The zebrafish (Danio rerio) is a powerful model organism for studies of the innate immune system. One apparent difference between human and zebrafish innate immunity is the cellular machinery for LPS sensing. In amniotes, the protein complex formed by TLR4 and myeloid differentiation factor 2 (Tlr4/Md-2) recognizes the bacterial molecule LPS and triggers an inflammatory response. It is believed that zebrafish have neither Md-2 nor Tlr4; Md-2 has not been identified outside of amniotes, whereas the zebrafish tlr4 genes appear to be paralogs, not orthologs, of amniote TLR4s. We revisited these conclusions. We identified a zebrafish gene encoding Md-2, ly96. Using single-cell RNA sequencing, we found that ly96 is transcribed in cells that also transcribe genes diagnostic for innate immune cells, including the zebrafish tlr4-like genes. In larval zebrafish, ly96 is expressed in a small number of macrophage-like cells. In a functional assay, zebrafish Md-2 and Tlr4ba form a complex that activates NF-κB signaling in response to LPS. In larval zebrafish ly96 loss-of-function mutations perturbed LPS-induced cytokine production but gave little protection against LPS toxicity. Finally, by analyzing the genomic context of tlr4 genes in 11 jawed vertebrates, we found that tlr4 arose prior to the divergence of teleosts and tetrapods. Thus, an LPS-sensitive Tlr4/Md-2 complex is likely an ancestral feature shared by mammals and zebrafish, rather than a de novo invention on the tetrapod lineage. We hypothesize that zebrafish retain an ancestral, low-sensitivity Tlr4/Md-2 complex that confers LPS responsiveness to a specific subset of innate immune cells.

Amniote innate immune systems are exquisitely sensitive to LPS, a component of the cell wall in Gram-negative bacteria (13). LPS is sensed by a protein complex composed of TLR4 (Tlr4) and Md-2 (also known as LY96 and ESOP-1) (1, 4). LPS binds in a pocket of Md-2, triggering dimerization of the Tlr4/Md-2 complex (Fig. 1) (5). This, in turn, activates a Myd88-dependent NF-κB response (6). When properly regulated, the LPS activation of Tlr4/Md-2 regulates microbiome populations (7), recruits neutrophils to sites of infection (8), and induces angiogenesis (9). When dysregulated, Tlr4/Md-2 activity induces septic shock (10, 11), plays roles in inflammatory disorders (11, 12), and is a key player in the tissue remodeling that accompanies tumorigenesis (13, 14).

The role of Tlr4/Md-2 in LPS sensing outside of amniotes remains poorly understood. Understanding this response in zebrafish (Danio rerio) is of particular interest as the zebrafish is a powerful model organism for studies of vertebrate innate immunity (15). Zebrafish have mature genetic resources, rapid generation time, clear embryos, and facile germ-free derivation (16, 17). The zebrafish is increasingly popular as a model for understanding host–microbe interactions (18) as well as a tool to understand the development of the innate immune system (16).

The zebrafish response to LPS is puzzling (19, 20). In some ways, it is similar to amniotes. As in amniotes, LPS triggers the expression of Myd88-dependent genes (2123). Further, the expression patterns of genes induced by LPS stimulation are highly similar between mouse and zebrafish (24). There are, however, several lines of evidence that suggest Tlr4/Md-2 is not involved. Most critically, the gene encoding the essential coreceptor Md-2 has not been identified in zebrafish and other ray-finned fishes (19, 20, 25). Further, zebrafish Tlr4 proteins do not activate NF-κB in response to LPS in ex vivo assays, even when complemented with a mouse or human Md-2 (19, 20). Finally, zebrafish do not have a direct ortholog to amniote tlr4. Rather, they possess three tandem tlr4-like genes (tlr4ba, tlr4bb, and tlr4al) that are thought to have arisen from an ancestral TLR lost in tetrapods but retained in ray-finned fishes (20). These observations have led to the hypothesis that zebrafish respond to LPS by a non–Tlr4/Md-2–dependent pathway.

We set out to carefully revisit these conclusions using resources unavailable when the initial investigations of zebrafish Tlr4 were performed. Using careful bioinformatics, we found an ortholog of the gene encoding Md-2 (ly96) in zebrafish and other ray-finned fishes. When cotransfected into mammalian cells, the zebrafish ly96 and tlr4ba genes activate NF-κB signaling in response to LPS. Single-cell RNA sequencing (RNA-Seq) experiments on larval zebrafish revealed that the gene is expressed in a small subset of cells that express the zebrafish tlr4-like genes and the macrophage-specific gene mpeg1.1 (26, 27). Physiologically, zebrafish larvae with loss-of-function ly96 mutations exhibited perturbed cytokine production in response to LPS but were not protected from LPS toxicity. Finally, we revisited the history of the tlr4 gene in zebrafish, finding that formation of an LPS-sensitive Tlr4/Md-2 complex is likely an ancestral feature shared by mammals and zebrafish, rather than a de novo invention on the tetrapod lineage. We hypothesize that zebrafish preserve an ancestral, low-sensitivity Tlr4/Md-2 complex that may play an LPS sensing role in a small population of innate immune cells.

We constructed curated databases of Md-1, Md-2, Tlr4, and Cd180 protein sequences from across the vertebrates. Cd180 and Md-1 are paralogs of Tlr4 and Md-2, respectively (28). We obtained amino acid sequences of these proteins from National Center for Biotechnology Information, Ensembl, Fish1TK, amphibian transcriptomes (2932), UniProt, and Zebrafish Information Network. We constructed a multiple sequence alignment for Tlr4 and Cd180 and for Md-2 and Md-1 using MSAProbs (33), followed by manual editing in MEGA (34). We manually trimmed the alignment to remove highly variable (and therefore unalignable) regions. We used PHYML (35, 36) with subtree pruning and regrafting to construct the maximum likelihood phylogeny. Pilot analyses revealed that the Jones–Taylor–Thornton substitution model with eight rate categories and a floating γ distribution parameter yielded the highest likelihood trees (3739). An Akaike information criterion test was used to control for overfitting (40). We rooted our trees at the duplication of these proteins in early vertebrates. Alignment figures in Supplemental Figs. 2 and 3 were made with JalView (41). All sequences and alignments are available for download at https://github.com/harmslab/vertebrate-tlr4-evolution.git.

For the ly96 synteny analysis, we used the Ensembl synteny module (42) to map homologs onto the chromosomes of species of interest. For the tlr4 synteny analysis, we took the 22 genes flanking human TLR4 (11 on each side) and the 22 genes flanking zebrafish tlr4. We used tblastn with default parameters as our basic local alignment search tool (BLAST) to search for these sequences against 11 vertebrate genomes. We discarded all hits with e value >0.001 and then calculated a running average of the log (e value) along each chromosome with a sliding window of 10,000 bases. Finally, we divided this running average by the maximum observed log (e value)/bp value across all genomes. This value occurs for the window centered on the zebrafish tlr4 gene. On the final relative scale, 0.0 indicates no hits observed in a given window, and 1.0 is the maximum e value per bp. The complete analysis pipeline is implemented in a collection of shell scripts and jupyter notebooks (https://github.com/harmslab/vertebrate-tlr4-evolution.git).

Whole 6 d postfertilization (dpf) zebrafish were euthanized by tricaine methane sulfonate overdose, flash frozen in 1 ml of TRIzol (Ambion), thawed, and homogenized. Chloroform (200 μl) was added to each tube followed by mixing, centrifugation at 12,000 × g for 10 min at 4°C, transfer of the aqueous phase to a separate tube, addition of 200 μl ethanol, and binding of sample to an RNeasy Mini Kit column (QIAGEN). RNA was washed and eluted according to the manufacturer’s instructions and treated with RQ1 DNase (Promega). RNA was reverse transcribed into cDNA using Superscript II Reverse Transcriptase (Invitrogen) and an oligodeoxythymidine (19) primer and then amplified by PCR using gene-specific primers for zebrafish ly96 (5′-TGTATGGCATCTGAGAAAGCAGA-3′ and 5′-AAGAGCAGGGGGAAACAGTC-3′) and the housekeeping gene b2m (5′-ACGCTGCAGGTATATTCATC-3′ and 5′-TCTCCATTGAACTGCTGAAG-3′). PCR products were separated by electrophoresis on a 6% bis-acrylamide (19:1) gel that was stained with 1× SYBR Green 1 Nucleic Acid Gel Stain (Invitrogen) and imaged using an AlphaImager HP (Alpha Innotech). The identity of the ly96 RT-PCR product was verified by Sanger sequencing.

Single-cell analysis of transcription patterns of ly96, tlr4ba, tlr4bb, and tlr4al was performed using the recently released Zebrafish Single-Cell Transcriptome Atlas (43). Briefly, dissociated cells were run on a 10× Chromium platform using v2 chemistry. Dissociated samples for each time point (1, 2, and 5 dpf) were submitted in duplicate to determine technical reproducibility. The resulting cDNA libraries were sequenced on either an Illumina HiSeq or an Illumina NextSeq. The resulting sequencing data were analyzed using the 10× Cellranger pipeline version 2.2.0 (44) and the Seurat software package for R v3.4.4 (45, 46) using standard quality control, normalization, and analysis steps. We aligned reads to the zebrafish genome, GRCz11_93, and counted the expression of protein coding reads. The resulting matrices were read into Seurat, in which we performed principal component analysis and uniform manifold approximation and projection analysis on the resulting dataset with 178 dimensions and a resolution of 13.0, which produced 220 clusters and one singleton. Differential gene expression analysis was performed using the FindAllMarkers function in Seurat and Wilcoxon rank-sum test.

Mammalian expression vectors containing human TLR4 and mouse Tlr4 were obtained from Addgene (https://www.addgene.org/; no. 13085 and no. 13086), originally deposited by Ruslan Medzhitov. Human CD14 and ELAM-Luc were also obtained from Addgene (https://www.addgene.org/; no. 13645 and no. 13029), originally deposited by Doug Golenbock. Human MD-2 was obtained from the DNASU Repository (https://dnasu.org/; HsCD00439889) and contains a C-terminal V5-tag. Mouse Md-2 (UniProt; https://www.uniprot.org/; no. Q9JHF9) and Cd14 (UniProt; https://www.uniprot.org/; no. P10810), opossum Md-2 (UniProt; https://www.uniprot.org/; no. F6QBE6) and Cd14 (National Center for Biotechnology Information; https://www.ncbi.nlm.nih.gov/; no. XP_007473804.1), and chicken Md-2 (UniProt; https://www.uniprot.org/; no. A0A1D5NZX9) and Cd14 (UniProt; https://www.uniprot.org/; no. B0BL87) were designed to be free of restriction sites, codon optimized for human expression, and purchased as mammalian expression vector constructs in pcDNA3.1 (+) from GenScript. Zebrafish tlr4ba (ensembl; https://ensembl.org; ENSDART00000044697.6) and ly96 (G) were also obtained from GenScript in pcDNA3.1 (+). Zebrafish tlr4bb was a gift from C. Kim. We recloned this protein from its original vector into pcDNA3.1 (+) to limit variability in expression because of differences in vector size and promoter.

