Abstract
A series of 3β-hydroxy steroid analogues possessing a contracted cyclopentane B-ring were prepared based on the initial activity screening of a recently reported naturally occurring marine 5(6 →7)abeo-sterol against Mycobacterium tuberculosis. All of the novel ring B abeo-sterols synthesized showed good inhibitory activity, whereas none of the starting steroids based on the common 3β-hydroxy-Δ5-cholestane nucleus, proved to be active. Therefore, the 5(6 →7)abeo-sterol nucleus present in compounds 3, 5, 7, 9, and 11 represents a novel scaffold for the development of new antitubercular agents.
Keywords: Synthesis, Tuberculosis, Cytotoxicity, Abeo-sterols, Mycobacterium tuberculosis
Tuberculosis is a common and deadly infectious disease caused by mycobacteria, mainly Mycobacterium tuberculosis. Tuberculosis most commonly attacks the lungs (as pulmonary TB) but can also affect the central nervous system, the lymphatic system, the circulatory system, the genitourinary system, bones, joints and even the skin. Over one-third of the world’s population has been exposed to the TB bacterium, and new infections occur at a rate of one per second.1 In 2004, mortality and morbidity statistics included 14.6 million chronic active TB cases, 8.9 million new cases, and 1.6 million deaths, mostly in developing countries.2 In addition, a rising number of people in the developed world are contracting tuberculosis because their immune systems are compromised by immunosuppressive drugs, substance abuse, or HIV/AIDS. The rise in HIV infections and the neglect of TB control programs have enabled a resurgence of tuberculosis.3 The emergence of drug-resistant strains has also contributed to this new epidemic with, from 2000 to 2004, 20% of TB cases being resistant to standard treatments and 2% resistant to second-line drugs.4
During a recent investigation, while searching for novel highly-active antitubercular natural products from the Caribbean Sea sponge Svenzea zeai, our research group isolated a complex mixture of sterols the structures of which, after careful purification, were confidently assigned by combined spectroscopic methods.5,6 Thus, the known sterol 1 was obtained as the main isolate (2.2% yield) from the same hexane-soluble fractions as minor sterol 2, which was isolated only in 0.007% yield. The MIC values for antitubercular activity of compounds 1 and 2 were determined as 120.1 and 7.8 μg/mL, respectively. Since compound 1, displaying only marginal activity against M. tuberculosis H37Rv, is a most plausible biosynthetic precursor to sterol 2, this finding suggested that in active steroids, maximum antimycobacterial activity could be attained upon contracting the cyclohexane B-ring, most likely as a result of increasing both the hydrophilic impact and rigidity of the steroidal backbone.
On the basis of reasonable activity shown by sterol 2, and the fact that no prior work has been conducted to explore the potential of 6-5-6-5 fused rings sterols as antitubercular agents, we prepared a small 5(6 →7)abeo-sterol library for antimycobacterial screening by the Institute for Tuberculosis Research of the University of Illinois at Chicago. The significantly enhanced antitubercular activity suggested the importance of having a contracted cyclopentane Bring in the steroid series described herein. Additionally, branching in the side chain at C24 in combination with the 5(6 →7)abeo-steroidal nucleus, appear to maximize the sterol’s ability to be effective against M. tuberculosis (Fig. 1).
Figure 1.
Molecular structures of the sterol analogues submitted for in vitro antimycobacterial and cytotoxicity screenings.
We carried out the synthesis of five structurally diverse 5(6 →7)abeo-sterols for screening against M. tuberculosis H37Rv (Mtb). Whilst all of the sterol analogues synthesized (3, 5, 7, 9, and 11) possess the same tetracyclic 6-5-6-5 steroidal backbone, each analogue has a distinct alkyl side chain at C17. For instance, compound 2, the only naturally occurring steroid in this series, has a double bond in the side chain at C24, whereas semi-synthetic sterols 3, 5, 9, and 11 possess either a carbonyl, a methyl or ethyl moiety at C24. Analogue 9, on the other hand, displays further alkyl substitution at C22 and C23 of the side chain, whereas abeo-steroid 7 [prepared as a reference analogue from commercially available cholesterol (6)] has no branching in the side chain at C24. Interestingly, except for the latter analogue, all of the abeo-sterols synthesized have shown ≥80% inhibitory activity against Mtb at a concentration of 8 μg/mL (Table 1). On the other hand, none of the starting Δ5-unsaturated steroids (1, 4, 6, 8, and 10) displays meaningful activity at ≤64 μg/mL.
