Neutrophils are vital in the host response to infection and inflammation. Neutrophil dysfunction is described at the extremes of age, in overwhelming infection, in chronic diseases and is associated with poor outcomes for patients (1, 2). In sepsis and other disease states there is broad dysfunction with all neutrophil functions reduced (3, 4). Ineffective neutrophil responses are harmful during the acute response where dysfunction may drive increased mortality and morbidity (5, 6). Then in the resolution phase persistent neutrophilic responses potentiates tissue damage as seen in acute respiratory distress syndrome (ARDS) (7) and unopposed neutrophil action in alpha-1 anti-trypsin deficiency (8). In patients with ARDS or sepsis, studies of neutrophil extracellular traps (NETs) demonstrate that there is an optimal level of ex-vivo function with poor outcomes for those with the highest or lowest levels of NETosis (6, 9). While alterations to a single cell function suggests involvement of a single molecular pathway, broad cellular dysfunction implicates a central driver such as altered metabolism. Immunometabolism is defined as the role of cellular metabolism in the regulation of immune cell function (10).

Murine work by Sadiku and colleagues has demonstrated that regulation of neutrophil function is linked to metabolism (11). Neutrophils are highly glycolytic (12) and inhibition of glycolysis using 2-deoxyglucose (2-DG) significantly reduces migration, phagocytosis, NETosis and oxidative burst. Neutrophils possess a functional network of mitochondria; however, mitochondrial oxidative phosphorylation does not appear to be required for adenosine triphosphate (ATP) synthesis to meet the energetic demand of neutrophils (13–15). The bulk of ATP is synthesised by glycolysis (16, 17). Sadiku et al. demonstrated that neutrophils from patients with chronic obstructive pulmonary disease had lower baseline ATP levels and exhibited reduced glycolysis in response to a pathogen compared to age matched controls (18). Moreover, the glycolytic capacity of neutrophils can be impacted by treatment with lipopolysaccharide (LPS) (19). Developing methods to understand neutrophil metabolism may provide insights into mechanisms underpinning disease and could offer a new therapeutic paradigm, improving outcomes for patients with acute and chronic disease.

Extracellular Flux (XF) analysis assesses real-time measurement of oxygen consumption rate (OCR) and proton efflux rate (PER) of cells. XF technology allows automated addition of different stimuli or inhibitors with flexibility over timing to visualise the effect of stimulation and inhibitory factors on cellular metabolism. XF technology requires low cell numbers relative to other forms of respirometry and allows for simultaneous measurement of PER and OCR as indicators for glycolysis and mitochondrial function, respectively. The streamlined workflow combined with user-friendly interface and analysis templates have seen a worldwide boom in the use of these instruments for probing cellular metabolism. In this manuscript we describe optimised protocols for using an XFe96 XF Analyser to probe (i) glycolysis in neutrophils under basal and stimulated conditions and (ii) phorbol 12-myristate 130-acetate (PMA)-induced oxidative burst. We highlight limitations of using XF technology to examine mitochondrial function in neutrophils. Finally, we provide an overview of how to analyze XF data and identify pitfalls of probing neutrophil metabolism with XF analysis.

Neutrophils are highly glycolytic cells (12); inhibition of glycolysis significantly impacts neutrophil effector functions (13). Neutrophils utilise glycogen stores to synthesise ATP in the absence of oxygen and are capable of gluconeogenesis to increase glycolytic capacity (11). The use of glycolytic inhibitors abolishes neutrophil ATP levels, while they remain unchanged after inhibition of mitochondrial respiratory complexes (33).

Activated neutrophils rapidly consume oxygen as reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase forms superoxide anions, reactive oxygen species and hydrogen peroxide, a process commonly referred to as oxidative burst. PMA is a protein kinase C activator and a robust activator of neutrophils, albeit non-physiological (26). In cell seeding densities beyond 5 x 10⁴ neutrophils/well, we saw a reduction in the size of the oxidative burst response. Typically, it would be anticipated that greater cell seeding densities would produce a larger oxidative burst response, however, due to cell overcrowding and/or reduced PMA stimulation relative to number of cells, this was not the case. We demonstrated that 5 x 10⁴ neutrophils/well combined with an injection strategy incorporating 2 µM rotenone plus 2 µM antimycin A followed by 160 nM PMA, produces a substantial and consistent oxidative burst response, without maximising the neutrophil reactive oxygen species (ROS) response. This allows for detection of both elevated and suppressed oxidative burst responses caused by disease states or exogenous treatments. We show that PMA concentrations lower than 160 nM do not produce a consistent peak following PMA injection, and therefore area under the curve analysis is unable to quantify a reliable oxidative burst response. Data in Figures 3 , 4 demonstrate that when assessing oxidative burst in neutrophils a multi-parameter approach should be taken as using a single parameter may lead to false assumptions as seen in the variability of time to max oxygen consumption (2B and 3A) which is due to detection of very small peaks at extremes of cell density and or PMA concentration, and are not likely to be a true difference in time to maximum oxygen consumption. PMA is a potent and non-physiological activator of neutrophils; therefore, we explored the utility of other stimuli in the oxidative burst assay. We demonstrated that priming with TNFα enhances detection of the increased oxygen consumption induced by fMLP and mtDAMPs. In unprimed cells there was a small increase in oxygen consumption following fMLP/mtDAMPs but this was well below the recommended limits of detection for OCR. Heat killed SP, NTHI, or LPS did not induce a detectable increase in oxygen consumption with or without priming. Heat killed bacteria and LPS have been used by others to induce oxidative burst, although with different methods (27, 28).

