main

Since January 2020 Elsevier has created a COVID-19 resource centre with
free information in English and Mandarin on the novel coronavirus COVID-

19. The COVID-19 resource centre is hosted on Elsevier Connect, the
company's public news and information website.
Elsevier hereby grants permission to make all its COVID-19-related

research that is available on the COVID-19 resource centre - including this
research content - immediately available in PubMed Central and other
publicly funded repositories, such as the WHO COVID database with rights
for unrestricted research re-use and analyses in any form or by any means
with acknowledgement of the original source. These permissions are
granted for free by Elsevier for as long as the COVID-19 resource centre
remains active.
INTERNATIONALREVIEW OFCYTOLOGY, VOL. 126
Glycosylation in Intestinal Epithelium

DOUGLAS AND JURGEN Row?
J. TAATJES*
Interdepartmental Electron Microscopy, Biocenter, University of Basel, CH-4056 Basel,
Switzerland
I. Introduction
A great deal of interest has been focused in the last few decades on elucidat-
ing the mechanisms and cellular locations of glycosylation reactions. This in-
terest has arisen as the result of many divergent disciplines converging on a
central, fundamental question: Where and how are sugar residues added on to
proteins and lipids during their journey from the sites of synthesis to delivery
at final destinations? This question is of practical concern to biochemists, cell
biologists, virologists, and cancer biologists, to name a few, and has emerged
from the realization that the sugar moieties of glycoconjugates are involved in
a myriad of events related to recognition phenomena. This is especially true of
oligosaccharide chains present on plasma membrane glycoconjugates. The first
version of the fluid mosaic model of the plasma membrane published in 1972
(Singer and Nicolson, 1972) depicting a lipid bilayer containing integral and
peripheral proteins, neglected the existence of the sugar constituents of these
molecules. Indeed, almost all plasma membrane proteins are glycoproteins.
Given the knowledge that the sugar moieties of glycoconjugates are found on
the extracellular side of plasma membranes, it is quite easy to envision their in-
volvement in recognition of another cell, bacterium, or virus (Rademacher et
a!., 1988). This idea took on added significance with reports from the 1960s
that cancer cell plasma membranes were different from those of normal cells
with respect to their oligosaccharide composition (Aub et al., 1963, 1965a,b;
Burger and Goldberg, 1967). Although time and much subsequent research
have revealed the oversimplification of this notion of malignancy and metasta-
sis, it nevertheless certainly generated a flurry of activity aimed at understand-
ing glycosylation. Perhaps of equal importance was the finding that many
pathogenic organisms bind to and gain entry to cells through recognition of
sugar molecules present at the cell surface (Paulson, 1985; Kocourek, 1986;
Sharon and Lis, 1989). Regardless of the driving forces, the result has been an
understanding in detail of some aspects of the process of glycosylation.
*Present Address: Department of Pathology, University of Vermont, Burlington, Vermont 05405

'Present Address: Department of Cell and Molecular Pathology, Institute of Pathology,University of
Zurich, CH-8091 Zurich, Switzerland
135
Copyright 0 1991 hy Academic Press.Inc.
All rights of reproduction in any form reserved.
136 DOUGLAS J. TAATES AND JURGEN ROTH
GLYCOSYLATION IN INTESTINAL EPITHELIUM 137
The intestinal epithelium represents a good model system in which to study

the glycosylation process. From a simplistic viewpoint, the intestinal epithelium
is composed of two basic epithelial cell types; the absorptive enterocyte and the
mucus-producing goblet cell (Figs. 1 and 2). These two cell types differ both in
form and function, yet are derived from a singular precursor cell (Leblond and
Messier, 1958; Cheng and Leblond, 1974). This affords the opportunity to study
glycosylation in neighboring but distinct cells. From a historical perspective, the
intestine was chosen as one of the earliest systems for the cytochemical investi-
gation of glycosylation. Indeed, in two classic papers investigating intestinal ep-
ithelial cells, Neutra and Leblond (1966a,b) provided the first demonstration
that complex sugars are added to glycoconjugates in the Golgi apparatus. They
injected tritiated hexoses (['H]glucose and [3H]galactose)into rats, and found
that autoradiographic grains were first localized over the Golgi apparatus cister-
nal stack. Radioactivity was then followed into the mucus droplets of the goblet
cell and into the plasma membrane of absorptive cells. Further, classical histo-
chemical staining techniques, applied to electron microscopically visualize peri-
odate-reactive carbohydrates (Rambourg er al., 1969), demonstrated reaction in
the Golgi apparatus and plasma membrane of intestinal epithelial cells, as previ-
ously reported for acid mucopolysaccharides using colloidal iron staining
(Revel, 1964; Wetzel er al., 1966; Berlin, 1967).
Likewise, .the basic morphology and physiology of the intestinal tract provide
an ideal system in which to study glycosylation as it relates to development and
differentiation (Fig. 3). The harsh environment of the intestinal lumen results in
continuous loss of epithelial cells through sloughing into the lumen.
Gastrointestinal epithelial renewal ensues through the processes of cell prolifer-
ation, migration, and differentiation (Eastwood, 1977). This renewal occurs in
discrete proliferative zones along the gastrointestinal tract. In the small intestine,
this proliferative zone is restricted to the base of the crypts, whereas in the large
intestine it is less restrictive, occurring in the basal two thirds of the crypt. The
definitive autoradiographic studies of Leblond and co-workers have established
that in both small (Cheng, 1974; Cheng and Leblond, 1974) and large intes-
tine (Chang and Leblond, 1971), a ring of undifferentiated stem cells situated in
the crypts divide to produce daughter cells committed to differentiate into
FIG. I . Low-power electron micrograph showing the surface epithelium from the chick duode-
num. In this and all subsequent micrographs, the tissue was lightly fixed in aldehydes (without post-
fixation in osmium tetroxide), followed by low temperature dehydration and embedding in Lowicryl
K4M. Thus, membrane delineation is different from that seen in routinely processed tissues. The
surface lining the intestinal lumen is composed mainly of two epithelial cell types: the mucus-pro-
ducing goblet cell interspersed among the absorptive enterocyte. Both cell types possess an elaborate
apical plasma membrane called the brush border (arrowheads), which is covered by a layer of mucus
rich in glycoconjugates. (md) Mucus droplets; ( n l ) goblet cell nucleus; (n2) absorptive cell nucleus;
(n3) nucleus of migrating lymphoid cell. X 1400. Bar = 6.7 l m .
138 DOUGLAS J. TAATJES AND JURGEN ROTH
B r u s h Border
. Multivesicular
Body
Apical Cytoplasmic
Mucus Droplets Vesicles
, Lysosome
Golgi Apparatus
Golgi Apparatus
W
FIG.2. Schematic drawing illustrating the morphology of the two basic epithelial cell types lin-
ing the intestinal tract. Compare this drawing with the electron micrograph presented in Fig. 1. Of
particular imponance with respect to glycosylation are the Golgi apparatus, apical and basolateral
plasma membrane domains, absorptive cell apical cytoplasmic vesicles, and goblet cell mucus
droplets.
absorptive (columnar), goblet, and endocrine cells. The mechanisms guiding

proliferation and migration are not fully understood (Potten and Loeffler, 1987).
however, migration time from crypt to surface epithelium is known to take 2-3
days in the rodent. Thus, in a longitudinal section along the crypt-to-surface
axis, cells in various degrees of differentiation may be observed, providing a
unique in viw system in which to investigate differentiation-related glycosyla-
tion events.
In this review we will attempt to assimilate the data available from biochem-
ical and cytochemical studies to present a picture of glycosylation reactions in
the intestinal epithelium. Very few generalizations will be attempted given the
variability of the oligosaccharide structures of glycans in different segments of

the gastrointestinal tract, as well as between animal species. Moreover, the very
fascinating discipline of cancer-related changes in intestinal glycosylation will
not be presented here (Boland and Kim, 1984; Bresalier et al., 1985; Kim,
1989a,b). A review by Weiser and co-workers, focusing on intestinal cell mem-
branes (but also discussing glycosylation) has been published in this series
(Weiser et al., 1986).
11. Overview of Glycosylation

A brief overview of what is known concerning glycosylation reactions is in
order for the understanding of material presented in the rest of this paper. It is
not in the realm of this review to present in great detail a treatise on glycosyla-
tion reactions in general. Interested readers may consult several excellent papers
on this subject published in the last decade (Hanover and Lennarz, 1981;
Hubbard and Ivatt, 1981; Kornfeld and Kornfeld, 1985; Hirschberg and Snider,
1987; Roth, 1987a). Most of the work investigating glycosylation has focused
on the glycosylation of proteins; not as much is known concerning the glycosy-
lation of lipids. For this reason, the vast majority of this article will focus on gly-
cosylation of intestinal proteins, with only infrequent reference to glycolipids.
Glycoproteins can be conveniently divided into two main classes based on the
nature of the covalent linkage between the amino acid in a peptide and the sugar
residue of an oligosaccharide. One class of glycoproteins are characterized by
oligosaccharide chains linked N-glycosidically from the sugar N-acetylglu-
cosamine to the arnide nitrogen of asparagine in the peptide (Fig. 4). A certain re-
striction seems to be placed on the asparagine residue to be glycosylated; it must
be part of the sequence -Asn-X-Sermr (where X can be any amino acid except
proline or aspartic acid). Interestingly, even when asparagine is found in the cor-
rect sequence it is not always glycosylated, suggesting that accessibility of the se-
quence may also be a factor. The asparagine-linked (N-linked) oligosaccharides
can further be divided into three subclasses depending upon the extent of trim-
ming and elongation reactions: (1) high mannose-type oligosaccharides which
contain only the sugars mannose and N-acetylglucosamine (Fig. 4C); (2) com-
plex-type oligosaccharides which contain the sugars galactose, fucose, and sialic
acid in addition to mannose and N-acetylglucosamine (Fig. 4B); and (3) hybrid-
type oligosaccharides which contain both high mannose- and complex-type units
(Fig. 4D). Although the three subclasses of N-linked oligosaccharides display a
wide variety of structural conformations, they nevertheless all share the common
core structure Manal-3(Manal-6)Manp1-4GlcNAc pl-4GlcNAc-Asn'.
'Abbreviations: Fuc, fucose; Gal, galactose; GalNAc, N-acetylgalactosamine; Glc, glucose;

GlcNAc, N-acetylglucosamine; Man, mannose; NeuSAc, neuraminic (sialic) acid; GDP-Man,
guanosine 5'-diphosphate mannose; UDP-Gal, uridine 5 '-diphosphate galactose; UDP-GlcNAc, uri-
dine 5 '-diphosphate N-acetylglucosamine.
GLYCOSYLATION IN INTESTINALEPITHELIUM 141
The second major class of glycoproteins are the so-called 0- or mucin-type

glycoproteins and comprise, among others, oligosaccharides attached via an 0-
glycosidic linkage from the sugar N-acetylgalactosamine to the hydroxyl group
of the amino acids serine or threonine in the peptide (Fig. 5).
A third, and novel type of glycoprotein has recently been described in which
N-acetylglucosamine is 0-glycosidically attached to the peptide (Torres and
Hart, 1984). This appears to be a most interesting type of glycoprotein since its
synthesis most likely occurs outside of the normal cellular glycosylation path-
way. Detailing this type of glycoprotein is beyond the scope of this review; how-
ever, this subject has recently been discussed by Hart and co-workers (Hart et
al., 1988, 1989).
A. ASSEMBLY LINKEDOLIGOSACCHARIDES
OF N-GLYCOSIDICALLY
The mechanisms of N-linked glycosylation reactions have been worked out in

great detail (Komfeld and Komfeld, 1985; Roth, 1987a) and will only be briefly
reviewed here. The classical reactions themselves are restricted to two intracel-
lular compartments, the rough endoplasmic reticulum and the Golgi apparatus.
The assembly of N-glycosidically linked oligosaccharide chains follows a
unique pathway involving a lipid-linked intermediate. The lipid employed is
dolichol phosphate, to which sugars are added in a stepwise manner, with the
first seven sugars (two N-acetylglucosamine and five mannose residues) derived
from the nucleotide sugars UDP-GlcNAc and GDP-Man, while the next seven
sugars are donated from the lipid intermediates Dolichol-P-Man and Dolichol-
P-Glc (Hirschberg and Snider, 1987). The resulting lipid-linked oligosaccharide
consists of the structure Glc,Man,GlcNAc,-P-P-Dolichol (Fig. 4A), and is
found in the lumen of the rough endoplasmic reticulum. Glycosylation of a
growing peptide chain occurs via the en bloc transfer of the oligosaccharide
chain from the lipid to an asparagine residue as the peptide is sequestered into
the lumen of the rough endoplasmic reticulum. Processing of the oligosaccha-
ride chain seems to begin immediately after its transfer to the peptide. The
FIG.3. Light micrographs of semithin sections (1 pm) from rat proximal colon illustrating the
basic organization of the intestinal mucosa. (a) The mucosa of the large intestine is composed of the
surface epithelium lining the intestinal lumen, and the goblet cell-rich crypt region. Beneath the
basement membrane of the surface epithelial cells is the lamina propria (Ip). (b) Higher magnifica-
tion of the boxed region from (a), demonstrating the organization of the surface epithelium. The
section had been incubated with the Lirnaxflavus lectin (specific for sialic acid residues), followed
by fetuin-gold complex and silver amplification. All structures containing sialic acid therefore
appear black in these photos. Note the intense staining of goblet cell mucus (md), brush border
(arrowheads), Golgi apparatus (arrows), and cells in the lamina propria. In contrast, note the lack of
nuclear staining (n). A goblet cell in the process of releasing mucus is denoted by an asterisk. x160
(a); x725 (b). Bars = 100 pm (a) and 13 pm (b). (b) (Reproduced with permission from Taatjes and
Roth, 1988.)
142 DOUGLAS J. TAATJES AND fiRGEN ROTH
A
I
A
I @ @
A A A A
I I I I
0 e e 0 0 0
I I I I I I
m m m
1 I
a e\Jo
‘e’
I
I
a-0
I
-hm-
0
I
D O e e
I \/ \/
0
‘e’ ‘ O L .
I I
rn rn
I I
.-a rn
I I
-As~I- -AstI-
FIG.4. Schematic representation of typical N-glycosidically linked oligosaccharides. (A) lipid-
linked oligosaccharide precursor; (B) complex-type oligosaccharide; (C) high mannose-type
oligosaccharide: and (D)
hybrid-type oligosaccharide. See Section I1 for details.
extent of processing is dictated by whether the oligosaccharide chain on the gly-

coprotein is destined to become a high mannose- or complex-type. In the case of
the high mannose-type, processing merely involves the removal of the three glu-
cose residues, as well as perhaps a few mannose residues. In contrast, the pro-
I A Sialic Acid
-Serr/Thr -
FIG.5. Schematic representation of typical O-glycosidically linked oligosaccharide. See Section
I1 for details.
cessing of complex-type oligosaccharides is more extensive and entails the re-

