The olfactomedin-4 positive neutrophil has a role in murine intestinal ischemia/reperfusion injury

Abstract

Olfactomedin-4 (OLFM4) identifies a subset of neutrophils conserved in both mouse and man, associated with worse outcomes in several inflammatory conditions. We investigated the role of OLFM4-positive neutrophils in murine intestinal ischemia/reperfusion (IR) injury. Wild-type (WT) C57Bl/6 and OLFM4 null mice were subjected to intestinal IR injury and then monitored for survival or tissues harvested for further analyses. In vivo intestinal barrier function was determined via functional assay of permeability to FITC-dextran. OLFM4 null mice had a significant 7-d survival benefit and less intestinal barrier dysfunction compared with WT. Early after IR, WT mice had worse mucosal damage on histologic examination. Experiments involving adoptive transfer of bone marrow demonstrated that the mortality phenotype associated with OLFM4-positive neutrophils was transferrable to OLFM4 null mice. After IR injury, WT mice also had increased intestinal tissue activation of NFκB and expression of iNOS, 2 signaling pathways previously demonstrated to be involved in intestinal IR injury. In combination, these experiments show that OLFM4-positive neutrophils are centrally involved in the pathologic pathway leading to intestinal damage and mortality after IR injury. This may provide a therapeutic target for mitigation of intestinal IR injury in a variety of common clinical situations.—Levinsky, N. C., Mallela, J., Opoka, A., Harmon, K., Lewis, H. V., Zingarelli, B., Wong, H. R., Alder, M. N. The olfactomedin-4 positive neutrophil has a role in murine intestinal ischemia/reperfusion injury.

Intestinal ischemia/reperfusion (IR) injury is a common clinical problem associated with high morbidity and mortality following a wide range of injuries, such as acute mesenteric ischemia secondary to embolism or thrombosis, trauma, sepsis, shock, or intestinal transplantation (1). The inflammatory cascade leading to IR injury results from an initial ischemic insult followed by a reperfusion insult of which neutrophils are a chief mediator (2). The initial ischemic injury causes cellular damage via ATP depletion and cellular acidosis. Reperfusion, followed by neutrophil influx and activation, with subsequent release of reactive oxygen species, then leads to a “second hit” resulting in increased tissue damage upon restoration of blood flow to the ischemic tissue (3). This leads to what is essentially a sterile inflammatory environment in the intestinal mucosa, at least initially, before subsequent breakdown in barrier function may lead to translocation of enteric bacteria or enterotoxins causing sepsis or multiorgan failure (3).

The concept of neutrophil heterogeneity has been gaining traction in recent years, with evidence supporting various potentially functionally distinct subsets of neutrophils. Two different subsets of neutrophils, N1 and N2, have been identified in models of myocardial ischemia (4). Other markers such as CD177, CD62L, and CXCR4 may have differential expression on subsets of neutrophils, but it is unclear whether these are different activation states of a single population of neutrophils or whether these represent unique neutrophil subtypes delineated during myeloid cell development (5–7).

Olfactomedin-4 (OLFM4) is a secreted glycoprotein belonging to a group of proteins identified by a conserved olfactomedin domain (8). OLFM4 is expressed in numerous human tissues, including breast, prostate, bone, and gastrointestinal epithelium, and changes in OLFM4 expression have been associated with malignancies in these tissues (9, 10). Clemmensen et al. (11) initially reported in the blood compartment, OLFM4 is expressed solely in neutrophils, and ∼25% of mature human neutrophils express OLFM4 within the specific granules. Several studies have subsequently shown that OLFM4 transcript is highly up-regulated during inflammatory states, such as sepsis, sepsis-induced acute respiratory distress syndrome, and respiratory syncytial virus (RSV) infection. Our group has recently shown that the OLFM4-positive subset of neutrophils is conserved in mice (12). Similar to humans, OLFM4 is expressed only in neutrophils and depending on the strain of mice, 6–30% of neutrophils express OLFM4, with C57Bl6 expressing in 6–8% of neutrophils. Although the exact function of OLFM4 remains elusive, previous studies have been consistent in that loss of OLFM4 is associated with a protective phenotype during inflammatory responses in various murine models (13–15). Given these findings, we investigated the role of OLFM4 in a noninfectious model and hypothesized that genetic ablation of OLFM4 would be protective in the setting of intestinal IR injury.

