Modulation of Neutrophil Activity by Soluble Complement Cleavage Products—An In-Depth Analysis

The cellular and fluid phase-innate immune responses of many diseases predominantly involve activated neutrophil granulocytes and complement factors. However, a comparative systematic analysis of the early impact of key soluble complement cleavage products, including anaphylatoxins, on neutrophil granulocyte function is lacking. Neutrophil activity was monitored by flow cytometry regarding cellular (electro-)physiology, cellular activity, and changes in the surface expression of activation markers. The study revealed no major effects induced by C3a or C4a on neutrophil functions. By contrast, exposure to C5a or C5a des-Arg stimulated neutrophil activity as reflected in changes in membrane potential, intracellular pH, glucose uptake, and cellular size. Similarly, C5a and C5a des-Arg but no other monitored complement cleavage product enhanced phagocytosis and reactive oxygen species generation. C5a and C5a des-Arg also altered the neutrophil surface expression of several complement receptors and neutrophil activation markers, including C5aR1, CD62L, CD10, and CD11b, among others. In addition, a detailed characterization of the C5a-induced effects was performed with a time resolution of seconds. The multiparametric response of neutrophils was further analyzed by a principal component analysis, revealing CD11b, CD10, and CD16 to be key surrogates of the C5a-induced effects. Overall, we provide a comprehensive insight into the very early interactions of neutrophil granulocytes with activated complement split products and the resulting neutrophil activity. The results provide a basis for a better and, importantly, time-resolved and multiparametric understanding of neutrophil-related (patho-)physiologies.


Introduction
Neutrophil granulocytes (neutrophils) and the complement system serve as the vanguard of cellular and humoral innate immunity [1][2][3][4]. The complement cascade consists of over 50 proteins, mainly synthesized in the liver, and includes several zymogens [3,5]. Upon activation, most complement proteins are divided into specific cleavage products, which in for the first time, highly resolved multiparametric insights into neutrophil activation after stimulation with complement activation products.

Blood Sampling
Following ethical approval by the Local Independent Ethics Committee of Ulm University (#459/18), informed written consent was obtained from healthy female and male human volunteers with aged 21-38 without acute medication or recent change in chronic medication and without clinical signs of infection. Peripheral venipuncture was performed as described by the guideline of the World Health Organization [36]. Blood was collected into syringes containing 3.2% trisodium citrate or 16 IE/mL lithium-heparin (Sarstedt, Nümbrecht, Germany). All blood samples were maintained at room temperature and processed within 10 min.

Measurement of Membrane Potential, Intracellular pH, Glucose Uptake, and Cellular Shape
Purified leukocytes were used to determine MP, pH i , GlcU, and cellular shape. To this end, 15 mL whole blood was diluted with 15 mL 0.9% NaCl and gently layered upon 7.5 mL 5% dextran solution for sedimentation (30 min, room temperature) followed by hypotonic lysis (10 s of resuspension in distilled and sterile water), and the leukocytes were adjusted to 2 × 10 6 cells/mL and stored in Hank's Balanced Salt Solution with calcium and magnesium (HBSS +/+ , Thermo Fisher Scientific, Darmstadt, Germany). Exemplary results of the purification process are shown in Supplementary Figure S1. All described buffers were adjusted to pH 7.3.

PCA Reveals a Common Neutrophil Activation Pattern
To identify a general signature of neutrophil activation, a panel of eleven antibodies was applied to neutrophils stimulated by eleven activators. Given the complexity of the generated data, dimensionality reduction was performed by means of PCA. The first three PCs were able to identify 84% of the variance in neutrophil activation. The first PC explained 62% of the variance, with changes in CD11b, CD10, and CD16 being identified as the most important markers of neutrophil activity (Figure 1a, Supplementary Table S2).

