A Short Review on Cardiotonic Steroids and Their Aminoguanidine Analogues

Abstract

A short review on cardiotonic steroids and their analogues is presented. The natu-ral, semisynthetic and synthetic derivatives, as well as their mechanism of action and struc-ture-activity relationships are shown, with a special reference to aminoguanidine deriva-tives.

1. Introduction	3.2.1. Structure-activity relationships
2. Mechanism of inotropic activity	3.2.2. Steroidal framework
2.1. Na+,K+-ATPase	3.2.3. Side chain at C-17
2.2. Isoenzymes	3.2.4. Sugar
3. Compounds displaying inotropic digitalis-like activity	3.2.5. Other substituents at position C-3
3.1. Natural digitalis-like compounds	3.2.6. Pregnane derivatives
3.1.1. Digitalis-like glycosides in vegetal species	3.3. Endogenous digitalis-like factors
3.1.2. Digitalis-like glycosides in animal species	4. Digitalis analogues bearing aminoguanidine moieties
3.1.3. Natural digitalis-like glycosides recently isolated	4.1. Aminoguanidine moiety in drugs
3.2. Semisynthetic and synthetic digitalis-like compounds	4.2. Aminoguanidines in digitalis compounds

1. Introduction

Digitalis glycosides are one of the most useful groups of drugs in therapeutics. The use of plants containing them has been known for a long time, but it was in 1785 when William Withering described the properties and medicinal uses of Digital [1], in particular for the treatment of dropsy, being aware of side effects and toxicity associated to high dosages. It was not until 1799 when the importance of the direct action of digitalis on the heart was considered [2].

Among the different cardiac glycosides present in Digitalis purpurea, gitaloxin (16-formylgitoxin) seemed to be the main responsible for therapeutic activity in the "foxglove tea" prescribed by Wither-ing, due to its water solubility [3]. Digitoxin is present in similar concentration, but its low water solu-bility makes it more difficult to remove in aqueous extracts. The use of dried leaves of this plant has remained as the main treatment of cardiac failure for a long time.

Digoxin and its derivatives (acetyl- and methyldigoxin) are the cardiac glycosides most currently used in therapeutics.

2. Mechanism of Inotropic Activity

Digitalis glycosides can be defined as allosteric inhibitors of Na⁺,K⁺-ATPase, they seem to be chemically inert [4] and not covalently bound to the enzyme [5].

According to the still most widely accepted mechanism of action for digitalis type compounds [6], they act through inhibition of Na⁺,K⁺-ATPase, thus raising indirectly the intracellular Ca²⁺ concen-tration ([Ca²⁺]_i). Therapeutic concentrations of digitalic compounds produce a moderate enzyme inhi-bition (about 30%). When the cell is depolarised, there is a lower amount of enzymes available for the restoration of the Na⁺/K⁺ balance. The remaining enzymes, non-inhibited, will act faster, because the high [Na⁺]_i and the ionic balance must be restored before the following depolarisation, although it will take longer than if every enzyme were available. This lag causes a temporary increase of [Na⁺]_i, reaching higher concentrations than if ATPase activity were not partially inhibited. This temporary in-crease of [Na⁺]_i, modifies [Ca²⁺]_i through a Na⁺/Ca²⁺ exchanger which allows Na⁺ exit from the cell in exchange for Ca²⁺, or Ca²⁺ exit from the cell in exchange for Na⁺, depending on the prevailing Na⁺ and Ca²⁺ electrochemical gradients [7]. This mechanism decreases exchange rate, or even re-verses exchanger ion transport, being Ca²⁺ carried into the cell; anyway increasing [Ca²⁺]_i and thus increasing contractile force.

When the concentration of digitalis compounds reaches toxic levels, enzyme inhibition is too high (>60%), thus decreasing Na⁺ and K⁺ transport to the extent that the restoring of normal levels during diastole is not possible before the next depolarisation. Then, a sustained increase of [Na⁺]_i, and thus of [Ca²⁺]_i, gives rise to toxic effects (i.e. arrhythmia) of these compounds.

2.1. Na⁺,K⁺-ATPase

Na⁺,K⁺-ATPase (EC 3.6.1.37) or sodium pump, is a carrier enzyme present in almost every animal cell. Its physiological function is to maintain the Na⁺ and K⁺ electrochemical gradients through the cell membrane, keeping low Na⁺ and high K⁺ intracellular concentrations. Such concentrations of ions, their gradients and the consequent membrane potential determine a broad range of cellular functions, as excitability of nerves and muscle cells, secondary active transport and cellular volume regulation. It is estimated to consume about 25% of total ATP consumed at rest. The enzyme was discovered by Skou in 1957 [8].

The minimum functional entity of the Na⁺,K⁺-ATPase is formed by one α and one β subunit. Each one of these subunits has different isoenzymes. The α subunit, with about 112 kD and 1012 aminoac-ids, presents ten transmembrane segments, with N–terminal and C–terminal ends into the cytoplasmic side [9]. It is the catalytic subunit, undergoing conformational changes relating the energy produced from ATP hydrolysis and enzyme phosphorylation to the transport of cations [10]. The β subunit (about 55 kD and 300 aminoacids) has just one transmembrane segment, a short N–terminal end to-wards the cytoplasmic side [11] and a large extracellular segment, bearing three disulphide bridges [12]. The β subunit is the limiting one for the formation of the functional enzyme. α and β subunits are synthesized in a different way and then they are included into the endoplasmic reticulum, where they are assembled, so determining the structural maturation of α subunit and the acquisition of its main functional features [13].

Both α and β subunits are glycosidated. For the β subunit, three to seven extracellular glycosidation sites have been found, depending on the considered isoenzyme. The α subunit is also glycosidated, al-though in this case by a monosaccharide (likely N-acetylglucosamine) and disposed towards the cyto-plasmic side [14].