MATERIALS AND METHODS

Animals and murine model of intestinal IR injury

All experiments were performed using C57Bl/6 mice [wild type (WT)] obtained from Charles River Laboratories (Wilmington, MA, USA) for use directly in experiments or as in-house breeding pairs. OLFM4 null mice were generated using the clustered regularly interspaced short palindromic repeats/CRISPR associated protein 9 (CRISPR/Cas9) method to target the fourth exon as previously described (12, 16). All mice received standard housing, were fed a diet of standard rodent chow, and regulated day-night cycling. All experiments were performed on 12–15-wk-old male mice unless otherwise noted. Experimental protocols were approved by the Cincinnati Children’s Hospital Medical Center Institutional Animal Care and Use Committee and performed in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD, USA).

Mice were subjected to intestinal IR as previously described by Gubernatorova et al. (17). Mice were given unlimited access to chow and water until the night before surgery, when food was removed. Mice were anesthetized using 1–2% isoflurane and a midline laparotomy incision was made. The cecum was exteriorized, followed by evisceration of the small bowel from the cecum to the duodenal-jejunal junction. The superior mesenteric artery was dissected from the mesentery and was clamped along with its proximal branches using three 60-g pressure microvascular clips (World Precision Instruments, Sarasota, FL, USA). The small bowel and cecum were covered with saline-soaked gauze and ischemic time allowed for 30 min, resulting grossly in a deep red small bowel. The clips were removed, the viscera replaced into the abdomen, and the abdomen was closed. Mice received 1.0 ml of 0.9% sodium chloride subcutaneously for resuscitation and subcutaneous analgesia. Mice were recovered under a warming lamp and were either monitored for 7 d for survival or euthanized 4 h after reperfusion for tissue harvest. Control animals were directly euthanized for tissue harvest.

Evaluation of in vivo intestinal permeability

Female mice were used for intestinal permeability experiments due to availability, and they were allowed only water overnight prior to IR procedure. They underwent oral gavage with 20 ml/kg of 25 mg/ml 40 kDa FITC-dextran (MilliporeSigma, Burlington, MA, USA) in PBS. Thirty minutes after oral gavage, the mice underwent intestinal IR and were allowed access to only water during recovery. Blood was collected at 4 h after reperfusion. Plasma was plated in 1:4 dilutions in PBS, along with standard dilutions of FITC-dextran (half-log serial dilutions from 3 μg/ml to 10 ng/ml) in a 1:4 dilution of control mouse plasma in PBS. Fluorescence spectrophotometry was performed with an excitation wavelength of 485 nm and an emission wavelength of 525 nm. Fluorescence results were extrapolated to serum concentration of FITC-dextran using the standard curve.

Histologic evaluation of intestinal I/R injury

Harvested small bowel was flushed with 5 ml of 0.9% sodium chloride to clear luminal content, and a portion of proximal intestine was placed into formalin for fixation. Tissue then underwent standard paraffin embedding and sections were stained with hematoxylin/eosin for histopathologic analysis of IR injury. Intestinal sections were graded by 4 blinded observers using the Chiu scale for intestinal IR injury, with the mean of the 4 scores taken as the final result (18, 19). The Chiu scale grades IR-mediated damage from 0 to 5. A score of 0 is normal; 1 shows early evidence of submucosal edema at the villus tip with formation of Gruenhagen’s space; 2 demonstrates enlargement of Gruenhagen’s space at the villus tip; 3 is associated with spread of submucosal edema to the midpoint of the villus; 4 has spread of submucosal edema to the base of the villus; 5 is the worst, with sloughing of the mucosal architecture and exposure of the lamina propria.

Adoptive transfer experiments

To determine whether phenotypic effects were due to OLFM4 expressed in the intestinal epithelium compared with the neutrophils, mice underwent bone marrow transplantation to create chimeric animals. Six-week-old male mice underwent bone marrow transplantation in the Cincinnati Children’s Hospital Medical Center Animal Bone Marrow Transplantation Core. Mice received whole-body irradiation of 11.75 gray fractionated into 2 doses, followed by transplantation of 5 × 10⁶ bone marrow cells from the appropriate donor mouse. WT C57Bl/6 mice were transplanted with marrow from OLFM4 null mice, whereas OLFM4 null mice received bone marrow from WT C57Bl/6 mice. This yielded chimeric mice, whereby WT mice expressed OLFM4 in the intestinal epithelium but not in the neutrophils (gut⁺/neutrophil⁻), and OLFM4 null animals only expressed OLFM4 in the neutrophils after transplantation (gut⁻/neutrophil⁺). Following bone marrow transplantation at 6 wk, mice were allowed to recover to an age of at least 12 wk prior to subsequent procedures.

