1. Introduction
About 96% of Caucasians display the carbohydrate antigen Sd
a on the erythrocytes, secretions [
1] and a few organs [
2]. This antigen, which behaves as a dominant genetic character, was discovered independently by two groups [
3,
4]. Among the 4% Sd
a-negative individual, only a few contains “natural” anti Sd
a antibodies in serum. This is in sharp contrast with the AB0 blood group system in which all individuals form “natural” antibodies against the antigens they don’t express. A stronger version of the Sd
a antigen, known as the Cad antigen, reacts more strongly with anti-Sd
a antibodies [
5]. The structural and molecular differences between Cad and Sd
a are still not clear [
6].
Basically, the structure of the Sd
a epitope is composed by a α2,3-sialylated galactose substituted by a β1,4-linked GalNAc [
7]. However, the epitope can be found at the end of different sugar structures, including type 1 and type 2 lactosaminic chains, as well as core 1, core 2 and core 3
O-linked structures [
8] and the glycolipid sialosylparagloboside [
9] (
Figure 1).
The UDP-GalNAc β1,4-N-acetylgalactosaminyltransferase 2 (B4GALNT2), encoded by the
B4GALNT2 gene is the only enzyme able to catalyze the final step of Sd
a biosynthesis. The enzymatic activity was first detected in Guinea-pig kidney [
10] and found to require the presence of a sialic acid α2,3-linked to galactose in the acceptor. The mouse
B4galnt2 cDNA [
11] and the human
B4GALNT2 cDNA were successively cloned [
12,
13]. The human
B4GALNT2 maps on 17q21.33, it is formed by at least 12 coding exons and generates transcripts diverging in their 5’- and 3’-UTR (
Figure 2). The presence of two alternative 5’-UTR is particularly relevant. In fact, the alternative use of two distinct first exons, both provided with a translation start codon, results in two polypeptides with different amino-terminal portions [
12,
13]. The mRNA forms containing either exon 1 long (1L) or exon 1 short (1S) encode for 566 aa or 506 aa long B4GALNT2 proteins. The long form is characterized by a cytoplasmic tail of 67 aa, which is among the longest among glycosyltransferases. Both isoforms are mainly localized in the Golgi. However, the long form displays also post-Golgi vesicles and plasma membrane localization [
14]. B4GALNT2 protein contains two unconventional
N-glycosylation sites. The first is occupied by a complex type chain, which is necessary for stability, proper intracellular localization and homodimer formation [
15].
The regulation of
B4GALNT2 expression is a complex issue. The genomic sequences upstream of the transcription start site(s) display the features of a CpG island and CpG shores. Methylation of these sequences can block
B4GALNT2 expression [
16,
17]. However, promoter methylation does not appear to be an all-or-none switch. In fact, data from The Cancer Genome Atlas (TCGA) reveal that the methylation status of specific sites in the promoter and inside the gene is more predictive of the expression level than methylation as a whole [
18,
19]. TCGA data have revealed also a potentially regulatory role of miR-204-5p [
19]. Another crucial role in
B4GALNT2 regulation can be played by transcription factors. The transcription factors ETS1 and, to a lesser extent, SP1 are necessary for gene transcription, although their different expression does not appear to be responsible for B4GALNT2 modulation in pathological conditions, such as colon cancer (see below) [
20].
Despite the fact that a small but significant percentage of population is Sd
a/B4GALNT2 negative without overt phenotypic consequences, the Sd
a antigen is involved in a broad variety of physio-pathological phenomena in humans and animals. The present review aims to focus on these physio-pathological phenomena, while a comprehensive history of the Sd
a antigen discovery has been published previously [
8,
21].
