A prospective analysis of lymphocyte phenotype and function over the course of acute sepsis

Patients older than 18 years of age admitted to the medical or surgical ICU at Barnes-Jewish Hospital with a diagnosis of sepsis were identified. Severe sepsis was defined using a consensus panel definition of the presence of systemic inflammatory response syndrome (SIRS) criteria, known or suspected source of infection and organ failure [31]. All patients had respiratory failure requiring mechanical ventilation on enrollment. Patients with active malignancy, HIV infection, chronic Hepatitis B or C infection or on immunosuppressive medications (except corticosteroids at a dose of < 10 mg prednisone or equivalent per day) were excluded. Consent was obtained from the patients legally authorized representative, as all patients enrolled were judged too acutely ill to provide valid consent. Onset of severe sepsis, defined by the time at which the subject met consensus criteria for severe sepsis was confirmed by the study physician to be within 24 hours of enrollment. Control subjects consisted of normal age-matched volunteers with no significant concomitant acute or chronic illnesses. Clinical data on the septic patients was captured daily. Patients were monitored daily for evidence of new infection, and considered to have developed a secondary infection if there were new positive cultures and/or clinical data such as new fever, elevated white blood cells (WBC) and new infiltrate on chest X-ray consistent with new infection. All protocols were approved by the Washington University Institutional Review Board.

Heparinized blood was collected through an indwelling central venous catheter, arterial line or venipuncture (septic patients) or by peripheral venipuncture (controls) on day 0 and again at the end of the study (day seven unless otherwise indicated). The blood was immediately transported and processed in the research laboratory. Peripheral blood mononuclear cells (PBMC) were isolated by ficoll-hypaque density gradient separation, using standard protocols. Plasma was collected and stored at -80°C for cytokine determinations. The cells were washed and resuspended in T cell media (Roswell Park Memorial Institute (RPMI) 1640 supplemented with 10% heat-inactivated FCS, hydroxyethyl piperazineethanesulfonic acid (HEPES), penicillin/streptomycin, L-glutamine and non-essential amino acids) and processed for staining, proliferation or cytokine secretion as described below.

All antibodies were purchased from BD Biosciences (San Diego, CA, USA), eBiosciences (San Diego, CA, USA) or Biolegend (San Diego, CA, USA). For surface marker staining 100 μl of whole blood was incubated with 10 μl human AB serum and the indicated fluorescently conjugated antibodies for one hour at room temperature followed by hypotonic red blood cell lysis. The cells were extensively washed in PBS containing 1% BSA and then resuspended in PBS containing 1% BSA and 1% paraformaldehyde (PFA). For intracellular staining of FoxP3, following surface staining with α-CD4 and α-CD25, cells were permeabilized with Fix/Perm buffer (eBiosciences) and incubated with either control or α-FoxP3 antibody for one hour at 4°C, washed then resuspended in PBS/BSA containing 1% PFA. Viable lymphocytes were identified by forward scatter (FSC) and side scatter (SSC) properties and exclusion of 7-aminoactinomycin D (7-AAD). T cells were identified as either CD4+ or CD8+. Regulatory T cells were identified as CD4+ and CD25+ that co-labeled with FoxP3. NK cells were identified as CD56+ while natural killer T (NKT) cells co-labeled with CD3 and CD8+ NK cells co-labeled with CD8. Dendritic cells were identified as lineage cocktail negative (CD3/CD19/CD16/CD14/CD20/CD56), HLA-DR+ and sub-typed as plasmacytoid dendritic cells (pDC, CD123+) or monocytoid dendritic cells (mDC, CD11c+). MDSC were identified as lineage cocktail negative, HLA-DR low/negative, CD33+ and CD11b+. All samples were analyzed by flow cytometry on a FACSCalibur cytometer (Becton Dickinson Corp., Mountainview, CA, USA). Data were further analyzed using WinList Software (Verity Corporation, Topsham, ME, USA).

