A Protein Associated with Toll-Like Receptor 4 (PRAT4A) Regulates Cell Surface Expression of TLR4¹

Innate immunity is the first line of defense against microbial infection (1). The Toll family of receptors plays an essential role in innate recognition of microbial products (2). TLRs are type I transmembrane proteins that contain a large, leucine-rich repeat in an extracellular region and a Toll/IL-1 receptor homology domain in a cytoplasmic region (3). Cell surface TLRs, including TLR1, TLR2, TLR4, TLR5, and TLR6, recognize bacterial products, whereas TLR3, TLR7, TLR8, and TLR9 reside in intracellular organella and recognize microbial nucleic acids. These subcellular distributions of TLRs are optimized for microbial recognition. TLR5, e.g., is expressed exclusively on the basolateral surface of intestinal epithelia where pathogenic Salmonella, but not commensal Escherichia coli, translocate TLR5 ligand flagellin across epithelia (4). The subcellular distribution of TLR9 is important for sensing viral infection and differential recognition of two distinct types of TLR9 ligands (5, 6, 7, 8). A/D-type CpG is retained for long periods in the endosomal vesicles of plasmacytoid dendritic cells (DCs),⁴ together with the complex consisting of downstream signaling molecules (8), whereas the B/K-type CpG is recognized by TLR9 in the lysosome (9, 10).

LPS, one of the most immunostimulatory glycolipids constituting the outer membrane of the Gram-negative bacteria, is recognized by the receptor complex consisting of TLR4 and MD-2 (3, 11, 12, 13, 14). MD-2 is an extracellular molecule that is associated with the extracellular domain of TLR4 and is indispensable for LPS recognition by TLR4 (15, 16, 17, 18). TLR4/MD-2 is similar to TLR9 in that LPS can be recognized at the two distinct sites. TLR4/MD-2 is expressed on the cell surface, and LPS recognition occurs on the cell surface in macrophage cells (19, 20, 21). In intestinal epithelial cells, TLR4/MD-2 resides in the Golgi apparatus and recognizes internalized LPS (20). Little is known about differences between LPS recognition/signaling on the cell surface and that in the Golgi apparatus. Considering roles for TLR9 subcellular distribution in CpG recognition, it is possible that these LPS responses are distinct from each other with regard to specificity or cytokine production.

The subcellular distribution of TLR4 is reported to be regulated by two molecules, the endoplasmic reticulum (ER) chaperon gp96 and MD-2. Randow and Seed (22) described a pre-B cell line deficient in the gp96 in which TLR4 failed to associate with MD-2, get to the cell surface, and respond to LPS (22). We previously demonstrated that TLR4 alone is little expressed on the cell surface in MD-2^−/− embryonic fibroblasts and that most TLR4s reside in ER or the Golgi apparatus without mature glycosylation, revealing an important role for MD-2 in cell surface expression of TLR4 (16). MD-2 association permits TLR4 glycosylation at Asn⁵²⁶ or Asn⁵⁷⁵, which facilitates cell surface expression of TLR4 (23). Despite an important role for MD-2 in cell surface expression of TLR4, several studies claimed that TLR4 alone may be expressed on the cell surface without MD-2 in a human kidney cell line (10, 24, 25), suggesting that another MD-2-like molecule regulates TLR4 trafficking to the cell surface. In this study we addressed this issue and described a novel molecule that is associated with TLR4 and regulates its expression on the cell surface.

Anti-Flag Ab and anti-Flag-agarose were purchased from Sigma-Aldrich. Rat anti-mouse TLR4 mAb (MTS510 and Sa15-21) and rat anti-mouse CD14 (Sa2-8) were established in our laboratory and described previously (19). Pre-made Northern blot for mouse normal tissues was purchased from Seegene. HEK293 cells were maintained in DMEM supplemented with 10% FCS and antibiotics. IL-3-dependent Ba/F3 cells were cultured in 10% FCS-RPMI 1640 supplemented with antibiotics, 100 μM 2-ME, and recombinant murine IL-3 (∼70 U/ml) at 37°C in a humidified atmosphere of 5% CO2. A rat was immunized with the GST-PRAT4A (protein associated with TLR4A) fusion protein and used for hybridoma production. The anti PRAT4A mAb (542; rat IgG) can be used for immunoprobing.

