Sequences at the Interface of the Fifth Immunoglobulin Domain and First Fibronectin Type III Repeat of the Neural Cell Adhesion Molecule Are Critical for Its Polysialylation *

Polysialic acid is an anti-adhesive glycan that modifies a select group of mammalian proteins. The primary substrate of the polysialyltransferases (polySTs) is the neural cell adhesion molecule (NCAM). Polysialic acid negatively regulates cell adhesion, is required for proper brain development, and is expressed in specific areas of the adult brain where it promotes on-going cell migration and synaptic plasticity. The first fibronectin type III repeat (FN1) of NCAM is required for polysialylation of the N-glycans on the adjacent immunoglobulin-like domain (Ig5), and acidic residues on the surface of FN1 play a role in polyST recognition. Recent work demonstrated that the FN1 domain from the unpolysialylated olfactory cell adhesion molecule (OCAM) was able to partially replace NCAM FN1 (Foley, D. A., Swartzentruber, K. G., Thompson, M. G., Mendiratta, S. S., and Colley, K. J. (2010) J. Biol. Chem. 285, 35056–35067). Here we demonstrate that individually replacing three identical regions shared by NCAM and OCAM FN1, 500 PSSP 503 (PSSP), 526 GGVPI 530 (GGVPI), and 580 NGKG 583 (NGKG), dramatically reduces NCAM polysialylation. In addition, we show that the polyST, ST8SiaIV/PST, specifically binds NCAM and that this binding requires the FN1 domain. Replacing the FN1 PSSP sequences and the acidic patch residues decreases NCAM-polyST binding, whereas replacing the GGVPI and NGKG sequences has no effect. The location of GGVPI and NGKG in loops that flank the Ig5-FN1 linker and the proximity of PSSP to this linker suggest that GGVPI and NGKG sequences may be critical for stabilizing the Ig5-FN1 linker, whereas PSSP may play a dual role maintaining the Ig5-FN1 interface and a polyST recognition site.

Introduction

The neural cell adhesion molecule (NCAM) is a member of the immunoglobulin superfamily of proteins (1). It engages in both heterophilic and homophilic interactions that allow cell-cell adhesion and signal transduction (for review, see Refs. 2 and 3). NCAM is also the primary substrate for the two polysialyltransferases (polySTs) ST8SiaIV/PST (PST) and ST8SiaII/STX (4,–8). These two enzymes are capable of attaching long chains of α2,8-linked sialic acid residues to N-linked glycans in the fifth immunoglobin-like domain (Ig5) of NCAM (9). These long chains of negatively charged sialic acid attenuate the interactions of NCAM and other cell surface adhesion molecules (10, 11, 16), and this is critical in developing embryos and neonates for proper neuronal cell migration and differentiation and brain development (for review, see Refs. 12 and 13).

Mice null for NCAM show mild defects such as a reduced olfactory bulb size, decreased mossy fiber fasciculation, and deficits in spatial learning (14). In contrast, mice lacking the two polySTs show severe defects in brain architecture, hydrocephaly, and most die within 4 weeks of birth (15). A triple knock-out of NCAM and the polySTs rescues the lethal phenotype, indicating that polysialic acid is needed to prevent early and inappropriate cell adhesion and differentiation (15). In mammals, polyST expression decreases soon after birth, and in the adult animal NCAM is mostly unpolysialylated except in a few specific areas of the brain that require on-going cell migration and synaptic plasticity, such as the hippocampus, hypothalamus, and olfactory bulb (13, 17,–20). In addition, polysialic acid is aberrantly expressed in several pediatric and adult cancers, such as neuroblastoma, small and non-small cell lung carcinoma, and Wilms’ tumor (21,–26). Polysialic acid re-expression in cancer cells has been suggested to increase tumor invasiveness and to promote tumor growth by down-regulating NCAM signaling that activates tumor suppression pathways (27).

Polysialic acid is expressed on a very small subset of mammalian proteins with NCAM being the major substrate for the polySTs. Polysialic acid is found on the O-glycans of dendritic cell neuropilin-2 (28) and the scavenger receptor, CD-36, in milk (29). It is also found on the N-glycans of the α subunit of the voltage-dependent sodium channel (30) and a small population of the synaptic cell adhesion molecule, SynCAM1, expressed in NG2 glia cells (31). In addition, the two polySTs are capable of autopolysialylation (32, 33). This limited number of polyST substrates as well as the observation that polysialic acid is added preferentially to glycans linked to NCAM as compared with free glycans (34, 35) led to the hypothesis that polysialylation is a protein-specific modification event requiring an initial protein-protein interaction between polyST and substrate.

