Ethics statement

The sand flies collection was performed following the guidelines of the Colombian decree N° 1376. No specific permits were required for this study. This study was conducted under the Ministry of Environment and Sustainable Development (MADS) permission contract No. 121 of 2016, OTROSí No-25, for access to genetic resources and its derivative products. Also, the insects were collected under resolution No. 0207 of 090320 of the Ministry of Environment and Sustainable Development (MADS) and Universidad Nacional de Colombia. Sand flies were collected on private property and permission was received from landowners before sampling.

Origin, survey, and identification of sand flies and mosquitoes

Each locality was selected based on the incidence of leishmaniasis, malaria, and several arboviruses in Colombia during the last four years and historical records of adult phlebotomines and mosquitoes (Culex and Anopheles) [63–66]. Sand flies and mosquitoes were collected in peri-urban locations and natural reserves of the Andean (Antioquia, Caldas, and Cundinamarca departments) and Caribbean regions (Sucre and Cordoba departments) during an entomological survey performed between the high rainfall periods from September to November 2018 and the months from March to May 2019, with a collection frequency of three days per study area (Fig 1) using CDC light traps, located indoors and near houses, overnight, between 17:00 and 06:00 h. Shannon traps were also used for collection near houses. Additionally, a diurnal collection using a mouth aspirator was done in the surroundings of nocturnal trapping sites [67].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Geographic locations of sampling sites and proportion by department of insect vectors (Phlebotomine sand flies and mosquitoes) infected with Wolbachia, Cardinium and Microsporidia in Colombia.

Link to the base layer of the map used: https://www.colombiaenmapas.gov.co/?e=-74.7117818345732,0.23307652371168444,-62.73668417832639,8.735414279337139,4686&b=dark-gray&u=0&t=2901&servicio=610.

https://doi.org/10.1371/journal.pntd.0009942.g001

Collected specimens were kept dry in 1.5 mL vials and transported to the laboratory with dry ice. Once at the laboratory, they were kept at −20°C. The head and last three abdominal segments of sand flies were removed from the specimens to perform the recognition of sex and taxonomic identification following the classification system of Galati, 2018 [68]. The thorax and remaining abdominal segments were stored at −20°C until DNA total extraction for diagnostics of bacterial endosymbionts and DNA barcode methodology [20]. For Culex and Anopheles the morphological keys of Lane 1953 and Forattini 2002 [69, 70], and González & Carrejo 2007 [71] were used, respectively, to perform the recognition of sex and taxonomic identification. Two legs were collected for DNA barcode methodology and the rest of the body for diagnostics of bacterial endosymbionts [20]. All mosquitoes and phlebotomines were processed using a chill table at −5°C during the taxonomical identification process.

DNA extraction from sand flies and mosquitoes for detection of endosymbionts

DNA extraction of all samples of sand flies and controls (Mosquitoes of Aedes) was performed according to the high salt concentration protocol [72]. The elution of total DNA was made in 50 μL of Buffer Tris-HCL 10 mM-EDTA 1 mM pH = 8. For Anopheles and Culex individuals, the body was processed with the ZR Tissues & Insect DNA mini-Prep (Zymo Research) extraction kit. The DNA concentration for the species of Anopheles and Culex varied approximately 45–280 ng/μl and for Lutzomyia species between 1 and 20 ng/μl on average measured using NanoDrop 2000 Spectrophotometer (Thermo Scientific NanoDrop, Wilmington, USA).

Detection of bacterial endosymbionts of sand flies and mosquitoes by conventional PCR

Wolbachia detection in the collected insects was performed by PCR using wsp81F and wsp691R primers to amplify a partial fragment (650 bp) of the gene coding for the main surface protein of endosymbiotic Wolbachia (wsp) (Table 1) [73]. As a PCR-positive control, in all trials DNA from Aedes (Stegomyia) aegypti larvae from Medellín (Antioquia, Colombia) infected with a reference strain of Wolbachia (Supergroup A, strain wMel) were included, and ultrapure water was used as a PCR-negative control. Microsporidia detection was done using ss18f and ss1492r primers of the small subunit rDNA (ssrDNA- fragment 1300 bp) (Table 1) [37, 74]. A plasmid with Microsporidia MB (from Anopheles arabiensis) cloned fragment was routinely used as a positive control and ultrapure water as a PCR-negative control.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Genes and primers used in polymerase chain reaction (PCR) assays to detect reproductive parasites in insect vectors (Phlebotominae Sand flies, Culex, Anopheles), DNA barcoding for the identification of species, screen of Leishmania infection and microbiota by Illumina MySeq from samples of several locations in Colombia.

https://doi.org/10.1371/journal.pntd.0009942.t001

To detect Cardinium, we used the primers CLO-F1 and CLO-R1, targeting a ~450 bp fragment of the 16S rRNA gene (Table 1) [24]. As a positive control, Cardinium DNA from a biting midge (Culicoides) was included and ultrapure water as a negative control. Also, the presence of Spiroplasma was investigated using SpixoF and Spixo R primers that amplify a ~800 bp fragment of the 16S rRNA gene (Table 1) [37]. As positive control, DNA of S. poulsoni (SPHY1) was included, and ultrapure water was used as a negative control. The reaction mix used and conditions of amplification for PCR for endosymbionts were done according to the conditions described previously [28, 37, 73, 74]. The amplified fragments were run on an agarose gel at 1% in 1× TBE buffer (40-mM Tris-Borate, 1 mM EDTA, pH 8.0) to verify the integrity of the fragments.

