Glycobiology Advance Access originally published online on December 15, 2006
Glycobiology 2007 17(3):249-260; doi:10.1093/glycob/cwl075
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Published by Oxford University Press 2006
Polysialic acid bioengineering of neuronal cells by N-acyl sialic acid precursor treatment
2 Institute for Biological Sciences, National Research Council of Canada, 100 Sussex Drive, Ottawa K1A 0R6, Canada
1 To whom correspondence should be addressed; Tel: +1 613 990 0821; Fax: +1 613 941 1327; e-mail: harry.jennings{at}nrc-cnrc.gc.ca
Received on February 24, 2006; revised on December 5, 2006; accepted on December 8, 2006
 |
Abstract |
|---|
 Top Abstract IntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
The inherent promiscuity of the polysialic acid (PSA) biosynthetic pathway has been exploited by the use of exogenous unnatural sialic acid precursor molecules to introduce unnatural modifications into cellular PSA, and has found applications in nervous system development and tumor vaccine studies. The sialic acid precursor molecules N-propionyl- and N-butanoyl-mannosamine (ManPr, ManBu) have been variably reported to affect PSA biosynthesis ranging from complete inhibition to de novo production of modified PSA, thus illustrating the need for further investigation into their effects. In this study, we have used a monoclonal antibody (mAb) 13D9, specific to both N-propionyl-PSA and N-butanoyl-PSA (NPrPSA and NBuPSA), together with flow cytometry, to study precursor-treated tumor cells and NT2 neurons at different stages of their maturation. We report that both ManPr and ManBu sialic acid precursors are metabolized and the resultant unnatural sialic acids are incorporated into de novo surface sialylglycoconjugates in murine and human tumor cells and, for the first time, in human NT2 neurons. Furthermore, neither precursor treatment deleteriously affected endogenous PSA expression; however, with NT2 cells, PSA levels were naturally downregulated as a function of their maturation into polarized neurons independent of sialic acid precursor treatment.
Key words: glycan bioengineering / human neurons / N-acyl mannosamines / polysialic acid / tumor cell
|
Introduction |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Sialic acids are the most ubiquitous sugars found in eukaryotic cells primarily residing in terminal positions of cell surface glycoconjugates, where they play critical roles in biological events such as cell–cell recognition, migration and homing, and protein stability, as well as serving as substrates for infectious agents (Varki 1992
; Keppler et al. 1995). Polysialic acid (PSA), a linear homopolymer composed of 
-(2-8)-linked N-acetyl-neuraminic acid (NeuAc) residues, is a unique biological form of sialic acid that is important during central nervous system development, neural regeneration and synaptic plasticity, as well as being an important cancer-associated antigen involved with metastasis (Theodosis et al. 1991; Szele et al. 1994; Tang et al. 1994; Daniel et al. 2001). PSA is most often found on neural cell adhesion molecules (PSA-NCAMs) as N-linked glycans and found most prevalently in embryonic tissues as opposed to its progressive loss in adult tissues, except for regions of high plasticity (Theodosis et al. 1991; Buttner et al. 2003).
Several studies have taken advantage of the permissiveness of the sialic and PSA biosynthetic pathways to remodel the cell surface landscape of tumor (Liu et al. 2000; Zou et al. 2004), neuronal (Charter et al. 2000; Mahal et al. 2001; Horstkorte et al. 2004), and glial cells (Schmidt et al. 1998) both in vitro and in vivo (Prescher et al. 2004). This can be performed by replacing N-acetyl-mannosamine (ManAc), the physiological precursor of sialic acid, by an exogenous source of unnatural N-acyl mannosamines resulting in the introduction of these unnatural sialosides into surface glycoconjugates (Keppler et al. 2001). PSA synthesis arises in part through the synergistic actions of two major polysialyltransferases ST8SiaII and ST8SiaIV. In vitro studies using soluble enzyme extracts suggested that ST8SiaII is biased toward initiation and lower levels of NCAM polysialylation, whereas ST8SiaIV is more associated with an enhanced ability to polysialylated oligomeric NCAM-sialic acids (Kitazume-Kawaguchi et al. 2001; Angata et al. 2002). Contrasting these findings are recent studies using selective polysialyltransferase gene-targeting studies in mice that suggested overall polysialylation levels in neonatal brain were reduced in ST8SiaII–/– relative to ST8SiaIV–/– knockouts, and PSAs contained similar degrees of polymerization within the mutant strains studied (Weinhold et al. 2005; Galuska et al. 2006). Despite the fact that several groups have reported success in using a variety of N-acyl mannosamines or their corresponding cytidine monophosphate (CMP) derivatives to successfully incorporate unnatural sialic acids into surface PSA (Charter et al. 2000; Liu et al. 2000; Mahal et al. 2001; Charter et al. 2002; Horstkorte et al. 2004), there is still some debate as to the effects of administering the unnatural precursor N-butanoyl-mannosamine (ManBu) on both de novo-modified PSA and native PSA biosynthesis. Mahal et al. (2001) originally reported that ManBu precursor treatment effectively halted PSA surface expression on several PSA-expressing cell types, including human NT2-derived neurons, whereas its N-acyl homolog N-propionyl-mannosamine (ManPr) was metabolized without altering endogenous PSA expression. It followed from these studies that the precursor molecule ManBu would be important not only in studying the role of PSA-NCAM in the central nervous system and disease, but also as a potential therapeutic. More recently, Horstkorte et al. (2004) re-examined the effects of ManBu precursor treatment on PSA-expressing cells, leading to mixed results. In some instances, ManBu precursor treatment did not affect surface PSA levels, but when using human NT2 neurons, precursor treatment resulted in the loss of native PSA expression consistent with the prior study. This PSA abrogative effect of ManBu, as well as other N-acyl mannosamines, was explained by differences in the levels of the polysialyltransferases expressed in these cells where they found specific inhibition of ST8SiaII by N-acyl mannosamines, whereas ST8SiaIV was more permissive.
