https://orcid.org/0000-0002-4870-3202
Abstract
Miltefosine is currently the only oral drug for visceral leishmaniasis, and although deficiency in an aminophospholipid/miltefosine transporter (MT) is sufficient to elicit drug resistance, very few naturally miltefosine-resistant (MIL-R) strains have yet been isolated. This study aimed to make a detailed analysis of the impact of acquired miltefosine resistance and miltefosine treatment on in vivo infection.
Bioluminescent versions of a MIL-R strain and its syngeneic parental line were generated by integration of the red-shifted firefly luciferase PpyRE9. The fitness of both lines was compared in vitro (growth rate, metacyclogenesis and macrophage infectivity) and in BALB/c mice through non-invasive bioluminescence imaging under conditions with and without drug pressure.
This study demonstrated a severe fitness loss of MT-deficient parasites, resulting in a complete inability to multiply and cause a typical visceral leishmaniasis infection pattern in BALB/c mice. The observed fitness loss could not be rescued by host immune suppression with cyclophosphamide, whereas episomal reconstitution with a wild-type MT restored parasite virulence, hence linking parasite fitness to MT mutation. Remarkably, in vivo miltefosine treatment or in vitro miltefosine pre-exposure significantly rescued MIL-R parasite virulence. The in vitro pre-exposed MIL-R promastigotes showed a longer and more slender morphology, suggesting an altered membrane composition.
The profound fitness loss of MT-deficient parasites most likely explains the low frequency of MIL-R clinical isolates. The observation that miltefosine can reverse this phenotype indicates a drug dependency of the MT-deficient parasites and emphasizes the importance of resistance profiling prior to miltefosine administration.
Introduction
Miltefosine was introduced in 2002 as the first and only oral drug for visceral leishmaniasis (VL) with a reported 6 month cure rate of 94%.1 Upon its broader use, treatment failures (TF) became increasingly frequent with recent reports indicating cure rates of 73% in India for Leishmania donovani2,3 and only 43% in Eastern Africa for L. infantum infections.3,4 Different factors play a role, including age and gender5 or low drug exposure.6 Some reports also link miltefosine-TF to a decreased miltefosine susceptibility of the parasite,4,7,8 while others question this.2,9 Although miltefosine’s long half-life (T1/2 150–200 h)10 in combination with the long and unsupervised treatment regimen of 28 days2,7,11 puts miltefosine at considerable risk for selection of drug-resistant parasites, surprisingly only four naturally fully miltefosine-resistant (MIL-R, IC50 >20 μM) strains have yet been identified: two L. infantum12,13 and two L. donovani isolates.14 Despite the scarce number of MIL-R clinical isolates, MIL-R strains have been experimentally selected in the laboratory and found to display similar characteristics.12,15–17 Previous experiments performed in our lab indicated that acquisition of full miltefosine resistance is accompanied by a fitness loss, which may explain the actual low prevalence of MIL-R strains in the field.18
Mutations in the miltefosine transporter (MT) or Ros3 subunit genes are the basis of miltefosine resistance in L. infantum and L. donovani.12–14,16,19,20 The LdMT/LiMT (MT) proteins belong to the P4-type ATPase subfamily and are responsible for the uptake of miltefosine and other phospholipids (phosphatidylcholine, phosphatidylethanolamine and phosphatidylserine) through a flippase mechanism.17 To be active in the plasma membrane, MT requires the functional β-subunit Ros3, which belongs to the CDC50/Lem3 protein family.17 As MT and Ros3 are mutually important for function and correct subcellular localization, alterations in either protein will therefore impact on phospholipid influx, parasite metabolism and membrane composition19,21,22 and potentially affect parasite fitness.
In the present study, the in vivo course of infection of an L. infantum WT (parent) strain and a syngeneic MIL-R counterpart were studied using bioluminescence imaging (BLI) in BALB/c mice, which is a widely used animal model for VL23–25 that is particularly suited for adopting the BLI technique.26 The observed severe fitness loss and the inability of MIL-R parasites to multiply in vivo, was found to be linked to MT deficiency. Remarkably, miltefosine itself was able to rescue partially this fitness loss in vivo.
Materials and methods
Ethics
The use of laboratory rodents was carried out in strict accordance to all mandatory guidelines (European Union directive 2010/63/EU on the protection of animals used for scientific purposes and the Declaration of Helsinki) and was approved by the ethics committee of the University of Antwerp (UA-ECD 2015–90).
Animals
Female BALB/c, C57Bl/6 and Swiss mice (6–8 weeks old; Janvier) were used for intraperitoneal starch stimulation and collection of peritoneal macrophages. Animals were randomly allocated and housed per experimental group in individually ventilated cages with environmental enrichment. Food for laboratory rodents (Carfil) and drinking water were available ad libitum. Animal welfare during the experiments was assessed daily using the ‘functional observation battery’ and body weight monitoring.
