Northwestern University Medical School Chicago, Illinois 60611-3008
Recently, Fuller et al. (1) reported the absence of mutations in the FSH-receptor in granulosa cell tumors. These findings differed from our previous study (2), which described a Phe-to-Ser mutation in codon 591 (F591S) in the 6th transmembrane domain of the FSH-receptor in approximately two thirds of sex cord tumors, including granulosa cell tumors.
Because of the discrepancy in these two studies, we analyzed additional samples of granulosa cell tumors and reanalyzed the original samples reported in our study. In our analyses of ten new specimens of granulosa cell tumors, we found that none of the new specimens had the F591S mutation. However, when we reanalyzed the DNA samples reported in our earlier study (2), the results were confirmed as originally reported. Although the F591S mutation had been confirmed multiple times and using several independent techniques, including direct DNA sequencing, oligonucleotide specific hybridization, and mutation-specific digestion with restriction endonucleases, we remained skeptical in view of our results with new specimens and in view of the findings of other groups. Therefore, all of the original paraffin blocks were re-extracted, DNA was isolated, and the sequence analyses were repeated. In this case, we did not find the F591S mutation in any of the specimens. Although there are several possible explanations, contamination during the initial paraffin extraction or DNA isolation process is most likely.
The source of DNA contamination remains elusive and was difficult to recognize because direct sequencing resulted in a characteristic heterozygous mutation pattern (2). Although there was no evidence of contamination in any of the more than 100 controls that were run during the sequence analyses, we did not anticipate the possibility that the original DNA specimens might be contaminated. Care was taken, for example, to use new blades for the sectioning of each specimen. How contamination occurred and the source of the contaminant remain unknown. To our knowledge, no plasmid with this sequence ever existed in the laboratory at the time of the DNA isolation. It is also unknown why the contamination affected only some of the samples.
Taken together, these analyses suggest that the F591S mutation that we identified originally was due to contamination that occurred during the handling of the paraffin blocks or during DNA isolation. Contamination did not occur during the PCR analyses for mutation detectionthe step at which controls are typically introduced. In addition to the implications for our own study, this problem underscores the potential difficulties associated with mutation detection by PCR. In retrospect, this erroneous detection of the F591S mutation explains its unusual functional effects and its occurrence in tumor types with disparate histological classifications. The F591S substitution inactivated the FSH-receptor rather than causing constitutive activation as occurs with activating mutations of the TSH-receptor or the LH receptor. We regret any confusion that this error has caused.
Footnotes
Received May 11, 1998. Address correspondence to: Dr. J. Larry Jameson, Center for Endocrinology, Metabolism, and Molecular Medicine, Northwestern University Medical School, Tarry Building 15, Room 709, 303 East Chicago Avenue, Chicago, Illinois 60611-3008.
References