Human Reproduction Vol.19, No.7 pp. 1647±1654, 2004
Advance Access publication June 4, 2004
DOI: 10.1093/humrep/deh193
Expression of leukemia inhibitory factor and its receptor is
not altered in the decidua and chorionic villi of human
anembryonic pregnancy
Hsin-Fu Chen1, Kuang-Han Chao1, Jin-Yuh Shew2, Yu-Shih Yang1 and Hong-Nerng Ho1,3
1
Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, College of Medicine and the
Hospital, National Taiwan University, 2Department of Biochemistry, College of Medicine, National Taiwan University, 3Graduate
Institute of Immunology, College of Medicine, National Taiwan University Corresponding author addressed at: National Taiwan
University Hospital, Department of Obstetrics and Gynecology, No.7 Chung-Shan S. Road, Taipei 100, Taiwan. Tel: 886 2
23123456 extension 5161; Fax: 886 2 23418557; E-mail: [email protected]
BACKGROUND: Uterine expression of leukemia inhibitory factor (LIF) is absolutely essential for mouse, and
critical for human, embryo implantation. However LIF is not required for post-implantation development of mouse
embryo. The objective of this study was to examine the role of LIF system in post-implantation stage of human
pregnancy. METHODS: Tissues from 25 patients with anembryonic pregnancy (AP; blighted ovum) and 25
matched patients with normal pregnancy (NP) were collected. LIF and its receptor b (LIF-Rb) expression in the
decidua and chorionic villi were analyzed by semi-quantitative reverse transcription and polymerase chain reaction
(RT±PCR), real-time quantitative PCR and immunohistochemical study. RESULTS: LIF mRNA levels were not different either between different tissues (decidua vs chorionic villi) or between different patients (NP vs AP). LIF-Rb
mRNA levels were signi®cantly higher in chorionic villi than in decidua but were not different between NP and AP.
Immunohistochemical staining supported these ®ndings and showed a predominate expression of LIF-Rb in the
trophoblast cells. CONCLUSIONS: This study concluded that at early human post-implantation stage, LIF is
produced from both decidua and chorionic villi and may exert its major action on trophoblasts. A baseline
expression of LIF and LIF-Rb is probably needed for early pregnancy, but AP cannot be accounted for by the
defective expression of either LIF or LIF-Rb in most circumstances.
Key words: anembryonic pregnancy/blighted ovum/chorionic villi/decidua/leukemia inhibitory factor
Introduction
Anembryonic pregnancy (AP; blighted ovum in a clinical term)
is a frequent presentation of ®rst-trimester abortion and it
accounts for ~15% of all spontaneous pregnancies. Excluding
those with chromosome abnormality, there remains a high
percentage of abortion occurring without discernible mechanisms. Recently defective expression of some growth factors
and/or cytokines is suggested to be attributable to these
reproductive failures. These include colony-stimulating factor
(CSF), leukemia inhibitory factor (LIF) and interleukin-11 (IL11) (Bhatt et al., 1991; Pollard et al., 1991; Stewart et al., 1992;
Bilinski et al., 1998; Robb et al., 1998). LIF and IL-11 are
related cytokines but exert different reproductive functions.
They however share a common signal transduction component,
gp130, in the formation of high-af®nity receptor complexes
(Yin et al., 1993). Recently, we reported that there is a
defective production of IL-11 by decidua and chorionic villi in
human anembryonic pregnancy (Chen et al., 2002a), indicating
a signi®cant role of IL-11 in maintaining early pregnancy after
embryo implantation.
It was found that expression of LIF in the endometrium is
absolutely essential for mouse embryo to implant (Bhatt et al.,
1991; Stewart et al., 1992). Subsequent studies identi®ed the
peak expression of LIF and its receptor (LIF-R) in the midluteal phase of human endometrium. Therefore it is suggested
that LIF plays a role in regulating human embryo development
at the peri-implantation stage (Cullinan et al., 1996). In
addition, LIF probably also exerts some function before
embryo implantation, since LIF and LIF-R transcripts were
detectable in early mouse and human embryos (Chen et al.,
1999). However the role of LIF in the fetal±maternal interface
after embryos have implanted has not been clari®ed. The
expression of LIF and LIF-R genes has been detected in the
decidua and chorionic villi of ®rst-trimester and term placentas
(Kojima et al., 1994; Kojima et al., 1995; Sawai et al., 1995a,b,
1997; Sharkey et al., 1999), and LIF regulates the growth and
differentiation of trophoblasts (Kojima et al., 1995; Ren et al.,
1997). These reports suggest that LIF may constitute one
mechanism for local control of trophoblast and endometrial
proliferation and therefore is likely associated with the
Human Reproduction vol. 19 no. 7 ã European Society of Human Reproduction and Embryology 2004; all rights reserved
1647
Hsin-Fu Chen et al.
continuation of pregnancy. However, an animal study has
basically excluded the role of LIF in the maintenance of
pregnancy in mice (Chen et al., 2000). These discrepant results
apparently needs further clari®cation.
In addition, decidual NK cells and T cells produce a variety
of cytokines including LIF (Jokhi et al., 1994), and the
production of LIF and type 2 T-helper cytokines are both
defective by decidual T cell clones in unexplained recurrent
abortions (Piccinni et al., 1998). However, the clinical
signi®cance of LIF involvement in the evolution and formation
of anembryonic pregnancy (and thus abortion) has not been
established. Therefore, the present study was designed to
examine the expression of LIF and its receptor in early human
pregnancy, in order that the role of LIF system in the formation
of anembryonic pregnancy might be clari®ed. To test this, we
examined and compared the LIF and LIF-R gene expression
and protein products in normal pregnancy (NP) and AP.
