Extracellular CIRP generates a pro-inflammatory Ly6G⁺CD11b^hi subset of low-density neutrophils in sepsis

Abstract

Extracellular cold-inducible RNA-binding protein (eCIRP) is a damage-associated molecular pattern. Neutrophils present in the mononuclear cell fraction of Ficoll gradient separation are called low-density neutrophils (LDNs). Here we report the novel role of eCIRP on LDN’s heterogeneity in sepsis. Sepsis was induced in male C57BL/6 WT and CIRP^−/− mice by cecal ligation and puncture (CLP). At 20 h after CLP, LDNs in the blood were isolated by Ficoll gradient separation, followed by staining the cells with anti-Ly6G, and anti-CD11b Abs and detection by flow cytometry. Sepsis or rmCIRP injection in mice resulted in significant increase in the frequency (%) and number of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDNs in the blood compared to sham or vehicle-treated mice. At 20 h of CLP, CIRP^−/− mice had significantly lower frequency and number of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDNs in the blood compared to WT mice. In sepsis mice or rmCIRP-injected mice, compared to Ly6G⁺CD11b^lo LDNs, the expression of CXCR4, ICAM-1, and iNOS and formation of ROS, and NETs in Ly6G⁺CD11b^hi LDNs in the blood were significantly increased. Treatment of WT bone marrow-derived neutrophils (BMDN) with rmCIRP increased Ly6G⁺CD11b^hi LDNs frequency, while treatment of TLR4^−/− BMDN with rmCIRP significantly decreased the frequency of Ly6G⁺CD11b^hi LDNs. BMDN’s stimulation with rmCIRP increased the expression of transcription factors in LDNs. eCIRP induces the formation of a pro-inflammatory phenotype Ly6G⁺CD11b^hi of LDN through TLR4. Targeting eCIRP may provide beneficial outcomes in sepsis by decreasing pro-inflammatory Ly6G⁺CD11b^hi LDNs.

Summary sentence:

eCIRP, released during sepsis induces the formation of a pro-inflammatory Ly6G⁺CD11b^hi subset of LDNs through binding to its receptor TLR4.

Graphical Abstract

Introduction

Sepsis is caused by body’s exaggerated immune response to an infection [1, 2]. Neutrophils are the first leukocytes to reach infected sites to fight against invading pathogen. Upon activation, neutrophils produce reactive oxygen species (ROS), myeloperoxidase (MPO), and neutrophil extracellular traps (NETs) to counteract pathogen [3–5]. Conversely, uncontrolled release of these molecules causes tissue injury [6, 7]. Recent studies revealed that neutrophils are a heterogeneous cell population, composed of subsets displaying distinct phenotypes and functionality [8, 9]. Pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs) play a critical role in neutrophil activation and heterogeneity [7, 9].

Neutrophils expressing intercellular adhesion molecule-1 (ICAM-1) produce excessive amount of NETs, causing inflammation and tissue injury in sepsis [7, 10, 11]. Neutrophils expressing CXCR4, a receptor allowing their clearance in the bone marrow are known as aged neutrophils [12, 13]. Neutrophil ageing is driven by the microbiota via toll like receptor 4 (TLR4) [13]. Aged neutrophils produce higher levels of NETs under inflammatory conditions [13]. Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature and mature cells that are found in cancer, autoimmune disease, and sepsis [14–17]. MDSCs are two types, granulocytic (G) and monocytic (M). G-MDSCs are phenotypically and morphologically similar to neutrophils, whereas M-MDSCs are similar to monocytes [14]. MDSCs were characterized by their ability to inhibit T cell functions [14].

Under the normal condition, circulatory neutrophils are mature and mostly sediment at the bottom of the gradient centrifugation; hence they are called normal-density or high-density neutrophils (HDNs) [18]. In inflammation, however, a fraction of neutrophils found in the mononuclear cell fraction of Ficoll gradient separation are called low-density neutrophils (LDNs) [18, 19]. Sagiv et al showed that both HDNs and LDNs express Ly6G⁺CD11b⁺ phenotype [20]. Ly6G⁺CD11b^hi HDNs and Ly6G⁺CD11b^hi LDNs are mature cells, which contain multi-lobed nucleus, while the size of Ly6G⁺CD11b^hi LDNs is larger than the size of Ly6G⁺CD11b^hi HDNs [20]. LDNs have been reported to be present in the blood of human patients suffering from systemic and local infection, autoimmune disease, and cancer [20–22]. Studies suggest that LDNs display an enhanced pro-inflammatory profile with an increased production of pro-inflammatory cytokines and NETs [21, 23]. Circulating LDN frequencies are correlated with disease state and severity in humans [20]. In sepsis, LDNs are expanded and have been shown to exhibit MDSC-like functions [24]. Studies show that LDNs are heterogeneous population of granulocytes containing immature and mature neutrophils as well as MDSCs [25]. Although LDNs have been better characterized in cancer and autoimmune diseases, where they play diverse roles [20, 21], considering LDN’s heterogeneity further research is needed to fully understand their role in inflammatory diseases.

Cold-inducible RNA-binding protein (CIRP) is an intracellular RNA chaperon [26], which serves as a DAMP when released outside the cells during inflammatory conditions [27, 28]. Extracellular CIRP (eCIRP) induces the production of pro-inflammatory cytokines and chemokines from macrophages, activates T cells to become Th1 cells [29], and neutrophils to generate NET-forming ICAM-1⁺ neutrophils by its receptor TLR4 [11]. eCIRP also activates immune cells via triggering receptor expressed on myeloid cells-1 (TREM-1) [30]. eCIRP has also been shown to promote macrophage endotoxin tolerance by binding to IL-6R ([31]). eCIRP’s role in LDN in terms of altering their phenotype and function in sepsis have not been studied.

