Wolbachia Endosymbiont of the Horn Fly (Haematobia irritans irritans): a Supergroup A Strain with Multiple Horizontally Acquired Cytoplasmic Incompatibility Genes

Horn flies, Haematobia irritans irritans, are obligate hematophagous parasites of cattle having significant effects on production and animal welfare. Control of horn flies mainly relies on the use of insecticides, but issues with resistance have increased interest in development of alternative means of control. Wolbachia pipientis is an endosymbiont bacterium known to have a range of effects on host reproduction, such as induction of cytoplasmic incompatibility, feminization, male killing, and also impacts vector transmission. These characteristics of Wolbachia have been exploited in biological control approaches for a range of insect pests. Here we report the assembly and annotation of the circular genome of the Wolbachia strain of the Kerrville, TX, horn fly (wIrr). Annotation of wIrr suggests its unique features, including the horizontal acquisition of additional transcriptionally active cytoplasmic incompatibility loci. This study provides the foundation for future studies of Wolbachia-induced biological effects for control of horn flies.

ABSTRACT

The horn fly, Haematobia irritans irritans, is a hematophagous parasite of livestock distributed throughout Europe, Africa, Asia, and the Americas. Welfare losses on livestock due to horn fly infestation are estimated to cost between $1 billion and $2.5 billion (U.S. dollars) annually in North America and Brazil. The endosymbiotic bacterium Wolbachia pipientis is a maternally inherited manipulator of reproductive biology in arthropods and naturally infects laboratory colonies of horn flies from Kerrville, TX, and Alberta, Canada, but it has also been identified in wild-caught samples from Canada, the United States, Mexico, and Hungary. Reassembly of PacBio long-read and Illumina genomic DNA libraries from the Kerrville H. i. irritans genome project allowed for a complete and circularized 1.3-Mb Wolbachia genome (wIrr). Annotation of wIrr yielded 1,249 coding genes, 34 tRNAs, 3 rRNAs, and 5 prophage regions. Comparative genomics and whole-genome Bayesian evolutionary analysis of wIrr compared to published Wolbachia genomes suggested that wIrr is most closely related to and diverged from Wolbachia supergroup A strains known to infect Drosophila spp. Whole-genome synteny analyses between wIrr and closely related genomes indicated that wIrr has undergone significant genome rearrangements while maintaining high nucleotide identity. Comparative analysis of the cytoplasmic incompatibility (CI) genes of wIrr suggested two phylogenetically distinct CI loci and acquisition of another cifB homolog from phylogenetically distant supergroup A Wolbachia strains, suggesting horizontal acquisition of these loci. The wIrr genome provides a resource for future examination of the impact Wolbachia may have in both biocontrol and potential insecticide resistance of horn flies.

IMPORTANCE Horn flies, Haematobia irritans irritans, are obligate hematophagous parasites of cattle having significant effects on production and animal welfare. Control of horn flies mainly relies on the use of insecticides, but issues with resistance have increased interest in development of alternative means of control. Wolbachia pipientis is an endosymbiont bacterium known to have a range of effects on host reproduction, such as induction of cytoplasmic incompatibility, feminization, male killing, and also impacts vector transmission. These characteristics of Wolbachia have been exploited in biological control approaches for a range of insect pests. Here we report the assembly and annotation of the circular genome of the Wolbachia strain of the Kerrville, TX, horn fly (wIrr). Annotation of wIrr suggests its unique features, including the horizontal acquisition of additional transcriptionally active cytoplasmic incompatibility loci. This study provides the foundation for future studies of Wolbachia-induced biological effects for control of horn flies.

INTRODUCTION

Flies from the genus Haematobia (Diptera: Muscidae) are obligate hematophagous ectoparasites of pastured cattle. Two prominent members of this genus are the horn fly, Haematobia irritans irritans, distributed throughout Europe, Africa, Asia, and the Americas (1), and the buffalo fly, Haematobia irritans exigua, which is widespread throughout Asia and Australia (2). Blood-feeding behavior from H. i. irritans results in severe welfare issues and economic losses to cattle industries, with annual estimates of up to ∼$1 billion (U.S. dollars) in North America and ∼$2.5 billion in Brazil (2–4). In Australia, H. i. exigua is estimated to cost the domestic cattle industry $98.7 million (Australian dollars) annually and is currently restricted to the northern part of the country (5). Control of Haematobia flies relies primarily on the use of chemical insecticides; however, reports of insecticide resistance suggest that alternative intervention strategies are required (2, 6, 7).

Wolbachia pipientis is an obligate, endosymbiotic, Gram-negative alphaproteobacterium estimated to infect between 40 and 70% of terrestrial arthropods (8, 9). Wolbachia infection in insects is known to selfishly alter host reproductive biology to transmit and persist in the next generation (10). One mechanism that drives transgenerational Wolbachia persistence is known as cytoplasmic incompatibility (CI) (11, 12). In CI, mating between a Wolbachia-infected male and a noninfected female (unidirectional CI) or a female infected with a different Wolbachia strain (bidirectional CI) results in embryo death (12). The commonly accepted model for CI is “mod/resc.” Here, mod stands for modification of sperm by a toxin in the Wolbachia-infected male and resc for a rescue of sperm by an antidote present in the egg (11, 13). Cellular studies have linked early embryonic death with defects in first zygotic mitosis, irregular chromosomal condensation postfertilization, and delayed histone deposition in the earlier interphase cell cycle (14–17). Two parallel studies recently identified the molecular mechanisms underpinning CI. Using a combined genomic and transcriptomic approach, LePage et al. identified two genes, cifA and cifB, in the prophage WO of Wolbachia strain wMel mediating CI (18), whereas Beckmann et al. demonstrated that two genes, cidA and cidB, cifA and cifB homologs, underpinned CI in the supergroup B Wolbachia strain wPip (19). Further experimental examination of the CI loci suggested a “two-by-one” model, whereby the cifA gene works as the rescue factor and cifA and cifB together instigate CI (20).

