TZD doses exceeded those required for maximal efficacy in early years of TZD use (Tamargo et al., 2014). Subsequently, TZD doses were substantially decreased in order to lower the incidence of side effects, with the overwhelming majority of TZD doses established at less than p.o. 50 mg/day (approximately 0.65 mg/kg/day; Musini et al., 2014; Tamargo et al., 2014; Vongpatanasin, 2014).

Bioavailability of TZD varies from 65% to 95% (Tamargo et al., 2014). Treatment of hypertension with p.o. hydrochlorothiazide 7.5–25 mg/day (approximately 0.1–0.3 mg/kg/day) resulted in a median plasma concentration of 0.26 µM (Sigaroudi et al., 2018). Consistent with the plasma hydrochlorothiazide level derived at these doses (Sigaroudi et al., 2018), treatment of hypertension and congestive heart failure with p.o. 75 mg/day (approximately 1 mg/kg/day) and 50 mg/day (approximately 0.7 mg/kg/day), respectively, yielded peak blood levels of approximately 1–3 µM (Beermann and Groschinsky‐Grind, 1977; Beermann and Groschinsky‐Grind, 1978; Beermann and Groschinsky‐Grind, 1979).

Acute TZD reduction of arterial pressure correlates with diuresis and the associated decreased plasma volume in hypertensive humans and animals ( Table 1g ; Figure 1 ). Exceptions to this correlation are TZD pressure reduction prior to decreased plasma volume (Gifford et al., 1961; Greene et al., 1961; Lockett and Nicholas, 1968).

Plasma volume repletion reversed acute TZD reduction of arterial pressure, consistent with the correlation between acute arterial pressure reduction and diuresis ( Figure 1 ). Specifically, after TZD treatment for 6 days, dextran infusion increased arterial pressure as well as the cardiac index and reduced total peripheral vascular resistance in humans (Dustan et al., 1959). However, after TZD treatment for 7 days, in which the plasma volume decrease was maintained at the initial reduced level, plasma and saline infusion did not alter arterial pressure (Hollander et al., 1960). Also, in barbiturate-anesthetized 1 kidney/DOCA hypertensive rats challenged with TZD for 5–8 days, hypotonic saline and dextrose infusion did not reverse the reduced arterial pressure, although hypertonic saline infusion increased systolic pressure, albeit not diastolic pressure (Daniel, 1962).

In contrast to acute TZD, chronic TZD reduction of arterial pressure appears dissociated from diuresis and decreased plasma volume ( Figure 1 ). With respect to diuresis, TZD chronic reduction in arterial pressure is maintained while diuresis is minimized in hypertensive humans ( Table 1h ). TZD also did not alter serum Na⁺ ( Table 1h ). Additionally, water and Na⁺ balance returned to normal during TZD challenge of hypertensive high Na⁺ diet/angiotensin II rats and spontaneously hypertensive rats, while reduced arterial pressure was maintained (Ballew and Fink, 2001; Jessup et al., 2008). Also consistent with the dissociation between diuresis and chronic TZD reduction of arterial pressure is the lesser diuretic efficacy of TZD than inhibitors of the luminal Na–K–Cl cotransporter (NKCC2) located in the thick ascending limb of Henle (loop diuretics), i.e., the NCC and NKCC2 regulate approximately 7% and 25% of Na⁺ absorption, respectively (Glover and O’Shaughnessy, 2013), while the antihypertensive efficacy of TZD was greater than that of NKCC2 inhibitors (Anderson et al., 1971; Araoye et al., 1978; Holland et al., 1979). Contributing to the minimal diuresis is increased expression of epithelial Na⁺ channel (ENaC) and Na⁺ anion exchanger, pendrin (Na et al., 2003; Alshahrani and Soleimani, 2017; Figure 1 ).

With respect to plasma volume, TZD-decreased plasma volume apparently returned to pre-TZD level despite continuous TZD for 1–12 months in hypertensive humans ( Table 1i ; Figure 1 ). Whether, in fact, plasma volume completely or only partially returned to pre-TZD level is unclear because a number of these studies demonstrated a tendency for reduced plasma volume. Moreover, the magnitude of plasma volume return to pre-TZD level may be overestimated by increased hemoglobin (Hilden et al., 1968; Leth, 1970). Along these lines, the TZD lowered plasma volume only partially returned at 2–3 months of TZD treatment (Winer, 1961; Hansen, 1968), and as determined at a single time point during TZD treatment, plasma volume was decreased at 12 and 24 months of TZD (Lund-Johansen, 1970; Tarazi et al., 1970; de Carvalho et al., 1977).

Also consistent with the only partial return of plasma volume with chronic TZD is the return of TZD-reduced arterial pressure following plasma volume expansion with dextran plus saline or glucose infusion at 2 weeks to 8 months of TZD treatment in hypertensive humans (Wilson and Freis, 1959). In partial agreement, following 2 weeks to 9 months of TZD treatment, dextran plus glucose infusion resulted in partial return of systolic pressure while the decreased diastolic pressure remained (Winer, 1961).

Additionally, plasma volume repletion with elevated dietary Na⁺ reversed TZD chronic reduction of arterial pressure in hypertensive humans (Wilkins et al., 1958). At 6 days post single TZD injection (i.v. 1 g of chlorothiazide), increased dietary Na⁺ of 140 mEq/day (1.8 g/day) from previously lowered 9 mEq/day (0.2 g/day), completely reversed the sustained reduction of arterial pressure to pre-TZD level (Wilkins et al., 1958). In partial agreement, greater amounts of dietary Na⁺ salt, i.e., 6–12 g/day (2.4–4.8 g/day Na⁺) only partially reversed TZD-mediated arterial pressure reduction, and complete reversal required 20 g/day (8 g/day Na⁺; Freis et al., 1958; Wilkins et al., 1958; Freis, 1959; Winer, 1961). Also, following withdrawal of TZD, plasma volume increases (Wilson and Freis, 1959; Tarazi et al., 1970).

Apparently inconsistent with the correlation between elevated Na⁺ and reversal of TZD-reduced arterial pressure, TZD reduction of arterial pressure remained despite prevention of Na⁺ loss by mineralocorticoid co-administration (Hollander et al., 1960). However, it is likely that TZD-decreased plasma volume was actually maintained because weight loss, an indirect measure of plasma volume, also remained following mineralocorticoid co-administration (Hollander et al., 1960).