moval of the three glucose residues and all but three of the mannose residues.
These trimming events are carried out by the enzymes glucosidase I, glucosidase
11, and ER mannosidases (Kornfeld and Kornfeld, 1985; Moremen and Touster,
1988). Following these trimming reactions, the glycoprotein is transported via
vesicles to the Golgi apparatus where further mannose residues are removed
(Moremen and Touster, 1988). Complex-type oligosaccharides are then com-
pleted in the Golgi apparatus by the addition of N-acetylglucosamine and the
terminal sugars galactose, fucose, and sialic acid.
B. ASSEMBLYOF O-GLYCOSIDICALLY
LINKED
OLIGOSACCHARIDES
In contrast to the abundant information available concerning the synthesis of
asparagine-linked oligosaccharides, less is known regarding the steps involved
in the synthesis of O-glycosidically linked oligosaccharides. The synthesis of
the oligosaccharide chain seems to be a late posttranslational event, appears to
occur by a sequence of classical glycosyl transfer reactions, and does not in-
volve an oligosaccharide-lipid intermediate (Hanover and Lennarz, 1981). The
initial event consists of the transfer of N-acetylgalactosamine from UDP-
GalNAc to serine or threonine in a polypeptide. There seems to be no require-
ment for a specific amino acid sequence surrounding the serine or threonine, as
is necessary for the asparagine in N-glycosidically linked oligosaccharides.
However, 0-glycosyl residues are often clustered in regions rich in the amino
acid proline, suggesting that a possible signal required for O-glycosylation is
contained in the secondary or tertiary structure of the polypeptide chain
(Hanover and Lennarz, 1981).
The precise cellular location of the onset of O-glycosylation has been diffi-
cult to ascertain. Depending upon the methodology employed, various investi-
gators have implicated the rough endoplasmic reticulum (Strous, 1979) or the
Golgi apparatus (Hagopian ef al., 1968; Kim et al., 1971; Hanover et af., 1980;
FIG.6. Demonstration of Helixpomario lectin-gold binding sites in sections from chick duode-
num. Gold particle label is detectable in goblet cell mucus (md) and Golgi apparatus (arrows). but
not in the rough endoplasmic reticulum (rer). At higher magnification (b), label in the Golgi appara-
tus is seen to be restricted to cis and trans (arrowheads) regions. X 5,500 (a); X 18.000 (b). Bars =
1.8 pm (a) and 0.6 pm (b). (Reproduced from the J. Cell B i d . . 1984,98, 399-406 by permission o f
the Rockefeller University Press.)
Roth, 1984; Deschuyteneer et al., 1988; Tooze et al., 1988) as the site of addi-
tion of N-acetylgalactosamine to the peptide chain. These conflicting results
are most probably due to the different methodologies employed, or to cell type
variability in glycosylation reactions (Section IV,B,2). The majority of the evi-
dence, though, points to the Golgi apparatus (or a pre-Golgi apparatus compart-
ment; see below) as the site of initiation of 0-glycosylation. Elhammer and
Kornfeld ( 1984) investigated the subcellular distribution of the initiating
polypeptide : N-acetylgalactosaminyltransferase as well as the subsequently
acting UDP-Gal : GalNAc-P- 1,3 galactosyltransferase by fractionating the total
microsomal membranes of mouse lymphoma BW5147 cells on linear sucrose
gradients. They found that the two transferases were present in membranes of
different densities. The galactosyltransferase codistributed with the “classical”
galactosyltransferase involved in asparagine-linked glycosylation and was
shown to reside in trans-Golgi apparatus cisternae of HeLa cells (Roth and
Berger, 1982), whereas the polypeptide : N-acetylgalactosaminyltransferase
distributed in a fraction intermediate between those containing galactosyltrans-
ferase activity and glucosidases I and 11, the latter indicative of the endoplasmic
reticulum (Lucocq et al., 1986). These results were interpreted to suggest that
the polypeptide : N-acetylgalactosaminyltransferase most likely was derived
from membranes representative of the cis-Golgi apparatus, and was contained
in a separate Golgi compartment from the galactosyltransferase.
In a carefully controlled complimentary biochemical approach, Abeijon and
Hirschberg (1987) reported that in rat liver the polypeptide : N-acetylgalac-
tosaminyltransferase activity (using apomucin as an exogenous acceptor) was
highly enriched in membranes derived from the Golgi apparatus compared to
those derived from the rough and smooth endoplasmic reticulum. Moreover,
they found that vesicles prepared from the Golgi apparatus were able to translo-
cate UDP-GalNAc into their lumen in an in vitro assay, at rates 4-6-fold higher
than those from rough and smooth endoplasmic reticulum. These results demon-
strated that at least in rat liver, all the cellular machinery necessary for the initi-
ation of 0-glycosylation was located within the Golgi apparatus.
An independent cytochemical investigation has also implicated the cis-Golgi
apparatus as the site for the onset of 0-glycosylation. Roth (1984) used a Helix
pomatia lectin-gold complex in a postembedding study as a probe for terminal
nonreducing N-acetylgalactosamineresidues. In goblet cells of chick and rat in-
testine he found label in the cis and trans regions of the Golgi apparatus, but not
in the endoplasmic reticulum (Figs. 6 and 7). He interpreted these findings to in-
dicate that the initial transfer of N-acetylgalactosamine to the peptide occurs in
the cis region of the Golgi apparatus, while the staining present in the trans re-
gion of the Golgi apparatus was representative of terminal N-acetylgalac-
tosamine residues added on to the oligosaccharide chain. These results were cor-
roborated and expanded upon by Deschuyteneer er al., (1988). They used
FIG.7 . Demonstration of H . pomuria lectin-gold binding sites in ultrathin section from rat colon.
Gold particle label is found over cis-cistemae and dilated trans-cistemae (arrowheads) of goblet cell
Golgi apparatus. as well as in mucus droplets (md). Note lack of staining over cisternae of rough en-
doplasmic reticulum (rer). X 11,500. Bar = 0.9 pm. (Adapted from the J. Cell B i d . , 1984, 98,
399-406 by permission of the Rockefeller University Press.)
immuno- and cytochemical means to investigate the site of addition of N-acetyl-

galactosamine to the peptide chain of porcine submaxillary gland mucin.
Employing an antibody which recognized only the deglycosylated form of
mucin (apomucin). they found that immunoreactivity was localized throughout
the rough endoplasmic reticulum (including the nuclear envelope) of mucous
cells. In contrast, staining with a H. pomaria lectin-gold complex was found in
the Golgi apparatus and mucus droplets, but not in the rough endoplasmic retic-
ulum, confirming the results obtained by Roth (1984) for intestinal goblet cells.
Interestingly, by treating thin sections with a cocktail of glycosidases,
Deschuyteneer et al., (1988) were able to detect immunoreactivity with the an-
tibody raised against apomucin in the cis-cistemae of the mucous cell Golgi ap-
paratus. These results provided very strong further support for the contention
that the onset of 0-glycosylation is a posttranslational event occurring most
likely in cis-cistemae of the Golgi apparatus.
Very recently, a third compartment has been implicated as the site of the
onset of 0-glycosylation. Tooze and co-workers ( 1988) investigated steps in
the 0-glycosylation of the El glycoprotein of coronavirus MHV-A59 using a

combination of pulse-labeling and morphological techniques. They found that
N-acetylgalactosamine was added to the E l protein about 10 min after synthe-
sis, followed about 10 min later by the addition of galactose and sialic acid,
producing the mature oligosaccharide. They then took advantage of tempera-
ture-sensitive steps in the exocytic transport system of virus release (Saraste
and Kuismanen, 1984) to block the movement of the glycosylated El protein
out of different compartments. Interestingly, they were able to identify a
smooth-membraned compartment situated between the endoplasmic reticulum
and the cis-Golgi apparatus where the El glycoprotein was found to already
possess N-acetylgalactosamine, but not galactose or sialic acid. They inter-
preted these results to indicate that the site of addition of the core N-acetyl-
galactosamine is this smooth membrane compartment, which most likely repre-
sents a budding compartment situated between transitional elements of the
rough endoplasmic reticulum and the cis side of the Golgi apparatus. However,
it should be borne in mind that these results were obtained from virally infected
cells, which may or may not behave as normal cells with respect to glycosyla-
tion reactions. Nevertheless, in light of these findings, the results of Elhammer
and Komfeld (1984) discussed above could also be interpreted to suggest that
the polypeptide : N-acetylgalactosaminyltransferase is housed within this pre-
Golgi apparatus budding compartment, rather than within cis-Golgi apparatus
cisternae. It seems likely, though, that the definitive answer defining where the
onset of 0-linked glycosylation occurs awaits the immunocytochemical local-
ization of the polypeptide : N-acetylgalactosaminyltransferase. Regardless,
subsequent 0-linked glycosylation reactions are known to occur in the Golgi
apparatus.
111. Methods Employed to Investigate Cellular Glycosylation

Reactions in Intestine
Investigations into intestinal glycosylation have mainly drawn upon biochem-

ical and morphological techniques for the demonstration of enzymes (glycosyl-
transferases) involved in glycosylation reactions, as well as the carbohydrate
products resulting from the enzymatic action. These techniques are not specific
to the study of intestinal tissues, but rather have been used to study glycosyla-
tion in general. The unique application of these methods to intestinal tissues
stems mainly from the following three areas: (1) separating activities from crypt
and villus portions of the small intestine; (2) separating activities from Golgi ap-
paratus membranes and plasma membranes; and (3) investigating glycosylation
changes in different intestinal regions.
A. BIOCHEMICAL
METHODS
Two main biochemical tacks have been followed in the investigation of in-
testinal glycosylation. The first was to assay for the activity of specific glycosyl-
transferases, followed by subsequent purification. These assays were usually
employed with the goal of measuring and comparing activities of various glyco-
syltransferases in the different intestinal segments, as well as crypt versus villus.
Moreover, such methods were also extensively used to compare changes in dif-
ferent glycosyltransferase activities during development (Section IV,A,2). The
main drawbacks of this method were the possible contamination of cellular frag-
ments which could lead to erroneous conclusions. as well as the inability to
identify which cells in the mixture actually contained the glycosyltransferase of
interest (Section IV,A, I ) . The second most popular method has been to measure
the incorporation of radiolabeled sugars into glycoproteins and glycolipids and
following their transport to final destinations (Section IV,A,3).
B. MORPHOLOGICAL
METHODS
As noted in the Introduction, classical morphological and cytochemical meth-
ods have been extensively applied to studies of intestinal glycosylation. These
studies will not be mentioned again here so that we may focus on more recent
types of cytochemical investigations. However, it should be borne in mind that
although we now have available more sensitive and specific probes, as well as
refined cytochemical techniques, the results obtained have dramatically sup-
ported the original conclusions drawn from the earlier studies.
Investigations employing lectins have proven to be of particular importance
for the study of intestinal glycosylation (Etzler and Branstrator, 1974). Lectins
are carbohydrate-binding proteins of nonimmune origin that agglutinate cells
and precipitate glycoconjugates (Goldstein and Poretz, 1986). When coupled to
an appropriate marker (Fig. 8) they can be used to demonstrate the cellular oc-
currence and distribution of specific sugar residues (Roth, 1978, 1987b). In in-
testinal studies they have been used to investigate Golgi apparatus glycosylation
events, to determine the pattern of sugar residue expression at the plasma mem-
brane of epithelial cells along the intestinal tract, and to monitor changes in the
glycosylation pattern of epithelial cells during development and neoplastic
transformation. For convenience, a list of lectins (with their saccharide specifici-
ties) mentioned in this article is presented in Table I.
A major advance for the field of glycosylation investigation in general (how-
ever, also applicable to intestinal studies; Sections IV,B,2; IV,C) was the devel-
opment of sensitive, yet routine methods for electron microscopic immunocyto-
chemistry. The first was the introduction of colloidal gold particles as an
electron dense marker for immunocytochemistry (Faulk and Taylor, 1971). This
GLYCOSYLATIONIN INTESTINALEPITHELIUM 149
GLYCOPROTEIN - G O L D
\
LECTIN - GOLD
"\
LECTIN
GLYCOPROTEINS
FIG.8. Schematic representation of lectin-gold techniques for the detection of sugar residues on
thin sections from tissues embedded in Lowicryl K4M. Lectins can be applied directly complexed
with particles of colloidal gold (left side of figure), or in a two-step cytochemical affinity technique
employing an unlabeled lectin followed by a glycoprotein-gold complex (right side of figure). A de-
tailed description of these procedures can be found in Roth er al. (1988a) and Roth (1989).
was followed by the introduction of the protein A-gold technique (Roth et al.,
1978) for the postembedding localization of antigenic sites (Fig. 9). The virtues
of this method have been detailed in many reviews (Roth, 1983a, 1986, 1989;
Bendayan, 1984) and will not be enumerated here. It suffices to say that the pro-
tein A-gold technique in conjunction with the introduction of the low tempera-
ture embedding methods employing Lowicryl K4M (Carlemalm et al., 1982)
provided the means necessary for the precise immunolocalization of the rela-
tively scarce glycosyltransferases. Of course, of equal importance was the pro-
duction and availability of highly specific anti-glycosyltransferase antibodies
(Roth and Berger, 1982; Weinstein et al., 1982; Shaper et al., 1985; Ulrich et al.,
1986). The combination of these innovations provided immediate dividends
PROTEIN A - GOLD
J \
6
, L ; N , T I G E N : h n ANTIBODIES n,
\A
TISSUE SECTION
FIG.9. Schematic representation of the protein A-gold technique for antigen localization on thin
sections from tissues embedded in Lowicryl K4M. This is a typical two-step technique in which sec-
tions are first incubated with an antibody, followed by protein A-gold complex which binds to the
Fc portion of the antibody. As illustrated on the right side of the figure, multiple antigens can be de-
tected on the same section by employing colloidal gold particles of different sizes. Details of this
technique can be found in Roth et a/. (1978) and Roth (1983a. 1989).
TABLE I
SACCHARIDE-BINDING
SPECIFICITES
OF VARIOUSLECTINS
Lectin Common name Abbreviation Nominal specificity"

Canar,alioensfformis Jack bean Con A aMan > aGIc > GlcNAc
Lens c~ilinaris Lentil LCL aMan > aGlc > GlcNAc
Pisum sarivrtm Pea PSL aMan > aGlc = GlcNAc
Triticum wlgure Wheat germ WGA GlcNAc ( ~ I P G l c N A c ) ,>~ ,
PGlcNAc > NeuSAc
Dolichos hiflorus Horse gram DBL GalNAc al.3GalNAc >>
aGalNAc
He1i.r pomatia Edible snail HPL GalNAc al.3GalNAc >
aGalNAc
G l w i n e ma.\ soybean SBL aGalNAc = PGalNAc
Griffonia - GSL I aGal>> aGalNAc
simpiicifolia I-B,
.Arachis hypogaea Peanut PNL CalPl,3GalNAc > a and PGal
Ricinus communis I Castor bean RCL I pGal> aGal>> GalNAc
Ricinus rommunis II RCL I I P and a G a l > GalNAc
Durura stramonium Thorn apple DSL GalPl,4GlcNAc = GlcNAc
(PI,4GlcNAc),.,
Lorus terragonolobus Asparagus pea LTL a-L-Fuc
U1e.r europaeus Gorse seed UEL I a-L-Fuc
Lima.\ flaws Slug LFL aNeuSAc > aNeuSGc
Samhucus nigra L. Elderberry SNL I NeuSAc a2,6Gal/GalNAc
Maackia amurensis - MAL NeuSAc a2,3Cal
uAbbreviations: Man, mannose; Glc, glucose; GlcNAc, N-acetylglucosamine; GalNAc, N-acetyl-
galactosamine; Gal, galactose; Fuc, fucose; NeuSAc, sialic acid.
with the publication of the first immunocytochemical localization of a glycosyl-

transferase at the electron microscopic level (Roth and Berger, 1982). This paper
provided a key piece of evidence which led to the subcompartmentation model
of the Golgi apparatus, which will be discussed in detail in Section IV,B,lb. At
the same time, the opportunity had now arisen to explore the subcellular distri-
bution of glycosyltransferases at the electron microscopic level in intestinal ep-
ithelial cells.
IV. Distribution of Intestinal Glycosyltransferases and

Their Saccharide Products
A. STUDIES ON WHOLE TISSUE
1 . Measurement of Glyco.$yltransferase Activities in Adult Animals

As mentioned earlier, many of the studies concerning glycosyltransferases in
intestinal cells sought to compare activities in the crypt with those in the villus,
or to compare glycosyltransferases among various segments of the intestinal
tract. Weiser (1 973a,b) compared the glycosyltransferase activity of mature cells
in the villus with immature cells of the crypt by measuring the incorporation of
radiolabeled monosaccharides into surface membrane glycoproteins, and using
this as a measure of the corresponding glycosyltransferase activity. He em-
ployed a separation method based on citrate and EDTA to dissociate cells, re-
sulting in epithelial cell fractions which defined a gradient of cells from villus
tips to crypts. His results demonstrated that the levels of N-acetylgalac-
tosaminyltransferase, galactosyltransferase, and fucosyltransferase were ap-
proximately 10-fold greater on crypt as compared to villus cells, whereas sialyl-
transferase activity was higher on villus cells. In a subsequent study, however,
Weiser and co-workers (1978) not only separated crypt from villus cells, but
also prepared membrane fractions, and reported that the basolateral plasma
membrane of villus cells was rich in galactosyltransferase activity. This discrep-
ancy with their previous results (Weiser, 1973a,b) was explained as resulting
from the presence of glycosidases on the microvilli of intact villus cells in the
earlier study which had interfered with the detection of glycosyltransferases on
the lateral plasma membrane. However, an alternative explanation was proposed
by Lau and Carlson (1981). They found that nucleotide pyrophosphatase, an en-
zyme that interferes with glycosyltransferase assays, is particularly enriched in
intestinal mucosa (especially at the villus tips). By assuring inactivation of this
enzyme, they determined that the activity of two galactosyltransferases (one act-
ing on asparagine-linked and the other acting on 0-linked oligosaccharides)
displayed essentially identical activities on both crypt and villus cells. They
advocated exercising caution when interpreting the measurement of intestinal