Moreover, there is a third γ subunit, a membrane protein that interacts only with Na⁺,K⁺-ATPase, thus modulating the enzyme transport activity [15]. It has 12-14 kD mass, with about 58-68 aminoac-ids. This γ subunit is an anphiphilic protein, bearing two structural domains: the amino terminal frag-ment is highly hydrophilic, carrying many charged aminoacids, and it must be extracellular; the car-boxy terminal end is hydrophobic and likely lay across the membrane [16]. The γ subunit only inter-acts with the α-β complex assembled and with functional ability, but not with separate α or β subunits, inducing enzyme activity [15b].

In addition to the protein subunits negatively charged phospholipids are needed for enzyme activity [17], even when the cause of this is not clear. The membrane lipid bilayer could have a structural effect on the protein, maybe allowing it to carry out the necessary conformational changes for activity. A regulatory effect of phospholipids on the enzyme has not been shown.

Na⁺,K⁺-ATPase acts as a dimer (αβ-αβ). The most widely accepted view related to such a dimer to act is a “flip-flop” model, in which both α subunits show complementary conformations [18]:

E₁E₂⇌E₂E₁

where E is the conformation of each α subunit. The observed co-operation between the members of the dimer [19] supports this model, which postulates that, when one of the α subunits is in the E₁ confor-mation, the other one is necessarily in the E₂. Different results have confirmed such a dimeric structure [20]. It seems that structural contacts between both αβ units are needed for a complete catalytic activ-ity of the enzyme.

Related to the transport activity, the enzyme takes out 3 Na⁺ in exchange for 2 K⁺ carried into the cell. So, it allows the restoration of the appropriate Na⁺:K⁺ ratio to maintain the transmembrane differ-ence of potential [21]. It requires ATP and Mg²⁺ for activity. Binding of ligands to the enzyme, in-cluding a phosphorylation step, leads to conformational changes associated to Na⁺ and K⁺ transport. The supposed mechanism of action currently accepted was firstly proposed by Albers [22] and Post [23]. This mechanism includes a step in which the enzyme, after leaving out 3 Na⁺ and before taking in 2 K⁺, can be bound, and thus inhibited, by digitalis glycosides or their analogues, preventing K⁺ bind-ing and then stopping enzyme activity.

2.2. Isoenzymes

Classification of α subunit isoenzymes was initially based on the electrophoretic mobility of their polypeptides and their sensitivity towards digitalis glycosides. Currently three isoenzymes are differ-entiated, α₁, α₂ and α₃ [24]. A fourth one has also been identified, although only at the genetic level [25]. The classification is based on the aminoacid number and the comparison between total and N-terminal end aminoacids sequence.

Related to the affinity found for the different α isoenzymes towards digitalis glycosides, α₁, present in almost every cell (it is the only one found in mammal kidney), shows the lowest affinity. α₂ is largely found in skeletal and cardiac muscle, fatty tissue and stomach, and its affinity is higher than that of α₁. α₃ isoenzyme is mainly expressed at the brain, but it is also found in other vertebrate tissues. Its affinity toward digitalis glycosides is the highest.

Among β isoenzymes, β₁ is present in all tissues [26], being the main isoenzyme found in mammal kidney. β₂ has been isolated from brain [27]. Recently, a third , β₃, has been identified in mammals, including the human being [28] and other animal species [29]. Any of α subunits can be assembled with any of β subunits, giving thus rise to different enzyme complexes.

The different Na⁺,K⁺-ATPase isoenzymes come from different genes, which in turn show a signifi-cant chromosome dispersion [30]. Each one of the genes corresponding to each α or β subunit is ex-pressed at different rates and in different degrees, depending on the type of tissue and its development period [31].

Na⁺,K⁺-ATPase is regulated by Na⁺ and K⁺ concentrations, as well as by several hormones, as al-dosterone, thyroid hormones, catecholamines and peptide hormones (vasopresin or insulin). Hormone regulation can be carried out at different levels, from cell surface to nucleus, and it can be expressed at short or long term [32].

3. Compounds Displaying Inotropic, Digitalis-like Activity

It is understood as “digitalis activity” the ability to inhibit Na⁺,K⁺-ATPase through acting onto the digitalis receptor, along with the ability to display a positive inotropic effect. Such an action is per-formed by several natural, semisynthetic and synthetic compounds [33]. Among the natural com-pounds, there are three groups: steroidal butenolides and pentadienolides, known as “cardiotonic ster-oids” or “digitalic compounds” and Erythrophleum alkaloids. The word “digitalis” is often used as a generic word for all cardiotonic steroids; similarly, the receptor for these compounds is generally known as “digitalis receptor”.

Digitalis glycosides are compounds bearing a steroidal genin or aglycone with one or several sugar molecules attached to position C-3. In the case of toad venom, sugar is replaced by suberylarginine.

Fig. 1 depicts two of the most therapeutically used digitalis glycosides, digoxin and methyldigoxin. They represent a very important group of drugs for the treatment of heart failure, but display a main disadvantage, which arises from their narrow therapeutic index, so they have to be administered under a strict supervision. The proximity between effective and toxic doses is the cause of severe adverse ef-fects to appear.

Na⁺,K⁺-ATPase inhibition at therapeutic doses is the cause of their positive inotropic effect, since only little changes in [Na⁺]_i are required for a large effect on contractile force [34]. Apart from this activity, they can act on other physiological systems, leading to adverse effects [35].

3.1. Natural digitalis-like compounds

This kind of compounds are found in both animal and vegetal species [36], being the most well known those obtained from Digitalis. In most cases they are cis-anti-trans-syn-cis steroids, bearing a hydroxyl group at position C14β, a butenolide or pentadienolide lactone ring at position C17β and a glycosidic moiety at position C3β carrying one or several sugar units.

Figure 2. Structure of cardenolides and bufadienolides.

Figure 2. Structure of cardenolides and bufadienolides.

Erythophleum alkaloids, some of them are shown in Fig. 3, display cardiotonic activity acting through the same mechanism, but have a different terpenic structure.

3.1.1. Digitalis-like glycosides in vegetal species

Vegetal species containing digitalis-like glycosides mainly belong to the Angiosperms, although some of them are found in certain Gymnosperms [37]. Digitalis-like glycosides isolated from different Fanerogam families [38] are shown in

Table 1. Fanerogam species containing digitalis-like glycosides.

