Folate Receptor Alpha Expression in Fallopian Tube: Further Evidence Supporting Tubal Contribution of “Ovarian” Endometriosis

Background Ovary is the most common organ site involved by endometriosis. We previously found that fallopian tube may contribute to the histogenesis of ovarian endometriosis. The nding was novel and requires further studies. We addressed this issue by examining a differentially expressed gene folate receptor alpha FOLR1 and its protein FRA in this study. Results A total of 144 tissue samples were studied. These included 32-paired 32-paired tubal-endometrial-ovarian endometriosis samples (n = 96), 18 samples of ovarian endometriosis without corresponding tubal or endometrium, and 30 ovarian tissue samples with ovarian surface epithelia but without endometriosis. Multiple comparisons among groups of ovarian endometriosis, normal fallopian tube and benign endometrium were performed. FOLR1 was highly expressed in the epithelia of fallopian tube and ovarian endometriosis, with paired endometrial samples showing a signicantly lower level of expression. Similar differential studies for FRA protein were performed through Western blot and immunohistochemistry (IHC). The expression of folate receptor alpha at both mRNA and protein levels in the tissues (fallopian tube or ovarian endometriosis vs. the endometrium) were signicantly different (p < 0.001). All ovarian surface mesothelial epithelia showed negative expression of FRA by IHC. Conclusions The results further support that the fallopian tube may contribute to the development of ovarian endometriosis. Understanding the tubal contribution to ovarian endometriosis should ultimately contribute to ongoing investigative efforts aimed at identifying alternative ways to prevent and treat endometriosis.

time, we assumed that the morphologic transitions from ovarian surface epithelia or ovarian epithelial inclusions (OEI) to initial endometriosis lesions represented a metaplastic process from OSE as discussed in 2005 [4]. However, nowadays we realize that the majority of epithelia attached on the ovarian surface and OEIs are derived from the fallopian tube. This was found when we studied the origin of OEIs from the fallopian tube epithelia vs ovarian surface mesothelia in 2011 [5].
The fallopian tube, particularly tubal mbriated end, has recently caught great attention in studies of adnexal pathology and the origin of ovarian serous cancers. This is mainly because most investigators believe that the majority ovarian serous cancers are actually derived from the fallopian tube, not the ovary as was historically asumed [6][7][8]. Similarly, in our study of the cellular origins of OEIs, we found that the majority OEIs (also called endosalpingiosis) display a phenotype suggesting they are derived directly from tubal epithelial cells rather than from ovarian surface mesothelial cells [5,9]. Indeed, microscopic examination of resected ovarian and tubal tissues frequently shows various stages of the following sequence: tubal epithelia mostly from mbriated end are frequently adherent on the ovarian surface, accompanied by invagination of morphologically similar epithelia onto ovarian surface, then invaginate into the cortex forming OEIs or endosalpingiosis [5,6]. From those ndings, we believe cellular metaplasia that eventuates in the formation of ovarian initial endometriosis occurs through the conversion of tuballike epithelia within the ovarian cortex into endometriosis-like tissue, rather than the conversion of ovarian surface mesothelial cells into endometriosis-like tissue. The potential for tubal mucosal epithelia to convert into endometriosis-like tissue is also supported by the phenomenon of "endometrialization" of the proximal end of the fallopian tube after a tubal ligation procedure [10,11]. Based in part on the aforementioned morphological ndings, we rst proposed the hypothesis that tubal epithelia contribute to the formation of ovarian endometriosis in 2014 and 2015 [9,12].
The biomarker FRA (folate receptor alpha), the product of the FOLR1 gene, has previously been shown to be differentially expressed between the fallopian tube and the endometrium [13,14]. In this study, we aimed to investigate FRA expression in tissue samples of human ovarian endometriosis and normal tubal and endometrial tissue samples to further examine the cellular origin of the ovarian endometriosis.

