Ethics statement

All animal studies were conducted in accordance with the GSK Policy on the Care, Welfare and Treatment of Laboratory Animals and were reviewed by the National Institute of Allergy and Infectious Diseases (NIAID) Animal Care and Use Committee (under the animal protocol LMVR4E).

Mice

Six-week-old female BALB/c mice were obtained from Charles River laboratories (catalog # BALB-F) and housed under pathogen-free conditions at the NIAID Twinbrook animal facility (Rockville, MD) at 18–23 °C room temperature, and 40–70% relative humidity, under 12 h dark/light cycles, and with water and food ad libitum.

Parasites

A cloned line of Leishmania major (WR 2885) was used⁴⁸. Promastigotes were maintained at 26 °C in Schneider’s insect medium supplemented with 20% heat-inactivated fetal bovine serum, 100 U/mL penicillin, and 100 mg/mL streptomycin (all Thermo Fischer Scientific).

Bacteria

An An. stephensi colony from GSK’s insectary at Global Health Medicines R&D site in Tres Cantos, Spain, gradually lost susceptibility to P. falciparum, but mosquito survival and reproductive ability were not compromised. Our collaborators isolated a Plasmodium refractoriness-promoting bacterium from the midguts of these mosquitoes⁵. Briefly, homogenized tissues from pupae, larvae, adult female mosquito midguts and water from larval breeding pans were transferred to sterile LB broth and incubated at 27 °C or at room temperature on a rotary shaker. When turbidity was observed, samples were plated on solid LB agar to obtain single colonies. Colonies were selected based on distinct morphology, transferred to fresh liquid media and subsequently identified both via biochemical characterization using API strips (Biomerieux) and via 16sRNA typing. A bacterium identified as Delftia tsuruhatensis was found to be predominantly present in all screened samples and the strain was designated as Delftia tsuruhatensis Tres Cantos 1 or TC1. A GFP-expressing Delftia tsuruhatensis was further generated and used for imaging purposes.

Sand flies

Phlebotomus duboscqi sand flies were mass-reared at the Laboratory of Malaria and Vector Research insectary as previously described in ref. ⁴⁹. Adult females were maintained on a 30% sucrose diet and were starved for 12 h before feeding.

Bacterial growth and preparation

D. tsuruhatensis was grown in LB liquid medium containing 50 µg/ml carbenicillin (both from Sigma-Aldrich, St. Louis, MO, USA) overnight at 28 °C under aeration conditions (250 rpm). The culture was then centrifuged at 4,000 × g for 5 min and washed 3 times with PBS. For bacterial feeding experiments the number of colony-forming units (CFUs) was adjusted via the measurement of the optical density (600 nm wavelength). The calculated volume was then re-suspended in sterile 5% sucrose solution. For experiments with killed bacteria (fig. S5), the bacterial suspension was incubated at 95 °C for 10 min prior to dilution in sucrose solution.

Bacterial supernatant preparation and fractionation

The bacterial excreted-secreted products were obtained as described elsewhere⁵. Briefly, D. tsuruhatensis TC1 bacteria were grown in LB liquid medium overnight (200 rpm, 28 °C), washed with PBS, resuspended in M9 medium (10⁹ CFUs/ml), and incubated (200 rpm, 28 °C) for 8 h. Afterward the TC1 supernatant was collected via a centrifugation step, and passed through a 0.22 μm filter. M9 culture medium without bacteria was subjected to exactly the same conditions and steps for the generation of the M9 control supernatant. For some experiments, we fractionated both the TC1 and M9 supernatants using sequential centrifugation steps in the context of Amicon Ultra ® Centrifugal filters (100 and 10 kDa cutoffs; Millipore Sigma, Burlington, MA, USA).

Sand fly bacterial (byproducts) feeding

P. duboscqi sand flies were allowed to feed on cotton rolls impregnated with a suspension of TC1 bacteria (WT or GFP-expressing; alive or dead), or E. coli, or an Ornithinbacillus massiliensis strain directly isolated from the midguts of sand flies (1 × 10⁸ CFU/ml in 5% sucrose solution) for 24 h; sand flies in the control group were given 5% sucrose alone. Bacterial feeding was done either one week prior to infection, 24 h prior to infection, 5 days post-infection, or 8 days post-infection. The bacterial supernatant was also given to sand flies in the same manner. In this case, sucrose was directly diluted into both TC1 or M9 control supernatants which served as the sugar vehicle. For these experiments, the supernatant was given to sand flies on a daily basis, after infection.