In addition to CI, other phenotypes of reproductive manipulation have been reported for Wolbachia, including male killing, parthenogenesis, and feminization (12). Wolbachia has also been demonstrated to confer protection against RNA virus infection in dipteran hosts (9, 10). Both CI and the ability of Wolbachia to restrict RNA viruses form the basis for the deployment of Wolbachia-infected Aedes aegypti mosquitoes for the control of dengue fever and other arboviruses worldwide (21, 22).

Wolbachia has been found to replicate in higher density in organophosphate-resistant Culex pipiens mosquitoes than in susceptible individuals (23, 24). However, no such association between insecticide resistance and Wolbachia density was observed in A. aegypti mosquitoes, suggesting that interactions between the insecticide resistance phenotype and Wolbachia dynamics is both host and Wolbachia strain dependent (25).

While H. irritans flies are not currently known to be vectors of pathogenic viruses in livestock, there exists significant interest in exploiting the CI phenotype of Wolbachia as a form of sterile insect technique in H. i. exigua in Australia. A comprehensive screen of H. i. exigua samples from 12 locations in Australia and also Bali, Indonesia, did not detect Wolbachia (26). In contrast, Wolbachia has been previously identified in many wild-caught populations of H. i. irritans from Mexico (27), field-caught and laboratory colonies from the United States (28, 29), both field-collected and laboratory colonies from Alberta, Canada (26, 30), and also from field-collected samples in Hungary (31).

Due to the intracellular nature of Wolbachia and presence of multiple insertion sequences (IS) within Wolbachia genomes, assemblies using only short-read chemistries are often highly fragmented (32). However, combining PacBio long-read sequencing and Illumina technologies has resulted in the closed and completed Wolbachia genome assembly (32, 33). The genome of the H. i. irritans Kerrville reference strain maintained at the USDA-ARS Knipling-Bushland U.S. Livestock Insects Research Laboratory (Kerrville, TX) was recently assembled using Pacific Biosciences (PacBio) SMRT technology and Illumina chemistries (34). Initial analysis of long- and short-read sequencing data indicated that a large portion of the reads in both libraries shared similarity to those for the Wolbachia endosymbiont of Drosophila simulans strain wRi (34, 35). During the deposition of the H. i. irritans genome in the NCBI database, Wolbachia contigs were removed (F. Guerrero, USDA, personal communication). To recover the Wolbachia genome, we extracted the Wolbachia sequences from the Kerrville H. i. irritans genome project and through bioinformatics analysis were able to assemble a high-quality and circularized wIrr genome. Further, we explored its phylogenetic relationship with the described Wolbachia strains and the possibility of induction of CI by this strain based on what is known about the genes responsible for CI.

RESULTS AND DISCUSSION

wIrr genome assembly, annotation, and genome features.

To extract and assemble the genome of Wolbachia from H. i. irritans, the genomic data from the Kerrville reference genome project (34) were trimmed and mapped against published Wolbachia genomes (32, 35–37) using Burrows-Wheeler Aligner (BWA-MEM) under relaxed mapping criteria (38). Initially, ∼10 million of ∼404 million paired-end Illumina reads and 128,203 of 4,471,713 (2.86%) PacBio reads mapped to representative supergroup A Wolbachia genomes. These reads were then extracted and de novo assembled using Unicycler, resulting in a singular, circularized draft assembly (39). Raw Illumina fastq reads were then iteratively mapped against this draft genome and polished using pilon (40). The final numbers of reads that mapped to the assembled Wolbachia genome were 140,429 out of 4,471,713 (3.1%) from the PacBio library, corresponding to an average coverage of ∼187×, and 10,285,275 out of 404,202,898 (2.54%) from the paired-end Illumina libraries, corresponding to an average coverage of ∼1,280×. The final wIrr genome is 1,352,354 bp, with a GC content of 35.3%, which is similar to findings for other previously assembled supergroup A Wolbachia strains (Table 1). The polished wIrr genome was then annotated using the NCBI prokaryotic genome annotation pipeline (41), which predicted that wIrr contains 1,419 genes, including 1,249 protein-coding genes and 129 pseudogenes, with 56 containing frameshifts, 93 incomplete, 12 with an internal stop, and 31 with multiple problems. The RNA gene repertoire of the wIrr genome was identified to include 34 tRNAs, 3 rRNAs (5S, 16S, and 23S), and also 4 other noncoding RNA genes such as RNase P RNA component class A (RNA family [Rfam] database no. RF00010), bacterial small signal recognition particle RNA (Rfam no. RF00169), 6S RNA (Rfam no. RF00013), and transfer mRNA (Rfam no. RF01849). Completeness of the wIrr genome was assessed by comparing the proteome against 221 single-copy orthologs derived from 1,520 proteobacterial species in the BUSCO pipeline (42). The BUSCO score for completeness of a model organism with a good reference genome is usually above 95%, but for the endosymbiotic bacteria with degenerated metabolic pathways, BUSCO scores can vary between 50% and 95% based on the genome size, presence of repetitive elements in the genome, and individual taxonomic placement (43). The completeness score for wIrr was 82.4%, which included 182 single-copy orthologs, 2 fragmented orthologs, and 37 missing orthologs (see Fig. S1 in the supplemental material), similar to the case with four other completed Wolbachia genome projects (wAu, wMel, wHa, and wRi).

TABLE 1.