Family	Species	Cardiotonic glycosides
Apocynaceae	Nerium oleander	Oleandrin, neriin, neriantin.
	Nerium odorum	Odoroside A and B.
	Strophantus gratus, S. kombe, S. his-pidus, S. sarmentosus, S. emini	Ouabain (G-strophantin), cymarin, sarmentocyma-rin, periplocymarin, K-strophantin.
	Acokanthera schimperi (A. ouabaïo), A. venenata, A. abyssinica	Ouabain.
	Thevetia nereifolia	Thevetin, cerberin, peruvoside.
	Thevetia yecotli	Thevetosin, thevetin A.
	Cerbera odollam	Cerberin.
	Cerbera tanghin	Tanghinin, deacetyltanghinin, cerberin.
	Adenium boehmanianum	Echujin, hongheloside G.
Asclepiadaceae	Periploca graeca	Periplocin.
	Periploca nigrescens	Strophantidin, strophantidol, nigrescin.
	Xysmalobium undulatum	Uzarin.
	Gomphocarpus fruticosus	Uzarin.
	Calotropis procera	Calotropin.
Brassicaceae	Cheiranthus cheiri	Cheiroside A, cheirotoxin.
Celastraceae	Euonymus europaeus, E. atropur-pureus	Eounoside, euobioside, euomonoside.
Crassulaceae	Kalanchoe lanceolata	Lancetoxin A and B.
	Kalanchoe tomentosa	Kalanchoside.
	Kalanchoe tubiflorum	Bryotoxin A-C.
	Kalanchoe pinnatum	Bryotoxin C, bryophyllin B.
	Tylecodon wallichii	Cotiledoside.
	Tylecodon grandiflorus	Tyledoside A-D, F and G.
	Cotyledon orbiculata	Orbicuside A-C.
Fabaceae	Coronilla sp.	Alloglaucotoxin, corotoxin, coroglaucin, glau-corin.
Iridaceae	Homeria glauca	Scillirosidin derivatives.
	Moraea polystachya, M. graminicola	Bovogenin A derivatives.
Liliaceae	Urginea scilla, U. maritima	Scillarene A and B, scilliroside, scillarenia, scillia-cinoside, scilliglaucoside, scilliglaucosidin, scil-liphaeosidin, scilliphaeoside, scillirosidin, scilliru-brosidin, scillirubroside, proscillaridin A.
	Urginea rubella	Rubelin.
	Convalaria majalis	Convalloside, convallatoxin.
	Bowiea volubilis, B. kilimand-scharica	Bovoside A, glucobovoside A, bovoruboside.
Moraceae	Antiaria africana, A. toxicaria	Antiarin α.
Ranunculaceae	Helleborus niger, H. viridis, H. foeti-dus	Helleborein, helleborin, hellebrin.
	Adonis vernalis, A. aestivalis, A.autumnalis, A. flammea	Adonidin, adonin, cymarin, adonitoxin.
Santalaceae	Thesium lineatum	Thesiuside.
Scrophulariaceae	Digitalis purpurea, D. lanata	Digitoxin, gitoxin, gitalin, digoxin, F-gitonin, digitonin, lanatoside A-C.

.

3.1.2. Digitalis-like glycosides in animal species

They occur mainly in toads, either as free genins (bufagins) or bound to suberylarginine (bufotox-ins). The genins are shown in

Table 2. Toad species having digitalis-like compounds.

Species	Genins
Bufo vulgaris	Bufotalin, bufotalinin, bufotalidin.
Bufo japonicus	Gamabufagin.
Bufo gargarizans	Cinobufagin.
Bufo marinus	Marinobufagin.
Bufo arenarum	Arenobufagin.
Bufo regularis	Regularobufagin.
Bufo valliceps	Vallicepobufagin.
Bufo quercicus	Quercicobufagin.
Bufo viridis	Viridibufagin.
Bufo sp.	Pseudobufotalin.

.

They have also been isolated from glow-worms and snakes, although in these cases only as genins.

3.1.3. Natural digitalis-like glycosides recently isolated

Currently, this kind of compounds continues to be isolated from vegetal and animal species. Some of such compounds identified during the last decade are:

From Digitalis canariensis: digitoxigenin-3-O-β-glucosyl-(1->4)-3’-O-acetyl-β-D-digitoxoside [39].

Figure 4. Structure of digitoxigenin-3-O-β-glucosyl-(1->4)-3’-O-acetyl-β-D-digitoxoside (VIII).

Figure 4. Structure of digitoxigenin-3-O-β-glucosyl-(1->4)-3’-O-acetyl-β-D-digitoxoside (VIII).

From Acokanthera spectabilis: acobioside A and 14-O-acetylacovenioside C [40].

Figure 5. Structure of acobioside A (IX) and 14-O-acetylacovenioside C (X).

Figure 5. Structure of acobioside A (IX) and 14-O-acetylacovenioside C (X).

From Asclepias vestita: 5,6-dehydrocalactin [41].

Figure 6. Structure of 5,6-dehydrocalactine (XI).

Figure 6. Structure of 5,6-dehydrocalactine (XI).

From Nierembergia aristata [42]:

Figure 7. Digitalis-like glycoside with cardiotonic activity isolated from Nierembergia aristata (XII).

Figure 7. Digitalis-like glycoside with cardiotonic activity isolated from Nierembergia aristata (XII).

From Periploca sepium [43]:

Figure 8. Digitalis-like glycoside displaying cardiotonic activity isolated from Periploca sepium (XIII).

Figure 8. Digitalis-like glycoside displaying cardiotonic activity isolated from Periploca sepium (XIII).

From Apocynum cannabinum [44]:

Figure 9. Cardiotonic digitalis-like glycosides isolated from Apocynum cannabinum (XIV and XV).

Figure 9. Cardiotonic digitalis-like glycosides isolated from Apocynum cannabinum (XIV and XV).

During the last decade some other compounds of natural origin have been described. Nevertheless, their structure is different from that of compounds above described, and their mechanism of action, even when they display positive inotropic activity, do not involve acting onto digitalis receptor, but on other physiological systems, so they have not been included in this review.