glycosyltransferase activities without recognizing the potential influence of nu-
cleotide pyrophosphatase.
In subsequent studies, Weiser’s group (Weiser el a/., 1987; Wilson el al.,
1987) took precautions against nucleotide pyrophosphatase activity and ana-
lyzed the rat intestinal distribution of two different galactosyltransferases; one
acting on N-linked oligosaccharides and the other acting on 0-linked (mucin-
type) oligosaccharides (this enzyme may be identical to that investigated by Lau
and Carlson mentioned above). They were still able to detect both crypt : villus
differences as well as differences among intestinal segments for both enzymes.
The galactosyltransferase acting on 0-linked oligosaccharides showed increased
activity in the duodenum and distal ileum of the small intestine, and the cecum
and proximal colon of the large intestine (Wilson eta/. , 1987). These areas of in-
creased activity corresponded to areas of increased mucus production.
Moreover, within the duodenum this galactosyltransferase showed a moderately
increased activity in cells from the crypt region as compared to those of the vil-
lus; however, no such difference was detectable in the jejunum or ileum. The
galactosyltransferase acting on N-linked oligosaccharides displayed highest ac-
tivities in the terminal ileum, cecum, and proximal colon, with lesser amounts
detected in the jejunum and duodenum (Weiser et af., 1987). Although they
could not demonstrate a difference in total homogenate galactosyltransferase ac-
tivity between crypt and villus cells, they found that assays for cell surface
galactosyltransferase revealed an elevation in the crypts (Section IV,C).
Kim and co-workers (1975) devised a planar sectioning technique utilizing a
mounted razor blade to cut frozen sections for the separation of crypt from villus
cells. Upon homogenization the sections were assayed for glycosyltransferase
activity. The results showed that sialyltransferase activity was enriched in crypt
cells. whereas galactosyltransferase activity was approximately equal in both re-
gions (Kim ef al.. 1975). These results are in contrast to those from Weiser’s
group mentioned above. The discrepancy may have arisen from the different
methodologies employed by the two groups for the separation of cell popula-
tions or from glycosyltransferase activity variation among intestinal sections.
Nevertheless, both separation techniques suffer from the questionable purity of
the fractions. Indeed, the importance of the purity of intestinal fractions cannot
be overstated, since elements other than intestinal epithelial cells have recently
been shown to be the major, if not the only source for sialyltransferases in rat
small intestine (Paulson et af., 1989).
Glycosyltransferase activities were also shown to vary from the proximal to
the distal regions of the rat small intestine (Morita et al., 1986). Specifically, ac-
tivities for two galactosyltransferases (acting on N- and 0-linked oligosaccha-
rides), two sialyltransferases (acting on N- and 0-linked oligosaccharides), fu-
cosyltransferase and N-acetylgalactosaminyltransferasewere consistently found
GLYCOSYLATIONIN INTESTINAL EPITHELIUM 153
to be higher in distal regions of the small intestine compared with proximal re-
gions. These results were corroborated by the carbohydrate analysis of brush
border membranes in proximal and distal small intestine. Interestingly, both sia-
lyltransferase enzymes displayed the lowest activity of all the glycosyltrans-
ferases assayed. In an immunocytochemical investigation (Section IV,C) we
found that the distribution of the a2,6-sialyltransferase was regionalized within
the rat intestine (Taatjes et al., 1988a). Although abundant staining was de-
tectable in all regions of the large intestine, no labeling was detectable in any
portion of the small intestine from the same animals. These results were corrob-
orated by direct measurement of enzymatic activity for the a2,6- and a2,3-sia-
lyltransferases. In this case, the activity of these two sialyltransferases (both act-
ing on N-linked oligosaccharides) was undetectable in rat small intestine. These
results would appear to conflict with those of Morita et al. (1986) mentioned
above, as well as those of Van Halbeek er al. (1983) who reported the presence
of sialic acid a2,3 linked to galactose in much glycoproteins from rat small in-
testine. The apparent discrepancy may be explained in part by the fact that we
examined sialyltransferase activity only for the mucosal surface of the intestine
scraped from the intestinal wall. Indeed, in a subsequent investigation, Paulson
et al. (1989) found that homogenates of the intestinal wall itself contain substan-
tial levels of this sialyltransferase in the small intestine. Moreover, in contrast to
the results of Morita er al. (1986) they found that the activity for a sialyltrans-
ferase (adding sialic acid in an a2,3 position to galactose in 0-linked oligosac-
charides) actually decreased in the mucosa from proximal to distal small intes-
tine. Ironically, no activity was detectable in the ileal mucosa, whereas
substantial activity was measurable in the ileum wall. Thus, results obtained
from intestinal homogenates or segments are not directly comparable. This also
indicates that these sialyltransferase enzymes are differentially expressed within
different regions of the small intestine, each having specialized functions, yet
identical cell types. These results have recently been further supported by in siru
lectin-binding studies (Section IV,A,6).
The situation is further complicated by the acceptor substrates used for the
measurement of glycosyltransferase activity. For instance, according to the
“one-enzyme one-linkage’’ hypothesis (Hagopian and Eylar, 1968) at least a
dozen different sialyltransferases must exist in order to form the known linkages
of sialic acid to penultimate sugars. Thus, fetuin, which is quite often employed
as an acceptor substrate for sialyltransferase activity, contains both N- and 0-
linked carbohydrate groups which are acceptors for at least four different sialyl-
transferases (Kim er al., 1975; Weinstein et af., 1982; Green et al., 1988). This
may explain in part the sometimes variable results reported using different tech-
niques with respect to glycosyltransferase distribution within the intestine.
Earlier studies demonstrating glycosyltransferase activity based upon known ac-
ceptor substrates may have actually been measuring the activity of several
different enzymes. However, due to antibody specificity, immunological-based

techniques are assaying for the presence of one highly linkage-specific glycosyl-
transferase. Therefore, these limitations should be borne in mind when interpret-
ing results from different investigators employing different techniques.
Much less information is available concerning the glycosyltransferase activity
in the large intestine. Freeman et al. (1978) have demonstrated that in adult rat
large intestine both gaiactosyltransferase and sialyltransferase activities were
significantly greater in proximal than in distal colonic mucosa. We have re-
ported (Taatjes e f al.. 1988a) enzymatic activity for two sialyltransferases acting
on asparagine-linked oligosaccharides in mucosal scrapings from rat large intes-
tine (Section 1V.C; Table IV).
2 , Measurement of Glvcosyltransferase Activity during Development
The activity of several glycosyltransferases has been found to vary during
postnatal development in both rat small and large intestine. Sialyltransferase ac-
tivity was present in increased levels during the suckling period, and decreased
5-fold during the subsequent weaning and adult periods (Chu and Walker, 1986).
On the other hand, fucosyltransferase activity was very low during suckling
phase, rapidly increased during weaning, and reached adult levels by 5 weeks of
age (Chu and Walker, 1986).
The activities of two galactosyltransferases, the UDP-Gal : GlcNAc(P1-4)-
galactosyltransferase, and UDP-Gal : GalNAc(p 1 -3)-galactosyltransferase have
also been found to be under developmental regulation (Ozaki et al., 1989). Both
glycosyltransferases demonstrated a marked elevation in activity after the wean-
ing period and into adulthood in all regions of the rat small intestine. All of these
results considered together demonstrate that activities for galactosyltransferase,
N-acetylgalactosaminyltransferase and fucosyltransferase all increase during
postnatal development of rat small intestine. On the other hand, sialyltransferase
activity declines during the same developmental period. These data correlate
well with known changes in terminal glycosylation of microvillar proteins dur-
ing postnatal development of rat small intestine (Sections IV,A,4 and 6).
The activity of galactosyltransferase has been measured in fetal and postnatal
rat large intestine (LaMont and Ventola, 1978). The activity in fetal ho-
mogenates increased 4-to 7-fold between 18 and 22 days, the last 4 days of ges-
tation. The enzyme activity then gradually increased postnatally, reaching adult
levels by day 15. Determination of the autoradiographic incorporation of
[3H]galactose into fetal large intestine glycoconjugates correlated well with the
increase in galactosyltransferase activity during this period (Rampal et al.,
1978). Interestingly, these autoradiographic studies revealed the selective incor-
poration of [3H]galactoseinto goblet cells but not into absorptive cells. These re-
sults therefore suggest that the maturation of fetal rat large intestine during the
last 4 days of gestation is accompanied by the appearance of goblet cells and en-
hanced mucus synthesis.
3. Analysis of Sugar Content of Membrane Glycoproteins in Adult Animals
Two basic biochemical methods have traditionally been applied in order to
determine the carbohydrate composition of intestinal membrane glycoproteins.
The first method requires the administration of radiolabeled sugar precursors
into the lumen of the intestine, followed by purification and analysis of the in-
corporation of the sugars into membrane glycoproteins. The second method em-
ploys the binding of labeled lectins to intestinal plasma membrane fractions.
In an early study, Kim and Perdomo (1974) traced the incorporation of
[14C]glucosamineinto the membranes of intestinal cells. They prepared three
membrane fractions, consisting of smooth, rough, and brush border membranes.
They observed incorporation first into smooth membranes, followed after a lag
period by entrance into rough and brush border fractions. Aside from the pecu-
liar late entrance into a rough membrane fraction, these results trace the transit
of ['4C]glucosamine-containingglycoconjugates from the Golgi apparatus to the
brush border. A further purification to distinguish apical (brush border) mem-
branes from basolateral was not attempted.
Quaroni and co-workers (Quaroni et al., 1980; Herscovics et al., 1980) took
these studies further by separating Golgi apparatus, apical, and basolateral mem-
branes, both in crypt and villus cells. They measured the incorporation of ~ - [ 5 , 6 -
3H]fucose and ~-[2-~H]mannose into intestinal membrane glycoproteins follow-
ing an intraperitoneal injection of these radiolabeled sugars. The incorporation of
mannose was roughly equal in crypt and villus cells, whereas fucose incorpora-
tion was elevated in the differentiated villus cells (Quaroni et al., 1980).
Fucosylated glycoproteins were originally detected in the Golgi apparatus and
basolateral membranes, followed by redistribution into villus membranes after
3 4 hr. In contrast, most mannose-labeled glycoproteins remained in the Golgi
and basolateral membrane fractions. They interpreted their results to indicate that
fucosylated glycoproteins represent a special class of membrane components that
appear with differentiation (absent in undifferentiated crypt cells) and are specif-
ically localized to the luminal portion of the intestinal cell plasma membrane. In
an accompanying paper, Herscovics et al. (1980) used similar techniques to
demonstrate that high-mannose oligosaccharides were the precursors of complex
oligosaccharides. Moreover, they provided evidence that luminal membranes of
both crypt and villus cells were greatly enriched in complex oligosaccharides as
compared with basolateral plasma membranes, but no qualitative changes were
found to occur during cellular differentiation. Thus, their results suggested that
intestinal epithelial cells were polarized with respect to plasma membrane glyco-
conjugate oligosaccharide composition (Section IV,A,5).
The second biochemical method, utilizing lectin binding to isolated cell

plasma membranes, is a more recent innovation, and has found widespread use
for the comparison of membrane glycoproteins among intestinal segments as
well as during maturation. Kim and co-workers (Morita et a [ . , 1986) examined
the reactivity of brush border membrane components with lectins, and the bind-
ing of brush border membrane-associated enzymes to Ricinus communis lectin I
(RCL I) and wheat germ agglutinin (WGA) in segments from the proximal and
distal small intestine. In addition they analyzed the carbohydrate composition of
the brush border membranes. Their results indicated that although brush border
membrane glycoproteins from distal portions of the small intestine contained
more complete oligosaccharide side chains, the glycoprotein profile on SDS gels
was less complex than in proximal small intestine. Specifically, more WGA and
succinylated-WGA-binding glycoproteins were present on brush border mem-
branes from proximal compared to distal segments. However, the binding of
RCL I to brush border membranes was two times higher in the distal as com-
pared to proximal small intestine. Overall sugar content was higher in distal
small intestine brush border membranes, reflected mainly by elevated galactose
and sialic acid content. The content of N-acetylglucosamine appeared equal in
the two intestinal segments. These results suggest that the carbohydrate content
of brush border membranes changes with the progression of the gastrointestinal
tract, with more distal regions of the small intestine containing more completed
oligosaccharide chains.
4 . Analysis of Sugar Content of Membrane Glycoproteins During Development
Similar biochemical methods to those just mentioned above have also been
used extensively to investigate changes in the glycosylation pattern of microvil-
lar proteins in postnatal intestine. Mahmood and Torres-Pinedo ( 1 983) incu-
bated microvillar membrane preparations from postnatal rats with radiolabeled
lectins to determine the carbohydrate profile of membrane glycoconjugates.
They found that the microvillus membrane of suckling rats (from birth to about
2 weeks of age) was rich in glycopeptides containing binding sites for peanut
lectin (PNL) in sialyl-substituted form. During the weaning phase (14-21 days
postnatal), the membranes lost about half of these binding sites, accompanied by
decreased sialic acid content and increased content of glycopeptides containing
unsubstituted binding sites for soybean lectin (SBL) and RCL I. Perhaps the
most important result from this study was the finding that the sialic acid content
of microvillar plasma membrane drastically decreases from the suckling to the
weaning period. Indeed, in a subsequent paper (Torres-Pinedo and Mahmood,
1984) they found that this decrease in microvillar plasma membrane sialic acid
content was accompanied by a dramatic rise in fucose content. They observed
that the binding of '2sl-labeled WGA to neuraminidase-sensitive sites in the mi-
crovillar membrane decreased markedly from early suckling to weaning ages.
At the same time, the binding of Iz5I-labeledUlex europaeus lectin I (UEL I) to

microvillar membranes showed an opposite increase from suckling to weaning
periods. This developmentally related shift from sialylation to fucosylation was
found for both glycoproteins and glycolipids of the microvillar membrane, sug-
gesting that it is a general phenomenon for membrane constituents. They postu-
lated that such a dramatic shift from a strongly acidic to a more neutral microvil-
lar glycocalyx could relate to the physiological changes occumng in the
intestine concomitant with development.
These studies were followed up by examining the carbohydrate profile of in-
dividual microvillar membrane proteins during postnatal development of the rat
small intestine. Srivastava et al. (1987) found that the terminal glycosylation of
several microvillar glycoproteins of >90,000 Da (most likely hydrolases) does
not reach complete maturation until after weaning, although their content within
the membrane has reached adult levels by this time. Moreover, several of these
glycoproteins were fully sialylated during the suckling period, whereas addition
of N-acetylgalactosamine and fucose continued well into the weaning period.
Buller el al. (1990) took such investigations one step further by examining the
glycosylation of a known glycoprotein, lactase-phlorizin hydrolase, during de-
velopment of the rat small intestine. Lectin binding to the enzyme immunopre-
cipitated from microvillus membranes revealed the presence of both N- and 0-
linked oligosaccharide chains containing mannose and galactose, which did not
vary throughout development. In contrast, the content of fucose and sialic acid
was developmentally regulated; sialic acid was present at weaning and declined
through adulthood, whereas fucose was not detectable until rats were 20 days of
age. Thus, by examining a single enzyme, it was established that the core N- and
0-linked oligosaccharide structures of this microvillar hydrolase remain con-
stant during development, whereas alteration in terminal glycosylation occurs
with a shift from sialic acid at suckling to fucose in adulthood. The above stud-
ies taken together show that a definite change in glycosylation occurs on spe-
cific microvillar membrane glycoproteins during the postnatal developmental
period.
In a related study (Jaswal et al., 1988), the content of sialic acid and fucose in
enterocytes was measured in crypt and villus cells from suckling and adult ani-
mals. In suckling animals, no change was found in the sialic acid content of en-
terocytes during progression from crypt to villus. In contrast, the sialic acid con-
tent decreased precipitously from crypt to villus in adult animals. The fucose
content of enterocytes from suckling animals was greater in the crypts than in the
villus, whereas in adult animals fucose content was much greater in the villus.
5 . Cytochemical Detection of Lectin Binding to Intestinal Cells in Situ
Much cytochemical data based upon lectin binding studies have contributed
to the understanding of intestinal glycosylation patterns. In the absence of
158 DOUGLAS J . TAATES AND JURGEN ROTH
glycosyltransferase enzymatic measurements, lectin-binding sites can be taken

