Normal pregnancy: mechanisms underlying the paradox of a ouabain-resistant state with elevated endogenous ouabain, suppressed arterial sodium calcium exchange, and low blood pressure

Abstract

Endogenous cardiotonic steroids (CTS) raise blood pressure (BP) via vascular sodium calcium exchange (NCX1.3) and transient receptor-operated channels (TRPCs). Circulating CTS are superelevated in pregnancy-induced hypertension and preeclampsia. However, their significance in normal pregnancy, where BP is low, is paradoxical. Here we test the hypothesis that vascular resistance to endogenous ouabain (EO) develops in normal pregnancy and is mediated by reduced expression of NCX1.3 and TRPCs. We determined plasma and adrenal levels of EO and the impact of exogenous ouabain in pregnancy on arterial expression of Na⁺ pumps, NCX1.3, TRPC3, and TRPC6 and BP. Pregnant (embryonic day 4) and nonpregnant rats received infusions of ouabain or vehicle. At 14–16 days, tissues and plasma were collected for blotting and EO assay by radioimmunoassay (RIA), liquid chromatography (LC)-RIA, and LC-multidimensional mass spectrometry (MS3). BP (−8 mmHg; P < 0.05) and NCX1.3 expression fell (aorta −60% and mesenteric artery −30%; P < 0.001) in pregnancy while TRPC expression was unchanged. Circulating EO increased (1.14 ± 0.13 nM) vs. nonpregnant (0.6 ± 0.08 nM; P < 0.05) and was confirmed by LC-MS3 and LC-RIA. LC-MS3 revealed two previously unknown isomers of EO; one increased ∼90-fold in pregnancy. Adrenal EO but not isomers were increased in pregnancy. In nonpregnant rats, similar infusions of ouabain raised BP (+24 ± 3 mmHg; P < 0.001). In ouabain-infused rats, impaired fetal and placental growth occurred with no BP increase. In summary, normal pregnancy is an ouabain-resistant state associated with low BP, elevated circulating levels of EO, two novel steroidal EO isomers, and increased adrenal mass and EO content. Ouabain raises BP only in nonpregnant animals. Vascular resistance to the chronic pressor activity of endogenous and exogenous ouabain is mediated by suppressed NCX1.3 and reduced sensitivity of events downstream of Ca²⁺ entry. The mechanisms of EO resistance and the impaired fetal and placental growth due to elevated ouabain may be important in pregnancy-induced hypertension (PIH) and preeclampsia (PE).

the maternal hemodynamic and endocrine environments undergo various adaptations to meet the metabolic demands of pregnancy. Cardiac output and blood volume increase by 30–50% (5) while peripheral vascular resistance declines, usually resulting in a net decrease in blood pressure (BP; Refs. 6, 19). During pregnancy, the activity of the renin-angiotensin system (RAS) increases and promotes salt and water retention (41, 46) with an increase in total blood volume (19). The generalized fall in systemic vascular resistance has been attributed, in part, to increased production of vasodilators including nitric oxide and relaxin and decreased sensitivity to vasoconstrictors (5). For example, RAS hormone activity is increased and remains elevated throughout pregnancy, while the vascular response to angiotensin II is severely blunted (6, 19). Because of the direct relationship between vascular function and BP, failure of the physiological mechanism(s) that mediates the fall in peripheral vascular resistance in pregnancy may lead to hypertension in pregnancy and preeclampsia.

The pathogenesis of pregnancy-induced hypertension (PIH) and preeclampsia (PE) remain unclear. A variety of factors have been implicated in PE (29, 32), and in addition, both PIH and PE have been associated with elevated circulating levels of cardiotonic steroids (CTS)-like materials and other Na⁺-pump inhibitors (2, 15, 16, 20, 34, 43, 50). However, the cause-effect relationships of Na⁺-pump inhibitors with BP in pregnancy are unclear, in part because some inhibitors appear to be significantly elevated even in normal pregnancy (15, 34, 48). Further, many measurements of the Na⁺-pump inhibitors were made with either highly nonspecific or nonselective assays (9, 21). In addition, not all Na⁺-pump inhibitors, even when present in elevated amounts in the circulation, can raise BP (11, 30, 36).

The chronic administration of ouabain induces hypertension in normal rodents (10, 11, 31, 37, 55, 56). Work in transgenic mice (10, 11) and pharmacological studies in rats (26, 27) demonstrates that the mechanism of ouabain-induced hypertension is mediated by the Blaustein mechanism (4), i.e., inhibition of vascular myocyte α₂-Na⁺ pumps and the promotion of Ca²⁺ and Na⁺ entry via the vascular spliced isoform of the sodium-calcium exchanger NCX1.3 (11, 27, 27, 44) and members of the transient receptor potential channels (TRPC3 and TRPC6; Refs. 44, 54). In view of the abovementioned clinical and experimental associations of ouabain-like materials with high BP, and a defined long-term pressor mechanism for endogenous ouabain (EO), we set out to determine the circulating levels of EO in pregnancy by highly specific assay methods. We also sought insights into the possible pathological actions of EO by looking at the hemodynamic and vascular responses to long-term elevations of plasma ouabain in pregnant rats and their nonpregnant counterparts. Our working hypothesis was that the prolonged superelevation of circulating ouabain in pregnancy would stimulate the expression of a number of key cellular entry pathways for Ca²⁺ and/or Na⁺ (e.g., NCX1, TRPC3, and TRPC6), raise BP, and cause intrauterine growth retardation, i.e., a PIH or PE-like state.

METHODS

Animals.

Timed pregnant embryonic day 3 (E3) and size-matched virgin Sprague-Dawley female rats (200–225 g; Charles River, Wilmington, MA) consumed standard rat chow (2018SX; Harlan Teklad, Madison, WI) and tap water ad libitum. They were maintained on a 12:12-h light-dark cycle and allowed to acclimate for 1 day before the start of the study.

The animals were randomized to four groups and weighed, and their BPs were recorded. Group 1 (nonpregnant rats; n = 9) and group 2 (pregnant rats; n = 10) received subcutaneous miniosmotic pumps (Alzet model 2002, Durect, Cupertino, CA) containing 0.9% sterile saline while group 3 (nonpregnant + ouabain rats; n = 11) and group 4 (pregnant + ouabain rats; n = 8) were implanted with pumps containing 0.85 mM plant ouabain (Sigma, St. Louis, MO) in a 0.9% sterile saline 1 day after initial BP measurements were taken. Ouabain was delivered at a rate of 21 μg·kg⁻¹·day⁻¹. In some experiments, animals were killed, and following dissection, the maternal heart and kidney and individual placental and fetal units (with attached cords) were weighed. All animal procedures were reviewed and approved by the University of Maryland Institutional Animal Care and Use Committee.

