New Research In
Physical Sciences
Social Sciences
Featured Portals
Articles by Topic
Biological Sciences
Featured Portals
Articles by Topic
 Agricultural Sciences
 Anthropology
 Applied Biological Sciences
 Biochemistry
 Biophysics and Computational Biology
 Cell Biology
 Developmental Biology
 Ecology
 Environmental Sciences
 Evolution
 Genetics
 Immunology and Inflammation
 Medical Sciences
 Microbiology
 Neuroscience
 Pharmacology
 Physiology
 Plant Biology
 Population Biology
 Psychological and Cognitive Sciences
 Sustainability Science
 Systems Biology
Polyclonality of familial murine adenomas: Analyses of mouse chimeras with low tumor multiplicity suggest shortrange interactions

Contributed by William F. Dove, March 31, 2005
Abstract
In previous studies demonstrating the polyclonal structure of familial intestinal adenomas, high tumor multiplicity made it difficult to eliminate the possibility that polyclonality arose by the random collision of distinct initiated clones as opposed to some form of clonal interaction. We sought to test further the random collision hypothesis. Chimeric mice carrying the multiple intestinal neoplasia (Min) mutation of the adenomatous polyposis coli gene (Apc) and homozygous for the tumor resistance allele of the Mom1 locus were established. These chimeras also display a strong propensity for tumors of polyclonal structure, despite their markedly reduced tumor multiplicity. Considering tumor sizes and multiplicities, the observed fraction of overtly polyclonal heterotypic adenomas was significantly higher than predicted by the random collision hypothesis. This finding supports models of polyclonality involving interaction among multiple initiated clones. The extent of clonal interaction was assessed by statistical analyses that relate the observed frequency of overtly polyclonal heterotypic tumors to the geometry of the chimeric patches and the pattern of underlying crypts. These statistical calculations indicate that the familial adenomas of the Apc ^{Min/+} mouse may commonly form through interactions between clones as close as 1–2 crypt diameters apart.
Most current models of tumorigenesis involve a monoclonal origin. The evidence for monoclonality at the earliest stages of spontaneous intestinal neoplasia is limited. Intestinal adenomas and carcinomas in carcinogentreated mice (1, 2) and large adenomas in humans (3) seem to be monoclonal. The clonality of normal and neoplastic tissue is assessed by the analysis of patches in chimeras or mosaics heterotypic for a clonal lineage marker. The ability to detect polyclonality in early human tumors is restricted by the large patch sizes of Xinactivation mosaics (4). The first indication that familial intestinal adenomas were commonly polyclonal in structure came from a study of a mosaic XY ↔ XO male patient with familial adenomatous polyposis (5). In that study, 5% of microadenomas were found to have a mixed karyotype, which led the authors to suggest that as many as 76% are polyclonal (ref. 5 but see ref. 6). Novelli et al. (5) suggested that primary adenoma formation involved cooperation between initiated cells or, alternatively, that the mixed karyotype arose by mosaicism within a clonal tumor through loss of the Y chromosome from subpopulations of neoplastic cells, owing to mitotic instability of the dicentric Y chromosome.
In a subsequent study using mice heterozygous for the multiple intestinal neoplasia (Min) allele of the adenomatous polyposis coli gene (Apc) and chimeric for the ubiquitously expressed lineage reporter gene Gt(ROSA)26Sor^{tm1Sor} (ROSA26) expressing lacZ, Merritt et al. (7) observed that 22 of 251 (9%) adenomas were overtly polyclonal by being heterotypic for the blue (ROSA26) and white (nonROSA26) neoplastic lineages. Four hypotheses were entertained:

Mosaicism within the adenoma by loss of the heterozygous lacZ marker: Unlike the dicentric Y chromosome involved in the XY ↔ XO study (5), the lacZ transgene shows no mosaicism in the C57BL/6 (B6) Apc ^{Min/+} mice carrying ROSA26/+, making this explanation implausible.

