Locus-specific recombination controlled by Cre recombinase primarily has been used to investigate gene function in transgenic animals. Inducible strategies for regulating recombinase expression now also allow the timing of changes in gene expression to be controlled in specific tissues or cell types. Previously, we have described one such transgenic line, Ahcre, in which constitutive recombination is induced in intestinal progenitors but not in differentiated Paneth cells (Ireland et al., 2004). Potentially, this pattern allows the inheritance of a Cre-activated reporter into the Paneth cell population to be followed to determine their turnover time. Our understanding of cellular longevity in epithelial tissues stems from a single approach, that of detecting the incorporation of nucleotide analogues into newly replicated DNA before cell division. The reliability of such approaches depends on the untested assumption that the rate of cell turnover is not perturbed. However, to date, there has been no independent method to confirm or refute the turnover time of any epithelial cell type.
Labeling studies using [3H]thymidine indicate that Paneth cells are nonproliferative and have a turnover time of 18–22 days (Cheng et al., 1969; Troughton and Trier, 1969; Cairnie, 1970; Bjerknes and Cheng, 1981). Here, we exploit the differential somatic induction of Cre recombinase in Ahcre mice to estimate Paneth cell turnover time independently of nucleotide incorporation. This time, 57 days, suggests that previous estimates of cell turnover of long-lived cell types may be underestimates and identifies a new strategy for determining cellular longevity in vivo.
Differential Reporter Expression in Intestinal Precursors and Paneth Cells
Ahcre/R26R-EYFP mice were treated with five daily doses of β-naphthoflavone (βNF) and killed 1 day after the last injection. The pattern and intensity of intestinal EYFP expression was assessed by direct viewing of EYFP fluorescence on the conventional microscope using a YFP filter set. The intensity of EYFP-positive cells was similar throughout the vertical crypt to villus axis (Fig. 1A) and the observed pattern of recombination was as previously described (Ireland et al., 2004). Paneth cells were clearly reporter-negative compared with neighboring intercalated cells, all or part of which contain the epithelial stem cells (Fig. 1A,B). There was no detectable elevated fluorescence in R26R-EYFP tissues not containing cre or in uninduced Ahcre/R26R-EYFP mice.
Paneth cells could inherit the recombined allele from precursors either directly or, if there are unrecombined precursors, after successive cellular generations. To determine the relative probability of these two possibilities, we determined the proportion of available precursors that are EYFP-positive after induction.
Ahcre/R26R-EYFP mice were sampled at day 5 after induction. Nuclei in cryostat sections were visualized with TOPRO-3, and the confocal microscope was used to collect images from 0.6-μm optical slices. The nuclei of the intercalated cells are elongated, having a distinctive alignment perpendicular to the basal lamina compared with those of the Paneth cells, which are rounded (Fig. 2A). This observation allowed initial identification and selection of intercalated cells on the basis of TOPRO-3 staining only, before scanning for EYFP, thereby eliminating any selection bias based on EYFP expression. After visualization of both EYFP and TOPRO-3 (Fig. 2B,C), the cell was scored for reporter expression. Only one intercalated cell per crypt was scored. Of 180 cells scored in total, 177 were reporter-positive and 3 reporter-negative. Therefore, 98.4% of the intercalated cells were recombined on induction with only 1.6% remaining unrecombined.
The 1.6% of unlabeled intercalated cells could in principle be part of an enriched subpopulation with a bias in their ability to persist and give rise to progeny, including Paneth cells. We conclude that this is not the case by reference to our original data (Ireland et al., 2004). This publication shows that induced Ahcre animals expressing β-galactosidase from the R26R reporter locus have a stable representation of recombinant epithelium at the 98% level up to 6 months after induction. Therefore, Paneth cells have a high probability of inheriting the recombined allele directly from the reporter-positive pool of available precursors.
