Study: Non-Classical monocytes display inflammatory features: Validation in Sepsis and Systemic Lupus Erythematous

Monocytes are a group of immune cells that originate in bone marrow and are released into peripheral blood, where they circulate for several days^1,2. They belong to the mononuclear-phagocyte system, which also include macrophages, dendritic cells and their bone-marrow precursors^3,4,5. Monocytes represent 5–10% of peripheral leucocytes and are probably best known for serving as a systemic reservoir of myeloid precursors that are needed for the renewal of tissue macrophages and dendritic cells^6,7,8,9. However, they also have other well documented functions in immune response against infection^10,11,12,13 and in pathogenesis of several inflammatory disorders. Although initially perceived as a homogeneous population, it has become increasingly apparent that monocytes display considerable heterogeneity with respect to their phenotype and function^1,2,14,15.

Abstract

Given the importance of monocytes in pathogenesis of infectious and other inflammatory disorders, delineating functional and phenotypic characterization of monocyte subsets has emerged as a critical requirement. Although human monocytes have been subdivided into three different populations based on surface expression of CD14 and CD16, published reports suffer from contradictions with respect to subset phenotypes and function. This has been attributed to discrepancies in reliable gating strategies for flow cytometric characterization and purification protocols contributing to significant changes in receptor expression. By using a combination of multicolour flow cytometry and a high-dimensional automated clustering algorithm to confirm robustness of gating strategy and analysis of ex-vivo activation of whole blood with LPS we demonstrate the following: a. ‘Classical’ monocytes are phagocytic with no inflammatory attributes, b. ‘Non-classical’ subtype display ‘inflammatory’ characteristics on activation and display properties for antigen presentation and c. ‘Intermediate’ subtype that constitutes a very small percentage in circulation (under physiological conditions) appear to be transitional monocytes that display both phagocytic and inflammatory function. Analysis of blood from patients with Sepsis, a pathogen driven acute inflammatory disease and Systemic Lupus Erythmatosus (SLE), a chronic inflammatory disorder validated the broad conclusions drawn in the study.

Introduction

In humans, monocytes have been divided into three subtypes based on relative surface expression of LPS co-receptor CD14 and FCγIII receptor CD16^16,17. The most predominant of the three, termed “classical monocytes”, express high levels of CD14 on their surface, are devoid of surface CD16 and account for approximately 80% of the total monocyte population. The remaining 20% express CD16 and have been further classified into two subtypes. The more abundant “nonclassical monocytes”, are characterized by very low expression of surface CD14 and high levels of CD16, whereas the third monocyte subtype, called “intermediate monocytes”, express high levels of both the receptors^16,17.

Over the years, a number of studies, often contradictory, aimed at functional characterization of the three monocyte subsets have been undertaken by different research groups^{18,19,20,21,22,23,24,25,26}, which has sparked an interest to investigate proportions of monocyte subsets in a wide variety of diseases. Relative percentages of monocyte subsets has been reported to understand pathogenesis of several infectious and metabolic diseases viz., sepsis^27,28,29, chronic liver disease^30,31,32, rheumatoid arthritis³³, atherosclerosis³⁴, filariasis³⁵ and obesity³⁶. However, lack of consistent gating strategies for quantification by flow cytometry has been a major limitation, leading to lack of reproducibility. Since CD16 is highly expressed on neutrophils and NK cells, these cell types need to be excluded from analysis for reproducible quantification and to avoid confounders. The other major limiting factor is that all functional studies so far have been performed with purified monocytes, contributing to highly variable results depending on method of isolation employed³⁷.

In this study, we used a combination of multicolour flow cytometry, multiparameter imaging cytometry and a high-dimensional automated clustering algorithm to confirm robustness of our gating strategy. Further, we have addressed these controversies on analysis of functional phenotype of human monocyte subsets in circulation. We demonstrate that gradient purified monocytes are significantly different from cells analysed in whole blood for receptor expression. Our findings suggest that ficoll purification of blood monocytes leads to a decrease in number of CD14+/CD16- classical monocytes with a concomitant expansion of CD14dim/CD16+ nonclassical monocytes. Moreover, gradient separation also contributed to increased surface expression of CD16. Therefore, in order to retain physiologically relevant monocyte function, we adopted whole blood stimulation and staining protocol. Using this method we show distinct phenotype and function of monocyte subsets in circulation with respect to inflammation, antigen presentation and phagocytosis. We unequivocally demonstrate that nonclassical and not intermediate subtype are the primary inflammatory monocytes and further validate by studying patients suffering from sepsis and systemic lupus erythematous (SLE).

