Stereoselective synthesis and structural analysis of polycyclic lactams derived from tetrahydroisoquinoline 1,2-and 1,3-diamines

.


Introduction
The cyclocondensation of various 1,2-difunctional compounds (1, amino alcohols, amino thiols or diamines, containing a primary amino group) comprise a well-established method for the preparation of 1,3-heterocycle-fused or -lactams with a nitrogen at the annelation (4). 1 The reaction is classified as a domino process, 2 since it presumably occurs in two steps: first a ringchain tautomeric intermediate (3A-3B) is formed, the equilibrium of which gradually shifts towards the cyclic form 4 in consequence of the practically irreversible intramolecular Nacylation. 3 The nitrogen-bridged or -lactams 4, both in racemic and in enantiomerically pure form, are often applied as intermediates in numerous synthetic processes involving various regioand stereocontrolled transformations. 4e recently described the domino ring-closures of 1-(aminomethyl)-and 1-(2-aminoethyl)-6,7-dimethoxy-1,2,3,4-tetrahydroisoquinoline with and -oxo acids, whereby tetra-and pentacyclic either angularly or linearly tetrahydroisoquinoline-condensed lactam derivatives were formed with excellent diastereoselectivities. 5 As a continuation of this work and in connection with our previous studies on the synthesis and structural analysis of tetrahydroisoquinolinecondensed saturated heterocycles, 6 we now report on the reactions of 1-substituted 1,2,3,4tetrahydroisoquinoline 1,2-and 1,3-diamines, bearing a methyl substituent or an aromatic ring in the side chain, with and -oxo acids.Our aim was to investigate the effects of the side chain substituents and the size of the lactam rings formed on the stereochemical outcome of the reactions, and the conformational and mass spectral behaviour of the tetra-and pentacyclic products.X = O, S, NR; R 1 = H, alkyl; R 2 = H, alkyl, aryl Scheme 1
When tetrahydroisoquinoline diamines 9 and 10 were boiled in toluene with 4-oxopentanoic acid, 3-benzoylpropanoic acid, 4-benzoylbutanoic acid, 2-formylbenzoic acid or 2-benzoylbenzoic acid, cyclocondensations generally took place within 2-12 h to furnish the corresponding angularly-condensed tetra-and pentacyclic lactams 11-20 (Scheme 3) in good yields.By a similar reaction of diamine 21 and 2-formylbenzoic acid, the angularly-condensed hexacyclic lactam 22 was obtained (Scheme 4).NMR spectra of the crude products 11-20 and 22 indicated that neither the methyl substituent nor the aromatic ring at the side chain influenced significantly the stereochemical outcome of the lactam formation, since the polycycles 11-20 and 22 were formed with practically full diastereoselectivity (de ~100%), affording only the isomers having the hydrogen at the annelation of the tetrahydroisoquinoline and imidazolidine/hexahydropyrimidine rings (H-10b for 11-14, H-11b for 15 and 16, H-4b for 17-20 and 22) and substituent R (H at 14b in 22) in the cis position.The observed high de values in favour of the formation of the cis lactams can be explained as a result of the kinetic control governing the second cyclization steps of the domino cyclocondensations.

