Abstract
It is now well understood that thrombocytes (nucleated platelets) express TLRs and respond to both bacterial and viral products. Release of proinflammatory molecules can be expected following relatively short exposure times to LPS, lipoteichoic acid (LTA), thymidine homopolymer phosphorothioate oligonucleotide [Poly(dT)], and polyinosinic-polycytidylic acid [Poly(I:C)]. This study reports the varied expressions of genes encoded for components of the TLR, nucleotide binding oligomerization domain–like receptor, and retinoic acid-inducible gene RIG–like receptor signaling pathways in response to the TLR ligands listed above. Highly sensitive RNA-sequencing technologies were used to analyze the complete transcriptome of thrombocytes treated with all four microbial products for a period of 1 h. A total of 14,326 gene transcripts were found in chicken thrombocytes across all ligand exposures. After 1 h of stimulation with ligands, 87, 138, 1013, and 22 genes were upregulated for LTA, LPS, Poly(dT), and Poly(I:C), and 12, 142, 249, and 16 genes were downregulated for LTA, LPS, Poly(dT), and Poly(I:C), respectively, with at least a 1-fold change relative to unexposed thrombocytes. Summarizations of biological processes, protein classes, and biochemical pathways reveal the role of chicken thrombocytes in proinflammatory responses linked to key signaling pathways. TLR, nucleotide binding oligomerization domain–like receptor, and retinoic acid-inducible gene RIG-like receptor pathways were mapped based on the transcriptome results with gene expression for common signal and proinflammatory mediators highlighted. The information reported in this study is useful for defining a limited set of proinflammatory molecules to evaluate in cases of either bacterial or viral disease monitoring.
Introduction
Enucleated thrombocytes or platelets are only found in mammals; nucleated thrombocytes in lower vertebrates such as reptiles, amphibians, fish, and birds are functionally comparable and primarily involved in hemostatic functions and wound healing (1). The role of thrombocytes in immunity was shown with evidence of phagocytic ability, followed by a role in the inflammatory response. More recently, we have reported the transcriptomic profile of the avian thrombocyte in an effort to more thoroughly understand the range of function for this cell type. A total of 10,041 constitutively expressed genes were identified in the thrombocyte, and top among the biological processes and biochemical pathways associated with the transcriptomic profile were functions with defined immunological activity, especially for proinflammatory response (2). This information in conjunction with our previously reported results has clearly established the thrombocyte as an important immunological effector cell (2, 3).
Thrombocytes have been shown to express, produce, or release a variety of mediators of inflammation, antimicrobial activity, and other immune-modulating activities (3). The discovery of pathogen recognition receptors (PRRs), such as TLRs, on these cells has led to a new understanding of the thrombocyte’s role in immune responses (4–10). Thrombocytes respond to LPS, lipoteichoic acid (LTA), thymidine homopolymer phosphorothioate oligonucleotide [Poly(dT)], and polyinosinic-polycytidylic acid [Poly(I:C)] (8, 11), and this stimulation takes place through TLR signal pathways, specifically TLR4, TLR2, TLR7, and TLR3, respectively. Use of inhibitors has provided evidence to map LPS stimulation of thrombocytes to activation through mitogen-activated protein kinase (ERK, MEK1, and p38 MAPK) and NF-κ L-chain–enhancer of activated B cells (NF-κB) pathways (8, 12).
Our laboratory has studied in vitro stimulation of chicken thrombocytes with bacterial and viral TLR ligands for several years to establish the proper role of this cell in immunity (2, 8, 11, 13). In this study, we have used RNAseq technology to characterize the transcriptome of in vitro stimulated chicken thrombocytes in response to 1-h exposure to LPS, LTA, Poly(dT), and Poly(I:C). Using the information generated in this study, we were able to construct pathway maps for various common PRR-associated molecules in chicken thrombocytes when exposed to the four treatments. The long-term aim of this research is to devise a method by which to evaluate select proinflammatory responses of thrombocytes and platelets that provides a clear prognostic test for bacterial versus viral infectivity as well as bacterial and viral specific causes of infectivity (i.e., Gram-positive versus Gram-negative and ssRNA versus dsRNA, respectively). Furthermore, such methodology would be beneficial for both veterinary and human medical applications for rapid-test diagnostics to intervene with the onset of septicemia. Such solutions include the identification of proinflammatory biomarkers encoded in genes expressed by thrombocytes.
