The pump, the exchanger, and the holy spirit: origins and 40-year evolution of ideas about the ouabain-Na⁺ pump endocrine system

Abstract

Two prescient 1953 publications set the stage for the elucidation of a novel endocrine system: Schatzmann’s report that cardiotonic steroids (CTSs) are all Na⁺ pump inhibitors, and Szent-Gyorgi’s suggestion that there is an endogenous “missing screw” in heart failure that CTSs like digoxin may replace. In 1977 I postulated that an endogenous Na⁺ pump inhibitor acts as a natriuretic hormone and simultaneously elevates blood pressure (BP) in salt-dependent hypertension. This hypothesis was based on the idea that excess renal salt retention promoted the secretion of a CTS-like hormone that inhibits renal Na⁺ pumps and salt reabsorption. The hormone also inhibits arterial Na⁺ pumps, elevates myocyte Na⁺ and promotes Na/Ca exchanger-mediated Ca²⁺ gain. This enhances vasoconstriction and arterial tone—the hallmark of hypertension. Here I describe how those ideas led to the discovery that the CTS-like hormone is endogenous ouabain (EO), a key factor in the pathogenesis of hypertension and heart failure. Seminal observations that underlie the still-emerging picture of the EO-Na⁺ pump endocrine system in the physiology and pathophysiology of multiple organ systems are summarized. Milestones include: 1) cloning the Na⁺ pump isoforms and physiological studies of mutated pumps in mice; 2) discovery that Na⁺ pumps are also EO-triggered signaling molecules; 3) demonstration that ouabain, but not digoxin, is hypertensinogenic; 4) elucidation of EO’s roles in kidney development and cardiovascular and renal physiology and pathophysiology; 5) discovery of “brain ouabain”, a component of a novel hypothalamic neuromodulatory pathway; and 6) finding that EO and its brain receptors modulate behavior and learning.

A new idea is the most quickly acting antigen known to science.

—Wilfred Trotter, FRS

Origins

In 1977, Professors Paul Horowicz and Paul De Weer (my Washington University colleague), the founding editor and associate editor, respectively, of AJP Cell Physiology, invited me to write a review article for their new journal. Paul De Weer, an Na⁺ pump expert, and I had discussed my ideas about the role of Na⁺ pumps and a postulated Na⁺ pump inhibitor/natriuretic hormone in the pathogenesis of hypertension. The present article is a personal account of the origin of those ideas and the developments in the field since my hypothesis was published in the journal’s first volume (29); Table 1 provides a timeline.

Table 1.

Milestones in elucidating the endogenous ouabain-Na⁺ pump endocrine system

Year	Key Investigator	Discovery	Reference
1785	Withering	Digitalis is a therapy for heart failure (HF)	(301)
1861	Kirk	Discovery that S. kombé seed juice (contains ouabain) is cardiotonic	(182, 204)
1888	Arnaud	Isolation (from ouabaio tree bark) and naming of ouabain	(8, 139)
1953	Schatzmann	Cardiotonic steroids (CTS) inhibit Na+ pumps	(255)
1953	Szent-Gyorgi	Digitalis may replace a missing endogenous compound	(274)
1957	Skou	Na+ pump is an Na-K-ATPase (NKA)	(264)
1967	Dahl	A humoral factor contributes to salt-dependent hypertension	(50)
1968	Reuter	Na/Ca exchanger (NCX) discovery;	(246)
1968/9	Baker/Blaustein	NCX discovery: “missing link” explains the cardiotonic effect of CTSs	(13, 14)
1973	Reuter/Blaustein	NCX is present in arteries and is postulated to modulate blood pressure (BP)	(245)
1975	Bricker	Natriuretic hormone (“third factor”) is a NKA inhibitor	(72)
1976	Haddy	Na+ pump-endogenous CTS-hypertension hypothesis	(113)
1977	Blaustein	Na+ pump-NCX-endogenous CTS-hypertension hypothesis	(29)
1978	Brody	Sustained hypertension requires the hypothalamic AV3V region	(39)
1982	Hamlyn/Blaustein	A circulating NKA inhibitor correlates with BP in humans	(123)
1983	Atkinson	Low (≤10–7 M) ouabain concentrations trigger cell proliferation	(10)
1984	Erdmann	Rat cardiac myocytes express ~90% low (α1) and ~10% high (α2) ouabain affinity Na+ pumps	(299)
1985–7	Lingrel/Shull	NKA is cloned; 3 catalytic (α) subunit isoforms (α1–α3) identified	(259, 260)
1989	Philipson	Ouabain upregulates NCX1 expression in cardiomyocytes	(287)
1989	Delva	Plasma ouabain-like compound (OLC) is elevated in pregnancy and more so in preeclampsia	(54)
1991	Hamlyn/Upjohn group/Blaustein	An endogenous CTS in humans identified as “ouabain” (EO)	(118)
1992	Gottlieb/Hamlyn	Plasma EO correlates inversely with cardiac index in HF patients	(104)
1993	Haddy/Pamnani/Hamlyn	Ouabain (chronic treatment) causes hypertension in rats	(314)
1993	Hamlyn/Manunta	Digoxin doesn’t induce hypertension, but antagonizes ouabain’s effect	(196)
1993/4	Hamlyn	Adrenal glands synthesize and secrete EO	(36, 162)
1994	EP and CE Gomez-Sanchez	Active anti-ouabain immunization attenuates DOCA-salt hypertension	(101)
1994	Leenen	“Brain ouabain” mediates salt-sensitive hypertension	(133)
1995	Hollenberg	Fab fragments (Digibind) attenuate DOCA + salt hypertension in rats	(156)
1995	Leenen	“Brain ouabain” mediates sympathetic hyperactivity in HF	(169)
1995	Rossi/Manunta/Hamlyn Hamilton	~50% of patients with essential hypertension or aldosterone-producing adenomas have elevated plasma EO that correlates with mean BP	(252)
1996–8	Askari/Xie	Ouabain binding to NKA activates protein kinase (PK) signaling cascades	(135, 153, 231)
1997–8	Defaye/Perrin	[14C]rhamnose is incorporated into EO in bovine adrenocortical cells	(217)
1997–8	Bianchi/Ferrari	Synthesis of rostafuroxin (PST2238), an ouabain antagonist but not NKA inhibitor, attenuates ouabain-induced hypertension	(79, 241)
1997–8	Blaustein/Juhaszova	α2 Na+ pumps and NCX1 colocalize at PLasmERosomes, and help regulate local [Ca2+]CYT and Ca2+ signaling	(33, 144, 145)
2000	Xie/Askari	Ouabain activates PK cascades independent of changes in [Na+] or [Ca2+]	(181)
2005	Lingrel	ACTH-hypertension requires α2 NKA ouabain binding sites and EO	(60)
2006	Hamlyn/Manunta/Hamilton	Plasma EO in humans is elevated by both high salt and salt depletion	(191)
2008	Lingrel/Lorenz	Digibind prevents natriuresis promoted by high ouabain affinity renal α1 NKA*	(184)
2009	Lichtstein	Anti-ouabain antibodies promote adrenal cortex, not medulla, hypertrophy	(213)
2009	Heiny/Lingrel	α2 NKA ouabain binding sites modulate skeletal muscle fatigue*	(231)
2010	Aperia	Ouabain rescues fetal kidney development in malnourished pregnant rats*	(174)
2010	Golovina	Arterial Ca2+ transporters are upregulated in ouabain- and salt-sensitive hypertension	(238, 322)
2010	Leenen	A novel hypothalamic slow neuromodulatory pathway [Ang II-AT1R-aldosterone-MR-ENaC-EO-(α2) Na+ pumps] sustains BP in hypertension*	(167)
2011	Lorenz/Lingrel	Pressure overload-induced cardiac hypertrophy and dysfunction requires ouabain-sensitive Na+ pumps and EO	(296)
2011	Van Huysse/Lingrel	NaCl-induced hypertension requires α2 NKA ouabain binding sites and EO	(286)
2011	Vorhees/Williams/Lingrel	Mice with ouabain-resistant α2 NKA exhibit specific behavioral abnormalities and learning impairment*	(254)
2012	Hamlyn	MS shows EO is elevated in pregnant rats; detection of two EO isomers in rat plasma	(138)
2013	Golovina/Hamlyn	Ouabain, not digoxin, upregulates arterial Ca2+ transporters and elevates BP via a Src-dependent mechanism; first evidence that an ion transporter can act as a biased hormone receptor	(323)
2013	Manunta/Gottlieb/Bianchi/Hamlyn	Preoperative plasma EO predicts acute kidney injury (AKI) in cardiac surgery patients; renal function is impaired in ouabain-hypertensive rats	(22)
2014	Ferrandi/Ferrari	Rostafuroxin attenuates ouabain-induced high BP and kidney injury	(78)
2014	Leenen/Hamlyn/Blaustein	A hypothalamic slow neuromodulatory pathway regulates plasma EO	(121)
2015	Lichtstein	In pregnant rats, anti-ouabain antibodies reduce fetal birth weight, and impair fetal kidney and liver development but increase cardiac growth*	(65)
2017	Scavone	α2 NKA silencing abolishes the anti-inflammatory and anti-apoptotic effects of nanomolar ouabain in cultured glial cells	(150)
2018	Blaustein	α2 NKA ouabain binding sites modulate basal BP*	Table 3

^*These data demonstrate that the EO-α2 Na⁺ pump endocrine system influences normal physiology by modulating, for example: 1) blood pressure; 2) skeletal muscle exercise endurance; 3) fetal renal, hepatic, and cardiac growth and development; 4) natriuresis (ouabain-sensitive α1 Na⁺ pumps); 5) long-term central sympathetic drive; and 6) behavior and learning.

Two key early influences were my experiences treating heart failure (HF) patients with cardiotonic steroids (CTSs, e.g., digoxin) as a medical student and intern, and my 18 months in Daniel Tosteson’s laboratory studying the Na⁺ pump, where I routinely used inhibition by the CTS ouabain (pronounced, “wä′bān”) (255) to measure pump activity. In 1958 I first met Tosteson who told me that “the Na-K-ATPase (NKA, or Na⁺ pump) has just been discovered (264). You’d better hurry up and work on it before everything about the pump is found out and there’s nothing left to do”–—so I went to work in his laboratory on a stipend from an NIH training grant. I’m still working on the pump 60 years later (43)!

