A novel model of common Toll-like receptor 4- and injury-induced transcriptional themes in human leukocytes

Healthy adult subjects were recruited by public advertisement and screened for inclusion in this study under approved guidelines of the Institutional Review Board of the Robert Wood Johnson Medical School. Written informed consent was obtained from all patients participating in the study. Inclusion criteria for the study were normal general health as demonstrated by medical history and physical examination, complete blood count, and basic metabolic panel within normal lab limits. Exclusion criteria included a history of any acute or chronic disease, arrhythmia, recent history of alcohol, drug or medication ingestion, pregnancy or prior exposure to endotoxin in the experimental setting.

Upon accrual to the study, the subjects were admitted to the Clinical Research Center (CRC) at UMDNJ-Robert Wood Johnson Medical School the afternoon prior to the study and a repeat examination confirmed that no changes in health status had occurred since enrollment. Female subjects underwent a urine pregnancy test. The subjects' characteristics are summarized in Table 1. The volunteer subjects were placed nil per os (NPO) at midnight prior to the endotoxin study day, and underwent intravenous fluid hydration (1 ml/kg-hr) until completion of the acute study phase. Following admission, subjects were randomized to one of two study groups. Subjects assigned to Groups B and D (Table 2) received a placebo infusion of physiologic saline prior to endotoxin administration. PBL samples obtained from these subjects prior to endotoxin infusion were used as baseline (Group A; Table 2). Subjects assigned to Groups C and E (Table 2) received continuous intravenous infusion of cortisol (3 μg/kg/min) for 12 hours starting six hours before endotoxin administration [17]. Subjects assigned to Groups B to E received a one-time intravenous dose (2 ng/kg) of endotoxin (NIH Clinical Center Reference Endotoxin; CC-RE-Lot2) at 0 hour (0900 clock time). Blood samples were drawn at six hours (Groups B and C; Table 2) and 24 hours (Groups D and E; Table 2) post-endotoxin.

Patients were accrued from the adult Surgical ICU at Robert Wood Johnson University Hospital under a protocol approved by the Institutional Review Board of the Robert Wood Johnson Medical School.

The patient demographic characteristics are described in Table 1. An anticipated ICU stay of at least 72 hours and anticipated ultimate survival were utilized as inclusion criteria. Patients were excluded if they had a suspected or confirmed infection, received an organ transplant, required more than six units of blood transfusions and/or had severe traumatic brain injury (admitting GCS < 8). Blood samples were first drawn within one to five days of ICU admission, and again five to seven days later.

Blood samples were drawn in EDTA tubes, and centrifuged at 400 × g for 10 minutes. The plasma was removed, and the red blood cell/leukocyte pellet was treated with bicarbonate-buffered ammonium chloride lysing solution (0.1% potassium bicarbonate; 0.826% ammonium chloride in H₂0) at a ratio of 1 part red blood cell/leukocytes to 20 parts lysing solution for 15 minutes in order to lyse the red blood cells. The leukocytes were then collected by centrifugation and washed once in lysing solution. After another centrifugation, a small aliquot of the leukocyte pellet was removed for performing a flow cytometric differential cell count on the healthy subjects. The leukocyte pellet was lysed in TRIzol™ solution (Sigma, St. Louis, MO, USA), sheared 10 times with an 18-gauge needle, and frozen at -70°C.

Steps outlined in this section were performed by the microarray core facility at this institution. Following fragmentation of the biotinylated cRNA, 15 μg was placed in hybridization cocktail, heated to 95°C, centrifuged and then hybridized to the Focus™ GeneChip microarray (Affymetrix) for 16 hours at 45°C. Chips were then washed, stained with streptavidin phycoerythrin and scanned on the Agilent Gene Array Scanner™ (Agilent Technologies).

We compiled a database that includes 38 Focus GeneChip^® microarrays (Affymetrix) derived from the study groups outlined in Table 2. The microarray data have been submitted to Gene Expression Omnibus [GEO:GSE22278]. The database includes two matching PBL samples obtained from five patients (Table 2). For four out of the five patients, the blood samples were obtained within 5 days (Group F) and 12 days of admission (Group G). The fifth patient was also sampled in the later phase but the microarray displayed a background level that precluded statistical analysis.

Focus Gene chip data CEL files were imported, grouped, and analyzed using GeneSpring™ software (Agilent Technologies). Primary analysis was carried out by log2 transformation followed by transformation to the median and RMA (quantile) normalization. Advanced significance analysis was performed on normalized-transformed data utilizing unpaired Student's t-tests. We further defined significantly expressed probes as those with a P-value < 0.05 and ≥1.2-fold change from baseline. Data were also exported for analysis by Ingenuity Pathway Analysis ™ (Ingenuity, Palo Alto, CA, USA) as previously described [3].

