Synthesis of chiral 1,4-oxazepane-5-carboxylic acids from polymer-supported homoserine

The preparation of novel 1,4-oxazepane-5-carboxylic acids bearing two stereocenters is reported in this article. Fmoc-HSe(TBDMS)-OH immobilized on Wang resin was reacted with different nitrobenzenesulfonyl chlorides and alkylated with 2-bromoacetophenones to yield N-phenacyl nitrobenzenesulfonamides. Their cleavage from the polymer support using trifluoroacetic acid (TFA) led to the removal of the silyl protective group followed by spontaneous lactonization. In contrast, TFA/triethylsilane (Et3SiH)-mediated cleavage yielded 1,4-oxazepane derivatives as a mixture of inseparable diastereomers. The regioselectivity/stereoselectivity depended on the substitution of the starting 2-bromoacetophenones and was studied in detail. Catalytic hydrogenation of the nitro group improved the separability of the resulting diastereomeric anilines, which allowed us to isolate and fully characterize the major isomers.


Introduction
Chiral seven-membered heterocycles bearing one or more heteroatoms in their skeleton are an interesting group of compounds with unique physico-chemical and biological properties. The prominent heterocyclic scaffold within this group is represented by 1,4-oxazepanes, which occur in both synthetic compounds and natural products (Fig. 1). [1][2][3][4][5][6] Compounds bearing 1,4-oxazepane scaffolds have been reported as potent anticonvulsants 7 and antifungal agents 5,8 or agents to treat inammatory bowel disease, 9 lupus nephritis 10 and respiratory diseases, including asthma and bronchiectasis. 11,12 Over the past decade, synthetic chemists have struggled to develop different strategies to access 1,4-oxazepanes from various starting materials.
The most robust synthetic approaches reported to date are based on intramolecular cyclization of alkenols, 13,14 alkynols 15,16 or hydroxyketones, 16 typically using Brönsted or Lewis acids; however, some alternative methods of limited applicability have also been described recently. [17][18][19][20][21][22][23][24][25] Although the synthetic availability of chiral 1,4-oxazepanes has already been determined, the decoration of the scaffold with reactive functional groups amenable to further diversication remains a challenging task due to the limited applicability of previously developed procedures for functionalizing starting materials. In our previous contribution, we reported the simple synthesis of chiral morpholines starting from resin-bound serine. 26 Using either TFA-or TFA/Et 3 SiH-induced cleavage of the corresponding polymer-supported intermediates (Fig. 2), we synthesized either dihydrooxazine-3-carboxylic acids or morpholine-3-carboxylic acids with full control of the newly formed stereocenter. Later, we extended this method to the simple synthesis of fused [6 + 7] 27,28 or [6 + 6] 29,30 morpholines. To eventually synthesize the corresponding homological compounds, we decided to use polymer-supported homoserine to access the 1,4-oxazepane-5-carboxylic acids suitable for further modication. In this article, we report on the applicability, regioselectivity and stereoselectivity of the proposed method. Fig. 1 Selected pharmacologically relevant compounds bearing a 1,4oxazepane scaffold. [1][2][3][4][5][6] Results and discussion Fmoc-HSe(TBDMS)-OH was prepared from homoserine in two steps 31,32 and immobilized on Wang resin using the 1-hydroxybenzotriazole (HOBt)/diisopropylcarbodiimide (DIC) technique to suppress racemization. The key intermediate 3a was synthesized according to our previous protocols 26 consisting of Fmoc-protective group cleavage, reaction with 2-nitrobenzenesulfonyl chloride (2-Ns-Cl) and alkylation using 2-bromoacetophenone (Scheme 1). Inspired by the smooth cyclization of serine-based analogs to 3,4-dihydro-1,4-oxazine-3carboxylic acids, 26 we hypothesized that exposure of resin 3a to TFA could yield the corresponding homologous product, i.e., 4,5,6,7-tetrahydro-1,4-oxazepine-5-carboxylic acid 4a. The reaction yielded a single product with 87% crude purity (calculated