Blood-borne sphingosine 1-phosphate maintains vascular resistance, blood pressure, and cardiac function in mice

Significance

Contraction of resistance arteries in peripheral tissues maintains blood pressure (BP) and facilitates cardiac output distribution according to local metabolic demands. Sphingosine 1-phosphate (S1P) is a signaling lipid that is released from red blood cells and that circulates in plasma bound to high-density lipoproteins and albumin. We report that mice lacking blood-borne S1P have reduced peripheral vascular resistance, BP, and cardiac contractile function but preserved cardiac output. Additional observations indicate that the pool of circulating S1P that is bound to albumin plays an essential role in maintaining vascular resistance through activation of vascular smooth muscle cell S1P receptors. These insights may lead to therapeutic strategies for BP optimization and heart failure.

Abstract

Sphingosine 1-phosphate (S1P) is a bioactive lipid that circulates in plasma bound to high-density lipoproteins (HDL) and albumin. Circulating S1P levels correlate positively with systolic blood pressure (BP) in hypertension and negatively with severity in septic shock and with left ventricular function in heart disease. In mice, isolated deficiency in HDL-S1P and endothelial cell S1P receptor (R)-1 both trigger hypertension, supporting an essential role for HDL-S1P in endothelial function. Physiological roles of albumin-S1P and myocyte S1PRs in the cardiovascular system remain incompletely defined. We report that mice lacking all circulating S1P pools display hypotension and lack of BP increase with age, which contrasts with HDL-S1P deficiency and suggests an essential role for albumin-S1P in cardiovascular homeostasis. Although cardiac output was preserved in a basal state, left ventricular systolic function and contractile reserve were reduced in the absence of circulating S1P. Cardiac function and BP were partially or fully normalized by transfusion of erythrocytes capable of S1P production. Hypotension was accompanied by reduced peripheral resistance, and albumin-S1P, but not S1P complexed to an HDL-like chaperone, dose-dependently increased vascular resistance in isolated perfused kidneys via S1PR3 and S1PR2. Epistatic analysis supported a critical role for S1PR3 in S1P-dependent BP maintenance and pointed to a distinct origin of the cardiac phenotype. We thus uncover an essential role for circulating S1P in maintaining BP and left ventricular systolic function in mice. Our results also highlight distinct functions for the pools of S1P bound to HDL and to albumin, carrying both diagnostic and therapeutic implications.

Vascular resistance generated by small arteries and arterioles is essential for blood pressure (BP) maintenance and for the distribution of cardiac output (CO). Opening and closing of resistance arteries directs blood flow according to metabolic demand. Impaired control of peripheral resistance is a hallmark of cardiovascular disease that can lead to hypoperfusion and, in severe cases, hypoxia and loss of organ function (1). A systemic drop in resistance can cause distributive shock during infections and allergic reactions (2, 3), while elevated resistance can contribute to hypertension, a major risk factor for stroke, myocardial infarction, and dementia (4). G protein–coupled receptors (GPCR) are key regulators of vascular resistance and therapeutic targets for BP control (5). Experimental findings in mice lacking specific G proteins highlight essential roles for GPCR signaling in endothelial cells (EC) and vascular smooth muscle cells (VSMC) in modulating vascular resistance, BP, and age- and diet-induced hypertension (6–9). The GPCRs mediating these functions remain to be fully defined.

Sphingosine 1-phosphate (S1P) is a bioactive lipid that circulates in plasma bound to apolipoprotein M (ApoM)-containing HDL and to albumin, also known as S1P chaperones (10, 11). S1P is supplied to plasma by erythrocytes, with possible contribution by ECs (10, 12, 13) and by platelets upon activation (14). Circulating S1P is reduced in trauma and septic shock (11, 15), elevated with exercise training (11) and in hypertension (16, 17), and correlates inversely with left ventricular (LV) contractile function and outcome in heart failure (HF) (18–20). Whether these changes directly impact cardiovascular risk or disease progression remains unclear. S1P regulates vascular and cardiac function via GPCRs: it dilates and constrict resistance arteries via EC S1PR1 and VSMC S1PR2&3, respectively, and modulates cardiomyocyte contractility and heart rate via S1PR1-3 (20, 21). S1P also supports vascular integrity and promotes endothelial nitric oxide synthase (eNOS) activation and vasodilation to HDL through EC receptors (22). Mice in which S1pr1 (encoding S1PR1) is deleted in EC (S1pr1^ECKO) display vascular leak, reduced NO bioavailability, and hypertension (23, 24). Deficiency in ApoM and thus in S1P binding to HDL also triggers hypertension (25, 26), underscoring a critical role for the HDL–S1P–S1PR1 axis in endothelial function. Naïve S1PR2-deficient mice show reduced renal and mesenteric artery resistance (27), yet S1PR2 and S1PR3 deficient mice both maintain normal BP and cardiac function (23, 27). In disease models, S1P and its myocyte receptors provide essential support for the recovery from experimental anaphylactic shock (28–30) and myocardial ischemia–reperfusion injury (20); S1P production and S1PR3 signaling contribute to hypertension induced by the glycoprotein sortilin (17); and vessel wall S1P production and S1PR2 signaling to increased myogenic tone in heart failure and diabetes (31–33). Thus, S1P provides essential support for endothelial function through S1PR1, and may also promote vascular resistance and cardiac function through S1PR2&3.

Similar BP elevations in EC S1PR1- and ApoM-deficient mice indicate that albumin-bound S1P cannot compensate for loss of HDL-S1P (25). Despite retaining ~50% of normal S1P levels, ApoM-null mice also exhibit vascular leak (11, 34). This may indicate that HDL receptors such as SR-BI are important for presenting S1P to S1PR1 (35). It also questions if albumin-S1P, which has a short plasma half-life, has physiological relevance or is mainly an intermediate in the redistribution and clearance of sphingolipids (36–38). Besides the poorly defined role for albumin-S1P, it is unclear to what extent VSMC and cardiomyocyte S1PRs are engaged by plasma or locally produced S1P.

