Reptile Toll-like receptor 5 unveils adaptive evolution of bacterial flagellin recognition

Toll-like receptors (TLR) are ancient innate immune receptors crucial for immune homeostasis and protection against infection. TLRs are present in mammals, birds, amphibians and fish but have not been functionally characterized in reptiles despite the central position of this animal class in vertebrate evolution. Here we report the cloning, characterization, and function of TLR5 of the reptile Anolis carolinensis (Green Anole lizard). The receptor (acTLR5) displays the typical TLR protein architecture with 22 extracellular leucine rich repeats flanked by a N- and C-terminal leucine rich repeat domain, a membrane-spanning region, and an intracellular TIR domain. The receptor is phylogenetically most similar to TLR5 of birds and most distant to fish TLR5. Transcript analysis revealed acTLR5 expression in multiple lizard tissues. Stimulation of acTLR5 with TLR ligands demonstrated unique responsiveness towards bacterial flagellin in both reptile and human cells. Comparison of acTLR5 and human TLR5 using purified flagellins revealed differential sensitivity to Pseudomonas but not Salmonella flagellin, indicating development of species-specific flagellin recognition during the divergent evolution of mammals and reptiles. Our discovery of reptile TLR5 fills the evolutionary gap regarding TLR conservation across vertebrates and provides novel insights in functional evolution of host-microbe interactions.


Toll-like receptors (TLR) are ancient innate immune receptors crucial for immune homeostasis and protection against infection. TLRs are present in mammals, birds, amphibians and fish but have not been functionally characterized in reptiles despite the central position of this animal class in vertebrate evolution. Here we report the cloning, characterization, and function of TLR5 of the reptile Anolis carolinensis (Green Anole lizard). The receptor (acTLR5) displays the typical TLR protein architecture with 22 extracellular leucine rich repeats flanked by a N-and C-terminal leucine rich repeat domain, a membrane-spanning region, and an intracellular TIR domain. The receptor is phylogenetically most similar to TLR5 of birds and most distant to fish TLR5. Transcript analysis revealed acTLR5 expression in multiple lizard tissues. Stimulation of acTLR5 with TLR ligands demonstrated unique responsiveness towards bacterial flagellin in both reptile and human cells. Comparison of acTLR5 and human TLR5
using purified flagellins revealed differential sensitivity to Pseudomonas but not Salmonella flagellin, indicating development of species-specific flagellin recognition during the divergent evolution of mammals and reptiles. Our discovery of reptile TLR5 fills the evolutionary gap regarding TLR conservation across vertebrates and provides novel insights in functional evolution of host-microbe interactions.
Toll-like receptors (TLRs) form a family of evolutionarily highly conserved innate immune receptors that play a crucial role in immune homeostasis and the response to infection 1,2 . TLRs are glycoproteins that typically consist of an extracellular sensor domain (ECD) composed of multiple leucine rich repeats (LRR), a transmembrane domain (TM) and an intracellular Toll/Interleukin-1 receptor (TIR) signalling domain 3 . The ECD senses the presence of conserved microbial structures in the environment and transduces this signal to the TIR domain which acts as a docking station for intracellular adapter proteins like Myeloid differentiation primary response gene 88 (MyD88). The formed complex then initiates a cascade of events that ultimately results in nuclear translocation of transcription factors like Nuclear factor kappa light chain enhancer of activated B cells (NF-κ B) that direct the innate and adaptive immune response 4 .
Throughout evolution, selective pressures exerted by microbes have driven diversification of the TLR ECD, resulting in a family of distinct receptors that recognize a variety of mainly microbial ligands 5 . For example, TLR4 binds bacterial lipopolysaccharide 6 ; TLR9 or 21 recognizes bacterial nucleic acid motifs 7,8 and avian TLR15 is uniquely activated by microbial proteases via cleavage of the receptor ectodomain 9 . TLR5 senses flagellin subunits 10 that make up the flagellum of certain bacterial species including Salmonella enterica and Pseudomonas aeruginosa. Besides structural diversity between TLR family members, coevolution with microbes has also led to adaptive evolution of individual TLRs 11-13 , leading to differential recognition of TLR ligands between animal species [14][15][16] .