PCR products of each endosymbiont were ligated into a pJET1.2/blunt cloning vector following the manufacturer’s instructions (Thermo Fisher Scientific, MA, USA) [75]. Plasmidial DNA from a positive colony was extracted using GenJet Plasmid Miniprep Kit (Thermo Fisher Scientific, MA, USA), and amplified using pJET1.2 forward and reverse primers. Clones with the partial products of endosymbionts were verified by sequencing in both directions, using the service of Macrogen Company in Korea. For each assay, a negative control (no DNA), as well as a positive control (control PCR product by the cloning kit), was included. The taxonomic list of insect species, distribution, the frequency or percentage infection of endosymbionts by locality and ecoregion, as well as the sex ratio for each endosymbiont, was consolidated systematically.

DNA barcoding identification of Phlebotominae sand flies and mosquitoes species

DNA samples of sand flies and mosquitoes positive for any endosymbionts and with taxonomic importance were used for PCR, corresponding to a barcode region of the COI (700 bp fragment) using LCO1490- HCO2198 primers (Table 1); the reaction mix, and conditions of amplifications were used according to the previously described [76, 77]. To amplify the mitochondrial 5′ COI gene region, PCR was performed using the following thermal profile: an initial denaturation of 5 min (94°C); followed by 35 cycles at 94°C for 1 min (denaturation), 45°C for 1 min (annealing), and 72°C for 1 min (extension); and a final extension at 72°C for 10 min. Each PCR cocktail had a final reaction volume of 50 μl and contained the following: 0.3 μL of Taq DNA polymerase (5 U/μL), 5 μL of 10× PCR buffer (NH4SO4), 5 μL of 2.5 mM MgCl2, 5 μL of 2 mmol/L dNTP, 2 μl of 10 μmol/L of each oligonucleotide (forward and reverse), 4 μL of the DNA template, and ddH2O to bring the volume to 50 μl [78]. The amplified fragments were observed on an agarose gel at 1% to verify the integrity of the fragments. Subsequently, double-stranded DNA was sequenced in both directions using a capillary electrophoresis system, using the service of Macrogen Company in Korea.

PCR amplification of the HSP-70 N Leishmania gene in female groups

PCR was done to screen Leishmania infection in Lutzomyia females. HSP70-F25 and HSP70-R617 primers were used to amplify a 593 bp partial segment of the HSP-70N gene (coding for Heat shock protein 70) (Table 1). PCR was done following the described conditions by Fraga et al., 2010 [79]. As a positive control, Leishmania panamensis (reference strain UA140) DNA and Leishmania braziliensis (reference strain UA 2903) DNA were used.

Molecular identification and phylogenetic analysis of endosymbionts based on the analysis of sequences

In this study we propose a genomic–phylogenetic species concept (GPSC) for the taxonomy of the S-symbionts [80]. The advantages of the GPSC are: i) The GPSC is a methodological species concept that relies heavily on macromolecular sequences (e.g. protein, RNA and DNA) that are not subject to biases of interpretation; ii) The organism’s ecology is encompassed by the GPSC in that adaptive radiations will bereflected in the alteration of gene and protein sequences in response to the organism’s niche; iii) The biogeography of an organism is considered by the GPSC in that genetic drift will result in a change in the gene and protein sequences of core genes; iv) The evolutionary history of the organism isconsidered through the phylogeny of gene and protein sequences; v) Sequence information is readily portable and globally accesible; vi) Taxa can be identified rapidly once the set ofrelevant genes for a have been identified.

Sequences of wsp gene and 16S rDNA obtained from Wolbachia or other endosymbionts, were directly compared with those in GenBank by BLASTn with megablast algorithm (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The sequences were edited using Bioedit software v.7.2.5 [81]. Alignments of sequences of endosymbionts obtained and reported in GenBank were performed using a Clustal W incorporated in MEGA X software [82]. Verification of recombination events and the presence of chimeras was performed using RDP4 (Recombination Detection Program version 4) software [83]. Patterns of genetic divergence as nucleotide composition, the number of haplotypes (H) and variable sites were evaluated using Bioedit v.7.2.5(82) and DNAsp 5.0 [84]. K2P genetic distances of Kimura-2-Parameters (Intra and interspecific) also were evaluated with the MEGA X software [82]. The K2P model will be calculate the number of differences accumulated between two sequences since their last common ancestor. The K2P model is probably the most widely used of all models of nucleotide substitution for estimating genetic differences (called genetic distances) and phylogenetic relationships. The web program FINGERPRINT (http://evol.mcmaster.ca/fingerprint/) was also used to depict the nucleotide heterogeneity and nucleotide diversity [85].

