Intramolecular Azide to Alkene Cycloadditions for the Construction of Pyrrolobenzodiazepines and Azetidino-Benzodiazepines

The coupling of proline- and azetidinone-substituted alkenes to 2-azidobenzoic and 2-azidobenzenesulfonic acid gives precursors that undergo intramolecular azide to alkene 1,3-dipolar cycloadditions to give imine-, triazoline- or aziridine-containing pyrrolo[1,4]benzodiazepines (PBDs), pyrrolo[1,2,5]benzothiadiazepines (PBTDs), and azetidino[1,4]benzodiazepines. The imines and aziridines are formed after loss of nitrogen from a triazoline cycloadduct. The PBDs are a potent class of antitumour antibiotics.


Synthesis and Reactivity of the Pyrrolo-Based Systems
The largest and most researched class of PBDs, those represented by abbeymycin (1) and DC-81 (2), are able to interact with nucleophilic guanidine residues in the minor groove of DNA, a process that is reliant upon the presence of an electrophilic imine or carbinolamine based functional group [5]. It is of note that the (S)-chiral center that is present in the 3-dimensional structure of these PBDs gives the molecules a shape which enables them to twist into the DNA minor groove meaning that most syntheses are based upon derivatives of (S)-proline. It is known that intramolecular 1,3-dipolar cycloadditions of azides to alkenes can lead to either imines or aziridines via triazoline intermediates [28,29]. On this basis, we anticipated that we could access imine or aziridine containing PBDs using intramolecular 1,3-dipolar cycloadditions between azides and alkenes using alkenes derived from (S)-proline. Whilst we were interested in the possibility of producing an imine, we were more intrigued by the possibility of producing an aziridine due the known propensity [30] of aziridines to function as electrophiles and the potential biological activity that a novel system of this type might offer. We settled upon two approaches, as shown in Scheme 1.
In the first system, (S)-prolinol (20) was converted into the unstable alkene 24 using a modified literature protocol [31]. Thus, S-prolinol (20) was protected as the carbamate 21, oxidized to the aldehyde 22 and then converted to the N-protected alkene 23 by Wittig reaction. In-situ deprotection and coupling of the deprotected alkene 24 to 2-azidobenzoyl chloride or 2-azidobenzenesulfonyl chloride 27a/b gave the cycloaddition precursors 14 and 15 in 32% and 20% yield, respectively, from the carbamate 23. Heating compound 14 in chloroform gave a ~1:1 mixture of the imine 25 and the aziridine 16 (55% combined yield) whereas heating compound 15 in the same solvent gave the aziridine 17 as the only isolable product in 44% yield.
We assume these reactions proceed via intramolecular 1,3-dipolar cycloaddition to give an intermediate triazoline which then undergoes loss of two nitrogen atoms. It is of note that Broggini explored a similar process with halogenated aryl systems and obtained the triazolines 26 [32]. The presence of the aziridine ring in compounds 16 and 17 was clear in the NMR spectra from the extra CH2 and CH groups and from the diagnostic lack of CH2 geminal coupling in the aziridine CH2.

