Cancer is a multifaceted disease characterized by uncontrolled proliferation of aberrant cells that can develop in diverse organs and tissues and then invade other parts of the body (Chambers et al., 2002; Hanahan and Weinberg, 2011). Despite advances in cancer research over the last few decades, efficient treatment for several cancers in many cases is still lacking. One of the reasons for this is the multifactorial nature of the disease, determined by genetic, epigenetic, and transcriptional deregulation combined with environmental insults (Hanahan and Weinberg, 2011; Almendro et al., 2013). This heterogeneity not only exists among different cancer types but also within the same cancer type. As a result, current approaches are evolving towards precision medicine, aimed at finding novel markers for patient stratification. To date, the treatments approved for anticancer therapy are based either on relatively non-specific chemotherapy drugs or on the selective targeting of biological processes such as chromatin remodelling, DNA repair, proteasomal degradation, immune surveillance or aberrant kinases involved in signal transduction (Zwick et al., 2002; Lapenna and Giordano, 2009). However, a large fraction of such treatments suffers from drug resistance and tumor relapse, thus motivating the identification of new molecular drug targets. In this review, we focus on the prospect of novel anticancer treatments that impact on intra-Golgi traffic and, in particular, on the cisternal maturation machinery (CMM) that controls intra-Golgi transport, and hence glycosylation.

Glycosylation is extensively altered during oncogenic transformation (Hakomori, 2002; Stanley, 2011; Pinho and Reis, 2015; Varki, 2017). Glycans are abundant polymers constituting the most abundant post-translational modifications of proteins and lipids, impacting a variety of biological functions (Varki and Gagneux, 2017). As compared to other processes, the role of glycosylation has been overlooked for several years in cancer biology and therapy. Analysis of glycans compared to other biopolymers is inherently complex due to their branched nature; moreover, this problem is compounded by the template-independent nature of their biosynthesis (Narimatsu et al., 2019; Schjoldager et al., 2020). As a result, glycan research often faces difficulties that can impede the development of projects by groups lacking full technical and scientific expertise in this area. Further, the contribution of aberrant glycans to tumorigenesis has not been clearly defined. Altered glycans have been used as tumor markers for several decades, e.g., carcinoembryonic antigen (CEA) 125, CEA19-9 (sLeA), and CEA72-4 (sTn) which are tumor-associated circulating serum O-glycans (Pinho and Reis, 2015). In most cases, it is unclear whether these aberrant glycans have a direct role in tumorigenesis, or they only act as intermediate players in the disease pathways initiated by other well-defined oncogenes. However, some studies have demonstrated how aberrant glycans can be “active players” in promoting oncogenesis (Nguyen et al., 2017; Rizzo et al., 2021). Truncated O-glycans frequently found in cancers have been observed to promote tumorigenesis by inducing EGFR and ErbB2 receptor activation (Freitas et al., 2019) and activation of MMP14 that increases the degradation of extracellular matrix (ECM) and initiates tumor growth in mice liver (Nguyen et al., 2017). Along the same line, glycan changes occurring in the later stages of cancer that were considered as “passive players” are emerging as determinants of drug resistance. For instance, increase in surface sialylation of N-glycans linked to plasma membrane (PM) receptors were shown to induce resistance to doxorubicin by acting as a barrier to chemotherapeutic agents (Ma et al., 2013; Ji et al., 2017). These glycan aberrations participate in tumorigenesis, and it is, therefore, conceivable that they may result from potential oncogenes acting upstream to the observed malignant phenotypes. Currently, there is no clear experimental evidence for glycosylation-based oncogenic mechanisms. As a basis for identifying errors in glycan biosynthesis and understanding how they occur, it is necessary to comprehend the underlying mechanisms of glycosylation.

