Female Infertility and Disrupted Angiogenesis Are Actions of Specific Follistatin Isoforms

Abstract

The circulating and tissue-bound forms of follistatin (FST315 and FST288, respectively) modulate the actions of activins. FST knockout (KO/null) mice, lacking both isoforms, die perinatally with defects in lung, skin, and the musculoskeletal system. Using constructs of the human FST gene engineered to enable expression of each isoform under the control of natural regulatory elements, transgenic mouse lines were created and crossed with FST null mice to attempt to rescue the neonatal lethality. FST288 expression alone did not rescue the neonatal lethality, but mice expressing FST315 on the KO background survived to adulthood with normal lung and skin morphology and partial reversal of the musculoskeletal defects noted in FST KO mice. The FST315 rescue mice displayed a short period of neonatal growth retardation, impaired tail growth, and female infertility. The latter may be due to failure of corpus luteum formation, a decline in the ovarian follicular population, and an augmented uterine inflammatory response to mating. Failure of corpus luteum formation and impaired tail growth indicate abnormal vascularization and suggest that FST288 is required for the promotion of angiogenesis. The augmented uterine inflammatory response may result from the failure of FST315 to modulate the proinflammatory actions of activin A in the uterus or may result from the altered steroid milieu associated with the ovarian abnormalities. Although we cannot definitively conclude that the remaining defects are due to the absence of a particular isoform or due to variable expression of each, these models have demonstrated novel physiological processes that are influenced by FST.

FOLLISTATIN (FST), A MONOMERIC glycoprotein originally purified from follicular fluid, inhibits FSH secretion (1,2). FST was subsequently shown to bind activin A and B to neutralize their actions (3,4). Activin A, a member of the TGF-β superfamily, was first identified by its ability to stimulate FSH secretion. Subsequently, activin A has been shown to affect many diverse cellular processes including development, reproduction, and inflammation (for reviews, see Refs. 5 and 6). FST also binds other members of the TGF-β family including myostatin (7) and bone morphogenetic proteins 4, 6, and 7, but their affinity for FST is at least 10-fold lower than the binding of activins to FST (8).

FST is highly conserved between species. The mouse and human FST genes show 97% amino acid homology but have distinguishing differences in their nucleotide sequences. FST is encoded by a single gene of six exons, producing two isoforms by alternate splicing events (9). The two isoforms have distinct biochemical properties. FST288, a glycoprotein of 288 amino acids, binds to heparan sulfate proteoglycans and is regarded as a predominantly tissue-bound form directing activin toward a lysosomal degradation pathway (10,11). FST315, a 315-amino-acid glycoprotein, does not bind to heparan sulfate proteoglycans and is predominantly a circulating form (12).

Targeted disruption of the FST gene (Fst^−/−) resulted in death of the live-born pups within a few hours of birth due to respiratory difficulty associated with a decreased mass of the diaphragm and intercostal muscles (13). These mice displayed growth retardation, skeletal defects of the hard palate and 13th rib, whisker and tooth abnormalities and shiny taut skin. More recent studies involving tissue-specific disruption of the FST gene in ovarian granulosa cells demonstrated that FST is required for ovarian development and function because the mice are subfertile while young and become sterile as they age due to premature ovarian failure (14).

To date, no specific in vivo functions of these biochemically distinct FST isoforms have been identified. To define distinct roles for each protein, we created isoform-specific transgenic mice expressing either human FST315 (hFST315) or human FST 288 (hFST288) in addition to mouse FST. This involved cloning of the human FST gene to form a 95-kb construct with significant sequence 5′ and 3′ of the six exons and subsequent modification of this construct to express either hFST315 or hFST288. These mice were then crossed with mice heterozygous for the targeted deletion of the FST gene (Fst^+/−) (13) and ultimately bred to generate mice expressing the engineered human isoform-specific transgenes on a mouse FST knockout background. The resulting lines, expressing transgenic hFST315 (tghFST315) or human FST288 (tghFST288) in the absence of the mouse FST gene, were evaluated to determine whether the specific isoforms could rescue the neonatal lethality of the phenotype in the knockout mice. Analysis of these lines provides evidence to suggest that the FST isoforms exert specific physiological actions and increase our understanding of the biology of FST and the processes that these isoforms modulate.

RESULTS

Analysis of the Mouse Phenotypes at Birth

As reported (13), Fst^−/− mice did not move freely, were cyanotic, and died within 4 h of birth due to respiratory failure. These FST null mice were smaller than wild-type mice, showing a taut skin and an abnormal tightly curled whisker pattern (Fig. 1A). The tghFST288 mice were very similar to Fst^−/− mice in size, skin, and whisker phenotype (Fig. 1A). However, the tghFST288 mice were not cyanotic at birth but died within 12 h if left with their mothers, some living for up to 24 h if isolated and placed in a warm environment, raising the possibility that they may have been too weak to prevent smothering. In contrast, the tghFST315 mice at birth were similar to wild-type mice in size, weight, skin tone, and physical activity (Fig. 1B). However, tghFST315 mice were distinguishable from wild-type mice by three findings as follows: 1) the whiskers were not completely straight but not tightly curled as was noted in the tghFST288 and Fst^−/− mice; 2) they had an obvious genital tubercle, in contrast to the Fst^−/− and tghFST288 mice where no genital tubercle was noted, but it was smaller than the prominent tubercle in wild-type mice (data not shown); and 3) the distal end of the tail was often shiny, red and shorter (Fig. 1C, inset).

Figure 1.

Analysis of the Mouse Phenotypes

A and B, Gross appearance of mice at d 0: FST null (Fst^−/−) mouse and tghFST288 mouse (A) and tghFST315 mouse and wild-type (Fst^+/+) mouse (B); C, 10-d-old mice: Fst^+/+ (left) and tghFST315 (right); tghFST315 mice show growth retardation and shorter, reddened tails of varying thickness often with a black tip (inset). D, Body weight and CRL of mice at birth: FST null (black bars) and tghFST288 animals (gray bars) weigh significantly less than wild-type mice (white bars) at birth, whereas the tghFST315 mice (hatched bars) are significantly heavier than Fst^−/− mice but not significantly different from either tghFST288 or wild-type mice. The body lengths of the Fst^−/− and tghFST288 mice are significantly lower than the wild-type and tghFST315 mice, whereas the tghFST315 mice are not significantly different from the wild-type mice at birth. E and F, Alizarin Red-stained ribs of tghFST288 (E) and tghFST315 (F) mice at birth. The partial 13th ribs, seen in tghFST288 mice, are indicated by the arrows, whereas tghFST315 typically have 13 complete ribs. G, Eye phenotype of 3-wk-old tghFST315 mice showing the decreased width of the palpebral fissure. H, In tghFST315 mice, a part in the fur was observed down the center of the back of the head and neck.