C57BL/6 mice were purchased from Japan SLC. TLR4^−/− mice were kindly provided by Dr. S. Akira (Osaka University, Osaka, Japan), and MD-2^−/− mice were established in our own laboratory (16). These mice were maintained in the animal facility at the Institute of Medical Science, the University of Tokyo, Tokyo, Japan. Bone marrow-derived DCs and macrophages were prepared as described previously (16). Briefly, bone marrow cells were plated at 1 × 10⁶ cells/ml in 24-well plates with 10% FCS-RPMI 1640 supplemented with 10 ng/ml recombinant murine GM-CSF (Genzyme/Techne) or 100 ng/ml recombinant murine macrophage CSF (M-CSF) (Genzyme/Techne). On day 7, cells were harvested and used for flow cytometry. For retroviral transduction, bone marrow cells during DC induction were transduced on days 1, 3, and 5 with retroviruses and 25 μl of N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP) transfection reagent (Roche) and centrifuged at 2000 rpm for 1 h at room temperature. After centrifugation, cells were cultured in 10% FCS-RPMI1640 supplemented with GM-CSF and cultured at 37°C in a humidified atmosphere of 5% CO2. On day 7, cells were harvested and used for flow cytometry. To detect DCs expressing short hairpin RNA (shRNA), we used a retroviral vector encoding GFP. In flow cytometry, the gate was set so that only DCs with GFP expression were analyzed (Fig. 7, A and B). To enrich DCs expressing shRNA, we used a retroviral vector with the puromycin resistance gene. Bone marrow was isolated from C57BL/B6 mice that had been injected i.p. 4 days earlier with 5 mg of 5-fluorouracil. After 48 h of culture, cells were transduced with retroviral supernatant on two successive days. After the second transduction, cells were washed and cultured in the presence of GM-CSF. During DC induction, puromycin (2μg/ml) was included to enrich DCs expressing shRNA (Fig. 7 C).

HEK293/mTLR4-GFP cells were collected up to ∼5 × 10⁹ cells. Cell lysate extraction and purification with the anti-GFP Abs were conducted as described in the paragraph title Immunoprecipitation and immunoprobing below in this section. Bound proteins were eluted and fractionated with the elution buffer containing 30 mM glycine/HCl (pH 2.5), 30 mM NaCl, and 0.1% Triton X-100. Each fraction was neutralized immediately with 1 M Tris-HCl (pH 8.0), and peak fractions were subjected to freeze drying, SDS-PAGE, electroblotting, and staining with Coomassie brilliant blue R-250 (Bio-Rad). The coprecipitated protein was excised, and its N-terminal amino acid sequence was determined by APRO Science.

Mouse PRAT4A cDNA was cloned into a retroviral vector, pMX-IRES-GFP. PRAT4A was tagged with the Flag epitope followed by the His₆ epitope (collectively designated hereafter as the FH epitope) at the C terminus (PRAT4A-FH). PRAT4A-FH was expressed in Ba/F3 cells stably expressing TLR4/MD-2 and CD14. Retroviral vectors were transfected into the Plat-E cells using the FuGENE 6 transfection reagent (Roche). Retroviruses were transduced into Ba/F3 cells, and stable cells expressing mouse PRAT4A were established by enriching GFP-positive cells with a cell sorter. Human and mouse PRAT4A cDNAs were also cloned into the expression vector pEFBOS. We also established retroviral expression vector (pMX-puromycin) encoding TLR4, MD-2, and TLR2, all of which were either tagged with the FH epitope (TLR4FH, MD-2FH, and TLR4FH) or without epitope tags (see Fig. 3 C).