Phylogenetic inference analysis for Wolbachia, Cardinium, and Microsporidia was performed using the Bayesian inference method with MrBayes 3.2 software [86] under the substitution model 4by4 (number of generations = 1’000.000), after the consensus trees were obtained, they were visualized and printed using FigTree V.1.1.4. We included a list of sequences previously registered for Wolbachia [20], Cardinium [22], Microsporidia [87] from arthropods for analysis. PCR products, and smaller edited sequences (< 550-600pb) were not considered in the phylogenetic analysis of Wolbachia. Alignments of sequences of wsp genes from sand flies and mosquitoes were performed as suggested Zhou et al., 1998 and Braig et al., 1998 [73, 88]. However, sequences of wsp gen less than 500 bp were also verified in BLASTn only to confirm the presence of Wolbachia and estimate infection rates. The phylogenetic relationship of Microsporidia included two fungi (Basidiobolus ranarum and Conidobolus coronatus) as outgroups. In the analysis of Wolbachia was used Wolbachia of Bemicia tabaci as outgroup. For phylogenetic inference of Cardinium was used as an outgroup Candidatus Amoebophilus asiaticus. The neighbor-joining (NJ) tree was built in MEGA X for analysis of sequences COI obtained to determine the limitation taxonomical for the identification of sand flies and mosquitoes.

Interactions between the microbiota-endosymbionts: 16S rRNA gene survey, and statistics

To evaluate the dynamics of the microbiota against the presence or absence of different endosymbiotic bacteria, DNA of sand flies, and mosquitoes positive for one or more endosymbionts were considered (Table 2). Total DNA was used as a template to generate amplicons of the hypervariable region V4 of the 16S rRNA gene according to the protocol described at EMP 16S Illumina Amplicon Protocol (https://www.earthmicrobiome.org/protocols-and-standards/16s/) [89] using primer 515F and primer 806 R with PrimeSTAR HS DNA Polymerase (Takara, Japan) cocktail mix using the manufacturer’s instructions [90]. The first PCR round consisted of an initial step of 5 min at 95°C, followed by 20 cycles of denaturation at 98°C for 10 s, annealing at 50°C for 15s, and extension at 72°C for 45 s, followed by an elongation step of 72°C for four minutes and a holding stage at 4°C until further processing. From these reactions, 1 μl was transferred to fresh PCR cocktail mixes containing each corresponding overlapping primer with two distinct indices and Illumina adapters for each sample [91], run for 10 cycles using the same temperature and times of the first round. The presence and expected size of PCR products were assessed by agarose gel electrophoresis. Controls negative and positive were used.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Samples of phlebotomine sand flies and mosquitoes that were positive for endosymbionts (Wolbachia*; Cardinium^, Microsporidia⁺) by PCR previously and selected for analysis of the total microbiota by Illumina MiSeq sequencing.

https://doi.org/10.1371/journal.pntd.0009942.t002

Amplified products were purified, normalized, and pooled using the SequalPrep Normalization Plate (Thermofisher Scientific, USA) and subjected to 250-bp paired-end Illumina MiSeq sequencing, using MiSeq Reagent Kit, (300-cycles). On negative and positive controls of sequencing, in this case they were not necessary since there were no negative control amplification signals and the sequencing kit produced sequences of high- quality and expected quantities, as it can be confirmed inspecting the raw data quality values and reads per sample. Sequencing was commissioned to Alianza Sanitizar SA. From demultiplexed 16S amplicon raw pair-end sequence datasets from each sample, we used the DADA2 software package (https://github.com/benjjneb/dada2) following a sequential pipeline for filtering, denoising, chimeras, and merging [92], resulting in assembled datasets for each amplicon with trimmed reads based on quality, cut out of the primer sequences, deleting potentially chimeric sequences, to detect the counts of each unique Amplicon Sequence Variant (ASV) across all samples and to classify them using RDP Naive Bayesian Classifier [93] having as taxonomic reference Silva database release 132 (https://www.arb-silva.de/documentation/release-132/).

In all cases, the coverage was above >0.9 with 10.000 reads, therefore Bray-Curtis dissimilarity matrix of non-rarified relative percentages was used. For Bray-Curtis dissimilarity between all pairs of samples at ASV level was calculated using phyloseq (default parameters, https://joey711.github.io/phyloseq/distance.html). Identical nodes on displayed UPGMA dendrogram clustering were obtained with data rarified to 19884 reads on each dataset. The phyloseq software package (https://joey711.github.io/phyloseq/) [94] and Microbiome Analyst (https://www.microbiomeanalyst.ca) [95], were used for estimating and obtain plot on alpha diversity (within-sample) metrics (Observed, Chao 1, Shannon, Simpson) and beta diversity between all communities (Hierarchical clustering analysis, Principal Coordinates Analysis of weighted Unifrac distances-PCoA, Heatmap). Non-metric Multidimensional Scaling (NMDS) was also obtained with Microbiome Analyst using Bray-Curtis index [95]. Differences in weighted and unweighted Unifrac distances between the groups of sand flies and mosquitoes were analyzed with analysis of similarity (Kruskal-Wallis test). A P value ≤0.05 was considered statistically significant. The core microbiota (Sample prevalence 20%; Relative abundance 0.01%) and linear discriminant effect size (LEfSe) analysis were implemented using Microbiome Analyst [95], to identify that were significantly and differentially abundant taxa between hosts. A threshold alpha value of 0.05 for the LSmeans & Tukey’s HSD post-hoc test was implemented. The random forest algorithm was also used to find significant correlations between bacterial taxa using the top 25 most abundant genera of the microbiota.