Cargoes are synthesized in the endoplasmic reticulum (ER) and then transported to the Golgi apparatus. Except for the cargoes that undergo N-glycosylation which experience an initial step of glycan addition before leaving the ER, most of the glycosylation pathways occur in the Golgi apparatus, (Varki and Gagneux, 2017). The Golgi is endowed with approximately 200 resident glycosyltransferases (glycoenzymes) (Varki, 2017), which catalyse the serial addition of sugars to the cargo in the lumen of the Golgi apparatus (Stanley, 2011). These enzymes are arranged in orderly succession along the Golgi cisternae, mirroring the order in which they act in elongating the growing glycan chains. Moreover, each Golgi cisterna is equipped to carry out a complex set of glycosylation reactions and thus to act as a specialized reaction chamber. Enzymes are enriched in 1–2 cisternae, and it is clear that some enzymes have a sharper localization than others (Tie et al., 2018; Pothukuchi et al., 2021). Within each cisterna, subsets of glycoenzymes are organized to catalyse serial reactions of the same glycosylation pathway, generating a functional processing module (Narimatsu et al., 2019) (Figure 1). The processing occurs during transport through Golgi cisterna, where cargo encounters glycoenzymes often organized in metabolic modules, and single monosaccharide units are sequentially added, which defines the cargo-specific sugar configuration or glycoform (Schjoldager et al., 2020) (Figure 1). Specific glycoenzymes compete for the same substrates, which may contribute to the formation of different glycosylation pathways and of glycoforms (variants of glycosylation chains). Glycosylated cargoes then exit from the trans-Golgi where they are sorted to the appropriate cellular destination. Hundreds of different glycoforms have been found linked to cargoes (Moremen and Haltiwanger, 2019). The glycoform repertoire of a given cell type is not a random collection of all possible synthesizable configurations, instead, they fall into a discrete distribution of variants, suggesting that the glycosylation process is highly regulated. Several factors have been proposed to regulate the overall glycan synthesis process, and the interested reader can refer to Pothukuchi et al. (2019). As the transcript level of different glycoenzymes does not quantitatively correlate with the glycans synthesized by a cell (Moremen and Haltiwanger, 2019), the intra-Golgi localization, and the control of enzyme levels by their degradation rate in lysosome or plasma membrane, have emerged as a new theme for regulating glycan synthesis (Pothukuchi et al., 2019). Both the levels and positioning of the enzymes is controlled by membrane transport mechanisms operating in the Golgi. Since the glycoenzymes are “single pass” membrane proteins with a short cytosolic tail and the catalytic domain-oriented towards the Golgi lumen, their levels and intra-Golgi positioning is indirectly controlled by proteins peripherally associated with the cytosolic side of the Golgi. Such peripherally associated Golgi proteins control the dynamics of glycoenzymes and constitute the components of the CMM (Tu et al., 2008; Rizzo et al., 2021). The process of cisternal maturation is broadly accepted as the predominant mechanism of intra-Golgi trafficking and is supported by solid experimental evidence in various cell types including mammalian and yeast cells (Bonfanti et al., 1998; Bonifacino and Glick, 2004; Rizzo et al., 2013). Transport by cisternal maturation represents the fundamental framework within which the glycosylation of most proteins and lipids are regulated at the Golgi apparatus.

The Golgi complex is composed of flattened, stacked membranous cisterna with cis- to trans-polarity (Varki, 2017) (Figure 2). Each Golgi cisterna is defined by its Golgi residents. The cisternal maturation mechanism proposes that new cisternae continuously form on the cis- face of the Golgi (by the homotypic fusion of vesicles coming from the ER) and disassemble at the trans-face, generating a flow of membranes from the cis- to the trans-Golgi up to the PM. Cargo proteins remain in the progressing cisterna until they leave the Golgi in post-Golgi carriers for their final destinations. At the same time, the cisternae change their “identity” (see below) from cis to medial to trans. Otherwise, glycoenzymes also move anterogradely along with the progressing cisterna for a given distance, but they do not leave the Golgi. When they reach the (cis, medial or trans) where they must reside, they are retained there by a mechanism based on active recycling via sorting in retrograde transporting vesicles. Thus, instead of moving forward like the cargoes, glycoenzymes are transported in vesicles to acceptor proximal compartments, to balance the anterograde flow of progression. Once incorporated into the acceptor cisterna, retainer adaptors such as GRASP55 prevent a further round of glycoenzyme recycling. The retaining action is required for dynamic compartmentalization of specific glycoenzymes within the Golgi compartment (Pothukuchi et al., 2021). The balance between anterograde movement of the cisterna and the retrograde recycling, along with the retention activity of retainers help to determine the localization of the enzymes in the stack (see below). The CMM, besides determining the positions of glycoenzymes in the Golgi apparatus also contributes substantially to controlling the level of glycoenzymes. At steady-state, a fraction of glycoenzymes constantly escapes the recycling mechanisms and is transported for lysosomal degradation, or to the PM where they are degraded by ectodomain shedding (Sun et al., 2021). This fraction is replenished by new synthesis, thus ensuring constant levels of each enzyme. However, if the recycling mechanisms of the glycoenzymes lose efficiency or slow down, the fraction of enzymes exported to the lysosomes or PM and degraded increases, thus decreasing the overall enzyme level in Golgi. Conversely, if the recycling mechanism is potentiated, the fraction of enzymes degraded over time decreases or is completely inhibited. As a consequence, enzyme levels increase, even massively, altering the molecular and functional profile of the glycans (Figure 3) (Rizzo et al., 2021). Thus, the intra-Golgi transport mediated by cisternal maturation controls both the levels and position of glycosylation enzymes in the Golgi.

