NOX1 selectively enriches for LGR5-expressing stem cells in the colon
To identify markers specifically enriched in colonic stem cells, we profiled the transcriptome of fluorescence-activated cell sorting (FACS)-isolated GFP and GFP cell populations from the small intestine and colon of our non-variegated Lgr5-2A-eGFP reporter mice (Fig. 1a). Comparison of their respective GFP and GFP transcriptomes revealed the signature of the LGR5-expressing stem cells in each region. As expected, there was a significant overlap with the previous colon LGR5 stem cell signature generated using Lgr5-eGFP-IRES-CreERT2 reporter mice, including many Wnt target genes (Extended Data Fig. 1a). The LGR5 stem cell signature in the colon overlapped with a published mouse CD44 colon stem cell signature, including validated stem cell markers such as achaete-scute family BHLH transcription factor 2 (ASCL2) (Supplementary Table 1). We identified the transcriptional signature of GFP stem cells for each tissue by screening it against the corresponding GFP cell population. We then screened the list of genes enriched in the colon GFP stem cells against the GFP stem cell transcriptional signature of the small intestine and pylorus to identify genes selectively upregulated in colonic stem cells (Fig. 1b and Supplementary Table 2). One of the most highly upregulated targets was Nox1, which is known to be enriched in the distal gastrointestinal tract (Fig. 1c). Nevertheless, a detailed characterization of the endogenous NOX1 expression pattern and functional evaluation of the stem potential of NOX1-expressing cells within the colon has not yet been performed.
To identify a more regionally restricted colon-specific stem cell marker, we conducted a new series of RNA sequencing (RNA-seq) analyses. Working with Lgr5-2A-eGFP mice, we separated the small intestine into three regions to mimic the anatomical separation of this tissue (that is, the duodenum, jejunum and ileum) and divided the colon into two regions (proximal and distal) (Fig. 1d). We then identified the gene signature for each region by comparing the GFP and GFP transcriptomes (Supplementary Tables 3 and 4). Finally, we compared the list of genes enriched in the proximal or distal colon against the GFP stem cell transcriptional signature of the small intestine (all regions) and pylorus. We also compared the list of genes enriched in the distal colon against that of the proximal colon and vice versa to isolate a specific gene signature for each region of the colon (Fig. 1e and Supplementary Tables 5 and 6). Although we failed to validate via quantitative PCR (qPCR) and RNAscope a robustly expressed, selective marker of LGR5 stem cells in the proximal colon (Extended Data Fig. 1b and Supplementary Table 5), we identified Npy1r as one of the most highly enriched genes in the LGR5 population of the distal colon (Fig. 1f and Supplementary Table 6). NPY1R is known to be expressed in a subset of enteroendocrine cells in the gastrointestinal tract, but no stem cell enrichment has been described.
To precisely define which lineages express NOX1 and NPY1R, we performed single-cell RNA-seq (scRNA-seq) on the epithelial fraction of the small intestine and colon. The small intestine was separated into three parts that were identical to the ones used for RNA-seq analysis (Fig. 1d). The colon was separated into four parts: the caecum and the proximal, middle and distal colon. Epithelial fractions were separated for each tissue the and cells were dissociated and embedded into chromium beads before RNA-seq. Using the shared nearest neighbour (SNN) modularity optimization-based clustering algorithm (implemented in Seurat), we identified 15, 14, 11 and 13 clusters for the caecum and proximal, middle and distal colon epithelium, respectively. Consensus lineage marker expression was plotted for each cluster of each tissue compartment (Fig. 2a) to assign clusters to their respective lineages (Fig. 2b). We then mapped LGR5, NOX1 and NPY1R expression distribution on uniform manifold approximation and projection (UMAP) plots generated for each tissue (Fig. 2c) against the stem cell clusters delineated according to their assigned lineage identities (Fig. 2b). We found that NOX1 was expressed across many clusters at a very low level but highly upregulated within the stem cell cluster for all parts of the colonic epithelium (Fig. 2c), in line with our RNA-seq data. In contrast, NPY1R-expressing cells overlapped with the stem cell cluster specifically in the middle and distal colon regions, with a higher expression in the distal colon, but not in the caecum or proximal colon (Fig. 2c). As was previously shown, a subpopulation of NPY1R cells was found in the chromogranin A (CHGA) enteroendocrine cell cluster of the colon (Fig. 2c and Extended Data Fig. 1c). Interestingly, we also found a subpopulation of NPY1R-expressing cells in the tuft cell cluster in the distal colon (Fig. 2c).