Diversity of sand flies and mosquitoes

Five hundred seventy-four individuals (292 and 282 from the Caribbean the Andean region, respectively) were collected and identified by classical taxonomy. 430 specimens from 12 genera of phlebotomines, 109 individuals from seven species of the genus Anopheles and 35 individuals corresponding to Culex nigripalpus (Table 3). Species with abundances greater than 50 specimens correspond to Pi. evansi, Lu. longipalpis and An. nuneztovari (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. List of species of phlebotomine sand flies, Culex nigripalpus and Anopheles tested for Wolbachia, Cardinium, Microsporidia and Spiroplasma in two ecoregions from Colombia.

https://doi.org/10.1371/journal.pntd.0009942.t003

Frequency and molecular detection of endosymbionts

Conventional PCR used in this study could detect Wolbachia, Cardinium, and Microsporidia in total DNA extracted from sand flies and mosquitoes using reference primers (S1 Fig). Only Wolbachia wsp primers showed some PCR products of unexpected size (S1C Fig. Line 11). Spiroplasma was not detected in the DNA samples using reference conditions. Correct cloning of PCR products from various endosymbionts from natural populations of sand flies and mosquitoes from Colombia was successfully performed and preserved (S1H Fig). In the Caribbean region, Wolbachia was detected in 9.83% of the sand flies, and 1.83% of the mosquitoes were positive (Table 3).

In contrast, lower Wolbachia infection values were found in sand flies (2.83%) and higher in mosquitoes (20%) in the Andean region (Table 3). The presence of Microsporidia in the Andean region (mosquitoes 2.8%, sand flies 0.8%) and Caribbean ecoregion (sand flies 1,1%) was low (Table 3. Cardinium was estimated with lower infection values in the Andean region (sand flies 0.4%) but with a significant presence in sand flies (8.74%) from the Caribbean ecoregion (Table 3).

The results by the department show the highest presence of Wolbachia in Sucre (11%) compared with Antioquia (1%) and the Córdoba (5%) (Fig 1). For Cardinium, similar infection rates (5%) for the Córdoba and lower levels in Sucre (2%) and Antioquia (1%) (Fig 1). Natural populations of sand flies and mosquitoes from Antioquia and Caldas showed lower levels of Microsporidia infection (1%) compared to Sucre (4%) (Fig 1).

Specific analyses show that Wolbachia was found in seven species of sand flies and two species of mosquitoes (Table 3). A high percentage of infection was observed in Ev. dubitans (100%), Pi. evansi (11.3%) and Br. hamata (16.6%). Importantly, a single diagnosed Mi. micropyga individual was infected with Wolbachia. The mosquitoes, Cx. nigripalpus (20%) and An. trianulatus (16.6%) also presented high levels of infection (Table 3). Species such as Pi. pia (2.1%), Pi. evansi (2.5%) and Cu. nigripalpus (2.8%) were the only ones that showed positive results for Microsporidia. The species Mi. cayennensis (population from the department of the Córdoba) presented a high percentage of Cardinium infection (45.83%) followed by Tr. triramula (50%) and Ps. shannoni (25%) (Table 2). Double infection with Wolbachia and Cardinium is reported in one Pi. evansi specimen from the Sucre department (S2 Table and Fig 1C and 1F). For the first time, Wolbachia, Cardinium, and Microsporidia endosymbionts are registered in different species of sand flies and mosquitoes (Table 3).

We observed a higher frequency of Wolbachia presence in female sand flies and mosquitoes (9.27%) (Table 4). Contrasting Cardinium, was found only in phlebotomines and this has a more equitable sex ratio of infection (Males = 3.43%; Females = 2.41%) (Table 4). Microsporidia was detected only in females (1.03%) (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The frequency of endosymbionts infection by sex in sand fly and mosquitoes from Colombia.

https://doi.org/10.1371/journal.pntd.0009942.t004

Molecular identification and phylogenetic relationships of endosymbionts

Fifteen Wolbachia wsp gene sequences (eight from phlebotomines sand flies; seven from Culex nigripalpus) presented high percentages of similarity (98% - 99%) compared with the GenBank sequences. The global nucleotide alignment (580 bp) of the wsp gene included an additional 63 sequences from 59 arthropod-derived Wolbachia strains located in super groups A and B, as seen in the phylogenetic relationships of Bayesian inference (Fig 2). Clade posterior probability ranged from 0.8 to 1, indicating a robust analysis of Wolbachia phylogeny (Fig 2). Host wsp sequences were determined for Cx. nigripalpus, Br. hamata, and Lu. gomezi from Antioquia (Andean Eco-region), located in supergroup A, while the wsp sequences of phlebotomines (Mi. micropyga and Ev. dubitans) from the Department of Cordoba (Eco-region Caribe) present a close relationship with strains of the "Leva" group located in supergroup B (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phylogenetic relationships of Wolbachia strains using Bayesian analysis and inferred with wsp sequences.

