Research ArticleEndocrinology and Metabolism

De Novo Biosynthesis and Radiolabeling of Mammalian Digitalis-Like Factors

Hassan M.A.M. Qazzaz, (Tim) Zhimin Cao, Duane D. Bolanowski, Barbara J. Clark, Roland Valdes
DOI: 10.1373/clinchem.2003.022715 Published February 2004

Abstract

Background: Digoxin-like immunoreactive factors (DLIFs) are endogenous mammalian cardenolides with structural features similar to those of the plant-derived digitalis compounds. DLIFs and their structurally related forms (Dh-DLIFs) may serve as effectors of ion-transport activity mediated by their interaction with Na,K-ATPase and thus play a role as a new hormonal axis. Although some evidence implicates the adrenal gland as a tissue source for the DLIFs, little is known about the biosynthetic pathway producing these compounds. We now demonstrate de novo biosynthesis of DLIF by incorporation of radioactive carbon (14C) into the structures of both DLIF and Dh-DLIF.

Methods: We used a combination of reversed-phase HPLC techniques to separate the radioactive DLIF components after incorporation of 14C into their structure by use of either [1,2-14C]acetic acid or [4-14C]cholesterol as precursors and a Y-1 mouse adrenocortical tumor cell line. We also stimulated and suppressed production of steroidogenesis by use of cAMP analogs and Mevastatin, respectively, to demonstrate the dependence of DLIF production on the cholesterol-dependent biosynthetic pathway. A combination of chromatographic mobility, immunoassays specific for digoxin and dihydrodigoxin, and deglycosylation using 5-sulfosalicylic acid were used to identify the DLIF and Dh-DLIF components.

Results: With cholesterol as precursor, the cells produced DLIF (7.5 mCi/mmol) with a labeling efficiency of 10%, whereas with acetate the cells produced DLIF (72.2 mCi/mmol) with a labeling efficiency of 0.08% of the total DLIF produced. The radiolabeled DLIF and Dh-DLIF molecules had identical chromatographic mobilities and stoichiometric removal of sugars as the previously characterized DLIFs isolated from different mammalian species and tissues. With radioactive cholesterol as precursor, the 14C was incorporated into the DLIF-genin portion of the compounds and not the sugars. Interestingly, treatment of Y-1 cells with 8-bromoadenosine 3′:5′-cAMP to stimulate steroidogenesis did not increase production of DLIF or Dh-DLIF but did increase production of progesterone. Mevastatin (5 μmol), an inhibitor of the enzyme hydroxymethylglutaryl-CoA reductase and thus of cholesterol biosynthesis, gave an 85% decrease in the production of 14C-DLIF and progesterone, but only a modest 15% decrease in 14C-Dh-DLIF production.

Conclusions: These data demonstrate that the adrenal cell has the cellular machinery necessary for de novo biosynthesis of DLIF and Dh-DLIF starting from a simple carbon pool and also support the concept that cholesterol is a major precursor of the DLIF compounds. This cell culture model provides a source of radiolabeled DLIF compounds for future experimental work.

Digitalis-like immunoreactive factor (DLIF)2 and its structurally related form dihydrodigoxin-like immunoreactive factor (Dh-DLIF) are endogenous mammalian cardenolides with molecular features remarkably similar to those of the plant-derived cardiac glycosides digoxin and dihydrodigoxin (1)(2). Cardiac glycosides regulate the activity of the sodium pump (Na,K-ATPase), and the α-subunit of Na,K-ATPase is a recognized functional receptor for these compounds (3)(4)(5)(6)(7). Control of sodium pump activity is believed to be an underlying mechanism in the pathophysiology of several diseases, including cardiovascular, neurologic, renal, hepatic, psychiatric, and metabolic disorders [for reviews, see Refs. (8),(9)]. The source of mammalian digitalis-like factors in a hormone-secreting tissue such as the adrenal, their presence in blood, a recently identified protein binder/carrier in serum (10)(11)(12)(13), and the tissue-specific distribution of Na,K-ATPase isoforms (14) strongly suggest the existence of a hormonal axis linking the adrenal cortex, the endogenous digitalis-like factors, and control of sodium pump activity (15)(16). As with the plant-derived cardenolides, the heterogeneous family of naturally occurring mammalian cardenolides (17)(18) have common molecular features essential for their biological activity. The two main structural features essential for preferential interaction with the sodium pump are a steroid four-ring structure with a cis-trans-cis configuration and a lactone ring attached at position C-17. In addition, depending on the specific type of cardenolide involved, linearly linked sugars extend from the aglycone C-3 position. Although considerable evidence now exists that mammals produce a family of digitalis-like factors and ouabain-like factors (OLFs), including several deglycosylated and apparent chemically reduced species (2)(19)(20)(21), their biosynthetic pathways have not been defined.

Structure–function studies (22)(23)(24)(25) and studies on the biosynthetic or metabolic pathways of the mammalian cardenolides have been limited primarily by the small amounts of material available for study (26) and the inability to track these compounds. We recently reported a chromatographic method for separating oxidized forms (DLIF and OLF) from their apparent chemically reduced components (Dh-DLIF and Dh-OLF) (2)(21). In the present study, we set out to address whether the DLIFs are synthesized de novo by adrenal cells, to determine whether their production is mediated by cAMP-dependent signaling as are the adrenal steroids, and to determine whether the production of these two DLIFs is dependent on the synthesis of cholesterol. Our working hypothesis was that if DLIFs are in fact synthesized in the adrenal cells and if they use a cholesterol-dependent pathway for their production, we would be able to incorporate radioactive carbon into the structure of DLIF starting with radiolabeled acetate or cholesterol as precursors. In addition, we also hypothesized that if the biosynthetic pathway is in fact dependent on formation of cholesterol as an intermediary precursor, then the production of DLIF (and Dh-DLIF) would be sensitive to intracellular inhibition of cholesterol synthesis. We found that both [1,2-14C]acetic acid and [4-14C]cholesterol were effective precursors for radiolabeling of the DLIFs. The synthesis of these DLIF molecules appears to be independent of cAMP signaling. Mevastatin, an inhibitor of cholesterol biosynthesis, suppressed secretion of DLIF and Dh-DLIF differently.