Decreased Na⁺ excretion as the result of Na⁺-restricted diet (urinary excretion from 3.75 g Na⁺/day to 2 g Na⁺/day) which was associated with reduced arterial pressure and increased renin activity and aldosterone plasma level, also failed to prevent further arterial pressure reduction by TZD in the absence or presence of angiotensin receptor antagonists in humans (Priddle et al., 1970; Vogt et al., 2008; Wang et al., 2015b). The TZD arterial pressure reduction was still associated with plasma volume contraction because Na⁺-restricted diet did not prevent TZD-increased Na⁺ excretion (urine volume not reported) and futher increased plasma renin activity and aldosterone levels (Priddle et al., 1970; Vogt et al., 2008).

Among the findings in support of an extrarenal, vascular mechanism mediating TZD chronic reduction of arterial pressure in hypertensive humans and animals is the maintained pressure reduction despite return of plasma volume towards pre-TZD levels ( Figure 1 ). Overall, however, it appears that a remaining component of TZD acute decreased plasma volume is required for chronic TZD reduction in arterial pressure ( Figures 1 , 2 ). Thus, TZD chronic reduction of arterial pressure is dependent upon decreased plasma volume both as a trigger for the subsequent maintained arterial pressure reduction and for its ability to sustain the reduction ( Figure 2 ).

Acute and chronic TZD failed to reduce arterial pressure in normotensive humans and animals despite the diuresis and decreased plasma volume ( Table 1j ; Figure 1 ). A supra-therapeutic TZD dose was required to reduce arterial pressure in dogs (Preziosi et al., 1959). Lack of TZD reduction of arterial pressure was not due to decreased diuresis because TZD caused similar magnitudes of changes in Na⁺ and K⁺ per body weight, as well as extracellular fluid volume and serum osmolarity, in normotensive subjects and hypertensive patients (Freis et al., 1960; Hollander et al., 1960; Figure 1 ). Diuresis due to NKCC2 inhibitors was also greater than diuresis due to TZD in normotensive subjects, while NKCC2 inhibitors still failed to reduce arterial pressure (Joubert et al., 1968).

In apparent contrast to the lack of TZD reduction of arterial pressure in normotensive humans and animals ( Table 1j ; Figure 1 ), in hypertensive humans. TZD and Na⁺-restricted diet elicited a steady reduction of arterial pressure (Priddle et al., 1970). Na⁺-restricted diet did not prevent TZD diuresis and natriuresis in rats (Hofmann and Sagartz, 1970) and, thus, were presumably also not prevented in humans.

Although TZD did not cause diuresis in normotensive mice, the absence of diuresis may actually reflect the relatively delayed assessment, i.e., day 3–4 post initial TZD (Alshahrani and Soleimani, 2017). This suggestion is supported by the return to baseline within 24 and 72 h of increased diuresis due to p.o. TZD q.d. (Fuchs et al., 1960; Claus-Walker et al., 1977) and 48 h p.o. TZD b.i.d. (Wilson and Freis, 1959). Also, ENaC and pendrin expression, which presumably compensates for TZD diuresis through increased Na⁺ reabsorption, increased on day 3 post initial TZD treatment (earliest time point assayed; Alshahrani and Soleimani, 2017; Figure 1 ). Along these lines, diuresis increased 2 h post a single i.v. injection of TZD in rats and ENaC activity remained unchanged (Frindt et al., 2017).

TZD failed to reduce arterial pressure in normotensive humans and animals despite similar magnitudes of diuresis and plasma volume decrease as in hypertensive humans and animals ( Figures 1 , 2 ). The lack of TZD reduction of arterial pressure suggests a balance between compensatory contraction of the plasma volume and TZD mitigation of the contracted volume through inhibition of vasoconstriction ( Figure 2 ). However, this balance is disrupted by further plasma volume contraction, which can occur, e.g., with increased Na⁺ loss, resulting in TZD reduction of arterial pressure ( Figure 2 ).

TZD-mediated increased diuresis and plasma volume reduction, and body weight loss were greater in responders than nonresponders (de Carvalho et al., 1977; Freis et al., 1988; Table 2 ). Also, TZD failed to decrease plasma volume in a study limited to nonresponders (Svendsen et al., 1983). The greater contracted plasma volume of responders was suggested to underlie increased TZD reduction of total peripheral vascular resistance (de Carvalho et al., 1977). Consistent with the lesser magnitude of plasma volume contraction in nonresponders is the lack of angiotensin II and aldosterone plasma level elevation (Svendsen et al., 1983; Constrictor Pathways In Vivo, Endogenous Activation; Table 2 ). The lesser decrease in plasma volume in nonresponders than responders was not associated with lesser decreases in serum K⁺ (de Carvalho et al., 1977; Svendsen et al., 1983).

In contrast, there was a tendency for increased plasma volume contraction in chronic TZD-treated nonresponders than responders (van Brummelen et al., 1980; Table 2 ). It was suggested that the tendency for greater plasma volume contraction in nonresponders, which was associated with a tendency for greater elevated plasma levels of angiotensin II (Constrictor Pathways In Vivo, Endogenous Activation; Table 2 ), resulted in increased peripheral vascular resistance (van Brummelen et al., 1980) and, thus, diminished TZD-decreased arterial pressure. On the other hand, body weight was not different in responders and nonresponders (van Brummelen et al., 1980; Table 1 ).

The dependency of TZD reduction of arterial pressure upon relaxation of vasoconstriction suggests that the inability of TZD to reduce arterial pressure in nonresponders may reflect the lack of compensatory vasoconstriction ( Table 2 ; Figure 2 ). That is, the inability of TZD to elicit diuresis results in lack of decreased plasma volume and, therefore, absence of compensatory vasoconstriction ( Table 2 ; Figure 2 ). This conclusion is supported by the lack of TZD arterial pressure reduction in normotensive humans and animals despite increased diuresis (above). Additionally, lack of diuresis in some types of hypertension associated with nonresponders could result from prevention of diuresis ( Table 2 ; Figure 2 ).