Epigenetic silencing of lacZ expression: When heterotypic tumors were analyzed for the marker D6Mit36 linked to the lacZ insertion, no white adenomatous regions were found that carried the 129/Sv allele linked to the ROSA26lacZ marker.

Random collision between independently arising neoplasms: The high tumor multiplicity in the chimeric mice of this study, and in the XY ↔ XO human with familial adenomatous polyposis studied by Novelli et al. (5), made this a possible explanation for the origin of polyclonal adenomas (see ref. 7).

Polyclonality via active interaction between multiple initiated clones: Several variations of this model were favored by Merritt et al. (7) including interaction between crypts promoting somatic loss of heterozygosity at the Apc locus; interaction during the transition to dysplastic growth, as in the “community effect” for developmental transitions (8); and interaction between initiated clones favoring tumor survival or growth.
The evaluation of polyclonality in both the familial adenomatous polyposis patient analyzed by Novelli et al. (5) and the chimeric B6 Min mice (mice carrying Apc ^{Min/+}) analyzed by Merritt et al. (7) was compromised by high tumor multiplicity. In the latter case, it was estimated that a model of passive clonal interaction could explain the overall frequency of heterotypic adenomas if the intestinal epithelium is heterogeneous with only a portion of the tissue susceptible to tumorigenesis. Regional heterogeneity could involve exposure to dietary carcinogens. Indeed, recent studies have found that adenomas in B6 Min mice form predominantly in the distal segment of the small intestine (9). We have therefore assessed the incidence of heterotypic tumors in chimeric Min mice in which the tumor multiplicity has been strongly reduced by homozygosity for the resistance allele of the Mom1 modifier locus (Mom1 ^{R}).
In this analysis at low multiplicity, we have explicitly studied the chimeric patch size distribution in the neighborhood of tumors. A statistical analysis of these data indicates that random collision between independent neoplasms is a highly unlikely explanation of the incidence of polyclonal adenomas in familial intestinal neoplasia. More likely, the early adenomas in Min mice involve an interaction among clones that lie as close as 1–2 crypt diameters apart.
Materials and Methods
Mouse Strains and Generation of Aggregation Chimeras. B6 Min mice were obtained from our colony that is maintained by continually backcrossing to B6/J mice (The Jackson Laboratory) imported every fifth generation. B6 mice heterozygous for the ROSA26 transgene expressing LacZ were obtained from The Jackson Laboratory. B6 Mom1^{R/R} mice were obtained from our congenic line that carries a ≈4 × 10^{6}bp region from the resistant AKR strain lying between markers D4Mit54 and D4Mit13 on mouse chromosome 4 (10, 11). The Mom1^{R} resistance region was made homozygous on the B6 Min and B6 ROSA26/+ backgrounds. Aggregation chimeras were generated by fusing embryos from two crosses: B6 Apc ^{+/+} Mom1^{R/R} × B6 Apc ^{Min/+} Mom1^{R/R} and B6 Apc ^{Min/+} Mom1^{R/R} × B6 ROSA26/+ Mom1^{R/R} .
Offspring were screened for Min and the ROSA26 transgene, using PCR assays of DNA isolated from toe clips. Ear snips were also taken from each animal and stained with XGal to ascertain βgalactosidase chimerism. The desired chimeric genotype B6 Apc ^{Min/+} Mom1^{R/R} ↔ B6 Apc ^{Min/+} ROSA26/+ Mom1^{R/R} was found at the expected frequency of 1/8.
Staining and Serial Sectioning Protocol. Whole mounts of the intestines were prepared and stained with XGal (United States Biological, Swampscott, MA; see Supporting Methods, which is published as supporting information on the PNAS web site). Sections were then postfixed in 10% formalin overnight and stored in 70% ethanol. The full intestinal tract was photographed and tumors were excised. All adenomas were embedded in paraffin, serially sectioned in toto, and arrayed as two 5μm sections per slide. Every seventh slide was counterstained with hematoxylin and eosin. When a more complete analysis was required, an additional slide was counterstained with nuclear fast red. Immunohistochemistry for Apc antibody 3122 was performed as described (7), with the exception of the fixation step (see Supporting Methods). With experience, staining for Apc is evident in XGalstained tissue.
Statistical Evaluation of the Random Collision Hypothesis. We investigated whether the polyclonality evident in a heterotypic tumor composed of both blue and white neoplastic lineages is a consequence of independent random initiation events occurring in sufficiently close proximity to result in a neoplasm scored as a single tumor (the random collision hypothesis). We tested this hypothesis by comparing the observed number of heterotypic tumors in a mouse with this quantity's posterior predictive distribution (12). The method was fully developed by Newton (M.A.N., unpublished work). Briefly, the method considers that multiple initiation events are distributed independently and uniformly at random throughout the 2D surface of the small intestine in each mouse. Monoclonal tumors would emerge from initiated clones that lie more than a distance δ from any other initiated