Báo cáo Y học: Human bile salt-stimulated lipase has a high frequency of size variation due to a hypervariable region in exon 11 pot

Human bile salt-stimulated lipase has a high frequency of size
variation due to a hypervariable region in exon 11
Susanne Lindquist
1
, Lars BlaÈ ckberg
2
and Olle Hernell
1
Departments of
1
Clinical Sciences, Pediatrics and
2
Medical Biosciences, Medical Biochemistry, Umea
Ê
University, Sweden
The a pparent molecular mass of human milk bile salt-
stimulated lipase (BSSL) varies between mothers. The
molecular basis for this is unknown, but indirect evidence
has suggested the dierences to reside in a region o f
repeats located in the C-terminal part of the protein. We
here report that a polymorphism within exon 11 of the
BSSL gene is the e xplanation for t he molecular v ariants of
BSSL found in milk. By Southern blot hybridization we
analyzed the BSSL gene from mothers known t o have
BSSL of dierent molecular masses in their milk.
A polymorphism was found within exon 11, previously
shown t o c onsist o f 1 6 n ear i dentical repeats of 3 3 bp each.
We detected deletions or, in one case, a n insertion corres-
ponding to the variation in molecular mass of the BSSL
protein found in milk from the respective woman. Fur-
thermore, we f ound that 56%, out of 295 individuals

studied, carry deletions or insertions within exon 11 in one
or both alleles of the BSSL gene. Hence, this is a hyper-
variable region a nd the current understanding that exon 11
in the human BSSL gene encodes 16 repeats is an over-
simpli®cation and needs to be revisited. Natural variation
in the molecular mass of BSSL may have clinical impli-
cations.
Keywords: BSSL; lipase; human milk; r epeats; po ly-
morphism.
Bile salt-stimulated lipase (BSSL) or carboxyl ester lipase
is a digestive enzyme secreted from exocrine pancreas in
all species examined. BSSL has a broad substrate
speci®city a nd contributes to the hydrolysis of dietary
mono-, di-, a nd tri-acylglycerols and is responsible for
digestion of fat-soluble vitamin esters and cholesterol
esters in the small intestine. In some species, including
humans, the gene is also expressed in the lactating
mammary gland and the resulting protein is a constituent
of the milk [1,2]. Milk BSSL is a major reason why
breast-fed infants digest and absorb fat more ef®ciently
than formula-fed infants [3]. Moreover, BSSL has been
detected in low, but signi®cant, levels i n s erum [4]. The
function of BSSL in serum is unknown, but it has been
suggested to in¯uence the level of serum cholesterol [5,6].
Deduced from the cDNA sequence, the human BSSL
protein consists of 722 amino acids with a predicted
molecular mass of 76 kDa [7±10]. The protein is, however,
abundantly glycosylated and the apparent molecular mass
on SDS/PAGE has been estimated to 120±140 kDa
[11,12]. Human BSSL has a unique primary structure as

compared to other mammalian lipases. The N-terminal
part of the protein shows striking homology to acetylcho-
linesterase and some other esterases [7]. The C-terminal
part has b een reported t o consist of a unique structure
with 16 proline-rich, O-glycosylated repeats of 11 amino-
acid residues each [7±10]. T he biological function of the
repeated region is not fully understood. It has been shown
that the repeats protect the protein from denaturation by
acid and from proteolysis by pepsin or pancreatic prote-
ases in vitro [13,14]. It has also been shown that t he
O-glycosylation of the repeated sequences is important for
secretion of rat pancreatic BSSL [15]. On the other hand,
we and others have shown that t he repeats are completely
dispensable for the typical functional properties of BSSL,
i.e. catalytic activity, bile-salt activation, heparin binding,
heat stability, stability at low pH and resistance to
proteolytic inactivation [16±18].
The BSSL protein i s well c onserved between species, but
the number of proline-rich repeats varies, from three in
cow and mouse [19,20] t o 16 in the human [7±10]. The
salmon enzyme seems to be completely devoid of repeats
[21].
The human gene encoding BSSL spans 9.8 kb and
consists of 11 exons [22]. The gene has been mapped to
chromosome 9q34- qter and the BSSL locus was shown to
exhibit a high degree of p olymorphism [23]. A co rrelation
between BSSL genotype and serum cholesterol levels has
been proposed [24,25] but to our knowledge, the polymor-
phism in the BSSL gene has not been further characterized
until now.

The carboxyl ester lipase like (CELL) gene is a ubiqui-
tously transcribed pseudogene for BSSL [22,26]. The
sequence of the CELL gene is in some parts identical to
BSSL, i.e. exons 1, 8 a nd 9, whereas there are some major
differences in other parts. A 4.8-kb fragment, spanning
exons 2±7 in the BSSL ge ne, i s deleted in CELL .Thereare
also several base substitutions within exons 10 and 11.
A region in exon 11, encoding the proline-rich repeats,
differs between BSSL and CELL.HumanBSSL has
previously been shown to carry 16 repeats, although in t his
Correspondence to S. Lindquist, Department of Clinical Sci ences,
Pediatrics, Umea
Ê
University, SE-901 85 Umea
Ê
,Sweden.
Fax: + 4 6 90 123728, Tel.: + 46 90 7852128,
E-mail:
Abbreviations: BSSL, bile salt-stimulated lipase; CELL, carboxyl ester
lipase like; FAPP, feto-acinar pancreatic protein.
Enzyme: b ile salt-stimulated lipase (EC 3.1.1.3).
(Received 2 6 June 2 001, revised 5 November 200 1, accepted
8 November 2001)
Eur. J. Biochem. 269, 759±767 (2002) Ó FEBS 2002
paper we show that the number of r epeats can vary between
individuals. In CELL, this region has been described as a
hypervariable region and the o verall number o f repeats are
fewer compared to BSSL.
Naturally occurring variants of BSSL, differing in
apparent molecular mass, have been described in human