At birth, the body weight and body length (crown-rump length, CRL) of Fst^−/− mice and tghFST288 mice were significantly lower than wild-type mice (P < 0.05) (Fig. 1D), suggesting that this transgene did not alter the phenotype of the Fst^−/− mice. In contrast, body weight and CRL of the tghFST315 mice were not significantly different from the wild-type mice, indicating that the availability of FST315 alone could rescue this aspect of the Fst^−/− phenotype (Fig. 1D).

The lung weights of the Fst^−/− and tghFST288 mice were significantly lower than the wild-type mice (P < 0.001) in contrast to wild-type and tghFST315 mice (Table 1). Liver weights were significantly lower in the Fst^−/− and tghFST288 mice (P < 0.001) as well as in tghFST315 mice (P < 0.05), compared with wild-type mice; similar data were noted for splenic weights (Table 1). There was no difference in the mean weights of the kidney and heart (Table 1). Despite the absence of the genital tubercle in the Fst^−/− mice and the tghFST288 mice, the general appearance of the bladder, ureter, and kidney, in both these lines, appeared normal.

Table 1.

Fixed Organ Weights of Day-0 Mice

Tissue	Fst+/+	Fst−/−	tghFST288	tghFST315
Lung	45.7 ± 8.9a	26.0 ± 4.8a	27.7 ± 4.7a	35.9 ± 8.0a
Liver	66.3 ± 5.0a	41.4 ± 9.3a	40.6 ± 6.2a	50.3 ± 3.3a
Spleen	2.1 ± 0.2a	1.1 ± 1.0a	0.6 ± 0.4a	1.3 ± 0.2a
Kidney	10.4 ± 1.0a	9.7 ± 1.3a	8.8 ± 1.0ia	9.9 ± 2.5a
Heart	10.4 ± 2.2a	8.6 ± 2.5a	10.0 ± 2.3a	7.1 ± 0.7a

Data are shown in milligrams (mean ± sd). Wild-type Fst^+/+ n = 6; Fst^−/− n = 11; tghFST288 n = 5; and tghFST315 n = 3. 

^aGroups with shared letter superscripts are not statistically different (P > 0.05). 

The lung histology in Fst^−/− mice showed markedly thickened interalveolar septa and a decrease in the size of the alveolar spaces (Fig. 2B), in contrast to the greater number of alveolar sacs and thin interalveolar walls in wild-type mice (Fig. 2A). The histology of the lungs in the tghFST288 mice (Fig. 2C) was similar to that of Fst^−/− mice. In contrast, no differences were observed between the lung morphology of the tghFST315 mice (Fig. 2D) and wild-type mice.

Figure 2.

Cross-Sections of Lung Stained with H&E from Day-0 Mice

A, Wild-type mouse; B, Fst^−/− mice show markedly thickened interalveolar septa (arrows) and a decrease in the number of the alveolar spaces in contrast to the large number of alveolar sacs and thin interalveolar walls in wild-type mice (A); C, the histology in the tghFST288 mice (C) is similar to that of Fst^−/− mice (B); D, no difference was observed between the lung morphology of the tghFST315 mice (D) and wild-type mice (A). Bar, 500 μm.

Histological analysis of the liver in the Fst^−/− mice showed that the hepatocytes were reduced in size and showed increased cell degeneration associated with dilated hepatic sinusoids when compared with wild-type mice (data not shown). The histology of the liver in the tghFST288 mice was similar to the Fst^−/− mice, whereas the tghFST315 mice were not different from the wild-type mice.

In Fst^−/− mice, microscopic examination of the skin showed decreased undulations, a thin and poorly differentiated epidermis, and decreased number of hair follicles, and the fibroblastic-muscular components of the dermis were poorly organized (Fig. 3B). The appearance of the skin, both macroscopically and microscopically, in the tghFST315 (Fig. 3D) mice did not differ from the wild-type mice (Fig. 3A), whereas the tghFST288 (Fig. 3C) mice showed taut shiny skin that was not different from that of the Fst^−/− mice. These histological appearances were consistent with the macroscopic appearances as shown in Fig. 1, A and B.

Figure 3.

Cross-Sections of Skin Stained with H&E from Day-0 Mice

A, Wild-type mouse (Fst^+/+); B, Fst^−/− mice. In the Fst^−/− mice, the skin showed decreased undulations, a thin and poorly differentiated epidermis, and decreased number of hair follicles (arrows), and the fibroblastic and muscular components (arrowheads) of the dermis were poorly organized. C and D, These changes were essentially identical in the tghFST288 mice (C) but were not present in the tghFST315 mice and more typical of wild-type mice (D). Bar, 100 μm.

As noted earlier (13), the number of ribs in the Fst^−/− mice varied between 12 ribs, a partial 13th rib, or 13 complete ribs. However, in the strain used in our experiments, the Fst^−/− mice had a partial 13th or 13 complete ribs. Wild-type mice consistently had 13 complete ribs, whereas tghFST288 mice had 12 or only a partial 13th (Fig. 1E), and tghFST315 mice had a partial 13th or 13 complete ribs (Fig. 1F and data not shown).