Leishmania parasites and transfection
Three L. infantum strains were used: (i) LEM3323 WT (MHOM/FR/96/LEM3323), a field isolate (CNRL, Montpellier France) from a French HIV patient;15 (ii) LEM3323 MIL-R, harbouring a frameshift mutation in the LiMT gene conferring miltefosine resistance;12 and (iii) a LEM3323 MIL-RLiMT rescue line with an episomal WT LiMT gene copy.12 The WT and MIL-R strain were transfected with the red-shifted firefly luciferase gene variant PpyRE927 that was codon optimized for expression in Leishmania (GenScript). For integration in the pLEXSY-hyg2 vector (Jena Bioscience), NcoI and NotI restriction sites were added to the 5′ and 3′ end respectively (GenScript). Twenty micrograms of pLEXSY-hyg2-PpyRE9 was digested with SwaI and the 6.8 kb fragment was gel-purified using the QIAquick Gel Extraction Kit (Qiagen) for transfection into parasites. For each strain, 1 × 108 procyclic promastigotes were electroporated (twice with 25 μF at 1500 V) in the presence of 10 μg linearized construct using the Bio-Rad GenePulse Xcell electroporation unit. Transfectants were selected under hygromycin B (Hyg) pressure and subcultured twice weekly in HOMEM promastigote medium (Life Technologies).
Clonal selection of transfected parasites
The presence and integration of the PpyRE9 gene was confirmed in monoclonal lines by conventional DNA PCR using primers targeting either PpyRE9 (FP: 5′-GATGAACATCTCCCAGCCGA-3′, RP: 5′-GGTAGTCCGTCTTGCTGTCC-3′) or the integration site (Jena Bioscience, F3001 FP and A1715 RP). Next, the in vitro light-producing capacity of all promastigote clones was compared using the ONE-Glo™ Luciferase Assay System (Promega). Luminescence [in relative luminescence units (RLU) p/s/cm2/Sr] was measured using the GloMax Explorer (Promega) after addition of 50 μL ONE-Glo substrate to an equal volume of promastigotes. The three most-light-producing clones were evaluated for their in vitro and in vivo infectivity and light-producing capacities. For the in vitro part, infections were evaluated in primary peritoneal mouse macrophages as described elsewhere.28,29 The infection ratio was determined by microscopy and luminescence measurement with the ONE GloTM Luciferase Assay System. To assess the in vivo infectivity, two BALB/c mice per clone were infected intravenously (iv) with 1 × 108 metacyclic promastigotes. BLI at 2, 6, 9 and 12 weeks post-infection (wpi) was performed after intraperitoneal (ip) injection of 150 mg/kg d-Luciferin (Beetle Luciferin Potassium Salt; Promega) in the IVIS® Spectrum In Vivo Imaging System (Perkin Elmer) under 2% isoflurane anaesthesia. Images were analysed using LivingImage v4.3.1 within organ-specific regions of interest. The selected representative clones are referred to as WTPpyRE9 and MIL-RPpyRE9. Stability of the integrated PpyRE9 expression cassette in the absence of antibiotic pressure was monitored over an extended period by luminescence detection and PCR.
Effect of transfection on in vitro strain infectivity and drug susceptibility profile
Drug susceptibility29 of the parent and the PpyRE9-transfected clones was compared for four reference drugs: pentavalent antimonials (SbV), trivalent antimonials (SbIII), miltefosine and paromomycin (highest concentration of 77 μg/mL, 88 μg/mL eq, 20 μM and 500 μM respectively). IC50 values were determined following Giemsa staining. Drug susceptibility of procyclic promastigotes was assessed upon incubation at 25°C for 72 h and addition of resazurin for 24 h before fluorescence reading.
Limit of detection of the BLI technique
In vitro
The limit of detection (LOD) was investigated with the IVIS® System using two substrates: the lysing ONE-GloTM Luciferase (Promega) and the non-lysing d-Luciferin (Promega). Luminescence was recorded over 1 min from 104, 103, 500, 100, 50, 10 and 1 procyclic promastigote(s) in 100 μL following the 1:1 addition of the ONE-GloTM substrate or the addition of 10 μL d-Luciferin (30 mg/mL).
In vivo
BALB/c mice were infected either iv or intradermally (id) with a 1:10 dilution series of WTPpyRE9 or MIL-RPpyRE9 promastigotes. A range from 108 to 105 or from 105 to 103 parasites were used for the iv and id infections respectively. Imaging was performed at 1 h post-infection (hpi) and 1 day post-infection (dpi).
In vitro parasite fitness analyses
Promastigote growth
The growth profile of WTPpyRE9 and MIL-RPpyRE9 parasites was assessed at different miltefosine concentrations (0, 1, 7, 20 and 40 μM). Promastigotes were quantified microscopically using a KOVA Glasstic counting slide (KOVA International) and by luminescence measurement using ONE-GloTM substrate.
