Toll-like Receptors of the Ascidian Ciona intestinalis

Key transmembrane proteins in the innate immune system, Toll-like receptors (TLRs), have been suggested to occur in the genome of non-mammalian organisms including invertebrates. However, authentic invertebrate TLRs have been neither structurally nor functionally investigated. In this paper, we originally present the structures, localization, ligand recognition, activities, and inflammatory cytokine production of all TLRs of the ascidian Ciona intestinalis, designated as Ci-TLR1 and Ci-TLR2. The amino acid sequence of Ci-TLR1 and Ci-TLR2 were found to possess unique structural organization with moderate sequence similarity to functionally characterized vertebrate TLRs. ci-tlr1 and ci-tlr2 genes were expressed predominantly in the stomach and intestine as well as in hemocytes. Ci-TLR1 and Ci-TLR2 expressed in HEK293 cells, unlike vertebrate TLRs, were localized to both the plasma membrane and endosomes. Intriguingly, both Ci-TLR1 and Ci-TLR2 stimulate NF-κB induction in response to multiple pathogenic ligands such as double-stranded RNA, and bacterial cell wall components that are differentially recognized by respective vertebrate TLRs, revealing that Ci-TLRs recognize broader pathogen-associated molecular patterns than vertebrate TLRs. The Ci-TLR-stimulating pathogenic ligands also induced the expression of Ci-TNFα in the intestine and stomach where Ci-TLRs are expressed. These results provide evidence that the TLR-triggered innate immune systems are essentially conserved in ascidians, and that Ci-TLRs possess “hybrid” biological and immunological functions, compared with vertebrate TLRs. Moreover, it is presumed that chordate TLR ancestors also acquired the Ci-TLR-like multiple cellular localization and pathogen-associated molecular pattern recognition.

Toll-like receptors (TLRs) 2 play pivotal roles in mammalian host defenses via the innate immune system. All TLRs are type I transmembrane proteins, containing an intracellular Toll/interleukin-1 receptor (TIR) domain and leucine-rich repeat (LRR) motifs in the extracellular domain (1)(2)(3). The diversity in the numbers and organization of LRR domains enables the specific and sensitive recognition of pathogenic molecular species by respective TLRs (1)(2)(3). TLRs are expressed mainly in immune cells such as lymphocytes, macrophages, and dendritic cells. TLR1, TLR2, TLR4, TLR5, and TLR6 are responsible for recognition of extracellular microbial pathogenic components on plasma membranes, whereas TLR3, TLR7, TLR8, and TLR9, present on endosomes, participate in defensive responses to viral DNA or RNA incorporated into the cytoplasm (1)(2)(3)(4)(5). Stimulation of TLRs initiates signal transduction pathways via adaptor proteins (MyD88, TIRAP, TRIF, and TRAM) followed by activation of a wide range of inducible transcriptional factors such as NF-B, AP-1 and IRF, leading to production of inflammatory cytokines, chemokines, and type I interferon (1)(2)(3)(4)(5).
Recent genomic surveys detected a great variety of TLR-like genes in fish and deuterostome invertebrates: ascidian (2 TLRrelated genes), amphioxus (72 TLR-like genes), sea urchin (222 TLR-like genes), and Fugu (23 TLR-like genes) (6 -9). Furthermore, primary structures, localization, authentic ligands, and/or the resultant signaling of several fish and cyclostome TLRs have been documented (9,10). These findings support the view that the TLR family is also involved in the innate immune system of non-mammalian animal species, and diverges in the number, structural organization, and biological roles among animal species. In insects and hydra, pathogenic components for Drosophila Toll and hydra Toll-related receptor were identified (11)(12)(13). However, the molecular mechanisms underlying their ligand recognition and signal transduction are distinct from those of vertebrate TLRs, leading to the presumption that they are not regarded as models of the vertebrate-type innate immune system. Moreover, the active protein forms, pathogen-associated molecular patterns (PAMPs), cellular localization of respective deuterostome invertebrate TLRs, or TLR-like receptors and an evolutionary origin of the vertebrate TLR family have never been structurally or functionally elucidated.