milk [27±29]. V ariants of h igher, as well as lower, molecular
mass than the most common 120-kDa variant were
detected. Occasionally, two different variants occurred
simultaneously in the s ame m ilk sample, e .g. a BSSL
variant of the most commonly occurring molecular mass
coexisted with a variant o f l ower or higher mass. The
differences in molecular masses w ere s hown to re side i n t he
C-terminal part of the protein, but could not be explained
by differences in carbohydrate content. Rather it was
speculated that it is the number of proline-rich repeats that
varies [28].
In t he p resent study we show that a hypervariable region
locatedtoexon11intheBSSL gene explain t he different
forms of BSSL found in human milk. Moreover, we show
that several m olecular variants occur and that some 56% of
the S wedish population do have variants different from the
most common one of 16 repeats. We speculate that this may
be of clinical signi®cance.
EXPERIMENTAL PROCEDURES
Collection of milk and blood samples
Human milk was collected via breast pump from healthy
women during their ®rst weeks of lactation. The milk was
either used immediately (for RNA preparation) or stored at
)20 °C until analyzed. Blood samples were collected in
vacutainerÒ tubes containing EDTA. The samples were
stored at )70 °C until DNA was isolated.
SDS/PAGE and Western blotting
One milliliter of h uman milk was centrifuged at 15 800 g for
10 min a nd the f at layer w as discarded. The s kimmed milk
was diluted 10-fold, after which 10 lLwasappliedtoa10%

SDS/PAGE [30]. After gel-electrophoresis, Western blotting
was performed using B SSL speci®c antibodies as pr eviously
described [28].
Probes for hybridization experiments
DNA fragments to be used as probes in Northern or
Southern blot hybridizations were obtained by PCR ampli-
®cation. The primers used for a mpli®cation of each probe
are list ed in Table 1. Plasm id pS146, c arrying the entire
BSSL cDNA [16], was used as template to create probe A
and probe B. Probe C was ampli®ed using a B SSL genomic
clone, pS453 (L. Hansson, Arexis AB, Mo
È
lndal, Sweden,
personal c ommunication) as template. PCR was performed
in a total volume of 30 lL (50 ng plasmid DNA, 10 m
M
Tris/HCl, p H 8.3, 1.5 m
M
MgCl
2,
50 m
M
KCl, 2 l
M
each of
dCTP, dGTP, and dTTP, 0 .82 l
M
[a-
32
P] dATP, 7.5 pmol

of each primer, and 2.5 U of Ta q polymerase). The reactions
were carried out for 30 cycles with d enaturation at 9 4 °Cfor
30 s, annealing at 55 °C for 1 min a nd extension at 72 °Cfor
1 min. The program ended with a n elongation a t 72 °C
for 7 min. The PCR products were puri®ed on a Sephadex
G-50 ÔNick columnÕ (Amersham Pharmacia Biotech,
Uppsala, Sweden) before used in hybridization experime nts.
RNA isolation and Northern blot hybridization
Total RNA was isolated from fresh human milk samples as
previously described [31]. RNA hybridization was per-
formed essentially as described in Sambrook et al.[32].
Approximately 20 lg of each RNA preparation was
separated on 1% agarose gels, blotted onto Hybond-N
®lters (Amersham International plc., Buckinghamshire,
UK) and hybridized to a [
32
P]dATP-labelled probe. After
hybridization, ®lters were washed and signals visualized
using a Molecular I mager (Bio-Rad L aboratories, Hercules,
CA).
32
P-Labelled k HindIII digested DNA was used as
molecular mass standard on the R NA gels.
DNA isolation and Southern blot hybridization
Genomic DNA was isolated from 10 m L EDTA-blood as
previously described [33]. For Southern blot analysis, 10 lg
of DNA was digested with appropriate restriction
enzyme(s). Digested DNA was separated on an agarose
gel, transferred to a Hybond-N ®lter (Amersham) and
hybridized to a [

32
P]dATP-labelled probe as described in
Sambrook et al . [32]. Pre-hybridization, and hybridization,
was performed at 42 °C in s tandard solutions, supplemen-
Table 1. Oligonucleotide primers used t o a mplify D NA probes. Position s refer to the sequence of the BSSL gene subm itted to the EMB L databank
[22], a ccession no. M94579.
Probe Primer sequence Positions
Probe A
BSSL03 5¢-GACCCCAACATGGGCGACTC-3¢ 10621±10640
BSSL04 5¢-GTCACTGTGGGCAGCGCCAG-3¢ 10793±10774
Probe B
SYM2677
a
5¢-tctagaagcttGGCGCCGTGTACACAGAAGGTGGG-3¢ 4047±4069
SYM2133: 5¢-GTTGGCCCCATGGCCGGACCCCAT-3 4752±4729
Probe C
SYM2143
a
5¢-cgggatccGAAGCCCTTCGCCACCCCCACG-3¢ 10201±10222
BSSL05 5¢-GGCCTCGTGGTGGGAGGCCCTT-3¢ 10336±10357
a
The ®rst 11 bases in primer SYM2677 and the ®rst eight bases in primer SYM2143 are linkers with no relevance for the application in this
paper.
760 S. Lindquist et al. (Eur. J. Biochem. 269) Ó FEBS 2002
ted with 50% formamide. After washing the ®lter, signals
were visualized using Molecular Imager (Bio-Rad).
32
P-Labelled k HindIII DNA was r un in parallel as a size
marker.
Cloning and DNA sequencing

