Identification of deleterious neutrophil states and altered granulopoiesis in sepsis

Abstract

Sepsis is a condition of high mortality arising from dysregulation of the host immune response. Gene expression studies have identified multiple immune endotypes but gaps remain in our understanding of the underlying biology and heterogeneity. We used single-cell multi-omics to profile 272,993 cells across 48 whole blood samples from 26 sepsis patients (9 with paired convalescent samples), 6 healthy controls and 7 post-surgery patients. We identified immature neutrophil populations specific to sepsis and demonstrated the immunosuppressive nature of sepsis neutrophils in vitro. An IL1R2+ neutrophil state was expanded in a transcriptomic sepsis endotype associated with increased early mortality (sepsis response signature 1, SRS1), together with enrichment of the response to IL-1 pathway in mature neutrophils, marking IL-1 out as a potential target for immunotherapy in SRS1 sepsis patients. We confirmed the expansion of immature neutrophils, specifically IL1R2+ neutrophils, in SRS1 in additional cohorts of patients (n = 906 RNA-sequencing samples, n = 41 CyTOF samples). Neutrophil changes persisted in convalescence, implicating disrupted granulopoiesis. Our findings establish a cellular immunological basis for transcriptomically defined sepsis endotypes and emphasise the relevance of granulopoietic dysfunction in sepsis, identifying opportunities for precision medicine approaches to the condition.

An immune dysfunction score for stratification of patients with acute infection based on whole-blood gene expression

“…First, SepstratifieR relies on bulk gene expression, and cannot establish which cellular alterations cause immune dysfunction. Combining SepstratifieR with single-cell technologies is a promising research avenue, as evidenced by our recent study describing differences in granulopoiesis between SRS groups ( 31 ). Second, SRSq does not capture the full heterogeneity of sepsis, and orthogonal axes of variation could be lost if focusing exclusively on SRSq.…”

“…We thank all the patients, patient families, nurses, and clinicians who participated in the GAinS and MARS studies, and the COMBAT 29 , Gary Mills 30,31 , John Humphreys 30,31 , Kelsey Armitage 30,31 , Shond Laha 32 , Jacqueline Baldwin 32 , Angela Walsh 32 , Nicola Doherty 32 , Stephen Drage 33 , Laura Ortiz-Ruiz de Gordoa 33…”

High-throughput mass spectrometry maps the sepsis plasma proteome and differences in response

“…We have previously reported sepsis response signatures (SRS) from leukocyte transcriptomic datasets associated with outcome and differential response to therapy (Antcliffe et al, 2019; Burnham et al, 2017; Davenport et al, 2016), including patients with evidence of granulopoietic dysfunction involving specific neutrophil subsets, relative immune compromise and high mortality (SRS1) or changes involving T cell and adaptive immunity (SRS2) measurable as a quantitative trait SRSq (Cano-Gamez et al, 2022; Kwok et al, 2022). We found component 92 strongly correlated with lower SRSq (more similar to SRS2) (P adj =1.6×10 -172 , rho=0.78 Spearman) which included increased apolipoprotein APOF and differential RNA abundance including repressive chromatin regulator SAMD1 and T cell regulator ligase RNF114 (Figure S6G).…”

“…Incomplete knowledge of pathophysiology and failure to differentiate individual patient variation in the nature and timing of maladaptive host responses within the heterogeneous clinical syndrome of sepsis currently limit clinical trials (Marshall, 2014; van der Poll et al, 2021; Stanski and Wong, 2019). Sepsis subphenotypes informative for immune response state, outcome and therapeutic response are proposed based on clinical, laboratory and molecular stratifiers (Baghela et al, 2022; Cano-Gamez et al, 2022; Davenport et al, 2016; Kwok et al, 2022; Scicluna et al, 2017; Seymour et al, 2019; Sweeney et al, 2018; Wong et al, 2019). However, establishing the nature of the sepsis host response has been limited by incomplete knowledge of the sepsis plasma proteome.…”

Inflammatory sub-phenotypes in sepsis: relationship to outcomes, treatment effect and transcriptomic sub-phenotypes

“…In GAinS, IFN-α and TNFα levels were higher in SRS1 compared with SRS2 although not reaching statistical significance. Further work is needed to understand the causal and temporal relationship between inflammation and transition into the SRS1 sub-phenotype with recent evidence, for example, of granulopoietic dysfunction involving specific neutrophil subsets (44).…”