My research and clinical experience with ouabain led me to ponder the question of how a pump inhibitor could enhance cardiac contraction (the positive inotropic effect) and thereby benefit patients with HF. The story of how I fortuitously found the answer by discovering the Na/Ca exchanger (NCX) (14), the “missing link,” while studying squid axon Na⁺ pumps as an NIH Special Fellow in Cambridge and Plymouth, England, is described elsewhere (28, 35). When we discovered the NCX, my coworker Peter Baker recommended I read Luttgau and Niedergerke’s paper on Na⁺-Ca²⁺ antagonism in cardiac muscle (189), which I did when a gale caused a brief hiatus in the squid supply. I immediately realized that Luttgau and Niedergerke had obtained indirect evidence of NCX in the heart. With the CTS conundrum on my mind I had an epiphany (28): Pump inhibition by CTS elevates the cytosolic Na⁺ concentration ([Na⁺]_CYT) and, via NCX, the cytosolic Ca²⁺ concentration ([Ca²⁺]_CYT). This in turn promotes cardiac contraction (i.e., the positive inotropic action). It was truly a “Eureka! moment.”

I spent a mini-sabbatical in Harald Reuter’s laboratory in Bern, Switzerland, in 1971 [Harald independently and simultaneously discovered NCX in 1966 (35, 246)]. We found that arterial smooth muscle (ASM) also expresses an NCX (35, 245) but, whereas NCX was readily accepted as a novel Ca²⁺ transport mechanism in mammalian cardiac muscle and squid nerve, the smooth muscle evidence encountered considerable skepticism (e.g., Refs. 211, 265, and see Refs. 28, 35). The original manuscript was rejected by Nature, Science, and Experientia and was eventually published in a nonrefereed journal (245) because Harald was invited to speak at a Royal Society conference. It wasn’t until NCX could be measured by molecular methods that the presence of NCX in smooth muscle became widely accepted (212, 266), but I can’t be too critical of skeptics because I, too, am often slow to accept novel ideas (30).

In the Discussion section of our 1973 article Harald and I speculated that ASM NCX might not only help to control vascular tone, but might also play a role in blood pressure (BP) regulation and the pathogenesis of hypertension (245). Although the original statement was undocumented, I continued to ponder this idea and started to search for evidence in the literature. This required me to spend a great deal of time in the library as there was no internet or PubMed. I also tried to reason from the viewpoint of comparative physiology. For example, I had studied muscles that contract tonically, namely, barnacle muscle (253) and ASM. Also, cardiac ventricular muscle is specialized so that it normally sustains [Ca²⁺]_CYT above the contraction threshold for many, many milliseconds so that sufficient force can develop to contract the entire ventricle during each systole. I was intrigued by the possibility that localized (sub-plasma membrane, sub-PM) elevation of [Na⁺]_CYT (i.e., [Na⁺]_sub-PM) might help to slow NCX-mediated Ca²⁺ extrusion and contribute to tonic contraction. In our 1973 paper (245) we postulated a “privileged communication” between the extracellular fluid and the sarcoplasmic reticulum (SR), involving the PM NCX (and, presumably, PM Na⁺ pumps and the tiny intervening space between the PM and SR). A similar idea, the ASM “superficial buffer-barrier”, was later proposed by Van Breemen (285). These views were bolstered by evidence for local Na⁺ gradients in heart (298) and ASM (235) and by discoveries about the subcellular localization of Na⁺ pumps and NCX in PM microdomains (see The Plot Thickens—A Lot, below). This concept differs from the more widely distributed “fuzzy space” (165) that has recently been questioned (90, 187).¹

The Pump, the Exchanger, and a Hypothesis

By the 1960s the role of dietary salt in the pathogenesis of essential hypertension (49) and its treatment with natriuretic agents (83) were well established, but the underlying mechanisms were unknown. In the late ’60s and early ’70s Guyton and his coworkers focused on the key role of the kidneys. They proposed that salt retention and plasma volume expansion led via arterial “autoregulation” to increased vasoconstriction and total peripheral vascular resistance (TPR) and thus to elevated BP (47, 108, 109). [Recall that BP = CO × TPR, where CO = cardiac output; if CO is constant, BP is directly proportional to TPR (18).] This “autoregulation theory” and the role of abnormal “pressure natriuresis” (115, 206) has dominated thinking on the mechanism of salt-dependent hypertension for the past half-century (47, 114, 207) although several groups have challenged this view (159, 209, 225). One critical problem is that “autoregulation” is a phenomenon, not a mechanism, as Nelson and colleagues showed by discovering specific mechanisms that regulate cerebral blood flow (183). Another problem is that the fundamental mechanisms that underlie the direct enhancement of systemic artery Ca²⁺ signaling, contractility, and tone in various forms of hypertension, including salt-dependent hypertension (178, 237, 238, 317), have been ignored. Further, brain mechanisms play a key role in hypertension (2, 38, 39, 143, 166, 197) and, when these central mechanisms are blocked, salt-induced hypertension can be averted (100, 286, 293) without evidence that they affect the renal and hemodynamic alterations postulated to trigger “autoregulation.”

I jumped into this abyss in 1977 with my AJP Cell Physiology article (29). One of its core concepts was the idea that the functional-coupling of Na⁺ pumps and NCX provides a unique connection between Na⁺ metabolism and ASM cell Ca²⁺ regulation and, thus, a link to arterial contraction, TPR, and BP. While scouring the literature for relevant clinical data, I found a number of reports that raised the possibility that a natriuretic hormone, or “third factor,” might be involved (37, 45, 54). Some studies even suggested that this factor was an “active sodium transport” (i.e., Na⁺ pump) inhibitor (72, 80) [the senior author, Neal Bricker, had taught me renal pathophysiology and clinical nephrology when I was a medical student!]. I leaped at the idea and wove a seemingly logical (but rather naïve) story about how salt retention, by an unknown mechanism, triggers the secretion of an endogenous Na⁺ pump inhibitor (i.e., a ouabain-like compound or “OLC”). I was unaware that Szent-Gyorgi had postulated an endogenous CTS 24 years earlier (274), and that Dahl had already shown that a “humoral agent” was instrumental in “salt” hypertension in rats (50). The OLC would be expected to: 1) inhibit Na⁺ pumps in the kidneys, thereby promoting natriuresis to compensate for the salt retention; and 2) inhibit Na⁺ pumps in arterial myocytes, thereby promoting Ca²⁺ entry via NCX [see Fig. 1, taken from the original article (29)]. The myocyte Ca²⁺ retention should, of course, increase arterial tone, TPR and BP. In fact Mason and Braunwald had shown that local ouabain infusion increases forearm vascular resistance in humans (198). Numerous novel findings have demonstrated that the model in Fig. 1 is overly simplistic. Nevertheless, the concepts of functional Na⁺ pump-NCX coupling and an OLC with vasotonic (and cardiotonic) actions have now been verified, and many of the details have been filled in.

Fig. 1.

Block diagram (1977 version) of Ca²⁺ regulation in arterial smooth muscle myocytes used to explain the hypothesis that Na⁺ pumps, Na⁺/Ca²⁺ exchangers (NCX) and an endogenous ouabain-like compound (OLC) help regulate myocyte Ca²⁺ and arterial tone (29). Some key Na⁺ and Ca²⁺ transport pathways are shown: 1) ouabain-sensitive Na⁺ pump; 2) NCX operating in the Ca²⁺ extrusion mode, although we now know that NCX usually operates in the Ca²⁺ entry mode in arteries with tone (31, 319); 3) sarcoplasmic reticulum (SR) Ca²⁺ pump (SERCA). Most of the Na⁺ and Ca²⁺ entry is through plasma membrane voltage-gated channels (k₁ and k₂, respectively) and Ca²⁺ is released from the SR through ryanodine receptors or inositol trisphosphate receptors (k₃). This diagram anteceded the discovery of Na⁺ pump isoforms and PLasmERosomes. [Modified from Blaustein (29).]

My 1977 paper (29) was published while I was abroad. Upon my return, my laboratory colleagues (at Washington University Medical School in St. Louis, MO) all stopped what they were doing to watch me enter my office at the back of the laboratory. They wanted to see my astonishment when I opened the door and found my desk buried under a mound of several thousand reprint request cards (the article has been cited >1,700 times). Interest in those ideas was soon reflected in de Wardener and MacGregor’s review, “Dahl’s hypothesis that a saluretic substance (plays a) role in essential hypertension” (55), and in Kaplan’s reference to my “biochemical hypothesis” (147).

The Hunt for the OLC (the Holy Spirit)

While my AJP Cell article was in press, Haddy and Overbeck (113) also proposed that an OLC is a critical player in the pathogenesis of salt-sensitive hypertension. They, however, suggested that the critical effects of pump inhibition were reduced electrogenic Na⁺ pump current and consequent myocyte depolarization (112), but this mechanism cannot work in the steady state. Partial pump block will cause a transient decline in Na⁺ transport (and pump current), but when [Na⁺]_CYT rises, pumped Na⁺ efflux must recover and must equal Na⁺ influx in the steady state to avoid continued net gain of Na⁺ (24). In other words, chronically, partially inhibited pumps operate with a higher [Na⁺]_CYT but unchanged pump current.

These proposals about the role of an OLC in hypertension spurred interest in purifying and identifying this compound. One consequence was that a postdoctoral applicant, John Hamlyn, arrived at my office in 1981, shortly after I moved to the University of Maryland School of Medicine in Baltimore. He told me he wanted to purify this new hormone. I thought to myself, “Sure, you and who else?” Nevertheless, he had some good ideas so I agreed to take him on and let him try, perhaps swayed a bit by the fact that he grew up in the ’60s in Plymouth, UK, the site of our NCX discovery. Little did I—or any of us—realize how terribly difficult this quest for the holy spirit would be.

Most opportunely, shortly after John’s arrival I was introduced to Avinoam Kowarski and Bruce Hamilton, two clinical endocrinology colleagues who were studying hypertension and had collected plasmas from a number of patients and controls. They agreed to collaborate and provided us with de-identified plasmas, and John devised a direct but complex biochemical assay to measure Na⁺-K⁺-ATPase (NKA) inhibitory activity in the plasmas. On several occasions I hovered over John’s shoulder as he performed the assay, and I marveled at his skill; it left me with no doubt about the quality of his data. My mentors, Howard Schneiderman, Dan Tosteson, David Goldman and Alan Hodgkin (26) had imbued me with the need to perform meticulous experiments so there could be no question about their validity. John exemplified this approach.