Confirmation That Paneth Cells Are Correctly Identified by Nuclear Morphology
Paneth cells are distinctive in position and morphology, and their nuclei are easily recognized in cross-section. To unequivocally confirm Paneth cell identity, sections from Ahcre/R26R-EYFP mice were stained with the lectin UEA-I, a recognized Paneth cells marker (Falk et al., 1994). Sections of intestine taken from mice sampled at day 5 after induction show reporter-positive and reporter-negative cells in the crypt base (Fig. 3A). The UEA-I gave very distinctive staining that clearly distinguished Paneth cells from the neighboring intercalated cells by staining of both secretory granules and the cell membrane (Fig. 3B,D). Cells with elongated nuclei positioned perpendicular to the basal membrane, visualized with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; Fig. 3C) did not stain with the lectin (Fig. 3D) and, therefore, were confirmed to be intercalated cells. All cells with round nuclei were stained with UEA-1 (Fig. 3D), confirming that these cells could be confidently identified by nuclear orientation alone.
Determination of Paneth Cell Longevity
To determine the time scale for reporter inheritance into Paneth cells, a cohort of Ahcre/R26R-EYFP animals was induced and killed at 5, 25, 35, 45, and 55 days after the first injection. Two mice were scored at each time point, with a minimum of 50 cells per mouse counted.
As with the intercalated cells, the cells were scored blind by first identifying Paneth cells by nuclear orientation using TOPRO-3 (Fig. 4A) then scanning to score the cell for reporter expression (Fig. 4B,C). One cell per crypt was scored, and this was randomly selected within the bottom five cell positions from the crypt base. Paneth cells in higher crypt positions were generally the first to inherit the marker, with those at the base acquiring it later. The frequency of recombined Paneth cells increased linearly with time, after an initial delay, from 2% at day 5. By 55 days after induction, 93% of Paneth cells were recombinant (Table 1; Fig. 4D). A linear regression line gave the best fit (r = 0.997) and indicated that 100% labeling would be achieved by 57 days. This interval was taken to be the Paneth cell turnover time.
Table 1. Increase in EYFP-Expressing Paneth Cells With Timea
Days after induction
Total cells counted
Number of EYFP-positive cells
% EYFP-positive Paneth cells
Ahcre/R26R-EYFP mice induced with five daily i.p. injections of β-naphthoflavone (βNF) 80mg/kg. The first time point is 1 day after the last injection.
Previous turnover times for Paneth cells have been based on [3H]thymidine labeling experiments (Cheng et al., 1969; Troughton and Trier, 1969; Bjerknes and Cheng, 1981). There are several reasons to suspect that [3H]thymidine may itself influence cell turnover in the intestine: (1) a proportion of crypt base cells are hypersensitive to irradiation and are killed at extremely low doses of [3H]thymidine (Potten, 1977), including doses well below those used in labeling studies (Chwalinski and Potten, 1989); (2) a subset of crypt base cells are known to be recruited into rapid cell cycle by small doses of radiation (Tsubouchi and Potten, 1985); (3) it is possible that [3H]thymidine may be reused after labeled nuclei are degraded, a point that may be particularly relevant to Paneth cells as these are known to phagocytose adjacent cells and may thereby access the radiolabel independently of DNA replication (Maurer, 1981).
Our determination of Paneth cell lifespan is three times longer than that previously calculated from continuous infusion studies. Two such studies at electron microscopic level showed an initial 2-day delay in the appearance of the thymidine label, analogous to the 5-day delay observed here (Cheng et al., 1969; Bjerknes and Cheng, 1981). Conceivably, this shorter time could be due to the recognition of Paneth cells at an earlier stage of maturation. Equally, it could be due to accelerated Paneth cell renewal caused by the tritiated label. That such accelerated renewal occurs is evident from comparison of the early rate, up to 10 days, of accumulation of labeled Paneth cells in the studies: 4–6% labeled Paneth cells per day in the earlier studies compared with 2% per day in this study.
