plasma, the majority of which are associated with lipo- proteins Table 2 [58,59]. Gangliosides are likely to be
present on newly secreted lipoproteins although there may also be transfer of gangliosides to lipoproteins in
plasma.
The synthesis of plasma gangliosides requires the addition of sialic acid to glycolipid core structure. This
reaction is under control of hepatic sialyltransferases, a family of enzymes that sialylate glycoproteins and gan-
gliosides [60,61]. The expression of individual sialyl- transferases is unregulated in certain disease states
resulting in an overproduction of individual gan- gliosides that increases the concentration of gan-
gliosides in plasma [6].
Gangliosides are
able to
influence lipoprotein
metabolism, though the mechanisms involved are not known. Filipovic et al. [62] showed that incubation of
LDL with ganglioside inhibits its binding and uptake by smooth muscle cells. Subsequent removal of the
sialic acid from the ganglioside-enriched LDL with neuraminidase abolished the inhibitory effect on the
uptake of LDL by smooth muscle cells. Neither free sialic acid nor other negatively charged amphipathic
molecules incubated with LDL had any effect on their uptake by smooth muscle cells demonstrating a gan-
glioside-specific effect. Millar et al. [63] showed that incubation of LDL with ganglioside resulted in a de-
creased interaction between LDL and arterial proteo- glycans while incubation with asialoganglioside resulted
in a slightly increased interaction. Thomas and Poznan- sky [64] have demonstrated that the addition of gan-
gliosides
to lipid
vesicles resulted
in increased
cholesterol transfer suggesting that lipoprotein-associ- ated
gangliosides can
influence the
activity of
cholesteryl ester transfer protein CETP.
4. Sialylation of lipoproteins
The sialic acid content of plasma lipoproteins varies considerably between different lipoprotein fractions and
within lipoprotein fractions [14,63,65]. As total sialic acid content of a lipoprotein fraction is determined by
the contribution of sialic acid from each of its con- stituents, the factors that influence the sialic acid con-
tent of lipoproteins must affect one or more of the following:
1. the sialylation of apolipoproteins prior to their se- cretion into plasma,
2. the quantity of sialic acid-containing apolipo- proteins and gangliosides on lipoproteins,
3. the sialic acid content of lipoprotein-associated apolipoproteins and gangliosides following their se-
cretion in plasma either by addition, removal or modification of sialic acid.
Each of these possible mechanisms for influencing lipoprotein-associated sialic acid will be expanded upon
in the discussion that follows.
4
.
1
. The sialylation of apolipoproteins prior to their secretion into plasma
The sialylation of apolipoproteins occurs in the post- Golgi compartment where sialic acid is added to core
oligosaccharide structures. Sialylated apolipoproteins can be divided into those in which all circulating
polypeptides are sialylated apo a, B-48, B-100, D, and J and those in which sialylated and nonsialylated
polypeptides are found in plasma apo A-II, C-III, C-III, and E Table 1. The non-sialylated apolipo-
proteins in this latter group that have been studied are non-glycosylated [21,29,30]. Sialylated apolipoproteins
have a minimum of one amino acid in a covalent linkage to a sialylated oligosaccharide. There may be
subpopulations of an apolipoprotein isolate that are sialylated at only one of several potential glycosylation
sites while other subpopulations are sialylated at two or more sites [2]. Furthermore, each sialylated oligosac-
charide may potentially contain one or more sialic acid residues [2,47]. The existence of multiple potential gly-
cosylation sites on an apolipoprotein combined with the variability in the sialylation of the oligosaccharides that
can occupy these sites results in the possibility of con- siderable heterogeneity in the sialylation of many
apolipoproteins.
In general, the particular type of sialylated oligosac- charides and its site of attachment to a protein is not
random but instead is influenced by physiological con- ditions under which the protein is synthesized [32]. The
degree of sialylation of apolipoproteins is likely related to the expression of individual sialyltransferases. Sialyl-
transferase activity in various rat tissues correlates with mRNA levels of this enzyme [4]. The differing levels of
expression of sialyltransferases and other glycosyltrans- ferases among different tissues and cell lines and be-
tween species results in a great degree of variability in the core oligosaccharide and further variability in the
sialylation of this core structure [4]. This variability is evident when comparing the sialylation of apo E in
different tissues and cell lines. Apo E has been reported to contain up to six sialic acids per polypeptide [66].
