Fusobacterium infection facilitates the development of endometriosis through the phenotypic transition of endometrial fibroblasts

Abstract

Retrograde menstruation is a widely accepted cause of endometriosis. However, not all women who experience retrograde menstruation develop endometriosis, and the mechanisms underlying these observations are not yet understood. Here, we demonstrated a pathogenic role of Fusobacterium in the formation of ovarian endometriosis. In a cohort of women, 64% of patients with endometriosis but <10% of controls were found to have Fusobacterium infiltration in the endometrium. Immunohistochemical and biochemical analyses revealed that activated transforming growth factor-β (TGF-β) signaling resulting from Fusobacterium infection of endometrial cells led to the transition from quiescent fibroblasts to transgelin (TAGLN)-positive myofibroblasts, which gained the ability to proliferate, adhere, and migrate in vitro. Fusobacterium inoculation in a syngeneic mouse model of endometriosis resulted in a marked increase in TAGLN-positive myofibroblasts and increased number and weight of endometriotic lesions. Furthermore, antibiotic treatment largely prevented establishment of endometriosis and reduced the number and weight of established endometriotic lesions in the mouse model. Our data support a mechanism for the pathogenesis of endometriosis via Fusobacterium infection and suggest that eradication of this bacterium could be an approach to treat endometriosis.

Full text

ENDOMETRIOSIS
Fusobacterium infection facilitates the development of
endometriosis through the phenotypic transition of
endometrial fibroblasts
Ayako Muraoka
1,2
, Miho Suzuki
1
, Tomonari Hamaguchi
3
, Shinya Watanabe
1
, Kenta Iijima
1
,
Yoshiteru Murofushi
1
, Keiko Shinjo
1
, Satoko Osuka
2
, Yumi Hariyama
4
, Mikako Ito
3
, Kinji Ohno
3
,
Tohru Kiyono
5
, Satoru Kyo
6
, Akira Iwase
7
, Fumitaka Kikkawa
2
, Hiroaki Kajiyama
2
,
Yutaka Kondo
1,8
*
Retrograde menstruation is a widely accepted cause of endometriosis. However, not all women who experience
retrograde menstruation develop endometriosis, and the mechanisms underlying these observations are not
yet understood. Here, we demonstrated a pathogenic role of Fusobacterium in the formation of ovarian endo-
metriosis. In a cohort of women, 64% of patients with endometriosis but <10% of controls were found to have
Fusobacterium infiltration in the endometrium. Immunohistochemical and biochemical analyses revealed that
activated transforming growth factor–β (TGF-β) signaling resulting from Fusobacterium infection of endometrial
cells led to the transition from quiescent fibroblasts to transgelin (TAGLN)–positive myofibroblasts, which
gained the ability to proliferate, adhere, and migrate in vitro. Fusobacterium inoculation in a syngeneic
mouse model of endometriosis resulted in a marked increase in TAGLN-positive myofibroblasts and increased
number and weight of endometriotic lesions. Furthermore, antibiotic treatment largely prevented establish-
ment of endometriosis and reduced the number and weight of established endometriotic lesions in the
mouse model. Our data support a mechanism for the pathogenesis of endometriosis via Fusobacterium infection
and suggest that eradication of this bacterium could be an approach to treat endometriosis.
Copyright © 2023 The
Authors, some
rights reserved;
exclusive licensee
American Association
for the Advancement
of Science. No claim
to original U.S.
Government Works
INTRODUCTION
Endometriosis is caused by endometrial-like tissue containing en-
dometrial glands and extensive fibrotic tissue growing outside the
endometrial cavity, most often in the pelvic peritoneum or
ovaries, resulting in chronic pelvic pain and infertility (1). It is a
common gynecological disease affecting 10 to 15% of women of re-
productive age (2). Several hypotheses have been proposed to
explain the cause of endometriosis, including retrograde menstru-
ation, coelomic metaplasia, and the presence of Mullerian remnants
(3). Of these, retrograde menstruation, where menstrual blood flows
back through the fallopian tubes and into the pelvic cavity instead of
out of the vagina, is widely accepted as a cause of endometriosis.
However, although most women of reproductive age experience ret-
rograde menstruation (3), only 10 to 15% of women develop endo-
metriosis. This suggests the existence of other mechanisms that
facilitate the development of endometriosis.
Although endometriosis outgrowths are benign, the abundant
desmoplastic stroma, where robust fibroblast proliferation and
migration are promoted by secreted growth factors, is clinically
problematic (4). The characteristics of the stromal fibroblasts are
key determinants of the growth of newly established endometriosis
and subsequently its progression (5). Recent studies have revealed
that quiescent fibroblasts are activated to differentiate into myofi-
broblasts during wound healing and chronic inflammatory condi-
tions such as fibrosis and cancer (6–8). Activated myofibroblasts
contain cytoplasmic microfilament–associated proteins such as α-
smooth muscle actin (αSMA; encoded by ACTA2), vimentin
(VIM), and transgelin [TAGLN; also known as smooth muscle
22α (SM22α)]. The latter is a calponin-related actin-binding cyto-
skeletal protein that is linked to increased cell motility and migra-
tion and is expressed in mesenchymal lineage cell types
(myofibroblasts and smooth muscle cells) (9,10). The transition
from quiescent fibroblasts to activated myofibroblasts is triggered
by chronic inflammation that involves the production of multiple
cytokines, including transforming growth factor–β (TGF-β) (6,
11,12).
Although a large proportion of the bacteria in the vagina are Lac-
tobacilli, studies have documented the presence of other types of
bacteria as well, such as Fusobacterium nucleatum, which may be
associated with vaginal dysbiosis in certain situations (13,14).
Species of the Fusobacterium genus are known to be common
members of the oral and gastrointestinal tract microbiota and
have a symbiotic relationship with its hosts (15). Although the
uterine cavity is believed to be almost sterile, a close association
between endometriosis and endometritis has been reported (16).
Recent highly sensitive 16Sribosomal RNA (rRNA) gene amplicon
sequencing has demonstrated that, although the amount of bacteria
in the uterus is about four orders of magnitude less than in the
vagina, certain microbial communities, including members of the
1
Division of Cancer Biology, Nagoya University Graduate School of Medicine, 65
Tsurumai-cho, Showa-ku, Nagoya 466-8550, Japan.
2
Department of Obstetrics
and Gynecology, Nagoya University Graduate School of Medicine, 65 Tsurumai-
cho, Showa-ku, Nagoya 466-8550, Japan.
3
Division of Neurogenetics, Nagoya Uni-
versity Graduate School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-
8550, Japan.
4
Department of Obstetrics and Gynecology, Toyota Kosei Hospital,
500-1, Ihohara, Zyosui-cho, Toyota 470-0396, Japan.
5
Project for Prevention of
HPV-related Cancer, Exploratory Oncology Research and Clinical Trial Center, Na-
tional Cancer Center, Kashiwanoha 6-5-1, Kashiwa 277-8577, Japan.
6
Department
of Obstetrics and Gynecology, Shimane University Facultyof Medicine, 89-1 Enya-
Cho, Izumo 693-8501, Japan.
7
Department of Obstetrics and Gynecology, Gunma
University Graduate School of Medicine, 3-39-22 Showa-machi, Maebashi 371-
8511, Japan.
8
Institute for Glyco-core Research (iGCORE), Nagoya University,
Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan.
*Corresponding author: Email: ykondo@med.nagoya-u.ac.jp
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 1 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Fusobacterium genus, can be detected (17). Although members of
the Fusobacterium genus have long been considered opportunistic
pathogens, recent studies revealed that some Fusobacterium species
such as F. nucleatum play an important role not only in periodon-
titis but also in carcinogenesis via induction of several inflammatory
cytokines, such as interleukin-6 (IL-6), IL-8, and tumor necrosis
factor (TNF) (15). In this study, we asked whether the endometrial
stromata of patients with endometriosis contained different types of
fibroblasts compared with people without endometriosis and
whether these fibroblasts might be associated with the presence of
specific pathogenic bacteria.
RESULTS
TAGLN is up-regulated in broblasts from patients with
endometriosis
First, we undertook gene expression profiling and compared uterus
endometrial fibroblasts from patients without endometriosis
(UTnon-FBs; n= 4, two in menstrual cycle proliferative phase
and two in secretory phase) with those from ovarian endometriotic
lesions [ovarian endometriotic fibroblasts (OVend-FBs); n= 4, also
two in each phase]. There was no significant difference in age, men-
strual period, gravidity and parity, or history of repeated surgery
between patients with UTnon-FBs and OVend-FBs (P> 0.05,
Fisher’s exact test; tables S1 to S3). We found that 1187 genes
were significantly up-regulated and 1084 down-regulated in endo-
metriotic tissues [defined as fold change (FC) > 2.0, P< 0.05; Fig. 1,
A and B, and fig. S1A]. The expression profiles in our dataset were
compared with three public datasets of normal human endometri-
um and endometriosis samples (GSE25628, GSE99949, and
GSE7305), which were derived from endometrial or endometriosis
tissue (18–20). This revealed that 10 genes were up-regulated and 3
were down-regulated in endometriosis samples relative to normal
endometrial tissues across all four datasets (table S4).
The expression of these 13 genes was validated by quantitative
reverse transcription polymerase chain reaction (qRT-PCR) in
primary UTnon-FBs (n= 14) compared with primary OVend-
FBs and primary fibroblasts from the corresponding eutopic endo-
metrial tissues [uterus endometrial fibroblasts with endometriosis
(UTend-FBs)] from patients with endometriosis (n= 14). Of
these 13 genes, there was a significant stepwise increasing mRNA
expression for only TAGLN, with the lowest amount in UTnon-
FBs, followed by UTend-FBs and OVend-FBs (P< 0.05 and P<
0.01 for UTnon-FBs versus UTend-FBs and UTend-FBs versus
OVend-FBs, respectively; Fig. 1C and fig. S1, B to D). TAGLN
protein amounts were also increasingly up-regulated in UTnon-
FBs and UTend-FBs, and the greatest abundance was found in
OVend-FBs (P< 0.01; Fig. 1D). Fibroblasts in only a part of the en-
dometrium expressed TAGLN, but this protein was present in most
fibroblasts in the corresponding ovarian endometriotic lesions from
the same patients with endometriosis (Fig. 1E and fig. S1F).
Gene expression profiling was further examined by single-cell
RNA sequencing (scRNA-seq). In the endometrium without endo-
metriosis (UTnon-EM) and ovarian endometriosis (OVend)
samples, scRNA-seq revealed that cells classified into groups, in-
cluding epithelial cells, macrophages/monocytes, mast cells, T
cells/natural killer (NK) cells, endothelial cells, or fibroblasts, ex-
pressed characteristic marker genes for each type of cell (fig. S2, A
to C) (21). Fibroblasts from UTnon-EM and OVend were classified
into distinct subpopulations (Fig. 1F). TAGLN was substantially up-
regulated in the subpopulation of fibroblasts from OVend (Fig. 1, F
and G). Nine of 10 genes identified as up-regulated by gene expres-
sion profiling were more highly expressed in fibroblasts from
OVend than from UTnon-EM (table S5).
Up-regulation of TAGLN promotes broblast proliferation
and mobility
TAGLN functions as an actin cross-linking protein of the calponin
family and is a marker of myofibroblasts (9,10). It is localized to the
cytoskeletal apparatus and confers contractile function on cells (22).
The established OVend-FB cell lines SC8 and SC10 retained higher
expression of TAGLN mRNA and protein than the UTnon-FB cell
lines MC1 and MC2 (Figs. 1A and 2, A and B). Characteristic ex-
pression profiles of myofibroblast markers, such as ACTA2,TAGLN,
platelet-derived growth factor subunit A (PDGFA), and VIM, in
these four cell lines revealed that SC8 and SC10 represented myofi-
broblasts, whereas MC1 and MC2 represented quiescent fibroblasts
(fig. S3, A to D) (6).