Postnatal Development of Wild-Type and tghFST315 Mice

Because the Fst^−/− and tghFST288 mice died at birth, further developmental comparisons were not possible. The tghFST315 mice showed a number of interesting phenotypes as they grew to adults. The tghFST315 mice showed growth retardation compared with the wild-type (Fig. 1C). Specifically, the body weights of male wild-type and tghFST315 mice were not significantly different at birth, then were significantly lower (P < 0.05) between 1 and 5 wk of age, but reached wild-type levels by 6 wk of age (Fig. 4A). Although the body lengths were not significantly different at birth, the body length (CRL) of tghFST315 males was significantly lower than wild-type males (P < 0.05) at 3–4 wk of age (Fig. 4C). Female tghFST315 mice had a significantly smaller CRL until 4 wk of age (P < 0.05), and their body weight remained significantly lower than wild-type mice until 6 wk of age (P < 0.05) (Fig. 4, B and D). For both male and female tghFST315 mice, tail lengths were significantly lower from birth (P < 0.05) compared with wild-type mice and remained significantly lower throughout life as tail length plateaued at 5–6 wk in all groups (Fig. 4, E and F). The tghFST315 mice (Fig. 1G) had a smaller palpebral fissure than wild-type mice, giving the appearance of microphthalmia. The horizontal and vertical dimensions of the palpebral fissures and orbital openings were measured in adult mice from the ages of 7–13 wk at dissection. The horizontal palpebral fissure of wild-type mice (n = 9) was 2.92 ± 0.17 mm (mean ± sd) in comparison with 1.33 ± 0.33 mm in tghFST315 mice (n = 8) (P < 0.0001). The vertical palpebral fissure measured 1.89 ± 0.15 mm in wild-type mice compared with 0.50 ± 0.08 mm in tghFST315 mice (P < 0.0001). The orbital opening of tghFST315 mice was also marginally smaller than wild-type mice, with the vertical plane of the orbital opening measuring 4.28 ± 0.64 mm in wild-type mice and 3.64 ± 0.58 mm in tghFST315 mice (P = 0.0496) and the horizontal plane measuring 3.94 ± 0.64 mm in wild-type and 3.37 ± 0.57 mm in tghFST315 mice (P > 0.05). The eyeball itself was of normal diameter in tghFST315 mice, measuring 3.01 ± 0.14 mm, in comparison with wild-type mice at 3.07 ± 0.10 mm (P > 0.05) The narrow palpebral fissure was consistent with the increased discharge from the eyes probably due to a reduced ability to clear foreign objects and secretions as well as being prone to eye infections. Tests of visual acuity suggested that vision was normal based on their ability to extend the forelimbs toward a wire grid. The tghFST315 mice also showed fur abnormalities in that a part appeared down the center of the back, beginning between ears and extending to about halfway between the ears and the base of the tail (Fig. 1H).

Figure 4.

Body Weight, Body Length, and Tail Length

Body weight, body length, and tail length of male wild-type (▪), male tghFST315 (□), female wild-type (•), and female tghFST315 (○) are shown. A, C, and E, Male mice (wild-type n = 6; tghFST315 n = 6 until 7 wk, n = 1 until 17 wk); B, D, and F, female mice (wild-type n = 6; tghFST315 n = 7 until 7 wk, n = 5 until 26 wk). A, The body weight of tghFST315 male mice was not significantly different from wild-type mice at birth, but between 1 and 5 wk of age was significantly lower (P < 0.05) and then attained wild-type levels from 6 wk of age. C, Body length of male wild-type and tghFST315 mice were not significantly different at birth; however, tghFST315 mice were significantly shorter at 3–4 wk of age (P < 0.05). Similarly, female tghFST315 mice (D) had a significantly shorter body length until 4 wk of age (P < 0.05), but their body weight (B) remained significantly lower until 6 wk of age (P < 0.05). For both male and female tghFST315 animals (E and F), tail length was significantly shorter from birth (P < 0.05) compared with wild-type mice and remained significantly shorter throughout life as tail length plateaued at 4–6 wk in all groups. Data are shown as mean ± sd.

The tails of many tghFST315 mice were slightly reddened and shiny at birth. These features became more marked over time in all of the tghFST315 mice, and the failure of tail growth was accompanied by irregularities in thickness and thinning at the tip. Around 7 d postpartum, the tail developed a red or black tip (Fig. 1C, inset) consistent with tissue necrosis, which disappeared by 3–4 wk, leaving the tail significantly shorter than wild-type mice thereafter.

At 7 wk, there were no significant differences in lung, liver, spleen, kidney, heart, diaphragm, calf muscle, and testis weights between wild-type and tghFST315 mice (data not shown).

Reproductive Function

The tghFST315 males were fertile with normal testicular histology. However, the females were infertile showing copulation plugs but not producing litters. Female tghFST315 mice were first exposed to a male at 45 d of age; however, the first copulation plugs were not detected until 53 d. Plugs were detected in these animals until 168 d of age, after which no plugs were detected despite exposure to proven fertile males for at least 7 d. In contrast, copulation plugs were detected in wild-type littermates as early as 42 d, and these mice remained fertile through 12 months of age.

Of 11 tghFST315 females first mated between 53 and 168 d of age, four plugged on the first night, one on the second, and four on the third, and two took more than 4 d to plug. This indicated that the animals were exhibiting estrus behavior, so patterns of vaginal smears in these animals were studied. In wild-type mice, the vaginal smears showed four distinct stages typical of normal estrous cycles, namely diestrus, proestrus, estrus, and metestrus. Our results indicate that the tghFST315 mice have abnormal cycles wherein both leukocytes and cornified epithelial cells were observed on each day, but uncornified epithelial cells were absent (see supplemental Fig. 2, published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org).

In view of this reproductive behavior and infertility, 7- to 8-wk-old wild-type and tghFST315 females were mated, confirmed by the presence of a copulatory plug and killed at 4 d post coitum (dpc). Flushing of the uteri and uterine horns recovered a mean of 9.2 ± 3.4 (n = 5, mean ± sd) blastocysts from wild-type mice but none from tghFST315 mice (n = 5).

Ovarian histology in the 7-wk-old wild-type mice at 4 dpc showed a normal complement of primary, secondary, and antral ovarian follicles and the presence of multiple corpora lutea (Fig. 5A). In striking contrast, the tghFST315 mice had no corpora lutea but showed several large cysts containing eosinophilic serous material consistent with follicles that have recently ruptured (Fig. 5B). Serial sections of these cysts did not demonstrate a retained ovum, and the surrounding stroma showed that the numbers of primordial, primary, and secondary ovarian follicles were substantially reduced in comparison with wild-type mice (Fig. 5A). To quantify the difference between wild-type and tghFST315 mice, ovarian follicular populations and corpora lutea were counted in every 10th section of an entire ovary, and the summed totals for each ovary are presented in Table 2. It should be noted that because of the methodology, large structures may have been counted more than once and will therefore be overrepresented in the results, and small structures may be underrepresented. However, the methodology used was identical for both wild-type and tghFST315 mice. The data clearly demonstrate that in the tghFST315 mice, there was a marked decrease in all types of ovarian follicles, corpora lutea were not identified, and the presence of cystic follicles was noted. The ovarian histological patterns of both the wild-type and tghFST315 mice at 6 months of age showed no differences to those at 8 wk, with no corpora lutea observed, and a markedly reduced number of ovarian follicles was noted.

Figure 5.

Ovarian Phenotype

A–C, Sections of ovary stained with H&E. A, Ovary of 7-wk-old wild-type mouse at 4 dpc showing corpora lutea (CL) and follicles at various stages of development (*). Bar, 100 μm. B, Wild-type ovary from 6-month-old virgin animal, showing follicles at various stages of development (*) and a corpus luteum (CL). Bar, 100 μm. C, Ovary of 7-wk-old tghFS315 mouse at 4 dpc. No corpora lutea are present; however, a serous cyst (Cy) is evident. Some follicles are present (*). Bar, 100 μm. D, RT-PCR of FST mRNA from adult ovary in wild-type and tghFST315. +RT represents the band amplified from cDNA, whereas −RT represents that there is no cDNA amplification. E, Western blot of FST expression in adult ovarian tissue. FST was detected in wild-type (Fst^+/+) and tghFST315 ovary.