The sialic acid content of apo E produced by HepG2 cells, monocytes, and astrocytes is much higher than
that found in plasma, indicating a lower degree of sialylation of the apo E synthesized by normal liver
[30,66 – 68]. In addition, the tissue-specific sialylation of proteins can be affected by factors such as acute alco-
hol consumption, inflammation, aging, and hyperc- holesterolemia [2,37,50,51].
Gangliosides are classified according to the structure of their oligosaccharide component [2]. Thus, their level
J .S
. Millar
Atherosclerosis
154 2001
1 –
13
Table 2 Plasma and lipoprotein ganglioside absolute and relative concentrations
a
GD
1a
Ganglioside concentration GM
2
+ MG
3
MG
4
GT
1b
GD
1b
GM
1
Average S.A.mol GM
3
Reference GD
3
Fraction nmolml
gangalioside 4.3–7.0
4.1–6.1 3.8–6.6
2.1–4.4 0.2–1.5
ND 1.3
[104] Plasma
55.4–66.6 4.0–8.9
17.2–21.3 ND
B 23
b
ND 3
1.3 3
[105] Plasma
B 23
b
ND 71
7.0–8.6 10.5+3.2
10–11 1
4–5 2–3
ND 1.4
[58] 46–48
24–26 Plasma
6–9 3
3 2
ND 1.2
12 [58]
18 11
VLDL 0.7
36 11
6.7 2
3 3
ND 1.2
[58] 45
13 9
LDL 2
16 3
3 4
ND 1.1
[58] HDL
2.6 34
20
a
ND, not detected; S.A., sialic acid.
b
Reported GD
1a
and GT
1b
as ‘Other gangliosides’.
of sialylation is, by definition, constant and for this reason will not be discussed further.
4
.
2
. Influence o6er the quantity of sialic acid-containing constituents on lipoproteins
The factors that control the affinity of sialic acid-con- taining constituents for lipoproteins are largely un-
known. Apo B-100 is a structural component of VLDL, LDL and Lpa, there being one apo B-100 polypeptide
per lipoprotein particle [15,69,70]. In addition, Lpa also has a single apo a polypeptide per particle [15].
Sialylated exchangeable apolipoproteins associate pri- marily with lipoproteins of specific size and lipid com-
position. Apo C-II, C-III and E are lipoprotein-bound in plasma, and associate primarily with VLDL and
HDL [71]. Apo D and J although largely found in HDL are associated with specific HDL subfractions
[25,72]. The di-sialylated form of apo C-III has a greater affinity for VLDL than the mono-sialylated or
non-sialylated forms and should result in relatively high VLDL sialic acid content in hypertriglyceridemia
[14,44].
A decrease in the concentration of B:C-III and B:E lipoproteins in plasma was reported for subjects under-
going treatment with the fibric acid derivative fenofi- brate [73]. These changes should result in decreases in
the sialic acid content of the apo B-containing lipo- protein fractions due to a lower content of sialylated
apolipoproteins apo C-III and E. Millar et al. [63] showed that the sialic acid content of VLDL, IDL and
LDL decreased following ciprofibrate treatment al- though apo C-III and E levels were not measured in
this study. Other treatments or interventions that result in changes in the content of sialylated apolipoproteins
within a lipoprotein fraction would be expected to be accompanied by corresponding changes in the sialic
acid content of that lipoprotein fraction.
Gangliosides are largely associated with lipoproteins in plasma with a small proportion bound to albumin
[58,59]. There are no apparent differences in the pro- portion of individual gangliosides found on VLDL and
LDL but there are differences between LDL and HDL [58]. This latter observation would suggest that ex-
change of gangliosides between lipoproteins, if it oc- curs, is carrier-mediated. In vitro data suggest that
ganglioside exchange between lipoproteins is both car- rier-mediated and spontaneous [74]. Therefore other
factors, possibly lipid composition, may affect any spontaneous transfer of gangliosides between lipo-
proteins. Overproduction of ganglioside GD
3
has been described in malignant melanoma resulting elevated
ganglioside levels in plasma, though it is not known how this is reflected in plasma lipoproteins [6].
4
.