Near-Real-Time Measurement of C5a-Induced Response Kinetics of Neutrophils
In total, 20 µL citrate-anticoagulated whole blood was diluted with 30 µL PBS −/− containing the following antibodies (all from BioLegend): CD11b (final concentration: 8 µg/mL), CD35 (8 µg/mL), CD10 (8 µg/mL), CD16 (16 µg/mL), CD62L (1 µg/mL), CD45 (Pacific Blue, 4 µg/mL, #368540), and CD66b (APC-Cy7, 8 µg/mL, #305126). Cells were stained in a light-protected water bath at 37 • C for 10 min. The stained blood was transferred into a 5 mL polystyrene round-bottom tube with a total volume of 1650 µL PBS +/+ . Prior to the transfer, the tube was perforated at a height of 4.5 cm using a soldering station (BASETech, Hirschau, Germany) forming a small whole with a diameter of approximately 7-10 mm. Following a resting period of 2 min in the water bath at 37 • C, the sample was placed in the manual acquisition port of the BD Lyric flow cytometer. The tube was surrounded by a temperature-controlled heating unit (#TC-124A Handheld Temperature Controller, 64-1545, Warner Instruments, Holliston, MA, USA) set to 37.3 • C. Acquisition with a medium speed was started and continued for 1320 s. After 55 s, a peripheral venous catheter (Vasofix ® Safety 1.30 × 45 mm G 18, #4268130S-01, B.Braun, Melsungen, Germany) was placed in the polystyrene tube during the acquisition. Attached to the peripheral venous catheter was a 1 mL syringe (BD Plastipak TM , #303172, Becton Dickinson S.A. Madrid, Spain) containing 150 µL stimulation mixture. Following a 60 s baseline measurement, the content of the syringe was rapidly injected into the polystyrene tube followed by a bolus of 200-300 µL air to ensure the proper mixing of the stimulants. The stimulation mixture consisted of C5a in PBS +/+ , which, after injection, resulted in a final C5a concentration of 10 nM.

Determination of the Half-Maximal Effective Concentrations
Half-maximal effective concentration (EC 50 ) was determined by fitting a Hill equation to normalized response values at various C5a concentrations (range of 10 pM-100 nM). For this purpose, the C5a-induced response was standardized by subtracting fluorescence signals of resting cells stimulated with PBS +/+ and normalizing to signals from cells stimulated with 100 nM C5a. The EC 50 was determined from a nonlinear regression fit applying a variable slope model (y = Bottom + (Top − Bottom)/(1 + 10ˆ((LogEC 50 − x) × HillSlope)) x = log of concentration, y = response) with min/max constraints (bottom = 0, top = 1). Data analysis and model fitting were performed with GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA).

Principal Component Analysis (PCA) for Dimensionality Reduction in Data from Multicolor Flow Cytometry
A training set was generated using stimulus versus control ratios of the median fluorescence intensities of neutrophils applying the formula: (stimulated/control) -1. All stimuli (PBS +/+ as control, fMLF, IL-8, PAF, C3a, C3a des-Arg, C3b, C3c, C4a, C5a, C5a des-Arg, and C3a + C5a) applied to samples from eight donors were included using the subsequent listed parameters: CD11b, CD35, CD10, CD16, CD62L, CD15, CD46, CD55, C3aR, C5aR1, and C5aR2. On the basis of the resulting dataset, the z-score of each entry within the respective parameter was calculated to normalize the input data. Subsequently, the principal components (PCs) were calculated by single value decomposition of the normalized input data (which comprised the whole training data set). On the basis of the PCs, the output values were calculated using a linear combination of the PCs and the training set. The output values were normalized, with unstimulated neutrophils set to 0% and fMLF-stimulated neutrophils to 100%. The resulting loading matrix is reported in Supplementary Table S2 and the first two PCA loading vectors can be visualized in Figure 5b.

