The sialotranscriptome of the gopher-tortoise tick, Amblyomma tuberculatum

Abstract

The gopher tortoise tick, Amblyomma tuberculatum, is known to parasitize keystone ectotherm reptile species. The biological success of ticks requires precise mechanisms to evade host hemostatic and immune responses. Acquisition of a full blood meal requires attachment, establishment of the blood pool, and engorgement of the tick. Tick saliva contains molecules which counter the host responses to allow uninterrupted feeding on the host. RNASeq of the salivary glands of Amblyomma tuberculatum ticks were sequenced resulting in 138,030 pyrosequencing reads which were assembled into 29,991 contigs. A total of 1875 coding sequences were deduced from the transcriptome assembly, including 602 putative secretory and 982 putative housekeeping proteins. The annotated data sets are available as a hyperlinked spreadsheet. The sialotranscriptome assembled for this tick species made available a valuable resource for mining novel pharmacological activities and comparative analysis.

Introduction

The gopher tortoise (Gopherus polyphemus), a keystone species native to the Southeastern United States, is listed as threatened under the US Endangered Species Act in the Western part of its range (Service, 1987) and the Eastern populations have been classified as a candidate for federal listing as a threatened species (Dodd Jr, 1995; Service, 2015). The gopher tortoise is a keystone species because its burrows are utilized by 350 vertebrate and invertebrate commensals (Catano and Stout, 2015; Johnson et al., 2017).

Amblyomma tuberculatum Marx, 1894 is commonly known as gopher tortoise tick for parasitizing the gopher tortoise. The gopher tortoise ticks feed for several days or weeks on the gopher tortoises and obtain a vast blood meal up to 100 times their unfed weight. Tick hematophagy (blood-feeding) is usually divided into slow feeding and fast feeding. Tick attachment involves the penetration of the skin by the chelicerae and the secretion of salivary cement, which helps affix the tick at the feeding site. Tick saliva also assists in feeding, its complex mixture of compounds disarming the host’s attempts to establish hemostasis and tissue repair. Tick saliva composition, as revealed by sialotranscriptome (from the Greek, sialo means saliva), indicates the presence of over 5,000 putative secreted peptides, containing representatives of dozens of protein families (Chmelar et al., 2016; Gulia-Nuss et al., 2016; Karim and Ribeiro, 2015; Karim et al., 2011). Sialomes of ixodid ticks are an invaluable resource for the tick biologists and will help in comparative analysis to better understand the complex biology of tick hematophagy and evolution. Very limited work has been done to investigate tick feeding on gopher tortoises (Budachetri et al., 2016; Cooney and Hays, 1972; Ennen and Qualls, 2011). In this study, we report an annotated catalogue of salivary gland expressed transcripts from adult female A. tuberculatum resulting from over 138,000 pyrosequencing sequences, representing, to our knowledge, the first analysis of a tick sialome from a parasite exclusively feeding on an ectotherm. This study is also relevant by describing the sialome of a tick feeding on its natural host, because tick sialomes may change according to the particular host that the tick is feeding (Narasimhan et al., 2019). This study complements the sialotranscriptome of other Amblyomma species (Aljamali et al., 2009; Batista et al., 2008; Esteves et al., 2017; Garcia et al., 2014; Karim and Ribeiro, 2015; Karim et al., 2011; Maruyama et al., 2010; Moreira et al., 2017; Ribeiro et al., 2011).

Materials and Methods

Ethics Statement

All animal experiments were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol for the collection of ticks from Gopher tortoises was approved by the Institutional Animal Care and Use Committee of the University of Southern Mississippi (protocol # 10042001).

Ticks

All ticks were collected directly from gopher tortoises that were trapped using a collapsible Tomahawk^R Model 18 Live Trap (81.28 X 25.4 X 30.48) covered with burlap and pine needles to provide shade. The site of collection was from Wiggins Airport (Wiggins, Mississippi) [30°50’32” N, −89°09’37”W] and Crossroads site within the DeSoto National Forest, MS [30°57’24”N, −89°06’32”W]. Ticks were manually pulled from the tortoise using fine forceps. The partially blood fed ticks were brought into the lab and cleaned as described earlier (Budachetri et al., 2016). Ticks were identified using standard morphological keys. Tick salivary glands were dissected from 10–12 engorged female ticks. The dissecting solution was ice cold 100 mM 3-(N-Morpholino)-propanesulfonic acid (MOPS) buffer containing 20 mM ethylene glycol bis-(β-aminoethyl ether)-N, N, N’, N’ -tetraacetic acid (EGTA), pH 6.8. After removal, glands were washed gently in the same ice-cold buffer. The dissected SGs were stored immediately after dissection in RNAlater (Ambion Inc., Austin, TX, USA) prior to isolating mRNA.

