ABSTRACT
The impact of Pneumocystis pneumonia (PcP) on morbidity and mortality remains substantial for immunocompromised individuals, including those afflicted by HIV infection, organ transplantation, cancer, autoimmune diseases, or subject to chemotherapy or corticosteroid-based therapies. Previous work from our group has shown that repurposing antimalarial compounds for PcP holds promise for treatment of this opportunistic infection. Following our previous discovery of chloroquine analogues with dual-stage antimalarial action both in vitro and in vivo, we now report the potent action of these compounds on Pneumocystis carinii in vitro. Identification of chloroquine analogues as anti-PcP leads is an unprecedented finding.
INTRODUCTION
Pneumocystis jirovecii pneumonia (PcP) is a potentially fatal opportunistic infection for immunocompromised individuals, such as those infected with HIV, or receiving immunosuppressive therapies for organ transplantation, cancer, or autoimmune diseases (1–3). Although fatality rates due to PcP in HIV-infected people have significantly dropped due to highly active antiretroviral therapies, PcP remains a major cause of death or of chronic lung dysfunction among people affected by diseases as diverse as lymphoma (4), rheumatoid arthritis (5), lupus erythematosus (6), or renal dysfunction due to IgA nephropathy (7). Moreover, atypical radiological presentation of PcP sometimes occurs, leading to delayed diagnosis and intervention, with frequently discouraging results (1, 7). Though considered an innocuous infection in immunocompetent people, P. jirovecii has been suggested as a morbidity factor in sudden infant death syndrome, given its high incidence in healthy infants (8).
P. jirovecii, formerly known as “P. carinii f. sp. hominis,” is the causative agent of PcP or pneumocystosis in humans and was originally identified as the pathogen responsible for the epidemics of interstitial plasma cell pneumonitis that infected premature infants in Europe during World War II (2, 9). Pneumocystis spp. are unicellular eukaryotic organisms now included in the fungal kingdom due to their high genetic homology with other fungi (2, 10). These fungi are thought to be host specific and infect a wide variety of mammals (11). P. jirovecii infects humans, while P. carinii and P. murina infect rats and mice, respectively. Preclinical drug screening is conducted using the latter two species, as a culture system to properly propagate P. jirovecii in vitro was developed only recently, and still requires validation (12).
Pneumocystis spp. have several features that are atypical among fungi, such as the lack of efficacy of common antifungal agents such as amphotericin B and the ketoconazoles, likely due to the lack of ergosterol as the major sterol in their membranes (13). Initially, these microorganisms were thought to be protozoans, since they were susceptible to antiprotozoals such as primaquine, trimethoprim, sulfa drugs, pentamidine, or atovaquone, among others; these are still present in anti-PcP therapies, where the recommended first-line regimen for PcP prophylaxis and treatment is the well-known trimethoprim-sulfamethoxazole (TMP-SMX) combination, used for more than 50 years as a wide-spectrum antibiotic, and also employed against malaria (13, 14).
Some reports documenting P. jirovecii gene mutations in the target of the sulfamethoxazole component, dihydropteroate synthase (15, 16), may indicate resistance of this pathogen to TMP-SMX, supporting earlier reports (17–19). Alternative options for salvage therapy include intravenous pentamidine, the adverse effects of which are usually worse than those of TMP-SMX, or primaquine-clindamycin combinations (20). Clinical use of the latter still lacks approval in several countries, partly because primaquine (PQ; compound 1) (Fig. 1) is hemotoxic, in particular for people with 6-glucose-phosphate dehydrogenase (21, 22).
