Distinct Levels of Radioresistance in Lgr5⁺ Colonic Epithelial Stem Cells versus Lgr5⁺ Small Intestinal Stem Cells

The differential response of mammalian tissues to high doses of ionizing radiation is posited in terms of differences in radiosensitivity of tissue stem cell compartments (1–3), although no definitive study corroborating this widely held hypothesis has ever been published. This is because until recently, validated solid tissue stem cells were not available for direct study. Although it is well documented that the large intestine is more radioresistant than the small intestine, mechanisms underlying large intestine radioresistance remain poorly understood (4, 5). Here, we compare two highly similar Wnt-driven Lgr5⁺ (also known as Gpr49; refs. 6, 7) intestinal stem cell (ISC) populations at the base of the crypts of the small and large intestine in an attempt to elucidate mechanisms controlling radiosensitivity.

Of currently studied adult stem cell populations in mammalian organs, response of the small intestinal Lgr5⁺ crypt base columnar cell (CBC) to genotoxic stress is perhaps the best studied. The small intestinal mucosa is a rapid turnover system, considered driven by mitotic activity of self-renewing Lgr5⁺ CBCs (6, 7). Prior to discovery of the Lgr5⁺ CBC, survival of ISCs to genotoxic stress was quantified indirectly using the microcolony assay (also termed clonogenic assay) of Withers and Elkind (8) by counting regenerating crypts in histologic sections at 3.5 days postradiation. The rationale of this assay, regarded as one of the best surrogate assays for adult mammalian stem cell responses for decades, is that it must have taken at least one stem cell to regenerate a complete crypt.

The discovery that the CBC, most often located between Paneth cells at the base of the small intestinal crypt (9), is a legitimate ISC represents a milestone in the gastrointestinal field and has permitted detailed analysis of its response to genotoxic stress. Although many groups have confirmed that the threshold for survivability of the small intestine to ionizing radiation is approximately 10% regenerative crypts at 3.5 days after irradiation as determined by the microcolony assay (10–12), we determined that the death of CBCs preceded crypt regeneration, detectable at 1 day postirradiation and maximal at 2 days (13, 14). Furthermore, loss of CBCs predicted outcome of the microcolony assay as one might anticipate if CBCs represented the relevant target for small intestinal survival. Loss of CBCs within the first 2 days postirradiation was biphasic, with about 30% of CBCs dying by apoptosis during the first 24 hours after 12 Gy, while cells are growth arrested and repairing DNA damage, followed by mitotic death of approximately 60% of CBCs during the subsequent 24 hours, leaving a residual of 10% of CBCs available for crypt regeneration. That CBCs were relevant ISCs for organ recovery was indicated by lineage-tracing studies postirradiation. At 15 Gy, a dose that reduces crypt counts by 95% to 99% in the Withers and Elkind analysis, and is not survivable, complete loss of CBCs was detected by 48 hours postirradiation. Further study revealed that of all the cell populations of the small intestinal crypt/villus unit, the Lgr5⁺ CBC was the most radioresistant due to efficient use of error-free homologous recombination to repair double-strand breaks (DSB).

Like the small intestine, the large intestine is maintained by a rapidly cycling population of Wnt-driven Lgr5⁺ colonic epithelial stem cells (CESC) that reside at the crypt base in between a cKit⁺ population of niche cells. In the current study, we profile these two highly similar Lgr5⁺ stem cell populations in the large and small intestine. We show that Lgr5⁺ CESCs in the distal large intestine are far more radiation resistant than their counterparts in the small intestine, differences attributable to substantive differences in cell cycle reentry and the propensity to undergo mitotic death. Activation of cell-cycle checkpoints by DSBs that transiently arrest cell-cycle progression is critical in optimizing DSB repair and maintaining genomic integrity (15). Accordingly, tight coordination between DSB repair and checkpoint recovery constitutes a generic function of a high-fidelity DNA damage response (DDR; ref. 16). Recently, however, a pathophysiologic alternative checkpoint dynamic has been described, termed checkpoint adaptation (reviewed in refs. 16, 17), an uncoupling of completion of DSB repair and checkpoint recovery that enables precocious cell-cycle progression in the presence of residual unrepaired DNA (18). G₂M transition in the presence of unrepaired DNA renders kinetochore/mitotic spindle dysfunction during segregation of damaged DNA strands (19), promoting toxic chromosomal rearrangements, genomic instability, and reproductive (also termed clonogenic) lethality (20). Hence, although the mechanism driving dysfunctional checkpoint adaptation remains to a large extent unknown (17), it predisposes affected cells to mitotic catastrophe. Here, we show CBCs are subject to checkpoint adaptation, which drives a unique mechanism of small intestinal tissue radiosensitivity, while CESCs maintain a fully controlled checkpoint recovery, coordinated with the completion of DSB repair, conferring a radioresistant colonic tissue phenotype.