Sand fly infection

After an overnight starving period, sand flies were infected by artificial feeding through a chick membrane on defibrinated rabbit blood (Spring Valley Laboratories, MD, USA) containing L. major promastigotes (5 × 10⁶/ml), as previously described in ref. ⁵⁰. In some instances, 50 µM harmane (Sigma-Aldrich, St. Louis, MO, USA), or 50% (bacterial) supernatant were added to the infected blood mixture. After infection, blood-fed females were sorted and kept on a 30% sucrose diet. At 7-, and 11/12-days post infection sand flies were collected to assess the infection status. The number of parasites was determined under the optical microscope using a Neubauer chamber, and the thickness of the anterior midgut was measured under a Leica DFC 7000 T stereomicroscope using the software Leica applications suite X, v3.7.5.24914. Additionally, at 11 days post-infection, sand flies were used for transmission experiments.

Pick up experiments

BALB/c mice were infected intradermally in the ear pinnae with 1000 L. major metacyclic parasites and kept with water and food ad libitum until the development of typical CL lesions. Then sand flies were allowed to take a bloodmeal from the ears with active lesions using vials with a meshed surface held in place by custom-made clamps. Blood-engorged sand flies were separated into 2 groups. One group was then allowed to feed on the bacteria via sugar-meal overnight (1 × 10⁸/ml in 5% sterile sucrose solution), while the other received 5% sterile sucrose solution alone. Sand flies from both groups were dissected 11 days post-infection to evaluate their infection status.

Non-infected second bloodmeal

After the sand flies had taken an infected blood meal, either by feeding on a glass feeder or an infected animal, they were kept on a 30% sucrose diet for 12 days. The sand flies were then allowed to blood engorge on an anesthetized naïve mouse for one hour, as reported elsewhere⁵¹. Blood-fed females were sorted and kept on a 30% sucrose diet, and 6 days post feeding sand flies were collected to assess the infection status.

Sand fly infection status evaluation

Under a stereomicroscope, sand fly midguts were dissected in PBS and transferred to individual microtubes (Denville Scientific) with 50 μL of formalin solution (0.005% in PBS). Midguts were homogenized and 10 μL loaded onto disposable Neubauer chambers (Incyto). Slides were observed under a phase contrast microscope (Zeiss) at 400× magnification. The total parasite numbers, and the metacyclic frequency⁴⁸ were determined.

Fluorescence imaging

Under a stereomicroscope, sand fly midguts were dissected in PBS, over a glass slide, and covered with a coverslip. Phase-contrast and epifluorescence imaging was carried oved immediately after dissection. Only intact midguts were used for image acquisition. Images were taken using a K5 camera coupled to a Thunder Imager microscope (Leica Microsystems, Wetzlar, Germany) using the Leica Application Suite X software platform. Images were processed with the Imaris software v10.2.0 (Oxford Instruments, Abingdon, UK). Images were exported using the Imaris snapshot tool. If present, the grey contour around an image is from the software display screen, which is kept for frame size uniformity. Images were pseudo-colored since the camera is a black and white one, for increased sensitivity. Histograms were adjusted equally for the channels across all images. All raw images are available upon request.

Metagenomics analysis – layout and samples

Control or D. tsuruhatensis-fed bacteria sand fly midguts were dissected under a sterile-like environment at different times post bacterial/blood feeding: after bacterial feeding but before blood feeding (D0), and after bacterial and blood feeding (D2, D5, D7, D9, and D12). Dissected midguts were washed three times in sterile PBS drops and then transferred to an Eppendorf containing 50 μL of sterile PBS; pools of 20 midguts were collected per condition at least in triplicate. The midgut pools were then macerated using a motorized pestle (Kimble Chase, Vineland, New Jersey). Genomic DNA was finally extracted using the EZNA Tissue DNA Kit (D3396-01; Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer’s instructions, and the samples were subjected to 16S rRNA amplification and sequencing as reported elsewhere¹¹. Overall, a total of 42 samples were analyzed. For the analysis, frequently, samples were grouped by time-point.

Metagenomics analysis – amplicon sequence variant calling, phylogenetic tree, and taxonomy classification

A total of 7,758,941 raw reads were generated. 16S rRNA amplicon reads were demultiplexed and trimmed using Novogene in-house scripts. The trimmed demultiplexed reads were then imported into QIIME2 version 2021.4⁵² for downstream analysis. DADA2⁵³ was used to call amplicon sequence variants (ASVs). Chimeric sequences were identified and removed by setting the flag “–p-min-fold-parent-over-abundance” to 10, which discards chimeric sequences showing an at least ten times lower abundant than their parent sequences. After DADA2 quality filtering, a total of 6,911,632 reads (164,562 ± 4,808 reads/sample) were retained and 3735 ASVs were called. A rooted phylogenetic tree was generated using FastTree⁵⁴ based on ASV multiple alignment with MAFFT⁵⁵. The ASVs were taxonomically classified with a Naïve Bayes classifier pre-trained on SILVA rRNA database (release 138 SSURef NR99)⁵⁶. All code used to do the metagenomics analysis is fully available⁵⁷.