Results
Multiple unique genes identi ed in the fallopian tube over the endometrium A total of 4114 and 3451 genes were identi ed from the tubal and endometrial samples, respectively. The gene expression pro les of the paired tubal and endometrial samples were compared using a Volcano Plot. The threshold for a differentially expressed genes between the tubal and endometrium was set at ≥ 2.0-fold level. There were 1796 genes identi ed with more than two-fold differential expressions between the two tissue samples. All differentially expressed genes were further scrutinized and some representative genes were listed and studied previously [9].
Compared with the endometrial samples, the tubes showed a total of 911 upregulated genes. These included genes that were ≥ 50-fold (n = 8), ≥ 20-fold (n = 28) and ≥ 2-fold (n = 875) comparatively upregulated in the fallopian tube. There were no genes with more than 50-fold upregulation in the endometrial sample relative to the tubal tissue. FOLR1 (gene bank accessing number NM_016729), which was not listed previously and represented another most differentially expressed gene, showed 55.6-fold increment in the tubal sample compared to the gene expression level at the endometrium (p = 0.012831689). FOLR1 gene detection in the fallopian tube, endometrium, and ovarian endometriosis by real-time PCR Real-time PCR was used to examine the level of FOLR1 expression in 15 paired fresh tissue samples (total of 45 samples, including 15 each of the fallopian tube, endometrium, and ovarian endometriosis specimens). FOLR1 was highly expressed in the fallopian tube compared with the paired endometrium, with fold increment ranging from 9 to 35 (average fold change = 18.25, p < 0.001). Similarly, the FOLR1 gene was highly expressed in the samples of ovarian endometriosis compared with the paired endometrium, with fold increment ranging from 3 to 20 (average fold change = 11.18, p < 0.001). However, although there were some expression level differences between the fallopian tube and ovarian endometriosis samples, the difference did not reach statistical signi cance (p = 0.125). The gene expression level from the 15-paired cases, together with pooled data, is shown in Fig. 1. Among these 15paired samples, 12 pairs (from 1 to 12) were in the proliferative phase and 3 (13 to 15) were in secretory phase of the menstrual cycle (Fig. 1). The gene expression levels of the two menstrual phases were compared and no signi cant differences (p > 0.10) in the organ-matched samples were found (data not shown).
FRA protein expression in the fallopian tube, endometrium, and ovarian endometriosis by Western blot FRA protein level examination for the 15-paired samples was performed by Western blot. FRA protein expression was signi cantly higher in the fallopian tube and ovarian endometriosis samples than that in the endometrium, with an average fold of increment of 12.36 (p = 0.001) and 9.72 (p = 0.022), respectively. As expected, there were no signi cant differences between the fallopian tube and ovarian endometriosis samples (p > 0.10). These results were compatible with the ndings from real-time PCR validation, indicating that FOLR1 or FRA do not change signi cantly at the transcriptional and posttranscriptional levels. Western blots of 6 representative paired samples and pooled data of FRA protein expression are shown in Fig. 2.
FRA expression in tissue sections of the fallopian tube, endometrium, ovarian endometriosis and ovarian surface epithelia by immunohistochemistry A total of 114 tissue samples were analyzed with anti-FRA antibodies by IHC. These tissue sections represented the 32 paired tissue sections of the fallopian tube, endometrium, and ovarian endometriosis and additional 18 samples of ovarian endometriosis. All 32 normal fallopian tube samples showed strong and diffuse FRA expression as expected. The observed staining pattern was mostly intense cytoplasmic, with some extending to the luminal border of tubal epithelia. The staining score ranged from 185 to 300 with an average of 228 for the tubes. There were minor differences in FRA staining between tubal secretory and ciliated cells, with the latter showing stronger staining intensity on the luminal border than the former. However, these staining differences were not statistically signi cant (data not shown). In addition, compared with the menstrual phases, proliferative versus secretory, we did not observe a signi cant difference of FRA staining pattern or intensity in the tubal and endometrial epithelia.
FRA expression in 32-paired and 18-non-paired ovarian endometriosis samples were also highly positive, with IHC scores ranging from 94 to 300 with an average of 160. The majority of the epithelial cells were positive for FRA stain. Among all the 50 cases with endometriosis, there were 38 cases showing staining intensity score 3, 8 score 2, and 4 score 1. There were no signi cant differences in FRA expression scores between the paired and non-paired endometriosis. FRA expression was negative in all Pax-8 negative OSE as well as in all Pax-8 negative OEIs (endosalpingiosis) from the 30 ovarian sections without endometriosis examined (data not shown).
In the eutopic endometrial tissue sections, FRA expression was largely absent, with IHC scores that ranged from 0 to 84 (average 42). Typically, only focal areas of endometrial glands were positive with weak intensity at the luminal apical borders of the epithelial cells. Among the 32 endometrial samples, 26 were in proliferative and 6 in secretory phase; Occasional weak staining was seen in samples of both proliferative and secretory endometria with no clearly discernible differences in staining frequency or intensity.
Compared with the endometrium, both tubal and ovarian endometriosis samples showed signi cantly higher FRA IHC scores with p = 0.00051 and p = 0.00174, respectively. Interestingly, the difference of FRA expression between the fallopian tube and ovarian endometriosis also reached a statistical signi cance (p = 0.05) with the expression was higher in the tubal samples. The data is summarized in Fig. 3, and representative IHC pictures of the FRA expression are presented in Fig. 4. Stromal cells and blood vessels were all negative for FRA expression in all samples examined.