We quantified the total number of LGR5 cells for each cluster and their expression status with respect to NOX1 and NPY1R expression (Extended Data Fig. 1d). We found that in the caecum 75.75% of the LGR5 stem cells co-expressed NOX1. Surprisingly, NOX1 was found to be co-expressed in only 29.15% of LGR5 stem cells in the proximal colon. This percentage increased towards the middle and distal part of the colon, with 84.52 and 77.73% of LGR5 stem cells co-expressing NOX1, respectively (Extended Data Fig. 1d). When looking at NPY1R expression in LGR5 stem cells, we showed that the percentage of co-expression increased from the middle to the distal colon (27.61 and 62.28%, respectively) and that the majority of the LGR5 stem cells (52.73%) co-expressed both NOX1 and NPY1R (Extended Data Fig. 1d). When looking at NOX1 and NPY1R cells for each cluster, we also found that the majority of these cells were within the stem cell cluster (Extended Data Fig. 1e,f). For the caecum, we found that 66.08% of NOX1 stem cells expressed LGR5. This percentage remained constant in the proximal colon (65.22%) and distal colon (67.06%), but decreased in the middle colon (42.62%) (Extended Data Fig. 1e). NPY1R stem cells were found to be present predominantly in the distal colon, with most co-expressing LGR5 and NOX1 (Extended Data Fig. 1f). Overall, these data show that NOX1 is expressed in the stem cell compartment of the entire large intestine, with the highest overlap in the caecum and distal colon (Extended Data Fig. 1d,e), whereas NPY1R expression is highly restricted to the distal colon stem cell compartment (Extended Data Fig. 1d,f).
For our small intestine scRNA-seq dataset, we again used the SNN modularity optimization-based clustering algorithm to identify 14, 15 and 16 clusters for the proximal, middle and distal small intestine epithelium, respectively. We plotted the key lineage marker expression for each cluster of each region (Extended Data Fig. 2a) in order to assign each cluster to its respective lineage identity on a UMAP plot (Extended Data Fig. 2b). We found NOX1 expression to be distributed at extremely low, almost undetectable levels throughout all clusters of each part of the small intestine (Extended Data Fig. 2c). In contrast, we did not detect any NPY1R cells in all regions of the small intestine except a subpopulation within the enteroendocrine cluster, which presented a very low level of NPY1R expression (Extended Data Fig. 2c). Taken together, these results demonstrate that NOX1 is expressed in stem cell compartments throughout the colon, with the highest enrichment present within the caecum and middle and distal regions. In contrast, NPY1R is selectively enriched in LGR5 stem cell compartments of the middle and distal colon, with the highest enrichment in distal regions.
Next, we validated our in silico transcriptomic analyses by first establishing endogenous Nox1 RNA expression profiles within different regions of the colon using RNAscope. In the caecum and distal colorectum, the Nox1 expression pattern was highly similar to that of Lgr5, with transcripts restricted to crypt bases (Fig. 3a). In the proximal colon, Nox1 expression was similarly restricted to crypt bases but was only detected in a minority of crypts and at notably lower levels than Lgr5 (Fig. 3a). In the middle colon, Nox1 expression extended beyond the Lgr5 stem cell compartment at the crypt base to encompass presumptive early transit-amplifying cells (Fig. 3a).
Nox1 expression was largely absent from the stomach, duodenum and jejunum (Extended Data Fig. 3a), except for intestinal crypts directly adjacent to Peyer's patches, where expression was restricted to crypt bases (Extended Data Fig. 3b). Limited Nox1 expression was also detected at crypt bases in the most distal part of the ileum, with robust expression present around the +4 to +6 positions immediately adjacent to the Paneth cell compartment and low levels in a subset of crypt base columnar cells (Extended Data Fig. 3b).
