Integrated analysis of sialotranscriptome and sialoproteome of the brown dog tick Rhipicephalus sanguineus (s.l.): Insights into gene expression during blood feeding

Abstract

Tick salivary glands secrete a complex saliva into their hosts which modulates vertebrate hemostasis, immunity and tissue repair mechanisms. Transcriptomic studies revealed a large number of transcripts coding for structural and secreted protein products in a single tick species. These transcripts are organized in several large families according to their products. Not all transcripts are expressed at the same time, transcription profile switches at intervals, characterizing the phenomenon of “sialome switching”. In this work, using transcriptomic and proteomic analysis we explored the sialome of Rhipicephalus sanguineus (s.l.) adult female ticks feeding on a rabbit. The correlations between transcriptional and translational results in the different groups were evaluated, confirming the “sialome switching” and validating the idea that the expression switch may serve as a mechanism of escape from the host immunity. Recombination breakpoints were identified in lipocalin and metalloprotease families, indicating this mechanism could be a possible source of diversity in the tick sialome. Another remarkable observation was the identification of host-derived proteins as a component of tick salivary gland content. These results and disclosed sequences contribute to our understanding of tick feeding biology, to the development of novel anti-tick methods, and to the discovery of novel pharmacologically active products.

1. Introduction

Ticks are exclusive blood sucking mites belonging to three different families, the Argasidae, or soft ticks, which feeds for minutes to hours in their vertebrate hosts, the Ixodidae, or hard ticks, which feeds for several days and the Nuttalliellidae family comprising of a single extant species [1].

The salivary glands of ticks are important in several ways for their survival, by helping the tick to collect water from humid air (when the tick spreads a hygroscopic saliva over its palps, re-ingesting it later with the absorbed water), during blood feeding (by antagonizing host hemostasis and modulating host immunity and tissue repair reactions), and to maintain ion and water metabolism during blood feeding (when most of the ingested water contained in the blood meal is reinjected in the host as saliva), and to assist tick reproduction (when saliva is spread over the sperm collected by male ticks before insertion into the female vaginal pore, a function similar to that performed by invertebrate accessory glands or vertebrate prostate glands) [2–5].

The feeding process of hard ticks is classically divided in three phases, the attachment phase, usually lasting one day, the slow feeding phase lasting several days, and the fast feeding, or “big sip”, usually lasting one day [4]. The name “slow feeding”, however, is a misnomer, since most of the blood removed from the host occurs during this phase but the majority of the meal protein is excreted in the feces and the blood water and ions is reinjected into the host as saliva [6].

Recent transcriptomic studies of the salivary glands of hard ticks have revealed a surprisingly large number of transcripts coding for many different families of proteins, totaling over one thousand different salivary secreted polypeptides per tick species [5,7–27]. Their coding genes appear to be evolving at a fast pace of evolution, due to relaxed and/or positive selection [28,29]. Interestingly, these transcripts are not all expressed at once, but rather at different stages of feeding [8,22,26,30], or when feeding on different hosts [31]. This process of sialome switching has been interpreted as a mechanism of immune evasion [22,26,30]. Proteomic studies of the tick salivary glands at different stages of feeding and when ticks were exposed to different hosts have also been performed [32–36], indicating in some cases a disagreement between the proteomic and transcriptomic findings [35]. These pioneer studies, however, were done with few developmental stages of feeding, and in most cases had only one biological replicate.

Rhipicephalus sanguineus sensu lato (s.l.) is a cosmopolitan tick species parasitizing domestic animals such as dogs and cats, and, occasionally, humans [37,38]. R. sanguineus can act as a vector of Rickettsia rickettsii to humans [39] and Babesia spp. to dogs and cats [40,41]. A limited Sanger-based sialotranscriptome has been previously performed [21], using salivary glands from ticks feeding for 3–5 days, and 5 days, leading to the description of 1024 transcripts, 26% of which were considered as coding for salivary secreted proteins. Moreover, a transcriptome of larvae was described more recently focusing in genes involved in acaricide resistance [42]. Using Illumina technology, we herein analyze the sialotranscriptome of R. sanguineus of adult female ticks at six different stages of feeding, and with three biological replicates. Uniquely, instead of grouping the different feeding stages by days of feeding, we grouped the ticks by their weight, as it is a better indicator of the physiological status of the tick feeding process [43]. Using the coding sequences derived from the transcriptome assembly to form a database for proteomic studies, we pursued a proteome analysis of the salivary glands, which confirmed the sialome switching at the proteomic level, and with a high degree of agreement between transcriptome and proteome expression levels.

2. Material and methods

2.1. Ethics statement

Animals used in the experiments were housed at Faculdade de Veterinária, Universidade Federal do Rio Grande do Sul (UFRGS) facilities. This study was conducted according to the ethical and methodological norms prescribed by the International and National Directives and Norms by the Animal Experimentation Ethics Committee of UFRGS. Protocol (number 27559) was approved by the Comissão de Ética no Uso de Animais – CEUA – UFRGS.

2.1.1. Ticks, tick feeding and salivary gland dissection

R. sanguineus (s.l.) strain used in this study belongs to the tropical lineage [44] and were collected in Rio de Janeiro, Brazil [45], and kept at the UFRGS. Unfed ticks were maintained at 28 °C and 85% relative humidity before infestation on rabbits. Adults ticks used for salivary gland extraction were restricted to feed onto the outer part of the ear of four naïve female New Zealand rabbits with orthopedic stockinet’s glued. A total of 25 adult females and 10 males (35 ticks per ear, 70 ticks per animal) were placed into the tick containment apparatus and allowed to attach. To compose the groups of ticks by a blood feeding index, partially fed ticks were collected randomly from different hosts during feeding, selected by their engorgement size, sorted by weight, and collected into three independent replicates, including: group unfed (UF) (n = 5 females per replicate), group G1 (n = 5 females per replicate, average 1.8 mg, collected at day 2 of feeding), group G2 (n = 5 females per replicate, average 3.6 mg, collected at day 6 of feeding), group G3 (n = 4 females per replicate, average 7.0 mg, collected at day 6 of feeding), group G4 (n = 2 females per replicate, average 10.9 mg, collected at day 8 of feeding), group G5 (n = 3 females per replicate, average 24 mg, collected at days 8 and 11 of feeding), group G6 (n = 2 females per group, average 36 mg. collected at days 6, 10 and 13 of feeding). Supplemental fig. 1 displays a sample of the ticks collected for each group. Ticks from this experiment were used exclusively to dissect salivary glands for extraction of total RNA used for RNAseq analysis.

