Research ArticleNutrition

A high-salt diet compromises antibacterial neutrophil responses through hormonal perturbation

See allHide authors and affiliations

Science Translational Medicine  25 Mar 2020:
Vol. 12, Issue 536, eaay3850
DOI: 10.1126/scitranslmed.aay3850

Salting neutrophils’ game

Sodium chloride (salt) has been shown to invigorate immune responses in various contexts. In contrast, Jobin et al. now show that salt can impair neutrophil antibacterial responses. Mice on a high-salt diet experienced exacerbated E. coli kidney or systemic Listeria monocytogenes infections due to reduced capacity of neutrophils to kill ingested bacteria. The neutrophil deficiencies were not due directly to salt or urea but instead were dependent on salt-induced hyperglucocorticoidism. In addition, neutrophils from healthy volunteers were less capable of controlling bacteria ex vivo after consumption of a high-salt diet. Given that the typical Western diet is replete with salt, these findings reveal that people might be making themselves more vulnerable to bacterial infections.

Abstract

The Western diet is rich in salt, which poses various health risks. A high-salt diet (HSD) can stimulate immunity through the nuclear factor of activated T cells 5 (Nfat5)–signaling pathway, especially in the skin, where sodium is stored. The kidney medulla also accumulates sodium to build an osmotic gradient for water conservation. Here, we studied the effect of an HSD on the immune defense against uropathogenic E. coli–induced pyelonephritis, the most common kidney infection. Unexpectedly, pyelonephritis was aggravated in mice on an HSD by two mechanisms. First, on an HSD, sodium must be excreted; therefore, the kidney used urea instead to build the osmotic gradient. However, in contrast to sodium, urea suppressed the antibacterial functionality of neutrophils, the principal immune effectors against pyelonephritis. Second, the body excretes sodium by lowering mineralocorticoid production via suppressing aldosterone synthase. This caused an accumulation of aldosterone precursors with glucocorticoid functionality, which abolished the diurnal adrenocorticotropic hormone–driven glucocorticoid rhythm and compromised neutrophil development and antibacterial functionality systemically. Consistently, under an HSD, systemic Listeria monocytogenes infection was also aggravated in a glucocorticoid-dependent manner. Glucocorticoids directly induced Nfat5 expression, but pharmacological normalization of renal Nfat5 expression failed to restore the antibacterial defense. Last, healthy humans consuming an HSD for 1 week showed hyperglucocorticoidism and impaired antibacterial neutrophil function. In summary, an HSD suppresses intrarenal neutrophils Nfat5-independently by altering the local microenvironment and systemically by glucocorticoid-mediated immunosuppression. These findings argue against high-salt consumption during bacterial infections.

INTRODUCTION

The high salt content of the Western diet poses various health risks and is thought to contribute to prosperity illnesses. The recommended daily salt intake is controversial (13). A high-salt diet (HSD) is thought to be proinflammatory and to stimulate immunity, especially in the skin, the sodium storage organ of the body (4), by engaging the transcription factor Nfat5 (nuclear factor of activated T cells 5) in local macrophages. HSD was shown to invigorate macrophage-driven immunity against cutaneous leishmaniasis (5), and high cutaneous sodium concentrations have recently been proposed to promote T helper 2 (TH2) cell–mediated atopic dermatitis (6). Furthermore, an HSD stimulated macrophages and/or TH17 cell responses (710) and altered the intestinal microbiome (11), leading to exacerbation of experimental multiple sclerosis. These findings suggested that the high salt content of a Western diet causes a general proinflammatory state and might contribute to the increasing incidence of autoimmunity. It is unclear whether an HSD can stimulate other immune cells as well.

The kidney maintains the salt and water homeostasis of the body. It conserves water by reabsorbing sodium chloride from the glomerular filtrate and accumulating it in the medulla, thereby establishing an osmotic gradient that drags filtered water from the tubular lumen back into circulation. Sodium homeostasis is regulated by the renin-angiotensin-aldosterone hormone system (RAAS). Detection of low sodium concentrations in the kidney causes the release of renin, which, through angiotensin II, stimulates aldosterone production in the adrenal gland. Aldosterone induces transporters in kidney tubular epithelial cells that reabsorb sodium (12). Under an HSD, the RAAS is suppressed, allowing for the excretion of excess sodium.

The effects of the high osmolarity in the renal medulla on immunity are controversial. High sodium has been reported to polarize medullary mononuclear phagocytes to an anti-inflammatory state in after transplantation (13). By contrast, it has recently been shown that renal tubular epithelial cells produced, in an osmolarity- and Nfat5-dependent manner, chemokines, which attracted proinflammatory monocytes (14). Analogous to the immunostimulatory function of sodium in the skin (5), it was proposed that the intrarenal sodium gradient establishes an anti-infectious microenvironment in the kidney (14). This was supported by the observation that kidney transplant patients treated with loop diuretics that can disrupt this gradient showed an increased incidence of urinary tract infections (UTIs) (15).

Bacterial infections of the urinary tract are among the most prevalent infections and affect more than 25% of the population, especially young females (1618). Uropathogenic Escherichia coli (UPEC) cause more than 70% of UTIs. UPEC ascension from the bladder into the kidney leads to pyelonephritis, a potentially life-threatening disease (18, 19). The innate immune defense against UTI relies on neutrophilic granulocytes, which clear UPEC by phagocytosis, and on mononuclear phagocytes such as macrophages or dendritic cells, which attract and activate the neutrophils by chemokines and cytokines (16, 20, 21). Treatment of pyelonephritis includes antibiotic drugs and high liquid intake to flush UPEC out of the kidney (22). In the present study, we tested whether an HSD can strengthen the intrarenal immune defense against pyelonephritis.

RESULTS

An HSD impairs neutrophil functionality and worsens pyelonephritis

To investigate the effect of increased sodium intake on pyelonephritis outcomes, we induced pyelonephritis in mice that had been exposed for 1 week to a widely used HSD regimen and in control mice under normal-salt diet (NSD) or low-salt diet. We ensured that fluid intake and, thus, urine production were similar between experimental groups (fig. S1A). Unexpectedly, we noted that kidneys of HSD-fed mice contained four to six times more UPEC compared with mice under NSD or low-salt diet, indicating exacerbation of infection (Fig. 1A). Microscopic analysis showed more bacteria accumulating in the pelvis (Fig. 1B), confirming more severe pyelonephritis on an HSD. A direct effect of the high salt concentration on UPEC growth was excluded by in vitro experiments (fig. S1B).

Fig. 1 HSD exacerbates pyelonephritis in mice.

