The estrogen-injected female mouse: new insight into the etiology of PCOS

Polycystic ovarian syndrome (PCOS) occurs in 5%–10% of all women of reproductive age [1, 2]. The disease begins at menarche, and symptoms generally include oligomenhorrhea, amenorrhea, anovulation, cystic ovaries, an elevated LH/FSH ratio, obesity, hirsutism, and insulin resistance. Cystic ovaries produce high levels of androstenedione, testosterone, and 17αOH-progesterone. The cysts themselves are remnants of atretic follicles, fluid filled and devoid of granulosa cells. As reported in the scientific literature in 1935 [3] the etiology of PCOS still remains obscure. Three major hypotheses are [4]: 1) PCOS is due to a primary neuroendocrine defect leading to an exaggerated LH pulse frequency and amplitude; 2) the disease is caused by a deficiency in insulin action leading to hyperinsulinemia; and 3) the primary fault occurs in the ovary and involves changes in FSH response.

Basic symptoms of PCOS such as anovulation and follicular cysts are produced in female mice by injecting them with estrogen, testosterone, or cortisone prior to 10-days of age [5, 6]. Significantly, during this same 10-day period the thymus gland is in its final stages of development. Interference with this process alters the evolution of 'self' versus 'nonself' recognition [7]. For example, thymectomy at 3-days of age prevents the production of regulatory T cells, and a number of autoimmune diseases ensue [8, 9]. Evidence presented herein suggests that steroids also forestall the production of regulatory T cells. The resultant autoimmune disease in this instance is PCOS.

Our first study (Table 1) was undertaken to determine if removing the thymus prior to estrogen injection could prevent anovulation and follicular cysts. Mice were thymectomized on day 3 (Tx-3) [10], and at 5–7 days of age given daily injections of estradiol-17β (E₂) (Treatment 7); the thymus was essentially replaced after steroid treatment by infusing the Tx-3 animals on day 15 with 20 × 10⁶ thymocytes from adult female donors [11]. Whereas all intact, 100-day-old female mice injected with E₂ at 5–7 days of age (Treatment 2) had ovaries containing follicular cysts, none were observed in 100-day-old Treatment 7 animals in which thymectomy preceded E₂ injection. In fact, the majority had ovaries with corpora lutea (CLs). Thus, when E₂ is unable to act on the thymus, ovulation occurs and follicular cysts do not develop. This implicates the thymus in the cysts' genesis. Note that infusion of thymocytes from 7-day-old females neither prevented cysts (Treatment 5), nor restored immunocompetency to Tx-3 mice (Treatment 4). The reason for this will be discussed later.

Study 2 demonstrated the ability of lymphocytes from E₂-injected mice to create follicular cysts when infused into Tx-3 animals. Female mice, injected at 5–7 days of age with either E₂ or diethylstilbestrol (DES), were killed at 100–110 days of age, and their lymphocytes (isolated from spleens) infused into 15-day-old, Tx-3 females. Control Tx-3 mice were given lymphocytes from vehicle-injected animals. Recipients were killed at 100 days of age, and ovaries from both donor and recipient mice examined for the presence of follicular cysts and CLs. All ovaries from E₂-injected (n = 10) and DES-injected (n = 20) donor-animals had follicular cysts; none had CLs. Seven of 10 of recipient mice infused with lymphocytes from E₂-injected animals, and 11 of 20 infused with lymphocytes from DES-injected animals, had ovaries containing follicular cysts. Taken in toto a reasonable explanation for these results is that the infusate from the E₂- and DES-injected animals contained autoreactive T cells, but lacked regulatory T cells; a combination that led to the creation of cysts in the recipient Tx-3 mice. The absence of cysts in some Tx-3 recipients is likely due to incomplete thymectomy. Thymic fragments are fully capable of producing regulatory T cells, as we have previously reported [12]. No cysts were seen in the ovaries of Tx-3 mice infused with lymphocytes from vehicle-injected animals. Figure 1 contains photomicrographs of ovaries from: (A) vehicle-injected donor and (B) Tx-3 recipient mice; (C) E₂-injected donor and (D) Tx-3 recipient mice; and, (E) DES-injected donor and (F) Tx-3 recipient mice. Note CLs in photomicrographs A and B and their absence in photomicrographs C through F.