Materials and Methods
Chickens
Three female Single Comb White Leghorn chickens (16 wk old) were randomly selected for blood collection in this study. The chickens were housed at the Clemson University Morgan Poultry Center, Clemson, SC, which is an Institutional Animal Care and Use Committee–approved animal facility operating under standard management practices adhering to the Association for Assessment and Accreditation of Laboratory Animal Care International criteria.
Thrombocyte isolation and in vitro stimulation
Syringes fitted with needles were used to collect 3 ml of whole blood from the wing vein of each chicken into 0.1 ml of 10% EDTA solution. The collected blood samples were stored on ice until they were brought back to the laboratory. Each blood sample was diluted (1:1) with calcium- and magnesium-free HBSS (Cambrex Bio Science Walkersville, Walkersville, MD). Diluted blood samples were then layered on a lymphocyte separation medium (density 1.077–1.080 g/ml; Mediatech, Herdon, VA), and centrifuged at 1700 × g for 30 min at 23°C to collect the thrombocyte-rich band as previously described by Scott and Owens (8). The isolated thrombocyte-enriched cell suspension is routinely 99% positive for the thrombocyte-specific marker CD41/61 (10, 14). Trypan blue solution (0.4% w/v in normal saline) was used for quantification of viable cell numbers on a SPolite Hemacytometer (Baxter Healthcare, McGaw Park, IL) with the aid of an upright light microscope. The isolated thrombocytes from each chicken were incubated with 1 μg/ml of ultra-pure LPS from Salmonella minnesota, 400 μg/ml of purified LTA from Staphylococcus aureus, 400 μg/ml of Poly(I:C), and 50 μM of Poly(dT) (InvivoGen, San Diego, CA). The control samples were incubated with HBSS only. The cell suspensions with and without ligands were incubated in sterile 1.5 ml microcentrifuge tubes (1 × 107 cells per tube) on a rocking platform (VWR, Suwanee, GA) at 41°C for 60 min. The concentration of ligands used and stimulation length was chosen based on previous experiments performed in our laboratory.
RNA isolation, quantification, and quality assessment
For RNA isolation after thrombocyte stimulation, cells were centrifuged at 5000 × g for 2 min to pellet. The pellets were stored in 100 μl of RNAlater (Qiagen, Valencia, CA), an RNA stabilizing solution. After 24 h at 4°C in RNAlater, the cells were centrifuged again to remove the supernatant and stored at −20°C until thawed for RNA isolation. The RNeasy Kit (Qiagen, Valencia, CA) was used according to the manufacturer’s protocol to isolate the total RNA from these samples. The RNA samples were treated with an on-column DNase (Qiagen) to remove any possible contamination from chicken genomic DNA. Isolated RNA samples were quantified and integrity validated on a Nano Drop 1000 Spectrophotometer (Thermo Scientific, Waltham, MA) and Bioanalyzer 2100 (Agilent Technologies).
Illumina library construction, RNA sequencing, and analysis
Each thrombocyte sample was normalized to a standard input concentration (1 μg of total RNA) and an Illumina compatible sequencing library was prepared robotically on a Microlab STAR (Hamilton) with the TruSeq stranded total RNA library prep kit (Illumina) following the manufacturer’s recommended procedures. The resulting sequencing libraries were assessed for size on a 2100 Bioanalyzer (Agilent) and sequence data collected on one lane of an Illumina HiSeq2500, with a 2 × 125 bp PE read on high-output mode. Raw sequence reads were assessed for run quality with the FastQC analysis package (15), then preprocessed to remove adapter and low-quality bases with the Trimmomatic software package (16). Processed reads were mapped to the Gallus_gallus-4.0 reference assembly (GenBank Assembly identifier GCA_000002315.2) (17) with the Burrows Wheeler-Aligner (18) short read aligner. Read abundance counts per exon were determined with the Subread (19), and differential gene expression determined with EDGER (20). Gene Ontology (GO) enrichment and analysis was performed with the Panther suite of analytical tools (21). We used the Panther-derived GO-slim terms for broad classification of molecular function, biological processes, and cellular components (21, 22). De novo transcriptome assembly of the thrombocyte was performed with Trinity (23).