Discussion
Endometriosis remains a serious problem for reproductive-age women. The etiology and pathogenesis of endometriosis have been the subject of numerous investigations, although answers remain unclear [4,9,12,15]. Starting in 2005, our group described the earliest morphologic changes of endometriosis in the ovary, which we named "initial endometriosis" [4]. That study, which was largely based on light microscopic observations, provided morphologic evidence that retrograde menstruation may not explain how the initial endometriosis forms either on the ovarian surface or within the ovarian cortex. That study indirectly provided some supportive evidence that ovarian endometriosis may be derived from the fallopian tube. Morphologically, foci of initial endometriosis show gradual transitions from a typical example of endosalpingiosis (tubal type epithelia with ovarian stroma but without appreciable vascular changes) to areas with a classic appearance of endometriosis (endometrioid epithelia associated with endometrioid stromal cells and increased density of microcapillary vessels) [4]. Such morphologic transitions suggest that the foci of ovarian endometriosis started from the site within the ovary, rather than being deposited from retrograde menstrual endometrial tissue.
From what being presented, we questioned the hypothesis of retrograde menstruation and proposed that tubal cells are a plausible tissue source for ovarian endometriosis [16]. Recent studies of the tubal origin of ovarian serous carcinomas [5,17,18] have highlighted many previously under-recognized biologic and physiologic properties of the fallopian tube. In vivo, the fallopian tube is in close spatial proximity to the ovary [5,7,17,19,20], tubal epithelia are easily sloughed from the tubal mucosae [17,21], and the majority of the endosalpingiosis or OEIs are originated from the fallopian tube. Moreover, endosalpingiosis or OEIs are readily observed in ovarian cortex in many ovarian endometriosis [5]. It is likely therefore that the relationship between endosalpingiosis and initial endometriosis represents a trans-differentiation process from the former to the latter, although detailed underlying mechanisms remain to be clari ed.
Given the well-known limitations of morphologic assessment to support the novel concept of tubal origin of ovarian endometriosis, we have shifted morphologic observations to molecular studies addressing the issue since 2014. In one of our previous studies, we identi ed a set of novel genes that are either highly expressed in the normal tubal or endometrial tissue through a gene differential array study [9]. In that study, FMO3 and DMBT1 were examined as the biomarkers to test the hypothesis if the fallopian tube contributes the formation of ovarian endometriosis. These biomarkers were then validated in those paired samples of ovarian endometriosis, normal tubal and endometrial tissue similar to the current study. We found that FMO3 was highly expressed in the fallopian tube while was low in the paired endometrial samples, whereas DMBT1 had the reverse ndings. Based on those observations, we concluded that tubal epithelia contribute at least partially to the development of ovarian endometriosis.
The current study used another biomarker, FRA (folate receptor alpha), which is highly differentially expressed in the tubal and endometrial tissues. FRA, the product of the FOLR1 gene, is a glycosylphosphatidylinositol (GPI)-anchored protein that binds plasma folate (5-methyltetrahydrofolate) and transports it into the cell via endocytosis [22]. Folate is essential for 1-carbon metabolism, transferring single carbon units in reactions involving purine and pyrimidine synthesis, DNA repair, and methylation of various biomolecules including DNA, proteins, phospholipids, and neurotransmitters [23,24]. Folate de ciency has been linked with dysregulation of these processes and, in some cases, is associated with an increased risk of developing cancer, including serous type epithelial ovarian tumors [14,[25][26][27]. However, no speci c data have been reported on the expression of FRA in normal fallopian tube relative to ovarian endometriosis or to evaluate what role, if any, FRA may play in the development of ovarian endometriosis.