Within the animal kingdom, the TLR repertoire varies among species. Regarding vertebrates, genome wide studies have identified 16 TLR types in lampreys compared to 20 in bony fish, 21 in amphibians and 10 in both humans and birds 4,[17][18][19][20] . The dynamic evolution between and within TLR family members and the conservation of TLRs across highly diverse animals underlines the importance of TLRs throughout vertebrate evolution.
However, one major gap in our knowledge on vertebrate TLR evolution is the complete lack of information about the structure, function, and ligand specificity of TLRs in any species of reptile. Reptiles have a unique physiology, being the only poikilothermic amniotes, and take a central position in vertebrate evolution 21 . The first reptiles evolved around 330 to 310 million years ago (Mya) from an amphibian-like ancestor 22 . Development of the amniotic egg and a water impermeable skin allowed these early reptiles to be the first vertebrates that could permanently colonize terrestrial habitats. This pioneering step must have brought the first reptiles into contact with the prehistorical terrestrial flora, fauna and microbiota that undoubtedly shaped the immune system of reptiles and descendant animals. Yet compared to other vertebrates our knowledge on the reptile immune system, especially concerning molecular insights in reptile microbe interactions, is marginal 21 .
In present study we report the molecular cloning, characterization and function of the first reptile TLR namely TLR5 of the 'New world' lizard Anolis carolinensis (acTLR5). Evidence is provided that acTLR5 is closely related to other TLR5 orthologs and responds to bacterial flagellin, even when expressed in human cells. Differential sensitivity of acTLR5 compared to human TLR5 to Pseudomonas aeruginosa but not Salmonella Enteritidis flagellins indicate host specific adaptation of flagellin recognition.

Results
Reptile cells respond to bacterial flagellin. To assess whether reptile cells respond to TLR ligands we first stimulated IgH-2 Iguana iguana cells carrying a NF-κ B luciferase reporter plasmid with the canonical mammalian TLR ligands; LTA (TLR2), Pam 3 CSK 4 (TLR2/1), FSL-1 (TLR2/6), LPS (TLR4), FliC (flagellin of Salmonella enterica serovar Enteritidis) (TLR5), CL097 (TLR7), ODN2006 (TLR9) and the avian TLR15 activator Proteinase K. None of these TLR agonists elicited significant NF-κ B activity except for bacterial flagellin (Fig. 1). In search for the putative TLR receptor conferring this response, and by absence of the I. iguana whole genome sequence, we interrogated the whole genome sequence of the related model organism Anolis carolinensis 23,24 , using BLAST with mammalian and chicken TLR protein sequences as queries. This search yielded nine putative TLR orthologs including a putative TLR5 ortholog (Genbank accession number: XP003216083.1), which was designated as acTLR5.
Expression and characterization of the actlr5 gene. To verify that the putative acTLR5 ortholog is expressed in vivo in the Anolis lizard, we tested total mRNA isolated from different organs of an adult male for the presence of the actlr5 transcript using RT-PCR with glyceraldehyde 3-phosphate dehydrogenase (acgapdh) as a control. Transcripts of actlr5 were detected in all the tissues tested including lung, heart, stomach, liver, spleen, kidneys, intestine and testis (Fig. 2), indicating that the gene product is expressed and may be functional in various tissues.
In order to examine the function of the acTLR5 we cloned the tlr5 gene from genomic DNA of an adult male A. carolinensis. The gene consisted of a single exon encoding a protein of 871 amino acids that contained typical TLR domains. These included an ECD (residues 28 to 634) containing 24 LRRs (including N-and C-terminal LRR) as found in other TLR5 orthologs 25 , a TM domain (residues 647 to 665) and an intracellular TIR signalling domain (residues 697 to 840). The amino acid sequence differed from the A. carolinensis reference sequence at positions: 471 (H471L), 550 (V550A), 642 (S642P) and 658 (F658Y), suggesting the existence of polymorphisms in TLR5 of A. carolinensis.