Cells were lysed on ice for 30 min with a lysis buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 7.6), 2 mM EDTA, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM PMSF, and 1% Triton X-100. Lysates were separated from debris by centrifuging at 13,000 rpm at 4°C for 10 min and then incubated with Ab-conjugated beads at 4°C for 2 h. Beads were washed three times, and bound proteins were boiled at 99°C for 5 min in the sample buffer. Samples were subjected to SDS-PAGE and Western blotting. Reagents used for immunoprobing were anti-Flag (M2), anti-mouse PRAT4A mAb, anti-mouse TLR4 mAb, anti-His₆ mAb, and anti-mouse MD-2 polyclonal Abs. The second Abs were goat anti-mouse IgG-alkaline phosphatase conjugate (American Qualex) and goat anti-rat IgG-alkaline phosphatase conjugate (American Qualex).

To investigate the biological function of PRAT4A, we purchased an RNA interference duplex (Invitrogen Life Technologies). The oligonucleotide sequences of target sequences for human and mouse PRAT4A were ggaggaagaccugacugaauuccuc (sense strand sequence for human PRAT4A, no.9) and accugacugaauuccucugcgccaa (sense strand sequence for mouse PRAT4A, no.11). Sequence-scrambled negative controls for no.9 and no.11 were ggaagaaguccgucauuaacggcuc and accgucauuaacuccgcgucgucaa, respectively. HEK293 transfectants were seeded onto 24- or 6-well plates in antibiotic-free medium on the day before RNA interference transfection. LipofectAMINE 2000 (Invitrogen Life Technologies) was used as a transfection reagent. Cells were collected 3–4 days after transfection and used for further analyses.

To inhibit mouse PRAT4A expression, we produced a retroviral vector expressing shRNA (26). The target sequences was 5′-gag ttt gaa gag gtg att gag-3′.

Premade Northern blot for mouse normal tissues was purchased from Seegene. The RNA blot was hybridized with a ³²P-labeled cDNA probe encoding mouse PRAT4A (from aa 1 to aa 166). The probe was labeled using a random primed labeling system (Rediprime II; Amersham Biosciences). Hybridization was conducted at 65°C for 17 h in a buffer containing 10% dextran sulfate, 1% SDS, 50 mM Tris-HCl (pH7.5), 1 M NaCl, and denatured salmon sperm DNA (100 μg/ml). Blots were washed three times in 2 × SSC/0.1% SDS at 65°C and three times in 0.2 × SSC/0.1% SDS for 10 min at 65°C before exposure to imaging plate (Fuji Film).

Total pure RNA (1 μg) from Ba/F3 or HEK293 cells was reverse transcribed into cDNA using ReverTra Ace-α (Toyobo) according to the manufacturer’s instruction. cDNA was diluted at 1/1, 1/4, and 1/16, and each diluted cDNA was used as PCR template. RT-PCR was conducted with initial denaturation at 94°C for 3 min followed by 25 cycles of 94°C for 30 s, 54°C for 30 s, and 72°C for 1 min. The primer pairs used were as follows: 5′-ggctcgagatggattcaatgcctgagcc-3′ and 5′-ggggatcctcagagctcatcagggg-3′ for human PRAT4A; 5′-tctactcaagccaggatgag-3′ and 5′-ttcctgccaattgcatcctg-3′ for human TLR4; and 5′-tatggacaggactgaacgtc-3′ and 5′-cctacaacaaacttgtctgg-3′ for human hypoxanthine phosphoribosyltrans-ferase (HPRT).

To determine the efficiency of gene silencing, semiquantitative PCR analyses were conducted using a 7300 fast real-time PCR System (Applied Biosystems) with TaqMan gene expression assays (Applied Biosystems) for mouse PRAT4A (Mm00511161). To normalize the amount of cDNAs, each sample was standardized by using internal TaqMan gene expression assays for mouse β-actin (Mm00607939).

Because HEK293 cells do not express endogenous MD-2, HEK293 cells expressing TLR4 responded to LPS/MD-2 but not to LPS alone (25), suggesting that TLR4 was expressed on the cell surface and bound to the LPS/MD-2 complex. We established HEK293 cells stably expressing mouse TLR4 fused with GFP. Established HEK293 cells, in which TLR4 expression was confirmed by GFP expression, were stained with two distinct mAbs to TLR4, Sa15-21 and MTS510. Cell surface TLR4 was detected by Sa15-21, but very weakly with MTS510 (Fig. 1,A, middle). When these cells were transiently transfected with an expression vector encoding mouse MD-2, cell surface TLR4 was apparently up-regulated, confirming a role for MD-2 in optimal expression of TLR4 on the cell surface (Fig. 1,A, bottom). MTS510 was able to detect cell surface TLR4 only when MD-2 was coexpressed (Fig. 1 A, bottom row).

