Cloningofaratgeneencodingthehisto-bloodgroupA enzyme
Tissue expressionofthegeneandoftheAandB antigens
Anne Cailleau-Thomas
1
,Be
´
atrice Le Moullac-Vaidye
1
,Je
´
zabel Rocher,
1
Danie
`
le Bouhours
2
,
Claude Szpirer
3
and Jacques Le Pendu
1
1
INSERM U419, Institut de Biologie, Nantes, France;
2
INSERM U539, Faculte
´
de Me
´
decine, Nantes Cedex, France;
3
IBMM, Universite
´
Libre de Bruxelles, Gosselies, Belgium
The complete coding sequence ofa BDIX ratgene homo-
logous to the human ABO gene was determined. Identifi-
cation ofthe exon–intron boundaries, obtained by
comparison ofthe coding sequence with rat genomic
sequences from data banks, revealed that therat gene
structure is identical to that ofthe human ABO gene. It
localizes to rat chromosome 3 (q11-q12), a region homolo-
gous to human 9q34. Phylogenetic analysis ofa set of
sequences available for the various members ofthe same
gene family confirmed that therat sequence belongs to the
ABO gene cluster. The cDNA was transfected in CHO cells
alreadystablytransfectedwithana1,2fucosyltransferase in
order to express H oligosaccharide acceptors. Analysis of the
transfectants by flow cytometry indicated that A but not B
epitopes were synthesized. Direct assay ofthe enzyme
activity using 2¢ fucosyllactose as acceptor confirmed the
strong UDP-GalNAc:Fuca1,2GalaGalNAc transferase
(A transferase) activity oftheenzyme product and allowed
detection ofa small UDP-Gal:Fuca1,2GalaGal transferase
(B transferase) activity. The presence ofthe mRNA and of
the AandBantigens was searched in various BDIX rat
tissues. There was a general good concordance between the
presence ofthe mRNA and that oftheA antigen. Tissue
distributions oftheAandBantigens in the homozygous
BDIX rat strain were largely different, indicating that these
antigens cannot be synthesized by alleles ofthe same gene in
this rat inbred strain.
Keywords:ABO;N-acetylgalactosaminyltransferase; histo-
blood group; antigen; rat.
Histo-blood groupantigensAandB are oligosaccharides
carried by glycolipids and glycoproteins or present as free
oligosaccharides in some biological fluids such as milk or
urine. The immunodominant AandB epitopes correspond
to the trisaccharides GalNAca1,3(Fuca1,2)Galb-and
Gala1,3(Fuca1,2)Galb-, respectively. In humans, the ABO
gene is polymorphic with A alleles encodingA transferases,
B alleles encodingB transferases and O alleles encoding
inactive products. TheA transferases catalyse the transfer of
an N-acetylgalactosamine to acceptor H substrates
(Fuca1,2Galb-) whereas theB transferases catalyse the
transfer ofa galactose to the same substrates [1]. ABH
antigens are found in many species and have a wide tissue
distribution. Their main sites ofexpression appear to be
epithelia in contact with the external environment such as
the gut, the higher respiratory tract andthe genito-urinary
tract. In some primates, they are present on the vascular
endothelium of all tissues and in chimpanzee, gorilla and
man they are additionally present on erythrocytes, hence
their name blood groupantigens [2].
The molecular genetic basis ofthe human ABO alleles
has been elucidated. Although many mutations have been
described to date, only some of these are functionally
relevant [3]. Functional analyses have been performed to
determine which amino-acids are responsible for theA or
B enzyme activities. These studies revealed that the amino-
acids at positions 266 and 268 and to a lesser extent at
position 235 were critical in determining whether the
enzyme transfers an N-acetylgalactosamine, a galactose or
both [4]. For example, the presence ofa glycine at position
268 allows the transfer of an N-acetylgalactosamine
(GalNAc). But this activity is modulated by amino-acids
at the other two positions as, depending on these, transfer
of Gal may become possible in addition to the transfer of
GalNAc.
The biological meaning ofthe ABO phenotypes is still
largely obscure. Yet, the above mentioned tissue distribu-
tion and some associations between the ABO polymor-
phism and infectious diseases suggest a role in the
interaction with pathogens. For example, a strong associ-
ation is found between blood group O and susceptibility to
cholera [5]. Moreover, various strains of bacteria can adhere
to either A, B or H antigens, suggesting that microbes use
them as receptors and that their host range may be
influenced by the individual blood group phenotype [6].
Histo-blood groupantigens or structurally related carbo-
hydrates may be present on pathogens. Protective antibod-
ies directed against these structures may therefore be
Correspondence to J. Le Pendu, Inserm U419,
Institut de Biologie, 9 Quai Moncousu, F-44093, Nantes, France.
Fax: + 33 240 08 40 82, Tel.: + 33 240 08 40 99,
E-mail: jlependu@nantes.inserm.fr
Abbreviations: A transferase, UDP-GalNAc:Fuca1,2GalaGalNAc
transferase; B transferase, UDP-Gal:Fuca1,2GalaGal transferase;
CHO, Chinese hamster ovary; GalNAc, N-acetylgalactosamine;
Gal, galactose; Fuc, fucose; FITC, fluorescein isothiocyanate;
AEC, 3-amino-9-ethylcarbazol.
Enzymes: UDP-GalNAc:Fuca1,2GalaGalNAc transferase
(EC 2.4.1.40).
(Received 22 March 2002, revised 12 June 2002, accepted 5 July 2002)
Eur. J. Biochem. 269, 4040–4047 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03094.x
differentially generated by the host, depending on the blood
group phenotype [7]. The differential host range for
adhesion of pathogens andthe presence of protective
antibodies mean that all individuals would not be equally
sensitive to a given pathogen. This would provide a
mechanism of protection against the pathogenic strain at
the level ofthe population anda selective force to maintain
polymorphism at the ABO locus [6].
In the present work, we report the isolation ofa BDIX rat
cDNA homologous to the human ABO sequences encoding
for an Ahisto-bloodgroup enzyme. Examination of the
tissue distribution ofthe corresponding mRNA andof the
A andBantigens in the BDIX strain ofrat was performed,
allowing some aspects ofthe ABO genetics in this species to
be considered.