We report that mice lacking all plasma S1P pools (plasma S1Pless mice) present low BP and compensated LV dysfunction. This contrasts with ApoM deficiency and points to an essential function for albumin-bound S1P. Bone marrow transplantation and erythrocyte transfusion experiments show that blood-borne S1P provides essential support of vascular resistance and LV contractile function. Analysis of receptor dependence indicates a dominant role for S1PR3 in arteriolar tone and BP maintenance, with distinct mechanisms underlying the cardiac phenotype.

Results

S1P Maintains BP and Contributes to Its Increase With Age in Mice.

S1P is synthesized from sphingosine, a ceramide metabolite, by sphingosine kinases (SphKs) 1 & 2 (encoded by Sphk1&2), and can be exported to blood via Spns2 and Mfsd2b (10, 39). To assess the net role of circulating S1P in BP regulation, we measured BP in mice lacking both HDL- and albumin-associated S1P generated by postnatal deletion of Sphk1&2 in S1P-exporting cells with Mx1Cre [Sphk1^f/f(−):2^−/−:Mx1Cre⁺] (12). Contrasting mice with HDL-selective S1P deficiency (25, 26), systolic (S), diastolic (D), and mean (M) BPs in adult male plasma S1Pless mice were 15 to 25 mmHg lower than littermate controls (Sphk1^f/+:2^−/−:Mx1Cre⁺) as assessed under anesthesia with aortic BP probes (Fig. 1A). A similar BP reduction was observed in male and female plasma S1Pless mice generated with conditional knockout alleles for both Sphks (Sphk1^f/−:2^f/−:Mx1Cre^+/−) to control for global Sphk2 deficiency (13, 29) (Fig. 1 B and C). Both strategies yield plasma S1P levels <5% of normal in males and females (12, 13, 29, 39). Lower BPs were confirmed in male mice by noninvasive tail-cuff plethysmography (Fig. 1D) and by twenty-four-hour telemetry recordings (Fig. 1E), which also revealed a blunted nocturnal BP increase (Fig. 1E and SI Appendix, Fig. S1A). S1P deficiency did not impact heart rates (HR, Fig. 1 A–C, and E). Intriguingly, SBP increased with age in controls, but not in plasma S1Pless mice (Fig. 1F, telemetry; SI Appendix, Fig. S1B, tail-cuff). Thus, S1P maintains BP in both sexes and contributes, either directly or indirectly, to its increase with age.

Fig. 1.

Plasma S1Pless mice have lower BP than age and sex-matched littermate controls. (A–C). Invasive central BP and HR measurements of anesthetized Sphk1^f/−:2^−/−:Mx1Cre⁺ (A) (n = 6) and Sphk1^f/−:2^f/−:Mx1Cre⁺ (B) (n = 11) males and Sphk1^f/−:2^f/−:Mx1Cre⁺ female (C) (n = 5) plasma S1Pless mice and their respective age and sex-matched littermate controls (n = 5,11,9). (D) Noninvasive tail cuff plethysmography-based SBP measurements of Sphk^f/−:2^f/−:Mx1Cre⁺ (n = 18) plasma S1Pless and littermate control males (n = 20). (E) Dark and light cycles invasive radiotelemetry-based central BP and HR measurements of Sphk1 ^f/(f)−:2^−/−:Mx1Cre+ (n = 18) plasma S1Pless and littermate control males (n = 18). (F) Correlation between age and SBP in plasma S1Pless mice and littermate controls presented in E. Each symbol represents one mouse. Mean + SEM shown. Statistical analysis was performed using the unpaired t test or Mann–Whitney; two-way ANOVA or Kruskal–Wallis followed by Sidak’s multiple comparison and Pearson correlation.

Blood Cell-Derived Circulating S1P Maintains BP in Mice.

The Mx1Cre driver used to generate plasma S1Pless mice provides efficient excision in the hematopoietic compartment, the main source of plasma S1P (12, 13), but may also target other cell types involved in BP regulation. Most importantly, loss of S1P production by lymphatic endothelial cells (LEC) in this model impairs lymphocyte trafficking, reduces peripheral blood white cell counts, and may impair lymphatic vascular development (12, 13, 40, 41). To define the Mx1Cre-sensitive sources of S1P that maintain BP, we measured BP in >5-mo-old anesthetized male mice with selective deletion of Sphk1&2 in EC (PdgfbCreERT2), VSMC and cardiomyocytes (Sm22Cre) or LEC (Lyve1Cre) with Cre drivers validated for excision efficiency [(13); SI Appendix, Fig. S1C]. Disabling S1P production in these cells did not reduce BP (Fig. 2A and SI Appendix, Fig. S2 A and B). To address sufficiency of hematopoietic S1P sources, we replaced SphK-deficient bone marrow cells (BMC) of irradiated S1Pless adults with wild-type BMC to restore S1P production in plasma but not lymph (12, 13, 29). BPs measured three months later were comparable to littermate controls (Fig. 2B). Plasma S1Pless mice had normal hematocrits and both LEC-selective Sphk knockouts and Sphk1^f/−:2^f/−:Mx1Cre⁺ mice with wild-type BMC showed leukopenia similar to plasma S1Pless mice, arguing against blood viscosity effects (SI Appendix, Fig. S2C). Normal BP was reported in mice lacking S1P production in erythrocytes (42); selective impairment in platelets (Pf4Cre) (29, 43) or myeloid cells (LysMCre) (13) also did not impact BP (SI Appendix, Fig. S2 D and E), indicating redundant hematopoietic S1P sources. Substantiating the above observations, plasma S1Pless mice generated by hematopoietic cell selective Sphk deletion (Sphk1^f/−:2^f/−:Vav1Cre⁺)(13) or by transplantation of irradiated wild-type mice with S1P-deficient BMC (Sphk1^f/−:2^f/−:Mx1Cre⁺)(13) had lower BP than their respective controls (SI Appendix, Fig. S2 F and G), and transfusion of erythrocytes from wild-type donors normalized SBP in plasma S1Pless mice (Fig. 2C). Plasma S1P levels 5 d posttransfusion were fivefold pretransfusion levels (Fig. 2C). Measurements at 48 h in a subsequent experiment (below) indicated full restoration of S1P levels at the time of SBP normalization. Thus, hematopoietic cell-derived S1P is essential and erythrocyte S1P sufficient for BP maintenance in naïve mice.