from HPLC-UV traces at 205-400 nm) and 74% overall yield (calculated from the 1 H NMR spectrum of the puried product). Although HRMS analysis corresponded to the molecular mass of suggested product 4a, NMR analysis (see ESI for details †) revealed preferential lactonization, which yielded compound 5a. Interestingly, when Et 3 SiH was added to the cleavage cocktail, a different course of reaction was observed. We received two chromatographically inseparable compounds with identical molecular masses (as indicated by HPLC-UV-MS analysis, the combined crude purity was 85%). To eventually improve the separability, we performed catalytic hydrogenation using palladium on carbon (Pd/C) in 2-isopropanol (IPA), 28 which again afforded a mixture of two isomers (the combined crude purity was 91%, and the ratio was 56 : 44, as calculated from HPLC-UV traces at 205-400 nm); however, in this case, reversephase chromatography (RP-HPLC) indicated possible separation. Consequently, the major isomer was successfully isolated using semipreparative RP-HPLC at 34% overall yield (calculated from the 1 H NMR spectrum of the puried product) and subjected to detailed NMR investigation.
We recorded and analyzed 1 H, 13  HMBC NMR data to determine the constitution. Complete assignment of the 1 H, 13 C and 15 N signals was possible and is shown in the ESI (Fig. S17-S19 and Table S2 †). In brief, by means of the homonuclear and heteronuclear correlation data, we identied the 1,4-oxazepane-5-carboxylic acid, phenyl and 2aminobenzenesulfonyl moieties indicating the structure of 7a. The connectivity between the oxazepane and phenyl rings was conrmed by three long-range 1 H-13 C correlations (see ESI, Fig. S18 †). Finally, the planar structure was established by the 1 H-1 H NOESY spectrum, which gave the key correlations between oxazepane protons and aminobenzenesulfonyl proton H 19 and correlations between oxazepane protons and phenyl protons H 9,13 ( Fig. 3 and S19 †). All the 1D and 2D NMR spectra can be found in the ESI (Fig. S20-S27 †).
To determine the conformation and relative conguration of 7a, we analyzed the 1 H-1 H coupling constants on the oxazepane ring and the NOE correlations in detail. Despite the spatial exibility of seven membered rings leading to multiple possible conformational states, the analysis of vicinal 1 H-1 H couplings indicated that the scaffold existed in the most energetically favourable chair conformation (Fig. 3). Since the conguration of the C5 stereocenter was dened by the conguration of the starting material (S), the conguration of the newly formed C2 stereocenter was assigned as R.  Although we did not isolate and analyze the minor isomer of 7a, in the case of 7b, RP-HPLC purication enabled the separation and isolation of both isomers. Thorough NMR structural analysis indicated the formation of C2 R,S diastereomers 7b 2R and 7b 2S ( Table 1). The 1 H NMR spectrum of 7b 2R was fairly similar to the spectrum of 7a; hence, we concluded that this compound had the same conguration as 7a. Compound 7b 2S was analyzed by 1D and 2D NMR spectroscopy (see ESI,  Table S3 †) to verify the planar structure, assign all the 1 H and 13 C signals and determine the relative conguration. Finally, we conrmed the conguration of 7b 2S by analyzing the vicinal 1 H-1 H couplings on the oxazepane ring and NOE correlations (Fig. 4). Table 1 summarizes the 1 H chemical shis, splitting patterns, and 2 J and 3 J homonuclear couplings observed in 7b 2R and 7b 2S . While the chemical shis differ rather insignicantly between the two diastereomers, the splitting pattern of signal H 5 and 3 J(H 5 -H a 6 ) and 3 J(H 5 -H b 6 ) can be used to differentiate between 7b 2R and 7b 2S . In the C2 R isomer 7b 2R , two relatively large couplings and one very small long-range coupling leading to a doublet of doublet of doublets are observed. However, in the case of the C2 S isomer 7b 2S , the vicinal coupling constants are smaller and equal or nearly equal, and hence, the signal takes the shape of a pseudotriplet (dd). The same trend was observed in all derivatives 6-7.