When all the results had been collected, John and I sat in my office and plotted the BP as a function of NKA inhibitory activity. The results were almost too good to believe: although the signals were generally small and there was a fair amount of scatter, there appeared to be a virtually linear correlation between BP and NKA inhibitory activity (Fig. 2)! The paper was published in Nature (123), accompanied by a “News and Views” commentary (96). Our article whetted everyone’s appetite and intensified the search for the endogenous OLC. In fact, three major pharmaceutical companies contacted me about the possibility of a collaboration. Scientists at the Upjohn Company (Kalamazoo, MI), led by Jerry Zinn and Donald DuCharme, proposed a joint effort, and I decided to work with them (an excellent choice in retrospect, as our Upjohn colleagues were wonderful people, and excellent scientists and colleagues). The decision to work with Upjohn was cemented when NIH (National Heart, Lung, and Blood Institute, NHLBI) turned down our two grant applications to purify the OLC. The rationale was (to paraphrase the Summary Statements) “Everybody else is doing this so why do we need another group in the field?” and “Everyone who has tried this has failed, so why waste money on anyone else?”

Fig. 2.

Mean arterial pressure (MAP) is directly correlated with plasma Na⁺-K⁺-ATPase (NKA) inhibitory activity. Data are from 20 normotensive (green and blue circles) and 26 hypertensive (red circles) subjects. Green circles indicate discrete single plasma samples; red and blue circles indicate plasma samples drawn slowly over 6 hours (“integrated samples”). The regression line is a least-squares fit to the data; the correlation between MAP and % NKA inhibition is highly significant (r = 0.73; P < 0.0005). Details are given in Ref. 123. [Modified from Hamlyn et al. (123) with permission from Macmillan Publishers Ltd., copyright 1982.]

During the next decade, Upjohn funded and participated in our research as follows: I was tasked with procuring, literally, tons of plasma. A good source was blood banks at our university and other nearby large medical institutions that performed plasmapheresis. John worked out the initial steps of extraction and purification on large, normal phase chromatographic columns, but the latter proved unsatisfactory. Accordingly, subsequent work was done on smaller, reverse-phase HPLC (high-pressure liquid chromatography) columns that gave far superior resolution and reliability. At every step, the eluates were carefully screened for NKA inhibitory activity, and the most promising, high-activity samples were saved for further analysis. Once false-positive had been excluded and the location of relevant biological activities was understood, the isolation steps worked out in Baltimore were refined and scaled up for batch processing at Upjohn.

The Competition Heats Up

In 1987 I attended a meeting in Switzerland and stopped off to visit Reuter in Bern. Harald asked me if I had seen the latest issue of Biochemical and Biophysical Research Communications. I hadn’t, so he escorted me to the library, retrieved the issue from the rack, and opened it to an article by Tamura and Inagami (278). I skimmed through the report in which they claimed to have purified an endogenous NKA inhibitor of molecular mass 336 Da, with “striking similarity to… ouabain,” and was alarmed. I immediately phoned Hamlyn in Baltimore. He hadn’t seen the article either, so I started to read it to him. Before I’d gotten very far, he interrupted me: “They missed it!” John, who has an exceptional understanding of steroid chemistry, recognized that they, too, had seen what appeared to be the same bioactive polar material eluting from their columns, but that they had been led astray by a co-eluting contaminant. When I returned to Baltimore, I sought John out. He wasn’t in his office, but an 8½ × 11 inch photocopy of an Inagami portrait was hanging from each of his bookcases—and from each of the eye-level shelves in his laboratory. As John explained, “I want to know who my competition is.”

In fact, there was no shortage of competition. In 1988, a prominent Tokyo group, Goto and Yamada, described a bioactive material in dog plasma that “closely resembles ouabain” (103). Their extracts, while impure, almost certainly contained OLC, but they, too, were thrown off the scent by a contaminant that had an estimated mass of only 343 Da; ouabain’s mass is 585 Da. The following year, another very active group, Hallaq and Haupert in Boston, described the ouabain-like positive inotropic effects of a partially purified material from bovine hypothalamus (116). Clearly, the key to winning the “race” was to put sufficient pure material in the hands of an analytical chemist who made a living by identifying small molecules.

A New Twist—and an Astonishing Answer

By 1989 we had 0.8 μg of a very highly enriched OLC with an inhibitory action on NKA indistinguishable from that of plant ouabain (120). At this point we were close, but not yet there. The OLC was purified by a factor of ~10 billion but wasn’t completely pure, and the 0.8 μg were insufficient for the best mass spectrometers (MS) at the time. We needed several micrograms of pure material to be sure not to get caught with a contaminant. Fred Mandel at Upjohn suggested employing the NKA, itself, as an affinity ligand. A slurry of NKA was used to bind up the OLC; after washing to remove unbound contaminants, the OLC was eluted from the NKA. Two more rounds of HPLC were sufficient to obtain pure OLC, which was submitted to the excellent Upjohn analytical chemistry facility for MS.

Don Ducharme phoned a couple of weeks later to tell me that they had identified the structure: “Are you sitting down? You’d better sit down. You won’t believe this. It’s ouabain!” It took him some time to convince me that he wasn’t joking (“Okay, Don, now really, what’s the structure?”). I immediately went to John’s office to inform him (he was then an assistant professor). John was just as incredulous; he phoned Don back for confirmation. The chemistry and mechanism of action had indicated that the endogenous molecule was very “ouabain-like”; nevertheless, we were all stunned that it was “indistinguishable from ouabain” (Fig. 3A) (118). This manuscript, too, had a rough time. It was submitted to Nature, and although one referee offered to write a “News and Views” commentary, another had “a gut feeling that (endogenous ouabain, EO) is an artifact.” Nature therefore rejected the manuscript so I consulted Professor Joseph Hoffman, a respected leader in the Na⁺ pump field, who communicated it to Proceedings of the National Academy of Sciences (PNAS; 118). Our report merited an editorial, “Welcome to ouabain,” in The Lancet (68); the original presentation to the American Heart Association High Blood Pressure Council was reported in The New York Times (7). John and our Upjohn colleagues also published full details of the EO purification and mass spectroscopy (119, 188, 199). The PNAS article and the earlier Nature (123) and AJP Cell (29) publications were all included among “The Top 100 Cited Scientific Reports… on Hypertension Research“ published in the past century (221). Regrettably, a change in leadership at Upjohn led the company to abandon our project shortly after the discovery of EO. Our Upjohn collaborators, Don DuCharme, Jim Ludens, Fred Mandel, and Rod Mathews, were dismayed because they described the EO study as “the most exciting project” in their careers at the company.

Fig. 3.

Molecular structures of ouabain (A), digoxin (B), marinobufagenin (C), and rostafuroxin (D). The sugars in ouabain and digoxin are indicated by “Rs.” [Reproduced from Schoner and Scheiner-Bobis (257).]

The skepticism voiced by some colleagues when we announced the result was understandable (a few thought I’d gone off the deep end). After all, ouabain had been used in Africa as an arrow poison for hunting and warfare since the 3rd century (215). Why would anyone have expected ouabain to be a normal human hormone? It seems weird, but doesn’t this simply reflect our prejudice because we learned that ouabain is a poison more than a century before discovering that it is a hormone?

Confirmation—Despite a Few Vocal Naysayers

Within a decade six groups on three continents had verified by MS and NMR spectroscopy that human, bovine, and rodent materials (plasma, urine, adrenals, and hypothalamic tissue) contain EO (Table 2A); the highest concentrations were found in the adrenals (118). A few outspoken investigators were unable to detect EO in human plasma and claimed that EO did not exist [Table 2C; (172)]. Regrettably, this distracted many researchers and clinicians outside the immediate field who could not evaluate or, more often, were unaware of the actual data. In the face of so much positive analytical data, including published MS and NMR spectra of EO (Table 2A), very rigorous experiments are needed to prove its absence. Could so many fine investigators have found the same artifact in specimens from different species? A more likely explanation is that the authors listed in Table 2C may have not detected EO because they did not replicate the original protocols; e.g., one used homogenized placenta (126) rather than plasma, and the others (12, 59, 173) relied on immunoassays. In contrast, many of those in Table 2A used the native receptor: i.e., they bioassayed eluates for NKA binding and/or inhibitory activity at each stage of HPLC separation (e.g., refs. 118, 120, 188, 232, 256, 276).

Table 2.

Analytic detection of EO in mammals

Senior/Key Author(s) (Location)	Year	Source	Material	Analytical Method	Reference
A. Positive detection of EO in mammals (published spectra/MS and NMR data)
1. Hamlyn/Blaustein/DuCharme (Baltimore, MD/Kalamazoo, MI)	1991	Human	Plasma	MS	(118)
2. Mathews/DuCharme/Hamlyn (Kalamazoo/Baltimore)	1991	Human	Plasma	MS	(199)
3. Inagami/Tamura (Nashville, TN)	1994	Rat	Urine	MS, NMR	(276)
4. Doris (Lubbock, TX)	1994	Bovine	Adrenal	——*	(59)
5. Defaye/Perrin (Grenoble, France)	1997	Bovine	Adrenal	MS	(232)
6. Schoner/Schneider (Giessen, Germany)	1998	Bovine	Adrenal	MS, NMR	(256)
7. Nakanishi/Kawamura/Haupert (New York, NY/Boston, MA)	1999	Bovine	Hypothalamus	NMR	(149)
8. Takahashi/Komiyama (Tokyo, Japan)	2000	Human	Plasma	MS	(155)
9. Manunta/Hamlyn/Pitzalis (Milan, Italy/Baltimore)	2005	Human	Plasma	MS	(234)
10. Hamlyn/Manunta (Baltimore/Milan)	2006	Human	Plasma	MS	(191)
11. Hamlyn/Jacobs (Baltimore/Milan)	2012	Rat	Plasma	MS	(138)
12. Hamlyn/Leenen/Blaustein (Baltimore/Ottawa, Canada)	2014	Rat	Plasma	MS	(121)
13. Hamlyn/Blaustein (Baltimore)	2016	Rat/Mouse	Plasma	MS	(117)
B. Identification of EO isomers (published spectra)
1. Hamlyn/Jacobs (Baltimore)	2012	Rat	Plasma	MS	(138)
2. Hamlyn/Leenen (Baltimore/Ottawa)	2014	Rat	Plasma	MS	(121)
C. Did not identify EO in human samples
1. Doris (Lubbock, TX)	1994	Human	Plasma	RIA†	(59)
2. Nicholls/Lewis (Christchurch, New Zealand)	1994	Human	Plasma	ELISA†	(173)
3. Hilton (London, UK)	1996	Human	Placenta	MS	(126)
4. Vogeser/Baecher (Munich)	2014	Human	Plasma	MS‡	(12, 289)

^*Identified by elution pattern.

