Placenta Defects and Embryonic Lethality Resulting from Disruption of Mouse Hydroxysteroid (17-β) Dehydrogenase 2 Gene

Abstract

Hydroxysteroid (17-β) dehydrogenase 2 (HSD17B2) is a member of aldo-keto reductase superfamily, known to catalyze the inactivation of 17β-hydroxysteroids to less active 17-keto forms and catalyze the conversion of 20α-hydroxyprogesterone to progesterone in vitro. To examine the role of HSD17B2 in vivo, we generated mice deficient in Hsd17b2 [HSD17B2 knockout (KO)] by a targeted gene disruption in embryonic stem cells. From the homozygous mice carrying the disrupted Hsd17b2, 70% showed embryonic lethality appearing at the age of embryonic d 11.5 onward. The embryonic lethality was associated with reduced placental size measured at embryonic d 17.5. The HSD17B2KO mice placentas presented with structural abnormalities in all three major layers: the decidua, spongiotrophoblast, and labyrinth. Most notable was the disruption of the spongiotrophoblast and labyrinthine layers, together with liquid-filled cysts in the junctional region and the basal layer. Treatments with an antiestrogen or progesterone did not rescue the embryonic lethality or the placenta defect in the homozygous mice. In hybrid background used, 24% of HSD17B2KO mice survived through the fetal period but were born growth retarded and displayed a phenotype in the brain with enlargement of ventricles, abnormal laminar organization, and increased cellular density in the cortex. Furthermore, the HSD17B2KO mice had unilateral renal degeneration, the affected kidney frequently appearing as a fluid-filled sac. Our results provide evidence for a role for HSD17B2 enzyme in the cellular organization of the mouse placenta.

HYDROXYSTEROID (17-β) DEHYDROGENASES (HSD17Bs) catalyze the conversion between 17-keto and 17β-hydroxysteroids. Both androgens and estrogens are highly active in the 17β-hydroxy configuration, whereas the affinity of the 17-keto derivatives for the corresponding receptors is markedly lower. Thus, the enzymes regulate the balance between biologically highly active and less active sex steroids, and they are considered to have an important role in regulating the ligand availability in various sex steroid-dependent tissues (1, 2). HSD17B enzymes belong to two families: aldo-keto reductases and keto steroid reductases (3). To date, in these two enzyme families 12 different enzymes with HSD17B activity have been characterized in mammals. The enzymes possess different substrate and cofactor specificities, different tissue distribution, and different preferences for the direction of the reaction (1, 4, 5). In addition to catalyzing reaction between sex steroids, some HSD17B enzymes have been shown to be involved in other metabolic pathways as well. For example, HSD17B4 is involved in β-oxidation of fatty acids (6) and HSD17B10 in oxidation of branched and straight fatty acids (7).

HSD17B1 and HSD17B3 have been shown to be involved in sex steroid biosynthesis in the mammalian gonads, HSD17B1 being expressed in the ovaries (8, 9, 10) and HSD17B3 in the testis (11). In line with their expression particularly in the gonads, the enzymes have shown to predominantly catalyze the reductive reaction from 17-keto to 17β-hydroxy forms in cultured cells, and this activity is essential for estradiol and testosterone biosynthesis. However, HSD17B1 is also found in several extragonadal tissues in human, e.g. the breast (12) and the endometrium (13). Therefore, it has been hypothesized that the enzyme could activate the circulating low-active 17-keto steroids locally in the target tissues.

Interestingly, several peripheral tissues, including sex steroid target tissues (1), express also HSD17B enzymes that possess mainly oxidative activity (converting the 17-hydroxy to 17-keto form), and several HSD17B enzymes with opposite activities may be expressed in the same tissues and cells at the same time (1, 5, 14). It is hypothesized, therefore, that HSD17B enzymes displaying opposite reaction directions could regulate temporal activation and inactivation of sex steroids, and therefore, regulate the ligand availability for steroid receptors (3).

HSD17B2 is one of the family members with oxidative activity, capable for catalyzing in vitro the conversion of estradiol, testosterone, and dihydrotestosterone to their less-active 17-keto forms, estrone, androstenedione, and 5α-androstanedione, respectively. However, the enzyme also possesses 20α-HSD activity in vitro, thereby, activating 20α-hydroxyprogesterone to progesterone [Michaelis-Menten constant (K_m) for estradiol, 0.21 ± 0.04; K_m for progesterone, 0.71 ± 0.06 (15, 16)]. The enzyme is widely expressed in the peripheral tissues both in human and in rodents. Studies in mice have revealed that the enzyme is especially expressed throughout the gastrointestinal tract (17, 18) and the liver (19), both in fetal and adult mice. Studies in human tissues have shown that the enzyme is also expressed in various estrogen and androgen target tissues such as the endometrium (20), the placenta (17), and the prostate (21), thus, putatively being involved in the hormonal regulation of these tissues (3). Despite several studies carried out to characterize the tissue distribution of the enzyme expression and to analyze the enzyme properties in vitro, little is known about the physiological role of HSD17B2 in vivo. We have recently generated mice ubiquitously expressing the human HSD17B2 (22), and in the present study we generated mice with inactivated Hsd17b2 gene. The data reveal a novel role for HSD17B2 independent on its action on sex steroids.