as indicative of specific glycosyltransferase activity. The earliest studies used
lectins conjugated to fluorescent dyes in a direct labeling technique, usually on
frozen or paraffin sections. Such investigations demonstrated differences in
lectin-binding patterns to intestinal epithelial cells in the various segments of the
intestinal tract. For instance, Etzler and Branstrator (1974) examined the binding
of FITC-conjugated lectins from Dolichos biflorus (DBL), Lotus terragonolobus
(LTL), Ricinus communis I, and Triticum vulgare (WGA) to the various regions
from rat small intestine. They observed differences in the binding of the lectins
to both the epithelial cell plasma membranes, as well as to the goblet cell mucus.
With respect to plasma membrane staining, LTL, RCL I, and WGA bound to the
microvillar portion of the epithelial cells lining the crypts and villi in the proxi-
mal regions of the small intestine. This pattern of staining was altered along the
first 15 cm of the small intestine, such that distal to this point the apical surfaces
of only those epithelial cells in the crypts and at the base of the villi reacted with
LTL and RCL I, while WGA stained the apical surfaces of cells lining the villi.
In the distal small intestine, LTL, RCL I, and WGA stained the cell surfaces of
only those epithelial cells at the base of the villi and in the crypts. DBL did not
stain the epithelial cell surface in any portion of the small intestine. With respect
to staining of the mucus content of goblet cells, WGA and DBL stained the gob-
let cells in proximal portions of the intestine, whereas in middle and distal re-
gions all four lectins were found to stain goblet cell mucus. These results sug-
gested that the content of complex carbohydrates in goblet cell mucus increases
from proximal to distal regions of the small intestine.
In a preembedding peroxidase study, Ovtscharoff and Ichev (1984) showed
that in rat small intestine (middle regions), the pea and soybean lectins stained
the microvillar membrane from epithelial cells in the crypts and lower villus
more intensely than those in the upper villus and lumen.
Essner et al. (1978) also used several lectins conjugated to FITC to investi-
gate the lectin-binding pattern to cryostat sections from portions of the
descending colon of the rat. Besides reactivity in goblet cell mucus and plasma
membrane, they identified cytoplasmic staining which they attributed to the
Golgi apparatus. Binding sites for the lectins from Glycine max (SBL) and
Dolichos hiflorus were observed in goblet cell mucus, apical and basolateral
plasma membranes, and in the apical cytoplasm, indicating the presence of ter-
minal nonreducing N-acetylgalactosaminyl residues at these sites. WGA, RCL I,
UEL I, and concanavalin A (Con A) all stained the cytoplasm of epithelial cells,
but did not, or only weakly, stain mucus droplets and plasma membranes.
Gorelick et al. (1982) examined lectin-binding patterns in the plasma mem-
brane and goblet cell mucus of epithelial cells in the various regions of guinea
pig large intestine. Staining of the brush border with the various rhodamine-la-
beled lectins tended to be heterogeneous across the regions of the large intestine.
GLYCOSYLATION IN INTESTINALEPITHELIUM 159
In general, though, Con A, WGA, RCL 11, and PNL stained the brush border of
the right colon, whereas RCL I and SBL stained the transverse colon intensely.
Limulin reacted with the brush border only in the left colon. Staining of the gob-
let cell mucus was even more variable. Gorelick et al. (1982) not only examined
the lectin binding to the goblet cell mucus in the various regions of the large in-
testine, they also separated the intestinal segments according to crypt regions:
basal, middle and apical. Again, the staining patterns varied not only among the
large intestinal segments, but also within crypt region of individual segments.
Noted exceptions were LTL and PNL which did not stain the goblet cell mucus
in any area examined. Moreover, goblet cells in the transverse and left portions
of the large intestine tended to react more intensely with the various lectins em-
ployed, suggesting a maturation of goblet cell mucus along the large intestine.
The paper by Gorelick et al. (1982) also served to usher in the modem approach
to investigating intestinal lectin-binding sites; namely, the postembedding appli-
cation of colloidal gold-labeled lectins to ultrathin sections. They prepared a
complex of colloidal gold particles with RCL I1 and applied this to sections from
intestine embedded in Epon-Araldite. They were able to demonstrate binding of
this complex to goblet cell mucus, apical plasma membrane, apical cytoplasmic
vesicles, and Golgi apparatus. Since this time, numerous papers have been pub-
lished utilizing both colloidal gold-labeled lectins and peroxidase-labeled lectins
at the light and electron microscopic level to investigate intestinal glycosylation
patterns. The main,benefit of such studies has been the increase in resolution
over fluorescence studies obtainable with these methods. Many of these studies
were interested in investigating the role of the Golgi apparatus in glycosylation,
and their results will be described in Section IV,B,3. In the current section we
will detail the results of these studies as they relate to plasma membrane and
mucous glycoconjugates.
Helixpomatiu lectin (HPL) binding has been observed (Fig. 10) in the mucus
and apical plasma membrane of chick duodenum (Roth, 1984), rat duodenum
(Ellinger and Pavelka, 1985), and rat jejunum (Murata er al., 1986). RCL I bind-
ing has been reported in the mucus and apical and basolateral plasma mem-
branes of chick (Roth, 1983b) and rat duodenum (Ellinger and Pavelka, 1985),
and the basolateral plasma membrane of rat proximal colon epithelial cells
(Roth et al., 1988a). Pavelka and Ellinger (1989b) have shown binding of
Erythrina cristagalli lectin (ECL) to apical and basolateral plasma membranes
and goblet cell mucus in rat duodenum, while Egea et al. (1989) have reported
identical results by using Daturu srramonium lectin (DSL). Lotus terrago-
nolobus lectin (LTL) binding to the apical plasma membrane and goblet cell
mucus has been described in chick duodenum (Roth, 1983b), while UEL I bind-
ing to apical plasma membrane and goblet cell mucus has been reported for rat
duodenum (Ellinger and Pavelka, 1988a). Binding of sialic acid-specific lectins
has also been documented in intestinal cells. Roth et ul. (1984) found binding of
the Limaxflavus lectin (LFL) to the apical plasma membrane and goblet cell
mucus in rat distal colon. Similar results were also found in rat proximal colon
(Taatjes and Roth, 1988). Recently, staining with Maackia amurensis lectin
(MAL) has been reported in the apical and basolateral plasma membranes and
goblet cell mucus in pig colon (Sata el al., 1989). Furthermore, we have recently
observed staining with Sambucus nigra L. lectin (SNL I) complexed with col-
loidal gold particles (Taatjes et al., 1988b) in mucus droplets, but not in the
plasma membrane of rat jejunal epithelial cells (Taatjes and Roth, 1990; Section
IV,A,6). The main message resulting from all of these studies is that lectins rec-
ognizing complex carbohydrate structures bind to plasma membranes and goblet
cell mucus in both small and large intestine from various species.
6 . CyfochemicalDetection of Lectin Binding to Intestinal Cells in Situ During
Development and Diferentiation
As described in Section IV,A,2, the activities of several glycosyltransferases
are altered during intestinal cell development. Such alterations are also reflected
in the modification of lectin binding to epithelial cells that occurs during postna-
tal development. In the rat small intestine, Etzler and Branstrator (1979) found
developmental changes in the binding of RCL I, LTL, and WGA. RCL I stained
the brush border of epithelial cells as early as 1 hr after birth. The staining be-
came patchy at the cell surface over the next few days, reacting uniformly with
the surface 5-14 days after birth. By 19-24 days postnatal, the epithelial cell
surface began to lose its ability to react with RCL I, and by 30 days postnatal,
the cell surfaces were no longer stained with RCL I. The onset of LTL staining
was a much later event, commencing between 11 and 19 days after birth. By 28
days after birth, regional differences were apparent with respect to LTL binding
to intestinal cell surfaces; brush borders of cells lining the villi in the distal por-
tion of the small intestine were no longer bound by LTL. Wheat germ agglutinin
(WGA) stained the brush border of epithelial cells from 1 hr after birth, until
about postnatal day 19 when cells lining the villi were no longer stained with
this lectin.
We have recently investigated the binding of sialic acid-specific and fucose-
specific lectins to developing rat small intestinal cells (Taatjes and Roth, 1990).
In line with the results detailed in Sections IV,A,2 and 3 concerning develop-
mental-related changes in sialyltransferase and fucosyltransferase activities, as
FIG. 10. Low-power electron micrograph demonstrating H . pomatiu lectin-gold binding sites in
chick duodenum. In the center of the micrograph a prominent goblet cell is displayed, with intense
gold particle labeling present in the goblet cell mucus (asterisk), in the Golgi apparatus, and in the
apical plasma membrane (arrowheads). Label is also observable in the apical plasma membrane of
adjacent absorptive cells (arrowheads). X 1,710. Bar = 5 p n . (Reproduced from the J . Cell B i d .
1984.98, 399406 by permission of the Rockefeller University Press.)
I62 DOUGLAS J. TAATES AND JURGEN ROTH
well as the lectin-binding results of Etzler and Branstrator (1974), we found that
binding of SNL I, LFL, and UEL I to intestinal cells changed with postnatal de-
velopment. SNL I (Fig. lla,b) and LFL stained the brush border and mucus
droplets in animals during the suckling phase. During weaning (day 23) we
found that individual epithelial cells were no longer stained with SNL I (Fig.
llc,d and Fig. 12) and LFL. By adulthood, staining with these two sialic acid-
specific lectins was restricted to goblet cell mucus and cells in the lamina pro-
pria and submucosa (Fig. 1 le,f). In contrast, binding of fucose-specific UEL I
was restricted to goblet cell mucus during the suckling phase, but by day 23
postnatal appeared in the brush border of some epithelial cells. In adults, intense
staining with UEL I was found in goblet cell mucus and in the brush border of
epithelial cells. All of these results taken together support the premise that dur-
ing postnatal development of rat small intestine, a progressive change from sia-
lylation to fucosylation of brush border glycoconjugates occurs.
Caldero et al. (1988) have performed a detailed investigation of changes in
glycoconjugate composition of the rat colonic mucosa during development.
They used a battery of eight fluorescein-conjugated lectins recognizing a variety
of sugar residues. Their results demonstrated that each lectin showed a unique
developmental staining pattern, including differences between the various re-
gions of the colon. In all cases, the adult pattern of staining was achieved 25-30
days after birth.
Differentiation-related changes in intestinal cell glycosylation patterns have
been described in adult animals during cell migration from crypts to the villus or
lumen. Some of these were already described above (Section IV,A,S; Etzler and
Branstrator, 1974). We have investigated the localization of LFL binding sites in
the plasma membrane of rat colonic epithelial cells during differentiation
(Taatjes and Roth, 1988). We found that in the crypt regions, goblet and absorp-
tive cell precursors were stained along their entire plasma membrane (Fig. 13);
that is, both apical and basolateral plasma membranes were stained. However,
when cells reached the zone of migration (Eastwood, 1977) the staining with
LFL became restricted to the apical plasma membrane (Fig. 13). This polarized
staining remained a feature of fully mature epithelial cells (both absorptive and
goblet) located at the intestinal lumen. These results suggest that a feature of
FIG.1 1 . Light micrographs illustrating the detection of SNL I-gold binding sites in epithelial
cells during postnatal development of rat jejunum. At postnatal day 1 (a,b), staining is present in the
epithelium along the apical plasma membrane (arrowheads)and in the goblet cell mucus (asterisk).
By postnatal day 23 (c,d), individual cells in the epithelium are not stained by the SNL I-gold com-
plex (arrows). In adult animals (e,9, the apical plasma membrane (arrowheads)of all epithelial cells
is not stained by SNL I-gold complex, whereas goblet cell mucus (asterisks) and the plasma mem-
brane of cells in the lamina propria (Ip) are intensely stained. a,c,e, Bright-field micrographs; b,d,f,
correspondingphase-contrast images. lp, lamina propria. X 368 (a-9. Bar = 5 pm.
164 DOUGLAS J . TAATJES AND JURGEN ROTH
fully differentiated colonic epithelial cells is the polarization of the plasma

membrane with respect to the distribution of sialic acid residues on membrane
glycoconjugates. A similar phenomenon was apparent in the small intestine, al-
though fully differentiated small intestinal epithelial cells display very sparse
LFL binding sites.
B. GOLGIAPPARATUS
In an early study, Kim and co-workers (1 97 1) investigated the subcellular dis-
tribution of the then called “multienzyme system” of glycosyltransferases in rat
small intestinal mucosal scrapings. They determined that the polypeptide : N-
acetylgalactosaminyltransferase, galactosyltransferase, N-acetylglucosaminyl-
transferase, and N-acetylgalactosaminyltransferase were enriched in a smooth
microsome fraction. This was the first detailed report of the localization of gly-
cosyltransferases in intestinal tissue, and quite accurately determined them to be
located in a fraction most likely representing Golgi apparatus membranes. In a
subsequent investigation, Kim and Perdomo (1974) investigated the intestinal
membrane distribution of five glycosyltransferases: two galactosyltransferases
(acting on N- and 0-linked oligosaccharides), sialyltransferase, fucosyltrans-
ferase, and N-acetylgalactosaminyltransferase.They found that all five enzymes
were enriched in a smooth membrane fraction (Golgi apparatus), with only
background amounts detected in a rough membrane fraction and a brush border
membrane fraction.
More recently the techniques of immuno- and lectin cytochemistry have
helped to unravel the pattern of glycosylation reactions in the intestinal cell
Golgi apparatus. However, before we begin to detail these intestinal studies, it
will be helpful to briefly review the concept of general Golgi apparatus glycosy-
lation as formulated by the assimilation of data from several different tech-
niques.
1. Subcompartmentation Model of the Golgi Apparatus
Several recent reviews have considered this topic in detail and the interested
reader should refer to them for more information (Dunphy and Rothman, 1985;
Farquhar, 1985; Kornfeld and Kornfeld, 1985; Roth, 1987a; Roth and Taatjes,
1989). Briefly, this model proposes that the Golgi apparatus cisternal stack is
FIG. 12. Demonstration of SNL I-gold binding sites in the jejunum of sections from postnatal
day 23 rat. The apical plasma membrane (brush border) of adjacent epithelial cells is shown. Large
gold particles (14-nm diameter) indicative of SNL I binding sites are restricted to the cell on the
right. To rule out possible processing artifacts, this section was also stained with RCL I/asialofe-
tuin-gold (small gold particles; 10-nm diameter). As can be seen, both cells are RCL I positive, in-
dicating that loss of binding sites is specific for SNL I. The lateral plasma membrane (arrowheads)
separating the two cells contains binding sites for both lectins. X 66,000. Bar = 0.15 pm.
FIG. 13. Detection of sialic acid residues with the Limo.rf7avus lectin/fetuin-gold technique at
the plasma membrane of rat colonic absorptive cells. in differentiated cells at the surface epithelium
(a). gold particle label is restricted to the apical plasma membrane (arrowheads). Note the lack of
staining in the basolateral plasma membrane (asterisks). In contrast, the basolateral plasma mem-
brane of undifferentiated absorptive cells from the crypts region (b) is intensely stained for sialic
acid residues (arrowheads). n, nuclei of absorptive cells. X 18,000 (a); X 55,000 (b). Bars = 0.6 p n
( a ) and 0.2 p n (b). (Reproduced with permission from Taatjes and Roth, 1988.)
GLYCOSYLATIONIN INTESTINAL EPITHELIUM I67
functionally subcompartmentalized with respect to the steps involved in the pro-

cessing of oligosaccharide side chains of glycoconjugates. Although based upon
the maturation of N-linked oligosaccharides of glycoproteins, support for this
model has also been presented for the case of 0-linked oligosaccharides of gly-
coproteins. Inherent in this concept is the premise that glycosyltransferases
which act early in the pathway are preferentially located in cis-cistemae of the
Golgi apparatus, whereas those acting at intermediate steps are located in middle
cisternae, and those acting at terminal steps are housed in trans-cisternae. This
compartmentation, or spatial separation, would allow the enzymes to act upon
an oligosaccharide chain in an “assembly line” progression, without risk of in-
terfering with the action of one another. This model is thus very attractive bio-
chemically, and indeed has received much experimental support. For instance, in
cell fractionation studies utilizing analytical sucrose gradients, activities for ear-
lier and later acting oligosaccharide-processing enzymes were detected in dis-
tinct Golgi apparatus fractions (Dunphy et al., 1981; Dunphy and Rothman,
1983; Goldberg and Komfeld, 1983). More direct evidence, however, has been
provided by numerous investigations analyzing the in situ cytochemical detec-
tion of various sugar residues with lectins (Pavelka, 1987; Roth er al., 1988b)
and by the immunolocalization of a few glycosyltransferases (Roth and Berger,
1982; Dunphy er al., 1985; Roth et al., 1985a). Indeed, the first direct demon-
stration of Golgi apparatus subcompartments was provided by the immunocyto-
chemical localization of galactosyltransferase by Roth and Berger (1982). They
found that galactosyltransferase immunoreactivity colocalized with thiamine
pyrophosphatase activity in one or two trans-Golgi apparatus cisternae in HeLa
cells. This result indicated that the Golgi apparatus contained at least two com-
partments with respect to glycosylation reactions: cis (defined as galactosyl-
transferase negative) and trans (defined as galactosyltransferase positive). The
number of identifiable subcompartments increased to three with the localization
of N-acetylglucosaminyltransferase I to middle cisternae of the Golgi apparatus
stack (Dunphy et al., 1985). Finally, the most distally acting glycosyltransferase,
sialyltransferase, was detected in two trans-cisternae and a complex trans-tubu-
lar network continuous with these cisternae in rat hepatocytes (Roth er al.,
1985a). Interestingly, sialyltransferase immunoreactivity was found in portions
of the Golgi apparatus stack which also contained cytochemically demonstrable
thiamine pyrophosphatase activity, a classical trans-Golgi marker, or acid phos-
phatase (CMPase) activity, a classical marker for the GERL element of the
Golgi apparatus. These results suggested that in hepatocytes the Golgi apparatus
is composed of three subcompartments with respect to glycosylation reactions:
cis, so far delineated by what it does not contain; middle, containing N-acetyl-
glucosaminyltransferase I; and trans, containing sialyltransferase. We include
the trans-tubular network (Rambourg and Clermont, 1990), or trans-Golgi net-
work (Griffiths and Simons, 1986) as part of the trans-Golgi apparatus since
functionally it is involved in sialylation as are trans-cisternae, and structurally it

is continuous with trans-Golgi cisternae (Roth et al., 1985a; Taatjes and Roth,
1986). This view differs from that of Griffiths and Simons (1986), who regard
the trans-Golgi network as perhaps a fourth Golgi subcompartment, separate
from trans-cistemae. Moreover, other investigators have proposed that galacto-
syltransferase is housed in trans-Golgi apparatus cisternae, whereas sialyltrans-
ferase is separated and housed in the more distally located trans-Golgi network
(Berger and Hesford, 1985; Thorens and Vassalli, 1986; Berger et al., 1987).
Several pieces of evidence contest this view. First, we have shown that sialyl-
transferase is not only localized in the trans-Golgi network of hepatocytes, but
also quite clearly in two trans-cistemae of the Golgi apparatus stack (e.g., Fig. 3
in Roth ef al., 1985a). Second, by double-labeling immunofluorescence we
found an identical codistribution of galactosyltransferase and sialyltransferase
irnmunoreactivity in cultured rat hepatocytes (Taatjes el al., 1987). Third, Geuze
and co-workers ( 1985) found that galactosyltransferase was detectable in the
trans-Golgi network, in addition to trans-cisternae in hepatoma cells and liver
hepatocytes. Fourth, galactose residues detected with RCL I, were found in the
trans-cisternae and trans-Golgi network of hepatocytes (Lucocq et al., 1987).
Clearly, the ability to resolve more and more Golgi apparatus subcompartments
will come with the introduction of more Golgi apparatus-specific antibodies. Of
certain importance for the previous discussion will be the simultaneous im-
munocytochemical demonstration of galactosyltransferase and sialyltransferase
in the same Golgi apparatus at the electron microscopic level. Due to the con-
straints placed upon imrnunocytochemical investigations by antibody cross-re-
activity with other animal species, this experiment has not proven possible.
Moreover, cell-specific variability with respect to the organization of Goigi ap-
paratus subcompartments may have been a factor in the above described dis-
crepancies.
2 . Immunoc:\.tochemicalLocalization of Glycosyltransferases in Golgi