Our recent focus has been on the remodeling of sialylglycoconjugates on the surface of cancer cells, using similar bioengineering strategies to introduce unnatural carbohydrate antigens. These modifications serve to increase the immunogenicity of these antigens or, alternatively, to metabolically label the tumor cells with nonnatural sialic acids, followed by their specific targeting with therapeutic antibody. We have applied this strategy both in vitro and in vivo to tumor cells carrying key sialylforms such as GD3, GM3, and PSA (Liu et al. 2000; Zou et al. 2004). During the course of our PSA investigations, we were greatly aided by the possession of mAb 13D9, which specifically recognizes an extended epitope [nine contiguous N-propionyl sialic acids N-propionyl-neuraminic acid (NeuPr)] (Pon et al. 1997; MacKenzie and Jennings 2003) that was developed in conjunction with our efforts to generate a Group B meningococcus vaccine (Jennings et al. 1986). With the use of this and other mAbs specific to the nonnatural sialylglycoconjugates (Zou et al. 2004), we were able to directly assess the effects of exogenous unnatural N-acyl mannosamine precursor treatment of tumor cells, using flow cytometry. We reasoned that mAb 13D9 in conjunction with mAb 735, which is specific to native PSA (Frosch et al. 1985), could be used to assess the controversial effect of ManBu precursor treatment on PSA biosynthesis, provided there is suitable cross-recognition of N-butanoyl-PSA (NBuPSA) (Horstkorte et al. 2004). We now provide evidence from this study that ManBu does not act as a general PSA chain terminator and that it, as well as ManPr, is metabolized and expressed as modified cell surface PSAs in both tumor cells and NT2 neurons, without reducing levels of native PSA expression.
|
Results |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
mAb 13D9 is capable of detecting NBuPSA polysaccharide on ManBu sialic acid precursor-treated RMA-s leukemia cells
mAb 13D9 specifically recognizes extended epitopes of N-propionyl-PSA (NPrPSA) and has been extensively characterized (Pon et al. 1997
; MacKenzie and Jennings 2003). To ascertain whether mAb 13D9 could be used as a probe to detect its related homolog NBuPSA, we first screened the mAb against a series of human serum albumin (HSA)–polysaccharide constructs by indirect enzyme-linked immunosorbent assay (ELISA) (Figure 1A and B). NPrPSA and NBuPSA modifications differ only by methylene unit extensions of the sialic acid N-acyl regions relative to native PSA. As expected, mAb 13D9 was highly specific to its homologous antigen NPrPSA contrasting its complete lack of recognition to native PSA. MAb 13D9 was, however, completely cross-reactive with NBuPSA and indeed its recognition of this antigen was virtually indistinguishable from NPrPSA (EC50 = 0.12 versus 0.17 µg/mL for NBuPSA versus NPrPSA, respectively). To further address the specificities of mAbs 13D9 or 735 (PSA-recognizing mAb) (Frosch et al. 1985), competitive inhibition ELISA was performed using either native PSA or synthetic NPr- and NBuPSA polysaccharides as binding inhibitors. Competitive inhibition of mAb 735 binding to PSA-HSA was only observed with PSA and not with either NPr- or NBuPSA inhibitors (1370 versus <0.1 µg–1 for 50% inhibition, respectively (Figure 1C). Conversely, both NPr- and NBuPSA were effective inhibitors against mAb 13D9 binding to NPrPSA-HSA (NPrPSA: 156 µg–1; NBuPSA: 340 µg–1 for 50% inhibition) or NBuPSA-HSA (NPrPSA: 31 µg–1; NBuPSA: 156 µg–1 for 50% inhibition) antigens, whereas N-acetyl polysialic acid (NAcPSA) was not (<0.1 µg–1 in all cases).
|
), NPrPSA (
), and NBuPSA (
). [(C) and (D)] Graded amounts of pure NAcPSA (
), NPrPSA (
), and NBuPSA (
) polysaccharides were used to competitively inhibit binding of mAb 735 to NAcPSA-HSA (C) or mAb 13D9 (D) to NPrPSA- and NBuPSA-HSA antigens in ELISA experiments. Bars represent the inverse amount of inhibitor required for 50% inhibition of mAb binding to each coating antigen without the presence of polysaccharide inhibitor.We next evaluated whether the PSA-expressing RMA-s murine tumor cell line would be capable of metabolizing ManPr and ManBu with their eventual conversion and surface expression of the unnatural PSAs. Untreated RMA-s cells expressed native PSA detected using mAb 735 coupled with flow cytometry and live-cell gating [mean channel fluorescence (mcf = 56); Figure 2A]. There was little-to-no background staining with mAb 13D9 on untreated RMA-s cells (mcf = 7.3), indicating that nonspecific binding was not a factor. Treatment of RMA-s cells with 10 mM ManPr for 3 days and staining with mAb 13D9 resulted in the expression of an extended epitope composed of NPr-sialic acid residues (mcf = 54.1; Figure 2B), as was previously observed in other studies (Pon et al. 1997
; MacKenzie and Jennings 2003). Noticeably, ManPr treatment marginally reduced native PSA levels (mcf = 46.5 versus 56 for treated versus untreated cells) but did not abolish its presence on the cell surface. Extending the length of precursor treatment to 5 days did not significantly increase NPrPSA or significantly decrease native PSA expression. RMA-s cells treated similarly with ManBu precursor and mAb 13D9 detected cell surface levels of NBuPSA (mcf = 61.7; Figure 2B). Consistent with ManPr treatment, ManBu precursor lowered but did not abolish the expression of native PSA expression (mcf = 38.9 versus 56 for treated versus untreated cells). To try and ascertain the nature of the neo-epitope formed on precursor-treated RMA-s cells, a series of competitive inhibition flow cytometry assays were performed using small PSAs of defined length (12–22 residues), containing mixed ratios of either NeuPr:NeuAc or N-butanoyl-neuraminic acid (NeuBu):NeuAc (Figure 2C–E). Consistent with our ELISA experiments, mAb 735 binding to untreated as well as to ManPr- and ManBu-treated RMA-s cells was only effectively inhibited by PSAs, in which all residues were of the native NeuAc form. Introduction of between 35% and 100% NeuPr or NeuBu residues significantly diminished the ability to block surface recognition. Conversely, in ManPr- and ManBu-treated cells (Figure 2D–E), only PSAs containing 80–100% NeuPr or NeuBu were able to compete for mAb 13D9 surface binding to the respective precursor-treated cells.
|
Overall, these results indicate that mAb 13D9 is a suitable antibody probe to detect not only its homologous NPrPSA antigen but also NBuPSA, and that both of these antigens were present on the surface of RMA-s tumor cells pretreated with the unnatural sialic acid precursors ManPr and ManBu, respectively (similar results were obtained with SH-SY5Y human neuroblastoma cells—data not shown). Furthermore, the de novo surface PSAs appear to be composed primarily of either contiguous NeuPr or NeuBu stretches on the basis of mAb 13D9 recognition, and neither precursor treatment appreciably affected the coexpression of native PSA, as observed by others (Mahal et al. 2001
).