Table 1.
In vitro M. tuberculosis growth inhibition by sterols 1–11a
| Compound | % Inhibition (μg/mL)
|
MICb μg/mL | Cytotoxicityc IC50, μg/mL (SI) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 128 | 64 | 32 | 16 | 8 | 4 | 2 | |||
| 1 | 97 | 40 | 43 | 33 | — | — | — | 120.1 | >128 (n.d.) |
| 2 | 99 | 99 | 99 | 105 | 91 | 69 | 52 | 7.8 | 51 (6.5) |
| 3 | 99 | 100 | 96 | 92 | 83 | 67 | 52 | 13.6 | 43.8 (3.2) |
| 4 | 45 | 48 | 31 | 32 | 7 | 8 | 31 | >128 | >128 (n.d.) |
| 5 | 99 | 99 | 99 | 98 | 99 | 97 | 29 | 3.8 | 26.6 (7) |
| 6 | 57 | 24 | 13 | 33 | — | — | — | >128 | >128 (n.d.) |
| 7 | 99 | 99 | 99 | 96 | 51 | 34 | 10 | 15 | >128 (>9) |
| 8 | 38 | 18 | 13 | 10 | — | — | — | >128 | >128 (n.d.) |
| 9 | 99 | 99 | 98 | 96 | 81 | 33 | 17 | 12.7 | 54.7 (4.3) |
| 10 | 58 | 41 | 31 | 21 | 23 | 14 | 15 | >128 | >128 (n.d.) |
| 11 | 99 | 100 | 99 | 99 | 97 | 94 | 39 | 3.9 | 42.3 (10.8) |
| RMPd | 100 | 100 | 100 | 99 | 94 | 73 | 49 | 0.06 | 89.3 (1488) |
Values are means of three experiments.
Lowest drug concentration that effected an inhibition of ≥90% relative to untreated cultures.
Cytotoxicity against VERO cells (ATCC CCL-81). Selectivity index (SI) = IC50/MIC. ‘n.d.’ indicates not determined.
Rifampin was used as a positive control.
The abeo-sterol analogues herein described were prepared in one-pot by the reactions of the corresponding 3β-hydroxy-Δ5-cholestanes7 with ozone at −78 °C in either Et2O or CH2Cl2 solution after reductive work-up with dimethylsulfide under mild conditions according to the literature procedures.8–10 Intramolecular aldol condensation of the intermediate ε-keto aldehydes (not isolated) under slightly basic conditions led to the desired abeo-sterols in a relatively low isolated yield, between 30% and 50%, depending on the purity of the starting sterols used. Final purification was subsequently achieved by silica gel flash column chromatography and the desired products thus obtained were used as such for rigorous structure characterization studies as well as antimycobacterial screening. The complete structural assignment of all of the synthetic abeo-steroid analogues described in this work was accomplished on the basis of comprehensive 1D and 2D NMR experiments involving 1H–1H COSY, DEPT, NOESY, 1H–13C COSY (HMQC), and HMBC spectra, in addition to IR, UV, and HR-MS measurements.11 In most cases, the 2D NMR spectra provided both the structure and the complete and unambiguous proton and carbon atom assignments.
Each of the starting 3β-hydroxy-Δ5-cholestanes (1, 4, 6, 8, and 10) and all of the synthesized 5(6 →7)abeo-sterols (3, 5, 7, 9, and 11) were screened for their antimycobacterial activity. The activity results of steroids 1–11 are presented in Table 1. The primary screen was conducted at 128, 64, 32, 16, 8, 4, and 2 μg/mL against Mtb H37Rv (ATCC 27294) in BACTEC 12B medium using the Microplate Alamar Blue Assay (MABA) as previously described.12 Because all of the synthesized abeo-sterols demonstrated 100% inhibition at ≥64 μg/mL, they were retested to determine the MIC, defined as the lowest concentration inhibiting growth by ≥90%. Rifampin (RMP) was used as a positive control during all the antituberculosis assays.
