MSCs—defined by their adherence to plastic, expression of a subset of cell surface markers, and ability to differentiate into adipogenic, osteogenic, and chondrogenic lineages —have to this date been isolated from several body tissues including bone marrow, fat, umbilical cord, placenta, amniotic fluid, umbilical cord blood, peripheral blood, and endometrium [32–35]. Bone marrow and adipose tissue have been the most common sources of clinical MSCs in horses, and they are also the most common sources used for clinical trials in humans. Collection of MSCs from these locations requires relatively invasive procedures involving sedation and local anesthesia, and carries the potential of postsurgical complications . Thus, alternative sources of equine MSCs, such as the endometrium, are desirable. A major advantage of isolating MSCs from the endometrium compared to bone marrow or adipose tissue is that cells can be harvested by biopsy collection [24, 37], which is a relatively noninvasive approach used routinely in horses for diagnostic purposes that does not require sedation or local anesthesia . In this study, we show for the first time that putative MSCs contained within the equine endometrium can be harvested and expanded in vitro, and have characteristics that may prove useful for tissue regeneration applications.
Endometrial MSCs had typical spindle-shaped morphology, indistinguishable from that of BM MSCs; however, they tended to grow faster than BM MSCs following initial seeding, as indicated by their mean doubling time values. In contrast, cloning efficiencies (CE) at passage 2 were similar for the two cell types, around 25–30%, and comparable to previous reports from 27%  to 34%  for equine BM MSCs. The faster initial growth of endometrial MSCs relative to BM MSCs may be conferred by their native in-vivo environment characterized by fast tissue turnover during the estrous cycle. If confirmed in future studies, this property of endometrial MSCs may provide an advantage over other MSC sources because it may allow shortening of the interval between collection of tissue samples and transplant of in-vitro expanded MSCs, which is a serious limitation of current BM and adipose MSC treatments in horses. In addition, based on cell yields obtained from 1 g of endometrial tissue (≥107 Muc-1– cells) and considering subsequent growth rates in culture (see Results), we estimate that a typical 0.2–0.4 g biopsy would readily yield >10 million cells after short-term expansion, a sufficient number for therapy applications in horses. Furthermore, when executed appropriately, the biopsy procedure does not result in damage or scarring of the uterus. Indeed, it has been shown that repeated collection of multiple biopsies (up to five each time) before estrus had no effect on subsequent pregnancy rates in mares .
Cells staining for CD44, CD105, CD146, and NG2 were located primarily around blood vessels within the equine endometrium, consistent with the identification of perivascular cells as native counterparts of MSCs in many different human tissues [7, 42], including the endometrium . By contrast, CD90 (clone OX7) followed a less restricted pattern throughout the stroma to include nonperivascular cells. The distinct abundance of CD90 compared to the other MSC markers tested suggests that this may not be an appropriate marker for equine MSCs in the endometrium.
Consistent with the definition of MSCs, endometrial stromal cells robustly maintained the expression of CD29, CD44, CD90, and CD105 in culture, as well as, to a lesser extent, perivascular markers, whilst having negligible expression of hematopoietic markers and MHC-II, in agreement with previous studies with human endometrial-derived MSCs [5, 25, 27, 43]. A limited number of studies have compared the features of endometrial MSCs with MSCs from other sources [23, 44]. Our finding based on results of flow cytometry and qPCR, showing that endometrial MSCs in culture display moderately higher levels of CD29, CD90, and CD105 but lower levels of NG2 than their BM counterparts, is consistent with data from Indumathi et al. . Whether this is indicative of differences in the abundance of stem cells between the two tissue sources or reflects tissue-specific changes in immunophenotype that may be induced in culture should be investigated in future studies.
That endometrial and BM MSCs have different properties was confirmed by the results of differentiation assays; specifically by the observation that while endometrial MSCs were able to undergo trilineage differentiation, their ability to generate cartilage was lower than that of BM MSCs based on a clearly reduced intensity of Alcian Blue staining in endometrial MSC-derived chondrogenic pellets (Fig. 5). In contrast, the opposite was observed in relation to the ability of MSCs to adopt a smooth muscle phenotype, as evidenced by a distinct increase in endometrial MSCs, but not in BM MSCs, in the levels of the mature smooth muscle marker, MYH11, after treatment with TGF-β1. There is evidence that significant differentiation bias can be conferred by the tissue of origin of MSCs . For example, while human multipotent cell populations from the myometrium and skeletal muscle had a similar immunophenotype and ability to differentiate into smooth muscle, only skeletal muscle-derived progenitors were able to undergo osteogenic and adipogenic differentiation . In light of this, a distinct ability of endometrial MSCs (compared to BM MSCs) to differentiate into smooth muscle may be related to the presence of a large smooth muscle component in the uterus, the myometrium. Whether our observation alternatively reflects the presence, natural or through contamination during sample collection, of myometrial precursor cells, different from MSCs, in the endometrial stroma needs to be investigated in future studies. Nonetheless, a reported intrinsic ability of human endometrial MSCs to differentiate into smooth muscle provides the rationale for specific therapeutic applications already being sought for these cells (e.g., pelvic organ prolapse) .