^†Not a definitive analytical method.

^‡The MS trace signal is broken at ∼5.0 min in the original publication (Fig. 2, upper spectrum) (12), but an ion current at that position is shown in spectra from two patients later published in a letter to the editor (289) (and see Ref. 117 online supplement Fig. S7). That ion current had the same mass/charge ratio as ouabain but eluted 0.3 min ahead of ouabain.

Ouabain, also called “g-strophanthin,” is an unusual steroid. It is highly polar and therefore can be challenging to separate by column chromatography, especially when it is present in a complex matrix such as plasma. Hamlyn recently identified two ouabain isomers in rodent plasma that are absent from commercial (plant) ouabain, one even more polar than ouabain [Table 2B; (121, 138)]. This is additional evidence that EO is of animal origin. In fact, one group that was unable to detect EO in human plasma (12) did observe an ion currrent peak in their MS-MS product ion spectra that eluted shortly ahead of ouabain ((289) and see footnote to Table 2 line C4). That ion current peak contained a product ion with the same mass/charge ratio as ouabain, and we have therefore suggested that it might correspond to the more polar of two ouabain isomers (27, 117) previously observed in rat plasma by multistage MS (121, 138).

Is EO the Only Endogenous CTS in Mammals?

The OLC has also been called a “digitalis-like compound” (CTSs obtain from plants of the genus, Digitalis) or DLC. As discussed in the next section, however, digitalis-like, very hydrophobic compounds (e.g., digoxin; Fig. 3B) do not mimic the unusually polar ouabain-like CTSs (from flora of the genera Strophanthus and Acokanthera) in some of their key actions. To date, there is negligible evidence that DLCs are endogenous in mammals. Plant ouabain comes from the bark of the Somali “ouabaio” tree, Acokanthera ouabaio (alt., A. schimperi), hence the name (“waabaayo” is “arrow poison” in Somali), and from the seeds of the flowering woody vines, Strophanthus gratus and Strophanthus kombé. The cardiotonic, pulse-slowing effect of juice from S. kombé seeds, which contains ouabain, was discovered by the British botanist, John Kirk, in 1861 (182, 204). Ouabain was first identified chemically and named by the French chemist, Arnaud, in 1888 (8, 139).

Another related CTS, marinobufagenin (MBG; Fig. 3C), a bufadienolide first identified in amphibia (15), has also been linked to hypertension (73, 74), and elevated levels of MBG and another bufadienolide, telocinobufagin, have been found in patients in renal failure (154). The linkage to hypertension is based mainly on immunologic (versus analytic) identification and on immuno-neutralization by “Digibind” or “DigiFab,” commercial anti-digoxin Fab fragments with much higher affinity for ouabain and digoxin than for MBG (239, 240). Also, MBG binds preferentially to α1 Na⁺ pumps (74, 75, 295). Here I focus on α2 Na⁺ pumps and EO (see The Plot Thickens – a Lot for a discussion of Na⁺ pump α-subunit isoforms). Many of the immuno-neutralization studies mentioned below indicate that the main endogenous CTS binding site ligand is an OLC. Nevertheless, there are likely to be other related endogenous Na⁺ pump ligands such as the two ouabain isomers mentioned above (121, 138).

An oft-stated concern about EO is that the biosynthetic pathway is unknown (172), but this is not entirely correct. The adrenals are an important source of EO because they have high concentrations (118, 256, 277). Adrenalectomy markedly reduces plasma EO in rats (118), and patients with adrenal insufficiency have very low circulating EO levels (268). Also, prolonged immuno-neutralization of EO with custom anti-ouabain antibodies increases adrenal cortical, but not medullary, weight by about 40%, suggesting feedback enhancement of EO production (213). There are several reports of hypertensive patients with adrenocortical tumors associated with excessive production of EO or endogenous “DLC” (190, 252, 258). Further, plasma aldosterone and EO are co-elevated in patients with essential hypertension (279). Also, excision of the adrenocortical tumors (some may even be “ouabainomas”; e.g., Ref. 190) markedly reduced plasma EO or DLC and attenuated the hypertension (190, 252, 258). (In this case, as in many other reports, “DLC” is simply used as the generic term for an endogenous substance that binds with high affinity to anti-digoxin fab fragments, rather than a reference to an entity chemically identified as “digitalis-like.”]

Primary cultured bovine adrenocortical cells exhibit net synthesis of EO (162, 232), as do the adrenals of awake dogs (36). Perrin and colleagues (232) showed that the biosynthesis of ouabain, like aldosterone, begins with cholesterol side chain cleavage to form pregnenolone, which is then converted to progesterone. The more distal steps where the pathway diverges from that of aldosterone synthesis, including the addition of the unsaturated five-member lactone ring and the hexose, l-rhamnose (ouabain is a rhamnoside; Fig. 3A), still need to be clarified. The observation that addition of rhamnose to the medium more than doubled the synthesis of ouabain, but not cortisol, by primary cultured bovine adrenocortical cells implies that coupling the sugar to the steroid is one of the late steps in EO biosynthesis (232). Regrettably, the research needed to identify the distal EO biosynthetic steps is unlikely to be performed in today’s funding climate, but that is surely not evidence that EO doesn’t exist. It is fascinating to contemplate the likelihood that the biosynthetic pathways for animal and plant ouabain evolved independently, and to consider how similar/different the pathways are.

The Proof of the Pudding Is in the Tasting

If EO is the Na⁺ pump endogenous ligand, and if it does promote the elevation of BP, as Haddy and Overbeck (113) and I (29) had postulated, there is a direct test: infuse ouabain into animals to elevate the plasma level and determine if it raises BP. Accordingly, Hamlyn, Haddy, and their colleagues showed that simply administering ouabain to normal outbred rats for 6–8 weeks causes hypertension (314). This landmark observation has now been replicated in numerous laboratories (reviewed in Refs. 92, 117). Because ouabain does not always raise BP in outbred rats, a few investigators have mistakenly concluded that the ouabain-hypertension link is unimportant (92). This view ignores the evidence that the susceptibility to ouabain-induced hypertension is genetically determined. As Hamlyn and colleagues showed, by inbreeding the least and the most ouabain-sensitive rats, one can generate a highly ouabain-resistant strain and one that is exquisitely ouabain-sensitive within just three generations (6).

Based on the knowledge that all CTSs are pump inhibitors (255) with a cardiotonic action (291), one might have anticipated that the influence of ouabain on BP depends exclusively on its inhibition of Na⁺ pumps and consequent NCX-mediated elevation of cytosolic Ca²⁺ [as postulated (29) and see Ref. 14]. In this case, all CTSs, which are also vasotonic (198, 283), would be expected to have a similar hypertensinogenic effect. When Hamlyn and his post-doc, Paolo Manunta, infused digoxin and related compounds into rats for 5 weeks, however, the BP did not rise (193, 196), even though these compounds, too, are vasotonic (267, 283). In fact, when digoxin was infused into ouabain-hypertensive rats, the BP fell; i.e., digoxin behaved like an ouabain antagonist (193, 196). Despite these novel findings, the original report (193) was initially rejected for publication as a full-length article. It was only published 7 years later, after another report had verified the antagonistic effect of digoxin on ouabain-induced and salt-dependent hypertension (131). Subsequently, Lichtstein and colleagues corroborated the ouabain-digoxin antagonism. They showed that in anesthetized guinea pigs given a toxic dose of digoxin, and in isolated cardiac muscle treated with toxic concentrations of digoxin or bufalin, low-dose ouabain attenuated the cardiotoxicity (214). Also, we later observed ouabain-digoxin antagonism in isolated arteries and in primary cultured neurons (267). In retrospect, these results should have been anticipated because digoxin has been used clinically for more than two centuries (301) without any evidence that it is hypertensinogenic. Some of the clinically beneficial effects of Digitalis steroids might even be explained by (unrecognized) ouabain-digoxin antagonism.

The implications of these straightforward observations are profound! First, because both digoxin and ouabain similarly inhibit the Na⁺ pump (267, 310), this inhibition alone cannot account for ouabain’s hypertensinogenic effect, but an explanation wasn’t recognized until a decade later (see The NKA Is More Than a Cation Pump, below). Second, these data demonstrate that the CTS-Na⁺ pump interactions for digoxin and ouabain are not identical; they support the view that the endogenous ligand is very “ouabain-like” and not “digoxin-like.” Third, the results suggest that digoxin or a digoxin-like molecule might be a therapeutically useful antihypertensive. In fact, digoxin has been used to treat hypertension (1). (During a luncheon following a seminar I gave at Johns Hopkins University three years ago, a post-doc from India told me that his mother, who had hypertension, had been treated with digoxin. “What happened?” I asked. “Her blood pressure came down” was his reply.) This concept was quickly recognized by others and led Bianchi and colleagues to develop rostafuroxin (PST2238; Fig. 3D), a derivative of digitoxigenin (241) that displaces [³H]ouabain from NKA at concentrations that don’t inhibit the ATPase (79). Rostafuroxin antagonizes ouabain-induced hypertension in rats (79) but may have too low an affinity for the pump (267) to be useful in unselected patients with essential hypertension (269). This drug, however, effectively lowers BP in hypertensive patients who carry specific isomorphisms (161). Nevertheless, a more effective EO antagonist is obviously needed.

In a prescient article, Manunta, Hamilton, and Hamlyn (192) identified some of the structural features of ouabain that are required for its hypertensinogenic effect. They showed that the aglycone ouabagenin is as effective as ouabain in inducing hypertension in the rat. Surprisingly, modifications of the normally unsaturated lactone ring—i.e., saturation, conformational constraint, or opening of the ring—strengthen the hypertensinogenic effect even though these alterations greatly weaken the ATPase inhibitory activity by increasing the EC₅₀ almost 100-fold. This, too, indicates that the hypertensinogenic effect is not tightly linked to the Na⁺ pump inhibitory activity but instead suggests that the effect is more closely related to EO-activated signaling mechanisms (see The NKA Is More Than a Cation Pump).

