature is silent on the preference of intramolecular spirocycle formation in cases like these (as opposed to intermolecular organometallic additions, which prefer  attack away from the C18 methyl group). However, C-16 migration was observed when (R)-3-azidobutanol (R)-10 was used in the reaction (Supplementary Figure 14), which is only consistent with -attack of the azide (in this case, R = Me in the above diagram). Moreover, the opposite regiochemistry was observed when (S)-3azidobutanol (S)-10 was used in the reaction (Supplementary Figure 15), also only consistent with  attack in that system. For those reasons, we propose that -attack is uniformly observed in all of the spirocycles formed upon attack at C-17 by these reagents.

Supplementary Figure 14.
Proposed mechanism for regioselectivity of 17-oxosteroid 4 with (R)-3azidobutanol (R)-10. In this example, the (R)-methyl group and the placement of the large C-18 carbon into an equatorial substituent clearly favor the conformation of the 1,3-oxazinane shown in G4I and experimentally was shown to lead to isomer G4 resulting from migration of C-16. Together, these facts support the assignment of attack onto the C-17 oxonium ion as shown.    Supplementary Figure 15. Proposed mechanism regioselectivity of 17-oxosteroid 4 with (S)-3azidobutanol (S)-10. In this example, the (S)-methyl group and the placement of the large C-18 carbon into an equatorial substituent clearly favor the conformation of the 1,3-oxazinane shown in H2I and experimentally was shown to lead to isomer H2 resulting from migration of C-18. Together, these facts support the assignment of attack onto the C-17 oxonium ion as shown. 

Caution: Although we have not experienced any untoward events with the compounds mentioned in this paper, azides and their precursors are known explosive hazards and should be used with appropriate safety precautions. Minimally, careful control of temperature and scale should be exercised. We do not recommend distillation of reaction mixtures that may contain residues of azide sources.
Reactions were performed under inert atmosphere (argon or nitrogen). Reactions were carried out in either flame-dried round bottom flasks or glass sample vials (TFE-lined cap). All chemicals were purchased from commercial sources and used without further purification. New containers of BF3OEt2, TfOH, and HFIP were used. Anhydrous CH2Cl2, MeOH, DMF and THF were purchased from Sigma-Aldrich and used as received. Thin-layer chromatography (TLC) was performed using commercial glass-backed silica plates (250 µM) with an organic binder. Visualization was accomplished with UV light, Seebach's stain or aqueous KMnO4 stain and heating. Purification was carried out by an automated flash chromatography/medium-pressure liquid chromatography (MPLC) system using normal phase silica gel flash columns (4, 12, 24, 40, or 80 g).
The infrared (IR) spectra were acquired as thin films using a universal ATR sampling accessory either on a PerkinElmer Spectrum One FT-IR spectrometer, Thermo Scientific Nicolet iS5 FT-IR spectrometer, or Bruker Alpha FT-IR spectrometer; the absorption frequencies are reported in cm -1 . All nuclear magnetic resonance spectra were recorded on a 400 MHz, 500 MHz with a dual carbon/proton cyroprobe, or 600 MHz with a dual carbon/proton cryoprobe instrument; Varian and Bruker instruments were used. NMR samples were recorded in deuterated chloroform (CDCl3) or deuterated dimethylsulfoxide (DMSO-d6). Chemical shifts are reported in parts per million (ppm) and referenced to the center line of solvent (CDCl3: δ 7.26 ppm for 1 H NMR and 77. 16  HRMS data were collected using two instruments. (1) Time-of-flight mass spectrometer (TOF) with an electrospray ion source (ESI). (2) Thermo LTQ Fourier transform ion cyclotron resonance (FT-ICR, 7T) with a heated electrospray ion source (HESI), electrospray ion source (ESI), atmospheric-pressure chemical ionization source (APCI), or atmospheric-pressure photoionization (APPI). Purity data were collected using two instruments. (1) Waters Acquity H-class UPLC-PDA detector coupled to the Thermo LTQ Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR, 7T) with a heated electrospray ion source (HESI). Samples were run on analytical Acquity UPLC BEH 2.1 × 50 mm, 1.7 µm, C18 column, and analytical Acquity UPLC HSS T3, 2.1 × 50 mm, 3.18 µm, C18 column, at 40 ˚C with mobile phases A (H2O + 0.1% formic acid) and B (MeCN + 0.1% formic acid). (2) Agilent 6110 Series LCMS with a UV detector and a single quadrupole mass spectrometer. Samples were run on an analytical Agilent Eclipse Plus 4.6 × 50 mm, 1.8 µm, C18 column at room temperature with mobile phases A (H2O + 0.1% acetic acid) and B (MeOH + 0.1% acetic acid). Melting points were determined in open capillary tubes using OptiMelt, an automated melting point apparatus, and are uncorrected.
Spectroscopic data for known compounds described in the paper match with those reported in the literature. 1 H NMR and 13 C NMR of known compounds are provided in the Supplementary Figures section.