Since FOLR2 and its protein product FRA were highly differentially expressed in the tube and the eutopic endometrium, we assessed the tubal and endometrial samples with the FRA antibody on 50 patients with ovarian endometriosis (32 paired and 18 non-paired). We found that all fallopian tubal epithelia expressed FRA with strong intensity. In contrast, FRA expression in the endometrium was either negative or weakly and focally positive. Therefore, FRA may be considered a tubal-speci c biomarker when compared to that of the endometrium. This result was quite consistent with whole-genome expression microarray analysis [9]. Among 18 non-matched ovarian endometriosis, 15 cases had strong or moderately membrane and intracellular staining. Only 3 (18%) samples showed a weak staining on apical luminal borders. Furthermore, Pax-8 positive epithelial cells on the ovarian surface and OEIs showed moderate to strong FRA expression, while Pax-8 negative epithelial cells on the ovarian surface and OEIs were negative for FRA, which is supportive of tubal origin of ovarian endometriosis. Morphologically ovarian surface like epithelia and OEIs have two origins with one originated from classic ovarian surface epithelia (OSE) (Pax-8 negative) and the other from fallopian tube (Pax-8 positive) [5]. Original OSE negative for FRA expression suggests that ovarian endometriosis is unlikely derived from OSE through a metaplastic process. In short, our ndings from this study further support the hypothesis that ovarian endometriosis is at least partially derived from the fallopian tube. In our previous study [9], we estimated that approximately 60% of ovarian endometriosis may be originated from the tubal epithelia based on the FMO3 and DMBT1 expression study. This is correlated to the number of OEIs from fallopian tube [5]. The design of the current study does not allow such quanti cation; however, the ndings bolster the argument that the majority of ovarian endometriosis comes from the fallopian tube rather from retrograde menstrual endometrium.
Two essential conditions must be met to consider the fallopian tube to be the origin of endometriosis when present in the ovary. First, tubal cells must enter the ovary. Frequent detachment of tubal epithelia from mbriated ends makes this possible. Tubal cells are easily retrieved by ushing the fallopian tube [17,21]. The process is further facilitated by juxtaposed spatial relationship between tubal mbria and ovarian surface [19,28]. The rupture of ovarian surface caused by ovulation [29,30] provides a favorable condition for tubal epithelia to implant onto the ovary then get into ovarian cortex. The latter process, from a morphologic perspective, has long been described as endosalpingiosis [31][32][33]. Second, endosalpingiosis or OEI must transform itself into endometriosis. The latter probably occurs through metaplasia or trans-differentiation, a process that is commonly seen in the Müllerian system [34], although detailed molecular mechanism remains elucidated. Initial endometriosis within the ovary describes the morphologic transition of OEIs, with some glands of OEIs displaying the earliest morphologic changes of endometriosis in only half of the gland [4]. Metaplasia from tubal epithelia is the most likely explanation for this morphologic observation, especially since transitional areas from normalappearing tubal epithelia to endometrial like tissue are commonly present [17,35]. In summary, a graphic abstract is illustrated in Fig. 5. We may conclude, therefore, endometriotic or endometrioid epithelial cells are likely originated from tubal epithelia. However, by all means, so far we don't have solid scienti c evidence that ovarian endometriosis are either derived from or not coming from the endometrial cells from retrograde menstruation. Further studies in this regard are needed.
The novel ndings of this study have provided further evidence supporting our previously proffered theory of the tubal contributions for ovarian endometriosis. Although our ndings remain preliminary, they might provide an alternative way of thinking about the etiology and pathogenesis of endometriosis that might aid the prevention and early treatment of ovarian endometriosis.