Alternatively, TZD may induce diuresis in TZD-insensitive hypertension with different etiologies. Thus, the absence of arterial pressure reduction could result from inhibitory effects post plasma volume depletion ( Figure 2 ). Additional studies of the effects of TZD in nonresponders with different hypertensive etiologies are warranted, including the mechanism underlying TZD reduction of arterial pressure in nonresponders with co-administration of other antihypertensive agents (Sander and Giles, 2011; Tamargo et al., 2014).

NCC activity determinations include increased NCC expression and phosphorylation, with the latter an indicator of NCC trafficking (Glover and O’Shaughnessy, 2013; Pathare et al., 2017). These two measurements may be unrelated because phosphorylation can be dissociated from increased NCC expression (Frindt et al., 2017). TZD binding to NCC, determined by high-affinity binding with ³H-metolazone (Beaumont et al., 1988), has also been used as an indirect measure of NCC expression and, thus, activity (Chen et al., 1990; Morsing et al., 1991). However, even though TZD increased NCC binding, transport was reduced (Morsing et al., 1991). Additional potential complications with binding studies include the inability to account for NCC isoforms along with selective isoform phosphorylation (Tutakhel et al., 2016; Pathare et al., 2017), as well as the influence of Na⁺ and Cl⁻ concentrations (Beaumont et al., 1988; Tran et al., 1990).

In responders, chronic TZD increased NCC expression and NCC phosphorylation, as assayed in urinary vesicular exosomes (Pathare et al., 2017).

Acute and chronic TZD increased NCC expression and phosphorylation in whole kidney and, in some studies, distal convoluted tubules from normotensive rats and mice (Northern and Western blot, in situ hybridization, and immunohistochemistry; Loffing et al., 1996; Na et al., 2003; Nijenhuis et al., 2005; Frindt et al., 2017). TZD also increased ³H-metolazone high-affinity binding to NCC in kidney membrane fractions from normotensive rats (Chen et al., 1990; Morsing et al., 1991).

Enhanced delivery of Na⁺ to the distal collecting duct likely underlies the increased NCC expression (Pathare et al., 2017) because furosemide, a diuretic that acts through inhibition of NKCC2, also increased NCC expression and binding in rat kidney (Chen et al., 1990; Na et al., 2003). Also, angiotensin receptor antagonists failed to increase NCC expression and phosphorylation even though arterial pressure was reduced to a similar magnitude as TZD (Pathare et al., 2017).

TZD-induced NCC expression and phosphorylation were greater in responders than nonresponders, as determined in urinary vesicular exosomes (Pathare et al., 2017; Table 1 ).

Increased NCC expression and phosphorylation due to TZD are consistent with the greater amount of diuresis in responders than nonresponders (Freis et al., 1988; Table 2 ; Figure 2 ; Diuresis and Plasma Volume). In normotensive animals, there is also a positive relationship between TZD-mediated NCC expression/phosphorylation and diuresis (Diuresis and Plasma Volume). NCC activation is restricted to agents that increase Na⁺ delivery to the distal collecting duct.

NCC knockout mice demonstrated changes similar to those observed in Gitelman’s syndrome, in which NCC is dysfunctional (below), including hypomagnesemia, hypocalciuria, and increased renin expression (determined in the kidney; Schultheis et al., 1998; Loffing et al., 2004; Alshahrani et al., 2017a). It should be noted, however, that TZD increased NCC expression and that NCC expression is decreased in Gitelman’s syndrome (Joo et al., 2007; Isobe et al., 2013; Yang et al., 2013; Frindt et al., 2017). This difference in NCC expression with TZD and Gitelman’s syndrome may reflect different mechanisms whereby TZD inhibit NCC activity (Payne, 2012) and of NCC dysfunction in Gitelman’s syndrome (Riveira-Munoz et al., 2007a, Riveira-Munoz et al., 2007b; Wang et al., 2015a).

Plasma bicarbonate levels were also elevated in NCC knockout mice, indicative of underlying, compensated alkalosis (Loffing et al., 2004). Additionally, aldosterone plasma levels were elevated (Loffing et al., 2004), although elevated aldosterone plasma levels were not consistently observed (Schultheis et al., 1998).

Basal arterial pressure: Basal arterial pressure was not significantly decreased in NCC knockout mice, although there was a tendency for decreased arterial pressure (Schultheis et al., 1998). The absence of hypotension in NCC knockout mice was attributed to compensation of mildly decreased intravascular volume (Schultheis et al., 1998; Loffing et al., 2004; Alshahrani et al., 2017a). Lack of hypotension in NCC knockout mice (Schultheis et al., 1998; Loffing et al., 2004; Alshahrani et al., 2017a), but hypotension in 50% of Gitelman’s syndrome patients (Cruz et al., 2001; below), may reflect differences in Na⁺ handling in mice and humans. This suggestion is supported by the arterial pressure reduction with Na⁺-restricted diet still containing 77.5% of normal in humans (Priddle et al., 1970), but the absence of arterial pressure reduction by more severe Na⁺-restricted diet of 0.01% and 0.1% in wild-type mice (Schultheis et al., 1998; Alshahrani et al., 2017a).

In contrast, arterial pressure was lower in NCC knockout mice with Na⁺-restricted diet (0.01% or 0.1%; Schultheis et al., 1998; Alshahrani et al., 2017a). Thus, it appears that compensatory vasoconstrictor pathways were unable to fully normalize the lower arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet despite further increased renin expression (Schultheis et al., 1998; Alshahrani et al., 2017a). Similarly, NCC/pendrin double-knockout mice, which demonstrated severe diuresis and natriuresis along with even greater renin expression, was associated with further reduced arterial pressure (Soleimani et al., 2012).

TZD on arterial pressure: Hydrochlorothiazide failed to reduce arterial pressure in NCC knockout mice, despite s.c. administration at the supra-therapeutic dose of 20 mg/kg, i.e., approximately 30-fold greater than the therapeutic dose (Alshahrani et al., 2017a; Dose/Concentration of TZD). In fact, 20 mg/kg of hydrochlorothiazide would result in a plasma concentration of approximately 50 µM (Dose/Concentration of TZD). Lack of hydrochlorothiazide reduced arterial pressure in NCC knockout mice may reflect conditions similar to those associated with normotensive subjects and animals ( Figure 2 ). That is, vasoconstriction available for TZD relaxation is minimal due to the relatively small magnitude of decreased intravascular volume and, thus, minimal compensatory vasoconstriction ( Figure 2 ).