The ascidian Ciona intestinalis is regarded as the invertebrate chordate most closely related to vertebrates (14,15). Moreover, C. intestinalis, unlike traditional protostome model animals including Drosophila melanogaster and Caenorhabditis elegans, has been found to possess the developmental, neuropeptidergic, and endocrine systems prototypic to those of vertebrates (14 -17), indicating the possibility that prototypes of vertebrate innate immune systems are also conserved in the protochordate, and that identification of Ciona TLRs provides critical clues to the investigation of the evolutionary origin of TLR pathways.  In the present study, we identified the structures, tissue distribution, cellular localizations, and ligand-specific activities of all Ciona TLRs, Ci-TLR1 and -2. Our data verified both similar and unique propensities of Ci-TLRs, compared with vertebrate TLRs, suggesting the original features of chordate TLRs. To the best of our knowledge, this is the first report on molecular and functional characterization of TLRs that are localized to the plasma membrane and endosomes and recognize multiple PAMPs.

EXPERIMENTAL PROCEDURES
Animals-Adults of C. intestinalis were cultivated and collected at the Maizuru Fisheries Research Station of Kyoto University, and kept in seawater at 18°C.
Cloning of Ci-TLRs-Total RNA was extracted from Ciona hemocytes. The complete sequences of the ci-tlr genes were obtained by 5Ј-and 3Ј-rapid amplification of the cDNA ends using a 3Ј-and 5Ј-rapid amplification of the cDNA ends system for the rapid amplification of cDNA ends kit (Invitrogen). The products were sequenced with primers used for RT-PCR and universal primers (SP6 and T7 primers).
Molecular Phylogenetic Tree of the TLR Family-The amino acid sequences were aligned using the CLUSTAL program. The alignment was checked by hand. After removing gaps, the verified alignments were used to construct phylogenetic trees. The trees were calculated using the MEGA program based on the neighbor-joining method.  (18).
Whole Mount in Situ Hybridization of Ci-TLR mRNAs-The Ci-TLR1 cDNA fragment (nucleotides 1115-1654) and the Ci-TLR2 cDNA fragment (nucleotides 1614 -2356) were inserted into the pCRII TOPO vector according to the manufacturer's instruction. Whole mount in situ hybridization was performed as previously reported (19,20).
Cell Culture and Transfection-HEK293 MSR cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum at 37°C in a humidified atmosphere of 5% carbon dioxide. The cells were transfected using F1-reagent (Targeting System, El Cajon, CA) with the pcDNA6/myc-His vector A containing no insert, ci-tlr1, or ci-tlr2 (Invitrogen). The stable transfectants were selected in medium containing 20 g/ml blasticidin. Each hTLR-stable HEK293 transfectant was purchased from InvivoGen (San Diego, CA).
NF-B Reporter Assay-Stable transfectants with empty vector, ci-tlr1 or ci-tlr2 were transiently transfected using F1-reagent with the NF-B-responsive reporter gene pNiFty-SEAP (InvivoGen). Induction of NF-B in these transfectants was evaluated in the presence or absence of each TLR ligand according to the manufacturer's instructions.
Quantification of Ci-TNF␣ in Ciona Tissues-Adult tissue was dissociated into cells, treated with each ligand, and incubated for 24 h at 25°C. Real-time PCR for Ci-TNF␣ using the SYBR Green master mixture (Applied Biosystem, Foster City, CA) was performed three times. As an internal control, Ci-␤actin was employed. All primers for real time PCR are shown under supplemental Table S1. To calculate relative quantification of gene expression, we used the comparative C T method.
Statistical Analysis-Data are expressed as mean Ϯ S.D. Differences in the mean values assessed for significance by oneway analysis of variance, Scheffe's F-test (see Fig. 6), and Fisher's PLSD (see Fig. 7). p values (p Ͻ 0.05) were considered significant.

RESULTS
Identification of ci-tlr Genes-We initially performed BLAST searches on the genome data base of C. intestinalis with each typical domain of various vertebrate TLRs as a query, and detected only two independent clones, gene numbers ci0100137574 and cibd077g18. This is compatible with a previous Ciona genome survey (6), and suggested that only two TLRlike genes are present in C. intestinalis. The two full-length cDNA sequences were cloned from Ciona hemocytes, and the open reading frames of ci0100137574 and cibd077g18 were found to encode 883 and 948 amino acids, respectively. SMART protein domain analyses revealed that both of the deduced proteins harbor an intracellular TIR domain, a transmembrane domain, and multiple extracellular LRRs, which are reminis-cent of vertebrate TLRs (Fig. 1). We thus designated the two genes as ci-tlr1 and ci-tlr2, respectively. Ci-TLR1 and Ci-TLR2 bear 7 and 13 putative LRRs, respectively (Fig. 1). Moreover, Ci-TLR1 contains one LRR C-terminal domain, which is flanked by a cysteine-rich domain, although three LRR C-terminal domains are present in Ci-TLR2 (Fig. 1). Sequence homology of Ci-TLRs to vertebrate counterparts is confounding, compared with vertebrate TLRs (Fig. 1). The overall amino acid sequences of Ci-TLR1 and -2 were most homologous to human TLR (hTLR)7 (26%) and hTLR8 (26%), respectively. The TIR domains of Ci-TLR1 and Ci-TLR2 were most similar to hTLR4 and hTLR6, respectively. The LRRs at 1-330 in Ci-TLR1 are most similar to hTLR5, whereas the LRR at amino acids 331-681 displays the highest similarity to hTLR9 in the corresponding region. In contrast, each portion of the LRR in Ci-TLR2 was most similar to hTLR8. The phylogenetic tree of TLRs and their putative orthologs (Fig. 2) demonstrate that Ci-TLRs undoubtedly belong to the vertebrate TLR family, and suggest the pivotal positions of Ci-TLRs as prototypes of vertebrate TLRs.