Materials and methods
Subjects and materials
In total, 25 patients with sonographically con®rmed AP (at 7±10
weeks of gestation) were recruited into the study group (AP group).
Twenty-®ve age-matched patients with NP (viable pregnancies at 7±
10 weeks of gestation, which were arti®cially terminated due to multiparity) were recruited as control group (NP group). The gestational
age (mean 6 SEM, 8.5 6 0.4 vs 8.0 6 0.5 weeks for AP and NP,
respectively) and maternal age (34.5 6 2.1 vs 32.4 6 1.9 years) were
comparable between study and control groups. For the con®rmation of
pregnancy status, the presence or absence of fetal heart motion was
documented by either trans-abdominal or trans-vaginal ultrasound.
Samples of decidua and chorionic villi were obtained during dilatation
and curettage. Informed consent was obtained from each patient who
participated in this study. This study was approved by the Ethical
Committee of the National Taiwan University Hospital in 1998.
Tissue processing
The gestational tissues (decidua and chorionic villi) were collected
aseptically, put into test tubes containing normal saline, and
immediately taken to the laboratory for processing. The preparation
of decidual and villus tissues was performed according to the previous
report (Chen et al., 2002a). Brie¯y, tissue mixtures were grossly
separated into decidua and chorionic villi, washed twice with Hank's
balanced salt solution (HBSS), and divided into portions for further
RNA extraction or formalin ®xation. Small pieces of each sample
from each patient were also stained with hematoxylin and eosin
(H&E) and examined by a pathologist to exclude any pathology.
Semi-quantitation of LIF and LIF-Rb mRNA in decidua and
chorionic villi by standard RT±PCR
The RNA from decidua or chorionic villi was obtained using the
protocol reported previously (Chadderton et al., 1997; Chen et al.,
2002a). Brie¯y, tissue samples were cut into pieces, washed with PBS
(diethyl pyrocarbonate-treated), and then homogenized in liquid
nitrogen. Subsequently, RNA was extracted using the Trizol reagent
(Life Technologies, Inc., Grand Island, NY), chloroform±
isopropanol±ethanol method (Chadderton et al., 1997; Chen et al.,
2002a). The purity and concentration of RNA were analyzed ®rst by
UV spectrophotometer at absorbance wavelengths of 260 nm and 280
nm (OD260 and OD280). An OD260/OD280 ratio of higher than 1.6 was
considered adequate. Furthermore the integrity of the RNA sample
1648
was veri®ed both by agarose gel electrophoresis and examination of
the RT±PCR product of S26, a housekeeping gene (Vincent et al.,
1993). Five micrograms of RNA samples were subsequently subjected
to ®rst strand cDNA synthesis using oligo(dT) primers and Superscript
II reverse transcriptase following the manufacturer's manual (First
Strand cDNA Synthesis Kit, Pharmacia Biotech). The resultant cDNA
template was further subjected to PCR ampli®cation with oligonucleotide primers designed according to the published human LIF and LIFRb cDNA sequences (Stahl et al., 1990; Gearing et al., 1991). For LIF
mRNA ampli®cation, the following primers were used: sense, 5¢CAGCATCACTGAATCACAGAGC-3¢; and antisense, 5¢-AGTATGAAACATCCCCACAGGG-3¢. The size of the PCR product was 560
base pairs (bp). For LIF-Rb mRNA ampli®cation, the following
primers were used: sense, 5¢-CAAAAGAGTGTGTCTGTGAG-3¢;
and antisense, 5¢-CCATGTATTTACATTGGC-3¢. The predicted size
of the product was 459 bp. For the purpose of semi-quantitation, the
S26 gene transcript was also co-ampli®ed in the same reaction to
control for the mRNA amount in the same sample. The primer set used
was designed according to the published cDNA sequence (Vincent
et al., 1993), and were: sense, 5¢-CCGTGCCTCCAAGATGACAAAG-3¢; and antisense, 5¢-ACTCAGCTCCTTACATGGGCTT-3¢. The predicted size of the PCR product was 371 bp. All
the above primer sets were designed to span at least two exons, in
order to exclude the possibility of amplifying contaminating genomic
DNA.
In a 50 ml reaction mixture, PCR was performed in the presence of
25 ml Taq Master Mix (QIAGEN, Valencia, USA), 3 ml ®rst strand
cDNA template, 200 ng/ml 5¢-primer, 200 ng/ml 3¢-primer, and 20 ml
d.d.H2O. A number of preliminary experiments were performed to
determine the PCR conditions and the cycle numbers that ensured the
PCR production in an exponential phase. The PCR was carried out in a
thermal cycler (GeneAmp PCR System 9700; PERKIN-ELMER,
Norwalk, CT) in the following sequences: denaturation at 94°C for
30±60 s, annealing at 55±65°C for 60±90 s, extension at 72°C for 30 s,
and a ®nal extension at 72°C for 15±30 min. The PCR was repeated for
28±38 cycles according to the gene of interest and the product was
examined using 1.5% agarose gel electrophoresis and stained with
SYBR green stain (SYBR Green I; Roche Molecular Biochemicals,
Indianapolis, IN). The signal intensity of each product was measured
by scanning the SYBR green luminescence using a phosphoimage
scanner (Storm 840; Molecular Dynamics, Sunnyvale, CA). A reverse
transcription reaction was carried out without the addition of reverse
transcriptase, and the resulting product was subjected to PCR to
exclude the possibility of genomic DNA contamination. PCR was also
performed without the presence of template DNA to test for crosscontamination of samples. All the samples were examined in duplicate
by two researchers, and the data were compared. It was found that the
data obtained by the two researchers were in most instances
comparable.