MATERIALS AND METHODS
Cloning ofarat ABO-like cDNA
An EMBL-3 rat genomic DNA phage library (Clontech)
was screened with a cDNA probe (P718), derived from the
human blood groupA transferase and corresponding to
nucleotides 148–865 ofthe human cDNA sequence [8].
Positive recombinants were purified, digested with EcoRI
and analyzed by Southern Blot using the pb718 probe. Two
hybridization-positive fragments (4.0 and 3.5 kb) were
obtained and digested by HinfI. The products were ligated
into the pUC18 vector (Pharmacia) and sequenced. One
clone, from the 4 kb fragment digest, contained an insert of
406 bp, a stretch of which was 84.4% similar to exon 6 of
the human blood groupAgene coding sequence. The
complete coding sequence was obtained by RACE/PCR
using primers deduced from this fragment. To this end, total
RNA from BDIX rat stomach was extracted using the SV
Total RNA isolation kit from Promega. This RNA
preparation was used to obtain mRNA using the Oligotex
kit from Qiagen. Double stranded cDNA synthesis, adaptor
ligation and RACE/PCR were performed using the Clon-
tech Marathon cDNA Amplification kit. Elongation in the
5¢ direction was performed using an inverted primer
deduced from the 406 bp fragment homologous to the
human Agene (GTCGATGTTGAAGGTCCCCTCCCA
GATG) andthe adaptor AP1 primer provided by the
supplier. The second nested PCR was performed using a
nested inverted primer deduced from the same fragment
sequence (TCCCAGATGATGGGAGCCACGCCAA
GG) andthe nested adaptor AP2 primer from the supplier.
Synthesis in the 3¢ direction was performed by the same
method using the following first primer
(CTTGTCTTCACTCCTTGGCTGGCTCCCAT) and
the adaptor AP1 primer, followed by a second nested
PCR with the second nested primer (CATCTGG
GAGGGGACCTTCAACATCGAC). The PCR was run
with the Advantage Polymerase (Clontech) andthe man-
ufacturer’s touchdown-RACE program. The PCR products
were ligated into the pUC18 vector and sequenced. To
obtain the full coding fragment, stomach cDNA was PCR
amplified using the following primers, deduced from the
sequence ofthe 5¢ and 3¢ RACE products: AC
CATCCCGGGCCTTGCATGGA (forward) and GCTA
CAGGTACCGCCTCTCCAA (reverse). The product was
ligated into the pUC18 vector and sequenced.
Sequence analysis
Multiple alignments were performed with the
CLUSTALW
program [9]. Genetic distances ofthe complete deduced
peptide sequences were calculated by the ÔNeighbor joiningÕ
method from the
PHILIP
program. Approximate location of
transmembrane regions were determined using the
TMHMM
,
TMAP
and
TOPPRED
2 programs. Determination ofthe exon/
intron boundaries were obtained by analysis ofthe rat
genomic sequences available in the NCBI database. The
programs used are all available from http://www.infobio-
gen.fr.
Chromosome localization
The Abo gene was first assigned to arat chromosome using
a panel of standard rat X mouse cell hybrids that segregate
rat chromosomes [10]. The hybrids were typed by PCR
with the following primers: 5¢-GGAGCAGCTGGAGT
CATG-3¢ and 5¢-GGTCATCCTGTATCCTTCA-3¢ (the 5¢
end of these primers corresponds to the positions 163 and
270, respectively, of [35104745] from the Rattus norvegicus
WGS trace database). For regional localization, the panel of
rat · hamster radiation cell hybrids [11] was typed in the
same manner. The mapping results were obtained from the
rat radiation hybrid map server at the Otsuka GEN
Research Institute (http://ratmap.ims.u-tokyo.ac.jp/menu/
RH.html) [11].
RT-PCR analysis
Total RNAs (1 lg) from various rat tissues listed below
were prepared using the SV Total RNA isolation System kit
from Promega and reverse transcribed at 42 °Cwiththe
M-MLV reverse transcriptase from Promega. Contaminat-
ing DNA had been removed by digestion with RNAse-free
DNAseI (10 unitsÆlg
)1
RNA) for 15 min at room temper-
ature. Amplification ofthe cDNA corresponding to the A
enzyme cDNA was performed with the following primers:
CAGACGGATGTCCAGAAAGTTG and: GCTACAG
GTACCGCCTCTCCAA. Amplification was performed
using the Advantage Polymerase (Clontech) with initial
denaturation at 94 °C 3 min, followed by 30 cycles of 94 °C
30 s, 64 °C45s,68°C 2 min. The amplification yields a
product of 929 bp. The control of cDNA quality was
performed by amplification of glyceraldehyde phosphate
dehydrogenase (GAPDH).
Transfection of CHO cells
The complete coding sequence oftheratAgene was
inserted into the pDR2 eukaryotic expression vector
(Clontech) deleted ofthe sequences lying between the
EcoRV and ClaI sites. Chinese hamster ovary carcinoma
cells, CHO cells, are devoid of a1,2fucosyltransferase
activity and therefore of ABH antigens. They were first
transfected using LipofectAMIN
TM
with therat a1,2fuco-
syltransferases cDNA FTB (GenBank accession number
AF131238) in the pBK-CMV expression vector (Gibco,
Paisley, UK) according to the manufacturer’s instructions.
Twenty-four hours later, fresh medium was added and 48 h
later, selective medium containing 0.6 mgÆmL
)1
G418
(Gibco) was added. Transfected cells expressing a1,2-linked
Ó FEBS 2002 Rathisto-bloodgroupAenzyme (Eur. J. Biochem. 269) 4041
fucose residues were selected by flow cytometry, using
fluorescein isothiocyanate (FITC)-labeled UEA-I, after
cloning by limiting dilutions as previously described [12].