Aer proving the structure and formation of the two diastereomers, we tried to improve the stereoselectivity by using a lower reaction temperature (0 C or À20 C); however, a nearly equal ratio of isomers was obtained. Furthermore, we tested the use of TMSOTf/Et 3 SiH as reported earlier for an analogical starting material, 17 but a complex mixture of unknown compounds was obtained.
To explain the different reaction outcomes depending on the composition of the cleavage cocktail, we suggested a reaction mechanism (Scheme 2). The protonation of intermediate 3a cleaved from the resin can be followed by intramolecular attack of ketone or carboxylic acid with the hydroxy group as the internal nucleophile. With respect to the higher electrophilicity of the ketone, we presume that the formation of intermediate B over E is preferred. Intermediate B can be further stabilized by the formation of intermediates C and D; however, due to their limited stability, all the reactions in pathway A-D are reversible, which can lead to regeneration of starting material A. The same is true for the formation of E from A. In contrast, the conversion of E to 5a is irreversible, as lactone does not undergo hydrolysis under the conditions used. In the presence of Et 3 SiH, the preferential formation of B is considered again; however, the subsequently formed intermediates C and D are attacked by triethylsilane as the external nucleophile, which leads to the formation of stable compound 6a. Consequently, the formation of lactone 5a was not observed in this case.
Aer the determination of the reaction outcome, we used different starting materials to evaluate the limitations and scope of the methodology and to reveal the structure-regioselectivity and structure-stereoselectivity relationships. For this purpose, Wang resin was replaced with Rink amide resin,  Wang-piperazine resin and BAL resin with immobilized propylamine 26 to alter the carboxylic group to carboxamides. Furthermore, variously substituted sulfonyl chlorides and fourteen 2-bromoacetophenones bearing electron-donating or electron-withdrawing groups were selected, including one heterocyclic derivative (thienyl) and another aliphatic derivative (Me; Fig. 5).
First, we tested the combination of 2-Ns-Cl and 2-bromoacetophenone with different resins. The use of Rink amide resin (R 1 ¼ NH 2 ) required repetition of the alkylation step and yielded the nal aniline 7b in 89% crude purity as a mixture of C2 R,S diastereomers in a ratio of 45 : 55. Its RP-HPLC enabled the separation of both isomers in 9% (C2 R isomer) and 4% overall yields (C2 S isomer, Table 2). In the case of the Wang-piperazine resin (R 1 ¼ piperazin-1-yl), the TFA/Et 3 SiH-mediated cleavage of the corresponding sulfonamide from the resin unexpectedly caused the cleavage of the piperazine moiety, and oxazepane-5carboxylic acid derivative 7a was obtained as a mixture of C2 R,S isomers in a ratio of 72 : 28. The major diastereomer was isolated in 10% overall yield. The use of BAL resin-immobilized propylamine failed in the alkylation step stage, which is in accordance with our previous results. 26 Aer that, various sulfonylating agents (Fig. 5) were tested, starting from intermediate 2 and using 2-bromoacetophenone as the alkylating agent. In the case of sulfonamides 3e and 3g bearing 4-methoxy-2-nitrobenzenesulfonyl group or tosyl group as R 2 , the alkylation required a longer reaction time (2 and 6 days, respectively) to completion. The TFA/Et 3 SiH cleavage of tosyl intermediate 3g yielded the desired oxazepane in 73% crude purity as a partially separable mixture of C2 R,S isomers in a ratio of 62 : 38. Its RP-HPLC purication yielded the major isomer C2 R in 18% overall yield. In the case of 3e and 3f bearing 4-methoxy and 4-chloro-2-nitrobenzene-sulfonyl as R 2 substituents, the nitro-oxazepane derivatives 6e and 6f were obtained as inseparable mixtures of C2 R,S diastereoisomers in 81-90% crude purities. Similar to previously reported results, 28 the hydrogenation of 6f led to undesired hydrogenolysis of the C-Cl bond (Scheme 3). For this reason, Pd/C was replaced with PtO 2 , which yielded aniline 7f in 86% crude purity as a mixture of C2 R,S diastereomers in a ratio of 70 : 30. In this case, the isolation of the major isomer from the diastereomeric mixture was Scheme 2 The hypothetical mechanism explaining the different reactivity of intermediate 3a.  Table 2 for the list of target compounds).