Depletion of TAGLN by small interfering RNA (siRNA) trans-
fection significantly decreased proliferation of OVend-FB cell lines
SC8 and SC10 (P< 0.05; Fig. 2C and fig. S4, A and B). TAGLN de-
pletion also impaired the migration ability of OVend-FB cell lines
and their attachment to mesothelial cells (Fig. 2, D and E). Recip-
rocally, overexpression of TAGLN by transfection of a TAGLN-ex-
pressing vector in the UTnon-FB cell lines MC1 and MC2 increased
proliferation, migration, and attachment (fig. S4, C to G).
TAGLN-expressing OVend-FBs promote endometrial cell
proliferation via IL-6
Colocalization of TAGLN with circular bundles of actin (αSMA)
microfilaments was observed in OVend-FBs that had a polygonal
shape (Fig. 2F). The abundance of both TAGLN and αSMA was in-
creased in cell lines SC8 and SC10 (Fig. 2B and fig. S5A). Depletion
of TAGLN by siRNA transfection in cell lines SC8 and SC10 result-
ed in marked morphological changes, with cells taking an elongated
spindle shape in which circular actin bundles were no longer
present and straight bundles were formed across the cell body,
with no change of αSMA expression (Fig. 2F and fig. S5B). These
morphological changes suggest that the cells may have decreased
contractility. The cell lines MC1 and MC2 overexpressing TAGLN
obtained a polygonal shape (fig. S5C).
The C-terminal calponin-like module (CLIK) of TAGLN (Fig.
2G) is known to interact with αSMA, which confers contractility
to cells (23). Overexpression of TAGLN-ΔCLIK in TAGLN-deplet-
ed SC8 and SC10 cells using si-TAGLN#2 (see Materials and
Methods) altered the colocalization of TAGLN and αSMA.
TAGLN accumulated at the centers of the cells, resulting in an en-
larged, flattened shape (Fig. 2H). Transfection of TAGLN-FL (full-
length TAGLN) into the MC2 cell line resulted in the acquisition of
a polygonal shape, whereas TAGLN-ΔCLIK took on an enlarged,
flattened shape (fig. S5D).
Myofibroblasts produce and secrete a number of cytokines when
they contract (24). We examined cytokines correlating with TAGLN
expression in cell lines of UTnon-FBs or OVend-FBs and found that
IL-6 was the most prominent among the 30 cytokines and growth
factors tested on the protein array (Fig. 2I and fig. S6A). However,
overexpression of TAGLN-ΔCLIK did not up-regulate IL-6 produc-
tion (fig. S6B) (25). Addition of recombinant IL-6 to the cell culture
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 2 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Fig. 1. Up-regulation of TAGLN in both eutopic endometrial and ovarian endometriotic broblasts from patients with endometriosis. (A) Schema of fibroblasts
and immortalized cell lines in this study. (B) Venn diagram of up-regulated genes in OVend-FBs combined with the public data. (C) mRNA expression of TAGLN in primary
UTnon-FBs (n= 14), UTend-FBs (n= 14), and OVend-FBs (n= 14). The yaxis indicates the mRNA of TAGLN relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
(D) TAGLN protein abundance in primary fibroblastsfrom each patient (in the left gel image, patient numbers correspond to table S1). The expression of TAGLN relativeto
GAPDH was calculated (right, n= 3 in each group). (E) Immunohistochemistry of TAGLN in endometrium without endometriosis (UTnon-EM), endometrium with endo-
metriosis (UTend-EM), and ovarian endometriosis (OVend) (left). IHC scores of UTnon-EM (n= 42), UTend-EM (n= 42), and OVend (n= 42) are shown as dot plots (right). (F)
UMAP plots of TAGLN expression within the fibroblast subpopulation (right). Orange- and turquoise-colored fibroblast cells were derived from OVend and UTnon-EM,
respectively (left). (G) Violin plots of total read count of TAGLN within fibroblasts. Scale bars, 100 μm. Error bars indicate SEM (C and D) or SD (E). *P< 0.05 and **P< 0.01.
Data were analyzed by two-tailed Student’sttest.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 3 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Fig. 2. Functional analyses of TAGLN in endometrial and endometriotic broblasts. (Aand B) TAGLN mRNA (A) and protein (B) in the MC1, MC2, SC8, and SC10 cell
lines. The yaxis indicates the mRNA of TAGLN relative to GAPDH (A). Ratios of TAGLN to GAPDH are shown below the gel images (B). (Cto E) Cell proliferation (C), migration
(D), and attachment assays (E) of the SC8 and SC10 cell lines after TAGLN depletion. The yaxis indicates cell proliferation rate relative to day 1 (C). The bar graphs show the
number of migrated or expressing GFP-attached cells (D and E). Scale bars, 100 μm. (F) Presence of TAGLN (green) and αSMA (red, left) and quantification of cell length
(right) in SC8 and SC10 cells with siRNA against TAGLN (#1, #2). For each sample, 30 cells were measured. Nuclei were stained with DAPI (blue). Insets, lower magnifications.
Scale bars, 50 μm. (G) Schema of full-length (FL) and truncated (ΔCLIK 152 to 201) TAGLN protein. CH, calponin homology; CLIK, C-terminal calponin-like module. (H)
Localization of exogenous Flag-tagged TAGLN. Scale bars, 50 μm. (I) Presence of TAGLN (green) and IL-6 (red) in SC10 after depletion of TAGLN (left). Mean fluorescent
intensity of each cell was quantified (right). Thirty cells were analyzed per sample. Nuclei were stained with DAPI. Scale bars, 100 μm. (J) Proliferation of MC2 and SC10 cells
with/without IL-6 treatment (10 ng/ml). The yaxis indicates cell proliferation rate relative to day 1. Error bars indicate SD (A, C to F, I, and J). *P< 0.05 and **P< 0.01. Data
were analyzed by one-way ANOVA (A, D to F, and I) or two-tailed Student’sttest (C and J), respectively.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 4 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

medium promoted the proliferation of all four cell lines tested (Fig.
2J and fig. S6C). Consistent with this in vitro analysis, clinical
samples also expressed IL-6 at increasing amounts in endometria
from patients without endometriosis (n= 10), more in endometria
(n= 10), and most in the corresponding ovarian endometriotic
lesions from patients with endometriosis (n= 16) (fig. S6D). Fur-
thermore, the protein expression of TAGLN was positively correlat-
ed with IL-6 in these tissues (fig. S6E).
TGF-β1 induces the expression of TAGLN
TGF-β1 is known to be a key growth factor that plays a major role in
endometriosis (26,27). Furthermore, TGF-β1 is known to induce
myofibroblast-associated genes (fig. S7A), including TAGLN, via
sma- and mad-related (SMAD) binding to its promoter (28). We
therefore asked whether TGF-β signaling induced TAGLN expres-
sion in the endometrium. Abundance of TGF-β1 and TGF-β recep-
tor 1 (TGF-βR1) was increased in both stromal cells and epithelial
cells in endometrial tissues and endometriotic lesions from patients
relative to normal endometrium (Fig. 3, A and B). Treatment with
TGF-β1 increased the abundance of TAGLN in MC1 and MC2 cell
lines. Up-regulation of TAGLN after TGF-β treatment was abol-
ished by SB431542, a specific inhibitor of SMAD2/3-mediated
signal transduction (Fig. 3C) (29). These changes in protein expres-
sion were associated with the enrichment of histone H3 lysine 27
acetylation (H3K27ac) in the enhancer regions and SMAD2/3 in
the promoter regions of the TAGLN gene in MC1 and MC2 cell
lines (Fig. 3D and fig. S7B). These findings suggested that epigenetic
regulation was involved in the transition from quiescent fibroblasts
into myofibroblasts via TGF-β signaling.
Fusobacterium infection inuences the endometrial
microenvironment
We hypothesized that TGF-β abundance in the endometrial micro-
environment might be associated with bacterial infection. We ana-
lyzed the publicly available dataset from a previous study
(PRJEB16013 and PRJEB21098: https://ebi.ac.uk/ena/browser/
home) (17), which revealed that five bacterial genera were signifi-
cantly increased in the endometria of patients with endometriosis
relative to healthy controls (P< 0.01) (table S6, see also Materials
and Methods). Among these bacterial genera, the top-ranked
species, Erysipelothrix, was not present in substantial amounts in
endometrial tissues from individuals either with or without endo-
metriosis, as assessed by qPCR analysis (fig. S8A). Therefore, we ex-
amined the second-ranking bacterium, Fusobacterium. This genus,
known as a symbiont organism, was detected by qPCR in our clin-
ical cohort of endometrial tissues (fig. S8A).
Among Fusobacterium species, the presence of F. nucleatum in
the vagina has previously been reported (13). We tested this using
the fluorescence in situ hybridization (FISH) probe EUB338, which,
based on publicly available bacterial 16SrRNA sequences, covers
most of the eubacteria (30), and probes for Fusobacterium species
(FUSO) (31) or for F. nucleatum (FUS664) (32–34). We found sig-
nificantly higher frequencies of Fusobacterium inside endometrial
tissues (27 of 42, 64.3%) and endometriotic tissues (22 of 42,
52.4%) from patients with endometriosis than in controls without
endometriosis (3 of 42, 7.1%) (P< 0.01; Fig. 4, A and B, and fig.
S8B). The amount of infiltrated Fusobacterium was significantly
higher in both tissues from patients with endometriosis than con-
trols (P< 0.01; Fig. 4B). It was notable that the EUB338, FUSO, and
FUS664 signals were almost completely colocalized (>96% of the
signals were merged), suggesting that the bacteria infiltrating the
endometrium were mainly F. nucleatum (fig. S8C).
Infiltration of F. nucleatum may induce innate immune respons-
es via its membrane lipopolysaccharide (35), and macrophages are
the most abundant immune cells within endometriotic lesions (36).
We observed increased numbers of macrophages infiltrating endo-
metrial tissues, with a further increase in staining in ovarian endo-
metriotic tissues compared with controls (fig. S9A). Increased
numbers of CD163-positive M2 macrophages, which are known
to produce TGF-β1 (37), were present in both of these tissues
from patients with endometriosis (Fig. 4, C and D). Immunohisto-
chemistry (IHC) analysis revealed that TGF-β1 in the stroma was
significantly different between Fusobacterium-positive and Fuso-
bacterium-negative tissues from both UTend-EM (endometrium
with endometriosis) and OVend (P< 0.01 and P< 0.05, respectively;
fig. S9B). In contrast, TGF-β1 in the epithelium was not different
between Fusobacterium-positive and Fusobacterium-negative
tissues, although TGF-β1 is also known to be secreted from epithe-
lial cells (P> 0.05; fig. S9B).
We cocultured cells from the macrophage cell line THP1 [THP-
1–derived macrophages (dTHP1)] with F. nucleatum or Lactobacil-
lus iners, which are the major indigenous bacteria of the vaginal mi-
crobiota in reproductive-age women, especially in Asian countries
(38,39). This revealed that even heat-killed F. nucleatum promoted
differentiation of dTHP1 cells into M2 macrophages and stimulated
the production of TGF-β1. In contrast, L. iners did not do so (P<
0.05; Fig. 4E). Furthermore, cell-free supernatants from the cocul-
tures of dTHP1 with F. nucleatum increased TAGLN protein ex-
pression in MC2 cells (P< 0.05; Fig. 4F and fig. S9C). The
amounts of Fusobacterium, numbers of CD163-positive macro-
phages, and IHC scores of TGF-β1 were positively correlated with
each other in the endometrial stromal tissues (Fig. 4G). Amounts of
Fusobacterium and TAGLN expression were significantly correlated
in both the endometrial tissues and endometriotic tissues from pa-
tients with endometriosis (P< 0.01 and P< 0.01, respectively),
whereas the abundances of both Fusobacterium and TAGLN were
low in controls (fig. S9D and table S7). Together, these data suggest
that F. nucleatum within the endometrium may influence TAGLN
abundance in fibroblasts via up-regulated TGF-β1 signaling.