The Wolbachia strains are identified by host species from which they were isolated (sand fllies symbol of green color; Culex symbol of orange color). Numbers in nodes represent Bayesian posterior probabilities. Reconstruction performed with MrBayes (version 3.0). The wMel positive control is indicated in red color. As out-group Wolbachia of Bemicia tabaci.

https://doi.org/10.1371/journal.pntd.0009942.g002

Seven wsp sequences obtained from Cx. nigripalpus, with all similar nucleotide sequence suggesting a new strain named wCnig, forming a putative new group Cnig (Fig 2). Cnig’s specific alignment (514 bp) suggests the presence of three haplotypes, 10 variant sites (S2 Fig), eight parsimoniously informative sites, and two singletons. Heterogeneity is distinguished at the beginning and end of the alignment (S2 Fig), indicating that K2P genetic distances ranged from 0.001 to 0.002 and the identity percentages between 98.33% and 100%. It was also determined that Wolbachia wsp gene sequences were obtained from Lu. gomezi corresponds to a variant or haplotype of the wOri strain, while Br. hamata corresponds to a putative new strain named wBham, which is also related to the wOri and wCauB strains of Tosogodes and Ephetesia, respectively (Fig 2). To confirm this result, 33 variable sites and three parsimoniously informative sites with a high number of singletons were determined in the specific alignment of this sequence grouping. The K2P genetic distances of wLgom with wOri and wCauB were 0.02, with wCnig 0.12 and 0.007 with wBham. The percentage of similarity of wLgom with wOri was 98.57%. In contrast, wBham presented distances with values of 0.17 with wCnig and 0.07 with wOri and wCauB. The percentage of similarity of wBham with wOri was 88.84%, as with wLgom, indicating a clear difference between these two strains.

Wsp sequences of Mi. micropyga and Ev. dubitans were consistently located in the Leva group, as mentioned above, indicating that nucleotide alignment revealed the existence of 29 variant sites, eight parsimoniously informative sites, and heterogeneity was appreciated throughout the alignment (S2 Fig). The K2P genetic distance values ranged from 0.001 to 0.003, indicating that the Ev. dubitans are similar to the wLeva strain (identity percentages 99.28% −98.8%), while the Mi. micropyga sequences are similar to the wLcy strain, where it presented identity percentages of 99.52%. The percentages of identity between sequences of Mi. micropyga and Ev. dubitans were 96.63% −98%.

Regarding Microsporidia, good quality 16S SSU rDNA sequences were obtained from PCR products detected in Pi. pia and Cx. nigripalpus with high percentages of similarity (99%) associated with Microsporidia sequences deposited in GenBank. The 16S SSU rDNA sequences from our study and 82 Microsporidia representing the five established clades derived from arthropods with different environmental origins were successfully aligned to achieve a data set of 1097 bp.

The phylogenetic relationships determined by Bayesian inference (Fig 3) suggest that Microsporidia of Pi. pia and Cx. nigripalpus was located in clade IV corresponding to the Terresporidia class (Terrestrial Origin) and closely related to Microsporidian MB from An. arabiensis (ID 2270798) with supports of the posterior probability of 0.86 (Fig 3). Other nearby Microsporidia correspond to Vitaforma, Endoreticulatus, and Cystoporogenes (Fig 3). The analysis of genetic variability that includes sequences from this grouping (Clade) indicated the existence of 296 variant sites, with heterogeneity distributed throughout the alignment (S2 Fig). Contrasting, the 16S SSU rDNA sequences of Microsporidia MB, Pi. pia and Cx. nigripalpus, they revealed only 62 variants and 16 parsimoniously informative sites, showing clear differences. The K2P distances show values of 0.002 and 99.46% identity percentages between Microsporidia of Pi. Pia and Cx. nigripalpus indicating that they corresponded to the same species, but different from Microsporidia MB because the K2P genetic distances were 0.056 with identity percentages between 94.40% and 97.20%. Concerning the other members of nearby Microsporidia, the minimum values of K2P distances were 0.09 with Vitaforma corneae and maximums of 0.167 with Endiriculatus bombicis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Bayesian inference and phylogenetic relationships of Microsporidian strains using sequences of the small subunit ribosomal DNA.

The Microsporidian strains are identified by host species from which they were isolated (sand fly symbol of green color; Culex nigripalpus of orange color). Numbers in nodes represent Bayesian posterior probabilities. Reconstruction performed with MrBayes (version 3.0). Clades I–V are indicated. Two fungi were used as outgroups in the analysis (Conidiobolus coronatos and Basidiobolus ranarun). Taxa without GenBank accession numbers are original for this dataset.

https://doi.org/10.1371/journal.pntd.0009942.g003

The initial analysis of partial 16S rDNA sequences of Cardinium found in phlebotomines revealed high and variable percentages of similarity (97.61% −98.49%) with the arthropod host sequences reported in Genbank. In total 14 only phlebotomine sequences were obtained and compared with 41 available Cardinium sequences and located in groups A, B, C, and E (Fig 4), the result of a nucleotide alignment of 467 bp, of which 368 sites were variables. Specifically, the sequences are categorized as the Cardinium of Tr. triramula and Ps. shannoni, they were located in group C with a high posterior probability of clade (0.93), indicating a close relationship with the Cardinium of Culicoides (Fig 4). In this group C data set, 13 variant sites (S2 Fig) and four parsimoniously informative sites were estimated, with K2P genetic distances ranging between 0.001 and 0.002, and identity percentages between 97.67% and 98.19%, suggesting which corresponded to the same Cardinium species, but in different hosts. Heterogeneity occurred mainly at the beginning of the alignment (S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Phylogenetic relationships of Cardinium strains using Bayesian analysis and based on 16S rDNA sequences.

