Bacterial Communities of Bartonella-Positive Fleas: Diversity and Community Assembly Patterns

Abstract

We investigated the bacterial communities of nine Bartonella-positive fleas (n = 6 Oropsylla hirsuta fleas and n = 3 Oropsylla montana fleas), using universal primers, clone libraries, and DNA sequencing. DNA sequences were used to classify bacteria detected in a phylogenetic context, to explore community assembly patterns within individual fleas, and to survey diversity patterns in dominant lineages.

Animals harbor microbial communities that play integral roles in organismal function; for example, aphids depend on intracellular endosymbionts for essential nutrients (20); termites harbor protists and endosymbionts to convert cellulose into palatable nutrients (18, 21); marine bivalves harbor endosymbionts in their gill tissue, and these organisms provision energy for the host (10); and photosynthetic microbial eukaryotes inhabiting corals account for the high rate of primary productivity and explain the existence of reefs (9).

Fleas, like other animals, harbor a bacterial community. Previous molecular surveys of flea-related bacteria have been screened for the presence or absence of specific pathogenic lineages (4, 12, 14, 17, 25, 26). However, within an individual flea, pathogens are members of bacterial communities, and they may interact with coexisting bacteria. Bacterial communities in other hematophagous invertebrates including ticks, mosquitoes, lice, and leeches have been targeted by using a clone library approach (1, 13, 24, 28).

Here, we used clone libraries and 16S rRNA gene sequence data to characterize bacterial communities in two flea species, Oropsylla hirsuta, collected from three black-tailed prairie dogs, Cynomys ludovicianus (from the family Sciuridae), and Oropsylla montana, collected from one rock squirrel, Spermophilus variegatus (also from the Sciuridae family). Prairie dogs occur in the plains, and rock squirrels occur in the foothills of the Rocky Mountains; we collected fleas in Boulder County, CO, where these two ecosystems converge. O. hirsuta bacteria were screened for the presence of Bartonella strains before selection for this study at the Centers for Disease Control and Prevention in Fort Collins, CO, by using Bartonella-specific primers to amplify the gltA gene (26). Fleas were kept at −20°C until they were identified based on morphology (3, 5).

We cleaned the fleas to minimize the contribution of ephemeral bacteria on external flea parts. Fleas were washed twice (0.133 M NaCl, 1.11% sodium dodecyl sulfate, 0.0088 M EDTA), treated with a lysozyme (11.6 mg/liter, for 30 min, at 37°C), and washed again (three washes in total). Before DNA was extracted (Qiagen DNeasy tissue kit) from the cleaned fleas, they were crushed and subjected to a lysozyme treatment (2 mg/ml, for 30 min, at 37°C) to lyse spore-forming bacteria.

16S PCR products were generated using the universal primers 27f (5′AGAGTTTGATCCTGGCTCAG) and 1492r (5′GGTTACCTTGTTACGACTT) (11). Purified PCR products were cloned using a pGem-T cloning kit (Promega). Cells were shipped to Functional Biosciences, Inc. (Madison, WI), for two-pass sequencing. Forward and reverse sequencing readings were assembled using Sequencher version 4.6 software (Gene Codes Corp.). We obtained 305 16S rRNA gene sequences (GenBank accession no. EU137334 to EU137638). Sequences were aligned by using the Greengenes NAST aligner (2) imported into ARB (15) and corrected by hand. Aligned sequences were checked for chimeras using Bellerophon software (6). Sequences were classified according to Greengenes phylogeny as implemented in ARB (15). Ten “families” of proteobacteria represent greater than 90% of the total discovered diversity (Table 1).

TABLE 1.