Phylogenetic analysis using full-length protein sequences of different TLR types from several vertebrates including fish, amphibians, birds and mammals clustered acTLR5 with other TLR5 orthologs and in particular with chicken TLR5 (Supplementary Fig. S1). BLAST analysis with the ECD, TM and TIR domains as separate queries indicated that all three domains of acTLR5 were most similar to (predicted) TLR5 sequences of other reptiles and birds and least similar to fish TLR5 (Table 1), fully in line with the evolutionary relationships among these vertebrates. acTLR5 is functional in reptile but also in human cells. Evidence for the function of acTLR5 was sought by introducing an expression vector carrying actlr5 (or a control plasmid without insert) together with a NF-κ B luciferase reporter plasmid into reptile IgH-2 cells. Stimulation of the mock-transfected cells with S. Enteritidis flagellin (FliC) increased NF-κ B activity in these cells, confirming the results depicted in Fig. 1. However, stimulation with S. Enteritidis flagellin significantly increased NF-κ B activity in acTLR5 transfected cells (p < 0.05) (Fig. 3a), indicating that recombinant acTLR5 is functional in the transfected reptile cells and responds to flagellin. To ensure the specificity of this response, cells were stimulated with FSL-1, a synthetic lipoprotein known to be recognized by TLR2 and TLR6 heterodimers. A high dose of FSL-1 yielded similar responses in empty vector and acTLR5 transfected cells (Fig. 3a) confirming the specificity of the flagellin-induced acTLR5 response.
Reptiles and mammals have evolved independently over more than 300 million years 22 . Yet, a sequence alignment of acTLR5 with human and other vertebrate TLR5 orthologs indicated strong conservation across vertebrates of a critical proline 15 and tyrosine 26 residue as well as a phosphorylation motif 27 in the TLR5 signalling domain ( Supplementary Fig. S2). To determine whether TLR5 signalling has evolved under strong functional constraint, the functioning of acTLR5 was determined in human HeLa-57A cells which do not endogenously express TLRs and stably express the NF-κ B luciferase reporter 28 . Stimulation with FliC, and not with other TLR ligands, yielded a strong increase in NF-κ B activity in acTLR5 transfected human cells compared to control cells carrying empty vector (Fig. 3b). This functionality of reptile TLR5 in human cells strongly suggests that the expression and trafficking of the receptor and its signalling properties as well as its ligand specificity have been functionally conserved across the reptile and mammalian lineage.
Finally, to verify that acTLR5 was also able to recognize native (non-recombinant) flagellin we incubated acTLR5 transfected cells with live wild-type S. Enteritidis (WT) or its isogenic flagellin deficient derivative (Δ fliC). Only incubation with wild-type S. Enteritidis resulted in NF-κ B activation in an acTLR5 dependent manner, confirming that TLR5 is a bonafide reptile receptor for bacterial flagellin (Fig. 3c).  Reptile and human TLR5 recognize the D1 domain in flagellin. Now that we had identified acTLR5 as a specific receptor for bacterial flagellin, we examined the conservation of residues involved in flagellin binding by aligning acTLR5 (and also chicken, African clawed frog and human TLR5) with zebrafish TLR5b of which the crystal structure in complex with flagellin has been determined 29 . The alignment showed that only 40% (18/45) of the zebrafish TLR5b-flagellin interacting residues resemble the residues at the same positions in acTLR5 and the other vertebrate TLR5 sequences ( Supplementary Fig. S2), suggesting a differential basis for the structural recognition of flagellin among these vertebrates.