Since hemotoxicity and low oral bioavailability are major downsides of PQ's use in clinical practice, we have been working on its chemical modification over the past decade in an attempt to improve its action on malaria (23–28), leishmaniasis (29, 30), and pneumocystosis (24, 25), with overall promising results. These successes provided the rationale to extend such “chemical recycling” efforts to other antimalarials, as chloroquine (CQ; compound 2 in Fig. 1), (31, 32) and mepacrine (MPR; compound 3 in Fig. 1) (32). This effort delivered N-acyl chloroquine analogues (compounds 2a to 2j in Fig. 1) as relevant antiparasitic leads, active against liver- and blood-stage Plasmodium parasites (31, 32), as well as Leishmania spp. (28), while being generally noncytotoxic (31, 32). Moreover, one of such leads, compound 2c, displayed remarkable in vivo performance in immunocompetent mice when encapsulated in conveniently functionalized liposomes specifically targeting Plasmodium-infected erythrocytes (33).
Classical antimalarial drugs and respective analogues, here under scrutiny as potential anti-Pneumocystis agents.
In view of the above, we investigated the anti-Pneumocystis activity of N-acyl chloroquine derivatives compounds 2a to 2j (Fig. 1) using a standardized in vitro assay with P. carinii (34). The study included N-acyl primaquine and N-acyl mepacrine analogues, compounds 1a/1b and 3a, respectively, as well as all relevant parent antimalarials, for comparison (see Fig. 1). The effects of replacing the covalent amide bond between the antimalarial pharmacophore and the cinnamoyl group in PQ derivatives 1a,b and CQ derivatives 2a,d by an ionic ammonium carboxylate bond, respectively producing PQ-derived ionic liquids 1′a and 1′b (26) and CQ-derived ionic liquids 2′a and 2′d (Fig. 1), were also assessed.
RESULTS
Chemical synthesis.The test compounds were prepared with a minimum 95% purity as described elsewhere and presented structural data consistent with previous reports (27, 31, 32). Novel ionic liquids 2′a and 2′d were successfully synthesized and presented adequate purity and spectral data, as described in detail in Materials and Methods.
Anti-Pneumocystis activity and cytotoxicity assays.All the test compounds (including parent drugs compounds 1 to 3) were first screened for their anti-Pneumocystis activity at 100 μg/ml (data not shown); remarkably, with the exception of PQ derivatives 1a, 1b, and 1′b and CQ derivative 2h, all compounds reduced P. carinii viability (ATP content) by at least 94% after 24 h and reached nearly 100% reduction after 72 h. Such potent anti-Pneumocystis activity of most compounds was confirmed by subsequent determination of their respective 50% inhibitory concentrations (IC50s) at 72 h, as shown in Table 1, which includes the compounds' toxicity for two mammalian cell lines (equally expressed as IC50 at 72 h), namely, A549 human lung adenocarcinoma and L2 rat pulmonary epithelial cells.
MICs at 50% (IC50) of test compounds at 72 h against P. carinii, A549 human lung adenocarcinoma cells, and L2 rat pulmonary epithelial cells
Interestingly, despite the fact that, in most cases, activity differences are not markedly significant within experimental error, a number of trends for structure-activity relationships (SAR) were apparent. (i) The anti-Pneumocystis activity of PQ (compound 1) is markedly decreased upon N-acylation of the drug's primary amine with a cinnamoyl group, as in compounds 1a and 1b; in turn, when the amide bond between PQ and the cinnamic acid building blocks is replaced by a noncovalent ionic bond, as in compounds 1′a and 1′b, the activity of the parent drug is restored. (ii) In contrast to the previous situation, the moderate anti-Pneumocystis activity of CQ (compound 2) is preserved or even improved in the respective N-acyl analogues, except when a flexible alkyl spacer between the 4-aminoquinoline core and the N-acyl group is absent, as in compound 2h. Moreover, replacement of the amide bond, as in compounds 2a and 2d, by its ammonium carboxylate ionic counterpart, as in compounds 2′a and 2′d, does not affect activity expressively. (iii) Within the subset of CQ analogues 2a to 2g, compounds 2b (R = Me) and 2d (R = OMe) are markedly more active than the others, with compound 2b being nontoxic to both