Fig. 2.

Erythrocyte-derived circulating S1P maintains BP. (A and B) Invasive central BP and HR measurements of anesthetized male mice lacking S1P production in lymphatic endothelial cells and perivascular macrophages (Sphk1^f/f:2^−/−:Lyve1Cre⁺; n = 5; A), nonhematopoietic cell targets of Mx1Cre recombinase (Sphk1^f/−:2^f/−:Mx1Cre⁺ mice transplanted with wild-type bone marrow cells (BMC); n = 5; B) and their respective littermate controls (Lyve1Cre⁻, n = 6; Sphk1^f/−:2^f/−:Mx1Cre⁻, n = 5). (C) Left: Experimental timeline of erythrocyte transfusion experiments. Middle: Tail-cuff based SBP measurements in plasma S1Pless (Sphk1^f/−:2^f/−:Mx1Cre⁺; n = 5) and control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 6) males pre- and posterythrocyte transfusion. Right: Analysis of plasma S1P levels two weeks before and 5 d post erythrocyte transfusion. Each symbol represents one mouse in bar graphs; mean + SEM shown. Statistical analysis was performed using unpaired t test (A and B), one-way or two-way ANOVA followed by Sidak’s multiple comparison (C).

S1P Maintains BP By Supporting Peripheral Vascular Resistance Via S1PR2&3.

Considering the established role for S1PR1 in NO-dependent BP regulation (23), we next addressed if SphK deficiency in hematopoietic cells enhances S1PR1-mediated eNOS activation and vasodilation by increasing the supply of sphingosine to vessel wall SphKs, thereby reducing BP. Inhibition of eNOS exacerbated rather than eliminated the difference in BP between plasma S1Pless and controls (Fig. 3A), arguing that plasma S1Pless mice have reduced rather than increased NO bioavailability. This is consistent with loss of HDL-S1P-mediated engagement of EC S1PR1 (25), but argues that the effect on BP is masked when also depleting albumin-S1P. We therefore addressed if circulating S1P provides continuous support for vascular tone and BP via VSMC S1PRs. Alexa647-albumin accumulated both in the endothelium and in Lyve1+ macrophages outside the VSMC layer of mesenteric arteries 60 min after i.v. administration to wild-type mice (SI Appendix, Fig. S3A), suggesting that albumin-bound plasma S1P may access VSMC receptors. Supporting this notion, Echo Doppler analysis revealed elevated renal and mesentery artery blood flow velocities and reduced total peripheral resistance in plasma S1Pless mice (Fig. 3B). To study mechanisms of S1P-induced vascular resistance ex vivo, we employed a perfusion system that measures renal vascular resistance established by interactions between myogenic tone and flow-mediated dilation (Fig. 3C). Infusion of BSA-S1P into phenylephrine-preconstricted wild-type kidneys caused a dose-dependent resistance increase starting in the low nanomolar range (Fig. 3D). The response was blunted in S1PR2- and in S1PR3-deficient kidneys, and abolished in kidneys from double knockouts (Fig. 3E). Vasoconstriction to albumin-S1P in S1pr1^ECKO kidneys was comparable to controls, indicating minimal S1PR1-dependent eNOS activation despite validated endothelial function in perfused kidneys (Fig. 3F, Materials and Methods). Thus, intravascular albumin-S1P triggers a robust S1PR2/3-mediated constrictor response at plasma concentrations. As predicted (23, 27), and consistent with receptor redundancy, mice with isolated S1PR2 deficiency in VSMC and cardiomyocytes (S1pr2^f/f Sm22Cre; SI Appendix, Fig. S3B) or S1PR3 deficiency in all cells (SI Appendix, Fig. S3C) nevertheless had normal BP. HRs were increased in S1PR3 knockouts (SI Appendix, Fig. S3C), suggesting possible functional compensation. A fivefold increase in renal S1pr2 mRNA abundance in S1PR3 knockouts relative to controls also indicated genetic compensation (SI Appendix, Fig. S3D). While we were able to study renal resistance in rare S1PR2/3 double knockout survivors, high embryonic and postnatal lethality precluded BP measurements (44, 45). To address roles of S1PR2&3 downstream of S1P, we instead crossed Sphk knockout alleles into S1pr2 or S1pr3 knockout backgrounds and induced Sphk deletion postnatally. Plasma S1P deficiency lowered BPs in S1PR2-deficient males and females (Fig. 3G and SI Appendix, Fig. S3E) but not in S1PR3-deficient mice of either sex (Fig. 3H and SI Appendix, Fig. S3F). Collectively, these data argue a critical role for circulating S1P in maintaining vascular resistance and BP through partially redundant actions of S1PR2&3.

Fig. 3.