Promastigote metacyclogenesis
Promastigote morphology was evaluated microscopically for WT, MIL-R, MIL-R + 40 μM miltefosine and MIL-RLiMT to assess metacyclogenesis. The flagellum/cell body length ratio of 100 Giemsa-stained promastigotes was determined by bright-field microscopy (Axiovert 200m®; Carl Zeiss) using the Zeiss Axiocam MRm® and the Zen® software.
Promastigote infectivity and intracellular replication
The in vitro infectivity and amastigote multiplication of WT, MIL-R and MIL-RLiMT was compared by Giemsa staining every 24 h.
In vivo parasite fitness analyses
Animal infections
BALB/c and C57Bl/6 mice were infected iv with 1 × 108 metacyclic promastigotes. Infections in immunosuppressed mice were following ip injection of 150 mg/kg Endoxan® (Baxter) at 2 days before infection followed by weekly injections. In case of miltefosine treatment, the drug was administered by oral gavage for 5 days at 40 mg/kg single dose starting from 3 dpi. In some experiments, parasites were pre-treated for 7 days with 40 μM miltefosine. Enrichment of metacyclic promastigotes was performed using peanut lectin (Sigma–Aldrich) agglutination.30 Coinfection studies were conducted with WTPpyRE9/MIL-R and WT/MIL-RPpyRE9. All animals were followed-up using BLI, greatly reducing the number of experimental animals and permitting more accurate monitoring of parasite dispersal.31,32 At final timepoints, organ parasite burdens were determined by microscopic evaluation of Giemsa-stained tissue imprints using the Stauber index.33 Promastigote back-transformation was performed in HOMEM medium at 25°C. Determination of organ burdens was also performed by Spliced Leader (SL)-RNA quantitative PCR (qPCR) as described elsewhere.34
Statistical analyses
Appropriate statistical tests including Mann–Whitney U, Kruskal–Wallis and two-way ANOVA were performed in GraphPad Prism 6. A mixed model ANOVA test was run in IBM SPSS Statistics v23. A 95% CI was used for all statistical tests.
Results
Generation of syngeneic bioluminescent reporter lines with a differential miltefosine susceptibility
Monoclonal reporter lines were generated for respectively the LEM3323 WT and the LEM3323 MIL-R (MT-deficient) strain, harbouring a genomic integrated copy of the PpyRE9 gene (Figure S1, available as Supplementary data at JAC Online). The light-producing capacity of both reporter lines was comparable (Figure S2A). No significant differences in drug susceptibility (Table 1) and macrophage infectivity indices (WT: 9.41 ± 0.22; MIL-R: 14.88 ± 4.99; WTPpyRE9: 9.91 ± 0.48; MIL-RPpyRE9: 13.25 ± 4.94) were detected between the parent and PpyRE9-transfected line.
Drug susceptibility profiles (IC50 values) of promastigotes and amastigotes of parent and PpyRE9-transfected WT and MIL-R lines
| Drug/strain | Amastigote assay | Promastigote assay | ||
|---|---|---|---|---|
| WT | MIL-R | WT | MIL-R | |
| SbV (eq./mL) | ||||
| parent | >77 | >77 | >77 | >77 |
| PpyRE9-transfected | >77 | >77 | >77 | >77 |
| SbIII (eq./mL) | ||||
| parent | >44 | >44 | >88 | >88 |
| PpyRE9-transfected | >44 | >44 | >88 | >88 |
| Miltefosine (µM) | ||||
| parent | 0.42 ± 0.03 | >40 | 8.19 ± 0.41 | >40 |
| PpyRE9-transfected | 0.63 ± 0.08 | >40 | 7.73 ± 0.55 | >40 |
| Paromomycin (µM) | ||||
| parent | 61.03 ± 7.44 | 50.53 ± 3.46 | 183.82 ± 11.49 | 156.74 ± 13.45 |
| PpyRE9-transfected | 60.18 ± 5.43 | 54.78 ± 6.35 | 147.32 ± 16.16 | 105.62 ± 6.22 |
| Drug/strain | Amastigote assay | Promastigote assay | ||
|---|---|---|---|---|
| WT | MIL-R | WT | MIL-R | |
| SbV (eq./mL) | ||||
| parent | >77 | >77 | >77 | >77 |
| PpyRE9-transfected | >77 | >77 | >77 | >77 |
| SbIII (eq./mL) | ||||
| parent | >44 | >44 | >88 | >88 |
| PpyRE9-transfected | >44 | >44 | >88 | >88 |
| Miltefosine (µM) | ||||
| parent | 0.42 ± 0.03 | >40 | 8.19 ± 0.41 | >40 |
| PpyRE9-transfected | 0.63 ± 0.08 | >40 | 7.73 ± 0.55 | >40 |
| Paromomycin (µM) | ||||
| parent | 61.03 ± 7.44 | 50.53 ± 3.46 | 183.82 ± 11.49 | 156.74 ± 13.45 |
| PpyRE9-transfected | 60.18 ± 5.43 | 54.78 ± 6.35 | 147.32 ± 16.16 | 105.62 ± 6.22 |
IC50 values are presented as mean ± SD and are the result of three independent repeats.
