Phosphocholine Steroid Conjugates: Are These Compounds the Mammalian Cardiotonic Steroids?

Cardiotonic glycoside toxins, originally isolated from plants or amphibians, have been proposed as mammalian cardiotonic hormones. This paper is a review and update of the discovery of [i] a new class of steroid hormones, [ii] the path for their biosynthesis and [iii] some preliminary data on their function. The compounds are phosphoester conjugates and share a characteristic structural feature, a lactone ring, with [a] one class of synthetic potassium sparing diuretics and with [b] plant and amphibian cardiotonic steroids. Purification was initially monitored by cross reaction with steroid sulfate assays and later with digoxin-specific assays. Six compounds were purified by HPLC to near homogeneity and characterized by Tandem mass spectroscopy (MS-MS) and 31P-NMR. Three were digoxin-like materials (DLM) with 23 carbon atoms. The two extra carbon atoms form a spiral lactone E-ring. Several additional spiral lactones have been identified by MS-MS. In a pilot study, based on MS analysis, we evaluated phosphocholine steroid levels in individual serum samples in patients with pre-eclampsia (n=20). The control group of normotensive pregnant women (n=20) was used to estimate the mean and standard deviation. Twelve of the women with pre-eclampsia had a z-score over 2 for at least one of the four phosphocholine steroids. In contrast, only 1 sample from the normotensive women had a z-score over 2. The observation that there are two patterns, one with elevated phosphocholine steroid levels and one without the elevation, suggests that there may be two different underlying causes of pre-eclampsia. We now need to extend the study to determine which pattern predicts progression to life-threatening hypertension in pregnant women. This pilot study illustrates that it is possible to evaluate individual endogenous cardiotonic hormones without relying on antibodies developed to plant or amphibian toxins.


Introduction
In the 18 th century, Withering recognized that an extract of foxglove, digoxin, was useful therapy for dropsy, now described as congestive heart failure [1]. In 1955, Szent-Gyorgyi proposed that digoxin wasn't really a drug but was a substitute for an unknown hormone [2]. This recognition led to a 35-year search for the endogenous hormone based on its detection by cross reaction with digoxin-specific antibodies [3]. In 1991, Hamlyn, starting with 80 liters of plasma, led a team that isolated 13 µ g of a material that seemed to be ouabain [4]. This was the initial observation that suggested that the cardiotonic steroids were not toxins but actually were endogenous hormones. More than 200 papers have now been published reporting endogenous ouabain levels in human disease based on cross reaction with an antibody developed to a cardiotonic steroid, typically ouabain or digoxin [5]. Although ouabain was characterized as an endogenous hormone, neither precursors nor metabolites of the steroid component have ever been detected by any direct analytical method [6].

Nomenclature
In total, this paper describes more than 20 novel compounds. For all but one, there is no trivial, recognized name and no familiar structure for any of them. For ease, phosphocholine esters will be designated as Cxxx; phosphoethanolamine esters will be designated as Exxx; Pxxx will be used to designate compounds for which the steroid fragment was observed but the conjugate was not confirmed by mass spectroscopy. For each symbol, the xxx will be the m/z from the steroid fragment as observed on mass spectroscopy in the cation mode. In most cases, the mass for the chemical formula of the steroid will be m/z= xxx +17 Da because the phosphoester fragment (m/z=184 Da) includes all four of the phosphate oxygen atoms, including the 3β-hydroxyl. For carboxyl steroid fragments, the mass for the chemical formula is m/z=xxx+15 Da as the carboxyl fragment must be protonated to generate a cation. Note that all molecules containing only carbon, hydrogen, and oxygen atoms must have chemical formulas with an 'even" mass. Both nitrogen and phosphorus generate 'odd' masses except when an even number of them are present in the molecule (1+1=2).
When the classical steroids were discovered, laboratories used mobility on thin layer chromatography (TLC) to identify individual steroids. However, as not all TLC methods were equivalent, the same compound was designated differently in different laboratories. The use of the mass of the MS fragment as part of the common name eliminates this source of confusion. The proposed stereochemistry is based on the stereochemistry of known enzymes catalyzing similar reactions. If the proposed stereochemistry is not correct, there must be another enzyme which has not yet been described.