Genome features of complete supergroup A Wolbachia strains

Strain designation	Host	Genome size (Mb)	G+C (%)	No. of:							BUSCO score (%)	Reference
				Coding genes (protein)	rRNA	tRNA	Other RNA	Prophage regions	Total genes	Total pseudo genes
wIrr	H. i. irritans	1.35	35.3	1,249	3	34	4	5	1,419	129	82.4	This study
wRi	D. simulans	1.44	35.2	1,254	3	34	4	4	1,403	108	81.9	35
wAu	D. simulans	1.26	35.2	1,099	3	34	4	3	1,265	125	81.9	32
wMel	Drosphila melanogaster	1.26	35.2	1,100	3	34	4	3	1,270	129	81.9	37
wHa	D. simulans	1.29	35.3	1,126	3	34	4	2	1,263	95	81.4	36

We compared the proteome of wIrr against those of the four completed and circularized supergroup A Wolbachia strains (wAu, wMel, wHa, and wRi) using the Orthovenn 2 web server (44). A total of 1,136 conserved orthologs were identified in all five strains. All five strains shared 810 orthologs, with 782 of these being single copy (Fig. S2). The wIrr genome has 1,005 orthologs comprising of 1,248 proteins, mostly involved in cell function and metabolism. While seven clusters of “singletons” were predicted to exist only in wIrr by Orthovenn, closer examination of six of these clusters suggested that these are transposable elements that are present but not annotated in the GenBank Wolbachia genome assemblies. These are discussed further below.

In addition to DNA sequencing data, we explored the transcriptional activity of wIrr in all life stages of H. i. irritans by mapping transcriptome RNA sequencing (RNA-Seq) data used to annotate the genome. As each sample was only sequenced once and poly(A) enriched, it is difficult to make differential gene expression analyses with the data or infer Wolbachia tissue distributions. However, it appears that wIrr is present and transcriptionally active in all life stages and all tissues dissected (Table 2). There is lower transcriptional activity in eggs and pupae than adults, and the highest normalized transcriptional activity was found in adult libraries at 2 h post-blood meal.

TABLE 2.

Transcriptional activity of wIrr in all life stages of the H. i. irritans Kerrville colony

Run accession no.	Sample	No. of:
		Reads total	Reads mapped to wIrr	Transcripts per million
SRR6231662	Malpighian tubules	34,421,850	60,443	1,755.94
SRR6231666	Salivary gland	27,461,942	2,580	93.94
SRR6231664	Adult 0 h postfeed	44,820,832	181,392	4,047.04
SRR6231671	Adult 24 h postfeed	47,070,858	188,823	4,011.46
SRR6231665	Adult 2 h postfeed	51,829,048	827,413	15,964.27
SRR6231670	Adult 4 h postfeed	50,512,062	205,766	4,073.60
SRR6231654	Egg 0 h	34,417,392	34,021	988.48
SRR6231655	Egg 2 h	38,158,656	48,477	1,270.40
SRR6231660	Egg 4 h	31,907,062	64,611	2,024.97
SRR6231661	Egg 9 h	33,832,868	61,261	1,810.69
SRR6231658	Midgut	33,327,082	26,281	788.57
SRR6231659	Legs	38,066,002	175,711	4,615.95
SRR6231663	Ovary	39,372,642	51,187	1,300.06
SRR6231668	Pupae 1 days	53,975,828	274,698	5,089.27
SRR6231669	Pupae 3 days	51,859,710	377,067	7,270.90
SRR6231667	Testes	87,918,073	1,167,195	13,275.93

Phylogenetic placement of wIrr suggests a close relationship with Drosophila supergroup A Wolbachia strains.

Since the discovery of Wolbachia within the gonads of the Culex pipiens mosquito, Wolbachia has taxonomically been considered a single species divided into 16 major supergroups (designated A to Q) (45, 46). While the suitability of classifying the supergroups into a single Wolbachia species is the subject of ongoing debate (47, 48), a universal genotyping tool has been developed to demarcate supergroups based on multilocus sequence typing (MLST) of five ubiquitous genes (gatB, coxA, hcpA, fbpA, and ftsZ) (49). Although MLST clearly demarcates Wolbachia strains in supergroups, it fails to reliably discriminate strains within supergroups with high phylogenetic support. A recent examination of these loci by Bleidorn and Gerth suggested that a number of alternative single-copy loci outperform these five genes (50). To construct a whole-genome phylogenetic analysis of wIrr, we used 79 of the 252 single-copy orthologs from nonrecombinant loci identified by Bleidorn and Gerth from 19 strains of Wolbachia (50). The phylogeny gives strong posterior probability support for the Wolbachia wIrr strain being basal to a clade containing wRec, wAu, and wMel in supergroup A (Fig. 1).

FIG 1.

The Wolbachia endosymbiont of Haematobia irritans irritans wIrr is related to Wolbachia endosymbionts from Drosophila hosts. Shown is a maximum clade credibility (MCC) tree resulting from BEAST analyses of 79 concatenated recombination free gene loci of supergroup A and B Wolbachia strains previously identified by Bleidorn and Gerth (50), resulting in an alignment of 4,968 bp. Posterior probability values are indicated at the nodes. wIrr indicated by an arrowhead, and branch lengths represent the genetic distances.

Natural Wolbachia transfer between hosts can be cladogenic (Wolbachia acquired during the speciation of hosts), introgressive (transfer during mating between closely related host species), or horizontal (possibly via shared food and ecological niche, wounds, and vectors) (51, 52). Concordance between the Wolbachia genome with the host’s mitochondrial and nuclear genome with consistent divergence time shows cladogenic transfer, whereas discordance suggests the possibility of horizontal transmission. Taxonomically, all Drosophilidae belong to the Ephydroidea superfamily of muscomorph flies, in which the Wolbachia strains wAu, wRi, wMel, and wRec have been identified. The Haematobia genus belongs to the superfamily Muscoidea in the insect order Diptera (53). Divergence estimates of Ephydroidea and Muscoidea inferred from mitochondrial genes suggest that the most recent common ancestor of all Haematobia and Drosophila species diverged sometime in the Paleocene, ∼60 million years ago (Mya) (54).