mammalian cell lines tested. The IC50s for this subset suggest that (a) the ideal length spacer between the 4-aminoquinoline core and the cinnamoyl group is butyl (n = 4), since activity slightly drops when this length is either decreased (compound 2b versus compound 2f) or increased (compound 2d versus compound 2g); notably, this was equally observed regarding anti-malarial activity (31); (b) there seems to be no obvious correlation between compounds' lipophilicity (clogP values, not shown) and anti-Pneumocystis activity, whereas former observations regarding the antimalarial activity of these compounds indicated that more lipophilic ones seemed preferred (31, 35); and (c) the influence of the stereoelectronic properties of the p-substituent R on the anti-Pneumocystis activity of compounds 2a to 2e was also surveyed by searching possible correlations between IC50s and classical descriptors, namely, the Hammett constant σpara, the inductive sigma constant σI, Charton's sterical constant σn, the Taft sterical parameter Es, the hydrophobicity parameter π, and molar refractivity (36). Interestingly, as shown in Fig. 2, a linear correlation (r2 = 0.96) seems to exist with the Hammett's constant σpara, where the positive slope of the linear fit suggests that electrodonating R groups favor anti-Pneumocystis activity. (iv) Replacement of the aryl group (compound 2a) by a nonaromatic cyclohexyl (compound 2i) or propyl (compound 2j) seems to slightly decrease activity, while improving selectivity compared to the rest of the compound 2a to 2g series, except when it comes to the best overall performer (activity plus selectivity), compound 2b. MPR (compound 3) and its N-acyl analogue 3a were included in this study only for comparison, since MPR was earlier reported as having low in vivo efficacy against Pneumocystis (37). Indeed, despite its marked activity, MPR was too toxic, whereas compound 3a was not, but it still did not outshine the activity and safety performance of its CQ counterparts compounds 2b and 2d.
Correlation between in vitro anti-Pneumocystis activity (IC50 at 72 h/μM) of compounds 2a to 2e and Hammett σpara constants for the aryl substituent group R.
Finally, all covalent compounds here reported were run through the following publicly available filters for Pan Assay Interference Compounds (PAINS) (38), namely, ZINC (http://zinc15.docking.org/patterns/home), (39) PAINS Remover (http://www.cbligand.org/PAINS/) (40), and Aggregator Advisor (http://advisor.docking.org) (41) to identify potential PAINS and aggregators. With the exception of compounds 3 and 3a, whose aminoacridine moiety triggered a PAINS alert (39, 40), and compound 2h, whose structural similarity to reported aggregators has been identified (39, 41), all compounds passed all filters used. In the case of organic salts 1′a, 1′b, 2′a, and 2′d, only ZINC was apparently able to conveniently process their ionic structures, returning no PAINS or aggregator alerts.
DISCUSSION
Globally, IC50s against P. carinii (Table 1) confirm that, with exception of compounds 1a and 1b (slightly active) and of compound 2h (inactive), all test compounds range from moderately to markedly active, particularly in the cases of CQ analogues 2b and 2d and of MPR (compound 3) and its analogue compound 3a. Moreover, the vast majority of the compounds displayed only mild or no cytotoxic activity against both mammalian cell lines used, the only exceptions being compounds 2a, 2c, 2e, and 3, which displayed IC50s below 10 μg/ml against A549 human lung carcinoma cells. Remarkably, one of the most active compounds against P. carinii, compound 2b, was also the most selective.
Altogether, these observations show that, in contrast to PQ, whose well-known anti-Pneumocystis activity is practically lost upon acylation of the drug's aliphatic amine, N-acyl chloroquine analogues, especially those embedding a cinnamoyl group, are quite promising anti-PcP leads. Interestingly, this is not the first time that N-cinnamoyl chloroquine analogues such as compounds 2a to 2g have performed surprisingly well against pathogens, e.g., P. carinii as here described, or against specific pathogen developmental stages, e.g., liver-stage plasmodia, as previously reported by us (31), for which PQ, not CQ, was one of the gold standards.