S1P regulates BP through S1PR2/3-mediated control of vascular tone. (A) Telemetry-based delta SBP before and during 5 d of L-NAME treatment (1 g/L in the drinking water) of plasma S1Pless (Sphk1^f/f:2^−/−:Mx1Cre⁺) and littermate control (Sphk1^f/f:2^−/−:Mx1Cre⁻) males. (B) Mean blood flow velocities (mBFVs) measured by ultrasound in right renal artery (RRA) and superior mesentery artery (SMA) of plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 10) and littermate control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 9) males and total peripheral resistance (TPR; mean arterial pressure/cardiac output) in a subset of these males (n = 5 to 6). Left panel: Representative Doppler images. (C) Schematic illustration of the kidney perfusion system used in D–F. (D) Normalized delta pressure with increasing concentrations of lipid free BSA or S1P-BSA infused into perfused wild-type kidneys. (E) Normalized delta pressure with increasing concentrations of S1P-BSA in kidneys from S1pr2 deficient (KO), S1pr3 deficient (KO), S1pr2/3 deficient (dKO), and shared littermate control (Ct) mice. (F) Normalized delta pressure with increasing concentrations of S1P-BSA infusion into S1pr1^ECKO (S1pr1^f/f:PdgfbCre⁺) and littermate control (S1pr1^f/f:PdgfbCre^-) kidneys. (G) Arterial pressures of anesthetized plasma S1Pless males generated in an S1PR2 deficient background (Sphk1^f/f:2^−/−: Mx1Cre⁺:S1pr2^−/−; n = 10) and of their respective littermate controls (Sphk1^f/f:2^−/−:Mx1Cre⁻:S1pr2^−/−; n = 8). (H) Arterial pressure of anesthetized plasma S1Pless males generated in an S1PR3 deficient background (Sphk1^f/f:2^−/−: Mx1Cre⁺:S1pr3^−/−; n = 9) and of their respective littermate controls (Sphk1^f/f:2^−/−:Mx1Cre⁻:S1pr3^−/−; n = 13). Each symbol represents one mouse. Mean + SEM shown. Statistical analysis was performed using unpaired t test (B, G, and H) and two-way ANOVA (A, and D–F).

Plasma Chaperones Dictate Vascular Responses to S1P.

The contrast between hypotension in mice lacking all plasma S1P and hypertension in mice lacking HDL-S1P only suggests a distinct role for albumin-S1P, potentially due to chaperone-dependent differences in S1PR engagement. We tested the capacity of S1P bound to recombinant ApoM (ApoM-Fc) (25) and to a recombinant ApoM-ApoA1 fusion protein (A1M) (46) to increase renal vascular resistance ex vivo relative to ApoM incapable of S1P binding [ApoM-TM; (25)], unloaded A1M (46), albumin-S1P, and unloaded albumin. While ApoM-Fc-S1P and albumin-S1P both increased resistance, A1M-S1P and negative controls did not (Fig. 4). As predicted (46), the preparations of albumin-S1P, ApoM-Fc-S1P, and A1M-S1P were equipotent at inducing AKT phosphorylation in cultured EC (SI Appendix, Fig. S4). Thus, while promoting EC signaling (46), ApoA1-mediated HDL binding to EC receptors may also reduce VSMC access and signaling. This aligns with BP observations and suggests that VSMC S1PRs are preferentially engaged by the albumin-associated S1P pool.

Fig. 4.

Plasma chaperones dictate vascular responses to S1P. Changes in renal vascular resistance (normalized delta pressure) upon infusion of wild-type murine kidneys with increasing concentrations of S1P associated with BSA, ApoM-Fc, ApoM-ApoA1 (A1M), ApoM incapable of S1P binding (ApoM-TM), S1P free BSA and S1P free A1M. Mean + SEM shown. Statistical significance was assessed by two-way ANOVA. Unless specifically indicated, P values denote comparison to respective control, i.e. BSA-S1P to unloaded BSA, ApoM-Fc-S1P to ApoM-TM and A1M-S1P to unloaded A1M.

Renal and Adrenal Function and RAAS Activity in Plasma S1Pless Mice.

The kidney plays a central role in BP regulation under control of the renin-angiotensin-aldosterone system (RAAS) (47). To address the impact of Mx1-Cre-mediated Sphk deletion on renal function and RAAS, we measured glomerular filtration rates (GFR) and blood and urine parameters and performed histological, immunohistochemical, and gene expression analyses of kidneys and adrenal glands from plasma S1Pless and control littermates. No difference was observed in expression of genes implicated in angiotensin responses and aldosterone production in adrenal glands, which also had normal architecture and abundance of aldosterone synthase expressing cells in the zona glomerulosa (Fig. 5A and SI Appendix, Fig. S5A). Accordingly, plasma aldosterone and corticosterone levels were normal (Fig. 5A). Renal architecture was also normal, with no difference in glomerular structure, juxtaglomerular pro-renin producing cells, or monocyte abundance (Fig. 5B and SI Appendix, Fig. S5 B and C). Expression of genes involved in renal responses to angiotensin/aldosterone and in electrolyte/water reabsorption were normal, except NKCC2, which was upregulated (SI Appendix, Fig. S5D). GFRs were normal (Fig. 5C), as were blood electrolytes, gases, creatine kinase (SI Appendix, Fig. S6A), urine output, electrolytes, creatinine and total protein (SI Appendix, Fig. S6B). A tendency to lower urinary sodium excretion aligned with NKCC2 upregulation (47). Intriguingly, despite normal pro-renin expression, equilibrium angiotensin peptides were increased in plasma S1Pless mice of both sexes (Fig. 5D and SI Appendix, Fig. S6C). Plasma renin activity was calculated to be higher, angiotensin-converting enzyme activity slightly elevated, and aldosterone/AngII ratios lower (Fig. 5 E and F and SI Appendix, Fig. S6C). Increased renin activity could be driven by renal autoregulation to maintain arteriolar tone and GFR. Accordingly, treatment with an angiotensin receptor blocker uncovered reduced GFR in plasma S1Pless mice (Fig. 5G) and further lowered SBP in plasma S1Pless mice relative to controls during both light and dark cycles (Fig. 5H). Thus, hypotension in plasma S1Pless mice is not caused by reduced RAAS activity or reduced sodium retention. Instead, RAAS activity and possibly sodium retention increase in the absence of plasma S1P, likely in attempt to normalize GFR and BP.