Where indicated, RNA was extracted as described above and reversed transcribed to cDNA using High capacity cDNA Archive kit™ (Applied Biosystems, Foster City, CA, USA). Gene expression was analyzed in duplicate by quantitative real-time polymerase chain reaction (qPCR) using inventoried TaqMan^® gene expression assays (Applied Biosystems) as described [16]. A list of the gene expression assays can be found in [16]. The relative gene expression analysis was performed using the 2^-ΔΔCT method [18]. The level of beta-2-microglobulin (B2M) expression was used as an internal reference [3, 19, 20].

Prior studies [3, 4] indicated a maximal change in PBL gene expression at the six-hour time point post endotoxin infusion in all volunteer subjects. Hence, this time-point was chosen to depict the influence of endotoxin. Expressed gene selection proceeded from the array database as outlined in Figure 1. Arrays representing PBL obtained after an in vivo endotoxin challenge (Group B), or antecedent cortisol plus endotoxin challenge (Group C), as well as those obtained from trauma patients within five days of admission (Group F), were independently compared to baseline. Gene probes that were significantly differentially expressed (both P ≤ 0.05 and >1.2 fold-change) were then selected (Figure 1a). Out of the 8,793 genes represented on the Focus GeneChip^® (Affymetrix) microarrays, 2,338 (27%) and 2,962 (34%) genes were differentially expressed, by the criteria described above, in PBL six hours after challenge with endotoxin, without or with cortisol, as compared to baseline (Figure 1a). Of these, 1,956 were common to PBL treated with endotoxin (Group B) and cortisol plus endotoxin (Group C) (Figure 1a).

Numerous genes (1,581; 18%) were also differentially expressed (both P ≤ 0.05 and >1.2 fold-change) in PBL obtained from trauma patients within the first five days of admission as compared to baseline values of normal subjects. Based on these similarities, 937 genes were significantly differentially expressed in all three groups (Group 1; Figures 1a). Scatter plot analyses revealed that the gene expression trends were highly correlated among the three groups (Figure 1b). These data suggest a significant commonality among differentially expressed genes during the early, dynamic phase of TLR4-induced inflammation resulting from endotoxin infusion, and those differentially expressed in PBL in the early post-trauma time period.

Next, we sought to determine which of the 937 genes that are differentially expressed during the peak of systemic inflammatory responses, and during the first several days after a trauma event, remain differentially expressed in PBL obtained at later time points of up to 12 days after ICU admission. To that end, we first selected 1,136 genes that were differentially expressed in PBL obtained from trauma patients after 9 to 12 days of admission (Figure 1c), and then identified genes that were common to both this later injury phase group and those genes defined as Group 1 genes (Figure 1d). This resulted in the identification of 445 genes (5.4%) that persisted in differential expression in response to TLR4-induced systemic inflammation and/or injury. We refer to this group of TLR4 and injury responsive genes as "TIR" genes. The 445 TIR genes are listed in Table S1, which can be found in Additional file 1.

The TIR genes selected as outlined in Figure 1, plus those from the 24 hours post-endotoxin groups (Table 2; Groups D and E) were subjected to hierarchical cluster analysis. As shown in Figure 2a, the clustering analysis defined two dominant groups. Cluster 1 included both baseline samples and all PBL samples derived from subjects at 24 hours after endotoxin. Cluster 2 included all the PBL samples derived from subjects at 6 hours post-endotoxin challenge as well as the trauma patient samples.

One strength of the present analysis is the identification of gene expression patterns common to both de novo endotoxin and injury-induced conditions. As a consequence, there is likely to be a lesser transcriptional influence of clinical management factors, such as prior transfusions of blood products, vasopressor use, or opiates and other therapeutics since these agents were not utilized in the endotoxin challenged subjects. While we cannot completely exclude interacting effects from interventions and therapies, the common transcriptional themes arising from the present analysis strongly suggest pathways dominated by endotoxin or other TLR4 agonist influences in vivo. Although it is documented that circulating endotoxin is frequently detectable in trauma/burn patients [21, 22] as well as in more heterogeneous ICU populations [23], the presence of detectable endotoxin is far from uniform in these patients. Since we did not measure endotoxin or other soluble factors, such as HMGB1, S100A/B, or acute phase proteins that may also serve as TLR4 activating ligands, we cannot further speculate on whether the derived leukocyte transcriptional signatures are attributable to endotoxin or other mediators.

We also examined the TIR gene expression trends using a published database [3] [GEO:GSE3284] that includes microarray data derived from four previously reported endotoxin challenged subjects at 0, 2 4, 6, 9 and 24 hours post challenge, and four control subjects studied at parallel time points. The TIR genes showed a robust response in all endotoxin-challenged subjects, and a return to baseline by 24 hours post treatment (Figure 2b). Furthermore, hierarchical cluster analysis revealed two dominant clusters. Cluster 1 included a total of 30 samples representing 26 control samples plus 4 PBL samples obtained from endotoxin challenged subjects at 24 hours post-infusion (Figure 2b). Cluster 2 included all the PBL samples obtained between two and nine hours post-infusion (Figure 2b). This significant degree of correspondence between a prior endotoxin challenged population and the present volunteers group confirms the fidelity of our baseline and endotoxin challenged-subjects analysis.