What was different in our approach that lead to the discovery
Rather than starting with cross reaction to an antibody to a cardiotonic steroid, our path to discovery was dependent on two physiological starting points. First, Dr. Chasalow was investigating serum from newborn infants with Smith-Lemli-Opitz syndrome (SLO) [7]. Dr. Chasalow had noted that patients with SLO syndrome were potassium wasting and, in the newborn period, were missing two, unknown, steroid-like materials that cross-reacted with various steroid-sulfate antibodies [8]. However, the compounds were not present in serum from 2-week old normal infants. Hence, it was not possible to collect enough serum for characterization of the compounds. Second, Dr. Bradlow had noted that fluids from human breast cysts could be characterized on the basis of their electrolyte compositions [9]. Type 1 fluids had high potassium and low sodium levels while Type 2 electrolytes had the reverse. Together, we speculated that, to account for the difference in electrolytes, Type 1 fluids would have a compound that caused potassium accumulation but that the compound was not present in patients with SLO syndrome. From co-operating physicians, Dr. Bradlow obtained samples of both types of breast fluid. Dr. Chasalow extracted them with the methods used with the blood from newborn infants. The Type 1 fluids were very rich in the compounds that the infants with SLO couldn't make [10]. This recognition provided a pathway to isolation that was distinctly different from methods suitable for isolation of digoxin or ouabain.

Isolation of C313, C341, C339, & C337
Our first source for successful isolation was outdated plasma obtained from our local blood bank [11]. As we couldn't recognize the DLM by UV absorption, we switched to LC-MS to supplement the DLM assay. There were two fractions. Fraction A had UV absorption over 240 nm, consistent with conjugated alkenes, but was not a DLM. Fraction B did not have UV absorption over 240 nm but had DLM.
Fraction A had a peak ion at m/z=496 Da which was attributed to M+1. The key observation was the presence of m/z=184 Da as the base ion. Authentic phosphocholine lipids all showed the base ion at m/z=184 Da. The difference between the base peak and the H+ mass ion was attributed to the steroid fragment. Thus, for the compound in Fraction A with m/z=496 Da, the unknown steroid fragment would have m/z=313 Da and a fragment at that mass was present in the MS. The symbol for this compound would be C313. When the mass of the 3-hydroxyl group (17 Da) was added, C313 would have m/z=330 Da for the intact steroid molecule.  Table 1 shows a trial and error analysis of possible chemical formulas containing only carbon, oxygen and hydrogen atoms with m/z = 330 Da. "C + O" is the sum of the contribution of the carbon and oxygen atoms to the mass of the steroid. Hreq is the number of hydrogen atoms to complete the molecule as observed by MS. Hmax is twice the number of carbon atoms plus two. Delta is the exact number of delta elements needed to provide the Hreq. The line in RED shows the only combination that can make a steroid-like molecule -C21H30O3 -and it would have a Delta of 7 -4 for the steroid rings, 2 for the alkenes at Δ5 and Δ7 and the ketone at carbon 20. Compound isolated by Shackleton from serum of a child with Smith-Lemli-Opitz syndrome [12]. Right Panel. Proposed structure of the H+ ion of C313 and which would have m/z = 496 Da [11].
The major component of Fraction B had an ion at m/z=524 Da. The steroid fragment had a mass of m/z=341 Da (524-183) and a symbol of C341. Table 2 shows the trial and error analysis for the composition of C341. Line 5 shows the only composition that fits the mass of 358 Da is C23H34O3 and Delta must be 7.  Delta  1  21  4  316  42  44  1  2  21  5  332  26  44  9  3  21  6  348  10  44  17  4  22  3  312  46  36  0  5  23  3  324  34  48  7  6  23  4  340  18  48  15  7  24  2  320  38  50  6  8  24  3  336  22  50  14 Trial and error analysis for the composition of a molecule with m/z=358 (C341). The line in RED shows the only combination that can make a steroid-like molecule. Carbon 17 (highlighted in yellow) is a spiral carbon. Spiral carbons form a bridge between two rings. The proposed stereochemistry is based on well-known enzymes: 3β-hydroxy, 5β-reduction, and 17α-hydroxylase. Note that direct reduction of Δ4-3 ketones leads to 5α-stereochemistry. The numbering scheme is based on the standard numbering of steroids. In order to keep carbon 21 in its usual place, the extra carbons have been assigned numbers based on their possible origin in acyl coenzyme A. Figure 3 shows the LC-MS chromatogram obtained from C341 (Ionotropin) [11]. Based on (a) the symmetrical peak shape and (b) the absence of ions at other retention times, the isolated material seemed to be essentially homogenous. In contrast to C313, C341 did not have a UV absorption band over 240 nm, indicating it did not have a conjugated double bond.
In addition to C341, fraction B had two less intense ions, one at m/z=520 Da (C337) and the other at m/z= 522 Da (C339). C337 had a UV band over 240 nm, indicating the presence of a conjugated double bond. The corresponding compositions from trial and error analysis are C23H30O3 and C23H32O3, respectively. Table 3 summarizes the MS fragmentation pattern of steroid phosphoesters. All of these compounds show similar fragmentation pattern, indicating that all are phosphocholine steroid esters. These three compounds plus C313 are the first description of endogenous steroid phosphoesters [11]. C337 (C23H30O3 ), C339 (C23H32O3), and C341(C23H34O3) were first isolated from Fraction B from human plasma. Later, multi-mg samples were isolated from porcine and bovine blood, which had been obtained at local slaughterhouses. Each of these compounds was also isolated to > 90% homogeneity from bovine adrenal extracts. DHEA-PC was synthesized in the laboratory [13]. The ions on line 3 have odd masses, suggesting the loss of trimethyl amine fragment from the phosphocholine steroid.