Expression of Ci-TLRs-Whole mount in situ hybridization demonstrated that ci-tlr1 and ci-tlr2 genes were expressed intensively in the stomach, intestine (Fig. 3, A and C), and numerous, but not all, hemocytes (Fig. 3, B and D) in Ciona juveniles. RT-PCR also showed that both Ci-TLRs were detected in various adult tissues with prominent expression in the stomach, and/or anterior and middle intestines (Fig. 4), whereas Ci-TLR2 expression was slightly less intensive than that of Ci-TLR1 (Fig. 4).
Cellular Localization of Ci-TLRs-To verify the intracellular localization of Ci-TLRs, we performed double immunofluorescent confocal microscopy observations of localization of the myc-tagged ci-tlr1 or ci-tlr2 stably expressed in HEK293 MSR cells (Fig. 5, A and D). Moreover, syntaxin-7, which was shown   OCTOBER 2, 2009 • VOLUME 284 • NUMBER 40

Ascidan TLRs
to be co-localized with human TLR3 (36), was stained as a marker for late endosomes (Fig. 5, B and E). Intriguingly, Ci-TLR1 and Ci-TLR2, unlike any mammalian TLRs, were found to be present on both the plasma membrane and a number of late endosomes (Fig. 5, C and F), whereas more intense expres-sion of Ci-TLR2 was detected in endosomes than in the plasma membrane (Fig. 5, C and F).
Elucidation of Pathogenic Ligands of Ci-TLRs-All mammalian TLRs eventually transactivate a transcriptional factor, NF-B, in response to their specific ligands (1-3). Moreover, the homologous ascidian genes encoding the innate immunity relevant adapter proteins, except MD-2 and transcriptional factors including NF-B for TLR-triggered signaling, were detected in the genome (6). Therefore, we employed a NF-B-responsive secreted alkaline phosphatase reporter assay to elucidate PAMPs of Ci-TLRs. HEK293 MSR cells were stably transfected with an empty, ci-tlr1, or ci-tlr2inserted pcDNA6.1 vector, and treated with 16 known TLR ligands ( Table 1) at concentrations that were found to exhibit significant activation of the respective hTLRs (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). Unexpectedly, Ci-TLR1 and Ci-TLR2 activated NF-B in response to multiple TLR ligands (Fig. 6), which are recognized by different mammalian TLRs. Zymosan (Saccharomyces cerevisiae cell wall) for hTLR2, heat-killed Legionella pneumophila (HKLP, a Gram-negative bacterium) for hTLR2, double-stranded RNA, poly(I-C) for hTLR3, S. typhimurium flagellin (the major component of the bacterial flagellar filament) for hTLR5 elicited a dose-dependent induction of NF-B in the ci-tlr1-or ci-tlr2expressing cells (Fig. 6). Additionally, the potencies of these four ligands at Ci-TLRs were comparable with those at hTLR2, -3, and -5 (results not shown). In contrast, no significant responses to these ligands were detected in empty vector-transfected cells (supplemental Fig. S1), and other ligands failed to induce NF-B transactivation in ci-tlr1-or ci-tlr2-transfected cells (supplemental Figs. S2 and S3). Notably, both Ci-TLRs showed equipotent NF-B activation in response to the same ligands (Fig. 6). These data clarified PAMPs of Ci-TLRs, and indicated that Ci-TLRs have more extensive PAMPs than vertebrate TLRs. In addition, these functional analyses proved that   Ci-TLR2 (D) were observed by immunostaining using anti-myc antibody followed by Alexa 488-conjugated secondary antibody (green). Late endosomes (red) were visualized with anti-syntaxin 7 antibody followed by Alexa 594-conjugated secondary antibody (B and E). The merged images clarify that Ci-TLR1 (C) and 2 (F) are localized to endosomes as well as plasma membrane. Scale bar, 10 m. Ci-TLRs, like vertebrate TLRs (1-3), directly recognize various pathogenic compounds and trigger signal transduction, which is in contrast with functionally characterized Drosophila Toll and hydra Toll-related receptors that require the specific associated proteins for ligand recognition and/or initiation of signal transduction cascades (11)(12)(13). These findings confirmed that Ci-TLRs share the essential ligand recognition and signaling mechanisms with the vertebrate TLR family. Induction of TNF␣ by Ci-TLR Ligands in Ciona Tissues-In the mammalian TLR signaling pathway, activation of NF-B leads to an increase in gene expression of an inflammatory cytokine, TNF␣. We examined whether gene expression of Ci-TNF␣ (37) was