Fig. 5.

Impact of circulating S1P deficiency on renal and adrenal function. (A) Representative images of sections of the adrenal cortex from plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 4) and control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 4) mice stained for aldosterone synthase (AS; magenta) and cell nuclei (DAPI, blue). Right, quantification of AS expression and mass spectrometry–based determination of plasma aldosterone and corticosterone in plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 6) and littermate control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 6) males. (B) Assessment of renin expression in sections of kidneys from plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺) and littermate control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻) males. Left, representative images of sections stained for endothelial cell nuclei (Erg+, red), renin (green), and vascular smooth muscle cells (ASMA+; blue). Right, quantification of renin on ASMA+ area (n = 4). (C) Glomerular filtration rate (GFR) determined by the renal FITC-sinistrin clearance in S1Pless males (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 4) and females (n = 5) and their respective littermate controls (n = 4, n = 7). (D) Equilibrium angiotensin metabolite profile of plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 6) and littermate control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 6) males. (E) Renin activity (PRA-S), calculated as the sum of AngI+AngII, and ACE activity (ACE-S) as ratio between Ang II and Ang I. (F) Aldosterone/AngII ratios from data in A and D. (G) GFR after 24 h losartan (20 mg/kg) treatment in the same mice as in C. (H) Telemetry-based night and day delta SBP after three consecutive days of losartan (20 mg/kg i.p.) treatment of plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 6) and littermate control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 7) males. Mean+or-SEM shown, with each symbol representing one mouse. Statistical analysis was performed using the unpaired t test or Mann–Whitney, one-way or two-way ANOVA followed by Sidak’s multiple comparison test.

Circulating S1P Does Not Modulate BP Responses to Dietary Salt or Angiotensin II Challenge.

A role for circulating S1P in vascular resistance and the spontaneous increase in BP with age may suggest causality in the correlation between circulating S1P and experimental and human hypertension (16, 17, 26). To test the role of circulating S1P in renal and adrenal responses to dietary salt, we challenged plasma S1Pless mice and littermate controls with low, normal, and high sodium diets (SI Appendix, Fig. S7A) (48, 49). Three weeks of low or high sodium did not affect SBP, which remained lower in plasma S1Pless mice independent of diet (SI Appendix, Fig. S7B). Plasma sodium and calcium were unchanged (SI Appendix, Fig. S7C), and potassium slightly lower in plasma S1Pless mice under low sodium diet (SI Appendix, Fig. S7C). Other physiological parameters, including partial O2 and CO2 pressure, HCO3, total CO2, and base excess extracellular fluid, were normal (SI Appendix, Fig. S7 D and E). Urine aldosterone increased with low and decreased with high sodium diet, independent of genotype, and body weights and urinary outputs were similar (SI Appendix, Fig. S7F). To address if circulating S1P—which is elevated in AngII-induced experimental hypertension (26, 50)—plays a role in AngII-induced BP changes, basal pressure, and responses to AngII infusion (500 ng/kg/min) for 4 wk was recorded by telemetry (SI Appendix, Fig. S7G). The relative increase in SBP with AngII did not differ between genotypes. A similar relative increase in BP was also observed in an acute, high-dose AngII model (1,500 ng/kg/min) (SI Appendix, Fig. S7H). Accordingly, dose responses to AngII and other ex vivo functional measures were normal in mesenteric resistance arteries from plasma S1Pless mice (SI Appendix, Fig. S7I). Although it does not exclude a role for S1P produced by Mx1Cre-resistant cell types, this argues against a critical role for circulating S1P or S1P-dependent lymphocyte egress in AngII-induced hypertension. To further address causality between plasma S1P and hypertension, BP was measured in SphK2-deficient mice, in which defective clearance increases plasma S1P levels to approximately twice normal levels (28, 37), the high end of that reported in experimental and clinical hypertension (16, 17, 26, 50). There was no difference in SBP between SphK2-deficient mice and littermate controls (Fig. 6J). Thus, circulating S1P does not play an essential role in renal or adrenal responses to changes in dietary salt, nor in AngII-driven hypertension. Moreover, increasing plasma S1P is not alone sufficient to induce hypertension.

Fig. 6.

Reduced cardiac function in plasma S1Pless mice. (A) Echocardiographic assessment of left ventricular (LV) internal diameter in end diastole (LVIDd) and end systole (LVIDs), ejection fraction (LVEF), heart rate (HR), and cardiac output (CO) in plasma S1Pless (Sphk1^f/−:2^f/−:Mx1Cre⁺; n = 12 to 16) and littermate control (Sphk1^f/−:2^f/−:Mx1Cre⁻; n = 12 to 16) males. (B) Echocardiographic and Doppler assessment of HR, delta LVEF, and delta cardiac output (CO) before and 10 min after dobutamine (1,5ug/g, i.p.) administration to plasma S1Pless (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁺; n = 7) and control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 6) mice. (C) Echocardiographic assessment of LVEF and LV mass/BSA in irradiated wild-type male recipients of Sphk1^f/−:2^f/−: Mx1Cre⁺ (n = 12) or Sphk1^f/−:2^f/−:Mx1Cre⁻ (n = 8) BMC. (D) Echocardiographic assessment of LVEF, LVIDd, and LVIDs in plasma S1Pless (Sphk1^f/−:2^f/−:Mx1Cre⁺; n = 9) and control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 8) males pre- and 48 h posterythrocyte transfusion. (E) Echocardiographic and Doppler assessment of HR, delta EF, and delta CO before and 10 min after dobutamine (1,5ug/g, i.p.) administration plasma S1Pless (Sphk1^f/−:2^f/−:Mx1Cre⁺; n = 7) and control (Sphk1^f/−:Sphk2^f/−:Mx1Cre⁻; n = 7) males 48 h posterythrocyte transfusion. (F) Plasma S1P levels in Sphk1^f/−:2^f/−:Mx1Cre⁺ mice (n = 6) and littermate controls (Sphk1^f/−:2^f/−:Mx1Cre⁻; n = 3) after erythrocyte transfusion (harvested at 48 h). (G) Volcano plot showing log2FC (fold change) against −log10 (Adj. p-value) of transcripts identified by RNASeq analysis. Blue symbols indicate a selection of downregulated genes and red symbols a selection of upregulated genes. Each symbol represents one mouse or gene. Mean + SEM shown. Statistical analysis was performed using the unpaired t test, one- and two-way ANOVA followed by uncorrected Fisher’s LSD.