Analysis of Kinetics of Neutrophil Activation
To analyze the near-real-time kinetics of whole blood without further fixation by multicolor flow cytometry according to the staining protocol as described above, a stepwise gating approach was conducted. First, a CD45-threshhold was established to exclude contaminating erythrocytes. Second, doublets were excluded based on the linearity of FSC-A and FSC-H. Third, neutrophils were identified as SSC high , CD16 high , CD45 high , and CD66b high cells. On the basis of this gating strategy, a median of 23.2 (21.5|27.1) neutrophils per seconds was measured ( Figure 6).     The identified neutrophils were further analyzed using a custom written, pythonbased flow cytometry analytics software "BFlow" (BFlow Project, www.bflow.science, last accessed 2 September 2022). For this purpose, a moving median analyzing nine consecutive neutrophils was calculated for each parameter. The C5a-induced change in neutrophil phenotype within the first 15 min was characterized regarding the time elapsed to the half-maximal response (Figure 7a), time to maximal slope (Figure 7b), the maximal slope relative to the baseline (Figure 7c), and the ratio of the plateau to the baseline (Figure 7d).

Statistical Analysis and Presentation
Data were analyzed using GraphPad Prism 8 and Microsoft Excel (version 16.32, Microsoft Corporation, Redmond, WA, USA). Data are reported as the median with whiskers indicating interquartile range, if not indicated otherwise. In the manuscript, data are reported as median (25th quantile|75th quantile). The present study followed a two stepped statistical approach: First, the listed stimuli were screened in blood from at least seven independent donors. Outliers were identified based on lab records (potential technical errors, such as low cell number in tube) and/or discussion of at least two authors (A.E.P.S. and D.A.C.M.). Disagreements were resolved by consulting a third author (L.W.). In the PCA, eliminated values were substituted by the mean of the particular parameter. The generated screening data were analyzed comparing the median fluorescence intensity (MFI) or the amount of % positive cells as appropriate with the respective control sample using Kruskal-Wallis test with uncorrected Dunn's test. This approach ensures conservative testing regarding missing potential positive stimuli, but also may include false-positive results due to repeated testing. To address this issue, a confirmatory dataset for the resulting hit regarding C5a was generated when generating the respective concentrationresponse curve. Here, the two groups of the results of 10 nM C5a and respective control cells were compared using the Mann-Whitney test to corroborate the findings of the screening (Supplementary Table S3). For the analysis of the resulting kinetics, the response of the neutrophils was divided into three parts (seconds in brackets refer to the total acquisition time with neutrophils being stimulated after 60 s): baseline (10-50 s), initial stimulation (60-90 s), and response kinetic (90-960 s). The baseline was defined as the median of the measurement within the seconds 10-50 s. The response kinetic was fit as a polynomial of the fifth order. The polynomial was fitted using the moving median of seconds 65-1140. However, because it is difficult to distinguish effects induced by either C5a or mixing of the cells, directly after stimulation, the seconds 60-90 (first 30 s after stimulation) were approximated by a linear curve starting from the baseline and ending at the polynomial-derived value of 90 s. Examples of the used fits are summarized in Supplementary Figure S5.

Statistical Analysis and Presentation
Data were analyzed using GraphPad Prism 8 and Microsoft Excel (version 16.32, Microsoft Corporation, Redmond, WA, USA). Data are reported as the median with whiskers indicating interquartile range, if not indicated otherwise. In the manuscript, data are reported as median (25th quantile|75th quantile). The present study followed a two stepped statistical approach: First, the listed stimuli were screened in blood from at least seven independent donors. Outliers were identified based on lab records (potential technical errors, such as low cell number in tube) and/or discussion of at least two authors (A.E.P.S. and D.A.C.M.). Disagreements were resolved by consulting a third author (L.W.). In the PCA, eliminated values were substituted by the mean of the particular parameter. The generated screening data were analyzed comparing the median fluorescence intensity (MFI) or the amount of % positive cells as appropriate with the respective control sample using Kruskal-Wallis test with uncorrected Dunn's test. This approach ensures conservative testing regarding missing potential positive stimuli, but also may include false-positive results due to repeated testing. To address this issue, a confirmatory dataset for the resulting hit regarding C5a was generated when generating the respective concentration-response curve. Here, the two groups of the results of 10 nM C5a and respective control cells were compared using the Mann-Whitney test to corroborate the findings of the screening (Supplementary  Table S3).