Materials and Methods

radioactive precursors

We used [1,2-14C]acetic acid (58 mCi/mmol; Amersham Life Sciences) and [4-14C]cholesterol (56 mCi/mmol; DuPont NEN, Life Science Products) as radioactive precursors. Before use these were checked for purity according to the manufacturers’ instructions.

equipment and materials

Solid-phase C18 cartridges (Sep-Pak) were obtained from Waters Associates as were the C18 reversed-phase μBondapak columns [300 × 3.9 mm (i.d.); 10-μm particle size], which were connected to a Waters 600E system controller and a Waters 996 photodiode array detector. Fractions collected with a Waters Fraction Collector (Millipore Corp.) were evaporated with a Jouan RC 10.22 Centrifugal Vacuum Concentrator connected to a Jouan Refrigerated Trap RCT 60. We used a 1900CA Liquid Scintillation Analyzer to measure the radioactive chromatographic fractions.

mouse y-1 cell line

Y-1 murine adrenocortical tumor cells were a generous gift from Dr. Bernard Schimmer (University of Toronto, Toronto, Ontario, Canada). The Y-1 cells were grown in DMEM supplemented with 100 mL/L fetal bovine serum (heat inactivated) plus antibiotics in 75-mm tissue culture flasks at 37 °C in 5% CO2 under a humid atmosphere. The cells were grown to 50–70% confluency and then washed twice with phosphate-buffered saline. Fresh serum-free DMEM medium was added to the cells, and incubation was continued for 48 h. For some experiments, either 1 mmol/L 8-bromoadenosine 3′:5′-cAMP (Sigma) or Mevastatin (Sigma) at 5 μmol/L was added to the medium, which was then added to the cells, and incubation was continued for 48 h. For radiolabeling experiments, 250 μCi of [1,2-14C]acetate or [4-14C]cholesterol was added to the serum-free medium, and the cells were incubated for 48 h. This provided 4.5 μmoles of cholesterol or 4.3 μmoles of acetate added to 1 mL of cell culture medium. After 48 h, the medium was collected, and the cells were harvested by scraping with a cell scraper and counted. Both cells and medium were stored at −80 °C until analysis. In experiments involving treatment of cultures with Mevastatin, 125 μCi of acetate was added to the serum-free medium in the presence and absence of Mevastatin (which provides 2.15 μmoles of acetate). The production of progesterone was measured directly in the cell medium by RIA as described previously (27). The production of the DLIFs was normalized to 1 × 106 Y-1 cells.

data analysis

The statistical significance of differences observed in the data shown in Tables 1⇓ and 2⇓ was analyzed by the Student unpaired t-test, with the null hypothesis rejected at the 5% confidence level. Data are expressed as the mean (SD).

View this table:
Table 1.

Production of DLIF and Dh-DLIF by mouse Y-1 adrenocortical cells.1

View this table:
Table 2.

Specific activity of radiolabeled DLIF and Dh-DLIF from Y-1 mouse adrenocortical cells.

separation and purification of compounds

Purification of DLIF components and steroids was accomplished by a multistep process involving first a solid-phase extraction followed by three sequential chromatographic separations.

Solid-phase extraction (C18 cartridge column).

The cell culture medium was extracted as reported previously by use of primed C18 reversed-phase solid-phase Sep-Pak extraction cartridges (actual volume, 10 mL) (20) but without the use of 5-sulfosalicylic acid (SSA). The cartridges were primed with 1 volume of CH3CN followed by rinsing with 2 volumes of deionized H2O. The sample was passed through the cartridge twice at a rate of 1 mL/min. The cartridge was then washed twice with 2–4 volumes of deionized H2O (typically 20–30 mL) before the compounds of interest were eluted with 20 mL of 600 mL/L CH3CN. To remove the CH3CN, the eluates were evaporated to dryness in a vacuum desiccator, reconstituted in deionized H2O, and filtered through a 0.22 μm filter (Whatman) for removal of particulates. All measured DLIFs and steroids were normalized to 1.0 × 106 Y-1 adrenal cells in culture.

HPLC reversed-phase chromatography steps.

We used three HPLC steps to purify DLIF and Dh-DLIF. In the first step (Fig. 1A⇓ ), the reconstituted eluates (600 mL/L CH3CN) from the solid-phase extractions indicated above were fractionated by a linear gradient of 200–800 mL/L CH3CN in H2O, and 1-mL fractions (1 mL/min) were collected over 40 min, evaporated, and reconstituted. In the second chromatographic step (Fig. 1B⇓ ), fractions of interest eluting at 25 min (from the linear gradient step 1 above) were further chromatographed in an isocratic mobile phase of 250 mL/L CH3CN in H2O for 40 min. In step 3 (Fig. 1C⇓ ), fraction 37 (containing DLIFs) from step 2 was rechromatographed with a mobile phase of 390 mL/L CH3CN in H2O for 20 min to further separate DLIF from Dh-DLIF and from any potentially remaining steroids not separated in the previous step. After this last HPLC separation, the purified DLIF and Dh-DLIF fractions (eluting at 6 and 4 min, respectively) were analyzed and measured by enzyme immunoassays for digoxin or dihydrodigoxin immunoreactivity and radioactivity depending on the experiment.

Figure 1.
Figure 1.

HPLC separation of 14C-labeled DLIF and Dh-DLIF from mouse adrenocortical Y-1 cell culture medium after incubation with [1,2-14C]acetate.