Mice were fed low-salt diet (LSD), NSD, or HSD for 1 week and infected with UPEC. Eighteen hours after infection, kidneys were analyzed. (A) Quantification of E. coli colony-forming units in kidney homogenates. (B) Representative microscopy pictures of kidney sections infected with E. coli expressing GFP. Dim green, autofluorescence; bright green, E. coli GFP stained with anti-GFP Alexa Fluor 488 antibody; red, CD11b-APC; blue, CD45-V421. Yellow rectangles indicate areas with visible UPEC. (C to E) Flow cytometric quantification of kidney dendritic cells (viable CD45+, MHCII+, and CD11c+) (C), inflammatory monocytes (viable CD45+, Ly6G, and Ly6C+) (D), and neutrophils (viable CD45+, Ly6Cint, and Ly6G+) (E) 18 hours after infection in mice fed NSD or HSD for 1 week. (F) Flow cytometric quantification of reactive oxygen species (ROS) production by kidney neutrophils using the DCFDA dye. (G) Flow cytometric quantification of inducible nitric oxide synthase (iNOS) expression by kidney neutrophils. (H) FACS plot of intracellular E. coli contained in kidney neutrophils. The numbers indicate percentages of neutrophils containing E. coli and geometric mean fluorescence intensities (MFIs). SSC, side scatter. (I) Quantification of bacterial viability in neutrophils isolated from infected kidneys based on gentamicin protection assay and phagocytosis. Live and dead bacteria were measured by flow cytometry (E. coli MFI). Live bacteria were measured by colony-forming units. Both measures were normalized to the average of the NSD group, and normalized live bacterial counts were divided by normalized total bacterial counts. MFI0, MFI of unstained sample; AU, arbitrary units. *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA (A) and unpaired Student’s t test (C to I). Bar graphs indicate means and SEM. The experiments were performed at least two times with seven mice per group. Graphs depict pooled experiments.

To clarify the underlying mechanisms, we examined the antibacterial immune response in the kidney. Neither intrarenal chemokines or proinflammatory cytokines (fig. S2A) nor the numbers of dendritic cells, inflammatory monocytes, or neutrophils showed major differences between infected mice on NSD and HSD (Fig. 1, C to E, and fig. S2, B and C). Intrarenal neutrophils produced similar amounts of reactive oxygen species (Fig. 1F), but somewhat lower amounts of inducible nitric oxide synthase (Fig. 1G), hinting at reduced bactericidal activity during HSD exposure. UPEC phagocytosis by intrarenal neutrophils from infected HSD mice seemed to be reduced but not statistically significant (Fig. 1H). However, given the higher bacterial numbers in kidneys of HSD mice, this suggested an inappropriately low phagocytic activity. Despite phagocytosing similar amounts of bacteria, the neutrophils isolated from the kidneys of HSD mice contained more viable bacteria, indicating an impaired ability to kill phagocytosed UPEC (Fig. 1I). Thus, an HSD aggravated experimental pyelonephritis by compromising the bactericidal functions of neutrophils.

An HSD increases Nfat5, but not sodium, in the kidney

Aggravation of pyelonephritis and reduced neutrophil functionality are inconsistent with the notion of a sodium-rich, Nfat5-mediated immunostimulatory microenvironment (14). We therefore asked whether such a microenvironment actually existed in the kidneys of mice on HSD. Intrarenal Nfat5 levels were higher on HSD, not only in the outer medulla but also in the cortex (Fig. 2A), which is usually normo-osmolar. Hence, we measured sodium in the kidneys of HSD-fed mice by atomic absorption spectroscopy. Unexpectedly, intrarenal sodium was not elevated (Fig. 2B), so sodium-mediated immunostimulation would not be a factor here. Another method, flow cytometric Asante sodium green staining of intrarenal neutrophils, revealed even a slight sodium decrease (Fig. 2C and fig. S3). Consistent with the absence of a proinflammatory microenvironment, kidney macrophages from mice fed an HSD expressed less tumor necrosis factor–α (TNFα) (fig. S4A). The ratio between intrarenal TH17 and regulatory T (Treg) cells, which has been described to be increased under an HSD in other organs (7, 8, 10), was not affected by this diet (fig. S4B).

Fig. 2 HSD-induced intrarenal effects.

Mice were fed NSD or HSD for 1 week, and kidneys were analyzed. (A) RT-PCR quantification of Nfat5 expression relative to Gapdh and control [kidney cortex (Cor) NSD] in kidneys of uninfected mice fed NSD or HSD for 1 week. OMed, outer medulla; IMed, inner medulla. (B) Atomic absorption spectroscopy quantification of sodium content in kidneys of uninfected mice. (C) Flow cytometric quantification of sodium (Asante NaTRIUM Green-2) in kidney neutrophils 18 hours after infection in mice fed NSD or HSD for 1 week. (D) Quantification of bacterial viability in bone marrow neutrophils exposed in vitro to +40 mM NaCl gentamicin protection assay and phagocytosis combined. *P < 0.05 by unpaired Student’s t test (A to D). Bar graphs indicate means and SEM. The experiments were performed at least twice with 5 (A and D), 17 (i.e., kidneys from 17 mice were pooled for one measurement) (B), and 3 (C) mice per group. Graphs depict pooled experiments.

We decided to compare these findings to the situation in the skin, which does accumulate sodium on an HSD (4, 5, 23, 24). This was also the case in our hands (fig. S4C). Although there was a slight tendency toward proinflammatory effects, neither TNFα production by skin macrophages (fig. S4D) nor the ratio between skin TH17 and Treg cells (fig. S4E) was altered. In addition, in the intestine, macrophage TNFα production was not significantly increased (fig. S4F), but the TH17/Treg ratio cells was increased (*P < 0.05, fig. S4G). These findings lend further support to previous studies documenting inflammatory alterations in organs that accumulate sodium under an HSD (79) and show that the kidney is not among these organs.

Another difference between our model and previous studies is the type of immune effector cell involved. When we cultured neutrophils and macrophages at high salt concentrations, only macrophages showed increased antibacterial functionality (Fig. 2D and fig. S4, H and I). These findings are consistent with previous reports showing that sodium can stimulate macrophages (5, 9, 14, 25), and indicate that this is not so for neutrophils. In conclusion, an HSD did not increase sodium in the kidney; moreover, sodium chloride would fail to stimulate neutrophils, the principal effector cells against pyelonephritis.

An HSD induces hyperglucocorticoidism that leads to Nfat5 up-regulation

The discrepancy between intrarenal Nfat5 and sodium concentrations implied that Nfat5 must have been increased by a factor other than sodium. Nfat5 was elevated also in the spleens of HSD-fed mice (Fig. 3A), hinting at a systemic factor. We decided to identify this factor by comparing the transcriptomes of neutrophils in the bone marrow of noninfected mice on NSD and HSD. RNA sequencing (RNA-seq) transcriptome analysis revealed up-regulation of known glucocorticoid target genes (26) such as Tsc22d3 [GILZ (glucocorticoid-induced leucine zipper); inhibits the key immune activator NFκB], Zfp36 [TTP (tristetraprolin)], Fkbp5 [Fkbp51 (FK506-binding protein 51)], or Per1 (a transcription factor that regulates circadian rhythms) (Fig. 3B) and alterations of genes involved in steroid biosynthesis (fig. S5A) in neutrophils from HSD-fed mice. Glucocorticoids were elevated in the blood of HSD-fed mice (Fig. 3C), confirming the recent proposition that an HSD causes hyperglucocorticoidism in mice and humans (23, 24).

Fig. 3 HSD-induced systemic effects.