The region o f r epeats in BSSL exon 11 was PCR ampli®e d
using t he Platinum Ò Pfx DNA polymerase (Life Technol-
ogies Inc., Gaithersburg, MD, USA). To improve ampli®-
cation of the extremely GC-rich repeats, betaine was a dded
to each reaction to a ®nal concentration of 2
M
(Sigma,
St Louis, MO, U SA). A pair of primers, referred to as
BSSL 12 and BSSL 14, was designed to cover the entire
sequence of repeats (BSSL 12: 5¢-ACCAAC TTCCT
GCGCTACTGGACCCTC-3¢;BSSL14:5¢-GGAGCC
CCTGGGGTCCCACTCTTGT-3¢). The P CR started with
adenaturationstep(96°C, 5 min) followed by 35 cycles
with denaturation (96 °C, 45 s) and annealing/elongation
(68 °C, 5 min). The reaction terminated b y a ®nal i ncuba-
tion at 68 °C for 10 min.
The PCR p roducts were sep arated on an agarose gel and
the fragments to be cloned were recovered using Gene-clean
II (BIO 101, Carlsbad, CA, USA). Cloning was p erformed
using t he pGEM Ò-T easy vector system II (Promega Co.,
Madison, WI, USA). Before ligation into the pGEMÒ-T
easy vector, the PCR fragments had to be modi®ed using the
A-tailing p rocedure for blunt-ended PCR fragments, as
recommended by Promega.
The cloned fragments were sequenced on both strands
using t he Big Dye terminator kit (PE Applied Biosystems,
Foster City, CA) supplemented with betaine to a ®nal
concentration of 1
M
(Sigma). BSSL 12 or BSSL 14

(described above) were used as primers, and t he DNA
was a mpli®ed for 30 cycles with denaturation a t 9 8 °C(30 s)
and annealing/elongation at 60 °C (5 min). The reactions
were analyzed on an ABI PRISM 377A DNA sequencer
(PE Applied Biosystems).
RESULTS
Expression of different BSSL variants in human milk
To con®rm the described heterogeneity in m olecular m ass o f
milk derived BSSL and select representatives for different
BSSL phenotypes we screened milk samples from nine
different mothers. The m ilk proteins were separated on
SDS/PAGE, electroblotted and immunostained with BSSL
speci®c antibodies (Fig. 1). The most c ommonly occurring
variant of BSSL migrated with an appar ent molecular mass
of  120 kDa (donors D11, D8, D7). A variant with an
apparently lower m olecular mass, i.e.  100 kDa, was
found in some milk samples, either as the only on e (donor
D2) or coexpressed with a variant of the most common
molecular m ass (donors D6 and D3). A single mother
(donor D1) had a varian t with higher molecular m ass
(160 kDa) than the most common one. This mother also
carried the 100 kDa variant in her milk. Donors D4 and D5
carried only the 120-kDa BSSL variant in their milk samples
(data not shown).
Analysis of BSSL transcripts in milk cells
Northern blot hybridization was performed on RNA
isolated from m ilk from four different mothers, D1 and
D6±D8 ( Fig. 2). A 2.8-kb transcript was detected in RNA
from mother s D7 a nd D8 when a fragment c omplementary
to a sequence immediately upstream t he repeats i n exon 1 1

was used as a probe (probe A; Fig. 3). A slightly shorter
transcript,  2.7 kb, was d etected in RNA isolated from
mother D6. The RNA isolated from mother D1 contained
two hybridizing transcripts, 2.7 and 3.0 kb in size, respec-
tively. To e xc lude the possibility that probe A had failed t o
detect any possible truncated BSSL mRNA we used
another probe, complementary to exon 2 to exon 4 in the
BSSL cDNA (probe B; Fig. 3). However, identical results
as with probe A w ere obtained using probe B in t he
Northern blot (data not shown).
Genetic variation occurs in exon 11 of the
BSSL
gene
To explore the possibility that genetic rearrangement(s)
within the BSSL gene might explain the occurrence of
Fig. 1. Naturally occurring variants of the BSSL protein in human milk.
Milk proteins from seven dierent dono rs (D6, D11, D1, D8, D3, D2
and D7) were separated on a 10% SDS/PAGE and immunostained
with BSSL speci®c antibodies. The molecular mass standards are
shownontheright.
Fig. 2. Northern blot analysis of total RNA from milk cells isolated
from four dierent mothers (D1, D6, D7 and D8). The RNA was
hybridized to a BSSL speci®c probe, P robe A. HindIII-cut k was u sed
as the m olecular mass m arker (M).
Ó FEBS 2002 Molecular mass variants of BSSL in human milk (Eur. J. Biochem. 269) 761
molecular mass variants of B SSL, we isolated DNA from
eight of the mothers (D1±D8) and p erformed Southern blot
hybridizations (Fig. 4). PstI digested DNA was hybridized
to a p robe complementary to a sequence in BSSL exon 11
(probe A; Fig. 3). According to the published BSSL