This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 35906-35916 | 35909 problematic and furnished the product C2 R in only 78% diastereomeric purity. In the case of 6e bearing a 4-methoxy group, hydrogenation using PtO 2 suppressed the previously reported demethylation, 28 and the nal compound 7e was obtained in 91% crude purity as a mixture of C2 R,S diastereomers in a ratio of 69 : 31. RP-HPLC purication enabled the separation of the major C2 R isomer in 21% overall yield.
Finally, we tested various 2-haloketones bearing electrondonating and electron-withdrawing groups in the o-, m-and p-positions, and one heterocyclic and aliphatic derivative was included. All these building blocks were tested in combination with intermediate 2 and 2-Ns-Cl. In the case of 7j and 7p bearing 2-Br-Ph or 4-Br-Ph as R 3 , PtO 2 had to be used to avoid undesired debromination. The preferential formation of oxazepane was observed in each case. In the case of 7p, inseparable C2 R,S diastereomers were isolated in 14% overall yield as diastereomeric anilines in a ratio of 42 : 58. In the case of 3i bearing 2-F-Ph as R 3 , the regioselectivity was compromised, and the TFA/ Et 3 SiH reaction yielded 1,4-oxazepane 6i accompanied by lactone 5i (26% according to HPLC). The following hydrogenation of the reaction mixture yielded the corresponding diastereomeric aniline 7i (69% combined crude purity) and aminolactone 8i, which spontaneously cyclized to benzothiadiazepine 1,1-dioxide 9i (26% according to HPLC-UV-MS analysis) (Scheme 4). 33 In this case, RP-HPLC purication enabled the separation of both diastereomers. Intermediates 3h (R 3 ¼ 2-Me-Ph) and 3j (R 3 ¼ 2-Br-Ph) yielded lactones 5h and 5j as the major products (72% and 64% crude purities). Their hydrogenation yielded the corresponding anilines 8h and 8j, which were cyclized to benzothiadiazepine 1,1-dioxides 9h and 9j. 33 Derivatives 9h and 9j were isolated using semipreparative RP-HPLC at 19% overall yield and fully characterized. Interestingly, in contrast to previously reported results, 33 compounds 9 were fully stable and did not undergo ring contraction to benzothiadiazine 1,1-dioxides.
In the case of 3k (3-F-Ph as R 3 ), the TFA/Et 3 SiH reaction and the subsequent hydrogenation yielded diastereomeric aniline 7k in a ratio of 64 : 36 and 70% combined crude purity, and the major isomer was isolated and fully characterized. On the other hand, intermediate 3l (3-MeO-Ph as R 3 ) yielded a mixture of oxazepane derivative 6l and lactone 11l in a ratio of approximately 1 : 1, and for this reason, the products were not isolated (Scheme 6). Intermediates 3m, 3o-q, t were synthesized from psubstituted 2-bromoacetophenones (4-Me-Ph, 4-F-Ph, 4-Br-Ph and 4-CF 3 -Ph as R 3 ) and 3-thienyl derivative yielded oxazepanes 7m, 7o-q, t as separable C2 R,S diastereomers in variable ratios ( Table 2) with combined crude purities in a range of 33-91%. In the case of oxazepane derivative 6r (4-NO 2 -Ph as R 3 ), hydrogenation using either Pd/C or PtO 2 led to cleavage of the heterocyclic scaffold, and compound 10r was isolated (Scheme 5) in 82% crude purity and 50% overall yield (Table 2).