Mice were fed NSD or HSD for 1 week, and spleen, bone marrow blood, and kidneys were analyzed. (A) RT-PCR quantification of spleen Nfat5 expression relative to Gapdh and control (NSD) under homeostatic conditions in mice. (B) Volcano plot showing differentially regulated genes in neutrophils isolated from the bone marrow of NSD- or HSD-fed uninfected mice and heatmap showing expression of selected genes in bone marrow neutrophils from HSD-fed mice. GC, glucocorticoid. (C) ELISA quantification of serum corticosterone in uninfected mice fed NSD or HSD for 1 week. (D) INTEGRA clinical analyzer quantification of urea content in kidneys of uninfected mice fed NSD or HSD for 1 week. (E) In silico prediction of glucocorticoid response elements in Nfat5 promoter. ARE, androgen response element; GRE, glucocorticoid response element; PRE, progesterone response element; Ar, androgen receptor; Nr3c1, glucocorticoid receptor; Nr3c2, mineralocorticoid receptor; Pgr, progesterone receptor. (F) RT-PCR quantification of spleen Nfat5, Tsc22d3 (TSC22 domain family protein 3), Per1 (period circadian protein homolog 1), and Sgk1 (serum and glucocorticoid-regulated kinase 1) expression relative to Gapdh and control (NSD + veh) in mice injected subcutaneously for 1 week daily with 60 μg of dexamethasone (dex) or PBS. *P < 0.05 and **P < 0.01 by unpaired Student’s t test. Bar graphs indicate means and SEM. The experiments were performed at least twice with four (A) or five (B to F) mice per group. Graphs depict pooled experiments.

These recent studies proposed that the hyperglucocorticoidism induces catabolic processes that elevate urea in the circulation and renal medulla (23, 24). We detected elevated urea concentrations not only in the renal medulla but also in the cortex of HSD-fed mice (Fig. 3D), similar to the pattern of intrarenal Nfat5. However, urea failed to increase Nfat5 expression in cultured kidney epithelial cells (fig. S5B), arguing against glucocorticoid-induced urea as the factor that had up-regulated Nfat5 in HSD-fed mice.

Examining the Nfat5 gene promoter region in databases revealed putative response elements for the glucocorticoid receptor Nr3c1 and mineralocorticoid receptor Nr3c2 (Fig. 3E), suggesting that glucocorticoids might directly induce Nfat5 expression. Injection of the glucocorticoid dexamethasone systemically up-regulated expression of Nfat5, Gilz, and Per1 (Fig. 3F). Moreover, it induced serum and glucocorticoid-regulated kinase 1 (SGK1) (Fig. 3G), a downstream target of Nfat5 (27) that has also been linked to HSD-aggravated autoimmunity (7, 8). These findings indicated that the systemic Nfat5 elevation in HSD-fed mice resulted from systemic hyperglucocorticoidism.

Mechanism of HSD-induced hyperglucocorticoidism

We wished to understand how an HSD causes hyperglucocorticoidism. We noted that glucocorticoid serum concentrations in HSD-fed mice were elevated only in the morning, when the diurnal glucocorticoid rhythm of nocturnal animals is at its minimum, but not in the evening (Fig. 4A). Moreover, adrenocorticotropic hormone (ACTH), the pituitary hormone that regulates this rhythm, was strongly suppressed (Fig. 4B), indicating that hyperglucocorticoidism did not originate from ACTH-stimulated glucocorticoid synthesis, which takes place in the adrenal zona fasciculata. Aldosterone is produced in the zona glomerulosa, from steroid precursors via the enzyme aldosterone synthase [Cyp11b2 (cytochrome P450 family 11 subfamily B Member 2)], which, in response to angiotensin II, converts the glucocorticoid corticosterone into aldosterone as the last step of mineralocorticoid synthesis. In HSD, angiotensin II serum concentrations (Fig. 4C), adrenal expression of Cyp11b2 (Fig. 4D), and aldosterone production (Fig. 4E) were suppressed, consistent with their role in sodium excretion. Cyp11b2 suppression caused accumulation of the aldosterone precursor corticosterone. We speculated that glucocorticoids produced by the mineralocorticoid synthesis pathway might have suppressed the ACTH regulatory circuit. If so, then stimulating Cyp11b2 should reenable corticosterone conversion into aldosterone. When we infused angiotensin II into mice on an HSD, adrenal Cyp11b2 expression was increased (Fig. 4F), aldosterone production was elevated (Fig. 4G), and the morning glucocorticoid concentrations were low again (Fig. 4H), implying that the diurnal rhythm was active again. These findings indicated that the hyperglucocorticoidism on an HSD was a biochemical consequence of the hormonal response of the adrenal gland required for salt excretion.

Fig. 4 Mechanism of the HSD-mediated glucocorticoid increase.

Mice were fed NSD or HSD for 1 week, and blood and adrenal glands were analyzed. (A to C) ELISA quantification of serum corticosterone at 9 a.m. and 6 p.m. (A), ACTH (B), and serum angiotensin II (C) in uninfected mice fed HSD and NSD for 1 week. (D) RT-PCR quantification of adrenal gland Cyp11b2 (aldosterone synthase gene) expression in uninfected mice fed HSD and NSD for 1 week. (E) ELISA quantification of serum aldosterone in uninfected mice fed HSD and NSD for 1 week. (F) RT-PCR quantification of adrenal gland Cyp11b2 expression in mice fed HSD and implanted with angiotensin II (AngII) osmotic minipumps (1.25 ng/min per gram) for 1 week. (G and H) ELISA quantification of serum aldosterone (G) and serum corticosterone (H) in mice fed HSD and implanted with angiotensin II osmotic minipumps (1.25 ng/min per gram) for 1 week. *P < 0.05, **P < 0.01, and ***P < 0.001 by unpaired Student’s t test (A to H). Bar graphs indicate means and SEM. The experiments were performed at least twice with five (A to C and E) and six (D and F to H) mice per group. Graphs depict pooled experiments.

Neither HSD-induced aldosterone decrease, nor glucocorticoid-induced urea, Nfat5, or glucosuria exacerbate pyelonephritis

We next asked which of the HSD consequences had exacerbated pyelonephritis, starting with hypoaldosteronism. In addition to its electrolyte-regulating function, aldosterone also has immunostimulatory properties (28), suggesting that its reduction in HSD might weaken the antibacterial defense. However, neither treating NSD-fed mice with the aldosterone inhibitor spironolactone nor providing aldosterone to HSD-fed mice altered the course of pyelonephritis (Fig. 5A), arguing against a role of aldosterone.

Fig. 5 Neither HSD-induced aldosterone decrease nor glucocorticoid-induced urea, Nfat5, or glucosuria exacerbate pyelonephritis.

Mice were fed NSD or HSD for 1 week and infected with UPEC. Eighteen hours after infection, kidneys were analyzed. Additional treatments are indicated below. (A) Quantification of kidney colony-forming units 18 hours after infection in mice treated for 1 week with 50 mg/kg per day of spironolactone (spiro) or 50 μg/kg per day of aldosterone. (B) Flow cytometric quantification of GFP+ UPEC phagocytosis by bone marrow neutrophils exposed to urea concentrations found in the OMed of mice fed NSD (+60 mM) or HSD (+80 mM). (C) In vitro bacterial viability in bone marrow neutrophils exposed to an additional 60 or 80 mM urea. Gentamicin protection assay and phagocytosis combined. (D) Flow cytometry quantification of dead bone marrow neutrophils after exposure to 60 and 80 mM urea for 3 hours in vitro. (E) Flow cytometry quantification of apoptotic and necrotic bone marrow neutrophils after exposure to 600 or 1200 mM urea for 3 hours in vitro. (F to H) Mice have been exposed to HSD for 1 week and injected intraperitoneally with vehicle or 80 mg/kg per day of arginase inhibitor N-ω-hydroxy-l-norarginine (NOHA). (F) Quantification of colony-forming units in kidney homogenates 18 hours after infection. (G) Flow cytometric quantification of kidney neutrophils (viable CD45+, Ly6Cint, and Ly6G+). (H) Quantification of bacterial viability in neutrophils isolated from infected kidneys based on gentamicin protection assay and phagocytosis. (I to K) Mice were exposed to NSD or HSD for 1 week and injected subcutaneously twice with 2 mg/kg of unspecific (Neg1) or Nfat5-targeting (iNFAT5) antisense oligonucleotides. (I) Quantification of kidney colony-forming units 18 hours after infection. (J) Flow cytometric quantification of kidney neutrophils. (K) Quantification of bacterial viability in neutrophils isolated from infected kidneys based on gentamicin protection assay and phagocytosis. (L) Absorbance quantification (wavelength, 570 nm) of UPEC growth in urine collected from mice fed NSD or HSD for 1 week. *P < 0.05 and ***P < 0.001 by one-way ANOVA (A, E, and I to K) and unpaired Student’s t test (B to D, F to H, and L). Bar graphs indicate means and SEM. The experiments were performed at least twice with five (A and L), four (B to D), three (E), six (F to H), and seven (I to K) mice per group. Graphs depict pooled experiments.