To study a role for MD-2 in cell surface expression of TLR4 on normal immune cells, we examined bone marrow-derived macrophages and DCs prepared from wild-type, TLR4^−/−, and MD-2^−/− mice. As published previously (16), MTS510 only bound to wild-type cells but not to TLR4^−/− or MD-2^−/− cells. In contrast, Sa15-21 was able to detect TLR4 on DCs or macrophages from MD-2^−/− mice (Fig. 1, B and C). Although a role for MD-2 in augmenting cell surface expression of TLR4 was confirmed, MD-2 was found to be dispensable for cell surface expression of TLR4 at least in certain types of cells such as HEK293 cells, bone marrow-derived DC, or macrophages.

We have previously demonstrated with Ba/F3 cells that the above two mAbs to TLR4, MTS510 and Sa15-21, reacted with TLR4/MD-2 but poorly with TLR4 alone as judged by immunoprecipitation (19, 27). Sa15-21 was not able to detect cell surface TLR4 alone on Ba/F3 cells by flow cytometry. Because Sa15-21 was found to react with TLR4 on HEK293 cells, MD-2-deficient macrophages, or DCs, Sa15-21 but not MTX510 was able to bind to TLR4 without MD-2. TLR4 without MD-2 expressed in Ba/F3 cells might be distinct in conformation from TLR4 on HEK293 cells, macrophages, or DCs.

We hypothesized that cell surface expression of TLR4 is regulated not only by MD-2 but also by another unknown molecule. With an aim to identifying such a molecule, we sought for a molecule coprecipitated with TLR4-GFP in HEK293 cells. An ∼40 kDa signal was consistently coprecipitated with TLR4/MD-2, and we therefore determined its N-terminal amino acid sequence. The determined amino acid sequence (EENDWVRLPS) was completely matched with aa 38–47 of a gene containing CAG trinucleotide repeats (28). A similar 40-kDa molecule was also found to be coprecipitated with TLR4 in Ba/F3 cells, and its N-terminal amino acid sequence was determined. The N-terminal amino acid sequence was precisely the same portion of the mouse homologue (Fig. 2,A) that was previously described as a retinoic acid-induced gene (29). We hereafter describe this molecule as PRAT4A, a protein associated with TLR4A. Data base analyses identified a molecule similar to PRAT4A. We named it PRAT4B, which was described elsewhere (30). The experimentally determined N-terminal amino acid sequence of PRAT4A clearly identified the first 37 amino acids as a signal sequence (Fig. 2,A, boxed area). The mature peptide has a canonical N-glycosylation site and six cysteine residues. PRAT4A is highly conserved among a variety of species, including cow, chick, pig, catfish, sea squirt, nematode, zebra fish, and fruit fly (29). We examined tissue distribution of mRNA for mouse PRAT4A. Abundant expression of PRAT4A mRNA was detected in all of the organs tested, including lymphoid organs such as spleen and thymus (Fig. 2 B).

We first confirmed the physical association between TLR4 and PRAT4A. HEK293 cells expressing mouse TLR4-GFP or TLR4-GFP/MD-2 were transiently transfected with human PRAT4A tagged with the Flag epitope followed by the His₆ epitope (PRAT4A-FH). TLR4-GFP and PRAT4A-FH were precipitated, and coprecipitated RPAT4-FH or TLR4-GFP was immunoprobed by the anti-Flag mAb or anti-GFP Ab, respectively (Fig. 3,A). Mouse TLR4-GFP was coprecipitated with PRAT4A-FH (Fig. 3,A, upper panel, lanes 4 and 6), and human PRAT4A-FH was coprecipitated with mouse TLR4-GFP (lower panel, lanes 4 and 6). The amount of PRAT4A coprecipitated with TLR4-GFP appeared to decrease when MD-2 was coexpressed (Fig. 3 A, upper, lanes 4 and 6).