NMR analysis of steroids with 23 carbon atoms.
1 H-NMR analysis is shown in Figure 4. The spectra are complex with many different types of splitting patterns, which is typical of steroids. Two features were identifiable. First, there were no long chain alkyl groups similar to a fatty acid or fatty alcohol. Second, there were three "unsplit" methyl groups as shown in the inset. Two of these could be assigned to the angular methyl groups at carbon 18 and carbon 19. The third one is consistent with the methyl group at carbon 21, which would be also be unsplit as the adjacent carbon could be either a ketone or an alkene. 13 C-NMR analysis provided a limited amount of new data. It was not consistent with a long chain alkyl group, thus eliminating structures similar to platelet activating factor and other phospholipids.
Analysis of the 31 P-NMR, shown in Figure 5, confirms the molecule has one phosphorus atom. 31 P-NMR chemical shifts vary up to several hundred, depending on the oxidation state of the phosphorus atom, confirming that these two molecules have phosphorus in a similar oxidation state.
If all of the other atoms were carbon, hydrogen, or oxygen, the mass would be odd. This indicates that there must be a nitrogen atom in the molecule because nitrogen is the only common atom that would also generate an odd molecular mass, and two odds make it even. Prior to these investigations, there were no known endogenous steroid phosphoesters. DHEA-3-phosphocholine ester was synthesized in the laboratory and used as a model compound [13].

Biosynthesis of steroids with 23 carbon atoms -part 1
There are two ways to generate a steroid with 23 carbon atoms: either by [a] adding two carbon atoms to a 21-carbon atom precursor or by [b] removing four carbon atoms from a sterol, perhaps cholesterol. Starting with side chain labeled cholesterol, Burstein could not find any evidence for a four carbon atom fragment [14] and we have not observed any ion consistent with cholesterol phosphocholine ester in any LC-MS chromatogram. Further, two of the isolated compounds, C313 and C337, have UV absorption characteristic of conjugated double bonds. As cholesterol does not have a similar UV absorption band, it can't be a precursor of C313. This suggestion is confirmed by the lack of reports of the use of cholesterol tracer to isolate a DLM, as has been done for the other steroid hormones. That dog didn't bark in the night. Slominski showed that the side chain cleavage enzyme cleaved 7-dehydrocholesterol to 7dehydropregnenolone [15]. His team also isolated its 17α-hydroxy metabolite [16], which we would designate P313, if it were phosphorylated. C337 and C339 would fit in the biosynthesis pathway between C313 and C341. The proposed pathway is shown in Scheme 1. The enzyme in the gap has not been described.

Scheme 1.
These four panels show the structures proposed for the first four steroid phosphoesters that we isolated.
The conversion of C337 to C339 is catalyzed by 7-dehydrostrol reductase [17]. The conversion of C339 to C341 is catalyzed by 5-sterol reductase [18]. However, the conversion of C313 to C337 has not been described -the gap. Figure 2 compares the proposed structure of C341 to that of spironolactone. The common feature of these compounds is the lactone E-ring. Table 4 summarizes the different observations that support the proposed structure for C341 (Ionotropin). The composition of the 341 Da fragment must be C23H34O3 with Δ =7. No other combination of carbon, hydrogen and oxygen atoms makes a molecule. See Table 2. Isomers are not prohibited. • C339 and C337 differ from C341 by 1 and 2 alkene groups, respectively, as illustrated in Scheme 1 • C341 cross-reacts with digoxin specific antibodies and is a DLM by definition, whether or not every structural element has been correctly identified.