An alternative test of the EO-hypertension hypothesis is to neutralize the circulating EO in hypertensive models (or humans). Indeed, immuno-neutralization of EO with Digibind significantly lowers BP in rats with DOCA-salt (156) or reduced renal mass (146) hypertension and prevents ACTH-induced BP elevation in mice (60). By the late 1990s the link between ouabain/EO and hypertension was already well documented in humans (195, 233, 252) and rodents—despite a few awkward and as yet unexplained observations such as why only about half of essential hypertensives and patients with aldosterone-producing adenomas have substantially elevated EO levels (252). Nonetheless, plasma EO correlates directly with BP in essential hypertensives and their offspring (194, 233).

The Plot Thickens—a Lot (and One Type of Na⁺ Pump Isn’t Enough)

In 1985 Lingrel and Shull cloned the Na⁺ pump catalytic (α) subunit which contains the transport machinery and ouabain binding site (259). By the early ’90s they had cloned all 4 α-subunit isoforms (α1–α4) and the small, glycosylated mammalian β-subunit isoforms (β1 and β2) (23, 180). The pumps are αβ dimers, and β is a chaperone that is required for maturation, intracellular routing, and activity of the pump, including ouabain binding (124). The pump complex may also include a small regulatory “FXYD family” protein, sometimes called the γ-subunit (91). Importantly, all cells express pumps with an α1-subunit, usually the predominant pumps [80–90% of total pumps in the heart (19, 299)]. Most cells also express pumps with an α2- or α3-subunit [the remainder (19, 141, 299)]. In addition to α1 pumps, for example, all types of muscle (cardiac, skeletal, and smooth), glial cells, and hippocampal neurons also express α2 Na⁺ pumps, while most neurons express α3 pumps (179, 203). The α1 pumps are “housekeepers” that maintain the low [Na⁺]_CYT in bulk cytosol and mediate net Na⁺ transport across epithelia (201). So what do the α2 and α3 Na⁺ pumps do?

While studying α-subunit distribution in ASM cells and neurons, we serendipitously discovered that the pumps with α2- or α3-subunits were confined to PM microdomains that were closely apposed to “junctional” SR or endoplasmic reticulum (S/ER) (144, 145). The NCX colocalized with α2 or α3 in these microdomains at PM-S/ER junctions, which we have called “PLasmERosomes” (33). In contrast, α1 pumps were more ubiquitously distributed in the PM (144). This was the first experimental support for our conjecture that there is preferential communication between the PM and S/ER at the PLasmERosomes (245). Further, the unusually low affinity of rodent α1 for CTSs (218) implicated α2 and/or α3 pumps and NCX in the cellular action of low-dose CTSs (9, 33).

Does EO Directly Affect Cardiovascular Function?

Crucially, Lingrel’s Na⁺ pump-cloning studies opened the door to genetic alteration of the pumps to facilitate the study of α isoform function. Mutations included the generation of mice with ouabain-resistant (R) α2 pumps [α1^R/Rα2^R/R mice (48, 61)] and R-α2 mice with ouabain-sensitive (S) α1 pumps (α1^S/Sα2^R/R, or “SWAP” mice (63)), to determine the role of the ouabain binding site (179). The α1^R/Rα2^R/R mice, with two amino acids normally found only in ouabain-resistant rodent α1 substituted into the α2 ouabain binding site (48), apparently have normal Na⁺-K⁺-ATPase activity (61), but they exhibit a baseline cardiovascular phenotype. Telemetry studies show that baseline mean BP (MBP) is 4–10 mm Hg higher in awake α1^R/Rα2^R/R than wild-type (WT) α1^R/Rα2^S/S mice (Table 3). Also, pregnant α1^R/Rα2^R/R dams have about 10 mm Hg lower systolic BP than do pregnant WT controls (228). Thus, the α2 ouabain binding site and, presumably, its endogenous ligand play a role in normal basal BP regulation.

Table 3.

Baseline MBP in awake WT (α1^R/Rα2^S/S) mice and ouabain-resistant α2 (α1^R/Rα2^R/R) mice measured by telemetry; ΔMBP is the difference between the two

Study	α1R/Rα2S/S (WT) MBP, mm Hg	α1R/Rα2R/R MBP, mm Hg	ΔMBP, mm Hg	Reference
1.*	104.0 ± 2.6 (11)†	108.2 ± 3.2 (10)	4.2	(185)
2.‡	115 ± 3 (8)	125 ± 5 (10)	10	(286)
3a.§	121 ± 8 (4)	128 ± 3 (5)	7	(168)
3b.§	115 ± 6 (5)	125 ± 4 (7)	10	(168)
4.§	97.8 ± 0.8 (7)	108.2 ± 0.8 (7)	10.4	(43)

Values are means ± SE. WT, wild type; MBP, mean blood pressure.

^*Mice were “paired for age, weight and sex.”

^†Numbers of mice shown in parentheses.

^‡Mouse sex not specified. §Male mice; 3a = Experiment no. 1 and 3b = Experiment no. 2. WT, wild type; MBP, mean blood pressure.

The α1^R/Rα2^R/R mice are resistant to hypertension induced by ouabain, ACTH, elevated intracerebroventricular (icv) NaCl, and high dietary NaCl (60, 62, 168, 185, 286). Even angiotensin (Ang) II-high salt hypertension, a common hypertension model, apparently requires S-α2 pumps for full expression (25, 187a). Trans-aortic constriction (TAC)-induced cardiac hypertrophy also is delayed/attenuated in these mice despite the TAC-induced BP elevation (296). Conversely, SWAP (α1^S/Sα2^R/R) mice, which have 5–10 times as many S-pumps as WT mice, exhibit a smaller BP rise but accelerated TAC-induced hypertrophy and HF (25, 296). In addition, ACTH-induced and salt-induced hypertension and TAC-induced cardiac hypertrophy, but not TAC-induced pressure overload, are all markedly attenuated by immuno-neutralization of EO (60, 168, 176, 286, 296). Also, active anti-ouabain immunization with an ouabain conjugate attenuates hypertension in salt-sensitive Dahl rats (101). Collectively, these seminal studies demonstrate that EO and its receptor site participate in the pathogenesis of hypertension, cardiac hypertrophy, and HF. Moreover, the SWAP mouse data may be relevant to humans and other non-rodent mammals that have a α1^S/Sα2^S/S Na⁺ pump genotype.

I first learned of the ACTH-hypertension results when Jerry Lingrel presented them in a symposium in 2004 honoring Joe Hoffman on his retirement. Joe, who was sitting in the front row, turned around and gave me a “thumbs-up” when Jerry finished speaking. It was welcome support for the EO-Na⁺ pump-hypertension hypothesis, when many had anticipated a negative result.

There is a fly in the ointment, however. The α1^R/Rα2^R/R mice apparently can develop DOCA-salt hypertension despite the ouabain resistance (186). The discrepancy between this result and the attenuation of DOCA-salt hypertension in rats by immuno-neutralization of EO with Digibind (156) needs to be resolved. Also, the apparent paradox that α1^R/Rα2^R/R mice have elevated basal BP (Table 3) yet are unable to develop several forms of hypertension requires an explanation.

The NKA Is More Than a Cation Pump

Several other unexpected and astonishing observations that revolutionized our conception of the CTS-Na⁺ pump endocrine mechanism were made in the 1990s. One is the discovery by Xie and Askari and colleagues that ouabain binding to Na⁺ pumps activates protein kinase (PK) cascades and cell signaling apparently independent of active cation transport (110, 135, 153, 181, 231, 305). The two pathways that have been identified, epidermal growth factor receptor (EGFR)/Src/MEK/ERK1/2 and PI3K1A/PDK/Akt (87, 111, 175, 303), are activated by ouabain in a variety of cell types, including cardiomyocytes, ASM cells, and kidney tubule cells. Ouabain signaling also generates reactive oxygen species (ROS) (307). Interestingly, ouabain may either activate (150) or suppress (53, 160) some signaling pathway-mediated responses, perhaps indicating that low EO concentrations are necessary, but higher concentrations are detrimental. Also, how the PK signaling pathways are linked to the Na⁺ pumps and activated by EO is still unsettled (87, 242, 292, 320). Some investigators have suggested or implied that the activation of PK signaling is mediated by ouabain concentrations below those needed to inhibit cation transport (148, 224, 261). This would be surprising, however, because there is evidence for only a single high-affinity binding site on each αβ dimer, and X-ray crystallography shows that it lies within the cation transport channel (163, 308). Another possibility is that Na⁺ pumps function as tetramers of the αβ protomer, i.e. (αβ)₄ (267).

At first I was quite skeptical that the pump did anything other than pump cations. After all, I had lived with that classical concept for about 40 years, and it’s hard to teach on old dog a new trick (30). Nonetheless, the discovery that prolonged ouabain, but not digoxin, treatment triggers PK-mediated ion transporter reprogramming, including upregulation of NCX1 both in vivo and in vitro in ASM cells (237, 238, 323) and cardiomyocytes (32) was the clincher. Our initial report was rejected by Journal of Physiology with the claim that EO doesn’t exist (yes, it’s getting monotonous!), but was published in AJP Heart and Circulatory Physiology (238) with an accompanying “Editorial Focus” (94). Importantly, here was a possible explanation for the fact that digoxin inhibits NKA and is vasotonic but does not raise BP: The PK-mediated upregulation of NCX1 and enhancement of ouabain-induced, NCX-mediated Ca²⁺ entry at PLasmERosomes may be crucial for BP elevation (see Ref. 31). In a striking turnabout, Journal of Physiology recently invited and published our review of the evidence that α2 and its ouabain binding site are pivotal in cardiovascular physiology and pathophysiology (31).

“Brain Ouabain”—Really?