BP measurements.

Systolic BP was measured by tail cuff using an IITC computerized BP monitor with automated cuff inflation and deflation (model 31; IITC Life Science, Woodland Hills, CA). Rats were preconditioned to tail cuff measurements. The animals basked under a 100-W lamp to stimulate tail blood flow. Initial measurements were taken at E4 or equivalent time in nonpregnant rats and E20 or equivalent. Once animals were acclimatized and BP readings were stable, three measurements were taken for each animal, and a mean value was recorded.

Solid phase extraction of blood and adrenal samples.

The methods used were similar to those described previously (37). Trunk blood was obtained from each rat by decapitation and collected in tubes with EDTA at E20 or the equivalent time point in nonpregnant rats. Blood was centrifuged at 3,000 g for 15 min, and plasma was separated and collected. Plasma volume was recorded, and a 0.1% solution of trifluoroacetic acid in nanopure water was added to plasma in a 1:1 ratio. For adrenals, the glands were excised, cleaned and rinsed in PBS, gently blotted, and weighed. Individual glands were homogenized in 1 ml 0.1% trifluoroacetic acid in water. Both plasma and adrenal samples were centrifuged for 30 min at 3,000 g, and the supernatants were applied to preconditioned solid phase extraction (SPE) C-18 columns (200 mg; Agilent, Santa Clara, CA). Columns were washed three times with 5 ml water and once with 3 ml 2.5% CH₃CN. EO and high polarity compounds were eluted with 3 ml 25% CH₃CN, and bufadienolide-like low polarity compounds were eluted with 3 ml 40% CH₃CN. The SPE eluates were dried by vacuum centrifugation for 18 h and reconstituted with nanopure water for radioimmunoassay (RIA) and radioreceptor assay (RRA).

RIA.

Aliquots (35 μl) of each reconstituted sample were incubated with 15 μl ³H-ouabain (30 Ci/mmol; Perkin Elmer) and 100 μl antibody mixture (2 μl polyclonal anti-ouabain serum, R7 primary antibody, and 38 μl coated anti-rabbit IgG secondary antibody in 30 ml of RIA buffer) at room temperature overnight (37). Samples were then harvested onto glass fiber filters (Brandel, Gaithersburg, MD) by being washed three times with nanopure water and placed in scintillation vials with 3 ml Safety-Solve (Research Projects International, Mount Prospects, IL) overnight. Bound ³H was determined by scintillation spectrometry in a β-counter (TA3000; Beckman Instruments). Samples were counted to accumulate 5–10K events (i.e., ∼2% coefficient of variation).

RRA.

Reconstituted samples (typically 312.5 μl plasma equivalents) were incubated with 40 μl ³H-ouabain (36.3 nM, 30 Ci/mmol, Perkin Elmer), 10 μl bovine adrenocortical membranes, and 125 μl MgPi buffer (pH 7.4) for 2 h at 37°C as described elsewhere (51). Nonspecific binding was determined by inclusion of 10 μM cold ouabain. Samples were harvested onto glass fiber filters (Brandel), and filters were soaked in vials with 3 ml scintillation solution (Safety-Solve, Research Projects International, Mount Prospect, IL). Bound ³H was determined by scintillation spectrometry (TA3000; Beckman Instruments, Fullerton, CA).

Liquid chromatography.

Equal sized pools of the SPE plasma and adrenal samples from pregnant and nonpregnant groups (typically 6 animals per pool) were chromatographed using a semipreparative (Ultrasphere, 10 × 250 mm) C18 high-performance liquid chromatography (LC) system (System Gold, Beckman Instruments, Fullerton CA) and a gradient solvent program similar to that described previously for human plasma EO (22). Thrity-second fractions (1.5 ml) were collected, dried by vacuum centrifugation, reconstituted in nanopure water, and assayed for the presence of ouabain immunoreactivity by offline RIA and offline mass spectrometry (MS).

MS.

Multidimensional MS analysis of the LC-separated SPE pools was performed offline using either a Bruker HCT Ultra or a Bruker Esquire LC ion trap mass spectrometer (Bruker, North Billerica, MA). In each case, the LC fractions were mixed with acetonitrile containing lithium carbonate and directly infused into the electrospray interface of the MS instrument. For EO and ouabain measurements, each LC fraction was monitored for molecular ions having a mass to charge ratio (m/z) equivalent to lithiated EO (i.e., m/z 591.3). Molecular ions at this m/z were then selected for collision-induced dissociation. The MS-MS product ion spectra were monitored for the lithiated steroid moiety of EO (or ouabain) at m/z 445.2. Molecular ions at this m/z were then selected for collision-induced dissociation, and the MS-MS-MS product ion spectra (i.e., MS3) were monitored for the most prominent product ion at 379.2 m/z corresponding to the removal of the lactone ring and sugar moiety from EO or ouabain. Calibration was performed by injecting known amounts of ouabain (1–100 fmol) into the MS and monitoring the intensity of the lithiated MS3 379.2 m/z product ion. Offline monitoring of LC fractions by MS was used to allow discrete interrogation of every stage of MS2 and MS3 operations and, when coupled with optimized spectral acquisition times and massive averaging, resulted in high quality spectra suitable for quantitation.

Western blot.