The significant activity of the test compounds against P. carinii may be different against P. jirovecii or P. murina, given the biological divergence among these species. Still, the predictive value of the in vitro methods applied here has been previously validated in vivo in a mouse model of Pneumocystis pneumonia (42–44).
In conclusion, N-cinnamoyl chloroquine analogues such as compounds 2a to 2g are a notable family of anti-infective compounds, in particular because they have displayed remarkable activity with pathogens against which CQ, or other related compounds, is not currently considered useful. In the case of PcP, these compounds hold great promise, and the best performers, compounds 2a and 2b, will be further assessed in in vivo assays. To the best of our knowledge, unveiling CQ analogues as potent anti-PcP leads is unprecedented.
Ongoing work is addressing evaluation of the potential activity of these CQ analogues against another concerning class of lung pathogens, mycobacteria (45), since CQ itself has been reported as potentially useful in the fight against these bacteria (46). Results from these studies should be available soon.
MATERIALS AND METHODS
Chemical synthesis. (i) Chemicals and instrumentation.Primaquine and chloroquine biphosphates, as well as mepacrine hydrochloride, were obtained from Sigma-Aldrich. The cinnamic acids were all from Acros Organics. The coupling agent 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethylaminium tetrafluoroborate (TBTU) was from Bachem, and all other chemicals were from Sigma-Aldrich. All solvents were of analytical grade and were purchased from VWR International. Nuclear magnetic resonance analyses were carried out on a Bruker Avance III 400 MHz spectrometer, and samples were prepared in either deuterated chloroform (CDCl3) or deuterated dimethyl sulfoxide [(CD3)2SO] with tetramethylsilane as an internal reference. Chemical shifts are reported downfield in parts per million (ppm), and the multiplicity of proton signals is indicated as s (singlet), d (doublet), dd (double doublet), t (triplet), or m (unresolved multiplet). High-resolution mass spectra were obtained on an LTQ Orbitrap XL/LTQ Tune Plus 2.5.5 spectrometer and processed using 2.1.0 Xcalibur software (Thermo Scientific). Compound purity was based on peak areas obtained through high-pressure liquid chromatography analyses that were run under the following conditions: 0 to 100% of solvent B in solvent A (A = H2O with 0.05% of trifluoroacetic acid; B = acetonitrile) in 30 min with a flow rate of 1 ml/min on a Merck-Hitachi Lachrom Elite instrument equipped with a diode array detector and thermostated (Peltier effect) autosampler, using a Purospher STAR RP-18e column (150 mm by 4.0 mm; particle size, 5 μm).
(ii) Synthesis of primaquine derivatives 1a, 1b, 1′a, and 1′b.N-Cinnamoyl-primaquine derivatives 1a and 1b were prepared by previously described methods, and their analytical and structural data were consistent with those formerly reported (27). Similarly, the synthesis of the ionic counterparts of compounds 1a and 1b (ionic liquids 1′a and 1′b, respectively) was carried out as reported earlier, and the compounds' structural and analytical data concurred with those previously published (28).
(iii) Synthesis of N-acyl-chloroquine (compounds 2a to 2j) and of N-acyl-mepacrine (compound 3a) analogues.The chemical synthesis procedures for compounds 2a to 2j and compound 3a were similar and are in accord with previously reported procedures, yielding the expected compounds, whose structural and analytical data agreed with those earlier described (31, 32, 35, 36, 47).