Table 2.

Ovarian Follicular and Corpora Lutea Data

Ovarian Structure	Primordial Follicle	Primary Follicle	Secondary Follicle	Antral Follicle	Corpus Luteum	Cyst
Wild-type	115.8 ± 16.1	102.8 ± 22.5	118.3 ± 10.6	117.3 ± 19.1	88.5 ± 23.7	0.00 ± 0.00
tghFS315	7.7 ± 6.2	1.0 ± 0.9	5.3 ± 2.5	0.00 ± 0.00	0.00 ± 0.00	25.7 ± 14.1

Shown are the summed data of ovarian follicular populations and corpora lutea observed in every 10th section of entire ovaries of wild-type and tghFS315 mice at 4 dpc (mean ± sem). Wild-type, n = 4; tghFS315, n = 3. 

When the reproductive tracts of virgin wild-type and tghFST315 animals were dissected out en bloc, it was clear that the tghFST315 mice have visibly shorter uterine horns and vagina (Fig. 6A). The endometrium of the young adult virgin tghFST315 mice (n = 3) was similar to that of wild-type mice, but the lumen of the uterus of tghFST315 mice showed the presence of small increases in the number of leukocytes (data not shown).

Figure 6.

The Female Reproductive Tract

A, Gross appearance of the female reproductive tract: tghFST315 mouse (left) and wild-type mouse (right). Note the shortened uterus and vagina in the tghFST315 mouse. B–G, Cross-sections of reproductive tract of females stained with H&E. B, uterus of 7-wk-old wild-type mouse at 4 dpc showing normal stromal and epithelial organization (arrows) and a clear lumen (Lu). Bar, 100 μm. C and D, Uterus of 7-wk-old tghFST315 mouse at 4 dpc. C, Leukocytic infiltration (LI) in the lumen (Lu) lined by columnar epithelium (arrows). Bar, 100 μm. D, Higher magnification of leukocytic infiltration (LI) in the lumen (Lu) of the uterus in 7-wk-old tghFST315 at 4 dpc. The epithelial lining (arrow) is disrupted (arrowhead), and the infiltrate extends into the endometrial stroma. Bar, 50 μm. E, Uterus of tghFST315 animal at 6 months of age. There was a huge infiltration of leukocytes (LI) extending to the myometrium (My). Bar, 200 μm. F, Higher magnification of the uterus of 6-month-old tghFST315 animal showing leukocytic infiltration (LI) and myometrium (My). Bar, 50 μm. Inset, Leukocytic infiltrates shown in higher magnification. Bar, 20 μm. G, Uterine lumen of a 12-month-old tghFST315 mouse. After mating, no plug was detected; however, the black arrows show sperm heads present within accumulated secretions and cell debris in the uterus 40 d after last mating. Bar, 50 μm.

In contrast to wild-type uteri (Fig. 6B), sections of 7-wk-old tghFST315 uteri at 4 dpc demonstrated a huge infiltration of leukocytes, predominantly neutrophils and macrophages, consistent with a substantial inflammatory response within the lumen (Fig. 6C) with disruption of the luminal epithelium (Fig. 6D).

Four female virgin tghFST315 mice at 5.5 months (n = 3) or 11 months (n = 1) were co-caged with proven sires for at least 7 d and comprehensively assessed for copulation plugs each morning. They did not show a copulation plug, and thus it was assumed that no mating event took place. When these mice were killed at 6 months (n = 3) or 12 months (n = 1), respectively, and compared with the tissues of 6-month-old virgin wild-type mice, their uteri were enlarged due to a substantial leukocytic infiltration (Fig. 6, E and F, inset), with neutrophils predominating centrally and lymphocytes at the periphery. The infiltration also extended to the myometrium. (Fig. 6, E and F). Furthermore, despite the failure to plug, all tghFST315 mice showed sperm in the uteri when killed at least 7 d after the males were removed. In fact, sperm heads were still observed in the uterine lumen of the 12-month-old tghFST315 mouse 40 d after the male was removed (Fig. 6G).

Expression of FST Isoforms

At birth, mouse FST288 (mFst288) and mouse FST315 (mFst315) mRNA expression were detected by RT-PCR in wild-type mice, and in the transgenic mouse lines expressing hFST288 and hFST315 on a wild-type background. However, when the transgenic mice were crossed onto the FST null (Fst^−/−) background, only the human FST gene expressing either hFST288 or hFST315 was detected, as shown in skin (Fig. 7A). FST mRNA was also detected in wild-type and tghFST315 adult ovary by RT-PCR (Fig. 5D).

Figure 7.

Expression of FST mRNA and Protein

A, Comparative mRNA expression of the different FST isoforms present at birth in the different mouse genotypes. Data are shown in skin and are representative of other tissues. Note the expression of mRNA for both mouse isoforms in the wild-type mice but the absence of Fst in the knockout. Also, note the presence of mRNA for hFST288 but not hFST315 in the tghFST 288 line and the presence of mRNA for hFST315 but not hFST288 in the tghFST 315 line. B, Western blot of FST expression at birth and at 7 wk of age in mouse liver homogenates. FST was detected at birth in wild-type (Fst^+/+) and for mouse FST heterozygotes (Fst^+/−) but not in mouse FST knockout mice (Fst^−/−). FST was detected in tghFST288 and tghFST315 at birth and at 7 wk in wild-type (Fst^+/+) and tghFST315 mice.

FST protein was detected by Western blot analysis in wild-type and heterozygous, tghFS288, tghFS315 mouse liver extracts from d-0 mice and in the wild-type and tghFST315 mice at 7 wk (Fig. 7B). FST protein was detected in adult ovaries of wild-type and tghFST315 mice (Fig. 5E). FST protein was not detected in Fst^−/− mice (Fig. 7B).