Separation of DLIFs requires three steps. (A), reversed-phase HPLC of Y-1 cell medium with a linear gradient of CH3CN (dashed line). Fraction 25 (min) contains the DLIFs (DLIF and Dh-DLIF) and steroids (20). (B), fraction 25 from A is further chromatographed to separate steroids and other compounds from DLIF (DLIF and Dh-DLIF coelute in fraction 37). (C), DLIFs (fraction 37) are rechromatographed with 390 mL/L CH3CN in H2O, and Dh-DLIF (4 min) and DLIF (6 min) separate. Percent DPM represents the total counts loaded on each column from the respective fraction in step above. These data represent the mean of four independent experiments listed in Table 2⇑ .

immunoassays for dlif, Dh-DLIF, and progesterone

We measured DLIF and Dh-DLIF immunoreactivity by digoxin and dihydrodigoxin immunoassays (2). In this study we used an enzyme immunoassay format with digoxin (or dihydrodigoxin) covalently bound to the microtiter plate to compete with DLIF or Dh-DLIF in samples (or for digoxin or dihydrodigoxin in calibrators) binding to a constant amount of anti-digoxin and anti-dihydrodigoxin polyclonal antisera in each plate well. The bound digoxin- and dihydrodigoxin-antibody complexes were detected by a secondary antibody-enzyme conjugate (goat anti-rabbit horseradish peroxidase conjugate). The breakdown of 3,3′,5,5′-tetramethylbenzidine substrate by the conjugated enzyme produces a color change with an intensity inversely proportional to the amount of analyte present in the well. Color was allowed to develop for 30 min, after which the reaction was stopped with a 3,3′,5,5′-tetramethylbenzidine buffer and the plate was then read at 450 nm. Our immunoassays have lower limits of quantification of 75 ng of digoxin-equivalent/L for DLIF (dig-equivalent) and 125 ng of dihydrodigoxin/L for Dh-DLIF (dihydrodigoxin-equivalent). The molar concentrations of DLIFs (DLIF and Dh-DLIF) were estimated by use of the previously reported similar molar absorptivities for digoxin and DLIF and dihydrodigoxin and Dh-DLIF (2)(19). We measured progesterone by a RIA with specific antibodies (Sigma) (27).

estimation of specific activity of radiolabeled DLIFs

We estimated the specific activities (mCi/mmol) of the radiolabeled mammalian cardenolides by dividing the radioactivity measured based on the molar concentration of the mammalian cardenolide (DLIF or Dh-DLIF). For example, when we used [4-14C]cholesterol as precursor, 4094 pg dig-equivalent of DLIF [equivalent to 4094 ng of DLIF; see Ref. (19)] measured 66 500 dpm. Use of an estimate of 780 g/mol digoxin gave a specific activity of 7.48 mCi/mmol (16.6 × 1012 dpm/mol divided by 2.22 × 109 dpm/mCi).

deglycosylation of 14c-dlif

We have previously demonstrated that DLIF isolated from mammalian tissue deglycosylated after incubation with acid into chromatographically well-defined genin, mono, and bis species while retaining the basic steroid and lactone ring structure in a manner identical to that of digoxin (19). To confirm that the radioactivity was linked to the DLIF molecule, we subjected purified 14C-DLIF (DLIF fraction 6 from Fig. 1C⇑ from the Y-1 cells (incubated with [4-14C]cholesterol) to acid treatment (10 g/L SSA) for 15 s and then immediately extracted the radiolabeled DLIF by use of reversed-phase C18 Sep-Pak cartridges as described above. The filtered sample was then chromatographically separated by reversed-phase HPLC with a linear gradient of 200–800 mL/L CH3CN over 40 min as in Fig. 1A⇑ . In all cases the background dpm were subtracted from the data presented.

Results

stimulation of y-1 mouse adrenocortical cells with cAMP

Y-1 cells in culture after incubation for 48 h produced and secreted DLIFs (DLIF and Dh-DLIF) into the medium. Table 1⇑ shows the results of several experiments in which purified DLIF and Dh-DLIF (pmoles) were assessed with and without exposure of the cell culture to cAMP. We observed no significant increase in the production of either DLIF or Dh-DLIF with cAMP stimulation. On the other hand, progesterone production was increased sevenfold. In both cases the DLIFs measured in the cell pellets were <20% of the concentrations in the medium. As a control, neither the fresh medium (without incubation with Y-1 cells) nor the medium or cell pellets taken after 10 min of incubation with the 70% confluent Y-1 cells showed detectable immunoreactivity for DLIFs (data not shown). These results indicate that DLIF is produced by this adrenal cell culture but that the production is not increased when the cells are exposed to cAMP.

production and chromatographic separation of radiolabeled DLIFs

Our working hypothesis was that if DLIF and Dh-DLIF are synthesized de novo as suggested in the above experiments, then incubating the cultured cells with a rich source of a radioactive carbon pool, such as acetate, would produce 14C-labeled DLIFs as products. We successfully obtained radiolabeled DLIFs (both DLIF and Dh-DLIF) after incubation of Y-1 cells with either [1,2-14C]acetate or [4-14C]cholesterol as the radioactive precursor. Fig. 1⇑ shows the radioactive HPLC fractions containing DLIF and Dh-DLIF through their successive purification processes. We used the three-step chromatographic sequence shown in Fig. 1⇑ , in which DLIF and Dh-DLIF initially coelute in the 25-min fraction of the linear gradient of 200–800 mL/L CH3CN in H2O (Fig. 1A⇑ ). The fraction containing the DLIFs was further purified by two subsequent isocratic chromatography steps as indicated (Fig. 1⇑ , B and C). Fraction 25 (Fig. 1A⇑ ), containing DLIF, Dh-DLIF, and other undefined radioactive products, retained 42.9% (1.8 × 106 dpm) of the total 14C-labeled compounds (4.2 × 106 dpm) obtained from the cell culture medium. However, after further purification, 15.8% of the radioactivity from fraction 25 subsequently eluted as DLIFs (DLIF plus Dh-DLIF) in fraction 37, and the remaining 14C-labeled compounds eluted earlier (step 2). The DLIFs (fraction 37 from step 2) when chromatographed in step 3 (Fig. 1C⇑ ) separated as two components; the first was Dh-DLIF, eluting at fraction 4 (containing 55% of the counts loaded on this column), and the second was DLIF, eluting at fraction 6 (25% of the counts loaded).