Plasma S1P Supports Cardiac Function and Reserve.

Reduced plasma ApoM and S1P levels and an inverse correlation between left ventricular ejection fraction (LVEF) and S1P levels in HF patients may suggest that plasma S1P supports cardiac function (18, 19). Baseline M-mode echocardiography revealed larger LV internal diameters (LVID) in both systole and diastole, and reduced LVEFs in plasma S1Pless males (Fig. 6A). Cardiac output (CO) was nevertheless maintained and HRs normal (Fig. 6A and SI Appendix, Fig. S8A), arguing against a cardiac origin of the BP phenotype. Septal and posterior wall thicknesses were unchanged, but LV mass slightly increased (SI Appendix, Fig. S8A), reflecting LV dilation. Females had normal cardiac function and mass (SI Appendix, Fig. S8B), further uncoupling cardiac and BP phenotypes. Reduced contractile function in male mice was paralleled by modest increases in fibrosis (SI Appendix, Fig. S8C), heart weight to tibia length ratios and cardiomyocyte size (SI Appendix, Fig. S8D). Thus, LV dilation compensates for systolic dysfunction to maintain LV stroke volume and sustain CO in plasma S1Pless males. Yet when challenged with a dobutamine stress test, they had a reduced capacity to increase LVEF and CO despite normal HR elevation, indicating reduced cardiac contractile reserve, a marker of subclinical LV dysfunction (Fig. 6B). LV dilation and contractile dysfunction were also observed in wild-type recipients of SphK1&2 deficient BMC (Fig. 6C), pointing to a role for hematopoietic cell-derived S1P also for this phenotype. Accordingly, Mx1Cre-mediated induction of an mTmG reporter showed little or no activity in VSMC or cardiomyocytes (SI Appendix, Fig. S8E). Moreover, transfusion of wild-type erythrocytes to 6- to 8-mo-old plasma S1Pless males restored plasma S1P levels, improved basal LV contractile function and normalized cardiac reserve 48 h later (Fig. 6 D−F), indicating that circulating S1P provides essential support for cardiac contractile function.

To address if the heart undergoes transcriptional reprogramming in the absence of circulating S1P, we performed bulk RNA-Seq and unbiased gene enrichment analysis on hearts from plasma S1Pless and littermate control males. Of 15341 genes with detectable expression, 1,040 were differentially expressed (DEGs). Of these, 95 showed a log2 [fold change (FC)] > 1, 91 of which were upregulated (Fig. 6G). Upregulated genes were enriched in gene ontology terms for immune responses (SI Appendix, Fig. S8F). Expression was only modestly enhanced (<3 fold) in all but 4 genes. Cross-referencing DEGs to scRNA-Seq databases and cell marker gene-sets indicated that the response could be accounted for by an increase in macrophages (SI Appendix, Fig. S8F). Increased expression of macrophage markers was confirmed by qPCR, persisted after erythrocyte transfusion, and was observed also in wild-type recipients of Sphk1&2 deficient BMC (SI Appendix, Fig. S9 A−C). Immunostaining of heart sections from the latter showed ~2× more mrc1/CD206+ tissue macrophages (SI Appendix, Fig. S9C). Thus, RNA-Seq analysis did not reveal transcriptional rewiring of cardiomyocytes in S1Pless mice, but an increase in cardiac macrophages that did not preclude a partial rescue of the phenotype by normalization of plasma S1P levels with S1P-producing erythrocytes.

Minimal structural and transcriptional consequences of S1P deficiency could suggest that S1P supports contractile function through S1PRs on cardiomyocytes, VSMCs, or ECs. Cardiac function was normal in mice with global S1PR3 deficiency (SI Appendix, Fig. S9D), S1PR2 deficiency in VSMC and cardiomyocytes (S1pr2^f/f:Sm22Cre⁺; SI Appendix, Fig. S9E) and S1PR1 deficiency in cardiomyocytes (SI Appendix, Fig. S9F). Unlike the BP phenotype, the cardiac phenotype induced by plasma S1P depletion was equally pronounced in the absence of S1PR3 (SI Appendix, Fig. S9G). Despite endothelial dysfunction (23, 24), S1pr1^ECKO mice had normal LV function and cardiac contractile reserve (SI Appendix, Fig. S9H). These observations argue against an essential role for any one S1P receptor in supporting cardiac contractile function, but do not exclude receptor redundancy (20).