Classification of 305 16S rRNA gene sequences from O. hirsuta and O. montana

Bacteria	No. of gene sequences (no. of fleas harboring the sequence) harbored bya:
	O. hirsuta	O. montana
Acinetobacter
Propionibacterineae
Propionibacteriaceae	2 (2)	0
Bacteroidetes
Bacteroidales
Porphyromonadaceae	2 (1)	0
Flavobacteriales
Flavobacteriaceae	3 (2)	0
Firmicutes
Bacilli
Lactobacillales	10 (2)	0
Staphylococcaceae	2 (2)	0
Lachnospiraceae
Anaerofilum	1 (1)	0
Ruminococcus	1 (1)	0
Sporobacter	1 (1)	0
Unclassified	1 (1)	0
Fusobacteria
Unclassified	6 (1)	0
Gemmatimonadetes
Gemmatimonadales	1 (1)	0
Proteobacteria
Alphaproteobacteria	172 (6)	61 (3)
Gammaproteobacteria, Betaproteobacteria	39 (5)	3 (2)
Total	241 (6)	64 (3)

^aNumbers in parentheses indicate the number of fleas in which a particular lineage was detected. A more detailed presentation of proteobacterial groups is presented in Fig. 1.

Bacterial diversity is not equally represented among individual fleas; one lineage (e.g., from Bartonella or Rickettsiales) typically dominates community membership (Fig. 1). This pattern has been found in symbiotic communities of other invertebrates. In individual wood-boring bivalves, multiple lineages often coexist, but one lineage typically dominates the symbiotic community (16). Competition among community members, precedence effects related to the timing of colonization, and physiochemical differences among hosts were suggested as possible explanations for these dynamics of dominance. Within fleas, another potential explanation is that bacterial lineages may respond differently to pulses of mammalian blood. Abundance data for lineages within a large number of fleas or laboratory studies are needed to further investigate community assembly patterns.

FIG. 1.

The proportions of 10 proteobacterial “families” within individual fleas are depicted. Samples are arranged based on Bartonella abundance data. Individual fleas tend to be dominated by one bacterial lineage. Numbers in parentheses indicate the total number of proteobacterial DNA sequences obtained from each flea. Fleas were collected from three different prairie dogs (prairie dog 1 harbored O. hirsuta fleas 1 to 3 [Oh_1 to Oh_3]; prairie dog 2, Oh_4; prairie dog 3, Oh_5 to Oh_6).

For dominant lineages (e.g., Bartonella and Rickettsiales), we used ModelGenerator (7) to determine the best model of evolution and MultiPhyl (8) to construct phylogenetic trees (Fig. 2 and 3). Prior to generating the Rickettsiales tree, we removed the most variable nucleotide positions using the LANE mask implemented in ARB.

FIG. 2.

Maximum likelihood (GTR plus G correction model) tree of Bartonella from O. hirsuta (Oh) and O. montana (Om), with a number of sequences from GenBank. The clade with both O. hirsuta and O. montana is indicated by Om/h. Bartonella strains obtained from GenBank in the tree include B. washoensis (w; accession no. AF070463), two Bartonella sequences from Spermophilus sp. in China (C; accession no. DQ641913 and DQ641912), B. quintana (accession no. M73228), B. koehlerae (accession no. AF076237), B. grahamii (accession no. Z31349), B. bacilliformis (accession no. M65249), B. clarridgeiae (accession no. X97822), and a Bartonella sequence from another O. hirsuta (accession no. DQ473482). The tree was constructed using all of the Bartonella sequences detected, but some were removed to minimize the figure's size.

FIG. 3.