Lgr5-lacZ and Lgr5-EGFP-ires-CreERT2/Rosa26-lacZ mice were genotyped and used as described previously (6, 13). All mice were maintained on a C57BL/6 background. Mouse protocols were approved by Memorial Sloan Kettering Cancer Center Institutional Animal Care and Use Committee.

Whole-body radiation (WBR) was delivered with a Shepherd Mark-I unit (model 68, SN643, J. L. Shepherd & Associates) operating ¹³⁷Cs sources at 1.72 Gy/minute. Distal colon irradiation was delivered with 250 kVp X-ray produced by X-Rad 320 with 0.25-mm Cu filtration as per published method with minor modifications. Briefly, anesthetized mice were placed and irradiated in a close-fitting jig. The jig was covered by a 5-mm thick lead shield except for a 2.0 × 1.0 cm rectangular aperture enabling precise targeting of the distal colon. Field uniformity and dose homogeneity were defined using a 16-mm thick Superflab phantom, with an IBA CC04 ionization chamber placed at 8 mm, and by exposing Kodak XV film for density contour using a densitometer. Crypt and ISC survival curves were calculated by least square regression analysis, with a modification of the FIT software program, as published previously (21). The program fits curves by iteratively weighted least squares to each set of dose survival data, estimates covariates of survival curve parameters and corresponding confidence regions, and plots the survival curve. It also derives curve parameters, such as the D₀, the reciprocal of the slope on the exponential portion of the curve, representing the level of radiosensitivity.

Intestinal tissue samples were fixed by 16- to 18-hour incubation in 4% freshly prepared neutral-buffered formaldehyde (21) and embedded in paraffin blocks. Transverse sections of the full intestinal circumference were prepared and stained with hematoxylin and eosin as described previously (13).

Lgr5-lacZ mice were euthanized after radiation, and four 2.5-cm sequential segments of proximal jejunum from the ligament of Treitz, or 2-cm segment of distal colon were obtained. Staining for the presence of β-galactosidase was as per ref. 6.

The microcolony assay was performed as described by Withers and Elkind (8). Briefly, at 3.5 days (small intestine) or 5 days (distal colon) after irradiation, intestines were obtained and hematoxylin and eosin stained as above. Surviving crypts were defined as containing 10 or more adjacent chromophilic cells and a lumen. The circumference of a transverse cross-section of the intestines was used as a unit. Number of surviving crypts was counted per circumference.

Lgr5-lacZ mice, 6 to 8 weeks old, were injected intraperitoneally with 100 μL of 20 mmol/L EdU solution 2 hours before sacrifice (22). Distal colons were collected for paraffin embedding and EdU staining. EdU was detected with the Click-iT EdU Cell Proliferation Kit according to the manufacturer's instructions (Invitrogen).

