Identification and Characterization of a Splicing Variant in the 5′ UTR of the Human TLR5 Gene

Toll-like receptors (TLRs) are essential components of the innate immune system. TLR5 is the receptor for flagellin, the principal protein component of bacterial flagella. The TLR5 gene has 6 exons. In an RT-PCR analysis, we found long TLR5 transcripts, in addition to those of the expected size (short TLR5 transcripts). A sequence analysis revealed that the long TLR5 transcripts contain a new exon of 94 nucleotides located between previously reported exons IV and V in the 5′ untranslated region (5′ UTR). A real-time PCR analysis of the two alternatively spliced variants in various cell lines showed that the long TLR5 transcripts are abundantly expressed in nonimmune cells. The ratios of long/short transcripts in human nonimmune cell lines, such as A549, T98G, HaCaT, H460, HEK-293, and Caco-2 cells, and primary mesenchymal stem cells were in the range of 1.25 to 4.31. In contrast, those of human monocytic THP-1 and U937 cells and E6.1 T cells and Ramos B cells were around 0.9. These ratios in human monocytic THP-1 cells were decreased by treatment with IFN-γ in a concentration-dependent manner. Based on our findings, we suggest that the newly found long TLR5 transcripts may be involved in the negative regulation of TLR5 expression and function.


Introduction
Toll-like receptors (TLRs) are innate immune receptors that consist of an extracellular domain for the recognition of pathogenic components and a cytoplasmic tail with a conserved Toll/IL-1 receptor (TIR) domain for the generation of intracellular signaling [1]. Upon TLR stimulation by pathogenic components, the TIR domain recruits signaling molecules to activate the transcription of diverse genes, including inflammatory and antimicrobial mediators [1]. TLR signaling must be tightly controlled to avoid the overproduction of proinflammatory mediators that would be harmful to the host. As a control mechanism, alternative splicing can be used to modulate the expression and function of TLRs. Members of the TLR signaling pathway are alternatively spliced at a high frequency, producing novel proteins that can change inflammatory outcomes. Alternative splicing has been found in mammalian TLR genes and their homologs in plants and Drosophila [2][3][4]. Mouse TLR4 has two splicing variants that are inducible by interferon-priming as well as LPS stimulation of primary macrophages [5]. Alternative splicing of human TLR has also been reported. Human TLR1, TLR2, TLR3 [6], and TLR9 [7] have two splicing variants, while TLR4 has four splicing variants, all of which change the length of the extracellular domain, but their functional significance has not been examined [5]. Alternative splicing of key TLR signaling components, such as MyD88 and IRAK, has also been reported. A splicing variant of MyD88, termed MyD88s, which lacks the intermediate region between the TIR domain and the death domain, inhibits inflammatory signals that are normally mediated by MyD88 in both mouse [8] and human [9] cells. IRAK also has splicing variants. Two splicing variants of murine IRAK2 are inhibitory [10], and a splicing variant of human IRAK1 is inhibitory [11]. These studies suggest that the splicing of TLR signaling molecules is involved in the resolution of TLR-directed immune responses.
In the present study, we identified and characterized a new splicing variant in the 5 untranslated region (5 UTR) of the human TLR5 gene.  Figure 1: The exon structure of the human TLR5 gene and detection of TLR5 splice variants in human monocytic THP-1 cell line. (a) Human TLR5 contains seven exons (depicted as boxes). Exons IV+ are alternative (marked by shading). (b) PCR was performed with primers designed to exons I and VI (dotted arrows), or (c) exon IV and exon V (solid arrows). The amplified products were visualized by gel electrophoresis. IV+ designates the newly discovered exon.

Primers.
To detect TLR5 alternative splicing, specific primers for RT-PCR were designed to bind to exons I and VI (Figure 1(a), dotted arrows) or exons IV and V (Figure 1(a), solid arrows). To quantify splicing variants of TLR5, specific primers were designed for real-time quantitative PCR (RT-qPCR). The primers that recognize the new exon (designated V in Figure 4) were used to detect long TLR5 transcripts (solid arrows). Short TLR5 transcripts were detected by primers recognizing the boundary between exons IV and V (dashed arrows). The primers for the TLR5 reference that recognize exon VII are represented as dotted arrows. The primer sequences are listed in Table 1.