3
. Addition, remo6al, or modification of apolipoprotein and ganglioside sialic acid following
secretion into plasma Native lipoproteins and apolipoproteins are com-
monly referred to as being desialylated, implying that they were secreted as sialylated entities and that sialic
acid was removed, either enzymatically or chemically, in plasma Fig. 3D [34,75,76]. This concept gained
favor in the lipoprotein field when it was noted that certain apolipoproteins secreted from cultured cells had
a higher sialic acid content than the same apolipo- proteins isolated from plasma [66]. The assumption was
made that the difference in sialic acid content between apolipoproteins secreted in vitro and those found in
vivo was due to loss of sialic acid in plasma. It is now known that the reason for this discrepancy, as de-
scribed in Section 4.1, is that there is differential expres- sion of sialyltransferases in liver and cultured cells. This
differential expression can result in variability in the sialylation of the same apolipoprotein secreted from
these two sources [3].
There is little evidence to support the action of desialylation in normal plasma. Neuraminidase, a lyso-
somal enzyme that cleaves sialic acid from oligosaccha- ride chains, is detectable in plasma but has a pH
optimum between 4 and 5 [77]. Hanson et al. [78] have detected increased plasma neuraminidase levels follow-
ing acute myocardial infarction supporting the concept that the enzyme has entered plasma following tissue
damage. Several studies have been unable to demon- strate desialylation of sialylated plasma proteins in
vivo, the exception being experimental peritonitis where there are high plasma levels of bacterial-derived neu-
raminidase [48,79,80].
Further evidence against the concept of desialylation occurring in plasma relates to the lack of glycosylation
of some proteins that have been called ‘desialylated’. If desialylation were to occur in plasma then the desialy-
lated fraction would remain glycosylated with the core oligosaccharide chain rather than being non-glycosy-
lated Fig. 3. Most proteins that have been desialylated in vitro and then reinjected into plasma are cleared
rapidly by the hepatic asialoglycoprotein receptor. This may explain the absence of desialylated proteins in
plasma. However, apo B-100 that has been desialylated in vitro and reinjected appears to be an exception,
having a normal, or near-normal clearance from plasma similar to what has been observed for desialy-
lated transferrin, a protein with similar complex-type oligosaccharides [81 – 84]. Despite this, no fully desialy-
lated oligosaccharide structures have been detected on LDL [20,34]. Bartlett and Stanley [34] have reported
that all LDL contain at least one mono-sialylated com- plex-type oligosacccharide and have interpreted this as
being a desialylated product. An alternative explana-
tion is that these are normal structures that are secreted on apo B-100 in addition to di-sialylated complex-type
oligosacccharides. Several studies report a decrease in the sialic acid
content of LDL following LDL oxidation [85 – 87]. There is no accompanying increase in free sialic acid
suggesting that sialic acid is oxidized to a product that is not detectable using conventional assays for sialic
acid [86,87]. Van Lenten and Ashwell [88] showed that oxidized sialic acid has a molar extinction coefficient
that is 45 lower than unmodified sialic acid when measured using the Warren method [89]. Thus, oxidized
sialic acid is underestimated using this method of mea- surement. The resorcinol method of Svennerholm is less
sensitive to this modification, oxidized sialic acid having a molar extinction coefficient that is 10 greater than
unmodified sialic acid [90]. It is not clear if product of sialic acid oxidation remains bound to the oligosaccha-
ride chain of its parent glycolipid or apolipoprotein [86,87]. Tertov et al. [86] showed that oxidatively
modified sialic acid is not found in the free form i.e. remains apolipoprotein- or glycolipid-bound while re-
sults from Tanaka et al. [87] were inconclusive.
Tertov et al. [86] have raised the possibility of trans- fer of sialic acid from lipoproteins to other plasma
constituents. Although Trypanosoma cruzi has been re- ported to express a trans-sialidase that transfers
protein-bound donor sialic acid to an acceptor glyco- protein, there is no evidence of this occurring in hu-
mans [2]. While sialyltransferases are demonstrable in plasma these are, as with neuraminidase, likely intracel-
lular enzymes that serve no function in plasma but have entered plasma in response to tissue injury [91]. Fur-
thermore, human sialyltransferases do not transfer protein-bound sialic acid to an acceptor but instead
transfer nucleotide-activated sialic acid CMP-neu- raminic acid, which is not found in plasma, to galacto-
sylated acceptors [61].
5. VLDL sialic acid