abstract
Thyroid hormones are critical to growth and development of the human fetus. Abnormal placental development, a major cause of intrauterine growth restriction (IUGR), is associated with a high perinatal mortality and morbidity. Thyroid status has been postulated to play a role in the pathogenesis of such morbidity. In the present study, we have investigated fetal thyroid function and placental expression of thyroid hormone receptor (TR) α and β variants during normal human pregnancy and in pregnancy associated with IUGR. Measurement of free thyroid hormones and TSH concentrations revealed significant rises in free T4 and free T3 between the second and third trimesters of normal pregnancy. Serum concentrations of free T4 and free T3 were lower in fetuses affected by IUGR, although serum TSH levels were not significantly different. Immunocytochemistry demonstrated the presence of TR α1, α2, andβ 1 proteins within the nuclei of trophoblast and stromal placental cells. Immunostaining for these TR variants increased with increasing gestation in normal placenta. Comparison of IUGR placental samples with normal samples revealed greater immunostaining for TR α1, α2, andβ 1 variants in IUGR. Examination of pretranslational expression of TRα 1, α2,β1, and β2 variants by semiquantitative RT-PCR revealed increasing expression of TR α1, α2, and β2 messenger RNAs with increasing gestation in normal pregnancy, which “mirrored” post-translational expression. However, and in contrast, there were no significant differences in expression of TR messenger RNAs in normal and IUGR placenta. The present findings of reduction in serum free thyroid hormones and increased expression of TR α and β proteins in association with IUGR highlight the potential importance of thyroid status in influencing long-term fetal outcome in this condition.
FETAL growth is dependent upon a number of endocrine, paracrine, and autocrine events within the fetoplacental unit (1), the precise mechanisms of which remain to be defined. Babies born with intrauterine growth restriction (IUGR) are major casualties of both perinatal and neonatal mortality. This pathological process also causes significant morbidity, with 10% of low birth weight babies having physical handicaps and a further 5% showing neurodevelopmental delays at the age of 3–4 yr (2). Thyroid status is one of several factors that have been postulated to play a critical role in the pathogenesis of such morbidity, especially with respect to growth and development of the central nervous system (3, 4).
Investigations in human pregnancy have indicated that TRH can be detected in the fetal hypothalamus by the end of the first trimester, at a time when the thyroid begins to concentrate iodine. It is possible to measure TSH in the pituitary and in serum in early pregnancy, and concentrations rise towards term, usually exceeding adult levels (5, 6). Concentrations of total T4 and T3 in fetal serum rise between the 10th and 30th wk of gestation, paralleled by rising serum thyroxine binding globulin (TBG) levels. Increases in free T4 and free T3 concentrations have also been observed in late gestation, but values for both total T3 and free T3 remain lower than those in the maternal circulation (6).
The use of cordocentesis for fetal blood sampling in pregnancies complicated by intrauterine fetal growth restriction has revealed a significant reduction in circulating free T4 and a modest elevation in TSH concentrations in hypoxemic, acidotic fetuses (7). Likewise, collection of umbilical cord blood from very low birth weight (VLBW) babies at delivery has revealed that total T4 and total T3 concentrations are significantly reduced. This has been postulated to largely reflect reductions in TBG, and free thyroid hormone concentrations are not markedly low compared with term infants (8). In addition, transient “hypothyroidism,” characterized by low total and free T4 concentrations, with or without a rise in TSH, frequently develops in small premature infants in the first two to three postnatal weeks (3, 4). While treatment of permanent and transient primary hypothyroidism (associated with a rise in serum TSH) is considered important in terms of long-term neurological outcome, the significance of “physiological” hypothyroxinemia of prematurity (with no rise in serum TSH) remains controversial (3, 9), as does the role of thyroxine therapy in this situation (10, 11). Severe hypothyroxinemia has, however, been specifically associated with poor neurological outcome (12).