(iv) Synthesis of the new chloroquine/cinnamic acid-derived ionic liquids 2′a and 2′d.Organic salts 2′a and 2′d, which are representative of the ionic counterparts of compounds 2a to 2g, were prepared as follows. Chloroquine was first obtained as a free base, by adding triethylamine (0.25 ml) to a suspension of chloroquine biphosphate (0.1298 g; 0.285 mol) in dichloromethane (4 ml). The mixture was stirred for 30 min in ice and in the dark, and then the organic layer was washed with water (three times using 10 ml each time), dried over anhydrous sodium sulfate, and evaporated to dryness under reduced pressure. Free chloroquine base, quantitatively obtained as a pale yellow oil, was next dissolved in methanol and titrated with an equimolar amount of the relevant cinnamic acid upon dropwise addition of a methanolic solution of the acid. The neutralization reaction proceeded in the dark at room temperature for 30 min; the methanol was removed by evaporation at reduced pressure, and the target organic salts 2′a and 2′d were obtained as chromatographically homogeneous colorless viscous liquids whose spectral data were in agreement with the target structures, as follows.
2-((7-Chloroquinolin-4-yl)amino)-N,N-diethylpropan-1-aminium cinnamate (compound 2′a).Colorless oil (51 mg, 81%) δH (400 MHz, DMSO-d6) 8.37 (s, 1H), 8.35 (d, 1H, J = 4.4 Hz), 7.76 (d, 1H, J = 2.2Hz), 7.66 (m, 2H), 7.55 (d, 1H, J = 16.0 Hz), 7.42 (m, 5H), 6.90 (d, 1H, J = 7.9 Hz), 6.52 (m, 2H), 2.42 (m, 6H), 1.69 (m, 1H), 1.50 (m, 4H),1.23 (d, 4H, J = 6.3 Hz), 0.91 (t, 6H, J = 7.1 Hz); δC (100 MHz, DMSO-d6) 167.8, 151.8, 149.5, 149.2, 143.2, 134.4, 133.3, 130.0, 128.8, 128.0, 127.4, 124.3, 123.8, 120.1, 117.5, 98.8, 51.9, 47.5, 46.1, 33.3, 23.0, 19.8, 11.2; (EI+) m/z calculated for C18H27ClN3+: 320.19, found 320.33; (EI−) m/z calculated for C9H7O2− 147.05, found 147.33.
2-((7-Chloroquinolin-4-yl)amino)-N,N-diethylpropan-1-aminium p-methoxycinnamate (compound 2′d).Colorless oil (52 mg, 87%) δH (400 MHz, DMSO-d6) 8.37 (s, 1H), 8.34 (d, 1H, J = 4.4Hz), 7.76 (d, 1H, J = 2.2Hz), 7.62 (d, 2H, J = 8.8 Hz), 7.51 (d, 1H, J = 15.9 Hz), 7.42 (dd, 1H, J = 9.0 Hz, J = 2.2 Hz), 6.96 (d, 2H, J = 8.8Hz), 6.89 (d, 1H, J = 8.0 Hz), 6.50 (d, 1H, J = 5.6 Hz), 6.36 (d, 1H, J = 15.9 Hz), 3.79 (s, 3H), 2.40 (m, 6H) 1.69 (m, 1H), 1.50 (m, 4H), 1.22 (m, 4H), 0.91 (t, 6H, J = 7.1 Hz); δC (100 MHz, DMSO-d6) 167.9, 160.8, 151.9, 149.5, 149.3, 143.4, 133.3, 129.8, 127.4, 126.9, 124.3, 123.7, 117.5, 116.9, 114.3, 98.8, 55.3, 52.0, 47.5, 46.1, 33.3, 23.3, 19.8, 11.5; (EI+) m/z calculated for C18H27ClN3+: 320.19, found 320.35; (EI−) m/z calculated for C10H9O3− 177.06, found 177.05.
