Figure 1.
Purification of peripheral blood Granulocytes.
Representative example of preparation of granulocytes by a one-step double density centrifugation method [12]. In the left-hand panel, a whole blood sample diluted 1∶3 in PBS is layered on a double density Ficoll gradient: poly (density 1113 g/L) and H (density 1077 g/L). In the right-hand panel one sees that, after centrifugation (500 g for 35 minutes at 22°C), erythrocytes (RBC), granulocytes (PMN), mononuclear cells (MNC) and plasma fractions are clearly separated. The relatively few erythrocytes contaminating the granulocyte fraction are subsequently removed by gentle osmotic lysis.
Figure 2.
Counting granulocytes that have a PIGA-inactivating mutation.
The left-hand panel shows the gating strategy, based on SSC and CD45, used to eliminate virtually all cells that are not granulocytes. The right-hand panel shows that granulocytes are positively identified as CD11b-positive; at the same time, GPI-negative granulocytes are identified (lower right quadrant) as cells that (while being among the 95% of cells with higher CD11b fluorescence) have a GPI-linked-PE fluorescence intensity lower than 4% of the geometric mean of the fluorescence values of all events. Mutant frequency, ƒ, was calculated as the number of negative events, as just defined, per million cells.
Figure 3.
Technical reproducibility and intra-individual variation of ƒ.
(A) In this plot we show ƒ values of 45 split samples (see Methods). For each sample the higher value is a filled circle and the lower value of ƒ is an empty square. (B) Spiked control. The X axis shows the number of the added PNH (GPI-negative) granulocytes to a control sample; the Y axis shows the number of GPI-negative granulocytes measured (after the subtraction of the ƒ value measured in the control sample). Square, diamond and circle indicate different samples. The line indicates the theoretical identity between added and measured GPI-negative cells. See “Materials and Methods” for details. (C) In this plot we show ƒ values for 27 subjects who have been tested at 3 different times. Each diamond represents one measurement of ƒ. White and black diamonds alternate in the interest of clarity. In panels A and C the data have been arranged in order of decreasing mean ƒ values, again in the interest of clarity.
Figure 4.
Histogram distribution of ƒ in a population of 142 healthy individuals.
A. On a linear scale ƒ has a highly asymmetric truncated distribution. B. On a log scale ƒ has a truncated distribution, where the truncation clearly results from the ‘zero’ values. C. After replacement of the zero values with values obtained by maximum likelihood estimation (see text) the distribution of ƒ on a log scale is not significantly different from a normal distribution.