Protein to Regulate Endotoxin Function Protein CRISPLD2 Is a Critical ...

The Journal of Immunology

The Novel Lipopolysaccharide-Binding Protein CRISPLD2 Is a Critical Serum Protein to Regulate Endotoxin Function1 Zhi-Qin Wang,2* Wen-Ming Xing,2* Hua-Hua Fan,2† Ke-Sheng Wang,* Hai-Kuo Zhang,* Qin-Wan Wang,* Jia Qi,* Hong-Meng Yang,* Jie Yang,† Ya-Na Ren,‡ Shu-Jian Cui,* Xin Zhang,* Feng Liu,* Dao-Hong Lin,‡ Wen-Hui Wang,‡ Michael K. Hoffmann,§ and Ze-Guang Han3* LPS is an immunostimulatory component of Gram-negative bacteria. Acting on the immune system in a systemic fashion, LPS exposes the body to the hazard of septic shock. In this study we report that cysteine-rich secretory protein LCCL domain containing 2 (CRISPLD2/Crispld2; human and mouse/rat versions, respectively), expressed by multitissues and leukocytes, is a novel LPS-binding protein. As a serum protein, median CRISPLD2 concentrations in health volunteers and umbilical cord blood samples are 607 ␮g/ml and 290 ␮g/ml, respectively. Human peripheral blood granulocytes and mononuclear cells including monocytes, NK cells, and T cells spontaneously release CRISPLD2 (range, 0.2– 0.9 ␮g/ml) and enhance CRISPLD2 secretion (range, 1.5– 4.2 ␮g/ml) in response to stimulation of both LPS and humanized anti-human TLR4-IgA Ab in vitro. CRISPLD2 exhibits significant LPS binding affinity similar to that of soluble CD14, prevents LPS binding to target cells, reduces LPS-induced TNF-␣ and IL-6 production, and protects mice against endotoxin shock. In in vivo experiments, serum Crispld2 concentrations increased in response to a nontoxic dose of LPS and correlated negatively with LPS lethality, suggesting that CRISPLD2 serum concentrations not only are indicators of the degree of a body’s exposure to LPS but also reflect an individual’s LPS sensitivity. The Journal of Immunology, 2009, 183: 6646 – 6656.

P

lant and animal life evolved in a symbiotic relationship with the microbial world. Higher animals as well as humans are exceedingly conscious of structural and functional microbial components and entrust them with vital roles in the regulation of their life functions (1, 2). Decades ago immunologists identified, in the cell wall of Gram-negative bacteria, a highly bioactive LPS composed of a monomorphic lipid core and a polymorphic polysaccharide coat that stimulates immune functions in a systemic rather than local fashion (3). LPS-producing Gram-negative bacteria pose a threat to the health of mammals and may kill them by septic shock (4, 5). Mammals express an LPS-binding protein (LBP)4 that is considered a critical molecule in LPS-elicited activation cascades. LBP assembles upon reaction with LPS and several molecules on the cell membrane, thus forming a molecular signal transduction *Shanghai-MOST Key Laboratory for Disease and Health Genomics, Chinese National Human Genome Center, Shanghai, China; †Department of Blood Engineering, Shanghai Blood Center, Shanghai, China; and ‡Department of Pharmacology and § Department of Microbiology and Immunology, New York Medical College, Valhalla, NY 10595 Received for publication July 18, 2008. Accepted for publication September 19, 2009. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1

This study was supported by Chinese High-Tech Research and Development Program Grants 2006AA02Z193 and 2006AA02A305, National Natural Science Foundation for Outstanding Youth Grant 30425019, Chinese National Key Program on Basic Research Grant 2004CB518605, and the Shanghai Commission for Science and Technology.

2

Z.-Q.W., W.-M.X., and H.-H.F. contributed equally to this work.

3

Address correspondence and reprint requests to Dr. Ze-Guang Han, ShanghaiMOST Key Laboratory for Disease and Health Genomics, Chinese National Human Genome Center at Shanghai, 351 Guo Shou-Jing Road, Shanghai 201203, China. E-mail address: [email protected]

4

Abbreviations used in this paper: LBP, LPS-binding protein; CHO, Chinese hamster ovary; CRISP, cysteine-rich secretory protein; CRISPLD2/Crispld2, CRISP LCCL domain containing 2 (human/mouse or rat); hTLR4, human TLR4; LTA, lipoteichoic

www.jimmunol.org/cgi/doi/10.4049/jimmunol.0802348

complex (6 –11) that initiates the clonal expansion of immunocytes or the secretion of immunoregulatory cytokines (11). LPS-binding reagents that down-regulate immunological LPS activities have been reported (12–15). Of note is CD6, belong to the scavenger receptor cysteine-rich superfamily, which is readily expressed on the surface of lymphocytes but maintains a low serum concentration (16). It exhibits significant LPS-binding affinity, and its soluble form has been shown to inhibit the induction of endotoxic shock in mice (17). Passively administered LPS-reactive Abs have also been shown to ameliorate Gramnegative sepsis (18 –20). In this study we present an LPS-binding molecule, cysteinerich secretory protein (CRISP) LCCL domain containing 2 (CRISPLD2 and Crispld2, representing human and mouse/rat versions, respectively), which was previously known for a variety of other functions. Because it is also known as late gestation lung 1 (Lgl1), it has been implicated in the development of rat lung (21– 23) and mouse kidney (24) and is thought to be involved in the development of nonsyndromic cleft lip with or without cleft palate (25). Its possession of two LCCL domains suggests a relationship to LPS. LCCL structures were initially described in the horseshoe crab (Limulus) factor C, which, by binding LPS, initiates the Limulus coagulation cascade to protect the crab against bacterial infection (26 –28). It seemed conceivable that mammalian Crispld2, with two LCCL domains, exhibits particular avidity for LPS. Several known LPS-binding components of pulmonary surfactant, such as surfactant proteins A and D, have been implicated in an Ab-independent host defense against pathogens (29, 30). Previously, the analysis of leukocytes from healthy human volunteers

acid; ORF, open reading frame; RU, relative response unit; SPR, surface plasmon resonance. Copyright © 2009 by The American Association of Immunologists, Inc. 0022-1767/09/$2.00

The Journal of Immunology i.v. administered bacterial endotoxin has shown the rapidly increased transcript of CRISPLD2 but not CRISPLD1 within 2 h (31), suggesting that transcription of CRISPLD2 in leukocytes is immediately initiated in response to LPS challenge. The present work introduces mammal CRISPLD2 as a major serum protein that acts as a natural LPS antagonist and promises to be of considerable preventative value against endotoxic shock.

Materials and Methods Bacterial components, mAbs, and rCRISP-3 protein Escherichia coli (O127:B8 and O55:B5) LPS was purchased from SigmaAldrich and dissolved in PBS at 1 mg/ml. Alexa Fluor 488-labeled E. coli (O55:B5) and Salmonella enterica serovar Minnesota LPS was purchased from Invitrogen and dissolved in PBS at a final concentration of 1 mg/ml. Purified Staphylococcus aureus lipoteichoic acid (LTA), ultra-pure Rhodobacter sphaeroides LPS, and chimeric anti-human TLR4 (hTLR4)IgA mAbs were purchased from InvivoGen and dissolved in endotoxinfree water according to the manufacturer’s instruction. Anti-hTLR4 (HTA125) and OKT3 mAbs were purchased from eBioscienc. Human IgA was purchased from Sigma-Aldrich. Recombinant CRISP-3 was purchased from R&D Systems.

Plasmid constructs and cell transfection cDNA with an entire open reading frame (ORF) of human CRISPLD2 was obtained from mixed tissues by PCR using the F1 forward primer 5⬘GCTGTCGCCGCTGCTACCGC-3⬘ and the R1 reverse primer 5⬘GACGCCCCTTCTCCCCTGGT-3⬘. The PCR product was inserted into the plasmid pGEM-T Easy Vector (Promega), which then served as the template in expression vector construction. To construct a recombinant prokaryote expression vector, partial human CRISPLD2 was inserted into pGEX-4T-1 (GE Healthcare) to generate the GST-CRISPLD2 (257– 497 aa) fusion protein, which contains two LCCL domains and was used for generating the rabbit anti-CRISPLD2 polyclonal Ab. For mammalian cell expression, human CRISPLD2 with the entire ORF was then subcloned into pcDNA3.1A-myc-His (Invitrogen) to generate the c-myc-His-tagged CRISPLD2. The pcDNA3.1A-CRISPLD2 was transfected into CHO cells by Lipofectamine (Invitrogen) and then selected with the antibiotic G418 (Invitrogen). These stable CHO cells with human CRISPLD2 were maintained with G418, and many monoclonal recombinant CHO cell lines were further screened out by limited cell dilution and Western blotting assays.