FST isoform mRNA expression was assessed by quantitative RT-PCR at 7 wk (Fig. 8, A–D). Based on copy numbers of FST isoforms per microgram RNA converted, in wild-type mice at 7 wk, mFst315 mRNA expression was proportionally higher than that of mFst288. Specifically, wild-type mFst315 expression was 10.0-fold higher than mFst288 in lung (P < 0.001), 7.1-fold higher in kidney (P < 0.001), 12.6-fold higher in muscle (P < 0.001), and 5.0-fold higher in skin (P < 0.001). In the wild type, mFst288 mRNA expression in the kidney was significantly higher than mFst288 mRNA in lung, skin, and skeletal muscle (P < 0.05). Skin mFst288 was significantly lower than muscle (P < 0.05) but not significantly different from lung. Additionally, mFst315 isoform expression in skin was significantly lower (P < 0.05) than mFst315 expression in other tissues. In the tghFST315 mice, neither mouse nor human FST288 mRNA expression was detected in lung, skin, muscle, or kidney (data not shown). The mRNA expression of the hFST315 transgene at 7 wk was considerably lower than the level of expression of the mFst315 isoform in wild-type mice at the same age. Specifically, relative to wild-type mFst315 mRNA expression, and normalized to GAPDH expression, tghFST315 was 886-fold lower in the kidney (Fig. 8B) (P < 0.001), 277-fold lower in the skin (Fig. 8C) (P < 0.001), and more than 1000-fold lower in muscle (Fig. 8D) (P < 0.001). Problems were experienced in normalizing FST mRNA expression in the lung to the expression of the housekeeping gene GAPDH, because there was a comparative decrease in the expression of this gene in the lung of the tghFST315 mice relative to that of wild-type mice. Unnormalized raw data suggested a more than 1000-fold decrease in transgene expression (P < 0.001) (Fig. 8A). Other housekeeping genes, such as β-actin, were investigated, but these also showed an unacceptable variation between wild-type and tghFST315 mice. In contrast, in tghFST315 mice, there was no significant difference in tghFST315 mRNA expression between skin, lung, muscle, and kidney (Fig. 8, A–D) in either the raw data or when normalized to GAPDH expression. This, in conjunction with the decreased tghFST315 expression compared with the wild-type mFst315 isoform, indicates that the natural human promoter driving FST expression has limited ability to drive expression.

Figure 8.

FST Isoform mRNA Expression at 7 Weeks

A, Lung; B, kidney; C, skin; D, skeletal muscle. Wild-type mouse Fst315 mRNA expression was 10.0-fold higher than mouse Fst288 mRNA in the lung (P < 0.001), 7.1-fold higher in kidney (P < 0.001), 12.6-fold higher in muscle (P < 0.001), and 5.0-fold higher in skin (P < 0.001). Wild-type mFst288 mRNA expression in the kidney was significantly higher than mFst288 mRNA in lung, skin, and skeletal muscle (P < 0.05), whereas skin mFst288 expression was significantly lower than muscle (P < 0.05) but not different from lung. However, mFst315 isoform expression in skin was significantly lower (P < 0.05) than mFst315 expression in other tissues. Analysis of the measured quantities of lung tghFST315 mRNA (A) revealed a more than 1000-fold decrease in transgene expression (P < 0.001). Analysis of mRNA expression normalized to GAPDH revealed tghFST315 mRNA expression was 886-fold lower in the kidney (B) (P < 0.001), 277-fold lower in the skin (C) (P < 0.001), and more than 1000-fold lower in muscle (D) (P < 0.001), all comparisons being with wild-type mFst315 expression. Groups with different letters are statistically significantly different. Data are shown as mean ± sem.

DISCUSSION

Our results provide further insights into the complex biology of FST through selective expression of the human FST isoforms under the control of endogenous promoters contained within the substantial upstream and downstream sequence in the engineered construct. We have clearly demonstrated that the perinatal lethal phenotype of the FST null (Fst^−/−) mouse can be rescued by introducing the circulating isoform of FST, FST315 (tghFST315), under this endogenous control. In contrast, the introduction of the tissue-bound form of FST, FST288 (tghFST288), under the same endogenous control, was able to extend the lifespan of the Fst^−/− mice by only 10–20 h. The failure of the transgene expressing hFST288 to rescue the neonatal lethal outcome in the Fst^−/− mice may be due to inadequate expression of hFST288 or alternatively may indicate the need for the FST315 isoform. Levels of tghFST288 mRNA expression appeared to be similar to wild-type mice of the same age, and Western blot data indicate that the tghFST288 protein is being transcribed; however, additional quantitative studies of expression at the protein level are not possible at this time because there are no assays specifically designed to measure each isoform. These data point to differing functional activities of these isoforms that became clearer from the studies of the postnatal tghFST315 mice.

The perinatal rescue of the Fst^−/− mice by hFST315 is attributed to the normal respiratory function of the tghFST315 mice, supported by the normal lung weights and histology of these rescued mice. The presence of the FST315 isoform would appear to be critical to the development of the normal lung because, in the absence of this isoform, the lung weights were markedly lower, and there are clearly demonstrable changes in the Fst^−/− and tghFST288 lung histology that would compromise gaseous transfer. Given the involvement of activin A in branching morphogenesis of other organs (15,16,17), and that activin A and FST are both present in mouse lung tissue (18), it is reasonable to propose that activin A and FST could be critical to this process in the lung. Furthermore, the original report of the Fst^−/− mouse phenotype raised the possibility that the perinatal respiratory failure could have been due to the poorly developed diaphragmatic and intercostal musculature (13). In this context, the diaphragmatic weights in the tghFST315 mice at 7 wk were not significantly different from wild type, raising the possibility that improved muscle mass may also have contributed to the improved lung function. However, because FST also binds myostatin (7), the decreased muscle mass of diaphragm could be secondary to increased myostatin activity.

The perinatal phenotype of the tghFST315 mice also demonstrated that the skin and whisker patterns approached those of the wild type. This suggests that the FST315 isoform expression is critical to the morphogenesis of these tissues in keeping with the observation that mRNA for mFst315 was more abundantly produced in these tissues compared with mFst288. It is of note that the weights of the liver and spleen were reduced at birth in the Fst^−/− and tghFST315 mice. The decrease in liver weight is consistent with a decrease in hepatocyte size and increased cell degeneration. These features may reflect the known action of activin A on hepatocytes where it causes apoptosis, an effect reversed by FST (19,20,21). Because hepatic weights were restored at 7 wk in the tghFST315 mice, the difference in weight at birth may reflect the lower expression of tghFST315 mRNA. Alternatively, because the tghFST315 mice lack FST288, this tissue-bound form may play a role during fetal development, a concept requiring more studies.

The existence of defects postnatally in tghFST315 mice indicates only a partial rescue. There is a period of transient growth retardation that affects both body length and body weight, the cause of which remains unknown because the tghFST315 mice appear to suckle normally. However, whatever the nature of the disturbed physiology, this does not persist beyond 7 wk of age.

