Differentiate and switch, a tale of two heads of a lipid

Abstract

Neuronal differentiation is an intricate process involving many factors and programs. One notable “curiosity” has been the observation that upon neuronal differentiation, stem cells switch the expression of their surface glycosphingolipids (GSLs) by substituting one class (the globo‐series) of GSLs by another (the ganglio‐series). Russo and colleagues show that there is an intricate dance between these two lipid series such that the globo products suppress neuronal differentiation via the master regulator AUTS2, which in turn suppresses the formation of the ganglio‐series. These findings open the door for further mechanistic studies on the roles of various GSLs in neuronal differentiation.

The study of sphingolipids and the brain has been intertwined for more than a century and a half! In one of the earliest studies of the chemistry of the brain, Thudicum, the prototypical physician scientist, published his “Treatise on the Chemical Constitution of the Brain” in 1884 (Thudicum, 1984) in which he introduced the world to the sphingolipids: lipidic materials that remained insoluble after hydrolysis of the tissue. He dubbed them sphingolipids based on the Greek Sphinx because of their enigmatic nature and function. In the succeeding decades, notable scientists have defined the structures of various sphingolipids, their metabolism, and many important functions.

Sphingolipids are structurally based on a ceramide backbone (composed of a sphingoid base and an N‐linked fatty acid) to which various headgroups are attached (Fig 1). The GSLs constitute a diversified class of sphingolipids in which the headgroup attached to the ceramide includes one or more glycose unit. GSLs are mostly based on lactosylceramide, modification of which generates distinct groups or “series” of GSLs, most notable of which are the ganglio‐series (gangliosides) in which the GSLs contain at least one sialic acid residue; and the globo‐series (globosides), based on Gal‐α1‐4‐LactosylCer. Two key enzymes regulate the formation of the founding lipids in each series: GM3S and Gb3S, respectively.

Figure 1. Interplay of glycosphingolipid metabolism, expression, and signaling in neuronal differentiation.

The scheme shows the production of globo‐series GSLs (globosides) as the “default” preferred products of metabolism of lactosylceramides in stem like undifferentiated cells. This results in inhibition of Auts2, suppression of generation of ganglio‐series GSLs (gangliosides), and suppression of differentiation. Loss of stem cell transcription factors and/or induction of differentiation results suppression of globoside synthesis, activation of Auts2, generation of gangliosides, and induction of differentiation.

It is now recognized that sphingolipids are critical structural and functional molecules. The “simple” backbone sphingolipids and their phosphorylated derivatives are engaged in a multitude of signal transduction pathways (Hannun & Obeid, 2018). GSLs, which are located predominantly on the outer leaflet of the plasma membrane, function in cell differentiation, cell adhesion, cell–cell interaction, and as surface receptors, immune modulators, and residents of specialized membrane sub domains (rafts and others) (Hakomori, 2008). In the brain, GSLs and especially galactosylceramide and its sulfated form, sulfatide, also play essential roles as components of the myelin sheath. Deficiency in the first enzyme in synthesis of the gangliosides (GM3S) in humans results in a severe disease, GM3 deficiency syndrome, in which the entire ganglio‐series is deficient.

The various functions of the sphingolipids are matched by an unprecedented structural complexity. We estimate that the ceramide backbone can exist in more than 200 variants (Hannun & Obeid, 2018), and likewise, the glycolipids exist as several dozen and perhaps up to 300 molecules with distinct headgroups (Hakomori, 2008). The combination of the two yields theoretically upward of 60,000 distinct species! It should be recognized that these variations (in both the backbone and the head group) are products of combinatorial metabolic pathways (Kolter et al, 2002; Hannun & Obeid, 2018).

Phenotypically, the complexity of structure of the glycosphingolipids is matched by drastic changes in the levels of individual molecules that occur during development and in cell and tissue differentiation. This for example has allowed their use as markers of specific cancers and of specific differentiated states. In an important study, Liang et al (2010), using advanced MALDI‐MS and MS/MS techniques, defined a significant class switch in glycosphingolipids from the globo‐ and lacto‐series to the ganglio‐series during neural differentiation of human embryonic stem cells with the expression of Gb3S being suppressed, whereas that of GM3S induced upon differentiation. The significance of this drastic switch or its mechanisms, however, was not elucidated.