Consecutive 5-μm thick sections were used for double staining of caspase-3 and Lgr5 in situ. Apoptotic crypt cells were identified by caspase-3 staining of the active fragment. Five-micron sections were deparaffinized and rehydrated through graded ethanol. Endogenous peroxidase activity was blocked for 5 minutes with PBS containing 0.3% H₂0₂. Antigen retrieval was performed by boiling sections for 15 minutes in 0.1 mol/L citrate buffer (pH 6.0). Nonspecific antibody binding was blocked for 30 minutes by incubation with 10% normal goat serum. Sections were incubated with rabbit anti-caspase-3 (Cell Signaling Technology, #9661; 1:100 dilution) overnight at 4°C in a humidified chamber. Primary antibody was visualized using biotinylated goat anti-rabbit secondary antibody and developed with an ABC Kit and DAB. In situ hybridization (ISH) employed the QuantiGene ViewRNA ISH Tissue 2-plex Assay Kit (QVT0012) according to the manufacturer's instructions using Lgr5 (mouse VB1-10818) and GAPDH (mouse VB1-10150) probes from Affymetrix. GAPDH ISH confirmed mRNA integrity of all samples.

Lgr5-EGFP-ires-CreERT2/Rosa26-lacZ mice were injected intraperitoneally with a single dose of tamoxifen (4 mg/25 g mouse) immediately after 19 Gy distal colon irradiation. Mice were sacrificed at days 0, 7, and 12 postirradiation, and distal colons were collected and stained for lacZ as per refs. 6 and 13.

Paraffin-embedded tissue sections were dewaxed, rehydrated, and stained with antibodies, using standard procedures (13). Primary antibodies for immunofluorescence were mouse monoclonal anti-H2AX Ser139 (JBW301, Upstate Biotechnology; dilution 1:2,000), polyclonal goat anti-lysozyme (Santa Cruz Biotechnology, sc-27958; dilution 1:500), polyclonal goat anti-cKit (R&D Systems, AF-1356; dilution 1:400). Multichannel fluorescence images were acquired using an upright widefield Zeiss Axio2 Imaging microscope with AxioCam MRm Camera.

Intestinal tissues, harvested and processed as above, were stained with hematoxylin and eosin as per ref. 13, and mitoses were scored by visual inspection at ×400 magnification. Normal mitoses were scored when a cell was undergoing mitosis and condensed chromosomes aligned symmetrically. Aberrant mitoses were scored when condensed chromosomes showed multipolar spindles, lagging or misaligned chromosomes, anaphase bridges, or micronuclei as per refs. 23 and 24.

Values represent mean ± SD or SEM. Differences were analyzed by Student t test. A P value ≤0.05 was considered significant.

Initial studies examined the dose response of intestinal crypts and Lgr5⁺ ISCs to WBR. The crypt regeneration process was studied using the microcolony assay, standardized to measure the number of regenerating crypts per intestinal circumference at day 3.5 postradiation for small intestine and at day 5 for large intestine (4), as a surrogate for stem cell survival. In the small intestine of Lgr5-lacZ transgenic reporter mice (Fig. 1A), 15 Gy WBR, which results in this strain in 100% lethality from the radiation gastrointestinal syndrome (13), yields total depletion of Lgr5⁺ ISCs (blue cells at the crypt base) and complete loss of crypts. In contrast, distal colon crypts are significantly more radioresistant, displaying a combination of degenerating crypts (thin walled, white arrows) and regenerating crypts (violaceous thick walled, black arrows) after 15 and 19 Gy (Fig. 1B). Consistent with Lgr5⁺ ISCs as representing critical determinants of the radioresistance of the colon, Lgr5⁺ ISCs remain detectable at 15 and 19 Gy (Fig. 1C). Radiation dose survival analysis of the small intestinal and colonic crypts and ISCs, performed over a wide range of doses (8–30 Gy; Fig. 1D), confirmed large intestinal crypts are markedly radioresistant with a D₀ = 2.1 ± 0.1 Gy compared with small intestinal crypts that manifest a D₀ = 0.9 ± 0.1 (P < 0.01). Even greater disparity was observed when comparing large and small Lgr5⁺ ISCs that display D₀ = 6.0 ± 0.3 Gy and D₀ = 1.3 ± 0.1 values, respectively (P < 0.01), a very large 4.6-fold difference.