•
There is no evidence that C341 is a glycoside rather than a phosphocholine ester.

•
Patients with 21-hydroxylase have C341 in their serum, confirming carbon 21 is a methyl group [11].
C363 is a carboxylic acid while C341 is a spiral lactone, which adds 18 Da, [ii] ] the C20-C23 alkene in C341 16 must be reduced in C363, which adds 2 Da and [iii] the carboxylic acid of C363 must be protonated, which 17 adds 2 Da more to its mass to get to C363. Figure 7 shows 3D representation of isomers of carboxylic acids 18 C357 and C359. The actual pair of interest is C361 and C363, but the figures were prepared with C357 and

19
C359 because the double bonds improve visualization of the structures. When the C20-C23 alkene bond is 20 reduced, as in C363, the two oxygen atoms (shown in red) can't get close enough to complete the E-ring

32
This discussion is important for several reasons: first, it confirms the C20-C-23 alkene in the spiral 33 lactones, and, second, it indicates that the carboxylic acids need not be on the biosynthesis pathway, 34 depending on which alkene is present. Note that C363 and C341 have both alkene bonds (Δ5-6 and Δ7-8)

35
reduced, as do both digoxin and ouabain. Thus, it is possible that C363 might fit into the binding site for 36 the phosphocholine portion of the cardiotonic steroids and block their function. 39 40 In addition to the phosphocholine esters, there was a group of compounds that did not have the.

49
Choline is an essential nutrient. As a consequence, the choline kinase pathway is probably not 50 functional in mammals. The basic pathway seems to be the Kennedy pathway -CDP-ethanolamine 51 donation to a sterol and subsequent N-methylation to generate the phosphocholine in situ [20]. There are 52 two ways to generate CDP-ethanolamine. It can be generated either from ethanolamine kinase or from 53 CDP-serine and decarboxylation. The latter pathway seems more likely because serine is not an essential 54 amino acid [21]. In either pathway, the final step would be N-methylation. There are several ACTH-dependent adrenal enzymes that N-methylate amines [22]. Thus, the presence of the Exxx compounds suggests the adrenal is the site of synthesis for the corresponding Cxxx compounds. Perhaps, the Exxx are confirm their freshness [23]. Neither extracts from the whites nor the yolks contained Exxx steroids, 66 indicating biosynthesis probably doesn't occur after the shell is formed. DLM was present in egg white 67 extracts but not in egg yolk extracts. MS-MS confirmed that the major Cxxx extracts in the yolk extracts was 68 C313, which is not a DLM. The egg white extracts contained C337, C339 and C361. These were present as 69 both the Na+ cation and the K+ cation. C337 and C339 are steroid spiral lactones and are DLM. C361 is a 70 carboxylic acid. This study shows the steroid phosphoesters are not limited to mammals but are also 71 present in avian species [24].

72
In preparing this review, I also reviewed the original extracts from turkey and chicken serum. Ions corresponding to C337 and C339, as both the Na+ and K+ ions, were present, but there was no C341 in any     Table 7 summarizes the biosynthetic path based on the compounds we have detected by MS-MS 122 analysis. Note the similarity to the compounds with 23 carbon atoms that we purified to homogeneity.

138
In the original paper describing the steroids with 23 carbon atoms, we proposed a malonic acid 139 condensation driven by decarboxylation [11]. Now that we know about the compounds with 24 carbon 140 atoms and possibly 25 carbon atoms, that process is no longer tenable. Scheme 2 shows the proposed role 141 of acyl coezyme A as the origin of the additional carbon atoms. Reduction at C20-C23 Ring closure -formation of lactone 7-dehydrosterol reduction 7-dehydrosterol reduction 5β-sterol reduction 5β-sterol reduction N-Methylation, the conversion of R3 from A to B, can probably occur at more than one step.

143
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 10 July 2020 doi:10.20944/preprints202007.0211.v1 The table shows the possible range of each of the R Groups for the structures shown in Scheme 2. For each R group there is more than one alternative. Additional R groups are possible but no others have been identified. The three R1 choices are the three most common acyl-Coenzyme A compounds.

Phosphoesters of high molecular mass
In addition to the major compounds in adrenals and ovaries, we have noted a few unexpected ions in the spectra from pregnant women. These ions are in the range characteristic of the steroid fragments and have solubility properties of phosphoesters but parent ions were not recognized under the conditions of analysis. When we recognized the role of acyl coenzyme A in the biosynthesis, we have proposed structures that fit the pattern for the two major unexplained peaks (figure 10).