Collectively, these expression analyses corroborate the in silico data, identifying Nox1 as being highly and selectively expressed in Lgr5 epithelial stem cell compartments throughout the adult colon, with the greatest enrichment present within the caecum and distal colorectum.
In contrast with Nox1, Npy1r expression was absent from the proximal colon and caecum of adult mice (Fig. 3a). In the middle colon, relatively weak Npy1r expression extended beyond the Lgr5 zone at the crypt base to encompass the lower third of the crypts in a pattern reminiscent of Nox1 expression in this region (Fig. 3a). More robust levels of Npy1r were present exclusively within the Lgr5 zone at crypt bases throughout the distal colorectum (Fig. 3a). Npy1r expression was absent from epithelia of the small intestine, stomach and bladder, but exhibited robust levels in bladder stroma where Npy1r is known to mark a subset of fibroblast cells (Extended Data Fig. 3c,d). Again, these expression analyses aligned well with our in silico data, identifying Npy1r as being highly enriched in Lgr5 epithelial stem cell compartments predominantly within the distal adult colorectum.
To further characterize the various NOX1- and NPY1R-expressing colon populations, we generated independent eGFP reporter mouse models via targeted integration of either 2A-eGFP or eGFP-IRES-CreERT2 cassettes at the Nox1 or Npy1r locus, respectively (Fig. 3b,c). GFP expression in adult Nox1-2A-eGFP reporter mice faithfully recapitulated endogenous NOX1 expression across all regions of the colon and ileum (Fig. 3b and Extended Data Fig. 4a). Similarly, GFP expression in Npy1r-eGFP-IRES-CreERT2 reporter mice was predominantly restricted to crypt bases of the middle and distal colorectum, recapitulating the endogenous expression pattern (Fig. 3c). Sporadic reporter expression was also observed scattered in stromal tissues throughout the colon and rectum (Fig. 3c) and bladder (Extended Data Fig. 4b).
Next, we sorted the GFP and GFP cell populations from the different adult reporter mice by FACS to document the lineage identities of the NOX1 and NPY1R colon populations using established markers via qPCR. For both NOX1-GFP and NPY1R-GFP cell populations, Nox1 and Npy1r, respectively, together with Lgr5 and the stem cell markers SPARC-related modular calcium-binding protein 2 (Smoc2) and Ascl2, were robustly enriched compared with in GFP populations (Extended Data Fig. 5a), further underscoring their relatively undifferentiated, stem-like identity. Npy1r was not enriched in the caecum, in line with our scRNA-seq and staining data (Extended Data Fig. 5a). Conversely, expression levels of the lineage markers downregulated in adenomas (Dra, also called Slc26a3 -- a colonocyte marker), Mucin2 (Muc2 -- a goblet marker) and regenerating family member 4 (Reg4 -- a deep crypt secretory cell marker) were markedly lower in GFP compared with GFP cell populations. The expression of Chga (an enteroendocrine marker) was also relatively low in NOX1 populations. In contrast, NPY1R populations in the distal colon displayed robust ChgA expression (Extended Data Fig. 5a), consistent with observations from scRNA-seq analyses documenting the presence of enteroendocrine cells within this population (Fig. 2 and Extended Data Fig. 1c).
Co-immunofluorescence (co-IF) of GFP with CHGA, MUC2 and DRA across the different regions of the colon from adult NOX1 reporter mice confirmed a lack of co-localization with these mature lineage markers, further highlighting the relatively undifferentiated status of NOX1 populations (Extended Data Fig. 5b). NPY1R-GFP colon populations similarly lacked co-expression of MUC2 or DRA markers, but did partially co-localize with CHGA, confirming that a subset of NPY1R cells are enteroendocrine cells (Extended Data Fig. 5c). A subset of NOX1-GFP and NPY1R-GFP cells overlapped with the proliferation marker Ki-67, highlighting their proliferative status (Extended Data Fig. 5b,c).
