- Top of page
- Materials and methods
Literature on the development of the human vagina is abundant; however, contributions concerning the prenatal development of the entire utero-vaginal anlagen (UVA) are rare or carried out in rodents. The primary epithelial characteristics in the adult vagina and uterus are determined during prenatal development and depend on epithelio-mesenchymal stroma interaction; thus an investigation summarizing the spatiotemporal distribution of relevant molecular markers in the entire human UVA will be of current interest. We phenotyped epithelial and mesenchymal characteristics in sagittal sections from 24 female fetuses of 14–34 weeks of gestation and two female newborns by immunostaining with cytokeratins 8, 13, 14 and 17, p63, bcl-2, bmp4, HOX A13, CD31, VEGF, SMA, Pax2 and vimentin. Epithelial differentiation followed a caudal-to-cranial direction in the UVA. Due to the cytokeratin profile of cytokeratins 8, 13 and 14, the characteristics of the different epithelial zones in the UVA could already be recognized in middle-age fetuses. Vaginal epithelium originated from the urogenital sinus in the lower portion and initiated the transformation of vimentin-positive Müllerian epithelium in the upper vaginal portion. During prenatal development the original squamo-columnar junction was clearly detectable from week 24 onwards and was always found in the cervical canal. Early blc-2 positivity within the surrounding mesenchyme of the entire vagina including the portio region pointed to an organ-specific mesenchymal influence. Prenatal findings in human specimens clearly show that fornix epithelium up to the squamo-columnar junction is of vaginal Müllerian origin, and the cervical epithelium cranial to the squamo-columnar junction is of uterine Müllerian origin and includes cells with enough plasticity to transform into squamous epithelium.
- Top of page
- Materials and methods
Anatomically, several internal female genital organs can be distinguished, the majority of which, i.e. vagina, uterus and uterine tube, develop from common embryonic structures, the fused Müllerian ducts. The morphology of the vagina and the uterine portions differs as to the epithelial lining and the muscular coats. The vagina is covered by a stratified squamous epithelium, the cervical canal is characterized by a columnar epithelium and cervical glands, and the uterine cavity is lined by a columnar epithelium and uterine glands that change in length during the uterine cycle (Marieb et al. 2008). There is a distinct border between the vaginal and cervical epithelium, which is called the original squamo-columnar junction (SCJ) and undergoes a positional change from the inside to the outside of the cervical canal during the different life (hormone) periods. According to their different functions, the walls of the vagina and uterus differ widely: whereas the smooth muscular coat of the vagina is thin, the uterine myometrium is bulky and consists of interlacing bundles.
Studies on the development of the entire human utero-vaginal anlagen (UVA) (Meyer, 1910; Hunter, 1930) or the human vaginal anlagen (VA) are rare (Mijsberg, 2007; Koff, 1933; Bulmer, 1957; Forsberg, 1973), mostly going back to the first decades of the last century; yet they are still the basis for embryological and clinical textbooks (Gray & Skandalakis, 1972; Ferris et al. 2004; Sadler,2004). With the exception of the early work of Meyer (1910) these studies do not lay special emphasis on the development of the transitory region between cervix and vagina. In contrast, Kurita's group has published several papers regarding the development of the female reproductive tract in mice including the vaginal/cervical transitory region (Kurita & Cunha, 2001; Kurita et al. 2005; Kurita, 2010). In summary, Kurita (2011) showed that during mice development the Müllerian ducts undergo a morphogenetic transformation from simple tubes into distinct organs and then differentiate to diverse epithelial cell types with a unique morphology in each organ. Within this process the transcription factor p63, induced by vaginal, cervical and uterine mesenchyme during development, was regarded to play a key role in the determination of epithelial cell fate (Kurita & Cunha, 2001). However, in adult mice, p63 is only expressed in the vaginal and cervical epithelium, but no longer in the uterine epithelium, when the developmental plasticity of epithelial cells is lost (Kurita et al. 2004). Kurita supposed rare epithelial plasticity to be restricted to small groups of so-called stem cells in cervix and uterus. In human specimens these cells are found to express combined CK17 and p63-positivity in the cervical epithelium (Martens et al. 2004, 2007), especially near the SCJ.