Some final comments of the discovery of the phospho spiral steroid conjugates
Where possible, in the upper left-hand corner, the codes generated by the instrument are included in the figures. These codes serve to validate the spectra.

Application of MS-MS to investigate spiral steroids in samples from individual patients
To date, investigation of mammalian cardiotonic steroids has been limited because the only method of detection and quantitation has been with antibodies for plant and amphibian toxins. There are no published investigations measuring DLM by LC-MS methodology in serum from individual patients or animals. As described in Section 2.1, there are many compounds that could be detected in various proportions by antibodies to various toxins. The circumstances are now different. We actually isolated >1 mg/L of C341 and C313 from porcine blood. With this huge endogenous level, we developed a MS based method of quantitating individual Cxxx levels in individual samples. The method was first used with samples from patients with documented enzyme defects. The following sections review the first investigations of specific spiral steroid phosphoesters in serum from patients.

Investigation of serum from pediatric patients
For this investigation, serum samples from 40 pediatric patients were collected in the normal course of patient care [11]. There were 5 groups (n=6-8) and a few samples from children with less common diagnosis. The big groups: were: (1) GH deficiency on therapy, (2) type 1 diabetes on therapy, (3) thyroid disorders on appropriate therapy and (4) patients with 21-hydroxylase deficiency on therapy. Most of samples were from pre-pubertal children. LC-MS analysis of these samples were essentially indistinguishable whether the patients were males or females. In all three panels, the central ion at m/z=496 Da is off scale. The m/z=496 Da ion is the H+ parent ion for C313, just like ions at m/z=520 Da, m/z= 522 Da, m/z=524 Da correspond to C337, C339, and C341.

Investigation of serum from patients with 7-dehydrosterol reductase deficiency.
Using our direct MS assay method, we evaluated spiral lactone biosynthesis in patients with 7dehydrosterol reductase deficiency (Smith-Lemli-Opitz syndrome) [27]. Dr. Porter was able to provide 12 samples, 6 from patients with the syndrome (3 males and 3 females) serum from older children with the syndrome and 6 from obligate heterozygotes (3 males and 3 females) but was not able to provide samples from the newborn period. The samples were evaluated without knowledge of the diagnosis but it was quite easy to assign the samples to the proper diagnostic category. Representative spectra are shown in Figure  10. Comparison of the spectra indicates that the serum from the affected patients have decreased serum levels of C341, C353, and C381. This suggests that these three compounds require 7-dehydrosterol reductase for their biosynthesis. Conversely, serum from affected patients has significant levels of C313 (peak at m/z=518 Da) and C329 (peak at m/z=475 Da), indicating that these two compounds probably have a Δ7-8 alkene element in their structure. The low levels of the steroids, rather than their complete absence, are caused by incomplete blockage of the reductase. During the fetal period, potassium is only provided by maternal plasma (5 mM K+) and a hormone that stimulates K+ recovery would be needed. High levels of precursors would be expected to be present [28]. After parturition, the need for a potassium sparing hormone is limited because mother's milk is high in potassium, reducing the need for a K+ sparing hormone. Thus, this pattern is consistent with our index case [8].