sequence [7±10] this probe was expected to hybridize to a
731-bp PstI fragment carrying all 16 repeats. Accordingly, a
0.7-kb PstI fragment was detected in all DNA samples
isolated from mothers c arrying the most common variant of
BSSL in their m ilk, i.e. D3±D8. However, this 0.7-kb PstI
fragment was not found in DNA from m other D2, carrying
only the low molecular mass variant in her m ilk. I nstead, D 2
and also the other mothers carrying low molecular mass
variants in their milks (D1, D3 and D6) carried a shorter
PstI fragment (0.6 kb). The mother with a high molecular
mass BSSL variant in her milk (D1), carried a longer
hybridizing PstI fragment (0.9 kb) not detected in any other
DNA s ample. Also a third PstI fragment (0.7 kb) was
detected in DNA from mother D1. In contrast to the 0.9
and 0.6 kb fragments t his 0.7-kb f ragment did not correlate
with any BSSL protein variant in milk from mother D1.
When the DNA samples were digested with Eco RI and
hybridized to probe C (Fig. 3) the hybridizing fragments
corresponded t o the products obtained with PstIdigestion
and probe A (Fig. 4b). DNA isolated from donors
expressing the most common BSSL variant in milk (D3±
D8) yielded a 2.2-kb EcoRI fragment when hybridized to
probe C. A shorter fragment ( 2.1 kb) was detected in DNA
isolated from donors c arrying the 100-kDa variant of BSSL
in milk (D1±D3 and D6). In DNA isolated from mother
D1, thre e EcoRI fragments were foun d to hybridize to
probe C (2.1, 2.2, and 2.4 kb, respectively).
Several other appropriate restriction enzymes and DNA
probes were used to cover the entire BSSL gene, looking for
additional genetic rearrangements. However, no genetic

variation was detected in any other part of the BSSL gene,
neither upstream nor downstream the repeats in exon 11
(data not shown). Hence, we conclude that rearrangements
(deletions and insertions) occur within the region carrying
the repeats in exon 11 of the BSSL gene.
PCR ampli®cation, cloning and DNA sequencing
of different BSSL alleles
To further characterize some of the rearrangements in BSSL
exon 11, we used PCR to amplify the region carrying the
repeats in DNA isolated from two mothers (D1 and D2)
(Fig. 5). According to t he published sequence [7±10] a
678-bp fragment was expected to amplify if all the 1 6 r epeats
(33 bp each) is present and if there is no deletions or
insertions. The results of the PCR con®rmed the Southern
blot results, i.e. both mothers carry a deletion within one
(D1) or both ( D2) alleles of their BSSL gene. In addition,
D1 also carries an insertion within a nother allele, shown by
the ampli®cation of a fragment  0.9kbinsize.Alsoin
concert w ith d ata f rom Southern blot, a third fragme nt
corresponding to the size of the wild-type a llele (678 bp) was
detected in DNA from D1.
The 0.6-kb PCR fragments, expected to carry the
proposed deletions, were cloned from each of t he samples
(D1 and D2) and the DNA sequenced. When the
sequences were aligned to the previously published
DNA s equence, it was con®rmed that t he deletions had
occurred within th e region of repeats (Fig. 6). However,
the deletions were not identical between the two samples.
The f ragment t hat was sequenced from mother D1 was
shown to carry a 98-bp deletion that changes the reading

frame of the gene and predicts a premature translational
stop after 632 amino acids (Fig. 7). The sequence f rom
mother D2 was essentially identical to D1 except that one
basepair less was deleted, i.e. a 97-bp deletion was found.
This difference predicts an even earlier translational stop,
i.e. after 610 amino acids. In both c ases t he d eletion
changes the reading frame and predicts a new C-terminal
tail (RAAHG). Besides the d eletions, t he seq uences of D1
and D 2 w ere i dentical to the published sequence e xcept
for one base substitution that does not affect the protein
sequence (Fig. 6).
The
BSSL
gene contains a hypervariable region
in exon 11
To estimate the frequency of the BSSL polymorphism in a
larger population, DNA was isolated from 2 95 healthy
blood donors, digested with PstI and hybridized to probe
A (Fig. 3) in Southern blot experiments. A high frequency
of variation was found. Only 131 out of the 295 (44%)
DNA samples showed a restriction pattern corresponding
to the published sequence, i.e. a PstI fragment  731 bp in
size. In 23 out of 295 ( 8%) analyzed DNA samples, a PstI
fragment considerably shorter than 731 bp was detected.
As many as 41% (121/295) o f the analyzed DNA samples
showed a heterozygous pattern with one PstI fragment
 731 bp in size and another fragment considerably
shorter. An increased length of the actual PstI fragment
was found in 21 out of 295 (7%) of the analyzed DNA
samples.

DISCUSSION
The BSSL locus is known t o exhibit a high degree o f
polymorphism [23], but whether this polymorphism affects
the BSSL coding region has not previously been shown.
Therefore, in the present paper we have investigated if the
Fig. 3. Schematic drawing of the genetic organization of the human BSSL gene, m odi®ed f rom Lidberg et al .[22].Exons are shown as boxes and
numbered 1±11. The r epeated reg ion in exon 1 1 (re p) is hatch ed. Horiz ontal bars show the p osition o f sequ ence homology t o probes A , B and C ,
used for hybridization experiments. C leavage sites f or PstI(P)andEcoRI (E) are marked.
762 S. Lindquist et al. (Eur. J. Biochem. 269) Ó FEBS 2002
occurrence of different BSSL variants in human milk is du e
to genetic variation with in the BSSL gene.
We collected milk samples and isolated RNA and DNA
from nine lactating mothers. Southern blot hybridization
experiments con®rmed the occurrence of allelic variance in
the BSSL gene. The variations were exclusively found
within a Pst I fragment covering a region of direct repeats i n
exon 11, and in each woman a correlation to the molecular
mass of BSSL in milk was evident. Mothers known to have
low molecular mass variant(s) of the BSSL protein i n their
milk were shown to carry a deletion,  0.1 kbinsize,within
this PstI fragment. Mothers with two differen t BSSL
variants in t heir milk, e.g. the most common 120-kDa
variant together with o ne of l ower mass, carried the deletion
in one of the alleles, whereas t he mother with only the low
molecular mass variant in milk (D2) carried deletions in
both alleles. In DNA isolated from mother D1, known t o
express a high molecular m ass variant (160 kDa) together
with a low molecular mass variant in her m ilk, PstI
fragments of 0 .6 and 0 .9 kb were detected. The sizes of these
fragments c orrespond to a 0.1-kb deletion i n one allele, and