A second independent infestation was performed to dissect salivary glands to obtain protein for LC-MS/MS analysis. For this second infestation, two naïve female New Zealand rabbits were used. A total of 20 adult females and 10 males (30 ticks per ear, 60 ticks per animal) were placed into the tick containment apparatus and allowed to attach. Partially fed ticks were collected from host during feeding, selected by their engorgement size, sorted by weight, and divided: group UF (n = 10 females), group G1 (n = 13 females, average 1.8 mg), group G2 (n = 5 females, average 3.6 mg), group G3 (n = 5 females, average 7.0 mg), group G4 (n = 4 females, average 10.9 mg), group G5 (n = 5 females, average 24 mg), group G6 (n = 4 females, average 36 mg). The measurements of the scutum length and tick length were made using the ImageJ software [46].

After removal from the host, ticks were rinsed with ethanol 70% following a second rinsing with nuclease-free water. Ticks were dissected within two hours after removal from the host. Tick salivary glands (SGs) were dissected in a fresh ice-cold PBS, pH 7.4. After dissection, salivary glands were washed gently in a fresh ice-cold PBS. After washing, dissected SGs were stored immediately in RNAlater (Invitrogen, Carlsbad, CA, USA) prior to extracting total RNA and protein.

2.1.2. RNA and protein extractions

After removal from RNAlater, SGs were gently washed in nuclease- free phosphate-buffered saline pH 7.4 (Thermo Fisher Scientific), and immediately transferred to TRIzol® reagent (Thermo Fisher Scientific). Total RNA and protein were isolated according to manufacturer’s specifications. Protein pellet was resuspended into Tris-HCl 100 mM, urea 8 M, pH 8.0. Protein concentration was measured using BCA Protein Assay Reagent Kit (Thermo Scientific Pierce), following the manufacturer’s recommendations and samples were stored at −80 °C upon usage.

2.1.3. Library preparation and sequencing

Tissue samples were submitted to the North Carolina State Genomic Sciences Laboratory (Raleigh, NC, USA) for Illumina RNA library construction and sequencing, as detailed in [47], except that single ended libraries were sequenced. Briefly, RNA samples were analyzed with an Agilent 2100 Bioanalyzer with an RNA 6000 Nano Chip (Agilent Technologies, USA). mRNA purification used the NEBNExt Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, USA). Libraries were constructed using the NEBNext Ultra Directional RNA Library Prep Kit (NEB) and NEBNext Mulitplex Oligos for Illumina (NEB). The libraries were sequenced in an Illumina HiSeq 2500 DNA sequencer, utilizing 125bp single end sequencing flow cell with a HiSeq Reagent Kit v4 (Illumina, USA).

2.1.4. Bioinformatic analysis

Bioinformatic analyses were conducted following the methods described previously [47], with some modifications. Briefly, the fastq files were trimmed of low quality reads (< 20), removed from contaminating primer sequences and concatenated for single-ended assembly using the Abyss (using k parameters from 21 to 91 in 5 fold increments) [48] and Trinity [49] assemblers. The combined fasta files were further assembled using a iterative blast and CAP3 pipeline as previously described [23]. CDS were extracted based on the existence of a signal peptide in the longer open reading frame (ORF) and by similarities to other proteins found in the Refseq invertebrate database from the National Center for Biotechnology Information (NCBI), proteins from Acari deposited at NCBI’s Genbank and from SwissProt.

Reads for each library were mapped on the deducted CDS using the RSEM software [50]. Heat maps were made using the software gplots [51] and statistical tests used the package edgeR [52], both running under the R environment [53].

Protein alignments were done using ClustalX [54], and phylogenies were inferred using the Mega v.7 package [55]. The Maximum Likelihood method based on the best nucleotide substitution matrix available for the alignment was used to infer the evolutionary history, as discovered by the Mega package. Recombination breakpoints in aligned coding sequences were determined by the RDP4 software [56].

2.1.5. LC-MS/MS analysis and data analysis

Approximately 6 μg of protein from each sample was adjusted to a final volume of 30 μL with 50 mM HEPES, pH 8.5 in 8 M urea. The samples were reduced in 5 mM DTT for 1 h at room temperature followed by alkylation with 15 mM iodoacetamide for 20 min. The concentration of urea was reduced to 1.5 M by the addition of 50 mM HEPES, pH 8.5 and 0.5 μg of LysC protease was added and incubated for 15 h at 30 °C. The urea concentration was reduced to 0.8 M with 100 mM HEPES, pH 8.0, 1 μg of trypsin was added and incubated for 6 h at 30 °C. The pH was adjusted to 2.5 with trifluoroacetic acid (TFA) and the samples were desalted on an Oasis HLB micro-elution plate. The peptides were eluted with 0.1% TFA, 50% acetonitrile (AcCN) and the solvent was removed under vacuum at 50 °C. The residue was dissolved in 0.1% formic acid, 3% AcCN for injection.

Data were collected using an Orbitrap Fusion Lumos mass spectrometer equipped with an EASY-Spray Ion Source and an EASY-nLC 1200 liquid chromatography system (Thermo Fisher Scientific). The mobile phase solvent contains water and 0.1% formic acid. The peptides (5 μL) were loaded onto trap column (PepMap 100 C18, particle size 3 μm, length 2 cm, inner diameter 75 μm, Thermo Fisher Scientific), and separated on analytical column (PepMap 100 C18, particle size 2 μm, length 25 cm, inner diameter 75 μm, Thermo Fisher Scientific) with a linear gradient of 0–40% acetonitrile for 80 min, followed by 40–80% for 5 min, holding at 80% for 5 min, 80–0% for 5 min, and holding at 0% for 5 min. Throughout this 100-min data acquisition, the flow rate was set at 300 nL/min and the analytical column temperature was set at 50 °C. The data acquisition was done with the standard data-dependent acquisition strategy, where the survey MS1 scan was done every 2 s with Orbitrap mass analyzer at 120,000 resolution and the data-dependent MS2 scans were done with Linear Ion Trap mass analyzer for multiply charged precursor ions isolated with the 1.6 m/z window using Quadrupole and fragmented by CID at 35% collision energy. The dynamic exclusion period was set at 15 s, and the EASY-IC internal calibration was utilized for Orbitrap scans.