To determine whether the structural differences in TLR5 have influenced flagellin recognition throughout the divergent evolution of reptiles and mammals we mapped the domain of flagellin that is recognized by acTLR5. For this, we took advantage of the fact that the D1 domain of flagellins of βand γ-Proteobacteria (incl. Salmonella, and Pseudomonas species) activate TLR5, whereas, due to compositional changes, the D1 domain of flagellins of αand ε-Proteobacteria (incl. Campylobacter species) escapes recognition by TLR5 30,31 . acTLR5 or human TLR5 (hTLR5) were transfected in HeLa-57A cells and stimulated with purified recombinant S. Enteritidis flagellin (FliC) or Campylobacter jejuni flagellin (FlaA). This showed that S. Enteritidis FliC but not Campylobacter FlaA activated NF-κ B in both acTLR5 and hTLR5 transfected cells (Fig. 4). To ascertain that the unresponsiveness of acTLR5 and hTLR5 to Campylobacter FlaA involved the FlaA D1 domain, we stimulated both TLRs with NHC flagellin. NHC is a chimeric flagellin based on Campylobacter FlaA in which the D1 domain was exchanged for the S. Enteritidis FliC D1 domain 31 . Indeed, this swapping of the D1 domain restored the activation of hTLR5 and acTLR5 (Fig. 4), indicating that both receptors recognize the D1 region of Salmonella but not Campylobacter flagellin (Fig. 4) and thus that this ability is conserved between reptiles and humans. The inability of acTLR5 and hTLR5 to recognize Campylobacter flagellin may further indicate that evasion of TLR5 detection by Campylobacter developed before the divergence of reptiles and mammals.   Fig. 5a to Fig. 4). Stimulation of the cells with the reptile C. fetus subsp. testudinum lysate did not activate acTLR5 or hTLR5 (Fig. 5a) suggesting a similar evasion of TLR5 recognition by this reptile strain as noted for mammalian and chicken derived Campylobacter strains 31 . The lysate of A. hydrophila activated acTLR5 and hTLR5 equally well (Fig. 5a). However in clear contrast, reptile derived P. aeruginosa (isolate 1) potently activated acTLR5 but failed to activate hTLR5 (Fig. 5a). Additional analysis using three extra reptile (isolates 2-4) and also four human P. aeruginosa isolates (isolates 1-4) indicated stronger activation of acTLR5 than hTLR5 by P. aeruginosa isolates, regardless of their reptile or human origin (Fig. 5b). The opposite response of these TLRs to the Pseudomonas and Salmonella lysates indicates that differential recognition of the lysates was not due to variable receptor expression. acTLR5 is more sensitive than hTLR5 to Pseudomonas flagellin. To verify that flagellin was the key determinant in the differential recognition of P. aeruginosa lysates by acTLR5 and hTLR5 and to exclude the destructive effect of flagellin degrading proteases potentially present in the lysates 33 , we cloned and purified recombinant flagellin of reptile and human P. aeruginosa isolate 1. Stimulation of transfected HeLa-57A cells with low concentrations (0.1-10 ng ml −1 ) of these purified P. aeruginosa flagellins revealed again stronger activation of acTLR5 compared to hTLR5 (Fig. 5c,d) and an opposite effect for S. Enteritidis flagellin (Fig. 5e). At high concentrations (100-1,000 ng ml −1 ), reptile but not human P. aeruginosa flagellin did yield a potent hTLR5 response.
The differential dose-dependent responses by acTLR5 and hTLR5 suggested that the receptors recognize the purified flagellins with a different sensitivity. To substantiate the apparent different sensitivity of acTLR5 and hTLR5 to the purified flagellins we set the response to 1,000 ng ml −1 flagellin at 100%. This revealed that the receptors had a similar relative sensitivity to S. Enteritidis flagellin (Fig. 5h). However, compared to hTLR5, acTLR5 showed a higher relative sensitivity to the human P. aeruginosa flagellin (p < 0.05) (Fig. 5f). Higher relative sensitivity of acTLR5 was also noted for the reptile P. aeruginosa flagellin (p < 0.05) (Fig. 5g), despite the fact that high doses of this flagellin induced stronger activation of hTLR5. Overall, these results show that acTLR5 is more sensitive than hTLR5 to P. aeruginosa but not S. Enteritidis flagellin.

