Association between dietary phosphate intake and skeletal muscle energetics in adults without cardiovascular disease

Abstract

Highly bioavailable inorganic phosphate (Pi) is present in large quantities in the typical Western diet and represents a large fraction of total phosphate intake. Dietary Pi excess induces exercise intolerance and skeletal muscle mitochondrial dysfunction in normal mice. However, the relevance of this to humans remains unknown. The study was conducted on 13 individuals without a history of cardiopulmonary disease (46% female, 15% Black participants) enrolled in the pilot-phase of the Dallas Heart and Mind Study. Total dietary phosphate was estimated from 24-h dietary recall (ASA24). Muscle ATP synthesis was measured at rest, and phosphocreatinine (PCr) dynamics was measured during plantar flexion exercise using 7-T ³¹P magnetic resonance (MR) spectroscopy in the calf muscle. Correlation was assessed between dietary phosphate intake normalized to total caloric intake, resting ATP synthesis, and PCr depletion during exercise. Higher dietary phosphate intake was associated with lower resting ATP synthesis (r = –0.62, P = 0.03), and with higher levels of PCr depletion during plantar flexion exercise relative to the resting period (r = –0.72; P = 0.004). These associations remain significant after adjustment for age and estimated glomerular filtration rate (both P < 0.05). High dietary phosphate intake was also associated with lower serum Klotho levels, and Klotho levels are in turn associated with PCr depletion and higher ADP accumulation post exercise. Our study suggests that higher dietary phosphate is associated with reduced skeletal muscle mitochondrial function at rest and exercise in humans providing new insight into potential mechanisms linking the Western diet to impaired energy metabolism.

NEW & NOTEWORTHY This is the first translational research study directly demonstrating the adverse effects of dietary phosphate on muscle energy metabolism in humans. Importantly, our data show that dietary phosphate is associated with impaired muscle ATP synthesis at rest and during exercise, independent of age and renal function. This is a new biologic paradigm with significant clinical dietary implications.

INTRODUCTION

Inorganic phosphate (Pi) is a common ingredient used for leavening, emulsifying, color and material preservation, pH adjustment, and taste enhancement in highly processed foods, including frozen meals, canned food, soda drinks, and bakery items. Thus, the average diet in the United States far exceeds the recommended daily allowance of dietary Pi intake of 700 mg of elemental phosphorus (1).^. Up to 25% of US adults consume phosphate at three to five times higher than the above-recommended daily allowance on a regular basis (2). The preponderance of Pi in the Western diet further exacerbates this excess due to the high bioavailability of Pi. However, the long-term health consequences of excess dietary phosphate consumption in the general population remain unknown.

Prior work from our group in animal models suggests that increased dietary phosphate induced skeletal muscle mitochondrial functional impairment, leading to reduced muscle oxygen uptake and exercise intolerance in normal mice, which is not explained by cardiac function (3). However, the relevance and translatability of the preclinical high dietary Pi model on energy metabolism in humans are unknown. Accordingly, we conducted a parallel translational research study to determine the association between dietary phosphate intake and skeletal muscle energetics in humans without a previous history of cardiovascular disease.