In this report, we aimed to study the effects of eCIRP on the phenotypic and functional alterations in LDN in sepsis. Here, we found two distinct LY6G⁺ LDN subsets, CD11b^hi and CD11b^lo were both elevated in sepsis and their elevation was directly associated with eCIRP’s elevation as CIRP^−/− mice showed reduced levels of the LDN. We further detected that Ly6G⁺CD11b^hi LDN produced significantly higher levels of ICAM-1, CXCR4, ROS, iNOS and NETs compared to Ly6G⁺CD11b^lo LDN in sepsis. Thus, identification of the eCIRP-induced Ly6G⁺CD11b^hi LDN as a novel pro-inflammatory LDN subtype may direct a new pathophysiological feature and therapeutic strategy in sepsis.

Materials and Methods

Experimental animals

C57BL/6 male mice were purchased from Charles River Laboratories (Wilmington, MA). CIRP^−/− mice were initially obtained from Dr. Fujita (Kyoto University, Kyoto, Japan). TLR4^−/− mice were obtained from Dr. Tracey (Feinstein Institutes for Medical Research, Manhasset, NY). Mice were housed in a temperature-controlled room with 12 h light/dark cycles and provided laboratory chow and drinking water daily. Age (8–12 weeks) matched healthy mice were used in all experiments. Female sex steroids exhibit diverse immune-modulating functions in both humoral and cell-mediated immune responses under normal conditions and various disease processes including sepsis [32, 33]. Given the impact of sex on sepsis pathogenesis, to generate consistent findings, only male mice were used. All experiments were performed in accordance with the guidelines for the use of experimental animals by National Institutes of Health (Bethesda, MD). All the experiments involving mice were approved by the Institutional Animal Care and Use Committees (IACUC) of the Feinstein Institutes for Medical Research.

Isolation of LDN from blood

Mice were anesthetized with 2% isoflurane inhalation. A heparin coated 23-gauge needle was penetrated into the heart from the right side of the sternum and collected 1 ml of blood. Collected blood was transferred into a conical 15 ml tube and adjusted the volume to 6 ml with 0.5% BSA (Fisher BioReagents, Pittsburgh, PA) in PBS. In this study, we used Histopaque reagent (Sigma-Aldrich, St. Louis, MO), which is same as Ficoll reagent. Histopaque gradient was prepared in a 15 ml conical tube by pipetting 3 ml of Histopaque-1119 first and overlaid with 3 ml of Histopaque-1077 carefully. Histopaque-1119 and Histopaque-1077 indicate Histopaque concentrations in g/ml as prepared and sold by the vendor (Sigma-Aldrich, St. Louis, MO). The blood-BSA solution was placed on the top of the Histopaque gradient without any mixture. Histopaque gradient with blood-BSA was centrifuged at 700 ×g for 30 min at room temperature with no brake. After centrifugation, low-density cells’ layer was seen around 6 ml mark of a 15 ml conical tube as a foggy ring. Low-density cells were collected from between 5 ml above the ring of low-density cells and 5 ml below the ring and transferred to 30 ml PBS in 0.5% BSA in a 50 ml tube to wash the cells. Collected cells were spun down at 400×g for 10 minutes at room temperature and aspirated supernatant and resuspended the pellet in RPMI 1640 medium (ThermoFisher Scientific, Waltham, MA) or fluorescence activated cell sorting (FACS) buffer for subsequent experiments.

Mouse model of sepsis

WT and CIRP^−/− mice were induced polymicrobial sepsis by cecal ligation and puncture (CLP) [30, 34, 35]. Mice were anesthetized with 2% isoflurane inhalation. The abdomen was shaved and made sterile with 10% povidone-iodine wash. After a 1 cm midline abdominal incision the cecum was exposed and ligated at 1 cm from the tip with 4–0 silk suture. Ligated distal part of the cecum was punctured once through and through using a 22-gauge needle. A small amount of cecal content was extruded softly and the ligated cecum was returned to the peritoneal cavity. The wound was closed in layers. CLP-operated animals were resuscitated with 0.5 ml of normal saline subcutaneously to avoid stress induced dehydration, and then returned to their cages with normal access to foods and drinking water. The inflammatory response in sepsis depends on bacterial load. Since the use of antibiotics reduce bacterial contents, and therefore inflammation, antibiotics were not used in the CLP-operated animals in order to allow for a rapid, robust, inflammatory response. With this model of sepsis, mice became severely sick as demonstrated by our previous study showing the increased levels of pro-inflammatory cytokines, chemokines, and organ injury markers in the blood and the increased expression of pro-inflammatory cytokines, chemokines, and myeloperoxidase in the lungs at 20 h of CLP [30]. Hence, the CLP model that we used in the current study was considered to be a severe sepsis model.