In contrast, hydrochlorothiazide (20 mg/kg) reduced arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet (Alshahrani et al., 2017a). Moreover, in NCC/pendrin double-knockout mice, hydrochlorothiazide elicited an even greater reduction in arterial pressure than in NCC knockout mice subjected to Na⁺-restricted diet (Alshahrani et al., 2017a). Thus, the magnitude of hydrochlorothiazide arterial pressure reduction correlated with greater underlying vasoconstriction as the result of increased intravascular volume depletion (Soleimani et al., 2012; Alshahrani et al., 2017a; Figure 2 ). Consistent with this suggestion is that prevention of intravascular volume depletion with increased dietary Na⁺ from 1% to 7% abolished both the lowered basal arterial pressure and TZD reduction of arterial pressure in NCC/pendrin double-knockout mice (Alshahrani et al., 2017a).

It was suggested that TZD reduction of arterial pressure in the absence of NCC provided evidence for an extrarenal TZD target (Alshahrani et al., 2017a). Further, hydrochlorothiazide arterial pressure reduction does not appear mediated by decreased plasma volume, due to decreased plasma Na⁺ (Alshahrani et al., 2017a). This conclusion is based upon the finding that urine production was actually decreased in NCC/pendrin double-knockout mice (Alshahrani et al., 2017a). The decreased urine production was attributed to lowered kidney perfusion with reduced arterial pressure (Alshahrani et al., 2017a). Also contributing to the decreased urine production was lowered water intake (Alshahrani et al., 2017a). Inexplicably, food intake was greatly decreased by hydrochlorothiazide (Alshahrani et al., 2017a).

Measurements of plasma Na⁺ would further test whether decreased plasma volume through lowered plasma Na⁺ was involved in hydrochlorothiazide reduction of arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet and NCC/pendrin double-knockout mice (Alshahrani et al., 2017a). In this regard, hydrochlorothiazide increased Na⁺ excretion in NCC knockout mice (Leviel et al., 2010; Eladari and Chambrey, 2011). Furthermore, the increased Na⁺ excretion by hydrochlorothiazide appears mediated through inhibition of the Na⁺-driven chloride–bicarbonate exchanger (NDCBE; SLC4A8e; Ware et al., 2017; Nadal et al., 2018; Xu et al., 2018) because increased Na⁺ excretion, at least at 50 mg/kg i.p. hydrochlorothiazide, was absent in NCC/NDCBE double-knockout mice (Leviel et al., 2010; Eladari and Chambrey, 2011; Eladari and Hübner, 2011; Sinning et al., 2017).

It should also be considered that the hydrochlorothiazide dose that reduced arterial pressure in the absence of NCC (Alshahrani et al., 2017a) yields a plasma concentration in the range that inhibits NDCBE. That is, hydrochlorothiazide doses that increased Na⁺ excretion in NCC knockout mice, i.p. hydrochlorothiazide 4 and 50 mg/kg (Leviel et al., 2010; Eladari and Chambrey, 2011), correspond to plasma concentrations of at least 10 and 100 µM, respectively (Dose/Concentration of TZD). In fact, the hydrochlorothiazide concentration used in the metabolic studies in NCC/pendrin double-knockout mice, 40 mg/kg, would yield a plasma concentration of approximately 100 µM (Alshahrani et al., 2017a; Dose/Concentration of TZD). Further, 100 µM hydrochlorothiazide inhibited NDCBE in perfused renal cortical collecting ducts of mice (Leviel et al., 2010). In comparison, the IC₅₀ for metolazone inhibition of Na⁺ uptake by NCC was 2 or 0.3 µM (depending upon the report; Sabath et al., 2004; Moreno et al., 2006). Also, essentially complete inhibition of NCC by metolazone occurred at 10 µM (Sabath et al., 2004; Moreno et al., 2006).

On the other hand, 10 mg/kg i.p. hydrochlorothiazide was selective for NCC as demonstrated in STE20/SPS1-related proline/alanine-rich kinase (SPAK)-dead knock-in mice (Glover and O’Shaughnessy, 2013). Specifically, in SPAK-dead knock-in mice, which demonstrated decreased NCC expression and phosphorylation leading to overall decreased NCC activity, hydrochlorothiazide saliuresis was absent (Rafiqi et al., 2010; Glover and O’Shaughnessy, 2013). Although NKCC2 expression and phosphorylation were also decreased in SPAK-dead knock-in mice (Rafiqi et al., 2010; Glover and O’Shaughnessy, 2013), saliuresis to the loop diuretic, furosemide, was not decreased (Rafiqi et al., 2010; Glover and O’Shaughnessy, 2013). However, whether NDCBE function is also decreased in SPAK-dead knock-in mice is not known (to our knowledge).

Although a lowered IC₅₀ for TZD inhibition due to increased NDCBE expression following NCC knockout (Eladari and Chambrey, 2011) could occur, depending upon the presence of NDCBE spare transporters, a lowered IC₅₀ does not appear to occur at least in Gitelman’s syndrome. In support of this conclusion is the decreased Na⁺ and Cl⁻ excretion with a single therapeutic dose of hydrochlorothiazide (p.o. 50 mg; Colussi et al., 1997; Peng et al., 2017; below).

While NDCBE-mediated TZD reduction of arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet and NCC/pendrin double-knockout mice (Alshahrani et al., 2017a) may occur through hyponatremia, it should also be considered that hyponatremia may not accompany increased Na⁺ excretion. However, TZD-induced hyponatremia has been associated with decreased NCC (Channavajjhala et al., 2018). In fact, hyponatremia was observed in approximately one of seven TZD-treated patients (Glover and Clayton, 2012; Burst et al., 2017). Further, in patients with demonstrated hyponatremia treated with hydrochlorothiazide plus amiloride, challenge with 50 mg of hydrochlorothiazide, albeit combined with 5 mg of amiloride, caused hyponatremia (Friedman et al., 1989).

A major caveat, however, with respect to the potential functional role of NDCBE in Na⁺ handling and, therefore, possible high dose TZD-induced saliuresis and/or natriuresis, is the location of NDCBE within the kidney. Specifically, it was suggested that NDCBE was located in β-intercalated cells of the cortical collecting duct, consistent with saliuersis/natriuresis regulation (Ware et al., 2017; Nadal et al., 2018). In direct contrast, we were unable to locate NDCBE in cortical collecting duct β-intercalated cells (Xu et al., 2018). Moreover, NDCBE was restricted to the basolateral membrane of the medullary collecting duct (Xu et al., 2018), thereby questioning the potential role of NDCBE in Na+ handling.