Human embryonic kidney cells (HEK293T/17, American Type Culture Collection CRL-11268) were maintained up to 30 passages in DMEM supplemented with 10% FBS at 37°C with 5% CO2. For each transfection, a confluent 100-mm plate of HEK293T/17 cells was treated at room temperature with 0.25% Trypsin–EDTA in HBSS and resuspended with an addition of DMEM + 10% FBS. This was diluted 4-fold into fresh medium, and 135 μl aliquots of resuspended cells were transferred to a 96-well cell culture–treated plate. Transfection mixes were made with 10 ng of tlr4, 1 ng of cd14, 10 ng of ly96, 1 ng of Renilla, 20 ng of ELAM-Luc, and 58 ng of pcDNA3.1 (+) per well for a total of 100 ng of DNA, diluted in OptiMEM to a volume of 10 μl/well. To the DNA mix, 0.5 μl per well of PLUS reagent was added followed by a brief vortex and room temperature incubation for 10 min. Lipofectamine was diluted 0.5 μl into 9.5 μl OptiMEM per well. This was added to the DNA + PLUS mix, vortexed briefly, and incubated at room temperature for 15 min. The transfection mix was diluted to 65 μl/well in OptiMEM and aliquoted onto a plate. Cells were incubated with transfection mix overnight (20–22 h) and then treated with LPS. Escherichia coli K-12 LPS (tlrl-eklps; InvivoGen) was dissolved at 5 mg/ml in endotoxin-free water, and aliquots were stored at −20°C. To avoid freeze–thaw cycles, working stocks of LPS were prepared at 10 μg/ml and stored at 4°C. To prepare treatments, LPS was diluted in 25% PBS and 75% DMEM. Cells were incubated with treatments for 4 h. The Dual-Glo Luciferase Assay System (Promega) was used to assay firefly and Renilla luciferase activity of individual wells. Each NF-κB induction value shown represents the firefly luciferase activity/Renilla luciferase activity, normalized to the buffer-treated transfection control to compare fold change in NF-κB activation for treatments.

Zebrafish experiments were approved by the University of Oregon Institutional Animal Care and Use Committee. Chop Chop (http://chopchop.cbu.uib.no) was used to design a guide RNA (gRNA) targeting the first exon of zebrafish ly96 (si:dkey-82k12.13, GRCz11). A gRNA template was generated by a template-free Phusion polymerase (New England BioLabs) PCR using a scaffold primer (5′-GATCCGCACCGACTCGGTGCCACTTTTTCAAGTTGATAACGGACTAGCCTTATTTTAACTTGCTATTTCTAGCTCTAAAAC-3′) and an ly96-specific primer (5′-AATTAATACGACTCACTATAGGGTATCAGATATGGCGCTTGTTTTAGAGCTAGAAATAGC-3′) and then cleaned using the QIAquick PCR Purification Kit (QIAGEN), transcribed using a MEGAscript Kit (Ambion), and purified by phenol–chloroform extraction and isopropanol precipitation. Cas9 RNA was made by linearizing the pT3TS-nls-zCas9-nls plasmid (47) with XbaI, purifying it using the QIAquick Gel Extraction Kit (QIAGEN), performing an in vitro transcription reaction using the T3 mMESSAGE Kit (Invitrogen), and purifying the RNA using the RNeasy Mini Kit (QIAGEN). AB strain zebrafish embryos were microinjected at the one-cell stage with 1–2 nl of a mixture containing 100 ng/μl Cas9 mRNA, 50 ng/μl gRNA, and phenol red and raised to adulthood. Fin DNA was amplified by PCR using primers specific to the targeted region (5′-CAAATTGGATTCACAACAGAGC-3′ and 5′-CCATGGAAAATCAATGAAAAGC-3′). Mosaic mutants were identified based on loss of an HaeII restriction site and were outcrossed to wild-type (WT) AB zebrafish to generate heterozygotes. Fish with loss-of-function mutations were identified by Sanger sequencing and further crossed to generate three independent homozygous ly96 mutant lines (Supplemental Fig. 1).

Larval 6 dpf fish were anesthetized in 168 mg/ml tricaine methane sulfonate in embryo medium (EM) and microinjected by cardiac ventricular injection with ∼2 nl of 100 μg/ml LPS from Salmonella enterica serotype typhimurium (L6511; Sigma-Aldrich) and transferred to fresh EM (22, 48). Six pools of five embryos of each genotype and condition were harvested at 6 h postinjection. Total RNA was purified using the same procedure as for gene expression analysis. Quantitative RT-PCR was used to measure il1b and mmp9 RNA levels in zebrafish tissues. RNA was treated with TURBO DNase (Ambion) according to manufacturer’s instructions and then reverse transcribed with an oligodeoxythymidine(20) primer using the Superscript III cDNA First Strand Synthesis Kit (Invitrogen). The quantitative PCR was set up using the KAPA SYBR FAST ABI PRISM Kit (KAPA Biosystems) according to manufacturer’s instructions and run on a Quant Studio 3 System (Thermo Fisher Scientific) using the default settings for SYBR Green reagents and the fast run mode in the QuantStudio Design and Analysis Software v.1.4.2. The comparative cycle threshold method was used to calculate relative mRNA levels. Each sample was run in triplicate, and data were normalized to the expression of the housekeeping genes eef1a1l1 and ppiab. The primers used for quantitative RT-PCR were published previously (49, 50). Md2 experiments were performed using AB strain fish, whereas myd88 experiments were on an AB × TU background.

We analyzed the outcome of these treatments on the log of the relative RNA level, using ANOVA accounting for the effects of genotype, treatment, RNA identity, and genotype:treatment cross-terms. We estimated the significance of individual factors using the Tukey honest significant difference approach. Statistical analyses were done using R 4.0.2 (46).

WT and homozygous ly96 mutant zebrafish embryos were grown under standard conditions in separate 10-cm petri dishes at a density of one fish per milliliter of EM, with 50 fish total per dish. At 5 dpf, LPS purified from E. coli 0111:B4 (L2630; Sigma-Aldrich) was dissolved in EM and added to dishes at a final concentration of 0.6 mg/ml, and control fish were mock treated with EM alone. Dead larvae, as determined by lack of heartbeat, were counted and removed at regular intervals from 16 to 48 h or from 16 to 72 h after the addition of LPS, at which time the experiment was terminated, and surviving fish were humanely euthanized.

The strongest argument against Tlr4/Md-2 performing LPS sensing in zebrafish is the presumed lack of Md-2. Md-2 is essential for LPS recognition by amniote Tlr4 as it contains the LPS binding pocket (Fig. 1). We therefore asked whether we could find a gene encoding Md-2 in bony fishes. By convention, the gene encoding Md-2 is known as ly96; therefore, throughout this manuscript, we will refer to the protein as Md-2 and its gene as ly96.

FIGURE 1.

LPS activation of amniote TLR4 requires cofactors MD-2 and CD14. (A) Schematic representation of LPS transfer from CD14 to the TLR4/MD-2 complex. LPS is brought by CD14 and loaded into MD-2, which binds to TLR4. Binding of LPS to the MD-2 coreceptor causes dimerization of the TLR4/MD-2 complex, activating a downstream inflammatory response. (B) The interface between human TLR4 (white and black) and MD-2 (dark gray) is extensive. Both are required to form a productive interaction with LPS (shown as spheres). Structure shown was made from Protein Data Bank 3FXI (82).

FIGURE 1.

LPS activation of amniote TLR4 requires cofactors MD-2 and CD14. (A) Schematic representation of LPS transfer from CD14 to the TLR4/MD-2 complex. LPS is brought by CD14 and loaded into MD-2, which binds to TLR4. Binding of LPS to the MD-2 coreceptor causes dimerization of the TLR4/MD-2 complex, activating a downstream inflammatory response. (B) The interface between human TLR4 (white and black) and MD-2 (dark gray) is extensive. Both are required to form a productive interaction with LPS (shown as spheres). Structure shown was made from Protein Data Bank 3FXI (82).

Close modal

We started by using the human MD-2 protein sequence to BLAST against the zebrafish genome and transcriptome. This returned no significant hits, so we took a more phylogenetically informed strategy. Relative to humans, the earliest branching, functionally characterized Tlr4/Md-2 complex is from chicken (Gallus gallus). We therefore “walked out” from amniotes toward fishes, starting with amphibians. We BLASTed the human MD-2 protein sequence against the Xenopus laevis genome. This revealed a hit to a hypothetical protein with 30% identity (OCT74818.1). When reverse BLASTed against the human proteome, this hit returned Md-2. To validate the amino acid sequence, we compared it with the sequences of functionally characterized Md-2 proteins from amniotes. We found that the X. laevis gene appeared to be N-terminally truncated. Using XenBase, we identified the full-length transcript in the transcriptome for X. laevis. By BLASTing against available amphibian transcriptomes (2932), we further identified putative Md-2 proteins in Rhinella marina, Hynobius retardatus, Odorrana margaretae, and Ichthyophis bannanicus (Supplemental Fig. 2).

With these putative amphibian Md-2 sequences in hand, we returned to our search for a zebrafish Md-2. A BLAST against a zebrafish transcriptome using the X. laevis sequence revealed a likely transcript (si:dkey-82k12.13, ENSDARG00000105462, 23% identity). We then searched additional fish transcriptomes available from the Fish-T1K project (51) and identified a set of transcripts from three species that matched Md-2 (Supplemental Fig. 2). The genes we identified in bony fishes that encode putative Md-2 proteins were highly diverged. On average, they exhibited only 26% identity against human Md-2 and only ∼40% identity relative to one another.

We continued to BLAST genomes and transcriptomes from species that diverged earlier than bony fishes relative to humans, including Chondrichthyes (cartilaginous fishes) and Agnatha (jawless fishes). We were unable to identify an Md-2 protein sequence or ly96 gene in either lineage. This is consistent with ly96 arising after the divergence of cartilaginous and bony fishes (∼470 million years ago) but before the divergence of bony- and ray-finned fishes (∼435 million years ago). The sequence resources for cartilaginous and jawless fishes remain relatively sparse, however, so we cannot exclude an earlier origin for this gene.

We next set out to assign the orthology of the putative ly96 genes from amphibians and bony fishes. In addition to returning ly96 sequences, our BLAST searches above returned sequences of ly86, a paralog of ly96 that encodes the protein Md-1. To determine if the newly identified zebrafish gene was ly86 or ly96, we constructed an alignment of 294 Md-1 and Md-2 protein sequences sampled from amniotes, amphibians, and bony fishes and then used this to infer a maximum likelihood phylogeny (Fig. 2A).

FIGURE 2.