Scheme 1. Synthesis of aziridinopyrrolobenzodiazepines and aziridinopyrrolobenzothiadiazepines.
Mass spectroscopy confirmed loss of two nitrogen atoms and precluded the triazoline. The relative stereochemical assignment in compounds 16 and 17 was made on the basis of the CH/CH coupling constant at 8-10 Hz, the lack of a CH to CH nOesy correlation and the presence of aziridine CH2 to pyrrolidine CH correlation, and the unequivocal assignment of Broggini's system by X-ray crystallography [32]. Stereochemistry at the aziridine nitrogen was not determined. Although we used (S)-prolinol as the starting material (due to its ease of availability), we did not seek to confirm absolute stereochemical assignments, although it is of note that Sato and co-workers reported that the alkene 24 retains the (S)-configured centre [31].
In our second approach, also shown in Scheme 1, we coupled (S)-prolinol to 2-azidobenzoyl chloride or 2-azidobenzenesulfonyl chloride 27a/b (80%-96%) and oxidized the alcohols 28 and 29 under Swern conditions to give the aldehydes 30 and 31 (72%-80%). Attempts to react the aldehydes 30 and 31 with (methylene)triphenylphosphorane were unsuccessful meaning that a more efficient route (compared to the synthesis and in-situ deprotection of 23) to the alkenes 14 and 15 was not possible. However, reactions of these aldehydes with (carbethoxymethylene)triphenylphosphorane in toluene were successful. In the case of reaction with aldehyde 30, the alkene 32 was isolated in 51% yield and then heated in chloroform, whereupon it converted into the aziridine 33 as a 1:1 mixture of diastereoisomers in 30% yield. In the case of aldehyde 31, the alkene 34 could not be isolated and the aziridine 35 was isolated as the only product in 34% yield from the aldehyde, this time as a single diastereoisomer, possibly due to the lower temperature at which reaction occurred. We assigned the aziridine-CH/pyrrolidine-CH stereochemistry on the same basis as that described above, but were unable to determine the stereochemistry at the CHCO2Et chiral center. We assume that these products arise as a result of azide to alkene cycloaddition and formation of an intermediate triazoline 39, which then undergoes loss of two nitrogen atoms. As discussed previously, this loss of nitrogen is known behavior in alkene to azide cycloaddition processes [28,29]. We also studied reactions involving (2-azido-4-benzyloxy-5-methoxy)benzoyl chloride 27c (X = CO, R 1 = OMe, R 2 = OBn) due to this having the substitution pattern present in the natural product DC-81 (2). This azide was available in six steps from commercially available 4-hydroxy-3-methoxybenzoic acid using a literature procedure [33,34]. Attempts to couple this acid chloride to the alkene 24 were unsuccessful and led to significant degradation of the azide staring material-possibly as a result of its intolerance to the extreme in-situ conditions under which the alkene 24 was generated [31]. However, coupling of the acid chloride 27c to prolinol was successful and subsequent oxidation gave the aldehyde 37. Attempted reaction of this aldehyde with (methylene)triphenylphosphorane was again unsuccessful. However, reaction with (carbethoxymethylene)triphenylphosphorane gave a new product. As was observed previously with aldehyde 31, the alkene 38 was not isolable. This time, however, the product was unexpectedly found to be the "reduced" pyrrolobenzodiazepine 40, formed as a single diastereoisomer in 21% yield from the aldehyde, with stereochemistry consistent with that discussed above (no CH to CH correlation by nOesy and a strong CH2CO2Et to pyrrolidine CH correlation). We assume that this product arises as a result of a free-radical based loss of nitrogen from the triazoline 39 followed by hydrogen abstraction from the toluene solvent rather than imine/aziridine formation as observed previously.

Synthesis and Reactivity of the Azetidino Based Systems
In order to obtain further examples of these cycloaddition processes, we next turned our attention to the use of 4-alkenyl-2-azetidinones (Scheme 2) as coupling partners for 2-azidobenzoyl chloride (27a). We chose azetidinone systems due to our long-standing interest in the chemistry and biological applications of the β-lactams and their derivatives [35,36], and also due to the lack of literature examples of azetidino analogues of the PBDs. The 4-alkenyl-2-azetidinones 41 were synthesized in 58%-60% yield using a [2+2]-cycloaddition between a diene and chlorosulfonyl isocyanate followed by work-up to remove the N-sulfonyl group. Whilst the 4-methyl substituted β-lactam 41a coupled successfully (65% yield) to give the 1-(2'-azidobenzoyl)-2-azetidinone 42a, we were unable to obtain the corresponding demethyl system 42b from the β-lactam 41b. We thus converted the azetidinones 41a/b into the corresponding azetidinethiones 43a/b by reaction with Lawesson's reagent, reasoning that this may result in a nitrogen atom that was more nucleophilic, and in fact, the two azetidin-2-thiones 43a/b reacted successfully with 2-azidobenzoyl chloride (27a) to give the 1-(2'-azidobenzoyl)-2-azetidinthiones 44a/b in 84% and 71% yield, respectively. Each of the three azido alkenes 42a and 44a/b behaved differently when heated to reflux in boiling solvents. Compound 42a gave the imine 45 (38%) when heated in boiling chloroform for 72 h; compound 44a when heated in boiling chloroform also gave the corresponding imine 47, but this time formed the isolable triazoline 46 (61% yield) after 72 h and gave the imine in 32% yield after a further 7 days in boiling chloroform; compound 44b was stable in chloroform at reflux, but gave the aziridine 48 (48% yield) upon 24 h of heating in toluene at reflux. The aziridine ring in compound 48 was apparent from the CH2CHCHCH2 connectivity in the system, the distinctive lack of geminal proton coupling for the aziridine CH2 and the loss of two nitrogen atoms in the mass spectrum. The product was formed as a single diastereoisomer according to the 1 H/ 13 C-NMR spectra. Stereochemistry at the aziridine nitrogen was not determined. The CH/CH stereochemistry was assigned by correlation to the systems described above on the basis of the similar 8.9 Hz coupling constant.