TZD reduction of arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet, but not NCC knockout mice with normal Na⁺ diet, and an even greater pressure reduction in NCC/pendrin double-knockout mice (Alshahrani et al., 2017a) are consistent with the dependency of TZD reduction of arterial pressure on decreased plasma volume per se and associated compensatory vasoconstriction (Diuresis and Plasma Volume; Figure 2 ). Thus, the considerable diuresis associated with less reduction of arterial pressure with NKCC2 inhibitors compared with TZD (Wilson and Freis, 1959; Anderson et al., 1971; Araoye et al., 1978; Holland et al., 1979; Diuresis and Plasma Volume) would reflect unopposed compensatory vasoconstriction with the NKCC2 inhibitors.

However, while TZD-reduced arterial pressure in NCC knockout mice with Na⁺-restricted diet and an even greater pressure reduction in NCC/pendrin double-knockout mice are suggestive of an extrarenal TZD target, an involvement of hyponatremia due to NDCBE inhibition cannot be completely eliminated because of the use of supra-therapeutic TZD dose (Alshahrani et al., 2017a). In fact, lack of TZD reduction of arterial pressure in Gitelman’s syndrome (below) at least indirectly supports the possibility of NDCBE inhibition in TZD reduction of arterial pressure. Overall, however, considering that NDCBE is apparently located on the basolateral membrane of medullary collecting duct (Xu et al., 2018) rather than in β-intercalated cells of the cortical collecting duct (Ware et al., 2017; Nadal et al., 2018), the function of NDCBE in response to hydrochlorothiazide needs to be revisited.

Gitelman’s syndrome is an autosomal recessive malady attributed to dysfunctional NCC, largely due to missense mutations (Riveira-Munoz et al., 2007a; Riveira-Munoz et al., 2007b; Wang et al., 2015a). These mutations result in undetectable or reduced NCC amounts in urinary exosomes (Joo et al., 2007; Isobe et al., 2013; Yang et al., 2013). Consistent with NCC dysfunction in Gitelman’s syndrome is the lesser increase in urinary excretion of Na⁺ and Cl⁻ to hydrochlorothiazide (single, therapeutic dose of p.o. 50 mg; Table 1k ; Dose/Concentration of TZD). In contrast, urinary excretion of Na⁺ and Cl⁻ increased in response to furosemide (loop diuretic; i.v. bolus infusion of 40 mg in Colussi et al., 1997 and i.m. 20 mg in Peng et al., 2016).

The phenotype associated with Gitelman’s syndrome includes hypomagnesemia, hypocalciuria, hypochloremia, and hypokalemic alkalosis ( Table 1I ; mimicked by NCC knockout mice; Schultheis et al., 1998; Alshahrani et al., 2017a; above). Vascular volume depletion is also associated with Gitelman’s syndrome, which results in elevated plasma renin levels/activity and aldosterone levels (Tsukamoto et al., 1995; Simon et al., 1996). Hypotension occurs in approximately 50% of patients with Gitelman’s syndrome (62% and 12% of patients and control subjects, respectively; Cruz et al., 2001). Based on measurements in neutrophils and monocytes from patients with Gitelman’s syndrome, the hypotension and absence of hypertension despite elevated plasma renin levels/activity and aldosterone levels were attributed to up-regulation/down-regulation of signaling pathways that would result in decreased vasoconstriction and basal vascular tone (Calò et al., 1998, Calò et al., 1999; Calò, 2006). These signaling pathways include decreased inositol trisphosphate release of intracellular Ca²⁺, decreased activation of protein kinase C, increased nitric oxide formation due to increased endothelial nitric oxide synthase expression, inhibition of oxidation signaling mechanisms (which mediate vasoconstriction to angiotensin II and other constrictors), decreased G-protein coupling through reduced α_q expression, decreased agonist activation of Rho kinase through down-regulation of RhoA/Rho kinase signaling, and increased heme-oxygenase-1 expression (Calò et al., 1998, Calò et al., 1999; Calò, 2006).

However, possibly inconsistent with the predicted decreased agonist constriction and/or basal vascular tone in patients with Gitelman’s syndrome (Calò et al., 1998, Calò et al., 1999; Calò, 2006) are 1) autonomic function was preserved in hypotensive, Gitelman’s syndrome patients (Sartori et al., 2007). Autonomic function was determined indirectly by reflex tests including the Valsalva maneuver, cold pressor, hand grip, and hyperventilation and by norepinephrine and arginine vasopressin plasma levels in response to the reflex tests (Sartori et al., 2007). And 2) basal blood flow in the forearm of two hypotensive patients with Gitelman’s syndrome did not differ from basal blood flow in normotensive and hypertensive subjects (Pickkers et al., 1998).

Also, possibly inconsistent with the potential extrarenal TZD target in NCC knockout mice subjected to Na⁺-restricted diet and in NCC/pendrin double-knockout mice (Alshahrani et al., 2017a), in two patients with Gitelman’s syndrome with 8 and 25 µg hydrochlorothiazide/min/dL forearm infusion (plasma hydrochlorothiazide concentrations of 2.7 and 11.0 µM, respectively), blood flow did not increase (Pickkers et al., 1998). Similarly, while 75 µg hydrochlorothiazide/min/dl infusion (38 µM of hydrochlorothiazide) increased forearm blood flow in Gitelman’s syndrome patients, the magnitude of increase was not different from the increased blood flow in normotensive and hypertensive subjects (Pickkers et al., 1998). The increased blood flow to 75 µg hydrochlorothiazide infusion likely resulted from non-specific effects to the approximately 20-fold greater than therapeutic plasma levels of hydrochlorothiazide (Pickkers et al., 1998; Dose/Concentration of TZD).