Relative Quantification of Alternative Splicing Variants by Real-Time Quantitative PCR.
Total RNA was extracted using the Qiagen RNeasy Kit (Hilden, Germany) according to the manufacturer's instructions. An SD2000 microspectrophotometer (Bioprince, Atlanta, GA, USA) was used to determine RNA concentrations. The cDNA was constructed from 2.5 g of total RNA using MMLV Reverse Transcriptase (GeneAll, Seoul, Korea) and an Oligo(dT) primer (Invitrogen Corp.) at 50 ∘ C for 1 h. The cDNA was amplified by PCR, and the PCR products were stained with Loading Star solution (Dynebio, Seongnam, Korea) and separated on 1.5% agarose gels.
RT-qPCR was conducted using the iQ5 multicolor RT-PCR detection system (Bio-Rad, Hercules, CA, USA) with the iQ SYBR Green Supermix (Bio-Rad). DNA amplification was performed using the primer sequences listed in Table 1, which were designed to specifically identify each of the alternative splicing variants.
Relative quantification of TLR5 transcripts generated by alternative splicing was performed according to previously  [12]. Briefly, in order to determine the relative proportions of TLR5 transcripts, a "never-spliced" exon of TLR5 was used as an internal reference, instead of a classical housekeeping gene, with a portion common to the long and short TLR5 transcripts, that is, exon VII. The real proportions of TLR5 transcripts were determined according to the principle that the sum of both the long and the short TLR5 transcripts equals the level of expression of a "neverspliced" TLR5 exon. Therefore, the sum of the ratios of the

Identification of Human TLR5 Splicing Variants.
The human TLR5 gene comprises 6 exons and the coding sequence is in exon VI [13]. In our RT-PCR experiment using primers complementary to exon I and exon VI of TLR5, we observed two main bands on the agarose gel.
Unexpectedly, a longer product (about 680 bp) was detected in addition to the product of the expected size (about 580 bp), suggesting an alternative splicing variant with an additional exon (Figure 1(b)). To confirm this, a new set of nested PCR primers was designed to amplify the region spanning exons IV through V (Figures 1(a) and 1(c), solid arrows) and two PCR products were detected (Figure 1(c)). A sequence analysis of these products revealed that the longer product contains an additional 94 bp between exon IV and exon V, and its sequences are exactly the same as sequences deposited in the NCBI database (NM 003268.5) (Figure 2). Therefore, our results indicate that human TLR5 has seven exons and a newly found exon lies between previously reported exons IV and V in the 5 UTR.

Expression of Human TLR5 Splicing Variants in Different
Cell Lines. To determine whether two main PCR products can be generated in other human cell lines, we performed an RT-PCR analysis of TLR5 expression in various human cell lines using the same PCR primers used in Figure 1(c). As shown in Figure 3, long TLR5 transcripts were observed in all eleven cell types examined. Short TLR5 transcripts were clearly detected in immune cells, such as THP-1, E6.1, U937, and Ramos cells, but were weakly detected in nonimmune cells, such as HaCaT, MSC, T98G, H460, HEK-293, and Caco-2 cells.
We determined the precise ratio of long TLR5 transcripts to short TLR5 transcripts (long/short TLR5 transcripts) in each cell line by RT-qPCR, as described in the Materials and Methods. The ratios of long TLR5 transcripts to short TLR5 transcripts in human immune cells, such as THP-1, U937, E6.1, and Ramos cells, were around 0.9, and those in nonimmune cells were 1.25 to 4.31 ( Figure 4, Table 3).