In each sample, the relative mRNA levels of LIF and LIF-Rb were
corrected by the mRNA amount, as re¯ected by the S26 level, using
the following formulae: LIF/S26 and LIF-Rb/S26, to represent LIF
mRNA and LIF-Rb mRNA levels, respectively.
Real-time quantitative PCR
Due to the inherently limited sensitivity of the semi-quantiation used
above, we further conducted a quantitative PCR to measure these gene
expressions. A real-time quantitative PCR system (ABI PRISM 7700
Sequence Detection System; PE Applied Biosystems) was used as
previously described (Gibson et al., 1996; Chen et al., 2002a). Brie¯y,
RNA was prepared and subsequently cDNA was obtained by random
hexamer priming. The primers and probes used in the PCR were
designed according to the TaqMan primer and probe design system
LIF in decidua and villi of anembryonic pregnancy
Table I. Sequences of TaqMan primers and probes
Gene
Sequence
Position
LIF
Forward primer: 5¢-TGGTTCTGCACTGGAAACATG-3¢
Reverse primer: 5¢-TGTAATAGAGAATAAAGAGGGCATTGG-3¢
Probe: 5¢-FAMa±AACCTCATGAACCAGATCAGGAGCCAACT±TAMRAa-3¢
Forward primer: 5¢-GGAGCCTCTGCGACTCATTC-3¢
Reverse primer: 5¢-GATGGTCGTTTCAAACATACGTAAA-3¢
Probe: 5¢-FAMa±CCTCCAGGACTGACTGCATTGCACA±TAMRAa-3¢
82±100
220±246
171±199
103±122
164±188
128±152
LIF-Rb
aFAM
and VIC are the reporter dyes and TAMRA is the quencher dye as detailed in the Materials and methods.
(PE Applied Biosystems) (Table I). Based on the manufacturer's
protocol, the FAM (6-carboxy¯uorescein) and VIC were used as the
reporter dyes and TAMRA (6-carboxy-tetramethyl-rhodamine) as the
quencher dye. The probes were labeled with both reporter dye and
quencher dye on the 5¢- and 3¢-ends, respectively. During PCR, the
reporter dye was released and the resultant ¯uorescence was detected
and could be quanti®ed. Before examinations on the study samples, a
number of preliminary experiments were done to determine the PCR
conditions. The PCR was carried out in a thermal cycler (ABI PRISM
7700 Sequence Detection System) in the following sequence: reaction
at 50°C for 2 min, at 95°C for 10 min, and subsequently the PCR was
repeated for 45 cycles at denaturation at 95°C for 15 s and annealing
and extension at 60°C for 1 min. The authenticity of PCR products was
veri®ed by 2% agarose gel electrophoresis and by direct sequencing.
Each sample was examined in triplicate and a mean value was
obtained. The relative concentration of each mRNA was subsequently
calculated according to the Manufacturer's User Manual. Brie¯y, the
threshold cycle (CT) values of the target gene (LIF or LIF-Rb) and the
internal control gene (b-actin) mRNA in the studied sample were ®rst
measured. The DCT value of the studied sample was calculated by the
following formula: DCT = CT of target gene ± CT of b-actin (this DCT
is designated as `sample DCT'). Similarly, the CT values of the target
gene and its respective b-actin of a positive control were also obtained
and the DCT was calculated (designated as `calibrator DCT'). DDCT
was then calculated using the following formula: DDCT = DCT
(sample) ± DCT (calibrator) and ®nally the relative value of each
mRNA was calculated by the formula: 2±DDCT. PCR without template
was used as a negative control (called no template control) to verify
experimental results.
Immunohistochemistry
The immunohistochemistry protocol was modi®ed from previous
reports (Hsu et al., 1981; Lessey et al., 1994a,b; Chen et al., 2002a)
and
a
Streptavidin/Biotin
Universal
Detection
System
(IMMUNOTECH, Marseille, France) was used. Brie¯y, formalin®xed, paraf®n-embedded tissue sections were deparaf®nized with
xylene, and hydrated with step-down concentrations of ethanol. The
sections were then incubated with 30% H2O2/70% methanol solution
at RT for 5 min, treated with protein blocking agent at RT for 5 min to
reduce non-speci®c binding of antibodies, and incubated overnight at
4°C with primary antibodies (for LIF detection, goat anti-human LIF
antibody, 1:25 dilution; R&D Systems Inc., Minneapolis, MN, USA;
for anti-LIF-R detection, goat anti-human LIFR antibody, 1:40±80
dilution; R&D Systems). On the next day, the sections were washed
with PBS, treated with polyvalent biotinylated secondary antibody,
which bound to primary antibody at RT for 10 min, and then treated
with streptavidin±peroxidase reagent, which bound to the secondary
antibody, at RT for 10 min. The colour was developed using DAB
chromogen at RT for 15 min and counterstained with hematoxylin for
1 min. An independent pathologist who was blind to the origins of the
Figure 1. Expression of LIF (560 bp), LIF-Rb (459 bp) and S26
(371 bp) mRNA in the decidua and chorionic villi of normal and
anembryonic pregnancy by standard RT±PCR using the primer sets
described in Materials and methods. The PCR product was resolved
in 1.5% agarose gel electrophoresis and stained with SYBR green
stain. S26 was used as an internal control. M, DNA marker; NP,
normal pregnancy; AP, anembryonic pregnancy.
samples reviewed these stained slides and graded their staining
intensity. Experiments without primary or secondary antibodies were
used as negative controls.