Phylogeny and synteny of the identified zebrafish protein support classifying it as an Md-2 (the ly96 gene). (A) Maximum likelihood phylogeny of Md-2 and Md-1 proteins. Wedges are collapsed clades of orthologs, with wedge height corresponding to the number of included taxa and wedge length indicating the longest branch length with the clade. Tree includes Md-2 sequences from mammals (83 taxa), sauropsids (53 taxa), amphibians (3 taxa), and fish (4 taxa), in addition to Md-1 sequences from mammals (84 taxa), sauropsids (58 taxa), amphibians (3 taxa), and fish (6 taxa). Support values are SH supports calculated using an approximate likelihood ratio test. The taxa included in each clade are noted on the tree by silhouettes of mammals (mouse), sauropsids (chicken), amphibians (frog), and fish (zebrafish). (B) Genomic organization of genes surrounding Md-2 in representative vertebrates. Arrows for genes represent the coding strand. Lengths are to scale. The genomes are aligned to the ly96 start site.

FIGURE 2.

Phylogeny and synteny of the identified zebrafish protein support classifying it as an Md-2 (the ly96 gene). (A) Maximum likelihood phylogeny of Md-2 and Md-1 proteins. Wedges are collapsed clades of orthologs, with wedge height corresponding to the number of included taxa and wedge length indicating the longest branch length with the clade. Tree includes Md-2 sequences from mammals (83 taxa), sauropsids (53 taxa), amphibians (3 taxa), and fish (4 taxa), in addition to Md-1 sequences from mammals (84 taxa), sauropsids (58 taxa), amphibians (3 taxa), and fish (6 taxa). Support values are SH supports calculated using an approximate likelihood ratio test. The taxa included in each clade are noted on the tree by silhouettes of mammals (mouse), sauropsids (chicken), amphibians (frog), and fish (zebrafish). (B) Genomic organization of genes surrounding Md-2 in representative vertebrates. Arrows for genes represent the coding strand. Lengths are to scale. The genomes are aligned to the ly96 start site.

Close modal

The putative amphibian and bony fish Md-2 sequences grouped with the tetrapod MD-2 sequences with strong support (Shimodaira–Hasegawa statistic [SH] = 0.99). The Md-1/Md-2 protein tree largely reproduced the species tree, with the exception of amphibians. On the Md-1 lineage, amphibians form a polytomy with fishes at the base of the tree; on the Md-2 lineage, they are placed inside the amniote clade with a relatively short internal branch. This is likely an artifact of the small number of amphibian sequences as well as the rapid evolution of the genes along these lineages.

The protein sequences of Md-1 and, particularly, Md-2 are evolving rapidly. The total branch lengths between the last common ancestor of Md-2 to its human and zebrafish descendants are 2.00 and 2.44, respectively. Put another way, the average site in the Md-2 sequence has changed its amino acid approximately two times over the last 430 million years. Only 7 of 160 positions in MD-2 are universally conserved across the clade.

The simplest scenario consistent with this tree is that an ancestral gene arose and duplicated to form ly86 and ly96 sometime before the evolution of bony vertebrates. Both genes were preserved in most descendant lineages. Another, more complicated, explanation would have the putative fish ly96 gene be an ohnolog of the tetrapod ly96 gene. In this scenario, the putative fish ly96 arose by two gene duplications prior to the divergence of bony- and ray-finned fishes: one that led to ly86 and ly96, followed by a second duplication of ly96. Tetrapods then lost one duplicate of ly96, and bony fishes lost the other.

To distinguish these possibilities, we investigated the synteny of the putative ly96 gene, comparing the location of the gene in the D. rerio and X. laevis genomes to the location of ly96 in several tetrapods (Fig. 2B). In five genomes sampled from across bony vertebrates (including X. laevis and D. rerio), the ly96 gene is located between tmem70 and jph1b. This provides strong evidence that these amphibian and fish genes are, in fact, orthologous to the amniote gene encoding Md-2.

This evolutionary scenario predicts we should find sequences for ly96 in early-branching tetrapods (such as the coelacanth and lungfish) and early-branching teleosts (such as the sturgeon and reedfish). We searched through genomes and transcriptomes of these, and other related species, for evidence of ly96. As a whole, this proved difficult. For example, in the coelacanth genome (LatCha1), tmem70 and jph1 were on different contigs (ENSLACG00000011765 and ENSLACG00000013907); therefore, ly96 may or may not be present. That said, by BLASTing directly against the short-read archive, we found RNA transcripts from the South American lungfish (Lepidosiren paradoxa) that yield MD-2 when reverse BLASTed against the G. gallus proteome (SRR2895254.17689241.2 and SRR2895254.44296697.1). This thus provides another piece of evidence that the zebrafish ly96 is indeed an ortholog to human ly96 and that this evolved prior to the divergence of tetrapods and teleosts.

Finally, we also looked for a second copy of ly96 that may have arisen through the genome duplication event that occurred along the zebrafish lineage (52). We examined the genomic location of the jph1a paralog, but we were unable to identify an additional gene with any similarity to ly96. It appears that an inversion may have occurred in this region, complicating identification by synteny alone. This said, no additional transcripts were identified within the zebrafish transcriptome with similarity to the identified zebrafish ly96 sequence. This is consistent with a loss of the duplicate copy of this gene.

We next asked whether zebrafish express ly96. To do so, we used the recently released Zebrafish Single-Cell Transcriptome Atlas (43). This dataset consists of single-cell RNA-Seq transcriptomes for 44,102 individual cells extracted from 1, 2, and 5 dpf zebrafish. The gray points in Fig. 3A, 3B show the entire Atlas; each point is a cell, plotted such that cells with similar transcription profiles appear near one another. Cluster identity can be established by examining differentially expressed transcripts and using these marker genes to assess cell type expression in vivo (43); this provides a means to assess which cell types express ly96 simply by asking which clusters possess ly96 transcripts.

FIGURE 3.

ly96 and tlr4 genes are expressed in immune cells. Each point in these plots is an individual cell characterized by single-cell RNA-Seq. The distance between the cells corresponds to the relative difference in their transcriptional profiles (43). (A) Yellow points indicate cells expressing ly96, and gray points show all 44,102 cells in the dataset. The two clusters in which ly96 is expressed (c71 and c212) are highlighted with black circles. (B) Colored points indicate cells expressing tlr4bb (green), tlr4al (magenta), or tlr4ba (orange); gray points and circles are identical to (A). (C) Enlarged views of clusters c212 and c71, separated by gene of interest. This includes the genes shown in (A and B) as well the macrophage marker, mpeg1.1 (light blue) (26, 27). The number in the bottom left of each table entry is the expression level of the gene within the cluster divided by its expression level in all other cells in the dataset. If there was no expression in cells outside the cluster, expression within the cluster was divided by the detection threshold (0.001), giving a minimum estimate for the enrichment. The background cells are now colored by the developmental stage from which the cell was isolated: 1 dpf (white), 2 dpf (light gray), or 5 dpf (dark gray). The dashed lines shown on c71 are approximate divisions between the age-dependent subclusters of c71.

FIGURE 3.

ly96 and tlr4 genes are expressed in immune cells. Each point in these plots is an individual cell characterized by single-cell RNA-Seq. The distance between the cells corresponds to the relative difference in their transcriptional profiles (43). (A) Yellow points indicate cells expressing ly96, and gray points show all 44,102 cells in the dataset. The two clusters in which ly96 is expressed (c71 and c212) are highlighted with black circles. (B) Colored points indicate cells expressing tlr4bb (green), tlr4al (magenta), or tlr4ba (orange); gray points and circles are identical to (A). (C) Enlarged views of clusters c212 and c71, separated by gene of interest. This includes the genes shown in (A and B) as well the macrophage marker, mpeg1.1 (light blue) (26, 27). The number in the bottom left of each table entry is the expression level of the gene within the cluster divided by its expression level in all other cells in the dataset. If there was no expression in cells outside the cluster, expression within the cluster was divided by the detection threshold (0.001), giving a minimum estimate for the enrichment. The background cells are now colored by the developmental stage from which the cell was isolated: 1 dpf (white), 2 dpf (light gray), or 5 dpf (dark gray). The dashed lines shown on c71 are approximate divisions between the age-dependent subclusters of c71.

Close modal

We found that ly96 is expressed in two clusters, denoted “c71” and “c212” (Fig. 3A, yellow points). Both of these clusters are annotated in the Atlas as putative macrophage cells based on their transcription profiles (43). ly96 is highly enriched in these clusters relative to other clusters. This can be measured by taking the ratio of the average expression level of ly96 for the cells in the cluster relative to the average expression level of ly96 in all other cells. This ratio is 235 for cluster c212 and ≥173 for cluster c71. For comparison, the well-established macrophage marker mpeg1.1 (26, 27, 53) has ratios of 102 and 202, respectively, for these same clusters (Fig. 3C).

We next investigated the expression of the tlr4bb, tlr4al, and tlr4ba genes. We found that tlr4bb and tlr4al had quite limited expression patterns (Fig. 3B, green and magenta), whereas tlr4ba was expressed broadly (Fig. 3B, orange). All three tlr4 genes were found in cluster c212, but only tlr4bb was found in cluster c71 (Fig. 3C).

The Atlas also has the potential to reveal time course information for the expression of these genes as it contains cells isolated from fish at 1, 2, and 5 dpf. We therefore shaded the cells within clusters c71 and c212 by their developmental time point (Fig. 3C). Cluster c212, where we observed overlapping expression for ly96 and all three tlr4 genes, consists entirely of cells isolated from 5 dpf zebrafish (Fig. 3C). Cluster c71 has three discrete subclusters corresponding to the age of the fish from which the cell was extracted. We see no ly96 in the 1 dpf subcluster, a small amount in the 2 dpf subcluster, and the highest level in the 5 dpf subcluster (Fig. 3C). Likewise, tlr4bb is expressed in the 5 dpf subcluster but no others. For comparison, the macrophage marker mpeg1.1 is found in all cells within c71 and c212, regardless of the age of the fish from which the cell was extracted.

These observations suggest that ly96 and all three tlr4 genes are expressed together in a subset of macrophage cells by 5 dpf (Fig. 3C, c212). Samples of later time points would be necessary to establish if these genes are at their full expression level by 5 dpf or if their expression level and cell type specificity continues to change as the fish develop.

Given the low sequence similarity between the zebrafish Md-2 protein and its amniote orthologs, it was not clear that the zebrafish Md-2 would be capable of mediating the Tlr4 response to Md-2. We therefore turned to an ex vivo cell culture assay to assess the ability of the zebrafish Md-2 to partner with zebrafish Tlr4ba and Tlr4bb for LPS activation. In this assay, we cotransfected genes encoding complex components into HEK293T cells and then used luciferase to quantify NF-κB output in response to exogenously applied LPS (6).