MATERIALS AND METHODS

Participants were recruited from the cohort pool previously enrolled in the Dallas Heart Study phase 2 (DHS-2) after providing informed consent. Thirteen participants without a history of cardiopulmonary disease were enrolled. The study was approved by the Institutional Review Board at the University of Texas Southwestern Medical Center. Details regarding DHS selection criteria, study design, and methods have been described elsewhere (4). The DHS is a multiethnic, population-based probability sample of Dallas County. Race was self-reported by study participants, and race categories were defined by investigators based on the US Office of Management and Budget’s Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity. The glomerular filtration rate was estimated using the CKD-EPI 2021 formula (5). Total dietary Pi intake was calculated via the Automated Self-Administered 24-hour (ASA24) Dietary Assessment Tool based on 1 day of food recall. It should be noted that there is no elemental phosphorus in our diet or the body but clinical reports conventionally (blood, urine, food, etc.) refer to the amount of phosphorus in a sample as a surrogate for phosphate. Pi intake reported in the current study refers to elemental phosphorus, which has a lower molecular weight than phosphate (31 vs. 95, respectively). To account for the correlation between total phosphorus intake and caloric intake, we evaluated the dietary phosphorus density (6) as the ratio of total phosphorus intake-to-total caloric intake, described previously (7). In our pilot studies, total daily phosphorus intake and phosphorus intake-to-total caloric intake ratio based on 1 day of food recall were found to be significantly correlated with urinary phosphorus to creatinine ratio (Supplemental Fig. S1, A and B, both P < 0.05). Diabetes was determined by an HbA1c of 6.5% or higher. Hypertension was determined with blood pressure ≥130/80 mmHg (8). All participants underwent blood/urine collections within 24 h of completion of dietary questionnaire and within 2 wk of echocardiography and magnetic resonance spectroscopy (MRS). Serum Klotho was measured using a highly specific synthetic anti-α-Klotho antibody (sb106) as previously described (9). Fibroblast growth factor 23 (FGF23) was measured using a C-terminal ELISA method (Immutopics International, Quidel Corporation, Athens, OH).

Echocardiography

A standard two-dimensional (2-D), M-mode, and Doppler echocardiography (from parasternal long- and short-axis, apical-4 and apical-2 chamber, long axis, and subcostal views) was performed using an iE33 ultrasound system (Philips Healthcare, Andover, MA) following the American Society of Echocardiography guidelines (10).

Magnetic Resonance Spectroscopy

The subjects underwent ³¹P MRS scan in the supine position with the calf muscle of their primary leg in the center of the detection radiofrequency (RF) coil on a 7-T human MRI scanner (Achieva; Philips Healthcare, Best, The Netherlands). Resting (adenosine triphosphate) ATP synthesis [Pi + (adenosine diphosphate) ADP→ATP] was determined in vivo by applying radiofrequency (RF) pulse to invert the ³¹P signal of the phosphate group at the gamma position of the ATP, denoted as γ-ATP, into a negative signal, whereas the free inorganic Pi is left unchanged (i.e., the Pi signal remains to be positive). Following the RF pulse, the subsequent chemical exchange between the positive Pi signal and the negative γ-ATP signal results in partial cancelation of the Pi signal and a fall in the amplitude. Under the influence of this exchange effect and the intrinsic magnetic relaxation effect, the Pi signal typically experiences an initial fall followed by a rise in positive signal pattern over time. This fall-rise pattern can be captured by the exchange kinetics by band inversion transfer (EBIT sequence), a wideband inversion-based magnetization transfer (MT) method, as previously described (11). Similarly, the creatine kinase (CK)-mediated ATP recycling by reaction (PCr + ADP→ATP + Cr) was also monitored by the MT effects between γ-ATP and PCr through PCr inversion to a negative signal. These magnetization transfer (MT) effects allow calculation of bidirectional exchange reactions Pi ↔ γ-ATP and PCr ↔ γ-ATP using two different MT modules. Module I was used to invert Pi and PCr while observing MT effects at γ-ATP spins. Module II was used to invert PCr and γ-ATP spins while observing MT effects at Pi. Both modules were performed at a constant TR = 7 s with 7 varying inversion delays (t_d) ranging from 30 ms to 5 s (logarithmically spaced) and each with an eight-scan average.

Dynamic ³¹P MRS scans.

After completion of the EBIT kinetic scans at resting state, subjects were asked to perform plantar flexion exercise in-magnet for 1 min (2-s cycle with 1 s of contraction followed by 1 s of relaxation for a total of 30 cycles) during the dynamic ³¹P scan at a constant TR of 2 s. Calf muscle excursion was performed by plantar flexion against a constant workload (20% of lean body weight estimated by Boer’s formula), using a magnet-compatible pulley system mounted on the scanner table. The ³¹P dynamic scans included 10 baseline scans without exercise, followed by 30 scans with exercise, and then 160 scans post exercise. Two bouts of exercise were performed by each subject with 10 min of rest between each bout and the results were averaged. The energy state of the calf muscle was quantified as the phosphorylation potential, [ATP]/([ADP][Pi]), which was used to calculate the Gibbs free energy of ATP hydrolysis.