Statistical analysis
Since a nonparametric testing was a more appropriate statistical
method due to the smaller case number, the Wilcoxon Rank-Sum Test
(the Mann±Whitney U test) was used to analyze the data. A P-value of
<0.05 was considered statistically signi®cant.
Results
The expression of LIF and LIF-Rb genes in decidua and
chorionic villi
PCR revealed that all the samples expressed LIF and LIF-Rb
mRNA. These included decidua and chorionic villi either from
NP or AP (Figure 1). In addition, S26, a housekeeping gene,
was detectable in all samples (Figure 1). Direct DNA
sequencing for the PCR products was done and it was found
1649
Hsin-Fu Chen et al.
Figure 2. Representative photographs of semi-quantitation of LIF
(560 bp) and LIF-Rb (459 bp) mRNA levels in the decidua and
chorionic villi. S26 (371 bp) was co-ampli®ed as an internal control.
Samples were subjected to RT±PCR using the primer sets described
in the Materials and methods. Different cycle numbers were done
(e.g. 26 and 28 cycles in these images) for the purpose of semiquantitation. The relative levels of each mRNA of interest (LIF and
LIF-Rb) were quanti®ed by the formulae: LIF/S26 and LIF-Rb/S26,
respectively. (A) LIF and S26 co-ampli®cation; (B) LIF-Rb and S26
co-ampli®cation; M, DNA marker.
that their sequences were identical to the published human LIF
and LIF-Rb cDNA sequences, respectively (data not shown).
Semi-quantitation of LIF and LIF-Rb mRNA
Representative photographs of the semi-quantitation of the LIF
and LIF-Rb mRNA are shown (Figure 2). Due to the lack of
suf®cient quantities of samples for study in some subjects,
there were in total 24 samples in the NP group and 23 samples
in the AP group available for analysis. For LIF mRNA, the
relative transcript levels were 2.3 6 0.5 (mean 6 SEM) and 2.5
6 0.3, respectively, in the decidua and chorionic villi of NP
(Figure 3A). The levels were 3.1 6 0.9 and 3.5 6 1.0,
respectively, in AP (Figure 3A). For LIF-Rb mRNA, the
relative transcript levels were 0.7 6 0.1 and 3.8 6 0.8,
respectively, in the decidua and chorionic villi of NP
(Figure 3B). The levels were 1.2 6 0.2 and 3.1 6 0.6,
respectively, in AP (Figure 3B). These data indicate that the
LIF-Rb mRNA level was higher in chorionic villi compared to
that in decidua, in either NP (0.7 6 0.1 vs 3.8 6 0.8,
respectively, for decidua and villi, P = 0.002) or AP (1.2 6 0.2
vs 3.1 6 0.6, P = 0.008). However, the levels of LIF mRNA
were not different within the same group (NP, 2.3 6 0.5 vs 2.5
6 0.3, P > 0.05 respectively in decidua and; AP, 3.1 6 0.9 vs
3.5 6 1.0, in villi P > 0.05) (Figure 3A). The comparison
between different groups (NP vs AP) neither showed any
difference of LIF mRNA levels, when either decidua (2.3 6 0.5
vs 3.1 6 0.9, P > 0.05) or chorionic villi (2.5 6 0.3 vs 3.5 6
1.0, P > 0.05) were concerned (Figure 3A). Similarly, the LIFRb mRNA levels were not different between groups (NP vs
AP) in either decidua (0.7 6 0.1 vs 1.2 6 0.2, P = 0.18) or
chorionic villi (3.8 6 0.8 vs 3.1 6 0.6, P = 0.50) (Figure 3B).
1650
Figure 3. Comparison of LIF and LIF-Rb mRNA levels in the
decidua and chorionic villi by semi-quantitative RT±PCR, using S26
as an internal control. Longitudinal axis indicates the relative levels
of LIF (A) and LIF-Rb (B) mRNA, which were calculated by the
formulae: LIF/S26 and LIF-Rb/S26, respectively. NP, normal
pregnancy; AP, anembryonic pregnancy. *P = 0.002 (comparison
between decidua and chorionic villi of NP for LIF-Rb). **P = 0.008
(comparison between decidua and chorionic villi of AP for LIF-Rb).
The bars represent the mean 6 SEM.
Quantitation of LIF and LIF-Rb mRNA by real-time
quantitative PCR
By real-time quantitative PCR, it was found that the relative
levels of LIF mRNA were 162.1 6 59.5 (mean 6 SEM) and
226.4 6 105.5, respectively, in the decidua and chorionic villi
of NP (n = 21) (Figure 4A). The levels were 926.3 6 487.0 and
401.8 6 210.7, respectively, in AP (n = 22) (Figure 4A). For
LIF-Rb, the relative transcript levels were 15.1 6 14.0 and
7333.2 6 3198.0, respectively, in the decidua and chorionic
villi of NP (Figure 4B). The levels were 0.3 6 0.2 and 1053.4
6 535.3, respectively, in AP (Figure 4B).
These data indicate again that the levels of LIF mRNA were
not different within the same group (NP, 162.1 6 59.5 vs 226.4
6 105.5, P = 0.60, respectively in decidua; AP, 926.3 6 487.0
vs 401.8 6 210.7, P = 0.35 in villi). The comparison between
different groups (NP vs AP) neither showed any difference of
LIF mRNA levels, when either decidua (162.1 6 59.5 vs 926.3
6 487.0, P = 0.13) or chorionic villi (226.4 6 105.5 vs 401.8 6
LIF in decidua and villi of anembryonic pregnancy
detectable in both the stroma cells and the luminal and
glandular epithelium but the vascular endothelium did not
show signi®cant immunoreactivity (Figure 5A and C). In
chorionic villi, the LIF protein was mostly localized to the
trophoblasts but the stroma cells and endothelium also showed
faint staining (Figure 5B and D). It was not microscopically
possible to identify any difference in LIF staining intensity
between NP and AP, in either decidua or chorionic villi
(Figure 5A±D) (Table II).
