Crystal structure and absolute configuration of (3S,4aS,8aS)-N-tert-butyl-2-[(S)-3-(2-chloro-4-nitrobenzamido)-2-hydroxypropyl]decahydroisoquinoline-3-carboxamide and (3S,4aS,8aS)-N-tert-butyl-2-{(S)-2-[(S)-1-(2-chloro-4-nitrobenzoyl)pyrrolidin-2-yl]-2-hydroxyethyl}decahydroisoquinoline-3-carboxamide

The crystal structure and absolute configuration of two new nelfinavir analogs have been determined. Intermolecular hydrogen bonding leads to two-dimensional sheets in one analog and one-dimensional chains in the other.


Chemical context
Nelfinavir (Viracept) is an FDA approved HIV protease inhibitor identified through structure-based design with a low nanomolar inhibitory concentration against the HIV aspartyl protease (Kaldor et al., 1997). Although nelfinavir is no longer recommended as a first-line treatment against HIV due to its inferior efficacy compared to alternative protease inhibitors (Panel on Antiretroviral Guidelines, 2015), it has been found to have a number of additional biological activities that may have therapeutic utility, including antiviral (against other human viruses) (Yamamoto et al., 2004;Kalu et al., 2014), anticancer (Gantt et al., 2013;Koltai, 2015), and antivirulence activity (Maxson et al., 2015). However, nelfinavir was originally designed with only the HIV protease active site in mind and the structure is likely not optimal for binding to the alternative targets involved in these other activities. We recently reported on the synthesis of a collection of nelfinavir analogs that may be of interest for efforts to repurpose the drug (Maxson et al., 2015).
The syntheses of the title compounds were achieved by a previously reported route that utilizes the configuration of the amino acid starting material to control the stereochemical outcome of the sodium borohydride reduction of the chloromethyl ketone (Kaldor et al., 1997). However, the reduction of compound (I), derived from achiral glycine, results in a racemic mixture (Fig. 1), while the reduction of compound (II), derived from l-proline, does not benefit from a strong ISSN 2056-9890 directing influence from the existing chiral center (Fig. 2). The products of the two reductions were carried forward through the remainder of each synthesis to generate the title compounds. The absolute configurations of compounds (I) and (II), as well as the conformations they adopt due to the increased flexibility and rigidity, respectively, relative to nelfinavir was investigated by X-ray diffraction.

Figure 2
The synthesis of (II).

Figure 3
Plot showing 35% probability ellipsoids for non-H atoms and circles of arbitrary size for H atoms for (I). Only the major component of disordered sites is shown.
There is disorder of the Cl group in (I) over two positions with the site occupancies refining to 0.941 (8) and 0.059 (8) for Cl1 and Cl1B, respectively. The nitro group is disordered over two positions, with the site occupancies refining to 0.60 (2) and 0.40 (2). The NO 2 group in one orientation is essentially coplanar with the phenyl ring [O1B-N1B-C4-C3; = 1(2) ] and in the other orientation is twisted slightly more out of plane [O1-N1-C4-C3; = À9.0 (13) ]. Both sixmembered rings of the decahydroisoquinoline group in (I) adopt a chair conformation, the dihedral angle between the best-fit planes of the cyclohexyl and piperidine moieties is 119.9 (15) . There is one intramolecular hydrogen-bonding interaction in (I) which involves the two carboxamide groups (N2-H2Á Á ÁO5; Table 1). The Flack x parameter of À0.008 (18) and the Hooft y parameter of À0.010 (19) indicate that the absolute configuration of (I) has been assigned correctly.
There are multiple disordered moieties in (II), the nitro group is disordered over two positions with the site occupancies for the two orientations refining to 0.967 (6) and 0.033 (8). In both orientations, the NO 2 group is twisted out of the plane of the phenyl ring; the major orientation is twisted out of the plane less [O1-N1-C3-C2; = 10.9 (4) ] than the minor orientation [O1B-N1B-C3-C2; = À26 (6) ]. The carbonyl C7-O3 group is disordered over two positions, with the site occupancies refining to 0.58 (2) and 0.42 (2). In the minor orientation, the CO group is nearly normal to the plane of the phenyl ring [O3B-C7B-C6-C5; = À89 (3) ], while the major orientation is significantly less out of plane [O3-C7-C6-C5; = À44 (3) ]. The final two disordered moieties of (II) are a portion of the pyrrolidin-2-yl group and the three methyl groups of tert-butyl. The C10 and C11 atoms of the pyrrolidin-2-yl group are disordered over two positions, with site occupancies of 0.669 (16) and 0.331 (16). The tert-butyl methyl groups are also disordered over two positions via a slight rotation around the N4-C24 bond, the site occupancies refining to 0.811 (17) and 0.189 (17). Similar to (I), both sixmembered rings of the decahydroisoquinoline group in (II) adopt a chair conformation, with a dihedral angle between the best-fit planes of the cyclohexyl and piperidine moieties of 116.3 (17) . There is one weak intramolecular hydrogenbonding interaction in (II), involving the N-tert-butyl carboxamide group and the 2-hydroxyl O atom (N4-H4CÁ Á ÁO4; Table 2). The Flack x parameter of 0.036 (19) and the Hooft y parameter of 0.03 (2) indicate that the absolute configuration of (II) has been assigned correctly.