In vitro assays. (i) Compound preparation.Test compounds were stored at −20°C without exposure to light until use. Immediately prior to the first screening, compounds were solubilized in 100% dimethyl sulfoxide at a concentration of 20 mg/ml. For testing, the compounds were diluted directly into culture medium at dilutions of 100, 10, 1, and 0.1 μg/ml. Culture medium consisted of RPMI 1640 containing 20% horse serum, 1% minimal essential medium (MEM) vitamin solution, 1% MEM nonessential amino acid solution (NEAA), 200 U/ml penicillin, 0.2 mg/ml streptomycin, and 50 μg/ml vancomycin. Negative controls were medium alone and medium with 10 μg/ml ampicillin. The positive control included 1 μg/ml pentamidine isethionate. Medium with vehicle alone (dimethyl sulfoxide) was tested at the highest dilution to identify any associated toxicity.
(ii) P. carinii ATP assay.P. carinii viability was assessed through a nonspecific ATP assay previously validated for this purpose (42–44, 48, 49). Briefly, cryopreserved and characterized P. carinii isolated from rat lung tissue was distributed into triplicate wells of 48-well plates with a final volume of 500 μl and final concentrations of 5 × 107 P. carinii nuclei/ml. Controls and compound dilutions were added, followed by incubation at 36°C and 5% CO2. At 24, 48, and 72 h, 10% of the well volume was removed, and the ATP content was measured using a Perkin-Elmer ATP-liteM luciferin-luciferase assay. The bioluminescence generated by the ATP content of the samples was measured with a BMG POLARstar OPTIMA spectrophotometer and expressed as relative light units. A sample of each group was examined microscopically on the final assay day to rule out the presence of bacteria or other fungi. A quench control assay to determine compound interference in the luciferin/luciferase revealed quenching with six compounds at the highest (100 μg/ml) concentration. No quench was observed at 50 μg/ml. Test concentrations of these compounds were lowered accordingly.
(iii) L2 and A549 toxicity assay.For mammalian cell toxicity testing, A549 cells were cultured in Dulbecco's modified Eagle medium and L2 cells with F-12 medium. Both were supplemented with 10% fetal calf sera, 1× MEM vitamins, and 1× NEAA. Cultured cells were plated at 2 × 105/ml and grown to confluent monolayers. The medium was removed and replaced with fresh medium containing controls and test compound at appropriate dilutions. Assays of three time points (24, 48, and 72 h) with triplicate wells were sampled. The medium was aspirated from the wells. Adherent cells were stained with MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] reagent (thiazolyl blue), followed by incubation for 1 h. The MTT reagent was aspirated from the cells, and MTT solvent was added to the stained cells, followed by rocking at room temperature until the crystals were dissolved. The plate was read in absorbance mode on a BMG POLARstar Optima spectrophotometer.
(iv) Calculations.For the P. carinii ATP assay, the background bioluminescence was subtracted, and triplicate well readings were averaged. Using EXCEL software, the percent reduction in ATP for all groups was calculated (vehicle control − experimental/vehicle control × 100). The results of triplicate assays were averaged, and the standard deviations were determined. For the cell toxicity assay, triplicate well readings were averaged. For each day's readings, the percent reduction in absorbance for all groups was calculated as follows: medium control − experimental/medium control × 100. In both assays, the IC50 was calculated using the InSTAT linear regression program.
ACKNOWLEDGMENTS
We thank Fundação para a Ciência e Tecnologia (FCT, Portugal) for funding Research Unit LAQV-REQUIMTE (UID/QUI/50006/2013) and for a doctoral grant to A.G. (PD/BD/135073/2017). We also thank the Comissão de Coordenação e Desenvolvimento do Norte (CCDR-N, Porto, Portugal) for funding project DesignBiotechHealth (Norte-01-0145-FEDER-000024). Studies performed in Cincinnati were supported by the Cincinnati Veterans Affairs Medical Center and the Cincinnati Education and Research for Veterans Foundation. M.T.C. received a VA Research Career Scientist Award.
FOOTNOTES
- Received 11 May 2018.
- Returned for modification 12 July 2018.
- Accepted 19 August 2018.
- Accepted manuscript posted online 10 September 2018.
- Copyright © 2018 American Society for Microbiology.