Tandem mass spectra were extracted from Thermo RAW files using RawExtract 1.9.9.2 [57] and searched with ProLuCID [58] against R. sanguineus database (71,643 entries) concatenated with Oryctolagus cuniculus from Uniprot [59] reference database (21,176 entries) and reverse sequences of all entries. The search space included all fully- tryptic and half-tryptic peptide candidates. Carbamidomethylation of cysteine was used as static modification. Data was searched with 50 ppm precursor ion tolerance and 0.4 Da fragment ion tolerance. The validity of the peptide spectrum matches (PSMs) generated by ProLuCID was assessed using Search Engine Processor (SEPro) module from PatternLab for Proteomics platform [60]. A cutoff score was established to accept a protein false discovery rate (FDR) of 1% based on the number of decoys. Results were post processed to only accept PSMs with < 10 ppm precursor mass error and proteins with a unique peptide. Normalized spectral abundance factors (NSAF) was used to represent relative abundance and secretion dynamics. Values were normalized calculating Z-score and values were used to generate heat maps using the heatmap2 function from the ggplot2 library in R.

2.1.6. Data availability

The transcriptome data was deposited to the National Institute for Biotechnology Information (NCBI) under Bioproject PRJNA606595 and Biosample accessions SAMN14115946, SAMN14115947, SAMN14115948, SAMN14115949, SAMN14115950, SAMN14115951, SAMN14115952, SAMN14115953, SAMN14115954, SAMN14115955, SAMN14115956, SAMN14115957, SAMN14115958, SAMN14115959, SAMN14115960, SAMN14115961, SAMN14115962, SAMN14115963, SAMN14115964 and SAMN14115965. The reads were deposited to the Short Reads Archive of the NCBI under accessions SRR11109985, SRR11109984, SRR11109973, SRR11109972, SRR11109971, SRR11109970, SRR11109969, SRR11109968, SRR11109967, SRR11109966, SRR11109983, SRR11109982, SRR11109981, SRR11109980, SRR11109979, SRR11109978, SRR11109977, SRR11109976, SRR11109975 and SRR11109974. This Transcriptome Shotgun Assembly project has been deposited at DDBJ/EMBL/GenBank under the accession GINV00000000. The version described in this paper is the first version, GINV01000000.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD018964.

3. Results

3.1. Overall description of the transcriptomes

Following primer removal and trimming of low-quality bases, we obtained over 687 million reads from 20 libraries, including two biological replicates for the unfed group (UF), and three replicates for each of groups 1–6. A minimum of 17,854,945 reads was found for the third replicate of Group 4, and a maximum of 144,760,365 reads was found for replicate 2 of Group 6 (Supplemental table 1). As indicated in the methods section, the groups were organized according to the tick weight following attachment to a rabbit, not to days of feeding as it is usual.

Following assembly of the reads, 71,643 coding sequences (CDS) were extracted. The RSEM software was used to map the reads from each library to these transcripts, 28,921 of which had a TPM (transcripts per million) of 10 or larger in at least one library. Notice that a TPM value of 1000 indicates the transcript to be represented by 1000 in one million total transcripts, or one in one thousand transcripts, which is equal to 0.1%. A TPM value of 10 then indicates the transcript expression to be 0.001% of the totality of transcripts. These transcripts and their comparisons to several databases are available as a hyperlinked spreadsheet (Supplemental spreadsheet 1).

The total number of reads mapped to these CDS totalled 357,388,554 reads, or 51.98% of the 687,502,190 reads. The unmapping of nearly half of the reads could be due to reads mapping to the 5′ or 3’ UTR of the CDS, which were not extracted, and also due to many reads deriving from non-coding RNA’s. These results were similar to those obtained following analysis of the sialotranscriptome of Rhipicephalus zambeziensis [19].

Functional classification of the 28,921 transcripts with TPM values larger than 10 showed that the two major functional groups are represented by CDS of the unknown class and those representing CDS coding for proteins that are probably secreted (Table 1). These two classed accrued nearly 50% of all mapped reads, as follows: The secreted class accounted for 4039 CDS, or 14% of the total CDS, accruing over 80 million reads, or 22.5% of the total reads. The unknown class accounted for 20,743 CDS, or 71% of all 28,921 CDS, and these accrued over 99 million reads, or 28% of all mapped reads. Twenty-four additional classes are shown in Table 1, which also includes 238 CDS probably deriving from transposable elements, and 16 CDS that are of probable viral origin.

Table 1.

Functional classification of coding sequences (CDS) originating from the Rhipicephalus sanguineus sialotranscriptome. Classes were sorted according to their accreted number of reads, shown in background color from red (maximum value) to blue (minimum value). Yellow background represents average values.

t1

Within the secreted class of transcripts, we found 42 families that have 4 or more members, including only those near full length in size. These 915 transcripts accrued ~60 million reads (Table 2). The top five families were identified as lipocalins, with 245 members and accruing 18.7% of the reads from this secreted group, Kunitz family of protease inhibitors, with 82 members and accruing 14.1% of the reads, glycine- rich protein family, with 67 members and accruing 38% of the reads, metalloprotease family, with 61 members and accruing 2.5% of the reads, and 8.9 kDa family, with 47 members and accruing 2.3% of the reads.

Table 2.

Families of secreted proteins within the sialotranscriptome of Rhipicephalus sanguineus. Protein families with 4 or more members are shown, ordered from the most abundant family and including the average number of reads accrued per each member, and the totality of reads accrued by the family.