There were no significant differences between the parent and transfected lines, both on promastigote or amastigote, for four reference drugs: pentavalent (SbV) and trivalent (SbIII) antimonials, miltefosine (MIL) and paromomycin (PMM).
Drug susceptibility profiles (IC50 values) of promastigotes and amastigotes of parent and PpyRE9-transfected WT and MIL-R lines
| Drug/strain | Amastigote assay | Promastigote assay | ||
|---|---|---|---|---|
| WT | MIL-R | WT | MIL-R | |
| SbV (eq./mL) | ||||
| parent | >77 | >77 | >77 | >77 |
| PpyRE9-transfected | >77 | >77 | >77 | >77 |
| SbIII (eq./mL) | ||||
| parent | >44 | >44 | >88 | >88 |
| PpyRE9-transfected | >44 | >44 | >88 | >88 |
| Miltefosine (µM) | ||||
| parent | 0.42 ± 0.03 | >40 | 8.19 ± 0.41 | >40 |
| PpyRE9-transfected | 0.63 ± 0.08 | >40 | 7.73 ± 0.55 | >40 |
| Paromomycin (µM) | ||||
| parent | 61.03 ± 7.44 | 50.53 ± 3.46 | 183.82 ± 11.49 | 156.74 ± 13.45 |
| PpyRE9-transfected | 60.18 ± 5.43 | 54.78 ± 6.35 | 147.32 ± 16.16 | 105.62 ± 6.22 |
| Drug/strain | Amastigote assay | Promastigote assay | ||
|---|---|---|---|---|
| WT | MIL-R | WT | MIL-R | |
| SbV (eq./mL) | ||||
| parent | >77 | >77 | >77 | >77 |
| PpyRE9-transfected | >77 | >77 | >77 | >77 |
| SbIII (eq./mL) | ||||
| parent | >44 | >44 | >88 | >88 |
| PpyRE9-transfected | >44 | >44 | >88 | >88 |
| Miltefosine (µM) | ||||
| parent | 0.42 ± 0.03 | >40 | 8.19 ± 0.41 | >40 |
| PpyRE9-transfected | 0.63 ± 0.08 | >40 | 7.73 ± 0.55 | >40 |
| Paromomycin (µM) | ||||
| parent | 61.03 ± 7.44 | 50.53 ± 3.46 | 183.82 ± 11.49 | 156.74 ± 13.45 |
| PpyRE9-transfected | 60.18 ± 5.43 | 54.78 ± 6.35 | 147.32 ± 16.16 | 105.62 ± 6.22 |
IC50 values are presented as mean ± SD and are the result of three independent repeats.
There were no significant differences between the parent and transfected lines, both on promastigote or amastigote, for four reference drugs: pentavalent (SbV) and trivalent (SbIII) antimonials, miltefosine (MIL) and paromomycin (PMM).
Characterization of the in vitro bioluminescent reporter detection
The IVIS® Spectrum detected a single parasite with the lysing ONE-GloTM Luciferase and 50 parasites with the non-lysing d-Luciferin for both WTPpyRE9 and MIL-RPpyRE9 (Figure S2A). Both assays showed a linear correlation between promastigote numbers and emitted luminescence (Figure S2B). The rapidly dividing log-phase promastigotes were more luminescent than metacyclic promastigotes, whereas intracellular amastigotes emitted the least (P ≤ 0.05) of all life cycle stages (Figure S2C). Luminescence of WTPpyRE9 and MIL-RPpyRE9 remained stable in promastigote cultures without selection pressure for at least 3 months, but rapidly declined afterwards (Figure S3A). The reverting reporter lines were cloned to investigate clonal differences in light production, presence of the PpyRE9 gene in the genome and Hyg resistance. Most clones (eight WT and eight MIL-R clones) lost the PpyRE9 and Hyg resistance genes, two WT clones retained the same characteristics as the original reporter line and two other WT clones showed an intermediate luminescence, PpyRE9 presence and Hyg resistance (Figure S3B, C).
Characterization of the in vivo bioluminescent reporter detection
Despite the observed loss of the transgene after long-term in vitro culture, bioluminescence remained stable for at least 5 months in vivo. The in vivo LOD depended on life cycle stage and the origin of the signal from superficial or deep visceral organs. Intravenously administered parasites nearly exclusively accumulate in the liver whereby BLI at 1 dpi indicated an LOD of ∼107 amastigotes (Figure 1a,b). In the spleen, an LOD of 5 × 104 amastigotes was estimated from infection experiments that compared BLI and microscopic or molecular detection methods. The signal of an intradermal inoculum of 104 parasites could be reliably detected in the ear at 1 dpi (Figure 1c,d). For all target organs, a significant positive correlation [P ≤ 0.01; Spearman correlation coefficients (rs): 0.741–0.935] was found between microscopy, SL-RNA qPCR and the BLI technique (Figure S4).