Anti-CRISPLD2 polyclonal antibody The recombinant pGEX-4T-1 containing GST-CRISPLD2 (257– 497 aa, with an LCCL domain) was transformed into E. coli BL21 for prokaryote expression. The GST-CRISPLD2 expression in bacteria was induced by isopropyl-␤-D-thiogalactopyranoside at 28°C, followed by centrifugation at 8,000 ⫻ g at 4°C for 10 min. The supernatant was discarded and the pellet was resuspended in PBS with 5 mM DTT and 100 ␮g/ml PMSF. After sonication, the lysate was centrifuged at 8,000 ⫻ g again at 4°C for 10 min. The GST-CRISPLD2 (257– 497 aa) in the supernatant was purified using an affinity column of glutathione-Sepharose 4B (GE Healthcare). Normal rabbits were immunized with the purified protein and the polyclonal Abs were purified from the sera using protein G-Sepharose (GE Healthcare).

Purification of rCRISPLD2 The serum-free cell medium of CHO cells with stably transfected pcDNA3.1A containing human CRISPLD2 was collected. The rCRISPLD2 in the supernatant was denaturalized by 4 M urea and purified by Ni-NTA metal-affinity chromatography (Qiagen). After two washes with 50 ␮l of washing buffer (50 mM Na2HPO4, 300 mM NaCl, and 4 M urea; (pH 7.5)), the purified protein was eluted in buffer with 250 mM imidazole (pH 7.8). The collected fractions were then pooled and dialyzed against PBS. SDSPAGE was used to evaluate the purity of CRISPLD2-myc-His fusion by Coomassie blue staining. The BCA assay kit (Pierce) was used to determine protein concentration.

Native PAGE mobility shift assay To evaluate the interaction of CRISPLD2 with LPS, a native PAGE mobility shift assay was used as described (7). Briefly, the purified rCRISPLD2 protein was incubated with either E. coli or S. enterica serovar Minnesota LPS at 37°C for 30 min. Bromophenol blue and glycerol were loaded into samples and electrophoresis was preformed on 10% native polyacrylamide gel to measure the mobility of the mixtures with different LPS doses in a running

6647 buffer without detergents by 140 V for 4 h, as compared with that of rCRISPLD2 alone. The gels were stained with silver and scanned by Image Scanner (GE Healthcare).

Surface plasmon resonance (SPR) assay To measure the affinity between CRISPLD2 and LPS, a SPR assay was performed to determine the recognition of LPS by CRISPLD2 with a Biacore 3000 biosensor instrument (Biacore/GE Healthcare) using a C1 sensor chip (GE Healthcare). Briefly, either rCRISPLD2 or unrelated IgG at 0.18 mg/ml diluted with acetate buffer (pH 4.5) was immobilized to the sensor chip by coupling with the primary amine group. Different concentrations of E. coli LPS were injected into the flow cells at a rate of 20 ␮l/min. Sterilized and deaerated PBS was used as the running buffer during experiments. The sensorgram and relative response units (RU) for the binding of LPS to immobilized CRISPLD2 were obtained by subtracting the background of the LPS binding to immobilized, unrelated IgG. Following removal of the unbound LPS by injection with PBS, the bound LPS was removed by quick injection with 50 mM NaOH at a flow rate of 50 ␮l/min. A saturation binding curve was depicted as a function of LPS concentrations vs RU.

LPS binding affinity analysis The dissociation constant of LPS (KD) was calculated by Scatchard plot analysis using the following formula: KD ⫽ ⫺1/slope ⫽ (RUmax ⫺ RU)/ (RU/ConcLPS), where RUmax is maximum response and ConcLPS is the concentration of LPS. In addition, KD can be calculated as the ratio of these two constants (koff/kon). In brief, the dissociation rate constant koff is described by the equation Rt ⫽ Rt0e⫺koff(t ⫺ t0), and Rt0 is the amplitude of the initial response. The association rate constant kon can be derived from the measured koff value using: Rt ⫽ Rmax [1 ⫺ e⫺(konC ⫹ koff)(t ⫺ t0)], where C is the concentration of LPS, Rmax represents the maximum binding capacity of LPS to immobilized CRISPLD2, and Rt represents the amount of LPS bound to the CRISPLD2 at time t. The dissociation rate constant koff is determined when the LPS passing over CRISPLD2 on the surface of chip is replaced by PBS alone.

Northern blot analysis and RT-PCR Northern blotting was performed by using a human multiple tissue Northern blot membrane (Clontech) according to the manufacturer’s instruction. The probe for detecting CRISPLD2 expression was amplified by PCR using the F2 forward primer 5⬘-TGGAATTCCGAGAAGAAACCTACA CTC-3⬘ and the R2 reverse primer 5⬘-TGCTCGAGATTCACTGCCTGAC AGCA-3⬘ labeled with [␣-32P]dCTP using a random primer DNA labeling kit (Takara). Hybridization with a probe to human ␤-actin in the same membrane was used as a loading control. For RT-PCR, total RNA from human immune cells and mice tissues was extracted by TRIzol reagent (Invitrogen). Reverse transcription was performed in 20-␮l reactions using 2 ␮g of total RNA. Each PCR was generally performed with ⬃30 –32 thermal cycles, and the PCR products were detected by electrophoresis on a 2% agarose gel with ␤-actin used as a loading control. The PCR primers used in this study for Crispld2 (mouse) were 5⬘-GCGGTCGACGCCAT GAGCTGCGTCCTG-3⬘ (forward) and 5⬘-GCGCTCGAGTCACTGCCT GACAGCAAAGATC-3⬘ (reverse), and the primers used for CRISPLD2 (human) were 5⬘-AGT/CTAGACATGAGCTGCGT-3⬘ (forward) and 5⬘GGGAATTCTCTGCCTGACAG-3⬘ (reverse). The PCR primers used for human TLR4 were 5⬘-CTGCAATGGATCAAGGACCA-3⬘ (forward) and 5⬘-TCCCACTCCAGGTAAGTGTT-3⬘ (reverse), those used for MD2 were 5⬘-TTCCACCCTGTTTTCTTCCA-3⬘ (forward) and 5⬘- TCATCA GATCCTCGGCAAAT-3⬘) (reverse), and the PCR primers used for ␤-actin were 5⬘-TGAACCCTAAGGCCAACCGTG-3⬘ (forward) and 5⬘-GCTC ATAGCTCTTCTCCAGGG-3⬘ (reverse).

Isolation of human immune cells and flow cytometry analyses Human peripheral blood was donated by healthy volunteers and PBMCs were isolated by Ficoll (Pharmacia/GE Healthcare) density gradient centrifugation. After whole blood was collected in acidic citrate dextrose (1:5 (v/v) blood) and centrifuged at room temperature to form a leukocyte layer, granulocytes were precipitated by Ficoll (Pharmacia/GE Healthcare) density gradient centrifugation from the resuspension of leukocytes layer, and then remaining erythrocytes were removed by hypotonic lysis. Generally, T cells, NK cells, and monocytes were isolated from human PBMCs by negative selection using specific Ab-coated magnetic beads (Dynabeads) according to manufacturer’s instructions (Invitrogen). To further confirm the up-regulation of CRISPLD2 secretion from purified human T cells by activation of TLR4, CD4 T cells (95% purity) were isolated twice by a CD4 T cell negative selection kit (Miltenyi Biotec). For RT-PCR analysis

6648

SERUM LIPOPOLYSACCHARIDE-BINDING PROTEIN CRISPLD2

of TLR4 transcript in T cells, the CD4⫹ T cells (98% purity) were first isolated by CD4 T cells negative selection kit and then, in second step, purified by a CD4 T cell positive selection kit. CD14⫹ cells, as a positive control for TLR4 transcript, were purified by a CD14 monocyte positive selection kit (Miltenyi Biotec). To assess CRISPLD2 on cell surfaces, PBMCs (106 cells/ml) in PBS with 1% inactivated human serum were seeded into 96-well plates and incubated with 15 ␮g/ml rabbit anti-CRISPLD2 polyclonal Ab for 30 min. After washing three times with buffer, the cells were stained with Cy2-labeled goat anti-rabbit IgG for 30 min. Normal rabbit IgG was used as a negative control. To determine intracellular CRISPLD2, immunofluorescence assays were used to simultaneously detect surface cluster of differentiation markers and intracellular CRISPLD2 of these cells. PBMCs were seeded into 96-well plates, stained with PE-labeled (red) anti-CD4, CD8, CD56, or CD14 mAb (BD Biosciences) in the presence of 2% inactivated human serum, and fixed with 4% paraformaldehyde in PBS. After washing twice with Perm/Wash buffer (BD Pharmingen), the cells were incubated with either rabbit anti-CRISPLD2 polyclonal Ab or normal rabbit IgG for at 15 ␮g/ml for 1 h. The cells were washed three times with 1⫻ Perm/Wash buffer again and stained by Cy2-labeled (green) goat anti-rabbit IgG for 30 min. T cell-, NK cell-, and monocyte-specific surface markers and the intracellular CRISPLD2 were detected by a FACSCalibur flow cytometer (BD Biosciences). T cells, NK cells, and monocytes were gated according to their characteristic surface marker vs forward scatter characteristics. Granulocytes were gated according to their typical forward vs side scatter characteristics.

