Ex vivo immune profiling in patient blood enables quantification of innate immune effector functions

Abstract

The assessment of a patient's immune function is critical in many clinical situations. In complex clinical immune dysfunction like sepsis, which results from a loss of immune homeostasis due to microbial infection, a plethora of pro- and anti-inflammatory stimuli may occur consecutively or simultaneously. Thus, any immunomodulatory therapy would require in-depth knowledge of an individual patient's immune status at a given time. Whereas lab-based immune profiling often relies solely on quantification of cell numbers, we used an ex vivo whole-blood infection model in combination with biomathematical modeling to quantify functional parameters of innate immune cells in blood from patients undergoing cardiac surgery. These patients experience a well-characterized inflammatory insult, which results in mitigation of the pathogen-specific response patterns towards Staphylococcus aureus and Candida albicans that are characteristic of healthy people and our patients at baseline. This not only interferes with the elimination of these pathogens from blood, but also selectively augments the escape of C. albicans from phagocytosis. In summary, our model could serve as a valuable functional immune assay for recording and evaluating innate responses to infection.

Figure 1

Comparison of the dynamics of host–pathogen interaction during C. albicans and S. aureus infection in healthy blood. (a–c) Results of fitting the state-based model (SBM) and the agent-based model (ABM) to experimental data. Simulated dynamics of the combined units (solid line) were obtained by fitting the SBM (dark color) and the ABM (light color) to the experimentally measured association kinetics (dotted line). Experimental data were gained from whole-blood infection assays with either C. albicans (green, n = 10) or S. aureus (red, n = 7). SBM: line thickness represents SD obtained by 50 simulations with transition rate values sampled within their corresponding SD. ABM: line thickness represents the standard deviations obtained from 30 stochastic simulations with the estimated diffusion coefficients. (a) and (b), dynamics of the combined units P_N and P_M, which correspond to the experimental data on pathogens associated with neutrophils and monocytes. (c) kinetics of the combined unit P_E together with experimentally measured kinetics of either fungal or bacterial cells in extracellular space. (d) Mean values (± SD) of transition rate values obtained by fitting the SBM to experimental data using simulated annealing. The rate of phagocytosis by neutrophils (${\varphi }_N$) and by monocytes (${\varphi }_M$) as well as the rate for immune escape ($\rho$) are depicted for infection scenarios with either C. albicans (green circle) or S. aureus (red circle). (e) Diffusion coefficients for neutrophils $(D_N)$ and monocytes $(D_M)$ were estimated by fitting the ABM to the experimental data for C. albicans (green circle) and S. aureus (red circle), respectively. Mean and SD are calculated from all parameter sets with a mean LSE within the SD of the optimal parameter set. (f) Association of viable and inactivated C. albicans with blood monocytes and neutrophils after 60 min quantified using flow cytometry. Significance was estimated using the unpaired, two-sided Student t test (***P < 0.001). (g) Release of monocyte-derived cytokines (TNF-α, IL-1β, IL-6) in plasma samples generated from 4 h whole-blood infection experiments in response to viable and inactivated C. albicans cells was investigated. Bars are shown as means ± SD of at least 3 independent experiments with whole blood from different donors.

Figure 2

Blood after surgery shows changes in cytokine profiles and peripheral differential cell counts. Blood samples from six HLM patients taken before cardiac surgery (pre-operative), immediately after surgery (post-operative) and one day after admission to intensive care (post-operative + 1d) were analyzed for (a) cytokine levels (MIP-1α, MIF, IL-8, IL-6, IL-10, G-CSF) using Luminex technology and (b) white blood cell (WBC) as well as neutrophil, lymphocyte and monocyte counts using an automated hematology analyzer. Reference ranges of leukocytes are indicated in red. The plasma concentrations of cytokines are presented in pg/ml, except for IL-8. Levels for IL-8 in post-operative and post-operative + 1d blood samples were normalized to respective plasma levels within pre-operative blood (set to 100%) for each donor and means ± SD of the calculated percentages are shown. Significance was estimated using the unpaired, two-sided Student t test and shown as *P < 0.05, ***P < 0.001.