Construction of heat maps and pathways
Results
TLR agonist–stimulated thrombocyte transcriptomes
A total of 14,326 transcripts were detected for both unstimulated and TLR agonist–stimulated chicken thrombocytes compared with the 17,108 total annotated gene sequences identified in the reference chicken annotation. For analysis of similarities and differences between the four treatments the visualization tool Circos was used (26). The Circos diagram (Fig. 1) displays the up- and downregulated genes among the four ligands used to treat thrombocytes. Overlain on Fig. 1 is a Venn diagram displaying the number of total upregulated genes within, between, and among the four ligands inducing expression in treated thrombocytes. Poly(dT) invoked the greatest transcriptional response followed by LPS, LTA, and Poly(I:C). Overall, there were only six shared genes induced to ≥1.0-fold change above the untreated control cells for all four TLR ligands. In comparing upregulated gene expression between bacterial (LTA and LPS) and viral [Poly(dT) and Poly(I:C)] ligand treatments, there are 35 and 5 common genes, respectively, induced in treated thrombocytes. Furthermore, the numbers of different upregulated genes between LTA (6) and LPS (53) are large, but the difference between Poly(dT) (948) and Poly(I:C) (3) is extreme.
A Circos plot representing distribution of all the gene transcripts up- or downregulated thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists (A). The Venn diagram in the center represents the comparison of upregulated gene transcripts within, between, and among the TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). The outer ideograms represent the reference Gallus gallus chromosomes, followed by an annotation track representing annotated genes (those in green are transcribed in the clockwise direction and orange are reverse). The colored tracks depict fold-changes in gene expression as a result of Poly(dT), Poly(I:C), LTA, and LPS treatment, respectively. The bacterial (orange) and viral (blue) genes identified in this study as biomarkers are genes that are common in the top-10 expressed genes in thrombocytes stimulated with each of the four TLR agonists. These biomarkers are highlighted in the table (B), demonstrating the chromosomal location, TLR ligand that it is specific for, Ensembl identifiers, gene name, and functional description of the marker if available.
A Circos plot representing distribution of all the gene transcripts up- or downregulated thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists (A). The Venn diagram in the center represents the comparison of upregulated gene transcripts within, between, and among the TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). The outer ideograms represent the reference Gallus gallus chromosomes, followed by an annotation track representing annotated genes (those in green are transcribed in the clockwise direction and orange are reverse). The colored tracks depict fold-changes in gene expression as a result of Poly(dT), Poly(I:C), LTA, and LPS treatment, respectively. The bacterial (orange) and viral (blue) genes identified in this study as biomarkers are genes that are common in the top-10 expressed genes in thrombocytes stimulated with each of the four TLR agonists. These biomarkers are highlighted in the table (B), demonstrating the chromosomal location, TLR ligand that it is specific for, Ensembl identifiers, gene name, and functional description of the marker if available.
The Circos plot in Fig. 1A also indicates several potential gene transcripts that can be used as biomarkers for early detection of infection due to the exposure of pathogenic compounds such as LPS and LTA (bacterial), or similarly to Poly(dT) and Poly(I:C) (viral). These biomarkers are highlighted in Fig. 1B. There is a total of six common genes among the top 10 upregulated genes in the two bacterial TLR ligands (LPS and LTA), and 14 genes that are common in the two viral TLR ligands [Poly(dT) and Poly(I:C)] for treated thrombocytes. Some of these genes are specific to immunity (e.g., IL-6, IL-8, and IL-12) whereas others are related to general cellular function, and there are genes with an unknown function included in this list of biomarkers. Upon further investigation of the novel and unknown genes, functional description of some novel genes were found [such as bacterial permeability–increasing proteins, LPS-binding proteins, Plunc (palate, lung and nasal epithelium clone) family members, sulfotransferase, and atypical dual specificity protein] and putative gene names were found for some of the unknown genes that are shown in parentheses, such as C-C motif chemokine 5, and E74-like ETS transcription factor 5 (Fig. 1B). In addition to the list of biomarkers suggested in Fig. 1, which is based on genes that were common between TLR ligand treatments, the top 10 up- and downregulated genes for each ligand are presented in Supplemental Table I.
To decipher the functional aspects of the thrombocyte genes, we organized the transcripts based on GO functional categories. The results indicated that these cells have a role in a rather broad range of different biological processes (Supplemental Tables II–IV). Each functional category was compared between the four TLR ligands (Figs. 2–4). Within common biological processes, the most abundant upregulated functions detected (Fig. 2A) were biological adhesion, biological regulation, cellular component organization, cellular processes, developmental process, immune system processes, localization, locomotion, metabolic processes, multicellular organismal processes, and response to stimulus. Similar common biological processes are evident among downregulated genes with the absence of biological regulation, immune system process, and locomotion (Fig. 2B). Among the four TLR ligands, Poly(dT) treatment of thrombocytes produced the largest number of both up- and downregulated genes for all biological process categories reported. It is not that the other three ligands did not induce or reduce expression of many genes, but rather the numbers of genes associated with Poly(dT) exposure were very high at over 100 upregulated genes for biological regulation, cellular processes, developmental processes, metabolic processes, and response to stimulus. There were 100 or more genes downregulated under each cellular process and metabolic process with Poly(dT) treatment. LPS was second in transcript abundance changes, whereas LTA and Poly(I:C) invoke more subtle transcriptome responses in either direction for most all the biological processes.