The effects of unnatural sialic acid precursor treatment in the NTera 2/cl.D1 system for deriving human neurons
Numerous studies examining human neuronal physiology have employed neurons generated from the pluripotent teratocarcinoma progenitor line NTera 2/cl.D1 (NT2) upon retinoic acid (RA) treatment (Andrews 1984; Lee and Andrews 1986). Neurons generated in this manner have been shown to possess many of the morphological and physiological features of primary human neurons (Pleasure et al. 1992). Studies investigating PSA-NCAM expression on NT2 neurons used neurons derived from RA-treated parent cells, following long-term culture, and are henceforth referred to as classically derived neurons (Mahal et al. 2001; Horstkorte et al. 2004). Because different effects of N-acyl mannosamine precursor treatment have been observed (Mahal et al. 2001; Horstkorte et al. 2004), we chose to re-address this question using a different system to generate NT2 neurons which allowed for the separate analysis of NT2 neurons at different stages of maturation. NT2 cells treated with RA and cultured in 100-mm bacterial plates minimized adherence and promoted the formation of large clusters of postmitotic neurospheres (Figure 3A). These cells are fully committed to a neuronal lineage possessing exclusive neural cell markers such as NSE and NF68 but lacking in MAP2ab protein associated with neuritogenesis (Horrocks et al. 2003), hence their designation as relatively immature. Plating of these neurospheres onto basement membranes such as matrigel rapidly promoted cytoskeletal changes and the extensive formation of axons and dendrites over the course of a few days (neurite-expressing neurospheres), which became increasingly more elaborate with time (Figure 3B). Generation of fully mature classical NT2 neurons was accomplished using standard techniques (Sandhu et al. 2003) or with slight modifications (Horrocks et al. 2003), both involving culture periods of in excess of 2 months before analysis. Morphologically, the fully mature neurons were much smaller compared with neurospheres and displayed extensive neuritic networks (Figure 3C).
|
With these variations in hand, we proceeded to address the effects of treating NT2 neurons with sialic acid precursors. Starting with the less mature neurospheres, we first evaluated surface expression of native PSA, using mAb 735 staining and flow cytometry (Figure 4A), revealing a homogeneous population of live cells [as determined by FS/SS and propidium iodide (PI) exclusion parameters] positive for PSA (mcf = 365). Subjecting neurospheres to 10 mM ManPr and ManBu precursors for 3 days and evaluation with mAbs 735 and 13D9 revealed two important surface staining patterns. NPrPSA was strongly detected on ManPr-treated cells (mcf = 235); however, somewhat surprisingly, NBuPSA was also present at high density on ManBu-treated cells (mcf = 318). Furthermore, neither ManPr nor ManBu precursor treatment abrogated the expression of native PSA on these cells and indeed PSA expression levels remained high, although somewhat diminished, compared with nontreated control cells (mcf = 149 and 144 versus 365 for ManPr, ManBu, and control cells, respectively).
|
We next evaluated the capacity of more mature neurons (neurite-expressing neurospheres and classical neurons; Figure 3B and C) to incorporate unnatural sialic acid precursors within growing surface PSA chains. Neurospheres were layered onto matrigel surfaces for 3–7 days to promote neurite expression, followed by sialic acid precursor treatment and flow cytometric analysis as before. A striking difference was observed in the forward scatter/side scatter (FS/SS) profile of neurospheres and neurite-expressing neurospheres (Figure 4A and B), indicating a shift to a smaller phenotype that was consistent with our morphological evaluations. Although NPrPSA or NBuPSA expression was still observed (mcf = 44 NPrPSA; 102 NBuPSA), unexpectedly the majority of cells did not express either polysaccharide (88% and 86% negative for NPrPSA and NBuPSA, respectively). When native PSA was assessed on the same precursor-treated cells, virtually the same staining pattern was observed, with the majority of cells staining negative for PSA expression. Significantly, matured neurons not subjected to prior precursor treatment also exhibited poor PSA expression on the majority of cells (80%). When classically derived NT2 neurons were similarly assessed using ManPr and ManBu precursor treatments, virtually identical results were obtained with neurite-expressing neurospheres, except that the loss of native PSA and the lack of newly formed NPrPSA or NBuPSA were more pronounced (Figure 4C). Also of note were the homogeneous small size and poor viability (approximately 50%) of the NT2 neurons at this stage of analysis and the presence of native PSA on contaminating NT2 astrocytes present within these latter cultures (data not shown).
Native PSA expression on human NT2 neurons is naturally downregulated as a function of maturation
On the basis of our observations that the level of native PSA found on NT2 neurons diminished as neurospheres matured on matrices regardless of sialic acid precursor treatment, we postulated that the loss of PSA is a natural consequence and occurs in conjunction with cytoskeletal changes and maturation to axon and dendrite expression by NT2 neurons. To address this hypothesis, we followed the expression of surface PSA on neurospheres as they matured over time on a matrigel surface. Initially, FS/SS settings were optimized and gates set to encompass the majority of cells which appeared as a homogeneous population (gate R4; 87% of cells; Figure 5). Within this gate, initial PSA expression was relatively high (mcf = 25), as had been observed previously in this study. After 18 h on matrigel, the FS/SS profile did not change significantly and the majority of cells fell within the neurosphere gate (72%) and remained positive for PSA (mcf = 21). After 3 days on matrigel, however, three populations of cells were discerned on the basis of their FS/SS scattergram. The majority of the initial NT2 neurospheres underwent a dramatic shift to a smaller, less granular profile (R1 gate), representing approximately 85% of the viable cells present, with only 9% of the original cell population falling within the neurosphere gate (R4). Additionally, a third minor population, which itself is positive for PSA and possessing even larger size and granularity profiles, was observed, which is consistent with astrocyte contamination on the basis of independent flow analysis of NT2 astrocytes derived for comparison purposes (data not shown) (Sandhu et al. 2002). Native PSA expression was localized to those cells remaining in the original neurosphere gate (R4 gate) and was not present on the morphologically smaller yet major cell subset (R1 gate). By day 7 on matrigel, a few cells (approximately 7%) remained in the neurosphere gate, which were the only source of PSA-positive material at levels comparable with initial values (mcf = 73 versus 25). From these studies, native PSA downregulation appears to be a natural event that proceeds in conjunction with neuronal maturation and expression of neuritic processes and is independent of unnatural sialic acid precursor treatment.