Immunoblotting analysis The samples of serum and plasma from human volunteers and mice were diluted 1/10 in PBS, separated by electrophoresis in 10% SDS-PAGE gels and then transferred onto nitrocellulose membranes (GE Healthcare). The references of rCRISPLD2 in PBS (0.3, 0.1, 0.03, and 0.01 mg/ml) were used as control for diluted samples. The electrophoresis was performed in loading buffer without DTT for retaining IgG above the 150 kDa position on a gel to avoid overlap of the IgG H chain with CRISPLD2/Crispld2. The membranes were blocked with 5% dry skim milk and incubated with rabbit anti-CRISPLD2 polyclonal Ab at 37°C for 2 h, followed by a secondary Ab coupled with fluorescein (Rockland) at 37°C for 1 h. The blotting membranes were finally scanned by the Odyssey infrared imaging system (LICOR Biosciences). The serum CRISPLD2 concentration, represented by the density of each spot, was calculated according to reference of the rCRISPLD2 standard.

Immunofluorescence assay and immunohistochemical staining For identification of Crispld2 in mouse and rat tissues, freshly sacrificed mice and rats were arterially perfused with 4% paraformaldehyde in PBS. The excised tissues were fixed in the same buffer. Paraffin-embedded tissue sections (6 ␮m) were mounted on glass slides, the sections were deparaffinized and rehydrated and the mouse placenta tissue was excised and frozen then sectioned with a cryostat at 6 ␮m. All tissue sections on glass slides were stained for 60 min with anti-CRISPLD Ab diluted in PBS containing 1% Tween 20. For immunofluorescent staining, Cy2-labeled second Ab was used for recognition of the primary Ab. Hoechst dye was used for nuclear staining. Immunofluorescence was visualized using a confocal microscope (LSM 510, Version 3.5 META; Zeiss). For immunochemical staining, the HRP-conjugated anti-rabbit secondary Ab (DakoCytomation) was used as the secondary Ab, and the signals were detected using a diaminobenzidine substrate kit (Vector Laboratories). Nuclear staining was conducted by hematoxylin.

LPS binding to immune cells PBMCs were seeded into 96-well plate at 2 ⫻ 106 cells/ml in PBS containing 0.3% BSA with different doses of rCRISPLD2 or rCRISP-3 (0, 0.018, 0.054, 0.178, 0.535, 1.78, and 5.35 ␮M) in the presence of Alexa Fluor 488-labeled E. coli LPS (1.0 and 3.0 ␮g/ml) or S. enterica serovar Minnesota LPS (9.0 ␮g/ml) for 30 min. After washing twice, the mean fluorescent intensity of Alexa Fluor 488 on monocytes and lymphocytes was examined by a FACSCalibur flow cytometer (BD Biosciences). Monocyte and lymphocyte regions were distinguished according to their forward vs side scatter characteristics. The fluorescence signal was represented by mean fluorescence intensity units.

Stimulation of immune cells in vitro For assessing CRISPLD2 secretion from cells in vitro, human granulocytes at 1 ⫻ 106 cells/ml, PBMCs, purified T cells, and NK cells at 2 ⫻ 106 cells/ml, and purified monocytes at 0.5 ⫻ 106 cells/ml were seeded into 96-well plates, stimulated by LPS and LTA, respectively, in RPMI 1640

medium containing 0.1% FCS, and incubated in 5% CO2 at 37°C. To determine the reduction by CRISPLD2 of LPS-induced TNF-␣ and IL-6, PBMCs at 1 ⫻ 106 cells/ml were incubated in the presence of multiple concentrations of rCRISPLD2 or rCRISP-3 in RPMI 1640 medium with 0.3% FCS in 5% CO2 at 37°C and stimulated by LPS (10 ␮g/ml). To assess the regulatory function of endogenous and exogenous CRISPLD2, human PBMCs were incubated in RPMI 1640 with 0.1% FCS in 5% CO2 at 37°C, and the other supplement regents, including human serum, rCRISPLD2, E. coli LPS, anti-CRISPLD2 Ab, and control Ab, were used.

ELISA Commercial TNF-␣ and IL-6 ELISA kits (R&D Systems) were used to measure human cytokines in culture supernatants and serum according to manufacturer’s instruction. For measuring CRISPLD2/Crispld2 in culture supernatants and mouse serum, appropriately diluted samples in sodium carbonate buffer (pH 9.5)-coated 96-well microtiter plates (Nunc MaxiSorp) overnight at 4°C. After washing twice with PBS, the plates were first incubated with 8% FCS in PBS for 60 min and subsequently with rabbit anti-CRISPLD2 Ab (10 ␮g/ml) in PBS with 8% FCS for 2 h. After washing four times, HRP-conjugated donkey anti-rabbit IgG Ab (Jackson ImmunoResearch Laboratories) was added. The tetramethylbenzidine substrate in sodium acetate/citric acid buffer (0.1 M; pH 6.0) was used for visualization. The catalytic action was stopped by 2.0 M sulfuric acid. OD was measured in a spectrophotometer at 450 nm with the correction wavelength set at 570 nm. A standard curve of rCRISPLD2 titration (2.0, 1.0, 0.5, 0.25, 0.125, 0.062, 0.031,and 0 ␮g/ml) was used for the quantification of CRISPLD2 and for the reference of mouse Crispld2. It should be pointed out that the concentration of bovine Crispld2 in FCS is very low, and the incubation with 8% FCS does not significantly increase the background in ELISA (see supplemental Fig. S7).5 For measuring anti-LPS Ab in mouse serum, LPS (100 ng/ml) in sodium carbonate buffer (pH 9.5) was plated in 96-well microtiter plates overnight, followed by treatment with 10% FCS in PBS for 60 min. Mouse sera in 1- to 100-fold dilutions in PBS were added to the immobilized LPS, and HRPconjugated donkey anti-mouse Ab (Jackson ImmunoResearch Laboratories) was used to react with plate-bound anti-LPS Abs.

Mice and in vivo LPS stimulation Six- to 8-wk-old BALB/c female mice were purchased from SIPPR-BK Experimental Animal Ltd. All mice were housed in the animal facility of Shanghai Jiao Tong University Medical School (Shanghai, China). All procedures with mice were performed according to guidelines of the Harvard Medical School Office for Research Subject Protection (Boston, MA) and approved by the ethics committee of the Chinese National Genome Center (Shanghai, China). To determine serum Crispld2 concentration in response to LPS and LTA, 6- to 8-wk-old BALB/c mice were injected i.p. with S. aureus LTA (0.1 mg/mouse) and E. coli-type LPS (0.015, 0.030, and 0.10 mg/mouse), respectively, and blood samples were taken from the tail vein on days 0, 1, 2, 5, 7, 12, and 31. For the prevention of LPS-induced endotoxin shock, 6- to 8-wk-old BALB/c female mice were injected i.p. with E. coli (O55:B5) LPS (0.45 mg/mouse; 22.5 mg/kg) in the absence or presence of recombinant human CRISPLD2 (molecular mass, 55 kDa; 1.4 mg/mouse; 70 mg/kg) and CRISP-3 (molecular mass, 25 kDa; 0.64 mg/ mouse; 32 mg/kg), respectively. To monitor the serum CRISPLD2 response to repeated LPS challenges, 6- to 8-wk-old BALB/c female mice were first injected i.p. with E. coli (055:B5) LPS (0.03 mg/mouse; 1.5 mg/kg). After 10 days, these mice were separated into four groups and again injected i.p. with LPS at 25, 40, 55, and 70 mg/kg, respectively. After 32 days, surviving mice received a third injection of LPS (45, 60, or 75 mg/kg). Serum Crispld2 and anti-LPS Ab were examined before i.p. injection of high doses of LPS.

Statistics analyses Student’s t test was used for the comparison of two independent groups. Statistical difference of survival rates between two groups with endotoxin shock was analyzed by a Kaplan-Meyer log-rank test. Statistic analyses on the correlation between serum Crispld2 concentrations vs lethal LPS dose were performed by GraphPad Prism 5 software. For all tests, a value of p ⬍ 0.05 was considered statistically significant.

Results Molecular analysis and LPS reactivity To examine the possibility that CDRISPLD2 is a LPS-binding protein, we isolated the whole ORF of human CRISPLD2 and stably 5

The online version of this article contains supplemental material.