up-regulated by the Ci-TLR ligands (Fig. 6) in the intestine or stomach, where ci-tlr1 and ci-tlr2 are abundantly expressed (Figs. 3 and 4). Real-time PCR revealed that poly(I-C) induced ϳ4and 10-fold Ci-TNF␣ expression in the anterior (Fig. 7A) and middle (Fig. 7B) intestine, respectively, compared with controls. Flagellin caused ϳ6-fold and more than 10-fold elevation of Ci-TNF␣ in the stomach (Fig. 7D) and middle intestine (Fig. 7E), respectively. In contrast, neither poly(I-C) nor flagellin induced any significant Ci-TNF␣ expression in the Ciona posterior intestine (Fig. 7, C and  F) where expression of ci-tlrs was not detected (Figs. 3 and 4). These results suggest that Ci-TLRs respond to the pathogenic ligands in Ciona tissues.

DISCUSSION
The innate immune system is expected to play pivotal roles in defensive responses to pathogens in invertebrates, where adaptive immune systems are absent. In vertebrates, the innate immunity is functionally linked to the adaptive immunity (1, 2). However, high molecular divergence in the numbers and organizations of the LRRs among fish and invertebrate TLR (or TLR-like) genes hinders sequence homology-based elucidation or prediction of their crucial functions: PAMPs, cellular localization, and signaling pathways (6 -9). To address these issues, we explored all TLRs of the ascidian C. intestinalis the closest group of vertebrates.
The striking feature is that Ci-TLRs show unprecedented cellular localization and PAMPs. Ci-TLRs are localized to both the plasma membranes and late endosomes (Fig. 5), whereas vertebrate TLRs have so far been shown to be present exclusively either on the plasma membrane (TLR1, -2, -4, -5, and -6) or on endosomes (TLR3, -7, -8, and -9). Trafficking and localization of hTLR3 and -7 to endosomes are likely to be regulated by several regions between the transmembrane and TIR domains (4,5), although the underlying molecular mechanism has yet to be elucidated. Because Ci-TLRs share no significant sequence homology with hTLRs in the corresponding region,  the molecular mechanism underlying the multiple localization of Ci-TLR awaits further study. Also of particular biochemical and immunological significance is the multiple PAMP recognition by Ci-TLRs. Ci-TLR1 and -2 are responsive to identical multiple PAMPs (Fig. 6), which are differentially recognized by respective vertebrate TLRs: poly(I-C) for hTLR3, flagellin for hTLR5, zymosan for hTLR2, and HKLP for hTLR2 (1-3, 9, 21, 22, 24, 26, 29). Such PAMPs of Ci-TLRs were not predicted solely on the basis of sequence homology and molecular phylogenetic analyses (6 -8) (Fig. 2) due to no consistency of PAMP specificity with their sequence homology to hTLRs (Figs. 1 and  6). For instance, Ci-TLRs elicit a prominent response to poly(I-C) (Fig. 6) that is the specific ligand for hTLR3 (26) despite no significant sequence homology of the LRRs of Ci-TLRs to those of hTLR3. Likewise, the Ci-TLR2 is devoid of any response to single-stranded RNA (supplemental Fig. S3), a typical pathogenic ligand for hTLR8 (see Table 1), whereas the LRRs of Ci-TLR2 display the highest amino acid similarity to those of hTLR8. However, PAMPs of Ci-TLRs are in good agreement with their cellular localization, given that poly(I-C) is recognized by an endosome TLR, hTLR3, and hTLR2 and hTLR5 interact with flagellin, zymosan, and HKLP on the cell surface. Collectively, these results show that Ci-TLRs are "functionally hybrid TLRs" of vertebrate cell surface TLRs and endosome TLRs. Furthermore, combined with the critical phylogenetic position of C. intestinalis as the closest relative to vertebrates (14 -16), the hybrid cellular localization and PAMPs of Ci-TLRs lead to the evolutionary view that ancestral TLRs, like Ci-TLRs, might have recognized multiple PAMPs and been present on both cell surfaces and endosomes in common antecedent species of ascidians and vertebrates, and the essential specific ligand recognition and localization of current vertebrate TLRs might have been established during the evolution of chordates. Further functional analyses of sea urchin, Amphioxus, and fish TLR(-like) genes will contribute to not only elucidation of their biological roles but also investigation of molecular and functional divergence of deuterostome TLRs or TLR-like genes.