To confirm the physical association between TLR4 and PRAT4A with another cell line and to study the interaction between PRAT4A and other TLRs, Ba/F3 transfectants expressing TLR4-FH, TLR4-FH/MD-2, MD-2-FH, or TLR2-FH were used for immunoprecipitation studies. Coprecipitation of endogenously expressed PRAT4A was detected by a mAb to PRAT4A. PRAT4A was coprecipitated with TLR4-FH and TLR4-FH/MD-2, but not with MD-2-FH, TLR2-FH, or RP105/MD-1 (Fig. 3,B, and data not shown), demonstrating specific association between TLR4 and PRAT4A. In contrast to the results with HEK293 cells, PRAT4A coprecipitated with TLR4 increased with MD-2 coexpression (Fig. 3,B, lanes 2 and 3). The association of human PRAT4A with mouse TLR4 might be weaker than that of mouse PRAT4A and may be affected by MD-2 association. Next, epitope-tagged PRAT4A (PRAT4A-FH) was expressed in Ba/F3 cells expressing TLR4, MD-2, and CD14. PRAT4A-FH was precipitated by the anti-Flag mAb, and coprecipitated TLR4 and MD-2 were immunoprobed (Fig. 3,C). Interestingly, the smaller species of TLR4 and MD-2 were coprecipitated with PRAT4A-FH (Fig. 3,C, lane 3). Considering that PRAT4A-FH was not associated with MD-2 alone (Fig. 3,B), PRAT4A-FH was likely to be associated with the hypoglycosylated form of TLR4 that was expressed either alone or as a complex with the hypoglycosylated form of MD-2. As published previously (19), coexpressed MD-2 facilitated TLR4 glycosylation as revealed by the larger form in SDS-PAGE (Fig. 3,B, lane 2 compared with lane 3). The glycosylated mature form was preferentially precipitated by the MTS510 mAb (compare Fig. 3,B, lane 2 with C, lane 1). Coprecipitation of PRAT4A-FH was hardly detected when the MTS510 mAb was used for immunoprecipitation of TLR4/MD-2 (Fig. 3 C, middle, lane 1). The MTS510 mAb would bind to the mature TLR4/MD-2 complex from which PRAT4A has been already dissociated, or the mAb might disrupt the association between PRAT4A and TLR4/MD-2.

To address a role for PRAT4A in cell surface expression of TLR4, we used siRNA for inhibiting the expression of endogenous PRAT4A. The effect of siRNA was validated by detecting the transiently expressed human PRAT4A protein (Fig. 4,A) or mRNA encoding endogenously expressed PRAT4A by RT-PCR (Fig. 4,B), both of which were apparently down-regulated by two distinct siRNA, siRNA1 and siRNA2, but not by sequence-scrambled control RNA, ctrl1 and ctrl2 (Fig. 4, A and B). The PRAT4A siRNA did not down-regulate mRNA expression encoding human TLR4-FH, indicating specific silencing by the siRNA (Fig. 4,B). HEK293 cells stably expressing human TLR4 were transiently transfected with the human PRAT4A siRNA. Cell surface TLR4 was detected by flow cytometry with an anti-human TLR4 mAb, HTA125 (Fig. 4,C). The PRAT4A siRNA down-regulated cell surface human TLR4 without influencing the expression of mRNA encoding hTLR4-FH (Fig. 4 B). Cell surface CD44 was not influenced by the siRNA, and the sequence-scrambled control siRNA had no effect on cell surface hTLR4-FH.