While circulating concentrations of free thyroid hormones are major determinants of cellular uptake of T4 and T3, many other factors may modulate thyroid hormone action at the tissue level, including deiodination of T4 to T3 and the expression and function of thyroid hormone receptors (TRs). The ontogeny of TR expression in fetal tissues has engendered much interest (9), but most studies have been confined to rats and other mammals, with little known about TR expression in the human fetus. Ligand binding studies have revealed the presence of high affinity nuclear binding sites for T3 in the human fetal brain after the tenth week of gestation (13), as well as in other tissues such as the placenta (14). In the human placenta, T3 has been noted to stimulate production of 17β estradiol and epidermal growth factor (15), indicating a possible role for T3 in the control of trophoblast growth and development, which in turn are known to be abnormal in IUGR (16). The placentae of pregnancies complicated by IUGR have distinct ultrastructural differences, especially those with abnormal umbilical blood flow. The terminal villi demonstrate reduced cytotrophoblast proliferation and villous angiogenesis with increased deposition of matrix proteins within the villous stroma. Such abnormalities of trophoblast development may be influenced by fetoplacental thyroid status, as well as by autocrine mechanisms.
To further address the functional significance of hypothyroxinemia and other possible abnormalities of circulating thyroid hormones and TSH in IUGR, we have investigated tests of fetal thyroid function and placental expression of TR α and β variants at different stages of normal human pregnancy and have compared findings with those from fetal blood and placental tissue associated with IUGR.
Subjects and Methods
Subjects and samples
Flowing blood samples were collected during cordocentesis (17) from both normal (appropriately grown) fetuses and those affected by IUGR, defined as outlined below. Cordocentesis was performed in a group of normal fetuses because of the need to exclude karyotypic abnormalities. Blood samples (800 mL surplus to that required for karyotyping) were obtained from 11 fetuses in the second trimester of pregnancy (median gestational age 22 wk, range 20–25 wk) and from 15 fetuses in the third trimester (median gestational age 31 wk, range 28–38 wk). Cordocentesis was also performed in a group of fetuses affected by IUGR (n = 15, median gestational age 29 wk, range 24–35 wk). The diagnosis of IUGR was made in the presence of 3 of 4 well-defined characteristics (18): 1) abdominal circumference (AC) measured by ultrasound less than the tenth centile for gestational age; 2) reduced fetal growth potential (ΔAC less than 1.5 standard deviations in 14 days); 3) presence of oligohydramnios, as measured by amniotic fluid index (19); 4) absent or reversed end diastolic flow in the umbilical artery Doppler studies. All fetuses included in the study were shown subsequently to be karyotypically normal.
In addition to the blood samples described above, fresh placental tissue was collected at surgical termination of first trimester (n= 15, median gestation 10 wk [range 7–12 wks]) and second trimester (n = 11, median gestation 15 wk [13–21 wks]) of normal pregnancies and after elective Cesarean section delivery (n = 21, median gestation 38 wk [27–41 wks]). Placental tissue was also collected at delivery from a group of those fetuses (n = 10) identified prospectively as being affected by IUGR. For comparison of findings from blood and placental samples between normal and IUGR fetal groups, normal fetuses were gestationally matched to fetuses affected by IUGR. There were no significant differences in the maternal ages of these fetal cohorts. As expected, median birth weights (normal 2174 g (range 819–3690 g)), IUGR 1179 g (range 501–1990 g) and placental weights (normal 456 g (range 434–551 g), IUGR 255 g (range 197–287 g)) were lower when pregnancy had been complicated by IUGR.
Measurement of circulating concentrations of free thyroid hormones and TSH
Circulating concentrations of free T4, free T3, and TSH were measured in sera collected from fetal blood samples. Free T4 and free T3 were measured by two-step, back-titrated, time-resolved fluoroimmunoassays, methods utilized in other investigations of fetal thyroid status (8) (Delfia, Wallac, Milton Keynes, UK) and TSH by time-resolved fluoroimmunometric assay (Delfia, Wallac, Milton Keynes, UK) calibrated against the WHO second IRP 80/558. Inter-assay coefficient of variation for free T4 was 6.1% at 18.3 pmol/L, for free T3 5.9% at 5.8 pmol/L, and for TSH 6.0% at 3.88 mU/L.