A number of phylodynamic analyses of Wolbachia genomes have attempted to reconstruct evolutionary timescales, albeit with limited concordance between analyses (55–57). One Bayesian analysis of time to most recent common ancestor (TMRCA) conducted by Gerth and Bleidorn (57) on the clade encompassing all Drosophila Wolbachia strains was dated at 48.38 Mya, with a range of 110 to 16 Mya. This fits within the mitochondrial divergence of Haematobia from Drosophila (∼60 Mya) and divergence of Wolbachia from supergroup A members (wMel, wRi, and wRec). Due to the limited genetic data available for Wolbachia infecting Haematobia, we did not make an attempt to formally test the mode of transmission through timescale estimates or phylogenetic discordance, as we may potentially incorrectly conclude the mode of transmission. Therefore, it still remains to be elucidated if the close genetic relationship between wIrr and other Wolbachia may be the result of codivergence or horizontally acquired in H. i. irritans.

The Kerrville Wolbachia wIrr strain is closely related to wild H. i. irritans Wolbachia strains from the United States, Mexico, Canada, and Hungary.

Previous publications have demonstrated the presence of Wolbachia from wild-caught and laboratory colonies of H. i. irritans through amplicon Sanger sequencing of samples (29, 31, 58) or identifying Wolbachia reads in pyrosequencing-based approaches or expressed sequence tags (EST) (27, 28). Currently avaliable GenBank data from Wolbachia of H. i. irritans are limited to partial fragments of the Wolbachia surface protein (wsp) gene (29, 58) or fragments of the 16S rRNA gene (31). BLASTn analysis of the wsp fragment sample of the Kerrville colony used by Jeyaprakash and Hoy, designated wIrr-A1 (GenBank accession number AF217714.1) (29), showed 100% identity with the wsp locus of wIrr (gene E0495; positions 1282799 to 1283488). A similar high nucleotide identity of the wsp fragment of H. i. irritans samples, originating from Lethbridge, Alberta, Canada, designated wIrr (GenBank accession number DQ380856.1), with the wIrr wsp was found: 99.64%, with only two nucleotide differences over an amplicon of 554 bp. In addition, the wIrr 16S rRNA gene (positions 882502 to 884006) and partial 16S rRNA fragments from two Wolbachia strains from H. i. irritans samples from Hungary (GenBank accession numbers EU315781.1 and EU315780.1) were 99.62% identical, with a sequence identity of 264/265 bases. While this suggests that the wIrr strain of Wolbachia is very closely related to the Canadian and Hungarian H. i. irritans samples, the nature of the amplicon size and the high nucleotide identity between strains make it difficult to state this with complete certainty.

As high-throughput sequencing allows for a closer examination of relatedness between the Kerrville wIrr Wolbachia strain and wild-caught H. i. irritans harboring Wolbachia, we reanalyzed EST, DNA sequencing (DNA-Seq), and RNA-Seq data with BLASTn using our wIrr genome as a query from a number of studies using wild-caught flies from Mexico, the United States and Uruguay (Table 3). We identified 5 EST fragments and 394 assembled Wolbachia RNA contigs from wild-caught H. i. irritans from two different studies of Louisiana State University Agricultural Center St. Gabriel Research Station (59, 60) and 4 EST fragments from a cattle farm in Ciudad Victoria, Tamaulipas, Mexico (61). Additionally, for six RNA-Seq libraries of newly emerged male and female horn flies wild-caught in Louisiana, 10% (on average) of each library could be mapped to the wIrr genome (Table S1) (59). All identified contigs shared closer nucleotide identity to the wIrr strain than any other Wolbachia genome deposited in the NCBI database (data not shown). Interestingly, we could not identify any assembled contigs or reads that mapped to the wIrr genome from salivary gland and midgut samples originating from wild-collected H. i. irritans from Canelones, Uruguay (62), suggesting that Wolbachia was either present in very low abundance in these samples or completely absent.

TABLE 3.

Metadata of available sequencing data of H. i. irritans samples

Location of H. i. irritans sample and collection date	Sample and type of sequencing	NCBI accession no.	Reference
Agricultural Center St. Gabriel Research Station Louisiana, USA; collected 2008	Larval and embryonic samples; poly(A)-enriched RNA-Seq, EST	Larval EST, FD457983–FD466257; embryonic EST, FD449556–FD457982	60
Agricultural Center St. Gabriel Research Station Louisiana, USA; collected 28 July 2010	Whole male and females; permethrin-treated surviving males and permethrin- and piperonyl butoxide-treated killed males; poly(A)-enriched RNA-Seq, Illumina Genome Analyzer II/Illumina HiSeq 2000	Assembled transcriptome, GGLM01000000; Bioproject, PRJNA429442	59
Agricultural Center St. Gabriel Research Station Louisiana, USA; collected 2010	Eggs, larvae, whole male and females; poly(A)-enriched RNA-Seq, 454	Male, SRR003192; female, SRR003191; egg, SRR003190; larvae, SRR003189	95
Ciudad Victoria, Tamaulipas, Mexico; collected prior to August 2010	Abdominal tissues of partially fed adult female; poly(A)-enriched RNA-Seq, EST	HO000420–HO001165, HO004499–HO004744	61
Pressler Cattle Ranch Kerrville, TX; originally collected 2003 and sampled 2010	Single male adult; random DNA sequencing using 454	SRA, SRR1578740	63
Canelones, Uruguay; collected 2016	Salivary glands and midgut samples, poly(A)-enriched RNA-Seq, Illumina HiSeq 2000	SRA salivary glands, SRR5136552, SRR5136553; SRA midguts, SRR5136554, SRR5136555	62

We conducted de novo assembly of the 454 DNA-Seq data originating from a single male H. i. irritans fly collected in 2003 from the Pressler Cattle Ranch in Kerrville, TX (63). Of 1,130 assembled contigs, 74 were identified through BLASTn analysis as having the closest bit score hit to the wIrr genome. As very few Wolbachia genome fragments were conserved from RNA-Seq and DNA-Seq assemblies, we could not construct a single phylogenetic tree for all the samples. However, the close identity of all available transcriptome and genomic data of wild-caught H. i. irritans flies from North American populations, including Mexican, to the Wolbachia Kerrville reference H. i. irritans strain suggests that likely they are also infected with the same wIrr strain.