clone. Likewise, biclonal tumors would emerge from the passive interaction between two initiated clones lying less than a distance δ from one another and more than that distance from any other initiated clones, and triclonal tumors would emerge from the interaction of three such clones. Higherlevel collisions were ignored in this approximation. Tumor size measurements indicated conservatively that two tumors will be called distinct if δ ≥ 1.5 mm. The formulae of Armitage (13) indicate the approximate expected numbers of monoclonal, biclonal, and triclonal tumors under random collision. Random geometric graph theory (14) justifies the Poisson distribution of the tumor counts. A randomeffects formulation unified the parameters of the different mice and was the basis of a Bayesian analysis to simulate the posterior predictive distribution of heterotypic tumor counts in each mouse. In this posterior simulation, the heterotypic counts are binomialsized subsets of the numbers of biclonal and triclonal tumors. Clones are assigned types independently, according to a Bernoulli distribution. The predictive P value is the probability, under completely random collision, that an animal presenting a certain total number of tumors will also present a heterotypic tumor frequency as large as or larger than that observed.
Chimeric Patch Analysis. Digitized images revealing the chimeric patch structure of the intestinal wall adjacent to tumors were derived from XGalstained wholemount normal tissue for all unambiguously phenotyped tumors in three of the informative chimeric mice. The tissue was photographed from the serosal side so that the patch structure was not obscured by villi. Basic image processing techniques were used to form binary (blue/white) pixel images, which were processed further to measure the statistical properties of the chimeric patch structure. Distance maps show for each pixel p in a binary image the distance d(p) that is the shortest distance from p to pixels of the opposite color. In addition to highlighting boundary regions typical of the loci of heterotypic tumors, these distances were useful in calculations to estimate the spatial extent of interaction among clones in polyclonal tumors. For a given image, let F(r) denote the cumulative proportion of distances d(p) that are no greater than r. It follows that F(r) is also the probability that a disk of radius r, centered randomly on a pixel within the image lattice, will be heterotypic in that it covers both blue and white pixels. Similarly, the complement 1F(r) is the probability that a randomly placed disk of radius r will be homotypic white or homotypic blue. These probabilities formed the basis of a likelihood calculation to estimate the possible spatial extent of interaction among clones of a polyclonal tumor. If polyclonality entails very shortrange interactions (i.e., a small disk in the model), then the observed heterotypic fraction will be low. Conversely, longerrange interactions yield higher heterotypic fractions. In this simple disk model, the number of heterotypic tumors is a binomial random variable with success probability θF(r), where θ is the fraction of polyclonal tumors. For F(r) we used the average distribution over all available images; imagespecific Fs did not improve the estimation.
As an alternative to interactions within a disk, a twopoint interaction model was considered in which the number of heterotypic tumors is again binomial but in which the success probability is θG(r), where G(r) is the proportion of pixel pairs r units apart that are of opposite color; a tumor formed solely by the interaction of exactly two clones originating r units apart would be heterotypic with this probability.
Crypt Reconstruction. XGalstained images reveal the chimeric patch structure but sometimes fail to display the underlying crypt layout. Instead, photographic data from a nonchimeric mouse expressing the cryptspecific ROSA11 transgene (15) show the expected crypt patterning, variable spacing in a roughly hexagonal layout. Reconstructing the crypt layout within the chimeric patch images provided useful information such as the fraction of crypts at chimeric boundaries and the fraction of neighboring crypt pairs that were heterotypic. We applied techniques from Bayesian image analysis (16) to derive crypt reconstructions for a subset of images. The dimension of each chimeric patch image dictated the expected number N of crypts presumed to reside in that region of the intestinal surface. Unknown crypt centers c = {c _{i}: i = 1, 2,.., N} were viewed as points of a spatial point process having probability density π (c), which we sampled by using the Metropolis algorithm to yield a single reconstruction ĉ obtained after 2 × 10^{6} singlecrypt Metropolis moves from a square lattice initial state (17). The form π (c) ∝ π _{0}(c)L(c) ensured that ĉ would exhibit roughly hexagonal patterning and also that the pixels representing a circular crypt i, centered at c _{i}, would be homotypic with high probability. We used a pairpotential Gibbs distribution wherein each pair of crypts d units apart contributes a factor h(d) to π _{0}(c). We set h(d) = 0 if d ≤ 4 pixels, h(d) = 1if d ≤ 5 pixels, and otherwise h(d) = exp{2(d – 5)} following