Mesenteric arteries and aortas from E20 rats or equivalent were carefully dissected, cleaned of fat and endothelium, and flash frozen in liquid nitrogen. Preparations were homogenized in ice-cold homogenization buffer (145 mM NaCl, 10 mM NaH₂PO₄, 10 mM NaN₃, 1% IGEPAL1, and EDTA-free protease inhibitor cocktail; Roche Diagnostics, Indianapolis, IN), pH 6.5, and centrifuged. Following protein quantification (Bio-Rad protein quantification kit; Bio-Rad, Hercules, CA), equal amounts of protein were loaded and separated by 7.5% Tris-glycine polyacrylamide gel electrophoresis (Bio-Rad) and transferred to a PVDF membrane (GE Health Care, Piscataway, NJ). Proteins were visualized using monoclonal mouse anti-NCX1.3 (Swant, Bellinzona, Switzerland); polyclonal rabbit anti-α₂-Na K-ATPase (Millipore, Billerica, MA); monoclonal mouse anti-α₁-Na,K-ATPase (gift of Dr. Thomas Pressley, Texas Tech University, Lubbock, TX); monoclonal mouse anti-sarco(endo)plasmic reticulum Ca²⁺-ATPase 2 (anit-SERCA2; Thermo Scientific, Rockford, IL); monoclonal mouse anti-plasma membrane Ca²⁺-ATPase (anti-PMCA; Thermo Scientific Rockford, IL); polyclonal rabbit anti-TRPC6 (Abcam, Cambridge, MA); and polyclonal rabbit anti-TRPC3 (Alomone, Jerusalem, Israel). Antibody against β-actin (Sigma-Aldrich) was used as a loading control. Bound antibody was detected with the ECL method (Millipore).

Statistical analysis.

Data (means ± SE) were analyzed using one-way and two-way ANOVA followed by paired and unpaired t-tests as appropriate (Systat, Evanston, IL). In those instances where group variances differed significantly, median values with 25 and 75% confidence intervals were compared by the Mann Whitney U-test. A P value of <0.05 was considered to be statistically significant.

RESULTS

Effect of prolonged infusions of ouabain on BP in pregnant and nonpregnant rats.

Figure 1A shows systolic BP during pregnancy and ouabain infusion. The baseline BPs taken on E4 (or time equivalent) were not significantly different among the four groups. Rats were then implanted with miniosmotic pumps containing ouabain or vehicle. Final systolic BP taken on E20 (or nonpregnant time equivalent) was significantly higher in nonpregnant rats infused with ouabain. In contrast, systolic BP fell similarly in pregnant rats and in pregnant animals chronically infused with ouabain. Figure 1B shows the change in systolic BP from the baseline for each group. Nonpregnant rats showed only a small nonsignificant increase in BP over the study period (3.1 ± 3.3 mmHg). In nonpregnant rats, the imposition of a ouabain infusion increased BP by 23.8 ± 3.4 mmHg (P < 0.001) compared with nonpregnant controls. In contrast, the BP in pregnant rats receiving ouabain declined 8.1 ± 3.2 mmHg from their baseline and the fall in BP was similar to that in vehicle-infused pregnant controls (−9.4 ± 2.6 mmHg; both P < 0.001 vs. nonpregnant). Thus chronic ouabain infusions raised BP in normal but not pregnant rats.

Fig. 1.

Systolic blood pressures in pregnant and nonpregnant rats and the impact of ouabain infusion. Baseline systolic blood pressures (BPs) were measured at embryonic day 4 (E4) or equivalent via tail cuff method. Mini osmotic pumps containing 0.85 mM ouabain or vehicle (0.9% saline) were implanted subcutaneously. Systolic BPs were taken at E4 and E20 or equivalent. BPs at both time points (A) and the change in BP (B) over the study are shown (means ± SE). *P < 0.001 vs. nonpregnant, pregnant, and pregnant + ouabain (O); **P < 0.001 vs. nonpregnant and nonpregnant + ouabain (O).

Plasma EO and ouabain levels in infused rats by RIA.

Figure 2 shows EO and/or ouabain levels in extracted plasma from pregnant rats at E20 or their equivalent nonpregnant time controls as determined by RIA. Plasma EO was significantly elevated in pregnancy compared with the nonpregnant group (1.14 ± 0.13 nM vs. 0.60 ± 0.08 nM ouabain equivalents, respectively; P < 0.05). In ouabain-infused nonpregnant rats, the total plasma ouabain + EO concentration reached 3.2 ± 0.3 nM and this value was significantly elevated compared with the total EO in both nonpregnant and pregnant controls (P < 0.001). The steady-state plasma levels in all ouabain-infused female rats were comparable to those previously reported in male rats with ouabain-induced hypertension (37). The plasma ouabain was significantly higher in infused pregnant (5.2 ± 0.4 nM ouabain equivalents; P < 0.01) vs. nonpregnant rats (3.2 ± 0.3 nM ouabain equivalents) receiving the same nominal dose of ouabain.

Fig. 2.

Plasma endogenous ouabain (EO) in pregnancy and the impact of prolonged infusion of ouabain. Data are for extracted plasma samples from E20 pregnant rats (see Fig. 1) or the equivalent time in nonpregnant rats (means ± SE). Average infused dose of ouabain (O) was 21.2 ± 1.2 μg·kg⁻¹·day⁻¹.

Polar and nonpolar CTS-like activity in pregnancy by RRA.

Figure 3 shows the polar CTS-like bioactivity in SPE-extracted plasma as determined by RRA. The EC₅₀ for ouabain in the RRA under the conditions used was 37 nM. Polar “ouabain-like” activity measured in pregnant rats at E20 (or equivalent) was dramatically higher (77 nM ouabain equivalents) vs. nonpregnant controls (12.9 nM). During chronic infusion of exogenous ouabain in nonpregnant and pregnant rats, the polar levels of CTS were 13.9 and 48.3 nM, respectively. Table 1 shows the bioactivity of nonpolar CTS-like materials present in the SPE plasma samples in our RRA system. In general, the measured bioactivity of the nonpolar SPE plasma samples was ∼7- to 70-fold lower than their corresponding polar SPE extracts, and no differences were found between the nonpolar treatment groups.

Fig. 3.

Plasma activity of polar cardiotonic steroids (CTS)-like materials in pregnancy. Extracted plasma samples were collected from E20 pregnant rats or the equivalent time point in nonpregnant rats that were infused for 2 wk with ouabain as indicated and measured by a radioreceptor assay standardized with exogenous ouabain. Results are expressed as ouabain equivalents (means ± SE). *P < 0.01 vs. nonpregnant.

Table 1.

Nonpolar CTS-like activity in plasma by RRA

Group	n	Nonpolar CTS-Like Activity, nmol ouabain equivalents
Nonpregnant	4	1.59 ± 0.67
Pregnant	6	0.92 ± 0.16
Nonpregnant + ouabain	5	1.06 ± 0.26
Pregnant + ouabain	6	1.80 ± 0.16

Data are means ± SE; n = number of animals used. Nonpolar cardiotonic steroids (CTS)-like materials were derived from solid phase extraction (SPE) plasma from the indicated groups. Apparent bioactivity of the plasma extracts was derived by interpolation from a ouabain standard curve. RRA, radioreceptor assay.