Statistical Analysis

Statistical analyses were performed using SAS version 9.4. Participant characteristics are reported as means ± SD, and number (percentage) for categorical data. Pearson’s correlation coefficient was used to determine the strength of association between markers of dietary Pi intake and muscle energetics at rest and during exercise. Since muscle mitochondrial function may be influenced by age and other relevant factors, multivariable linear regression analyses were performed to assess the associations between markers of Pi intake and muscle energetics after adjusting for relevant baseline characteristics and variables identified from the univariate model. Standardized β estimates with 95% confidence intervals (CIs) were reported for multivariable linear regression analyses. The standardized β predicted the change in the response variable for 1 SD of change in the explanatory variable (while controlling for the other variables).

RESULTS

We studied 13 individuals (46% female, 15% Black participants) with a mean age of 60.4 ± 17 yr, body mass index (BMI) of 27.6 ± 3 kg/m², and left ventricular ejection fraction (LVEF) of 64.3 ± 6.6%, respectively. Of the 13 participants, 3 had type 2 diabetes mellitus and 8 had hypertension; all were well-controlled. Among these participants, seven were taking antihypertensive medications and none were taking glucose-lowering medications. The estimated average dietary phosphorus intake was 1,414 ± 724 mg and average phosphorus intake-to-total caloric intake (Pi density) was 0.80 ± 0.3 mg/kcal. The remaining baseline characteristics are shown in Table 1.

Table 1.

Subject characteristics at screening

Variables	Means ± SD
Age, yr	60.4 ± 17.0
Sex (M/F)	7/6
Black participants, %	2/13 (15%)
Body mass index, kg/m2	27.6 ± 3.1
Systolic blood pressure, mmHg	128.3 ± 15.7
Diastolic blood pressure, mmHg	75.9 ± 7.3
Fasting plasma glucose, mg/dL	111.7 ± 19.9
Serum creatinine, mg/dL	1.01 ± 0.40
eGFR, mL/min/1.73 m2	82.4 ± 22.3
Serum triglyceride, mg/dL	107.1 ± 59.9
Total cholesterol, mg/dL	173.9 ± 34.2
Total daily Pi intake, mg/day	1,414 ± 724
Pi intake/total caloric intake, mg/kcal	0.80 ± 0.3
High-density lipoprotein concentration, mg/dL	51.7 ± 9.4
Left ventricular end diastolic volume, mL	87.6 ± 17.6
Left ventricular end systolic volume, mL	31.9 ± 10.7
Left ventricular ejection fraction, %	64.3 ± 6.6
Type 2 diabetes mellitus, %	3/13 (23%)
Hypertension, %	8/13 (62%)
Resting de novo ATP synthesis, mM/min	19.3 ± 11.5
PCr concentration at rest, mM	28.0 ± 4.0
PCr concentration at end exercise, mM	16.2 ± 4.3
ADP concentration at rest (µM)	10.6 ± 4.0
ADP concentration at end exercise, µM	87.2 ± 54.2

n = 13 subjects. ADP, adenosine diphosphate; ATP, adenosine triphosphate; eGFR, estimated glomerular filtration rate; PCr, phosphocreatinine; Pi, inorganic phosphate.

Association between Dietary Pi Intake and Skeletal Muscle Energetics

Resting ATP synthesis is inversely associated with age (r = –0.75, P = 0.002; Fig. 1A). In contrast, eGFR is positively correlated with resting ATP synthesis (r = 0.58, P = 0.03; Fig. 1B). Dietary Pi intake, on the other hand, is inversely associated with resting ATP synthesis (r = –0.62, P = 0.03; Fig. 1C). In contrast, there was no correlation between ATP resynthesis catalyzed by CK reaction and phosphate intake (Fig. 1D), age, eGFR, or other demographic variables (data not shown). The association between dietary phosphate intake and resting ATP synthesis remained significant after adjusting for potential confounders such as age and eGFR (Table 2).