Regulation of Human TLR5 Alternative Splicing by IFN-
. Since IFN-is an important activator of macrophages and downregulates TLR5 expression [14], we investigated whether IFN-influences TLR5 alternative splicing. THP-1 cells were stimulated with 10 ng/ml IFN-for 0-24 h and collected for RT-QPCR analysis. The ratio of long/short TLR5  -GACTTCCC TACTCTCATC TGCTCCAGCC CCCTGACCTT ATAGTTGCCC AGCTTTCCTG GCAATTGACT TTGCCCATCA  TLR5 lower band  -TGACTTCC-CTATCTCATC TGCTCCAGCC CCCTGACCTT ATAGTTGCCC AGCTTTCCTG GCAATTGACT TTGCCCATCA  TLR5 upper band  TTGACTTCCG TTCTCTCATC TGCTCCAGCC CCCTGACCTT ATAGTTGCCC AGCTTTCCTG GCAATTGACT TTGCCCATCA   90  140  130  120  110  100  150 Figure 2: Alignment of sequences of two TLR5 splicing variants with TLR5 sequence from NCBI. Two PCR product bands of the amplification from exon IV to exon VI were collected and sequenced with automated sequencing methods by Applied Biosystems 3730xl DNA Analyzer. These two sequences were then compared to the NCBI TLR5 sequence (NCBI reference sequence: NM 003268.5) by using BioEdit software version 7.2.5. The detailed sequences are listed in Table 2. transcripts began to decline at 3 h, reached 60% of normal ratio at 6 h, and then returned to normal at 24 h after IFNtreatment ( Figure 5(a)). The ratios of long/short TLR5 transcripts decreased by treatment with IFN-in a concentrationdependent manner and were reduced by approximately 50% compared to normal levels at a concentration of 50 ng/ml ( Figure 5(b)).

Discussion
In this study, we found a new splicing variant of human TLR5 with an extra exon; it included 94 nucleotides and was located between previously reported exons IV and V in the 5 UTR. Since genetic variants of TLR5 have been clinically associated with disease outcomes such as obesity, type 2 diabetes, and colorectal cancer [15], it is important to determine the functional role of alternative splicing variants of TLR5. Based on the sequences in the NCBI database, human TLRs have a small number of exons (2-6 exons). The coding regions of TLR3 and TLR4 consist of more than two exons, while those of other TLRs consist of only the last exon (TLR1, TLR2, TLR5, TLR6, and TLR10) or the last exon plus a small portion of the second last exon (TLR7,  Figure 3: Expression of TLR5 splicing variants in different cell lines. Eleven cell lines were cultured at specific conditions. Cells were collected and 2.5 g mRNA was used for RT-PCR; 3 l of each cDNA was used for PCR with alternative splicing detecting primer (35 cycles, 60 ∘ C annealing temperature). GAPDH was used as control.
TLR8, and TLR9). Therefore, TLR5 is not expected to produce variable proteins, while TLR3 and TLR4 are predicted to express a variable protein. In fact, human TLR4 has four reported splicing variants with different extracellular leucinerich repeat lengths, but their functional significance has not been evaluated [5]. Although the functional significance of alternative splicing at the 5 UTR of TLR genes remains to be elucidated, it is a common feature of mouse and human TLR genes [5], suggesting that alternative splicing of TLR genes plays an important role in the precise control of immune activation [16]. Since alternative splicing of the 5 UTR influences mRNA stability and translation and thus alters the amount of protein translated [17,18], differential expression levels of alternative splicing variants of TLR5 in various cell types, especially between immune and nonimmune cells, observed in our study may reflect intrinsic differences of TLR5 protein expression and function between cell types. The preferential expression of long TLR5 transcripts in nonimmune cells compared to immune cells suggests that the long TLR5 transcripts are selectively induced in nonimmune cells and may function differentially from the short TLR5 transcripts. IFN--induced reduction in the expression of the long TLR5 transcripts of human monocytes also suggests a different function from that of the short TLR5 transcripts in activated macrophages.
In  : Ratios of long TLR5 transcripts to short TLR5 transcripts in several different cell lines. Expression of long TLR5 transcripts was quantified by RT-qPCR using primers recognizing exon V (primer pairs were represented by solid arrows), while expression of short TLR5 transcripts was quantified using primers recognizing the boundary between exons IV and VI (primer pairs were represented by dashed arrows). Exon VII was used as a reference gene (primer pairs were represented by dotted arrows). The expression level of short TLR5 transcripts was arbitrarily set to 1. * indicates < 0.05 as compared with THP-1.
were different in distinct cell types. Based on our finding that long TLR5 transcripts were predominantly expressed in nonimmune cells and were significantly reduced by treatment with the proinflammatory cytokine IFN-, we cautiously speculate that the newly detected long TLR5 transcripts may be involved in the negative regulation of the expression and function of TLR5. To confirm this, further studies of the protein expression patterns in different cell types and functional analyses of splicing variants using HEK-293 cells transfected with cloned splicing variants are required.

Disclosure
Thi Xoan Hoang and Cao Nguyen Duong should be considered co-first authors. Results have been presented in part at a poster session of Innate Immunity conference held in Barcelona, Spain, 2015.