Concurrent with the determination of MICs, compounds 1–11 were tested for cytotoxicity (IC50) in VERO cells in order to investigate the effect of contracting the cyclohexane B-ring on cytotoxic activity.13 The results described herein (Table 1) show that all of the starting steroids (1, 4, 6, 8, and 10) lacked significant cytotoxicity (IC50’s > 128 μg/mL). However, upon derivatization, analogues 3, 5, 9, and 11 showed some cytotoxicity (IC50’s 26.6–54.7 μg/mL). Remarkably, the absence of appreciable toxicity in analogue 7 (IC50 > 128 μg/mL) against mammalian cells in the VERO cell assay suggests that functionalization (branching or oxidation) at C24 of the flexible alkyl side chain alone might explain the mild toxicity displayed by most abeo-sterols.
In conclusion, during a MABA bioassay against M. tuberculosis (H37Rv), all of the abeo-sterol analogues synthesized showed minimum inhibitory concentrations of 3.8–15 μg/mL, while the starting 3β-hydroxy-Δ5-cholestanes had MIC values >128 μg/mL, respectively (Table 1). The present data suggest that the 5(6 →7)abeo-steroidal nucleus inherently enhances antimycobacterial activity and thus represents a novel scaffold for the development of clinically useful agents for tuberculosis. Further correlations of structural features and the MICs of the six abeo-sterols scrutinized during this study suggest that, in addition to the 5(6 →7)abeo-steroidal moiety, the presence of a methyl or ethyl residue at C24 is required for superior activity. The fact that modifications in the side chain can affect the antitubercular activity can be related to the permeability of the compounds through the Mtb membrane. Noteworthy is the activity of compound 3, which has two carbonyl groups and shows the worse selectivity index. Cytotoxicity results for a mammalian cell line indicate that in general strongly antitubercular abeo-sterols are also mildly cytotoxic. Although cytotoxicity of mammalian cells is low relative to toxicity to Mtb, it is not clear yet whether the observed activity of abeo-sterols is unique to mycobacterial targets. This study has demonstrated the future potential for development of B-ring abeo-sterol analogues as antimycobacterial agents and suggests new directions for the rational design of additional steroids that are active against the tubercle bacillus. In addition, the finding that abeo-sterol derivatives are inhibitors of Mtb is potentially very significant, as interest has been building recently in the metabolism of cholesterol by Mtb and the importance of such metabolism for intracellular survival in vivo.14
Acknowledgments
Work for this project at the UPR was supported by funding from the NIH-SCORE Grant S06GM08102.
References and notes
- 1.Raviglione MC, O’Brien RJ. Tuberculosis. In: Kasper DL, Braunwald E, Fauci AS, Hauser SL, Longo DL, Jameson JL, Isselbacher KJ, editors. Harrison’s Principles of Internal Medicine. McGraw-Hill Professional; 2004. p. 953. [Google Scholar]
- 2.World Health Organization (WHO). Tuberculosis Fact sheet No. 104—Global and Regional Incidence, March 2006.
- 3.Iademarco MF, Castro KG. Semin Respir Infect. 2003;18:225. doi: 10.1053/s0882-0546(03)00074-4. [DOI] [PubMed] [Google Scholar]
- 4.Emergence of Mycobacterium tuberculosis with extensive resistance to secondline drugs—worldwide, 2000–2004. MMWR Morb Mortal Wkly Rep. 2006;55:301. [PubMed] [Google Scholar]
- 5.Wei X, Rodríguez AD, Wang Y, Franzblau SG. Tetrahedron Lett. 2007;48:8851. [Google Scholar]
- 6.Steroids are important natural products that possess diverse biological activity, including cytotoxic and antimycobacterial activities. For recent articles related to the antitubercular properties of naturally occurring steroids, see: Cantrell CL, Rajab MS, Franzblau SG, Fronczek FR, Fischer NH. Planta Med. 1999;65:732. doi: 10.1055/s-1999-14053.Cantrell CL, Rajab MS, Franzblau SG, Fischer NH. J Nat Prod. 1999;62:546. doi: 10.1021/np980288u.Rugutt JK, Rugutt KJ. Nat Prod Lett. 2002;16:107. doi: 10.1080/10575630290020000.Wächter GA, Franzblau SG, Montenegro G, Hoffmann JJ, Maiese WM, Timmermann BN. J Nat Prod. 2001;64:1463. doi: 10.1021/np010101q.Cantrell CL, Franzblau SG, Fischer NH. Planta Med. 2001;67:685. doi: 10.1055/s-2001-18365.Okunade AL, Elvin-Lewis MPF, Lewis WH. Phytochemistry. 2004;65:1017. doi: 10.1016/j.phytochem.2004.02.013.Woldemichael GM, Gutierrez-Lugo MT, Franzblau SG, Wang Y, Suarez E, Timmermann BN. J Nat Prod. 2004;67:598. doi: 10.1021/np0303411.Suksamrarn A, Buaprom M, Udtip S, Nuntawong N, Haritakun R, Kanokmedhakul S. Chem Pharm Bull. 2005;53:1327. doi: 10.1248/cpb.53.1327.