a 0 .2-kb i nsertion in another a llele, and are likely to encode
the low and high molecular mass variants detected in milk
from D1, respectively. However, a third, unexpected PstI
Fig. 4. Southern blot hybridization. (A) DNA
isolated from d ono rs D1±D8 w as digested
with PstI a nd h ybrid ized to probe A (T ab le 1,
Fig. 3). Hybridizing fragments shorter t han
0.56 kb correspond t o fragments within the
pseudogene CELL. HindIII-cut k was used
as molecular m ass marker (M). (B) DN A
isolated from th e same donors a s above was
digested with EcoRIandhybridizedtoprobe
C(Table1,Fig.3).HindIII-cut k was used
as molecular m ass marker (M).
Ó FEBS 2002 Molecular mass variants of BSSL in human milk (Eur. J. Biochem. 269) 763
fragment, was detected in DNA from mother D1. This
fragment was  0.7 kb, corresponding to a fragment size
probably carrying the most commonly occurring 16 repeats.
This fragment probably originates from a duplication of the
BSSL gene, or at least of the 5¢ end, in one of the alleles.
This duplicated copy was not expressed as on ly two variants
of the protein were found in milk from this donor.
A DNA fr agment covering the deletions in exon 11 was
PCR ampli®ed, cloned and sequenced from two different
mothers (D1 and D2). The DNA sequences con®rmed t he
deletions and the deduced amino-acid sequences predicted
BSSL variants of considerably lower molecular m ass, i.e. the
variants were predicted to be truncated after 632 and 610
amino acids, respectively. Hence, these BSSL variants are
truncated within the region of proline r ich repeats and the

number o f repeats is decreased from 16 t o 8.5 and 6.5,
respectively, in the D1 and D2 variants. In both variants, a
new C-terminal sequence consisting of ®ve a mino acids
(RAAHG) is created due to the delet ions. In the wild-type
protein (the most common v ariant), the C -terminal consists
Fig. 6. The DNA sequence of the rep eated
region c arrying a de letion in exon 11 from
mother D1 and D2 w as aligned t o the published
BSSL sequence (w t). The repeats are
numbered 1±16 according to the wt sequence.
Alignments were performed using the
program
BESTFIT
from the U niversity of
Wisconsin
GCG
software packa ge. Dots
represent gaps that were inserted to improve
alignment. The position of primers 12 and 14
used f or ampli®cation and sequencing of the
fragments a re marked. A n asterisk ( *) marks
the position of the single base subs titution
detected in D1 an d D2.
Fig. 5. PCR analysis of BSSL exon 11. DNA f rom mother D1 and D2
was ampli®ed using a pair of primers covering the entire region of
repeats i n t he BSSL ge ne. Three independent reactions were run from
each mother. Lane 1±3, D1; lane 4±6, D2. The shortest fragments,
 0.6 kb, we re subsequently cloned and seque nced.
764 S. Lindquist et al. (Eur. J. Biochem. 269) Ó FEBS 2002
of the 16 repeats followed by a hydrophobic tail of 11 amino

acids. The low molecular mass variants expressed by
mothers D1 and D2 d id not react with speci®c antibodies
directed towards t his tail ( M. Stro
È
mqvist, AstraZeneca
R&D,Mo
È
lndal, Sweden, personal communication) con-
®rming a different sequence of t he tail. The f unction of the
tail has previously been discussed in the literature. Deletion
of the tail by in vitro mutagenesis of the human enzyme was
shown to signi®cantly decrease expression of the protein,
presumably by affecting mRNA s tability [16]. F rom studies
on the crystal structure of bovine BSSL it was concluded
that the terminal six hydrophobic amino acids physically
block a putative oxyanion hole at t he active site. Calcula-
tions indicated that removal of this hexapeptid e exposes a
large hydrophobic area on the protein surface suggesting
that displacement of this re gion can play a role in the
stability and function of BSSL [34].
The s ize o f the BSSL transcript has p reviously been
estimated to be  2.5 or 2.9 kb [7,9]. We detected BSSL
transcripts of 2.8 kb in Northern blots p erformed on RNA
from two mothers (D7 and D8) known to have the most
common BSSL genotype and phenotype i n milk. RNA
isolated fr om mother D1, expressing t he high molecular
mass varian t together with a low molecular m ass v ariant of
BSSL in milk, carried two transcripts that hybridized to the
BSSL-speci®c probe. The sizes of these two transcripts were
estimated t o b e 2.7 and 3 .0 kb, respectively. Accordingly, we