Recombinant murine CIRP injection in mice

Recombinant murine CIRP (rmCIRP) was prepared in-house and the quality control assays were performed as described previously [27]. WT mice were anesthetized with 2% isoflurane inhalation, and 5 mg/kg BW of rmCIRP or the same amount of vehicle (normal saline) was injected intravenously (i.v.) through the jugular vein. The single dose of rmCIRP (5 mg/kg BW) administration in mice through i.v. was chosen from our previous studies, which showed that treatment of mice with this dose of rmCIRP was sufficient to induce systemic inflammation and acute lung injury in mice [30, 36]. After 5 h, under 2% isoflurane inhalation anesthesia, blood was collected via cardiac puncture for analysis.

Isolation of LDN from BMDN

Mice were euthanized and the femurs and the tibias were dissected. Bone marrow was flushed out from femurs and tibias with RPMI 1640 medium using a 25-gauge needle. Cell suspensions were filtered through 70 μm cell strainer (Corning, NY). BMDN were enriched by negative selection using EasySep mouse neutrophil isolation kit (Catalog no.: 19762; STEMCELL, Vancouver, BC, Canada). The numbers of isolated BMDNs were counted using a microscope (Eclipse TS100; Nikon, Tokyo, Japan). The purity of the sorted neutrophils was assessed by labeling the cells with anti-mouse Ly6G Ab (clone 1A8; Biolegend, San Diego, CA) using BD LSRII flow cytometer (BD Biosciences, San Jose, CA). BMDN suspension was divided into 5 ml tube as each tube containing 2×10⁶ cells/ml with RPMI 1640 medium. PBS or rmCIRP (1 μg/ml) was added to BMDN-RPMI 1640 and incubate 4 h at 37 °C in 5% CO₂ humidified incubator. After PBS or rmCIRP stimulation, BMDN was washed and resuspended with 1 ml of 0.5% BSA in PBS, transferred into a 15 ml conical tube and adjusted the volume to 6 ml with 0.5% BSA in PBS. The Histopaque gradient was prepared in a 15 ml conical tube as described above. BMDN with 0.5% BSA in PBS was placed on the top of the Histopaque gradient without any mixture and then centrifuged at 700 ×g for 30 minutes at room temperature with no brake. After centrifugation, low-density cells were collected from between 5 mm above the ring of low-density cells and 5 mm below the ring and transferred to a 50 ml tube with 30 ml 0.5% BSA in PBS to wash the cells. Collected cells were spun down at 400×g for 10 min at room temperature and aspirated supernatant and resuspended the pellet with respective solution for the subsequent experiment.

Assessment of frequency and number of LDN in sepsis

Sepsis was induced in C57BL/6 male WT and CIRP^−/− mice by CLP. LDN of WT and CIRP^−/− sham and CLP mice were isolated from 1 ml blood. LDN were stained with APC anti-mouse Ly-6G Ab (clone 1A8; Biolegend) and PerCP/Cyanine5.5 anti-mouse/human CD11b antibody (clone: M1/70; Biolegend) and then accessed by flow cytometry. The population of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were assessed with frequency and real number. The real number was calculated by using Precision Count Beads™ (Catalog no.: 424902; BioLegend). More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo software (Tree Star, Ashland, OR).

Assessment of LDN in rmCIRP-treated mice

C57BL/6 male WT mice were i.v. injected through the jugular vein with rmCIRP (5 mg/kg) and after 5 h, LDN in blood were assessed. Process of isolating LDN and assessing frequency and real number of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were described above. More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo software (Tree Star).

Stimulation of BMDN of WT and TLR4^−/− mice with rmCIRP and assessment of LDN

A number of 2×10⁶ BMDN isolated from WT and TLR4^−/− mice in RPMI 1640 medium were stimulated with PBS or rmCIRP (1 μg/ml). After 4 h 37 °C incubation, the cells were washed and resuspended with 1 ml of 0.5% BSA in PBS. LDN was isolated using the Histopaque gradient as described above. LDN were stained with APC anti-mouse Ly-6G Ab (clone 1A8; Biolegend) and PerCP/Cyanine5.5 anti-mouse/human CD11b antibody (clone: M1/70; Biolegend), and accessed by flow cytometry. The frequency of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were assessed as described above. More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo software (Tree Star).

Assessment of CXCR4 and ICAM-1 expression by flow cytometry

LDN isolated from the blood of sham, CLP, and rmCIRP-treated (i.v.) WT mice were evaluated for CXCR4 and ICAM-1 expression. After isolating LDN via Histopaque gradient, cells were stained with APC anti-mouse Ly-6G Ab (clone 1A8; Biolegend), PerCP/Cyanine5.5 anti-mouse/human CD11b antibody (clone: M1/70; Biolegend), Brilliant Violet 421™ anti-mouse CD184 (CXCR4) antibody (Clone: L276F12; Biolegend), and PE anti-ICAM-1 Ab (Clone: 3E2, BD Biosciences). After gating Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo, the frequency of CXCR4 and ICAM-1 positive population in each group were assessed by flow cytometry. More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo software (Tree Star).

Assessment of iNOS expression by flow cytometry

After isolating LDN from the blood of sham and CLP mice, cells were stained with APC anti-mouse Ly-6G Ab (clone 1A8; Biolegend) and PerCP/Cyanine5.5 anti-mouse/human CD11b antibody (clone: M1/70; Biolegend). Cells were then fixed and permeabilized using 2% paraformaldehyde and 0.1% tween20, followed by the intracellular staining with PE-anti-mouse iNOS Ab (Cat. No.: sc-651; Santa Cruz Biotechnology, Santa Cruz, CA). After gating Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo, the frequency of iNOS positive population was assessed by flow cytometry. More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo (Tree Star).