Disturbances in the balance of circulating and cardiac sphingolipids could contribute to impaired cardiac function after Mx1Cre-mediated Sphk1&2 deletion. Plasma S1Pless mice have higher plasma sphingosine (13) and ceramide (SI Appendix, Fig. S10A) levels. However, no significant change was observed in wild-type recipients of Sphk1&2 deficient BMC (SI Appendix, Fig. S10A), which present a similar cardiac phenotype (Fig. 6C). A similar pattern was observed for cardiac sphingosine and ceramide (SI Appendix, Fig. S10B). Erythrocyte transfusion partially normalized plasma sphingosine and ceramide levels (SI Appendix, Fig. S10C). However, cardiac levels of sphingosine and ceramide remained high 48 h later (SI Appendix, Fig. S10D). Thus, although Mx1Cre-mediated Sphk1&2 gene deletion increases the levels of sphingosine and ceramide in plasma and heart, this increase is not attributable to gene deletion in the hematopoietic compartment and does not track with the cardiac nor the BP phenotype.

These observations reveal a role for circulating S1P in maintaining LV contractile function and cardiac reserve in mice, but do not pinpoint the underlying mechanism.

Discussion

We report that circulating S1P is required to maintain vascular tone, BP, and cardiac function in the optimal physiological range. In the context of published data, this indicates that HDL- and albumin-bound S1P play distinct roles in the optimization of vascular tone and BP. It further points to the diagnostic relevance of plasma S1P and its distribution between chaperones, and potential benefits of targeting VSMC S1P receptors for microvascular dysfunction and BP normalization.

Plasma S1Pless mice generated by different genetic strategies displayed hypotension and lack of BP elevation with age. Bone marrow transplantation, erythrocyte transfusion, and selective impairment of S1P synthesis in hematopoietic and vascular cells confirmed that the phenotype stemmed from plasma S1P loss. Although transfusion of wild-type erythrocytes was sufficient to restore BP, normal BP in mice with erythrocyte (42)—endothelial—or platelet-selective Sphk deficiency indicates that no single cellular S1P source is essential for BP maintenance. BP reduction was not explained by established roles for S1P in supporting endothelial function or lymphocyte trafficking. Analyses of vascular, renal, adrenal, and cardiac function instead indicated that low BP results from a loss of peripheral vascular resistance. This was supported by ex vivo studies demonstrating the capacity of S1P to increase renal vascular resistance dependent on S1PR2&3, genetic compensation by upregulation of S1pr2 in S1PR3 deficient kidneys, and elimination of the BP-lowering effect of S1P deficiency in the absence of S1PR3, which scRNA-Seq studies have revealed to be the principal VSMC S1P receptor (51).

Plasma S1P deficiency was associated with elevated plasma renin activity, increased renal NKCC2 expression and a tendency to enhanced sodium reabsorption, while other renal, adrenal, and RAAS parameters remained normal. Because renin activation and sodium retention increase BP (47), these findings likely reflect compensatory responses to maintain BP and GFR upon loss of S1P-mediated vasoconstriction, particularly in S1P-sensitive glomerular arterioles (52). Supporting this notion, AT1R blockade further lowered BP and exposed reduced GFR in plasma S1Pless mice. NKCC2 upregulation, if validated at the activity level, may disturb circadian changes in salt excretion (53) and thus help explain blunted diurnal changes in BP in these mice. Single-cell RNA-Seq of murine and human kidneys shows S1pr1 expression in EC and immune cells, S1pr3 in VSMC and potentially other cells expressing VSMC markers and renin, S1pr2 at low levels in the same cells, S1pr4&5 in immune cells, and little or no S1PR expression in tubular epithelium (54, 55). Since EC S1pr1 deletion raises BP (23) and does not alter renal vascular resistance to S1P ex vivo (Fig. 3F), and since renin activity is increased rather than decreased, we suggest that VSMC S1PR3 is the main renal target of plasma S1P implicated in BP elevation.

Plasma S1P deficiency reduced LV contractility and inotropic reserve in male mice, which were partially or fully restored by erythrocyte transfusion. This indicates that circulating S1P also provides trophic support for LV contractile function, supporting clinical studies linking low plasma S1P and ApoM to impaired LV function and poor heart failure outcomes (18, 19). Normal LV function in females, preserved resting cardiac output in males and persistent LV dysfunction in an S1pr3 null background all argue that heart defects do not account for reduced BP. The mechanistic basis for the cardiac phenotype and male-specific sensitivity to plasma S1P loss remain to be determined. S1pr1^ECKO mice showed normal LV function and inotropic reserve despite vascular integrity and hypercapnia-induced hyperemia impairments similar to plasma S1Pless mice (13, 29), arguing against a role for endothelial dysfunction. LV performance was also normal in mice with isolated deficiency of S1PR1, 2, or 3 in cardiomyocytes or all cells. As roles for cardiomyocyte S1PR2&3 in I/R injury were only revealed in double knockouts (20), compound cardiomyocyte-specific deletions are currently being investigated to address redundancy and genetic compensation. Despite equally efficient depletion of plasma S1P, females exhibited normal LV function, less profound BP reduction and greater renin compensation and show less vascular leak than males in preliminary analysis, suggesting that sex hormone-dependent mechanisms may render them less sensitive to S1P loss. Sex does not influence total plasma S1P in humans, yet ApoM levels rise in women post menopause, suggesting a potential relative increase in endothelium protective HDL-S1P in older women (11, 56).

S1P circulates in plasma bound to HDL, albumin, and other minor carriers (10, 11). HDL-S1P has received most attention because of the role of S1P in HDL-mediated vascular protection, therapeutic effects of ApoM-S1P in experimental models, and S1PR1-deficiency-like vascular phenotypes in ApoM-deficient mice (11, 22, 25, 35). Yet our results indicate that it is the remaining plasma S1P—accounting for ~40% of total and mainly albumin bound—that maintains normal BP homeostasis. The contrast between hypotension observed in mice lacking all plasma S1P in this study and hypertension reported with isolated HDL-S1P deficiency (25, 26), as well as the capacity of S1P to promote renal vascular resistance when bound to albumin but not to an HDL-like fusion protein, support distinct roles for the two S1P pools. While HDL may stabilize S1P and promote its presentation to EC S1PR1 via SR-BI (35), this could also restrict access to VSMC receptors, whereas albumin is known to cross the endothelium via caveolae-mediated transport (57). S1PR2/3-dependent vasoconstriction to albumin-S1P in the perfused kidney indicates that S1P can also cross the endothelium, although this remains to be directly demonstrated. Thus, while HDL-S1P supports endothelial function, albumin-S1P appears to be more important for vascular contraction.