In vivo LOD at 1 h and 1 day after iv (a, b) and id infection (c, d). (a, c) Representative BLI pictures using an exposure time of 15 min. (b, d) RLU values in liver (b) and ear (c), the background RLU value is depicted as a red dotted line. (a, b) At 1 hpi, 106 parasites could be detected for both WTPpyRE9 and MIL-RPpyRE9 in the liver, while at 1 dpi (amastigote stage), the LOD was 107 amastigotes for both strains. (c, d) At 1 hpi, 103 parasites could just be detected, while at 1 dpi (amastigote stage), the LOD was 104 amastigotes for both lines. Two BALB/c mice were used per inoculum size for each strain, from which one representative mouse was shown. Twelve mice were used for the assessment of the LOD upon iv infection, another 12 for the same experiment upon id infection. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
MIL-R acquisition and exposure to miltefosine modify the in vitro parasite characteristics
Promastigote growth
MIL-RPpyRE9 promastigotes could cope with up to 40 μM miltefosine without impact on growth and light production (Figure S5). MIL-RPpyRE9 attained lower maximal densities (5.98 × 107/mL ± 9.19 × 106 versus 7.71 × 107/mL ± 1.07 × 107) and was accompanied by a significantly lower luminescence (4.91 × 107 ± 2.15 × 106 versus 7.00 × 107 ± 1.45 × 106, P ≤ 0.05).
Parasite morphology and metacyclogenesis
The MIL-R strain showed a significantly lower rate of metacyclogenesis and did not reach a flagellum/cell body ratio of ≥2 as its cell body remained significantly longer (P ≤ 0.0001, Figure 2a,b). Episomal rescue of MIL-RLiMT enhanced metacyclogenesis (P ≤ 0.001, Figure 2b), with cell body lengths intermediate between those of WT and MIL-R and achieving an average flagellum/cell body ratio of ≥2 after 240 h (Figure 2a,b). Remarkably, in vitro exposure of MIL-R promastigotes to miltefosine modified the parasite morphology into a form with a longer cell body (P ≤ 0.0001, Figure 2a,c), hence strongly affecting the morphometric quantification of metacyclics.
Cell body length (a), flagellum/cell body length ratio (b), morphology (c) and percentage metacyclics (d) of WT, MIL-R, MIL-R + 40 μM MIL and MIL-RLiMT. (a) Cell body of WT shortened as a function of time and was significantly smaller over all timepoints. Episomal rescue with LiMT in MIL-RLiMT parasites resulted in an intermediate cell body length. Addition of MIL to the MIL-R parasites induced a morphological change featuring a significantly longer cell body. (b) WT showed a significantly higher rate of metacyclogenesis and reached an average flagellum/cell body ratio of ≥2 within 216 h in vitro culture. MIL-R line did not reach a ratio of ≥2 and addition of MIL resulted in a reduction of the flagellum/cell body ratio. Episomal LiMT rescue in the MIL-RLiMT strain improved metacyclogenesis with average flagellum/cell body ratio of ≥2 attained after 240 h. (c) MIL-R promastigotes at 144 h in culture without (1) and with addition of 40 μM MIL (2): upon MIL addition, MIL-R parasites morphologically change to a longer and more slender shape (×63, Giemsa staining). (d) MIL-RLiMT showed an intermediate percentage of metacyclics. (a, b, d) At least 50 parasites were measured per timepoint. Data are summarized as mean ± SD, *P ≤ 0.05, ***P ≤ 0.001 and ****P ≤ 0.0001. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Macrophage infectivity
Both WT and MIL-R were able to infect and multiply in vitro in primary macrophages. At all timepoints, MIL-R amastigotes showed a significantly lower rate of multiplication (Figure 3c) compared with the WT counterparts. The episomally rescued MIL-RLiMT showed slightly enhanced infectivity characteristics (Figure 3a,b).
Percentage of infected macrophages (a), infection ratio (b) and amastigote multiplication (c) upon infection with WT, MIL-R or MIL-RLiMT promastigotes. (a) Percentage of infected macrophages did not significantly differ between WT, MIL-R and MIL-RLiMT, although it was slightly higher for WT. (b) Infection with WT resulted in a significantly higher macrophage infection ratio compared with the MIL-R, while the MIL-RLiMT showed an intermediate infection ratio. (c) Intracellular amastigotes increased during the first 72 h for both strains, although WT multiplied to significantly higher burdens than MIL-R. Results are depicted as the mean ± SD of four (a, b) or two (c) independent repeats, *P ≤ 0.05, **P ≤ 0.01.