RESULTS

Deletion of Hydroxysteroid (17-β) Dehydrogenase 2 Is Associated with Embryonic Lethality and Growth Retardation

The genetic manipulation of Hsd17b2 was designed to prevent the expression of the exons 4–6, resulting in lack of active enzyme expression. The gene disruption was confirmed by Southern blot (Fig. 1C) and PCR-based analyses (data not shown). Absence of the full-length transcript in the liver [highly expressing the Hsd17b2 in wild-type (WT) mice] and in the kidney was demonstrated by Northern blot analysis (Fig. 1D). We then genotyped 178 mice at the age of 2 wk from heterozygous intercrosses in hybrid genetic backround (C57Bl/129), and the number of HSD17B2 knockout (KO) mice identified was lower (6%, 11 of 178) than expected (25%). The result indicated that the Hsd17b2^−/− allele produced a recessive phenotype, resulting into lethality in close to 76% of the HSD17B2KO mice during embryonic development or during the first postnatal days. To determine the time of embryonic death, pregnant mice were killed and 250 embryos were collected at different embryonic stages. At the blastocyst stage (n = 35) the percentages of homozygous, HO −/−, heterozygous, HE +/−, and WT +/+ genotypes were 26%, 48%, and 26%, respectively, which is in agreement with a Mendelian transmission of the recombinant allele (Table 1). However, the percentage of live HSD17B2KO embryos decreased to 14% at embryonic d 11.5 (E11.5), to 10% at E15.5, and to 6% at E17.5 (Fig. 2A and Table 1). The results indicate a loss of HSD17B2KO embryos at E9–E12 onward, and thus markedly reduced number of HSD17B2KO embryos survived until birth. There was no evidence for gender influence in survival of the HSD17B2KO mice. Furthermore, the number of dead embryos and reabsorbed deciduas at the different stages of embryonic development were inversely related to the percentage of living HSD17B2KO embryos (Fig. 2, B and C). At every developmental stage studied, HSD17B2KO embryos were slightly growth retarded as compared with their WT or HE littermates.

Fig. 1.

Generation of the HSD17B2KO Mice

Panel A, Strategy used to disrupt mouse Hsd17b2 gene by homologous recombination in ES cells. Structures of the WT Hsd17b2 gene, the targeting vector (pACYC177-NEO-HSD17B2), and the targeted locus after homologous recombination are presented. The exons 4 and 5 are presented as black boxes and introns 3 and 4 with a thin line. The mutated Hsd17b2 allele was confirmed by PCR with primers (genom3 and NEO1an) indicated by arrows. Genotyping was carried out with PCR primers ex4s, ex4a, and NEOs1, to simultaneously detect both the WT and KO allele. Presence of the mutated allele was confirmed by Southern blot analysis with a 5′-probe and NEO probe, indicated by the solid black lines. Only restriction sites relevant for the cloning are shown: XmnI and BamHI. Panel B, Schematic picture of ET-cloning used to generate the targeting vector. Recombinogenic Escherichia coli bacteria (indicated as large ovals) were used as host, for ET-recombination steps. Linear DNA used in subcloning of the targeted arms was generated by PCR. It contained a replication origin (ori), a selectable marker (Amp), and flanking short homology regions (A and B). The homology regions were chosen to define the boundaries of the DNA fragment of interest of the intact circular BAC clone. The linear targeting fragment was electroporated into ET-competent host bacteria, containing the plasmid encoding for recombination enzymes (E), to create a recombinant plasmid pACYC177-HSD17B2. In the second ET-cloning step the linear targeting fragment contained a neomycin resistance cassette and flanking short homology regions (C and D; see panel A). The DNA fragment was electroporated into the ET-competent host bacteria together with pACYC177-HSD17B2 plasmid. This created the final targeting vector pACYC177-HSD17B2-NEO (see Materials and Methods for more detailed description). Panel C, Southern blot analysis with the 5′-probe of XmnI-digested genomic DNA from WT ES clones (lanes 1 and 2), targeted heterozygous ES clone 11D4 (lane 3), and BAC clone used as control (lane 4). Southern blot analyses of the targeted ES cells were confirmed with DNA obtained from mice at the F2 generation; WT (lane 6), HE (lane 7), and HO (lane 8). The expected sizes of the XmnI fragments that hybridize with the probe on Southern analysis are indicated for WT (6.3 kb) and mutant Hsd17b2 (8.2 kb) alleles. Panel D, Northern blot analysis of RNA from the kidney and liver of WT and homozygous HSD17B2KO animals with the Hsd17b2-cDNA probe. Ex 4, Exon 4; Amp, ampicillin; hmA–hmD, homology regions A–D.