To provide up-to-date molecular correlates for the human utero-vaginal development that can be compared with the results of the above-mentioned cell biological studies in rodents we investigated the prenatal development of the entire UVA in complete series of human fetuses from week 14 to newborn and described our findings concerning the epithelial and mesenchymal differentiation. By application of antibodies raised against various cytokeratins, progenitor cells, transcription factors and smooth muscle actins as well as antibodies indicating cell survival, angiogenesis and mesenchymal origin (see Table 1), we were able not only to characterize the epithelial and mesenchymal structures at a defined stage but also to obtain detailed information about their spatiotemporal distribution. We presume that there is an interaction between the formation of mesenchyme and epithelial tissues, the knowledge of which is of high clinical relevance in regions of transformation (vagina/cervix).
Table 1. Antibodies used in immunohistochemistry
|Antibody (catalogue number)||Host||Dilution in IHC||HIER||Supplier||Marker for|
|BCl-2 (760-4240)||Mouse||RTUa||CC1 standard||Ventana, Mannheim, Germany||Cell survival (Hockenbery et al. 1990)|
|BMP4 (AP15370PU-N)||Rabbit||1 : 200||CC1 mild||Acris, Herford, Germany||Basal/progenitor cell compartments (Ter Harmsel et al. 1996)|
|CD 31 (SIG-3632-26)||Mouse||RTU||CC1 mild||Covance, Dedham, MA, USA||Endothelial cells and angiogenesis (Parums et al. 1990)|
|Cytokeratin 8 (760-2637)||Mouse||RTU||CC1 standard||Ventana||Non-squamous epithelium (Gown & Vogel, 1984)|
|Cytokeratin 13 (E018)||Mouse||RTU||CC1 short||Linaris, Wertheim-Bettingen, Germany||Non-cornified squamous epithelia, expressed in exocervix (Moll et al. 1982)|
|Cytokeratin 14 (RTU-LL002)||Mouse||RTU||CC1 mild||Novocastra, Newcastle upon Tyne, UK||Stratified epithelial cell types (Purkis et al. 1990)|
|Cytokeratin 17 (790-4560)||Rabbit||RTU||CC1 standard||Ventana||Basal/stem cells in complex epithelia (Smedts et al. 1992)|
|HOX A13 (sc-133669)||Rabbit||1 : 400||CC1 mild||Santa Cruz Biotechnology, Santa Cruz, CA, USA||Homeobox-gene, development of the upper vagina (Taylor et al. 1997)|
|p63 (MS-1084-P)||Mouse||1 : 400||CC1 standard||Termo Fisher Scientific||Basal/progenitor cells of many epithelial tissues (Yang et al. 1999)|
|Pax2 (18-0483)||Rabbit||1 : 80||CC1 standard||Invitrogen Corporation, Camarillo, CA, USA||Ductal and mesenchyme of urogenital system including the MDs and Wolffian ducts (Shapiro et al. 2004)|
|SMA (E046)||Mouse||RTU|| ||Linaris||Smooth muscle actins, myofibroblasts and myoepithelial cells (Skalli et al. 1986)|
|VEGF (E2614)||Rabbit||1 : 160||CC1 standard||Spring Bioscience, Fremont, CA, USA||Induction of angiogenesis (Ferrara et al. 2003)|
|Vimentin (E034)||Mouse||RTU||CC1 short||Linaris||Cells originating in the mesenchyme (Osborn et al. 1984)|
|Rabbit polyclonal lgG (ab27478)||Rabbit||1 : 200||CC1 short, mild and standard||Abcam, Cambridge, USA||Immunoglobulin G rabbit isotype control|
- Top of page
- Materials and methods
The present study was undertaken to examine the development and differentiation of the human utero-vaginal epithelia and mesenchyme. We here report that:
- Epithelial differentiation runs in a caudal-to-cranial direction: vaginal epithelial differentiation is followed by cervico-uterine epithelial differentiation, and the characteristics of the different zones of the UVA can be recognized in middle-aged human fetuses (Table 2: marked in red).
- A dual mechanism for vaginal epithelialization is seen: the most caudal vaginal epithelium is exclusively UGS-epithelium-derived; all the upper vaginal epithelium is a converted or transformed Müllerian epithelium (Table 2: marked in lilac).
- The epithelium of the fornices, the portio (ectocervix) up to the SCJ at the beginning of the endocervical canal is of vaginal origin.
- Mesenchymal input seems to be important for the regionalization of Müllerian epithelium (vagina, cervix and uterus).
- The original SCJ is always situated in the cervical portion, i.e. within the cervical canal during prenatal life, and is clearly detectable from week 24 onwards (Table 2: asterisk).