We started by coexpressing zebrafish Md-2 and Tlr4ba or Tlr4bb and then measuring NF-κB activation in response to exogenously applied LPS. We saw no activation (Fig. 4A). This result was unsurprising as this experiment attempted to activate a Tlr4/Md-2 complex without Cd14, an important peripheral protein that brings LPS to Tlr4/Md-2 complexes in amniotes, dramatically increasing the NF-κB response (Fig. 1) (5458). We thus cotransfected tlr4ba or tlr4bb with zebrafish ly96 and human CD14. In this context, we observed potent activation of NF-κB in response to LPS for tlr4ba but not tlr4bb (Fig. 4A). To verify that the activation of Tlr4ba required Md-2 rather than merely CD14, we tested the activation of Tlr4ba and CD14 without transfecting ly96; this complex did not respond to LPS (Fig. 4A). We then verified that the zebrafish Tlr4ba/Md-2 complex, complemented with human CD14, exhibited a dose-dependent response to LPS (Fig. 4B). The concentration of LPS needed for activation of the zebrafish Tlr4ba/Md-2 complex was much higher than that needed for activation of the human proteins in these cells but consistent with what has been observed for other species (59).

FIGURE 4.

LPS activates the zebrafish Tlr4ba/Md-2 in a functional assay. (A) Activation of zebrafish Tlr4ba and Tlr4bb in the presence and absence of zebrafish Md-2 and human CD14. Points are the technical replicates from three biological replicates. Bold lines are the mean of the biological replicates. Error bars are an SE on the mean of the biological replicates. (B) Dose dependence of LPS response by zebrafish Tlr4ba/Md-2 in the presence (circles) and absence (squares) of human CD14. (C) Zebrafish Tlr4ba/Md-2 complemented with Cd14 proteins from amniotes (Homo sapiens, Mus musculus, Monodelphis domestica, and Gallus gallus). (D) Zebrafish Tlr4ba complemented with species-matched Md-2/Cd14 pairs taken from amniotes. Statistically significant differences (single-tailed Student t test) are noted on each panel. *p < 0.05.

FIGURE 4.

LPS activates the zebrafish Tlr4ba/Md-2 in a functional assay. (A) Activation of zebrafish Tlr4ba and Tlr4bb in the presence and absence of zebrafish Md-2 and human CD14. Points are the technical replicates from three biological replicates. Bold lines are the mean of the biological replicates. Error bars are an SE on the mean of the biological replicates. (B) Dose dependence of LPS response by zebrafish Tlr4ba/Md-2 in the presence (circles) and absence (squares) of human CD14. (C) Zebrafish Tlr4ba/Md-2 complemented with Cd14 proteins from amniotes (Homo sapiens, Mus musculus, Monodelphis domestica, and Gallus gallus). (D) Zebrafish Tlr4ba complemented with species-matched Md-2/Cd14 pairs taken from amniotes. Statistically significant differences (single-tailed Student t test) are noted on each panel. *p < 0.05.

Close modal

Our results support the hypothesis that zebrafish Tlr4ba/Md-2 can activate in response to LPS; however, this could only be done with the presence of a supporting mammalian protein (human CD14). To determine if this was an artifact of the human protein, we tested the LPS activation of Tlr4ba/Md-2 in the presence of human, mouse, opossum, and chicken Cd14. We found that all but the chicken Cd14 were able to support the activation of the complex (Fig. 4C). Thus, the activity of the zebrafish Tlr4ba/Md-2 complex does not depend exclusively on human CD14 but can instead be supported by diverse Cd14 molecules.

Given the importance of Cd14 in this assay, we looked for evidence of a zebrafish cd14 gene; however, we were unable to locate such a gene. The inability to detect a cd14 in fish may be due to rapid evolution of this gene because the most recent common ancestor, or, alternatively, Cd14 may have arisen as a supporting molecule for LPS recognition after the divergence of tetrapods. The requirement for Cd14 in these experiments could be a problem with the heterologous cell line (these experiments were done in human cells) or a missing alternate secondary cofactor (such as a fish LPS binding protein).

Finally, to see if zebrafish Tlr4ba behaved similarly to amniote Tlr4, we investigated whether Md-2 from other species could act in concert with zebrafish Tlr4ba. We cotransfected tlr4ba with human, mouse, or opossum ly96 genes. We saw complementation by both mouse and opossum Md-2 for LPS activation of zebrafish Tlr4ba (Fig. 4D). This suggests that, despite lineage-specific coevolution leading to incompatibility between some interspecies pairs of Tlr4 and Md-2 (60), the core biochemical features that allow assembly of an LPS-sensitive Tlr4/Md-2 complex arose over 400 million years ago before the tetrapod/teleost split.

We next probed for a physiological role for Md-2 in larval zebrafish. We used CRISPR-Cas9–based mutagenesis to establish three independent zebrafish lines with mutations in the first exon of the ly96 gene. The mutations were expected to induce a loss of function through removal of the start codon (ly96A/A) or through a frame shift and premature stop codon (ly96B/B and ly96C/C) (Supplemental Fig. 1). Using RT-PCR primers downstream of the targeted region, we demonstrated that ly96 mRNA is expressed in mutant larval zebrafish (Fig. 5A).

FIGURE 5.

ly96 mutations perturb the transcriptional response to LPS injection in larval zebrafish. (A) mRNA transcript level for each ly96 mutant zebrafish. Rows show amplicons with different PCR primers (ly96 or b2m) with and without reverse transcriptase (RT). Columns show fish genotype. (B and C) Graphs show the relative RNA levels of mmp9 (gray) and il1b (black) for WT and mutant fish in the presence or absence of LPS treatment. RNA levels were normalized to the WT values within each panel. Points represent RNA pooled from five embryos (six pools for each treatment); solid black lines are the mean of the log10(RNA level) for each genotype/treatment. The p values were calculated using the Tukey honest significant difference method for the indicated genotype x treatment effect within the ANOVA model. Experiments in (B) were done using AB × TU fish; experiments in (C) were done using AB fish. *0.05 > p > 0.005, **0.005 > p > 0.0005, ***p < 0.0005.

FIGURE 5.

ly96 mutations perturb the transcriptional response to LPS injection in larval zebrafish. (A) mRNA transcript level for each ly96 mutant zebrafish. Rows show amplicons with different PCR primers (ly96 or b2m) with and without reverse transcriptase (RT). Columns show fish genotype. (B and C) Graphs show the relative RNA levels of mmp9 (gray) and il1b (black) for WT and mutant fish in the presence or absence of LPS treatment. RNA levels were normalized to the WT values within each panel. Points represent RNA pooled from five embryos (six pools for each treatment); solid black lines are the mean of the log10(RNA level) for each genotype/treatment. The p values were calculated using the Tukey honest significant difference method for the indicated genotype x treatment effect within the ANOVA model. Experiments in (B) were done using AB × TU fish; experiments in (C) were done using AB fish. *0.05 > p > 0.005, **0.005 > p > 0.0005, ***p < 0.0005.

Close modal

To determine if deletion of ly96 alters cytokine expression in response to LPS, we injected 6 dpf larval zebrafish with LPS and measured the RNA levels of the downstream genes mmp9 and il1b. As a control, we included myd88−/− fish, which should have perturbed innate immune transcriptional responses (21). Fig. 5B shows the effect of the LPS treatment on the RNA levels of mmp9 (gray) and il1b (black) for WT and myd88(−/−) fish. The LPS treatment causes a roughly 3-fold increase in transcript levels of both of these genes in WT fish. The myd88(−/−) knockout fish have depressed mmp9 and il1b RNA levels relative to WT in the absence of LPS treatment (p = 1.9 × 10−5) but still respond to LPS (p = 0.047).

We repeated these experiments for the ly96A/A, ly96B/B, and ly96C/C fish (Fig. 5C). We found that ly96A/A zebrafish had normal basal mmp9 and il1b RNA levels, but the RNA levels were unresponsive to LPS treatment. The ly96B/B mutant behaved similarly to the myd88(−/−) mutant as it had significantly depressed basal mmp9 and il1b RNA levels, but it still responded to LPS. The mean basal levels of mmp9 and il1b were slightly lower for the ly96C/C mutant than for WT but not significantly so. This mutant remained responsive to LPS.

We next tested the role of ly96 in LPS-induced septic shock in larval 5 dpf zebrafish. We first treated 5 dpf larval WT zebrafish with LPS and followed their survival over time. None of the treated WT fish survived more than 48 h; the median survival time was 30 h (Fig. 6A). As a control, we also tested the LPS response for myd88−/− fish. As has been observed previously (21), these showed a modest but significant increase in survival (Fig. 6A). This was consistent with LPS inducing a response that involves a myd88-dependent pathway.

FIGURE 6.

ly96 mutations affect LPS survival only moderately in larval zebrafish. (AD) Curves show survival of WT (black) and mutant (red) zebrafish in the presence of 0.6 mg/ml LPS (solid line) or mock treatment (dashed line). The genotype is indicated on each panel. The p value was determined by comparing the matched survival curves by a log-rank Mantel–Cox test. The experiments shown in (A–C) were performed in parallel, whereas the experiments in (D) were performed at a later date with an LPS lot that showed lower potency, necessitating a longer treatment time. (A–D) represent averages of one, five, five, and three experimental repeats, respectively.

FIGURE 6.

ly96 mutations affect LPS survival only moderately in larval zebrafish. (AD) Curves show survival of WT (black) and mutant (red) zebrafish in the presence of 0.6 mg/ml LPS (solid line) or mock treatment (dashed line). The genotype is indicated on each panel. The p value was determined by comparing the matched survival curves by a log-rank Mantel–Cox test. The experiments shown in (A–C) were performed in parallel, whereas the experiments in (D) were performed at a later date with an LPS lot that showed lower potency, necessitating a longer treatment time. (A–D) represent averages of one, five, five, and three experimental repeats, respectively.

Close modal

We then tested the three ly96 mutant zebrafish lines for their susceptibility to LPS-induced septic shock. The results were mixed. Compared with matched WT controls, ly96A/A zebrafish survived for slightly longer (Fig. 6B), ly96B/B zebrafish survived similarly (Fig. 6C), and ly96C/C zebrafish survived shorter (Fig. 6D). This is consistent with some pathway other than Tlr4/Md-2 being the primary route for LPS-induced death in larval zebrafish.

Finally, we revisited the idea that the evolutionary history of zebrafish tlr4 genes implies that they do not act as LPS sensing molecules. Previous authors suggested that an ancestral TLR gene duplicated in the ancestor of bony vertebrates (∼450 million years ago) and that the two paralogs were then differentially lost on the mammalian and bony fish lineages (20), respectively. This early divergence, before the evolution of ly96, may suggest very different functional roles for mammalian versus fish tlr4s.