To investigate the impact of Lgr5⁺ ISCs on regeneration of the colon after high dose ionizing radiation in the absence of concomitant damage to the small intestine, we engineered a specialized colorectal jig modified from published literature (25, 26). The jig was designed to frame the distal colon, based on anatomic and CT (27) data, localizing a large intestinal area restricted to the pelvis. Radiation was delivered through a 2.0 × 1.0 cm aperture, thus specifically targeting the distal colon while avoiding other regions of the intestinal tract (Fig. 2A). Whereas orthovoltage radiation, as used here, also causes epilation restricted to the exposed skin, Fig. 2B shows the rectangle of fur loss at 40 days, serving as a biological marker of the incident beam. Figure 2C presents the total large intestinal section in the standardized Swiss roll configuration (28) and reveals radiation damage to intestinal mucosa at day 5 postirradiation was confined to distal colonic tissue (Fig. 2C).

Using this strategy, Lgr5-lacZ mice were treated with 19 Gy WBR, a dose normally lethal to mice from the small intestinal radiation gastrointestinal syndrome at 5 days. Figure 2D reveals that 5 days is the nadir for both crypts and CESCs in the distal colon after 19 Gy, with recovery after 7 days. Unlike the small intestine in which loss of CBCs and crypts occur in parallel (13), in the large intestine, <2% of crypts survive at day 5 after 19 Gy (Fig. 2E, top), yet >40% of CESCs survive within the injured mucosa (Fig. 2E, bottom). The most parsimonious interpretation of these data is that Lgr5⁺ CESCs constitute the most radioresistant population within colonic crypts, which then serve as a stem cell pool to regenerate crypts postradiation, preserving large intestinal mucosa integrity. Furthermore, many crypts display clustering of LacZ⁺ CESCs at the colonic crypt base at day 10 after 19 Gy (Fig. 2D), consistent with the extensive crypt fission often observed at this time postirradiation.

To establish whether surviving Lgr5⁺ CESCs result in colonic tissue regeneration after irradiation, we performed lineage-tracing assays using Lgr5-EGFP-ires-CreERT2/Rosa26-lacZ mice. Figure 3 shows Lgr5⁺ CESC-derived lacZ lineage-tracing crypt units after 19 Gy, indicating that Lgr5⁺ CESCs represent a population, at least in some crypts, that regenerates epithelium postirradiation.

Although our prior studies revealed small intestinal Lgr5⁺ CBCs show biphasic apoptotic/mitotic lethality postradiation, it should be noted that mammalian cells not undergoing apoptosis normally experience growth arrest of about 1 hour per Gy delivered, a number confirmed in our prior study in the small intestines (13). During growth arrest, mammalian cells attempt to repair potentially lethal DNA DSBs, constituting a generic response of eukaryotic cells to ionizing radiation (15). Initial screening studies quantifying the extent of Lgr5⁺ ISC loss in the small and large intestine during the first and second 24-hour time periods after 19 Gy (Fig. 4A and B) revealed a similar loss of about one third of small and large ISCs during the first 24-hour time period, while there was disparate loss during the second 24-hour period. Whereas the remaining two thirds of CBCs in the small intestine were deleted during the second 24-hour period, no significant loss occurred in the large intestine. Because these studies depended on the use of ISH (Lgr5 mRNA), we repeated these studies identifying Lgr5⁺ ISCs using the LacZ reporter gene and obtained virtually identical results (Fig. 4C).

To understand the mechanism of ISC death, we first compared induction of apoptosis in the small and large Lgr5⁺ ISC populations. Figure 4D shows that in the current study, 19 Gy causes an immediate apoptotic response in the ISCs of the small intestine detected by caspase-3 staining, consistent with our published data (13). The pattern and extent of caspase-3⁺ Lgr5⁺ CESCs in distal colon was similar, peaking at 6 hours postradiation, and not statistically different from the small intestine (0.79 ± 0.08 apoptotic CBCs/crypt vs. 0.64 ± 0.08 apoptotic CESCs/crypt; P > 0.05). Thus, differences in apoptotic cell death are likely not the reason for differences in radiosensitivity profiles.