Assessment of ROS expression by flow cytometry

LDN from the blood of WT mice subjected to sham and CLP surgery were evaluated for oxidative stress. After isolating LDN, cells were stained with APC anti-mouse Ly-6G Ab (clone 1A8; Biolegend), PerCP/Cyanine5.5 anti-mouse/human CD11b Antibody (clone: M1/70; Biolegend) and CellROX™ Green Reagent, for oxidative stress detection (Catalog no: C10444; ThermoFisher Scientific, Waltham, MA). After gating Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo, the frequency of reactive oxygen species (ROS) positive population was assessed by flow cytometry. More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo (Tree Star).

Assessment of NETs by flow cytometry

LDN isolated from the blood of CLP mice were fixed with 2% paraformaldehyde for 20 min at room temperature, blocked for 30 min with 2% bovine serum albumin in PBS at 37° C. Without a permeabilization step, the cells were then stained with APC anti-mouse Ly-6G Ab (clone 1A8; Biolegend), PerCP/Cyanine5.5 anti-mouse/human CD11b antibody (clone: M1/70; Biolegend), FITC anti-mouse MPO Ab (clone: mAb 8F4; Hycult Biotech, Uden, The Netherlands) and rabbit anti-histone H3 (CitH3) Ab (Catalog no: 5103: Abcam) followed by staining with PE-donkey anti-rabbit IgG (Clone: Poly4064; Biolegend). After gating Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo, the frequency of both MPO and CitH3 positive population was evaluated as NETs by flow cytometry. More than 30,000 events were acquired using a BD LSR Fortessa Flow Cytometry Analyzer (BD Biosciences) and the data were analyzed by FlowJo software (Tree Star).

RT2 profiler PCR array experiment

A number of 2×10⁶ BMDN isolated from WT mice were stimulated with PBS or rmCIRP (1 μg/ml). After 4 h incubation, the cells were washed and resuspended with 1 ml of 0.5% BSA in PBS. LDN was isolated using the Histopaque gradient as previous described. Total RNA was extracted from isolated LDN using GE Healthcare illustra™ RNAspin Mini Isolation Kit (GE Healthcare Bioscience, Piscataway, NJ). cDNA was synthesized using RT2 First strand Kit (QIAGEN, Hilden, Germany). PCR components mix were prepared with RT2 SYBR Green Master mix, synthesized cDNA and RNase free water (QIAGEN). PCR components mix were dispensed into the RT2 profiler PCR array mouse transcription factors (QIAGEN, Hilden, Germany). Amplification and analysis were conducted in a Step One Plus real-time PCR machine (Applied Biosystems, Waltham, MA). Relative gene expression levels were calculated using the ddCT method. Relative expression of mRNA was expressed as fold change in comparison with PBS treated group.

Statistical analysis

Data represented in the figures are expressed as mean ± SE. Multiple groups were compared by one-way ANOVA using the Student-Newman-Keuls (SNK) test. The Student’s t test was applied for two-group comparisons. P value of less than 0.05 was considered to be statistically significant. All statistical analyses were carried out using GraphPad Prism graphing and statistical software (GraphPad Software).

Results

Sepsis increases Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood.

At various time point after sepsis, blood was collected from mice and subjected to Ficoll density gradient centrifugation to harvest the mononuclear cell fraction. These cells were stained with anti-Ly6G and -CD11b Abs and detected by flow cytometry. In sham mice, the frequencies of Ly6G⁺CD11b^hi and Ly6G⁺CD11^lo LDN were found to be 0.18% and 2.1%, respectively, and their numbers were 0.12 × 10⁵/ml, and 1 × 10⁵/ml, respectively (Fig 1A-E). We found that sepsis significantly increased the frequencies of both Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood in a time-dependent manner, where the highest increase in their frequency by a mean value of 8.7%, and 27.8%, respectively occurred at 20 h of CLP (Fig 1A-C). Compared to sham mice, the numbers of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood were also increased at highest levels of 1.5 × 10⁵/ml, and 5 × 10⁵/ml, respectively, at 20 h of CLP (Fig 1A, D, E).

Figure 1. Sepsis increases Ly6G⁺CD11b^hi LDN in the blood.

At various time points (5, 10, and 20 h) after sepsis induced by CLP, 1 ml blood was collected from mice and subjected to Ficoll density gradient centrifugation to harvest the mononuclear cell fraction including LDN followed by the staining of these cells with anti-Ly6G and -CD11b Ab and detection by flow cytometry. The population of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were assessed with frequency and number. (A) Representative dot blots indicating forward and side scatter, Ly6G and CD11b frequencies from sham and CLP mice at various time point are shown. (B-E) Bar diagram representing the mean frequencies and real number of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo populations are shown. The real number was calculated by using Precision Count Beads™. Experiments were repeated three times using 3 samples/group each time. The figures represent the results of all experimental iterations combined together. Data are expressed as mean ± SE (n=9 mice/group) and compared using one-way ANOVA (*p < 0.05 vs. sham mice).

CIRP^−/− mice have less blood Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in sepsis

Blood eCIRP is increased during sepsis [34]. We determined the optimum increase in the frequency and number of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood at 20 h of CLP (Fig 1). Therefore, to identify the relationship between eCIRP and LDN in sepsis, we induced sepsis in WT and CIRP^−/− mice and assessed Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood. As shown in Fig 2A-E, after 20 h of sepsis both WT and CIRP^−/− mice showed an increase in the frequency and number of Ly6G⁺CD11b^hi and CD11b^loLDN in the blood compared to sham mice. By contrast, the frequency of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood of CIRP^−/− sepsis mice were significantly decreased by the relative percentages of 77.5%, and 71.2%, respectively, and the number by 72.6%, and 58.6%, respectively, compared to WT sepsis mice (Fig 2A-E). These data clearly determine the link between eCIRP and Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood during sepsis.