We set out to better resolve when the zebrafish tlr4 paralogs arose relative to their mammalian counterparts, particularly with regard to the evolution of ly96. As with our analysis of Md-2, we started with a phylogenetic tree and then turned to synteny. For the phylogenetic tree, we constructed a multiple sequence alignment of bony vertebrate protein sequences containing 278 Tlr4 sequences and 189 Cd180 sequences as an outgroup (Cd180 is the most closely related paralog to Tlr4) (61). In the resulting maximum likelihood tree, Tlr4 and Cd180 form distinct, well-supported clades (Fig. 7A). Within the Tlr4 clade, zebrafish Tlr4ba, Tlr4bb, and Tlr4al are part of a monophyletic group with other Tlr4s from fish. It appears that the duplication of the zebrafish tlr4-like genes occurred in the ancestor of Cypriniformes as the duplicate is seen in species closely related to zebrafish but not in more distant species, such as the catfish, pike, and gar. It also appears that tlr4al is a lineage-specific duplicate of tlr4ba as these proteins group strongly with one another to the exclusion of other tlr4-like protein sequences within the Cypriniformes.

FIGURE 7.

Zebrafish tlr4 paralogs evolved within the ray-finned fishes. (A) Maximum likelihood phylogeny for 467 Tlr4 and Cd180 protein sequences. SH supports are indicated on the tree. Wedges are clades, with the length indicating the maximum branch length from the ancestor of the clade. The taxonomic distribution and number of genes within each wedge are indicated on the plot. The clade containing human TLR4 is highlighted in purple; the three zebrafish tlr4 sequences are highlighted in orange. (BG) Hits for human (purple) and zebrafish (orange) gene sets on six representative chromosomes taken from five species. The species and chromosome are indicated at the top of each plot. The x-axis denotes position on the chromosome. Triangles indicate gene start positions. The green arrow indicates the location of a Tlr4 gene. The y-axis is a running average of the BLAST e value for each gene set along the genome (see 2Materials and Methods). The numbers on the plot indicate the number of human and zebrafish hits within the region shown. (H and I) Each row shows the chromosome with the most BLAST hits from the human (H) or zebrafish (I) gene set. Columns indicate specific genes from the set, with names denoted below. A colored square indicates a gene found somewhere on the chromosome. A green square is a tlr4 gene. The species tree is shown on the left; the chromosome number is on the right. (J) Schematic representation of a plausible scenario for the history of the tlr4 gene. Times are taken from Hughes et al. (83) and timetree.org (84).

FIGURE 7.

Zebrafish tlr4 paralogs evolved within the ray-finned fishes. (A) Maximum likelihood phylogeny for 467 Tlr4 and Cd180 protein sequences. SH supports are indicated on the tree. Wedges are clades, with the length indicating the maximum branch length from the ancestor of the clade. The taxonomic distribution and number of genes within each wedge are indicated on the plot. The clade containing human TLR4 is highlighted in purple; the three zebrafish tlr4 sequences are highlighted in orange. (BG) Hits for human (purple) and zebrafish (orange) gene sets on six representative chromosomes taken from five species. The species and chromosome are indicated at the top of each plot. The x-axis denotes position on the chromosome. Triangles indicate gene start positions. The green arrow indicates the location of a Tlr4 gene. The y-axis is a running average of the BLAST e value for each gene set along the genome (see 2Materials and Methods). The numbers on the plot indicate the number of human and zebrafish hits within the region shown. (H and I) Each row shows the chromosome with the most BLAST hits from the human (H) or zebrafish (I) gene set. Columns indicate specific genes from the set, with names denoted below. A colored square indicates a gene found somewhere on the chromosome. A green square is a tlr4 gene. The species tree is shown on the left; the chromosome number is on the right. (J) Schematic representation of a plausible scenario for the history of the tlr4 gene. Times are taken from Hughes et al. (83) and timetree.org (84).

Close modal

We next set out to identify when the zebrafish tlr4 ohnologs evolved by investigating the genomic context for tlr4 genes in 11 genomes, each with a complete chromosome assembly. We selected a set of 22 genes flanking human tlr4 and a set of 22 genes flanking the three zebrafish tlr4 genes. Notably, there were no shared homologs between the sets, demonstrating the radical difference between the genomic contexts of human and zebrafish tlr4. We then used these sets of genes to BLAST each of the 11 genomes and calculated a running average for the BLAST e values along each chromosome. This allowed us to assess the overall similarity of genomic regions to either the human or zebrafish tlr4 context. Fig. 7B–G shows representative traces for six chromosomes taken from five species. We were able to distinguish two distinct contexts for tlr4 genes. In some organisms (human and frog, for example), tlr4 is surrounded by hits from the human gene set (Fig. 7B, 7D). In other organisms (zebrafish and pike, for example), tlr4 is surrounded by hits from the zebrafish gene set (Fig. 7C, 7E).

To place our results in their evolutionary context, we plotted our BLAST output against the phylogeny for our chosen species. For each species, we displayed the chromosome with the most hits from the human set (Fig. 7H) and the chromosome with the most hits from the zebrafish set (Fig. 7I). We made an exception for the pike, displaying the chromosome with the tlr4 gene (linkage group 5), not the chromosome with the most zebrafish hits (linkage group 6). We indicated whether a gene from the human or zebrafish set was seen somewhere on that chromosome by coloring the square corresponding to that gene.

Four species had tlr4 in a human-like context: human, chicken, frog, and gar. None of these species, including the gar, had a duplicate copy of tlr4 in a zebrafish-like context. The human-like context of the gar gene is shown in Fig. 7F, whereas the lack of tlr4 in the most zebrafish-like region of the gar genome is shown in Fig. 7G. The remainder of the ray-finned fishes had tlr4 in either a zebrafish-like context (catfish, zebrafish, and pike) or had no tlr4 gene at all (bonytongue, cod, and puffer).

The most parsimonious history consistent with the observed distribution across genomes is shown in Fig. 7J. In this scenario, tlr4 arose in a genomic context similar to the one preserved in humans. This occurred after the divergence of bony and cartilaginous fishes (∼475 million years ago) but before the divergence of ray-finned and lobe-finned fishes (∼430 million years ago). The ancestral genomic context was preserved in tetrapods, including humans. It was also maintained in the ray-finned fishes for ∼130 million years, as indicated by the location of the tlr4 gene in the gar genome. Then, sometime between 300 and 250 million years ago, the tlr4 gene was both duplicated into the genomic context observed in zebrafish as well as lost from the ancestral context. Between 150 and 100 million years ago, a tandem duplication occurred within the Cypriniformes fishes, leading to the tandem copies of tlr4ba and tlr4bb observed in zebrafish, carp, and other Cypriniformes fishes. Finally, after the divergence of zebrafish and carp, a second tandem duplication of tlr4ba led to the formation of tlr4al.

This revised evolutionary history places the evolution of the zebrafish tlr4 paralogs much later than was previously hypothesized (20). Importantly, the duplication of TLR4 occurred after the evolution of Md-2, meaning that the formation of the Tlr4/Md-2 complex likely predates the duplication event. Thus, the interaction with Md-2 and the ability to activate with LPS were an ancestral feature of zebrafish Tlr4 rather than something that could only be gained in parallel along the tetrapod and bony fish lineages.

Our observations led us to reevaluate the decade-old idea that Tlr4 does not participate in the LPS-induced inflammatory response in zebrafish. We have identified the zebrafish gene encoding the Tlr4 coreceptor Md-2 (ly96). The gene, like tlr4ba, tlr4bb, and tlr4al, is expressed in zebrafish cells that also express a collection of macrophage genes. In concert with zebrafish Tlr4ba, zebrafish Md-2 is capable of activating NF-κB signaling in an ex vivo functional assay. Zebrafish ly96 loss-of-function mutants have a perturbed transcriptional response to LPS challenge, suggesting that the Md-2 protein is involved in LPS sensing in vivo. Finally, a careful phylogenetic analysis suggests that the mammalian and zebrafish tlr4 genes are not as evolutionarily distinct as previously thought. Although not direct orthologs, the zebrafish paralogs evolved well after ly96 and likely preserve an ancestral LPS recognition activity.

Our work demonstrates that, given the correct context, zebrafish Tlr4ba and Md-2 form a functional complex that recognizes LPS and activates NF-κB signaling. Further, the molecular basis for the interaction between the partners appears to have been conserved for the last 450 million years; zebrafish Tlr4ba is compatible with mouse and opossum Md-2 (Fig. 4D). This is despite the fact that the orthologous proteins from each species have only ∼20% identity at the amino acid sequence level. The simplest explanation for this observation is that the ability of Tlr4/Md-2 to activate in response to LPS is an ancestral feature of the protein complex.

This said, the overall significance of LPS signaling through Tlr4/Md-2 in zebrafish remains unclear. Both our cell culture functional assay and our studies of larval zebrafish with ly96 loss-of-function mutations gave mixed results. We will discuss each in turn.

In our functional assays, we had to add a mammalian Cd14 to activate NF-κB signaling through zebrafish Tlr4ba/Md-2 (Fig. 4C). In amniotes, Cd14 delivers LPS directly to Md-2 (Fig. 1). We could find no ortholog to Cd14 in the zebrafish genome.

One possibility is that the human cell line used for the functional assays is missing some critical component for the delivery of LPS and assembly of the active dimer. Tlr4, Md-2, and Cd14 are the necessary and sufficient set of amniote proteins that confer an LPS-dependent NF-κB response in HEK293T cells. It could be that some other nonhomologous protein plays the role of Cd14 in zebrafish.

Another possibility is that LPS is not a zebrafish Tlr4ba/Md-2 agonist in vivo. We showed that we can activate the complex in a human cell line given an appropriate delivery molecule and a high enough LPS concentration. But under physiological conditions, the Tlr4ba/Md-2 complex could respond to some other chemically similar ligand. This would not be surprising; changes in ligand specificity have been observed across Md-2 in the amniotes (62). There is also some evidence that zebrafish Tlr4ba may be antagonized by LPS in vivo (19). This would be compatible with another ligand activating the complex and LPS competing and activating at a lower level than can be achieved by the native ligand.

Finally, our observation that Tlr4ba activates NF-κB with both mouse and opossum Md-2 directly contrasts previous work that showed that the Tlr4ba/mouse Md-2 complex could not activate NF-κB (Fig. 4C) (19, 20). The key difference between our experiments and those done previously is the sequence of Tlr4ba used. Previous investigators used a construct that was ∼75 aa shorter than tetrapod Tlr4s. This construct is missing both the signal peptide required to target Tlr4ba to the cell surface and a region of the protein that is likely critical for Md-2 binding (Supplemental Fig. 3). In contrast, we used a full-length open reading frame (ENSDART00000044697.6, GRCz10). The difference in our constructs arises because the previous analysis relied on cDNA that, apparently, captured an alternate splice variant of tlr4ba.

Larval zebrafish ly96 loss-of-function mutants did not exhibit appreciably altered death rates upon exposure to LPS compared with WT (Fig. 6B–D). This is consistent with a previous morpholino study that knocked down tlr4ba and observed no change in sensitivity to LPS (19). This contrasts with mice, however, in which knockout of Ly-96 is protective against endotoxic shock (4), and disruption or knockout of Tlr4 leads to hyporesponsiveness to LPS (1, 63). We also found mixed results for cytokine expression in the zebrafish ly96 mutants following LPS challenge (Fig. 5B, 5C); in ly96A/A larvae, LPS-induced expression of inflammatory cytokines is ablated, in ly96B/B larvae, we observe a response similar to the myd88−/− background, and in ly96C/C larvae, the response is similar to WT.