The retrograde recycling of the enzymes required for cisternal maturation is mediated mainly by the Coat protein I (COPI) complex. Indeed, mutation of a COPI subunit can impair the maturation of cisterna as observed in yeast cells (Matsuura-Tokita et al., 2006). Recent studies have described multiple COPI dependent and independent pathways operating at the Golgi (see below) to promote enzyme recycling (Tu et al., 2008; Liu et al., 2018; Witkos et al., 2019; Zhao et al., 2019; Casler et al., 2021; Rizzo et al., 2021; Sardana et al., 2021), leading to a recent proposal that cisternal maturation is driven not by a single pathway but a mosaic of several likely independent recycling pathways (Rizzo et al., 2021).

Initial models of cisternal maturation had hypothesized a COPI dependent recycling of the enzymes within the Golgi likely mediated by their direct interaction with the COPI complex (Glick et al., 1997). Indeed, a differential affinity of enzymes to the COPI coat was hypothesized to regulate the differential localization of enzymes along the cisterna. Further direct binding of at least some enzymes to the COPI coat has been described (Liu et al., 2018). Later studies in yeast proposed a variation to this general theme by identifying for the first time an adaptor, Vps74, that links the enzymes to the COPI coat. Vps74 binds to a specific motif present in the cytosolic tails of glycosylation enzymes and simultaneously binds to the COPI coat through its N-terminus (Tu et al., 2012). Thus, by bridging the coat with the client glycoenzymes, this recycling adaptor allows the recruitment of a specific subset of glycosylation enzymes into the COPI recycling pathway. Golgi Phosphoprotein 3 (GOLPH3), the human Vps74 homolog, was shown to act on a group of glycosylation enzymes among which there is a subset of enzymes which catalyse serial reactions responsible for the synthesis of specific glycosphingolipids (GSLs) and which represent an enzymatic module (Rizzo et al., 2021). The evidence supports a model by which the localization of GOLPH3 clients in the trans-Golgi cisterna depends on the localization of GOLPH3 itself in the same cisterna. GOLPH3 localization is, in turn, determined by the presence in the trans-cisterna of the phosphatidylinositol 4-phosphate [PI(4)P] (Wood et al., 2009; Rahajeng et al., 2019). PI(4)P is a trans-Golgi identity marker produced by the PI4KIIIβ kinase, by which GOLPH3 is recruited to the membrane. Thus, at each transport round, GOLPH3 is recruited to the trans-Golgi by PI(4)P (Dippold et al., 2009). Once recruited, GOLPH3 incorporates its client enzymes into COPI-dependent retrograde vesicles for transport to the proximal cisterna (presumably the maturing medial-elements) (Eckert et al., 2014). In this cisterna, PI(4)P is degraded, causing GOLPH3 to detach and release the GOLPH3 clients in the (medial) cisternal membrane. At the same time the trans-cisterna disassembles and the medial-cisterna completes maturation into a trans-element, where PI(4)P is produced again, and GOLPH3 is recruited to start a new transport round. In this scheme, glycoenzyme localization is driven by the production of identity markers [e.g., PI(4)P] on maturing cisterna, which is itself driven by the Rab conversion process (see below); and the GOLPH3 clients cycle between trans- and medial-cisterna, with a prevalent steady-state localization in the trans-Golgi because of the underlying synthesis and degradation cycle of PI(4)P in the trans- and medial-cisterna, respectively. Similar mechanisms likely function at other levels of the Golgi complex, and other components may be involved in the recruitment of recycling adaptors or directly COPI, but this remains to be tested experimentally. Indeed, yeast Erd1 was recently found to act as a transmembrane adaptor molecule to links client enzymes to COPI coat surprising by binding to the adaptor Vps74 (Sardana et al., 2021). Thus, the COPI-dependent recycling is a mosaic of at least three different pathways (one adaptor independent and two adaptor dependent pathways). There are likely to be other COPI-dependent pathways as for instance GORAB and Scyl1 were found to scaffold COPI coat on trans-Golgi to promote recycling of ST6GAL1 enzyme (Witkos et al., 2019). Whether they also act as a recycling adaptor is still to be understood. Of note, while we classify them as three different COPI pathways—it includes the possibilities of three distinct COPI vesicles, three distinct pathways of recruiting Golgi residents into the same COPI vesicle and all the variations in between. In addition to biochemical evidence, morphological evidence provided by Cryo-electron microscopy studies show how COPI-coated vesicles changed their luminal density, membrane thickness and size during progression through the Golgi stack, in a manner that reflects corresponding changes in the morphology of the donor cisterna (Bykov et al., 2017), supporting the idea of the existence of distinct intra-Golgi COPI vesicles serving particular transport routes.