RNA and Library preparation

These were done essentially as reported before (Karim et al., 2011). Briefly, RNA Poly A+ mRNA was isolated using the Illustra™ QuickPrep micro mRNA purification kit (GE Healthcare, Piscataway, NJ, USA. mRNA quality was determined by lab-on-chip analysis using the 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA). The mRNA quantity was determined by Nanodrop.

Library preparations for GS FLX titanium (Roche/454 Life Sciences, Branford, CT, USA) sequencing were developed in the Center for Genomics and Bioinformatics, Indiana University. The cDNA library was normalized using Trimmer cDNA normalization kit (Evrogen, Moscow, Russia) according to manufacturer’s instructions. Normalized DNA was fragmented by sonication, and ends were enzymatically blunted and ligated to customized 454 adaptors. Emulsion PCR reactions were performed according to the manufacturer (Roche 454 Life Sciences). Sequencing adapters (A and B) were automatically removed from the reads using signal processing software (Roche 454 Life Sciences).

Bioinformatics Analysis

Assembly of all reads were done using the assemblers Abyss and Soapdenovo-Trans with every other kmer (-k program switch) parameter from 17 to 85 (Birol et al., 2009; Liu et al., 2011; Luo et al., 2012; Miller et al., 2010; Simpson et al., 2009). Resulting contigs were re-assembled by a pipeline of blastn and cap3 assembler (Huang and Madan, 1999) as described earlier (Karim et al., 2011). Coding sequences were extracted based on blastx (Altschul et al., 1997) results deriving from several database matches, including a subset of the non-redundant protein database of the National Center for Biotechnology Information (NCBI) containing tick and other invertebrate sequences, as well as the Swissprot and Gene Ontology (GO) databases. The longest open reading frame was also extracted if it had a signal peptide indicative of secretion as evaluated by version 3.0 of the SignalP program (Nielsen et al., 1999).

Data sharing

All coding sequences are available for browsing in additional file S1 which also contains hyperlinks to several databases, as explained previously (Karim et al., 2011; Ribeiro et al., 2004). This project was registered at the National Center for Biotechnology Information (NCBI) under Bioproject ID PRJNA595760 and Biosample ID SAMN13569030. Deducted coding sequences and their translations were deposited to DDBJ/EMBL/GenBank under the accession GIDH00000000. The version described in this paper is the first version, GIDH01000000.

Results and Discussion

A total of 138,030 pyrosequencing reads were assembled into 29,991 contigs. A subset of this data containing 1,875 coding sequences larger than 150 nucleotides and mapping at least 2 reads was assembled from 64,852 reads. (Spreadsheet S1 and Table 1). This subset maps 47 % of all raw reads. Five broad categories of expressed genes are observed in spreadsheet S1 and summarized in Tables 2. The putatively secreted (S) category contained ~ 59% of the reads and ~32% CDS; the housekeeping (H) category had ~39% reads with ~52% CDS; ~2% of reads were not classifiable, constituting 265 contigs which were assigned to the unknown (UK) class; Transposable elements (TE class) accounted for 25 CDS and 93 reads, and 2 rickettsial reads derived from microbes.