Metagenomics analysis – microbial diversity calculation, statistical testing, and differential abundance testing

The diversity metrics were calculated using the QIIME2 core-metrics-phylogenetic function after rarefying the feature table to a subsampling depth of 149,894 reads/sample. All 42 samples were retained after rarefaction. The rarefied feature table was used to compute Shannon’s diversity metrics⁵⁸, observed features, Faith’s phylogenetic diversity index⁵⁹, Pielou’s evenness index⁶⁰, and UniFrac distance⁶¹. To statistically compare each alpha diversity metrics between groups, Wilcoxon rank-sum tests⁶² were applied to compare group median values. To visualize the dissimilarities in microbial communities across groups, the weighted and unweighted UniFrac distance metrics was used to generate PCoA coordinates after Cailliez transformation⁶³ to correct for negative eigenvalues. Permutational multivariate analysis of variance (PERMANOVA)⁶⁴ and permutational multivariate analysis of group dispersion homogeneity (PERMDISP)⁶⁵ were applied to compare the centroid location and within-group dispersion level, respectively, in UniFrac distance metrics across groups. Analysis of compositions of Microbiomes with Bias Correction (ANCOM-BC)⁶⁶ was also used to find differentially abundant genera between conditions on each day. The unrarefied features table was used as input to ANCOM-BC as per the standard recommendations. All code used to do the metagenomics analysis is fully available⁵⁷.

Shotgun metagenomic sequencing – data preprocessing, assembly, binning, and classification

To reach bacterial identification at the species level when applicable, we conducted shotgun metagenomic sequencing. The quality check, trimming and human contamination removal were performed using the MetaWRAP⁶⁷ read_qc module. All samples were pooled and co-assembled using metaSPAdes⁶⁸. The contigs were then binned using the metaWRAP binning module, based on a combination of CONCOCT⁶⁹, maxbin2⁷⁰, and MetaBAT2⁷¹. The draft bins were then consolidated using the metaWRAP bin_refinement module that merges bin sets generated from the three binning algorithms into one set of de-replicated and improved bins. A total of four high quality metagenome-assembled genomes (MAGs) were acquired, each with a completeness > 98% and a contamination level < 2.2%. The MAGs were then taxonomically classified with GTDB-Tk v2⁷² (version 2.1.1) using the “–full_tree” mode. GTDB-Tk provides accurate taxonomic assignment of MAGs based on the concordance of placement of MAGs in the GTDB reference phylogenetic tree and the average nucleotide identity to representative reference genomes. The abundance of MAGs in samples were quantified using the metaWRAP quant_bins module. Particularly, within the module, salmon⁷³ was used to estimate the coverage of each contig in every sample; the coverage value of contigs within each bin was then length-weighted averaged to estimate the mean abundance of each bin in each sample. All code used to do the metagenomics analysis is fully available⁵⁷.

Leishmania in vitro inhibition assays

Leishmania major were “seeded” into 24-well plates in Schneider’s insect medium supplemented with 10% heat-inactivated fetal bovine serum, 100 U/mL penicillin, and 100 mg/mL streptomycin (all Thermo Fischer Scientific) at the density of 1 to 2 x 10⁵ parasites/mL. Increasing concentrations of TC1 supernatant or harmane were added to the wells; in parallel, the respective amount of M9 supernatant or DMSO vehicle were added to control wells. In some instances, TC1 and M9 supernatant fractions were also added at a final concentration of 50%. The parasite numbers were counted every day up to seven days after plating.

Sand fly fitness evaluation

Bacteria-fed or control sand flies, not fed on blood, artificially fed on non-infected blood, or artificially fed on blood containing L. major promastigotes (5 x 10⁶/ml) were placed into cardboard pints (100 flies per pint) and kept on a 30% sucrose diet for 11–18 days. Sand fly mortality was recorded daily, and the dead sand flies were removed also daily. Additionally, for the infected sand flies only, at day 12 post-infection, a second bloodmeal was given (as above described) and the mortality was also recorded daily only considering the sorted blood fed sand flies in the following 6 days.

Transmission of L. major via sand fly bites

Only blood fed females were used in transmission experiments. At day 11 post feeding, 20 sand flies were applied to each mouse ear, using vials with a meshed surface held in place by custom-made clamps, and allowed to feed. The number of blood-fed flies was determined by observing them under a stereomicroscope.