expected to ®nd two transcripts in RNA from mother D6,
known to have two BSSL variants (100 + 120 k Da) i n
milk, and to carry t he exon 11 deletion in one of the BSSL
alleles. However, only one transcript was detected. The
band representing this transcript is however broad and we
believe that the resolution of the gel was insuf®cient t o
separate the two proposed transcripts.
The frequency of variation in exon 11 of the BSSL gen e
was determined by Southern blot experiments with DNA
isolated from 295 blood donors. When compared to the
published s equence [7±10] 56% of the individuals examined
carried genetic variations within the repeats. These data
con®rm that BSSL is located in a hypervariable region [23]
but also shows t hat the polymorphism is due to deletions or
insertions within the BSSL coding sequence. Hence, we
conclude that exon 11 in th e BSSL gene consists of a
hypervariable r egion and that the current understanding
that exon 11 of the human gene encodes 16 proline-rich
repeats is an oversimpli®cation and needs to be revisited.
This high frequency of variation in the BSSL gene
corresponds very well with a previous study on incidence
of molecular forms of BSSL in human milk [29]. This study
showed that 50% of the milk samples contained BSSL
variants with a molecular m ass different to the m ost
common variant.
An onco-fetal variant of BSSL, d enoted feto-acinar
pancreatic p rotein (FAPP), has been d etected in human
embryonic and fetal pancreas and in pancreatic tumoral
cell lines [35,36]. FAPP and BSSL are structurally closely
related, but are distinguished by a monoclonal antibody

directed towards a fucosylated epitope, present on FAPP
but not on BSSL [37]. Compared to BSSL, FAPP h as
lower enzymatic activity against ester substrates, and is
poorly secreted [36,37]. The cDNA sequence of FAPP is
identical to that of BSSL except for a 330-bp deletion in
the C-terminal repeated region [38,39]. T he fact that we
now show that  50% of a Swedish population c arry a
deletion in the r epeated region of the BSSL gene makes i t
tempting to speculate that FAPP is identical to a naturally
occurring low molecular mass variant of BSSL. The
characteristic FAPP epitope should then result from tissue
speci®c glycosylation, rather than structural features of the
protein. If so, th e concept of F APP being an onco-fetal
variant of BSSL, exclusively expressed in proliferating cells
such as embryonic and fetal panc reas as well a s pancreatic
tumoral cells, should b e r e-evaluated. The human hepa-
toma cell line HepG2 also expresses a BSSL isoform of
lower molecular m ass [ 40]. The cDNA sequence of t his
isoform contained only one ÔrepeatÕ.
The obvious question is, of course, whether there are
biological phenotypes associated with speci®c BSSL vari-
ants? As mentioned above, it has been proposed that there is
no signi®cant difference in enzymatic activity, bile salt
stimulation, pH stability a nd temperature stability b etween
BSSL of the most common molecular mass and variants of
lower or higher m ass [27,28]. However, some l ow molecular
mass variants with only half the speci®c activity compared
to the most common variant have been isolated and the
concentration of BSSL was co nsiderably lower in milk from
mothers carrying only low molecular mass variant(s) [28].

A possible explanation of th ese somewhat contradictory
results could be the presen ce o r absence of t he most
C-terminal 11 amino acids, re ferred to as t he tail. Two l ow
molecular mass variants characterized in this paper were
bothshowntolacktheÔnormalÕ C-terminal tail, whereas
Stro
È
mqvist et al. [ 28] showed that the tail is p resent in other
low molecular mass variants. From the crystal structure of
bovine BSSL the tail was suggested to be involved in the
active site machinery [34].
Finally, a positive correlation has b een demonstrated
between BSSL activity in serum, assayed as cholesterol
esterase activity, and serum cholesterol levels [5,6]. More-
over, in vitro BSSL was shown to transform larger LDL
particles to smaller, more atherogenic LDL particles [41].
Considering the data presented in the present paper, it is
interesting to note that an association between BSSL
genotype and serum lipid levels has been suggested [24,25].
Taken together, we have shown that the molecular mass
variants of BSSL found in milk results from a polymor-
phism in the BSSL gene. This strongly suggests that BSSL
variants described in other tissues, such as the onco-fetal
protein FAPP, is due to the s ame frequently occurring
Fig. 7. Comparison of the deduced amin o-acid sequences of the repeats
in B SSL from the published sequence carrying 16 repeats (w t), and the
shorter v ariants from mother D1 and D2.
Ó FEBS 2002 Molecular mass variants of BSSL in human milk (Eur. J. Biochem. 269) 765
genetic variation. Whether this polymorphism is correlated
to activity of the enzyme a nd to serum lipid levels is

currently under investigation.
ACKNOWLEDGEMENTS
We are grateful to Yvonne Andersson for excellent technical assistance
andtoMatsStro
È
mqvist for fruitful d iscussions. Grants from t he
Swedish Medical Research Council (05708 and 12721), Astra-Ha
È
ssle
AB, PPL therapeutics, Margarinindustrin, Stiftelsen Oskarfonden,
Va
È
sterbotten County Council, and The Swedish Society for Medical
Research (postdoctoral f ellowship to S . L .) supported this w ork.
REFERENCES
1. Hernell, O. & O livecrona, T . ( 1974) Human milk l ipases, II: Bile
salt-stimulated lipase. Bioc him. Biophys. Acta 369, 234±244.
2. Bla
È
ckberg, L., A
È
ngquist, K.A., Hernell, O. (1987) Bile- salt-stimu-
lated lipase i n human milk: evidence f or its synthesis in the lac-
tating mammary g land. FEBS Lett. 21 7 , 37±41.
3. Bernba
È
ck, S., Bla
È
ckberg, L., Hernell, O. (1990) The complete
digestion of human milk triacylglycerol in vitro requires gastric