Mammalian cells undergo mitotic arrest postirradiation to facilitate DNA repair, which can be evaluated indirectly by examining kinetics of resolution of γH2AX ionizing radiation–induced repair foci (IRIF), a quantitative surrogate of the repair process. Figure 5A shows that CBCs and CESCs resolve IRIF at different rates with CBCs resolving IRIF more slowly, resulting in statistical differences in number of cells displaying detectable γH2AX foci by 12 hours postirradiation (Fig. 5B; P < 0.01), a difference that persists until at least 18 hours, a time at which CESCs have fully resolved IRIF.

To evaluate potential differences in the period of growth arrest postirradiation, we employed EdU staining to identify (Fig. 6A) and quantify (Fig. 6B) cells undergoing mitosis. Although both small and large intestinal crypts show numerous dividing ISCs and progenitors preceding 19 Gy irradiation, at 12 hours postirradiation, cycling cells are not detected in either small or large intestinal crypts. However, CBCs begin cycling after 12 hours, and by 15 hours, maximal division is reinstituted (1.7 ± 0.1 Edu⁺ CBCs/crypt). By comparing γH2AX data in Fig. 5 with Edu data in Fig. 6, it is estimated that approximately 60% of 19 Gy–irradiated CBCs have not completed DNA repair at the time cell division reinitiates, indicating CBCs are subject to checkpoint adaptation. Consistent with this notion, we find that over the ensuing 24 hours, all of the mitotically active small intestinal CBCs die precipitously. In contrast, CESCs begin to exit growth arrest at 24 hours postirradiation (0.5 ± 0.1 Edu⁺ CESCs/crypts) and reach preradiation cycling levels by 48 hours after 19 Gy. Whereas, the γH2AX data in Fig. 5 indicate DSB repair is complete by 18 hours, a time preceding checkpoint recovery initiation, these data indicate that large-bowel CESCs do not undergo checkpoint adaptation.

Although checkpoint adaptation predisposes affected cells to genomic instability and mitotic death (17), we quantified chromosomal aberrations in the crypts of the small and large intestine postirradiation (29). In unirradiated mice, all crypt cells undergoing mitosis display symmetrically aligned chromosomes (Fig. 7A). At 24 hours after 19 Gy WBR, mitotic cells in both small and large intestine display aberrant mitoses. Common abnormal mitoses include anaphase bridges, multipolar spindles, misaligned chromosomes, and chromosomal lagging (23, 24). Consistent with our γH2AX/Edu data indicating CBCs but not CESCs are subject to checkpoint adaptation, we find significantly more aberrant mitotic figures in small intestinal crypts compared with large intestinal crypts, 0.41 ± 0.04 versus 0.05 ± 0.01 aberrant mitoses/crypt, respectively, an 8-fold difference (Fig. 7B, P < 0.01).

This study provides a unique comparison of two highly similar Lgr5⁺ adult stem cell populations. Although it would have been reasonable to anticipate that the large intestinal Lgr5⁺ ISC would be more radioresistant than the small intestinal Lgr5⁺ CBC, much of what has been learned in the course of the current studies could not have been anticipated. Indeed, we find D₀ values of dose survival curves confirm colonic CESCs to be significantly more radioresistant than small intestinal CBCs, which according to classical tenets of mammalian radiobiology would be interpreted as the two ISC populations differing in inherent capacity and fidelity to repair radiation-induced DSB lesions (3, 30–33). Surprisingly, however, we find differences in radiosensitivity result in large part, if not exclusively, from disparities in kinetics of checkpoint recovery. In this regard, small intestinal CBCs begin to exit growth arrest at 12 hours after 19 Gy, and by 15 hours, CBCs are dividing at a rate equivalent to that prior to irradiation. Previous studies by Chwalinski and Potten (34) demonstrated synchronous checkpoint recovery of murine small intestinal crypt epithelium with mitotic reentry in vivo after a delay of approximately 1 hour per Gy, confirmed in our current and previous (13) studies. Here, we show this checkpoint recovery kinetic associates with the majority of CBCs initiating cell cycling while still manifesting unrepaired DNA DSBs, as judged by residual γH2AX foci (Fig. 5B shows 60% unrepaired at 12 hours), indicating the Lgr5⁺ CBC population is subject to the aberrant DDR of checkpoint adaptation. In contrast, the colonic CESC exhibits DDR-regulated checkpoint recovery only at 48 hours after 19 Gy (∼2-hour growth delay per Gy), coordinated with completion of DSB repair, as evidenced by resolution of γH2AX foci returning to background levels by 18 hours postradiation. The net outcome of the disparate modes of checkpoint recovery is total depletion of the CBC population by 48 hours and small intestinal organ lethality resulting from denudation of the mucosa, while a significant surviving fraction of CESCs regenerate colonic mucosa and rescue organ function, providing a mechanistic basis for the observed colonic radioresistance relative to the small intestines. The discovery that small intestinal CBCs succumb to checkpoint adaptation is the first demonstration that this aberrant cell-cycle response may drive mammalian tissue radiosensitivity.