|
Native and modified PSA can be detected on NT2 neurons by western blot analysis
To date, most studies assessing the impact of unnatural sialic acid precursors have relied on western blot analysis of whole-culture lysates (Mahal et al. 2001
; Charter et al. 2002; Horstkorte et al. 2004). We chose to perform similar analyses, using our NPrPSA- and NBuPSA-specific mAb 13D9, along with mAb 735 specific to native PSA, to evaluate the PSA expression on neurospheres, neurite-expressing neurospheres, and classically derived neurons (Figure 6A). When protein lysates from a 3-day treatment of neurospheres with 10 mM ManPr and ManBu were probed with mAb 13D9, a broadband with an apparent molecular weight from 190 to >230 kDa was detected corresponding to the respective N-acyl PSA-NCAM molecule (lanes 2 and 3), but not for control untreated cells (lane 1). The NBuPSA-NCAM band appeared at slightly lower molecular weights relative to NPrPSA-NCAM (Mr = 223 versus 231 kDa), suggesting a lower degree of NCAM substitution. Probing lysates with mAb 735 revealed identical high molecular weight bands for all cell preparations, indicating neither precursor treatment had any effect on native NCAM polysialylation (lanes 4–6). Probing with NCAM-specific mAb OB11 revealed an equivalent molecular weight profile on all preparations at around Mr = 180–200 kDa, typical of low or nonpolysialylated NCAM (Buttner et al. 2003). We were unable to detect the highly polysialylated species of NCAM owing to the masking nature of PSA (Daniel et al. 2001). In addition, OB11 detected faint bands (Mr = 140 kDa) from ManPr- and ManBu-treated neurospheres (lanes 8 and 9) corresponding to a lower molecular weight isoform of NCAM. When more mature neurons were screened (neurite-expressing or classically derived neurons), virtually identical immunoblot profiles were obtained when compared with the less mature neurospheres (Figure 6B). High-molecular-weight native PSA-NCAM (Mr > 200 kDa) was detected on either preparation, regardless of treatment with either ManPr or ManBu (lanes 6–10), and high-molecular-weight NPrPSA or NBuPSA-NCAM (Mr > 200 kDa) was detected specifically on ManPr- or ManBu-treated cells, respectively (lanes 1, 2, and 4). Our immunoblot studies are consistent with our flow cytometry results in that NT2 neurons are able to incorporate ManPr and, in particular, ManBu metabolites into downstream NPrPSA and NBuPSA-NCAM glycoproteins. Furthermore, the consistent molecular weight pattern of native PSA observed, regardless of the maturity of the neurons and the nature of precursor treatment, seemed to indicate that native PSA expression was not adversely affected by prior precursor treatment.
|
, untreated; neuron-Pr, ManPr treated; and neuron-Bu, ManBu treated.)Both native N-acetyl- and N-butanoyl-PSA chains are present as functional epitopes on ManBu precursor-treated NT2 neurospheres
We have been successful in detecting NBuPSA and PSA on ManBu-treated NT2 neurons directly by using flow cytometry and indirectly by western blots of cell lysates. However, we wanted to satisfy ourselves that both of these polysaccharides were present in sufficient quantities on the surface of NT2 neurons that they in turn could behave as functional determinants. Addressing this question, we performed chromium release cytotoxicity studies on neurospheres treated with 10 mM ManBu for 4 days (Figure 7). In the presence of graded amounts of mAb 13D9 and baby rabbit complement for 4 h, we observed a dose-dependent specific killing of ManBu-treated neurospheres (31.2% specific cytotoxicity at 0.5 µg mAb 13D9), whereas untreated neurospheres were unaffected (data not shown). Similarly, in the presence of mAb 735, dose-dependent killing of the same ManBu-treated neurospheres was also observed (34.9% specific cytotoxicity at 0.5 µg mAb 735). In all instances, the level of spontaneous cell death was <20%. Taken together, these results demonstrate that both engineered NBuPSA and residual native PSA are sufficiently present on the surface of NT2 neurospheres to be able to mediate functional cell killing, suggesting the effects of PSA engineering can be more extensively studied in this neurological model.
|
|
Discussion |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
We and others have amply demonstrated the flexibility of the sialic acid biosynthetic pathway to incorporate nonnatural sialic acids into PSA (Charter et al. 2000
; Liu et al. 2000; Mahal et al. 2001; Charter et al. 2002; Horstkorte et al. 2004). In this study, we have taken advantage of mAb 13D9, which is specific for extended epitopes of NPrPSA (Pon et al. 1997) and which we have now shown to be fully cross-reactive with NBuPSA, to characterize the effects of ManBu sialic acid precursor treatment on several cell lines, including embryonal carcinoma stem cell-derived human neurons. We show that both ManPr and ManBu are efficiently utilized by both tumor cells and relatively immature neurons, given that these cells readily express detectable and functional amounts of NPrPSA and NBuPSA on their cell surfaces; the composition of which requires a high percentage of NeuPr or NeuBu residues for mAb 13D9 recognition. Although ManBu precursor treatment results in lower expression of PSA on NT2 neurons, neither of the small molecule sialic acid precursors behaved as metabolic inhibitors nor abrogated PSA expression, as was observed by others (Mahal et al. 2001; Horstkorte et al. 2004). Moreover, taking advantage of a system that allowed us to generate human NT2 neurons at different stages of maturation (Horrocks et al. 2003), we have for the first time characterized time-related surface PSA expression in this cell system and have demonstrated that PSA downregulation is a natural consequence of NT2 neuron maturation correlating with neuronal cytoskeletal rearrangements and neurite expression.