Immunohistochemistry

Deparaffinized slides underwent heat-induced epitope retrieval in citrate buffer (Thermo Fisher Scientific, Waltham, MA, USA) for 20 min. Slides were blocked with Bloxall blocking solution (Vector Laboratories, Burlingame, CA, USA), avidin/biotin blocks (BioCare Medical, Burlingame, CA, USA), and with 5% fetal bovine serum in PBS with 0.5% Tween-20. Primary antibody was added and incubated overnight at 4°C. OLFM4 was stained using 1:400 rabbit anti-mouse OLFM4 mAb (39141; Cell Signaling Technology, Danvers, MA, USA). Following overnight incubation, the secondary biotinylated goat anti-rabbit IgG (BA-1000; Vector Laboratories) was added at 1:1000 and incubated for 4 h at room temperature. Slides were incubated with streptavidin- horseradish peroxidase (HRP) (BD Biosciences, San Jose, CA, USA), then developed with DAB-Quanto (Thermo Fisher Scientific) and counterstained with hematoxylin.

Quantitative PCR of intestinal tissue cytokines and plasma cytokine measurement

At tissue harvest, 1-cm-length specimens from the proximal jejunum were placed into 1 ml of Trizol Reagent (Thermo Fisher Scientific) and stored at −20°C until use. The specimens were processed per the manufacturer’s provided protocol for RNA extraction. RNA was then prepared for RT-PCR amplification using the SuperScript IV System (Thermo Fisher Scientific). cDNA was stored at −20°C until further use for quantitative PCR. Quantitative PCR analysis of intestinal tissue cytokine expression was performed using the TaqMan Mouse Immune Response Array 96-well Fast Plate (4418856; Thermo Fisher Scientific) per the manufacturer’s provided protocol on the QuantStudio 6 Flex Platform (Thermo Fisher Scientific).

Whole blood was collected in heparinized tubes at the time of euthanizing. Plasma cytokines were measured using custom plates from MilliporeSigma using a Luminex 200 (Luminex, Austin, TX, USA).

Western blotting

Portions of proximal jejunum from IR and control mice were homogenized in lysis buffer (Cell Signaling Technology) with protease inhibitors and protein concentrations measured by bicinchoninic acid (BCA) Protein Assay Kit (Pierce, Rockford, IL, USA). Forty micrograms of protein per specimen were prepared in NuPage reducing agent and loaded and run on NuPage Bis-Tris Gel for 60 min (Thermo Fisher Scientific). Protein was then transferred under semidry conditions onto a nitrocellulose membrane and treated with SuperSignal Western Blot Enhancer (Pierce). The membrane was blocked with 5% milk in PBST and 10 μg of anti-iNOS IgG (MAB9502; R&D Systems, Minneapolis, MN, USA) for overnight incubation at 4°C. The membrane was incubated with HRP-conjugated goat anti-rabbit IgG (MilliporeSigma) for 1 h at room temperature and was developed with SuperSignal substrate (Pierce) prior to imaging. The membrane was then stripped and β-actin primary antibody added for loading control. Western blotting of OLFM4 from neutrophil pellets and cell culture medium followed the above protocol with rabbit anti-mouse OLFM4 mAb (39141; Cell Signaling Technology) for the primary antibody.

Assay for activation of NFκB

The relative levels of NFκB p50 activation were assayed in proximal intestinal tissues after IR using a commercially available ELISA-based colorimetric assay per the manufacturer’s provided protocol (TransAM 41096; Active Motif, Carlsbad, CA, USA). Whole tissue lysates were prepared from snap-frozen intestinal tissue, and 40 μg of protein was added to each well of a plate precoated with an NFκB consensus binding site oligonucleotide. Detection was via the provided anti-NFκB p50 primary and HRP-conjugated secondary antibodies. Developing and stopping solutions were added and the absorbance recorded at 450 nm. Levels of NFκB activation were normalized to WT control.