As shown in Table 2⇑ , both [1,2-14C]acetic acid and [4-14C]cholesterol served as effective precursors for incorporation of 14C into the structure of the DLIFs. The specific activities of 14C-DLIF and 14C-Dh-DLIF were 7.5 and 5.5 mCi/mmol when [14C]cholesterol was used as a precursor and 72.2 and 51.5 mCi/mmol when [14C]acetate was used as precursor. Stoichiometrically, one would expect a maximum of 18 carbons to be radiolabeled when acetate is used as a precursor compared with 1 carbon being labeled when cholesterol is used as a precursor. To compare the relative amounts of DLIFs being labeled, we found that 10% of the total DLIFs were labeled when [4-14C]cholesterol was used as precursor and 0.08% when [1,2-14C]acetic acid was used (Table 3⇓ ). When we started with [4-14C]cholesterol (250 μCi, or 4.5 μmoles of [14C]cholesterol), we observed that ∼0.01% of the initial amount of radioactivity added to the cell culture medium eluted in fraction 6 (Fig. 1C⇑ ). Considerably less (0.00005%) of the radioactivity eluted in that same chromatographic fraction when we used [1,2-14C]acetic acid (250 μCi, or 4.3 μmoles of [14C]acetate) in the original incubation (Table 3⇓ ).

View this table:
Table 3.

Relative amounts of cholesterol and acetate as precursors used in biosynthesis of DLIF.1

To confirm that the radioactive carbons were in fact incorporated in the structure of DLIF, we subjected the radiolabeled DLIF obtained from fraction 6 (Fig. 1C⇑ ) to acid hydrolysis as described previously (19). This process has been shown to remove the sugars without disrupting the genin portion of the molecules (19). In this experiment, we used DLIF material obtained from the incubations with [4-14C]cholesterol with the expectation that if cholesterol is a precursor, then only the genin portion of DLIF would be radiolabeled and thus the radioactivity would be conserved in each chromatographic fraction obtained after deglycosylation. As shown in Fig. 2⇓ , 14C-DLIF when exposed to acid hydrolysis fractionated into components that eluted at 23, 19, and 15 min (no other radioactive peaks were observed in this chromatographic step). These fractions corresponded exactly to those of DLIF-bis (two sugars), DLIF-mono (one sugar), and DLIF-genin (no sugars), respectively (19). We also observed conservation in the total counts of 14C-DLIF before (12 200 dpm) and after (11 476 dpm) treatment with SSA (DLIF-genin, 3388 dpm; DLIF-mono, 1142 dpm; DLIF-bis, 1971 dpm; DLIF, 4975 dpm) and in the total amount of immunoreactivity (total before, 182 pg of dig-equivalents and total after 177 pg of dig-equivalents). We also found that the ratios of dpm to the amount of DLIF before (67 dpm/dig-equivalents of DLIF) and after (64.8 dpm/dig-equivalents of DLIF-genin) treatment with SSA were similar.

Figure 2.
Figure 2.

Deglycosylated components (DLIF-bis, -mono, and -genin) produced after acid treatment of 14C-DLIF.

Purified 14C-DLIF from HPLC (Fig. 1C⇑ ) treated with 10 g/L SSA for 15 s and chromatographed with a linear gradient of 200–800 mL/L CH3CN in H2O for 30 min. □, untreated 14C-DLIF; ▪, respective components obtained after treatment with acid (see Materials and Methods). The background dpm were subtracted from the data presented. Total DLIFs measured was 182 pg of dig-equivalents before acid treatment and 177 pg of dig-equivalents after acid treatment.

inhibition of cholesterol synthesis and the secretion of dlif and Dh-DLIF

If cholesterol is an intermediate in the pathway between acetate and the DLIFs, inhibiting the synthesis of cholesterol should reduce the production of DLIFs. We tested the effect of Mevastatin (5 μmol/L), an inhibitor of the enzyme hydroxymethylglutaryl-CoA (HMG-CoA) reductase and thus of cholesterol biosynthesis, on [1,2-14C]acetate incorporation into DLIF and Dh-DLIF. As shown in Fig. 3⇓ , Mevastatin reduced the production of both DLIF and Dh-DLIF. However, Mevastatin appeared to suppress production of the DLIF species by 85.3%, whereas the Dh-DLIF was suppressed by only 15%. As a control in the these same experiments and as expected, Mevastatin (5 μmol/L) inhibited the production of progesterone by 57% as measured by a RIA specific for progesterone (data not shown).

Figure 3.
Figure 3.

Effect of Mevastatin (5 μmol/L) on production of 14C-labeled DLIF and Dh-DLIF in mouse Y-1 cells with [1,2-14C]acetate as precursor.

Note the apparent selective inhibition of DLIF vs Dh-DLIF after separation by the methods illustrated in Fig. 1C⇑ . Progesterone was decreased by 57% (from 4 ng/106 cells to 1.8 ng/106 cells). These data represent the mean of two independent experiments.

Discussion

The focus of this study was to demonstrate de novo biosynthesis of DLIFs (DLIF and Dh-DLIF) by mouse Y-1 adrenocortical cells in culture by use of radioactive acetate and cholesterol as precursors. This approach would also have the advantage of producing radiolabeled DLIFs, which could then be tracked chromatographically in addition to monitoring their other known properties. With this model, we explored the dependence of DLIF synthesis on cAMP signaling and on the regulation of cholesterol synthesis. Our initial working hypothesis was that if DLIFs are synthesized in the adrenal cells, we would be able to incorporate 14C into the structure of DLIF starting with either radiolabeled acetate or cholesterol as carbon source. In addition, we also speculated that if cholesterol is an intermediary in the synthesis of DLIFs, their production could be regulated by intracellular inhibition of cholesterol synthesis. The results of our experiments indicate that both DLIF and Dh-DLIF are produced de novo by these cells in culture, that production of DLIF is not stimulated by cAMP, and that inhibition of cholesterol synthesis seems to significantly reduce the amount of DLIF but only modestly reduce the amount of Dh-DLIF produced by the adrenal cell.

The premise of “endogenous” digitalis-like or ouabain-like compounds having a mammalian tissue source has been a subject of considerable controversy since the discovery of these compounds in mammals (17)(18). Although there is evidence to support this idea, most of it has been clouded by suspicion of extra vivo sources (e.g., diet or contamination) (26)(28) or by certain structural features of the plant-derived cardenolides not consistent with what is known about mammalian steroid biosynthesis (the details are discussed below). Although the fine structural features of DLIFs as a family of compounds have not been definitively determined, there is very persuasive evidence that their structures are remarkably similar to those of the plant-derived digitalis compounds (2)(19)(29). The structure of digoxin, depicted in Fig. 4⇓ , is used as a proposed “scaffold working structure” for DLIF (30). The putative structure includes features considered essential for the biological activity of the cardenolides: a characteristic five-ring structure to which is attached linearly linked sugars extending from the aglycone C-3 position, and the aglycone portion consisting of the four-ring steroid skeleton (cis-trans-cis conformation) with an unsaturated lactone ring attached at position C-17. We have recently shown that, as with digoxin, several forms of mammalian DLIFs exist that include a series of deglycosylated species: DLIF-genin, -bis, and -mono components (19) as well as a dihydrodigoxin-like form, Dh-DLIF (2). Evidence of a metabolic conversion in vitro of dihydrodigoxin (a dihydro species) to a molecule with DLIF-like properties by a cytochrome P450-dependent reaction has also been reported (2).