Differential Changes in Neutrophil Cell Physiology upon Stimulation with Various Complement Cleavage Products
In general, the activation of neutrophils is linked to changes in MP, pH i , cellular shape, and GlcU, as shown previously [8][9][10][11][12]41]. Using neutrophils within a population of purified leukocytes, the screening of various complement activation products revealed a unique response profile for C5a and C5a des-Arg, whereas the C3 cleavage products (C3a, C3b, C3c, and C3a des-Arg) and C4a had no detectable effect at a concentration of 100 nM (Figure 1). At 10 nM, C5a and C5a des-Arg elicited a depolarization of the neutrophil MP comparable 10 µM fMLF (Figure 1a). The addition of 100 nM C3a to 10 nM C5a did not further enhance (or decrease) cellular depolarization compared to C5a alone. Similarly, C5a and C5a des-Arg increased the pH i , the cellular size as measured by FSC, and the glucose uptake (Figure 1c,e,g). Time courses of C5a-induced changes in cellular physiology followed a clear time pattern, which was similar to the response after stimulation with fMLF (Figure 1b,d,f). The comparisons of the magnitude of the complement-mediated cellular response is summarized in Figure 1a,c,e at the time points of the peaks for fMLF stimulation. The response of neutrophils to all measured complement cleavage products were analyzed after 1, 5, 10, 30, and 60 min. However, no further relevant response was captured during this period (data not shown). Neutrophil activation by C5a regarding MP, pH i , FSC, and GlcU was concentration-dependent ( Figure 1h). Of note, a higher EC 50 concentration of C5a was necessary to induce depolarization (6.8 nM, 95% CI 4.7-9.8 nM), when compared (p = 0.008, Mann-Whitney test, analyzing samples individually shown as summarized in Figure 1h) to the EC 50 determined for alkalization (1.3 nM, 95% CI 1.0-1.6 nM).

The Impact of Complement Cleavage Products on Neutrophil Phenotype as Well as Altered Cellular Effector Functions
Neutrophil activation is defined by altered cellular effector functions and reflected by changes in the surface expression of activation markers and complement receptors as well as altered cellular effector functions [23,42]. The analysis of activation markers and cellular effector functions revealed a marked response of neutrophils in whole blood to C5a and C5a des-Arg for indicators associated with chemotactic activity, including CD10 upregulation (Figure 2a), CD15 upregulation (Figure 2b), and CD62L downregulation (Figure 2c). None of the screened complement cleavage products enhanced PNC formation (Figure 2g). C5ainduced increased ROS production, although at a lower level as by either fMLF or PAF (Figure 2d). C5a and C5a des-Arg induced increased the expression of the phagocytosis receptor CD16 and an enhanced phagocytotic activity (Figure 2f,e). As expected, the latter was also increased after exposure to C3b, known as the classical opsonin. All C5a-induced changes were unaffected when co-stimulating neutrophils with C3a and C5a.

Modulation of Complement Receptors and Regulators on Neutrophils by C5-Cleavage Products
Subsequently, the impact of complement cleavage products on the surface expression of complement receptors and regulators on neutrophils in whole blood was analyzed. Both C5a and C5a des-Arg increased CD11b, CD35, CD46, and CD55 expression, while both decreased the expression of C5aR1 expression (Figure 3). Of note, C5a-induced alterations of the MFI did not greatly affect the percentage of positive neutrophils within the respective marker as summarized in Supplementary Table S1. For example, neutrophil stimulation with C5a reduced CD62L surface expression, but the majority of neutrophils remained positive for this marker as indicated by respective isotype control staining. C3aR expression remained unaffected after stimulation with either complement cleavage products or fMLF, IL-8, or PAF (Figure 3c). Figures 1-3 report the statistical results of the screening procedure. As indicated in the methods section, the findings regarding C5a were confirmed using a second dataset reported in Supplementary Table S3. Figure 4 summarizes the relative change in neutrophil cell physiology, activation markers, effector functions, complement receptors, and regulators after stimulation with complement cleavage products or respective positive controls. Of note, the simultaneous stimulation of neutrophils by C3a and C5a neither increased nor decreased neutrophil activation beyond the level of stimulation with C5a alone.