hTLR2 is shown to be activated by multiple pathogenic ligands (Table 1). Intriguingly, two ligands for TLR2, zymosan and HKLP, exhibit prominent activation Ci-TLRs (Fig. 6), whereas other hTLR2 ligands including HKLM, 1,3-␤-glucan, and HKSA are devoid of any activity at Ci-TLRs (supplemental Figs. S2 and S3). These findings support the view that PAMP recognition of TLRs is highly likely to be diverged in accordance with the difference in spectra of pathogenic microbes for the host organisms; pathogens for ascidians in the ocean should not completely coincide with those for mammals on the land. In this regard, the present data emphasize the importance in elucidation of PAMPs of TLRs in non-mammalian organisms of interest by functional expression assays (Fig. 6), not only by sequence homology analyses including molecular phylogenetic analyses.
Also of interest is that Ci-TLRs exhibited the same PAMP recognition (Fig. 6) despite the considerable difference in the number and organization of LRRs (Fig. 1). In the past few years, the tertiary structures of several TLR-ligand complexes have been resolved (38 -41). Elucidation of tertiary structures of Ci-TLRs bound to each of their ligands is expected to provide novel insights into PAMP recognition modes of TLRs and the evolutionary process of ligand recognition of chordate TLRs.
Vertebrate TLRs are present in various tissues, such as lung, kidney, small intestine, stomach, and testis as well as in hemocytes (1-3, 9, 10). ci-tlrs are also expressed predominantly in the stomach, intestine, and the hemocytes (Figs. 3  and 4). Moreover, up-regulation of TNF␣ was demonstrated in the Ciona stomach and anterior and middle intestine in response to the ligands of Ci-TLRs (Fig. 7). Unfortunately, neither gene silencing nor gene knock-out techniques for adult ascidians including C. intestinalis have ever been established, and we cannot absolutely exclude the possibility that other receptors than Ci-TLRs are also involved in the induction of Ci-TNF␣ in these tissues. Instead, we also showed that no up-regulation of TNF␣ was detected in the posterior intestine where ci-tlrs were not expressed (see Figs. 3, 4, and 7). Collectively, these results support the view that the TLR-mediated innate immune functions are conserved in ascidian tissues. Furthermore, the present study suggests that ancestral TLRs were originally responsible for the innate immune systems mainly in the digestive tissues and in hemocytes, leading to expansion of the TLR-mediated innate immunity system into more various tissues. Intriguingly, Ci-TLRs display moderate differences in their expression (see Fig. 4) despite almost identical ligand recognition (see Figs. 6 and 7); Ci-TLR1 is expressed much more in the anterior intestine, compared with Ci-TLR2 (see Fig. 4). Recently, TLRs have also been shown to recognize several high concentration endogenous ligands including heme and endogenous double-stranded RNA resulting from apoptosis (42)(43)(44). These findings suggest another biological role for TLRs as a sensor of the unusual generation of endogenous components. Together, it is presumed that the expression profiles for Ci-TLRs are implicated with their different functions in recognition of endogenous ligands. Elucidation of the biological significance of Ci-TLRs in the innate immune system and their ultimate endogenous roles is currently in progress.
In conclusion, we have identified novel TLRs, Ci-TLRs, in the ascidian C. intestinalis. The unique hybrid cellular localization and PAMP recognition of Ci-TLRs pave the way to investigate an evolutionary origin of vertebrate TLRs and species-specific molecular divergences of TLRs that are correlated with lifespans and environmental pathogens specific to each organism. Our study also sheds new light on TLRs of lower animals. For instance, unique and diverse PAMP recognition modes of TLRs in each species are expected to provide novel targets for highly specific agonists and antagonists of TLRs in a certain animal species, which will lead to the unprecedented protection and/or expansion of endangered species or elimination of invading organisms with high efficiency and species specificity. Such studies are underway.