Table 1.

Genotype Distribution of the Mice from HE Breedings

	No. of Mice	Genotype of Mice (%)
				+/+	+/−	−/−
3.5 dpc	38	27	55	23
11.5 dpc	72	24	58	11
15.5 dpc	58	28	62	8
16.5–18.5 dpc	79	31	63	6
Postnatal, 1d	203	38	56	6

dpc, Days post coitus.

Fig. 2.

Fetal Loss in HSD17B2KO Mice

A, Percentage of live HSD17B2KO mice. Division between embryonic and postnatal days is indicated by a dotted line. Eight to ten pregnant female mice were used for each age group studied. Total number of embryos genotyped was 247. B, Uterus of WT and (C) of HE mouse at E17.5 with fetuses inside. Two of the dead fetuses in the HE mouse (white arrows) were found to be HO (HSD17B2KO) in subsequent genotyping.

At birth, HSD17B2KO mice were distinguishable from their littermates by their smaller size. The mean body weight of HO newborns was 23% smaller as compared with their WT or HE littermates (body weight: WT, 1.3 ± 0.14 g, n = 34; HO, 1.0 ± 0.14 g, n = 9; P < 0.002; Fig. 3, A and B). Organ weights of the kidney, spleen, adrenal, and brain were measured but a significant reduction in weight was found only for the kidney (WT, 6.9 mg ± 0.99; HO, 4.6 ± 0.29 mg; P < 0.001) and brain (WT, 90.9 ± 5.48 mg; HO, 71.1 ± 8.56 mg; P < 0.001, Fig. 3, D and E), However, when compared with body weight, the organ weights in the KO and WT mice were not statistically different (data not shown). All HSD17B2KO mice that survived until birth died or had to be killed before the age of 4 months, due to the enlarged head and severe hydrocephalus leading to sleepiness, problems with balance, and poor coordination.

Fig. 3.

Growth Retardation and Reduced Brain and Kidney Weights in Newborn HSD17B2KO Mice

A, Body weights of WT (+/+) and HSD17B2KO (−/−) mice. B, Appearance of the WT and HSD17B2KO mice at the age of 1 d. Brain (C) and kidney (D) weight of WT and HSD17B2KO mice at the age of 1 d. ***, P < 0.001; **, P < 0.002; n = 9–30 per group; bars indicate mean ± sd.

Placenta Defects in HSD17B2KO Mice

In normal mouse placenta, HSD17B2 expression is found throughout the pregnancy. Thus, it was interesting to observe that the placentas of HSD17B2KO mice at age E17.5 were smaller in size as compared with the HE or WT mice placentas (WT, 77.4 ± 8.12 mg; HE, 78.2 ± 10.40 mg; HO, 66.9 ± 7.11 mg, P < 0.05). At the age of E12.5 the HSD17B2KO and WT placentas were identical in their histological appearance. However, at late gestation (E17.5), the HSD17B2KO placentas showed variable structural abnormalities in all three major layers: the decidua, spongiotrophoblast, and labyrinth (Fig. 4, A, B, D, and G). These defects included liquid-filled cysts in the junctional region and basal layer (Fig. 4, C, F, I, J, and M), increased volume of the spongiotrophoblast layer (Fig. 4, H and I) and a disorganized labyrinthine layer (Fig. 4, J–M). Extensive trophoblast aggregates were present in the labyrinthine region (Fig. 4, H–K), and the junction between the spongiotrophoblast and labyrinthine layer was disrupted. Also the basal layer, which includes spongiotrophoblast, glycogen cells and trophoblast giant cells, was increased in size and unorganized in structure (Fig. 4, J–M).

Fig. 4.

Histological Examination of HSD17B2KO Mice Placentas at the Age of E17.5

A, D, and G, Hematoxylin and eosin-stained sections of placentas of the WT and (B, C, and E–M) HSD17B2KO littermates. A and D, The normal appearance of the labyrinthine zone (Lab), junction zone (Jun), and decidua (Dec) observed in WT placenta was disrupted in HSD17B2KO placenta. The junctional region, which includes spongiotrophoblast, glycogen cells, and trophoblast giant cells, was increased (white arrow, E and K) and unorganized (B, E, and H), and some of the HSD17B2KO placentas showed impaired formation of the labyrinth layer (B, E, H, K, and M). Excessive cysts formation (black arrows) was also detected in HSD17B2KO placentas. Scale bar in D–F is 500 μm; scale bar in G–M, 200 μm.