MIL-R acquisition due to MT deficiency impacts on in vivo infectivity
Although the MIL-R strain was able to multiply in vitro in macrophages, it failed to infect successfully the BALB/c mice. Infection with WTPpyRE9 resulted in a typical VL visceralization pattern with maximal liver and bone marrow burdens at 3 wpi, followed by a rise in splenic parasite load reaching a peak at 8–10 wpi (Figure 4). In contrast, MIL-RPpyRE9 failed to establish in the liver with a rapid decline to undetectable bioluminescence by 6 wpi. Spleen and bone marrow signals were only rarely detected very early in infection and did not increase over time, indicating a severe fitness loss (P ≤ 0.0001 for liver and bone marrow, P ≤ 0.01 for spleen) (Figure 4). The course of infection of WTPpyRE9 remained unaltered upon coinfection with MIL-R (Figure S6). As MIL-R parasites were less able to differentiate into metacyclics, an additional experiment was conducted to exclude that the defective infection could have resulted from a lower number of metacyclics in the infection inoculum. Peanut lectin agglutination was used to enrich metacyclic promastigotes and the infectious dose was normalized to the same number of metacyclic forms. Under those conditions, the same relative differences between WTPpyRE9 and MIL-RPpyRE9 in infection kinetics and parasite burdens were recorded (Figure S7) with significantly lower MIL-RPpyRE9 burdens in liver and bone marrow (P ≤ 0.05 and P ≤ 0.0001, respectively).
Infection course of WTPpyRE9 and MIL-RPpyRE9 in liver (a), spleen (b) and bone marrow (c) in BALB/c mice. WT caused a typical VL infection pattern with a peak in liver (a) and bone marrow (c) at 3 wpi, and a subsequent spleen peak ∼8 wpi (b). MIL-RPpyRE9 failed to visceralize with decreasing liver burdens from 1 dpi onwards and undetectable levels by 6 wpi (a). Any spleen and bone marrow signal was only detectable early in infection and did not increase over time (b, c). (a–c) MIL-RPpyRE9 amastigote burdens were significantly lower than those of WTPpyRE9 in all target organs. Results are pooled data from three different experiments with at least three animals per group. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001. Fitness loss phenotype was observed in 15 independent experiments, in BALB/c as well as C57Bl/6 mice, with at least three animals per group.
MIL-R-associated fitness loss cannot be rescued by cyclophosphamide-mediated host immune suppression
The infectivity of MIL-R was also evaluated under conditions of immune suppression with cyclophosphamide. WTPpyRE9 liver and bone marrow burdens were elevated over the course of infection in suppressed hosts (Figure 5). MIL-RPpyRE9 fitness was not restored upon cyclophosphamide treatment, but the liver signal still persisted until 12 wpi (Figure 5a, b). Nevertheless, parasite levels never rose above of those recorded at 1 dpi, indicating that the MIL-RPpyRE9 parasites were unable to multiply effectively in vivo despite cyclophosphamide treatment (Figure 5a–d).
Effect of host immune suppression (IS) on the infection course of WTPpyRE9 and MIL-RPpyRE9 in BALB/c mice. (a) BLI using an exposure time of 15 min in four groups (panels): WTPpyRE9 and MIL-RPpyRE9 in both normal and cyclophosphamide-treated (immune suppressed) mice. (b–d) Background-subtracted RLU values (ΔRLU) in the liver (b), spleen (c) and bone marrow (d). IS had a major effect on WTPpyRE9 liver burdens, which kept rising until the experimental endpoint and were significantly higher than in WTPpyRE9-infected control mice (a, b). Effect on WTPpyRE9 spleen and bone marrow burdens was not very pronounced (a, c, d). MIL-RPpyRE9 fitness was not restored upon IS as liver burdens still decreased from 1 dpi onwards and no spleen and bone marrow burdens could be detected (a–d). Generally, organ burdens in both groups of MIL-RPpyRE9-infected mice were significantly lower than those in both groups of WTPpyRE9-infected mice (b–d). One mouse in the cyclophosphamide-treated WTPpyRE9-infected group suffered from a prolapse and was therefore euthanized after 6 wpi (indicated with an X). Each experimental group consisted of three BALB/c mice, bringing the total number of animals for this experiment to 12. *P ≤ 0.05, **P ≤ 0.01 and ****P ≤ 0.0001. IS, immune suppression. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Inserting a functional LiMT gene in MIL-R parasites partially restores fitness
The role of LiMT deficiency in MIL-R-associated fitness loss was investigated by episomal reconstitution. Owing to the lack of a bioluminescent MIL-RLiMT reporter line, parasite burdens were determined by microscopic counting and SL-RNA qPCR at 3 and 10 wpi (corresponding to the previously recorded liver and spleen peaks respectively). At both timepoints and in both organs, MIL-RLiMT was consistently recovered by both quantification techniques and achieved organ burdens that were intermediate between those of WT and MIL-R infections (Figure 6). These results indicate a partial rescue of the MIL-R fitness loss by restoring LiMT function.