- 7.The starting sterols 24-methylene-cholesterol (1) and 24-α-ethyl-cholesterol (4) were extracted, respectively, from the Caribbean marine sponges S. zeai and Myrmekioderma styx. Commercial cholesterol (6) was purchased from Aldrich Chemical Company. Gorgosterol (8) and 24-α-methyl-cholesterol (10) were extracted from the Caribbean gorgonian octocoral Gorgonia mariae. Abeo-sterol 3 was prepared from precursor 1 as described.
- 8.A stream of dry oxygen containing about 1–2% ozone was bubbled for 15 min into a solution of starting 3β-hydroxy-Δ5-unsaturated sterol (50.0 mg) in Et2O or CH2Cl2 (25 mL) kept at −78 °C. After allowing the temperature to warm up to 25 °C, the solvent was evaporated and the oily residue obtained was stirred with a mixture of dimethylsulfide (0.5 mL) and Et2O (5 mL) for 24 h at 25 °C to afford the ozonolyzed product. Upon removal of excess reagent under reduced pressure, 1.0 M KOH in MeOH (5 mL) was added and the resulting mixture stirred for 5 min at 25 °C. At this point, routine monitoring by TLC typically revealed no starting material left and the presence of a strongly UV active product. After quenching with saturated aqueous NH4Cl followed by extraction with Et2O (3 × 10 mL), the combined ether layers were washed with brine, dried (MgSO4), and concentrated in vacuo. Purification by silica gel flash column chromatography (hexane/EtOAc 90:10 or hexane/acetone 90:10) afforded the desired 5(6→7)abeo-sterol derivatives (15.5–26.0 mg; 30–50% isolated yield) as colorless oils with 97–98% purity.
- 9.Baldwin JE, Lusch MJ. Tetrahedron. 1982;38:2939. [Google Scholar]
- 10.Miyamoto T, Kodama K, Aramaki Y, Higuchi R, Van Soest RWM. Tetrahedron Lett. 2001;42:6349. [Google Scholar]
- 11.Compound 3: Colorless oil; −3.0 (c 0.8, CHCl3); IR (film) 3407, 2926, 2868, 1712, 1675, 1598, 1465, 1380, 1262, 1072, 802 cm−1; UV (MeOH) λmax 256 (ε7400) nm; 1H NMR (500 MHz, CDCl3) δ9.97 (s, 1H), 3.70 (m, 1H), 3.47 (m, 1H), 2.70–2.35 (br m, 3H), 2.10 (br m, 2H), 2.05–1.12 (broad envelope, 19H), 1.09 (d, J = 5.0 Hz, 6H), 0.93 (s, 3H), 0.92 (d, J = 5.0 Hz, 3H), 0.73 (s, 3H); 13C NMR (125 MHz, CDCl3) δ 215.4 (C), 189.6 (CH), 168.9 (C), 139.3 (C), 70.9 (CH), 60.1 (CH), 55.2 (CH), 54.4 (CH), 46.3 (C), 46.1 (CH), 45.2 (C), 40.8 (CH), 39.8 (CH2), 37.4 (CH2), 36.2 (CH2), 35.2 (CH), 33.9 (CH2), 31.3 (CH2), 29.9 (CH2), 28.5 (CH2), 26.5 (CH2), 20.7 (CH2), 18.7 (CH3), 18.4 (CH3), 18.3 (CH3), 15.6 (CH3), 12.5 (CH3); HREI-MS m/z [M]+ Calcd for C27H42O3 