Next, we examined consequences of HSD-induced hyperglucocorticoidism, starting with the intrarenal urea elevation. Urea is a chaotropic agent that can impair actin polymerization in neutrophils (29), so high intrarenal urea concentrations might impair actin-dependent phagocytosis of UPEC by neutrophils. Neutrophils cultured with 80 mM urea, the highest concentration detected in mice under HSD, showed reduced phagocytosis (Fig. 5B) and intracellular killing of UPEC (Fig. 5C), without signs of cell death (Fig. 5D). Only physiologically excessive urea caused neutrophils to die (Fig. 5E). In contrast, macrophage function was not affected by urea (fig. S6, A and B).

To investigate the functional relevance of high intrarenal urea in pyelonephritis, we treated mice on an HSD with N-omega-hydroxy-l-norarginin (NOHA), an arginase inhibitor we and others previously used to inhibit urea formation (23). NOHA treatment seemed to somewhat improve pyelonephritis, but this was not statistically significant (Fig. 5F). Likewise, we failed to detect significant changes in intrarenal numbers of neutrophils (Fig. 5G) nor in their bactericidal function in NOHA-treated mice (Fig. 5H). These findings indicated that urea-mediated neutrophil suppression cannot fully explain the exacerbation of pyelonephritis under HSD.

It appeared theoretically possible that glucocorticoid-induced Nfat5 might be immunosuppressive, unlike sodium-induced Nfat5, which is proinflammatory (5, 9, 25). To address this, we decided to reduce Nfat5 expression comparable to NSD with Nfat5-specific antisense oligonucleotides (fig. S7A) that allowed in vivo Nfat5 reduction depending on the duration of treatment (fig. S7, B and C). Thereby, we decreased medullary Nfat5 expression by 50%, i.e., to the concentration we had observed in NSD mice. However, Nfat5 normalization altered neither the course of pyelonephritis (Fig. 5I), nor intrarenal neutrophil numbers (Fig. 5J), nor their antibacterial function (Fig. 5K), arguing against an effect of the glucocorticoid-induced Nfat5 increase in pyelonephritis.

Glucosuria is another typical consequence of hyperglucocorticoidism. In patients with diabetes mellitus, glucosuria is believed to increase the UTI risk by promoting UPEC growth in the urine (30). However, HSD-fed mice were normoglycemic (fig. S8A), and their urinary glucose concentrations were hardly increased (2.5×) (fig. S8B), much less so than in diabetic mice and humans. Moreover, UPEC cultured in their urine grew at the same rate as in urine from NSD-fed mice (Fig. 5L), excluding glucosuria as a major pathogenic factor in HSD.

HSD-induced hyperglucocorticoidism directly suppresses neutrophils and aggravates bacterial infections

Glucocorticoids are clinically used as potent immunosuppressants (31), suggesting that they might directly impair the neutrophil response in HSD. In support of this, gene set enrichment analysis (GSEA) of our RNA-seq transcriptome analysis of bone marrow neutrophils revealed down-regulation of various innate immune functions, especially phagosome maturation, important for bactericidal activity of neutrophils. Several other gene sets relevant for such activity were moderately reduced as well (table S1). Immediate early response genes were up-regulated, suggesting that neutrophils from HSD mice developed while being exposed to a systemic stress–associated factor, such as glucocorticoids. In support of this, dexamethasone treatment inhibited neutrophil, but not macrophage function in vitro (Fig. 6Aand fig. S9), and aggravated pyelonephritis after injection into NSD-fed mice (Fig. 6B). Moreover, neutrophils from these mice were impaired in their ability to kill phagocytosed UPEC (Fig. 6C), mimicking the phenotype of neutrophils in HSD mice. Treating HSD mice with the glucocorticoid receptor inhibitor, mifepristone, improved pyelonephritis (Fig. 6D), and neutrophils from these mice were superior at eliminating UPEC in comparison to neutrophils from control mice fed HSD (Fig. 6E). This indicated that hyperglucocorticoidism was the factor that impaired the pyelonephritis defense during an HSD.

Fig. 6 HSD-induced hyperglucocorticoidism suppresses the defense against pyelonephritis and listeriosis.

Mice were fed NSD or HSD for 1 week and analyzed or infected with UPEC or L. monocytogenes and analyzed. Additional treatments are indicated below. (A) Quantification of in vitro bacterial viability in bone marrow neutrophils exposed to different dexamethasone concentrations. Gentamicin protection assay and phagocytosis combined. (B) Quantification of kidney colony-forming units 18 hours after infection in mice injected subcutaneously for 1 week daily with 60 μg of dexamethasone or vehicle. (C) Quantification of bacterial viability in neutrophils isolated from infected kidneys based on gentamicin protection assay and phagocytosis combined. (D) Quantification of kidney colony-forming units 18 hours after infection in mice injected subcutaneously for 1 week daily with 1 mg of mifepristone (mif) or vehicle. (E) Quantification of bacterial viability in neutrophils isolated from infected kidneys based on gentamicin protection assay and phagocytosis combined. (F) Quantification of in vitro bacterial viability in blood neutrophils from uninfected mice fed NSD or HSD for 1 week. Gentamicin protection assay and phagocytosis combined. (G) Gene set enrichment analysis (GSEA) plot for gene ontology (GO) biological process phagosome maturation in neutrophils isolated from the bone marrow of NSD- or HSD-fed uninfected mice. NES, normalized enrichment score; FDR, false discovery rate. (H) Quantification of in vitro phagosomal pH in blood neutrophils from uninfected mice fed NSD or HSD for 1 week. (I) Quantification of spleen colony forming units 3 days after L. monocytogenes infection [5 × 104 per mouse intraperitoneally (i.p.)]. Mice were fed NSD or HSD for 1 week. (J) Quantification of bacterial viability in neutrophils isolated from infected spleens based on gentamicin protection assay and phagocytosis combined. (K) Quantification of spleen colony-forming units 3 days after L. monocytogenes infection (5 × 104 per mouse i.p.). Mice were fed HSD and injected subcutaneously for 1 week with 1 mg of mifepristone daily. (L) Quantification of bacterial viability in neutrophils isolated from infected spleens based on gentamicin protection assay and phagocytosis combined. *P < 0.05, **P < 0.01, and ***P < 0.001 by unpaired Student’s t test (A to L). Bar graphs indicate means and SEM. The experiments were performed at least twice with five (A and H to L), seven (B to E), and three (F) mice per group. Graphs depict pooled experiments.