PCA Reveals a Common Neutrophil Activation Pattern
To identify a general signature of neutrophil activation, a panel of eleven antibodies was applied to neutrophils stimulated by eleven activators. Given the complexity of the generated data, dimensionality reduction was performed by means of PCA. The first three PCs were able to identify 84% of the variance in neutrophil activation. The first PC explained 62% of the variance, with changes in CD11b, CD10, and CD16 being identified as the most important markers of neutrophil activity (Figure 5a, Supplementary Table S2). Figure 5b visualizes the first two principal component vectors of the loading matrix, summarizing how each characteristic parameter influenced the respective principal component. The loading summarizes surface antigens, which responded similarly upon stimulation with complement cleavage products. Smaller angles between vectors indicate the similar behavior of the respective surface marker.
Orthogonal angles indicate no dependence of the respective antigens. The longer the vector in the x-axis and/or y-axis, the stronger is its influence in PC1 and/or PC2, respectively.

In-Depth Analysis of the C5a-Induced Response
Finally, a detailed investigation of the C5a-induced cellular activation pattern was conducted analyzing the concentration-dependency and the kinetics in a near-real-time manner. All C5a-induced effects displayed a concentration-response relationship.
The C5a-induced response of neutrophils demonstrated a well-defined time pattern ( Figure 6). All measured responses (CD11b, CD35, CD10, CD16, FSC, and CD62L) started within minutes after stimulation with C5a. However, while CD35, CD11b, CD10, and CD16 changed gradually, reaching a plateau after 10-15 min, the response of FSC and CD62L were much more rapid and attained a plateau within the first 5 min. Comparing the C5a-induced response in regard to the time (for example, until reaching its maximal slope, time to complete half of the maximal change, maximal slope, and its overall ratio), it revealed highly uniform responses in neutrophils, irrespective of the donors investigated ( Figure 7). Among the analyzed surface markers, CD11b displayed the most rapid and strongest response in neutrophils after exposure to C5a.