Table 1:

Length statistics for raw reads, assembled contigs and extracted coding sequences (CDS)

	Raw reads	Contigs	Extracted CDS (1)
Total number of sequences	138,030	29,991	1,875
Average length	323.73	335.23	443.45
Median size	459	471	501
L50	417	400	393
< 100	31,527	8,977	0
101–200	11,090	2,634	168
201–300	8,112	1,282	410
301–400	13,957	2,121	381
401–500	60,992	10,189	332
501–600	12,305	2,063	209
601–700	28	947	174
701–1000	14	1,254	146
1001–2000	5	510	52
2001–3000	0	13	3
3001–5000	0	1	0
Total	138,030	29,991	1,875

⁽¹⁾Only CDS larger than 150 nt with 2 or more reads were retained

The CDS derived from assembly of 64,852 reads, or 46.98% of all raw reads

Table 2:

Five broad categories of expressed genes from the sialotranscriptome of Amblyomma tuberculatum

Class	Number of CDS	Number of reads	%CDS	% Reads
Secreted	602	37,724	32.107	58.489
Housekeeping	982	25,362	52.373	39.322
Transposons	25	93	1.333	0.144
Rickettsia	1	2	0.053	0.003
Unknown	265	1,317	14.133	2.042
Total	1,875	64,498	100	100

The following is a guide to browsing spreadsheet S1.

Housekeeping (H) Genes

A total of 982 CDS ascribed to H-class genes expressed in the SGs of Am. tuberculatum were further characterized into 22 subgroups according to function (Table 3 and spreadsheet S1). CDS of the H class represents 52.37% of all CDS and 39.32% of total reads (Table 3). As shown in Table 3, detoxification and oxidant metabolism / detoxification enzymes contain altogether 624 reads which include sulfotransferases and selenoproteins. Sulfotransferases have been shown to play a role in dopamine detoxification (Pichu et al., 2011; Yalcin et al., 2010), a neurotransmitter associated with tick salivary secretion (Kaufman, 1978). Enzymes which neutralize oxidative and endoplasmic reticulum (ER) stress such as superoxide dismutase (cytoplasmic), superoxide dismutase (mitochondrial), glutathione peroxidase, thioredoxin peroxidase, salivary selenoprotein precursor, selenoprotein K and selenoprotein T are also present in the sialotranscriptome of A. tuberculatum. Carboxypeptidases and leukotriene hydrolases were also detected under amino acid metabolism related genes. Energy and lipid metabolism genes altogether contain 2,450 reads including pyrophosphatases which promotes the rapid hydrolysis of inorganic pyrophosphate and activates fatty acids for beta oxidation. Several immunity-related proteins such as TNF receptor-associated factor-like protein, Toll like receptors, interferon alpha-inducible proteins etc. were annotated. Other additional classes of housekeeping sequences related to protein synthesis machinery, energy metabolism, protein modification, signal transduction, protein export and transcription machinery were also found abundantly expressed in salivary glands.

Table 3:

General classification of extracted coding sequences (CDS) from the sialotranscriptome of Amblyomma tuberculatum

Class	Number of CDS1	Total reads2
Putative secreted protein
Enzymes
5′-nucleotidase	2	41
Dipeptidyl peptidase (Kininase?)	2	5
Endonuclease	1	27
Sphingomyelinase and lipases	2	12
Metalloproteases	16	498
M13 peptidases - may be for peptide processing	12	855
Protease inhibitor domains
Serpins	1	23
Kunitz domain containing	42	3388
Kazal domain	2	18
TIL domain	18	1005
Cystatins	5	77
Thyropins	2	23
Basic tail	9	507
Lipocalins	76	17105
Antigen 5	2	8
Immunity related
Antimicrobial peptides
Defensins	4	23
Microplusin	6	200
5.3 kDa peptide	1	4
Pathogen recognition domains
ML domain	3	43
Peptidoglycan recognition	1	15
Ixoderins	2	8
Mucins	20	194
Tick specific proteins
Glycine rich superfamily	21	594
8.9 kDa protein family	19	4763
Evasins	20	5804
12 kDa family	2	4
Other secreted (many may be fragments of larger proteins with false signal P)	311	2480
Housekeeping proteins
Cytoskeletal proteins	31	446
Detoxification	16	501
Oxidant metabolism/Detoxification	11	123
Extracellular matrix	11	59
Immunity	8	60
Amino acid metabolism	15	188
Carbohydrate metabolism	20	175
Energy metabolism	75	2212
Intermediary metabolism	1	3
Lipid metabolism	29	238
Nucleotide metabolism	11	101
Nuclear regulation	30	446
Protein export	70	745
Protein modification	64	1058
Proteasome machinery	50	620
Protein synthesis machinery	145	5696
Signal transduction	89	911
Storage	4	225
Transcription factor	16	153
Transcription machinery	75	651
Transporters and channels	46	342
Unknown conserved	116	8206
Unknown conserved membrane protein	49	2203
Unknown product	265	1317
Transposable element	25	93
Rickettsial	1	2
Total	1875	64498

¹⁾Total number of coding sequences found for the indicated class.