To evaluate the stem cell potential of NOX1-expressing colon cells, we FACS-sorted NOX1-GFP and NOX1-GFP cells from the caecum and distal colon of adult Nox1-2A-eGFP reporter mice and assessed their capacity to form long-term organoid cultures in vitro (Fig. 4a). The organoid outgrowth efficacy of plated GFP cells relative to GFP cells was 11-fold higher in the caecum and 5.5-fold higher for the distal colon (Fig. 4b). Importantly, the limited numbers of caecum and colon organoids generated from GFP cells could not be maintained beyond the first passage, in contrast with GFP cell-derived organoids, which could be cultured long term (more than ten passages) (Fig. 4c).
We also FACS sorted NPY1R-GFP and NPY1R-GFP cells from the distal colon of adult Npy1r-eGFP-IRES-CreERT2 reporter mice (Fig. 4d). The organoid outgrowth efficacy of plated GFP cells relative to GFP cells was 5.5-fold higher (Fig. 4e). Again, the limited number of GFP cell-derived organoids could not be maintained beyond the first passage, in contrast with the GFP cell-derived organoids that could be passaged long term (Fig. 4f).
To formally document the in vitro multipotency of the plated NOX1 and NPY1R cells, we performed a differentiation assay. Organoids derived from either NOX1 or NPY1R cells were cultured in growth medium for 2 d before switching to a differentiation medium for 3 and 5 d for the distal colon and caecum, respectively (Fig. 4g). qPCR analyses on NOX1 cell-derived caecum and distal colon organoids documented a major decrease in both Nox1 and Lgr5 expression and a concomitant increase in differentiated lineage marker expression (Fig. 4h,i). NPY1R cell-derived distal colon organoids similarly displayed a marked increase in differentiated lineage marker expression, with only a modest accompanying decrease in Npy1r, probably due to the presence of Npy1r-expressing enteroendocrine cells (Fig. 4j). Collectively these results identify NOX1 populations from both the caecum and distal colon and the NPY1R population from the distal colon as being highly enriched for cells with robust stem cell potential in vitro.
To functionally evaluate the endogenous stem cell identity of NOX1-expressing cells within the adult colonic epithelium, we generated a Nox1-2A-CreERT2/Rosa26-tdTomato (tdTom) mouse model to track the contribution of endogenous NOX1-expressing cells to homeostatic epithelial renewal via in vivo lineage tracing (Fig. 5a). At 24 h after tamoxifen induction, cells at the base of the crypts of the colon expressed tdTom, with only a subset of labelled crypts in the proximal colon, recapitulating the expression profiles observed via RNAscope analyses and the eGFP reporter model (Fig. 5b). One week after induction, tdTom-labelled progeny had expanded to encompass most of the crypts (Fig. 5b). One month after tamoxifen induction, the vast majority of crypts and the associated surface epithelium in the caecum (97.7%), middle (91.7%) and distal colon (92.5%) and rectum (75%), together with scattered crypts in the proximal colon, were entirely tdTom positive (Fig. 5b and Extended Data Fig. 6a). Co-IF analyses of tdTom with major cell lineage markers confirmed overlapping expression in tracing units across all regions of the colon (Fig. 5c). Collectively, these observations identify NOX1 populations as harbouring self-renewing, multipotent stem cells contributing to epithelial homeostasis throughout the colon.