We cannot rule out the possibility that these mild LPS responses are due to an experimental artifact. Another possible explanation for the mild LPS response is that zebrafish may have retained a second copy of the ly96 gene from the teleost genome duplication that compensates for the loss of function of the targeted copy. We were unable to find any evidence of such a gene; however, the challenge of finding the original ly96 gene means that we cannot rule this out. A further possibility is that the mutants that we generated may not represent a complete loss of function. For example, use of a potential alternative start codon 17 aa downstream of the normal start codon could produce a truncated protein (ly96A/A). Although this would be missing N-terminal amino acids that are known to be critical for Md-2 function in other systems (Supplemental Fig. 1), these amino acids may not be necessary in zebrafish.

It is also important to note that we tested a single early developmental time point (5 dpf). As we and others have shown, the expression patterns for tlr4 genes in zebrafish change during larval development (43, 64, 65). It could be that Tlr4ba/Md-2, although expressed in larvae (Fig. 3C), is not yet a large player in LPS sensing. Our expression analysis revealed that in larval zebrafish, ly96 is much more restricted in its expression pattern than the corresponding mammalian genes (66). There may, in fact, be specific subtypes of macrophages that express ly96 and tlr4s (and are defective in LPS sensing in the ly96 mutants) but remain invisible at the level of LPS-induced death and overall cytokine expression. Higher-resolution studies of LPS-induced inflammation at multiple developmental time points will be required to sort this out.

Given the quantities of LPS needed for in vivo studies, the use of ultrapure LPS was not possible for in vivo experiments. There are a few limitations associated with the use of non-ultrapure LPS; most notably, there has been concern that LPS-induced toxicity may be due in part to peptidoglycan (PGN) contamination in standard preparations of LPS. To address this concern, we have previously directly assessed the role of PGN in LPS-induced toxicity in zebrafish and shown that PGN does not induce toxicity in zebrafish within 24 h (21). It is possible, however, that contaminants in the preparation of LPS used could explain the inability to fully ablate LPS-induced toxicity and expression of inflammatory cytokines in response to LPS. In future studies, the use of different LPS preparations and testing for signaling with LPS from different bacterial species may further clarify the role of LPS signaling in zebrafish.

Another difficulty would be if other non-Toll–like pathway(s) are involved in LPS signaling zebrafish. If true, even if the Tlr4ba/Md-2 complex contributes to the LPS-induced inflammatory response, removing it might have a relatively small effect on cytokine production or LPS-induced toxicity. We see evidence for such a pathway as addition of LPS dramatically increased death rate even in a myd88−/− background (Fig. 6A). One possibility is that this occurs by intracellular sensing of LPS via caspases and inflammasomes (67). Various studies have shown inflammasome signaling to be widespread in zebrafish larvae (68) and Il-1r to be required to prevent cell death in response to infection in multiple cells (69). Intracellular sensing may be much more important in fish than mammals; zebrafish have 385 of these putative intracellular sensors, whereas humans have 22 (70).

The presence of Md-2 in zebrafish indicates that both Tlr4 and Md-2 existed together in the last common ancestor of bony vertebrates. Because descendants along both the tetrapod and ray-finned fish lineages activate with LPS, the ability to respond to LPS is likely an ancestral function that has been conserved for 435 million years.

That said, these proteins have evolved significantly since this shared ancestor. Along the tetrapod lineage, a supporting collection of proteins evolved. Cd14 arose through a duplication within the TLR family and is now an essential component of the Tlr4/Md-2 complex, delivering LPS to Md-2 in a coordinated fashion (55, 56). Tetrapods also acquired lipid binding protein, improving LPS delivery (54, 71). Amniotes then further adjusted the Tlr4/Md-2 proinflammatory response through the addition of amniote-specific damage-associated molecular pattern molecules such as S100A9 (60), which amplify LPS-induced inflammation (72). All the while, mutations to Md-2 changed its specificity for LPS and its chemical analogs (73). For example, humans acquired unique lipid IVa antagonism sometime after the divergence of humans and mice (74, 75).

The changes that occurred along the ray-finned fish lineages are not yet clear. Did they acquire supporting LPS delivery molecules analogous to Cd14? Has the specificity of Md-2 fluctuated in ray-finned fishes as it has along the tetrapod lineage? Further work is needed to answer these questions.

We hypothesize, however, that ray-finned fishes maintain an ancestral, low-sensitivity Tlr4/Md-2 LPS sensing complex. Fish have previously been shown to be relatively resistant to septic shock (76, 77), with high concentrations of LPS needed to activate teleost leukocytes (24, 7880). This parallels the observation that early diverging tetrapods, such as amphibians, also require high doses of LPS to trigger an inflammatory response (81). This could be explained if ray-finned fishes do not have specialized machinery to deliver LPS to the complex but instead use Tlr4/Md-2 as a simple LPS sensor. Other observations consistent with a relatively primitive Tlr4/Md-2 LPS response in zebrafish are the fact that Tlr4 was lost independently along multiple fish lineages (19, 20, 70) (Fig. 7J) as well as the existence of parallel LPS sensing pathways in zebrafish (68). If the Tlr4/Md-2 complex is peripheral to the LPS response in ray-finned fishes, it could be lost with minimal fitness consequences. In contrast, Tlr4/Md-2 became progressively more central to the LPS response along the mammalian lineage and, as a result, has been highly conserved.

Much more work is needed to fully understand the role of LPS signaling in zebrafish. Our work gives the first evidence, to our knowledge, for the presence of Md-2 outside of amniotes. Knowledge of this essential cofactor for LPS signaling through Tlr4 opens the door for future studies of LPS sensing through Tlr4 in fish.

We thank Kristi Hamilton and Lila Kaye for assistance with zebrafish LPS survival assays and Rose Sockol and the University of Oregon Aquatic Animal Care Services staff for fish husbandry. We thank Prof. Carol Kim for sharing the tlr4bb plasmid.

This work was supported by an American Heart Association grant (AHA-15BGIA22830013 to M.J.H.), National Institute of General Medical Sciences grants (NIH-T32GM007413 to A.N.L., NIH-F32DK107318 to M.N.H., and P50GM09891 to K.G.), and a National Institutes of Health Office of the Director grant (NIH-R24OD026591 to A.C.M.). M.J.H. is a Pew Scholar in the Biomedical Sciences, supported by The Pew Charitable Trusts. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The online version of this article contains supplemental material.

Abbreviations used in this article:

BLAST

basic local alignment search tool

dpf

day postfertilization

EM

embryo medium

gRNA

guide RNA

PGN

peptidoglycan

RNA-Seq

RNA sequencing

SH

Shimodaira–Hasegawa statistic

WT

wild-type.