The wIrr genome has undergone significant genome rearrangements compared to other Wolbachia genomes.

In bacterial genome evolution, horizontal gene transfer (64, 65) and genetic vehicles such as bacteriophages, plasmids, and transposons (mobile element) (65–68) contribute to changes in the bacterial genome. Due to the intracellular niche of the endosymbiont, the evolution of Wolbachia genomes is highly dependent on bacteriophages and transposable elements, with both contributing to sometimes as much as 21% of the genome (65). Whole-genome comparisons of nucleotide synteny between wIrr and wMel and wRi were carried out using MAFFT v. 7 (69). We did not analyze the synteny between wRec (70) and wIrr because the genome is fragmented and yet to be circularized. It appears that while wIrr maintains between 90 and 99% nucleotide identity with the other two strains, wIrr has undergone a high degree of genome rearrangement (Fig. 2A and B). In comparison, wMel and wRi show very similar genome arrangements (Fig. 2C). Similar genomic rearrangement has been previously seen while comparing wPip and wMel, wMel and wBm, and wUni and wVitA (71–73).

FIG 2.

The wIrr strain has undergone genome rearrangements compared to other supergroup A Wolbachia strains. Genomes were compared using the MAFFT (v. 7) algorithm. Shown are dot plots of LAST comparisons with a threshold score of 39 and an E value of 8.4e−11 for wIrr compared to wMel (GenBank ID NC_002978.6) (A), wIrr compared to wRi (GenBank ID NC_012416) (B), and wRi compared to wMel (C). Similarities in the forward orientation (red) and similarities suggesting inversions (blue) are shown.

Expansion of insertion sequence elements in the wIrr genome is associated with a divergent CifB homolog.

Insertion sequences (IS) are diverse transposable elements in bacterial genomes (65, 74). Considerable variation in the IS element composition in Wolbachia genomes is speculated to contribute to diversification or speciation of closely related strains, and IS elements can cause the disruption of protein coding genes, leading to pseudogenes (32, 36). To compare the IS element loads between wIrr and other supergroup A Wolbachia species, wRi, wAu, wMel, and wHa IS elements were identified and searched against the IS finder database using the ISsaga web server (74) (File S1). A total of 283 open reading frames (ORFs) related to IS elements were identified in the wIrr genome, including 61 complete ORFs and 150 partial IS elements. Maximum copies of IS elements were from IS630 (111 copies), which belongs to the Tc1/mariner (class II) transposon family, and ssgr IS1031 (109 copies), which is from the IS5 family. Comparative analyses between wIrr and other supergroup A Wolbachia strains identified 12 conserved IS families between all genomes IS66_ssgr_ISBst12, ISL3, IS5_ssgr_IS1031, IS4_ssgr_IS4, IS4_ssgr_IS231, IS3_ssgr_IS3, IS110, IS110_ssgr_IS1111, IS4_ssgr_IS50, IS630, IS481, and IS5_ssgr_IS903. However, two IS families were identified as exclusive to wIrr: IS5_ssgr_IS427, which has one complete ORF and three partial ORFs, and the IS5_ssgr_ISL2, with two partial ORFs. We manually extracted the IS5_ssgr_IS427 annotations, and within one of the identified loci (positions 632890 and 630128), a disrupted IS5-like element was found with only one single hit. Based on BLASTn similarity (query length, 100%; nucleotide identity, 80.39%; E value, 0), this element is from the Wolbachia endosymbiont of Brugia malayi isolate TRS (GenBank accession number CP034333.1) (72). Immediately after this transposable fragment is the sequence for protein E0495_03245 (Fig. 3A); on the basis of BLASTp analysis, this 546-amino-acid (aa) protein appears to be a truncated CI factor, encoded by cifB, belonging to the wHa Wolbachia endosymbiont of D. simulans (GenBank accession number WP_144054595.1; query cover, 98%; percent similarity, 65.71%; E value, 0.0). We examined the transcriptional activity of this cifB gene by mapping the RNA-Seq data of all life stages to this region of the genome. One paired read mapped to this gene at this location, suggesting reduced transcriptional activity compared to those of other cifA and cifB loci (Fig. 3B). However, due to the nature of the RNA-Seq library preparation with poly(A) enrichment, we are unable to make strong conclusions and comparisons between different genomic loci. The length of IS elements varied between 174 and 1,743 bp, having a median size of 348 bp. The total burden of IS elements on the wIrr genome is 115,692 bp, which is 8.55%. This is similar to the IS element percentage found in wRi (9%), which is double the IS element loads of wMel (4.3%), wHa (4.4%), and wAu (4.4%). These lineage-specific attainments and losses of IS elements, as well as length of the IS element, size, and family distribution, are well documented across Wolbachia strains (36). The discovery of a single IS element shared between wIrr (supergroup A) and Wolbachia from Brugia malayi (supergroup D) is of particular interest. B. malayi is a filarial nematode that relies on a hematophagous mosquito host as a vector. Potentially, the gain of this IS element may have arisen through coinfection of H. i. irritans with a distantly related nematode species, as it seems unlikely to have been independently lost in all other supergroup A genomes. While H. i. irritans is known to vector Stephanofilaria sp. nematodes (75), the presence or absence of Wolbachia within these nematodes is yet to be characterized, and therefore formal testing of IS acquisition cannot be undertaken. Further assembly and genetic characterization of filarial nematodes and their Wolbachia endosymbionts would allow for a better understanding of interaction between the H. i. irritans, Stepahnofilaria, and Wolbachia.