NMR spectroscopic structure determination
The solution structures of products 11−20 and 22 were determined by 1 H and 13 C NMR spectroscopy in CDCl3 solution.The 1 H and 13 C chemical shifts of these compounds were assigned with the help of standard 2D NMR correlation spectra (dqf-COSY, NOESY, multiplicity-edited HSQC and HMBC).The stereo assignment of diastereotopic protons, and the analysis of relative configuration and conformation was based on NOEs and JH,H-coupling constant values.Accurate 1 H chemical shifts and J H,H -coupling constants were obtained from computer simulation and iteration of the 1D 1 H NMR spectra.The NMR data are presented in Tables 1-3.Below, the diastereotopic hydrogens which are on the same side of the polycycle as the isoquinoline bridgehead hydrogen will be labelled as syn or s, and as anti or a otherwise.
Compounds 11-14 contain three asymmetric carbons (C3a, C10b, C11), as do 15-16 (C4a, C11b, C12), 17-18 (C4b, C5, C11b) and 19-20 (C4b, C5, C12b), whereas 22 contains two (C4b, C14b).Of these, the R-bearing carbon atom (C3a for 11 etc.) is a new stereo centre formed during the domino ring-closure and its configuration is, a priori, unknown.The syntheses were highly stereo-selective, so that only one of the two possible epimers was produced to a notable extent.For each product, proton-proton NOE correlations proved that the relative positions of the substituents carried by the three asymmetric carbons (two for 22) is all-cis, as shown in Schemes 2 and 3.These NOEs also verify the configuration of the starting diamines 9 and 10.The preferred conformations of the polycyclic products could be deduced from the Jcoupling constants and NOE correlations between protons, and by using the structural information thus obtained as constraints in molecular mechanics modelling.Compound 11 showed an NOE correlation between all pairs of 3a-Me, 5-Hsyn and 10b-H, indicating their mutual spatial proximity, and also displayed large 3 J H,H (5s,6a) and 3 J H,H (10b,11a) coupling values implying these hydrogens to be antiperiplanar.On the other hand, the 5-Hsyn proton of 12-14, and the analogous proton of derivatives 15-20 and 22 (6-Hsyn of 15 etc.), typically did not share an NOE correlation with the substituent R or the isoquinoline bridgehead proton.Instead, these derivatives showed an NOE between 5-Hanti and 11-Hanti (12-14), 6-Hanti and 12-Hanti (15-16) etc.The derivatives 12-20 and 22 also had large 3 JH,H(5a,6s) (12-14), 3 JH,H(6a, etc. values (as opposed to large 3 JH,H(5s,6a) etc.).These data suggest two conformations populated by these compounds: 11 prefers a conformation in which the ring B (tetrahydropyridine) is approximately trans-fused with the ring C (imidazolidine) whilst the other derivatives prefer a cis-fused conformation between rings B and C (the latter being either imidazolidine, n = 0, or hexahydropyrimidine, n = 1).Ring B adopts a half-chair conformation with the two aromatic carbons and the aliphatic carbons attached to them defining a plane, and the remaining methylene carbon (C5 etc.) and nitrogen (N4 etc.) lying above and below this plane.In 11, it is the methylene carbon (C5) which lies on the same side of the plane as the bridgehead hydrogen (10b-H), whereas in 12-20 and 22 the syn atom is nitrogen (12-14: N4,  15-16: N5, 17-18: N12, 19-20: N13, 22: N15).These trans-B/C and cis-B/C conformations are depicted in Figures 1 and 2 for 11 and 16, respectively. Tere is a possibility for an equilibrium between the trans-B/C and cis-B/C conformers, involving the ring inversion of ring B combined with an umbrella inversion of its nitrogen atom.In a pure trans-B/C conformation, 3 JH,H(5s, or the analogous coupling (15-20, 22) would be large while the corresponding 3 JH,H(5a,6s) etc. coupling would be small.In the cis-B/C conformation, the opposite would be true.In case of a fast equilibrium, the observed values of these coupling constants become population-weighted averages of the "large" (ca.12.2 Hz) and  "small" (ca. 1 Hz) limiting value.Using the J-data from Table 2 this allows for a rough estimate: 11 is ca.85% trans-B/C, 13 and 17 are 80-85% cis-B/C and the rest are 90-100% cis-B/C.Thus, increasing the size of the substituent R or adding a condensed benzene ring (ring E) on the lactam ring seems to shift the equilibrium towards the cis-fused conformation.This is consistent with our previous findings on similar structures which lack the methyl substituent at ring C. 5 Comparing the results, this methyl substitution in itself seems to cause a shift towards the cis-B/C conformation.The terminal six-membered piperidine ring in 15 and 16 (ring D) displayed large vicinal H,H-coupling constants within the anti-syn-anti sequence of its protons (i.e. 3 J H,H (2a,3s) and 3 JH,H(3s,4a)), implying their trans-diaxial relationship.This is consistent with a chair or halfchair (with the amide group close to planar) conformation, with R in a (pseudo)axial position.When the terminal ring D is five-membered (pyrrolidine; derivatives 11-14) the J-coupling between the syn protons (2s,3s) is large, as is the coupling between the anti protons (2a,3a).Therefore, in the preferred conformation of this ring the syn hydrogens nearly eclipse each other (as do the anti hydrogens).For 11 and 12 (13 and 14), the eclipsed conformation is slightly opened so that the 2s and 3a (2a and 3s) hydrogens become more antiperiplanar.The ortho protons of the phenyl group in phenyl-substituted (R = Ph) derivatives 16 and 20 were non-equivalent and heavily broadened at 298 K, indicating that the rotation of this substituent is restricted and takes place in the slow-intermediate NMR time scale.The other phenylsubstituted compounds (12, 14, 15, 18) did not show similar signal splitting.Apparently, there are steric 1,3-diaxial interactions present when the rings C and/or D are six-membered (hexahydropyrimidine and piperidine, respectively), which in case of 16 and 20 slow down the rotation of the phenyl substituent sufficiently to cause the observed 1 H signal splitting of the ortho protons.All compounds except 13, 17, 18 and 22 give also some amount of the ion m/z 176 which for 11, 16, 18 and 20 (n = 0, 1, 0 and 1; R = Me, Ph, Ph and Ph, respectively The hexacyclic lactam is greatly stabilized by the benzo-fusions which is proved by the fact that its base peak is [M−H] + and also the molecular ion is very strong (72%).This is on line with the observation that otherwise it gives a great number of medium or weak ions (Tables 4 and 5).As to the other compounds 11, 13, 14, 17 and 18 give relatively few and mainly low mass fragments whereas compounds 12, 15, 16, 19, and 20 give rather many further fragments (Table 5).The common features for the first-mentioned group seem to be that n = 0 (11, 17 and 18) or n = 1 and R = Me or Ph (13 and 14).As to the second group the common features are n = 0 or 1 and R = Ph (12, 15, 16 and 20), for 19 n = 1 and R = H.If we compare this situation with the groups in Scheme 1 12, 17 and 18 (n = 0 and R = Ph) fall out from these groups probably due to the fivemembered ring in 12 and the benzo-fusion (and n = 1) in 17 and 18.On the other hand 15, 16, 19 and 20 consist of a six-membered oxo-ring + R=Ph (15 and 16) and a benzo-fusion with n = 1 (19 and 20).

