Genomic data

To construct and compare the phylogenetic relationships between the Acari, 12 genomes were downloaded from the NCBI database for use in the analyses (S1 Table). Two Arachnid outgroups namely the Arizona bark scorpion (Centruroides sculpturatus) and the American house spider (Parasteatoda tepidariorum) were also downloaded. Assembly completeness of each genome was assessed using BUSCO v3 (Benchmarking Universal Single-Copy Ortholog) [17] with the Arthropoda dataset. Comparative analysis of D. pteronyssinus genome assemblies were conducted against previously published D. pteronyssinus strains [18, 19].

Phylogenetic analysis

Phylogenomic analysis of D. pteronyssinus was undertaken with reference to 11 other species (subclass Acari), consisting of six Parasitiformes and five Acariformes (S1 Table). Supermatrix and supertree phylogenomic methods were employed to infer the evolutionary relationships. Suitable phylogenetic markers were selected by locating single copy orthologs (from BUSCO analysis above) in the Acari and outgroup genomes. Single copy orthologs that were ubiquitously present (n = 111) were aligned using MUSCLE [20]. Individual gene families alignments were subsequently concatenated to yield a supermatrix 77,878 amino acids in length. This supermatrix was used to reconstruct a maximum likelihood phylogenomic species tree using RAxML [21] utilising the LG+G+I+F model as selected by ProtTest [22], branch supports were determined using 100 bootstrap replicates.

Single copy ortholog families, present in at least four species (n = 2,796), were identified and individually aligned using MUSCLE. Subsequent phylogenies were generated using FastTree [23]. A supertree was constructed using the matrix representation with parsimony (MRP) method implemented in Clann [24] using the 2,796 gene trees as input with 100 bootstrap replicates. The resultant phylogeny was visualised and annotated using the Interactive Tree of Life (iTOL) [25].

Proteo-genomic analysis of assembly completeness

Proteo-genomic software Peppy [26] was used to generate peptide databases from three D. pteronyssinus assemblies [18, 19, 27]. LC-MS/MS spectra derived from proteomic experimentation on D. pteronyssinus airmid were searched against the six-frame translated genomes (maximum FDR 0.01; precursor tolerance 2000; fragment tolerance 300; digestion rules–cleavage acid R & K, missed cleavages 1; static mods -mod C 57.021464). For completeness and comparative purposes, the spectra were searched against the corresponding translated predicted protein-coding genes for each assembly. Output files were filtered to locate unique peptides and corresponding genomic locations.

Annotation of predicted proteome

Annotation of predicted proteins was achieved using BLAST2GO Version 5.0 [28] to sequentially search SwissProt (Downloaded; 14/01/2018) then NCBInr (Downloaded; 30/08/2017) database. Gene Ontology (GO) terms were assigned to predicted proteins (GO cut off 55, GO weight 5, E value hit filter E^-06 and default computational evidence codes). InterPro Scan was used to identify; families, domains, sites and repeats in predicted proteins (CCD, HAMAP, HMMPanther, HMMpfam, FPrintscan, BLASTPromDom) and performed secretion peptide prediction using SignalP ver4.0 in parallel. Mapping feature facilitated mapping of GO terms to enzyme codes.

D. pteronyssinus specific proteins

D. pteronyssinus predicted proteins without BLASTp homology (n = 3,906) to proteins in NCBInr/SwissProt were searched (tBLASTn; E value ≤ E^-05) against Acari genomes for presence of homologs in closely related species (S1 Table). Predicted proteins without significant homology were considered D. pteronyssinus-specific proteins (n = 1,848) and were then searched against other the other two available D. pteronyssinus genome assemblies [18, 19]. Proteins without homology to predicted proteins in these assemblies were considered D. pteronyssinus airmid-specific proteins.

Identification of LEA-like proteins

LEA homologs were identified by performing BLASTp (E-value ≤ 1E^-05) searches [29] against the Late Embryogenesis Abundant Proteins Database (LEAPdb) [30].

Mite group allergen orthologs

The majority of mite allergens identified to date have been assigned to groups (Group 1–33) in accordance with their order of discovery [7]. Chan et al (2015) reported a further seven non-chronological allergens for D. farinae [31]. Mite group allergen (MGA) orthologs were identified by performing BLASTp searches of query FASTA files containing MGA from D. pteronyssinus & D. farinae (when no other sequences were available for D. pteronyssinus) against the Acari genomes utilised in this study (S1 Table). Allergen orthologs had to satisfy the following criteria; have reciprocal best BALST hits (RBH) with an allergen (E-value ≤ 1E^-05) and a minimum alignment length of 80 amino acids with an identity ≥ 35% in accordance with FAO/WHO guidelines [32]. Allergens that had a RBH but did not meet the alignment and identity criteria were considered RBH homologs, these were included in the sequence similarity network and visualised using Gephi [33]. Each protein was represented by a node, two proteins were connected by an undirected edge if they were homologous (BLASTp ≥ 1E^-05).

Allergens and predicted allergens

D. pteronyssinus airmid MGA and non-chronological allergens were located by performing a local BLASTp (E-value ≤ 1E-03) search of the D. pteronyssinus predicted proteome against the FASTA sequence file containing the MGAs located above. The match with the smallest E-value was chosen as designated MGA. These sequences were annotated as “Der p1 Allergen” or Der f 22 like-allergen etc. for D. farinae based BLAST hits. Subsequent BLASTp hits were considered MGA homologs and were assessed for potential cross-reactivity with MGA in accordance with FAO/WHO guidelines [32]. Sequence similarity between known allergens (Uniprot “allergenome”) and D. pteronyssinus airmid predicted proteins was assessed in accordance with FAO/WHO guidelines [32].