List of known compounds
The following steroid intermediates and substrates: S1 7 , 1 8 , 2 9 , 3 10 , S12 11 , Beck1 12 , S13 13 , S15 11,14 , 5b 15 The following hydroxyalkyl azides: 6 16  Preparation of steroidal ketones (known compounds) 5α-Cholestan-3-ol, S1 7 To a solution of cholesterol (2.03 g, 5.25 mmol) in anhydrous EtOH (55.0 mL, 0.1 M) was added 10% Pd/C (220 mg, 2.07 mmol, 0.40 equiv) under argon. The reaction mixture was degassed and charged with hydrogen. The reaction mixture was stirred under balloon-pressure hydrogen overnight. The reaction mixture was filtered over Celite and concentrated. Purification was carried out by an automated MPLC system using a 40 g normal phase silica column with gradient elution from 0-10% EtOAc/hexanes to afford S1 as a white amorphous solid (1.86 g, 4.79 mmol, 91% yield). Characterization data were consistent with reported data. 5α-Cholestan-3-one, 1 8 To a solution of S1 (1.57 g, 4.04 mmol) in anhydrous CH2Cl2 (50.0 mL) was added Celite and PCC (1.31 g, 6.07 mmol, 1.5 equiv) at 0 ˚C. The reaction mixture was allowed to room temperature and stirred overnight. The reaction mixture was filtered through a short pad of Celite and concentrated. Purification was carried out by an automated MPLC system using a 40 g normal phase silica column with gradient elution from 0-10% EtOAc/hexanes to afford 1 as a white amorphous solid (1.45 g, 3.75 mmol, 93% yield). Characterization data were consistent with reported data; mp 129-130 ˚C. 5α-Dihydrotestosterone, 2 9 Following a literature procedure 20 , 5α-Dihydrotestosterone 2 was prepared as described. To a solution of testosterone (570 mg, 1.98 mmol) in anhydrous THF (6.0 mL) was condensed liquid NH3 (~30 mL) at -78 ˚C. To the cooled solution was added pieces of lithium wire (83.0 mg, 12.0 mmol, 6.0 equiv). The blue solution was stirred at -78 ˚C for 20 min, and was allowed to -35 ˚C and stirred for an additional 2 h. The reaction mixture was quenched by adding solid NH4Cl slowly (until the disappearance of blue). The resulting mixture was diluted with H2O and EtOAc. The organic layer was washed with brine, dried over anhydrous Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 24 g normal phase silica column with gradient elution from 0-30% EtOAc/hexanes to afford 2 as a white amorphous solid (331 mg, 1.14 mmol, 58% yield). Characterization data were consistent with reported data; mp 179-181 ˚C. 5β-Dihydrotestosterone, 3 10 To a solution of testosterone (2.30 g, 8.00 mmol) in anhydrous THF (80.0 mL, 0.1 M) was added 10% Pd/C (341 mg, 3.20 mmol, 0.40 equiv) under argon. The reaction mixture was degassed and charged with hydrogen. The reaction mixture was stirred under balloon-pressure hydrogen overnight. The reaction mixture was filtered over Celite and concentrated. Purification was carried out by an automated MPLC system using an 80 g normal phase silica column with gradient elution from 0-20% EtOAc/hexanes to afford 3 as a colorless amorphous solid (1.51 g, 5.20 mmol, 65% yield) and 2 as a white amorphous solid (414 mg, 1.43 mmol, 18% yield). Characterization data were consistent with reported data.