The distribution of gene transcripts according to the GO-slim categories of biological processes (BPs) in chicken thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Only the common upregulated (A) and downregulated (B) BPs among the four treatments are displayed. The complete list of individual BPs for all four TLR agonists is shown in Supplemental Table II.
The distribution of gene transcripts according to the GO-slim categories of biological processes (BPs) in chicken thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Only the common upregulated (A) and downregulated (B) BPs among the four treatments are displayed. The complete list of individual BPs for all four TLR agonists is shown in Supplemental Table II.
The distribution of gene transcripts according to the GO-slim categories of protein classes in chicken thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Only the common upregulated (A) and downregulated (B) protein classes among the four treatments are displayed. The complete list of individual protein class for all four TLR agonists is shown in Supplemental Table III.
The distribution of gene transcripts according to the GO-slim categories of protein classes in chicken thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Only the common upregulated (A) and downregulated (B) protein classes among the four treatments are displayed. The complete list of individual protein class for all four TLR agonists is shown in Supplemental Table III.
The distribution of gene transcripts according to the GO-slim categories of common biochemical pathways in chicken thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Only the common upregulated biochemical pathways among the four treatments are displayed. The complete list of individual biochemical pathways for all four TLR agonists is shown in Supplemental Table IV.
The distribution of gene transcripts according to the GO-slim categories of common biochemical pathways in chicken thrombocytes in response to 1-h stimulation with bacterial or viral TLR agonists LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Only the common upregulated biochemical pathways among the four treatments are displayed. The complete list of individual biochemical pathways for all four TLR agonists is shown in Supplemental Table IV.
The categorical lists of common protein classes representing the most genes both up- (Fig. 3A) and downregulated (Fig. 3B) for all four agonists contain some identifiers related to signaling through pathways linked to TLR. There are protein classes for cell adhesion molecules, cytoskeletal proteins, defense or immunity proteins, enzyme modulators, hydrolases, and transferase for upregulated genes. Among the downregulated genes identified, protein classes such as cell adhesion molecules, enzyme modulators, extracellular matrix proteins, nucleic acid binding, receptors, transcription factors, and transferases are displayed in the figure. In four of six protein classes shown for upregulated genes, there are >40 genes expressed with Poly(dT) treatment of the cells whereas the other two classes are composed of 20 and 21 genes, respectively. Overall, LPS induced the second largest number of genes to be upregulated. LTA and Poly(I:C) both affected upregulation of <10 genes for different protein classes in thrombocytes. There were seven protein classes with downregulated gene expression across all four ligand treatments. Poly(dT) exposure of thrombocytes resulted in far greater numbers of downregulated genes for four of the protein classes presented in Fig. 3B. The other three protein classes varied in downregulated gene numbers by ligand treatment.
When gene numbers for common biochemical pathways were sorted, there was only one pathway (integrin signaling) that was significantly downregulated for all four TLR agonists with just one gene. The pathways with significant upregulated genes are angiogenesis, blood coagulation, CCKR signaling map, inflammation mediated by chemokine and cytokine signaling, and IL signaling (Fig. 4). Again, Poly(dT) treatment produced the most upregulated genes across all listed pathways, whereas LTA and LPS gave the second most, and Poly(I:C) the fewest.
Genes and pathways connected with proinflammatory immune function
For the heat maps, the fold-change expression data were uploaded to generate side-by-side graphics to demonstrate the relative magnitude of gene expression for all identifiable components of the TLR signal pathway (Fig. 5A). In addition to TLR signaling, we detected gene transcripts for nucleotide binding oligomerization domain–like receptor (NLR; including inflammasome) and retinoic acid-inducible gene 1–like receptor (RLR) pathways, which share very important signal components with the TLR pathway. Therefore, heat maps for NLR, inflammasome, and RLR pathways were also constructed for each TLR ligand used for thrombocyte treatment, and color-coded for expression levels (Fig. 5B, 5C).