D. pteronyssinus airmid culture

D. pteronyssinus airmid were obtained from cultures housed at airmid healthgroup ltd (Dublin, Ireland) and maintained on diet composed of dried porcine liver and yeast, house dust mite maximal media (HDMMM, airmid healthgroup ltd, Ireland) at 75% relative humidity and 25°C.

Harvest of mites and spent culture medium

Mites from replicate cultures (n = 5) were separated from spent culture medium (SM) by sieving, and saline floatation method [16], washed with distilled water and surface-sterilised by submersion in 70% ethanol (3 min) followed by washing with sterile distilled water. An average of 10.8 mites were present per mg of spent medium. Mites were then snap frozen in liquid nitrogen, and lyophilised. SM (n = 5) was divided into aliquots (200 mg) and stored at -70°C prior to protein extraction.

Protein extraction

Lyophilised mite bodies (MB) were ground to a fine powder. Proteins were extracted from MB (25 mg) by addition of 500 μl lysis buffer and quantified according to methods described by Owens et al (2015) [34]. Proteins from spent culture medium (SM, n = 5) were extracted by addition of glass beads (50 mg; 0.5 mm BioSpec Products) and 1000 μl lysis buffer [34], followed by bead-beating (30 Hz; 5 min; MM300, Retsch®). MB (n = 2) and SM lysates (n = 2) were pooled for gel filtration (one biological replicate). Protein lysates for shotgun proteomic analysis (n = 4) were normalised (MB; 0.4 mg/ml, SM 0.3 mg/ml), then prepared for digestion according to Owens et al (2015) [34].

Culture media contaminants

Proteins deriving from HDMMM (200 mg) were extracted (n = 1) and prepared for shotgun proteomic analysis as per methods for spent growth culture medium.

House dust protein extracts

Der p 1 positive (0.2–16.94 μg Der p 1 per gram house dust) house dust protein extracts (n = 21) were provided by airmid healthgroup ltd (Dublin, Ireland).

Gel filtration chromatography

Gel filtration chromatography was carried out using an ÄKTA purifier coupled with a Superdex 200 10/300 GL column (GE Healthcare, Germany), equilibrated in PBS. Filtered specimens (0.22 μm), were injected (500 μl) and separated (flow rate 0.4 ml/min) with absorbance monitored at 215, 254 and 280 nm. Fractions were collected (between ~ 6 and 32 ml) and stored at -20°C until further analysis.

Proteolytic digestion of protein extracts for proteomic analysis

Specimens for proteomic analysis were prepared for LC-MS/MS as described by Owens et al (2015) [34].

Nano-flow liquid chromatography electro-spray ionization tandem mass spectrometry (LC-MS/MS) analysis

Peptide mixtures were analysed using a Thermo Fisher Q-Exactive mass spectrometer coupled to a Dionex RSLC nano for LC-MS/MS analysis. LC gradients operated from 3–40% acetonitrile over 40 min, with data collection using a Top15 method for MS/MS scans [35].

Representative proteome

LC-MS/MS spectra obtained from proteomic analysis of D. pteronyssinus airmid were randomised into 5 groups (14–20 files each). Spectra were searched using Sequest HT engine within Proteome Discoverer (Version 1.4) against D. pteronyssinus airmid predicted proteome (peptide filters; set to medium peptide confidence and protein filters; set to two peptides per protein). Protein molecular weight and pI for predicted proteome and representative proteome were calculated using JVirGel [36].

MaxQuant and perseus data analysis

Protein identification and label free quantitative (LFQ) analysis was conducted using MaxQuant (Version 1.6.1.0; http://maxquant.org/), statistical analysis of MaxQuant output data was performed by Perseus (Version 1.6.2.2) as described in O’Keeffe et al (2014) [37].

Culture media contaminants database

As it was not possible to fully remove culture media from specimens prior to proteomic analysis, a custom contaminates database was generated. This allowed for exclusion of protein identifications deriving from culture media (HDMMM) which contained porcine liver and baker’s yeast. Spectra obtained from LC-MS/MS of HDMMM were interrogated (MaxQuant and Perseus) against combined proteomic database of Sus scrofa and Saccharomyces cerevisiae, resulting in the identification of 2,135 proteins (min. 1 peptide). These proteins were added to MaxQuant contaminants database to generate a custom contaminants database (n = 2,380 contaminants).

Data analysis of D. pteronyssinus proteomes

Spectra obtained from LC-MS/MS were interrogated (MaxQuant and Perseus) using either standard contaminant (HD samples) or custom contaminant (MB & SM samples) databases. Proteins were considered present when a minimum of two peptides (1 unique) for each parent protein was observed. Proteins meeting the following criteria were included in the analysis; (i) identified in two of the four non-fractionated whole protein extracts or (ii) identified in one chromatographic fraction.

Qualitative assessment of mite group allergen localisation

A qualitative assessment of localisation of mite group allergens to MB or SM was conducted. An allergen was considered present in MB/SM proteome if it was; (i) absent from one dataset or (ii) found at higher LFQ intensity and ms/ms count.