To further analyze the placental defects, we performed in situ hybridization and RT-PCR analysis for expression of several trophoblast markers (Fig. 5). In situ hybridization revealed the increased expression of a spongiotrophoblast-specific marker, Tpbpa (23)(Fig. 5A), which indicated an expansion of the spongiotrophoblast cell layer in the KO mice. The expression of PL-1, a, marker gene for trophoblast giant cells (24), was reduced. Furthermore, the expression of fetal placental labyrinth marker gene Esx1 (25) was decreased in HSD17B2KO placenta as compared with the WT (Fig. 5B). These results confirmed that both the basal and labyrinth layers were affected by the absence of functional HSD17B2.

Fig. 5.

Analyses of the Expression of Trophoblast Marker Genes in Placentas at E17.5

A, Serial sections of WT and HSD17B2KO placentas were probed with the antisense RNA probe for Tpbpa. The expression of the spongiotrophoblast marker (Tpbpa) indicated that the spongiotrophoblast layer was increased in HSD17B2KO placenta when compared with WT. B, The trophoblast marker genes were also analyzed by RT-PCR. The expression levels of PL-1 and Esx-1 were present in the HSD17B2KO placenta at E17.5 but were significantly decreased in the HSD17B2KO placenta at E17.5.

The data in vitro have suggested a role for HSD17B2 in the inactivation of sex steroids and in the conversion of 20α-dihydroprogesterone to progesterone. Accordingly, we found that intraplacental level of progesterone in HSD17B2KO mouse placenta at E17.5 was slightly elevated compared with the WT mice (WT, 0.407 ± 0.043 nm; HO, under 0.3 nm). Placental estradiol levels in both the KO and WT placentas were under detection limit (0.05 nm) of the assay used.

Furthermore, we treated the pregnant mice with ICI 182,780, a potent antiestrogen, or with progesterone from E7.5 to E17.5. However, no KO embryos with a rescued phenotype were observed (Table 2). Furthermore, the placentas from the rescue experiments appeared to be identical in morphology to those of untreated or placebo-treated mice at E17.5 (data not shown). Thus, the cause of embryonic death does not appear to be due to the excess of estradiol, or lack of progesterone.

Table 2.

Effect of Maternal Drug Treatment in Survival of 17.5 dpc Homozygous Embryos

Drug	Treatment	Genotype of Breeding Pair	No. of Embryos	Genotype of Living Embryos at 17.5 dpc (%) (+/+)/(+/−)/(−/−)
No treatment	No	WT	56	100/0/0
No treatment	No	HE	35	34/60/6
ICI1	1.5 mg/12 d	WT	72	100/0/0
ICI1	1.5 mg/12 d	HE	69	37/57/5
Progesterone	5 mg/12 d	WT	52	100/0/0
Progesterone	5 mg/12 d	HE	63	39/56/6
Placebo	No	WT	52	100/0/0
Placebo	No	HE	71	35/58/7

The number of pregnant females used for analyses was from five to eight. All data are presented as mean ±sd. dpc, Days post coitus.

¹ICI 182,780.

HSD17B2KO Mice Have Abnormalities in Their Brain Structure and Defects in the Metanephric Kidney

In histological analysis, newborn HSD17B2KO mice showed no obvious abnormalities in tissues other than kidneys and brain. Although the HSD17B2KO mice were significantly smaller, the bone and cartilage structure showed no obvious morphological abnormalities (data not shown). A severe hydrocephalus was evident at the end of fetal life in all HSD17B2KO mice. Sagittal sections revealed that the brains of HSD17B2KO mice were histologically abnormal, with enlarged lateral ventricle at one side (Fig. 6, A and B). The thalamus of some HSD17B2KO mice also appeared to be enlarged (Fig. 6B). Furthermore, in the developing cerebral cortex, an altered layering of cells was detected in HSD17B2KO mice together with abnormal laminar organization and increased cellular density [WT (Fig. 6, C and E) vs. HSD17B2KO (Fig. 6, D and F)]. Hyperplasia and disrupted cellular strand were evident in most cases in the occipital and parietal cortical areas. The number of small cells in all the cortical layers was increased but was most prominent in layers II–IV (Fig. 6, D and E). However, the increased cortical cell mass was not due to decrease in apoptosis, because there was no difference in number of apoptotic cells between WT and HSD17B2KO brains (data not shown).

Fig. 6.

Morphological Analysis of the Brain in HSD17B2KO Mice

A, 1-d-old WT mouse demonstrating normal brain organization. B, 1-d-old HSD17B2KO mice showing the enlargement of the ventricle (black arrow) and the thalamus (white arrow). C and E, Normal organization of cells in brain cortex in 1-d-old WT mouse. D and F, An altered layering of nerve cells was detected in the brains of 1-d-old HSD17B2KO mice (white arrows in panel D and black arrows in panel F). Black line in panel D shows the cortical layers II–IV). G, HSD17B2KO embryo at the age of E17.5 exhibited exencephaly (white arrow). H, Brains of HSD17B2KO mice at age of 7 wk. Enlargement of ventricles progressively increased during the postnatal age resulting in a severe hydrocephalus (white arrow).