a proposal of Ripley (18). These parameters were derived from the observed layout of 283 crypt centers in the ROSA11 image data using the method of maximum pseudolikelihood (19). The likelihood factor L(c) scored crypt layout c highly if the induced crypts were homotypic in the chimeric patch image. Specifically, we took w(i) and b(i) equal to the numbers of white and blue pixels, respectively, in a cryptsized circle centered at c _{i}, and we took W and B equal to the numbers of white and blue pixels in the intercryptal spacings. With p equal to the overall proportion of blue pixels, and with ε = 0.02 a small impurity rate, the chimeric patch likelihood was: The rationale was that image pixels were partitioned into those close to crypt centers c _{i} and others, in intercryptal space, that could be simply treated as independent and identically distributed Bernoulli trials. Conceptually, crypt i realized a Bernoulli trial to be either blue or white, and then the nearby pixels retained this color except for an ε rate of impurity. Reconstructions sampled from the posterior distribution were subjected to Delaunay triangulation to identify a crypt neighborhood system (20). All statistical computations were done by using the R system (www.rproject.org).
Results
Characterization of Chimeras and Patch Patterns Generated. In this study, we have introduced the Mom1^{R} resistance allele in the homozygous form to reduce the intestinal tumor multiplicity in the B6 background effectively from 130 ± 35 to 16 ± 7 (a factor of 8). In addition, all animals in the study were dissected between 60 and 72 days, minimizing tumor size to ≤3 mm in maximum diameter. Together, these changes enable a very strong test of the random collision hypothesis for the development of polyclonal adenomas.
Wholemount intestinal preparations of animals uniformly carrying the Min allele and chimeric for ROSA26 were analyzed. Five of seven chimeras were found to carry white, blue, and heterotypic tumors when stained with XGal. The other two chimeras that overtly showed only white and heterotypic tumors were also included. In this study, a broad spectrum of chimerism between the ROSA26 and WT lineages was observed. Wholemount examination of intestinal tissue revealed ratios of white to blue tissue ranging from 20% to 85% (Table 1). We explicitly documented the sizes and contours of intestinal patches to develop an analysis of clonal interaction.
Histological Examination and Immunohistochemical Analysis of Adenomas. Intestinal tumors were initially identified by examining XGalstained wholemount sections under a dissecting microscope. All adenomas were photographed, excised, and serially sectioned through the entire tumor (see Materials and Methods). In contrast to previous work, the fixation and staining protocols (see Supporting Methods) permitted much deeper penetration of XGal staining within adenomas (Fig. 1). Tissue within tumors was assessed for neoplastic transformation by inspection of stained slides (see Materials and Methods) based on histopathologic characteristics such as architectural distortion, nuclear atypia, and increased nuclear/cytoplasmic ratio. In some instances, XGalstained sections were counterstained with nuclear fast red to clarify the distribution of XGal staining. We frequently observed adenomas in which histologically normal crypts seemed to lie within the tumor. We examined two heterotypic and two white adenomas by immunohistochemistry with the polyclonal Apc antibody 3122. Within the histologically unambiguous heterotypic adenomas, both the blue and the white tumor tissue were found to be negative for Apc. Our results were consistent with the report of Merritt et al. (7) that all lineages within polyclonal adenomas lose Apc expression. In some tumors composed of both XGalpositive and negative tissue components, it was uncertain based on histologic appearance alone whether both tissue types were transformed. Twelve such ambiguous tumors were observed (Table 1).
Analysis of Tumor Multiplicity. In total, 112 adenomas of the small intestine were identified from the seven chimeric animals having the desired genotype (B6 Apc ^{Min/+} Mom1^{R/R} ↔ B6 Apc ^{Min/+} Mom1^{R/R} ROSA26/+). Of these 112 adenomas, 54 appeared to be homotypic white, 24 homotypic blue, 22 heterotypic, and 12 ambiguous (Table 1). The frequency of unambiguously heterotypic tumors varied between 8% and 63% among the seven animals, with an overall average of 22%. Considering the low tumor multiplicity and the small size of the tumors, this heterotypic fraction is far above that expected by chance if polyclonal tumors arose by the random collision of initiated clones (posterior predictive P < 0.01 for six of seven chimeric mice).
Because heterotypic tumors are overtly polyclonal, an estimated lower bound on the polyclonal fraction is the heterotypic fraction, 8–63% with an average of 22%. A more refined estimate is not possible without assumptions on the nature of polyclonality.