A number of recent publications in the stem cell literature have attempted to profile the radiosensitivity of ISC populations and their cognate intestinal mucosa (30, 31). However, the criteria for defining organ radiosensitivity and the correlation of ISC lethality to organ damage in these studies did not appear to comply with the rigorous criteria, developed and validated over the past four decades by the radiation community, that are used to define inherent radiosensitivity of normal and tumor tissues (3, 32, 33, 35). Hence, although a recent study argues that Lgr5⁺ stem cells of the colon are more radiosensitive than KRT19⁺ stem cells of the colon based on differences in post-12 Gy lineage tracing (35), we show that at this dose level colonic Lgr5⁺ cells are totally resistant to radiation-induced mitotic lethality and crypt ablation, and hence, there is no stress-induced incentive for lineage tracing. We thus posit that the reason lineage tracing is observed in KRT19⁺ cells is that KRT19 cells undergo significant damage at 12 Gy, leading to KRT19 lineage tracing during the rapid division phase of population recovery. Another less likely possibility is that radiation differentially affects the lineage-tracing assay in the KRT19⁺ and Lgr5⁺ compartments. Whether or not murine KRT19⁺ and Lgr5⁺ stem cells manifest different radiosensitivities or tumors derived thereof will require performance of classic radiation dose survival assays.

In summary, our data broaden and refine the phenotypic features of the Wnt-driven Lgr5⁺ stem cell populations in the small and large intestine, leading to the discovery that Lgr5⁺ ISCs in adult murine small intestine are radiosensitive owing to engagement of checkpoint adaptation. These studies thus show that despite a high degree of similarity in vivo and in vitro between the Lgr5⁺ CBCs and CESCs, there is nonetheless a major difference in their in vivo response to genotoxic stress, indicating radiation phenotype is organotypic rather than generic for genetic drivers, at least in the context of the Lgr5–Wnt program.

No potential conflicts of interest were disclosed.

Conception and design: G. Hua, Z. Fuks, P.B. Paty, R. Kolesnick

Development of methodology: G. Hua, C. Wang, Y. Pan, Z. Zeng, A. Haimovitz-Friedman, Z. Fuks

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Hua, C. Wang, M.L. Martin

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Hua, S.G. Lee, M.L. Martin, A. Haimovitz-Friedman, Z. Fuks, R. Kolesnick

Writing, review, and/or revision of the manuscript: G. Hua, Z. Fuks, R. Kolesnick

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Pan

Study supervision: G. Hua, A. Haimovitz-Friedman, Z. Fuks, P.B. Paty, R. Kolesnick

This work was supported by funds from Mr. William H. Goodwin and Mrs. Alice Goodwin and the Commonwealth Foundationfor Cancer Research and The Experimental Therapeutics Center of MSKCC (R. Kolesnick), CA008748, a NIH/NCI Cancer Center Support Grant P30 (MSKCC), a grant from National Natural Science Foundation of China (31470826), and a gift from the Virginia and D.K. Ludwig Fund for Cancer Research (Z. Fuks).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.