TR α and β expression in placenta determined by immunocytochemistry and by Western blotting
Full thickness placental biopsies were excised from the central (peri-umbilical) region of freshly delivered placentae. These were washed thoroughly in ice-cold, physiological (0.9%) saline and either fixed in formalin or separated from overlying fetal membranes and snap frozen in small pieces (1 g) before storage at −80 C. The formalin-fixed placental biopsies were subsequently embedded in paraffin wax and stored at room temperature.
For immunocytochemistry, 5 μm formalin fixed sections of normal placentae and diseased placentae were dewaxed and microwaved in citrate buffer (0.01 m, pH 6) at 750W for 30 min, and allowed to cool. Sections were treated with methanol-H2O2 (1:1000) to block endogenous peroxidase activity. After washing in phosphate buffered saline (PBS), slides were incubated with specific rabbit polyclonal antibodies to human TR α1, α2, and β1[ Affinity Bioreagents, NJ; 1:200 in 10% normal goat serum (20)] overnight at 4 C, as performed in our laboratory previously (21, 22). Primary antibody was omitted during incubation of control sections and was replaced by nonimmune serum. Biotinylated secondary antibody was added to sections for 30 min, followed by addition of the avidin-biotin complex (Dako Duet, Strept ABC Complex, Dako Ltd., Copenhagen, Denmark). Slides were developed using 3,3′ diaminobenzidine for 5 to 10 min and counterstained with Meyers Haematoxylin. Photographs were taken at 400x magnification. Immunostaining was quantified by a single experienced observer (JV) under “blind” conditions (22), with an overall assessment of the presence or absence of specific immunostaining for each section examined, and an assessment of staining of specific placental cell types (see below).
To investigate TR expression using Western blot analysis, nuclear proteins were extracted from fresh frozen placental tissue using the methods of Samuels and Tsai (23). Briefly, placental fragments were homogenized in STM buffer (0.25 m sucrose, 20 mm Tris, 1.1 mm magnesium chloride pH 7.85) at 4 C and resuspended in STM + Triton X-100 (0.5%) on two occasions to obtain a clean nuclear preparation. Purified nuclei were incubated in lysis buffer (STM + 0.5 mm PMSF, 5 mm DTT, 0.4 m potassium chloride, 20% glycerol) and were left agitating vigorously for 15 min on ice. Nuclear chromatin was removed by centrifugation and the supernatant containing nuclear proteins was quantified by optical densitometry at 595 nm (OD595) and stored at −70 C.
Specificity of the antibodies employed in immunocytochemistry was confirmed by Western blot analysis of nuclear protein preparations of placentae performed by SDS-PAGE on discontinuous acrylamide gels using established methods (21, 22, 24).
Pretranslational expression of TR variants determined by semiquantitative RT-PCR
Total RNA was isolated from fresh frozen placental tissue using a single step acid guanidinium phenol-chloroform extraction technique (25). Briefly, fragments of placenta were homogenized in an Ultraturax homogenizer in the presence of RNAzol‘ B (Biotecx Laboratories, Houston, TX). Total RNA was extracted by adding 0.1 vol chloroform and vortexing before centrifugation at 8000 rpm for 20 min at 4 C. The aqueous phase was transferred into an equal volume of isopropanol and RNA was allowed to precipitate at −20 C overnight. Following centrifugation, the RNA pellet was washed with 75% ethanol and allowed to dry before being resuspended in diethyl pyrocarbonate (DEPC) treated water. Total RNA was quantified by optical density measurement at 260 nm, and the integrity of RNA was verified by agarose gel electrophoresis with ethidium bromide staining.
Reverse transcription was performed using avian myeloblastosis virus (AMV) reverse transcriptase (Promega, Madison, WI) in a total reaction volume of 50 mL/2 mg of placental total RNA was added to 60 pmol oligo(dT)15 (Promega) primer, and the volume was adjusted to 32 mL with DEPC water. The solution was heated to 77 C for 10 min to allow for primer annealing. The mixture was cooled to room temperature and 10 mL of 5x AMV reverse transcriptase buffer (Promega) was added along with 5 mL of deoxynucleotide triphosphate (dNTP) mix (20 mmol each; Boehringer Mannheim, Germany). Fifty units of placental ribonuclease inhibitor (RNasin®, Promega) and finally 15 units of AMV reverse transcriptase (Promega) were added to the reaction mixture. Incubation for 70 min at 42 C was carried out in a thermal cycler, and the RT reaction was terminated by heating to 77 C for 10 min. RT was performed on each RNA sample with no AMV RT added. PCR carried out on this resultant RT product generated no PCR product, confirming absent genomic DNA contamination.