Neutrophil purification and stimulation

WT and OLFM4 null mice were euthanized, and neutrophils were harvested from the long bones by flushing with PBS. Neutrophils were then washed in PBS and separated via density gradient centrifugation by layering 1.077 density Histopaque on top of 1.119 Histopaque (MilliporeSigma). The neutrophils were quantified and resuspended in Rosewell Park Memorial Institute (RPMI) 1640 medium. WT and OLFM4 null neutrophils were then stimulated with 100 nM phorbol myristate acetate (PMA) for 2 h. Afterward, the neutrophils were pelleted and lysed in RIPA buffer. The residual culture medium was concentrated and added to RIPA buffer. To localize OLFM4 following neutrophil activation, Western blotting for OLFM4 was performed as previously described.

Stimulation of RAW cells with recombinant OLFM4

To evaluate whether secreted OLFM4 may have an effect on surrounding inflammatory cells, murine RAW264 cells were grown in tissue culture with DMEM with 10% fetal bovine serum and plated in 6-well plates at 1 × 10⁶ cells/well until reaching 50–60% confluency. They were then stimulated overnight with 10–100 ng/ml of recombinant murine OLFM4 (Creative Biomart, Shirley, NY, USA) at 37°C. Control treatments included medium alone or 1 μg/ml of LPS. After overnight stimulation, the cell pellets were lysed and protein quantitated by BCA assay. A protein sample (75 μg) was loaded onto a gel for Western blotting for iNOS as previously described.

Data analysis

Continuous variables were analyzed by a Student’s t test when data were normally distributed or Wilcoxon rank sum testing when data were non-normally distributed. Survival analysis was performed via the Kaplan-Meier method with the log-rank test. Statistical analyses were carried out using SigmaPlot v.14.0 (Systat Software, San Jose, CA, USA) and results were plotted using Prism 8 (GraphPad Software, La Jolla, CA, USA).

RESULTS

OLFM4 null mice are protected from IR injury

We first performed IR in groups of WT and OLFM4 null mice to assess for differences. After the IR procedure, mice were monitored daily for clinical deterioration and mortality for 7 d, after which any surviving animals were euthanized. At 7 d after reperfusion, OLFM4 null animals had a survival benefit, with a mortality of 33% (n = 7/21) compared with 68% (n = 13/19) in WT mice (P = 0.028) (Fig. 1A).

Figure 1.

OLFM4 null mice are protected from intestinal IR injury. A) Kaplan-Meier survival curve demonstrating mortality at 7 d following IR injury. OLFM4 null mice have a significant survival benefit following intestinal IR injury compared with WT mice undergoing IR (33 vs. 68%, P = 0.028). B) Assay of intestinal permeability by plasma levels of FITC-dextran at 4 h following reperfusion. WT mice have significantly higher plasma FITC-dextran concentration at 4 h after reperfusion compared with OLFM4 null mice (414 ± 414 vs. 162 ng/ml ± 147, P = 0.031; Wilcoxon rank sum).

Following these findings, we evaluated in vivo functional changes in intestinal permeability as a potential mechanism leading to the increased mortality in the WT mice, using FITC-dextran gavage prior to IR. Four hours after IR procedure, plasma samples were analyzed for FITC-dextran concentration. WT mice had higher plasma levels of FITC-dextran compared with OLFM4 null mice at 4 h after reperfusion (414 ng/ml ± 414 vs. 162 ng/ml ± 147, P = 0.031) (Fig. 1B) indicative of increased intestinal permeability in WT animals following IR.

We evaluated histologic differences following intestinal IR. Proximal intestinal sections harvested at 4 h after IR were stained with hematoxylin and eosin for histologic scoring of IR-mediated damage (Fig. 2A). Compared with WT mice, OLFM4 null mice demonstrated less histologic damage by the Chiu scale (n = 8 each group, mean injury score = 3.1 ± 1.5 vs. 1.3 ± 1.1, P = 0.019) (Fig. 2B). Notably, injury scores for OLFM4 null mice following IR were not different from either the WT or null control tissues at 4 h. At 24 h after reperfusion, there was no difference in mean injury scores after IR injury due to restitution of the mucosal layer (unpublished results). To evaluate if intestinal IR induced lung or liver injury, we also harvested lung and livers from animals 4 h after IR injury. Injury scores for the liver and lungs were not different between WT and OLFM4 null animals (Supplemental Fig. S1).

Figure 2.