The systemic elevation of glucocorticoid levels implied that the neutrophil-mediated antibacterial defense should be compromised system-wide. In support of this, the microbicidal activity of neutrophils was reduced also in the circulation of HSD-fed mice (Fig. 6F). Moreover, the expression of gene sets associated with phagosome maturation was lower in neutrophils from HSD mice (Fig. 6G), and their phagosomal pH was increased (Fig. 6H and fig. S10A) to an extent that permits better E. coli growth (fig. S10B) (32), reinforcing the notion that neutrophils were functionally compromised during an HSD.

We therefore infected HSD mice with the bacterium Listeria monocytogenes that causes systemic infections, which may be fatal, especially in children, the elderly, and in immunocompromised patients (33). The early innate immune response against Listeria is glucocorticoid sensitive and depends on neutrophils (3335). One hundred– to 1000-fold higher colony-forming unit (CFU) numbers were detected in spleens (Fig. 6I) and livers (fig. S11A) of Listeria-infected HSD-fed mice. Consistent with the reduced intracellular killing of UPEC in pyelonephritis, neutrophils were less able to eradicate phagocytosed Listeria (Fig. 6J). Glucocorticoid receptor inhibition mostly prevented the exacerbation of listeriosis by HSD (Fig. 6K and fig. S11B) and reinvigorated the antibacterial neutrophil response (Fig. 6L). These findings demonstrated that HSD-induced glucocorticoids compromised the antibacterial immune response systemically.

An HSD abrogates the diurnal glucocorticoid rhythm and reduced bactericidal activity of neutrophils in humans

Last, we wished to corroborate our findings with epidemiological data in humans. However, there are no publications examining a potential association between salt intake and UTI risk in humans. Therefore, we inspected online databases on salt consumption (36) and UTI incidence world-wide (37) and found a positive correlation between these two parameters (fig. S12). Although these studies did not discriminate between cystitis and pyelonephritis and represent countries and not individuals, they were consistent with a positive correlation between salt consumption and the incidence of pyelonephritis.

To more directly address this possibility, we performed a clinical study with healthy volunteers, who consumed additional 6 g of sodium per day for 1 week as previously described (11). This amount is equivalent to 1.5 to 2 large fast food hamburger meals. After this 1-week period on HSD, the salt concentration in urine increased (Fig. 7A), and blood aldosterone decreased (Fig. 7B), as expected. In addition, in humans, an HSD increased the serum corticosterone evening concentrations, i.e., at the minimum of the diurnal rhythm, compared to the concentrations before starting the HSD (Fig. 7C), almost to morning concentrations (8 a.m.; 44.8 ± 5.1 versus 39.1 ± 5.9 ng/ml; n = 10, not significant), indicating loss of the diurnal glucocorticoid rhythm, as we had observed in the mice. The number of blood neutrophils remained unaffected (Fig. 7D). When blood neutrophils were exposed to UPEC ex vivo, their capacity to digest intracellular bacteria was lower in almost every individual (Fig. 7E), confirming in the human system that an HSD reduces the bactericidal activity of neutrophils.

Fig. 7 An HSD causes hyperglucocorticoidism and worsens ex vivo antibacterial neutrophil function in humans.

Analysis of urine and plasma values of 10 healthy human volunteers that consumed 6 g/day of NaCl for 1 week in addition to their normal diet. Quantification of sodium content in morning urine (A), aldosterone (B), and corticosterone (C) in plasma collected in the evening, numbers of blood neutrophils (live, CD16+, and CD15+ cells). (D) Quantification of in vitro bacterial viability in blood neutrophils. (E) Gentamicin protection assay and phagocytosis combined. *P < 0.05 paired Student’s t test. Dots indicate each of the 10 individual human volunteers.

DISCUSSION

The Western diet contains high amounts of salt and cholesterol, which are widely considered to be proinflammatory (7, 8, 11, 38). In particular, an HSD resulted in sodium accumulation in the skin, and this has been shown to strengthen the macrophage-driven defense against experimental cutaneous leishmaniosis (5). In line with these findings, it was recently proposed that the intrarenal sodium gradient improves the local defense against bacterial infection by attracting macrophages (14). An HSD is linked with cardiovascular side effects, hampering its use in the treatment of infections (39). Our present study revealed that an HSD is not only proinflammatory but also has distinct systemic immunosuppressive effects that compromise the antibacterial immune defense. This effect emerged as a direct consequence of the physiologic hormonal response that initiates the excretion of excess sodium. Lowering RAAS-regulated mineralocorticoid synthesis caused accumulation of precursors with glucocorticoid functionality, which abolished the physiologic diurnal ACTH-driven glucocorticoid rhythm and suppressed neutrophils systemically, both in mice and in human volunteers.

Various forms of stress may cause hyperglucocorticoidism (40). However, stress stimulates an ACTH increase that increases glucocorticoids, whereas, in our study, ACTH concentrations were decreased in mice under an HSD, excluding that stress was the cause of the elevated glucocorticoids.

Our results also shed new light on Nfat5, which is often viewed as a sodium- or osmotic stress–specific transcription factor. The disconnection between sodium and Nfat5 expression in the kidneys of HSD-fed mice implied that Nfat5 is regulated by a different factor, which we identified as glucocorticoids. When we reduced Nfat5 expression by antisense oligonucleotides to NSD levels, pyelonephritis was unaltered, indicating that the glucocorticoid-induced Nfat5 increase lacked functional relevance in this infection. Previous studies had deleted the Nfat5 gene by cell-specific knockout methods and found an exacerbation of pyelonephritis (14). One possible explanation for this discrepancy is that gene knockout removes the Nfat5 gene from targeted cells entirely. However, Nfat5 has important protective functions against hyperosmotic stress, and its complete loss would render targeted cells vulnerable in the hyperosmolar microenvironment of the renal medulla (41). The use of inhibitory antisense oligonucleotides circumvents this problem by normalizing rather than annulling gene expression in targeted cells. On the other hand, potential immunomodulatory side effects of such oligonucleotides need to be considered.

These findings raised the question of why an HSD promoted immunity in previous studies (5, 7, 8, 11) but not in ours on pyelonephritis. One difference is the organ examined: An HSD causes sodium accumulation in the intestinal lumen and in the skin, and consequently, it can stimulate gut or dermal immunity by activating the Nfat5 pathway (58). By contrast, we found that sodium did not accumulate in the kidney during an HSD, ruling out sodium-mediated stimulation of intrarenal immunity. This counterintuitive finding can be explained by the normal physiologic response to an HSD, i.e., the suppressed RAAS activity, which reduces sodium reabsorption in the renal medulla, to allow for salt excretion. The need to switch off sodium transporters implies that this osmolyte cannot be used to build the medullary osmotic gradient during an HSD. Recent groundbreaking studies have offered a solution to this conundrum: The kidney builds the intrarenal osmotic gradient during an HSD from urea, another well-known osmolyte, which is produced in response to hyperglucocorticoidism (23, 24). This metabolic adjustment may be one of the physiologic purposes of the spill-over hyperglucocorticoidism during an HSD, namely, to induce the production of an organic osmolyte that can substitute for sodium (23, 24). However, while both osmolytes are effective at renal water conservation, only sodium induced Nfat5 expression and stimulated macrophages, consistent with previous findings (5).