F. nucleatum infection promotes endometriosis in vivo
Next, we asked whether bacterial infection promoted endometriosis
in a syngeneic mouse model in which F. nucleatum were intravagi-
nally introduced into the uteri of donor BALB/c mice (Fig. 5A). En-
dometriotic lesions in these models were induced by peritoneal
injection of minced endometrial tissue by transferring endometrial
tissues from one animal to a syngeneic animal with an intact
immune system (40). Stromal cells in endometrial tissues were val-
idated to express both estrogen receptor α (ERα) and β (ERβ) (fig.
S10A). First, we determined the volume of transferred endometrial
tissues from donor mice. Transferring of 1 or 2 mg of tissues from F.
nucleatum–infected uteri of donor mice resulted in the formation of
multiple endometriotic lesions, whereas tissues from F. nucleatum–
uninfected uteri did not do so, even when stimulated by estrogen
(Fig. 5B).
To ensure that the endometriotic lesions were stably established
in all recipient mice regardless of F. nucleatum infection, 8 mg of
endometrial tissues were used for this transfer model. In the
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 5 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

recipient endometriotic lesions, F. nucleatum was also observed, as
was seen in the donor uterus (Fig. 5C). More abundant M2 macro-
phage infiltration, expression of TGF-β1, and TAGLN-positive my-
ofibroblasts were observed in both the endometrial stromata of F.
nucleatum–infected donor uteri and the endometriotic lesions es-
tablished from F. nucleatum–infected uteri relative to the controls
(P< 0.01; Fig. 5, D to G, and fig. S10B). Immunofluorescence anal-
ysis revealed infiltration of F. nucleatum organisms colocalized with
TAGLN-positive cells in the F. nucleatum–infected endometrial
stroma (Fig. 5H). High TAGLN protein expression was observed
only in stromal cells but not in the epithelial cells in both the
donor uterus and endometriotic lesions after infection with F. nu-
cleatum (Fig. 5H). F. nucleatum–infected endometrium showed
high TGF-β1 abundance in the stroma with TAGLN expression
(fig. S10C). In comparison with F. nucleatum,L. iners did not
lead to M2 macrophage infiltration, induction of TGF-β1, or
increased numbers of TAGLN-positive myofibroblasts after bacte-
rial injection (Fig. 6, A to D, and fig. S10B). Consequently, L. iners
injection did not facilitate development of endometriotic lesions
relative to controls (Fig. 6E). We also tested the ability of a different
Gram-negative bacterium, Escherichia coli, to form endometriotic
lesions using the mouse model. However, this organism had no
effect on numbers and weights of endometriotic lesions, similar
to L. iners (Gram-positive bacteria) (fig. S10D). These data
support F. nucleatum as a candidate bacterium associated with es-
tablishment of endometriosis.
In the clinical setting, the presence of Fusobacterium in the
vaginal swab samples from patients with endometriosis was signifi-
cantly greater than from patients without endometriosis (P= 0.023;
table S8). However, hematogenous transmission of Fusobacterium
from the oral cavity to the uterus is also a possible route (41). Inoc-
ulation of donor BALB/c female mice with F. nucleatum via the
Fig. 3. TGF-β1 induces expression of TAGLN in vitro. (Aand B) Representative images of immunohistochemical analysisof TGF-β1 (A) and TGF-βR1 (B) in UTnon-EM (n=
44), UTend-EM (n= 37), and OVend (n= 49). Scale bars, 100 μm (left). Tissues were classified by the IHC score of TGF-β1 (A) and TGF-βR1 (B) in stromal lesions of each
sample (±, 0 to 100; 1+, 101 to 200; 2+, 201 to 300; right). (C) Abundance of TAGLN, phospho-SMAD2 (P-SMAD2), and total SMAD2 (SMAD2) in MC1 and MC2 cells after
treatment with TGF-β1 (10 ng/ml) and SMAD2 inhibitor (SB431542, 1 μM). Ratios of TAGLN to GAPDH and P-SMAD2 to SMAD2 are shown below each gel image. (D) ChIP-
qPCR analyses of H3K27ac and SMAD2/3 enrichment in the TAGLN enhancer and promoter regions, respectively. MC1 and MC2 cells were treated with either TGF-β1 or
SB431542 (SMAD2-specific inhibitor). Error bars indicate SEM. *P< 0.05 and **P< 0.01. Data were analyzed by Fisher’s exact test (A and B) or two-tailed Student’sttest (D),
respectively.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 6 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Fig. 4. Fusobacterium infection inuences the microenvironment of the endometrium. (A) FISH analysis using the EUB338 probe for whole bacteria (green) and
FUSO for Fusobacterium species (magenta) in UTnon-EM, UTend-EM, and OVend. Nuclei were stained with DAPI (blue). Patient numbers (No.) correspond to table S1. (B)
Quantification of the numbers of FUSO probe spots in (A) (n= 42 in each group). (C) Abundance of CD163 in each tissue (n= 42, left). Middle horizontal line inside the box
indicates the median value. Bottom and top of the box indicate the 25th and 75th percentiles, respectively. Ends of the whiskers indicatethe minimum and maximum of
all data, respectively (right). (D) Immunofluorescence staining of TGF-β1 (green) and CD163 (magenta) in UTnon-EM, UTend-EM, and OVend. Nuclei were stained with
DAPI (blue). (E) Abundance of TGF-β1 (44-kDa precursor and 13-kDa cleaved mature proteins) and CD163 in dTHP1 cells cocultured with F. nucleatum (F. nuc.), L. iners (L.
ine.), or heat-killed F. nuc. or L. ine., as analyzed by Western blotting (left). The relative protein expression of CD163 to GAPDH was quantified (right). (F) Abundance of
TAGLN in MC2 cells treated with the indicated conditioned medium (left). The relative protein expression of TAGLN to GAPDH was quantified (right). (G) Dot plot analysis
between amounts of Fusobacterium, CD163-positive cells, and TGF-β1 in UTnon-EM (n= 42), UTend-EM (n= 37), and OVend (n= 42). The x,y, and zaxes indicate the
numbers of FUSO probe spots, numbers of CD163-positive cells, and IHC scores of TGF-β1. Pearson‘s correlation coefficients are shown in the upper left of each square.
Error bars indicate 95% confidence interval (B) and SD (E and F). *P< 0.05, **P< 0.01. Data were analyzed by Wilcoxon rank-sum test (B) and two-tailed Student’sttest (C
and E) and one-way ANOVA (F), respectively. Scale bars, 100 μm.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 7 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Fig. 5. F. nucleatum infection promotes endometriosis in mice. (A) Donor mice were transvaginally infected with F. nucleatum (F. nuc.) or L. iners (L. ine.) (red arrow).
Endometrial tissues were intraperitoneally injected into recipient mice (white arrow).Gray triangles indicate 17β-estradiol treatments. To ensurethe endometriotic lesions
established in all the recipient mice, 8 mg of endometrial tissue were inoculated (Cto H). (B) Number and weights of lesions from recipient mice processed in (A) (n= 5 in
each group). (C) Detection of Fusobacterium with EUB338 (green) and FUSO (magenta) probes in endometria of donor mice infected without and with F. nuc. (left: CTRL
and F. nuc., respectively) and endometriotic lesions (right: CTRL and F. nuc., respectively). DAPI (blue). Scale bars, 100 μm. (D and E) Presence of CD163, TGF-β1, and TAGLN
in endometria from the infected donor mice. Scale bars, 50 μm. Numbers of infiltrated macrophages and IHC scores of TGF-β1 and TAGLN are shown in the boxplot (n= 6
in each group). Middle horizontal line inside the box indicates the median value. Bottom and top of the box indicate the 25th and 75th percentiles, respectively. Ends of
the whiskers indicate the minimum and maximum of all data, respectively. (F and G) Presence of CD163, TGF-β1, and TAGLN in endometriotic lesions. Scale bars, 500 μm
(H.E., hematoxylin and eosin) and 50 μm (CD163, TGF-β1, and TAGLN). Numbers of infiltrating macrophages and IHC scores of TGF-β1 and TAGLN are shownin the boxplot
(n= 6 in each group). (H) Immunofluorescence of TAGLN (green) and Fusobacterium (magenta) in the endometria from infected mice. Nuclei were stained with DAPI
(blue). Scale bars, 100 μm. **P< 0.01. Data were analyzed by two-tailed Student’sttest. *P< 0.05.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 8 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

jugular vein showed that 56% exhibited F. nucleatum colonization of
the endometrium after 2 weeks (Fig. 6F). Endometriotic lesions es-
tablished from the F. nucleatum–positive endometrial tissues were
significantly larger than those from the F. nucleatum–negative
tissues in this hematogenous transmission model (P< 0.01; Fig. 6F).
To examine the requirement for TAGLN-expressing fibroblasts
in endometriosis formation after F. nucleatum infection in vivo,
TAGLN was depleted by siRNA in the tissues from F. nucleatum–
infected uteri of donor mice. Depletion of TAGLN significantly
reduced the numbers and weights of endometriotic lesions in com-
parison with the tissues treated with control siRNA (P< 0.01; fig.
S11, A to D). Reciprocally, forced expression of TAGLN in tissues
from donor uteri without F. nucleatum infection facilitated the in-
creased numbers and weights of endometriotic lesions in vivo (fig.
S11, E to H). These data implicated TAGLN protein expression in
endometriosis after F. nucleatum infection.
Antibiotic treatment that eradicates F. nucleatum results in
reduced endometriosis in a mouse model
Last, we explored the effects of antibiotic treatment to eradicate F.
nucleatum in the syngeneic mouse model. Two different types of
antibiotics, metronidazole (MZ) and chloramphenicol (CP), to
which F. nucleatum is sensitive, were administered transvaginally
every day for 5 days after infection was established in donor mice
Fig. 6. Eects of bacterial infection in a syngeneic mouse model of endometriosis. To ensure that the endometriotic lesions were established in all the recipient mice,
8 mg of endometrial tissue were inoculated (Ato D). Donor mice were transvaginally infected with L. iners (L. ine.) as is illustrated in Fig. 5A (A to D). (A) Presence of CD163,
TGF-β1, and TAGLN in endometria from the infected mice. Scale bars, 50 μm. (B) Numbers of infiltrated macrophages and IHC scores of TGF-β1 and TAGLN are shown in
the boxplot (n= 6 in each group). (C) Presence of CD163, TGF-β1, and TAGLN in endometriotic lesions of recipient mice. Scale bars, 500 μm (H.E.) and 50 μm (CD163, TGF-
β1, and TAGLN). (D) Numbers of infiltrating macrophages and IHC scores of TGF-β1 and TAGLN are shown in the boxplot (n= 6 in each group). (E) Relative weight of
endometriotic lesions from recipient mice processed in Fig. 5A (n= 6 in each group). (F) Infection rate of F. nucleatum (F. nuc.) in the endometria of donor mice after
intravenous injection. The numbers (left graph) and weights (right graph) of each endometriotic lesion from recipient mice. (−), n= 9 in the uninfected group and (+), n=
12 in the infected group. Uninfected group indicates donors without presence of F.nuc. in their endometrial tissues even after intravenous injection of F. nuc. The amount
of inoculation in Fig. 6F was 2 mg. Error bars indicate SEM (E and F). **P< 0.01; n.s., not significant. Middle horizontal line inside the box indicates the median value.
Bottom and top of the box indicate the 25th and 75th percentiles, respectively. Ends of the whiskers indicate the minimum and maximum of all data, respectively. Data
were analyzed by two-tailed Student’sttest.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 9 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Fig. 7. Antibiotic treatment that eradicates F. nucleatum reduces endometriotic lesion weight in mice. (A) Donor mice were infected with F. nucleatum (red arrow).
Endometrial tissues were injected into the recipients 1 week after infection (white arrow). To ensure that the endometriotic lesions established in all the recipient mice, 8
mg of endometrial tissue were inoculated (A to E). Three weeks after inoculation, recipient mice were treated with oral antibiotics for 5 days (green arrow). Gray triangles
indicate 17β-estradiol treatments. (B) Detection of Fusobacterium with EUB338 (green) and FUSO (magenta) probes in endometriotic lesions of recipient mice at day 1
(left) and day 21 (right) after treatment with either MZ or CP. Nuclei are stained with DAPI (blue). Scale bars, 100 μm. (C) Presence of CD163, TGF-β1, and TAGLN in
endometriotic lesions of recipients at day 1 (top left) and day 21 (bottom left) after treatment. Scale bars, 1 mm (H.E.) and 50 μm (CD163, TGF-β1, and TAGLN).