414.3134, found 414.3135; EI-MS m/z [M]+ 414 (57), 396 (25), 367 (26), 353 (6), 329 (12), 287 (9), 269 (15), 229 (9), 161 (18), 129 (28), 95 (49), 83 (73), 55 (100). (b) Compound 5: Colorless oil; −8.6 (c 0.9, CHCl3); IR (film) 3376, 2957, 2869, 1675, 1464, 1381, 1072 cm−1; UV (MeOH) λmax 256 (ε 8600) nm; 1H NMR (500 MHz, CDCl3) δ 9.97 (s, 1H), 3.70 (m, 1H), 3.47 (br d, J = 15.0 Hz, 1H), 2.55 (dt, J = 4.0, 11.0 Hz, 1H), 2.08 (br m, 3H), 2.00–0.99 (broad envelope, 23H), 0.93 (s, 3H), 0.92 (d, J = 6.5 Hz, 3H), 0.85 (t, J = 5.0 Hz, 3H), 0.82 (d, J = 6.5 Hz, 3H), 0.80 (d, J = 6.5 Hz, 3H), 0.73 (s, 3H); 13C NMR (125 MHz, CDCl3) δ 189.7 (CH), 168.9 (C), 139.3 (C), 70.9 (CH), 60.2 (CH), 55.3 (CH), 54.5 (CH), 46.3 (C), 46.2 (CH), 46.1 (CH), 45.2 (C), 39.8 (CH2), 36.2 (CH2), 36.1 (CH), 33.9 (CH2), 33.8 (CH2), 31.3 (CH2), 28.9 (CH), 28.5 (CH2), 26.6 (CH2), 26.5 (CH2), 23.0 (CH2), 20.7 (CH2), 19.6 (CH3), 19.0 (CH3), 18.9 (CH3), 15.6 (CH3), 12.5 (CH3), 12.3 (CH3); HREI-MS m/z [M]+ Calcd for C29H48O2 428.3654, found 428.3654; EI-MS m/z [M]+ 428 (61), 410 (34), 381 (46), 353 (9), 337 (10), 269 (15), 145 (27), 107 (30), 91 (37), 85 (100). (c) Compound 7: Colorless oil; −42.5 (c 0.8, CHCl3); IR (film) 3410, 2951, 2867, 1676, 1465, 1381, 1169, 1073 cm−1; UV (MeOH) λmax 256 (ε 14,200) nm; 1H NMR (500 MHz, CDCl3) δ9.97 (s, 1H), 3.71 (m, 1H), 3.46 (ddd, J = 2.0, 4.5, 14.5 Hz, 1H), 2.55 (dt, J = 4.0, 11.0 Hz, 1H), 2.08 (br m, 3H), 1.96–0.95 (broad envelope, 22H), 0.93 (s, 3H), 0.92 (d, J = 6.5 Hz, 3H), 0.86 (d, J = 6.5 Hz, 3H), 0.85 (d, J = 6.5 Hz, 3H), 0.73 (s, 3H); 13C NMR (125 MHz, CDCl3) δ 189.7 (CH), 168.9 (C), 139.3 (C), 70.9 (CH), 60.2 (CH), 55.4 (CH), 54.5 (CH), 46.3 (C), 46.2 (CH), 45.2 (C), 39.8 (CH2), 39.5 (CH2), 36.2 (CH2), 36.1 (CH2), 35.6 (CH), 33.9 (CH2), 31.3 (CH2), 28.5 (CH2), 28.0 (CH), 26.6 (CH2), 23.9 (CH2), 22.8 (CH3), 22.5 (CH3), 20.7 (CH2), 18.9 (CH3), 15.6 (CH3), 12.5 (CH3); HREI-MS m/z [M]+ Calcd for C27H44O2 400.3341, found 400.3343; EI-MS m/z [M]+ 400 (22), 382 (9), 353 (12), 149 (10), 145 (10), 120 (25), 118 (27), 95 (26), 87 (100). (d) Compound 9: Colorless oil; +1.8 (c 1.0, CHCl3); IR (film) 3391, 2956, 2870, 1677, 1460, 1382, 1160, 1072 cm−1; UV (MeOH) λmax 256 (ε 3000) nm; 1H NMR (500 MHz, CDCl3) δ 9.97 (s, 1H), 3.70 (m, 1H), 3.47 (br d, J = 15.0 Hz, 1H), 2.54 (m, 1H), 2.10–1.05 (broad envelope, 19H), 1.01 (d, J = 5.0 Hz, 3H), 0.94 (d, J = 5.0 Hz, 3H), 0.93 (s, 3H), 0.90 (s, 3H), 0.86 (d, J = 5.0 Hz, 3H), 0.78 (d, J = 5.0 