Circulating EO and its isomers in pregnancy.

Figure 4 shows EO in pregnant and nonpregnant plasma samples as determined by both offline LC-MS3 and offline LC-RIA. Figure 4A shows the high performance LC elution of plant ouabain with elution beginning at 34.0 min, peaking at 34.3 min and baseline at 34.8 min. As there is a small ∼30-s delay from the ultraviolet detector to the fraction collector, ouabain (and EO) will appear at the LC fraction collector beginning at 34.5 and ending at 35.3 min. Thus the peak concentrations of ouabain and EO are expected in LC fraction 34.5 (which covers the time bin from 34.5–35.0 min) with some residual material in fraction 35.

Fig. 4.

Detection and quantitation of plasma EO in pregnancy by offline liquid chromatography (LC)-RIA and LC-mass spectrometry (MS). High performance-LC of ouabain. A, left: sharp elution of exogenous ouabain measured at 220 nm (red) and 300 nm (green) in response to an acetonitrile gradient. A, right: enlargement showing elution of ouabain begins and ends at 34 and 35 min with a maximum at 34.3 min. B, top: offline LC-MS3 screening and quantitation for molecular product ions of EO (379.2 m/z) in solid phase extraction (SPE) plasma samples from nonpregnant and pregnant rats. Peaks corresponding to EO were found at 34.5 min in samples from nonpregnant and pregnant rats. Additional and/or more prominent peaks corresponding to EO isomers were found at 32.5 and 38 min in pregnancy. B, bottom: offline LC-RIA of the same SPE plasma samples screened for ouabain immunoreactivity and the dashed line is ∼2 SD above background. In each case values are expressed as femtomoles ouabain equivalents per fraction. C: individual spectra showing ESI-MS3 molecular product ions produced by 5 nM ouabain (left) and the offline LC fractions at 34.5 (middle) and 38 min (right) originating from the LS separation of SPE plasma samples from pregnant rats in B.

Figure 4B, top, shows the results of the offline MS3-based interrogation of the LC separation of the SPE plasma pools from pregnant and nonpregnant rats. Prominent MS3 product molecular ions corresponding to EO were observed in fraction 34.5 in both pregnant and nonpregnant samples. The EO content was significantly higher in the pregnant group (1,044 fmol/fraction) vs. the nonpregnant (198 fmol/fraction). An additional MS3 molecular product ion whose m/z values are identical to EO and is thus an EO isomer were detected peaking in fraction 38 in both pregnant and nonpregnant groups. This F38 isomer is slightly less polar than EO on C18 and its plasma levels were increased ∼90-fold in pregnancy. In addition to this remarkably prominent isomer, another less abundant isomer that was slightly more polar than EO on C18, was detected in fraction 32.5. Figure 4B, bottom, shows the results of screening the same LC fractions by ouabain RIA. As expected, the peak immunoreactivity representing EO eluted in fraction 34.5 and, in full agreement with the LC-MS3 data, the plasma levels were increased in pregnancy. In the nonpregnant state, the major peak of ouabain immunoreactivity accounted for 44.51% of the total immunoreactivity in plasma. In pregnancy, the peak at fraction 34.5 explained 60% of the total immunoreactivity while the peak at 38 min accounted for 20% of the total immunoreactivity. Additional small peaks of immunoreactivity were observed in both sample groups, but there was no obvious correlation between them with the possible exception of fraction 38. When the ouabain immunoreactivity in F38 was compared with the MS3 quantitation, the apparent crossreactivity of the F38 isomer relative to ouabain was ∼0.5%. The low crossreactivity explains why it has not been previously detected in RIA-based studies. A small peak of immunoreactivity was also noted at 33.5 min in the nonpregnant sample, and this retention is similar to both the dihydro derivative of EO and the aglycone of EO (not shown). However, we were unable to detect molecular ions of those entities at their expected m/z ratios by MS2 and MS3. Hence, the general nature of these and the other small peaks of ouabain immunoreactivity scattered throughout the run is not known. No above threshold MS3 or immunoreactive peaks were observed in blank runs when SPE extracts of water were carried through offline LC and subsequent assay (not shown).

Figure 4C shows the MS3 spectra for plant ouabain, EO in fraction 34.5, and the prominent EO isomer eluting in fraction 38. In each spectra, identical signature MS3 product ions at 427.3, 409.3, 397.3, 383.3, 379.3, and 361.3 were observed. Unique to the isomer in fraction 38 is the appearance of an additional prominent molecular product ion at m/z 417.3. Note also that the absolute abundance of all signature MS3 ions in the F38 sample is dramatically greater than those for EO in fraction 34.5 reflecting its dramtically increased presence in pregnancy. The 417.2 m/z MS3 molecular ion in fraction 38 was absent in the EO isomer in fraction 32.5. Thus there are, in addition to EO, two chromatographically distinct isomers of EO present in pregnancy, the most prominent of which can also be distinguished from EO also by its unique MS3 spectrum.

Adrenal EO in pregnancy.

Table 2 shows the average weight of the adrenals was greater (39.3 ± 1.3 mg) in pregnant vs. nonpregnant (33.2 ± 1.6 mg; P < 0.05) rats. In spite of the adrenal enlargement, the ratio of adrenal weight to total body weight was significantly less in pregnant vs. nonpregnant rats (0.011 vs. 0.013%; P < 0.05). Figure 5 shows the EO content in the maternal adrenals measured by both offline LC-MS3 (A) and LC-RIA (B). The major peak of EO immunoreactivity was observed in fraction 34.5 in samples from both pregnant and nonpregnant rats and explained 76.5 and 38.3% of the total recovered immunoreactivity, respectively. The adrenal EO content in pregnancy was elevated as determined by both LC-RIA and LC-MS3 (MS3 values: 528.2 vs. 194.3 fmol/fraction). Similar to the plasma samples, the same EO isomer detected by MS3 in fraction 38 was also present the adrenal samples from pregnant and nonpregnant rats. However, the adrenal levels of this EO isomer were 100-fold less abundant than in plasma and were not increased in pregnancy. In addition, the levels and crossreactivity of this isomer were not sufficient to be detected with the RIA.

Table 2.