6649 Table I. Affinity of LPS to its receptors and soluble binding proteins Immobilized

CRISPLD2 CRISPLD2 LBP CD14 CD14 MD-2 TLR4 a b

LPS KD (M)a

Refs. ⫺6

2.46 ⫾ 0.8 ⫻ 10 1.33 ⫾ 0.9 ⫻ 10⫺6 1.40 ⫻ 10⫺8 0.98 ⫻ 10⫺5 0.87 ⫾ 0.2 ⫻ 10⫺5 2.33 ⫾ 0.9 ⫻ 10⫺6 1.41 ⫾ 0.7 ⫻ 10⫺5

Fig. 1D Fig. 1Bb 8 8 32 32 32

All KD values were measured by SPR assay. Data from three independent experiments (mean ⫾ SD).

rized in Table I, indicating that the affinity of CRISPLD2 for LPS is similar to that for CD14 and MD2 but lower than that for LBP (8, 32). To assess the molecular phylogeny of CRISPLD2, we also compared CRISPLD2 with homologous protein sequences deposited in GenBank by using bioinformatic tools. We found that CRISPLD2 is highly conserved throughout evolution (supplemental Fig. S2). Tissue expression and serum concentration

FIGURE 1. CRISPLD2 binds LPS but not S. aureus LTA. A, Reaction with S. enterica serovar Minnesota (S. minnesota) and E. coli LPS shifts the mobility of rCRISPLD2 in a native PAGE mobility shift assay. B and C, Sensorgrams of E. coli LPS (B) and S. aureus LTA (C) binding to immobilized CRISPLD2. Either E. coli LPS or S. aureus LTA (0.3, 1.0, 3.0, 10, and 30 ␮M) was passed over immobilized CRISPLD2. The dissociation constant of LPS, KD, was calculated as the ratio of these two constants (koff/kon) according to the binding parameters from a sensorgram. D, RU that reflects E. coli LPS binding activity was measured by a BIAcore instrument when E. coli LPS (1.0, 3.0, 10, 30, and 100 ␮M) was passed over the CRISPLD2 chip. E, Scatchard plot analysis and the KD of E. coli LPS binding to CRISPLD2 (mean ⫾ SD).

transfected it into CHO cells using the mammalian expression plasmid pcDNA3.1 for expressing the recombinant c-myc-His tagged CRISPLD2. rCRISPLD2 was purified from culture medium and evaluated by electrophoresis on SDS-polyacrylamide gels and by immunoblotting assays using c-Myc Abs and homemade anti-CRISPLD2, respectively (supplemental Fig. S1). To assess potential CRISPLD2 LPS-binding activity, we used an EMSA on native polyacrylamide gel using purified rCRISPLD2 together with LPS from E. coli or S. enterica serovar Minnesota. The electrophoretic mobility of the CRISPLD2 band was significantly shifted by E. coli and S. enterica serovar Minnesota LPS (Fig. 1A), indicating that CRISPLD2 did interact physically with LPS. To further confirm the finding, we performed a SPR assay by using a BIAcore 3000 biosensor to quantitatively assess the recognition of E. coli LPS by CRISPLD2; S. aureus LTA was used as a control. The sensorgrams showed that E. coli LPS, but not S. aureus LTA, binds to immobilized CRISPLD2 (Fig. 1, B and C). The relative RU of LPS binding to immobilized CRISPLD2 was measured by subtracting a background of LPS binding to immobilized, unrelated IgG. The binding of E. coli LPS to immobilized CRISPLD2 occurs in a dose-dependent manner (Fig. 1D). Therefore, we tentatively assessed an apparent KD that is 2.46 ⫾ 0.84 ␮M for the binding of LPS to CRISPLD2 (Fig. 1E). The LPS affinity to CRISPLD2 was compared with LPS affinity to known soluble binding proteins and receptors by using an SPR assay as summa-

RT-PCR technology was applied to demonstrate the occurrence of Crispld2 mRNA in six mouse tissues (supplemental Fig. S3A). In addition, Crispld2 was immunofluorescently identified in tissue sections of mouse lung, placenta, and small intestine (supplemental Fig. S3B), and also immunohistochemically identified in rat lung and intestine (supplemental Fig. S3C). Northern blot analysis revealed that the CRISPDL2 gene is transcribed in leukocytes and many human organs; it is most pronounced in the placenta and also in heart, lung, small intestine, and blood leukocytes, but far less pronounced in liver, kidney, spleen, thymus, colon, muscle, and brain (supplemental Fig. S3D). CRISPLD2 was identified and quantified in healthy human serum (range, 384 –790 ␮g/ml; n ⫽ 7) and plasma (range, 402–998 ␮g/ml; n ⫽ 3) by using an immunoblotting assay (Fig. 2A). However, the concentrations of CRISPLD2 in infant umbilical cord blood plasma (range, 149 – 426 ␮g/ml; n ⫽ 5) are lower than those in adult blood plasma (Fig. 2B). As compared with published serum concentrations of known LBPs, the serum levels of CRISPLD2 are substantially higher than those of LBP, soluble CD14, bactericidal/permeability-increasing protein, and soluble CD6 (supplemental Table I, Fig. 2, A and B, and Refs. 16, 33, and 34). Moreover, human peripheral blood granulocytes and mononuclear cells, including monocytes, NK cells and T cells, express CRISPLD2 that was identified by RT-PCR (Fig. 2C) in concordance with published microarray data for gene expression (35–38). CRISPLD2 was readily seen in the cytoplasm of PBMCs (Fig. 2, D and F), including monocytes, NK cells, and T cells (supplemental Fig. S4F), by using intracellular Ab staining, but we were unable to detect it on the cell surface by using extracellular Ab staining (Fig. 2E). Interestingly, intracellular Crispld2 was also found in lamina propria lymphocytes in the small intestines of mouse and rat (supplemental Fig. S3, B and C). Soluble rCRISPLD2 interferes with LPS target cell binding The ability of soluble rCRISPLD2 to intercede in the interaction of LPS with its target cells was tested under several experimental conditions. Fig. 3, A and B, revealed that the presence of CRISPDL2, but not CRISP-3, in the culture medium inhibited the reaction of two LPS preparations with cellular LPS receptors on monocytes in a dose-dependent fashion. Accordingly, the release of the cytokines TNF-␣ (Fig. 3C) and IL-6 (Fig. 3D) subsided in

6650


FIGURE 2. CRISPLD2 occurs in cells and in serum. A, Immunoblots (IB) of 1/10 diluted sera from healthy volunteers (right; n ⫽ 7) and the rCRISPLD2 as reference (left) were used for assessing the quantity of the protein in human sera. B, Comparison of CRISPLD2 plasma levels in the blood of healthy adults (n ⫽ 3) with the levels infant umbilical cord blood (n ⫽ 5). The density of spots on film was scanned and calculated by using an Odyssey infrared imaging system. No DTT was in the loading buffer. Recombinant human CRISPLD2 served as a reference standard. C, CRISPLD2 mRNA from subpopulations of human white blood cells, including granulocytes, PBMCs, monocytes, T cells, and NK cells was subjected to RT-PCR. ␤-Actin served as the loading control. RNA extracted from CHO cell line with stable CRISPLD2 transfection was used as positive control (⫹), and clean water was used as a negative control (⫺). D, Confocal microscope photographs of human PBMCs fixed with 4% paraformaldehyde, permeated with 0.1% Triton, and stained with anti-CRISPLD2 Ab as compared with normal rabbit IgG; goat anti-rabbit IgG-Cy2 Ab was used as a secondary Ab (red). The photographs are in 200-fold (panels 1 and 3) and 1000-fold (panels 2 and 4) magnification. E, PBMCs were directly stained with rabbit anti-CRISPLD2 Ab (open area) or a control Ab (shaded areas) in PBS. F, PBMCs were permeabilized by Perm/Wash buffer and stained with rabbit anti-CRISPLD2 Ab (open area) or control Ab (shaded areas) in Perm/Wash buffer; Cy2-labled anti-rabbit IgG Ab served as a secondary Ab. A fluorescent signal was only obtained in the cytoplasm and not on the surface of cells.

the presence of the LPS-binding molecule but not in the presence of CRISP-3, a cysteine-rich secretory protein without LCCL domain (Fig. 3, C and D). TLR4 has been found on surface of NK cells (39). Fluorescencelabeled E. coli and S. enterica serovar Minnesota LPS were detectable on CD56⫹ cells and CD56⫺ lymphocytes in our experiment (supplemental Fig. S4, D and E), although the LPS fluorescence intensity on these cells was far lower than that on monocytes (supplemental Fig. S4, A–E). Expectedly, rCRISPLD2 prevented LPS binding to CD56⫹ cells and CD56⫺ lymphocytes (supplemental Fig. S4, D and E). To determine whether CRISPLD2 is capable of dislodging LPS after it has reacted with cellular surface receptors, PBMCs were incubated with labeled LPS for 30 min before unbound LPS was removed and CRISPLD2 was added. Fig. 3, E and F, suggest that in sufficient doses, competitive removal of bound LPS from monocyte surfaces occurs in the presence of CRISPLD2. Fig. 3G revealed that LPS-induced TNF-␣ production was significantly altered even when CRISPLD2 was added after LPS. However, the reduced CRISPLD2 inhibitory activity at 6 h after LPS may reflect a prior commitment of target cells to LPS response rather than consolidation of LPS binding. LPS and an anti-hTLR4-IgA mAb up-regulate CRISPLD2 secretion In the course of these experiments, it was noted that not only does CRISPDL2 curtail LPS bioregulatory function, but also in turn LPS up-regulates the release of CRISPLD2 by granulocytes and PBMCs, including purified monocytes, T cells, and NK cells. Spontaneously, the cells emitted lower quantities of CRISPLD2

into the culture medium, but quite substantial amounts accumulated in the presence of LPS (Fig. 4, A–E). It is possible that LPS up-regulates CRISPLD2 release in such a short time (Fig. 4, A–E) due to constant transcription and translation of CRISPLD2 in the these cells (Fig. 2, C–F, and supplemental Fig. S4F). The upregulation of CRISPLD2 was observed not only in leukocyte culture but also in animals. The Crispld2 levels rose substantially in the serum of LPS-treated mice and seemed to remain higher over 1 wk. Fig. 4G shows immunoblot data obtained with sera from mice taken several days after treatment with nontoxic doses of LPS. A similar picture presents itself in an ELISA of Crspld2 in such sera (Fig. 4F). Rhodobacter sphaeroides LPS is the potent antagonist of toxic LPS in both human and murine cells and prevents LPS-induced shock in mice (40). In the present study we found that ultra-pure R. sphaeroides LPS alone up-regulated CRISPLD2 secretion and induced significantly lower TNF-␣ production (Table II). To determine whether the LPS-induced up-regulation of CRISPLD2 secretion could be blocked by a neutralizing mAb against TLR4, we used an anti-TLR4 (HTA125) mAb and a humanized chimeric anti-hTLR4-IgA mAb in vitro experiments. According to the reference data from the manufacturer (InvivoGen), the efficacy of the humanized chimeric IgA mAb is 100-fold higher than that of the HTA125 mAb for blocking LPS-induced intracellular activation. In concordance with the reference data, the anti-hTLR4-IgA mAb effectively suppressed E. coli LPS-induced TNF-␣ and IL-6 production in our experimental system (Table II). Unexpectedly, both anti-TLR4 mAbs were failed to block E. coli LPS-induced upregulation of CRISPLD2 secretion. Interestingly, we found that the anti-hTLR4-IgA mAb alone, but not anti-TLR4 (HTA125) mAb or