Conclusions
The folate receptor alpha gene FOLR1 identi ed through a differential gene array analysis is highly expressed in ovarian endometriosis and the fallopian tube epithelial cells. However, the expression level of FOLR1 and its protein FRA is signi cantly lower in the endometrium and ovarian surface epithelia. These novel ndings further support that the fallopian tube is likely the cellular source of ovarian endometriosis. Understanding the tubal contribution to ovarian endometriosis should ultimately contribute to ongoing investigative efforts aimed at identifying alternative ways to prevent and treat endometriosis.

Tissue specimens
Fresh tissue samples, including normal fallopian tube mbria and corresponding endometrium were obtained from pathology specimens within 30 minutes of their surgical resections as described previously [9]. A total of 114 formalin xed and para n-embedded tissues were obtained from Henan Provincial People's Hospital, Zhengzhou, China. Among 114 samples, 96 were derived from 32 patients with each patient contributing 1 ovarian endometriosis, 1 endometrium and 1 fallopian tube sample and 18 patients each contributing one sample of ovarian endometriosis only. Among the 32-paired samples, 15 had fresh tissue samples available in addition to the formalin-xed para n sections. The remaining 18 patients had ovarian endometriosis from formalin-xed tissues only. In addition, 30 ovarian tissue sections containing Pax-8 negative OSE were included for FRA IHC stains. The representativeness of the ovarian endometriosis foci (n = 15) for quantitative PCR and Western blot analysis were con rmed by examining the corresponding frozen sections under the microscope. Foci of endometriosis, comprised predominantly of epithelial and stromal tissue, were manually dissected by removing non-endometriotic ovarian tissue.
Patients' age ranged from 22 to 50 years (mean 34.5). The reasons for total hysterectomy and bilateral salpingo-oophorectomy were prolonged ovarian endometriosis or benign gynecologic disorders including leiomyomata and benign ovarian cysts. No patient received hormonal treatment in 12 months prior to the surgery. Pathologic diagnosis of ovarian endometriosis were con rmed by two pathologists (RZ and WZ). Endometriosis, tubal mucosa and endometrium with both glandular epithelia and stroma were con rmed to be present under microscope. Determination of menstrual cycle phase (proliferative or secretory) for the 15-paired cases with both fresh and formalin-xed tissues was made by examining hysterectomy specimens microscopically. Patients with a history of malignancy or tubal ligation were excluded. The research protocol was approved by the institutional review board of Henan Provincial People's Hospital at Zhengzhou, Henan Province, China.

Microarray and Data Analysis
In order to identify tissue-speci c biomarkers, we compared the gene expression pro les between the fallopian tube and the endometrium from patients without evidence of endometriosis by gene array analysis, as previously described [9]. The endometriosis samples were manually dissected after microscopic con rmation. Three pairs of fresh samples of fallopian tube mbria and corresponding endometrial specimens mentioned above were sent to Kang Chen Bio-Tech (Shanghai, China) to perform whole-genome expression microarray analysis using the Agilent array platform (Agilent Technologies, Palo Alto, CA, USA). To limit potential confounding interferance, endometrial samples showing tubal metaplasia were excluded from the study. Total RNA from three pairs of hand-dissected epithelial samples under routine microscopy were prepared by using TRIzol (Invitrogen, Gaithersburg, MD, USA), further quanti ed by the NanoDrop 1000, and RNA integrity was con rmed by standard denaturing agarose gel electrophoresis. The Human Gene Expression Array was manufactured by Agilent with a cluster of 41000 genes and their corresponding transcripts and they are present in those public domain annotations.
Array hybridization and sample labeling were done according to the protocol provided by Agilent Technologies (Palo Alto, CA, USA) as described elsewhere [36]. Median normalization and subsequent data processing were analyzed by using the GeneSpring GX software (version 11). Highly differentially expressed genes were selected for further analysis after normalization of the raw data. Differentially expressed genes were identi ed through fold change ltering. Hierarchical clustering analyses were carried out also by using the Agilent GeneSpring GX software. Analysis of gene ontology and pathway studies were performed by using the standard enrichment computation method. A gene was considered to be highly differentially expressed between the tubal mbria and the endometrium if there was a ≥ 2fold difference in its expression level between or among the tissue categories, and the p-value less than 0.05 was considered signi cant.