Figure 4.
Figure 4.

Working hypothetical model for production of DLIF in adrenal cells.

Cholesterol is viewed as a proximal precursor, with a series of subsequent enzymatic reactions (see Fig. 5⇓ ) leading to the production of DLIF and Dh-DLIF as two distinct species. The possible roles of progesterone and pregnenolone as proximal precursors with addition of a lactone ring cannot be ruled out at this time.

Do adrenal cells have the machinery to synthesize digitalis-like compounds? Our present knowledge of steroid biosynthesis may argue against this. The steroid rings comprising cholesterol and known mammalian steroids have a trans-trans-trans configuration on the A-B-C-D rings (the 5,6 double bond ensures that ring A remains in a plane similar to that of ring B) (31). None of the common steroids are cis-trans-cis as are the plant-derived compounds (32), nor do they possess closed lactone rings attached to carbon 17. Additionally, digitoxose sugars are not known to exist in mammals. However, recent evidence may suggest otherwise (30). The production of compounds such as DLIF would then need to entail as yet undefined cellular processes if indeed they have structural properties similar to those of plant-derived digitalis. For example, transformation of the cholesterol backbone into a cis-trans-cis conformation and closing of the side chain of cholesterol into a lactone ring may be key elements in the formation of DLIF compounds. We found no increase in the final yield of DLIF after stimulation with 8-bromoadenosine 3′:5′-cAMP. Y-1 cells increase the synthesis of progesterone and its derivatives in response to adrenocorticotropin hormone (ACTH) (33). ACTH binds a specific G-protein-coupled 7-transmembrane receptor that activates adenylate cyclase and produces the increase in intracellular cAMP (34)(35). When we treated Y-1 cells with 8-bromoadenosine 3′:5′-cAMP, a cAMP analog, we observed a fivefold increase in steroids, as expected (36), but no increase in DLIF was detected (Table 1⇑ ). The distribution of DLIF and steroids between the incubation medium and cell pellets remained the same. No detectable DLIF was found in the cell medium before culture or in the cell pellets or medium after 10 min of incubation, at which time values were below the lower limit of detection in both the digoxin and dihydrodigoxin enzyme immunoassays. This provides evidence that the synthesis of the DLIFs is indeed de novo in this cell system but is not dependent on cAMP stimulation.

We also found that inhibition of cholesterol synthesis by Mevastatin, an inhibitor of the enzyme HMG-CoA reductase, reduced the amount of total DLIFs (DLIF plus Dh-DLIF) produced by 58%, as it did progesterone. Interestingly, the inhibition seemed to be selective in that DLIF was reduced by 85.3% whereas Dh-DLIF was reduced by only 15% (Fig. 3⇑ ). The apparent selectivity in suppression of these two species suggests the possibilities of either independent pathways for production of these two DLIF compounds or that the process of conversion of one to the other may be altered by the actions of Mevastatin; this is unknown at this time.

Do both acetic acid and cholesterol efficiently serve as carbon pools for the biosynthesis of DLIFs and is cholesterol itself involved in the biosynthesis of DLIFs? Results of our present study using the Y-1 adrenal cell culture system show that both radiolabeled acetate and cholesterol can serve as precursors to produce 14C-labeled DLIFs (see Table 2⇑ ). This is not surprising because acetic acid as a carbon pool would also generate cholesterol. However, the fact that when we started with radiolabeled cholesterol as a precursor, the labeling seemed limited to the aglycone portion of the DLIFs and not the sugars (Fig. 2⇑ ) would suggest that cholesterol is likely a necessary precursor. Additionally, if cholesterol is a necessary precursor, then when starting with a basic carbon pool such as acetate, we would expect that inhibition of the cholesterol synthesis pathway by Mevastatin would also suppress the synthesis of DLIFs, as we observed (Fig. 3⇑ ). If the major pathway from acetate to DLIF was independent of (i.e., before) the HMG-CoA reductase step, DLIF synthesis would be less likely to be suppressed by Mevastatin. Future experiments might include testing the nonspecific effects of Mevastatin (37). However, the observed decreases in progesterone with Mevastatin in our experiment indirectly support that Mevastatin did decrease cholesterol synthesis. Additionally, the need to then convert the cholesterol to form a closed lactone ring implies that side chain cleavage is not required for subsequent synthesis of the DLIF compounds. This would then suggest that production of DLIF would not be dependent on a cAMP-signaling mechanism, as our data indicate (Table 1⇑ ).

Comparing the relative amounts of 14C incorporated into DLIF when acetate and cholesterol are used as the precursors may provide clues to the efficiency of the labeling and the biosynthetic pathways. Given that acetyl-CoA is the two-carbon precursor for cholesterol biosynthesis, if we assume that the first steps in the synthesis of DLIF are conversion of the acetate carbon pool to cholesterol and then cholesterol to DLIF, one can predict that a maximum of 18 carbons are radiolabeled when acetate is used compared with 1 carbon being labeled when cholesterol is used as a precursor. With the consideration that 18 moles of acetate yields 1.0 mole of cholesterol and that 1.0 mole of cholesterol yields 1.0 mole of DLIF, our calculations indicate that when [1,2-14C]acetic acid is used as the precursor, only 0.08% of the DLIF is radiolabeled compared with 10% of the total DLIF synthesized being labeled when [4-14C]cholesterol is used as precursor (Table 3⇑ ). This suggests that for a first-order approximation, starting with cholesterol as a precursor is considerably more efficient for synthesis of DLIFs. Stoichiometrically, one would expect a maximum of 18 carbons to be labeled when acetate is used as precursor and only 1 carbon when cholesterol is used as precursor. As observed in Table 2⇑ , the ratio of dpm between the two conditions is 9.6 (72.2/7.5) for DLIF and 9.4 for Dh-DLIF. The mean number of carbons needed to build cholesterol or DLIF from the available pool is 10 carbons (range, 2–18 carbons), which is approximately what we observe.