In the HSD17B2KO mice born alive, the enlargement of ventricles progressively increased during the postnatal age, resulting in hydrocephalus with markedly enlarged bilateral ventricles covered with only a thin cortical mantel (Fig. 6H). The mouse model appears as a phenocopy of human hydrocephalus with several symptoms typical for this disorder, such as abnormal enlargement of the head and problems with balance. Interestingly, at the age of E15–E17.5, some of the HSD17B2KO mouse embryos exhibited exencephaly (Fig. 6G).

The HSD17B2KO mice displayed marked abnormalities in their kidneys. The newborn HSD17B2KO mice frequently had only one kidney, indicating unilateral degeneration of the whole kidney while the ureter was still detected (Fig. 7, A and B). The remnant tissues of the degenerated kidney were frequently liquid filled whereas the other remaining kidney was pale in color (Fig. 7, A and B). The weight of the remaining kidney of the newborn HSD17B2KO mice was also significantly reduced as compared with the WT littermates (P < 0.001, Fig. 3D). In 1-d-old HSD18B2KO mice the developing cortical layer, where the new tubules form, was normal whereas the tubules in the kidney medulla and the renal pelvis were enlarged (Fig. 7, C–F). The stromal cells at the developing medullary region had differentiated normally as judged by histological inspection but the cells were disorganized, which is likely due to the changes in the tubules and collecting duct (Fig. 7F).

Fig. 7.

Deficiency in Hsd17b2 Function Leads to Defects in the Metanephric Kidney

A, Urogenital block of WT mice at age of 1 d demonstrating the kidneys (Kid), ureters (Ur), ovaries (Ov), with their ducts, bladder (Bl), and the adrenals (Ad). B, Urogenital block of HSD17B2KO mice at the age of 1 d revealing unilateral renal degeneration with only a remnant ureter present (white arrow). Note that the HSD17B2KO kidney (black arrow) is paler as compared with the WT kidneys (A). C and E, 1-d-old WT mouse demonstrating normal organization of the cortical (Co) and medullary (Me) region composed of several glomerulae (Gl), segments of the nephrons and collecting ducts at the region close to renal pelvis (Re). D and F, HSD17BKO mice with enlargement of the tubular segments of the nephrons (black arrows) and disorganization of the developing collecting duct epithelium at the same stage (black star). Glomerulogenesis takes place normally.

DISCUSSION

In the present study, we describe the generation of KO mice for Hsd17b2 by disrupting the gene in embryonic stem (ES) cells. The data indicate that the HSD17B2 enzyme plays an essential role in proper cellular organization of mouse placenta, and in the hybrid genetic background the enzyme deficiency leads to embryonic lethality in most of the homozygous embryos. Hsd17b2 is expressed in several epithelial cell types in developing mouse embryo (17). On E11.5, the enzyme expression appears in the liver, the esophagus, and the intestine and is thereafter, highly expressed in these tissues throughout the embryonic development. At later embryonic stages (E15.5–E16.5), the mRNA has been also detected in epithelial cells of the stomach, the tongue, the oropharynx, and the nasopharynx (17). However, histologically these tissues are normal in the newborn HSD17B2KO mice.

Previous studies have shown that distribution of Hsd17b2 expression vary during mouse placental development (19). In the chorioallantoic placenta and in midgestation, the enzyme is expressed in trophoblast giant cells (19). At E12.5, the expression appears in the labyrinth region and toward the end of pregnancy the expression is strongest at this region (19). Thus, the first sign of Hsd17b2 expression in the placenta is detected simultaneously with the expression in the fetus (19). Interestingly, occurrence of the embryonic deaths in the HSD17B2KO mice agrees well with the shift in the Hsd17b2 expression from the maternal to the fetal part of placenta, and with the Hsd17b2 expression in the fetus. Expression of Hsd17b2 in the placenta and embryo, together with the oxidative HSD17B activity associated with the enzyme, has suggested a role for the enzyme in reducing the sex steroid exposure of the fetus (26). Estrogen excess has been shown to cause placental thrombosis and spontaneous fetal loss in estrogen sulfotransferase, Sult1e1, KO mice (27), and fetal death in mice lacking the Srd5a 1, the gene that encodes steroid 5α-reductase type 1 (28). Placentas of HSD17B2KO mouse display histological malformations but no signs of thrombosis. Mice carrying a null mutation of the progesterone receptor gene exhibit several reproductive abnormalities, including anovulation, uterine hyperplasia, and lack of mammary gland development, but homozygous embryos for the progesterone receptor mutation developed normally to adulthood without defects in the placenta function (29, 30). Furthermore, the treatment of pregnant female mice with antiestrogen or with progesterone did not prevent the fetal loss of the HSD17B2KO mice. Thus, the cause of HSD17B2KO embryonic deaths did not appear to be due to the lack of progesterone or increased action of estradiol. However, the results do not exclude the involvement of placental HSD17B2 in the embryonic survival via pathways other than those mediated via estrogen or progesterone receptors.