The Cardinium strains are identified by host species from which they were isolated (sand flies symbol of green color). Numbers in nodes represent Bayesian posterior probabilities. The reconstruction was performed using MrBayes (version 3.0). As an outgroup was used Candidatus Amoebophilus asiaticus. Taxa without GenBank accession numbers are original for this data set.

https://doi.org/10.1371/journal.pntd.0009942.g004

Two subgroups of Cardinium sequences, the first only associated with Mi. cayennensis (Subgroup 1, the posterior probability of clade = 1) and the second drift of Mi. cayennensis and Pi. evansi (Subgroup 2, the posterior probability of clade = 1) and one sequence near Ps. shannoni were located phylogenetically in group A (Fig 4). Further analysis suggests the presence of 17 variant sites in subgroup 1 and a single parsimoniously informative site (S2 Fig), while K2P distances were 0.001 between all sequences in this subgroup confirming that they corresponded to a single new strain called Cardinium of Mi. cayennensis sub-1. In subgroup 2, only three variable sites were determined, and the genetic distances ranged from 0.003 to 0.009, with the sequence of Mi. cayennensis MtCor125 is the most divergent. These results confirmed the presence of another genetic variant called the Cardinium of Mi. cayennensis sub-2. Our results contribute to the significant presence of Cardinium diversity in phlebotomine sand flies.

DNA barcoding for the identification of Sand flies and mosquitoes’ species

A total of 38 sequences barcode (COI gene) of phlebotomines and five sequences of mosquitoes positive in the diagnosis of endosymbionts and other species with difficulties in taxonomic determination by morphology were correctly analyzed and allowed species differentiation. Other sequences available at Genbank for phlebotomines from Colombia were included in the Neighbor-Joining comparative analysis (Fig 5). The analysis showed high bootstrap values (100%) for taxonomic groupings of sand flies species and Cx. nigripalpus (Fig 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Neighbor-joining analysis of mitochondrial COI sequences among sand flies species (symbol of green color) and mosquitoes of genus Culex (symbol of orange color) from Colombia, based on the model of Kimura’s two-parameter model (K2P).

Bootstrap support was estimated with 1000 replicas, only bootstrap values >50% being shown above the branch. As an outgroup was used Tipula maxima.

https://doi.org/10.1371/journal.pntd.0009942.g005

Interactions between the microbiota-endosymbionts in sand flies and mosquitoes

A total of 11 samples of the DNA of phlebotomines (n = 8) and Cx. nigripalpus (n = 3) that were positive for Wolbachia, Microsporidia, and Cardinium, were selected to evaluate the microbiota behavior by amplifying the V4 region of the 16S rRNA (Table 4).

The summary of results obtained (total reads, the number of ASVs, top five Phyla-Family-Genus) from 16S rRNA gene amplicon sequencing of 11 samples of insects, untreated and treated based on quality and taxonomy classification, is shown in the S1 Table and S3 Fig.

At the phylum level, the microbiota is dominated by five bacterial groups: Proteobacteria (67.6%), Firmicutes (17.9%), Actinobacteria (7.4%), Bacteroidetes (1.7%) and Cyanobacteria (1.5%), also with a 3,25% fungal phyla Microsporidia, making the 99.4% of the entire population (S4 Fig).

The total microbiota profiles at the level of genus show high percentages of infection by Wolbachia in Lu. gomezi, Ev. dubitans, Mi. micropyga, Br. hamata, and Cx. nigripalpus, with relative abundances of ASVs of 30%-83% (Fig 6). A high significance in public health represents the first record of the natural infection of Rickettsia ASVs in sand flies from Colombia, mainly in populations of Pi. pia (Department of Caldas) with a relative abundance of 58.8% (Fig 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Microbiota composition and diversity of wild specimens of sand flies and Culex nigripalpus found with endosymbiont in several regions or departments from Colombia.