Heat maps displaying up- and downregulation of common gene transcripts associated with PRRs in chicken thrombocyte treated with LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Molecules associated with TLRs, NLRs (including inflammasome), and RLR pathways are displayed in (A)–(C) respectively. The heat maps were generated using the HeatmapGenerator (24).
Heat maps displaying up- and downregulation of common gene transcripts associated with PRRs in chicken thrombocyte treated with LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). Molecules associated with TLRs, NLRs (including inflammasome), and RLR pathways are displayed in (A)–(C) respectively. The heat maps were generated using the HeatmapGenerator (24).
Inspection of the TLR heat map (Fig. 5A) reveals that each agonist induces its own unique pattern of gene expression for the various components of the pathway, respective to the initiating TLR of the cell. Overall, Poly(dT) induced gene expression for the TLR pathway very differently to the other three agonists (Fig. 5A). Although differential gene expression patterns were observed between the four treatment groups, RIPK3 appeared to be upregulated in all four treatments in the heat map for NLR and inflammasome (Fig. 5B). The gene expression pattern between the two bacterial and two viral treatments appeared similar when compared with the overall bacterial and viral products. In the case of the RLR heat map, the Poly(I:C) gene expression profile appeared to be more similar to the two bacterial treatments when compared with Poly(dT) (Fig. 5C).
The expressions of genes for downstream proinflammatory responses are presented in Table I, which are shown with the log2 fold-change (and p values). Based on the expected gene products resulting from signaling via the four proinflammatory pathways, it is evident that the array of gene expressions attributable to the respective TLR ligands are acting through the receptors of thrombocytes. These expression values serve to quantify the color code for the pathway signaling found in Fig. 6, whereas many more thrombocyte genes expressed outside these specific pathways are influenced by ligand treatment as well.
Category . | Gene ID . | Ensembl ID . | LogFC (p value) . | |||
---|---|---|---|---|---|---|
LTA . | LPS . | Poly(dT) . | Poly(I:C) . | |||
Proinflammatory cytokine | IL1B | ENSGALG00000000534 | 4.99 (3.93 × 10−33) | 6.54 (1.35 × 10−48) | 0.92 (0.01) | 3.43 (1.07 × 10−18) |
IL6 | ENSGALG00000010915 | 4.81 (2.50 × 10−9) | 9.63 (2.36 × 10−36) | −3.25 (0.12) | 1.44 (1.85) | |
IL12B | ENSGALG00000001409 | 5.94 (4.83 × 10−9) | 4.97 (4.14 × 10−7) | −1.04 (0.37) | 0.97 (1.04) | |
IL18 | ENSGALG00000007874 | 0.21 (0.41) | 0.36 (0.16) | −0.40 (0.12) | 0.12 (0.24) | |
Colony factor | CSF3 | ENSGALG00000026420 | 6.14 (1.16 × 10−4) | 9.31 (1.03 × 10−11) | 5.75 (6.05 × 10−4) | 4.01 (3.13 × 10−2) |
Chemokine | IL8L1 | ENSGALG00000011668 | 5.73 (2.75 × 10−19) | 7.24 (8.55 × 10−27) | 0.12 (0.83) | 2.41 (3.45 × 10−5) |
IL8L2 | ENSGALG00000026098 | 5.27 (3.10 × 10−22) | 6.73 (2.91 × 10−31) | 1.81 (1.78 × 10−4) | 2.94 (2.70 × 10−9) | |
CCL20 | ENSGALG00000003003 | 1.11 (1.71 × 10−3) | 4.80 (3.05 × 10−33) | −2.17 (7.40 × 10−8) | 0.39 (1.25) | |
Costimulatory molecule | CD40 | ENSGALG00000007015 | 0.99 (1.17 × 10−4) | 1.84 (1.27 × 10−12) | 0.18 (0.47) | 0.50 (0.04) |
CD80 | ENSGALG00000015474 | 0.49 (0.10) | 0.19 (0.51) | 0.77 (0.01) | 0.17 (0.32) | |
CD86 | ENSGALG00000014362 | 0.93 (0.33) | 1.50 (0.11) | 1.46 (0.12) | −0.48 (0.22) |
Category . | Gene ID . | Ensembl ID . | LogFC (p value) . | |||
---|---|---|---|---|---|---|
LTA . | LPS . | Poly(dT) . | Poly(I:C) . | |||