The remaining studies determined the role of steroids in altering the ability of the thymus to produce regulatory T cells. Normally, thymocytes do not exit the thymus until they become mature T cells. However, estrogen is reported to increase the gland's vascular permeability [13]. This could cause early thymocyte discharge and contravene production of regulatory T cells. Female mice were injected at 5–7 days of age with E₂, testosterone (T), cortisol, or progesterone. At 12-days of age they were killed, thymuses and spleens removed and weighed, and thymocytes and splenocytes counted. Mice injected with E₂ had 50 million fewer thymocytes than control animals (p < 0.025) (Table 2). Mice injected with cortisol and progesterone had 30 million and 10 million fewer thymocytes, respectively (N.S.). Notably, animals injected with T had increased numbers of thymocytes. This was unexpected and necessitated a more detailed assessment of the effect of T on the thymus; for comparison, a similar analysis was made of the effect of E₂.

Female mice were injected at 5–7 days of age with E₂ or at 0–3 days of age with T (T having been shown to have a slower time course for producing cysts [5]). Beginning 1 day after the last E₂ injection and 2 days after T, and at intervals thereafter, the animals were killed, thymuses removed and thymocytes counted. Response to E₂ was rapid. Beginning two days after the last injection, thymuses of E₂-injected animals averaged 50 million fewer thymocytes than those of control mice (Fig. 2A). Response to T was quite different (Fig. 2B). Instead of a decrease, thymocyte levels initially increased, reaching a zenith four days after the last injection. At this juncture, thymuses of T-injected mice contained 20 million more thymocytes than those of control animals. Thymocyte levels subsequently fell, and after five days thymuses of T-injected mice contained 27 million fewer thymocytes than those of control mice. The slower rate of thymocyte loss, relative to the rapid loss induced by E₂, is reflected in the ability of each steroid to cause anovulation and follicular cysts. T is less effective than E₂, and produces a "delayed anovulation syndrome" [5, 14, 15].

In adult mice, thymocyte loss after E₂ administration is unevenly distributed. While the cortex experiences some loss of thymocytes, the greatest decline occurs in the medulla [16]. Prior to E₂, this region of the thymus contains substantial levels of CD4⁺CD8⁺ double positive (DP) thymocytes and CD4⁺ and CD8⁺ T cells. After E₂, all three subsets suffer a 99% reduction. Their lowered levels reflect a drastically reduced role for the medulla in T cell development. The loss of thymocytes was interpreted as being due to a blockage in early T cell development [16]. However, a significant influx in thymocytes precedes any loss (Fig. 2B). This response is best explained as resulting from an increase in the thymus' vascular permeability. Data from Abo [17, 18] support this interpretation. In his study of extrathymic T cell maturation, Abo noted that sinusoids of mouse liver contain lymphocytes that increase in number after E₂ injection. Their increase occurs in parallel with thymocyte loss and thymic involution. FACS analysis reveals them as being a mixed population of CD4^-CD8^- double negative (DN) cells with α/β T cell receptor (TcR), and CD4⁺ and CD8⁺ T cells [17]. Since DN cells are normally found in the cortex of the thymus, their appearance in the liver suggests a premature exit at, or near, the corticomedullary junction. Further evidence of a truncated pathway is indicated by the observation that estrogen-injected mice have significantly higher splenocyte levels than control animals (Table 3).