A second difference between our study and previous reports showing proinflammatory consequences of an HSD is the immune cell type involved. Although sodium activated macrophages in our present study and previous ones (5, 14, 25), it failed to stimulate neutrophils, the relevant immune effector cells in pyelonephritis, listeriosis (35), and other bacterial infections. Moreover, glucocorticoids and urea inhibited bacterial digestion by neutrophils, but either stimulated this process in macrophages (25, 42) or had no effect on them in our present study. Macrophages are critical for the defense against leishmaniosis, in contrast to neutrophils, which are detrimental in this disease (5, 43). Different target cells may also explain why the previously reported aggravation of experimental multiple sclerosis under an HSD (711) was not antagonized by the hyperglucocorticoidism resulting from this diet: Experimental multiple sclerosis depends on T cells and macrophages but not on neutrophils. Consistent with this interpretation, steroids are not recommended as effective long-term multiple sclerosis therapy (44).

Third, we found that the osmolytes sodium and urea had distinct direct immunologic effects on neutrophils, with urea being inhibitory, whereas sodium had little effect. However, the intrarenal urea elevation did not fully explain the exacerbation of pyelonephritis during an HSD. Instead, the glucocorticoids themselves were mainly responsible for suppressing the antibacterial neutrophil response, as evident from our experiments using inhibitors for both pathways.

In conclusion, we have demonstrated that an HSD does not strengthen neutrophil responses but entails hormonal and metabolic alterations that systemically compromise their antibacterial activity (fig. S13). Hence, the recent proposition that the intrarenal sodium gradient establishes a defense zone against pyelonephritis (14) must not be misinterpreted as implying that an HSD will support the defense against pyelonephritis. Instead, such a diet may be harmful in bacterial infections of the urinary tract and in other organs that do not accumulate sodium. Future studies are warranted to identify which infections beyond pyelonephritis and listeriosis are aggravated by an HSD. Moreover, our findings did not establish epidemiologically that an HSD is associated with a higher incidence or severity of bacterial infections. Therefore, clinical studies in patients with UTI are needed to test clinical implications of our experimental observations. We found that low salt intake was not beneficial either, presumably because it cannot reduce glucocorticoid abundance below that was maintained by the ACTH regulatory hormone circuit. Last, our findings in mice and humans indicate that the Western diet is not exclusively proinflammatory (7, 8, 11, 38) because its high salt content has selective immunosuppressive consequences as well. This may be relevant for the current discussion about the recommended daily sodium intake.

MATERIALS AND METHODS

Study design

The objective of the study was to evaluate the influence of HSD on immune response during bacterial infection of the kidney. To investigate that, mice were fed NSD or HSD for 1 week, and their blood, spleen, adrenal glands, and kidneys were analyzed under homeostatic conditions as well as 18 hours after infection with UPEC. Once the parameters affected by HSD were established, various drugs were given to normalize them, and pyelonephritis outcome was analyzed. Because the results showed that glucocorticoids, which are systemic, were responsible for pyelonephritis worsening, a systemic infection with L. monocytogenes was performed. Last, we investigated human neutrophils from healthy volunteers consuming an additional 6 g of NaCl per day ex vivo. All data have been replicated in independent experiments with three or more mice, which were assigned randomly to the experimental groups. Rules for stopping data collection were determined by ethics application for the project. Primary data are reported in data file S1.

Clinical study design

The study in humans was approved by the ethics committee of the Medical Faculty of the Friedrich Wilhelm University, Bonn, Germany. All volunteers provided written informed consent before the study. Ten healthy males and females (four and six, respectively), ages 20 to 50, took part. Initially, a morning urine sample and an evening blood sample were collected. After 7 days of consuming 6 g of NaCl per day in form of tablets three times a day in addition to their normal diet, another morning urine sample and an evening blood sample were obtained. Neutrophils from the blood were isolated with the EasySep Direct Human Neutrophil Isolation Kit (STEMCELL Technologies) and cultured with green fluorescent protein (GFP)–expressing UPEC (45). For measuring phagocytosis, 2 × 105 neutrophils were cultured with 5 × 106 heat-killed GFP-expressing E. coli for 3 hours. After that, the cells were stained in fluorescence-activated cell sorting (FACS) buffer containing gentamicin and analyzed by flow cytometry. For measuring intracellular bacterial survival, 2 × 105 neutrophils were cultured with 2 × 106 viable bacteria for 2 hours. Subsequently, cells were incubated with gentamicin (50 μg/ml; Sigma-Aldrich) at 37°C for 1 hour to kill extracellular bacteria. Next, cells were incubated for 3 hours and lysed with saponin-based buffer (BD Biosciences). The suspension was plated on LB-agar CPSE plates (BioMérieux) and cultured overnight at 37°C before counting CFUs. Intracellular bacterial viability rate was calculated as follows: Total bacteria were measured by flow cytometry (E. coli mean fluorescent intensity), and live bacteria were measured by CFU. To calculate intracellular bacterial viability rates, both measures were normalized to the average of the control (placebo) group, and normalized live bacterial counts were divided by total bacterial counts. Urine sodium content was analyzed with INTEGRA clinical analyzer c501 (cobas). Plasma corticosterone and aldosterone were evaluated by enzyme-linked immunosorbent assay (ELISA; Enzo Life Sciences).

Animals

The experiments were performed according to the recommendations of the Federation of European Laboratory Animal Science Association and approved by Behörde für Gesundheit und Verbraucherschutz Hamburg and Landesamt für Natur, Umwelt und Verbraucherschutz north rhine-westphalia comities. Seven- to 10-week-old mice were bred at the central animal facilities of Bonn University or purchased from JANVIER LABS. Age-matched C57BL/6 mice were used for all the experiments.

Pyelonephritis model

Pyelonephritis was induced as described before (45, 46). Briefly, female mice were anesthetized with isoflurane and 1010 CFUs of UPEC strain 536 (provided by J. Hacker) were instilled into the bladder. The procedure was repeated 3 hours later. Water intake was equalized in all pyelonephritis groups during the last 10 hours before analysis to ensure similar urine flow and to avoid differences in flushing bacteria from the urinary tract.

High-salt diet

Mice were given low-salt diet (0.1% Na), NSD (0.3% Na), or HSD (1.71% Na + 0.9% NaCl autoclaved tap water) (Sniff) ad libitum for 1 week.

Listeriosis infection model

Mice were injected intraperitoneally with 5 × 104 CFUs of L. monocytogenes (EGDe strain) (47). Three days later, the mice were euthanized, and the organs were processed for flow cytometry or CFU assessment on BHI agar plates.

In vivo treatments of mice

Dexamethasone (60 μg per mouse; Jenapharm) in phosphate-buffered saline (PBS) was administered subcutaneously every day for 1 week to mice on NSD. Mifepristone (1 mg per mouse per day; Sigma-Aldrich) in dimethyl sulfoxide was injected subcutaneously for 1 week to mice receiving HSD. Angiotensin II osmotic minipumps (1.25 ng/min per gram) purchased from ALZA Corporation were implanted in mice fed HSD for 1 week as described (48). Mice on NSD were given spironolactone (0.1 mg/ml; Hexal) in drinking water ad libitum, corresponding to a dose of 50 mg/kg per day. Mice on HSD were given aldosterone (0.2 μg/ml; Sigma-Aldrich) in drinking water ad libitum, corresponding to a dose of 50 μg/kg per day. Nfat5 antisense oligonucleotides or control oligonucleotides were designed, synthesized (QIAGEN), and applied similarly to a previously report (49), except that they were injected subcutaneously (2 mg/kg) on days 5 and 3 before pyelonephritis induction. To assess Nfat5 knockdown, kidney tissue was pulverized, and the QuantiGene SinglePlex Gene Expression Assay (Thermo Fisher Scientific) was performed according to the manufacturer’s protocol. Mice on HSD were injected intraperitoneally with 80 mg/kg per day of NOHA (Bachem) for 1 week.

