Apparent anomalies in nuclear feulgen-DNA contents. Role of systematic microdensitometric errors

The Feulgen-DNA contents of human leukocytes, sperm, and oral squames were investigated by scanning and integrating microdensitometry, both with and without correction for residual distribution error and glare. Maximally stained sperm had absorbances which at lambdamax exceeded the measuring range of the Vickers M86 microdensitometer; this potential source of error could be avoided either by using shorter hydrolysis times or by measuring at an off-peak wavelength. Small but statistically significant apparent differences between leukocyte types were found in uncorrected but not fully corrected measurements, and some apparent differences disappeared when only one of the residual instrumental errors was eliminated. In uncorrected measurements, the apparent Feulgen-DNA content of maximally stained polymorphs measured at lambdamax was significantly lower than that of squames, while in all experimental series uncorrected measurements showed apparent diploid:haploid ratios significantly greater than two. In fully corrected measurements no significant differences were found between leukocytes and squames, and in four independent estimations the lowest diploid:haploid ratio found was 1.99 +/- 0.05, and the highest 2.03 +/- 0.05. Discrepancies found in uncorrected measurements could be correlated with morphology of the nuclei concerned. Glare particularly affected measurements of relatively compact nuclei such as those of sperm, polymorphs and lymphocytes, while residual distribution error was especially marked with nuclei having a high perimeter:area ratio (e.g. sperm and polymorphs). Uncorrected instrumental errors, especially residual distribution error and glare, probably account for at least some of the previously reported apparent differences between the Feulgen-DNA contents of different cell types. On the basis of our experimental evidence, and a consideration of the published work of others, it appears that within the rather narrow limits of random experimental error there seems little or no reason to postulate either genuine differences in the amounts of DNA present in the cells studied, or nonstoichiometry of a correctly performed Feulgen reaction.