Natural Killer Cell Lysis Assay Protocol

Browne et al. [1 (link)] analysed the correlation matrix for eight measures of immune system function of 72 females with breast cancer, recorded during investigation of the physiological consequences of a psychological intervention [3 (link),4 (link)]. Four ⁵¹Cr-release measures of natural killer cell lysis were obtained using effector (NK cell) to target cell (K562 human myeloid cell) ratios of 100:1, 50:1, 25:1 and 12.5:1. Following Browne et al. [1 (link)] we designate these measures by their effector to target (E:T) ratios, NK100, NK50, NK25 and NK12 respectively. Similarly, natural killer cell lysis measured in the presence of recombinant interferon gamma (rIFNγ) using E:T ratios of 50:1, 25:1, 12.5:1 and 6.25:1, are designated IFN50, IFN25, IFN12, and IFN6 respectively. Lower E:T ratios are used in the presence of rIFNγ because rIFNγ increases NK cells' ability to rupture target cells.
The correlations reported in Browne et al.'s [1 (link)] Table 1 indicate that the four NK measures correlate highly with one another (average r = 0.852), and that the four rIFNγ enhanced NK measures also correlate highly with one another (averaging 0.960). However, the low correlations between the sets of NK and rIFNγ measurements (averaging only .111) indicate that the two sets of measurements reflect relatively distinct aspects of natural killer cell functioning. Browne et al. [1 (link)] viewed this as justifying the use of an exploratory two-factor model (Figure 1) which, unfortunately, was significantly inconsistent with the data (χ² = 103.59, degrees of freedom (df) = 13, and probability p < 10^-15). The small but significant residual differences between the data correlations and the correlations implied by the two-factor model were dismissed by Browne et al.[1 (link)] as "negligible from a practical point of view". SEMNET discussion of this model prompted Hayduk to investigate whether some unrecognized measurement feature was producing the significant, even if seemingly slight, ill fit.
Andersen, Farrar, Golden-Kreutz, Kutz, MacCallum, Courtney & Glaser [3 (link)] provide a description of the reasonably standard procedures used to obtain the Browne et al. [1 (link)] data. Peripheral blood leukocytes (PBLs) were obtained from 60 mL of venous blood, counted so that a known number of PBLs could be suspended in medium and incubated with either additional medium or additional medium plus rIFNγ. K562 target cells (a human myeloid cell line sensitive to NK cell activity) were labelled with ⁵¹Cr and aliquoted with the effector cells (either the NK, or the rIFNγ activated NK cells) in the ratios reported above. The cell mixture was centrifuged to ensure cell surface contact, and incubated to provide an opportunity for the NK cells to bind and rupture the target cells, thereby releasing the radioactive target cell cytoplasm. Gamma radioactivity of the supernatant collected from a second centrifuging indicated the effectiveness of the NK or rIFNγ-activated-NK cells at lysing the target cells, with larger measurements corresponding to more effective NK cell activity.