Hz, 3H), 0.72 (s, 3H), 0.45 (dd, J = 5.0, 10.0 Hz, 1H), 0.20 (br m, 2H), –0.13 (dd, J = 5.0, 10.0 Hz, 1H); 13C NMR (125 MHz, CDCl3) δ 189.7 (CH), 168.8 (C), 139.3 (C), 70.9 (CH), 60.2 (CH), 57.1 (CH), 54.3 (CH), 50.8 (CH), 46.3 (C), 46.2 (CH), 45.6 (C), 39.9 (CH2), 36.2 (CH2), 35.0 (CH), 33.9 (CH2), 32.1 (CH), 32.0 (CH), 31.3 (CH2), 28.7 (CH2), 26.8 (CH2), 25.9 (C), 22.2 (CH3), 21.5 (CH3), 21.4 (CH3), 21.3 (CH2), 20.7 (CH2), 15.6 (CH3), 15.3 (CH3), 14.3 (CH3), 12.6 (CH3); HREI-MS m/z [M]+ Calcd for C30H48O2 440.3654, found 440.3653; EI-MS m/z [M]+ 440 (10), 422 (7), 414 (21), 400 (27), 367 (13), 353 (13), 287 (22), 269 (22), 107 (39), 95 (56), 83 (59), 55 (100). (e) Compound 11: Colorless oil, −2.0 (c 1.0, CHCl3); IR (film) 3415, 2953, 2869, 1675, 1463, 1381, 1261, 1166, 1074 cm−1; UV (MeOH) λmax 256 (ε 5000) nm; 1H NMR (500 MHz, CDCl3) δ 9.96 (s, 1H), 3.71 (br m, 1H), 3.46 (dd, J = 5.0, 15.0 Hz, 1H), 2.53 (dt, J = 5.0, 10.0 Hz, 1H), 2.08 (m, 3H), 2.05–0.94 (broad envelope, 21H), 0.93 (s, 3H), 0.93 (d, J = 6.5 Hz, 3H), 0.85 (br t, J = 5.0 Hz, 6H), 0.78 (d, J = 5.0 Hz, 3H), 0.72 (s, 3H); 13C NMR (125 MHz, CDCl3) δ 189.7 (CH), 168.9 (C), 139.3 (C), 70.9 (CH), 60.1 (CH), 55.2 (CH), 54.4 (CH), 46.3 (C), 46.1 (CH), 45.2 (C), 39.8 (CH2), 39.1 (CH), 36.1 (CH2), 36.0 (CH), 33.9 (CH2), 33.7 (CH2), 31.4 (CH2), 31.2 (CH), 30.7 (CH2), 28.7 (CH2), 26.6 (CH2), 20.7 (CH2), 20.5 (CH3), 19.1 (CH3), 17.6 (CH3), 15.6 (CH3), 15.4 (CH3), 12.5 (CH3); HREI-MS m/z [M]+ Calcd for C28H46O2 414.3498, found 414.3500; EI-MS m/z [M]+ 414 (47), 396 (15), 367 (24), 353 (16), 287 (8), 269 (11), 133 (20), 120 (63), 118 (64), 95 (47), 87 (100).
- 12.Collins LA, Franzblau SG. Antimicrob Agents Chemother. 1997;41:1004. doi: 10.1128/aac.41.5.1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cantrell CL, Lu T, Fronczek FR, Fischer NH, Adams LB, Franzblau SG. J Nat Prod. 1996;59:1131. doi: 10.1021/np960551w. [DOI] [PubMed] [Google Scholar]
- 14.(a) Yang X, Dubnau E, Smith I, Sampson NS. Biochemistry. 2007;46:9058. doi: 10.1021/bi700688x. [DOI] [PMC free article] [PubMed] [Google Scholar]; (b) Van der Geize R, Yam K, Heuser T, Wilbrink MH, Hara H, Anderton MC, Sim E, Dijkhuizen L, Davies JE, Mohn WW, Eltis LD. Proc Natl Acad Sci USA. 2007;104:1947. doi: 10.1073/pnas.0605728104. [DOI] [PMC free article] [PubMed] [Google Scholar]; (c) Pandey AK, Sassetti CM. Proc Natl Acad Sci USA. 2008;105:4376. doi: 10.1073/pnas.0711159105. [DOI] [PMC free article] [PubMed] [Google Scholar]