LIF-Rb protein was also detectable in the decidua and
chorionic villi of both NP and AP (Figure 5E±H). However, in
striking contrast to that of LIF protein, the LIF-Rb staining was
much less intense in decidua compared to that in chorionic villi
(Figure 5E±H). In the chorionic villi, the intense staining was
mostly localized to the trophoblasts. However, when tissues
from NP and AP were compared, the staining intensity was
comparable in most instances, when either decidua (Figure 5E
and G) or chorionic villi (Figure 5F and H) was concerned
(Table II). As expected, the negative controls (Figure 5I and J)
did not show any immunoreactive staining.
Discussion
Figure 4. Comparison of LIF and LIF-Rb mRNA levels in the
decidua and chorionic villi by real-time quantitative PCR, using
b±actin as an internal control. Longitudinal axis indicates the
relative values of LIF (A) and LIF-Rb (B) mRNA, which were
calculated by the formula: 2±DDCT, as described in the Materials and
methods. NP, normal pregnancy; AP, anembryonic pregnancy. *P =
0.006 (comparison between decidua and chorionic villi of NP for
LIF-Rb). **P = 0.03 (comparison between decidua and chorionic
villi of AP for LIF-Rb). The bars represent the mean 6 SEM.
210.7, P = 0.44) were concerned (Figure 4A). Similar to the
trend found by semi-quantitation in the previous section, the
LIF-Rb mRNA level was statistically lower in decidua than
that in chorionic villi, in either NP (15.1 6 14.0 vs 7333.2 6
3198.0, P = 0.006) or AP (0.3 6 0.2 vs 1053.4 6 535.3, P =
0.03) (Figure 4B). However, comparison between different
groups of patients did not show difference in LIF-Rb mRNA
levels in decidua (15.1 6 14.0 vs 0.3 6 0.2, respectively in NP
and AP, P = 0.20) (Figure 4B). Though there was a trend
towards lower level of LIF-Rb mRNA in chorionic villi of AP
(Figure 4B), the difference did not reach statistical signi®cance
(7333.2 6 3198.0 vs 1053.4 6 535.3, P = 0.07) (Figure 4B).
Immunohistochemical study for LIF and LIF-Rb proteins
in decidua and chorionic villi
Immunohistochemical study identi®ed that the immunoreactive LIF was found in both decidua and chorionic villi of both
NP and AP (Figure 5A±D). In decidua, the LIF staining was
LIF is transiently expressed in the glandular epithelium of mice
at ovulation and again on the fourth day of pregnancy, when the
embryo is prepared to implant (Bhatt et al., 1991; Stewart et al.,
1992). LIF-de®cient female mice (LIF±/±) are fertile, but their
blastocysts cannot implant unless they are treated with LIF
(Stewart et al., 1992). In the human, LIF expression in the
endometrium is up-regulated at the peri-implantation stage,
suggesting that LIF may also be important for the human
embryo to implant (Charnock-Jones et al., 1994; Cullinan et al.,
1996). However, it has not yet been established whether
continuous expression of LIF during the entire pregnancy is
required for embryonic development. It is neither clear whether
AP has a causal relationship with the defective production of
LIF-related genes. The present study identi®ed the expression
of LIF and LIF-Rb genes in human decidua and chorionic villi
of early gestation. However, LIF mRNA levels were not
different either in different tissues (decidua vs chorionic villi)
or between different groups of patients (NP vs AP). In contrast,
a differential spatial distribution of LIF-Rb mRNA level was
found, showing a signi®cantly lower level in decidua than in
chorionic villi. One of our previous studies (Wu et al., 2001)
also demonstrated that serum levels of LIF neither changed
before and after embryo implantation nor were related to the
occurrence of abortion in an in vitro fertilization (IVF)
program. All these data suggest a role but do not de®ne the
function of LIF system genes in post-implantation stage. While
the major target of LIF is likely to be on the trophoblasts, the
absence of differences in LIF or LIF-Rb levels between NP and
AP suggests that in human, AP cannot be accounted for by the
defective expression of either LIF or LIF-Rb in the maternal±
fetal interface in most circumstances.
Discrepant results have been reported in a mouse model,
which suggested that after implantation, LIF is neither required
by the mouse embryo for further development nor for the
maintenance of pregnancy (Chen et al., 2000). That speci®c
1651
Hsin-Fu Chen et al.
Figure 5. Representative photomicrographs of immunohistochemical study of LIF and LIF-Rb in the decidua and chorionic villi from normal
pregnancy (NP) and anembryonic pregnancy (AP). Formalin-®xed, paraf®n-embedded tissue samples were treated with either anti-LIF
antibody or anti-LIF-R antibody, respectively, using the Streptavidin±Biotin Universal Detection System (Immunotech) as described in the
Materials and methods. The color was developed using DAB chromogen and counterstained by hematoxylin. (A±D) LIF examination. (E±H)
LIF-Rb examination. (I±J) Representative negative controls. A, C, E, G and I were decidua tissues. B, D, F, H and J were chorionic villi
tissues. A, B, E and F were from NP. C, D, G and H were from AP. Brown color indicates positive staining (arrow). Magni®cation, 3200.