The results from this study would indicate that in both murine and human PSA-expressing tumor cells, as well as in NT2-derived human neurons, both ManPr and ManBu are metabolized and incorporated into cell surface PSA chains. Furthermore, with the exception of NT2 neurons (see below), neither precursor treatment affects the endogenous expression of native PSA despite the use of high precursor concentrations (10 mM) for period up to 5 days. This contrasts with the result of Mahal et al. (2001), who originally found that treating either tumor or NT2 neurons with ManBu effectively halted native PSA expression, whereas ManPr not only preserved PSA expression, but was also metabolized and incorporated into growing PSA chains. Furthermore, although our results partially support the recent investigation by Horstkorte et al. (2004) in which they determined that ManBu precursor treatment of HL-60 tumor cells resulted in NCAM modification with NBuPSA, they are not consistent with their reports of abrogative effects of both ManPr and ManBu treatment on native PSA expression in NT2 neurons. We were facilitated in this study by mAb 13D9, which we have now shown to be fully cross-reactive with both NPrPSA and NBuPSA, thus allowing us to directly and quantitatively survey the cell surface environment of either ManPr or ManBu precursor-treated cells via flow cytometry. A flow cytometric approach also offered the additional advantages of localizing expression patterns to viable cells and to subset populations, thereby excluding other contaminating cell types or debris present within the cultures. This contrasts the approaches used in other studies (Mahal et al. 2001; Charter et al. 2002; Horstkorte et al. 2004) that analyzed whole-culture extracts by western blot, which poses a significant risk of the inclusion of contaminating parameters and therefore producing summary results. The presence of mAb 13D9 surface binding to ManPr or ManBu precursor-treated cells implies three critical events: (1) both ManPr and ManBu are metabolized and converted into CMP versions of NeuPr and NeuBu, respectively, (2) both of these CMP derivatives are acceptable substrates for at least one of a number of potential polysialyltransferases at play, and (3) oligomeric NeuPr and NeuBu are suitable PSA acceptor molecules for further NeuPr and NeuBu chain elaboration by the polysialyltransferases. We make this latter assumption on the basis of the exquisite and well-characterized binding properties of mAb 13D9 that cannot bind to less than nine contiguous NeuPr residues (Pon et al. 1997; MacKenzie and Jennings 2003). Support for this observation at the cell surface level (i.e. through in vivo labeling) comes from the results of our flow inhibition studies using small PSAs encompassing 1–2 binding epitopes for either mAb 735 or 13D9, in which we randomly replaced native NeuAc residues with increasing percentages of either NeuPr or NeuBu. We observed effective inhibition of mAb735 binding to cellular substrates only with PSA inhibitors composed exclusively of NeuAc residues which fell off quickly as the percentage of NeuPr or NeuBu increased to 35% or greater. Similarly, mAb 13D9 binding to surface components of ManPr- and ManBu-treated cells was disrupted only by PSA inhibitors with high NeuPr or NeuBu content. Taken together, de novo surface PSAs synthesized under the influence of either ManPr or ManBu are composed of either NeuPr or NeuBu exclusively, or at the very least, with long contiguous stretches of the modified sialic acids interspersed with regions containing endogenously produced NeuAc. Moreover, the finding that mAb 735 binding requires NeuAc homogeneity in the cellular PSA substrates substantiates the case that mAb 735 does not recognize PSA containing nonnatural building blocks, consistent with our in vitro ELISA studies.
In the special case of human NT2 neurons, we were fortunate to have employed an alternate method to generate these cells from the parent stem cell line, which complemented our direct analysis methodology, and provided a somewhat different perspective on the effects of N-acyl mannosamine treatment relative to prior studies (Mahal et al. 2001; Horstkorte et al. 2004). Initial interpretation of our results obtained by treating classically derived NT2 neurons with ManPr and ManBu precursors suggested that they were entirely consistent with these prior studies, in that the unnatural sialic acid precursors substantially reduced native PSA expression and were only poorly incorporated into de novo-modified PSAs. However, when we analyzed NT2 neurons at earlier points in their maturation (neurosphere stage), not only did ManPr and ManBu precursor treatment result in surface levels of NPrPSA and NBuPSA sufficient to serve as functional determinants, but neither treatment seriously affected the cell surface density of native PSA. This clearly indicates the capacity of NT2 neuronal cells to metabolize unnatural precursor molecules and to not inhibit native PSA expression (at least at this stage of their maturation), which is consistent with our prior results with murine and human tumor cells (Liu et al. 2000; and this study). Interestingly, both ManPr and ManBu precursor treatment resulted in lower levels of native PSA surface expression on NT2 and RMA-s tumor cells when compared with untreated controls, with the effect being largest upon ManBu treatment. This observation can be rationalized, given the observed production of de novo nonnatural PSAs and the potential for competition of glycosylation sites on NCAM molecules (Bork et al. 2005). Moreover, NCAM may be differentially substituted upon ManPr versus ManBu precursor treatment since we also observed in immunoblot studies a slight but discernible band shift toward lower molecular weight for NBuPSA-NCAM, which in turn may be a reflection of lower polysialyltransferase kinetics for NeuBu residues as previously hypothesized (Horstkorte et al. 2004). We found that the ability to metabolize the N-acyl mannosamine precursors was intimately associated with the maturation status of the NT2 cells, being greatest with committed neuronal cells at relatively early stages and lowest with fully mature polarized neurons. PSA expression and the ability to engineer it with sialic acid precursors resided only on populations that preserved the relatively immature neurosphere phenotype, a population that is severely depleted in classically derived NT2 neurons such as those used in prior studies (Mahal et al. 2001; Horstkorte et al. 2004).
Interestingly and to the best of our knowledge, not yet reported were our observations that native PSA expression on untreated NT2 neurons appeared to be maturation dependent and was naturally downregulated coincident with cytoskeletal rearrangements and the formation of extensive neurites, at least in the context of maturation on matrigel surfaces. This is in keeping with the well-known loss of PSA during fetal development and in ex vivo neuronal studies (Szele et al. 1994; Nakayama et al. 1998). That this occurs independent of sialic acid precursor treatment and was pronounced on long-term neuronal cultures containing the greatest proportion of fully mature polarized neurons suggests that the majority of similar stage neurons that were previously characterized (Mahal et al. 2001) were naturally devoid of surface PSA. This observation represents a complicating factor when trying to interpret the effects of ManPr and ManBu on neuronal PSA expression, especially given that only a minor fraction is PSA-positive and may help to explain some of the discrepancies between this and previous studies. Indeed, we also observed from our immunoblot analyses a relative trend for faster migration of NBuPSA-NCAM, thus reflecting lower levels of ManBu-induced polysialylation, which is somewhat, but certainly not as dramatic an effect, as was observed previously (Horstkorte et al. 2004). We also cannot rule out the possibility that PSA-NCAM expression on mature NT2 neurons, especially those found on neuritic processes, may be sensitive and lost in response to cell harvesting prior to our analysis by flow cytometry. Conversely, NCAM or PSA-NCAM extracted from whole-culture lysates suffers from the incertitude of its source, from viable versus nonviable cells, intracellular versus extracellular, or from differing cell types—all factors that can be discerned by the use of flow cytometric analyses. For instance, we often observed contaminating cell types, morphologically resembling NT2 astrocytes which themselves express PSA-NCAM, and in our hands, were found most prevalently in classically derived neuronal cultures such as those previously studied. This summary effect obtained upon whole-culture extraction helps to reconcile what appears to be high-level PSA-NCAM expression on classically derived NT2 neurons in our immunoblot analyses, as opposed to our flow cytometric observation of low PSA levels found specifically on discrete, viable mature NT2 neurons. Moreover, PSA-NCAM was assessed in a qualitative manner on the various maturation subgroups in which PSA-NCAM signal was the determining factor in immunoblot analysis rather than relative amount determinations.