OLFM4 null mice have less histologic evidence of injury. A) Representative light microscopy images of mouse intestinal tissue stained with hematoxylin and eosin. Healthy control tissue for WT (top left) and OLFM4 null (bottom left) are demonstrated. Following IR injury, WT mice have severe injury up to complete loss of mucosal architecture (top right), whereas OLFM4 null mice have minimal visible evidence of IR injury (bottom right). B) Mean injury scores of WT and OLFM4 null mice after injury as graded with the Chiu score by 4 blinded observers. WT mice had significantly higher mean injury scores compared with OLFM4 null mice (3.1 ± 1.5 vs. 1.3 ± 1.1, P = 0.019; Student’s t test). Furthermore, OLFM4 null mice had similar mean injury scores as healthy controls.

OLFM4 is present in increased amounts in injured intestinal tissues

Intestinal tissue sections underwent immunohistochemical staining for OLFM4 at 4 h after reperfusion. Healthy control WT sections demonstrated specific staining for OLFM4 in the bases of the intestinal crypts, consistent with previous reports (20). After IR injury, specific staining for OLFM4 was noted to expand further up the intestinal crypts and was also present intraluminally. OLFM4 null animals did not demonstrate any positive staining for OLFM4 at the crypts or intraluminally, confirming the specificity of the WT OLFM4 staining (Fig. 3A–C).

Figure 3.

Representative immunohistochemical sections stained for OLM4 in mouse intestine. A) In healthy control WT mice, OLFM4 is located in the bases of the crypts. B) Following IR injury in WT mice, OLFM4 expression appears to extend further up the crypts, and OLFM4 is found secreted into the lumen (double arrow). OLFM4-positive neutrophils are also noted infiltrating the base of the crypts after IR injury (single arrows). C) OLFM4 null mice after IR injury shown for comparison, where OLFM4 is not present. D) Western blotting of neutrophil cell pellets or culture medium following stimulation with 100 nM PMA. Unstimulated WT neutrophil pellets have high levels of OLFM4 compared with stimulated. Culture medium from stimulated WT neutrophils has OLFM4, whereas there is no OLFM4 in the medium from unstimulated neutrophils. There is no OLFM4 in the null pellets or medium.

We next determined whether neutrophils might be partially responsible for increased OLFM4 staining seen after IR injury. To test this, we stimulated neutrophils harvested from WT and OLFM4 null animals in vitro and assessed for OLFM4 secretion. After 2 h of stimulation with 100 nM PMA, neutrophils from both WT and OLFM4 null mice were pelleted and lysed, and supernatants were concentrated for Western blot analysis. In unstimulated conditions, Western blot of WT neutrophil pellets was heavily positive for OLFM4, whereas there was no positive staining for OLFM4 in the supernatant. After PMA stimulation, the WT neutrophil pellets stained less heavily for OLFM4, and OLFM4 was now detected in the cell culture medium. OLFM4 was not detected from OLFM4 null neutrophil pellets under both unstimulated and stimulated conditions, and cell culture medium control was also negative for OLFM4 (Fig. 3D). Although neutrophilic OLFM4 cannot account for all the specific OLFM4 staining, we can conclude that OLFM4-positive neutrophils will secrete OLFM4 into the environment in response to insults that may affect inflammatory responses of other tissues.

Adoptive transfer of OLFM4-positive neutrophils to OLFM4 null mice increases IR-mediated mortality

To evaluate whether OLFM4-positive neutrophils were the chief contributors to the increased IR injury, adoptive transfer experiments were performed to produce chimeric mice. WT mice were transplanted with bone marrow from OLFM4 null mice to produce mice with OLFM4-positive intestinal epithelium and OLFM4-negative neutrophils (gut⁺/neutrophil⁻). Meanwhile, OLFM4 null mice were transplanted with bone marrow from WT mice to yield animals with OLFM4-negative intestinal epithelium and OLFM4-positive neutrophils (gut⁻/neutrophil⁺). Successful engraftment was confirmed via flow cytometry, which demonstrated lack of OLFM4 + neutrophils in the gut⁺/neutrophil⁻ mice and the appropriate subset of OLFM4-positive neutrophils in the gut⁻/neutrophil+ mice (Fig. 4A, B). After appropriate recovery from adoptive transfer procedure, chimeric mice underwent IR injury procedure, and they were monitored for clinical deterioration or mortality until 7 d. OLFM4 null mice that received OLFM4-positive (WT) bone marrow had greater mortality compared with WT mice receiving OLFM4 null bone marrow [57% (n = 4/7) vs. 9% (n = 1/11), P = 0.027] (Fig. 4C).