Numbers of infiltrated macrophages and IHC scores of TGF-β1 and TAGLN are shown in boxplots (right, n= 6 in each group). Middle horizontal line inside the box
indicates the median value. Bottom and top of the box indicate the 25th and 75th percentiles, respectively. Ends of the whiskers indicate the minimum and
maximum of all data, respectively. (Dand E) Numbers (left) and weights (right) of endometriotic lesions in recipients treated with each antibiotic (n= 6 in each
group) at day 1 (D) and day 21 (E). Error bars indicate SD (D and E). *P< 0.05 and **P< 0.01. Data were analyzed by one-way ANOVA.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 10 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

(fig. S12A). After a week of antibiotic treatment, F. nucleatum infil-
tration was no longer present, and M2 macrophage infiltration,
TGF-β1 expression, and TAGLN expression were all decreased in
both the endometria of donor mice and endometriotic lesions in
recipient mice (fig. S12, B to D). Consistently, recipient mice,
which received endometrium from F. nucleatum–infected donor
mice treated with either MZ or CP antibiotics, developed signifi-
cantly fewer endometriotic lesions weighing less than in the
control group without antibiotics (P< 0.01; fig. S12E).
We further explored the effects of antibiotic treatments 3 weeks
after the inoculation of endometrium, when the endometriotic
lesions had already developed (Fig. 7, A and B). MZ or CP was ad-
ministered orally for 5 days to mice with endometriotic lesions es-
tablished from F. nucleatum–infected uteri. Antibiotic treatment
decreased the amount of F. nucleatum, M2 macrophage infiltration,
TGF-β1 abundance, and TAGLN protein expression in the endo-
metriotic lesions 21 days after initiation of treatment, but not
after only 1 day (Fig. 7, B and C, and fig. S13, A and B). Antibiotic
treatment resulted in a significant reduction in the weights of endo-
metriotic lesions after 21 days of treatment (P> 0.05 and P< 0.01,
respectively; Fig. 7, D and E). Together, these data suggest that MZ
or CP antibiotic treatment may facilitate treatment of
endometriosis.
DISCUSSION
In this study, we further the understanding of the pathogenesis of
endometriosis by showing that Fusobacterium infection of the en-
dometrium may contribute to disease. Increased numbers of myo-
fibroblasts with high TAGLN expression were found in the
endometrial microenvironment where Fusobacterium was present.
The reason why Fusobacterium preferentially infects the endometria
of some patients is uncertain. Some work has shown intrauterine
infection of F. nucleatum due to hematogenous transmission of
the bacteria from the oral cavity. Such transmission mostly occurs
during pregnancy, when blood flow to the placenta is generally in-
creased (42). Because patients with endometriosis are often nonpar-
ous, transmission through the vagina may also be considered. The
presence of Fusobacterium in the vaginal swab samples from pa-
tients with endometriosis was significantly greater than from pa-
tients without endometriosis (P= 0.023; table S8), supporting the
possibility of a vaginal transmission route. Menstrual cycle–associ-
ated changes in mucin conformation and immune cell function,
both barriers against bacterial infection, have been demonstrated
in the female reproductive tract, which is associated with the indi-
vidual’s susceptibility to infections (43,44). F. nucleatum appears to
damage the intestinal barrier of tumors and induce aberrant inflam-
mation (45–47). These pathogenic roles of Fusobacterium may be
due to its strong adhesion to epithelial tissues and its invasive abil-
ities (48,49).
We found that F. nucleatum, even when heat-killed, effectively
stimulated the production of TGF-β1 from M2 macrophages and
activated TGF-β signaling in vitro. A recent study showed that
Gram-negative bacteria, including F. nucleatum, induce innate
immune responses via recognition by Toll-like receptor 4 (TLR4)
of their cell walls, which are composed of lipopolysaccharide (41).
This enhances M2 macrophage polarization within the tumor
immune microenvironment (50). Thus, Fusobacterium infection
appears to create a TGF-β1 signaling–enriched environment in
the endometrium.
TAGLN is also expressed in certain other fibroblasts, such as
cancer-associated fibroblasts (CAFs), which have a role in creating
extracellular matrix structures and in metabolic and immune repro-
gramming of the tumor microenvironment (6). Comprehensive
scRNA-seq analysis revealed that one subtype of CAF in the micro-
environment of colorectal cancer expressed cytoskeletal genes and
markers of activated myofibroblasts, such as ACTA2,TAGLN, and
PDGFA, whereas another subtype expressed genes related to extra-
cellular matrix remodeling (51). Fibrosis of the surrounding tissue is
a hallmark of peritoneal or ovarian endometriosis. Despite some
similarities between fibrosis-associated myofibroblasts and CAFs,
invasive and proliferative abilities appear to be less potent in endo-
metriosis than malignancy (6,52). Although functional differences
between these two types of fibroblasts are yet to be clearly defined at
the molecular basis (6), both fibrosis-associated myofibroblasts and
CAFs nevertheless gain enhanced proliferative properties and are
functionally diverse populations different from other subtypes of fi-
broblasts (6).
Two theories of the pathogenesis of ovarian endometriomas had
been considered: invagination of shed endometrial cells derived
from retrograde menstruation into the ovarian cortex and surface
epithelial transdifferentiation to endometrial-lined ovarian cysts
(coelomic metaplasia) (3). Our data are consistent with the
former. However, it is also plausible that coelomic metaplasia
may be caused by Fusobacterium infection. To date, the stimuli
causing transformation of coelomic epithelium into endometrial-
type glands remain unidentified (53). Fusobacterium may trigger
the transdifferentiation of ovarian surface epithelial cells. A recent
study showed that bacterial infection induced transdifferentiation of
epithelial cells during colon tumorigenesis (54). The possibility that
Fusobacterium is a trigger of metaplasia needs to be explored in
the future.
We investigated the roles of F. nucleatum and TAGLN in the ini-
tiation and development of endometriosis using a mouse model,
because F. nucleatum is not present in laboratory mice, and
TAGLN is well conserved between humans and mice (55).
Despite being the most frequently applied animal model valuable
for the study of endometriosis, mice do also have some limitations.
Endometriosis is a condition in which endometrial tissue aberrantly
grows ectopically. However, mice lack a menstrual cycle and do not
develop spontaneous endometriosis (56). Therefore, endometriotic
lesions in our mouse models have to be induced by peritoneal in-
jection of minced endometrial tissue, a limitation of which is that
this contains myometrial cells that are not present in refluxed men-
strual tissue. Using the mouse model that we used in the current
study, earlier work has revealed several important features of the
human disease, including the hormonal resistance of endometriosis
(57–59). Recently, Greaves et al. (60) developed a mouse model of
endometriosis that closely reflects the human condition and has re-
vealed that immune complement in the menstrual endometrium
contributes to the appearance of pathology. In this model, inoculat-
ed endometrium is removed from the myometrial layer. Further val-
idation of these findings using this model will need to be conducted
in the future.
Treatment options for endometriosis are currently based on hor-
monal therapy, such as long-term ovulation suppression (52).
However, creating a relatively hypoestrogenic environment can
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 11 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

lead to adverse effects, and women cannot become pregnant during
treatment (5). Surgical treatment is generally necessary for intracta-
ble pelvic pain. However, a high recurrence rate is a major concern
after surgical removal of endometriotic lesions (52). Our data reveal
a potential pathogenic mechanism of endometriosis involving Fu-
sobacterium infection, and eradication with MZ or CP could repre-
sent an option to improve treatment of this disease. Consistent with
our findings, a recent empirical study showed that MZ treatment
reduced the growth of endometriotic lesions in a mouse model, al-
though the potential involvement of pathogenic bacteria and the
underlying mechanisms were not elucidated (61). Thus, combining
antibiotics with other therapeutics, such as hormone treatments,
might be another approach and will hopefully be examined in
future clinical trials.
We found that 64% of patients with endometriosis had Fusobac-
terium in their endometria, supporting the idea that endometriosis
is a multifactorial disease and that its pathogenesis is difficult to
ascribe to a single factor. In addition, bacterial infections other
than Fusobacterium might also be involved in causing endometri-
osis. Recent studies have shown that various different bacteria may
be present in the urine, vagina, or uterus from patients with endo-
metriosis, although F. nucleatum has not been identified as a causal
bacterium for the disease (62–65). Because the amount of bacteria
in the uterus is quite low (17), differences in the methodology used
in these studies may explain the divergent results regarding the mi-
crobiota repertoire. Nevertheless, we documented higher frequency
of Fusobacterium in both the endometrium and ovarian endometri-
otic tissues and documented pathogenic effects in endometrium in-
fected with pure cultured F. nucleatum in an in vivo mouse model.
Furthermore, antibiotic treatment effective against F. nucleatum
reduced lesion weight. Therefore, Fusobacterium may have a path-
ogenic role in endometriosis rather than simply flourishing in the
environment of the endometrium in reproductive-age women with
endometriosis.
This study has some limitations that need to be considered when
interpreting the results. First, as already mentioned, the study lacks
direct evidence supporting the hypothesis that the presence of Fu-
sobacterium in the endometrium promotes endometriosis after ret-
rograde menstruation. Considering the many studies documenting
mechanisms responsible for endometriosis (1–5), further precise
studies are needed. Second, clinical studies are required to ascertain
whether treatment with antibiotics against Fusobacterium repre-
sents a bona fide effective treatment for patients with endometriosis.
In conclusion, we presented evidence that myofibroblasts ex-
pressing TAGLN promote endometrial cell survival at ectopic
sites. These cells are induced by TGF-β signaling, which can be ac-
tivated by Fusobacterium infection. Although further studies are
needed, our data suggest that targeting Fusobacterium in the endo-
metrium by antibiotic treatment may be a therapeutic option for
patients with endometriosis.
MATERIALS AND METHODS
Study design
We designed a series of studies to identify mechanisms involved in
the development of endometriosis caused by the phenotypic transi-
tion of endometrial fibroblasts induced by TGF-β and to test the
hypothesis that endometrial Fusobacterium infection promotes
these phenotypic transitions. We first designed an observational
study in humans to identify genes specifically expressed by endome-
trial fibroblasts in patients with endometriosis. We next designed a
functional study of human endometrial fibroblasts strongly express-
ing TAGLN. We then proceeded to investigate upstream targets to
determine whether the phenotypic transition was caused by TGF-β
and systemic and local inflammatory responses after Fusobacterium
infection. We finally designed in vivo analyses to determine whether
Fusobacterium infection promotes endometriosis development in
mice and to explore the effectiveness of antibiotics as a nonhormon-
al treatment for endometriosis. Mice were randomized to their re-
spective treatment groups. Pathological quantification, microarray,
and scRNA-seq analyses were performed by blinded investigators.
Analysis of experimental data from the mice was not blinded. All
experiments were performed in triplicate. Although formal statisti-
cal methods were not used to predetermine sample size, sample
sizes were chosen on the basis of estimates from pilot experiments
and previously published results.
Animal experiments were performed using protocols approved
by the Institutional Animal Care and Use Committee of Nagoya
University Graduate School of Medicine (no. 20165). All samples
were collected with written informed consent after approval by
the Institutional Review Board of the Nagoya University Graduate
School of Medicine (nos. 2014-0134, 2017-0497, 2017-0503, and
2021-0178).