4. How B4GALNT2/Sda Play a Role in Cancer
B4GALNT2/Sd
a display a strong dependence on the neoplastic transformation. The dramatic down-regulation of B4GALNT2 in colorectal cancer (CRC) [
51] is due to a marked mRNA reduction [
52]. The Sd
a antigen is also down-regulated in CRC [
53,
54]. The availability of large public databases, such as TCGA, reporting clinical and molecular data of hundreds of patients, has confirmed the marked reduction of B4GALNT2 mRNA in CRC samples. In most of patients, B4GALNT2 is completely switched off. However, those patients retaining a nearly normal expression level, display a much longer overall survival [
19,
55]. These patients display high levels of genes related to normal colon functions, such as mucus secretion, ion transport and proper glycosylation, while genes associated with tumor progression, such as
WIF1 and
TSIX were poorly expressed [
19]. In ulcerative colitis, a pre-neoplastic intestinal inflammatory condition,
B4GALNT2 is also transcriptionally modulated [
56]. In colon, the Sd
a biosynthesis is strictly connected with that of the sialyl Lewis x (sLex) [Siaα2,3Galβ1,4(Fucα1,3)GlcNAc] antigen [
57] (
Figure 3), which act as ligands for the cell adhesion molecules of the selectin family, playing a physiological role in leukocyte extravasation and a pathological role in metastasis formation [
58].
Due to steric hindrance, the biosynthesis of sLe
x and of Sd
a antigens are mutually exclusive. In fact, GalNAc (Sd
a) or Fuc (sLe
x) are added by the respective glycosyltransferases on two adjacent sugars (
Figure 3). The Sd
a epitope, but not sLe
x, is expressed by mucins from normal colon, while structures in which the β4-linked GalNAc to galactose and Fuc α3-linked to GlcNAc are in adjacent positions have never been detected [
59]. The transition from normal colon to CRC is accompanied by a switch from
O-linked chains with core 3 structures, which are often decorated by Sd
a, to those with core 2 structures, which are often terminated by sLe
x antigens [
60]. The core 3 to core 2 transition is mainly due the down-regulation of core 3 biosynthesis rather than to the up-regulation of core 2 biosynthesis [
60] (
Figure 3). In addition, the B4GALNT2 level plays a role in keeping low sLe
x biosynthesis in normal colon by competing with FUT6 for the same substrate acceptor(s) [
61]. Thus, the switch from core 3/Sd
a in normal colon from core 2/sLe
x in CRC is due to down-regulation of B3GNT6 and B4GALNT2, rather than to the up-regulation of core 2/sLe
x synthases and FUT6 in cancer (
Figure 3). Interestingly, the core-3 synthase B3GNT6 and B4GALNT2 display a grossly proportional positive relationship with B4GALNT2 in the colon cancer (COAD) TCGA cohort [
19].
In the stomach, the level of
B4GALNT2 expression level is about 50-fold lower than in colon while in the near totality of gastric cancer samples it is virtually switched off [
18]. Consistently, the Sd
a antigen which is carried in normal gastric mucosa by the glycolipid sialylparagloboside, is lost in gastric cancer [
62].
Even if the kidney is a major site of B4GALNT2/Sd
a expression, no data have been reported so far on its modulation in malignancy. TCGA reports two kidney cancer cohorts: the kidney renal clear cell carcinoma (KIRC), which accounts for 70-80% of the cases, and the kidney renal papillary carcinoma (KIRP), accounting for 10-15%. As shown in
Figure 4, the B4GALNT2 expression was markedly reduced in cancer samples of both cohorts, although at a variable degree. In KIRC patients, the B4GALNT2 relationship with survival is complex, and non-significant. In fact, in the first 1500 days, high B4GALNT2 expressers display better survival, while after 1500 days the opposite occurs. Interestingly, the putative tumor suppressor gene
HEPACAM2 is expressed only by high
B4GALNT2 expressers. Like in colon cancer (see above) TSIX was strongly down-regulated in high B4GALNT2 expressers. In KIRP, the association of high B4GALNT2 expression and longer survival is highly significant (
Figure 4).