Class	Average Number of Reads per CDS	SE	N	Total number of reads	Percentage of this group
Lipocalin	45,877	9530	245	11,239,766	18.71
Kunitz	103,602	45,692	82	8,495,355	14.14
GRP	341,156	106,131	67	22,857,468	38.04
Metalloprotease	24,959	3980	61	1522,472	2.53
8.9 kDa	29,749	17,901	47	1,398,198	2.33
BTSP	33,885	9372	44	1,490,951	2.48
AlaRich	7985	1193	34	271,500	0.45
DAP-36	35,187	16,422	23	809,295	1.35
Evasin	27,745	16,615	21	582,646	0.97
M13_peptidase	32,070	9163	18	577,258	0.96
28 kDa	31,532	6607	17	536,052	0.89
Defensin	7207	3246	17	122,527	0.20
Ixodegrin	13,508	4518	16	216,131	0.36
Mucin	95,544	36,085	15	1,433,163	2.39
OneOfEach	27,022	11,343	15	405,329	0.67
TIL	25,574	9702	15	383,612	0.64
Cystatin	3603	1746	13	46,840	0.08
Carboxypeptidase_inhibitor	2713	981	13	35,271	0.06
Antigen-5	17,443	3562	12	209,317	0.35
Serine carboxypeptidase	8541	3030	12	102,490	0.17
Cytotoxin	12,159	1916	10	121,595	0.20
23–24 kDa	7152	1693	10	71,518	0.12
Serpin	14,320	3982	8	114,562	0.19
Rhiapp specific protein	639	220	8	5110	0.01
Down syndrome cell adhesion molecule	58,904	30,921	7	412,326	0.69
Toll-like	34,472	12,061	6	206,833	0.34
Ficolin/Ixoderin	32,515	12,359	6	195,091	0.32
Transposon	26,131	13,231	6	156,784	0.26
Superoxide dismutase	19,797	6965	6	118,780	0.20
Microplusin	11,645	5430	6	69,872	0.12
Rapp-40-287	1522	462	6	9135	0.02
SPARC/Kazal	77,755	35,892	5	388,774	0.65
Endonuclease	38,671	24,806	5	193,354	0.32
Mys-30-94	20,766	12,113	5	103,829	0.17
Thyropin	19,202	7985	5	96,009	0.16
Mys-30-60	1593	377	5	7965	0.01
35–224 Madanin	1,208,799	542,032	4	4,835,197	8.05
Serine protease	24,451	8522	4	97,804	0.16
Insulin_growth_factor	18,747	9464	4	74,988	0.12
IPPase	13,388	5220	4	53,552	0.09
Serum amyloid A	2560	166	4	10,242	0.02
12 kDa	641	178	4	2563	0.00
Total			915	60,081,522	100

SE = standard error of the number of reads per coding sequence (CDS); N = number of sequences.

3.2. An insight into the structure of the families coding for secreted salivary proteins in R. sanguineus

3.2.1. Lipocalins

The lipocalin family in ticks have been associated with a kratagonist function towards biogenic amines [61,62] and eicosanoids [63,64], as well as having toxin properties in soft ticks [65]. This family has the highest number of transcripts identified in tick sialotranscriptomes [30,66,67]. Of the 245 transcripts attributed to the lipocalin family found in this sialotranscriptome, 55 are full length or near full-length members of the subfamily characterized by having the PFAM domain “pfam02098, His_binding, Tick histamine binding protein”. Phylogenetic analysis of these 55 coding sequences showed remarkable low bootstrap support for the branches containing most of the transcripts. Indeed, only 15 of the 55 sequences were within a clade with three or more members, totalling five clades (Supplemental fig. 2).

3.2.2. Metalloproteases

Transcripts coding for salivary zinc-dependent metalloproteases are abundantly found in tick sialotranscriptomes [30,66,67], and their function was associated with a fibrinolytic and anti-angiogenic activities [68–71]. The sialotranscriptome of R. sanguineus here described allowed for the identification of 46 full length or near full length coding for metalloproteases. The phylogenetic analysis produces a tree that is deeper than that observed for the lipocalins. Forty of the 46 sequences are organized within seven clades containing three or more sequences (Supplemental fig. 3).

3.2.3. Cystatins

Tick salivary cystatins were first discovered in Ixodes scapularis and found to affect immunity and inflammation by counteracting cysteine proteases found in leukocytes [72–78]. They are normally expressed in much lower levels than lipocalins or metalloproteases, and the family is not as numerous. The phylogram of 13 cystatin coding sequences indicated four divergent clades (Supplemental fig. 4).

3.3. An insight on the sialotranscriptome switch of R. sanguineus

The edgeR package was used to determine the differential expressed CDS in paired comparisons (UF × SG1, SG1 × SG2, SG2 × SG3, SG4 × SG5 and SG5 × SG6). The multi-dimensional scaling plot shows good clustering of the biological replicates, indicating that the consolidation of samples by weight was a good choice, instead of days of feeding (Fig. 1). Out of 28,921 CDS having a TPM > 10 in at least one library, 17,590 were found differentially expressed (DE) (Fig. 2), including 1945 DE CDS found in paired comparisons at a FC larger than 128 (Table 3). When comparing the transcript class frequencies with FC > 4 with those with FC > 128 by a Χ² test (adjusted for multiple testing with the Bonferroni correction), it is evident that the secreted class was represented in the FC > 128 at a higher frequency (28%) than the frequency found for the FC > 4 group (19%) (false discovery rate (FDR) = 2.07e⁻¹¹). Other statistically significant changes occurred in the groups: protein export machinery, energy metabolism, protein modification machinery, proteasome machinery, transporters/storage, carbohydrate metabolism and lipid metabolism, all significantly decreasing their frequency in the FC > 128 group when compared to the FC > 4 group (Tables 3 and 4).

Fig. 1.

Principal component analysis of the transcriptome of Rhipicephalus sanguineus when the different libraries (determined by Groups) were contrasted using the edgeR program. Group unfed female (UF), partially fed ticks groups G1 (1.8 mg), G2 (3.6 mg), G3 (7.0 mg), G4 (10.9 mg), G5 (24 mg) and G6 (36 mg).

Fig. 2.

Heat plot of differentially expressed transcripts from the sialotranscriptome of Rhipicephalus sanguineus obtained from unfed ticks (UF) or at 6 partially fed ticks with different weights (G1 - G6). Two biological replicates are shown for the unfed (UF) group, and three for each of the six feeding groups.