Maximum likelihood (GTR plus G correction model) tree of Rickettsiales detected in O. hirsuta (Oh) and O. montana (Om). Except for R. typhi, R. prowazekii, Rickettsiales bacterium “Montezuma” (R. ‘montezuma’), Orientalis tsutsugamushi, Anaplasma sp., Ehrlichia sp., and C. caryophilus, the taxon names indicate the genus of the host in which the bacteria were found (e.g., Wolbachia bacteria were found in a Nasonia parasitoid wasp host). Sequences from GenBank in the tree include Caedibacter caryophilus (accession no. AY753195), O. tsutsugamushi (accession no. U17258), Rickettsia typhi (accession no. AE017197), R. prowazekii (accession no. M21789), Rickettsiales bacterium “Montezuma” (accession no. AF493952), Anaplasma sp. (accession no. AY527214), Ehrlichia sp. (accession no. DQ324367), Hyalomma (accession no. AM181356), Acanthamoeba (accession no. AF132137), Anopheles (accession no. AY837738), Oh_GenBank (accession no. AY335925), Phtiropsylla (accession no. AY335932), Pulirritans (accession no. AY335926), Nephila (accession no. AF232234), Drosophila (accession no. AY833061), Nasonia (acces sion no. M84688), Myrmeleon (accession no. DQ068803), Paramecium (accession no. X58198), Louse 1 (accession no. AF467369), and Louse 2 (accession no. AF467370). The tree was constructed using all of the Rickettsiales sequences detected, but some were removed to minimize the figure's size.

In the Bartonella phylogenetic tree, the Bartonella strain from an O. hirsuta flea is in a clade with a strain isolated from a Spermophilus squirrel in China, and the Bartonella strain from O. montana is in a clade with B. washoensis (with the exception of one sequence that clustered with the O. hirsuta strains) (Fig. 2). Interestingly, the Bartonella strain from an O. hirsuta flea in GenBank (accession no. DQ473482) clusters with Bartonella clarridgeiae; this suggests that specific strains of Bartonella have not coevolved with specific flea species. Our finding that the Bartonella strain in Boulder, CO, is closely related to a strain detected in China suggests that Bartonella dispersed widely and frequently.

The Rickettsiales phylogenetic tree shows that the Rickettsiales strains detected in this study are unique relative to those of the known diversity (Fig. 3). The Rickettsiales sequences detected in O. hirsuta are most closely related to a sequence detected in the bacteria of a pocket gopher louse (95% sequence similarity) but are not closely related to any other known diversity. Both of the Rickettsiales lineages detected in O. montana are related to those of Wolbachia, but they are more closely related to the Wolbachia lineage detected in flies than to that of Wolbachia from fleas.

Strains of Rickettsiales and Bartonella have different transmission strategies that may affect the accumulation of genetic diversity within a lineage. Bartonella bacteria are horizontally transmitted, and fleas acquire Bartonella lineages through the environment, from host blood, flea feces, etc. Rickettsial lineages, on the other hand, have either a strictly vertical transmission strategy or a mixture of vertical and horizontal transmission strategies (23). We investigated diversity patterns using PAUP version 4.0b10-Altivec (Sinauer Associates) to generate pairwise distances (general time reversible [GTR] plus gamma correction) between all members of each Bartonella and Rickettsiales lineage. We used full sequences for these comparisons (i.e., no LANE mask). A cumulative frequency distribution of pairwise distances depicts patterns of genetic diversity within a lineage (Fig. 4). The average genetic similarity is greater within the Bartonella lineages than within the Rickettsiales lineages. Vertically transmitted bacterial lineages have been found to have a higher rate of evolution (19, 22, 27), and this may explain the higher diversity found in Rickettsiales.

FIG. 4.

Cumulative frequency distributions of pairwise genetic distances (GTR plus G corrected) for Bartonella and Rickettsiales strains from each flea species show that Rickettsiales lineages harbor more genetic diversity than Bartonella lineages. As genetic diversity within a lineage increases, the frequency distribution shifts to the right. Distributions of Bartonella genetic distances are presented as black lines, and Rickettsiales distances are presented as gray lines. Distributions for the two Bartonella lineages almost completely overlap and fall to the left of those for the Rickettsiales lineages. As an example of how to interpret this figure, the data point marked by the arrow indicates that 33.33% of the O. montana (Om) Rickettsiales (Rick) group 1 (G1) shares 99.65% or more sequence similarity.