Figure 2. CIRP^−/− mice have decreased contents of blood Ly6G⁺CD11b^hi LDN.

At 20 h of CLP or sham operation in mice, 1 ml blood were collected from WT and CIRP^−/− mice. After harvesting the mononuclear cell fraction including LDN by Ficoll density gradient centrifugation followed by the staining of these cells with anti-Ly6G and -CD11b Abs, the population of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were assessed with frequency and real number by flow cytometry. (A) Representative dot blots indicating forward and side scatter, Ly6G and CD11b frequencies from WT and CIRP^−/− mice of sham and CLP mice are shown. (B-E) Diagrammatic presentation of the quantitative mean values of the (B) frequencies and (D) numbers of Ly6G⁺CD11b^hi populations in blood from WT and CIRP^−/− mice are shown. (C-E) Diagrammatic presentation of the quantitative mean values of the (C) frequencies and (E) numbers of Ly6G⁺CD11b^lo populations in blood from WT and CIRP^−/− mice are shown. Experiments were repeated three times using 3 samples/group each time. The figures represent the results of all experimental iterations combined together. Data are expressed as mean ± SE (n = 9 mice/group) and compared using one-way ANOVA (*p < 0.05 vs. WT or CIRP^−/− sham; #p < 0.05 vs. WT CLP).

Treatment of WT mice with rmCIRP increases Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood

We injected mice with rmCIRP and after 5 h of injection blood was collected and subjected to Ficoll density gradient centrifugation to harvest the mononuclear cell fraction. We found that the mice injected with rmCIRP significantly increased the frequency of both Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood of 3.7%, and 6.8%, respectively (Fig 3A, B, D), and their numbers of 0.22 × 10⁵/ml, and 0.4 × 10⁵/ml, respectively, compared to PBS-injected mice (Fig 3A, C, E). Here, we noticed that the numbers of blood Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDNs in rmCIRP injected mice were remarkably lower than the CLP mice (Fig 1D, E), which we think could be due to less severe inflammation in rmCIRP injected mice (sterile model) than CLP mice (non-sterile model). These findings indicate that eCIRP directly induces the increase in the frequency and number of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in the blood.

Figure 3. Treatment of WT mice with rmCIRP increases Ly6G⁺CD11b^hi LDN in the blood.

Wild type mice were subjected to intravenous injection (i.v.) of rmCIRP (5 mg/kg) or PBS as vehicle. After 5 h, 1 ml blood was collected from rmCIRP-treated or PBS-treated mice. After harvesting the mononuclear cell fraction including LDN by Ficoll density gradient centrifugation from 1 ml blood followed by the staining of these cells with anti-Ly6G and -CD11b Abs, the population of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were assessed with frequency and real number by flow cytometry. (A) Representative dot blots indicating forward and side scatter, Ly6G and CD11b frequencies from PBS-treated and rmCIRP-treated mice are shown. (B-C) Diagrammatic presentation of the quantitative mean values of the (B) frequencies and (C) numbers of Ly6G⁺CD11b^hi LDN populations in blood from PBS-treated and rmCIRP-treated mice are shown. (D-E) Diagrammatic presentation of the quantitative mean values of the (D) frequencies and (E) numbers of Ly6G⁺CD11b^lo LDN populations in blood from PBS-treated and rmCIRP-treated mice are shown. Experiments were repeated two times using 2–3 samples/group each time. The figures represent the results of all experimental iterations combined together. Data are expressed as mean ± SE (n = 5 mice/group) and compared using Student’s t-test (*p < 0.05 vs. PBS-treated mice).

Ly6G⁺CD11b^hi LDN express aged and inflammatory surface markers in sepsis

Neutrophils expressing CXCR4 are aged neutrophils, which exhibit pro-inflammatory activity [37]. The ICAM-1⁺ neutrophils have also been shown to play pro-inflammatory role in sepsis. We induced sepsis in mice or injected rmCIRP in mice and assessed CXCR4 and ICAM-1 expression in Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo LDN in blood. We found that the Ly6G⁺CD11b^hi LDN in the blood of sepsis mice showed significantly increased frequency of CXCR4 and ICAM-1 expressing neutrophils compared to Ly6G⁺CD11b^lo LDN (Fig 4A, C, D). Similarly, in rmCIRP injected mice we also found that the Ly6G⁺CD11b^hi LDN in the blood had significantly increased frequencies of CXCR4 and ICAM-1 expressing neutrophils compared to Ly6G⁺CD11b^lo LDN (Fig 4B, E, F). These data clearly suggest that the Ly6G⁺CD11b^hi LDN are comparatively more aged and pro-inflammatory than the Ly6G⁺CD11b^lo LDN.

Figure 4. Ly6G⁺CD11b^hi LDN express aged and inflammatory surface markers in sepsis.