GL, gland; ST, stroma; TR, trophoblast; EN, vascular endothelium. Bar = 50 mm.
study (Chen et al., 2000) basically excludes the active role of
LIF in the continuation of pregnancy. However, in view of the
signi®cant expression and possible function of LIF-related
genes in the placenta and decidua during pregnancy reported in
this study and others (Jokhi et al., 1994; Kojima et al., 1994;
Arici et al., 1995; King et al., 1995; Sawai et al., 1997; Sharkey
et al., 1999; Modric et al., 2000), it remains premature to
conclude that LIF action is totally unnecessary after implantation in other species. For example, greatest steady-state LIF
mRNA levels were evident in the porcine placenta at postimplantation stages, suggesting its important function (Modric
et al., 2000). In human, LIF-R mRNA was localized in the
1652
villous and extravillous trophoblast throughout pregnancy and
strong expression of LIF mRNA was detected in decidual
leukocytes, especially at the implantation sites, suggesting that
LIF may mediate interactions between maternal decidual
leukocytes and invading trophoblast (Sharkey et al., 1999).
Piccinni and coauthors showed that progesterone-mediated
production of both IL-4 and LIF might contribute to the
development and maintenance of successful human pregnancy
(Piccinni et al., 1998). Therefore, in the consideration of the
discrepant results from different studies, species-related factor
should be one to be explored. Actually there are good
evidences that LIF action differs in different species. For
LIF in decidua and villi of anembryonic pregnancy
Table II. Results of immunohistochemical study (LIF and LIF-Rb proteins)
Tissue/Gene
Normal pregnancy (n)
Anembryonic pregnancy (n)
Villi (8)
LIF
LIF-Rb
Trophoblast
+++
+++
Decidua (8)
Stroma
±/+
±/+
Gland
+++
+
example, LIF alone can support the self-renewal without
differentiation of a mouse embryonic stem cell line (which is
derived from pre-implantation embryo) in a feeder-free culture
condition. In contrast, feeder cells are generally indispensable
for a human embryonic stem cell line to proliferate without
differentiation (Reubinoff et al., 2000). As yet we have not
encountered a case with extreme down-regulation or alteration
of either LIF or LIF-R. Therefore we cannot predict what will
happen if baseline levels of LIF and LIF-R are not expressed in
early human gestation. However, according to the present
study and those previous reports (Jokhi et al., 1994; Kojima
et al., 1994; Arici et al., 1995; Sawai et al., 1997; Piccinni et al.,
1998; Sharkey et al., 1999; Chen et al., 2002b), it is probably
reasonable to suggest that a baseline expression of LIF and
LIF-Rb is important for the continuation of human pregnancy.
Further study will be necessary to identify the implication of
the difference of LIF action on human and mouse.
A number of known or unknown factors including certain
cytokines may prove to be the mechanism leading to human AP
or abortion. However, we did not ®nd a signi®cant difference in
LIF or LIF-Rb expression between NP and AP in this study. It
is reasoned that though LIF action may play an active role in
the maintenance of human pregnancy, the occurrence of AP is
not usually associated with a defective production of LIF or
LIF-Rb. As far as our knowledge is concerned, this is one of the
rare human reports regarding the investigation of LIF role in
the evolution of AP or abortion. However, the results from this
study are not completely agreeable with those of a previous
human study which showed a decreased production of LIF by
decidual T cell clones from women with unexplained recurrent
abortion (Piccinni et al., 1998). Several mechanisms probably
can explain these contradictory ®ndings. The ®rst is the
difference in patient populations (®rst trimester abortion in this
study vs unexplained recurrent abortion in the Piccinni study).
We cannot exclude the possibility that heterogeneity of the
patient population in ®rst trimester abortion will compromise
the analysis. However it is not always practical to identify a
pure population of unexplained recurrent aborters, especially
when the subject number needs to be beyond a threshold level
for analysis. Secondly, the study by Piccinni et al. examined
the in vitro function using clones of T cells. Their approach was
different from that of the present study, which directly used an
in vivo sample (decidua and chorionic villi) for examination. In
addition, it is quite likely that decidual T cell may not be the
only cell type that produces LIF in the human endometrium or
placenta (Croy et al., 1991; Jokhi et al., 1994; Laird et al.,
1997). For example, in situ hybridization and immunohistochemical studies identify most LIF production in human
Villi (8)
Stroma
++/+++
+
Trophoblast
+++
++/+++
Decidua (8)
Stroma
±/+
±/+
Gland
+++
+
Stroma
++
+
endometrium to the glandular epithelium (Laird et al., 1997).
In this study, we also showed the production of LIF by both the
stroma and epithelial cells in decidua and by trophoblast cells
in chorionic villi, by immunohistochemical study. The possibility therefore cannot be excluded that the decreased LIF
production by decidual T cell in recurrent aborters (Piccinni
et al., 1998) may in some circumstances be overcome by
exaggerated or supplemental production from other cell types.
Since en bloc tissues (decidua and chorionic villi) were
examined in this study, it is conceivable that if the rescue
mechanism (i.e. supplemental production from other cells)
actually exists, real-time quantitative PCR will not detect the
difference in LIF gene expression between NP and AP. In a
logical consideration based on clinical conditions, one has
good reason to speculate that this type of rescue mechanism
would be present if LIF-producing cell types other than T cells
remained functionally intact. Therefore further study to
identify the presentation in a more puri®ed patient population
(for example, unexplained recurrent abortion) and a detailed
comparison between localized and general environment would
be needed to clarify this issue.