Analysis of Chimeric Patch Patterns. Because polyclonal tumors can be heterotypic only if they form at the boundary between patches of the two chimeric types, we investigated the pattern of patches in a series of images of the intestinal epithelium adjacent to tumors. Fig. 2 illustrates features of the chimeric pattern revealed in a digitized version of the XGal staining in a wholemount section near one tumor and also summarizes characteristics of the chimeric patch structure from similar images adjacent to 47 unambiguously phenotyped tumors in chimeras 100, 122, and 154. These digitized images averaged 7.9 ± 6.9 mm^{2} in size.
Fig. 2 C–E illustrates the observed spatial patterning in terms of the minimum distance d(p) from a pixel p to a point of the opposite type. In this sample image, the average minimum distance was 80 μm, indicating that patches are typically small; pixels were usually close to a boundary. Fig. 2F summarizes such computations for images adjacent to the 47 tumors. As expected, boundary distances increased in images that were dominated by one color; also heterotypic tumors were more frequent in regions where pixels were closer to boundaries.
Analysis of Clonal Interaction via the Disk Model. We combined the image data (Fig. 2) with the tumor counts (Table 1) to estimate the spatial extent of interactions among clones in polyclonal tumors. Fig. 3A shows loglikelihood curves calculated for the radius parameter r of the simple disk model, supposing different levels of polyclonality. The calculation determined how best to explain the observed frequencies of homotypic and heterotypic tumors if polyclonal tumors form by the interaction of all of the clones contained within a disk of radius r that is positioned randomly in a field with the patch characteristics of the observed images. The maximumlikelihood estimate of r depends on the assumed level of polyclonality. For example, if all tumors were polyclonal, the estimate would be 30 μm (95% confidence interval 22–45 μm). This result indicates that very shortrange interactions can explain the heterotypic fraction. Indeed longerrange interactions would be unlikely in this model, owing to the relatively high rate of homotypic tumors, which, in this first case, are all assumed to be polyclonal. Alternatively, if the level of polyclonality is 0.5, some of the homotypic tumors would be monoclonal, and the estimated interaction radius among polyclonal tumors would be 68 μm (95% confidence interval 43–109 μm). Finally, if the level of polyclonality were as low as the frequency of overtly heterotypic adenomas (0.22) the interaction radius would be larger.
Fig. 3B conveys how these estimated interaction distances relate to the crypt layout from the ROSA11 mouse. A twopoint interaction model was also considered in which polyclonal interaction occurs between two and only two clones separated by a distance r. Qualitatively this model mirrors the previous disk model, although it gives limited information on the upper bound to the range of clonal interaction (data not shown).
Analysis of Clonal Interaction by Crypt Reconstruction. The Metropolis sampling method was used to simulate the crypt arrangement in 17 representative images from three of the seven chimeric mice (images adjacent to two blue, two white, and two heterotypic tumors from animals 100, 122, and 154, except just one blue tumor in 154). Delaunay triangulation of the reconstructed crypt centers provided a neighborhood system that, as expected, showed a roughly hexagonal pattern. Fig. 4 A and B shows the triangulated reconstruction for one image. The neighborhood system surrounding a typical crypt includes first, second, and higherorder neighbors (Fig. 4C ). On average over all analyzed images, we computed the number of nthorder neighbors (Fig. 4D ) and the proportion of nthorder neighbors that are isotypic to the centering crypt (Fig. 4E ). Fig. 4F shows the heterotypic fraction under various scenarios of polyclonal tumor formation. For example, if, on average, 50% of all first and secondorder neighboring crypts were to participate in a polyclonal tumor with an initiated crypt, then the probability that this tumor would be heterotypic is ≈60%. This calculation revealed that the observed heterotypic fraction (22%) is easily explained by intercryptal interactions spanning one or two crypt diameters.
The tumor distribution throughout the intestinal tract is significantly different for heterotypic tumors compared with homotypic tumors. Although Min tumor multiplicity is elevated in the distal region of the small intestine (9), the proportion of heterotypic tumors was higher in the proximal region. (Overall, 11 heterotypic and 13 homotypic tumors in the proximal half versus 11 heterotypic and 65 homotypic tumors in the distal half; P = 0.001, χ^{2} test.) An ANOVA indicated no significant difference in patch structure between the proximal and distal regions of the small intestine that could explain these data. In addition, no significant difference in tumor size could be documented between homotypic and heterotypic tumors when compared segment by segment, although the statistical power of this test was limited by the low tumor multiplicity. These findings also weigh against the random collision hypothesis.
Discussion
In this study we have sought first to test rigorously the random collision hypothesis for the formation of polyclonal familial intestinal adenomas. The random collision hypothesis being tested states that every crypt in the intestine acts independently with an equal probability of generating an adenoma. Several lines of evidence against this hypothesis have emerged.