Another unanticipated development in the mid-90s was Frans Leenen’s discovery of the role of “brain ouabain” in the pathogenesis of hypertension (129, 130, 132). This followed earlier reports of an OLC in the hypothalamus (275, 282). But why was brain EO necessary, as Leenen’s experiments clearly suggested (130), if most of the circulating EO comes from the adrenal cortex (36, 118, 268) where it is synthesized (162), and if circulating EO/ouabain has a direct vasotonic effect (9, 238, 317, 318)? How did brain ouabain fit into the “big picture” if circulating EO directly augments Ca²⁺ signaling and activates the arterial (and cardiac) PK-dependent Ca²⁺ transporter reprogramming? Despite these key, unresolved questions, Leenen, and Hamlyn and I, proceeded on our separate ways for more than a decade without trying to reconcile our different viewpoints. Then, in 2010 Leenen invited me to participate in a satellite conference he organized in conjunction with the International Society on Hypertension’s Vancouver meeting. Following our divergent presentations he asked me if there wasn’t some way to reconcile our differing views. For the first time this got me to think seriously about the dilemma (30). I invited Frans to Baltimore, and John and I initiated a collaboration with him. This led to a jointly authored review in which we postulated that Leenen’s novel hypothalamic pathway, involving “brain EO” (88), helps regulate adrenal secretion of EO and circulating EO levels (34). We then embarked on a collaborative study that tested this hypothesis (121). Specifically, we found that icv infusion of low-dose Ang II elevated BP and plasma EO and upregulated arterial expression of NCX1 and SERCA type 2 (SERCA2). All these effects were blocked by icv infusion of an aldosterone synthase inhibitor or a mineralocorticoid receptor (MR) blocker (121). Yet despite all this evidence for a novel EO-Na⁺ pump endocrine mechanism, grant applications, including the renewal of our Program Project Grant (which generated much of these data) and a new collaborative program with Leenen, were rejected. For example, despite two enthusiastic reviews of one application, the primary reviewer stated (Summary Statement) that “It is intuitive that these... (central nervous system, CNS) pathways likely exist… Hence, the innovation is modest at best.” Anything is obvious when you’re given the answer [in this case, the preliminary data that supported a novel explanation for the paradoxical finding that high salt and salt restriction both elevate plasma EO (191)].

EO: a “Missing Screw” in Congestive Heart Failure?

In his 1953 monograph, Szent-Gyorgi suggested that in HF there might be a “missing screw” that exogenous CTSs are able to replace (274). Therefore, following the discovery of EO, Hamlyn, Manunta, and Stephen Gottlieb, a University of Maryland cardiologist, measured EO levels in HF patients. The results were clear: EO was elevated, not reduced (as implied by Szent-Gyorgi), in HF patients, and the level correlated inversely with cardiac index (104). Was this already a hint of EO-digoxin antagonism? A later corroborative study from Milan demonstrated that EO correlates directly with the extent of cardiac hypertrophy in patients with end stage renal disease (270). In these patients and in patients undergoing cardiopulmonary bypass surgery (22) the highest EO levels were associated with the most severe cardiac disease and the worst clinical outcomes (263).

The NIH (NHLBI) funded the Digitalis Investigative Group (DIG) clinical trial (1990 to 1998), a randomized digoxin/placebo test on 7,788 HF patients, to investigate the effectiveness of digoxin therapy [https://clinicaltrials.gov/ (58)]. The presence of EO and occurrence of ouabain-digoxin antagonism surely raise fundamental questions about the interpretation and significance of this trial’s outcome (93, 229).

During the past 15 years it has become apparent that the pathogenesis of hypertension and HF involves renin-Ang-aldosterone system (RAAS)-mediated activation of the same central pathways that drive sympathetic input to the arteries and heart (Fig. 4) (57, 226, 227, 300, 313, 321). Thus, in retrospect it is not surprising that Leenen found that “brain ouabain” participates in the same CNS mechanisms that enhance sympathetic nerve activity (SNA) in HF (169) and hypertension (129, 130, 132). After all, similar agents, including Ang-converting enzyme (ACE) inhibitors, Ang receptor blockers (ARBs), and MR blockers, are all effective therapies for both hypertension and HF (42, 140, 157, 170). Furthermore, the progression from uncontrolled hypertension to cardiac hypertrophy and then HF is well documented (46, 128, 290, 316). Additionally, as noted above (The Plot Thickens), the cardiac hypertrophy that results from TAC-induced pressure overload in mice also depends in part on the α2 Na⁺ pump ouabain binding site and on EO (25, 296). Indeed, in SWAP mice with S-α1 Na⁺ pumps the progression to HF is accelerated (296). This is a further manifestation of the EO-ouabain-sensitive Na⁺ pump-hypertension-HF relationship.

Fig. 4.

The centrally controlled, parallel sympathetic nerve and slow neurohumoral pathways that regulate both arterial and cardiac function, and participate in the pathogenesis of hypertension and heart failure (HF). Angiotensin II (Ang II) and high dietary salt (↑[Na⁺]) are convergent signals that act via hypothalamic Ang-type 1 receptors (AT₁R) to activate central nervous system (CNS) sympathetic pathways; the role of the kidneys in the retention of salt has been omitted for the sake of simplicity. The increased sympathetic nerve activity (SNA) promotes vasoconstriction and increases cardiac rate and contractile force. Prolonged stimulation of hypothalamic AT₁R also activates a novel neurohumoral pathway (box at top right) that includes aldosterone (Aldo), mineralocorticoid receptors (MR), epithelial Na⁺ channels (ENaC), endogenous ouabain (EO), and α2 Na⁺ pumps. This hypothalamic pathway feeds back (green line, “+”) to modulate Ang II-activated SNA and also promotes adrenal secretion of EO, triggered by, e.g., ACTH, adrenal SNA, and/or Ang II. The elevated plasma EO acutely inhibits α2 Na⁺ pumps in both the heart and arteries, and the rise in intracellular Na⁺ rapidly induces Na/Ca exchanger (NCX)-mediated Ca²⁺ gain, and cardiotonic and vasotonic effects. Prolonged plasma EO elevation also activates an α2 Na⁺ pump-associated protein kinase cascade (e.g., EGFR/Src/MEK/ERK-mediated) that increases cardiac and arterial NCX expression, and arterial sarcoplasmic reticulum (SR) Ca²⁺ pump (SERCA2) expression. In arteries with tone, NCX normally favors Ca²⁺ entry and helps to sustain cytosolic Ca²⁺ ([Ca²⁺]_CYT) above contraction threshold. The EO-induced NCX and SERCA2 upregulation enhances Ca²⁺ signaling and helps the very modestly increased SNA to increase vascular tone and resistance and elevate blood pressure. In the heart, NCX promotes Ca²⁺ extrusion during diastole, but prolonged α2 pump inhibition by EO reduces the Na⁺ gradient driving force so that [Na⁺]_CYT and diastolic [Ca²⁺]_CYT are both elevated; consequently, cardiac relaxation is slow and/or incomplete. Also, cardiac SERCA2 expression is usually reduced in HF (perhaps due to the high EO), as are SR Ca²⁺ stores and [Ca²⁺]_CYT transients, and systolic function is impaired. The diastolic dysfunction and attenuated cardiac contraction and stroke volume contribute to HF. These cellular mechanisms are discussed further in a recent review (31). [Modified from Blaustein et al. (31) with permission.]

Initially, I questioned the EO-HF link, so John Hamlyn and I explored the effects of coronary occlusion in mice. We found that the resultant HF is associated with elevated plasma EO and enhanced expression of both cardiac and arterial NCX1 (32). This upregulation of NCX1 is seen in HF (219, 249, 271) as well as in hypertension (237, 322, 323) (see The NKA Is More Than a Cation Pump). However, whereas SERCA2 expression in ASM is significantly upregulated as in several hypertension models (237, 238, 322), cardiac SERCA2 is decreased in the HF mice (32) as others have observed in human and experimental HF (219, 271). Not surprisingly, the signaling mechanisms appear to be cell type-specific.

How Does the Brain Talk to the Heart and Arteries?

The preceding discussion about the cardiovascular consequences of interfering with the EO-Na⁺ pump mechanism raises critical questions about how this system works. In other endocrine systems, either absence or excess of the hormone can have pathophysiological consequences as in, for example, hypo- and hyperthyroidism. This is also the case for the EO-Na⁺ pump system; furthermore, the pathophysiological consequences are not always “all-or-none.” For example, ouabain resistance in mice delays/attenuates TAC-induced cardiac hypertrophy but does not prevent it (296; see also Ref. 25). Also, transient or modest elevation of EO may be beneficial (e.g., the cardiotonic effect), but greater and more sustained elevation may have pathological consequences (hypertension, cardiac hypertrophy, and HF). How can we explain these seeming contradictions?

The key role of the sympathetics in cardiovascular control is widely accepted, but it seems logical that other mechanisms also contribute to these functions that are so vital for survival. In this regard the EO-Na⁺ pump system operates in conjunction with and in parallel to other mechanisms, especially the sympathetic nervous system (Fig. 4). Whereas α1^R/Rα2^R/R mice have only a modest baseline BP phenotype (Table 3; see The Plot Thickens), α-adrenoceptor blockade markedly reduces arterial tone (302, 315) and lowers BP (105, 220). Enhanced sympathetic drive in hypertension and HF is maintained at least in part by activation of a novel, slow neuromodulatory pathway in the hypothalamus (Fig. 4, top right corner), one component of which is locally secreted (paracrine) “brain EO” (88, 121, 300). Interruption of this pathway by lesioning the anteroventral third ventricle (AV3V) could explain why AV3V lesions prevent the maintenance of established hypertension (39). The neuromodulatory pathway feeds forward via Ang II onto the brainstem sympathetic system that drives both the heart and arteries (171). In addition to modulating SNA, the slow pathway regulates plasma EO, perhaps via SNA to the adrenals and Ang II (31, 121). This endocrine mechanism serves as a parallel pathway to the sympathetics in influencing cardiac and vascular signaling (Fig. 4; see also Ref. 31). Block of the neuromodulatory pathway therefore attenuates hypertension and HF by both central and peripheral mechanisms (121, 134) and has obvious implications for hypertension and HF therapy (166).