Apparatus of Intestinal Epithelial Cells
To this date, only two glycosyltransferases have been immunocytochemically
detected in the Golgi apparatus of intestinal cells; yet, their localization has
yielded most interesting results. After having reported on the localization of sia-
lyltransferase in hepatocytes, we sought to expand on these findings by perform-
ing similar localizations on intestinal cells. When we examined the Golgi appara-
tus distribution of sialyltransferase in goblet cells from the rat colon, we observed
the expected result (Fig. 14). Namely, immunoreactivity was restricted to trans-
Golgi apparatus cisternae (Roth et al., 1986). Likewise, sialic acid residues, as
detected with LFL, were localized to trans-cisternae. However, quite surprisingly
when we examined neighboring absorptive cells a quite different pattern of label-
ing emerged: the entire Golgi apparatus cisternal stack (with the exception of the
fenestrated first cis-cistema) was labeled (Figs. 14 and 15a). In a fashion similar
FIG.14. Immunocytochemical localization of sialyltransferasein surface epithelial cells from rat

proximal colon. Gold particle label is restricted to trans-Golgi apparatus cistemae in a goblet cell
(gc), whereas in a neighboring absorptive cell (ac) label is detectable throughout the Golgi apparatus
cistemal stack (with the exception of the fenestrated first cis-cistema). Label is also present in the
goblet cell mucus droplets (md) and in the lateral plasma membrane separating the two cells (arrow-
heads). X 25,500. Bar = 0.4 pm. (Reproduced with permission from Roth er al., 1986.)
to that observed in goblet cells, the distribution of sialic acid residues in absorp-
tive cells was found to mirror that of the sialyltransferase enzyme (Fig. 15b).
Quantitative evaluation of the distribution of sialyltransferaseimmunolabel in the
Golgi apparatus of absorptive versus goblet cells confirmed the differential label
I70 DOUGLAS J. TAATJES AND JURGEN ROTH
FIG. 15. Gold particle label for sialyltransferase (a) and sialic acid residues (b) is distributed
throughout the absorptive cell Golgi apparatus cistemal stack (with the exception of the fenestrated
first cis-cistema) from rat proximal colon. X 63,000 (a): X 66,500 (b). Bar = 0.16 pm (a) and 0.15
pm (b). (Reproduced with permission from Roth PI of., 1986.)
TABLE I1
QUANTIFICATION OF IMMUNOLABEL FOR SIALYLTRANSFERASE IN
RAT PROXIMAL COLONICEPITHELIAL CELLS?
Absorptive cell (n = 26) Goblet cell ( n = 23)
Golgi apparatus Gold Total Gold Total
cistema” particle/pm length (p) particle/pm length (pm)
0.21 +/- 0.08 115.5 0.45 +I- 0.24 148.5
2.55 +/- 0.18 134.3 0.16 +/- 0.08 148.5
2.34 +/- 0.22 138.4 0.15 +/- 0.08 154.3
3.42 +/- 0.33 132.1 0.16 +/- 0.09 154.3
3.63 +/- 0.24 126.8 0.18 +/- 0.09 151.7
4.14 +/- 0.28 124.1 0.18 +/-0.10 148.5
6.20 +/- 0.41 125.7 0.43 +/- 0.27 147.3
4.50 +/- 0.26 147.9
6.64 +/- 0.33 148.5
“From Roth er nl. (1986).
”Cistema 1 designates the fenestrated first cis-cisterna and the following numbers the subsequent
cistemae toward the trans side of the Golgi apparatus.
observed on micrographs (Table 11). Moreover, it was apparent that although the
labeling was diffuse throughout the Golgi apparatus of absorptive cells (with the
exception of the fenestrated first cis-cistema), the labeling intensity increased
gradually from the cis to the trans side. This was the first demonstration of an ap-
parent lack of subcompartmentation for a glycosyltransferase within the Golgi
apparatus cistemal stack. In the same study, we found that another terminal gly-
cosyltransferase, the blood group A N-acetylgalactosaminyltransferase,was dis-
tributed in strikingly different patterns in the Golgi apparatus of absorptive versus
goblet cells from human intestine, restricted to trans-cisternae in goblet cell Golgi
apparatus, and diffusely localized throughout the absorptive cell Golgi apparatus
(Fig. 16). In accordance with the matching labeling in the Golgi apparatus for sia-
lyltransferase and sialic acid residues in rat intestine, the distribution of blood
group A substance (detected with a monoclonal antibody) mirrored that of the
blood group A N-acetylgalactosaminyltransferasein both absorptive and goblet
cells. These results were then confirmed and extended in further studies of human
intestinal cells (Roth er al., 1987, 1988~).Thus, for both N-linked oligosaccha-
ride processing (sialyltransferase),as well as 0-linked (blood group A N-acetyl-
galactosaminyltransferase) a terminal glycosyltransferase was not distributed in
the Golgi apparatus cisternal stack as would be predicted by the subcompartmen-
tation model. The implications of these findings for the elaboration of oligosac-
charide side chains of glycoconjugates in intestinal absorptive cells are not clear.
Although the subcompartmentation of glycosyltransferases would serve to pre-
vent competing reactions which could alter the normal processing of oligosac-
FIG.16. Immunocytochemical localization of blood group A al.3-N-acetylgalactosaminyltrans-

ferase in absorptive cell Golgi apparatus from human ileum. Label is present throughout the cistemal
stack and trans-tubular network of the Golgi apparatus. Note the complexity of the structures at the
trans side of the Golgi apparatus. X 48,000.Bar = 0.2 pm. (Reproduced with permission from Roth
tv a/., 1986.)
charides, it is not clear that normal processing requires such subcompartmenta-

tion. Other mechanisms such as the differential expression of the levels of two
competing glycosyltransferases could favor one terminal glycosylation pattern
over another. Besides, it is not known if the consequences of having oligosaccha-
rides with one type of terminal structure would have functional significance over
another for most proteins. It should be emphasized, though, that more insight into
these questions awaits the immunolocalization of other glycosyltransferases in a
variety of cell types.
Indirect support for these results has emerged from recent lectin-binding in-
vestigations. Diffuse labeling throughout the Golgi apparatus cisternal stack has
been observed with RCL I in mouse epididymal cells (Yokoyama et al., 1980)
and rat absorptive intestinal cells (Pavelka and Ellinger, 1986). Hedman er al.
( 1986) observed label with LFL throughout the cisternal stack with the excep-
tion of one cis-cisterna in 3T3 cells. Similarly, Roth and co-workers (Lee et al.,
1989) found that LFL labeled the entire Golgi apparatus cisternal stack in CHO
cells. Moreover, in the same study (Lee et al., 1989) CHO cells were transfected
with a cDNA coding for the P-galactoside a2,6-sialyltransferase. This enzyme
competed with the endogenous P-galactoside a2,3-sialyltransferase for the ter-
mination of oligosaccharide chains. This competition was assessed by the bind-
ing of SNL I (specific for NeuSAc a2,6-Gal/GalNAc sequences) to sections
from wild type and transfected cells. While SNL I did not stain wild type CHO
cells, the Golgi apparatus of transfected cells was labeled throughout the entire
cisternal stack. Thus, in both wild type and transfected cells, sialic acid residues
were not restricted to trans-cisternae of the Golgi apparatus.
Such lectin-binding studies identify any glycoconjugate in the Golgi appara-
tus carrying the required sugar residues. Although the nature and extent of recy-
cling of glycoconjugates from the plasma membrane through the Golgi appara-
tus remains controversial (Farquhar, 1985; see Snider and Rogers, 1985, 1986;
Neefjes et al., 1988; Reichner et al., 1988, for disparaging views) such recycling
could at least in part explain the presence of complex-type oligosaccharide
chains in the middle and cis regions of the Golgi apparatus cisternal stack. For
this reason, we feel that it is most important to determine the intra-Golgi appa-
ratus distribution of a particular glycosyltransferase before surmising that the
pattern of glycoconjugate localization represents the site of glycosyltransferase
activity.
3. Demonstration of Lectin-Binding Sites in lntestinal Cell Golgi Apparatus
In contrast to the relatively few investigations detailing the localization of
glycosyltransferases within the intestinal cell Golgi apparatus, many studies
have employed lectins for the demonstration of sugar residues therein (Pavelka,
1987). Preembedding methods employing peroxidase-conjugated lectins, as
well as postembedding methods employing colloidal gold-labeled lectins and
glycoproteins have been used. Although the methods and animal species inves-
tigated may differ among the various investigators, the lectin-binding patterns to
intestinal goblet and absorptive cell Golgi apparatus may be summarized as in
Table 111. As can be seen from the table, the interpretations of lectin-binding
studies by various investigators tend to overlap, but also display variability.
Such discrepancies may result from species variability, variability among in-
testinal segments as well as crypt versus villus regions, and methodology (pre-
versus postembedding, tissue fixation and processing, probe preparation, etc.).
Moreover, probably of equal importance is the very subjective nature of the in-
terpretation of lectin labeling patterns within the Golgi apparatus. It may be
rather easy to distinguish between cis- and trans-sides of the Golgi apparatus
cisternal stack, yet what defines where the cis region ends and middle begins, or
where middle ends and trans begins? How many cisternae compose the desig-
TABLE 111
LECTINBINDING
TO GOLGIAPPARATUS
IN
INTESTWALEPITHELIAL
CELLS"
Goblet cell Absorptive cell
Lectin cis middle trans cis middle trans References
ConA nd nd nd + +I- - Pavelka and Ellinger (1985)
SBL + + - nd nd nd Tsuyama et al. (1986)
PSL nd nd nd + +I- - Pavelka and Ellinger (1989a)
LCL nd nd nd + - - Pavelka and Ellinger (1989a)
HPL + +I- +/- nd nd nd Murata et a / .(1986)
t +I- t + - - Pavelka and Ellinger (1985);
Ellinger and Pavelka (1988b)
+ - + nd nd nd Roth ( 1984)
GSL I + + - nd nd nd Ellinger and Pavelka (1988b)
RCL I - +I- + - + +/- Pavelka and Ellinger
(1985, 1989b)
- + - nd nd nd Tsuyama ef a / . (1986)
LFL - - + +I- + + Roth et al. (1986)
UEL I - - t - +I- + Ellinger and Pavelka (1988a)
O+, Staining present; -, not detected; nd, not determined.
nated cis, middle, and trans regions of the stack? In the absence of specific
markers, these borders seem to be arbitrarily defined by individual investigators.
The situation is further complicated by cell type variability. For instance, some
cell types such as hepatocytes may contain Golgi apparatus with as few as three
cistemae, whereas the Golgi apparatus of goblet cells may possess up to 20 cis-
ternae. The number of cistemae within a given cell may also vary depending
upon the functional condition of the cell. Finally, the plane of section must be
considered when interpreting the number of cistemae within, as well as the ori-
entation of the Golgi stack. This may necessitate examining serial sections in
order to exclude the possibility of missing a particular region of the cisternal
stack in a given section. For instance, Orci et al. (1986) performed a serial sec-
tioning analysis of the transport of horseradish peroxidase from the cell surface
to the Golgi apparatus in insulin-secreting B cells. Their results showed quite
convincingly that what appeared to be a cis- or trans-cisterna in a random sec-
tion could always be traced to a position in the Golgi stack intermediate (i.e.,
middle cistemae) between the cis and trans poles. Taking all of these points into
consideration, and assuming some subjectivity on our part, we propose the fol-
lowing scheme for the localization of sugar residues within the intestinal cell
Golgi apparatus. In absorptive cells, mannose/glucose residues are restricted to
cis and middle portions of the cisternal stack (Pavelka and Ellinger, 1985,
1989a); N-acetylgalactosamine residues are concentrated in cis and trans regions
(Pavelka and Ellinger, 1985; Ellinger and Pavelka, 1988b); galactose residues to
trans- and variably middle cistemae (Pavelka and Ellinger, 1985); fucose
residues to middle/trans regions (Ellinger and Pavelka, 1988a); and sialic acid
residues diffuse throughout the stack, but concentrated in trans-cistemae (Roth
et al., 1986). In goblet cells, mannose/glucose residues are restricted to cis/mid-
dle portions of the stack (Tsuyama et al., 1986); N-acetylgalactosamineresidues
to cis and trans regions (Roth, 1984); galactose residues to middle/trans portions
of the stack (Pavelka and Ellinger, 1985, 1989b); fucose residues to trans-cister-
nae (Ellinger and Pavelka, 1988a); sialic acid residues to trans-cisternae (Roth
er af., 1986); sialic acid a2,3-linked and a2,6-linked to galactose concentrated
in trans-cisternae (Sata et al., 1989; Taatjes and Roth, 1990). In a single study,
Ellinger and Pavelka ( 1988b) have reported that a-galactose residues as de-
tected with GrifSoonia simplicifolia isolectin I-B4, are restricted to cis- cisternae
in intestinal goblet cells.
C. POST-GOLGI
APPARATUS
DISTRIBUTION
OF GLYCOSYLTRANSFERASES:
FACTOR ARTIFACT?
It is well established that glycosyltransferases exist outside of their usual lo-
cation as Golgi apparatus integral membrane proteins; specifically in cellular
plasma membranes and in soluble form in a number of secretions, predomi-
nantly milk and colostrum (Andrews, 1970; Barker er al., 1972; Paulson er al.,
1977), and serum (Hudgin and Schachter, 1971; Fujita-Yamaguchi and Yoshida,
1981; Kaplan el al., 1983). Immunocytochemical methods have indicated the
presence of galactosyltransferase (Pestalozzi et a f . , 1982; Davis et af., 1984;
Roth et al., 1985b; Shaper er al., 1985; Bayna er al., 1988), N-acetylgalac-
tosaminyltransferase (Balsam0 er al., 1986), blood group A N-acetylgalac-
tosaminyltransferase (Roth et al., 1987, 1988c), and sialyltransferase (Roth er
al., 1986; Taatjes and Roth, 1988; Taatjes et al., 1988a) at the plasma membrane
of many cell types. A detailed discussion of cell surface glycosyltransferases is
beyond the scope of this review. However, interested readers should consult sev-
eral excellent reviews of this area (Pierce er al., 1980; Strous, 1986; Shur, 1989).
In this section we will focus on the evidence pertaining to the presence of gly-
cosyltransferases outside of the Golgi apparatus in intestinal cells. Such evi-
dence has been presented from three types of experiments: (1) measurement of
glycosyltransferase activities in plasma membrane fractions; (2) autoradio-
graphic detection of glycosyltransferase activity in plasma membranes; and (3)
in situ immunocytochemical localization of glycosyltransferases.
Many studies have reported the presence of glycosyltransferase activity in the
plasma membranes of intestinal epithelial cells, and these results were already
presented in Sections IV,A, 1 and 2. Briefly, activities for galactosyltransferase
and sialyltransferase have been detected on the apical and basolateral plasma
membranes of intestinal epithelial cells (Weiser, 1973a,b; Weiser er al., 1978).
Using a different methodology, Bennett ei al. (1987) have reported on the exis-
tence of an active sialyltransferase at the microvillar surface of rat intestinal ab-
sorptive cells. They injected CMP-['Hlsialic acid into the intestinal lumen, fol-
lowed by visualization of autoradiographic products at the light microscopic
level. Injection period was restricted to 5 min to ensure that reaction product re-
flected cell surface phenomena and not activity from the Golgi apparatus. They
observed a moderate autoradiographic reaction at the microvillar surface of
small intestinal absorptive cells, yet found no reaction at the luminal surface of
epithelial cells from gallbladder, ciliary body, and iris. Likewise, the injection of
UDP-['H]galactose resulted in no reaction at the cell surface of all these cells,
including intestinal absorptive cells. They attributed these results to reflect the
presence of a sialyltransferase capable of sialylating endogenous acceptors at
the luminal surface of small intestinal absorptive cells. In light of immunocyto-
chemical results to be discussed below, it would have been of interest if Bennett
and co-workers had examined reaction in the large intestine as well.
As mentioned previously, several immunocytochemical investigations at both
the light and electron microscopic levels have reported the presence of im-
munoreactivity for glycosyltransferases at the plasma membrane, as well as
other post-Golgi apparatus sites of intestinal epithelial cells (Pestalozzi et al.,
1982; Roth et al., 1985b. 1986, 1987, 1988~;Taatjes and Roth, 1988; Taatjes er
al., 1988a). Berger and co-workers (Pestalozzi ei al., 1982), observed at the light
microscopic level label with an affinity-purified galactosyltransferase antibody
at the apical, but not basolateral plasma membrane of human jejunal entero-
cytes. Roth ei al., (1985b) performed a similar investigation using postembed-
ding protein A-gold immunocytochemistry at the electron microscopic level.
An affinity-purified antibody against human milk galactosyltransferase was ap-
plied to thin sections from human duodenum embedded in Lowicryl K4M.
Intense gold particle label was observed at the apical (brush border), as well as
basolateral plasma membrane of enterocytes. The intensity of label decreased on
the lateral plasma membrane as it approached the basal membrane. Staining was
completely abolished by preabsorption of the antibody with purified galactosyl-
transferase antigen. However, the validity of these results as representing true
cell surface, or ecto-galactosyltransferase has recently been challenged. First,
Boyle ef al. ( 1986) using analytical subcellular fractionation techniques, re-
ported that galactosyltransferase activity was confined to the Golgi apparatus
fraction in human jejunal biopsy homogenates, with no significant amount de-
tectable in the brush border membrane fraction. They postulated that the staining
observed by Roth et al. (1985b) was probably due to contaminating im-
munoglobulins present in the milk used as the source of the galactosyltrans-
ferase antigen. An alternative explanation for their inability to detect significant
amounts of galactosyltransferase activity in their plasma membrane fraction
could have resulted from failure to block endogenous intestinal nucleotide py-
GLYCOSYLATlON IN INTESTINAL EPITHELIUM 177
rophosphatase. As reported by Lau and Carlson (1981) (and discussed in Section