Validation of Microarray Data by Real-Time PCR
Multiple differentially expressed genes were veri ed and published previously [9]. In the current study, we examined another highly differentiated gene FOLR1 to further test the possibility of tubal contribution of ovarian endometriosis.
To verify the gene expression data obtained from the microarray, real-time PCR was performed on FOLR1 gene, using total RNAs from 15-paired tubal mbria and corresponding endometrial samples. Among the 15-paired specimens, 12 were in the proliferative phase and the remaining 3 were secretory endometria, which were determined based on endometrial morphology. The FOLR1 gene was selected in this study because it was 56-fold more expressed in the tubal samples compared with the endometria by gene array study. In addition, the antibody against corresponding protein FRA suitable for immunohistochemistry on FFPE tissue samples was recently available. FRA represents one of the glycosylphosphatidylinositol (GPI)-anchored receptors that bind plasma folate (5-methyltetrahydrofolate) with high a nity (K D ~1nM), and transport it into the cell via endocytosis [22]. Together with folate receptor beta, gamma, and delta, FRA lies in tandem on chromosome 11q13 [22], although much less is known about those other family members. FRA has been shown to be expressed in normal placenta, fallopian tube, kidney, lung, breast and choroid plexus [14].
GAPDH was used as internal control. Real-time PCR was done on 15-paired ovarian endometriosis, tubal, and endometrial samples. Primers were designed using Primer 3 software and the sequences were as follows.
Quanti cation of the ampli ed products were normalized against GAPDH (ΔCt). Paired two-tailed t test was used to compare relative mRNA expression levels of the studied tissue samples. Statistical signi cance was de ned as a p value < 0.05.

Western Blot Analysis
Monoclonal antibodies against FRA, 1:500 dilution, was obtained from Dr. Daniel O'Shannessy (Department of Diagnostics Development, Morphotek Inc., Exton, Pennsylvania). All samples mentioned above were subsequently evaluated for protein expression by Western blot, as described elsewhere [38].
GAPDH antibody (Abcam, USA) was used as the loading control.

Immunohistochemistry
Among many differentially expressed genes, folate receptor-alpha (FRA) was one of the most highly differentially expressed. Immunohistochemistry (IHC) stain with antibodies against FRA, mentioned above, was performed as described previously [14]. Normal fallopian tube samples served as a positive control since it is ubiquitously expressed in tubal epithelial cells [14]. Negative controls were carried out by replacing primary antibodies with class-matched mouse and rabbit IgGs on parallel sections. The subcellular staining localization for FRA was both cytoplasmic and membranous.
IHC stained slides were evaluated and scored by counting at least 500 epithelial cells independently by two pathologists (RZ and WZ). Positive FRA expression was de ned as discrete membrane and intracellular (mainly cytoplasmic) brown color with at least weak intensity within the epithelial cells. IHC scoring criteria have been described previously [14]. Brie y, the staining intensity was scored as 0 if negative, 1 if with weak intensity, 2 with moderate intensity, and 3 if strong and intensely stained. Tissues with score 3 staining showed an intensely positive pattern that was readily identi able at low magni cation (4× objective) under light microscope. Sometimes a complete circumferential staining was seen. Score 2 was only visible at the 10× objective level and it was typically localized to the apical luminal or occasionally to the lateral cell border. In contrast, score 1 stain was generally limited to the luminal borders, or required 20× to 40× objectives to con rm. Percentage of the positive cells at each intensity was calculated for each case. The nal score was summarized by multiplying the staining intensity and the percentage of the positive cells for each sample [9].

Statistical Analysis
Multiple comparisons among categories of ovarian endometriosis, benign tubal and endometrial tissues were carried out by using student t-test and SPSS statistical software program version 13.0 (SPSS,        (7), and endosalpingiosis (8). Low-panel shows endometriosis develops from initial endometriosis within the ovary.
Endosalpingiosis like gland start to have endometrioid stroma and increased micro capillary vessels (9), more apparent in a magni ed view (10), endometriosis without evidence of bleeding (11), and typical appearance of endometriosis in the ovary (12).