We further explored the clonal dynamics of lineage tracing using the Nox1-2A-CreERT2 driver by administering a low dose of tamoxifen (0.1 mg per 30 g bodyweight) to initiate recombination at low frequency. This tamoxifen dose is sufficient to activate tdTom expression within single cells at colonic crypt bases after 24 h in both the middle and distal colon (top panel of Extended Data Fig. 6b). As expected, tracing units initiated from these single NOX1 cells had expanded upwards from the crypt base after 1 week (middle panel of Extended Data Fig. 6b). A subset of the tracing units further expanded to encompass entire crypts by 1 month (bottom panel of Extended Data Fig. 6b), confirming that individual NOX1-expressing stem cells have the potential to populate entire colonic crypts during epithelial renewal. With this low level of induction, the proportion of crypts containing tdTom cells rapidly decreased over time, from 95.66 ± 3.44 and 87.33 ± 6.15% of crypts at 24 h to 48.44 ± 14.37 and 44 ± 8.88% of crypts after 1 week and 31.11 ± 9.11 and 34.44 ± 7.40% of crypts after 1 month in the middle and distal colon, respectively (Extended Data Fig. 6c). The decreases in the number of tdTom crypts 24 h and 1 month after induction in the middle and distal colon were statistically significant (P < 0.0001), probably reflecting known stem cell competition dynamics between labelled and non-labelled colonic stem cells in these crypts. In contrast, using the same low tamoxifen dose, we found that a larger pool of NOX1-expressing cells were labelled at crypt bases of the caecum after 24 h (Extended Data Fig. 6b), resulting in a higher frequency of retained lineage tracing after 1 month (70.4 ± 10.98% tdTom caecum crypts after 24 h and 61.33 ± 9% after 1 month; P = 0.0674) (Extended Data Fig. 6c). Thus, NOX1-expressing cells present distinct lineage tracing dynamics in different regions of the colon, with more efficient recombination and retention of clones within the caecum.
In the small intestine, long-term tracing was only observed in the distal ileum and crypts adjacent to the Peyer's patches (Extended Data Fig. 7a,b), in line with the endogenous NOX1 expression observed within the LGR5 stem cell compartments of those regions (Extended Data Figs. 3b and 4a). Following induction with 3 mg tamoxifen per 30 g bodyweight, lineage tracing in the ileum originated predominantly from presumptive early transit-amplifying cells immediately adjacent to the Paneth cell compartment, although tdTom expression was also observed in limited numbers of crypt base columnar cells interspersed with Paneth cells at the crypt base after 24 h (Extended Data Fig. 7a). Accordingly, limited numbers of tdTom ribbons spanning crypt-villus units present in the distal ileum after 1 week were retained after 1 month (Extended Data Fig. 7a). In contrast, multiple tdTom cells were observed at the base of crypts adjacent to Peyer's patches after 24 h (Extended Data Fig. 7b). After 1 week, tracing units had expanded along the entire crypt-villus axis and the follicle-associated surface epithelium (Extended Data Fig. 7b). By 1 month, tracing units had further expanded to completely encompass the Peyer's patch-associated crypt-villus epithelium and the follicle lining epithelium (Extended Data Fig. 7b). These data highlight Nox1-2A-CreERT2 as being a highly selective colonic stem cell driver that readily facilitates recombination throughout the colon, along with accompanying sporadic recombination within the ileum.
To evaluate the endogenous stem cell identity of NPY1R colon cells, we performed in vivo lineage tracing using Npy1r-eGFP-IRES-CreERT2, tdTom mice under homeostatic conditions (Fig. 6a). Due to the relatively low endogenous expression levels of NPY1R in the colon, we increased the tamoxifen dose to 4 mg per 30 g bodyweight to ensure robust recombination levels. At 24 h post-induction, single tdTom cells were visible at the crypt bases within the middle colon, whereas a higher recombination frequency was achieved throughout the distal colon and rectum, with multiple tdTom cells present at most crypt bases (Fig. 6b). After 1 week, basal tdTom tracing units had expanded to reach the surface epithelium in multiple crypts in both the middle and distal colon and rectum, albeit at a higher frequency within more distal and rectal regions (Fig. 6b). The tracing frequency in the distal colon and rectum was maintained after 1 month, with 74.8% of colon crypts and associated surface epithelium displaying uniform tdTom expression and 81.8% of rectum crypts retaining tdTom labelling. Completely traced crypts were also present in the middle colon at this time point, albeit at a reduced frequency (22.7%) reflecting the lower induction efficacy in this region (Fig. 6b and Extended Data Fig. 8a). As expected, recombination within the crypts of the proximal colon was exclusively initiated within infrequent presumptive enteroendocrine cells at 24 h, but this was not retained at later time points (Fig. 6b).
To formally document the multipotency of NPY1R colon cells, we performed red fluorescent protein (RFP) co-IF with CHGA, DRA and MUC2. Each lineage marker could be detected within tdTom tracing units present in the middle and distal colon, confirming that NPY1R cells can generate the various functional lineages of the colonic epithelium (Fig. 6c). We therefore conclude that NPY1R crypt-base-resident populations in the middle and distal colorectum are highly enriched for long-term self-renewing, multipotent cells contributing to homeostatic epithelial renewal.