Supramolecular features
The extended structure of (I) is a two-dimensional sheet of hydrogen-bonded molecules extending in the ac plane ( Fig. 5a). Each molecule of (I) is hydrogen bonded to four neighboring molecules via O-HÁ Á ÁO and N-HÁ Á ÁO interactions; the details of these interactions can be found in Table 1. The two-dimensional layers stack in an ABAB pattern along the crystallographic b axis (Fig. 5b). The layers are separated by the bulky decahydroisoquinoline groups, which protrude above and below the sheets. The layers alternate between these bulky groups pointing 'left' and 'right', this along with a slight offset between the A and B layers allows them to interdigitate.
The extended structure of (II) is a one-dimensional chain of hydrogen-bonded molecules extending parallel to the crystallographic a axis (Fig. 6a). Each molecule of (II) is hydrogen bonded to two neighboring molecules via O-HÁ Á ÁO interactions, the details of these interactions can be found in Table 2. The one-dimensional chains are separated by the bulky decahydroisoquinoline groups and the tert-butyl groups, which prevent the chains from linking via further hydrogenbonding interactions (Fig. 6b). Plot showing 35% probability ellipsoids for non-H atoms and circles of arbitrary size for H atoms for (II). Only the major component of disordered sites is shown. Table 1 Hydrogen-bond geometry (Å , ) for (I). Symmetry codes: (i) x; y; z þ 1; (ii) x þ 1 2 ; Ày þ 3 2 ; Àz þ 2. Table 2 Hydrogen-bond geometry (Å , ) for (II).

Database survey
A search of the Cambridge Crystallographic Database (CSD; Groom & Allen, 2014) returns only three crystal structures with the N-(tert-butyl)decahydroisoquinoline-3-carboxamide core. One of the structures is N-(tert-butyl)decahydroisoquinoline-3-carboxamide (CSD refcode COVYAO;Zhao et al., 2006). The other two molecules are nelfinavir derivatives like (I) and (II), which were isolated during optimization of the synthesis. The difference between these two molecules comes via the substitution at the N-atom position of the decahydroisoquinoline group. One compound has a 3-amino-2-hydroxy-4-(phenylsulfanyl)butyl group in this position (CSD refcode QONJUY; Inaba et al., 2000) and the other has a 3-acetoxy-2-(3-acetoxy-2methylbenzoylamino)-4-(phenylsulfanyl)butyl group at the Natom position (CSD refcode GONKOJ; Inaba et al., 1998). Each of these molecules has intramolecular N-HÁ Á ÁO hydrogen bonding. In QONJUY it involves the two carboxyamide groups similar to the situation in compound (I). In GONKOJ it involves the N-tert-butyl carboxamide group and the 2-hydroxyl O atom similar to the situation in compound (II). The core structure of each of these previously reported materials is similar to (I) and (II) in that both six-membered rings of the decahydroisoquinoline groups adopt chair conformations. The dihedral angle between the best-fit planes of the cyclohexyl and piperidine moieties for the 3-amino-2hydroxy-4-(phenylsulfanyl)butyl-substituted molecule is 117.1 (18) . Similarly, this angle for the 3-acetoxy-2-(3-acetoxy-2-methylbenzoylamino)-4-(phenylsulfanyl)butyl-substituted molecule is 116.8 (14) .