A) Relative abundance of ASVs that were called to the taxonomic rank of genus. The ASVs with abundances above 1% of the rarefied count matrix were selected. See Table 4 for the nomenclature of samples in detail. NA: ASVs without taxonomic assignment.

https://doi.org/10.1371/journal.pntd.0009942.g006

In summary, there are clear differences in the composition and diversity of microbiota at the intra-specific (Figs 6 and 7) and interspecific (Fig 8) levels in sand flies and Cx. nigripalpus, which may depend in the first instance on the presence and load of natural infection of endosymbionts (Fig 6). For example, Pi. evansi, revealed two samples positive with Wolbachia and one with Microsporidia by PCR, but have a low relative abundance of these ASVs (0.04%-0,2%) in their microbiota, with differences in the composition of ASVs such as Acinetobacter, Streptococcus, Staphylococcus, Ralstonia, Pseudomonas and Aeromonas (Fig 6). This pattern is also evident in samples of Cx. nigripalpus, suggesting that ASVs Lactococcus, and Aeromonas are the ones that fluctuate significantly (Fig 6). In contrast, when Wolbachia levels exceed 30% in Ev. dubitans, Mi. micropyga, Br. hamata, and Cx. nigripalpus, and Rickettsia exceed 50% of the composition in Pi. pia, significantly reduces the relative abundance of ASVs listed above (e.g., Ralstonia, Acinetobacter, Staphylococcus, Aeromonas) (Fig 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Alpha diversity (intra-diversity) of bacterial composition using amplicon sequence variants (ASVs) in sand flies and Cx. nigripalpus from Colombia by ecoregion and presence of endosymbiont.

a) Species richness was assessed by Chao1 (estimates diversity from abundance data), b) Observed OTU (sum of unique OTUs in each sample), c) Shannon-Wiener (sum of species proportion logarithms) and d) Simpson indexes. e) Alpha diversity with all indexes but associated with the presence of endosymbionts. Higher index values indicate richer and more diverse bacterial composition.

https://doi.org/10.1371/journal.pntd.0009942.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. β-diversity analysis (between groups) of microbial communities associated with the groups of samples established of sand flies and Culex nigripalpus found with endosymbiont in several ecoregions from Colombia.

a) Hierarchical clustering analysis of ASVs at the genus level. b) Principal Coordinate Analysis (PCoA) with ASVs at the genus level (Weighted Unifrac Distance, PERMANOVA, F = 2.4744; R2 = 0.18363; p-value = 0.007). c) Beta-diversity visualized using Non-metric Multidimensional Scaling (NMDS). The plot was obtained with Bray-Curtis Dissimilarity distances (PERMANOVA, F = 1.1429; R2 = 0.094118; p-value = 0.324; [NMDS] Stress = 0.11348).

https://doi.org/10.1371/journal.pntd.0009942.g008

The alpha diversity analysis is consistent with composition profiles, and we also observed a possible effect due mainly to host and abundance of Wolbachia (Fig 7A–7E) on the diversity of bacterial communities within sand flies and Cx. nigripalpus. The alpha diversity per host shows a value lower of Chao 1 and Observed for sand flies in the Carribbean ecoregion (Fig 7A and 7B). Samples as Pi. pia has these indeces were higher than the rest of phlebotomines or mosquitoes. Cx. nigripalpus, Br. hamata and Pi. evansi have the Shannon and Simpson indexes higher with respect to other sand flies (Fig 7C and 7D).

We found that Wolbachia infection decreased the diversity and number of taxa (Observed = 50–200), the adjusted richness (Chao 1 = 50–200) as well as the uniformity and/or distribution of ASVs in different samples, as seen in the Shannon index (Fig 7E). This impact of Wolbachia is associated with the Simpson index, which shows an increase in the dominance of ASVs, mainly in sand flies (Simpson index = 07–09). In contrast, the sand flies and Cx. nigripalpus that have other endosymbionts (Cardinium/Microsporidia) show a high diversity (Shannon index = 2–4.5; Chao 1 index = 200–450), suggesting the possibility of finding more rare species in their niches and a high richness (Fig 7E).

The hierarchical analysis of clusters (Beta Diversity) derived from the comparison of ASVs at level of genus (Fig 8A), shows three groups (a = Sand flies and Cx. nigripalpus from Andean ecoregion; b = Sand flies from Caribbean region; c = Sand flies and Cx. nigripalpus from both ecoregions), suggesting that the composition of the prokaryotic microbiota is also determined for geographical distribution (ecoregion Andean or Caribbean), but also show in the influence of the host. The Principal Coordinate Analysis (PCoA) is consistent, which showed statistically significant differences (PERMANOVA, F = 2.4744; R2 = 0.18363; p-value = 0.007) between the microbiota of sand flies and mosquitoes depending on its origin and possibly for the abundance or presence of some endosymbionts (Wolbachia, Rickettsia) (Fig 8B). However, the NMDS plot was performed to assess significance among Wolbachia and bacterial community composition (Fig 8C), did not exhibit significant clustering in either between groups of sand flies or mosquitoes (PERMANOVA, F = 1.1429; R2 = 0.094118; p-value = 0.324; [NMDS] Stress = 0.11348). Only some samples, as Cx. nigripalpus and Mi. Micropyga showed differences visually significant on Wolbachia in this analysis. Now, the Kruskal–Wallis test, which showed statistically significant interspecific differences between the sand flies and mosquito total microbiota (***p<0.0005), using weighted and unweighted Unifrac distances.