Investigation of serum from pregnant women
During pregnancy, pre-eclampsia is the single largest cause of morbidity and mortality. It effects 3 to 5% of pregnancies [29]. Diagnosis is based on the occurrence of both hypertension and proteinuria after 20 weeks of gestation. There is no FDA approved diagnostic method and no therapy. As gestation proceeds, maternal blood pressure increases. If the blood pressure becomes life threatening, the baby is delivered, immediately. After parturition, there is a long-term 2-4 fold increased risk of poor cardiac function for both the affected infant and the mother. The mother also has a 10-fold increased risk of end stage kidney failure [30]. The underlying biochemistry of the enhanced risks is completely unknown. DLM levels increase during gestation [31] but the pattern is very variable [32]. In animal models, administration of either ouabain or marinobufagenin leads to kidney failure [33]. During the second and third trimester, aldosterone signaling is blocked. Pregnant women get pseudo-hypoaldosteronism (high aldosterone levels, but no mineralocorticoid function) and hypoaldosteronism (low aldosterone levels and no mineralocorticoid function. Thus, there are two different unexplained events during normal pregnancy [34]. During the third trimester, whatever its basis, the absence of mineralocorticoid function causes Na+ wasting and is a key regulatory factor in amniotic fluid production. At birth, cord serum has high levels of DLM and, although aldosterone levels are normal, infants are Na+ wasting and fail to grow until, during the 2 nd week post-partum, DLM is no longer present in serum and Na+ wasting ends [28].
It is our theory that these are all related events and dependent on an unidentified spiral steroid hormone [28]. Cardiotonic steroids, detected as assays for DLM, ouabain and marinobufagenin, are all detecting the same hormone. However, no cardiotonic steroid has ever been detected in a vertebrate by any technique other than cross-reaction with antibodies specific to one of the plant or amphibian cardiotonic steroids.
For this investigation, we acquired 40 samples from Global Alliance for the Prevention of Prematurity and Stillbirth (GAPPS) [35]. Twenty samples were from women diagnosed with pre-eclampsia and 20 samples were from normotensive pregnant women who were matched for gestational age and sex of the fetus. For analysis, after addition of miltefosine, each sample was extracted with acetonitrile and analyzed as described in the Methods section. Peak intensities were compared to the intensity of the miltefosine peak at m/z=430 Da. This ion is the Na+ ion. The low intensity of the ion at m/z=446 Da, which would be the corresponding K+, ion confirms the low concentration of K+ ions in the original serum samples. This simplifies the analysis.  @ No ions characteristic of intact steroid esters of either intact phosphoethanolamine or phosphocholine were present. However, in the mass spectrometer, other phosphoethanolamine steroid esters fragment by separation of the steroid fragment. The trial and error analysis for 370 Da (P353) and for 398 Da (P381) only identifies one possible composition. The proposed composition is consistent with a spiral steroid which could be synthesized with the Scheme 2 pathway. We have not identified an appropriate rich source for isolation and have not confirmed that inference.
NP -Not present Figure 14. The relative intensities of the different ions present in serum from pregnant women [35] Left Panel: analysis of serum samples from normotensive pregnant women Right Panel: analysis of serum samples from women with pre-eclampsia. Method of evaluation: For each spectrum, the intensity of each ion was compared to the intensity of the ion generated by miltefosine. The mean and standard deviation of ions from the normotensive women were calculated and used to generate Z-scores for each of the 40 samples. The Z-scores for each sample were graphed as a cluster.
In the samples from the normotensive women, only one sample had a single value with a Z-score >2. The results evaluating the samples from the women with pre-eclampsia are shown on the right. Twelve of the samples had at least one sample with a z-score over 2. That distribution is statistically significant at the o.o5 level. Thus, there seems to be two different diseases. One disease associated with normal levels of PC steroids and a second disease with high levels of PC steroids. However, high levels of C313 and K475 could not account for the high DLM levels in cord serum because neither compound is detected by digoxin specific antibodies [36].
Originally, we thought that the K475 compound was derived from the K+ ion of C313 (m/z=534 Da) after the loss of its trimethylamine fragment. However, GAPPS advised us that no K+ containing reagents were used to prepare any serum samples. Secondly, a magical process would be needed to select only one phosphocholine ion for binding to K+ and not any other, such as miltefosine. We now think that the K475 compound derives from C329 -the 11-hydroxy sister of C313. We now have three sub-groups of patients with pre-eclampsia: Type 1 with high levels of C313, Type 2 with high levels of C329 and Type 3 with low levels of both C313 and C329.
The gestational stage of the women who provided serum for our investigation ranged from 24 to 27 weeks. Symptoms of pre-eclampsia continue to develop as pregnancy progresses. Further changes in the spiral steroids, both qualitative and quantitative, would be expected during the 3 rd trimester.
• Which sub-group is at increased risk of developing renal or cardiac disorders?
• Does persistence after parturition explain the long term increased risk of renal failure?
• Are there differences in the spiral steroids in the fetal circulation?
The ratio of fetal blood volume to maternal blood volume is about 50:1. It is quite possible for the fetus to use C313 or C329 for local synthesis of a spiral lactone for an unknown function. Perhaps C329 binds to the mineralocorticoid receptor and blocks aldosterone signaling. Both compounds have 11βhydroxyl groups. Could that be part of the nuclear receptor binding site?