PCR was carried out in a total volume of 50 mL, and a hot start technique was employed in all instances. For the purposes of semiquantitation, PCR reaction components were premixed (to generate master mixes) before addition to individual PCR tubes to minimize pipetting errors, and all samples underwent PCR at the same time in the same experiment. Each PCR tube contained 60 pmol each of the specific TR α1, α2, β1, and β2 forward and reverse primers described previously (22), 1 mL of dNTP mix (20 mmol each; Boehringer Mannheim) and between 2–10% of the RT reaction product. Volume was adjusted to 44.5 mL with water. The reaction constituents were pulse centrifuged, mineral oil was over-laid, and the PCR mixture was heated to 95 C for 6 min and cooled to 72 C before adding PCR buffer and Taq. A dilution of 0.5 units of Taq DNA polymerase (Boehringer Mannheim) in 5 mL of 10x PCR reaction buffer (Boehringer Mannheim) was prepared and was added through the mineral oil layer to each of the PCR reaction mixtures. PCR cycling was commenced using a melt temperature of 94 C in all instances, and extension was carried out at 72 C. PCR products were run on 2% agarose gels stained with ethidium bromide. Appropriate negative control PCR reactions were run in parallel.
All RT reactions utilized oligo(dT)15 primed RNA to minimize the variations in RT efficiency seen when using specific RT primers. Once PCR conditions for each primer pair had been optimized, comparative kinetic analyses were performed as described previously (22, 26, 27) to determine the phase during which there was exponential generation of PCR product before reaching plateau. It was at this point that the PCR was terminated allowing for semiquantitative data to be obtained.
Ethidium bromide stained gels were visualized under UV light, and the image was digitized and stored on computer disk. Gelbase/Gelblot software (Ultra Violet Products, Cambridge, UK) was used to measure luminescence of PCR generated bands, which in turn is a measure of the quantity of PCR product (28). Values for each PCR product were expressed as a fraction of the quantity of β-actin, which was used as an internal standard to correct for sample to sample variation in RNA degradation. PCR products were run in triplicate on 2% agarose gels, and the mean luminescence value was recorded in each case.
Intergroup comparisons for continuous variables with a nonparametric distribution were made using the Mann-Whitney U-test to determine significant differences between the data sets. For such data, median values and ranges are described. If normal distribution was noted by a measure of skewness, then mean ± se are described. Categorical data were analyzed using Fisher’s exact test and both odds ratios and 5th–95th centile confidence intervals. Statistical significance was taken as P < 0.05, unless otherwise stated. Ethical approval for the study was obtained from the South Birmingham Ethics Committee.
Results
Thyroid hormone and TSH concentrations in serum from normal and growth restricted fetuses
Measurement of free thyroid hormone and TSH concentrations in serum collected from normal fetuses revealed a significant rise in serum free T4 between the second (n = 11) and third trimesters (n = 15) of pregnancy (10.4 (7.2–11.6) pmol/L, median (range) vs. 12.6 (10–20), P < 0.01, Mann-Whitney U test). A similar increase in serum free T3 concentrations was observed (1.9 (1.2–2.3) pmol/L vs. 3.01 (2.1–4.1), P < 0.0001), in parallel with a small but nonsignificant rise in serum TSH concentrations (4.8 (3.7–10.1) mU/L vs. 6.2 (4.1–11.2), P = 0.14).
Comparison of tests of thyroid function in groups of gestationally matched fetuses with normal growth or affected by IUGR (n = 15, mean gestational age 29.3 wk, range 24–35) demonstrated that serum concentrations of free T4 were significantly lower in fetuses affected by IUGR (9.8 (4.9–12.8) pmol/L, [median and ranges] vs. 12.0 (8.5–20.9) pmol/L, P < 0.001 by Mann-Whitney U test), as were serum concentrations of free T3 [2.0 (1.0–5.5) pmol/L vs. 2.9 (2.0–7.5) pmol/L, P < 0.05]. In contrast, serum TSH values were not significantly different [6.5 (0.5–12.0) mU/L vs. 6.0 (2.5–11.2), P = 0.7].