If Cre or EYFP were toxic to intercalated cells, there would be selection for nonexpressing cells. However, the results show that only 1.6% of intercalated cells remain unrecombined and presumably do not express Cre. If these survived preferentially, we would expect to see an increased proportion of nonrecombined epithelium appear subsequently, which is not observed. In the inducible Ahcre system, cre is expressed for a minimal time and is not induced in Paneth cells (Campbell et al., 1996; Ireland et al., 2004). For these reasons, we believe the current method to be a highly physiological means of determining Paneth cell turnover time.
It is not clear why the Paneth cells are uninducible in the Ahcre mouse. Inducible cells must contain both a functional Ah receptor and Ah receptor nuclear translocator (ARNT) that are expressed at different levels in a cell-type and tissue-specific manner (Carver et al., 1994; FitzGerald et al., 1996). Certainly, clear differences in the levels of induction of the Ah promoter between tissues have been reported previously, and some tissues, including heart and muscle, are completely nonresponsive (Campbell et al., 1996). It is possible that Paneth cells do not express the Ah receptor, the ARNT or both. Alternatively, an Ah receptor repressor has been identified in numerous nonresponsive cells that is capable of binding the xenobiotic response elements within the Ah promoter, thereby preventing transcriptional up-regulation (Mimura et al., 1999; Tsuchiya et al., 2003; Korkalainen et al., 2004).
The contrast between the longevity of Paneth cells compared with the other cell types of the intestinal epithelium has always been striking. Paneth cells have been associated with intestinal inflammation in diverse disease situations (Stamp et al., 1992; Tanaka et al., 2001). Mutation of NOD2 and altered levels of TNFα and defensins have been implicated in both animal models and human disease (Lala et al., 2003; Ogura et al., 2003; Wehkamp et al., 2004, 2005). At the cellular level, Paneth cell hyperplasia is induced by T cells in a murine model of intestinal infection (Kamal et al., 2001) and Paneth cell metaplasia is observed in the colon of inflammatory bowel disease patients (Haapamaki et al., 1997, 1999; Cunliffe et al., 2001). Therefore, the regulation of Paneth cell formation and longevity, as well as function, are recognized as important determinants in the etiology of inflammatory disease, investigations of which will need to take account of the greater stability of this population. This strategy could include using the Cre-based approach described here to detect changes in the kinetics of Paneth cell formation accompanying inflammation in mouse models.
The development of increasing numbers of inducible mouse strains for the up-regulation of Cre activity may permit reciprocal patterns of induction to be identified in other precursor/progeny cell populations to allow their relationships to be defined in quantitative terms. Candidates for such studies include the origins of pancreatic beta cells from ductal cells, the conditional stem cell properties of hepatic oval cells, and the longevity of parietal cell and zymogen cells in the glandular stomach.
Double transgenic Ahcre/R26R-EYFP mice were generated by crossing Ahcre mice (Ireland et al., 2004) with the R26R-EYFP reporter strain (Srinivas et al., 2001). For induction of the Ah promoter, double-transgenic Ahcre/R26R-EYFP animals were induced with five daily i.p. injections of 80 mg/kg βNF (Sigma, Dorset, UK) dissolved in corn oil (8 mg/ml).
Reporter Visualization and Immunohistochemistry
Tissue from Ahcre/R26R-EYFP animals was fixed in 4% paraformaldehyde for 48 hr at 4°C then equilibrated in 20% sucrose/phosphate buffered saline. Tissue was rolled in a “swiss-roll” style, flash frozen in liquid nitrogen–cooled 2-methylbutane (Sigma-Aldrich), and stored at −80°C. Cryostat sections were cut at 10 μm. Ulex europaeus agglutinin (UEA-I) biotin conjugate (Sigma #L-8262) at 1:1,000 was applied to air-dried, frozen sections for 1 hr at room temperature. Streptavidin–Alexa Fluor 633 conjugate (Molecular Probes, The Netherlands #S-21375) was applied at 1:200 for 15–30 min at room temperature. Nuclei were stained with TOPRO-3 (Molecular Probes) at 1:5,000 or DAPI and mounted in Vectashield (Vector Laboratories Ltd, UK). Sections were examined on a Zeiss Axioplan2 microscope or a Zeiss LSM confocal microscope.