Chimeric mice were established in which the tumor multiplicity was decreased to 16 ± 7 over the entire intestinal tract, a factor of 8 lower than in our previous study (7). This reduction in multiplicity, accompanied by a reduction in tumor size, permitted an explicit calculation of the expected probability distribution for the number of heterotypic tumors. The observed numbers were far higher than predicted by the random collision hypothesis (P < 0.01).
A modification of the random collision hypothesis considers that the intestinal tract is regionally heterogeneous in tumor susceptibility (21). Again, our findings weigh against this regionally restricted collision hypothesis: heterotypic tumors were observed more frequently in the proximal region of the small intestine, where tumor multiplicity is lower than in the distal region (9).
Ruling out both the random and the regionally restricted collision hypotheses, we have next analyzed clonal interaction models by methods of spatial statistics. Through several different statistical analyses, we connected the chimeric patch structure to the heterotypic counts and thus measured the likely spatial extent over which clones interact in the establishment or growth of tumors. The disk model calculations approximate this interaction by supposing that a polyclonal tumor is formed through the participation of crypts that lie within a disk placed uniformly on the intestinal surface. With a relatively high polyclonal fraction, very small disks can explain the observed heterotypic fraction.
More detailed inferences were possible when the crypt pattern, from a ROSA11 mouse, was superimposed on the chimeric pattern (Fig. 4). This analysis demonstrates that, even if an initiated crypt interacts with only a small fraction of its nearest neighboring crypts, the predicted heterotypic fraction will meet the observed 22%.
Our studies of familial intestinal adenomagenesis indicate the importance, if not the necessity, of shortrange clonal interactions. These interactions can either involve groups of nearestneighbor monoclonal crypts or else distinct clones within the single polyclonal crypts that are common in neonates before crypt purification (22). Our analysis of chimeric patch patterns is consistent with a proportion of polyclonal adenomas ranging from the overtly heterotypic incidence of 22% to 100%. Although these chimeras have permitted spatial resolution superior to that of Xinactivation mosaics in the human (4), still higher resolution will be needed to refine this estimate, to determine whether polyclonality is a necessary condition for adenomagenesis, and to ascertain whether polyclonal single crypts are important progenitors.
The inference of shortrange clonal interactions in familial intestinal neoplasia seems inconsistent with the observation that each participating clone has lost the expression of the WT Apc allele (ref. 7 and this study). Hits in Apc/APC occur early during tumorigenesis in the mammalian intestine (14, 23). Haigis and Dove (24) demonstrated that the WT allele of Apc is lost by somatic recombination during intestinal tumorigenesis in B6 Min mice. How can multiple lossofheterozygosity events arise simultaneously, either in a very small neighborhood or within a single polyclonal crypt? One possibility is microheterogeneity in tumor susceptibility. Regions of stroma may promote loss of heterozygosity in the associated epithelium. Alternatively, loss of Apc/APC in one epithelial clone may induce the expression of growth factors that affect mitotic rates in neighboring clones. The normal intestinal epithelium adjacent to tumors is often hyperplastic (25). Thus, the multiple hits in Apc/APC that occur in polyclonal familial adenomas in the mammalian intestine may not be independent of one another.
This microheterogeneity would give rise to epithelial neoplasms that are focal even if not clonal. If several clones within a focus simply share a neighborhood or niche, but do not exchange biologically relevant signals, this focal interaction would have a passive character. On the other side, an apparently clonal early neoplasm may involve interaction between progenitors. For example, crypt fission is frequent in the colon of familial adenomatous polyposis patients (26). If initiated sister crypts formed by fission were to cooperate in adenomagenesis, the emergent tumors would be monotypic for a lineage marker.
Clearly the analysis of clonal interactions in the establishment, growth, and progression of intestinal neoplasms (familial, sporadic, and carcinogeninduced) deserves the continuing attention of cancer biologists and statisticians.
Acknowledgments
We thank the McArdle Histotechnology Group, in particular Jane Weeks and Harlene Edwards, for outstanding assistance. Our manuscript has received careful critical review from Norman Drinkwater, Richard Johnson, Cheri Pasch, and Alexandra Shedlovsky. Our research was supported by National Cancer Institute Grants R37CA63677, U01CA84227, K08CA84146, and R01CA64364. This is publication no. 3622 from the Laboratory of Genetics, University of Wisconsin.
Footnotes