Additionally, based upon the acute hydrochlorothiazide reduction of arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet and in NCC/pendrin double-knockout mice (Alshahrani et al., 2017a; above), TZD may have been predicted to reduce arterial pressure in patients with Gitelman’s syndrome. However, hydrochlorothiazide infusion into the forearm also did not reduce arterial pressure in the two Gitelman’s syndrome patients (as well as the normotensive and hypertensive subjects; Pickkers et al., 1998). The limited duration of hydrochlorothiazide infusion, i.e., 5 min, however, may have been insufficient to determine whether arterial pressure was actually reduced (Pickkers et al., 1998). Arterial pressure was also not reduced following diagnostic challenge with p.o. 50 mg of hydrochlorothiazide in Gitelman’s syndrome patients (L Chen and G Colussi, personal communications), with heart rate also remaining unchanged (G Colussi, personal communication). Lack of hydrochlorothiazide arterial pressure reduction in patients with Gitelman’s syndrome appeared independent of basal arterial pressure, as hydrochlorothiazide failed to reduce arterial pressure in Gitelman’s syndrome patients with low-normal mean arterial blood pressures, ranging at basal from 73 to 96 mmHg (n = 5; G Colussi, personal communication).

On the other hand, plasma volume depletion in patients with Gitelman’s syndrome may be insufficient to demonstrate TZD reduction of arterial pressure, consistent with the inability of TZD to reduce arterial pressure in NCC knockout mice with normal Na⁺ diet (Alshahrani et al., 2017a). However, significant plasma volume depletion likely occurs in Gitelman’s syndrome based on the elevated plasma renin levels/activity and aldosterone levels (Tsukamoto et al., 1995; Simon et al., 1996). Further, treatment of Gitelman’s syndrome, a condition associated with high Na⁺ intake, includes ad libitum dietary Na⁺ (Blanchard et al., 2017). Despite the increased Na⁺ intake, the intake apparently is insufficient to fully compensate for the decreased plasma volume (Diuresis and Plasma Volume) and, therefore the contribution of the decreased plasma volume to the hypotension.

Compensation of contracted plasma volume in Gitelman’s syndrome appears to occur through elevated aldosterone and renin and possibly other vasoconstrictor pathways (Simon et al., 1996). Hypotension associated with Gitelman’s syndrome, along with lack of hypertension despite increased renin activity and aldosterone plasma levels (Simon et al., 1996), suggest both an incomplete compensation and a prevention of increased constriction due to reduced vascular contractility (Calò et al., 1998, Calò et al., 1999; Calò, 2006). However, unaltered autonomic nervous system regulation of arterial blood pressure and absence of changes in basal and TZD-mediated forearm blood flow in patients with Gitelman’s syndrome (Pickkers et al., 1998; Sartori et al., 2007) do not appear to support a mechanism whereby up-regulation or down-regulation of these non-selective vascular inhibitory and excitatory signaling pathways, respectively, underlies hypotension and inhibition of anticipated elevated arterial pressure due to apparent activation of the renin–angiotensin II–aldosterone pathway (Calò et al., 1998, Calò et al., 1999; Calò, 2006).

Thus, it is tempting to conclude that selective inhibition of the renin–angiotensin II–aldosterone pathway underlies the compromised vascular constriction. This conclusion appears consistent with lower plasma renin and urine aldosterone levels as predictors of greater TZD efficacy of arterial pressure reduction (Chapman et al., 2002; Diuresis and Plasma Volume). A further speculation is that alterations in blood electrolytes and pH associated with Gitelman’s syndrome, i.e., hypomagnesemia, hypocalciuria, hypochloremia, and hypokalemic alkalosis ( Table 1l ), underlie possible compromised renin–angiotensin II–aldosterone pathway. However, it appears that these conditions do not decrease the formation of angiotensin II (Tsukamoto et al., 1995).

The absence of TZD reduction of arterial pressure in Gitelman’s syndrome is not supportive of an extrarenal TZD target suggested by TZD reduction of arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet and in NCC/pendrin double-knockout mice (Alshahrani et al., 2017a). However, this apparent lack of support could be explained if intravascular volume contraction is less in Gitelman’s syndrome than in knockout mice models (above).

An additional assessment of contracted plasma volume in response to TZD is determination of compensatory constrictor pathways. Moreover, TZD may inhibit the compensatory vasoconstriction, eliciting reduced vascular tone and resulting in only partial compensation of the arterial pressure reduction. Additionally, these same constrictor pathways may be activated in some types of hypertension. Thus, determination of the effects of TZD on these constrictor pathways is fundamental to the identification of the TZD extrarenal target. The involvement of constrictor pathways can be assessed in part through determinations of endogenous levels of humoral and neurogenic constrictor factors and effects of TZD on constriction by these factors. Major compensatory constrictor pathways include the renin–angiotensin II–aldosterone and autonomic nervous systems. Overall, however, the relevance of plasma renin activity and angiotensin II plasma levels is unclear due to the importance of intrarenal versus systemic levels of these factors (Reudelhuber, 2013).

TZD inhibition of the compensatory contracted plasma volume as well as constriction underlying certain types of hypertension may involve direct inhibition of vascular constriction. This assessment can be achieved in part through the effects of TZD on exogenous agonist pressor responses.

Numerous findings describe the effects of acute TZD on agonist-elevated and agonist-reduced vascular bed and mean arterial pressure ( Table 1q ). However, based upon the close correlation between acute TZD reduction of arterial pressure and diuresis, the extrarenal-mediated TZD reduction of arterial pressure is most likely absent during acute TZD (Diuresis and Plasma Volume). Thus, findings presented are limited to effects of chronic TZD on pressor responses ( Table 3 ).

Numerous studies investigated TZD inhibition of vascular contractility in vitro ( Tables 1r and 5 ). Moreover, a number of these studies, but not all, demonstrated TZD inhibition of constriction ( Tables 1s and 5 ). Proposed mechanisms underlying the inhibition of vascular contractility include carbonic anhydrase inhibition, increased nitric oxide release, BKCa²⁺ channel activation, decreased extracellular/intracellular Na⁺, decreased sensitization to intracellular Ca²⁺, RhoA and Rho kinase down-regulation, and inhibition of voltage-dependent Ca²⁺ channel activation ( Table 1s ).