Surface expression of CXCR4 and ICAM-1 were assessed in CD11b^hi and CD11b^lo LDN in blood from WT CLP and rmCIRP-treated mice. (A-B) Representative dot blots indicating forward and side scatter, Ly6G, CD11b, CXCR4 and ICAM-1 expression in WT CLP and WT rmCIRP-treated mice are shown. (C-D) Bar diagram representing the mean frequencies of CXCR4 and ICAM-1 expressing cells within Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo populations in blood from WT CLP mice are shown. Experiments were repeated three times using 3–4 samples/group each time. The figures represent the results of all experimental iterations combined together. Data are expressed as mean ± SE (n = 10 mice/group) and compared using Student’s t-test (*p < 0.05 vs. CD11b^lo ). (E-F) Bar diagram representing the mean frequencies of CXCR4 and ICAM-1 expressing cells within Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo populations in blood from WT rmCIRP-treated mice are shown. Experiments were repeated two times using 2–3 samples/group each time. The figures represent the results of all experimental iterations combined together. Data are expressed as mean ± SE (n = 5 mice/group) and compared using Student’s t-test (*p < 0.05 vs. CD11b^lo).

Ly6G⁺CD11b^hi LDN produce increased levels of iNOS, ROS, and NETs

We next determined the synthesis of pro-inflammatory effector molecules such as iNOS, ROS, and NETs in Ly6G⁺CD11b^hi LDN and compared with Ly6G⁺CD11b^lo LDN isolated from sepsis mice. We found that Ly6G⁺CD11b^hi LDN in the blood of sepsis mice had significantly increased frequency of iNOS, ROS, and NET-forming neutrophils compared to Ly6G⁺CD11b^lo LDN (Fig 5A-F). These data clearly suggest that the Ly6G⁺CD11b^hi LDN produce more pro-inflammatory mediators than Ly6G⁺CD11b^lo LDN.

Figure 5. Ly6G⁺CD11b^hi LDN produce increased levels of iNOS, ROS, and NETs.

Production of iNOS, ROS, and NETs in CD11b^hi LDN of sepsis mice blood were assessed by flow cytometry. (A-C) Representative dot blots indicating forward and side scatter, Ly6G, CD11b, iNOS, ROS, and NETs in Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo cells are shown. Nets was evaluated as MPO and CitH3 double positive. (D-F) Bar diagram representing the mean frequencies of iNOS, ROS and NETs expressing cells within Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo populations in blood from WT CLP mice are shown. Experiments were repeated 2–3 times. The figures represent the results of all experimental iterations combined together. Data are expressed as means ± SE (n = 5–10 mice/group) and compared by using Student’s t-test (*p < 0.05 vs. CD11b^lo).

eCIRP increases Ly6G⁺CD11b^hi LDN through TLR4

We have previously shown that eCIRP induces its pro-inflammatory function by binding to TLR4/MD2 complex in the macrophages [27]. We therefore assessed the frequency of Ly6G⁺CD11b^hi PMN following in vitro treatment of the BMDN isolated from WT and TLR4^−/− mice with rmCIRP. We found that the WT BMDN treated with rmCIRP significantly increased the frequency of Ly6G⁺CD11b^hi LDN as compared to PBS-treated control. However, when TLR4^−/− BMDN were treated with rmCIRP the frequency of Ly6G⁺CD11b^hi LDN was significantly lower compared to WT BMDN by relative decrease of 62.3% (Fig 6A, B). We also assessed the expression of transcription factors downstream of TLR4 signaling that are implicated in inflammation. To do this, we treated BMDN with PBS or rmCIRP and after 4 h we isolated LDN from each group and assessed the expression of transcription factors using PCR array. We found that BMDN stimulated with rmCIRP increased the expression of several transcription factors implicated in inflammation i.e., Ets2, Irf1, Junb, Jund, Nfkb1, Rel, Rela, STAT1, and STAT2 by >2 fold compared to PBS-treated BMDN (Fig 6C). These data suggest that eCIRP increases Ly6G⁺CD11b^hi LDN through TLR4.

Figure 6. eCIRP increases Ly6G⁺CD11b^hi LDN through TLR4.

BMDN (2 × 10⁶) isolated from WT and TLR4^−/− mice were treated with rmCIRP (1 μg/ml) or PBS for 4h. After stimulation, LDN was harvested by Ficoll density gradient centrifugation followed by the staining of these cells with anti-Ly6G and -CD11b Abs. The frequency of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo were assessed by flow cytometry. (A) Representative dot blots indicating forward and side scatter, Ly6G and CD11b frequencies within Ly6G⁺ populations from WT mice BMDN treated PBS and rmCIRP and TLR4^−/− mice of BMDN treated with PBS and rmCIRP are shown. (B) Diagrammatic presentation of the quantitative mean values of the frequencies of Ly6G⁺CD11b^hi and Ly6G⁺CD11b^lo populations in WT and TLR4^−/− mice BMDN followed by stimulation with PBS and rmCIRP are shown. Experiments were repeated 2 times. The figures represent the results of all experimental iterations combined together. Data are expressed as means ± SE (n = 3–6 mice/group) and compared using Student’s t-test (*p < 0.05 vs. WT BMDN treated with rmCIRP). (C) Expression of transcription factors downstream of TLR4 signaling. BMDN obtained from WT mice were stimulated with PBS or rmCIRP. After 4 h incubation, LDNs were isolated using the Histopaque gradient. Total RNA was extracted from isolated LDNs, cDNA was synthesized and PCR array of mouse transcription factors were performed. Relative expression of mRNA was expressed as fold change in comparison with PBS treated group. Data are expressed as mean ± SE (experiment was performed 3 times using n=1 sample/group in each time).