Discussion
In the present study, C5a and C5a des-Arg elicited a rapid, multivariate response in neutrophils in contrast to C3a, C3a des-Arg, C3b, C3c, and C4a, which did not induce such effects. The predominant role of C5a and C5a des-Arg is in accordance with previous findings [8,10,24,41,42]. The response of neutrophils induced by C5a and C5a des-Arg was comparable to effects elicited by fMLF or PAF single readouts as well as their combination in principal components. However, neither C5a nor C5a des-Arg promoted PNC formation, which implicated that a platelet-neutrophil interaction is not a hallmark of complement activity and, therefore, is likely to be neither a diagnostic marker nor largely involved in complement-mediated diseases, at least not in the whole blood assay used here. Interestingly, a study using isolated neutrophils, platelets, and serum found that complement activation appears to be involved in PNC formation on endothelial cells [43].
The C5a-induced response of neutrophils was further assessed and confirmed to be concentration-dependent with EC 50 s ranging over approximately 10 -8 -10 -9 M. This finding is accordance with other studies reporting an EC 50 for C5a-binding to C5aR1 of approximately 1 nM and to C5aR2 of approximately 3-10 nM [19,37,38]. Previous studies reported C5a concentrations of approximately 10 nM during inflammation [19,39,44,45]. Interestingly, a higher EC 50 concentration of C5a was determined to induce depolarization when compared to alkalization. A current widely accepted hypothesis is that neutrophil depolarization is associated with the activity of the NADPH oxidase (NOX), which is crucially involved in ROS generation [13,[46][47][48][49]. NOX activity results in electron extrusion, which must be compensated, which is likely mediated by proton extrusion via voltagegated proton channels and/or sodium-proton exchanger 1 (NHE1) activity [13,46,49]. However, surprisingly, C5a-induced alkalization occurred at lower concentrations than the depolarization. Currently, we cannot exclude that this is due to limitations of the measurement methods, for example, the response time of the used fluorescent probes to monitor depolarization and alkalization. Further studies with direct methods, for example, combining patch-clamp with the measurement of extracellular proton concentrations, are warranted to elucidate this issue.
Concomitant with a previous study [9], this study analyzes in-depth the rapid kinetic of the C5a-induced response. Interestingly, the C5a-induced response of neutrophils appears to have three response types: For depolarization, there is an initial peak within the first 2 min, with a return to baseline after 5 min. Alkalization, changes in FSC, and CD62L shedding can be observed starting within the first minutes, reaching plateaus after approximately 5 min. By contrast, the C5a-induced CD11b, CD35, CD10, and CD16 upregulation occurred more gradually and was delayed. In this context, further studies need to clarify whether these dynamics are similar using lower C5a concentrations and to link the different kinetics observed in cellular physiology and phenotype to specific functional consequences.
This study has several strengths and limitations. The role of C5a is highlighted synchronically by analyzing the response of neutrophils regarding cellular biological parameters and changes in cellular phenotype. Although the findings were generated solely by flow cytometry, this is a well-established and clinically applied method to analyze neutrophils. In addition, our findings are in accordance with previous studies, for example, also reporting C5a but not C3a-related effects on neutrophils [21,50]. Moreover, flow cytometry allows the non-invasive analysis of non-adhering neutrophils. Most readouts were captured using diluted whole blood, which allowed the stimulation of the neutrophils after venipuncture. This decision was based avoiding purification steps that might artificially activate neutrophils and to take the short lifespan of neutrophils into account. Moreover, this ensured (to a certain degree) the presence of plasma proteins potentially interacting with complement cleavage products, thereby more closely mimicking the physiological situation. For MP, pH i , FSC, and GlcU, the isolation of leukocytes was chosen because the removal of erythrocytes by lysing agents, for example, ammonium chloride, might interact with cellular parameter, in particular the pH i . The further purification of neutrophils within the leukocyte population was omitted to reduce potential artificial activation and delay due to the isolation process. Of note, GlcU was determined indirectly using the fluorescent dye 2NBDG, the validity of which as a surrogate for GlcU was recently challenged [51]. Further studies need to confirm some specific findings by other methods, for example, depolarization by the patch-clamp technique, cellular activation by analyzing humoral activation markers of neutrophils, or alterations of cell metabolism by metabolic flux analysis. Furthermore, we analyzed blood from young healthy volunteers, thereby potentially overlooking the effects of certain pre-existing complement activation during acute and/or chronic comorbidities. Moreover, our analysis focused on central complement cleavage products; however, we neither analyzed all complement proteins nor reflected the fact that complement system activation results in a parallel increase in the concentration of several complement cleavage products, nor were complement components bound to any surfaces (which would provide clustering of C3 opsonins). In addition, this study used an innovative approach to analyze the response of neutrophils in near-real-time. Finally, the conducted PCA revealed a specific change in the neutrophil phenotype upon activation. The corresponding loading values of the used activation markers and their resulting vectors on the loading plot contribute to a better understanding of hallmarks of neutrophil activity and to identify potential relevant and redundant markers for the immunomonitoring of complement-driven neutrophil activation.

Conclusions
C5a and C5a des-Arg but neither C3a nor C4a induced a multiparametric response in neutrophils as demonstrated for cell physiological parameters, neutrophil phenotype, and cellular effector functions. The findings including the reported EC 50 values of C5a and the C5a-induced response kinetic of neutrophils have implications for complement-driven pathologies, for example, regarding the design and evaluation of an intervention. Further studies need to assess the findings in other cell types, such as other leukocytes and platelets.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cells11203297/s1, Supplemental Table S1. Summary of the neutrophils positive for the used surface markers; Supplemental Table S2. Principal component analysis; Supplemental Table S3. Statistical analysis for C5a-induced effect on neutrophils; Supplemental Figure S1. Representative result of the leukocyte isolation procedure and summary of the applied gating strategy; Supplemental Figure S2 Data Availability Statement: All data will be made available upon reasonable request by emailing the corresponding author.