TR α and β expression determined by immunocytochemistry and Western blotting
Immunocytochemistry confirmed the presence of TR α1, α2 andβ 1 proteins within the nuclei of syncytiotrophoblast, cytotrophoblast, and stromal cells throughout the villous core by the third trimester of pregnancy (Fig. 1). The majority of nuclei were stained, but some nuclei were negative; there was no obvious difference in the degree of immunostaining of trophoblast and stromal cells. In addition to specific nuclear staining, weak granular staining of placental cell cytoplasm was also observed. Examination of immunostaining of sections of paraffin-fixed placental tissue from normal first (n = 15), second (n = 11), and third trimester (n = 21) pregnancies revealed an increase in the proportion of samples with detectable TR protein with increasing gestational age (Fig. 1, Fig. 2). No TRβ 2 protein could be detected in human placenta by either Western analysis or immunocytochemistry using a polyclonal antibody directed against TR β 2 (kindly provided by Dr Paul Yen, Boston, MA) (29).
Immunostaining of representative samples of first, second and third trimester normal placenta for the presence of TR α1, α2, or β1 protein. Increasing (brown) nuclear immunostaining for each TR variant is observed with increasing gestation. Negative controls (primary antibody omitted) are also shown.
Percentage of samples of normal human placenta from the first (T1, n = 15), second (T2, n = 11), and third (T3, n = 21) trimesters of pregnancy which demonstrated nuclear immunostaining for TR TR α1, α2 or β1 protein.* P < 0.01 compared with T1 using Fisher’s exact test. * = trimesters.
Comparison of TR immunostaining in placentae from fetuses affected by IUGR (n = 10) with gestationally matched placentae (Fig. 3, Fig. 4) revealed that a higher proportion of IUGR samples expressed TR α1, α2, and β1 variants than normal placental samples. The distribution of TR expression amongst different placental cell types was examined by counting the number of cells immunopositive for each TR variant within 100 trophoblast and 100 stromal cells in 4 randomly selected, high power fields from a section of each IUGR or normal placental sample illustrated in Figure 4. There was a significant increase in the percentage of immunopositive cells for all TRs investigated (TR α1 IUGR 46.8 ± 2.9%, mean ± sevs. 22.3 ± 5.4%, P < 0.0001, Fisher’s exact test; α2 IUGR 21.6 ± 7.5% vs. 6.7 ± 3.4%, P < 0.01; TR β1 IUGR 27.2 ± 6.0% vs. 10.1 ± 4.1%, P < 0.01). Similar findings were observed when trophoblast and stromal cells were considered separately (data not shown).
Immunostaining of representative samples of normal placenta and placenta associated with IUGR for the presence of TR TR α1, α2 or β1 protein, demonstrating a higher proportion of nuclear immunostaining in association with IUGR. Negative controls (primary antibody omitted) are also shown.
Percentage of samples of placenta associated with IUGR (n = 10) and gestationally matched normal placenta (n = 10), which demonstrated nuclear immunostaining for TR TRα 1, α2 or β1 protein. **P < 0.01,*** P < 0.0001 compared with normal placenta using Fisher’s exact test.
TR α1, α2, and β1 proteins were visualized as bands of similar size to those reported by ourselves in other tissues (21, 22) and proteins of similar size were observed in normal and IUGR placental samples. No protein bands were visualized in control specimens in which the primary antibody had been omitted.
Pretranslational expression of TR variants determined by semiquantitative RT-PCR
Messenger RNAs (mRNA) of appropriate size encoding all TR variants were detected in all samples of placental tissue (22). Examination of the level of expression of specific TR mRNAs (corrected for expression of β-actin) revealed significant increases in TR α1, α2, and β2 expression with increasing gestational age, although a different pattern of expression with increasing gestation was observed for TRβ 1 (Fig. 5). There was no significant difference in the level of expression determined semiquantitatively of TR mRNAs when IUGR and gestationally-matched normal placental samples were compared (Fig. 6).