The involvement of B4GALNT2/Sd
a in breast cancer is suggested by TCGA data, showing that in normal breast tissue
B4GALNT2 is virtually not expressed. However, in a minority of breast cancer cases it reaches a relevant level, suggesting an ectopic expression of the gene. In contrast with colon cancer, patients expressing high
B4GALNT2 display shorter overall survival and a gene expression profile closely associated with malignancy [
18]. The overexpression of the
KRT20gene, encoding cytokeratin 20, which is about 500-fold higher in high- than in low expressers is particularly relevant.
An association does not necessarily imply a causal relationship. To establish the existence of a causal nexus between B4GALNT2 and malignancy relationship, the ability of the Sd
a antigen to modulate the behavior of cancer cells was investigated by gene transfer experiments. Previous experiments in colon cancer have shown that forced
B4GALNT2 expression by gene transfer results in a strong inhibition of sLe
x antigen [
53,
64] and in a reduction of the metastatic potential of transfected cells [
64,
65]. This finding raises a key question: is this effect attributable to inhibition of sLe
x antigen or is it independent? To answer this question, the two colon cancer cell lines SW480 and SW620 (the former derived from a primary tumor and the latter from a metastasis of the same patient), originally devoid of both sLe
x and Sd
a antigens, were forced to express either Sd
a by B4GALNT2 transfection or sLe
x by transfection with the main sLe
x synthase FUT6 [
66,
67]. Consistent with previous data obtained with the LS174T cell line [
55], in SW480 and SW620 B4GALNT2 expression reduced stemness [
67]. Importantly, this model showed that the phenotypic effects were oriented toward reduced malignancy independently on sLe
x inhibition [
67]. Consistent with TCGA data, experimental studies with triple negative breast cancer cell lines have shown a positive correlation between high B4GALNT2 and malignancy [
68,
69]. In particular, it has been shown that the B4GALNT2 protein is able to interact with proteins of the major histocompatibility complex (HLA-B) [
69]. Thus, in both colon- and breast cancer, experimental data show that the level of B4GALNT2 expression strongly affects the behavior of cancer cells.
7. How B4GALNT2/Sda Affects Xenotransplantation
The availability of organs for transplantation from deceased persons represents a strongly limiting factor in modern surgery. The use of pig organs may be a possible solution but is strongly limited by the occurrence of hyperacute rejection. This reaction, takes place when the recipient’s contains preformed antibodies against surface antigens of the donor organ, and leads to complement-mediated rejection in a very short time. In pig organs, there are at least three carbohydrate antigens toward which human blood can trigger the hyperacute rejection. One of them is the so called “Galili antigen”, consisting in a terminal α1,3-linked galactose residue whose addition is mediated by the α-galactosyltransferase encoded by the
GGTA1 gene. During evolution,
GGTA1 has been inactivated in a precursor of humans and Old World primates (gorillas, bonobos, chimpanzees). Consequently, these species do not express the Galili antigen on their cells, while a significant percentage of their circulating antibodies is able to react with it. This is probably due to cross-reactivity with microorganism antigens and lack of tolerance by the human immune system against this antigen. A second type of non-human carbohydrate is represented by glycoconjugates terminating with N-glycolylneuraminic acid. This type of sialic acid is present in animals but not in human glycoconjugates because of the inactivation of the
CMAH gene, which encodes CMP-N-acetylneuraminic acid hydroxylase. This event occurred after divergence of humans from the other Old World apes. Surprisingly, the third is the Sd
a antigen encoded by porcine
B4GALNT2 [
80,
81]. In fact, human cells forced to express porcine
B4GALNT2 display increased complement-mediated lysis with serum from primates pre-immunized with pig organs [
82]. Blood cells from
CMAH/GGTA1/B4GALNT2 triple KO pigs display reduced reactivity with human plasma than cells from
CMAH/GGTA1 double KO pigs [
83,
84,
85,
86]. It is not clear why the human immune system is not tolerant toward a self antigen when it is synthesized by pig B4GALNT2. Whatever the reason, the presence of the Sd
a antigens on pig cells is an obstacle that must be removed before a pig-to-human transplantation is considered.