Hamlyn et al. (38) have shown that production of mammalian OLF from adrenal glands was not significantly increased after treatment with a cAMP analog. However, other studies have reported increases in OLF. For example, increased OLF was reported in animals and/or culture models using ACTH (39)(40); angiotensin II (41); combinations of ACTH, angiotensin II, aldosterone, and cortisol (42)(43); and progesterone and pregnenolone (44).

Lastly, are radioactive carbons in the structure of DLIF? Although advanced biophysical methods would be required to definitely demonstrate that radioactive carbons are indeed part of the DLIF molecules, we believe that the evidence presented here is, although not definitive, sufficiently convincing. In all of our chromatographic separations, the radioactivity followed the immunoreactivity, but the most convincing evidence we present is the simultaneous and exclusive tracking of the radioactivity and immunoreactivity after exposure of 14C-DLIF to acid hydrolysis of the sugar components. On the basis of well-documented previous studies, the chromatographic profile we observed after acid-induced deglycosylation was characteristic of both digoxin and DLIF (19). Three aspects of our experiments support the conclusion that DLIF is radiolabeled: (a) the constancy of the ratio of radioactivity to immunoreactivity in all four fractions obtained after acid hydrolysis; (b) the lack of other radioactive or immunoreactive fractions detected in the entire chromatograph after hydrolysis of fraction 25 in Fig. 2⇑ ; and (c) within experimental error, the observed conservation of both total radioactivity and immunoreactivity before and after acid exposure and chromatographic separation. It would be extremely unusual that other non-DLIF but radiolabeled compounds produced by the cells would provide all of these properties. Thus, these data strongly support the premise that the DLIFs produced by the Y-1 cells are indeed radiolabeled with 14C. We have previously reported the production of DLIF deglycosylated species (mono-, bis-, and genin-DLIF) by Y-1 cells in culture (45). In our present experiments, however, the chromatographic separation we used for isolating DLIFs from other radiolabeled compounds produced by Y-1 cells was not optimized for separation of these deglycosylated species during the isolation process.

In summary, the release of radiolabeled DLIF from Y-1 cells into the culture medium confirms that the adrenal cell is a biosynthetic site for production of both DLIF and Dh-DLIF. These data strongly support the concept that cellular machinery exist in the adrenal cell to provide de novo synthesis of DLIFs starting from a simple carbon pool. Of interest is that cholesterol does not have a cis-trans-cis conformation, as suspected for DLIF, but served as an effective precursor for synthesis of DLIF. This suggests the strong likelihood that either DLIF may not necessarily be a cis-trans-cis molecule or that an isomerization reaction is required in the biosynthesis of the DLIF-compounds. In Figs. 4⇑ and 5⇓ , we propose a working model for the biosynthesis of DLIFs from several known molecular intermediates as possibilities for a cholesterol-dependent pathway for synthesis of two DLIF species. Our data for DLIF production based on a mouse adrenal cell culture model are consistent with those of Hamlyn et al. (38) for OLF, but do not at this time exclude pregnenolone as a precursor. Pregnenolone is less likely than cholesterol to be a precursor because of the need for a lactone ring, which may be derived from closing of the cholesterol side chain as opposed to subsequent attachment. However, the work of Lichtstein et al. (46) suggests that removal of the cholesterol side chain may be involved in the production of some digitalis-like compounds. The issue remains controversial. Our present work provides several avenues and tools for future studies to dissect the molecular mechanisms and identify subcellular fractions and enzymes involved in the biosynthesis of these important mammalian compounds. Such characterization may include confirmation using 13C substrates with mass spectrometric analysis to identify intermediate products as well as use of selected enzyme inhibitors to characterize the enzymes involved in the biosynthetic pathways. Additionally, the approach we use demonstrates that 14C-labeled DLIFs can be produced naturally in vitro, which now provides a source of radiolabeled DLIF compounds for development of immunoassays.

Figure 5.
Figure 5.

Proposed enzymatic conversions to produce DLIF compounds from cholesterol as precursor in adrenocortical cells.

A series of enzymatic reactions involves converting cholesterol (all-trans) into a cis-trans-cis configuration, closing the lactone ring, and adding sugars to produce either DLIF or Dh-DLIF as two distinct species. Enz, enzyme.

Acknowledgments

We gratefully acknowledge the skilled technical assistance of Rebecca Combs in maintaining and growing the Y-1 murine adrenocortical tumor cells. This work was supported by Public Health Service Grants HL R01-32617 (to R.V.), HL R01-59404 (to R.V.), NIH DK 51656 (to B.C.), and NSF EPSCoR OSR-9452895 (to R.V.).

Footnotes

  • ↵1 Current address: Laboratory of Molecular Diagnostics, Wadsworth Center for Laboratories and Research, New York State Department of Health, Albany, NY.

  • 1 All measurements are from the cell culture medium. DLIFs measured in the cell pellets were <20% of the values in medium: with cAMP, DLIF, 0.09 (0.02) nmol/106 cells; Dh-DLIF, 0.28 (0.08) nmol/106 cells; without cAMP, DLIF, 0.11 (0.02) nmol/106 cells; Dh-DLIF, 0.23 (0.06) nmol/106 cells.

  • 2 Number of independent experiments for each condition.

  • 3 NS, not significant.

  • 1 Number of independent experiments performed.

  • 1 In these radiolabeling experiments, a total of 250 μCi of [1,2-14C]acetate (4.3 μmoles) or [4-14C]cholesterol (4.5 μmoles) was added to the serum-free medium. Note that when [4-14C]cholesterol is used as precursor, only 10% of the total DLIF isolated is radiolabeled and when [1,2-14C]acetic acid is used, only 0.08% of the isolated DLIF is labeled. See Materials and Methods for additional details.

  • 2 Measured by radioactivity.

  • 3 Measured by immunoreactivity.