We next asked whether PRAT4A regulates cell surface expression of TLR4/MD-2. We established two Ba/F3 cell lines, PRAT4A8 and PRAT4A11, stably expressing TLR4/MD-2 and the mouse PRAT4A shRNA. The effect of the PRAT4A shRNA was confirmed by semiquantitative PCR (Fig. 5,A). The PRAT4A mRNA was reduced to 15% (PRAT4A8) or 13% (PRAT4A11) by the shRNA when compared with the Ba/F3 transfectant without transduction or with the Ba/F3 transfectant expressing an empty vector (Fig. 5,A). PRAT4A mRNA reduction by ∼85% appeared to be sufficient for down-regulating cell surface TLR4/MD-2 on Ba/F3 cells to the undetectable level by flow cytometry (Fig. 5,B). Such a drastic down-regulation was not observed in cell surface TLR2 or CD43 (Fig. 5,B, and data not shown). We asked whether the lack of cell surface TLR4/MD-2 influenced LPS responses. Because the Ba/F3 transfectants had been transfected with a luciferase reporter construct driven by NF-κB activity, transfectants were stimulated with lipid A, LPS, another TLR4/MD-2 ligand (Taxol), or TNF-α. Luciferase activity in the cell lysate was determined (Fig. 5,C). Whereas Ba/F3 transfectants without any treatment or with an empty control vector both responded to the TLR4 ligands, those with the PRAT4A shRNA did not respond to lipid A or LPS from E. coli even at a concentration as high as 10 μg/ml (Fig. 5 C). Similar results were obtained with another TLR4/MD-2 ligand (Taxol) (data not shown). PRAT4A knockdown had no effect on NF-κB activation induced by TNF-α (data not shown).

To ask how the PRAT4A shRNA led to the lack of cell surface TLR4/MD-2 and hyporesponsiveness to LPS, the expression of TLR4 and MD-2 was studied. RT-PCR revealed that the amount of mRNA encoding TLR4 or MD-2 was not down-regulated by the PRAT4A shRNA (Fig. 6,A). Because TLR4 and MD-2 had been tagged with the FH epitope, these molecules were precipitated with the anti-Flag mAb and immunoprobed with the anti-His₆ mAb, respectively. The amount of TLR4 protein was decreased in one of the transfectants, PRAT4A11, but not in the other transfectant, PRAT4A8, when compared with Ba/F3 cells without transduction or with a control vector (Fig. 6,B, lanes 4 and 5). The MD-2 protein was comparably expressed in all the lines. Despite the expression of TLR4, PRAT4A was not coprecipitated with TLR4 in the transfectants expressing PRAT4A shRNA, demonstrating its silencing effect (Fig. 6,B, middle). The MTS510 mAb was specific for the TLR4/MD-2 complex and the preferentially precipitated mature form of TLR4 (compare TLR4 in Fig. 6,B, lane 2, between the upper and lower panels). Despite the expression of TLR4 and MD-2, TLR4 was much less precipitated by the MTS510 mAb in the transfectants with RPAT4A shRNA (Fig. 6 B, lower panel), indicating the lack of mature TLR4. These results demonstrate an important role for PRAT4A in TLR4 maturation.

To confirm that the role for PRAT4A on TLR4 was not restricted to in vitro cell lines such as Ba/F3 or HEK293 cells, bone marrow-derived DCs were transduced with retrovirus encoding the PRAT4A shRNA or an empty control vector. Cell surface expression of TLR4 alone, TLR4/MD-2, or TLR2 was determined by flow cytometry analyses. TLR4 alone was detected on MD-2-deficient DCs without any treatment or with an empty control vector but not on those with the PRAT4A shRNA (Fig. 7,A). Similarly, cell surface TLR4/MD-2 on wild-type DCs was drastically down-regulated with the PRAT4A shRNA (Fig. 7,B). The control vector had no effect on cell surface expression of TLR4/MD-2. To make sure that the TLR4 protein is expressed in PRAT4A-silenced DCs, DCs expressing the PRAST4A shRNA were enriched and subject to immunoprecipitation with the anti-TLR4 mAb Sa15-21 and to immunoprobing with anti-TLR4 Ab (Fig. 7,C). We were able to detect the TLR4 protein in DCs expressing the RPAT4A shRNA, but the amount of TLR4 protein in PRAT4A-silenced DCs was much less than the amount in control cells, DCs without any treatment, or DCs expressing an empty vector. When compared with TLR4/MD-2, cell surface TLR2 was much less influenced by the PRAT4A shRNA (Fig. 7 B). However, we cannot exclude a possibility that PRAT4A knockdown had a small but significant effect on cell surface expression of TLR2. This issue is currently being studied further.