Is there a role for spiral steroids in Post-Trauma Stress Disorder (PTSD)?
PTSD has been a major interest at the VA Medical Center. Diagnosis is routinely made on the basis of DSM-5 criteria. Once the diagnosis has been established, endocrinologists have noted symptoms that seem to be common in affected patients, typically: (a) inadequate GH secretion, (b) lack of REM sleep, (c) inadequate production of cortisol, despite over secretion of ACTH [37] and (d) hypertension [38]. The steps (a)-(b) -(c) -(x) -(y) -(d) seem to form a cascade but the connection between (c) and (d) has not been recognized. The observation that individuals who experience lack of REM sleep are GH deficient is well known. Less well known is that GH deficient patients secrete a higher proportion of corticosterone to cortisol, then do individuals with adequate GH secretion [39]. Although cortisol is about 4 times more potent than corticosterone as a glucocorticoid, corticosterone is more than ten times more potent than cortisol as mineralocorticoid. This is step (x). Over production of corticosterone causes sodium retention.
Step (y) is synthesis of an endogenous potassium sparing hormone. This restores the electrolyte balance but causes hypertension. The recognition of (x) and (y) completes the endocrine cascade. A new approach to therapy for these patients could be to interfere with the endocrine cascade. In summary, the new understanding of the cardiotonic steroids, particularly the function of Ionotropin (C341) as a potassium sparing hormone, suggests that interfering with the endocrine cascade might provide a new therapeutic approach to PTSD [40].

Discussion
In summary, there are at least 3 classes of spiral lactones -C341 (23 carbon atoms), C353 (24 carbon atoms) and C383 (25 carbon atoms). Based on the various compounds we have characterized, starting with precursors with 21 carbon atoms, the additional carbons are added by condensation of an acyl coenzyme A derivative. These match with the three most common acyl coenzyme A derivatives, which are acetyl, propyl, and acetoacetyl. There are two 21 carbon precursors, C313 and C329, C329 has an extra hydroxy group that might be at carbon 11. These are both present in serum but the 11β-hydroxy could also be added later. There are at least 4 members of each class, for example -C337, C339, C341 and C363. There may also be matching compounds with 11β-hydroxyl groups.
Based on the isolation of the Exxx compounds, the Cxxx compounds are not synthesized by transfer of phosphocholine directly from other phosphatidyl lipids. The steroidogenic tissues (specifically the adrenal and gonads) all have high concentrations of Exxx compounds but the corresponding serum concentration is minimal. Methylation of norepinephrine to epinephrine is regulated by ACTH and cortisol. These could all be targets for future development of therapeutic agents and for a better understanding of many disease processes.
The observation that there are multiple classes of spiral lactones explains some of the conflicting observations about cardiotonic steroids. It also suggests that investigators should not rely on a single antibody to evaluate the concentration of the active material. Finally, the recognition of both PC and PE spiral steroids points to a new site for regulation. Although most of the steroid hormones function via nuclear receptors, that is clearly not the case for all of the phosphoester compounds. C341 is present in serum at a concentration of 1 mg/L. If it functioned molecule for molecule with a nuclear receptor, it would require 2000 mg of receptors for each liter of blood. Hence, the steroid phosphoesters must function by a different mechanism. One mechanism that we already have recognized is via the NaK-ATPase. However, that is only half the story because the apparent half-life is too long to regulate the changes that must take place when we eat or drink. Although the compounds we have isolated and detected in serum are present in high concentration, there may be others that function by classical receptor mechanisms. Because of the relative small size of gonads and the fetus, when compared to an adult weighing 50 kg, this is particularly likely during reproductive processes including fetal development.