For each cell line, a strongly expressing clone was selected
and transfected a second time using the same procedure
with theratAenzyme cDNA in the pDR2 vector. Forty-
eight hours later, cells were cultured in selective medium
containing 0.6 mgÆmL
)1
hygromycin. After cloning by
limiting dilutions, strongly A antigen expressing transfec-
tants were selected. Control transfected cells were prepared
by transfection with the empty vectors. These stable
transfectants were cultured in RPMI 1640, 10% fetal
bovine serum, 2 m
ML
-glutamine, free nucleotides
(10 lgÆmL
)1
), 100 UÆmL
)1
penicillin and 100 lgÆmL
)1
streptomycin (Gibco) supplemented with 0.25 mgÆmL
)1
G418 and 0.2 mgÆmL
)1
hygromycin. They were cultured at
confluence after dispersal with 0.025% trypsin in 0.02%
EDTA. Cells were routinely checked for mycoplasma
contamination by Hoescht 33258 (Sigma, St Louis, MO,
USA) labeling.
Cytofluorimetric analysis
Viable cells (2 · 10
5
per well) were incubated with antibod-
ies at the appropriate dilutions in NaCl/P
i
containing 0.1%
gelatin for 1 h at 4 °C. Optimal concentrations of antibodies
were chosen after serial dilutions to obtain the strongest
positive signal without cell death. The anti-A mAb 3–3 A
was obtained from J. Bara (INSERM U482, Villejuif,
France). It recognizes all types ofAantigens [13] and does
not show any detectable cross reactivity with B epitopes as
judged from enzyme immunoassay with synthetic oligosac-
charides and immunostaining of human tissues of known
ABO phenotypes. The anti-B mAb ED3 is a gift from
A. Martin (CRTS, Rennes, France). It recognizes all types
of Band shows no detectable cross-reactivity with
A epitopes or with the Gala1,3Gal epitope [14]. Following
the first incubation with monoclonal antibodies, after three
washes with the same buffer, a second 30 min incubation
was performed with the FITC-labeled anti-(mouse Ig) Ig
under the same conditions. After washings in the same
buffer, fluorescence analysis was performed on a FACScan
(Beckton–Dickinson) using the
CELLQUEST
program.
Detection ofenzyme activity
Confluent transfected CHO cells were rinsed with ice-cold
NaCl/P
i
, pH 7.2, then recovered by scraping. After
washing with ice-cold NaCl/P
i
, cells were solubilized in
50 m
M
cacodylate pH 7.0, containing 2% (v/v) Triton
X-100 on ice for 30 min Following a centrifugation at
13 000 g for 10 min, the supernatant was collected and used
as a crude enzyme preparation. Protein concentration was
determined using bicinchoninic acid. The reaction mixture
contained: 50 lg protein extracts, 30 m
M
MnCl
2
,5m
M
ATP, 10 m
M
NaN
3
,5m
M
2¢ fucosyllactose and 20 l
M
UDP-
D
-[
14
C]N-acetylgalactosamine (55 mCiÆmmol
)1
,ICN,
Costa Mesa, CA, USA) or 20 l
M
UDP-
D
-[
14
C]galactose
(278 mCiÆmmol
)1
, NEN, Chemical Center, Dreieichen-
dain, Germany) in a final volume of 50 lLandwas
incubated at 37 °C for 16 h. After incubation, the reaction
mixture was quenched with 750 lL distilled water and
applied to an AG1-X8 column, chloride form, 100–200
mesh (Bio-Rad, Hercules, CA, USA). The radiolabeled
product was then eluted with 1 mL water and counted in
5 mL scintillation liquid (Ready Safe
TM
, Beckman, Palo
Alto, CA, USA). Background levels of radioactivity were
obtained from controls without exogenous acceptor.
Values obtained for the controls were then subtracted
from those obtained for the assays.
Immunohistological analysis
Tissues from 2- to 3-month old rats were collected and
immediately frozen or paraffin embedded. Sections (5 lm)
werepreparedandwashedinNaCl/P
i
. Endogenous
peroxidase was inhibited using methanol/H
2
O
2
0.3% for
20 min. Sections were then washed in NaCl/P
i
for 5 min
and covered with NaCl/P
i
/BSA 1% for 20 min at room
temperature in a moist chamber. After washing in NaCl/P
i
,
sections were covered with either the primary antibodies
diluted in NaCl/P
i
/BSA 1% and left at 4 °Covernight.
Sections were then rinsed thrice with NaCl/P
i
and incubated
with biotinylated anti-(mouse IgG) Ig (Vector Laboratories,
Burlingame, CA, USA) diluted at 1/100 for 60 min at room
temperature. After washing in NaCl/P
i
, the sections were
covered with peroxidase-conjugated avidin (Vector Labo-
ratories) diluted at 1 : 1000 for 45 min, washed with NaCl/
P
i
and reactions were revealed with 3-amino-9-ethylcarbazol
(AEC). Counterstaining was performed with Mayer’s
hemalun.
RESULTS
Sequence analysis
A coding sequence homologous to the human ABO coding
sequences was isolated (GenBank accession AF264018). It
possesses an open reading frame of 1047 bp, 77% identical
with the human Agene coding region. Comparisons of this
sequence with genomic sequences available in the rat
genome data bank allowed determination ofthe exon/
intron boundaries as they conform to the GT-AG consensus
rule (Fig. 1). Thegene organization appears similar to that
of the human ABO gene with seven exons [15]. The same
analysis was performed on the mouse A/B gene orthologous
to the human ABO gene [16]. This gene lacks exon 4. At the
amino-acid level, therat sequence shows 70, 71 and 77%
identity with the human, pig and mouse sequences, respec-
tively (Fig. 2A, Table 1). Like all glycosyltransferases the
Fig. 1. Comparison oftherat Abo gene exonic structure with those of
the human and mouse orthologs. Organization ofthe human and mouse
genes has been previously reported. Exons are represented by boxes
numbered in bold. Nucleotide numbers limiting the exons are noted in
superscript. For the mouse gene, a search in the mouse genomic data
bank confirmed the absence of one exon, corresponding to exon 4 of
the other two species, as the genomic sequence separating mouse exons
3 and 4 revealed no potential exonic sequence with homology to the
human andrat exon 4.