Table 2. Summarized spatiotemporal distribution patterns of molecular markers in the entire UVA. Respective immunoreactivities were categorized in a semiquantitative manner: Ø, No immunoreactivity; + Ø, low; +, moderate; ++, high immunopositivity; bcl-2 pattern of mesenchyme is lost; × SCJ is clearly detectable
Following the detailed investigations of Meyer (1910), Koff (1933) and Bulmer (1957), our study, based on nearly complete prenatal stages, is the first to contribute to the development of the human UVA. In the last decade, functional morphological studies concerning UVA development models were published by Kurita's group (Kurita & Cunha, 2001; Kurita et al. 2005; Kurita, 2010), who focused on mice and mice models and only in a few cases compared them with human probes. Although these publications present clear and strong results concerning the rodents, they suffer from the weakness that their human specimens were not complete, and their results cannot safely be connected and compared with the cell-biological findings in the laboratory animals or to clinical pathological aspects. A recent publication on human uterine development (Martens et al. 2007) lacks the younger human stages and does not consider the entire UVA. We are the first to analyse the immunohistochemistry of different cytokeratins, vimentin, Pax2, p63, bcl-2, bmp4 and HOX A13 in the entire human UVA in order to find new aspects as to the heterogeneity of the different epithelia in the UVA and their possible interaction with the underlying mesenchyme.
Our results clearly show that epithelial differentiation in the UVA does occur caudo-cranially, regardless of where the cells originate from. This is in accordance with the literature (Meyer, 1910; Forsberg, 1973; Orvis & Behringer, 2007; Kurita, 2010) and supports the results of our last study on early VA development (Fritsch et al. 2012) where we proposed that the caudo-cranial differentiation process is initialized by cells or a cellular stimulus from the UGS epithelium, itself is of endodermal origin. This is supported by the new results presented here – that in weeks 16/17 a group of UGS-derived squamous epithelial cells proliferates and extends into the lower vagina and initializes an epithelial differentiation first within the VA and then within the uterine portions.
We can be sure now that there is a dual mechanism of vaginal epithelialization, supporting the clinical compartment theory of the distal vagina (Höckel et al. 2011): the cells that extend into the caudal-most vaginal portion are CK 13-, 14- and p63- positive and vimentin-negative. However, those cells that are situated within the solid vaginal portion are vimentin-positive. Vimentin is a characteristic intermediate filament of mesoderm-derived cells, highly indicative of its derivation from the paramesonephric system (Van der Putte, 2005). The vimentin-positivity we noted therefore seems to be characteristic of the mesoderm-derived Müllerian epithelium and is observed neither in the UGS epithelium nor in the anorectal epithelia (not shown). The Müllerian-derived epithelium shows an ascending distribution of p63- (Table 2: marked in yellow) and CK 14-positive cells (Table 2: marked in red) that was shown by Kurita & Cunha (2001), which we assumed to be derived from the UGS in early vaginal development (Fritsch et al. 2012). In weeks 24/25 the solid Müllerian epithelium is transformed into a squamous epithelium by the ingrowth of blood vessels (Table 2: marked in pink). This transformed Müllerian squamous epithelium is found throughout the upper vagina, in the fornices, the ectocervix and in the entrance of the cervical canal. Our results revealed the Müllerian origin of this epithelium based on vimentin (Van der Putte, 2005) and Pax2 immunoreactivities that were found in the upper vagina, the cervix and the uterine segment at week 14. With increasing age the expression of Pax2 protein decreased in a caudo-cranial direction, and at week 24 it was restricted to the epithelium of the uterine segment.
In the adult, p63-positive cells are considered to be reserve cells in the vaginal and ectocervical epithelium as well as in subcolumnar cells at the SCJ (Martens et al. 2004). Furthermore, Martens et al. (2004) pointed out that in cervical intra-epithelial neoplasia, p63-immunostaining is strongly expressed irrespective of grade. Our results regarding human UVA development show an ascending p63-distribution from the vagina to the uterus in early fetal life (Table 2: marked in yellow). In later stages, p63-positive cells are regularly found in the basal cells of the vagina and the fornix (ectocervix), and they are scattered in the cervical and uterine epithelium (Table 2: marked in yellow). The latter cells are supposed to maintain the developmental plasticity, i.e. to be the targets of cervical or uterine squamous metaplasia (Kurita, 2011). As shown by Martens et al. (2004), HPV target cells are qualified not only by p63 but also by CK 17. We have shown that during prenatal life CK 17-positivity is first found in the lower vagina (Table 2: marked in violet), ascends towards the upper vagina, the fornix and portio, and is found in subcolumnar cells in the cervical epithelium in late fetal life. This finding is identical with the situation shown in the adult where CK 17 staining is only found in (endo)cervical reserve cells and reserve cell hyperplasia (Martens et al. 2004).