Preparation of hydroxyalkyl azides
3-Azidopropanol, 6 16 3-Azidopropanol 6 was prepared following a previously published procedure. Characterization data were consistent with reported data.

Synthesis of C9 and D9 via nucleophilic addition of hydride:
To a solution of iminium residue in anhydrous MeOH (2.0 mL) at 0 ˚C was added NaBH4 (30.0 mg, 0.799 mmol, 4.0 equiv) cautiously. The reaction mixture was stirred at room temperature for overnight. The reaction mixture was partitioned between saturated solution of NaHCO3 (5 mL) and CH2Cl2 (20 mL). The organic layer was separated, dried over Na2SO4, filtered, and concentrated. Purification of analogs was carried out by an automated MPLC system on normal phase silica gel flash columns.

Synthesis of C10
and D10 via nucleophilic addition of hydride: 10% Pd/C (10.6-21.0 mg, 0.100-0.197 mmol, 1.0 equiv) was added to a solution of iminium ether in anhydrous EtOH (2.0-4.0 mL). The reaction was stirred under an atmosphere of hydrogen for 24 h. Purification of analogs was carried out by an automated MPLC system on normal phase silica gel flash columns.
General procedure C for the PCC oxidation of A11, B11, C16, and D16: To a slurry solution of A3, B3, C2, or D2 (53.0-93.0 mg, 0.099-0.212 mmol) in anhydrous CH2Cl2 (6.0-10.0 mL) and Celite at 0 ˚C was added PCC (43.0-190.0 mg, 0.200-0.880 mmol, 2.0-4.0 equiv). The brown reaction mixture was allowed to room temperature over 30 min and stirred overnight. The reaction mixture was diluted with CH2Cl2, filtered over Celite and concentrated. Purification of analogs was carried out by an automated MPLC system on normal phase silica gel flash columns.
General procedure D for the optimization of G1: To a solution of trans-androsterone 4 (0.150 mmol, 1.0 equiv) and 3-azidopropanol 6 (0.300-0.450 mmol, 2.0-3.0 equiv) in solvent (0.38 mL, 0.4 M) in a nitrogen-flushed, two-dram vial at room temperature (unless otherwise noted) was added acid catalyst (0.0825-1.05 mmol, 0.55-7.0 equiv) dropwise. The vial was capped, and the reaction mixture was stirred at room temperature for 24 h. The solvent was removed under nitrogen, and an aqueous solution of 15% KOH (3.0 mL) was added to the iminium residue. The reaction mixture was vigorously stirred at room temperature for 24 h. The reaction mixture was diluted with CH2Cl2 (50 mL), dried over anhydrous Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 4 g normal phase silica gel flash column with gradient elution from 0-5% MeOH/CH2Cl2.
Step 2: Addition of nucleophiles: 2.1. Synthesis of G9 and I10 via nucleophilic addition of sulfide: Na2S (195-234 mg, 2.50-3.00 mmol, 10.0 equiv) was added to a solution of iminium ether in anhydrous THF (5.0 mL). The reaction mixture was stirred at 65 ˚C for 24 h. The reaction mixture was diluted with Et2O (20 mL), and was washed with a saturated solution of NH4Cl (3 × 5 mL), brine (5 mL), dried over Na2SO4, filtered, and concentrated. Purification of analogs were carried out by an automated MPLC system on normal phase silica columns.