4042 A. Cailleau-Thomas et al. (Eur. J. Biochem. 269) Ó FEBS 2002
rat enzyme presents a short N-terminal intracytoplasmic
domain, a transmembrane domain followed by a stem
region anda catalytic domain, the latter domain presenting
the highest identity among the four sequences. Of note, the
presence ofa conserved potential N-glycosylation site at
the beginning ofthe catalytic domain. Previous studies of
the human ABO transferases underscored the major
importance ofthe amino-acid at position 268 with a glycine
determining A activity whereas an alanine determines B
activity. A glycine is present at the homologous position in
the rat sequence (position 263), suggesting a potential A
enzyme activity.
At present, the ABO gene family is known to comprise
four members, the ABO gene itself, the a3galactosyltrans-
ferase (pseudo B) gene [17], the aN-acetylgalactosaminyl-
transferase or Forssman synthetase gene [18] and the
a-galactosyltransferase iGb3 synthetase gene [19]. Phylo-
genetic analysis of this gene family revealed a clear
distinction between the four genes, with the new rat
sequence falling within the ABO cluster (Fig. 2B).
Chromosome localization oftherat Abo gene
The gene was first assigned to rat chromosome 3, using a
panel of 16 standard rat X mouse cell hybrid clones
segregating rat chromosomes. No discordant clone was
obtained for chromosome 3, while at least two discordant
clones were counted for each other chromosome (data not
shown). Chromosome placement by radiation hybrid map-
ping confirmed this result, with a precise localization
between D3Rat54 (at 58cR, lod score ¼ 5.67) and
D3Rat50 (at 62cR, lod score ¼ 5.12). This position corre-
sponds to the centromeric region ofthe chromosome (bands
3q11–q12). This rat chromosome region is known to be
homologous to the human region 9q34 [20,21], where the
human ABO gene resides [22].
Determination oftheenzyme activity
In order to study the functional characteristics ofthe rat
A-like gene, the cDNA was transfected in CHO cells already
stably transfected with an a2fucosyltransferase cDNA,
namely therat FTB [23]. The presence of this fucosyltrans-
ferase in CHO cells allows expressionof H histo-blood
group structures which are compulsory precursors ofthe A
and B antigens. Stable doubly transfected cells were isolated
and tested by flow cytometry for their expressionofA or B
antigens (Fig. 3). A positive control cell line (MT-450)
known to constitutively express both antigens indicated that
the antibodies readily detected their respective epitopes
when present [24]. The doubly transfected CHO cells were
strongly labeled by the anti-A reagent, but not significantly
Fig. 2. Comparison ofthe amino acid sequences oftheA or cis A/B
transferases in four species (A) and phylogenetic analysis ofthe ABO
gene family (B). (A) Identical residues are marked by arrows. The
transmembrane regions are highlighted in grey, the conserved N-gly-
cosylation site is boxed andthe amino-acids corresponding to positions
266 and 268 ofthe human sequence are labeled in white on black
boxes. The numbering corresponds to that ofa consensus sequence.
(B) Genetic distances were calculated from
CLUSTALW
multiple align-
ments using the Ôneighbour joiningÕ method from the sequences listed
in Table 1. The scale bar represents the number of substitutions per site
for a unit branch length.
Table 1. Identification ofthe ABO gene family sequences used for the
phylogenetic analysis.
Enzyme Species GenBank/EBI
A transferase Human J05175
A transferase Rat AF264018
A transferase Pig AF050177
A(cis A/B) transferase Mouse AB041039
A-like
a
Human M65082
Gal transferase Platyrrhini S71333
Gal transferase Marmoset A56480
Gal transferase Cow J04989
Gal transferase Pig L36152
Gal transferase Mouse M85153
Gal transferase Rat AF520589
IGb3 synthetase Human AL513327
IGb3 synthetase Rat AF246543
Forssman synthetase
a
Human AF163572
Forssman synthetase Dog CFU66140
a
Pseudogenes.
Ó FEBS 2002 Rathisto-bloodgroupAenzyme (Eur. J. Biochem. 269) 4043
by the anti-B reagent, suggesting that the new rat gene
encodes an enzyme with the catalytic activity oftheA histo-
blood group transferase. To confirm this result, the enzyme
activity was directly assayed on cell extracts of CHO
transfectants (Fig. 4). No transfer of galactose or
N-acetylgalactosamine could be detected on extracts from
the control FTB transfected cells using 2¢ fucosyllactose as
acceptor. However, cell extracts from the double transfec-
tants showed a high N-acetylgalactosaminyltransferase
activity anda weak galactosyltransferase activity. These
results indicate that the new ratenzyme is indeed an A histo-
blood group transferase with a small B transferase activity.
Tissue expressionofthe mRNA andofthe A
and Bantigens in the rat
An RT-PCR analysis was performed to determine the
tissue expressionoftherat Abo gene. Primers were chosen
to encompass different exons so as to confirm amplifica-
tion of cDNA and lack of contamination by genomic
DNA. A band at the expected size was detected in various
tissues as indicated in Table 2. A strong signal was
obtained in the oesophagus, the stomach, the colon, the
Fig. 3. Cytofluorimetric analysis of cell surface
A andBantigensof stably transfected CHO
cells. CHO cells previously transfected with
the rat FTB cDNA encoding an a1,2fucosyl-
transferase were cotransfected with therat A
enzyme cDNA. Stable transfectants were iso-
lated and one of these clones was used for
analysis. The MT-450 cell line, a mammary
carcinoma cell line from the w/Fu rat strain,
was used as positive control. TheAand B
antigens were detected using the 3–3 A and
ED3 Mabs, respectively. The Logs of fluo-
rescence intensities are plotted against cell
numbers. Negative controls were performed in
absence of primary antibody.