Phylogenomic assessment of D. pteronyssinus reveals closest relative to be Euroglyphus maynei

Supermatrix and supertree phylogenomic methods were employed to infer the evolutionary relationships between the Acari species that have genome data available (S1 Table). Both supertree and supermatrix methodologies generated phylogenies with identical topologies and similarly high levels of bootstrap support (BP) for the monophyly of the Parasitiformes and Acariformes superorders (Fig 1A, 100% BP). Within the Parasitiformes superorder, Ixodida and Mesotigmata are monophyletic (Fig 1A, 100% BP). Within the Acariformes superorder, Trombidiformes and Sarcoptiformes are also found to be monophyletic (100% BP) although only a single representative of the Trombidiformes (Tetranychus urticae) is represented in our dataset (Fig 1A).

Download:

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

Fig 1. Arachnida phylogenomic species tree and network of mite group allergen orthologs in genomes of twelve Acari and two Arachinds.

A. Phylogenomic species tree of 11 Acari species inferred using maximum-likelihood supermatrix (111 ortholog families, 77,878 aa aligned, LG+G+I+F model) and supertree (2,796 ortholog families) methods. Bootstrap support values are given for each branch, with values above branches corresponding to the supermatrix phylogeny and values below corresponding to the supertree phylogeny. Both supermatrix and supertree methods infer a strongly supported (87% and 100% bootstrap support respectively) sister group relationship between D. pteronyssinus and E. maynei to the exclusion of D. farinae. Therefore, in evolutionary terms D. pteronyssinus and E. maynei are more closely related to one another even though D. pteronyssinus and D. farinae belong to the same genus. B. Network map depicting abundance of mite group allergen homologs in each species. Homologous allergen group together forming an interconnected network distinct from non-homologous allergens i.e 2 & 22, groups 3, 6 & 9, groups 5 & 21, and groups 15 & 18.

https://doi.org/10.1371/journal.pone.0216171.g001

To date, phylogenetic studies of Acari (mites and ticks) have been restricted to multi-locus studies utilising a small number of genes, due to the absence of full genome sequences [38–40]. Our phylogeny infers a strongly supported (87% and 100% BP for supermatrix and supertree methods, respectively) sister group relationship between D. pteronyssinus and Euroglyphus maynei to the exclusion of D. farinae. This confirms previous studies which utilised only two and six genes but observed the same phylogenetic relationship between D. pteronyssinus and E. maynei [38, 41]. Therefore, in molecular evolutionary terms, D. pteronyssinus and E. maynei are more closely related to one another, even though D. pteronyssinus and D. farinae are currently classified within the same genus. Furthermore, the phylogeny confirms the paraphyly of the Dermatophagoides genus as previously reported [41].

Comparison of D. pteronyssinus airmid predicted proteome with other D. pteronyssinus assemblies

D. pteronyssinus airmid genome assembly [27] completeness was assessed by proteo-genomics. Peppy software [26] facilitated the mapping of 615,150 spectra (28,001 non-redundant) to the predicted proteome and 402,998 (21,505 non-redundant) to the assembly. Peptides that spanned intron-exon junctions identified in the predicted proteome were not mapped to the assembly. Of the 21,505 peptides mapped to the genome assembly, 96.2% were also identified in predicted proteins (n = 20,683). Several peptides (n = 65) were located adjacent to predicted genes indicating that 65 gene models may need to be extended. The predicted proteome of D. pteronyssinus airmid [27] incorporates 8.3 million amino acids, this is significantly more than the Liu et al. (2018) and Randall et al. (2018) assemblies, which have 6.7 million and 5.9 million amino acids respectively. Moreover, despite having fewer predicted proteins than the other two available D. pteronyssinus proteomes (i.e. 12,530 versus 15,846 proteins [19] and 19,368 [18]), D. pteronyssinus airmid has on average longer protein coding genes (557aa vs. 425aa & 304aa respectively). These results indicate that differences in gene calling methodologies are most likely responsible for the differences in the number of protein coding genes.

Proteogenomic comparison of the predicted proteomes reveals the highest number of proteins were identified using the Waldron et al (2017) proteome (n = 4,581), followed by Randall et al (2018) with 4,416, then Liu et al (2018) with only 3,408 proteins identified. Therefore, unsurprisingly proteogenomic analysis of the available D. pteronyssinus genome assemblies against our protein samples which are derived from D. pteronyssinus airmid reveals the Waldron et al (2017) assembly and predicted proteome to be the most appropriate for our analyses, as higher numbers of protein identifications are uncovered relative to the other current assemblies [18, 19, 27].

Mite group allergen orthologs in arachnidia

The presence of common MGA in the genomes of Arachnida species (S1 Table) was investigated, identifying multiple putative cross-reactive MGA orthologs in mite species. Most MGA had numerous orthologs distributed across all species (S1 Data) with presence closely linked to phylogeny. To help visualise the abundance of MGAs in the different Arachnida species a homology network was generated. Our results show that D. farinae, D. pteronyssinus, and E. maynei contained at least one MGA ortholog for all groups investigated, with the exception of Groups 23 & 24 in E. maynei. Group 7 and 14 allergens are only located in the Sarcoptiformes subset of Acariformes. Homologous allergens, Der p 5 and Der p 21, were present in the closely related Acariformes D. pteronyssinus, D. farinae, E. maynei and Psoroptes ovis, but absent from the other species (Fig 1B and S1 Data). Group 23 allergens are specific to D. farinae and D. pteronyssinus. More distantly related species from the Parasitiformes superorder were either missing orthologs of particular allergens such as Der p 4, 5 & 21 or contained RBH homologs only, Der p 1, 2, 22, 23 & 27 for example, (S1 Data). Serine proteases (groups 3, 6 & 9) appear to be expanded in some species, with a minimum of 10 homologs in D. pteronyssinus to a maximum of 28 in ixodes scapularis (Fig 1B and S2 Data).