Circulating EO acts directly on the heart and arteries in two ways. First, by inhibiting α2 Na⁺ pumps, EO exerts acute cardiotonic and vasotonic actions [Fig. 4, bottom half); (31)]. Second, by activating PK cascades, EO triggers reprogramming of Ca²⁺ transporter protein expression exemplified by increased NCX1 in both the heart and arteries, and increased arterial but decreased cardiac SERCA2 (see “chronic” pathways in Fig. 4, bottom). In arteries with tone, because NCX1 normally mediates net Ca²⁺ entry, the upregulation of both NCX1 and SERCA2 contributes to enhanced Ca²⁺ signaling and increased tone and fosters hypertension. In the heart, however, although NCX1 normally mediates Ca²⁺ extrusion, primarily, the combination of Na⁺ pump inhibition and increased NCX1 expression tends to promote net Ca²⁺ gain, and thereby increases ventricular “stiffness” and reduces diastolic dilation (Fig. 4; bottom right) (86, 95, 200, 236). Also, the decreased SERCA2 expression reduces Ca²⁺ storage and signaling (20, 21, 273, 280). Thus the elevated plasma EO initially contributes to enhanced Ca²⁺ signaling and cardiac workload, and to hypertrophy by a sustained cardiotonic effect as well as the protein reprogramming. But, if EO elevation is greatly prolonged, and especially if the EO level continues to rise, this will eventually increase cardiac stiffness, reduce Ca²⁺ signaling, and cause HF (additional details in Ref. 31). The key role of EO and α2 Na⁺ pumps is indicated by the results of studies on TAC-induced cardiac hypertrophy. Contrary to the dogma that cardiac stretch directly induces the hypertrophy (223, 297, 309), pressure overload-induced hypertrophy is attenuated/delayed when EO is immuno-neutralized, or its receptor is knocked out or mutated to an ouabain-resistant form (248, 296).

Is EO Exclusively a Cardiovascular Hormone?

Ouabain-sensitive Na⁺ pumps are expressed in all tissues. Therefore it is appropriate to ask whether ouabain-resistant mutants and EO immuno-neutralization induce dysfunction in organ systems other than the cardiovascular system. An affirmative answer would help validate the EO-Na⁺ pump endocrine concept. The CNS might, however, be protected from circulating EO by the blood-brain barrier, which polar EO does not permeate—except at the circumventricular organs, which lack the barrier. Nevertheless, as discussed next, several groups explored the idea that brain neurons and/or astrocytes might express an EO-dependent phenotype.

EO is a mood modulator.

Following up on their earlier studies demonstrating learning and behavior alterations in heterozygous mice with a single null mutation in α2 (or α3) Na⁺ pumps (210), Lingrel and coworkers discovered that α1^R/Rα2^R/R mice have a behavioral phenotype: They exhibit decreased locomotion in a novel environment, a blunted acoustic startle reflex, enhanced response to methamphetamine (an indirect sympathomimetic), and impaired egocentric learning (254). Importantly, these specific behavioral and memory deficits, which are likely related to α2 Na⁺ pump expression in hippocampal pyramidal neurons and/or brain astrocytes, differ from those in heterozygous α2 mice (α2^S/−) (254). The α2^S/− mice, which have a null mutation in one α2 allele and express only half the normal complement of S-α2 pumps, exhibit increased anxiety, decreased locomotor activity, and impaired spatial learning (137, 210). In fact, a difference would be expected because α2^S/− is equivalent to elevation of the EO level (i.e., greater inhibition of α2) but reduction of EO-activated signaling, while α2^R/R is equivalent to elimination of EO. Thus the α2 ouabain binding site and, presumably, EO are associated with specific brain functions.

Low-dose icv ouabain infusion induces mania-like behavior in rats (69, 70). Also, immuno-neutralization of EO with icv anti-ouabain antibodies attenuates amphetamine-induced hyperactivity (another mania model) and is antidepressive in the Flinders Sensitive Line of genetically depressed rats (97, 127). An important caveat to these experiments, however, is that most neurons express (S) α3 Na⁺ pumps which have a different behavioral phenotype from α2. For example, α3 heterozygotes (α3^S/−) have spatial learning and memory deficits, and increased locomotor activity and locomotor response to methamphetamine (210). Heterozygous Mishkin mice, with an inactivating point mutation in one α3 gene (44), also display mania-like behavior resembling that of bipolar patients in the manic phase, with enhanced exploratory locomotion in a novel environment and increased sensitivity to amphetamine. These symptoms are treatable with mood-stabilizing drugs such as lithium and valproic acid, and with rostafuroxin (151). Because ouabain infusion and EO immuno-neutralization should affect both α2 and α3 pumps, the results of those tests support the view that both EO and its binding sites are important for behavior and learning, but they do not distinguish between the roles of the α2 and α3 Na⁺ pumps. [Regrettably, no α3^R/R mouse has yet been generated.]

The source of the EO that regulates the neurons involved in behavior and learning is unknown. Circulating EO might influence hypothalamic neurons in the circumventricular organs that participate in cardiovascular control (294), if only as feedback to help regulate the plasma EO level. However, EO-regulated neuronal activity in other areas of the brain surely require local EO synthesis. This circumstantial evidence that EO is synthesized and secreted in the brain complements Leenen’s findings (“Brain Ouabain”—Really?).

EO and α2 modulate skeletal muscle fatigue.

In skeletal muscle, α2, not α1, Na⁺ pumps are the most prevalent, and the α1:α2 ratio is approximately 1:6 (125), but this can vary depending on fiber type and conditions. Most skeletal muscle α2 pumps are, however, located in intracellular membranes, and exercise training and insulin promote the trafficking of α2 pumps to the PM (89, 136, 272).

EO and the α2 ouabain binding site modulate exercise endurance. α1^R/Rα2^R/R mice have fewer exercise failures on a treadmill than do WT mice (243). Conversely, mice with targeted knockout of skeletal muscle α2 Na⁺ pumps (SkM-α2^−/− mice) fatigue faster, i.e., they have more treadmill failures than do WT mice (244). Further, the increased fatigability of isolated extensor digitorum longus muscle from SkM-α2^−/− mice is mimicked by treating WT extensor digitorum longus muscle with ouabain (244). Thus, reduction of EO binding enhances, and inhibition or knockout of α2 pumps curtails exercise endurance.

EO is a stress hormone that modulates neuroinflammatory responses.

Numerous reports demonstrate that EO is a stress hormone. Plasma EO is elevated not only in hypertension, and cardiac hypertrophy and failure, but following cardiac surgery (263), in renal failure (122), in ectopic corticotropin syndrome (102), in normal human and rodent pregnancy and even more so in preeclampsia (56, 138), following intense exercise (17), and during salt depletion as well as dietary salt excess (191).

Many reports indicate that low dose ouabain/EO is a neuroprotective and antiapoptotic agent whose effects appear to be mediated by the Ca²⁺-dependent nuclear transcription factor, NF-κB (4, 53, 148, 175, 261). Several in vivo and in vitro studies also show that ouabain/EO has an anti-inflammatory action (53, 160, 224). On the other hand, some in vivo and in vitro studies indicate that ouabain/EO can activate proinflammatory responses (148, 150, 152). For example, in primary cultured glia, knockout of α2 Na⁺ pumps by siRNA prevents activation of the NF-kB-mediated inflammatory response to lipopolysaccharides (150). This is clearly a “hot area”, and new studies are needed to elucidate how EO can elicit both pro- and anti-inflammatory responses. Is this simply a concentration-dependent effect (as seems unlikely) or are other factors involved?

EO is a growth factor: role in the kidneys.

The preceding observations demonstrate that the actions of EO are widespread. Additionally, several groups have shown that low nanomolar ouabain/EO is a growth factor in a variety of tissues (10, 11, 66, 98, 217, 257). For example, ouabain promotes cyst growth and cystic fluid secretion in renal tissue from patients with an autosomal dominant polycystic disease mutation, apparently via the EGFR/Src/MEK/ERK signaling pathway (142, 216).

But is EO a growth factor in vivo? To determine the role of EO in pregnant rats, Lichtstein and colleagues immuno-neutralized the EO by intraperitoneal infusion with anti-ouabain antibodies. This treatment (vs. control IgG) significantly retards fetal development and reduces kidney and liver growth (65). Conversely, prolonged infusion of ouabain in rats increases heart and kidney weights, and these effects are abrogated by rostafuroxin (77). In vitro, immuno-neutralization of EO in growth medium containing fetal bovine serum attenuates the growth of several types of primary cells in culture (65, 66). Further, Aperia and coworkers report that malnourishment of pregnant rats by feeding a low-protein diet impairs fetal kidney growth, and that subcutaneous infusion of low-dose ouabain in protein-restricted dams protects fetal development (174). In vitro, nanomolar ouabain sustains kidney development and attenuates apoptosis of renal epithelial cells and podocytes (40, 174, 175). These latter effects appear to be due to upregulation of NF-κB (174).

It has been suggested that activation of signaling cascades by low-dose (nanomolar) ouabain in rodents is mediated by α1 pumps (174, 304), but this is at odds with the finding that ouabain-induced hypertension, which apparently depends on this signaling (31), is averted in α1^R/Rα2^R/R mice (62). Thus, the low-dose (nanomolar) ouabain/EO-induced effects on rodent kidneys needs to be reconciled with reports that R-α1 is the only Na⁺ pump α isoform detected in this tissue (71, 202). Others, however, suggest that very low levels of α2 and/or α3 are also expressed in some rat renal tubule segments (5, 76); also, α3 has been observed in rabbit kidneys in which S-α1 is the predominant isoform (16). This issue may be settled by the aforementioned report that in mouse glial cells, which express R-α1 (80–90% of pumps) and S-α2 Na⁺ pumps (99), silencing α2 with siRNA prevents ouabain from activating PK signaling and the neuroinflammatory pathway (150). Thus, as in rodent hearts (25, 31) and glial cells, a tiny minority of pumps (the S-α2 or S-α3) might mediate the signaling actions of EO in rodent kidneys. Nevertheless, given these examples of EO acting as a growth factor in the kidneys and liver, and in view of the broad distribution of ouabain-sensitive Na⁺ pumps, it seems that additional examples of this action are likely to be identified in other tissues. Clearly, we have much to learn about this aspect of EO function.

If the ouabain/EO level is too high, however, the hormone can stimulate generation of ROS and trigger apoptosis (66, 158, 284, 306, 307). Manunta and colleagues found that acute kidney injury, which may occur (e.g.) as a serious complication of cardiopulmonary bypass surgery (250, 251), is associated with high levels of circulating EO (22, 262). In rats with ouabain-induced hypertension or renal ischemia-reperfusion injury, renal podocyte nephrin, a key component of the glomerular filtration barrier, and creatinine clearance are reduced and proteinuria is increased. All these signs of impaired barrier function are prevented or reversed by treating the rats with rostafuroxin (78, 288).