3.2. Semisynthetic and synthetic digitalis-like compounds

Apart from digitalis-like compounds isolated from natural sources, there is a wide range of semi-synthetic and synthetic analogues. These have been prepared by introducing different modifications on the original compounds, thus trying to determine which are the essential structural requirements for displaying cardiotonic activity and, on the other hand, to obtain derivatives with less toxicity, i.e., a broader therapeutic index. From the research on those compounds some structure-activity relationships (SAR) have been deduced.

3.2.1. Structure-activity relationships (SAR)

SAR studies on steroidal cardiac glycosides [6][45] have led to the conclusions exposed below and the proposal for the existence of a receptor model containing different areas:

-

A first one for the interaction with the steroidal framework, through hydrophobic binding.

-

A second one for interaction with the lactone ring showing two binding points, an electrostatic in-teraction through the electron-deficient β-Carbon atom and a Hydrogen bond to the carbonylic Oxygen atom.

-

A third one for interaction with sugar: hydrophobic binding through C5' and Hydrogen bond to C3'-OH.

-

A forth one, from which, Hydrogen bonds to the β face of the molecule could be established, in the case of having such groups.

3.2.2. Steroidal framework

In Digitalis-like glycosides, the steroidal framework is considered the pharmacophoric moiety, re-sponsible for the activity of these compounds [46]. Specifically, the 5β,14β-androstane-3β,14-diol skeleton has shown the same binding properties to the enzyme as digitalis compounds. As deduced from energetic calculations related to pharmacophore-receptor interaction, most of interaction energy arises from Van der Waals forces, which are very dependent on the distance [47]. Electrostatic mo-lecular potential calculations, in turn, show that the steroidal framework is surrounded by a positive field, its charge arising from C-H bonds, so it is supposed that aminoacids interacting with such a moi-ety should bear negative charges.

Moreover, structure bending, owed to the cis junctions between A/B and C/D rings, is an essential request for the highest interaction energy. Any change on that spatial disposition, modifying A and/or B rings related to B-C plane, decreases the interaction energy [48]. This can be observed in the case of variation of the configuration 5β to 5α or introduction of a Δ⁵ double bond, even when such a decrease is not very high. Similarly, a higher interaction energy decrease is observed if the geometry of D ring is modified, through the conversion of 14β-OH to 14α-OH or the introduction of double bonds Δ⁸⁽¹⁴⁾, Δ¹⁴, Δ^14,16, Δ^8,14,16 or Δ¹⁶. Such a decrease can be partially related to the flexibility conferred to the molecule by double bonds, in contrast with the rigidity of the steroidal framework. This fact can also explain some reduction in the specificity (discrimination among ATPases from different origins).

The presence of OH groups at different positions of the steroidal skeleton reduces, in general, the interaction energy, though it depends on the location and spatial disposition of such OH groups. This fact seems to be due to decreasing of steroidal positive potential field and steric hindrance. Further-more, the effect of OH groups varies depending on whether the cardenolide or the pregnane framework are considered, which in turn could depend on the characteristics of the side chain at position C17, bulky and rigid or small and flexible. The OH group at position C14β is not an essential feature for inotropic activity, although when it is replaced by a Hydrogen atom potency decreases considerably [49]. Particularly, an OH group at position C16β reduces potency in a large extent, but if such an OH is esterified, potency increases. Several explanations have been proposed for this fact: the existence of a binding site for such ester or the presence of the unesterified 16β-OH group that would make possi-ble a Hydrogen bond to the OH at position C14β, producing a modification of the spatial disposition of ring D and thus the interaction of the lactone with its receptor [50].

Conversion of 3β-OH to 3α-OH brings about a strong decrease of interaction energy; such a de-crease is not so high if conversion goes towards 3-ketone or ester formation. The later effect can be owed to transformation of an unpaired Hydrogen bonding moiety into a strong Hydrogen bond-acceptor one.

The change of the A/B junction does not necessarily imply a decrease of activity of aglycones, al-though it does for the corresponding glycosides, indicating that the main influence of A/B junction arises from its ability to place the sugar into a suitable position.

Although it was firstly thought that only C/D cis steroids (5β,14β) could display cardiotonic activ-ity, it has been observed later that some C/D trans steroids (5β,14α) show a similar or even stronger interaction with the digitalis receptor. The low potency of pregnanes with this configuration is owed to the side chain at C17, which is not contributing to interaction, and its replacement by a lactone ring does not increase such energy, being even possible to cancel out the activity. Thus, the spatial disposi-tion of either the lactone ring or its equivalent moiety is essential for digitalis activity.

In the same way, D-seco and D-homosteroids have been evaluated [51], concluding that the D ring is not essential for cardiotonic activity. Nevertheless, for D-homoderivatives, less potent than their par-ent homologues, C/D cis junction is needed for displaying cardiotonic activity, similarly to classic digitalis compounds.

Simplified cardiotonic steroid analogues with a non-steroidal framework have also been synthesized and evaluated as inotropic agents. For instance, diterpene analogues, lacking the D ring, bearing A/B trans junction and other structural features considered important for positive inotropic activity, as the lactone ring and the hydroxyl group at C14β, have been synthesized [52]. However, none of these diterpene analogues displayed a significant inotropic activity.

Other simplified analogues with non-steroidal framework are hydroindene derivatives, which mimic C/D cis rings. Moreover, they carry a hydroxyl group at an equivalent position to the C14β of steroids, and an open chain fragment replacing the lactone ring at the equivalent position. Several families of these analogues have been synthesized [53] and their conformational study has been carried out, taking into account the stability of different conformers depending on substituents attached to the hydroin-dene skeleton [54]. Inotropic activity of these analogues has also been evaluated [55].