Human tissue samples
Samples from all patients were obtained from Nagoya University
Hospital and Toyota Kosei Hospital in Japan. Seventy-six patients
without endometriosis and 79 patients diagnosed with endometri-
osis were enrolled in this study. Paired primary endometrial and en-
dometriosis samples were taken from the 79 patients who
underwent surgical removal of the uterus together with the endo-
metriotic lesions. Because the uterus is retained during the surgical
treatment in younger women for possible future pregnancies, the
median age of the patients studied here was >40 years. In the
control group without endometriosis, patients had undergone
surgery for leiomyoma, adenomyosis, cervical dysplasia, or cervical
cancer and had no history of endometriosis. Detailed patients’in-
formation is provided in table S1. The vaginal samples were collect-
ed using cotton swab (DNA/RNA Shield Collection Tube With
Swab, ZYMO RESEARCH) to absorb vaginal secretions. All pa-
tients had regular menstrual periods and had not received any hor-
monal treatment for at least 3 months before surgery or at the time
of collecting the samples. No patients had received treatment with
antibiotics or anti-inflammatory drugs for 1 month before surgery
or at the time of collecting the samples. Primary fibroblasts were
collected from each sample after shaking the sample for 1 hour at
37°C in Dulbecco’s modified Eagle’s medium (DMEM)/F12-con-
taining collagenase (200 U/ml; Thermo Fisher Scientific) and
straining through a 40-μm nylon mesh as was reported previously
(21). Primary human mesothelial cells were obtained from the
tumor-free omenta of female patients with serous adenocarcinoma
without genetic modification (66).
Cell lines
The UTnon-FB lines (MC1 and MC2) have been established previ-
ously (67). The OVend-FB lines (SC8 and SC10) were established in
the current study using the same method as used for MC1 and MC2
(67). Briefly, all four cell lines were immortalized by transfection
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 12 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

with telomerase reverse transcriptase (TERT), cyclin D1, and
mutant CDK4 genes. Clinical data of the patients from which the
four cell lines were derived are shown in table S9. Immortalized fi-
broblasts were maintained in DMEM/F12 supplemented with 10%
fetal bovine serum (FBS; Thermo Fisher Scientific) and 1% antibi-
otic-antimycotic solution that includes penicillin, streptomycin,
and amphotericin (Anti-anti, Thermo Fisher Scientific). The
human monocyte cell line THP-1 was purchased from the JCRB
Cell Bank (JCRB0112.1, human acute monocytic leukemia cell
line, Osaka, Japan) and maintained in RPMI 1640 medium supple-
mented with 10% FBS and 1% Anti-anti. The cell line was authen-
ticated through short tandem repeat profiling by the JCRB Cell
Bank and was mycoplasma-free. All cells were cultured at 37°C in
a humidified incubator with 5% CO
2
.
Plasmid construction
A human TAGLN vector, pcDNA3.4-TAGLN, and an N-terminal
Flag-tagged human TAGLN vector, pcDNA3.4-Flag-TAGLN-FL,
were constructed by inserting a full-length human TAGLN cDNA
fragment from pCMV-SPORT6.1-hTAGLN (IRAK141P09,
RIKEN, Japan) into pcDNA3.4. A partial deletion construct of
TAGLN, pcDNA3.4-Flag-TAGLN-ΔCLIK, was created by inverted
PCR using PrimeSTAR Max DNA Polymerase (TAKARA) and li-
gation of the PCR products. pCAG-mTAGLN was constructed by
inserting a full-length mouse Tagln cDNA into a pCAG-HIVgp
vector using In-Fusion HD Cloning Kits (TAKARA). The pCAG–
green fluorescent protein (GFP) vector was used as a control. The
primer sequences for the inverted PCR and mouse TAGLN are
shown in table S10. All constructs were verified by Sanger
sequencing.
Cell manipulation
Transfections with the plasmids and siRNAs were carried out using
Lipofectamine 3000 (Thermo Fisher Scientific) or ScreenFect A
Plus (FUJIFILM Wako Pure Chemical) according to the manufac-
turer’s protocol. We used 50 nM siRNA targeting TAGLN (#1,
si14039 and #2, si14040; Thermo Fisher Scientific) or negative
control siRNA (Silencer Select Negative Control siRNA, 4390844,
Thermo Fisher Scientific). Cells stably overexpressing TAGLN
were generated by transfecting MC1 and MC2 with pcDNA3.4-
TAGLN, followed by geneticin selection (100 μg/ml; Thermo
Fisher Scientific).
Bacterial strains and culture
F. nucleatum (JNBP_02614), L. iners (GAI_11032), and E. coli K-12
(GTC_2003) were obtained from Gifu University Center for Con-
servation of Microbial Genetic Resource, Organization for Research
and Community Development, Japan (https://pathogenic-bacteria.
nbrp.jp/bacteria/bacteriaAllItemsList.jsp). F. nucleatum and L. iners
were first cultured under anaerobic conditions at 37°C in Brucella
agar (Kyokutoseiyaku). We then used liquid cultures based on brain
heart infusion (Nissui Pharmaceutical Co) broth medium, supple-
mented with hemin (10 μg/ml, Thermo Fisher Scientific), menadi-
one (5 μg/ml, Nacalai Tesque), L-cysteine (1 μg/ml, Sigma-Aldrich),
and resazurin (2 μg/ml, FUJIFILM Wako Pure Chemical) as an an-
aerobic indicator. F. nucleatum colonies were transferred to the
liquid culture medium and maintained at 37°C under anaerobic
conditions with constant stirring on a magnetic stirrer. The bacte-
rial cell concentrations were measured by assessing optical density
at 660 nm (Miniphoto518R, TAITEC) or McFarland standard 0.5
(BioMérieux). E. coli K-12 was cultured under aerobic conditions
at 37°C in L-Broth agar (MP Biomedicals).
Animal experiments
We used an endometriosis model in which endometrial tissues from
estrus-stage donor mice are intraperitoneally inoculated into the re-
cipient mice (40). Endometriosis induction surgery was performed
using 6-week-old BALB/c female mice (Japan SLC, Shizuoka) given
17β-estradiol (100 μg/kg per mouse per week; Fuji Pharma) subcu-
taneously after ovariectomy of donor mice. Endometrial tissues
were collected from one donor, minced using fine scissors, and in-
jected intraperitoneally by syringe into two recipients, followed by
ovariectomy of recipient mice under anesthesia and pain control
and subcutaneous injection with 17β-estradiol (100 μg/kg) once a
week. Four weeks after the intraperitoneal inoculation, endometri-
otic lesions of the recipient mouse were detected as a cyst composed
of ERα- and ERβ-positive stromal cells (fig. S10A) (57). Endometri-
otic lesions, which were located on the superficial layers of the peri-
toneum, mesentery, and near the ovaries, were counted and
measured by a micro-weighing instrument (ENTRIS64-1S, Sartor-
ius) after removal from the abdominal cavity. All mice were main-
tained in a specific pathogen–free biosafety level 2 biohazard
facility. Experimental mice were cohoused and exposed to a 12-
hour light/12-hour dark cycle with unrestricted access to water
and food. The ambient temperature was restricted around 25°C,
and the room humidity ranged from 40 to 70%.
Animal experiment 1
Four different amounts of endometrial tissues—
1
/
16
(1 mg),
1
/
8
(2 mg),
1
/
4
(4 mg), and
1
/
2
uterus (8 mg)—were titrated to determine the
appropriate amount to use for the endometriosis formation model.
For each amount, five mice were included in both the F. nuclea-
tum–uninfected and F. nucleatum–infected groups. Because we in-
tended to examine and compare differences in the established
endometriotic lesions inoculated from the donor mice with and
without Fusobacterium infection, a minimally sufficient amount of
endometrial tissues was used in most of our experimental condition
(almost 8 mg).
Animal experiment 2
Transvaginal bacterial infections of donor mouse endometrium
were performed four times over 1 week beginning at 6 weeks of
age. A sham group was treated with phosphate-buffered saline
(PBS), and for infection, F. nucleatum,L. iners, or E. coli K-12 at
10
7
colony-forming units (CFUs)/10 μl of PBS per day were
applied intravaginally. A schema of the mouse endometriosis
model with transvaginal infection is shown in Fig. 5A (n= 6 each
for four experimental groups: CTRL, L. ine.,E. coli K-12, and F.
nuc.). We administered a 2- or 8-mg amount of endometrial
tissues for each experiment.
Animal experiment 3
For hematogenous administration, 10
7
CFUs of F. nucleatum in 10
μl of PBS were injected into the jugular veins of mice using a 34-
gauge microneedle. Bacterial injections into donor mice were per-
formed four times over 1 week. One week after infection, 2 mg of the
donor mouse endometrial tissue were inoculated into the recipient
mice. Before inoculation, a part of the donor tissue was examined
for the presence of F. nucleatum by FISH analysis. No mouse expe-
rienced a deterioration in health after the hematogenous adminis-
tration of F. nucleatum.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 13 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Animal experiment 4
Antibiotics for treating the endometria of donor mice, containing
MZ (17 μg per mouse per day; FUJIFILM Wako Pure Chemical) or
CP (7 μg per mouse per day; Sigma-Aldrich), were introduced
transvaginally every day for 5 days after infection with F. nucleatum.
Endometrial tissues were intraperitoneally injected into the recipi-
ent mouse 1 week after antibiotic treatments. A schema for the
mouse transvaginal antibiotic treatment model is shown in fig.
S12A (n= 6 each for the three experimental groups: PBS, MZ,
and CP). To treat endometriosis, 4 weeks after endometriosis induc-
tion surgery and F. nucleatum infection, mice were treated with an-
tibiotics containing MZ (17 μg per mouse per day) or CP (7 μg per
mouse per day) for 5 days by oral gavage. After antibiotic treat-
ments, endometriotic lesions were harvested and evaluated for
numbers and total weight. A schema for the mouse oral antibiotic
treatment model is shown in Fig. 7A (n= 6 each for the three ex-
perimental groups: PBS, MZ, and CP).
Animal experiment 5
Electroporation was used to introduce pCAG-mTAGLN (n= 4
mice) or pCAG-GFP (n= 4 mice) control vectors and the overex-
pressed mouse-Tagln gene into the minced donor mouse uterus
tissues without Fusobacterium infection. For electroporation, after
five rectangular pulses (40 V, 5-ms duration with an interval of 50
ms) for 5 s, the direction of the electrical field was reversed, and
three pulses were applied with identical settings (CUY700P2L and
CUY701P5E, Nepa gene) (68). Electroporation was also used to
perform Tagln knockdown experiments in the minced uteri from
Fusobacterium-infected donor mice under the same conditions.
siRNAs against mouse-Tagln (#1: SASI_Mm01_00135382, n= 6
mice and #2: SASI_Mm01_00135383, n= 6 mice; Sigma-Aldrich)
or negative control siRNA (Silencer Select Negative Control
siRNA, 4390844, Thermo Fisher Scientific; n= 6 mice) were used
here. The efficiency of overexpression and depletion was validated
after 24 hours by qPCR and Western blotting (fig. S11).
ChIP assay
Chromatin immunoprecipitation (ChIP) assays were performed ac-
cording to previously published methods (69). UTnon-FB cell lines
(MC1 and MC2) were treated with either TGF-β1 or SB431542 for
48 hours. Ten percent of each lysate was used as an input control.
The putative enhancer and promoter regions were identified by the
enrichment of H3K27ac and SMAD2/3 in fibroblasts by reference to
a public database (SCREEN database, https://screen.encodeproject.
org/; fig. S7B) (70). Primer sets for ChIP-qPCR are shown in table
S10. We used anti-H3K27ac (39133, Active Motif) and anti-
SMAD2/3 (#8685, Cell Signaling Technology) antibodies for
target lesions and anti-IgG (immunoglobulin G) (PM035, MBL In-
ternational) antibody as the negative control.
Immunouorescence
Cell cultures were fixed with 2% paraformaldehyde for 10 min.