FIG 3.

Expansion of IS elements in wIrr genome is associated with a cifB homolog with limited transcriptional activity. (A) Schematic diagram of genomic loci in wIrr associated with the IS5_ssgr_IS427 IS family identified by ISsaga and BLASTn hits against the wHa genome (GenBank ID NC_021089.1) and the wBm genome (GenBank ID CP034333.1). (B) Transcriptional activity of the putative CifB homolog E0495_03245 was explored through pooling RNA-Seq reads originating from all tissues and developmental stages of all H. i. irritans libraries that were mapped to the wIrr genome. The resultant BAM files were visualized with Integrated Genomics Viewer (IGV v. 2.5.2). Forward mapped reads are shown in red; reverse orientation reads are shown in blue. Light blue and red regions indicate a mapping quality number of 0 (MQ = 0), which indicates that the read maps to multiple regions on the genome.

The prophage regions of wIrr have a reduced eukaryotic association module.

Wolbachia bacteriophages or prophages (WO) have been widely reported for strains from supergroups A, B, and F; however, they have been lost in supergroup C and D strains (76). The tripartite relationship between Wolbachia WO and arthropod hosts is of great interest, as it has been shown that many genes located within prophage regions of Wolbachia genomes include eukaryotic association genes and toxin-antitoxin modules (77), and there is interest in utilizing WO as a candidate for Wolbachia genetic transformation (76, 78). Using the Phaster web server, we identified five potential WO regions in the wIrr genome. The largest is a 60.8-kb region designated “intact” by Phaster with 68 ORFs from positions 359527 to 420415 having head, baseplate, tail, and virulence genes and IS630 family transposons (79). The other four were ∼7-kb incomplete prophage regions containing 10, 9, 12, and 8 ORFs positioned at 613245 to 620397, 859203 to 866672, 903423 to 910665, and 1241523 to 1247571, respectively, in the wIrr genome. Supergroup A members wMel, wRi, wAu, and wHa have between two and four variable WO phage regions with at least one presumed intact and other WO-like degenerated phage regions (32, 35–37). We compared the intact putative prophage region of WOIrr with the predicted WO phage regions from wMel (WOMelB) and completely sequenced WO phage region from wVitA (WOVitA), to identify the conserved region using reciprocal BLASTn analysis (37, 65, 80). The conserved phage regions were visualized using Easyfig (Fig. 4). Previously, it has been reported that a eukaryotic association module (EAM) is present in the WO phage genomes from Wolbachia characterized by proteins that are enriched with eukaryotic-like domains (77). EAMs are enriched for ankyrin repeat domains (ANK), which are involved in regulation of cell cycle, promotion of protein-protein interactions, and Wolbachia-induced reproductive phenotypes (81–83) and vary widely between strains (84). Reciprocal blast comparisons between WOVitA and WOIrr (Fig. 4A) suggest that there is a reduction of the EAM in WOIrr. In WOVitA, the EAM lies between the hypothetical protein gwv_1089 and the patatin-like phospholipase family protein gwv_1104. In WOIrr, there is no ankyrin repeat with PRANC domain protein (gwv_1092), ankyrin, and tetratricopeptide repeat family protein (gwv_1093). These are also not found in chromosomal wIrr. Additionally, the only protein that is conserved in the region corresponding to the EAM of WOVitA and the WOirr prophage is the patatin-like phospholipase E0495_02120 which shares 78.88% pairwise protein identity with BLASTp analysis with gwv_1104. The EAM is also missing in WOMelB (Fig. 4B).

FIG 4.

Gene order comparisons between WO prophages. Shown are reciprocal BLASTn analyses of Comparisons between WOVitA (GenBank ID KX522565) and WOIrr (A) and Comparisons between WOMelB (GenBank ID NC_002978.6) and WOIrr (B). Genomic loci in WO prophages were analyzed using Easyfig and matching loci with the maximum E value (0.001). Regions of nucleotide identity are indicated by gray shading from 100 to 65%. Annotations of genes are colored based on automated NCBI annotation and manual Pfam protein database curation. The predicted eukaryotic association module (EAM) is shown on WOVitA.

Horizontal acquisition of Wolbachia cytoplasmic incompatibility loci in wIrr.

To explore the genetic diversity of CI genes in wIrr, we explored orthologous clusters for the previously described CI genes. In addition to the truncated cifB (E0495_03245) gene, we found two complete and genetically distant CI operons in wIrr, with one located within the WOIrr region (gene identifiers [ID] E0495_02160 and E0495_02165) and the second (gene ID E0495_02270 and E0495_02275) downstream of the WOIrr region. BLASTp analysis of the predicted protein sequences (Table 4) indicated that these CI genes are not duplications as previously reported for wRi (35).

TABLE 4.