6651

FIGURE 3. CRISPLD2 prevents the binding of LPS to target cells and inhibits the release of proinflammatory factors. A, Soluble rCRIPLD2 but not rCRISP-3 prevents Alexa Fluor 488-labeled E. coli LPS (1.0 ␮g/ ml) binding to human PBMCs. Monocytes and lymphocytes were gated respectively, and mean fluorescence intensity (MFI) was determined by using flow cytometry analysis. Open symbols represent lymphocytes, and closed symbols represent monocytes. The absence of LPS (No LPS) was used as a negative control. B, The same experimental system was used with Alexa Fluor 488-labeled S. enterica serovar Minnesota (S. minnesota) LPS (9.0 ␮g/ml). C and D, Soluble rCRISPLD2, but not rCRISP-3, inhibits E. coli (Ec) LPS (10 ng/ml)induced TNF-␣ (C) and IL-6 (D) release from human PBMCs. Open symbols represent the absence of LPS stimulation used as a negative control, and closed symbols represent the presence of E. coli LPS. E and F, Soluble rCRISPLD2 dislodges cell-bound LPS from monocyte surfaces. PBMCs were incubated with Alexa Fluor 488-labeled E. coli (3.0 ␮g/ml) or S. enterica serovar Minnesota (S. m) (9.0 ␮g/ml) LPS for 30 min before unbound LPS was removed and rCRISPLD2 was added. After incubation with rCRISPLD2 for 30 min at 4°C, the bound LPS on monocytes (closed symbols) and lymphocytes (open symbols) were analyzed by using flow cytometry. The absence of LPS (No LPS) was used as a negative control. G, Recombinant CRISPLD2 interferes with LPS-induced cytokine production even when added several hours after LPS induction. Human PBMCs were stimulated with E. coli LPS (100 ng/ml) and rCRISPLD2 was added at the indicated time points. Cytokines released in the culture supernatants at 18 h were measured by ELISA, and the absence of rCRISPLD2 (⫺) was used as a positive control.

control human IgA, was capable of up-regulating CRISPLD2 secretion (Table II and supplemental Fig. S5A). Moreover, the chimeric anti-hTLR4-IgA mAb synergized the E. coli LPS-induced up-regulation of CRISPLD2 release (Table II). Together, the results in Table II suggested that the up-regulation of CRISPLD2 secretion can be triggered by TLR4-mediated signaling that seems to be irrelevant to TNF-␣ production. It is known that purified S. aureus LTA is a ligand of CD6 (17) that can activate TLR2 but no other TLRs, including TLR4 (41). We investigated whether purified S. aureus LTA up-regulates CRISPLD2 release in vivo and in vitro. Fig. 4, A and B, and supplemental Fig. S5B show that purified S. aureus LTA (1 ␮g/ml) failed to up-regulate CRISPLD2 release from PBMCs, purified CD4 T cells, and monocytes. Furthermore, Crispld2 serum levels

of purified S. aureus LTA-treated mice (100 ␮g/per-mouse) did not significantly rise (Fig. 4F), implying that S. aureus LTA binding to CD6 and S. aureus LTA-induced activation of TLR2 is irrelevant to the up-regulation of CRISPLD2 release. Whether the activation of other TLRs, except TLR4 and TLR2, up-regulate CRISPLD2 release is to be further investigated. To determine whether the TLR4/MD2 receptor complex is expressed on T lymphocytes, RT-PCR was applied to demonstrate the occurrence of TLR4 and MD2 mRNA in PBMCs, purified CD4⫹ T cells (98% pure), and CD14⫹ monocytes (96% pure), and PE-labeled anti-TLR4 mAb was also used to determine whether TLR4 anchors on the surfaces of CD4⫹ and CD8⫹ T cells and CD14 monocytes in PBMCs. Supplemental Fig S4G revealed that TLR4 and MD2 transcripts were detected

6652


FIGURE 4. LPS promotes human granulocytes and PBMCs to secrete CRISPLD2 and increases the CRISPLD2 serum levels. A and B, Human PBMCs (A) and monocytes (B) were stimulated with E. coli LPS and S. aureus LTA, respectively, and CRISPLD2 in supernatants were quantitated at various time points. C–E, Human granulocytes (C), NK cells (D), and T cells (E) were stimulated with or without E. coli LPS, and the CRISPLD2 concentrations in culture medium were quantitated by ELISA (mean of three experiments ⫾ SD). F and G, Up-regulation of Crispld2 expression in mouse serum after administration of a nontoxic dose of LPS. BALB/c mice were injected i.p. with S. aureus LTA (100 ␮g/mouse) and E. coli type LPS (15, 30 and 100 ␮g/mouse) respectively. F, Western blotting (IB, immunoblotting) was used to determine Crispld2 in mice sera (E. coli LPS 100 ␮g/ mouse) at days (D) 0, 1, 2, 5 and 7. The mice sera were diluted 1/10 in PBS for PAGE separation, so real concentrations of Crispld2 are ⬃10 times higher. G, Serum Crispld2 from mice treated with S. aureus LTA (100 ␮g/mouse) or E. coli LPS (15, 30 and 100 ␮g/mouse) was measured by ELISA at different time points. Recombinant human CRISPLD2 was used as reference for western blot and ELISA.

in CD4⫹ T cells as compared with CD14 monocytes and PBMCs. Moreover, supplemental Fig S4H demonstrated that TLR4 was found on the cell surfaces of CD4 and CD8 T lymphocytes via immunofluorescence assay, which is consistent with the published literature (42– 44). To further confirm LPS binding to TLR4 on T cells, the anti-hTLR4-IgA mAb was used to interfere with the binding of FITClabeled E. coli LPS to PBMCs stained with PE-conjugated anti-CD4 and CD8 mAbs, respectively. The results revealed that anti-hTLR4IgA mAb, but not control IgA, partially blocked the binding of E. coli LPS-FITC to CD4⫹ and CD8⫹ T lymphocytes (supplemental Fig. S4I), suggesting that LPS can bind to TLR4 on the surface of T cells.

To exclude the possibility that the up-regulation of CRISPLD2 secretion from T cells is a secondary response or side effect of contamination that could stimulate T lymphocytes, in the present work CD4⫹ T cells from PBMCs were purified twice by negative selection to reach high purity (95%), and the anti-hTLR4-IgA mAb was extensively dialyzed to exclude possible contamination. Also, a stimulatory anti-CD3 mAb (OKT3) was used to mimic contamination for activating the TCR/CD3 complex. The results showed that the stimulation of anti-hTLR4-IgA mAb, but not pure S. aureus LTA or human IgA, effectively increased the CRISPLD2 secretion from purified CD4 T cells in a short time, as compared with


6653

Table II. Both R. sphaeroides LPS and anti-hTLR4-IgA Ab up-regulate CRISPLD2 secretion and reduce E. coli LPS-induced TNF-␣ and IL-6 production b

Treatmenta

Concentration (␮g/ml)

CRISPLD2b (␮g/ml)

TNF-␣ (ng/ml)

IL-6b (ng/ml)

Controlc E. coli LPS R. sphaeroides LPS E. coli LPS ⫹ R. sphaeroides LPS Anti-hTLR4-IgA Anti-hTLR4-IgA E. coli LPS ⫹ anti-hTLR4-IgA E. coli LPS ⫹ anti-hTLR4-IgA

1 1 1⫹1 1 5 1⫹1 1⫹5

0.37 ⫾ 0.10 1.59 ⫾ 0.26 1.40 ⫾ 0.06 1.65 ⫾ 0.09 1.66 ⫾ 0.04 1.93 ⫾ 0.21 2.25 ⫾ 0.24e 2.55 ⫾ 0.43e

0.04 ⫾ 0.01 8.56 ⫾ 0.62 3.45 ⫾ 0.27d 3.29 ⫾ 0.07d 0.08 ⫾ 0.03 0.13 ⫾ 0.01 3.17 ⫾ 0.09d 2.63 ⫾ 0.07d