Annotation of D. pteronyssinus airmid predicted proteome

Predicted proteome annotations.

Multi-database Blast2GO workflow enabled annotation of 96% of the predicted proteome (n = 11,996, S3 Data). Gene Ontology (GO) terms were assigned to 68.2% of proteins (n = 8,546, Fig 2A). InterPro Scan (IPS) assigned IPS annotations to 95.5% of predicted proteins (n = 11,971), 6, 874 with IPS GO terms and IPS IDs to a further 3,804 proteins (Fig 2B). SignalP4.0 identified eukaryotic secretion signals in 10.3% of predicted proteins (n = 1,293, Fig 2B). Enzyme codes (EC) were assigned to 21.5% of the predicted proteins (n = 2,689, Fig 2C) with hydrolases (EC:3.0) representing the largest enzyme category (n = 1,244). Putative peptidase activity was identified in 377 predicted proteins comprising almost 3% of the total predicted proteome. Peptidase EC (EC:3.4) were assigned to 275 peptidases, the remaining peptidases (n = 102) were identified by GO annotations. Enzymes have a propensity to cause allergy [42]. The potent peptidase activity of Der p 1 has been shown to disrupt numerous immune system processes [43] and it is thought lesser studied peptidases may have a similar effect [42]. Enzymes, particularly those with predicted secretion peptides, should be considered in the context of patient exposure as they are more likely to be excreted into house dust and therefore may augment the immune response.

Download:

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

Fig 2. Annotation of D. pteronyssinus airmid predicted proteome.

Annotation of D. pteronyssinus airmid proteins conducted using Blast2GO to search SwissProt, NCBI and InterPro databases for homology to known proteins and functional domains. A. GO annotations were assigned to 68.2% of the predicted proteome (n = 8,546). A small number of proteins had Blast hits with proteins in NCBInr or SwissProt databases but were not assigned a GO term (n = 372). The remaining proteins (n = 3,612) were not homologous with proteins in these databases. Most of the predicted proteins were annotated with more than one GO category (Biological Processes, Molecular function, Cellular component) with median number of assigned GO terms being 9 (Range: 0–164). B. InterPro scan feature of Blast2GO enabled assignment of InterPro IDs to 95.5% of predicted proteins/proteome? (n = 11,971). InterPro GO IDs were assigned to 55% (n = 6, 874) of the predicted proteome and predicted eukaryotic secretion peptides to 10.3% of the predicted proteome (n = 1,293). A further 3, 804 proteins were assigned InterPro Scan (IPS) IDs. Multi-database Blast2GO workflow enabled 96% of the predicted proteome to be assigned some form of annotation (n = 11,996). A small number of proteins had BLAST hits with proteins in NCBInr/SwissProt databases but were not assigned a GO term (n = 372). Several predicted proteins (n = 294) were assigned GO annotations but BLASTp hit alignments did not meet the required threshold of significance. The remaining proteins (n = 3,612) were not homologous with proteins in these databases C. Enzyme code classes assigned to D. pteronyssinus airmid predicted proteins (n = 2,689), representing 21.5% of D. pteronyssinus airmid predicted proteome. Hydrolyses (EC:3.0) formed the largest enzyme category (n = 1,244) and Isomerases (EC:5.0) the smallest (n = 87). D.Venn diagram depicting overlapping categorisation of allergenic and predicted allergenic proteins. D. pteronyssinus airmid predicted proteins were assigned into unambigous allergen goupings based upon BlastP homology to known allergens. We found full predicted proteins for all reported MGA (Groups 1–11, 13–16, 18, 20–33) and Seven non-chronalogical allergens (81.5–100% identiy). Subsequent blast hits were considered MGAH. Predicted proteins with potential cross-reactivity [32] with allergens from other species (Uniprot “allergenome”) were annotated as Allergen homolog (AH). Many Allergenic/potentially allergenic proteins were present in multiple allergen analyses, in total 332 allergenic/potentially allergenic proteins were identified.

