CXCR4+ PD-L1+ neutrophils are increased in non-survived septic mice

The dysregulated host response to infections can lead to sepsis, a complex disease characterized by a spectrum of clinical phenotypes associated with the host immune response variability and outcomes. This heterogeneity poses challenges for implementing specific therapeutic approaches. While clinical sepsis phenotypes are well distinguished, the mechanisms driving these heterogeneous responses remain poorly understood. Using an unbiased experimental approach, we analysed immune cell activation profiles in survived and non-survived CLP-septic to gain insights into the immunological mechanisms by which neutrophils contribute to the hyperinflammatory septic phenotype. Our finds reveal that non-survived septic mice exhibit increased frequencies of immature CXCR4+ PD-L1+ neutrophils and monocytes in the bloodstream, accompanied by an accumulation of trafficking-specific CXCR4+ PD-L1+ neutrophils into the lungs. The increased PD-L1 expression on CXCR4+ neutrophils is associated with increase of IFN-gamma signaling pathways. Additionally, the IFN-gamma and LPS promote an activation profile of CXCR4+ PD-L1+ neutrophils, exhibiting a phenotype associated with inflammation and organ damage. Notably, abrogating the IFN-gamma reduced susceptibility to CLP-sepsis and diminished PD-L1 expression on CXCR4+ neutrophils. This study provides molecular and functional insights into the immune cell activation profiles associated with the worsening of the septic hyperinflammatory phenotype experimental model. The CXCR4+ PD-L1+ neutrophils population and elevated plasmatic IFN-gamma levels highlighted here represent promising targets for therapeutic modulation in clinical sepsis hyperinflammatory phenotype. Isolation of leukocytes and single-cell RNA sequencing: Sepsis was induced using the cecal ligation and puncture (CLP) model, as previously described (Rittirsch et al., 2009). The mice were anesthetized by inhalation of 1.5 % of isoflurane, and two punctures were made through the ligated cecum using an 18-gauge needle. After the surgery, 1 mL of saline and tramadol 25 mg/kg was injected subcutaneously. After 6 hours of the surgery, antibiotic treatment with 30 mg/kg ertapenem (Merck Sharp & Dohme) was administered, followed by six additional doses every 12 hours. After 12 hours of the CLP-sepsis induction, 50 uL of blood samples were collected from the retro-orbital plexus of anesthetized mice, and the sepsis survival rate was monitored for up to 7 days. Leukocytes were isolated using positive magnetic isolation of CD45+ cells, following the manufacturer’s instructions (Miltenyi Biotec, cat#130-052-301). Leukocytes from each mouse were the multiplexed with a molecular tag following the 10x Genomics 3’ CellPlex Kit (10x Genomics, cat#1000261), and cells from 10 mice were encapsulated in two lanes of a 10× Chromium instrument, each 5 mice per lane. Libraries were constructed using the Chromium Next GEM Single Cell 3' Kit v3.1 following the manufacturer’s instructions (10x Genomics, cat#1000269), and were subsequently sequenced on the HiSeq 4000 (Illumina). Single-cell RNA sequencing data pre-processing: Base call files were converted in the Cell Ranger to FASTQ files using the mkfastq function. The reads were then aligned to the mm10 Mus musculus transcriptome using the Cell Ranger software pipeline (version 6.1.2) provided by the 10x Genomics to generate raw genes by cell matrices of unique molecular identifier counts for each sample. The samples were then grouped in health controls, survived, and non-survived, and were read into R using the read10x function (Hao et al., 2021), resulting in a count matrix with 32285 genes and 8248 cells. A Seurat object was then created, and a filter was applied to remove the low-quality cells by excluding those based on unique molecular identifier counts (nCount_RNA < 10000), the number of unique genes (nFeature_RNA between 200 and 3000), and the percentage of mitochondrial gene expression (percent.mt < 10). The remaining expression values were then normalized using the SCTransform (Hafemeister & Satija, 2019).