Correction of acute and chronic hypotension represents an important clinical challenge with few effective therapies. In sepsis, plasma S1P levels decrease and correlate inversely with severity (15, 29, 58). S1P levels also drop rapidly in experimental anaphylaxis, and S1P deficiency exacerbates vasoplegia and sensitizes to mortality (29). Recent studies of the distribution of S1P between carriers in trauma and sepsis showed that while albumin-S1P drops sharply, there is a concomitant relative increase in HDL-S1P (15, 58). We speculate that this could shift S1P signaling from contractile to dilatory, worsening outcome, and that albumin-S1P/ApoM-S1P ratios correlate with BP. Plasma albumin depletion in chronic kidney disease is a risk factor for adverse outcome that is aggravated by hemodialysis and may contribute to intradialytic hypotension (59). Albumin infusion has shown benefit for severe sepsis, where it remains controversial (60), and for intradialytic hypotension (59). We speculate that S1P enrichment could improve benefit. In mice, albumin-S1P infusion accelerates BP normalization in anaphylaxis models, while S1PR2/3 deficiencies sensitize (28, 29, 39), and inhibition of S1P lyase and S1PR3 agonism both improve survival in experimental sepsis (61). In addition to normalizing albumin-S1P, this could suggest benefit of inhibiting S1P degradation or activating S1PR2/3, yet both carry risk of fibrosis (62).

A role for S1P signaling in BP maintenance independent of RAAS may also suggest diagnostic and therapeutic potential in hypertension. Assessment of S1P distribution between carriers may clarify S1P-BP correlations (16, 17). Enhancing S1PR1 activation with ApoM-Fc-S1P provides prolonged BP lowering in experimental hypertension (25); SphK1 and VSMC S1PRs represent potential targets to reduce VSMC S1P signaling. Although S1PR3 dominates S1P-mediated vascular tone under homeostasis, S1PR2 may contribute more in metabolic and cardiovascular disease (31,32, 33, 63), requiring dual targeting. Vesicular transport that carries albumin across the endothelium increases both in experimental hypertension (64) and aging (65), potentially explaining more pronounced BP-lowering effects of S1P deficiency with age. Although SphK2-deficient mice show normal BP despite elevated plasma S1P, it is possible that higher plasma S1P contributes to BP elevation when endothelial transcytosis is also increased. Induced S1P synthesis in the vessel wall could also enhance VSMC S1PR signaling to increase vascular resistance in hypertension (17, 66).

By signaling through S1PR1 on ECs, plasma S1P supports flow-mediated vasodilation (23). Through S1PR2&3 on VSMCs and yet-to-be-defined cardiac receptors, it also modulates arterial resistance and LV function to maintain optimal BP and cardiovascular homeostasis. Our findings suggest that distinct chaperone-bound pools mediate complementary physiological effects: HDL-S1P vasodilatory and albumin-S1P contractile, warranting further studies on albumin- and HDL-S1P pools and their distribution in cardiovascular disease.

This study has important limitations that will need to be addressed in future work, including its entirely preclinical nature, unresolved mechanisms of cardiac dysfunction in mice lacking circulating S1P, and redundancy and genetic compensation rendering it difficult to isolate roles for S1PRs in vascular and cardiac phenotypes.

Materials and Methods

Please refer to the SI Appendix for detailed Methods. Generation of plasma S1Pless mice via compound Sphk1 and Sphk2 gene deletion by inducible and constitutive strategies with Mx1- and Vav1Cre, quantification of S1P levels in these mice and irradiated wild-type recipients of their BMC and characterization of nonhematopoietic Cre-mediated excision by the Cre drivers was reported (12, 13, 29, 30, 43). Other mouse lines are described in SI. Controls were sex and age-matched littermates for all studies including BMC donors. As plasma S1Pless females display normal cardiac function, molecular and structural analysis of the heart was limited to male mice. Generation and characterization of recombinant ApoM-based proteins, S1P loading, and functional analysis was reported (25, 46). BP was measured by tail cuff plethysmography in conscious restrained mice or with aortic pressure transducers either directly connected to a recording device in anesthetized mice or connected to radiotelemetry transmitters in conscious freely moving mice; LV systolic function, cardiac output, and peripheral hemodynamics by ultrasound B-mode imaging and color-Doppler recordings; GFR by transcutaneous monitoring of FITC-Sinistrin clearance; renal vascular resistance and reactivity with a pressure transducer during renal perfusion; mesenteric artery resistance function by wire myography; sphingolipids, angiotensin peptides, aldosterone, and corticosterone by LC-MS/MS; blood electrolytes and gases using iSTAT analysis; blood, tissue, and urine collection, tissue histology, immunohistochemistry, Western blot, RNA isolation, quantitative PCR, and bulk RNA Seq and analyses by standard techniques.

Study Approval.

The study was approved by the animal care and use committee of the Paris Descartes University and by the French Department of Education (02822.02; 03474.02; 03560.02; 28539-202006271229617) and conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

Statistical Analyses.

Statistical analyses were performed with Prism 10 software (Prism, GraphPad, San Diego). A value of P < 0.05 was considered statistically significant. Parametric tests were applied only if normal distribution could be confirmed.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Bulk RNA sequencing data have been deposited in NCBI/GEO (GSE289865) (67). Study data are included in the article and/or SI Appendix.