In addition to COPI-dependent pathways, yeast studies have also shown the existence of COPI-independent recycling pathways especially operating at the trans-Golgi. For instance, a COPI-independent pathway mediated by Dop1 for recycling enzymes from trans-Golgi network (TGN) to Golgi was identified in yeast (Zhao et al., 2019). Furthermore, recycling of classes of enzymes in the late Golgi was shown to be mediated by the adaptor protein complex-1 (AP-1) clathrin adaptor along with Ent5, a clathrin adaptor. The AP-1/Ent5- mediates two sequential intra-Golgi recycling pathways that define the polarized distribution within Golgi of two classes of Golgi residents (Casler et al., 2021). Thus, there are at least three different COPI-independent recycling pathways shown to operate in the Golgi. Of note, multi-transmembrane proteins TM9SF2 and LAPTM4A were shown to be necessary for the Golgi localization of a subset of glycosylation enzymes (Yamaji et al., 2019). Whether these proteins act in a COPI-dependent or independent manner is not clear, even though TM9SF2 is equipped with a motif for interacting with COPI (Woo et al., 2015; Yamaji et al., 2019). In general, it is becoming apparent that there is still a lot of intra-Golgi recycling pathways to discover.

While these multiple recycling pathways at the Golgi still act within the cisternal maturation framework of intra-Golgi transport, they change it in a subtle but important manner- the recycling of glycosylation enzymes is not mediated by single homogenous house-keeping pathway, but rather by a sophisticated mosaic of pathways that act in a coordinated manner and can likely be subjected to independent regulatory controls to precisely build the glycan profile of the cell.

The mammalian Golgi stack is thought to consist of three distinct compartments - cis-, medial- and trans-Golgi and so the transport of cargoes from cis to trans-Golgi will require at least three distinct maturation steps, each with its set of above-mentioned molecular components (Figures 1, 2). Of note in yeast recent studies propose there could just be three steps in this maturation scheme that are required to transport cargoes from cis to trans-Golgi (Day et al., 2013). But when we look at the number of molecular components available in the Golgi, it does not necessarily reflect the three-step pathway. For instance, while one could expect the requirement of three distinct Rab proteins corresponding to the three distinct maturation steps, we find close to two dozen Rab proteins localized to the Golgi apparatus (Barr, 2009; Thomas and Fromme, 2020), which is an order of magnitude more than the minimum required. Similarly, there are more than expected number of Golgin tethers. Biochemical evidence shows how cis-localised GM130 tethers vesicles originating from ER, while TMF, Golgin-84 and GMAP210 mediate different intra-Golgi transport routes. GMAP210 and golgin-84 tether preferentially a population of presumably COPI vesicles originating from medial-Golgi which could dock with the early Golgi compartments, while TMF tethers late derived COPI vesicles, which dock with the medial-Golgi compartment. A similar situation is present with other tethers in other districts in the Golgi (Wong and Munro, 2014). It is to be noted this excess number of Rabs and tethers is still valid if we remove the isoforms of the same gene from consideration. Similarly, even the COPI coat subunits have distinct isoforms and that are distributed non-uniformly over the Golgi stack (Moelleken et al., 2007). Whether this corresponds to distinct specificities for the Golgi residents incorporated into the vesicles is yet to be resolved. While at first sight it might seem confounding, we propose that this goes in line with the proposal that the cisternal maturation is a coordinated event of multiple maturation pathways each characterized by a distinct Rab, glycoenzyme adaptors and possibly tethering molecules. The multiple pathways may operate completely distinctly or converge into the same COPI vesicle. In either case, the existence of this mosaic of maturation pathways suggests a potential for specific regulation of a subset of glycosylation enzymes.