The heatmap cluster analysis (Fig 9) at the genus level with 10 samples of sand flies and three samples of Cx. nigripalpus includes 54 bacterial genera were overlapped differentially according to all species (host). Between five and ten bacterial genera contributed 90% abundance of bacterial communities per sample of insects. No bacterial genus served as biomarkers were found in unique bacteria of sand flies or mosquitoes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Heatmap based on the microbiota composition at the genus level illustrating common and endemic bacterial community abundances according to insects, the presence of endosymbionts and categorization according to its importance as a vector.

Hierarchical Ward’s linkage clustering based on Pearson’s correlation coefficient of the microbial taxa abundance. Red and blue colors represent positive and negative correlations, respectively. The color scale represents the scaled abundance of each variable, denoted as Z-score, red indicates high abundance, and blue indicating low abundance.

https://doi.org/10.1371/journal.pntd.0009942.g009

The core microbiota of sand flies and Cx. nigripalpus, mainly show some ASVs (Aeromonas, Ralstonia, Corynebacterium, Acinetobacter, Pseudomonas) with relative abundance and prevalence significant (S5 Fig), and other taxa as Lactococcus, Empedobacter that has a consistent behavior with Linear discriminant analysis effect size (LEfSe), that was performed to identify specialized bacterial groups in sand flies and mosquitoes at each ecoregion. At genus level, the LEfSe showed 14 bacterial taxa with significant differences in insects of Andean ecoregion (Fig 10), such as Aeromonas, Lactococcus, Empedobacter, Veillonella, among others. Whereas Herbaspirillum was the ASVs most abundant and have significant differences in insects from Caribbean ecoregion (Fig 10).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Linear discriminant analysis effect size (LEfSe) of gut microbiota in sand flies and Culex nigripalpus de Colombia.

https://doi.org/10.1371/journal.pntd.0009942.g010

The random forest analysis was highly reproducible and shows interactions among some taxa (Wolbachia, Ralstonia, Rickettsia, Acinetobacter, Aeromonas) that act as significant features in the microbiota in sand flies and Culex nigripalpus de Colombia (Fig 11). Accuracy and robustness of the random forest revealed that Wolbachia have positive interactions only with Weisella, Micrococcus and Corynebacterium (Fig 11A). ASVs of Ralstonia and Aeromonas, are also important in the structure and composition of the microbiota, for having a greater number of positive interactions, mainly with Methylobacterium, Sphingomonas, Rahenella for the case of Ralstonia and Brucella, Empedobacter, Gemmella, Dechlorosoma Achromobacter and Dolosigranulim for the case of Aeromonas (Fig 11B and 11C). The endosymbiont Rickettsia presented strong interaction with Streptococcus and Staphylococcus (Fig 11D), while that Acinetobacter presented strong interaction with Halomonas and Comamonas (Fig 11E). ASV of Lactococcus was abundant but show no interaction with other ASV the microbiota (Fig 11F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. The random forest analysis shows positive interactions and significant features among some taxa in the core microbiota in sand flies and Culex nigripalpus de Colombia.

The top 25 of ASVs most abundant are considered. The y axis, from top to bottom, displays the taxa ranked by their importance or association with other ASV and ecoregion.

https://doi.org/10.1371/journal.pntd.0009942.g011

A deep and specific search shows an ASV corresponding to Candidatus cardinium, only in Pi. evansi with low relative abundance (0.2%; Reads = 149) (Table 5) and show a higher pattern of relative abundance of the ASV Microsporidia in Pi. pia (23%; Reads = 136062) (Table 5), Cx. nigripalpus (11%; Reads = 38938) and Pi. evansi (0.2%; Reads = 1017). Also, the Microsporidia infection is revealed in Ev. dubitans (2%; Reads = 12663) and Br. hamata (0.1%; Reads = 240) (Table 5). Microsporidia was detected only at the level of Phylum (S4 Fig). Other endosymbionts, as such Flavobacterium, were found in Pi. pia (0.03%; Reads = 43), Ev. dubitans (0.34%; Reads = 2356) and ps. panamensis (0.8%; Reads = 3531) (Table 5). In Cx. nigripalpus (0.29%; Reads = 1386) and Pi. pia (0.03%; Reads = 160) ASVs of Asaia, were also determined (Table 5). We also showed the first Bartonella record in Cx. nigripalpus (Department of Antioquia) with a lower relative abundance (0.2%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Presence and number of reads (relative abundance) of several endosymbionts in the microbiota of sand flies and Culex nigripalpus.

ASVs at the level of the genus.

https://doi.org/10.1371/journal.pntd.0009942.t005

We also explored the presence of archaea in sand flies and mosquito’s species (S6 Fig). Archaeal ASVs showed a low overall abundance relative to ASVs compared to bacteria. The most abundant ASVs belonged to Crenarchaeota (83%), Euryachaeota (16%), and Thaumarchaeota (1%) (S6 Fig). Crenarchaeota (Family Thermoproteaceae) was mainly found in Ev. dubitans (reads of ASV = 195) with 0.04% of total microbiota, while Euryachaeota (genus Halalkalicoccus and Halococcus) was found in Pi. evansi (ASV = 42; 0.01%). Crenarchaeota and Thaumarchaeota (Family Nitrososphaeraceae) also have a presence in Pi. pia but in low abundance.