https://doi.org/10.1371/journal.pone.0216171.g002

Allergens and predicted allergens

We searched the D. pteronyssinus airmid proteome for the presence of 37 prevously reported mite allergens [7, 31]. MGA were identified for D. pteronyssinus airmid, we found full protein sequences (n = 37) for all reported MGA; Groups 1–11, 13–16, 18, 20–33 and seven non-chronological allergens (81.5–100% identity, Table 1). Subsequent BLAST hits (E-value ≤1E^-03) were considered MGA homologs (n = 233, Table 1 and S5 Data). Der p 1-like cysteine proteases were represented by 31 homologs (20.9–63.7% identity), several were found in clusters of 2–3 adjacent protein coding genes (n = 13). MGA homologs with high sequence similarity to the query MGA (> 67% identity) met criteria for being considered isoallergens [55]. Isoallergens were identified for eight different MGAs (Table 1). Der p 28 has five isoalllergens (69.4–85.7% identity) and two were identified for Der p 29 (75.7–88.1% identity). One-third of MGA homologous proteins met the criteria for potential allergenicty [32] and were annotated as MGA Homologs (MGAH) (n = 93, S5 Data). In addition, many D. pteronyssinus airmid predicted proteins (n = 326, S6 Data) also exhibited significant similarity [32] to allergens from other species, suggesting they may be cross-species allergens. Most of these allergen homologs (AH) had multiple high scoring alignments with non D. pteronyssinus allergens (n = 991, Range:1–204, Median: 4). Significant overlap was seen between allergenic (MGA) and predicted allergenic (MGAH & AH) proteins with many being observed in more than one category, illustrated in Fig 2D. The structure and function of a protein has important implications for allergenicity, most allergenic proteins are limited to just 2% of protein families, with cross-reactivity linked to protein family rather than allergen source [6, 56, 57]. Most potential allergens highlighted in this study had predicted biochemical functions that placed them in well defined allergen families [6]. For example, predicted enolases DERPT_G12026 and DERPT_G4831 have high levels of sequence similarity (67–87%) with enolases from up to 19 phylogenetically distinct species (S6 Data). Enolase has long been recognised as a major cross-reacting allergen in plants, fungi, fish, and arthropods [58]. Moreover, the presence of at least one putative cross-reactive enolase ortholog [32] in all 12 Acari and two Arachnid outgroups (S2 Data) highlights the importance of this pan-allergen protein family. Several cyclophilins (n = 12) were annotated as putative cross-reactive proteins, of note Der f 29 like allergen (DERPT_G9923) exhibited sequence homology (52–83% identity) with cyclophilins from 12 different species (S6 Data). Cross-reactive cyclophilins from HDM, mouse, humans and fungi are well reported in the literature [59–61]. HDM and fungi are frequently co-present in HD [62] with Alternaria and Aspergillus spp. being the most common source of mould allergens [63]. D. pteronyssinus airmid predicted proteins exhibited homology to Aspergillus fumigatus (n = 19) and Alternaria alternata (n = 6) allergens. Homology between HDM and fungal proteins may play a role in reported fungal exacerbation of HDM-induced asthmatic symptoms [64, 65]. Excretion of putative allergenic proteins into HD via faecal particles would implicate a route of exposure and therefore has significant implications for allergy. Eukaryotic secretion peptides were predicted in 20.7% of allergenic/potentially allergenic proteins. Even if these putative allergens were unable to induce immune responces in their own right, co-presence with other immune modulators may be involved in bystander sensitization. Der p 1 accumulates in HD, levels exceeding 2 μg/g of dust are considered hazardous to the occupants [1]. In addition to being a potent activator of the immune system, Der p 1 has an adjuvant effect, enhancing IgE production against bystander molecules that may be present in HD [66]. Therefore, any protein accumulating in HD should be considered in the context of being a bystander allergen candidate.

Download:

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

Table 1. Mite group allergens (MGA), isoallergens and MGA homologs.

https://doi.org/10.1371/journal.pone.0216171.t001

Proteomic characterisation of D. pteronyssinus airmid

Representative proteome.

Analysis of LC-MS/MS spectra obtained from proteomic analysis of D. pteronyssinus airmid resulted in the high confidence identification of 3,931 proteins (S7 Data). This representative proteome comprised 31.4% of the predicted proteome of D. pteronyssinus airmid with experimental evidence for expression. Protein molecular weight and pI were widely distributed in the representative proteome (Range: 5.25–3086.2 kDa, pI 3.26–12.85) and similar to that of the predicted proteome (Range: 3.94–3086.2 kDa, pI 2.63–13.27), confirming the protein extraction methods were optimal for the characterisation of D. pteronyssinus airmid proteome (Fig 3A). Establishing a representative proteome, that reflects the methodological limitations of protein extraction and identification is essential for subsequent enrichment analyses. This background proteome is defined by Bessarabova et al (2012) as “the complete set of proteins known to be expressed in an organ/tissue/body liquid/cell line of sample origin” [67]. To date, proteomic investigation of D. pteronyssinus has extended to a few discrete studies [68–71]. Laboratory HDM populations have been shown to have different reproduction rates to wild-type strains [72] and isolated populations to give rise to geographical allergen variants [7]. Experimental examination of uncharacterised D. pteronyssinus airmid-specific proteins, particularly those with evidence for expression, may provide useful insights into the genes evolving within isolated populations. Our representative proteome may be expanded by employing alternative protein extraction methods, use of trypsin alternative or multi-protease digestion and depletion of high abundance proteins [73–75].

Download:

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

Fig 3. Proteomic characterisation of D. pteronyssinus airmid.