²⁾Total number of illumina reads mapping to the total number of CDS of the indicated class.

Secreted (S) class of expressed genes

A total of 602 CDS attributed to secreted (S)-class genes expressed in the SGs of Am. tuberculatum were further characterized into several subgroups based on their function (Table 3 and Spreadsheet S1). It covers 32.1% of CDS and 58.5% of total reads. Among the secreted (S) class of expressed genes are metalloproteases, M13 peptidases, lipocalins, Kunitz domain containing proteins, TIL domain containing proteins, evasins and 8.9 kDa protein family, altogether covering for ~86% of the reads attributed to the S class.

Metalloproteases

Metalloproteases of the reprolysin superfamily have been previously described in tick sialomes. In I. scapularis, a recombinant enzyme of this family was shown to have fibrinolytic properties and to inhibit angiogenesis (Francischetti et al., 2003, 2005). The tick metalloproteases are distinct reprolysins that can be readily identified by the domain cd04272 from the CDD, annotated as “ZnMc_salivary_gland_MPs, Zinc-dependent metalloprotease, salivary gland_MPs. Metalloproteases secreted by the salivary glands of arthropods”. The A. tuberculatum sialotranscriptome reveals 16 CDS coding for members of this family, accruing 1.32% of the reads associated with the S class (Table 3). Phylogenetic analysis of near full-length A. tuberculatum sequences of this class with their best matches from the Acari database reveals 5 distinct clades with strong bootstrap support (named I-V on Supplementary figure 1), and one clade with less bootstrap support named A (Supplementary. figure 1). Clades I-IV and A are exclusively metastriate, while clade V has both prostriate and metastriate sequences, distinctly arranged in subclades. Five A. tuberculatum sequences are found in clades I, V and A. AtSigP-642 belongs to clade I, a clade characterized by metastriate sequences containing a carboxyterminal extension rich in Ser and Thr as visible within the red box in supplemental alignment S1. The Netoglyc server (Hansen et al., 1998) indicates 15 sites for N-galactosylation of the carboxyterminus of AtSigP-642, representing these metalloproteases to have a mucin tail, as previously pointed out before (Francischetti et al., 2009). AtSigP-168 and AtSigP-210 are closely related sequences (~10% amino acid difference) found in clade A and may represent alleles or more probably a recent gene duplication. AtSig-2297 is related to a R. pulchellus metalloprotease, with strong bootstrap support also in clade A. Finally, AtSigP-1811 is found in the metastriate subclade of clade V and may represent descent of the most primitive type of metalloprotease that gave rise to the prostriate expansion.

M13 metalloproteases/peptidases of the neprilysin family are usually membrane bound ecto enzymes associated with the hydrolysis of peptide hormones (Ouimet, 2004). Previous disclosure of tick sialomes (Francischetti et al., 2011; Karim et al., 2011) revealed highly expressed members of this family suggesting their role in their hosts, and indeed they lack indication of transmembrane domains or glyco inositol-phosphate anchors. The sialotranscriptome of A. tuberculatum reveals 12 CDS coding for M13 zinc peptidases, accreting 855 reads, or 2.27 % (Table 3) of all reads associated with secreted products, more than the classical salivary metalloproteases that have been characterized in I. scapularis and I. ricinus as fibrinolytic and anti-angiogenic enzymes (Francischetti et al., 2003; Kotsyfakis et al., 2015; Ribeiro et al., 2006; Schwarz et al., 2013); these are represented in the A. tuberculatum sialotranscriptome by 16 CDS assembled from 498 reads. The phylogenetic reconstruction of the M13 peptidases from this sialotranscriptome data (Supplementary figure 2) showed Clades I and II to contain distinct subclades of vertebrates and invertebrates, indicating the ancient origin of these two subfamilies. Clade III contains insect and worm sub clades, and clades IV and V are purely from ticks. Two of the A. tuberculatum sequences form a subclade of clade V, where two copies of M13 sequences are also found for A. maculatum and A. triste, signifying a gene duplication with possible gene conversion maintaining the paralogs within species-specific clades. This contrasts with clade IV sequences, where two or more gene products are represented for R. pulchellus, A. cajennense (sensu lato) and A. triste, but the subclades are multi-specific, indicating their equal evolutionary rate of change after the species diverged from the common ancestor.