In trying to reconcile the disparate effects of ManPr and ManBu precursor treatment on PSA expression in tumor cells versus NT2-derived neurons, Horstkorte et al. (2004) looked at the expression levels of the polysialyltransferases ST8SiaII and ST8SiaIV. They showed that the predominant polysialyltransferase ST8SiaII expressed in NT2 neurons can be inhibited by N-acyl analogs of sialic acid, whereas the much less abundant ST8SiaIV is fully permissive for these substrates, proposing this as an explanation for the observed PSA loss upon N-acyl mannosamine treatment within these cells. Following from their findings, our data suggest that both polysialyltransferases are still operational in morphologically immature neurospheres since ManPr and ManBu both serve as polysialyltransferase substrates, but upon further maturation, ST8SiaII most probably becomes increasingly dominant and, correspondingly, sensitive to the inhibitory effects of N-acyl sialic acids. The dominance of ST8SiaII in fully mature NT2 neurons (Horstkorte et al. 2004) and our observed natural loss of PSA independent of precursor treatment can be partially rationalized on the basis of in vitro studies demonstrating that this enzyme is associated with shorter, as opposed to longer, forms of PSA with ST8SiaIV (Angata et al. 2002). However, contrasting evidence, using ST8SiaIV knock-out mice in which overall neonatal brain PSA levels and length were found to be relatively unaffected (Galuska et al. 2006), demonstrates that the mechanism behind PSA downregulation is more complex than a simple change in polysialyltransferase expression patterns and requires further study. Regardless of the exact mechanism, in the scenario in which mature neurons are naturally downregulating their PSA and are associated with increasing ST8SiaII dominance, the observed differential inhibitory effects of ManPr and ManBu on PSA expression (Mahal et al. 2001; Horstkorte et al. 2004) may be more focused and specific to this unique and minor subset, rather than a generalized phenomenon. Our results also raise the question as to whether ManPr or ManBu precursors may influence growth, growth rates, or maturation rates of our neurosphere population, which in turn may also help to explain the differences observed between our study and others. Since differentiation of NTera 2/cl.D1 by RA results in postmitotic neurospheres that possess a full complement of neuronal markers (Horrocks et al. 2003), sialic acid precursor treatment will therefore not impinge upon cell growth or growth rates. However, it is not presently known whether N-acyl mannosamines may affect NT2 neuronal differentiation, and indeed, the differential effects observed with these precursors on PSA expression may be the result of enhancing the natural neuronal maturation process (and therefore loss of PSA expression), as has been previously observed with neuron-like rat PC-12 cells (Buttner et al. 2002). To gain further insight into these dynamic relations between cell maturity, enzyme status, and PSA expression, a comparative evaluation of ST8Sia II and IV expression levels at the different stages of maturity is of course necessary and this is the subject of current investigations, along with the application of our model toward neuronal function such as cell migration and synaptic function studies.
In summary, we have demonstrated using flow cytometry and specific antibodies to different cell surface PSAs that the unnatural sialic acid precursors ManPr and, significantly, ManBu are incorporated into growing PSA chains at a minimum of 8–10 contiguous residue blocks and do not substantively interfere with PSA biosynthesis. Metabolism of these precursors spans both the murine and human PSA synthesis pathways, and we show evidence for the first time that this includes NT2-derived human neurons prior to extensive neurite production, which itself is associated with a natural PSA downregulation independent of sialic acid precursor treatment. These results are significant since the ability to control or modify cellular PSA expression can have determining influences on brain development, neural regeneration, and synaptic plasticity (Szele et al. 1994; Tang et al. 1994), and PSA expression is negatively associated with metastatic events in several malignancies (Theodosis et al. 1991; Daniel et al. 2001). Thus, the permissive nature of the sialic and PSA biosynthetic pathways can be exploited not only to help us study further their roles in events such as cellular adhesion and migration, axonogenesis, and myelination, but also to flag and then target carbohydrate-based tumor-associated antigens (Charter et al. 2000; Liu et al. 2000; Zou et al. 2004).
|
Materials and methods |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Antibodies, cell lines, and reagents
The neuraminic acid precursors ManPr and ManBu were synthesized as described (Keppler et al. 1995
). The mouse leukemic cell line (RMA-s) and the human teratocarcinoma line NTera 2/cl.D1 were generous gifts of Drs Hans Ljunggren (Karolinska Institute, Stockholm, Sweden) and Marianna Sikorska (NRC-IBS, Ottawa, Canada), respectively. The -(2-8)-PSA-specific mAb 735 was a gift of Dr R. Gerady-Schahn (Medizinishe Hochschule, Hanover, Germany). The synthesis of the fully N-propionylated or butanoylated forms of the polysaccharide (NPrPSA; NBuPSA), their HSA conjugates, as well as the NPrPSA-specific mAb 13D9 have all been described previously (Pon et al. 1997). NCAM-specific mAb OB11 was obtained from Sigma (Oakville, Canada) and goat antimouse IgG (H + L)-Pe or alkaline phosphatase from Cedarlane (Hornby, Canada). Fetal bovine serum (FBS) (Sigma), tissue culture media, and reagents were obtained from Invitrogen (Burlington, Canada), unless otherwise specified.
Mixed N-acyl polysaccharide inhibitors
Colominic acid (250 mg; Nacalai Tesque, Kyoto, Japan) was depolymerized in NaOAc buffer (0.2 M; pH 5.0) at 70°C for 30 min, prior to size fractionation on a Superdex G-75 column (Amersham Biosciences, Baie d'Urfé, Canada) equilibrated in phosphate-buffered saline (PBS). Fractions corresponding to (NeuAc)12–(NeuAc)22 were pooled on the basis of Kav comparisons obtained from a Superose-12 (Amersham) HPLC system calibrated with defined NeuAc oligosaccharides. The desalted fraction was dissolved in NaOH (2 M; 6 mL) containing 2 mg/mL NaBH4, divided into four fractions, and heated at 110°C for 0.5, 1, 1.5, and 6.5 h. The pH of each time point fraction was lowered to pH 8 with HCl prior to treating it with either 10 eq. of propionic- or butyric-anhydride (Sigma), as described (Pon et al. 1997). 1H-NMR analysis was used to characterize the mixed NPr:NAc and NBu:NAc PSA inhibitors, affording inhibitors with 0%, 35%, 65%, 80%, and 100% incorporation of either NeuPr or NeuBu within the (NeuAc)12-22 oligosaccharide framework.