Fig. 3A depicts the normal developmental history of CD4⁺ and CD8⁺ T cells. Development begins when bone marrow precursors enter the thymus at the outer, sub-capsular region [19]. As the cells then migrate through the cortex they undergo a number of developmental stages, first as CD3^-CD4^-CD8^- triple negative (TN) thymocytes, and later, after CD3 expression, as DN thymocytes. During transit the α/β TcR is formed. DN thymocytes express CD4 and CD8 and, as DP thymocytes, interact with cortical epithelial cells to undergo positive selection. Thymocytes survive this process if their TcR, along with CD8 and/or CD4 accessory molecules, selectively binds to MHC class I and class II positive epithelial cells. Surviving DP thymocytes then pass through the corticomedullary junction into the medulla, where they undergo negative deletion. The TcR this time is exposed to self-antigens produced by medullary epithelial cells (MECs). Synthesis of self-antigens, such as tissue proteins, is under the control of the autoimmune regulator (Aire) promoter [20]. It should be noted that binding of the TcR to self-antigens in conjunction with MHC class I and class II molecules causes self-destruction and elimination of most autoreactive T cells. Regulatory T cells control those that are not eliminated and that enter into the system. Regulatory T cells comprise less than 10% of CD4⁺ T cells, and are identified by their expression of the CD25 antigen (interleukin-2 receptor α-chain) [9]. The CD4⁺CD25⁺ T cells are unique in that the binding of their TcR to self-antigens results in activation and conversion to a nonproliferative (anergic) state [21, 22]. Thus activated they are capable of preventing the activation of autoreactive CD4⁺ and CD8⁺ T cells [21]. All mature CD4⁺ and CD8⁺ T cells exit the thymus at the corticomedullary junction [19].

Fig. 3B shows the proposed thymocyte maturation pathway after an increase in thymic vascular permeability. Prothymocytes continue to enter at the outer, sub-capsular region, and proceed through stages of TN development, CD3 expression, and formation of the α/β TcR. However, instead of expressing CD4 and CD8, the majority of CD3⁺CD4^-CD8^- (DN) cells exit at the corticomedullary junction, bypassing positive selection and negative deletion. Their subsequent development into T cells has been proposed to occur in the sinusoids of the liver [17, 18]. The loss of 50 million thymocytes in E₂-injected mice (Fig. 2A) is likely due, in large part, to this altered pathway. The long-lasting nature of the new pathway is indicated by the 48 million (average) fewer thymocytes found in 60- and 100-day-old perinatally E₂-injected mice (Table 3).

The ability of CD4⁺CD25⁺ regulatory T cells to exert control over CD4⁺_Autoreactive and CD8⁺_Autoreactive T cells is attained in a region of the thymus (medulla) that is not fully functional until 7–10 days postpartum [7, 19]. Tx-3, therefore, permanently negates development of CD4⁺CD25⁺ regulatory T cells [9]. E₂ administration achieves this same result. Instead of physical removal of the thymus, however, E₂ initiates an altered thymocyte maturation pathway and de facto circumvention of the medulla. The involvement of CD8⁺_Autoreactive T cells in the formation of follicular cysts is suggested by the report that CD8⁺ T cells selectively infiltrate ovarian tissue during cyst formation [23]. In a previous study, we made an analysis of the combined effects of Tx-3 and E₂ on ovarian pathology [12]. Tx-3 by itself was found to have an incidence of ovarian dysgenesis, of 45%. The Tx-3 mice having normal ovaries (55%) were found at autopsy to contain thymic fragments. A second group, in which E₂ was injected after Tx-3, had a similar efficacy of ovarian dysgenesis (46%). However, instead of the remaining animals having normal ovaries, they either had ovaries that were dysgenic and cystic (39%), or ovaries that were cystic (15%). These results provided a basic understanding of the sequence in T cell development, as is diagrammed below.

CD4⁺_Autoreactive T cells are the first produced by the thymus. Tx-3 at point A prevents development of CD8⁺_Autoreactive cells, T_{Reg CD4} cells, and T_{Reg CD8} cells. This results in ovarian dysgenesis. Tx-3 or E₂ administration at point B forestalls development of T_{Reg CD4} cells and T_{Reg CD8} cells. This causes both ovarian dysgenesis and follicular cysts. E₂ injection at point C prevents development of T_{Reg CD8} cells, and follicular cysts are produced. Ovarian dysgenesis does not occur because T_{Reg CD4} cells are produced by the medulla prior to E₂ intervention. Shown in study 1 (Table 1), ovaries of mice in Treatment 5 have the same pathologies as described for the Tx-3 + E₂ group (dysgenic, dysgenic + cystic, and cystic) [12]; whereas, the ovaries of Treatment 7 animals are normal. This indicates that the infusate from 7-day-old donors, given to Treatment 5 mice lacked T_{Reg CD4} and T_{Reg CD8} cells; whereas, the infusate from adult donors, given to Treatment 7 animals contained T_{Reg CD4} and T_{Reg CD8} cells. These same results occurred in Treatment 4 and Treatment 6 animals, and for the same reason.