  • ↵2 Nonstandard abbreviations: DLIF, digoxin-like immunoreactive factor; Dh-DLIF, dihydrodigoxin-like immunoreactive factor; OLF, ouabain-like factor; SSA, 5-sulfosalicylic acid; dig-equivalent, digoxin-equivalent; HMG-CoA, hydroxymethylglutaryl-CoA; and ACTH, adrenocorticotropin.

References

  1. ↵
    Shaikh IM, Lau BWC, Siegfried BA, Valdes R, Jr. Isolation of digoxin-like immunoreactive factors from mammalian adrenal cortex. J Biol Chem 1991;266:13672-13678.
    OpenUrlAbstract/FREE Full Text
  2. ↵
    Qazzaz HMA, Jortani SA, Poole JM, Valdes R, Jr. Evidence for the presence of a reduced form of digoxin-like immunoreactive factor (dihydro-DLIF) in mammalian tissue. Clin Chem 1996;42:270-278.
    OpenUrlAbstract/FREE Full Text
  3. ↵
    Arystarkhova E, Gasparian M, Modyanov NN, Sweadner KJ. Na+, K+-ATPase extracellular surface probed with a monoclonal antibody that enhances ouabain binding. J Biol Chem 1992;267:13694-13701.
    OpenUrlAbstract/FREE Full Text
  4. ↵
    Schultheis PJ, Wallick ET, Lingrel JB. Kinetic analysis of ouabain binding to native and mutated forms of Na+, K+-ATPase and identification of a new region involved in cardiac glycoside interactions. J Biol Chem 1993;268:22686-22694.
    OpenUrlAbstract/FREE Full Text
  5. ↵
    Palasis M, Kuntzweiler TA, Arguello JM, Lingrel JB. Ouabain interactions with the H5–H6 hairpin of the Na+, K+-ATPase reveal a possible inhibition mechanism via the cation binding domain. J Biol Chem 1996;271:14176-14182.
    OpenUrlAbstract/FREE Full Text
  6. ↵
    Price EM, Lingrel JB. Structure-function relationships in the Na+, K+-ATPase α subunit: site-directed mutagenesis of glutamine-111 to arginine and asparagine-122 to aspartic acid generates a ouabain-resistant enzyme. Biochemistry 1988;27:8400-8408.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  7. ↵
    Kaplan JH. Biochemistry of Na,K-ATPase. Annu Rev Biochem 2002;71:511-535.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  8. ↵
    Doris PA, Bagrov AY. Endogenous sodium pump inhibits and blood pressure regulation: an uptake on recent progress. Proc Soc Exp Biol Med 1998;218:156-157.
    OpenUrlAbstract/FREE Full Text
  9. ↵
    Hollenberg NK, Graves SW. Endogenous sodium pump inhibition: current status and therapeutic opportunities. Prog Drug Res 1996;46:9-42.
    OpenUrlPubMed Order article via Infotrieve
  10. ↵
    Valdes R, Jr, Graves SW. Protein binding of endogenous digoxin-immunoreactive factors in human serum and its variation with clinical condition. J Clin Endocrinol Metab 1985;60:1135-1143.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  11. ↵
    Antolovic R, Kost H, Mohadjerani M, Linder D, Linder M, Schoner WA. Specific binding protein for cardiac glycosides exists in bovine serum. J Biol Chem 1998;273:16259-16264.
    OpenUrlAbstract/FREE Full Text
  12. ↵
    Antolovic R, Bauer N, Mohadjerani M, Kost H, Neu H, Kirch U, et al. Endogenous ouabain and its binding protein: effects of physical exercise and study on the globulin’s tissue distribution. Hypertens Res 2000;23:S93-S98.
  13. ↵
    Parhami-Seren B, Haberly R, Margolies MN, Haupert GT, Jr. Ouabain-binding protein(s) from human plasma. Hypertension 2002;40:220-228.
    OpenUrlAbstract/FREE Full Text
  14. ↵
    Rose AM, Valdes R, Jr. Understanding the sodium pump and its relevance to disease. Clin Chem 1994;40:1674-1685.
    OpenUrlAbstract/FREE Full Text
  15. ↵
    Blaustein MP. Physiological effects of endogenous ouabain: control of intracellular Ca stores and cell responsiveness. Am J Physiol 1993;264:1367-1387.
    OpenUrl
  16. ↵
    Blaustein MP. Endogenous ouabain: role in the pathogenesis of hypertension. Kidney Int 1996;49:1748-1753.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  17. ↵
    Dmitrieva RI, Doris PA. Cardiotonic steroids: potential endogenous sodium pump ligands with diverse function. Exp Biol Med 2002;227:561-569.
    OpenUrlAbstract/FREE Full Text
  18. ↵
    Schoner W. Endogenous cardiac glycosides, a new class of steroid hormones. Eur J Biochem 2002;269:2440-2448.
    OpenUrlPubMed Order article via Infotrieve
  19. ↵
    Qazzaz HMA, Goudy SL, Valdes R, Jr. Deglycosylated products of endogenous digoxin-like immunoreactive factor (DLIF) in mammalian tissue. J Biol Chem 1996;271:8731-8737.
    OpenUrlAbstract/FREE Full Text
  20. ↵
    Qazzaz HMA, Valdes R, Jr. Simultaneous isolation of endogenous digoxin-like immunoreactive factor (DLIF), ouabain-like factor (OLF), and deglycosylated congeners from mammalian tissues. Arch Biochem Biophys 1996;328:193-200.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  21. ↵
    Qazzaz HMA, El-Masri MA, Valdes R, Jr. Secretion of a lactone-hydrogenated ouabain-like effector of sodium, potassium-adenosine triphosphate activity by adrenal cells. Endocrinology 2000;141:3200-3209.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  22. ↵
    Tao Q-F, Hollenberg NK, Price DA, Graves SW. Sodium pump isoform specificity for the digitalis-like factor isolated from human peritoneal dialysis. Hypertension 1997;29:815-821.
    OpenUrlAbstract/FREE Full Text
  23. ↵