Fig. 4. Enzymatic assays of CHO transfectant cell extracts. The same
stable transfectants of CHO cells used for the cytofluorimetric analysis
of theA or Bantigensexpression (Fig. 3) were used to assay the
enzyme activity. Cell extracts were prepared as described in the
Materials and methods section. Activities were determined using
2¢ fucosyllactose as acceptor substrate and either UDP-[
14
C]galactose
or UDP-[
14
C]N-acetylgalactosamine as donor substrates. The product
of the reaction was separated on AG1-X8 anion exchange columns.
The background was determined in absence of acceptor substrate and
its value was deduced from the values obtained in the presence of the
acceptor. Values ofthe specific activities are given in
pmolÆh
)1
Æmg protein
)1
of either [
14
C]galactose or [
14
C]N-acetylgalac-
tosamine transferred.
Table 2. Tissue distribution ofthe Abo mRNA andoftheAand B
antigens in the BDIX rat. Transcripts were detected by RT-PCR from
total RNA extracts of various rat tissues using specific primers. An
indication ofthe intensity ofthe detected band is given, with +++
corresponding to the strongest signal and – to no detectable signal. The
A andBantigens were detected by immunohistochemistry on frozen
tissue sections using well characterized specific monoclonal antibodies.
In tissues positive for both AandB antigens, the cellular distribution
of the two antigens is not always identical (see text for details). The
labeling by the anti-A antibody was the same on paraffin embedded
sections, but no B antigen was detected on such sections. ND ¼ not
done.
Tissue mRNA A antigen B antigen
Tongue ND +++ –
Oesophagus ++ ++ –
Stomach ++ ++ ++
Small intestine – – –
Caecum ND +++ +++
Large intestine +++ +++ ++
Pancreas + +++ ++
Parotid gland ND +++ –
Submaxillary gland +– ++ ++
Liver – – –
Trachea ND – –
Lung – – –
Kidney ++ – ++
Urinary bladder ++ – –
Ovary – – –
Uterus ++ +++ –
Testis – – –
Seminal vesicle ND +++ –
Thyroid gland ND +++ –
Parathyroid gland ND +++ –
Brain – – –
Muscle +– – –
Skin – – –
Spleen +– – –
Thymus ++ ++ –
Lymph node – – –
4044 A. Cailleau-Thomas et al. (Eur. J. Biochem. 269) Ó FEBS 2002
kidney, the urinary bladder, the uterus andthe thymus. A
weaker signal was obtained from the pancreas and very
weak, barely detectable, signals were visible from a
salivary gland, muscle and spleen. The presence of this
mRNA was compared with that oftheAandB histo-
blood groupantigens in the various rat tissues. The A
antigen appeared to have a wider distribution than the B
antigen as shown in Table 2 and Fig. 5. There was a
general agreement between the presence oftheA antigen
and that ofthe Abo gene mRNA. Indeed, theA antigen
was expressed in the oesophagus, the stomach, the large
intestine (Fig. 5a), the pancreas, the uterus (Fig. 5g), the
seminal vesicle (Fig. 5h) andthe thymus (Fig. 5e), while it
was essentially absent from the small intestine like the
mRNA, although a few glands were positive (Fig. 5c).
However, a strong antigen expression was noted in the
submaxillary gland whereas the mRNA was barely
detected. Conversely the mRNA was readily detected in
the kidney andthe urinary bladder whereas the antigen
was not. TheB antigen was detected in fewer tissues than
the A antigen. It could not be found in the tongue, the
oesophagus, the parotid gland, the uterus, the seminal
vesicle, the thyroid and parathyroid glands and the
thymus where theA antigen was present. Inversely,
the B antigen but not theA antigen could be detected in
the kidney. In this organ its distribution was limited to
some tubules ofthe medulla (Fig. 5f). In those tissues
where both theAandBantigens were present, their
cellular distribution differed. For example, in the stomach,
the A antigen was mainly present on cells ofthe neck area
and on the parietal cells ofthe gastric glands whereas the
B antigen was present on the pits epithelial cells and
the chief cells ofthe glands. In the large intestine and the
caecum, theA antigen was strongly expressed throughout
the mucosa whereas presence oftheB antigen was
restricted to the surface epithelium (Fig. 5a,b). In the
submaxillary gland, theA antigen was expressed in the
secretory cells andtheB antigen in the duct cells. The A
and Bantigens were only coexpressed in the pancreas
acinar cells (Fig. 5d) and in some sebaceous glands of the
skin.
DISCUSSION
A rat cDNA homologous to the human ABO gene has been
cloned. It encodes a protein with a strong A transferase and
a small B transferase activity in vitro. The human A enzyme
also is known to possess a small B transferase activity [25].
Yet, this activity is not comparable with that of cis A/B
human alleles or ofthe mouse enzyme which are char-
acterized by their ability to transfer galactose and
N-acetylgalactosamine about equally well [16]. That the
rat enzyme described here is truly an Aenzyme is confirmed
by the fact that upon transfection into CHO cells, A antigen
was readily detected whereas B antigen was not. In addition,
there were BDIX rat tissues strongly expressing A antigen
and no detectable B antigen.
The ABO gene structure is conserved between rat, pig
and man while one exon (exon 4) has been lost in the mouse.
As the mouse cDNA encodes an active enzyme, it is clear
that this exon is not required for enzyme activity. This is not
surprising as it corresponds to a part ofthe stem region of
the enzyme. Nevertheless, it could affect the type of
structures used by the mouse enzyme as acceptor substrates,
i.e. glycolipids vs. glycoproteins or the type of precursor.
Previous detailed studies aimed at defining the amino-acid
residues that determine theA or B specificity ofthe human
enzymes have been performed by site-directed mutagenesis.
From this work, it could be concluded that residues at
positions 266 and 268 were critical and that the residue at
position 235 was influencial [3]. At the position correspond-
ing to amino-acid 235 ofthe human sequence, therat and
pig A enzymes have a glycine residue like the human A and
mouse cis A/B enzymes. Ofthe three positions, 268 is
considered the most important and, as noted above, a
glycine at this position characterizes theA activity. The rat
sequence reported herein has a glycine at the position
equivalent to human 268 in accordance with its A enzymatic
activity. Similar to the pig enzyme [26], but at variance with
the human A enzymes, it has an alanine at position 266. In
man, as well as in anthropoid apes, a leucine is present at
this position, confirming that it is indeed less important than
the amino-acid at position 268 in determining theA or B
specificity ofthe enzyme.