Discussion
Reptiles form a large group of vertebrates with a central position in vertebrate evolution and a unique physiology, being the only ectothermic amniotes. Despite this, relatively few studies have investigated the reptile immune system and detailed molecular characterizations of reptile immune molecules are scarce. Here we report a detailed functional characterization of the first TLR in reptiles. Our characterization of TLR5 of the lizard A. carolinensis fills the evolutionary gap of functional TLRs across vertebrates and provides a novel view on the reptile immune system at a molecular level. Evidence is provided that acTLR5 is expressed and functional in reptile as well as human cells and responds to bacterial flagellin. Our results indicate that TLR5 structure, function and signalling are highly conserved throughout evolution, although differences in relative sensitivity of reptile and human TLR5 to Pseudomonas but not Salmonella flagellin point to bacterial species dependent adaptations in flagellin recognition by reptile and human TLR5.
The reptile tlr5 gene was cloned from an Anolis carolinensis lizard. Support for its identification as tlr5 ortholog included a strong phylogenetic relationship of the full-length protein with the well-characterized chicken TLR5 15,31 . The ECD and TIR domain of the cloned acTLR5 were highly similar to a putative TLR of the Burmese python (snake), suggesting that the gene is present in other reptiles as well. Lizards, snakes and tuatara form the group of Lepidosauria that diverged approximately 270 Mya from their bird and crocodile sister group; the Archosauria 34 . Lizards and snakes thereafter diverged approximately 180 Mya 35,36 . The phylogeny of these species is reflected by the high similarity of acTLR5 with the putative snake and chicken TLR5, suggesting that TLR5 underwent a constrained evolution according to species divergence.
Functional evidence for identifying the cloned Anolis gene as a TLR5 ortholog was provided by the responsiveness of acTLR5 transfected cells to bacterial flagellin, thus far the only known TLR5 ligand. Activation of NF-κ B in acTLR5 expressing cells was observed upon stimulation with wild type but not flagellin-deficient Salmonella as well as with purified recombinant Salmonella and Pseudomonas flagellins, thereby excluding non-specific activation of NF-κ B. The results indicate that acTLR5 senses flagellins of different bacterial species and is capable of initiating a signalling cascade required to evoke an immune response. In mammals, flagellin recognition by TLR5 is indispensable for an adequate immune response to infection with flagellated bacteria [37][38][39][40] . As A. carolinensis tissues express the actlr5 gene in vivo (Fig. 2) acTLR5 may have a similar function in reptiles.
A striking finding that underpins the evolutionary conservation of the TLR system is the functional expression of reptile TLR5 in a human cell background. The first step in TLR5 mediated NF-κ B activation is the recruitment of the intracellular MyD88 adapter protein to the TLR5 TIR domain 41 . Comparison of the TIR domains of reptile and human TLR5 revealed a high overall sequence similarity (85%) and conservation of specific amino acid residues that are critical for TLR5 signalling 15,26,27 . In addition, both the TIR domain and MyD88 have been shown to evolve under strong functional constraint [42][43][44] . Together, this may explain the successful activation of NF-κ B by acTLR5 in human cells. The compatibility of reptile TLR5 with human intracellular proteins suggests that the TLR5 signalling system was already functional in the common ancestor of reptiles and mammals and provides support for the functionally constrained evolution of TLR5 signalling at least throughout the divergent evolution of reptiles and mammals. Here it may be noteworthy that efforts to functionally express intact TLR5 from fish or amphibians in human cells have thus far not been reported.
Bioinformatics analysis indicated that the ECD of acTLR5 contained a N-and C-terminal LRR separated by 22 consecutive LRRs which is a typical feature of chicken 15 and other vertebrate TLR5 orthologs 25 . In line with the apparent conserved structure of the ECD, both reptile and human TLR5 recognized and responded to the D1 domain of Salmonella but not Campylobacter flagellin. This finding demonstrates that throughout 300 million years of divergent evolution, reptile and human TLR5 have conserved the ability to recognize flagellin at its D1 domain and hence the flagellin D1 domain of certain bacterial species has remained a critical activator of TLR5.