Ratios of expression of mRNAs encoding specific TR α1, α2 or β1 or β2 TRs to β-actin mRNA in normal human placenta associated with the first (T1, n = 15), second (T2, n = 11), and third (T3, n = 21) trimesters. Results shown are means±se. *P < 0.05,** P < 0.01 compared with first trimester (T1) by Kruskal-Wallis ANOVA.
Ratios of expression of mRNAs encoding specific α1, α2 or β1 or β2 TRs to β-actin mRNA in IUGR and gestationally matched normal human placenta (n = 10). Results are shown as means±se. P = NS comparing IUGR with normal sample results.
Discussion
In the present studies we have examined fetal tests of thyroid function and placental TR expression in normal pregnancy and in pregnancy complicated by IUGR. Measurement of circulating concentrations of free thyroid hormones and TSH in serum from normal fetuses confirmed an increase in free T4 and free T3 concentrations with increasing gestational age, in parallel with a small rise in serum TSH, findings in accord with other studies (6, 8). The relatively small sample volume obtained at cordocentesis determined that it was not feasible to employ a direct equilibrium dialysis method for measurement of free thyroid hormone concentrations in fetal serum. We therefore used a method for estimation of free thyroid hormone concentrations employed in a recent study of cord blood from normal and low birth weight infants (8). Results for free T4 obtained with this method of two-step, back-titration, immunoassay have been compared with equilibrium dialysis (8, 30) and have been shown to correlate with free thyroid hormone concentrations measured using equilibrium dialysis. Underestimates of free thyoid hormones have been reported with related methods in several clinical situations, including nonthyroidal illness (31), a factor that should be taken into account when interpreting absolute concentrations from immunoassays reported here. We did not measure reverse T3 concentrations in our samples.
In parallel with an increase in serum free T4 and free T3 concentrations from the second to the third trimesters of normal pregnancy, we demonstrated an increase in TR α1, α2, andβ 1 protein in placenta with increasing gestational age, when assessed semiquantitatively by immunocytochemistry. TR expression was detected in the fetal placenta as early as the first trimester, before significant production of thyroid hormones from the fetal thyroid, in accord with similar observations in the rat (32). The apparent increase in placental TR α and β variant expression with increasing duration of gestation is also in agreement with the findings of studies of TR expression in various rat tissues (9) and with the results of a study of TR expression in the human brain, which demonstrated an increase in receptor expression between the 10th and 16th wk of fetal development (13). An increase in TR α1, α2, and β1 protein in the present studies was accompanied by a parallel increase in TR α1, α2, andβ 2 specific mRNAs determined by semiquantitative RT-PCR. These findings suggest that increased TR expression with increasing fetal age reflects increased TR gene transcription, although the apparent discrepancy between the ontogeny of TR β1 protein and mRNA remains unexplained. The physiological relevance of expression of TR β2 mRNA in the placenta is unknown. TR β2 expression has been considered to be largely pituitary specific, although the presence of TR β2 mRNA in extrapituitary tissues has been reported in the rat, including in the central nervous system and liver (33, 34). The present study reports for the first time the presence of TR β2 mRNA in human placenta; failure to demonstrate TR β2 protein by Western blotting and immunocytochemistry suggests relatively low abundance and argues against an important physiological role in this tissue.
In the present studies, placenta was utilized both to further elucidate mechanisms of trophoblast development and as a source of fetal tissue. Further studies are required to determine whether the pattern of TR expression is similar in other fetal tissues, especially the central nervous system. The demonstration of parallel increases in circulating free thyroid hormone concentrations and placental TR expression supports an increasing role for thyroid hormone action in“ target” fetal tissues during development. The role of thyroid hormones in controlling trophoblast growth and development also warrants further investigation. Trophoblast has a high binding capacity for T3, and it has been suggested that placenta is a thyroid hormone dependent tissue (14). Maruo et al. (15) have reported stimulation by thyroid hormones of trophoblast endocrine function, with enhanced production of human placental lactogen and human chorionic gonadotropin. It has also been shown that T3 enhances the production of epidermal growth factor, a potent trophoblast mitogen (35). Thyroid hormones may thus have an important role in villous development and placentation, in part mediated by the interaction between endocrine and autocrine factors (36).