↵ ¶ To whom correspondence should be addressed. Email: dove{at}oncology.wisc.edu.

↵ ‡ A.T.T. and R.B.H. contributed equally to this work.

Author contributions: A.T.T., R.B.H., L.C., and W.F.D. designed research; A.T.T., R.B.H., and L.C. performed research; A.T.T., R.B.H., S.S., and M.A.N. contributed new reagents/analytic tools; A.T.T., R.B.H., L.C., W.F.D., R.S., M.K.W., S.S., and M.A.N. analyzed data; and A.T.T., R.B.H., L.C., W.F.D., and M.A.N. wrote the paper.

Abbreviations: Apc, adenomatous polyposis coli gene; B6, C57BL/6; Min, multiple intestinal neoplasia; Min mice, mice carrying Apc ^{Min/+}; ROSA26, Gt(ROSA)26Sor^{tm1Sor} .
 Copyright © 2005, The National Academy of Sciences
References
 ↵
 ↵

↵
Fearon, E. R., Hamilton, S. R. & Vogelstein, B. (1987) Science 238 , 193–197. pmid:2889267

↵
Novelli, M., Cossu, A., Oukrif, D., Quaglia, A., Lakhani, S., Poulsom, R., Sasieni, P., Carta, P., Contini, M., Pasca, A., et al. (2003) Proc. Natl. Acad. Sci. USA 100 , 3311–3314. pmid:12610207
 ↵

↵
Newton, M. A. (2005) On Estimating the Polyclonal Fraction in LineageMarker Studies of Tumor Origin, University of Wisconsin, Department of Statistics, Technical Report 1099 (University of Wisconsin, Madison).