However, i) supra-therapeutic TZD dose (Dose/Concentration of TZD) was required in a majority of studies to demonstrate inhibition of vasoconstriction ( Table 5 ). Even though rabbits challenged with hydrochlorothiazide for 7 weeks with the supra-therapeutic dose of 40 mg/kg/day failed to increase acetylcholine relaxation of isolated femoral artery, while incubation of rabbit isolated femoral artery with 100 µM hydrochlorothiazide enhanced acetylcholine relaxation (Sládková et al., 2007; Vascular Contractility In Vitro, In Vivo TZD on Vascular Contractility Determined In Vitro; Tables 4 and 5 ). Indeed, the requirement of supra-therapeutic concentrations in vitro is similar to findings in humans of TZD-increased forearm blood flow, which required at least 38 µM of hydrochlorothiazide (Pickkers and Hughes, 1995; Pickkers et al., 1998), and in dogs, which required 9–18 times the usual therapeutic dose of chlorothiazide (Overbeck and Haddy, 1960). Concentration–response (inhibition of constriction) curves to TZD were not performed in a number of investigations ( Table 5 ). Thus, TZD potency in these investigations remains to be established. ii) The magnitude of TZD inhibition of constriction depended significantly upon the TZD (Afsar et al., 2016). iii) Conclusions regarding mechanism underlying the inhibition apparently depended upon the contractile agent and/or vessel as observed, e.g., with the role of BKCa²⁺ channel inhibition (Calder et al., 1992, Calder et al., 1993, Calder et al., 1994; Colas et al., 2000a; Table 5 ). iv) In vitro effects may differ from in vivo effects. For example, the proposal that TZD decreased constriction through BKCa²⁺ channel inhibition based upon in vitro findings (Colas et al., 2000a; Table 5 ) is not supported by observations that i.p. paxilline, a BKCa²⁺ channel blocker, and a BKCa²⁺ channel/NCC/pendrin in triple-knockout mice did not prevent acute, hydrochlorothiazide reduction of arterial pressure (Alshahrani et al., 2017a). And v) vessels were not obtained from animals chronically treated with TZD ( Table 5 ).

Generally consistent with the dependency of TZD reduction of arterial pressure on altered regulatory protein expression is the minimal and greatly reduced arterial pressure with acute infusion and with 3-day TZD treatment, respectively, in hypertensive, angiotensin II-salt rats (Ballew and Fink, 2001). Additionally, TZD inhibition of vasoconstriction was attributed to decreased expression of proteins that sensitize agonist constriction, RhoA and Rho kinase (Zhu et al., 2005; Table 4 ). This conclusion was based in large part on decreased expression of these proteins with TZD challenge of cultured vascular smooth muscle cells (Zhu et al., 2005). Consistent with these findings is the more recent demonstration that chronic hydrochlorothiazide reduced arterial pressure in the hypertensive 1 kidney/DOCA-salt rat and decreased the associated elevated Rho-kinase (as well as gene expression of markers of remodeling) in the aorta (Araos et al., 2016).

On the other hand, decreased Rho-kinase would be expected to inhibit and enhance agonist constriction and relaxation, respectively (Nakamura et al., 2003). However, 1) chronic TZD failed to consistently inhibit constriction and relaxation was not enhanced (Constrictor Pathways In Vivo, Exogenous Agonist; Vascular Contractility In Vitro, In Vivo TZD on Vascular Contractility Determined In Vitro; Tables 3 and 4 ); 2) while Rho-kinase inhibition reduces arterial pressure in rats with normal arterial pressure (Dhaliwal et al., 2007), chronic (and acute) TZD do not reduce arterial pressure in rat and other animals and in human (Introduction; Diuresis and Plasma Volume; Table 1j and d, Figure 1 ; unreported in Araos et al. (2016) is whether hydrochlorothiazide decreased mean arterial pressure in sham operated rats); 3) although hydrochlorothiazide in vitro at clinically relevant concentrations decreased Rho-kinase expression and agonist constriction, based upon the varied inhibitory effects of TZD on vasoconstriction in vitro as well as the use in many studies of supra-clinical TZD concentrations (Vascular Contractility In Vitro, In Vitro TZD on Vascular Contractility; Tables 1r, s , and 5 ), it is should be considered that inhibition of Rho-kinase activity by TZD is indirect and requires chronic exposure (Araos et al., 2016). In contrast, the Rho-kinase inhibitor, fasudil (Y27632), causes direct inhibition (Nakamura et al., 2003) and 4) intrarenal rather than systemic angiotensin II may be the important modulator of vascular contractility (Reudelhuber, 2013; Constrictor Pathways In Vivo, Endogenous Activation). Thus, the possible role of decreased Rho-kinase activity in chronic TZD reduction of arterial pressure remains unclear.

TZD injection into the hypothalamus reduced arterial pressure in spontaneously hypertensive rats (Bergmann and Altshuler, 1996). However, in hypertensive patients, TZD was neutral with respect to central systolic pressure (Morgan et al., 2004; Mackenzie et al., 2009; Kwon et al., 2013; Manisty and Hughes, 2013).

Direct TZD application to the carotid artery sinus in the presence and absence of norepinephrine and epinephrine failed to reduce arterial pressure in dogs with acute hypertension due to vagi-aortic nerve denervation (Preziosi et al., 1959; Preziosi et al., 1961). Concomitant with the findings related to baroreceptor reflexes (Preziosi et al., 1959; Preziosi et al., 1961), while TZD increased the gain of renal nerve activity and heart rate in spontaneously hypertensive rats, the increase was non-selective in that similar findings were observed with several classes of anti-hypertensive agents (Kumagai et al., 1996).

TZD did not lower arterial pressure when infused into the carotid artery in a dog preparation with ligated internal and external carotid arteries with and without carotid sinus denervation (Preziosi et al., 1961). Consistent with these findings TZD were without effect on isolated guinea pig and rabbit heart (Preziosi et al., 1959).

It is unlikely that TZD act centrally at vasomotor centers to decrease peripheral vascular resistance. Further, the relevance of determinations of TZD effects in normotensive animals is unclear because TZD do not lower arterial pressure in normotensive humans and animals (Diuresis and Plasma Volume).

In terms of baseline protein expression in responders and nonresponders, responders had greater mRNA baseline levels of a number of proteins including vasodilator-stimulated phosphoprotein (VASP) and metabolites of the sphingolipid metabolic pathway (Turner et al., 2013; Shahin et al., 2017; Sá et al., 2018). It was suggested that increased baseline VASP expression and the sphingolipid metabolic pathway were responsible for TZD efficacy of arterial blood pressure reduction (Shahin et al., 2017).