Proinflammatory cytokine | IL1B | ENSGALG00000000534 | 4.99 (3.93 × 10−33) | 6.54 (1.35 × 10−48) | 0.92 (0.01) | 3.43 (1.07 × 10−18) |
IL6 | ENSGALG00000010915 | 4.81 (2.50 × 10−9) | 9.63 (2.36 × 10−36) | −3.25 (0.12) | 1.44 (1.85) | |
IL12B | ENSGALG00000001409 | 5.94 (4.83 × 10−9) | 4.97 (4.14 × 10−7) | −1.04 (0.37) | 0.97 (1.04) | |
IL18 | ENSGALG00000007874 | 0.21 (0.41) | 0.36 (0.16) | −0.40 (0.12) | 0.12 (0.24) | |
Colony factor | CSF3 | ENSGALG00000026420 | 6.14 (1.16 × 10−4) | 9.31 (1.03 × 10−11) | 5.75 (6.05 × 10−4) | 4.01 (3.13 × 10−2) |
Chemokine | IL8L1 | ENSGALG00000011668 | 5.73 (2.75 × 10−19) | 7.24 (8.55 × 10−27) | 0.12 (0.83) | 2.41 (3.45 × 10−5) |
IL8L2 | ENSGALG00000026098 | 5.27 (3.10 × 10−22) | 6.73 (2.91 × 10−31) | 1.81 (1.78 × 10−4) | 2.94 (2.70 × 10−9) | |
CCL20 | ENSGALG00000003003 | 1.11 (1.71 × 10−3) | 4.80 (3.05 × 10−33) | −2.17 (7.40 × 10−8) | 0.39 (1.25) | |
Costimulatory molecule | CD40 | ENSGALG00000007015 | 0.99 (1.17 × 10−4) | 1.84 (1.27 × 10−12) | 0.18 (0.47) | 0.50 (0.04) |
CD80 | ENSGALG00000015474 | 0.49 (0.10) | 0.19 (0.51) | 0.77 (0.01) | 0.17 (0.32) | |
CD86 | ENSGALG00000014362 | 0.93 (0.33) | 1.50 (0.11) | 1.46 (0.12) | −0.48 (0.22) |
ID, identifier; LogFC, log2 fold change.
Pathways associated with PRR in chicken thrombocytes treated with LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). LTA stimulates TLR 2 (A), LPS stimulates TLR4 (B), Poly(dT) stimulates TLR7 (C), and Poly(I:C) stimulates TLR3 (D). In addition to the specific TLR that each ligand binds, the pathway figure includes other PRR-associated molecules for TLRs, NLRs (including inflammasome), and RLRs. Gene expression in each case is color-coded to show upregulation (red, fold-change ≥0.5), unchanged (purple, fold-change >−0.5 to <0.5), and downregulation (blue, fold-change ≤−0.5) for receptors, pathway components, and proinflammatory products. These pathways were constructed using the PathVisio software (25).
Pathways associated with PRR in chicken thrombocytes treated with LPS, LTA, thymidine homopolymer Poly(dT), and Poly(I:C). LTA stimulates TLR 2 (A), LPS stimulates TLR4 (B), Poly(dT) stimulates TLR7 (C), and Poly(I:C) stimulates TLR3 (D). In addition to the specific TLR that each ligand binds, the pathway figure includes other PRR-associated molecules for TLRs, NLRs (including inflammasome), and RLRs. Gene expression in each case is color-coded to show upregulation (red, fold-change ≥0.5), unchanged (purple, fold-change >−0.5 to <0.5), and downregulation (blue, fold-change ≤−0.5) for receptors, pathway components, and proinflammatory products. These pathways were constructed using the PathVisio software (25).
Graphic diagrams of TLR, NLR, inflammasome, and RLR pathways are provided in Fig. 6 using information from Fig. 5 and Table I. Pathway diagrams for LTA and Poly(I:C) (Fig. 6A, 6D) present gene expression on a different range and scale to the heat maps (Fig. 5). There is minimal up- or downregulation of most pathway components except for upregulation of few members of the TLR, NLR, and RLR pathways, and not in the inflammasome for LTA and Poly(I:C). LPS and Poly(dT) affect gene expression more prominently in both the up- and downregulated directions for TLR, NLR, and RLR pathways (Fig. 6B, 6C), with little to no effect on the inflammasome. Proinflammatory cytokines are upregulated with LTA, LPS, and Poly(I:C) treatment, whereas most genes for chemokines and the CSF3 gene are upregulated by all four ligands. Of unique interest is the expression of genes for costimulatory molecules typically found on professional APCs. LTA and LPS treatments upregulated the expression of genes for CD40 and CD86, whereas Poly(dT) treatment resulted in higher gene expression for CD86. Poly(I:C) did not affect CD40, CD80, or CD86.