Lipocalin Family

Proteins of the lipocalin family have been found in tick sialomes as product of dozens of genes in each species (Kim et al., 2016; Mans et al., 2008; Paesen et al., 1999; Paesen et al., 2000; Sangamnatdej et al., 2002; Tirloni et al., 2014). A few have been characterized functionally which collectively pointed their kratagonistic role (from the Greek kratos = seize) in sequestering host agonists such as serotonin, histamine, leukotrienes and thromboxanes (Andersen and Ribeiro, 2017; Ribeiro and Arca, 2009). The sialotranscriptome of A. tuberculatum identified 76 CDS coding for members of this family, representing 45% of the reads attributed to the secreted class (Table 3). Alignment of 26 A. tuberculatum sequences comprised of more than 150 amino acids with 549 best matches of the Acari database (Supplemental alignment S2) reveals 8 blocks of sequence similarities that are separated by insertions (marked with red lines on the top of the alignment). These insertions are usually flanked by glycine or proline residues (shown within the red boxes in the alignment). The first represents the signal peptide, distinguishable by the turquoise color of the leucine and other hydrophobic amino acids. Because lipocalins contain eight-stranded antiparallel-barrel arranged beta strands (Schlehuber and Skerra, 2005), the 7 blocks of conservation after the signal peptide represent these helices, perhaps the larger block (number 5) containing 2 helices, and the insertions represent loops between the helices. The phylogenetic reconstruction is complex and shallow, producing twelve large clades (marked A-L in the supplemental figure), most having under 50% bootstrap support (Supplementary. figure. 3). This may derive from the fast evolution of this group of lipocalin-coding genes that makes phylogenetic reconstruction difficult. However, sixteen clades with strong bootstrap support are shown, named with Roman numerals. Most clades are metastriate specific and group closely related species, thus within a shorter evolutionary time. The few clades with prostriate sequences are marked “Prot” on the phylogenetic tree. The alignment and phylogeny indicate the complexity of the tick lipocalin superfamily, but at the same time points to robust clades of sequences that could indicate a common function, helping to choose sequences for functional characterization of these proteins.

Kunitz domain containing proteins

These protease inhibitors are the second largest family among secreted salivary proteins. In ticks, Kunitz inhibitors possess one to seven Kunitz domains and most of them have anti-coagulant property which inhibit various proteases of the coagulation cascade (Assumpcao et al., 2016; Chmelar et al., 2012; Corral-Rodríguez et al., 2009; Francischetti et al., 2004; Francischetti et al., 2002; Mans et al., 2002; Nazareth et al., 2006; Paesen et al., 2007; Valdes et al., 2013). Apart from protease inhibition, Kunitz domains also inhibit ion channel (Castaneda and Harvey, 2009; Harvey, 1997; Kwong et al., 1995; Lucchesi and Moczydlowski, 1991). A study on Rhipicephalus appendiculatus tick has suggested a vasodilator function of these proteins by activating maxiK channels in an in vitro system (Paesen et al., 2009). Other members of this family also bind to integrin and demonstrate anti-platelet activity (Chmelar et al., 2019; Mans et al., 2002).

These properties make Kunitz-Domain Protease Inhibitors significantly important target molecules in tick research. A Kunitz-domain protein Ir-CPI (Ixodes ricinus contact phase inhibitor), a very efficient inhibitor of the coagulation cascade (Decrem et al., 2009) is being used for pre-clinical testing. Some of the Kunitz inhibitors such as Ixolaris and Amblyomin-X have also demonstrated anti-cancer therapeutic potential (Barboza et al., 2015; Carneiro-Lobo et al., 2009; Carneiro-Lobo et al., 2012; Chudzinski-Tavassi et al., 2016; de Oliveira Ada et al., 2012; de Souza et al., 2016). Highest potential of these inhibitors lies in their anti-hemostatic properties which could be exploited in drug development for several human diseases. In this study, these proteins accrued 8.98% of the total reads i.e. 3388, and numbered 42 CDS (Table 3).