Interestingly, despite different amino acid compositions of the ECD, reptile and human TLR5 showed equal sensitivity to flagellin of Salmonella enterica serovar Enteritidis. Pet reptiles are frequently reported as carriers of zoonotic Salmonella serovars that can cause salmonellosis in humans but are generally considered non-pathogenic in healthy reptiles [45][46][47][48][49] . The principles underlying resistance or tolerance of reptiles to Salmonella are unknown but may relate to the poikilothermic nature of reptiles since Salmonella virulence is influenced by environmental temperature 50,51 . Yet, the fact that reptile and human TLR5 show a similar relative sensitivity to S. Enteritidis flagellin may suggest that flagellin recognition does not play a significant role in the differential susceptibility to Salmonella infection observed between reptiles and humans.
In contrast to Salmonella flagellin, reptile and human TLR5 showed a differential sensitivity to flagellin of P. aeruginosa clinical isolates. P. aeruginosa is a common bacterium that resides in diverse environments including water and soil and is an opportunistic pathogen of both reptiles and humans 52,53 . Why reptile TLR5 is more sensitive to P. aeruginosa flagellin than human TLR5 remains to be elucidated but it may suggest that throughout host-microbe coevolution, P. aeruginosa has exerted a stronger selective pressure on the evolution of acTLR5 than on hTLR5. Indeed, in silico studies indicate that among primates 54 and galloanserae birds 13 TLR5 undergoes diversifying, adaptive evolution through positive selection, a process most likely driven by host specific coevolution with flagellated bacteria. A similar process in reptiles may explain the observed differences in P. aeruginosa flagellin recognition between the Anolis and human TLR5.

Methods and Materials
Isolation of Anolis carolinensis DNA and RNA. Anolis tissue samples were obtained from a healthy male Anolis carolinensis lizard that had been euthanized by intra-coelomic injection of pentobarbital (200 mg kg −1 BW, Euthanimal ® , Alfasan International, The Netherlands). Organs were directly frozen in liquid nitrogen. Genomic DNA was isolated using the high pure template kit (Roche) according to the manufacturer's instructions. RNA was extracted from tissue lysed with RLT buffer (1% β-mercaptoethanol) (Qiagen) in 1.4 mm Fastprep lysing matrix tubes (MPbio) in a Magna Lyser centrifuge (6,500 × g, 40 s, RT) (Roche). Total RNA was isolated using the RNeasy mini kit (Qiagen) following the manufacturer's instructions, treated with DNase I (1 U mg −1 RNA, Thermo Scientific) and stored at − 80 °C until use. Cloning of A. carolinensis tlr5. The A. carolinensis tlr5 gene (actlr5) was amplified from genomic DNA (500 ng) by PCR in 50 μ l volume containing 1X Phusion polymerase buffer, dNTP's (0.2 mM each), MgCl 2 (50 mM), Phusion hot start II high fidelity polymerase (1 Unit, Thermo Scientific) and 20 μM of forward (5′ -CCGGATCCATGAAAAAGATGCTTCATTATCTCTTC-3′ ) and reverse (5′ -CCGCGGCCGCAAGAGATTGTGACTACTTT-3′ ) primer (Life Technologies). Underlined sequences in the forward and reverse primer indicate BamHI and NotI restriction sites, respectively. The bold GC in the reverse primer substituted an AG in the tlr5 gene, thereby replacing the terminal stopcodon for a cysteine. PCR conditions were: one cycle for 1 min at 98 °C followed by 35 cycles of 30 s at 98 °C, 30 s at 54 °C, 90 s at 72 °C and one final extension step of 10 min at 72 °C. The PCR product was purified from gel using the GeneJet gel extraction kit (Thermo Scientific) and ligated into a pTracer-CMV2Δ GFP/3×FLAG 8 using the BamHI and NotI restriction sites, yielding pTracer 3 × FLAG-actlr5 carrying actlr5 with a C-terminal 3×FLAG tag. The plasmid was propagated in DH5-α. The cloned actlr5 gene sequence was verified by DNA sequencing (Macrogen). The sequence was deposited in Genbank (accession number: KT347095).

Construction, expression and purification of recombinant His-tagged flagellins.