Table 3.

Functional classification of the Rhipicephalus sanguineus transcripts for all coding sequences and for the differentially expressed CDS, including all those that are significantly differentially expressed, and within these, those achieving more than 4× fold change (FC), more than 16 × FC, etc. up to 128 × FC.

Class	Number of transcripts
	All	All Significant	4× FC	8× FC	16× FC	32× FC	64× FC	128× FC
Unknown	20,743	11,580	4734	2922	2146	1759	1482	1247
Secreted	4039	2519	1371	1108	940	794	680	554
Unknown, conserved	439	361	100	43	23	15	13	11
Signal transduction	394	326	86	42	27	20	14	8
Transcription machinery	421	350	71	24	11	7	6	6
Protein synthesis machinery	381	340	66	20	5	1	1	1
Extracellular matrix/cell adhesion	192	170	105	87	67	50	38	33
Transposable element	238	172	79	61	44	37	26	19
Protein export machinery	287	250	68	19	7	3	2	2
Metabolism, energy	264	227	58	17	11	5	3	3
Protein modification machinery	199	170	68	39	26	14	5	3
Proteasome machinery	188	150	42	11	4	3	3	3
Transporters/storage	142	129	44	26	14	13	10	8
Immunity	95	84	50	38	32	29	23	20
Metabolism, carbohydrate	124	107	47	22	13	6	3	3
Metabolism, lipid	129	114	48	19	8	2	2	2
Oxidant metabolism/detoxification	97	86	42	23	21	16	13	8
Nuclear regulation	130	105	24	8	3	3	2	2
Cytoskeletal	130	102	21	8	4	4	3	3
Metabolism, amino acid	86	77	22	11	9	6	5	2
Metabolism, nucleotide	68	57	20	12	8	7	6	4
Transcription factor	70	54	7	2	2	2	2	2
Metabolism, intermediate	27	25	9	2	2	1	0	0
Viral	16	16	10	8	5	3	1	1
Nuclear export	15	14	3	0	0	0	0	0
Storage	7	5	4	4	3	3	1	0
Total	28,921	17,590	7199	4576	3435	2803	2344	1945

Table 4.

Functional classification of the sialotranscriptome of Rhipicephalus sanguineus represented as percentage of transcripts according to their ascribed class. The columns represent results for All transcripts, for all differentially expressed transcripts predicted by edgeR, and those having a fold change (FC) higher than 4×, 8×,..128× . The last column shows the false discovery rate (FDR) resulting from a X² test contrasting the expressions 4× and 128× for each functional class. The color gradients shows background colors from red (maximum value) to blue (minimum value). Yellow background represents average values.

t2

To verify whether the DE transcripts were organized in clusters associated with the different library groups, the normalized TPM values of 3435 transcripts having a significant FC > 16 were submitted to the CLICK algorithm of the expander program, which identified 10 clusters of transcripts (Fig. 3). Clusters that were unique to the UF, SG1, SG2, SG3 and SG5 are seen in Fig. 3 A, B, D, F, H. Other clusters contained transcripts of 2 or more groups. The results of the paired contrasts and Expander clusterization (Tables 3 and 4) indicated the occurrence of sialome switch within R. sanguineus, with remarkable values of differential expression, exceeding in many cases over 100-fold.

Fig. 3.

Clusters of 16 x differentially expressed transcripts from the sialotranscriptome of Rhipicephalus sanguineus according to their library groups, as determined by the click algorithm of the program Expander.

3.3.1. An insight into the sialome switch of the lipocalins of R. sanguineus

The assembled transcriptome of R. sanguineus identified 245 transcripts coding for lipocalins, 216 of which (88%) were found significantly DE in at least one paired comparison. To visualise the expression patterns of these DE transcripts, we made graphs of their normalized average TPM values found within each of the 7 groups of ticks. Supplemental fig. 5 displays the graphed results for 53 lipocalins that peak their expressions within tick group 1, and Supplemental fig. 6 displays the results for 30 lipocalin coding transcripts that peak within tick group 2. Notice that while some lipocalin transcripts could be found expressed in two or more contiguous groups, some were narrowly expressed in a single group. This narrow distribution was mostly seen in the lipocalins peaking within group 1, which shows the longer distance from its neighbours in the MDS plot (Fig. 1). Finally, Supplemental fig. 7 displays 77 lipocalins with DE FC > 16 that were found in expander clusters 3–9.

3.3.2. An insight into the sialome switch of the Kunitz-coding transcripts of R. sanguineus

The assembled transcriptome of R. sanguineus identified 82 transcripts coding for proteins containing Kunitz domains, 74 of which (90%) were found significantly DE in at least one paired comparison. Of these 74, 37 are at least DE with a FC > 16. Supplemental figs. 8 and 9 display the plots of the normalized average TPM values for these transcripts according to their tick groups.

3.3.3. An insight into the sialome switch of the cystatin-coding transcripts of R. sanguineus

There were 13 transcripts coding for cystatins in the assembled sialotranscriptome, all of which were DE in at least one paired comparison. The graph of the DE transcripts coding for cystatins is shown in Supplemental fig. 10.

3.3.4. Transcripts possibly involved with sialome switching in R. sanguineus

The mechanism of sialome switching in ticks remains a mystery. How are the genes of the same family turned on and off? It has been suggested that either epigenetic mechanisms or more classical signal-transduction/transcription factors could be involved in controlling the sialome switch [26]. Table 5 displays the transcripts found in the sialotranscriptome of R. sanguineus that are associated with epigenetic regulation (DNA methylation and histone modification machinery) [79]. Table 6 displays transcripts associated with transcription factors that might be associated with the regulation of gene expression in tick salivary glands.

Table 5.

Transcripts possibly associated with epigenetic gene expression within the salivary glands of Rhipicephalus sanguineus.