lipase, pancreatic colipase-dependent lipase, and bile salt-stimu-
lated lipase. J. Clin. Invest. 85, 1221±1226.
4. Bla
È
ckberg, L., Blind, P.J., L jungberg, B., Hernell, O. (1985) On the
source of bile salt-stimulated lipas e in human milk: a study b ased
on serum concentration s as determined by sandwich enzyme-
linked immunosorbent assay technique. J. Pediatr. Gastroenterol.
Nutr. 4, 441±445.
5. Brodt-Eppley,J.,White,P.,Jenkins,S.,Hui,D.Y.(1995)Plasma
cholesterol esterase l evel is a determinant for an atherogenic
lipoprotein pro®le in normolipidemic human subjects. Biochim.
Biophys. A cta 1272, 6 9±72.
6. Caillol, N., Pasqualini, E., Mas, E., Valette, A., Verine, A.,
Lombardo, D . (1997) Pancreatic bile salt-dependent lipase a ctivity
in serum of normolipidemic patients. Lipids 32, 1147±1153.
7. Nilsson, J., B la
È
ckberg, L., Carlsson, P., Enerba
È
ck, S., Hernell, O.,
Bjursell, G. (1990) cDNA cloning of human-milk bile-salt-stimu-
lated lipase a nd evid ence fo r i ts identity to pancreatic carb oxylic
ester hydrolase. Eur. J . Biochem. 192, 543±550.
8. Hui, D.Y. & Kissel, J.A. (1990) Sequence identity between human
pancreatic cholesterol esterase and bile salt-stimulated milk lipase.
FEBS Lett. 276, 131±134.
9. Baba, T., Downs, D., Jackson, K.W., Tang, J ., Wang, C S. (1991)
Structure of human milk bile salt activated lipase. Biochemistry 30 ,
500±510.

10. Reue,K.,Zambaux,J.,Wong,H.,Lee,G.,Leete,T.H.,Ronk,
M., S hively, J.E., Sternby, B., Borgstro
È
m, B., Ameis, D., Schotz,
M. (1991) cDNA cloning of carboxyl ester lipase from human
pancreas reveals a unique proline-rich repeat unit. J. Lipid Res. 32,
267±276.
11. Bla
È
ckberg, L. & Hern ell, O. (1981) The bile-salt-stimulated lipase
in human milk. Puri®cation and charact eriz at ion. Eur. J. Biochem.
116, 221±225.
12. Wang, C S. & Johnson, K. (1983) Puri®cation o f human m ilk bile
salt-activated lipase. Anal. Biochem. 133, 457±461.
13. Loomes, K.M. (1995) Structural organiz ation of human bile-
salt-activated lipase probed by limited proteolysis and expres-
sion of a recombinant truncated variant. Eur. J. Bioc hem. 23 0,
607±613.
14. Loomes, K .M., Senior, H.E., West, P.M., Robe rton, A .M. (1999)
Functional prote ctive role for mucin gl ycosylated repetitive
domains. Eur. J. Biochem. 266, 105±111.
15. Bruneau, N., Nganga, A., Fisher, E.A., Lombardo, D. (1997)
O-Glycosylation of C-terminal tandem-repeated sequences regu-
lates the secretion of rat pancreatic bile salt-dependent lipase.
J. Biol. Chem. 272, 27353±27361.
16. Hansson, L., Bla
È
ckberg, L., Edlund, M ., Lundberg, L ., Stro
È
mq-

vist, M., Hernell, O. (1993) Recombinant human milk bile salt-
stimulated lipase . J. Biol . Chem. 268, 26692±26698.
17. Downs, D ., Xu, Y Y., Ta ng, J ., Wang, C S. ( 1994) Prol ine-rich
domain and glycosylation are no t essential for the en zymic activity
of bile salt-activated lipase. Kinetic studies of T-BAL, a truncated
form o f the enzyme, exp ressed in Escherichia coli. Biochemistry 33,
7979±7985.
18. Bla
È
ckberg, L ., Stro
È
mqvist, M., Edlund, M ., Juneblad, K., Lund-
berg, L., Hansson, L., Hernell, O. (1995) Functional properties are
retained in the absence of glycosylation and the unique proline-
rich repeats. Eur. J. Biochem. 228 , 817±821.
19. Kyger, E.M., W iegand, R.C., Lange, L .G. ( 1989) Cloning of the
bovine pancreatic cholesterol esterase/lysopho spholipase.
Biochem. Biop hys. Res. Commun. 164, 1302 ±1309.
20. Lidmer, A S., Kannius, M., Lundberg, L., Bjursell, G., N ilsson, J.
(1995) Molecular cloning and characterization of the mouse
carboxyl este r lipase gene and evidence for e x pression in the la c -
tating mammary g land. Genomics 29, 1 15±122.
21. Gjellesvik, D.R., Lorens, J.B., Male, R. (1994) Pancreatic car-
boxylester lip ase from A tlantic salmon ( Salmo salar). cD NA
sequence and computer-assisted modelling of tertiary structure.
Eur. J. Bioc hem. 226, 603±612.
22. Lidberg, U., Nilsson, J., Stro
È
mberg,K.,Stenman,G.,Sahlin,P.,
Enerba