Is EO a natriuretic agent?

In 1977 I suggested that the endogenous Na⁺ pump inhibitor should be a natriuretic as well as a hypertensinogenic agent (29). But, is there evidence that the low concentrations in the circulation directly inhibit Na⁺ reabsorption by the kidneys? During intravenous (iv) infusion of anti-ouabain antibodies (passive immunization) in normal rats, a transient anti-natriuresis (i.e., salt retention) was observed a day after starting the EO immuno-neutralization while BP remained normal (213). This suggests that EO may be a natriuretic agent, although the mechanism is unclear because virtually all the Na⁺ pumps responsible for Na⁺ reabsorption in the rodent renal tubule epithelium are R-α1 (202). The transience of the effect is expected with constant dietary salt because, following a brief net gain of Na⁺ and attendant anions, a new steady state should soon be achieved so that Na⁺ intake is once again balanced by output.

In an isolated case report, an 11-year-old patient who developed hyponatremia due to salt wasting following the removal of a large brain tumor was treated with iv Digibind; this reduced the natriuresis and corrected the hyponatremia (205). Here, too, the mechanism for the anti-natriuresis was not resolved but, in this case, could have been due to immuno-neutralization of nanomolar EO that normally binds to α1 Na⁺ pumps in the human kidney tubule epithelium. Furthermore, saline loading enhances natriuresis in SWAP mice, and the fact that this effect is attenuated by infusion of Digibind (184), indicates that EO is natriuretic.

In sum, the data from mutant mice and from EO immuno-neutralization studies are powerful evidence that EO, and α2 and α3 Na⁺ pumps, are components of a major endocrine/paracrine system with critical cardiovascular, renal, and CNS influences (31, 179). This endocrine system has been largely overlooked, perhaps in part because of the insistence by some that “endogenous ouabain is not ouabain” (Ref. 172; see Confirmation—Despite a Few Vocal Naysayers). In practice, however, the exact structure of EO seems secondary to validating the concept of an EO-Na⁺ pump endocrine system.

Ouabain: a Therapeutic Agent?

Circulating EO, like most hormones, is expected to have a normal physiological concentration range. Plasma levels that are too high or too low are pathophysiological as summarized in this review: e.g., elevated EO leads to reduced cardiac index (104) and increased left ventricular hypertrophy (234), and low EO impairs fetal development (65, 174). In the latter case, ouabain would appear to be the ideal replacement therapy [recall Szent-Gyorgi’s “missing screw” (274)] provided that the normal plasma level and the appropriate dosage are known.

While they are equipotent (64, 267, 311, 312), ouabain acts much more rapidly than digoxin (81) and the absorption (only ≈2–2.5% vs. ≈70–80% for digoxin) and bioavailability of orally administered ouabain (106) are far less than for digoxin (247). The latter difference is reflected in the recommended oral (or sublingual) dosage (6–12 mg for ouabain vs. 0.375–0.5 mg for digoxin) (84, 106, 107). Ouabain was therefore recommended primarily for acute iv administration in heart failure emergencies with an oral digitalis preparation then used to maintain the cardiotonic effect (84). Ever since 1885 when Fraser popularized it as a cardiotonic agent (82, 204) some physicians have advocated use of oral ouabain (85). Digitalis preparations are, however, far more widely prescribed, and ouabain is no longer approved for clinical use in the US, perhaps because iv EO may cause sudden death (3). [After all, ouabain has been used as an arrow poison for centuries because of its proven efficacy (215)!] Ouabain’s apparently narrow therapeutic window and its low and variable absorption rate are fundamental concerns. Moreover, there has been no controlled clinical trial of oral ouabain efficacy; all evidence is anecdotal. To address these concerns we need: 1) a convenient EO assay; 2) solid information about the normal plasma EO level under a variety of conditions (as noted, EO is normally elevated during intense exercise and during pregnancy); and 3) a well-controlled clinical trial including plasma EO measurements before and during therapy.

From a Na⁺-Hypertension Link to a Novel Endocrine System: Summary and Conclusions

This article summarizes the origins and development of ideas and accrual of knowledge about a newly recognized endocrine system that has been hidden in plain sight. During the past 25 years numerous ground-breaking discoveries about Na⁺ pumps and CTSs, and their influences on cardiovascular, renal, and brain physiology and pathophysiology have helped to elucidate the EO-Na⁺ pump endocrine system. The mammalian Na⁺ pump is a most unusual hormone receptor because it functions both as a cation transporter that does not require the hormone for activity (although some isoforms can be modulated by EO), and as an EO-activated signal transducer that appears to function independently of the cation transport. The α-subunit is actually an ancient protein: it is found in Archea and bacteria as well as eukaryotes. Indeed, hydra, nematode, and fruit fly NKA α-subunits have substantial sequence homology and tertiary structure similarity to the human NKA α (208, 222, 281). The invertebrate β- and human β-subunit sequences are, however, more distinct (222). Nevertheless, ouabain also inhibits the hydra and nematode NKAs (41, 51, 52). Thus, CTS binding activity has been conserved through evolution despite the occasional appearance of ouabain resistant pumps, as in rodents and insects (51), but the significance of CTS binding in invertebrates and lower vertebrates is unknown. It should be very instructive to learn when and how the endogenous CTS-Na⁺ pump couple evolved into a hormone-receptor complex. When, during evolution, did EO first appear? And when did EO-activated cell signaling evolve?

The functional consequences of interfering with the EO-Na⁺ pump (hormone-receptor) interaction, either by mutation of the ligand binding site to an ouabain-resistant form or by immuno-neutralization of EO, demonstrate that EO and Na⁺ pumps are an endocrine/paracrine system with very broad impact. The idea that an endogenous Na⁺ pump inhibitor/CTS is a vasotonic agent that plays a role in the pathogenesis of hypertension has been validated (31), but the mechanism is far more complicated than originally envisioned (29). This is exemplified by the multitude of unexpected, dogma-defying findings (see Table 1). These include: cloning of the Na⁺ pump α and β subunit isoforms; discovery that hormone ouabain (EO) is synthesized in the brain as well as the adrenals, and that circulating EO is regulated by a novel neurohumoral pathway in the hypothalamus; discovery that EO and brain Na⁺ pumps modulate behavior and learning; identification of the ouabain-Na⁺ pump interaction as an activator of PK cascades and cell signaling; recognition that the actions of ouabain-like and digoxin-like CTSs are not identical, and that DLCs do not activate PK cascades; discovery that ouabain/EO can be both pro- and anti-inflammatory; and recognition that EO is a stress hormone and a growth factor and is essential for normal fetal development. It is difficult to grasp the full scope of this novel humoral system with all of its complexity and myriad actions.

The past 40 years have been an exciting roller coaster ride, but our odyssey is far from complete. We undoubtedly will learn much more about EO as a hormone when a specific antagonist becomes available, as happened with the RAAS when ARBs like losartan were developed. We also need a convenient commercial assay for EO. There are still numerous questions to answer and controversies to resolve, starting with elucidation of the steps in the synthesis of EO from progesterone. A major concern is whether funds will be available for the requisite studies that utilize standard methods and are not very “glitzy.” Because of a research funding squeeze the US is starting to lose its leadership in the EO-Na⁺ pump field despite its obvious clinical relevance, just as research on this novel endocrine system is picking up in Canada, Israel, Italy, Japan, Sweden, and the UK.

The EO-Na⁺ pump system influences the physiology of most, if not all, organs in the body. The complexity of this system is illustrated by the role of EO and Na⁺ pumps in hypertension (Fig. 4), from whence I began with a far more simplistic view (Fig. 1). Too often we lose sight of Page’s prescient insight that hypertension is a mosaic (230). Multiple factors (genetic, environmental, immunologic, renal, neural, endocrine, cardiac, and vascular) are essential contributors to the BP elevation—even when a single cause such as a monogenic defect in salt transport (177) can be identified as the initiating element. A plethora of components and interacting pathways that influence BP have been elucidated, but they are generally studied one bit at a time, and they can’t all be “essential.” Which ones are really important? It appears that we need a new, more quantitative approach, a new way to juxtapose all the pieces to determine how these elements all work together to regulate BP. It’s time for the engineer-biologists: we need to learn how to “fix a radio” (164). Here is the essence and future of (integrative) physiology!

The history of progress with the EO-Na⁺ pump endocrine system calls to mind T. H. Huxley’s admonition (letter to C. Kingsley, 1860): “Sit down before fact as a little child, be prepared to give up every preconceived notion, follow humbly wherever and to whatever abysses nature leads, or you shall learn nothing.”—but be sure of the “facts.”

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.P.B. drafted, edited and revised the manuscript, and approved the final version.

Glossary

α1^R/Rα2^R/R

Ouabain-resistant (R-) α2 Na⁺ pump mice

α2^S/− (α3^S/−)

Heterozygous α2 (α3) Na⁺ pump mice

ACE

Angiotensin-converting enzyme

Ang

Angiotensin

ARB

Ang receptor blocker

ASM

Arterial smooth muscle

BP

Blood pressure

CNS

Central nervous system

CO

Cardiac output

CTS

cardiotonic steroid

CYT

Cytosolic

DLC

Digitalis-like compound

ELISA

Enzyme-linked immunosorbent assay

EO

Endogenous ouabain

HF

Heart failure

HPLC

High pressure liquid chromatography

icv

Intracerebroventricular

iv

Intravenous

MBP

Mean BP

MBG

Marinobufagenin

MR

Mineralocorticoid receptor

MS

Mass spectroscopy

NCX

Na/Ca exchanger

NKA

Na⁺-K⁺-ATPase (Na⁺ pump)

NMR

Nuclear magnetic resonance

PK

Protein kinase

PM

Plasma membrane

RAAS

Renin-angiotensin-aldosterone system

RIA

Radioimmunoassay

ROS

Reactive oxygen species

SERCA

Sarco-/endoplasmic reticulum Ca²⁺ pump

SNA

Sympathetic nerve activity

SR

Sarcoplasmic reticulum

SWAP (α1^S/Sα2^R/R)

R-α2 Na⁺ pump mice with ouabain-sensitive (S-) α1 Na⁺ pumps

TAC

Trans-aortic constriction

TPR

Total peripheral resistance

WT

Wild type (α1^R/Rα2^S/S) mice.