While recent decades have witnessed a revolution in therapeutic strategies yielding significant clinical responses measured in terms of tumor regression and disease-free survival, overall survival has failed to substantially improve. The recent identification of tumor cell subpopulations with the unique ability to fuel tumor growth (i.e.
CSC) may shed light on the disconnect between response rates and overall survival. That is, therapies that fail to adequately target CSC populations, which represent a minority of most epithelial tumors, will fail to eliminate those cells capable of regenerating the tumor after therapy has ceased. Furthermore, survival of these long-lived cells in the presence of toxic therapeutic agents provides an ideal selective vice for additional mutations. Familiarity with both the means of resistance to a particular chemotherapeutic agent and the phenotypic identity of those cells that harbor resistance mechanisms should help facilitate the discovery of therapies better able to clear minimum residual disease and prolong overall survival. Although the cancer stem cell paradigm explains tumor heterogeneity, provides rationale for how genetic mutations might be accumulated over long time periods and suggests resistance to chemotherapeutics/radiation may be inherent and not acquired properties of specific tumor cell subpopulations in epithelial tumors 
, a few key tenets of this theory have not been supported experimentally. Having previously identified surface markers that reliably identify colorectal cancer stem cells (CoCSC)
, we show here that colorectal tumors are enriched for CSC following chemotherapeutic regimens that halt, or at least slow, tumor growth. ALDH1 enzymatic activity, which is generally highest in CoCSC, appears to play a major role in mediating resistance to CPA, as its inhibition in vitro
, and reduced expression in vivo
, sensitizes colorectal tumor cells to the bioactive metabolite of CPA. This chemotherapeutic resistance mechanism does not appear universal; however, as altered cytotoxicity profiles were not observed with other chemotherapeutic agents (e.g.
Irinotecan) en lieu
of ALDH1 inhibition.
ESA and CD44 demarcate the subpopulation of cells with tumorigenic ability in all colorectal tumors examined to date. Overexpressed in a number of epithelial tumors and suggested to be an important prognostic marker of tumor progression 
, CD166 (i.e.
ALCAM) appears to further segregate TG from NTG cells when used in combination with ESA and CD44 
. Consistent with the hypothesis that TG cells are more resistant to chemotherapy and the association between CD166 expression and poor outcome, the tumorigenic CD166+
subset of ESA+
cells appeared more resilient to not only CPA, but also Irinotecan.
Like the normal colon crypt, which is predominantly composed of two different cell lineages (i.e.
absorptive colonocytes and goblet cells), colorectal adenocarcinomas appear to contain both immature and mature colorectal cell lineages that are somewhat unstructured in their organization. Proto-oncogenes such as c-Myb not only coordinate normal development of the distal colon, but altered expression is commonly associated with hyperproliferation of immature colorectal cells and overt cancer 
. c-Myc has recently been demonstrated to mediate nuclear β-catenin-mediated tumorigenesis in the APC-deficient mouse model of intestinal neoplasia 
. In fact, MYB
appears to be a downstream target of c-Myc. Of significant interest in studies performed here is the observation that MYC
expression is elevated in TG versus NTG cells from xenogeneic colorectal tumors, and that MYC
levels are further increased in residual cells with the CoCSC phenotype. Conversely, MYC
levels do not change, or are reduced in the NTG contingent of tumor cells, which themselves are progeny of CoCSC. These observations corroborate past studies demonstrating the association of the MYB
proto-oncogenes with cancer, but underscore the distinction between TG versus NTG cells, in that those cells most resistant to therapy (i.e.
CoCSC) also most resemble stem/progenitor cells in their phenotype and gene expression profiles.
Establishing mechanisms of resistance to chemotherapeutic drugs can be difficult, especially with heterogeneous xenogeneic tumors. However, in vitro culture conditions that facilitate colony formation with an input of tumorigenic colorectal tumor cells have been established and offer a new approach to the characterization of underlying mechanisms of drug resistance. Culture of CoCSC in these conditions is herein demonstrated to generate morphologically and histologically diverse tumors from limiting dilutions of colorectal tumor cells. That is, minimally cultured individual colonies generated in vitro are able to generate tumors in vivo that resemble the parental adenocarcinomas from which they were obtained. We further show for the first time that tumor heterogeneity can result from a single CoCSC using classical lentiviral insertion site analysis. Because in vitro colonies are highly enriched for TG ESA+CD44+CD166+ cells, the fate of these cells in defined culture conditions can now be assessed without the dangers encompassing extended in vitro culture and passaging.