Cell culture
NT2 neurons were derived from the embryonic carcinoma NTera 2/cl.D1 cell line, using different methods, and in this study are referred to as neurospheres, neurite-expressing neurospheres, and classical neurons (Pleasure et al. 1992; Horrocks et al. 2003; Sandhu et al. 2003). Single-cell suspensions from confluent NTera 2/cl.D1 were prepared and seeded into 100-mm bacterial grade Petri plates (5 x 105 cells) in complete Dulbecco's Modified Eagle's Medium (cDMEM; high glucose; 10% FBS; 2 mM glutamine; 100 U/mL penicillin G; 100 µg/mL streptomycin). Following a 2-day culture (37°C/5% CO2), RA (10 µM) was added and the cells cultured for a further 14 days with periodic media changing (cDMEM + RA). Large neurospheres (>100 µm) were collected onto nylon strainers and back-washed into fresh bacterial plates (neurospheres) or were matured by plating and further culture (3–7 days) into 100-mm tissue-culture-grade Petri plates freshly coated with poly-D-lysine (10 µg/mL; overnight absorption and desiccation) and matrigel (0.3 mg/mL) and containing mitotic inhibitors cytosine ß-D-arabinofuroside (AraC) (1 µM; first 7 days only), Uridine (10 µM), and 5-fluoro-2'deoxyuridine (FDU) (10 µM) (neurite-expressing neurospheres). Classical neurons were generated by RA treatment of NTera 2/cl.D1 cells for 4 weeks, selective panning of differentiated neurons in the presence of mitotic inhibitors (4 weeks), and maturation on matrigel as described (Horrocks et al. 2003; Sandhu et al. 2003). RMA-s leukemic cells were maintained by twice-weekly subculture with RPMI-1640 containing 10% heat-inactivated FBS, 2 mM glutamine, 100 U/mL penicillin G, and 100 µg/mL streptomycin (cRPMI).
Remodeling of cell surface PSA on RMA-s or NTera 2-derived neurons was performed essentially as before (Zou et al. 2004). Briefly, cells in either T-25 tissue culture flasks or Petri plates were treated with either pure ManPr or ManBu (10 mM) in cDMEM or cRPMI for 3–5 days (37°C/5% CO2), harvested by trypsinization or chemical dissociation using Neurocult, following the manufacturer's recommendations (Stem Cell Technologies, Vancouver, Canada), followed by further analysis.
Flow cytometry
Single-cell suspensions of neurospheres, neurite-expressing neurospheres, or classical neurons were prepared either by trypsinization (0.15% trypsin-EDTA) and trituration or by chemical dissociation, as described earlier. RMA-s or neuronal cells (1 x 106) were typically suspended in staining buffer [SB; 1% bovine serum albumin (BSA), 0.01% NaN3 in PBS], treated with a titrated amount of the appropriate mAb (0.1–4 µg; 30 min/4°C), washed (SB; 2 times), and further incubated with (R)-phycoerythrin-labeled goat antimouse IgG secondary antibody (30 min/4°C). In competitive inhibition flow experiments, graded amounts of the size-defined mixed N-acyl PSA inhibitors were coincubated with untreated, ManPr, or ManBu-treated RMA-s cells and mAbs 735 or 13D9 to verify specific binding. Data acquisition was achieved using a BD FacsCalibur flow cytometer, with analysis using CellQuest Pro software (BD BioSciences, Mississauga, Canada). Live-cell gating was achieved by excluding PI-positive materials [10% (v/v) final using a 0.1 mg/mL solution; Sigma) in combination with FS/SS strategies. Typically, 10 000 events within the gates of interest were acquired. Percentage inhibition was calculated as:
|
|
Native PSA time course study
NT2 neurospheres were prepared as mentioned earlier and a portion of these cells was transferred to a 100-mm tissue-culture-grade Petri plate freshly coated with poly-D-lysine and matrigel and then incubated at 37°C/5% CO2. After 4 days of culture, a second fraction of NT2 neurospheres was transferred to a matrix-coated 100-mm Petri dish and this procedure was repeated 2 days later. On day 7, all plates were harvested, including a sample of neurospheres not subjected to matrigel maturation. Cells were chemically dissociated, followed by staining with mAb735 for native PSA expression and analyzed by flow cytometry. Flow cytosettings were optimized for the initial neurosphere sample and maintained for all subsequent time periods.
Complement-dependent cytotoxicity
To assess functional cell surface polysaccharide epitopes, complement-dependent cytotoxicity was measured using a chromium release assay. Freshly harvested RMA-s (log-phase growth) or dissociated NT2 neurons (native and precursor modified) were pelleted by centrifugation, followed by treating each cell pellet with Na2[51Cr]O4 (100 µCi; Amersham Biosciences) for 1 h at 37°C and removal of excess label by washing (PBS; 2 times). Labeled cells (50 000) were distributed in quadruplicate into 96-well U-bottom plates and mixed with graded amounts of mAbs 735 or 13D9 and freshly thawed baby rabbit complement (5% final; Cedarlane). Spontaneous release was calculated from wells treated with complement alone and maximum release from wells treated with cetrimide (0.25% final). After a 4-h incubation (37°C), cDMEM (50 µL) was added to each well, the plates gently centrifuged (300 rpm, 3 min), and supernatants removed (50 µL) and mixed with scintillation cocktail (150 µL). Radioactivity was enumerated using a MicroBeta Trilux scintillation counter (Perkin Elmer, Woodbridge, Canada) and the percentage of specific lysis calculated according to:
|
|
ELISA assays
ELISA was employed to determine the extent of cross-reactivity with either mAb 735 or 13D9 to PSA, NPrPSA, or NBuPSA polysaccharides. Ninety-six-well high binder Nunc ELISA plates (VWR, Mont-Royal, Canada) were coated (4°C/overnight) with HSA conjugates of PSA, NPrPSA, or NBuPSA (1 µg/well/100 µL PBS). After washing with PBS–Tween 20 (PBS-T, 0.05%), wells were subsequently blocked with 1% skim milk in PBS (200 µL) for 1 h at 37°C. Serial dilutions of antibody in PBS/1% skim milk (100 µL) were added to washed wells and incubated for 60 min at room temperature. Plates were further washed (PBS-T) before the addition of a 1:2000 dilution in PBS/1% skim milk of goat antimouse IgG(H + L)-alkaline phosphatase (100 µL) and incubation at room temperature for 60 min. After washing (PBS-T), color was revealed using p-nitrophenyl phosphate substrate (100 µL; Mandel, Guelph, Canada), the reaction stopped after 10 min with the addition of 5% EDTA (100 µL), and measured at 405 nm with a 96-well plate reader (Bio-Tek EL800, Fisher Scientific, Ottawa, Canada).