Flow cytometry

The organs were digested for 40 min at 37°C in media containing deoxyribonuclease and collagenase (Sigma-Aldrich) to obtain single-cell suspensions. Dead cells were excluded from the analysis via Fixable Viability Dye eFluor 780 (eBioscience). All antibodies were purchased from BioLegend, BD Biosciences, and Abcam. Leukocytes were stained with CD45-V421 antibody (clone 30-F11). Neutrophils were identified as Ly6Cintermediate (clone HK1.4 PerCP-Cy5.5) and Ly6Ghigh (clone 1A8 APC). Dendritic cells were stained with CD11c PerCP-Cy5.5 (clone HL3) and MHCII fluorescein isothiocyanate (FITC) (clone M5/114.15.2). Reactive oxygen species production was assessed with 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) from Thermo Fisher Scientific. The following antibodies were also used: TNF APC (allophycocyanin) (clone MP-XT22), iNOS (inducible nitric oxide synthase) FITC (clone 6/iNOS/NOS type II), and CD11b PE-Cy7 (clone M1/710). Neutrophil sodium content was evaluated using Asante NaTRIUM Green-2 (TEFLabs). This dye was present in the medium used to digest the kidneys and subsequently was removed by washing to avoid an influence of the washing buffer. Apoptotic cells were identified with Annexin V–FITC staining (BD Biosciences) and Hoechst (Thermo Fisher Scientific). For intracellular staining, cells were fixed in PBS containing 2% glucose, 1% formaldehyde, and 0.02% sodium azide, permeabilized with 1× Perm/Wash buffer (eBioscience), and stained for 45 min at room temperature in 1× Perm/Wash buffer. Phagosomal pH was determined using pHrodo Green E. coli BioParticles (Thermo Fisher Scientific) after 1 hour of incubation according to the manufacturer’s protocol. Human neutrophils were stained with CD16-APC (clone 3G8) and CD15-PB (clone MMA) (BioLegend). Samples were measured with BD FACSCanto II (BD Biosciences) and analyzed with FlowJo software, version 10.4.2.

Phagocytosis of UPEC

We determined the uptake of UPEC by intracellular staining with anti–E. coli lipopolysaccharide (LPS) antibody (clone 2D7/1) labeled with the Alexa Fluor 488 Antibody Labeling Kit (Thermo Fisher Scientific) and analysis via flow cytometry. This staining protocol was validated by adding increasing amounts of UPEC to neutrophils and showing a positive correlation to the flow cytometric signal (fig. S14A). We also showed that the staining intensity of nonpermeabilized neutrophils was less than 5% of that after intracellular staining (fig. S14B), excluding that extracellular UPEC or debris thereof had a major influence on analysis.

In vivo intracellular bacterial viability assay

Infected kidney homogenates were incubated for 1 hour at 37°C in RPMI 1640 media (Thermo Fisher Scientific) containing gentamicin (50 μg/ml; Sigma-Aldrich) to kill extracellular bacteria. Kidney neutrophils were isolated using the Anti-Ly6G MicroBead Kit (Miltenyi) according to the manufacturer’s protocol. Neutrophils were stained intracellularly with anti–E. coli LPS antibody as described above to obtain a readout for phagocytosis. Moreover, isolated neutrophils were washed with PBS and lysed in saponin-based buffer (BD Biosciences). Subsequently, the suspension was plated on LB-agar CPSE plates (BioMérieux) and cultured overnight at 37°C. Last, CFUs were counted manually, yielding the amount of viable intracellular bacteria. The intracellular bacterial viability rate was calculated as described above.

In vitro assays with macrophages and human embryonic kidney 293 cells

For gentamicin protection assay, bone marrow–derived macrophages were generated in Teflon Bags and infected with a multiplicity of infection of 100 E. coli HB101 as described (50). Together with the bacteria, salt (+0 and + 40 mM), urea (60 and 80 mM), and dexamethasone (250 and 750 ng/ml) were added, and the infection was synchronized by centrifugation. The cells were then incubated for 1 hour in the incubator. Afterward, cells were washed twice with PBS and incubated for 1 hour in medium containing gentamicin (100 μg/ml). Salt, urea, and dexamethasone were added again. After 1 hour of incubation, cells were lysed with PBS containing 0.1% Triton X-100 and 0.05% Tween 80 on ice. Serial dilutions were generated and plated on Mueller Hinton II agar plates. After 24 hours at 37°C, CFUs were counted.

For uptake assay, bone marrow–derived macrophages were generated in Teflon Bags and incubated with Latex beads (Sigma-Aldrich) for 30 min at 37°C in low adherence plates. Salt, urea, and dexamethasone were added simultaneously after the incubation time. Human embryonic kidney 293 kidney epithelial cells were exposed to +0 or + 40 mM NaCl or 60 and 80 mM urea for 4 hours, and Nfat5 was quantified by reverse transcription polymerase chain reaction (RT-PCR).

Microscopy

Immunofluorescence microscopy on 5-μm Tissue-Tek (Sakura) kidney cryosections was performed using E. coli GFP and anti-GFP antibody (Life Technologies) and CD11b-APC and CD45-V421 antibodies (BioLegend). An LSM 780 (ZEISS) microscope was used for imaging.

Real-time PCR analysis

RNA was isolated with NucleoSpin RNA (MACHEREY-NAGEL). The High-Capacity cDNA Reverse Transcription Kit was from Life Technologies. RT-PCR primers for Gapdh, Nfat5, Tsc22d3, Sgk1, and Per1 were from QIAGEN, and Cyp11b2 primer was acquired from Bio-Rad. SYBR Green was produced by Life Technologies. The measurement was performed with LightCycler 480 II (Roche).

Kidney sodium content

Kidneys were divided into the cortex, outer medulla, and inner medulla with a borer sampler and snap-frozen, and the weight was determined. Subsequently, they were dried for 72 hours at 90°C. The dry weight was determined. Ashing was performed for 24 hours at 190° and then at 450°C. The tissues were further ashed for 48 hours at 600°C and solved in 5% HNO3. Atomic absorption spectrometry was performed with Model 3100 (PerkinElmer) as described (4) to determine sodium concentration.

Kidney urea content

Kidneys were divided into the renal cortex, outer medulla, and inner medulla with a borer sampler and frozen in liquid nitrogen. After weight determination, tissues were lysed in radioimmunoprecipitation assay buffer (Thermo Fisher Scientific), homogenized with ULTRA-TURRAX (Sigma-Aldrich), and centrifuged, and the supernatant was collected and stored at −80°C until analysis with an INTEGRA clinical analyzer c501 (cobas).

Evaluation of substances in serum, urine, and kidney homogenate

Corticosterone, aldosterone, angiotensin II, and ACTH were evaluated by ELISA (Enzo Life Sciences). Urine glucose was measured with the Glucose Colorimetric Assay Kit from Cayman Chemical. Kidney supernatant cytokine concentrations were assessed with LEGENDplex (BioLegend).