More commonly used methods to measure oxidative burst detect generation of ROS by a range of fluorescent, chemiluminescent probes rather than detecting environmental oxygen consumption (34). While these methods can be highly specific this can also be considered a weakness if assay conditions are not carefully controlled. Different probes have varying selectivity for different species of ROS and vary in their ability to detect intra and extracellular ROS (34), this is frequently an issue in commercially available kits where the selectivity is not always clearly documented. In addition, chemiluminescent probes such as luminol and lucigenin are prone to errors related to redox recycling as the probes can generate radicals that produce superoxide (35). Using the XF analyser to assess oxidative burst allows for measurement of the total oxygen uptake from the environment which may better represent total ROS generation, especially when mitochondrial uptake of oxygen is abolished. This assay typically uses half the cell number of chemiluminescent assays and enables simultaneous indication of glycolysis through PER data.

Neutrophils rely on glycolysis for synthesis of ATP. In the traditional glycolytic rate test (22) used in other cell types, injection of oligomycin (ATP synthase inhibitor) or rotenone and antimycin A pushes glycolysis to its capacity, as cells are unable to produce ATP via mitochondrial respiration and therefore upregulate glycolysis to meet the energetic demands of the cell. When testing this assay on human neutrophils, we observed no change in glycolysis after oligomycin injection. A recent publication has highlighted that by inhibiting mitochondrial respiration using mitochondrial respiratory complex inhibitors in resting state cells, cells only shift to glycolysis as a compensatory mechanism for basal ATP demand requirements, thus these agents may not actually shift cells to maximum glycolytic capacity as previously thought (36). This confirms our findings, highlighting the need for a stimulant to increase ATP demand. Here we used 160 nM PMA, to increase glycolytic activity in neutrophils and assess downstream glycolytic parameters. Seeding more than 2 x 10⁵ neutrophils per well to probe glycolytic activity resulted in inconsistent and lower glycolytic responses. Similarly, to the oxidative burst test, this is likely due to overcrowding of cells in the bottom of the well, which compromises the monolayer required to run these assays.

The manufacturer states a limit of detection of 20 pmol O₂/min/well for basal OCR from cells assessed with the XFe96 and XFp analyzers. OCR below this threshold cannot be interpreted reliably and should be used with caution to infer mitochondrial function. Here, neutrophil mitochondrial metabolism is too low to detect using current XF analyzers – OCR are below the limit of detection and do not change in response to mitochondrial stressors. Moreover, this was not overcome by increasing cell seeding density. In fact, our data showed that increases in cell seeding density yields amplified the low levels of contamination by mononuclear cells which appear to skew mitochondrial respiration tests using XF analysis. These data highlight the importance of using highly pure neutrophil preparations when performing metabolic assays and suggest caution to be taken when inferring mitochondrial respiration from OCR of neutrophil preparations.

Of interest, in two recent studies, a typical mitochondrial OCR trace using XFe96 technology was reported in neutrophils seeded between 0.5-1.5 x10⁵ cells/well with isolation purities >95% (19, 37). At similar cell seeding densities, in our study we were unable to achieve equivalent traces. Such responses were only apparent with much less pure (<70%) neutrophil preparations. We show that basal OCR are higher in preparations with greater PBMC contamination at levels typically deemed acceptable for neutrophil assays (7E), however these data should be interpreted with caution given the lower limit of detection for XF analyzers. Given that PBMCs have higher rates of mitochondrial respiration than neutrophils, our data demonstrated that contamination with PBMCs has significant implications when assessing mitochondrial respiratory parameters from isolated neutrophils. Notably, in both papers, basal OCR are below the optimal minimum detection threshold for the XFe96 analyser. Additionally, the results as presented are problematic. For example, Porter et al. show higher OCR in neutrophils after rotenone and antimycin A injection than basal OCR, suggesting a negative basal rate of mitochondrial respiration (38). Unless non-mitochondrial oxygen consumption is stimulated in such experiments, for example by increased ROS, such finding is not theoretically possible. Nonetheless, if increased ROS are responsible for such increases in OCR this needs to be corrected to accurately calculate parameters of mitochondrial function.