We additionally performed lineage tracing in the bladder to assess the behaviour of resident NPY1R populations. After 24 h induction, tdTom cells were detected within the stroma of the epithelium. This labelling was retained over time after 1 week and 1 month of tracing, indicating that NPY1R marks long-lived stromal cells in the bladder (Extended Data Fig. 8b). We performed co-IF staining of RFP with lineage markers. RFP cells failed to co-localize with the epithelium marker E-cadherin and the smooth muscle marker alpha smooth muscle actin. In contrast, RFP cells were uniformly positive for the mesenchyme marker vimentin, confirming the stromal identity of NPY1R bladder cells (Extended Data Fig. 8c).
LGR5 stem cells in the small intestine are major cells of cancer origin following targeted dysregulation of Wnt signalling in vivo using Lgr5-CreERT2 drivers. The ensuing rapid tumour growth in the small intestine causes rapid lethality, precluding rigorous evaluation of LGR5 stem cell populations as cancer origins in the different regions of the colon. Having established the selective enrichment of NOX1 in stem cells within the caecum and middle and distal colon, we reasoned that the Nox1-2A-CreERT2 driver should facilitate selective targeting of oncogenic mutations to the genome of colon stem cells as an essential prerequisite to evaluating their role in driving colon cancer initiation. Hyperactivation of Wnt signalling via loss-of-function mutation of APC is a highly prevalent early event in human CRC. We therefore generated a Nox1-2A-CreERT2;Apc mouse model that incorporates a tdTom reporter allele to facilitate the visualization of any tumour formation in the gastrointestinal tract following targeted transformation of NOX1 stem cells (Fig. 7a). Non-induced adult mice were healthy, displayed very low levels of spontaneous recombination (as evidenced by <1 in 500 tdTom colon crypts) and presented no evidence of Wnt pathway hyperactivation with the absence of β-catenin accumulation within all regions of the colonic epithelium (Extended Data Fig. 9a). One month after tamoxifen administration, nucleocytoplasmic accumulation of β-catenin (β-cat) was apparent throughout the caecum and within discrete regions of the middle and distal colon (Fig. 7a). Co-IF for Ki-67 and β-catenin revealed that a subset of the β-cat cells were actively proliferating in the different regions (Fig. 7b). Compared with control tissues, these Wnt-hyperactivated crypts were highly disorganized and displayed prominent signs of hyperplasia (Fig. 7a,b). In line with the higher efficiency of recombination within the caecum (Extended Data Fig. 6b,c), the caecum presented more polyps and a markedly higher proportion of crypts displaying nucleocytoplasmic β-catenin accumulation compared with crypts present in the middle and distal colon (Fig. 7a). To a lesser extent, we note an accumulation of β-catenin in the Peyer's patches of the small intestine within small proliferative cell clusters (Extended Data Fig. 9b,c). Crucially, unlike previous LGR5-driven cancer models in which small intestinal tumours rapidly develop, we did not detect any nucleocytoplasmic accumulation of β-catenin within the ileum, even after 1 month of induction (Extended Data Fig. 9b), highlighting the selectivity of the Nox1-2A-CreERT2 driver for initiating cancers within the colon as an essential prerequisite to modelling more advanced cancers in the caecum and distal colon via the incorporation of additional, physiologically relevant mutations.
Given the highly efficient recombination achieved in the caecum relative to the distal colon, we reasoned that reducing the tamoxifen induction dose would enable selective targeting of mutations affecting the caecum epithelium to facilitate the generation of a caecum-specific cancer model. As caecum cancer presents the highest rate of Kras mutation compared with other forms of CRC, we incorporated a conditional Kras allele into our Nox1-2A-CreERT2, Apc mouse model (Fig. 7c). Following low dose tamoxifen induction of adult mice at 1 mg per 30 g bodyweight, we noted a major decrease of β-cat polyps in the distal colon (one polyp was detected over four independent mouse colons), with no accompanying hyperactivation of phosphorylated MAPK (pMAPK). In contrast, the entire caecum epithelium harboured β-cat polyps displaying robust hyperactivation of pMAPK, indicating a high recombination efficacy for both Apc and Kras alleles in the tissue (Fig. 7c). No β-cat polyps were visible in the small intestine for both the ileum and Peyer's patches, and pMAPK activity was confined to the crypt base columnar cells and transit-amplifying compartment, as seen in non-induced mice (Extended Data Fig. 9d).