At E15.5 onward, Hsd17b2 expression is also detected in the epithelium of the kidney where it is detected as a spot-like pattern at regions adjacent to developing glomeruli. Weak expression of the gene is also found in the epithelial cells of the central lumen of the kidney (19). The expression pattern of mouse Hsd17b2 in fetal kidney is in line with the findings in adults where moderate expression is found in the thick descending or ascending limbs of the loops of Henle (17). The expression of Hsd17b2 in the kidney appears to be essential for the kidney development as the HSD17B2KO mice present with disrupted kidney architecture. HSD17B2KO mice frequently presented with unilateral renal degeneration, with the affected kidney appearing as a fluid-filled sac. Progressive enlargement of tubules is a sign of hydronephrosis. Histological analysis of the HSD17B2KO mice urinary tract revealed no sign of physical obstruction other than enlarged tubules, collecting duct, and pelvis, suggesting that hydronephrosis may have caused a functional obstruction. The lack of expression of HSD17B2 in epithelial cells in loops of Henle may affect the normal reabsorption and secretion function leading to abnormal collection of urine in the kidney. However, the kidney dysfunction seems unlikely to be responsible for embryonic lethality, because mice with bilateral renal agenesis, such as Pax-2 deficient mice (31), do not show embryonic lethality. Various placental dysfunctions have been shown to lead to kidney lesions such as glomerulosclerosis (32, 33, 34), and thus, the malformation observed in the HSD17B2KO mice kidneys could partially be due to the abnormal function of placenta, observed in HSD17B2KO mice.

Histological analyses of the brains of 1-d-old HSD17B2KO mice revealed that their neuronal migration is disorganized. Control of the migration is highly orchestrated and dependent on both genetic and environmental factors. The critical role of neuronal migration in the brain development is evident from the variety of malformations occurring after abnormal migration (35). There are no previous data of the role of HSD17B2 in neuronal migration or in brain disorders. However, there is evidence for the role of sex steroids in brain development, and it has been shown that estrogen receptor 2 (ESR2, ERβ) is necessary for late embryonic development of the brain and it is involved in both neuronal migration and apoptosis (36).

In addition to the HSD17B2KO mice, several other mouse models with hydrocephalus have been described. The phenotype seems to arise as a result of the disruption of genes with vast majority of functions. These include DNA polymerase β (Pol β), involved in base excision repair in mammalian cells (37), Foxj1, regulating T cell activation (38), and L1, a member of the Ig superfamily, expressed in the developing central and peripheral nervous system (39). At present, none of the above mentioned gene defects could be linked to the action of HSD17B2.

Phylogenic analysis indicates that HSD17B2 is a close homolog of retinoid-converting enzymes in Caenorhabditis elegans, displaying 65% sequence identity to retinol dehydrogenase type 1 (40, 41, 42). Furthermore, similarly to HSD17B2, several retinoic acid-metabolizing enzymes belong to the family of aldo-keto reductases both in rodents and humans (43). Interestingly, the phenotypes identified in the transgenic mice expressing human HSD17B2 (22) and in the HSD17B2KO mice are closely similar to some of the phenotypes obtained in mice with altered retinoic acid metabolism (44, 45, 46, 47). In in vitro studies, the HSD17B2 expression has been showed to be regulated by retinoids (48). It is therefore possible that some of the phenotypes observed in conditions with a misbalanced retinoid signaling are mediated via misexpression of HSD17B2.

The present study showed that HSD17B2KO mice exhibit fetal and perinatal mortality. The KO mice were growth retarded and suffered from abnormalities in the placenta, the kidney, and the brain. The phenotypes observed indicate either a novel role for sex steroid metabolism on embryonic development or, more likely, a novel function for HSD17B2 enzyme.

MATERIALS AND METHODS

Construction of the Hydroxysteroid (17-β) Dehydrogenase 2 Targeting Vector

A bacterial artificial chromosome (BAC) clone containing the mouse Hsd17b2 gene was isolated by screening the mouse genomic 129/SvJ library (Genome Systems, St. Louis, MO), using a fragment of mouse Hsd17b2 cDNA (NCBI, GenBank accession no. NM 008290) as a probe. A 6.5-kb fragment of the BAC clone containing the exon 4 of the Hsd17b2 gene and the flanking regions was subcloned into pACYA177 plasmid (New England Biolabs, Beverly, MA) by ET cloning (GeneBridges, Dresden, Germany; Fig. 1, A and B) (49, 50, 51). The PCR product used in subcloning was amplified from pACYA177 plasmid using the primers (A) 5′-CAG AAA ATG GAA GCA GTA AAG TTA AAC TCT CCT CCA GAG ACT TCA GGA GAG GCA GAC CTC AGC GCT AG-3′ and (B) 5′-TTT CAT TGT AAT ATA TTA TAT CTA GAA GTT TTG GCA ATA TTT TTA CAA TGT GAA GAC GAA AGG GCC TC-3′ where the homology arms to the introns 3 (primer A) and 4 (primer B) of the Hsd17b2 gene are indicated in bold, and the PCR primers for an ampicillin resistance (amp) gene and the ori in pACYA177-plasmid are in italics (Fig. 1B).