0.11 ⫾ 0.02 10.5 ⫾ 0.28 7.98 ⫾ 0.22d 7.25 ⫾ 0.59d 0.11 ⫾ 0.03 0.07 ⫾ 0.03 7.06 ⫾ 0.45d 5.27 ⫾ 0.68d

a

Human PBMCs were treated by the reagents as indicated in table for 12 h. CRISPLD2, TNF-␣, and IL-6 were measured by ELISA in culture supernatants. Ten microliters of PBS was added in culture as control. Data (mean ⫾ SD of triplicate wells) are from two independent experiments. d Value of p ⬍ 0.005 as compared with that from E. coli LPS treatment alone. e Value p ⬍ 0.05 as compared with that from E. coli LPS treatment alone. b c

that of the ultra pure R. sphaeroides LPS (supplemental Fig. S5B), whereas the OKT3 mAb-induced activation of T cells did not increase CRISPLD2 secretion although the proinflammatory mediators, including TNF-␣ from PBMCs, were induced (supplemental Fig. S5A), suggesting that the activation of TLR4 on T cells can directly lead to the up-regulation of CRISPLD2 secretion. The collective results, as shown in Fig. 4 and supplemental Fig S5, A and B, seem to exclude the possibility that the up-regulation of CRISPLD2 secretion is a secondary response, such as proinflammatory mediator-induced response, or a side effect of contamination, such as the direct stimulation of Ags via the TCR/CD3 complex or unknown molecules (⬍20 kDa). However, we cannot exclude the possibility that the LPS-induced up-regulation of CRISPLD2 secretion can be mediated by the activation of other LPS-receptors such as CD6, which binds to LPS and LTA, respectively, via TLR4-independent and nonoverlapping extracellular sites and then delivers intracellular signaling in the presence of LPS (17). The question of whether LPS-activated CD6 can up-regulate CRISPLD2 secretion should be further investigated in the future. CRISPLD2 is a major biological regulator of LPS function It is possible that endogenous CRISPLD2 may down-regulate LPS immunostimulatory activities. If indeed the LPS sensitivity of monocytes in PBMCs was to correlate negatively with the availability of endogenous CRISPLD2 in culture medium, then experimentally interfering with the interaction between LPS and CRISPLD2 should increase LPS sensitivity. To test this hypothesis, human PBMCs were incubated with increasing concentrations of LPS in the presence or absence of CRISPLD2-specific Abs, and then the release of CRISPLD2 and TNF-␣ were examined 18 h later. The experiment revealed that PBMCs spontaneously released CRISPLD2 into culture medium and that the higher doses of LPS, ⬎10 ng/ml, up-regulated the release (Fig. 5A). In concordance with our hypothesis, the efficacy of LPS was increased in the presence of Ab against CRISPLD2 as compared with the control Ab, suggesting indeed that the antiCRISPLD2 Ab can remove an effective safety device against LPS (Fig. 5B). Intruding germs reach their cellular targets via the bloodstream. Previous analysis revealed substantial but varying CRISPLD2 quantities in the serum of humans (Fig. 2A). It seemed of interest to determine whether CRISPLD2 molecules contained in the serum provide animals and humans with a natural shield against the effects of LPS. Human PBMCs were challenged under culture con-

ditions with 0.1% FCS in the presence or in the absence of 10 – 15% human serum. Using three different samples, Fig. 5C shows that human serum suppressed LPS-induced TNF-␣ release readily. However, the suppressive effect was reversed by the addition of the Ab against CRISPLD2, but not the control Ab. The evidence revealed that the endogenous CRISPLD2 in sera down-regulated LPS-induced TNF-␣ production. To further confirm this result, rCRISPLD2 was added into the culture system in the presence or absence of 10% human serum. Fig. 5D demonstrates that the increased concentrations of rCRISPLD2 inhibit LPS-induced TNF-␣ production in a dose dependent manner and that the antiCRISPLD2 Ab reverses the inhibitory effects of not only endogenous CRISPLD2 but also those of additional rCRISPLD2 in culture medium. CRISPDL2 protects mice against endotoxic shock As the above experiments demonstrated that CRISPLD2 specifically inhibits biological functions of LPS, we examined whether CRISPLD2 can ameliorate the devastating effects of systemic LPS action in generating toxic shock in mice. Fig. 6A revealed that one i.p. injection of 450 ␮g of E. coli LPS killed 80% of the treated animals within a few days. Injection of E. coli LPS in combination with 1.4 mg of rCRISPLD2 reduced the death rate to ⬍20%, suggesting that CRISPLD2 exerts life-saving effects of anti-endotoxin in vivo (Fig. 6A). By contrast, CRISP-3 as a control has no protective effect. It has previously been shown that pretreatment of mice with small, nontoxic doses of LPS render recipients resistant to a lethal challenge with LPS (45, 46). This finding was consistent with our notion that LPS stimulates the generation of endogenous CRISPLD2. We therefore considered the possibility that CRISPLD2 elevation induced by a low dose of LPS may account for the protection against the lethal effect of a high dose of LPS. We found that the serum of mice that have been treated with a small dose of LPS more than doubled its CRISPLD2 expression (Fig. 4G and Table III). At the same time, we also noticed that CRISPLD2 was not the only LBP that accumulated in the serum. LPS-binding IgG rose by an even higher margin (supplemental Fig. S6). Table III showed that under such conditions mice experienced substantial protection against LPS-induced lethal shock. Plotted against a lethal LPS dose, it was shown that CRISPLD2 serum levels as well as LPS-reactive IgG serum levels correlate similarly and reciprocally with LPS lethality (Fig. 6B and supplemental Fig. S6). However, it has been known preadministration of the J5 LPS vaccine increases the concentration of serum anti-LPS core Abs in individuals (20). Whether both LPS-binding structures synergistically account for enhanced LPS resistance is to be further determined.

6654


FIGURE 5. Endogenous CRISPLD2 interferes with LPS-induced TNF-␣ response. A, PBMCs were stimulated with the increased concentration of E. coli LPS as indicated. The endogenous CRISPLD2 released in the culture supernatants was quantified at 18 h by ELISA, and the medium cultured without cells was used as a negative control. B, In the same experiment, anti-CRISPLD2 polyclonal Abs (䡲) and unrelated polyclonal Abs (E) were added into cell cultures as compared with cultures without Abs (F). TNF-␣ released in the culture supernatants at 18 h was measured by ELISA. C, Human serum (HS) down-regulates LPS-induced TNF-␣ production. Sera from three healthy volunteers were used in the experiment. PBMCs were stimulated by E. coli LPS (100 ng/ml) with anti-CRISPLD2 polyclonal Ab or a control Ab (both 500 ␮g/ml) in the presence or absence of 15% human serum. TNF-␣ in culture supernatants were measured at 18 h by ELISA according to the protocol provided by the manufacturer (R&D Systems); LPS alone (open bars), LPS with control IgG (gray bars), LPS with the anti-CRISPLD2 polyclonal Ab (black bars) are shown; ⴱⴱ, p ⬍ 0.05; ⴱⴱⴱ, p ⬍ 0.005 vs absence of human serum. D, Anti-CRISPLD2 polyclonal Abs against endogenous CRISPLD2 and rCRISPLD2 unleashes LPS immunostimulatory activities. In the same culture system with or without human serum (HS) 10% (#11), PBMCs were stimulated by E. coli LPS (10 ng/ml) in the presence of anti-CRISPLD2 polyclonal Ab or a control Ab (both 500 ␮g/ml) with additional rCRISPLD2 at the indicated concentrations. TNF-␣ released in the culture supernatants at 18 h was measured by ELISA. LPS with human serum (open bars), LPS with human serum plus control IgG (gray bars), LPS with human serum plus anti-CRISPLD2 Ab (black bars), and LPS with only 0.1% FCS (diagonal bars) are shown; ⴱⴱ, p ⫽ 0.048; and ⴱⴱⴱ, p ⫽ 0.00036; vs presence of anti-CRISPLD2 Ab (mean of three experiments ⫾ SD).

Discussion CRISPLD2 is a molecule recognized in the literature, but no LPSrelated functions are attributed to it (21–25). We were intrigued by the fact that it contains two LCCL domains, which are characteristic LPS-binding structures (26). Our data indeed show, for the first time, that CRISPLD2 is a potent LPS-binding protein that exhibits significant LPS-binding affinity. A major LPS-binding protein in the mammalian body, one known to guide the transduction of signals involved in eliciting LPS-dependent immune functions, is LBP (6, 7). CRISPLD2 shares with LBP none of the tasks examined here. On the contrary, the postulate that it could oppose LBP-mediated immunostimulation is supported by the finding that the polyclonal Ab against the CRISPLD2 region containing LCCL domains unleashes LPS immunostimulatory activities (Fig. 5). The mechanism of this antagonism is unclear. CRISPLD2 could inhibit LBP-mediated immune functions by competing for the target molecule, by facilitating negative signaling, or by using both mechanisms. CRISPLD2 is not the only molecule the body may engage in the fight against bacterial LPS. We noted in our experiments that LPSreactive Igs have a similar effect. Recently, it was also reported

that CD6 acts as negative regulator of LPS function and endotoxic shock induction (17). However, it may be argued, in favor of CRISPLD2, that CD6 is highly expressed on cell membranes and poorly present in serum (16), whereas CRISPLD2 lacking membrane expression is amply represented in the serum. The downside of LPS-reactive Abs is the fact that they are primarily directed against the immunodominant species-specific oligosaccharide side chains, in contrast to CRISPLD2, and protect selectively, namely against the immunizing strain. The immunological effects of LPS are associated with the core glycolipid region in which there is little strain variation. Experimental vaccines have been produced using an E. coli J5 mutant that lacks side strains to its core. J5specific Abs have been demonstrated to provide some protection against endotoxic shock (20). Given the fact that CRISPLD2 effectively blocks LPS-induced immune functions, the uneven organ distribution of the molecule in the body may indicate that organs expressing high levels of CRISPLD2 would benefit from a persistent down-modulation of LPS immunostimulatory activity. The desirability of an abundant presence of CRISPLD2 in the placenta would seem evident from the fact that LPS is a strong inducer of fetal abortion (47).