Discussion

LDNs are heterogeneous cell population containing immature and mature neutrophils, and MDSCs [18, 22]. Due to their heterogeneity, LDNs exhibit diverse pathophysiological functions in various diseases, for example in cancer they play immunosuppressive role [20], while in lupus they cause organ damage through their heightened pro-inflammatory response [19, 21]. However, the function of LDN in these diseases cannot be matched with their role in sepsis as the immune phenomenon of sepsis varies from an early hyperdynamic phase characterized by an overwhelming inflammatory response to a late hypodynamic phase resulting in immunosuppression [1, 38]. Thus, the role of these neutrophils in sepsis remains elusive. In this study, we identified two distinct subtypes of Ly6G⁺ LDNs: CD11b^hi and CD11b^lo Ly6G⁺ LDNs in the blood of 20 h CLP-induced sepsis mice, a time point of CLP where mice show an exaggerated inflammatory response [30]. Interestingly, we found that eCIRP, a DAMP which is released during sepsis is involved in the generation of CD11b^hi and CD11b^lo Ly6G⁺ LDNs. We identified that eCIRP induces the formation of CD11b^hi LDNs through binding to its receptor TLR4 and by increasing the expression of downstream transcription factors STATs, Jun, Rel, and Nfkb which are directly linked with inflammation. We further characterized them in terms of having CXCR4 that reflects their aging and ICAM-1 indicating pro-inflammatory status and producing pro-inflammatory mediators such as iNOS, ROS, and NETs. We found that LY6G⁺CD11b^hi LDNs expressed increased levels of CXCR4, ICAM-1, iNOS, ROS, and NETs compared to LY6G⁺CD11b^lo LDNs, as summarized in Fig 7. Identification of a novel eCIRP-induced pro-inflammatory LY6G⁺CD11b^hi subset of LDN reveals a previously unknown pathophysiology of sepsis.

Figure 7. Summary of findings.

eCIRP, a DAMP which is released during sepsis induces the formation of Ly6G⁺CD11b^hi LDNs through binding to its receptor TLR4 and by increasing the expression of downstream transcription factors STATs, Jun, Rel, and Nfkb which are directly linked with TLR4 pathway. Ly6G⁺CD11b^hi LDNs express increased levels of CXCR4, ICAM-1, iNOS, ROS, and NETs compared to Ly6G⁺CD11b^hi LDNs.

The role of MDSCs in various inflammatory diseases including sepsis has been studied [14, 16, 17], but their phenotypic characteristics were minimally identified. In mice, MDSCs are mainly defined as Gr1⁺ CD11b⁺ cells, where Gr1 or granulocyte receptor-1 antigen consists of Ly-6G and Ly-6C antigens [14]. PMN- or G- MDSCs are CD11b⁺ Ly6G⁺ Ly6C^lo cells and M-MDSCs are CD11b⁺ Ly6G⁻ Ly6C^hi cells [14]. In the current study, since we only focused on Ly6G⁺ cells in the low-density area of Ficoll gradient, the presence of M-MDSCs which are Ly6G⁻ cells were excluded from our experimental LDNs. However, in addition to Ly6G Ag, since the G-MDSCs also express CD11b [14], the presence of G-MDSCs in our experimental cells cannot be ruled out. In the CD11b expressing Ly6G⁺ LDNs we found two distinct groups, one expressing high level of CD11b and the other expressing low level of CD11b. We then compared the other phenotypic markers and functional differences between CD11b^hi and CD11b^lo populations of Ly6G⁺ LDNs. We found that both these subsets of LDNs expressed CXCR4, ICAM-1, iNOS, ROS, and NETs at different intensities, and the highest expression of these markers were found in Ly6G⁺CD11b^hi LDNs. These data indirectly indicate that the Ly6G⁺CD11b^hi LDNs may not exhibit MDSC-like suppressive function as they express the inflammatory markers and mediators at higher intensities compared to Ly6G⁺CD11b^lo LDNs. Moreover, we also found that Ly6G⁺CD11b^hi LDNs produced increased levels of iNOS that could probably distinguish them from the G-MDSCs as other studies reported decreased expression of iNOS is the hallmark feature of G-MDSCs [14]. Since the function of MDSCs was determined by their ability to suppress T cell function [15, 16], further studies of CD11b^hi and CD11b^lo Ly6G⁺ LDNs on T cell function may reveal whether or not these cells were functionally related to MDSCs.

Neutrophil ageing as determined by the presence of CXCR4 on their cell surface is directly linked with increased NET formation [13]. LPS induces the expression of CXCR4 on neutrophils through TLR4 [13], but whether or not the DAMPs like eCIRP induces neutrophil aging was not known previously. We found that LY6G⁺CD11b^hi LDNs expressed significantly increased levels of CXCR4 than LY6G⁺CD11b^lo LDNs. We therefore can assume that CXCR4 expressing LY6G⁺CD11b^hi LDNs may play pro-inflammatory role in sepsis through the increased release of NETs. Additionally, our study identified that eCIRP can induce the expression of aged marker CXCR4 on the LDNs in sepsis through its receptor TLR4. Besides our identification of CXCR4 expression in LDNs, a recent study revealed that the expression of CXC chemokine receptor 2 (CXCR2) in LDNs was lower than HDNs [20]. ICAM-1 mainly expressed on the endothelial cells facilitates leukocyte migration by interacting with lymphocyte function-associated antigen-1 (LFA-1) [39–41], a receptor found on the leukocytes. However, a small portion of neutrophils expressing ICAM-1 after stimulation of neutrophils with LPS or eCIRP were shown to be implicated in sepsis [10, 11]. The ICAM-1 expressing neutrophils produce increased levels of iNOS, ROS, and NETs [10, 11]. Here, we found that LY6G⁺CD11b^hi LDNs expressed increased levels of ICAM-1 compared to LY6G⁺CD11b^lo LDNs. The increased production of ROS, iNOS, and NETs in LY6G⁺CD11b^hi LDNs can be correlated with the NET-forming ICAM-1⁺ neutrophils. Since ROS serves as a positive regulator of NET formation [7], increased ROS production by LY6G⁺CD11b^hi LDNs may be the plausible mechanism of increased NET formation.