Figure 1.

Summary data (n = 13) in resting muscle showing significant correlation between de novo ATP synthesis flux and age (A), eGFR (B), and dietary Pi intake (C). In contrast, no correlation between CK-mediated ATP resynthesis rate and Pi intake (D). Relationship between each dataset was determined using Pearson’s correlation coefficient. eGFR, estimated glomerular filtration rate.

Table 2.

Adjusted multivariable linear regression analyses of mitochondrial function at rest with markers of Pi intake

Variable	Adjusted for Age		Adjusted for Age + eGFR
	β (95% CI)*	P Value	β (95% CI)*	P Value
Outcome: resting de novo ATP synthesis
Pi/kcal ratio	–0.39 (–0.80, 0.02)	0.09	–0.47 (–0.86, –0.08)	0.047
Total daily Pi intake	–0.30 (–0.70, 0.10)	0.17	–0.48 (–0.87, –0.10)	0.038

*β (standardized regression coefficients) and 95% CIs. ATP, adenosine triphosphate; CI, confidence interval; eGFR, estimated glomerular filtration rate; PCr, phosphocreatinine; Pi, inorganic phosphate. Boldface indicates significant values. Statistical significance is indicated by boldface for P < 0.05.

To further clarify the role of dietary phosphate excess on muscle energy utilization during exercise, dynamic ³¹P MRS spectroscopy during plantar flexion exercise was performed in 13 subjects. We found that higher dietary Pi intake was associated with greater PCr depletion at end-of-exercise relative to resting PCr (Fig. 2A), and higher ADP levels within 1 min of exercise and at end-of-exercise relative to rest (Fig. 2, B and C). No correlation was noted between dietary Pi intake and resting ADP concentration (Fig. 2D) or LVEF (data not shown). Likewise, there was no correlation between PCr depletion at end-of-exercise or ADP at end-of-exercise relative to rest with age, eGFR, or other demographic variables (data not shown). The association between phosphorus density and the PCr at end exercise relative to resting PCr ratio remains significant after adjusting for age and eGFR (Table 3). Similarly, the association between phosphorus density and the ADP concentration within 1 min of exercise in the calf muscles remains significant after adjusting for age and eGFR (Table 3).

Figure. 2.

Summary data (n = 13) in skeletal muscle at end exercise (ex) showing significant inverse correlation between Pi intake and (A) the ratio of PCr at end ex to PCr at rest as well as positive correlation between dietary Pi intake and (B) the ratio of muscle ADP at end ex to ADP at rest, and (C) ADP concentration at end exercise. No correlation between resting ADP concentration and (D) Pi intake. Relationship between each dataset was determined using Pearson’s correlation coefficient. PCr, phosphocreatinine.

Table 3.

Adjusted multivariable linear regression analyses of mitochondrial function at end exercise with markers of Pi intake

Variable	Adjusted for Age		Adjusted for Age + eGFR
	β (95% CI)*	P Value	β (95% CI)*	P Value
Outcome: PCr exercise/PCr rest ratio
Pi/kcal ratio	–0.76 (–1.25, –0.26)	0.01	–0.71 (–1.25, –0.18)	0.03
Total daily Pi intake	–0.40 (–0.93, 0.14)	0.18	–0.32 (–0.95, 0.30)	0.34
Outcome: ADP concentration at end exercise
Pi/kcal ratio	0.72 (0.20, 1.24)	0.02	0.74 (0.16, 1.31)	0.03
Total daily Pi intake	0.19 (–0.39, 0.78)	0.53	0.15 (–0.54, 0.84)	0.67

^*β (standardized regression coefficients) and 95% CIs. ADP, adenosine diphosphate; CI, confidence interval; eGFR, estimated glomerular filtration rate; PCr, phosphocreatinine; Pi, inorganic phosphate. Boldface indicates significant values. Statistical significance is indicated by boldface for P < 0.05.