General Information
Reactions were performed in oven-dried glassware under nitrogen dried through 4 Å molecular sieves and delivered through a gas manifold. Work-up procedures were carried out in air. Anhydrous grade solvents were freshly distilled using a continuous still under nitrogen. Chloroform was dried over 4 Å molecular sieves or distilled over phosphorus pentoxide (3% w/v). Dichloromethane, acetonitrile, ethyl acetate and toluene were distilled over calcium hydride (5% w/v) over 4-6 h. Diethyl ether and THF were pre-dried over sodium wire, and then distilled over sodium wire (1%-2% w/v) with benzophenone (0.2%-0.3% w/v) as an indicator. Other anhydrous solvents were purchased from Fisher Chemicals or Sigma-Aldrich. All reactions were monitored by TLC, which was carried out on 0.20 mm Macherey-Nagel Alugram ® Sil G/UV254 silica gel-60 F254 precoated aluminium plates and visualisation was achieved using UV light or vanillin stain. Column chromatography was performed on Merck silica gel (0.063-0.200 mm, 60 Å). NMR spectra were recorded on a Bruker DPX-400 or on a Bruker Avance 500 instrument. IR spectra were recorded on a Nicolet 380 FT-IR instrument as a drop for oils and liquids or as neat powders for solids. Mass spectra were recorded on a Bruker Daltonics micrOTOF mass spectrometer operating at a positive ion mode under an electrospray ionization (ESI +) method. High resolution mass spectra were recorded on a Finnegan MAT 900 XTL instrument operated by the EPSRC National Mass Spectrometry service at the University of Swansea. 1-(2'-Azidobenzoyl)pyrrolidine-2carbaldehyde [33,34], 2-azidobenzenesulfonic acid [37] and 2'-azido-4-(benzyloxy)-5-methoxybenzoic acid [33,34] were synthesized as described in the literature. The synthesis of N-2-ethenyl-1ethoxycarbonylpyrrolidine (23) and its deprotection have been described previously [31], but full details are included below as we found some modification was necessary. Due to ease of availability, we started our synthesis with (S)-prolinol, but we did not seek to confirm the absolute stereochemical integrity of  [27]. Similarly, we have described the synthesis and use of compounds 41 and 43 before as precursors in other chemistry [36], but have not before given detailed experimental procedures. Compounds 32-35, 38, 40, 42 and 44-48 are previously unreported.

Synthesis and Reactivity of the Pyrrolo-Based Systems
N-(Ethoxycarbonyl)-prolinol (21). To a stirring solution of S-prolinol (20, 1.00 g, 9.89 mmol) in 4 M NaOH (7 mL) was added ethyl chloroformate (1.13 mL, 1.29 g, 11.9 mmol) over 10 min at 0 °C. The reaction was stirred at 0 °C for 30 min followed by 30 min at ambient temperature. The reaction solution was neutralized with 2M HCl, the aqueous phase was separated and extracted with DCM (3 × 10 mL). The combined organic layers were dried (MgSO4), filtered and concentrated under reduced pressure to give the crude product 21 (1.64 g, 96%) as a yellow oil which was used directly in the next step. NMR: 02 mmol) in anhydrous THF (5 mL) was added dropwise over 10 min. The whole was allowed to warm to −10 °C, and maintained at that temperature for 2 h and then the mixture was allowed to reach room temperature overnight. The mixture was quenched with saturated aqueous NH4Cl (20 mL) and the aqueous layer was separated and extracted with EtOAc (3 × 10 mL). The combined organic layers were dried (MgSO4), filtered, concentrated under reduced pressure and purified by silica chromatography (50 g) (EtOAc/Hex; 3:2) yielding an air sensitive product as a mixture of rotamers in the form of a pale orange oil (536 mg, 45%  Aziridinopyrrolobenzothiadiazepine (17). The N-(2'-azidobenzenesulfonyl)-2-ethenylpyrrolidine (15) (125 mg, 0.449 mmol) was dissolved in toluene (10 mL) and heated to reflux for 18 h. The reaction was allowed to reach room temperature before the solvent was removed and the crude was purified by silica chromatography (25 g