6655 Table III. Correlation between serum Crispld2 level and LPS sensitivity LPS LPS Pretreatment Serum Crispld2 Administration Outcome a (0.03 mg) (OD 450 –570 nm) (mg/mouse)b (Survival/total mice)c No No Yes Yes Yes Yes

0.134 – 0.166 0.130 – 0.201 0.223– 0.327 0.323– 0.446 0.338 – 0.386 0.351– 0.371

0.5 0.8 0.5 0.8 1.1 1.4

0/4 0/4 3/4 4/6 4/4 0/2

a Serum Crispld2 was detected by ELISA assays on the 8th day after LPS pretreatment. b LPS was injected i.p. on the 10th day after LPS pretreatment. c Outcome on the 15th day after LPS pretreatment.

FIGURE 6. rCRISPLD2 protects mice against endotoxin shock. A, Mice were injected i.p. with E. coli LPS (450 ␮g/mouse, 22.5 mg/kg; n ⫽ 22) alone or in combination with rCRISPLD2 (1.4 mg/mouse, 70 mg/kg; n ⫽ 17) or CRISP-3 (0.64 mg/mouse, 32 mg/kg; n ⫽ 7). The percentages of surviving animals are plotted in a Kaplan-Meyer survival curve; p ⬍ 0.001 vs LPS with PBS or CRISP-3. B, The lethal LPS doses correlate positively with serum Crispld2 concentrations. Serum CRISPLD2 was measured by ELISA on the 8th day after i.p. preinjection with or without a nontoxic dose of LPS (0.03 mg/mouse). Lethal doses of LPS (0.5, 0.8, 1.1, or 1.4 mg/mouse) were injected i.p. on the 10th day after i.p. preinjection. The outcome was accounted on the 15th day after LPS pretreatment. Serum Crispld2 in surviving mice was examined again on the 30th day, surviving mice received a third injection of LPS (0.9, 1.2 or 1.5 mg/ mouse) on the 32nd day, and the outcome was accounted again on the 37th day. The data of lethal LPS dose vs serum Crispld2 concentration from each mouse were recorded. Together, all data were plotted by lethal LPS doses vs serum Crispld2 concentrations. Value of p ⫽ 0.00009 and value of F ⫽ 35.4. GraphPad Prism 5 software was used for statistic analysis. In regression analysis, logarithmic trendline program in Excel was best-fit to the values array in the plot, with the correlation coefficient R2 indicated on the graph.

cess of a systemic functional presence of LPS in Gram-negative bacteria. The prevention of endotoxin shock by CRISPLD2 is unequivocally demonstrated by our findings in the experimental mouse model. A drawback in conventional sepsis treatment is the fact that, once the first indications of an immanent shock are detected, the avalanche of immunological events is already in full motion and it can no longer be stopped. The most effective protection against these avalanches lies in preventing their initiation. If we were to draw the conclusion from our experiments that the risk of acquiring Gram-negative sepsis is high when CRISPLD2 serum levels are low and, vice versa, the risk is low when CRISPLD2 serum levels are high, it would logically follow that the best precaution a physician can take against the onset of Gram-negative sepsis is to assure that a patient possesses sufficient amounts of CRISPLD2 in his or her serum. We noted in these mice (Table III) that small nontoxic doses of LPS boost the serum expression of CRISPLD2 and protect animals against the lethal effect of the immunostimulatory agent. We recognize that E. coli LPS itself can hardly be considered a desirable therapeutic reagent, but it is conceivable that other less toxic microbial products or preparations may also up-regulate CRISPLD2 serum levels. A clinical concept of guarding against Gram-negative sepsis would thus entail the careful tracking of CRISPLD2 serum levels in patients with increased risk of developing Gram-negative septic shock and attempts to boost CRISPLD2 expression once its level falls below a critical threshold.

Acknowledgments We thank Q. N. Zhuang (GE Healthcare) for technical assistance on SPR.

Disclosures The authors have no financial conflict of interest.

The observations in mice regarding endotoxic shock prevention through CRISPLD2 treatment may have implications for improving strategies in treating this devastating clinical event in humans. Our data show that CRISPLD2 serum levels fluctuate in considerable amplitude, possibly due to communicating with a sensed microbial environment. Current results in this present work indicate that a large majority of white blood cells spontaneously release CRISPLD2 and, upon stimulation with LPS, enhance release of the LPS-binding molecule that would occur in vivo and may contribute to the serum CRISPLD2 level. We do not exclude the possibility that the heart, small intestine, and lung also release CRISPLD2 into body fluid because of apparent CRISPLD2 transcription in these tissues. Given the fact that CRISPLD2 blocks the immunostimulatory activity of LPS in cell culture, it would seem obvious that CRISPLD2 protects the body as well against the ex-

References 1. Danhorn, T., and C. Fuqua. 2007. Biofilm formation by plant-associated bacteria. Annu. Rev. Microbiol. 61: 401– 422. 2. Hooper, L. V., T. Midtvedt, and J. I. Gordon. 2002. How host-microbial interactions shape the nutrient environment of the mammalian intestine. Annu. Rev. Nutr. 22: 283–307. 3. Raetz, C. R., C. M. Reynolds, M. S. Trent, and R. E. Bishop. 2007. Lipid A modification systems in gram-negative bacteria. Annu. Rev. Biochem. 76: 295–329. 4. Beutler, B., and E. T. Rietschel. 2003. Innate immune sensing and its roots: the story of endotoxin. Nat. Rev. Immunol. 3: 169 –176. 5. Bosshart, H., and M. Heinzelmann. 2007. Targeting bacterial endotoxin: two sides of a coin. Ann. NY. Acad. Sci. 1096: 1–17. 6. Weiss, J. 2003. Bactericidal/permeability-increasing protein (BPI) and lipopolysaccharide-binding protein (LBP): structure, function and regulation in host defence against Gram-negative bacteria. Biochem. Soc. Trans. 31: 785–790. 7. Hailman, E., H. S. Lichenstein, M. M. Wurfel, D. S. Miller, D. A. Johnson, M. Kelley, L. A. Busse, M. M. Zukowski, and S. D. Wright. 1994. Lipopolysaccharide (LPS)-binding protein accelerates the binding of LPS to CD14. J. Exp. Med. 179: 269 –277.