Figure 4.

Confirmation of successful engraftment following bone marrow transplantation and survival analysis of OLFM4 chimeric mice. A) Immunohistochemical stain for OLFM4 shows expected presence of OLFM4 in the crypt bases of gut+/neutrophil- mice (top) following IR. Flow cytometry demonstrates lack of OLFM4-positive neutrophils in WT mouse after bone marrow transplantation of OLFM4 null marrow (bottom, cells obtained from bone marrow). B) Staining for OLFM4 shows presence of OLFM4-positive neutrophils infiltrating villus after IR injury (arrows) (top). Flow cytometric confirmation of engraftment with the OLFM4-positive neutrophil subset demonstrated in gut⁻/neutrophil⁺ chimeras (bottom, cells obtained from bone marrow). C) Survival analysis of chimeric mice showing decreased survival in gut-/neutrophil + mice after IR injury [57% (n = 4/7) vs. 9% (n = 1/11), P = 0.027].

WT mice demonstrate increased expression of iNOS following IR injury

In order to start to understand the mechanism for increased injury after IR in WT animals, we performed a broad PCR array for inflammation-related genes comparing WT and OLFM4 null animals. Quantitative PCR analysis of intestinal tissue after IR demonstrated differential expression of several inflammatory pathway genes of interest (Fig. 5A). Relative to hypoxanthine-guanine phosphoribosyltransferase (Hprt) expression, OLFM4 null mice had increased expression of Nfkb1 (relative units, 0.785 ± 0.050 vs. 0.540 ± 0.089, P = 0.014), whereas WT mice had increased expression of Ikb (0.127 ± 0.004 vs. 0.104 ± 0.007, P = 0.006). There was no difference in the expression of Nfkb2 between WT or OLFM4 null mice (0.634 ± 0.079 vs. 0.791 ± 0.114, P = 0.122). Actual NFκB activity was then assessed by transcription factor binding using the TransAM Assay, which demonstrated increased levels of NFκB activation in the proximal intestinal tissue of WT mice following IR injury (absorbance 450 nm, 0.122 ± 0.040 vs. 0.083 ± 0.054, P = 0.014) (Fig. 5B).

Figure 5.

Evaluation of inflammatory pathways of interest following intestinal IR injury. A) Quantitative PCR array demonstrated differential expression of Nfkb1, Ikb, and Nos2 genes after IR injury (*P = 0.014, **0.006; Student’s t test). B) There is increased activation of NFκB in WT mice following IR injury (P = 0.014; Student’s t test). C) Quantitative PCR demonstrates higher expression of Nos2 in WT compared with OLFM4 null mice after IR injury (P = 0.009; Student’s t test). D) Western blotting confirms increased levels of iNOS protein in the intestinal tissue of WT mice after IR injury (P = 0.05). E) Luminex assay showing murine plasma Mip-1a levels at baseline and after IR injury (P = 0.046; Student’s t test). F) Western blot of RAW cells after stimulation with 10–100 ng/ml recombinant OLFM4 shows production of iNOS from stimulated cells compared with no iNOS production from unstimulated cells.

Quantitative PCR analysis also showed that the most notable difference between WT and OLFM4 null animals was in the expression of Nos2, which was highest in the WT mice (0.0165 ± 0.00345 vs. 0.00546 ± 0.00223, P = 0.009) (Fig. 5C). Western blot analysis of proximal intestinal tissue after IR injury further confirmed increased levels of iNOS in WT mice compared with OLFM4 null mice (iNOS units relative to actin, 2.04 ± 1.46 vs. 0.26 ± 0.26, P = 0.049) (Fig. 5D).

We also tested for differences in plasma cytokine and chemokine expression before and after IR injury. We collected plasma 4 h post-IR to assess for differences in immune activation. Although our array included TNFa, IL-1b, IL-4, IL-6, IL-17A, IFNg, macrophage inflammatory protein 1α (MIP-1a), and matrix metallopeptidase 8 (MMP8), only MIP-1a was different between WT and OLFM4 null animals. At 4 h, MIP-1a was significantly lower in OLFM4 null animals (0.52 ± 0.44 vs. 1.66 ± 1.14 pg/ml, P = 0.049) (Fig. 5E).