To further determine the role of phosphate regulators on the muscle energetics, we first determined a correlation between phosphorus intake and FGF23 as well as serum Klotho. We found a significant inverse correlation between phosphorus intake and Klotho (Fig. 3B) but no correlation between phosphorus intake and FGF23 (Fig. 3A). Furthermore, there was a significant correlation between serum Klotho and PCr Ex/PCr rest ratio (Fig. 3C) as well as ADP Ex/ADP rest (Fig. 3D, both P < 0.05) but no correlation between Klotho and de novo ATP synthesis or CK-mediated ATP resynthesis (Fig. 3, E and F). Similarly, there was no correlation between FGF23 and PCr Ex/PCr rest ratio, ADP Ex/ADP rest, de novo ATP synthesis, or CK-mediated ATP resynthesis (Fig. 3, G–J).

Figure 3.

Summary data (n = 13) showing significant inverse correlation between Pi intake and Klotho (A), but no correlation between Pi intake and FGF23 (B); significant correlation between Klotho and the ratio of PCr at end exercise (ex) to PCr at rest (C), and inverse correlation with the ratio of muscle ADP at end ex to ADP at rest (D). No correlation between Klotho and de novo ATP synthesis (E) or CK-mediated ATP resynthesis (F). No correlation between FGF23 and markers of all muscle bioenergetics at rest or during exercise (G–J). Relationship between each dataset was determined using Pearson’s correlation coefficient. FGF23, fibroblast growth factor 23; PCr, phosphocreatinine.

DISCUSSION

There are three main findings of our study. 1) Dietary phosphorus excess is associated with impaired muscle ATP synthesis at rest in subjects without previous cardiopulmonary disease, impaired left ventricular or renal function. 2) Dietary phosphorus excess is associated with a rapid decline in PCr during submaximal plantar flexion exercise, suggesting impaired phosphorus energetics including mitochondrial function during exercise. 3) Higher serum Klotho levels are correlated with higher PCr after exercise and lower accumulation of ADP post exercise compared with pre exercise. 4) The multiple associations observed are independent of potentially relevant factors that influence muscle energetics including age and renal function.

Studies from our group and others have shown that dietary phosphorus loading at the level that mimics US consumption led to a downregulation of numerous genes that regulate fatty acid (FA) availability, leading to a reduction in total body fat oxidation and muscle mitochondrial dysfunction in mice (3, 12). In addition, in vitro studies showed that high-phosphate culture media acutely impair fatty acid oxidation and oxygen consumption in isolated mitochondria. Accordingly, a high-phosphate diet reduces resting V̇o₂ oxygen uptake (V̇o₂) and maximal exercise capacity in mice fed with a high-phosphate diet. There is increasing evidence that healthy skeletal muscle possesses metabolic flexibility in “choosing” fuels, that is, switching between fatty acid and carbohydrate metabolism, depending on substrate availability and exercise intensity and demand (13). This ability to modify substrate oxidation in response to changes in nutrient availability is markedly attenuated in older and sedentary individuals (14–16). Our data on de novo ATP synthesis in resting skeletal muscle in humans are consistent with previous data showing reduced mitochondrial function in resting muscle in mice exposed to dietary phosphate at two- to threefold higher than considered to be adequate for their health.

We chose de novo ATP synthesis as a surrogate for resting mitochondrial function in the present study because it is the key energy production reaction (Pi + ADP => ATP) in skeletal muscle at rest and occurs primarily as a consequence of oxidative phosphorylation. This reaction efficiently captures energy from the oxidation of fats and carbohydrates in mitochondria.

During the rest-to-exercise transition, PCr serves as an immediate high-energy phosphate buffer to maintain ATP synthesis and levels to meet abrupt increase in energy demand. This is an instantaneous, albeit limited, fuel source as compared with carbohydrates and fats, and this supply becomes critical during transient underperfusion when oxygen availability is restricted.