Discussion

Existence of heterogeneity in human peripheral blood monocytes based on size and density was reported almost three decades ago, showing that human monocytes in circulation consisted principally of two subtypes – a large and a small sized one – that differed in their phagocytic and inflammatory capabilities^{42,43,44,45,46,47}. Later, with the advent of flow cytometry, a more robust identification of blood monocyte subsets was performed based on differential expression of surface CD14 and CD16⁴⁸. This initial study showed existence of at least two distinct monocyte subsets – CD14+/CD16- and CD14+/CD16+- that were strikingly different in their function. Subsequently, the CD16+monocytes were shown to be further composed of two different populations¹⁴. One was shown to express equal levels of CD14 and CD16 (CD14+/CD16+), while the other population was characterized by very low surface expression of CD14 (CD14dim/CD16+).Given the importance of monocytes in immune function and their role in pathogenesis of a wide variety of diseases, it became essential to precisely characterize human blood monocyte subsets. To this end, a large number of studies were undertaken by different research groups with a view to functionally characterize blood monocyte subsets^{18,19,20,21,22,23,24,25,49}. However, there exists little consensus among most of published literature. One of the reasons is lack of a consistent gating strategy for immunophenotyping by flow cytometry. This issue was addressed by a recent review that proposed a common gating strategy to identify monocyte subpopulations in human circulation¹⁶. Accordingly, in the current study, a negative gating strategy was used to sequentially exclude neutrophils, NK cells, B cells and T cells from analysis followed by inclusion of HLA-DR+++monocytes that were further discriminated on a bivariate scatterplot of CD14 vs. CD16 to finally yield three clearly distinguishable monocyte subpopulations. According to proposed nomenclature^16,17, they were classified as ‘classical’ (CD14+/CD16-), ‘intermediate’ (CD14+/CD16+) and ‘nonclassical’ (CD14dim/CD16+). In accordance with previously published reports, classical monocytes were the most numerous, constituting about 80–90% of blood monocytes, with intermediate and nonclassical subtypes making up the rest.

A contentious issue with analysis of high dimensional flow cytometry data is the use of manual gating which is subjective and relies on visual inspection that varies between users leading to errors^50,51. This problem is especially evident during identification of rare cell types. Therefore, to validate our gating strategy we employed an automated clustering algorithm called spanning tree progression analysis of density-normalized events (SPADE)³⁸. SPADE involves density-dependent downsampling of raw flow cytometry data followed by agglomerative clustering based on relative expressions of different cellular antigens. This leads to construction of minimum spanning trees connecting the clusters that allows easy visualization of rare events. The similarities between bivariate plots created through SPADE and manual gating validated our manual gating strategy. A further confirmation of robustness of manual gating was obtained through high resolution imaging cytometry in which each individual population of cell was visualized and were clearly distinguishable via a combination of cell surface marker expression and nuclear morphology.

A growing body of opinion suggests that the primary reason for discrepancies between previously published reports on monocyte subset function is use of in vitro purified monocytes in culture and that whole blood stimulation and analysis could be a better alternative^39,40. A number of studies have shown benefits of whole blood culture and stimulation over purified cells for in vitro analysis of immune cell function^{37,52,53,54,55}. In the present study, we observed that purification of peripheral blood mononuclear cells (PBMCs) by ficoll-density gradient centrifugation led to considerable changes in relative percentages of monocyte subtypes – classical monocytes decreased with a simultaneous increase in percentage of nonclassical monocytes. When compared with whole blood analysis, our results provide evidence to the notion that gradient purification of monocytes can lead to experimental artefacts that can confound analysis of monocyte function.

Analysis of surface molecules on monocyte subpopulations revealed differential expression of toll-like receptors (TLRs), scavenger receptors and co-stimulatory molecules among the three monocyte subsets. Intermediate and nonclassical monocyte subsets expressed more TLRs 2, 4, 5, co-stimulatory molecules CD80, CD86 and HLA-DR than the classical subset, suggesting their role in antigen presentation. On the other hand, higher expression of scavenger receptors CD36 and CD163 on classical monocytes was suggestive of their predominantly phagocytic function. Our observation that CD80 and CD86 is differentially expressed among monocyte subsets is in disagreement with an earlier published report²³ that showed no difference in expression of CD80 and CD86 in purified monocyte subsets. Interestingly, the observed difference in expression of these two molecules within monocyte subsets was no longer present following stimulation of monocytes with LPS, suggesting that the earlier observation may have been a consequence of activation during purification process.