IV,A, 1), when measuring glycosyltransferase activities in tissues rich in nu-
cleotide pyrophosphatase activity (such as intestinal mucosa), precautions must
be taken, including inclusion of EDTA and soybean trypsin inhibitor, to ensure
that glycosyltransferase degradation does not occur.
A second, and perhaps more troublesome critique, has been the recent revela-
tion that polyclonal antibodies raised against glycoproteins may contain clones
directed against carbohydrate epitopes of the antigen (Feizi and Childs, 1987).
Human milk galactosyltransferase possesses blood group-related carbohydrate
structures as part of its oligosaccharide constituency. Indeed, Feizi and co-work-
ers (Childs et al., 1986) have reported immunochemical data demonstrating that
the affinity-purified antibody raised against the human milk galactosyltrans-
ferase used in the above-mentioned studies (Pestalozzi et al., 1982; Roth et al.,
1985b) contains a minor population of antibodies directed against the blood
group-related carbohydrate moiety of the enzyme. When used in immunofluo-
rescence experiments, these antibodies against carbohydrate epitopes of galacto-
syltransferase intensely stained the brush border of intestinal epithelial cells
(Childs et al., 1986). This staining could be abolished by preabsorbing the anti-
bodies with blood group substances, suggesting that the staining did not reflect
galactosyltransferase immunoreactivity at the brush border, but rather that of
blood group substances. How do these results relate to those published by Roth
et al. (1985b) described above? Perhaps a reevaluation is necessary, employing
a galactosyltransferase antibody preabsorbed with blood group carbohydrate
structures. Alternatively, antibodies could be raised against a deglycosylated
form of the enzyme and used for immunocytochemistry.However, it is not clear
what effect removal of the carbohydrate moieties would have on the folding and
three dimensional conformation of the enzyme. Resulting antibodies could po-
tentially recognize antigenic structures not present on the molecule in situ.
Perhaps the best method for resolving this discrepancy would be to utilize the
recent successful cloning of several galactosyltransferases (Narimatsu et al.,
1986; Shaper et al., 1986, 1988; Masri et al., 1988; Nakazawa et al., 1988;
D’Agostaro et al., 1989; Masibay and Qasba, 1989) to produce polypeptide epi-
tope-purified antibodies (Taatjes et al., 1988a) recognizing only the protein por-
tion of the enzyme as related below.
During our studies on the subcompartmentation of sialyltransferase in the
Golgi apparatus of intestinal epithelial cells, we noted predominant staining over
a variety of post-Golgi apparatus structures, including plasma membrane and
mucus droplets (Roth et al., 1986). In view of the findings of Childs et al. (1986)
concerning the contamination of galactosyltransferase antisera with carbohy-
drate-directed antibodies noted above, we sought to determine if the immunola-
beling we observed for sialyltransferase outside of the Golgi apparatus repre-
sented true sialyltransferase enzyme, or rather was due to nonspecific
cross-reaction with carbohydrate antigens. For this purpose we took advantage

of the recent cloning of the gene for this particular sialyltransferase (Weinstein
er al., 1987) to prepare polypeptide epitope-purified polyclonal sialyltransferase
antibodies (Fig. 17) by adsorption to a recombinant P-galactosidase-sialyltrans-
ferase fusion protein produced in Escherichia coli (Taatjes et al., 1988a).
Because the fusion protein is nonglycosylated, the resulting purified antibodies
recognize only protein epitopes of the sialyltransferase. Using these antibodies
for immunoelectron microscopy, we observed immunoreactivity to the sialyl-
transferase polypeptide in several post-Golgi apparatus structures, in addition to
the Golgi apparatus, in both absorptive and goblet cells from the rat colon
(Taatjes et al., 1988a). In absorptive cells, labeling was found in the apical and
basolateral plasma membranes, lysosomes, and multivesicular bodies, and at the
FIG. 17. Characterization of a P-galactosidase-sialyltransferase fusion protein epitope purified

antibody ( h STI). SDS-polyacrylamide gels of rat liver Golgi apparatus (lanes 2 , 4 , 6 ) and purified
Galpl.4ClcNAc a-2.6 sialyltransferase (lanes I . 3, 5 ) were stained by Coomassie blue (lanes I and
2) or processed as Immun-blot with fusion protein epitope-purified antibody (lanes 3 and 4) or by
antibody mock-purified with P-galactosidase without fused sialyltransferase (h GTI I; lanes 5 and
6). The h ST1 antibody recognizes both the purified and Golgi apparatus forms of sialyltransferase,
while the h GT11 control antibody shows only a background level of staining. (Reproduced with
permission from Weinstein ef a/., 1987.)
GLYCOSYLATION IN INTESTINALEPITHELIUM I79
limiting membrane of apical cytoplasmic vesicles (Fig. 18). In goblet cells (Fig.
19a), label was detected in the apical and basolateral plasma membranes and in
mucus droplets (both in the lumen and at the limiting membrane). Surprisingly,
label was undetectable (including the Golgi apparatus) in all regions of the small
intestine from the same animals (Fig. 19b), as presented in Table IV, and previ-
ously detailed in Section IV,A, 1.
FIG.18. Immunocytochemical localization of sialyltransferase with P-galactosidase-sialyltrans-

ferase fusion protein epitope-purified antibody in an absorptive cell from rat proximal colon.
Sialyltransferase immunoreactivity is detectable in the apical plasma membrane (asterisk) and along
the inner aspect of the limiting membrane of apical cytoplasmic vesicles (arrowheads). X 50,000.
Bar = 0.2 pn. (Reproduced with permission from Taatjes et al., 1988a.)
DOUGLAS J . TAATJES AND JURCEN ROTH
TABLE IV
DISTRIBUTION
OF SIALYLTFWNSFERASES IN RAT INTESTINE'
Sialyltransferase activityb
Intestinal segment GalPl,3(4)GlcNAc a2.3-ST' GalPl,4GlcNAc a2.6-STd
Duodenum 0 0
Jejunum 0 0
Ileum 0 0
Colon 12 14
"From Taatjes et al. ( 1988a).
hActivity expressed as picomole of [14C]Neu5Actransferred per milligram of protein/hour for both
sialyltransferases.
'A value of 0 indicated activity not detected with limit of detection at 10 pmol of ['4C]Neu5Ac/
milligram of proteinhour.
dA value of 0 indicates activity not detected with limit of detection at 1 pmol of [14C]Neu5Ac/
milligram of protein/hour.
Thus, the powerful combination of molecular cloning and immunocytochem-

istry provided very strong support to the contention that glycosyltransferases are
also housed in cellular locations distal to the Golgi apparatus. However, the
question must be asked whether the post-Golgi apparatus localizations of sialyl-
transferase in rat intestine are functionally significant. If we consider first the
label present in the mucus droplets, it is possible that this luminal sialyltrans-
ferase continues its function in the sialylation of glycoproteins. Recently,
Paulson and co-workers (Colley et al., 1989) investigated the conversion of
membrane-bound Golgi apparatus sialyltransferase to a secretory form of the
protein. By replacing the NH,-terminal signal anchor with the cleavable signal
peptide from y-interferon, and transfecting CHO cells with this sialyltransferase
expression vector, they were able to show that this construct was secreted from
the cell with a half time of 2-3 hr. Most importantly, this secreted form of sia-
lyltransferase contained the catalytic portion of the enzyme and was enzymati-
cally active. By analogy, the sialyltransferase located in the mucus may repre-
sent a form of the enzyme rendered soluble by cleavage of the NH,-terminal
signal anchor by an endogenous protease in the trans region of the Golgi appa-
ratus. Indeed, precedence for such a situation has been documented for the con-
FIG. 19. Immunocytochemical demonstration of sialyltransferase in goblet cells using P-galac-

tosidase-sialyltransferase fusion protein epitope-purified antibody. (a) Gold particle label indicative
of sialyltransferase immunoreactivity is found in trans-cistemae of the Golgi apparatus (facing the
mucus droplets), in the mucus droplet lumen (md), and along their limiting membrane (arrowheads),
and along the lateral plasma membrane (arrows) of goblet cells from rat proximal colon. In contrast,
immunoreactivity for sialyltransferase is undetectable in both the Golgi apparatus and mucus
droplets (md) of goblet cells from rat jejunum (b). X 50.000 (a and b). Bar = 0.2 pm. (Reproduced
with permission from Taatjes et al., 1988a.)
version of the blood group A N-acetylgalactosaminyltransferasefrom a mem-

brane-associated to a nonmembrane-associated form in the trans-tubular net-
work of human intestinal goblet cells (Roth et al., 1988~).In this case, active en-
zyme has been directly demonstrated in the mucin released from the goblet cells
(Omtoft et al., 1987). Of course, even if the enzyme is active, continued glyco-
sylation would require the transport of the appropriate nucleotide sugar from the
cytoplasm into the lumen by a nucleotide sugar antiport protein (Hirschberg and
Snider, 1987). Thus, whether or not the sialyltransferase plays a functional role
in the mucus is still an open question. Once it is released with the mucin into the
lumen of the gut, it most likely has no catalytic activity due to lack of substrates,
and is unlikely to play any role in the physical properties of the much since it
would be such a minor component of the total protein (estimated at less than
0.0001% by activity).
The demonstration of sialyltransferase at the apical and basolateral plasma
membrane of intestinal cells provides further evidence for the existence of
ecto-glycosyltransferases. The unambiguous existence of cell surface glyco-
syltransferases has been difficult to establish since Roseman ( 1970) first pro-
posed their role in cell recognition and adhesion. However, in recent years
Shur and co-workers, in a series of elegant studies, have succeeded in demon-
strating the role of cell surface galactosyltransferase in such diverse functions
as fertilization, preimplantation embryonic development, implantation, mes-
enchymal cell migration on substrates, and growth control in normal, neoplas-
tic, and metastatic cells (Shur, 1989). Presently, it is not clear if ecto-sialyl-
transferase plays a functional role in intestinal cells or if its occurrence simply
reflects its less restricted distribution in the post-Golgi apparatus membranes
of these cells. In this respect, it would be of interest to determine if other gly-
cosyltransferases in these cells have similar or different distributions, since
similar distributions would favor the view that their existence on the cell sur-
face is a consequence of an alteration in the underlying mechanism which
would restrict their subcellular localization to the Golgi apparatus. Indeed, this
view is supported by the immunocytochemical localization of the blood group
A N-acetylgalactosaminyltransferase in both the apical and basolateral plasma
membrane of human intestinal epithelial cells (Roth et al., 1987). On the other
hand, Lopez ez a/. (1989) have reported that at least in F9 embryonal carci-
noma cells the levels of cell surface (ecto-galactosyltransferase) and Golgi
apparatus galactosyltransferase change relative to one another during cell dif-
ferentiation, suggesting that these functionally and distinct pools of galacto-
syltransferase are independently and differentially regulated. This would indi-
cate that we should not necessarily think of ecto-glycosyltransferases as
representing nonspecific vesicular transport of the Golgi apparatus form of the
enzyme to the plasma membrane, but rather as an independently regulated en-
tity of its own.
V. Effects of Exogenous Agents on Intestinal Glycosyltransferase

Activity and Glycosylation
A. HORMONES
In adult rats, a continuous subcutaneous administration of testosterone for
fourteen days resulted in qualitative and quantitative changes in the glycosphin-
golipid composition of rat small intestinal mucosa (Dahiya et al., 1989). These
changes were accompanied by increases in the enzymatic activities of CMP-N-
acetylneuraminic acid : lactosyfceramide sialyltransferase and UDP-galactose :
lactosylceramide galactosyltransferase. The authors proposed that testosterone
induced the activities of the two glycosyltransferases to increase, resulting in
changes in intestinal mucosa glycosphingolipid composition. In a similar inves-
tigation, Dudeja et al. (1988) analyzed the activities of the same two glycosyl-
transferases reported above, in Golgi apparatus membranes, in response to sub-
cutaneous administration of the synthetic glucocorticoid dexamethasone. They
found that the activities of both glycosyltransferases were elevated in response
to dexamethasone administration. They speculated that the increase in galacto-
syltransferase activity may have resulted from an increased membrane fluidity
caused by the dexamethasone. However, they could not attribute the increase in
sialyltransferase activity to the same cause.
Several investigations have been aimed at examining the effect of hormone
administration on glycosylation activity in developing intestine. As described in
Sections 1V,A,2 and 3, the postnatal development of rat small intestine is char-
acterized by a decrease in sialyltransferase activity, with a concomitant increase
in fucosyltransferase activity. A postnatal injection of cortisone caused preco-
cious changes in the activities of sialyltransferase and fucosyltransferase in the
mucosal fractions from 2-week-old rats (Chu and Walker, 1986). Specifically,
cortisone administration resulted in a 50% decrease in sialyltransferase activity
and an 8-fold increase in fucosyltransferase activity as compared to control ani-
mals. Likewise, glycosidic-bound sialic acid content was significantly de-
creased, while glycosidic-bound fucose content significantly increased in the
hormone-treated animals. Walker and co-workers (Ozaki et al., 1989) have also
shown that cortisone injection into suckling rats causes a precocious increase in
the activities of two developmentally regulated galactosyltransferases:the UDP-
Gal : GlcNAc (~1-4)-galactosyltransferase was increased 2.7-fold and the UDP-
Gal : GalNAc(~1-3)-galactosyltransferaseactivity was increased 1.8-fold.
In an earlier study, Mahmood and Torres-Pinedo (1985) injected suckling rats
with cortisone, thyroxine, epidermal growth factor, or insulin and measured the
effect on the intestinal microvillar membrane content of sialic acid and fucose,
as well as subsequent lectin binding. Cortisone treatment was found to lower
sialic acid content and raise fucose content of microvillar membranes, as well as
increase the incorporation of ['Hlfucose into these membranes. These results

were also reflected in the binding of '251-labeledlectins to purified microvillar
membrane preparations; cortisone administration decreased the binding of
WGA to microvillar membranes, while increasing the binding of UEL I and
PNL. Thyroxine treatment had a similar effect as cortisone on membranous fu-
cose content and UEL I binding, but did not alter the incorporation of ['Hlfucose
into membranes or the sialic acid content of membranes. Epidermal growth fac-
tor and insulin did not affect any of these parameters. Thus, these results demon-
strated that only cortisone administration to suckling rats induced precocious
changes in sialic acid and fucose content of microvillar membranes normally as-
sociated with postnatal intestinal development.
Kolinska e? al. (1988) examined the effect of hydrocortisone administration
on sialyltransferase activity in the crypts versus villus of 10-day-old rat small in-
testine. They found that the decrease in sialyltransferase activity induced by hy-
drocortisone administration occurred mainly in the crypt cells.
B. DRUGSAND OTHERNoxlOuS STIMULI

Treatment of rats with the microtubule-disrupting drug colchicine or with tur-
pentine results in an increase in the serum level of sialyltransferase activity
(Mookerjea et al.. 1977; Kaplan ez al., 1983). Ratnam et al. (1987) speculated
that some of this increase in serum sialyltransferase may result from secretion
from the small intestine. They injected rats with colchicine and then 4 hr later
measured the activity of the a2,6-sialyltransferase in the homogenates from je-
junal slices. They found that secretion of soluble sialyltransferase into the
medium was elevated in the animals treated with colchicine, as compared to
control animals. A similar increase in intestinal and serum sialyltransferase ac-
tivity has also been shown to be induced by inflammation caused by a standard-
ized 25% body surface area thermal injury in rats (Chu e? al., 1988). These re-
sults thus suggested that intestinal sialyltransferase may form part of the acute
phase response to inflammation. However, we believe that this sialyltransferase
originates from cells of the lamina propria, and not from intestinal epithelial
cells. As pointed out in Sections IV,A,l and IV,C, immunoreactivity and enzy-
matic activity for the a2,6-sialyltransferase were undetectable in rat small in-
testinal mucosa.
Colchicine is normally used as a depolymerizing agent for microtubules to
study cellular processes which may be microtubule-dependent. Indeed, Hugon
e?al. (1987) performed such a study on mouse jejunal epithelial cells to investi-
gate the role of microtubules in the migration of glycoproteins from the Golgi
apparatus to the apical and basolateral plasma membranes. They examined by
autoradiography the incorporation of [3H]fucose into glycoconjugates in ex-
plants of mouse jejunum cultured in a medium containing colchicine. They
GLYCOSYLATIONIN INTESTINAL EPITHELm 185
found that colchicine inhibited the labeling of the brush border by 67%, while
labeling of the basolateral plasma membrane increased 114%. Similar results
were also obtained with the microtubule-disrupting drug nocodazole. These re-
sults suggested that some glycoproteins destined for the apical plasma mem-
brane may be rerouted to the basolateral plasma membrane in the presence of
colchicine and thereby suggests a role for microtubules in the transport of gly-
coconjugates from the Golgi apparatus to the apical plasma membrane in polar-
ized intestinal absorptive cells. Similar effects of colchicine on glycoprotein mi-
gration were reported earlier for human jejunal biopsies in culture (Blok et
af.,1981) and for rat small intestine (Ellinger et af., 1983).
The effect of polyamine deficiency on Golgi apparatus membranes and galac-
tosyltransferase activity in mouse small intestinal epithelial cells was studied by
Sakamaki et al. (1989). They produced polyamine-deficient cells by injecting
two inhibitors of polyamine synthesis, ethylglyoxal bis(guany1hydrazone) and
a-difluoromethylornithine into mice. Polyamine deficiency produced swelling
of the Golgi apparatus membranes (demonstrated by electron microscopy) ac-
companied by a decrease (to approximately 55% of the control value) in galac-
tosyltransferase activity. These results suggested that galactosyltransferase ac-
tivity is diminished in swollen Golgi apparatus membranes.
Umesaki and Ohara (1989) investigated in detail treatments which lead to an
increase in GDP-fucose : asialo GH,a(1-2)fucosyltransferase activity in rat
small intestinal mucosa. The increase in this particular fucosyltransferase activ-
ity was manifested by alteration in the neutral glycolipids of the microvillar
plasma membrane. Factors shown to cause an increase in fucosyltransferase ac-
tivity were microbial contamination of germ-free mice, weaning (see Section
IV,A,2), intraperitoneal injection of the protein synthesis inhibitors cyclohex-
imide or emetine (although repeated injection of cycloheximide every 2 hr re-
sulted in a repression of fucosyltransferase activity), injection of a soluble frac-
tion from a small intestinal homogenate, and mechanical injury to the intestinal
mucosa. They also analyzed the composition of the glycolipids in mucosal frac-
tions after such treatments and found an increase in their fucose content. Finally,
by separating crypt from villus cells, they found that fucosyltransferase activity
was increased in villus cells as compared to crypt cells. They attributed these
findings to indicate that the increase in fucosyltransferase activity in response to
various stimuli is preferentially localized to the postmitotic epithelial cells lo-
cated on the villus.
VI. Differentiation and Glycosylation in Intestinal Cell Culture Systems
Although the morphology and physiology of the intestinal tract are quite
amenable for studying differentiation events (Section I), pitfalls of using such an
organ system in biochemical studies are numerous, i.e., obtaining pure cell pop-
ulations, experimental manipulation of cells, and difficulty in administering ex-
ogenous agents, to name a few. For these reasons, alternative experimental sys-
tems such as intestinal organ (Quaroni, 1985) and cell culture (Rousset, 1986)
have been introduced. The recent introduction of several stable cell lines derived
from intestinal epithelial cells has made this a particularly fruitful avenue of re-
search.
The oligosaccharide composition of cell surface glycopeptides was investi-
gated in confluent and subconfluent cultures of the rat small intestinal epithelial
cell line IEC-6 by measuring the incorporation of D-[2-3H] mannose and by gly-
copeptide sensitivity to various oligosaccharide processing enzymes (Sasak et
al., 1982). They found that confluent cells contained a much higher proportion
of complex oligosaccharides in glycopeptides of the plasma membrane than did
subconfluent cells. Only minor differences were observed between total man-
nose-labeled glycopeptides from confluent and subconfluent cultures, suggest-
ing that the cell surface changes were mainly due to differences in biosynthesis
of the carbohydrate moieties and not to the formation of different glycoproteins.
Moreover, this alteration in oligosaccharide composition of cell surface gly-
copeptides was shown to be dependent upon cell density and not on the growth
rate of the cells. Interestingly, Sasak et al. (1982) also were able to draw a cor-
relation between degree of cell adhesion to the substratum and cell surface
oligosaccharides: confluent cultures containing cell surface glycopeptides with
complex-type oligosaccharide structures were more adherent than their subcon-
fluent counterparts displaying more high mannose-type oligosaccharides.
Several recent reports have documented the relationship between cell differ-
entiation and the extent of processing of N-linked oligosaccharides in the human
colon cancer cell line HT-29 (Trugnan et a / . , 1987; Ogier-Denis et al., 1988,
1989). HT-29 cells remain undifferentiated in media containing glucose, but un-
dergo differentiation when glucose is removed from the media. Trugnan er al.
( 1987) examined the biosynthesis of sucrase-isomaltase, a microvillar mem-
brane protein taken as a marker for differentiated intestinal epithelial cells in
v i w , in both differentiated and undifferentiated HT-29 cells. In contrast to the
normal processing and expression of this enzyme at the cell surface in differen-
tiated HT-29 cells, in undifferentiated cells no enzyme was detectable at the
plasma membrane. They showed that the failure to detect membrane expression
was not due to lack of synthesis, but rather to abnormal posttranslational pro-
cessing. Indeed, as compared to the enzyme synthesized and expressed in differ-
entiated HT-29 cells, sucrase-isomaltase produced in the undifferentiated cells
displayed ( I ) an impairment of the conversion from high mannose to complex
form of the enzyme; (2) abnormal complex form glycosylation; and (3) rapid in-
tracellular degradation of both high mannose-type and complex-type enzymes.
In a subsequent paper (Ogier-Denis et al., 1988), this group investigated
whether the impairment in glycosylation noted in undifferentiated HT-29 cells