Figure 3

Surface phenotypes of monocytes and neutrophils after surgery indicate an increased number of immature cells in blood of HLM patients. Flow cytometry analysis of CD66b⁺ neutrophils (a) and CD14⁺ monocytes (b) from whole blood of HLM patients taken before cardiac surgery (pre-operative), immediately after surgery (post-operative) and one day after admission to intensive care (post-operative + 1d) are shown. (a) Immature phenotype of CD66b⁺ blood neutrophils was analyzed by surface expression of CD10, CD16 and CD62L. Representative zebra plots show a change of the surface phenotype by the shift of the population to the lower right quadrant for CD10 and CD16 and to the higher right quadrant for CD62L. (b) Surgery-induced changes on CD14⁺ blood monocytes are represented by zebra plots of HLA-DR and CD62L expression patterns. Left plots show proper gate setting defined by isotype control stainings.

Figure 4

Time courses of pathogen association to immune cells observed in whole-blood samples of HLM patients. Blood samples were taken before cardiac surgery (pre-operative), immediately after surgery (post-operative) and one day after admission to intensive care (post-operative + 1d). Time-resolved experimental data (dotted line) were obtained by whole-blood infection assays with either C. albicans (a–c) or S. aureus (d–f). Data points and error bars refer to the means and SD of blood samples from six HLM patients. The simulated dynamics of the combined units (solid line) were obtained by fitting the state-based model (SBM, dark color) and the agent-based model (ABM, light color) to the experimental data. The thickness of the results from the SBM represents the SD obtained by 50 simulations with transition rate values that were sampled within their corresponding SD. The thickness of the results from the ABM represents the SD obtained from 30 stochastic simulations of the ABM with the estimated diffusion coefficients. (g) Transition rate values of the SBM resulting from fitting the model to experimental data of either C. albicans or S. aureus infection in blood samples from HLM patients. The transition rate values are given for the phagocytosis rate ${\varphi }_N$ of neutrophils and the phagocytosis rate ${\varphi }_M$ of monocytes. (h) The diffusion coefficients are given for neutrophils $D_N$ and monocytes $D_M$. Mean and SD are calculated from all parameter sets with a mean LSE that lies within the SD of the optimal parameter set.

Figure 5

Results of fitting the agent-based model (ABM) to the experimental data from C. albicans and S. aureus infection using the method of adaptive regular grid search. The parameter space is shown for fitting the ABM to experimental data, where blood samples from healthy donors and HLM patients before surgery (pre-operative), immediately after surgery (post-operative) and one day after admission to intensive care (post-operative + 1d) were infected with either C. albicans (left column) or S. aureus (right column). Colors of the points refer to the weighted least squares error $E\left(\overrightarrow{p}\right)$ for each parameter set $\overrightarrow{p}=\left(D_N,,,D_M\right)$. The optimal parameter set is marked with a white dot. All parameter sets with a mean LSE that lies within the SD of the optimal parameter set are marked with a black dot.

Figure 6

Changes in cytokine secretion and innate immune cell activation in whole blood from HLM patients after surgery. Blood samples from HLM patients were taken before cardiac surgery (pre-operative, non-filled bars), directly after surgery (post-operative, light grey bars) and one day after admission to intensive care (post-operative + 1d, dark grey bars) and either mock-infected, treated with C. albicans or S. aureus for 4 h. (a) Plasma levels of TNF-α, IL-1β, IL-6 and IL-10 were quantified and bars are shown as means ± SD. Results are presented as pg/ml; N/A stands for values not available. (b) Surface marker expression was analyzed on the total monocyte population and on pathogen-associated neutrophils by flow cytometry. Data shown are mean fluorescence intensity (MFI) ± SD. Significance is shown as *P < 0.05; **P < 0.01; ***P < 0.001, unpaired, two-sided Student t test.

Figure 7

Time courses of extracellular pathogens observed in whole-blood samples of HLM patients. Blood samples were taken (a) before cardiac surgery (pre-operative), (b) immediately after surgery (post-operative) and (c) one day after admission to intensive care (post-operative + 1d). Time-resolved experimental data (dotted line) were obtained by whole-blood infection assays with either C. albicans (green) or S. aureus (red). Data points and error bars refer to the means and SD of blood samples from six HLM patients. The simulated dynamics of the combined units (solid line) were obtained by fitting the state-based model (SBM, dark color) and the agent-based model (ABM, light color) to the experimental data. The thickness of the results from the SBM represents the SD obtained by 50 simulations with transition rate values that were sampled within their corresponding SD. The thickness of the results from the ABM represents the SD obtained from 30 stochastic simulations of the ABM with the estimated diffusion coefficients. (d) Transition rate values for immune evasion $\rho$ of the SBM resulting from fitting the model to experimental data of either C. albicans or S. aureus infection in blood samples from HLM patient.