3.2.3. Side chain at C17

The lactone ring at C17β has been considered for a long time to be responsible for inotropic activ-ity, bringing about conformational changes on the enzyme that would give rise to its inhibition [56]. Indeed, that is the most differentiating feature from steroid hormones, and its contribution to the inter-action energy (-20.5 or -27.6 kJ/mol for the butenolide and pentadienolide, respectively) is as large as that of the whole steroidal framework (-22.7 kJ/mol) [57]. Then, a very significant research was car-ried out on simple lactones, with the aim of assigning the cardiotonic activity of digitalis compounds to the lactone moiety [58], but none of them showed such an activity. This, together with the fact that there is not a “pre-existing” binding site for the lactone ring on the receptor, indicates that for large en-ergetic contribution of such lactone ring it is required the “building up” of its binding site through a conformational change promoted by the steroidal binding.

Surrounding the lactone ring there is a negative electrostatic potential. One of the interaction forces between the lactone ring and its binding site seems to be a Hydrogen bond, owing to Oxygen ability to act as acceptor for such bonds. In fact, there is a high negative potential surrounding both lactone Oxy-gen atoms [59]. Such Oxygen atoms can participate in up to two Hydrogen bonds, thus increasing di-rectionality of bond formation [60]. Together with those considerations, involving of lactone ring in formation of Hydrogen bonds is confirmed by the fact that most of the substituents at C17β (except OH, OCOMe and COOMe) increase interaction energy, but the highest increase is found with bute-nolide and pentadienolide [45b].

The lactone ring has been replaced by other heterocycles in some semisynthetic derivatives. In the case of pyridine or piridazine rings, energetic contribution is only partially kept, even when Nitrogen can also act as Hydrogen bond acceptor. Replacement by furan decreases interaction energy by 2.8 kJ/mol, the approximate energetic contribution of lactone Hydrogen bond [57]. Replacement of Oxy-gen by Nitrogen (furanone by pyrrolidone) decreases in large extent enzyme inhibition [58b], even when replacement by Sulphur does not modify interaction energy (because Sulphur can also act as Hy-drogen bond acceptor [57]).

Figure 10. Semisynthetic derivatives with heterocycles replacing the lactone ring.

Figure 10. Semisynthetic derivatives with heterocycles replacing the lactone ring.

Some other modifications carried out on the lactone ring [46a] [61] are shown in Fig, 11.

In addition, it has to be considered the strong dependence of Hydrogen bond energy on distance and angle of such a bond, which can explain the large variations observed when modifying spatial disposi-tion of carbonylic Oxygen [62], e.g., changing the attachment position of the lactone ring from C20 to C22 (actodigin and related compounds [63]) or to C21.

Reduction of the Δ²⁰⁽²²⁾ double bond, leading to loss of flat geometry at such bond and to modifica-tions of spatial disposition of Oxygen atoms, decreases the interaction energy. This can be explained partially by changes of Hydrogen bond accepting properties of Oxygen atoms and partially to steric hindrance.

Rotational energy barrier for C17-C20 bond is not high in comparison with the interaction energy between cardenolides and their receptor [64], so the proximity of the compound to its asymmetric re-ceptor potential, which determines the stereospecificity of binding site, can promote the conforma-tional disposition of the lactone ring in the binding to receptor. Data coming from different studies suggest that such conformation is 14,21, i.e., H21 and HO14 in the same direction. Moreover, it seems that a certain degree of conformational freedom is required for such interaction, so compounds can adopt the most suitable 14,21 conformation for binding to the receptor.

Replacement of the lactone ring by open chains, bearing an electron rich heteroatom and a conju-gated double bond gives high interaction energies. Thomas et al. suggested that, together with a Hy-drogen bond, there is a dipole-dipole interaction in lactone binding [6], but they place it at C20, while it is very much likely located at double bond, showing a strong electronic density decrease area at the butenolide plane [65].

Those open chains at C17β can adopt, through rotation about C22-C23 bond, a butenolide-like con-formation, thus contributing similarly to the interaction with the receptor [66]. α,β-unsaturated methyl ester is, after α,β-unsaturated nitrile, the most potent analogue. Nevertheless, α,β-unsaturated esters show decreased potency with increasing size in alkyl chain (Fig. 12).

Insertion of a methylene between C17 and C20, or modification of the lactone ring attachment from C17β to C17α lead to the loss of lactone ring contribution to interaction energy [68]. Several inactive derivatives modified at the C17 chain are shown in Figure 13.

Aminoacids involved in lactone binding have to be close to the steroid binding site, located at the first Na⁺,K⁺-ATPase transmembrane segment. It is supposed that the Hydrogen bonds to the lactone Oxygen atoms are established by the amide groups of Gln119 or Asn122. The strong electronic defi-ciency borne at the butenolide plane could be complemented by the negative charge of Asp121, which is closely related to ouabain binding.

3.2.4. Sugar

Sugar attachment to the steroid modifies both pharmacokinetics and pharmacodynamics of digitalis glycosides. Free genins are absorbed faster than glycosides, stored in larger extent at the central nerv-ous system and more easily metabolized to less active 3α-OH epimer, in consequence the action of free genins is fast and short-lasting but no free of toxic effects.

Significance of the sugar moiety for digitalis activity is well established, though sugars by them-selves do not display any activity. Indeed, it seems that rhamnose and glucose do not bind to Na⁺,K⁺-ATPase [69]. Sugar binding site is not “pre-existing” at the receptor, coming up during glycoside binding. The observed potency sequence is: monosaccharide-aglycone > disaccharide-aglycone > tri-saccharide-aglycone >> aglycone, indicating that only the first sugar molecule is involved in receptor binding [70]. Decreasing of the interaction energy related to the second and third sugar units can be due to increased rotational and translational entropies. The distance from the lactone carbonylic Oxy-gen to the Oxygen at C4’ of the first sugar is close to 19Å, thus indicating the receptor length.

There is a wide variety of sugars carried by digitalis glycosides, from the very often found in Na-ture, as glucose, galactose, mannose or rhamnose, to other not as usual, as digitoxose, thevetose, cyma-rose, allose or altrose. Both β-D- and α-L-sugars may contribute to increase the interaction energy.

Figure 14. Structure of some of the sugars present in digitalis glycosides.

Figure 14. Structure of some of the sugars present in digitalis glycosides.