Synthesis of G10-G11
and I11-I12 via nucleophilic addition of hydride: To a solution of iminium residue in anhydrous MeOH (2.0 mL) at 0 ˚C was added NaBH4 (151 mg, 0.400 mmol, 2.0 equiv) or NaBD4 (167 mg, 0.400 mmol, 2.0 equiv) cautiously. The reaction mixture was stirred at room temperature for 3 h. The reaction mixture was partitioned between saturated solution of NaHCO3 (5 mL) and CH2Cl2 (20 mL). The organic layer was separated, dried over Na2SO4, filtered, and concentrated. Purification of analogs was carried out by an automated MPLC system on normal phase silica gel flash columns.

Synthesis of G12
and I13 via nucleophilic addition of azide: NaN3 (39.0-130 mg, 0.600-2.00 mmol, 4.0 equiv) was added to a solution of iminium ether in anhydrous DMF (2.0-5.0 mL). The reaction mixture was stirred at 70 ˚C for 24 h. The reaction mixture was partitioned between Et2O (40 mL) and H2O (20 mL). The organic layer was washed with H2O (2 × 5 mL), brine (5 mL), dried over Na2SO4, filtered, and concentrated. Purification of analogs was carried out by an automated MPLC system on normal phase silica gel flash columns.

Synthesis of G13-G17 and I14-I16 via nucleophilic addition of para-substituted benzenethiolates:
Sodium thiobenzolate (0.56-0.70 M in DMF, 0.54-0.67 mL, 2.5 equiv) was added to a solution of iminium ether in anhydrous DMF (2.0 mL). The reaction mixture was stirred at 75 ˚C for 24 h. The reaction mixture was partitioned between EtOAc (40 mL) and H2O (15 mL). The organic layer was washed with a saturated solution of NaHCO3 (3 × 5 mL), brine (5 mL), dried over Na2SO4, filtered, and concentrated. Purification of analogs was carried out by an automated MPLC system on normal phase silica gel flash columns.

17β-Hydroxy-5α-androstane-derived A-Ring Lactam, C17
A solution of C16 (87.0 mg, 0.200 mmol) and sodium hydride (60% dispersion in mineral oil, 38.0 mg, 1.60 mmol, 4.8 equiv) in anhydrous THF (8.0 mL) was heated at 65 ˚C for 2 h. The reaction was cooled to room temperature and quenched with a saturated solution of NH4Cl (5 mL). The reaction mixture was diluted with CH2Cl2 (20 mL) and washed with saturated solution of NH4Cl (2 x 5 mL), H2O (5 mL) and brine (5 mL). The combined organic layer was dried over anhydrous Na2SO4 filtered, and concentrated. The crude residue was purified by an automated MPLC system using a 12 g normal phase silica column with gradient elution from 0-5% MeOH/CH2Cl2.
3β-Hydroxy-5α-androstane-derived D-Ring Lactam, Beck1 12 Following the literature procedure 29 , to a solution of S12 (106 mg, 0.349 mmol) in anhydrous THF (7.0 mL, 0.05 M) at 0 ˚C was added SOCl2 (0.25 mL, 10.0 equiv). The reaction mixture was stirred for 2 h at 0 ˚C, and stirred at room temperature overnight. The reaction mixture was terminated with a saturated aqueous solution of NaHCO3 (20 mL) and H2O (15 mL). The aqueous layer was extracted with CH2Cl2, and the combined organic layers were washed with a solution of saturated NaHCO3, H2O, brine, dried over Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 12 g normal phase silica column with gradient elution of 0-5% MeOH/CH2Cl2. Concentration of appropriate fractions afforded product as a cream-colored amorphous solid (53.0 mg, 0.174 mmol,

3β-Hydroxy-5α-androstane-derived D-Ring Lactam, Beck2
Following a literature procedure 29 , to a solution of S14 (139 mg, 0.400 mmol) in anhydrous THF (5.0 mL) at 0 ˚C was added SOCl2 (0.29 mL, 10.0 equiv). The reaction mixture was stirred for 2 h at 0 ˚C, and stirred at room temperature overnight. The reaction mixture was terminated with saturated aqueous solution of NaHCO3 (20 mL) and H2O (15 mL). The aqueous layer was extracted with CH2Cl2, and the combined organic layers were washed with a solution of saturated NaHCO3, H2O, brine, dried over Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 12 g normal phase silica column with gradient elution of 0-2% MeOH/CH2Cl2. Concentration of fractions afforded product as a cream-colored amorphous solid (60. 9