There are three known NCAM isoforms. NCAM180 and NCAM140 are transmembrane proteins that vary in the length of their cytoplasmic tail (tail), whereas NCAM120 associates with the membrane via a glycosylphosphatidylinositol anchor (1). The extracellular portion of all three NCAM isoforms is constant and is composed of five Ig domains and two fibronectin type III repeats. Although NCAM contains six consensus sites for N-linked glycosylation, the majority of polysialic acid is added to glycans attached to Asn449 and Asn478, the fifth and sixth N-glycosylation sites located on Ig5 (9). Previous work from our laboratory has shown that the first fibronectin type III repeat (FN1) of NCAM is required for the polysialylation of the N-glycans on the adjacent Ig5 domain (38). A truncated NCAM protein consisting of just the Ig5-FN1-transmembrane (TM)-cytoplasmic tail (tail) (NCAM4, ) can be polysialylated, but proteins containing only Ig5-TM-tail or lacking the FN1 domain (ΔFN1, ) are not (38, 39). In addition a truncated NCAM protein consisting of FN1-FN2-TM-tail (NCAM7, ) is weakly polysialylated on FN1 O-glycans. These results suggest that the minimal domains required for NCAM N-glycan polysialylation are Ig5-FN1 and suggest a model in which the recognition and binding of the FN1 domain by the polyST positions it to polysialylate N-glycans on Ig5, whereas in the absence of Ig5 the polySTs are able to bind FN1 and polysialylate O-glycans on FN1 (38,–40). To investigate this model, a closer examination of NCAM FN1 was required.

Fibronectin type III repeats are present in up to 2% of all human proteins (41), yet very few proteins are polysialylated, implying that NCAM FN1 is intrinsically unique, or in the context of NCAM, it allows for specific protein interaction with the polySTs. Molecular modeling and structural analysis of NCAM FN1 showed two novel features not seen in other fibronectin type III repeats, a surface acidic patch composed of Asp506, Asp520, Glu521, and Glu523 and an α-helix linking strands 4 and 5 of the FN1 β sandwich (39, 40). Two other unique FN1 sequences, 510PYS512 and 516QVQ518 (QVQ), were also identified (42). The role of all four sequences in NCAM polysialylation was analyzed by mutagenesis and the creation of chimeric proteins. Although the PYS sequence was shown to be required for the maintenance of a conformation of the FN1 domain that enhanced the biosynthesis of sialylated O-glycans and promoted O-glycan polysialylation, the acidic surface patch, α-helix, and QVQ sequence were shown to play roles in polyST recognition and positioning (39, 40, 42).

We found that when either the FN1 α-helix or QVQ sequence of NCAM was replaced with alanines, the polySTs continued to recognize the mutant proteins but added polysialic acid to O-glycans on FN1 rather than N-glycans on Ig5 (39, 42), suggesting that these sequences play a role in positioning the polySTs for N-glycan polysialylation. The importance of the FN1 acidic patch varies in N-glycan and O-glycan polysialylation. Replacing the acidic patch residues with alanine led to a dramatic decrease in, or even eliminated, the O-glycan polysialylation of the truncated NCAM7 protein (FN1-FN2-TM-tail) (Ref. 42; see also and C). In contrast, replacing these residues with alanine only slightly decreased NCAM N-glycan polysialylation in WT NCAM (Ref. 39; see B). However, replacing these residues with arginine dramatically decreased NCAM N-glycan polysialylation and suggested that additional FN1 residues are likely to function together with the acidic patch to allow the polySTs to modify Ig5 N-glycans (Ref. 39; see B).

In the course of investigating the requirements for NCAM polysialylation, we found that the FN1 domain from the unpolysialylated olfactory cell adhesion molecule, OCAM, could partially replace the FN1 domain of NCAM to allow N-glycan polysialylation (42). This result lead us to consider the possibility that sequences common to NCAM and OCAM FN1 domains are critical for polyST recognition and polysialylation.

In this study we evaluate three identical regions shared by NCAM and OCAM FN1 domains and their role in NCAM polysialylation and polyST binding. General requirements for polyST-NCAM binding are also investigated. Our results show that all three regions are critical for NCAM polysialylation and suggest that two sequences which form loops flanking the Ig5-FN1 linker function to stabilize this region, whereas the a third sequence, which is adjacent to the Ig5-FN1 junction, may play a dual role maintaining both the Ig5-FN1 interface and a polyST binding surface.