In parallel to impaired resting ATP synthesis, we also found an association between dietary Pi intake and an accelerated decline in skeletal muscle PCr during dynamic exercise in which the muscle PCr store is depleted with repeated muscle contraction, suggesting a reduced oxidative capacity (17). A higher level of muscle [ADP] at end-of-exercise also suggests impaired mitochondrial function (note that the calculation of [ADP] includes the measurement of PCr depletion based on CK reaction as well as the measurement of pH based on the chemical shift of Pi) (18, 19). This is significant given that previous studies have shown a faster decline in PCr after exercise in seniors than in younger volunteers (20, 21). Furthermore, there is an association between exercise training and an attenuation in PCr during exercise, which is indicative of a reduced reliance on energy buffer storage (PCr) to supply energy to contracting muscles (22, 23). Taken together, our study suggests that higher dietary Pi intake tends to impair skeletal muscle energetics that resembles the effect of aging.

Dietary phosphate loading triggers increases in regulatory phosphaturic substances, such as parathyroid hormone (PTH) and fibroblast growth factor 23 (FGF23), which serve to maintain phosphate balance. High dietary phosphate also suppressed serum Klotho akin to what was previously observed in rodent studies (24), which may impact muscle metabolism and exercise capacity. Klotho is a cytoprotective protein and circulating Klotho was inversely related to dietary phosphate intake (Fig. 3B). There is a plethora of studies examining the positive effect of Klotho on exercise and how exercise stimulates Klotho, which is summarized in some recent reviews (25, 26). There are data on multiple fronts including cross-sectional associations between Klotho and exercise capacity and frailty (27–30), preclinical animal models (31, 32), and direction addition in vitro to cells/tissues (33). Klotho appears to guard against age-related decline in physical activity and running endurance in mice (34). Klotho supplementation protects against age-related decline in progenitor cell mitochondrial dysfunction and impaired muscle regeneration (33, 35). Thus, Klotho downregulation may contribute in part to reduced muscle mitochondrial function associated with a high-phosphate diet.

Our study is limited by a small sample size, its cross-sectional and retrospective nature, and its reliance on dietary recall, which limits the ability to have high accuracy on dietary intake and conclude on causal relationships. The inferred muscle mitochondrial function from muscle ATP dynamics in the present study was assessed in the specific group of leg muscles, which may not reflect all muscle mitochondrial function. Nevertheless, using the 7-T MRI system allows us to assess muscle ATP production at rest and during exercise with high sensitivity as compared with the conventional 3 T and 1.5 T platforms. Dietary phosphate intake was estimated from the food recall, which may be susceptible to recall bias of the meal size or portion. In addition, bioavailability of phosphate contained in different food items may vary greatly depending on the source with the highest (80%–100%) bioavailability from the inorganic phosphate found in preservatives, lower (<40%) bioavailability from the organic phosphate, and the lowest from plant-derived phytates (36). Nevertheless, significant correlation between phosphate intake and urinary Pi/Cr ratio in our study validates that the estimated intake is a reliable estimate of dietary phosphate loading. The measurement of phosphaturic hormones was limited to FGF23 and Klotho, as such, the role of other hormones, such as PTH and vitamin D, cannot be excluded. Future observational studies and, importantly, randomized clinical trials are needed to verify the findings in diverse populations. Moreover, the direct and indirect mechanisms of how phosphate affects muscle energetics require elucidation.

DATA AVAILABILITY

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

SUPPLEMENTAL DATA

Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.25146011.

GRANTS

The study was supported by NIH R01HL159994, Charles and Jane Pak Center for Mineral Metabolism and Clinical Research, and the UT Southwestern O’Brien Kidney Research Core Center.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

J.R., C.M., A.P, A.S., J.D.B., and W.V. conceived and designed research; U.B.P., T.J., J.R., and W.V. performed experiments; J.M.G. and W.V. analyzed data; J.M.G., J.P., J.R., C.M., A.P., J.D.B., O.W.M., and W.V. interpreted results of experiments; J.M.G. and W.V. prepared figures; J.M.G., A.A., O.W.M., and W.V. drafted manuscript; J.M.G., J.R., L.S., C.M., A.P., A.S., J.D.B., O.W.M., and W.V. edited and revised manuscript; J.M.G., J.R., L.S., C.M., A.P., A.S., J.D.B., O.W.M., and W.V. approved final version of manuscript.