PBMCs were resuspended in T cell media at 2 × 10⁶ cells/ml and 50 μl added to each well of a round bottom 96-well tissue culture plate. The cells were either left unstimulated or treated with α-CD3 (1 μg/ml, OKT3, eBiosciences) in combination with α-CD28 (1 μg/ml, clone CD28.2, eBiosciences) or with phorbol myristate acetate (PMA) (5 ng/ml) plus ionmycin (0.4 μg/ml). After 48 hours, 1 μCi of tritiated thymidine (³H-TdR) was added to each well and allowed to incubate overnight. Cells were harvested onto glass microfiber filters and counts of incorporated ³H-TdR determined by liquid scintillography. For determination of cytokine production, cells were cultured as described for proliferation and culture supernatant obtained at five hours and 48 hours following stimulation. Levels of cytokine present in plasma and culture supernatants were determined using the cytokine bead array (BD Biosciences) according to the manufacturer's protocol using the human Th1/Th2/Th17 kit that measures IL-2, IL-4, IL-6, IL-10, IL-17A, TNF-α and IFNγ. For some samples, the cells were allowed to rest overnight in fresh media, then stimulated as above and culture supernatant collected after five hours for cytokine determinations.

Admissions to the medical and surgical ICU were screened daily for potential subjects. Those who met enrollment criteria were identified and their legally authorized representative approached to obtain informed consent. Twenty-four subjects were enrolled and initial samples collected. Of these 24 subjects, second samples were collected from 16 at day seven while five subjects died prior to the collection of a second sample. One subject had the second sample collected on day nine and two subjects had samples collected immediately prior to the withdrawal of life-sustaining therapies (on day two and day five). The clinical characteristics of the septic population are shown in Table 1. As expected, the patients were acutely ill with a median Acute Physiology and Chronic Health Evaluation II (APACHE II) score of 20 and a mortality of just over 40%. A number of comorbidities were present as is typical for this patient population. Three subjects were on low dose corticosteroids at the time of enrollment. Control subjects were identified and selected to approximately age- and sex-match the patient population. Only those with no underlying chronic health conditions were included as controls.

At enrollment the median total WBC count was 12,900 cell/μl with an absolute lymphocyte count (ALC) of 971 cells/μl. Half of the subjects were lymphopenic with an ALC of < 1,000. Over the first week, the ALC increased on average by 340 cells/μl, but this change was not statistically significant. To determine the distribution of lymphocyte subsets as well as expression of cell surface molecules on specific cell populations, blood was obtained within 24 hours of the onset of severe sepsis and extensively characterized by flow cytometry (Figures 1 and 2). An identical analysis was performed on healthy control subjects to provide a comparison population. The markers used to identify each cell population are described in the methods. As shown in Figure 1, there was a reduction in the percentage of CD4+ T cells and a trend towards a decrease in CD8+ T cells in the blood of acutely septic patients. A corresponding decrease in innate immune cells was also observed in the blood of septic patients with the percentage of NK cells and dendritic cells (both pDC and mDC) being reduced. Expression of CD86, a ligand for the T cell costimulatory receptor CD28, was decreased on mDC. While there was no difference in the percentage of Treg cells at day 0, there was a trend towards an increase in CD11b+ myeloid derived suppressor cells, a population implicated in mediating immune suppression in cancer and infection [18, 19] in acutely septic subjects.

In the acute phase of sepsis, both lymphocyte activation and depletion has been documented [2, 3]. T cells isolated from the blood of acutely septic individuals also had elevated expression of CD25, as a percentage of CD4+ T cells (Figure 1), and of CD69 (Figure 2), as both a percentage of positive cells and expression level per CD4+ T cell (MESF), indicative of an activated phenotype. Interestingly, we also observed a trend in elevated expression, as determined by MESF (P = 0.07), of Programmed Death Receptor-1 (PD-1), an important negative regulator of lymphocyte function, in the septic individuals compared to healthy subjects. The other major inhibitory receptors expressed by T cells are BTLA and CTLA-4. The majority of T cells expressed BTLA (> 90%) in both septic and non-septic individuals. There was a higher percentage of CD8+ T cells that expressed LAG-3 as well as elevated per cell expression when compared to age-matched healthy individuals. No other significant differences in CD8+ T cell expression of either activation or inhibitory receptors were observed. These data suggest that the observed changes are specific and not due to non-specific upregulation of all cell surface receptors.