In several publications (Cunha, 1976; Kurita & Cunha, 2001; Kurita, 2011) it has been shown that in rodents the functional differentiation of the vaginal and uterine epithelia requires organ-specific factors. For the first time now, our results present evidence of this phenomenon in the human UVA. Bcl-2-positivity is found in the surrounding mesenchyme of all the VA as well as in the portio region until weeks 19/20, whereas there is no bcl-2-positivity in the neighbouring mesenchyme of the cervical and uterine portion of the UVA (Table 2: arrow). Vaginal mesenchyme bcl-2-positivity is lost in weeks 24/25 when the differentiation process into squamous epithelium is in progress. As bcl-2 is an apoptosis regulator and not tissue-specific, our results can only be considered an indirect hint, further underlined by the position of the border of the vaginal muscular coat illustrated by the SMA staining pattern in weeks 13/14 and 16/17. Up to week 24 the vaginal epithelium is also bcl-2-positive (Table 2: marked in blue), the following differentiation and transformation process in the human vagina seems to be supported by HOX A13 and bmp4 (Table 2: marked in green) which, according to Cai (2009), play a decisive role in the conversion of the intermediate mesoderm nature of the Müllerian duct in mice. According to our results, HOX A13-positivity in human specimens is not restricted to the vagina but is also found in the cervical and uterine portion (Table 2: marked in black) just at the time when the glandular structures develop and migrate from the epithelial surface into the underlying mesenchyme. In contrast to the literature (Jaubert et al. 2009), smooth muscle appearance in the UVA was detected in early fetal life, whereas the completion of the uterine muscular coat about week 24 was found to be in accordance with the literature. Our results clearly show that the mesenchymal surroundings of vagina and uterus are different. Therefore we suppose that the different mesenchymal environment may drive the epithelial cells to different cell fates within the vaginal anlagen and the uterine segment of the UVA. To examine the epithelial–mesenchymal interactions further investigations using other methods (e.g. in situ hybridization) will be necessary before using this study as the basis for revealing the epithelial differentiation influenced by the adjacent mesenchyme.
The original SCJ is situated within the cervical canal during all stages of fetal life. In the newborn this border descends towards the vagina. Thus our results are in complete agreement with those of Meyer (1910), gained from the observation of more specimens than we had at our disposal. Ferris et al. (2004), however, proposed a variable position of the SCJ in late fetal life and were not able to explain why squamous epithelial cells partially replace the ‘Müllerian’ columnar epithelium in the fetal cervix. We think that the SCJ may have been confused with the border of the two squamous vaginal epithelia, and that this may have led to a misleading interpretation. We have shown that the cervical glands appear in the newborn, and that they grow caudally towards the cervical orifice; consequently the SCJ descends towards the fornices. This process cannot be considered to represent a replacement of epithelia (Ferris et al. 2004) but must be seen as a displacement or dislocation of the squamous cervical epithelium. Malpica & Robboy (2009) pointed out that during adolescence cervical growth leads to a descending original SCJ and an exposure of cervical tissue outside the cervical os, i.e. to a repositioning of cervical epithelium to a vaginal environment. In accordance to Martens et al. (2004) we have shown that the cervical epithelium includes cells with the plasticity to transform into squamous epithelium.
In the course of our investigations we found that there is a probable dual mechanism causing vaginal epithelialization, but we also considered the possibility of a second dual mechanism in which the human cervix develops into three compartments: (i) the Müllerian columnar epithelium of the uterus and cervix, (ii) the Müllerian squamous epithelium of the cervix and the upper vagina, and (iii) the vaginal squamous epithelium of the lower vagina. This approach is an interesting one and might offer explanations concerning the genesis/development of lesions and carcinomata in this region. However, as pursuing this was far outside the scope of this study, we intend to follow up our present investigations with another study considering not only the theory of this approach but also its clinical consequences, ranging from human papillomavirus to carcinomata of the cervix and vagina, thus our findings concern data which may become of lifelong clinical relevance for affected persons.