Impact of pregnancy on adrenal mass

	Nonpregnant (n = 6)	Pregnant (n = 7)
Mean animal weight, g	256.7	342.9
Mean adrenal weight, mg	33 ± 2	39 ± 1*
Ratio (adrenal/animal wt), %	0.013%	0.011%*

Data are means ± SE; n = number of animals used. Maternal right adrenal weight and total body weight were measured at embryonic day 20 or equivalent in pregnant and nonpregnant rats, and the ratio of adrenal weight to total body weight was calculated.

^*P < 0.05 vs. nonpregnant.

Fig. 5.

Analysis of rat adrenal SPE samples by offline LC-MS3 and LC-RIA. A: quantitation of adrenal EO by offline LC-MS3 of SPE samples from nonpregnant and pregnant samples. Quantitation is based upon the intensity of the main product ion at a mass to charge ratio of 379.2. B: offline LC separation and RIA based quantitation of adrenal SPE from nonpregnant and pregnant rats. Ouabain immunoreactivity is expressed as ouabain equivalents (fmol) per fraction. In each case, the dashed line is ∼2 SD above background.

Effect of pregnancy and ouabain infusion on vascular expression of ion transporters, cardiac and renal mass and placental and fetal growth.

Figure 6 shows the Western blot analysis for several vascular proteins involved in Ca²⁺ and Na⁺ transport normalized by actin controls in mesenteric arteries and aorta. In pregnancy, NCX1.3 expression decreased by 30% in mesenteric artery (Fig. 6A) and 60% in aorta (Fig. 6B) compared with vessels from nonpregnant animals (both P < 0.001). Interestingly, the expression of α₂-Na⁺ pumps did not change in mesenteric arteries (Fig. 6C) but was decreased by 30% in aorta (Fig. 6D) during pregnancy (P < 0.0001 vs. nonpregnant). However, in pregnancy the expression of α₁-Na⁺ pumps was decreased by 50% in both mesenteric arteries (Fig. 6E) and aorta (Fig. 6F; P < 0.001 and P < 0.05, respectively) compared with nonpregnant vessels. Protein expression of SERCA2, PMCA, TRPC-6, and TRPC-3 was not altered in response to pregnancy in mesenteric arteries or aortas (Table 3).

Fig. 6.

Effect of pregnancy on expression of vascular sodium calcium exchange (NCX1.3) and Na⁺ pump isoforms in aorta and pooled mesenteric arteries. Western blots showing NCX1.3 expression in mesenteric artery (A) and aorta (B). α₂-Na⁺ pump expression in mesenteric artery (C) and aorta (D) and α₁-Na⁺ pump expression in mesenteric artery (E) and aorta (F). In each image, arteries for nonpregnant arteries are at left, and their pregnant counterparts are at right. Bottom bands are β-actin expression and used as a reference in each case. *P < 0.001 (A and B), *P < 0.0001 (D), *P < 0.01 (E), *P < 0.05 vs. nonpregnant vessels (F).

Table 3.

Expression of aortic and mesenteric artery myocyte Ca ²⁺ transporters in normal pregnancy

Ca2+ Transport Protein	%Expression Aorta (n = 6)	%Expression Mesenteric Artery (n = 6)
PMCA	113 ± 7	96 ± 6
SERCA2	96 ± 6	98 ± 1
TRPC6	92 ± 1	99 ± 3
TRPC3	93 ± 5	100 ± 3

Data are means ± coefficient of variation; n = number of animals used. Data show the arterial expression of additional transport proteins in pregnant animals normalized to their respective nonpregnant counterparts. PMCA, plasma membrane Ca²⁺-ATPase; SERCA2, sarco(endo)plasmic reticulum Ca²⁺-ATPase 2; TRPC, transient receptor-operated channel.

The Western blot analysis documenting the effect of ouabain infusions in nonpregnant and pregnant rats is shown in Figure 7. Relative to their nonpregnant controls, pregnant rats showed a 45% (aorta) and 35% (mesenteric artery) decrease in NCX1.3 expression with no change in TRPC6 expression. In response to ouabain infusion that generated the superelevated plasma levels shown in Fig. 2, the expression of NCX1.3 was significantly increased in both nonpregnant and pregnant animals. The levels of NCX1.3 expression were comparable in both ouabain-infused groups but were highest (∼1.7-fold) in the aorta of nonpregnant rats. However, the magnitude of the increase in NCX1.3 expression was typically greater than twofold in the ouabain-infused pregnant rats because the baseline expression in pregnancy was lower. Expression of TRPC6 was also increased in both ouabain-infused groups.

Fig. 7.

Effect of ouabain infusion on NCX1.3 and transient receptor-operated channel 6 (TRPC6) expression in aorta and mesenteric arteries. Western blots showing NCX1.3 expression in aorta (A) and mesenteric artery (B) and TRPC6 expression in aorta (C) and mesenteric artery (D). Bottom band is GAPDH. NP, nonpregnant; P, pregnant; NP + O, ouabain-infused nonpregnant; P + O, ouabain-infused pregnant. A: *P < 0.01 vs. NP, NP + O, and P + O; **P < 0.01 vs. NP, P, and P + O; ***P < 0.01 vs. NP, P, and NP + O. B: *P < 0.01 vs. NP, NP + O, and P + O; **P < 0.01 vs. NP and P. C: *P < 0.01 vs. NP, P. D: *P < 0.01 vs. NP, P, and P + O; **P < 0.01 vs. NP, P, NP + O.

As shown in Figure 8, significantly lower median placental and fetal weights were found in ouabain-infused pregnant rats, indicating reduced growth. The use of median as opposed to mean values reflects the greater variance in the ouabain-infused group. In the ouabain-infused group, placentas lacking attached fetal tissue were noted (arrows). These placentae were much smaller than their attached counterparts and likely reflect prior fetal death and resorption. Consistent with this interpretation (Table 4), the number of placentae was normal while the number of fetal units per mother was less in the ouabain-infused group. Ouabain infusions led to lower maternal body weight; most of which was due to reduced placental and fetal mass. The number of observed placentae was not different in ouabain-infused rats, reflecting the imposition of the high ouabain state well after conception and implantation. However, the 2-wk period of ouabain administration reduced the number of fetal units. No effects of ouabain administration were observed on maternal heart or kidney weights.

Fig. 8.