The TIR genes group includes 272 downregulated and 173 upregulated genes (Table S1, which can be found in Additional file 1). The most striking feature of this group of differentially expressed genes is the abundance of RPL (ribosomal proteins associated with large 60S ribosomal subunit) and RPS genes (ribosomal proteins associated with small 40S ribosomal subunit) (for a recent review see [24]). Furthermore, 50 of the 53 RPL/RPS genes are downregulated. Among the downregulated TIR genes are also three EIF/EEF genes, which encode translation initiation factors, and six HNRNP genes, which regulate pre-mRNA processing and other aspects related to mRNA metabolism (for example, [24, 25]).

The expression data were analyzed through the use of Ingenuity Pathway Analysis (Ingenuity^® systems) as previously described (for example, [3, 26]). This analysis classified the TIR genes into five main modules, each representing 140 genes (the maximum number of genes that the program associates with each module). Three out of the top five modules, which include approximately 230 TIR genes in total, are related to protein synthesis pathways. Two additional pathways, a lipid metabolism pathway, and a cellular assembly and organization pathway, included, respectively, 71- and 68-TIR gene matches.

The top matching module shown in Figure 3 includes 99 TIR genes. Myc, a global transcription regulator of many cellular processes, including ribosomal biogenesis and protein synthesis (for example, [24]), is the focal point for the most densely populated node encompassing numerous RPL/RPS genes. This large number of suggested interactions is not surprising given that more that 600 genes, including 48 transcription factors, were identified as direct Myc-regulated gene targets in human B lymphoid tumor cells alone [27]. Furthermore, TIDBase, a web-based public resource supported by the type 1 diabetes (T1D) research community [28], identified more than 1,400 Myc-related interactions. We speculate that the implied reduction of PBL protein synthesis capacity is highly significant. A decline in transcripts associated with transcription was first observed in PBL obtained from endotoxin-challenged subjects [3]. However, the endotoxin-induced changes in PBL gene expression were all transient, with recovery within 24 hours. By contrast, the identification of a similar and persistent gene expression signature in PBL obtained from trauma patients 1 to 12 days post-admission clearly suggests that the translational function of circulating leukocytes is consistently reprogrammed to a lower state.

Importantly, among the upregulated TIR genes were several genes that are known to be associated with glycolysis. These include PFKFB3, encoding 6-phosphofructo-2-kinase (PFK-2), and HK3, encoding hexokinase 3. PFK-2 is a bifunctional enzyme that catalyzes the synthesis and degradation of fructose 2,6-biphosphate, which in turn, stimulates 6-phosphofructo-1-kinase, the key regulator of mammalian glycolysis [29]. An increase in PFKFB3 (also known as iPFK2) expression has been documented in endotoxin-treated cultured human monocytes [30]. Hexokinase 3 phosphorylates glucose to produce glucose-6-phosphate, the first intermediate in glycolysis. We also observed an upregulation of SLC2A3, encoding the glucose transporter Glut 4, and PDK3, encoding pyruvate dehydrogenase kinase (PDK). PDK is an inhibitor of pyruvate dehydrogense complex, which is positioned at the junction between glycolysis and the TCA cycle [31]. In cancer cells, an increase in PDK3 expression was associated with an increase in lactic acid production, which is indicative of a decrease in mitochondrial respiration [32]. These collective changes in gene expression predict an increase in glucose consumption and glycolysis. This possibility is supported by studies in endotoxin-challenged rats, wherein an increase in glucose utilization in multiple organs was observed within hours of an endotoxin or TNFα challenge [33, 34]. These data suggest that the systemic conditions induced by acute TLR4 ligation, resulting in enhanced PBL glycolysis, also persist for an extended period after trauma.

Included among the suppressed TIR genes was also Rorα, one of the key regulators of the circadian clock [35]. The circadian clock is an autoregulatory feedback network of transcription factors and proteins whose activity and/or availability cycle with a periodicity of approximately 24 h [36‐38]. The central "master" clock controlling behavioral circadian rhythms is located in the suprachiasmatic nucleus (SCN) within the brain hypothalamus [39, 40]. The central clock both regulates and receives inputs from peripheral clocks present in most tissues, including peripheral blood leukocytes [41‐44]. Multiple circadian clock genes, including Clock, Cry1, Cry2, Per3, and Rora, are significantly suppressed within two hours after an endotoxin-challenge and remain suppressed for up to 17 hours post-infusion [16]. We therefore sought to determine the status of Clock, Cry1, Cry2, Per3, and Rora expression in a subset of these surgical ICU patient samples. Our analysis revealed a significant and uniform reduction in PBL clock gene expression during the first week of ICU admission (Figure 4). Bmal1, the only gene not affected in endotoxin-challenge PBL [16], was also not reduced in PBL obtained from patients. Several genes, including Cry1, Per3, and Rora remained suppressed in the patients studied for at least an additional week during ICU admission (Figure 4). Our analysis thus suggests that the transient decline in circadian clock gene expression in PBL first noted during systemic inflammation induced by TLR4 activation [16] persists for an extended period in patients with injury induced systemic inflammation.