After incubating with 0.1% Triton X-100 on ice for 2 min, the
cells were blocked for 15 min using 2% gelatin. Samples were
washed with PBS/glycine and incubated with primary antibody
against TAGLN (ab14106, 1:100, Abcam, Cambridge, UK), αSMA
(M0851, 1:100, Agilent Technologies, Santa Clara, CA), IL-6
(MAB2062, 1:25, R&D Systems, Minneapolis, MN or ab214429,
1:500, Abcam), or FLAG (F1804, 1:100, Sigma-Aldrich) or with
control serum for 1 hour at room temperature. Cells were then
incubated with 4′,6-diamidino-2-phenylindole (DAPI; #4083,
1:1000, Cell Signaling Technology, Danvers, MA) and the combina-
tion of Alexa Fluor 488–labeled anti-mouse secondary antibody (A-
11029, 1:500, Thermo Fisher Scientific) and Alexa Fluor 546–
labeled anti-rabbit secondary antibody (A-11010, 1:500, Thermo
Fisher Scientific) or the combination of Alexa Fluor 488–labeled
anti-rabbit secondary antibody (A-11070, 1:500, Thermo Fisher
Scientific) and Alexa Fluor 546–labeled anti-mouse secondary an-
tibody (A-11018, 1:500, Thermo Fisher Scientific) for 45 min at
room temperature. Coimmunostaining of tissue samples with
anti-TAGLN and anti–IL-6 antibody, with anti-CD163
(ab182422, 1:100, Abcam) and anti–TGF-β1 (ab92486, 1:100,
Abcam), or with anti-TAGLN and anti–TGF-β1 was performed
using the Opal Multi-plex IHC Detection Kit (PerkinElmer,
Waltham, MS) following the manufacturer’s protocol. For the im-
munofluorescence negative control, we stained without primary an-
tibody as a negative control. Images were acquired using a Leica
DMI6000B. The integrated density of the fluorescent signal of
cells was measured using ImageJ software using region of interest
polygon selection (71). The IL-6 and TAGLN fluorescence score
was calculated using the following equation: score = Σ Pi (i),
where i= intensity of staining (1, 2, or 3 as weak, moderate, or
strong, respectively), and Pi is the percentage of stained cells for
each intensity. We calculated the average fluorescence scores of at
least five different fields of one section.
Immunohistochemistry
Staining was performed on 5-μm sections of mouse and human
tissues. After deparaffinization, antigen retrieval was performed in
a microwave oven in 10 mM sodium citrate buffer (pH 6.0) or tris/
EDTA buffer (pH 9.0). After blocking in 3% H
2
O
2
for 20 min, spec-
imens were incubated with the primary antibodies as follows:
TAGLN (ab14106, 1:200, Abcam), CD68 (M0876, 1:100, Agilent
Technologies), CD163 (ab182422, 1:100, Abcam), TGF-β1
(ab92486, 1:100, Abcam), TGF-βR1 (ab31013, 1:200, Abcam),
ERα (ab32063, 1:200, Abcam), ERβ (#PA1-310B, 1:50, Thermo
Fisher Scientific), and CD10 (sc9149, 1:100, Santa Cruz Biotechnol-
ogy). After subsequent incubation with horseradish peroxidase–
conjugated secondary antibody, specimens were subjected to diami-
nobenzine (DAB) (Agilent Technologies) and hematoxylin stain-
ing. The TAGLN IHC score was calculated using the following
equation: IHC score = Σ Pi (i), where i= intensity of staining (pos-
itive equal to muscle cells: 0 or 1) and Pi is the percentage of stained
cells. The TGF-β1 and TGF-βR1 IHC score was calculated using the
following equation: IHC score = Σ Pi (i), where i= intensity of stain-
ing (1, 2, or 3 as weak, moderate, or strong, respectively) and Pi is
the percentage of stained cells for each intensity. We calculated the
total TAGLN IHC scores or average TGF-β1 or TGF-βR1 IHC
scores of at least five different fields of one section using an
Olympus VS120 microscope at ×200 magnification.
Cytokine analysis
UTnon-FB cell lines (MC1 and MC2) were transfected with
pcDNA3.4-TAGLN to overexpress TAGLN for 2 days. OVend-FBs
(SC8 and SC10) were treated for 2 days with 50 nM siRNA targeting
TAGLN (#2; si14040) or negative control siRNA to knock down
TAGLN. Cytokines secreted by the cells were detected in condi-
tioned medium using RayBiotech Human Neuro Discovery Array
C2 according to the manufacturer’s instructions (RayBiotech
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 14 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Inc.). The signal intensity of each spot, which represents the secret-
ed cytokines, was evaluated by subtracting the background and nor-
malized to positive controls using ImageJ software.
Fluorescence in situ hybridization
FISH was performed using formalin-fixed paraffin-embedded en-
dometrial and endometriosis specimens. Sections were hybridized
with the 5′FAM–labeled universal bacterial probe EUB338 (30), the
5′Cy3–labeled Fusobacterium genus–specific probe FUSO (31), and
the 5′Cy5–labeled F. nucleatum–targeted probe FUSO664 (32–34).
The sequences of the FISH probes were obtained from probeBase
(http://probebase.csb.univie.ac.at/): pB-00159 for EUB338, pB-
00782 for FUSO, and pB-1346 for FUS664. A scrambled probe
(5′-CAATTGGGCCCGCTTTAACCCAATCTC-3′) served as the
nonspecific negative control. Slides were deparaffinized and
treated with 0.2 M HCl for 20 min and then hybridized overnight
with the indicated FISH probes at a concentration of 10 ng/μl at
56°C in hybridization buffer [0.9 M NaCl, 20 mM tris-HCl (pH
7.2), and 0.1% SDS]. Slides were washed for 20 min at 58°C in
wash buffer [0.9 M NaCl and 20 mM tris-HCl (pH 7.2)]. Tissue sec-
tions were counterstained with DAPI and mounted. Coimmunos-
taining with the anti-TAGLN antibody and FISH probe (FUSO
probe) was performed without antigen retrieval using the Opal
Multi-plex IHC Detection Kit (PerkinElmer) following the manu-
facturer’s protocol. Images were acquired using a Leica DMI6000B.
Bacterial spots were counted by ImageJ software using polygonal
region of interest selection. We calculated the total bacterial spot
numbers in at least five different fields of one section.
Microarray analysis
RNA microarray assays were performed as described previously
(72). The RNA was amplified into complementary RNA and
labeled according to the Agilent One-Color Microarray-Based
Gene Expression Analysis protocol (Agilent Technologies).
Labeled samples were purified using RNeasy Mini Kits (Qiagen)
and hybridized to SurePrint G3 Human Gene Expression 8x60K
v3 array slides (G4851C, Agilent Technologies) at 65°C with rota-
tion at 10 rpm for 17 hours. The arrays were scanned using an
Agilent Microarray Scanner (G2565BA, Agilent Technologies).
The scanned images were analyzed using Feature Extraction soft-
ware, version 12.0 (Agilent Technologies), with background correc-
tion. Data analysis was performed with GeneSpring GX, version
14.9 (Agilent Technologies).
Single-cell analysis
Samples for scRNA-seq were prepared using the same digestion
method as for the RNA microarray assays in this study. After remov-
ing the red blood cells, cell viability was quantified using the trypan
blue exclusion method. Cell concentration was adjusted for targeted
sequencing of 10,000 cells per sample using Chromium Single Cell
3′Reagent Kits v3.1 (10x Genomics) to prepare libraries. The librar-
ies were sequenced in the Center for Omics and Bioinformatics,
Graduate School of Frontier Sciences, University of Tokyo using
the Novaseq 6000 system (Illumina). The fastq files were mapped
to the reference genome provided by 10x Genomics (GRCh38).
The read count was quantified using the Cell Ranger version 6.0.0
count pipeline (10x Genomics) with default parameters. The fol-
lowing data analysis was performed with Python version 3.9.7 and
Scanpy version 1.8.2. For quality control, cells with greater than
9000 or fewer than 2000 detectable genes or whose mitochondrial
contribution exceeded 10% of transcripts were removed. After inte-
grating UTnon-EM and OVend data, the expression profile was
normalized to counts per 10,000 and then log-transformed. The
highly variable genes (HVGs) were calculated using in-built func-
tions (scanpy.pp.highly_variable_genes) with default parameters.
Principal components analysis (PCA) was performed on the expres-
sion profile of HVGs, and batch effects were corrected using
Harmony version 1.0. Dimension reduction was conducted with
Uniform Manifold Approximation and Projection (UMAP). We
performed Leiden clustering and manually annotated these clusters
on the basis of marker genes; for epithelial cells, EPCAM and
KRT18; for macrophages/monocytes, CD68 and MS4A7; for T
cells/NK cells, CD2,CD3D,CD3E,CD3G,TRDC,KLRC1,
FCGR3A, and CEACAM1; for MAST cells, KIT and TPSB2; for en-
dothelial cells, AQP1,MTCT1,CDH5, and PECAM1; for fibroblasts,
COL1A1,COL3A1, and COL1A2.
Bacterial selection methods
We downloaded the publicly available raw data of the endometrial
bacterial analysis of the target patients from the database [European
Nucleotide Archive under study numbers PRJEB16013 and
PRJEB21098; dataset from a previous study (17)] and reanalyzed
them ourselves. Each datum was referenced to VITCOMIC2
(http://vitcomic.org/) to identify the bacterial genera. The read
count of each bacterium was divided by the total bacterial read
count to calculate the percentage of each bacterium present. The
top five bacterial genera significantly present in UTend-EM com-
pared with those without endometriosis were selected (P< 0.01;
table S6).
Bacterial infection experiments in vitro
THP-1 monocytes were differentiated into macrophages (dTHP1)
by treatment with 10 nM phorbol 12-myristate 13-acetate (PMA)
(Sigma-Aldrich) for 48 hours (73). For bacterial and dTHP1 cocul-
ture experiments, dTHP1 cells were mixed with live or heat-killed F.
nucleatum or L. iners at a multiplicity of infection (MOI) of 10 bac-
teria:1 dTHP1 cell for 48 hours. Heat-killed bacteria were made by
heating at 60°C for 30 min. The culture medium was harvested and
used for the incubation with UTnon-FBs.
Statistical analysis
Data are presented as the mean ± SD or SEM. Microsoft Excel and R
were used to generate graphs and to perform statistical analyses. P
values are indicated as n.s., P> 0.05, *P< 0.05, and **P< 0.01. Data
were analyzed by either a two-tailed Student’sttest, Fisher’s exact
test, one-way analysis of variance (ANOVA), or Wilcoxon rank-sum
test. The statistical methods used in this study are described in each
figure legend. All experiments were performed in triplicate.
Supplementary Materials
This PDF le includes:
Supplementary Materials and Methods
Figs. S1 to S13
Tables S2 to S10
Other Supplementary Material for this
manuscript includes the following:
Table S1
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 15 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Data file S1
MDAR Reproducibility Checklist
View/request a protocol for this paper from Bio-protocol.
REFERENCES AND NOTES
1. L. C. Giudice, L. C. Kao, Endometriosis. Lancet 364, 1789–1799 (2004).
2. L. C. Giudice, Clinical practice. Endometriosis. N. Engl. J. Med. 362, 2389–2398 (2010).
3. S. E. Bulun, Endometriosis. N. Engl. J. Med. 360, 268–279 (2009).
4. P. Vercellini, P. Vigano, E. Somigliana, L. Fedele, Endometriosis: Pathogenesis and treat-
ment. Nat. Rev. Endocrinol. 10, 261–275 (2014).
5. F. L. Cousins, D. F. O., C. E. Gargett, Endometrial stem/progenitor cells and their role in the
pathogenesis of endometriosis. Best Pract. Res. Clin. Obstet. Gynaecol. 50, 27–38 (2018).
6. R. Kalluri, The biology and function of fibroblasts in cancer. Nat. Rev. Cancer 16,
582–598 (2016).
7. L. M. Coussens, Z. Werb, Inflammation and cancer. Nature 420, 860–867 (2002).
8. M. B. Buechler, R. N. Pradhan, A. T. Krishnamurty, C. Cox, A. K. Calviello, A. W. Wang, Y. A.
Yang, L. Tam, R. Caothien, M. Roose-Girma, Z. Modrusan, J. R. Arron, R. Bourgon, S. Muller,
S. J. Turley, Cross-tissue organization of the fibroblast lineage. Nature 593,
575–579 (2021).
9. L. Li, J. M. Miano, P. Cserjesi, E. N. Olson, SM22 alpha, a marker of adult smooth muscle, is
expressed in multiple myogenic lineages during embryogenesis. Circ. Res. 78,
188–195 (1996).