In relation with the sugar type, highest potency is conferred by 6’-deoxysugars, so 5’-methyl group is important for binding. Moreover, an equatorial 4’-OH group is also important for binding, and not a 3’-OH, while the 4’-OH axial group is much less effective [71]. In spite of these general considera-tions, the influence of different substituents depends on the regarded sugar [72]. Sugars as rhamnose or thevetose can increase potency several times, while mannose has no effect [73]. On the other hand, fu-ranose rings do not seem to be as active as pyranose rings.

The highest interaction with the receptor is found with α−L-rhamnose and its 4’-amino-4’-deoxy derivative (although if other sugars are analyzed, potency depends on the aglycone structure). So, it seems that sugar interaction comes through 4’-OH or 4’-NH₂ of rhamnose and receptor aminoacids acting as Hydrogen bond donors. Since the NH₂ group is even better than the OH group, it is supposed that binding to an acid proton (i.e., a carboxylic group), would represent the strongest bond [65]. In fact, attachment to C3β of substituents such as glucuronide, sulphate or esters from carboxylic acids decreases or eliminates activity.

As pointed out before, replacement of sugars by aminosugars sometimes increases activity. For ex-ample, compounds shown in Fig. 15 have a better therapeutic index and a 10-fold higher potency than digoxin [74].

Digitalis-like compounds bearing two-point attached sugars can also display high activity, e.g., gomphoside (Fig. 16).

Some studies suggest that recognition of certain structural sugar features depends on the studied isoenzyme, but in the case of the main structural features such a dependence does not exist [75].

3.2.5. Other substituents at position C3

Although most of the described substituents are sugars, other moieties have been attached at that position. There have been mainly described esters (Fig. 17) showing different lengths and functionali-zations at the end of the chain, alcohol, ketone, carboxylic acid and amide [6]. The amide series has elicited the highest activity.

Recently, some 3β-aminoalkylthio moieties have been introduced, with the aim of checking the in-fluence of a free or alkylated amino group [76], showing in most cases a higher affinity than for the corresponding 3β-OH and 3β-SH derivatives. Previously, there has been only described a strophantidin derivative of this type [77], with affinity higher than that of the corresponding genin.

Some C3-amino derivatives have also been described [78].

The sugar binding site would be located at the third transmembrane segment, which includes ami-noacids such as Glu307 and Tyr308. Both of them bear Hydrogen atoms than can be used for Hydro-gen bonding [79]. Then, steroid binding to its site promotes folding of the first transmembrane segment to the third one, building up a three-dimensional environment that becomes the whole digitalis recep-tor. Such a conformational change would affect the geometry of the catalytic centre, dramatically de-creasing enzyme affinity for its substrate, ATP [80].

3.2.6. Pregnane derivatives

Pregnane derivatives eliciting digitalis-like activity are interesting because they can represent a structural possibility for endogenous digitalis-like compounds [81]. There have been identified some pregnane derivatives which could display such an activity, and their ability to displace ³H-ouabain from its Na⁺,K⁺-ATPase binding site has been evaluated [82]. These results gave rise to a systematic search for synthetic pregnane analogues.

Thus, several families of 5β,14β-pregnane and 21-norpregnane derivatives, carrying at C20 hy-droxyl, amine, acetamide, nitro and other functional groups, were prepared (Fig. 18). Both series have shown a very similar activity related to inhibition of ³H-ouabain binding [83].

The nitro group has shown to be the best substituent, being genins much less potent than glycosides. Rhamnose is the sugar conferring the highest potency. Related to the stereochemistry, β disposition is most favourable. With these substituents, specifically, changing from 17β to 17α implies a 20-fold de-crease in potency.

Comparison between pregnane (α or β) and the corresponding 21-norpregnane derivatives indicates a higher potency for the latter [84]. This effect could be due either to the restriction in rotation pro-duced by methyl at C21, preventing the tightest binding of C20 group, or to the own methyl preventing the strongest binding to receptor. Nevertheless, conformational study of these compounds shows, for 21-norderivatives, a closer conformation to less potent 20α than to more potent 20β. 20-nitro-14-hydroxy-3β-(α-L-rhamnopiranosyloxy)-21-nor-5β-14β-pregnane showed a similar potency to digi-toxin in ³H-ouabain-binding displacement.

Preliminary studies indicate that some of these derivatives (nitro, amine) could have a better therapeutical index than digitoxin.

Pregnane derivatives bearing amine group at C14β and different moieties at C3 and C17 have been prepared [85]. Among them, the best therapeutical index has been shown by compound LND623 [86], being the 20α-OH epimer more potent than the 20β-OH [87]; this suggest that only the 20α-OH could accept a favourable Hydrogen bond, possibly acting through formation of one or two Hydrogen bonds.

Figure 19. Structure of some 14β-aminopregnane derivatives.

Figure 19. Structure of some 14β-aminopregnane derivatives.

Related to the stereochemistry of C/D rings junction, megestrol acetate and chlormadinol acetate (CMA), bearing C/D trans junction, inhibit Na⁺,K⁺-ATPase [88]. CMA showed a potency similar to digitoxigenin in in vitro ³H-ouabain binding assays.

Figure 20. Structure of megestrol acetate (XX) and chlormadinol acetate (CMA, XXI).

Figure 20. Structure of megestrol acetate (XX) and chlormadinol acetate (CMA, XXI).

A detailed analysis of the CMA structure [89] indicates that, overlapping B and C rings onto those of digitoxigenin places the C20 carbonyl group of CMA close to the lactone carbonyl of digitoxigenin. CMA displayed a negative inotropic effect in contractility assays, being turned into positive when gly-cosidated at C3, possibly because in that way its ability to go into the cell and produce intracellular ef-fects that give rise to negative inotropy was reduced [82b] [90]. Spirolactones, with also C/D trans junction, display a positive inotropic activity as well [91].

Figure 21. Structure of spirolactone (XXII).

Figure 21. Structure of spirolactone (XXII).

On the other hand, recent studies carried out on 17β-(O-aminoalkyl)-oxime derivatives [92] showed that C/D trans analogues are significantly less potent than the corresponding C/D cis, although only slightly less potent than digoxin.