Phylogenetic analysis including sequences ofthe four
known members ofthe ABO gene family confirmed that
Fig. 5. Immunohistochemical analysis oftheAandBantigens expres-
sion in BDIX rat tissues. Frozen BDIX rat tissues sections were incu-
bated with an anti-A (a, c, e, g, h) or an anti-B (b, d, f) Mab and their
binding was detected as described in the Ômaterials and methodsÕ
section. In the large intestine, theA antigen is detected throughout the
mucosa (a) whereas theB antigen is restricted to the surface epithelium
(b). In the small intestine, theAandBantigens are absent except for a
few glands displaying A antigen (c). In the pancreas, both antigens are
present on the acinar cells as illustrated for theB antigen (d). In the
thymus some medullary epithelial cells express theA epitopes (e). In
the kidney medulla, some tubules are B positive (f). TheA antigen is
present in the uterine epithelium (g) andthe seminal vesicle (h).
Ó FEBS 2002 Rathisto-bloodgroupAenzyme (Eur. J. Biochem. 269) 4045
these four genes can be clearly separated across various
species. In addition, despite the small number of sequences
available for two ofthe genes, genetic distances among
species for each gene were quite similar, suggesting that they
evolved at about equal rates. From this, one can speculate
that the four members ofthe ABO gene family are
submitted to the same kind of selective pressure.
We generally observed a good concordance between the
presence ofthe mRNA andoftheA antigen in tissues.
Nevertheless, there were some discrepancies in tissues such
as the kidney andthe urinary bladder where the mRNA was
easily detected but which did not express A or B epitopes.
The presence ofa glycosyltransferase mRNA and the
corresponding glycan structure may not always correlate as
the mRNA may not necessarily be translated or the
appropriate precursor glycans may not be available. In the
present case, H antigen, the precursor oftheA or B antigens
is not synthesized in therat urinary bladder epithelium (data
not shown). Some H antigen should be present in the kidney
as B antigen was detected. This is in accordance with the fact
that therat FTA a1,2fucosyltransferase mRNA can be
detected in the BDIX rat kidney [23]. One would thus expect
to find some A antigen in therat kidney. However, the A
enzyme mRNA may not be expressed in the same cells as
the FTA mRNA. Further studies by in situ hybridization
are required to clarify this point.
In humans, A or Bantigens are present on glycolipids
as well as on glycoproteins and based on various types of
precursors [27]. In rats, theAandBantigens have been
characterized on glycolipids [28–30]. A polymorphism of
the expressionofthe A-active glycolipids has been found
among various strains of inbred rats [31]. TheA antigen is
also present on glycoproteins as it can be detected by
Western blotting on various bands from the transfected
cells (data not shown) or other A positive cell types from
BDIX rats [32,33] and as it has been characterized on
glycopeptides from Sprague–Dawley rats. At variance, the
B antigen could not be found on such glycopeptides [34].
In addition, theA antigen is still strongly detected on
paraffin embedded rattissue sections [35]. Paraffin
embedding-deparaffination is known to remove glyco-
lipids, therefore the remaining antigenic activity should
correspond to glycoproteins, while frozen sections contain
both glycoproteins and glycolipids. In the present study,
the B reactivity was only detected on frozen sections and
not on paraffin embedded sections, suggesting that it is
restrited to glycolipids. The tissues that express most ABH
antigens in humans are quite similar to those that express
A antigen in therat although there are some notable
differences. Unlike humans, rats do not present these
antigens on erythrocytes, the vascular endothelium or the
epidermis. There are also some regional differences. Most
strains of rats, like the BDIX strain used in this study, do
not express theA antigen in the small intestine although
they express H antigen. Inversely, Aand H antigens are
present in the large intestine down to the rectum whereas
they are absent from the human rectal and distal colonic
epithelia [36]. Another major difference between humans
andratsintermsoftissueexpressionoftheAandB
antigens is that in man AandB epitopes are expressed in
the same cell types whereas in therat they are not. It had
been observed earlier that B, but not A, antigen is
developmentally expressed on rat cochlear hairy cells and
on olfactory cells [37,38]. In the present study, we noticed
that with the exception ofthe pancreas andthe skin, in rat
organs, theAandBantigens never codistributed at the
cellular level. It is unlikely that mAb ED3, the anti-B that
we used, detected the related pseudo-B or Gala1,3Gal
epitope as the antibody did not react with the synthetic
disaccharide (data not shown) and as this potentially
cross-reactive epitope is expressed on therat vascular
endothelium to which mAb ED3 did not bind. Neverthe-
less, the possibility that mAb ED3 detects another B-like
structure cannot be completely eliminated. BDIX is a rat
strain that has been generated in the 1940s and is inbred.
Therefore, in these animals, theB antigen cannot be
synthesized by theenzyme product of an allele at the Abo
locus. It is to be expected that another gene encodes a
galactosyltransferase with Bhisto-bloodgroup activity.
The enzyme activity oftherat pseudo B a3galactosyl-
transferase has not been reported as yet. One possibility is
that, unlike in other species, this enzyme could have the
ability to transfer a galactose residue in a1,3 linkage to
a1,2fucosylated precursor glycolipids. It could also be that
one ofthe two other known members ofthe family,
namely the iGb3 synthetase andthe Forssman synthetase,
could have a dual specificity. Alternatively, there could
exist another gene in therat genome encodinga B-like
blood group transferase that would act exclusively on
glycolipids. These possibilities remain to be tested.
Availability oftheratAgene sequence will allow the
search of genetic polymorphisms at the Abo locus in this
species. Comparisons ofthetissueexpression across species,
as well as ofthe sequences and genetic polymorphisms of
various mammals should help understanding the biological
significance of ABO antigens during evolution. The results
presented here should be useful in such future analyses.