Enter Russo et al (2018) who pursue the mechanisms of this switch. They define a major component of a circuit that begins with loss of globosides which results in induction of the epigenetic modifier AUTS2 leading to induction of GMS3S and formation of gangliosides (Fig 1).

They first corroborate the results of Liang et al (2010) on the switch in GSLs to the ganglio‐series and the requisite changes in their enzymes of synthesis in murine embryonic stem cells (El4‐mESCs). They then provide evidence that the switch to synthesis of ganglio‐series is a result and not an inducer of differentiation (using various inhibitors of GSL synthesis as well as silencing of GM3S). In contrast, treatment of cells with individual GSLs of the globo‐series (but not of the ganglio‐series) inhibited neural differentiation (monitored by evaluation of specific markers) and inhibited the induction of GM3S and the switch to the ganglio‐series. Together, these results demonstrate a clear role for globosides in maintaining an undifferentiated state. They also suggest that the downstream mediator(s) of the action of globo GSLs is a broader regulator of differentiation rather than being only a specific regulator of GM3S.

In search of such mediators, Russo et al (2018) conducted studies in HeLa cancer cells. Interestingly, the authors provide evidence that in a mixed population of cells, individual cells express either ganglio or globo‐series GSLs but not both, a phenomenon previously noted in Vero cells (Majoul et al, 2002). They isolated these specific populations and found them to be selectively enriched in the machinery required for the synthesis of the respective GSLs. Moreover, inhibiting or silencing components in the pathway leading to synthesis of globo GSLs induced the enzymatic and lipidic switch to the ganglio‐series, thus disclosing similar relationships to embryonic stem cells.

Mechanistically, the authors then show that globo‐series GSLs e.g. Gb3 act (indirectly) on the GM3S promoter to silence it, and therefore they embarked on a microarray analysis that luckily yielded a small number of genes whose expression increased upon silencing globo synthesis. In particular, they focus on Auts2 (autism susceptibility candidate 2) as a candidate mediator of action of globo GSLs on the GM3S promoter. AUTS2 is known to control the expression of many neuronal genes through stimulating histone acetylation, and indeed, Russo et al (2018) find that AUTS2 binds the GM3S promoter and induces expression of GM3S through acetylation with a prominent role for the p300 acetyltransferase. Moreover, the relationship of Gb3S and AUTS2 extended to regulation of additional genes, many of which are prominent in brain.

Armed with these insights, the authors then validate these mechanistic relationships using mouse brain by showing that AUTS2 does bind the GM3S promoter in adult mouse brain; they also note an increase in Auts2 expression upon neural differentiation which followed a drop on globosides. Administration of globosides to E14mESCs prevented the increase in Auts2. Knockdown of Auts2 prevented the increase in ganglio‐series GSLs and inhibited expression of differentiation markers.

This is elaborate and impressive work that defines novel and important connections between at least three major players: GB3S and globosides; AUTS2 (and coregulators p300 and PRC1); and GM3S and the globosides (Fig 1). In this context, it is interesting that mutations in GM3S, AUTS2, and p300 all have human neurologic diseases with overlapping manifestations.

These studies open the way for additional important questions. First, how do globosides regulate AUTS2 expression? These are lipids that exist primarily on the surface of cells. Somehow these lipids must initiate signaling that results in reprogramming. This conundrum with GSLs showing “signaling” functions without having clear receptors has been noted repeatedly in the past, with regulation of specific receptors such as EGFR being proposed (Hakomori, 2008). As a corollary, do individual globosides regulate distinct programs or do they overlap in functions? Are there targets in addition to AUTS2?

Another important question relates to how is GB3S synthase suppressed in the early phases of the differentiation process. In this context, a study by Ojima et al (2015) suggests that the primary stemness transcription factors (Oct4, Klf4, Sox2, and c‐Myc) may promote production of globosides, possibly through maintaining the activity of GB3S. Alternatively, early regulators of neural differentiation may suppress the expression and activity of GB3S, thus launching the pathway identified by Russo et al (2018).

Tantalizingly, the intimate relationship to AUTS2 raises possibilities of roles of defects in GSLs in autism pathogenesis.

These are all exciting questions, arising from this work, that merit serious considerations.

The EMBO Journal (2018) 37: e99221