Sepsis is the 10th leading cause of death in the United States , and the mortality rate has not changed substantially in many years, in spite of the use a wide range of experimental therapeutics. Biological based therapies that work well in rodent models when given before challenge have typically not worked in human sepsis patients. This could be due to ineffectiveness of administration of these drugs after sepsis is established or the complex biological response with many partially redundant mediators, causing inhibition of a single molecule to be ineffective. A better understanding of the full range of gene expression changes associated with sepsis will be necessary to understand these issues, and this was the purpose of the present study.
Sepsis is a result of a complex systemic dysregulation of inflammation that results when the host is unable to contain an infection that usually starts in a primary organ. The inflammatory response to sepsis involves two distinct phases which are not always mutually exclusive. The initial acute phase is the systemic inflammatory response, which is closely followed by the compensatory anti-inflammatory response . The dysregulation of the immune response observed in sepsis is a combination of the local immune response at the primary site of infection and the systemic immune response. The site of primary infection plays a key role as it produces inflammatory molecules that modulate the systemic immune response. The cellular environment of each primary organ involved in sepsis including the immunological functional cells (e.g., the resident macrophages) is very specific also the microbial aetiology leading to the infection also differs considerably. Therefore it is expected that the local immune response in sepsis varies depending on the site of primary infection and thus it is important to evaluate the organ specific immune responses.
The abdomen is the second most common primary site of infection in sepsis. Intra-abdominal infections account for about 20% of all sepsis cases. We used a widely accepted [19, 20, 42–44] model of intraperitoneal inoculation of a non-pathogenic strain of E. coli to induce sepsis. Two early time points (1 h and 2 h) and one late time point (18 h) were used to evaluate the changes in the transcriptional profiles of cells in the peritoneal cavity of the infected animals. As described in the results, the early time points that are associated with increased bacterial counts are expected to reflect an acute inflammatory response. In our experience the sepsis model used in this study is associated with very low mortality, only one out of the ten mice in the survival study succumbed to infection (Additional file 1: Figure S1). This is expected since a non-pathogenic strain of E. coli was used for infection and most animals were able to fully clear the bacterial load by about 21 h post infection . Thus the 18 h time point chosen in the study gives a view of the transcriptional changes involved in initiation of the resolution of infection. The coordinated response of professional phagocytes (macrophages, monocytes and neutrophils) is essential for the resolution of infection. The bacteria were recognized by the resident peritoneal macrophages because of their proximity to the site of infection. Thus, most of the cells in the peritoneal fluid at the early time points were macrophages (Figure 1B). The significantly increased expression of markers associated with macrophages was also reflected in the microarray data at the early time points. The cytokines released by the macrophages recruit the neutrophils from the blood stream to the site of infection. So as the infection progresses, the cell population in peritoneal fluid predominantly consists of neutrophils (Figure 1C). This shift was also reflected in the microarray data at 18 h. Most of the cells at the early time points were macrophages while at 18 h neutrophils were the predominant cell type in the peritoneal fluid (Figure 1B, 1C). The 18 h time point in our mouse model is expected to correspond to the initiation of the resolution of infection. The significant DE genes at this time point were found to be associated with cell death of immune cells. The apoptosis of neutrophils is critical for the resolution of infection as it ends the sustained neutrophil recruitment and the phagoctytic clearance of the apoptotic neutrophils also re-programs the macrophages to an anti-inflammatory phenotype .
We compared our results with a recent meta-analysis of 12 studies that evaluated the blood profiles from early and late stages of sepsis from human subjects . In this meta analysis, there was no clear transition between pro- and anti-inflammatory phases at early or late stages. The up-regulation of TNF-α, IL-1α, IL-1β, CXCL-10, SOCS3, IL-10 at 1 and 2 h in the present study was similar to that observed in blood profiles of sepsis patients at early stages [11, 46]. Expression of these genes peaked at 2 h and was still up-regulated at 18 h in this study. Decreased expression of CD69, CD36 (DE 1.78 and −5.53 respectively at 18 h) and MHC II genes (H2-OB, Figure 3) in peritoneal cell profiles is consistent with the blood profiles of sepsis patients at late stages of sepsis . However, there are some differences in expression of some genes at 18 h, as would be expected due to differences in inflammatory responses in different anatomical locations [8–10]. For example, TNF-α levels in peritoneal cells is elevated at 18 h in our study (DE 5.57), but meta-analysis identified a decreased expression of TNF-α in blood profiles [36, 47]. The CCR2 and S100A11 genes are up-regulated in early stages of sepsis in blood  but down-regulated in the peritoneal cells in our sepsis model. Despite a number of similarities, gene expression in peritoneal cells differs from the patterns in blood, suggesting that cell type and anatomical location as well as species differences influence the specific characteristics of the local inflammatory response. Nevertheless, there were many similiarities between the early and late responses to sepsis in humans and mice.
An important finding of this study is the identification of 20 transcription regulators among the DE genes, whose role in sepsis was not previously known. Some of the transcription regulators expressed uniquely at 2 and 18 h are known to be involved in T cell regulation. Transcription factor 7 (TCF-7) was up-regulated and Kruppel like factor 10 (KLF10) was down-regulated at 2 h. TCF-7 is a critical regulator of T cell specification  while KLF10 induces TGF-β1 expression. TGF-β1, an anti-inflammatory cytokine, also regulates T cell activation . ZBTB7B, a zinc finger protein crucial for the commitment of MHCII restricted thymocytes to a CD4+ lineage  was up-regulated at 18 h. Strawberry notch homolgue2 (SBNO2) was also up-regulated at 18 h. SBNO2 is induced by IL-10 and contributes to its downstream anti-inflammatory effects . It is not clear if the expression of T cell-related transcription regulators indicates the presence of T cells in the peritoneal cavity or if these transcription regulators are also expressed in macrophages, and have not been recognized as having a role in these cells. Lymphocytes are present in the peritoneal cavity (~5-10% of total cells, data not shown), and we have previously observed expression of genes unique to T cells (e.g., TCR), so it is possible that T cell activation is an important event in sepsis. It is also possible that a set of transcription regulators not previously known to be expressed by macrophages are expressed and are potentially important in regulating inflammation. Whether changes in expression of these regulators at 18 h is due to altered expression in macrophages or differential patterns of expression in neutrophils vs. macrophages, they are significant compared to naive mice, thus implicating them in the host response to sepsis. It is not clear if these transcription factors are activated by known or previously unrecognized signaling pathways. We report here that the best known transcription factor associated with sepsis, NF-κB, exhibits no alteration in transcript quantity. However, transcriptomics does not reveal information about activation of signaling or transcription related molecules. Our previous study demonstrated that sepsis substantially increases activation of NF-κB using a transgenic reporter mouse model . This serves as a reminder that the changes in the quantity of transcripts for transcription factors may not correspond with activation of those transcription factors. All other things being equal, increased expression should yield increased activity, but activity needs to be measured for each transcription factor implicated in this study. This will be done in future studies.