An examination of the literature indicates a strong likelihood of thymus involvement in estrogen and/or testosterone-induced anovulation in other animal species. For example, Kincl et al. [24, 25] reported that anovulation in E₂- and T-injected female rats could be prevented by thymocyte infusion. Notably, only thymocytes from adult donors were effective. Thymocytes from 5-day-old animals did not prevent anovulation. In primates the thymus undergoes its final development prenatally [7]. Steroid action would thus occur in utero. This could explain why injections of testosterone propionate (TP) given to pregnant rhesus monkeys on gestational day's 40–55, produces anovulatory female offspring [26, 27]. The female offspring have enlarged ovaries with multiple small follicles; an elevated LH/FSH ratio; and, high levels of serum 17αOH-progesterone and testosterone.

Additional evidence of steroid influence in utero is detailed in reports of the consequences of using DES in pregnant women [28‐35]. Prescribed from the 1940s until 1971, DES was banned by the FDA due to the large number of reproductive problems in daughters exposed in utero. Problems included an increased rate of primary infertility, oligomenhorrhea, amenorrhea, high levels of androstenedione and testosterone, facial hirsutism, and an elevated LH/FSH ratio. These symptoms are all associated with the formation of cysts [36, 37]. Notably, exposure to DES on gestational weeks 9 through 12 produced the highest rate of infertility [35]. This timeframe is coincident with the final developmental stages of the thymus [7].

The identity of the self-antigen(s) that CD8⁺_Autoreactive T cells regard as nonself, is at present, a matter for conjecture. MECs synthesize approximately 300 ectopic tissue proteins [20]. At least two are involved in autoimmune disease. A peptide epitope of insulin initiates CD8⁺_Autoreactive T cell destruction of pancreatic β cells [38], and zona pellucida glycoprotein 3 (ZP3) is implicated as the self-antigen involved in ovarian dysgenesis [39]. Synthesized in the ovary by the oocyte and granulosa cells [40], ZP3 is a prime candidate for the self-antigen involved in the formation of follicular cysts. Destruction of granulosa cells by CD8⁺_Autoreactive T cells would seriously impair the follicle's capacity to synthesize estrogen. Restoration of this ability might explain why injections of FSH cause ovulation in clomiphene-resistant PCOS women without intervention by either exogenous LH or hCG [41].

In conclusion: we have proposed that follicular cysts formed in a popular animal model of PCOS represent an autoimmune disease initiated by steroid administration. An increased incidence of autoimmune disease in DES-exposed women [42], lends further support for the autoimmune nature of PCOS. As maternally derived androgens and estrogens diffusing into the fetal area are limited by the amnion [43], and are normally at nanogram levels, it is unlikely that this source of steroid causes PCOS. The reproductive problems observed with DES came from milligram levels [44]. Potential sources of steroids at this level are phytoestrogens, contained in food supplements and ingested by some pregnant women. The Centers for Disease Control and Prevention, for example, report that 10% of representative samples of women in the United States contain urinary levels in the milligram range, of phytoestrogens found in flax seed [45]. Flax seed and soy bean products cause reproductive problems in female rats [46] and mice [47], and mice suffer thymocyte loss and thymic atrophy when given genistein, the phytoestrogen contained in soy beans [47]. Our future research will determine whether or not phytoestrogens cause anovulation and follicular cysts when administered to female mice during the thymus' critical period. We will also be investigating the impact of adrenal corticoids. While the bulk of this paper has concentrated on the role of gonadal steroids, the observation that adrenal steroids diminish thymic/spleen weight and numbers of thymocytes/splenocytes (Table 2), and can instigate cyst formation [6], raises the possibility that severe stress during pregnancy may be a factor in PCOS development.