We hypothesized that specific changes in receptor expression patterns over the course of acute sepsis would be detected. In particular, we were interested in whether receptor ligand pairs that had been implicated in inhibiting lymphocyte function were upregulated over time. Therefore, we repeated the flow cytometric analysis on samples obtained later in the disease. Except as noted in the methods, the repeat analysis was performed on day seven for all subjects in whom a second sample was able to be collected. As presented in Figure 3, changes in the expression of several receptors were noted over this time interval. On both CD4+ and CD8+ T cells, we observed a trend towards a decrease in per cell expression of the IL-7 receptor (CD127) and a decrease in the overall percentage of CD8+ cells that expressed the IL-7R. Expression of TIM-3 and LAG-3, as determined by quantitative MESF values was also increased on CD4+ T cells. On CD4+ and CD8+ T cells, we detected an increase in expression of CTLA-4 by MESF as well as an increase in the percentage of cells expressing CTLA-4. Models of cell exhaustion have reported persistent elevated expression of the activation marker CD69. We found that patients with sepsis did have a higher expression of CD69 on CD4+ T cells at the onset of sepsis, as well as a higher circulating percentage of CD69 expressing CD4+ T cells and this did not significantly change over the course of the acute illness (data not shown). The percentage of pDC that expressed the ligand for PD-1, PD-L1, increased over the duration of sepsis, as did expression of CD86 on mDC (Figure 4). Although we did not detect any difference in the percentage of Treg cells at the onset of sepsis, this increased significantly over the interval in septic patients but not in normal controls (Figure 4). This pattern of increased inhibitory receptor expression is consistent with our findings in patients who died of sepsis, and is compatible with the phenotype of immune cell exhaustion. Smaller, but statistically significant changes in the expression of a few markers were also detected in control subjects, highlighting the need for caution in interpreting these results.

We measured the level of a panel of seven circulating cytokines in the plasma of patients with sepsis and controls. Of those measured, only IL-6 and IL-10 were different with the remaining cytokines being very low or undetectable in either subject population (Figure 5 and Additional File 1, Table S1). Both IL-6 and IL-10 were significantly higher in patients with sepsis and both decreased markedly from enrollment to the end of the study. Three of the four patients with the highest IL-6 levels at enrollment died, two during the first week and one later in the hospitalization. We did not detect any relationship of IL-10 levels or the IL-6:IL-10 ratio to clinical outcome (data not shown).

We had previously reported that PBMCs isolated from patients who died of sepsis had a profound global reduction in cytokine secretion when stimulated in vitro [20]. To determine whether the defect was present in patients with acute sepsis, we performed a similar analysis on PBMCs isolated at enrollment and at the end of the study. In contrast to our findings with post-mortem specimens, in this study we found that cytokine secretion from PBMCs activated in vitro was largely preserved with the notable exception of IFNγ (Figure 6 and Additional File 2, Table S2). Secretion of IFNγ following 48 hours of stimulation with α-CD3/α-CD28 antibody was markedly reduced compared to normal controls in samples obtained at either enrollment or at the end of study participation.

One potential explanation for the observed defect in IFNγ secretion was that factors present or absent in the septic milieu might reversibly impair cytokine secretion. To test this, the cells were either stimulated immediately or rested overnight in fresh media and then stimulated for five hours with α-CD3/α-CD28. As shown in Figure 6C, the rested cells from patients with sepsis, but not controls, had significantly greater secretion of IFNγ (rested: 16.1 ± 23.1 pg/ml versus no rest: 5.7 ± 15.7 pg/ml P < 0.01). The fact that overnight rest in fresh media restored IFNγ secretion to levels near those of controls subjects suggests that the impairment may in fact be reversible, perhaps by removal of an inhibitory factor or addition of a growth factor present in fresh media.

As an additional measure of lymphocyte function, we assessed the proliferative capacity of PBMCs following in vitro stimulation (Figure 6D). The mean proliferation of PBMCs stimulated with α-CD3/α-CD28 from septic individuals was similar to non-septic individuals at enrollment and at the end of the study. Therefore, although there was a decrease in IFNγ secretion, there is not a global impairment in all cellular functions in this patient population.

We were interested in determining which, if any, of the immunologic variables measured correlated with a clinically significant outcome. All patients with sepsis were followed over the course of their hospitalization for overall mortality or evidence of a new infection. In addition, we determined if patients were re-admitted within 28 days of the enrollment with a new infection. In the course of the study ten patients died as a result of their illness, six patients developed a second infection (two of whom died. Two infections occurred after discharge and resulted in readmission within 28 days.) and ten patients survived without a second infection. We analyzed whether any parameters we measured at enrollment correlated with either second infection and mortality together or mortality alone. Consistent with previous reports, patients who died or had second infections had higher plasma IL-6 levels than those who had an uneventful recovery (Figure 7A) [32]. Interestingly, those patients who developed a second infection or died had significantly less IFNγ secretion from in vitro stimulated PBMCs than those who recovered uneventfully (Figure 7B). These data are supportive of the hypothesis that functional immune cell impairment may predispose to a poor outcome.