ACKNOWLEDGEMENTS
The authors are grateful to Drs J. Bara and A. Martin for their
generous gift of antibodies, to Dr M. Cle
´
ment for help with the pictures,
to Mrs P. Fichet and S. Minaut for great animal care and to Pascale
Van Vooren for excellent technical assistance. They thank Dr J F.
Bouhours for helpul discussions. The work was supported by grants
from the Association for International Cancer Research (AICR), the
Association pour la Recherche sur le Cancer (ARC) andthe Fund for
Scientific Medical Research (FRSM, Belgium). C. S. is a Research
Director ofthe National Fund for Scientific Research (FNRS,
Belgium).
REFERENCES
1. Watkins, W.M. (1999) A half century of blood-group antigen
research. Some personal recollections. Trends Glycosci. Glycotech.
11, 391–411.
2. Oriol, R., Mollicone, R., Couillin, P., Dalix, A.M. & Candelier,
J.J. (1992) Genetic regulation oftheexpressionof ABH and Lewis
antigens in tissues. APMIS. 100 (Suppl. 27), 28–38.
3. Yamamoto, F. (2000) Molecular genetics of ABO. Vox Sang. 78,
91–103.
4. Yamamoto, F. & McNeill, P.D. (1996) Amino acid residue at
codon 268 determines both activity and nucleotide-sugar donor
substrate specificity of human histo-bloodgroupAandB trans-
ferases. J. Biol. Chem. 271, 10515–10520.
5. Glass, R.I., Holmgren, J., Haley, C.E., Khan, M.R., Svennerholm,
A.M.,Stoll,B.J.,BelayetHossain,K.M.,Black,R.E.,Yunus,M.
4046 A. Cailleau-Thomas et al. (Eur. J. Biochem. 269) Ó FEBS 2002
& Barua, D. (1985) Predisposition for cholera of individuals with
O blood group. Possible evolutionary significance. Am.J.Epi-
demiol. 121, 791–796.
6. Marionneau, S., Cailleau-Thomas, A., Rocher, J., Le Moullac-
Vaidye, B., Ruvoe
¨
n-clouet, N., Cle
´
ment, M. & Le Pendu, J. (2001)
ABH and Lewis histo-bloodgroup antigens, a model for the
meaning of oligosaccharide diversity in the face ofa changing
world. Biochimie 83, 565–573.
7. Preece, A.F., Strahan, K.M., Devitt, J., Yamamoto, F.F. &
Gustavson, K. (2002) Expressionof ABO or related antigenic
carbohydrates on viral envelopes leads to neutralization in the
presence of serum containing specific natural antibodies and
complement. Blood 99, 2477–2482.
8. Yamamoto, F., Marken, J., Tsuji, T., White, T., Clausen, H. &
Hakomori, S.I. (1990) Cloningand characterization of DNA
complementary to human UDP-GalNAc: Fuca1–2Gala1–3Gal-
NAc transferase (histo-blood groupA transferase) mRNA. J. Biol.
Chem. 265, 1146–1151.
9. Thompson, J.D., Higgins, D.G. & Gibson, T.J. (1994) ClustalW:
improving the sensitivity of progressive multiple sequence align-
ment through sequence weighting, position specific gap penalties
and weight matrix choice. Nucleic Acids Res. 22, 4673–4680.
10. Szpirer,J.Levan,G.Tho
¨
rn, M. & Szpirer, C. (1984) Gene map-
ping in therat by mouse-rat cell hybridization: synteny of the
albumin and alpha-foetoprotein genes and assignment to chro-
mosome 14. Cytogenet. Cell Genet. 38, 142–149.
11. Watanabe, T.K. Bihoreau, M.T. McCarthy, L.C. Kiguwa, S.L.
Hishigaki, H. Tsuji, A. Browne, J. Yamasaki, Y. Mizoguchi-
Miyakita, A. Oga, K. Ono, T. Okuno, S. Kanemoto, N. Takah-
ashi, E. Tomita, K. Hayashi, H. Adachi, M. Webber, C. Davis, M.
Kiel,S.Knights,C.Smith,A.Critcher,R.Miller,J.Thangarajah,
T. Day, P.J.R. Hudson, J.R. Irie, Y. Takagi, T. Nakamura, Y.
Goodfellow, P.N. Lathrop, G.M. Tanigami, A. & James, M.R.
(1999) A radiation hybrid map oftherat genome containing 5,255
markers. Nat. Genet. 22, 27–36.
12. Goupille, C. Marionneau, S. Bureau, V. Hallouin, F. Meichenin,
M. Rocher, J. & Le Pendu, J. (2000) a1,2fucosyltransferase
increases resistance to apoptosis ofrat colon carcinoma cells.
Glycobiol. 10, 375–382.
13. Bara, J. Gautier, R. Le Pendu, J. & Oriol, R. (1988) Immuno-
chemical characterization of mucins. Polypeptide (M1) and poly-
saccharide (A and Le
b
)antigens.Biochem. J. 254, 185–193.
14. Le Pendu, J. Le Cabellec, M. & Bara, J. (1997) Immunohisto-
logical analysis of antibodies against ABH and other glycocon-
jugates in normal human pyloric and duodenal mucosae. Transfus
Clin. Biol. 1, 41–46.
15. Yamamoto, F. McNeill, P.D. & Hakomori, S.I. (1995) Genomic
organization ofthe human histo-bloodgroup ABO genes.
Glycobiol. 5, 51–58.
16. Yamamoto, M. Lin, X H. Kominato, Y. Hata, Y. Noda, R.
Saitou, N. & Yamamoto, F. (2001) Murine equivalent of the
human histo-bloodgroup ABO gene is a cis-AB gene that encodes
a glycosyltransferase with both AandB transferase activity.
J. Biol. Chem. 276, 13701–13708.
17. Joziasse, D.H. Shaper, J.H. Van den Eijnden, D.H. Van Tunen,
A.J. & Shaper, N.L. (1989) Bovine alpha1–3galactosyltransferase:
isolation and characterization ofa cDNA clone. Identification of
homologous sequences in human genomic DNA. J. Biol. Chem.
264, 14290–14297.