In addition, it is known that both in mice and human
(Cullinan et al., 1996), there are two isoforms of LIF, diffusible
and matrix-associated, which may exert different biological
effects. The present study did not speci®cally differentiate
them and therefore we cannot exclude the possibility that
expression and production of respective isoforms might be
different in the study subjects. Therefore future detailed
analyses of the LIF isoforms will probably provide further
information about this issue. In addition, gp130, an important
part of the LIF receptor system, should also be studied.
In this study, we used two experimental methods to quantify
the concentration of target mRNA, one by conventional semiquantitation and the other by real-time quantitative PCR. Some
scientists argue that there are limitations about the sensitivity
and speci®city of semi-quantitation method. The present study
did show some essential differences between these two
methods, but the semi-quantitation method still identi®es a
trend in mRNA level which was re¯ected by real-time
quantitative PCR. Therefore, the validity of the data obtained
in this study was strengthened by the concomitant use of the
two methods. However, due to the potentially higher sensitivity
of real-time quantitative PCR, this method would probably
provide a clearer insight into the signi®cance of the data in
some circumstances.
In conclusion, the present study suggests that LIF-related
genes may play a functional role in the maintenance of human
pregnancy, at least in the ®rst trimester. At this stage, the LIF
1653
Hsin-Fu Chen et al.
action may be predominantly on the trophoblasts through
binding to LIF-R. However, there is no absolute indication that
defective production of LIF or LIF-R is a major cause of
abortion in early human pregnancy.
Acknowledgements
This work was supported by grants from the National Science Council
of the Republic of China (NSC90-2314-B-002-147 and NSC91-2314B-002-374) and National Taiwan University Hospital (91A10-6).
References
Arici A, Engin O, Attar E and Olive DL (1995) Modulation of leukemia
inhibitory factor gene expression and protein biosynthesis in human
endometrium. J Clin Endocrinol Metab 80,1908±1915.
Bhatt H, Brunet LJ and Stewart CL (1991) Uterine expression of leukemia
inhibitory factor coincides with the onset of blastocyst implantation. Proc
Natl Acad Sci USA 88,11408±11412.
Bilinski P, Roopenian D and Gossler A (1998) Maternal IL-11Ralpha function
is required for normal decidua and fetoplacental development in mice. Gene
Dev 12,2234±2243.
Chadderton T, Wilson C, Bewick M and Gluck S (1997) Evaluation of three
rapid RNA extraction reagents: relevance for use in RT-PCRs and
measurement of low level gene expression in clinical samples. Cell Mol
Biol 43,1227±1234.
Charnock-Jones DS, Sharkey AM, Fenwick P and Smith SK (1994)
Leukaemia inhibitory factor mRNA concentration peaks in human
endometrium at the time of implantation and the blastocyst contains
mRNA for the receptor at this time. J Reprod Fertil 101,421±426.
Chen HF, Shew JY, Ho HN, Hsu WL and Yang YS (1999) Expression of
leukemia inhibitory factor and its receptor in preimplantation embryos.
Fertil Steril 72,713±719.
Chen HF, Lin CY, Chao KH, Wu MY, Yang YS and Ho HN (2002a)
Defective production of interleukin-11 by decidua and chorionic villi in
human anembryonic pregnancy. J Clin Endocrinol Metab 87,2320±2328.
Chen HW, Chen JJ, Tzeng CR, Li HN, Chang SJ, Cheng YF, Chang CW,
Wang RS, Yang PC and Lee YT (2002b) Global analysis of differentially
expressed genes in early gestational decidua and chorionic villi using a 9600
human cDNA microarray. Mol Hum Reprod 8,475±484.
Chen JR, Cheng JG, Shatzer T, Sewell L, Hernandez L and Stewart CL (2000)
Leukemia inhibitory factor can substitute for nidatory estrogen and is
essential to inducing a receptive uterus for implantation but is not essential
for subsequent embryogenesis. Endocrinology 141,4365±4372.
Croy BA, Guilbert LJ, Browne MA, Gough NM, Stinchcomb DT, Reed N and
Wegmann TG (1991) Characterization of cytokine production by the
metrial gland and granulated metrial gland cells. J Reprod Immunol 19,149±
166.
Cullinan EB, Abbondanzo SJ, Anderson PS, Pollard JW, Lessey BA, Stewart
and CL. (1996) Leukemia inhibitory factor (LIF) and LIF receptor
expression in human endometrium suggests a potential autocrine/
paracrine function in regulating embryo implantation. Proc Natl Acad Sci
USA 93,3115±3120.
Gearing DP, Thut CJ, VandeBos T, Gimpel SD, Delaney PB, King J, Price V,
Cosman D and Beckmann MP (1991) Leukemia inhibitory factor receptor is
structurally related to the IL-6 signal transducer, gp130. EMBO J 10,2839±
2848.
Gibson UE, Heid CA and Williams PM (1996) A novel method for real time
quantitative RT±PCR. Genome Res 6,995±1001.
Hsu SM, Raine L and Fanger H (1981) A comparative study of the peroxidaseantiperoxidase method and an avidin-biotin complex method for studying
polypeptide hormones with radioimmunoassay antibodies. Am J Clin Pathol
75,734±738.
Jokhi PP, King A, Sharkey AM, Smith SK and Loke YW (1994) Screening for
cytokine messenger ribonucleic acids in puri®ed human decidual
lymphocyte populations by the reverse-transcriptase polymerase chain
reaction. J Immunol 153,4427±4435.
King A, Jokhi PP, Smith SK, Sharkey AM and Loke YW (1995) Screening for
cytokine mRNA in human villous and extravillous trophoblasts using the
reverse-transcriptase polymerase chain reaction (RT±PCR). Cytokine
7,364±371.
