The full-length CD44 gene comprises 20 exons, with the constant exons 1–5 and 16–20 encoding the N-terminal and C-terminal domains respectively, which are homologous domains shared by all CD44 family members [10]. The smallest and the most expressed CD44 isoform is the CD44 standard (CD44s), constructed of ten constant exons with no variant exons [11]. The other isoform, the CD44 variant (CD44v), differs from CD44s by the insertion or excision of alternatively spliced exons between the N-terminal and C-terminal domains [12]. Tolg et al. [13] confirmed that besides ten constant exons, the mouse and rat genome has at least ten variant exons, all of which can be combined alternatively into CD44 mRNA. They suggested that the variant exons be numbered by the exon code v1 to v10. Screaton et al. [14] described the structure of the human CD44 gene, reporting that it contains 19 exons crossing some 50 kilobases of DNA with ten constant exons and nine variant exons coded v2–v10 [15,16]. CD44v isoforms may contain a single variant exon such as CD44v3 or CD44v6, or a combination of variant exons such as CD44v3–v7 and CD44v8-v10. Individual cells can continually alter the splicing of CD44 pre-mRNA, resulting in many possible combinations of these variant exons, giving the potential for great diversity [12].

The CD44 protein has four primary characteristic regions: the extracellular region, the stem region (standard stem region and/or variable stem region), the transmembrane region (TM), and the short C-terminal intracellular/cytoplasmic (CP) region [17]. The extracellular part consists of seven extracellular domains (1–5, 6 and 7 of the constant exons) including N-terminal domains (ligand-binding region). The stem region (alternative splicing area) has an insertion of one or more of the variant exons between exon 5 and exon 6. The transmembrane region is encoded by a single exon (exon 8), whereas the cytoplasmic region is encoded by exon 10 or exon 9. However, exon 9 is spliced out in almost all CD44 cDNA isoforms [12,18]. Several isoforms of the human CD44 molecule are associated with tumour progression and stemness in various cancers, such as breast cancer [19], gliomas [20,21], head and neck squamous cell carcinoma [22], pancreatic cancer [23,24], prostate cancer [25] and colorectal cancer [26,27] (Figure 1 and Table 1). The complexity of the CD44 protein is further augmented by post-translational modifications including variance glycosylation with O-glycans, N-glycans and glycosaminoglycans, such as chondroitin sulphate and heparan sulphate [17]. Due to these side-chain attachments, the conserved format of CD44 (37 kDa) is enlarged to 80–100 kDa with some isoforms surpassing 200 kDa due to a high level of glycosylation [12]. An illustration of CD44 protein structure is shown in Figure 2.

CD44 is significantly expressed in lymphocytes, smooth muscle, fibroblasts and various types of epithelia and is involved in lymphocyte homing, cell adhesion and aggregation, cell migration, leukocyte activation, lymphopoiesis and myelopoiesis, angiogenesis and cytokine release [12,69]. CD44s was initially isolated from haematopoietic cells even though it is expressed in several other tissues including the liver, lung, pancreas, skin and central nervous system [12]. CD44s is expressed in adult tissues and embryo tissues from day 9.5 post coitum, whereas numerous isoforms of CD44v show a highly specialised expression pattern and are already in the egg cylinder at day 6.5 of development [70]. In contrast to CD44s, CD44v isoforms distribution is more restricted to a selected range of cells during specific stages of activation, maturation or development including macrophages, activated lymphocytes, keratinocytes and some epithelial cells such as in the stomach, bladder and uterine cervix [12] and many carcinomas. In normal tissues, CD44 isoforms play a role in hyaluronic acid (HA) metabolism regulation, whereby loss of CD44 expression disrupts HA metabolism and impairs hair regrowth, wound healing and keratinocyte proliferation [71].

Numerous studies indicated that lymphoma, breast, colon and endometrial cancer have elevated levels of CD44 mRNA [69]. Increasing evidence also suggests that CD44 is extensively overexpressed in other cancer types including gallbladder, prostate, ovarian, oral squamous cell carcinoma and gastric cancer, correlating with aggressive biological behaviour and a poor prognosis [72]. The role of CD44 in tumours is not well defined, however, elevated levels of CD44 are associated with numerous malignant tumours. The physiological functions of CD44 indicate that it is involved in the metastasis of tumours [12]. For instance, lung adenocarcinoma cells show a high expression of CD44v, which correlates with enhanced CSCs characteristics, proliferation and resistance to chemotherapeutics [73], whereas these variants, especially CD44v6, are closely related to metastasis of pancreatic carcinoma cells [69]. Many studies have investigated CD44 expression levels in several cancers in comparison to their adjacent normal tissues and explored the relationship with tumour progression and clinicopathological outcomes by mining various publicly available databases, including The Cancer Genome Atlas (TCGA), Tumour Immune Estimation Resource (TIMER) database, Oncomine database, Gene Expression Profiling Interactive Analysis (GEPIA), In silico Transcriptomics (IST) database, R2 online database, SAGE Genie tools, and Human Gene Expression Map (HGEM) (Table 2 and Figure 3).