WT, MIL-R and MIL-RLiMT burdens in the liver (a, b) and spleen (c, d) at 3 and 10 wpi, determined by either microscopy (a, c) or SL-RNA qPCR (b, d). At both timepoints, in both organs and for both quantification techniques, MIL-RLiMT showed intermediate burdens between WT and MIL-R. Each experimental group and timepoint consisted of three BALB/c mice, bringing the total number of animals for this experiment to 18. *P ≤ 0.05; ND, not detected.
In vitro miltefosine pre-exposure and in vivo miltefosine treatment increase the fitness of the MIL-R strain
As miltefosine was found to induce morphological changes in MIL-R parasites (see Figure 2), the effects of in vivo miltefosine treatment and in vitro miltefosine preconditioning on MIL-R fitness were investigated in BALB/c mice. While miltefosine treatment was very effective in eliminating WTPpyRE9 to undetectable levels within 1 week after the end of treatment at 2 wpi (Figure S8), MIL-RPpyRE9 parasite burdens in liver (P ≤ 0.05) and bone marrow (P ≤ 0.001) increased after 5 days of miltefosine treatment (Figure 7). In vitro miltefosine pre-exposure also rendered MIL-RPpyRE9 parasites more infective (P ≤ 0.05, Figure 8). The increased expansion was most prominent in the bone marrow between 4 dpi and 1 wpi (P ≤ 0.01) with stably elevated levels up to 4 wpi (Figure 8a,d). Also in C57Bl/6 mice, miltefosine pre-exposure in combination with in vivo miltefosine treatment confirmed the fitness recovery with detection of significantly increased organ burdens (Figure S9).
Effect of in vivo MIL treatment on MIL-RPpyRE9 parasite fitness in BALB/c mice. (a) BLI using an exposure time of 15 min in two groups (panels): MIL-RPpyRE9 in untreated and MIL-treated mice. (b–d) Background-subtracted RLU values (ΔRLU) in the liver (b), spleen (c) and bone marrow (d). MIL-RPpyRE9 burdens in all target organs increased during the first round of MIL treatment (a–d; red arrow). In the absence of MIL, these burdens rapidly diminished again. Each experimental group consisted of six BALB/c mice, bringing the total number of animals for this experiment to 12 (no repeat). An independent repeat experiment was performed with three mice/group. *P ≤ 0.05, **P ≤ 0.01 and ***P ≤ 0.001. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Effect of in vitro miltefosine pre-exposure on MIL-RPpyRE9 fitness in BALB/c mice. (a) BLI using an exposure time of 15 min in two groups (panels): mice infected either with untreated (control) MIL-RPpyRE9 parasites or with pre-MIL MIL-RPpyRE9 parasites. (b–d) Background-subtracted RLU values (Δ RLU) in the liver (b), spleen (c) and bone marrow (d). In vitro pre-exposed parasites were more infective in vivo than the control MIL-RPpyRE9 line, marked by a significantly higher liver signal (a, b; orange arrow), significantly higher spleen burdens (a, c) and significantly increased bone marrow burdens (a, d; orange arrow). Each experimental group consisted of three BALB/c mice, bringing the total number of animals for this experiment to six. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001. Fitness gain upon miltefosine exposure was observed in four independent experiments, with at least three animals per group. This effect was also confirmed by conventional microscopic counting. pre-MIL, miltefosine-pretreated. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Discussion
The long elimination half-life10 and the long treatment schedule2,11 put miltefosine at considerable risk for the development of resistance. Until now, only four natural MIL-R strains have been isolated,13,14,35 which is surprising given that mutations in a single transporter complex are sufficient to confer full resistance.12–14,16,19 This study therefore investigated the impact of MIL-R on parasite fitness, i.e. the complex interaction in a given environment between different factors needed for survival, pathogenesis, reproduction and transmission.18,36,37
To be able to monitor closely the course of infection, bioluminescent versions of the WT and MIL-R strains (WTPpyRE9 and MIL-RPpyRE9) expressing the red-shifted PpyRE9 firefly luciferase27 were constructed. The LOD of the BLI technique was 107 amastigotes in the liver and 5 × 104 amastigotes in the spleen. In contrast, as few as 100 PpyRE9-expressing Trypanosoma brucei could be detected in vivo,38,39 probably reflecting the relative lower metabolic activity of amastigotes.40 When using a non-modified firefly luciferase in L. infantum, the LODs were substantially (4-fold) higher.32 In line with previous reports,32,41 an excellent correlation was found between the in vivo BLI signal and the parasite load in the major target organs.