Conclusions
Our results demonstrate that 1-substituted 1,2,3,4-tetrahydroisoquinoline 1,2-and 1,3-diamines, bearing a methyl substituent or an aromatic ring in the side chain, were conveniently transformed to tetra-, penta-and hexacyclic lactams by domino cyclocondensations with acyclic or aromatic or -oxo acids.The NMR analyses proved that each cyclocondensations took place with practically full diastereoselectivity (de ~100%) in favour of the cis isomer.Compound 11 prefers a conformation with a trans-B/C ring fusion, whereas the other derivatives (12-20, 22) favour cis-B/C fusion.These conformers can change into each other through a ring-inversion of ring B combined with an umbrella inversion of its nitrogen atom, but the population of the preferred conformer is always 80-100%.Increasing the size of substituent R, or adding a condensed benzene ring at the lactam ring, shifts the conformational equilibrium towards the cis-B/C conformer.

Experimental Section
General.Melting points were measured on a Kofler hot-plate microscope apparatus and are uncorrected.Column chromatography was performed with silica gel 60 (0.063-0.200).For routine thin-layer chromatography (TLC), Silica gel 60 F254 plates (Merck, Germany) were used.Elemental analyses were performed with a Perkin-Elmer 2400 CHNS elemental analyser.Compounds 10 7 and 21 8 were prepared according to known procedures.

NMR spectra
The NMR spectra were recorded in CDCl3, DMSO-d6 or D2O solutions on JEOL JNM-LA400, Bruker AVANCE DRX 400 and AVANCE 500 spectrometers.Chemical shifts are given in  (ppm) relative to TMS (CDCl3 and DMSO-d6) or to TSP (D2O) as internal standards.The spectra of the products were acquired without sample spinning at 298 K.The NMR experiments consisted of standard 1 H NMR (using a 30º flip angle and a 5 s pulse repetition time), 13 C NMR with broad-band proton decoupling, dqf-COSY, 1D and 2D NOESY (with a mixing time of 0.3 s), multiplicity-edited HSQC (optimized for a one-bond coupling of 145 Hz and set to show CH and CH3 signals positive and CH2 signals negative), and HMBC (optimized for long-range couplings of 8 Hz with a low-pass J-filter optimized to remove signals due to one-bond coupling around 145 Hz) measurements.Proton chemical shifts δH and proton-proton coupling constants JH,H were extracted by using the spectral simulation and analysis tool PER included in the PERCH NMR software package (version 2008.1). 12The initial guess for the NMR parameters was obtained manually from the 1 H spectra or by trial-and-error in case of crowded/complicated spectral regions, which was then refined iteratively by using the integral-transform and total lineshape-fitting modes of the software.Manual adjustment and iterative fitting of parameters was repeated until good visual comparison was achieved between the calculated and the observed spectra.Structural models were obtained from molecular-mechanics modelling (MM+ force field) by using the HyperChem 7.0 software.Distance and torsion angle restraints (with the software's default force constants) were applied in the initial modelling based on the observed NOE and J-coupling constant data, and when a satisfactory structure was obtained it was refined without restraints to yield the final model.