The resulting plasmid pACYA177-HSD17B2 was targeted using the ET cloning technique. Exon 4, containing the active center of the enzyme, was disrupted by inserting a neomycin resistance (neo) gene that was under the control of the mouse phosphoglycerol kinase 1 (PGK1) promoter for eukaryotic expression and under the tn5 promoter for bacterial expression. The neomycin dual transcription unit used was amplified from pPGK-loxP-neo plasmid (GeneBridges) using the primers (C) 5′-TAAT CTC CCA TTC TTT CCA TAG GCA CGG TTC CAC TTC AAA TGA CAT CAG CTG AAG CAG GGA TTC TGC AAA C-3′ and (D) 5′-CTC CTG TCT GAT GAT TGT TGA GAA CAT GGT TAG AGC TGC CTT TGT GGC TGC GCG GAT TTG TCC TAC TCA GG-3′ where the homology arms to the target gene are indicated in bold, and the PCR primers for pGK-loxP-NEO are shown in italics (Fig. 1B). The final targeting vector contained 1.7- and 4.6-kb long homology regions upstream and downstream of the exon 5 of mouse Hsd17b2, respectively (Fig. 1A).

Gene Targeting in ES Cells and Generation of Hsd17b2-Deficient Mice

The linearized targeting vector for Hsd17b2 was introduced into AB 2.2 ES cells (Lexicon Genetics, TX) by electroporation (one pulse of 230 V, 500 μF; Gene Pulser; Bio-Rad Laboratories, Hercules, CA). Cells were subsequently cultured for 8 d in the presence of a selective antibiotic, G418 (0.35 mg/ml; Invitrogen Life Technologies, Paisley, UK), after which 250 clones were manually picked, expanded, and stored frozen at −70 C. Homologous recombination was screened by PCR and Southern blot analysis.

Animals

C57BL/6N mice used for blastocyst donors were obtained from Charles River Laboratories (Wilmington, MA). Homozygous (HO) HSD17B2KO, heterozygous (HE), and WT mice were obtained from the heterozygous breeding. All the studies were carried out in mixed (SV129/C57BL/6N 1:1) strain. HSD17B2KO mice born alive were carefully monitored twice a day, and special nursing was needed, including prolonged breast feeding until the age of 4–6 wk and softening of the food pellets by water. The mice were fed with a Soya-free diet, and they were maintained in a specific pathogen-free stage at Central Animal Laboratory at the University of Turku. All studies carried out with the mice were approved by the ethical committee for experimental animals, University of Turku (Turku, Finland), complying with international guidelines on the care and use of laboratory animals.

Genotyping by PCR and Southern Blot Analyses

PCR screening of the recombinant allele was carried out on ES cells to identify the clones with homologous recombination. For PCR amplification an antisense primer NEO1an within the Neo cassette and a sense primer Genom3, specific for the Hsd17b2 gene at the 5′-flanking sequence of the targeting construct, were used (Fig. 1A and Table 3). The PCR conditions (40 cycles) were 96 C for 2 min, followed by 94 C for 30 sec, 58 C for 30 sec, and 72 C for 4 min and a final extension at 72 C for 10 min.

Table 3.

Primers and Annealing Temperatures Used

A.
NEO1an	5′-GTTTGCAGAATCCCTGCTTC-3′
Genom3	5′-CATGGTTCTAGAGGGCCACA-3′
ex4s	5′-GGCACGGTTCCACTTCAAAT-3′
ex4a	5′-CGGGTTTGATGGCACAACTT-3′
NEOs1	5′-GACCCATGGCGATGCCTGCTTG-3′
B.
P1	5′-TCCGCAGGAATGCAATTGTTGCTG-3′
P2	5′-GGGAAAGCATTACAAGTCTGGTTC-3′	Tm 60
E1	5′-TTGCTAGAGTGGAGCTTGCC-3′
E2	5′-GCCTAAATGGTGGAGGCATTC-3′	Tm 60
β-A	5′-TTGGCCTTAGGGTTCAGGGGG-3
β-A	5′-CGTGGGCCGCCCTAGGCACCA-3′	Tm 60

Table 3A lists PCR primers used in ES cell screening and mouse genotyping; Table 3B lists semiquantitative RT-PCR primers. Tm, Melting temperature.