Discussion
Experimentation with the chicken thrombocyte is proving to be a valuable resource in understanding the proinflammatory responsiveness of a cell type typically viewed as involved in blood clotting alone. The phagocytic capacity of this cell has been examined by several researchers over the past 50 y, but we have only recently discovered that the proinflammatory capacity of the thrombocyte is quite similar to platelets. Uniquely different is that the thrombocyte is nucleated and gene expression is inducible within minutes of exposure to bacterial and viral products (i.e., TLR agonists and ligands) resulting in production of reactive inflammatory molecules (e.g., cytokines) (13).
Based on our experience in working with thrombocytes and previous studies conducted in our laboratory (8, 10, 11, 13), stimulation with these TLR ligands for 1 h is more than sufficient to induce these cells to produce proinflammatory responses. However, those past assessments of thrombocytes were focused on assay systems to detect innate immunity. With the full transcriptome approach of this report, we have been able to examine a greater range of gene expressions induced by the ligands. Genes associated with a variety of biological processes, protein classes, and biochemical pathways have been demonstrated to be up- and downregulated (Figs. 2–4). It is quite evident that proinflammatory reactivity is at the top of the list of responses, but many other genes outside of immunity alone are found to be highly expressed (Supplemental Table I). This is also noticeable in the list of suggested biomarkers for early detection of bacterial or viral infection (Fig. 1). The most upregulated gene transcripts with LTA treatment are a combination of cytokine and a few receptors. LPS-induced gene expression is represented by cytokines as well as receptors, enzymes, and regulatory proteins. Exposure of thrombocytes to Poly(dT) produced upregulation of many genes encoded for similar types of proteins observed with LPS treatment, however, certain cytokines such as IL-6 and -12 are downregulated. A shorter list of genes is designated with Poly(I:C) stimulation.
Given that we chose to treat thrombocytes with TLR ligands, it was more than apparent that an examination of gene expression for components of signaling pathways leading to proinflammatory gene expression from TLR 2, 3, 4, and 7 to the nucleus would be mapped. Generally, pathogen-associated molecular patterns (PAMPs) initiate TLRs to activate different transcription factors (e.g., AP1, NF-κB, IFN regulatory factors), and the expression of proinflammatory cytokines (e.g., IL-6, IL-1β, and IL-12), and innate anti-viral genes such as type I IFNs (IFN-α and IFN-β). Avian TLR 2 and 4 are involved in recognition of bacterial components (e.g., glycolipids, peptidoglycan, LTA, and lipoproteins) whereas TLR 3 and 7 function in recognition of RNA viral products (e.g., double stranded RNA, imiquimod, and single stranded RNA) (27, 28). The results of the RNAseq procedure were analyzed and aligned with a known database to find the representation of up- and downregulated genes for signaling pathways containing the highest numbers of genes influenced by the ligand treatments of thrombocytes. Thorough inspection of the TLR-associated pathway (Fig. 6) reveals that each ligand induces its own unique pattern of gene expression for the various component members of the pathway respective to the initiating TLR of the cell, which is indicative of bacterial versus viral product ligands as well as the MyD88-dependent and independent pathway distinction. Poly(I:C) is just one of the ligands that binds to TLR 3 and transduces signaling via the MyD88-independent pathway. Secondarily, the Poly(I:C) heat map rankings differ from the two bacterial ligands (LTA and LPS), and between the latter two there are ranking differences and similarities. For the most part, Poly(dT) induces a far different level of gene expression but the relationships among the various genes are not clustered in such a way as to result in a greater emphasis on one area of the pathway than another.
Among all the gene transcripts detected, GO-slim clustering of biological processes showed upregulation of 18, 13, 57, and 4 for LPS, LTA, Poly(dT), and Poly(I:C), respectively, for genes related to immune system processes. Other biological processes that ranked high in our assessments are those expected to be influenced by TLR ligand treatment, which provides evidence for thrombocytes as important immune cells. Among all biochemical pathways revealed by GO-slim analysis (Supplemental Table IV), the greatest numbers of genes up- (417) and downregulated (113) were associated with cellular processes during Poly(dT) treatment. Among all protein classes revealed by GO-slim analysis (Supplemental Table III), the greatest numbers of genes up- (89) and downregulated (51) were associated with nucleic acid binding during Poly(dT) treatment. Overall, for the single-stranded viral analog that is an agonist for TLR 7, Poly(dT) appeared to influence gene expression in thrombocytes in all functional categories.