In the cases of 3n (4-MeO-Ph as R 3 ) and 3s (4-NH 2 -3,5-diCl-Ph as R 3 ), we observed reversed regioselectivity, and aer TFA/ Et 3 SiH cleavage, lactone derivatives 11n and 11s were formed exclusively; however, 11s was accompanied with oxazepane 6s as the minor product (30% crude purity). Due to the excess triethylsilane in the reaction mixture, the abovementioned derivatives were obtained as phenylethyl derivatives 11 (Scheme 6). The exposure to TFA/Et 3 SiH had to be prolonged to make the reduction quantitative. In combination with the outcome obtained from intermediate 3l (3-MeO-Ph as R 3 ), we can state that electron-donating R 3 groups in the m-or p-positions contribute to the formation of lactones, as they probably diminish the reactivity of adjacent ketones toward nucleophilic addition, which suppresses the formation of intermediate C (Scheme 2).
Similar to previously reported results, 34 the alkylation step with 2-chloroacetone provided the corresponding oxazepane 6u in only 50% conversion. For this reason, the product was not isolated.

Conclusions
To conclude, we developed a simple methodology to prepare 2phenyl-substituted-1,4-oxazepane-5-carboxylic acid derivatives. Although the formation of the oxazepane scaffold was nonstereoselective, as in the case of serine-based analogs leading to chiral morpholines, 26 the separation and full characterization of major diastereomers was feasible. The developed strategy provided rather minor limitations, e.g., competitive lactonization in the cases of m-and p-electron-donating groups as R 3 substituents. Importantly, the developed protocols and corresponding intermediates can be applied for the synthesis of differently fused 1,4-oxazepanes based on previously reported approaches targeted to fused morpholines. [27][28][29][30] Experimental section

General information
Solvents and chemicals were purchased from Sigma-Aldrich (Milwaukee, WI, http://www.sigmaaldrich.com), Acros Organic (Geel, Belgium, http://www.acros.com) and Fluorochem (Had-eld, United Kingdom, http://www.uorochem.co.uk). Wang resin (100-200 mesh, 1% DVB, 1.4 mmol g À1 ), Rink resin (100-200 mesh, 1% DVB, 0.4 mmol g À1 ) and aminomethyl resin (100-200 mesh, 1% DVB, 0.98 mmol g À1 ) were obtained from AAPPTec (Louisville, KY, http://www.aapptec.com). Solid-phase synthesis was carried out in plastic reaction vessels (syringes, each equipped with a porous disk) using a manually operated synthesizer (Torviq, Niles, MI, http://www.torviq.com). All reactions were carried out at ambient temperature (23 C) unless stated otherwise. The synthesis of Fmoc-HSe(TBDMS)-OH, 31,32 immobilization of Fmoc-HSe(TBDMS)-OH on the resin 26 and N-phenacyl sulfonamides 3a-u 26 was performed according to these reported protocols. The LC-MS analyses were carried out on UHPLC-MS system consisting of UHPLC chromatograph Acquity with photodiode array detector and single quadrupole mass spectrometer (Waters), using X-Select C18 column with the mobile phase consisting of 10 mM ammonium acetate (AmAc) in H 2 O and MeCN. The ESI source operated at discharge current of 5 mA, vaporizer temperature of 350 C and capillary temperature of 200 C. For the LC/MS analysis, a sample of resin ($5 mg) was treated with TFA in CH 2 Cl 2 , the cleavage cocktail was evaporated under a stream of nitrogen, and cleaved compounds extracted into MeCN/H 2 O (20% or 50%; 1 mL). Purication was carried out on C18 semipreparative RP-HPLC with the gradient of 10 mM aqueous AmAc and MeCN, ow rate 15 mL min À1 