To further demonstrate the value of using the Nox1-2A-CreERT2 driver to generate more advanced cancers, we incorporated the conditional knockout mutant allele p53 into our model. We reasoned that incorporating a single floxed Apc allele to this compound mutant model would ensure a limited tumour load resulting from sporadic loss of heterozygosity of the second wild-type Apc allele, thereby ensuring a lifespan compatible with cancer progression, as previously shown. Indeed, induction of the resulting Nox1-ATK mice resulted in the formation of an average of four to six large tumours in the caecum after 3 months, including advanced tumours independently characterized by a certified animal pathologist as being T2-stage adenocarcinomas undergoing robust invasion of the underlying submucosal layers. Immunohistochemistry (IHC) analyses confirmed successful recombination of the three conditional alleles in these tumours and loss of heterozygosity of Apc, as highlighted by β-catenin accumulation, pMAPK expression and loss of p53 expression (Fig. 7d). These results collectively identify NOX1 stem cells in the caecum and distal colon as key sources of cancer following in vivo mutation. Importantly, lower tamoxifen induction doses facilitate highly efficient targeting of conditional mutations affecting the caecum to selectively drive advanced cancer in this region.
The well-established gene signature for distal CRC involves the sequential mutation of Apc, Kras, Trp53 and Smad4 alleles. First, we incorporated the conditional Apc and conditional Kras alleles by crossing with our Npy1r-eGFP-IRES-CreERT2, LoxP-Stop-LoxP (LSL)-tdTom mice (Fig. 8a). Non-induced mice showed no ectopic Cre activation in the colon (Extended Data Fig. 10a). Because of the low level of NPY1R expression, we increased the tamoxifen dosing regimen to three consecutive doses of 4 mg per 30 g bodyweight every 2 d. Due to the relatively slow growth of distal colon tumours compared with cancers originating in the caecum (Fig. 7b), mice were harvested at an extended timepoint of 8 weeks after induction to allow sufficient growth for detection. Nucleocytoplasmic accumulation of β-catenin was readily apparent only in the rectum, probably reflecting the relatively high levels of NPY1R-driven Cre expression levels in this region (Fig. 8b). pMAPK hyperactivation was detected throughout the entire rectum epithelium (Fig. 8b). Co-IF imaging of β-catenin and pMAPK showed that β-catenin-accumulating cells were hyperactivated for pMAPK (Fig. 8b). Co-IF imaging with the Ki-67 marker revealed that these Wnt-hyperactivated cells were highly proliferative (Fig. 8b). No β-catenin accumulation was observed in the bladder after 8 weeks of induction (Extended Data Fig. 10b).
Finally, we incorporated the p53 conditional knockout allele into the Npy1r-eGFP-IRES-CreERT2, Apc, Kras mouse model. Npy1r-eGFP-IRES-CreERT2 was used as a homozygote to ensure sufficient Cre expression for the necessary multi-allelic recombination. Although the Npy1r-CreERT2 homozygote is effectively a knockout, Npy1r-knockout mice were healthy and fertile, as previously shown. The colon was healthy and did not present any visible defects. At 3 months following tamoxifen induction, mice presented with a high tumour load in the rectal area with frequent rectal prolapses. Tumour features were independently characterized by a certified animal pathologist as adenocarcinomas displaying an onset of invasion through the submucosa layers (Fig. 8c). Thus, NPY1R stem cells in the rectum serve as important sources of cancer formation following dysregulation of Wnt signalling in vivo. Incorporation of relevant compound conditional mutant alleles to the mouse validates the Npy1r-Cre driver as a powerful tool with which to model more advanced stages of rectal cancer originating from stem cells.