Correspondence re: N. N. Mahmoud et al., Genotype-Phenotype Correlation in Murine Apc Mutation: Differences in Enterocyte Migration and Response to Sulindac. Cancer Res., 59: 353–359, 1999.

Articles recently published in Cancer Research and elsewhere (1, 2, 3, 4, 5) suggest that inheritance of a mutant allele of the apc gene in ApcMin^+/− mice exerts a dominant-negative effect in gastrointestinal epithelial cells in the normal appearing mucosa. More specifically, the data show a decrease ranging from 50 to 80% in levels of apoptosis of cells in the normal mucosa, and this is interpreted as an important factor in the cause of tumor formation in these animals.

In all mouse models with an inherited mutation of apc—the ApcMin (6), Apc^Δ716 (7), and Apc1638 mouse (8, 9)—all of the normal tissues, including cells in the gastrointestinal mucosa, are heterozygous for apc and are apc^+/−. However, even the smallest of the gastrointestinal tumors that form have lost or inactivated the wild-type allele (7), similar to the situation reported for loss of the wild-type APC allele in adenomas and carcinomas of patients with the syndrome familial adenomatous polyposis (FAP), in which there is inheritance of a mutant APC allele (10). Moreover, specific evidence has been presented that demonstrates that a transgenic model in which a mutant apc allele is introduced into animals that have two wild-type alleles has no dominant-negative effect (11).

Thus, it is difficult to interpret the dominant-negative effect suggested in references 1, 2, 3, 4, 5 in the context of these other data. Additionally, it would be helpful if the authors could resolve a discrepancy in their data. The latest paper reports a low apoptotic index (approximately 5%) in the mucosa of control, apc^+/+ mice, consistent with the work of many others (see Refs. 12, 13, 14, 15, 16). However, in all of the earlier papers (2, 3, 4, 5), apoptosis in control animals was an order of magnitude higher, and the decreases in the mucosa of ApcMin^+/− mice were seen in comparison with this high level. This high level of apoptosis would be inconsistent with the growth and development of normal animals. In addition, these reports state that the calculation of percent apoptotic cells is based on the number of apoptotic cells divided by total crypt and villus cells, and that all apoptotic cells were located in the upper third of the villi. This implies that essentially all of the cells in the upper third of the villus are apoptotic. Such extensive apoptosis is not a feature of the histology of the gastrointestinal tract.

Technical problems that contribute to a large overestimate of the apoptotic index would make it difficult to measure accurately any affect of the apc mutation. Because the question of whether mutations in apc can exert a dominant-negative effect is critical in understanding how this gene initiates tumor formation, it is important that the authors resolve this issue.

The relative decrease in TUNEL¹ staining by the intestine of Min/+ mice compared to their wild-type littermates is only one of several observations supporting a dominant negative effect by the truncated APC protein resulting from the Min mutation. When we compare the histologically-normal enterocytes of Min/+ mice with their littermates lacking the mutant allele (+/+), we not only see decreased staining by TUNEL, we also find decreased proliferation as measured by proliferating cell nuclear antigen staining (in a distribution opposite that of staining by TUNEL) and decreased enterocyte migration as measured by a timed assay following bromodeoxyuridine labeling (1, 2). Moreover, we do not find this discrepancy in tissues from the Apc1638N mouse, an animal whose germline Apc mutation does not produce a detectable truncated protein (2). Further support for a dominant negative effect by some Apc mutations was reported in a recent paper by Dihlmann, et al. (3). This group expressed truncated APC proteins together with wild-type APC in a colorectal cancer cell line that lacked endogenous wild-type APC. Using β-catenin-/Tcf-mediated gene transcription as a read-out, they showed that a mutant APC truncated at codon 1309 strongly inhibited wild-type APC activity, whereas gene products from APC mutant at codon 386 or 1465 produced minimal inhibition of wild-type activity. Interestingly, in humans the 1309 mutation is associated with a severe polyposis phenotype, whereas the 386 and 1465 mutations are associated with attenuated polyposis. In the transgenic model described in Dr. Augenlicht’s letter, a third-copy Apc minigene truncated at codon 716 was introduced (4). This was accompanied by two wild-type alleles. These animals contained no tumors, although Apc⁷¹⁶ knockout mice developed many intestinal tumors. This result may be due to a gene dosing effect, as it is possible that the level of truncated APC protein produced by the minigene was too low to exert a dominant negative effect in the face of two functioning wild-type alleles.

To address the concerns over our TUNEL data, I will provide some background on the use of this technique in our laboratory. We measure apoptosis in the enterocytes of Min/+ mice using a commercially available TUNEL kit. We have found that the percentage of the total enterocyte population staining by this method can be quite variable, depending upon the age of the reagents, the particular batch used, and most important, the processing time of the specimen. Because of this, we do not claim that our data can be interpreted as showing the absolute percentage of apoptotic cells in the mucosa. Rather, this method is used to compare the relative apoptotic index of animals that differ in their germline Apc status. To minimize error when making comparisons, we stain tissues harvested and processed in the same manner on the same day and use a single batch of reagents. All of the specimens are scored by blinded observers.

Over several years of performing the TUNEL stain, our technique has admittedly improved. As Dr. Augenlicht points out, the percentage of TUNEL positive cells in all animals assayed has fallen as we adopted shorter specimen-processing times. In our most recent report, the staining in the mucosa of the wild-type animal was approximately 12% and was only 3% in the Min/+ mouse. Despite these changes in technique, what has not changed, throughout all of our studies, is the finding that there is significantly less TUNEL staining (and by inference, less apoptosis) in the histologically normal mucosa of Min/+ mice compared with their wild-type littermates. The distribution of TUNEL staining is always toward the tip of the villus, where apoptosis is to be expected. We also consistently observe an increase in TUNEL stain when animals are treated with chemopreventive doses of sulindac or other nonsteroidal anti-inflammatory drugs (1, 2, 5678). This finding is in keeping with the effect of these drugs on colorectal cancer cell lines and in humans with familial adenomatous polyposis (9,10,11 12).