Main

CD44, a class I transmembrane glycoprotein that binds to hyaluronan, has attracted much interest due to its role in cancer and stem cells. Initial studies have suggested that CD44 promotes tumor metastasis.1 This function is supported by further studies demonstrating the association of CD44 expression with poor patient survival in multiple cancer types, including colon cancers.2 More recently, CD44 emerged as a useful marker in certain solid tumors for enriching cancer stem cells (CSCs), the highly tumorigenic cancer cell population within a tumor mass. CSCs in both breast and colon cancers were indeed shown to express CD44.3, 4 It is worth noting that CD44 expression is not limited to CSCs, as many non-CSCs can also express CD44.5 Thus, specific identification of CSCs requires combination of CD44 and other markers. However, CD44 is functionally involved in the maintenance and homing of CSCs as studies in breast and prostate cancer cell lines have shown that CD44 is required for tumorigenicity, and its expression is repressed by tumor suppressors p53 and microRNA-34a.6, 7 In addition, CD44 participates in leukemia stem cell homing to their niches.8, 9 This effect may be mediated in part by interactions with hyaluronan and also through specific fucosylation of CD44 standard isoform (CD44s) that confers E-selectin ligand binding activity and ability to interact more closely with bone marrow endothelium.10

In addition to its multifunctional cell adhesion properties, CD44 signaling also connects extracellular matrixes with intracellular signaling networks.2, 5 The diverse functions of CD44 are enabled, in part, by the expression of multiple isoforms through alternative mRNA splicing. CD44s, the shortest CD44 molecule, is encoded by 10 exons. Between the standard exons 5 and 6, up to 10 variant exons can be inserted in various combinations to form CD44 variants (CD44v).2, 5 These extra exons allow CD44 to interact with multiple signaling molecules in addition to its main ligand hyaluronan. For example, CD44v3 can associate with heparin-binding proteins such as fibroblast growth factor 2, whereas CD44v6 binds to hepatocytes growth factor and vascular endothelial growth factor.2 These distinct properties of CD44 isoforms suggest they may have specific physiological functions in vivo. However, the exact function of CD44 isoforms remains largely undefined.

The report by Zeilstra et al.11 sheds light on distinct roles of CD44s and CD44v isoforms in intestinal stem cells and in intestinal tumor formation. As a target of Wnt signaling pathway, CD44 is expressed in intestinal crypts and overexpressed in colon cancers, which often have hyperactivation of the Wnt signaling pathway.11 Using a well-defined intestinal stem cell marker Lgr5,12 Zeilstra et al.11 separated mouse intestinal stem cells and transit-amplifying cells and performed isoform-specific reverse transcription–PCR. This analysis revealed that intestinal stem cells do not express CD44s, but rather CD44v isoforms, including a long isoform that contains variant exons 4–10 (CD44v4–10) that is only expressed in the stem cell compartment. In contrast, transit-amplifying cells express CD44s and other shorter CD44v isoforms, but not CD44v4-10. Human cryptic foci, which are microadenomas retaining a stem cell program, similarly express CD44v4-10 but not CD44s.

To dissect the function of CD44v4–10 and CD44s in vivo, Zeilstra et al.11 generated mice that solely express either CD44v4–10 or CD44s by knocking in specific cDNAs into the Cd44 locus. Interestingly, the animals that are deficient in either CD44s or CD44v developed a normal intestinal stem cell compartment, which is consistent with the earlier finding that pan-CD44 knockout mice develop intestine crypts normally.11 However, when bred into the Apcmin/+ background, CD44s-expressing mice have significantly reduced tumor burden compared with wild-type mice. Similar defects were previously observed in pan-CD44 knockout mice.11 In contrast, mice expressing CD44v4–10 have a tumor burden similar to wild-type mice. This shows that CD44v4–10, but not CD44s, is the functional CD44 isoform that contributes to intestinal tumor formation.

These studies provide convincing genetic evidence of differential functions of CD44 isoforms in intestinal tumorigenesis. However, the precise mechanism by which CD44v promotes intestinal tumorigenesis remains to be elucidated. Deficiency of CD44v does not affect cell proliferation or survival in the intestinal crypt of Apcmin/+ mice, suggesting that other mechanism(s) are involved. Given the important role of intestinal stem cells in adenomatous polyposis coli mutation-induced tumor formation13 and the specific expression of CD44v4–10 in intestinal stem cells, it would be interesting to investigate whether CD44 regulates the self-renewal and differentiation of stem cells. Although the intestinal stem cell compartment appears phenotypically normal in CD44s knock-in mice, it would be worthwhile to determine whether there is any defect in stem cell activity, especially under stress or wounding conditions or upon oncogenic insult.

Intestinal stem cells express a few other shorter CD44 variants in addition to v4–10. Can any of these shorter isoforms replace CD44v4–10 in intestinal tumor initiation? Identifying the exact exon(s) that are required for tumor formation may help pinpoint the specific CD44-mediated signaling pathway(s) that are involved in tumorigenesis. Another intriguing question for future investigation is whether CD44s plays any role in intestinal tumor progression at all. Interestingly, CD44s was shown to be required for epithelial–mesenchymal transition.14 Given the contribution of epithelial–mesenchymal transition in tumor progression and metastasis, the function of CD44s in late-stage tumor progression needs to be further investigated. Because the Apcmin/+ model forms mostly adenomas, other tumor models recapitulating invasive and metastatic intestinal cancers would be required to address this question.

Overall, the study by Zeilstra et al.11 highlights interesting aspects of isoform-specific functions of CD44 (Figure 1). The CD44 isoform-specific knock-in models will likely become a valuable genetic tool for dissecting the functions of CD44s and CD44v in various normal physiological and pathological settings. Such studies have been performed previously by interfering with CD44 function with variant-specific blocking antibodies. Although this approach is somewhat useful in the hematopoietic system, it is not as effective in solid tissues where it most likely does not completely block CD44 function.2 The present study also raises an intriguing possibility that normal and CSCs of other tissues may express specific CD44v isoforms rather than CD44s. Thus far, most of CSC studies that defined CD44 as a CSC marker have used pan-CD44 antibodies. Because CD44s is widely expressed by various cell types, pan-CD44 antibodies do not faithfully distinguish CSCs and non-CSCs. Thus, it is worth to explore if CD44 variant-specific antibody can detect normal and CSCs more specifically. Finally, it is now recognized that CD44 is part of a larger alternative splicing program regulated by ESRP1 and ESRP2.15 Thus, it is conceivable that other genes may also be alternatively spliced between intestinal stem cells and non-stem cells. Identifying these genes by global profiling approaches, such as RNA-seq, may uncover novel stem cell markers and regulators.

Figure 1
figure 1

Alternative CD44 splicing in intestine. Intestinal stem cells and transit-amplifying cells express distinct CD44 isoforms. The stem cell isoform CD44v4–10, but not CD44s, contributes to the tumor formation.

Full size image