In addition to genes related to TLR signaling, we detected transcripts for NLR, inflammasome, and RLR. Similar to pathogen recognition in mammals, TLRs, NLRs, inflammasomes, and RLRs are involved in the innate pathogen pattern recognition in birds (27–30). Whereas TLRs detect PAMPs either on the cell surface or in the lumen of intracellular vesicles, NLRs and RLRs detect intracellular PAMPs in cytosol. The NLRs have been shown to respond to intracellular pathogens and play important roles in distinct biological processes ranging from the regulation of Ag presentation, sensing metabolic changes in the cell, modulation of inflammation, embryo development and cell death, to the differentiation of the adaptive immune response (31). Inflammasomes are multiprotein complexes that consist of caspase-1, an apoptosis-associated speck-like protein containing a caspase-associated recruitment domain, and an upstream activator, such as an NLR, which when activated converts IL-1β and IL-18 from their immature proforms (32). The RLRs are generally involved in the recognition of viral RNA that results in the production of IFNs (33). In this study, molecules associated with PRR activation were up- or downregulated depending on the treatment. However, IFN gene transcripts were detected at very low expression levels inconsistently among the samples across animals and treatments. Minimal counts for IFN-α were uncovered in two Poly(I:C)-treated samples by data sifting.
Many molecules in the NLR and inflammasome pathway that are common in all four treatments have been detected (Fig. 5B) in this study and used in the construction of the PRR pathways (Fig. 6) in chicken thrombocytes. Upon careful examination of the pattern of common genes expressed in the PRR pathways in thrombocytes, there were noticeable differences in patterns of gene expression between bacterial and viral treatment. Both LPS and LTA, the two bacterial products, resulted in upregulation of proinflammatory cytokines, chemokines, and costimulatory molecules. Poly(I:C), the synthetic analog of double-stranded RNA virus, showed an expression profile of these PRR-associated molecules that was more similar to those of LPS and LTA (Fig. 6D). However, the array of gene expression for the same PRR-associated molecules for Poly(dT) was relatively different. There was downregulation of cytokines such as IL-6, IL-12, and chemokines such as chemokine (C-C motif) CCL20, and upregulation of an RLR family member, melanoma differentiation–associated gene 5 (MDA5), that were not observed in the other treatments (Fig. 6C). RLRs such as RIG-I and MDA5 have been shown to recognize distinctly different viral and synthetic RNA including Poly(I:C) depending on the length of the nucleotide strand (33). Therefore, the model for the RLR pathway presented in this study may change depending on type of virus or viral product that enters the thrombocyte cytosol. The same argument may be applicable to the other model PRR pathways presented in this study depending on the type of PAMP. In addition, this model is constructed based on the response of thrombocytes at 1 h of TLR agonist treatments; the response may differ at other time points.
In our previous work with these four ligands, we documented gene expression through quantitative real-time PCR for IL6 and NOS2 (13), which limited our ability to understand the full range of changes occurring in thrombocytes stimulated by LTA, LPS, Poly(dT), and Poly(I:C). With RNAseq of both untreated and treated cells, we not only observe >14,000 gene expressions, but are able to map the varied influences on the activity of the thrombocyte. In all, there is a greater influence on processes and pathways linked to proinflammatory responses. Additional data in the transcriptomic dataset implies a more specialized role of this cell in Ag processing and presentation, which is being investigated to establish the assistance that thrombocytes could offer to adaptive immunity. In addition, we were able to suggest potential gene transcripts that can be used as biomarkers for early detection of infection due to exposure of pathogenic compounds similar to the TLR ligands used in this study.
In conclusion, to our knowledge, this is the first representation of the various PRRs in chicken thrombocytes when exposed to four different PAMPs. There are genes highly relevant to activating other immune cells, as seen among the highest expressed genes for bacterial versus viral influences. Our report on thrombocyte transcriptomics pointed out the nature of this cell as an immune effector type (2), and the results presented in this study further document the role of the thrombocyte as having far-reaching effects on immunity for both bacterial and viral infection. It is envisaged that growing evidence for a specialized role for thrombocytes outside of blood clotting and phagocytosis will reveal a more centralized position for thrombocytes (and platelets) in immunity.
Footnotes
This work was supported by both internal and external funds secured through Clemson University.
The online version of this article contains supplemental material.
Abbreviations used in this article:
References
Disclosures
The authors have no financial conflicts of interest.