È
ck, S., Bjursell, G. (1992) Ge nom ic organization, sequ en ce
analysis, and chromosomal localization of the human carboxyl
ester lipase (CEL)geneandtheCEL-like (CELL) g en e. Genomics
13, 630±640.
23. Taylor, A.K., Zambaux, J.L., Klisak, I ., Mohandas, T., Sparkes,
R.S., Schotz, M.C., Lusis, A.J. (1991) Carboxyl ester lipase: a
highly polymorphic locus on human chromosome 9qter. Ge nomics
10, 425±431.
24. Hui, D.Y. (1996) Molecular biology of enzymes involved with
cholesterol ester hydrolysis in m ammalian tissues. Biochim.
Biophys. A cta 1303, 1±14.
25. Aleman-Gomez, J.A., Colwell, N.S., V yas, K., Borecki, I.,
Shonfeld, G ., Lange, L .G., Kumar, V.B. (1999) Restriction frag-
ment length polymorphism o f the human panc re atic cholesterol
esterase gene and i ts association with s erum lipid levels. Life Sci.
64, 2419±2427.
26. Nilsson, J., Hellquist, M., Bjursell, G. (1993) The human carboxyl
ester lipase-like (CELL) ge ne is ubiq uitously expresse d and con -
tains a hypervariable region. Genomics 17, 416±422.
27. Swan, J.S., Homan, M .M., Lord, M.K., Poechmann, J.L. (1992)
Two forms of human milk bile-salt-stimulated lipase. Biochem.
J. 283, 119±122.
28. Stro
È
mqvist, M ., Hernell, O., Hansson, L., Lindgren, K., Skytt, A
Ê
.,
Lundberg,L.,Lidmer,A S.,Bla
È

ckberg, L. ( 1997) Naturally
occurring variants of human milk bile salt-stimulated lipase. Arch.
Biochem. Biop hys. 347, 30±36.
29. McKillop, A.M., O'Hare, M.M.T., Craig, S., Dodge, J.A.,
Halliday, H.L. (1998) Incidence of molecular forms of bile salt-
stimulated lipase i n preterm and term h uman milk. Pediatr. Res.
43, 101±104.
30. Laemmli, U.K. (1970) Clea vage of structural prote ins during the
assembly of the head of b ac teriophage T4. Nature 22 7, 680±685.
31. Lind quist, S., Hansson, L., Hernell, O., Lo
È
nnerdal, B., Normark,
J., Stro
È
mqvist, M., Bergstro
È
m, S. (1994) Isolation of mRNA and
genomic DNA from epithelial cells in the human milk and
ampli®cation by PCR. Bio t echniq ues 17, 6 92±694.
32. Sambrook, J., Fritsch, E.F., M aniatis, T. (1989) Molecular Clon-
ing: a Laboratory Manual, 2nd ed n. Cold Spring. H arbor Labo-
ratory Press, Cold Spring Harbor, New York.
33. Balciun iene, J., Johansson, K., Sandgren, O., Wachtm eister, L.,
Holmgren, G., Forsm an, K . ( 1995) A gene for autosomal domi-
nant progressive cone dystrophy (CORD5) maps to chro mosome
17p12-p13. Genomics 30, 281±286.
766 S. Lindquist et al. (Eur. J. Biochem. 269) Ó FEBS 2002
34. Chen, J.C H., Mier cke, L.J.W., K rucinski, J., Starr, J.R., Saenz,
G., Wang, X., Spilburg, C.A., Lange, L .G., Ellsworth, J.L.,
Stroud, R.M. (1998) Structure of bovine pancreatic cholesterol

esterase at 1.6 A
Ê
: novel structural features involved in lipase
activation. Biochemistry 37, 5 107±5117.
35. Escribano, M.J. & Imperial, S. (1989) Puri®cation and molecular
characterization of FAP, a feto-acinar protein associated with the
dierentiation of human pancreas. J. Biol. Chem. 264, 21865±
21871.
36. Mazo, A., Fujii, Y., Shimotake, J., Escribano, M.J. ( 1991)
Expression of fetoacinar pancreatic (FAP) protein in the pancre-
atic human tumor cell line BxPC-3. Pancreas 6, 37±45.
37. Mas, E ., Abouakil, N., Roudani, S., Mirall es, F., Guy-Crotte, O.,
Figarella, C ., Escribano, M .J., Lombardo, D . (1993) H uman
fetoacinar pancreatic p rotein: an on cofetal glycoform o f the nor-
mally secreted pancreatic bile-salt-dependent lipase. Biochem.
J. 289, 609±615.
38. Pasqualini, E., Caillol, N., Panicot, L., Mas, E., Lloubes, R.,
Lombardo, D. (1998) Molecular cloning of the oncofetal isoform
of the human pancreatic bile salt-dep endent lip ase. J . Biol. C hem .
273, 282 08±28218.
39. Pasqualini, E., Caillol, N., Panicot, L., Valette, A., Lombardo, D.
(2000) Expression of a 70-kDa immunoreactive f orm of b ile salt-
dependent lipase by human pancreatic tumoral M ia PaCa-2 cells.
Arch. B iochem. Biophys. 375, 90±100.
40. Ve
Â
rine, A., Bruneau, N., Valette, A., Le Petit-Thevenin, J.,
Pasqualini, E., Lombardo, D. ( 1999) Immunodetection and mole-
cular cloning of a bile-salt-dependent lipase isoform in HepG2
cells. Bioche m. J. 342, 179±187.

41. S hamir, S., Johnson, W.J., Morlock-Fitzpatrick, K., Zolfaghari,
R.,Ling,L.,Mas,E.,Lombardo,D.,Morel,D.W.,Fisher,E.A.
(1996) Pancreatic carboxyl ester lipase: a c irculating enz yme tha t
modi®es normal and oxidized lipoproteins in vitro. J. Clin. Invest.
97, 1696 ±1704.
Ó FEBS 2002 Molecular mass variants of BSSL in human milk (Eur. J. Biochem. 269) 767