The Na/K pump maintains the concentration gradients of Na+ and K+ ions across the surface membrane of animal cells , and a substantial amount of surface-expressed NaK, especially in cancer cells, has been suggested to function as non-canonical cardiotonic steroid-binding receptors  that activate multiple signaling cascades [8–11].
Multiplex gene expression analysis demonstrated a decade ago that various cardiotonic steroids inhibit prostate target genes . Several studies have demonstrated that various cardiotonic steroids are able to sensitize apoptosis-resistant cancer cells to pro-apoptotic stimuli [38–40] and directly induce apoptosis in lymphoma  and leukemia  cells. Cardiotonic steroids can also induce cancer cell death through Src- or MAPK-mediated inhibition of p53 expression , the inhibition of general protein synthesis , the inhibition of HIF-1a synthesis , sustained and irreversible autophagy [10, 46] and lysosomal membrane permeabilization .
Some cardiotonic steroids are also able to overcome the MDR phenotype. Indeed, while ouabain activates the MDR phenotype , 19-hydroxy-2″-oxovoruscharine  and the hellebrin / hellebrigenin pair (Figure 4) display similar and marked anticancer activity in chemosensitive versus MDR cancer cells.
Thus, cardiotonic steroids display pleotropic anticancer effects, and we recently reviewed all of the patents filed in this field, along with their potential applications in oncology .
We previously reported that the 19-hydroxy-2″-oxovoruscharin cardenolide induced marked decreases in [ATP]i in various cancer cell types, while much weaker effects were observed in normal cells [10, 12]. This observation was confirmed in the present study with respect to gamabufotalin-rhamnoside / gamabufotalin and hellebrin / hellebrigenin. We provide evidence in this study that the previously observed drop in intracellular ATP in tumor cells exposed to cardiotonic steroids is unlikely to arise from an alteration in the glycolytic flux. The extent of glucose-to-lactate conversion remained unaltered in HT29 colon cancer cells treated either with cardenolides or bufadienolides. Instead, we found a dramatic reduction in the oxygen consumption rate in cells treated with cardenolides and bufadienolides, reflecting a direct impact on the mitochondrial oxidative phosphorylation. While these effects on cell respiration were quite similar for each compound tested, it should be emphasized that bufadienolides (glycosylated or not) were used at a concentration in the low nanomolar range (10–30 nM), while cardenolides were used at concentrations ~3-fold (ouabain), 10-fold (digoxine) and 150-fold higher (ouabagenin, digoxigenin). This observation again supports the specific profile of bufadienolides and identifies tumor cell oxidative metabolism as a major target of these drugs.
The glycosylation patterns of cardiotonic steroids markedly influence their anticancer activity profiles. For example, Langenhan et al.  demonstrated that the glycorandomization of digitoxin leads to analogs that display significantly enhanced anticancer activity and tumor specificity when compared to digitoxin. In addition, changes in NaK expression dictate the growth regulatory effects of ouabain on cells . In the current study, the quantitative determination of α1, α2 and α3 subunits at the mRNA level in the various cancer cell lines that were used clearly indicated that all of the cell lines expressed the α1 subunit, though in a heterogeneous manner (Figure 2B), but did not express the α2 or α3 subunits (or they expressed the α3 subunit in very low amounts) (data not shown).
Table 2 includes previously published data  and shows that ouabain could display weak selectivity for the α1 subunit, while digoxin shows a 3- to 4-fold selectivity for α2/α3. The other cardenolides, uzarigenin and gitoxin, do not show isoform selectivity, although the glycosides have much higher affinities compared to the aglycones. For the bufadienolide gamabufotalin / gamabufotalin-rhamnoside pair, the rhamnoside appears to show marginal selectivity for α1 when compared to α2 but not to α3. There is also a clear difference in that the glycoside shows a much higher affinity than the aglycone (Table 2). By contrast, the hellebrin / hellebrigenin pair is anomalous in that the glycoside does not show a higher affinity for 3H-ouabain displacement compared to the aglycone, and in NaK inhibition assays the aglycone is even somewhat superior to the glycoside (Figure 3). Of all the cardiotonic steroids analyzed, the hellebrin / hellebrigenin pair displayed the highest selectivity for the NaK α1 subunit at an approximately 2 times higher affinity for the α1 than for the α2 and α3 subunits (Table 2).
The features of binding and inhibition of the human α1β1 complex of the different cardiac glycosides are reflected in the IC50 concentrations for growth inhibition. A positive correlation was shown between the IC50 for growth inhibition and Ki for inhibition of the purified α1β1 complex, and there is no evidence for a pattern typical of inhibition of α2 or α3 (Figure 2A). For example, digoxin shows a higher Ki and IC50 compared to ouabain. This is typical for α1, whereas digoxin should show a lower Ki and IC50 than ouabain in the cases of α2 and α3 subunits . The data for the bufadienolides are consistent with these conclusions, especially the conclusion that a lower Ki for α1β1 is associated with a lower IC50 for growth inhibition. The anomaly in binding and inhibition occurs with the hellebrin / hellebrigenin pair, a feature that may be related to the fact that the second sugar in hellebrin is glucose. In systematic studies, it has been found that glucose is not an optimal glycoside derivative for the binding or inhibition of renal NaK α1β1 [52, 53]. It has also been previously shown that the relative effects of glycoside and aglycone on Ki for inhibition of the renal NaK vary markedly in the function of different cardiac steroids . The parallel behavior between binding and inhibition of α1β1 versus cancer cell growth inhibition is also observed for the hellebrin / hellebrigenin pair. The findings that mouse cancer cells display high IC50 values for growth inhibition by different cardiac glycosides and that the growth effects of gamabufotalin-rhamnoside are in the μM range (compared to the nM range for the human cancer cells) also demonstrate the association of cancer cell growth inhibition and inhibition of the low affinity cardiac glycoside-binding rodent α1β1 complex. The gamabufotalin-rhamnoside is therefore a useful tool to check this association in rodent cells because even a “low affinity” effect is in the μM concentration range.