1654
Kojima K, Kanzaki H, Iwai M, Hatayama H, Fujimoto M, Inoue T, Horie K,
Nakayama H, Fujita J and Mori T (1994) Expression of leukemia inhibitory
factor in human endometrium and placenta. Biol Reprod 50,882±887.
Kojima K, Kanzaki H, Iwai M, Hatayama H, Fujimoto M, Narukawa S,
Higuchi T, Kaneko Y, Mori T and Fujita J (1995) Expression of leukaemia
inhibitory factor (LIF) receptor in human placenta: a possible role for LIF in
the growth and differentiation of trophoblasts. Hum Reprod 10,1907±1911.
Laird SM, Tuckerman EM, Dalton CF, Dunphy BC, Li TC and Zhang X
(1997) The production of leukaemia inhibitory factor by human
endometrium: presence in uterine ¯ushings and production by cells in
culture. Hum Reprod 12,569±574.
Lessey BA, Castelbaum AJ, Buck CA, Lei Y, Yowell CW and Sun J (1994a)
Further characterization of endometrial integrins during the menstrual cycle
and in pregnancy. [See comments]. Fertil Steril 62,497±506.
Lessey BA, Castelbaum AJ, Buck CA, Lei Y, Yowell CW and Sun J (1994b)
Further characterization of endometrial integrins during the menstrual cycle
and in pregnancy. [See comments]. Fertil Steril 62,497±506.
Modric T, Kowalski AA, Green ML, Simmen RC and Simmen FA (2000)
Pregnancy-dependent expression of leukaemia inhibitory factor (LIF), LIF
receptor-beta and interleukin-6 (IL-6) messenger ribonucleic acids in the
porcine female reproductive tract. Placenta 21,345±353.
Piccinni MP, Beloni L, Livi C, Maggi E, Scarselli G and Romagnani S (1998)
Defective production of both leukemia inhibitory factor and type 2 T-helper
cytokines by decidual T cells in unexplained recurrent abortions. Nat Med
4,1020±1024.
Pollard JW, Hunt JS, Wiktor-Jedrzejczak W and Stanley ER (1991) A
pregnancy defect in the osteopetrotic (op/op) mouse demonstrates the
requirement for CSF-1 in female fertility. Dev Biol 148,273±283.
Ren SG, Melmed S and Braunstein GD (1997) Decidual leukemia inhibitory
factor production and action on human chorionic gonadotropin secretion at
different stages of gestation in vitro. Early Pregnancy 3,102±108.
Reubinoff BE, Pera MF, Fong CY, Trounson A and Bongso A (2000)
Embryonic stem cell lines from human blastocysts: somatic differentiation
in vitro. [See comments]. [Erratum appears in Nat Biotechnol 2000;
18:559]. Nat Biotechnol 18,399±404.
Robb L, Li R, Hartley L, Nandurkar HH, Koentgen F and Begley CG (1998)
Infertility in female mice lacking the receptor for interleukin 11 is due to a
defective uterine response to implantation. Nat Med 4,303±308.
Sawai K, Matsuzaki N, Kameda T, Hashimoto K, Okada T, Shimoya K,
Nobunaga T, Taga T, Kishimoto T and Saji F (1995a) Leukemia inhibitory
factor produced at the fetomaternal interface stimulates chorionic
gonadotropin production: its possible implication during pregnancy,
including implantation period. J Clin Endocrinol Metab 80,1449±1456.
Sawai K, Azuma C, Koyama M, Ito S, Hashimoto K, Kimura T, Samejima Y,
Nobunaga T and Saji F (1995b) Leukemia inhibitory factor (LIF) enhances
trophoblast differentiation mediated by human chorionic gonadotropin
(hCG). Biochem Biophys Res Commun 211,137±143.
Sawai K, Matsuzaki N, Okada T, Shimoya K, Koyama M, Azuma C, Saji F
and Murata Y (1997) Human decidual cell biosynthesis of leukemia
inhibitory factor: regulation by decidual cytokines and steroid hormones.
Biol Reprod 56,1274±1280.
Sharkey AM, King A, Clark DE, Burrows TD, Jokhi PP, Charnock-Jones DS,
Loke YW and Smith SK (1999) Localization of leukemia inhibitory factor
and its receptor in human placenta throughout pregnancy. Biol Reprod
60,355±364.
Stahl J, Gearing DP, Willson TA, Brown MA, King JA and Gough NM (1990)
Structural organization of the genes for murine and human leukemia
inhibitory factor. Evolutionary conservation of coding and non-coding
regions. J Biol Chem 265,8833±8841.
Stewart CL, Kaspar P, Brunet LJ, Bhatt H, Gadi I, Kontgen F and Abbondanzo
SJ (1992) Blastocyst implantation depends on maternal expression of
leukaemia inhibitory factor. [See comments]. Nature 359,76±79.
Vincent S, Marty L and Fort P (1993) S26 ribosomal protein RNA: an
invariant control for gene regulation experiments in eucaryotic cells and
tissues. Nucleic Acids Res 21,1498.
Wu MY, Chen HF, Chen SU, Chao KH, Yang YS and Ho HN (2001) Increase
in the production of interleukin-10 early after implantation is related to the
success of pregnancy. Am J Reprod Immunol 46,386±392.
Yin T, Taga T, Tsang ML, Yasukawa K, Kishimoto T and Yang YC (1993)
Involvement of IL-6 signal transducer gp130 in IL-11-mediated signal
transduction. J Immunol 151,2555±2561.
Submitted on November 7, 2003; accepted on February 2, 2004