    Butt AN, Tennant BP, Gillingwater SD, Shepherd PS, Swaminathan R. Binding of ouabain and human ouabain-like substance to different Na+,K+-ATPase isoforms. Hypertens Res 2000;23(Suppl):S45-S51.
  24. ↵
    Fedorova OV, Kolodkin NI, Agalakova NI, Lakatta EG, Bagrov AY. Marinobufagenin, an endogenous α-1 sodium pump ligand, in hypertensive Dahl salt-sensitive rats. Hypertension 2001;37:462-466.
    OpenUrlAbstract/FREE Full Text
  25. ↵
    Manunta P, Hamilton BP, Hamlyn JM. Structure-activity relationships for the hypertensinogenic activity of ouabain: role of the sugar and lactone ring. Hypertension 2001;37:472-477.
    OpenUrlAbstract/FREE Full Text
  26. ↵
    Doris PA. Regulation of Na+, K+-ATPase by endogenous ouabain-like materials. Proc Soc Exp Biol Med 1994;205:202-212.
    OpenUrlAbstract/FREE Full Text
  27. ↵
    Resko JA, Norman GD, Niswender GD, Spies HG. The relationship between progestins and gonadotropins during the late luteal phase of the menstrual cycle in rhesus monkeys. Endocrinology 1974;94:128-135.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  28. ↵
    Tamura M, Harris TM, Phillips D, Blair IA, Wang YF, Hellerqvist CG, et al. Identification of two cardiac glycosides as Na+-pump inhibitors in rat urine and diet. J Biol Chem 1994;269:11972-11979.
    OpenUrlAbstract/FREE Full Text
  29. ↵
    Komiyama Y, Nishimura N, Dong XH, Hirose S, Kosaka C, Masaki H, et al. Liquid chromatography mass spectrometric analysis of ouabain-like factor in biological fluid. Hypertens Res 2000;23:S21-S27.
  30. ↵
    Qazzaz HMA, Lane AN, Valdes R, Jr. Structural identification of digoxin-like immunoreactive factors (DLIFs) using NMR spectroscopy. Clin Chem 2003;49(Suppl 6):A130.
    OpenUrl
  31. ↵
    Heftmann E. Steroid biochemistry 1970:85-96 Academic Press New York. .
  32. ↵
    Greeff K. Handbook of experimental pharmacology, Vol. 56. Cardiac glycosides, part 1 1981:13-24 Springer-Verlag Berlin. .
  33. ↵
    Kowal J, Fiedler R. Adrenal cells in tissue culture. I. Assay of steroid products; steroidogenic resposes to peptide hormones. Arch Biochem Biophys 1968;128:406-421.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  34. ↵
    Schimmer BP, Schulz P. The role of cAMP and cAMP-dependent protein kinase in forskolin’s actions on Y1 adrenocortical tumor cells. Endocr Res 1985;11:199-209.
    OpenUrlPubMed Order article via Infotrieve
  35. ↵
    Wong M, Krolczyk AJ, Schimmer BP. The causal relationship between mutations in cAMP-dependent protein kinase and the loss of adrenocorticotropin-regulated adrenocortical functions. Mol Endocrinol 1992;6:1614-1624.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  36. ↵
    Waterman MR, Bischof LJ. Mechanisms of ACTH(cAMP)-dependent transcription of adrenal steroid hydroxylases. Endocr Res 1996;22:615-620.
    OpenUrlPubMed Order article via Infotrieve
  37. ↵
    Brown MS, Faust JR, Goldstein JL. Induction of 3-hydroxy-3-methylglutaryl coenzyme a reductase activity in human fibroblasts incubated with compactin (ML-236B), a competitive inhibitor of the reductase. J Biol Chem 1978;253:1121-1128.
    OpenUrlFREE Full Text
  38. ↵
    Hamlyn JM, Lu Z, Manunta P, Ludens JH, Kimura K, Shah JR, et al. Observation on the nature, biosynthesis, secretion, and significance of endogenous ouabain. Clin Exp Hypertens 1998;20:523-533.
  39. ↵
    Hinson JP, Harwood S, Dawnay AB. Release of ouabain-like compound (OLC) from the intact perfused rat adrenal gland. Endocr Res 1998;24:721-724.
    OpenUrlPubMed Order article via Infotrieve
  40. ↵
    Beck M, Szalay KS, Nagy GM, Toth M, de Chante R. Production of ouabain by rat adrenocortical cells. Endocr Res 1996;22:845-849.
    OpenUrlPubMed Order article via Infotrieve
  41. ↵
    Shah JR, Laredo J, Hamilton BP, Hamlyn JM. Effects of angiotensin II on sodium potassium pumps, endogenous ouabain and aldosterone in bovine zona glomerulosa cells. Hypertension 1999;33:373-377.
    OpenUrlAbstract/FREE Full Text
  42. ↵
    Pena C, Rodriguez de Lores Arnaiz G. Differential properties between an endogenous brain Na+,K+-ATPase inhibitor and ouabain. Neurochem Res 1997;22:379-383.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  43. ↵
    Li M, Wong KS, Martin A, Withworth A. Adrenocorticotrophin-induced hypertension in rats. Am J Hypertens 1994;7:59-68.
    OpenUrlAbstract/FREE Full Text
  44. ↵
    Perrin A, Brasmes B, Chambaz EM, Defaye G. Bovine adrenocortical cells in culture synthesize an ouabain-like compound. Mol Cell Endocrinol 1997;126:7-15.
    OpenUrlCrossRefPubMed Order article via Infotrieve
  45. ↵
    Cao TZ, Qazzaz HMA, Bolanowski DD, El-Masri MA, Clark BJ, Valdes R, Jr. Ex vivo production of digoxin-like immunoreactive factors (DLIF) by mouse adrenal cortex tumor (Y-1) cells [Abstract]. Clin Chem 1999;45(Suppl 6):A83.
    OpenUrl
  46. ↵
    Lichestein D, Steinitz M, Gati I, Samuelov S, Deutsch J, Orly J. Biosynthesis of digitalis-like compounds in rat adrenal cells: Hydroxycholesterol as possible precursor. Life Sci 1998;62:2109-2126.
    OpenUrlCrossRefPubMed Order article via Infotrieve
View Abstract
Back to top

In this issue

Clinical Chemistry: 50 (3)
Vol. 50, Issue 3
March 2004
Print
Article Alerts
Citation Tools

Jump to section

Subjects