This hypothesis has an important implication in oncology, since it changes the perspective of cisternal maturation from house-keeping pathway that can barely be altered without dire consequences to the cell to a complex mosaic of pathways each when altered affecting only a specific set of enzymes but not the cisternal maturation and cargo transport as such. This can potentially lead to specific alterations in glycoenzyme levels and their intra-Golgi positioning with a subsequent change in the glycan profile of the cell. We discuss the potential of the CMM components to contribute to oncogenesis from this mosaic maturation point of view.

Several studies have underlined how alterations of glycoenzyme trafficking affect their position inside the Golgi cisterna, affecting the sequence of glycosylation reactions and the final cargo-specific glycan outcome (Chia et al., 2019; Rizzo et al., 2021). In this scenario, we have recently focused on identifying plausible oncogenes operating within the mechanism of glycoenzyme trafficking. We have particularly focused on the Golgi protein GOLPH3, a well-known oncogene (Scott et al., 2009). When amplified in certain tumors, GOLPH3 exerts its oncogenic activity by influencing glycan biosynthesis (see below). GOLPH3 is thus the first demonstrated glyco-oncogene. Similarly, several CMM genes that could support and drive tumorigenesis through altered glycan synthesis are over-expressed due to gene amplification in several solid tumors (Figure 5). Thus, the alteration of expression in CMM genes could support oncogenesis by altering the dynamics and levels of relevant glycoenzymes. These findings imply that previously unknown oncogenes may be hidden in the CMM or glycoenzyme genes. These genes have yet to be discovered or simply overlooked, despite being part of the glycan synthesis machinery, even though they are not glycoenzymes. Indeed, current therapies aim at cancer glycans exposed on cancer cells, rather than targeting potential oncogenes involved in their underlying synthetic machinery. To date, only a limited number of glycoenzymes inhibitors are currently being evaluated as anticancer drugs in clinical trials, such as 2-Fluorofucose (2FF), which is a fucosylation inhibitor used alone or in combination with pembrolizumab (anti-PD1) to treat specific carcinomas (NCT02952989). Our goal is to develop a framework for the analysis of the oncogenic potential of CMM genes and reveal novel Golgi-related oncogenes. Several preclinical studies highlight the active role of these genes in tumorigenesis through the regulation of trafficking and other cellular functions. Below, we discuss the oncogenic potential of different categories of CMM genes, found frequently amplified in cancers. We do not attempt to exhaustively survey the potential glyco-oncogenes. Rather we wish to discuss the possible oncogenic mechanism of action, due to intra-Golgi glycoenzyme recycling and synthesis of pro-tumorigenic glycoforms, of a limited number of prototypes belonging to the main classes of Golgi machinery proteins.

Two main types of adapters have been described so far, the recyclers and the retainers. In the following we will define their mechanism of action based on the knowledge acquired about the best characterized of these types of proteins.

In this review, we have discussed how components of CMM that are frequently amplified in cancers (Figure 5) might represent oncogenes that act by deregulating glycan synthesis. The intra-Golgi glycoenzyme recycling mediated by CMM (e.g., Arf1, PI4KIIIβ, PITPNC1, GOLPH3, and Rab1) might affect the position and levels of glycoenzyme, as shown for some of them, thereby altering glycosylation flux and impacting the overall glycan outcome. Aberrant glycans exposed on cancer cells, in turn, exert their tumorigenic functions by different mechanisms [reviewed elsewhere (Mereiter et al., 2019). GOLPH3 and Arf1 have been previously shown to contribute to tumorigenesis through altered glycosylation suggesting a plausible oncogenic mechanism shared by other trafficking components when amplified in cancer (Chia et al., 2021; Rizzo et al., 2021).