Consistent with our findings, Chacko et al. show that PER is not responsive to oligomycin in neutrophils (39). Moreover, neutrophil OCR is not sensitive to mitochondrial compounds, corroborating our findings and agreeing with the central dogma for a minimal role of mitochondrial oxidative phosphorylation in neutrophil energy generation. Our data, combined with existing evidence, suggests that mitochondrial ATP synthesis is below the limit of detection for XF analyzers, and thus play an insignificant role in supplying ATP to meet basal energetic demands in mature neutrophils. Mitochondrial ATP synthesis does appear to be required for messenger purposes (13, 40). Further, mitochondrial respiration appears to be important in supporting oxidative burst in the absence of glucose by providing NADPH oxidase (41).

XF analyzers account for the pH and oxygen content in background wells, which contain no cells. This enables background correction of the values to ensure readouts represent actual metabolism rather than changes in ambient O₂ concentration or pH fluctuations affecting media parameters. For adherent cell cultures, normalisation is required to account for well-well cell density differences across the plate. When using neutrophils, cells are plated from pre-diluted homogenous cell suspensions at defined cell seeding densities per well and therefore normalisation to cell number, protein content or DNA will have modest effects on OCR or ECAR. This is, of course, if even cell seeding across the plate is obtained by the experimenter. Reporting ECAR values (traces or extrapolated values) are no longer accepted as representative for glycolysis. Media often has a large buffering capacity, which by its nature can mask pH changes, and subsequent acidification rates. Therefore, by calculating and applying a buffering factor for each specific media to account for the buffering capacity, ECAR values can be converted to PER, a more representative readout for glycolytic tests (when corrected for any mitochondrial-linked PER).

XF analyzers allow for high throughput data and multiple metabolic tests to be performed on a single assay plate, with a minimal demand for cell numbers. The dynamics of oxygen consumption and change in pH over the duration of the assay allows for the depiction of cellular metabolism and metabolic responses over time. XF analyzers accurately predict lactate production compared to direct lactate measurement (42). Other methods to assess glycolysis include direct measurement of glucose and or lactate in assay media, use of glucose analogs, measurement of rate limiting glycolytic enzymes, glucose tracing and metabolite quantification. Each of these has strengths and limitations, we recommend a combined approach to interrogation of metabolism for robust data. These are discussed in detail in (31). Alternative respirometers such as the Oroboros O2k analyser require 2mL of solution, with a recommended 2-4 million cells (based on HL-60 data) per chamber (43, 44), more than 40x the requirement for a well of a XFe96 cell microplate. XF technology also incorporates a pH probe to allow the determination of the glycolytic pathway alongside oxygen consumption simultaneously, without the need for additional modules. As mentioned previously, neutrophils are notoriously easy to activate by mechanical, chemical, and biological stressors. Respirometers such as Oroboros allow for inhibitor/stimulant injections but operate using continuous stirring motions, which may activate neutrophils, skewing metabolic readouts. Adherent cells lend themselves naturally to XF assays, as they can be cultured in a monolayer directly in the cell plate. By incorporating a molecular adhesion step, cells in suspension can be manipulated into a monolayer to perform metabolic assays without exposing them to rigorous mixing. It has been demonstrated previously that plate adhesion using Cell-Tak™ does not affect neutrophil ROS production (37).

XF analyzers are highly sensitive; however, care must be taken in ensuring an evenly distributed monolayer of cells is formed. If cells are not distributed directly under the probe, or if they begin to detach from the base of the well due to disruptions, probes can lack fine discrimination. Further, introduction of pharmaceutical treatments or other stimulants/inhibitors requires further assay validation to ensure they do not interact with the probes themselves, which are sensitive to pH. Despite these limitations, we have demonstrated how XF technology can be harnessed for neutrophil bioenergetic analysis. XF analyzers are costly to run compared to other respirometers, however the high throughput and automated nature of the system allows for efficient data collection.

In conclusion, we have presented optimised protocols for the assessment of neutrophil glycolysis and oxidative burst using XF technology. The metabolic tests outlined here can be applied to any project context by incorporating pharmacological treatments, disease state neutrophils or neutrophils from different developmental stages. We demonstrate robust methodology to achieve neutrophil monolayers to perform metabolic assays for assessing glycolytic rate and oxidative burst. We highlight the challenges of these assays and mitigating strategies when assessing mitochondrial respiration. Overall, XF technology represents a powerful platform with a user-friendly interface and data analysis templates which has the potential to revolutionise our understanding of neutrophil metabolism. The protocols described in our manuscript provide a template to enhance discovery and ensure a standardised approach to neutrophil measurements using XF analysis.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by University of Birmingham Research Ethics Committee (ERN 12-1185R2). The patients/participants provided their written informed consent to participate in this study.

AJ and FG performed all experiments. JB, AS, AJ and FG designed the study with critical review by DT, ES and CM. Data analysis was undertaken by AJ, FG, and JB. All authors contributed to the article and approved the submitted version.