Khosrotehrani K, Johnson KL, Cha DH, Salomon RN, Bianchi DW. Transfer of Fetal Cells With Multilineage Potential to Maternal Tissue. JAMA. 2004;292(1):75-80. doi:10.1001/jama.292.1.75
Author Affiliations: Departments of Pediatrics and Obstetrics and Gynecology (Drs Khosrotehrani, Johnson, Cha, and Bianchi), and Department of Pathology (Dr Salomon), Tufts-New England Medical Center, Boston, Mass.
Context During pregnancy, fetal CD34+ cells enter the maternal circulation,
persist for decades, and create a state of physiologic microchimerism. Many
studies have confirmed the residual presence of fetal cells in maternal blood
and tissues following pregnancy. Fetal cells may respond to maternal injury
by developing multilineage capacity in maternal organs.
Objective To verify that fetal microchimeric cells express markers of epithelial,
leukocyte, and hepatocyte differentiation within maternal organs.
Design, Setting, and Patients Archived paraffin-embedded tissue section specimens from 10 women who
had male offspring and were previously found to have high numbers of microchimeric
cells, and 11 control women who had no prior male pregnancies. Male cells
were identified by fluorescence in situ hybridization, using X and Y chromosome–specific
probes, followed by histologic and immunochemical studies using anticytokeratin
(AE1/AE3) as a marker of epithelial cells, anti-CD45 as a leukocyte marker,
and heppar-1 as a hepatocyte marker.
Main Outcome Measure Percentage of microchimeric cells expressing nonhematopoietic markers.
Results A total of 701 male (XY+) microchimeric cells were identified (mean
[SD], 227  XY+ cells per million maternal cells). In maternal epithelial
tissues (thyroid, cervix, intestine, and gallbladder), 14% to 60% of XY+ cells
expressed cytokeratin. Conversely, in hematopoietic tissues, such as lymph
nodes and spleen, 90% of XY+ cells expressed CD45. In 1 liver sample, 4% of
XY+ cells expressed heppar-1. Histologic and immunochemical evidence of differentiation,
as assessed by independent observers, was highly concordant (κ = 0.72).
Conclusion The detection of microchimeric male cells, bearing epithelial, leukocyte,
or hepatocyte markers, in a variety of maternal tissue specimens suggests
the presence of fetal cells that may have multilineage capacity.
Fetal cells enter the maternal circulation during all pregnancies.1,2 They can persist in maternal blood
or tissues for decades, creating a state of physiologic microchimerism in
the parous woman.3 Recent studies detected
male cells of presumed fetal origin in 30% to 50% of healthy women who had
prior male pregnancies.4 The long-term consequences
of fetal cell microchimerism for maternal health are only beginning to be
appreciated. Fetal microchimeric cells are present in higher numbers in women
with some autoimmune diseases, such as systemic sclerosis, than in control
groups.5,6 We have also observed
fetal cells in the tissues of women with nonautoimmune disorders, such as
hepatitis C7 and cervical cancer.8 Thus,
we developed an alternate hypothesis in which fetal cells were associated
with the maternal response to injury as opposed to causing disease.
During pregnancy, the fetal cells that enter the maternal circulation
are predominantly of hematopoietic origin, such as nucleated red blood cells,
lymphocytes, or hematopoietic stem cells.9,10 Trophoblasts
and mesenchymal stem cells also circulate within maternal blood.11,12 Following
pregnancy, male fetal cells have been demonstrated in the CD34+ compartment.3,13,14 They have also been
found in various sorted subsets of maternal peripheral mononuclear blood cells,
such as T, B, and natural killer cells, or cells that express the CD4 or CD8
antigens,15,16 suggesting that
fetal microchimeric cells may be capable of engraftment and differentiation
along the hematopoietic pathway.
Little information is available on the phenotype of fetal microchimeric
cells in nonhematopoietic tissues and most published studies suggest that
fetal cells express hematopoietic markers.17,18 In
contrast, we previously reported that the male cells of presumably fetal origin
observed in the thyroid of a woman affected with a multinodular goiter had
a follicular morphology.19 We therefore tested
our hypothesis by examining tissue specimens from women, affected with a variety
of diseases, who had male offspring to determine the morphology, cell surface,
and intracellular phenotype of fetal cells within maternal organs.