Southern blot analyses were performed on genomic DNA to confirm homologous recombination in PCR-positive ES clones as well as in the tissues of the HSD17B2KO mice. Genomic DNA was digested with XmnI, size fractioned in 0.8% agarose gel, and transferred onto Hybond-N⁺ membrane (GE Healthcare, Buckinghamshire, UK). The membranes were hybridized overnight at 42 C in UltraHyb buffer according to manufacturer’s instructions (Ambion, Austin, TX) with a 400-bp PCR fragment of the intron 3 of the Hsd17b2 gene (Fig. 1A) labeled with [α-³²P]dCTP (random-primed method, Promega Corp., Madison, WI). The signal was detected with Fuji Bas 5000 phosphor imager (Fuji Film Ltd., Tokyo, Japan).

Genotyping of the mice was performed on DNA extracted from yolk sacs of the embryos or earmarks of 2-wk-old mice. For this, two PCRs applying primers ex4s, ex4a, and NEOs1 were performed (Fig. 1A and Table 3). The 1900-bp and 450-bp PCR products specific to the recombinant allele were distinguishable from the 173-bp fragment corresponding to the WT allele. The PCR conditions (25 cycles) were as follows: 94 C for 4 min, followed by 94 C for 30 sec, 60 C for 30 sec, and 72 C for 2 min and a final extension at 72 C for 10 min.

Northern Blot and RT-PCR Analyses

For Northern blot analysis, total RNA (10 μg/lane) from liver and kidney was resolved on an agarose gel, and the RNA was transferred onto Hybond-N⁺ membrane (GE Healthcare). The membranes were hybridized overnight at 42 C in UltraHyb buffer according to manufacturer’s instructions (Ambion) with [α-³²P]CTP-labeled 734-bp cDNA fragment of mouse Hsd17b2. The nonspecific signal was removed by a series of washes with 2× saline sodium citrate and 0.5% sodium dodecyl sulfate. Hybridization results were visualized by autoradiography using a Fuji Bas 5000 phosphor imager (Fuji).

For RT-PCR analysis total RNA from placenta was isolated using RNeasy minikit (QIAGEN, Hilden, Germany). Total RNA (1 μg) was treated with deoxyribonuclease (DNase I Amplification Grade Kit; Invitrogen Life Technologies). The gene-specific primers used in the RT-PCR were P1 and P2 for PL-1, E1 and E2 for Esx1, and for the internal control bAkt1 and bAkt2 for β-actin. The primers and annealing temperatures used are described in Table 3. After 30 cycles, the amplification products were resolved in agarose gel and visualized with ethidium bromide.

Morphological and Histological Analyses

Pregnant female mice were euthanized by cervical dislocation and blastocysts at E3.5, and embryos at E11.5, E15.5, and E17.5, together with the yolk sack, were dissected free from the uterine tissue and placed in ice-cold PBS, and macroscopic abnormalities were evaluated. For histological analysis, the embryos and the placentas were fixed in 4% paraformaldehyde in PBS for 24–48 h at 4 C and embedded in paraffin. Similarly, various organs from newborn mice were weighed, fixed with 4% paraformaldehyde in PBS, and embedded in paraffin. The embedded tissues and whole embryos were cut in 4-μm sections and stained with Harris’s hematoxylin and eosin (BDH Ltd., Poole, UK). For bone and cartilage analyses, newborn mice were euthanized, skinned, eviscerated, and fixed with 100% EtOH for 24 h. The cartilages were stained with 0.3% Alcian Blue, and the bones were counterstained in 0.1% Alizarin Red solution. The gender of the newborns was determined by the presence of the ovaries or the testes. Apoptosis was studied using ApopTag Peroxidase In situ Oligo Ligation (ISOL) Apoptosis Detection Kit (Chemicon International, Inc., Temecula, CA) according to the manufacturer’s instructions.

Hormone Measurements

Placentas were homogenized in 1 ml PBS for the analysis of the intratissular estradiol and testosterone concentrations. The steroids were extracted from the homogenates or from amniotic fluid using diethyl ether, the organic phase was evaporated into dryness, and the steroids were solubilized in PBS. Estradiol and progesterone concentrations were measured by Delfia kits following the manufacturer’s protocol (Wallac Oy, Turku, Finland). Measurements of testosterone concentrations were performed by a RIA, as described previously (52).

Steroid Treatments

On E7.5 (plug day = E0.5), HE and WT female mice (mated with HE and WT males, respectively) were implanted with pellets containing either an antiestrogen, ICI 182,780 (1.5 mg/12-d release, Innovative Research of America, Sarasota, FL) or progesterone (5.0 mg/21-d release, Innovative Research of America). Control mice were implanted with a placebo pellet. The females were euthanized on pregnancy d 17.5, and living embryos of various genotypes were determined by the presence of a beating heart. Amniotic fluids and placentas were collected for further analyses.

Statistical Analysis

Statistical analyses were performed using the SigmaStat program (SYSTAT Software, Inc., San Jose, CA). The results for body and organ weights were analyzed by Student’s t test or Mann-Whitney rank sum test with a limit of significance set at P < 0.05. One-way ANOVA was used to analyze the data from antiestrogen and progesterone treatment study.