OLFM4 stimulation induces iNOS production in RAW macrophages in vitro

After stimulation of murine RAW cells with 10–100 ng/ml recombinant murine OLFM4 or 1 μg/ml of LPS, cells were lysed and Western blotted for the presence of iNOS. Western blot demonstrated expression of iNOS in RAW cells stimulated with murine OLFM4 or LPS, whereas unstimulated RAW cells did not produce iNOS (Fig. 5F).

DISCUSSION

We set out to test the role of OLFM4-positive neutrophils in a model of intestinal IR, a form of sterile inflammation. Several studies have shown that the OLFM4 null mouse is protected from infection, but none have tested a sterile inflammatory model. Here we show that, like models of infection and sepsis, OLFM4 null animals exhibit a survival benefit at 7 d after IR injury. We demonstrate that at 4 h after IR injury, WT mice have increased IR-mediated damage, and from a functional standpoint, WT mice also have increased intestinal permeability after IR injury. The protective phenotype from OLFM4 null animals is lost after bone marrow transplantation of OLFM4-positive neutrophils, resulting in a similar mortality rate to nontransplanted WT mice. As a first step to understanding the mechanism for OLFM4 in inflammation, we have shown that WT mice have increased NFκB activity and expression of Nos2 following IR injury, and that soluble OLFM4 will stimulate murine RAW macrophages to produce iNOS.

In this model of intestinal IR, OLFM4 is expressed by both the intestinal epithelial cells and infiltrating neutrophils. The role for OLFM4 in intestinal epithelium is not fully understood because mice deficient in OLFM4 were found to have no phenotype by 1 group and to have an increased propensity for forming gastrointestinal tumors following dextran sodium sulfate colitis by another group (20, 21). However, this phenotype took weeks to develop. Here we exposed the OLFM4 null mouse to IR and found protection from intestinal damage, even at a short time interval of 4 h. This suggests that perhaps neutrophils rather than intestinal expression of OLFM4 may be responsible for the phenotype.

OLFM4-positive neutrophils may be functionally different from OLFM4-negative neutrophils. In adult patients with septic shock, progression to acute respiratory distress syndrome was associated with increased expression of neutrophil OLFM4 (22). Likewise, children with respiratory syncytial virus infection who had higher expression of OLFM4 from peripheral blood leukocytes were more likely to progress toward severe disease (23). Moreover, increased percentage of OLFM4-positive neutrophils and increased serum OLFM4 levels were associated with a higher likelihood of a complicated intensive care unit course and mortality in pediatric septic shock (24). Finally, we recently showed that this subset of neutrophils is evolutionarily conserved from the mouse to man (12). Whereas previous work by Welin et al. (25) did not demonstrate any functional differences between human OLFM4-positive and OLFM4-negative neutrophils in terms of phagocytosis and migration, they did note the presence of OLFM4 in neutrophil extracellular traps (NETs). However, definitive comparisons of OLFM4-positive and OLFM4-negative neutrophils have been hindered by the fact that identifying OLFM4-positive neutrophils currently requires fixation and permeabilization, making side-by-side comparisons of live cells difficult.

The mechanism of action for OLFM4 has been equally difficult to determine. Studies in mice have shown OLFM4 to potentially play a role in down-regulation of the innate immune response in murine Helicobacter pylori infection by inhibiting cathepsin G (13–15). Another study showed that OLFM4 may be involved in superoxide production to augment neutrophil apoptosis (26). However, these studies were carried out assuming that all neutrophils expressed OLFM4, whereas the finding that only a subset of murine neutrophils express OLFM4 was only recently published (12). Here we show that OLFM4-positive neutrophils can secrete OLFM4 into the environment with stimulation. Although yet to be directly proven, this OLFM4 may be important in stimulating downstream inflammatory responses.