Identification of glucocorticoid responsive and related elements on Nfat5 promoter

The Genomatix database (Munich, Germany) was screened for known and validated promoters of murine Nfat5 (mm38) that contain glucocorticoid responsive elements to enable transcriptional regulation after glucocorticoid treatment. The promoter GXP_142189 had androgen response element (ARE), glucocorticoid response element (GRE), and progesterone response element (PRE) sites, which are categorized as glucocorticoid responsive, and related elements. All three sites were found in close proximity to each other, upstream of the transcriptional start side (TSS). Binding motifs for the transcription factors Ar (androgen receptor), Nr3c1, Nr3c2, and Pgr were found in glucocorticoid responsive elements, suggesting active participation in the transcriptional regulation of Nfat5.

E. coli growth in NaCl-supplemented media and urine

A total of 5.0 × 107 UPEC/ml was added to LB media supplemented with additional NaCl or to collected urine. They were cultured for 5 hours, and at different time points, optical density was measured (wavelength, 570 nm) with Safire2, Tecan (Thermo Fisher Scientific).

In vitro culture

Bone marrow or blood neutrophils were isolated with the Neutrophil Isolation Kit (Miltenyi) and cultured with GFP-expressing UPEC. Isolated neutrophil viability was measured by flow cytometry and was typically ~80%. For the phagocytosis assay, 105 neutrophils were cultured in RPMI 1640 GlutaMAX medium (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific) with 2.5 × 106 heat-killed E. coli for 3 hours. Afterward, the cells were stained in FACS buffer containing gentamicin and measured with a flow cytometer. For the gentamicin protection assay, 105 neutrophils were cultured with 2 × 105 living bacteria for 2 hours. Subsequently, the cells were incubated with gentamicin (50 μg/ml; Sigma-Aldrich) at 37°C for 1 hour and washed. Next, they were incubated for 3 hours and lysed with saponin-based buffer (BD Biosciences). Subsequently, the suspension was plated on LB-agar CPSE plates (BioMérieux) and cultured overnight at 37°C before counting CFUs.

3′mRNA-seq of bone marrow neutrophils

Bone marrow neutrophils were isolated from NSD- or HSD-fed mice by cell sorting (live, Ly6G+ cells, purity 99%). Cells were lysed in 200 μl of RLT buffer (QIAGEN). Total RNA was isolated using Zymo-Spin I columns (Zymo Research) and eluted in 8 μl of ribonuclease–free H2O. RNA concentrations were determined using Qubit (Lifetech). The 3′ mRNA-seq library was prepared using the forward 3′ mRNA-Seq Library Prep Kit for Illumina (Lexogen GmbH) according to the manufacturer’s protocol. Size distribution and yield of the 3′mRNA-seq library after the PCR step were determined by the D1000 high-sensitivity tape station (Agilent) before pooling of the barcoded libraries. The pooled 3′ mRNA-seq libraries were loaded on the Illumina HiSeq 2500 platform and analyzed by a 50-cycle high-output run (Next generation sequencing core facility at the Univesrity clinic, Bonn). Raw data are available through the European Nucleotide Archive (ENA) archive, accession: PRJEB28204, link: www.ebi.ac.uk/ena/data/view/PRJEB28204. FASTQ files were aligned to the mm10 mouse reference genome using the Star Aligner. Further computational 3′ mRNA-seq analysis was performed with the Bioconductor/R computing environment. The voom method of the limma package was used for normalization and linear modeling with quantile normalization. To obtain mRNA expression values, the read counts per million were transformed to log2 values according to the voom algorithm. Differentially expressed genes were determined by an empirical Bayes moderated t test statistic using the limma package. For GSEA, we used the gene set collection gene ontology (GO) biological processes provided by the Molecular Signature Database from the Broad Institute (https://software.broadinstitute.org/gsea/msigdb/). We used the t test statistic values as input ranking parametric for GSEA preranked gene list mode and default settings. For GSEA, we downloaded the javaGSEA stand-alone version from http://software.broadinstitute.org/gsea/downloads.jsp running on Java 8 (Oracle).

Statistical analysis

The experiments were repeated at least twice with at least three mice per group. Results are expressed as means ± SEM. Statistical differences between two groups were calculated using Student’s t test. Differences between three or more groups were calculated with one-way analysis of variance (ANOVA). Pearson’s correlation test was applied to test correlation significance. GraphPad Prism, version 6, was used to perform all statistical testing. Statistical significance was indicated as follows: *P < 0.05, **P < 0.01, and ***P < 0.001.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/12/536/eaay3850/DC1

Fig. S1. Influence of different salt concentrations on UPEC growth and measurement of urine excretion.

Fig. S2. No influence of an HSD on intrarenal cytokine production and neutrophil numbers.

Fig. S3. Sodium staining in kidney neutrophils.

Fig. S4. High salt effect on macrophages and T cells.

Fig. S5. HSD effect on steroid biosynthesis and urea effect on Nfat5.

Fig. S6. Effect of urea on macrophage function.

Fig. S7. Antisense oligonucleotide–mediated NFAT5 knockdown in the kidney.

Fig. S8. An HSD does not cause major changes in serum and urine glucose concentrations.

Fig. S9. Effect of dexamethasone on macrophage function.

Fig. S10. Measuring intraphagosomal pH.

Fig. S11. An HSD impairs the defense against Listeria in the liver.

Fig. S12. Epidemiologic correlation between human salt intake and UTI incidence in various countries.

Fig. S13. Mechanism of HSD-mediated glucocorticoid increase and pyelonephritis/listeriosis exacerbation.

Fig. S14. Validation of the method to measure UPEC phagocytosis.

Table S1. GO analysis of genes down-regulated in bone marrow neutrophils upon HSD exposure.

Data file S1. Primary data.

REFERENCES AND NOTES

Acknowledgments: We thank V. Kotov, S. Leopold, and M. Blankart for technical assistance; A. Böhner and M. Leweke for assistance in the clinical study; and T. Klockgether for expert neurologic advice on multiple sclerosis. Funding: We acknowledge the support by the Central Animal Facility, the Next Generation Sequencing Core Facility, and the Flow Cytometry Core Facility of the Medical Faculty at Bonn University. C.K. was supported by the Deutsche Forschungsgemeinschaft (DFG) (Gottfried-Wilhelm Leibniz Award, SFB1192 project A8 and SFBTR57 project 10), the Bundesministerium für Wirtschaft (grant ZF4507501587), and the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 668036 “RELENT”). J.J. was supported by DFG grant JA 1993/4-1. U.W. was supported by DFG grant SFB1192 project B7. N.G., Z.A., M.H., and C.K. are funded by the DFG under Germany‘s Excellence Strategy (EXC 2151) 390873048. Author contributions: K.J. and C.K. designed the study. K.J. performed most experiments. N.E.S., S. Schwab, M.E., M.A., O.B., D.H., P.N., M.R., U.W., Z.A., M.M., and J.J. performed individual experiments. K.J. and C.K. wrote the manuscript. K.J., C.K., S. Sivalingam, S. V. Schmidt, and M.H. generated the figures. K.H., S. Schwab, N.G., U.W., Z.A., and J.J. provided crucial ideas. K.J., N.S., S. Schwab, C.W., and C.K. designed, applied for, and performed the clinical study. All authors discussed and interpreted the results. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper, the Supplementary Materials, or ENA archive (accession: PRJEB28204, link: www.ebi.ac.uk/ena/data/view/PRJEB28204).

Stay Connected to Science Translational Medicine

Navigate This Article