Because all the aforecited considerations, further research is needed in order to know the influence of C/D junction on digitalis-like activity.

Figure 22. Structure of some C/D trans derivatives.

Figure 22. Structure of some C/D trans derivatives.

3.3. Endogenous digitalis-like factors

The digitalis binding site is strongly conserved in a wide variety of organisms. This fact, together with its high sensitivity to low substrate concentrations and high specificity of interaction substrate-receptor, has given rise to the speculation of binding site being a receptor for an endogenous ligand, in the same way as the opioid receptor gave rise to idea and further discovery of endogenous opioids (en-dorphins and enkephalins). The ability of some animal species to synthesize cardiotonic steroids, e.g. toads, shows that vertebrates are able to synthesize steroids with digitalis structure.

Systematic search for such endogenous digitalis-like compounds was started by Szent-Györgyi in 1953 [93]. Since then, a large amount of EDLF (endogenous digitalis-like factors) have been described [94], having been their levels related to hypertension in animal and human models. Most recent results point to ouabain-like factors (OLF) [95], probably ouabain stereoisomers, isolated from hypothalamus [96] and human plasma [97], which are increased in certain types of hypertension [98].

Nevertheless, it has to be taken into account the possibility of an alteration of putative ELDF during its isolation, as well as it is possible that extracted compounds are precursors or metabolites eliciting a lower activity than the primary ligand (specially when they have been isolated from urine).

The most suitable test for classifying a compound as digitalis-like factor is its ability to bind to the digitalis receptor. So, hypothalamic inhibitor has shown negative results [99]. They seem to be several physiological Na⁺,K⁺-ATPase inhibitors, which have not necessarily a structural relationship with digitalis compounds [100].

However, some authors are uncertain about the existence of such an inhibitor and its role in hyper-tension [101], indicating that it is possible that the digitalis receptor is only an enzyme feature needed for its activity as cationic pump. Moreover, it is argued that isolated EDLF amounts are so small that they do not seem to be important for playing a physiological role, even when it can be considered the possibility of a located release or a selective affinity for one of Na⁺,K⁺-ATPase isoenzymes. Another argument is that cardiac glycosides do not produce hypertension at therapeutic doses. Some authors even speculate with certain C/D trans steroidal hormones as models for the chemical structure of en-dogenous digitalis [88]. They have also been proposed synthetic candidates as EDLF analogues [102].

Although there are many results supporting the existence of an endogenous Na⁺,K⁺-ATPase in-hibitor in physiological situations and in pathogenesis of hypertension and other diseases, many ques-tions remain to answer [103], so further research is needed to definitively stablish the existence of such an inhibitor.

4. Digitalis Analogues Bearing Aminoguanidine Moieties

4.1. Aminoguanidine moiety in drugs

Aminoguanidine is present in a wide variety of drugs eliciting very different activities. Its most characteristic feature is a high basicity, arising from its ability to delocalize the positive charge of pro-tonated derivatives on three guanidine Nitrogen atoms [104]. However, most of these compounds dis-playing a pharmacological activity are not strong bases, since any electron-withdrawing group bound to any of the Nitrogen atoms decreases dramatically basicity [105].

Among therapeutic groups including aminoguanidine derivatives appear adrenergic agonists (aganodine), antihypertensives (guanabenz, guanoxabenz, idralfidine), antiiflammatory and analgesic drugs (praxadine, apazone), antibacterials (ambazone, chloroazodine), antiparasitic drugs (robenidine), antineoplastic compounds (bisantrene, mitoguazone), antihistamine, antivirals, platelet inhibitors, anti-arrhytmic, inmunosuppressors... [106].

Aminoguanidine by itself also displays pharmacological activities, such as nitric oxide synthase in-hibition, histidine decarboxylase inhibition, catalase inhibition, diamine oxidase inhibition, glycation inhibition, S-adenosylmethionin decarboxylase inhibition, antioxidant activity, antiaterogenic effect, etc.

4.2. Aminoguanidines in digitalis compounds

Among guanylhydrazone derivatives displaying cardiotonic activity, the 3,20-bisguanylhydrazones of prednisone and prednisolone (PGB) have shown activity related to inotropic effect and Na⁺,K⁺-ATPase inhibition, even without bearing any of structural requirements that seem to be essential for classic cardiotonic steroids.

Figure 23. Structure of 3,20-guanylhydrazones of prednisone (XXIII) and prednisolone (XXIV).

Figure 23. Structure of 3,20-guanylhydrazones of prednisone (XXIII) and prednisolone (XXIV).

It was later observed that PGB differs from ouabain in not showing high and low affinity for differ-ent Na⁺,K⁺-ATPase isoenzymes, although it showed large differences related to potency if different species are considered, and it also inhibited ³H-ouabain binding [107]. This can be interpreted as either PGB binds to the digitalis receptor, but some parts of the enzyme not affecting to isoenzyme differen-tiation are involved, or it binds to a different site from digitalis receptor and then induces conforma-tional changes inhibiting ouabain binding.

The guanylhydrazone moiety is a strong base, so bisguanylhydrazones can act as biscationic com-pounds, having then the steroid a passive role as bridge between such cations. That possibility is sup-ported by the fact that changes in steroidal framework, related to stereochemistry or substituents at-tached to it, do not affect significatively its activity [108], while it does for digitalis glycosides. Bis-guanylhydrazone of 5β- or 5α-pregnanedione do not show a significative difference related to their activity. Enzyme sequentiation stablishes that the first and second extracellular segments bear several aminoacids carrying free carboxylic groups [109], which could interact with both guanylhydrazone ends of the compound.

However, replacement of digitoxigenin lactone ring by a guanylhydrazone group results in a com-pound with decreased digitalis-like activity (about 20% of parent compound), which showed a high sensitivity to modification in steroidal framework, indicating a possible binding to digitalis receptor [39]. If so, it could indicate that such receptor would include a moiety able to display a strong electro-static interaction with the guanylhydrazone cation. It has to be pointed out that the corresponding semicarbazone, similar but non-ionic, does not elicit any activity.