Despite the diversity of the Kunitz domain containing peptides in ticks, the phylogenetic reconstruction of A. tuberculatum protein sequences (Supplementary figure 4) with their best matching sequences from the NCBI Acari database indicates 9 robust clades of sequences (100% bootstrap support) containing at least one A. tuberculatum sequence and one or more sequences from other tick species, all from metastriate ticks except clade IX were sequences from I. ricinus and I. scapularis are also found, and may represent uniquely conserved sequences, probably of a housekeeping function. Clade VII is interesting as it has three clear sub-clades, (A) containing A. maculatum, A. triste, A. americanum and A. tuberculatum sequences, (B) containing only A. maculatum and A. triste sequences, and (C) containing R. pulchellus, A. americanum, A. cajennense (sensu lato), A. parvum and H. marginatum sequences. The only species appearing in both sub clades A and C is A. americanum, the sole sequence of subclade A being the most basal in the clade VII, whereas the sequences from A. maculatum and A. triste are found in the sister clades indicating evolutionarily closer relationship than those of A. americanum and A. tuberculatum. (Supplementary figure 4)

8.9 kDa protein family

This is a protein family of secreted peptides of molecular weight near 10 kDa but can also have proteins of double size containing two domains. This protein family is exclusive to hard ticks. An earlier work (Francischetti et al., 2009) has described the identification of sixty members of this protein family from prostriate and metastriate ticks. The A. tuberculatum sialotranscriptome reveals 19 CDS coding for members of this family, accruing 12.63% of the reads i.e. 4,763 reads associated with the secreted class. In a previous study, phylogenetic analysis of these proteins has shown robust clades with metastriate and prostriate sequences indicative of multiple round of gene duplication and faster evolution of these proteins (Karim et al. 2011).

Evasins

Evasins are considered as potential target molecules in tick research because they inhibit inflammatory responses (Charo and Ransohoff, 2006) associated with some chemokine classes. Evasins are a family of chemokine-binding proteins which were first identified in Rhipicephalus sanguineus (Deruaz et al., 2008; Frauenschuh et al., 2007) and more recently biochemically characterized in other tick species (Hayward et al., 2017). Evasin homologs are present in Amblyomma and Ixodes species, but homology in Ixodes is weaker (Frauenschuh et al., 2007). Evasin-1 and evasin-4 bind specifically to CC chemokines, and evasin-3 binds to CXC chemokines (Deruaz et al., 2008; Frauenschuh et al., 2007). Evasin-1 has therapeutic potential due to its ability to inhibit T-cell, neutrophil, macrophage migration and the production of inflammatory cytokines (Chemlar et al. 2019). The A. tuberculatum sialotranscriptome reveals 20 CDS coding for members of this family, accruing 15.39% of the reads i.e. 5,804 reads associated with the secreted class. In a previous study, phylogenetic analysis indicated the diversity of the Evasin family, corroborated in this study by the finding of at least 18 different genes coding for evasins within A. maculatum (Karim et al., 2011).

Conclusion

In this study, the sialotranscriptome of adult female A. tuberculatum collected directly from gopher tortoises, we report an annotated catalogue of expressed transcripts from over ~138,000 pyrosequencing reads assembled into 29,991 contigs. Putatively secreted (S) category and housekeeping (H) category both together cover for 98% of the mapped reads, and 84% of the coding sequences (CDS). 41% of the reads from the putative housekeeping class were unknown conserved. Many orphan proteins have been found in this study which are secretory in nature, indicative of Amblyomma-specific proteins. Information related to the mechanism of tick feeding on ectotherm gopher tortoises is limited. Comparisons of the sequences of the different proteins families with those from ticks feeding on mammalian hosts may shed light on the evolution of blood feeding by ticks. This study will help in understanding the complex biology of tick hematophagy. On the other hand, knowledge on the physiology of hemostasis, inflammation and immunity in tortoises is virtually unknown when compared to the same subjects in mammals. The repertoire of salivary proteins from A. tuberculatum may result in the development of tools to investigate tortoise hemostasis and inflammation, as happened by the use of a bird-feeding vampire bat salivary protein which helped to determine the plasminogen activation pathway in birds (Cartwright and Hawkey, 1969).