A. Proteomic Strategy for characterisation of D. pteronyssinus airmid. Flow diagram depicting proteomic strategy for characterisation of D. pteronyssinus airmid excretome and mite body proteome utilising whole sample analysis and sample fractionation analysis. A1. Gel filtration chromatography of D. pteronyssinus airmid whole body homogenate. Whole body homogenate protein extract (4.5 mg; 500 μl injection) separated by size exclusion chromatography (Superdex 200 10/300 GL; 0.4 ml/min flow rate). 0.5 ml fractions collected between 6.21 and 32.26 ml post-injection. A2. SDS-PAGE analysis of D. pteronyssinus airmid whole body homogenate fractions. SDS-PAGE (4–20%) with silver staining of whole body homogenate protein extract (Lane 1; 10 μg) from D. pteronyssinus airmid and fractions from gel filtration (6.2–22.2 ml post-injection) of same (Lanes 2–13, 17 μl loaded). SDS-PAGE visualisation of fractions (0.5ml) depicting fractionation of complex protein mixture into reduced complexity extracts containing proteins of approximately similar sizes. Fractions, 7.2–32.2 ml, post injection (n = 50) were pooled to give 1–1.5 ml aliquots (n = 21),100 μl was processed for LC-MS/MS. A3. Gel filtration chromatography of D. pteronyssinus airmid spent growth medium (SM). Protein extract of spent growth media (0.2 mg; 500 μl injection) separated by size exclusion chromatography (Superdex 200 10/300 GL; 0.4 ml/min flow rate). 0.5 ml fractions collected between 6.1 and 32.1 ml post-injection for SDS-PAGE analysis. Proteins in gel filtration fractions were acidic and therefore unsuitable for analysis by SDS-PAGE. Therefore, fractions were pooled and analysed according the methods used for mite body extract fractions. Fractions, 7.1–32.1 ml, post injection (n = 50) were pooled to give 1–1.5 ml aliquots (n = 21), 100 μl was processed for LC-MS/MS. B. Alignment of theoretical molecular mass and pI from D. pteronyssinus airmid PP (n = 12, 530) and representative proteome (n = 3,931) Software: http://www.juirgel.de Hiller et al., (2006). Proteins > 900 kDa excluded for graphing purposes (DERPT_G11449, DERPT_G9606, DERPT_G4775, DERPT_G8007,3088 kDa, 2021 kDa, 1122 kDa & 1001 kDa respectively). Predicted proteins with strong proteomic evidence for expression (2+ medium confidence peptides) accounted for 31.4% of the predicted proteome (n = 3, 931). Calculated molecular mass and pI (Range: 5.25–3086.2 kDa, pI 3.26–12.85) of representative proteome was approximate to that of the predicted proteome (Range: 3.94–3086.2 kDa, pI 2.63–13.27). Median protein size in the D. pteronyssinus airmid PP was 54.25 kDa. The largest predicted protein coding gene was DERPT_G11449 at 79,548 bp and encoded a paladin (Titin) protein homolog (26,516 a.a, 3086.2 kDa, pI 4.93). The smallest predicted protein was DERPT_G12515 (36 a.a, 3.94 kDa, pI 6.04) of unknown function, encoded by a 106 bp gene. The median protein pI was calculated to be 7.42, the protein with lowest predicted pI of 2.63 was DERPT_G12367 (50 a.a, 5.45 kDa), DERPT_G11425 had the highest predicted pI at 13.27 (67 a.a, 6.94 kDa). C. Venn Diagram depicting overlap between Mite Body (MB), Spent Culture Medium (SM) and House Dust (HD) Proteomes, totalling 1,952 proteins identified.

https://doi.org/10.1371/journal.pone.0216171.g003

Wildtype proteome of D. pteronyssinus.

Extending the relevance of D. pteronyssinus proteomics to the environment is essential, as very little is known about which HDM components are present in inhalable air [10]. Proteomic analysis of 21 Der p 1 positive HD samples revealed the presence of 150 D. pteronyssinus proteins (S8 Data), the ten most abundant are given in Table 2, with sequence coverage ranging from 9.6–73.3%. Here, it can be seen that allergens Der p 1, 2, 5, 14 and Der f 6 like allergen are amongst the most abundant D. pteronyssinus proteins in HD. Der p 1 and Der p 2 are considered major allergens while Der p 5, 6 and 14 are mid-tier IgE binders, all were amongst the ten most abundant proteins in HD [7]. Five non-allergenic proteins were also highly abundant, suggesting that sensitisation and IgE binding is a result of the unique properties of the protein rather than its abundance in house dust. Allergenic/potentially allergenic molecules accounted for almost 51% of all proteins identified in HD (n = 76). The predicted functions of many of these proteins place them into well-established allergen families, therefore their presence in HD and sequence similarity to known allergens make them interesting targets for future studies seeking to identify new allergens. Comparative analysis between allergenic and non-allergenic components of HD may reveal epitopes or structural characteristics common to inhalant allergens [10]. Our data illustrates the utility of high sensitivity protein MS as a novel way to identify HDM products in the wild-type environment and has significant implications for the development of immunotherapies that mimic natural exposure. Previously, researchers have examined numerous commercially available diagnostic and therapeutic HDM extracts, finding many were lacking important allergens and some had several fold variations in Der p 1 to Der p 2 ratios [76]. For example, the 2017 international consensus (ICON) report on the clinical consequences of mite hypersensitivity states that the “development of objective methods to assess allergen exposure and environmental control outcomes” are unmet and in need in mite allergy research [1]. Furthermore, the European medicines agencies guideline on the clinical development of products for specific immunotherapy for the treatment of allergic diseases, states that for seasonal allergies “it is mandatory to document the exposure to the relevant allergens” and “it is recommended to document the exposure level for the individual patient especially for the evaluation periods to evaluate the variation of indoor allergens” [77]. Generating diagnostic and therapeutic HDM extracts with allergen content and ratios that mimic natural exposure is of great importance. Characterisation of the factors affecting allergen repertoire and accumulation in different home microhabitats may give rise to much needed strategies for reducing allergen exposure for sensitised individuals [78].