Figure 24. Structure of guanylhydrazone (XXV) and semicarbazone (XXVI) of digitoxigenin.

Figure 24. Structure of guanylhydrazone (XXV) and semicarbazone (XXVI) of digitoxigenin.

The activity of some monoguanylhydrazones at C17 of the pregnane system [110] (bearing C/D trans junction and variability in A/B junction -cis, trans, quasiplanar-) have also been studied. These compounds inhibit Na⁺,K⁺-ATPase, but they produce a transient increase of myocardial contractility followed by a strong negative inotropic effect.

So, related to the influence of C/D junction on the activity of 20-guanylhydrazones, the C/D cis junction confers specificity of action, whereas the guanylhydrazone side-chain confers on the steroid with a C/D trans junction the non-specific ability to inhibit various ATPases.

A wide variety of 3,20-bisguanylhydrazones display a positive inotropic effect on cardiac muscle. The variation of distance between both guanylhydrazone moieties by modification of attachment posi-tion at the steroidal framework does not alter, within certain limits, activity. Moreover, some bisgua-nylhydrazones of non-steroidal molecules also display cardiotonic activity; e.g., some diphenylbisgua-nylhydrazones and bisguanylhydrazones of bi- and tricyclic molecules are active [108].

On the other hand, certain monoguanylhydrazones at position C3 are also active. Specifically, 3-guanylhydrazone of 3-oxodigitoxigenin is active, while 3-guanylhydrazone of the corresponding 14-deoxy-14α-cardenolide is not.

Figure 25. Structure of 3-guanylhydrazones of 3-oxodigitoxigenin (XXVI) and 14-deoxy-14α-digitoxigenin (XXVII).

Figure 25. Structure of 3-guanylhydrazones of 3-oxodigitoxigenin (XXVI) and 14-deoxy-14α-digitoxigenin (XXVII).

Hydrogenation of guanylhydrazone moiety to yield bisguanidineamine derivatives does not involve a significative loss of activity. Nevertheless, replacement of such guanylhydrazone by other related structures, as hydrazine or S-methylthiosemicarbazone, gives inactive compounds.

Pharmacological assays carried out with 3,20-bisguanylhydrazones indicate that therapeutical in-dexes for these compounds are about two times better than those of natural cardiac glycosides are.

Among all these derivatives, the 3,20-bisguanylhydrazone of triamcinolone was selected to carry out pharmacological assays on human beings, for having shown best results in animal models. Never-theless, although results on cardiac function were satisfactory, increasing the outgoing cardiac flow in 32% without giving rise to bradycardia, its too short-lasting action made it a non-suitable substitute for digitalis.

Figure 26. Structure of 3,10-bisguanylhydrazone of triamcinolone (XXIX).

Figure 26. Structure of 3,10-bisguanylhydrazone of triamcinolone (XXIX).

In the same way, some studies on the guanylhydrazone of 2-digitoxigenone and several Schiff bases of 3-digitoxigenone have been carried out [111], using guinea-pig atria for inotropism assays and ³H-ouabain displacement as a measure of interaction with digitalis receptor. These compounds showed an intermediate activity between 3-digitoxigenone and ouabagenin. Although absolute potency of these derivatives varies depending on the considered species, the relative order of potency remains similar for all of them, indicating that they do not show any special preference for one of the isoenzymes in any of studied species.

Among the simplified, non-steroidal, digitalis glycosides analogues, the guanylhydrazone and bis-guanylhydrazone derivatives of C/D cis hydroindenes have shown the best results, significantly im-proving contractile force without affecting the cardiac rate [55].

The first and second Na⁺,K⁺-ATPase extracellular fragments are thought to be responsible of digi-talis binding to enzyme [112]. These fragments bear aminoacids carrying six free carboxylic acids, giving rise to different binding ways to ligands bearing different ionisation degrees. Because of that, it is interesting to study the influence of basicity of substituents attached to the digitalis steroidal frame-work. So, a large amount of cardenolide analogues, carrying a guanylhydrazone moiety or derivatives of it replacing the lactone ring, have been synthesized; QSAR studies on several of these compounds have been carried out [83f, 67] [113], from which the following conclusions have been extracted:

*

A good correlation is observed between receptor binding and Van der Waals volume and molar refractivity of different fragments including guanylhydrazone or derivatives of it at-tached to position C17β of digitalis compounds, as well as between such a binding and pK_a and PA (proton affinity) values. The two later parameters are the most important related to the affinity showed by these compounds.

*

The positive charge borne by guanylhydrazone (or derivatives) moiety is a very significant factor. The more protonated it is, the higher is the affinity. This one also rises with basicity.

*

Increasing of substituent bulk, especially when it is accompanied by an increase in the delocalization of π−electrons, produces a decreased affinity, indicating that either the re-ceptor volume for this fragment is small, or decreasing of substituent bulk allows stronger electrostatic interactions between enzyme carboxylate and the protonated hydrazone of li-gand.

Other studies indicate that alkyl or aryl derivatives of guanidinium act as competitive antagonists of Na⁺ ions at Na⁺,K⁺-ATPase, then blocking occlusion and active cation transport [114]. Among them, those bearing two guanidine groups are more effective (about two times) than those with just one, be-ing hydrophobicity of the molecule, rather than distance between guanidine groups, determinant for binding affinity. These derivatives act by stabilisation of E1 enzyme conformation, binding to it at the cytoplasmic side.

In summary, it seems likely that steroidal guanylhydrazones inhibit Na⁺,K⁺-ATPase by binding to the digitalis receptor. It is significant that derivatives bearing open chains, lactone ring isosters, are more effective than guanylhydrazone derivatives when bound to “classical” cardiotonic steroids, but they are inactive if bound to C/D trans steroids. This suggests that a close proximity to the receptor is more decisive for lactone isosters than for guanylhydrazone derivatives, so the main importance of C/D cis configuration in classic cardiotonic compounds rises from their ability to place the lactone ring very close to receptor. This argument is not valid for bisguanylhydrazones of 3,20-diketosteroids, which produce enzyme inhibition, but by a different mechanism than that involved in the binding of cardiac glycosides to the digitalis receptor [84].