10. S. J. Assinder, J.-A. Stanton, P. D. Prasad, Transgelin: An actin-binding protein and tumour
suppressor. Int. J. Biochem. Cell Biol. 41, 482–486 (2009).
11. J. M. Carthy, TGFβsignaling and the control of myofibroblast differentiation: Implications
for chronic inflammatory disorders. J. Cell. Physiol. 233, 98–106 (2018).
12. E. Tarbit, I. Singh, J. N. Peart, R. B. Rose’Meyer, Biomarkers for the identification of cardiac
fibroblast and myofibroblast cells. Heart Fail. Rev. 24, 1–15 (2019).
13. M. S. Payne, J. P.Newnham, D. A. Doherty, L. L. Furfaro, N. L. Pendal, D. E. Loh, J. A. Keelan, A
specific bacterial DNA signature in the vagina of Australian women in midpregnancy
predicts high risk of spontaneous preterm birth (the Predict1000 study). Am. J. Obstet.
Gynecol. 224, 206.e1–206.e23 (2021).
14. K. Agarwal, L. S. Robinson, S. Aggarwal, L. R. Foster,A. Hernandez-Leyva, H. Lin, B. A. Tortelli,
V. P.O’Brien, L. Miller, A. L. Kau, H. Reno, N. M. Gilbert, W. G. Lewis, A. L. Lewis, Glycan cross-
feeding supports mutualism between Fusobacterium and the vaginal microbiota. PLOS
Biol. 18, e3000788 (2020).
15. C. A. Brennan, W. S. Garrett, Fusobacterium nucleatum—Symbiont, opportunist and on-
cobacterium. Nat. Rev. Microbiol. 17, 156–166 (2019).
16. K. N. Khan, A. Fujishita, M. Kitajima, K. Hiraki, M. Nakashima, H. Masuzaki, Intra-uterine
microbial colonization and occurrence of endometritis in women with endometriosis.
Hum. Reprod. 29, 2446–2456 (2014).
17. C. Chen, X. Song, W. Wei, H. Zhong, J. Dai, Z. Lan, F. Li, X. Yu, Q. Feng, Z. Wang, H. Xie, X.
Chen, C. Zeng, B. Wen, L. Zeng, H. Du, H. Tang,C. Xu, Y. Xia, H. Xia, H. Yang, J. Wang, J. Wang,
L. Madsen, S. Brix, K. Kristiansen, X. Xu, J. Li, R. Wu, H. Jia, The microbiota continuum along
the female reproductive tract and its relation to uterine-related diseases. Nat. Commun. 8,
875 (2017).
18. S. Crispi, M. T. Piccolo, A. D’Avino, A. Donizetti, R. Viceconte, M. Spyrou, R. A. Calogero, A.
Baldi, P. G. Signorile, Transcriptional profiling of endometriosis tissues identifies genes
related to organogenesis defects. J. Cell. Physiol. 228, 1927–1934 (2013).
19. A. V. Predeus, E. S. Vashukova, A. S. Glotov, M. M. Danilova, N. S. Osinovskaya, O. V. Maly-
sheva, N. Y. Shved, N. Ganbarli, M. I. Yarmolinskaya, T. E. Ivashchenko, V. S. Baranov, Next-
generation sequencing of matched ectopic and eutopic endometrium identifies novel
endometriosis-related genes. Russ. J. Genet. 54, 1358–1365 (2018).
20. A. Hever, R. B. Roth, P. Hevezi, M. E. Marin, J. A. Acosta, H. Acosta, J. Rojas, R. Herrera, D.
Grigoriadis, E. White, P. J. Conlon, R. A. Maki, A. Zlotnik, Human endometriosis is associated
with plasma cells and overexpression of B lymphocyte stimulator. Proc. Natl. Acad. Sci.
U.S.A. 104, 12451–12456 (2007).
21. J. Ma, L. Zhang, H. Zhan, Y. Mo, Z. Ren, A. Shao, J. Lin, Single-cell transcriptomic analysis of
endometriosis provides insights into fibroblast fates and immune cell heterogeneity. Cell
Biosci. 11, 125 (2021).
22. R. Liu, M. M. Hossain, X. Chen, J. P. Jin, Mechanoregulation of SM22α/Transgelin. Bio-
chemistry 56, 5526–5538 (2017).
23. R. Chakraborty, F.Z. Saddouk, A. C. Carrao, D. S. Krause, D. M. Greif, K. A. Martin, Promoters
to study vascular smooth muscle. Arterioscler. Thromb. Vasc. Biol. 39, 603–612 (2019).
24. J. Baum, H. S. Duffy, Fibroblasts and myofibroblasts: What are we talking about?
J. Cardiovasc. Pharmacol. 57, 376–379 (2011).
25. A. Farmawati, Y. Kitajima, T. Nedachi, M. Sato, M. Kanzaki, R. Nagatomi, Characterization of
contraction-induced IL-6 up-regulation using contractile C2C12 myotubes. Endocr. J. 60,
137–147 (2013).
26. V. J. Young, S. F. Ahmad, W. C. Duncan, A. W. Horne, The role of TGF-βin the pathophys-
iology of peritoneal endometriosis. Hum. Reprod. Update 23, 548–559 (2017).
27. C. O. A. Omwandho, L. Konrad, G. Halis, F. Oehmke, H.-R. Tinneberg, Role of TGF-betas in
normal human endometrium and endometriosis. Hum. Reprod. 25, 101–109 (2010).
28. S. Chen, M. Kulik, R. J. Lechleider, Smad proteins regulate transcriptional induction of the
SM22alpha gene by TGF-beta. Nucleic Acids Res. 31, 1302–1310 (2003).
29. G. J. Inman, F. J. Nicolas, J. F.Callahan, J. D. Harling, L. M. Gaster, A. D. Reith, N. J. Laping, C. S.
Hill, SB-431542 is a potent and specific inhibitor of transforming growth factor-beta
superfamily type I activin receptor-like kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol.
Pharmacol. 62, 65–74 (2002).
30. R. I. Amann, In situ identification of micro-organisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. Mol. Microb. Ecol. Manual 3, 1–15 (1995).
31. P. T. Sunde, I. Olsen, U. B. Gobel, D. Theegarten, S. Winter, G. J. Debelian, L. Tronstad, A.
Moter, Fluorescence in situ hybridization (FISH) for direct visualization of bacteria in
periapical lesions of asymptomatic root-filled teeth. Microbiology (Reading) 149,
1095–1102 (2003).
32. T. Thurnheer, R. Gmur, B. Guggenheim, Multiplex FISH analysis of a six-species bacterial
biofilm. J. Microbiol. Methods 56, 37–47 (2004).
33. L. Karygianni, M. Follo, E. Hellwig, D. Burghardt, M. Wolkewitz, A. Anderson, A. Al-Ahmad,
Microscope-based imaging platform for large-scale analysis of oral biofilms. Appl. Environ.
Microbiol. 78, 8703–8711 (2012).
34. K. Bao, N. Bostanci, T. Thurnheer, J. Grossmann, W. E. Wolski, B. Thay, G. N. Belibasakis, J.
Oscarsson, Aggregatibacter actinomycetemcomitans H-NS promotes biofilm formation
and alters protein dynamics of other species within a polymicrobial oral biofilm. NPJ
Biolms Microbiomes 4, 12 (2018).
35. D. Grenier, L. Grignon, Response of human macrophage-like cells to stimulation by Fu-
sobacterium nucleatum ssp. nucleatum lipopolysaccharide. Oral Microbiol. Immunol. 21,
190–196 (2006).
36. J. Vallve-Juanico, S. Houshdaran, L. C. Giudice, The endometrial immune environment of
women with endometriosis. Hum. Reprod. Update 25, 564–591 (2019).
37. M. Benoit, B. Desnues, J.-L. Mege, Macrophage polarization in bacterial infections.
J. Immunol. 181, 3733–3739 (2008).
38. M. I. Petrova, G. Reid,M. Vaneechoutte, S. Lebeer, Lactobacillus iners: Friend or foe? Trends
Microbiol. 25, 182–191 (2017).
39. J. Ravel, P. Gajer, Z. Abdo, G. M. Schneider, S. S. K. Koenig, S. L. McCulle, S. Karlebach, R.
Gorle, J. Russell, C. O. Tacket, R. M. Brotman, C. C. Davis, K. Ault, L. Peralta, L. J. Forney,
Vaginal microbiome of reproductive-age women. Proc. Natl. Acad. Sci. U.S.A. 108(suppl. 1),
4680–4687 (2011).
40. T. Kato,K. Yasuda, K. Matsushita, K. J. Ishii, S. Hirota, T. Yoshimoto, H. Shibahara, Interleukin-
1/-33 signaling pathways as therapeutic targets for endometriosis. Front. Immunol. 10,
2021 (2019).
41. H. Liu, R. W. Redline, Y. W. Han, Fusobacterium nucleatum induces fetal death in mice via
stimulation of TLR4-mediated placental inflammatory response. J. Immunol. 179,
2501–2508 (2007).
42. E. L. V. Haar, J. So, C. Gyamfi-Bannerman, Y. W. Han, Fusobacterium nucleatum and adverse
pregnancy outcomes: Epidemiological and mechanistic evidence. Anaerobe 50,
55–59 (2018).
43. J. V. Fahey, C. R. Wira, Effect of menstrual status on antibacterial activity and secretory
leukocyte protease inhibitor production by human uterine epithelial cells in culture.
J. Infect. Dis. 185, 1606–1613 (2002).
44. C. R. Wira, M. Rodriguez-Garcia,M. V. Patel, The role of sex hormones in immune protection
of the female reproductive tract. Nat. Rev. Immunol. 15, 217–230 (2015).
45. A. D. Kostic, E. Chun, L. Robertson, J. N. Glickman, C. A. Gallini, M. Michaud, T.E. Clancy, D. C.
Chung, P. Lochhead, G. L. Hold, E. M. El-Omar, D. Brenner, C. S. Fuchs, M. Meyerson, W. S.
Garrett, Fusobacterium nucleatum potentiates intestinal tumorigenesis and modulates
the tumor-immune microenvironment. Cell Host Microbe 14, 207–215 (2013).
46. L. Liu, L. Liang, H. Liang, M. Wang, B. Lu, M. Xue, J. Deng, Y.Chen, Fusobacterium nucleatum
aggravates the progression of colitis by regulating M1 macrophage polarization via AKT2
pathway. Front. Immunol. 10, 1324 (2019).
47. L. Ai, Y. Ren, M. Zhu, S. Lu, Y. Qian, Z. Chen, A. Xu, Synbindin restrains proinflammatory
macrophage activation against microbiota and mucosal inflammation during colitis. Gut
70, 2261–2272 (2021).
48. Y. W. Han, W. Shi, G. T. Huang, S. K. Haake, N. H. Park, H. Kuramitsu, R.J. Genco, Interactions
between periodontal bacteria and human oral epithelial cells: Fusobacterium nucleatum
adheres to and invades epithelial cells. Infect. Immun. 68, 3140–3146 (2000).
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 16 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

49. J. Strauss, G. G. Kaplan, P. L. Beck, K. Rioux, R. Panaccione, R. Devinney, T. Lynch, E. Allen-
Vercoe, Invasive potential of gut mucosa-derived Fusobacterium nucleatum positively
correlates with IBD status of the host. Inamm. Bowel Dis. 17, 1971–1978 (2011).
50. T. Chen, Q. Li, J. Wu, Y. Wu, W. Peng, H. Li, J. Wang, X. Tang, Y. Peng, X. Fu, Fusobacterium
nucleatum promotes M2 polarization of macrophages in the microenvironment of colo-
rectal tumours via a TLR4-dependent mechanism. Cancer Immunol. Immunother. 67,
1635–1646 (2018).
51. H. Li, E. T. Courtois,D. Sengupta, Y. Tan, K. H. Chen, J. J. L. Goh, S. L. Kong, C. Chua, L. K. Hon,
W. S. Tan, M. Wong, P. J. Choi, L. J. K. Wee, A. M. Hillmer, I. B. Tan, P. Robson, S. Prabhakar,
Reference component analysis of single-cell transcriptomes elucidates cellular heteroge-
neity in human colorectal tumors. Nat. Genet. 49, 708–718 (2017).
