A method for identification and analysis of non-overlapping myeloid immunophenotypes in humans

Abstract

The development of flow cytometric biomarkers in human studies and clinical trials has been slowed by inconsistent sample processing, use of cell surface markers, and reporting of immunophenotypes. Additionally, the function(s) of distinct cell types as biomarkers cannot be accurately defined without the proper identification of homogeneous populations. As such, we developed a method for the identification and analysis of human leukocyte populations by the use of eight 10-color flow cytometric protocols in combination with novel software analyses. This method utilizes un-manipulated biological sample preparation that allows for the direct quantitation of leukocytes and non-overlapping immunophenotypes. We specifically designed myeloid protocols that enable us to define distinct phenotypes that include mature monocytes, granulocytes, circulating dendritic cells, immature myeloid cells, and myeloid derived suppressor cells (MDSCs). We also identified CD123 as an additional distinguishing marker for the phenotypic characterization of immature LIN-CD33+HLA-DR- MDSCs. Our approach permits the comprehensive analysis of all peripheral blood leukocytes and yields data that is highly amenable for standardization across inter-laboratory comparisons for human studies.

Fig 1. The development of a 10-color protocol to enumerate lymphocytes, monocytes, and granulocytes.

Antibodies corresponding to the TBNK/M/G protocol were added directly to fresh peripheral blood. The red blood cells lysed, and run on the Gallios cytometer with the addition of fluorescent counting beads. A. Fluorescent beads are gated on FL2 by time, and total beads collected by FL4 and forward scatter (FS) B. Gating strategy for the enumeration of leukocyte subsets. The gates for the selected populations are listed above each bivariate dot plot or histogram. [WBCs] = total leukocytes gated by forward and side scatter; [MNCs] = mononuclear cells.

Fig 2. Radar plots permit simultaneous visualization of leukocyte compartments.

Fresh blood was stained with the TBNK/M/G protocol and analyzed per Fig. 1. A. Gated populations from histograms for selected markers were color coded for radar plot analyses. Lymphocytes (orange) and CD14pos (purple) were gated from mononuclear cells ([MNCs]). CD3 (red), CD19 (blue), and CD56 (black) were gated from the Lymphocyte gate. B. Bivariate dot plot and radar plot comparisons. Populations were either gated from the MNC or Lymphocyte gate. Radar plots show populations on 3, 4, and 6 axes. The markers for the 6 axes include CD3, CD4, CD8, CD14, CD19, and CD56. C. Radar plot analyses of longitudinal data from a patient and a healthy control. The arrangement of the 6 axes in each plot is identical for each sample. D. Graphical representations of the longitudinal cell count data collected from 5 DLBCL patients and 10 healthy volunteer controls (HV). * = P value < 0.05.

Fig 3. The TBNK/M/G protocol and radar plot analysis are ideal for comparisons of different biological samples.

Radar plot analyses on (A) peripheral blood and bone marrow samples from a multiple myeloma patient and (B) peripheral blood and pleural fluid from a mesothelioma patient. The radar plot configuration includes CD3 (red), CD19 (blue), CD56 (black), CD14 (purple), and CD15 (granulocytes, brown) gated from total CD45+ cells ([CD45]) and the axes are identical across all samples. (C) Summary of cell count data from each of the examples.

Fig 4. Characterization of myeloid marker staining patterns on whole blood.

Peripheral blood from a healthy control sample was processed via the lyse/wash Myeloid protocol. Histograms were generated from each of 9 antibodies used to delineate myeloid populations. In most cases regions (R1, R2, and R3) were created for each peak of expression including peaks with no expression (N). Other regions represent populations of cells falling between the negative/positive peaks or populations that do not have a clear peak. Density gradient plots were created from cells gated from each region plotted by forward and side scatter. A representative example from a healthy volunteer is shown.

Fig 5. 10-color analysis of myeloid cells reveals complexity of myeloid phenotypes.

Peripheral blood from a healthy control sample was processed via the lyse/wash Myeloid protocol. A. Bivariate plots were used to delineate and color-code myeloid populations. CD33 was used to distinguish myeloid cells from lymphoid cells. LIN2 was also used as a surrogate for CD14. A plot of LIN2 and HLA-DR from the Myeloid gate was used to further separate cells into mature monocytes (LIN2+, [Monocytes]), LIN2+HLA-DR^lo/neg monocytes ([A] in dark green), LIN2-CD33+HLA-DR- myeloid derived suppressor cells (MDSC in black), and circulating dendritic cells (DC in red). Monocytes were further gated into classical monocytes LIN2+CD16- (blue), intermediate monocytes (orange), and non-classical monocytes (purple). Granulocytes were gated from CD45+ high side scatter populations and colored brown. CD33 by CD11b dot plots from mononuclear cells ([MNCs]) and total leukocytes ([CD45]) using the same color coding scheme listed above. B. Radar plots of myeloid cells were generated from MDSCs, DC, Monocytes and total myeloid cells ([Myeloid]). Selected markers and axes were arranged uniquely for each myeloid subset. C. Histograms showing representative examples of CD123 expression on LIN^-HLA-DR^-CD33⁺ ImMC MDSC (black), LIN⁺CD33⁺ monocytes (white), and CD15⁺CD66b⁺ granulocytes (brown) from a healthy volunteer and DLBCL patient. The graph shows the geometric mean fluorescence intensity of CD123 on LIN^-HLA-DR^-CD33⁺ MDSC (ImMC), LIN⁺CD33⁺ monocytes (white), and CD15⁺CD66b⁺ granulocytes and CD123⁺CD11c^- plasmacytoid dendritic cells (pDC) from healthy volunteers (n = 11). * = p<0.05 as determined by two-tailed Wilcoxon matched-pairs signed rank test.

Fig 6. Additional analyses of monocytes reveals diversity of surface expression on classical, intermediate, and non-classical monocytes.

A. Examples of monocyte gating from a healthy volunteer control and DLBCL patient. Mononuclear cells were gated from CD45+ WBCs ([WBCs]). CD14⁺ monocytes were gated from [MNCs]. CD14⁺ cells were either plotted with HLA-DR to identify HLA-DR^lo/neg populations or CD16 to delineate CD14⁺CD16^- classical monocytes, CD14⁺CD16⁺ intermediate monocytes, and CD14^loCD16⁺ non-classical monocytes. B. Histogram overlays of HLA-DR, CD86, TNFR2, and CD40 on a healthy volunteer control and DLBCL patient. Classical monocytes were colored in red, intermediate monocytes colored in black, and non-classical monocytes were colored blue.

Fig 7. Identification of potential mechanistic associations of immune phenotypes using the combination of relational data and cell counts.

Immune phenotype values were measured using the TBNK and T cell-2 protocols from 64 healthy volunteers. The percentages of PD-1, CTLA4, and CD28 positive CD8 or CD4 cells were plotted against either total (A) CD8 cells/ml and (B) CD4 cells/ml. Populations were tested for statistical significance using the Spearman correlation test.