1
Poltorak
,
A.
,
X.
He
,
I.
Smirnova
,
M.-Y.
Liu
,
C.
Van Huffel
,
X.
Du
,
D.
Birdwell
,
E.
Alejos
,
M.
Silva
,
C.
Galanos
, et al
.
1998
.
Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene.
Science
282
:
2085
2088
.
2
Schletter
,
J.
,
H.
Heine
,
A. J.
Ulmer
,
E. T.
Rietschel
.
1995
.
Molecular mechanisms of endotoxin activity.
Arch. Microbiol.
164
:
383
389
.
3
Ulevitch
,
R. J.
,
P. S.
Tobias
.
1995
.
Receptor-dependent mechanisms of cell stimulation by bacterial endotoxin.
Annu. Rev. Immunol.
13
:
437
457
.
4
Nagai
,
Y.
,
S.
Akashi
,
M.
Nagafuku
,
M.
Ogata
,
Y.
Iwakura
,
S.
Akira
,
T.
Kitamura
,
A.
Kosugi
,
M.
Kimoto
,
K.
Miyake
.
2002
.
Essential role of MD-2 in LPS responsiveness and TLR4 distribution.
Nat. Immunol.
3
:
667
672
.
5
Park
,
B. S.
,
J.-O.
Lee
.
2013
.
Recognition of lipopolysaccharide pattern by TLR4 complexes.
Exp. Mol. Med.
45
: e66.
6
Chow
,
J. C.
,
D. W.
Young
,
D. T.
Golenbock
,
W. J.
Christ
,
F.
Gusovsky
.
1999
.
Toll-like receptor-4 mediates lipopolysaccharide-induced signal transduction.
J. Biol. Chem.
274
:
10689
10692
.
7
Yiu
,
J. H. C.
,
B.
Dorweiler
,
C. W.
Woo
.
2017
.
Interaction between gut microbiota and toll-like receptor: from immunity to metabolism.
J. Mol. Med. (Berl.)
95
:
13
20
.
8
Fan
,
J.
,
A. B.
Malik
.
2003
.
Toll-like receptor-4 (TLR4) signaling augments chemokine-induced neutrophil migration by modulating cell surface expression of chemokine receptors. [Published erratum appears in 2003 Nat. Med. 9: 477.]
Nat. Med.
9
:
315
321
.
9
Murad
,
S.
2014
.
Toll-like receptor 4 in inflammation and angiogenesis: a double-edged sword.
Front. Immunol.
5
:
313
.
10
Leon
,
C. G.
,
R.
Tory
,
J.
Jia
,
O.
Sivak
,
K. M.
Wasan
.
2008
.
Discovery and development of toll-like receptor 4 (TLR4) antagonists: a new paradigm for treating sepsis and other diseases.
Pharm. Res.
25
:
1751
1761
.
11
Kuzmich
,
N. N.
,
K. V.
Sivak
,
V. N.
Chubarev
,
Y. B.
Porozov
,
T. N.
Savateeva-Lyubimova
,
F.
Peri
.
2017
.
TLR4 signaling pathway modulators as potential therapeutics in inflammation and sepsis.
Vaccines (Basel)
5
:
34
.
12
Zettel
,
K.
,
S.
Korff
,
R.
Zamora
,
A. E.
Morelli
,
S.
Darwiche
,
P. A.
Loughran
,
G.
Elson
,
L.
Shang
,
S.
Salgado-Pires
,
M. J.
Scott
, et al
.
2017
.
Toll-like receptor 4 on both myeloid cells and dendritic cells is required for systemic inflammation and organ damage after hemorrhagic shock with tissue trauma in mice.
Front. Immunol.
8
:
1672
.
13
Gruffaz
,
M.
,
K.
Vasan
,
B.
Tan
,
S.
Ramos da Silva
,
S.-J.
Gao
.
2017
.
TLR4-mediated inflammation promotes KSHV-induced cellular transformation and tumorigenesis by activating the STAT3 pathway.
Cancer Res.
77
:
7094
7108
.
14
Koliaraki
,
V.
,
N.
Chalkidi
,
A.
Henriques
,
C.
Tzaferis
,
A.
Polykratis
,
A.
Waisman
,
W.
Muller
,
D. J.
Hackam
,
M.
Pasparakis
,
G.
Kollias
.
2019
.
Innate sensing through mesenchymal TLR4/MyD88 signals promotes spontaneous intestinal tumorigenesis.
Cell Rep.
26
:
536
545.e4
.
15
Yoder
,
J. A.
,
M. E.
Nielsen
,
C. T.
Amemiya
,
G. W.
Litman
.
2002
.
Zebrafish as an immunological model system.
Microbes Infect.
4
:
1469
1478
.
16
Renshaw
,
S. A.
,
N. S.
Trede
.
2012
.
A model 450 million years in the making: zebrafish and vertebrate immunity.
Dis. Model. Mech.
5
:
38
47
.
17
Hall
,
C.
,
M. V.
Flores
,
A.
Chien
,
A.
Davidson
,
K.
Crosier
,
P.
Crosier
.
2009
.
Transgenic zebrafish reporter lines reveal conserved toll-like receptor signaling potential in embryonic myeloid leukocytes and adult immune cell lineages.
J. Leukoc. Biol.
85
:
751
765
.
18
Burns
,
A. R.
,
K.
Guillemin
.
2017
.
The scales of the zebrafish: host-microbiota interactions from proteins to populations.
Curr. Opin. Microbiol.
38
:
137
141
.
19
Sepulcre
,
M. P.
,
F.
Alcaraz-Pérez
,
A.
López-Muñoz
,
F. J.
Roca
,
J.
Meseguer
,
M. L.
Cayuela
,
V.
Mulero
.
2009
.
Evolution of lipopolysaccharide (LPS) recognition and signaling: fish TLR4 does not recognize LPS and negatively regulates NF-kappaB activation.
J. Immunol.
182
:
1836
1845
.
20
Sullivan
,
C.
,
J.
Charette
,
J.
Catchen
,
C. R.
Lage
,
G.
Giasson
,
J. H.
Postlethwait
,
P. J.
Millard
,
C. H.
Kim
.
2009
.
The gene history of zebrafish tlr4a and tlr4b is predictive of their divergent functions.
J. Immunol.
183
:
5896
5908
.
21
Bates
,
J. M.
,
J.
Akerlund
,
E.
Mittge
,
K.
Guillemin
.
2007
.
Intestinal alkaline phosphatase detoxifies lipopolysaccharide and prevents inflammation in zebrafish in response to the gut microbiota.
Cell Host Microbe
2
:
371
382
.
22
van der Vaart
,
M.
,
J. J.
van Soest
,
H. P.
Spaink
,
A. H.
Meijer
.
2013
.
Functional analysis of a zebrafish myd88 mutant identifies key transcriptional components of the innate immune system.
Dis. Model. Mech.
6
:
841
854
.
23
van der Vaart
,
M.
,
C. J.
Korbee
,
G. E. M.
Lamers
,
A. C.
Tengeler
,
R.
Hosseini
,
M. C.
Haks
,
T. H. M.
Ottenhoff
,
H. P.
Spaink
,
A. H.
Meijer
.
2014
.
The DNA damage-regulated autophagy modulator DRAM1 links mycobacterial recognition via TLR-MYD88 to autophagic defense [corrected]. [Published erratum appears in 2014 Cell Host Microbe 16: 141.]
Cell Host Microbe
15
:
753
767
.
24
Forn-Cuní
,
G.
,
M.
Varela
,
P.
Pereiro
,
B.
Novoa
,
A.
Figueras
.
2017
.
Conserved gene regulation during acute inflammation between zebrafish and mammals.
Sci. Rep.
7
:
41905
.
25
Sepulcre
,
M. P.
,
G.
López-Castejón
,
J.
Meseguer
,
V.
Mulero
.
2007
.
The activation of gilthead seabream professional phagocytes by different PAMPs underlines the behavioural diversity of the main innate immune cells of bony fish.
Mol. Immunol.
44
:
2009
2016
.
26
Ellett
,
F.
,
L.
Pase
,
J. W.
Hayman
,
A.
Andrianopoulos
,
G. J.
Lieschke
.
2011
.
mpeg1 promoter transgenes direct macrophage-lineage expression in zebrafish.
Blood
117
:
e49
e56
.
27
Ferrero
,
G.
,
E.
Gomez
,
S.
Lyer
,
M.
Rovira
,
M.
Miserocchi
,
D. M.
Langenau
,
J. Y.
Bertrand
,
V.
Wittamer
.
2020
.
The macrophage-expressed gene (mpeg) 1 identifies a subpopulation of B cells in the adult zebrafish.
J. Leukoc. Biol.
107
:
431
443
.
28
Kimoto
,
M.
,
K.
Nagasawa
,
K.
Miyake
.
2003
.
Role of TLR4/MD-2 and RP105/MD-1 in innate recognition of lipopolysaccharide.
Scand. J. Infect. Dis.
35
:
568
572
.
29
Qiao
,
L.
,
W.
Yang
,
J.
Fu
,
Z.
Song
.
2013
.
Transcriptome profile of the green odorous frog (Odorrana margaretae).
PLoS One
8
: e75211.
30
Matsunami
,
M.
,
J.
Kitano
,
O.
Kishida
,
H.
Michimae
,
T.
Miura
,
K.
Nishimura
.
2015
.
Transcriptome analysis of predator- and prey-induced phenotypic plasticity in the Hokkaido salamander (Hynobius retardatus).
Mol. Ecol.
24
:
3064
3076
.
31
Nourisson
,
C.
,
M.
Carneiro
,
M.
Vallinoto
,
F.
Sequeira
.
2014
. De novo transcriptome assembly and polymorphism detection in ecologically important widely distributed Neotropical toads from the Rhinella marina species complex (Anura: Bufonidade). In Congreso Latinoamericano de Herpetologia, August 1, 2014–30 September 2014. Cartagena, Colombia. DRYAD, Dataset. Available at: .
32
Arthofer
,
W.
,
B. L.
Banbury
,
M.
Carneiro
,
F.
Cicconardi
,
T. F.
Duda
,
R. B.
Harris
,
D. S.
Kang
,
A. D.
Leaché
,
V.
Nolte
,
C.
Nourisson
, et al
Genomic Resources Development Consortium
.
2015
.
Genomic resources notes accepted 1 August 2014-30 September 2014.
Mol. Ecol. Resour.
15
:
228
229
.
33
Liu
,
Y.
,
B.
Schmidt
,
D. L.
Maskell
.
2010
.
MSAProbs: multiple sequence alignment based on pair hidden Markov models and partition function posterior probabilities.
Bioinformatics
26
:
1958
1964
.
34
Tamura
,
K.
,
G.
Stecher
,
D.
Peterson
,
A.
Filipski
,
S.
Kumar
.
2013
.
MEGA6: molecular evolutionary genetics analysis version 6.0.
Mol. Biol. Evol.
30
:
2725
2729
.
35
Guindon
,
S.
,
J. F.
Dufayard
,
V.
Lefort
,
M.
Anisimova
,
W.
Hordijk
,
O.
Gascuel
.
2010
.
New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0.
Syst. Biol.
59
:
307
321
.
36
Le
,
S. Q.
,
O.
Gascuel
.
2010
.
Accounting for solvent accessibility and secondary structure in protein phylogenetics is clearly beneficial.
Syst. Biol.
59
:
277
287
.
37
Le
,
S. Q.
,
O.
Gascuel
.
2008
.
An improved general amino acid replacement matrix.
Mol. Biol. Evol.
25
:
1307
1320
.
38
Hordijk
,
W.
,
O.
Gascuel
.
2005
.
Improving the efficiency of SPR moves in phylogenetic tree search methods based on maximum likelihood.
Bioinformatics
21
:
4338
4347
.
39
Anisimova
,
M.
,
O.
Gascuel
.
2006
.
Approximate likelihood-ratio test for branches: a fast, accurate, and powerful alternative.
Syst. Biol.
55
:
539
552
.
40
Akaike
,
H.
1998
.
Information theory and an extension of the maximum likelihood principle.
In
Selected Papers of Hirotugu Akaike.
K.
Tanabe
,
G.
Kitagawa
, eds.
Springer
,
New York
, p.
199
213
.
41
Waterhouse
,
A. M.
,
J. B.
Procter
,
D. M. A.
Martin
,
M.
Clamp
,
G. J.
Barton
.
2009
.
Jalview Version 2--a multiple sequence alignment editor and analysis workbench.
Bioinformatics
25
:
1189
1191
.
42
Herrero
,
J.
,
M.
Muffato
,
K.
Beal
,
S.
Fitzgerald
,
L.
Gordon
,
M.
Pignatelli
,
A. J.
Vilella
,
S. M. J.
Searle
,
R.
Amode
,
S.
Brent
, et al
.
2016