or by normal phase by silica gel chromatography. Residual solvents (H 2 O and AmAc buffer) were lyophilized by the ScanVac Coolsafe 110-4 working at À110 C. All 1D and 2D NMR experiments were performed with using ECX500 spectrometer (JEOL RESONANCE, Tokyo, Japan) at magnetic eld strength of 11.75 T corresponding to 1 H and 13 C resonance frequencies of 500.16 MHz and 125.77 MHz at 27 C. Chemical shis (d) are reported in parts per million (ppm) and coupling constants (J) are reported in Hertz (Hz). The signals of MeCN-d 3 were set at 1.94 ppm in 1 H NMR spectra and at 118.26 ppm in 13 C NMR spectra. 15 N chemical shis were referenced to external 90% formamide in DMSO-d 6 at 112.00 ppm. 35 The assignment of 1 H, 13 13 C NMR spectra. HRMS analysis was performed using LC-MS (Dionex Ultimate 3000, Thermo Fischer Scientic, MA, USA) with Exactive Plus Orbitrap high-resolution mass spectrometer (Thermo Exactive Plus, Thermo Fischer Scientic, MA, USA) operating at positive or negative full scan mode (120 000 FWMH) in the range of 100-1000 m/z with electrospray ionization working at 150 C and the source voltage of 3.6 kV. Chromatographic separation was performed on column Phenomenex Gemini (C18, 50 Â 2 mm, 3 mm particle) with isocratic elution and mobile phase (MP) containing MeCN/ 10 mM AmAc (80 : 20; v/v). The samples were dissolved in the initial MP. The acquired data were internally calibrated with phthalate as a contaminant in MeOH (m/z 297.15909). IR spectra were measured by DRIFT (Diffuse Reectance Infrared Fourier Transform) on a Thermo Nicolet AVATAR 370 FTIR spectrometer. Absorbance peaks (wavenumbers) are reported in reciprocal centimeters (cm À1 ) and transmittances (T) are reported in percentages (%). Specic optical rotations were measured on Automatic Compact Polarimeter POL-1/2 (ATAGO, Japan) with LED Light Source and 589 nm interference lter at 24 C. The length of cuvette was 2 cm and specic optical rotations are reported as follows: [a] T D , concentration (g mL À1 ) and solvent.
General method for calculation of yields using 1 H NMR. 1 H NMR spectra of external standard at three different concentration levels were measured. In each spectrum, the solvent signal was integrated followed by the integration of selected H Ar signal of external standard. Ratios of solvent/ standard signal areas along with known quantity of standard were used to construct a calibration curve. Then, 1 H NMR spectra of studied sample were measured and the ratio of solvent/sample (selected H Ar signal) areas was determined. Using the calibration curve, the quantity of compound in the sample was calculated.
General procedure for cleavage the TBDMS protecting group and lactonization 5a-u. The polymer-supported intermediate 3a-u (500 mg) was cleaved in the mixture of TFA/CH 2 Cl 2 (5 mL, 50%) for 1 h at room temperature. Then the resin was washed three times with fresh cleavage cocktail (5 mL) and the combined fractions were evaporated using a stream of nitrogen, lyophilized overnight and puried by semipreparative RP HPLC. General procedure for cyclization to 1,4-oxazepanes 6a-u. The polymer-supported intermediate 3a-u (500 mg) was cleaved in TFA/Et 3 SiH/CH 2 Cl 2 (5 mL, 10 : 1 : 9) for 30 min (except for 3n and 3s) or 24 h (for derivatives 3n and 3s) at room temperature. Then the resin was washed three times with fresh cleavage cocktail (5 mL) and the combined fractions were evaporated using a stream of nitrogen and lyophilized overnight. The crude products were puried using RP-HPLC.