Individual male (XY+) microchimeric cells were evaluated for their cell
surface and intracellular phenotype. We selected tissue samples from 10 women,
previously studied by our laboratory group that had significant and easily
detectable male cell microchimerism.7,8,19- 21 All
of the initial studies documenting the presence of microchimerism included
appropriate control patients, including women who had no prior male pregnancies.
In the present study, we contemporaneously analyzed skin and cervical tissue
from 11 women with no known history of a male pregnancy. We obtained approval
from the institutional review board and written informed consent from all
patients who underwent surgery or biopsies. To the extent possible, we obtained
complete pregnancy histories from study participants, including the number
of sons, daughters, and abortions (spontaneous and elective), as well as the
possibility of other sources of microchimeric cells. None of the women had
a twin brother or had received an organ transplant at the time of tissue collection.
One woman had a history of blood transfusion from a donor of unknown sex.
We performed fluorescence in situ hybridization (FISH) analysis of the
tissue sections, as previously described,21,22 with
simultaneous immunolabeling.23 We tested 3
different mouse monoclonal IgG1 antibodies: AE1/AE3 anticytokeratin (Chemicon
International, Temecula, Calif) was used to identify epithelial cells, anti-CD45
(Dako, Carpintera, Calif) to identify leukocytes, and heppar-1 (Dako) to identify
hepatocytes. In all experiments, a mouse IgG1 (BD Bioscience, San Diego, Calif)
was used as an isotypic control.
Following hybridization and immunostaining, we included tissue sections
for subsequent analysis if the following criteria were met: FISH, immunostaining,
FISH Criteria. During the hybridization procedure,
there was minimal loss of cells and more than 75% of nuclei contained FISH
signals. Male cells had 2 different-colored FISH signals, representing both
the X and Y chromosomes, and an intact nuclear border. We recorded the coordinates
of microchimeric cells, which allowed us to retrieve 701 (97.9%) of 716 cells
on the slide. We also estimated the total number of nuclei in each section
by counting them in 10 fields at 400× magnification and counting the
number of fields to cover the whole tissue section. We then extrapolated the
frequency of male cells among a million maternal cells for each tissue section.
Immunostaining Criteria. We considered the
immunostaining results to be positive if target areas were stained and nontarget
areas were not stained. For CD45, the target areas were defined as nucleated
cells inside blood vessels and nontarget areas were defined as any epithelial
tissue. For heppar-1 and cytokeratin, the target area was defined as liver
parenchyma or epithelial area, respectively, and nontarget areas were defined
as cells inside blood vessels. In addition, to further prove the specificity
of our antibodies, we performed immunostaining with the anticytokeratin antibody
on liver, lymph node, and spleen tissue, and with heppar-1 on skin, spleen,
heart, and thyroid tissue. We also performed 2 series of immunostaining experiments
on a cord blood sample obtained during a full-term cesarean delivery with
all the antibodies described above to determine if circulating fetal cells
express hepatocyte or epithelial cell markers.
Morphologic Criteria. After evaluating both
FISH and immunostaining results, we stained tissue sections with hematoxylin
and eosin. We then relocated microchimeric male cells based on their slide
coordinates and assessed morphology and relative location within a section
using a light microscope. Cells that were not part of the section were excluded.
Morphology and immunostaining were independently evaluated by 2 investigators
(K.K. and R.N.S.).