The involvement of these pathways in TZD reduction of arterial pressure warrants additional investigation. Indeed, it remains unclear whereby, e.g., increased sphingolipid metabolism would result in greater efficacy of TZD reduction of arterial pressure (Shahin et al., 2017).

It is unlikely that TZD selectively inhibit the presumably different TZD targets associated with each of the multiple types of hypertension sensitive to reduction by TZD ( Table 1c ; Introduction). For example, hypertension varies with respect to the magnitudes of sympathetic nerve activity and vascular contractile sensitivity ( Table 1t ). Thus, as illustrated in the working model ( Figure 2 ), TZD reduction of arterial pressure is unrelated to direct inhibition of a specific constrictor pathway but rather through a common effect that inhibits constriction to multiple pathways. These multiple pathways underlie constriction (1) to TZD-decreased plasma volume (consistent with earlier proposals of Tapia et al., 1957; Freis et al., 1958; Dustan et al., 1959; Tarazi et al., 1970) and (2) responsible for the numerous TZD-sensitive types of hypertension ( Table 1c ; Introduction). Alternatively, a particular pathway sensitive to TZD inhibition, e.g., renin–angiotensin II, could underlie the multiple types of hypertension associated with responders.

In any case, the above proposal suggests that TZD reduction of arterial pressure is mediated by inhibition of underlying compensatory vasoconstrictor pathways ( Figure 2 ). Indeed, removal of the compensatory constriction, through repletion of plasma volume by infusion with dextran plus saline/glucose or elevated dietary Na⁺, reversed TZD chronic reduction of arterial pressure (Freis et al., 1958; Wilkins et al., 1958; Wilson and Freis, 1959; Winer, 1961). Also consistent with a common TZD target is enhancement of agonist constriction with myogenic tone (Faber and Meininger, 1990).

It should also be noted that, despite the potential attractiveness of the suggestion that NCC serves as the extrarenal target underlying TZD chronic arterial pressure reduction based upon the lower arterial pressure in Gitelman’s syndrome (Calò, 2006), this suggestion is inconsistent with restriction of NCC expression to the kidney, i.e., as evidenced by lack of NCC expression and binding in the aorta, as well as the adrenal gland, brain, heart, intestine (large and small), liver, lung, pancreas, placenta, salivary gland, skeletal muscle, spleen, and testis in humans and/or rats (Beaumont et al., 1988; Hebert and Gamba, 1995; Mastroianni et al., 1996).

As illustrated in the working model of Figures 2A–D :

(A) Normotensive subjects/animals: The proposed dependency of TZD arterial pressure reduction on inhibition of compensatory vasoconstrictor pathways (Guiding Parameters (a) Therapeutic TZD dose/concentration) suggests that the absence of TZD reduction of arterial pressure in normotensive subjects and animals (Diuresis and Plasma Volume) is due to complete mitigation by compensatory vasoconstriction of the TZD inhibition of constriction ( Figure 2A ).

(B) Decreased plasma volume: In contrast to the lack of TZD reduction of arterial pressure in normotensive subjects/animals, TZD reduced arterial pressure following enhanced plasma volume depletion, as observed in NCC knockout mice subjected to Na⁺-restricted diet and NCC/pendrin knockout mice (Alshahrani et al., 2017a). Thus, even greater compensatory vasoconstriction is required to limit the decrease in arterial pressure (Alshahrani et al., 2017a). However, a further decrease in plasma volume due to TZD cannot be fully compensated for by the already enhanced magnitude of compensatory vasoconstriction. The lack of complete mitigation of the plasma volume reduction, therefore, fails to fully compensate for TZD inhibition of vasoconstriction ( Figure 2B ). A caveat to this explanation is the possible mediation of TZD reduction of arterial pressure in NCC knockout mice subjected to Na⁺-restricted diet and NCC/pendrin knockout mice (Alshahrani et al., 2017a) through NDCBE inhibition ([Na⁺/Cl⁻ Contransporter (NCC; SLC12A3)]).

(C) TZD-sensitive hypertension: Similar to TZD reduction in arterial pressure following enhanced plasma volume depletion (Alshahrani et al., 2017a), in TZD-sensitive hypertension, the magnitude of constriction due to hypertension plus decreased plasma volume in response to TZD cannot fully mitigate TZD inhibition of constriction ( Figure 2C ). Thus, TZD reduces the elevated arterial pressure ( Figure 2C ). This proposal further suggests, therefore, that reduction of arterial pressure in hypertension requires an underlying component of reduced plasma volume. In this regard, this suggestion is consistent with the reversal of the TZD lowered arterial pressure with plasma volume repletion (Diuresis and Plasma Volume). It should also be considered, however, that a particular pathway sensitive to TZD inhibition, e.g., renin–angiotensin II, could underlie the multiple types of hypertension associated with responders (Guiding Parameters (a) Therapeutic TZD dose/concentration).

(D) TZD-insensitive hypertension: Considering the suggested requirement for an underlying component of reduced plasma volume to lower elevated arterial pressure (C above), the inability to decrease arterial pressure in some types of hypertension may be due to prevention of contracted plasma volume ( Figure 2D ). The lack of contracted plasma volume is consistent with the greater TZD-increased diuresis, plasma volume reduction, and body weight loss in responders than nonresponders and the inability of TZD to decrease plasma volume in nonresponders (de Carvalho et al., 1977; Svendsen et al., 1983; Freis et al., 1988; Table 2 ; Diuresis and Plasma Volume). Alternatively, the constrictor pathways underlying TZD-insensitive hypertension may overcome the TZD inhibition of constriction ( Figure 2D ).

Clearly, the identity of the extrarenal site of TZD inhibition of vascular tone remains for further investigation. Moreover, this site is presumably active in Gitelman’s syndrome. TZD-insensitive hypertensive models should also assist in this identification through delineation of factors in common with, and different from, TZD-sensitive hypertensive models and unique to particular models. Additionally, several genetic signals associated with sensitivity to TZD reduction of arterial pressure have been identified (Melander et al., 2013; Shahin and Johnson, 2016; Sá et al., 2018). Whether these or other signals underlie TZD inhibition of vasoconstriction remains for further investigation. Identification of the TZD extrarenal target will provide the basis for conversion of nonresponder to responder and the development of novel antihypertensive therapies acting through this site.