Construction of recombinant His-tagged flagellin of S. Enteritidis (FliC), C. jejuni (FlaA) and chimeric NHC has been described previously 15,31 . The flagellin gene of both reptile and human P. aeruginosa isolate 1 was amplified from genomic DNA by PCR in 50 μ l volume containing 1X Dreamtaq polymerase buffer, dNTP's (0.2 mM each), Dreamtaq polymerase (1 Unit) and 20 μ M of forward (5′ -AAACCATGGCCTTGACCGTCAACAC-3′ ) and reverse (5′ -AAAGAGCTCGCGCAGCAGGCTCAGAAC-3′ ) primer. Underlined sequences in the forward and reverse primer indicate NcoI and SacI restriction sites, respectively. PCR conditions were: one cycle for 3 min at 95 °C followed by 35 cycles of 30 s at 95 °C, 30 s at 64 °C, 2 min at 72 °C and a final extension step for 10 min at 72 °C. PCR products were ligated into the pET101/D-TOPO (Promega) expression vector using NcoI and SacI restriction enzymes. Ligation into the pET101/D-TOPO vector added a C-terminal His-tag to the flagellin gene and the plasmids were transformed into E. coli BL21 star (DE3).
Protein expression was induced by growing log phase cultures in the presence of 1 mM IPTG (Thermo Scientific) for 4 h at 37 °C. For flagellin purification bacteria were pelleted (4,400 × g, 15 min, 4 °C), resuspended in 10 ml cold DPBS with protease inhibitor cocktail (Roche), spun down (4,400 × g, 15 min, 4 °C) and incubated (RT) under end-over-end rotation for 16 h in 8 M urea buffer (8 M urea, 100 mM NaH 2 PO 4 , 100 mM Tris-HCl, pH 8). After removal of cell debris (5,300 × g, 30 min, RT) supernatant was incubated with Ni 2 + -NTA agarose beads (Qiagen). After 2 h the beads were washed with 4 × 4 ml of 8 M urea buffer pH 6.3. Flagellins were eluted with 4 × 0.5 ml of 8 M urea buffer pH 5.9 followed by 4 × 0.5 ml of 8 M urea buffer pH 4.5. Collected fractions were checked for purity on SDS-PAGE and pure fractions were pooled and concentrated using Amicon YM-30 filters (Millipore). Protein concentration was measured by BCA assay. Concentrated flagellins were diluted to the desired concentration and stored (− 20 °C) as aliquots in 4 M urea, 100 mM NaH 2 PO 4 , 10 mM Tris-HCl, pH 9.
Luciferase assay. Twenty-four hours after transfection cells were re-distributed in a 48-well plate. After 24 h cells were washed twice with medium without FCS and stimulated with the indicated TLR ligands or live bacteria in 500 μ l medium without FCS (for stimulation with LPS, medium did contain FCS). After 5 h at 37 °C (HeLa-57A) or 10 h at 30 °C (IgH-2), cells were washed with DPBS and lysed with reporter lysis buffer (100 μ l, Promega) at − 80 °C for at least 1 h. After thawing, cell lysate (20 μ l) was mixed with luciferase reagent (50 μ l, Promega) and luciferase activity was measured in a luminometer (TD20/20, Turner designs). Experiments with bacterial lysates and purified P. aeruginosa flagellins were performed in 96-well plates in 250 μ l volumes. Cells were lysed in 50 μ l reporter lysis buffer. Luciferase activity in these experiments was measured with a TriStar 2 luminometer (Berthold) by mixing 15 μ l cell lysate with 37 μ l luciferase reagent. Values obtained from the TriStar 2 were 1000 times higher compared to the values obtained from the TD20/20 but relative sensitivity and accuracy between the two luminometers was equal. Results were expressed in relative light units (RLU) or % RLU in experiments with purified P. aeruginosa flagellins. Percent RLU was calculated by dividing the RLU obtained from each concentration of flagellin over the value obtained from stimulation with 1 μ g ml −1 flagellin which was set at 100%.
Statistics. Statistical analysis were performed using Graphpad 6 (Prism) software. Differences between two groups were tested with unpaired Student t-tests. A probability (p) value of < 0.05 was considered significant.