Refinement
Crystal data, data collection and structure refinement details for (I) and (II) are summarized in Table 3. Structural models consisting of the target molecules were developed for (I) and (II). Several disordered sites on each molecule were modeled with disorder. In each case, like distances were restrained to be similar. Since the major and minor components of each disordered site are in such close proximity to each other, the displacement parameters were constrained to be equal. Methyl H atom positions, R-CH 3 , were optimized by rotation about R-C bonds with idealized C-H, R-H and HÁ Á ÁH distances. All hydroxy and amine H atoms were located in a difference Fourier map in good hydrogen-bonding environments (Hamilton & Ibers, 1968) and their distances were allowed to refine. The O4-H4B distance in (II) was restrained to be 0.84 (2) Å . The remaining H atoms were included as riding idealized contributors. Methyl, hydroxy and amine H atomU values were assigned as 1.5 times U eq of the carrier atom; remaining H atom U values were assigned as 1.2 times the carrier atom U eq . On the basis of 2237 unmerged Friedel opposites, the fractional contribution of the inverted twin component was negligible (Flack, 1983;Flack & Bernardinelli, 2000) for (I). The absolute structure parameter y was calculated using PLATON (Spek, 2009). The resulting value was y = À0.010 (19), indicating that the absolute structure has been determined correctly (Hooft et al. 2008). On the basis of 2720 unmerged Friedel opposites, the fractional contribution of the inverted twin component was negligible (Flack, 1983;Flack & Bernardinelli, 2000) for (II). The absolute structure parameter y was calculated using PLATON (Spek, 2009). The resulting value was y = 0.03 (2) indicating that the absolute structure has been determined correctly (Hooft et al. 2008

Computing details
For both compounds, data collection: APEX2 (Bruker, 2014); cell refinement: SAINT (Bruker, 2014); data reduction: SAINT (Bruker, 2014), XPREP (Bruker, 2014), and SADABS (Bruker, 2014  where P = (F o 2 + 2F c 2 )/3 (Δ/σ) max = 0.001 Δρ max = 0.34 e Å −3 Δρ min = −0.43 e Å −3 Absolute structure: Flack (1983); Hooft et al. (2008); 2720 Friedels Absolute structure parameter: −0.008 (18) Special details Experimental. One distinct cell was identified using APEX2 (Bruker, 2014). Four frame series were integrated and filtered for statistical outliers using SAINT (Bruker, 2014) then corrected for absorption by integration using SHELXTL/XPREP V2005/2 (Bruker, 2014) before using SAINT/SADABS (Bruker, 2014) to sort, merge, and scale the combined data No decay correction was applied. Geometry. All e.s.d.'s (except the e.s.d. in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell e.s.d.'s are taken into account individually in the estimation of e.s.d.'s in distances, angles and torsion angles; correlations between e.s.d.'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.'s is used for estimating e.s.d.'s involving l.s. planes. Refinement. Structure was phased by direct (Sheldrick, 2015). Systematic conditions suggested the ambiguous space group. The space group choice was confirmed by successful convergence of the full-matrix least-squares refinement on F 2 . The final map had no significant features. A final analysis of variance between observed and calculated structure factors showed no dependence on amplitude or resolution.

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å 2 )
x y z U iso */U eq Occ.  Special details Experimental. One distinct cell was identified using APEX2 (Bruker, 2014). Four frame series were integrated and filtered for statistical outliers using SAINT (Bruker, 2014) then corrected for absorption by integration using SHELXTL/XPREP V2005/2 (Bruker, 2014) before using SAINT/SADABS (Bruker, 2014) to sort, merge, and scale the combined data. No decay correction was applied. Geometry. All e.s.d.'s (except the e.s.d. in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell e.s.d.'s are taken into account individually in the estimation of e.s.d.'s in distances, angles and torsion angles; correlations between e.s.d.'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.'s is used for estimating e.s.d.'s involving l.s. planes. Refinement. Structure was phased by direct (Sheldrick, 2015) methods. Systematic conditions suggested the ambiguous space group. The space group choice was confirmed by successful convergence of the full-matrix least-squares refinement on F 2 . The final map had no significant features. A final analysis of variance between observed and calculated structure factors showed little dependence on amplitude and resolution.