The effect of MIL-R was very pronounced in vivo with a complete failure to visceralize and expand into the target organs. The BLI signal disappeared ∼6 wpi, evolving to undetectable parasite levels by 12 wpi on the basis of a very sensitive SL-RNA qPCR assay.34 In contrast, WTPpyRE9 produced a typical VL infection pattern with liver and bone marrow burdens peaking at 3 wpi, followed by a maximal spleen signal ∼8 wpi. Host immune suppression with cyclophosphamide could not restore MIL-R parasite multiplication but prolonged the presence of parasites in the organs. Episomal reconstitution with a functional LiMT copy restored MIL-R virulence in vivo, producing intermediate parasite burdens between the WT and MIL-R lines. The episomal rescue rather than chromosomal integration most likely explains why the rescue was partial. Nevertheless, these results strongly suggest a causal link between MIL-R fitness loss and LiMT mutation. Given that LiMT also transports phospholipids,17 deleterious mutations indeed seem to impact phospholipid influx, parasite metabolism and membrane composition19,21,22 resulting in an attenuated phenotype.
Evaluation under miltefosine pressure indicated that MIL-R promastigotes undergo morphological changes, showing a longer and more slender shape. This could result either from the incorporation of the alkylphospholipid drug into the membrane or from drug-induced alterations of the membrane composition/architecture. Making use of the functional green fluorescent miltefosine analogue BODIPY-miltefosine,42 this study also explored its potential for incorporation into the membrane. However, the bulky BODIPY group in the terminal position of the drug alkyl chain modified the characteristics and abrogated the capacity to induce morphological changes in the MIL-R line, whereas the WT parasites remained fully susceptible (data not shown). In MIL-S parasites, miltefosine is known to cause a decrease in several membrane phospholipids and amino acids, whereas sphingolipids and sterols increase.43 Effects on MIL-R parasites could be different as these lack a functional MT.44
Quite surprisingly, the treatment with miltefosine rescued the observed MIL-R fitness loss in all target organs. In addition, in vitro miltefosine pre-exposure of parasites alone or in combination with in vivo treatment resulted in pronounced effects. The phenotype of drug dependence of a drug-resistant microorganism has already been described for a streptomycin-resistant Mycobacterium tuberculosis,45 enteroviruses46 and an HIV-1 variant.47 To our knowledge, this has yet not been reported for Leishmania. Although the mechanism remains to be unravelled, it is plausible that plasma membrane compositional changes in the MIL-R parasite alter the interaction with the host’s (innate) immune system, resulting in an enhanced parasite control. Miltefosine interactions with the plasma membrane might functionally complement the parasite and render them less susceptible to immune recognition/elimination.
It would be of particular interest to investigate whether our findings apply to a broader panel of MIL-R Leishmania strains. Beside the impact of various SNPs in MT and/or Ros3, the impact of a rapid adaptive modification by aneuploidy at the level of chromosome 13 (harbouring the MT gene) deserves further exploration. Other causes of reduced miltefosine susceptibility, such as increased efflux through multidrug resistance-like protein (MRP) and/or reduced influx through ABCF2, may have an impact on parasite fitness. Indeed, L. donovani clinical isolates without MT/Ros3 mutations from miltefosine-TFs, showed an increased metacyclogenesis and in vitro macrophage infectivity.8 It is obviously worrying that fitness loss due to acquired resistance could be compensated by the drug itself, emphasizing the need for rational drug use. The observations with BODIPY-miltefosine suggest that it could be possible to synthetize miltefosine analogues without the detrimental features but with retained antiparasitic activity. Altogether, these observations re-emphasize the importance of miltefosine resistance profiling prior to miltefosine administration.
Acknowledgements
We thank Dr Luis Rivas for providing BODIPY-miltefosine, Dr Laurence Lachaud for the L. infantum LEM3323 strain, Dr Francisco Gamarro and Dr Santiago Castanys for providing the MIL-RLiMT and Florence Kauffmann and Dr Stefan Magez for initial support with the transfections. We also thank Pim-Bart Feijens, Rik Hendrickx and Mandy Vermont for their excellent technical assistance.
Funding
This work was supported by the Research Fund Flanders (FWO) by two projects (grant numbers G051812N and 12I0317N) and scholarships to E. E. (grant number 11V4315N), L. Van Bockstal (grant number 1136417N) and S. H. (grant number 12I0317N); a research fund of the University of Antwerp supporting G. C. (grant number TT-ZAPBOF 33049); and a Vlaamse Interuniversitaire Raad (VLIR) Global Minds Small Research Grants project. The donors had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. LMPH is part of European Cooperation in Science and Technology (COST) Action CM1307 (Targeted chemotherapy towards diseases caused by endoparasites) and is a partner of the Antwerp Drug Discovery Network (ADDN, www.addn.be) and the Excellence Centre ‘Infla-Med’ (www.uantwerpen.be/infla-med).
Transparency declarations
None to declare.
References
Author notes
Louis Maes and Guy Caljon authors made an equal contribution.