Competitive inhibition ELISAs were performed essentially as described earlier, with the exception that graded amounts of polysaccharide inhibitors were preincubated (1 h; 4°C) with a standardized dilution of mAb [optical density (OD) = 0.8 against its homologous antigen-HSA conjugate] prior to their addition into PSA-, NPrPSA-, or NBuPSA-HSA-coated wells. The percentage of inhibition was calculated as:
|
|
Western blot analyses
Precursor-treated NT2 neurospheres, neurite-expressing neurospheres, and classical neurons were harvested either by extensive washing or trypsin treatment (0.15%) and pelleted by centrifugation. Cellular protein was obtained by the addition of 50 µL lysis buffer (40 mM Tris–HCl, pH 7.4; 2% Nonidet P40; 20 mM EDTA; 1 mM phenylmethylsulfonyl fluoride), vigorous mixing, and collection of protein-laden supernatant (14 000 rpm/5 min), essentially as described (Daniel et al. 2001). Protein was estimated using the bicinchonic acid microassay using BSA as standards (Fisher Scientific). Positive controls consisted of PSA or NPrPSA-HSA glycoconjugates. Samples were boiled (5 min) in reducing buffer and resolved using 7% sodium dodecyl sulfate polyacrylamide electrophoresis (SDS–PAGE). Proteins were transferred onto polyvinylidene difluoride (PVDF) membranes for 60 min (15 V), using a semidry transblot apparatus (BioRad, Mississauga, Canada). Blocked membranes (1% BSA in Tris–HCl (pH 8)/0.05% Tween 20) were then developed with mAb 735 (2.5 µg/mL), 13D9 (2.1 µg/mL), or OB11 (1:1000), followed by goat antimouse IgG-alkaline phosphatase (1:1000; Cedarlane) and revealed with alkaline phosphatase substrate (Fisher Scientific).
|
Conflict of interest statement |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
None declared.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
The authors would like to thank Ms Chantelle Cairns (ELISA), Qingling Yang (provision of mAb 13D9), and Tom Devecseri (figure preparation) for all their help. This is National Research Council of Canada publication 42514.
|
Abbreviations |
|---|
BSA, bovine serum albumin; CMP, cytidine monophosphate; CNs, classical neurons; DMEM, Dulbecco's Modified Eagle's Medium; EDTA, ethylenediaminetetraacetic acid; ELISA, enzyme-linked immunosorbent assay; FBS, fetal bovine serum; FS/SS, forward scatter/side scatter; HSA, human serum albumin; mAb, monoclonal antibody; ManAc, N-acetyl-mannosamine; ManBu, N-butanoyl-mannosamine; ManPr, N-propionyl-mannosamine; mcf, mean channel fluorescence; NAcPSA, N-acetyl-polysialic acid; NBuPSA, N-butanoyl-polysialic acid; NCAM, neural cell adhesion molecule; NENs, neurite-expressing neurons; NeuAc, N-acetyl-neuraminic acid; NeuBu, N-butanoyl-neuraminic acid; NeuPr, N-propionyl-neuraminic acid; NPrPSA, N-propionyl-polysialic acid; OD, optical density; PBS, phosphate-buffered saline; PI, propidium iodide; PSA, N-acetyl-polysialic acid; PVDF, polyvinylidene difluoride; RA, retinoic acid; SB, staining buffer; SDS–PAGE, sodium dodecyl sulfate polyacrylamide electrophoresis.
|
References |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Andrews PW. (1984) Retinoic acid induces neuronal differentiation of a cloned human embryonal carcinoma cell line in vitro. Dev Biol 103:285–293.[CrossRef][ISI][Medline]
Angata K, Suzuki M, Fukuda M. (2002) ST8Sia II and ST8Sia IV polysialyltransferases exhibit marked differences in utilizing various acceptors containing oligosialic acid and short polysialic acid. The basis for cooperative polysialylation by two enzymes. J Biol Chem 277:36808–36817.
Bork K, Reutter W, Gerardy-Schahn R, Horstkorte R. (2005) The intracellular concentration of sialic acid regulates the polysialylation of the neural cell adhesion molecule. FEBS Lett 579:5079–5083.[CrossRef][ISI][Medline]
Buttner B, Kannicht C, Reutter W, Horstkorte R. (2003) The neural cell adhesion molecule is associated with major components of the cytoskeleton. Biochem Biophys Res Commun 310:967–971.[CrossRef][ISI][Medline]
Buttner B, Kannicht C, Schmidt C, Loster K, Reutter W, Lee HY, Nohring S, Horstkorte R. (2002) Biochemical engineering of cell surface sialic acids stimulates axonal growth. J Neurosci 22:8869–8875.
Charter NW, Mahal LK, Koshland DE, Bertozzi CR. (2000) Biosynthetic incorporation of unnatural sialic acids into polysialic acid on neural cells. Glycobiology 10:1049–1056.
Charter NW, Mahal LK, Koshland DE, Bertozzi CR. (2002) Differential effects of unnatural sialic acids on the polysialylation of the neural cell adhesion molecule and neuronal behavior. J Biol Chem 277:9255–9261.
Daniel L, Durbec P, Gautherot E, Rouvier E, Rougon G, Figarella-Branger D. (2001) A nude mice model of human rhabdomyosarcoma lung metastases for evaluating the role of polysialic acids in the metastatic process. Oncogene 20:997–1004.[CrossRef][ISI][Medline]
Frosch M, Gorgen I, Boulnois GJ, Timmis KN, Bitter-Suermann D. (1985) NZB mouse system for production of monoclonal antibodies to weak bacterial antigens: isolation of an IgG antibody to the polysaccharide capsules of Escherichia coli K1 and group B meningococci. Proc Natl Acad Sci USA 82:1194–1198.
Galuska SP, Oltmann-Norden I, Geyer H, Weinhold B, Kuchelmeister K, Hildebrandt H, Gerardy-Schahn R, Geyer R, Muhlenhoff M. (2006) Polysialic acid profiles of mice expressing variant allelic combinations of the polysialyltransferases ST8SiaII and ST8SiaIV. J Biol Chem 281:31605–31615.
Horrocks GM, Lauder L, Stewart R, Przyborski S. (2003) Formation of neurospheres from human embryonal carcinoma stem cells. Biochem Biophys Res Commun 304:411–416.