Synthesis and Reactivity of the Azetidino-Based Systems
To a stirred solution of isoprene (2.33 g, 3.43 mL, 34.41 mmol) in dry diethyl ether (15 mL) at -78 °C was added a solution of chlorosulfonyl isocyanate (4.88 g, 3.01 mL, 34.01 mmol) in dry ether (10 mL) dropwise over one hour. The reaction mixture was allowed to warm to −10) dropwise over one hour. The reaction mixture was allowed to warm to −10 °C and then the reaction flask was transferred to an ice-salt bath and stirred for 30 min. The cooled solution was added dropwise to a vigorously stirred solution of water (50 mL), sodium carbonate (9.00 g), sodium sulfite (6.01 g) and ice (30 g) over 10 min. The mixture was stirred at −10 °C for 1 h and then allowed to warm to room temperature and extracted with diethyl ether (6 × 20 mL). The combined organic extracts were dried (MgSO4) and the solvent removed under reduced pressure to give the product as a pale yellow  Methyl-4-ethenyl-1-azetidin-2-thione (43a). To the azetidin-2-one (0.55 g, 4.97 mmol) in dry THF (15 mL) was added Lawesson's reagent (1.00 g, 2.48 mmol) and the whole was stirred at room temperature for an hour before being heated to reflux for two h. The reaction was cooled to ambient temperature before being concentrated under reduced pressure and purified by silica chromatography (75 g) (EtOAc/petroleum ether; 1:3) to give the product as a yellow oil (0.35 g, 56%). The reaction was higher yielding (up to 70%) on a larger scale (2 g of lactam), but less convenient to purify (stench). NMR: δH (400 MHz, 4-Ethenyl-2-azetidinone (41b). To 1,3-butadiene (10 mL) condensed into anhydrous ether (40 mL) at −10 °C, was added a solution of chlorosulfonyl isocyanate (3.0 mL, 34.5 mmol) in anhydrous ether (10 mL) over one hour, under an atmosphere of nitrogen. The temperature was maintained at −10 °C for a further period of 3 h and warmed slowly to room temperature overnight, to produce a clear, yellow solution. The solution was added to an ice cold mixture of water (70 mL), ice (30 g), NaHCO3 (9.0 g) and Na2SO3 (6.0 g) and stirred for one hour at −10 °C. The reaction was allowed to warm to room temperature before being extracted with ether (6 × 20 mL). The combined organic layers were dried over anhydrous sodium sulfate, filtered and evaporated to dryness under reduced pressure to yield 4-ethenyl-2-azetidinone (2.01 g, 60% yield) as a clear yellow oil

4-Ethenylazetidin-2-thione (43b).
To the lactam (1.00 g, 10.3 mmol) dissolved in anhydrous THF (15 mL), was added Lawesson's reagent (2.08 g, 5.15 mmol) in one portion, with stirring. The reaction was stirred at ambient temperature, under an inert atmosphere of dry nitrogen for one hour, before being heated at reflux for an additional hour. The reaction was monitored for completion by tlc and was subsequently cooled to room temperature, to give a crude product as a clear orange liquid.

Author Contributions
Hemming designed the project and is the principal and corresponding author and wrote the text. Chambers, O'Gorman and Jamshaid conducted the practical chemistry shown in Schemes 1 and 2 and provided the experimental procedures and characterization data.