NFκB is one of the chief transcription factors activated in response to tissue injury and the subsequent inflammatory insult (27). It has been shown to become highly activated in murine models of intestinal IR, with both local and systemic injury ameliorated by use of NFκB inhibitors (28, 29). NFκB is also known to be critical for the expression of murine iNOS (30). In the current work, NFκB activation is higher in the intestinal tissue of WT mice after IR compared with OLFM4 null mice, and this potentially drives increased expression of iNOS in the intestinal tissues leading to worse damage and breakdown in barrier function. We also noted increased expression of the chemokine Mip-1a [C-C motif chemokine ligand 3 (CCL3)] in WT mice relative to OLFM4 null mice. Given the many sources and stimulants for induction of Mip-1a, it is difficult to know if this directly related to OLFM4 secretion or related to the greater injury in WT mice.

iNOS is the type 2 isoform of the family of NO synthases and is activated in numerous cell types including endothelium, smooth muscle, and macrophages following inflammatory insult to produce large amounts of NO (30). Past studies show a role for iNOS in intestinal IR injury leading to increased mucosal damage and bacterial translocation (31). The presumed mechanism to this injury is increased levels of NO causing either direct tissue injury or decreased blood flow to the mesenteric vessels due to systemic effects (31, 32). In our study, we found that WT mice had increased expression of the gene Nos2 and increased intestinal tissue levels of iNOS protein, along with increased intestinal permeability after IR injury. Other studies have noted the roles of neutrophilic iNOS in the production of NO during inflammatory insults. One recent study by Ode et al. (33), noted that intercellular adhesion molecule 1 (ICAM1)⁺ neutrophils stimulated with cold-inducible RNA-binding protein (CIRP) had increased production of iNOS and NETs in a murine sepsis model. Previous work of our laboratory and others have demonstrated the presence of OLFM4 in neutrophilic NETs (12, 25).

As OLFM4-positive neutrophils have otherwise been shown to be functionally similar to OLFM4-negative neutrophils, we evaluated the potential effect that secreted OLFM4 may have on surrounding inflammatory cells, such as macrophages. After incubation with recombinant murine OLFM4, RAW cells, a murine macrophage cell line, showed increased production of iNOS similar to that with LPS stimulation. This experiment provides a potential explanation for the profound phenotypic differences among WT and OLFM4 null mice, accounting for the fact that OLFM4-positive neutrophils only make up ∼5–8% of neutrophils in the WT C57Bl/6 strain (12). However, there are likely other factors as we tested for differences in iNOS signal in our groups of mice following marrow transplantation and found iNOS expression in both groups (unpublished results). We believe these findings are confounded by intestinal injury during the whole-body irradiation the animals receive prior to transplantation, thus confounding the study. These results do suggest that other molecules beside iNOS likely have a role.

The major limitation of this work is that OLFM4 is produced by both intestinal tissue and the neutrophils, and it is difficult to ascertain whether intestinal or neutrophil-derived OLFM4 is playing the major role. In our adoptive transfer experiments, we were able to show that OLFM4 null mice that received OLFM4-positive neutrophils had decreased survival, similar to that of untransplanted WT mice. Furthermore, WT mice that were transplanted with OLFM4-null neutrophils had the best survival of any of our trials. We have also shown that upon activation, neutrophils secrete OLFM4 into the environment and that soluble OLFM4 will stimulate a murine macrophage cell line to produce iNOS. This supports the role of the neutrophil-derived OLFM4 as the major driver for the increased injury. However, because absolute chimerism is not guaranteed via adoptive transfer, the definitive studies will await the availability of a floxed OLFM4 mouse in which OLFM4 can be specifically ablated in neutrophils alone.

In summary, we show that WT mice have worse IR injury and mortality compared with OLFM4 null mice, and that this mortality phenotype can be transferred to OLFM4 null animals with bone marrow transplant of WT neutrophils. We outline a potential mechanism for this, whereby activated neutrophils at the site of injury secrete OLFM4 into the environment, which leads to increased iNOS production by either injured tissue or macrophages, with the ultimate consequence of decreased intestinal barrier function. The OLFM4-positive neutrophil is clinically relevant and may serve as a therapeutic target for a variety of inflammatory conditions not limited to IR injury and sepsis.

Glossary

HRP

horseradish peroxidase

IR

ischemia/reperfusion

MIP-1a

macrophage inflammatory protein 1α

NET

neutrophil extracellular trap

OLFM4

olfactomedin-4

PMA

phorbol myristate acetate

WT

wild type

AUTHOR CONTRIBUTIONS

N. C. Levinsky and M. N. Alder designed and performed the research, performed statistical analysis, and prepared the manuscript; J. Mallela, A. M. Opoka, K. Harmon, and H. V. Lewis performed the research and reviewed the manuscript; and B. Zingarelli and H. R. Wong designed the research and critically reviewed the manuscript.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.