18. Xu, H. & Storch, T., Yu.M. Elliott, S.P. & Haslam, D.B. (1999)
Characterization ofthe human Forssman synthetase gene. J. Biol.
Chem. 274, 29390–29398.
19. Keusch,J.J.Manzella,S.M.Nyame,K.A.Cummings,R.D.&
Baenziger, J.U. (2000) Expressioncloningofa new member of the
ABO blood group glycosyltransferases, iGb
3
synthase, that directs
the synthesis of isoglobo-glycosphingolipids. J. Biol. Chem. 275,
25308–25314.
20. Nilsson, S. Helou, K. Walentinsson, A. Szpirer, C. Nerman, O. &
Stahl, F. (2001) Rat-mouse and rat-human comparative maps
based on gene homology and high-resolution zoo-FISH. Genomics
74, 287–298.
21. RATMAP Rat Genome Database, http://ratmap.Generalgu.se/
22. Henske, E.P., Ozelius, L., Anderson, M.A. & Kwiatkowski, D.J.
(1992) A radiation-reduced hybrid cell line containing 5 Mb/17 cM
of human DNA from 9q34. Genomics 13, 841–844.
23. Bureau, V., Marionneau, S., Cailleau-Thomas, A., Le Moullac-
Vaydie, B., Liehr, T. & Le Pendu, J. (2001) Comparison of the
three rat GDP-
L
-fucose: b-
D
-galactoside 2-a-
L
-fucosyltransferases
FTA, FTB and FTC. Eur. J. Biochem. 268, 1006–1019.
24. Sleeman, J.P., Kim, U., Le Pendu, J., Howells, N., Coquerelle, T.,
Ponta, H. & Herrlich, P. (1999) Inhibition of MT-450 rat mam-
mary tumour growth by antibodies recognising subtypes of blood
group antigen B. Oncogene 18, 4485–4494.
25. Yates, A.D. & Watkins, W.M. (1982) The biosynthesis of blood
group B determinants by the blood groupAgene specified a-3-
N-acetylgalactosaminyltransferase. Biochem. Biophys. Res. Com-
mun. 109, 958–965.
26. Meijerink, E., Neuenschwander, S., Dinter, A., Yerle, M.,
Stranzinger, G. & Vogeli, P. (2001) Isolation ofa porcine UDP-
GalNAc transferase cDNA mapping to the region ofthe blood
groupEAAlocusonpigchromosome1.Anim. Genet. 32, 132–138.
27. Clausen, H. & Hakomori, S.I. (1989) ABH and related histo-
blood group antigens; immunochemical differences in carrier iso-
types and their distribution. Vox Sang. 56, 1–20.
28. Breimer, M.E., Hansson, G.C., Karlsson, K.A. & Leffler, H.
(1982) Glycosphingolipids ofrat tissues. Different composition of
epithelial and nonepithelial cells of small intestine. J. Biol. Chem.
257, 557–568.
29. Hansson, G.C., Bouhours, J.F. & Angstro
¨
m, J. (1987) Char-
acterization of neutral blood group B-active glycosphingolipids of
rat gastric mucosa. J. Biol. Chem. 262, 13135–13141.
30. Hansson, G.C. (1983) The structure of two blood group A-active
glycosphingolipids with 12 sugars anda branched chain present
in the epithelial cells ofrat small intestine. J. Biol. Chem. 258,
9612–9615.
31. Bouhours, D., Hansson, G.C. & Bouhours, J.F. (1995) Structure
and genetic polymorphism of blood group A-active glycoshingo-
lipids oftherat large intestine. Biochem. Biophys. Acta 1255,
131–140.
32. Laferte, S., Prokopishyn, N.L., Moyana, T. & Bird, R.P. (1995)
Monoclonal antibody recognizing a determinant on type 2 chain
blood groupAandB oligosaccharides detects oncodevelopmental
changes in azoxymethane-induced rat colon tumors and human
colon cancer cell lines. J. Cell. Biochem. 57, 101–119.
33. Me
´
noret, A., Otry, C., Labarrie
`
re, N., Breimer, M.E., Piller, F.,
Meflah, K. & Le Pendu, J. (1995) Theexpressionof carbohydrate
blood groupantigens correlates with heat resistance. J. Cell Sci.
108, 1691–1701.
34. Norrsell, H., Bengtsson, J., Jovall, P.A. & Hansson, G.C. (1992)
N-linked glycopeptides with blood group determinants lacking
neuraminic acid from the epithelial cells ofrat small and large
intestine. Eur. J. Biochem. 203, 285–293.
35. Hallouin, F., Goupille, C., Le Cabellec, M., Bara, J. & Le Pendu,
J. (1997) ExpressionofAand H blood-group andof CD44 anti-
gens during chemical rat colonic carcinogenesis. Glycoconj. J. 14,
801–808.
36. Ravn, V. & Dabelsteen, E. (2000) Tissue distribution of histo-
blood group antigens. APMIS 108, 1–28.
37. Gil-Loyzaga, P., Pujol, R., Mollicone, R., Dalix, A.M. & Oriol, R.
(1989) Appearance ofBand H blood groupantigens in the
developing cochlear hair cells. Cell Tissue Res. 257, 17–21.
38. Astic, L., Le Pendu, J., Mollicone, R., Saucher, D. & Oriol, R.
(1989) Cellular expressionof H andBantigens in therat olfactory
system during development. J. Comp. Neurol. 289, 386–394.
Ó FEBS 2002 Rathisto-bloodgroupAenzyme (Eur. J. Biochem. 269) 4047
. Cloning of a rat gene encoding the histo-blood group A enzyme Tissue expression of the gene and of the A and B antigens Anne Cailleau-Thomas 1 ,Be ´ atrice Le Moullac-Vaidye 1 ,Je ´ zabel. transferase) activity of the enzyme product and allowed detection of a small UDP-Gal:Fuca1,2GalaGal transferase (B transferase) activity. The presence of the mRNA and of the A and B antigens was searched. various BDIX rat tissues. There was a general good concordance between the presence of the mRNA and that of the A antigen. Tissue distributions of the A and B antigens in the homozygous BDIX rat