6656


8. Thomas, C. J., M. Kapoor, S. Sharma, H. Bausinger, U. Zyilan, D. Lipsker, D. Hanau, and A. Surolia. 2002. Evidence of a trimolecular complex involving LPS, LPS binding protein and soluble CD14 as an effector of LPS response. FEBS. Lett. 531: 184 –188. 9. Gioannini, T. L., and J. P. Weiss. 2007. Regulation of interactions of Gramnegative bacterial endotoxins with mammalian cells. Immunol. Res. 39: 249 –260. 10. Miyake, K. 2006. Roles for accessory molecules in microbial recognition by Toll-like receptors. J. Endotoxin Res. 12: 195–204. 11. Kitchens, R. L., and P. A. Thompson. 2005. Modulatory effects of sCD14 and LBP on LPS-host cell interactions. J. Endotoxin Res. 11: 225–229. 12. Kim, H. S., J. H. Cho, H. W. Park, H. Yoon, M. S. Kim, and S. C. Kim. 2002. Endotoxin-neutralizing antimicrobial proteins of the human placenta. J. Immunol. 168: 2356 –2364. 13. Macagno, A., M. Molteni, A. Rinaldi, C. Rossetti, and F. Sallusto. 2006. A cyanobacterial LPS antagonist prevents endotoxin shock and blocks sustained TLR4 stimulation required for cytokine expression. J. Exp. Med. 203: 1481–1492. 14. Tan, N. S., B. Ho, and J. L. Ding. 2000. High-affinity LPS binding domain(s) in recombinant factor C of a horseshoe crab neutralizes LPS-induced lethality. FASEB. J. 14: 859 – 870. 15. Rosenfeld, Y., N. Papo, and Y. Shai. 2006. Endotoxin (lipopolysaccharide) neutralization by innate immunity host-defense peptides: peptide properties and plausible modes of action. J. Biol. Chem. 281: 1636 –1643. 16. Ramos-Casals, M., J. Font, F. Lozano, and M. Ingelmo. 2001. High circulating levels of soluble scavenger receptors (sCD5 and sCD6) in patients with primary Sjo¨gren’s syndrome. Rheumatology 40: 1056 –1059. 17. Sarrias, M. R., M. Farno´s, R. Mota, J. Ye´lamos, and F. Lozano. 2007. CD6 binds to pathogen-associated molecular patterns and protects from LPS-induced septic shock. Proc. Natl. Acad. Sci. USA 104: 11724 –11729. 18. Ziegler, E. J., J. A. McCutchan, J. Fierer, M. P. Glauser, J. C. Sadoff, H. Douglas, and A. I. Braude. 1982. Treatment of gram-negative bacteremia and shock with human antiserum to a mutant Escherichia coli. N. Engl. J. Med. 307: 1225–1230. 19. Baumgartner, J. D., M. P. Glauser, J. A. McCutchan, E. J. Ziegler, G. van Melle, M. R. Klauber, M. Vogt, E. Muehlen, R. Luethy, R. Chiolero, et al. 1985. Prevention of gram-negative shock and death in surgical patients by antibody to endotoxin core glycolipid. Lancet 2: 59 – 63. 20. Cross, A. S., S. Opal, P. Cook, J. Drabick, and A. Bhattacharjee. 2004. Development of an anti-core lipopolysaccharide vaccine for the prevention and treatment of sepsis. Vaccine 22: 812– 817. 21. Oyewumi, L., F. Kaplan, and N. B. Sweezey. 2003. Lgl1, a mesenchymal modulator of early lung branching morphogenesis, is a secreted glycoprotein imported by late gestation lung epithelial cells. Biochem. J. 376: 61– 69. 22. Oyewumi, L., F. Kaplan, S. Gagnon, and N. B. Sweezey. 2003. Antisense oligodeoxynucleotides decrease LGL1 mRNA and protein levels and inhibit branching morphogenesis in fetal rat lung. Am. J. Respir. Cell Mol. Biol. 28: 232–240. 23. Nadeau, K., R. P. Jankov, A. K. Tanswell, N. B. Sweezey, and F. Kaplan. 2006. Lgl1 is suppressed in oxygen toxicity animal models of bronchopulmonary dysplasia and normalizes during recovery in air. Pediatr. Res. 59: 389 –395. 24. Quinlan, J., F. Kaplan, N. Sweezey, and P. R. Goodyer. 2007. LGL1, a novel branching morphogen in developing kidney, is induced by retinoic acid. Am. J. Physiol. Renal. Physiol. 293: F987–F993. 25. Chiquet, B. T., A. C. Lidral, S. Stal, J. B. Mulliken, L. M. Moreno, M. Arcos-Burgos, C. Valencia-Ramirez, S. H. Blanton, and J. T. Hecht. 2007. CRISPLD2: a novel NSCLP candidate gene. Hum. Mol. Genet. 16: 2241–2248. 26. Trexler, M., L. Banyai, and L. Patthy. 2000. The LCCL module. Eur. J. Biochem. 267: 5751–5757. 27. Nakamura, T., F. Tokunaga, T. Morita, and S. Iwanaga. 1988. Interaction between lipopolysaccharide and intracellular serine protease zymogen, factor C, from horseshoe crab (Tachypleus tridentatus) hemocytes. J. Biochem. 103: 370 –374. 28. Nakamura, T., F. Tokunaga, T. Morita, S. Iwanaga, S. Kusumoto, T. Shiba, T. Kobayashi, and K. Inoue. 1988. Intracellular serineprotease zymogen factor C, from horseshoe crab hemocytes: its activation by synthetic lipid A analogues and acidic phospholipids. Eur. J. Biochem. 176: 89 –94.

29. Stamme, C., M. Mu¨ller, L. Hamann, T. Gutsmann, and U. Seydel. 2002. Surfactant protein A inhibits lipopolysaccharide-induced immune cell activation by preventing the interaction of lipopolysaccharide with lipopolysaccharide-binding protein. Am. J. Respir. Cell Mol. Biol. 27: 353–360. 30. Ohya, M., C. Nishitani, H. Sano, C. Yamada, H. Mitsuzawa, T. Shimizu, T. Saito, K. Smith, E. Crouch, and Y. Kuroki. 2006. Human pulmonary surfactant protein D binds the extracellular domains of Toll-like receptors 2 and 4 through the carbohydrate recognition domain by a mechanism different from its binding to phosphatidylinositol and lipopolysaccharide. Biochemistry 45: 8657– 8664. 31. Calvano, S. E., W. Xiao, D. R. Richards, R. M. Felciano, H. V. Baker, R. J. Cho, R. O. Chen, B. H. Brownstein, J. P. Cobb, S. K. Tschoeke, et al. 2005. A networkbased analysis of systemic inflammation in humans. Nature 437: 1032–1037. 32. Shin, H. J., H. Lee, J. D. Park, H. C. Hyun, H. O. Sohn, D. W. Lee, and Y. S. Kim. 2007. Kinetics of binding of LPS to recombinant CD14, TLR4, and MD-2 proteins. Mol. Cells 24: 119 –124. 33. Zweigner, J., H. J. Gramm, O. C. Singer, K. Wegscheider, and R. R. Schumann. 2001. High concentrations of lipopolysaccharide-binding protein in serum of patients with severe sepsis or septic shock inhibit the lipopolysaccharide response in human monocytes. Blood 98: 3800 –3808. 34. Juffermans, N. P., A. Verbon, S. J. van Deventer, W. A. Buurman, H. van Deutekom, P. Speelman, and T. van der Poll. 1998. Serum concentrations of lipopolysaccharide activity-modulating proteins during tuberculosis. J. Infect. Dis. 178: 1839 –1842. 35. Martinez, F. O., S. Gordon, M. Locati, and A. Mantovani. 2006. Transcriptional profiling of the human monocyte-to-macrophage differentiation and polarization: new molecules and patterns of gene expression. J. Immunol. 177: 7303–7311. 36. Sung, L. Y., S. Gao, H. Shen, H. Yu, Y. Song, S. L. Smith, C. C. Chang, K. Inoue, L. Kuo, J. Lian, et al. 2006. Differentiated cells are more efficient than adult stem cells for cloning by somatic cell nuclear transfer. Nat. Genet. 38: 1323–1328. 37. Hanna, J., P. Bechtel, Y. Zhai, F. Youssef, K. McLachlan, and O. Mandelboim. 2004. Novel insights on human NK cells’ immunological modalities revealed by gene expression profiling. J. Immunol. 173: 6547– 6563. 38. Hyrcza, M. D., C. Kovacs, M. Loutfy, R. Halpenny, L. Heisler, S. Yang, O. Wilkins, M. Ostrowski, and S. D. Der. 2007. Distinct transcriptional profiles in ex vivo CD4⫹ and CD8⫹ T cells are established early in human immunodeficiency virus type 1 infection and are characterized by a chronic interferon response as well as extensive transcriptional changes in CD8⫹ T cells. J. Virol. 81: 3477–3486. 39. O’Connor, G. M., O. M. Hart, and C. M. Gardiner. 2006. Putting the natural killer cell in its place. Immunology 117: 1–10. 40. Qureshi, N., B. W. Jarvis, and K. Takayama. 1999. Nontoxic RsDPLA as a potent antagonist of toxic lipopolysaccharide. In Endotoxin in Health and Disease, H. Brade, S. M. Opal, S. N. Vogel, and D. C. Morrison, eds. Marcel Dekker, New York, pp. 687– 697. 41. Schwandner, R., R. Dziarski, H. Wesche, M. Rothe, and C. J. Kirschning. 1999. Peptidoglycan- and lipoteichoic acid-induced cell activation is mediated by Tolllike receptor 2. J. Biol. Chem. 274: 17406 –17409. 42. Cairns, B., R. Maile, C. M. Barnes, J. A. Frelinger, and A. A. Meyer. 2006. Increased Toll-like receptor 4 expression on T cells may be a mechanism for enhanced T cell response late after burn injury. J. Trauma 61: 293–299. 43. Lewkowicz, P., N. Lewkowicz, A. Sasiak, and H. Tcho´rzewski. 2006. Lipopolysaccharide-activated CD4⫹CD25⫹ T regulatory cells inhibit neutrophil function and promote their apoptosis and death. J. Immunol. 177: 7155–7163. 44. Komai-Koma, M., D. S. Gilchrist, and D. Xu. 2009. Direct recognition of LPS by human but not murine CD8⫹ T cells via TLR4 complex. Eur. J. Immunol. 39: 1564 –1572. 45. Beeson, P. B. 1947. Tolerance to bacterial pyrogens. I. Factors influencing its development. J. Exp. Med. 86: 29 –37. 46. Cavaillon, J. M., C. Adrie, C. Fitting, and M. Adib-Conquy. 2003. Endotoxin tolerance: is there a clinical relevance? J. Endotoxin. Res. 9: 101–107. 47. Deb, K., M. M. Chaturvedi, and Y. K. Jaiswal. 2004. Comprehending the role of LPS in Gram-negative bacterial vaginosis: ogling into the causes of unfulfilled child-wish. Arch. Gynecol. Obstet. 270: 133–146.