Each microchimeric cell received a score of 0 if hematopoietic or 1
if epithelial or hepatocyte. The concordance between the morphology and immunohistochemical
assessments were compared for all cells that had both criteria scored by estimating
the κ value. In thyroid specimens, the cells were also evaluated as
being inside or outside the diseased area of the tissue section. Thyroid samples
were the only specimens in which the pathologic area (adenomatous tissue)
could be clearly distinguished from healthy surrounding tissue. All other
specimens contained exclusively diseased tissue. We compared the frequency
of cytokeratin-positive microchimeric cells inside and outside the diseased
area by using the Kruskal-Wallis test.24
We performed FISH analyses and identified a total number of 701 XY+
cells (mean [SD], 227  XY+ microchimeric per million maternal cells)
in archived paraffin-embedded tissue section specimens from 10 women (mean
age, 51.7 years; range, 34-74 years) who had male offspring. We subsequently
evaluated the cell surface and intracellular phenotype of the XY+ cells by
immunolabeling, morphology, and relative location within the sample (Table 1). We also performed FISH analysis
on tissue biopsies from women who had no history of a male pregnancy (n =
11) and found no XY+ cells (Table 2).
Anticytokeratin did not stain hematopoietic tissues, such as lymph node
or spleen, but did stain biliary epithelium as expected. Antihepatocyte antibody
(heppar-1) was specific for liver and did not stain any of the additional
tissues tested (skin, heart, thyroid, and spleen). In addition, unlike anti-CD45,
anticytokeratin and heppar-1 antibodies did not stain cord blood cells. In
90% of cases in which there was positive immunochemical staining of a cell,
independent histological assessment of that cell after hematoxylin and eosin
staining was substantially concordant with regard to morphology (κ =
We found a mean (SD) frequency of 190 (157) XY+ microchimeric cells
per million maternal cells among 3 women with multinodular goiters who underwent
partial thyroidectomy. In each of the 3 women, 14% to 60% of the XY+ cells
stained positively with cytokeratin, a marker of epithelial differentiation
(Figure 1). In 1 case, some of the
XY+ fetal cells that expressed cytokeratin were integrated into a thyroid
follicle. In 2 of 3 thyroid specimens, none of the microchimeric cells expressed
CD45, a common leukocyte antigen. A large inflammatory infiltrate was observed
in the third woman's thyroid; 67% of the XY+ cells expressed CD45.
We also analyzed the differentiation pattern of XY+ cells, according
to their physical location within a pathologic or healthy area. The 3 thyroid
specimens studied included a macroscopically visible adenoma surrounded by
healthy thyroid tissue. Histological examination of these 3 specimens revealed
that most of the microchimeric cells (114 of 150 cells successfully relocated)
were not part of the adenomatous tissue but were in the surrounding healthy
thyroid tissue. Interestingly, fetal cells inside the adenoma (36 of 150)
had a significantly higher percentage of cytokeratin expression than cells
outside the adenoma (92% vs 17%, respectively; P<.001).
The reverse situation was found for CD45: XY+ cells outside the adenoma more
frequently expressed CD45 than cells inside the adenoma (32% vs 3%, respectively; P<.001).
We also analyzed other epithelial tissues, such as cervical epithelium
specimens from 3 women, and digestive epithelial (gallbladder, intestine)
tissues from 2 women. We found a comparable pattern of differentiation, 20%
to 56% of the XY+ cells expressed cytokeratin and 30% to 55% expressed CD45.
In hematopoietic tissues, such as lymph nodes and spleen from 2 women,
90% of the XY+ cells expressed CD45. None of the cells expressed cytokeratin.
We also performed double staining (CD45 and cytokeratin) in most tissues;
microchimeric cells never stained positively with both antibodies.
In liver specimens of 2 women (patients G and I), most of the XY+ cells
expressed CD45 (Figure 2). In 1
woman, 4% of the fetal microchimeric cells stained with the hepatocyte marker
heppar-1. These cells had a morphology compatible with that of hepatocytes
The use of stem cells as a novel treatment for repair of diseased organs
in the human is an area of intense interest for the worldwide scientific community,
as well as the lay public and many governments. In this study, we show that
XY+ microchimeric cells in maternal tissues, acquired most likely through
pregnancy, express leukocyte, hepatocyte, and epithelial markers. These data
suggest that pregnancy may result in the physiologic acquisition of a fetal
cell population with the capacity for multilineage differentiation. We have
coined the term pregnancy-associated progenitor cells to
describe this population.