was specific for sucrase-isomaltase, or a general glycosylation defect. They
found that there is an overall defect in the processing of N-linked oligosaccha-
rides (Section 11) which is manifested by alterations in three processing steps:
( 1) incorporation of ~-[2-~H]mannose into glycoproteins; (2) conversion of
high-mannose chains to complex-type N-linked glycans; and (3) trimming of
high-mannose chains at the level of conversion of Man,,-GlcNAc,-Asn to
Man,-GlcNAc,-Asn. This particular trimming reaction was elaborated on in an-
other report (Ogier-Denis et al., 1989), where it was suggested that it may rep-
resent an important regulatory point in the conversion of undifferentiated to dif-
ferentiated cells. These results, therefore, suggest that there is an impairment in
the conversion of high mannose forms of N-linked oligosaccharides into com-
plex-type in undifferentiated HT-29 cells. Future investigations on these cul-
tured intestinal cells should help to unravel in more detail the cellular mecha-
nisms involved in terminal differentiation as it relates to glycosylation.
VII. Concluding Remarks
Although much effort has been directed toward elucidating glycosylation

mechanisms and patterns in intestinal cells, unequivocal answers have not been
forthcoming. This lack of emergence of a unifying concept underlying intestinal
cell glycosylation may be the result of many divergent factors. Conflicting results
concerning the activities of glycosyltransferasesin intestinal homogenates almost
certainly results from variation in methodologies; i.e., (1) mucosal scrapings rep-
resenting mostly epithelial cells versus homogenates containing submucosa and
lamina propria; (2) failure to allow for endogenous intestinal enzymes which
could potentially degrade glycosyltransferases; (3) variation in acceptor sub-
strates employed, resulting in the measurement of different glycosyltransferases
within the same class; and (4)different techniques for the separation of crypt ver-
sus villus epithelial cells. Similar technique-related problems could explain the
variation in expression of intestinal carbohydrates. However, this is more likely
due to the inherent variability in glycosylation expressed in a given cell type.
Many detailed investigations have revealed a marked degree of glycoconjugate
heterogeneity, not only among similar cell types from different species
(Holthofer, 1983; Schulte and Spicer, 1983a,b; Spicer er al., 1987), but also
among supposedly homogeneous cell populations within a given organ (Watanbe
er al., 1981; Spicer et al., 1981; LeHir er al., 1982; Roth er al., 1983; Brown et
al., 1985; Roth and Taatjes, 1985; Roth er al., 1988b; Taatjes et al., 1988b). Such
heterogeneity may reflect blood group specificities, environmental or genetic
variation, differentiation state of the cell, or pathological influences. However, a
rapidly emerging concept suggests that the glycoconjugate repertoire displayed
188 DOUGLAS J. TAATJES AND mRGEN ROTH
by a given cell reflects its endogenous expression of glycosyltransferases. This

concept has recently been discussed in detail by Paulson (1989; Paulson et al.,
1989) and by Rademacher et al. (1988), and will not be elaborated on here. Given
the role played by terminal oligosaccharide structures in cell-cell recognition
phenomena (Section I), the expression of glycosyltransferases would appear to
occupy a key position in the posttranslational processing of glycoconjugates and
thus influence cellular functions. Does this then mean that the carbohydrate por-
tion of all types of glycoconjugates is important for their biological functioning?
Certainly this is not the case for all glycoconjugates, and is an important area of
concern in biotechnology. The importance of glycosylation in intestinal systems
is mostly unknown at this point, although the well-documented shift from sialy-
lation to fucosylation during rat postnatal development (Sections IV,A,2,4, and 6)
has been attributed to represent the change in physiological functioning of the in-
testine during the weaning phase (Torres-Pinedo and Mahmood, 1984). It seems
probable that the application of cDNA probes for various glycosyltransferases to
intestinal systems (Paulson er a/., 1989) as well as the development of chimeric
and transgenic mice (Gordon, 1989; Trahair et a/., 1989) will provide exciting
opportunities in the future for the investigation of the importance of intestinal
glycosylation in a myriad of functions.
ACKNOWLEDGMENTS
The original research described in this paper has received generous continual support from the
Swiss National Science Foundation. We would like to thank Daniel Wey, Michele von Turkovitch,
and Linda Barcornb for preparing the figures and photographs.
REFERENCES
Abeijon. C.. and Hirschberg. C. B. (1987).J . Biol. Chem. 262,41534159.
Andrews, P. (1970). FEBS Lett. 9.297-300.
Aub, 1. C.. Tieslau. C., and Lankester, A. (1963). Proc. Natl. Acad. Sci. U.S.A. 50, 613-619.
Aub, J . C., Sanford. B. H., and Cote, M. N. (1965a). Proc. Nut/. Acad. Sci. U S A . 54,396-399.
Aub. J . C., Sanford. B. H., and Wang. L. (1965b). Proc. Natl. Acad. Sci. U S A . 5 4 , 4 W 0 2 .
Balsarno, J . . Prati. R. S., Ernmerling, M.R., Grunwald, G. B.. and Lilien, J. (1986). J . Cell Biochetn.
32, 135-141.
Barker. R., Olsen. K. W., Shaper, J. H.. and Hill, R. L. (1972). J. Biol. Chem. 247.7 135-7147.
Bayna. E. M.,Shaper. J. H.. and Shur, B. D. (1988).Cell 53, 145-157.
Bendayan. M. (1984). J. Electron Microsc. Tech. 1, 243-270.
Bennett, G . , Haddad. A., and Leblond, C. P. (1987). J. Hisrochem. Cytochem. 35,861-864.
Berger. E. G.. and Hesford, F. (1985). Proc. Natl. Acad. Sci. U S A . 82,473-739.
Berger, E. G., Thumher, M.. and Muller, U. (1987). E.rp. Cell Rex. 173,267-273.
Berlin, J . D. (1967). J. Cell B i d . 32, 76&766.
Blok, J.. Ginsei, L. A., Mulder-Stapel, A. A.. Ondewater, J. J. M., and Daems, W. T. (1981). Cell
Tissue Res. 215. 1-12.
Boland. C. R., and Kim, Y. S. (1984). I n "Markers of Colonic Cell Differentiation" (S. R. Wolrnan
and A. J. Mastromarino, eds.), pp. 253-266. Raven, New York.
Boyle, F.,Snape, S., Duane, P., Cook, N., and Peters, T. (1986). Biosci. Rep. 6, 171-175.
Bresalier, R. S., Boland, C. R., and Kim, Y.S. (1985). J. Nail. Cancer Inst. 75,249-259.
Brown, D., Roth, J., and Orci, L. (1985). Am. J. Physiol. 248, C348-C356.
Buller, H. A., Rings, E. H. H. M., Pajkrt, D., Montgomery, R. K., and Grand, R. J. (1990).
Gastroenterology 98,667475.
Burger, M. M., and Goldberg, A. R. (1967). Proc. Natl. Acad. Sci. USA. 57, 359-366.
Caldero, J., Campo, E., Calomarde, X.,and Torra, M. (1988). Histochemistry 90, 261-270.
Carlemalm, E., Garavito, M., and Villiger, W. (1982). J . Microsc. (Oxford) 126, 123-143.
Chang, W. W. L., and Leblond, C. P. (1971). Am. J. A m . 131,73-100.
Cheng, H. (1974). Am. J. Anat. 141,481-502.
Cheng, H., and Leblond, C. P. (1974). Am. J. Anat. 141,537-562.
Childs, R. A., Berger, E. G., Thorpe, S. J., Aegerter, E., and Feizi, T. (1986). Biochem. J. 238,
605411.
Chu, S.-h. W., and Walker, W. A. (1986). Biochim. Biophys. Acta 883,496-500.
Chu, S.-h. W., Carter, E. A., Tompkins, R. G., and Burke, J. F. (1988). Biochem. Biophys. Res.
Commun. 153,377-381.
Colley, K. J., Lee, E. U., Adler, B., Browne, J. K., and Paulson, J. C. (1989). J. Biol. Chem. 264,
17619-1 7622.
D’Agostaro, G . , Bendiak, B., and Tropak, M. (1989). Eur. J . Biochem. 183,211-217.
Dahiya, R., Ahlawat, R. S., and Sharma, A. (1989). Biochem. Cell Biol. 67,4247.
Davis, B. W., Berger, E. G., Locher, G. W., Zeller, M., and Goldhirsch, A. (1984). J. Histochem.
Cytochem. 32,92-96.
Deschuyteneer, M., Eckhardt, A. E., Roth, J., and Hill, R. L. (1988). J. Biol.Chem. 263,2452-2459.
Dudeja, P. K., Dahiya, R., Brown, M. D., and Brasitus, T. A. (1988). Biochem. J. 253,401-408.
Dunphy, W. G., and Rothman, J. E. (1983). J. Cell Biol. 97,270-275.
Dunphy, W. G., and Rothman, J. E. (1985). Cell 42, 13-21.
Dunphy, W. G., Fries, E., Urbani, L. J., and Rothman, J. E. (1981). Proc. Natl. Acad. Sci. U.SA. 78,
7453-7457.
Dunphy, W. G., Brands, R., and Rothman, J. E. (1985). Cell 40,463472.
Eastwood, G. L. (1977). Gastroenterology 72,962-975.
Egea, G., Goldstein, I. J., and Roth, J. (1989). Histochemistry 92, 515-522.
Elhammer, A., and Komfeld, S. (1984). J . Cell Biol. 98, 327-331.
Ellinger, A., and Pavelka, M. (1985). Histochem. J. 17, 1321-1336.
Ellinger, A., and Pavelka, M. (1988a). Eur. J. Cell Biol. 47, 62-71.
Ellinger, A., and Pavelka, M. (1988b).In “Lectins: Biology, Biochemistry, Clinical Biochemistry” (T. C.
Bog-Hansen and D. L. J. Freed, eds.), Vol. 6, pp. 709-714. Sigma Chem. Co., St. Louis, Missouri.
Ellinger, A., Pavelka, M., and Gangl, A. (1983). J. Ultrastruct. Res. 85, 260-271.
Essner, E., Schreiber, J., and Griewski, R. A. (1978). J. Histochem. Cytochem. 26,452-458.
Etzler, M. E., and Branstrator, M. L. (1974). J. Cell Biol. 62,329-343.
Etzler, M. E., and Branstrator, M. L. (1979). Ciba Found. Symp. No.70.51-68.
Farquhar, M.G. (1985). Annu. Rev. Cell Biol. 1,447488.
Fauk, W. P., and Taylor, G. M. (1971). Immunochemistry 8, 1081-1083.
Feizi, T., and Childs, R. A. (1987). Biochem. J. 245, 1-1 1.
Freeman, F. J., Kim, Y., and Kim, Y. S. (1978). Cancer Res. 28,3385-3390.
Fujita-Yamaguchi, Y., and Yoshida, A. (1981). J. Biol. Chem. 256,2701-2706.
Geuze, H. J., Slot, J. W., Strous, G. J. A. M., Hasilik, A., and von Figura, K. (1985). J. Cell Biol.
101,2253-2262.
Goldberg, D. E., and Komfeld, S. (1983). J. Biol. Chem. 258,3159-3165.
Goldstein, I. J., and Poretz, R. D. (1986). In “The Lectins: Properties, Functions, and Applications
in Biology and Medicine” (I. E. Liener, N. Sharon, and 1. J. Goldstein, eds.), pp. 33-248.
Academic Press, Orlando, Florida.
Gordon, J. 1. (1989).J . Cell Biol. 108, 1187-1 194.