In contrast, tail development in the tghFST315 mice during the first 3 wk, characterized by reddening and the development of blackened tips, is abnormal with all changes consistent with impaired vascular development. This temporally limited abnormality results in a permanent shortening of the tails in the tghFST315 mice, and the vascular abnormality has a parallel in the reproductive phenotype discussed below.

There are several likely causes of the infertility in female tghFST315 mice. First, we noted the gross abnormalities in the reproductive tract in these mice as shown by the shorter uterine horns and vagina. This observation raises the potential that the activin FST system can influence Mullerian duct development as well as fetal ovarian development as shown by Yao and colleagues (22). Second, there is profound follicular depletion that leads to premature ovarian failure because the older mice fail to display estrous behavior. The degree of follicular development is consistent with some behavioral estrus; however, the mice are unable to develop corpora lutea, thus lacking the progesterone that prepares the uterus for implantation. Third, there is evidence of an inflammatory response in the uterus that is exacerbated by mating.

Our data in the tghFST315 mice show excessive leukocyte infiltration of the uterus that is particularly evident after mating. Mating in normal mice induces an inflammatory response in the uterine mucosa, predominantly due to high molecular weight seminal vesicular proteins (23,24,25). However, normally, this is transient, and the leukocyte response diminishes by the time of implantation in contrast to our observations in the tghFST315 mice (24,25). This could result for two reasons or a combination of both. First, this excessive inflammatory response in the tghFST315 mice may represent a response to the actions of activin A, which is known to be proinflammatory and is released within 50 min of an inflammatory stimulus that is incompletely modulated by the levels of FST315 or may require the presence of both isoforms (26). Both activin A and FST mRNA and protein have been identified in the endometrium (17,27,28), but the isoform of FST was not specified in these studies. No studies to date have monitored the response of activin A in the circulation or in uterine fluid or tissue during the normal estrous cycle or in response to mating. Such studies are clearly necessary to assist in the further interpretation of our data.

Second, the distribution of leukocytes in the normal mouse uterus has been shown to change during the estrous cycle in an estrogen-dependent manner (29). Using an ovariectomized mouse model, estrogen has been shown to induce an inflammatory response in the uterus, whereas progesterone antagonizes this action (30). Mice lacking a progesterone receptor show similarities to our tghFST315 mice, with significant leukocytic infiltration in both the lumen and endometrium of the uterus (31). In our tghFST315 mice, the presence of cornified epithelial cells in each daily vaginal smear suggests constant estrogen production. Because tghFST315 mice do not develop corpus luteum, they also will not secrete progesterone in the luteal phase or after mating. Consequently, it is possible that the inflammation seen in the tghFST315 uterus is due to unopposed estrogen action. These mice would be unable to antagonize the estrogen-induced inflammation during the estrous cycle or after mating (23,24,25). These data indicate that more studies are warranted to examine the role of the activin A-FST proteins and the manner in which they interact with estradiol and progesterone in managing the uterine response to mating.

Curiously, no copulatory plug was found in the older nonvirgin tghFST315 mice despite the presence of sperm in the uterus indicating that mating had occurred. It has been suggested that the variations in leukocyte number in the vagina may affect the length of time the plug remains in the vaginal tract (32). The augmented inflammatory infiltrate seen in the uteri, oviducts, and surrounding adipose tissue of the tghFST315 mice may have caused the copulatory plug to disintegrate before the animals were examined. It is well recognized that liquefaction of cells and tissues accompanies a severe inflammatory reaction.

In the Fst^−/− mice, there are markedly reduced numbers of primordial ovarian follicles at birth (22). This was supported by the profound decrease, supported by quantitative data, in the number of ovarian follicles noted in the tghFST315 mice at both 7 wk and in the older mice between 6 and 12 months. It is unclear why the hFST315 isoform was unable to restore the follicular ovarian follicular population in the Fst^−/− mice. We show that the transgene is expressed in the ovaries of the tghFST315 mice at both the mRNA and protein levels. However, detailed quantitative studies of this expression were not possible due to the unavailability of isoform-specific assays and the difficulty of matching the stage of the estrous cycle in the tghFST315 and wild-type mice. Furthermore, expression of FST during fetal ovarian development is required for normal ovarian development (22) and may be controlled by promoter sequences that were not included in our construct. Also, it is possible that the site of insertion of the transgene and the number of copies may have produced a pattern of expression that was distinct from the wild type. Finally, the absence of the FST288 isoform may be critical for normal ovarian morphogenesis and corpus luteum formation and is absent in the tghFST315 mice.

The ovarian phenotype identified is consistent with the observations of ovarian follicular development in the granulosa cell conditional FST knockout (14), because those female mice were subfertile, but some did achieve small litters, clearly indicating the presence of a functional corpus luteum. This contrasts with the absence of corpora lutea in the ovaries of the tghFST315 mice.

The failure of tghFST315 ovaries to form a corpus luteum was not due to a retained oocyte because the large antral cystic follicles, on serial sectioning, did not contain a retained ovum, which suggests that ovulation took place, but reorganization of the ovulatory follicle into a corpus luteum did not. Normal development of the corpus luteum requires migration of the theca interna cells, accompanied by new capillaries, to join with the remaining granulosa cells of the follicle. Given that the secondary ovarian follicles of tghFST315 females have a normal theca interna, the inability to form a corpus luteum is likely to result from the failure of angiogenesis required to form the corpus luteum. This finding may have a similar etiology to the impaired tail growth in these tghFST315 mice, also implicating abnormal vascular development.

It has been demonstrated that both activin A and FST are produced in the vascular endothelium (33,34). However, the presence of corpora lutea in the conditional granulosa cell knockout seems to contradict this hypothesis of an angiogenic basis for the failure of corpus luteum formation in the tghFST315 mice. However, it is important to note that both forms of FST would still be produced in the vascular endothelium of mice with a conditional knockout of the FST gene in granulosa cells. This availability of both FST isoforms from the vascular endothelium could support the angiogenesis required for the formation of the corpus luteum in the granulosa cell conditional knockout. Experiments using mice with a conditional knockout of the FST gene in endothelial cells could confirm this proposal.

It is well recognized that activin stimulates granulosa cell differentiation, the expression of FSH receptors, and aromatase activity, whereas FST, by neutralizing these actions, stimulates progesterone secretion (for review, see Ref. 35). Our data provide new insights into the effects of FST in the development of the corpus luteum. FST has been identified in blood vessels of many tissues (33,34), including during wound healing (36). However, information about specific isoform expression is unavailable apart from the localization of mRNA for both FST315 and -288 to granulosa cells (37). These data would support the hypothesis that the failure of angiogenesis may be due to the absence of the tissue-bound form of FST, FST288, in the tghFST315 mice.