Effect of ouabain infusion on placental and fetal weight in pregnancy. Distribution of placental (top) and fetal weights (bottom) in ouabain- and vehicle-infused (controls) pregnant rats. All infusions started on E4 (i.e., post conception and implantation) and measurements were taken on E19. Boxes show the median values (middle line) and their 25 (bottom line) and 75% (top line) confidence intervals. Placental numbers are ouabain infused (n = 54) and controls (n = 141). Fetal data are ouabain-infused (n = 51) and controls (n = 141). Arrows indicate placentae with no accompanying fetus (NF). The variance of the values is greater and downwardly skewed in the ouabain-infused animals and the median values are significantly lower as indicated by the relevant P values.

Table 4.

Impact of ouabain infusion on maternal, placental, and fetal parameters in normal pregnancy

Parameter	Controls	Ouabain Infused	Significance (P Value)
Maternal weight	351 ± 2.1	338 ± 5.3	0.015
No. placentas/mother	14.2 ± 0.2	13.5 ± 0.3	NS
No. fetal units/mother	14.2 ± 0.2	12.75 ± 0.48	0.006
Maternal heart weight, mg	995 ± 63	990 ± 37	NS
Maternal left kidney weight, mg	947 ± 20	947 ± 18	NS

Data are means ± SE. NS, not significant.

DISCUSSION

One of the initial goals of this study was to investigate circulating EO in pregnancy using highly specific analytical assay methods. In addition, we sought insights into possible pathological actions of EO by looking at the hemodynamic and arterial myocyte protein responses to long-term elevations of plasma ouabain in pregnant rats and their nonpregnant counterparts. Our expectation was that the prolonged superelevation of circulating ouabain during pregnancy would eventually lead to a PE-like state with elevated BP, proteinuria, and intrauterine growth retardation (IUGR). As investigation of the mechanism of PIH and PE is difficult in humans, we anticipated that a ouabain-induced model of PIH or PE would be useful in understanding more about the functional significance of the reported association of endogenous CTS, and EO in particular, in pregnancy-related hypertension. Indeed, we show analytical proof that pregnancy per se is a state with elevated circulating EO and is accompanied by significant amounts of two novel isomers of EO. Further, the circulating levels are further superelevated by chronic ouabain infusion. However, and to our great surprise, pregnant rats not only developed resistance to the chronic vasopressor effects of their own EO but also to the addition of ouabain, so that BP declined normally during pregnancy in both groups. However, the ouabain-infused mothers-to-be showed increased arterial NCX1.3 and TRPC6 expression and reduced placental growth and IUGR. Thus the superelevated plasma levels of ouabain (and by inference EO in PIH and PE) are of pathological consequence for the fetus in pregnancy. Further, the mechanisms that lead to hypertension and IUGR are, in principle, separable.

Endogenous CTS and BP in pregnancy.

A number of observations are pertinent to our study paradigm. When compared with paired nonpregnant time controls, all our pregnant rats experienced a modest but significant decline in BP, in agreement with other studies in rats (3, 7, 12, 39) and in humans (6, 13, 19). Pregnancy is associated with vasodilation, volume expansion, and a small decrease in BP, and, despite increases in plasma renin activity, angiotensin II, aldosterone, and norepinephrine, the pressor response to angiotensin II and norepinephrine is dramatically decreased (5, 8, 18, 40). The decline in peripheral vascular resistance in the gravid circulation is likely multifaceted, involving the creation of a low resistance placental vasculature as well as decreased maternal vascular tone and reactivity.

Prior work (2, 9, 21, 34, 43) has suggested that CTS-like materials are increased in pregnancy and further raised in PE where they often correlate positively with BP. However, none of those prior studies employed analytical assay systems. Instead, most used highly nonspecific assay systems to measure ouabain-like materials (9, 20). Those studies using more specific immunoassays disagreed on whether EO either was (48) or was not (34) increased in pregnancy. Other work has also implicated marinobufagenin (MBG), a bufodienolide, in the clinical pathogenesis of PE (1, 34) and elevated levels of MBG and digoxin-like materials have been described in pregnancy and become superelevated in PE (20, 34, 43). We did not measure marinobufagenin or digoxin in this study; however, our radioreceptor assay of the nonpolar SPE samples (which would include digoxin and marinobufagenin) suggests that these materials in rat plasma during normal pregnancy are collectively much less biologically active in our RRA system compared with their polar SPE (“ouabain-like”) counterparts. Also, the bioactivity of the nonpolar materials in our RRA system was not elevated, either by pregnancy or by chronic ouabain infusion, although it is important to realize that our RRA (EC₅₀ ∼37 nM ouabain) is relatively insensitive. Thus there may be nonpolar CTS entities that are elevated in pregnancy, but these would have to exhibit either a much lower binding affinity than ouabain and/or be present in amounts much lower than EO.

Pressor mechanism of EO and ouabain and the impact of pregnancy.

Elevated levels of EO circulate in 40–50% of Caucasian patients with essential hypertension and correlate positively with BP and left ventricular mass (37, 38, 42, 45). Moreover, hypertension can be induced in nonpregnant rodents by chronic low dose administration of ouabain in the periphery (11, 30, 31, 37, 55, 56). Blaustein (4) proposed that chronic volume expansion raises the plasma levels of an Na⁺-pump inhibitor that secondarily augments Ca²⁺ entry via NCX1.3. The resultant increases in cytoplasmic Ca²⁺ and vascular tone raise BP. Our observation of elevated levels of EO in the high fluid volume state of pregnancy supports the initial feature of this hypothesis while the increase in ACTH that occurs in pregnancy also is a plausible explanation.

Additional independent lines of research confirm the specific roles of the α₂-Na⁺ pumps NCX1.3 and TRPC6 in raising vascular tone and BP in ouabain-induced hypertension, and some forms of salt-sensitive and ACTH-dependent hypertension (10, 11, 26, 2758). In addition, reduced expression of NCX1.3 is linked with salt-resistance, low BP, and inability to develop hypertension (27). Thus there is a large body of experimental evidence showing that the level of NCX1.3 expression (and hence the amount of Ca²⁺ influx) has a profound influence on BP. Figure 9 provides a summary of the mechanism for the hypertensinogenic effects of ouabain in the nonpregnant state and shows the changes we observed in normal pregnancy. The upregulation of vascular NCX1.3 expression in high ouabain models of hypertension (44, 58) led us to expect that NCX1.3 would be similarly increased in pregnant rats where EO is physiologically elevated. Thus we were surprised to find that both the vascular expression of NCX1.3 and BP declined in pregnancy irrespective of the high ambient EO. The behavior of NCX1.3 and BP in the pregnant state is thus highly consistent with the abovementioned view that NCX1.3 expression is a key determinant of BP. However, our results make it clear that the gravid state in both rodents and humans is apparently a unique physiological condition in which plasma EO is inversely related with BP. Notably, the decline in the vascular expression of NCX1.3 in pregnancy was specific; the expression of other key Ca²⁺ transporting proteins measured (SERCA2, PMCA, TRPC3, and TRPC6) was unchanged. The downregulation of both vascular isoforms of the Na⁺ pump further suggests that the overall demand for monovalent ion transport and vascular metabolism is diminished in pregnancy.