Gorelick, F. S.. Sarras. M. P.. and Jamieson, J. D. (1982). J . Hisrochem. Cyrochern. 30, 1097-1 108.
Green, E. D.. Adelt, G., Baenziger, J. U.,Wilson, S., and Van Halbeek, H. (1988). J . B i d . Chem.
263, 18253-18268.
Grifiths, G., and Simons, K. (1986). Science 2 3 4 , 4 3 8 4 3 .
Hagopian, A,, and Eylar, E. H. (1968). Arch. Biochem. Biophys. 128,422433.
Hagopian, H. B.. Bosman, B. H., and Eylar, E. H. (1968). Arch. Biochem. Biophys. 128,387-396.
Hanover. J. A,. and Lennarz, W. J. (1981). Arch. Biochem. Brophys. 211, 1-19.
Hanover, J. A,, Lennarz. W. J., and Young. 1. D. (1980). 1.B i d . Chem. 255,67 1 3 4 7 16.
Hart, G. W., Holt, G. D., and Haltiwanger, R. S. (1988). Trends Biochem. Sri. 13,380-384.
Hart. G. W.. Haltiwanger. R. S., Holt, G. D., and Kelley. W. G. (1989). Annu. Rev. Biochem. 58,
841-874.
Hedman. K., Pastan, I., and Willingham, M. C. (1986). J . Hisrochem. Cytnchern. 34, 1069-1077.
Herscovics, A.. Bugge, B., Quaroni, A,, and Kirsch, K. (1980). Biochem. J . 192, 145-153.
Hirschberg. C. B., and Snider, M. D. (1987). Annu. Rev. Biochem. 56,63-87.
Holthofer, H. (1983).J. Hisrochem. Cgtochem. 31, 531-537.
Hubbard, S . C., and Ivatt, R. J. (1981). Annu. Rev. Biochem. 50,555-583.
Hudgin, R. L.. and Schachter, H. ( 1971). Con. J . Biochem. 49, 829-837.
Hugon, J. S.. Bennett, G.. Pothier, P., and Ngoma. Z. (1987). Cell Tissue Res. 248,653662.
Jaswal. V. M. S., Babbar. H. S., and Mahmood, A. (1988). Biochem. Med. Merhods B i d . 39,
105-1 10.
Kaplan, H. A., Woloski, B. M. R. N. J., Hellman, M.. and Jamieson, J. C. (1983).J. Biol. Chem. 258,
I 1505-1 1509.
Kim. Y. S. (1989a).In “Modem Concepts in Gastroenterology” (E. Shaffer. N. Williams, and A. B.
R. Thomson, eds.), Vol. 2, pp. 197-208. Raven. New York.
Kim. Y. S. (l989b). Gasrroenrerul. Inr. 2, 101-106.
Kim. Y.S.. and Perdomo, J. M. (1974). FEBS Lett. 44, 309-312.
Kim. Y. S.. Perdomo. J., and Nordberg. J. ( 1971).J . B i d . Chem. 246,5466-5476.
Kim. Y. S., Perdomo, J.. Ochoa, P., and Isaacs, R. A. (1975). Biochim. Biophys. Acra 391, 39-50.
Kocourek, J. (1986). I n “The Lectins: Properties, Functions, and Applications in Biology and
Medicine” (I. E. Liener. N. Sharon, and I. J. Goldstein, eds.), pp. 1-32. Academic Press,
Orlando. Florida.
Kolinska. J.. Ivanov, S., and Chelibonova-Lorer, H. (1988). FEBS Letr. 242, 57-60,
Komfeld. R., and Komfeld, S. (1985).Annu. Rev. Biochem. 54,63 1 4 6 4 .
LaMont. J. T.. and Ventola, A. (1978). Am. J. Physiol. 235, E213-E217.
Lau. J. T. Y., andCarlson. D. M. (1981).J.Biol. Chem. 256,7142-7145.
Leblond. C. P.,and Messier, 8 . (1958). Anar. Rec. 132, 247-259.
Lee. E. U.. Roth. J.. and Paulson, J. C. (1989). J. B i d . Chem. 264, 13848-13855.
LeHir, M.. Kaissling. B., Koeppen, B. M., and Wade, J. B. (1982). Am. J . Phgsiol. 242. CI 17-Cl29.
Lopez. L. C . . Maillet. C. M.. Oleszkowicz. K., and Shur. B. D. (1989). Mol. Cell. B i d . 9,
237G-2377.
Lucocq, J. M.. Brada. D., and Roth, J. (1986).1.Cell B i d . 102, 2137-2146.
Lucocq, 1. M., Berger. E. G.. and Roth, J. (1987).J. Hisrochem. Cytochem. 35, 67-74.
Mahmood, A.. and Torres-Pinedo, R. (1983). Biochem. Biophys. Res. Commun. 113,400-406.
Mahmood, A.. and Torres-Pinedo, R. (1985). Pediorr. Res. 19,899-902.
Masibay. A. S.. and Qasba, P. K. (1989). Proc. Not/. Acod. Sci. U.S.A. 86,5733-5737.
Masri. K. A., Appert. H. E.. and Fukuda. M. N. (1988). Biochem. Biophys. Res. Commun. 157,
657463.
Mookerjea, S.. Marshall, J. W.. and Collins, J. M. (1977). Biocheni. Biophys. Res. Commun. 78,
309-3 16.
Moremen, K., and Touster, 0. (1988). In “Protein Transfer and Organelle Biosynthesis” (R. C. Das
and P. W. Robbins, eds.), pp. 209-240. Academic Press, San Diego, California.
Morita, A., Miura, S., Erickson, R. H., Sleisenger, M. H., and Kim, Y. S. (1986). Biochim. Biophys.
Acra 883,506-5 16.
Murata, F., Suganuma, T., Tsuyama, S., Ishida, K., and Funasako, S. (1986). J. Elecrron Microsc. 35,
29-37.
Nakazawa, K., Ando, T.,Kimura, T., and Narimatsu, H. (1988). Biochem. Biophys. Res. Commun.
104, 165-168.
Narimatsu, H., Sinha, S., Brew, K., Okayama, H., and Qasba, P. K. (1986). Proc. Narl. Acad. Sci.
U.S.A. 83,4720-4724.
Neefjes, J. J., Verkerk, J. M. H., Broxterman, H. J. G., van der Marel, G. A., van Boom, J. H., and
Ploegh, H. L. (1988).J. Cell Biol. 107,79-87.
Neutra, M., and Leblond, C. P.(1966a).J . Cell Biol. 30, 119-136.
Neutra, M., and Leblond, C. P. (1966b).J. Cell Biol. 30, 137-150.
Ogier-Denis, E., Codogno, P., Chantret, I., and Trugnan, G. (1988). J . Biol. Chem. 263,6031-6037.
Ogier-Denis, E., Bauvy, C., Aubely, M., Codogno, P., Sapin, C., Rousset, M., Zweibaum, A., and
Trugnan, G. (1989). J . Cell. Biochem. 41, 13-23.
Orci, L., Ravazzola, M., Amherdt, M., Brown, D., and Perrelet, A. (1986). EMBO J. 5, 2097-2101.
Orntoft, T. F., Greenwell, P., and Watkins, W. M. (1987). Glycoconjugates, Proc. Int. Symp. 9rh.
Lille, Fr. Abstr. E 108.
Ovtscharoff, W., and Ichev, K. (1984).Acra Hisrochem. 74, 21-24.
Ozaki, C. K., Chu, S.-h. W., and Walker, W. A. (1989). Biochim. Biophys. Acra 991,243-247.
Paulson, J. C. (1985). In “The Receptors” (P. M. Conn, ed.), Vol. 2, pp. 131-219. Academic Press,
New York.
Paulson, J. C. (1989). Trends Biochem. Sci. 14,272-276.
Paulson, J. C., Beranek, W. E., and Hill, R. L. (1977).J. Biol. Chem. 252,2356-2362.
Paulson, J. C., Weinstein, J., and Schauer, A. (1989).J. Biol. Chem. 264, 10931-10934.
Pavelka, M. (1987).Adv. Anar. Embryo/. Cell Biol. 106, 1-94.
Pavelka, M., and Ellinger, A. (1985). J . Hisrochem. Cyrochem. 33,905-914.
Pavelka, M., and Ellinger, A. (1986). Eur. J. Cell Biol. 41,27&278.
Pavelka, M., and Ellinger, A. (1989a).J . Hisrochem. Cyrochem. 37,877-884.
Pavelka, M., and Ellinger, A. (1989b). In “Electron Microscopy of Subcellular Dynamics” (H.
Plattner, ed.), pp. 199-218. CRC Press, Boca Raton, Florida.
Pestalozzi, D. M., Hess, M., and Berger, E. G. (1982). J . Hisrochem. Cyrochem. 30, 1146-1 152.
Pierce, M., Turley, E. A., and Roth, S. (1980).Int. Rev. Cyrol. 65, 1 4 7 .
Potten, C. S., and Loeffler, M. (1987). J. Theor. Biol. 127, 381-391.
Quaroni, A. (1985). J . Cell Biol. 100, 1611-1622.
Quaroni, A., Kirsch, K., Herscovics, A., and Isselbacher, K. J. (1980). Biochem. J . 192, 133-144.
Rademacher, T. W., Parekh, R. B., and Dwek, R. A. (1988). Annu. Rev. Biochem. 57,785-838.
Rambourg, A., and Clermont, Y.(1990). Eur. J . Cell Biol. 51, 189-200.
Rambourg, A., Hernandez, W., and Leblond, C. P. (1969). J. Cell Biol. 40,395414.
Rampal, P. J., LaMont, J. T., and Trier, J. S. (1978). Am. J . Physiol. 235, E207-E212.
Ratnam, S., Nagpukar, A., and Mookerjea, S. (1987). Biochem. Cell Biol. 65, 183-187.
Reichner, J. S., Whiteheart, S. W., and Hart, G. W. (1988).J . Biol. Chem. 263, 16316-16326.
Revel, J. P. (1964). J. Microsc. (Paris)3,535-544.
Roseman, S . (1970). Chem. Phys. Lipids 5,270-297.
Roth, J. (1978). Exp. Parhol., Suppl. No.3. Fischer, Jena, Germany.
Roth, J. (1983a).In “Techniques in Immunocytochemistry” (G. R. Bullock and P. Petrusz, eds.), Vol.
2, pp. 217-284. Academic Press, London.
Roth, J. (1983b). J. Hisrochem. Cyrochem. 31,987-999.
Roth, J. (1984).J. Cell Biol. 98, 399406.

Roth, J. ( I 986). J . Microsc. (OAjord) 143, 125-1 37.
Roth, J. ( 1987a).Biochim. Biophys. Acra 906,405436.
Roth, J. (1987b).Scunninx Microsc. 1,695-704.
Roth, J. ( 1989).Methods Cell Biol. 31, 5 13-55 1.
Roth. J.. and Berger, E. G. (1982). J . Cell Biol. 92,223-229.
Roth, J., and Taatjes. D. J. (1985).Eur. J. Cell B i d . 39,449-457.
Roth, J.. and Taatjes, D. J. (1989). in “Electron Microscopy of Subcellular Dynamics” (H. Plattner,
ed.), pp. 2 19-237. CRC Press, Boca Raton, Florida.
Roth, J.. Bendayan, M.. and Orci, L. (1978).J. Histoehem. Cytochem. 26, 1074-1081.
Rorh, J., Brown. D., and Orci, L. (1983).J . CelI Biol. 96, 1189-1 196.
Roth, J.. Lucocq, J. M., and Charest. P. M. (1984).J . Histochem. Cytochem. 32, 1167-1 176.
Roth, J., Taatjes. D. J., Lucocq, J. M., Weinstein, J., and Paulson, J. C. (1985a).Cell 43,287-295.
Roth, J.. Lentze, M. J., and Berger, E. G. (1985b).J . Cell Biol. 100, 118-125.
Roth, J.. Taatjes, D. J., Weinstein, J., Paulson, J. C., Greenwell, P.,and Watkins, W. M. (1986).J .
Biol. Chem. 261, 14307-14312.
Roth, 1.. Greenwell, P.. and Watkins, W. M. (1987).Biochem. Soc. Trans. 15, 599-601.
Roth, J.. Lucocq, J. M.. and Taatjes, D. J. (1988a).Arm Histoehem.. Suppl. 36, 81-99.
Roth. 1.. Taatjes. D. J., Lucocq. J. M., and Charest, P. M. (l988b). Acta Hisrochem.. Suppl. 36,
125-140.
Roth. J., Greenwell, P., and Watkins, W. M. (1988~). Eur. J . Cell B i d . 46, 105-1 12.
Rousset, M. (1986).Biochimie 68, 1035-1040.
Sakamaki. Y.. Terao. K., 110, K.. Kashiwagi, K., and Igarashi, K. (1989).Biochem. Phurmacol. 38,
1083-1 089.
Saraste. J.. and Kuismanen. E. (1984).Cell 38,535-549.
Sdsak, W.. Herscovics, A., and Quaroni, A. (1982). Biochem. J . 201,359-366.
Sata. T.. Lackie, P. M., Taatjes, D. J., Peurnans. W., and Roth, J. (1989).J. Hixtochem. Cyrochem. 37,
1577-1588.
Schulte, B. A.. and Spicer, S. S. (1983a).J . Hisrochem. Cytoehem. 31, 391403.
Schulte. B. A,, and Spicer, S. S. (l983b). Am. J . Anut. 168,345-362.
Shaper, N. L.. Mann. P. L., and Shaper. J. H. (1985).J . Cell. Biochem. 28,229-239.
Shaper. N. L.. Shaper. J. H.. Meuth. J. L., Fox, J. L., Chang, H., Kirsch, I., and Hollis, G. (1986).
Proc. Natl. Acad. Set. U.S.A. 83, 1573-1577.
Shaper, N. L., Hollis. G. F.. Douglas. J. G.. Kirsch, 1. R.. and Shaper. J. H. (1988)..I. Biol. Chem.
263, 10420-10428.
Sharon, N.. and Lis. H. (1989).Science 246, 227-234.
Shur, B. D. (1989).Biochim. Biophys. Acta 988,389409.
Singer, S . J., and Nicolson. G. L. (1972).Science 175,720-731.
Snider, M. D.. and Rogers, 0. C. (1985).J . Cell Biol. 100,826434,
Snider. M. C . , and Rogers. 0. C. (1986).J. Cell Biol. 103, 265-275.
Spicer, S. S.. Baron, D. A.. Sato. A.. and Schulte, B. A. (1981). J . Histochem. Cytochem. 29,
994-1002.
Spicer. S. S., Erlandsen. S. L., Wilson, A. C.. Hammer, M. F.. Hennigar. R. A., and Schulte, B. A.
(1987).J. Hisrochem. Cytoehem. 35, 1231-1244.
Srivastava.0. P.. Steele. M. I., and Torres-Pinedo, R. (1987).Biochim. Binphys. Acta 914, 143-151.
Strous, G. J. A. M. (1979). Pmc. Natl. Acad. Sci. U.S.A. 76,2691-2698.
Strous,G. 1. A . M . 0986).CRCCrir. Rev. Biochem. 21, 119-151.
Taatjes, D. I., and Roth. J. (1986).Eur. J. Cell Bin/. 42, 344350.
Taatjes, D. J.. and Roth, J. (1988). Eur. J . Cell Biol. 46, 289-298.
Taatjes, D. J.. and Roth, J. (1990). Eur. J . Cell B i d . 53, 255-266.
Taatjes, D. J., Roth, J., Weinstein, J., Paulson, J. C., Shaper, N. L., and Shaper, J. H. (1987). Eur. J .
Cell Biol. 44, 187-1 94.
Taatjes. D. J., Roth, J., Weinstein, J., and Paulson, J. C. (1988a).J. Biol. Chem. 263,63024309.
Taatjes, D. J., Roth, J., Peumans, W., and Goldstein, I. J. (1988b). Histochem. J. 20,478-490.
Thorens, B., and Vassalli, P. (1986). Nature (London) 321,618420,
Tooze, S. A., Tooze, J., and Warren, G. (1988). J. Cell Biol. 106, 1475-1487.
Torres, C.-R., and Hart, G. W. (1984).J.Biol.Chem. 261,3308-3317.
Torres-Pinedo, R., and Mahmood, A. (1984). Biochem. Biophys. Res. Commun. 125,546553.
Trahair, J. F., Neutra, M. R., and Gordon, J. 1. (1989). J . Cell B i d . 109, 3231-3242.
Trugnan, G., Rousset, M., Chantret, I., Barbat, A., and Zweibaum, A. (1987). J. Cell Biol. 104,
1199-1205.
Tsuyama, S., Suganuma, T., Ihida, K., and Murata, F. (1986). Acru Histochem. Cyrochem. 19,
555-565.
Ulrich, J. T., Schenck, J. R., Rittenhouse, H. G., Shaper, N. L., and Shaper, J. H. (1986). J . Biol.
Chem. 261,7975-7981.
Umesaki, Y., and Ohara, M. (1989). Biochim. Biophys. Actu 1001, 163-168.
Van Halbeek, H., Genvig, G. J., Vliegenthart, J. F. G., Smits, H. L., van Kerkof, P. J. M., and
Kramer, J. F. (1983). Biochim. Biophys. Acfu 737, 107-116.
Watanabe, M., Muramatsu, T., Shirane, H., and Ugai, K. (1981). J. Hisrochem. Cytochem. 29,
779-790.
Weinstein, J., de Souza-e-Silva, U., and Paulson, J. C. (1982).J . Biol. Chem. 257, 13835-13844.
Weinstein, J., Lee, E. U., McEntee, K., Lai, P., and Paulson, J. C. (1987). J. Biol. Chem. 262,
17735-1 7743.
Weiser, M. M. (1973a).J . Biol. Chem. 248,25362541.
Weiser, M. M. (1973b). J. Biol. Chem. 248,2542-2548.
Weiser, M. M., Neumeier, M. M., Quaroni, A., and Kirsch, K. (1978). J. Cell Biof. 77,722-734.
Weiser, M. M., Walters, J. R. F., and Wilson, J. R. (1986). In?.Rev. Cyrol. 101, 1-57.
Weiser, M. M., Majumdar, S., Wilson, J. R., and Luther, R. (1987). Biochim. Biophys. Acru 924,
323-33 1.
Wetzel, M. G., Wetzel, B. K., and Spicer, S. S. (1966). J . Cell Biol. 30, 299-315.
Wilson, J. R., Deinhart, J. A., and Weiser, M. M. (1987). Biochim. Biophys. Acru 924,332-340.
Yokoyama, M., Nishiyama, F., Kawai, N., and Hirano, H. (1980). Exp. Cell Res. 125,47-53.

main

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

main

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

main

Uploaded by

Copyright:

Available Formats

Since January 2020 Elsevier has created a COVID-19 resource centre with

free information in English and Mandarin on the novel coronavirus COVID-

Elsevier hereby grants permission to make all its COVID-19-related

Glycosylation in Intestinal Epithelium

*Present Address: Department of Pathology, University of Vermont, Burlington, Vermont 05405

The intestinal epithelium represents a good model system in which to study

absorptive (columnar), goblet, and endocrine cells. The mechanisms guiding

variability of the oligosaccharide structures of glycans in different segments of

11. Overview of Glycosylation

'Abbreviations: Fuc, fucose; Gal, galactose; GalNAc, N-acetylgalactosamine; Glc, glucose;

The second major class of glycoproteins are the so-called 0- or mucin-type

The mechanisms of N-linked glycosylation reactions have been worked out in

extent of processing is dictated by whether the oligosaccharide chain on the gly-

cessing of complex-type oligosaccharides is more extensive and entails the re-

immuno- and cytochemical means to investigate the site of addition of N-acetyl-

the 0-glycosylation of the El glycoprotein of coronavirus MHV-A59 using a

111. Methods Employed to Investigate Cellular Glycosylation

Investigations into intestinal glycosylation have mainly drawn upon biochem-

Lectin Common name Abbreviation Nominal specificity"

with the publication of the first immunocytochemical localization of a glycosyl-

IV. Distribution of Intestinal Glycosyltransferases and

A. STUDIES ON WHOLE TISSUE

1 . Measurement of Glyco.$yltransferase Activities in Adult Animals

advocated exercising caution when interpreting the measurement of intestinal

different enzymes. However, due to antibody specificity, immunological-based

The second biochemical method, utilizing lectin binding to isolated cell

At the same time, the binding of Iz5I-labeledUlex europaeus lectin I (UEL I) to

glycosyltransferase enzymatic measurements, lectin-binding sites can be taken

fully differentiated colonic epithelial cells is the polarization of the plasma

functionally subcompartmentalized with respect to the steps involved in the pro-

functionally it is involved in sialylation as are trans-cisternae, and structurally it

2 . Immunoc:\.tochemicalLocalization of Glycosyltransferases in Golgi

FIG.14. Immunocytochemical localization of sialyltransferasein surface epithelial cells from rat

FIG.16. Immunocytochemical localization of blood group A al.3-N-acetylgalactosaminyltrans-

charides, it is not clear that normal processing requires such subcompartmenta-

rophosphatase. As reported by Lau and Carlson (1981) (and discussed in Section

cross-reaction with carbohydrate antigens. For this purpose we took advantage

FIG. 17. Characterization of a P-galactosidase-sialyltransferase fusion protein epitope purified

FIG.18. Immunocytochemical localization of sialyltransferase with P-galactosidase-sialyltrans-

Thus, the powerful combination of molecular cloning and immunocytochem-

FIG. 19. Immunocytochemical demonstration of sialyltransferase in goblet cells using P-galac-

version of the blood group A N-acetylgalactosaminyltransferasefrom a mem-

V. Effects of Exogenous Agents on Intestinal Glycosyltransferase

increase the incorporation of ['Hlfucose into these membranes. These results

B. DRUGSAND OTHERNoxlOuS STIMULI

VI. Differentiation and Glycosylation in Intestinal Cell Culture Systems

whether the impairment in glycosylation noted in undifferentiated HT-29 cells

VII. Concluding Remarks

Although much effort has been directed toward elucidating glycosylation

by a given cell reflects its endogenous expression of glycosyltransferases. This

Gordon, J. 1. (1989).J . Cell Biol. 108, 1187-1 194.

Roth, J. (1984).J. Cell Biol. 98, 399406.

You might also like