Methods:
Extraction: Tissue or blood was homogenized with a mixture of acetonitrile/water, to yield a final ratio of about 4:1. In batches, the extract was applied to a preparative amino carbohydrate HPLC column (22x500 mm) from Alltech (Chicago). Aliquots of each fraction were tested with a saturated silver nitrate reagent. When the test was positive (white precipitate forms), the application of the crude extract was ended and the column eluted with a 40-minute gradient of water/acetonitrile (20/80) to water/acetonitrile (60/40). Individual fractions were analyzed by LC-MS. Fractions containing phosphocholine esters were combined and concentrated under vacuum. Three passes were necessary three to get a single peak. The final step was isocratic elution at water/acetonitrile (18/82). As an example, starting with 10 liters of porcine blood, the final sample weighed about 8 mg. If the isolation process recovered 50% of the amount in porcine blood, then the original serum concentration was about 1.6 mg/L (160 µ g/dL) or 3µ M.
The original analytical LC-MS procedure was as follows: [a] 0.1 ml of serum from a patient sample was pipetted into a 12x75 mm tube, [b] to each tube 0.4 mL of acetonitrile was added, [c] each tube was vortexed, [d] the sample was filtered through a syringe filter, and [e] subject to LC-MS after transfer to an injection vial (MS was Waters, APCI in the positive ion mode). Accelerating voltages were adjusted to minimize fragmentation. This method was used for the pediatric samples and for the original purification of the steroid phosphocholine esters.
NMR spectra were obtained by Acorn Laboratories (Livermore, CA). Trial and error analysis table of possible chemical formulas containing only carbon, oxygen and hydrogen atoms. Each line shows a possible combination of carbon and oxygen atoms necessary to generate a specific mass. The C&O column shows how many Da would be provided by the specific combination of carbon and oxygen atoms on that line. H-Req shows the required number of hydrogen atoms necessary to complete the molecule based on the difference between C&O and 330, the indicated mass of the steroid fragment. Delta is the sum of rings, double bonds and other features that provide the required number of hydrogens. Two hydrogens would be needed for each Delta. Because a steroid has 4 rings, Delta can't be less than 4 nor more than 12 without an aromatic A ring. The line in RED in each table shows the only combination that can make a steroid-like molecule. Multiple isomers are possible but only one chemical formula is possible. The analytical method won't always provide a unique composition, but it does for C341 and most of the other phosphoesters.
Analytical method for analysis of individual serum samples, 10 µ l of a 0.2 mg/ml solution of miltefosine (Cayman Chemical, Ann Arbor, MI) was added to 0.2 ml of serum. The serum was extracted with 0.8 ml of acetonitrile, centrifuged and filtered with Whatman Syringe filters with 0.2µ m pore size. The filtrate was analyzed by direct injection into the electrospray source of a LTQ-XL quadrupole ion trap mass spectrometer (MS) (Thermo Scientific, San Jose, CA). Flow rate was 10 µ l/min. The capillary temperature was 275 C. Spray voltage was 4.4 volts. Ten mass spectra scans were collected and averaged for quantification. Each sample was also analyzed when 60 volts was applied at the source to fragment the molecules. Representative peaks were selected for MS/MS to identify parent/fragment ion relationships. For selection of the parent ions for collision-and induced dissociation, the isolation width was set to 2. All spectra were generated in the positive ion mode.

What did we Discover
• Spiral steroids with 23 carbon atoms • Phosphoester conjugates • Precursors in bovine adrenal cortex and ovarian extracts • Related spiral steroids with 24 and 25 carbon atoms • A new class of 11β-hydroxy steroids • Serum concentrations sufficient for detection by mass spectroscopy with less than 1 ml of serum • Evaluation of serum concentration may be useful for disease diagnosis • Many new targets for transformational development

Overview
Although it is unlikely that every proposed structure is correct, certain basic elements should not be ignored: • Ionotropin, C341 is a an endogenous DLM. Based on its detection by digoxin specific assays, it is a digoxin-like material. It was isolated from human, bovine, and porcine serum. LC-MS analysis shows the absence of major contamination. There is no evidence for a glycoside component, as is present in the plant cardiotonic glycosides. Isolated Ionotropin inhibits NaK-ATPase just like other cardiotonic steroids. In brief, Ionotropin fits the definition of a cardiotonic steroid. • Ionotropin is a phosphocholine ester. It fragments in two ways: loss of trimethylamine and loss of phosphocholine. The 31 P-NMR spectrum shows the presences of a phosphate. A synthetic phosphocholine ester, DHEA-PC, fragments in the same pattern and also has a similar 31 P-NMR. • Based on the a-phosphocholine, the mass of the steroid portion of Ionotropin is 358 Da. This is equivalent to calculating the mass of digoxigenin by subtracting the mass of the a-glycone from the mass of digoxin. • The trial and error method of determining the chemical composition produces only a single result. This need not be the general case, but it is here. This doesn't mean that that there couldn't be isomers but that wouldn't change the points above. Ionotropin would still be an endogenous mammalian cardiotonic steroid. Other scientists are welcome to propose an alternate structure consistent with the overall pattern of steroid phosphocholine esters that is described here.

The Significance
Our efforts have concentrated on the isolation, characterization and biosynthesis of these compounds. Nicholls has proposed that the plant and amphibian cardiotonic steroids are not truly endogenous [41]. This paper reviews the discovery process for the spiral steroid phosphoesters. The data satisfies Nicholls guidelines for an endogenous cardiotonic steroid: (1) isolation and structural identification, (2) precursors and biosynthesis, and (3) correlation with at least one disease process. The hormonal function and regulation of these novel hormones presents the possibility of improved understanding of many endocrine processes. This discovery puts the role of plant and amphibian toxins in a new light. Investigators have now to identify which one is associated with a particular function and/or disorder.