An important component of the present study was a comparison of thyroid status in the appropriately grown and the growth retarded fetus. The presence of IUGR was defined prospectively using stringent criteria to indicate a pathological state. This is important as the results of studies of IUGR have been confounded by less clear definition of the clinical condition. We demonstrated a significant reduction in concentrations of both free T4 and free T3 in serum from fetuses affected by IUGR when compared with gestationally- matched normal fetuses. These results are in agreement with those of a study of “small-for-gestational-age fetuses” (7), which also demonstrated a reduction in free T4 concentrations, although that study did not reveal a reduction in free T3. Furthermore, Thorpe Beeston et al. (7) reported an increase in serum TSH in their small for gestational age cohort, whereas there was no significant difference in serum TSH in our IUGR and normal fetal groups. These discrepancies may reflect the application of different diagnostic criteria in the two studies or the use of a fetal “normal range” by Thorpe Beeston et al. (7) compared with specific matching of groups according to duration of gestation in the present study.
The reduction in circulating free thyroid hormone concentrations found in the present study is also in agreement with the results of a number of postnatal studies of very low birth weight babies (8, 12), which have revealed the presence of hypothyroxinemia, either at birth or developing in the early neonatal period. The present findings of reduction in circulating thyroid hormones in the absence of a rise in circulating TSH are also in accord with the findings in adults with a variety of “nonthyroidal” illnesses, an association termed the“ sick euthyroid” syndrome. The physiological significance of a reduction in serum thyroid hormones in association with various illness states remains a subject of debate (37) and has prompted investigation of the role of thyroid hormone supplementation in both the newborn (11) and adults (38), as well as investigation of “tissue” thyroid status in “nonthyroidal” illness (21). Many authors support the view that reductions in circulating thyroid hormones in association with illness do not result in hypothyroidism at a tissue level. The present findings of an increase in TR α1 and TR β1 expression in fetal tissue in IUGR (a finding observed in both trophoblast and stromal cells within the placenta) might provide an explanation for maintenance of tissue euthyroidism in the face of reduced circulating concentrations of receptor ligand. It should be noted that increased expression of the functional α1 and β1 thyroid receptors was associated with increased expression of the nonligand binding α2 variant, which may inhibit transcriptional regulation by functional TRs (39). Such findings indicate that further investigation of “tissue” thyroid status in the IUGR fetus is important, as well as investigation of TR expression in other fetal tissues. An unexpected finding in the present studies was the lack of correlation between TR protein and mRNA expression in IUGR fetal placenta, in contrast with findings in the normal placenta. Apparent discrepancies between TR protein and mRNA expression have been noted previously by several other groups (40–42), including studies of receptor expression in tissues of the developing rat (42). Such discrepancies suggest the operation of translational or post-translational factors in determining the expression of mRNA as protein. Such factors might include variation in sequestration of mRNAs in cytosol, efficiency of translation at a ribosomal level, post-transcriptional modification of the translation product, and alterations in the stability of receptor protein.
In conclusion, we have demonstrated increasing circulating concentrations of free thyroid hormones with increasing gestational age in the normal human fetus, with an associated increase in TR α andβ expression in fetal placenta. We have also demonstrated a reduction in serum concentrations of free thyroid hormones in the fetus affected by intrauterine growth restriction, in this case associated with an increase in TR expression in the placenta. The present findings in IUGR, highlight the potential importance of thyroid status in determining long-term fetal outcome and emphasise the role of further investigation of the pathogenesis and functional significance of disturbance of thyroid status in this common condition with considerable associated morbidity and mortality.
Acknowledgments
The authors wish to thank the staff and patients at Birmingham Womens Hospital, UK. Thanks are especially extended to Professor M. Whittle for his support and encouragement.
N. G. was supported by a Smith & Nephew Research Fellowship.
D. A. S. was funded by the West Midlands Perinatal Audit, United Kingdom.
Gaffney G.
van
Whittle MJ.