Fig. 9.

Summary of the chronic vascular pressor mechanism of EO and ouabain in nonpregnant rats and comparison with pregnancy-evoked changes. Left: known mechanism linking elevated EO (or ouabain) with increased BP in nonpregnant rodents (see text for references). Major events leading to increased BP are shown in red boxes. Binding of EO increases the concentration of EO-α2 Na⁺ pump complexes causing reduced pumping activity and activation of a phosphorylated form of SRC (pSRC; Ref. 54). pSRC-triggered events include altered transcription, translation and trafficking. Thus pSRC provides the most plausible link between elevated plasma EO and enhanced expression of NCX1.3, TRPC3, and TRPC6, and hence increased Ca²⁺ influx, raised arterial tone and elevated BP. Increased expression of α₂-Na⁺ pumps (dashed line) is a feedback mechanism that helps restore Na⁺ pump activity towards normal. Right: mechanism showing changes in normal pregnancy. Red, green, and blue boxes show events that raise, lower, or have no impact on BP, respectively. Elevated levels of circulating EO evoked by pregnancy increase EO-α₂-Na⁺ pump complexes, reduce Na⁺ pump activity, and activate pSRC. Pregnancy-related factors (yellow boxes) impact the relationship between EO and BP. Box 1 are factors (e.g., ACTH and volume) thought to drive the pregnancy-evoked rise in EO. Box 2 factors modify transcription, translation, and trafficking events, reduce the impact of pSRC signaling on expression of NCX1.3 and α₂-Na⁺ pumps, and lower Ca²⁺ influx. Box 3 factors reduce the Ca²⁺ sensitivity of downstream events that impact contractility. The origin and identities of the boxed factors is under investigation.

In light of the abovementioned parallel declines in NCX1.3 and BP in normal pregnancy, we were surprised when the infusion of exogenous ouabain raised NCX1.3 and TRPC6 expression in the aorta and mesenteric artery in pregnant rats. Also, the magnitude of the increase in expression was similar to that in nonpregnant female rats that developed hypertension. Taken together, these results indicate that the pregnancy-related factors that suppress NCX1.3 expression in normal pregnancy (Fig. 9) can be overcome by superelevated levels of ouabain. Thus the results from the ouabain infusions imply that the threshold concentration of plasma EO needed to upregulate NCX1.3 expression in pregnant rats is higher than is achieved physiologically in pregnancy; indeed they are more representative of the superelevated levels of EO in PIH and PE. Another important observation is that BP rose (>22 mmHg) in the nonpregnant rats given ouabain infusions as expected but declined normally in the infused pregnant rats as shown in Fig. 1, A and B. Since NCX1.3 and TRPC6 were significantly elevated in the infused pregnant rats, the pressor resistance to ouabain (and EO) must be strongly influenced also by events that are downstream of NCX1.3-mediated Ca²⁺ influx.

Ca²⁺ is the trigger for vascular contraction and reduced NCX1.3 activity and hence Ca²⁺ influx will almost certainly underlie part of the active vasodilation, reduced reactivity to agonists, and diminished phenylephrine-stimulated Ca²⁺ entry in aorta in pregnancy (7). Events downstream from Ca²⁺ entry including reduced Ca²⁺ sensitivity and myogenic tone have been described in uterine arteries during pregnancy (52, 53). The extent to which the functional changes in uterine arteries occur in other arterial beds such as those we used in our studies is not known but appears likely as vascular reactivity is reduced in many arterial segments. The increases in plasma ouabain in the infused animals were similar to those previously reported in male rats (37). Further, the doses we infused, as well as the steady-state circulating levels achieved, are maximally pressor (35). Hence, the ease with which ouabain induced hypertension in nonpregnant females (Fig. 1) is in striking contrast to the prominent resistance of the pregnant females. Is the resistance to the vasopressor actions of ouabain applicable to other CTS? MBG administration per se raises BP and is claimed to create a PE-like syndrome in pregnant rats (24). Thus the answer appears to be no. Alternative explanations include the much greater lipophilic nature of MBG, which may allow easy entry into cells and, that MBG may have an additional mechanism(s) of action in a stress signaling pathway (47).

Maternal, placental, and fetal impact of superelevated plasma ouabain in pregnancy.

In spite of the inability of ouabain infusions to raise BP in pregnancy, the superelevated plasma levels of this CTS were not without deleterious effects. Modest but distinct decreases in maternal weight gain, reduced placental mass, and IUGR were noted at death. While no fetal structural abnormalities were noted, some infused animals had small placentae that lacked attached fetal structures. Thus, in addition to the IUGR, some fetal mortality had occurred with fetal resorption; this interpretation is consistent also with a significant decrease in the average number of fetal units in the ouabain-infused rats (Table 4). Our observations that elevated ouabain has both adverse effects and undesirable outcomes in normal pregnancy contrasts with other work in a malnutrition model where comparable infusions of ouabain restored embryonic kidney development (33). Worldwide, malnutrition is extraordinarily common and, among many effects, results in low nephron numbers, increased susceptibility to renal injury, and disease. The suggestion that ouabain be considered as a therapeutic tool in malnourished pregnant individuals (33) is intriguing but based on our work deserves caution.

EO and EO isomers in pregnancy.