Transcript name	Annotation
Rs-116658_FR3_1-2041	Chromatin remodeling complex WSTF-ISWI small subunit
Rs-182989_FR3_48-542	DNA methyltransferase 1-associated protein-1
Rs-119673_FR3_78-274	heterochromatin-associated protein hp1
Rs-138721_FR3_6-538	Histone deacetylase complex catalytic component RPD3
Rs-89535_FR2_92-653	histone-lysine N-methyltransferase PRDM9-like
Rs-88148_FR3_105-361	Methyl-CpG binding transcription regulators
Rs-133231_FR2_1-702	NAD-dependent histone deacetylases and class I sirtuins (SIR2 family)
93106_FR2_23-447	Phosphatidylserine-specific receptor PtdSerR contains JmjC domain
91046_FR1_1-320	Sirtuin 4
Rs-76172_FR2_23-420	Sirtuin 5

Table 6.

Transcription factor-coding transcripts found in the salivary glands of Rhipicephalus sanguineus.

Transcript name	Annotation
150871_FR1_1-632	Transcription factor CA150
167428_FR2_41-174	Transcription factor PBX and related HOX domain proteins
88577_FR3_1-153	Phosphatidylserine-specific receptor PtdSerR, contains JmjC domain
92854_FR2_29-739	Heat shock transcription factor
96041_FR3_1-152	RNA polymerase II transcription elongation factor DSIF/SUPT5H/SPT5
96060_FR3_66-187	Transcription factor e(y)2
Rs-100916_FR2_1-328	Transcriptional activator FOSB/c-Fos and related bZIP transcription factors
Rs-100931_FR3_1-341	Transcription factor GT-2 and related proteins
Rs-103965_FR3_6-947	Transcription factor NFAT, subunit NF90
Rs-105917_FR3_339-722	Transcription factor LIM3, contains LIM and HOX domains
Rs-107136_FR1_40-299	RNA polymerase II transcription factor complex subunit
Rs-112269_FR2_1-381	Basic region leucine zipper transcription factor
Rs-112270_FR3_14-407	Basic region leucine zipper transcription factor
Rs-113329_FR1_61-366	Transcription factor XBP-1
Rs-115972_FR3_1-274	CREB/ATF family transcription factor
Rs-116088_FR2_41-465	Transcription factor PBX and related HOX domain proteins
Rs-117324_FR3_27-552	CREB/ATF family transcription factor
Rs-117326_FR3_14-540	CREB/ATF family transcription factor
Rs-119605_FR3_37-278	bHLHZip transcription factor BIGMAX
Rs-119650_FR1_31-170	Transcription factor, subunit of SRB subcomplex of RNA polymerase II
Rs-120403_FR1_6-193	bZIP transcription factor MafK
Rs-121027_FR1_55-1220	Transcription factor TMF, TATA element modulatory factor
Rs-121251_FR1_1-575	Apoptosis antagonizing transcription factor/protein transport protein
Rs-121258_FR2_1-464	Apoptosis antagonizing transcription factor/protein transport protein
Rs-126695_FR3_68-392	Transcription factor of the Forkhead/HNF3 family
Rs-127921_FR1_18-193	Transcription factor MBF1
Rs-128460_FR3_69-821	Regulatory protein MLP and related LIM proteins
Rs-131367_FR2_52-256	Basic region leucine zipper transcription factor
Rs-131369_FR1_30-270	Basic region leucine zipper transcription factor
Rs-133834_FR3_313-885	Hypoxia-inducible factor 1/Neuronal PAS domain protein NPAS1
Rs-139212_FR1_68-294	Transcription factor containing NAC and TS-N domains
Rs-145690_FR2_39-687	Alternative splicing factor ASF/SF2 (RRM superfamily)
Rs-147431_FR1_17-245	Basic region leucine zipper transcription factor
Rs-151159_FR3_33-393	cAMP response element binding protein and related transcription factors
Rs-152328_FR1_1-439	HMG-box transcription factor
Rs-158588_FR1_1-268	Nucleosome-binding factor SPN, POB3 subunit
Rs-160619_FR1_304-853	Splicing factor 3b, subunit 1
Rs-167325_FR3_105-531	Activating transcription factor 4
Rs-169240_FR3_154-301	Transcription factor NERF and related proteins, contain ETS domain
Rs-169255_FR3_67-287	Transcription factor NERF and related proteins, contain ETS domain
Rs-169445_FR3_1-403	Basic region leucine zipper transcription factor
Rs-170419_FR1_127-306	RNA polymerase II general transcription factor BTF3 and related proteins
Rs-175974_FR3_68-493	C2H2-type Zn-finger protein
Rs-183043_FR3_1-407	Transcription factor NFAT, subunit NF45
Rs-68622_FR2_34-231	Transcription factor IIB
Rs-76207_FR4_79-197	Transcription factor e(y)2
Rs-76208_FR1_121-238	Transcription factor e(y)2
Rs-81156_FR3_88-263	Nucleosome-binding factor SPN, POB3 subunit
Rs-81157_FR2_51-290	Nucleosome-binding factor SPN, POB3 subunit
Rs-85500_FR3_97-942	WD40 repeat protein

3.3.5. Proteome studies reveal sialome switching in R. sanguineus

In parallel with the transcriptome studies, proteome analysis of the salivary glands of adult female R. sanguineus was performed, using groups of ticks achieving the same range of weight gain after attachment to a rabbit. Following a shotgun proteomic approach, the generated raw data was used to query the “de novo” transcriptome assembly described above using the PatternLab software [60] which allows normalization by the normalized spectral abundance factor (NSAF) approach, which takes into account a protein’s length during the normalization process [80]. A total of 2125 transcripts were identified as coding for the peptides found by the proteomic study (Supplemental spreadsheet 2), 1745 of which matched transcript translations with a TPM value of 10 or more. To gain insight into broad relationships of secretion dynamics of tick proteins with the tick feeding processes, Z-score statistics normalized NSAF values were visualized on heat maps (Fig. 4). Indeed, the PCA plot (Fig. 5) showed remarkable clustering of the replicates, revealing the sialome switching at protein level as well.

Fig. 4.

Heat map of the normalized spectral abundance factor (NSAF) values for 2125 peptides identified by mass spectrometry in seven triplicate groups of Rhipicephalus sanguineus ticks (unfed ticks (UF) or 6 partially fed ticks with different weights (G1 - G6)). For more details, see text.

Fig. 5.