52. S. E. Bulun, B. D. Yilmaz, C. Sison, K. Miyazaki, L. Bernardi, S. Liu, A. Kohlmeier, P. Yin, M.
Milad, J. Wei, Endometriosis. Endocr. Rev. 40, 1048–1079 (2019).
53. T. Kurita, Normal and abnormal epithelial differentiation in the female reproductive tract.
Dierentiation 82, 117–126 (2011).
54. C. Kong, X. Yan, Y.Zhu, H. Zhu, Y. Luo, P. Liu, S. Ferrandon, M. F.Kalady, R. Gao, J. He, F. Yin, X.
Qu, J. Zheng, Y. Gao, Q. Wei, Y. Ma, J. Y.Liu, H. Qin, Fusobacterium nucleatum promotes the
development of colorectal cancer by activating a cytochrome P450/epoxyoctadecenoic
acid axis via TLR4/Keap1/NRF2 signaling. Cancer Res. 81, 4485–4498 (2021).
55. D. A. Morrison, S. A. Lacks, W. R. Guild, J. M. Hageman, Isolation and characterization of
three new classes of transformation-deficient mutants of Streptococcus pneumoniae that
are defective in DNA transport and genetic recombination. J. Bacteriol. 156,
281–290 (1983).
56. A. D. Greene, S. A. Lang, J. A. Kendziorski, J. M. Sroga-Rios, T. J. Herzog, K. A. Burns, En-
dometriosis: Where are we and where are we going? Reproduction 152, R63–R78 (2016).
57. S. J. Han, S. Y. Jung, S.-P. Wu, S. M. Hawkins, M. J. Park, S. Kyo, J. Qin, J. P. Lydon, S. Y. Tsai, M.-
J. Tsai, F. J. DeMayo, B. W. O’Malley, Estrogen receptor βmodulates apoptosis complexes
and the inflammasome to drive the pathogenesis of endometriosis. Cell 163,
960–974 (2015).
58. S. J. Han, S. M. Hawkins, K. Begum, S. Y.Jung, E. Kovanci, J. Qin, J. P. Lydon, F. J. DeMayo,B. W.
O’Malley, A new isoform of steroid receptor coactivator-1 is crucial for pathogenic pro-
gression of endometriosis. Nat. Med. 18, 1102–1111 (2012).
59. Y. Zhao, P. Gong, Y. Chen, J. C. Nwachukwu, S. Srinivasan, C. Ko, M. K. Bagchi, R. N. Taylor,
K. S. Korach, K. W. Nettles, J. A. Katzenellenbogen, B. S. Katzenellenbogen, Dual suppres-
sion of estrogenic and inflammatory activities for targeting of endometriosis. Sci. Transl.
Med. 7, 271ra9 (2015).
60. E. Greaves, F. L. Cousins, A. Murray, A. Esnal-Zufiaurre, A. Fassbender, A. W. Horne, P. T. K.
Saunders, A novel mouse model of endometriosis mimics human phenotype and reveals
insights into the inflammatory contribution of shed endometrium. Am. J. Pathol. 184,
1930–1939 (2014).
61. S. B. Chadchan, M. Cheng, L. A. Parnell, Y. Yin, A. Schriefer, I. U. Mysorekar, R. Kommagani,
Antibiotic therapy with metronidazole reduces endometriosis disease progression in mice:
A potential role for gut microbiota. Hum. Reprod. 34, 1106–1116 (2019).
62. J. M. Wessels, M. A. Dominguez, N. A. Leyland, S. K. Agarwal, W. G. Foster, Endometrial
microbiota is more diverse in people with endometriosis than symptomatic controls. Sci.
Rep. 11, 18877 (2021).
63. K. Akiyama, K. Nishioka, K. N. Khan, Y. Tanaka, T. Mori, T. Nakaya, J. Kitawaki, Molecular
detection of microbial colonization in cervical mucus of women with and without endo-
metriosis. Am. J. Reprod. Immunol. 82, e13147 (2019).
64. N. Le, M. Cregger, V. Brown, J. Loret de Mola, P. Bremer, L. Nguyen, K. Groesch, T. Wilson, P.
Diaz-Sylvester, A. Braundmeier-Fleming, Association of microbial dynamics with urinary
estrogens and estrogen metabolites in patients with endometriosis. PLOS ONE 16,
e0261362 (2021).
65. K. N. Khan, A. Fujishita, H. Masumoto, H. Muto, M. Kitajima, H. Masuzaki, J. Kitawaki, Mo-
lecular detection of intrauterine microbial colonization in women with endometriosis. Eur.
J. Obstet. Gynecol. Reprod. Biol. 199, 69–75 (2016).
66. H. Kajiyama, F. Kikkawa, O. Maeda, T. Suzuki, K. Ino, S. Mizutani, Increased expression of
dipeptidyl peptidase IV in human mesothelial cells by malignant ascites from ovarian
carcinoma patients. Oncology 63, 158–165 (2002).
67. A. Muraoka, S. Osuka, T. Kiyono, M. Suzuki, A. Yokoi, T. Murase, K. Nishino, K. Niimi, T. Na-
kamura, M. Goto, H. Kajiyama, Y. Kondo, F. Kikkawa, Establishment and characterization of
cell lines from human endometrial epithelial and mesenchymal cells from patients with
endometriosis. F S Sci. 1, 195–205 (2020).
68. I. Kawabata, T. Umeda, K. Yamamoto, S. Okabe, Electroporation-mediated gene transfer
system applied to cultured CNS neurons. Neuroreport 15, 971–975 (2004).
69. H. Totani, K. Shinjo, M. Suzuki, K. Katsushima, S. Mase, A. Masaki, A. Ito, M. Ri, S. Kusumoto,
H. Komatsu, T. Ishida, H. Inagaki, S. Iida, Y. Kondo, Autocrine HGF/c-Met signaling pathway
confers aggressiveness in lymph node adult T-cell leukemia/lymphoma. Oncogene 39,
5782–5794 (2020).
70. ENCODE Project Consortium, J. E. Moore, M. J. Purcaro, H. E. Pratt, C. B. Epstein, N. Shoresh,
J. Adrian, T. Kawli, C. A. Davis, A. Dobin, R. Kaul, J. Halow, E. L. Van Nostrand, P. Freese, D. U.
Gorkin, Y. Shen, Y. He, M. Mackiewicz, F. Pauli-Behn, B. A. Williams, A. Mortazavi, C. A. Keller,
X.-O. Zhang, S. I. Elhajjajy, J. Huey, D. E. Dickel, V. Snetkova, X. Wei, X. Wang, J. C. Rivera-
Mulia, J. Rozowsky, J. Zhang, S. B. Chhetri, J. Zhang, A. Victorsen, K. P. White, A. Visel, G. W.
Yeo, C. B. Burge, E. Lecuyer, D. M. Gilbert, J. Dekker,J. Rinn, E. M. Mendenhall, J. R. Ecker, M.
Kellis, R. J. Klein, W. S. Noble, A. Kundaje, R. Guigo, P. J. Farnham, J. M. Cherry, R. M. Myers, B.
Ren, B. R. Graveley, M. B. Gerstein, L. A. Pennacchio, M. P. Snyder, B. E. Bernstein, B. Wold,
R. C. Hardison, T. R. Gingeras, J. A. Stamatoyannopoulos, Z. Weng, Expanded encyclo-
paedias of DNA elements in the human and mouse genomes. Nature 583, 699–710 (2020).
71. J. Schindelin, I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch, C.
Rueden, S. Saalfeld, B. Schmid, J. Y. Tinevez, D. J. White, V. Hartenstein, K. Eliceiri, P. Tom-
ancak, A. Cardona, Fiji: An open-source platform for biological-image analysis. Nat.
Methods 9, 676–682 (2012).
72. Y. Tasaki, M. Suzuki, K. Katsushima, K. Shinjo, K. Iijima, Y. Murofushi, A.Naiki-Ito, K. Hayashi,
C. Qiu, A. Takahashi, Y. Tanaka, T.Kawaguchi, M. Sugawara, T. Kataoka,M. Naito, K. Miyata, K.
Kataoka, T. Noda, W. Gao, H. Kataoka, S. Takahashi, K. Kimura, Y. Kondo, Cancer-specific
targeting of taurine-upregulated gene 1 enhances the effects of chemotherapy in pan-
creatic cancer. Cancer Res. 81, 1654–1666 (2021).
73. Y. Xue, H. Xiao, S. Guo, B. Xu, Y. Liao, Y. Wu, G. Zhang, Indoleamine 2,3-dioxygenase ex-
pression regulates the survival and proliferation of Fusobacterium nucleatum in THP-1-
derived macrophages. Cell Death Dis. 9, 355 (2018).
Acknowledgments: We thank the Center for Omics and Bioinformatics, Graduate School of
Frontier Sciences, University of Tokyo for the sequencing and its analysis. Funding: This work
was supported by the Grant-in-Aid for Scientific Research,the Japan Society for the Promotion
of Science (19K22674 to A.I., 20H03511 to Y.K., and 20K20598 to Y.K.) and the Research Grant of
the Princess Takamatsu Cancer Research Fund (15-24712 to Y.K.). Author contributions:
Conceptualization: A.M., M.S., and Y.K.Methodology: A.M., M.S., T.H., S.W., K.I., Y.M.,K.S., S.O., M.I.,
T.K., and S.K. Investigation: S.O., Y.H., A.I., H.K., and F.K. Funding acquisition: A.I. and Y.K. Project
administration: A.M. and Y.K. Supervision: Y.K. and K.O. Writing—original draft: A.M., M.S., and
Y.K. Writing—review and editing: A.M., M.S., and Y.K. Competing interests: A patent, Method
for detecting bacteria of genus Fusobacterium in order to diagnose endometriosis (WO2023/
042714), was submitted (international publication date, 23 March 2023). Data and materials
availability: The microarray and single-cell data have been deposited in the Genomic
Expression Archive (GEA) under accession number E-GEAD-426 and DRA013464, respectively.
The 16Ssequencing data of the endometrial microenvironment were derived from the
European Nucleotide Archive (ENA; https://ebi.ac.uk/ena/browser/home) under study numbers
PRJEB16013 and PRJEB21098 (17). Bacteria (F. nucleatum and L. iners) were obtained from the
Center for Conservation of Microbial Genetic Resource, Organization for Research and
Community Development, Gifu University under a materials transfer agreement (MTA).
Submitted 24 May 2022
Resubmitted 07 November 2022
Accepted 24 May 2023
Published 14 June 2023
10.1126/scitranslmed.add1531
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Muraoka et al.,Sci. Transl. Med. 15, eadd1531 (2023) 14 June 2023 17 of 17
Downloaded from https://www.science.org at Nagoya University on June 14, 2023

Use of this article is subject to the Terms of service
Science Translational Medicine (ISSN ) is published by the American Association for the Advancement of Science. 1200 New York Avenue
NW, Washington, DC 20005. The title Science Translational Medicine is a registered trademark of AAAS.
Copyright © 2023 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works
Fusobacterium infection facilitates the development of endometriosis through the
phenotypic transition of endometrial fibroblasts
Ayako Muraoka, Miho Suzuki, Tomonari Hamaguchi, Shinya Watanabe, Kenta Iijima, Yoshiteru Murofushi, Keiko Shinjo,
Satoko Osuka, Yumi Hariyama, Mikako Ito, Kinji Ohno, Tohru Kiyono, Satoru Kyo, Akira Iwase, Fumitaka Kikkawa, Hiroaki
Kajiyama, and Yutaka Kondo
Sci. Transl. Med., 15 (700), eadd1531.
DOI: 10.1126/scitranslmed.add1531
View the article online
https://www.science.org/doi/10.1126/scitranslmed.add1531
Permissions
https://www.science.org/help/reprints-and-permissions
Downloaded from https://www.science.org at Nagoya University on June 14, 2023