To explore the specific nature of the materials measured in the plasma of the pregnant rats with our ouabain RIA, we used liquid chromatography and mass spectrometry (Fig. 4). As shown in Fig. 4B, MS3 studies confirmed the elevated circulating levels of EO in pregnancy reported by the RIA (Fig. 2) and also, for the first time, we detected two pregnancy-specific novel isomers of EO, one of which was dramatically elevated in the circulation of pregnant rats. We also observed elevated EO in the adrenal glands of pregnant rats. Several conclusions can be drawn from these observations: 1) as determined by LC-MS3, the new compounds are isobaric with EO and plant ouabain and hence are highly structurally similar. They have identical parent mass to charge ratios, dissociate to form the same signature product ions in both MS2 and MS3, and are, therefore, steroidal isomers of EO. The isomers are distinct from all known ouabain- or digoxin-like compounds and, by virtue of their parent mass and chromatographic polarity, are very different from the 370-molecular weight bufadienolide-like material isolated from placenta (23) and the pregnancy-related digoxin-like materials (20). 2) Furthermore, the EO isomers are readily distinguished from EO and ouabain; one is slightly more polar and the other slightly less polar than EO. 3) Neither of the EO isomers was readily detected by our ouabain antibody when assayed in the LC-RIA format. Taking the most prominent EO isomer in fraction 38, the apparent crossreactivity (based on the amounts reported by the LC-MS3 and LC-RIA methods) is ∼0.5% of ouabain itself. 4) The biological potency of these isomers, especially that in fraction 38, seems the most likely explanation for the increased CTS-like activity detected in the polar SPE samples from pregnant rats by the RRA (Fig. 3). Comparison of the MS and RRA data shows that the biological activity of these isomers is ≥10-fold weaker than ouabain or EO; pure materials will be needed to determine their precise bioactivity. 5) Both EO isomers were found in elevated amounts in the circulation but, in contrast to EO, not in the adrenal glands of pregnant rats (c.f, Figs. 4B and 5). This suggests that EO is of adrenal origin, while the bulk of the novel EO isomers must originate elsewhere, probably from the fetoplacental unit. The functional role of these isomers in pregnancy is unexplored; one intriguing possibility, by analogy with the anti-ouabain compound PST2238 (17), is that they are endogenous EO antagonists that may help to block the pressor effect of high plasma EO in the maternal circulation. Another, possibility is that the physiologically elevated levels of these EO isomers support normal growth and development of the fetus (33). Elucidation of their structures and chemical synthesis will be needed before their role in pregnancy can be further explored.

Endogenous and exogenous CTS as agonists and antagonists in experimental models of PE.

There are two contemporary models of preeclampsia in the rat, the DOCA-salt paradigm (24, 25, 50) and the reduced uterine perfusion pressure model (29, 35). There are no measurements of EO or assessment of its vasopressor activity in either model. However, it is noteworthy that the bufadienolide resibufagenin (RBG) prevents or reverses the PE-like syndrome in the DOCA-salt model where MBG is thought to be elevated (24, 49). By itself, this observation would not be especially noteworthy were it not for the fact that the antihypertensive effect of RBG is strikingly reminiscent of the antagonism by digoxin and its analogs of ouabain-induced hypertension (30, 36). Further, it is remarkable that the same structural features may explain the effect: for example, digoxin, RBG, and PST2238, all of which are antihypertensive, share a 5β-oriented proton in the A ring while the pressor CTS whose hemodynamic effects they antagonize almost invariably have a 5β-hydroxyl group. Thus it is clear that within both the cardenolides and bufadienolides, there are hitherto unrecognized structure-specific bioactivities. One explanation for the differences in the long-term hemodynamic effects of EO vs. digoxin, or MBG vs. RBG, is the observation that these CTS prefer one or the other of the two vascular Na⁺ pump α-subunit isoforms. For example, MBG may be selective for the rat α₁-isoform while ouabain (and EO) prefer the α₂-isoform (14). Yet another explanation, and one we favor, is that ligand binding to a single receptor may generate multiple signals. An example is the β-adrenergic receptor which produces a variety of ligand-specific signaling events (28). Indeed, ouabain and digoxin both inhibit Na⁺ pumps with similar affinity, yet paradoxically, nanomolar concentrations of digoxin block ouabain-induced signaling in arterial myocytes in which the α₂-Na⁺ pump isoform is the only feasible target (58). Thus the α₂-Na⁺ pump distinguishes among similar CTS ligands and generates different signal events (58) that lead to very different hemodynamic outcomes (30, 36). This may be of particular importance because both EO and/or MBG are potential vasopressor agents in PE and the paradoxical effects of digoxin and RBG may therefore be therapeutically relevant. Another implication of the present work is that in PE, the mechanism for the diminished vasopressor response to CTS is overcome otherwise the positive relationship between EO and BP in the clinical studies would not be present. By analogy, the various factors currently implicated in PE such as angiotensin II receptor autoantibodies, sFlt-1, and/or endothelin (32) would not be expected to have significant vasopressor effects in pregnancy until such time as the mechanisms for the diminished vascular contractility can be defeated. Our results highlight the importance of knowing more about the fundamental physiology of pregnancy as a way to gain new insights into the etiologies of PIH and PE and their potential therapy.

Conclusions.

Normal pregnancy is a ouabain-resistant state associated with physiologically elevated levels of circulating EO and two novel isomers. This physiologically high EO state has low BP with no adverse maternal or fetal consequences. The paradoxical inverse relationship between EO and BP reflects functional reprogramming of the arterial vascular system in pregnancy that switches off the long-term pressor effects of EO (and ouabain) to protect the mother and the fetus. The molecular mechanism of the vasopressor resistance involves altered Ca²⁺ transporting proteins and reduced efficacy of downstream Ca²⁺ mechanisms that support arterial contraction and BP. In the presence of higher, pathologically relevant levels of plasma ouabain (i.e., EO in PIH and PE), there is impaired fetal and placental growth. The mechanisms that govern plasma EO and cellular Ca²⁺ and the arterial resistance in pregnancy may provide a basis for understanding more about the abnormal vascular function in PIH and PE.

GRANTS

This study was supported in part by US Public Health Service Grants HL-75584, HL-045215, and HL-078870 (to J. M. Hamlyn) and National Institute of General Medical Sciences Initiative for Maximizing Student Development Grant R25-GM55036 (to B. E. Jacobs).

DISCLOSURES

B. E. Jacobs is the recipient of an International Travel Award from Proctor and Gamble.

AUTHOR CONTRIBUTIONS

Author contributions: B.E.J. and J.M.H. conception and design of research; B.E.J., Y.L., M.P., and J.M.H. performed experiments; B.E.J., Y.L., and J.M.H. analyzed data; B.E.J., V.A.G., and J.M.H. interpreted results of experiments; B.E.J. and J.M.H. prepared figures; B.E.J. and J.M.H. drafted manuscript; B.E.J., V.A.G., and J.M.H. edited and revised manuscript; J.M.H. approved final version of manuscript.