Cytoplasmic incompatibility genes identified in wIrr and their closely related proteins

Gene	Size (aa)	Positions	Top BLASTp hit, query cover, % identity, GenBank ID
E0495_02160 cidA	483	414360–415811	cidA wPip, 100%, 82.89%, AGR50404.1
E0495_02165 cifB	1,132	415858–419216	cifB wHa, 99%, 91.81%, WP_144054595.1
cifA E0495_02270	474	441815–443239	cifA wMel, 100%, 99.79%, WP_044471237.1
cifB E0495_02275	1,166	443315–446,815	cifB wMel, 99%, 99.40%, AYE93038.1
E0495_03245	546	629487–631127	cifB wHa, 98%, 65.71%, WP_144054595.1

The CI genes of Wolbachia have been grouped into four different phylogenetic groups (types I to IV) (18, 85); therefore, we conducted a Bayesian phylogenetic analysis of the complete CI genes of wIrr (Fig. 4). For one set of CI genes (genes E0495_02270 and E0495_02275) located just outside the predicted WO region of wIrr, both copies of cifA and cifB genes phylogenetically clustered to a clade encompassing wRi and wMel (Fig. 5), showing high posterior probability support. The second cifA and cifB gene set, which sits within the WOIrr region (genes E0495_02160 and E0495_02165), groups phylogenetically with Wolbachia cifA and cifB genes originating from WOSol, which is a WO phage infecting the Wolbachia strain (wSol) from the fig wasp (Ceratosolen solmsi) (86), and wHa, which is a Wolbachia strain infecting D. simulans (Fig. 1) from a Wolbachia clade evolutionarily distinct from wIrr. Based on phylogenetic discordance and high posterior support for ancestors of the cifA and cifB genes, there is reasonable support for the horizontal acquisition of the E0495_02160 and E0495_02165 genes in the wIrr genome from other distantly related Wolbachia species. This report is similar to another independent acquisition of CI genes in the Wolbachia endosymbionts of the Drosophila yakuba clade which cause weak intra- and interspecific CI (87). Cooper et al. assembled the genomes of wYak variants and demonstrated that while there appears to be another CI locus in these genomes, the presence of an inversion introduces several stop codons within the cidB^wYak-clade locus relative to the same region in cidB^wMel, speculated to potentially render this gene nonfunctional (88). To support the horizontal acquisition of these CI loci in wYak, Cooper and colleagues compared pairwise differences between the homologs and nuclear genes. By comparison, both genes within the CI loci in wIrr are seemingly complete, with no premature stops, and presumed to encode functional proteins. Previous studies have suggested that the CI gene sets cifA and cifB vary in copy number across CI-inducing Wolbachia strains and are directly correlated with the extent of CI (strong or weak) (85). The acquisition of a second set of CI genes corroborates previously unpublished experiments conducted in which wIrr Wolbachia from the Kerrville reference strain demonstrated a strong CI phenotype (F. Guerrero, USDA, personal communication). The transcriptional activities of the CI genes have previously been explored by Lindsey et al., who demonstrated that both cifA and cifB show differential transcriptional activities across host development (85). Again, RNA-Seq data of all life stages were mapped to the wIrr genome, and we examined the mapped reads at these two CI loci. Reads mapped exclusively to one CI region, and very few reads mapped to both (MAPQ score 0). In general, the cifA gene was more transcriptionally active than the cifB gene in both loci, as also previously reported (Fig. 6) (85). The evidence of two transcriptionally active CI loci may explain the high incidence of Wolbachia in wild-caught specimens of H. i. irritans, as Wolbachia has been identified in 100% of all collected individuals from Hungary (10/10) (31), as well as all 15 tested horn flies from two wild locations in Alberta, Canada, and also in 54/55 individuals tested in two independent screens of the laboratory colony of Lethbridge Research Centre, Alberta, Canada (26).

FIG 5.

The Wolbachia endosymbiont wIrr has horizontally acquired a second cytoplasmic incompatibility loci. Shown are maximum clade credibility trees resulting from BEAST analyses of cifA (A) and cifB (B) homologs with type numbers as designated by Lindsey et al. (85). Posterior probability values are indicated at the nodes. wIrr CI genes are indicated by arrowheads, and branch lengths show genetic distances.

FIG 6.

Both cytoplasmic incompatibility loci are transcriptionally active in the Wolbachia strain of wIrr. Pooled RNA-Seq reads originating from all tissues and developmental stages of all H. i. irritans libraries were mapped to identified CI loci in wIrr genomes. Resultant BAM files were visualized with Integrated Genomics Viewer (IGV v. 2.5.2). Forward mapped reads are shown in red; reverse orientation reads are shown in blue. Light blue and red reads indicate a mapping quality number of 0 (MQ = 0), which indicates that the read maps to multiple regions on the genome.

Conclusion.

In this study, we assembled and annotated a high-quality genome of the Wolbachia endosymbiont of H. i. irritans designated wIrr. Phylogenetic analysis of the wIrr strain suggests that the wIrr belongs to a well-supported supergroup A lineage that includes the well-studied wMel, wAu, and wRi Wolbachia strains from Drosophila spp. Comparative genomics of wIrr indicated acquisition of additional transcriptionally active CI loci. Phylogenetic analysis indicates either horizontal acquisition of these genes from a closely related Wolbachia strain or the potential loss of CI loci in other Wolbachia strains infecting Drosophila spp. The wIrr genome has undergone significant reassortment compared to closely related and completely assembled strains. Additional analysis of available and deposited sequencing data from wild-caught and laboratory H. i. irritans colonies suggest that wIrr is the most closely related to wild U.S. and Mexican samples and a close relative of Canadian and Hungarian samples. This study provides the foundation for future functional studies of effects that Wolbachia may have on life history traits of H. i. irritans such as insecticide resistance and for evaluating the contribution of wIrr toward population control.

MATERIALS AND METHODS

Genomic DNA and RNA sequencing data.