↵
Merritt, A. J., Gould, K. A. & Dove, W. F. (1997) Proc. Natl. Acad. Sci. USA 94 , 13927–13931. pmid:9391129
 ↵

↵
Haigis, K. M., Hoff, P. D., White, A., Shoemaker, A. R., Halberg, R. B. & Dove, W. F. (2004) Proc. Natl. Acad. Sci. USA 101 , 9769–9773. pmid:15210940

↵
Gould, K. A., Luongo, C., Moser, A. R., McNeley, M. K., Borenstein, N., Shedlovsky, A., Dove, W. F., Hong, K., Dietrich, W. F. & Lander, E. S. (1996) Genetics 144 , 1777–1785. pmid:8978063
 ↵

↵
Gelman, A., Carlin, J. B., Stern, H. S. & Rubin, D. B. (2004) Bayesian Data Analysis (Chapman & Hall, Boca Raton, FL).

↵
Armitage, P. (1949) Biometrika 36 , 257–266. pmid:15402066

↵
Penrose, M. (2003) Random Geometric Graphs (Oxford Univ. Press, Oxford).
 ↵

↵
Besag, J. (1986) J. R. Stat. Soc. B 48 , 259–302.

↵
Liu, J. S. (2001) Monte Carlo Strategies in Scientific Computing (Springer, New York).

↵
Ripley, B. D. (1977) J. R. Stat. Soc. B 39 , 172–212.

↵
Besag, J. (1974) J. R. Stat. Soc. B 36 , 192–236.

↵
Lee, D. T. & Schacter, B. J. (1983) Int. J. Comput. Information Sci. 9 , 219–242.

↵
Dove, W. F., Cormier, R. T., Gould, K. A., Halberg, R. B., Merritt, A. J., Newton, M. A. & Shoemaker, A. R. (1998) Philos. Trans. R. Soc. London B 353 , 915–923. pmid:9684289

↵
Schmidt, G. H., Winton, D. J. & Ponder, B. A. J. (1988) Development (Cambridge, U.K.) 103 , 785–790.

↵
Levy, D. B., Smith, K. J., BeazerBarclay, Y., Hamilton, S. R., Vogelstein, B. & Kinzler, K. W. (1994) Cancer Res. 54 , 5953–5958. pmid:7954428
 ↵
 ↵
 ↵
Citation Manager Formats
More Articles of This Classification
Biological Sciences
Medical Sciences
Related Content
 No related articles found.
Cited by...
 Mutational Heterogeneity in APC and KRAS Arises at the Crypt Level and Leads to Polyclonality in Early Colorectal Tumorigenesis
 Oncogenic Kras drives invasion and maintains metastases in colorectal cancer
 Regulated Expression of Chromobox Homolog 5 Revealed in Tumors of ApcMin/+ ROSA11 Gene Trap Mice
 Understanding Intratumoral Heterogeneity: Lessons from the Analysis of AtRisk Tissue and Premalignant Lesions in the Colon
 Transformation of epithelial cells through recruitment leads to polyclonal intestinal tumors
 Interpretation of clonality and Xchromosome inactivation assays urge attention
 Monoallelic silencing and haploinsufficiency in early murine intestinal neoplasms
 Clonal Structure of CarcinogenInduced Intestinal Tumors in Mice
 Cyp1b1 Exerts Opposing Effects on Intestinal Tumorigenesis via Exogenous and Endogenous Substrates
 Cell Lineage Analysis of a Mouse Tumor
 Expansion of a mutated clone: from stem cell to tumour
 Suppression of Intestinal Neoplasia by Deletion of Dnmt3b
 Opposing effects of DNA hypomethylation on intestinal and liver carcinogenesis