3β-Hydroxy-5α-androstane-derived D-Ring Lactam, Intra1
Following a literature procedure 31 , Intra1 was prepared as described. To a solution of S17 (48.5 mg, 0.099 mmol) in HFIP (0.5 mL) at 0 ˚C under nitrogen atmosphere was added TiCl4 (1.0 M in CH2Cl2, 149 µL, 0.149 mmol, 1.5 equiv). The vial was capped, and the reaction mixture was stirred at room temperature for 24 h. The reaction mixture was concentrated under a stream of nitrogen. The residue was diluted with CH2Cl2, washed with a solution of saturated NH4Cl (2 × 5 mL), solution of saturated NaHCO3 (5 mL), brine (5 mL), dried over Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 12 g normal phase silica column with gradient elution of 0-4% MeOH/CH2Cl2. Concentration of fractions afforded product as a white amorphous solid (28.7

3-Methoxy-1,3,5-estratriene-derived D-Ring Lactam, Intra2
Following a literature procedure 31 , Intra2 was prepared as described. To a solution of S19 (95.7 mg, 0.260 mmol) in HFIP (2.0 mL) at 0 ˚C under nitrogen atmosphere was added TiCl4 (1.0 M in CH2Cl2, 130 µL, 0.130 mmol, 0.5 equiv). The vial was capped, and the reaction mixture was stirred at room temperature for 24 h. The reaction mixture was concentrated under nitrogen. The residue was diluted with CH2Cl2, washed with a solution of saturated NH4Cl (2 × 8 mL), solution of saturated NaHCO3 (8 mL), brine (8 mL), dried over Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 12 g normal phase silica column with gradient elution of 0-70% EtOAc/hexanes. Concentration of fractions afforded product as a white amorphous solid (61.

3-Hydroxy-1,3,5-estratriene-derived D-Ring Lactam, Intra3
To a solution of Intra2 (51.5 mg, 0.152 mmol) in anhydrous CH2Cl2 (5.0 mL) at -78 ˚C was added BBr3 (1.0 M in CH2Cl2, 1.20 mL, 1.20 mmol, 8.0 equiv). The reaction mixture was stirred at -78 ˚C for 1 h, warmed to room temperature over 4 h, and continued stirring at room temperature for 1 h (pinkishorange suspension). The reaction mixture was quenched with two drops of water and MeOH (2 mL). The solvent was removed under nitrogen, the residue was redissolved in MeOH, and loaded on silica gel for purification. Purification was carried out by an automated MPLC system using a 12 g normal phase silica column with gradient elution of 0-3% MeOH/CH2Cl2. Concentration of fractions afforded

3β-Hydroxy-5α-androstane-derived D-Ring Lactam, Beck1
To a solution of H5 (85.0 mg, 0.200 mmol) in anhydrous THF (4.0 mL) was condensed liquid NH3 (~15 mL) at -78 ˚C. To the cooled solution was added pieces of sodium metal until the solution turned blue. The blue solution was stirred at -78˚C for 2 h, and was quenched by adding solid NH4Cl slowly (until the disappearance of blue). The resulting mixture was diluted with H2O (20 mL) and EtOAc (30 mL). The organic layer was washed with brine, dried over anhydrous Na2SO4, filtered, and concentrated. Purification was carried out by an automated MPLC system using a 12 g normal phase silica column with gradient elution from 0-5% MeOH/CH2Cl2 to afford Beck1 as a white amorphous solid (56.8 mg, 0.186 mmol, 93% yield, UPLC/HRMS purity: ≥99.5%). Characterization data were consistent to Beck1. HRMS (FT-ICR, HESI) m/z: [M + H] + calcd for C19H32NO2 306.2428, found 306.2423.