Download:

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

Table 2. Top 10 most abundant D. pteronyssinus proteins identified in house dust protein extracts.

https://doi.org/10.1371/journal.pone.0216171.t002

Proteome of laboratory-reared D. pteronyssinus airmid.

D. pteronyssinus airmid whole protein lysates (WPL) were analysed directly (shotgun) and subjected to fractionation by size exclusion chromatography resulting in identification of 1,948 MB and SM proteins by high sensitivity protein mass spectrometry. Protein identification was confirmed by detection of at least two peptides per protein, and percentage sequence coverage ranged from 0.2 to 99.4% across the entire protein dataset. Gel filtration fractionation of MB extracts and SM (Fig 3A1–3A3) led to the unique identification of 248 and 105 proteins (18% of total identified proteins) from chromatographic fractions. Of the proteins identified, 1,076 proteins (58.2%) were uniquely found in MB extracts (S9 Data), while 99 proteins (0.8%) were solely identified as secreted proteins (S10 Data). Faecal rich SM was obtained by sieving to remove large mites, therefore smaller mites remained. In a previous proteomic study of D. pteronyssinus faeces, it was suggested that the method of faeces sample collection is beneficial over the sieving technique, which unavoidably contains mite bodies and growth media [69]. While culture media-derived proteins (contaminants) will also be detected in proteomic analysis, it is possible to differentiate true dust mite proteins from contaminants by use of a culture media contaminants database as demonstrated in our study. Several functional terms assigned to proteins were differentially represented in MB (S9A–S9C Data) and SM (S10A–S10C Data) proteomes compared to the RP.

Several GO terms were differentially represented (Over-represented: n = 161, Under-represented: n = 6, S9A Data), of note, GO Cellular Component terms cytosol (Fisher’s P = 5.3E-49, GO:0005829) and mitochondrion (Fisher’s P = 8.5E-33, GO:0005739) were the most highly over-represented terms. This finding supports our proteomic strategy, showing significant enrichment of GO terms associated with intracellular activities. EC were highly represented in the MB proteome, assigned to 44.6% of proteins identified (n = 824). Only two enzyme names were found to be differentially represented, acting on peptide bonds (Fisher’s P = 1.28E-05) was over-represented and transferring phosphorous-containing groups (Fisher’s P = 1.03E-04) underrepresented (S9B Data). NAD(P)-binding domain superfamily (IPR036291) was the most highly over-represented IPS ID (Fishers P = 1.63E-12) of 25, with mobidb-lite IPS ID (Fishers P = 5.58E-17) for Intrinsic disorder protein sequences, representing the most significantly underrepresented of 30 IPS IDs (S9C Data).

All MGA apart from Der p 28 were found in the MB, six were among the most abundant proteins identified (S3 Table). In the absence of Der p 28, nine Der p 28 homologs were identified including four isoallergens (Table 1 and S9 Data). MGA were also found to be amongst the most abundant proteins in the SM, including Der p 2 and Der p 14 (S4 Table). Allergenic/potentially allergenic molecules were highly represented in both MB and SM proteomes accounting for 11.4% and 18.5% of all proteins identified. Homologs of Der p 1 (n = 17), Der p 2 (n = 2) and Der p 23 (n = 1) were identified in the SM. Given that allergens Der p 1and Der p 2) account for up to 60% of IgE reactivity in HDM sensitised individuals [5], and Der p 23 sensitivity is seen in 79% of HDM allergic patients [79] examination of cross-reactivity between these MGA and excreted homologs is warranted.

MB and SM proteomes were abundant in enzymes, which accounted for 44.6% and 52.4% of all proteins identified respectively (S9 and S10 Datas), some of these enzymes may be involved in digestion. HDM have long been associated with feeding on shed skin present in HD. While they have been observed to eat skin, the poor nutritional value of keratin makes it unlikely to be a primary food source. Rather, HDM are trophic generalists, they feed on organic debris associated with their proximity to humans [16]. In the laboratory, D. pteronyssinus have been grown on diverse culture media including various combinations of wheat bran, wheat flour, dog food, rodent chow, ground porcine liver, dried egg powder, defatted skin scales, and fish food. Most research groups use dried yeast to supplement diets and improve mite population growth [72, 80–83]. HDM have been observed to feed on bacteria and fungi in laboratory experiments [84, 85]. Whether HDM consume bacteria or fungi in a wildtype setting as a nutrient source needs further experimental evidence. Expression and excretion of bacterial and fungal degrading enzymes may indicate a role in digestion [85].