Several reports clearly indicate that activin A and FST are involved in the control of angiogenesis and endothelial biology. Two conflicting hypotheses have emerged. One suggests that activin A inhibits angiogenesis and that FST is proangiogenic (34,38,39,40,41,42). The other proposes that activin A supports angiogenesis and is essential for the branching of capillaries induced by vascular endothelial growth factor (VEGF) (43). In a murine model of corneal damage, support for a proangiogenic role for activin resulted from the observation of an increased area of neovascularization and enhanced expression of VEGF stimulated by activin, but no data were reported concerning FST (43). In vitro, activin was equivalent to VEGF in stimulating markedly enhanced capillary tubulogenesis of bovine aortic endothelial cells, and when tubulogenesis was induced by VEGF (44), activin A increased the expression of VEGF and its receptors, Flt-1 and Flk-1. Both these actions of activin A were inhibited by FST, although the FST isoform used was not specified. Similar actions of activin A on fibroblast growth factor-2 (FGF2)-induced tubulogenesis of bovine aortic endothelial cells were recently reported and could also be inhibited by FST but the isoform was not specified (45). In contrast, activin inhibited endothelial cell proliferation and was expressed constitutively in migrating and resting endothelial cells (34,40). FST was expressed in migrating endothelial cells but decreased at growth arrest and was inducible by basic FGF, stimulating a burst of angiogenesis in a rabbit cornea model (34). Furthermore, both in vitro and in vivo activin inhibited the basic FGF2-induced sprouting angiogenesis (39). Several studies have also shown that activin can inhibit angiogenesis in neuroblastomas both in vitro and in vivo and can induce hemorrhage (38,41), with one study indicating that the favorable response was due to both antiangiogenic and antimitotic actions (41). Others have shown that high expression of activin in neuroblastomas can predict a more favorable outcome (42). Our data from the use of tghFST315 mice, which express the human FST315 transgene as both mRNA and protein in the ovary strongly support the view that activin A, unopposed by the absence of FST288, the tissue-bound form of FST, inhibits angiogenesis as shown by the abnormalities of perinatal tail growth and the failure of the corpus luteum to form. Alternatively, one would have to propose that the level of expression of the FST315 transgene is insufficient to support corpus luteum formation despite maintaining adequate ovarian follicle populations to allow follicular maturation, estrus, and mating to occur.

It is possible that the lower levels of FST315 in the endometrium may be responsible for the augmented uterine inflammation together with the absence of the FST288 isoform. It is clear that more studies are necessary to explore these possibilities, and it will be essential that both forms of FST should be assessed for their action on the complex process of angiogenesis in other models. It should be emphasized that these novel actions of FST and its isoforms would not have been revealed without the successful rescue of the perinatal lethal phenotype of the Fst^−/− mice achieved in our study. As such, our study provides additional data to implicate FST and activin in angiogenesis in vivo and expands the information linking activin to the control of the inflammatory process.

MATERIALS AND METHODS

Construction of the FST Transgenes

The PAC-FS vector (a generous gift from Dr. Panos Ioannou, Murdoch Institute, Melbourne, Australia), which included the human FST locus plus approximately 45 kb upstream and 45 kb downstream sequences, was genetically engineered to express specific FST isoforms. Exons 5, 6a, and 6b and exons 5 and 6b (Fig. 9A) were amplified by RT-PCR from human testis total RNA (BD Biosciences Clontech, Palo Alto, CA). Using enzyme restriction mapping, these PCR fragments were separately cloned into the human FST gene to replace exons 5, 6a, and 6b and intron 5 (Fig. 9B). This fragment then replaced the human FST gene in the PAC-FS vector.

Figure 9.

Schematic of the Generation of the Constructs

A, The structure of the FST gene. Alternative splicing occurs at the 3′-terminal of the gene between exon 5 and exon 6. The splicing out of intron 5, generating the stop codon, TGA, immediately after the last amino acid of exon 5, leads to the termination of the coding sequence for a precursor of 317 amino acids, the COOH-terminal truncated form. On the other hand, when exon 6a is spliced out together with intron 5, the stop codon, TAA, is in exon 6b, and a precursor of 344 amino acids is generated. Cleavage of the signal peptide (29 amino acids) generates the mature FST isoforms of 288 (FST288) and 315 (FST315). B, Exons 5, 6a, and 6b (resulting in FST288 protein) or exons 5 and 6b (FST315) were amplified from human testis cDNA. White, Exon 5; gray, exon 6a; black, exon 6b. Each of these fragments was cloned into a construct containing exons 2–5 of the human FST gene using the restriction enzymes EcoRV and XbaI. This modified human FST gene was cloned into the PAC-FST construct to replace the FST gene. The modified PAC-FST construct was linearized with BsiWI before pronuclear microinjection into FVB ovum.

Generation and Screening of the Transgenic Mice

Modified constructs were microinjected into the pronucleus of FVB/N ova by IngenKO Pty. Ltd. (Melbourne, Australia). Multiple lines were generated, and we have focused on a single line for each transgene, which is representative of all of the other lines. FVB/N mice carrying the construct were crossed with Fst^+/− mice (13) to create 129/C57BL6/FVB/N mice carrying the construct and heterozygous for the endogenous mouse FST (F1). F1 mice were backcrossed with Fst^+/− to create mice that carried the modified human construct but did not carry the endogenous mouse FST gene. Mice carrying only human FST288-specific construct were designated tghFST288 whereas mice carrying human FST315-specific construct were designated tghFST315. Transgenic mice were genotyped by PCR using primers specific for the mouse FST gene, the NEO replacement cassette (which is specific for the knockout construct used to target endogenous FST gene) (13), and the human transgene (Table 3). Animals were generated under the ethics approval by Monash Medical Centre Committee A and housed in standard conditions with a 12-h light (0800–2000 h), 12-h dark cycle, in accordance with The Australian and New Zealand Council for the Care of Animals in Research and Teaching guidelines.

Table 3.

Primer Sequences

Experiment	Target	Primer Position	Primer Sequence 5′ to 3′
Genotyping	Mouse Fst gene	Upstream	TGTGCCTCTTTCCAACTCCT
		Downstream	ATCTATCGCCCTTGGGTCTT
	NEO cassette	Upstream	TGCTGACCTGCTGGATTACA
		Downstream	CTGCATTGTTTTGCCAGTGT
	Human FST transgene	Upstream	TCCTCAGGTGTGCTACTGGA
		Downstream	CAAAGGCTATGTGAACACTGAA
Real-time PCR	Mouse Fst	Upstream	CCTCCTGCTGCTGCTACTCT
		Downstream	CTAGTTCCGGCTGCTCTTTG
	Mouse Fst288	Upstream	AGAGGTCGCTGCTCTCTCTCTG
		Downstream	CCGAGATGGAGTTGCAAGT
	Human FST transgene	Upstream	AAGACCGAACTGAGCAAGGA
		Downstream	GACCCTTCCAGGTGATGTTG
	Human FST288transgene	Upstream	TCCCTCTGTGATGAGCTGTG
		Downstream	TTTGTTTTTGGCATCTGCTG
	Mouse Gapdh	Upstream	TACTGGCATCTTCACCACCA
		Downstream	GTGAGCTTCCCATTCAGCTC

Annealing temperature for all primers used in genotyping was 57 C. Annealing temperature for all primers used in real-time PCR was 60 C. 