Principal component analysis plot of the normalized spectral abundance factor (NSAF) values of 2125 polypeptides identified by mass spectrometry in seven triplicate groups (unfed ticks (UF) or 6 different weights (G1 - G6)) of Rhipicephalus sanguineus ticks. For more details, see text.

To get insights on the relation between transcriptomic and proteomic data, and using solely the transcripts with a TPM value of 10 or more, we normalized the transcriptome TPM values as well as the proteome NSAF values within each tick group, taking as a value of 100 the largest TPM or NSAF value for each measured sample in all tick groups. There was a positive correlation between the maximum TPM value for a transcript and the maximum NSAF value for the same transcript, with a correlation coefficient of 0.549 (P < 1e-6), and a coefficient of determination (R²) equal to 0.301. A correlation analysis between the TPM and NSAF values of the 1745 transcripts identified 221 that were positively correlated when their TPM values where compared to their NSAF values for each tick group (Supplemental fig. 11 having a p value smaller than 0.1). Notice that many of these correlations refer to an expression that is found in a single group. Thirty per cent of these transcripts coded for secreted proteins. Conversely, there were 17 transcripts where the correlation was negative with a p value smaller than 0.1 (Supplemental fig. 12). Of these 17 transcripts, 13 had a maximum TPM on the unfed group, with no or little NSAF values, which increased in values within group 1, indicating that the transcript was present in unfed ticks but with little protein expression, which took over after the tick started feeding. All these negatively correlated transcripts belong to the housekeeping class.

Of the 1745 transcript products that were identified by MS/MS and correlated with TPM value, 1507 showed no statistically significant correlation between the TPM and NSAF results as a function of the 7 tick groups. Inspection of the graphs plotting the normalized TPM and NSAF results (Supplemental spreadsheet 2, column TU) indicated that despite the non-significant statistical correlation result, the curves were relatively similar. Indeed, if we compare the peaks of the TPM and NSAF results, 49% of the 1507 results are concordant or found in the neighbor tick group (Supplemental spreadsheet 2 and Supplemental fig. 13). A X² test contrasting the observed peak differences with an expected uniform peak distance indicated a highly significant departure of the random expectation (P < .00001).

3.3.6. Host-derived proteins are present into R. sanguineus salivary glands

Presence of host-derived proteins was already described in several species as a component of tick saliva [81–83]. This observation raises questions regarding of whether host-derived proteins could be originated by regurgitation and/or mouthpart contamination. Here we investigated for the presence of host proteins in the tick salivary gland homogenate, not saliva, excluding the possibility of regurgitation or mouthpart contamination. Host-derived proteins were identified as a component of the tick salivary gland content (Supplemental spreadsheet 2, worksheet “Host”). Out of the 72 matches of rabbit proteins found by MS/MS, we excluded eight that were at least 90% identical to R. sanguineus proteins (tubulins, actin beta, elongin A, tyrosine 3- monooxygenase and proteasome 26S subunit) and another 17 that matched skin proteins (keratins, desmoglein and plakoglobin), which are common contaminants in MS/MS experiments [84]. The NSAF levels for the remaining 47 host proteins, which includes serum albumin, immunoglobulin G chains, hemoglobin, hemopexin, lactoferrin, annexin, fibrinogen, antithrombin, among others are shown in Supplemental fig. 14. These results suggest that the presence of host proteins in tick saliva may be a real and common recycling system present in ticks, and not a result of contamination during saliva collection.

3.3.7. An insight on the diversity of salivary proteins of R. sanguineus

On a recent review (Ribeiro and Mans, submitted) it was suggested that intra-gene recombination events could be driving the diversification of the tick sialome, as evidenced in the abundant protein families of lipocalins and metalloproteases which had indications of several breakpoints as determined by the RDP4 pipeline [56]. The less abundant and less expressed cystatin protein family, however, showed no signs of recombination. Here, using the coding sequences for lipocalins (Supplemental fig. 2), metalloproteases (Supplemental fig. 3) and cystatins (Supplemental fig. 4) we found four recombination events for the lipocalins, five for the metalloproteases and zero for the cystatin family (Supplemental file 3).

4. Discussion

In this work we explored the sialotranscriptome and sialoproteome of adult female R. sanguineus tick, aiming at (1) disclosing its salivary repertoire of transcripts and polypeptides and (2) shedding light into the rate and mechanisms associated with the tick sialome switch. Accordingly, from a total of 71,643 transcripts, 28,921 had a TPM of 10 or larger in at least one library (Supplemental spreadsheet 1). The translated peptides from these transcripts served as a target database for proteomic studies, leading to the identification of a total of 2125 proteins. From these, 1745 polypeptides derived from transcripts having a minimum TPM value of 10 in any library. Of these, 221 had their group expression values for TPM and NSAF significantly correlated, indicating a synchronicity, or better, a “syngroupicity” of transcript and peptide expression. On the other hand, there were 17 transcripts that were deemed negatively correlated, the majority of which had a peak transcription in the unfed group, and a peak translation in the first tick group, indicating the transcript was “waiting” for the feeding to start to be translated. The non-correlated transcripts were found to have a non-random departure of their peak values: The NSAF peaks were significantly associated with the corresponding vicinal TPM peaks.

Previous tick sialotranscriptomes done at different days of feeding indicated vast changes in transcript composition as a function of feeding time, characterizing the sialome switching phenomenon [22,85]. This variability was also observed within proteomic studies of tick saliva. The salivary composition is time-dependent, and it changes in quantity and quality during the feeding process [86,87]. Moreover, the composition of saliva is differentially expressed when ticks are exposed to different host species [31,32].