The present study identified RPAT4A as a novel molecule that is associated with TLR4 and regulates its cell surface expression. This conclusion was based on the following results. First, physical association between TLR4 and PRAT4A was shown in human and mouse cell lines, HEK293 cells, and Ba/F3 cells (Fig. 3). Second, gene silencing of the PRAT4A mRNA down-regulated cell surface TLR4 alone on human HEK293 cells and mouse MD-2-deficient DCs (Figs. 4 and 7). Third, PRAT4A knockdown decreased cell surface TLR4/MD-2 on Ba/F3 transfectants and wild-type DCs. Fourth, mature TLR4/MD-2 precipitated by the MTS510 mAb was reduced by the PRAT4A knockdown (Fig. 6).

PRAT4A binds hypoglycosylated TLR4 with or without MD-2 (Fig. 3,C), but not with mature TLR4/MD-2. PRAT4A knockdown decreased the amount of mature TLR4 precipitated with the MTS510 mAb, but not that of immature TLR4 (Fig. 6,B). These results suggest a role for PRAT4A in TLR4 maturation/glycosylation. Although the MTS510 mAb was unable to precipitate TLR4/MD-2 in Ba/F3 transfectants with RPAT4A shRNA (Fig. 6 B), it is possible that hypoglycosylated TLR4 is still associated with hypoglycosylated MD-2 even in the lack of PRAT4A. A role for PRAT4A in the assembly of the TLR4/MD-2 complex remains to be clarified.

An ER chaperon, gp96, was shown to have a similar role. Randow and Seed (22) described a pre-B cell line, deficient in gp96, in which TLR4 failed to associate with MD-2, get to the cell surface, and respond to LPS. Altered localization of gp96 from ER to the membrane led to the redistribution of TLR4 (20). gp96 regulates the subcellular distribution of not only TLR4 but also TLR2, TLR1, and some integrins (22). Although PRAT4A is not associated with TLR2, it remains to be determined whether PRAT4A is associated with other TLRs, including TLR1 or TLR6.

A majority of TLR4 was detected inside the cells in the Golgi apparatus (16) as a hypoglycosylated immature form as judged by SDS-PAGE analyses (Fig. 3,B, lane 3). Overexpression of MD-2 increased the mature form of TLR4 on the cell surface, but most TLR4 still remained immature in the Golgi apparatus (Fig. 3 B, lane 2) (16). There must be machinery regulating the subcellular distribution of TLR4 between Golgi/ER and the cell surface. The Golgi apparatus is reported to be another site allowing TLR4/MD-2 to recognize LPS. TLR4 in intestinal epithelial cells resides in the Golgi apparatus, and internalized LPS is recognized in the Golgi apparatus (20). The molecular mechanisms by which TLR4/MD-2 stays in Golgi/ER or traffics back from the cell surface are currently unknown. PRAT4A is likely to be a component of the machinery facilitating TLR4/MD-2 trafficking to the cell surface. Further study might reveal a difference in cytokine production between TLR4/MD-2 on the cell surface and in the Golgi apparatus.

PRAT4A knockdown led to the profound defect in LPS responsiveness that is probably due to the impaired maturation of TLR4, leading to the lack of mature TLR4/MD-2 on the cell surface (Fig. 5). PRAT4A was required for cell surface expression of TLR4/MD-2, not only in Ba/F3 cells but also in normal DCs (Fig. 7), suggesting a role for PRAT4A in DC response to LPS. We are currently studying a role for PRAT4A in LPS responses by DCs or macrophages by using shRNA. Up-regulation or down-regulation of PRAT4A may influence LPS responses. PRAT4A was previously shown to be induced by all-trans retinoic acid (29). It is important to study induction and reduction of PRAT4A mRNA upon a variety of stimulations.

In conclusion, the present study identified a novel molecule that is associated with TLR4 and regulates cell surface expression of TLR4. This molecule might reveal a novel mechanism regulating the subcellular distribution of TLR4.

We thank Dr. Shizuo Akira (Osaka University, Osaka, Japan) for providing us with TLR4-deficient mice.

The authors have no financial conflict of interest.