Mass spectral measurements
The electron ionization mass spectra (Tables 4 and 5) were recorded on a VG ZABSpec mass spectrometer (VG Analytical, Division of Fisons, Manchester, UK) equipped with the Opus V3.3X program package (Fisons Instruments, Manchester, UK).The ionization energy used was 70 eV and the source temperature was 160 °C.The accelerating voltage was 8 kV and the trap current was 200 mA.Perfluorokerosene was used to calibrate the mass scale.A small amount of solid sample dissolved in MeOH was placed in a capillary tube and the solvent was evaporated off with hot air.Thereafter, the sample was introduced into the ionization chamber via the solid inlet.The fragmentation pathways were confirmed by linked scans at constant B/E or B 2 /E (first field-free region, FFR1) without collision gas.The low-resolution B/E and B 2 /E spectra were measured with a resolving power of 3000 (10% valley definition).The accurate masses were determined by voltage scanning with a resolving power of 6000-10,000.The compositions of the molecular ions based on their accurate masses are given in Table 6.For the low-resolution spectra, consecutive scans selected from the stable and constant part of the total ion current chromatogram were averaged to obtain more reproducible abundances.For accurate masses and linked scans, 10 scans were averaged to minimize noise and to eliminate random peaks.Since our VG ZABSpec was taken out of use, the mass spectrum of 11 was measured on a Bruker micrOTOF-Q ESI-HRMS instrument (Bruker Daltonics, Bremen, Germany).Mass spectrometer was controlled by Compass 1.3 for micrOTOF software package (Bruker Daltonics).The sample of 11 was dissolved in HPLC-grade acetonitrile (VWR International, Leuven, Belgium).The sample was then introduced to source using infusion pump.Positive ionization mode was used for MS analysis.The capillary voltage was maintained at -4500V and the end plate offset at -500 V. Nitrogen was used as nebulizer and drying gas.The pressure for the nebuliser gas was set at 0.4 bar.The drying gas flow rate was 4.0 L/min and the drying gas temperature 200°C.Mass detection was performed in the m/z range 50-1500.The resolving power of the mass spectrometer was typically 8000-12000.Sodium formate clusters were used for external calibration of the m/z range.Compass DataAnalysis 4.0 (Bruker Daltonics) was used for interpreting the mass data.For collision induced dissociation (CID) MS/MS measurements, the collision energy in the m/z range 50-1500 varied from 20 eV or from 30 eV (m/z 50) to 75 eV (m/z 1500).Argon was used as collision gas.The mass spectrometer was operated in data-dependent mode to automatically select the most abundant precursor ions.7.28-7.38(5H, m, C6H5). 13C NMR (100 MHz, CDCl3): C 19. 1, 35.6, 41.3, 51.0, 56.2, 56.3,  67.3, 111.9, 112.5, 121.1, 128.4,128.6, 128.9, 131.8, 136.6, 148.1, 149.5, 156.4,172.9.Anal.Calcd.for C21H26N2O5 C, 65.27; H, 6.78; N, 7.25; Found C, 65.02; H, 6.49; N, 7.13%.
Pure diamine base 9 was obtained from the above dihydrobromide by alkaline treatment (20% NaOH), extraction (CH2Cl2) and evaporation under reduced pressure.The free base was dried in a vacuum desiccator for 24 h before further transformations.
General procedure for the preparation of tetra-, penta-and hexa-cyclic lactams 11-20 and 22 A mixture of diamine 9 or 10 or 21 (3 mmol) and the corresponding or -oxo acid (3 mmol) was refluxed in toluene (40 mL) until no more starting materials could be detected by TLC (3-12  h).In case of 21, some crystals of p-toluenesulfonic acids were also added to the reaction mixture prior to reflux.The solvent was then evaporated off and the oily or solid residue (the NMR spectrum of which was applied for determination of the diastereomeric ratios) was purified by means of column chromatography.9), 205 (20), 204 (30), 202 (13), 193 (16), 192 (98), 190 (62), 189 (20)

2 Figure 1 .
Figure 1.The preferred conformation, trans-B/C, of 11 in CDCl3 solution at 298 K as obtained from MM optimization with constraints from NMR spectroscopic analysis.Relevant NOE correlations are indicated with double-headed arrows.