Transcriptomes done with salivary glands of single ticks, artificially fed or fed on a rabbit, demonstrated a large variance on transcript expression and suggested that the switches occurred in a time frame well below 12 h [26]. Aiming at reducing this variance, we planned our libraries to be built from ticks having similar weights instead of similar times of feeding, as the physiological status of the tick may be better defined by its weight gain rather than the time it has been since commencement of feeding, as previously indicated by the “critical weight” of ticks that determine their host detachment and salivary gland degeneration [43]. Indeed, the PCA plots for the transcripts and polypeptides (Figs. 1 and 5) showed remarkable clustering of the replicates, even though individual groups contained ticks feeding at several days’ difference from each other but having similar weights. The distance between these clusters indicates that the larger difference occurred between the unfed and fed groups, and, within the fed groups, between group 1 vs group 2, followed by groups 2 vs 3 and 3 vs 4, 5 and 6. The last three groups formed distinguished clusters, but were located very near each other in the PCA plots. The statistically significant differentially expressed transcripts as well as the clusters identified by the Expander program followed this pattern. Notably, there were many transcripts and peptides that were found in a single group, most often at groups 1 and 2. These results indicate that several sialome switches must have occurred between groups 1 and 2, 2 and 3 and 3 and 4, raising the possibility that there are hundreds of novel transcripts yet to be discovered. A future and improved experimental design would be to have the transcriptome and proteome done on single gland pairs, similar to performed by Perner et al. [26], or with an specific type of acini, since each type has a specific function and composition [88,89], from ticks weighting from 1.8 mg to 40 mg at 10% weight increments, a minimum of 35 ticks or libraries, not an impossible task.

It has been proposed that the sialome switching mechanisms in ticks could be associated with classical transcription factor regulation and/or epigenetic regulation [26]. Tables 5 and 6 presents transcripts associated with these processes. RNAi experiments targeting these transcripts may shed light into the mechanism of sialome switching.

In addition to tick proteins, many host proteins have been described as components of tick saliva in different tick species [81–83,90–92], suggesting the presence of a recycling system of host proteins during tick feeding [83]. Immunoglobulins, serum albumin, enzymes, serine protease inhibitors, and other host blood proteins were found [33,34]. The pattern of appearance of host proteins in the salivary glands of R. sanguineus is intriguing (Supplemental spreadsheet 2 and Supplemental fig. 14). First, it implies a mechanism of midgut transportation of the host proteins to hemolymph, and from the hemolymph to the salivary glands. Digestive cells of the cattle tick, Rhipicephalus microplus, were shown to transport hemoglobin and albumin by distinct mechanisms [93], indicating the possibility of direct transport of host blood proteins to the hemocoel. Lacking still is the knowledge of how these host proteins are acquired by the salivary glands, and how their voyage is accomplished through the cell. Are they inserted into the endosomal compartment and later secreted together with salivary gland-synthesized proteins? If so, can the host proteins be glycosylated in this process? Can the anti-tick immunoglobulins bind with the newly synthesized tick proteins? Will this binding trigger a misfold reaction [94]? Another intriguing observation is the presence of host haptoglobin, hemopexin and lactoferrin in the tick’s saliva. These proteins may act as scavengers of hemoglobin degradation products that are the substrate of heme-oxygenase, an enzyme associated with tissue repair [95] which was found activated at skin sites that were fed by sand flies [96]. Finally, while the majority of the 47 host proteins were found in all tick groups, some of them had distinct peaks that could reflect a selective control of protein transport through the midgut or salivary gland uptake.

Finally, the finding of intra-genic recombination breakpoints on the abundantly expressed lipocalin and metalloprotease families raises the possibility that non-homologous recombination events may be a mechanism increasing the diversity of salivary transcripts in ticks. Whether these events are meiotic or somatic remain to be elucidated. If the events are somatic, then the recombinant transcripts would only exist transiently in the salivary glands and attempts to map their genomic location would fail. In meiotic recombination the recombinants should be mappable to the genome. However, the diversity or heterozygosity of these recombinants may be so large as to be very difficult to map all recombinant products. For example, if a gene has one recombinant breakpoint, we could imagine one chromosome representing A-B and another one a-b, thus four types Ab, aB, ab and AB would be produced, thus a single heterozygotic tick has only the maximum ability to recover 50% of the possible forms. With two recombination breakpoints, nine recombinant forms would exist, and with three breakpoints we would have 16 possible forms, etc. If we multiply these considerations by all genes having intragenic recombination breakpoints, several thousand different genomes would emerge. Accordingly, mapping of these recombinant coding sequences to a single genome would be futile. Indeed, a calculation of the successful genome mapping of transcripts coding for lipocalins and ribosomal proteins from Ixodes scapularis is consistent with this scenario: the program BLAT [97] was used to map the transcripts to the available genome then the exons were summed up to find the percent coverage for the transcripts. While we found an average of 51% genomic coverage for 721 transcripts coding for ribosomal proteins, we found only 31% genomic coverage for lipocalins (Ribeiro, in preparation). Notice that the published genome is a draft that was expected to cover 57% of the tick genome [98]. Thus, the mapping of 51% of the ribosomal proteins is an expected result. This indicates that the lipocalins are underrepresented in the published genome in relation to the more conserved ribosomal proteins. Working with single pair of glands it would be possible to use its mRNA to build a transcriptome library and save its DNA for attempting salivary genomic mapping by PCR methods. Carcass DNA from the same tick would serve as a non-salivary gland genomic control. This experimental design hopefully will allow to distinguish between somatic and meiotic recombination as driving the diversity of transcripts in tick salivary glands.

Significance:

Ticks are a burden by themselves to humans and animals, and vectors of viral, bacterial, protozoal and helminthic diseases. Their saliva has anti-clotting, anti-platelet, vasodilatory and immunomodulatory activities that allows successful feeding and pathogen transmission. Previous transcriptomic studies indicate ticks to have over one thousand transcripts coding for secreted salivary proteins. These transcripts code for proteins of diverse families, but not all are transcribed simultaneously, but rather transiently, in a succession. Here we explored the salivary transcriptome and proteome of the brown dog tick, Rhipicephalus sanguineus. A protein database of over 20 thousand sequences was “de novo” assembled from over 600 million nucleotide reads, from where over two thousand polypeptides were identified by mass spectrometry. The proteomic data was shown to vary in time with the transcription profiles, validating the idea that the expression switch may serve as a mechanism of escape from the host immunity. Analysis of the transcripts coding for lipocalin and metalloproteases indicate their genes to contain signals of breakpoint recombination suggesting a new mechanism responsible for the large diversity in tick salivary proteins. These results and the disclosed sequences contribute to our understanding of the success ticks enjoy as ectoparasites, to the development of novel anti-tick methods, and to the discovery of novel pharmacologically active products.