Currently, there are a few inhibitors directed against the glycan synthetic machinery, but none are utilised as anticancer drugs. Few glycoenzyme inhibitors are currently being evaluated as drugs in clinical trials, such as fluorinated sugars analogs (van Wijk et al., 2015). The compound T3Inh-1 was shown to specifically block GALNT3 to reduce breast cancer cell growth in vitro (Song and Linstedt, 2017), and Swainsonine, a glycoenzyme specific inhibitor, showed anticancer effects on patients, with low toxicity (Galustian et al., 1994; Goss et al., 1994). The GSL inhibitor Miglustat is regularly used in clinics for the treatment of congenital disorders of glycosylation (CDG) and lysosomal storage diseases (Patterson et al., 2007). There are no trials in which Miglustat is used for anticancer treatments despite the established role of GSL in tumor progression. Given the scarcity of approved inhibitors of the underlying synthetic machinery, current anticancer interventions generally target directly the aberrant glycoforms exposed by cancerous cells, as is the case of the approved drug Dinutuximab which targets the GD2 ganglioside expressed on neuroblastoma cells, or for specific CAR-T recognizing an aberrant O-glycoform on MUC1 which is still in clinical trials (Yu et al., 2010; Posey Jr et al., 2016). Aberrant glycosylation plays an important role in Trastuzumab resistance as observed in ErbB2-positive unresected advanced gastric cancer patients. ST6Gal1-dependent sialylation within ErbB2 N-glycan site was shown to block the mAb-binding ectodomain, thereby antagonizing the pro-apoptotic effects of Trastuzumab (Duarte et al., 2021). A better understanding of cancer heterogeneity based on the oncogenic role of CMM or glycoenzymes, will improve our comprehension of the role of glycans in cancer biology. We expect that a detailed understanding of the mechanism(s) by which the CMM influences the synthesis of onco-glycans will reveal 1) novel markers to aid in the clinical stratification of patients, 2) new molecular targets enabling the precise manipulation of the synthetic glycan machinery, and 3) new drugs that will act synergistically to current treatment regime to develop personalized targeted anticancer therapies.

Treatment of certain cancers might benefit from glycoenzyme specific inhibitors when an individual glycoenzyme is aberrantly expressed. Instead, alterations of components of the CMM impact entire enzymatic modules, each consisting of different types of enzymes. This condition might be not easily modified by inhibitors targeting single glycoenzymes, as multiple co-occurring onco-glycoforms might be simultaneously active. Indeed, altered level of individual glycoenzyme may play a modest role in influencing the final glycosylation flux. Whereas, altered levels of CMM components are expected to induce an extensive remodelling of glycosylation reactions, as the CMM modulates simultaneously the dynamics of several sequentially acting enzymes which constitute a processing module. In this scenario, gross variations in an individual enzyme level are required to alter the flux through its subset of reaction network to affect the final glycosylation outcome. Conversely, a small fluctuation in the level and position of each glycoenzyme belonging to the same processing module, as observed upon CMM gene amplification, can lead to a significant effect on the metabolic flux of glycosylation, when the entire glycoenzymatic module is enhanced. Mechanistically, CMM components are limiting factors wherein copy number alterations of any one component might simultaneously enhance the membrane bound fraction of others, thus enhancing the entire COPI-dependent recycling process (we expect this to be true also in the case of amplification of components of COPI-independent recycling machinery). Interestingly, the specific intra-Golgi recycling machinery for all Golgi glycoenzymes are not known. For instance, MGAT5 is a putative glyco-oncogene whose overexpression enhances the synthesis of β1-6 branched N-glycans linked to adhesion molecules and RTKs, thus promoting tumor growth (Granovsky et al., 2000; Lau et al., 2007). Intriguingly, the amplification of CMM genes responsible for MGAT5 dynamics might increase the levels of this enzyme, phenocopying the effect on glycosylation observed upon MGAT5 overexpression, and promoting cancer. Alternative regulatory mechanisms based on loss of function mutations within the CMM might affect intra-Golgi glycoenzyme dynamics. Loss of function mutations within the recycling machinery of EXT1 and EXT2, known to act as tumor suppressors regulating proteoglycan synthesis, might lead to enzyme mislocalization outside Golgi, thus promoting sarcomas (McCormick et al., 2000). Similarly, the loss of function mutations or genetic deletions of retainers, such as GRASP55 might lead to excessive intra-Golgi recycling of specific glycoenzymes, thereby affecting glycosylation, as already shown for a few glycoenzymes belonging to GSL metabolism (Pothukuchi et al., 2021).

Further efforts are needed to elucidate the oncogenic potential of amplified CMM components and discover novel mechanisms impacting onco-glycan synthesis. These components, which were overlooked for years, might represent key molecules for understanding cancer heterogeneity and, by linking glycomics and cancer, might aid in the patient stratification and establishing personalized treatments.