.
Ensembl comparative genomics resources.
Database (Oxford)
2016
: bav096.
43
Farnsworth
,
D. R.
,
L. M.
Saunders
,
A. C.
Miller
.
2020
.
A single-cell transcriptome atlas for zebrafish development.
Dev. Biol.
459
:
100
108
.
44
Zheng
,
G. X. Y.
,
J. M.
Terry
,
P.
Belgrader
,
P.
Ryvkin
,
Z. W.
Bent
,
R.
Wilson
,
S. B.
Ziraldo
,
T. D.
Wheeler
,
G. P.
McDermott
,
J.
Zhu
, et al
.
2017
.
Massively parallel digital transcriptional profiling of single cells.
Nat. Commun.
8
:
14049
.
45
Satija
,
R.
,
J. A.
Farrell
,
D.
Gennert
,
A. F.
Schier
,
A.
Regev
.
2015
.
Spatial reconstruction of single-cell gene expression data.
Nat. Biotechnol.
33
:
495
502
.
46
R Core Team
.
2017
.
R: A Language and Environment for Statistical Computing.
R Foundation for Statistical Computing
,
Vienna, Austria
.
47
Jao
,
L.-E.
,
S. R.
Wente
,
W.
Chen
.
2013
.
Efficient multiplex biallelic zebrafish genome editing using a CRISPR nuclease system.
Proc. Natl. Acad. Sci. USA
110
:
13904
13909
.
48
Konantz
,
J.
,
C. L.
Antos
.
2014
.
Reverse genetic morpholino approach using cardiac ventricular injection to transfect multiple difficult-to-target tissues in the zebrafish larva.
J. Vis. Exp.
DOI: 10.3791/51595.
49
Vanhauwaert
,
S.
,
G.
Van Peer
,
A.
Rihani
,
E.
Janssens
,
P.
Rondou
,
S.
Lefever
,
A.
De Paepe
,
P. J.
Coucke
,
F.
Speleman
,
J.
Vandesompele
,
A.
Willaert
.
2014
.
Expressed repeat elements improve RT-qPCR normalization across a wide range of zebrafish gene expression studies.
PLoS One
9
: e109091.
50
van Soest
,
J. J.
,
O. W.
Stockhammer
,
A.
Ordas
,
G. V.
Bloemberg
,
H. P.
Spaink
,
A. H.
Meijer
.
2011
.
Comparison of static immersion and intravenous injection systems for exposure of zebrafish embryos to the natural pathogen Edwardsiella tarda.
BMC Immunol.
12
:
58
.
51
Sun
,
Y.
,
Y.
Huang
,
X.
Li
,
C. C.
Baldwin
,
Z.
Zhou
,
Z.
Yan
,
K. A.
Crandall
,
Y.
Zhang
,
X.
Zhao
,
M.
Wang
, et al
.
2016
.
Fish-T1K (Transcriptomes of 1,000 Fishes) project: large-scale transcriptome data for fish evolution studies.
Gigascience
5
:
18
.
52
Woods
,
I. G.
,
C.
Wilson
,
B.
Friedlander
,
P.
Chang
,
D. K.
Reyes
,
R.
Nix
,
P. D.
Kelly
,
F.
Chu
,
J. H.
Postlethwait
,
W. S.
Talbot
.
2005
.
The zebrafish gene map defines ancestral vertebrate chromosomes.
Genome Res.
15
:
1307
1314
.
53
Bernut
,
A.
,
J.-L.
Herrmann
,
K.
Kissa
,
J.-F.
Dubremetz
,
J.-L.
Gaillard
,
G.
Lutfalla
,
L.
Kremer
.
2014
.
Mycobacterium abscessus cording prevents phagocytosis and promotes abscess formation.
Proc. Natl. Acad. Sci. USA
111
:
E943
E952
.
54
Heumann
,
D.
,
R.
Lauener
,
B.
Ryffel
.
2003
.
The dual role of LBP and CD14 in response to Gram-negative bacteria or Gram-negative compounds.
J. Endotoxin Res.
9
:
381
384
.
55
Triantafilou
,
M.
,
K.
Triantafilou
.
2002
.
Lipopolysaccharide recognition: CD14, TLRs and the LPS-activation cluster.
Trends Immunol.
23
:
301
304
.
56
Kim
,
S. J.
,
H. M.
Kim
.
2017
.
Dynamic lipopolysaccharide transfer cascade to TLR4/MD2 complex via LBP and CD14.
BMB Rep.
50
:
55
57
.
57
Shapiro
,
R. A.
,
M. D.
Cunningham
,
K.
Ratcliffe
,
C.
Seachord
,
J.
Blake
,
J.
Bajorath
,
A.
Aruffo
,
R. P.
Darveau
.
1997
.
Identification of CD14 residues involved in specific lipopolysaccharide recognition.
Infect. Immun.
65
:
293
297
.
58
Juan
,
T. S.-C.
,
E.
Hailman
,
M. J.
Kelley
,
S. D.
Wright
,
H. S.
Lichenstein
.
1995
.
Identification of a domain in soluble CD14 essential for lipopolysaccharide (LPS) signaling but not LPS binding.
J. Biol. Chem.
270
:
17237
17242
.
59
Zhou
,
Y.
,
Q.
Liang
,
W.
Li
,
Y.
Gu
,
X.
Liao
,
W.
Fang
,
X.
Li
.
2016
.
Characterization and functional analysis of toll-like receptor 4 in Chinese soft-shelled turtle Pelodiscus sinensis.
Dev. Comp. Immunol.
63
:
128
135
.
60
Loes
,
A. N.
,
J. T.
Bridgham
,
M. J.
Harms
.
2018
.
Coevolution of the toll-like receptor 4 complex with calgranulins and lipopolysaccharide.
Front. Immunol.
9
:
304
.
61
Divanovic
,
S.
,
A.
Trompette
,
S. F.
Atabani
,
R.
Madan
,
D. T.
Golenbock
,
A.
Visintin
,
R. W.
Finberg
,
A.
Tarakhovsky
,
S. N.
Vogel
,
Y.
Belkaid
, et al
.
2005
.
Negative regulation of toll-like receptor 4 signaling by the toll-like receptor homolog RP105.
Nat. Immunol.
6
:
571
578
.
62
Muroi
,
M.
,
T.
Ohnishi
,
K.
Tanamoto
.
2002
.
MD-2, a novel accessory molecule, is involved in species-specific actions of Salmonella lipid A.
Infect. Immun.
70
:
3546
3550
.
63
Hoshino
,
K.
,
O.
Takeuchi
,
T.
Kawai
,
H.
Sanjo
,
T.
Ogawa
,
Y.
Takeda
,
K.
Takeda
,
S.
Akira
.
1999
.
Cutting edge: toll-like receptor 4 (TLR4)-deficient mice are hyporesponsive to lipopolysaccharide: evidence for TLR4 as the Lps gene product.
J. Immunol.
162
:
3749
3752
.
64
Meijer
,
A. H.
,
S. F.
Gabby Krens
,
I. A.
Medina Rodriguez
,
S.
He
,
W.
Bitter
,
B.
Ewa Snaar-Jagalska
,
H. P.
Spaink
.
2004
.
Expression analysis of the toll-like receptor and TIR domain adaptor families of zebrafish.
Mol. Immunol.
40
:
773
783
.
65
Jault
,
C.
,
L.
Pichon
,
J.
Chluba
.
2004
.
Toll-like receptor gene family and TIR-domain adapters in Danio rerio.
Mol. Immunol.
40
:
759
771
.
66
Vaure
,
C.
,
Y.
Liu
.
2014
.
A comparative review of toll-like receptor 4 expression and functionality in different animal species.
Front. Immunol.
5
:
316
.
67
Yang
,
D.
,
X.
Zheng
,
S.
Chen
,
Z.
Wang
,
W.
Xu
,
J.
Tan
,
T.
Hu
,
M.
Hou
,
W.
Wang
,
Z.
Gu
, et al
.
2018
.
Sensing of cytosolic LPS through caspy2 pyrin domain mediates noncanonical inflammasome activation in zebrafish.
Nat. Commun.
9
:
3052
.
68
Forn-Cuní
,
G.
,
A. H.
Meijer
,
M.
Varela
.
2019
.
Zebrafish in inflammasome research.
Cells
8
:
901
.
69
Mazon-Moya
,
M. J.
,
A. R.
Willis
,
V.
Torraca
,
L.
Boucontet
,
A. R.
Shenoy
,
E.
Colucci-Guyon
,
S.
Mostowy
.
2017
.
Septins restrict inflammation and protect zebrafish larvae from Shigella infection.
PLoS Pathog.
13
: e1006467.
70
Jones
,
J. D. G.
,
R. E.
Vance
,
J. L.
Dangl
.
2016
.
Intracellular innate immune surveillance devices in plants and animals.
Science
354
:
aaf6395
.
71
Bernheiden
,
M.
,
J.-M.
Heinrich
,
G.
Minigo
,
C.
Schütt
,
F.
Stelter
,
M.
Freeman
,
D.
Golenbock
,
R. S.
Jack
.
2001
.
LBP, CD14, TLR4 and the murine innate immune response to a peritoneal Salmonella infection.
J. Endotoxin Res.
7
:
447
450
.
72
Ehrchen
,
J. M.
,
C.
Sunderkötter
,
D.
Foell
,
T.
Vogl
,
J.
Roth
.
2009
.
The endogenous toll-like receptor 4 agonist S100A8/S100A9 (calprotectin) as innate amplifier of infection, autoimmunity, and cancer.
J. Leukoc. Biol.
86
:
557
566
.
73
Ohto
,
U.
,
K.
Fukase
,
K.
Miyake
,
T.
Shimizu
.
2012
.
Structural basis of species-specific endotoxin sensing by innate immune receptor TLR4/MD-2.
Proc. Natl. Acad. Sci. USA
109
:
7421
7426
.
74
Walsh
,
C.
,
M.
Gangloff
,
T.
Monie
,
T.
Smyth
,
B.
Wei
,
T. J.
McKinley
,
D.
Maskell
,
N.
Gay
,
C.
Bryant
.
2008
.
Elucidation of the MD-2/TLR4 interface required for signaling by lipid IVa.
J. Immunol.
181
:
1245
1254
.
75
Anderson
,
J. A.
,
A. N.
Loes
,
G. L.
Waddell
,
M. J.
Harms
.
2019
.
Tracing the evolution of novel features of human toll-like receptor 4.
Protein Sci.
28
:
1350
1358
.
76
Iliev
,
D. B.
,
J. C.
Roach
,
S.
Mackenzie
,
J. V.
Planas
,
F. W.
Goetz
.
2005
.
Endotoxin recognition: in fish or not in fish?
FEBS Lett.
579
:
6519
6528
.
77
Berczi
,
I.
,
L.
Bertók
,
T.
Bereznai
.
1966
.
Comparative studies on the toxicity of Escherichia coli lipopolysaccharide endotoxin in various animal species.
Can. J. Microbiol.
12
:
1070
1071
.
78
Stafford
,
J. L.
,
K. K.
Ellestad
,
K. E.
Magor
,
M.
Belosevic
,
B. G.
Magor
.
2003
.
A toll-like receptor (TLR) gene that is up-regulated in activated goldfish macrophages.
Dev. Comp. Immunol.
27
:
685
698
.
79
MacKenzie
,
S.
,
J. V.
Planas
,
F. W.
Goetz
.
2003
.
LPS-stimulated expression of a tumor necrosis factor-alpha mRNA in primary trout monocytes and in vitro differentiated macrophages.
Dev. Comp. Immunol.
27
:
393
400
.
80
Pelegrín
,
P.
,
J.
García-Castillo
,
V.
Mulero
,
J.
Meseguer
.
2001
.
Interleukin-1beta isolated from a marine fish reveals up-regulated expression in macrophages following activation with lipopolysaccharide and lymphokines.
Cytokine
16
:
67
72
.
81
Ishii
,
A.
,
M.
Kawasaki
,
M.
Matsumoto
,
S.
Tochinai
,
T.
Seya
.
2007
.
Phylogenetic and expression analysis of amphibian Xenopus toll-like receptors.
Immunogenetics
59
:
281
293
.
82
Park
,
B. S.
,
D. H.
Song
,
H. M.
Kim
,
B.-S.
Choi
,
H.
Lee
,
J.-O.
Lee
.
2009
.
The structural basis of lipopolysaccharide recognition by the TLR4-MD-2 complex.
Nature
458
:
1191
1195
.
83
Hughes
,
L. C.
,
G.
Ortí
,
Y.
Huang
,
Y.
Sun
,
C. C.
Baldwin
,
A. W.
Thompson
,
D.
Arcila
,
R.
Betancur-R
,
C.
Li
,
L.
Becker
, et al
.
2018
.
Comprehensive phylogeny of ray-finned fishes (Actinopterygii) based on transcriptomic and genomic data.
Proc. Natl. Acad. Sci. USA
115
:
6249
6254
.
84
Hedges
,
S. B.
,
J.
Dudley
,
S.
Kumar
.
2006
.
TimeTree: a public knowledge-base of divergence times among organisms.
Bioinformatics
22
:
2971
2972
.

The authors have no financial conflicts of interest.

Supplementary data