Tissue Collection and Processing

Organs were dissected and processed for either frozen or fixed tissue. For RNA and protein extraction, fresh tissue was collected from newborn (n = 3) and 7-wk-old (n = 5) animals, snap frozen immediately in an isopropanol/dry ice slurry, and then stored at −80 C or stored in RNAlater Stabilization Buffer (QIAGEN, Hilden, Germany) until processing. For fixed tissue, newborn animals were decapitated and fixed whole in fresh Bouin’s fixative. Organs were dissected 24–36 h later. Tissues from 7-wk-old or adult mated animals were dissected and fixed in Bouin’s fixative for an appropriate amount of time. All tissues, except female reproductive tracts, were weighed before histological processing. Sections of the Bouin’s fixed tissues were stained with hematoxylin and eosin (H&E) or periodic acid-Schiff for histological analysis by light microscopy.

Ovarian follicle populations, corpora lutea, and cystic structures were counted in every 10th periodic acid-Schiff-stained 3-μm serial section of each ovary. These data were summed to give the number of observed ovarian structures per ovary. Data are presented as the mean ± sem number of each structure seen per ovary. It should be noted that because of the methodology, large structures may have been counted more than once, and will therefore be overrepresented in the results, and small structures may be underrepresented.

Morphological Analysis of Transgenic Mice

Body weight, body length, and tail length of mice were measured weekly until 7 wk of age and then approximately biweekly until 6 months of age. The body length (CRL) was measured from the tip of the nose to the end of the fur line on the tail. Tail length was measured from the end of the fur growth to the tip of the tail. To perform skeletal evaluation, the newborn mice were decapitated, and the skeletons were stained with Alizarin Red (Chroma IF583; Sigma Chemical Co., St. Louis, MO) as described previously (46). Ribs were counted under a dissection microscope to allow detection of partial 13th ribs, which may not be visible by the naked eye. A crude assessment of visual acuity was made based on the ability of mice to extend their forelimbs toward a wire grid (cage lid) using a visual placing test as described in the SHIRPA protocols for mouse phenotypic analysis (47). The detailed protocol is available at http://www.mgu.har.mrc.ac.uk/facilities/mutagenesis/mutabase/shirpa_1.html#table.

Assessment of Reproductive Phenotype

Daily vaginal smears were performed on 7- to 8-wk-old mice (wild-type, n = 5; tghFST315, n = 4) from 1600–1700 h using a method modified from (48). A glass Pasteur pipette was flamed to smooth the opening used the collect the cells, which were then spread onto a glass slide coated with 3-amino propyltriethoxy silane (Sigma Aldrich, St. Louis, MO). The smears were allowed to air dry and then stained with H&E. From 6 wk to 12 months, males and females from the wild-type and transgenic lines were mated and the outcomes observed. The tghFST315 males and females were allowed to mate with proven fertile mice, and all females were assessed daily by experienced staff for plug formation using vaginal probes as well as for the successful completion of pregnancy.

Female mice were allowed to mate at 7–8 wk of age and were killed at 4 dpc. Their uterine tubes and uteri were flushed with PBS, and the eluate was examined for blastocysts. Sections from the ovaries, oviducts, and uteri were examined by light microscopy.

RNA Isolation and Quantitative RT-PCR

RNA was extracted from tissues using TRIzol (Invitrogen, Carlsbad, CA) or RNAeasy Mini kit (QIAGEN) from RNAlater-stabilized tissues (QIAGEN) in accordance with the manufacturer’s instructions. RNA was DNase treated with TURBO DNase (Ambion, Austin, TX). cDNA was made from 1 μg RNA using Superscript III (Invitrogen). mRNA expression was quantitated with SYBR Green (Roche Diagnostic Co., Mannheim, Germany) in a Roche LightCycler system. RT-PCR bands in adult ovary were generated using the human FST transgene primer pair (Table 3).

In a wild-type mouse, endogenous mFst315 mRNA is difficult to determine directly in the presence of endogenous mFst288. Thus, primers were designed to specifically detect total Fst expression and Fst288 expression; Fst315 expression was estimated by measuring the total Fst mRNA expression and subtracting the Fst288 component. Similarly, primers were designed to detect transgenic human FST and hFST288 expression. Primer sequences for all targets can be found in supplemental data. Tissue FST mRNA values were calculated by using a PCR product standard curve consisting of serial dilutions from 1 × 10⁻² ng to 1 × 10⁻⁹ ng. For each PCR product, including GAPDH, copy number was calculated using the following formula: grams of cDNA PCR product/molecular mass of PCR product × N_A (6.02214 × 10²³). Fst315 copy numbers were then calculated as described. Data are presented as FST isoform copy number normalized to GAPDH copy number. However, due to unexplained discrepancies in lung GAPDH expression, FST expression in the lung is presented as FST isoform copy number per microgram of RNA converted and not relative to GAPDH expression.

Western Blot Analysis

Qualitative Western blot analysis was performed as described previously (49) on adult ovary extracts and mouse liver extracts from d-0 and 7-wk-old animals. Total protein amounts loaded in each gel lane were equal as determined by DC protein assay (Bio-Rad, Hercules, CA). Western blot analysis was performed using a monoclonal antibody (5B5) or polyclonal rabbit antibody JM19 (50). The antibodies were raised against full-length human recombinant FST288 and detect both FST288 and FST315.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 2.01. Where data were collected from both the left and right sides of one animal, e.g. analysis of the ophthalmic defect, data were averaged per animal before being statistically analyzed. Data were analyzed by one-way ANOVA with Tukey’s post hoc tests or by unpaired t tests. Results were considered statistically significant at P < 0.05. Groups that are represented by different letters are significantly different (P < 0.05). Data are presented as mean ± sd or sem.

Supplemental Data

Supplemental data include four figures showing, in detail, the construction of the construct used in creating the transgenic mice and two figures showing vaginal smears from wild-type and tghFST315 mice (published as supplemental data).