Post-transmission follow-up and experimental endpoints

Animals were monitored on a weekly basis to follow the development of lesions caused by L. major infection; images of individual ears were captured weekly using a smartphone. Animals were euthanized 4 weeks post infection for parasite burden determination.

Parasite burden determination

Euthanasia was performed by cervical dislocation under isoflurane anesthesia in all cases (Piramal Healthcare). Ears were collected, immersed in 70% ethanol and then in PBS. Ear cell suspensions were obtained as previously described in ref. ⁷⁴. Ear parasite burdens were assessed by the limit dilution method. Briefly, ear cell suspensions were serially diluted in 96-well plates containing 200 µL Schneider’s insect medium supplemented with 20% heat-inactivated fetal bovine serum, 100 U/mL penicillin, and 100 mg/mL streptomycin (all Thermo Fischer Scientific), and incubated for 14 days at 26 °C. The total number of parasites per ear was calculated as previously⁵⁰.

Modelling

To explore the potential impacts of the bacterial infection on Leishmania spp. parasite transmission we used analytical results from a mathematical model that we developed to study the dynamics of leishmaniasis transmission by vectors in a community of two vertebrate hosts⁷⁵, where one is a reservoir host while the other is an incidental host (e.g. a host that only gets infected but cannot infect sand fly vectors). In this model the basic reproduction number (R₀, the average number of new infections in a susceptible population following the introduction of an infected host⁷⁶), is defined by the following equation:

where V is the population size of vectors, B the population size of reservoirs, γ the recovery rate of reservoirs, μ the mortality rate of vectors, and β the transmission rate, which is an abstraction of the processes that following the contact between vectors and reservoirs can lead to infections in an uninfected host.

To assess the impact of the bacteria in transmission first we assume that \(C=\frac{{BV}}{\gamma }\) is a constant that doesn’t change when the bacteria is introduced in an ecosystem with Leishmania spp. transmission. Under this assumption Eq. (1) becomes:

Here it is important to highlight that \({R}_{0}\propto \frac{{\beta }^{2}}{\mu }\) when more than 2 vertebrate host species are present⁷⁷. Therefore, the use of Eq. (2) is better suitable when we do not know exactly the vertebrate host species diversity in a transmission context (which is often the case thinking on leishmaniasis). Furthermore, with Eq. (2) we can then explore how R₀ changes in response to changes in μ and β. For μ this can be done by assuming that in a natural transmission system, μ is given by the survival of sand flies in our control group. We can then estimate the change in R₀ assuming that changes in sand fly mortality, μ^#, will be the same in the field as the ones observed in the laboratory when sand flies are colonized with D. tsuruhatensis. From Eq. (2) we can expect that if μ increases when sand flies are colonized with the bacteria (particularly Leishmania-infected sand flies), R₀ should decrease. For β, we can assume that its magnitude will change following proportional changes in the dissemination of Leishmania spp. infections. After introducing D. tsuruhatensis, we can then assume that \({\beta }^{\#}={\beta }^{2}\frac{{SF}+}{{SF}-}\), where SF+ is the proportion of disseminated Leishmania spp. infections in vectors when bacteria are present, while SF- is the proportion of the disseminated infections in the absence of the bacteria. Thus, the basic reproduction number after the introduction of the bacteria, \({R}_{0}^{\#}\), becomes:

Using Eq. (3), we explored changes in R₀ using baseline estimates from the field, including a study where R₀ was based on a cross-sectional study of Leishmania spp. exposure in dogs from Panama¹⁵, a multi-species leishmaniasis outbreak from Venezuela^17,78, and the analysis of a time series of human leishmaniasis cases from Costa Rica¹⁸. In the supplementary materials we include the code used to illustrate how changes in β and μ can lead to changes in R₀ by plotting changes over a surface that represents the relationship of R₀ to β and μ. Additionally, the parameter estimates used in our modelling approach are also listed in our supplementary materials, in Table S2.

Data representation and statistics

Results obtained in at least 2 independent experiments are shown per individual sand fly/mouse, with a representation of the group mean/median value ± 95% confidence interval (CI), unless otherwise stated. Statistical analysis was performed using GraphPad Prism software v6.01. All datasets were first subjected to normality tests (Shapiro-Wilk or Kolmogorov-Smirnov). The Unpaired t-test (parametric; always two-tailed) was used to access statistical differences when all groups in a dataset showed normal distributions. The Mann-Whitney test (nonparametric; always two-tailed) or Kruskal-Wallis test, the latter with post-hoc analysis (Dunn’s test), were used to access statistical differences when at least one group in a dataset did not show a normal distribution. A p-value ≤ 0.05 was considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.