The Kerrville reference H. i. irritans strain is a closed fly colony which has been maintained at the USDA-ARS Knipling-Bushland U.S. Livestock Insects Research Laboratory since 1961 (34). Genomic DNA from unfed adult flies of mixed-sex originating from this strain was subjected to whole-genome sequencing, and previously deposited in the National Center for Biotechnology Information Sequence Read Archive (SRA) (accession number PRJNA30967) (34). Briefly, these data include two PacBio runs: one 10-kb and two 20-kb insert libraries. Libraries of 10 kb were sequenced using C2 chemistry and P4 polymerase (P4-C2; Pacific Biosciences of California, Inc.), whereas C3 chemistry and P5 polymerase (P5-C3; Pacific Biosciences of California, Inc.) were used for both 20-kb libraries with 3 h of movie time. The 10-kb libraries and two of the 20-kb libraries were sequenced on 12 SMRTCells, 4 SMRTCells, and 8 SMRTCells, respectively, and all the sequences were finally pooled and uploaded under the same accession number (SRA accession number SRR6231657). For Illumina sequencing, one short-insert paired-end library and one mate-paired end library with a 6- to 12-kb insert size were sequenced as 100-nucleotide (nt) paired ends on the HiSeq2000 and uploaded under the same accession number (SRA accession number SRR6231656). Additional RNA sequencing data from different life stages and tissues of the horn flies were sequenced on an Illumina HiSeq 2000 using 2 × 100-nt configuration and are available with the above Illumina read accession number (SRA accession number SRR6231656).

Wolbachia genome assembly, polishing, and RNA-Seq analysis.

Raw fastq files originating from Illumina and PacBio sequencing data were imported to the Galaxy Australia web server (https://usegalaxy.org.au/, v. 19.05; accessed between May and October 2019). The Nextra universal transpose Illumina sequence adapters were removed and reads were quality trimmed using Trimmomatic (Galaxy v. 0.36.4) under the following conditions: sliding window = 4 and average quality = 20 (89). Resultant clean reads were mapped to the genomes of wRi (35) and wAu (32) using BWA-MEM (Galaxy v. 0.7.17.1) (38) under default parameters and under simple Illumina mode and PacBio mode (-x pacbio) for subsequent libraries. Mapped reads were extracted using a BAM filter (Galaxy v. 0.5.9) and were then assembled using Unicycler (Galaxy v. 0.4.1.1) (39). For RNA-Seq analysis, we also used BWA-MEM (Galaxy v. 0.7.17.1) (38) under default parameters and under simple Illumina mode. To visualize mapped RNA-Seq data, resultant BAM files were visualized with Integrated Genomics Viewer (IGV v. 2.5.2).

Genome annotation and comparative genomics.

Coding regions and noncoding RNAs (ncRNAs) of the assembled wIrr genome contig were annotated using the NCBI prokaryotic genome annotation pipeline (41). To assess the quality of the assembly, BUSCO v. 3.1.0 was used to search for orthologs of the near-universal, single-copy genes in the BUSCO proteobacterium database (42). As a control, we performed the same search using the reference genomes for wRi (35), wAu (32), wMel (37), wHa (36), and wNo (36) as well as the complete wAlbB genome (33). Identification of phage and prophage regions of wIrr was conducted using the PHASTER web platform (https://phaster.ca/; accessed 4 September 2019) (79). Groupings of orthologous clusters were identified using the Orthovenn2 web server (https://orthovenn2.bioinfotoolkits.net/; accessed 5 May 2019) (44) under the following conditions: E value of 1e−2 and inflation value of 1.5. Insertion sequence (IS) elements of wIrr were identified using the ISsaga web server platform (http://issaga.biotoul.fr/; accessed 8 August 2019) (90). For nucleotide synteny plots of wIrr, MAFFT (https://mafft.cbrc.jp/alignment/server/; accessed 8 July 2019) (91) was used to align wIrr and other genomes and then results were visualized by dot plots of matches (without extensions) identified using the LAST algorithm, which compares sequences by adaptive and fixed-length seeds (score = 39; E value = 8.4e−11). Comparisons between the putative prophage regions of wIrr were examined using BLASTn and visualized using Easyfig v. 2.2.2 (92).

Phylogenetic analyses of wIrr and cytoplasmic incompatibility loci.

For full-genome phylogenetic analyses, we used 79 nonrecombinant gene loci, which have been previously determined by Bleidorn and Gerth to perform well from 19 strains of Wolbachia (50). These were downloaded (https://github.com/gerthmicha/wolbachia-mlst; accessed September 2019), aligned using MUltiple Sequence Comparison by Log-Expectation (MUSCLE v. 3.8.98) installed as part of the CLC Genomics Workbench (v. 11.0.1) (93), and concatenated. The resultant alignment was analyzed using Bayesian evolutionary analysis by sampling trees (BEAST v. 2.5.1) (94), split into individual codon positions with linked site model and unlinked clock model under the General Time Reversible and Gamma = 4 nucleotide substitution model. Clock rates were drawn from a log-normal distribution. Additional parameters were a chain length of 10 million steps sampling every 10,000 steps under a Yule model. For phylogenetic placement of the CI genes within wIrr, identified Cif homologs were first aligned using MUSCLE (93) and also subjected to BEAST (94) with 10 million steps with a pre-burn-in of 100,000 with sampling being conducted every 20,000 steps under a Yule model and a general empirical model of protein evolution (WAG) amino acid substitution model. For both BEAST runs, convergence for all parameters as well as stationary distributions of the MCMC chain were inspected using Tracer v. 1.7.1 (effective sample sizes of >400). The maximum clade credibility (MCC) tree (i.e., the tree with the largest product of posterior clade probabilities) was selected from the posterior tree distribution using the program TreeAnnotator (included in the BEAST package) after a 10% burn-in. Resultant MCC trees were then visualized using FigTree v. 1.4.4.

Data availability.

PacBio and Illumina raw sequencing data are available from the NCBI Sequece Read Archive under accession numbers SRR6231657 and SRR6231656, respectively. The assembled Wolbachia pipientis wIrr strain has been deposited in GenBank under the accession number CP037426. Additional sequencing data and metadata used for validation are available in the supplemental material.