Our study was based on a small number of patients already selected for
having high numbers of microchimeric cells. We recognize that there is an
inherent selection bias in the study women, but to perform the study, adequate
numbers of fetal microchimeric cells must be present in the tissue to be further
analyzed. Therefore, the conclusions drawn from our study may only apply to
women with high numbers of microchimeric cells.
Most of the women did not have any additional sources of microchimerism,
such as solid organ transplantation. One of the 10 patients had a history
of blood transfusion. Transfusion-associated microchimerism is highly unlikely
to develop unless large quantities of blood are transfused in the setting
of trauma.25 Therefore, it is most likely that
the XY+ cells in this study are fetal in origin.
In almost all tissues, XY+ cells bearing CD45, the common leukocyte
antigen, were observed at variable frequencies. These results are consistent
with previous findings that suggest that fetal microchimeric cells are originally
blood cells, including hematopoietic progenitor cells.3,10,14,15
XY+ microchimeric cells that expressed cytokeratin, a marker of epithelial
cell differentiation, were never observed in hematopoietic tissues (eg, lymph
node). The concordance of morphological and immunohistochemical findings supports
the idea that some fetal cells may have an epithelial phenotype. In 1 woman,
in whom higher numbers of microchimeric cells were present, we were also able
to detect cells with evidence of a hepatocyte marker. We also show that fetal
cord blood cells do not express epithelial or hepatocyte markers, suggesting
that the microchimeric fetal cells acquire these markers in the environment
of maternal tissues.
Our study did not determine the type of fetal progenitor cells originally
transferred during the pregnancies of the women. Fetal blood contains a variety
of stem cell types, including mesenchymal stem cells and hematopoietic stem
cells.26 During pregnancy, fetal hematopoietic
and mesenchymal progenitor cells circulate within maternal blood and can be
cultured in maternal peripheral blood for up to 6 months after delivery.10,12,14
Fetomaternal transfusion may be even higher after an elective termination
of pregnancy.27 We have shown previously by
meta-analysis that a reproductive history that includes an elective termination
or an early fetal loss is associated with a higher incidence of microchimerism
in maternal tissues.28 The CD34+ fetal
cells are present in maternal blood for decades after delivery in 75% of women
studied,3 as well as in the CD34+–enriched
cell fraction of women undergoing granulocyte colony-stimulating factor bone
marrow stimulation.13 Our results imply but
do not prove that fetal CD34+ hematopoietic stem cells that persist
post partum may have multilineage capacity. Another possibility is that pregnancy
results in the acquisition of a different type of circulating stem cell, perhaps
from the placenta, which has epithelial characteristics.
The nonhematopoietic morphology and phenotypes of the fetal cells that
we observed may result from different mechanisms. Fetal progenitor cells could
transdifferentiate into hematopoietic, hepatic, or epithelial cells. They
could also adopt the host tissue phenotype by fusing with hepatocytes or epithelial
cells.29 In our identification of microchimeric
XY+ cells based on X and Y chromosome FISH signals, we never detected an XY+
cell with an interphase karyotype suggestive of a fused nucleus (XXXY) or
having 2 separate nuclei. However, we cannot exclude the possibility that
some fetal and maternal cells fuse their cytoplasm, especially in dense tissues,
such as liver, in which the outer limits of each cell are hard to distinguish.
Whatever the mechanism involved, we believe that the idea of fetal cells expressing
nonhematopoietic markers is novel and may have important long-term health
implications for the woman who has undergone pregnancy by providing her with
a younger population of cells that may have different capabilities in the
response to tissue injury.
In conclusion, we have shown that fetal cells, in a variety of maternal
tissues, have morphologic and protein expression characteristics of not only
hematopoietic but also epithelial and hepatic cells. These data suggest that,
at least in some women after pregnancy, fetal cells transferred during pregnancy
develop multilineage capacity either by cell fusion or transdifferentiation.
Further study of naturally occurring fetal cell microchimerism may be useful
in determining the characteristics of the specific progenitor cell population
and the exact mechanisms involved in its apparent differentiation.