Figure 2 .
Figure 2. The preferred conformation, cis-B/C, of 16 in CDCl3 solution at 298 K as obtained from MM optimization with constraints from NMR spectroscopic analysis.Relevant NOE correlations are indicated with double-headed arrows.
20) or 14b (22) as well as n = 0 or 1 (Scheme 3) seems to have fairly decisive effect on the type of fragments obtained.For all compounds, ring C splits into relatively strong ion a+1 which is even the base peak when n = 1 and R = Me or Ph and into [M−a] +• which in turn is the base peak when n = 0 and R = H or Ph (17 and 18).Furthermore, 11, 16, 19, 20 and 22 give some amount of the ion a−1 which is the base peak for 19 (n = 1 and R = Ph).The complementary ion [M−a] +• is the base peak for 17 and 18 (n = 0 and R = H or Ph) and also rather strong for the other compounds except 11 and 22.All compounds except 11, 13, 17 (n = 0, 1, 0 and R = Me, Me and H, respectively) and 22 (H at C-14b) give relatively strong complementary ion [M−(a+1)] + which is even the base peak for 14 and 20 (n = 1 and R = Ph).Compounds 14, 19 and 20 (n = 1; R = Ph, H, Ph, respectively) give also a weak ion [M−(a−1)] + and compounds 12, 14, and 15 (n = 0, 1 and 0, respectively and R = Ph) a weak ion [M−(a+2)] +• .Ring C additionally splits into ion b, which contains the CH3CH-group (Scheme 5) more than ion a. Ion b is the base peak for 12 (n = 0 and R = Ph) and abundant for 15 (n = 0 and R = Ph) and present also in the spectra of 17 and 18 (n = 0, R = H and Ph).It is interesting that compounds 11, 15 and 19 (n = 0 and 1; R = Me, Ph and H, respectively) give the ion b−1 of which, however, only 15 (n = 0, R = Ph) exhibited also the ion b.

Table 1 .
Proton chemical shifts δ H (in ppm, CDCl 3 , 298 K) of 11-22.The label s(a)indicates that the hydrogen is syn (anti) with respect to the isoquinoline bridgehead hydrogen.The chemical shifts marked with an asterisk (*) may be interchanged

Table 2 .
Proton-proton coupling constants JH,H (in Hz, CDCl3, 298 K) of 11-22.The label s(a)indicates that the hydrogen is syn (anti) with respect to the bridgehead hydrogen

Table 3 .
Carbon chemical shifts δC (in ppm, CDCl3, 298 K) of 11-22.The assignment between shifts marked with a similar symbol is uncertain

Table 4 .
Main fragments [m/z(RA %)] from the studied compounds under electron ionization.

Table 6 .
Composition of the molecular ions .05mol) was added dropwise at a rate low enough to keep the internal temperature below -10 °C.After 5 min, a solution of homoveratrylamine (9.06 g, 0.05 mol) in CH2Cl2 (50 mL) was added dropwise, the internal temperature being kept below 0 °C.When the addition was complete, the reaction mixture was heated under reflux for 5 min.The mixture was allowed to cool down to room temperature and CHCl3 (250 mL) was added.The mixture was next washed with saturated NaHCO3 solution (3 × 75 mL) and water (2 × 75 mL), and then dried (Na2SO4), and the solvent was removed in vacuo to give a crude oily product, which crystallized on treatment with Et2O.The crystals were filtered off, washed with Et2O and recrystallized from EtOAc.
NaBH4 (3.10 g, 82 mmol) was added in small portions.The resulting mixture was stirred for 3 h with ice-water bath cooling and for 3 h at ambient temperature, and then evaporated in vacuo.The residue was dissolved in 5% HCl (150 mL), and the solution was made alkaline with 20% NaOH while cooled, and then extracted with CHCl3 (4 × 150 mL).The combined organic extracts were dried (Na2SO4) and evaporated in vacuo to give an oily product, containing diastereomers 8a and 8b in a ca.8 : 1 ratio.The oil crystallized on treatment with a 2:1 mixture of n-hexane and Et 2 O.The crystalline product, which was filtered off and washed with n-hexane, proved to be diastereomerically pure 8a.The crude crystalline product was used in the next step without further purification.