We identified a mechanism by which eCIRP induced the formation of LY6G⁺CD11b^hi LDN from naïve BMDN. We showed that eCIRP utilized TLR4 for generating LY6G⁺CD11b^hi LDN. We noticed that the frequency of LY6G⁺CD11b^hi LDN in in vitro rmCIRP-stimulated BMDN were 5-fold higher than in vivo rmCIRP-injected mice blood. This could be due to different experimental conditions, optimum dose and time-point of rmCIRP’s stimulation. We further identified the downstream molecules the transcription factors that were upregulated in eCIRP-induced LDN. Identification of these transcription factors to generate eCIRP-mediated LY6G⁺CD11b^hi LDN supports the fact that these neutrophils are inflammatory as these transcription factors are mainly relevant to inflammation [42–46]. In the present study we elucidated the role of eCIRP on the heterogeneity LDN in terms of generating CD11b^hi and CD11b^lo populations and identified that LY6G⁺CD11b^hi LDN are more pro-inflammatory. However, since the mature and majority of the circulatory neutrophils sediment at the high density area (the bottom layer) of the Ficoll gradient, a comparison between the LY6G⁺CD11b^hi population in LDN and HDN will be an interesting study to revel eCIRP’s role in neutrophil heterogeneity and function. Here, we only assessed the blood LY6G⁺CD11b^hi LDN, although neutrophils infiltration in various tissues especially in lungs during sepsis is common and have been implicated to aggravated inflammation. In our previous study, we identified NET-forming ICAM-1⁺ neutrophils in the lungs during sepsis [11], since LY6G⁺CD11b^hi LDN express ICAM-1 and produce NETs and are responsive to eCIRP stimulation, it is reasonable that these LDN can also be found in lungs in sepsis. Determination of LY6G⁺CD11b^hi LDN in lungs will implicate the pathophysiology of tissue damage by these cells in sepsis.

During inflammation, the bone marrow neutrophils are released into the peripheral circulation [47]. As such, we cannot exclude the fact that the increased release of bone marrow neutrophils during sepsis could contribute to the increase in Ly6G⁺CD11b^hi LDNs in the blood in sepsis. However, we showed that stimulation of freshly isolated BMDN with rmCIRP significantly increased the % of Ly6G⁺CD11b^hi LDNs. In sepsis, since the eCIRP level is elevated in the circulation [27, 34], the circulatory eCIRP may directly induce the generation of Ly6G⁺CD11b^hi LDNs in the blood. In the flow cytometry gating, there might be some CD11b^lo cells overlapping with the CD11b^hi population. We gated Ly6G⁺CD11b^hi LDNs based on the expression of CD11b at high levels which ascended from the center of the population to the right end of the cells in the dot blot. We assessed the expression of CXCR4, ICAM-1, iNOS, ROS, and NETs in Ly6G⁺CD11b^lo and Ly6G⁺CD11b^hi LDNs and found that Ly6G⁺CD11b^hi LDNs exhibited increased expression of these markers, supporting the notion that Ly6G⁺CD11b^hi LDNs are distinct cell population. Akin to our gated Ly6G⁺CD11b^hi LDNs, a recent study identified LDNs based on the expression of CD11b, in which the boundary between CD11b^lo and CD11b^hi populations was found to be overlapped with each other [20]. Further studies to determine other markers may be helpful for stringent separation of these two subsets. Our results revealing the elevated expression of certain transcription factors in rmCIRP stimulated neutrophils may further identify the new markers of Ly6G⁺CD11b^hi LDNs.

In human, eCIRP is elevated in sepsis patients and directly relevant to the disease severity [27, 34]. Our study of identifying the subsets of mouse LDNs as induced by eCIRP in sepsis further focuses on identifying valuable insights that may be applicable to human LDNs in sepsis patients to develop novel therapeutics targeting eCIRP and LY6G⁺CD11b^hi LDNs.

Abbreviations

eCIRP

Extracellular cold-inducible RNA-binding protein

LDNs

low-density neutrophils

CLP

cecal ligation and puncture

BMDN

bone marrow-derived neutrophils

ROS

reactive oxygen species

MPO

myeloperoxidase

NETs

neutrophil extracellular traps

PAMPs

pathogen-associated molecular patterns

DAMPs

damage-associated molecular patterns

ICAM-1

intercellular adhesion molecule-1

TLR4

toll like receptor 4

MDSCS

myeloid-derived suppressor cells

TREM-1

triggering receptor expressed on myeloid cells-1

rmCIRP

recombinant murine CIRP

LFA-1

lymphocyte function-associated antigen-1