Intracellular ALDH enzymes oxidize aldehydes to carboxylic acids and carry out various catabolic processes, including ethanol and amine catabolism and conversion of vitamin A to retinoic acid 
. ALDH enzyme activity can also protect cells from the cytotoxic affects of CPA, as a subset of ALDH enzyme family members can catabolize the bioactive metabolite of CPA, aldophosphamide/4-HC 
. Although ALDH1A1
, and ALDH5A1
gene products can degrade biologically active CPA metabolites 
, DEAB appears to specifically inhibit ALDH1 
. Like hematopoiesis, intestinal epithelium is resilient following damage incurred during CPA therapy 
. Hematopoietic stem cells (HSC) have high ALDH1 activity and can be isolated from bone marrow based on that unique trait 
. Similarly, we suggest that stem/progenitor cells in the base of colon crypts, like CoCSC, may have high ALDH1 activity, thus providing protection against CPA-induced cytotoxicity and tissue ablation during therapy. As in vitro
experiments suggest, inhibition of ALDH activity in vivo
sensitizes tumors to CPA therapy; however, the half-life of DEAB in vivo
is extremely short 
and such studies cannot be done. Knock down of ALDH1A1
expression using a lentiviral-based shRNA approach in UM-C6 tumors demonstrated that CoCSC can be sensitized to CPA in vivo
, as there was no noticeable enrichment of CoCSC in tumors from CPA- versus vehicle-treated mice. Furthermore, unlike for 4-HC in vitro
, DEAB was unable to alter colorectal tumor cell sensitivity to Irinotecan, suggesting that the target of this inhibitor (ALDH1 enzyme activity) plays an important role in resistance to CPA, and resistance to Irinotecan appears to involve another mechanism.
Like DEAB, retinoic acid (a vitamin A/retinaldehyde metabolite and product of ALDH activity) appears to decrease ALDH1 and ALDH3 protein levels via a feedback mechanism that can sensitize cells to CPA-induced cytotoxicity 
. Retinoic acid, in the form of ATRA, is used with great success in the clinical setting for a subset of acute promyelocytic leukemia patients who have chromosomal translocations involving the retinoic acid receptor-α gene, RARα 
, but the use of retinoids in solid tumors has not been promising to date 
. As we demonstrated here, either DEAB or ATRA alone sensitize colorectal tumor cells to 4-HC, and the combination of both appears synergistic. Because normal stem cell populations, such as HSC, neural stem cells and in all likelihood, intestinal stem cells, have high ALDH activity, its inhibition by ATRA as a pre-therapeutic regimen to CPA may also negatively impact normal stem cell populations. Nevertheless, as shown here, detailed study of rare tumor populations responsible for fueling tumor growth can provide mechanistic insights not only into tumorigenesis, but resistance mechanisms to common therapies. These inherent resistance mechanisms can include drug specific catabolic enzyme activity; such as that of ALDH1.
Here we show that xenogeneic colorectal tumors investigated to date contain a subset of TG ESA+CD44+ cells with high ALDH activity, and that this subpopulation is enriched in xenogeneic tumors from mice treated with CPA. These observations are supported by qRT-PCR using TG or NTG cells isolated by FACS, which show that ALDH1A1 is the predominant cytoplasmic ALDH enzyme in colorectal tumors and its expression is further increased in residual tumor cells following therapy with CPA; consistent with the phenotypic increase in ALDH+ cells among the CoCSC phenotype. Importantly, however, ALDH1 activity alone does not confer tumorigenicity nor demarcate TG cells. When tumorigenicity of CD44+ versus CD44− ESA+ALDH+ cells is compared, only the CD44+ subset is able to initiate actively growing tumors. Secondly, extended inhibition of ALDH1 activity with DEAB in vitro does not appear to alter cell proliferation or survival, as its presence in Irinotecan combination studies for 7 days in vitro did not differ from control. Furthermore, initiation of tumorigenesis with ALDH1A1- versus Luciferase-targeted shRNA containing cells was identical, demonstrating that ALDH1 enzymatic activity is not requisite in the absence of CPA exposure.
The advent of flow cytometry and cell sorting has revolutionized the study of developmental biology and disease, particularly in the hematopoietic system. Hematologic malignancies are among the best understood of the neoplastic diseases precisely because hematopoietic cells are easy to obtain and the in vivo and in vitro assays to determine the fate and potential of these cells have been developed. Similarly, the field of solid tumor biology has begun to enter an era where the cells responsible for fueling tumor growth can be identified, isolated, and their characteristics tested both in vivo and in vitro. Here we demonstrate for the first time that CoCSC are responsible for fueling both tumor growth and heterogeneity, and are enriched in residual tumors following chemotherapy. We also reveal that inherent resistance mechanisms differentially expressed within tumor subpopulations, such as ALDH1 enzymatic activity in CoCSC, can explain the inability of chemotherapeutic agents to improve overall survival despite tumor regression. In addition to providing evidence supported by serial transplantation studies for a previously unsupported tenet of the “cancer stem cell hypothesis”, we identify a major CSC-specific mechanism of resistance to a classical chemotherapeutic agent and establish experimental platforms both in vitro and in vivo for testing of novel agents either alone or in combination with standard of care therapies. Closer scrutiny of both normal tissue-resident stem cells and CSC will lead to a better understanding of disease mechanisms and, ultimately, better therapies.