A constant source of controversy regarding function of subsets of human monocyte the type of cytokines produced by them. While inflammation is a complex phenomenon involving several receptors, mediators and pathways, monocytes that synthesize molecules such as TNF-α, IL-1β, IL-6 etc. have been classified and designated as ‘inflammatory subtype’ by us in this manuscript. The current study demonstrates that nonclassical monocytes are the primary producers of TNF-α and IL-1β upon activation as shown by intracellular cytokine staining. This observation is markedly different from an earlier observation²¹ that classified intermediate monocytes to be primarily responsible for inflammatory cytokine production while the non-classical monocytes showed patrolling behaviour. It is pertinent to note that these authors had used bead purified monocytes in vitro and measured released cytokines in supernatants. Another interesting observation was increased IL-10 production by intermediate subsets, which differs from earlier studies that demonstrated classical monocytes as principal producers of IL-10^21,23. While whole blood assays performed in this study is closer to physiological condition (unlike bead purified monocyte subsets) the discrepancy between earlier studies^21,23 and our current observation could also be due to difference in the assay system adapted for measuring cytokines. While the previous reports had measured secreted cytokines by ELISA in the current study we have quantified intracellular cytokines by flow cytometry. Intracellular cytokines do not necessarily correlate with secreted cytokines, particularly in the case of IL-1β secretion which is tightly regulated. Thus the observed differences in TNF-α, IL-1β and IL-10 positive monocyte subsets could be a result of the two different assay systems used in the investigations. Another probable cause could be a result of observations made at different time points as most studies reported in literature were all 18 hour stimulations followed by analysis of cytokines whereas our observations were at much earlier times points.

There was concordance between scavenger receptor expression and functional phagocytosis in different monocyte subsets. Classical and intermediate monocytes were found to be highly phagocytic while nonclassical monocytes were poorly phagocytic, an observation in agreement with previous reports²¹. Our data also provide circumstantial evidence to the hypothesis that intermediate monocytes may not be a distinct endpoint of differentiation rather a developmental stage between classical and nonclassical subsets. Principal components analysis and hierarchical clustering revealed the intermediate and classical monocytes to be closely linked whereas the nonclassical subset formed a distant cluster. This close relationship between intermediate and classical subsets has been suggested earlier²¹. It would be of interest to investigate if intermediate monocytes can switch to-and-fro between classical and nonclassical subtypes, i.e. between a predominantly phagocytic and a primarily inflammatory phenotype depending on different activation cues.

Flow cytometry-based detection of immune cell types in the context of various diseases is a useful way to diagnose severity and outcome in a clinical setting. We validated our findings in two inflammatory disease conditions, viz. sepsis and SLE. Our results indicate that expansion of CD16+ monocytes can be used to determine an inflammatory condition that is consistent with published reports^27,28,29,33. However, a question that remained unanswered was which one of the CD16+ monocyte subsets expand during inflammation? Our data show that while in an acute inflammation like sepsis both subsets increase in percentage with a concomitant decrease in classical monocyte percentage, in a chronic inflammatory situation like SLE, only the nonclassical subset is expanded. Although it is not clear whether such increase in inflammatory cell types is a cause or consequence of disease, it tends to suggest that increase in intermediate monocytes could be a differentiating factor between acute and chronic inflammation.

We conclude that ‘Classical’ monocytes are phagocytic with no inflammatory attributes, ‘Non-classical’ subtype display ‘inflammatory’ characteristics on activation and exhibit properties for antigen presentation while ‘Intermediate’ monocytes constitute a very small percentage in circulation (under physiological conditions) and appear to be a minor transitional subset that displays both phagocytic and inflammatory function. Given the importance of understanding monocyte subtypes in several human diseases (infectious as well as metabolic) very large number of reports have begun to clutter medical literature using un-validated flow cytometric assays. This highly validated phenotypic and functional characterization of Human monocyte subtypes should bring clarity to investigators in human immunology.