Numerous glycoside hydrolases (EC:3.2.1) were identified in the predicted proteome (n = 57) and proteomic datasets (RP, MB & SM) summarised in Table 3. Each glycoside hydrolase enzyme sub-family were represented by at least one proteomic identification. Two predicted lysozymes (EC:3.2.1.17) and one 14.5 kDa bacteriolytic enzyme, Der p38 (DERPT_G10989) were identified with evidence for high expression, these enzymes may be responsible for bacteriolytic activity in HDM extracts [86]. Excreted proteins with predicted activities against major components of fungal cell walls were identified and include; chitinases (n = 5), four glycoside hydrolase family 16 members with putative β-1,3 glucanase activity, β-mannosidase (n = 1), α-mannosidases (n = 4), chitosanase (n = 1) and α-N-Acetyl hexosaminidaseine (n = 2). Carbohydrate metabolism GO terms were among the 318 over-represented GO terms (Fisher’s exact test < 0.05 FDR) in the excretome and included carbon utilization, hydrolase activity hydrolyzing O-glycosyl compounds, chitin metabolic process, chitin binding and starch metabolic process (S10A Data). Of the putative enzymes identified in the SM proteome (n = 457), enrichment analysis showed 35 to be over-represented, many of which related to carbohydrate digestion; Alpha-glucosidase, Chitinase and Alpha-mannosidase (S10B Data). The presence of a predicted secretion signal was a strong indicator of excretion as 22.5% of all proteins identified in the SM proteome contained predicted secretion signals. Moreover, secretion signal peptides were the most highly over-represented IPS ID (Fisher’s P = 1.08E-26) of 61 (S10C Data).

Download:

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

Table 3. Carbohydrate degrading enzymes.

https://doi.org/10.1371/journal.pone.0216171.t003

The expression of the numerous carbohydrate active enzymes listed above provides compelling evidence to support observations of D. pteronyssinus feeding on fungi and bacteria and demonstrate that they possess the necessary enzymes to utilise bacteria and fungi as a nutrient source [85]. Feeding on bacteria or fungi within wild-type microhabitats may alter allergen repertoire between homes, as diet has been demonstrated to alter allergen production in laboratory HDM cultures [87]. This new insight compounds the necessity for characterising factors affecting HDM allergen production within the home.

The process of chitin synthesis and remodelling is an integral part of the growth and development of all arthropods. Chitin remodelling enzymes include chitinase, β-N-Acetylhexosaminidase and the highly conserved chitin synthase, a key enzyme in the insect biosynthetic pathway [88]. Proteomic profiling of D. pteronyssinus airmid facilitated identification of eight predicted chitinases (EC:3.2.1.14, five β-N-Acetylhexosaminidases (EC:3.21.52) and two chitin synthases (IPR004835) (Table 3) putatively involved in chitin remodelling. These enzymes represent important biocontrol targets, as chitin is absent from vertebrates, dysregulation of these enzymes could provide a much needed method of curtailing HDM populations in the homes of sensitised individuals [12, 88].

Mite group allergen localisation

All proteins are synthesised in the MB prior to excretion, however excreted proteins are likely to accumulate in growth medium and HD. Data regarding localisation of MGA are limited [7]. Localisation is often linked to the degree of protein allergenicity, identifying sites of MGA accumulation may reveal trends of exposure that can be applied to assessing new allergens. In our analysis, the majority of MGA were detected in MB and SM proteomes, the relative amounts in each proteome was used to infer localisation. Proteomic assessment of localisation showed sero-dominant allergens Der p 1, Der p 2 and Der p 23 to accumulate in SM (Table 4), with Der p 1 and Der p 2 being the two most abundant proteins (S4 Table). Der p 23 has previously been reported to be found only in low quantities in SM relative to Der p 2 [79], we observed the same trend, more Der p 2 ms spectra (n = 486) were detected than for Der p23 (n = 90). Allergens Der p 3, Der f 6 like allergen, Der p 9, Der p 15 and Der p 28 were also found to accumulate in the SM (Table 4). Localisation of Der p 3 to SM is consistent with previous observations for Der f 3 [89], the serine peptidases Der p 3, Der f 6 like allergen and Der p 9 were all among the top 10 most abundant SM proteins (S4 Table). Despite a different methodological approach, another study also found Der p 1, Der p 2, Der p 6 and Der p 15 to be major proteins in D. pteronyssinus faeces and SM [69].

Download:

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

Table 4. Proteomic evaluation of mite group allergen localization in D. pteronyssinus airmid.

https://doi.org/10.1371/journal.pone.0216171.t004

Analysis of localisation must not be restricted to laboratory-based studies and should include environmental reference samples where possible as demonstrated for allergens Der p 5 and Der p 21. Initial proteomic assessment found Der p 5 and Der p 21 were not excreted under laboratory conditions as they were absent from SM (S10 Data). However, proteomic analysis of HD revealed Der p 5 to be among the top 10 most abundant proteins (Table 2) and Der p 21 the 63^rd most abundant, both were identified in at least 16 of the Der p 1 positive dust samples (S8 Data). The location of Der p 5 and Der p 21 in laboratory cultures, present in MB and absent from SM, would indicate they are not excreted, however, their presence in HD shows that there must be other factors that contribute to accumulation of non-excreted allergens in the home. These allergens may accumulate in HD as the mite bodies begin to degrade. The accumulation of dead mites in laboratory cultures is avoided by regular sub-culturing of mites. Cross-referencing SM with MB associated proteins as demonstrated in Table 4, can infer localisation and can allow for novel insights into modes of accumulation when compared to wild-type dust samples. Our work has demonstrated the utility of high sensitivity mass spectrometry in characterising the complex proteomes of D. pteronyssinus. Laboratory cultures show parallels in protein expression with wild-type samples, as most proteins were identified in two or more of the proteomes (Fig 3C) with only four proteins uniquely identified in HD. Additional research is required to characterise the various microhabitats of HDM and factors affecting allergen presence in the home.